TRANSIENT REPROGRAMMING FOR MULTIFACETED REVERSAL OF AGING

PHENOTYPES

A DISSERTATION

SUBMITTED TO THE DEPARTMENT OF APPLIED PHYSICS

AND THE COMMITTEE ON GRADUATE STUDIES

OF STANFORD UNIVERSITY

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS

FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

TAPASH SARKAR

MAY 2019

© 2019 by Tapash Jay Sarkar. All Rights Reserved. Re-distributed by Stanford University under license with the author.

This dissertation is online at: http://purl.stanford.edu/vs728sz4833

ii I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Vittorio Sebastiano, Primary Adviser

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Andrew Spakowitz, Co-Adviser

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Vinit Mahajan

Approved for the Stanford University Committee on Graduate Studies. Patricia J. Gumport, Vice Provost for Graduate Education

This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file in University Archives.

iii iv

Abstract

Though aging is generally associated with tissue and organ dysfunction, these can be considered the emergent consequences of fundamental transitions in the state of cellular physiology. These transitions have multiple manifestations at different levels of cellular architecture and function but the central regulator of these transitions is the epigenome, the most upstream dynamic regulator of gene expression. Reproduction is the only general phenomena in nature where the age of

(parental) cells is truly reset - to produce an embryo and ultimately an age 0 offspring - and core to this process is a dramatic reprogramming of the epigenome.

Here we present a technology that captures part of this age reset mechanism but using a transient reprogramming - to drive more youthful phenotypes but without the full reset back to an embryo. This reprogramming technology is distinct from previous anti-aging/pro-longevity interventions as instead of just modulating a few identified aging pathways, reprogramming engages a global and balanced state transition, in the case of reproduction, or state perturbation, in our transient approach, which we show leads to a multifaceted age reversal effect at the DNA, metabolic, whole cell and local environmental levels. We further discuss the emergent tissue and organ level benefits when transplanted with cells undergoing this treatment.

v

Acknowledgments

Thanks to our collaborators in the …

Sebastiano Lab, Stanford University

Vittorio Sebastiano

Jens Durruthy-Durruthy

Horvath Lab, University of California, Los Angeles

Steve Horvath

Bhutani Lab, Stanford University

Nidhi Bhutani

Shravani Mukherjee

OneSkin Technologies

Carolina Oliveira

Alessandra Zonari

Chu Lab, Stanford University

Constance Chu

Eleonora Migliore

Erika Leonardi

vi

Rando Lab, Stanford University

Tom Rando

Marco Quarta

Alex Colville

Patrick Paine

Linda Doan

Mahajan Lab, Stanford University

Vinit Mahajan

Katherine Wert

Palmer Lab, Stanford Univeristy

Theo Palmer

Aditya Asokan

And to the services and resources provided by …

ESI Biotechnologies

Nakuchi Lab, Stanford University

Girihlet Inc

Human Immune Monitoring Facility, Stanford University

vii

As well as to our funding sources

National Science Foundation

Glenn Foundation for Medical Research

American Federation of Aging Research

viii

Table of Contents

1) Introduction

1.1) Motivation 1

1.2) Background 2

1.3) Approach 5

2) Technology

2.1) Background 6

2.2) Results 8

2.3) Discussion 13

2.4) Methods 15

3) DNA Level

3.1) Background 17

3.2) Results 21

3.3) Discussion 32

3.4) Methods 34

4) Metabolic Level

4.1) Background 38

4.2) Results 42

4.3) Discussion 50

4.4) Methods 51

5) Niche Level

5.1) Background 54

5.2) Results 57 ix

5.3) Discussion 63

5.4) Methods 65

6) Metabolic Level

6.1) Background 67

6.2) Results 69

6.3) Discussion 73

6.4) Methods 75

7) Conclusion

7.1) Concepts 78

7.2) Future Directions 79

8) Supplemental Figures 82

9) References 93

x

Figures

11 Figure 1 Verification of Results from In Vitro Sysnthesis of Reprogramming

12 Figure 2 Verification of identity retention and lack of transformation in treated cells and derived tissue by histology.

13 Figure 3 Verification of identity retention and lack of transformation in treated cells and derived tissue by gene expression

26 Figure 1 Combined single cell distritubtions of epigenetic, laminar and DNA damage markers.

27 Figure 2 Combined single cell distritubtions for treated and untreated cells from

HGPS patients.

29 Figure 3 Transcriptomic landscape in aging subspace.

30 Figure 4 Transcriptomic landscape of treatment

32 Figure 5 Methylation Clock analysis

46 Figure 6 Combined single cell distritubtions and bulk redout of nutrient sensing, mitochondria activity and protien clearence

49 Figure 7 Combined single cell distritubtions and bulk redout of redox and ECM metabolism parameters.

59 Figure 8 Evaluation of senescence parameters in aged monolayer endothelial cells and fibroblast-keratinocyte constructs

61 Figure 9 Evaluation of senescence parameters in fat pat derived mesenchymal stem cells. xi

62 Figure 10 Intracellular expression and extracellular secretion of pro-inflammatory factors.

71 Figure 11 First injury response after transplanting in treated, aged and young mouse

MuSC

72 Figure 12 Second injury (mouse) and human MuSC transplant results.

82 Supplementary Figure 1 Demonstration of in vivo, in situ transfection

83 Supplementary Figure 2 Comparison of Shorter vs Extended Treamtment

(Fibroblast)

85 Supplementary Figure 3 Comparison of Shorter vs Extended Treamtment

(Endothelial Cell)

86 Supplementary Figure 4 Effects of control transfections (Fibroblasts).

88 Supplementary Figure 5 Effects of control transfections (Endothelial Cells).

89 Supplementary Figure 6 Time course study of retentiation of rejuvenative benefits

(Fibroblasts).

91 Supplementary Figure 7 Time course study of retentiation of rejuvenative benefits

(Endothelial Cells).

92 Supplementary Figure 8 Example single patient distribution comparison

1

1) Introduction

1.1) Motivation

Progression into middle and old age is marred by multitude of diseases and dysfunctions like arthritis, cardiovascular disease, muscle atrophy and skin aging [Jaul and Barron, 2017]. Research to combat these conditions has been largely specialized to develop solutions for each specific “downstream” conditions. However, intervention into physiological process of aging itself, as the “upstream” driver of all these conditions, has remained relatively unexplored compared to the dramatic advances in medicine over the last few decades [Finkel, 2005]. Though admittedly the lack of scientific consensus on the defining biomarker of aging presents a key obstacle to this day, a more general impediment was the long-standing dogma on the irreversibility of age, irreversible like time itself as an almost entropic concept [Newman et al., 2016]. Though there existed previous work studying interventions to slow or delay aging, the first demonstration of age reversal, or rejuvenation, came around the turn of the new millennium [Mahmoudi and Brunet, 2012]. In their landmark heterochronic parabiosis study, merging the vasculature of mice of different ages together (2-3 month young with 19-26 month old), the Rando and Weismann labs at Stanford demonstrated that the youthful blood from the young parabiont was sufficient to reverse multitude of aging phenotypes in the liver and muscle of the old parabiont [Conboy et al., 2005]. This approach demonstrated that extracellular factors, cytokines, growth factors, exosomes etc., were key to establishing and progressing aging and likewise youth. Since then the list of tissues that have been shown to exhibit a similar youthful reversal by this approach has grown to also include brain, bone, pancreas and heart [Castellano et al., 2015]. This work introduced a greater 2 aspiration for the field of ageing from a just mitigating age related dysfunction and extending lifespan, to the possibility of rejuvenation and health restoration. Since then, another contender has also risen in senolytic treatments, which we will discuss in further detail in later chapters but essentially presents another way of promoting a youthful environment by clearing the cells which most emphatically secrete aging paracrine factors [Jeon et al., 2017]. The work of this thesis highlights a third possible contender that is in many ways the complement of these approaches as it introduces a method to reverse aging in a cell autonomous fashion rather than by affecting the extracellular environment. It is also, as we will see, the one proposed rejuvenation method that is based on a rejuvenation mechanism already routinely employed in nature.

1.2) Background

When considering exceptions to one-way progression of aging in nature, one is typically predisposed to think of the obscure corner cases, specifically organisms considered to be “biologically immortal”. The select organisms, specific species like the jellyfish Turritopsis dohrnii or the general hydra genus, have evolved the ability to reset the age related dysfunction of their cells by altering their epigenetic identity, through differentiation of stem cells to somatic cells and trans-differentiation from one type of somatic cell to another [King and Newmark, 2012 ], [Piraino et al., 1996]. This process can be safely and frequently employed in these cases because the organisms are much simpler and there is less structure and organization needed to be re-established [Petralia et al., 2014]. In more complex organisms trans-differentiation does not play a functional role and stem cells, though present, have much more limited potential and activity

3 specifically because errors in cell type, function or growth are more severe when the tissues are more diverse and complex; cancer is the most blatant example of such a failure mode. However, this same mechanism of epigenetic resetting for rejuvenation does in fact occur for more advanced animals but in very specific circumstance and is crucial to the survival of each species. Instead of resetting all the cells in a complex organisms to try to rejuvenate it at the risk of disrupted or aberrant tissue function, evolution has chosen to reserve this process to a select and isolated group of cells that when epigenetically reset yield young tissue for not just one but every organ in the body.

The caveat is that body is not the original organism but its progeny, ie the process we are alluding to is sexual reproduction: the reset of aged human germ cells of the reproductive system to the pluripotent embryo from which every tissue for a new, age 0 organism is produced [O’Neil, 2015]. A key aside, one could also hypothesize that the germ cells, sperm and egg, themselves somehow don’t age at all, however there have been multiple studies showing evidence to the contrary [Eichenlaub-Ritter et al., 2011], [Sharma et al.,

2015]. Even if germ cells do age at slower rate than somatic cells, passing down even the smallest amount of age to each offspring would lead accumulated dysfunction over generations and ultimately yield nonviable offspring and extinction.

We will discuss epigenetics in greater detail in a later chapter, but from an rejuvenation science standpoint the epigenome presents linchpin target both as 1) the most upstream level of cellular information processing, including signals for aging, where regulation of the gene expression and cell function is programmed and 2) as established architecture which is be methodically “reprogrammed” in nature for broader, different or healthier functionality. (The specific terminology of reprogramming has

4 come to denote the specific transition to an embryonic like state, rather than transitioning to other differentiated states.) Deconstructing this process, researchers first replaced the sperm cell contribution, ie its nucleus, with that of a somatic cell in a technique called somatic cell nuclear transfer (SCNT) [Willadsen, 1986]. Though initially there were some difficulties with the procedure and publicized examples like Dolly the sheep in the mid-1990’s died relatively young, advancements in the technology have established the method to produce viable offspring with no evidence of inherited, advanced age

[Burgstaller and Brem, 2017]. This proved that the maternal factors from the originally oocyte were sufficient to drive both the reprograming and rejuvenation. A decade after

Dolly, Shinya Yamanaka’s lab at Kyoto University achieved the most thorough deconstruction to date – reducing the reprogramming environment to just a handful of transcription factors sufficient to drive fibroblasts to embryonic-like induced pluripotent stem cells (iPSCs), though albeit with a much lower efficiency [Takahashi et al., 2007].

The efficacy of this technology would be expanded upon in the following years - generalizing the starting cell type, demonstrating differentiation capacity, and, most pertinently, establishing the youthful properties of iPSCs when derived from aged, even centenarian donor cells [Khazaei et al., 2015], [Williams et al., 2012], [Yagi et al., 2012].

This last point was established both by batteries of cellular assays in vitro as well as blastocyst complementation experiments to generate youthful tissue in vivo [Kang et al.,

2009], [Studer et al., 2015].

5

1.3) Approach

Our focus the rejuvenation aspect of reprogramming actually presents a divergence in the motivation for the science. The general reprogramming field was developed because of the identity erasure or “de-differentiation” aspect of reprogramming which enabled any type of somatic cell, usually one that was easy to harvest and culture like fibroblasts, to revert to an iPSC where it could then be directed to differentiate to another cell type of interest, one that would be otherwise difficult to attain from patients like neurons [Kiefer et al., 2011]. Age reset was just seen as a side effect of this process. However, to apply reprogramming specifically for rejuvenation, the key imperative is scaling the technology to rejuvenate entire tissues and organs. In this regard, changing cell type is a hindrance as it would disrupt the cell diversity, function and organization of the tissue and would be intractable to reestablish, harkening back to the discussion with biological immortality. In addition, even if one opted to employ directed differentiation to re-derive the various cell identities, such protocols are still being developed and optimized for each cell types and, so far, cannot be multiplexed to easily produce a deterministic mixture of different cell types, let alone in an organized and structured fashion to rebuild a tissue [Inoue and Yamanaka, 2011]. Thus our interest in developing this science is to extricate the intrinsic age reset phenomena from the loss of cell identity that accompanies it during reprogramming. The direct approach would require one to trace out the full dynamics of reprogramming in the epigenetic space, then identify and selectively drive only the transitions that instigate the age reversal and not those that erase cell identity. This epigenetic spatial segregation itself may not be fundamentally possible, and may involve a lot of networks also affecting cell state.

6

Furthermore identifying transcription factor cocktails to selectively drive these transitions would difficult in silico and expensive in vitro with combinatorial screening. We instead hypothesized that, instead, temporal segregation may already occur to a degree and would present a more tractable initial solution. Empirically and intuitively it was well established that there is a window during the early stages of reprograming in which the process could be terminated and the cells would retain their identity. This window has been demarcated colloquially by the so called “Point of No Return” (PNR) beyond which cells would start to become locked in undifferentiated and exotic states and a small percentage would continue to iPSC [Nagy and Nagy, 2010]. Then the question simply reduced to whether a “transient reprogramming” regimen, initiating reprogramming but terminating within this early window for cell identity retention, drives measurable and enduring rejuvenative shifts. Through this body of work, we sought to evaluate the existence and scope of these rejuvenative shifts using a number human cells types, both stem and differentiated, across the nine categories of known cellular aspects of aging with intracellular and tissue level consequences.

2. Technology

2.1 Background

Integral to the experimental design was a precise temporal control of the expression of the reprogramming factors. The majority of time evolution studies on reprogramming dynamics had utilized inducible lentivirus constructs, to deliver and integrate the core Yamanaka factors – OCT4, SOX2, KLF4 and cMYC - into the target cell genome [Hockemeyer et al., 2008]. The use of a high titre lentivirus ensured a high

7 transduction efficiency and the inheritance of the construct after mitotic division; the usual target for these experiments were easily maintainable and expandable cell types like fibroblasts [Somer et al., 2009]. The inducible feature was incorporated most often through the Tet-On system. Here the promoter of the factors, usually organized in a polycistronic operon, is a tetracycline responsive element (TRE) which is induced by the binding of a reverse tetracycline controlled trans-activator protein (rtTA) [Zhou et al.,

2006]. The binding affinity of this rtTA protein is enhanced when the protein itself is bound to tetracycline or, its more commonly used surrogate, the antibiotic doxycycline.

(The “reverse” tTA nomenclature stems from the first discovered trans-activator proteins that were in activated by doxycycline as a Tet-Off system [T Das et al., 2016].)

Exogenous doxycycline can be efficiently administered in vitro and in vivo, thus allowing for experimenter controlled reprogramming factor expression. However this system has a number of drawbacks. First, the size of entire payload is past the payload size of lentivirus can effectively deliver, typically 10 kilobase pairs [Al Yacoub et al., 2007].

The system require all four reprogramming factors, their inducible promoter(s), any intercistronic elements if implemented as a polycistronic cassette, the rtTA element and its own constitutive promoter [Yamaguchi et al., 2012]. Thus this Tet-On system is often implemented through multiple viruses and thus increase the stress and decreasing viability in the recipient cells and reducing overall efficiency, as multiple constructs need to be successfully integrated in to the same cells for the system to function [Rauser,

2017]. Second, the rtTA protein unbound by doxycycline still has some binding potential which could trigger unwanted, leaky expression and the stronger the rtTA gene promoter, the more protein product and the higher this potential, establishing a tradeoff between

8 strong expression and tight regulation [Costello et al., 2018]. Third, the use of lentivirus does not establish uniformity in expression. Lentiviruses have a scattered genomic integration pattern which yields variability in cell copy number and, like every gene insertion approach, ultimately still relies on the transcriptional dynamics of the target cells, which means possible delays in transcription and silencing [Ellis, 2005], [Kang et al., 2015]. Fourth and possibly most concerning, this pattern of integration can lead to insertional mutogeneisis, which compounds onto concerns of teratoma formation from de-differentiation [Bayart, 2015].

All these factors become significant issues when trying to investigate a consistent and tightly regulated time evolution during reprogramming. The use of these inducible lentivirus solutions was motivated by the interest in studying the series of events that eventually led up to iPSC rather than the need to specifically engage just a subset of those events. Thus we had to first identify and develop a reprogramming modality that was more amenable to our studies and an eventual therapeutic strategy.

2.2 Results

To address these issues, we explored mRNA reprogramming [Warren et al.,

2010]. This method uses lipid based transfection technology to deliver the mRNA transcripts for the reprogramming factors into the cell cytosol. This involves encasing the anionic mRNA molecules in the hydrophilic interior of a cationic synthetic liposome so that it may pass through cell membrane and be released into the cytoplasm by endocytosis [Hajj and Whitehead, 2017]. This transfection technology was optimized for high efficiency in delivery and reduced toxicity [Avci-Adali et al., 2014]. The introduced

9 mRNA can then be directly translated into protein by the cell’s ribosomes, thus bypassing the need for nuclear translocation, genomic integration and cell-mediated transcription/post-transcriptional modification [Kim and Eberwine, 2010]. This in turn speeds up the time to protein synthesis and increases the uniformity of expression across cells and for the most part virtually every cell is transfected. The dosage can be more consistently controlled by titrating mRNA quantity, which correlates to protein yield more so than DNA-based methods. Most importantly, the time course is regulatable as the protein product peaks after an overnight period (16-17 hours) coinciding with the mean mRNA lifetime [Leonhardt et al., 2014] with the mean lifetime of the protein then approximately 36 hours after that point. Thus reprogramming factor expression can be reliably suspended for time course studies, assuming no endogenous gene upregulation of the factors - which would of course entail passing the PNR. Finally, given that the mRNAs are do not genomically integrate or even enter the nucleus, there is no significant risk for insertional mutagenesis and hence they present a much more clinically translatable, “foot-print” free approach.

For the specific reprogramming cocktail, we employed a mixture of 6 factors, moving beyond the Yamanaka core and also including NANOG and LIN28. The addition of these factors has been shown to be critical for improving reprogramming efficiency for aged cells, particularly by overcoming the pro-senescence stress response as cells undergo reprogramming [Lapasset et al., 2011]. We will discuss senescence in a later chapter, but it essentially acts as a shutdown response to prevent the proliferation of highly damaged and potentially oncogenic cells, which show equally drastic shifts in gene expression [Krizhanovsky et al., 2008]. Young cells can overcome this barrier

10 continue onto iPSC, however aged cells already begin with a significant pre-senescent, if not committed senescent, population and are more sensitive to this additional pro- senescent response, thus exhibiting poorer iPSC yield [Mahmoudi and Brunet, 2012].

NANOG and LIN28 were known additional factors characterizing the embryonic state and have particular utility in regards to cell cycle. NANOG has been shown to accelerate reprogramming independent of cell cycle and thus can circumvent the induced shutdown to continue the reprogramming process [Hanna et al., 2009]. Alternatively, LIN28 does advance reprogramming through cell cycle manipulation, specifically by regulating cyclins A and B as well as cdk4 to boost mitosis as it does in embryonic cells [Xu et al.,

2009]. This combination of 6 factors has even shown to be effective in reprograming an entirely senescent population, though as we’ll discuss in later chapters, the pre-senescent cells may be a more appropriate target [Lapasset et al., 2011]. Thus we synthesized a mixture of the 6 factors using in vitro transcription [Figure 1 A and B]. We choose to mix

6 monocistronic mRNAs instead of synthesizing one polycistronic payload given the relatively high number of mNRAs encapsulated by each lipid complex and delivered to each cell as well as the relative ease of delivering smaller mRNAs [Leonhardt et al.,

2014]. We further verified widespread expression of each of the 6 factors was widespread across the population of cells treated with the mixture [Figure 1 C].

