A Systematic Analysis of Nuclear Heat Shock Protein 90 (Hsp90) Reveals A

Max Planck Institute of Immunobiology und Epigenetics Freiburg im Breisgau

A systematic analysis of nuclear Heat Shock Protein 90 (Hsp90) reveals a novel transcriptional regulatory role mediated by its interaction with Host Cell Factor-1 (HCF-1)

Inaugural-Dissertation

to obtain the Doctoral Degree Faculty of Biology, Albert-Ludwigs-Universität Freiburg im Breisgau

presented by Aneliya Antonova born in Bulgaria

Freiburg im Breisgau, Germany March 2019

Dekanin: Prof. Dr. Wolfgang Driever Promotionsvorsitzender: Prof. Dr. Andreas Hiltbrunner

Betreuer der Arbeit:

Referent: Dr. Ritwick Sawarkar Koreferent: Prof. Dr. Rudolf Grosschedl Drittprüfer: Prof. Dr. Andreas Hecht

Datum der mündlichen Prüfung: 27.05.2019

AFFIDAVIT

I herewith declare that I have prepared the present work without any unallowed help from third parties and without the use of any aids beyond those given. All data and concepts taken either directly or indirectly from other sources are so indicated along with a notation of the source. In particular I have not made use of any paid assistance from exchange or consulting services (doctoral degree advisors or other persons). No one has received remuneration from me either directly or indirectly for work which is related to the content of the present dissertation.

The work has not been submitted in this country or abroad to any other examination board in this or similar form.

The provisions of the doctoral degree examination procedure of the faculty of Biology of the University of Freiburg are known to me. In particular I am aware that before the awarding of the final doctoral degree I am not entitled to use the title of Dr.

Aneliya Antonova

iii

ACKNOWLEDGEMENTS

First of all, I would like to express my gratitude to my supervisor Dr Ritwick Sawarkar for giving me the opportunity to carry out this challenging project in his lab. Thank you for guiding my way from a graduate to an independent and responsible scientist. I learned a lot from you!

I am very grateful to all the former and present members of the lab for their scientific and moral support. Dear Parul, thank you for being such a supportive partner in that tough journey, a great friend and a valuable example. Barbara, Prashant, Fernando, Sergio, Erik, and Rebecca, I am glad I met every single one of you.

I wish to thank the members of my thesis committee Prof Dr Rudolf Grosschedl and Prof Dr Andreas Hecht for their guidance. I would also like to acknowledge Dr Magdalena Baer- Rademacher, Dr Monika Lachner, Dr. Kyle Austin and Lisa Breitner, the coordinators of the International Max Planck Research School for Molecular and Cellular Biology (IMPRS- MCB) for always being there to help.

I would like to thank my family and friends, who were so supportive and motivating during these years. My very special thanks go to my husband Nikolay Antonov who was always there to share my excitement in good times and hold my hand in bad days. Thank you for believing in me! And finally, thank you my little Thoma for being the sunshine in my days. I cannot wait to hear you saying “My mom is a scientist”.

CONTENTS

List of figures………………………………………………………………………………..viii

List of tables…………………………………………………………………………………...x

Abbreviations…………………………………………………………………………………xi

Abstract……………………………………………………………………………………...xiii

Zusammenfassung…………………………………………………………………………...xv

CHAPTER I Systematic analysis of nuclear heat shock protein 90 (Hsp90) 1. Introduction ...... 1 1.1. Molecular chaperones ...... 1

1.2. Hsp90 ...... 3

1.2.1. Structure and conformational cycle ...... 4 1.2.2. Function ...... 4 1.2.3. Regulation of Hsp90 ...... 7 1.2.4. Chaperone cycle ...... 13 1.2.5. Hsp90 isoforms ...... 15 1.1.6. Hsp90 and cancer ...... 16 1.2.7. Interactome studies ...... 18 1.2.9. Nuclear Hsp90 ...... 23 1.2.9. Challenges and open questions ...... 27 2. Aims ...... 30 2.1. Nuclear Hsp90α and Hsp90β interactomes in human cells...... 30

2.2. Nuclear Hsp90α and Hsp90β PTMs...... 30

3. Materials and Methods ...... 31 3.1. Cell Culture and Cell Lines ...... 31

3.2. Antibodies ...... 31

3.3. Cell Fractionation ...... 31

3.4. SDS-PAGE ...... 33

3.5. Silver staining ...... 33

CONTENTS

3.6. Western Blotting ...... 34

3.7. Nuclear Protein Affinity Purification coupled to Mass Spectrometry ...... 34

3.7.1 Stable Cell Lines Generation ...... 34 3.7.2 Nuclear Protein Extract Preparation ...... 35 3.7.3 Protein Affinity Purification ...... 35 3.7.4. Protein Digestion and Mass Spectrometry Analysis ...... 36 3.8. SILAC labelling ...... 37

3.9. Di-Gly Immunoprecipitation ...... 37

3.10. RNA isolation ...... 38

3.11. Immunofluorescence ...... 38

3.12. Calcium Phosphate Transfection ...... 39

3.13. Synthetic Lethality Screen ...... 39

4. Results ...... 40 4.1. Nuclear Hsp90 physical interactome investigated by affinity purification ...... 40

4.1.1. Nuclear Hsp90α interactions in normal growth conditions ...... 42 4.1.2. Nuclear Hsp90β interactions in normal conditions ...... 51 4.1.3. Nuclear Hsp90α and Hsp90β interactions in stress conditions ...... 53 4.2. Nuclear Hsp90 interactome investigated by revealing Hsp90-dependent ubiquitylated proteome ...... 56

4.3. Hsp90 functional interactome investigation ...... 72

4.4. Hsp90 PTMs ...... 73

4.5. Summary ...... 75

5. Discussion and Outlook ...... 77 5.1. Nuclear Hsp90 interaction network ...... 77

5.1.1. Collaboration with nuclear protein folding and degradation machineries ...... 77 5.1.2. Role in gene expression regulation...... 80 5.1.3. RNA metabolism ...... 81 5.1.3. Novel opportunities of anti-cancer therapies ...... 84 5.1.4. Hsp90 during proteotoxic stress ...... 85

CONTENTS

5.2. Nuclear Hsp90 PTMs ...... 85

CHAPTER II Hsp90 chaperones the transcriptional co-factor Host Cell Factor 1 (HCF-1) 1. Introduction ...... 90 1.1. HCF-1 role in HSV gene expression ...... 90

1.2. HCF-1 protein structure ...... 90

1.3. HCF-1 polypeptide cleavage ...... 92

1.4. HCF-1 role in cell cycle ...... 93

1.5. HCF-1 interaction partners ...... 94

2. Aims ...... 96 3. Materials and Methods ...... 97 3.1. Cell Culture and Cell Lines ...... 97

3.2. siRNA-mediated protein knockdown ...... 97

3.3. ChIP-qPCR and ChIP-seq ...... 97

3.3.1. Preparation of chromatin ...... 97 3.3.2 Chromatin Immunoprecipitation ...... 99 3.3.4. ChIP-qPCR ...... 99 3.3.5. ChIP-seq data analysis (done by Barbara Hummel, MPI-IE, Freiburg) ...... 99 3.4. Chem-seq ...... 100

4. Results ...... 101 4.1. HCF-1 and Hsp90 share multiple common interactors ...... 101

4.2. HCF-1 is a client of Hsp90 ...... 102

4.3. Hsp90 and HCF-1 co-localize at chromatin ...... 103

4.4. HCF-1 chromatin binding and effect on gene expression is Hsp90-dependent ...... 105

5. Discussion and Outlook ...... 107 CONCLUDING REMARKS ...... 110 REFERENCES ...... 113 APPENDIX ...... 137

vii

LIST OF FIGURES

Figure 1. Characteristics of different chaperone machineries Figure 2. Models for chaperone-assisted folding of newly synthesized polypeptides in eukaryotes. Figure 3. Structure, conformational cycle and function of Hsp90. Figure 4. Hsp90 PTMs. Figure 5. Hsp90 chaperone cycle illustrated by the stabilization of steroid receptors. Figure 6. Hsp90 chaperones diverse transcription factors and epigenetic modifiers. Figure 7. Nuclear Hsp90 affinity purification. Figure 8. Nuclear Hsp90α interactome analysis. Figure 9. Overview of nuclear Hsp90α interaction network. Figure 10. Nuclear Hsp90α interaction network GO Enrichment Analysis. Figure 11. Nuclear Hsp90α interacts with several chromatin remodelling/ modifying complexes. Figure 12. Nuclear Hsp90β interactome analysis. Figure 13. Overview of nuclear Hsp90β interaction network. Figure 14. Nuclear Hsp90α (A) and Hsp90β (B) affinity purification upon heat shock – experimental replicates overlap. Figure 15. Investigation of Hsp90-dependent ubiquitylated proteome. Figure 16. Hsp90 and proteasomal inhibition effect on cellular proteins ubiquitylation. Figure 17. Hsp90 and proteasomal inhibition effect on nuclear proteins ubiquitylation. Figure 18. Hsp90 and proteasomal inhibition effect on cellular proteins K48 poly- ubiquitylation Figure 19. Different stresses effect on nuclear proteins K48 poly-ubiquitylation Figure 20. SILAC-based MS identification of insoluble nuclear proteins upon Hsp90 and proteasomal inhibition. Figure 21. Aggresome formation upon simultaneous Hsp90 and proteasome inhibition. Figure 22. Post-translational modification sites of nuclear Hsp90α and Hsp90β in normal growth and stress conditions. Figure 23. Role of VP16-induced complex in the herpes simplex virus (HSV) lytic cycle. Figure 24. HCF-1 protein structure. Figure 25. HCF-1 is a client of Hsp90.

viii

LIST OF FIGURES

Figure 22. Post-translational modification sites of nuclear Hsp90α and Hsp90β in control and stress conditions. Figure 23. Role of VP16-induced complex in the herpes simplex virus (HSV) lytic cycle. Figure 24. HCF-1 protein structure. Figure 25. HCF-1 is a client of Hsp90. Figure 26. Hsp90 and HCF-1 co-localize at chromatin. Figure 27. HCF-1 association to chromatin and expression of corresponding genes is Hsp90- dependent.

LIST OF TABLES

Table 1. The range of Hsp90 clients and corresponding disease association. Table 2. Chaperones and co-chaperones identified as potential nuclear Hsp90α interactors. Table 4. Transcription factors/ co-factors identified as potential nuclear Hsp90α interactors. Table 3. Ubiquitin-proteasome system components identified as potential nuclear Hsp90α interactors. Table 5. Previously known Hsp90 clients identified in the insoluble protein fraction upon 8h- treatment. Table 6. Nuclear proteins identified upon 8h-treatment. Table 7. Overlap between Hsp90 physical and functional interactors. Table 8. Hsp90 and HCF-1 common interactors.

ABBREVIATIONS

Hsp90 - heat shock protein 90 Pi - inorganic phosphate Hsp - heat shock proteins SHRs - steroid hormone receptors sHsp - small heat shock proteins RNA - ribonucleic acid HSR - heat shock response DNA - deoxyribonucleic acid TRiC - TCP-1 ring complex HSF1 - heat shock factor 1 NTD - amino-terminal domain PTMs - post-translational modifications MD - middle domain Ser - serine CTD - carboxy-terminal domain Thr – threonine TPR - tetratricopeptide repeat Tyr - tyrosine CS - Chord and Sgt1 DNA-PK - DNA-activated protein kinase ADP - adenosine diphosphate NLS - nuclear localization signal ATP - adenosine triphosphate NES - nuclear export signal ATPases - adenosine triphosphatases HtpG - high-temperature protein G Hop - Hsc70/ Hsp90-organizing protein piRNA - Piwi-interacting RNA Cdc37 - cell division cycle 37 homolog snoRNA - small nucleolar RNAs PP5 -serine/threonine-protein Phosphatase 5 pre-rRNA - preribosomal RNA PPIase - peptidylprolyl isomerase miRNA – micro RNA CYP40 - cyclophilin 40 siRNA - small (or short) interfering RNA Aha1 - activator of Hsp90 ATPase protein 1 RISC - RNA-induced silencing complex p23 - prostaglandin E synthase 3 PARP1 - poly(ADP-ribose)polymerase 1 R2TP - Ruvbl1-Ruvbl2-Rpap3-Pih1d1 XRCC1 - X-ray repair cross- Pih1d1 - Pih1 domain-containing protein 1 complementing protein 1 Rpap3 - RNA Polymerase II associated LUMIER - luminescence-based protein 3 mammalian interactome mapping PIKK - phosphatidylinositol-3 kinase- di-Gly - di-glycine group related protein kinase polymerase II TF - transcription factor AARSD1 - alanyl-tRNA synthetase domain- PcG - Polycomb containing protein 1 TRx - trithorax ER - endoplasmic reticulum H3K4me3 H3K27me3 Grp94 - 94 kDa glucose-regulated protein SMYD - SET and MYND domain- TRAP1 - tumor necrosis factor type 1 containing methyltransferase receptor-associated protein H3 - histone 3

ABBREVIATIONS

NF-κB - nuclear factor kappa-light-chain- PBS - phosphate buffer saline enhancer of activated B cells RFP - red fluorescent protein DSP - dithiobis succinimidyl propionate HA - hemaglutinine HCF-1 - host cell factor 1 TNE - total nuclear extract SILAC - stable isotope labeling with amino NUP - nucleoplasm acids in cell culture CHR - chromatin HEK293 - human embryonic kidney cells SNE - soluble nuclear extract K562 - chronic mylogenous leukemia cells INE - insoluble nuclear extract DMEM - Dulbecco's Modified Eagle's SCE - soluble cytosolic extract Medium ICE - insoluble cytosolic extract IMDM - Iscove's Modified Dulbecco's PML - promyelocytic leukaemia Medium YY1 - transcriptional repressor protein Y BZ - bortezomib PQC - protein quality control TCE - total cell extract Hsp90i – Hsp90 inhibition

xii

ABSTRACT

Molecular chaperones are defined as factors that help other proteins to acquire their functionally active conformation. Most chaperones bind nascent polypeptides that emerge directly from the ribosome. Hsp90, however, is a specialized chaperone. It does not take part in the early stage of protein folding but stabilizes partially mature metastable structures. It binds a restricted set of proteins collectively called clients - transcription factors, kinases, ubiquitin ligases, metabolic enzymes and others. Notably, many Hsp90 clients contribute to oncogenic transformation and maintenance of malignant phenotypes. Thus, the chaperone has turned into an attractive target in cancer therapy.

More than three decades of study revealed the central importance of Hsp90 in numerous cellular processes and carcinogenesis. Nonetheless, despite its crucial roles in biology and medicine, the complexity of the system left many aspects of its function, regulation and structure uncovered. In addition, Hsp90 is traditionally viewed as a cytoplasmic chaperone but cumulative evidences suggest nuclear localization and important roles in nuclear events. Understanding mechanisms by which Hsp90 regulates transcription may offer additional opportunities for cancer therapy involving Hsp90 inhibition.

The present work aimed at shedding more light on the nuclear pool of Hsp90. In the first part of my thesis I employed physical and genetic interactome screens and found that Hsp90 forms a complex network of interactors. Its binding partners fall into the following main categories: (i) chaperones and co-chaperones, (ii) components of the protein degradation machinery, (iii) DNA repair proteins, (iv) RNA metabolism factors and (v) transcription regulators. In addition, I characterized nuclear Hsp90 post-translational modifications (PTMs) by confirming previously known and identifying novel PTMs.

In the second part of my thesis I utilized biochemical and computational approaches and presented evidence that nuclear Hsp90 stabilizes the abundant transcriptional co-factor HCF- 1 and thus supports its transcriptional regulation activity. These findings shed more light on HCF-1 mode of action and imply a role of nuclear Hsp90 in the modulation of chromatin dynamics together with HCF-1.

In summary, our results provide a resource for more detailed mechanistic and functional studies of nuclear Hsp90 and illustrate an example of chromatin-associated action and transcription regulation role of Hsp90 via its interaction with HCF-1. Notably, the

xiii

ABSTRACT

interactome data obtained in this work can be exploited as a tool to investigate additional opportunities for cancer therapy involving Hsp90 inhibition.

Part of the present work was included in a manuscript submitted for publication in February, 2019 to the Molecular Cell journal (Antonova et al., 2019; see Appendix).

xiv

ZUSAMMENFASSUNG

Molekulare Chaperone sind definiert als Faktoren, die anderen Proteinen helfen, ihre funktionell aktive Konformation zu erhalten. Die meisten von ihnen binden naszierende Polypeptide, die direkt aus dem Ribosom hervorgehen. Hsp90 stellt jedoch ein besonders spezialisiertes Chaperon dar. Es nimmt nicht am frühen Stadium der Proteinfaltung teil, stabilisiert aber teilweise reife, metastabile Strukturen. Darüber hinaus bindet es eine begrenzte Menge von Proteinen, die kollektiv als „Clients“ bezeichnet werden – dazu zählen unter anderem Transkriptionsfaktoren, Kinasen, Ubiquitinligasen und Stoffwechselenzyme. Bemerkenswerterweise tragen viele Hsp90-Clients zur onkogenen Transformation und Aufrechterhaltung von malignen Phänotypen bei. So hat sich das Chaperon zu einem attraktiven Ziel in der Krebstherapie entwickelt.

Über drei Jahrzehnte Forschung zeigen die zentrale Bedeutung von Hsp90 in zahlreichen zellulären Prozessen und in der Karzinogenese. Trotz seiner wichtigen Rolle in der Biologie und Medizin sind die genaue Funktion, Regulierung und Struktur von Hsp90 jedoch noch nicht aufgeklärt. Darüber hinaus wird Hsp90 traditionell als zytoplasmatisches Chaperon angesehen, es gibt jedoch immer mehr Beweise, welche auf eine nukleare Lokalisierung und die damit verbundenen wichtigen Funktionen im Nukleus hindeuten.

Die vorliegende Arbeit zielte darauf ab, die bisher vernachlässigte Rolle von Hsp90 im Nukleus zu charakterisieren. Im ersten Teil meiner Arbeit beschreibe ich, dass Hsp90 ein komplexes Netzwerk mit verschiedenen Interaktoren bildet, die in die folgenden Hauptkategorien fallen: (i) Chaperone und Co-Chaperone, (ii) Komponenten der Proteinabbau-Maschinerie, (iii) DNA-Reparatur-Proteine, (iv) RNA-Stoffwechselfaktoren und (v) Transkriptionsregulatoren. Darüber hinaus wurden von mir in dieser Arbeit sowohl bekannte Hsp90-posttranslationale Modifikationen bestätigt als auch neue charakterisiert.

Im zweiten Teil meiner Arbeit konnte ich zeigen , dass nukleäres Hsp90 den Transkriptions-Co-Faktor HCF-1 in seiner Transkriptionsregulationsaktivität unterstützt. Diese Ergebnisse geben mehr Aufschluss über die Wirkungsweise von HCF-1 und implizieren eine zusätzliche Rolle von nukleärem Hsp90 bei der Modulation des Chromatins.

Zusammenfassend liefern die von uns erbrachten Daten einen guten Ausgangspunkt für weiterführende mechanistische und funktionelle Untersuchungen von nukleärem Hsp90. Meine Arbeit zeigt ein Beispiel für die Rolle von Hsp90 während der Transkriptionsregulation und die Chromatin-assoziierte Wirkung von Hsp90 durch seine

ZUSAMMENFASSUNG

Wechselwirkung mit HCF-1. Zusätzlich können die in dieser Arbeit generierten Interaktom- Daten genutzt werden, um potentielle Anti-Krebstherapien, durch die Inhibierung von Hsp90, zu entwickeln.

xvi

CHAPTER 1 INTRODUCTION

CHAPTER I

Systematic analysis of nuclear heat shock protein 90 (Hsp90)

1. Introduction

1.1. Molecular chaperones

Proteins are large, complex biomolecules that perform a vast array of functions within organisms [1]. They are required for the proper structure, function, and regulation of not only individual cells but body’s tissues and organs. Certain proteins act as enzymes and control diverse biochemical reaction (glucose oxidase, kinases, nucleic acid polymerases, etc.). Others are structural components providing desired organization and strength to cells and tissues (actin, tubulin, collagen, keratin, etc.). Many proteins are involved in the process of cell signalling and signal transduction (receptors, hormones, etc.), intra- and inter-cellular transport (transmembrane proteins, haemoglobin, etc.) as well as protection upon pathogen invasion (antibodies, complement proteins, etc.). Regardless of their role, every protein must acquire its characteristic three-dimensional structure in order to be functional [1]. The folding process begins co-translationally as the nascent polypeptide emerges from the ribosome. Despite its spontaneous nature protein folding requires assistance as in the highly crowded cellular environment protein aggregation competes with productive folding even at physiological temperatures [2]. That determines a need for support provided by a group of proteins called molecular chaperones [3].

Molecular chaperones, commonly referred to as heat shock proteins (Hsp) are defined as factors that interact with, stabilize or help another protein to acquire its functionally active conformation, without being present in its final structure [4, 5]. They are classified according to molecular weight into five major and broadly conserved families - Hsp100s, Hsp90s, Hsp70s, Hsp60s, and small Hsp (sHsp) (FIGURE 1), and often cooperate with each other. For example, a polypeptide that is bound by Hsp70 at an early stage of folding can be transferred to Hsp90 or to the TCP-1 ring complex (TRiC) to reach its final native state (FIGURE 2) [6, 7]. Some chaperones use cycles of ATP binding and hydrolysis to fold non-native polypeptides. They are referred to as “foldases” - Hsp70, Hsp90, TRiC, etc. Others simply have a “handover” role. They bind folding intermediates to prevent aggregation and are referred to as “holdases” - Hsp40, Hsp33, etc. [8].

CHAPTER 1 INTRODUCTION

Figure 1. Characteristics of different chaperone machineries. Molecular chaperones are classified according to molecular weight into five distinct chaperone families. Different co-factors (termed co- chaperones) provide them assistance in protein folding (image from [9]).

