University of Groningen Tryptophan Metabolism in Aging and Disease

Home , Cysteine dioxygenase, HAAO, Life extension

University of Groningen

Tryptophan metabolism in aging and disease Michels, Helen

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record

Publication date: 2017

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA): Michels, H. (2017). Tryptophan metabolism in aging and disease: Insights from Caenorhabditis elegans. [Groningen]: University of Groningen.

Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

Download date: 12-11-2019 Tryptophan metabolism in aging and disease Insights from Caenorhabditis elegans

PhD thesis

Helen Michels

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 1 ISBN: 978-94-028-0754-7

Comic design: Sebastian Busch Layout: Midas Mentink, PersoonlijkProefschrift.nl Printing: Ipskamp Drukkers, Enschede, the Netherlands

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 2 Tryptophan metabolism in aging and disease

Insights from Caenorhabditis elegans

PhD thesis

to obtain the degree of PhD at the University of Groningen on the authority of the Rector Magnificus Prof. E. Sterken and in accordance with the decision by the College of Deans.

This thesis will be defended in public on

Wednesday, 27 September 2017 at 12:45

Helen Michels

born on 30 June 1985 in Münster, Germany

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 3 Supervisors Prof. E.A.A. Nollen Prof. I.P. Kema

Assessment Committee Prof. R.P.H. Bischoff Prof. B.M. Bakker Prof. R. Korswagen

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 4 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 5 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 6 TABLE OF CONTENTS Overview Outline of this thesis 10

Chapter I: Introduction 14

Chapter II: Genetic screens in Caenorhabditis elegans models for 56 neurodegenerative diseases

Chapter III: Identification of an evolutionary conserved structural loop 82 that is required for the enzymatic and biological function of tryptophan 2,3-dioxygenase

Chapter IV: Tryptophan 2,3-dioxygenase regulates health- and lifespan 100 independently of kynurenine, serotonin and tryptamine signaling in C. elegans

Chapter V: Tryptophan 2,3-dioxygenase regulates motility and lifespan 118 independently of protein translation in C. elegans

Chapter VI: Transcriptome profiling inC. elegans reveals that TDO and 132 other amino acid dioxygenases regulate mitochondrial capacity and healthspan

Chapter VII: A screening-based platform for the assessment of cellular 164 respiration in Caenorhabditis elegans

Chapter VIII: Increase in number and complexity of modified tryptophan 206 residues during aging in C. elegans

Chapter IX: Discussion and perspectives 222

Summary 244 Samenvatting 248 List of publications 252 Appendix 256 Aknowledgement 266

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 7 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 8 Overview

Outline of this thesis

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 9 Chapter I starts by introducing the different hallmarks of aging. It describes how they might contribute to the aging process and the pathology of age-related diseases. The chapter focuses especially on the theories of aging that are based on loss in proteostasis, mitochondrial dysfunction and ROS production, and changes in nutrient sensing and processing. The chapter continues with the impact of amino acids on cellular metabolism emphasizing findings on the tryptophan 2,3-dioxygenases (TDO, TDO-2 in C. elegans), tryptophan metabolism and its metabolites. Finally, it lists open questions on the mechanisms by which depletion of tdo-2 leads to a health- and lifespan extension that led to the aim of this study.

Chapter II explains the power of using C. elegans for high-throughput screens in order to discover routes that have implications in neurodegenerative diseases. It compares the two most frequently used screening methods - EMS mutagenesis and RNA interference – by describing the advantages and disadvantages for each method. We finish by summarizing findings in the field of neurodegeneration that are based on genetic screens in C. elegans models for Parkinson’s, Alzheimer’s and polyglutamine diseases.

Chapter III gives insight in the molecule structure of TDO-2 by comparing different C. elegans deletion mutants that were generated by the CRISPR/Cas9 system. It shows that 3 amino acids situated in the loop between two alpha- helices, predicted to bind haem, are essential for the enzyme function.

Tryptophan is a precursor for a range of metabolites including serotonin, tryptamine and the downstream products of the kynurenine pathway called kynurenines. Chapter IV investigates the possibility that an increase in the amount of free tryptophan and changes in the metabolite levels might be responsible for the health- and lifespan increasing effects of tdo-2 depletion. The results indicate that the two phenotypes are independent of tryptophan metabolites.

Tryptophan is the least abundant essential amino acid. Chapter V deals with the possibility that an increase in free tryptophan levels might influence translational rates within the cell and that this might be the underlying mechanism for the health- and lifespan benefits intdo-2 depleted animals. The results show that the function of TDO-2 in regulating health- and lifespan in C. elegans is independent of changes in cytosolic and mitochondrial translation.

Chapter VI further explores the cellular and transcriptomic changes in tdo-2 depleted animals and reveals that TDO-2 is involved in the regulation of mitochondrial activity. It further finds that two other dioxygenases, HPD-1 and CDO-1, are sharing the same phenotypes when depleted and that the regulation of the mitochondrial respiration

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 10 might be via similar mechanisms. It suggests that this is only true for enzymes that combine the dioxygenase activity with amino acid degradation and therefore O proposes a general role of amino acid dioxygenases in control of mitochondrial activity, health- and lifespan in C. elegans.

Chapter VII presents the establishment of a protocol to measure C. elegans’ respiration with the help of the Seahorse XF cycler that was initially constructed for cell-based assays. It shows optimizations of every single step including the use of different drugs. The chapter gives a guideline how to implement this method to answer whether a treatment or genetic modification might influence general but also mitochondrial specific respiration.

Chapter VIII switches the focus from free tryptophan to single tryptophan residues within proteins. Preliminary results show that the amount and variety of modification on tryptophan residues in a polypeptide increases with age. It suggests a range of assays to test whether and how depletion of tdo-2 might change the number and composition of these modifications and whether this might have implication on the aging process.

Chapter IX summarizes the findings of this thesis and discusses the open question and next steps of research on how TDO-2 influences cellular processes especially on the level of mitochondria to improve the animals’ fitness and to extend its lifespan. It describes the resemblance of amino acid dioxygenases and a possible new view on aging. It concentrates on the possibilities how the findings might help to understand the molecular basis of aging and age-related diseases.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 11 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 12 Chapter I

Introduction

Helen Michels1

1European Research Institute for the Biology of Aging, University of Groningen, University Medical Centre Groningen, Laboratory of Molecular Neurobiology of Aging, The Netherlands

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 13 The increasing age of the world’s population is accompanied by increasing numbers of individuals suffering from age-related pathologies such as neurodegenerative diseases, cancer and type 2 diabetes. This development demands increasing efforts from society, either financially or in terms of more specialized or prolonged care for the elderly. In turn, this raises questions such as how do age-related diseases develop? How do we age in general? And how can we modulate age-related processes to prolong the lifespan in such a health-promoting way that it is worth living longer and that less care is needed?

This chapter summarizes the findings of aging research over the past decades by dividing them into nine different hallmarks of aging. It also emphasizes findings on the three hallmarks that are relevant to this thesis: age-related loss of proteostasis, dysfunction of mitochondria, and changes in nutrient sensing and processing.

The chapter then continues by describing the effects of amino acids on the aging process. It focuses on tryptophan, an essential amino acid that controls translational rates and is a precursor for bioactive amines that are involved in intracellular signaling and neurotransmission. These include serotonin and the downstream products of the kynurenine pathway such as the neuroactive molecules kynurenic acid, quinolinic acid and kynurenine. The end of the chapter focuses on one central enzyme of tryptophan metabolism: tryptophan 2,3-dioxygenase (TDO in humans, TDO-2 in the nematode Caenorhabditis elegans). TDO is the first enzyme of the kynurenine pathway and its activity strongly influences free tryptophan levels and the formation of the above-mentioned tryptophan metabolites. The findings presented in this thesis help to elucidate how TDO and tryptophan metabolites are involved in healthy aging and tries to understand the molecular processes that are involved in the TDO- dependent control of health- and lifespan.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 14 The increasing age of the world population is accompanied by an increasing incidence in age-related pathologies that include diabetes type 2, cancer and neurodegenerative diseases such as Parkinson’s (PD), Alzheimer’s (AD) and Huntington’s (HD). The increasing burden for society has stimulated research that seeks to understand the 1 processes that are involved in the aging of an organism, the changes that cause age-related diseases and finally, how lifespan be prolonged in such a way that life at advanced ages is still enjoyable. Below I summarize the key aspects of aging research, categorized into nine hallmarks of aging. The emphasis is on protein homeostasis, mitochondrial regulation and nutrient balance, three hallmarks that are directly linked to tryptophan metabolism and the findings of this thesis (Sections 3,4,5).

1. Aging – a post-reproductive phenomenon Aging can be defined as a progressive loss of physiological integrity, leading to impaired function and increased vulnerability to death (López-Otin et al., 2013).

Humans form an exception when it comes to aging. In a natural environment, animals face a high risk of death due to external causes such as predators, competition, constantly changing pathogens and extreme changes in the environment. Therefore, in such an environment animals have a poor chance of actually reaching an age at which age-related diseases develop (Blagosklonny, 2006). However, humans – as well as domesticated and laboratory animals – have a low extrinsic risk of dying. We usually die of senescence-associated diseases such as cancer, stroke, myocardial infarction, heart failure, circulatory disorders, infections and the complications of diabetes (Blagosklonny, 2006). Thus, in a biological context, aging is a rather unusual phenomenon.

In addition, the post-reproductive phase of human lifespan is not of interest in the context of evolution and, therefore, few control mechanisms have been optimized during the evolutionary process (Parsons, 2007; Blagosklonny, 2006; Kirkwood, 2005). One theory of aging developed over the past few years – the theory of antagonistic pleiotropy – therefore postulates that the parts of our metabolism that regulate development and reproduction are not reduced during aging, thereby possibly causing cellular exhaustion, unbalanced protein homeostasis, cell damage and the development of diseases (first mentioned already by Williams, 1957). A lifespan screen in C. elegans has supported this hypothesis, by showing that genes that are essential for development and that are highly evolutionary conserved have the potential to increase longevity when downregulated post-developmentally (Curran and Ruvkun, 2007). Aging is therefore a stochastic and passive process (Charmpilas et al., 2015).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 15 C. elegans is a powerful model organism when it comes to studies of aging and the identification of aging-related genes. The first long-lived strains were isolated more than 30 years ago. Mapping of the mutations led to identification of the age- 1 gene, which encodes a member of the insulin-like signaling pathway (Klass, 1983; Friedman and Johnson, 1988). Several systematic screens for longevity genes followed (Lee et al., 2003; Hamilton et al., 2005; Curran and Ruvkun, 2007). Yanos and colleagues summarized the findings of longevity screens in C. elegans and the dependency of the aging process on external factors such as food availability and temperature. They also showed the strong connectivity between the different pathways identified in these screens, thereby emphasizing that aging is a process that takes place at multiple molecular levels (Yanos et al., 2012).

2. Hallmarks of aging – an explanation of how we age? Carlos López-Otin and colleagues have clearly categorized the different areas of aging-focused research into nine major aspects that they call “hallmarks of aging”. A hallmark of aging is defined by López-Otin et al. as a factor or process that manifests during normal aging whose experimental aggravation accelerates aging but whose experimental amelioration retards the aging process, thus prolonging healthy lifespan (López-Otin et al., 2013). The authors mention the following nine hallmarks: genomic instability, telomere attrition, epigenetic alterations, loss of proteostasis, deregulated nutrient sensing, mitochondrial dysfunction, cellular senescence, stem cell exhaustion and altered intercellular communication. The hallmarks of aging are highly entangled and it is not always possible to place findings in only one of the categories. That is why the authors proposed a further three-way classification that distinguishes between primary hallmarks, which lead to damaging events that accumulate over time; antagonistic hallmarks, which are in theory beneficial but can change to negative events when unbalanced; and integrative hallmarks, which arise when primary and antagonistic processes can no longer be compensated for by tissue homeostatic mechanisms (See Figure 1; López-Otin et al., 2013). Below, I will give a summary of all hallmarks of aging and I will then discuss the hallmarks “loss of proteostasis” (primary hallmark, see Figure 1-7), “mitochondrial dysfunction” (antagonistic hallmark, see Figure 1-8) and “deregulated nutrient sensing” (antagonistic hallmark, see Figure 1-9) in more detail.

2.1 Genomic instability Aging can be seen as an accumulation of genetic damage throughout life (Moskalev et al., 2012, primary hallmark, see Figure 1-1). The challenge for the organism is to constantly protect DNA from exogenous physical, chemical and biological agents as well as from endogenous aspects including DNA replication errors, spontaneous

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 16 primary hallmarks

1. Genomic 2. Telomere 3. Epigenetic 7. Loss of instability attrition alteration proteostasis 1

antagonistic hallmarks

4. Cellular 8. Mitochondrial 9. Deregulated senescence dysfunction nutrient-sensing

integrative hallmarks

5. Stem cell 6. Altered cellular exhaustion communication

Figure 1: The nine hallmarks of aging. Primary hallmarks (1-3; 7) of aging are defined as damaging events that accumulate over time and lead to accelerated aging. Antagonistic hallmarks of aging (4; 8-9) are in theory beneficial as they represent attempts to rebalance molecular homeostasis. They can however reduce lifespan when uncontrolled, unbalanced or dysfunctional. Integrative hallmarks of aging (5-6) arise when primary events can no longer be compensated for by antagonistic processes; they can result in a reduced lifespan. Adopted from López-Otin et al., 2013

hydrolytic reactions and reactive oxygen species (ROS) (Hoeijmakers, 2009). Failure to protect DNA leads to genomic instability. Genetic lesions in nuclear as well as in mitochondrial DNA can have a major impact on cellular homeostasis and can accelerate aging and/or result in all kind of diseases (Gregg et al., 2012; Park and Larsson, 2012).

2.2 Telomere attrition One of the earliest theories of aging – known as the Hayflick limit – describes the limitation in the numbers of cell cycles based on the shortening of telomeres (Hayflick and Moorhead, 1961, primary hallmark, see Figure 1-2). The Hayflick limit is however not universal: many cells do not proliferate regularly and those that do so frequently have active telomerases that prevent telomere shortening (Tang et al., 2001; Parrinello et al., 2003). Telomere shortening has however been observed during aging in most studied organisms including humans and mice (Blasco, 2007). DNA damage at telomeres is associated with the development of a range of diseases that are linked to a loss of regenerative capacity of different tissues (summarized in Armanios and Blackburn, 2012).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 17 2.3 Epigenetic alterations Age-related alterations in DNA methylation patterns, in post-translational modifications of histones, and in chromatin remodeling are collectively called epigenetic alterations (Talens et al., 2012, primary hallmark, see Figure 1-3). Such alterations can involve multiple enzyme families such as DNA methyltransferases, histone acetylases, deacetylases (including sirtuins), methylases and demethylases, and also protein complexes of chromatin remodeling (summarized in López- Otin et al., 2013). Epigenetic alterations lead to changes in the transcriptome which consequently alters cellular homeostasis, thereby promoting aging and the development of diseases (Harries et al., 2011; Nicholas et al., 2010). It is important to mention the existence of transgenerational epigenetic memory, which also influences cellular homeostasis and longevity in subsequent generations. This epigenetic inheritance has been demonstrated for example in C. elegans, where starvation- induced longevity in the parental generation was able to increase the lifespan in up to three normally fed follow-up generations (see Section 5, Greer et al., 2011).

2.4 Cellular senescence Cellular senescence (antagonistic hallmark, see Figure 1-4) can be seen as a program for development that has not been turned off. In other words, the combination of excessive individual cell growth and cell-cycle arrest leads to cell senescence. A senescent cell is therefore a cell that has large morphology and a high metabolic activity, that secretes mitogens and is resistant to apoptosis (Blagosklonny, 2006).

2.5 Stem cell exhaustion Aging is also characterized by a decline in the regenerative potential of tissues (López- Otin et al., 2013, integrative hallmark, see Figure 1-5). For example, hematopoiesis is reduced during aging leading to diminished production of adaptive immune cells (Shaw et al., 2010) which are crucial for maintaining the health and function of an organism.

2.6 Altered cellular communication Aging also occurs at the organismal level. Intercellular communication (e.g. endocrine, neuroendocrine or neuronal) is crucial for cells to send each other signals about their metabolic state and to control the immune surveillance machinery of an organism. Age-dependent alterations in intercellular communication therefore crucial and can have a strong impact on the survival of the animal. (Laplante and Sabatini, 2012, integrative hallmark, see Figure 1-6). The term “inflammaging” comprises the changes in inflammation during aging. Inflammaging has multiple causes, including

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 18 the accumulation of inflammation-induced tissue damage, the failure of the immune system to clear pathogens and dysfunctional host cells, the secretion of pro- inflammatory cytokines by senescent cells, and defects in the autophagic response, the latter also in neighboring cells (Salminen et al., 2012). Alterations in the control 1 of inflammation are associated with various age-related diseases, including type 2 diabetes (Barzilai et al., 2012) and atherosclerosis (Tabas, 2010).

3. Protein homeostasis – loss of fine-tuning during aging Protein homeostasis, also called proteostasis, is defined as the homeostatic mechanisms that maintain the conformation, concentration, interactions, localization and, hence, function of proteins (see Figure 2; Kim and Strange, 2013). The proteostasis network is highly conserved throughout evolution and coordinates the integrated activities of gene transcription and RNA metabolism as well as protein synthesis, folding, assembly, trafficking, disassembly, repair and degradation (summarized in Balch et al., 2008).

B mRNA A J

translation insoluble chaperone- aggregate small dependent non-coding folding mRNA RNA chaperone- C independent folding soluble misfolded/ damaged protein chaperones e.g. HSPs transcriptional control folded protein E D autophagosome

refolding F I

G poly- protein ubiquination transport H

complex degradation formation via autophagy protein degradation interactions via proteasome

Figure 2: Overview of protein homeostasis Protein homeostasis (proteostasis) includes all homeostatic mechanisms that maintain the conformation,concentration, interactions, localization and thus the function of proteins. This includes the transcription of mRNA, which is regulated by diverse proteins of the transcriptional machinery and small non-coding RNAs (A). The translation into a polypeptide occurs at the ribosome (B) after which it is further folded (C) into a functional protein. Misfolded proteins can be refolded (D) and then sent to their final location in the cell (E), where they might be part of a protein complex (F) or interact with diverse cellular components (G). A damaged or un-refoldable protein will be degraded/recycled via the proteasome (H) or autophagosomal/lysosomal pathways (I). Large amounts of misfolded proteins can also be enclosed in insoluble, compact protein aggregates (J).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 19 The ability to detect and stabilize cellular and molecular protein damage – especially damage induced by environmental stress – is crucial for cell function and survival (Lamitina et al., 2006). For example, the protein damage that can occur during aging, for example modifications caused by ROS, can affect cargo recognition by transport proteins and interfere with their efficiency (Vanhooren et al., 2015, see Figure 2-E). Key components that respond to protein damage and stress are molecular chaperones (including the heat-shock family of proteins) and proteins of the proteasome and lysosome (Hartl et al., 2011). The coordination of all these processes is important for maintaining cellular protein homeostasis (Hartl et al., 2011).

Protein homeostasis is maintained by several signaling pathways as well as by ATP levels, amino acid pools, metabolites, lipids and ion balance (summarized in Powers et al., 2009). For example, it has been suggested – but not demonstrated – that the decrease in ATP during aging might be responsible for the lower abundance of ATP- dependent chaperones (Brehme et al., 2014).

Proteostasis can also be communicated in a cell non-autonomous manner, whereby the proteostatic balance is regulated in a network of multiple cells (Murphy and Bloom, 2006; Prahlad et al., 2008). Components of the proteosatsis network can even travel from one cell to another (summarized in Kaushik and Cuervo, 2015). This exchange of components may well involve exosomes functioning as vehicles (Lo Cicero et al., 2015) and the formation of nanotubes between cells (Astanina et al., 2015; Wang and Gerdes, 2015). The nature of material exchange has gained attention in recent years as several aggregation-prone proteins involved in neurodegenerative diseases have been detected in exosomes, and it has been suggested that some of these disease proteins are able to spread from cell to cell in a prion-like manner (Russo et al., 2012).

Furthermore, the proteostasis network has to make the grade that arises by various demands of different tissues (Su et al., 2004). The differences in composition might explain why some tissues are more susceptible to protein misfolding diseases. In this context, it has been found that the activity of chaperone-mediated autophagy that is involved in the removal of the PD-related protein alpha-synuclein is lower in aggregation-prone brain regions in mice (Malkus and Ischiropoulos, 2012).

Loss-of-protein function or gain-of-toxic function interfere with the cell’s function and can facilitate the development of diseases such as cancer, diabetes type 2 or neurodegeneration (Powers et al., 2009). Alterations in the composition and concentration of the proteostasis network as well as accumulation of intracellular damage are also observed during aging (summarized in Koga et al., 2011; Kaushik and Cuervo, 2015). For example, it has been shown that the expression of ATP- dependent chaperones and their induction are reduced in aged brains. It has been

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 20 suggested that this could be a consequence of reduced ATP availability (Brehme et al., 2014). Attempts to improve proteostasis during aging have been made in model organisms. In this respect, while the production, folding, transport and degradation systems have been extensively studied, potential improvements in protein refolding 1 have not been widely tested and may well form an alternative strategy (Kaushik and Cuervo, 2015). The following paragraphs summarize how the different components of the proteostasis network (see also Figure 2) change with age and could be involved in the development of neurodegenerative diseases.

3.1 Loss of translational control In 1963, Orgel postulated the error catastrophe theory of aging. This theory describes an age-related decline in accuracy during translation which might be a source of aging-associated degenerative changes (Orgel, 1963, see Figures 2-A and B). In many tissues, the rates of protein synthesis and turnover significantly decline during aging (Dwyer et al., 1980; Rattan, 1996; Tavernarakis, 2008). The protein synthesis process is highly surveyed and controlled for quality both at the pre-translational and translational level. The consequence is that less than 2% of the transcribed genome is finally translated into a functional protein (Kapranov et al., 2007). Alterations in this quality control process have been also reported for aging (Vargas and Castaneda, 1983; Kimball et al., 1992; Webster and Webster, 1983; summarized in Tavernarakis, 2007).

3.2 Dysfunctional protein processing For proteins in eukaryotic cells, proper folding and function has to take place under various physiological conditions and in crowded environments – partly as a result of cellular compartmentalization (e.g. endoplasmic reticulum, nucleus and mitochondria). A wide range of macromolecules have evolved to assist with protein folding, assembly, disassembly and protein transport (see Figures 2-C to G). They comprise more than 1000 chaperones (general and specialized), folding enzymes and degradation components, as well as trafficking subunits (summarized in Powers et al., 2009).

Much research into such chaperones and their link with aging has been done in model animals. The overexpression of chaperones extends lifespan in flies and worms (Morrow et al., 2004; Walker and Lithgow, 2003) and long-lived mouse strains show an upregulation of heat-shock proteins (Swindell et al., 2009). The master regulator of the heat-shock response is the transcription factor HSF-1 and its activation increases stress resistance and extends lifespan in nematodes (Hsu et al., 2003).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 21 The speed and efficiency of protein folding in a cell is affected by many factors, including the rate of translation (see Figure 2-B). For example, during stresses, proteostasis mechanisms are upregulated (e.g. induction of the unfolded protein response) to maintain proteome solubility and functionality (Kaushik and Cuervo, 2015). By inhibiting the protein synthesis that occurs in response to diverse environmental stressors, the cell can limit stress-induced protein damage. Inhibition of translation reduces energy usage and, of course, the number of proteins that might get damaged during a stress period (Holcik and Sonnenberg, 2005). Kim and Strange showed that reducing translational rates by inhibition of translational components using RNA interference (RNAi) reduced the formation and growth of protein aggregates, including those of the neurodegenerative disease- associated proteins Huntingtin and alpha-synuclein (Kim and Strange, 2013). The stress-induced reduction in translation was mediated by the activation of a kinase called GCN- 2 (general control nonderepressible-2 kinase), which inhibits general translation and enables the transcription of specific stress-response genes by activation of the transcription factor GCN-4 (Kim and Strange, 2013).

3.3 Decrease in protein recycling and degradation Besides production and quality control of proteins by ribosome and chaperones, proteostasis is also maintained by the removal of damaged cellular proteins and whole dysfunctional organelles (see Figure 2). It has also been postulated that aging is triggered by increasing levels of damaged proteins (e.g. through ROS damage) and that the proteostasis network is overwhelmed by their abundance (summarized in Powers et al., 2009). In this context, it should be mentioned that many studies on protein damage and damaging agents involve in vitro experiments or include technical procedures that might cause damage to the protein of interest (e.g. protein damage due to reactive nitrogen species (RNS), summarized in Nuriel et al., 2011). Therefore, those results must be interpreted with care. Nevertheless, research has shown that aging is accompanied not only by an accumulation of damaged proteins (Smith et al., 1991) but also by a reduction in the function of degradation systems (Lipinski et al., 2010; Gavilán et al., 2012).

Proteins are usually degraded via the ubiquitin proteasome system or via autophagosomal/lysosomal mechanisms (Hartl et al., 2011, see Figures 2-H and I). The two systems communicate with each other and can compensate for one another (Park and Cuervo, 2013) and both show an activity decline during aging (Lipinski et al., 2010; Gavilán et al., 2012). In addition, the ability to compensate for the other’s low efficiency seems to be lost during aging, as demonstrated in aged mice (Gavilán et al., 2015; Schneider et al., 2015).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 22 In terms of the ubiquitin proteasome system, overexpression of several of its components has been shown to extend lifespan in nematodes, yeast and flies (Liu et al., 2011; Chondrogianni et al., 2015; Madeo et al., 2015). When Walther and colleagues studied proteome composition during aging in C. elegans, they found 1 that the abundance of ribosomal proteins decreases while levels of proteasome complexes increase (Walther et al., 2015). This increase in complexes is proposed to be a compensating effect to counteract age-dependent activity decline (Rubinsztein et al., 2011).

In terms of autophagosomal mechanisms, induction of autophagy has resulted in life- and healthspan-extending effects in multiple organisms, including yeast, nematodes, flies and mice (Harrison et al., 2009; Madeo et al., 2010; summarized in Rubinsztein et al., 2011). However, elevating the rate of degradation could have opposing effects, changing not only the abundance of toxic species but also the level of folded, functional proteins, which in turn might have severe consequences for the cell’s health (Powers et al., 2009). Pharmaceutical attempts to restabilize protein homeostasis during aging and in disease, including the use of pharmacologic chaperones, have been diverse and are summarized in Powers et al. (2009). The first results are promising and such compounds may well facilitate the folding of specific proteins. For example, Loo and colleagues have shown that the application of pharmacologic chaperones improves and corrects the folding of the cystic fibrosis- related protein CFTR (Loo and Clarke, 2011).

3.4 Aggregation – the last attempt of proteostasis? Loss of proteostasis can eventually result in the accumulation and aggregation of overrepresented, misfolded or damaged proteins (see Figure 2-J). Even small changes in protein homeostasis might lead to aggregation. Thus, aging highly affects this solubility of proteins. It has been shown that the aggregation process can have a snowball effect, whereby single molecules act as a seed and then sequester more and more proteins into an aggregate (Chiti and Dobson, 2009).

Aggregation of proteins does not necessarily only occur under disease conditions. Different stressors can induce protein aggregation in the cell as well as protein accumulation. Aggregation might primarily be a protection mechanism for the cell. Aggregates encapsulate stress-inducing components such as damaged or misfolded proteins so that they can no longer interact with other cellular structures. This is especially important when the capacity of protective proteostasis mechanisms (such as repair and degradation) is reached (Powers et al., 2009; Walther et al., 2015; David, 2012). In this context, the application of aggregation inhibitors might worsen the cellular state, as they might increase the levels of smaller, toxic species (Power

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 23 et al., 2009).

Regulating functions for aggregation have also been found in yeast, where 25 proteins that are involved in gene expression regulation and signal transduction can aggregate to become inactivated in response to environmental changes (Halfmann et al., 2012). In mammals, aggregation has also been identified as a way of storing peptide hormones to make them quickly accessible when needed (Maji et al., 2009).

Several groups have tried to identify the “aggregating proteome”. In general, all proteins seem to have the capacity to aggregate depending on pH, temperature, denaturation state or concentration (David, 2012). However, some proteins are more prone to aggregation. This group is predicted to be significantly enriched in beta- sheets (David et al., 2010; Reis-Rodrigues et al., 2012). The tendency of the proteome to aggregate also increases in older organisms (Walther et al., 2015). Experiments in a number of different organisms have demonstrated that proteins known to regulate lifespan are overrepresented in the pool of aggregating proteins. Reducing their levels extends lifespan, e.g. in C. elegans, suggesting that the aggregation process might be involved in lifespan determination (David, 2012). A screen in human cells, which looked at the aggregate content after inducing aggregation by overexpressing artificial beta-sheet proteins, revealed that co-sequestered proteins are large in size and enriched in unstructured regions. This group of proteins contained a large overrepresentation of key players of the cellular protein network, having functions in chromatin organization, transcription, translation, maintenance of cell architecture, and protein quality control (Olzscha et al., 2011).

3.5 Loss of proteostasis in neurodegenerative diseases Neurodegenerative diseases such as AD, HD and PD are accompanied by the formation of aggregates (Weidner et al., 2011). Depending on the disease these aggregates are composed of different types of proteins and occur at different locations of the brain (summarized in Lane et al., 2015). It is still under debate whether the aggregates are toxic or whether the aggregation process is a protective mechanism that isolates toxic species.

For AD, two types of aggregates have been described: intracellular tangles, which are composed of hyperphosphorylated microtubule-associated protein Tau; and insoluble extracellular plaques, which in contrast contain amyloid-beta, a cleavage product of the amyloid precursor protein (APP) (summarized in Mairet-Coello et al., 2013). Familial cases of AD are associated with mutations in three different proteins: APP, which is cleaved by beta and gamma-secretase to produce amyloid- beta, and presenilin 1 and 2 (PS1 and PS2), components of the gamma-secretase complex (summarized in Lin and Beal, 2006). The expression levels of these three

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 24 proteins can influence the proteostasis network. For example, it has been shown that constitutive expression of amyloid-beta leads to a reduction in ribosomal activity (Ben-Zvi et al., 2009; Kirstein-Miles et al., 2013).

For HD, a proteomic study has shown that expanded Huntingtin, the protein that 1 is responsible for toxicity, preferentially interacts with intrinsically disordered proteins and proteins related to energy metabolism, protein trafficking, RNA posttranscriptional modifications and cell death (Ratovitski et al., 2012). These interactions might alter protein homeostasis which might explain part of the toxicity in HD. For example, as seen for amyloid-beta, the expression of poly-glutamine (Huntingtin) has also been shown to result in reduced ribosomal activity (Ben-Zvi et al., 2009; Kirstein-Miles et al., 2013).

For PD, links have been found with mutations in several different genes including a mutation in parkin, a E3 ubiquitin ligase (Winklhofer, 2007). This mutation changes the activity of the ubiquitin proteasome system and therefore destabilizes protein homeostasis (Lim and Tan, 2007). A second PD-related protein is alpha-synuclein. This protein is intrinsically disordered but adopts a highly helical structure when binding to lipid surfaces (Davidson et al., 1998; Fusco et al., 2014). It has been suggested that it interferes with membrane structure and might be involved in membrane-associated processes such as neurotransmitter release at the synapse (Clayton and George, 1998; Nemani et al., 2010). On the other hand, an imbalance in lipid/protein ratio as a consequence of alterations in proteostasis has been described to induce the formation of aggregates containing not only alpha-synuclein but also lipid molecules (Galvagnion et al., 2015).

4. An imbalance in mitochondrial function during aging Mitochondria are important for ATP production and are involved in cell signaling, autophagy, metabolite production, branched amino acid catabolism, iron and calcium homeostasis, lipid balance, steroid production and cell death (Figure 3-A, summarized in Demine et al., 2014). The mitochondrion therefore appears to be a central organelle of the cell that is strongly connected to all kinds of cellular processes (Perier and Vila, 2012). Mitochondria also occupy a large part of the volume of a cell (2-20%, depending on species and tissue), which underlines their dominant role (Brand, 2014). Alterations in mitochondrial function are observed during aging and in a number of age-related pathologies (Lane et al., 2015): cancer, neurodegenerative diseases, type 2 diabetes and other metabolic disorders have all been linked to defects in mitochondrial homeostasis (summarized in Wallace, 2005).

Mitochondrial dysfunction can be caused by various alterations. These alterations include mitochondrial uncoupling, mitochondrial depolarization, destabilization

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 25 energy balance I iron homeostasis Ca2+ homeostasis mitophagy A F lipid balance fusion steroid production

control of apoptosis

fission ATP G membrane potential H mt protein mtDNA mtUPR nc protein D mitochondrial E transport ROS GCN-2 ROS damage super- hypoxia complexes B mitohormesis

glucose

amino acids C fatty acids

NADH

Figure 3: Overview of mitochondrial metabolism Mitochondria are the major energy source of the cell, but also control iron, calcium and lipid homeostasis as well as the production of steroids and apoptotic processes (A). Mitochondria are also a source of ROS, small molecules that are highly reactive and can therefore induce damage but that also signal stress cascades such as hypoxia, which can lead to mitohormetic effects (B). Major substrates are NADH, glucose, amino acids and fatty acids (C). Mitochondrial proteins are nuclear (nc)- encoded and mitochondrial (mt)- encoded. A mitochondrion can contain several copies of the mitochondrial genome (D). An imbalance between the nuclear- and mitochondrial-encoded proteins induces the mt unfolded protein response (UPR) to restore the protein balance (E). Individual mitochondria can fuse to multi-mitochondrial networks, thereby enabling the exchange of DNA and metabolites and improving mitochondrial function (F). On the other hand, mitochondrial fission (G) is necessary for long distance transport e.g. through axons (H), but also to separate individual, non-functional mitochondria for degradation via mitophagy (I).

of mitochondrial supercomplexes, inhibition of the oxidative phosphorylation, fragmentation of the mitochondrial network, mutations in nuclear or mitochondrial- encoded mitochondrial-related genes, and the accumulation of protein aggregates in the mitochondria (summarized in Demine et al., 2014).

A dysfunction of mitochondria is signaled by a number of different mechanisms. These include the production of ROS, changes in membrane potential, changes in levels of acetyl-CoA, NAD and calcium, or changes in the location, number and shape of the mitochondria (summarized in Yun and Finkel, 2014; Chandel, 2015). Mitochondrial stress can be communicated within the cell but also cell non-autonomously. It has been suggested that such stress is signaled by so-called mitokines (Durieux et al., 2011). The existence of such factors, however, has not yet been demonstrated.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 26 Mitochondrial dysfunction is usually measured in terms of the activity of oxidative phosphorylation, ATP production and respiration, as well as through other parameters such as membrane potential, proton leakage and levels of ROS (summarized in Demine et al., 2014). 1

4.1 Reactive oxygen species – killing factor number one or hormetic signaling? Hormesis is defined as a coordinated response to mild stress that appears to make the cell less susceptible to subsequent perturbations (Yun and Finkel, 2014). Mitochondrial- specific hormesis is called mitohormesis (Yun and Finkel, 2014). Evidence from studies in C. elegans suggests that the activation of cytoprotective responses – through stress signals such as induction of chaperones, xenobiotic detoxification or antioxidant defense – is linked to lifespan-extending effects (summarized in Yun and Finkel, 2014). Hormetic reactions can have cell non-autonomous effects as it has been shown that hormesis in one tissue is enough to increase the age of a whole organism (Durieux et al., 2011). Central molecules in the induction of mitohormetic effects are thought to be ROS , as described in more detail below.

Mitochondria are the major source of ROS (Figure 3-B). It has been estimated that 2-4% of oxygen consumed by a human is reduced to superoxide and other ROS (Chance et al, 1979), mainly at complexes I and III of the electron transport chain (Jezek, 2005). For a long time, ROS and RNS have been seen as the main sources of cellular damage. Indeed, as far back as 60 years ago Harman formulated the “ROS theory of aging”, which postulated that age-related cellular damage is caused by free radicals (Harman, 1956). ROS and RNS can react with all three types of macromolecules in the cell: nucleic acids, lipids and proteins (e.g. Ishii et al., 2007; Levine and Stadtman, 2001). These kind of oxidation events such as e.g. carbonylation of proteins usually reduces the function and stability of the specific target (Rebrin et al., 2007). Levine and Stadtman suggested two possibilities that might cause enzymes to be less active during aging: decreased protein turnover resulting in longer-lived proteins, and increased protein oxidation (Levine and Stadtman, 2001). However, one does not necessarily exclude the other: A reduction in protein turnover increases the half-life of proteins. ROS and RNS lead to damage that is promoted in an environment where the damaged protein is not efficiently degraded and replaced, for example due to decreased protein turnover.

Oxidative stress can be a consequence of respiration defects in the mitochondria and can result in misfolding and aggregation of mitochondrial proteins, which might eventually lead to irreversible damage of mitochondria (e.g. Fritz et al., 2011; Kong et al, 2010; summarized in Tatsuta, 2009). The control mechanisms that are involved

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 27 in maintaining the function of mitochondria can remove damaged proteins or even the entire organelle (summarized in Segref et al., 2014). Their reduced efficiency during aging – possibly through ROS-induced damage – might explain the age- dependent functional decline in mitochondrial activity (Morimoto and Cuervo, 2014; Segref et al., 2014).

Mitochondrial ATP production is linked to the life expectancy of an organism. Longevity screens, primarily RNAi and mutagenesis screens in C. elegans, have revealed that a relatively large proportion of the genes that are involved in lifespan regulation are genes of the mitochondrial electron transport chain (ETC) (Lee et al., 2003; Hamilton et al., 2005; Hansen et al., 2005; Curran and Ruvkun, 2007). The lifespan-extending effects seen in these experiments as a result of attenuated mitochondrial function were accompanied by low ATP production, small body size, slow growth rate and reduced reproduction levels, as well as sterility in some cases (summarized in Yanos et al., 2012).

An explanation frequently given for this mitochondrial-dependent lifespan extension has been the reduced production of ROS that often accompanies the decreased activity of the ETC. However, in the past few experiments have been performed to prove this theory. Several research groups have now finally shown the opposite to be true: that an increase in ROS production might even increase lifespan in yeast and C. elegans by inducing mild stress responses (hormesis effect, see above) (Doonan et al., 2008; Van Raamsdonk and Hekimi, 2009; Mesquita et al., 2010). ROS – produced by complex I of the ETC – also mediate lifespan-extending effects in flies (Scialò et al., 2016). Several studies have shown that there is indeed no correlation between ROS levels and the length of life of an organism (Lee et al. 2003; Yee et al., 2014; Van Remmen et al., 2003; Zhang et al., 2009; Perez et al., 2009). It has also been shown that ROS might have compartment-specific effects. Whereas mitochondrial ROS have been shown to increase lifespan, ROS generated in the cytosol appear to decrease lifespan. However, since these studies were based on the activity of mitochondrial and cytosol-specific superoxide dismutases (SODs) and did not measure the levels of ROS under different conditions (Schaar et al., 2015), level-dependent toxicity cannot yet be excluded.

It has also been shown that ROS even have signaling functions in the cell and are therefore involved in the regulation of cellular processes. For example, an imbalance of nuclear and mitochondrial-encoded proteins leads to problems in the assembly of ETC complexes which might result in an inefficient transfer of electrons and lead to the production of ROS (for more details see 4.2.3). ROS molecules then communicate this state to the cytosol where they induce the mitochondrial unfolded protein response (Figure 3-E). GCN2, a cytosolic regulator of translation that responds to the increase in ROS levels, adjusts the translational rates and induces

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 28 the transcription of certain stress-response genes in order to restore the balance between nuclear and mitochondrial-encoded proteins (Baker et al., 2012). Although the signaling cascade of the here described mitochondrial unfolded protein response (mtUPR) has as yet only partially been identified, it is subject of several studies. 1 Another example of the cellular functions of ROS is their communication with some chaperones: some molecules of the chaperone machinery adapt to ROS stress by using protein oxidation to induce structural changes in the chaperone, thereby stabilizing it and improving function (Dahl et al., 2015). Oxidative stress reduces cellular ATP levels, resulting in an environment of low energy and high ROS levels. But, in order to properly fold polypeptides into proteins, chaperones need energy and under conditions of low energy however, there is the risk that polypeptides and aggregation-sensitive folding intermediates might accumulate and finally aggregate. In this situation, energy-independent chaperones (small heat shock proteins (sHsp)), so-called “holdases”, can bind those structures to prevent their aggregation without refolding them. When energy becomes available again, the load is transferred to energy-dependent chaperones that finish the folding to a protein (Voth et al., 2014). In particular, under oxidative stress, a special group of chaperones becomes active: these specific chaperones known as conditionally disordered chaperones (Dahl et al., 2015) have been shown to have a four-cysteine-residue motif that is highly sensitive to oxidation by ROS. This therefore allows for a post-translational redox switch that induces structural changes by modulating disulfide bonds. These changes in quaternary structure switch the chaperones to energy-independent holdases which can remain bound to protein folding intermediates (Voth et al., 2014).

Increased ROS levels are also seen in hypoxia (reduced oxygen availability in the cell). Hypoxia activates the transcription factor known as hypoxia-inducible factor (HIF), which induces the transcription of hypoxia-related stress-response genes. Transient hypoxia can extend the lifespan of C. elegans (Schieber and Chandal, 2014). This special form of lifespan extension is dependent on mtUPR and mTOR (mammalian target of rapamycin) activity and has been shown to be inhibited by antioxidant treatment, suggesting once more a mitohormetic effect signaled by ROS (Schieber and Chandal, 2014).

To summarize, since Harman formulated his theory of aging in 1956, there have been several key developments in our knowledge of how ROS might be involved. Firstly, we now know that ROS are not only toxic molecules that induce global uncontrolled damage, they also function as signaling factors that induce a cellular stress response, possibly also through damage to key proteins. Secondly, and as a direct result, they are involved in the control of cellular functions and can increase lifespan. This has necessitated an update to the ROS-based theory of aging: whereas low amounts of ROS are involved in cellular signaling and promote longevity, excessive amounts of

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 29 ROS lead to increased damage to macromolecules in the cell and might cause cell death.

4.2 ROS-independent mitochondrial dysfunctions As described above, aging can be aggravated by dysfunctional mitochondria independently of ROS (Edgar et al., 2009; Hiona et al., 2010). Such mitochondrial defects include mutations and deletions in mtDNA; aberrant production of mitochondrial proteins; oxidation of mitochondrial proteins; destabilization of respiratory complexes and alterations in the formation of supercomplexes; changes in the lipid composition of the mitochondrial membrane, which alters interorganellar communication; age-dependent variations in fusion and fission events; and defects in quality control by mitophagy (Wang and Klionsky, 2011).

4.2.1 Alterations in metabolites and energy sources Mitochondria use pyruvate, a product of the metabolism of glucose during glycolysis, fatty acids for mitochondrial beta oxidation and glutamine for glutaminolysis, three processes that enable them to produce ATP under aerobic conditions (Figure 3-C, summarized in Demine et al., 2014). The main substrate is NADH, which is formed out of NAD+. In C. elegans, NAD+ levels decline with age and restoring them can extend lifespan (Mouchiroud et al., 2013).

NADH and FADH2 feed the ETC with electrons at complex I or II. Electrons are transported from one complex to the next, thereby freeing energy to generate a proton gradient between the mitochondrial matrix and the inner-mitochondrial membrane space. The complexes of the electron transport chain can form supercomplexes, so- called respirasomes, to better channel the transport of electrons from one complex to another (Schagger and Pfeiffer, 2001). The formation of those supercomplexes is facilitated by other proteins and certain structures of the lipid membrane (Strogolova et al., 2012). Electrons are finally transferred to molecular oxygen in complex IV. Protons of the proton gradient are used by the ATPase (sometimes also called complex V) to produce ATP. This whole process of oxidative phosphorylation is dependent on the availability of oxygen.

However, an organism is also able to generate ATP under anaerobic conditions. Pyruvate, the product of glycolysis, is then converted into lactate by lactate dehydrogenase (summarized in Demine et al., 2014). Even though the resulting amount of energy is higher under aerobic conditions – i.e. through active oxidative phosphorylation – organisms without mitochondria are known to be viable (Williamson, 2002). Furthermore, cancer cells often shift their metabolism towards lactate production to generate ATP (Warburg effect) (Warburg, 1956).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 30 4.2.2 Genetic alterations within the mitochondria Since mitochondria are derived from a symbiotic relationship between eukaryotes and aerobic bacteria, they still contain their own DNA as well as their own transcription and translation machinery (Figure 3-D, summarized in DiMauro and Schon, 2003). 1 Mutations in mitochondrial DNA do not necessarily cause dramatic changes in cellular function as mitochondria contain several copies of their genome. Furthermore, since a cell contains a high number of mitochondria, one mutation is easily compensated for by hundreds of unmutated copies (summarized in Payne and Chinnery, 2015). It has been suggested that in order to have potential effects on mitochondrial function a mutation must reach a threshold, thus, a certain fraction of total mitochondrial DNA has to carry the same mutation. However this threshold might be lower if a tissue is highly dependent on oxidative metabolism, as is the case for the brain, heart and skeletal muscle (Perier and Vila, 2012). Mutations in mitochondrial DNA accumulate during aging and correlate with the decline in mitochondrial function (Corral-Debrinski et al., 1992). It might be that this is a consequence of age-dependent changes in the quality control of mitochondrial transcription and translation, as well as reduced degradation and protein production. These age-related developments lead to a longer median life of mitochondria, and changes in fusion and fission of the mitochondrial network help mutations to spread and accumulate (Perier and Vila, 2012). For example, it has been suggested that in a young healthy cell, fusion events might be needed to dilute mutations and/or fission events might occur to isolate those mitochondria that have a mutated genome (Figure 3-F to I, Payne and Chinnery, 2015). Through fission, isolated single mitochondria as well as smaller multi-mitochondrial complexes can be sent for degradation by mitophagy, a process that recycles impaired mitochondria (Figures 3-G and I, Palikaras et al., 2015).

4.2.3 An imbalance in the coordination of nuclear and mitochondrial translation The mitochondrial unfolded protein response (mtUPR) is the cellular response to an imbalance in the expression of mitochondrial proteins (Figure 3-E, Houtkooper et al., 2010). The mtUPR has several triggers and is highly influenced by the balance between mitochondrial autophagy (mitophagy) and mitochondrial biosynthesis (summarized in Yun and Finkel, 2014).

More than 1000 nuclear genes encode mitochondrial proteins (Vinuela et al., 2010). However, there is a small group of proteins that are encoded by the mitochondrial genome and synthesized by the mitochondrial translational machinery (summarized in Lin and Beal, 2006). This group of proteins (37 in total) is made up of 13 proteins in the electron transport chain (ETC) and 24 translational components (ribosomal

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 31 subunits and mitochondrial-specific tRNAs; summarized in Yun and Finkel, 2014). For mitochondria to function properly it is therefore very important that nuclear and mitochondrial translation are coordinated, thereby enabling the assembly of complexes that contain nuclear as well as mitochondrial-encoded proteins. For example, complex IV of the ETC is composed of 10 nuclear and 3 mitochondrial- encoded proteins. As described above (Section 4.1), it has been shown that this communication between nuclear and mitochondrial translation is at least partially mediated by ROS (Baker et al., 2012) but also by specialized signaling molecules such ATFS-1 (activating transcription factor associated with stress-1) (Haynes et al., 2010). The induction of mtUPR has been demonstrated to increase lifespan (Hamilton et al., 2005). This can apparently also be achieved by cell non-autonomous signals as the activation of mtUPR in the intestine has been shown to be sufficient to increase organismal lifespan (Durieux et al., 2011).

4.2.4 Dysfunctional inter-mitochondrial communication, transport and recycling Mitochondrial function and quality control are maintained by the dynamics of the mitochondrial network, i.e. fusion and fission events. These processes ensure adequate mitochondrial function at a specific time and location (Perier and Vila, 2012). For example, fusion enables the exchange of mitochondrial DNA and proteins with neighboring mitochondria and can restore the functionality of a single mitochondrion (Figure 3-F, Detmer and Chan, 2007; Schon and Przedborski, 2011). Fission is important for long distance transport of mitochondria, e.g. along axons (Figure 3-H, Brown et al., 2006). Defects in fission can therefore have consequences for the rate of neurotransmission: energy and calcium homeostasis is needed at the synapse, both processes of the mitochondria that can be altered if mitochondria are not properly transported (Verstreken et al., 2005). Mitochondrial fission can also be induced by a loss in membrane potential, which subsequently leads to the removal of single mitochondria by mitophagy (Figure 3-I, Veatch et al., 2009). The number of single, thus fragmented mitochondria increases with aging, either to compensate for reduced ATP production or as a result of impaired removal of less active mitochondria (Payne and Chinnery, 2015). In neurodegenerative diseases, mitochondria have been shown to have more fragmentation/fission than their counterparts in healthy cells. These higher levels of fission might have effects on inter-mitochondrial communication. Or they could be a consequence of higher mitochondrial damage and dysfunction, thus of increased activation of mitophagy (Otera et al., 2013).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 32 4.3 Mitochondrial dysfunction in neurodegenerative diseases The increasing levels of dysfunctional mitochondria with increasing age – due to alterations in the structure, dynamics, gene expression and metabolism of mitochondria – may well aid the development of age-related diseases (Gomes et 1 al., 2013). Indeed, mitochondrial dysfunction has been demonstrated to already occur early in the pathogenesis of neurodegenerative disorders and might be a cause for the toxicity observed in those diseases (summarized in Lin and Beal, 2006). Neurons are vulnerable to mitochondrial impairment as they are highly dependent on energy, especially for long distance cargo transport and synapse activity (Perier and Vila, 2012). Oxidative stress, possibly induced by dysfunctional mitochondria, is also associated with many age-related diseases, including neurodegeneration in the pathology of AD, PD and HD (summarized in Lin and Beal, 2006).

Besides having general age-related effects, for example on energy balance which affects organismal health, many disease-related proteins have been shown to be directly associated with the mitochondria and to directly interact with them (summarized in Lin and Beal, 2006). Although this interaction might be part of their normal function, changes in the interaction might be involved in the pathology of specific diseases (e.g. Lutz et al., 2009).

Evidence for links between the disease-related proteins seen in AD, PD and HD and the mitochondria is accumulating, as discussed below.

First, APP, implicated in AD, has an endoplasmic reticulum/mitochondrial targeting sequence. Overexpression of APP in mice inhibits the mitochondrial protein import machinery which is important for mitochondrial function and ATP production (Anandatheerthavarada et al., 2003). Furthermore, amyloid-beta, presenilins and other components of the gamma-secretase complex have been localized at the mitochondria (Lustbader et al., 2004; Hansson et al., 2004).

Second, the expansion mutation in the Huntingtin protein (high number of glutamine repeats) seems to increase the interaction with mitochondrial proteins involved in oxidative phosphorylation (subunits of the ETC and ATPase), which might cause the mitochondrial dysfunction that is observed in HD pathology (Ratovitski et al., 2012). In accordance with this, a decline in activity of complex II and complex III of the ETC has been found in human brains of HD patients (Gu et al., 1996).

Third, in PD, nine nuclear genes have so far been identified as either being mutated or known to increase the risk of disease. Among those, alpha-synuclein, parkin, DJ- 1, PINK1 (phosphatase and tensin homolog (PTEN)-induced kinase 1), LRRK2 (leucine-rich-repeat kinase 2) and HTRA2 are directly or indirectly involved in mitochondrial regulation (summarized in Lin and Beal, 2006).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 33 For example, alpha-synuclein binds to mitochondrial membranes and can cause fragmentation of mitochondria (Kamp et al., 2010; Nakamura et al., 2011). Overexpression of alpha-synuclein in mice impairs mitochondrial function and increases oxidative stress (Song et al., 2004). In isolated rat brain mitochondria under oxidative stress, aggregated alpha-synuclein is also able to induce the release of cytochrome c (Parihar et al., 2008). Cytochrome c release can provoke apoptosis, a finding confirmed in rat substantia nigra overexpressing alpha-synuclein (Yamada et al., 2004). Deficiency of or mutations in the Parkin gene also lead to mitochondrial dysfunction and oxidative stress (Pesah et al., 2004; Palacino et al., 2004). Parkin is an E3 ubiquitin ligase that also ubiquitinates proteins of mitochondria with low mitochondrial membrane potential to target the whole organelle for degradation (Youle and Narendra, 2011). Parkin and the other PD-associated protein PINK1 are involved in the regulation of mitophagy. PINK1 accumulates at the mitochondria in response to a loss in membrane potential and then recruits Parkin to ubiquitinate mitochondrial membrane proteins to induce the degradation of damaged mitochondria (Dzamko et al., 2014). This suggests that dysfunctional Parkin, as is the case when it is mutated in PD, might highly influence the quality of mitochondria in the cell and might be associated with an increase in dysfunctional, fragmented mitochondria (Lutz et al., 2009). In support of this, experiments have shown that stimulating mitochondrial fusion by pharmacological inactivation of Drp1, a protein involved in mitochondrial fission, can offset mitochondrial damage induced by PINK1 and Parkin (Cui et al., 2010).

And a last example is DJ-1, another PD-related protein, has been suggested to function as a sensor for oxidative stress as it contains a sensitive cysteine residue that reacts with ROS. The acidification of this residue leads to DJ-1 relocalizing to the mitochondria (Canet-Aviles et al., 2004).

Thus, several PD-associated proteins are linked to the mitochondria. This suggests that the mitochondrion is indeed a central organelle when it comes to the development of this neurodegenerative disease. PD mainly affects dopaminergic neurons. Under non-disease conditions, these neurons not only have smaller mitochondria and slower transport of mitochondria along the axons (Kim-Han et al., 2011), they also have fewer mitochondria in the cell body and dendrites (Liang et al., 2007) than their non- dopaminergic counterparts. This might explain their higher sensitivity to changes in mitochondrial metabolism and transport relative to other types of neurons.

PD is accompanied by a deficiency of complex I and this has been found in the brain, platelets, and skeletal muscles of patients (Parker et al. 1989). Induction of complex I deficiency by different compounds or mutations leads to parkinsonism in mammals, a state that is accompanied by PD-related phenotypes (Langston et al., 1983) suggesting that the mitochondrial dysfunction might be causative.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 34 Experiments that have tried to bypass ATP production via deficient complex I, in order to restore mitochondrial ATP synthesis, for example by promoting the activity of complex II by feeding energy carriers, have shown improvement of PD-associated phenotypes (Tieu et al., 2003). This also supports the hypothesis that complex I 1 deficiency contributes to the disease. It has been suggested that the toxicity is caused by ATP depletion. However, toxic effects of low ATP levels are only seen if complex I activity is reduced by more than 50% (Davey and Clark, 1996). Since earlier studies have shown that ATP levels in the mitochondria of PD patients are only reduced by about 25-30% (Parker et al. 1989), ATP depletion cannot be the only cause of disease toxicity. An alternative explanation might be the generation of ROS, levels of which are dramatically increased after complex I inhibition (Ramsay et al., 1987) but which might also happen outside of the mitochondria (Zhou et al., 2008).

Another hypothesis is that changes in the recycling of proteins might also be causing PD-related toxicity, as has been suggested for Parkin mutations. General autophagic degradation including mitophagy is also found to be impaired in PD (Dehay et al., 2010; Vila et al., 2011). As mitophagy is necessary for a constant turnover of mitochondria and for adjustments to the energy demands in the cell, as well as for removal of defective mitochondria, lower levels of mitophagy might contribute to an accumulation of dysfunctional mitochondria in PD.

5. Nutrient availability, nutrient sensing and nutrient processing – how the diet influences the aging process One would assume that maintenance of high levels of translation is essential for renewing the pool of fresh and intact proteins (Charmpilas et al., 2015). Protein synthesis levels are highly dependent on nutrient availability, especially on nutrients required for energy metabolism and on amino acids for building proteins. Surprisingly, the attenuation of protein synthesis (e.g. during dietary restriction) has been shown to be beneficial for lifespan, probably shifting energy resources to repair and stress-response mechanisms (Tavernarakis, 2007).

Dietary restriction increases health- and lifespan in all eukaryotic species that have been tested (summarized in López-Otin et al., 2013). These studies demonstrate that nutrient availability and nutrient sensing is central to aging. One hypothesis relating to the lifespan-extending effects of dietary restriction is that it mimics defensive responses by minimizing cell growth and metabolism, a process also seen in the context of systemic damage (Garinis et al., 2008). The process by which dietary restriction extends lifespan is thought to involve the following components: growth hormones; IGF-1 (insulin-like growth factor 1) signaling; insulin signaling, including downstream effectors such as AKT, mTOR and FOXO; metabolites such as glucose,

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 35 lipids, amino acids; and the cellular processes that require these metabolites, including protein synthesis and the production of ATP and biomass (summarized in López-Otin et al., 2013).

As described above (Section 2.3), nutrient availability and nutrient sensing can have effects on metabolism and lifespan over several generations. This epigenetic inheritance has been demonstrated in C. elegans, where dietary restriction-triggered lifespan extension is still observed in up to three generations of normally fed animals following starvation of their predecessors (Greer et al., 2011).

5.1 GCN-2 and mTOR – the key sensors of nutrient availability Nutrients that are taken up in the diet, particularly amino acids, are mainly sensed by GCN-2 (general control nonderepressible -2 kinase) or mTOR (mammalian target of rapamycin). These two kinases then control global translational activity and energy metabolism to maintain cellular homeostasis (Sattlegger and Hinnebusch, 2000). While both kinases have been linked to longevity, their interplay has still barely been studied.

For example, mTOR has been suggested to be one of the key players in the regulation of aging (Blagosklonny, 2006). It integrates mitogenic, oxidative and metabolic signaling by promoting the conversion of nutrients into cellular biomass while inhibiting autophagy and lysosomal biogenesis (Laplante and Sabatini, 2012). It is therefore a central driving force for development and consequently for postdevelopmental aging, in which such processes play a crucial role (Blagosklonny, 2006). Inhibition of mTOR prolongs lifespan in worms, flies and mammals (Hara et al., 2002; Kapahi et al., 2004; Liang et al., 2003). On the other hand, inhibition of mTOR attenuates lifespan benefits caused by dietary restriction, suggesting a phenocopied effect (Johnson et al., 2013).

The activity of mTOR is regulated by several kinases. These include the AMP- activated protein kinase (AMPK) (summarized in Bost et al., 2016). AMPK is activated by decreased energy availability (ratio AMP:ATP) which might be a consequence of reduced levels of energy sources in the diet (Houtkooper et al., 2010). Therefore, AMPK could be a link between dietary restriction and mTOR. The upregulation of AMPK favors healthy aging (Mair et al., 2011). Anne Brunet’s group has shown that AMPK is required for the lifespan extension caused by dietary restriction (Greer et al., 2007). This effect is dependent on the FOXO transcription factor DAF-16 (see Section 5.2), that can be phosphorylated and activated by AMPK (Greer et al., 2007).

Besides the role of AMPK and mTOR in lifespan regulation, activation of the

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 36 mTOR pathway has been also found in various diseases. For example, mTOR seems to increase the production of Tau in neurons, which might contribute to AD (An et al., 2003). And in a fly model of tauopathy (AD), TOR activation causes neurodegeneration (Ravikumar et al., 2004). Furthermore, in a fly model of HD, 1 inhibition of TOR protects against neurodegeneration, probably by increasing the autophagic clearance of toxic proteins (Berger et al., 2006).

5.2 Nutrient sensing and signaling via the insulin/IGF-1 signaling pathway and the FOXO transcription factor DAF-16 Another pathway involved in the extension of lifespan through dietary restriction is the insulin/IGF-1 signaling (IIS) pathway, which integrates nutrient availability and the transcriptional response, thereby modulating cellular metabolic activity. The IIS pathway functionally interacts and communicates with the mTOR pathway at multiple levels (Wullschleger et al., 2006). When altered, the ISS nutrient-sensing pathway has been shown to increase lifespan in worms, flies and mice (e.g. Kenyon et al., 1993; Tatar et al., 2011; Holzenberger et al., 2003) and this lifespan extension protects against age-related pathologies (e.g. Killick et al., 2009; Cohen et al., 2009). Other studies have shown that this might involve autophagy, transcription, cell- cycle inhibition and decreased translational activity including ribosome synthesis (Blagosklonny, 2006; Anisimov et al., 2010; Bove et al., 2011).

The very end of the ISS pathway involves the activity of transcription factors that modulate the transcriptional response to metabolic needs. Among these is the family of FOXO transcription factors. FOXO transcription factors are highly conserved and are involved in the response to several stresses (including insulin signaling via the IIS pathway). They do so by upregulating the transcription of a series of target genes, including ROS-detoxification genes (Oh et al., 2006; Barzilai et al., 2012). The best studied FOXO transcription factor is DAF-16 in C. elegans. DAF-16 regulates the transcription of diverse cellular processes, including metabolism, the oxidative stress response, detoxification and immunity (Dong et al., 2007; McElwee et al., 2003; Murphy et al., 2003).

Only a few longevity-promoting factors in C. elegans have been found to act independently of the FOXO transcription factor DAF-16 (Hamilton et al., 2005). One of them is mTOR (Vellai et al., 2003). Others include factors that help to control the activity of the mitochondrial oxidative phosphorylation machinery (Feng et al., 2001; Dillin et al, 2002; Lee et al., 2003) and the hypoxic response (Yanos et al., 2012).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 37 5.3 Nutrient sensing and mitochondrial activity As mentioned above, mitochondria use different metabolites to produce ATP. Thus, mitochondrial activity is highly dependent on the availability of nutrients in the diet. Several signaling pathways are involved in the communication between nutrient sensing and mitochondrial energy metabolism. For example, a state of low energy can also influence translational activity and biomass production, as about half of a cell’s total energy is used for translation (Charmpilas et al., 2015). Mitochondrial mutants in C. elegans are typically smaller in size and have lower reproductive rates (Dancy et al., 2014), both of which might be a consequence of reduced protein synthesis. A possible link between energy sensing and protein synthesis is – again – AMPK, which regulates mTOR activity (Bost et al., 2016). This could be mediated by the downregulation of components of the ETC, which have been shown to trigger longevity in multiple studies (Lee et al., 2003; Hamilton et al., 2005; Curran and Ruvkun, 2007). In most cases this happens independently of the IIS pathway (Lakowski and Hekimi, 1996; Dillin et al., 2002). The low energy state might again activate AMPK (see Section 5.1). Alternatively, a low energy state might be detected by sirtuins via high NAD+ levels (inactivation of mitochondria inhibits the conversion of NAD+ to NADH) (Houtkooper et al., 2010). Sirtuins are deacetylases that regulate the metabolic response; they do so by promoting mitochondriogenesis, enhancing antioxidant defense and increasing fatty acid oxidation (Fernandez-Marcos and Auwerx, 2011). An upregulation of sirtuins improves healthy aging (Rodgers et al., 2005) and both sirtuins and AMPK engage feedback loops that connect the two signaling pathways (Price et al., 2012).

5.4 Nutrient-specific diets and their implications in lifespan control Dietary restriction increases lifespan in a broad range of organisms (see Section 5). In recent years, research groups have tried to find out which essential factors in a diet contribute to longevity. For this reason, diets that either miss or are increased in a certain component have been studied for their effect on lifespan. A central group of components in food are amino acids. Levels of free amino acids in an organism are known to change with age, and in mutant animals that have an extended lifespan the majority of amino acid pools are upregulated (Fuchs et al., 2010). It is not yet known whether the increased amino acid pools are due to reduced rates of protein synthesis or to an upregulation in amino acid synthesis or a downregulation in amino acid degradation. However, these findings suggest that the availability of amino acids might be crucial for the survival of an organism.

At this point, studies on specific amino acids are still rather inconclusive and sometimes contradictory. For example, the restriction of methionine or tryptophan has been suggested to extend lifespan in higher eukaryotes (Orentreich et al., 1993;

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 38 Ooka et al., 1988), whereas increased levels of proline and tryptophan extend lifespan in C. elegans (Zarse et al., 2012; Van der Goot et al., 2012). Feeding mice with high amounts of branched chain amino acids (leucine, isoleucine and valine) can extend the average lifespan of male mice, and this is accompanied by increased 1 mitochondrial biogenesis and decreased ROS production (D’Antona et al., 2010). In line with this, in yeast, supplementation with branched amino acids and threonine also extends chronological lifespan (Alvers et al., 2009). But others have found that addition of serine, threonine or valine decreases yeast’s chronological lifespan (Mirisola et al., 2014) while deprivation of methionine, aspartate, glutamate and asparagine has lifespan-extending effects (Wu et al., 2013; Powers et al., 2006). Similar effects were seen in another study, in which depletion of methionine extended the chronological lifespan of yeast cells and delayed senescence of mammalian cell cultures (Johnson and Johnson, 2014; Koziel et al., 2014).

Claire Edwards and her colleagues performed a systematic screen to study the effect on lifespan of supplying C. elegans with different amounts of amino acids as an addition to their diet (Edwards et al., 2015). They found that low levels of supplementation of most amino acids (except for phenylalanine and aspartate) increased the nematode’s lifespan. However, the lifespan-increasing effects declined with higher amounts of amino acids (Edwards et al., 2015) suggesting that the effect on lifespan depends on amino acid balance, which might explain the protective and toxic effects of the same amino acid seen in different studies.

It is also necessary to mention that the studies performed by Edwards et al. and by others did not check whether an amino acid had been taken up by the animals (internal amino acid levels) or whether the feeding might have induced a general stress response due to changes in the animals’ diet. For example, food avoidance behavior could provoke starvation effects, and it has been shown that just small changes in the bacterial composition of feed for C. elegans can activate the animals’ innate immune response (Alper et al., 2008). Furthermore, when interpreting such studies, one should also keep in mind that the amino acid pools are connected and that the animals themselves can generate any missing non-essential amino acids. It is therefore possible, for example, that methionine is converted to cysteine (summarized in Jung, 2015). Changes in metabolism can also affect amino acid pools which then has consequences for other cellular processes. For example, reduced ATP production in the mitochondria leads to an upregulation of branched amino acids such as leucine, valine and isoleucine, which can be used to supply the mitochondria with fresh substrate (Falk et al., 2008). Thus, there might be secondary effects that influence the outcome of an experiment.

All these results show that supplementation or deprivation studies depend strongly on the levels of the component of interest and that their interpretation requires well-

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 39 controlled experiments and tests (e.g. metabolite measurements).

Besides changes in the diets of model organisms, genetic studies on genes that affect nutrient transport and metabolism have also been performed. For example, the knockdown of aromatic amino acid transporters has been shown to extend the lifespan of C. elegans, suggesting that low levels of tryptophan might be beneficial (He et al., 2014). In line with this finding, inhibition of the major tryptophan- degrading pathway in C. elegans – which leads to an increase in tryptophan levels – also results in lifespan extension (Van der Goot et al., 2012). And a decrease in tyrosine degradation leads to increased longevity (Ferguson et al., 2013).