11

A)

B) C)

Figure 1 Verification of Results from In Vitro Sysnthesis of Reprogramming mRNAs A) shows systhesized in vitro transcription size verification after tail PCR reaction step while B) shows size verification of final mRNA product. C) shows protein production verification of cells transfected with the mRNAs. Finally in terms of the duration, we first noted from the literature that the PNR both in vitro and in vivo was purported to occur around the end of the first third of reprogramming [Ohnishi et al., 2014], [Stradtfield et al., 2008]. In concordance, previous work in our lab using specifically the four factor (OKSM) mRNA reprogramming reached iPSC colony formation after 12 days and expressed endogenous HPATs, pluripotency-associated lncRNAs, after 5 days, which concurs with the first third PNR window [Durruthy-Durruthy et al., 2016]. Thus for our studies we looked at two to four days of reprogramming, adjusting based on optimal effect or cell culture conditions.

Given the mrna degradation time course and to ensure stability in expression levels, we transfected every 24 hours with off the shelf transfection reagents and supplemented the

12 cells with the type 1 interferon binding protein B18R, a standard addition to mRNA transfection reduce toxicity from the stress of transfection itself. With the fibroblasts and endothelial cells, specifically, we also employed a 4-hour media change after each transfection to avoid toxicity of repeated transfection, an option offered by the transfection reagent manufactures. For later studies we found this was unnecessary and the cells survived multiple transfections. In most cases, a serum free media was used to allow for reprogramming, and after the tested duration, transfections were ceased cells were switch backed to complete media specific to that lineage; the exceptions were in the case of chondrocytes and MSCs where the cell identity was already fragile due to their diseased condition so serum was used throughout. Cell identity verification was established both through transcriptomic sequencing of the cells directly and by histopathological analysis of tissues with treated cells as shown in [Figure , 3], these and other verifications will be given more context in later chapters.

A) ` B)

Figure 2 Verification of identity retention and lack of transformation in treated cells and derived tissue by histology. Histology show no abberent or dysplasic growth after treating fibroblasts and keritinocytes in artifical skins(A)) and in tissue derived (GFP+) from treated muscle stem cells (B))

13

A) B)

Figure 3 Verification of identity retention and lack of transformation in treated cells and derived tissue by gene expression. A) and B) show p-value for variation in cell identy markers between treated and untreated cells fibroblasts and endothelial cells after bulk RNA sequencing.

2.3 Discussion

The choice of mRNA technology gave us substantial freedom in time course and dosage and further enabled us to work with many different cell types, as we will discuss in the coming chapters. Its transitory, non-integrative nature was also in line with our overall motivation of moving towards a clinical therapy. The results presented here are based on the methodology in vitro treatment for in vivo transplantation, but a more long- term goal would be direct in vivo delivery of the factors for in situ treatment. Before any issues specific to reprogramming can be studied, the first obstacles that arise are in regard to the difficulty of delivery and transfection in vivo. A majority of in vivo technologies also use lipid nanoparticle based solutions and show results in terms of luciferase imaging signal, or other indirect approaches, but do not follow the studies up with

14 histology [Fenton et al., 2016], [McKinlay et al., 2017]. Luciferase can produce a halo of signal with a relatively small number of affected cells [Omokoko et al., 2016]. Only by subsequent immunohistochemistry for individual cell signal can one actually determine the efficiency of the delivery method, hence why we followed up our own luciferase imaging in a later chapter with GFP imaging. For most of the technologies we tested, we were able to reproduce the luciferase signal but found very little (<5%) of cells actually harboring the construct, which is sufficient for purposes like inoculation but not of course our application [Pascolo 2008]. With one technology, however, we were able to see a substantial number of cells transfected across multiple tissues and when combining multiple mRNAs [

Supplementary Figure 1]. However, this experiments are still very early stage and consistency with each injection is still an issue that needs to be resolved, hence the results’ exclusion from the main text.

Even if delivery can be optimized, the need for repeated delivery to the same cells presents another layer of difficulty. The worst case, though unlikely, scenario would be one in which the probability that a given cell will be transfected is uncorrelated between subsequent transfections. In this case the probability would scale down exponentially with number of transfections. For instance, to get at least a 50-50 chance for a cell to undergo a full 4 days of reprogramming would require at least a 85% efficient transfection, yet by the same token the chance that the cell would not undergo at least 1 transfection is less than 0.1%. A somewhat orthogonal issue is accumulated tissue damage that comes with each delivery, given that every route we explored was in situ

15 injection. More systemic delivery may be an option like using less disruptive means like intravenous injection, however biodistrubtion studies already show that these routes tend to aggregate mainly in the liver and spleen rather than distributing evenly throughout the body, again by luciferase imaging only [Broos et al., 2016]. A way to possibly circumvent these issues of efficiency and damage compounding is to using so called

“self-amplifying mRNA” solutions. The approach involves repurposing RNA replication mechanisms of a modified positive-sense ssRNA virus, specifically based on the

Venezuelan equine encephalitis, with the desired mRNA payload [Davis et al., 2002]. As the system never uses a DNA intermediate, it cannot integrate into the host genome, and instead relies on a feedback between the host ribosomes and the viral genome to replicate the RNA [Kinney et al., 1989]. Thus, one viral transduction can provide a continuous feed of new reprogramming mRNA but for the limited lifetime of the virus, which has been optimized for to drive cells all the way to iPSC in vitro [Yoshioka et al., 2015].

Further work would be necessary to tune the lifetime for our shorter window and evaluate efficacy in vivo, but this system could provide a more amenable solution with just a single injection.

2.4 Methods

In vitro transcription

Protocol was first described in [Durruthy-Durruthy et al., 2014] and [Ramathal et al.,

2016] and repeated here. First, a DNA vector containing “backbone sequence” was designed. PCR amplification of the ORF of interest was performed with the specific forward primer and reverse primer. Both, backbone clone and amplified ORF were

16 digested. Next, transformation of template clone and amplification of positive clones was carried out, followed by a final digest to cut out ORF+UTR. A polyA tail was added at the 3′ end with a Tail PCR and this template was subject to in vitro transcription reaction. Synthesis for mRNA was carried out with the MEGAscript T7 kit (Ambion) according to the manufacturer's instructions with slight modifications. A custom ribonucleoside blend was comprised of 6 mM 3′-0-Me-m7G(5′)ppp(5′)G ARCA cap analog (New England Biolabs), 7.5 mM of adenosine triphosphate and 1.5 mM of guanosine triphosphate (Ambion), 7.5 mM of 5-methylcytidine triphosphate and pseudouridine triphosphate (TriLink Biotechnologies). Reactions were incubated for 4 h at 37°C, followed by DNase treatment for 15 min at 37°C. DNase treated RNA was purified using the MEGAclear kit (Ambion). Correct RNA synthesis and RNA purification was verified and quantified using a Nanodrop (A230/A260 between 1.7–2.0) and concentration was adjusted to 100 ng/ml. RNA reprogramming cocktails were prepared by pooling individual 100 ng/μl RNA stocks to produce a 100 ng/μl total blend.

Stocks were stored at −80°C. RNAse mediated RNA degradation was prevented by cleaning the working space and the instruments with RNaseZap (Ambion).

Denaturing formaldehyde-agarose gel electrophoresis mRNAs were analyzed to verify templates for the IVT reaction and correct size of the transcripts. A 1.5% denaturing formaldehyde-agarose gel was prepared dissolving 0.75 g agarose in 36 ml DI water, 5 ml 10X MOPS running buffer (Ambion) and 9 ml 37% formaldehyde (12.3 M, Sigma-Aldrich). mRNA samples (1 μg) were mixed with 3x the volume of Formaldehyde Loading Dye (Ambion) and 0.25 μl ethidium bromide (10

17 mg/ml, Bio-Rad) followed by heat denaturation for 15 min at 70°C. RNA ladder (RNA

Millenium marker, Ambion) was treated like mRNA samples and used for size comparison.

mRNA Transfection. Cells were transfected using either mRNA-In (mTI Global Stem) for fibroblasts and chondrocytes, to reduce cell toxicity, and Lipofectamine

MessengerMax (Thermo Fisher) for endothelial cells and MuSCs, which were more difficult to transfect, using manufacturer’s protocol; for fibroblasts and endothelial cells

NutriStem media was used as serum free alternative to sustain the cell during reprogramming, with additional controls to test its effect, while the cell types specific media was used in subsequent studies with other cell types. Culture medium was changed for fibroblasts and endothelial cells 4 hours after transfection, but not for chondrocytes or

MuSCs as overnight incubation was needed to produce a significant uptake of mRNA.

Efficiency of delivery was confirmed by both GFP mRNA and immunostaining for individual factors in OSKMLN cocktail.

3. DNA Level

3.1 Background

Given that reprogramming is itself fundamentally an epigenetic re-landscaping, the DNA level presented the natural starting point for evaluating transient reprogramming. This encompasses the aging phenotypes of epigenetic dysregulation,

18 which alters gene expression, as well as DNA and nuclear lamina structural damage and telomere shortening, both which threaten genomic integrity.

Epigenetic alterations, broadly defined, encompass the gamut of mitotic and meiotically heritable physical and chemical changes to DNA and its supporting architecture that alter the transcriptional potential of the genome without altering the baseline DNA sequence itself. As such, this level represents the most upstream regulation of gene expression and the caveat of heritability implies an enduring profile, though one that does change and respond with environmental and internal triggers. Epigenetics most notably establishes cell type and functionality, but also markedly evolves with age. DNA methylation, the most localized form of epigenetic manipulation to inhibit transcription, shows a global pattern of hypomethlyation and some promoter specific hypermethylation in numerous cells types across aged tissues. [Johnson et al., 2012] This has been hypothesized to be linked to declining levels and efficacy of DNA methyltransferases with aging, specifically DNMT1 (DNA methlytransferase 1) which facilitates the transfer of methylation patters to daughter cells during mitosis [Fraga and Esteller, 2007].

Histones, which act as the spools around which DNA is wound and compacted into nucleosomes, are the next scale of epigenetics and also exhibit age-related modifications, for instance the decrease of the gene repressive H3K27 and H3K9 trimethylation and increase in gene activating H4K20 trimethylation and H3S10 phosphorylation [D’Aquila et al., 2013]. Key to these alterations is the age related shift in histone modifying enzymes, like EZH1 or the Sirtuins (which we will also discuss at the Metabolic Level)

[Calvanese et al., 2009]. Then in the more microarchitecture epigenetic scale, aging correlates to a general loss of transcriptionally inactive heterochromatin, driven by a loss

19 of histone marks and proteins which maintain condensed structure, but also the formation of specific foci, like those associated to senescence [Tsurumi and Li, 2012]. All of these trends correlate to the broader theme of a loss in gene silencing with age and overall increase in transcriptional noise which then dysregulate gene expression and ultimately cell functionality.

These epigenetic changes are viewed as alterations that are dynamic and readily reversible. However aging also is characterized by substantial damage accumulation, to the underlying DNA code and the laminar architecture which supports and houses it.

Direct damage to DNA from natural metabolic activities, stochastic chemical

(endogenous and environmental) and radiative interactions regularly occurs throughout life on the order of 105 total molecular lesions per cell per day [Barnes and Lindahl,

2004]. There are number of enzymatic repair mechanisms in place to fix these aberrations but some of these processes loose efficiency with age. A prime example is an accumulation of DNA double strand brakes linked to declining Homologous

Recombination (HR) and Non-Homologous End Joining (NHEJ) mechanisms during cell cycle [White and Vigj, 2016]. In addition, the nuclear lamina suffers a loss of associated proteins and polypeptides and develops age related conformational defects like folding and blebbing (though this dysmorphic shape is more prominent in accelerated aging diseases like Hutchinson-Gilford progeria syndrome) [Scaffidi and Misteli, 2006]. This impairs the lamina’s function as the mechanical scaffolding that supports the DNA, its chromatin architecture and its repair processes. Ultimately these factors lead to poorer genomic integrity which impairs cell functionality, decreases the number of viable progeny and can lead to apoptosis, cell cycle arrest or cancer. This is the central idea in

20 the DNA Damage Theory of Aging that leads to tissue wide consequences in organs like brain, liver, muscle and kidney [Freitas and De Magalhães, 2011]

In contrast to this stochastic and undesired damage, the genome also undergoes routine attrition of its chromosomal end caps, known as telomeres. During cell division, a helicase enzyme unidirectionally unzip a segment of DNA while the DNA polymerases builds off of both template strands moving in the 5’ -> 3’ direction. The polymerase moving opposite to the helicase direction must repeatedly attach, synthesize and detach to keep up with the helicase. This re-attachment is instructed but RNA primers which repeatedly bind near the helicase position, however they themselves cover up a segment of DNA that must be subsequently filled in on the next synthesis step. This works for all sites except the end of the chromosome, where there is not further upstream position for mRNA to bind and thus the polymerase doesn’t transcribe this segment. Based on this fact, the genome has evolved to include junk DNA, repeats of non-coding TTAGGG sequences, to buffer the chromosome ends. When these telomeres are completely exhausted after so many divisions, this leads to cell cycle arrest. This is primarily an in vitro phenotype of extensive culture known as the Hayflick limit [Shay and Wright,

2000]. In vivo a majority of cells don’t reach near this threshold and highly dividing cells have a telomere extension enzyme called telomerase, an explicit design choice by nature cognizant of the potential failure mode [Garber, 2012]. Thus, this counters the general perception that regular, cell cycle driven telomere attrition drives of aging. However, just like other segments of DNA, telomeres can be subject to stochastic damage and more recent studies have this modality of attrition with age related defects in tissue such as

21 cardiac muscle and liver [Sahin and DePinho, 2012]. Thus telomere attrition can be more accurately bundled with the general DNA damage aging hallmark.

These aspects comprise the first set of analysis to investigate the aging implications of our transient reprogramming technology. A positive effect at this core

DNA level should drive at improved conditions the subsequent levels.

3.2 Results

To establish a generalizable effect of the transient reprogramming treatment, we began with two different cell types that are commonly used in traditional reprogramming studies. These were dermal fibroblasts and vascular endothelial cells. Our sources were young and aged healthy patients, healthy at least in regard to the cell types of interest, as well an additional set of fibroblast from patients with Hutchinson-Gilford progeria syndrome. These cell types were chosen since both are easy to attain in large quantities from tissue biopsies, maintainable and expandable in culture and have been employed in previous reprogramming studies. Furthermore, they could be interesting therapeutic targets. They are relatively easy to target in an eventual in vivo application – through intradermal or intravenous injections respectively – and also play a key role in defining and influencing the niche environment for the stem and parenchymal cells of many tissues. Though of foreseeably a large number would have to be treated given their expansive populations.

Our initial metrics were fluorescence immunostaining assays with single cell analysis using high content imaging. This was chosen over qPCR or western blot techniques for this capability of single cell resolution. Full iPSC reprogramming is a very

22 low efficiency process with only a small subpopulation of cells undertaking the full transition while the rest fail to continue at various points in the process [Zunder et al.,

2015]. This dropping out of the population has been linked to an a phase of stochastic gene expression changes during early reprogramming; for cells that make it past the early transitions, there exists a more stereotyped, deterministic phase that engages the full pluripotency network [Buganim et al., 2012], [Yamanaka, 2009]. However, the gene expression panels analyzed in these time course studies were focused on those that define cell identity or characterize cell identity transitions, rather than genes linking to cell age or functionality. In addition, this work was done with lentiviral vector, which has a different time dynamic than our mRNA technology. Thus, we had little background knowledge as to the heterogeneity of, specifically, age-related changes early in reprogramming and hence the need for single cell analysis. This technology gave us the added benefit of being able to scale down the total cell number need for each measurement. Another similar option would have been flow cytometry analysis, but our technique allowed asses the cells directly on the plate itself with minimal disturbance and thus avoiding the additional stress and cell loss that comes with trypsinization, suspension and flow.

We first evaluated the cellular distribution of epigenetic markers H3K9me3

(histone 3 lysine 9 trimethylation) and the gamma isoform of heterochromatin protein 1

(HP1) on the young, aged and treated cells, metrics previous studied in the context of full reprograming a differentiation of healthy and progeria cells [Miller et al., 2013]. The trimethylation of lysine 9 of histone 3 enables the binding at the chromodomain of HP1, which is crucial for maintenance of centromeric heterochromatin [Fischle et al., 2005].

23

This heterochromatin is specifically meant to be constitutive and remain condensed, unlike facultative heterochromatin which can convert to euchromatin depending on cell conditions. So, the age-related loss of these factors can be viewed as weakened transcriptional silencing leading to increased noise [Miller et al., 2013]. In addition the gamma isoform of HP1 in particular has also been linked to transcriptional elongation at euchromatin locations, thus its reduction further decreases the transcriptional signal to complement the increasing noise [Vakoc et al., 2005]. With our treatment, however, we observed the population distribution of these levels shifted away from the aged population distribution and towards that of young cells; [Figure 13 A and B] summarizes both assay results for the three cohorts (young, aged and treated), normalized and combined across the patients for each cell type. This effect was evident even with the 2 day treatment but more pronounced with the 4 day treatment – both with 2 days relaxation after treatment (relaxation begins the day following the last transfection)

[Supplementary Figure 2 A and B],[Supplementary Figure 3 A and B]. This contrasts with the unchanged, and in some cases, negatively affected aged cells with control GFP transfections [Supplementary Figure 4 A and B],[Supplementary Figure 5 A and B].

Furthermore, the changes remained significant on into 4 and 6 days of relaxation, though diminished over time [Supplementary Figure 6 A and B],[Supplementary Figure 7 A and

B]. For the remaining manuscript, treatment is defined as 4 days treatment, 2 days relaxation, unless otherwise specified. A key observation was that these shifts in the direction of the youthful cell distributions were unimodal (a trend seen across assays in this and later chapters); [Supplementary Figure 8 ] shows a representative set of all the single patient H3K9me3 signal distributions for fibroblasts with and without treatment.

24

This allayed the concern that only a stochastically determined subpopulation that would benefit from treatment.

Then for the nuclear and DNA integrity phenotypes, we evaluated the levels of lamina-associated polypetide 2 alpha (LAP2α) and phosphorylated histone H2AX respectively. LAP2α is a nucleoskelatal protein, along with its binding partner Lamin

A/C, help to form and maintain the lamina-chromatin architecture; its loss has been linked to aging, progeria, and cell cycle arrest in general [Dechat et al, 2000],[Miller et al., 2013],[Pekovic et al., 2007],[Vidak et al., 2000]. As such, we observed a similar distribution shift towards youthful levels of this protein with treatment [Figure 13 C]; time course and controls are in [Supplementary Figure 2 C],[Supplementary Figure 3

C],[Supplementary Figure 4 C],[Supplementary Figure 5 C],[Supplementary Figure 6

C],[Supplementary Figure 7 C]. The phosphorylation of histone H2AX, initiated by

HP1β, is a key intermediary in the cells’ response to a double strand breaks as it recruits

MDC1 and a host of subsequent proteins to form the MRN complex that initiates repair by HR or NHEJ [Podhorecka et al, 2010]. Thus, it is commonly used as a surrogate probe for the prevalence of double strand breaks. We did not however see a reduction in the distribution of phosphorylated H2AX foci with treatment, data not shown. Given our technology transiently alters epigenetics, it is not surprising that more underlying genetic damage may not be directly affected, at least for this short window of treatment.