The critical importance of molecular chaperones goes beyond nascent protein folding. Living cells are constantly challenged by changing environmental conditions and provoked to maintain protein homeostasis (proteostasis) under diverse circumstances. Certain stimuli, such as sudden increase in temperature (termed ‘heat shock’), toxins, oxidative stress, heavy metals, and pathogenic infections may induce massive protein misfolding and deleterious aggregation [10]. That damage on cellular proteins strongly hampers their biological functions and leads to essential cellular structures/processes breakdown. In addition, the accumulation of unfolded/aggregated protein species is very toxic [11]. What a cell experiences in the described scenario is defined as proteotoxic stress [12, 13]. Fortunately, not every instance of proteotoxic stress proves fatal. Cells quickly respond to the insult by activating a transcriptional program known as the heat shock response (HSR) [14, 15]. It boosts surviving and recovery by elevating the expression levels of several key protein

CHAPTER 1 INTRODUCTION

groups – molecular chaperones, components of the protein degradation machinery, metabolic enzymes, regulatory proteins (such as transcription factors and protein kinases), transport and detoxification factors, cell organization proteins (such as cytoskeleton components) and DNA/ RNA repair factors [16, 17]. Thus, chaperones strongly contribute to cell survival upon proteotoxic stress by restoration of the normal protein homeostasis - they carry out refolding of misfolded proteins, solubilization of aggregates and target terminally damaged proteins to degradation. Importantly, since abnormal proteostasis is often associated with different disease states in human such as Alzheimer’s, Huntington’s, Parkinson’s and cancer, chaperones are found misregulated in a number of disorders [18-20].

Figure 2. Models for chaperone-assisted folding of newly synthesized polypeptides in eukaryotes. Polypeptide chains reach their native states (N) in a reaction assisted by the Hsp70 and Hsp40 chaperones. A fraction of these must be either transferred for folding to Hsp90 or passed on to the chaperonin TRiC with the assistance of prefoldin (PFD). Both TRiC and PFD belong to the Hsp60 family of chaperones (image from [5]).

In human, the complex chaperone network (or the “chaperome”) is encoded by 332 genes [20]. Several exhaustive reviews provide detailed information on individual molecular chaperones structure, function and significance [3, 16, 21, 22]. In my thesis I will focus on the Hsp90 chaperone machinery.

1.2. Hsp90

Similar to other molecular chaperones, Hsp90 was discovered because of the specific, elevated expression during the HSR [23-25]. The chaperone comprises around 1 %–2 % of total protein in unstressed cells, and is one of the most abundant cellular proteins [26, 27]. It

CHAPTER 1 INTRODUCTION

is ubiquitously expressed from bacteria to human, being absolutely essential for viability in eukaryotes [28, 29] but dispensable under non-heat stress conditions in bacteria [30, 31].

1.2.1. Structure and conformational cycle

Hsp90 functions as a homodimer [32]. Each monomer consists of three highly conserved domains (FIGURE 3A): • amino-terminal domain (NTD) - facilitates ATP (adenosine triphosphate) binding [33]. • middle domain (MD) - important for ATP hydrolysis and binding of clients. It is connected to the NTD by a long, flexible, charged linker that modulates the corresponding domains contact and biological function [34-36]. • carboxy-terminal domain (CTD) - responsible for dimerization [37]. Its MEEVD motif is important for the crucial interaction with tetratricopeptide repeat (TPR) domains- containing co-factors (co-chaperones) [38].

Hsp90 belongs to the GHKL (Gyrase, Hsp90, Histidine Kinase, MutL) superfamily of ATPases (adenosine triphosphatases) [39, 40] and its ATPase activity proves to be essential in vivo [41, 42]. It exhibits low ATP affinity and low enzymatic activity in absence of activating co-factors [33, 43, 44]. Its ATP-binding pocket resides in the NTD. In the absence of ATP, the homodimer adopts a V-shaped open conformation [45, 46]. Upon ATP binding all three domains undergo large and dynamic rearrangements in order to reach a closed state in which the formation of the split ATPase domain is completed (FIGURE 3B) [47]. The loop containing several conserved amino acids, termed the ‘lid’ region, closes over the ATP-bound state. This is how an intermediate state is reached. Further structural rearrangements favor NTDs dimerization (closed 1 state) and subsequent association with the middle domains (closed 2 state). This chain of events leads to a closed state and represents the rate-limiting step in Hsp90 conformational cycle [48]. It allows ATP hydrolysis to occur [49], after which the NTDs dissociate, ADP (adenosine diphosphate) and Pi (inorganic phosphate) are released and Hsp90 returns to its open conformation (FIGURE 3B). It was recently shown that the amount of time that Hsp90 spends in different steps of the cycle, and not the overall speed, is important for its in vivo function [50, 51].

1.2.2. Function

Hsp90 chaperone binds to and stabilizes a large number of proteins collectively called clients (substrates) – a comprehensive literature-based list has been maintained by the lab of Didier Picard (www.picard.ch/Hsp90Int/index.php). Hsp90 clients are defined as proteins

CHAPTER 1 INTRODUCTION

that directly interact with the chaperone and at the same time are destabilized and depleted upon Hsp90 inactivation. This postulate has been confirmed for nearly every Hsp90 substrate studied in detail and turned into a dogmatic hallmark of Hsp90 dependency [52-55]. Several hundred proteins with extremely diverse biological significance are described as clients of Hsp90. Historically, protein kinases and steroid hormone receptors (SHRs) were the first and most extensively studied Hsp90 substrates [56-58]. Recent quantitative analysis of Hsp90 clients revealed that approximately 60 % of the human kinome and 7 % of human transcription factors associate with the chaperone [59]. The same study identified about 30 % of human E3 ubiquitin ligases as Hsp90 interactors. In addition, the long and diverse list of Hsp90 clients includes cytoskeletal proteins, RNA processing factors, telomerase, kinetochore components and many others (www.picard.ch/Hsp90Int/index.php). This makes the chaperone a central modulator of important processes such as cell signal transduction, DNA repair, transcription regulation, development, immune response, stress regulation and others [60].

Figure 3. Structure, conformational cycle and function of Hsp90. Schematic representation of domain organization of the Hsp90 homodimer (A), its conformational cycle (B) and mode of action (C) (image from [61]).

CHAPTER 1 INTRODUCTION

Despite the apparent heterogeneity, the clientele of Hsp90 is limited compared to the general chaperone Hsp70, for example. Unlike Hsp70, which appear to interact with all unfolded proteins, Hsp90 interacts with a restricted set of substrates [6]. It seems to be specialised in stabilising late folding intermediates - partially folded and conformationally labile structures [51, 62]. It maintains and prevents from degradation/aggregation metastable conformations that favour protein activation (by ligand binding for example as in the case of steroid hormone receptors [63]), translocation across membranes (nuclear export of 60S ribosomal subunits [64]), incorporation into a protein complex (RNA polymerase II complex assembly[65]), and interaction with other factors (XRCC1 and DNA polymerase β interaction in DNA damage repair [66]) (FIGURE 3C).

In contrast to Hsp70, for which the mechanism of client protein recognition is generally understood (binding to hydrophobic sequences regardless of secondary and tertiary structure) [6], determinants of client recognition by Hsp90 have remained elusive. The chaperone binds to and stabilizes a large number of proteins with unrelated amino acid sequences and functions, including protein kinases and steroid hormone receptors [60]. Although structurally unrelated, the kinase domain and the ligand-binding domain of steroid hormone receptors must have something in common that Hsp90 recognizes. Surprisingly, even within a particular client protein family some members strongly interact with the chaperone, others display a weaker affinity and third - do not associate at all [67]. Therefore, despite the overall conservation of the protein fold, there must be specific sequence or structural features that distinguish client proteins from non-clients. Finally, Hsp90 associates with a large number of co-factors (co-chaperones), which regulate its function and seem to provide specificity to client protein interactions. An intriguing work by Taipale et al., 2012 provided evidence to establish Hsp90 client recognition as a combinatorial process - a co-chaperone provides recognition of particular types of fold, whereas thermal and conformational stability determines the extent of interaction with the chaperone [68].

Hsp90 is a specialized but absolutely essential protein folding tool in eukaryotes. It is a key regulator of proteostasis under both physiological and stress conditions in eukaryotic cells. Its function becomes even more critical in times of cellular proteotoxic stress - then it contributes more generally to coping with stress-induced structural abnormalities in proteoins [69].

An intriguing aspect of Hsp90 function links the chaperone to morphological evolution. Susan Lindquist was the first to raise the question of whether Hsp90 protein stabilization

CHAPTER 1 INTRODUCTION

activity could affect evolution by buffering the potentially detrimental effects of spontaneous mutations on protein stability [70]. It was demonstrated that inhibiting or deleting Hsp90 in various species revealed phenotypes that under normal conditions are suppressed (“buffered”) by the action of Hsp90. Thus, Hsp90 has been proposed to function as an evolutionary capacitor allowing natural populations to accumulate genetic variation, which could be potentially released as phenotypic variation under stressful conditions that compromise Hsp90 function. An intriguing recent work by our lab demonstrates that in a similar fashion Hsp90 buffers the regulatory effects of mammalian endogenous retroviruses [71].

1.2.3. Regulation of Hsp90

Hsp90 function is regulated at several levels – transcription, post-translational modifications (PTMs) and interaction with co-factors called co-chaperones [72].

1.2.3.1. Transcriptional regulation of Hsp90

In most eukaryotes, Hsp90 expression approximately doubles in response to environmental stress due to induction by the stress-related transcription factor heat shock factor 1 (HSF1). Intriguingly, HSF1 proves to be an Hsp90 client in normal conditions. The current model suggests that monomeric HSF1 is kept in an inactive state by its binding to Hsp70 and Hsp90 in the cellular cytosol [73]. Upon proteotoxic stress chaperones dissociate in order to deal with the increasing number of misfolded proteins around. Thus, HSF1 is released, forms trimers and migrates towards the nucleus to induce the expression of heat shock genes. Once the cell returns to normal state, the availability of free chaperones leads again to the downregulation of the transcriptional factor. This is how Hsp90 links the stress status of the cell directly to the expression of heat shock proteins and has an important role in regulating its own transcription. This simple commonly accepted model has been recently revised and supplemented with additional layers of complexity [74]. Cumulative evidence suggest that HSF1 activity and thus Hsp90 expression level are additionally affected by factors such as PTMs (phosphorylation, sumoylation, ubiquitylation), protein trafficking, degradation and interaction with co-factors [74].

1.2.3.2. Post-translational modifications of Hsp90

Various PTMs such as phosphorylation, acetylation, methylation, ubiquitination, SUMOylation, oxidation and S-nitrosylation fine-tune Hsp90 function [75, 76]. Phosphorylation has been the most widely studied one with a number of serine (Ser),

CHAPTER 1 INTRODUCTION

threonine (Thr), and tyrosine (Tyr) phosphorylation sites identified - a comprehensive list can be found in a study by Mollapour et al., 2011 [77].

Steady-state phosphorylation level of Hsp90 could be altered by a few kinases some of which appear to be its clients (FIGURE 4). Phosphorylation affects its activity, interaction with co-chaperones/nucleotides/clients or localization and modulate its function upon specific stimuli [78]. A number of studies have shown that chaperone phosphorylation positively regulates its interaction with distinct client proteins both in vitro and in vivo [79-82]. Functional proteomics studies have demonstrated that Hsp90 could be phosphorylated by B- Raf and Akt, both clients of the chaperone. Barati et al., 2006 reported that Akt phosphorylates several stress-induced chaperones including Hsp90α and Hsp90β [83]. Another study showed Hsp90α Ser-263 as one of the targets of B-Raf in melanoma cells [84].

Hsp90 phosphorylation has been shown to modulate its function in response to environmental stimuli. For instance, c-Src kinase phosphorylates Hsp90β Tyr-301 in response to vascular endothelial growth factor receptor-2 (VEGFR-2) activation [85]. This leads to increased association between Hsp90β and eNOS and higher eNOS activity. Another example shows that DNA-activated protein kinase (DNA-PK) isolated from HeLa cells phosphorylates two threonine residues (Thr-5 and Thr-7) of Hsp90α in response to cellular DNA damage [86-90]. In addition, heat shock has been found to cause decrease of Hsp90 phosphorylation in HeLa and yeast cell suggesting that PTMs dynamics could be critical for the HSR [91, 92].

An intriguing aspect of Hsp90 PTMs importance is their contribution to chaperone localization. Hsp90α phosphorylation status for example, has been shown to affect its translocation to the cell surface in diabetes mellitus, hyperglycaemia and cancer [81, 93]. This effect is mediated via Thr-90 modification by PKA. Several exciting studies reveal further that phosphorylation-dependent regulation of Hsp90 secretion in keratinocytes and cancer cells may play an important role in wound healing and tumor metastasis, respectively [94, 95]. Hsp90 PTMs dynamics could also be the factor to facilitate its shuttling between nucleus and cytosol since no nuclear localization signal (NLS) or nuclear export signal (NES) have been identified so far. Mollapour et al., 2010 for example demonstrated that Swe1- mediated phosphorylation of nuclear Hsp90 in yeast cells is a “switch-off” mechanism during S phase of the cell cycle causing chaperone to translocate to cytoplasm [96]. There it gets poly-ubiquitylated and degraded by proteasomes. In addition, a study in human cells

CHAPTER 1 INTRODUCTION

demonstrates Hsp90α Thr-7 phosphorylation immediately after DNA damage and subsequent accumulation in repair foci [89]. Quanz et al., 2012 speculate that Thr-7 modification occurs in cytosol and could be the possible reason behind Hsp90 translocation to the nucleus [89].

Figure 4. Hsp90 PTMs. Domain localization of phosphorylated (black), S-nitrosylated (green), oxidated (brown) and acetylated (pale blue) residues of Hsp90α and Hsp90β (N – N-terminal domain, M – middle domain, C – C-terminal domain). Kinases known to phosphorylate Hsp90 are shown in red. S indicates serine, T – threonine, Y - tyrosine and K - lysine residues. (image from [78]).

Interestingly, Hsp90 phosphorylation emerged very critical in relation to pharmacological inhibition of the chaperone. A study showed that CK2 phosphorylates Ser- 231 and Ser-263 in the charged-linker of Hsp90α [97]. Equivalent residues in Hsp90β (Ser- 226 and Ser-255) are also phosphorylated in untransformed but not in leukemic cells [98]. Several studies [77, 81, 96, 99-101] demonstrated that altered phosphorylation status of Hsp90 could affect drug binding and confer cancer cells sensitivity to Hsp90 inhibition (Hsp90i). This raised the intriguing possibility of modulating certain Hsp90 PTMs states to

CHAPTER 1 INTRODUCTION

sensitize cancer cells to Hsp90 inhibitors. Indeed, recent evidence suggests that inhibition of enzymes responsible for certain Hsp90 PTMs can act synergistically with Hsp90 inhibitors and provide a novel therapeutic strategy against cancer [102].

Hsp90 acetylation was reported for the first time in 2002 by Yu et al. [103]. Later on, the chaperone was shown to possess a number of additional acetylated lysine residues some of which deposited by p300 [104, 105]. Several studies demonstrated that increased steady state acetylation of the chaperone destabilizes its interaction with clients (androgen receptor, Raf- 1, and mutant p53, etc.) and co-chaperones and decreases binding to ATP [103, 105, 106]. Overall Hsp90 acetylation dramatically affects its function, but the identity and importance of individual acetylated residues have not been extensively studied.

Increasing evidence suggests that Hsp90 could also be methylated. Abu-Farha et al., 2011 reported that SMYD2 could methylate Ser-209 and Ser-615 on Hsp90α [107]. A recent study by Hamamoto et al. described Ser-531 and Ser-574 of Hsp90β to be methylated and important for dimerization and chaperone complex formation [108]. In addition, Donlin et al., 2012 demonstrated the importance of Hsp90 methylation by SMYD2 for muscle functions [109]. However, methylation still remains poorly investigated.

In conclusion, the diverse biological processes Hsp90 influences justify the necessity of a wide range of mechanisms to regulate its function [75]. Consequently, a vast array of PTMs modulate Hsp90 activity/interactions/localization and intriguingly – binding to inhibitory drugs. Hsp90 PTMs investigation should progress further in order to characterize responsible modifiers, regulatory mechanisms and possible effect on drug binding.

1.2.3.3. Hsp90 co-chaperones

Co-chaperones are defined as non-client proteins that physically interact with a chaperone in a reversible manner to assist its function. To date, more than 20 co-chaperones have been identified. They regulate the function of Hsp90 in different ways such as modulation of its ATPase activity and conformational cycle as well as recruitment of specific clients [110]. Individual co-chaperones bind to specific conformational states and must associate with and dissociate from Hsp90 in a sequential manner [111]. Binding sites for co-chaperones reside in all three domains of the chaperone. Some compete with each other, whereas others could bind simultaneously or even synergistically. This demonstrates that the functions of different co-chaperones present in the same complex are integrated in the pro-

CHAPTER 1 INTRODUCTION cessing of a client protein [112]. Several co-chaperones possess one of the very characteristic TPR or CS (Chord and Sgt1) domains that facilitate Hsp90 binding [113, 114].

Interestingly, Hsp90 co-chaperones first appear in eukaryotes - yeast already contain a co-chaperone complement that is similar to that of human cells [111]. In bacteria the directionality of its reaction cycle is ensured by a deterministic, ratchet-like mechanism strongly influenced by nucleotide binding and turnover [115]. Thus, during evolution Hsp90 machinery has developed to a more flexible and less nucleotide-controlled system. Surprisingly, recent evidence suggest that evolution could also favour a chaperone substitution by a PTM [116]. Zuehlke et al., 2017 provided evidence that support the concept of evolutionary replacement of HCH1 co-chaperone in yeast by reversible phosphorylation of Hsp90-Y627 in higher eukaryotes.

Some of the well-studied Hsp90 co-chaperones will be discussed below:

• Hop (Hsc70/ Hsp90-organizing Protein) It is well-accepted that some clients are first bound to Hsp70 and then transferred to Hsp90. Hop is a co-chaperone that functions as an adaptor for the two chaperones [117-119]. It binds Hsp90 via its TPR domain. Furthermore, Hop inhibits the ATPase activity of the chaperone by acting as a non-competitive inhibitor. That maintains Hsp90 in an open conformation and allows client transfer. Hop deletion in mice is embryonically lethal, which highlights its importance in higher eukaryotes [120].

• Cdc37 (Cell Division Cycle 37 Homolog) Cdc37 is a specialized co-chaperone recruiting diverse kinases to Hsp90 for further maturation. It binds to the NTD of the chaperone, which leads to partial inhibition of Hsp90 ATPase activity by abrogation of both lid closure and NTD dimerization. It also wraps around the MD, forming extensive contacts with both clients and Hsp90 [121]. In addition, Cdc37 has some properties of a chaperone and may itself contribute to the folding reaction [122].

• PP5 (Serine/threonine-protein Phosphatase 5) PP5 is another TPR-containing co-chaperone. It modulates Hsp90 conformational cycle by dephosphorylating the chaperone as well as Cdc37 [123, 124]. The phosphorylation status of Hsp90 has been reported to influence its conformational dynamics as well as maturation of clients [125].

CHAPTER 1 INTRODUCTION

• FKBPs (FK506-binding Proteins) Several member of the so called FK506-binding proteins (FKBPs) family of immunophilins [126] that possess peptidylprolyl isomerase (PPIase) activity have been described as Hsp90 co-chaperones. These include FKBP51 (also known as FKBP5), FKBP52 (also known as FKBP4), FKBP36, FKBP38 and cyclophilin 40 (CYP40)) in vertebrates. In addition, they themselves possess chaperone activity and thus may also be involved in the selection and recruitment of clients [127].

• Aha1 (Activator of Hsp90 ATPase protein 1) Unlike Hop and Cdc37, Aha1 co-chaperone is a strong activator of Hsp90 ATPase activity [128]. Similar to Cdc37, one Aha1 molecule interacts with the Hsp90 dimer in an asymmetric manner binding to both the MD as well as the NTD [129]. Aha1 drives the Hsp90 conformational cycle towards closed 1 state by displacing Hop [130] and interfering with client binding [131].

• p23 (Prostaglandin E synthase 3) p23 is a co-chaperone that acts at a late stage of the chaperone cycle. It stabilizes the closed 2 state by binding in a groove that is formed between the dimerized NTDs of Hsp90 [132]. That interaction occurs through its CS domain. The co-chaperone reduces the ATPase activity of Hsp90 and thus drives the progression of the reaction cycle towards maturation of clients [133]. In addition, p23 exhibits chaperone activity itself [127, 134], which links the protein to a broad range of processes including chromatin remodeling and ribosome biogenesis [135-137].

• R2TP complex (Ruvbl1-Ruvbl2-Rpap3-Pih1d1) R2TP is the most complicated of the various Hsp90 co-chaperones identified so far, being a multiprotein complex itself. It was initially discovered in yeast in 2005 and described as a novel Hsp90 co-chaperone [138]. Later on, proteomic analysis of human Hsp90 binding partners identified the human orthologues of complex components as Hsp90 interactors [139].

R2TP complex in human consists of four proteins: Pih1d1 (Pih1 Domain-Containing Protein 1), Rpap3 (RNA Polymerase II Associated Protein 3) and two ATPases - Ruvbl1 and Ruvbl2. The first two are the ones providing physical association with Hsp90 through a CS and TPR domains, respectively. It is a specialized co-chaperone, which supports Hsp90 in the assembly of diverse complexes [140] and thus links it to cellular processes such as box C/D

CHAPTER 1 INTRODUCTION snoRNP biogenesis, apoptosis, phosphatidylinositol-3 kinase-related protein kinase (PIKK) signalling, and RNA polymerase II assembly [141].

Notably, R2TP was shown to inhibit Hsp90 ATPase activity [142] – a shared property of co-chaperones, which are thought to facilitate client loading onto Hsp90 (such as Cdc37 and Hop) [143, 144]. Furthermore, a recent study added an important structural basis for understanding how R2TP couples an Hsp90 dimer to a diverse set of client proteins [145]. These observations suggest that the Pih1-Tah1 co-chaperone could act as a platform for the recruitment and loading of client proteins onto the Hsp90 chaperone.

• CHIP (C terminus of Hsp70-interacting protein) CHIP is an E3 ubiquitin ligase that acts as a co-chaperone to facilitate the degradation of clients in a chaperone-dependent manner. It binds both Hsp90 and Hsp70. Its interplay with Hsp90 determines the fate of numerous clients that are eventually either stabilized or degraded [146].

In summary, co-chaperones are the most important regulators of Hsp90 and together they build up the complex Hsp90 chaperone machinery [61]. That powerful system allows the stabilization of numerous proteins which go through the tightly-regulated chaperone cycle.

1.2.4. Chaperone cycle

For 30 years now, steroid hormone receptors (SHRs) have been the best characterized examples of Hsp90 clients and our current mechanistic understanding of the chaperone cycle is largely based on detailed studies om them [147]. Many monomeric SHRs bind the chaperone at their near native but still inactive state [147]. According to in vitro experiments, SHRs must pass through several complexes with different co-chaperone compositions chronologically to reach their active conformation (FIGURE 5).