6. Tryptophan and its metabolites in cellular signaling Tryptophan is the least abundant essential amino acid. Besides being used for the production of proteins, tryptophan is also a precursor for a range of signaling molecules (Figure 4). It can be converted into the neurotransmitters serotonin and melatonin and into the neuromodulator tryptamine (summarized in Van der Goot and Nollen, 2013). However, most of the free tryptophan that is not used for translation enters the kynurenine pathway (Anderson et al., 2013). Whereas this pathway was long thought to be mainly a source of the coenzyme NAD+ (Schwarcz et al., 2012), other metabolites of the kynurenine pathway have now been found to be associated with neuromodulation: kynurenic acid (KYNA) has neuroprotective properties and 3-hydroxykynurenine (3-HK) and quinolinic acid (QUIN) are neurotoxic at high concentrations (summarized in Maddison and Giorgini, 2015). The effects of these metabolites have been linked to three distinct properties: their antagonistic actions on neuronal receptors (e.g. N-methyl-D-aspartate (NMDA)) (Schwarcz et al., 2012); their ability to modulate the redox state of the cell by ROS-generating or ROS-detoxifying properties (Ocampo et al., 2014); and/or their localization close to the mitochondria and their effects in modulating mitochondrial activity (Baran et al., 2003). As tryptophan as well as kynurenine can cross the blood brain barrier (summarized in Maddison and Giorgini, 2015), changes in serum levels of metabolites of the kynurenine pathway can highly influence their abundance in the brain: Alterations in tryptophan metabolism during aging have been reported. For example, people of advanced age have a higher kynurenine/tryptophan ratio, which has been suggested to be due to increased tryptophan degradation (Frick et al., 2004).

In a variety of different diseases, measuring the levels and ratios of metabolites has become a relatively common technique for deducing information about the health of an individual (summary of studies in Bohar et al., 2015). As a result, alterations in tryptophan metabolism have been found in several diseases. For example, increased activity of IDO (indolamine 2,3-dioxygenase) and TDO (tryptophan 2,3-dioxygenase)

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 40 human/ C. elegans

AADC/ ? TPH/TPH-1 TPH/TPH-1 tryptamine tryptophan 5-hydroxytryptophan serotonin 1 IDO/ TDO/TDO-2

N-formylkynurenine

AFMID/AMFD-1

KYNU/FLU-2

KAT-I/ KAT-II/ KAT-III/ NKAT-1/ NKAT-2/ TATN-1 anthranillic acid kynurenine kynurenic acid

KMO/KMO-1

KAT-I/ KAT-II/ KAT-III/ NKAT-1/ NKAT-2/ TATN-1 3-hydroxykynurenine xanthurenic acid

KYNU/FLU-2

3-hydroxyanthranillic acid

HAAO/HAAO-1

ACMSD/Y71D11A.3 quinolinic acid picolinic acid

NAD+

Figure 4: Overview of tryptophan metabolism The metabolism of the essential amino acid tryptophan is highly conserved from nematodes to humans. The major breakdown products are serotonin, tryptamine and the metabolites of the kynurenine pathway.Humans, however, possess besides TDO (tryptophan 2,3-dioxygenase) a second enzyme, IDO (indolamine 2,3-dioxygenase), which is also able to convert tryptophan into N-formylkynurenine.

has been found in different kinds of cancer, thereby increasing the immune tolerance of the tumor (Hjortsø et al., 2015; Metz et al., 2012). Patients with cardiovascular disease also have reduced tryptophan levels (Wirleitner et al., 2003). In type 2 diabetes, patients with high plasma kynurenine levels are more likely to have severe insulin resistance (Oxenkrug, 2015). Other research has shown that patients with AD have low tryptophan and high kynurenine levels, and that these levels correlate with their cognitive decline (Widner et al, 1999). In other patients with AD, high concentrations of QUIN have been found in amyloid plaques (Ting et al., 2009). Furthermore, in the early phase of HD, elevated levels of QUIN and 3-HK have been

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 41 measured, as well as low levels of KYNA (Beal et al., 1990; Pearson et al, 1992; Guidetti et al., 2004). Based on these findings in different diseases, enzymes of the kynurenine pathway have become popular targets in drug development (summarized in Schwarcz et al., 2012).

Most of the above-mentioned studies are however correlative, and therefore do not provide much insight into the mechanisms that might underlie the aging process or the progression of a certain disease. Research is still needed in order to elucidate whether the metabolic changes observed are a cause or consequence of a certain disease, and what exactly happens at the cellular level. Only a few studies in worms, flies and more recently in mice have started to elucidate the importance of tryptophan metabolism in homeostasis and signaling (Van der Goot et al., 2012; Campesan et al., 2011; Zwilling et al., 2011; Opitz et al., 2011; Breda et al., 2016, for more details see Sections 6.1 and 6.2). One of the main developments in this field comes from studies on a central enzyme of the kynurenine pathway called kynurenine monooxygenase (KMO; Figure 4), and the findings regarding its involvement in neurodegenerative diseases. KMO mutations in yeast have been shown to suppress HD-related proteotoxicity (Giorgini et al., 2005) and inhibition of KMO in flies can reduce neurodegeneration in a HD model; this was found to be connected to increased levels of KYNA relative to 3-HK (Campesan et al., 2011). Furthermore, inhibition of KMO has also been shown to reduce neurotoxicity in mouse models of AD and HD (Zwilling et al., 2011).

6.1 The central enzymes of tryptophan metabolism – tryptophan 2,3-dioxygenase (TDO) and indolamine 2,3-dioxygenase (IDO) In mammals, the conversion of tryptophan to N-formylkynurenine (first step of the kynurenine pathway) can be catalyzed by two different enzymes: tryptophan 2,3-dioxygenase (TDO) and indolamine 2,3-dioxygenase (IDO) (Figure 4; Higuchi and Hayaishi, 1967). Whereas TDO is found in almost all species, IDO only exists in higher organisms, suggesting that TDO developed earlier in evolution (Yuasa and Ball, 2015). Besides catalyzing the same enzymatic conversion of tryptophan, there is no significant homology between the two enzymes (Dang et al., 2000). While TDO is thought to regulate systemic tryptophan levels, IDO is usually activated locally to deplete tryptophan in certain tissues (summarized in Van der Goot and Nollen, 2013). Furthermore, TDO and IDO differ in their expression in different tissues and are regulated via different stimuli. TDO is mainly expressed in the liver and induced by glucocorticoids that are secreted in response to stress (Danesch et al., 1987). Recently, TDO expression has also been found in the brain (Kanai et al., 2009). In contrast, IDO is ubiquitously expressed except in the liver (Haber et al., 1993) and is stimulated by immunoregulatory molecules such as interferon-γ

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 42 (Yoshida et al., 1981; Hayaishi and Yoshida, 1978). Whereas TDO needs molecular oxygen for its activity, IDO preferentially uses superoxide but can also use oxygen (Dang et al., 2000). Furthermore, TDO is strictly limited to tryptophan as substrate, whereas IDO can also bind and metabolize other tryptophan metabolites such as 1 tryptamine and serotonin (Sono et al., 1996).

TDO and IDO have been the subject of several studies in various model organisms. Both knockout mice for TDO and IDO are viable and do not show any severe morphological or behavioral abnormalities (Kanai et al., 2009; Mellor et al., 2003). C. elegans and Drosophila melanogaster possess only TDO (TDO-2 in C. elegans, vermillion in D. melanogaster) (summarized in Van der Goot and Nollen, 2013). Depletion of TDO in those model organisms has been shown to increase lifespan (Van der Goot et el., 2012; Campesan et al., 2011; Oxenkrug, 2010). Research on IDO has been done in mammalian model systems. It has been found that IDO is involved in immune regulation via macrophages and it is stimulated by INF-γ and other immune stimuli (Yoshida et al., 1981; Havaishi and Yoshida, 1978). It has been proposed that IDO depletes tryptophan levels locally to induce cell-cycle arrest in T cells, which consequently results in their apoptosis (Mellor et al., 2001). Fougeray and colleagues found that the INF-γ-induced depletion of tryptophan activates GCN-2, which is further needed to activate autophagic processes also related to macrophage activity (Fougeray et al., 2012). IDO might directly be involved in pathogen defense by the same mechanism, depleting tryptophan in the environment of pathogens that are unable to generate their own tryptophan and that are then inhibited in their proliferation (Gupta et al., 1994). However, the hypothesis of local tryptophan depletion still needs to be proven. The role of IDO and tryptophan in the immune response described here has also been shown to be important during pregnancy. IDO is upregulated in the placenta and inhibits the activity and proliferation of maternal T cells to prevent the detection and destruction of the allogeneic fetus. Inhibition of IDO in pregnant rats leads to the loss of the embryos (Munn et al., 1998; Munn et al., 2005).

6.2 The role of TDO in aging and age-related diseases As mentioned above, in recent years TDO has become the subject of several studies on aging and age-related diseases. The depletion of TDO as well as supplementation of tryptophan has shown reduced toxicity in a fly model for HD as well as in worm models for AD, HD and PD (Campesan et al., 2011; Van der Goot et al., 2012; Breda et al., 2016). Whereas the protective effects in flies have been linked to changes in the levels of metabolites of the kynurenine pathway (Breda et al., 2016), the healthspan effects in worms have been found to be independent of the kynurenine pathway, suggesting the involvement of additional pathways (Van der Goot et al.,

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 43 2012). It is also worth mentioning that TDO depletion has beneficial effects on life- and healthspan not only in disease models but also in wild type nematodes. This suggests that TDO is a general regulator of the proteotoxicity that is induced by diseases or seen as an effect of aging (Van der Goot et al., 2012).

It is however not yet fully clear, how TDO mediates its proteoprotective effects. Does it involve other tryptophan-metabolizing pathways or signaling by tryptophan metabolites? Does a change in tryptophan availability induce translational changes that might be beneficial for the animal’s health- and lifespan? Are completely new pathways involved? In particular, aging-associated processes that have not been linked to tryptophan metabolism before? The goal of this thesis is to understand the cellular mechanisms that increase healthspan and extend lifespan in C. elegans as a result of TDO depletion.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 44 REFERENCES

• Alper, S et al.: Identification of innate immunity genes 433-42 (2006). and pathways using a comparative genomics approach. • Blagosklonny, MV: Aging and immortality: quasi- Proc Natl Acad Sci U S A 105(19), 7016-21 (2008). programmes senescence and its pharmacologic 1 • Alvers, AL et al.: Autophagy and amino acid inhibition. Cell Cycle 5(18), 2087-102 (2006). homeostasis are required for chronological longevity • Blasco, MA: Telomere length, stem cells and aging. in Saccharomyces cerevisiae. Aging Cell 8(4), 353-69 Nat Chem Biol 3, 640-649 (2007). (2009). • Bohar, Z, Toldi, J, Fülöp F, Vécsei, L: Changing the • An, WL et al.: Upregulation of phosphorylated/ face of kynurenines and neurotoxicity: therapeutic activated p70 S6 kinase and its relationship to considerations. Int J Mol Sci 16, 9772-9793 (2015). neurofibrillary pathology in Alzheimer’s disease. Am J Pathol 163, 591-607 (2003). • Bost, F, Decoux-Poullot, AG, Tanti, JF, Clavel, S: Energy disruptors: rising stars in anticancer therapy?. • Anandatheerthavarada, HK, Biswas, G, Robin, Oncogenesis 5, (2016). MA, Avadhani, NG: Mitochondrial targeting and a novel transmembrane arrest of Alzheimer’s amyloid • Bove, J et al.: Fighting neurodegeneration with precursor protein impairs mitochondrial function in rapamycin: mechanistic insights. Nat Rev Neurosci 12, neuronal cells. J Cell Biol 161, 41-54 (2003). 437-452 (2011). • Anderson, G et al. Increased IL-6 trans-signaling in • Brehme, M et al.: A chaperome subnetwork safeguards depression: focus on the tryptophan catabolite pathway, proteostasis in aging and neurodegenerative disease. melatonin and neuroprogression. Pharmacological Cell Rep. 9, 1135-1150 (2014). Reports 65, 1647-1654 (2013). • Brand, MD: The role of mitochondria in longevity and • Anisimov, VN et al.: Rapamycin extends maximal healthspan. Longev Healthspan 3, 7 (2014). lifespan in cancer-prone mice. Am J Pathol 176, 2092- • Breda, C et al.: Tryptophan-2,3-dioxygenase 2097 (2010). (TDO) inhibition ameliorates neurodegeneration by • Armanios, M, Blackburn, EH: The telomere syndrome. modulation of kynurenine pathway metabolites. Proc Nat Rev Genet 13, 693-704 (2012). Natl Acad Sci U S A 113(19), 5435-40 (2016). • Astanina, K, Koch, M, Jungst, C, Zumbusch, A, • Brehme, M et al.: A chaperome subnetwork safeguards Kiemer, AK: Lipid droplets as a novel cargo of proteostasis in aging and neurodegenerative disease. tunneling nanotubes in endothelial cells. Sci. Rep. 5, Cell Rep 9, 1135-1150 (2014). 11453 (2015). • Brown, H, Sullivan, PG, Geddes, JW: Synaptic • Baker, BM, Nargund, AM, Sun, T, Haynes, CM: mitochondria are more susceptible to Ca2+ overload Protective coupling of mitochondrial function and than nonsynaptic mitochondria. J Biol Shem 281, protein synthesis via the eIF2alpha kinase GCN-2. 11658-11668 (2006). PLoS Genet 8(6), (2012). • Campesan, S et al.: The kynurenine pathway • Balch, WE, Morimoto, RI, Dillin, A, Kelly, JW: modulates neurodegeneration in a Drosophila model Adapting proteostasis for disease intervention. Science of Huntington’s disease. Curr Biol 21, 961-966 (2011). 319, 916-919 (2008). • Canet-Aviles, RM et al.: The Parkinson’s disease • Ban, H et al.: Arginine and Leucine regulate p70 S6 protein DJ-1 is neuroprotective due to cysteine-sulfinic kinase and 4E-BP1 in intestinal epithelial cells. Int J acid-driven mitochondrial localization. Proc Natl Mol Med 13(4), 537-43 (2004). Acad Sci U S A 101, 9103-9108 (2004). • Baran, H et al.: Kynurenines and the respiratory • Chandel, NS: Evolution of mitochondria as signaling parameters on rat heart mitochondria. Life Sci 72(10), organelles. Cell Metab 22(2), 204-6 (2015). 1103-15 (2003). • Chance, B, sies, H, Boveris, A: Hydroxyperoxide • Barzilai, N, Huffman, DM, Muzumdar, RH, Bartke, metabolism in mammalian organs. Physiol Rev 59, A: The critical role of metabolic pathways in aging. 527-605 (1979). Diabetes 61, 1315-1322 (2012). • Charmpilas, N, Daskalaki, I, Papandreou, ME, • Beal, MF et al.: Kynurenine pathway measurements Tavernarakis, N: Protein synthesis as an integral in Huntington’s disease striatum: evidence for reduced quality control mechanism during ageing. Ageing formation of kynurenic acid. J Neurochem 55, 1327- Research Reviews 23, 75-89 (2015). 1339 (1990). • Chiti, F, Dobson, CM: Amyloid formation by globular • Ben-Zvi, A, Miller, EA, Morimoto, RI: Collapse of proteins under native conditions. Nat Chem Biol 5, 15- proteostasis represents an early molecular event in 22 (2009). Caenorhabditis elegans aging. Proc. Natl. Acad. Sci U • Chondrogianni, N, Georgila, K, Kourtis, N, S A 106, 14914-14919 (2009). Tavernarakis, N, Gonos, ES: 20S proteasome • Berger, Z et al.: Rapamycin alleviates toxicity of activation promotes life span extension and resistance different aggregate-prone proteins.Hum Mol Genet 15, to proteotoxicity in Caenorhabditis elegans. FASEB J.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 45 29, 611-622 (2015). approaches. Metabolites 4, 831-878 (2014). • Clayton, DF, George, JM: The synucleins: a family • Detmer, SA, Chan, DC: Functions and dysfunctions of proteins involved in synaptic function, plasticity, of mitochondrial dynamics. Nat Rev Mol Cell Biol 8, neurodegeneration and disease. Trends Neurosci 21, 870-879 (2007). 249-254 (1998). • Dillin, A et al.: Rates of behavior and aging specified • Cohen, E et al.: Reduced IGF-1 signaling delays age- by mitochondrial function during development. associated proteotoxicity in mice. Cell 139, 1157-1169 Science 298, 2398-401 (2002). (2009). • DiMauro, S, Schon, EA: Mechanisms of disease: • Corral-Debrinski, M et al.: Mitochondrial DNA Mitochondrial respiratory-chain diseases. N Engl J deletions in human brain : regional variability and Med 348, 2656-2668 (2003). increase with advanced age. Nature Genet 2, 324-329 • Dong, MQ et al.: Quantitative mass spectrometry (1992). identifies insulin signaling targets in C.elegans. • Cui, M, Tang, X, Christian, WV, Yoon, Y, Tieu, K: Science 317, 660-663 (2007). Perturbations in mitochondrial dynamics induced by • Doonan, R et al.: Against the oxidative damage theory human PINK1 can be rescued by the mitochondrial of aging: superoxide dismutases protect against division inhibitor mdivi-1. J Biol Chem 285, 11740- oxidative stress but have little or no effect on life span 11752 (2010). in Caenorhabditis elegans. Genes Dev 22, 3236-3241 • Curran, SP and Ruvkun, G: Lifespan regulation by (2008). evolutionarily conserved genes essential for viability. • Durieux, J, Wolff, S and Dillin, A: The cell non- PLoS Genet. 3(4), (2007). autonomous nature of electron transport chain- • Dahl, JU, Gray, MJ, Jakob, U: Protein quality control mediated longevity. Cell 144 (1), 79-91 (2011). under oxidative stress conditions. J Mol Biol 427(7), • Dwyer, BE, Fando, JL, Wasterlain, CG: Rat brain 1549-1563 (2015). protein synthesis declines during postdevelopmental • Dancy, BM, Sedensky, MM, Morgan, PG: Effects of aging. J Neurochem 35, 746-749 (1980). the mitochondrial respiratory chain on longevity in • Dzamko, N, Zhou, J, Huang, Y, Halliday, GM: C. elegans. Experimental Gerontology 56, 245-255 Parkinson’s disease-implicated kinases in the brain; (2014). insights into disease pathogenesis. Front Mol Neurosci • Danesch, U et al.: Glucocorticoid induction of the rat 7, 57 (2014). tryptophan oxygenase gene is mediated by 2 widely • Edgar, D et al.: Random point mutations with major separated glucocorticoid-responsive elements. EMBO effects on protein-coding genes are the driving force J 6, 625-630 (1987). behind premature aging in mtDNA mutator mice. Cell • Dang, Y, Dale, WE, Brown, OE: Comparative Metab 10, 131-138 (2009). effects of oxygen on indoleamine 2,3-dioxygenase • Edwards, C et al.: Mechanisms of amino acid-mediated and tryptophan 2,3-dioxygenase of the kynurenine lifespan extension in Caenorhabditis elegans. BMC pathway. Free Radical Biology and Medicine 28, 615- Genetics, (2015). 624 (2000). • Falk, MJ et al.:Metabolic pathway profiling of • D’Antona, G et al.: Branched-chain amino acid mitochondrial respiratory chain mutants in C. elegans. supplementation promotes survival and supports Mol Genet Metab 93(4), 388-397 (2008). cardiac and skeletal muscle mitochondrial biogenesis in middle-aged mice. Cell Metab 12(4), 362-72 (2010). • Feng, J, Bussiere, F, Hekimi, S: Mitochondrial electron transport is a key determinant of life span in • Davey, GP, Clark, JB: Threshold effects and control Caenorhabditis elegans. Dev Cell 1, 633-644 (2001). of oxidative phosphorylation in nonsynaptic rat brain mitochondria. J Neurochem 66, 1617-1624 (1996). • Ferguson, AA et al.: TATN-1 mutations reveal a novel role for tyrosine as a metabolic signal that • David, DC et al.: Widespread protein aggregation as influences developmental decisions and longevity in an inherent part of aging in C. elegans. PLoS Biol 8, Caenorhabditis elegans. PLoS Genet 9(12), (2013). (2010). • Fernandez-Marcos, PJ, Auwerx, J: Regulation of PGC- • David, DC: Aging and the aggregating proteome. 1alpha, a nodal regulator of mitochondrial biogenesis. Front Genet. 3, 247 (2012). Am J Clin Nutr 93, 884S-890 (2011). • Davidson, WS, Jonas, A, Clayton, DF, George, JM: • Fougeray, S et al.: Tryptophan depletion and the kinase Stabilization of alpha-synuclein secondary structure GCN2 mediate IFN-gamma-induced autophagy. J upon binding to synthetic membranes. J Biol Chem Immunol 189, 2954-2964 (2012). 273, 9443-9449 (1998). • Frick, B et al.: Increasing production of homocysteine • Dehay, B et al.: Pathogenic lysosomal depletion in and neopterin and degradation of tryptophan with Parkinson’s disease. J Neurosci 30, 12535-12544 older age. Clin Biochem 37, 684-687 (2004). (2010). • Friedman, DB, Johnson, TE: Three mutants that • Demine, S, Reddy, N, Renard, P, Raes, M, Arnould, extend both mean and maximum life span of the T: Unraveling biochemical pathways affected by nematode, Caenorhabditis elegans, defins the age-1 mitochondrial dysfunctions using metabolomic gene. J Gerontol 43(4), B102-9 (1988)

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 46 • Fritz, KS et al.: 4-hydroxynonenal inhibits SIRT3 via 1544-55 (2005). thiol-specific modification. Chem Res Toxicol 24(5), • Hansen, M et al.: New genes tied to endocrine, 651-662 (2011). metabolic, and dietary regulation of lifespan from a • Fuchs, S et al.: A metabolic signature of long life in Caenorhabditis elegans genomic RNAi screen. PLoS Caenorhabditis elegans. BMC Biol 8, 14 (2010). Genet 1, 119-128 (2005). • Fusco, G et al.: Direct observation of the three regions • Hansson, CA et al.: Nicastrin, presenilin, APH-1, and 1 in alpha-synuclein that determine its membrane-bound PEN-2 form active gamma-secretase complexes in behavior. Nat Commun 5, 3827 (2014). mitochondria. J Biol Chem 279, 51654-51660 (2004). • Galvagnion, C et al.: Lipid vesicles trigger alpha- • Hara, K et al.: Raptor, a binding partner of target of synuclein aggregation by stimulating primary rapamycin (TOR), mediates TOR action. Cell 110, nucleation. Nat Chem Biol 11(3), 229-34 (2015). 177-89 (2002). • Garinis, GA, van der Horst, GT, Vijg, J, Hoeijmakers, • Harman, D: Aging: a theory based on free radical and JH: DNA damage and ageing: new-age ideas for an radiation chemistry. J Gerontol 11, 298-300 (1956). age-old problem. Nat Cell Biol 10, 1241-1247 (2008). • Hartl, FU, Bracher, A, Hayer-Hartl, M: Molecular • Gavilán, MP et al.: Age-related differences in the chaperons in protein folding and proteostasis. Nature dynamics of hippocampal proteasome recovery. J 475, 324-332 (2011). Neurochem 123(4), 635-44 (2012). • Harries, LW et al.: Human aging is characterized by • Gavilán, E et al.: Age-related dysfunctions of the focused changes in gene expression and deregulation autophagy lysosomal pathway in hippocampal of alternative splicing. Aging Cell 10, 868-878 (2011). pyramidal neurons under proteasome stress. Neurobiol. • Harrison, DE et al.: Rapamycin fed late in life extends Aging 36, 1953-1963 (2015). lifespan in genetically heterogenous mice. Nature 460, • Giorgini, F, Guidetti, P, Nguyen, Q, Bennett, SC, 392-395 (2009). Muchowski, PJ: A genomic screen in yeast implicates • Hayaishi, O, Yoshida, R: Specific induction of kynurenine 3-monooxygenase as a therapeutic target pulmonary indoleamine 2,3-dioxygenase by bacterial for Huntington disease. Nat Genet 37, 526-531 (2005). lipopolysaccharide. Ciba Found Symp 65, 199-203 • Gomes, AP et al.: Declining NAD(+) induces a (1978). pseudohypoxic state disrupting nuclear-mitochondrial • Hayflick, L, Moorhead, PS: The serial cultivation of communication during aging. Cell 155, 1624-1638 human diploid cell strains. Exp Cell Res 25, 585-621 (2013). (1961). • Greer, EL et al.: An AMPK-FOXO pathway mediates • Haynes, CM, Yang, Y, Blais, SP, Neubert, TA, Ron, longevity induced by a novel method of dietary D: The matrix peptide exporter HAF-1 signals a restriction in C. elegans. Curr Biol 17(19), 1646-56 mitochondrial UPR by activating the transcription (2007). factor ZC376.7 in C. elegans. Mol Cell 37, 529-540 • Greer, EL et al.: Transgenerational epigenetic (2010). inheritance of longevity in Caenorhabditis elegans. • He, C et al.: Enhanced longevity by ibuprofen, Nature 479, 365-371 (2011). conserved in multiple species, occurs in yeast through • Gregg, SQ et al.: A mouse model of accelerated liver inhibition of tryptophan import. PLoS Genet 10(12), aging caused by a defect in DNA repair. Hepatology (2014). 55, 609-621 (2012). • Higuchi, K, Hayaishi, O: Enzymatic formation of • Gu, M et al.: Mitochondrial defect in Huntington’s D-kynurenine from D-tryptophan. Arch Biochem disease caudate nucleus. Ann Neurol. 39, 385-389 Biophys 120(2), 397-403 (1967). (1996). • Hiona, A et al.: Mitochondrial DNA mutations induce • Guidetti, P, Luthi-Carter, RE, Augood, SJ, Schwarcz, mitochondrial dysfunction, apoptosis and sarcopenia R: Neostriatal and cortical quinolinate levels are in skeletal muscle of mitochondrial DNA mutator increased in early grade Huntington’s disease. mice. PLoS ONE, (2010). Neurobiol Dis 17, 455-461 (2004). • Hjortsø, MD et al.: Tryptophan 2,3-dioxygenase • Gupta, SL et al.: Antiparasitic and antiproliferative (TDO)-reactive T cells differ in their functional effects of indoleamine 2,3-dioxygenase enzyme characteristics in health and cancer. OncoImmunology expression in human fibroblasts. Infect Immun 62, (2015) 2277-2284 (1994). • Hoeijmakers, JH: DNA damge, aging, and cancer. N • Haber, R, Bessette, D, Hulihan-Giblin, B, Durcan, MJ: Engl J Med 361, 1475-1485 (2009). Identification of tryptophan 2,3-dioxygenase RNA in • Holcik, M, Sonnenberg, N: Translational control in rodent brain. J Neurochem 60, 1159-1162 (1993). stress and apoptosis. Nat Rev Mol Cell Biol 6, 318- • Halfmann, R et al.: Prions are a common mechanism 327 (2005). for phenotypic inheritance in wild yeast. Nature 482, • Holzenberger, M et al.: IGF-1 receptor regulates 363-368 (2012). lifespan and resistance to oxidative stress in mice. • Hamilton, B et al.: A systematic RNAi screen for Nature 421, 182-187 (2003). longevity genes in C. elegans. Genes Dev. 19(13),

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 47 • Houtkooper, RH, Williams, RW, Auwerx, J: Metabolic key regulator of proteostasis. EMBO J. 32, 1451-1468 networks of longevity. Cell 142, 9-14 (2010). (2013). • Hsu, AL, Murphy, CT, Kenyon, C: Regulation of aging • Klass, MR: A method for the isolation of longevity and age-related disease by DAF-16 and heat-shock mutants in the nematode Caenorhabditis elegans and factor. Science 300, 1142-1145 (2003). initial results. Mech Ageing Dev. 22, 279-286 (1983). • Ishii, Y et al.: Quantification of nitrated tryptophan • Koga, H, Kaushik, S, Cuervo, AM: Protein homeostasis in proteins and tissues by high-performance liquid and aging: The importance of exquisite quality control. chromatography with electrospray ionization tandem Ageing Res Rev. 10, 202-215 (2011). mass spectrometry. J Phram Biomed Anal 44, 150-159 • Kong, X et al.: Sirtuin 3, a new target of PGC-1 alpha, (2007). plays an important role in the suppression of ROS and • Jezek, P: Mitochondria in homeostasis of reactive mitochondrial biogenesis. PLoS One 5(7), (2010). oxygen species in cell, tissue and organism. Int J • Koziel, R et al.: Methionine restriction slows down Biochem Cell Biol 37, 2478-2503 (2005). senescence in human diploid fibroplasts. Aging Cell • Johnson, JE, Johnson, FB: Methionine restriction 13, 1038-1048 (2014). activates the retrograde response and confers both • Lamitina, T, Huang, CG, Strange, K: Genome- stress tolerance and lifespan extension to yeast, mouse wide RNAi screening identifies protein damage as a and human cells. PLoS ONE 9, (2014). regulator of osmoprotective gene expression. Proc • Johnson, SC, Rabinovitch, PS, Kaeberlein, M: mTOR Natl Acad Sci U S A 103(32), 12173-8 (2006). is a key modulator of ageing and age-related disease. • Langston, JW, Ballard, P, Irwin, I: Chronic Nature 493, 338-345 (2013). parkinsonism in humans due to a product of • Jung, YS: Metabolism of sulfur-containing amino meperidine-analog synthesis. Science 219, 979-980 acids in the liver: a link between hepatic injury and (1983). recovery. Biol Pharm Bull 38, 971-974 (2015). • Lakowski, B, Hekimi, S: Determination of life-span in • Kamp, F et al.: Inhibition of mitochondrial fusion by Caenorhabditis elegans by four clock genes. Science alpha-synuclein is rescued by PINK1, Parkin and DJ- 272, 1010-3 (1996). 1. EMBO J 29, 3571-3589 (2010). • Lane, RK, Hilsabeck, T, Rea, SL: The role of • Kanai, M et al.: Tryptophan 2,3-dioxygenase is a key mitochondria dysfunction in age-related diseases. modulator of physiological neurogenesis and anxiety- Biochimica et Biophysica Acta 1847, 1387-1400 related behavior in mice. Mol Brain 2, 8 (2009). (2015). • Kapahi, P et al.: Regulation of lifespan in Drosophila • Laplante, M, Sabatini, DM: mTOR signaling in by modulation of genes in the TOR signaling pathway. growth control and disease. Cell 149, 274-293 (2012). Curr Biol 14, 885-90 (2004). • Lee, SS et al.: A systematic RNAi screen identifies a • Kapranov, P et al.: RNA maps reveal new RNA classes critical role for mitochondria in C. elegans longevity. and a possible function for pervasive transcription. Nat Genet. 33(1), 40-8 (2003). Science 316, 1484-1488 (2007). • Levine, RL, Stadtman, ER: Oxidative modification of • Kaushik, S, Cuervo, AM: Proteostasis and aging. Nat protein during aging. Exp Gerontol 36(9), 1495-502 Med 21(12), 1406-15 (2015). (2001). • Kenyon, C et al.: A C. elegans mutant that lives twice • Liang, CL, Wang, TT, Luby-Phelps, K, German, DC: as long as wild type. Nature 366, 461-464 (1993). Mitochondria mass is low in mouse substantia nigra • Killick, R et al.: Deletion of Irs2 reduces amyloid dopamine neurons: Implications for Parkinson’s deposition and rescues behavioural deficits in APP disease. Exp Neurol 203, 370-380 (2007). transgenic mice. Biochem Biophys Res Commun 386, • Liang, H et al.: Genetic mouse models of extended 257-262 (2009). lifespan. Exp Gerontol 38, 1353-64 (2003). • Kim, H and Strange, K: Changes in translation rate • Lim, KL, Tan, JM: Role of ubiquitin proteasome modulate stress-induced damage of diverse proteins. system in Parkinson’s disease. BMC Biochem 8, S13 Am J Physiol Cell Physiol 305(12), (2013). (2007). • Kimball, SR, Vary, TC, Jefferson, LS: Age-dependent • Lin, MT and Beal, F: Mitochondrial dysfunction and decreases in the amount of eukaryotic initiation factor oxidative stress in neurodegenerative diseases. Nature 2 in various rat tissues. Biochem J 286, 263-8 (1992). 443(7113), 787-95 (2006). • Kim-Han, JS, Antenor-Dorsey, JA, O’Malley, KL: • Lipinski, MM et al.: Genome-wide analysis reveals Parkinsonian mimetic, MPP+, specifically impairs mechanisms modulating autophagy in normal brain mitochondrial transport in dopamine axons. J Neurosci aging and in Alzheimer’s disease. Proc. Natl. Acad. 31, 7212-7221 (2011). Sci. U S A 107, 14164-14169 (2010). • Kirkwood, TB: Understanding the odd science of • Liu, G et al.: Rogers, J, Murphy, CT, Rongo, C: EGF aging. Cell 120, 437-47 (2005). signaling activates the ubiquitin proteasome system to • Kirstein-Miles, J, Scior, A, Deuerling, A, Morimoto, modulate C. elegans lifespan. EMBO J. 30, 2990-3003 RI: The nascent polypeptide-associated complex is a (2011).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 48 • Liu, S et al.: Parkinson’s disease-associated kinase pathway targeted by D-1-methyl-tryptophan. PINK1 regulates Miro protein level and axonal OncoImmunology, 1460-1468 (2012). transport of mitochondria. PLoS Genet. 8, (2012). • Mirisola, MG et al.: Serine- and threonine/valine- • Lo Cicero, A, Stahl, PD, Raposo, G: Extracellular dependent activation of PDK and Tor orthologs vesicles shuffling intercellular messages: for good or converge on Sch9 to promote aging. PLoS Genet for bad. Curr. Opin. Cell Biol. 35, 69-77 (2015). 10(2), (2014). 1 • Loo, TW, Clarke, DM: Repair of CFTR folding • Morimoto, RI, Cuervo, AM: Proteostasis and the aging defects with correctors that function as pharmacologic proteome in health and disease. J. Gerontol. A Biol. chaperones. Methods Mol Biol 741, 23-37 (2011). Sci. Med. Sci. 69, S33-S38 (2014). • López-Otin, C, Blasco, MA, Partridge, L, Serrano, M, • Morrow, G, Samson, M, Michaud, S, Tanguay, RM: Kroemer, G: The Hallmarks of Aging. Cell 153(6), Overexpression of the small mitochondrial Hsp22 1194-1217 (2013). extends Drosophila life span and increases resistance • Lustbader, JW et al.: ABAD directly links Abeta to to oxidative stress. FASEB J 18, 598-599 (2004). mitochondrial toxicity in Alzheimer’s disease. Science • Moskalev, AA et al.: The role of DNA damage and 304, 448-452 (2004). repair in aging through the prism of Koch-like criteria. • Lutz, AK et al.: Loss of Parkin or PINK1 Ageing Res Rev., (2012) function increases Drp1-dependent mitochondrial • Mouchiroud, L et al.: The NAD(+)/sirtuin pathway fragmentation. J Biol Chem 284, 22938-22951 (2009). modulates longevity through activation of • Maddison, DC, Giorgini, F: The kynurenine pathway mitochondrial UPR and FOXO signaling. Cell 154, and neurodegenerative disease. Semin Cell Dev Biol 430-441 (2013). 40, 134-41 (2015). • Munn, DH et al.: Prevention of allogeneic fetal • Madeo, F, Tavernarakis, N, Kroemer, G: Can rejection by tryptophan catabolism. Science 281, 1191- autophagy promote longevity? Nat Cell Biol 12, 842- 1193 (1998). 846 (2010). • Munn, DH et al.: GCN2 kinase in T cells mediates • Madeo, F, Zimmermann, A, Maiuri, MC, Kroemer, G: proliferative arrest and anergy induction in response Essential role for autophagy in life span extension. J. to indoleamine 2,3- dioxygenase. Immunity 22, 633- Clin. Invest. 125, 85-93 (2015). 642 (2005). • Mair, W et al.: Lifespan extension induced by AMPK • Murphy, CT et al.: Genes that act downstream of DAF- and calcineurin is mediated by CRTC-1 and CREB. 16 to influence the lifespan of Caenorhabditis elegans. Nature 470, 404-408 (2011). Nature 424, 277-283 (2003). • Mairet-Coello, G et al.: The CAMKK2-AMPK kinase • Murphy, KG, Bloom, SR: Gut hormones and the pathway mediates the synaptotoxic effects of Abeta regulation of energy homeostasis. Nature 444, 854-59 oligomers through tau phosphorylation. Neuron 78(1), (2006). 94-108 (2013). • Nakamura, K et al.: Direct membrane association • Maji, SK et al.: Functional amyloids as natural storage drives mitochondrial fission by the Parkinson disease- of peptide hormones in pituitary secretory granules. associated protein alpha-synuclein. J Biol Chem 286, Science 325, 328-332 (2009). 20710-20726 (2011). • Malkus, KA, Ischiropoulos, H: Regional deficiencies • Nemani, VM et al.: Increased expression of in chaperone-mediated autophagy underlie alpha- alpha-synuclein reduces neurotransmitter release synuclein aggregation and neurodegeneration. by inhibiting synaptic vesicle reclustering after Neurobiol. Dis. 46, 732-744 (2012). endocytosis. Neuron 65, 66-79 (2010). • McElwee, J, Bubb, K, Thomas, JH: Transcriptional • Nicholas, A et al.: Age-related gene-specific changes outputs of the Caenorhabditis elegans forkhead protein of A-to-I mRNA editing in the human brain. Mech DAF-16. Aging Cell 2, 11-121 (2003). Ageing Dev 131, 445-447 (2010). • Mellor, AL et al.: Prevention of T cell-driven • Nuriel, T, Hansler, A, Gross, SS: Protein complement activation and inflammation by nitrotryptophan: Formation, significance and tryptophan catabolism during pregnancy. Nat Immunol identification.J Proteomics 74(11), 2300-2312 (2011). 2, 64-68 (2001). • Ocampo, JR et al.: Kynurenines with neuroactive • Mellor, AL et al.: Cutting edge: induced indoleamine and redox properties: relevance to aging and brain 2,3 dioxygenase expression in dendritic cell subsets diseases. Oxid Med Cell Longev, (2014). suppresses T cell clonal expansion. J Immunol 171, • Oh, SW et al.: Identification of direct DAF-16 targets 1652-1655 (2003). controlling longevity, metabolism and diapause by • Mesquita, A et al.: Caloric restriction or catalase chromatin immunoprecipitation. Nat Genet 38, 251- inactivation extends yeast chronological lifespan by 257 (2006). inducing H2O2 and superoxide dismutase activity. • Olzscha, H et al.: Amyloid-like aggregates sequester Proc Natl Acad Sci U S A 107, 15123-15128 (2010). numerous metastable proteins with essential cellular • Metz, R et al.: Ido inhibits a tryptophan sufficiency functions. Cell 144(1), 67-78 (2011). signal that stimulates mTOR – a novel IDO effector • Ooka, H, Segall, PE, Timiras, PS: Histology and

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 49 survival in age-delayed low-tryptophan-fed rats. Mech • Pesah, Y et al.: Drosophila parkin mutants have Ageing Dev 43(1), 79-98 (1988). decreased mass and cell size and increased sensitivity • Opitz, CA et al.: An endogenous tumour-promoting to oxygen radical stress. Development 131, 2183-2194 ligand of the human aryl hydrocarbon receptor. Nature (2004). 478(7368), 197-203 (2011). • Powers, ET, Morimoto, RI, Dillin, A, Kelly, JW, Balch, • Orentreich, N, Matias, JR, DeFelice, A, Zimmermann, WE: Biological and chemical approaches to diseases JA: Low methionine ingestion by rats extends life of proteostasis deficiency. Annu. Rev. Biochem. 78, span. J Nutr 123(2), 269-74 (1993). 959-91 (2009). • Orgel, LE: The maintenance of the accuracy of protein • Powers, RW, Kaeberlein, M, Caldwell, SD, Kennedy, synthesis and its relevance to ageing. Proc Natl Acad BK, Fields, S: Extension of chronological life span Sci U S A 49, 517-21 (1963). in yeast by decreased TOR pathway signaling. Genes Dev 20(2), 174-84 (2006). • Otera, H, Ishihara, N, Mihara, K: New insights into the function and regulation of mitochondrial fission. • Prahlad, V, Cornelius, T, Morimoto, RI: Regulation Biochim Biophys Acta 1833(5), 1256-68 (2013). of the cellular heat shock response in Caenorhabditis elegans by thermosensory neurons. Science 320, 811- • Oxenkrug, GF: The extended life span if Drosophila 14 (2008). melanogaster eye-color (white and vermilion) mutants with impaired formation of kynurenine. J Neural • Price, NL et al.: SIRT1 is required for AMPK Transm. 117(1), 23-26 (2010). activation and the beneficial effects of resveratrol on mitochondrial function. Cell Metab. 15, 675-690 • Oxenkrug, GF: Increased plasma levels of xanthurenic (2012). and kynurenic acids in type 2 diabetes. Mol Neurobiol 52(2), 805-10 (2015). • Ramsay, RR, Kowal, AT, Johnson, MK, Salach, JI, Singer, TP: The inhibition site of MPP+, the neurotoxic • Palacino, JJ et al.: Mitochondrial dysfunction and bioactivation product of 1-methyl-4-phenyl-1,2,3,6- oxidative damage in parkin-deficient mice. J Biol tetrahydropyridine is near the Q-binding site of NADH Chem 279, 18614-18622 (2004). dehydrogenase. Arch Biochem Biophys 259, 645-649 • Palikaras, K, Lionaki, E, Tavernarakis, N: Coordination (1987). of mitophagy and mitochondrial biogenesis during • Ratovitski, T et al.: Huntingtin protein interactions ageing in C. elegans. Nature 521(7553), 525-8 (2015). altered by polyglutamine expansion as determined by • Parihar, WD, Parihar, A, Fujita, M, Hashimoto, M, quantitative proteomic analysis. Cell Cycle 11(10), Ghafourifar, P: Mitochondrial association of alpha- 2006-21 (2012). synuclein causes oxidative stress. Cell Mol life Sci 65, • Rattan, SI: Synthesis, modifications, and turnover of 1272-1284 (2008). proteins during aging. Exp Gerontol 31, 33-47 (1996). • Park, C, Cuervo, AM: Selective autophagy: talking • Ravikumar, B et al.: Inhibition of mTOR induces with the UPS. Cell Biochem. Biophys. 67, 3-13 (2013). autophagy and reduces toxicity of polyglutamine • Park, CB, Larsson, NG: Mitochondrial DNA mutations expansions in fly and mouse models of Huntington in disease and aging. J Cell Biol. 193, 809-818 (2011). disease. Nat Genet 36, 585-95 (2004). • Parker, WD, Boyson, SJ, Parks, JK: Abnormalities of • Rebrin, I, Bregere, C, Kamzalov, S, Gallaher, TK, the electron transport chain in idiopathic Parkinson’s Sohal, RS: Nitration of tryptophan 372 in succinyl- disease. Ann Neurol 26, 719-723 (1989). CoA:3-ketoacid CoA transferase during aging in rat • Parrinello, S et al.: Oxygen sensitivity severely limits heart mitochondria. Biochemistry 46, 10130-10144 the replicative lifespan of murine fibroblasts. Nat Cell (2007). Biol 5, 741-7 (2003). • Reis-Rodrigues, P et al.: Proteomic analysis of age- • Parsons, P: Antagonistic pleiotropy and the stress dependent changes in protein solubility identifies theory of aging. Biogerontology 8, 613-617 (2007). genes that modulate lifespan. Aging Cell 11, 120-127 (2012). • Payne, BAI, Chinnery, PF: Mitochondrial dysfunction in aging: much progress but many unresolved • Rubinsztein, DC, Marino, G, Kroemer, G: Autophagy questions. Biochimica et Biophysica Acta 1847, 1347- and aging. Cell 146, 682-695 (2011). 1353 (2015). • Rodgers, JT et al.: Nutrient control of glucose • Pearson, SJ, Reynolds, GP: Increased brain homeostasis through a complex of PGC-1alpha and concentrations of a neurotoxin, 3-hydroxykynurenine, SIRT1. Nature 434, 113-118 (2005). in Huntington’s disease. Neurosci Lett 144, 199-201 • Russo, I, Bubacco, L, Greggio, E: Exosomes- (1992). associated neurodegeneration and progression of • Perez, VI et al.: The overexpression of major Parkinson’s disease. Am. J. Neurodegener. Dis. 1, 217- antioxidant enzymes does not extend the lifespan of 225 (2012). mice. Aging Cell 8, 73-75 (2009). • Salminen, A, Kaarniranta, K, Kauppinen, A: • Perier, C, Vila, M: Mitochondrial biology and Inflammaging: disturbed interplay between autophagy Parkinson’s disease. Cold Spring Harb Perspect Med and inflammasomes.Aging 4, 166-175 (2012). 2(2), (2012). • Sattlegger, E and Hinnebusch, AG: Separate domains

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 50 in GCN1 for binding protein kinase GCN2 and resolution in atherosclerosis. Nat Rev Immunol. 10, 36- ribosomes are required for GCN2 activation in amino 46 (2010). acid-starved cells. EMBO J 19(23), 6622-33 (2000). • Talens, RP et al.: Epigenetic variation during the • Schaar, CE et al.: Mitochondrial and cytoplasmic ROS adult lifespan: cross-sectional and longitudinal data have opposing effects on lifespan. PLoS Genet 11(2), on monozygotic twin pairs. Aging Cell 11, 694-703 (2015). (2012). 1 • Schagger, H, Pfeiffer, K: the ratio of oxidative • Tang, DG, Tokumoto, YM, Apperly, JA, Lloyd, AC, phosphorylation complexes I-V in bovine heart Raff, MC: Lack of replicative senescence in cultured mitochondria and the composition of respiratory chain rat oligodendrocyte precursor cells. Science 291, 868- supercomplexes. J Biol Chem 276, 37861-37867 71 (2001). (2001). • Tatar, M et al.: A mutant Drosophila insulin • Schieber, M, Chandal, NS: TOR signaling couples receptor homolog that extends life-span and impairs oxygen sensing to lifespan in C. elegans. Cell Rep neuroendocrine function. Science 292, 107-110 9(1), 9-15 (2014). (2001). • Schneider, JL et al.: Loss of hepatic chaperone- • Tatsuta, T: Protein quality control in mitochondria. J mediated autophagy accelerates proteostasis failure in Biochem 146, 455-461 (2009). aging. Aging Cell 14, 249-264 (2015). • Tavernarakis, N: Protein synthesis and aging: eIF4E • Schon, EA, Przedborski, S: Mitochondria: The next and the soma vs. germline distinction. Cell Cycle 6, (neurode)generation. Neuron 54, 823-827 (2011). 1168-1171 (2007). • Schwarcz, R, Bruno, JP, Muchowski, PJ, Wu, • Tavernarakis, N: Ageing and the regulation of protein HQ: Kynurenines in the mammalian brain: when synthesis: a balancing act? Trends Cell Biol 18, 228- physiology meets pathology. Nat Rev Neurosci 13(7), 235 (2008). 465-77 (2012). • Tieu, K et al.: D-beta-hydroxybutyrate rescues • Scialò, F et al.: Mitochondrial ROS produced via mitochondrial respiration and mitigates features of reverse electron transport extend animal lifespan. Cell Parkinson’s disease. J Clin Invest 112, 892-901 (2003). Metab. 23(4), 725-34 (2016). • Ting, KK et al.: Effect of quinolinic acid on human • Segref, A et al.: Pathogenesis of human mitochondrial astrocytes morphology and functions: implications in diseases is modulated by reduced activity of the Alzheimer’s disease. J Neuroinflamm 6, 36 (2009). ubiquitin/proteasome system. Cell Metab. 19(4); 642- • Van der Goot, AT et al.: Delaying aging and the aging- 52 (2014). associated decline in protein homeostasis by inhibition • Shaw, AC, Joshi, S, Greenwood, H, Panda, A, Lord, of tryptophan degradation. Proc Natl Acad Sci U S A JM: Aging of the innate immune system. Curr Opin 109(37), 14912-7 (2012). Biol 8, 703-713 (2010). • Van der Goot, AT and Nollen, EAA: Tryptophan • Smith, CD et al.: Excess brain protein oxidation and metabolism: entering the field of aging and age-related enzyme dysfunction in normal aging and Alzheimer pathologies. Trends Mol Med 19(6), 336-44 (2013). disease. Proc. Natl Acad. Sci. U S A 88, 10540-10543 • Vanhooren, V et al.: Protein modification and (1991). maintenance systems as biomarkers of ageing. Mech. • Song, DD, Shults, CW, Sisk, A, Rockenstein, E, Ageing Dev, (2015). Masilah, E: Enhanced substantia nigra mitochondrial • Van Raamsdonk, JM, Hekimi, S: Deletion of the pathology in human alpha-synuclein transgenic mice mitochondrial superoxide dismutase sod-2 extends after treatment with MPTP. Exp Neurol 186, 158-172 lifespan in Caenorhabditis elegans. PLoS Genet 5(2), (2004). (2009). • Sono, M, Roach, MP, Coulter, ED, Dawson, JH: • Van Remmen, H et al.: Life-long reduction in MnSOD Heme-containing oxygenases. Chem Rev 96, 2841- activity results in increased DNA damage and higher 2887 (1996). incidence of cancer but does not accelerate aging. • Strogolova, V, Furness, A, robb-McGrath, M, Garlich, Physiol Genomics 16, 29-37 (2003). J, Stuart, RA: Rcf1 and rcf2, members of the hypoxia- • Vargas, R and Castaneda, M: Age-dependent decrease induced gene 1 protein family, are critical components in the activity of protein-synthesis initiation factors in of the mitochondrial cytochrome bc1-cytochrome c rat brain. Mech Aging Dev 21, 183-91 (1983). oxidase supercomplex. Mol Cell Biol 32, 1363-1373 (2012). • Veatch, JR, McMurray, MA, Nelson, ZW, Gottschling, DE: Mitochondrial dysfunction leads to nuclear • Su, AI, et al.: A gene atlas of the mouse and human genome instability via an iron-sulfur cluster defect. protein-encoding transcriptomes. Proc Natl Acad Sci Cell 137, 1247-1258 (2009). U S A 101, 6062-67 (2004). • Vellai, T et al.: Genetics: influence of TOR kinase on • Swindell, WR, et al.: Endocrine regulation of heat lifespan in C. elegans. Nature 426, 620 (2003). shock protein mRNA levels in long-lived dwarf mice. Mech Ageing Dev. 130, 393-400 (2009). • Verstreken, P et al.: Synaptic mitochondria are critical for mobilization of reserve pool vesicles at Drosophila • Tabas, I: Macrophage death and defective inflammation neuromuscular junctions. Neuron 47, 365-378 (2005).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 51 • Vila, M, Bové, J, Dehay, B, Rodriguez-Muela, N, of tryptophan in coronary heart disease. Eur J Clin Boya, P: Lysosomal membrane permeabilization in Invest 33, 550-554 (2003). Parkinson’s disease. Autophagy 7, 98-100 (2011). • Wu, Z, Song, L, Liu, SQ, Huang, D: Independent and • Vinuela, A, Snock, LB, Riksen, JA, Kammenga, JE: additive effects of glutamic acid and methionine on Genome-wide gene expression regulation as a function yeast longevity. PLoS One 8(11), (2013). of genotype and age in C. elegans. Genome Res 20, • Wullschleger, S et al.: TOR signaling in growth and 929-937 (2010). metabolism. Cell 124, 471-484 (2006). • Voth, W et al.: The protein targeting factor Get3 • Yamada, M, Iwatsubo, T, Mizuni, Y, Mochizuki, H: functions as ATp-independent chaperone under Overexpression of alpha-synuclein in rat substantia oxidative stress conditions. Mol Cell 56, 116-127 nigra results in loss of dopaminergic neurons, (2014). phosphorylation of alpha-synuclein and activation of • Walker GA, Lithgow, GJ: Lifespan extension in C. caspase-9: Resemblance to pathogenetic changes in elegans by a molecular chaperone dependent upon Parkinson’s disease. J Neurochem 91, 451-461 (2004). insulin-like signals. Aging Cell 2, 131-139 (2003). • Yanos, ME, Bennett, CF, Kaeberlein, M: Genome- • Wallace, DC: A mitochondrial paradigm of metabolic wide RNAi longevity screens in Caenorhabditis and degenerative diseases, aging, and cancer: a dawn elegans. Curr Genomics 13(7), 508-18 (2012). for evolutionary medicine. Annu. Rev. Genet 39, 359- • Yee, C, Yang, W, Hekimi, S: The intrinsic apoptosis 407 (2005). pathway mediates the pro-longevity response to • Walther, DM et al.: Widespread proteome remodeling mitochondrial ROS in C. elegans. Cell 157 (4), 897- and aggregation in aging C. elegans. Cell 161(4), 919- 909 (2014). 32 (2015). • Yoshida, R et al.: Induction of pulmonary indoleamine • Wang, K, Klionsky, DJ: Mitochondria removal by 2,3-dioxygenase by interferon. Proc Natl Acad Sci U S autophagy.Autophagy 7, 297-300 (2011). A 78, 129-132 (1981). • Wang, X, Gerdes, HH: Transfer of mitochondria via • Youle, RJ, Narendra, DP: Mechanisms of mitophagy. tunneling nanotubes rescues apoptotic PC12 cells. Cell Nat Rev. Mol Cell Biol 12, 9-14 (2011). Death Differ. 22, 1181-1191 (2015). • Yuasa, HJ and Ball, HJ: Efficient tryptophan- • Warburg, O: On the origin of cancer cells. Science 123, catabolizing activity is consistently conserved through 309-314 (1956). evolution of TDO enzymes, but not IDO enzymes. J. • Webster, GC, Webster, SI: Decline in synthesis of Exp. Zool. (Mol. Dev. Evol.), (2015). elongation factor one (EF-1) precedes the decreased • Yun, J and Finkel T: Mitohormesis. Cell Metab 19(5), synthesis of total protein in aging Drosophila 757-66 (2014). melanogaster. Mech Aging Dev 22, 121-8 (1983). • Zarse, K et al.: Impaired insulin/IGF1 signaling • Weidner, AM et al.: RNA oxidation adducts 8-OHG extends life span by promoting mitochondrial and 8-OHA change with Abeta42 levels in late-stage L-proline catabolism to induce a transient ROS signal. Alzheimer’s disease. PLoS ONE 6, (2011). Cell Metab 15(4), 451-65 (2012). • Widner, B et al.: Degradation of tryptophan in • Zhang, Y et al.: Mice deficient in both Mn superoxide neurodegenerative disorders. Adv Exp Med Biol 467, dismutase and glutathione peroxidase-1 have increased 133-138 (1999). oxidative damage and a greater incidence of pathology • Williams, GC: Pleiotropy, natural selection, and the but no reduction in longevity. J Gerontol A Biol Sci evolution of senescence. Evolution 11, 398 (1957). Med Sci 64, 1212-1220 (2009). • Williamson, D: The curious history of yeast • Zhou, C, Huang, Y, Przedborski, S: Oxidative stress in mitochondrial DNA. Nat Rev Genet 3(6), 475-81 Parkinson’s disease: a mechanism of pathogenic and (2002). therapeutic significance. Ann NY Acad Sci 1147, 93- 104 (2008). • Winklhofer, KF: The parkin protein as a therapeutic target in Parkinson’s disease. Expert Opin Ther Targets • Zwilling, D et al.: Kynurenine 3-monooxygenase 11, 1543-52 (2007). inhibition in blood ameliorates neurodegeneration. Cell 145, 864-874 (2011). • Wirleitner, B et al.: Immune activation and degradation

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 52 1

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 53 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 54 Chapter II

Genetic screens in Caenorhabditis elegans models for neurodegenerative diseases

Olga Sin1,2, Helen Michels1, Ellen A.A. Nollen1

1European Research Institute for the Biology of Aging, University of Groningen, University Medical Centre Groningen, Laboratory of Molecular Neurobiology of Aging, Antonius Deusinglaan 1, 9700 AD Groningen, The Netherlands 2Graduate Program in Areas of Basic and Applied Biology, Abel Salazar Biomedical Sciences Institute, University of Porto, 4099-003 Porto, Portugal

Sin O, Michels H, Nollen EA. Genetic screens in Caenorhabditis elegans models for neurodegenerative diseases. Biochim Biophys Acta. 2014 Oct; 1842(10): 1951-1959.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 55 Caenorhabditis elegans comprises unique features that make it an attractive model organism in diverse fields of biology. Genetic screens are powerful to identify genes and C. elegans can be customized to forward or reverse genetic screens and to establish gene function. These genetic screens can be applied to “humanized” models of C. elegans for neurodegenerative diseases, enabling for example the identification of genes involved in protein aggregation, one of the hallmarks of these diseases. In this review, we will describe the genetic screens employed in C. elegans and how these can be used to understand molecular processes involved in neurodegenerative and other human diseases.

This article is part of a Special Issue entitled: From Genome to Function.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 56 INTRODUCTION

1.1. Caenorhabditis elegans as a model organism Sydney Brenner first introduced the nematode C. elegans as a genetic model organismin 1965 and since then the model has been extensively used in very 2 diverse fields of research, from developmental biology to ecotoxicology, aging and neuroscience [1]. This has resulted in several breakthroughs in biomedical science, which include the discovery of genetic regulators of programmed cell death, the use of the green fluorescent protein as a protein marker, and the discovery of RNA interference. Indeed, this nematode combines a number of characteristics that make it an advantageous model, anatomically and genetically, which are summarized in Box 1. Moreover, the characteristics of this invertebrate make it an easy experimentalmodel to study biological processes in a relatively cheap, quick, and easy way.

C. elegans is a small, free-living nematode of about 1–1.5 mm in length that can be found in temperate soil environments feeding on different bacteria, including Escherichia coli. It exists in two sexual forms, as a hermaphrodite or as a male. The former is self-fertile, able to produce its own sperm and eggs and is the predominant adult form. Although males are rare (about 0.02%), their abundance in the offspring can be increased to 50% by mating with hermaphrodites [2]. The length of the life

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 57 cycle of wild type N2 C. elegans strains and its lifespan depends on the growth temperature. Grown at 20 °C, hermaphrodites usually lay 300–350 eggs and once the eggs hatch, it takes about three days to develop from a larva to an adult. The average lifespan of this organism can vary between 18 and 20 days [3,4]. At higher temperatures, the life cycle is shortened and the lifespan decreased. One major advantage of C. elegans is that it has a well-dissected and predetermined anatomy. The adult hermaphrodite has exactly 959 somatic cells and 302 neurons [1,5,6]. Its transparent body enables one to easily follow cell fate or expression of fluorescently tagged proteins of interest in the living animal. Moreover, C. elegans was the first multicellular organism to have the complete genome sequenced and this gave rise to several databases and resources that are currently available online for the scientific community ([7], see “Online links” at the end of this article).

Genetic screens are widely used in C. elegans to discover gene function. It can be easily applied to discover which gene mutations are responsible for a specific phenotype of interest (forward genetics) or, conversely, the gene function can be purposely altered to assess what is the consequence in terms of development, behavior or alterations in specific biological processes (reverse genetics). The two major genetic screens employed are ethyl methane sulfate (EMS) screens and (genome-wide) RNAi screens. They have been fundamental not only to dissect nematode genetics but also to identify genes involved in aging, development, DNA damage response, and signal transduction, amongst other biological processes [8–13].

1.2. C. elegans homology to humans For the scope of this review, we shall explore the rationale and the basic procedures for both methods not only highlighting their advantages but also pinpointing the drawbacks. Next, we will explore how genetic screens can help us gain insight into the molecular and cellular mechanisms of human diseases. Specifically, we will focus on the application of genetic screens to discover potential disease-modifier genes by exemplifying studies on C. elegans models for neurodegenerative diseases.

There are a significant number of proteins that are evolutionary conserved between C. elegans and humans. At the time that the C. elegans genome sequencing was complete, 36% of C. elegans proteins (from a set composed of 18,891 protein sequences) were found to have homologs in humans (set composed of 4979 protein sequences), by pairwise comparison (The C. elegans Sequencing Consortium, 1998 [7]). Thereafter, this percentage was increased to 83% due to the much larger human gene dataset available to perform the comparison [14]. A more recent study estimated that 38% of the 20,250 C. elegans protein coding genes had unique corresponding functional orthologs in humans (7663 unique hits) [15]. In a nutshell, biological

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 58 processes unraveled in the invertebrate C. elegans can provide insight into human biology.

2. Genetic screens Genetic screens in C. elegans are well-established and commonly used to assess gene function in any biological process of interest. High-throughput (semi-) automatized 2 setups and screening methods enable hundreds of parallel experiments in microtiter plates. In a screen, wild type animals are mutagenized or treated with RNAi and then scored for phenotypical changes. Below we describe two types of genetic screens that are most frequently used: EMS mutagenesis and RNA interference (RNAi). The characteristics of both type of screens are summarized in Table 1.

2.1. EMS mutagenesis The most commonly used method to mutate the genome of C. elegans is the treatment with EMS. The mutagen induces mutations in the sperm and oocytes of hermaphrodites. Sydney Brenner tested systematically different mutagens, but researchers are mostly using EMS because of its relatively low toxicity and relatively good efficiency (summarized in [16]). The hermaphroditism of C. elegans allows easy maintenance of a mutation as a homozygous worm will pass it to all the progeny through self-fertilization.

Mutations can be identified using a simple F2 screen firstly described by Brenner in 1974 [1]. Thousands of copies of any particular gene can be analyzed in a typical EMS screen. The frequency of a null mutation at any particular locus of the genome is one for every 2000 copies by using standard concentrations (50 mM) of the mutagen. That means that one can expect to identify 6mutations per particular gene in a typical experiment of 12,000 haploid genomes (reviewed in [16]). The mutagenized worms

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 59 are placed on Petri dishes and grown for two generations to produce homozygous mutants (Fig. 1). Worms from the F2 generation showing a specific phenotype of interest are further singled to new plates to determine whether the phenotype is transmitted to the next generation.

Once a worm with a specific phenotype is isolated, the responsible mutation needs to be identified. By using single nucleotide polymorphisms (SNPs) of the Hawaiian wild type strain in comparison to the Bristol strain (natural variation wild type) it is possible to map a mutation first to a certain chromosome [17] and then in several steps to a specific region on that chromosome. When a mutation is mapped to a gene region, sequencing or the specific knockdown of every single gene in that area by RNAi can be used to identify the mutated gene. The development of new sequencing

Fig. 1. High-throughput mutagenesis screen in C. elegans: Animals (P0) are treated with a mutagen (e.g. EMS) to produce progeny (F1) containing mutations in alleles of different genes. These animals get progeny (F2) of different genotypes by self-fertilization which are further scored for a specific phenotype. Positive evaluated animals are singled to validate a homozygous mutation by breeding through of the phenotype to the next generation (F3). Once a mutant is isolated, the mutation site needs to be mapped to a specific genomic site to continue with functional studies (adapted from [16]).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 60 methods like deep sequencing in the last decade facilitates the identification of mutations and can save laborious fine mapping [18–20]. It is important to keep in mind that an isolated, mutated animal can have several mutations at different loci. For further interpretation, it is therefore necessary to backcross the animals several times with wild type strains. The importance for controlling the genetic background was shown by Burnett and colleagues. They demonstrated that a described lifespan 2 extension [21] by overexpression of SIR-2 disappeared after several backcrossings [22]. Tissenbaum and Guarente further showed that the overexpression of SIR-2.1 slightly increases lifespan but to a much lesser extent than the transgenic animals used in their first publication [23]. Deep sequencing methods can be used to monitor mutations in the background. Preferentially, one should use independent mutant or deletion alleles of a gene to confirm results.

In the first EMS screens, 619 mutants were identified with visible phenotypes especially from the uncoordinated class [1]. This group of genes impairs wild type movements when mutated. Under laboratory conditions proper moving is not essential as food is plentiful and sex is dispensable so therefore maintenance and characterization of mutants that may not survive in non-laboratory conditions are possible. Many of these mutants have revealed important information about molecules and mechanisms involved in human disease. For example, one gene of the uncoordinated class is unc-2 (uncoordinated 2) and encodes for a homolog of the voltage-sensitive calcium-channel alpha-1 subunit (human P/Q calcium channel CACNA1A) [24]. Missense mutations in the CACNA1A calcium channel in humans are associated with a rare form of migraine [25] which is often associated with low levels of serotonin [26]. unc-2 mutants show neuronal migration defects similar to serotonin-deficient mutants [27] and UNC-2 is required for the desensitization to the two neurotransmitters dopamine and serotonin [28]. Studies using C. elegans showed that UNC-2 interacts with the transforming growth factor (TGF)-β, a pathway that is required for movements through regulation of serotonin levels probably through the modulation of the expression of tph-1 (tryptophan hydroxylase), the enzyme that converts tryptophan into serotonin [29]. There are elevated levels of TGF-β1 in migraine patients compared to those of pain-free individuals [30].

In addition to mutations in the uncoordinated class, Brenner also identified mutants with aberrant appearance like animals with small bodies, blistered cuticles, twitching muscles, rolling locomotion, long bodies, dumpy bodies, forked heads or bent heads [1].

EMS screens are often used to identify different mutations with the same phenotype to further investigate if those genes function in the same processes. Using this approach, John Sulston and H. Robert Horvitz searched e.g. for mutants that show defects in the differentiation of avulva from epidermal cells [31]. Molecular follow-

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 61 up studies revealed that animals that lack a vulva had mutations in two signaling pathways: the epidermal growth factor (EGF)/RAS pathway and the Notch signaling pathway (reviewed in [32–34]), both having major roles in cell fate determination. These studies in C. elegans have increased the understanding of these molecular pathways involved in oncogenesis in humans (reviewed in [35,36]).

Another possibility to find genes of the same genetic pathways are enhancer or suppressor screens. In this case the mutagenesis occurs on a non-wild type strain whose genetic composition is known and causes a defined phenotype. Like this, one can screen for mutations in this genetic background that enhance or suppress (reverse) that phenotype. With this approach one is able to show that two genes not only act in the same pathway but also their hierarchy which means that one is acting upstream of the other (summarized in [16]). However, one should still keep in mind that it might also be possible that some proteins result in the same phenotype when mutated even though they do not necessarily function in the same pathway.

Although EMS mutagenesis is a powerful tool to generate a high number of mutations and high-throughput screens to identify mutants with a specific phenotype it also has some limitations. It has been estimated that about 30% of the genes in C. elegans can be mutated to a visible phenotype [37] (some mutations might result e.g. in a lethal phenotype as it is the case for a number of developmental genes) and it needs to be mentioned that the identification of the same mutations which indicates a saturation of the screen is no guarantee that some other genes might be missed in this screen. High-throughput screens are only a starting point for further detailed experiments at molecular levels.