Finally, telomere length as measured by the QFISH (quantitative fluorescence in situ hybridization) technique did not show appreciable change with treatment [Figure 13

D]. It should be noted that there was little change in aged and young by this measure as well, which in fact was observed in previous studies [Miller et al., 2013]. QFISH is a

25 relatively low resolution metric to evaluate this property and will only pick up substantial differences in telomere lengths as the technique has a variability typically ranging from

0.5 to 3.5 kilo base pairs [Canela et al., 2007]. Differences of this scale would be more clear-cut if examining the same cells at an early passage vs an advanced passage. Cells more actively proliferate and erode telomeres in vitro, thus causing a reduction on the order of 2.5- 7 kilobase pairs between seeding and reaching the Hayflick limit [Shen et al., 2009],[Young and Black, 2004]. However, the rate of telomere erosion in vivo is only estimated to be 25-30 base pairs per year, so detection of differences was a possibility with QFISH. but more sensistive measurements (like qPCR methods) may be more appropriate [Shammas, 2011]. Of course, if the changes were at a finer scale, then the effect of treatment would also be diminished and thus less meaningful. If anything, the confirmation that there are no large changes in telomere length is a positive result. For the cells to extend their telomeres on the scale of multiple kilobases, they would have to actively re-engage telomerase, which is otherwise only active in stem cells or hyperproliferive cells (cancer). Thus these results suggest that treated and relaxed cells are neither de-differentiated nor cancerous.

Interestingly, when we applied the same set of metrics to fibroblasts from chronologically young patients, but with the diagnosis of Hutchinson-Gilford Progeria

Syndrome, we did not see this same improvement in epigenetic and lamina phenotypes

[Figure 14]. This was in agreement with previous in vitro work on fully reprogramming progeria fibroblasts to iPSC [Miller et al., 2013]. In their study, the induced pluripotent cells did show youthful properties, as the LMNA mutation was hypothesized be silenced in this state. However, once the cells were differentiated back to fibroblasts, they

26 regained their accelerated aging phenotypes. This was the motivation of the work, as the goal was to derive aged cell types for study which were otherwise difficult to culture from primary tissues (they were specifically focused on iPSC derived aged neurons).

Reprogramming erased the aging phenotypes of normally aged cells so the lamin mutation was used because it was preserved after the process. This is, however, at odds with in vivo studies where cyclic partial transgenic reprogramming showed a number of tissue improvements and lifespan extension in LAKI (progeria) mice without de- differentiation [Ocampo et al., 2016]. It is yet to be confirmed if the mutation is transiently silenced in this scenario, but if it is, then perhaps the in vivo environment helps maintain this silencing and prevents it from being re-expressed after reprogramming ends. The reconciliation of these in vitro and in vivo progeria reprogramming studies still remains an open question.

A) B)

27

C) D)

Figure 13 Combined single cell distritubtions of epigenetic, laminar and DNA damage markers. Distributions on the left are the fibroblasts and on the right are endothelial cells. A) and B) show the treated populations epigenetic marks moving toward the young population. C) shows laminar protein levels in the treated population restored to youthful levels while D) shows little change in the prevalence of DNA double strand breaks with treatment. E) shows no signifcant change in telemore length could be detect with QFISH technique with either aging or treatment.

A) B) C)

Figure 14 Combined single cell distritubtions for treated and untreated cells from HGPS patients. A) and B) show now significant change in heterochromatin markers while C) shows LAP2α even slightly decreases with treatment.

28

We then proceeded to investigate how the DNA level changes suggested by these metrics translated into global shifts in gene expression patterns. This analysis was also used to verify the retention of cell identity. We performed 80 base pair paired-end bulk

RNA sequencing on both cell types for the same three cohorts: young, aged, and treated.

First, we compared the quantile normalized reads of the 16K+ genes in young and aged cells for each cell type (“Y vs A”). Screening with a significance criteria of p <.05 and a log fold change cutoff +/-0.5, we found 737 genes (4.18%) for fibroblasts (527 upregulated, 210 downregulated) and 441 genes (2.57%) in endothelial cells (260 upregulated, 181 down regulated) significantly differed between young and aged cells

[Figure 15 A and B]. When we mapped the polarity of expression above or below the mean of each of these significant genes (also including in the treated cells), we observed a clear similarity between treated and young cells as opposed to aged cells for both fibroblasts and endothelial cells [Figure 15 C and D]. We further performed principal component analysis in this space of genes. As expected, the young and aged populations were separable along the first principal component, which explained 83.42% of variance in fibroblasts and 81.16% of variance in endothelial cells, but more intriguingly, the treated cells also shifted closer to the younger population along PC1 [Figure 15 E and F].

29

Fibroblast Endothelial Cell

A) B)

C) D)

E) F)

Figure 15 Transcriptomic landscape in aging subspace. Bulk RNA sequencing was performed on 3 samples from each cohort, young aged and treated, because of cost considertations. Fibroblast results are shown on the right and endothelial cells results are on the left. A) and B) show volcano plots identifying significantly differnetially

30 expressed genes between aged and young patient samples. Significance was defined using groupwise comparsion as as p<.05 and log fold change of +/- 0.5 are shown by light blue dashed lines. Genes in green are upregulated in young cells while genes in orange were downregulated. In the subspace of significant genes, C) and D) show polarity of expression above or below mean expression for young and aged as well as treated samples. Each column represents a different patient. E) and F) show samples projection onto principal components 1 and 2 analyzing in subspace of significant genes. Consistent with the previous color scheme, blue samples are young, red are aged and light blue are treated. Using the same significance criteria as previously defined, we then compared the treated and aged populations (“T vs A”) and found that 939 genes (5.32%) in fibroblasts

(745 upregulated, 194 downregulated) and 322 gene (1.88%) in endothelial cells (129 upregulated, 193 down regulated) were significantly differentially expressed [

Figure 16 A and B]. When we compared the profiles “Y vs A” and “T vs A” in each cell type, we observed a 46.4% overlap for fibroblasts and 28.6% overlap for endothelial cells. For the vast majority of these genes the polarity of change in gene expression with treatment matched that of youth (i.e. if upregulated in young then upregulated in treated); less than 0.5% moved oppositely in either cell types [

Figure 16 C and D].

Fibroblasts Endothelial Cells

A) B)

31

C) D)

Figure 16 Transcriptomic landscape of treatment. Bulk RNA sequencing was performed on 3 samples from each cohort, young aged and treated, because of cost considertations. Fibroblast results are shown on the right and endothelial cells results are on the left. A) and B) show volcano plots identifying significantly differnetially expressed genes between treated and aged patient samples. Significance was defined using pairwise comparsion as p<.05 and log fold change of +/- 0.5 are shown by light blue dashed lines. Genes in green are upregulated in young cells while genes in orange were downregulated. Yellow genes are cell types specific markers. In the subspace of significant genes, C) and D) shows averaged fold change of expression comparing treated vs aged and young vs aged. All significant genes in treated vs aged comparison are shown in grey and those that are also signifcant for young vs aged are shown in light blue

As a final evaluation at the DNA level, the effect of treatment on the “methylation age” of the cells with methylation clock analysis was ascertained. This metric was built using the methylation of CpG sites, sites where cytosine followed by guanine in the 5’ ->

3’ direction. These often occur in high density in regions called CpG islands, usually situated near and associated with the expression of gene promoters [Antequera, 2003]. A general correlation between the global pattern of CpG methylation and age served as grounds to develop machine learning algorithms that could use linear combination of methylation states with a nonlinear transform as a predictor of cell age [Horvath 2013].

We first applied the Horvath algorithm, which incorporates 353 CpG sites, and found a significant decrease in predicted methylation age after treatment for the fibroblasts and

32 endothelial cells (with and average decrease of around 2 years for fibroblasts and 5 years for endothelial cells) [Figure 17 A]. Though we observed a relative decrease the absolute baseline age prediction was off (r2 value of .113 with the actual patient age), possibly because this clock was trained on fresh tissue rather than cultured cells. To address this we tried another clock using 618 sites developed with our collaborators at OneSkin who use in vitro skin models, which we will delve into in a later section. Running the same data through their algorithm yielded a better estimation of absolute age for primarily fibroblasts but also some of the endothelial cells (r2 value of .855), and confirmed the decrease of approximately 2 years in fibroblasts [Figure 17 B]. This is not a full measure of epigenetic age, as DNA methylation is a localized feature unlike larger scale histone and heterochromatin architecture, though it does hint at a global aging epigenetic effect with transient reprogramming.

A) B) Figure 17 Methylation Clock analysis. A) show comparison between chronological age and predicted methlyation age by the Horvath 2013 clock of 4 aged samples of fibroblast (right) and 4 aged samples of endothelial cells (left) along with their treated counterparts, again given cost considerations. Best fit line and R2 value also shown. B) show comparison between chronological age and predicted methlyation age by proprietary Oneskin clock on 4 aged samples of fibroblasts (right) and 2 aged samples of endothelial cells (left) along with their treated counterparts, with strong coorelation between methylation and chonological ages. Best fit line and R2 value also shown.

33

3.3 Discussion

These initial investigations demonstrated that transient reprogramming on aged cells re-established you a youth-like characteristics at the DNA level (epigenetic, transcriptomic and nuclear lamina). There are of course many more aspects at this level that can be investigated, additional histone and DNA integrity markers, different levels of sequencing like ChipSeq and ATACseq, evaluation of miRNAs and non-coding RNAs etc. However, the motivation of this study was to evaluate and demonstrate a multifaceted rejuvenation profile for transiently reprogrammed cells, thus we choose to move on instead to subsequent levels.

Current work in this direction has been focused on analyzing the sequencing data generated here and in planned future runs for new transcription factors that can drive, specifically and solely the rejuvenative process. The most basic motivation for this is to develop a “2.0 rejuvenating” mRNA transcription factor cocktail that avoids the use of some of the potentially oncogenic reprogramming factors, like cMYC and SOX2 [Lu et al., 2010],[Wei et al., 2005]. In addition, if these new factors can be shown to leave the differentiation programming untouched, they would open up the possibility of longer, even sustained expression in target cells for a stronger more rejuvenation focused effect.

It is also intriguing from a basic science perspective, all the previous work studying the epigenetic/transcriptomic manifestations of aging has always been studying the forward direction (young to old). This has led to the identification of transcription factors that increase, like NF-κB (nuclear factor kappa-light-chain-enhancer of activated B cells), or decrease, like the Forkhead box (FOXO) family, with aging [Alvarez‐Garcia

2016],[Helenius 2001]. If we now study the epigenome/transcriptome of reprogramming,

34 the reverse process (old to young), may identify transcription factors whose change in expression drives rejuvenation. Yet, the processes of aging is not the exact inverse of the process of reproduction (the basis of reprogramming). Thus, there is good reason to expect that rejuvenation is driven by new and different factors, instead just inverting the core changes in gene expression seen with aging, and these factors would not otherwise be discovered by just studying the forward process. Thus, our understanding of aging transitions so far may only be half the picture, hence our excitement at moving forward with this work.

3.4 Methods

Antibodies. The following antibodies were used in this study. The source of each antibody is indicated. H3K9me3 (Abcam #ab8898 1: 4000), LAP2α (Abcam #ab5162

1:500), HP1γ (Millipore Sigma #05-690 1:200), Lamin A/C (Abcam #ab40567 1:200),

GFP (Invitrogen, #A11122, 1:250)

Human Fibroblast Isolation and Culture: Tissue isolation was performed at the Stanford

Hospital and supplement with additional patient samples from Coriell Institute. Patients were biopsied from mesial aspect of mid-upper arm or abdomen using 2 mm-punch biopsies from both male and female patients 60-70 year olds (n=8), 25-35 year olds (n=3) and Hutchinson-Gilford progeria syndrome patients (n=4). Cells were cultured out from these explants and maintained in Dulbeco’s Modified Eagle Medium supplemented with non-essential amino acids, 10% fetal bovine serum and 1% Penicillin/Streptomycin. Cells were cultured at 37°C with 5% CO2.

35

Human Endothelial Cell Isolation and Culture: Tissue isolation was performed at the VA

Hospital and supplemented with additional patient samples from Coriell Institute. Iliac crest and tibialis anterior muscle arteries and veins were biospied from 45-60 year olds

(n=7) and 15-25 year olds (n=3). Tissue was digested with collagenase and cells released from the lumen were used to initiate cultures. Plates for seeding were coated with 2% gelatin, then washed with PBS before use. Cells were maintained in Medium 199 supplemented with 2 mM L- glutamine, 15% fetal bovine serum, 0.02 mg/ml Endothelial

Growth Supplement, 0.05 mg/ml Heparin and 1% Penicillin/Streptomycin. Cells were cultured at 37°C with 5% CO2.

Immunocytochemistry. Cells were washed with HBSS/Ca++/Mg++ then fixed with 15% paraformaldehyde in PBS for 15 minutes. Cells were then blocked for 30 minutes to an hour with a blocking solution of 1% BSA and 0.3% Triton X-100 in PBS. Primary antibodies were then applied in blocking solution and allowed to incubate overnight at 4 o C. The following day, the cells were washed with HBSS/Ca++/Mg++ before switching to the corresponding Alexa Flour-labeled secondary antibodies and incubated for 2 hours.

The cells were then washed again and stained with DAPI for 30 minutes and switched to

HBSS/CA/MG for imaging on the Operetta high content imaging system. Columbus software was to identify single cells, utilizing DAPI to delineate nuclear boundaries, and to calculate the signal statistics for each cell. Data processing and statistical analyses were performed using Matlab R2017a (MathWorks Software).

36

Statistics. Statistical analysis was done as describe previously in [Miller et al. 2013] as state summarized below. Box distribution plots display the fluorescence intensity quantification of 100 cells from each patient. Distributions were compared by statistical analysis by using multiple comparison ANOVA. Arbitrary units for frequency distributions of different cell types should not be compared because staining was performed at different times. Matlab 2017 (MathWorks) was used for data presentation and analysis.

RNA-Sequencing. Cells were washed and digested by TRIzol (Thermo). Total RNA was isolated using the Total RNA Purification Kit (Norgen Biotek Corp) and RNA quality was assessed by the RNA analysis screentape (R6K screentape, Agilent); RNA with

RIN > 9 was reverse transcribed to cDNA. cDNA libraries were prepared using 1 μg of total RNA using the TruSeq RNA Sample Preparation Kit v2 (Illumina) with the added benefit of molecular indexing. Prior to any PCR amplification steps, all cDNA fragment ends were ligated at random to a pair of adapters containing an 8bp unique molecular index. The molecular indexed cDNA libraries were than PCR amplified (15 cycles) and then QC’ed using a Bioanalzyer and Qubit. Upon successful QC, they were sequenced on an Illumina Nextseq platform to obtain 80-bp single-end reads. The reads were trimmed by 2 nt on each end to remove low quality parts and improve mapping to the genome.

The 78 nt reads that resulted were compressed by removing duplicates, while keeping track of how many times each sequence occurred in each sample in a database. The unique reads were then mapped to the human genome using exact matches. This misses reads that cross exon–exon boundaries, as well as reads with errors and SNPs/mutations,

37 but it does not have substantial impact on estimating the levels of expression of each gene. Each mapped read was then assigned annotations from the underlying genome. In case of multiple annotations (e.g. a miRNA occurring in the intron of a gene), a hierarchy based on heuristics was used to give a unique identity to each read. This was then used to identify the reads belonging to each transcript and coverage over each position on the transcript was established. This coverage is non-uniform and spiky. Therefore, we used the median of this coverage as an estimate of the expression value of each gene. In order to compare the expression levels in different samples, quantile normalization was used.

Ratios of expression levels were then calculated to estimate the log (base 2) of the fold- change. Data processing and statistical analyses were performed using Matlab R2017a

(MathWorks Software). For statistical analysis, multiple comparison ANOVA was used.

DNA methylation sequencing. Cells were washed and genomic DNA was isolated using the DNeasy Blood and Tissue kit (Qiagen) and DNA concentration was assessed by

Quant iT Broad Range dsDNA Kit (Invitrogen). The human Illumina Infinium EPIC

850K chip was applied to purified DNA samples. The raw image data were normalized using the "preprocessQuantile" normalization method implemented in the "minfi" R package [Touleimat and Tost, 2012], [Fortin et al., 2017]. Several DNAm based biomarkers have been proposed in the literature which differ in terms of their applicability (most were developed blood).We applied previously defined mathematical algorithm is used to combine the methylation levels of 353 CpG into an age estimate (in units of years) which is referred to as epigenetic age or DNAm age. In our secondary analysis, we used a proprietary skin based epigenetic clock (based on 618 CpGs) from

38

OneSkin. To properly account for the dependence structure in the data, we used linear mixed effects models to regress DNAm age (dependent variable) on treatment status, chronological age, and individual identifier (coded as random effect). Toward this end, we used the "lmer" function in the "lmerTest" R package.

4. Metabolic Level

4.1 Background

From the baseline DNA and nuclear changes, we transitioned to studying the overall cell metabolism, the regulation of the energy and protein content for cell functionality. Age-related changes in this regard, include poorer nutrient regulation from sensing to activation of metabolic responses, less efficient mitochondria functionality and energy production, and diminished proteolytic activity for the clearance of waste.

Nutrient regulation is a bridge that relates nutrient responsive gene pathways modulated by the previous DNA level changes in the nucleus, to downstream metabolic consequences like mitochondria biogenesis and the regulation of autophagy in the cytoplasm. Three such gene networks that have been shown to alter with aging are IIS pathway which responds to insulin and IGF-1, the AMPK pathway which responds to

AMP levels and the SIRT1 pathway which responds to NAD+ levels. The levels of IGF-1

(insulin like growth factor 1) secreted by the liver are continuously reduced with age after adulthood. This is further tied to reduced levels of Growth Hormone released by the aged pituitary gland and this phenomena is collectively referred to as somatopause, borrowing form the terminology of female menopause [Junnila et al., 2013]. Conversely,

39 circulating levels of insulin secreted from the pancreas seem to remain steady with age and, in some studies, increase due to reduced insulin clearance [Chang and Halter et al.,

2003]. With aging, the downstream recipient cells in a variety of tissue, muscle, liver, kidney etc., also show less sensitivity in response to IGF-1 and insulin which should stimulate the PI3K/Akt/mTOR network [Barzilai]. Conversely, the protein complex

AMP-K (adenosine monophosphate associated kinase) in its different variations, 12 versions total given combinations of its alpha, beta and gamma subunit isoforms, works suppress these same downstream genes in response to high AMP (and ADP)/ATP ratios

(ie energy poor conditions) [Willows et al., 2017]. The sensitivity of this kinase has also been shown to decline with age [Salminen and Kaarniranta, 2012]. The ultimate function of the PI3K/Akt/mTOR network is to stimulate cell proliferation, energy production, and reduce protein catalysis to allow for cell growth [Jason and Cui, 2016]. Thus, a poor response to both stimulation (by IGF-1 and insulin) and suppression (by AMPK) together lead to this aging phenotype of dysregulated nutrient sensing. The key intervention that has been sought to address this aging defect is caloric restriction and its mimetics, which of course suppresses this growth pathway as cells are methodically deprived of calories.

These interventions extend lifespan in a variety of model organisms by the principle that a slower metabolic/growth rate also means less cellular damage from metabolic byproducts and promote more mitochondria and protein turnover [Blagosklonny, 2010].

This represents the key aging theme of antagonistic pleiotropy, in which certain gene pathways act as double edged swords, since they benefit the body early on in life

(promoting growth and development) while weakening it later on (from accumulated damage and exhaustion that was not present earlier on) [Blagosklonny, 2010]. So even

40 though caloric restriction is understandable as an approach to disengage this pathway later on in life, it is distinct from true rejuvenation as, by definition, it imposes a different, slowed growth tendency to reduce the progress of age related damage rather than restoring the youthful growth environment with the low, youthful, levels of harmful byproducts.