Initially, SHRs bind Hsp70/Hsp40 in the so called “early complex”. That interaction facilitates hormone receptors recruitment to Hsp90. Later on, the adaptor protein Hop provides physical link between the early complex and the chaperone to form the “intermediate complex”. Hop co-chaperone stabilizes Hsp90 dimer open conformation and inhibits its ATPase activity. Subsequent binding of a PPIase (peptidylprolyl isomerase) leads to the formation of an asymmetric Hsp90 intermediate complex and SHRs appear to bind to Hsp90 at this stage. In the last step of the chaperone cycle binding of ATP and p23 favors the acquisition of Hsp90 “closed” conformation which weakens the binding of Hop and therefore promotes its dissociation. Potentially another PPIase associates to form the “late complex”

CHAPTER 1 INTRODUCTION

together with Hsp90 and p23. That allows ATP hydrolysis and client stabilization to occur. Eventually p23, PPIase, and the fully folded and stabilized SHR are released. Now, the SHRs have attained their hormone-binding conformation. Once released from Hsp90, they could bind the appropriate steroid hormone, resulting in dimerization and activation. The Hsp90 dimer, on the other hand, adopts an open conformation ready to enter another cycle [147] (FIGURE 5).

Figure 5. Hsp90 chaperone cycle illustrated by the stabilization of steroid receptors. Hop co- chaperone facilitates the transfer of native but still inactive state glucocorticoid receptor (GR) from the Hsp70/Hsp40 chaperone system to Hsp90. ATP, p23 and PPIase binding induce the formation of a closed complex. Subsequent ATP hydrolysis and p23 dissociation allow GR stabilization and release (image from [61]).

Hsp90 chaperone cycle slightly varies according to the substrates to be stabilized or resident cell type. Thus, displacement of Hop and Hsp70 could also be achieved by Aha1 co- chaperone in particular cases but not by the combined action of a PPIase and p23. In addition, different client recruiters could participate apart from Hsp70 – Cdc37 and R2TP complex, for example. Intriguingly, the AARSD1 (alanyl-tRNA synthetase domain-containing protein 1) co-chaperone displays an example of tissue specificity displacing the more ubiquitous p23 during muscle differentiation [148].

CHAPTER 1 INTRODUCTION

1.2.5. Hsp90 isoforms

Mammalian cytosolic Hsp90 is encoded by two genes - Hsp90α and Hsp90β [149]. The primary amino acid sequences of the two isoforms are 86 % identical and they are mutually expressed in all tissues in mice with few exceptions. In heart and muscle tissues Hsp90α has reduced levels compared to Hsp90β [148, 150]. Stage and lineage-regulated expression during mouse germ cell differentiation and embryogenesis has been demonstrated as well at both mRNA and protein level. Hsp90α is expressed at high levels in the germinal compartment of the testis, particularly in germ cells in meiotic prophase, whereas Hsp90β is expressed predominantly in the somatic compartment of the testis [151, 152].

Hsp90α is the more stress-inducible isoform due to the heat-inducible cis-acting elements in its promoter, whereas Hsp90β functions more as a “housekeeping chaperone” [149, 153, 154]. Mostly in vitro studies show that each isoform displays the same preferences for co-chaperones and clients, with the exception that some client substrates, such as c-Src and A-raf, preferentially bind Hsp90α following heat shock [155]. However, some cellular processes appear to specifically require Hsp90α, such as caspase-2 activation [156], maturation of metalloprotease 2 in the extracellular matrix and invasiveness of some metastatic cancers [157]. Taken as a whole, these in vitro parameters predict relatively similar phenotypes for mice deficient in Hsp90α or Hsp90β. Surprisingly, this is not the case. Hsp90β knockout mice displays early embryonic lethality due to an inability of the embryo to develop a placenta [158]. Notably, Hsp90α could not compensate for this defect. Contrary to what has been expected, an exciting work by Grad et al., 2010 revealed that Hsp90α deficiency still sustains life and maintains a normal phenotype into adulthood, with the exception of sterility in male mice due to failure of spermatogenesis [150]. It was later on proposed that this phenotype could be explained by either estrogen receptor abnormality [159] or abolishment of piRNA (Piwi-interacting RNA)-mediated retrotransposon silencing [160].

In addition to the cytoplasmic Hsp90, higher eukaryotes also possess distinct isoforms localized in the endoplasmic reticulum (ER) and in mitochondria. In the ER, Grp94 (94 kDa glucose-regulated protein) [161, 162] acts as a molecular chaperone, which is involved in the maturation of proteins such as nascent immunoglobulins, collagen, MHC class II, HSV-1 glycoproteins, and others [163, 164]. This ER-specific Hsp90 isoform is served by ER- specific co-chaperones such as Mzb1 (marginal zone B and B1 cell specific protein) and Cnpy3 (canopy FGF signalling regulator 3) [165, 166].

CHAPTER 1 INTRODUCTION

Trap1 is an Hsp90 homologue localized in mitochondria displaying high homology to bacterial chaperone HtpG (high-temperature protein G) [167, 168]. It is involved in the maintenance of mitochondrial function and polarization [169]. Its contribution to malignant phenotypes maintenance has made it a potential target for cancer treatment [170-172].

1.1.6. Hsp90 and cancer

The importance of Hsp90 in many aspects of human biology, ranging from cancer to infectious diseases to neurodegeneration has been recently reviewed (TABLE 1) [61]. Hsp90 role in cancer will be discussed below.

Several distinctive and complementary capabilities of cancer cells that enable tumor growth and metastasis has been characterized over the years and described as the six hallmarks of cancer [173] - self-sufficient in growth signals, evading apoptosis, sustained angiogenesis, limitless replicative potential, tissue invasion and metastasis, and insensitivity to anti-growth signals. These features could partially be attributed to the presence of abnormal proteins, such as mutated or fusion species. In addition to the rapid cellular proliferation, these destabilizing oncogenic mutations put tumor cells in a constant stressed state and make proteostasis maintenance even more challenging. Thus, as a major factor in protein homeostasis sustenance Hsp90 plays an important part in the survival of cancer cells. Furthermore, a number of Hsp90 clients such as protein kinases, telomerase, hypoxia- inducible factor 1α (HIF1α), UHRF1, p53, c-Myc are strongly involved in tumor growth [174]. Notably, the first client protein of Hsp90 to be identified in vivo in vertebrates was the oncogenic viral tyrosine kinase v-Src [56] . One of the reasons for this preferential interaction is probably due to the destabilising effects of amino acid mutations. In agreement with this hypothesis, it has been shown in vitro that the stability of v-Src is significantly reduced compared to its cellular counterpart c-Src [175]. Dependence of tumor cells on Hsp90 and increased expression levels associated with decreased survival have been well-documented [176, 177]. Thus, it does not come as a surprise that the chaperone has turned into an attractive target in cancer therapy. More than 50 clinical trials have been performed or are in progress, using 18 different Hsp90 inhibitors (https://clinicaltrials.gov) [74, 178, 179]. Most of them target the ATP-binding pocket within the NTD of the chaperone. The mode of ATP binding to Hsp90 is unusual, as the binding pocket in the NTD accommodates the base and the sugar, but the phosphates point outwards and the γ-phosphate only becomes buried once the MD associates with the NTD. That ensures specific inhibition of Hsp90 chaperone activity and minimizes side effects [33].

CHAPTER 1 INTRODUCTION

Client Function Diseases Steroid hormone receptors Glucocorticoid receptor Response to glucocorticoids Cushing syndrome Mineralocorticoid receptor Response to mineralocorticoids Chronic kidney disease Progesterone receptor Response to progesterone Cancer Androgen receptor Response to androgens Spinobulbar muscular atrophy Oestrogen receptor Response to oestrogens Cancer Kinases AKT (also known as PKB) Mitogen signalling Cancer CDK4 Cell cycle control Cancer ERBB2 EGF receptor Cancer HCK Immune response Cancer JAK1 and/or JAK2 Cytokine signalling Cancer SRC Constitutively active Tyr kinase Cancer BRAF Mitogen signalling Cancer BCR–ABL Constitutively active Tyr kinase Cancer E3 ubiquitin ligases MDM2 p53 degradation Cancer UHRF1 DNA methylation Cancer Transcription factors p53 Tumour suppressor protein Cancer OCT4 Embryonic development Cancer HIF1α Angiogenesis Cancer PPARα, PPARβ, PPARγ Fat and glucose metabolism Diabetes mellitus STAT2, STAT3, STAT5 Cytokine signalling Cancer Others Argonaute 1, Argonaute 2 RNA interference Calcineurin Immune response Rheumatoid arthritis Calmodulin Various signalling pathways CFTR Chloride channel Cystic fibrosis HSF1 Regulation of heat shock response NLR proteins Innate immune response Inflammation TERT Telomere maintenance Cancer eNOS Nitric oxide synthesis RAD51 and/or RAD52 DNA repair Cancer Tau Microtubule stabilization Alzheimer disease

Table 1. The range of Hsp90 clients and corresponding disease association (table from [61]).

Intriguingly, cancer cells have been shown to be more sensitive to chemical inhibitors of Hsp90 than normal cells [180]. This could be attributed to the presence of Hsp90 in multi- molecular complexes in tumor cells, which may enhance its affinity to inhibitors [181]. An exciting recent study by Rodina et al., 2016 described the ‘epichaperome’ - a tight physical interaction of the Hsp70 and Hsp90 machineries together with their co-chaperones. That super complex has been identified in a large number of tumors and appears more engaged in client interactions. It promotes survival but at the same time makes tumor cells more sensitive to Hsp90 inhibitors [182].

CHAPTER 1 INTRODUCTION

Hsp90 inhibitors range from natural compounds such as geldanamycin and radicicol to designed synthetic inhibitors, such as PU-H71. Most of them compete with ATP for the ATP- binding pocket in the N-terminal domain of the chaperone [52, 183-187]. However, a new generation of Hsp90 inhibitors, such as novobiocin binds to the C-terminal domain, thus disrupting interactions with the large group of TPR-containing co-chaperones [188]. They manage to overcome one of the major drawbacks of inhibitors that target the N terminus of Hsp90 - the induction of a heat shock response [189]. More-specific inhibitors disrupt Hsp90–Cdc37 binding [187] or target different isoforms [172]. In addition, combined inhibition of Hsp90 together with the proteasome, Hsp70, oncogenic kinases, HDACs for example, have also been investigated [190-194]. The resulting enthusiasm for Hsp90 as a promising target in oncology justifies the need for more detailed information about its interaction network within the cell.

1.2.7. Interactome studies

To better understand Hsp90 role in normal cells and in cellular pathologies, numerous studies have utilized proteomics assays and related high-throughput tools to characterize its physical and functional protein interactions. Despite the huge efforts, Hsp90 interactions mapping still remains particularly challenging due to the following issues: • Hsp90 is notoriously “sticky” - it possesses a highly charged surface capable of binding a wide range of proteins thus introducing false positive hits [195]. • The transient nature of Hsp90-client interactions makes their identification particularly challenging. • Hsp90 clients strength of binding to and dependence on the chaperone has been shown to vary widely [68]. Thus, preserving certain interactions will be highly dependent on the experimental conditions chosen. • Many clients are characterized by low abundance and rapid turn-over. Some display a cell-type or cell-cycle specific expression pattern or are overexpressed only upon defined (predominantly stress) stimuli. • Some interactions with client proteins are dependent upon the nucleotide- bound state of Hsp90 [196]. • It is hard to detect Hsp90 upon a certain client pull-down – the chaperone is simultaneously involved in too many diverse interactions and its levels in that particular complex are usually below the detection limit.

CHAPTER 1 INTRODUCTION

In the last three decades, both small- and large-scale experiments contributed towards unveiling the complex picture of Hsp90 interactome. Diverse approaches have been involved in order to circumvent the issues described above. The following section provides a representative overview of what has been achieved so far.

Standard interactome mapping techniques. Interestingly, Hsp90 was initially discovered in Rous sarcoma virus-transformed chicken cells in 1981 [56]. It co-precipitated during affinity purification of the protein Tyr kinase v-Src - the first molecularly characterized oncogene, and was subsequently characterized as a heat shock protein [25]. In a similar fashion, the discovery of the first Hsp90 clients occurred by serendipity, as the chaperone was frequently co-purified during the isolation of certain proteins. It was also found in complex with other virally encoded Src-like Tyr kinases [197] and several nuclear hormone receptors [57, 58, 198]. Thus, it came as no surprise that early investigations of Hsp90 function focused on two main protein classes: protein kinases [199] and transcription factors [54]. Later on, the discovery of novel Hsp90 substrates occurred rationally rather than coincidentally.

More recently, large-scale studies tremendously contributed to the current knowledge of Hsp90 interaction network. Yeast proved to be the ideal system to study Hsp90 interactors by large-scale genetic and proteomic studies due to the excellent tools available. A decent number of studies in mammalian cells were conducted too. Combining genome-wide two- hybrid screens, large-scale protein affinity purifications and genetic and chemical-genetic interactions Zhao et al., 2005 generated an extensive database from yeast. The work demonstrated that Hsp90 is physically or functionally linked to approximately 10 % of the yeast proteome [138]. The same year Millson et al. expanded that database by performing yeast-two hybrid screens using an Hsp90-mutant, which stabilises Hsp90-client interaction [200]. Parallel work by Falsone et al., 2005 using gel-based proteomics technique provided a co-immunoprecipitation derived snapshot of the Hsp90 interactome from human HEK293 cells [201]. The group described DNA-activated protein kinase catalytic subunit as a client of Hsp90 and demonstrated that its levels are sensitive to Hsp90 inhibition only in HeLa cells but not in HEK293 cells referring to the tumorgenicity of this chaperone.

Despite the methodical heterogeneity and the different origin of the cellular preparations, these pioneer large-scale interactome studies confirmed Hsp90 involvement in transcriptional regulation and signal transduction and suggested possible links to gene translation, metabolism, cell cycle, cytoskeleton maintenance, DNA processing and intracellular transport. Thus, they showed that Hsp90 clients are by far not limited to kinases and

CHAPTER 1 INTRODUCTION

transcription factors and paved the road for future investigation. In the following years, additional large-scale studies re-established and expanded Hsp90 clients repertoire to include RNA processing proteins, modulators of protein turnover via ubiquitination and proteasome degradation [55, 68, 139, 202-206]. An impressive study by Taipale et al., 2012 examined 355 protein kinases (69 % of the human kinome), 498 E3 ligases (69 % of all) and 1,303 transcription factors 79 % of all) for Hsp90 binding in HEK293 cells. As a result, 7 % of transcription factors, 30 % of ubiquitin ligases and 60 % of kinases were found to interact with the chaperone and principles of substrate recognition were investigated [68]. To make the picture even more complex certain Hsp90 interactors have been found to display preferences towards different nucleotide-bound forms of the chaperone [204]. Even more exciting, small molecule Hsp90 inhibitors provided an additional approach to investigate Hp90 interactome and were shown to preferentially pull down oncoprotein Hsp90 clients as compared to non-oncoprotein clients [203, 207]. That finding once again highlighted the strong potential of Hsp90 inhibitors in cancer therapy. Finally, a great number of small-scale targeted studies have shed light on the functional significance of particular Hsp90-client interactions in details. Some of the most significant ones will be further discussed in the course of my thesis.

As the complex interaction network of Hsp90 unveiled it became apparent that one major role of the chaperone is to facilitate assembly/disassembly of protein complexes [208] - by stabilizing an unstable protein and facilitating its incorporation into the complex or by promoting change in complex composition. It does so with the crucial assistance of various co-chaperones. In all cases, Hsp90 and co-chaperones do not appear to be part of the final assembled complex. Several systematic studies in yeast by the laboratory of Walid Houry established Hsp90 role in snoRNA (small nucleolar RNAs) complex assembly. This happens in collaboration with the Pih1 and Tah1 (Pih1d1 and Rpap3 in human) co-chaperones identified by the same group [138, 209-211] and links the chaperone to pre-rRNA processing. Additional studies established Hsp90 role in different aspects of RNA processing by promoting diverse RNPs complexes formation [212-215]. Several exciting studies revealed a role of Hsp90/Hsp70 chaperone machinery in RISC complexes formation thus linking the chaperone to miRNA/siRNA biology [216-219]. Hsp90 inhibition also causes the disappearance of the so called P bodies - cytoplasmic messenger ribonucleoprotein (mRNP) foci containing translationally repressed mRNAs and RNA processing factors [217, 220]. Small-scale interactome studies has also proposed that Hsp90 activity is required for

CHAPTER 1 INTRODUCTION

assembly of RNA polymerase II complex in cytosol [65], phosphatidylinositol-3 kinase- related protein kinase family proteins (PIKKs) [221, 222] and kinetochore [223, 224]. Furthermore, Hsp90 and p23 collaborative work appears critical for telomerase assembly, subsequent nuclear import and activity [225-229]. 26S proteasome assembly has also been shown to be Hsp90-dependent [230, 231].

Several studies suggested that Hsp90 contributes to intracellular trafficking [64, 232]. Probably the most prominent example is the efficient and rapid (half-life of 5 min) nuclear translocation of steroid receptors which is dependent on the Hsp90-FKBP4 heterocomplex [233-235]. Hsp90 inhibition in this case hampers translocation and leads to proteasomal degradation of the receptors [54, 236].

Hsp90 was also found to chaperone components of the Rab GTPase cycle [237, 238]. Hsp90 plays a crucial role during the nuclear export of 60S ribosomal subunits [64]. Furthermore, Falsone et al., 2005 identified the nuclear export factor CRM1 as a potential Hsp90 interactor and suggest a role for the chaperone as a ‘‘joker’’ appointed to direct the informational flux between genome and proteome at several levels [201].

Investigation of Hsp90 interactions and their functional consequences established a role of the chaperone in DNA damage response [239]. It was found to accumulate at site of DNA damage and to chaperone PRKDC (DNA-activated protein kinase catalytic subunit) - a molecular sensor for DNA damage involved in DNA non-homologous end joining [89, 90]. Another important facilitator of DNA damage response is the scaffold protein XRCC1. Upon DNA damage and subsequent poly(ADP-ribose)polymerase 1 (PARP1) activation XRCC1 tightly binds to DNA polymerase β and recruits it to foci of DNA damage. That interaction is modulated by Hsp90, which stabilizes XRCC1 monomeric form and thus protects it from CHIP-mediated degradation [66]. A large-scale study exploiting Hsp90 inhibition revealed further that a considerable number of DNA damage response players get unstable upon depletion of the chaperone and thus could be considered potential clients [240].

Hsp90 chaperone system is an example of a great collaborative work. Numerous co- factors contribute to Hsp90 chaperone cycle, activity and specificity towards clients [111]. That is why it is of significant importance to not only focus on Hsp90 interactome as an isolated case but also investigate it as part of the bigger picture. An impressive study by Taipale et al., 2014 did so by combining mass spectrometry and quantitative high-throughput LUMIER assays to systematically characterize the complex chaperone/co-chaperone/client

CHAPTER 1 INTRODUCTION

interaction network in human cells [59]. It provided firm evidence to strengthen the hypothesis that like the well-known kinase specificity of Cdc37, other co-chaperones also have distinct specificities. Probably this is how different co-chaperones preferences contribute to Hsp90 specificity towards particular client categories and contribution to very specific biological processes.

Apart from its indispensable role for cell well-being in normal conditions, Hsp90 is essential for cell survival and recovery in critical times of environmental stress as discussed in section 1.2 of chapter 1. Although a well-established knowledge, the precise contribution of Hsp90 in protecting cells from stress remains undefined [69]. Performing genome-wide chemical genetic screen in yeast McClellan et al., 2007 provided a list of functional interactors of Hsp90 upon heat shock. They revealed that in such conditions Hsp90 is required for progression of the cell cycle, meiosis, and cytokinesis and that yeast harbouring a mutant allele of Hsp90 display morphological abnormalities consistent with defective cell cycle progression [241].

Protein-protein interactions cross-linking techniques. Despite the challenges arising from the unstable and transient nature of Hsp90 interactions, only a limited number of proteomics studies employed protein crosslinking techniques to stabilize covalently Hsp90-interactor protein complexes and thus minimize the possibility of the native complexes undergoing dissociation during cell lysis and biochemical fractionation [242, 243]. Several studies from the past were able to demonstrate the existence of a 300 kDa complex between Hsp90, the non-activated glucocorticoid receptor and the FKBP4 co-chaperone both in intact cells and in cell extracts [244-246] employing protein chemical crosslinking strategies. A more recent study by Song et al., 2012 revealed the formation of a 240-kDa heteroconjugate of the molecular chaperone Hsp90 in different cell types upon protein-protein crosslinking [247]. These studies implied the potential of chemical crosslinking to unveil novel Hsp90 interactors if combined with MS protein identification. That could turn into a promising approach to expand the range of tools available to study Hsp90 interactome.

Hsp90-dependent ubiquitylated proteome. As shown by small- and large-scale studies, abrogation of Hsp90-client interactions leads to clients destabilization, subsequent poly- ubiquitylation and degradation by the proteasome [55, 248-252]. That sequence of events provides scientists an attractive tool to identify Hsp90 interactors in addition to the standard interactome mapping techniques [203, 253]. As demonstrated in a few studies the turnover of proteins already destabilized due to Hsp90 inhibition could be halted by inactivating the

CHAPTER 1 INTRODUCTION

proteasome as well [254, 255]. That leads to Hsp90 clients accumulation as ubiquitin-tagged aggregates in the detergent insoluble fraction of cells and could be further identified by MS analysis. A crucial advantage of the approach is the fact that it offers physical interaction- independent identification of Hsp90 clients. In the light of the very unstable and transient nature of chaperone-client interactions that could be highly valuable.

Falsone et al., 2007 employed the double inhibition strategy in HeLa cells to reveal the Hsp90-dependent ubiquitylated proteome – a dataset expected to represent a large number of chaperone clients [254]. They were able to identify previously known interactors such as dual specificity mitogen-activated protein kinase kinase 1 (MAP2K1), cell division protein kinase 6 (Cdk6), alpha-enolase (ENO1), tyrosine-protein kinase CSK and propose novel ones. Investigating large-scale changes in the ubiquitinome upon Hsp90 inhibition, Quadroni et al., 2015 demonstrated a powerful tool to complement Falsone et al., 2007 experimental setup and circumvent a major drawback of the work described above – low sensitivity due to lack of ubiquitin species enrichment step [255]. They took advantage of a newly available antibody-based method and enriched ubiquitylated species making use of the di-glycine group (di-Gly) preserved at sites of lysine ubiquitination upon protein endonuclease digestion [256, 257]. Thus, the study provided a detailed picture of Hsp90-inhibition induced protein- specific and global changes in the ubiquitinome [255] .