2.2. RNA interference RNA interference was first discovered and investigated in C. elegans and published in 1998 by Andrew Fire et al. [38] (Nobel Prize in Physiology and Medicine in 2006). The discovery of dsRNA-mediated gene silencing has revolutionized genetic studies in C. elegans, as well as in other model organisms. Similar to EMS screens, RNAi screens can be used to identify genes that, when depleted, result in a certain phenotype or enhance or suppress a mutant phenotype.

RNAi in C. elegans is systemic, which, to date, is not the case for any other animal models. Therefore it is sufficient to introduce dsRNA into one specific tissue to get RNA silencing also in distant cells because of an amplification process called transitive RNAi [39]. This systemic effect is advantageous for large-scale genome- wide RNAi screens in C. elegans.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 62 There are different methods that describe how to silence gene expression in C. elegans. The dsRNA can be delivered into the worm by (1) injection into any region [38], (2) feeding with dsRNA-producing bacteria [40], (3) soaking in dsRNA [41] or (4) in vivo production of dsRNA from transgenes under the control of specific promotors [42]. Cell-specific factors seem to regulate thereby the entry and export of dsRNA [39,43,44]. Some cell types (e.g. neurons) seem not to respond well to 2 systemically delivered RNAi [42]. The use of RNAi enhanced mutants (e.g. eri- 1 mutant or mutants of the retinoblastoma pathway that are described to enhance RNAi especially in nervous tissue) might circumvent this problem[45,46]. In addition, Calixto and colleagues generated transgenic animals overexpressing the transmembrane protein SID-1 which is an essential component for systemic RNAi in the neurons. This modification increased the response to dsRNA delivered by feeding. It seemed that the expression of SID-1 in the neurons, on the other hand, decreased the RNAi effect on non-neuronal cells which might be useful for studying the function of essential genes in the neurons. This effect could be even increased when using a sid-1 mutant background [47]. Durieux and colleagues further used this mutant that is insensitive for systemic RNAi to investigate the knockdown of one of the cytochrome c oxidase-1 subunits in specific tissues by controlling the expression of dsRNA via tissue-specific promotors [48]. The tissue specific expression of SID-1 (not only in neurons) in a sid-1 knockout background probably also enables to study tissue-specific effects especially of essential genes.

Especially the possibility to feed animals with dsRNA-producing bacteria enables to perform high-throughput RNAi screens in C. elegans (Fig. 2) [49,50]. For efficient induction of RNA interference the choice of the dsRNA-coding region is essential. In C. elegans, long dsRNA fragments (more than 100 bp) trigger gene silencing via RNAi. For most genes dsRNA is about 200–1000 nucleotides or even longer and covers exon-regions of the targeted gene. The fragment should only target one gene. Once the coding region is chosen it can be cloned into a specific vector encoding the production of the specific dsRNA (summarized in [51]). The L4440 vector contains two bacteriophage T7 RNA polymerase promotors flanking the multiple cloning site in which the cDNA of a specific gene has been inserted. The construct can be transformed into E. coli strain HT115 (ED3). This strain is deficient for the bacterial RNA polymerase III and its production of bacteriophage T7 polymerase from the construct can be induced by the addition of isopropyl β-D-1- thiogalactopyranoside (IPTG). The bacteria are then synthesizing two complementary RNA strands that form a duplex RNA which can mediate RNAi [38].

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 63 Fig. 2. High-throughput RNAi screen in C. elegans: Age-synchronized animals are transferred to microtiter plates containing different clones of HT115 E. coli bacteria. Every clone produces a specific dsRNA which is taken up by the nematodes and induces a knockdown of the corresponding gene. Positive hits in the phenotypic screen are finally confirmed by sequencing the bacterial clone and repeating the specific knockdown in single experiments. Starting point of the RNAi feeding (possible at any developmental stage) and time point of the phenotypic scoring depend on the experiment setup.

RNAi libraries are commercially available which includes one library of bacteria clones containing cDNAs of 17,575 genes which represents about 87% of the C. elegans genome[49] and one library including clones of 11,800 C. elegans genes [52]. Positive scored clones subsequently can be sequenced to confirm that they target the predicted gene. To prevent any further off-target effects of the dsRNA and therefore false-positive results, one should consider generating a second RNAi construct targeting the mRNA of the same gene [52]. Besides the coding regions, the 3′UTR of mRNA might as well be a suitable target as RNA localization elements for the transport of the mRNA or regulation elements of eukaryotic gene expression are typically located in this region (summarized in [53, 54]).

Dissolving adult hermaphrodites with hypochloride (bleach) will yield only fertilized eggs and can be used to age-synchronize the animals for a screen.

Gene knockdown by RNAi can be induced at different developmental stages in contrast to EMS mutagenesis, which generates stable mutations that are present at all stages. Thus, the examination of the function of a gene that is transcribed at different developmental stages is possible. This is especially interesting when an active gene is essential at early developmental stages [55]. To investigate the effect of gene depletion by RNAi during embryonic development it is necessary to feed the parental worm with the specific bacterial strain.

Another possibility to study the effect of gene depletion at a certain timepoint was previously described by Calixto and colleagues, by performing a temperature-

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 64 sensitive conditional knockdown. They could induce the knockdown of a gene by controlling via temperature the expression of RDE-1, a C. elegans argonaut protein which is required for RNA interference. This resulted in active RNAi at 15 °C but not at 25 °C. Furthermore, they observed that the switching ON and OFF is much faster than transferring animals from RNAi-mediating bacteria to non-RNAi inducing bacteria and vice-versa [56]. 2 The dilution of bacteria containing a specific RNAi construct with non-RNAi mediating bacteria may decrease the efficiency of knocking down a certain gene. In that case of mild RNA interference, lethality effects and other very strong phenotypes are reduced and it might still be possible to study the function of these special genes.

The target of RNAi is known. This is one major difference to EMS mutagenesis that cannot be directed to specific genes. Therefore, besides genome-wide RNAi screens, one can also screen in a smaller subset of candidate genes for example based on previous microarray data, GWAS data, interactome studies, etc. Colaiácovo and colleagues performed for example a RNAi screen to check for germline phenotypes in a subset of genes that were generated by a previous microarray analysis of Reinke et al. that was focusing on germline-enriched gene expression [57,58]. In another screen we were looking for genes that, when knocked down, increased the number of alpha-synuclein inclusions in a Parkinson’s disease C. elegans model [59]. This group of 80 genes was then further used for a second RNAi screen in order to find candidates that, when knocked down, induced motility changes in the disease background [60].

RNAi efficiency of bacterial clones in the library can differ. Whereas some dsRNAs induce gene silencing closely to a knockout of a gene, others only generate a mild knockdown. One should always be aware that RNAi is only silencing gene activity and that it is not a full knockout of a gene. It is estimated that about 10–30% of candidates are scored as false negatives as the RNAi is not efficient enough to result in an obvious phenotype. On the other hand the percentage of false positives is relatively low (0.4%) [61]. It is also important to keep in mind that RNA interference is acting at the mRNA level and therefore only influencing the expression of a protein. That means that the stability of a protein is highly influencing the RNAi effect as already generated proteins and their activity are not affected anymore.

Results can also differ from one experiment to the other using the same bacteria clone to silence a specific gene. For example the freshness of the material can be crucial (IPTG, Ampicillin, RNAi construct containing bacteria) [51]. In contrast, a knockout mutant e.g. by EMS mutagenesis has the advantage that it results in a stable genotype. Results of the RNAi screen should therefore be confirmed by single experiments with the candidate genes, preferentially with a gene mutant strain.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 65 A clear phenotype for scoring is mandatory for any successful screen. An obvious easy-to-recognize phenotype as well as automation of scoring facilitates the screening process.

The small size of the animals, the variety of simple phenotypes that are often results of one single gene disruption or silencing, the hermaphroditic reproduction, the homology to higher organisms (see above) and the knowledge of the C. elegans genome, cell-distribution and nematode anatomy make this animal an optimal model organism to identify the function of genes via any kind of high-throughput screen.

3. From genome to function: what have genetic screens taught us? One of the advantages of C. elegans is that it is amenable to generate “humanized” models of human diseases. For the purpose of this review, we will describe as an example C. elegans models of neurodegenerative diseases. Neuropathological hallmarks found in the human brain can be successfully recapitulated in the nematode, such as protein aggregation [62]. Indeed, one of the common features in neurodegenerative diseases is the presence of protein aggregates in the brains of affected patients. These structures originate from protein misfolding and aggregation of so-called “aggregation-prone proteins”. To name a few, these can be the amyloid- beta in Alzheimer’s disease (AD), mutant huntingtin in Huntington’s disease (HD) and alpha-synuclein in Parkinson’s disease (PD) [63]. By mechanisms that are still to be unraveled, these aggregation-prone proteins adopt a distinct conformation, which is thought be a toxic gain-of-function [64,65]. The general understanding is that aggregation (or inclusion formation) renders cellular protection by sequestering misfolded proteins, therefore preventing potentially toxic protein–protein interactions [65,66].

Several nematode models have been generated to recapitulate molecular aspects of diseases, including HD, PD, AD and muscular dystrophy [59,67–72]. Although they do not feature clinical aspects of the disease, they provide the means to understand the molecular mechanisms in these diseases. Genetic screens performed in some of these models represent quick, unbiased methods that have enabled insights into the underlying mechanisms of neurodegeneration. Indeed, many of the disease modifiers discovered in C. elegans were found to be reproducible in human cell-based models and other animal models such as mice, strengthening the validity of using this small organism to study complex human diseases, as summarized in Table 2 [69,73–78].

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 66 2

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 67 3.1. C. elegans models for polyglutamine diseases Polyglutamine diseases comprise a subset of neurodegenerative disorders that include HD, spinocerebellar ataxias (−1, −2, −6, −7, −17), Machado–Joseph disease (also know as spinocerebellar ataxia 3) and spinobulbar muscular atrophy [79]. The common characteristic of polyglutamine diseases is an abnormal expansion of CAG triplets (which encode glutamine) in the coding region of the disease gene. Although the length of the CAG repeatmay vary fromindividual to individual, the threshold to develop disease is around 40 CAG repeats (except for SCA6), which cause a polyglutamine expansion in the protein that is prone to aggregate. The larger the CAG repeat the earlier onset will occur and the more severe the disease phenotype will be. A more detailed and complete information on polyglutamine diseases is reviewed elsewhere [79].

In C. elegans, several different models have successfully recapitulated protein aggregation. Similarly to what occurs in humans, the length of the CAG repeats also determines the aggregation phenotype in C. elegans. At least three models have been generated to induce polyglutamine associated toxicity in neurons by expressing expanded polyglutamine stretches in ASH sensory neurons, touch receptor neurons or the entire nervous system of C. elegans [68,80–82]. Polyglutamine aggregation has been modeled in the body wall muscle cells of C. elegans [68]. In this model, expanded polyglutamine stretches are fused to a yellow fluorescent protein (YFP) under the unc-54 promotor, which is specific to the body wall muscle. The aggregation and toxicity phenotype is polyglutamine length-dependent. As the animal ages, the accumulation of protein aggregates increases, which is associated with toxicity [68]. This model has been widely used for genetic screens to discover enhancers and/ or suppressors of polyglutamine proteotoxicity. Two genome-wide RNAi screens revealed modifier genes and classified them according to their biological function [83,84]. In the first screen, Q35 animals were fed with dsRNA-producing bacteria and scored for genes that, when downregulated, provoked premature polyglutamine aggregation [83]. The major functional classes included RNA synthesis and processing, protein synthesis, folding, transport and degradation and components of the proteasome. In the second screen, the authors sought for genes that drive aggregation in Q35 animal and therefore the selection was made for genes that suppressed polyglutamine induced aggregation when downregulated [84]. With this study, a new subset of modifier genes was recently found to belong to broader biological functions, namely cell cycle, cell structure, protein transport and energy and metabolism [83,84]. Therefore, proteotoxicity is not derived only from protein- related processes but rather a more diverse spectrum of biological functions that also have an effect on protein misfolding and aggregation. Interestingly, nine of these recently identified modifier genes were able to fold misfolded proteins back into the

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 68 native state when constitutively expressed in misfolding mutants [84].

Forward genetics have also been used to identify modifiers of proteotoxicity. One such screen consisted in treating Q40-expressing worms with EMS. The aim was to find positive regulators of aggregation by selecting genes that suppressed protein aggregation when chemically mutated by EMS. The screen revealed MOAG-4 (modifier of aggregation) as a general aggregation-promoting factor in polyglutamine, 2 Parkinson’s and Alzheimer’s disease models [76]. Inactivating MOAG-4 alleviated from polyglutamine-induced aggregation and toxicity; moreover, this effect was functionally conserved in the human orthologs SERF1A and SERF2. A recent follow-up on one of these orthologs showed that SERF1A is a specific aggregation promoting factor, since it was able to bind specifically to amyloidogenic proteins, including alpha-synuclein, prion protein, amyloid-beta and huntingtin, but not to non-amyloidogenic proteins [85].

Genetic screens have also been used to find regulators of proteotoxicity using the C. elegans neuronal system. An RNAi screen performed in a C. elegans model expressing 128 polyQ stretches in the touch receptor neurons resulted in 662 genes that either enhanced or suppressed neuron toxicity, as measured by loss of touch response [77]. Comparison of these disease modifier genes to gene expression data in two mouse models of HD showed that there was an overlap of 49 genes that were dysregulated in the striatum of either model, emphasizing the power of using C. elegans to find novel regulators of proteotoxicity relevant in human diseases.

3.2. C. elegans models for Parkinson’s disease Parkinson’s disease (PD) is the second most common neurodegenerative disease (after Alzheimer’s disease) that affects 1% of the population over the age of 50. Clinically, it is characterized by resting tremors, rigidity, bradykinesia and postural instability [86,87]. The defects in the motor system result from the progressive loss of dopaminergic neurons in the substantia nigra pars compacta (SNpc), which project and innervate the neurons in the caudate and putamen. Consequently, there is a reduction of dopamine levels, which is the neurotransmitter that plays a role in the coordination of body movements. Besides motor disabilities, PD patients can experience non-motor symptoms such as autonomic dysfunction, sleep disturbances and neuropsychiatric symptoms [88]. Most cases of PD are sporadic (about 95%) with unknown etiology. It has been suggested that disease can result from the accumulation of toxins (pesticides and heavy metals) over the years. Only 5% of PD has a familial origin and is associated with genetic mutations [88]. However, there are neuropathological hallmarks common to both sporadic and familial forms of PD. These are the loss of dopaminergic neurons in the SNpc, that result from

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 69 the degeneration of the nigrostriatal pathway which leads to the motor symptoms described earlier as well as the formation of intraneuronal protein aggregates known as Lewy bodies and Lewy neurites in the surviving neurons, which contain alpha- synuclein.

Alpha-synuclein is a small (140 amino acids) natively soluble, monomeric protein that is predominantly expressed in the brain and is enriched in presynaptic terminals [89]. Although the precise function of this protein remains unclear, it is thought to be involved in the regulation of dopamine neurotransmission, vesicular trafficking and modulation of synaptic function and plasticity [90–92]. Three different mutations in the alpha-synuclein gene (A53T, A30P and E46K) cause autosomal-dominant PD [93–95] and genomic duplications and triplications of the gene have also been identified; suggesting that overproduction of wild type alpha-synuclein is sufficient to cause disease [96,97].

Genetic screens performed with this model have been supporting an important relationship between alpha-synuclein and vesicle transport. The “humanized” model of C. elegans for PD expresses the human alpha-synuclein fused to YFP in the body wall muscle. Phenotypically, immobile YFP-positive foci can be seen in the muscle cells and these foci increase in number and correlate with age-dependent toxicity. An unbiased genome-wide RNAi screen with this model showed 80 modifier genes that, when suppressed, provoked premature alpha-synuclein inclusion formation [59]. A follow-up on those modifier genes revealed tdo-2, a gene involved in tryptophan degradation, as a general regulator of protein homeostasis during aging [60]. Moreover, 49 of the original 80 modifier genes had human homologs, which were enriched for genes related to vesicular trafficking functions. In another screen using a similar model, nematode genes orthologous to human familial PD genes were preselected to perform a hypothesis-based RNAi screen [69]. A subset of candidate genes from the initial screen was then further analyzed in another C. elegans model, expressing alpha-synuclein in the dopamine neurons, in order to assess their relevance at the neuronal level. This study revealed five candidate genes that were able to protect from alpha-synuclein-induced dopaminergic neurodegeneration. Again, the most representative class of genes here was associated with vesicular trafficking, with the exception of the autophagy-related gene Atgr7, of which the mammalian ortholog (Atg7) was previously implicated in neurodegeneration in mice [78]. Also, a serine/threonine kinase involved in axonal elongation, UNC-51, was found to be homologous to the previously associated risk factor ULK-2, as revealed by a genome wide association study performed in PD patients [98]. Parallel to these findings, Kuwahara et al. were able to pinpoint two genes, apa-2 and aps- 2, that when knockdown by RNAi increase alpha-synuclein induced neurotoxicity in a C. elegans model expressing the transgene in the whole nervous system[70].

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 70 These two genes encode for subunits of the AP-2 adaptor complex, which mediates the internalization of cargo into the cell from the extracellular space via clathrin- mediated endocytosis [99].

3.3. C. elegans models for Alzheimer’s disease Alzheimer’s disease (AD) is the most prevalent neurodegenerative disease, which is 2 predicted to affect 66 million people worldwide by 2030 [100]. It represents the most common form of dementia, leading to clinical symptoms such as memory loss and mood swings. Aging and lifestyle are risk factors for development of AD, but 70% of the cases are attributable to genetics [101]. The main neuropathological features are the presence of extracellular amyloid-beta plaques, which consist of an accumulation of aggregated amyloid-beta, and intraneuronal tangles of hyperphosphorylated tau. Mutations in several genes can lead to the development of AD, including mutations in genes encoding for the amyloid-precursor protein (APP), presenelin 1 (PSEN1) and presenelin 2 (PSEN2). These genes are part of the APP cleavage pathway and mutations in these genes promote the processing of APP towards the amyloidogenic pathway, promoting the formation of amyloid-beta. Amyloid-beta peptides can have different lengths, including 40 or 42 amino acids. Amyloid-beta 42 is the most common species found in the amyloid plaques, indicating its propensity to rapidly aggregate in comparison to amyloid-beta 40.

Tau is encoded by the microtubule-associated tau protein (MAPT) gene and predominantly expressed in the nervous system. As to its function, it is known to associate and stabilize microtubules. It has been already classified as one of the risk genes for developing AD by at least two independent studies [102,103].

There are several models in C. elegans that express either human amyloid-beta or tau. In the first case, the worms express amyloid-beta 3–42 in the body wall muscles which causes the progressive accumulation of amyloid-beta 3–42 in the muscle cells and paralysis, which worsens with aging [71,104]. There have been several variations to this model, either combined with inducible systems, driving expression in neurons or, more recently, expressing full-length amyloid-beta 1–42 [105–107]. On the other hand, tau-expressing models have been specific to neuronal cells and the phenotype is either worsening of uncoordinated movement or insensitivity to the touch response due to transgene expression [75,108,109].

Genetic screens in C. elegans models for AD have been scarce. So far, there has been no genetic screen performed in any of the models expressing amyloid-beta. There is only one report on genome-wide RNAi done in a tau-expressing model [75]. Sixty modifier genes were discovered to belong to several functional classes including, kinases, chaperones, proteases and phosphatases. Of these, 38 had homologs in

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 71 humans but, more importantly, 6 had already been associated with disease, either in humans or other animal models. One of these modifiers was the nicotinic acetylcholine receptor alpha-7 (nAchR), a ligand-gated ion channel expressed in the human brain and known to contribute to tau phosphorylation [110].

4. Final considerations Genetic screens are powerful means to find genes involved in a certain biological process of interest and their function. The fact that C. elegans is a tractable system to model human diseases further allows one to perform genome-wide screenings in a relatively quick and unbiased manner. Genetic screens can have two outcomes, both being equally informative. On the one hand, new genes are discovered and therefore novel pathways are implicated, giving fresh perspectives on the biological process being studied. On the other hand, genetic screens that reveal genes already known to be associated with disease strengthen the importance of those genes in pathogenesis. Many screens that start with a genome-wide approach end up with an extensive list of candidate genes that are classified according to their functional class.

From here, a selection of these genes should be refined and prioritized in order to study further their individual contribution to pathogenesis. One of the critical points is considering those that might have significance at the mammalian level. Additionally, if the human gene can replace the function of the endogenous one, it demonstrates evolutionary conservation of function and enables one to extrapolate findings fromsmall organisms to complex human diseases. It is, therefore, essential to validate the genes from the screen in higher organisms. However, it is also often that a screen might reveal genes that do not have a direct sequence homolog in mammals. Nevertheless, they may be indicators of other genes that may be functional orthologs or otherwise regulators of genes with a role in human disease (e.g. transcription factors).

Another contribution of genetic screens may be to provide novel targets for drug development [111]. Although not all features of a complex human disease are fully recapitulated in the nematode, one can argue that its simplicity can be advantageous. Especially, because analysis of the expression of the causative gene and its interactors or modifiers can be done without other confounding factors inherent to the complexity of the human biology.

Genome-wide association studies (GWAS) have gained importance in the last ten years, becoming one of the forefront strategies to find common genetic factors associated with susceptibility to develop disease. One challenge now is to establish the functional consequences of these genetic variations. Coupling genetic screens or candidate gene approach in C. elegans to find causative genes may represent a

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 72 quick and inexpensive way to assess functional relevance of associated variations and consequently obtain concrete targets to act upon. For instance, a genome-wide toxicity screen in yeast revealed 6 modifiers of amyloid-beta toxicity that were previously identified as risk factors in GWAS [112]. Importantly, those modifiers were functionally conserved from yeast, to C. elegans and to rat. Another study by Shulman et al. showed, for the first time, a link between an AD risk factor and a 2 causative gene by functional screening in the fly [113].

Although this review focused in C. elegans models for neurodegenerative diseases, it should be noted that C. elegans is a model organism for other human diseases as well. C. elegans has been used to model certain aspects of cancer, diabetes, obesity, polycystic kidney disease, muscular dystrophy and innate immunity, to name a few. A more complete view of these different disease models is summarized elsewhere [111].

All in all, genetic screens in small organisms such as C. elegans can not only aid to dissect fundamental biological questions but also have the versatility of being adapted to model complex human diseases, such as neurodegenerative diseases. Moreover, its attributes make it a tractable system to drug target discovery and compound screening, emphasizing the potential of this organism to extrapolate findings from small organisms to higher vertebrates.

Online links – Textpresso, a full text literature searches of C. elegans (http://www.textpresso.org/)

– Worm Interactome Database (http://interactome.dfci.harvard.edu/C_elegans/ index.php)

– The Caenorhabditis Genetic Center, with an extensive list of strains (http://www. cbs.umn.edu/CGC/)

– Wormbase, a complete database of genetics, genomics and biology of C. elegans (www.wormbase.org)

– Wormbook, a comprehensive, open-access collection of original, peer-reviewed chapters covering topics related to the biology of C. elegans and other nematodes (http://wormbook.org)

– C. elegans Gene Knockout Consortium, which creates knockout strains (http:// celeganskoconsortium.omrf.org/)

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 73 – National Bioresource Project, which generates, collects, stores and distributes deletion mutants of C. elegans (http://www.shigen.nig.ac.jp/c.elegans/index.jsp)

– Wormatlas, a database of behavioral and structural anatomy of C. elegans (http:// www.wormatlas.org)

ACKNOWLEDGEMENTS This project was funded by the European Research Council (ERC) starting grant (to E.A.A.N.) and by Fundação para a Ciência e Tecnologia (SFRH/BPD/51009/2010) (to O.S.).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 74 REFERENCES

1. S. Brenner, The genetics of Caenorhabditis elegans, proteomics, Genome Res. 10 (2000) 703–713. Genetics 77 (1974) 71–94. 15. D.D. Shaye, I. Greenwald, OrthoList: a compendium 2. D. Fay, Genetic mapping and manipulation: Chapter of C. elegans genes with human orthologs, PLoS ONE 1-Introduction and basics, in: WormBook (Ed.), The C. 6 (5) (2011) e20085. elegans Research Community, WormBook, February 16. Jorgensen, Mango, The art and design of genetic 17 2006, http://dx.doi.org/10.1895/wormbook.1.90.1, screens: Caenorhabditis elegans, Nat. Rev. Genet. 3 2 (http://www.wormbook.org). (5) (2002) 356–369. 3. S. Wolff, H. Ma, D. Burch, G.A. Maciel, T. Hunter, 17. Davis, et al., rapid single nucleotide polymorphism A. Dillin, SMK-1, an essential regulator of DAF-16- mapping in C. elegans, BMC Genomics 6 (2005) 118. mediated longevity, Cell 124 (2006) 1039–1053. 18. Sarin, et al., Caenorhabditis elegans mutant allele 4. T. Gidalevitz, N.Wang, T. Deravaj, J. Alexander- identification by whole-genome sequencing, Nat. Floyd, R.I.Morimoto, Natural genetic variation Methods 5 (10) (2008) 865. determines susceptibility to aggregation or toxicity in a C. elegans model for polyglutamine disease, BMC 19. Shen, et al., Comparing platforms for C. elegans Biol. 11 (2013) 100. mutant identification using high-throughput whole- genome sequencing, PLoS ONE 3 (12) (2008) e4012. 5. D.L. Riddle, T. Blumenthal, B.J. Meyer, et al., (Eds.), C. elegans II, 2nd edition, Section I, The Biological 20. Bigelow, et al., A software, MAQGene, facilitating C. Model Cold Spring Harbor Laboratory Press, Cold elegans whole genome sequence analysis for mutant Spring Harbor (NY), 1997. identification, Nat. Methods 6 (8) (2009) 549. 6. S.F. Gilbert, Developmental Biology, Early 21. Tissenbaum, Guarente, increased dosage of sir-2 gene Development of the Nematode Caenorhabditis extends lifespan in Caenorhabditis elegans, Nature 410 elegans, 6th edition, Sinauer Associates, Sunderland (6825) (2001) 227–230. (MA), 2000. 22. Burnett, et al., Absence of effects of Sir-2 7. The C. elegans sequencing consortium. Genome overexpression on lifespan in C. elegans and sequence of the nematode C. elegans: a platform for Drosophila, Nature 477 (7365) (2011) 482–485. investigating biology, Science 282 (1998) 2012–2018. 23. Viswanathan, Guarente, Regulation of Caenorhabditis 8. A. Dillin, A.L. Hsu, N. Arantes-Oliveira, J. Lehrer- elegans lifespan by sir-2.1 transgenes, Nature 477 Graiwer, H. Hsin, A.G. Fraser, R.S. Kamath, J. (7365) (2011) E1–E2. Ahringer, C. Kenyon, Rates of behavior and 24. Bargmann, Neurobiology of the Caenorhabditis aging specified by mitochondrial function during elegans genome, Science 282 (1998) 2028–2033. development, Science 298 (2002) 2398–2401. 25. Ophoff, et al., Familial hemiplegic migraine and 9. S.S. Lee, S. Kennedy, A.C. Tolonen, G. Ruvkun, DAF- episodic ataxia type-2 are caused by mutations in 16 target genes that control C. elegans life-span and the Ca2+ channel gene CACNL1A4, Cell 87 (1996) metabolism, Science 300 (2003) 644–647. 543–552. 10. Y. Kim, H. Sun, Functional genomic approach to 26. Ferrari, et al., Serotonin metabolism inmigraine, identify novel genes involved in the regulation of Neurology 39 (1989) 1239–1242. oxidative stress resistance and animal lifespan, Aging 27. Kindt, et al., Serotonin promotes G(o)-dependent Cell 6 (2007) 489–503. neuronal migration in Caenorhabditis elegans, Curr. 11. B. Hamilton, Y. Dong, M. Shindo, W. Liu, I. Odell, Biol. 12 (20) (2002) 1738–1747. G. Ruvkun, S.S. Lee, A systematic RNAi screen for 28. Schafer, Kenyon, A calcium-channel homologue longevity genes in C. elegans, Genes Dev. 19 (2005) required for adaptation to dopamine and serotonin in 1544–1555. Caenorhabditis elegans, Nature 375 (6526) (1995) 12. P. Zipperlen, A.G. Fraser, R.S. Kamath, M. Martinez- 73–78. Campos, J. Ahringer, Roles for 147 embryonic lethal 29. Estevez, et al., The voltage-gated calcium channel genes on C. elegans chromosome I identified by RNA UNC-2 is involved in stress mediated regulation of interference and video microscopy, EMBO J. 20 tryptophan hydroxylase, J.Neurochem. 88 (2004) (2001) 3984–3992. 102–113. 13. C.D. Keating, N. Kriek, M. Daniels, N.R. Ashcroft, 30. Bo, et al., Cerebrospinal fluid cytokine levels in N.A. Hopper, E.J. Siney, L. Holden-Dye, J.F. Burke, migraine, tension-type headache and cervicogenic Whole-genome analysis of 60 G protein-coupled headache, Cephalalgia 29 (3) (2009) 365–372. receptors in Caenorhabditis elegans by gene knockout 31. Sulston, Horvitz, Abnormal cell lineages in mutants of with RNAi, Curr. Biol. 13 (2003) 1715–1720. the nematode Caenorhabditis elegans, Dev. Biol. 82 14. C.H. Lai, C.Y. Chou, L.Y. Chang, C.S. Liu,W. Lin, (1) (1981) 41–55. Identification of novel human genes evolutionarily 32. Wang, Sternberg, et al., Pattern formation during C. conserved in Caenorhabditis elegans by comparative elegans vulval induction, Curr. Top. Dev. Biol. 51

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 75 (2001) 189–220. 421 (6920) (2003) 231–237. 33. Beitel, et al., The Caenorhabditis elegans gene lin-1 51. Wang, Barr, RNA interference in Caenorhabditis encodes an ETS-domain protein and defines a branch elegans, Methods Enzymol. 392 (2005) 36–55. of the vulval induction pathway, Genes Dev. 9 (1995) 52. Rual, et al., Toward improving Caenorhabditis elegans 3149–3162. phenome mapping with an ORFeome-based RNAi 34. Jacobs, et al., Gain-of-function mutations in the library, Genome Res. 14 (10B) (2004) 2162–2168. Caenorhabditis elegans lin-1 ETS gene identify a 53. Eliscovich, et al., mRNA on the move: the road to C-terminal regulatory domain phosphorylated by ERK its biological destiny, J. Biol. Chem. 288 (28) (2013) MAP Kinase, Genetics 149 (1998) 1809–1822. 20361–20368. 35. Egan, Weinberg, The pathway to signal achievement, 54. Barrett, et al., Regulation of eukaryotic gene Nature 365 (6449) (1993) 781–783. expression by the untranslated gene regions and other 36. Pawson, Hunter, Signal transduction and growth non-coding elements, Cell. Mol. Life Sci. 69 (21) control in normal and cancer cells, Curr. Opin. Genet. (2012) 3613–3634. Dev. 4 (1) (1994) 1–4. 55. Maine, Studying gene function in Caenorhabditis 37. Waterston, et al., The Genome, in: Riddle, Blumenthal, elegans using RNA-mediated interference, Brief. Meyer, Priess (Eds.), C. elegans II, vol. 33, Cold Funct. Genomic Proteomic 7 (3) (2008) 184–194. Spring Harbor Laboratory Press, New York, 1997, pp. 56. Calixto, et al., Coditional gene expression and RNAi 23–45. using MEC-8-dependent splicing in C.elegans, Nat. 38. Fire, et al., Potent and specific genetic interference Methods 7 (5) (2010) 407–411. by double- stranded RNA in Caenorhabditis elegans, 57. Colaiácovo, et al., A targeted RNAi screen for genes Nature 391 (6669) (1998) 806–811. involved in chromosome morphogenesis and nuclear 39. Timmons, et al., Inducible systemic RNA silencing in organization in the Caenorhabditis elegans germline, Caenorhabditis elegans, Mol. Biol. Cell 14 (7) (2003) Genetics 162 (1) (2002) 113–128. 2972–2983. 58. Reinke, et al., A global profile of germline gene 40. Timmons, et al., Ingestion of bacterial expressed expression in C. elegans, Mol. Cell 6 (2000) 605–616. dsRNAs can produce specific and potent genetic 59. T.J. van Ham, K.L. Thijssen, R. Breitling, R.M. interference in Caenorhabditis elegans, Gene 263 Hofstra, R.H. Plasterk, E.A. Nollen, C. elegans (1–2) (2001) 103–112. model identifies genetic modifiers of alpha-synuclein 41. Tabara, et al., RNAi in C. elegans: soaking in the inclusion formation during aging, PLoS Genet. 4 genome sequence, Science 282 (5388) (1998) 430– (2008) e1000027. 431. 60. van der Goot, et al., Delaying aging and the aging- 42. Tavernakis, et al., Heritable and inducible genetics associated decline in protein homeostasis by inhibition interference by double-stranded RNA encoded by of tryptophan degradation, Proc. Natl. Acad. Sci. U. S. transgenes, Nat. Genet. 24 (2) (2000) 180–183. A. 109 (37) (2012) 14912–14917. 43. Winston, et al., Systemic RNAi in C. elegans requires 61. Simmer, et al., Genome-wide RNAi of C. elegans the putative transmembrane protein SID-1, Science using the hypersensitive rrf-3 strain reveals novel gene 295 (5564) (2002) 2456–2459. functions, PLoS Biol. 1 (1) (2003) E12. 44. Hunter, et al., Systemic RNAi in Caenorhabditis 62. T.J. van Ham, R. Breitling, M.A. Swertz, E.A. Nollen, elegans, Cold Spring Harb. Symp. Quant. Biol. 71 Neurodegenerative diseases: lessons from genome- (2006) 95–100. wide screens in small model organisms, EMBO Mol. 45. Simmer, et al., Loss of putative RNA-directed RNA Med. 1 (2009) 360–370. polymerase RRF-3 makes C. elegans hypersensitive to 63. C. Soto, Unfolding the role of protein misfolding in RNAi, Curr. Biol. 12 (15) (2002) 1317–1319. neurodegenerative diseases, Nat. Rev. Neurosci. 4 46. Kennedy, et al., A conserved siRNA-degrading RNase (2003) 49–60. negatively regulates RNA interference in C. elegans, 64. F. Chiti, C.M. Dobson, Protein misfolding, functional Nature 427 (6975) (2004) 645–649. amyloid, and human disease, Annu. Rev. Biochem. 75 47. Calixto, et al., Enhanced neuronal RNAi in C. elegans (2006) 333–366. using SID-1, Nat. Methods 7 (7) (2010) 554–559. 65. B. Bolognesi, J.R. Kumita, T.P. Barros, E.K. Esbjorner, 48. Durieux, et al., The cell-non-autonomous nature of L.M. Luheshi, D.C. Crowther, M.R. Wilson, C.M. electron transport chain mediated longevity, Cell 144 Dobson, G. Favrin, J.J. Yerbury, ANS binding reveals (1) (2011) 79–91. common features of cytotoxic amyloid species, ACS Chem. Biol. 20 (2010) 735–740. 49. Kamath, et al., Effectiveness of specific RNA-mediated interference through ingested double-stranded RNA in 66. M. Arrasate, S. Mitra, E.S. Schweitzer, M.R. Segal, S. Caenorhabditis elegans, Genome Biol. 2 (1) (2001) Finkbeiner, Inclusion body formation reduces levels (research 0002.1-0002.10). of mutant huntingtin and the risk of neuronal death, Nature 14 (2004) 805–810. 50. Kamath, et al., Systematic functional analysis of the Caenorhabditis elegans genome using RNAi, Nature 67. C. Bessou, J.B. Giugia, C.J. Franks, L. Holden-Dye, L. Ségalat, Mutations in the Caenorhabditis elegans

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 76 dystrophin-like gene dys-1 lead to hyperactivity program? Trends Cell Biol. 23 (2013) 168–174. and suggest a link with cholinergic transmission, 80. P.W. Faber, J.R. Alter, M.E. MacDonald, A.C. Hart, Neurogenetics 2 (1998) 61–72. Polyglutamine-mediated dysfunction and apoptotic 68. J.F. Morley, H.R. Brignull, J.J. Weyers, R.I. Morimoto, death of a Caenorhabditis elegans sensory neuron, The threshold for polyglutamine expansion protein Proc. Natl. Acad. Sci. U. S. A. 96 (1999) 179–184. aggregation and cellular toxicity is dynamic and 81. J.A. Parker, J.B. Connolly, C. Wellington, M. Hayden, influenced by aging in Caenorhabditis elegans, Proc. J. Dausset, C. Neri, Expanded polyglutamines in Natl. Acad. Sci. U. S. A. 99 (2002) 10417–10422. Caenorhabditis elegans cause axonal abnormalities 69. S. Hamamichi, R.N. Rivas, A.L. Knight, S. Cao, and severe dysfunction of PLM mechanosensory 2 K.A. Caldwell, G.A. Caldwell, Hypothesis-based neurons without cell death, Proc. Natl. Acad. Sci. U. RNAi screening identifies neuroprotective genes in a S. A. 98 (2001) 13318–13323. Parkinson’s disease model, Proc. Natl. Acad. Sci. U. S. 82. H.R. Brignull, F.E. Moore, S.J. Tang, R.I. Morimoto, A. 105 (2008) 728–733. Polyglutamine proteins at the pathogenic threshold 70. T. Kuwahara, A. Koyama, S. Koyama, S. Yoshina, display neuron-specific aggregation in a pan-neuronal C.H. Ren, T. Kato, S. Mitani, T. Iwatsubo, A systematic Caenorhabditis elegans model, J. Neurosci. 26 (2006) RNAi screen reveals involvement of endocytic 7597–7606. pathway in neuronal dysfunction in alpha-synuclein 83. E.A. Nollen, S.M. Garcia, H.G. van Haaften, S. Kim, transgenic C. elegans, Hum. Mol. Genet. 17 (2008) A. Chavez, R.I. Morimoto, R.H. Plasterk, Genome- 2997–3009. wide RNA interference screen identifies previously 71. C.D. Link, Expression of human beta-amyloid peptide undescribed regulators of polyglutamine aggregation, in transgenic Caenorhabditis elegans, Proc. Natl. Proc. Natl. Acad. Sci. U. S. A. 101 (2004) 6403–6408. Acad. Sci. U. S. A. 92 (1995) 9368–9372. 84. M.C. Silva, S. Fox, M. Beam, H. Thakkar, M.D. 72. N. Wittenburg, S. Eimer, B. Lakowski, S. Roèhrig, Amaral, R.I. Morimoto, A genetic screening strategy C. Rudolph, R. Baumeister, Presenilin is required identifies novel regulators of the proteostasis network, for proper morphology and function of neurons in C. PLoS Genet. 7 (2011) e1002438. elegans, Nature 406 (2000) 306–309. 85. S.F. Falsone, N.H. Meyer, E. Schrank, G. Leitinger, 73. E. Cohen, J. Bieschke, R.M. Perciavalle, J.W. Kelly, C.L. Pham, M.T. Fodero-Tavoletti, M. Holmberg, A. Dillin, Opposing activities protect against age-onset M. Dulle, B. Scicluna, B. Gesslbauer, H.M. Rückert, proteotoxicity, Science 313 (2006) 1604–1610. G.E. Wagner, D.A. Merle, E.A. Nollen, A.J. Kungl, 74. E. Cohen, J.F. Paulsson, P. Blinder, T. Burstyn- A.F. Hill, R. Cappai, K. Zangger, Cell. Reprogram. 2 Cohen, D. Du, G. Estepa, A. Adame, H.M. Pham,M. (2012) 358–371. Holzenberger, J.W. Kelly, E.Masliah, A. Dillin, 86. L.M. de Lau, M.M. Breteler, Epidemiology of Reduced IGF-1 signaling delays age-associated Parkinson’s disease, Lancet Neurol. 5 (2006) 525–535. proteotoxicity in mice, Cell 139 (2009) 1157–1169. 87. N.N. Inamdar, D.K. Arulmozhi, A. Tandon, S.L. 75. B.C. Kraemer, J.K. Burgess, J.H. Chen, J.H. Thomas, Bodhankar, Parkinson’s disease: genetics and beyond, G.D. Schellenberg, Molecular pathways that influence Curr. Neuropharmacol. 5 (2007) 99–113. human tau-induced pathology in Caenorhabditis 88. S. Przedborski,W. Dauer, Parkinson’s disease: elegans, Hum. Mol. Genet. 15 (2006) 1483–1496. mechanisms and models, Neuron 39 (2003) 889–909. 76. T.J. van Ham,M.A. Holmberg, A.T. van der Goot, E. 89. K.K. Chung, Y. Zhang, K.L. Lim, Y. Tanaka, H. Huang, Teuling,M. Garcia-Arencibia, H.E. Kim, D. Du, K.L. J. Gao, C.A. Ross, V.L. Dawson, T.M. Dawson, Parkin Thijssen, M. Wiersma, R. Burggraaff, P. van Bergeijk, ubiquitinates the α-synuclein-interacting protein, J. van Rheenen, G.J. van Veluw, R.M. Hofstra, D.C. synphilin-1: implications for Lewy-body formation Rubinsztein, E.A. Nollen, Identification of MOAG-4/ in Parkinson disease, Nat. Med. 7 (2001) 1144–1150. SERF as a regulator of age-related proteotoxicity, Cell 20 (2010) 601–612. 90. A. Abeliovich, Y. Schmitz, I. Fariñas, D. Choi- Lundberg, W.H. Ho, P.E. Castillo, N. Shinsky, J.M. 77. F.X. Lejeune, L. Mesrob, F. Parmentier, C. Bicep, Verdugo, M. Armanini, A. Ryan, M. Hynes, H. R.P. Vazquez-Manrique, J. Alex Parker, J.P. Vert, Phillips, D. Sulzer, A. Rosenthal, Mice lacking alpha- C. Tourette, C. Neri, Large-scale functional RNAi synuclein display functional deficits in the nigrostriatal screen in C. elegans identifies genes that regulate the dopamine system, Neuron 25 (2000) 239–252. dysfunction of mutant polyglutamine neurons, BMC Genomics 13 (2012) 91. 91. C. Lavedan, The synuclein family, Genome Res. 8 (1998) 871–880. 78. M. Komatsu, Q.J. Wang, G.R. Holstein, V.L. Friedrich Jr., J. Iwata, E. Kominami, B.T. Chait, K. Tanaka, Z. 92. D.F. Clayton, J.M. George, The synucleins: a family Yue, Essential role for autophagy protein Atg7 in the of proteins involved in synaptic function, plasticity, maintenance of axonal homeostasis and the prevention neurodegeneration and disease, Trends Neurosci. 21 of axonal degeneration, Proc. Natl. Acad. Sci. U. S. A. (1998) 249–254. 104 (2007) 14489–14494. 93. M.H. Polymeropoulos, C. Lavedan, E. Leroy, S.E. Ide, 79. E.S. Blum, A.R. Schwendeman, S. Shaham, PolyQ A. Dehejia, A. Dutra, B. Pike, H. Root, J. Rubenstein, disease: misfiring of a developmental cell death R. Boyer, E.S. Stenroos, S. Chandrasekharappa, A. Athanassiadou, T. Papapetropoulos, W.G. Johnson,

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 77 A.M. Lazzarini, R.C. Duvoisin, G. Di Iorio, L.I. Golbe, 105. C.D. Link, A. Taft, V. Kapulkin, K. Duke, S. Kim, R.L. Nussbaum, Mutation in the alpha-synuclein gene Q. Fei, D.E. Wood, B.G. Sahagan, Gene expression identified in families with Parkinson’s disease, Science analysis in a transgenic Caenorhabditis elegans 276 (1997) 2045–2047. Alzheimer’s disease model, Neurobiol. Aging 24 94. R. Krüger, W. Kuhn, T. Müller, D. Woitalla, M. (2003) 397–413. Graeber, S. Kösel, H. Przuntek, J.T. Epplen, L. Schöls, 106. Y. Wu, Z. Wu, P. Butko, Y. Christen, M.P. Lambert, O. Riess, Ala30Pro mutation in the gene encoding W.L. Klein, C.D. Link, Y. Luo, Amyloid-beta-induced alpha synuclein in Parkinson’s disease, Nat. Genet. 18 pathological behaviors are suppressed by Ginkgo (1998) 106–108. biloba extract EGb 761 and ginkgolides in transgenic 95. J.J. Zarranz, J. Alegre, J.C. Gómez-Esteban, E. Caenorhabditis elegans, J. Neurosci. 26 (2006) 13102– Lezcano, R. Ros, I. Ampuero, L. Vidal, J. Hoenicka, O. 13113. Rodriguez, B. Atarés, V. Llorens, E. Gomez Tortosa, 107. G. McColl, B.R. Roberts, T.L. Pukala, V.B. Kenche, T. del Ser, D.G. Muñoz, J.G. de Yebenes, The new C.M. Roberts, C.D. Link, T.M. Ryan, C.L. Masters, mutation, E46K, of alpha-synuclein causes Parkinson K.J. Barnham, A.I. Bush, R.A. Cherny, Utility of an and Lewy body dementia, Ann. Neurol. 55 (2004) improved model of amyloid-beta (Aβ1-42) toxicity 164–173. in Caenorhabditis elegans for drug screening for 96. M.C. Chartier-Harlin, J. Kachergus, C. Roumier, Alzheimer’s disease, Mol. Neurodegener. 7 (2012) 57. V. Mouroux, X. Douay, S. Lincoln, C. Levecque, L. 108. T. Miyasaka, Z. Ding, K. Gengyo-Ando, M. Oue, Larvor, J. Andrieux,M. Hulihan, N.Waucquier, L. H. Yamaguchi, S. Mitani, Y. Ihara, Progressive Defebvre, P. Amouyel, M. Farrer, A. Destée, Alpha- neurodegeneration in C. elegans model of tauopathy, synuclein locus duplication as a cause of familial Neurobiol. Dis. 20 (2005) 372–383. Parkinson’s disease, Lancet 364 (2004) 1167–1169. 109. R. Brandt, A. Gergou, I.Wacker, T. Fath, H. 97. A.B. Singleton,M. Farrer, J. Johnson, A. Singleton, Hutter, A Caenorhabditis elegans model of tau S. Hague, J. Kachergus,M. Hulihan, T. Peuralinna, hyperphosphorylation: induction of developmental A. Dutra, R. Nussbaum, S. Lincoln, A. Crawley, M. defects by transgenic overexpression of Alzheimer’s Hanson, D. Maraganore, C. Adler, M.R. Cookson, M. disease-like modified tau, Neurobiol. Aging 30 (2009) Muenter, M. Baptista, D. Miller, J. Blancato, J. Hardy, 22–33. K. Zwinn-Hardy, Alpha-synuclein locus triplication 110. H.Y. Wang, W.W. Li, N.J. Benedetti, D.H.S. causes Parkinson’s disease, Science 302 (2003) 841. Lee, Alpha(7) nicotinic acetylcholine receptors 98. Fung, et al., Genome-wide genotyping in Parkinson’s mediate beta-amyloid peptide-induced tau protein disease and neurologically normal controls: first stage phosphorylation, J. Biol. Chem. 278 (2003) 31547– analysis and public release of data, Lancet Neurol. 5 31553. (2006) 911–916. 111. T. Kaletta, M.O. Hengartner, Finding function in novel 99. B.M. Collins, A.J. McCoy, H.M. Kent, P.R. Evans, D.J. targets: C. elegans as amodel organism, Nat. Rev. Drug Owen, Molecular architecture and functional model of Discov. 5 (2006) 387–398. the endocytic AP2 complex, Cell 109 (2002) 523–535. 112. S. Treusch, S. Hamamichi, J.L. Goodman, K.E. 100. K. Bettens, K. Sleegers, C. Van Broeckhoven, Genetic Matlack, C.Y. Chung, V. Baru, J.M. Shulman, A. insights in Alzheimer’s disease, Lancet Neurol. 12 Parrado, B.J. Bevis, J.S. Valastyan, H. Han, M. (2013) 92–104. Lindhagen-Persson, E.M. Reiman, D.A. Evans, D.A. 101. C. Ballard, S. Gauthier, A. Corbett, C. Brayne, D. Bennett, A. Olofsson, P.L. DeJager, R.E. Tanzi, K.A. Aarsland, E. Jones, Alzheimer’s disease, Lancet 377 Caldwell, G.A. Caldwell, S. Lindquist, Functional (2011) 1019–1031. links between Aβ toxicity, endocytic trafficking, and Alzheimer’s disease risk factors in yeast, Science 334 102. A.J. Myers, M. Kaleem, L. Marlowe, A.M. Pittman, (2011) 1241–1245. A.J. Lees, H.C. Fung, J. Duckworth, D. Leung, A. Gibson, C.M. Morris, R. de Silva, J. Hardy, The 113. J.M. Shulman, P. Chipendo, L.B. Chibnik, C. Aubin, H1c haplotype at the MAPT locus is associated with D. Tran, B.T. Keenan, P.L. Kramer, J.A. Schneider, Alzheimer’s disease, Hum. Mol. Genet. 14 (2005) D.A. Bennett, M.B. Feany, P.L. De Jager, Functional 2399–23404. screening of Alzheimer pathology genome-wide association signals in Drosophila, Am. J. Hum. Genet. 103. T.M. Caffrey, R. Wade-Martins, Functional MAPT 88 (2011) 232–238. haplotypes: bridging the gap between genotype and neuropathology, Neurobiol. Dis. 27 (2007) 1–10. 104. G. McGoll, B.R. Roberts, A.P. Gunn, K.A. Perez, D.J. Tew, C.L. Masters, K.J. Barnham, R.A. Cherny, A.I. Bush, The Caenorhabditis elegans Aß1-42 model of Alzheimer disease predominantly expresses Aß3-42, J. Biol. Chem. 284 (2009) 22697–22702.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 78 2

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 79 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 80 Chapter III

Identification of an evolutionary conserved structural loop that is required for the enzymatic and biological function of tryptophan 2,3-dioxygenase

Helen Michels1, Renée I. Seinstra1, Joost Uitdehaag2, Mandy Koopman1, Martijn van Faassen3, Céline N. Martineau1, Ido P. Kema3, Rogier Buijsman2, Ellen A.A. Nollen1

1European Research Institute for the Biology of Aging, University of Groningen, University Medical Center Groningen, Laboratory of Molecular Neurobiology of Aging, The Netherlands 2Netherlands Translational Research Center B.V., Oss, The Netherlands 3University of Groningen, University Medical Center Groningen, Department of Laboratory Medicine, The Netherlands

Michels H, Seinstra RI, Uitdehaag J, Koopman M, Van Faassen M, Martineau CN, Kema IP, Buijsman R, Nollen EAA. Identification of an evolutionary conserved structural loop that is required for the enzymatic and biological function of tryptophan 2,3-dioxygenase. Sci Rep. 2016 Dec 20;6:39199.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 81 The enzyme TDO (tryptophan 2,3-dioxygenase; TDO-2 in Caenorhabditis elegans) is a potential therapeutic target to cancer but is also thought to regulate proteotoxic events seen in the progression of neurodegenerative diseases. To better understand its function and develop specific compounds that target TDO we need to understand the structure of this molecule. In C. elegans we compared multiple different CRISPR/Cas9-induced tdo-2 deletion mutants and identified a motif of three amino acids (PLD) that is required for the enzymatic conversion of tryptophan to N-formylkynurenine. Loss of TDO-2’s enzymatic activity in PDL deletion mutants was accompanied by an increase in motility during aging and a prolonged lifespan, which is in line with the previously observed phenotypes induced by a knockdown of the full enzyme. Comparison of sequence structures suggests that blocking this motif might interfere with haem binding, which is essential for the enzyme’s activity. The fact that these three residues are situated in an evolutionary conserved structural loop of the enzyme suggests that the findings can be translated to humans. The identification of this specific loop region in TDO-2–essential for its catalytic function–will aid in the design of novel inhibitors to treat diseases in which the TDO enzyme is overexpressed or hyperactive.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 82 Genetic variation, environmental factors and aging can lead to malfunctioning or hyperactive proteins, thereby provoking reactions in the cell that might underlie certain diseases. Understanding the structure of such proteins is required when aiming to generate disease-specific compounds that interfere with the protein’s function.

One such protein, that has recently attracted the attention as a potential therapeutic target, is tryptophan 2,3-dioxygenase (TDO in humans, TDO-2 in C. elegans) (e.g. ref. 1 and summarised in refs 2 and 3), a cytosolic enzyme that converts tryptophan into formyl kynurenine, thereby catalysing the first rate-limiting step of the kynurenine 3 pathway4. In animal models, inhibition or depletion of TDO has been shown to suppress disease phenotypes and to prolong lifespan5–9. This is thought to be mediated by inhibition of different tryptophan-related pathways that underlie the diseases. For example, since the kynurenine pathway is used to convert 95% of cellular tryptophan, it greatly influences the availability of this amino acid for other tryptophan-related metabolic pathways, including, but not limited to, the production of serotonin10. A lack of serotonin has, among others, been associated with depression disorders, which are usually treated with selective serotonin re-uptake inhibitors (SSRIs) to increase serotonin levels (summarised in ref. 11). Secondly, the kynurenine pathway produces metabolites called kynurenines, and an imbalance of certain kynurenines such as kynurenic acid, 3-hydroxykynurenine and quinolinic acid is thought to have profound effects on brain function, may cause a range of neuron-associated diseases (e.g. ref. 12 and summarised in ref. 13) and was recently found being associated with diabetes type 214. Thirdly, overexpression of TDO in certain cancers is thought to increase the tumour’s resistance to the T-cell-mediated immune response which might be dependent on tryptophan in the microenvironment15. Since the majority of the phenotypes described above are accompanied by overexpression or hyperactivity of TDO, it is possible that inhibiting the activity of TDO would increase tryptophan and subsequently serotonin levels and control the production of kynurenines.

The development of such compounds requires a thorough understanding of the structure of TDO. Fortunately, the conversion of tryptophan by TDO is evolutionarily conserved from yeast to humans16, which means that studies in other organisms may provide useful information on both its structure and function. For example, from such studies we know that TDO is mainly present as homodimers that can form tetramers to be enzymatically active4 and that the enzymatic reaction it catalyses is haem and oxygen-dependent17,18. The haem and tryptophan binding sites in human TDO are also known, and the enzyme’s 3D structure was first described by Forouhar and colleagues for TDO in the bacterium X. campestris19. While the crystal structure of human TDO has now also been reported20, it still remains to be determined which specific amino acids in the TDO polypeptide chains are critical for its enzymatic activity.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 83 We created multiple different CRISPR/Cas9-induced deletion mutants of tdo-2 in the nematode C. elegans. A comparison of these mutants revealed that three amino acid residues in the loop that is close to the haem moiety are essential for the enzymatic activity of TDO-2. The effect of the deletion of those three amino acid residues was as severe as that seen for larger truncations of the enzyme. We also found that the loop structure in which these three amino acid residues reside, is evolutionary conserved in human.

These findings will aid in the development of very specific inhibitors of TDO to intervene with certain disorders.

RESULTS AND DISCUSSION We mutated tdo-2 in C. elegans with the help of the CRISPR/Cas9 method. To increase the chances of generating truncation mutants we chose a target sequence that was situated the closest to the transcription start. The CRISPR/Cas9 system induces double strand breaks at the targeted sequence. Mistakes during the repair process can lead to mutations, mainly deletions. We got three different mutants with deletions in exon 3 (Fig. 1a), namely tdo-2 (ΔPLD), which lacked 9 bp, tdo-2 (del), which lacked 14 bp, and tdo-2 (del B), which lacked 28 bp. We predicted that the tdo- 2 (del) and tdo-2 (del B) deletions led to a frame shift during translation, resulting in an early STOP codon. Indeed, upon immunoblotting we were unable to detect complete TDO-2 proteins in these mutants (Fig. 1b). However, the mutant with the 9 bp deletion (tdo-2 (ΔPLD)) produced a protein that was probably missing exactly three amino acid residues (Fig. 1a) but that was still detectable with the anti-TDO-2 antibody (Fig. 1b).

We then determined whether the deletions resulted in complete knockout mutations. Figure 1c illustrates the first part of the kynurenine pathway whereby TDO (TDO- 2 in C. elegans) converts tryptophan into formylkynurenine, which is further processed to kynurenine. To this end, we measured tryptophan and kynurenine levels in C. elegans lysates. Surprisingly, we saw higher tryptophan levels and reduced kynurenine levels not only in the samples from animals whose TDO-2 proteins had large truncations, but also in the samples from the ΔPLDmutants, suggesting that the three amino acids missing in this mutant are essential for tryptophan conversion by TDO-2 (Fig. 1d).

We then checked if the phenotypes of the mutants were the same as those described previously when non-mutated animals were treated with tdo-2 RNAi8. Indeed, all three mutants produced fewer hatched progeny and had an extended reproductive

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 84 a tdo-2 (2328 bp)

exon 1 exon 2 exon 3 exon 4 exon 5 exon 6 exon 7

WT accgagcaaatcactttgctcgacacaatgagtccattggattttgtggatttcagaaagtatttga tdo-2 (ΔPLD) accgagcaaatcactttgctcgacacaatgag ttttgtggatttcagaaagtatttga tdo-2 (del) accgagcaaatcactttgctcgacacaatga gtggatttcagaaagtatttga tdo-2 (del B) accgagcaaatcac tttgtggatttcagaaagtatttga

translation

WT …lkivsgldrmtkilsllteqitlldtmspldfvdfrkyltpasgfqslqfrvlenk… tdo-2 (ΔPLD) …lkivsgldrmtkilsllteqitlldtms-- - fvdfrkyltpasgfqslqfrvlenk… 3 tdo-2 (del) …lkivsgldrmtkilsllteqitlldtmsgfqkvfdtsfgipessvpcigeST OP tdo-2 (del B) …lkivsgldrmtkilsllteqitlwisesiSTOP

b WT del B del ΔPLD 46kDa

c human/ C. elegans d tryptophan levels kynurenine levels * * * 100 * ** ** ** ** 0,5 60 WT tryptophan 80 * tdo-2 (ΔPDL) ** * * 0,4 50 * ** ** tdo-2 (del) * **

60 ** 40 0,3 tdo-2 (del B) TDO/TDO-2 * 40 *

30 * * * ** µmol/L

0,2 * * µmol/L ** ** * * * ** **

20 20 ** formylkynurenine body bends/ 30 sec 0,1 10 0 0 day 2 day 40 day 8 AFMID/AMFD-1 day 1 day 4 day 8 day 1 day 4 day 8

kynurenine

Figure 1. Deletion of PLD sequence impairs conversion of tryptophan by TDO-2. (a) CRISPR/Cas9-induced deletions in exon 3 of the tdo-2 gene. (b) Immunodetection of the TDO-2 enzyme in worm lysates from day 4 of adulthood (Cropped image, for full length blot see Supplementary Fig. 1a). (c) Schematic representation of TDO-mediated conversion of tryptophan to kynurenine in humans and C. elegans.(d) Tryptophan and kynurenine levels in worm lysates. Error bars in all panels = SEM. Statistics for all panels, one-way ANOVA with post-hoc Bonferroni (comparison with WT): * < 0.05; ** < 0.01; *** < 0.001.

lifespan (Fig. 2a). They also had increased motility (Fig. 2b and Supplementary Fig. S1). Comparing the relationship between time and genotype, the results suggest that differences in motility start to appear at day 4 of adulthood and become more consistent at higher ages, suggesting an age-dependent or accumulative effect. All three mutants show similar results (see legends of Fig. 2b and Supplementary Fig. S1). Finally, the different mutations all similarly extended lifespan (Fig. 2c). To note, the lifespan extension in the three different tdo-2 mutants was similar to each other but less when compared to the lifespan extension by RNAi knockdown starting after development8 and also less robust (Supplementary data Table S2). This suggests

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 85 a number of hatched progeny total 140 100 300 WT 120 80 tdo-2 (ΔPDL) 250 *** *** 100 tdo-2 (del) *** 60 tdo-2 (del B) 200 ***

80 *** *** *** *** 40 *** 150

60 *** *** 20 ***

body bends/ 30 sec 100 40 ns ns ns 0 ***

*** 50 day20 2 day 4 day 8 *** ** ** ** number of hatched progeny hatched of number 0 0 L4-D1 D1-D2 D2-D3 D3-D4 D4-D5 D5-D6

b 100 WT * ** * *** 80 * tdo-2 (ΔPDL) ** * ** * *** tdo-2 (del) 60 tdo-2 (del B) ** *** ** 40 *

20 body bends/ 30 sec

0 day 4 day 6 day 8 day 12

Copy of Survival of Data 1:Survival proportions Copy of Survival of Data 1:SurvivalCopy of SurvivalproportionsCopy of Data of Survival1:Survival of proportions Data 1:SurvivalCopy of proportions SurvivalCopy of Data of Survival1:Survival of proportionsData 1:Survival proportions 100 100 c 100 WT 100 100 100 WT WT WT WT WT WT tdo-2 (?PLD) 80 tdo-2 (ΔPLD) tdo-2 (?PLD) tdoOW716-2 (del) OW716 tdoOW715-2 (del B) OW715

60 50 50 50 50 50 40 Percent survival Percent survival 20 Percent survival Percent survival Percent survival Percent survival

0 0 0 0 0 0 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 Days of adulthood Days of adulthood Days of adulthood Days of adulthood Days of adulthood Days of adulthood

Figure 2. PLD motif in TDO-2 is required for regulation of reproduction, motility and lifespan. (a) Reproductive behaviour. (b) Motility. Results of interaction by two-way ANOVA considering genotype and time: p < 0.001, p < 0.001, p 0.5205. (c) Survival curve of wild type and tdo-2 mutated animals. Error bars in all panels = SEM. Statistics for all panels (comparison with WT): ** < 0.01; *** < 0.001.

that these germline mutations in tdo-2 may have unknown additional developmental and health consequences that influence lifespan. Still, taken together our results show that the 9 bp Δ PLD deletion is just as effective in mediating TDO-2-specific phenotypes as the removal of almost the entire protein, as it is the case for the tdo- 2(del) and tdo-2(del B) mutants, indicating that it is an inactivating mutation. The results also suggest that loss of enzymatic activity of TDO-2 is responsible for the observed phenotypes.

Next, to determine why these three amino acid residues might be essential for enzyme activity, we studied the sequence and structure of the TDO-2 molecule. When we aligned the sequences of TDO of different organisms with that of C. elegans we found that human and C. elegans show a sequence identity of 48.4%. Figure 3a depicts a part of the amino acid sequences, highlighting the residues involved in tryptophan binding (yellow) and haem binding (red) as described by Forouhar et al.19. Residues forming part of binding sites for both tryptophan and haem are shown in

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 86 orange (The complete sequence alignment is shown in Supplementary Fig. S3). The sequence comparison shows a strong homology around the site of the PLD deletion (Fig. 3a, highlighted in blue). Whereas the frameshift mutants are lacking most of the tryptophan and haem-binding sites (everything downstream of the PLD site), the PLD mutation is located in a region surrounded by binding sites for tryptophan and haem.

In order to understand the implications of the PLD deletion, we then studied 3D structures. Because the X-ray structure of C. elegans TDO-2 is not known, we 3 generated a model based on the 3D structure of human TDO (ref. 20 and Fig. 3b). When enzymatically active, TDO forms a tetramer, as illustrated in the structure depicted in Fig. 3b (monomers in grey, orange, blue and light red). The structure shows that the sequence has a flexible loop (MSPLDF-motif, depicted in green) that connects two alpha-helices that together interact with the haem moiety (green), which is important for the catalytic activity of TDO (ref. 19 and Fig. 3b). This loop is enlarged in Fig. 3c to show the interactions between different TDO monomers that stabilize the final enzyme structure as well as the binding to tryptophan (orange) and haem (green, Fig. 3c).

The tdo-2 (del) and tdo-2 (delB) mutant proteins are missing the entire ‘lid’ of the catalytic site, which is found on the C-terminal side of the MSPLDF loop (indicated red in Fig. 3b), explaining the lack of catalytic activity and the increased levels of tryptophan in these mutant animals.

We also generated a model for the PLD deletion mutant. The PLD deletion leaves the TDO-2 structure largely intact (Fig. 3d, right-hand panel). However, it shortens the MSPLDF loop (compare enlarged images of WT and tdo-2 (ΔPLD) models in Fig. 3d, left and right-hand panels) and removes Asp135 (signified by the D in PLD). This Asp135 residue interacts with lysine (Lys106) and arginine (Arg113) on a neighbouring chain and, the absence of such interaction most likely destabilises the protein’s quaternary structure (Fig. 3b). The PLD deletion also removes Pro133 (signified by the P in PLD), and proline is an amino acid residue known to provide rigidity to turns in protein structures. Based on the important structural functions of these two amino acids, we suggest that the PLD deletion introduces too much flexibility in the MSPLDF loop connecting helix α C and α D of the C. elegans TDO-2 structure (nomenclature of the alpha helices according to ref. 19) (Fig. 3c). Furthermore, as demonstrated in the X. campestris TDO X-ray structure19, the binding of tryptophan to an Arg in the α D helix by salt bridges is key to substrate-enzyme binding (Arg 117 in X. campestris as indicated in Fig. 3c, which is equivalent to Arg140 in C. elegans and Arg 144 in H. sapiens). Subtle disturbances in the positioning of this helix could be the main cause of the protein’s loss of enzymatic activity. In support of this, Dolušić et al.21 have shown that TDO inhibitors 680C91 and LM10 both strongly

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 87 a H. sapiens R. norvegicus M. musculus C. elegans D. melanogaster D. rerio X. laevis

b c

WT (C. elegans) tdo-2 ΔPLD e

WT (H. sapiens) tdo-2 ΔALD

Figure 3. Deletion of PLD motif in TDO-2 is predicted to destabilise the interaction with haem which is essential for enzymatic activity. (a) Part of a multiple sequence alignment of TDO of different organisms and C. elegans TDO-2; yellow = tryptophan binding site, red = haem binding site, orange = binding site for haem and tryptophan, blue = residues missing in mutant tdo-2 (ΔPLD). (b) X-ray structure of the human TDO tetramer (PDB identifier 4PW8, separate monomers in light red, blue, grey and orange), in which a haem group has been added to the model according to the 3D structure of TDO from X. campestris (green, PDB identifier NW7). Positioned nearby the haem, also in green, is the loop of residues MSPLDF which includes the PLD residues missing in the tdo-2 (ΔPLD) mutant. Amino acids on the C-terminal side of this loop are indicated in red. (c) Enlarged image of the X-ray structure of X. campestris TDO which contains haem (green) and the substrate L-Trp (orange, PDB identifier 1NW8)19. The PSE motif (yellow) is equivalent to the PLD motif inC. elegans. Through Glu 112, the motif interacts with monomer D of the tetramer, whereas through Tyr 113 and Arg 117, it forms part of the TDO active site. (d) Left: Enlarged image of MSPLDF loop in wild type C. elegans TDO-2. Right: Enlarged image of MSPLDF loop in C. elegans tdo-2 (ΔPLD) mutant. (e) Left: Enlarged image of MTALDF loop in wild type human TDO. Right: Enlarged image of MTALDF loop in human tdo (ΔALD) mutant.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 88 interact with the α D helix, thereby emphasizing the importance of our finding for the design of novel specific TDO inhibitors21.