The SIRT1 (Sirtuin 1) pathway responds to NAD+ (nictonamine levels from redox metabolism and affects the same nutrient sensing network. However, Sirt1 is a histone deacetylase (HDAC), which means, more than just triggering specific gene cascades as the other pathways do, SIRT1 modifies the epigenome; in fact the Sirtuins are the only

HDACs that respond to metabolic triggers instead of Zn2+ levels [Seto and Yoshida,

2014]. Thus Sirtuins also affects a host of other networks for DNA damage repair, NF-kB and inflammation, senescence, oxidative stress and autophagy, telomere stability and other age related factors [Liu et al., 2013],[Poulose and Raju, 2015]. SIRT1 expression wanes with age at both the transcriptional and translational levels and a number of studies showned increased expression of SIRT1, activated by either natural supplements (like resveratrol or omega-3 fatty acids) or synthetic small molecules (like SRT1720, SRT2104 or SRT2379) can increase lifespan as well as reduce aging phenotypes in model organisms [Saunders and Verdin 2007],[Villalba and Alcaín, 2012]. As such, it was the more interesting target to us a potential rejuvenation route rather than the other longevity pathways.

In terms of energy metabolism, aged mitochondria exhibit a lower transmembrane potential, decreased levels of ATP (Adenosine Triphosphate) and accumulation of harmful reactive oxygen species (ROS) byproducts of oxidative phosphorylation

41

[Chistiakov et al., 2014], [Sugrue and Tatton, 2001]. This has been linked to a transition from the orthodox configuration - a thinner, tighter membrane and wider matrix volume/surface area for a higher membrane potential and more efficient energy production - to a more condensed condensed configuration - trading off the level of energy production for a wider membrane and a condensed matrix which, in part, shields mitochondrial DNA from harmful molecules like ROS [Suhr et al., 2010]. As mentioned, these ROS byproducts exemplify the concept of antagonistic pleiotropy and furthermore constitute the basis for the Free Radical Theory of Aging. This idea holds that the nature of ROS as free radicals, compounds missing a single valence electron, makes them the key instigator of damage to both mitochondrial and nuclear DNA as well as various lipids and proteins through aberrant ionization, and thus is a driver of aging [Harman, 2003].

The theory has been further extended to correlate to the pathology of multiple age related diseases, like arthritis and atherosclerosis [Szeto, 2006].

Protein metabolism alters with age both through changes in the distribution of proteins translated, harkening back to the DNA level shifts in gene expression, as well as changes in cytoplasmic proteolytic activity, which normally prevents proteotoxicity and replenishes the pool of usable peptides. Aging correlates with diminished activity in both the autophagasome-lysosome and ubiquitin-proteasome systems. Macroautophagy is the primary mode of autophagy and aging correlates with decreased formation of the autophagasomes, which engulf and transfer degraded, long-lived proteins/organelles to lysomes for break down by hydrolytic enzymes [Rubinsztein et al., 2015]. Secondary modes of autophagy are the direct engulfment by lysosomes (microautophagy) and chaperone mediated migration of proteins to the lysosome. The levels of the heat shock

42 protein chaperones, specifically, has also shown an age-related decline, leading to impairments in protein stabilization as well as proteolysis [Calderwood et al., 2009]. The proteasome pathway, alternatively, targets more rapid turnover proteins by marking them with an ubiquitin tag which allows for recognition, unfolding and degradation into peptides by protease enzymes in the proteasome complex. Aged cells accumulate an excess of ubiquitinated proteins which overwhelms and interferes with proteasome function [Kevei and Hoppe, 2014]. It should be noted that ubquitinization plays a role in multiple processes, for instance the DNA damage signaling response of the aforementioned histone H2AX, but this overwhelming age related ubiquitination hasn’t been observed in other processes; this distinction may lie in the specific lysine residue that is ubiquitinated as primarily K48 and K29 are the signposts for proteasome deconstruction [Kristariyanto 2015], [Rogakou et al., 1998]. When these pathways grow less active with age, the aggregated waste can become toxic and has also been implicated in a number of age related disease like Alzheimer’s and cataracts [Costello et al., 2013],

[Zhang et al., 2017].

These metabolic aspects were our second level of analysis, as the subsequent downstream consequences of our epigenetic reprogramming. As these aspects have been shown to be directly tied to the pathology of various age-related diseases, we thought to add diseased samples, in addition healthy aged cells, into our evaluation and explore the broader implications of treatment.

43

4.2 Results

We continued to study the same cell types for this analysis of transient reprogramming’s effect on aging metabolism. We also broadened the study to include the diseased, osteoarthritic chondrocytes. Although not exclusive to the elderly population, osteoarthritis (OA) is strongly associated with aging [Chen et al. 2017]. OA is a disease characterized by increased cartilage loss resulting in joint pain and degeneration.

Chondrocytes are the cells within the articular cartilage which are responsible for maintaining the extracellular matrix (ECM) that comprises the semi-solid insulation of joints, as opposed to the synovial fluid that bathes the two articular surfaces and provides additional shock absorption [Ralphs and Benjamin, 1994]. The healthy turnover of this

ECM wanes with the progression of this disease, which can be linked to age related shifts in the intracellular metabolism of the chondrocytes as well as their intercellular maintenance of the inflammatory enjoyment [Mendel et al., 2015]. Thus, they provided interesting target for age and disease reversal by reprogramming.

To asses these phenotypes we used live cell imaging, again evaluating at the single cell level but also moved towards aggregate live cell measurements as the effects became more clearly established at the population level effect and qPCR, as it was the preferred modality of our collaborators in the Bhutani Lab. Nutrient sensing was the exception, as we had to histochemically stain for SIRT1 in the cell nuclei. Here we began to see cell type specific effects, as we found the distribution of SIRT1 start to increase in the treated aged fibroblast, but not in the endothelial cell populations, though our and previous work established an age-related decrease in both cell types [Figure 18 A]. In addition, we observed that treatment specifically decreased the mitochondrial superoxide

44 levels, the primary reactive oxygen species (ROS), in aged fibroblast rather than endothelial cell distributions [Figure 18 B]. We measured this using a flouregenic dye, which is oxidized by superoxide and fluoresces after binding to the mitochondrial nucleic acids (thus giving a localized superoxide abundance estimate) [Yousefi et al., 2009].

However, we did see an increase in mitochondrial membrane potential in both treated cell type distributions, though milder in endothelial cells, as measured by the increased uptake of Tetramethylrhodamine, methyl ester [Figure 18 C]. This is a cationic molecule that is sequestered by membrane bound proton pumps, just like hydrogen ions used to establish the membrane potential gradient (thus higher uptake correlates to higher maintain membrane potential for ATP production) [Huang, 2002]. We did not, however, see an accompanying in the total mitochondria content in either cell type. In contrast, the clearance activity of both cell types seems to be boosted with treatment. Treated fibroblasts and endothelial cell distributions exhibited increased autophagic activity as noted by the greater prevalence of lipidated LC3 positive vesicles (i.e. forming or established autophagosomes) [Figure 18 D], [Levine and Kroemer, 2008]. In addition, treatment boosted the aggregate chymotrypsin-like activity of the aged cells - one of the three forms of cleavage of the proteasome 20S, specifically acting on the hydrophobic amino acid residues on peptide bonds as the digestive enzyme chymotrypsin does [Figure

18 E], [Tanaka, 2009]. In the case of the fibroblasts the possible link between these observations is that increased SIRT1 is noted drive mitochondrial biogenesis to form new and healthy (higher membrane potential) mitochondria [Yuan et al., 2016]. While the increased clearance, specifically autophagy, has been previously demonstrated to target degraded mitochondria (mitophagy), by studies using transmission electron microscopy

45

(TEM) to image the contents of autophagasomes and lysomes during fibroblast reprogramming to iPSC [Xiang et al., 2017]. Hence the mitochondrial ROS levels could be reduced by the clearing of degraded mitochondria balanced by production of new healthy ones, maintaining the overall mitochondrial content of the cells. The endothelial cell don’t have the explicit tie to mitochondria biogenesis through SIRT1 so alternative mechanisms must be in play to boost the membrane potential. One known route is that mitochondria can interconvert from condensed to orthodox based on ADP levels, which could be modulated by treatment and cause some mitochondria to undergo this energetic transition contributing to a higher membrane potential, though not the degree caused by biogenesis for fibroblasts [Mannella et al., 2008]. Without the additional generation of new mitochondria, it would not be prudent clear out the old mitochondria, hence the unchanged ROS levels and overall mitochondria content, and instead the elevated proteolytic clearance could be focused on removal of cell type specific proteins, another noted target during reprogramming. For these parameters time course and controls are in

[Supplementary Figure 2 D through G],[Supplementary Figure 3 D through

F],[Supplementary Figure 4 D through H],[Supplementary Figure 5 D and

E],[Supplementary Figure 6 D through H],[Supplementary Figure 7 D through F].

46

A) B)

C) D)

E)

47

Figure 18 Combined single cell distritubtions and bulk redout of nutrient sensing, mitochondria activity and protien clearence. Distributions on the left are the fibroblasts and on the right are endothelial cells. A) show the treated fibroblasts starting to show a significant increase in the distribution of SIRT1 levels with treatment. B) shows again a fibroblasts specific reduction in the distribution of mitochondria ROS levels with treatment while C) shows an increase in the mitochondria membrane potential distribution in both cell types, though a smaller effect in endothelial cells. D) and E) show increase in both the single cell distribution of authophagasome formation levels and the bulk population level chymotrypsin like proteasome acitivity in both cell types towards the youthful values.

From here, we expanded the study to assess the potential for transient reprogramming to reverse the exacerbated metabolic defects of the age-related OA in chondrocytes. Thus we studied effect of treatment in the context the metabolic drivers of the disease: poorer intracellular redox metabolism which diminishes cell health and signaling as well as poorer extracellular ECM anabolism and catabolism which directly manifests as cartilage degradation [Kim et al., 2015]. Here, we found the optimal duration of treatment was actually 3 days, given that after 4 days the stress of transfection itself was causing a “dedifferentiation” or loss of chondrocyte markers. This greater instability and tendency to dedifferentiate is a known property of OA chondrocytes kept in monolayer culture, correlating to their dysfunctional phenotype, thus finding the optimal treatment meant balancing out the additional stress of transfection, studied with gfp control transfections [Sandell and Aigner, 2001]. In terms of redox metabolism, oxidative stress is a key driver of the OA pathology as it engages pathways to reduce cell proliferation and increase pro-inflammatory signaling which alter the chondrocytes’ role in cartilage maintenance [Lepetsos and Papavassiliou, 2016]. Transient reprogramming was able to reduce the superoxide (ROS) levels that drive this stress correlated with increased expression of the antioxidant superoxide dismutase 2 (SOD2) [Figure 19 A and

48

B]. With these changes we saw more energetic cells with elevated ATP production, observed using a glycerol assay which scavenges and dephosphorylates ATP to produce a colorimetric byproduct proportional to its concentration [Figure 19 C], [Yang et. al

2013]. This contributed to an increase cell proliferation [Figure 19 D] as well as reduced inflammatory signal, which we will discuss in the next section. Then we investigated how these improvements translated changes in ECM maintenance. For ECM anabolism we looked at gene expression for COL2A1 which initiates the production of type II collagen, one of the key structural component of cartilage, as well as expression of SOX9, which maintains the chondrogenic identity and function in producing matrix and was one of the markers used to ensure cell identity retention (COL2A1 is a direct target of SOX9 transactivation) [Tchetina, 2010]. Treatment elevated COL2A1 expression towards the healthy controls and didn’t show a consistent change SOX9 expression, which control transfections diminished [Figure 19 E and F]. In contrast, we did not observe a significant reduction in expression for the catabolic matrix metalloproteinase, MMP3 and

MMP13 [Figure 19 G and H]. These cells were taken from late stage patients, whose OA had developed to such a degree that they required full knee replacement surgery. Though our epigenetic intervention may drive the reversion of the some the drivers of the disease, genetic predisposition and environmental factors like mechanical stressors, poor diet and inflammation all played a role in progressing these patient cells to such an advanced diseased state [Johnson and Hunter, 2014]. Thus it is understandable that all aspects of this diseased state may not be effectively addressed by our treatment alone. As noted, we did observe a reduction in the inflammatory signaling of the treated chondrocytes, which

49 we will discuss in the next section, and this shift can start to influence the joint environment to yield more gradual benefits over time.

A) B) C)

D) E) F)

Figure 19 Combined single cell distritubtions and bulk redout of redox and ECM metabolism parameters. A) shows the treated OA chondrocytes exhibit a reduction in the distribution of mitochondria ROS levels correlated to the overexpression of the antioxidant SOD2 gene in B). C) and D) show resulting increases in bulk levels of ATP and cell proliferation. For healthier matrix anabolism, D) shows an elevation of COL2A1 expresion with treatment while E) shows SOX9 expression didn’t show a consistent change, though in some cases significantly increased with treatment.

50

4.3 Discussion

The findings demonstrate the improvement of metabolic phenotypes upon transient reprogramming and as downstream consequences of the epigenetic changes from chapter two. These aspects of aging are more well-known because the direct efforts to counteract them, like caloric restriction and dietary supplements, as well as their role in disease pathology, like in osteoarthritis shown here. We will discuss some more disease related factors that are affected with treatment, but it should be noted, as seen here, that not all aspects of the diseased state were affected by our treatment. The process of reproduction accounts for the need to reverse the accumulation of age related dysfunction in its reprogramming protocol, however such a transition was not designed to handle diseased conditions and indeed poor/diseased metabolic health has been identified as a failure mode of embryogenesis, both in vivo and in vitro [Kasum et al., 2018], [Ramsay et al., 2002]. Furthermore, it is crucial to note that disease is truly classified as a “state” whereas aging is not. This implies that the onset of disease is definitive transition has occurred in the cells and that they have crossed a threshold after which there is a runaway progression to an attractor, diseased state [Healy and Engler, 2009]. This isn’t of course a pure physical or chemical system which that could be exactly defined this way, and even in biology there are more clear-cut examples of state, for instance cell type itself. Yet, the dramatic and systemic transition that occurs with most diseases in a relatively short period of time, at least compared to the lifelong process of aging, lends credence to this sort of classification. Aging as a more gradual process cannot be truly characterized as a state transition save in one aspect – senescence, which we will discuss in the next section as prudent to remove but not necessarily reverse. Full reprogramming to iPSC is a state

51 transition and transient reprograming is so potent because it briefly taps into the global dynamics of that process, but ultimately its effect and our goal is more tantamount to a perturbation to offset and reverse the gradual drift that is aging. Thus, optimally it could work to reduce disease phenotypes, but a curative objective is beyond the scope of the technology. A more inspired application in regards to disease maybe a prophylactic approach, in which treatment is initiated with advanced age but before the onset of these diseases as a means to restore youthful functionality and prevent the onset of disease pathology.

4.4 Methods

Human Articular Chondrocyte Isolation and Culture: Tissue isolation was performed at the Stanford Hospital. Human OA chondrocytes were derived from discarded tissues of

OA patients (50-72 years of age, n=6) undergoing total knee arthroplasty. Cartilage pieces were shaved off bone by scalpel, taking care to avoid any fat, then digested with collagenase in DMEM/F12 media (supplemented with 25mg/ml ascorbate, 2mM L- glutamine, 1% penicillin/streptomycin antibiotics and 10% fetal bovine serum) for one to two days until shavings were substantially dissolved. Supernatant from cultures was strained, filtered and centrifuged and the cells were then resuspended in fresh media. The chondrocytes were cultured in high density monolayer in at 37oC with 5% CO2.

Mitochondrial Membrane Potential and Content. Tetramethylrhodamine Methyl Ester

Perchlorate (Thermo) for membrane potential and Mitotracker Green (Thermo) were added to cells in complete media and serum free media respectively. Cells were incubated

52 for 30 minutes at 37oC in 5% CO2 and washed 2 times with HBSS/Ca+/Mg+ before staining for 15 minutes using CellTracker Deep Red. Finally, cells were imaged in fresh

HBSS/Ca+/Mg+ using the Operetta High Content Imaging System (Perkin Elmer).

Columbus software was to identify single cells, utilizing CellTracker to delineate cell boundaries, and to calculate the signal statistics for each cell. Data processing and statistical analyses were performed using Matlab R2017a (MathWorks Software).

Mitochondrial ROS Measurement. Cells were washed with HBSS/Ca+/Mg+ and then switched to HBSS/Ca/Mg containing MitoSOX (Thermo). Cells were incubated for 10 minutes at 37oC in 5% CO2. Cells were then washed twice with HBSS/Ca+/Mg+, and stained for 15 minutes using CellTracker Deep Red. Finally, cells were imaged in fresh

HBSS/Ca/Mg using the Operetta High Content Imaging System (Perkin Elmer).

Columbus software was to identify single cells, utilizing CellTracker to delineate cell boundaries, and to calculate the signal statistics for each cell. Data processing and statistical analyses were performed using Matlab R2017a (MathWorks Software).

Autophagosome Formation Staining. Cells were washed with HBSS/Ca+/Mg+ and switched to a staining solution containing a proprietary fluorescent LC3 based autophagosome marker (Sigma). The cells were then incubated at 37oC in 5% CO2 for

20 minutes, washed 2 times using HBSS/Ca+/Mg+, and stained for 15 minutes using

CellTracker Deep Red cell labeling dye. Cells were then switched to HBSS/Ca/Mg for single cell imaging using the Operetta High Content Imaging System (Perkin Elmer) using cell tracker. Columbus software was to identify single cells, utilizing CellTracker to

53 delineate cell boundaries, and to calculate the signal statistics for each cell. Data processing and statistical analyses were performed using Matlab R2017a (MathWorks

Software).

Cell Number and Proteasome Activity Measurement. Wells were stained with PrestoBlue

Cell Viability dye (Life Technologies) for 10 minutes. Well signals were read using a

TECAN fluorescent plate reader as a measure of cell count. For proteasome activity, cells were then washed with HBSS+/Ca+/Mg+ before switching to original media containing the chymotrypsin like fluorogenic substrate LLVY-R110 (Sigma). Cells were then incubated at 37oC in 5% CO2 for 2 hours before signals were again read on the TECAN fluorescent plate reader. Readings were then normalized by PrestoBlue cell count.

Gene expression analysis: Total RNA was purified using RNeasy Plus Mini kit (Qiagen) and cDNA was prepared with First-strand cDNA synthesis kit (Applied Biosystems). The quantitative polymerase chain reaction (qPCR) was performed using VeriQuest

Mastermix (Thermo Fisher Scientific) for SYBR Green and Taqman primer sets respectively. The relative gene expression was analyzed by the ΔΔCt method and normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH).

ATP Assay: ATP in the chondrocytes was measured using colorimetric assay using ATP assay kit (Abcam). Cells were washed in cold phosphate buffered saline, homogenized and centrifuged to collect the supernatant. ATP reaction mix and background control

(50 µL) was added to the wells and incubated for 30 min in dark. The plate was read at

54

OD 570 nm using SpectraMax M2e (Molecular Devices, Sunnyvale, CA). The mean optical density was used to estimate of the intracellular ATP concentration relative to the standard curve.

Antibodies. The following antibodies were used in this study. The source of each antibody is indicated. SIRT1 (Abcam #ab7343 1:200);

5. Niche Level

5.1 Background

The manifestations of aging discussed so far have been limited to an intracellular scope, however they both affect and are affected by the cell’s surrounding micro-niche.

There are two aging manifestations, which can be thought to most directly play a role in this interaction – the increasing senescent populations and biased or inflammatory intercellular communication. The former corresponding to an induced cell cycle suspension that often contributes to signaling shifts in the latter.

Senescence is the one aspect of aging that can described as a state, specifically of permanent mitotic arrest for otherwise proliferative cells. This is distinct from quiescence, in which proliferation is suppressed and the cell remains in G0 but is not permanently suspended, and is also distinct from post-mitotic cell types, cells like neurons which do not divide after terminal differentiation [Demaria et al., 2015].

Contributing factors thought to drive this transition the accumulated DNA damage, oxidative stress, replicative exhaustion and cytokine/growth factor alterations with age

55

[McHugh and Gil, 2018]. The key pathways that respond and engage cell cycle arrests mechanism are the p16 and p21 pathways. Upregulation of the p16INK4a pathway suppress cyclin D/CDK4 to inhibit G1 phase progression while upregulation of the p53/p21 pathway to suppress cyclin E/CDK4 to inhibit S phase progression [Li et al., 2013].