In conclusion, the highly diverse interactome studies conducted so far revealed that Hsp90 is involved in nearly every major physiological process in yeast and human. The current list Hsp90 interactors includes more than 200 proteins and this number is growing steadily. A manually-curated and literature-based list compiled by the laboratory of Didier Picard could be found at www.picard.ch/Hsp90Int/index.php.

1.2.9. Nuclear Hsp90

Hsp90 is a highly abundant protein that predominantly localizes in the cellular cytosol. It has been traditionally viewed as a cytoplasmic chaperone despite numerous reports from the past revealing its presence inside the nucleus of normal and cancer cells. Early studies from nearly 30 years ago were able to show that both Hsp90α and Hsp90β isoforms could be found at very low levels in the nucleus of unstressed mammalian and Drosophila melanogaster cells [258-261]. Intriguingly, their nuclear abundance increases rapidly in response to stress and returns to basic levels upon recovery [258, 259, 262-265]. Furthermore, heat shock was shown to even stimulate its translocation to chromatin [259, 266]. Despite its well-established

CHAPTER 1 INTRODUCTION

nuclear localization, Hsp90 does not possess a nuclear localization signal. In fact, it appears to bear sequences in the C-terminus that suppress nuclear uptake [267]. Thus, nuclear localization might depend on co-chaperones or associated clients [268].

Surprisingly, despite all the evidence for the existence of Hsp90 nuclear pool it remained elusive for the upcoming years. This could be attributed to Hsp90 low abundance in that cellular compartment and the corresponding technical limitations at that time. Nuclear Hsp90 knowledge remained almost exclusively limited to its interplay with steroid receptors. Glucocorticoid, androgen and estrogen steroid receptors are among the earliest identified nuclear Hsp90 clients and the first instances of Hsp90-regulated transcriptional events [245, 269-276]. The chaperone has been found to modulate steroid receptors assembly, disassembly, nuclear import and activity [233-236, 277-279]. It maintains the receptor in a stable inactive complex awaiting high-affinity hormone binding and rapid activation. To date, chaperoning by the Hsp90 machinery has been attributed to all classical steroid receptors [63].

In the last decade numerous reports suggested important roles of Hsp90 in nuclear events [280]. Nuclear Hsp90 has been found to regulate the function of the general transcription machinery, diverse transcription factors as well as epigenetic factors thus contributing to regulation of gene expression (FIGURE 6). Notably, many nuclear resident clients have been found to drive and maintain oncogenic transformation [281]. In addition to hormone receptors, Hsp90 physically interacts and affects the functioning of a variety of TFs, such as Stat3, Stat5, NF-kB, Hsf1, p53, c-Myc and others [73, 207, 281-283]. Stat3 represent a class of TFs activated upon tyrosine phosphorylation, typically in response to cytokine stimulation [284]. Shah et al., 2002 has shown that Stat3 signalling is a chaperoned pathway that involves Hsp90 as a mechanism to preserve signalling during fever [285, 286]. Both Stat2 and Stat3 do not fit the criteria of classical Hsp90 client proteins since their concentrations remain essentially unchanged upon Hsp90 inhibition. This sheds light on an additional mode of action – apart from promoting protein stability Hsp90 maintains client proteins in an active configuration.

NF-κB is a pivotal TF regulating the expression of pro-inflammatory cytokines and chemokines. Hsp90 promotes NF-κB activation modulating its association with the IKK complex and the subsequent activation of the latter [207, 287]. In addition, certain stimuli promote recruitment of Hsp90 and NF-κB effectors to the TNF-α promoter, in tandem with the unliganded ER [288].

CHAPTER 1 INTRODUCTION

A key mechanism to regulate proteins function is modulating their abundance. Given the protein stabilization activity of Hsp90 and its close collaboration with the ubiquitin- proteasome system, the chaperone offers a quick way of fine-tuning cellular processes by either stabilizing a client or handing it over to the degradation machinery. Its interplay with the CHIP ubiquitin ligase has been shown to modulate the function of the intrinsically unstable Myc protein. Thus, the chaperone enables its ability to integrate diverse signalling inputs and coordinate cellular responses [289, 290]. Furthermore, Hsp90 has been found associated with DNA-bound Myc-containing complexes [291].

Figure 6. Hsp90 chaperones diverse transcription factors and epigenetic modifiers. Hsp90 interacts with either soluble (1) or chromatin-tethered (2) transcription factors (TF). Hsp90 interaction with epigenetic client proteins within the nuclear compartment (5) may repress (6) or activate (7) gene expression - most notably the expression of tumor-suppressor and oncogenic driver genes. Collectively, these Hsp90-dependent nuclear events support a variety of key tumor-supportive pathways (enclosed boxes) critical for cancer progression and therapeutic resistance (image from [280]).

Another intriguing role for Hsp90 in the nucleus appears to be its interaction with the general transcription machinery. It not only mediates Pol II assembly in the cytoplasm and

CHAPTER 1 INTRODUCTION

nuclear import of the fully assembled holoenzyme [292] but regulates Pol II pausing via stabilization of the negative elongation factor (NELF) complex [293].

While the connections to transcription are becoming well-established, a new intriguing lead in the field of nuclear Hsp90 investigation emerges to attract attention – its affiliation to the world of epigenetics. The impressive work of Zhao et al., 2005 mentioned earlier not only created a huge and reliable Hsp90 interactome database in yeast but also linked the chaperone to chromatin biology. The group identified two novel Hsp90 co-factors, Pih1 and Tah1 (Pih1d1 and Rpap3 in human). They were found to interact both physically and functionally with DNA helicases Rvb1/Rvb2, which are key components of several chromatin- remodelling factors. This implied possible role of Hsp90 in epigenetic gene regulation. Later on the human homologs of Pih1, Tah1, RVB1L and RVB2L were identified as Hsp90 interacting partners [139]. Early reports already gave a subtle hint towards Hsp90 possible role in epigenetic events by demonstrating its interactions with histone proteins [294, 295] and speculating on its potential role of a facilitator of histone–DNA binding [296, 297]. In addition, Hsp90 induces chromatin accessibility changes through interactions with chromatin- modifying/remodeling complexes that could either activate (Smyd3, trithorax/MLL, RSC complex) [135, 298, 299] or repress (Ezh2) [300] gene expression (FIGURE 6).

Polycomb (PcG) and trithorax (Trx) families of proteins are key catalytic components of epigenetic complex controlling cell-lineage specification during normal development. Trx facilitates transcription activation by depositing H3K4me3 marks [301] whereas PcG modulates transcription repression by depositing H3K27me3 marks [302]. Interestingly, mutations of Trx family members and Hsp90 in Drosophila generate similar phenotypes [303]. In agreement with this observation Trx was later on identified as an Hsp90 client and observed together with the chaperone at the promoters of Trx target genes [304]. The homologous human protein MLL1 was identified as an Hsp90 client in the same study and subsequently shown to facilitate the heat-shock response induced by Hsp90 inhibition [305]. In humans again, the catalytic component of Polycomb-repressive complex 2 (Prc2) named methyltransferase enhancer of zeste homolog 2 (Ezh2) has been identified as an Hsp90 client [300]. Intriguingly, in a fashion similar to Trx discussed above, Drosophila Polycomb protein was required for gene repression during heat stress [306].

Smyd2 and Smyd3 members of the SET and MYND domain-containing methyltransferases provide another example to link Hsp90 to chromatin modifications. They are both reported to interact with the chaperone [107, 307]. These enzymes methylate histone

CHAPTER 1 INTRODUCTION

and non-histone targets and control gene expression and DNA damage response. Interestingly, Hsp90 was also identified as a substrate for Smyd-directed methylation [107, 308] similarly to its phosphorylation by well-established kinases clients.

In conclusion, several studies link Hsp90 to transcription regulation at different levels. Given its chaperoning activity and interplay with the ubiquitin-proteasome system, it could be the factor providing an opportunity for a quick adjustment of gene expression according to cellular requirements. Together with the protein degradation machinery it could facilitate regulated protein degradation [309, 310] - a mechanism to affect gene expression regulators and cause rapid change in transcription in response to environmental stimuli. Thus the cell could react accordingly and adjust quickly enough.

1.2.9. Challenges and open questions

Despite the large amount of data available, Hsp90 interaction network cannot be considered complete yet. Certain clients might have easily escaped identification due to the difficulties in the interactome studies described earlier. At the same time, false positive hits may have been considered due to lack of direct binding confirmation. On the other hand, the understanding of Hsp90 interactome data remains challenging due to multiple functions of the chaperone and the complexity and diversity of the structures in which it participates. Finally, in contrast to yeast, the high throughput analysis of mammalian Hsp90 interactors is limited.

Notably, the two related Hsp90α and Hsp90β isoforms emerged as a result of a duplication event about 500 million years ago coinciding with the emergence of vertebrates [311]. That time permitted the development of some divergent functions for the two proteins that have only recently been investigated [149]. Accumulating evidence strongly hint at isoform-specific functions [156, 312-314]. Strikingly, GCUNC45A was found to be the first human Hsp90 co-chaperone to differentiate between Hsp90α and Hsp90β isoforms binding preferentially Hsp90β in vitro [315]. However, most of the Hsp90 studies conducted so far did not take into account the potential for isoform-specific interactions and functions. The scarce knowledge obtained so far comes from isolated studies either lacking large-scale perspective or isoforms side-by-side comparison. A good starting point for the investigation of Hsp90α and Hsp90β functional differences in mammals will be the comparison of their corresponding interactors side by side in a single system.

CHAPTER 1 INTRODUCTION

Strikingly, Hsp90 interactions under proteotoxic stress have been barely tackled. McClellan et al., 2007 genome-wide chemical genetic screen in yeast discussed in section 1.2.7 of chapter 1 remains the only large-scale study to provide a subtle hint of Hsp90 role upon proteotoxic stress [241]. However, it does not provide any evidence for physical interactions. In addition, no Hsp90 interactome data upon stress exists for mammalian cells.

Surprisingly, most of the large-scale proteomic studies which investigated direct interactions of Hsp90 failed to detect a considerable number of potential nuclear clients. That holds true even for the well-established clients category of transcription factors. These observations are consistent with the fact that proteomics techniques usually detect only the most abundant proteins. Nuclear Hsp90 clients definitely appear underrepresented in whole- cell preparations. This justifies the need for a large-scale approach to study the interactions of nuclear Hsp90 exclusively. That kind of study was done so far neither in yeast nor in human cells. It promises to bring exciting novel observations given the hints provided by individual small-scale investigations of nuclear Hsp90. What makes nuclear Hsp90 role upon stress even more intriguing is chaperone rapid translocation to the nucleus in this condition. The significance of the observation is still not clear. Is it possible that Hsp90 contributes to the so crucial rapid gene expression changes to ensure cellular adjustment and survival? A step to bring us closer to that understanding will be to reveal how nuclear Hsp90 interaction network changes in response to stress compared to normal conditions.

An important question regarding nuclear Hsp90 function remains largely unclear: does Hsp90 modulate the stability and activity of soluble, non-DNA-bound forms of clients or does Hsp90 directly interact with chromatin-bound forms? An early report in Drosophila documented Hsp90 association with specific chromatin domains engaged in active gene transcription following heat shock, known as heat shock-induced puffs [316]. Only few studies have confirmed Hsp90 recruitment to particular promoter sites in human so far [207, 277, 288, 291]. The single genome-wide investigation so far provides a slightly more detailed picture of Hsp90 localization at chromatin [317]. However, it provides data with much room for improvement due to the challenges regarding Hsp90 ChIP-seq analysis. Genome-wide occupancy analysis will be required to reveal Hsp90-bound chromatin loci and clearly understand its role at chromatin.

Hsp90 is predominantly localized and characterized in the cellular cytosol. Most of the PTMs findings described in section 1.2.3.2 of chapter 1 originate from studies focusing either on the cytosolic or the whole-cell pools of the chaperone. Both the conditions do not favor

CHAPTER 1 INTRODUCTION

nuclear Hsp90 PTMs identification due to its lower abundance and thus underrepresentation in the corresponding samples analyzed. Limited investigation offers scarce details on nuclear Hsp90 PTMs. Recent studies revealed Hsp90α Thr-5 and Thr-7 phosphorylation during apoptosis and upon DNA damage [89, 90]. Interestingly, work by Abu-Farha et al., 2011 demonstrates that Lys-209 methylation probably occurs in the nucleus and is dependent on the presence of histone H3 [107]. The so far poorly investigated nuclear Hsp90 pool increasingly attracts attention and PTMs description will further shed light on its characteristics.

CHAPTER 1 AIMS

2. Aims

In the light of the open questions discussed above it becomes clear that a comprehensive characterization of nuclear Hsp90 is needed. Therefore, in this part of my thesis, I investigated:

2.1. Nuclear Hsp90α and Hsp90β interactomes in human cells.

Such efforts would enhance our understanding of the following aspects: • Nuclear Hsp90 function • Nuclear Hsp90α and Hsp90β clients differences and their functional consequences • Interactions dynamics upon proteotoxic stress and their contribution to stress response • Nucleus-resident co-chaperones that fine tune Hsp90 function

2.2. Nuclear Hsp90α and Hsp90β PTMs.

Such investigation would further shed light on: • Nuclear Hsp90 fine-tuning by PTMs • Nuclear Hsp90α and Hsp90β PTMs differences and their functional consequences • PTMs dynamics in response to proteotoxic stress and their contribution to stress response • PTMs putative effect on Hsp90 entry to or exit from the nucleus • Cytosolic and nuclear Hsp90 PTMs comparison

CHAPTER 1 MATERIALS AND METHODS

3. Materials and Methods

3.1. Cell Culture and Cell Lines

T-REx™-293 cells (Thermo Fisher Scientific, #R710-07) were grown in DMEM (Sigma Aldrich, #D5671) supplemented with 10 % Fetal Bovine Serum (Sigma Aldrich, #F0804), 2 mM L-Glutamine (Sigma Aldrich, #G7513), Penicillin/ Streptomycin (Sigma Aldrich, #P4333) and 5 μg/ml blasticidin (Omnilab, #A3784,0010). K-562 cells (ATCC® CCL-243™) were grown in IMDM (Gibco, #21980065) supplemented with 10 % Fetal Bovine Serum (Sigma Aldrich, #F0804) and Penicillin-Streptomycin (Sigma Aldrich, #P4333). All the cell lines were maintained at 37 °C and 5 % CO2.

3.2. Antibodies

Anti-H3 (ab70550), anti-SNRP70 (ab83306), anti-Hsp90α (ab79849 and ab115645), anti-Hsp90β (ab53497 and ab80159), anti-CDC37 (ab2800), anti-Hsp70 (ab2787), anti-p23 (ab2814), anti-CHIP (ab134064) and anti-gamma H2A.X (ab26350) antibodies were purchased from Abcam. Anti-Ubiquitin (Z0458) antibody was purchased from Dako. Anti- Ubiquitin, Lys48-specific (05-1307) antibody was obtained from Merck. Anti-HCF1 antibodies were purchased from R and D Systems (AF6254) and Bethyl Laboratories (A301- 399A). Anti-HA (H3663) and anti-GAPDH (G8795) antibodies were obtained from Sigma. ECL anti-mouse (NA931V), anti-rat (NA935) and anti-rabbit (NA934) HRP-linked antibodies were purchased from GE Healthcare. Several antibodies were kind gifts from in- house based laboratories: anti-caspase-3 antibody (Prof. Grosschedl), anti-p62 (Dr. Pichler).

3.3. Cell Fractionation

Cells either remained resting (grown at 37 ) or were subjected to specific treatment prior to fractionation. For heat shock experiments℃ cells were exposed to 43 C for 10, 20, 30, 60 or 90 min. Cells heat shock recovery experiments cells were exposed to ⁰43 for 90 min and then recovered at 37 for 1, 2, or 4 hours. For Hsp90-dependent ubiquitylated℃ proteome analysis Bortezomib (50nM)℃ and NVP-AUY922 (500nM; Novartis Pharmaceuticals) treatment was used alone or in combination for 4, 8 or 16 hours. To test the effect of different stresses on nuclear protein ubiquitination cells were treated with either DTT (10 mM, 30 min), tunicamycin (5 μg/ml, 1.5 hours), H2O2 (50 μM, 30 min), sorbitol (200 mM, 30 min), camptothecin (1 μM, 3 hours) or etoposide (5 μM, 9 hours), starved (grown in serum-free media for 90 min ) or subjected to Arginine analogue-caused proteotoxic stress (1mM CAV

CHAPTER 1 MATERIALS AND METHODS

instead of normal Arginine in growth medium, 12 hours). To investigate nuclear proteins ubiquitination upon Hsp90i prerequisite cells were pre-treated with either cycloheximide (30 μg/ml, 3 hours), ivermectin (25 μM, 1.5 hours), importazol (40 μM, 1.5 hours), 2-DO8 (30 μM, 1.5 hours), actinomycin D (1 μg/ml), JNKi (50 μM, 1.5 hours), or nocodazol (10 μg/ml, 1.5 hours) and then subjected to heat shock (43 for 30 min).

Upon treatment, cells were either scraped ℃(HEK) or collected by centrifugation (K562) and washed twice with ice cold PBS PH 7.9 (0.1mM CaCl2, 1mM MgCl2). All the buffers used for the subsequent fractionation were supplemented with the following protease inhibitors: 3 µg/ml aprotinin, 10 µg/ml leupeptin, 1 µg/ml pepstatin, 0.1 mM PMSF. For most of the experiments fractionation was performed according to Margueron et al., 2008 [318]. An aliquot of the collected and washed cells was taken out in RIPA buffer (50 mM Hepes pH

7.9, 150 mM NaCl, 0.5 % sodium deoxycholate, 0.1 % SDS, 1% Triton-X, 5 mM MgCl2, 10 % glycerol) and saved as total cell extract (TCE). Remaining cells were resuspended in

Buffer A (10 mM Hepes pH 7.9, 5mM MgCl2, 0.25 M Sucrose) and incubated on ice for 10 minutes. NP-40 detergent was added to the cell suspension to a final concentration of 0.1% for HEK and 0.3 % for K562. Cells were then passed through an 18G needle (10 times for HEK and 20 times for K562) and centrifuged at 2000 xg for 10 min at 4 . The resulting

supernatant was saved as cytosol (CYT). The nuclear pellet was washed twice℃ with buffer A. An aliquot was taken out in RIPA buffer and saved as total nuclear extract (TNE). The rest

was resuspended in 0.5M buffer B (20 mM Hepes pH 7.9, 0.5 M KCl, 1.5 mM MgCl2, 0.1 mM EDTA, 10 % Glycerol) and incubated for 30 min at 4 (end-over-end rotation). The suspension was centrifuged and supernatant was saved as nucleoplasm℃ (NP). The chromatin pellet was resuspended in 2M buffer B (20 mM Hepes pH 7.9, 2 M KCl, 1.5 mM MgCl2, 0.1 mM EDTA, 10 % Glycerol), and incubated for 30 min at 4 (end-over-end rotation). TCE,

TNE and chromatin (CHR) were sonicated for 15 cycles (30℃ sec on/30 sec off, high power) using Bioruptor sonicator. Samples were centrifuged at 20 000 xg for 10 min to remove debris and protein amount was determined by Nanodrop.

For nuclear Hsp90-dependent ubiquitylated proteome analysis cell fractionation was performed as described by Falsone et al., 2007 [254]. All the buffers used were supplemented with 2 mM sodium orthovanadate, 50 mM sodium fluoride, 2mM NEM (Sigma Aldrich # E3876) and 5 μM IAA in addition to the protease inhibitors already described. Cells and nuclei were resuspended in Loading buffer (50 mM Tris pH 6.8, 2 % SDS, 10 % Glycerol), heated at 95 for 10 min and sonicated to generate TCE and TNE. Additional soluble (SNE)

℃ 32

CHAPTER 1 MATERIALS AND METHODS

and insoluble (INE) nuclear extracts were generated. For this nuclei were resuspended in Lysis buffer (50 mM Tris pH 7.5, 1 % NP-40, 2 mM EDTA, 100 mM NaCl) and incubate for 20 min at 4 (end-over-end rotation). Benzonase (Sigma Aldrich, #E1014) was added to a

final concentration℃ of 100 U/ml and samples were incubated for 30 min on ice. Following sonication extracts were spun at 10 000 xg for 10 min. Resulting supernatant was saved as SNE and pellet – washed in PBS, resuspended in IEF buffer (7 M Urea, 2 M thiourea, 2 % CHAPS) and saved as INE. Nucleoplasm and chromatin were isolated as described earlier. Additional soluble (SCE) and insoluble (ICE) chromatin extracts were generated. For this chromatin was resuspended in Lysis buffer and incubate for 20 min at 4 (end-over-end

rotation). Benzonase was added to a final concentration of 100 U/ml and℃ samples were incubated for 30 min on ice. Following sonication extracts were spun at 10 000 xg for 10 min. Resulting supernatant was saved as SCE and pellet – washed in PBS, resuspended in IEF buffer and saved as ICE.

3.4. SDS-PAGE

Polyacrylamide gels of different percentage were casted according to Rotiphorese® Gel 30 (ROTH, #30291) instruction manual. Separating gels comprised of the following components: 30 % acrylamide mix, 1.5 M Tris-HCl pH 8, 0.1 % SDS. Stacking gels (5%) contained 30 % acrylamide mix, 1 M Tris-HCl pH 6.8, 0.1 % SDS). Polymerization of gels was catalysed by 0.1 % APS and 0.01 % TEMED. Gels were immerged in SDS-PAGE running buffer (25 mM Tris, 192 mM glycine, 0.1 % SDS). Protein samples were mixed with 4X loading buffer (50 mM Tris-HCl pH 6.8, 2 % SDS, 10 % glycerol, 1 % β- mercaptoethanol, 0.02 % bromophenol blue) heated at 95 ℃ for 5 min, briefly spinned down and loaded on gels. Electophoresis was then run at 150-180 V for 60-90 min.

3.5. Silver staining

Upon electrophoresis polyacrylamide gel was fixed in fixation solution (10 % HAc, 45

% MeOH) for 15 min and incubated with Farmer’s reducer solution (30 mM K3Fe(CN)6, 30

mM Na2S2O3) for 2 min. Several washes in milliQ water with gentle shaking were performed until the yellow colour completely disappeared. Gel was then transferred into 0.1 % (g/v)

silver nitrate for 20 min. It was subsequently washed with milliQ water and 2.5 % Na2CO3 - 30 sec each wash. Silver stained gel was then developed with 0.1 % formaldehyde in 2.5 %

Na2CO3. Reaction was stopped by adding excess of 10 % HAc. Gel was washed with milliQ water and stored in 1 % HAc at 4 °C.