Upon studying the conservation of the PLD sequence in several species, we noted that in humans and many other species the proline is replaced by an alanine (resulting in ALD, Fig. 3a). Therefore, a model of human TDO with an ALD deletion (hTDO (Δ ALD)) (Fig. 3e) was generated as well. The ALD deletions shortens the loop connecting the α C and α D helices similar to the PLD deletion in the C. elegans enzyme. Loss of the aspartate (D) leads to a loss of connections to Arg 375, Lys 110 3 and Arg 117 in the neighbouring monomer, resulting in expected loss of the stability of the quaternary structure. Though the alanine in ALD does not provide intrinsic loop stability as the proline in PLD does, it is part of the N-terminal moiety of an α -helix. Disruption of this N-terminus through deletion of the ALD sequence will lead to destabilization of the entire helix and will impact the interaction of Arg 117 (X. campestris numbering, Arg140 in C. elegans and Arg 144 in H. sapiens) with the substrate (Fig. 3c–e). Deletion of the ALD motif will, therefore, likely abrogate activity in human TDO and other ALD containing orthologs.

Of the three CRISPR/Cas9-induced mutations in the tdo-2 gene studied here, two resulted in truncated proteins (early stop codon mutations). While the third mutant was lacking only three amino acids (PLD motif), it also had deficits when considering the conversion of tryptophan to kynurenine and a concomitant increase in motility and a prolonged lifespan. Thus, we have identified a stretch of amino acids in the TDO polypeptide chain that is situated in an evolutionary conserved loop that appears to be critical for this protein’s enzymatic activity.

The early stop codon mutations seem to distort the top half of the active site of TDO-2, preventing binding of tryptophan and haem. Depletion of the PLD motif introduces flexibility in the loop connecting the α D and α C helixes, thereby likely disturbing the α D helix and destabilizing the quaternary structure of the TDO-2 complex.

Interestingly, the loop containing the ALD motif of TDO has never before been identified as being crucial for enzyme activity. This finding will help to develop a new family of TDO-specific inhibitors that can target and destabilize this particular loop region, with the overall aim of treating TDO hyperactivity and tryptophan- associated diseases.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 89 METHODS Media and strains. Animals were maintained at 20 °C on Nematode Growth Medium (NGM) and fed with E. coli OP50.

Animals were age synchronised by hypochlorite treatment. The following strains were used in this study: Wild type N2, tdo-2 (del B) = OW715 tdo-2 (zg216)III; tdo-2 (del) = OW716 tdo-2 (zg217)III; tdo-2 (ΔPLD) = OW717 tdo-2 (zg218)III.

CRISPR/Cas9-induced mutagenesis. The target sequence was chosen as described in ref. 22. A target sequence in exon 3 (GCTCGACACAATGAGTCCAT) was selected for cloning. The target sequence was tested to be unique in the C. elegans genome (blast search on wormbase.org) to reduce the risk of off-target mutations. The target sequence was cloned into the pPD162 vector (C. elegans CRISPR/ Cas9 vector from ref. 22) using the Q5 site-directed mutagenesis kit (New England Biolabs, E0554S). We used a forward primer including the target sequence (5′ -ctcgacacaatgagtccatGTTTTAGAGCTAGAAATAGCAAGT-3′ ) and a general reverse primer (5′- CAAGACATCTCGCAATAGG-3′) . Successful cloning was confirmed by sequencing at the site of integration as well as sequencing different sites of the vector.

Young adult wild type animals were injected with the CRISPR/Cas9 injection mix (50 ng/μ l of pPD162 including the target sequence and 10 ng/μl of the CFP injection marker pPD136.61 (Fire Lab Vector Kit, 1999, linearised with ScaI) diluted in injection buffer23). The fluorescent injection marker was only used to control for successful injection. Single fluorescent progeny were isolated and checked for heterozygous mutations by sequencing and by the SURVEYOR® mutation detection kit from Transgenomics. Single progeny of heterozygous mutants were further isolated on individual plates and analysed for homozygous mutations (SURVEYOR® mutation detection and sequencing).

Metabolite measurements. Age-synchronised animals were transferred to NGM medium containing 5-fluoro-2′ deoxy-uridine (FUDR) on day 1 of adulthood (day after larval stage 4) to prevent offspring from growing. On day 4 of adulthood, about 2000 animals/experimental group were washed off the plates with PBS and extensively washed to remove as much bacteria from cuticle and gut as possible. cOmpleteTM protease inhibitor (Roche) was then added and the samples were snap frozen. Only samples from the same experiment round were compared. The samples were lysed by sonication, excessive worm debris was removed and the protein concentration was determined. The protein concentration of the different samples was equalised and tryptophan and kynurenine levels were measured by LC-MS/MS as described before8.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 90 Immunodetection of TDO-2 in C. elegans. Lysates were generated as described under metabolite measurements. Proteins were denatured for 10 minutes at 70 °C. Proteins were separated by SDS-PAGE (25 μg total lysate per lane) and blotted to a PVDF membrane. The blot was blocked with 5% milk powder in PBS-Tween (0.1%). For immunodetection of TDO-2, we generated a C. elegans-specific polyclonal antibody in rabbits (Eurogentec, NL13061) against the peptide SEHSNLSHSQSSESD. The antibody was diluted 1:10,000 in 5% milk/PBST and binding took place at room temperature for 1 hour. To visualise the binding of the anti-TDO-2 antibody, a HRP- 3 coupled anti-rabbit antibody (mouse) was used.

Reproductive lifespan assay. Ten animals per condition were tested for their reproduction behavior by transferring them every day to a fresh NGM plate. Eggs that were left behind could hatch and the number of hatched progeny was counted for the specific time period. Experiments were repeated and one representative graph is shown. Statistical analysis was by one-way ANOVA with post-hoc Bonferroni algorithm.

Motility. Age-synchronised animals were grown on FUDR-containing NGM medium from day 1 of adulthood onwards to prevent the growth of the offspring. Swimming assays were performed on 15 animals per experimental group on different days of adulthood as indicated. For the assay, a single worm was placed in a drop of M9 buffer and was allowed to acclimate for 30 seconds before its swimming movements were counted for another 30 seconds. Experiments were repeated 3 times and one representative graph is shown. Statistical analysis was by two-way ANOVA with post-hoc Bonferroni algorithm.

Lifespan experiments. 100 animals per experimental group were scored for their lifespan behaviour. Worms were age-synchronised and transferred to FUDR- containing NGM medium on day 1 of adulthood to prevent offspring from growing. The different mutations of the experimental groups were blinded before starting the experiment. At each time point, living animals were counted and dead animals (no longer showing nose touch response) removed. Animals that disappeared during the assay were excluded from the analysis. Experiments were performed at 20 °C and were repeated five times. One representative curve is shown. Curves were generated using Graphpad Prism software. Statistical analysis was by Log-rank (Mantel-Cox) algorithm.

Sequence alignment. The protein sequences for TDOs were retrieved from www.uniprot.org (H. sapiens: P48775, C. elegans: Q09474D. rerio: Q7SY53, D. melanogaster: P20351, X. leavis: Q5U4U6, M. musculus: P48776, R. norvegicus: P21643, accessed 18 Oct. 2016). Protein sequences were aligned using ClustalW2 or Clustal Omega. The default settings were applied (ClustalW accessed 18 July 2014,

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 91 Clustal Omega 18 Oct. 2016)24. Haem and tryptophan binding sites were adopted from ref. 19.

Homology models for wild type and mutant C. elegans TDO-2. BLAST was used to search for the TDO molecule that in terms of X-ray structure had the most homology to C. elegans TDO-225. This turned out to be human TDO (ref. 20; PDB identifier 4PW8) with a sequence identity of 48.4%. On basis of this structure, models of the wild type and mutant proteins were built automatically using the SwissModel server26 and checked manually. In addition, we used the same crystal structure (4PW8) to model the structure of a human TDO where amino acids ALD were deleted at the equivalent position of PLD in the C. elegans enzyme. However, the human TDO X-ray structure (PDB identifier 4PW8) does not contain a haem moiety, which is critical for the catalytic activity of the enzyme. We therefore used the Coot program to superimpose the models of all TDO structures on the basis of secondary structures27. This demonstrated that the haem moiety of X. campestris TDO19 is perfectly positioned over the catalytic site of C. elegans TDO-2. This haem moiety was therefore added to our 3D models of wild type and mutant C. elegans TDO-2. All pictures were generated using Pymol software28.

ACKNOWLEDGEMENTS We thank Yin Fai Chan for experimental support. We thank Sally Hill for editing the manuscript. This project was funded by the European Research Council (ERC) starting grant (to EAAN) and by support from the university alumni chapter “Gooische Groningers” through the Ubbo Emmius Fonds (to EAAN).

AUTHOR CONTRIBUTIONS H.M., R.I.S. and E.A.A.N. designed the experiments. H.M., R.B., M.K., J.U. and E.A.A.N. wrote the manuscript. H.M. and C.N.M. generated the CRISPR/Cas9 knockout mutants. H.M., and R.I.S. performed the motility, lifespan and reproduction assays and prepared the samples for metabolite measurements. M.K. performed lifespan assays and did statistical analyses. Mv.F. and I.K. measured the metabolite levels. Mv.F. performed the sequence alignment and R.B. and J.U. did the structure analysis including the molecular 3D models.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 92 REFERENCES

1. Seegers, N. et al. High-throughput fluorescence-based 15. Pilotte, L. et al. Reversal of tumoral immune resistance screening assays for tryptophan-catabolizing enzymes. by inhibition of tryptophan 2,3-dioxygenase. Proc Natl J Biomol Screen 19(9), 1266–74 (2014). Acad Sci USA 109(7), 2497–502 (2012). 2. Zhai, L. et al. Molecular pathways: Targeting IDO1 16. Yuasa, H. J. & Ball, H. J. Efficient tryptophan- and other tryptophan dioxygenases for cancer catabolizing activity is consistently conserved through immunotherapy. Clin Cancer Res 21(24), 5427–33 evolution of TDO enzymes, but not IDO enzymes. J. (2015). Exp. Zool. (Mol. Dev. Evol.) (2015). 3. Platten, M., von Knebel Doeberitz, N., Oezen, I., 17. Zhang, Y. et al. Crystal structure and mechanism Wick, W. & Ochs, K. Cancer immunotherapy by of tryptophan 2,3-dioxygenase, a heme enzyme 3 targeting IDO1/TDO and their downstream effectors. involved in tryptophan catabolism and in quinolinate Front Immunol 5, 673 (2015). biosynthesis. Biochemistry 46(1), 145–55 (2007). 4. Ren, S. & Correira, M. A. Heme: A regulator of rat 18. Chauhan, N. et al. Reassessment of the reaction hepatic tryptophan 2,3- dioxygenase. Archives of mechanism in the heme dioxygenases. J. Am. Chem. Biochemistry and Biophysics 377, 195–203 (2000). 131, 4186–4187 (2009). 5. Oxenkrug, G. F. The extended life span if Drosophila 19. Forouhar, F. et al. Molecular insights into melanogaster eye-color (white and vermilion) mutants substrate recognition and catalysis by tryptophan with impaired formation of kynurenine. J Neural 2,3-dioxygenase. Proc Natl Acad Sci USA 104, 473– Transm. 117(1), 23–26 (2010). 478 (2007). 6. Zwilling, D. et al. Kynurenine 3-monooxygenase 20. Meng, B. et al. Structural and functional analysis inhibition in blood ameliorates neurodegeneration. of human tryptophan 2,3-dioxygenase. Proteins: Cell 145, 864–874 (2011). Structure, Function and Bioinformatics 82, 3210– 7. Campesan, S. et al. The kynurenine pathway 3216 (2014). modulates neurodegeneration in a Drosophila model of 21. Dolušić, E. et al. Tryptophan 2,3-dioxygenase (TDO) Huntington’s disease. Curr Biol 21, 961–966 (2011). inhibitors. 3-(2-(pyridyl)ethenyl)índoles as potential 8. Van der Goot, A. T. et al. Delaying aging and the anticancer immunomodulators. J Med Chem 54(15), aging-associated decline in protein homeostasis by 5320–34 (2011). inhibition of tryptophan degradation. Proc Natl Acad 22. Dickinson, D. J., Ward, J. D. & Goldstein, B. Sci USA 109(37), 14912–7 (2012). Engineering the Caenorhabditis elegans genome 9. Breda, C. et al. Tryptophan-2,3-dioxygenase unsins Cas9-triggered homologous recombination. Nat (TDO) inhibition ameliorates neurodegeneration by Methods 10(10), 1028–34 (2013). modulation of kynurenine pathway metabolites. Proc 23. Mello, C. C., Kramer, J. M., Stinchcomb, D. & Natl Acad Sci USA 113(19), 5435–40 (2016). Ambros, V. Efficient gene transfer in C. elegans: 10. Van der Goot, A. T. & Nollen, E. A. A. Tryptophan extrachromosomal maintenance and integration of metabolism: entering the field of aging and age-related transforming sequences. EMBO J 10(12), 3959–70 pathologies. Trends Mol Med 19(6), 336–44 (2013). (1991). 11. Reinhold, J. A. & Rickels, K. Pharmacological 24. Larkin, M. A. et al. Clustal W and Clustal X version treatment for generalized anxiety disorder in adults: un 2.0. Bioinformatics 23(21), 2947–8 (2007). update. Expert Opin Pharmacother 16(11), 1669–81 25. Camacho, C. et al. BLAST+ : architecture and (2015). applications. BMC Bioinformatics 10, 421 (2008). 12. Wu, W. et al. Expression of tryptophan 2,3-dioxygenase 26. Schwede, T., Kopp, J., Guex, N. & Peitsch, M. C. and production of kynurenine pathway metabolites in SWISS-MODEL: an automated protein homology- triple transgenic mice and human Alzheimer’s disease modeling server. Nucleic Acids Res 31, 3381–3385 brain. PLoS One 8(4) (2013). (2003). 13. Maddison, D. C. & Giorgini, F. The kynurenine 27. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. pathway and neurodegenerative disease. Semin Cell Features and development of Coot. Acta Cryst D66, Dev Biol 40, 134–41 (2015). 486–501 (2010). 14. Oxenkrug, G. F. Increased plasma levels of xanthurenic 28. DeLano, W. L. The Pymol molecular graphics system. and kynurenic acids in type 2 diabetes. Mol Neurobiol DeLano Scientific, San Carlos, CA, USA, http://www. 52(2), 805–10 (2015). pymol.org (2002).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 93 SUPPLEMENTARY MATERIAL

Supplementary Figure S1 (a) Full blot of figure 1b, 3 min exposure (b) Replicate blot, including tdo-2 RNAi validation of the antibody, 1 min exposure

a b 100 100 * * * *

WT ** WT ** ** * ** ** *

80 tdo-2 (ΔPDL) 80 * tdo-2 (ΔPDL) * * ** * ** * ** ** ** tdo-2 (del) tdo-2 (del) 60 tdo-2 (del B) 60 tdo-2 (del B)

40 40

20 20 body bends/ 30 sec body bends/ 30 sec

0 0 day 2 day 4 day 8 day 4 day 6 day 8 day 12

Supplementary Figure S2 (a) Motility replicate 1; Statistics, two-way ANOVA (Genotype, Time, Interaction: p <0.001) with post-hoc Bonferroni (compared to WT) (b) Motility replicate 3; Statistics, two-way ANOVA (Genotype: p<0,001, Time: p<0,001, Interaction: p =0.2572) with post-hoc Bonferroni (compared to WT) , Error bars in panels= SEM. Statistics for all panels (comparison with WT): * <0.05; ** <0.01; *** <0.001

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 94 74 78 71 92 85 86 72 67 67 97 96 94 89 91 93 % scoredworms % value 0.4479 0.0650 0.0024 0.0002 0.0022 0.0006 0.0004 0.0592 0.0080 0.0021 0.0001 P <0.0001 <0.0001 <0.0001 <0.0001 3 0.52 0.44 0.53 0.32 0.44 0.43 0.28 0.25 0.21 0.34 0.31 0.29 0.41 0.47 0.43 SEM ------± Mean Mean 20.26 +/ 20.26 +/ 21.08 21.24 +/ 21.24 20.72 +/ 20.72 +/ 21.75 21.97 +/ 21.97 19.97 +/ 19.97 +/ 19.87 19.84 +/ 19.84 20.82 +/ 20.82 +/ 18.60 18.51 +/ 18.51 20.59 +/ 20.59 +/ 20.82 20.30 +/ 20.30 19.0 22.0 22.0 20.0 22.0 22.0 20.0 20.0 20.0 20.0 18.0 18.0 21.0 21.0 21.0 Median (del) (del) (del) (del) (del) (del B) (del (del B) (del (del B) (del (del B) (del (del B) (del ( ΔPLD ) ( ΔPLD ) ( ΔPLD ) ( ΔPLD ) ( ΔPLD ) 2 2 2 2 2 2 2 2 2 2 2 - 2 - 2 - 2 - 2 ------versus - - - - tdo tdo tdo tdo tdo tdo tdo tdo tdo tdo tdo tdo tdo tdo tdo mutated animals. * Experiments with more than 25% of the animals excluded because of rupture of rupture of because excluded animals of the 25% than more with Experiments * animals. mutated tdo-2 91 91 91 95 93 worms % scored% 0.38 0.30 0.33 0.35 0.31 SEM - - - - - ± Mean Mean 20.16 +/ 20.16 18.89 +/ 18.89 21.32 +/ 21.32 19.66 +/ 19.66 18.59 +/ 18.59 22.0 20.0 20.0 20.0 18.0 Median WT Str ain WT WT WT WT Summary of lifespan assays comparing wild type with type wild assays comparing lifespan of Summary * Experiment number Experiment 1* 2 3 4 5‡ Supplementary Table S3 Supplementary Table the animals.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 95 H. sapiens R. norvegicus M. musculus C. elegans D. melanogaster D. rerio X. laevis

H. sapiens R. norvegicus M. musculus C. elegans D. melanogaster D. rerio X. laevis

Supplementary Figure S4 Multiple sequence alignment of TDO; yellow = tryptophan binding site, red = haem binding site, orange = binding site for haem and tryptophan, blue = residues missing in mutant tdo-2 (ΔPLD).

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 96 3

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 97 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 98 Chapter IV

Tryptophan 2,3-dioxygenase regulates health- and lifespan independently of kynurenine, serotonin and tryptamine signaling in C. elegans

Helen Michels1, Renée I. Seinstra1, Dennis H. Lentferink1, Annemieke T. van der Goot1, Martijn van Faassen2, Ido P. Kema2, Ellen A.A. Nollen1

1European Research Institute for the Biology of Aging, University of Groningen, University Medical Centre Groningen, Laboratory of Molecular Neurobiology of Aging, The Netherlands 2University of Groningen, University Medical Centre Groningen, Department of Laboratory Medicine, The Netherlands

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 99 Due to a higher life expectancy, the number of people with age-related pathologies is increasing. Understanding the biological mechanisms that underlie the aging process will help to identify the cause of these diseases and reveal possibilities to treat them.

TDO (tryptophan 2,3- dioxygenase 2; TDO-2 in C. elegans) is the first enzyme of the kynurenine pathway, which regulates health- and lifespan in different model organisms. The benefits of TDO depletion have been suggested to depend on changes in levels of tryptophan metabolites, such as serotonin or the downstream products of the kynurenine pathway, summarized as the kynurenines, but evidence for this has been lacking.

This chapter demonstrates that serotonin and kynurenines in C. elegans do not play a role. Namely, depletion of other enzymes of the kynurenine pathway, which generate kynurenines, does not affect lifespan in a TDO-2 dependent manner. Furthermore, inhibition of serotonin synthesis showed a lifespan-extending effect but, again, independent of TDO-2. Healthspan regulation by TDO-2, as measured by the ability of animals to move as they age, was also independent of the presence of tryptamine, serotonin synthesizing enzymes, serotonin levels, and serotonin signaling pathways.

In summary, we show that the health- and lifespan-extending effects of reduced tdo-2 levels are mediated independently of the kynurenines, serotonin and the tryptophan- derived neuromodulator tryptamine. This suggests a new function of this enzyme independently of major tryptophan-degrading pathways, which can be used.

100

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 100 INTRODUCTION The tryptophan 2,3-dioxygenase (TDO; TDO-2 in C. elegans) is the rate-limiting enzyme of the kynurenine pathway (Figure 1A), mediating the first reaction of the degradation of tryptophan (Higuchi and Hayaishi, 1967). Whereas flies or worms only have TDO, mammals also express a second enzyme, ideolamine-2,3- dioxygenase (IDO), that only shares minor homology with TDO but contains a comparable active site and catalyzes the same reaction: the oxidation of tryptophan (Dang et al., 2000). 95% of free tryptophan that is not used for protein synthesis is degraded via the kynurenine pathway (Anderson et al., 2013). The activity of the first enzymes of 4 this pathway (TDO and IDO) is therefore highly influencing the homeostasis of tryptophan metabolism and controls internal free tryptophan levels (Van der Goot et al., 2012). Whereas IDO can bind different tryptophan-related molecules, including serotonin and tryptamine (Sono et al., 1996), TDO shows high specificity and only binds tryptophan itself (Sono et al., 1996). It was suggested that IDO developed in mammals due to its higher tissue complexity (Yuasa and Ball, 2015): While TDO is expressed in liver and brain to control global tryptophan levels (Danesch et al., 1987), IDO is ubiquitary expressed (Thackray et al., 2008) and activated in response to different stimuli (Yoshida et al., 1981; Hayaishi and Yoshida, 1978), e.g. to regulate the macrophage-mediated immune response (Mellor et al., 2001). The lack of IDO in flies and worms (summarized in Van der Goot and Nollen, 2013) makes them suitable candidates to study the role of tryptophan metabolism and especially the function of TDO.

Depletion of tdo-2 increased health- and lifespan in C. elegans (Van der Goot et al., 2012). It can very well be that changes in the metabolite composition might be responsible for the health- and lifespan-extending effects that we observe in tdo-2 depleted animals (Van der Goot et al., 2012) as tryptophan is a precursor for a range of metabolites, including the downstream products of the kynurenine pathway, the neurotransmitter serotonin and the neuromodulator tryptamine (summarized in Van der Goot and Nollen, 2013).

Metabolites of the kynurenine pathway, so-called kynurenines, and their altered abundance in different age-related diseases including neurodegeneration were already discussed in detail in the introduction of this thesis. Some kynurenines have been described as neurotoxic (3-hydroxy-kynurenine and quinolinic acid) whereas others (kynurenic acid) showed neuroprotective properties which has been linked to their agonistic respectively antagonistic binding to neuronal receptors (Schwarcz et al., 2012) as well as their pro- or anti-oxidant effects on reactive oxygen species (Ocampo et al., 2014). Depletion of tdo-2 changes the composition of these kynurenine metabolites (Van der Goot et al., 2012, Breda et al., 2016). In flies,

101

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 101 these metabolite changes have been shown to be at least partially responsible for healthspan benefits in model of Huntington’s disease (Breda et al., 2016). This is not the case for C. elegans (Van der Goot et al., 2012). We previously showed that only a knockdown of tdo-2 and none of the other tested enzymes of the kynurenine pathway (expected changes in metabolite levels were recorded) resulted in a motility increase. An additional depletion of tdo-2 in animals that show a mutation in kynurenine pathway genes improved their motility (Van der Goot et al., 2012), suggesting that the motility benefits of tdo-2 depletion are independent of the kynurenine pathway. It was however not tested yet, whether the lifespan extension, observed in flies and nematodes (Oxenkrug, 2010; Van der Goot et al., 2012), might be a result of the shift of kynurenine metabolites.

How could the changes in metabolite composition lead to the health- and lifespan extending effects that we observe in tdo-2 depleted animals? Which pathways might be controlled by the different tryptophan metabolites?

One such candidate is the aryl hydrocarbon receptor (AHR). AHR is highly conserved and was first identified as a transcription factor that is activated by different xenobiotics and pollutants inducing the detoxification response (summarized in Barouki et al., 2012). The unbound receptor is cytosolic and stabilized by a chaperone complex that facilitates the binding of ligands. Once a ligand attaches to the receptor, the complex falls partially apart and AHR translocates to the nucleus (Blankenship and Matsumura, 1997; Dong et al., 2011). In the nucleus, the remaining complex subunits dissociate and AHR bind to the AHR nuclear translocator (ARNT). The AHR/ARNT complex then associates with different co-activator and/or co- repressors and initiates the transcription of genes that are needed for the response to xenobiotic stress (Whitlock, 1999). Recent experiments showed that AHR is also able to bind endogenous metabolites and compounds besides the high variety of xenobiotics (Ocampo et al., 2014). That way, it was shown that AHR has other biological functions in cellular proliferation and migration, immune regulation and neuronal development and activity beside its role in detoxification (summarized in Barouki et al., 2012). Different tryptophan metabolites such as tryptamine (Nguyen and Bradfield, 2008), kynurenic acid (Moroni et al., 2012), and kynurenine (Opitz et al., 2011) were also found among those endogenous ligands. It was suggested that other tryptophan metabolites or tryptophan itself might be also part of the “ligandome” of AHR (pers. communication M. Platten). So far, a range of ligands of AHR have been identified, the biological function however of these interactions is not fully understood. In the group of tryptophan metabolites that bind or might bind to AHR, only the interaction with some kynurenines has been further studied. The binding of kynurenine to AHR reduces the expression of certain target genes to suppress the anti-tumor immune response in TDO-overexpressing tumors. Thus,

102

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 102 kynurenine promotes tumor cell survival and migration via AHR (Opitz et al., 2011). This example shows that the abundance of certain tryptophan metabolites is highly influencing the activity of AHR and suggests that changes in tryptophan metabolite composition induced by tdo-2 depletion could lead to health- and lifespan benefits via the AHR pathway.

Another pathway that might be involved in the control of health- and lifespan is induced by the tryptophan metabolite serotonin. Serotonin belongs to a group of monoamine compounds that are formed out of their respective precursor amino acids. This group also includes norepinephrine and dopamine (summarized in Capuron et al., 2011). In neurons members of this group of monoamine molecules 4 act as neurotransmitters mediating signals across the synaptic cleft (Berry, 2004). Serotonin can bind to 7 subfamilies of serotonin receptors composed of at least to 14 different members. Most of them are G-coupled receptors with exception of HTR3. The composition of receptors can be diverse and differs in different tissues and cell types (summarized in Cowen and Sherwood, 2013). Tryptophan is converted to 5-hydroxytryptophan and subsequently to 5-hydroxytryptamine (serotonin) by tryptophan hydroxylases (TPH). There are two TPH enzymes identified in human. TPH1 is expressed in peripheral tissue while TPH2 shows a neuronal localization (Ruddick et al., 2006). While serotonin itself cannot cross the blood brain barrier, tryptophan and the intermediate 5-hydroxytryptophan can (summarized in Mosienko et al., 2015). For this reason, changes in the abundance of serotonin precursors which occur when depleting tdo-2 are also affecting serotonin levels in the brain and control brain activity. The neurotransmitter function of serotonin in the brain has e.g. been associated with cognitive ability and regulation of the mood of an individual (Capuron et al., 2011). In C. elegans, the interplay of serotonin and dopamine were also described to control the animals’ motor function, in particular the transition from crawling to swimming movements and vice versa (Vidal-Gadea et al., 2011). Could the changes in serotonin levels also be responsible especially for the healthspan benefits (increased motility at higher ages) that we observe when depletingtdo-2 ?

Besides its role in the brain, serotonin has been identified as a global signaling molecule in a high variety of different organs, mediating diverse signals and influencing local metabolism (summarized in El-Merahbi et al., 2015). 98%of peripheric serotonin is stored in platelets whereas the remaining 2% is used as a hormone (autacoid function= local produced and local acting hormone) all over the organism (Berger et al., 2009). For example, serotonin transmits the nutrient- availability and composition of incoming food in the gut via the gut neuronal system to the rest of the organism (Neunlist and Schemann, 2014). It also controls the secretion of insulin from pancreatic cells, therefore influencing the glucose and lipid metabolism of the organism (Paulmann et al., 2009; Rorsman and Braun, 2013). This

103

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 103 is connected to its function in white adipocytes where it regulates the metabolism of fatty acids (Li et al., 2013), its function in brown adipocytes where it controls the glucose uptake (Oh et al., 2015) as well as its function in liver cells where serotonin inhibits the glucose uptake and stimulates the gluconeogenesis (Sumara et al., 2012). Furthermore, serotonin was described to promote glycolysis in muscle cells (Coelho et al., 2007). And finally, it is involved in the regulation of cytokine production in macrophages, therefore also being involved in the immune response of an organism (de las Casas-Engel et al., 2013). Taken all together, this shows that changes in serotonin levels by depleting tdo-2 and inhibiting the degradation of tryptophan via the kynurenine pathway, can highly influence the organismal homeostasis at several levels and might influence an animals’ behavior including its health- and lifespan.

Another possible route that might be responsible for the tdo-2 related phenotypes might be controlled by levels of tryptamine. Tryptamine belongs to a second group of endogenous amine compounds in neurons, the so-called trace amines. They are generated by enzymatic decarboxylation of their respective precursor amino acid (summarized in Mosienko et al., 2015). Trace amines have been considered as metabolic by-products for several decades due to their very low levels (Durden and Davis, 1993). However, the identification of trace amine specific receptors revived the research around this group of molecules (Borowsky et al., 2001; Bunzow et al., 2001). It was found that they function as neuromodulators, enhancing the excitatory activity of specific neurotransmitters which means that they have no effect on neurotransmission on their own but in combination with neurotransmitters such as serotonin or dopamine (Berry, 2004). Inhibition of synaptic signaling via neurotransmitters resulted in activation of the synthesis of trace amines by aromatic L-amino acid decarboxylase (AADC). The increase in trace amines then stimulated the neurotransmission in return (Jones, 1983). This feedback mechanism makes the inhibition of neurotransmission difficult (Berry, 2004). Deregulation of trace amines have been linked to several neuron-related diseases such as schizophrenia or depression as well as neurodegenerative disorders such as Parkinson’s disease (summarized in Berry, 2004). Could it be that an increase of tryptamine induced by inhibition of tryptophan degradation by depleting tdo-2 is responsible for the health- and lifespan benefits and especially for these that we observed in various models of age-related diseases (e.g. Parkinson’s disease model)?

We show here, that a depletion in tdo-2 results in health- and lifespan benefits independently of changes in levels of serotonin and tryptamine and metabolites of the kynurenine pathway.

104

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 104 RESULTS AND DISCUSSION In this chapter, we tested whether the phenotypes that we observe in tdo-2 depleted animals are caused by changes in tryptophan metabolites (Figure 1A) and a result of a signaling cascade that has been induced by one such a tryptophan metabolite.

First, we tested if a shift in kynurenine metabolites is responsible for the lifespan- extending effects of tdo-2 depletion. For these experiments we used mutants where metabolite measurements showed that the specific enzyme was indeed fully blocked (Van der Goot et al., 2012). We previously showed that the increase in motility was independent of kynurenines (Van der Goot et al., 2012). We observed comparable 4 effects by testing mutants of the kynurenine pathway for their lifespan behavior: none of them increased lifespan except for deletion mutants for tdo-2 (Figure 1B). Furthermore, a depletion of tdo-2 in animals with different mutations in kynurenine pathway genes still extended lifespan (Figure 2). This suggests that changes in the metabolite levels of the kynurenine pathway are also not involved in the regulation of lifespan by tdo-2 in C. elegans and that the phenotypes are controlled independently of the kynurenine pathway.

A B 100 WT tdo-2 (del) 80 tdo-2 (del B) 60

40 HUMAN/ C. elegans

percent survival 20

0 serotonin tryptamine tryptophan 0 10 20 30 40 50 days of adulthood TDO/ IDO/ TDO-2

100 100 AFMID/ AMFD-1 WT WT amfd-1 (del) kmo-1 (del) 80 80 kynurenine kynurenic acid 60 60 KMO/ KMO-1 KYNU/ FLU-2 40 40 3-hydroxyanthranillic percent survival 20 percent survival 20 kynurenine acid 0 0 KYNU/ FLU-2 0 10 20 30 40 0 10 20 30 40 days of adulthood days of adulthood 3-hydroxyanthranillic acid 100 WT 100 WT HAAO/ HAAO-1 flu-2 (del) haao-1 (del) 80 80

60 60

40 40 percent survival 20 percent survival 20

0 0 0 10 20 30 0 10 20 30 40 days of adulthood days of adulthood

Figure 1. Influence of the kynurenine pathway on lifespan (A) Schematic representation of the kynurenine pathway in human and C. elegans. (B) Survival curves of mutants of the kynurenine pathway.

105

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 105 WT/ ctr RNAi WT/ ctr RNAi 100 100 A WT/ tdo-2 RNAi WT/ tdo-2 RNAi 80 amfd-1 (del)/ ctr RNAi 80 kmo-1 (del)/ ctr RNAi amfd-1 (del)/ tdo-2 RNAi kmo-1 (del)/ tdo-2 RNAi 60 60

40 40 percent survival percent survival 20 20

0 0 0 10 20 30 40 0 10 20 30 40 days of adulthood days of adulthood

WT/ ctr RNAi 100 100 WT/ ctr RNAi WT/ tdo-2 RNAi WT/ tdo-2 RNAi 80 flu-2 (pm)/ ctr RNAi 80 haao-1 (del)/ ctr RNAi flu-2 (pm)/ tdo-2 RNAi haao-1 (del)/ tdo-2 RNAi 60 60

40 40 percent survival 20 percent survival 20

0 0 0 10 20 30 40 50 0 10 20 30 40 days of adulthood days of adulthood

Figure 2. TDO-2 regulates lifespan independently of the kynurenine pathway (A) tdo-2 dependency of survival curves of mutants of the kynurenine pathway.

These results were slightly surprising as experiments in other model organisms showed an involvement of the kynurenine pathway in lifespan regulation. Depletion of cinnabar (KMO in flies) increased longevity in Drosophila (Oxenkrug, 2010). It is not clear why similar phenotypes were not observed in C. elegans taking into account the high conservation of the kynurenine pathway. It does not mean that the kynurenine pathway is not involved in any phenotype in C. elegans, but they do not control the motility and lifespan effects as it is the case when inhibiting tdo-2. It is possible that the kynurenine metabolites are more important in Drosophila. One could imagine e.g. that the neuroprotective KAT-branch got more importance with the development of more specialized neuronal activity. Besides those differences in phenotypes, experiments in all organisms however showed similar effects when depleting TDO, suggesting overlapping biological functions at least for this enzyme throughout evolution.

In line with testing mutants of the kynurenine pathway, we also assayed ahr-1(del) mutants as it has been shown that this receptor is activated by several different tryptophan metabolites such as kynurenines (Nguyen and Bradfield, 2008; Opitz et al., 2011). The question was if AHR-1 might be activated by tdo-2 depletion because of changes in the kynurenines or increased tryptophan levels or alterations in other tryptophan metabolites levels. We tested two different ahr-1(del) mutants (Figure 3B). Both mutants were not capable to restore lifespan (Figure 3C) or healthspan of tdo-2 depleted animals (Figure 3D) to normal levels. Therefore, we concluded that the regulation of these phenotypes by tdo-2 is independent of a transcriptional response caused by activation of AHR-1.

106

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 106 A tryptophan metabolites translocation to the nucleus transcriptional AHR-1 response

C 4 ahr-1 (del. A) ahr-1 (del. B)

D ns 70 ns 60 ns 50 *** *** 40 *** 30 *** ** 20 *

body bends/30 sec bends/30 body 10 0 RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi -2 -2 -2 -2 -2 -2 -2 -2 -2 ctr ctr ctr ctr ctr ctr ctr ctr ctr tdo tdo tdo tdo tdo tdo tdo tdo tdo WT WT WT (del A) (del A) (del A) (del B) (del B) (del B) -1 -1 -1 -1 -1 -1 ahr ahr ahr ahr ahr ahr

day 4 day 8 day 12

Figure 3. TDO-2 regulates health- and lifespan independently of AHR-1 (A) Schematic model of AHR-1 activation by tryptophan metabolites and others. (B) Intron/exon structure of AHR-1 including the location of the mutations of the deletion mutants. (C) Survival curves of ahr-1 deletion mutants treated with ctr or tdo-2 RNAi. (D) Motility/ healthspan behavior including tdo-2 dependency. Error bars= SEM, statistics: *<0.05, **<0.01, ***<0.001.

Second, as mentioned before, a knockdown of tdo-2 increases organismal tryptophan levels. We showed before that supplying wild type animals with increasing amounts of tryptophan could phenocopy the motility increase that was achieved by tdo-2 depletion (Van der Goot et al., 2012). We therefore asked whether the free tryptophan might be converted into other metabolites independently of the kynurenines.

107

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 107 Tryptophan is a precursor for the neurotransmitter serotonin and the neuromodulator tryptamine (Figure 4A). The idea that inhibition of tryptophan degradation via the kynurenine pathway is increasing free tryptophan levels and subsequently increasing the amount of free tryptophan levels in the brain (competition with other large neutral amino acids (LNAA) for active transport via the blood brain barrier) was a common strategy to elevate serotonin levels in neuron (summarized in Young, 2013). We observed, that by depleting tdo-2 in C. elegans, thus increasing tryptophan levels, internal serotonin levels are slightly increased (Van der Goot et al., 2012).

We tested whether the production of serotonin might be responsible for the phenotypes we observe. A tph-1(del) mutant also lives longer (Figure 4B) than wild type animals. However, we showed before that the lifespan extension by tdo- 2 depletion is independent of the serotonin synthesis (tdo-2 depletion effect on lifespan is not inhibited by tph-1(del) mutation; Van der Goot et al., 2012) suggesting that there are two independent mechanisms that lead to the same lifespan-extending phenotype. The interplay between the two neurotransmitters serotonin and dopamine are controlling the coordination of movements in C. elegans (Vidal-Gadea et al., 2011). A tph-1(del) mutant has severe deficits in motility which can be partially rescued by tdo-2 depletion (Van der Goot et al., 2012) which, again, shows that healthspan regulation by TDO-2 is independent of serotonin. However, the motility defects are as severe and swimming movements are at their lower limit that the interpretation of such an experiment is not as clear and should be supported by other experiments. We therefore asked whether an increase in serotonin by feeding would result in a higher motility: However supplementing wild type animals with serotonin resulted in increased internal serotonin levels but did not increase motility (Figure 4C) which supports previous results from genetic experiments.

Supplementation of serotonin is one possibility to increase internal levels. However, serotonin is only soluble under very acidic conditions which might influence the behavior of the animals. Therefore, it might be still a good idea to increase serotonin levels by higher TPH-1 activity. This was tested by the group of Hiroyuki Hasegawa. Instead of elevating the amount of precursor molecules (tryptophan), they tried to modulate the activity of TPH, the enzyme that produces serotonin, by supplying a stimulating co-factor (Kobayashi et al., 1991). Furthermore, it is possible to increase the amount of free serotonin in the synaptic cleft by inhibiting the active removal via the serotonin transporter SERT (De Montigny et al., 1990). An overexpression of SERT might be another possibility to reduce serotonergic signaling.

We finally excluded serotonin-dependent neurotransmission being responsible for the phenotypes of tdo-2 depletion in C. elegans by inhibiting genes that encode serotonergic receptors: Deletions in genes that encode for serotonergic receptors also increase the motility of the animals, but a knockdown of tdo-2 could increase

108

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 108 motility in all cases (Figure 4D).

But here again, results have to be interpreted with care because it was shown though, that inhibition of the serotonergic receptors also increased the levels of free serotonin in the synaptic cleft (summarized in El-Merahbi et al., 2015). Taken into account, that there is a high number of different serotonergic receptors that also show overlapping expression profiles in different tissues (Cowen and Sherwood, 2013), we might have provoked an increased signaling through other serotonergic receptors by inhibiting specific ones. Therefore, it might be necessary to block several receptors

aromatic A B 100 WT L-amino tryptophan 4 tph-1 (del) decarboxylase hydroxylase 80

60 AADC TPH-1 40 L-tryptamine L-tryptophan serotonin

percent survival 20

0 0 10 20 30 40 days of adulthood

C serotonin D RNAi from L4 addition internal ctr RNAi tdo-2 RNAi from L4 serotonin 80 *** levels 100 ns *** * 800 60 *** *** 80 *** * *** * *** *** *** * 600 60 40 * * 40 400 20 20 200 sec 30 bends/ body body bends/ 30 sec 30 bends/ body serotonin (in nM) (in serotonin

0 0 0 WT (del) (del) (del) (del) (del) (del) (del) (del) (del) (del) (del) (del) (del) (del) 0 mM 0 mM -5 -1 -1 15 mM -5 15 mM -7 -1 -1 -2 -3 -4 -5 -6 -7 40 - ser ser cat ser ser ser ser ser ser pah dop lgc mod mod (del)/

E tryptamine -1 addition ser from L4 internal tryptamine ns levels 50 *** 60 50 40 40 30 30 20 20

10 tryptamine (in nM) body bends/ 30 sec 30 bends/ body 10 0 0 0 mM 0 mM 1 mM tryptamine 1 mM tryptamine 35 mM tryptophan

Figure 4. TDO-2 regulates health independently of serotonin and tryptamine (A) Schematic representation of tryptophan conversion to serotonin and tryptamine. (B) Survival curves of tph-1 deletion mutants compared to WT animals. (C) Motility/ healthspan on day 4 of adulthood of wild type animals fed with serotonin and their internal serotonin levels. (D) Motility/ healthspan on day 4 of adulthood of different serotonergic receptor mutants treated with ctr and tdo-2 RNAi. (E) Motility/ healthspan on day 4 of adulthood of wild type animals treated with tryptophan (positive control) and tryptamine and their internal tryptamine levels. Error bars= SEM, statistics: *< 0.05, ***<0.001.

109

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 109 simultaneously to assess the consequences of inhibiting serotonergic signaling. It is not clear, whether an inhibition of all serotonergic receptors might be possible concerning the global role of serotonin (El-Merahbi et al., 2015).

The sum of the here shown experiments however suggests that the beneficial effects on motility by tdo-2 depletion in C. elegans are not likely to be related to serotonin signaling. Tryptamine is a trace amine that only occurs in very low, almost non- detectable levels (Durden and Davis, 1993). We were therefore not able to check whether a knockdown of tdo-2 would result in increased levels of tryptamine. The corresponding enzyme that converts tryptophan into tryptamine is also not identified in C. elegans yet. We therefore could only supply animals with external tryptamine. Even though we could measure a high increase in internal tryptamine levels, we did not see an effect on motility as it was the case for tryptophan (Figure 4E). It is though very likely that the healthspan benefits bytdo-2 depletion are independent of tryptamine synthesis.

Taken together, the increase in motility and the extension in lifespan in tdo-2 depleted animals is most likely independent of the known tryptophan metabolites serotonin and tryptamine as well as the kynurenines.

110

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 110 Experimental procedures Media and strains Animals were maintained at 20˚C on Nematode Growth Medium (NGM) and fed with E. coli OP50.

Animals were age synchronized by hypochlorite treatment.

The following strains were used in this study:

Wild type N2; tdo-2 (del B) =OW715 tdo-2 (zg216)III; tdo-2 (del) =OW716 tdo-2 (zg217)III; OW479 haao-1 (tm4627)V; OW478 kmo-1 (tm4529)V; OW477 amfd-1 4 (tm4547)IV; CB1003 flu-2 (e1003)X; ZG24 ahr-1 A (ia3)I; CZ2485 ahr-1 B (ju145) I; DA1814 ser-1 (ok345)X; RB1690 ser-2 (ok2103)X; DA1774 ser-3 (ad1774) I; AQ866 ser-4 (ok512)III; RB2277 ser-5 (ok3087)I; FX2146 ser-6 (tm2146)IV; DA2100 ser-7 (tm1325)X; DA2109 ser-7 (tm1325) ser-1 (ok345)X; MT9668 mod- 1 (ok103)V; RB785 dop-5 (ok568)V; MT14678 lgc-40 (n4545)X; MT9772 mod-5 (n3314)I; LC73 pah-1 (tm520)II; RB681 cat-1 (ok411)X. RNA interference RNAi experiments were performed on NGM plates containing isopropylthio-β-D- galactoside (IPTG, 15mg/L) and 50μg/ml ampicillin. Plates were seeded with RNAi bacteria. Prior to the experiment, the plates were kept at room temperature for 2 days to allow the production of dsRNA by the bacteria. Plates were produced as freshly as possible (maximum 1 week before start of the experiment). Age-synchronized animals were either grown directly on RNAi food from larval stage L1 onwards or transferred to the RNAi treatment at larval stage L4 or day 1 of adulthood (indicated for specific experiments). For each experiment, the time point at which RNAi treatment started was determined depending on the results of a test run to see whether such treatment affected development. Animals of day 1 of adulthood (day after L4 stage) were transferred to RNAi plates containing FUDR to prevent the offspring from growing. RNAi clones were used from the Ahringer C. elegans RNAi (Fraser et al., 2000). All clones were verified by sequencing and animals were scored for associated phenotypes during the experiments. Metabolite feeding Serotonin and tryptamine was supplemented from larval stage 4 onwards. Animals were kept on FUDR-containing plates from day 1 of adulthood onwards to prevent offspring from growing. Worms were transferred to fresh metabolite plates every day. Serotonin and tryptamine were added to the plates every day prior to the worm transfer. Serotonin (15mM) and tryptamine (1mM) were solved in a 0,1 M HCl solution. 300µl of the solution was pipetted on a 6cm plates seeded with OP50, distributed and plates were stored until the liquid was completely dried into the

111

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 111 medium. Control plates were consequently treated with 0,1 M HCl solution. Worms were washed off the previous plate with M9 and transferred to the fresh plate by pipetting. Motility assays were performed on day 4 of adulthood as well as samples were collected for metabolite measurement. Metabolite measurements Age-synchronized animals were transferred to NGM medium containing 5-fluoro- 2’deoxy-uridine (FUDR) on day 1 of adulthood (day after larval stage 4) to prevent offspring from growing. About 2000 animals on day 4 of adulthood were washed of the plates with PBS and extensively washed to remove as much bacteria from cuticle and gut. cOmpleteTM protease inhibitor (Roche) was finally added and the samples were snap frozen. Only samples from the same experiment were compared. The samples were lysed by sonication, excessive worm debris was removed and the protein concentration was determined. The protein concentration of the different samples was equalized and metabolite levels were measured by LC-MS as described before (Van der Goot et al., 2012). Motility/ healthspan Age-synchronized animals were grown on FUDR-containing NGM medium from day 1 of adulthood to prevent the growth of the offspring. Swimming assays were performed on 15 animals per experimental group on different days of adulthood. A worm was placed in a drop of M9. The animal was allowed to acclimate for 30 seconds before its swimming movements were counted for another 30 seconds. Experiments were repeated and one representative graph is shown. Statistic analysis was done with a one-way ANOVA with post-hoc Bonferroni algorithm. Lifespan experiments 100 animals per experimental group were scored for their lifespan behavior. Worms were age-synchronized and transferred to FUDR-containing NGM medium on day 1 of adulthood to prevent offspring from growing. On every time point, living animals were counted and dead animals (no response to nose touch anymore) removed. Animals that disappeared during the assay were excluded from the analysis. Experiments were performed at 20˚C and were repeated. One representative curve is shown. Curves were generated by Graphpad Prism.

112

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 112 REFERENCES

• Anderson, G et al.. Increased IL-6 trans-signaling in macrophage polarization through HTR2B and HTR7. depression: focus on the tryptophan catabolite pathway, J Immunol 190, 2301-2310 (2013). melatonin and neuroprogression. Pharmacological • De Montigny, C, Chaput, Y, Blier, P: Modification of Reports 65, 1647-1654 (2013). serotonergic neuron properties by long-term treatment • Barouki, R, Aggerbeck, M, Aggerbeck, L, Coumoul, with serotonin reuptake blockers. J Clin Psychiatry 51, X: The aryl hydrocarbon receptor system. Drug Metab 4-8 (1990). Drug Interact 27(1), 3-8 (2012). • Dong, B et al.: FRET analysis of protein tyrosine • Berger, M, Gray, JA, Roth, BI: The expanded biology kinase c-Src activation mediated via aryl hydrocarbon of serotonin. Annu Rev Med 60, 355-366 (2009). receptor. Biochim Biophys Acta 1810, 427-31 (2011). • Berry, MD: Mammalian central nervous system trace • Durden, DA, Davis, BA: Determination of regional amines. Pharmacologic amphetamines, physiologic distributions of phenylethylamine and meta- and neuromodulators. J Neurochem 90(2), 257-71 (2004). para-tyramine in rat brain regions and presence in 4 • Blankenship, A, Matsumura, F: human and dog plasma by an ultra-sensitive negative 2,3,7,8-Tertrachlorodibenzo-p-dioxin (TCDD) causes chemical ion gas chromatography-mass spectrometric an Ah receptor-dependent and ARNT-independent (NCI-GC-MS) method. Neurochem Res 18, 995-1012 increase in membrane levels and activity of p60(Src). (1993). Environ Toxicol Pharmacol 3, 211-20 (1997). • El-Merahbi, R, Löffler, M, Mayer, A, Sumara, G: The • Borowsky, B et al.: Trace amines: identification of a roles of peripheral serotonin in metabolic homeostasis. family of mammalian G protein-coupled receptors. FEBS Letters 589, 1728-1734 (2015). Proc Natl Acad Sci U S A 98, 8966-8971 (2001). • Fraser, AG et al.: Functional genomic analysis • Breda, C et al.: Tryptophan-2,3-dioxygenase of C. elegans chromosome I by systematic RNA (TDO) inhibition ameliorates neurodegeneration by interference. Nature 408, 325-30 (2000). modulation of kynurenine pathway metabolites. Proc • Giorgini, F, Guidetti, P, Nguyen, Q, Bennett, SC, Natl Acad Sci U S A 113(19), 5435-40 (2016). Muchowski, PJ: A genomic screen in yeast implicates • Bunzow, JR et al.: Amphetamine, kynurenine 3-monooxygenase as a therapeutic target 3,4-methylenedioxyamphetamine, lysergic acid for Huntington disease. Nat Genet 37, 526-531 (2005). diethylamide, and metabolites of the catecholamine • Hayaishi, O, Yoshida, R: Specific induction of neurotransmitters are agonists of a rat trace amine pulmonary indoleamine 2,3-dioxygenase by bacterial receptor. Mol Pharmacol 60, 1181-1188 (2001). lipopolysaccharide. Ciba Found Symp 65, 199-203 • Campesan, S et al.: The kynurenine pathway (1978). modulates neurodegeneration in a Drosophila model • Higuchi, K, Hayaishi, O: Enzymatic formation of of Huntington’s disease. Curr Biol 21, 961-966 (2011). D-kynurenine from D-tryptophan. Arch Biochem • Capuron, L et al.: Chronic low-grade inflammation in Biophys 120(2), 397-403 (1967). elderly persons is associated with altered tryptophan • Jones, RSG: Trace biogenic amines: a possible and tyrosine metabolism: role of neuropsychiatric functional role in the CNS. Trends Pharmacol Sci 4, symptoms. Biol Psychiatry 70, 175-182 (2011). 426-429 (1983). • Coelho, WS, Costa, KC, Sola-Penna, M: Serotonin • Kobayashi, T, Hasegawa, H, Kaneko, E, Ichiyama, stimulates mouse skeletal nuscle 6-phosphofructo- A: Gastrointestinal serotonin: depletion due 1-kinase through tyrosine-phosphorylation of the to tetrahydrobiopterin deficiency induced by enzyme altering its intracellular localization. Mol 2,4-diamino-6-hydroxypyrimidine administration. J Genet Metab 92, 364-370 (2007). Pharmacol Exp Ther 256(2), 773-9 (1991). • Cowen, P, Sherwood, AC: The role of serotonin in • Li, Q, Hosaka, T, Harada, N, Nakaya, Y, Funaki, cognitive function: evidence from recent studies M: Activation of Akt through 5-HT2A receptor and implications for uncerstanding depression. J ameliorates serotonin-induced degradation of Psychopharmacol 27(7), 575-83 (2013). insulin receptor substrate-1 in adipocytes. Mol Cell • Danesch, U et al.: Glucocorticoid induction of the rat Endocrinol 365, 25-35 (2013). tryptophan oxygenase gene is mediated by 2 widely • Mellor, AL et al.: Prevention of T cell-driven separated glucocorticoid-responsive elements. EMBO complement activation and inflammation by J 6, 625-630 (1987). tryptophan catabolism during pregnancy. Nat Immunol • Dang, Y, Dale, WE, Brown, OE: Comparative 2, 64-68 (2001). effects of oxygen on indoleamine 2,3-dioxygenase • Moroni, F, Cozzi, A, Sili, M, Mannaioni, G: Kynurenic and tryptophan 2,3-dioxygenase of the kynurenine acid: a metabolite with multiple targets in the brain pathway. Free Radical Biology and Medicine 28, 615- and periphery. Journal of Neural Transmission 119(2), 624 (2000). 133-139 (2012). • De las Casas-Engel, M et al.: Serotonin skews human • Mosienko, V et al.: Life without brain serotonin:

113

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 113 Reevaluation of serotonin function with mice deficient • Sono, M, Roach, MP, Coulter, ED, Dawson, JH: brain serotonin synthesis. Behavioural Brain Research Heme-containing oxygenases. Chem Rev 96, 2841- 277, 78-88 (2015). 2887 (1996). • Neunlist, M, Schemann, M: Nutrient-induced changes • Sumara, G, Sumara, O, Kim, JK, Karsenty, G: Gut- in the phenotypes and function of the enteric nervous derived serotonin is a multifunctional determinant to system. J Physiol 592(14), 2959-2965 (2014). fasting adaptation. Cell Metab 16, 588-600 (2012). • Nguyen, LP, Bradfield, CA: The search for endogenous • Thackray, SJ, Mowat, CG, Chapman, SK: Exploring activators of the aryl hydrocarbon receptor. Chem Res the mechanism of tryptophan 2,3- dioxygenase. Toxicol 21, 102-116 (2008). Biochemical Society Transactions 36, (2008). • Ocampo, JR et al.: Kynurenines with neuroactive • Van der Goot, AT et al.: Delaying aging and the aging- and redox properties: relevance to aging and brain associated decline in protein homeostasis by inhibition diseases. Oxid Med Cell Longev, (2014). of tryptophan degradation. Proc Natl Acad Sci U S A • Oh, CM et al.: Regulation of systemic energy 109(37), 14912-7 (2012). homeostasis by serotonin in adipose tissues. Nat • Van der Goot, AT and Nollen, EAA: Tryptophan Commun 6, 6794 (2015). metabolism: entering the field of aging and age-related • Opitz, CA et al.: An endogenous tumour-promoting pathologies. Trends Mol Med 19(6), 336-44 (2013). ligand of the human aryl hydrocarbon receptor. Nature • Vidal-Gadea, A et al.: Caenorhabditis elegans selects 478(7368), 197-203 (2011). distinct crawling and swimming gaits via dopamine • Oxenkrug, GF: The extended life span if Drosophila and serotonin. Proc Natl Acad Sci U S A 108(42), melanogaster eye-color (white and vermilion) mutants 17504-9 (2011). with impaired formation of kynurenine. J Neural • Whitlock, JP: Induction of cytochrome P4501A1. Transm. 117(1), 23-26 (2010). Annu Rev Pharmacol Toxicol 39, 103-25 (1999). • Paulmann, N et al.: Intracellular serotonin modulates • Yoshida, R et al.: Induction of pulmonary indoleamine insulin secretion from pancreatic beta-cells by protein 2,3-dioxygenase by interferon. Proc Natl Acad Sci U S serotonylation. PLoS Biol 7, (2009). A 78, 129-132 (1981). • Rorsman, P, Braun, M: Regulation of insulin secretion • Young, SN: Acute tryptophan depletion in humans: a in human pancreatic islets. Annu Rev Physiol 75, 155- review of theoretical, practical and ethical aspects. J 179 (2013). Psychiatry Neurosci 38(5), 294-305 (2013). • Ruddick, JP et al.: Tryptophan metabolism in the • Yuasa, HJ and Ball, HJ: Efficient tryptophan- central nervous system: medical implications. Expert catabolizing activity is consistently conserved through Rev Mol Med 8(20), 1-27 (2006). evolution of TDO enzymes, but not IDO enzymes. J. • Schwarcz, R, Bruno, JP, Muchowski, PJ, Wu, Exp. Zool. (Mol. Dev. Evol.), (2015). HQ: Kynurenines in the mammalian brain: when • Zwilling, D et al.: Kynurenine 3-monooxygenase physiology meets pathology. Nat Rev Neurosci 13(7), inhibition in blood ameliorates neurodegeneration. 465-77 (2012). Cell 145, 864-874 (2011).

114

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 114 4

115

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 115 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 116 Chapter V

Tryptophan 2,3-dioxygenase regulates motility and lifespan independently of protein translation in C. elegans

Helen Michels1, Renée I. Seinstra1, Tristan de Jong2, Dennis H. Lentferink1, Mandy Koopman1, Victor Guryev2, Ellen A.A. Nollen1

1European Research Institute for the Biology of Aging, University of Groningen, University Medical Centre Groningen, Laboratory of Molecular Neurobiology of Aging, The Netherlands 2European Research Institute for the Biology of Aging, University of Groningen, University Medical Centre Groningen, Laboratory of Genome Structure of Aging, The Netherlands

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 117 Understanding the cellular mechanisms that regulate aging gained more and more importance because of an increasing average age of our population and higher incidences of age-related pathologies.

Depletion of the tryptophan-degrading enzyme TDO-2 (tryptophan 2,3-dioxygenase) increases lifespan and delays age-related diseases such as neurodegeneration and cancer in model organisms, which includes C. elegans and mice. By reducing TDO- 2, the major tryptophan-degrading pathway –the kynurenine pathway- is inhibited and tryptophan levels are increased. Tryptophan is the least abundant essential amino acid and its levels are therefore suggested to be of high importance for protein translational control. A reduction of TDO-2 might therefore be beneficial for health by increasing protein translational rates. Here, we show that the health-and lifespan increasing effects achieved by a depletion of tdo-2 in C. elegans are independent of tryptophan uptake via transporter F23F1.6, the loading of free amino acids to tRNAs as well as cytosolic and mitochondrial translation controlled by AMPK, GCN2 and mTOR.

118

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 118 INTRODUCTION Depletion of the tryptophan-degrading enzyme TDO (tryptophan 2,3-dioxygenase) increases lifespan and – as well as supplementation of tryptophan - delays pathological features of age-related diseases such as neurodegeneration or cancer in multiple model organisms (Giorgini et al., 2005; Campesan et al., 2011; Zwilling et al., 2011; Van der Goot et al., 2012; Breda et al., 2016).

This chapter focuses on the question whether the effects of tdo-2 depletion and tryptophan supplementation are a result of changes in protein translation. We tested the uptake of tryptophan into the cell, protein translation by modulating for example the abundance of different tRNA synthetases as well as the involvement of key regulators of cytosolic and mitochondrial translation, namely mTOR, GCN-2 and 5 AMPK and components that insure the balance of the two types of protein translation.

Tryptophan is the least-abundant essential amino acid. Its uptake into the cell is facilitated via membrane-crossing transporters. There are two transporters described in C. elegans: C50D2.2 and F23F1.6. Chong He and colleagues showed that a reduced uptake of tryptophan via these transporters prolongs the lifespan of the animals (He et al., 2014). Within the cell, tryptophan is used for several metabolic pathways but also for the synthesis of proteins. 95% of tryptophan that is not used for protein production is metabolized via the kynurenine pathway (Anderson et al., 2013). TDO (TDO-2 in C. elegans) is the first rate-limiting enzyme of this pathway. The depletion of TDO-2 results in increased organismal tryptophan levels in C. elegans (Van der Goot et al., 2012). This pool of tryptophan is then possibly available for the synthesis of other tryptophan metabolites such as e.g. serotonin but also for the production of proteins.

During translation free amino acids are binding to their specific transfer RNA (tRNA) which is catalysed by specific tRNA synthetases. The amino acid-bound tRNA is then transported to the ribosome where the amino acid is integrated into the growing polypeptide chain of a new protein (summarized in Pang et al., 2014). The cell possesses two independent translation machineries: a cytosolic- and a mitochondrial- specific system. The interplay between these two is especially important for a balanced synthesis of mitochondrial proteins that are nuclear (translation in the cytosol) as well as mitochondrial (translation in the mitochondria) encoded. The mitochondrial unfolded protein response (UPR) is a stress signal transduction pathway that gives feedback about the proper activity of the mitochondria and a possible underlying imbalance of nuclear and mitochondrial encoded proteins (Houtkooper et al., 2010). The induction of UPR has been shown to increase lifespan in C. elegans (Hamilton et al., 2005). Is it possible that the depletion of tdo-2 induces an imbalance of cytosolic

119

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 119 and mitochondrial protein translation which results in mitochondrial UPR and is responsible for the lifespan-extending phenotype of tdo-2 depleted animals?

Two kinases – the general control nonderepressible-2 kinase 2 (GCN2, GCN-2 in C. elegans) and the mammalian target of rapamycin (mTOR, LET-363 in C. elegans) - are the major regulators of cytosolic translation (Sattlegger and Hinnebusch, 2000). In the case of amino acid deprivation, the number of free tRNAs increases which activates GCN2 (Rousakis et al., 2013). It was shown that GCN2 reacts especially on free tryptophanyl-tRNAs (Fougeray et al., 2012). As a consequence, GCN2 phosphorylates the alpha-subunit of eukaryotic translation initiation factor 2 (eIF2α) and inhibits general, cytosolic translation (Lee et al., 2010). In parallel the machinery prioritizes the translation of specific mRNAs that are needed to react on the stress condition e.g. GCN4 in yeast (Natarajan et al., 2001) or ATF4 in mammals (Harding et al., 2003).

In contrast, mTOR counteracts GCN2 and activates translation in response to sufficient nutrient availability (Sattlegger and Hinnebusch, 2000). Both kinases are also influenced by the activity of the mitochondria. The ADP/ATP ratio stimulates the phosphorylation of AMP-activated protein kinase (AMPK, AAK-2 in C. elegans), a signal that is further transduced to mTOR to communicate the energy availability in the cell that is needed for a higher protein synthesis (summarized in Bost et al., 2016). Furthermore, mTOR not only controls the cytosolic translational rate but interacts also with the mitochondrial transmembrane protein VDAC and influences the mitochondrial specific translation (Head et al., 2015). In addition, if an imbalance of cytosolic and mitochondrial translation activates the mitochondrial unfolded protein response, a ROS-mediated signal provokes the inhibition of cytosolic translation via GCN2 (Baker et al., 2012).

Here, we show that the health-and lifespan increasing effects achieved by a depletion of tdo-2 in C. elegans are independent of tryptophan uptake via transporter F23F1.6, the loading of free amino acids to tRNAs as well as cytosolic and mitochondrial translation controlled by AMPK, GCN2 and mTOR.

120

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 120 RESULTS AND DISCUSSION Depletion of tdo-2 leads to increased organismal tryptophan levels and the accompanying beneficial effects on healthspan could be phenocopied by supplying wild type animals with increasing amounts of tryptophan (Van der Goot et al., 2012). Tryptophan is the least abundant essential amino acid. Analyzing the coding regions of the C. elegans genome, we show that tryptophan is also the least encoded amino acid (Figure 1A) but almost 90% of all C. elegans’ proteins contain at least one tryptophan residue (Figure 1B). This pinpoints the importance for this amino acid in protein production and suggests that changes in tryptophan levels might influence translational rates. It was suggested that tryptophan is unique in the sense that it can alter protein synthesis (Young, 2013). Supplementation of tryptophan could 5 increase protein production e.g. in rodent liver (Sidransky et al., 1968) and rodent brain (Jorgensen and Majumdar, 1976; Majumdar and Nakhla, 1977), coincidently those tissues that also express TDO (Danesch et al., 1987; Kanai et al., 2009). Those studies were however limited to certain tissues, it is therefore not clear if there might be tissue-specific difference in the effect of tryptophan on translational activity.

The question was whether changes in translation might be responsible for the health- and lifespan increase that we see in tdo-2 depleted animals. We indeed observe that CRISPR/Cas9 tdo-2 deletion mutants are shorter and contain only about half the amount of proteins of wild type animals of the same age (Figure 1C), suggesting a direct or indirect effect on protein production in those animals. Surprisingly, we expected an increase in translational rates provoked by higher internal levels of free tryptophan but we observed a decreased protein content. However, when we knocked down tdo-2 by RNAi we do not observe those differences in size and protein content (Figure 1C) even though we still observe a strong effect on health- and lifespan, suggesting that the effect in the deletion mutants was indirect via development (Van der Goot et al., 2012).

It is possible that the depletion of tdo-2 affects only a fraction of the proteome. For that reason we assessed different components of the translational control and machinery and tested the two translational pathways (Figure 1D): cytosolic and mitochondrial translation that is mainly regulated by two kinases: GCN-2 and mTOR (let-363 in C. elegans).

As tryptophan is an essential amino acid, we tested whether the uptake of tryptophan into the cell is required for the effects of tdo-2 depletion by depleting the tryptophan transporter F23F1.6 (Figure 2A). As it was described before, depletion of F23F1.6 results in a lifespan increase (He et al., 2014). However, we still observe a further lifespan extension when also depleting tdo-2 (Figure 2B). We therefore concluded

121

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 121 A B

C length protein content/ D * worm GCN-2 1 1,5 *** mTOR mitochondrial translation translation (LET-363) VDAC 1 0,5 increase - 0,5 arbitrary units fold (normalized to WT) to (normalized 0 0 WT WT WT (del) (del) (del) RNAi RNAi -2 -2 -2 -2 ctr tdo tdo tdo tdo

Figure 1. The role of tryptophan in translation (A) Abundance of coding residues for specific amino acids in C. elegans genome. (B) Abundance of proteins that contain specific amino acids. (C) Length and protein content of wild type vs.tdo-2 depleted animals on day 4 of adulthood. (D) Key players of translational control. Error bars = SEM; statistics: * <0.05, *** <0.001.

that the possible changes in the uptake of tryptophan are not causative for the effects on lifespan in tdo-2 depleted animals.