Longer term engagement of these pathway can be reinforced by the downregulation of the histone methyltransferases EZH2, which causes the loss of repressive H3K27me3 at the INK4a/ARF locus that regulates both pathways [Criscione et al., 2016]. In addition, there are a host of other pathways that can engage senescence but are more responsive to ongenenic triggers rather than other forms of insult, like the p15INK4B or the p27Kip1 pathways [Munoz-Espin and Serrano, 2014]. This localized changes disrupt the mitotic process and go on to trigger a larger scale shift in the epigenome that formally entrenches the state of senescence, aspects like the condensation of senescence associated heterochromatin foci and the detachment of peripheral regions of heterochromatin from the nuclear lamina (generally silenced regions call lamina-associated domains) [Corpet and Stucki, 2014]. These global structural changes drive the gene expression shifts that truly characterize senescence, beyond just the replicative behavior.

These senescent and pre-senescent cells are probably the most dramatic contributors to a general aging shift in signaling within the cellular micro-niche, through their altered secretome. Like the epigenome, the cell secretome is also a dynamic and complex information space. If the epigenome is the programing of each cellular “server” which needs to be reprogrammed to youthful functionality than the secretome is the biological “wifi” for the micro-niche that needs to be filtered of all the accumulated noise with age; the two are intimately linked and feedback information to each other [Kittan et

56 al., 2013], [Yasmin et al., 2014]. That noise results from components like pro- inflammatory cytokines, growth factors, enzymes and other soluble factors released by stressed cells, from accumulated ROS or proteotoxicity, and damaged cells, like senescent cells, as means to acutely engage neighboring and immune cells for damage control, clearance, and repair [Freund et al., 2010], [Ponnappan and Ponnappan, 2011].

The intracellular aging processes we have discussed so far drive the accumulation of such cells and thus overburden both the paracrine and immune responses leading to chronic inflammation of tissues, termed “inflammaging” [Salvioli et al., 2013]. This both feeds back to drive more intracellular aging phenotypes as interferes with the regular extracellular communication for tissue functionality [Xia et al., 2016]. Aging also drives tissue specific alterations in communication like the loss of synaptic plasticity in the brain and of the quiescence environment in stem cell niches [Petralia et al., 2014],

[Chakkalakal, 2010]. It further impacts larger scale communication like the onset of endocrine somatopause, as mentioned last chapter, and immunosenescence, a loss of leukocyte lineage cells and their maintenance of adaptive immunity as well as a dysregulation of dendrite cells and macrophages which moderate innate immunity (thus contributing to inflammaging) [Fulop et al., 2018].

These aspects which influence the aging niche were the third level of analysis by which we evaluated the benefits of transient reprograming. As noted, information in the intracellular epigeneome and the extracellular niche have a “chicken-and-the-egg” relationship, especially in driving the manifestation of aging. For these experiments we sought to examine how we could exploit this feedback for rejuvenation, with the epigenome is a more contained and methodically manipulable target.

57

5.2 Results

For this niche level we began by evaluating the same aged cell types, fibroblasts and endothelial cells, but then transitioned to more progressed manifestations of these effects. We evaluated treatments benefit towards these parameters in skin aging using in vitro organoid cultures. These organoids are co-cultures formed by embedding aged fibroblasts in primarily collagenous dermal matrix and then, after a few days to grow and stabilize, seeding a layer of aged keratinocytes on top of this dermal layer [Sriram et al.,

2015]. These steps are performed while submerged in culture media, as is normally the case in cell culture. However, they are done so on a well insert so that, after the co- culture is established, it can then be lifted out of the media to establish an air liquid interface, with just the bottom surface allowing for nutrient exchange with the media.

This air interface provides cues similar to those in vivo for epidermal stratification, into stratum corneum, stratum granulosum etc [Arnette et al., 2016]. This allows for not only the measure of intracellular aspects, like senescence parameters, but also cumulative effect on skin architecture, like thickness, morphology and integrity, which drive probably the most widely, publically perceived notions of aging like skin sagging or wrinkling. In addition we again looked at Osteoarthritic phenotypes (with OA itself being well-known as chronic inflammatory condition). We further studied chondrocytes sampled from OA patients but also their resident mesenchymal stem cells (MSCs) responsible to replenishing degraded cartilage as well as damaged bone with new, healthy cells. The activity of these cells is directly affected by the aging and OA driven changes at the intracellular and stem cell niche level [Marędziak et al., 2016]. This set of investigations encompassed the most cell types and collaborators, reflecting how much

58 interest there is in rejuvenating these parameters in aging and a variety of dysfunction and diseases.

We began by testing our treated fibroblasts and endothelial cells for senescence with SA-β-Gal (senescence-associated beta-galactosidase) staining. This technique was developed on the empirical finding that specifically at pH 6.0, only cells that were mitotically arrested, as measured by incorporation of proliferative marker H-thymidine, display β-galactosidase activity which then targets galactose molecules like lactose

[Dimri et al., 1995]. Thus presenting cells in a pH 6.0 solution with a chromogenic substrate like X-gal will distinguish the senescent cells in the population [Itahana et al.,

2007]. We saw a significant SA-β-Gal+ population in our aged endothelial cells, which treatment reduced, but not in our aged patient fibroblasts [Figure 20 A]. This is understandable as in some cases just the act of culturing in vitro selects for the best cells of a population and thus may weed out the senescent population. Instead, our collaborators at OneSkin proposed to use their fibroblasts samples, which they verified to have a SA-β-Gal+ cells, and extend the experiment beyond just treating a cell monolayer and instead evaluate efficacy in tissue engineered skin organoids derived from these cells.

We tested the benefits of treatment after the skins were formed which presented a new challenges of tissue penetration and bio distribution. We were able to get substantial transfection of both layers which then correlated to benefits like the reduction of senescence population, and levels of p16 and MMP1 (the pro-inflammatory matrix catalysis enzyme matrix metalloproteinase 1) [Figure 20 B - D]. In addition, we saw an overall improvement in histology score, which compiles multiple parameters like cell morphology, stratification and layer thickness, with [Figure 20 E] showing an example

59 skin. We also took this opportunity to evaluate our treatment against other aging skin interventions including retinoic acid and pep14 (a senolytic agent), and we found the senescence and inflammatory factor reduction to be comparable but a more substantial benefit with our treatment in terms of histology score.

A) B)

C) D)

E)

60

Figure 20 Evaluation of senescence parameters in aged monolayer endothelial cells and fibroblast-keratinocyte constructs. A) shows treatment reduced senescent population in aged endothelial cells but not in our samples of aged fibroblasts, which did not have significant starting senescence population. B), C) and D) demonstrate reduction in senescence, p19 and MMP1 with the mRNA treatment on tissue engineered artificial skins along with a comparison to alternative skin treatments E) shows and example Hemotoxylin and Eosin staining for all treatment cases, showing general advantage of our treatment as compared to alternative approaches in terms of histological parameters and scoring. Then, further building on our Osteoarthritis work, we isolated MSCs from infrapatellar fat pad from OA patients with our collaborators in the Chu lab. To evaluate their senescent profile of these cells, we stained for single cellular distributions of p16 and p21 with and without treatment, for these samples we weren’t able to get healthy reference samples at the time of writing given the limited availability of such tissue. We did see a significant reduction in the p16 distribution but the reduction in p21 was more profound [Figure 21 A and B]. This could be for the benefit, as p21 is the downstream actor of the aforementioned p53 which is more responsive to non-oncogenic drivers of senescence in comparison to p16 [Martínez-Zamudio et al., 2016]. This may suggest that this treatment may be more prominently targeted the cells by mitigating aged-related stressors, like ROS and inflammatory pressure, rather than suppressing cancer detection pathway; a corresponding reduction in ROS levels for the treated cells was also observed

[Figure 21 C]. This further correlated this reduction in the senescent population and an increase in cell proliferation [Figure 21 D]. These changes with treatment don’t directly affect the OA cartilage but they do indicate a more active MSC population to produce new chondrocytes to improve the tissue.

61

A) B)

C) D)

Figure 21 Evaluation of senescence parameters in fat pat derived mesenchymal stem cells. A) and B) show distribution shifts for each patient with and without treatment for p16 and p21. C) and D) shows mean ROS singal and percent senesccence diminshed with treatment We then looked at the differentiated chondrocytes to evaluate how treatment

effected their role in biasing the MSC niche and the cartilage environment with

inflammatory factors. We evaluated the effects of treatment on the expression of RANKL

and iNOS2. RANKL is an tumor necrosis factor that stimulates dendritic cell maturation,

by binding to the surface receptor RANK, to initiate the immune response and is itself

62 responsive to other secreted inflammatory factors [Akiyama et al., 2012], [Srinivasan,

2013]. In addition, RANKL produced by osteoblasts normally drives osteoclast activation, thereby balancing bone anabolism and catabolism, but OA elevated RANKL produced by chondrocytes can promote additional catabolism [Martínez-Calatrava et. al,

2012]. We observed a suppression of RANKL expression with treatment to healthy levels

[Figure 22 a]. This could serve to reduce the inflammatory signaling in OA cartilage and slow bone break down at joints that are already susceptible to fracture due to the diminished cartilage [Kwan Tat et al., 2009]. iNOS2 is a synthase responsible to the production the inflammatory mediator nitric oxide which then goes on to inhibit new

ECM synthesis, drives the catabolic activity of MMP’s and furthermore reacts with superoxide to promote cell apoptosis [Vuolteenaho et al., 2012]. Treatment also reduced iNOS2 levels towards healthy levels, another route to reduce inflammation [Figure 22 b].

Finally, we profiled the culture supernatant of the treated, diseased and the healthy cells, and saw that treatment reduced the extracellular concentrations of a number of secreted factors [Figure 22 c].

A) B) C)

Figure 22 Intracellular expression and extracellular secretion of pro-inflammatory factors. A) and B) show treatment reduces expression of inflammatory factors RANKL

63 and iNOS2 respectively to healthier levels while C) shows a reduction inflammatory factors secreted from the treated cells Both senescence and inflammatory signaling are examples of antagonistic pleiotropy. They exist for the purpose of regulating growth and response under stressors that are manageable during youth and adulthood, but accumulate and interfere with function later on in life [Goto, 2008]. The alteration of these niche level phenotypes works in concert with degradation in metabolic parameters, from the last chapter, to drive disease phenotypes [Lepetsos and Papavassiliou, 2016]. Thus the ability of treatment to affect both metabolic and niche aspects in the cell types affected by age related dysfunction and disease, like skin aging and osteoarthritis shown here, are promising signs towards a possible therapeutic route.

5.3 Discussion

Improving the cellular niche by treating the cells biasing its makeup is a key modality to scale up the impact of treatment. Another approach that was hinted at but will be fleshed out in the next chapter is the treatment of stem cells, as presumably the benefits of transient reprogramming will be inherited by their differentiated progeny. This niche affecting strategy starts to intersect with other strategies for extracellular rejuvenation, most notably heterochronic parabiosis. Those experiments showed that the import of younger systemic milieu was sufficient restore the youthful state of multiple tissues, without any direct intracellular intervention [Rebo et al., 2016]. Our technology would be act in the reverse scenario, intracellular changes driving youthful extracellular shifts. In our opinion, this is the more realizable approach. Young blood cannot be constantly imported from a donor, so some facsimile with key factors must be

64 synthesized. However, the list and concentrations of these factors is likely far too complex and dynamic (given the complexity of aging itself) to do so efficaciously

[Castellano, 2019]. Instead, driving the cells through an equally complex and dynamic epigenetic shift with just a handful of transcription factors is more accessible and possible because the cells themselves have the information on to make the transition, even transiently, and can do so in a balanced and comprehensive manner (ie use a few simple reset commands to run pre-established protocols in these biological computers).

An outstanding question remains in regards to the observed reductions in senescent population. This could be achieved through non-senescent cells recovering from the influence of their senescent neighbors after treatment and proliferating to crowding them out. Alternatively, it could be possible that transient reprogramming is directly removing the epigenetic markers of senescence cells and this re-entering the cell cycle. The latter might be concerning if these cells have oncogene induced senescence or in general have excess levels of DNA damage. As treatment did not elevate DNA repair in chapter 2, these populations should remain arrested to prevent tissue dysfunction. The stronger effect in p21 reduction over p16 reduction, might hint towards former but more rigorous lineage tracing studies are needed to investigate this matter.

A possible way to sidestep this issue is by working with emerging technology of senolytics, which is focused on inducing apoptosis in senescent cells to improve the tissue microenvironment [Xu et al., 2018]. Studies are still ongoing on how to more effectively and selectively target senescent cells, but even with perfect clearance, the process of rejuvenation still relies on the remaining cells to both reconstruct the tissues and restore the youthful environment [Kirkland and Tchkonia, 2015]. Yet these cells are

65 still in an aged extracellular environment and retain the biases from a lifetime in that environment. Applying transient reprogramming after such a treatment, could help to restore functionality in the remaining cells and allay concerns of triggering the proliferation of oncogenic cells.

5.4 Methods

SA-β-Gal Histochemistry. Cells were washed twice with 0.22 µm filtered PBS then fixed with proprietary fixation buffer (Sigma) in PBS for 6 minutes. Cells were rinsed 3 times with filtered PBS before staining with solution containing X-gal chromogenic substrate

(Sigma), also run through with 0.22 µm filters. Plates were kept in the staining solution,

Parafilmed to prevent from drying out, and incubated overnight at 37oC with ambient

CO2. The next day, cells were washed again with filtered PBS before switching to a 70% glycerol solution for imaging under a Leica bright field microscope. Identification and analysis of stained population was done using ImageJ software.

Cytokine Profiling. This work was performed together with the Human Immune

Monitoring Center at Stanford University. Cell media was harvested and spun at 400 rcf for 10 minutes at room temperature. The supernatant was then snap frozen with liquid nitrogen until analysis. Analysis was done using the human 63-plex kit

(eBiosciences/Affymetrix). Beads were added to a 96 well plate and washed in a Biotek

ELx405 washer. Samples were added to the plate containing the mixed antibody-linked beads and incubated at room temperature for 1 hour followed by overnight incubation at

4°C with shaking. Cold and room temperature incubation steps were performed on an

66 orbital shaker at 500-600 rpm. Following the overnight incubation, plates were washed in a Biotek ELx405 washer and then biotinylated detection antibody added for 75 minutes at room temperature with shaking. Plates were washed as above and streptavidin-PE was added. After incubation for 30 minutes at room temperature, wash was performed as above and reading buffer was added to the wells. Each sample was measured in duplicate.

Plates were read using a Luminex 200 instrument with a lower bound of 50 beads per sample per cytokine. Custom assay Control beads by Radix Biosolutions were added to all wells.

Skin Organoid Culture: The specific protocol for organoid culture is proprietary but based on the optimization of hydrogel culture models. [Carlson et al., 2017] presents a general review of the procedure, paraphrased here. A thin, acellular layer of collagen was first constructed. A collagen gel embedded with human dermal fibroblasts was layered onto the acellular layer. While submerged in medum for 7 days, dermal fibroblasts remodel the collagen matrix, causing it to contract away from the walls of the insert. The contracted collagen forms a plateau. Keratinocytes were then added to the center of the plateau of contracted collagen and allowed to attach to the collagen to create a monolayer. Tissues were then raised to an air-liquid interface to initiate stratification.

Keratinocytes stratify and differentiate and form a suprabasal layer and further exposure to the air-liquid interface and additional feedings with cornification medium results in an increase in the thickness of the spinous and cornified layers of the tissue.

67

6. Tissue Level

6.1 Background

The age related shifts mentioned in previous chapters aggregate over masses of cells and their local environment to eventually manifest as tissue level consequences, but the final aging hallmark correlates to a more direct route for tissue level aging consequences from changes in a relatively small population of cells, chiefly the exhaustion of number (self-renewal) and/or competency (differentiation) of stem cell pools. The defining feature of stem cells are their dual capacities to both self-renew their own population as well as to differentiate into somatic cells that will regenerate and replenish tissues, hence the broad scope of benefit from these typically small populations of cells. Aging of stem cells have multiple modes of dysfunction. Self-renewal can be diminished, like in the case of muscle stem cells (MuSCs), as aging correlates with more cell divisions favoring progenitors over new stem cells and poorer maintenance of quiescence when the stem cells are not needed for regeneration [Brack and Muñoz-

Cánoves, 2015]. Differentiation capacity is impaired in stem cells as they exhibit poorer activation in response to mitogenic cues during regeneration and even undergo senescence, like aged MSCs from last chapter and aged MuSCs as well [Zaim et al.,

2012], [Dumont et al., 2015]. In addition, aged multipotent stem cells also show more progeny of one differentiation fate vs another, for instance hematopoietic stem cells

(HSCs) produce more myeloid than lymphoid lineages with age [Pang et al., 2011]. It is still undetermined wither this lineage exhaustion is due a biasing towards the myeloid fate or if there exists subpopulation of HSCs that are already primed for lymphoid division and this subpopulation diminishes with age [Challen et al., 2010]. Either way

68 these alterations in differentiation lead to an exhaustion in stem cell capacity. Epigenetics is critical in each of these scenarios as differentiation is fundamentally and epigenetic transition and mitotic suppression involves epigenetic modifications, as discussed in the extreme case of permanent suppression in senescent cells. Both metabolic alterations, in terms of energy metabolism and protein clearance from chapter 3, as well as niche and systemic factors, like those from chapter 4, cause epigenetic alterations to key development pathways like WNT, Notch and FGF-p38-MAPK [Ermolaeva et al., 2018].

This compounds with the accumulated epigenetic drift from each cell cycle as well as

DNA damage driven changes in plasticity to ultimately manifest this stem cell exhaustion. Furthermore, as stem cells are a quickly emerging modality for therapy

(treating a small number of extracted cells in vitro and transplanting for tissue wide benefits in vivo), diminished yield due to the above factors is a pressing concern and additional age-correlated impairments in cell homing, engraftment and survival after transplantation present another form of exhaustion in viability [Liang et al., 2005].

These aspects of aging stem cell functionality were our final study on the efficacy of transient reprograming. Treating stem cells would be the most accessible route towards deploying transient reprogramming as a therapy as it minimizes the amount of cells needed to be targeted, and most likely screened for safety. In addition the nature of stem cells as already being in a relatively plastic epigenetic state, should make them more amenable for reprogramming to rejuvenate but at the same time could make them more susceptible to de-differentiation.

69

6.2 Results

To truly evaluate the benefits of treatment on stem cells we has to assess the effects of treated cells on tissue in vivo. Given our limitations on in vivo delivery, the most amenable design was to extract stem cells, maintain their potency and treat them in vitro and transplant them back in vivo. We needed a model system where this was done routinely without significant loss in potency from culture and without the need for expansion and the expense of treating millions of cells, as would be necessary for a MSC treatment [Olsen et al., 2018]. We found such a system in MuSCs through our collaborators in the Rando Lab. As noted, MuSc exhibit an age related loss of both self- renewal and differentiation potential and 100K to, in some cases, as low as 1K cells were all that were need to be extracted, kept in culture and transplanted for muscle regeneration experiments after damaging with a myotoxin [Xu et al., 2015]. Crucially, these cells were amenable to repeated transfection, unlike our attempts with HSCs which had poor transfection efficiency and viability. The subsequent requirement of transplantation hosts of course required us to move away from purely human studies, and use immunocompromised young and later on aged mice. However we extracted the aged

MuSCS from both mouse and human donors.

We first tested the effect of transient reprogramming on mouse-derived MuSCs from tibialis anterior (TA) skeletal muscle. The limitations on how long the cells could be kept quiescent in culture set the optimal transient reprogramming duration to the maximum of 2 days [Quarta et al., 2016]. To track the cells and cell derived tissue and thus we added a transduction step for lentiviral GFP and luciferase genes in all cohorts before injection into the young immunocompromised hosts. We also injured the mice

70 with a mytoxin before injecting the cells to initiate the repair response [Hill et al., 2003].