CHAPTER 1 MATERIALS AND METHODS

3.6. Western Blotting

Protein gels, nitrocellulose membranes (GE Healthcare, #10401396) and filter paper were briefly rinsed in transfer buffer (25 mM Tris, 192 mM glycine, 20 % methanol). Western blotting sandwich was assembled in the following order: filter paper-gel-membrane- filter paper and placed in the running chamber full with running buffer. Protein transfer was carried out at 300 mA for 60 min. Nitrocellulose membrane was then blocked with blocking solution (5 % non-fat milk in PBST) for 1 hour or ON. Primary antibody incubation was carried out in blocking solution for 1 hour or ON. Following 3 washes per 10 min in PBST membranes were exposed to a secondary antibody (1:10 000 dilution in blocking solution). After 3 washes per 10 min with PBST membranes were developed using ChemiDoc Imaging System (Biorad) and one of the following developing reagents: SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher, #34095) and Amersham ECL Western Blotting Detection Reagents (GE Healthcare, #RPN2209).

3.7. Nuclear Protein Affinity Purification coupled to Mass Spectrometry

All the buffers used (except for elution buffer) were supplemented with the following protease inhibitors: 3 µg/ml aprotinin, 10 µg/ml leupeptin, 1 µg/ml pepstatin, 0.1 mM PMSF. For Hsp90α and Hsp90β interactome analysis buffers A, B and D contained ATP regeneration system components (10 mM ATP, 0.2 M phosphocreatine, 1 mg/ml creatine kinase) to keep ATP concentration high enough for the ATP-dependent cycle of the chaperone. For Hsp90α and Hsp90β post-translational modifications (PTM) analysis all the buffers used (except for elution buffer) contained protein phosphatases (2 mM sodium orthovanadate, 10 mM sodium fluoride, 2 mM sodium pyrophosphate) and acetylases (10 mM sodium butyrate) inhibitors kindly provided by the MS facility in-house.

3.7.1 Stable Cell Lines Generation

Hsp90α, Hsp90β, p23, Cdc37, Hsp70, RPAP3 and RFP cDNAs were cloned into pC5 destination vector. Flp-In HEK293 T-REx cells (Thermo Fisher #R78007) stably expressing the tet repressor were co-transfected with the corresponding expression plasmids and the pOG44 vector (for co-expression of the Flp-recombinase) using the Lipofectamine 3000 (Life Technologies, #L3000-015) according to manufacturer’s protocol [319] . Two days after transfection, cells were selected in zeocin-containing medium (100 μg/ml) for 2–3 weeks to generate stable cell lines. Resulting HA-Strep tagged protein clones were used for further experiments.

CHAPTER 1 MATERIALS AND METHODS

3.7.2 Nuclear Protein Extract Preparation

Flp-In HEK293 T-REx cells stably expressing Hsp90α, Hsp90β or RFP were grown to 70 % confluency. Recombinant protein expression was induced by adding 1 µg/ml tetracyclin (Sigma Aldrich, #T7660-5G) directly into growing medium. For each experimental condition 10x 15-cm plates (around 200 x 106 cells) were used. Cells were collected by scraping 16 hours upon induction. For the heat shock experiments the last 90 min of induction were carried out at 43 to mimic stress conditions. Nuclei were then isolated as already described

in section 3.3 of ℃chapter 1). Nuclear protein extracts were generated using the classical high- salt Dignam extraction protocol [320] with slight modifications. Briefly, nuclei were resuspended in 400 μl of buffer B per plate (20 mM Hepes pH 7.9, 420 mM KCl, 1.5 mM

MgCl2, 0.1 mM EDTA, 10 % glycerol) and incubate for 30 min at 4 (end-over-end

rotation). Suspension was diluted 1:2 with 800 μl of buffer D per plate (20℃ mM Hepes pH

7.9, 1.5 mM MgCl2, 10% glycerol). To minimize binding of endogenous biotinylated proteins to the Strep-Tactin resin 100 µg/ml of Avidin (IBA, #2-0204-015) were added. Following benzonase addition (10 U/ml) samples were centrifuged at 70 000 xg for 30 min at 4 .

Supernatant was recovered and immediately used for protein affinity purification. ℃ For DSP (dithiobis succinimidyl propionate; Thermo Fisher, #22585) cross-linking experiments cells were either grown at 37 or subjected to heat shock (43 for 90 min).

Cells were exposed to 2 mM DSP in pre-warmed℃ PBS for cross-linking for 30℃ min (for heat shocked cells these were the last 30 min of heat shock treatment). DSP was then quenched with 50 mM Tris pH 7.4 for 10 min at RT. Cells were collected by scraping and washed in 50mM Tris, pH 7.4. Pellet was resuspended in buffer A (1ml per plate) and incubated on ice for 10 min. The suspension was then supplemented with 0.5 % NP-40, passed through an 18G needle (20 times) and spun down at 2 000xg for 10 min at 4 . Obtained nuclei were washed twice with buffer A, resuspend in 3 ml of RIPA buffer, supplemented℃ with 100 U/ml of benzonase and incubate 30min on ice. Material was sonicated 10 cycles, 30sec/30sec and centrifuged at 70 000 xg for 30min at 4 . Supernatant was recovered and immediately used for protein affinity purification. ℃ 3.7.3 Protein Affinity Purification

Affinity purification of recombinant HA-Strep-tagged proteins was performed utilizing the IBA GmbH Strep-tag® - Strep-Tactin system (IBA, #2-1201-025). Briefly, Strep-Tactin sepharose columns were washed twice with 1 ml of washing buffer (20 mM Hepes pH 7.9,

CHAPTER 1 MATERIALS AND METHODS

1.5 mM MgCl2, 10 % Glycerol, 150 mM KCL, 0.1 % NP-40). Sample was applied to the column (1 ml at a time) to pass through gravity force. Following six washes with washing buffer to eliminate contaminating proteins bound complexes were released with elution buffer (20 mM Hepes pH 7.9, 1.5 mM MgCl2, 10% glycerol, 150 mM NaCl, 20 mM desthiobiotin, 0.1% NP-40). A small aliquot was kept to be run on a gel and visualized by silver staining. The remaining amount of the eluates was snap frozen in liquid nitrogen and kept at – 80 until needed.

3.7.4. Protein℃ Digestion and Mass Spectrometry Analysis

Proteins from the obtained eluates were precipitated by addition of TCA to a final concentration of 25 %. Following 1 hour incubation on ice samples were centrifuged at 18 000 xg for 15 min at 4 °C. Resulting pellet was washed three times with acetone (- 20 °C pre- cooled) and air-dried. Upon reconstitution proteins were digested either in-solution or in-gel for MS analysis.

In-solution digestion was performed for recombinant Hsp90 interactome and Hsp90- dependent ubiquitylated proteome analyses. TCA-precipitated protein pellet was resuspended in Urea buffer (10 mM Hepes, pH 7.9, 6 M Urea/ 2 M Thiourea) and supplied with TCEP to a final concentration of 5 mM and incubate at 25°C in Thermomixer (max speed) for 20 min. Iodoacetamide was added to a final concentration of 10 mM and incubation continued for another 15 min (in the dark). Each samples was then supplemented with 250 ng of LysC endopeptidase (Wako, #125-02543) and incubate at 25°C in Thermomixer (max speed) for 3 hours. 250 ng trypsin (Promega #V5113) was added upon dilution with 200 µl of ABC buffer

(50 mM NH4HCO3 in MilliQ water) to drop urea concentration down to 2 M. ON digestion was stopped by addition of 10 % TFA to a final concentration of 1%. To further reduce urea concentration 200 µl of 0.1 % TFA was added. Samples were spun at 13 000 xg for 10 min at RT. Supernatant was recovered and mixed with stage buffer A [(0.5 % HAc, 0.02 % TFA) - 80ul of buffer per 100ul of sample] before loaded on stage tips. Stage tips were prepared using a solid phase extraction disk (Sigma Aldrich, #66883-U) wet in methanol, excising a small disk and inserting it into a low binding tip. Tip was successively washed with 80 μl of 100% methanol, Stage buffer B and Stage buffer A by spinning down at 15000rpm for 2 min. Samples was then loaded on stage tips by spinning down at 1 000 xg for 4-5 min. Stage tips were washed with Stage buffer A (3x 80 μl). Bound peptides were eluted with elution buffer (0.1 % TFA, 50 % ACN), speed vacuumed to get rid of ACN reduce volume and stored at 4 before MS analysis. They were eventually identified by nanoLC-MS analysis

℃ 36

CHAPTER 1 MATERIALS AND METHODS

on an Agilent1200 nanoHPLC system interfaced with a Thermo Fisher OrbitrapXL+ETD mass spectrometer.

In-gel digestion was performed for Hsp90 PTMs analysis. TCA-precipitated protein pellet was resuspended in NuPAGE LDS Running Buffer (NOVEX by Life Technologies, #NP0001) supplemeted with 10 mM DTT and boiled for 5 min at 95 . They were submitted to the in-house MS facility for further processing. Briefly, proteins℃ were run on a gel, stained with Coomassie and bands corresponding to Hsp90 were cut. Upon in-gel digestion by LysC and trypsin endopeptidases peptides were identified by nanoLC-MS analysis on an Agilent1200 nanoHPLC system interfaced with a Thermo Fisher OrbitrapXL+ETD mass spectrometer.

3.8. SILAC labelling (Stable isotope labelling with amino acids in cell culture)

SILAC was done as described by Ong et al., 2002 [321]. HEK293 cells were grown in medium supplemented either with L-arginine-13C6 and L-lysine-2H4 (heavy) or with unlabelled L-arginine and L-lysine (light) for at least five cell divisions. Light-medium grown cells were treated with Bortezomib (50nM) and heavy-medium – with Bortezomib (50 nM) and NVP-AUY922 (500 nM.) Treatment lasted for either 4 or 8 hours. Then insoluble (INE) nuclear extracts were generated (as described in section 3.3 of chapter 1), prepared for (as described in section 3.7.4 of chapter 1) and subjected to Mass Spectrometry analysis.

3.9. Di-Gly Immunoprecipitation

All the buffers used for the subsequent fractionation were supplemented with 3 µg/ml aprotinin, 10 µg/ml leupeptin, 1 µg/ml pepstatin, 0.1 mM PMSF, 2 mM sodium orthovanadate, 50 mM sodium fluoride, 2mM NEM and 5mM IAA. Insoluble (INE) nuclear extracts were generated as described in section 3.3 of chapter 1 and further processed for MS analysis of ubiquitylated species as already described [322, 323]. Briefly, they were treated consecutively with 1 mM DTT and 5.5 mM CAA for 35 min and 45 min, respectively (protected from light). Samples were diluted with ABC buffer to reduce urea concentration to 2 M. Proteins were proteolysed with 250 ng of LysC endopeptidase and incubate at 25 °C in Thermomixer (max speed) for 3 hours. 250 ng trypsin was added for ON digestion. The reaction was stopped by addition of 10 % TFA to a final concentration of 1 %. Samples were spun at 13 000 xg for 10 min at RT to remove precipitates. Peptides were purified using reverse-phase Sep-Pak C18 cartridges (Waters, #WAT036815) by gravity flow. For this cartridges were equilibrated with 3 ml 100 % methanol, 6 ml 0.1 % TFA, 6 ml 80 % ACN.

CHAPTER 1 MATERIALS AND METHODS

Peptides were loaded, washed with 12 ml MilliQ water and eluted with 3 ml 80 % ACN. Eluates were speed vacuumed to reach a volume of 200 μl and supplemented with 200 μl

immunoprecipitation buffer (50 mM MOPS pH 7.2, 10 mM NaPO4, 50 mM NaCl). Centrifugation at 14 000 xg was carried out to remove insoluble material. Supernatant was recovered and incubated with di-Gly antibody (Lucerna, #30-1000) ON with end-over-end rotation at 4 . Antibody-peptide complexes were captured by Protein G sepharose beads

(GE Healthcare,℃ #17061801). Beads were washed three times with immunoprecipitation buffer followed by three washes with MilliQ water. Enriched peptides were eluted with 0.15 % TFA and further purified on stage tips as described in section 3.7.4 of chapter 1.

3.10. RNA isolation

RNA isolation was done using TRI-Reagent (Sigma Aldrich, #T9424) according to manufacturer’s instructions. For HEK293, cell lysis was done directly on the cell culture dish. 1ml of TRI-Reagent was used per 10 cm. For K562, cells were pelleted by centrifugation and resuspended in1ml of TRI-Reagent per 10x 106 cells. Samples were incubated at room temperature for 5 min to dissociate nucleo-protein complexes. 0.2ml of chloroform was added per 1ml of TRI-Reagent used for cell lysis. Samples were shaken vigorously 15 seconds, incubated at room temperature for 10 min and spun down at 12 000 x g for 10 min at 4 °C. The upper aqueous phase was transferred into a fresh tube and supplemented with isopropanol (0.5 ml per 1 ml of TRI-Reagent used). Sample was mixed by inverting and incubated at room temperature for 5-10 min. RNA pellet was washed with 1 ml of 75 % ethanol by centrifuging at 10 000 xg for 10 min at 4 °C. It was air dried and re-dissolved in water. RNA concentration was measured using Nanodrop 1000.

3.11. Immunofluorescence

Microscopy slides or cover slips were coated with fibronectin (1ug/ml in PBS; Sigma Aldrich, #F2006-1MG) for 1h at RT and washed with PBS. HEK cells were plated and grown in DMEM. When desired 2 % PFA fixation was performed for 15 min at RT. Following two brief washes with PBS cells were permeabilized with 0.25 % Triton-X for 10 min at RT. Slides/cover slips were washed 3 times per 5 min with PBS and exposed to blocking solution (1 % BSA in PBST) for 30 min and incubated with a primary antibody of interest in blocking solution for an hour or ON. Cells were then washed 3 times per 5 min with PBS and stained with a secondary antibody conjugated to a fluorescent dye (1:200 dilution in blocking solution). Following 3 washes per 5 min in PBS cells were exposed to 0.1 μg/ml DAPI for 1

CHAPTER 1 MATERIALS AND METHODS

min to stain DNA, briefly washed again, and mounted with Fluoromount G (eBioscience, #00-4958-02). Coverslips were sealed with nail polish to prevent drying and stored at 4

until inspection. ℃ 3.12. Calcium Phosphate Transfection

0.8x 106 cells were plated in a 6-well plate. Next day growth medium was changed 1h

prior to transfection. 180 μl of transfection solution [5 μg of DNA, 0.3 M CaCl2, 2X HBS

buffer (280 mM NaCl, 1.5 mM Na2PO4)] per well was prepared and added to cells. Growth medium was changed 5 hours post transfection and cells were analysed 24-48 hours post transfection.

3.13. Synthetic Lethality Screen

The shRNA screen was performed in the laboratory of Cornelius Miething (UniKlinik, Freiburg). It was carried out according to Scuoppo et al., 2012 [324] with minor modifications. Briefly, oligonucleotides that target around 700 different transcription regulation factors were synthesized on a 55k customized oligonucleotide array (Agilent Technologies). Barcoded PCR products were individually cloned in TRE/BAF vector (allowing doxycycline inducible expression) to generate a plasmid shRNA library. Viruses carrying individual shRNAs were produced by transient transfection of Platinum-E retroviral packaging cells using TurboFect (Thermo Fisher #R0532) and used to infect K562 cells. ShRNA-mediated knockdown of selected targets was achieved upon doxycycline induction (1 µg/µl doxycycline final concentration). Cells were subsequently treated with either DMSO (mock treatment) or with 15 nM NVP-AUY922 (Hsp90 inhibition). Four days upon treatment cells were harvested and barcode populations were deep-sequenced.

CHAPTER 1 RESULTS

4. Results

4.1. Nuclear Hsp90 physical interactome investigated by affinity purification

Given the huge amount of proteomics data available for HEK293T cells [59, 68, 201, 325], we decided to exploit that particular cell line for the nuclear Hsp90 interactome study. The chosen Flp-In HEK293 T-REx expression system [319] offered an additional advantage - rapid generation of stable cell lines that ensure inducible homogenous expression of proteins of interest. The N-terminal tagging of Hsp90 left its C-terminus unmodified ensuring successful formation of functional dimers. Recombinant Strep-HA-tagged Hsp90α or Hsp90β (indicated by an arrow) were expressed upon doxycycline induction at levels lower than their endogenous counterparts to avoid introduction of artefacts due to overexpression (FIGURE 7A). Strep-HA-tagged RFP cell line was generated as well and used as a negative control – that low molecular weight protein (~27 kDa) is easily expressed, does not aggregate and has no cytotoxic effects.

To specifically identify nuclear Hsp90 interactors and not allow the much more abundant cytosolic ones to become an obstacle, our experimental approach involved cellular fractionation (FIGURE 7B). Nuclei were isolated and nuclear proteins were extracted by the classical high-salt Dignam protocol [320]. Hsp90-bound complexes were pulled down by Strep-Tactin affinity purification taking advantage of the Strep-tag for an efficient purification step. For this purpose, the influence of ionic strength, pH and temperature as well as of additives like detergents and metal ions was tested. The finally optimised conditions involving molybdate to stabilize Hsp90 interactions [326, 327] were applied for affinity purification. To investigate how proteotoxic stress alters chaperone interactions and how these changes contribute to cell survival and recovery two experimental conditions were analysed : normal growth control (Ctl; cells grown at 37 °C) and heat shock treatment (HS; cells incubated at 43 °C for 90 min). Hsp90-bound complexes were subsequently visualized by silver staining (red arrows indicate the position of the purified recombinant proteins) (FIGURE 7C). Two well-defined bands of a molecular weight around 100 kDa were observed upon Hsp90 pull down. This is due to the fact that Hsp90 functions as a dimer which in the recombinant cell lines is apparently formed by one endogenous (the lower band) and one recombinant (the upper band) monomer (FIGURE 7C). Notably, the amount of Hsp90 pulled down upon heat shock seems to be higher which comes in agreement with the well-documented increased expression of the chaperone upon proteotoxic stress.

CHAPTER 1 RESULTS

Hsp90-bound complexes were eventually identified by Mass Spectrometry. A ratio between the peptide intensity in Hsp90 and RFP pull-down for individual proteins was used as a criterion - only proteins displaying a log2 value higher than two were considered enriched in the Hsp90 pull-down and subjected to further analysis. Four independent biological replicates for each of the conditions and isoforms investigated were analysed.

Figure 7. Nuclear Hsp90 affinity purification. Recombinant Hsp90α and Hsp90β were stably expressed upon doxycycline induction in HEK293 cell lines at levels lower than their endogenous counterparts as demonstrated by a western blot using total cell extracts (TCE). Arrows indicate the position of recombinant proteins (A). The experimental layout involved nuclei isolation, generation of nuclear protein extracts by high-salt extraction, Strep-Tactin affinity purification of the recombinant proteins and Mass Spectrometry (MS) identification of the pulled down interactors (B). Hsp90-bound complexes were visualized by silver staining. Arrows indicate the position of the isolated recombinant proteins.

CHAPTER 1 RESULTS

4.1.1. Nuclear Hsp90α interactions in normal growth conditions

Our replicates of Hsp90α pull-down showed an overlap of around 202 proteins, which were from here on considered as high-confidence interactors (FIGURE 8A). These 202 proteins comprises around 40 % of the proteins identified for every single biological replicate (except for replicate 3) – a relatively good share given the transient and unstable nature of Hsp90 interactions expected to cause a considerable variability between replicates. Cellular Compartment GO Enrichment analysis revealed that proteins comprising the overlap are predominantly nuclear (FIGURE 8B). This pointed to the purity of the nuclear extracts generated and excluded the possibility of cytosolic contamination. This is of critical importance for nuclear Hsp90 investigation given the high abundance of the chaperone in the cellular cytosol. The comparison of this high-confident hit list with a publicly available Hsp90 interactome database maintained by Didier Picard revealed that 19 out of 202 proteins have already been described as Hsp90 clients (FIGURE 8C). The list includes proteins from well-established Hsp90 interactors categories:

• protein folding - Hsp70 and Cdc37 • cell signalling – CSNK2A2, and IRS4 • DNA damage response – XRCC1, PRKDC and Rad50 • RNA processing - snRNP200, PRPF8, PRPF19, SF3B3, MEPCE, PRPF4B • transcription regulation - MTA1 transcription cofactor, HDAC1, HMGA1, TRRAP and POLR2A • protein degradation - CHIP ubiquitin ligase

One should keep in mind that the database combines findings from different cell lines, model organism, protein extraction and purification methods and most importantly – primarily cytosolic or whole-cell extracts investigations. This explains the small overlap between the database and our nuclear Hsp90 interactome. Notably, there are a large number of novel interactions identified by us but not characterized as Hsp90 clients so far. That demonstrates the advantage of the experimental approach we applied towards revealing the interactions of nuclear Hsp90 exclusively. During the course of our investigation a study by Ding et al., 2016 confirmed another interactor identified by us as Hsp90 client – UHRF1 (ubiquitin-like PHD and RING finger domain-containing protein 1) required for the maintenance of DNA methylation [53, 328]. In addition, BCLAF1 was also identified as dependent on the chaperone {Zhou, 2018 #26}. The above-mentioned points suggested high relevance of the

CHAPTER 1 RESULTS

obtained nuclear Hsp90 interactome data and gave us confidence to continue with further analysis.

Figure 8. Nuclear Hsp90α interactome analysis. Venn diagram showing the overlap among proteins found to physically interact with Hsp90α (A). That list is highly enriched in nuclear proteins according to the performed cellular compartment GO enrichment analysis (B). Comparison with the publicly available Hsp90 interactome database (www.picard.ch/Hsp90Int/index.php) reveals a big number of novel Hsp90 clients (C).

The 202 highly-confident Hsp90α interactors form a network with several well-defined clusters (FIGURE 9):

Chaperones and co-chaperones. We were able to show that nuclear Hsp90 interacts with the well-documented co-chaperone partners of its cytoplasmic pool - Hsp70, Cdc37 [6]. Both of the proteins are predominantly cytosolic but a small fraction locates to the nucleus. They are known to hand over different substrates to Hsp90 and as it seems that collaboration remains important for its nuclear pool as well. Hsp70, Cdc37 and serine/threonine-protein phosphatase PP1-alpha catalytic subunit (PPP1CA) [329] were the only known Hsp90 co- chaperone that we identified in the nucleus. However, we noticed the presence of another member of the FK506-binding proteins (FKBPs) family of immunophilins [126] among Hsp90 interactors - FKBP3 (also known as FKBP25). A few members of the family have

CHAPTER 1 RESULTS

been described as Hsp90 co-chaperones – FKBP51, FKBP52, FKBP36 and FKBP38 as described in section 1.2.3.3 of chapter 1. FKBP3 is the first of the immunophilins discovered in the nucleus [330]. Its interaction with DNA and chromatin-related proteins hints to possible roles in chromatin remodelling [330-334]. Based on the above-mentioned points we speculate that FKBP3 is a novel compartment-specific co-chaperones to serve nuclear Hsp90 needs.