Are the phenotypes then a result of an aberrant loading of tRNAs and an accompanying inhibition of the kinase GCN-2 (Figure 3A)? We knocked down all the known tRNA synthetases in C. elegans separately (Figure 3B and S1). We could not observe any correlation between the activity of tRNA synthetases that are responsible for tRNA loading and a certain swimming behavior. Most depletions had no effect on motility. However, some showed a reduced swimming ability whereas the knockdown of the tRNA synthetases for tryptophan (wars-1) and for cysteine (cars-1) had increased numbers of body bends (Figure 3B). This was in contrast to what we would have expected when tRNA loading would be involved in the healthspan benefits: If higher levels of free tryptophan would result in more loading of tryptophan tRNAs, inhibition of this process should block the phenotype, thus decrease motility. These results indicate that the increase in healthspan by tdo-2 depletion is independent of a possible change in tRNA loading. Furthermore, we tested a subset of tRNA synthetases for their influence on lifespan (Figure 3C). A depletion of tdo-2 resulted in all cases in a lifespan extension independent of a reduction in any tRNA synthetase. This suggests that the lifespan- extending phenotype of tdo-2 depletion is also independent of tRNA loading.

122

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 122 Finally, we asked whether the phenotypes caused by tdo-2 depletion would be a result of an inhibition of GCN-2 which would mean that a gcn-2(del) mutant should show comparable phenotypes as tdo-2 depleted animals. Gcn-2(del) mutants are long and skinnier than wild type animals suggesting that the genetic mutation has indeed an effect on the activity of the protein. Furthermore, we examined two independent mutants: Both mutants show comparable results suggesting that the observed effects are indeed a consequence of gcn-2 depletion. GCN-2 gets activated by amino acid deprivation and free tRNAs (Rousakis et al., 2013), thus we would rather expect an inhibition of GCN-2 because of more tryptophan and more loaded tryptophanyl tRNAs. Both gcn-2(del) mutants neither did show an increase in motility (Figure 3D) nor increased lifespan (Figure 3E) and a depletion of tdo-2 could still improve health- as well as lifespan in those animals. Therefore, we concluded that the health- and lifespan benefits in tdo-2 depleted animals are independent of translation regulated 5 by the cytosolic kinase GCN-2.

A B 100 WT/ ctr RNAi tdo-2 (del)/ ctr RNAi 80 WT/ F23F1.6 RNAi

tryptophan cytosol tryptophan 60 tdo-2 (del)/ F23F1.6 RNAi F23F1.6 40

TDO/TDO-2 percent survival 20 0 0 10 20 30 40 days of adulthood

Figure 2. TDO-2 regulates lifespan independently of tryptophan transport into the cell (A) Schematic representation of transporter (F23F1.6) –dependent tryptophan transport into the cell. (B) Survival curves of wild type and tdo-2 mutant animals treated with ctr and F23F1.6 RNAi.

Translation is also controlled by mTOR (let-363 in C. elegans). mTOR regulates cytosolic as well as mitochondrial translation, the latter being dependent on an interaction with the mitochondrial membrane protein VDAC (Figure 4A). A knockdown of let-363 did result in morphology changes (small, skinny animals) but not result in major changes in motility and an additional depletion in tdo-2 still increased the healthspan of those animals (Figure 4B). In line with these results, we found that the healthspan as well as the lifespan of vdac depleted animals could still be increased be an additional depletion of tdo-2 (Figure 4C). Furthermore, a reduction of the mitochondrial ribosomal subunit mrps-5 (Figure 4D) or the mitochondrial specific tRNA synthetasepus-1 (Figure 4E) had no influence on either the healthspan or the lifespan benefits by tdo-2 depletion. Experiments on mitochondrial acitivity (chapter VI) showed that the RNAi indeed inactivated the tested proteins and resulted in changes in mitochondrial respiration as it was described before (Houtkooper et al.,

123

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 123 A tryptophan

aa aa translation tRNA tRNA synthetase tRNA

GCN-2

B ns ns WT tdo-2 (del) ** *** *** *** *** *** * * * *** *** *** * ** *** *** * *** ** ** * ** * * *** -3 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -2 -1 -1 -1 -2 -1 ctr RNAi: ctr ctr aat iars fars lars tars xpo pus rars sars cars aars dars ears kars yars pars nars qars vars hars gars wars mars wars

100 C WT/ ctr RNAi 100 WT/ ctr RNAi tdo-2 (del)/ ctr RNAi 80 tdo-2 (del)/ ctr RNAi 80 WT/ mars-1 RNAi WT/ wars-1 RNAi tdo-2 (del)/ wars-1 RNAi 60 tdo-2 (del)/ mars-1 RNAi 60

40 40 percent survival percent survival 20 20

0 0 0 10 20 30 40 50 0 10 20 30 40 days of adulthood days of adulthood

100 100 WT/ ctr RNAi WT/ ctr RNAi tdo-2 (del)/ ctr RNAi 80 tdo-2 (del)/ ctr RNAi 80 WT/ cars-1 RNAi WT/ yars-1 RNAi 60 tdo-2 (del)/ cars-1 RNAi 60 tdo-2 (del)/ yars-1 RNAi

40 40 percent survival percent survival 20 20

0 0 0 10 20 30 40 0 10 20 30 40 50 days of adulthood days of adulthood

D E 100 WT/ ctr RNAi WT/ tdo-2 RNAi 80 gcn-2 (del)/ ctr RNAi ns 60 gcn-2 (del)/ tdo-2 RNAi ns 60 *** * *** 40 50 * percent survival 20 40 *** * * 30 0 0 10 20 30 40 50 20 days of adulthood

day 4 day 8 day 12

Figure 3. TDO-2 regulates health- and lifespan independently of GCN-2-dependent translation (A) Schematic model of GCN-2-dependent translation. (B) Motility/ healthspan of wild type and tdo-2 mutant animals depleted of different tRNA synthetases on day 4 of adulthood. (C) Survival curves of wild type and tdo-2 knockout animals depleted of certain tRNA synthetases. (D) Motility/ healthspan of gcn-2 mutant animals. (E) Survival curves of gcn-2 mutant animals. Error bars=SEM; statistics: * <0.05, ** <0.01, *** <0.001.

124

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 124 A GCN-2 doxycycline

mTOR PUS-1 mitochondrial translation translation (LET-363) VDAC MRPS-5

B C

WT WT tdo-2 (del) tdo-2 (del) 60 *** 60 *** *** 100 WT/ ctr RNAi *** 50 50 80 tdo-2 (del)/ ctr RNAi WT/ vdac RNAi 40 40 60 tdo-2 (del)/ vdac RNAi 30 30 40 20 20 5

percent survival 20 body bends/ 30 sec 30 bends/ body 10 sec 30 bends/ body 10 0 0 0 0 10 20 30

RNAi days of adulthood RNAi RNAi RNAi ctr ctr 363 - vdac let D E WT tdo-2 (del) WT tdo-2 (del) 70 *** 100 WT/ ctr RNAi 200 60 *** tdo-2 (del)/ ctr RNAi *** ** 80 50 WT/ mrps-5 RNAi 150 40 60 tdo-2 (del)/ mrps-5 RNAi 100 30 40 20 50 relative motility percent survival 20

body bends/ 30 sec 30 bends/ body 10 0 0 0 0 10 20 30 40 50 RNAi RNAi RNAi days of adulthood RNAi -5 -1 ctr ctr pus mrps

Figure 4. TDO-2 regulates health- and lifespan independently of mitochondrial translation (A) Schematic model of cytosolic and mitochondrial translation. (B) Motility/ healthspan of mTOR/let-363 depleted animals on day 4 of adulthood. (C) Motility/ healthspan (day 4) and survival curve of vdac-depleted animals. (D) Motility/ healthspan (day 4) and survival curves of mrps-5 depleted animals. (E) Relative motility/ health span (day 4) of pus-1 depleted worms. Error bars=SEM; statistics: ** <0.01, *** <0.001.

2013). Taken all results together suggest that the increase of lifespan and motility in tdo-2 depleted animals is independent of cytosolic as well as mitochondrial translation controlled by the kinase mTOR.

We finally tested whether the kinase AMPK aak-2( in C. elegans) is involved in the observed phenotypes. AMPK couples energy availability with translational activity in the cell (Figure 5A). Animals with reduced levels of tdo-2 showed a lowered

125

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 125 mitochondrial respiration and reduced ATP levels (chapter VI of this thesis) and might therefore influence the translational activity via AMPK. For our experiments, we used an AMPK mutant (aak-2(del)) that had been used and shown before to inactivate AMPK activity (Greer et al., 2007). A depletion of aak-2 could not inhibit the tdo-2 depletion-induced increases in health- or lifespan (Figure 5B and C). This indicates an alternative regulation independent of the kinase AMPK.

To summarize, possible alterations in translation are most likely not involved in the health- and lifespan regulation by TDO-2.

A LET-363 DAF-16 B maximum C mitochondrial capacity 100 WT/ ctr RNAi 40 WT/ tdo-2 RNAi AAK-2 80 aak-2 (del)/ ctr RNAi 30 60 aak-2 (del)/ tdo-2 RNAi ADP 20 ETC 40

ATP worm OCR/ 10 percent survival 20 0 0 RNAi RNAi 0 10 20 30 40 RNAi RNAi -2 -2 days of adulthood ctr ctr tdo tdo WT (del) -2 aak

Figure 5. TDO-2 regulates lifespan independently of AMPK (A) Schematic model of AMPK (AAK-2 in C.elegans) - mediated synchronization of energy availibility and translation. (B) Maximum mitochondrial oxygen consumption rate (OCR) of aak-2 mutant animals on day 4 of adulthood. (C) Survival curves of aak-2 knockout animals. Error bars=SEM; statistics: *** <0.001.

126

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 126 Experimental procedures Media and strains Animals were maintained at 20˚C on Nematode Growth Medium (NGM) and fed with E. coli OP50.

Animals were age synchronized by hypochlorite treatment.

The following strains were used in this study:

Wild type N2; tdo-2 (del) =OW716 tdo-2 (zg217)III; RB967 gcn-2 A (ok871)II; RB980 gcn-2 B (ok886)II; TG38 aak-2 (gt33)X; Protein content determination Animals were age-synchronized and transferred to FUDR-containing medium on 5 day 1 of adulthood (one day after L4 stage) to prevent offspring from growing. On day 4 of adulthood, exact numbers of worms (400, 800 and 1200) were transferred to plates without bacteria. They were subsequently washed off with M9 buffer and remaining animals on plate and pipet tip were subtracted from the total number of worms. The animals were sonicated and the total amount of isolated protein was determined by taking sample volume and protein concentration into account. The relative concentration of mutant animals compared to wild type animals was calculated for the three different numbers of starting material (400, 800 and 1200 animals) and the average is shown. RNA interference RNAi experiments were performed on NGM plates containing isopropylthio-β-D- galactoside (IPTG, 15mg/L) and 50μg/ml ampicillin. Plates were seeded with RNAi bacteria. Prior to the experiment, the plates were kept at room temperature for 2 days to allow the production of dsRNA by the bacteria. Plates were produced as freshly as possible (maximum 1 week before start of the experiment). Age-synchronized animals were either grown directly on RNAi food from larval stage L1 onwards or transferred to the RNAi treatment at larval stage L4 or day 1 of adulthood (indicated for specific experiments). For each experiment, the timepoint at which RNAi treatment started was determined depending on the results of a test run to see whether such treatment affected development. Animals of day 1 of adulthood (day after L4 stage) were transferred to RNAi plates containing FUDR to prevent the offspring from growing. RNAi clones were used from the Ahringer C. elegans RNAi library (Fraser et al., 2000) unless otherwise indicated (see below). All clones were verified by sequencing and animals were scored for associated phenotypes during the experiments. The let-363 RNAi clone was generated by PCR using forward primer 5’-cgatggacgaacagatatagc-3’and reverse primer 5’- gcacaaattcttggtagaagctgc-3’ and cloning into the L4440 vector, followed by transformation into HT115 bacteria.

127

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 127 Cloning was confirmed by sequencing. Motility/ healthspan Age-synchronized animals were grown on FUDR-containing NGM medium from day 1 of adulthood to prevent the growth of the offspring. Swimming assays were performed on 15 animals per experimental group on different days of adulthood. A worm was placed in a drop of M9. The animal was allowed to acclimate for 30 seconds before its body bends were counted for another 30 seconds. Experiments were repeated and one representative graph is shown. Statistical analysis was done with a one-way ANOVA with post-hoc Bonferroni algorithm. Lifespan experiments 100 animals per experimental condition were scored for their lifespan behavior. Worms were age-synchronized and transferred to FUDR-containing NGM medium on day 1 of adulthood to prevent offspring from growing. On every time point, living animals were counted and dead animals (no nose touch response anymore) removed. Animals that disappeared during the assay were excluded from the analysis. Experiments were performed at 20˚C and were repeated. One representative curve is shown. Curves were generated by Graphpad Prism software. Oxygen consumption rates Oxygen consumption rates (OCR) were measured using the Seahorse XF96 cycler on day 4 of adulthood unless otherwise indicated. OCR was measured as pMoles/ min/worm. Respiration behaviour was optimized for wild type and tdo-2 deletion mutants at room temperature (the detailed protocol can be found in chapter VII of this thesis). The following cycles were used (1 loop = mixing: 2 minutes, waiting: 30 sec, measuring: 2 minutes): (1) calibration (2) equilibration (3) basal respiration = measurement for 5 loops (4) maximum mitochondrial capacity = first injection of FCCP, then measurement for 9 loops (5) non-mitochondrial respiration = first injection of sodium azide, then measurement for 4 loops. Mitochondrial OCR was determined by subtracting non-mitochondrial OCR from basal respiration levels. Maximum mitochondrial capacity was calculated by subtracting non-mitochondrial OCR from maximum FCCP-induced respiration. The following concentrations were used for injection: 20μl of 100μM FCCP and 20μl of 400mM sodium azide. Each experimental group was tested in 6-8 wells, each containing 10-30 worms/well (counted afterwards). Only conditions of the same run can be compared. Experiments were repeated and one representative graph is shown with error bars = SEM.

128

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 128 REFERENCES

• Anderson, G et al.. Increased IL-6 trans-signaling in • Houtkooper, RH, Williams, RW, Auwerx, J: Metabolic depression: focus on the tryptophan catabolite pathway, networks of longevity. Cell 142, 9-14 (2010). melatonin and neuroprogression. Pharmacological • Houtkooper, RH et al.: Mitonuclear protein imbalance Reports 65, 1647-1654 (2013). as a conserved longevity mechanism. Nature • Baker, BM, Nargund, AM, Sun, T, Haynes, CM: 497(7450), 451-7 (2013). Protective coupling of mitochondrial function and • Jorgensen, AJF, Majumdar, APN: Bilateral protein synthesis via the eIF2alpha kinase GCN-2. adrenalectomy: Effect of tryptophan force-feeding PLoS Genet 8(6), (2012). on amino acid incorporation into ferretin, transferrin • Bost, F, Decoux-Poullot, AG, Tanti, JF, Clavel, S: and mixed proteins of liver, brain and kidneys in vivo. Energy disruptors: rising stars in anticancer therapy?. Biochem Med 16, 37-46 (1976). Oncogenesis 5, (2016). • Kanai, M et al.: Tryptophan 2,3-dioxygenase is a key • Breda, C et al.: Tryptophan-2,3-dioxygenase modulator of physiological neurogenesis and anxiety- (TDO) inhibition ameliorates neurodegeneration by related behavior in mice. Mol Brain 2, 8 (2009). modulation of kynurenine pathway metabolites. Proc • Lee, SJ, Hwang, AB, Kenyon, C: Inhibition of Natl Acad Sci U S A 113(19), 5435-40 (2016). 5 respiration extends C. elegans life span via reactive • Campesan, S et al.: The kynurenine pathway oxygen species that increase HIF-1 activity. Curr Biol modulates neurodegeneration in a Drosophila model 20, 2131-2136 (2010). of Huntington’s disease. Curr Biol 21, 961-966 (2011). • Majumdar, PN, Nakhla, AM: Influence of tryptophan • Danesch, U et al.: Glucocorticoid induction of the rat on the activity of acetylcholinesterase in the brain of tryptophan oxygenase gene is mediated by 2 widely well-fed normal and adrenalectomized rats. Biochem separated glucocorticoid-responsive elements. EMBO Biophys Res Commun 76, 71-7 (1977). J 6, 625-630 (1987). • Natarajan, K et al.: Transcriptional profiling shows • Fougeray, S et al.: Tryptophan depletion and the kinase that Gcn4p is a master regulator of gene expression GCN2 mediate IFN-gamma-induced autophagy. J during amino acid starvation in yeast. Mol Cell Biol Immunol 189, 2954-2964 (2012). 21, 4347-4368 (2001). • Fraser, AG et al.: Functional genomic analysis • Pang, YLJ, Poruri, K, Martinis, SA: tRNA synthetase: of C. elegans chromosome I by systematic RNA tRNA aminoacylation and beyond. Wiley Interdiscip interference. Nature 408, 325-30 (2000). Rev RNA 5(4), 461-480 (2014). • Giorgini, F, Guidetti, P, Nguyen, Q, Bennett, SC, • Rousakis, A et al.: The general control Muchowski, PJ: A genomic screen in yeast implicates nonderepressible-2 kinase mediates stress response kynurenine 3-monooxygenase as a therapeutic target and longevity induced by target of rapamysin for Huntington disease. Nat Genet 37, 526-531 (2005). inactivation in Caenorhabditis elegans. Aging Cell • Greer, EL et al.: An AMPK-FOXO pathway mediates 12(5), 742-51 (2013). longevity induced by a novel method of dietary • Sattlegger, E and Hinnebusch, AG: Separate domains restriction in C. elegans. Curr Biol 17(19), 1646-56 in GCN1 for binding protein kinase GCN2 and (2007). ribosomes are required for GCN2 activation in amino • Hamilton, B et al.: A systematic RNAi screen for acid-starved cells. EMBO J 19(23), 6622-33 (2000). longevity genes in C. elegans. Genes Dev. 19(13), • Sidransky, H et al.: Effect of dietary tryptophan on 1544-55 (2005). hepatic polyribosomes and protein synthesis in fasted • Harding, HP et al.: An integrated stress response mice. J Biol Chem 243, 1123-32 (1968). regulates amino acid metabolism and resistance to • Van der Goot, AT et al.: Delaying aging and the aging- oxidative stress. Mol Cell 11, 619-633 (2003). associated decline in protein homeostasis by inhibition • He, C et al.: Enhanced longevity by ibuprofen, of tryptophan degradation. Proc Natl Acad Sci U S A conserved in multiple species, occurs in yeast through 109(37), 14912-7 (2012). inhibition of tryptophan import. PLoS Genet 10(12), • Young, SN: Acute tryptophan depletion in humans: a (2014). review of theoretical, practical and ethical aspects. J • Head, SA et al.: Antifungal drug itraconazole targets Psychiatry Neurosci 38(5), 294-305 (2013). VDAC1 to modulate the AMPK/mTOR signaling axis • Zwilling, D et al.: Kynurenine 3-monooxygenase in endothelial cells. Proc Natl Acad Sci U S A 112(52), inhibition in blood ameliorates neurodegeneration. (2015). Cell 145, 864-874 (2011).

129

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 129 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 130 Chapter VI

Transcriptome profiling in C. elegans reveals that TDO and other amino acid dioxygenases regulate mitochondrial capacity and healthspan Short title: Healthspan control by amino acid dioxygenases

Helen Michels1, Mandy Koopman1, Renée I. Seinstra1, Martijn van Faassen3, Tristan de Jong2, Tobias Ackermann4, Dennis H. Lentferink1, Laura Noth- durft1, Céline N. Martineau1, Cornelis F. Calkhoven4, Victor Guryev2, Ido P. Kema3, Ellen A.A. Nollen1

1European Research Institute for the Biology of Aging, University of Groningen, University Medical Centre Groningen, Laboratory of Molecular Neurobiology of Aging, Antonius Deusinglaan 1, 9700 AD Groningen, The Netherlands 2European Research Institute for the Biology of Aging, University of Groningen, University Medical Centre Groningen, Laboratory of Genome Structure of Aging, Antonius Deusinglaan 1, 9700 AD Groningen, The Netherlands 3University of Groningen, University Medical Centre Groningen, Department of Laboratory Medicine, Hanzeplein 1, 9713 GZ Groningen, The Netherlands 4European Research Institute for the Biology of Aging, University of Groningen, University Medical Centre Groningen, Laboratory of Gene Regulation in Aging and Age-related Diseases, Antonius Deusinglaan 1, 9700 AD Groningen, The Netherlands

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 131 In an era of increasing life expectancy, improving the quality of life of the elderly population requires understanding of the biological mechanisms of aging and age- related diseases.

The kynurenine pathways of tryptophan degradation is involved in aging and age-related pathologies. In model organisms, depletion of the tryptophan- 2,3 dioxygenase (TDO)- the first enzyme of the kynurenine pathway- increases lifespan, and protects against neurodegenerative proteotoxicity and tumors (1, 2, 3, 4, 5, 6). Alterations in the relative abundance of kynurenine metabolites play a role but have not explained all protective effects (2, 3, 4, 7, 8, 9). Using transcriptome profiling in C. elegans we identified kynurenine-independent responses to tdo-2 depletion. We find that tdo-2 regulates mitochondrial respiration and that this function is shared by other amino acid dioxygenases, hpd-1 and cdo-1. Depletion of hpd-1 and cdo-1 also resulted in lifespan extension and suppression of age- and Parkinson’s disease- related motility decline previously observed for tdo-2 depletion. Depletion of other enzymes that just consume oxygen or just degrade amino acids do not show these phenotypes, suggesting that the integrated use of oxygen and amino acids by amino acid dioxygenases controls mitochondrial function and determines organismal health and lifespan. Because the amino acid dioxygenases are evolutionary conserved in human, our results provide the rationale to explore their inhibition as a therapeutic strategy to slow aging and ameliorate age-related pathologies.

Significance statement Tryptophan metabolism is involved in various age-related pathologies such as neurodegeneration, cancer and diabetes type 2, which attracted the attention for therapeutic interventions. In model organisms for cancer and neurodegeneration, inhibition of the tryptophan-degrading enzyme, tryptophan- 2,3 dioxygenase (TDO), improves health. Tryptophan metabolites play a role but do not explain all protective effects. Here, we identified a tryptophan metabolite-independent function of TDO: the regulation of mitochondrial energy metabolism. Other amino acid dioxygenases regulated mitochondrial function, health- and lifespan in a similar manner. Our results suggest that the integrated use of oxygen and amino acids determines healthy lifespan and that inhibition of amino acid dioxygenases can be explored as a therapeutic strategy to slow aging and ameliorate age-related pathologies.

132

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 132 Tryptophan 2,3-dioxygenase (TDO-2 in the nematode C. elegans; TDO in human) is a haem-containing, oxygen-dependent, cytosolic enzyme that degrades tryptophan into formylkynurenine as a first step in the kynurenine pathway (10). TDO regulates systemic levels of tryptophan, an essential amino acid that regulates translation and is a precursor of other biogenic amines, including serotonin and tryptamine (reviewed in 5). An increase in tryptophan degradation via the kynurenine pathway is thought to be involved in various age-related diseases. For example in certain cancers, an increase in TDO activity is believed to cause tumoral immune tolerance (3, 7); in type 2 diabetes, patients with high plasma kynurenine levels are more likely to have severe insulin resistance (11); and patients with neurodegenerative diseases have been found to have an imbalance in kynurenine metabolites (12, 13, 14). Inhibition of TDO in model organisms increased their lifespan (1,4) and protected against neurodegenerative proteotoxicity and cancer (2, 3, 4, 6). Whereas the relative abundance of metabolites of the kynurenine pathway has been demonstrated to play a role for the toxicity especially in flies (2, 6), this could not explain all proteoprotective 6 effects in other organisms such as e.g. Caenorhabditis elegans (4), suggesting that there might be kynurenine pathway independent mechanisms as well.

TDO-2 regulates mitochondrial respiration and ATP levels The regulation of proteotoxicity and lifespan by TDO-2 in C. elegans appears independent of the kynurenine pathway, and also independent of serotonin and tryptamine (4 and Supplemental Information Fig. S1). We generated CRISPR/Cas-9 tdo-2 deletion mutants, which showed an increase in motility and lifespan extension similar to that in animals in which tdo-2 was depleted by RNAi (Supplemental Information Fig. S1C, S2). tdo-2(del) mutants also had an extended lifespan when combined with knockdown of the regulator of translation gcn-2 or knockdown of tRNA synthetases, indicating that lifespan extension was also independent of the role of tryptophan in protein translation (Supplemental Information Fig. S3).

To uncover the cellular responses mediating the health and lifespan benefits of tdo-2 depletion that are independent of the kynurenine and serotonin pathway, we performed transcriptome profiling. In order to identify only those changes that are related to tdo-2, we combined tdo-2 depletion with mutants of the serotonin (tph-1(del)) and kynurenine pathway (kmo-1(del)) (Fig. 1A, transcriptome profiles are available online: EBI ArrayExpress, accession number E-MTAB-4695). In tdo-2-depleted animals, we saw no alterations in the expression of genes of stress pathways known to play a role in lifespan extension (example genes in Supplemental Information Fig. S4B), i.e. genes related to the heat shock response, the unfolded protein responses of the endoplasmic reticulum and mitochondria, and the oxidative stress response (15, 16).

133

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 133 Mitochondrial oxygen consumption rates (OCR). Mitochondrial (D) Heatmap of differentially expressed genes on day 4 of adulthood. Blue = Heatmap of differentially (B) lysates on day 4 of adulthood. C. elegans ATP levels in levels ATP (C) ) Schematic representation of tryptophan-metabolising pathways; sample choice for transcriptome analysis. pathways; sample choice for transcriptome of tryptophan-metabolising ) Schematic representation Figure 1. TDO-2 regulates mitochondrial function and ATP levels ATP mitochondrial function and TDO-2 regulates 1. Figure (A expression. expression, yellow = higher than average lower than average = larval stage, D day of adulthood. = wild type, KD knockdown, ctr control, L WT Statistics for all panels: ** <0.01, *** <0.001. Error bars = SEM.

134

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 134 Among the most prominent changes in tdo-2-depleted animals was an increase in transcripts encoding three of the five major enzymes involved in mitochondrial beta- oxidation (Fig. 1B and Fig. S4A). Such an increase either occurs when cells shift to beta-oxidation for energy supply – resulting in an increase in oxygen consumption – or is a response to other mitochondrial alterations that would reduce respiration. When we measured mitochondrial oxygen consumption rates (OCR) and ATP levels, we found that, up to mid-adulthood, tdo-2-depleted animals had far lower levels of ATP (Fig. 1C) and mitochondrial respiration (Fig. 1D) than those seen in WT animals. We also found that, starting from day one of adulthood, RNAi depletion of tdo-2 reduced mitochondrial respiration (see also Fig. 3F), indicating that this reduction in mitochondrial respiration was independent of developmental effects of tdo-2 depletion. These findings suggest that TDO-2 regulates mitochondrial function.

TDO-2 regulates mitochondrial capacity 6 We detected no significant changes in the RNA expression of mitochondrial genes that are independent of kmo-1(del) or tph-1(del) that might indicate that a simple reduction in the number of the mitochondria is not likely to be responsible for the decrease in respiration (genes for oxidative phosphorylation in Supplemental Information Fig. S4C). Alternatively, the cause of reduced respiration may be impairment in oxygen-dependent biochemical pathways involved in ATP synthesis within the mitochondria. This can be studied using the ionophore FCCP (carbonyl cyanide p-trifluoromethoxyphenylhydrazone) that induces proton leakage in the inner mitochondrial membrane and therefore disrupts ATP synthesis via the electron transport chain (17). An organism typically reacts to such proton leakage by boosting mitochondrial respiratory pathways in an attempt to regenerate the proton gradient and restore ATP levels, which results in a strong increase in mitochondrial oxygen consumption (Fig. 2A). Treatment of wild type animals with FCCP resulted in the anticipated increase in respiration of more than two-fold, whereas in tdo-2 mutated animals the increase was minimal (Fig. 2B). These data suggest that tdo-2 deficiency affects the respiratory capacity of the mitochondria.

TDO-2 regulates mitochondrial capacity independently of major tryptophan signaling pathways, translation control and ROS production Depletion of tdo-2 in combination with a mutation in one of the genes involved in the serotonin and the kynurenine pathways, namely tph-1(del), afmd-1(del), kmo- 1(del) or ahr-1(del), still reduced mitochondrial respiration to a degree similar to that seen for tdo-2 depletion in wild type animals (Fig. 2C,D). The fact that the absence of genes involved in the serotonin and the kynurenine pathways could not restore

135

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 135 136

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 136 Reaction to proton leakage induced by FCCP on day 4 of induced by FCCP to proton leakage Reaction (B) 6 Schematic representation of the mitochondrial beta-oxidation and maximum mitochondrial respiration on respiration mitochondrial and maximum beta-oxidation mitochondrial of the representation Schematic (G) Maximum mitochondrial respiration on day 4 of adulthood. Statistics for all panels: ns = not significant, * <0.05, ** <0.01, *** <0.001. Error bars SEM. (H) Maximum mitochondrial respiration on day 4 of adulthood. respiration mitochondrial Maximum (C-F) ) Schematic overview of the different types of oxygen consumption rates (OCR) measured by the seahorse XF96 cycler. types of oxygen consumption rates (OCR) measured by the seahorse XF96 cycler. overview of the different ) Schematic Figure 2. TDO-2 regulates mitochondrial capacity and via a subunit of ETC complex IV TDO-2 regulates 2. Figure (A adulthood. day 4 of adulthood.

137

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 137 respiration to normal levels indicates again that the mitochondrial phenotype was independent of these pathways.

Depletion of amino acid responding cytosolic or mitochondrial regulators of protein synthesis – including the inhibitor of general translation gcn-2 and the mTOR homologue let-363 – as well as knockdown of the mitochondrial ribosomal protein mrps-5 (18, 19) resulted in additive reduction of respiration when combined with tdo- 2 knockdown (Fig. 2E,F and Supplemental Information Fig. S3D-H). Collectively, these results show that the mitochondrial phenotype induced by tdo-2 depletion is independent of tryptophan-dependent pathways and translation and is also additive to other pathways affecting respiration.

Mitochondrial-mediated, mitohormetic lifespan control has been associated with the altered production of reactive oxygen species (ROS) (20, 21, 22). Using a

sensor strain for the ROS H2O2 (36), we found that tdo-2 depleted animals did not show an increase in the amount of that specific ROS (Supplemental information S5D). Furthermore, neither deletion of genes required for ROS-mediated lifespan extension – including the superoxide dismutase genes sod-1, sod-2, sod-3 and ced- 13 (23, 24) – nor the addition of the anti-oxidant NAC (N-acetyl cysteine) could abolish the effects of tdo-2 depletion on the mitochondria or on lifespan. Moreover,

tdo-2 depleted animals do not show an increased sensitivity when treated with H2O2 previously shown to be an indicator of increased oxidative stress (42). These findings indicate that the tdo-2-dependent decrease in mitochondrial respiration and lifespan extension are not mediated by ROS (Supplemental Information Fig. S5).

TDO-2 might regulate mitochondrial capacity via a subunit of ETC complex IV Since the reduction in mitochondrial respiration due to tdo-2 depletion did not appear to involve either tryptophan-dependent pathways or ROS, we next searched for other parts of the respiratory machinery where tdo-2 depletion might intervene (Overview mitochondria in Supplemental Information Fig. S5A). Depletion of genes of the beta-oxidation pathway, including acs-2 and kat-1, had no effect on the respiration of tdo-2 mutant animals (Fig. 2G). This observation supports the idea that the upregulation of beta-oxidation genes in tdo-2-depleted animals is a consequence of reduced ATP production rather than of a defect in glycolysis, and suggests that tdo-2 might intervene in a process downstream of glycolysis and beta-oxidation. Therefore, we tested the involvement of components of the electron transport chain (ETC). Knockdown of subunits of the four complexes of the ETC reduced mitochondrial respiration, however an additional knockdown of tdo-2 further reduced the mitochondrial oxygen consumption rates only for three subunits

138

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 138 but not for subunit F26E4.6 (Cox7C in human) (Fig. 2H). F26E4.6 is a subunit of cytochrome c oxidase, the oxygen-dependent complex IV of the ETC, depletion of which showed lifespan-increasing effects in multiple longevity screens in C. elegans (25, 26, 27). Our results suggest that this subunit of complex IV might regulate the process by which tdo-2 depletion influences mitochondrial respiration.

TDO-2 shares functions in respiration control and health- and lifespan regulation with other amino acid-degrading dioxygenases Having excluded an involvement of the kynurenine pathway, biogenic amines, and protein synthesis, we then wanted to know which other aspects of TDO-2’s enzymatic activity might be responsible for the mitochondrial phenotype observed. TDO-2 combines amino acid catabolism with dioxygenase activity. This combination of activities is shared with other enzymes, including HPD-1 (4-hydroxyphenyl pyruvate dioxygenase, HPPD in human), the second enzyme in tyrosine metabolism (Fig. 3A), 6 and CDO-1 (cysteine dioxygenase, CDO in human), which degrades cysteine (Fig. 3B). When we depleted hpd-1 and cdo-1 we also found reduced oxygen consumption rates, similar to that observed for the tdo-2(del) mutants (Fig. 3C), suggesting that these three enzymes share a function in mitochondrial regulation.

The question remaining was what was responsible for the regulation of the three phenotypes: the process of oxygen usage or amino acid degradation in itself, or the coupling of oxygen consumption and amino acid availability? We therefore tested a range of other oxygen-consuming enzymes that function independently of amino acid degradation. We found 33 annotated oxygen-consuming genes in C. elegans from which we choose 5 representative candidates covering most cellular functions where oxygen consumption is involved in addition to the three amino acid dioxygenases. Of the tested proteins, only depletion of the amino acid dioxygenases reduced mitochondrial-specific respiration (Fig. 3D) and all are having similar effects on the respiration of F26E4.6 depleted animals (subunit of ETC complex IV) (Fig. 3E), supporting the idea that these three enzymes share the same regulation mechanism to control mitochondrial respiration.

If all three amino acid dioxygenases share a common mechanism in regulating mitochondrial activity, do they also share the ability to regulate health- and lifespan? Indeed, we observed that depletion of hpd-1 (Fig. 3I) or cdo-1 (Fig. 3J) increased lifespan in C. elegans and improved motility during aging (Fig. 3G). This effect was also seen when the genes were knocked down at day one of adulthood, after development is complete (Fig 3F; Supplemental Information Fig. S6C,D). The fact that other enzymes that either are only oxygen-dependent (Fig. 3G) or only degrade amino acids (Fig. 3H) did not share this role in healthspan regulation points to a

139

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 139

RNAi -1 cdo

RNAi -1 ** hpd RNAi

RNAi

RNAi -2 ** tdo -1 -2

RNAi

** RNAi ctr tdo cdo ctr 40 WT/ WT/ WT/ WT/ WT/

maximum mitochondrial respiration RNAi from day 1 day from RNAi 40 30 20 10 OCR/ worm OCR/

(del) 1 - hpd RNAi F26E4.6

(del) 2 - tdo RNAi F26E4.6 F26E4.6 20

RNAi F26E4.6 1+ - cdo

RNAi ctr+ F26E4.6 F26E4.6 ctr+ 10 days of adulthood of days ns ns ns

WT RNAi F26E4.6

RNAi ctr (del) 1 - hpd 0

80 60 40 20 RNAi ctr (del) 2 - tdo

100 survival percent

RNAi -1 cdo ctr+

J RNAi -2 tdo

RNAi -1 cdo

WT *** RNAi

RNAi ctr -2 RNAi ctr tdo RNAi 0

-2

50 40 30 20 10 RNAi OCR/ worm OCR/ (del)/ (del)/

40 tdo ctr

-1 -1 E

RNAi 268 - let WT/ hpd hpd WT/ WT/

30 RNAi -2 gbh

ns RNAi -2 pdi

20 RNAi -1 nmad

RNAi K07E1.1

ns days of adulthood of days

10 RNAi -1 hpd

RNAi -1 cdo

*** RNAi -2 tdo

80 60 40 20

*** survival percent

100 RNAi ctr I

0 (del) -2

20 10 cat 30 OCR/ worm OCR/

ns (del) -1 basl D

(del) -2 hdl RNAi -1 cdo

RNAi -1

hpd *** (del) -1 hdl

RNAi -2 tdo

WT WT *** RNAi ctr 0

20 80 60 40

0 sec 30 bends/ body

40 30 20 10 OCR/ worm OCR/ H

RNAi 268 - let

ns RNAi -2 gbh

-1

ns RNAi -2 pdi -1

ns RNAi -1 nmad

/ HPD /

ns RNAi K07E1.1 / CDO

ns RNAi -1 HPPD hpd

** acetoacetate CDO RNAi -1 cdo

RNAi -2 tdo

/ C. elegans / C. elegans

*** RNAi tyrosine ctr cysteine homogentisate fumarylacetoacetate human human - cysteine sulfinic acid

fumarate 0

4 60 30 70 50 40 20 10 body bends/ 30 sec 30 bends/ body B G A

140

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 140 Maximum (D) Motility/healthspan Schematic overview of (B) (G, H) Motility/healthspan of a Parkinson’s Parkinson’s of a Motility/healthspan C. elegans. (K) -depleted animals. -depleted

cdo-1 6 and hpd-1 Maximum mitochondrial capacity on day 4 of adulthood upon depletion of amino acid of amino upon depletion on day 4 of adulthood capacity mitochondrial Maximum (E) Survival curves of curves Survival Maximum mitochondrial oxygen consumption rate (OCR) on day 4 of adulthood. oxygen consumption rate (OCR) on day 4 of adulthood. Maximum mitochondrial

(I-J)

* RNAi -1

hpd (C)

*** RNAi -1 cdo

** RNAi -2 tdo

YFP RNAi ctr

C. elegans.

***

RNAi -1

hpd day 8

** RNAi -1 cdo

** of amino acid dioxygenases after development. depletion capacity; Maximum mitochondrial RNAi -2 tdo

on day 4 of adulthood. of adulthood. 4 on day RNAi ctr (F) -Syn::YFP

Summary. Statistics for all panels: ns = not significant, * <0.05, ** <0.01, *** <0.001. Error bars SEM. Summary.

(H) α

RNAi -1 hpd (L) ns

RNAi -1 cdo ns

RNAi -2 tdo ns

YFP RNAi ctr

RNAi -1 hpd

day 4

* RNAi -1 cdo ns

RNAi -2 tdo ns

RNAi ctr -Syn::YFP α 0

60 50 40 30 20 10 body bends/ 30 sec 30 bends/ body enzymes acid-degrading amino and (G) K ) Schematic representation of the pathway tyrosine degradation including HPPD/HPD-1representation ) Schematic (4-hydroxyphenyl pyruvate dioxygenase) in human and disease model. RNAi treatment started after development. Figure 3. TDO-2 shares functions in respiration control and health- and lifespan regulation with other amino acid-degrading dioxygenases with other and health- lifespan regulation control functions in respiration TDO-2 shares 3. Figure (A by CDO/CDO-1 cysteine degradation (cysteine dioxygenase) in human and enzymes. of oxygen-consuming upon depletion on day 4 of adulthood capacity mitochondrial subunit F26E4.6. with complex IV dioxygenases in combination of oxygen-consuming

141

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 141 specific function for amino acid dioxygenases.

We finally tested whether the depletion of the two other dioxygenases (CDO-1 and HPD-1) also protects against proteotoxicity in a C. elegans model of Parkinson’s disease, as it was described for tdo-2 before (4). Indeed, depletion of all three amino acid dioxygenases suppressed the accelerated decline in motility during aging in alpha-synuclein expressing animals (Fig. 3K). The general health- and lifespan extension and the suppression of age-related motility decline could be still observed when depletion of the amino acid dioxygenases starts at day 1 after development (Fig. 3F and K, Supplemental information Fig. S6C-D).

All results taken together indicate that amino acid dioxygenases integrate amino acid availability and oxygen consumption to control mitochondrial respiration and regulate health- and lifespan in C. elegans (Fig. 3L).

In humans, the enzymes TDO, its functional homologue IDO, and HPPD are known to be associated with several diseases and therefore they attracted thus the attention as targets for therapeutic interventions (summarized in 28, 29) . An inhibitor for HPPD (which was initially designed as a fungicide in plant agriculture) is used to reduce toxic downstream products in hepatorenal tyrosinemia, a severe tyrosine- related metabolic disease that starts early in childhood (30). In mice, inhibiting TDO in IDO-positive tumours suppresses tumour growth (3), and inhibitors for TDO and IDO are in clinical use to treat tumours (31, 32, 33, 34). The protective effects of e.g. TDO inhibition have been so far attributed to tryptophan increase and kynurenine depletion (3 and summarized in 31). Since in many tumours and other inflammatory diseases mitochondrial activity is increased – thereby securing the high energy demand required for cancer growth and metastasis (35) – global suppression of mitochondrial respiration may contribute to the protective effects of TDO inhibition.

The remarkable functional resemblance of HPPD and CDO to TDO and the profound effect of their inhibition on mitochondrial function, healthspan and lifespan make these amino acid dioxygenases powerful candidate targets for pharmacological inhibition to increase healthspan and improve quality of life in the elderly.

142

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 142 METHODS

Media and strains Nematodes were grown at standard conditions at 20˚C. Animals were age-synchronized by hypochlorite treatment, hatched overnight in M9 buffer and subsequently cultured on Nematode Growth Medium (NGM) seeded with Escherichia coli strain OP50. On day 1 of adulthood (one day after larval stage L4), animals were transferred to NGM plates containing 5-fluoro-2’deoxy-uridine (FUDR) to prevent the offspring from growing. The following strains were used in this study:

N2; ZG24 ahr-1 A (ia3)I; CZ2485 ahr-1 B (ju145)I; RB967 gcn-2 (ok871)II; OW479 haao-1 (tm4627)V; OW478 kmo-1 (tm4529)V; OW477 amfd-1 (tm4547)IV; CB1003 flu-2 (e1003)X; TDO (del B) =OW715 tdo-2 (zg216)III; TDO (del) =OW716 tdo-2 6 (zg217)III; RB1899 acs-2 (ok2457)V; VS24 kat-1 (tm1037)II; GR1321 tph-1 (mg280) II; GA187 sod-1 (tm776)II; GA184 sod-2 (gk257)I; GA186 sod-3 (tm760)X; GA480 sod-2 (gk257)I/sod-3 (tm760)X; MD792 ced-13 (sv32)X; jrIs1[Prpl-17::HyPer] (36); VC1539 hpd-1 (ok1955)III; VC217 hdl-2(ok440)IV; CB112 cat-2(e1112)II; RB865 basl-1(ok703)III; RB1029 hdl-1(ok956)IV; OW40 zgIs15[P(unc-54)::αsyn::YFP] IV; OW450 rmIs126[P(unc-54)Q0::YFP]V

RNA interference RNAi experiments were performed on NGM plates containing isopropylthio-β-D- galactoside (IPTG, 15mg/L) and 50μg/ml ampicillin. Plates were seeded with RNAi bacteria. Prior to the experiment, the plates were kept at room temperature for 2 days to allow the production of dsRNA by the bacteria. Plates were produced as freshly as possible (maximum 1 week before start of the experiment). Age-synchronized animals were either grown directly on RNAi food from larval stage L1 onwards or transferred to the RNAi treatment at larval stage L4 or day 1 of adulthood (indicated for specific experiments). For each experiment, the timepoint at which RNAi treatment started was determined depending on the results of a test run to see whether such treatment affected development. Animals of day 1 of adulthood (day after L4 stage) were transferred to RNAi plates containing FUDR to prevent the offspring from growing. RNAi clones were used from the Ahringer C. elegans RNAi library (37) unless otherwise indicated (see below). All clones were verified by sequencing and animals were scored for associated phenotypes during the experiments.The let-363 RNAi clone was generated by PCR using forward primer 5’-cgatggacgaacagatatagc-3’and reverse primer 5’- gcacaaattcttggtagaagctgc-3’ and cloning into the L4440 vector, followed by transformation into HT115 bacteria.

143

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 143 Cloning was confirmed by sequencing.The cdo-1 RNAi clone was generated by a nested PCR. For the first PCR forward primer 5’-gcctaaaaaaccgtgttctgc-3’ and reverse primer 5’-ggatgtgatccgacttctacagg-3’ were used. The follow-up PCR used forward primer 5’- gttcaaattcgtgaaatcttcc-3’ and reverse primer 5’- ctgtagtcaactttcttgccg-3’. The PCR product was cloned into the L4440 vector and transformed into HT115 bacteria. Cloning was confirmed by sequencing.

Supplementation N-acetyl cysteine (NAC), an anti-oxidant, was dissolved in water and added to the warm medium after autoclaving to a final concentration of 10mM or 20mM. Doxycycline, an inhibitor of mitochondrial translation (19), was dissolved in water and added to the warm medium before pouring the plates at final concentrations of 10μM or 20μM. Plates were prepared 2 days in advance and kept in the dark before starting the experiment to minimize degradation of compounds. Animals were supplemented with doxycycline from larval stage 4 onwards and supplementation with NAC started on day 1 of adulthood.

Preparation of C. elegans lysates for metabolite measurement About 2000 animals/experimental group were collected for lysate preparation in PBS and extensively washed for several hours to remove as much bacterial material as possible from the digestion system and sample. Only samples from the same run were compared later. cOmpleteTM protease inhibitor (Roche) was finally added to the samples. The samples were snap frozen in liquid nitrogen and stored at -80˚C until use. From this point on, the samples were constantly kept on ice. Worm samples were sonicated, excessive worm debris was removed and the concentration was determined. Lysates were equalized in concentration and tryptophan and kynurenine levels were measured by LC-MS/MS as described before (4).

Motility assay Animals were scored at day 4 of adulthood unless otherwise indicated. Animals were placed in a drop of M9 and allowed to adjust for 30 sec, after which the number of body bends was counted for another 30 sec. 15 animals were scored per experimental group. Treatments respectively mutations were blinded before counting body bends. Assays were repeated in three independent experiments and the data of one representative experiment is shown.

144

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 144 RNA sequencing Animals for RNA isolation were collected in M9 on day 4 of adulthood. Only samples from the same run were compared later. Animals were washed a few times and then snap frozen in liquid nitrogen. Total RNA was extracted using TRIzol Reagent (Life Technologies) according to the manufacturer’s description. The TruSeq Sample Preparation V2 Kit (Illumina) was used for polyA RNA sequencing. cDNA libraries were subjected to high-throughput single-end sequencing (50 bp) in an Illumina HiSeq 2500 instrument. The RNA sequencing dataset has been submitted to EBI ArrayExpress, under accession number E-MTAB-4695. RNA sequencing data was mapped to the WBCel235 genome reference using the TopHat 2.0.9 program (38) and gene annotation from Ensembl release 73. Per-gene expression data was normalized as fragments per million mapped (FPM). Differentially expressed genes were predicted using the CuffDiff program version 2.1.1. Data visualization and statistical tests were conducted using R scripts. The heatmap was generated with 6 Cluster 3.0 and visualized with Tree View.

Generation of TDO-2 deletion mutants by CRISPR/ Cas9-induced mutagenesis The choice of target sequence was performed as described in 39. One of the target sequences selected for cloning was in exon 3 (GCTCGACACAATGAGTCCAT). To reduce the risk of off-target mutations, the target sequence was checked to ensure it was unique to the C. elegans genome (blast search on wormbase. org). The target sequence was cloned into the C. elegans CRISPR/Cas9 vector pPD162 (39) using the Q5 site-directed mutagenesis kit (New England Biolabs, E0554S). We used a forward primer that included the target sequence (5’- ctcgacacaatgagtccatGTTTTAGAGCTAGAAATAGCAAGT – 3’) and a general reverse primer (5’-CAAGACATCTCGCAATAGG-3’). Successful cloning was confirmed by sequencing the site of integration of the target sequence as well as different sites of the vector. Young adult wild type animals were injected with injection buffer (40) containing 50 ng/µl of pPD162 incl. target sequence and 10 ng/µl of the CFP injection marker pPD136.61 (Fire lab Vector kit, 1999, linearized with ScaI). Single fluorescent progeny were transferred to individual plates and checked for heterozygous mutations by the SURVEYOR® mutation detection kit from Transgenomics and by sequencing around the target sequence. Single progeny of heterozygous mutants were further transferred to individual plates and analysed for homozygous mutations (SURVEYOR® mutation detection and sequencing).

145

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 145 Reproduction rates To determine the number of progeny, every day individual worms were transferred to a fresh plate and the number of living progeny that hatched from the remaining eggs were counted 2-3 days later. 10 worms/experimental group were tested for each experiment. Reproduction rates were determined in three independent replicates and one representative curve is shown with error bars = SEM.

Seahorse respiration studies Oxygen consumption rates (OCR) were measured using the Seahorse XF96 cycler on day 4 of adulthood unless otherwise indicated. OCR was measured as pMoles/min/worm. Respiration behaviour was optimized for wild type and tdo- 2 deletion mutants at room temperature (41). The following cycles were used (1 loop = mixing: 2 minutes, waiting: 30 sec, measuring: 2 minutes): (1) calibration (2) equilibration (3) basal respiration = measurement for 5 loops (4) maximum mitochondrial capacity = first injection of FCCP, then measurement for 9 loops (5) non-mitochondrial respiration = first injection of sodium azide, then measurement for 4 loops. Mitochondrial OCR was determined by subtracting azide-independent OCR from basal respiration levels. Maximum mitochondrial capacity was calculated by subtracting non-mitochondrial OCR from maximum FCCP-induced respiration. Paraquat was injected after measurement of basal respiration levels and before injection of FCCP and measured for six loops. The following concentrations were used for injection: 20μl of 100μM FCCP, 20μl of 400mM sodium azide and 20μl of 200mM paraquat. Each experimental group was tested in 6-8 wells, each containing 10-30 worms/well (counted afterwards). Only conditions of the same run can be compared. Experiments were repeated and one representative graph is shown with error bars = SEM.

ATP measurements On day 4 of adulthood, about 1000 animals/experimental group were collected in PBS, extensively washed to remove as many bacteria as possible and finally snap frozen in PBS plus cOmpleteTM protease inhibitor (Roche). Frozen samples were boiled for 15 minutes to destroy ATPases and release ATP into the medium. About 500μl of ice-cold PBS was added and samples were subsequently kept on ice. Excessive worm debris was removed and the protein concentration was determined. ATP levels were determined using the ATP determination kit from Molecular Probes and subsequently normalized to protein levels. Measurements were repeated with individual biological replicates.

146

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 146 Lifespan experiments Experiments were performed at 20˚C. 100 animals per experimental group were tested in one single experiment. Lifespan assays were repeated independently and one representative curve is shown. Animals were treated with RNAi starting from larval stage L1 onwards unless otherwise indicated. Animals were transferred to FUDR-containing plates on day 1 of adulthood. At each time point, the surviving animals were counted and dead animals removed (those no longer responding to nose touch). Animals that disappeared during the assay were excluded from the analysis. Treatments respectively mutations were blinded before starting the experiment. Lifespan curves were generated using GraphPad Prism software.

ROS levels To estimate the levels of H O we made use of the jrIs1[Prpl-17::HyPer] strain recently 2 2, 6 described by 36. In short, the jrIs1 strain expresses a hydrogen peroxide-specific sensor (HyPer) that is composed of a circularly permutated yellow fluorescent protein

that is inserted into the H2O2-sensitive regulatory domain of the bacterial transcription factor OxyR. Since HyPer is under the control of the promoter of the large ribosomal subunit (17S), it is constitutively expressed in the cytosol (36). Oxidation of HyPer

by H2O2 induces changes in the fluorescent properties of the protein, shifting the excitation wavelength of 405 to ~490 nm. jrIs1 nematodes were age-synchronized and treated with RNAi from L1 onwards. Treatment with atp-3 RNAi was always started at adulthood day 1. At adulthood day 4, worms were collected in M9, washed multiple times and then mounted on 2% agarose pads. We paralyzed the worms with 10 mM tetramisole hydrochloride (levamisole: Sigma-Aldrich) and directly acquired high resolution images with the DeltaVision Elite microscopy system based on an Olympus IX-71 inverted epifluorescence microscope, using a 40X UPLFLN/1.3 objective. For illumination, we used the FITC filter set (475/28 excitation, 525/48 emission) and DAPI filter set with adapted emission (390/18 excitation, 525/48 emission), with the light source set at minimal intensity to avoid ROS induction. Images were analysed with Fiji, an open-source packaged version of ImageJ. First, the channels of the z-stacks were split. Next, z-stacks were stitched, generating maximum projections of the DAPI and FITC channel. Subsequently, the image calculator in Fiji was used to divide the values of the stitched FITC images by the DAPI ones. Outlines of the worms were selected and the mean ratiometric values of the images were calculated. Stitched images were also converted into intensity- normalized ratio (INR) images, providing a heat map of the ratios. Experiments were repeated 3 times with 10-15 worms/experimental group.

147

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 147 H2O2 stress assay To perform stress/death assays, we used a protocol adapted from 42. Briefly, synchronized adults (day 4) were collected, extensively washed and then pipetted (40 µl) into the wells of a 96-well microtiter plate at about 5-10 worms/per well.

Subsequently, H2O2 (10 mM) was dissolved in standard M9 buffer and 40 µl of

this solution added to the wells already containing worms, producing a final H2O2 concentration of 5mM. The number of surviving worms was counted every hour until most of the worms were dead (6 h). Experiments were repeated 3 times with 80-160 worms in total per experimental group.

Statistics Statistical analysis for lifespan experiments and stress assays was performed with a log-rank Mantel-Cox test. Data comparison for motility/healthspan as well as for respiration experiments was done by a one-way ANOVA with post-hoc Bonferroni algorithm. If relationships are not otherwise indicated, p values reflect a comparison with the control or wild type conditions.

148

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 148 REFERENCES

1. Oxenkrug, GF (2010) The extended lifespan of 17. Zubovych, I.O., Straud, S., Roth, M.G (2010) Drosophila melanogaster eye-color (white and Mitochondrial dysfunction confers resistance to vermilion) mutants with impaired formation of multiple drugs in Caenorhabditis elegans. Mol Biol kynurenine. J Neural Transm. 117(1): 23-26. Cell. 21: 956-969. 2. Campesan, S et al. (2011) The kynurenine pathway 18. Fougeray, S et al. (2012) Tryptophan depletion and modulates neurodegeneration in a Drosophila model the kinase GCN2 mediate IFN-gamma-induced of Huntington’s disease. Curr. Biol. 21: 961-966. autophagy. J Immunol 189: 2954-2964. 3. Pilotte, L et al. (2012) Reversal of tumoral immune 19. Houtkooper, RH et al. (2013) Mitonuclear protein resistance by inhibition of tryptophan 2,3-dioxygenase. imbalance as conserved longevity mechanism. Nature Proc Natl Acad Sci U S A 109(7): 2497-502. 497(7450): 451-7. 4. Van der Goot, AT et al. (2012) Delaying aging and the 20. Doonan, R et al. (2008) Against the oxidative damage aging-associated decline in protein homeostasis by theory of aging: superoxide dismutases protect against inhibition of tryptophan degradation. Proc Natl Acad oxidative stress but have little or no effect on life span Sci U S A 109(37): 14912-7. in Caenorhabditis elegans. Genes Dev 22: 3236-3241. 5. Van der Goot, AT and Nollen, EAA (2013) Tryptophan 21. Van Raamsdonk, JM, Hekimi, S (2009) Deletion of metabolism: entering the field of aging and age-related the mitochondrial superoxide dismutase sod-2 extends pathologies. Trends Mol Med 19(6): 336-44. lifespan in Caenorhabditis elegans. PLoS Genet 5(2). 6 6. Breda, C et al. (2016) Tryptophan-2,3-dioxygenase 22. Mesquita, A et al. (2010) Caloric restriction or catalase (TDO) inhibition ameliorates neurodegeneration by inactivation extends yeast chronological lifespan by modulation of kynurenine pathway metabolites. Proc inducing H2O2 and superoxide dismutase activity. Natl Acad Sci U S A. Proc Natl Acad Sci U S A 107: 15123-15128. 7. Opitz, CA et al. (2011) An endogenous tumour- 23. Yee, C, Yang, W, Hekimi, S (2014) The intrinsic promoting ligand of the human aryl hydrocarbon apoptosis pathway mediates the pro-longevity receptor. Nature 478(7368): 197-203. response to mitochondrial ROS in C. elegans. Cell 157 8. Zwilling, D et al. (2011) Kynurenine 3-monooxygenase (4): 897-909. inhibition in blood ameliorates neurodegeneration. 24. Schaar, CE et al. (2015) Mitochondrial and Cell 145(6): 863-874. cytoplasmic ROS have opposing effects on lifespan. 9. Maddison, DC, Giorgini, F (2015) The kynurenine PLoS Genet. 11(2). pathway and neurodegenerative disease. Semin Cell 25. Lee, SS et al. (2003) A systematic RNAi screen Dev Biol 40: 134-41. identifies a critical role for mitochondria in C. elegans 10. Ren, S and Correira, MA (2000) Heme: A regulator of longevity. Nat Genet. 33(1): 40-8. rat hepatic tryptophan 2,3- dioxygenase. Archives of 26. Hamilton, B et al. (2005) A systematic RNAi screen Biochemistry and Biophysics 377: 195-203. for longevity genes in C. elegans. Genes Dev. 19(13): 11. Oxenkrug, GF (2015) Increased plasma levels of 1544-55. xanthurenic and kynurenic acids in type 2 diabetes. 27. Curran, SP and Ruvkun, G (2007) Lifespan regulation Mol Neurobiol 52(2): 805-10. by evolutionarily conserved genes essential for 12. Guidetti, P, Luthi-Carter, RE, Augood, SJ, Schwarcz, viability. PLoS Genet. 3(4). R (2004) Neostriatal and cortical quinolinate levels 28. Platten, M, von Knebel Doeberitz, N, Oezen, I, are increased in early grade Huntington’s disease. Wick, W, Ochs, K (2015) Cancer immunotherapy by Neurobiol Dis 17(3): 455-461. targeting IDO1/TDO and their downstream effectors. 13. Stoy, N et al. (2005) Tryptophan metabolism and Front Immunol 5:673. oxidative stress in patients with Huntington’s disease. 29. Moran, GR (2005) 4-hydroxyphenylpyruvate J Neurochem 93: 611-623. dioxygenase. Archives of Biochemistry and Biophysics 14. Wu, W et al. (2013) Expression of tryptophan 433: 117-128. 2,3-dioxygenase and production of kynurenine 30. Kitagawa, T (2012) Hepatorenal tyrosinemia. Proc. pathway metabolites in triple transgenic mice and Jpn. Acad. 88. human Alzheimer’s disease brain. PLoS One 8(4). 31. Platten, M, Wick, W, Van den Eynde, BJ (2012) 15. Segref, A et al. (2014) Pathogenesis of human Tryptophan catabolism in cancer: beyond IDO and mitochondrial diseases is modulated by reduced tryptophan depletion. Cancer Res 72(21): 5435-40. activity of the ubiquitin/proteasome system. Cell 32. Metz, R et al. (2012) IDO inhibits a tryptophan Metab. 19(4): 642-52. sufficiency signal that stimulates mTOR – a novel IDO 16. Jovaisaite, V, Mouchiroud, L, Auwerx, J (2014) The effector pathway targeted by D-1-methyl-tryptophan. mitochondrial unfolded protein response, a conserved OncoImmunology: 1460-1468. stress response pathway with implications in health and diseases. J Exp Biol. 217(1): 137-143.

149

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 149 33. Bessede, A et al. (2014) Aryl hydrocarbon receptor 39. Dickinson, DJ, Ward, JD, Goldstein, B (2013) control of a disease tolerance defence pathway. Nature Engineering the Caenorhabditis elegans genome 511: 184-90. using Cas9-triggered homologous recombination. Nat 34. Zhai, L et al. (2015) Molecular pathways: Targeting Methods 10(10): 1028-34. IDO1 and other tryptophan dioxygenases for cancer 40. Mello, CC, Kramer, JM, Stinchcomb, D, Ambros immunotherapy. Clin Cancer Res 21(24): 5427-33. V (1991) Efficient gene transfer in C. elegans: 35. Bost, F, Decoux-Poullot, AG, Tanti, JF, Clavel, S extrachromosomal maintenance and integration of (2016) Energy disruptors: rising stars in anticancer transforming sequences. EMBO J 10(12): 3959-70. therapy?. Oncogenesis 5. 41. Koopman M et al. (2016) A screening-based 36. Back, P et al. (2012) Exploring real-time in vivo redox platform for the assessment of cellular respiration in biology of developing and aging Caenorhabditis Caenorhabditis elegans. Nat Protoc., accepted for elegans. Free Radical Biology and Medicine 52(5): publication. 850-859. 42. Possik, E and Pause, A (2015) Measuring Oxidative 37. Fraser, AG et al. (2000) Functional genomic analysis Stress Resistance of Caenorhabditis elegans in 96- of C. elegans chromosome I by systematic RNA well Microtiter Plates. J. Vis. Exp. (99). interference. Nature 408: 325-30. 38. Kim, D et al. (2013) TopHat2: Accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol 14(4).

150

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 150 We thank the Caenorhabditis Genetics Center, funded by the NIH National Center for Research Resources, and the group of Bart Braeckman for C. elegans strains. We thank Yin Fai Chan for experimental support and Klaas A. Sjollema for microscopy support. We thank the sequencing facility of the European Research Institute of the Biology of Aging for performing the RNA sequencing. We thank Sally Hill for editing the manuscript. This project was funded by a European Research Council (ERC) starting grant (to EAAN) and by support from the university alumni chapter ‘Gooische Groningers’ through the Ubbo Emmius Fonds (to EAAN).

Authors’ contributions: HM, MK, RIS and EAAN designed the experiments. HM and EAAN wrote the manuscript. HM and RIS performed health- and lifespan experiments and prepared the samples for transcriptome analysis. HM, MK, CFC and TA designed and performed the respiration assays. MK tested for ROS levels and performed the ROS stress assays. MK designed the respiration model and final model. HM and CNM generated the CRISPR knockout mutants. DHL performed the 6 lifespan assays of the kynurenine pathway, as well as the GCN-2 experiments. LN assayed the motility of amino acid-degrading enzymes. VG and TdJ did the RNA sequencing analysis. MvF and IK performed the metabolite measurements.

The RNA sequencing dataset has been submitted to EBI ArrayExpress, under accession number E-MTAB-4695.

151

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 151 SUPPLEMENTARY MATERIAL RNAi RNAi -2 -2 RNAi RNAi ctr tdo tdo ctr RNAi RNAi -2 RNAi -2 RNAi (del)/ (del)/ (del)/ (del)/ (del)/ (del)/ (del)/ (del)/ ctr tdo -1 -1 -1 ctr tdo -1 40 WT/ WT/ WT/ kmo kmo haao WT/ WT/ WT/ haao 40 30 30 20 20 days of adulthood of days 10 10 days of adulthood of days 0 0 0

80 60 40 20 80 60 40 20

100 survival percent 100 percent survival percent RNAi RNAi -2 RNAi -2 RNAi ctr tdo tdo ctr RNAi RNAi -2 -2 RNAi RNAi (del)/ (del)/ (pm)/ (pm)/ tdo ctr ctr tdo -1 -1 50 -2 -2 WT/ WT/ WT/ WT/ flu flu WT/ WT/ WT/ amfd amfd 40 40 30 30 20 20 days of adulthood of days days of adulthood of days 10 10 0 0 0 0

80 60 40 20 80 60 40 20

100 percent survival percent percent survival percent 100 B -2 -2 -1 -1 / FLU / -1 / TDO / serotonin / AMFD / / HAAO KYNU / TPH / acid - TPH TDO/ IDO TDO/ AFMID HAAO anthranillic acid hydroxy - - 3 anthranillic tryptophan kynurenine -1 -2 hydroxy - / C. elegans 3 kynurenine / FLU / / KMO / human KMO KYNU tryptamine A

152

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 152 RNAi. tdo-2 depletion. 40 RNAi tdo-2 -2 RNAi tdo ctr 30 response transcriptional del. A del. B del. 20 days of adulthood of days 10 ) in combination with (del. B) (del. -1 ahr-1 0 ahr translocation to the nucleus the to 0

20 40 60 80 100 survival percent -1 40

AHR 6 RNAi -2 RNAi 30 tdo ctr Survival curves of kynurenine mutants treated with control and Survival curves of kynurenine mutants treated

(B) 20 tryptophan metabolites days of adulthood of days C. elegans. 10 (del. A) (del. -1 ahr -1 0 0

40 80 60 -1 AHR 100 survival percent ahr D Survival curves of mutants the aryl-hydrocarbon receptor 1 ( (D) mutants. 50 (del B) (del) -2 -2 40 tdo-2 (del) WT tdo tdo 30 20 days of adulthood of days 10 0 0

80 60 40 20 100 survival percent C Schematic representation of the kynurenine, serotonin and tryptamine pathways in human and of the kynurenine, serotonin and tryptamine Schematic representation Survival curves of two CRISPR/Cas9-induced Figure S1. TDO-2 regulates lifespan independently of the kynurenine pathway and AHR-1 pathway and lifespan independently of the kynurenine TDO-2 regulates S1. Figure (A) (C)

153

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 153 (del B) (del) day 8 -2 -2 WT tdo tdo day 4 Motility/healthspan. Error Motility/healthspan. day 1 kynurenine levels (D)

0,1 0,0 0,4 0,3 0,2

mol/L μ C total 0 50 100 200 150 300 250 (del B) (del (del) -2 -2 WT tdo tdo D6 - day 8 D5 lysates on day 4 of adulthood. of 4 on day lysates D5 - day 4 D4 (del B) (del (del) C. elegans C. -2 -2 day 1 D4 - WT tdo tdo tryptophan levels D3

20 10 50 40 30 60 mol/L μ B D3 - D2 D2 - D1 exon 7 (2328 bp) -2 D1 - tdo L4 Tryptophan and kynurenine levels in levels kynurenine and Tryptophan STOP number of hatched progeny hatched of number exon 6 0 (B,C) 20 80 60 40 100 120 140 gene. E exon 5 tdo-2 translation STOP gfqkvfdtsfgipessvpcige exon 4 day 12 wisesi (del) (del B) -2 -2 exon 3 day 8 WT tdo tdo day 6 exon 2 …lkivsgldrmtkilsllteqitlldtmspldfvdfrkyltpasgfqslqfrvlenklgvrqerriky… …lkivsgldrmtkilsllteqitlldtms …lkivsgldrmtkilsllteqitl accgagcaaatcactttgctcgacaca atgagtccattggattttgtggatttcagaa agtatttga accgagcaaatcac tttgtggatttcagaaagtatttga accgagcaaatcac accgagcaaatcactttgctcgacaca atga gtggatttcagaaagtatttga Reproduction rates. Error bars = SEM. day 4 WT WT exon 1 (E) 2 (del) 2 (del) motility/ healthspan - - 2 (del B) (del 2 2 (del B) (del 2 - - 0 tdo tdo 20 80 60 40

100 tdo sec 30 bends/ body tdo Location of the CRISPR/Cas9-induced deletions in the in CRISPR/Cas9-induced deletions of the Location A D Figure S2. CRISPR/Cas9-induced TDO-2 deletion mutants show same phenotypes as RNAi-depleted animals S2. CRISPR/Cas9-induced Figure (A) bars = SEM.