This cleared the niche to allow for engraftment and enabled us to first evaluate treatments benefits on differentiation capacity. To study the repair dynamics over time, we added D- luciferin enzyme intravenously which then cleaved the luciferase gene product to give off a bioluminescent signal [Wu et al., 2001]. We found that the muscles transplanted with the treated MuSCs showed a rapid spike in signal even above those with the young

MuSCs around day 4 but eventually leveling off and matching that of the young MuSC transplants on day 7 and day 11 [Figure 23 A]. All the while, both the young and the treated were above the aged cell transplant baseline. This suggests transient reprogramming boosted activation and proliferation, leading to faster derivation of new tissue that led to the increased signal. This was then confirmed by immunofluorescence on the tissue after harvesting on day 11. We found that the treated and young MuSC recipients showed a greater number of GFP+ (donor derived) myofibers which furthermore exhibited an increased cross-sectional areas when compared to the aged

MuSC recipients [Figure 23 B and C]. In addition, we also transplanted these cells into a limited number of aged immunocompromised mice to test a functional readout of muscle improvement. The TA muscles were harvested 30 days after the transplantation and evaluated by in vitro electrophysiology for tetanic force output, the sustained contractile force exerted after high frequency electrical stimulus [Croes and von Bartheld, 2007]. For addition comparison, we also harvested the TA of undamaged (therefore not transplanted) aged and young mice. In concordance with the previous results, the treated MuSC recipient muscle exhibited a similar degree of force exertion as the young untransplanted mice, both above the aged transplanted and untransplanted output levels [Figure 23 D].

71

A) B)

C) D)

Figure 23 First injury response after transplanting in treated, aged and young mouse MuSC A) shows bioluminsensce imaging time course showcasing faster tissue regeneration in young and treated MuSC recipients. B) and C) show increase fiber number and cross sectional area derived from treated and young vs aged MuSCs. D) shows in vitro force exertion capacity improved for muscle derived from the treated, trasnplanted MuSCs to match the young baseline levels, both above the aged baseline and aged MuSC transplanted muscles. These results hint that treatment improved activation and differentiation capacity to restore immediate tissue function but to evaluate whether treating the MuSCs traslanted to long term benefits, we he had to evaluate if the population persisted or if they exhausted in the initial damage repair response. Therefore, without any additional treatment, we injured the mice again 60 days later and found again similar elevation in signal for the treated and young MuSC recipients versus the aged MuSC recipients

[Figure 24 A]. This implied that the treated population persisted and had persistent

72 improvements. However there still exists an ambiguity as one possibility is that the treated cells self-renewed to sustain an improved stem population – which was responsible for this long-term benefit. Alternatively, it is also possible that some subpopulation of the MuSCs never proliferated but instead immediately went into quiescence after transplantation for the first injury and were later activated in the second injury. This is less likely, though, as the time in culture and the injured environment both drive the MuSCs away from quiescence [Cornelison, 2008]. To truly evaluate this question would require lineage tracing using another extraction procedure, but the second transplant models for this system have not been developed as there is poor MuSC yield from the first host, even with young cell transplants. We also assessed the harvested tissue a these time points and another time point 3 months after transplantation to confirm that there were no neoplastic lesions or teratomas. Furthermore, to verify that these results translated, we repeated the experiments from human MuSC donors and again we saw the treated cells display a faster regeneration dynamic than the untreated [Figure 24

B]. Given the availability of samples, we also tried treating then transplanting young human MuSCs and interestingly we found the benefit was truly restricted to treating the age stem cells [Figure 24 C].

A) B)

73

C)

Figure 24 Second injury (mouse) and human MuSC transplant results. A) shows that second injury 60 days later still displays improved response, bioluminescence signal, in the treated and young MuSC recipients vs those that received aged MuSCs. B) corroborates improved regeneration result in treated aged human MuSCs while C) show exclusivity of benefit to treating aged MuSCs vs treating younger MuSCs.

6.3 Discussion

The utility of treating stem cells is the scalability of benefits. Here we showed that just treating 100K cells conveyed lasting benefits to an entire muscle. However the scalability of this treatment could mean more than just this single muscle benefit, it could help with the fundamental problem of stem cell numbers. The amount of MuSCs extracted here from small patient biopsies were sufficient to reconstitute the muscle of a smaller model organism, but the number required to regenerate the same muscle in a human will likely be orders of magnitude larger. This would necessitate either taking much more muscle from the patient, which is not feasible, or somehow expanding the stem cell population in vitro. For many stem cell types like MuSCs, keeping the cells in culture leads to expansion by way of differentiation not self-renewal. Though we can temporarily maintain quiescence, the potency of these cells are lost because it is difficult to simulate the dynamic and complex niche that promotes self-renewal in vivo. However if treated human cells retained their rejuvenated phenotypes while self-renewing in vivo

74 in mouse hosts, as shown here, then there might be the possibility of establishing a breeder system. The stem cells would be treated, transplanted and harvested, likely in multiple cycles with multiple hosts, before yielding a sufficient size pool for human transplant. After each harvest, the cells could even be treated again to boost the effect.

The obstacle here, as noted, was the poor yield of subsequent harvests after transplant.

However, this was likely a limitation of scale. If a larger intermediate host were used, perhaps pig or horse, and greater numbers cells were initially transplanted, then a breeder system might be sustainable [Harding et al., 2013]. The transient reprogramming might even be critical to such a system, as the continued extraction and transplantation could yield a loss of potency, which is an epigenetic characteristic that would need to be restored.

Alternatively, though the paradigm here was large scale repair after dramatic injury, aging itself drives more gradual muscle wasting like atrophy or sarcopenia [Evans,

2010]. Most patients still have functional muscle tissue but the overall mass and strength of that tissue is declining. Thus event supplementary muscle fiber growth triggered by treated and transplanting MuSCs may provide benefit, albiet in a more gradual manner.

The key to such and application, however, is to determine if that engraftment occurs, without clearing out a niche for the stem cells. Perhaps coupling a minor induced injury might be a way to boost regeneration/engraftment and would already be in the realm of accepted practice, as the modality of therapies like acupuncture [Shi et al., 2017]. Either way, the targeted approach of transiently reprogramming stem cells is likely the first accessible application for this technology.

75

6.4 Methods

Mice. Aged C57BL/6 male mice (n=4) and recipient NSG mice were obtained from

Jackson Laboratory. Young recipient NOD/MrkBomTac-Prkdcscid mice were obtained from Taconic Biosciences. Mice were housed and maintained in the Veterinary Medical

Unit at the Veterans Affairs Palo Alto Health Care Systems.

Skeletal muscle MuSC isolation and culture. Tissue isolation was performed at the VA

Hospital. The human muscle biopsy specimens were taken from patients 10-30 years of age, n=2; 30-55 years of age, n=2; 60-80 years of age, n= 3. While skeletal muscle (n=4) was dissected the tibialis anterior of aged C57BL/6 mice. Both tissues were then mechanically dissociated to yield a fragmented muscle suspension. This was followed by a 45-50 minute digestion in a Collagenase II-Ham’s F10 solution (500 units per ml).

After washing, a second digestion was performed for 30 minutes with Collagenase II

(100 units per ml) and Dispase (2 units per ml). The resulting cell suspension was washed, filtered and stained with VCAM-biotin, CD31-FITC, CD45-APC and Sca-1-

Pacific-Blue antibodies, all at dilutions of 1:100. The resulting cell suspension was then washed, filtered and stained with anti-CD31-Alexa Fluor 488, anti-CD45-Alexa Fluor

488, anti-CD34-FITC, anti-CD29-APC and anti-NCAM-Biotin antibodies. Unconjugated primary antibodies were then washed and the cells were incubated for 15 min at 4°C in streptavidin-PE/Cy7 to detect NCAM-biotin. Cell sorting was performed on calibrated

BD-FACS Aria II or BD FACSAria III flow cytometers equipped with 488-nm, 633-nm and 405-nm lasers to obtain the MuSC population. A small fraction of sorted cells was plated and stained for Pax7 and MyoD to assess the purity of the sorted population. The

76 rest of the cells were then seeded on Poly-D-Lysine and ECM coated wells and cultured in media defined in [Quarta et al., 2016] at 37°C with 5% CO2.

Lentiviral Transduction. Luciferase and GFP protein reporters were subcloned into a third generation HIV-1 lentiviral vector (CD51X DPS, SystemBio). Cells were incubated with 5 μl of concentrated virus per well and 8 μg/mL polybrene (Santa Cruz

Biotechnology). Plates were spun for 5 min at 3200g, and for 1 hour at 2500g at 25°C.

Cells were then washed with fresh media two times, scraped from plates, and re- suspended in the final volume according to the experimental conditions.

Bioluminescence Imaging. Bioluminescent imaging was performed using the Xenogen

IVIS-Spectrum System (Caliper Life Sciences). Mice were anesthetized using 2% isoflurane at a flow rate of 2.5 l/min and injected intraperitoneal with D-Luciferin (50 mg/ml) dissolved in sterile PBS before imaging. Analysis was performed using Living

Image Software.

Tetanic force measurement. To measure the force, we isolated the TA in a bath of oxygenated Ringer’s solution and stimulated it with plate electrodes. Immediately after euthanasia, the distal tendon of the TA, the TA, and the knee (proximal tibia, distal femur, patella, and associated soft tissues) were dissected out and placed in Ringer’s solution (Sigma) maintained at 25°C with bubbling oxygen with 5% carbon dioxide. The proximal tibia was sutured to a rigid wire attached to the force transducer and the distal tendon was sutured to a rigid fixture. No suture loops or slack was present in the system.

77

The contralateral limb was immediately dissected and kept under low passive tension in oxygenated Ringer’s solution bath until measurement. Supramaximal stimulation voltage was found and the active force-length curve was measured in a manner similar to the in vivo condition. After measurement, the muscle was dissected free and the mass measured. An Aurora Scientific 1300-A Whole Mouse Test System was used to gather force production data.

Histology. TA muscles were carefully dissected weighed, and fixed for 5 hours using

0.5% paraformaldehyde and subsequently transferred to 20% sucrose overnight. Muscles were then frozen in OCT, cryosectioned at a thickness of 10 μm, and stained. For colorimetric staining with Hematoxylin and Eosin (Sigma) or Gomorri Trichrome

(Richard-Allan Scientific). For immune staining, tissue was stained against anti-GFP and

DAPI. Slides were imaged using Zeiss fluorescence microscope and analyzed using accompanying software.

Antibodies. The following antibodies were used in this study. The source of each antibody is indicated. GFP (Invitrogen, 1:250); Luciferase (Sigma-Aldrich, 1:200);

Collagen I (Cedarlane Labs, , 1:200); HSP47 (Abcam, 1:200), Laminin (Abcam, 1:1000), anti-CD31-Alexa Fluor 488 (clone WM59; BioLegend; 1:75), anti-CD45-Alexa Fluor

488 (clone HI30; Invitrogen; 1:75), anti-CD34-FITC (clone 581; BioLegend; 1:75), anti-

CD29-APC (clone TS2/16; BioLegend; 1:75) and anti-NCAM-Biotin (clone HCD56;

BioLegend; 1:75), anti- CD31-Alexa Fluor 488 (clone WM59; BioLegend; 1:75), anti-

CD45-Alexa Fluor 488 (clone HI30; Invitrogen; 1:75), anti-CD34-FITC (clone 581;

78

BioLegend; 1:75), anti-CD29-APC (clone TS2/16; BioLegend; 1:75) and anti-NCAM- iotin (clone HCD56; BioLegend; 1:75).

7 Conclusions

7.1 Concepts

This body of work has demonstrated that transient reprogramming, expressing reprogramming factors for a duration just short enough to retain cell identity, can revert a number of aging and age related disease phenotypes. The underlying trend that seems to be supported by this work is that the early reprogramming mechanisms predominantly focus on improving health and fitness in each cell first, so that they can proceed with the de-differentiation process; though it is possible, though unlikely, that the two processes are unrelated and the former just occurs at a faster rate than the later. To demonstrate the generalizability of the technology, we analyzed 5 different treated cell types from aged and, in some cases, diseased human donors. We showed improvements at the DNA, metabolic, cell niche and tissue levels, demonstrating this epigenetic intervention has dramatic consequences. We utilized an mRNA approach to provide a foot-print free, temporally regulatable method and verified retention of cell identity signature and function integration into tissue, making the approach more palatable for clinical translation than viral and genome integration approaches. We also investigated potential key targets for widespread impact, like the treatment of stem, senescent and secretory cell types.

79

7.2 Future Directions

There are still many aspects that have to be developed and optimized for an actual therapeutic. We alluded to them in each section and we summarize them here. 1) There is a sustained concern of transformation. Though the risk for dedifferentiation is lowered, it is still present and moreover the reprogramming factors themselves containing proto- oncogenes is fundamental issue. Thus transitioning to a cocktail that truly extricates the rejuvenation from any form or transformation, will likely be a necessity for a FDA approved clinical therapy. To gain insight into this potential 2.0 cocktail, as mentioned, we are opening the black box of this transient reprogramming process and studying the genes transiently engaged during this progression. This involves studying the time breadth of both transiently reprogrammed aged and young cells, as the genes engaged in both can be considered the de-differentiation genes and subtracting these out leaves those only engaged in the former and thus potential candidates for the rejuvenation mechanism.

There are still many additional complexities with this, like the question of granularity of sampling (daily, half day or more frequent) depending on how fast these transient genes are expressed as wells as the synchrony of the changes between young and old cohorts and even within the cohorts. Thus, this work is a largely exploratory effort which will develop and expand as we go. 2) Delivery of mRNAs in vivo, though possible, still has a large variability which will need to be controlled. The degradation of the mRNA works negatively in this regard. Though it prevents a runaway reprogramming, it also requires multiple transfections of the same cells. Thus transfection efficiency must high, since if we at minimum assuming the likelihood of each transfection is uncorrelated, the probability of the same cell being hit repeatedly reduces exponentially. Towards this aim

80 we are developing alternative solutions, like self-replicating mRNAs and inducible AAV solutions (generally more accepted than other viral vectors). 3) Any cell intrinsic mechanism will always have to compete the extracellular aged environment still remains an outstanding challenge. Previous in vivo work has been in transgenic mice, where expression has been guaranteed throughout the body. Long term, cyclic expression of these reprogramming factors was strong enough to eventually overcome the artificially aged progeria environment but may not be able to do so in a true aged environment.

Furthermore, it is highly unlikely that this therapy would ever be cleared as a transgeneic modification of embryo’s, a volatile topic itself regardless of which genes are altered.

Thus it will never be the case where every cell in the body is transiently reprogramming and synergizing to overcome the environment, but rather cells will be treated in vitro and transplanted cells or specific parts of tissues will be targeted with high efficiency in situ.

How these localized interventions interplay with the aged environment is largely unexplored. The majority of our transplantation experiments were into young hosts but the last set of experiments transplanting into aged dystrophic mice may be a promising sign for an enduring rejuvenation effect. However it is key to note that we injured virtually the entire TA muscle fibers with our mycotoxin injection to clear out a niche and regenerate the tissue. Thus the transplanted treated/untreated cells really remake the local microniche with their youthful/aged phenotype. This much tissue of course cannot be cleared when treating a human patient. Furthermore, we’ve only observed the early period of recovery after injury and transplantation, so likely the local microniche created is still the dominant effect but the aged systemic environment will take over. Thus more work needs to be conducted on the effects and persistence of treating different scales of

81 cells. 4) In all the applications we tested with mRNAs, the treatement period was one stretch followed by relaxation. However work by the Belmonte group used repeated cyclic application (2 days reprogramming 5 days off) was necessary for the accumulated tissue level benefits in this progeria model. While the use of mNRAs and transplant models might differ in their dynamics, their still would be improved by a cyclic administration for compounding benefits. Because of our focus on human cells and the limitations of in vivo mRNA delivery, our treatment was always in vitro, and in all cases we wanted to avoid the artifacts of long time in culture – extensive passaging, in vitro aging, progression away from the in vivo phenotype etc. Thus our previous studies were not permissible to repeated cyclic administration. As we now develop the technology for in vivo delivery, exploring this extended regimens will likely be the modus operandi in which to evaluate the benefits of treatment, as given the strong aging biased environment, repeated treatment will likely be a necessity. This will of course further raise concerns of accumulate progression of dedifferentiation or transformation with each cycle, which will need to be addressed. There is also potentially a trend of diminishing returns after a threshold of cycles. However, one rather optimistic hypothesis is that the repeated application drives the body to maintain a youthful homeostasis after which treatment can be suspended and potential applied after a months or years later. Ultimately homeostasis is the true rejuvenation, on that has yet to be shown by any anti-aging intervention. 5)

The effects that are weaker or not addressed with transient reprogramming, like DNA damage or telomere attrition, may be complemented by other solutions. In addition developing technologies like senolyitcs and young blood factors may synergize with developing that transient reprogramming. Yet combining these technologies may also

82 lead to destructive interference or possibly a transformative state and runaway growth.

Thus developing optimal combination and timing is crucial and will require specialized expertise as these are all new and emerging technologies. Yet this will be a necessity as none of these technologies is completely comprehensive and likely the eventual holistic therapy will require a panel of interventions in an optimized treatment plan.

8. Supplementary Figures

A)

B) C)

83

Supplementary Figure 1 Demonstration of in vivo, in situ transfection A) and B) demonstrate high transfection of GFP mRNAs to hypothalmus and retina respectively while C) demonstrate a combination of mRNAs can be delivered as in the case of muscle with both GFP and mCherry mRNAs

A) B)

C) D)

84

E) F)

G)

Supplementary Figure 2 Comparison of Shorter vs Extended Treamtment (Fibroblast) Each assay was tested with both two and four days of reprogramming, both with two days of relaxation. In general the longer period of treatement lead to the stronger effect.

85

A) B)

C) D)

86

D) E)

Supplementary Figure 3 Comparison of Shorter vs Extended Treamtment (Endothelial Cell) Each assay was tested with both two and four days of reprogramming, both with two days of relaxation. In general the longer period of treatement lead to the stronger effect.

A) B) C)

87

D) E) F)

G)

Supplementary Figure 4 Effects of control transfections (Fibroblasts). Cells were subjected to same treatment but with control (GFP) mRNAs for 4 day reprogramming and 2 day relaxation time point. Generally accumulated stress of transfection diminished the state of the cells, hence the baseline of comparison used to established efficacy were the untransfected aged cells

88

A) B) C)

D) E) F) Supplementary Figure 5 Effects of control transfections (Endothelial Cells). Cells were subjected to same treatment but with control (GFP) mRNAs for 4 day reprogramming and 2 day relaxation time point. Generally accumulated stress of transfection diminished the state of the cells, hence the baseline of comparison used to established efficacy were the untransfected aged cells

89

A) B)

C) D)

90

E) F)

G) H)

Supplementary Figure 6 Time course study of retentiation of rejuvenative benefits (Fibroblasts). Cells revcieving four days of treatment were cultured out and allowed to reslax for 2, 4 and 6 days. Effects were buy and large still significant out into day 6 through with some reduction.

91

A) B)

C) D)

92

E) F)

Supplementary Figure 7 Time course study of retentiation of rejuvenative benefits (Endothelial Cells). Cells revcieving four days of treatment were cultured out and allowed to reslax for 2, 4 and 6 days. Effects were buy and large still significant out into day 6 through with some reduction.

Supplementary Figure 8 Example single patient distribution comparison. Distribution data was accumulated for each patient in the treated aged and young cohorts. Depicted is an example showing the distribution wide shift in H3K9me3 for fibroblasts, a general trend amongst the parameters showing a change, hinting at a broadly engaged rejuvenative effect. The lefttmost plot is an example of the young distribution.

93

9 References

Akiyama, T., Shinzawa, M., & Akiyama, N. (2012). RANKL-RANK interaction in immune regulatory systems. World journal of orthopedics, 3(9), 142.

Alvarez‐Garcia, O., Matsuzaki, T., Olmer, M., Masuda, K., & Lotz, M. K. (2017). Age‐ related reduction in the expression of FOXO transcription factors and correlations with intervertebral disc degeneration. Journal of Orthopaedic Research, 35(12), 2682-2691.