Figure 9. Overview of nuclear Hsp90α interaction network. Shown are the 202 proteins that physically interact with nuclear Hsp90α in normal growth conditions. The network was generated using String database [335]. Nodes represent individual proteins and lines indicate interactions between proteins. Nodes colours correspond to the main functional modules of the network shown on the left-hand side.

In an attempt to reveal additional novel Hsp90 co-chaperones for the nuclear compartment, we performed PFAM domain GO annotation analysis looking for proteins possessing TPR or CS domains – characteristic domains shared by many Hsp90 co- chaperones. The analysis did not retrieve any CS domain-containing proteins but detected the following proteins possessing a TPR domain – CSTF3, PRPF6, TRRAP. While TRRAP is a

CHAPTER 1 RESULTS documented client of Hsp90 [222], the functional relationship of the other two factors with Hsp90 remains to be investigated.

Protein Symbol Protein Name Protein Family Protein Location predominantly cytosolic; HSPA8 Heat shock cognate 71 kDa protein Hsp70 family chaperone low nuclear abundance HSPA5 78 kDa glucose-regulated protein Hsp70 family chaperone endoplasmic reticulum predominantly cytosolic; CDC37 Hsp90 co-chaperone Cdc37 co-chaperone low nuclear abundance Serine/threonine-protein phosphatase PPP1CA co-chaperone cytosol and nucleus PP1-alpha catalytic subunit predominantly cytosolic; CCT2 T-complex protein 1 subunit beta chaperonin low nuclear abundance predominantly cytosolic; CCT7 T-complex protein 1 subunit eta chaperonin low nuclear abundance NPM1 Nucleophosmin chaperone predominantly nuclear

NPM3 Nucleoplasmin-3 chaperone cytosol and nucleus

Table 2. Chaperones and co-chaperones identified as potential nuclear Hsp90α interactors.

As discussed previously, a study by Taipale et al., 2014 [59] established the NUDC family of co-chaperones as factors that specifically recognize β-propeller domains such as Kelch (for NUDCD3), WD40 (for NUDC) and RCC1 (for NUDCD2) domains. We were able to detect a considerable number of WD40 proteins in our nuclear Hsp90 interactome dataset - CSTF1, BUB3, WDR82, SMU1, RBBP4/7, CORO1C, PRPF19/4, SF3B3, DDB1, SSBP1, WDR5, NEDD1, DYNC1/2. This implies a collaboration of Hsp90 with NUDC co-chaperone in the nucleus. The inability to pull down NUDC itself is consistent with the challenge to detect Hsp90 co-chaperones in co-precipitation experiments experienced both by us and other researchers in the field. Our preliminary nuclear Hsp90 interactome analysis upon protein crosslinking (data not shown) was able to not only pull down NUDC but a few other known Hsp90 co-chaperones in addition – Hop, Aha1, FKBP4, FKBP5, AH receptor-interacting protein (AIP), protein phosphatase 5 (PP5), and suppressor of G2 allele of SKP1 homolog (SUGT1).

Our interactome data suggests that nuclear Hsp90 interacts with the following chaperones in addition to hsp70 - nucleoplasmin 3 (NPM3), nucleophosmin (NPM1) and two components of the chaperonin TRiC/CCT complex. This does not come as a surprise since Hsp are known to exert their functions in close collaboration with other components of the

CHAPTER 1 RESULTS

cellular folding machinery [336]. However, the functional significance of these interactions so far remains unclear.

The Hsp70 endoplasmic reticulum isoform (HSPA5) detection as part of nuclear Hsp90 interaction network could be attributed to ER co-purification together with nuclei and will not be considered further.

Ubiquitin-proteasome system. Several components of the proteasome were co-purified with nuclear Hsp90 (TABLE 3). This does not come as a surprise given the close collaboration of the chaperone and the protein degradation machinery in order to take care of unfolded proteins [146]. As our data implies, that interplay does not happen in cytosol only, but in the nucleus as well. Furthermore, Hsp90 has been shown to assist 26S proteasome assembly in cytosol [230] and probably does so in the nucleus. It could also facilitate individual components nuclear entry. The E3 ubiquitin ligase CHIP shown to bridge the Hsp90 chaperone and proteasome systems [146] was also detected as interaction partner in the nucleus. In addition, four more E3 ubiquitin ligases were identified – CBLL1, RING1, SRSF3, and PRPF19. Notably, PRPF19 has been recently established as Hsp90 client [59]. Currently, the relationship with the others remains unclear. The ubiquitin fusion degradation protein 1 homolog (UFD1L) detection as part of nuclear Hsp90 interaction network could be attributed to ER co-purification together with nuclei, similar to HSPA5 and will not be considered further.

DNA damage response. We could identify the previously known Hsp90 client XRCC1 in our dataset [66]. It is involved in DNA damage response as described earlier in collaboration with PARP1 and DNA ligase 3. Both proteins are also part of the interaction network probably due to co-purification together with XRCC1. Two additional hits further link Hsp90 to DNA damage response - DNA damage-binding protein 1 (DDB1) and mediator of DNA damage checkpoint protein 1 (MDC1). Overall, these observations re-establish Hsp90 role in DNA damage repair [240].

RNA metabolism. The biggest cluster comprising the nuclear Hsp90 interaction network proved to consist of RNA processing proteins. Therefore, it came as no surprise that the top categories of GO annotation molecular function, biological process and KEGG pathways analysis pointed to important roles of nuclear Hsp90 in RNA metabolism (FIGURE 10). Individual interactors linked nuclear Hsp90 to: • mRNA splicing – SRSF3 spicing factor, RNA helicases DDX17, DDX5 and others

CHAPTER 1 RESULTS

• mRNA stability and transport to cytoplasm - heterogeneous nuclear ribonucleoproteins M/K/R/U, polymerase delta-interacting protein 3, THOC1 and others • mRNA cleavage and polyadenylation - cleavage and polyadenylation specificity factor subunit 1/2/3/7 and others • RNA degradation - exosome complex components RRP44 and RRP45, ZCCHC8

Protein Symbol Protein Name Protein Location

predominantly cytosolic; PSMC1 26S protease regulatory subunit 4 low nuclear abundance predominantly cytosolic; PSMC3 26S protease regulatory subunit 6A low nuclear abundance predominantly cytosolic; PSMC4 26S protease regulatory subunit 6B low nuclear abundance predominantly cytosolic; PSMA3 Proteasome subunit alpha type-3 low nuclear abundance predominantly cytosolic; PSMA7 Proteasome subunit alpha type-7 low nuclear abundance predominantly cytosolic; PSMD11 26S protease regulatory subunit 11 low nuclear abundance predominantly cytosolic; STUB1 E3 ubiquitin- protein ligase CHIP low nuclear abundance

RING1 E3 ubiquitin- protein ligase RING1 cytosol and nucleus

CBLL1 E3 ubiquitin- protein ligase Hakai nucleus

PRPF19 Pre-mRNA-processing factor 19 nucleus

SRSF3 Serine/arginine-rich splicing factor 3 nucleus

UFD1L Ubiquitin fusion degradation protein 1 homolog endoplasmic reticulum

Table 3. Ubiquitin-proteasome system components identified as potential nuclear Hsp90α interactors.

Overall, our observations confirm Hsp90 role in different aspects of RNA metabolism established by numerous studies from the past [203, 204, 254]. The identification of such big number of RNA-processing factors may be due to the fact that they usually operate as part of multi-component complexes that Hsp90 pulls down as a whole. Thus, many of them may appear indirect interactors. It remains to be investigated which of these factors directly interact with the chaperone. Another possibility could be that a client has already bound both

CHAPTER 1 RESULTS

Hsp90 and RNA. Thus, all the additional factors tethered to different sites of the RNA molecule are also pulled down. The use of benzonase (an enzyme that digest all forms of DNA and RNA) in our nuclear protein extract preparation protocol makes this possibility less likely. However, it remains to be investigated whether benzonase treatment was sufficient and which of the RNA processing factors are direct interactors.

Kinases. We identified four kinases as nuclear Hsp90α interactors - PRPF4B, TRRAP, PRKDC and CSNK2A2. They are all part of the serine/threonine protein kinase family and have been previously documented as Hsp90 clients (www.picard.ch/Hsp90Int/index.php). That small number might seem surprising given that kinases are one of the most common Hsp90 interacting protein classes. The inability to detect other kinases may reflect their low abundance in the nucleus and transient nature of interaction with Hsp90.

Figure 10. Nuclear Hsp90α interaction network GO Enrichment Analysis.

Transcription regulation. The second well-defined cluster in our dataset is comprised of factors regulating gene expression - transcription factors/co-factors, chromatin remodellers and chromatin modifiers (FIGURE 8; highlighted in yellow). Several transcription factors/co- factors exerting their effect by either activating or repressing gene expression were identified as novel Hsp90 interactors (TABLE 4), as detailed below.

CHAPTER 1 RESULTS

• General transcription machinery

TAF15 (TATA-binding protein-associated factor 2N) and TAF4 (transcription initiation factor TFIID subunit 4) are both subunits of the transcription factor IID (TFIID), which binds to the core promoter to position the RNA polymerase properly, serves as the scaffold for assembly of the remainder of the transcription complex, and acts as a channel for regulatory signals [337]. Furthermore, two subunits of RNA Pol II – POLR2A and POLR2E, were also identified as Hsp90 interactors. Apart from transcription initiation, nuclear Hsp90α could potentially be linked to elongation and RNA polymerase II pausing [338-340]. In that respect, we were able to demonstrate association with components of the PAF1 (RNA polymerase II- associated factor 1) complex – PAF1, CTR9 and CDC73, as well as FACT (facilitates chromatin transcription) complex – SUPT16H and SSRP1.

Protein Symbol Protein Name Protein Family RAI1 Retinoic acid-induced protein 1 transcription factor TAF15 TATA-binding protein-associated factor 2N transcription cofactor BCLAF1 Bcl-2-associated transcription factor 1 transcription factor POU2F1 POU domain, class 2, transcription factor 1 transcription factor SARNP SAP domain-containing ribonucleoprotein transcription factor MORF4L1 Mortality factor 4-like protein 1 transcription factor DEK Protein DEK transcription factor BPTF Bromodomain PHD finger transcription factor transcription factor NCOR1 Nuclear receptor corepressor 1 transcription cofactor HCF-1 Host cell factor 1 transcription cofactor PQBP1 Polyglutamine-binding protein 1 transcription cofactor

Table 4. Transcription factors/co-factors identified as potential nuclear Hsp90α interactors.

• Transcription factors

DEK - the protein is involved in chromatin organization [341]. A chromosomal aberration results in the formation of a DEK-NUP214 fusion gene is found in a subset of acute myeloid leukemia [341, 342].

BPTF (bromodomain PHD finger transcription factor) - it is highly similar to the largest subunit of the Drosophila NURF (nucleosome remodelling factor) complex. The complex, which catalyses nucleosome sliding on DNA and interacts with sequence-specific transcription factors, is necessary for the chromatin remodelling required for transcription

CHAPTER 1 RESULTS

[343]. It may also regulate transcription through direct binding to DNA or transcription factors [344].

POU2F1 (POU domain class 2 transcription factor 1; known as Oct-1) - Oct-1 is a ubiquitous transcription factor that binds octamer motifs in promotors. It has been found to activate the expression of genes for some snRNAs, H2B and immunoglobulins [345-347]. A study by Tantin et al., 2005 suggests that Oct-1 modulates the activity of genes important for the cellular response to oxidative stress and thus acts as a stress sensor [348]. Thus, Hsp90 potential stabilization effect on Oct-1 could contribute to the proper implementation of the oxidative stress response.

RAI1 (Retinoic acid-induced protein 1) - Rai1 exerts its effect on transcription through the basic transcriptional machinery and chromatin remodelling. It has been described as a transcriptional regulator of circadian clock components [349].

BCLAF1 (Bcl-2-associated transcription factor 1) - BCLAF1 is a poorly studied transcriptional repressor that interacts with several members of the Bcl2 family of proteins [350]. Overexpression of this protein induces apoptosis. The protein localizes to dot-like structures throughout the nucleus, and redistributes to a zone near the nuclear envelope in cells undergoing apoptosis.

NCOR1 (Nuclear receptor co-repressor 1) - NCOR1 is a transcriptional co-factor which assists nuclear receptors in the down regulation of gene expression [351]. It appears to recruit histone deacetylases to DNA promoter regions [352].

MORF4L1 (Mortality factor 4-like protein 1) - MORF4L1 (MRG15) is a component of the NuA4 histone acetyltransferase (HAT) complex which is involved in transcriptional activation by acetylation of nucleosomal histones H4 and H2A. This complex may be required for the activation of transcriptional programs associated with oncogene and proto- oncogene mediated growth induction, tumor suppressor mediated growth arrest, apoptosis, and DNA repair [353, 354]. It has also been found to interact with Sin3A HDAC complex [355, 356].

Sp1, Sp4, Sp8 and Sp9 (Specific protein 1/4/8/9) - our interactome data suggests that Hsp90 may interact with a member of the large multigene family of Sp/Kruppel-like factor (KLF) transcription factors playing important roles in cell cycle regulation and growth control, hormonal activation, apoptosis, and angiogenesis [357, 358]. The high degree of sequence similarity between Sp1, Sp4, Sp8 and Sp9 did not allow discrimination between them since

CHAPTER 1 RESULTS

the MS-identified peptides could potentially belong to any of them. Further investigation is needed to precisely identify the direct interactor.

HCF-1 - HCF-1 is a chromatin-associated scaffold protein [359] originally discovered and widely characterized as a host-cell factor required for herpes simplex virus (HSV) transcription [360-363]. It has a well-established role in cell cycle progression – it binds to certain transcription factor, recruits histone remodelling/modifying complexes and thus modulates chromatin structure in a cell cycle dependent manner [364] (details in chapter 2).

Intriguingly, complex enrichment analysis [365] revealed the following chromatin remodelling/modifying complexes as potential nuclear Hsp90α interactors – Sin3A, Set1, pBAF, NuRD and CoREST (FIGURE 11A). Several observations point to the specificity of these associations: (i) the big number of complexes subunits being pulled down simultaneously (FIGURE 11A), (ii) the absence of interaction with other highly abundant chromatin remodelling/modifying complexes (such as NSL histone acetylation complex) (FIGURE 11B), (iii) the presence of interaction with specific components that define the identity of a complex and distinguish it from its relatives (such as the case of MLL1 versus Set1, and BAF versus pBAF) (FIGURE 11B).

In addition, Hsp90α was also found to interact with Ran – a GTPase involved in nucleocytoplasmic transport, participating both in the import and the export from the nucleus of proteins and RNAs [366]. It switches between a cytoplasmic GDP- and a nuclear GTP- bound state by nucleotide exchange and GTP hydrolysis. Nuclear import receptors such as importin beta bind their substrates only in the absence of GTP-bound Ran and release them upon direct interaction with GTP-bound Ran while export receptors behave in the opposite way. Thereby, Ran controls cargo loading and release by transport receptors in the proper compartment and ensure the directionality of the transport [366]. Hsp90-Ran interaction may illustrate an example of chaperone-mediated client stabilization or could point to Ran- dependent nucleocytoplasmic shuttling of Hsp90. This remains to be investigated further.

4.1.2. Nuclear Hsp90β interactions in normal conditions

Hsp90β pull-down replicates displayed an overlap of around 49 proteins (FIGURE 12A) – a number that surprisingly comprises only 10-25 % of the proteins identified for every single biological replicate. The better results obtained for the parallel Hsp90α pull-down suggest that the poor reproducibility between replicates is not due to experimental issues. A possible explanation could be based on the work of Prince et al., 2015 demonstrating that

CHAPTER 1 RESULTS

Hsp90α binds clients with greater relative interaction strength than does Hsp90β [367]. Thus, certain Hsp90β interactions may have been lost during sample preparation. Stochastic nature of losses may explain the variability between replicates. However, that question was not addressed further in the course of my thesis. We rather focused on those 49 Hsp90β interactors – they were identified across all the four replicates and could be clearly considered as highly confident hits.

Figure 11. Nuclear Hsp90α interacts with several chromatin remodelling/ modifying complexes. Complexes that specifically interact with Hsp90 (A) compared to their relatives whose defining components were not among Hsp90 interactors (B). Red boxes indicate subunits that interacted with Hsp90. Grey boxes indicate subunits that did not interact with Hsp90. Boxes highlighted in yellow indicate subunits that define the identity of a particular complex.

Comparison of our nuclear Hsp90β interaction network with the publicly available Hsp90 interactome database compiled by Didier Picard revealed that most of the comprising proteins have not been described as Hsp90 clients yet (44 hits) (FIGURE 12B). In addition, the overlap re-established 5 of the known interactors - Cdc37 co-chaperone, cyclin-dependent kinases Cdk1 and Cdk11B, transcription regulator HMGA1 and signal transduction regulator YWHAZ. Similar to Hsp90α interaction network, GO Enrichment analysis revealed that the

CHAPTER 1 RESULTS

high-confident Hsp90β interactors are predominantly nuclear and that the top categories of GO annotation molecular function, biological processes and KEGG pathways analysis pointed to important roles of nuclear Hsp90β in RNA metabolism (data not shown). In line with this, the interaction network designed using String Protein Database [335] displayed a well-defined cluster of RNA processing factors (FIGURE 13). Components of the proteasome, co-chaperones, histones and a few transcription regulators complemented the list of nuclear Hsp90β interactors.

Figure 12. Nuclear Hsp90β interactome analysis. Overlap between replicates (A) and with Picard Hsp90 interactome database (B).

Comparison of Hsp90α and Hsp90β interaction networks revealed that all the Hsp90β interactors are common for both the isoforms which clearly establishes these 49 proteins as high-confident nuclear Hsp90 substrates regardless of the isoform investigated. For example, both the isoforms interact with Cdc37 co-chaperone, proteasome components, transcription regulators such as DEK, HMGA1 and PAF1, and multiple RNA processing factors. Our analysis did not identify any Hsp90β-specific substrates. We also cannot comment on Hsp90α–specific interactors with confidence due the issues with Hsp90β interactome data discussed previously - it is likely that future work could significantly expand Hsp90β interactors list and change the status of certain hits regarding isoform specificity.

4.1.3. Nuclear Hsp90α and Hsp90β interactions in stress conditions

In an attempt to shed light on nuclear Hsp90 role in cell survival and recovery upon proteotoxic stress, we exposed HEK293 cells to heat shock (43 °C for 90 min) and performed affinity purification of both the isoforms (FIGURE 7). Strikingly, the four replicates subjected to MS analysis displayed a negligible overlap (FIGURE 14). This could be attributed to both technical and biological reasons. However, the better results obtained for

CHAPTER 1 RESULTS

the parallel Hsp90 pull-down in non-heat shock conditions suggest that the poor reproducibility between replicates is probably not due to experimental issues.

Figure 13. Nuclear Hsp90β interaction network in normal growth conditions. The network was generated using String Protein Database [335]. Nodes represent individual proteins and lines indicate interactions between proteins.

A possible explanation could arise from the following hypothesis:

• Chaperone titration model. The widely accepted chaperone titration model was proposed more than 25 years ago [368- 370]. This is the phenomenon thought to largely contribute to HSF1 regulation in both pro- and eukaryotic organisms [73, 371, 372]. It assumes that upon heat shock chaperones are titrated away from their clients by the increasing number of unfolded/misfolded proteins. Once proteostasis is restored, the interactions could be resumed. Over the years, Hsp90 has been shown to interact with more than 200 substrates. According to the titration model, one could expect that these interactions become even more diverse upon proteotoxic stress due to massive unfolded/misfolded proteins accumulation. In addition, stress response dramatically increases cellular system dynamics. Thus, the notoriously weak Hsp90 interactions could

CHAPTER 1 RESULTS become even more dynamic and transient, and further challenge the acquisition of reproducible interactome data.

• Hsp90 presence in inclusion bodies Hsp90 is known to be involved in the formation of stress granules and P-bodies under proteotoxic stress [373]. Its localization within these inclusion bodies containing aggregated proteins could be another reason for the lack of reproducibility in our interactome study.

Figure 14. Nuclear Hsp90α (A) and Hsp90β (B) affinity purification upon heat shock – experimental replicates overlap.

• Stress-induced weakening of chaperone-client interactions Hsp90 is constitutively phosphorylated on serine and threonine residues [78]. Interestingly, rapid dephosphorylation occurs in response to heat stress [374]. In line with this observation, Mollapour et al., 2010 were able to demonstrate that Hsp90 threonine phosphorylation is lost upon heat shock treatment [100]. Intriguingly, the same study established a connection between the phosphorylation status of Hsp90 and substrates binding - a single conserved tyrosine residue in the N-domain of the chaperone was shown to permit prolonged association with some client proteins. Thus, we could speculate that the heat shock treatment we applied caused dephosphorylation of Hsp90 leading to a generalized decrease in client binding. As a result, interactions of the chaperone become hard to be detected in a reproducible manner.

Notably, the overlap of Hsp90β replicates comprises of two co-chaperones – Cdc37 and Hop. This does not come as a surprise since Hsp90 interacts with co-chaperones much strongly than it does with clients. Interestingly, the general Hop co-chaperone escaped

CHAPTER 1 RESULTS

detection under basal but was pulled down in stress conditions. However, Hop is known to be stress inducible which increases its concentrations and favours detection. The overlap of Hsp90α replicates includes the splicing factor Sugp1 (SURP and G-patch domain-containing protein 1) in addition to Cdc37 and Stip1. It is tempting to speculate that Sugp1 could also be a co-chaperone - a novel one meant to serve Hsp90 nuclear pool. Importantly, a closer inspection of individual replicates revealed well-documented Hsp90 interactors as well as potential clients identified by the affinity purification in normal growth conditions – PCNA, PAF1, UHRF1, SWI-SNF, DEK, CHIP, CBLL1, SIN3A, RAI1, HDADC1, NCOR1, HMGA, H2A, H2B, ENO1 and OCT-1. This shows that despite the poor overlap our data cannot be considered irrelevant. However, at this stage we cannot draw detailed conclusions about Hsp90 interaction network upon stress – at least not before coming up with a strategy to obtain a reproducible interactome data.