154

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 154 RNAi RNAi 1 1 - - RNAi RNAi RNAi RNAi 2 - RNAi RNAi yars ctr wars ctr 1 1 ctr - - tdo RNAi RNAi RNAi 50 40 50 (del)/ (del)/ 2 (del)/ (del)/ RNAi - yars ctr ctr wars (del)/ 2 2 (del)/ 2 2 - - - - ctr tdo 2 2 - - WT/ WT/ WT/ WT/ tdo tdo tdo tdo 40 40 gcn 30 WT/ WT/ gcn 30 30 20 20 20 days of adulthood days of adulthood

days of adulthood 6 10 10 10 0 0 0 0 0

80 40 20 40 20

60 60

80 40 20 80 60 percent survival percent

100 survival percent 100 100 survival percent B RNAi 1 - RNAi RNAi 1 - RNAi cars ctr RNAi 1 - RNAi RNAi mars ctr 1 - 50 40 (del)/ (del)/ cars ctr RNAi 2 2 - - (del)/ (del)/ translation mars ctr WT/ WT/ 2 2 tdo tdo - - 40 30 WT/ WT/ tdo tdo

30 tRNA 2 - aa 20 GCN 20 days of adulthood 10 days of adulthood 10 tRNA aa 0 0 synthetase 0

40 20

60 80 40 20

60 80 percent survival percent tRNA percent survival percent 100 100 A C

155

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 155 ),

mars-1

complex V complex RNAi 2 - tdo

(del) 2 - asg RNAi ctr

***

RNAi 2 - tdo

RNAi ctr *** maximum mitochondrial respiration

50 40 30 20 10 C/ worm OCR/

Maximum mitochondrial capacity on day capacity mitochondrial Maximum

RNAi vdac (E-I) (del) *** 2

- RNAi 1 ctr - WT tdo *** PUS MRPS maximum mitochondrial capacity

50 40 20 10 30 C/ worm OCR/

RNAi 1 - pus Survival curves animals depleted of tRNA synthetases for methionine ( of tRNA depleted Survival curves animals (del) ***

- (C)

RNAi WT ctr tdo *** doxycycline maximum mitochondrial capacity

5 0

30 25 20 15 10 C/ worm OCR/ mitochondrial translation mutant animals. mutant G

gcn-2

RNAi 5 - mrps *** (del)

2 -

RNAi VDAC ctr WT tdo *** Schematic representation of parts of cytosolic and mitochondrial translation. mitochondrial and of cytosolic of parts representation Schematic (D) maximum mitochondrial capacity Survival curves of 0 ).

40 30 20 10 C/ worm OCR/ (B) yars-1 F

363) -

µM mTOR 20 (LET ** (del)

2 -

µM 10 ) and tyrosine ( tyrosine ) and WT tdo **

treatment

µM 0 0 cars-1 doxycycline doxycycline 2 - *** maximum mitochondrial capacity GCN 5

translation 25 20 15 10 30 ), cysteine ( ), cysteine C/ worm OCR/ E D wars-1 ) Schematic model of GCN-2-dependent translation. ) Schematic Figure S3. TDO-2 regulates lifespan independently of translation TDO-2 regulates S3. Figure (A ( tryptophan 4 of adulthood. OCR= oxygen consumption rate. Statistics for all panels: ns = not significant, ** <0.01, *** <0.001. Error bars SEM.

156

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 156 A fatty acid CoA

acetyl-CoA acyl CoA synthetase (acs-2)

3-ketoacyl CoA thiolase (kat-1) acyl-CoA acyl CoA dehydrogenase (acdh-2)

3-ketoacyl-CoA 2-trans-enoyl-CoA

3-hydroxyacyl- enoyl CoA hydratase CoA dehydrogenase L-3-hydroxy-acyl-CoA (hacd-1)

WT B C ratio tdo-2 RNAi/ ctr RNAi

RNAi kmo-1 tph-1 2 - RNAi WT (del) (del) tdo ctr 6 sod-2 ROS gst-4 vdac-1 hypoxia hsp-4 ER UPR hsp-6 mt UPR hsp-16.2 hsp-60 heat shock hsp-70

0 1 2

Figure S4. Tdo-2 depletion-induced transcript changes

(A) Schematic model of the mitochondrial beta oxidation cycle. Underlined genes were upregulated in tdo-2 depleted animals. (B) Expression of selected stress response genes on day 4 of adulthood. Blue = lower than average expression, yellow = higher than average expression. (C) Relative expression (tdo-2 RNAi/ ctr RNAi) of different components of oxidative phosphorylation on day 4 of adulthood.

157

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 157 A E maximum mitochondrial capacity *** *** *** 30 * ***

10 OCR/ worm OCR/

0 RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi -2 -2 -2 -2 -2 ctr ctr ctr ctr ctr tdo tdo tdo tdo tdo / WT (del) (del) (del) (del) (del) -3 -1 -2 -3 B ROS detoxification -2 sod sod sod sod

SOD-1 sod

paraquat

ETC H2O2 treatment NAC H 100 WT/ ctr RNAi SOD-2 WT/ tdo-2 RNAi ROS detoxification WT/ atp-3 RNAi SOD-3 50 C 100 WT/ ctr RNAi percent survival WT/ tdo-2 RNAi 80 0 sod-1 (del)/ ctr RNAi 0 2 4 6 8 sod-1 (del)/ tdo-2 RNAi 60 hours of treatment

40 I 0mM NAC percent survival 20 10mM NAC ns 0 70 0 10 20 30 40 ns ns 60 days of adulthood 50 100 WT/ ctr RNAi 40 WT/ tdo-2 RNAi 80 30 sod-2 (del)/ sod-3 (del)/ ctr RNAi 20 60 sod-2 (del)/ sod-3 (del)/ tdo-2 RNAi

body bends/ 30 sec 30 bends/ body 10 40 0 percent survival 20 WT 2 (del) - 13 (del) 0 - tdo

0 10 20 30 40 ced days of adulthood

D *** 2 ns

1,5

Ox/ Red Hyper0,5

0 RNAi RNAi RNAi -2 -3

158 ctr atp WT/ ctr RNAi WT/ tdo-2 RNAi WT/ atp-3 RNAi tdo WT

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 158 A E maximum mitochondrial capacity F G ctr RNAi 100 WT/ ctr RNAi *** *** *** WT/ tdo-2 RNAi tdo-2 RNAi 30 80 * *** ced-13 (del)/ ctr RNAi *** 50 60 ced-13 (del)/ tdo-2 RNAi *** 20 40 40 30 10 OCR/ worm OCR/ percent survival 20 20 0 worm OCR/ 0 10 0 10 20 30 40 RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi days of adulthood 0 -2 -2 -2 -2 -2 ctr ctr ctr ctr ctr WT (del) tdo tdo tdo tdo tdo 13 - / J 0mM NAC ced WT 20mM NAC (del) (del) (del) (del) (del) -3 -1 -2 -3 B ROS detoxification -2 50 *** sod sod sod ns sod

SOD-1 sod 40 paraquat 30 6 ETC H2O2 treatment ns NAC H ns 100 WT/ ctr RNAi 20 WT/ tdo-2 RNAi SOD-2 worm OCR/ ROS detoxification WT/ atp-3 RNAi 10 SOD-3 50 0 WT C WT 100 2 (del) 2 (del) percent survival - WT/ ctr RNAi - tdo

WT/ tdo-2 RNAi tdo 80 0 sod-1 (del)/ ctr RNAi 0 2 4 6 8 20mM 60 sod-1 (del)/ tdo-2 RNAi hours of treatment D paraquat D *** ** 40 2 2 * I 0mM NAC ns ns percent survival 1,5 20 10mM NAC 1,5 ns 70 0 1 1 0 10 20 30 40 ns ns

60 Ox/ Red Hyper days of adulthood 0,5

Ox/ Red Hyper0,5 50 100 0 WT/ ctr RNAi 40 0 RNAi RNAi WT/ tdo-2 RNAi RNAi -2 80 30 -3 ctr RNAi RNAi sod-2 (del)/ sod-3 (del)/ ctr RNAi RNAi atp tdo -2 -3

20 ctr 60 sod-2 (del)/ sod-3 (del)/ tdo-2 RNAi WT/ ctr RNAi WT/ tdo-2 RNAi WT/ atp-3 RNAi atp WT/ ctr RNAi WT/ tdo-2 RNAi WT/ atp-3 RNAi tdo WT body bends/ 30 sec 30 bends/ body 10 40 WT 0 percent survival 20 WT Figure S5. TDO-2 regulates mitochondrial activity independently of ROS production 2 (del) - 13 (del) 0 - (A) Schematic overview of mitochondrial energy metabolism. (B) Schematic overview of mitochondrial reactive oxygen tdo

0 10 20 30 40 ced species production and detoxification. NAC = N-acetyl-cysteine. (C) Survival curves of different sod mutant animals days of adulthood combined with tdo-2 depletion. (D) H2O2 levels in reporter strain. (E) Maximum mitochondrial capacity on day 4 of adulthood. (F) Survival curves of ced-13 mutant animals. (G) Maximum mitochondrial capacity on day 4 of adulthood. (H) D *** 2 Stress response to H2O2 treatment. (I) Motility/healthspan on day 4 of adulthood. (J) Maximum mitochondrial capacity on ns day 4 of adulthood. Statistics for all panels: ns = not significant, * <0.05, *** <0.001. Error bars = SEM. 1,5

Ox/ Red Hyper0,5

0 RNAi RNAi RNAi -2 -3 ctr 159 atp WT/ ctr RNAi WT/ tdo-2 RNAi WT/ atp-3 RNAi tdo WT

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 159 RNAi -1 RNAi ctr cdo 50 WT/ WT/ WT/ Survival curves when Survival 40 (D) 30 20 days of adulthood of days

RNAi -1 cdo

RNAi -1 hpd 0 *** ns

0 RNAi from day 1 day from RNAi RNAi -2

tdo 80

60 40 20

100

RNAi ctr * survival percent

day 12

RNAi -1 cdo

RNAi -1 *** hpd

* RNAi -2 tdo RNAi

RNAi ctr day 8

-1 ***

RNAi

RNAi -1 cdo ctr hpd ns

RNAi -1 hpd

RNAi -2 tdo 30 ** WT/ WT/ WT/

RNAi * ctr day 4 RNAi from day 1 day from RNAi

50 30 40 20 10 body bends/ 30 sec 30 bends/ body 20

RNAi 268 - let

Motility/healthspan when depleting amino acid dioxygenases after development. dioxygenases after acid amino when depleting Motility/healthspan 10 ns

days of adulthood of days RNAi 2 - gbh

(C)

ns RNAi 2 - pdi

*** RNAi 1 - nmad 0

ns 0

RNAi from day 1 day from RNAi RNAi

K07E1.1 80 60 40 20

100 percent survival percent

ns RNAi 1 - hpd

RNAi 1 - cdo

***

RNAi RNAi 2 - tdo

-2

RNAi

mitochondrial respiration ***

RNAi - ctr 40 ctr tdo non WT/ WT/ WT/

0 4 3 2 1 5 C/ worm OCR/ 30 B

RNAi 1 - cdo

RNAi 1 - *** hpd

days of adulthood of days ** RNAi 2 -

tdo 10

** RNAi - ctr non mitochondrial respiration 0

0 RNAi from day 1 day from RNAi 2 4 0 3 1

C/ worm OCR/ 80 60 40 20

100 percent survival percent Non-mitochondrial oxygen consumption (OCR) on day 4 of adulthood. Non-mitochondrial D A Figure S6. TDO-2 shares functions in respiration control and health- and lifespan regulation with other amino acid-degrading dioxygenases with other and health- lifespan regulation control functions in respiration TDO-2 shares S6. Figure (A,B) depleting dioxygenases after development. Error bars = SEM. Statistics: ns=not significant, *<0.05, ** <0.01, *** <0.001.

160

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 160 6

161

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 161 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 162 Chapter VII

A screening-based platform for the assessment of cellular respiration in Caenorhabditis elegans

Mandy Koopman1, Helen Michels1, Beverley M Dan- cy2, Rashmi Kamble3, Laurent Mouchiroud4, Johan Auwerx4, Ellen A A Nollen1 & Riekelt H Houtkooper3

1Laboratory of Molecular Neurobiology of Ageing, European Research Institute for the Biology of Ageing, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 2Laboratory of Cardiac Energetics, National Heart, Lung and Blood Institute, Bethesda, Maryland, USA. 3Laboratory Genetic Metabolic Diseases, Academic Medical Center, Amsterdam, the Netherlands. 4Laboratory of Integrative and Systems Physiology, Ecole Polytechnique Federale de Lausanne, Lausanne, Switzerland.

Koopman M, Michels H, Dancy BM, Kamble R, Mouchiroud L, Auwerx J,

Nollen EAA, Houtkooper RH. A screening-based platform for the assessment of cellular respiration in Caenorhabditis elegans. Nat Protoc. 2016 Oct; 11(10): 1798-816.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 163 Mitochondrial dysfunction is at the core of many diseases ranging from inherited metabolic diseases to common conditions that are associated with aging. Although associations between aging and mitochondrial function have been identified using mammalian models, much of the mechanistic insight has emerged from Caenorhabditis elegans. Mitochondrial respiration is recognized as an indicator of mitochondrial health. The Seahorse XF96 respirometer represents the state-of-the-art platform for assessing respiration in cells, and we adapted the technique for applications involving C. elegans. Here we provide a detailed protocol to optimize and measure respiration in C. elegans with the XF96 respirometer, including the interpretation of parameters and results. The protocol takes ~2 d to complete, excluding the time spent culturing C. elegans, and it includes (i) the preparation of C. elegans samples, (ii) selection and loading of compounds to be injected, (iii) preparation and execution of a run with the XF96 respirometer and (iv) postexperimental data analysis, including normalization. In addition, we compare our XF96 application with other existing techniques, including the eight-well Seahorse XFp. The main benefits of the XF96 include the limited number of worms required and the high throughput capacity due to the 96-well format.

164

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 164 Mitochondria have long been considered the crucial organelles responsible for the production of ATP1. To perform their key roles in energy production, mitochondria use an intricate system that encompasses the breakdown of substrates—such as glucose, fatty acids and amino acids—coupled to the generation of ATP via oxidative phosphorylation2,3 (Fig. 1a,b). In the past few years, it has become increasingly apparent that mitochondria, the ‘powerhouses’ of the cell, are not only involved in the production of cellular energy, but are also intimately involved in a variety of processes and signaling pathways including, but not limited to, apoptosis and calcium homeostasis (as reviewed in refs. 4–6). As a consequence, it does not come as a surprise that mitochondrial dysfunction is associated with a broad spectrum of dis eases, including inherited metabolic diseases, as well as common conditions such as aging, neurodegenerative diseases, cancer and diabetes7–14.

Figure 1 | A schematic overview of glycolysis and oxidative phosphorylation. (a) The mitochondrion with its characteristic structure and the translocation of pyruvate, which is the result of glycolysis. During glycolysis, H+ is generated, which causes extracellular acidification. b( ) Energy-related metabolic pathways that take place in the mitochondrial matrix. Oxidative phosphorylation requires the consumption of O2. Compounds known to inhibit the different complexes (as shown with the roman numerals) are listed in gray (rotenone, antimycin A, oligomycin and azide). FCCP is an uncoupler reagent and transfers H+ back to the mitochondrial matrix without generating ATP. Acyl-CoA, Acyl-Coenzyme A.

Numerous in vitro, in situ and in vivo methodologies have been developed to assess various aspects of mitochondrial function1,15,16. Assays include the estimation of mitochondrial membrane potential 17, the determination of the production and/ or concentration of ATP with bioluminescence18 and the determination of calcium

165

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 165 retention19. Although these assays measure one specific aspect of mitochondrial function, they do not cover the full extent of mitochondrial health. The oxygen consump tion rate (OCR) of mitochondria provides an additional layer of complexity that is dependent on many sequential reactions in mitochondria. As such, the measurement of mitochondrial respiration at the level of the electron transport chain (Fig. 1b) has emerged as a proxy for mitochondrial health20–22. Indeed, alterations in the rate of oxygen consumption are described as an informative indicator of mitochondrial dysfunction23. It was the introduction of the Seahorse extracellular flux analyzers (hereafter called XF respirometers) that allowed measurement of

O2 consumption in multiple parallel wells without the need to lyse cells or isolate mitochondria1,24.

Although initially developed for cultured or isolated cells, XF respirometers can also be used for whole organisms such as C. elegans, when taking some important adjustments into account25–28. C. elegans is often used as a model organism to study mitochondria in the context of longevity29. It is the ease of genetic modifications (e.g., RNAi, CRISPR) that renders the exploration of mitochondria-related pathways in worms readily accessible30–32, especially considering that many of the genes involved in human metabolism are conserved in C. elegans33–36. Consequently, we can use C. elegans as a simple model to study the complex interplay between genes and mitochondrial function, which is usually measured at the level of gene expression (e.g., heat-shock protein induction) or phenotypic characterization of live animals (e.g., locomotion or pharyngeal pumping rates). Nevertheless, the progress of the mitochondrial field in C. elegans was long delayed by the limited options for the dedicated measure of metabolism, in particular mitochondrial activity. Hence, we set out to develop a platform using Seahorse respirometry, which was successfully applied in several projects15,26,27. Here we describe in detail how Seahorse XF96 respirometers can be used to study mitochondrial biology in C. elegans. Moreover, we also provide detailed information about how parameters can by optimized for individual worm strains.

Theoretical background of XF96 respirometers The XF96 respirometers provide researchers with the possibility of performing real- time measurements of the OCR and extracellular acidification rate (ECAR). The OCR is predominantly the result of mitochondrial respiration through oxidative phosphorylation (Fig. 1b), whereas ECAR is predominantly the result of glycolysis (Fig. 1a). The system uses a 96-well plate combined with a sensor cartridge containing an equal number of individual sensor probes.

At the tip of each of the sensor probes, there are two separate polymer-embedded

166

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 166 + fluorophores that are sensitive to either 2O or H . The sensitivity of theses probes is based on the quenching chemistry of the fluorophores in response to 2O and H+ in

+ the assay medium. The concentrations of O2 and H are therefore directly linked to the intensity of the fluorescence signal relative to a standard solution. Moreover,

any changes in the signal can be inferred to be proportional to changes in O2 or

+ + H concentrations, and therefore rates of change in O2 or H concentrations can be obtained by following the fluorescent signal over time.

During a measurement cycle, the sensor cartridge is lowered into the wells, creating a transient microchamber of a defined volume (Fig. 2a). Fiber-optic bundles are inserted into the probes, and light is pumped into the sensor to excite the embedded fluorophores. The resulting emitted light is transmitted back through the fiber-optic bundle and measured by the detector. The fluorescent emission is measured for a

specified period (measuring time; Fig. 2b). The slope of the linear decline (for O2 concentration over time) or incline (for H+ concentration over time) provides the basis for calculating the OCR (in picomoles per minute) and ECAR (milli pH units per minute), respectively (Fig. 2b). The exact algorithm that is used to calculate the 7 OCR in XF respirometers is described in Gerencser et al.22.

Figure 2 | The measurement units of XF respirometers. (a) Dashed red box shows a close-up of the probes of the sensor cartridge (solid red box). During the measuring phase, probes of the sensor cartridge are lowered, thereby creating a transient microchamber (from resting phase to measuring phase). Light is transmitted through the fiber-optic bundles to excite the embedded fluorophores, which are sensitive to O2 and H+, at the tip of the probe. Emitted light then passes back up the optic bundle to the detector. Compounds can be injected during the experimental assay via ports at the base of the probe. (b) During a measuring cycle, oxygen is consumed and the concentration of the analyte drops. A representative measurement is shown. The mixing phase represents the stage at which the probe is repeatedly moving between the resting position and the measuring position, allowing for mixing of the medium. During the measuring phase, the O2 concentration is estimated nine times per loop, when measuring for 2 min. These nine estimations of the oxygen concentration (or ticks, as directly derived from the ‘level data’ in the Excel file) together are used to calculate one OCR per loop. Here, the linear decline of O2 concentration (red dotted line) provides the basis for the OCR estimation.

167

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 167 After several minutes of continuous measurements, the sensor cartridge is raised and subsequently allowed to move up and down for a specified time interval, which causes the larger volume of medium above to mix with the medium in the transient

+ micro chamber, thereby restoring the O2 and H concentrations (mixing time; Fig. 2b). To get multiple, steady estimations of the OCR and ECAR, the measuring and mixing cycle can be repeated (looped). In addition, the presence of an integrated drug delivery system in the sensor cartridge plate allows sequential addition of up to four compounds/solutions per well at user-defined time points. Selection of the compounds to be added, and the sequence in which they are added, makes it possible to differentiate between several aspects of cellular respiration; for example, non- mitochondrial res piration can be determined by blocking mitochondrial respiration with a compound (e.g., rotenone or antimycin A)37–40 (Fig. 1b).

Comparison with other in vivo respirometry methods Several methods are available that measure respiration of living samples; these can

be divided into two general groups: methods that involve O2-dependent quenching of porphyrin-based phos phors (using a Seahorse Bioscience XF respirometer and

Luxcel MitoXpress) and those that involve amperometric O2 sensors (using Clark electrodes, including the widely adopted Oroboros system)1,41. Historically, the amperometric approach has been the main method used to assess mitochondrial respiration in C. elegans. For the amperometric approach, nematodes are delivered

into a single respiratory chamber, which is separated from two half-cells by O2-

permeable material. In this way, only O2 can diffuse from the assay medium through

the membrane. When a small voltage is applied to the half-cells, O2 is reduced by

electrons at the cathode, yielding hydrogen peroxide. Subsequently, H2O2 oxidizes the Ag (silver) of the Ag/AgCl anode, which results in an electrical current that

is proportional to the O2 pressure—and thus concentration—in the experimental respiratory chamber.

Apart from the detection modality, differences between XF respirometric methods appear in relation to the number of worms per assay, number of replicates, multiple or real-time measurements, and the ability to inject compounds during an experiment (Table 1). The Clark electrode approach requires thousands (~2,000–5,000) of worms in a single chamber to obtain an estimation of the OCR42. Performing multiple measurements and biological rep licates, and comparing conditions provide the biggest challenges within the Clark electrode method, as the traditional setup allows the measurements of only one sample at a time. By contrast, an XF96 respirometer requires ~10–20 worms per well to acquire a reproducible OCR; measurements can be easily and quickly (in the order of minutes) repeated in an automated way; and as XF respirometers can analyze whole plates at the same time, ~96 conditions/

168

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 168 replicates can be tested at once. An additional difference is the presence of drug-injection ports that can be programmed to inject compounds into all 96 wells at time points that are specified a priori during an XF respirometer experiment. Clark electrode systems also allow injection of compounds, and even offer flexibility with respect to the timing, dosing and number of additions, as compounds are injected manually during the course of the assay. However, precise timing of manual additions between replicate experiments may be challenging. More similar to the Seahorse XF 7 respirometer method is the Luxcel

MitoXpress O2 consumption assay,

which relies on O2-dependent quenching of porphyrin-based phosphor. The MitoXpress kits provide a way of performing real-time analysis of cellular respiration via an oxygen- quenching fluorophore system. Worms are placed into the wells of a 96- or 384-well plate, the kit reagents are added and measurements are made in a fluorometric plate reader. Multiple conditions and replicates can be tested side by side in the wells of a single plate, but repeated measurements over time are more challenging, as there is typically no automatized mixing system integrated into the plates or plate readers

to restore basal O2 levels. In addition, single estimation of the OCR takes >90 min, whereas careful estimations of the OCR in the XF respirometer approach take only 2–5 min of measuring time. Finally, the use of compounds to assess

169

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 169 multiple aspects of mitochondrial function related to oxygen consumption is limited, as the compounds need to be injected manually immediately before the start of the experiment.

Advantages and limitations of XF respirometry On the basis of the comparisons with other respirometric methods, the XF respirometer approach offers important benefits for the measurement of mitochondrial respiration in C. elegans, espe cially when considering the ease of performing repeated measurements, comparing conditions and replicates, and investigating several aspects of mitochondrial function. This allows a higher throughput for screening-based applications—e.g., genome-wide screens to find genetic interactors affecting mitochondrial bioenergetics and compound screens for mitochondrial toxicity. Indeed, XF respirometry has already been successfully applied to study the effects of compounds or gene knockdowns on mitochondrial function in C. elegans15,26,27.

In vivo assessment of mitochondrial function in C. elegans also has, however, some clear limitations. First of all, oxygen consumption of an intact organism (e.g., basal respiration) does not strictly reflect mitochondrial respiration. For instance, cells possess a variety of oxygenases that also consume oxygen and thereby contribute

Figure 3 | Distinguishing non-mitochondrial respiration from mitochondrial respiration. (a) The oxygen consumption in cells and organisms is not due to mitochondrial respiration only. Therefore, one should add a compound (e.g., sodium azide) to block mitochondrial respiration, in order to distinguish mitochondrial oxygen consumption from other oxygen- consuming processes. (b) Layout of the drug injection ports per well; indexed as A–D when considering the notch at the bottom left.

170

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 170 to the OCR43. Hence, residual respiration should be estimated in the presence of effective electron transport inhibitors in order to distinguish non-mitochondrial respiration from mitochondrial respiration (Fig. 3a). In cells, the use of four compounds (making use of the drug injection ports; Fig. 3b) to distinguish between and to quantify several aspects of cellular respiration is well established. These aspects include, but are not limited to, the following: (i) respiration linked to ATP production (using oligomycin); (ii) maximal respiration (using carbonyl cyanide- 4(trifluormethoxy)phenylhydrazone: FCCP); and (iii) mitochondrial and 7 non-mitochondrial respiration (using antimycin A and rotenone; Table 2; Fig. 1b; refs. 37–40). Nevertheless, C. elegans has traditionally been considered a poor candidate for drug-related assays because of the relatively inefficient uptake of compounds caused by impermeability of the cuticle to non–water soluble compounds44–46. Consequently, it does not come as a surprise that, except for FCCP, the compounds typically used in cultured cells are inefficient in acutely inhibiting specific aspects of mitochondrial function in C. elegans (Supplementary Fig. 1)28. However, there are alternative compounds, such as the complex IV/V inhibitor sodium azide, that can be used to help decipher aspects of mitochondrial respiration in C. elegans28,47 (Fig. 1b).

One can argue that isolating mitochondria of C. elegans would represent a more sophisticated way to assess mitochondrial func tion. Isolated

171

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 171 mitochondria make exploration of the molecular (super)complexes involved in oxidative phosphorylation possi ble48. Changes in respiratory chain complexes and anaplerotic enzymes can be identified using combinations of substrates and inhibitors specific to the isolated mitochondria49, which allows careful dissection of the respiratory chain. Nevertheless, the clear advantages of isolated mitochondria are also accompanied by several disadvantages. First of all, existing methods often require large amounts of sample because of a poor yield of mitochondrial content, and they do not prevent biased selection toward specific mitochondrial populations50–53. Second, the isolation process may decrease mitochondrial membrane integrity and disrupts both the mitochondrial environment and the cellular/organismal context, which affects mitochondrial morphology, respiration and reactive oxygen species (ROS) production52,53. In addition, mitochondrial isolation procedures are technically challenging in C. elegans compared with tissue samples or even cells, precluding the use of such a strategy for routine biochemical analysis54. As such, it is clear that the XF respirometers provide the user with a platform for performing relatively quick high-throughput screens in an organismal context. In this way, interactions of the mitochondria with the rest of the cell are maintained, and so is the physiological relevance.

Finally, in addition to the aforementioned considerations for using XF respirometers for assessing mitochondrial function in C. elegans, some critical technical ‘limitations’ should be mentioned. As XF respirometers were originally developed for cells, the tem perature control/heater of the machine is designed to work at 37 °C. Considering that nematodes are typically cultured at 20–25 °C, and their physiology is dependent on the environmental temperature, the lack of tight temperature control could affect the outcome of the experiments. Depending on the environmental setting of the XF respirometer, turning the heater off will result in temperatures of ~20–25 °C, which will increase during the run by ~2–4 °C. We have found that temperatures in the range of 20–25 °C will not cause any significant changes in the respiration rate of wild-type C. elegans (Fig. 4). Nevertheless, it is strongly advised to compare strains within only one plate to account for masked effects of temperature on respiration. Another aspect to consider is that ECAR cannot be assessed in worms when using the pro tocol that we describe here, as worms are assayed in a buffered medium (M9, see below) that precludes changes in H+ concentration. A method for measuring ECAR in C. elegans with an XFe24 respirometer has been described recently that uses unbuffered water as the assay medium28. Nonetheless, even when it is possible to estimate the ECAR, it is unclear whether this represents a valid marker for glycolytic rate. Further validation of these parameters will hopefully shed light on this option and expand the toolkit for measuring metabolism in C. elegans.

172

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 172 Figure 4 | Temperature does not influence the OCR of C. elegans. Up to 25 °C, the basal respiration of nematodes (N2) is not influenced by the temperature. Bars represent mean ± s.e.m. One-way ANOVA (not significant), n = 7–9 wells. The dashed red line represents the OCR at 20 °C. 7 Alternative applications The XF96 respirometer method provides a flexible platform for performing genome- wide screens or compound screens in relation to mitochondrial respiration. As the experimental setup requires only that organisms be viable in liquid culture and have a size that fits the transient microchamber (XF96: 3 μl and XF24: 7 μl), mitochondrial respiration can also be estimated in species other than C. elegans. Therefore, Seahorse XF respirometers might open new avenues for other well-established or upcoming species that may be used in metabolic or aging studies, such as the flat- worm Macrostomum lignano55, Danio rerio embryos56,57 and even Nothobranchius furzeri embryos58. Although XF respirometers are, to our knowledge, not used in M. lignano or N. furzeri research yet, it has been shown that respiratory kinetics in embryos from D. rerio can be investigated with XF respirometers57.

Overview of the procedure Performing an XF respirometer experiment with C. elegans is technically a simple procedure that consists of six main steps: (i) pre-experimental procedures; (ii) preparation of the XF respirometer; (iii) loading of compounds; (iv) collection and loading of worm samples; (v) the actual XF experiment; and (vi) postexperimental data analysis. An overview of the experimental procedure is given in Figure 5. Optimization of steps i–iv is required in order to obtain reliable estimations of the OCR and to reduce variation within runs. Some of the preparatory steps are very similar to those used in cells49. However, many of the exact parameters and steps cannot be translated one-on-one to C. elegans, and it is therefore highly

173

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 173 Figure 5 | A schematic overview of an XF respirometer experiment. Overview of experimental procedures involved in preparing and performing an XF assay with C. elegans. The procedure consists of (1) synchronizing worms; (2) aging worms; (3) preparing compounds to be injected; (4) hydrating the probes of the cartridge (Steps 1–4 in the PROCEDURE); (5) preparing the XF respirometer (Steps 5–14 in the PROCEDURE); (6) loading the compounds in the injection ports (Steps 15–20 in the PROCEDURE); (7) collecting, washing and loading worm samples (Steps 21–35 in the PROCEDURE); and (8) executing the actual respirometer run and performing data analysis (Steps 36–46 in the PROCEDURE). See the main text for a more detailed explanation of all steps involved.

174

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 174 recommended to take note of our proposed optimization steps before continuing to the PROCEDURE section.

(i) Pre-experimental procedures. The pre-experimental procedures should be performed 1 d before the actual XF experiment, but it will in general take more than 1 d of preparation. Comparing OCR of different worm strains requires strict age synchrony, as respiration changes significantly during development and aging Fig.( 6). Hence, it is important that careful phenotyping be per formed before the respiration experiments in order to anticipate potential differences in the rate of development and worm size. When assessed, worm strains can be age-synchronized (hypochlorite treatment, or egg laying) several days before the experiment59. To avoid offspring or eggs from hatching, one can use plates containing 5-fluoro-2′-deoxyuridine (FUdR; starting from young adults)60. Here, it is important to realize that FUdR can alter the biology of the worms, as evidenced by reported changes to life span, worm size and metabolite levels61,62. Other techniques for maintaining synchronized worm cultures are described in Gruber et al.63. In addition to preparing the worm samples, the cartridge needs to be prepared in advance in order to hydrate the sensor probes. 7 The sensor cartridges should be incubated with Seahorse Bioscience XF96 calibrant

solution, pH 7.4, at 37 °C without CO2 (to keep the calibration temperature constant and optimal for XF respirometers). It is recommended that the probes be hydrated overnight, but a minimum hydration time of 4 h is required, and, if necessary, the hydration can be prolonged to 72 h, in which case we advise sealing the plate with Parafilm and storing it at 4 °C. 1 d before the assay, one can also prepare stock solutions of the compounds to be injected during the Seahorse experiment, including

Figure 6 | Respiration rates change during development and with age in C. elegans. Basal respiration changes during development (from L4 to adult stage) and aging in N2 worms. All adult stages show significantly higher respiration than the larval stage L4. In addition, respiration is significantly higher at day 4 of adulthood (D4) as compared with day 1 (D1;P < 0.05), as indicated by ‘a’. Bars represent mean ± s.e.m. One-way ANOVA (P < 0.001) with post hoc Tukey test, n = 6–8 wells. **P < 0.01, ***P < 0.001.

175

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 175 FCCP (10 mM in DMSO: 1,000× the intended working concentration) and sodium azide (400 mM in dH2O: 10× working concentration). It is important to realize that compounds are ten times diluted when they are injected in the wells (because of the volume already present). Therefore, stock solutions need to have a concentration that is ten-fold higher than the intended working concentration. Moreover, for compounds that are dissolved in DMSO, it is strongly suggested that one make a stock that is 1,000× concentrated, to ensure that the DMSO concentration is ≤1% (dilute 100× to get the stock required for injection)64.

(ii) Preparation of the XF respirometer. In anticipation of the respiration measurements, the XF96 software should be prepared and programmed. Although the exact steps are described in the PROCEDURE, there are some important aspects that require clarification. As XF respirometers were initially developed for cell- related research, the machine has an automatic heater and temperature control system to maintain a temperature of 37 °C. For Seahorse experiments with C. elegans, it is essential that the temperature control system and heater be turned off in advance, thus allowing the machine to adapt to room temperature (~20–25 °C). Be aware that every time you restart the respirator, the temperature control system and the heater have to be turned off.

The ‘assay wizard’ embedded in the Seahorse software provides a user-friendly interface for generating an experimental template with all the commands to control the XF96 respirometer during the experiment. Many of the tabs and fields are filled at the user’s discretion, including, for instance, user info, medium info and plate layout. Although these items can be helpful in tracing the experimental conditions, they do not affect the XF run. However, it is important to adjust three tabs: ‘General’, where you fill in the date and name of your experiment; ‘Background correction’, where you select the wells that function as correction wells (these contain buffer only); and ‘Protocol’, where you specify the order and duration of mixing, measuring and waiting time, as well as the injection of compounds. The ‘Protocol’ tab provides you with several commands that can be executed by the machine. Although most of them are self-explanatory, we provide here a brief overview:

Calibrate. This is always the first command of each XF assay; it calibrates the sensors for the planned experiment. This takes ~15–20 min.

Equilibrate. This is always the second step after calibration. It consists of three loops of 2 min each of mixing and waiting. Equilibration is performed to stabilize the temperature of the plate.

Mix. This command sets the mixing of the contents of the wells by gentle up/down + motion of the sensor sleeves. Mixing is necessary to restore analyte (O2 and H )

176

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 176 distribution and concentration in the wells and to mix injected compounds.

Wait. This command instructs the machine to wait for x number of minutes.

Measure. This command instructs the machine to do the actual measurements to determine the OCR and ECAR.

Inject. This command instructs the machine to inject compounds that are loaded in the ports into the wells. Here you can also choose which port should be injected (A, B, C or D).

Loop. This command is used to repeat a subset of commands x number of times.

The duration of the commands ‘Mix’, ‘Wait’ and ‘Measure’ should be carefully optimized to ensure a stable OCR over time; see Optimization of parameters. In addition, multiple loops should be performed to obtain not one but multiple estimations of the OCR. When the experimental template is ready, the actual run can be directly started; however, to avoid unnecessary heating, it is important to start the run only when the samples are almost ready. 7

(iii) Loading compounds. The benefit of working with XF96 respirometers is the possibility of using the integrated drug delivery system that allows sequential injection of up to four compounds. With the use of specific mitochondrial toxins, a more complete profile of mitochondrial function can be generated. When an XF experimental template is generated, it is possible to include the command ‘Injection of port x’, where x refers to a port indexed with a specific letter: A–D. The drug ports are organized in the following way around each probe: A = top left, B = top right, C = lower left and D = lower right (Fig. 3b). To facilitate the loading procedure, the sensor cartridge is accompanied by two loading guides—i.e., lids with holes that allow easy access to pipette only in the port of choice. When the sensor cartridge is loaded with the compound(s), the plate can be placed back in an incubator at 37

°C without CO2. It is important to understand that the injection is initiated through a pneumatic mechanism. As a consequence, when using a specific injection port, this port should be filled for all 96 wells, even if those wells are not used for the oxygen consumption measurement.

There are several compounds that can be injected to assess different aspects of respiration. A drug that we have extensively tested and that is often used in combination with XF96 respirom eters is FCCP. FCCP is a proton ionophore, which means that it is able to transport hydrogen ions across membranes. FCCP is therefore an uncoupling agent, as it disrupts ATP synthesis by transporting hydrogen ions across the inner mitochondrial membrane instead of the proton channel of the ATP synthase (complex V)65. Consequently, administration of FCCP results in the collapse

177

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 177 of the mitochondrial membrane potential, thereby inducing a rapid increase in the consumption of energy and oxygen without the generation of ATP. FCCP injection can be used to calculate the maximal and spare respiratory capacity of worms. The latter is defined as the quantitative difference between the initial basal OCR and the maximal uncontrolled OCR (induced by FCCP; Fig. 7). Another drug that we have tested is sodium azide. Sodium azide blocks both cytochrome c oxidase (complex IV) and the ATP synthase (complex V), thereby shutting down the whole electron transport chain47. Therefore, we use sodium azide to estimate non-mitochondrial respiration—i.e., respiration caused by other cellular processes than oxidative phosphorylation (Figs. 3a, 7).

Figure 7 | Typical example of OCR profiles in C. elegans respirometry. Starting with the basal respiration, the spare capacity and maximal (mitochondrial) respiratory capacity (plateau phase) can be estimated by injecting FCCP. Non- mitochondrial respiration can be distinguished from mitochondrial respiration by injecting sodium azide. Dashed red lines indicate injections of compounds.

(iv) Collecting and loading worm samples. When collecting the nematodes for the XF respirometric assay, several aspects should be taken into account. First, before removing worms from their plates, it is important to make note of any abnormalities—e.g., infections by fungi and the amount of food present (starva tion). These observations may be critical in the interpretation of the data. Furthermore, collected worms should be washed multiple times to get rid of Escherichia coli. However, we have shown that the contribution of the E. coli to respiration is marginal and therefore will not affect the respiration data extensively (Supplementary Fig. 2). By estimating the number of worms per volume unit, one can estimate the volume required for a specific number of worms per well. This number needs to be optimized per strain and age; see Optimization of parameters.

178

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 178 Worms are loaded on a freshly opened Seahorse Bioscience cell culture microplate (the utility plate), keeping the total volume per well at 200 μl (= microliters of worms + microliters of M9). Use an XF96 plate layout (Supplementary Fig. 3) to keep track of which samples were placed in which wells. Visual inspection of the wells should show approximately the same number of worms in each experimental well and no worms in the background correction wells (typically the four corner wells: A1, A12, H1 and H12). The time between collecting the worms from their plate and loading them into a XF96 utility plate should be as short as possible, but not shorter than 30 min, to ensure that the worms have cleared their gut and the marginal oxygen consumption due to E. coli is further reduced. The OCR in the worms has a slight tendency to drop over time; therefore, a much longer waiting time, e.g., 2 h, should be avoided as much as possible.

(v) Starting the XF experiment. When the XF respirometer is prepared in advance, one can start the assay directly. We suggest starting the XF experimental assay when worms are collected, but not yet loaded (PROCEDURE), to avoid unnecessary heating of the machine before running the actual experi ment. Calibration, the 7 first command of an experimental template, takes ~20 min, during which the worms can be transferred to the utility plate. Just before starting calibration, we advise that one briefly place the sensor cartridge with calibrant solution at 37 °C, so that the calibration will take place at ~37 °C, which is the optimal temperature for the instrument’s determination of actual oxygen concentrations. However, as the protocol is set to room temperature (i.e., the heater is turned off), the interior of the wells and machine will actually be ~20–25 °C during the measurements. Once the worms are loaded into the wells and the calibration has completed, the experimental part of the XF assay can start. The results of the XF assay can be followed in real time, but it should be mentioned that the OCR is represented per well and not per worm or μg protein. In addition to that, the XF software can be unstable during the assay, causing it to crash, which is why we discourage using the software during the run.

(vi) Postexperimental data analysis. In order to interpret the data and compare strains or treatments, the OCR per well should be converted to OCR per worm (or OCR per microgram of protein). Therefore, the number of worms per well needs to be counted. This can be done directly, or one can capture images of each well using a microscope and count the number of worms per well at a later date. Use the XF96 plate layout sheet (Supplementary Fig. 3) to fill in the number of worms per well. Some mutant or RNAi-treated worms are smaller than the age-matched controls. In this case, correcting for protein levels can provide additional information, although mitochondrial deficiency through RNAi leads to a reduced OCR with both normalization protocols66. In addition, differences in mitochondrial activity can be picked up by normalization to mtDNA content67.

179

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 179 While inspecting the wells and/or images, also make note of any anomalies observed in the wells, such as dust or dirt, bacteria, embryos or larvae, as these may be important when interpreting the respiration data in the next step. An Excel spreadsheet is generated by the program after the XF assay is done. There will be different tabs in this spreadsheet, starting with ‘Data viewer’. Although most of the information within the tabs is self-explanatory, a quick overview is given:

- Data viewer. When the Seahorse XF96 software is installed on the same computer that you use for analysis, there will be a Seahorse plugin in the ‘Data viewer’ tab (Windows only). Here, the data representation is identical to that on the XF respirometer computer. Although it gives a quick summary of the run, it is not corrected for worm number or protein concentration.

- Assay configuration. The XF assay template is summarized here. The protocol, groups, background wells and all other parameters that were filled in are collected in this overview tab.

- Levels. This is one of the most important tabs, as it provides an overview of the oxygen concentration, pH and temperature per well and per time point.

- Rate data. This is the tab with the actual measurements of OCR and ECAR per well and time point. For C. elegans, only the OCR column is of importance. Wells 1–12 are at the first row of the plate (A1–A12), wells 13–24 are on the second row (B1–B12) and so on. Please note that as wells 1, 12, 85 and 96 are empty, these values should all be at zero.

- Rate data (Plate view). This tab also shows the actual measure ments of OCR and ECAR, but this time in a plate view, in which respiration rates match the appropriate wells.

- Time events. This gives a written overview of all the steps in the Seahorse protocol.

- Calibration. This gives an overview of the calibration parameters.

The ‘Rate data’ tab is needed to do the actual calculations. The OCR values per well and time point can be divided by the number of worms in a specific well. Please note that the OCR and ECAR values are typically not stable until the second or third measurement cycle (or loop); thus, it is strongly recommended that the first basal measurement be excluded from analysis in order to establish a consistent baseline. Take the average of all the basal measurements per well for a good estimation of the OCR.

Experimental design Overview of parameter optimization. There might be differences between C. elegans strains with respect to size and respiration rates. Therefore, it is essential to first optimize some important parameters for the strains of interest. These parameters include, but are not limited to, the timing for commands in the XF respirometer

180

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 180 experimental template (i.e., “Measuring’, ‘Mixing’ and ‘Waiting time’), the number of worms per well, the stability of the OCR over time and the concentrations of compounds to be injected. By taking into account the parameters and optimiza tion steps, one can adjust the protocol to each individual worm strain and also adapt the protocol to other organisms. The step-by-step PROCEDURE can easily be used to do optimization runs when parameters in specific steps are adjusted, as described in the following sections. Table 3 shows the suggested start 7 conditions, including optimal drug concentrations and timing, the XF template and the number of worms, when working with the N2 strain. This set of parameters also works for worms treated with RNAi, the mev-1(kn1) mutant and several other mutants.

Optimization of assay timing. Optimizing the timing of the XF assay commands ‘Mixing’, ‘Measuring’ and ‘Waiting’ requires interpretation of two aspects of the OCR: the course of the changing oxygen concentration and the stability of the OCR over time. On the basis of our experience, we advise one to start with the fol lowing protocol: (1) calibrate, (2) equilibrate, (3) loop x times, (3a) mix for 2 min, (3b) measure for 3 min, (4) loop end and (5) program end. Set x in loops to 20, pipette 10–25 worms per well (Optimization number of worms),

181

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 181 start the run (PROCEDURE) and look for the factors mentioned above (the course of the changing oxygen concentrations and the stability of the OCR over time).

When investigating the course of changing oxygen concentrations, it is essential to determine whether the parameters within a loop (measuring, mixing and, when applicable, waiting time) are optimal. During the measurement step, oxygen is consumed in the transient microchamber, and the oxygen concentration drops. During mixing, the oxygen concentration should go back to baseline level (Fig.

8a). When the O2 concentration fails to reach the baseline after each mixing cycle, this often indicates that the mix ing time is too short. Alternatively, if the oxygen

Figure 8 | Detailed tracing of oxygen levels is required to identify experimental issues. (a) The oxygen concentration at the start of each loop (measurement cycle) should be the same—i.e., it should touch the imaginary baseline (red dotted line). The O2 concentration is estimated nine times per loop, when measuring for 2 min. These nine measurements (ticks) together are used to calculate the OCR per loop. After each measurement cycle, mixing takes place (as indicated by the gray

area) to restore the initial O2 concentration. (b) The graph shows data from a single well, in which the oxygen concentration does not return back to baseline after a loop because of an excess of worms. Moreover, in the last three loops, the oxygen concentration drops below 0, which causes an underestimation of the OCR. Loop: 2 min of mixing; 3 min of measuring. (c) The OCR over time when a protocol with or without waiting time is used (both with 2 min of mixing and 2 min of measuring). Clearly, introducing a waiting step decreases the fluctuation over time. Moreover, when applying linear regression to the OCR data with a waiting step (red dotted line), the slope does not significantly differ from 0, indicating that the OCR is stable over time (single-well data). (d) Data from a single well that shows clear fluctuation in the steepness of the slopes per loop. Loop: 2 min of mixing; 3 min of measuring. The variation in these slopes is probably caused by worms not always being in the transient microchamber.

182

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 182 concentration drops to zero during the measurement phase (Fig. 8b), this may cause an underestimation of the OCR, and the measuring time or the number of worms per well should be decreased. The raw data concerning oxygen concentration per time unit can be found in the ‘Levels’ tab of the generated Excel result sheet. Add a filter to the column ‘Well’ in order to follow the changes in oxygen concentration over time per specific well.

The stability of the OCR over time can be studied by looking at the OCR estimations over the duration of the experiment. As no compounds are injected during the optimization runs, basal respiration should remain stable over time (at least up to 60 min, Fig. 8c, red dotted line). Small fluctuations in OCR typically occur across the different time points, usually without any overall upward or downward trend, but more drastic variation indicates that something is not optimal (Fig. 8c, black line). Outlier time points may be attributed to worms that occasionally stick to the probe during mixing, or remain outside the transient microchamber during measuring (Fig. 8d). To reduce this possibility, one can introduce the command ‘Waiting time’ between the mixing and measuring time, allowing the worms to sediment after 7 mixing and ensuring that all worms are in the transient microchamber during the measurements. In our experience, a ‘Waiting time’ of 30 s is optimal for N2 worms.

Optimization of worm number. When the values for the commands ‘Measuring’, ‘Mixing’ and ‘Waiting’ are optimized, the influence of the number of worms per well can be studied. Using too many worms will exhaust the oxygen pool, thereby contributing to an underestimation of the OCR. To establish the appropriate number of worms per well, load a 96-well plate with different numbers of worms—e.g. ~5, 10, 15, 20, 25 and 30 worms per well. Start an XF assay with the optimized XF template (measuring, mixing and waiting time) and loop approximately ten times. Calculate the OCR per worm for each well (average over the last nine loops), and plot this against the number of worms. For N2 worms, adult day 1 (D1), a clear regression with a good fit can be generated Fig.( 9a), although two nonstochastic subpopulations can be distinguished (Fig. 9b). With segmented regression, one can distinguish a clear cutoff ~23 worms per well, implying that the OCR is reliable up to 23 worms, at least for N2 worms at D1 (Fig. 9c,d). The optimal number does change with age—e.g., for L4s, up to 50 worms can be used (Supplementary Fig. 4).

Optimization of compound concentrations. Titration assays for compounds of interest should be performed before XF assays, to ensure the optimal working concentration. When using new compounds, one can often find commonly used concentrations of the drug of interest in cells, worms or other organisms. Use this concentration as a starting point, and generate a dose–response curve (half- logarithmic scale concentrations) around this value, keeping in mind that worms often require higher doses. Program the machine to loop five times (with the just-

183

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 183 Figure 9 | The OCR correlates with the number of worms. (a) Linear regression of the OCR as a dependent variable of the number of worms per well, n = 48 wells. Although the R2 value is relatively high, the fit is not optimal, as shown in b. (b) The residual plot. There is no stochastic distribution, but a clear biased distribution toward two populations (red dotted circles). (c) When excluding the negative residual population, the fit of the regression line is much better. d( ) More than 25 worms per well (D1) renders OCR measurement inaccurate. In a and c, the gray area shows the confidence interval (95%) and the dotted line shows the linear regression. Bars represent mean ± s.e.m.

optimized numbers for mixing, waiting and measuring and worm number), inject the compound and loop another 15 times to see its effect on respiration. It is strongly suggested that the specific drug concentration that maximizes the effect at the lowest concentration possible be chosen. By looping 15 times, the kinetics of the compound effect can be determined. As previously mentioned, C. elegans has a cuticle that shows clear impermeability toward non–water soluble compounds. Consequently, if a compound is injected and no clear respiratory response is observed, it does not exclude involvement of the targeted cellular process in C. elegans respiration; it may just be a matter of uptake efficiency. Here, the use of bus mutants, which have increased cuticle permeability68, may help distinguish between these two possibilities. In addition, actual uptake efficiency can be determined with HPLC or liquid chromatography–tandem mass spectrometry.

We have already shown that the optimal concentrations of FCCP and sodium azide in N2 worms are 10 μM and 40 mM, respectively (Fig. 10a,b). Other compounds, such

184

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 184 as rotenone and antimycin A, that are typically used for XF experiments with cultured cells appear to be highly inefficient in our Experimental Setup Supplementary( Fig. 1). Possible interference of sequential treatments should be carefully assessed when new compound combinations are used. For instance, it was reported that sodium azide is less effective when injected in sequence with FCCP28, although we were unable to reproduce this effect (Fig. 10c). Finally, we have tested the effect of glucose pretreatments on the OCR of C. elegans. In multiple model organisms, increasing the glucose content causes a shift from oxidative phosphorylation toward glycolysis, resulting in a decreased OCR42,69. With the XF96 respirometer, we could also detect a decreased OCR when worms were fed with 1 or 2% (wt/vol) glucose starting from the L1 stage until D4 (Fig. 10d). Clearly, prefeeding with compounds can affect cellular respiration and therefore provide researchers with a long-term intervention method whose effects can be analyzed with XF respirometers. 7

Figure 10 | The effects of specific compounds on the OCR in C. elegans. (a) Titration of FCCP and its effect on the OCR. 7.5–12.5 μM FCCP causes a significant increase in the OCR. One-way ANOVA (P < 0.001) with post hoc Tukey test, n = 8–16 wells. (b) Titration of sodium azide and its effect on the OCR. 5–80 mM sodium azide causes a significant decrease in the OCR. One-way ANOVA (P < 0.001) with post hoc Tukey test. a indicates that 40–80 mM causes a significant decrease in OCR compared with 5 mM sodium azide treatment, n = 14–16 wells. (c) There is no significant difference in the effect of sodium azide between when the compound was injected alone and when it was injected in sequence with FCCP; Student’s t-test, n = 15–16 wells. (d) Feeding worms from L1 on with different concentrations of glucose (in the agar medium) causes a significant decrease in basal respiration at D4. Bars are mean ± s.e.m. One-way ANOVA P( < 0.001) with post hoc Tukey test, n = 8 wells. ***P < 0.001. NS, nonsignificant.

185

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 185 Controls and replicates. Experimental control groups are helpful for determining whether the experiment worked and to assess variability from assay to assay. Typically, including an established mutant/RNAi condition with decreased respiration helps to determine whether the experimental setup is sensitive enough. Moreover, such a control may facilitate the relative comparison between different experiments, although we discourage directly comparing conditions between different plates or experiments. Established controls are gas-1(fc21)/CW152, mev-1(kn1)/TK22 and mrps-5 RNAi. ~6–8 replicates per group (per individual experiment) are sufficient to estimate the OCR with relatively small well-to-well intra-assay variability. It is strongly recom mended that experimental setups be repeated at least three times, so one does not have to rely on one experiment with 6–8 technical replicates when drawing conclusions.

MATERIALS

REAGENTS - Carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP; e.g., Abcam, cat. no. ab120081) ! CAUTION FCCP can be acutely toxic even at low doses. Personal protective equipment should be worn at all times while handling this reagent; wear gloves and protective clothing.

- Calibration solvent, pH 7.4 (Seahorse Bioscience, cat. no. 100840-000)

- Sodium azide (e.g., Sigma-Aldrich, cat. no. S2002) ! CAUTION The electron transport chain (ETC) inhibitor sodium azide can be acutely toxic even at low doses. Personal protective equipment should be worn at all times while handling this reagent; wear gloves and protective clothing. Sodium azide requires extra caution, as it changes rapidly into a toxic gas when mixed with water or acids.

- DMSO (e.g., Thomas Scientific, cat. no. C987Y85)

- KH2PO4 (e.g., Merck Millipore, cat. no. 1048731000)

- Na2HPO4 (e.g., Acros Organics, cat. no. 424380010)

- NaCl (e.g., Merck Millipore, cat. no. 1064041000)

- KCl (e.g., Sigma-Aldrich, cat. no. P9333-500)

- MgSO4 (e.g., Fisher Chemicals, cat. no. M1000/60)

186

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 186 - dH2O

- Triton X-100 (e.g., Thermo Fisher Scientific, cat. no. 215680010)

- C. elegans N2 worms (Caenorhabditis Genetics Center: C. elegans wild isolate) CRITICAL N2 worms are used in the described protocol, but the protocol can be used for any other strain or for N2 worms treated with RNAi from the Ahringer library70,71 (Source BioScience)

- C. elegans mev-1(kn1) (Caenorhabditis Genetics Center: cat. no. TK22)

EQUIPMENT - Bioscience Seahorse XF96 Flux Analyzer (or XFe96)

- Heratherm incubator (Thermo Fisher Scientific, model no. 50125882 or similar) - XFe96 extracellular flux assay kits (Seahorse Bioscience, cat. no. 102416-100) 7 - XF96 Cell Culture Microplates (Seahorse Bioscience, cat. no. 101085-004)

- Centrifuge (e.g., Thermo Fisher Scientific, model no. SL40R)

- Tubes, 15 ml (e.g., Cellstar, cat. no. 188271)

- Tubes, 1.5 ml (e.g., Greiner Bio-One, cat. no. 616201)

- Serological pipettes, 10–25 ml (e.g., Sarstedt, cat. nos. 86.1254.001 and 86.1685.001)

- Multichannel pipette, 20–200 μl (e.g., Eppendorf, cat. no. N21475D)

- Pipettes, 20, 200 μl

- Freezer (−20 °C)

- Vortex

- Dissection microscope

- (Optional) Parafilm CRITICAL For equipment and protocols related to C. elegans maintenance, kindly refer to Stiernagle et al.72.

REAGENT SETUP

M9 buffer (1×) Add 3 g of KH2PO4, 6 g of Na2HPO4 and 5 g of NaCl to 1 L of H2O.

Sterilize the buffer by autoclaving, and add 1 ml of 1 M MgSO4 after cooling. The

187

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 187 buffer can be stored at room temperature for several months.

FCCP stock (10 mM and 100 μM) Mix 2.54 mg of FCCP in 1 ml of DMSO and make 25-μl aliquots; these can be stored at −20 °C for up to 2–4 months. Dilute 25 μl of the FCCP stock (10 mM) in 2,475 μl of M9 buffer (= 100 μM) up to 24 h before use; ~2.5 ml of 100 μM FCCP solution is required per assay (25 μl of the stock in DMSO).

Sodium azide stock (400 mM) Add 26 mg of sodium azide per 1 ml of dH2O. The solution can be stored at room temperature for up to 24 h before use. 2.5 ml of 400 mM of sodium azide solution is required per assay.

1× PBS + 0.05% (vol/vol) Triton X-100 To make PBS, add 8 g of NaCl, 0.2 g of

KCl, 1.44 g of Na2HPO4 and 0.24 g of KH2PO4 (pH 7.4) to 1 l dH2O. Store PBS at room temperature or at 4 °C for up to 6 months. Add 0.5 ml of Triton X-100 to 1 l of PBS (0.05%); this solution can be stored at 4 °C for up to 1 month.

PROCEDURE CRITICAL C. elegans age-synchronization should take place in advance depending on the age of interest. Before starting the experimental procedures on day 2 of the protocol, make sure that the worms are at the appropriate age.

Hydration of probes ● TIMING 12–24 h; day 1 1| Open an unused flux package (green lid; XFe96 extracellular flux assay kit). In the package, you will find a utility plate with a green sensor cartridge on top, a lid and two drug-loading guides.

2| Remove the green sensor cartridge, but place it upside down so that the probe surface will not be scratched.

3| Use a multichannel pipette to add 200 μl of the Seahorse Bioscience XF96 calibrant (pH 7.4) solution to each well of the utility plate.

4| Place the sensor cartridge back on the utility plate so that the probes dip into the wells. Place the lid on top and incubate the cartridge overnight at 37 °C without CO2.

CRITICAL STEP One should be careful not to disturb the sensor probes’ surface.

CRITICAL STEP Cartridges cannot be reused. When performing multiple runs, make sure to hydrate multiple cartridges.

188

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 188 PAUSE POINT Hydration can take as little as 4 h and as long as 72 h (when hydrating for >24 h, seal the plate with Parafilm and store it at 4 °C. Rewarm to 37 °C before use).

Preparation of the XF96 respirometer: temperature control and background correction ● TIMING 5 min; day 2 5| Launch the XF software, and when prompted, select the ‘Standard’ software settings. Either create a new account or proceed by clicking on ‘Seahorse guest’.

CRITICAL STEP Always use the standard software package and not the software associated with one of the so-called ‘stress kits’.

6| Turn the heater and temperature control off. Click on the instrument icon in the lower-right corner and subsequently select ‘Administration’ > ‘Temperature’ > ‘Heater and temperature control’. The buttons should be gray when they are turned off. 7 CRITICAL STEP This is a critical step for working with C. elegans, as use of the heater is designed for cell culture at 37 °C. The temperature of the machine can be directly followed on its display.

? TROUBLESHOOTING

7| In the same screen (administration), click on ‘Background correction’. Make sure that background correction is on and that the correction wells are set to A1, A12, H1 and H12.

8| Leave the instrumental setup mode by clicking on ‘End instrument set-up mode’.

Preparation of the XF96 respirometer: creating an XF assay ● TIMING 10–20 min; day 2 9| Click on the ‘Assay wizard’ icon (next to the instrument icon). In the ‘General’ tab, fill in at least the assay name and results file name, for instance, using the following format: ‘YearMonthDate_Strains/conditions’—e.g., ‘20160515_N2_ L4’.

10| The tabs ‘Cells’, ‘Media’ and ‘Compounds’ can be ignored. However, make sure to record the compounds to be injected, their concentrations and the port in which they will be loaded.

11| Proceed to the ‘Background correction’ tab and double-check that the

189

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 189 correction is ‘on’ and the corner wells are selected (A1, A12, H1 and H12).

12| (Optional) Wells can be labeled in the ‘Groups and labels’ tab. Select the wells with technical replicates and give them a name and color. During the run, you can see the average OCR of the groups involved. When labeling is not done before the assay, it can still be done afterward.

CRITICAL STEP The OCR values that can be followed in real time are not corrected for the number of worms and are therefore not directly comparable.

13| Go to the ‘Protocol’ tab. By clicking on the commands, you can add instructions to the current protocol. For the commands ‘Measure’, ‘Mix’ and ‘Wait’, one must also fill in the duration. Always start with (1) ‘Calibrate’ and (2) ‘Equilibrate’. For N2 worms, use the parameters shown in Table 3. For advice on designing assay protocols for different strains, see Experimental design.

CRITICAL STEP When not injecting, or when injecting multiple compounds, make sure to exclude or include these commands in your protocol, respectively.

14| When the appropriate protocol is generated, go to the last tab, ‘End’, and click the button ‘End Assay Wizard’. A screen with a green start button will appear under the ‘Run’ tab.

CRITICAL STEP Do not click the ‘Start’ button yet. To avoid unnecessary heating of the machine, you should postpone starting the machine until <20 min before the worm assay plate will be ready.

Loading of compounds to be injected (optional) ● TIMING 30 min; day 2 15| Take the hydrated utility plate from Step 4 with the sensor cartridge from the incubator.

16| If desired, thaw one 25-μl aliquot of FCCP (10 mM) and dilute it 100-fold in M9 buffer (2,475 μl), as described in the Reagent Setup (Table 4).

17| Slowly pipette 22 μl of diluted FCCP (100 μM) into each ‘port A’ on the cartridge (Table 4), including those belonging to the background correction wells. During pipetting, insert the pipette tips into the holes and dispense the content. (Optional) The loading guide can be used to ensure pipetting into the correct port.

CRITICAL STEP A multichannel pipette can be used for this purpose. Make sure to avoid allowing the liquid to adhere to the sides of the port and avoid creating bubbles, as compounds are injected into the XF96 in a pneumatic manner. In addition,

190

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 190 be aware that the ports have a hole at the bottom for injection.

CRITICAL STEP As the compounds are injected using a pneumatic mechanism, it is essential to add the same volume of liquid to all injection ports—i.e., when injecting FCCP into port A, all of the other ports should contain the same amount of a compound or a vehicle.

CRITICAL STEP Do not use automated pipettes to load the compounds, as this will increase leakage through the bottom of the ports.

18| Load 24 μl of sodium azide (400 mM) in the same manner as described for FCCP into each ‘port B’ of the green sensor cartridge. 7 CRITICAL STEP As FCCP is injected first, the volume in the well is not 200 μl anymore, but it is 220 μl. Therefore, the injection volume of sodium azide is higher to ensure a correct final concentration.

19| Visually inspect the injection ports A and B for even loading; the liquid should be down at the bottom of the port.

20| Place the lid on top of the cartridge. Carefully place the sensor cartridge (still on top of the utility plate) back into the non-CO2 incubator at 37 °C, and keep it there until calibration is started.

CRITICAL STEP Keep the plate for at least 10 min at 37 °C to ensure that the compounds are warmed up. In this way, calibration will take place at ~37 °C, which is the optimal calibration temperature for XF respirometers (the heater is still off).

191

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 191 Preparation/loading of samples and starting of the XF assay ● TIMING 2–3 h; day 2 21| Make note of the developmental stages seen on the worm agar plates.

CRITICAL STEP Strict age synchrony is required for reliable estimations of the OCR. Make sure that worms are synchronized (by hypochlorite treatment, or egg laying and/or FUdR) and that they are of the age of interest on the day of the XF assay.

22| Use sufficient (2–4 ml per 9-cm plate) M9 buffer to rinse the worms off the plate with gentle agitation, and collect the worms for each sample in a 15-ml tube.

23| Spin the worms down in a centrifuge (20 °C, ~500g for 2 min).

24| Remove the supernatant and add fresh M9 buffer. Repeat Steps 23 and 24 two more times.

25| Because there may be eggs or even small larvae present in your samples when you are testing aged worms (even when using FUdR), add M9 to the tube, mix gently and allow the worms to settle for a few minutes by gravity. When the worms have sunken into a loose pellet, aspirate the supernatant, which usually contains smaller animals and eggs.

26| Repeat Step 25.

27| Resuspend the worms in 1.5–3 ml of M9, depending on the density and the size of the worm pellet.

28| Mix the worm suspension carefully and pipette two 3-μl drops onto an empty Petri dish, then count the number of worms per drop using a microscope.

CRITICAL STEP Wash the pipette tip with PBS + 0.05% (vol/vol) Triton X-100 before you pipette the worms, or use low-retention tips to avoid worms sticking to the tip. Otherwise, you may use wide-bore tips or tips with the ends cut off using a razor blade.

29| Calculate the volume of sample that should be loaded to acquire 10–20 worms per well. Subtract this volume from 200 μl (the maximal volume per well); this is the volume of M9 that will be added per well.

CRITICAL STEP The volume of M9 that should be loaded into the wells needs to be calculated for each sample separately. Make a clear pipetting scheme and fill in the scheme shown in Supplementary Figure 3. In addition, include control strains (e.g., mev-1 mutant, gas-1 mutant and mrps-5 RNAi) to ensure correct interpretation

192

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 192 of the results later on.

30| Before loading the worms, click ‘Start’ to start the actual XF assay. Be sure to make the appropriate entries in the ‘Save directory’ and ‘Save name’ commands before you click ‘Start’ again.

31| Follow the prompts on the screen for sensor cartridge calibration. When the loading door opens, take the sensor cartridge from the non-CO2 incubator (Step 20) and place it on the tray.

CRITICAL STEP The notch of the plate should point toward the front of the machine.

CRITICAL STEP Make sure to remove the lid before inserting the plate into the machine. ? TROUBLESHOOTING 7 CRITICAL STEP Wash the pipette tip with PBS + 0.05% (vol/vol) Triton X-100 before pipetting the worms, or use low-retention tips to avoid worms sticking to the tip. Otherwise, you may use wide-bore tips or tips with the ends cut off using a razor blade.

33| Pipette 200 μl of M9 into the background correction wells: A1, A12, H1 and H12.

34| By visual inspection with a dissection microscope, inspect all wells and make sure that the wells A1, A12, H1 and H12 are empty and that the experimental wells contain ~10–25 worms (see Table 3 for appropriate worm numbers per well).