Al Yacoub, N., Romanowska, M., Haritonova, N., & Foerster, J. (2007). Optimized production and concentration of lentiviral vectors containing large inserts. The journal of gene medicine, 9(7), 579-584.

Antequera, F. (2003). Structure, function and evolution of CpG island promoters. Cellular and Molecular Life Sciences CMLS, 60(8), 1647-1658.

Arakaki, N., Nishihama, T., Owaki, H., Kuramoto, Y., Suenaga, M., Miyoshi, E., ... &

Higuti, T. (2006). Dynamics of mitochondria during the cell cycle. Biological and

Pharmaceutical Bulletin, 29(9), 1962-1965.

Arnette, C., Koetsier, J. L., Hoover, P., Getsios, S., & Green, K. J. (2016). In vitro model of the epidermis: connecting protein function to 3D structure. In Methods in enzymology(Vol. 569, pp. 287-308). Academic Press.

94

Avci-Adali, M., Behring, A., Keller, T., Krajewski, S., Schlensak, C., & Wendel, H. P.

(2014). Optimized conditions for successful transfection of human endothelial cells with in vitro synthesized and modified mRNA for induction of protein expression. Journal of biological engineering, 8(1), 8.

Barnes, D. E., & Lindahl, T. (2004). Repair and genetic consequences of endogenous

DNA base damage in mammalian cells. Annu. Rev. Genet., 38, 445-476.

Barzilai, N., Huffman, D. M., Muzumdar, R. H., & Bartke, A. (2012). The critical role of metabolic pathways in aging. Diabetes, 61(6), 1315-1322.

Bayart, E., & Cohen-Haguenauer, O. (2013). Technological overview of iPS induction from human adult somatic cells. Current gene therapy, 13(2), 73-92.

Blagosklonny, M. V. (2010). Calorie restriction: decelerating mTOR-driven aging from cells to organisms (including humans). Cell Cycle, 9(4), 683-688.

Blagosklonny, M. V. (2010). Revisiting the antagonistic pleiotropy theory of aging:

TOR-driven program and quasi-program. Cell Cycle, 9(16), 3171-3176.

Brack, A. S., & Muñoz-Cánoves, P. (2015). The ins and outs of muscle stem cell aging. Skeletal muscle, 6(1), 1.

95

Broos, K., Van der Jeught, K., Puttemans, J., Goyvaerts, C., Heirman, C., Dewitte, H., ...

& Breckpot, K. (2016). Particle-mediated intravenous delivery of antigen mRNA results in strong antigen-specific T-cell responses despite the induction of Type I interferon. Molecular Therapy-Nucleic Acids, 5, e326.

Buganim, Y., Faddah, D. A., Cheng, A. W., Itskovich, E., Markoulaki, S., Ganz, K., &

Jaenisch, R. (2012). Single-cell expression analyses during cellular reprogramming reveal an early stochastic and a late hierarchic phase. Cell, 150(6), 1209-1222.

Burgstaller, J. P., & Brem, G. (2017). Aging of cloned animals: A mini-review.

Gerontology, 63(5), 417-425.

Calderwood, S. K., Murshid, A., & Prince, T. (2009). The shock of aging: molecular chaperones and the heat shock response in longevity and aging–a mini- review. Gerontology, 55(5), 550-558.

Calvanese, V., Lara, E., Kahn, A., & Fraga, M. F. (2009). The role of epigenetics in aging and age-related diseases. Ageing research reviews, 8(4), 268-276.

Carlson, M. W., Alt‐Holland, A., Egles, C., & Garlick, J. A. (2008). Three‐dimensional tissue models of normal and diseased skin. Current protocols in cell biology, 41(1), 19-9.

96

Canela, A., Vera, E., Klatt, P., & Blasco, M. A. (2007). High-throughput telomere length quantification by FISH and its application to human population studies. Proceedings of the National Academy of Sciences, 104(13), 5300-5305.

Castellano, J. M. (2019). Blood-Based Therapies to Combat Aging. Gerontology, 65(1),

84-89.

Castellano, Joseph M., Elizabeth D. Kirby, and Tony Wyss-Coray. "Blood-borne revitalization of the aged brain." JAMA neurology 72.10 (2015): 1191-1194.

Chakkalakal, J. V., Jones, K. M., Basson, M. A., & Brack, A. S. (2012). The aged niche disrupts muscle stem cell quiescence. Nature, 490(7420), 355.

Challen, G. A., Boles, N. C., Chambers, S. M., & Goodell, M. A. (2010). Distinct hematopoietic stem cell subtypes are differentially regulated by TGF-β1. Cell stem cell, 6(3), 265-278.

Chang, A. M., & Halter, J. B. (2003). Aging and insulin secretion. American Journal of

Physiology-Endocrinology And Metabolism, 284(1), E7-E12.

Chen, D., Shen, J., Zhao, W., Wang, T., Han, L., Hamilton, J. L., & Im, H. J. (2017).

Osteoarthritis: toward a comprehensive understanding of pathological mechanism. Bone research, 5, 16044.

97

Chistiakov, D. A., Sobenin, I. A., Revin, V. V., Orekhov, A. N., & Bobryshev, Y. V.

(2014). Mitochondrial aging and age-related dysfunction of mitochondria. BioMed research international, 2014.

Conboy, I. M., Conboy, M. J., Wagers, A. J., Girma, E. R., Weissman, I. L., & Rando, T.

A. (2005). Rejuvenation of aged progenitor cells by exposure to a young systemic environment. Nature, 433(7027), 760.

Cornelison, D. D. W. (2008). Context matters: in vivo and in vitro influences on muscle satellite cell activity. Journal of cellular biochemistry, 105(3), 663-669.

Corpet, A., & Stucki, M. (2014). Chromatin maintenance and dynamics in senescence: a spotlight on SAHF formation and the epigenome of senescent cells. Chromosoma, 123(5), 423-436.

Costello, A., Lao, N. T., Gallagher, C., Capella Roca, B., Julius, L. A., Suda, S., ... &

Clynes, M. (2018). Leaky expression of the TET‐On system hinders control of endogenous miRNA abundance. Biotechnology journal, 1800219.

Costello, M. J., Brennan, L. A., Basu, S., Chauss, D., Mohamed, A., Gilliland, K. O., ...

& Kantorow, M. (2013). Autophagy and mitophagy participate in ocular lens organelle degradation. Experimental eye research, 116, 141-150.

98

Criscione, S. W., Teo, Y. V., & Neretti, N. (2016). The chromatin landscape of cellular senescence. Trends in Genetics, 32(11), 751-761.

Croes, S. A., & von Bartheld, C. S. (2007). Measurement of contractile force of skeletal and extraocular muscles: effects of blood supply, muscle size and in situ or in vitro preparation. Journal of neuroscience methods, 166(1), 53-65.

D’Aquila, P., Rose, G., Bellizzi, D., & Passarino, G. (2013). Epigenetics and aging. Maturitas, 74(2), 130-136.

Davis, N. L., West, A., Reap, E., MacDonald, G., Collier, M., Dryga, S., ... & Cecil, C.

(2002). Alphavirus replicon particles as candidate HIV vaccines. IUBMB life, 53(4‐5),

209-211.

Dechat, T., Korbei, B., Vaughan, O. A., Vlcek, S., Hutchison, C. J., & Foisner, R. (2000).

Lamina-associated polypeptide 2alpha binds intranuclear A-type lamins. Journal of cell science, 113(19), 3473-3484.

Demaria, M., Desprez, P. Y., Campisi, J., & Velarde, M. C. (2015). Cell autonomous and non-autonomous effects of senescent cells in the skin. Journal of Investigative

Dermatology, 135(7), 1722-1726.

99

Dimri, G. P., Lee, X., Basile, G., Acosta, M., Scott, G., Roskelley, C., ... & Pereira-

Smith, O. (1995). A biomarker that identifies senescent human cells in culture and in aging skin in vivo. Proceedings of the National Academy of Sciences, 92(20), 9363-9367.

Dumont, Nicolas A., Yu Xin Wang, and Michael A. Rudnicki. "Intrinsic and extrinsic mechanisms regulating satellite cell function." Development 142.9 (2015): 1572-1581.

Durruthy-Durruthy, J., Briggs, S. F., Awe, J., Ramathal, C. Y., Karumbayaram, S., Lee,

P. C., ... & Wu, J. C. (2014). Rapid and efficient conversion of integration-free human induced pluripotent stem cells to GMP-grade culture conditions. PloS one, 9(4), e94231.

Durruthy-Durruthy, J., Sebastiano, V., Wossidlo, M., Cepeda, D., Cui, J., Grow, E. J., ...

& Au, K. F. (2016). The primate-specific noncoding RNA HPAT5 regulates pluripotency during human preimplantation development and nuclear reprogramming. Nature genetics, 48(1), 44.

Eichenlaub-Ritter, U., Wieczorek, M., Lüke, S., & Seidel, T. (2011). Age related changes in mitochondrial function and new approaches to study redox regulation in mammalian oocytes in response to age or maturation conditions. Mitochondrion, 11(5), 783-796.

Ellis, J. (2005). Silencing and variegation of gammaretrovirus and lentivirus vectors. Human gene therapy, 16(11), 1241-1246.

100

Ermolaeva, M., Neri, F., Ori, A., & Rudolph, K. L. (2018). Cellular and epigenetic drivers of stem cell ageing. Nature Reviews Molecular Cell Biology, 1.

Evans, W. J. (2010). Skeletal muscle loss: cachexia, sarcopenia, and inactivity. The

American journal of clinical nutrition, 91(4), 1123S-1127S.

Fenton, O. S., Kauffman, K. J., McClellan, R. L., Appel, E. A., Dorkin, J. R., Tibbitt, M.

W., ... & Anderson, D. G. (2016). Bioinspired alkenyl amino alcohol ionizable lipid materials for highly potent in vivo mRNA delivery. Advanced materials, 28(15), 2939-

2943.

Fischle, W., Tseng, B. S., Dormann, H. L., Ueberheide, B. M., Garcia, B. A.,

Shabanowitz, J., ... & Allis, C. D. (2005). Regulation of HP1–chromatin binding by histone H3 methylation and phosphorylation. Nature, 438(7071), 1116.

Fortin, J. P., Triche Jr, T. J., & Hansen, K. D. (2016). Preprocessing, normalization and integration of the Illumina HumanMethylationEPIC array with minfi. Bioinformatics, 33(4), 558-560.

Fraga, M. F., & Esteller, M. (2007). Epigenetics and aging: the targets and the marks. Trends in Genetics, 23(8), 413-418.

101

Freitas, A. A., & De Magalhães, J. P. (2011). A review and appraisal of the DNA damage theory of ageing. Mutation Research/Reviews in Mutation Research, 728(1-2), 12-22.

Freund, A., Orjalo, A. V., Desprez, P. Y., & Campisi, J. (2010). Inflammatory networks during cellular senescence: causes and consequences. Trends in molecular medicine, 16(5), 238-246.

Fulop, T., Larbi, A., Dupuis, G., Le Page, A., Frost, E. H., Cohen, A. A., ... & Franceschi,

C. (2018). Immunosenescence and inflamm-aging as two sides of the same coin: friends or foes?. Frontiers in immunology, 8, 1960.

Garber, K. (2012). At loose ends: telomere theories of aging and cancer begin to converge.

Goto, M. (2008). Inflammaging (inflammation+ aging): a driving force for human aging based on an evolutionarily antagonistic pleiotropy theory?. Bioscience trends, 2(6).

Hajj, K. A., & Whitehead, K. A. (2017). Tools for translation: non-viral materials for therapeutic mRNA delivery. Nature Reviews Materials, 2(10), 17056.

Hanna, J., Saha, K., Pando, B., Van Zon, J., Lengner, C. J., Creyghton, M. P., ... &

Jaenisch, R. (2009). Direct cell reprogramming is a stochastic process amenable to acceleration. Nature, 462(7273), 595.

102

Harding, J., Roberts, R. M., & Mirochnitchenko, O. (2013). Large animal models for stem cell therapy. Stem cell research & therapy, 4(2), 23.

Harman, Denham. "The free radical theory of aging." Antioxidants and Redox

Signaling 5.5 (2003): 557-561.

Healy, Brian C., and David Engler. "Modeling disease‐state transition heterogeneity through Bayesian variable selection." Statistics in medicine 28.9 (2009): 1353-1368.

Helenius, M., Kyrylenko, S., Vehviläinen, P., & Salminen, A. (2001). Characterization of aging-associated up-regulation of constitutive nuclear factor-κB binding activity. Antioxidants and Redox Signaling, 3(1), 147-156.

Hill, M., Wernig, A., & Goldspink, G. (2003). Muscle satellite (stem) cell activation during local tissue injury and repair. Journal of anatomy, 203(1), 89-99.

Hockemeyer, D., Soldner, F., Cook, E. G., Gao, Q., Mitalipova, M., & Jaenisch, R.

(2008). A drug-inducible system for direct reprogramming of human somatic cells to pluripotency. Cell stem cell, 3(3), 346-353.

Horvath, S. (2013). DNA methylation age of human tissues and cell types. Genome biology, 14(10), 3156.

103

Huang, Shu-Gui. "Development of a high throughput screening assay for mitochondrial membrane potential in living cells." Journal of biomolecular screening 7.4 (2002): 383-

389.

Inoue, H., & Yamanaka, S. (2011). The use of induced pluripotent stem cells in drug development. Clinical Pharmacology & Therapeutics, 89(5), 655-661.

Itahana, K., Campisi, J., & Dimri, G. P. (2007). Methods to detect biomarkers of cellular senescence. In Biological Aging(pp. 21-31). Humana Press.

Jason, S. L., & Cui, W. (2016). Proliferation, survival and metabolism: the role of

PI3K/AKT/mTOR signaling in pluripotency and cell fate determination. Development,

143(17), 3050-3060.

Johnson, A. A., Akman, K., Calimport, S. R., Wuttke, D., Stolzing, A., & De Magalhaes,

J. P. (2012). The role of DNA methylation in aging, rejuvenation, and age-related disease. Rejuvenation research, 15(5), 483-494.

Johnson, V. L., & Hunter, D. J. (2014). The epidemiology of osteoarthritis. Best practice

& research Clinical rheumatology, 28(1), 5-15.

104

Jaul, E., & Barron, J. (2017). Age-related diseases and clinical and public health implications for the 85 years old and over population. Frontiers in public health, 5, 335.

Jeon, Ok Hee, Chaekyu Kim, Remi-Martin Laberge, Marco Demaria, Sona Rathod, Alain

P. Vasserot, Jae Wook Chung et al. "Local clearance of senescent cells attenuates the development of post-traumatic osteoarthritis and creates a pro-regenerative environment." Nature medicine 23, no. 6 (2017): 775.

Junnila, R. K., List, E. O., Berryman, D. E., Murrey, J. W., & Kopchick, J. J. (2013). The

GH/IGF-1 axis in ageing and longevity. Nature Reviews Endocrinology, 9(6), 366.

Kang, L., Wang, J., Zhang, Y., Kou, Z., & Gao, S. (2009). iPS cells can support full-term development of tetraploid blastocyst-complemented embryos. Cell stem cell, 5(2), 135-

138.

Kang, X., Yu, Q., Huang, Y., Song, B., Chen, Y., Gao, X., ... & Fan, Y. (2015). Effects of integrating and non-integrating reprogramming methods on copy number variation and genomic stability of human induced pluripotent stem cells. PLoS One, 10(7), e0131128.

Kasum, M., Orešković, S., Čehić, E., Lila, A., Ejubović, E., & Soldo, D. (2018). The role of female obesity on in vitro fertilization outcomes. Gynecological Endocrinology, 34(3),

184-188.

105

Kiefer, J. C. (2011). Primer and interviews: promises and realities of induced pluripotent stem cells. Developmental Dynamics, 240(8), 2034-2041.

Kevei, É., & Hoppe, T. (2014). Ubiquitin sets the timer: impacts on aging and longevity. Nature structural & molecular biology, 21(4), 290.

Khazaei, M., Siddiqui, A., & Fehlings, M. (2015). The potential for iPS-derived stem cells as a therapeutic strategy for spinal cord injury: opportunities and challenges. Journal of clinical medicine, 4(1), 37-65.

Kim, H., Kang, D., Cho, Y., & Kim, J. H. (2015). Epigenetic regulation of chondrocyte catabolism and anabolism in osteoarthritis. Molecules and cells, 38(8), 677.

Kim, T. K., & Eberwine, J. H. (2010). Mammalian cell transfection: the present and the future. Analytical and bioanalytical chemistry, 397(8), 3173-3178.

King, R. S., & Newmark, P. A. (2012). The cell biology of regeneration. J Cell

Biol, 196(5), 553-562.

Kinney, R. M., Johnson, B. J., Welch, J. B., Tsuchiya, K. R., & Trent, D. W. (1989). The full-length nucleotide sequences of the virulent Trinidad donkey strain of Venezuelan equine encephalitis virus and its attenuated vaccine derivative, strain TC-

83. Virology, 170(1), 19-30.

106

Kittan, N. A., Allen, R. M., Dhaliwal, A., Cavassani, K. A., Schaller, M., Gallagher, K.

A., ... & Jarai, G. (2013). Cytokine induced phenotypic and epigenetic signatures are key to establishing specific macrophage phenotypes. PloS one, 8(10), e78045.

Kirkland, J. L., & Tchkonia, T. (2015). Clinical strategies and animal models for developing senolytic agents. Experimental gerontology, 68, 19-25.

Kristariyanto, Y. A., Rehman, S. A. A., Campbell, D. G., Morrice, N. A., Johnson, C.,

Toth, R., & Kulathu, Y. (2015). K29-selective ubiquitin binding domain reveals structural basis of specificity and heterotypic nature of k29 polyubiquitin. Molecular cell, 58(1), 83-94.

Krizhanovsky, V., Xue, W., Zender, L., Yon, M., Hernando, E., & Lowe, S. W. (2008,

January). Implications of cellular senescence in tissue damage response, tumor suppression, and stem cell biology. In Cold Spring Harbor symposia on quantitative biology (Vol. 73, pp. 513-522). Cold Spring Harbor Laboratory Press.

Kwan Tat, S., Amiable, N., Pelletier, J. P., Boileau, C., Lajeunesse, D., Duval, N., &

Martel-Pelletier, J. (2009). Modulation of OPG, RANK and RANKL by human chondrocytes and their implication during osteoarthritis. Rheumatology, 48(12), 1482-

1490.

107

Lapasset, L., Milhavet, O., Prieur, A., Besnard, E., Babled, A., Aït-Hamou, N., ... &

Lehmann, S. (2011). Rejuvenating senescent and centenarian human cells by reprogramming through the pluripotent state. Genes & development, 25(21), 2248-2253.

Leonhardt, C., Schwake, G., Stögbauer, T. R., Rappl, S., Kuhr, J. T., Ligon, T. S., &

Rädler, J. O. (2014). Single-cell mRNA transfection studies: delivery, kinetics and statistics by numbers. Nanomedicine: Nanotechnology, Biology and Medicine, 10(4),

679-688.

Lepetsos, P., & Papavassiliou, A. G. (2016). ROS/oxidative stress signaling in osteoarthritis. Biochimica et Biophysica Acta (BBA)-Molecular Basis of

Disease, 1862(4), 576-591.

Levine, B., & Kroemer, G. (2008). Autophagy in the pathogenesis of disease. Cell, 132(1), 27-42.

Li, X., Xu, H., Xu, C., Lin, M., Song, X., Yi, F., ... & Cao, L. (2013). The yin-yang of

DNA damage response: roles in tumorigenesis and cellular senescence. International journal of molecular sciences, 14(2), 2431-2448.

Liang, Y., Van Zant, G., & Szilvassy, S. J. (2005). Effects of aging on the homing and engraftment of murine hematopoietic stem and progenitor cells. Blood, 106(4), 1479-

1487.