In summary, the nuclear Hsp90 interaction network investigated by recombinant protein AP revealed an intriguing overview of potential Hsp90 clients and suggested important roles of the chaperone in critical processes such as transcription regulation, chromatin dynamics and RNA metabolism. However, one should keep in mind that some binding partners might have easily escaped their isolation due to transient interaction with the chaperone. In addition, Hsp90 AP was performed in mild conditions to preserve interactions and thus leading to a higher background. That could have hampered genuine interactors identification and introduced false positives as well. Finally, many proteins interact with Hsp90 indirectly within complexes, making unambiguous determination of direct binding difficult. The points mentioned above justify the need for an independent approach to confirm previous observations and identify novel genuine Hsp90 substrates.

4.2. Nuclear Hsp90 interactome investigated by revealing Hsp90-dependent ubiquitylated proteome

Hsp90 clients are defined as proteins that directly interact with the chaperone and at the same time are destabilized, poly-ubiquitylated and degraded by the proteasome upon Hsp90 inactivation [55, 248-252].The turnover of proteins already destabilized due to Hsp90 inhibition could be halted by inactivation of the proteasome as well (FIGURE 15A). Thus, Hsp90 clients accumulate as ubiquitin-tagged aggregates in the detergent insoluble fraction of cells as demonstrated by Falsone et al., 2007 [254]. They could be further enriched to increase sensitivity for subsequent identification as done by Quadroni et al., 2015 [255]. This provides an interactome investigation approach with a critical advantage - physical

CHAPTER 1 RESULTS

interaction-independent identification of Hsp90 clients. In the light of the very unstable and transient nature of chaperone-client interactions that could be highly valuable.

Figure 15. Investigation of Hsp90-dependent ubiquitylated proteome. Schematic representation of the idea behind the double inhibition strategy for Hsp90 clients identification (A) and the experimental workflow (B). Simultaneous Hsp90 and proteasome inhibition leads to Hsp90 clients accumulation as ubiquitin-tagged aggregates in the detergent insoluble fraction of cells (A). SILAC- labelled cells were treated for 8 hours with Bortezomib (BZ) or NVP-AUY922 and mixed equally. Nuclei were isolated and insoluble nuclear extracts were prepared. After digestion with trypsin and Lys-C, di-Gly-modified peptides were enriched and analysed by Mass Spectrometry (B).

In an attempt to reveal Hsp90-dependent nuclear ubiquitylated proteome we combined features of Falsone et al., 2007 and Quadroni et al., 2015 experimental setups [254, 255] (FIGURE 15B). The analysis was done in HEK cells grown in SILAC media in order to achieve protein labelling and subsequently to obtain quantitative data. We aimed for identification of ubiquitylated proteins from the insoluble nuclear fraction enriched upon Hsp90 and proteasome simultaneous inhibition (double inhibition) compared to proteasome only inhibition (single inhibition) (FIGURE 15B). We took advantage of a newly available antibody-based method and enriched ubiquitylated species making use of the di-glycine group (di-Gly) preserved at sites of lysine ubiquitination upon protein endonuclease digestion [256, 257]. The described di-Gly immunoprecipitation picks up only proteins that have been previously ubiquitylated and could contribute to genuine Hsp90 clients identification upon

CHAPTER 1 RESULTS

double inhibition. In addition, that kind of enrichment reduces the background and boosts protein MS identification. The following inhibitors were used: • Bortezomib (BZ) - an anti-cancer drug and the first therapeutic proteasome inhibitor to be used in humans [375, 376]. The boron atom within BZ catalytically binds the active site of the 26S proteasome with high affinity and specificity, thereby resulting in cell cycle arrest and apoptosis. • NVP-AUY922 (LC Laboratories) – an Hsp90 inhibitor tested for treatment of cancer in phase II clinical trials. It was discovered through a collaboration between The Institute of Cancer Research and the pharmaceutical company Vernalis, and licensed to Novartis [377-380].

In addition to the double inhibition, a set of controls was set up to reduce to minimum false positive and false negative hits consideration (FIGURE 16A):

• DMSO treatment – not expected to alter global protein ubiquitylation • NVP-AUY922 treatment – a negligible effect anticipated since destabilized proteins are rapidly degraded by the proteasome • BZ treatment – a considerable increase of ubiquitylation of all proteins that are rapidly turned over irrespectively of Hsp90 binding is expected

Inhibitors treatment conditions (concentration and duration) were chosen upon inhibitors screen experiments. Both visual inspection and apoptosis/DNA damage markers detection (caspase-3 and γH2AX, respectively) established them as not lethal and not causing serious deleterious effects (FIGURE 16B).

To get a global picture of Hsp90 inactivation effect on protein ubiquitylation in different cellular compartments, HEK cells were treated for 8 hours with either 50nM BZ, 500nM NVP-AUY922 or a combination of both, then fractionated and resulting fractions were run on SDS-PAGE (FIGURE 16B). Ubiquitylation levels were inspected by Western blotting. It is worth mentioning that the anti-ubiquitin antibody used at this point recognized all the possible ubiquitylated species - regardless the number of ubiquitin tags added and amino acid residues targeted (Lys-48, Lys-61, Lys-29, etc.). We were very excited to observe massive increase in ubiquitylation level in the nucleus upon double inhibition compared to single treatments. In contrast, both total cell and cytosolic extracts did not show such dynamics. The same experiment was done in K562 cells to figure out whether this is a cell-type specific effect or a general observation. The obtained results confirmed ubiquitination level changes

CHAPTER 1 RESULTS

in that cell type as well (FIGURE 16C). For all the future experiments we focused on HEK cells only since nuclear Hsp90 interactome was characterized by us in HEK293T cells (section 4.1 of chapter 1). To rule out off-target effects of Hsp90 inhibition we tested another Hsp90 inhibitor (17-AAG) in combination with BZ and could reproduce the observations (FIGURE 16D).

We were also able to demonstrate that ubiquitylation of nuclear proteins under the described experimental condition happens gradually over time. Kinetics experiments revealed that increase could already be detected 4 hours upon double treatment (FIGURE 17B). We then went further investigating whether detected ubiquitylated proteins reside in the soluble or insoluble nuclear protein fractions. In agreement with a previous study [254] we could demonstrate that Hsp90 inactivation destabilizes a large number of proteins which accumulate as ubiquitylated species in the insoluble protein fraction when the proteasome is not functional as well (FIGURE 17C). We could observe PML (promyelocytic leukaemia) protein accumulation in the insoluble fraction demonstrating that nuclear inclusion bodies such as PML bodies copurify with that fraction. As reported earlier by Falsone et al, 2007, Hsp70 presence could also be detected by western blotting [254]. In addition, we were able to see that Akt kinase, a well-known client of Hsp90 and thus a positive control accumulates there (FIGURE 17C).

Nuclear proteins which stability is affected by Hsp90 inhibition could originate from either nucleoplasm or chromatin. Nuclear fractionation experiments revealed an intriguing observation – double inhibition leads to a huge increase in chromatin proteins ubiquitylation implying important role of Hsp90 in chromatin biology (FIGURE 17D). To our surprise, nucleoplasmic proteins behaviour was exactly the opposite – double treatment reduced ubiquitylation levels. This could probably be explained by the rapid aggregation of destabilized proteins upon double inhibition and subsequent aggregates co-purification with chromatin. If so the chromatin fraction is supposed to contain both nucleoplasmic and chromatin proteins affected by Hsp90 inactivation. We could also observe PML protein accumulation in the chromatin fraction suggesting that nucleoplasmic inclusion bodies (such as PML bodies) copurify with chromatin. For the reasons mentioned above samples sent for MS analysis were obtained from TNE with no additional separation into nucleoplasm and chromatin.

CHAPTER 1 RESULTS

Figure 16. Hsp90 and proteasomal inhibition effect on cellular proteins ubiquitylation. Effect of treatment (A) on protein ubiquitylation levels in different cellular compartments of HEK (B) and K562 (C) cells. NVP-AUY922 effect was replicated by using 17-AAG – another Hsp90 inhibitor (D). The arrow indicates an unspecific band recognized by the GAPDH antibody.

CHAPTER 1 RESULTS

Protein poly-ubiquitylation could serve multiple purposes according to the particular lysine (K) residue that every next ubiquitin moiety is attached to form a poly-ubiquitin chain [381]. Ubiquitin has seven lysine residues (K6, K11, K27, K29, K33, K48, and K63), each of which can participate in further ubiquitination. K48 poly-ubiquitin chains regulate protein levels by inducing protein degradation by the proteosome. K63 chains are involved in processes such as endocytic trafficking, inflammation and DNA repair while K11-linked chains are implicated in mitotic regulation and endoplasmic reticulum associated degradation [381]. That is why we considered very important reproducing the observations described above in the context of protein ubiquitylation for degradation. Probing the same samples with anti-ubiquitin K48 antibody confirmed that K48 ubiquitylation contributes significantly to

Figure 17. Hsp90 and proteasomal inhibition effect on nuclear proteins ubiquitylation. Effect on treatment (A) on protein ubiquitylation levels in total nuclear extracts (B), insoluble/soluble nuclear proteins (C) and nucleoplasm/chromatin fractions (D).

CHAPTER 1 RESULTS

total ubiquitylation increase observed previously (FIGURE 18). That strengthened the hypothesis of Hsp90 affecting the stability of a large number of substrates in the nucleus.

Hsp90 and the proteasome have well-established roles in proteostasis and thus cell well- being. The combination of protein destabilization and accumulation of toxic ubiquitylation species within the cell could cause severe stress. To evaluate whether that nuclear ubiquitination increase we observe is mostly attributed to Hsp90 clients modification but not a general secondary effect due to cellular efforts to cope with stress we decided to examine how different cellular stresses affect nuclear ubiquitylation (FIGURE 19A): • Heat shock of 90 min • ER stress - dithiothreitol (DTT) or tunicamycin treatment

• Oxidative stress - H2O2 treatment • Starvation – incubation in serum-free media • DNA damage – camptothecin treatment • Osmotic stress – sorbitol treatment • Proteotoxic stress – incubation in canavanine-containing culture media. This arginine analogue does not allow proper protein folding.

To our satisfaction, diverse stresses were not shown to modulate the same effect as double inhibition (FIGURE 19B). The only exception proves to be the heat shock treatment (cells exposed to 43 °C for 90 min). However, it does not come as a surprise given the global protein misfolding it is known to cause. In addition, we could demonstrate once again that the increase in ubiquitylation in doubly inhibited cells is not due to DNA damage since the DNA damaging reagent camptothecin did not induce ubiquitylation increase (FIGURE 19B).

Proteotoxic stress induces severe destabilization and ubiquitylation of diverse classes of proteins. Newly synthesized polypeptides have been established as one of the most vulnerable ones [382, 383]. That is why it was important to confirm that the ubiquitylation increased we observed is not mostly contributed by newly synthesized proteins but rather an Hsp90-specific effect. We performed double inhibition experiment in the presence of translation inhibitor (cycloheximide) and did not observe any ubiquitylation decrease in comparison to double inhibition only. Transcription inactivation stimulated by actinomycin D did not show an effect as well (FIGURE 19C).

CHAPTER 1 RESULTS

Figure 18. Hsp90 and proteasomal inhibition effect on cellular proteins K48 poly-ubiquitylation Effect on treatment (A) on protein K48 ubiquitylation levels in different cellular compartments (B), insoluble/soluble nuclear proteins (C) and nucleoplasm/chromatin fractions (D).

The degradation of misfolded proteins in different cellular compartments is served by unique degradation pathways. Recent studies, mainly in yeast, have shown that nuclear ubiquitin-proteasome system is particularly active and targets misfolded nuclear as well as certain cytosolic proteins which are transported to the nucleus prior to their degradation [384, 385]. To test whether ubiquitylated cytosolic proteins contribute to nuclear ubiquitylation increase we inhibited pharmacologically importin α/β-mediated nuclear import [386] in addition to Hsp90 and proteasome double inactivation. Obtained results demonstrate that ubiquitylated cytosolic proteins do not enter the nucleus upon double inhibition and thus do not contribute to nuclear ubiquitylation increase (FIGURE 19D).

After exploring the global effect of Hsp90 inhibition on nuclear proteins ubiquitylation we went further to identify the proteins that get ubiquitylated upon Hsp90 and proteasome

CHAPTER 1 RESULTS inhibition but not upon proteasome inactivation only, and thus establish some of them as clients of the chaperone. For that purpose, HEK cells were grown in SILAC media prior to treatment to allow quantitative MS protein identification – heavy-isotope labelling for double inhibition and light-isotope labelling for BZ treatment (FIGURE 17B). BZ treatment of cells started 30 min before the inhibition of Hsp90 with NVP-AUY922, to ensure sufficient proteasomal inactivation. Combined inhibitors were incubated for additional 8 hours. Nuclei were then isolated and the insoluble nuclear protein fraction obtained upon 8 hours of

Figure 19. Different stresses effect on nuclear proteins K48 poly-ubiquitylation. Effect of treatment (A) on protein ubiquitylation levels in total nuclear extracts. treatment was analysed – proteins were digested, ubiquitinated peptides were enriched by di- Gly IP and subjected to MS analysis. To our surprise, only two proteins were identified – polyadenylate-binding protein 4 (PABPC4) and histone H2B, both enriched in doubly

CHAPTER 1 RESULTS

inhibited cells as compared to singly inhibited ones. That probably reflects potential issues of the protocol applied – not sufficient starting material resulting in poor enrichment, C18 columns peptide purification causing considerable loss, unsatisfactory di-Gly antibody performance, ubiquitin tag enzymatic removal during experimental procedure, difficulties in detecting di-Gly peptides etc. To address these issues we decided to first go back to insoluble nuclear extracts to reveal the identity of comprising proteins and the abundance of ubiquitylated species without any enrichment steps. Insoluble nuclear extracts obtained from SILAC labelled cells upon 4 hours of inhibitors treatment were subjected to MS analysis, and 558 proteins in total were identified. MS-identified peptides intensity was used to quantitative analyse the data. A ratio between heavy-media and light-media peptide intensity was

calculated. Log2 of the ratio was established as a cut off – only proteins displaying a log2 value higher than two were considered enriched upon double treatment. To our surprise, only two out of 558 proteins detected were found enriched in the double treatment sample - PABPC4 and histone H2B. These proteins were the same as the ones detected in the initial di-Gly immunoprecipitation experiment described above. The poorly studied PABPC4 is a cytosolic protein involved in cytoplasmic regulatory processes of mRNA metabolism [387] and thus probably appears as a cytosolic contaminant. Nevertheless, it was exciting to discover that both of the enriched proteins appear Hsp90 interactors according to Hsp90 affinity purification done earlier (section 4.1 of chapter 1). However, no di-Gly peptides of any of these 558 proteins were detected. This could be attributed to insufficient amount of starting material resulting in very low di-Gly peptides abundance and thus problematic detection. In an attempt to enhance ubiquitylated and thus differentially occurring proteins abundance, we went further increasing the starting material for the experiment and prolonging treatment to 8 hours. As expected, the following MS analysis detected a considerably higher number of insoluble proteins (1715) compared to the 4-hour treatment with a significant overlap (523) (FIGURE 20A). Of the 1715 proteins, 84 were enriched upon simultaneous inhibition of Hsp90 and the proteasome. Given the selectivity of NVP-AUY922 for Hsp90-inhibition we assumed that most of these 84 proteins are genuine Hsp90 substrates. They were considered as the most promising hits and thus further characterized. We used String Protein Database [335] to design their interaction network and detected two well- defined clusters which. KEGG Pathways GO annotation analysis revealed as proteasome- (highlighted in green) and metabolic pathways-related (highlighted in yellow). The observation came in agreement with previous studies employing Hsp90 inhibition strategies [254, 255]. Biological Process GO annotation analysis demonstrated enrichment of categories

CHAPTER 1 RESULTS

such as protein poly-ubiquitylation, DNA damage and mitotic cells cycle – processes for which Hsp90 involvement has been already described [240, 388].

To link our data with Hsp90 interactome studies done so far, we compared the highly promising hit list of 84 proteins enriched upon simultaneous inhibition of Hsp90 and the proteasome to Hsp90 interactome dataset generated by us (section 4.1 of chapter 1) or the one compiled by Didier Picard (www.picard.ch/Hsp90Int/index.php). To our surprise, there was a negligible overlap with our Hsp90 affinity purification derived interactome data - 26S proteasome non-ATPase regulatory subunit 3 (PSMC3) and 7SK snRNA methylphosphate capping enzyme (MEPCE). The overlap with Picard´s database revealed that 10 out of 84 enriched proteins have already been described as Hsp90 clients (TABLE 5). The number is low but one should keep in mind the following important notes: • Picard´s dataset combines findings from different cell lines, model organism, protein extraction and purification methods and most importantly – primarily cytosolic or whole- cell extracts investigations. On the other hand, what we are trying to identify are nuclear clients, which have been still poorly characterized. Therefore, a significant overlap is less likely to occur.

Figure 20. SILAC-based MS identification of insoluble nuclear proteins upon Hsp90 and proteasomal inhibition. Overlap between proteins identified in the insoluble nuclear fraction upon

CHAPTER 1 RESULTS

4h- and 8h-treatment with BZ and NVP-AUY922 (A) and 8h-treatment hits interaction network (B)[389].

• Some Hsp90 clients might have escaped ubiquitylation. They may have co-aggregated with other proteins or because of a progressive breakdown of the ubiquitin conjugating machinery. That raises a hypothesis that among the non- ubiquitylated proteins there might indeed be genuine Hsp90-clients. Consistent with this assumption, Falsone et al, 2007 demonstrated accumulation of non-ubiquitylated Cdc2 and CDK4 in the insoluble protein fraction [254]. Both of the proteins are well-established Hsp90 clients [248, 390]. We ourselves also observed non-ubiquitylated Akt kinase (Hsp90 client) accumulation in the insoluble fraction (FIGURE 17). • Some Hsp90 clients might still have not aggregated and thus remained in the soluble protein fraction. • Non-proteasome dependent degradation such as autophagy cannot be excluded [391, 392] • Possible de-ubiquitylation of clients occurring within the cell or during sample preparation.

Protein Symbol Protein Name Protein Location IKBKG NF-kappa-B essential modulator cytosol IKBKB Inhibitor of nuclear factor kappa-B kinase subunit beta cytosol PSAT1 Phosphoserine aminotransferase cytosol MTHFD1 C-1-tetrahydrofolate synthase cytosol LDHA L-lactate dehydrogenase A chain cytosol CDK6 Cyclin-dependent kinase 6 cytosol and nucleus N1PKN1 Serine/threonine-protein kinase N1 cytosol and nucleus UBA1 Ubiquitin- like modifier- activating enzyme 1 nucleus PFAS Phosphoribosylformylglycinamidine synthase nucleus

MEPCE 7SK snRNA methylphosphate capping enzyme nucleus Table 5. Previously known Hsp90 clients identified by mass spectrometry in the insoluble protein fraction of HEK cells upon 8-hour simultaneous Hsp90 and proteasome inhibition.

Subcellular localization assessment of the proteins comprising the overlap with Picard´s database revealed a very critical observation – half of the proteins are cytosolic according to the Human Protein Atlas (a platform developed with the aim to map all the human proteins in cells, tissues and organs; https://www.proteinatlas.org) (Table 5). That encouraged us to go back to the 4- and 8-hour treatment protein datasets and perform cellular compartment GO enrichment analysis. Surprisingly, a significant number of proteins are cytosolic (Fig. 21A).

CHAPTER 1 RESULTS

That probably points to cytosolic contamination of obtained insoluble nuclear protein extracts. To examine this possibility HEK cells subjected to BZ only or double treatment were immunostained to reveal ubiquitylated proteins localization within the cell. Interestingly, apart from ubiquitin distribution throughout the cell very intense signal originated from a cytoplasmic inclusion body likely attached to the nucleus (FIGURE 21B). Literature survey suggested that the observed formation could be what has been previously described as “aggresome” [393, 394] (FIGURE 21C).

Figure 21. Aggresome formation upon simultaneous Hsp90 and proteasome inhibition. Cellular component GO enrichment analysis of components identified upon 4 and 8 hours of treatment (A). Ubiquitin and DAPI staining of HEK cells isolated nuclei (B). Aggresome formation and localization

CHAPTER 1 RESULTS

(C) (image from [395]). CHIP-GFP expressing HEK cells and isolated nuclei (D). Arrows in (B) and (D) point to aggresomes.

The aggresome is an inclusion body shown to occur as a defence mechanism upon massive protein misfolding and aggregation. It allows ubiquitinated/aggregated proteins sequestration away from the still functional cellular machineries and thus reduces their cytotoxic effect [393, 394, 396-399]. It appeared that single inhibition of the proteasome or Hsp90 are known to favour the formation of aggresomes as an alternative form of protein disposal [400-403]. This does not come as a surprise given the induced release of unstable proteins by Hsp90 inhibition and simultaneously the great challenge a cell is facing upon proteasome inactivation. Consistent with these observations our anti-ubiquitin immunofluorescence in HEK cell detected aggresomes in both of the experimental conditions. Furthermore, we found several constituents of aggresomes in our detergent insoluble preparations - sequestrosome-1 (SQSTM1), centrosomal protein 72 (CEP72), cytoplasmic dynein 2 (DYNC2H1), cytoplasmic dynein 1 (DYNC1H1) and vimentin. Vimentin forms the shell of aggresomes [393] and as shown earlier it co-purifies with nuclei and ends up in both insoluble nuclear extract and chromatin (FIGURE 17B). Finally, we detected a considerable amount of 26S proteasome subunits which are known to constitute a notable fraction of the aggresomes [404]. These observations could explain the observed cytoplasmic contamination in our insoluble nuclear extracts. We hypothesized that aggresome remains attached to the nucleus during nuclei isolation. To explore this possibility we took advantage of a recombinant HEK cell line designed to express GFP-tagged CHIP ubiquitin ligase (previously generated in the lab by Dr. Fernando Aprile-Garcia). CHIP is a protein previously shown to accumulate in aggresomes [405]. Thus, we used it to track down the inclusion body localization upon sample preparation. Cells were either subjected to simultaneous Hsp90 and proteasome inactivation or DMSO treatment as a control. We could see that aggresome forms upon double inhibition, remains attached to the nucleus upon its isolation and could also be observed in nucleoplasmic protein extracts (FIGURE 21D). Our efforts to isolate aggresome-free nuclei by modifications in the protocol (different cell lysis buffers, different detergents, different types of mechanical force applied to isolate nuclei) failed. In conclusion, aggresomal proteins definitely end up in total- and insoluble nuclear extracts and chromatin fractions. To our regret, that means a massive contamination with potential cytosolic Hsp90 clients (within the aggresome). This will most probably severely hamper the detection of the nuclear Hsp90 clients and make data interpretation complicated.