35| Wait at least 2 min from when the worms are pipetted until the plate is loaded, to allow the worms to settle by gravity to the bottom of the wells.

36| By this point, the calibration should be completed. Note: when calibration is completed before the samples are ready, the instrument stays closed and waits until samples are ready to load. Open the loading door by clicking ‘OK’ on the screen; this will release the utility plate containing calibration buffer from the XF respirometer, while the cartridge stays inside the machine. Replace the calibrant plate with the worm plate.

CRITICAL STEP The green sensor cartridge is not ejected; it stays in the machine. Only the utility plates are swopped. Do not forget to remove the lid of the worm plate.

193

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 193 37| Close the door by clicking on ‘Continue’; the equilibrate step of the protocol will now be executed.

38| Once the run is completed, follow the prompts to remove the cell plate and the sensor cartridge from the XF96 respirometer.

39| Keep the worm plate and use a dissection microscope to count the number of worms per well. Make note of the numbers on the scheme shown in Supplementary Figure 3. While inspecting the number of worms, also make note of any anomalies observed in the wells (e.g., larvae, embryos and bacteria), as this probably influenced the OCR measurements. (Optional) Capture images of each well using a microscope, and count the number of worms per well at a later date.

CRITICAL STEP When sodium azide is injected during the XF assay, worms are paralyzed, which makes counting easier. Without injection of sodium azide, it is suggested that the plate be cooled down before counting the worms—as this also immobilizes the worms—or, alternatively, you can add sodium azide to the wells manually.

? TROUBLESHOOTING PAUSE POINT At this point the experimental part of the PROCEDURE is completed; the postexperimental analysis can be performed at any moment from here on.

Postexperimental analysis ● TIMING 1–2 h 40| Open the Excel spreadsheet that is automatically exported to the directory selected by the user.

CRITICAL STEP When you are performing the analysis on a computer without XF software, copy the important values in the ‘Rate data’ tab to another Excel file to avoid error messages.

? TROUBLESHOOTING 41| Copy the OCR values per well (with all separate measurement points) to a new Excel tab in such a way that the OCR values of a single well are collected below each other (in a column).

42| Divide all OCR values for each well at each time point by the number of worms in that well. (Optional) Determine the protein content with a BCA assay in exactly 1,000 worms to estimate the protein concentration per nematode.

194

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 194 Subsequently divide the OCR per worm by this number.

CRITICAL STEP When nematode strains differ in size, the optional protein determination step is recommended, although we suggest reporting both OCR per worm and OCR per milligram of protein.

43| Estimate the basal OCR for each well by averaging the OCR per worm of measurement points (loops) 2–5.

CRITICAL STEP The basal OCR values are typically not stable until the second measurement loop. In addition, the first measurement after injection is typically not stable. Exclude these measurement points when calculating average OCRs.

CRITICAL STEP The measurement points that should be used for calculations of the OCR are dependent on the selected protocol. Our calculations are based on our proposed protocol, as shown in Table 3.

44| To estimate the FCCP-induced OCR per worm, find a stable plateau phase of three points within measurements 7–14, and average these numbers. 7

? TROUBLESHOOTING 45| To calculate non-mitochondrial respiration per worm (sodium azide–induced), take the average of measurements 17 and 18.

? TROUBLESHOOTING 46| Use the values obtained in Steps 43–45 to calculate mitochondrial respiration for each well by subtracting the average sodium azide–induced OCR from the average basal OCR. To calculate the maximal respiratory capacity, subtract the sodium azide–induced OCR from the FCCP-induced average OCR. To calculate spare respiratory capacity, subtract the average of the basal OCR from the FCCP- induced average OCR. Non-mitochondrial respiration is defined as the sodium azide–induced OCR.

? TROUBLESHOOTING

? TROUBLESHOOTING Troubleshooting advice can be found in Table 5.

● TIMING

Age-synchronization and aging of worms: ~2–20 d.

195

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 195 196

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 196 Steps 1–4, hydration of probes: 12–24 h (day 1)

Steps 5–8, preparation of the XF96 respirometer: temperature control and background correction: 5 min (day 2)

Steps 9–14, preparation of the XF96 respirometer: creating a XF assay: 10–20 min (day 2)

Steps 15–20, loading of the compounds to be injected (optional): 30 min (day 2)

Steps 21–39, preparation/loading of samples and starting of the XF assay: 2–3 h (day 2)

Steps 40–46, postexperimental analysis: 1–2 h

ANTICIPATED RESULTS 7

Interpretation and considerations The goal of using the XF96 respirometers for C. elegans is to estimate the OCR, which mirrors cellular respiration of the worm. Clearly, including established mitochondrial mutants such as gas-1 or mev-1 or mrps-5 RNAi in the experimental setup allows the user to interpret whether the assay is sensitive enough to detect differences in respiration. Basal OCR should therefore differ between N2 and mitochondrial mutant worms (Fig. 11a,b). Note that absolute OCR values can differ between assays because of changes in temperature and the use of different sensor cartridges. Therefore, mutants are ideally analyzed side by side on the same plate and on the same day. The injection of FCCP typically increases the OCR two- to three-fold, but the effect is somewhat delayed when compared with experiments performed with cells, which is likely due to a difference in uptake efficiency. The OCR in N2 worms drops ~75–90% when sodium azide is added, which highlights that most of the respiration is caused by mitochondrial function (Fig. 11b). The quality of the different OCR readings can be verified by looking at the raw data— e.g., the variation in different time points and the course of the oxygen concentration, as described in detail in the Optimization section. The variation between different time points should be small (<15%), except for the initial time points after compound injection (particularly for FCCP), and the oxygen concentration should restore itself with each loop (Fig. 8).

To further verify and test the ability of XF96 respirometers to measure oxygen consumption and especially differences in the OCRs, we tested RNAi directed toward

197

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 197 several proteins in the electron transport chain, including cyc-1 (cytochrome c1, complex III), cco-1 (cytochrome c oxidase, complex IV), nuo-1 (NADH:ubiquinone oxidoreductase, complex I), atp-3 (ATP synthase subunit δ, complex V) and F26E4.6 (cytochrome c oxidase subunit VIIc, complex IV; Figs. 1b, 11c,d). Clearly, similar to compounds that inhibit specific electron transport chain proteins Fig.( 1b; Table 2), knocking down proteins associated with oxidative phosphorylation causes a striking decrease in basal respiration. These results underline the strength of XF respirometers in combination with C. elegans, as the effects of genetic interventions on cellular respiration can be assessed side by side in a quick and straightforward manner when following our developed protocol.

One important consideration that should be taken into account is the interpretation of mitochondrial respiration, spare respiratory capacity, maximal respiratory capacity and non-mitochondrial respiration. These terms were once coined to de scribe respiration effects in isolated mitochondria with exogenously controlled energy

Figure 11 | Genetic manipulation of the electron transport chain lowers the OCR in C. elegans. (a) Basal respiration is significantly lower in the mev-1 strain compared with N2 worms. Student’s t-test (P < 0.001), n = 8 wells. (b) mev-1 mutants have lower basal respiration, lower maximal respiration, lower spare capacity, but comparable non-mitochondrial respiration (n = 8 wells). Dashed red lines indicate injections of compounds. (c) Basal respiration is significantly lower when proteins in the electron transport chain are knocked down. One-way ANOVA (P < 0.001) with post hoc Tukey test, n = 6–8 wells. (d) A knockdown of atp-3 in complex V causes a significant decrease in basal respiration. Bars are mean ± s.e.m. Student’s t-test (P = 0.032), n = 8 wells. Ctrl, control. *P < 0.05; ***P < 0.001.

198

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 198 sources. It is likely that results from using the more complex intact animal should be interpreted differently than those from isolated mitochondria in a defined medium. The question that directly arises is whether treatment with FCCP and sodium azide indeed accurately reflects these different aspects of cellular respiration. This argues caution against overinterpretation of the OCR values in terms of selective association with aspects such as maximal respiratory capacity and would argue for relying on basal respiration only. Clearly, basal respiration provides the user with the most readily interpretable number from a XF96 experiment.

Scaling down: from an XF96 to an XFp respirometer The possibility of assessing respiration in 96 parallel wells makes the XF96 respirometer attractive for moderately large screens and the comparison of many conditions at the same time. However, when, for example, only two conditions need to be compared, 8–12 wells may be sufficient. In this specific case, it is not efficient to fill only a small fraction of the 96 wells, and it is not advisable to reuse plates. Here, the recently launched XFp respirometers provide an alternative. The 7 XFp respirometer works with eight-well plates (including two wells for background correction) in which the separate wells are the same size as the wells of the XF96 plates. The user interface of the XFp respirometer is intuitive, the machine is easy to handle and, because of the limited number of wells, the time for manual transfer of worms to the eight-well plate and the eventual starting of the assay is relatively short. Using the XFp, which was operated in a cold room so that the operating temperature was stable at 20 °C, we have confirmed that knockdown of mrps-5 reduces OCR in C. elegans (Fig. 12a), similar to the effects we had observed earlier using the XF96 respirometer26. Furthermore, the optimal FCCP concentrations are similar to those used in XF96 respirometers, and the number of worms correlates well with the OCR (Fig. 12b). The main limitation of the XFp respirometer is, however, the number of assay wells. Sample-to-sample variation in the XFp respirometer is similar to that observed in XF96 respirometers, but the limited number of wells (six experimental wells) can preclude sufficient technical replicates when analyzing mobile samples such as worms. As a consequence, although massive differences in respiration are easily picked up using the XFp respirometer (Fig. 12a), small changes in OCR might be missed. As such, the XF96 is more practical than the XFp respirometer for measuring worms, and these benefits outweigh the difference in investment that must be made when purchasing the XF96 respirometer and its consumables. As a final consideration, the same optimization steps can be used to measure respiration with the 24-well respirometer (XF24), which has larger wells that require more worms per well73.

199

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 199 Figure 12 | XFp respirometers provide an alternative to XF96 respirometers. Data acquired in XFp respirometers. (a) Nematodes treated with mrps5 RNAi show significant lower respiration than worms treated with control RNAi. Student’s t-test (P < 0.0001). This is comparable to what was previously described in XF96 respirometers26. Bars are mean ± s.e.m. (b) Linear regression of the OCR as a dependent variable of the number of worms per well. Data from two separate runs are combined (n = 10 wells). Gray area shows the confidence interval (95%) and the dotted line shows the linear regression.***P < 0.001.

AUTHOR CONTRIBUTIONS M.K. performed most of the experiments, optimized the final experimental pipeline and wrote the manuscript together with R.H.H., and with contributions from all authors. H.M. and R.K performed experiments using the XFp/XF96 respirometers. L.M., R.H.H., B.M.D. and J.A. pioneered the use of XF respirometers with C. elegans and developed the initial protocols. M.K., H.M., E.A.A.N., B.M.D. and R.H.H. have been extensively involved in discussions and interpretations of results.

200

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 200 REFERENCES

1. Perry, C.G.R., Kane, D.A., Lanza, I.R. & Neufer, behaviour and aging in the nematode P.D. Methods for assessing mitochondrial function Caenorhabditis elegans. EMBO J. 18, 1783–1792 in diabetes. Diabetes 62, 1041–1053 (2013). (1999). 2. Hatefi, Y. The mitochondrial electron transport 17. Kamo, N., Muratsugu, M., Hongoh, R. & and oxidative phosphorylation system. Annu. Rev. Kobatake, Y. Membrane potential of mitochondria Biochem. 54, 1015–1069 (1985). measured with an electrode sensitive to tetraphenyl 3. Brand, M.D. & Nicholls, D.G. Assessing phosphonium and relationship between proton mitochondrial dysfunction in cells.Biochem. J. electrochemical potential and phosphorylation 435, 297–312 (2011). potential in steady state. J. Membr. Biol. 49, 105– 121 (1979). 4. Goldenthal, M.J. & Marín-García, J. Mitochondrial signaling pathways: a receiver/integrator organelle. 18. Drew, B. & Leeuwenburgh, C. Method Mol. Cell Biochem. 262, 1–16 (2004). for measuring ATP production in isolated mitochondria: ATP production in brain and liver 5. Galluzzi, L., Kepp, O. & Kroemer, G. Mitochondria: mitochondria of Fischer-344 rats with age and master regulators of danger signalling. Nat. Rev. caloric restriction. Am. J. Physiol. Regul. Integr. Mol. Cell Biol. 13, 780–788 (2012). Comp. Physiol. 285, R1259–R1267 (2003). 6. Rizzuto, R., De Stefani, D., Raffaello, A. & 19. Pan, X. et al. The physiological role of Mammucari, C. Mitochondria as sensors and mitochondrial calcium revealed by mice lacking regulators of calcium signalling. Nat. Rev. Mol. the mitochondrial calcium uniporter. Nat. Cell Cell Biol. 13, 566–578 (2012). Biol. 15, 1464–1472 (2013). 7. Fernyhough, P., Roy Chowdhury, S.K. & Schmidt, 20. Chance, B. & Williams, G.R. Respiratory enzymes 7 R.E. Mitochondrial stress and the pathogenesis in oxidative phosphorylation. I. Kinetics of oxygen of diabetic neuropathy. Expert Rev. Endocrinol. utilization. J. Biol. Chem. 217, 383–393 (1955). Metab. 5, 39–49 (2010). 21. Wu, M. et al. Multiparameter metabolic 8. Ferreira, I.L., Resende, R., Ferreiro, E., Rego, analysis reveals a close link between attenuated A.C. & Pereira, C.F. Multiple defects in energy mitochondrial bioenergetics function and enhanced metabolism in Alzheimer’s disease. Curr. Drug glycolysis dependency in human tumor cells. Am. Targets 11, 1193–1206 (2010). J. Physiol. Cell Physiol. 292, C125–C136 (2007). 9. Jarrett, S.F., Lewin, A.S. & Boulton, M.E. The 22. Gerencser, A.A. et al. Quantitative microplate- importance of mitochondria in age-related and based respirometry with correction for oxygen inherited eye disorders. Opthalmic Res. 44, 179– diffusion. Anal. Chem. 81, 6868–6878 (2009). 190 (2010). 23. Will, Y., Hynes, J., Ogurtsov, V.I. & Papkovsky, 10. Kawamata, H. & Manfredi, G. Mitochondrial D.B. Analysis of mitochondrial function using dysfunction and intracellular calcium dysregulation phosphorescent oxygen-sensitive probes. Nat. in ALS. Mech. Ageing Dev. 131, 517–526 (2010). Protoc. 1, 2563–2572 (2006). 11. Ren, J., Pulakat, L., Whaley-Connell, A. & Sowers, 24. Nadanaciva, S. et al. Assessment of drug-induced J.R. Mitochondrial biogenesis in the metabolic mitochondrial dysfunction via altered cellular syndrome and cardiovascular disease. J. Mol. Med. respiration and acidification measured in 96-well 88, 993–1001 (2010). platform. J. Bioenerg. Biomembr. 44, 421–437 12. Rosenstock, T.R., Duarte, A.I. & Rego, A.C. (2012). Mitochondrial-associated metabolic changes and 25. Yamamoto, H. et al. NCoR1 is a conserved neurodegeneration in Huntington’s disease – from physiological modulator of muscle mass and clinical features to the bench. Curr. Drug Targets oxidative function. Cell 147, 827–839 (2011). 11, 1218–1236 (2010). 26. Houtkooper, R.H. et al. Mitonuclear protein 13. Vafai, S.B. & Mootha, V.K. Mitochondrial imbalance as a conserved longevity mechanism. disorders as windows into an ancient organelle. Nature 497, 451–457 (2013). Nature 491, 374–383 (2012). 27. Mouchiroud, L. et al. The NAD+/sirtuin pathway 14. Andreux, P.A., Houtkooper, R.H. & Auwerx, modulates longevity through activation of J. Pharmacological approaches to restore mitochondrial UPR and FOXO signaling. Cell 154, mitochondrial function. Nat. Rev. Drug Discov. 12, 430–441 (2013). 465–483 (2013). 28. Luz, A.L. et al. Mitochondrial morphology and 15. Andreux, P.A. et al. A method to identify fundamental parameters of the mitochondrial and validate mitochondrial modulators using respiratory chain are altered in Caenorhabditis mammalian cells and the worm C. elegans. Sci. elegans strains deficient in mitochondrial Rep. 4, 05285 (2014). dynamics and homeostasis processes. PLoS One 16. Felkai, S. et al. CLK-1 controls respiration, 10, e0130940 (2015).

201

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 201 29. Lee, S.S. et al. A systematic RNAi screen identifies Caenorhabditis elegans. Methods Cell Biol. 48, a critical role for mitochondria in C. elegans 187–204 (1995). longevity. Nat. Genet. 33, 40–48 (2002). 45. Burns, A.R. et al. A predictive model for drug 30. Fire, A.Z. Gene silencing by double-stranded bioaccumulation and bioactivity in Caenorhabditis RNA. Cell Death Differ. 14, 1998–2012 (2007). elegans. Nat. Chem. Biol. 6, 549–557 (2010). 31. Grishok, A., Tabara, H. & Mello, C.C. Genetic 46. Zheng, S.-Q., Ding, A.-J., Li, G.-P., Wu, G.- requirements for inheritance of RNAi in C. S. & Luo, H.-R. Drug absorption efficiency in elegans. Science 287, 2494–2497 (2000). Caenorhabditis elegans delivered by different 32. Dickinson, D.J., Ward, J.D., Reiner, D.J. & methods. PLoS One 8, e56877 (2013). Goldstein, B. Engineering the Caenorhabditis 47. Massie, M.R., Lapoczka, E.M., Boggs, K.D., Stine, elegans genome using Cas9-triggered homologous K.E. & White, G.E. Exposure to the metabolic recombination. Nat. Methods 10, 1028–1034 inhibitor sodium azide induces stress protein (2013). expression and thermotolerance in the nematode 33. Wadsworth, W.G. & Riddle, D.L. Developmental Caenorhabditis elegans. Cell Stress Chaperones 8, regulation of energy metabolism in Caenorhabditis 1–7 (2003). elegans. Dev. Biol. 132, 167–173 (1989). 48. Lanza, I.R. & Nair, K.S. Functional assessment of 34. Li, J. et al. Proteomics analysis of mitochondria isolated mitochondria in vivo. Methods Enzymol. from Caenorhabditis elegans. Proteomics 9, 4539– 457, 349–372 (2009). 4554 (2009). 49. Salabei, J.K., Gibb, A.A. & Hill, B.G. 35. Mullaney, B.C. & Ashrafi, K.C. elegans fat storage Comprehensive measurement of respiratory and metabolic regulation. Biochim. Biophys. Acta activity in permeabilized cells using extracellular 1791, 474–478 (2009). flux analysis. Nat. Protoc. 9, 421–438 (2014). 36. Dancy, B.M., Sedensky, M.M. & Morgan, P.G. 50. Piper, H.M. et al. Development of ischemia- Mitochondrial bioenergetics and disease in induced damage in defined mitochondrial Caenorhabditis elegans. Front Biosci. (Landmark subpopulations. J. Mol. Cell Cardio. 17, 885–896 Ed.) 20, 198–228 (2015). (1985). 37. Hill, B.G., Dranka, B.P., Zou, L., Chatham, J.C. & 51. Picard, M. et al. Mitochondrial functional Darley-Usmar, V.M. Importance of the bioenergetics impairment with aging is exaggerated in isolated reserve capacity in response to cardiomyocyte mitochondria compared to permeabilized stress induced by 4-hydroxynonenal. Biochem. J. myofibers. Aging Cell 9, 1032–1046 (2010). 424, 99–107 (2009). 52. Picard, M. et al. Mitochondrial structure and 38. Dranka, B.P. et al. Assessing bioenergetic function function are disrupted by standard isolation in response to oxidative stress by metabolic methods. PLoS One 6, e18317 (2011). profiling. Free. Radic. Biol. Med. 51, 1621–1635 53. Picard, M., Taivassalo, T., Gouspillou, G. & (2011). Hepple, R.T. Mitochondria: isolation, structure and 39. Sansbury, B.E., Jones, S.P., Riggs, D.W., Darley- function. J. Physiol. 598, 4413–4421 (2011). Usmar, V.M. & Hill, B.G. Bioenergetic function in 54. Kayser, E.B., Morgan, P.G., Hoppel, C.L. & cardiovascular cells: the importance of the reserve Sedensky, M.M. Mitochondrial expression and capacity and its biological regulation. Chem. Biol. function of GAS-1 in Caenorhabditis elegans. J. Interact. 191, 288–295 (2011). Biol. Chem. 276, 20551–20558 (2001). 40. Readnower, R.D., Brainard, R.E., Hill, B.G. & 55. Rivera-Ingraham, G.A., Bickmeyer, U. & Abele, Jones, S.P. Standardized bioenergetics profiling of D. The physiological response of the marine adult mouse cardiomyocytes. Physiol. Genomics platyhelminth Macrostomum ligano to different 44, 1208–1213 (2012). environmental oxygen concentrations. J. Exp. Biol. 41. Wilson, D.F., Vinogradov, S., Lo, L.W. & Huang, L. 216, 2741–2751 (2013). Oxygen dependent quenching of phosphorescence: 56. Steele, S.L., Prykhozhij, S.V. & Berman, J.N. a status report. Adv. Exp. Med. Biol. 388, 101–107 Zebrafish as a model system for mitochondrial (1996). biology and diseases. Transl. Res. 163, 79–98 42. Schulz, T.J. et al. Glucose restriction extends (2014). Caenorhabditis elegans life span by inducing 57. Gibert, Y., McGee, S.L. & Ward, A.C. Metabolic mitochondrial respiration and increasing oxidative profile analysis of zebrafish embryos. J. Vis. Exp. stress. Cell Metab. 6, 280–293 (2007). 14, e4300 (2013). 43. Ehrismann, D. et al. Studies on the activity of 58. Hartman, N. et al. Mitochondrial DNA copy the hypoxia-inducible-factor hydroxylases using number and function decrease with age in the an oxygen consumption assay. Biochem. J. 401, short-lived fishNothobranchius furzeri. Aging Cell 227–234 (2007). 10, 824–831 (2011). 44. Rand, J.B. & Johnson, C.D. Genetic pharmacology: 59. Porta-de-la-Riva, M., Fontrodona, L., Villanueva, interactions between drugs and gene products in A. & Cerón, J. Basic Caenorhabditis elegans

202

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 202 methods: synchronization and observation. J. Vis. reference population. Cell 158, 1415–1430 (2014). Exp. 10, e4019 (2012). 67. Weimer, S. et al. D-glucosamine supplementation 60. Mitchell, D.H., Stiles, J.W., Santelli, J. & extends life span of nematodes and ageing mice. Sanadi, D.R. Synchronous growth and aging Nat. Commun. 5, 3563 (2014). of Caenorhabditis elegans in the presence of 68. Partridge, F.A., Tearle, A.W., Gravato-Nobre, fluorodeoxyuridine.J. Gerontol. 34, 28–36 (1979). M.J., Schafer, W.R. & Hodgkin, J. The C. elegans 61. Davies, S.K., Leroi, A.M. & Bundy, J.G. glycosyltransferase BUS-8 has two distinct and Fluorodeoxyuridine affects the identification essential roles in epidermal morphogenesis. Dev. of metabolic responses to daf-2 status in Biol. 317, 549–559 (2008). Caenorhabditis elegans. Mech. Ageing Dev. 133, 69. MacVicar, T.D. & Lane, J.D. Impaired OMA1- 46–49 (2012). dependent cleavage of OPA1 and reduced DRP1 62. Rooney, J.P. et al. Effects of 5’-fluoro-2- fission activity combine to prevent mitophagy deoxyuridine on mitochondrial biology in in cells that are dependent on oxidative Caenorhabditis elegans. Exp. Gerontol. 56, 69–76 phosphorylation. J. Cell Sci. 127, 2313–2325 (2014). (2014). 63. Gruber, J., Ng, L.F., Poovathingsal, S.K. & 70. Fraser, A. et al. Functional genomic analysis of Halliwell, B. Deceptively simple but simply C. elegans chromosome I by systematic RNA deceptive – Caenorhabditis elegans lifespan interference. Nature 408, 325–330 (2000). studies: consideration for aging and antioxidant 71. Kamath, R.S. et al. Systematic functional analysis effects. FEBS Lett. 583, 3377–3387 (2009). of the C. elegans genome using RNAi. Nature 421, 64. Boyd, W.A. et al. A high-throughput method for 231–237 (2003). assessing chemical toxicity using a Caenorhabditis 72. Stiernagle, T. Maintenance of C. elegans. elegans reproduction assay. Toxicol. Appl. in WormBook (ed. The C. elegans Research Pharmacol. 245, 153–159 (2010). Community) doi:10.1895/wormbook.1.101.1 7 65. Zubovych, I.O., Straud, S. & Roth, M.G. (2006). Mitochondrial dysfunction confers resistance to 73. Dancy, B.M. et al. Glutathione S-transferase multiple drugs in Caenorhabditis elegans. Mol. mediates an ageing response to mitochondrial Biol. Cell 21, 956–969 (2010). dysfunction. Mech. Ageing Dev. 153, 14–21 66. Wu, Y. et al. Multilayered genetic and omics (2015). dissection of mitochondrial activity in a mouse

203

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 203 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 204 Chapter VIII

Increase in number and complexity of modified tryptophan residues during aging in C. elegans

Helen Michels1, Jörg Reinders2, Renée I. Seinstra1, Mandy Koopman1, Ido P. 3 1 2 Kema , Ellen A.A. Nollen , Peter J. Oefner

1Department of Molecular Neurobiology of Aging, European Research Institute for the Biology of Aging, University of Groningen, University Medical Centre Groningen, The Netherlands 2Institute of Functional Genomics, University of Regensburg, Regensburg, Germany 3University of Groningen, University Medical Centre Groningen, Department of Laboratory Medicine, The Netherlands

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 205 Aging became one of the major risk factors for a variety of age-dependent diseases in the last decades. Understanding the processes of aging helps to learn more about the development of those diseases.

Protein homeostasis and protein quality control systems undergo major changes during aging processes. Modifications of proteins are sometimes essential for their functions but might be also a product of damage to the molecule and an accompanying failure of the quality control system.

Tryptophan residues have been shown to be modified in a number of proteins. Whether these modifications are of a functional nature or a damage-associated phenotype still needs to be ascertained. However, the detection of tryptophan modifications was limited to subsets of modifications so far as previous studies mainly focused on one single modification or a single protein of interest as the detection usually demanded concentration of the sample.

Here, we show that, due to the development of mass spectrometry methods, the detection of different tryptophan modifications is possible in a proteomic approach without any concentration or specification. We further show that tryptophan-modified peptides and the complexity of tryptophan modifications accumulate during aging. We further discuss the possibility that the modifications might be causing aging as proteins are less functional and how the activity of the tryptophan degrading enzyme tryptophan 2,3-dioxygenase 2 (TDO-2) might influence this process.

206

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 206 INTRODUCTION Understanding the mechanism of how we age is a central puzzle piece to decode processes that might lead to the development of age-related diseases.

The protein quality control system is of high interest as it undergoes major changes during aging (Lipinski et al., 2010; Gavilán et al., 2012). An age-related decline leads to unrequested changes in the cell that promote the decline of the cell and organism. However, the cell developed mechanisms during evolution that counteract those processes and are slowing down the collapse of the organism. This is e.g. thought to be the case during the formation of aggregates to reduce the amount of unfolded/ misfolded peptides when the quality control system fails, occuring e.g. during the development of neurodegenerative diseases (Chiti and Dobson, 2009). Posttranslational modifications can be of functional nature as they control for the activity of an enzyme (Johnson and Eyers, 2010, Young et al., 2010), however a protein might undergo modifications also when it ages as it is constantly exposed to modifying agents such as e.g. reactive oxygen or reactive nitrogen species (ROS or RNS) (summarized in Levine and Stadtman, 2001, Nuriel et al., 2011). The oxidative stress response might be even mediated through oxidation of cysteine on specific 8 chaperones (summarized in Dahl et al., 2015). For example, the eukaryotic chaperone GET3 usually helps proteins to anchor in membranes in its reduced form but once it gets oxidized at a motif that contains four highly sensitive cysteine residues, it changes its conformation and switches to an ATP-independent chaperone, a holdase. Holdases prevent proteins from aggregation during oxidative stress, when the lack of energy inhibits the activity of other chaperones (Voth et al., 2014). Furthermore, the cell developed mechanisms to repair oxidized residues within proteins, e.g. methionine sulfoxide reductase reverses oxidation events at methionine residues (Stadtman et al., 2005). These modifications of amino acid residues gain especially importance when the protein degradation and renewal is impaired as it is the case in an aged cell. Modification-damaged proteins accumulate and show reduced functionality as it was e.g. shown for the human lens (Hains and Truscott, 2007).

Up to date, there are several hundreds, mainly functional, posttranslational modifications described (Creasy and Cottrell, 2004). First publications on oxidized residues were already published in the 1970s (summarized in Dreyfus and Khan, 1978, Rothstein, 1977). In the 1990s the focus changed more and more to specific amino acids such as methionine (summarized in Stadtman et al., 2005), tyrosine (summarized in Greenacre and Ischiropoulos, 2001) and later tryptophan (summarized in Yamakura and Ikeda, 2006). Tryptophan and its metabolite can be categorized into two different types of modifications (Figure 2): First, they can occur as side chains being attached by deamination of cysteine, lysine and histidine residues in

207

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 207 A B

deamination of cysteine, lysine and histidine modifications residues of tryptophan residues

Figure 2. Types of tryptophan modifications. (A) Tryptophan metabolites form side-chains. (B) The tryptophan residue is modified.

a polypeptide (Garner et al., 1999). Or second, the tryptophan residue itself might be modified, often in an oxidized form (e.g. Møller and Kristensen, 2006, Brégère et al., 2008, types of modified tryptophan residues are summarized in Figure 3). Some of these modifications might be functional and are increasing the activity of specific proteins. This was e.g. suggested for the mitochondrial protein Succinyl- CoA:3-ketoacid coenzyme A transferase (SCOT) that showed increased activity when tryptophan at position 372 was subsidized by 5-hydroxy-6-nitrotryptophan

OXIDATION ; ; hydroxytryptophan 3α-hydroxypyrolloindole 2-oxindolyalanine

DIOXIDATION

N-formylkynurenine

TRP à KYN

tryptophan kynurenine

TRPà 3HK

3-hydroxykynurenine

NITROLYSATION

nitrotryptophan

Figure 3. Possible modifications on tryptophan residues within a protein.

208

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 208 (Rebrin et al., 2007). But it has also been shown that induction of oxidative stress increases the number of tryptophan-modified proteins. It was suggested that this is due to the molecule structure of tryptophan and that tryptophan is prone to react due to its aromatic structure and the electron-richness of the indole ring (Naismith, 2012). However, in vivo studies on protein modifications are limited and publications mainly focused on specific type of modifications or a single protein as the detection methods were restricted and it was necessary to concentrate for the molecule of interest in a sample (Figure 1, summarized in Nuriel et al., 2011).

Here, we show in a preliminary test experiment that we can globally detect different tryptophan modification on different protein peptides without concentrating for a certain protein or modification in the sample. We show that the detectable modified peptides and the complexity of modifications increases with age. We further discuss open questions and how to proceed to test whether changes in tryptophan metabolism - by depleting tdo-2- are involved in the formation of tryptophan modifications during aging.

A B C 8

antibody antibody against against protein X modification

analysis

analysis analysis

Identification Identification global picture of different of different of modified proteins modifications proteins with and type of at protein X modification modifications

Figure 1. Types of assays in research around tryptophan modifications. (A,B) Assays with concentration step. (C) Assay without concentration step.

209

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 209 RESULTS Previous studies showed that proteins might contain modified tryptophan residues that interfere with their functionality as it was shown for proteins of the human lens (Hains and Truscott, 2007) or the mitochondrial protein SCOT (Rebrin et al., 2007). However these studies usually focused on a specific protein or type of modification and the detection was only possible due to concentrating the protein or for certain modifications in a sample.

We wondered if it is possible to detect tryptophan modifications (here: modified tryptophan residues) without concentration step and how a global picture would look like? Are specific peptides prone to be modified and how does the stateof modifications change with age? We applied mass spectrometry-based label-free quantification using a shotgun proteomics approach (SWATH-mass spectrometry (Gillet et al., 2012) on a TripleTOF5600+™ coupled to nano-LC) to quantify both proteins and protein modifications in parallel.

We generated C. elegans protein lysates from animals at different lifetime stages. We separated the proteins and tried to detect peptides with modified tryptophan residues in different subfractions (Figure 4-A). This preliminary test showed that we were

A B protein X 30

L1 L4 D1 D4 D8 25

20 peptides with modified tryptophan residues 15 IMDCPADLIYNEPLLICDWR 10

lysates PASLVFNPTILVCDWPR 5

total number of identified peptides 0 L4 D1 D4 D8

C nitrotryptophan protein X formaldehyde adduct identification of Trp à Kyn peptides with modified dioxidation tryptophan residues peptide with modified tryptophan residues oxidation 70 IMDCPADLIYNEPLLICDWR 60 50 different modifications at tryptophan residues 40 comparison 30 age IMDCPADLIYNEPLLICDWoxR 20 IMDCPADLIYNEPLLICDWdioxR 10

IMDCPADLIYNEPLLICDKynR total number of modifications 0 L4 D1 D4 D8

Figure 4. Detection of modified tryptophan residues during aging (A) Experimental setup (B) absolute number of modified peptides inC.elegans lysates at different ages (C) number of different types of modifications on detected peptides in C. elegans lysates at different ages

210

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 210 able to detect different protein peptides with modified tryptophan residues (Figure 3-B) as well as different types of modifications on the same peptide (Figure 4-C). Furthermore, the number of modified peptides (Figure 4-B) and the complexity of modifications (Figure 4-C) increased with the age of the animals, proposing a higher abundance in the total proteome in older organisms.

We then adjusted our protocol based on the preliminary findings (Figure 5). We included more time points as the accumulation is strongest at younger ages and seems to reach a plateau at higher ages (Figures 4-B and C). We also repeated the detection steps several times to increase the number of detectable modified peptides. And finally we decided to quantify unmodified and modified peptides to learn more about the ratios (see discussion).

Another question was whether a modulation in TDO-2 activity (by RNAi or a knockout mutation) or feeding of additional tryptophan could interfere with the formation of tryptophan modifications by changing the amount of free tryptophan in the cell and whether this might correlate with the lifespan-extending effect by tdo-2 depletion. We therefore added samples from tdo-2 depleted or tdo-2 mutated animals respectively as well as samples from animals that were exposed to tryptophan (Figure 8 5). Samples were collected in several replicates and prepared for analysis however final results are not available yet.

211

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 211 L1 L4 D1 D2 D4 D6 D8

lysates

several detection rounds

Identification and quantification of tryptophan-containing peptides

Identification and quantification of modified tryptophan residues

Ratios modified/ unmodified

Comparison: - age - ctr vs. tdo-2 RNAi - WT vs. tdo-2 (del) - WT vs. WT/Trp fed

Figure 5. Adjustments to the experimental set up and sample choice

212

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 212 DISCUSSION Here, we could show for the first time - in a preliminary experiment - that it was possible to detect modified tryptophan residues without any concentration step. This enables to test a proteome for their diverse modifications and different modified proteins at the same time (see Figure 1). Furthermore, we show that the detectable modified peptides as well as the complexity of modifications increases with age in C. elegans.

It was not surprising to find that the number of detectable modified peptides was increased during aging as it was suggested already in the introduction. Most studies describe oxidative modifications (carbonylation of proteins) that are most likely a result of oxidative damage (summarized in Levine and Stadtman, 2001). Adachi and colleagues (1998) showed that the number of carbonylated proteins might be linked to the lifespan of animals (here: C. elegans) as the long-lived age-1 mutant had less modified proteins with age whereas the short-lived mev-1 mutant shows higher levels of modified proteins compared to control animals (Adachi et al., 1998). Our results suggest that the age-dependent increase in protein modifications also appears to be the case for tryptophan-specific modifications. The adaptations that 8 were used for the additional runs might enable to identify enough peptides to also analyse the data in the sense of other tryptophan-independent modifications such as e.g. carbonylated methionine to compare our data to previous studies. It would be as well interesting to test the proteome of mutants that affect lifespan as it was done by Adachi and coworkers (1998) to see whether the number of tryptophan modifications also correlate with the lifespan-expectancy.

Our preliminary run also showed that the complexity of tryptophan modifications increases with the age of the organism. We find more double- and triple-oxidized tryptophan residues compared to proteins of younger animals (Figure 4-C). This might reflect a longer half-life of the proteins that enables several modifications at one tryptophan residue over time.

It was suggested that especially oxidation-derived modification are mainly a product of handling the samples and technical artifacts (Perdivara et al., 2010). This might be true for a certain percentage of modifications we detected but does not explain the differences between the here analyzed samples as e.g. the increase that we see with higher ages of the animals.

The adjustments to the experimental setup were supposed to confirm previous findings and answer open questions. For example, we included a quantification step to determine the amount of a specific peptide/ protein compared to the amount of its modified form. With this we would like to find out whether we can detect

213

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 213 more modified peptides because the events of specific modifications occur more often during aging or are less repaired or whether only certain peptides are more abundant as the expression of the protein is increased with age? The quantification of unmodified and modified peptides might give us answers in this direction.

And what are the proteins that are more prone to be modified? Do they have a longer half-life that modifications on tryptophan residues can accumulate? Goto et al. (1999) described that some proteins are more susceptible to oxidative modifications such as carbonylation (Goto et al., 1999). It remains to be elucidated whether those proteins or the unmodified proteins have a special function in the cell. Itmight be very well possible that central players that regulate the homeostasis of the cell are protected either through their special amino acid sequence and folding or by control mechanisms that induce a fast replacement of modified molecules. It was e.g. suggested that heme is protective against nitrosylation events (Herold and Shivashankar, 2003). Finally, such a list of modified proteins can be compared to other proteomic studies to make connections between amino acid modifications and e.g. certain diseases. One example would be the aggregating proteome that was recently identified (Walther et al., 2015).

In general, tryptophan residues can usually be found as being buried inside a protein due to their high hydrophobicity, being well protective from e.g. ROS or RNS and therefore less accessible for modifications (Oobatake and Ooi, 1988). But the tryptophan residues that are exposed on the proteins surface are often involved in interaction with other molecules, such as e.g. DNA, RNA (Pérez-Cañadillas, 2006) or lipid membranes (Chan et al., 2006). A modification at these sites might be crucial as it might interfere with the function of the protein even though the active site of the protein is still intact. Furthermore, tryptophan-tryptophan motifs stabilize the quaternary structure of proteins especially in loops between two beta-sheets by interacting with surrounding amino acids of the peptide chain (Santiveri and Jiménez, 2010). Alterations in the tryptophan-tryptophan motif by modifications might influence the stability and consequently the functionality of the protein.

The preliminary experiment showed that a global detection was possible but we probably only found the most abundant modified peptides. To get a bigger picture to answer some of the mentioned question we inserted several detection cycles for the next experiment round to increase the number of detected peptides.

How do tryptophan residues get modified? The first idea would be that the amino acid is more prone to react because of its aromatic structure and that with a longer half- life of the protein the chance of being oxidized increases. It would be interesting to determine the half-life of those proteins that we detected to be modified at tryptophan residues. Does the half-life increases during aging because of reduced translational

214

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 214 activity and decreased degradation in aged animals?

Besides the hypothesis of being modified posttranslational, it might be also possible that the modification already occurs during the translation process (co-translational). It was shown e.g. that the tRNA for serine is not highly specific for only serine but can also bind the non-protein amino acid β-N-methylamino-L-alanine (BMAA), a serine metabolite that is produced by cyanobacteria , endosymbionts of cycads. This binding leads to a misincorporation of the metabolite instead of serine during translation (Dunlop et al., 2013). What about if this is also the case for the tryptophan-specific tRNA? Experiments showed that this happens in bacteria where nitro-tryptophan was bound to tRNAs even with a higher affinity than tryptophan itself (Buddha and Crane, 2005). Cellular tryptophan decreases during aging and tryptophan metabolites get more abundant. In order to still produce a protein, it might be a compromise to a situation with low tryptophan levels and the system adapted in a way that it inserts tryptophan metabolites in the growing polypeptide instead of tryptophan. This might be reasonable as tryptophan is the least abundant essential amino acid and a lack of tryptophan might result in a total breakdown of the translational machinery. It would be interesting to isolate charged tRNAs and analyse their cargo as they did it for serine (Dunlop et al., 2013). And to test whether manipulating the ratios of 8 tryptophan metabolites influences the choice of cargo, either by inhibiting specific enzymes or by feeding certain metabolites.

As mentioned before, a lot of tryptophan residues are less accessible for modifications as they are located inside the protein (Oobatake and Ooi, 1988). Do we find these residues modified anyway? That would also support the idea that the modification already occurred during translation when the protein structure was not assembled yet.

If both theories are true, it would mean that during aging the proteins get in general older and more prone to be modified but also that changes in the aged metabolome might influence the quality of newly synthesized proteins.

A question that remains is whether the accumulation of modified tryptophan residues is a phenotype of aging of whether it is a reason for cellular decline. A first step would be to see whether it is possible to induce the modifications by either reducing translation or protein quality control, increasing oxidation events by e.g. ROS or by manipulating the balance of tryptophan and tryptophan metabolites and see whether this has any influence on health- and lifespan.

As a depletion of tdo-2 increases internal tryptophan level and induces an increase in health- and lifespan in C .elegans (Van der Goot et al., 2012), it might be also possible that it affects the number of modifications at tryptophan residues. Whether

215

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 215 the possible changes in tryptophan modifications by tdo-2 depletion might be involved in the health- and lifespan benefits has to be elucidated. Experimental procedures Media and strains Animals were age synchronized by hypochlorite treatment. Animals were maintained at 20˚C on Nematode Growth Medium (NGM) and fed with E. coli OP50. Animals were transferred to NGM medium containing 5-fluoro-2’deoxy-uridine (FUDR) on day 1 of adulthood (day after larval stage 4) to prevent offspring from growing.

The following strains were used in this study: Wild type N2, CRISPR/Cas9 mutant tdo-2 (del B) =OW715 tdo-2 (zg216)III; CRISPR/Cas9 mutant tdo-2 (del) =OW716 tdo-2 (zg217)III; CRISPR/Cas9 mutant tdo-2 (ΔPLD) =OW717 tdo-2 (zg218)III. RNA interference RNAi experiments were performed on NGM medium containing isopropylthio-β- D-galactoside (IPTG, 15mg/L) and 50μg/ml ampicillin. Plates were seeded with RNAi bacteria and kept at room temperature for two days to enable the production of dsRNA. Age-synchronized animals were exposed to the RNAi treatment from larval stage 1 onwards and transferred to FUDR-containing RNAi plates on day 1 of adulthood (one day after L4 stage) to avoid the development of progeny. Tryptophan feeding L-tryptophan was added to the medium prior to autoclaving in a concentration of 10mM or 20mM respectively. Tryptophan application started at larval stage L4. Metabolite measurements About 5000 animals per experimental group were washed of the NGM plates with PBS and subsequently extensively washed to remove as much bacteria from cuticle and gut. cOmpleteTM protease inhibitor (Roche) was finally added and the samples were snap frozen. Only samples from the same experiment were compared. The samples were lysed by sonication, excessive worm debris was removed and the protein concentration was determined. The protein concentration of the different samples was equalized and analyzed for tryptophan-modified peptides. Detection of tryptophan modifications and analysis 100 µg total protein lysate per sample were subjected to tryptic digestion using the FASP-protocol (Wisniewski et al., 2009). Subsequently, 2 µg of each sample was measured by nano-LC-MS/MS using the SWATH technology as published previously (Simbuerger et al., 2016). Briefly, 2 µg per sample were injected for separation by nano-RP-chromatography. The directly coupled TripleTOF5600+™ mass spectrometer repeatedly measured a 50 ms TOF MS spectrum and MS/MSALL-

216

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 216 spectra in the mass range of 400-1000 m/z using SWATH windows of 15 m/z width. Libraries were generated using 4 h- and 2 h-IDA-runs of pooled samples; therefore, sample replicates from each condition were pooled and measured using a TOP30 method. Database searches against the wormbase (June 2016, www.wormbase.org) were accomplished using Protein Pilot 4.5 software (Sciex, Darmstadt, Germany). SWATH data was processed using PeakView/Skyline.

AUTHORS’ CONTRIBUTIONS: HM, JR, PO and EAAN designed the experiments. HM and EAAN wrote the manuscript. HM, RIS, MK collected the samples and prepared them for analysis. JR performed the mass spectrometry analysis.

ACKNOWLEDGEMENTS 8 We thank Yin Fai Chan for experimental support. This project was funded by the European Research Council (ERC) starting grant (to EAAN) and by support from the university alumni chapter “Gooische Groningers” through the Ubbo Emmius Fonds (to EAAN).

217

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 217 REFERENCES

• Adachi, H, Fijiwara, Y, Ishii, N: Effects of oxygen on methemoglobin catalyze the isomerization of protein carbonyl and aging in Caenorhabditis elegans peroxynitrite to nitrate. Biochemistry 42(47), 14036- mutants with long (age-1) and short (mev-1) life spans. 46 (2003). J Gerontol A Biol Sci Med Sci 53(4), 240-4 (1998). • Johnson, H and Eyers, CE: Analysis of post- • Brégère, C, Rebrin, I, Sohal, RS: Detection and translational modifications by LC-MS/MS. Methods characterization of in vivo nitration and oxidation of Mol Biol 658, 93-108 (2010). tryptophan residues in proteins. Methods Enzymol 441, • Levine, RL and Stadtman, ER: Oxidative modification 339-49 (2008). of proteins during aging. Experimental Gerontology • Buddha, MR, Crane, BR: Structure and activity of an 36, 1495-1502 (2001). aminoacyl-tRNA synthetase that charges tRNA with • Lipinski, MM et al.: Genome-wide analysis reveals nitro-tryptophan. Nat Struct Mol Biol 12(3), 274-5 mechanisms modulating autophagy in normal brain (2005). aging and in Alzheimer’s disease. Proc. Natl. Acad. • Chan, DI, Prenner, EJ, Vogel, HJ: Tryptophan- and Sci. U S A 107, 14164-14169 (2010). arginine-rich antimicrobial peptides: structures and • Møller, IM and Kristensen, BK: Protein oxidation in mechanisms of action. Biochim Biophys Acta 1758(9), plant mitochondria detected as oxidized tryptophan. 1184-202 (2006). Free Radical Biology and Medecine 40, 430-435 • Chiti, F, Dobson, CM: Amyloid formation by globular (2006). proteins under native conditions. Nat Chem Biol 5, 15- • Naismith, JH: Tryptophan oxygenation: mechanistic 22 (2009). considerations. Biochem Soc Trans 40(3), 509-14 • Creasy, DM, Cottrell, JS: Unimod: Protein (2012). modifications for mass spectrometry. Proteomics 4(6), • Nuriel, T, Hansler, A, Gross, SS: Protein 1534-6 (2004). nitrotryptophan: Formation, significance and • Dahl, JU, Gray, MJ, Jakob, U: Protein quality control identification.J Proteomics 74(11), 2300-2312 (2011). under oxidative stress conditions. J Mol Biol 427(7), • Oobatake, M, Ooi, T: Characteristic thermodynamic 1549-1563 (2015). properties of hydrated water for 20 amino acid residues • Dreyfus, JC, Kahn, A, Schapira, F: Posttranslational in globular proteins. J Biochem 104(3), 433-9 (1988). modifications of enzymes. Curr Top Cell Regul 14, • Perdivara, I, Deterding, LJ, Przybylski, M, Tomer, 243-97 (1978). KB: Mass spectrometric identification of oxidative • Dunlop, RA, Cox, PA, Banack, SA, Rodgers, KJ: The modifications of tryptophan residues in proteins: non-protein amino acid BMAA is misincorporated into chemical artifact or post-translational modification? J human proteins in place of L-serine causing protein Am Soc Mass Spectrom 21(7), 1114-1117 (2010). misfolding and aggregation. PLoS One 8(9), (2013). • Pérez-Cañadillas, JM: Grabbing the message: • Garner, B et al.: Identification of gluthathionyl-3- structural basis of mRNA 3’UTR recognition by Hrp1. hydroxykynurenine glucoside as a novel flourophore EMBO J 25(13), 3167-78 (2006). associated with aging of the human lens. J Biol Chem • Rebrin, I, Brégère, C, Kamzalov, S, Gallaher, TK, 274(30), 20847-54 (1999). Sohal, RS: Nitration of tryptophan 372 in succinyl- • Gavilán, MP et al.: Age-related differences in the CoA:3-ketoacid CoA transferase during aging in rat dynamics of hippocampal proteasome recovery. J heart mitochondria. Biochemistry 46(35), 10130-44 Neurochem 123(4), 635-44 (2012). (2007). • Gillet, LC et al.: Targeted data extraction of the MS/ • Rothstein, M: Recent developments in the age-related MS spectra generated by data-independent acquisition: alteration of enzymes: a review. Mech Ageing Dev 6, a new concept for consistent and accurate proteome 241-257 (1977). analysis. Mol Cell Proteomics 11(6), (2012). • Santiveri, CM, Jiménez, MA: Tryptophan residues: • Goto S et al.: Carbonylated proteins in aging and Scarce in proteins but strong stabilizers of beta-hairpin exercise: immunoblot approaches. Mechanisms of peptides. Biopolymers 94(6), 779-90 (2010). Aging and Development 107, 245-253 (1999). • Simbürger, JM, Dettmer, K, Oefner, PJ, Reinders, J: • Greenacre, SA, Ischiropoulos, H: Tyrosine nitration: Optimizing the SWATH-MS-workflow for label-free localization, quantification, consequences for protein proteomics. J Proteomics 145, 137-40 (2016). function and signal transduction. Free Radic Res 34, • Stadtman, ER, Van Remmen, H, Richardson, A, Wehr, 541-581 (2001). NB, Levine, RL: Methionine oxidation and aging. • Hains, PG, Truscott, RJ: Post-translational Biochimica et Biophysica Acta 1703, 135-140 (2005). modifications in the nuclear region of young, aged, and • Van der Goot, AT et al.: Delaying aging and the aging- cataract human lenses. J Proteome Res 6(10), 3935-43 associated decline in protein homeostasis by inhibition (2007). of tryptophan degradation. Proc Natl Acad Sci U S A • Herold, S, Shivashankar, K: Metmyoglobin and 109(37), 14912-7 (2012).

218

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 218 • Voth, W et al.: The protein targeting factor Get3 analysis. Nat Methods 6(5), 359-62 (2009). functions as ATp-independent chaperone under • Yamakura, F, Ikeda, K: Modification of tryptophan and oxidative stress conditions. Mol Cell 56, 116-127 tryptophan residues in proteins by reactive nitrogen (2014). species. Nitric Oxide 14, 152-161 (2006). • Walther, DM et al.: Proteome imbalance is linked • Young, NL, Plazas-Mayorca, MD, Garcia, BA: with proteostasis decline and aggregation in aging C. Systems-wide proteomic characterization of elegans. Cell 161(4), 919-932 (2015). combinatorial post-translational modification patterns. • Wisniewski, JR, Zougman, A, Nagaraj, N, Mann, M: Expert Rev Proteomics 7(1), 79-92 (2010). Universal sample preparation method for proteome

219

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 219 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 220 Chapter IX

Discussion and perspectives

Helen Michels1

1European Research Institute for the Biology of Aging, University of Groningen, University Medical Centre Groningen, Laboratory of Molecular Neurobiology of Aging, The Netherlands

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 221 The previous chapters summarized the different theories of aging, the current knowledge on amino acid metabolism and its influence on aging, in particular on tryptophan and its breakdown products, and the research that has been performed on TDO-2 in C. elegans in the context of this thesis. The following chapter will connect and discuss the findings of the different chapters, summarizes open questions and suggests follow-up experiments that might help to better understand how TDO-2 – and the other two amino acid dioxygenases - control health- and lifespan.

9.1 Mechanism of lifespan regulation by TDO-2 Aging is a very complex phenomenon, which is also reflected in the broad range of lifespan-regulating genes that have been identified so far mainly in studies with C. elegans (summarized in Yanos et al., 2012). As already described in the introduction, there are several pathways that influence the aging process and there has not been only one ultimate aging gene been identified. Even though these studies suggest that some pathways and organelles take a more dominant role, the amount of recent publications on research in the field of aging shows that we are still far from having identified all players that contribute to the aging process. This thesis focuses on a new player in the aging process being connected to several cellular pathways: TDO- 2.

We have identified TDO-2 as a regulator of health and lifespan in C. elegans. Depletion of TDO-2 suppresses age-related phenotypes, which raises the question why this gene exists. One possible explanation is that post-reproductive phenotypes, such as lifespan, are not selected for by evolution. Therefore, genes that are beneficial during development and reproduction of the organism, but not afterwards, may not be down-regulated post reproduction (Parsons, 2007; Blagosklonny, 2006; Kirkwood, 2005; Curran and Ruvkun, 2007). Such mechanisms has been proposed also based on findings in C. elegans, in which knockdown of several developmental genes late in life extended the animals lifespan (Curran and Ruvkun, 2007). TDO-2 may fall into this category of genes: we mainly find positive effects of TDO-2 depletion on health- and lifespan, suggesting that this gene is not essential, but see a lowered in fertility of the organism, which might explain why the gene has been conserved during evolution (chapter III and paragraph 9.3). Important to note is that we work under laboratory conditions with low numbers of infectious microorganisms and plenty of food so we cannot exclude that TDO-2 is essential for the survival in the wild.

Independently of this, a central question of this study was do elucidate the cellular mechanism by which TDO-2 can regulate lifespan.

222

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 222 Depletion of TDO-2 appears to affect several cellular aspects of aging. First, depletion of TDO-2 protects against age-related proteotoxicity, suggesting a role in protein homeostasis (Campesan et al., 2011; Van der Goot et al., 2012; Breda et al., 2016; discussion on TDO-2’s role in neurodegeneration in paragraph 9.9). It is, however, unclear through which process it interferes with protein homeostasis. We find that changes in translation are not responsible nor can we find an upregulation of stress response pathways, such as the heat shock response, ER stress, and mtUPR pathways that maintain protein homeostasis in a cell (Chapter IV, transcriptome profile). In addition to this, chaperones, previously shown to protect against age- related proteotoxicity, do not play a role in the proteoprotective effects of tdo-2 depletion (Van der Goot et al., 2012).

Depletion of tdo-2 has been shown to increase the inclusion formation of the PD- associated protein alpha-synuclein (Van der Goot et al., 2012). A possible role in the aggregation process could be an explanation for the proteoprotective effects. On the other hand, depletion of tdo-2 does not affect the inclusion formation of HD- associated Huntingtin, while it still improves the health of animals expressing this toxic protein (Van der Goot et al., 2012). Thus, a possible involvement in aggregation processes cannot be the only explanation. Second, depletion of TDO, and other amino acid dioxygenases, results in phenotypes 9 that resemble long-lived mitochondrial mutants (summarized in Yanos et al., 2012): they are small in size (chapter V), have reduced reproduction rates (chapter III) and decrease mitochondrial respiration and capacity and often show an increased longevity (chapter VI). On the other hand, the amino-acid dioxygenases show some differences compared to the mitochondrial genes: their depletion after development still results in a lifespan extension (chapter VI) whereas it has been shown for other mitochondrial genes, that the lifespan extension can only be achieved when they are depleted during a narrow time window in development (Dillin et al., 2002; Rea et al., 2007). This may indicate that their lifespan regulation is unrelated to their effect on the mitochondria or that this represents a new post reproductive mechanism by which mitochondria regulate lifespan.

One possible mechanism by which mitochondria regulate lifespan is by regulating the levels of reactive oxygen species (ROS) (see introduction). TDO-2 may act by regulating ROS as well in three different ways. It can use ROS instead of molecular oxygen to convert tryptophan into formyl-kynurenine (Ocampo et al., 2014). It might be therefore possible that ROS levels are increased when inhibiting TDO- 2’s activity. On the other hand, the activity of TDO-2 controls tryptophan levels, a highly reactive molecule due to its aromatic structure that has been shown to have anti-oxidant properties (Tsompo et al., 2009). Tryptophan levels decrease with age which might have effects on the redox state of the cell. By increasing tryptophan

223

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 223 levels through inhibition of TDO-2, it might a counteracting process by lowering ROS levels which could possibly add to the lifespan-extending phenotype. And finally, by knowing that TDO-2 regulates mitochondrial respiration, the enzyme could again have a general influence on the redox state of the cell. We performed a range of experiments to test these hypotheses (supplement of Chapter VI) which in total suggest that the phenotypes of depleted tdo-2 are not affected by ROS and the oxidative damage theory cannot be applied.

In this context, it has to be emphasized that we only tested the role of ROS for TDO-2. Does the depletion of the other two amino acid dioxygenases show similar results? And what happens with the redox state when feeding free amino acids such as tryptophan? This is especially of interest when it comes to cdo-1 depletion. CDO-1 converts cysteine towards taurine (Brand et al., 1998; Satsu et al., 2003). By inhibiting this enzymatic step, more free cysteine is available for alternative reaction such as e.g. the conversion towards glutathione, a very potent antioxidant (Taylor, 1958).

This paragraph demonstrates that the effects of TDO-2 depletion are not linked to a single previous described aging pathway. It is a protein that is linking several metabolic processes in the cell making the effects on aging complex and difficult to analyze.

9.2 Regulation of mitochondrial respiration by amino acid dioxygenases This study shows that depletion of the three amino acid dioxygenases TDO-2, CDO- 1 and HPD-1 results in the same phenotypes suggesting a common mechanism. The central organelle of this study is the mitochondrion.

So far, we could identify one single subunit of complex IV of the ETC to possibly be involved in the regulation of mitochondrial respiration. It is not clear whether other subunits of complex IV or subunits of other complexes are also involved. To answer this, it would be necessary to systematically knockdown different subunits and check whether this has an effect on the regulation of respiration by amino acid dioxygenases. Once components are known, it still needs to be assessed whether amino acid dioxygenases control respiration directly or whether this is a consequence of a cascade of events in the cell. For this, the localization of amino acid dioxygenases within the cell is giving a first hint: are amino acid dioxygenases located at the mitochondria? This could be either done in cells or by tagging the specific proteins inC. elegans e.g. by a GFP knock-in by the CRISPR/Cas-9 system. If the signaling is indirect, through which pathway? Are there metabolites involved?

224

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 224 As shown in chapter IV, animals depleted for amino acid dioxygenases react less well to FCCP than wild type animals. FCCP induces a fast proton leakage and the ability of an organism to react on this drug by boosting mitochondrial respiration can be used to detect a dysfunction of the mitochondria, because increasing the amount of mitochondria is not possible in such a short period. A reduction in mitochondrial oxygen consumption, however, can also be a result of less mitochondria in the cell, which we have not been able to exclude yet. We started to investigate the quantity of mitochondria by measuring mitochondrial protein levels. It has to be mentioned here, that C. elegans research is still limited in the availability of specific antibodies that we could only test for MTCO-1, a subunit of complex IV of the ETC. Preliminary runs showed no reduction in MTCO-1 levels in tdo-2 depleted animals (data not shown). Reporter strains with fluorescent-labeled mitochondria can be used in addition to determine the amount of mitochondria or measure the ratios of nuclear and mitochondrial DNA of wild type versus amino acid dioxygenases-depleted animals.

In this context, the mitochondrial network might also affect the productivity of mitochondria. Studies on subunit F26E4.6 in C. elegans (the subunit of complex IV that we identified in chapter IV to act in line with the amino acid dioxygenases) have shown a highly fused mitochondrial network when depleting F26E4.6 (Smith et al., 2015). Does the depletion of amino acid dioxygenases also change the mitochondrial 9 fusion/fission processes in the cell? Reporter C. elegans strains with fluorescent labeled mitochondria can help to answer this question. Preliminary test runs, however, suggest that the depletion of amino acid dioxygenases have no obvious effect on the mitochondrial network or mitochondrial number. But these experiments should be further optimized and repeated.

The enzymatic reactions of TDO and IDO have been extensively studied on the molecule’s level (summarized in Thackray et al., 2008). It seems to be that these two enzymes form a special group of dioxygenases enzymes in the way of how they bind to ferrous heme (Geng and Liu, 2014). The structures and chemical reactions of HPD and CDO on the other hand have been hardly studied. It would be interesting to see whether they show comparable characteristics as TDO and IDO and whether this explains how they modify mitochondrial respiration?

The biosynthesis of heme is located to the mitochondria where heme also is a crucial factor for the assembly of the complexes of the electron transport chain (summarized in Duggan et al., 2011). It has been demonstrated that aβ that is imported to the mitochondria binds heme which has been proposed as an explanation for reduced ETC function during Alzheimer’s disease (Santos et al., 2010). Could it be that the binding of heme by dioxygenases is influencing mitochondrial function or is mitochondrial activity controlled by dioxygenases by balancing the heme availibility?

225

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 225 In the context of molecules’ structures, Salminen and colleagues have found that a group of dioxygenases, mainly non-heme dioxygenases, do use oxoglutarate as cofactor, a mechanism that enables them to induce transcriptional changes in response to oxygen changes and alterations in mitochondrial activity (Salminen et al., 2015): oxoglutarate is a product of the TCA cycle. Its level reflects thus the activity of the TCA cycle. Do amino acid dioxygenases belong to that group as well? Do they maybe function in a comparable manner? What would be the critical co-substrate? Oxoglutarate as it is for the other non-heme dioxygenases? One of the targets of these oxoglutarate-binding dioxygenases is a collagen hydroxylase which is crucial for the synthesis of collagens. During aging, the composition and organization of collagens in various tissues such as skin, bones and the vascular wall is altered which impacts the aging process (Salminen et al., 2015). Whether collagens also affect mitochondrial performance has not been tested yet but could very well be as those molecules strongly influence the structure and functionality of membranes. The transcriptome profiling of tdo-2 depleted animals revealed that the expression of genes involved in the synthesis of structural components, mainly collagens, are decreased (Chapter IV). This opens a new hypothesis: Could the aging process and the mitochondrial performance be affected by the changes in collagen composition?

Another aspect of mitochondrial dysfunction is the misbalance of protein production that is encoded by the nucleus on the one hand and the mitochondrion on the other hand (mtUPR, see introduction). We approached this theory by testing various components of the translational control and machinery (Chapter VI). Furthermore, could it be that the amino acid dioxygenases only control one of the two groups of proteins (nuclear- of mitochondrial encoded) which would lead to the alterations in respiration?

We showed that of the tested proteins, only TDO-2, HPD-1 and CDO-1, combining oxygen consumption with amino acid degradation, do have the here described effects on mitochondrial respiration (Chapter VI). Do they perhaps function as sensors for nutrients and oxygen, communicating simultaneously the availability of amino acids and oxygen that is needed to generate ATP for amino acid processing? Are they still able to control mitochondrial respiration when one of the two characteristics is inhibited? Is it possible to generate a mutant that still binds tryptophan but not oxygen and vice versa? Like this, it can be assessed whether the combined binding of both molecules is necessary for the regulation of mitochondrial functions. Unfortunately, studies on the enzymatic structure of TDO showed that the binding of oxygen or tryptophan is strongly connected (Chauhan et al., 2009), suggesting that such a mutant is unlikely because a substrate-binding mutant is probably automatically not binding oxygen and vice versa. It might be therefore better to modulate the oxygen and amino acid availability of the environment (providing high levels of amino acids but reducing

226

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 226 oxygen availability at the same time) to test whether amino acid dioxygenases are indeed important for signaling this condition e.g. to the mitochondria. The use of a mammalian cell culture model would facilitate the testing. However, one has to keep in mind that oxygen availability controls not only energy metabolism but also has a fundamental role in the regulation of gene expression (summarized in Salminen et al., 2015). Changes in oxygen levels might therefore provoke unexpected responses that are not necessarily linked to the mechanisms regulated by amino acid dioxygenases. If the three enzymes do act as sensors, amino acid dioxygenases build a new category of enzymes integrating nutrient with oxygen availability contributing to the control of metabolism in a similar manner as e.g. mTOR that is connecting nutrient availability to energy levels in a cell (Laplante and Sabatini, 2012).

9.3 Downregulation of the reproduction system and the immune response The transcriptome analysis of tdo-2 depleted animals revealed that mitochondrial beta oxidation is upregulated. We first focused on these findings as mitochondrial function had been linked to lifespan regulation before (Lee et al., 2003; Hamilton et al., 2005; Hansen et al., 2005; Curran and Ruvkun, 2007). However, the transcriptome profiling also gave other insights in the expression patterns that still can be explored. We found e.g. that some genes that are involved in reproduction (e.g. major sperm proteins) are 9 downregulated in tdo-2 depleted animals which fits with the low reproduction rates of those animals (chapter III). It could be a tryptophan metabolism-specific effect as melatonin levels are possibly changed, a molecule that is strongly involved in the regulation of reproductive behavior (Waldhauser et al., 1993). On the other hand, melatonin-related effects are not really likely as the observed transcript changes are independent of the melatonin/serotonin pathway: we still find those changes in the tph-1(del) mutant. Could it then be that the phenotypes are a secondary effect induced by low energy availability due to reduced mitochondrial activity? Several mitochondrial mutants have been described to show reduced fertility before (Yanos et al., 2012). This is also the case for tdo-2 depleted animals (Chapter III). It is possible that the organism reduces its non-essential, energy-consuming functions such as reproduction to save ATP for other acute tasks. The same might be the case for the immune response of the animal: we see a downregulation of immune system components in tdo-2 depleted animals. Especially under laboratory conditions, the maintenance of an active immune system is less needed. However, this raises the question whether tdo-2 depleted animals and possibly other mitochondrial mutants are more prone for infections then wild type animals and how this affects their ability to survive in the wild.

227

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 227 9.4 Amino acid dioxygenases share cellular functions: Do they communicate? The fact that the amino acid dioxygenases TDO-2, CDO-1 and HPD-1 seem to share similar functions in the cell, raises the questions whether there is a cross-talk between those enzymes? All three enzymes share a common expression profile: they are mainly expressed in the liver (Danesch et al., 1987; Hosokawa et al., 1990; Ruetschi et al., 1993), a mitochondrial-rich tissue (Forner et al., 2006). In addition, the transcriptome profile showed thathpd-1 is upregulated in tdo-2 depleted animals so that there might be indeed a feed-back mechanism. In the case of cdo-1, we could not detect sufficient cdo-1 transcripts in none of the conditions, being wild type or tdo-2 depleted. We also observed very low transcript levels of cdo-1 when we cloned the sequence into the RNAi vector and amplification was only possible with a nested-PCR reaction. This makes it hard to test for changes in expression when one of the other amino acid dioxygenases is depleted. It seems however that cdo-1 transcript levels are higher in human cells. It might be therefore better to perform those experiments in a cell model.