108

Liu, L., Cheung, T. H., Charville, G. W., Hurgo, B. M. C., Leavitt, T., Shih, J., ... &

Rando, T. A. (2013). Chromatin modifications as determinants of muscle stem cell quiescence and chronological aging. Cell reports, 4(1), 189-204.

Liu, T. F., & McCall, C. E. (2013). Deacetylation by SIRT1 reprograms inflammation and cancer. Genes & cancer, 4(3-4), 135-147.

Lu, Y., Futtner, C., Rock, J. R., Xu, X., Whitworth, W., Hogan, B. L., & Onaitis, M. W.

(2010). Evidence that SOX2 overexpression is oncogenic in the lung. PloS one, 5(6), e11022.

Mahmoudi, S., & Brunet, A. (2012). Aging and reprogramming: a two-way street. Current opinion in cell biology, 24(6), 744-756.

Mannella, C. A. (2008). Structural diversity of mitochondria. Annals of the New York

Academy of Sciences, 1147(1), 171-179.

Marędziak, M., Marycz, K., Tomaszewski, K. A., Kornicka, K., & Henry, B. M. (2016).

The influence of aging on the regenerative potential of human adipose derived mesenchymal stem cells. Stem cells international, 2016.

109

Martínez-Calatrava, M. J., Prieto-Potín, I., Roman-Blas, J. A., Tardio, L., Largo, R., &

Herrero-Beaumont, G. (2012). RANKL synthesized by articular chondrocytes contributes to juxta-articular bone loss in chronic arthritis. Arthritis research & therapy, 14(3), R149.

Martínez-Zamudio, R. I., Robinson, L., Roux, P. F., & Bischof, O. (2017). SnapShot: cellular senescence pathways. Cell, 170(4), 816-816.

McHugh, D., & Gil, J. (2018). Senescence and aging: causes, consequences, and therapeutic avenues. J Cell Biol, 217(1), 65-77.

McKinlay, C. J., Vargas, J. R., Blake, T. R., Hardy, J. W., Kanada, M., Contag, C. H., ...

& Waymouth, R. M. (2017). Charge-altering releasable transporters (CARTs) for the delivery and release of mRNA in living animals. Proceedings of the National Academy of

Sciences, 114(4), E448-E456.

Mendel, O. I., Luchihina, L. V., & Mendel, W. (2015). Aging and osteoarthritis. Chronic nonspecific inflammation as a link between aging and osteoarthritis (a review). Advances in Gerontology, 5(4), 252-260.

Miller, J. D., Ganat, Y. M., Kishinevsky, S., Bowman, R. L., Liu, B., Tu, E. Y., ... &

Taldone, T. (2013). Human iPSC-based modeling of late-onset disease via progerin- induced aging. Cell stem cell, 13(6), 691-705.

110

Munoz-Espin, D., & Serrano, M. (2014). Cellular senescence: from physiology to pathology. Nature reviews Molecular cell biology, 15(7), 482.

Nagy, A., & Nagy, K. (2010). The mysteries of induced pluripotency: where will they lead?. Nature methods, 7(1), 22.

Newman, A. B., Sanders, J. L., Kizer, J. R., Boudreau, R. M., Odden, M. C., Zeki Al

Hazzouri, A., & Arnold, A. M. (2016). Trajectories of function and biomarkers with age: the CHS All Stars Study. International journal of epidemiology, 45(4), 1135-1145.

Ocampo, A., Reddy, P., Martinez-Redondo, P., Platero-Luengo, A., Hatanaka, F.,

Hishida, T., ... & Araoka, T. (2016). In vivo amelioration of age-associated hallmarks by partial reprogramming. Cell, 167(7), 1719-1733.

Ohnishi, Kotaro, et al. "Premature termination of reprogramming in vivo leads to cancer development through altered epigenetic regulation." Cell 156.4 (2014): 663-677.

Olsen, T. R., Ng, K. S., Lock, L. T., Ahsan, T., & Rowley, J. A. (2018). Peak MSC—are we there yet?. Frontiers in medicine, 5.

Omokoko, T. A., Luxemburger, U., Bardissi, S., Simon, P., Utsch, M., Breitkreuz, A., ...

& Sahin, U. (2016). Luciferase mRNA transfection of antigen presenting cells permits

111 sensitive nonradioactive measurement of cellular and humoral cytotoxicity. Journal of immunology research, 2016.

O'Neill, Chris. "The epigenetics of embryo development." Animal Frontiers 5.1 (2015):

42-49.

Pang, W. W., Price, E. A., Sahoo, D., Beerman, I., Maloney, W. J., Rossi, D. J., ... &

Weissman, I. L. (2011). Human bone marrow hematopoietic stem cells are increased in frequency and myeloid-biased with age. Proceedings of the National Academy of

Sciences, 108(50), 20012-20017.

Pascolo, S. (2008). Vaccination with messenger RNA (mRNA). In Toll-Like Receptors

(TLRs) and Innate Immunity(pp. 221-235). Springer, Berlin, Heidelberg.

Petralia, R. S., Mattson, M. P., & Yao, P. J. (2014). Aging and longevity in the simplest animals and the quest for immortality. Ageing research reviews, 16, 66-82.

Pekovic, V., Harborth, J., Broers, J. L., Ramaekers, F. C., van Engelen, B., Lammens, M.,

... & Markiewicz, E. (2007). Nucleoplasmic LAP2α–lamin A complexes are required to maintain a proliferative state in human fibroblasts. The Journal of cell biology, 176(2),

163-172.

112

Petralia, R. S., Mattson, M. P., & Yao, P. J. (2014). Communication breakdown: the impact of ageing on synapse structure. Ageing research reviews, 14, 31-42.

Piraino, S., Boero, F., Aeschbach, B., & Schmid, V. (1996). Reversing the life cycle: medusae transforming into polyps and cell transdifferentiation in Turritopsis nutricula

(Cnidaria, Hydrozoa). The Biological Bulletin, 190(3), 302-312.

Podhorecka, M., Skladanowski, A., & Bozko, P. (2010). H2AX phosphorylation: its role in DNA damage response and cancer therapy. Journal of nucleic acids, 2010.

Ponnappan, S., & Ponnappan, U. (2011). Aging and immune function: molecular mechanisms to interventions. Antioxidants & redox signaling, 14(8), 1551-1585.

Poulose, N., & Raju, R. (2015). Sirtuin regulation in aging and injury. Biochimica et

Biophysica Acta (BBA)-Molecular Basis of Disease, 1852(11), 2442-2455.

Quarta, M., Brett, J. O., DiMarco, R., De Morree, A., Boutet, S. C., Chacon, R., ... &

Heilshorn, S. (2016). An artificial niche preserves the quiescence of muscle stem cells and enhances their therapeutic efficacy. Nature biotechnology, 34(7), 752.

Ralphs, J. R., & Benjamin, M. (1994). The joint capsule: structure, composition, ageing and disease. Journal of anatomy, 184(Pt 3), 503.

113

Ramathal, C. Y., Durruthy-Durruthy, J., Pera, R. A. R., & Orwig, K. E. (2016). U.S.

Patent Application No. 14/904,396.

Ramsay, J. E., Ferrell, W. R., Crawford, L., Wallace, A. M., Greer, I. A., & Sattar, N.

(2002). Maternal obesity is associated with dysregulation of metabolic, vascular, and inflammatory pathways. The Journal of Clinical Endocrinology & Metabolism, 87(9),

4231-4237.

Rauser, B. S. (2017). Direct conversion of somatic cells utilizing CRISPR/Cas9 (Doctoral dissertation, Universität München).

Rebo, J., Mehdipour, M., Gathwala, R., Causey, K., Liu, Y., Conboy, M. J., & Conboy, I.

M. (2016). A single heterochronic blood exchange reveals rapid inhibition of multiple tissues by old blood. Nature communications, 7, 13363.

Rogakou, E. P., Pilch, D. R., Orr, A. H., Ivanova, V. S., & Bonner, W. M. (1998). DNA double-stranded breaks induce histone H2AX phosphorylation on serine 139. Journal of biological chemistry, 273(10), 5858-5868.

Rubinsztein, D. C., Mariño, G., & Kroemer, G. (2011). Autophagy and aging. Cell, 146(5), 682-695.

114

Sahin, E., & DePinho, R. A. (2012). Axis of ageing: telomeres, p53 and mitochondria. Nature reviews Molecular cell biology, 13(6), 397.

Salminen, A., & Kaarniranta, K. (2012). AMP-activated protein kinase (AMPK) controls the aging process via an integrated signaling network. Ageing research reviews, 11(2),

230-241.

Salvioli, S., Monti, D., Lanzarini, C., Conte, M., Pirazzini, C., Giulia Bacalini, M., ... &

Bucci, L. (2013). Immune system, cell senescence, aging and longevity-inflamm-aging reappraised. Current pharmaceutical design, 19(9), 1675-1679.

Sandell, L. J., & Aigner, T. (2001). Articular cartilage and changes in arthritis: cell biology of osteoarthritis. Arthritis Research & Therapy, 3(2), 107.

Saunders, L. R., & Verdin, E. (2007). Sirtuins: critical regulators at the crossroads between cancer and aging. Oncogene, 26(37), 5489.

Scaffidi, P., & Misteli, T. (2006). Lamin A-dependent nuclear defects in human aging. Science, 312(5776), 1059-1063.

Seto, E., & Yoshida, M. (2014). Erasers of histone acetylation: the histone deacetylase enzymes. Cold Spring Harbor perspectives in biology, 6(4), a018713.

115

Shammas, M. A. (2011). Telomeres, lifestyle, cancer, and aging. Current opinion in clinical nutrition and metabolic care, 14(1), 28.

Sharma, R., Agarwal, A., Rohra, V. K., Assidi, M., Abu-Elmagd, M., & Turki, R. F.

(2015). Effects of increased paternal age on sperm quality, reproductive outcome and associated epigenetic risks to offspring. Reproductive Biology and Endocrinology, 13(1),

35.

Shay, J. W., & Wright, W. E. (2000). Hayflick, his limit, and cellular ageing. Nature reviews Molecular cell biology, 1(1), 72.

Shen, J., Gammon, M. D., Terry, M. B., Wang, Q., Bradshaw, P., Teitelbaum, S. L., ... &

Santella, R. M. (2009). Telomere length, oxidative damage, antioxidants and breast cancer risk. International journal of cancer, 124(7), 1637-1643.

Shi, P., Sun, L. L., Lee, Y. S., & Tu, Y. (2017). Electroacupuncture regulates the stress- injury-repair chain of events after cerebral ischemia/reperfusion injury. Neural regeneration research, 12(6), 925.

Sommer, C. A., Stadtfeld, M., Murphy, G. J., Hochedlinger, K., Kotton, D. N., &

Mostoslavsky, G. (2009). Induced pluripotent stem cell generation using a single lentiviral stem cell cassette. Stem cells, 27(3), 543-549.

116

Srinivasan, P. C. (2013). The role of inflammatory cytokines and the RANKL-RANK-

OPG molecular triad in periodontal bone loss-a review. J Clin Cell Immunol S, 13, 2.

Sriram, G., Bigliardi, P. L., & Bigliardi-Qi, M. (2015). Fibroblast heterogeneity and its implications for engineering organotypic skin models in vitro. European journal of cell biology, 94(11), 483-512.

Stadtfeld, M., Maherali, N., Breault, D. T., & Hochedlinger, K. (2008). Defining molecular cornerstones during fibroblast to iPS cell reprogramming in mouse. Cell stem cell, 2(3), 230-240.

Studer, L., Vera, E., & Cornacchia, D. (2015). Programming and reprogramming cellular age in the era of induced pluripotency. Cell stem cell, 16(6), 591-600.

Sugrue, M. M., & Tatton, W. G. (2001). Mitochondrial membrane potential in aging cells. Neurosignals, 10(3-4), 176-188.

Suhr, S. T., Chang, E. A., Tjong, J., Alcasid, N., Perkins, G. A., Goissis, M. D., ... &

Cibelli, J. B. (2010). Mitochondrial rejuvenation after induced pluripotency. PloS one, 5(11), e14095.

Szeto, H. H. (2006). Cell-permeable, mitochondrial-targeted, peptide antioxidants. The

AAPS journal, 8(2), E277-E283.

117

Tanaka, K. (2009). The proteasome: overview of structure and functions. Proceedings of the Japan Academy, Series B, 85(1), 12-36.

Takahashi, K., Tanabe, K., Ohnuki, M., Narita, M., Ichisaka, T., Tomoda, K., &

Yamanaka, S. (2007). Induction of pluripotent stem cells from adult human fibroblasts by defined factors. cell, 131(5), 861-872.

Tchetina, E. V. (2010). Developmental mechanisms in articular cartilage degradation in osteoarthritis. Arthritis, 2011.

T Das, A., Tenenbaum, L., & Berkhout, B. (2016). Tet-On systems for doxycycline- inducible gene expression. Current gene therapy, 16(3), 156-167.

Touleimat, N., & Tost, J. (2012). Complete pipeline for Infinium® Human Methylation

450K BeadChip data processing using subset quantile normalization for accurate DNA methylation estimation. Epigenomics, 4(3), 325-341.

Tsurumi, A., & Li, W. (2012). Global heterochromatin loss: a unifying theory of aging?. Epigenetics, 7(7), 680-688.

118

Vakoc, C. R., Mandat, S. A., Olenchock, B. A., & Blobel, G. A. (2005). Histone H3 lysine 9 methylation and HP1γ are associated with transcription elongation through mammalian chromatin. Molecular cell, 19(3), 381-391.

Vidak, S., Kubben, N., Dechat, T., & Foisner, R. (2015). Proliferation of progeria cells is enhanced by lamina-associated polypeptide 2α (LAP2α) through expression of extracellular matrix proteins. Genes & development, 29(19), 2022-2036.

Villalba, J. M., & Alcaín, F. J. (2012). Sirtuin activators and inhibitors. Biofactors, 38(5),

349-359.

Vuolteenaho, K., Moilanen, T., Knowles, R. G., & Moilanen, E. (2007). The role of nitric oxide in osteoarthritis. Scandinavian journal of rheumatology, 36(4), 247-258.

Warren, L., Manos, P. D., Ahfeldt, T., Loh, Y. H., Li, H., Lau, F., ... & Daley, G. Q.

(2010). Highly efficient reprogramming to pluripotency and directed differentiation of human cells with synthetic modified mRNA. Cell stem cell, 7(5), 618-630.

Wei, D., Kanai, M., Huang, S., & Xie, K. (2005). Emerging role of KLF4 in human gastrointestinal cancer. Carcinogenesis, 27(1), 23-31.

119

Wei, W., Hemmer, R. M., & Sedivy, J. M. (2001). Role of p14ARF in replicative and induced senescence of human fibroblasts. Molecular and cellular biology, 21(20), 6748-

6757.

White, R. R., & Vijg, J. (2016). Do DNA double-strand breaks drive aging?. Molecular cell, 63(5), 729-738.

Willadsen, S. M. (1986). Nuclear transplantation in sheep embryos. Nature, 320(6057),

63.

Williams, L. A., Davis-Dusenbery, B. N., & Eggan, K. C. (2012). SnapShot: directed differentiation of pluripotent stem cells. Cell, 149(5), 1174-1174.

Willows, R., Navaratnam, N., Lima, A., Read, J., & Carling, D. (2017). Effect of different γ subunit isoforms on the regulation of AMPK. Biochemical Journal,

BCJ20170046.

Wu, J. C., Sundaresan, G., Iyer, M., & Gambhir, S. S. (2001). Noninvasive optical imaging of firefly luciferase reporter gene expression in skeletal muscles of living mice. Molecular Therapy, 4(4), 297-306.

Xia, S., Zhang, X., Zheng, S., Khanabdali, R., Kalionis, B., Wu, J., ... & Tai, X. (2016).

An update on inflamm-aging: mechanisms, prevention, and treatment. Journal of immunology research, 2016.

120

Xiang, G., Yang, L., Long, Q., Chen, K., Tang, H., Wu, Y., ... & Liu, W. (2017).

BNIP3L-dependent mitophagy accounts for mitochondrial clearance during 3 factors- induced somatic cell reprogramming. Autophagy, 13(9), 1543-1555.

Xu, B., Zhang, K., & Huang, Y. (2009). Lin28 modulates cell growth and associates with a subset of cell cycle regulator mRNAs in mouse embryonic stem cells. Rna, 15(3), 357-

361.

Xu, M., Pirtskhalava, T., Farr, J. N., Weigand, B. M., Palmer, A. K., Weivoda, M. M., ...

& Onken, J. L. (2018). Senolytics improve physical function and increase lifespan in old age. Nature medicine, 24(8), 1246.

Xu, X., Wilschut, K. J., Kouklis, G., Tian, H., Hesse, R., Garland, C., ... & Hoffman, W.

Y. (2015). Human satellite cell transplantation and regeneration from diverse skeletal muscles. Stem cell reports, 5(3), 419-434.

Yagi, T., Kosakai, A., Ito, D., Okada, Y., Akamatsu, W., Nihei, Y., ... & Okano, H.

(2012). Establishment of induced pluripotent stem cells from centenarians for neurodegenerative disease research. PloS one, 7(7), e41572.

121

Yamaguchi, T., Hamanaka, S., Kamiya, A., Okabe, M., Kawarai, M., Wakiyama, Y., ...

& Kato-Itoh, M. (2012). Development of an all-in-one inducible lentiviral vector for gene specific analysis of reprogramming. PLoS One, 7(7), e41007.

Yamanaka, S. (2009). Elite and stochastic models for induced pluripotent stem cell generation. Nature, 460(7251), 49.

Yang, Z., Fujii, H., Mohan, S. V., Goronzy, J. J., & Weyand, C. M. (2013).

Phosphofructokinase deficiency impairs ATP generation, autophagy, and redox balance in rheumatoid arthritis T cells. Journal of Experimental Medicine, 210(10), 2119-2134.

Yasmin, R., Siraj, S., Hassan, A., Khan, A. R., Abbasi, R., & Ahmad, N. (2015).

Epigenetic regulation of inflammatory cytokines and associated genes in human malignancies. Mediators of inflammation, 2015.

Yoshioka, N., Gros, E., Li, H. R., Kumar, S., Deacon, D. C., Maron, C., ... & Dowdy, S.

F. (2013). Efficient generation of human iPSCs by a synthetic self-replicative RNA. Cell stem cell, 13(2), 246-254.

Young, H. E., & Black Jr, A. C. (2004). Adult stem cells. The Anatomical Record Part A:

Discoveries in Molecular, Cellular, and Evolutionary Biology: An Official Publication of the American Association of Anatomists, 276(1), 75-102.

122

Yousefi, S., Mihalache, C., Kozlowski, E., Schmid, I., & Simon, H. U. (2009). Viable neutrophils release mitochondrial DNA to form neutrophil extracellular traps. Cell death and differentiation, 16(11), 1438.

Yuan, Y., Cruzat, V. F., Newsholme, P., Cheng, J., Chen, Y., & Lu, Y. (2016).

Regulation of SIRT1 in aging: Roles in mitochondrial function and biogenesis. Mechanisms of ageing and development, 155, 10-21.

Zaim, M., Karaman, S., Cetin, G., & Isik, S. (2012). Donor age and long-term culture affect differentiation and proliferation of human bone marrow mesenchymal stem cells.

Annals of hematology, 91(8), 1175-1186.

Zhang, Y., Chen, X., Zhao, Y., Ponnusamy, M., & Liu, Y. (2017). The role of ubiquitin proteasomal system and autophagy-lysosome pathway in Alzheimer’s disease. Reviews in the Neurosciences, 28(8), 861-868.

Zhou, X., Vink, M., Klaver, B., Berkhout, B., & Das, A. T. (2006). Optimization of the

Tet-On system for regulated gene expression through viral evolution. Gene therapy, 13(19), 1382.

Zunder, E. R., Lujan, E., Goltsev, Y., Wernig, M., & Nolan, G. P. (2015). A continuous molecular roadmap to iPSC reprogramming through progression analysis of single-cell mass cytometry. Cell Stem Cell, 16(3), 323-337.