What effect do we see when depleting more than one amino acid dioxygenase? Depleting hpd-1 in tdo-2(del) mutants does not lead to an additional motility increase as it does in wild type animals (Appendix Figures A7 and A8). The same effect can be observed when depleting tdo-2 in hpd-1(del) animals. Analogous to the findings on healthspan, depletion of hpd-1 in tdo-2(del) animals and depletion of tdo-2 in hpd-1(del) mutants does not reduce oxygen consumption rates any further. This suggests that there is indeed a connection between those two enzymes. Based on the findings of the transcriptome profile, the depletion of tdo-2 might influence the expression of hpd-1.

Assuming that there is indeed a communication between the three enzymes and possible compensation for the deficiency of one enzyme, what happens when deleting all three enzymes? Would the animal be able to develop and survive? Would there be any reproduction at all (which complicates the generation of such a mutant)? Would the animal even be fine and extremely long-lived?

9.5. Are changes in tryptophan levels responsible for the phenotypes of tdo-2 depletion? The findings that tryptophan supplementation increases the healthspan ofC. elegans in wildtype and PD animals (Van der Goot et al., 2012; introduction and paragraph 9.9) lead to the question if changes in tryptophan levels are responsible for the other phenotypes we observe when depleting tdo-2.

228

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 228 The observation that tdo-2 depletion and tryptophan feeding result in the same phenotype might lead to the conclusion that tryptophan levels are indeed the central factor. But one should keep in mind that we are not looking at necessarily at the same processes when depleting tdo-2 or when feeding tryptophan. In both cases, an increase in internal tryptophan levels can be observed. However, in the case of tryptophan feeding, we also increase the levels of the metabolites of the kynurenine pathway as it is completely active and even stimulated by the higher levels of substrate.

With this background, tryptophan feeding could be theoretically seen as a state with hyperactive TDO-2 as it is described that tryptophan stimulates its degradation mainly via activation of the kynurenine pathway (Danesch et al., 1987). This would suggest to measure the expression levels and enzymatic activity of TDO-2 when supplying tryptophan in C. elegans. If TDO-2 is indeed hyperactive, it would mean that depletion of tdo-2 but also stimulation of TDO-2 by tryptophan supply both result in similar phenotypes such as e.g. motility increase. The generation of an overexpression mutant either by integration of an extrachromosomal array or by single gene knock-in could give additional information. Does an overexpression of TDO-2 result in more mitochondrial activity? Or is a specific balanced level of the enzyme the optimum for maximal mitochondrial functionality? 9 Furthermore, effects that accompany tryptophan feeding could be still dependent on other metabolite changes as degradation pathways are still active. In order to exclude such an effect, we also tried to combine tryptophan feeding with a depletion of TDO- 2. However, feeding the standard concentrations of tryptophan (10mM and 20mM) had toxic effects when combined with depleted tdo-2. It could be that internal levels are reaching a toxic levels when the major degradation pathway is inhibited. It is therefore worthwhile to try lower concentrations and to repeat the assays.

Amino acid supplementation is indeed activating a range of signaling cascades in the cell. The systematic feeding of amino acids by Claire Edwards showed that tryptophan feeding increases the expression of small heat shock proteins and the ER stress response and that the lifespan extension is dependent on GCN-2 and AAK-2 (Edwards et al., 2015). Our results on tdo-2 depleted animals show the opposite: we did not find changes in the heat shock response, ER stress response nor did GCN-2 or AAK-2 affect the lifespan regulation by TDO-2. The study of Claire Edwards is missing some critical controls to actually show that the effects are provoked directly by tryptophan and not a general stress effect by changes in the animal’s nutrition and environment that might also result in a lifespan extension (Edwards et al., 2015). In addition to that, it should still be tested whether phenotypes of tryptophan-fed animals are dependent on serotonin signaling, translation etc. as we performed it for tdo-2 depleted animals.

229

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 229 We finally assayed whether tryptophan supplementation has an influence on oxygen consumption rates (Appendix Figure A9). It has been shown that amino acid supplementation has an effect on oxygen consumption rates or ATP levels (Edwards et al., 2015). This overlaps partially with our findings: tryptophan-fed animals do not show alterations in basal mitochondrial respiration (Appendix Figure A9). However, the application of the Seahorse analyzer (Chapter VII) gave us the possibility to test the respiration rates under stressed conditions (proton leakage): tryptophan- fed animals showed a decreased maximum mitochondrial capacity, similar to tdo-2 depleted animals (Chapter VI). This suggests that tryptophan indeed has effects on mitochondrial function which is especially critical under stressed conditions with high mitochondrial activity.

Even though, the study of Claire Edwards did not find a direct link between amino acids and respiration, it showed that the lifespan-extending effects of amino acid supplementation could be inhibited by mutations in the electron transport chain of the mitochondria (Edwards et al., 2015). This suggests indeed a link between amino acid metabolism, mitochondrial function, and aging supporting our findings on amino acid dioxygenases.

9.6. Do changes in certain amino acid levels mediate the effects on health- and lifespan when depleting amino acid dioxygenases? The fact that tryptophan metabolism is involved in health- and lifespan regulation, leads to the question whether other amino acids and their metabolic pathways might have similar effects. First attempts to systematically feed all amino acids separately and test the animals for their motility showed that further optimization is strongly needed before any conclusions can be drawn. Especially the uptake of the amino acid has to be guaranteed. However, genetic inhibition of certain enzymes involved in the degradation of specific amino acids or classes of amino acids (Chapter VI) showed that they do not regulate healthspan as TDO-2 does. However, the findings that the two other amino acid dioxygenases CDO-1 and HPD-1 seem to share similar functions as TDO-2 might propose that at least the metabolism of cysteine and tyrosine are also important for the organismal health- and lifespan determination, which is similar as for the metabolism of tryptophan. Supplementation of cysteine and tyrosine will be required to confirm this similarity.

In addition, when it comes to amino acid homeostasis, the feedback response can play a crucial role and induce side-effects that we might not consider when interpreting results. Could it be that we have to look at effects that are induced by level changes of another amino acid? As already mentioned in the introduction, amino acids are strongly linked and might be even converted into another amino acid. Methionine and

230

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 230 serine can be e.g. precursors for cysteine (summarized in Jung, 2015). Furthermore, cellular processes can affect the amino acid homeostasis and induce shifts: Mutants of the respiratory chain have e.g. typically increased levels of branched amino acids (leucine, valine, isoleucine) (Falk et al., 2008). By generating an amino acid profile oftdo-2(del) and tdo-2 RNAi-depleted animals, we found that they also show increased levels of those three amino acids (Appendix Figure A10). It was suggested that the increase in production of branched amino acids is a compensation effect to react on low mitochondrial activity: branched amino acids can be used by the cell to produce substrates for mitochondrial ATP production (Falk et al., 2008). It might be that the tdo-2 depletion dependent increase in levels of leucine, valine, isoleucine is responsible for the induction of pathways that are linked to these three amino acids and not directly to tryptophan and might result in the observed phenotypes.

9.7. Other tryptophan metabolites: kynurenine and serotonin

It has to be emphasized that this study focusses on certain phenotypes, which are healthspan and lifespan extension and control of mitochondrial respiration. For these phenotypes, we could show that the kynurenine pathway, the serotonin and tryptamine pathway as well as AHR-1, GCN-2, mTOR and AAK-2 (AMPK) are not involved in the regulating mechanism. That does not mean that other functions of TDO-2 are also independent of these candidates. We do observe e.g. that tdo- 9 2(del) animals are smaller in size (Chapter V), which suggests actual alterations in development. Transcriptome profiling showed a long list of differentially expressed genes that were dependent on either the kynurenine pathway or the serotonin pathway (partially shown in heatmaps of Chapter VI). We did not focus on those hits as we showed before that the health- and lifespan benefits in C. elegans were independent of those pathways. But the data could become of high interest again when focusing on other phenotypes.

Focusing on the kynurenine pathway, C. elegans results differ from other organisms especially from those found in flies: the group of F. Giorgini showed that health- and lifespan benefits in Drosophila are at least partially dependent on the metabolite changes within the kynurenine pathway (Breda et al., 2016). Their explanation is that the ratio between the neuroprotective kynurenic acid-producing “kat”-branch and the neurotoxic 3HK-producing “kmo”-branch (see Figure 4 in the introduction of this thesis) is shifted towards kynurenic acid (Breda et al., 2016). However, it is not shown whether this shift is due to less 3HK being produced or more kynurenic acid being synthetized or a combination of both. Could it be that the “kmo”-branch is first deactivated to make sure that kynurenic acid is still produced? It could be that an organism that is strongly dependent on kynurenic acid for proper brain function and has control mechanisms that make sure that its production is maintained as long as possible even though tryptophan that enters the kynurenine pathway is scare e.g.

231

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 231 when TDO is inhibited. If this is the case, the kynurenine pathway got a higher importance during evolution from organisms with basic neuronal structures (e.g. C. elegans) to organisms with already complex brain organization (e.g. Drosophila). This might explain the differences we see between C. elegans and flies.

In addition, when it comes to the kynurenine pathway, it seems that small amounts of tryptophan are also converted to kynurenine in a non-enzymatic manner. When we characterized the CRISPR/Cas-9 mutants, we found a constant level of kynurenine, independent of tryptophan levels and TDO-2 activity (Chapter III). This suggests that a full inhibition of this metabolic step is rather unlikely, it is only possible to strongly decrease the speed by blocking the specific enzymes.

Does this altogether mean that the work in C. elegans is not needed to better understand cellular mechanism that are finally meant to better understand human aging? Could it even be that TDO depletion has negative effects in higher organisms e.g. because kynurenic acid production is decreased? At this point, it has to be emphasized that tryptophan feeding as well as TDO depletion has similar positive effects on health- and lifespan in both organisms (C. elegans and flies) suggesting that there are proteoprotective, kynurenine pathway-independent mechanisms. Of course, this does not explain the possible consequences in mammals but so far, the depletion of TDO in different mouse models also did not result in any severe, obvious phenotypes suggesting that TDO depletion is at least not deteriorating the organism’s health (Kanai et al., 2009; Mellor et al., 2003; personal communication M. Platten).

Earlier studies in yeast flies and mice focused on KMO as the central enzyme of the kynurenine pathway. They found health-and lifespan benefits when depleting KMO (Giorgini et al., 2005; Campesan et al., 2011; Zwilling et al., 2011). We could not confirm these findings in C. elegans (Van der Goot et al., 2012 and Chapter IV). As already mentioned the depletion of TDO has proteoprotective effects in various organisms (Oxenkrug, 2010; Campesan et al., 2011; Van der Goot et al., 2012; Breda et al., 2016). Is it now better to target KMO or TDO? In nematodes at least, TDO-2 seems to be the better target: the transcriptome profiling showed that a high range of expression levels of stress response pathways are indeed altered in tdo-2 depleted animals, but these changes are strongly dependent on kmo-1 (changes are the opposite in kmo-1(del) animals, data not shown), suggesting that these alterations are linked to the kynurenine pathway. This might indicate that tdo-2 depletion is less invasive by less inducing stress response pathways and effects might be more controllable.

232

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 232 9.8. Depletion of amino acid dioxygenases affects health- and lifespan and mitochondrial respiration: are the phenotypes connected and how? All metabolic processes in a cell are strongly linked but every branch has first to be analyzed separately before possible connections can be drawn.

One major question of this study was in this context whether the health- and lifespan regulation and the control of respiration are interconnected. Unfortunately, this question could not be answered yet. As all three phenotypes showed to be independent of the above-mentioned signaling cascades (kynurenine, serotonin, translation), an interconnection is possible. Especially a connection between mitochondrial activity and lifespan control is likely as a lot of mitochondrial mutants also show lifespan-extending phenotypes (summarized in Yanos et al., 2012). It would be interesting to see whether the complex IV subunit F26E4.6 - that seems to be involved in the regulation of mitochondrial activity by TDO-2 and the other two amino acid dioxygenases - is also able to affect the regulation of health- and lifespan. Furthermore, other mitochondrial components might be identified to be part of the mitochondrial regulation by amino acid dioxygenases. Do they also influence organismal health and longevity?

If the three functions of TDO-2 are interconnected, the longevity effect would be last in the cascade. This can be concluded from assays on the dependency on the 9 transcription factor DAF-16: whereas oxygen consumption rates and motility are controlled independently of DAF-16, the longevity effect is influenced by DAF- 16 (Van der Goot et al., 2012 and appendix of this study). Thus, mitochondrial respiration control and healthspan regulation are either independent or upstream of the lifespan control.

What if TDO-2’s function on mitochondria is independent of the role in proteotoxicity? How can TDO-2 affect e.g. the aggregation process independently of the mitochondria? Analysis of protein structures revealed that tryptophan residues are common motifs that have been found to stabilize the loops between two beta strands (Santiveri and Jiménez, 2010). Beta strands are overrepresented in aggregation-prone proteins (David et al., 2010; Reis-Rodrigues et al., 2012). It might be possible that TDO-2 affects the stability of those loops by e.g. changing the availability of tryptophan. And how about modifications at those tryptophan residues as we detected them in chapter VIII? Would such a modification change the aggregation properties of a protein? It will be interesting to analyze the list of modified peptides and compare them with list of aggregation-prone proteins that have been generated by other (Walther et al., 2015).

In the context of modifications at tryptophan residues, it has to be determined whether the depletion of tdo-2 or supplementation with tryptophan change these? If yes, are

233

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 233 the tryptophan modifications responsible for the health- and lifespan-extending effects of tdo-2 depletion? If the tryptophan modifications are a consequence of translational rates and the composition and availability of tryptophan metabolites, as we suggested it in chapter VIII (co-translational theory), then it is not likely that they affect health and lifespan, just as the changes in translation (Chapter V) and composition of metabolites (Chapter IV) have been excluded for those phenotypes. However, we also see that tryptophan supplementation improves healthspan and that the mutants of the kynurenine and serotonin pathway do not at all show similar motility rates or share similar life expectations (Van der Goot et al., 2012 and chapter IV). It would be interesting to test the modifications of tryptophan residues in those mutants. Even though, it might not explain the lifespan extension of tdo-2 depleted worms, it might give clues for the shorter lifespan of some mutants. Are perhaps some modifications more severe than others? Further analysis of modifications at tryptophan residues may yield new insights in age-related protein modifications and reveal new aspects of proteostatic mechanisms.

9.9. Age-related diseases: Is the control of mitochondrial respiration by TDO-2 linked to the regulation of proteotoxic processes? TDO-2 was initially found in a screen for Parkinson’s disease-associated alpha- synuclein toxicity and has been identified as a general regulator of proteotoxic processes (Van der Goot et al., 2012). Furthermore, depletion of tdo-2 also improved the motility and lifespan of wild type animals (Van der Goot et al., 2012 and this thesis) suggesting common proteotoxic mechanisms in neurodegenerative diseases and aging.

The performed transcriptome profile of tdo-2 depleted animals was initially meant to find new candidates that would give insights in the processes that lead tothe observed proteoprotective effects (improvement of motility) in those animals. We systematically checked the candidate list for effects on healthspan (motility screen, data not shown) but could not find any consistent results. We then changed the strategy and checked for functional groups in the candidate list which brought us to the here presented findings that TDO-2 is involved in the regulation of mitochondrial activity. This raises the question whether this mitochondrial function is connected to its role in proteotoxicity. As described in detail in the introduction, mitochondrial activity is altered in a variety of age-related diseases, among them neurodegenerative disorders such as Parkinson’s, Huntington’s and Alzheimer’s disease (see introduction paragraph 4.3). A connection between mitochondrial regulation and proteotoxic control by TDO-2 is therefore hypothetically possible.

234

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 234 Unfortunately, we could not fully answer this question yet but experiments were planned and already partially prepared. We crossed for example our CRISPR-Cas9- induced deletion mutants with a neuronal Huntington’s disease model of the research group of Christian Neri (Parker et al., 2007) to initially confirm previous results that show that depletion of tdo-2 by RNAi improves the neuronal touch response of these animals (Van der Goot et al., 2012) but also to have a second disease-specific readout. It would be interesting to see whether subunit F26E4.6 of complex IV of the ETC - involved in TDO-2’s regulation of the mitochondrial respiration (Chapter VI) - also has effects on neuronal performance in this HD model.

Another idea would be to test various disease models for their mitochondrial respiration. It has been suggested that oxidative stress by the mitochondria promotes the progression of AD (Li et al., 2004; Lovell et al., 2004; Tamagno et al., 2005; Velliquette et al., 2005). And cellular models of AD show lowered respiration at the level of complex I and IV (Keil et al., 2004; Devi et al., 2006). Is a depletion in tdo-2 able to restore these defects? If this is true and the reason for proteoprotective effects, then it could be also possible that any other gene that has similar effects on respiration as TDO-2 should also improve the health state of the diseased animals. Two of these candidates would be the here presented dioxygenases CDO-1 and HPD-1 whose depletion have comparable effects on wildtype animals as TDO-2 depletion (Chapter VI). First tests suggest indeed that depletion of these two genes also improves health 9 parameters in models of Parkinson’s and Alzheimer’s disease (chapter IV and data not shown). Further experiments also with the neuronal Huntington’s disease model are planned.

It has also been suggested that toxic effects in neurodegenerative diseases are linked to increased levels of reactive oxygen species (ROS). Even though, we did

not observe a change in H2O2 levels in wildtype animals when depleting tdo-2, we cannot exclude such an effect in the disease models as well as when depleting the other dioxygenases. Measurements of ROS in various disease models also in combination with a depletion of one or more dioxygenases could possibly answer whether ROS are at least partially responsible for the proteoprotective effects in dioxygenase-depleted animals.

At least for Parkinson’s disease it has been shown that the recycling rates and cleaning-up of dysfunctional mitochondria is decreased (Malkus and Ischiropoulos, 2012). Does a depletion of tdo-2 affect these processes? The transcriptome profiling of wildtype animals showed no changes in the expression of autophagy-related genes but it might be still possible that autophagic rates are changed under disease conditions. It would be interesting to see whether tdo-2 depletion changes the ratio between functional and dysfunctional mitochondria. One possible measurement of functional mitochondria is to test for a proton gradient with specific dyes.

235

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 235 TMRM (tetramethylrhodamine-methylester) is for example a dye that changes the fluorescence spectrum based on the acidification of the environment and is commonly used for mitochondrial staining (Ishigaki et al., 2016; Monteith et al., 2013). TMRM staining is also possible in living C. elegans (personal communication S. Henau) and might help to give more insights in how tdo-2 depletion affects mitochondrial function.

Also in the context of disease-related proteotoxic effect, one open question is still whether the observed effects by tdo-2 depletion are indirectly an effect of increased tryptophan levels. Supplementation of tryptophan to a PD model improved the healthspan of these animals (Van der Goot et al., 2012). In humans however, tryptophan levels are rather constant in the brain even under tryptophan starvation conditions (personal communication M. Platten). Is it even possible to modify tryptophan conditions then to e.g. treat neurodegenerative diseases?

Attempts to feed tryptophan to other disease models (HD and AD models) and wildtype animals to check for changes in phenotypes resulted in some difficulties: Long-term feeding with tryptophan has toxic effects (data not shown). Other groups observed similar effects with higher concentration of tryptophan (Edwards et al., 2015). It is possible to still adjust the conditions (kill the bacteria that they cannot convert tryptophan, short term feeding for a certain period to check whether this is enough for long-term effects such as lifespan extension etc.) but so far we are not able to completely exclude that the increase in tryptophan might be, at least partially, responsible for the observed proteoprotective effects.

9.10. Amino acid dioxygenases in higher organisms: the translation of results The ultimate goal of the here presented research is the understanding of cellular processes that prolong human healthspan. Especially when working with rather simple model organisms such as C. elegans, one should always keep in mind whether the findings can be translated and that research findings still have to be translated to higher organisms.

When it comes to the long-lived nature of mitochondrial C. elegans mutants, Gallo and colleagues added a completely new aspect to the discussion: they have found that C. elegans possesses an alternative ATP generating pathway (glyoxylate pathway) that does not exist in humans (Gallo et al., 2011). Thus, under conditions of mitochondrial dysfunction, nematodes are able to switch to this alternative pathway whereas humans can develop severe metabolic syndromes (Wallace, 1999). They imply that the research on mitochondria in C. elegans might therefore not be representative for the human situation (Gallo et al., 2011). This alternative pathway,

236

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 236 however, cannot have such an universal impact, as ATP levels are still dramatically reduced in mitochondrial mutants, a compensatory mechanism is thus inefficient.

Second, it is important to know which molecules in the cell do control mitochondrial function, as they might be targets for drugs for mitochondrial diseases and it is of special importance to know possible side-effects of treatments. As described in chapter III, TDO is such a candidate for drug development to treat a range of diseases. If a TDO inhibitor is used for treatment, it is important to know that this treatment might also affect the cellular energy balance as mitochondrial function is simultaneously reduced.

And last, some diseases like some cancer types are linked to an uncoupled mitochondrial energy metabolism. In these conditions, the modulation of mitochondrial regulators might be a therapeutic strategy.

Nevertheless, studies on TDO-2 in C. elegans do not capture all aspects of the human variant of the pathway because nematodes have only one tryptophan-degrading dioxygenase (Yuasa and Ball, 2015). and humans have a second enzyme: IDO (Higuchi and Hayaishi, 1967, see introduction). It is therefore of high importance to also control for IDO levels in translational studies with cells or mice. It might be that IDO takes over some functions of TDO when the latter is depleted. It is possible to measure metabolite levels to control for a complete inhibition of tryptophan 9 degradation. And finally, it might be even necessary to reduce not only TDO but also IDO to see effects and draw conclusions.

Furthermore, if the combination of amino acid degradation and oxygen use leads to overlapping functions of respective enzymes, then IDO would also fall into this category: it binds oxygen and degrades tryptophan. It would be interesting to see if the depletion of IDO has similar effects as the reduction of the other amino acid dioxygenases on proteotoxicity and mitochondrial function.

9.11. Final conclusion To summarize, the goal of this thesis was to find cellular mechanisms that can explain the health- and lifespan benefits oftdo-2 depleted animals. So far, we are not able to draw any final, exclusive conclusion but we identified a new cellular player being possibly involved: the mitochondrion. TDO-2 does control mitochondrial respiration in C. elegans. Whether this affects health- and lifespan of the organism still needs to be elucidated. We further found two other amino acid dioxygenases, CDO-1 and HPD-1, that share functions with TDO-2 in the regulation of mitochondrial respiration as well as in the control of health- and lifespan in C. elegans. Again, details on regulation mechanisms or signaling pathways have not been resolved but

237

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 237 a range of assays was performed that suggest that obvious pathways - such as ROS signaling – are most probably not involved (see also Appendix). This thesis has laid the foundation for future identification of new health-and lifespan extending mechanisms based on inhibition of amino acid dioxygenases.

238

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 238 LITERATURE

• Blagosklonny, MV: Aging and immortality: quasi- some C. elegans mitochondrial mutants explained by programmes senescence and its pharmacologic activation of an alternative energy-producing pathway. inhibition. Cell Cycle 5(18), 2087-102 (2006). Mechanisms of Ageing and Development 132, 515-518 • Böttinger, L et al.: Phosphatidylethanolamine and (2011). cardiolipin differentially affect the stability of • Geng, J, Liu, A: Heme-dependent dioxygenases in mitochondrial respiratory chain supercomplexes. J tryptophan oxidation. Archives of Biochemistry and Mol Biol 423(5), 677-86 (2012). Biophysics 544, 18-26 (2014). • Brand, A, Leibfritz, D, Hamprecht, B, Dringen, R: • Giorgini, F, Guidetti, P, Nguyen, Q, Bennett, SC, Metabolism of cysteine in astroglial cells: synthesis Muchowski, PJ: A genomic screen in yeast implicates of hypotaurine and taurine. J Neurochem 71, 827-32 kynurenine 3-monooxygenase as a therapeutic target (1998). for Huntington disease. Nat Genet 37, 526-531 (2005). • Breda, C et al.: Tryptophan-2,3-dioxygenase • Hamilton, B et al.: A systematic RNAi screen for (TDO) inhibition ameliorates neurodegeneration by longevity genes in C. elegans. Genes Dev. 19(13), modulation of kynurenine pathway metabolites. Proc 1544-55 (2005). Natl Acad Sci U S A 113(19), 5435-40 (2016). • Hansen, M et al.: New genes tied to endocrine, • Campesan, S et al.: The kynurenine pathway metabolic, and dietary regulation of lifespan from a modulates neurodegeneration in a Drosophila model Caenorhabditis elegans genomic RNAi screen. PLoS of Huntington’s disease. Curr Biol 21, 961-966 (2011). Genet 1, 119-128 (2005). • Chauhan N et al.: Reassessment of the reaction • Higuchi, K, Hayaishi, O: Enzymatic formation of mechanism in the heme dioxygenases. J. Am. Chem. D-kynurenine from D-tryptophan. Arch Biochem 131, 4186-4187 (2009). Biophys 120(2), 397-403 (1967). • Curran, SP and Ruvkun, G: Lifespan regulation by • Hosokawa, Y, Matsumoto, A, Oka, J, Itakura, H, evolutionarily conserved genes essential for viability. Yamaguchi, K: Isolation and characterization of a PLoS Genet. 3(4), (2007). cDNA for rat liver cysteine dioxygenase. Biochem • Danesch, U et al.: Glucocorticoid induction of the rat Biophys Res Commun 168, 473-478 (1990). tryptophan oxygenase gene is mediated by 2 widely • Ishigaki, M et al.: STED super-resolution imaging 9 separated glucocorticoid-responsive elements. EMBO of mitochondria labeled with TMRM in living cells. J 6, 625-630 (1987). Mitochondrion 28, 79-87 (2016). • David, DC et al.: Widespread protein aggregation as • Jung, YS: Metabolism of sulfur-containing amino an inherent part of aging in C. elegans. PLoS Biol 8, acids in the liver: a link between hepatic injury and (2010). recovery. Biol Pharm Bull 38, 971-974 (2015). • Devi, L, Prabhu, BM, Galati, DF, Avadhani, NG, • Kanai, M et al.: Tryptophan 2,3-dioxygenase is a key Anandatheerthavarada, HK: Accumulation of amyloid modulator of physiological neurogenesis and anxiety- precursor protein in the mitochondria import channels related behavior in mice. Mol Brain 2, 8 (2009). of human Alzheimer’s disease brain is associated • http://www.kegg.jp/kegg-bin/highlight_pathway with mitochondrial dysfunction. The Journal of ?scale=1.0&map=map00190&keyword= Neuroscience 26(35), 9057-9068 (2006). phosphorylation; accessed 20 Oct 2016 • Dillin, A et al.: Rates of behavior and aging specified • Keil, U et al.: Amyloid beta-induced changes in nitric by mitochondrial function during development. oxide production and mitochondrial activity lead Science 298, 2398-401 (2002). to apoptosis. The Journal of Biological Chemistry • Duggan, AT, Kocha, KM, Monk, CT, Bremer, K, 279(48), 50310-50320 (2004). Moyes, CD: Coordination of cytochrome c oxidase • Kirkwood, TB: Understanding the odd science of gene expression in the remodeling of skeletal muscle. aging. Cell 120, 437-47 (2005). J Exp Biol 214, 1880-7 (2011). • Laplante, M, Sabatini, DM: mTOR signaling in • Edwards, C et al.: Mechanisms of amino acid-mediated growth control and disease. Cell 149, 274-293 (2012). lifespan extension in Caenorhabditis elegans. BMC Genetics, (2015). • Lee, SS et al.: A systematic RNAi screen identifies a critical role for mitochondria in C. elegans longevity. • Falk, MJ et al.:Metabolic pathway profiling of Nat Genet. 33(1), 40-8 (2003). mitochondrial respiratory chain mutants in C. elegans. Mol Genet Metab 93(4), 388-397 (2008). • Li, F et al.: Increased plaque burden in brains of APP mutant MnSOD heterozygous knockout mice. J. • Forner, F, Foster, LJ, Campanaro, S, Valle G, Mann Neurochem 89, 1308-1312 (2004). M: Quantitative proteomic comparison of rat mitochondria from muscle, heart, and liver. Moll Cell • López-Otin, C, Blasco, MA, Partridge, L, Serrano, M, Proteomics 5(4), 608-19 (2006). Kroemer, G: The Hallmarks of Aging. Cell 153(6), 1194-1217 (2013). • Gallo, M, Park, D, Riddle, DL: Increased longevity of

239

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 239 • Lovell, MA, Xiong, S, Xie, C, Davies, P, Markesbery, (2010). WR: Induction of hyperphosphorylated tau in primary • Satsu H, Terasawa, E, Hosokawa, Y, Shimizu, M: rat cortical neuron cultures mediated by oxidative Functional characterization and regulation of the stress and glycogen synthase kinase-3. J Alzheimers taurine transporter and cysteine dioxygenase in human Dis 6, 659-671 (2004). hepatoblastoma HepG2 cells. Biochem J 375, 441-447 • Malkus, KA, Ischiropoulos, H: Regional deficiencies (2003). in chaperone-mediated autophagy underlie alpha- • Smith, RL, De Vos, WH, de Boer, R, Manders, EM, van synuclein aggregation and neurodegeneration. der Spek, H: In vivo visualization and quantification of Neurobiol. Dis. 46, 732-744 (2012). mitochondrial morphology in C. elegans. Methods Mol • Mellor, AL et al.: Cutting edge: induced indoleamine Biol 1265, 367-77 (2015). 2,3 dioxygenase expression in dendritic cell subsets • Tamagno, E et al.: Beta-site APP cleaving enzyme up- suppresses T cell clonal expansion. J Immunol 171, regulation induced by 4-hydroxynonenal is mediated 1652-1655 (2003). by stress-activated protein kinases pathways. J • Monteith, A et al.: Imaging of mitochondrial and non- Neurochem 92, 628-636 (2005). mitochondrial responses in cultured rat hippocampal • Taylor, DW: Effects of tocopherols, methylene blue neurons exposed to micromolar concentrations of and glutathione on the manifestations of oxygen TMRM. PLoS One 8(3), (2013). poisoning in vitamin E-deficient rats.J Physiol 140(1), • Ocampo, JR et al.: Kynurenines with neuroactive 37-47 (1958). and redox properties: relevance to aging and brain • Thackray, SJ, Mowat, CG, Chapman, SK: Exploring diseases. Oxid Med Cell Longev, (2014). the mechanism of tryptophan 2,3- dioxygenase. • Oxenkrug, GF: The extended life span if Drosophila Biochemical Society Transactions 36, (2008). melanogaster eye-color (white and vermilion) mutants • Tsompo, A et al.: Tryptophan released from mother’s with impaired formation of kynurenine. J Neural milk has antioxidant properties. Pediatr Res 66(6), Transm. 117(1), 23-26 (2010). 614-8 (2009). • Parker, JA et al.: Huntingtin-interacting protein 1 • Van der Goot, AT et al.: Delaying aging and the aging- influences worm and mouse presynaptic function associated decline in protein homeostasis by inhibition and protects Caenorhabditis elegans neurons against of tryptophan degradation. Proc Natl Acad Sci U S A mutant polyglutamine toxicity. J Neurosci 27(41), 109(37), 14912-7 (2012). 11056-64 (2007). • Velliquette, RA, O’Connor, T, Vassar, R: Energy • Parsons, P: Antagonistic pleiotropy and the stress inhibition elevates beta-secretase levels and activity theory of aging. Biogerontology 8, 613-617 (2007). and is potentially amyloidogenic in APP transgenic • Rea, SL, Ventura, N, Johnson, TE: Relationship mice: Possible early events in Alzheimer’s disease between mitochondrial electron transport chain pathogenesis. J Neurosci 25, 10874-10883 (2005). dysfunction, development, and life extension in • Waldhauser, F, Ehrhart, B, Forster, E: Clinical aspects Caenorhabditis elegans. PLoS Biol 5(10), (2007). of the melatonin action: impact of development, aging • Reis-Rodrigues, P et al.: Proteomic analysis of age- and puberty, involvement of melatonin in psychiatric dependent changes in protein solubility identifies disease and importance of neuroimmunoendocrine genes that modulate lifespan. Aging Cell 11, 120-127 interactions. Experientia 49, 671-681 (1993). (2012). • Wallace, DC: Mitochondrial diseases in man and • Ruetschi, U et al.: Human 4-hydroxyphenylpyruvate mouse. Science 283(5407), 1482-8 (1999). dioxygenase: primary structure and chromosomal • Walther, DM et al.: Proteome imbalance is linked localization of the gene. Europ. J. Biochem. 213, 1081- with proteostasis decline and aggregation in aging C. 1089 (1993). elegans. Cell 161(4), 919-932 (2015). • Santiveri, CM, Jiménez, MA: Tryptophan residues: • Yanos, ME, Bennett, CF, Kaeberlein, M: Genome- Scarce in proteins but strong stabilizers of beta-hairpin wide RNAi longevity screens in Caenorhabditis peptides. Biopolymers 94(6), 779-90 (2010). elegans. Curr Genomics 13(7), 508-18 (2012). • Salminen, A, Kauppinen, A, Kaarniranta, K: • Yuasa, HJ and Ball, HJ: Efficient tryptophan- 2-oxoglutarate-dependent dioxygenases are sensors catabolizing activity is consistently conserved through of energy metabolism, oxygen availability, and iron evolution of TDO enzymes, but not IDO enzymes. J. homeostasis: potential role in the regulation of aging Exp. Zool. (Mol. Dev. Evol.), (2015). process. Cell. Mol. Life Sci 72, 3897-3914 (2015). • Zwilling, D et al.: Kynurenine 3-monooxygenase • Santos RX et al.: A synergistic dysfunction of inhibition in blood ameliorates neurodegeneration. mitochondrial fission/fusion dynamics and mitophagy Cell 145, 864-874 (2011). in Alzheimer’s disease. J. Alzheimers Dis 20, 401-412

240

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 240 9

241

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 241 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 242 Summary

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 243 The increasing lifespan of the world’s population in the last few decades, correlates with an increasing number of people with age-related pathologies such as cancer, type 2 diabetes, and neurodegenerative diseases, like Alzheimer’s and Parkinson’s disease. This development demands increasing efforts in health care, financially and in the terms of care as elderly patients need more specialized and prolonged help. In order to prevent, cure or ameliorate these diseases, it is necessary to understand the age-related biological changes that cause them. With such knowledge, ideally, one would like to modulate the aging process itself and increase overall fitness and quality of life, also at higher ages.

This thesis is aimed to better understand how the nutrient tryptophan and its metabolism influences the aging process and how it functions in the development of age-related neurodegenerative pathologies.

The thesis starts by summarizing the different hallmarks of aging and gives details on the current research on aging theories such as loss in proteostasis, mitochondrial dysfunction and ROS production, and changes in nutrient sensing and processing. Chapter I further introduces the field of amino acid metabolism in aging research and summarizes findings on tryptophan, its metabolites and tryptophan-associated enzymes such as tryptophan 2,3- dioxygenase (TDO, TDO-2 in C. elegans) which has been shown to regulate health- and lifespan in various organisms.

The model organism of choice in this thesis is C. elegans. Chapter II shows the power of this small nematode in initial screens that led to major discoveries in the field of aging and neurodegeneration.

The thesis continues by new findings in the molecule structure of TDO, the major tryptophan- degrading enzyme. Chapter III predicts that the deletion of three amino acids which are situated in a loop between two alpha-helices, influences the ability to bind the co-factor haem. We show that these three amino acids are essential for the enzyme function and present a new target for the development of TDO-inhibitory drugs.

Subsequent chapters aim to understand how TDO-2, tryptophan and its metabolism regulate health- and lifespan in C. elegans. They show that these functions are independent of major tryptophan metabolites such as kynurenine, serotonin and tryptamine (chapter IV) as well as changes in nutrient-dependent cytosolic and mitochondrial translation (chapter V). We finally present a new function of TDO- 2 in regulating mitochondrial respiration and find functional similarities between the tryptophan-degrading enzyme TDO-2 and two other amino acid-degrading dioxygenases HPD-1 and CDO-1, suggesting a general role of amino acid dioxygenases in control of mitochondrial activity, health- and lifespan in C. elegans

244

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 244 (chapter VI).

This thesis continues by presenting a new protocol that we have established to measure respiration in C. elegans, which has been instrumental to identify the role for TDO-2, HPD-1, and CDO-1 in mitochondrial respiration (chapter VII).

Finally, chapter VIII changes the focus from free tryptophan to tryptophan residues within proteins. How does the availability of tryptophan and its metabolites influence the composition of tryptophan-containing proteins and how do tryptophan modifications change with age of an organism? Preliminary results suggest that the number of tryptophan modifications as well as the diversity of these modifications increase with the organism´s age. How tryptophan availability and the activity of TDO might influence the process still needs to be elucidated.

This thesis finishes by summarizing the findings of this thesis and discussing open questions. I propose future steps to elucidate how TDO-2 controls cellular processes and how this might be linked to aging and health regulation (chapter IX).

245

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 245 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 246 Samenvatting

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 247 De toegenomen levensverwachting van de wereldbevolking in de afgelopen decennia heeft geleid tot een toename van mensen die lijden aan leeftijd gerelateerde ziekten zoals kanker, type 2 diabetes en neurodegeneratieve ziekten zoals de ziekten van Alzheimer en Parkinson.

Als gevolg van deze ontwikkelingen komt er een grotere druk op de gezondheidszorg te liggen, en zal er meer langdurige, gespecialiseerde hulp voor ouderen nodig zijn. Dientengevolge zal de financiële druk op de maatschappij toenemen. Om deze aandoeningen te kunnen voorkomen, behandelen of genezen is het belangrijk om de veroudering gerelateerde biologische veranderingen die eraan ten grondslag liggen beter te begrijpen. De verkregen kennis zou idealiter worden gebruikt om het verouderingsproces te moduleren zodat de algehele gezondheid en kwaliteit van leven worden verhoogd, ook op hogere leeftijd.

Dit proefschrift is gericht op het beter begrijpen hoe de voedingsstof tryptofaan en zijn metabolisme het verouderingsproces beïnvloeden en hoe ze kunnen bijdragen aan de ontwikkeling van leeftijd gerelateerde neurodegeneratieve ziekten.

Het proefschrift begint met het samenvatten van de verschillende kenmerken van veroudering en gaat in op de huidige visie op veroudering, zoals het verlies van proteostase, mitochondriële disfunctie en ROS-productie, alsook op veranderingen in voedingsstofdetectie en de verwerking hiervan. Hoofdstuk I gaat verder in op aminozuurmetabolisme in veroudering. Verder wordt ingegaan op tryptofaan, tryptofaanmetabolieten en tryptofaan-gerelateerde enzymen zoals tryptofaan 2,3-dioxygenase (TDO, TDO-2 in C. elegans) waarvan is aangetoond dat het een rol speelt in de gezondheids- en levensduur van verschillende organismen.

Het modelorganisme waar verder mee wordt gewerkt in dit proefschrift is C. elegans. Hoofdstuk II illustreert de kracht van dit kleine wormpje. C. elegans is een veelzijdig modelorganisme dat zich uitstekend leent om genetische experimenten in uit te voeren. Dit heeft geleid tot grote ontdekkingen op het gebied van veroudering en neurodegeneratie.

Het proefschrift gaat verder met nieuwe ontdekkingen rondom de moleculaire structuur van TDO, het belangrijkste enzym voor tryptofaanafbraak. Hoofdstuk III voorspelt dat de verwijdering van drie aminozuren, die gelegen zijn in een lus tussen twee alfahelices, de binding met de cofactor heem beïnvloed. We laten zien dat deze drie aminozuren essentieel zijn voor de enzymfunctie van TDO en een nieuw doelwit vormen voor TDO-remmende medicijnen.

In de hierop volgende hoofdstukken proberen we uit te vinden hoe TDO-2, tryptofaan en het bijbehorende metabolisme, gezondheids- en levensduur reguleren in C. elegans. Zij laten zien dat deze functies onafhankelijk zijn van de belangrijkste

248

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 248 tryptofaanmetabolieten zoals kynurenine, serotonine en tryptamine (Hoofdstuk IV), maar ook onafhankelijk van veranderingen in nutriëntafhankelijke cytosolische en mitochondriële translatie (Hoofdstuk V). Tot slot laten we een nieuwe functie van TDO-2 zien in de regulatie van mitochondriële respiratie en vinden we functionele overeenkomsten tussen het tryptofaan-afbrekende enzym TDO-2 en twee andere aminozuur-afbrekende dioxygenases, HPD-1 en CDO-1, wat een algemene rol van aminozuur-dioxygenases suggereert in mitochondriële activiteit, gezondheids- en levensduurregulatie in C. elegans (Hoofdstuk VI).

Dit proefschrift gaat verder met een nieuw protocol voor het meten van ademhaling in C. elegans, dit heeft een grote bijdrage geleverd aan de identificatie van de rol van TDO-2, HDP-1 en CDO-1 in mitochondriële ademhaling (Hoofdstuk VII).

Tot slot verlegt Hoofdstuk VIII de focus van vrije tryptofaan naar tryptofaan- residuen in eiwitten. Hoe beïnvloedt de beschikbaarheid van tryptofaan en zijn metabolieten de compositie van tryptofaan-bevattende eiwitten en hoe veranderen tryptofaan-modificatiesmet de leeftijd van een organisme? Voorlopige resultaten suggereren dat het aantal tryptofaan-modificaties en de diversiteit van deze modificaties vermeerderen met leeftijd. Hoe de beschikbaarheid van tryptofaan en de activiteit van TDO dit proces kan beïnvloeden moet nog verder worden uitgezocht.

Dit proefschrift eindigt met een samenvatting van dit proefschrift en bediscussieert de open vragen. Toekomstige stappen om uit te vinden hoe TDO-2 cellulaire processen beïnvloedt en hoe dit aan veroudering en gezondheid gerelateerd kan worden wordt A besproken in Hoofdstuk IX.

249

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 249 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 250 List of publications

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 251  Van der Goot AT, Zhu W, Vázquez-Manrique RP, Seinstra RI, Dettmer K, Michels H, Farina F, Krijnen J, Melki R, Buijsman RC, Ruiz Silva M, Thijssen KL, Kema IP, Neri C, Oefner PJ, Nollen EA. Delaying aging and the aging-associated decline in protein homeostasis by inhibition of tryptophan degradation. Proc Natl Acad Sci U S A. 2012 Sep 11; 109(37): 14912- 7.

 Sin O, Michels H, Nollen EA. Genetic screens in Caenorhabditis elegans models for neurodegenerative diseases. Biochim Biophys Acta. 2014 Oct; 1842(10): 1951-1959.

 Waschbüsch D, Michels H, Strassheim S, Ossendorf E, Kessler D, Gloeckner CJ, Barnekow A. LRRK2 transport is regulated by its novel interacting partner Rab32. PLoS One. 2014 Oct 31; 9(10): e111632.

 Koopman M, Michels H, Dancy BM, Kamble R, Mouchiroud L, Auwerx J, Nollen EAA, Houtkooper RH. A screening-based platform for the assessment of cellular respiration in Caenorhabditis elegans. Nat Protoc. 2016 Oct; 11(10): 1798-816.

 Michels H, Seinstra RI, Uitdehaag J, Koopman M, Van Faassen M, Martineau CN, Kema IP, Buijsman R, Nollen EAA. Identification of an evolutionary conserved structural loop that is required for the enzymatic and biological function of tryptophan 2,3-dioxygenase. Sci Rep. 2016 Dec 20;6:39199.

 Michels H, Koopman M, Seinstra RI, Van Faassen M, De Jong M, Ackermann T, Lentferink DH, Nothdurft L, Martineau CN, Calkhoven CF, Guryev V, Kema IP, Nollen EAA. Transcriptome profiling in C. elegans reveals that TDO and other amino acid dioxygenases regulate mitochondrial capacity and healthspan. Manuscript in preparation

 Visscher M, De Henau S, Wildschut M, Van Es R, Dhondt I, Michels H, Kemmeren P, Nollen EAA, Braeckmann BP, Burgering B, Vos H, Dansen T. Proteome-wide changes in protein turnover rates in C. elegans models of longevity and age-related disease. Cell Rep. 2016 Sep 13;16(11):3041-51.

252

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 252  Jansen IE, Ye H, Heetveld S, Lechler MC, Michels H, Seinstra RI, Lubbe SJ, Drouet V, Lesage S, Majounie E, Gibbs JR, Nalls MA, Ryten M, Botia JA, Vandrovcova J, Simon-Sanchez J, Castillo-Lizardo M, Rizzu P, Blauwendraat C, Chouhan AK, Li Y, Yogi P, Amin N, van Duijn CM; International Parkinson’s Disease Genetics Consortium (IPGDC), Morris HR, Brice A, Singleton AB, David DC, Nollen EA, Jain S, Shulman JM, Heutink P. Discovery and functional prioritization of Parkinson’s disease candidate genes from large-scale whole exome sequencing. Genome Biol. 2017 Jan 30; 18(1):22.

PATENTS  Nollen, EAA, Michels H. Compositions and methods of prolonging the lifespan of a subject.

Application No./ Patent No.: 16172714.4-1466

253

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 253 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 254 Appendix

Helen Michels1, Renée Seinstra1, Jens Seidel1, Arwen W. Gao2

1European Research Institute for the Biology of Aging, University of Groningen, University Medical Centre Groningen, Laboratory of Molecular Neurobiology of Aging, The Netherlands 2Laboratory Genetic Metabolic Diseases, Academic Medical Center, Amsterdam, the Netherlands.

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 255 A MPC-1 PDHA-1

MPC-2 PDHB-1 glucose pyruvate pyruvate acetyl-CoA PYK-1 MCT-4 C30H6.7

DLD-1

B motility/ health span glucose feeding WT tdo-2 (del) WT tdo-2 (del) WT tdo-2 (del) *** 60 *** *** 70 *** *** 70 * *** *** *** 60 50 *** *** 60 50 50 40 * 40 40 30 30 30 20 20 20 body bends/ 30 sec 30 bends/ body body bends/ 30 sec 30 bends/ body

10 10 sec 30 bends/ body 10 0 0 0 0% 0,5% 1% RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi -1 -1 -1 -1 -2 -4 ctr ctr dld mct mpc mpc pdha pdhb C30H6.7

C maximum mitochondrial respiration

glucose feeding * WT tdo-2 (del) * 40 40 120 * *** *** * *** 100 *** 30 30 *** 80 *** 60 20 20 OCR/ worm OCR/

40 worm OCR/

relative OCR 10 10 20 0 0 0 1% 1% 0% 0% RNAi RNAi RNAi RNAi RNAi RNAi RNAi 0,5% 0,5% RNAi -2 -2 -1 -2 -4 ctr ctr ctr tdo tdo WT mct (del) mpc mpc -2 WT (del) tdo -1 pyk

Figure A1. TDO-2 mediated regulation of mitochondria and healthspan is not via glycolysis and pyruvate transport (A) Schematic model of glycolysis and pyruvate transport. (B) Motility/ helathspan.(C) Maximum mitochondrial capacity on day 4 of adulthood. OCR= oxygen consumption rate. Statistics for all panels: * <0.05, *** <0.001. Error bars = SEM.

256

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 256 A +O2 acetyl-CoA

glucose pyruvate -O2

lactate LDH-1

B C maximum mitochondrial respiration

WT WT tdo-2 (del) tdo-2 (del) *** 70 *** 40 *** 60 *** 30 50 40 20 30

20 worm OCR/ 10

body bends/ 30 sec 30 bends/ body 10 0 0 RNAi RNAi RNAi RNAi -1 -1 ctr ctr ldh ldh

Figure A2. TDO-2 mediated regulation of respiration and healthspan is independent of anaerobic lactate synthesis (A) Schematic model of lactate synthesis. (B) Motility/ healthspan. (C) Maximum mitochondrial capacity on day 4 of adulthood. OCR= oxygen consumption rate. Statistics for all panels: *** <0.001. Error bars = SEM. A

257

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 257 A

ASC-2 CPT-2 ECH1.2 KAT-1

fatty acid acyl-CoA ACDH-12 acetyl-CoA

HACD-1

B WT tdo-2 (del) C *** 60 *** *** 100 WT/ ctr RNAi ** WT/ tdo-2 RNAi 50 80 acs-2 (del)/ ctr RNAi 40 60 acs-2 (del)/ tdo-2 RNAi 30 40 20

percent survival 20

body bends/ 30 sec 30 bends/ body 10 0 0 0 10 20 30 40 days of adulthood RNAi RNAi RNAi RNAi -2 12 ctr - cpt ech1.2 acdh 100 WT/ ctr RNAi WT/ tdo-2 RNAi *** 80 60 *** *** kat-1 (del)/ ctr RNAi kat-1 (del)/ tdo-2 RNAi 50 60

40 40

30 percent survival 20 20 0 10 body bends/ 30 sec 30 bends/ body 0 10 20 30 40 0 days of adulthood RNAi RNAi RNAi RNAi RNAi RNAi -2 -2 -2 ctr ctr ctr tdo tdo tdo WT (del) (del) -2 -1 kat acs

Figure A3. TDO-2 regulates health- and lifespan independently of mitochondrial beta oxidation (A) Schematic model of mitochondrial beta oxidation. (B) Motility/ healthspan. (C) Survival curves. Statistics for all panels: ** <0.01 ***, <0.001. Error bars = SEM.

258

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 258 A NADH CTS-1 electron acetyl-CoA transport IDH-2 chain FADH2

B C WT maximum tdo-2 (del) mitochondrial respiration 80 *** *** 60 50 70 *** *** 50 * 60 40 50 40 *** 30 40 30 30 20 20 20 worm OCR/ 10

body bends/ 30 sec 30 bends/ body 10

10 sec 30 bends/ body 0 0 0 RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi -2 -2 -1 -2 -2 ctr ctr ctr ctr ctr cts tdo tdo tdo tdo WT WT (del) (del) -2 -2 idh idh D 100 WT/ ctr RNAi tdo-2 (del)/ ctr RNAi 80 WT/ cts-1 RNAi 60 tdo-2 (del)/ cts-1 RNAi

percent survival 20 A 0 0 10 20 30 40 days of adulthood

Figure A4. TDO-2 regulates mitochondrial respiration, health- and lifespan independently of TCA cycle (A) Schematic model TCA (citric acid cycle) cycle. (B) Motility/ healthspan. (C) Maximum mitochondrial capacity on day 4 of adulthood. OCR= oxygen consumption rate. (D) Survival curves. (F). Statistics for all panels: * <0.05, *** <0.001. Error bars = SEM.

259

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 259 A Trp

TDO-2

AMFD-1

KMO-1

HAAO-1

NAD+ ATP

Figure A5. Animals with depleted enzymes of the kynurenine pathway do not show lower ATP levels as tdo-2 depleted animals (A) Relative ATP levels of mutants of the kynurenine pathway

maximum maximum A mitochondrial B mitochondrial capacity capacity ctr RNAi tdo-2 RNAi 40 35 30 30

25 20 20 OCR/ worm OCR/ 10 15 ns OCR/ worm OCR/ 10 0 5 RNAi RNAi RNAi RNAi -2 0 -2 ctr WT tdo-2 ctr tdo tdo (del) WT (del) 16 - daf

Figure A6. TDO-2 regulates mitochondrial capacity independently of the FOXO transcription factor DAF-16 (A) Maximum mitochondrial capacity on day 4 of adulthood of tdo-2(del) and RNAi-depleted animals (B) Maximum mitochondrial capacity on day 4 of adulthood. Statistics for all panels: ** <0.01, *** <0.001. Error bars = SEM.

260

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 260 A L1 L4 D1 D2 D3 D4 D7 D8 D11

B C

D E F B C D E F RNAi from RNAi from RNAi from L4 RNAi from day 1 RNAi from day 4 day 2 day 3 ns 80 40 100 ns 40 30 * * 80 60 30 30 ns *** 20 60 *** 40 20 20 * 40 10 20 10 10 20 body bends/ 30 sec 30 bends/ body body bends/ 30 sec 30 bends/ body sec 30 bends/ body sec 30 bends/ body body bends/ 30 sec 30 bends/ body 0 0 0 0 0 RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi -2 -2 -2 -2 -2 -2 -2 -2 ctr ctr ctr ctr ctr ctr ctr ctr tdo tdo tdo tdo tdo tdo tdo tdo 2 days of RNAi 3 days of RNAi 4 days of RNAi 6 days of RNAi 6 days of RNAi 4 days of RNAi 5 days of RNAi 7 days of RNAi

Figure A7. Depletion of tdo-2 at later lifespan stages still leads to an increase in healthspan (A) Experimental overview; arrows indicate start of RNAi treatment and timepoints for motility scoring. (B-F) Motility/ healthspan; start of RNAi treatments and scoring intervals differ. Statistics for all panels: ns= not significant; *<0,05, ***<0.001; Error bars = SEM. A

A day 4 of adulthood day 8 of adulthood B maximum mitochondrial capacity 70 70 40 60 60 50 50 30 40 40 20 30 30

20 20 worm OCR/ 10 body bends/ 30 sec 30 bends/ body body bends/ 30 sec 30 bends/ body 10 10 0 0 0 RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi RNAi 1 1 1 1 1 1 1 1 1 1 1 1 -2 -2 -2 1 1 1 1 1 1 -2 -2 -2 -2 -2 -2 ------ctr ctr ctr ctr ctr ctr ctr ctr ctr tdo tdo tdo tdo tdo tdo tdo tdo tdo hpd hpd hpd cdo hpd hpd hpd cdo cdo hpd hpd hpd cdo cdo cdo cdo cdo cdo

WT tdo-2 hpd-1 WT tdo-2 hpd-1 WT tdo-2 hpd-1 (del) (del) (del) (del) (del) (del)

Figure A8. Amino acid dioxygenases regulate healthspan and mitochondrial capacity (A) Motility/ healthspan when depleting one or two amino acid dioxygenases at the same time. (B) Maximum mitochondrial capacity on day 4 of adulthood when depleting one or two amino acid dioxygenases at the same time. Error bars in all panels = SEM

261

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 261 A WT 10mM Trp 20mM Trp 25

OCR/ worm OCR/ 10 FCCP

0 capacity maximum respiration mitochondrial mitochondrial

Figure A9. Tryptophan supplementation does not affect basal mitochondrial respiration but reduces mitochondrial capacity (A) Oxygen consumption rates on day 4 of adulthood. Supplementation with tryptophan from larval stage 4 onwards.

900

800 ctr RNAi/ sample 1

700 ctr RNAi/ sample 2

600 tdo-2 RNAi/ sample 1

500 tdo-2 RNAi/ sample 2

400

nmol / mg of protein 300

200

100

0 lysine valine serine proline glycine leucine alanine tyrosine arginine ornithine glutamine isoleucine tryptophan asparagine methionine aspartic acid glutamic acid phenylalanine

Figure A10. Amino acid profile of control and tdo-2 depleted animals Amino acid levels of control and tdo-2 depleted animals. Isolation of total protein at day 4 of adulthood. RNAi from L4. Samples from two independent experiments.

262

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 262 A

263

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 263 513375-L-bw-michels Processed on: 29-8-2017 PDF page: 264 Acknowledgements

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 265 Looking back at this thesis, we prepared about 200.000 agar plates, preparing more than 5700 liters of medium and about 3000 liters of M9 buffer, used 500 liters LB medium to grow bacteria, touched more than 50.000 times the nose of the nematodes with an eyelash, picked about 30.000 animals and counted their body bends and handled in total 160 million animals. Yes, high numbers and I am talking about “we” as this work could not have done by one person. I am really thankful for all the people that supported my work and would like to mention a few of them.

Dear Ellen, thank you very much for giving me this wonderful, challenging project. Thank you for your guidance in the last years, for your constant optimism and your talent to motivate everybody. I usually left our discussion rounds with the high motivation to continue and to follow up even though I sometimes had the feeling of not moving forward before. Thank you also for sharing your private moments and letting us know your wonderful husband and children and showing me that it is possible to follow a career and having a family. Thank you for your patience and your positive way of pushing me, discussing every tiny result, listening to my thoughts and for all the time you spent on reading my texts. The TDO-2 project did not always follow the easiest way and I am grateful that you gave me the opportunity to try a lot of things, new methods and to contact people around the world to get help and to travel to Tokio, France, Berlin, the USA and Spain to exchange with other scientists. I wish you all the best, spectacular results and a lot of success in the coming years with the research but also great moments with your family.

Dear Ido, thank you very much to take the role as my second promotor. I really enjoyed our meetings, having your comments and different type of view on the project. I know that you have your own big group and obligations everywhere but you always took the time to join, discuss and think of the project.

Dear Cor, I would also like to thank you as my third unofficial promotor. Thank you for your comments, help and for your relaxed look at everything. I wish you and your group a lot of great findings in the future and a relaxed time with Christine and your children.

I would also like to thank the reading committee, Prof. Dr. Rik Korswagen, Prof. Dr. Babara Bakker and Prof. Dr. Rainer Bischoff to take the time and effort to critically evaluate my thesis and discuss it with me.

Thank you as well to all the people from other groups and universities who joined the TDO-2 project and helped me with my findings: There are Peter Oefner, Jörg Reinders and Nadine Nürnberger from Regensburg who helped me with the proteomic approaches. I am really curious what the aging profile will result in! There are Christian Neri and Franscesca Farina from Paris. Thank you for showing me

266

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 266 your lab and having me in Paris. Your strains are not forgotten, I promise! Thank you as well to Tobias Dansen, Marieke Visscher and Sascha de Henau from Utrecht for the work on the protein turn over profiles. And thank you toRiekelt Houtkooper from Amsterdam for your interest in my work, the time for discussions and Arwen Gao for the amino acid profiles of “my animals” and thank youIliana Chatzispyrou for the very nice company in Spain.

Of course, there are also a lot of people from the University of Groningen and ERIBA. Thank you Victor for your patience to introduce a naïve biologist to the world of informatics and working on the RNA profile with me. I think, this was the key experiment that pushed my project finally giving us an idea of how to proceed. And thank you Tristan for your enthusiasm and your quick work on all the big data sets. Thank you Cor and especially Tobias for helping us to develop the respiration studies with the Seahorse analyzer and the discussions on mitochondrial biology. And thank you Ido and Martijn for the numerous metabolite measurements and your constant input.

And finally thank you toYinfai . You had a big part of all the work that I mentioned at the beginning: preparing all the medium, pouring thousands of plates and supporting me and the others in all the things that are extremely time-consuming. I also enjoyed our private talks in the worm lab, thinking about the future and what to expect of your life. Thank you also for always having an ear for everything and helping me with my moving back to Germany. You are welcome to visit me here and I hope to A see you soon again!

I wish all of you the very best for your research and your professional and private careers.

And finally, the TDO-2 team from the Nollen lab!

Thank you Annemieke to introduce me to the project, your patience to explain everything to everyone (“Yoda”) and your will to share the project with me which was your project before and which you started mainly on your own. I hope, you are still feeling well in the USA and everything developed as you have imagined it.

Thank you to my wonderful Renée. You helped me so much in the last years with all my experiments and I could always count on you. For me, it was perfect team work and it made all the long hours behind the microscope easier and nicer when we shared them and could chat together. Thank you for being a friend in the last years! For helping me with the Dutch administration, being there for my moving and being generous in all situations. I loved our private 30 minutes in the morning when we were the first ones in the lab and I always enjoyed our lunches. Thank you as well for the moments outside of ERIBA and our “Nederlandse Dinsdag”. I know, we

267

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 267 did not do it consequently.  I was thinking to write you these lines in Dutch but I would like you to not struggle with all my little mistakes. Thank you also for sharing the love to books with me and introduce me to Dutch books. I enjoyed our little club together with Gertrud, sharing books, new stories and talk about them. You are always welcome to visit me in Münster and I hope to still see you a lot in the future.

And of course thanks to my students Dennis, Laura, Laas, Mandy and Jens. I enjoyed having you around me, teaching you, seeing your skills being all so different and trying to help you with the tasks that were not so easy for you and it is great still hearing from you.

Dear Dennis, I wish you a great PhD and all the best for your future. I hope you stay as relaxed and “normal” (this is meant as a compliment!) as you are, having your career in mind but also always knowing that family and friends are the things that matter in life.

Liebe Laura, ich wünsche dir eine tolle und erfolgreiche Zeit in Groningen. Ich freue mich immer, wenn du dich bei mir meldest und wünsche dir eine tolle Zukunft mit deinem Mann und all deinen Tieren. Ich habe selten so einen disziplinierten, zielstrebigen Menschen erlebt, der aber auch alles Lustige im Leben mitnimmt und mit dem man gerne zusammen ist. Vielleicht sehen wir uns ja irgendwann mal wieder.

Dear Mandy, the “great Mandy”! I told you once that I was a bit skeptical when I first met you. But I also told you that this completely changed when we worked together. I am still sorry that you started your internship by pouring thousands of plates and that you were thrown into the craziness of C. elegans work directly. HTTW! However, you have chosen a PhD in our group afterwards, still working with the little worms so it has not been as repulsive as I thought. Thank you for your help with the TDO-2 project, thank you for taking the time and effort to establish the Seahorse respiration studies and producing so many data of which I could always be sure that they were thought through, performed perfectly and seen critically before sharing them. Thank you for your constant good humor and infecting everybody with it. I wish you all the best for your future being sure that nobody has to worry about it! 

Und lieber Jens, vielen Dank auch dir für deine Hingabe an dieses Projekt. Ich weiβ, dass du in dieser Zeit wenig geschlafen hast und tausend Ideen hattest, immer enthusiastisch warst und es aber mit Sicherheit nicht die einfachsten Monate für dich waren. Ich wünsche auch dir ganz viel Erfolg in Köln mit deiner Doktorarbeit.

Besides to all the people that actively worked on the TDO-2 project, there are always people next to it, supporting you, giving the right hint at the right moment, keeping your mind free and just being there if you need somebody to talk, a moment to shut off and to relax.

268

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 268 In this context, I would like to thank the rest of the Nollen group for open discussions, nice free time activities and funny Sinterklaas parties. Thank you Esther, Ale, Olga, Wytse, Anita, Kees, Freek and Mats and all the students joining and leaving. There are not many left of the original group but we had a good time together and I wish you all the best for your future!

Merci beaucoup à toi, Céline, pour toujours avoir du temps pour écouter, aider dans tous les situations et pour être la partie française de ma vie aux Pays-bas. Je pense que tu sais comment cela est important pour moi. Je te souhaite un temps avec succès mais aussi agréable à Bruxelles et j’espère qu’on se voit un jour, peut-être en Belgique, peut-être aux Pays-bas, peut-être en Allemagne, peut-être en France, on ne sait jamais.

Thank you to all the people of the ground floor lab for the ambience in the lab. Good luck with all coming experiments!

Thank you, Gertrud, for your positivity and your open mind towards everybody. It was nice to have you as a member of the “early bird” club and sharing books and ideas with you. I hope, your renovations are going the way you expected it and we see us soon in Groningen or in Münster.

Thanks a lot to the people from the management and administration: Especially thanks to Annet and thanks to Sylvia for taking care of so many things and thanks to Craig. A Thanks a lot to some other people from ERIBA having a good time besides science: there are the members of the PhD committee, including Magda, Sarah, Jacob, Niek, David, Clémence, Aaron and Tristan as well as Sandra, Edyta, Mathilde, Anton, Stijn, Christine, Tobias and others. Some of you moved to the next institutes all over the world already, others are still there… I wish you a joyful life and success!

And finally, there are family and friends. They are not directly involved in science but make an important part of your life, finding a balance to finish such a thing like a PhD. So, I like to mention a few of them.

Liebe Kira, liebe Nicole, wir haben uns zwar selten gesehen in den letzten Jahren, trotzdem ist es schön zu wissen, dass wir uns noch genauso verstehen wie vor mehr als 5 Jahren, wo wir alle unserem Master entgegen gefiebert haben. Ich bin nun die dritte im Bunde: Drei Doktortitel, zwei Hochzeiten und ein Baby… klingt fast wie ein Filmtitel!  Ich freue mich, dass wir wieder näher zusammen wohnen und uns jetzt wieder mehr sehen.

Liebe Catti, natürlich gehörst du in solch eine Auflistung. Vielen Dank, dass du meine Freundin bist und das nun schon seit über 15 Jahren! Vielen Dank, dass du die in den

269

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 269 letzten Jahren immer Zeit genommen hast, insbesondere, wenn ich deine Planung durch einen Spontanbesuch in Münster mal wieder aus dem Konzept gebracht habe. Ich glaube, ich habe in den letzten Jahren besonders an solchen Wochenenden extrem wenig Schlaf bekommen, weil es einfach schön und gemütlich war und ist, mit dir und Maik rumzuhängen, zu quatschen und die Zeit zu vergessen. Manchmal frage ich mich, wie es sein kann, dass uns irgendwie nie der Gesprächsstoff ausgeht… Es ist toll Freunde zu haben, auf die man hunderprozentig zählen kann und mit denen die Zeit so kurzweilig ist! Schön, dass wie nun wieder näher zusammen wohnen und diese Momente nicht mehr so selten sind.

Und zum Schluss meine liebe Familie. Vielen Dank an euch, dass ich in eurer Gegenwart einfach nur ich sein kann und manchmal auch die Wissenschaft vergessen durfte. Ich weiβ, es hat doch alles länger gedauert als geplant, aber nun ist es geschafft. Viele von euch haben mitgefiebert und immer Interesse gezeigt, auch wenn meine Arbeit vielleicht nicht von allen so leicht zu verstehen war. Es ist schön wieder bei euch in der Nähe zu wohnen und dass wir uns so häufig sehen. Und es ist wunderbar, Tante zu sein und meinen kleinen Kilian aufwachsen zu sehen.

Liebe Mama, lieber Papa, ich hoffe, es ist euch bewusst, auch, wenn wir es euch nicht immer sagen: Ihr seid groβartige Eltern! Wir werden von euch in jeder Lebenslage bedingungslos unterstützt und ihr seid immer für uns da. Ihr habt uns immer vermittelt, dass im Leben alles möglich ist und vieles für uns möglich gemacht. Und gleichzeitig werdet ihr auch eurer Rolle gerecht, kritisch Themen anzusprechen, die man nicht immer so gerne hört. Ich habe nun 25 Jahre Ausbildung hinter mir und hatte nie das Gefühl, dass ich dazu nicht in der Lage bin. Ihr gebt einen die Sicherheit dazu. Vielen Dank für all die Unterstützung und die Geduld und ich freue mich nun, dass wir so eng zusammen wohnen, häufiger sehen und ich mit dir, Papa, nun jeden Tag zusammen arbeite und ich mit deiner und eurer Hilfe nun einen neuen Abschnitt in meinem Leben begonnen habe.

Thanks a lot, vielen Dank, hartelijk bedankt, merçi beaucoup!

270

513375-L-bw-michels Processed on: 29-8-2017 PDF page: 270