CHAPTER 1 RESULTS

The cytosolic contamination in our Hsp90-dependent ubiquitylated proteome experiments raised a critical concern: Could the described approach be used at all to identify nuclear clients of Hsp90? In the light of the recent findings the answer is no - not at least before finding a way to obtain aggresome-free nuclei. In addition, efforts will be needed to optimize di-Gly peptides enrichment and identification. However, the protein dataset obtained after 8 hours of treatment could still prove useful. Interpretation does not promise to be straightforward but a careful inspection detects nuclear proteins (TABLE 6) as potential novel Hsp90 clients. Notably, a member of the leucine-rich repeat (LRR) family of proteins - LRRC40, proves a promising lead - (i) it has been shown to interact with the SGT1 co- chaperone of Hsp90 [59] and (ii) Hsp90 is able to associate with the characteristic LRR domain [59]. Thus, it could well be that SGT1 co-chaperone associates with LRRC40 and transfers it to Hsp90 for conformational stabilization. During the course of the current study LRRC40 dependency of the chaperone was confirmed by Savitski et al., 2018 in both Jurkat and MCF-7 cells [406]. In addition, we could detect two transcription factors as potential novel Hsp90 clients - TCF20 and ZNF581, as well as factors that could complement Hsp90 role in replication - DNA replication licensing factor MCM4, PAF1 complex component RTF1 and DNA replication initiation player ORC-2 (TABLE 6).

Protein Symbol Protein Name Classification RTF1 RNA polymerase-associated protein RTF1 homolog Transcription ZNF581 Zinc finger protein 581 Transcription TCF20 Transcription factor 20 Transcription MCM4 DNA replication licensing factor MCM4 Replication ORC2 Origin recognition complex subunit 2 Replication TRMT112 Multifunctional methyltransferase subunit TRM112-like protein tRNA modification FYTTD1 UAP56-interacting factor mRNA export LRRC40 Leucine-rich repeat-containing protein 40 Unknown Table 6. Nuclear proteins identified upon 8h-treatment.

Although quite cumbersome the described approach still possesses clear advantages for Hsp90 clients identification. It offers physical interaction-independent identification of Hsp90 clients. In the light of the very unstable and transient nature of chaperone-client interactions that could be highly valuable. Our analysis could provide the first picture of the reorganization in the nuclear ubiquitinome that follows Hsp90 inhibition. The accomplished

CHAPTER 1 RESULTS

so far could be used as a starting point for future experiments in the lab. To achieve an optimal performance one should work on the following aspects:

• Obtaining aggresome-free nuclei • Sufficient preservation of native ubiquitylation In addition to cysteine protease inhibitors, such as iodoacetamide (IAA) and N- ethylmaleimide (NEM) used so far one could also add PR-619 - a cell permeable broad spectrum deubiquitylating enzymes (DUBs) inhibitor inducing the accumulation of polyubiquitylated proteins in cells without directly affecting proteasome activity. One could take advantage of the well-established PTMScan® Ubiquitin Remnant Motif (K-ε-GG) Kit #5562 offered by Cell Signaling Technology to isolate, identify, and quantitate di-Gly peptides with a high degree of specificity and sensitivity [257, 407].

• Effective clients destabilization Clients strength of binding to- and dependence on Hsp90 has been shown to vary significantly [68]. It is thus likely that at higher inhibitor concentrations many more clients would be affected.

In summary, nuclear Hsp90 interactome analysis that we performed (described in sections 4.1 and 4.2 of chapter 1) revealed a complex network of nuclear proteins that physically interact with the chaperone – transcription (co)factors, chromatin modifying/remodelling complexes, kinases, protein degradation factors, other molecular chaperones and RNA processing proteins. Thus, our data implies Hsp90 involvement in gene expression regulation at different levels, nuclear proteins turnover and RNA metabolism. To strengthen these findings and obtain further insight into the function of nuclear Hsp90, the applied standard interactome mapping affinity purification technique should be complemented/ supported by an approach to prove functional connection between the chaperone and its potential nuclear clients. The so called synthetic lethality approach, for example involves simultaneous inactivation of two factors (by means of gene knockdown/knockout/mutation or protein inactivation) [408] - if a deficiency in only one of these factors does not lead to cell/ organismal death but the combination does, that speaks for a functional interaction/ relationship between these two factors. Several studies applied a combination of physical and functional interactors investigation with a focus on either the whole-cell or the cytosolic Hsp90 pool mainly in yeast [138, 241].

CHAPTER 1 RESULTS

4.3. Hsp90 functional interactome investigation

To obtain further insight into the function of nuclear Hsp90 we investigated its functional interactors utilizing the above-described synthetic lethality approach in collaboration with the laboratory of Cornelius Miething (UniKlinik, Freiburg). We particularly focused on transcriptional regulators – around 700 proteins able to affect gene expression (transcription factors/co-factors and chromatin remodelers/modifiers) were simultaneous inactivated (by shRNA) together with Hsp90 (by pharmacological inhibition) in K562 cells. A total of 161 factors proved vital for cell growth and viability under conditions where Hsp90 function is compromised thus reporting a functional interaction. In order to establish a highly confident list of factors interplaying at both physical and functional levels with the chaperone we compared these functional interactors with the nuclear Hsp90 interactome dataset generated by us (section 4.1 of chapter 1). The overlapping revealed 13 common hits related to transcriptional regulation, protein degradation and RNA metabolism (TABLE 7). Notably, our physical and functional Hsp90 interactome data confirmed the E3 ubiquitin ligase UHRF1 relationship with the chaperone. The presence of such a well-established client of Hsp90 [53] in the 13-hit list demonstrates that the factors for which we demonstrated both physical and functional connection with the chaperone are most likely novel genuine Hsp90 clients. In addition, the shRNA screen results strengthened our suggestion that pBAF chromatin remodeling and SET1A histone methylation complexes are novel clients of the chaperone (section 4.1 of chapter 1). This is due to the findings that PBRM1 (protein polybromoo-1; pBAF complex component) and WDR82 (SET1A complex component) were also among the functional interactors (TABLE 7).

Intriguingly, a closer inspection of Hsp90 physical and functional interactors datasets accompanied by a detailed literature survey revealed an exciting observation. Our data demonstrates that the transcriptional co-factor HCF-1 (host cell factor 1) interacts both physically and functionally with Hsp90. In addition, HCF-1 has been previously shown to interact with and recruit SET1A histone methylation and SIN3A histone deacetylation complexes to chromatin [364]. Notably, we demonstrated that the same complexes are part of nuclear Hsp90 interaction network (FIGURE 11). A look at the HCF-1-related literature published so far revealed that in a study by Wysocka et al., 2003 HCF-1 was shown to copurify not only with components of the aforementioned complexes but Hsp90 and Hsp70 as well [364]. However, this connection was not a focus of their work and was not investigated.

CHAPTER 1 RESULTS

Overall, these findings point to intriguing collaboration of Hsp90 and HCF-1 possibly at chromatin level. This lead was further followed and described in chapter 2 of the thesis.

Protein Symbol Protein Name Protein Family HCF-1 Host cell factor 1 Transcription regulation WDR82 WD repeat-containing protein 82 Transcription regulation HDAC2 Histone deacetylase 2 Transcription regulation PBRM1 Protein polybromo-1 Transcription regulation UHRF1 E3 ubiquitin- protein ligase UHRF1 Transcription regulation DMAP1 DNA methyltransferase 1-associated protein 1 Transcription regulation BPTF Nucleosome-remodeling factor subunit BPTF Transcription regulation MBD2 Methyl-CpG-binding domain protein 2 Transcription regulation FUBP3 Far upstream element-binding protein 3 Transcription regulation NONO Non-POU domain-containing octamer-binding protein Transcription regulation SET Protein SET Transcription regulation SRRT Serrate RNA effector molecule homolog RNA metabolism EXOSC10 Exosome component 10 RNA metabolism Table 7. Overlap between Hsp90 physical and functional interactors.

4.4. Hsp90 PTMs

To reveal nuclear Hsp90 PTMs recombinant Hsp90α and Hsp90β isoforms were stably expressed in T-REx™-293 cells (FIGURE 7). Resulting HA-Strep-tagged proteins were eventually enriched from nuclear protein extracts via Strep-Tactin affinity purification and subjected to MS analysis for PTMs identification. To investigate whether proteotoxic stress alters chaperone modifications and how these changes could contribute to stress response two experimental conditions were analysed : normal growth control (CTL; cells grown at 37°C) and heat shock treatment (HS; cells incubated at 43 °C for 90 min). Three independent biological replicates for each Hsp90 isoform and experimental conditions were analysed and compared. MS analysis revealed that nuclear Hsp90 is modified at several sites – phosphorylated, methylated and acetylated residues were identified (FIGURE 22). All the phosphorylated and acetylated residue detected in normal growth conditions have already been reported in the past for cytosolic Hsp90 (distinguished by check marks on FIGURE 22). Notably, we discovered three novel methylation sites. In addition, there is no data on Hsp90 PTMs upon proteotoxic stress so we consider all the findings in this condition as novel ones.

In our search for nuclear Hsp90 phosphorylation sites three serine residues (Ser) within the charged linker were found modified (FIGURE 22). Ser-255 phosphorylation was detected

CHAPTER 1 RESULTS

for Hsp90β, whereas Ser-231 and Ser-263 – for Hsp90α. These modifications were already reported for cytosolic Hsp90 by Lees-Miller and Anderson in 1989 [409]. We confirmed them in both CTL and HS conditions but with different intensities. Comparison of peptides intensity showed that Hsp90α Ser-231 was slightly less abundant than Ser-263 in CTL condition and both increased in response to stress (FIGURE 22). In contrast, Hsp90β Ser-255 decreased upon heat shock.

Figure 22. Post-translational modification sites of nuclear Hsp90α and Hsp90β in normal growth and stress conditions. Phosphorylated serine (S) residues (orange), acetylated lysine (K) residues (green) and methylated lysine (K) or arginine (R) residues (blue) with the corresponding domain localization are shown (N – N-terminal domain, M – middle domain, C – C-terminal domain). The check mark denotes modification sites that have already been described in literature.

Our data revealed that nuclear Hsp90 is not only phosphorylated but subjected to acetylation and methylation as well (FIGURE 22). Lys-559 of Hsp90β was found acetylated in both Ctl and HS conditions with no change in abundance under proteotoxic stress. No acetylation of Hsp90α was detected. Lys-550 of Hsp90β was found methylated in both CTL and HS conditions with no change in abundance, whereas Lys-284 and Arg-502 appeared CTL and HS specific, respectively (FIGURE 22). Surprisingly, methylation of residues other than lysine and arginine (glutamate, glutamine and aspartate) were detected. This could probably be explained as an artefact introduced in-vitro during sample preparation [410, 411]. It remains to be investigated whether the identified methylation sites are genuine or false positive modification sites.

CHAPTER 1 RESULTS

4.5. Summary

The systematic analysis of nuclear Hsp90 that we performed included characterization of its physical interactome (employing different approache), its functional interactors and PTMs. Thus, we provided a further step in our understanding of the Hsp90 chaperone system in the nucleus.

Nuclear Hsp90 interactome analysis in normal growth conditions revealed a complex network of interactors falling into the following main categories: (i) chaperones and co- chaperones, (ii) components of the protein degradation machinery, (iii) DNA repair proteins, (iv) RNA metabolism factors and (v) transcription regulators. It re-established the previously known role of the chaperone in gene expression regulation, DNA repair and RNA metabolism. The analysis confirmed well-documented collaboration with other chaperones/co-chaperones (such as Hsp70, Cdc37 and PP1) and the protein degradation machinery (proteasome components and ubiquitin ligases). More importantly, we identified a considerable number of potential novel clients that particularly link Hsp90 to transcription regulation at different levels. The chaperone interacts with POLR2A and POLR2E subunits of RNA Pol II, components of the basic transcriptional machinery (such as TAF4 and TAF15), transcription elongation factors (components of PAF1 and FACT complexes), chromatin remodeling/modifying complexes ( pBAF, Set1, Sin3A, CoREST, NuRD) and transcription factors/co-factors (HCF-1, Oct1, BCLAF1, DEK, etc.). Several of these physical interactors appear to also functionally interact with the chaperone according to the synthetic lethality screen that we performed (such as HCF-1, HDAC2, WDR82, PBRM1, BPTF, etc).

Nuclear Hsp90 interactome analysis upon proteotoxic stress revealed that most probably Hsp90 turns into a less specialized chaperone in response to proteotoxic stress. It generally binds misfolded proteins that massively accumulate in these conditions in addition to its distinguished clients. This could be a transformation mechanism to ensure cell survival and recovery.

We discovered that nuclear Hsp90 is modified at several sites – phosphorylated, methylated and acetylated residues were identified (FIGURE 22). We re-established previously known phosphorylated (S231 and S263 for Hsp90α; S255 for Hsp90β) and acetylated (K559 for Hsp90β) cites. Notably, we identified three novel methylation sites for Hsp90β (R503, K284, K550) with R503 modified only upon stress conditions.

CHAPTER 1 RESULTS

Overall, our Hsp90 interactome and PTMs data provide a resource for future mechanistic and functional studies of nuclear Hsp90 as suggested in the following “Discussion and Outlook” section. As proof of concept, we have followed one interactor to link nuclear Hsp90 with gene regulation (chapter 2).

CHAPTER 1 DISCUSSION AND OUTLOOK

5. Discussion and Outlook More than three decades of study revealed the central importance of Hsp90 in numerous cellular processes and carcinogenesis. Nonetheless, despite its crucial roles in biology and medicine, the complexity of the system has left many aspects of its function, regulation and structure uncovered. In addition, Hsp90 is traditionally viewed as a cytoplasmic chaperone but cumulative evidences suggest nuclear localization and important roles in nuclear events. The present work sheds more light on the poorly-studied Hsp90 nuclear pool by characterizing its interaction partners and PTMs.

5.1. Nuclear Hsp90 interaction network

In the present work we investigated the interaction partners of Hsp90α and Hsp90β – the chaperone isoforms present in the nucleus. Hsp90β interaction network appeared not only less complex but actually a subset of Hsp90α network. That is the reason why in the following discussion I will be referring to Hsp90α interaction partners only. It remains to be confirmed by an additional approach (Hsp90α and Hsp90β immunoprecipitation, for example) that nuclear Hsp90β interaction network is indeed not as complex as the one of Hsp90α.

We were able to identify a considerable number of novel Hsp90 clients and to demonstrate that nuclear Hsp90 forms a complex network of interactors falling into the following main categories: (i) chaperones and co-chaperones, (ii) components of the protein degradation machinery, (iii) DNA repair proteins, (iv) RNA metabolism factors and (v) transcription regulators.

5.1.1. Collaboration with nuclear protein folding and degradation machineries

Molecular chaperones and the ubiquitin-proteasome system (UPS) act together in protein homeostasis maintenance - a collaboration that ensures protein folding, trafficking and degradation according to cellular demands. This established protein quality control (PQC) system is particularly crucial in the cytoplasm. Being the primary site of protein synthesis, the cytoplasm faces a high burden of nascent proteins that must be folded immediately after translation or rapidly degraded in case of fatal misfolding. Although the existence and functional purpose for nuclear translation remains controversial [412, 413], the nucleus possesses its own distinct PQC system [414-417] to manage misfolded proteins as they arise locally - chaperones to unfold/refold a misfolded protein or direct it to the UPS for degradation in case rescue is not possible. In the present work we were able to position the

CHAPTER 1 DISCUSSION AND OUTLOOK

molecular chaperone Hsp90 as part of the nuclear PQC network by revealing its interaction with components of the protein folding and degradation machineries.

According to our data, Hsp90 interacts with several other chaperones (Hsp70, NPM1 and NPM3), co-chaperones (Cdc37 and PPP1CA) and the chaperonin TRiC/CCT in the nucleus. Hsp70 and Cdc37 are well-established cytosolic partners of Hsp90 [6, 121]. The conserved nature of these interactions reflects the crucial role of both the factors in the Hsp90 chaperone machinery regardless of the cellular compartment. They are known to recruit substrates to Hsp90 and obviously this relation remains important in the nucleus as well. Given the specialized role of Cdc37 in handing over kinases to the chaperone one could expect that many nucleus-resident kinases are dependent on Hsp90. We could identify four serine/threonine protein kinases - PRPF4B, TRRAP, PRKDC and CSNK2A2, as nuclear Hsp90 interactors. They have been previously documented as Hsp90 clients (www.picard.ch/Hsp90Int/index.php). The inability to detect other kinases may reflect their low abundance in the nucleus and transient/unstable interaction with the chaperone. Kinobeads [418] may prove a useful tool to boost the identification of Hsp90-interacting kinases despite their low abundance. The technology features kinase inhibitors covalently attached to Sepharose beads for affinity enrichment of kinases. Thus, nuclear kinome enrichment followed by Hsp90 purification will recover kinases that interact with the chaperone. Furthermore, the same strategy applied upon Cdc37 inactivation may show whether this co-chaperone is indeed the one handing over kinases to Hsp90 in the nucleus.

Our interactome data suggests that in addition to Hsp70, nuclear Hsp90 interacts with the following chaperones - NPM1, NPM3 and TRiC/CCT complex. It could well be that TRiC/CCT hands over substrates for stabilization to Hsp90. Thus, the chaperone may support TRiC/CCT role in complex assembly, RNA processing or STAT3 activation [419-421]. In the same fashion one could speculate that Hsp90 collaborates with NPM1 and NPM3 in histones chaperoning and DNA repair [422].

Hsp90 is known to act as part of a complex machinery together with its regulatory partners known as co-chaperones [111]. Notably, Cdc37 together with serine/threonine-protein phosphatase PP1-alpha catalytic subunit (PPP1CA) [329] were the only known Hsp90 co- chaperones that we identified in the nucleus. The inability to detect additional co-chaperones such as p23, Aha1, RPAP3, etc. in our recombinant Hsp90 pull-down experiment may be caused by the introduced Strep-tag® which could hamper corresponding interactions. In addition, the transient nature of the Hsp90/co-chaperone complexes combined with the

CHAPTER 1 DISCUSSION AND OUTLOOK

expected low nuclear abundance of the latter proteins represents another challenge to overcome. Furthermore, many traditionally cytosolic co-chaperones may not be present in the nucleus. Instead, there might be compartment-specific co-chaperones to serve nuclear Hsp90 needs similar to Mzb1 (marginal zone B and B1 cell specific protein) and Cnpy3 (canopy FGF signalling regulator 3) co-chaperones for Grp94 in the ER [165, 166]. In that respect, we identified a novel interaction partner that belongs to the FK506-binding proteins (FKBPs) family of immunophilins [126] called FKBP3. A few members of the family have already been characterized as Hsp90 co-chaperones [126]. This makes us suggest that FKBP3 could not only be a co-chaperone but also the first co-factor that specifically serves Hsp90 nuclear pool and possibly supports its role in chromatin remodelling [330-334]. To prove this, we first need to show in a reciprocal pull-down experiment that FKBP3 binds Hsp90. Second, we should demonstrate that Hsp90i does not affect FKBP3 stability – a key feature to distinguish Hsp90 clients from co-chaperones. Then, an in vitro protein refolding assay (luciferase or β- galactosidase refolding assay, for example [423]) in the presence or absence of FKPB3 could demonstrate whether Hsp90 needs that factor to successfully fold a client. In addition, Hsp90 interaction network could be investigated upon FKBP3 inactivation to test whether known clients are still able to bind the chaperone. FKBP3 interaction partners could be revealed as well to figure out whether this factor communicates with components of the Hsp90 chaperone machinery and known clients. That could be an intriguing follow-up to reveal the first nucleus-specific co-chaperone of Hsp90. The experiments mentioned above could be successfully applied to not only FKBP3 but additional co-chaperone candidates such as the ones shown to interact with Hsp90 in the present work and to possess a TPR domain - CSTF3 and PRPF6.

In line with the close collaboration between chaperones and the UPS in proteostasis maintenance, we showed that Hsp90 interacts with several components of the UPS in the nucleus. That includes both proteasome components and E3 ubiquitin protein ligases (TABLE 3). That interplay may contribute to a quick adjustment of gene expression - together with the protein degradation machinery Hsp90 could affect the stability of gene expression regulators at chromatin level. Thus, according to cellular demands a TF could be either stabilized by Hsp90 or handed over to the UPS for ubiquitylation and subsequent regulated protein degradation [309, 310]. This mechanism allows rapid changes in transcription which is particularly critical during response to environmental stimuli.

CHAPTER 1 DISCUSSION AND OUTLOOK

Hsp90 has been shown to assist 26S proteasome assembly in cytosol [230] and probably does so in the nucleus. It could also facilitate individual components nuclear entry. It might be of particular interest to investigate the relationship between Hsp90 and CBLL1, RING1 and SRSF3 given that around 30 % of human E3 ubiquitin protein ligases are stabilized by the chaperone [59]. It is possible that Hsp90 promotes CBLL1 and RING1 stability/activity by maintaining the conformation of their common RING domain. Interestingly, in addition to CBLL1 and RING1 several more RING domain-containing proteins co-purified with Hsp90 implying a chance of Hsp90 chaperoning that particular domain. Regardless of the corresponding mechanism, CBLL1 for example may provide a link of Hsp90 to RNA splicing and cell cycle regulation being part of Wilms' tumor 1-associating protein (WTAP) complex [424, 425]. Notably, a second member of that complex co-purified with nuclear Hsp90 – BCLAF1. Acting as an oncogene [426], CBLL1 could provide an opportunity for anti-cancer activity of Hsp90 pharmacological inhibitors. RING1, on the other hand, as part of the Polycomb group (PcG) multiprotein PRC1-like complex could involve the chaperone in epigenetic transcriptional repression.

Overall, the present work shed light on Hsp90 partnership with components of the PQC system in the nucleus – chaperones, co-chaperones and UPS components. We would like to study further the complex and interconnected network that ensure protein homeostasis in the nuclear compartment. For that purpose, we cloned additional chaperones and co-chaperones (Hsp70, Cdc37, RPAP3 and p23) in the same Flp-In HEK293 T-REx expression system [319] as we did for Hsp90 in order to study their nuclear interactomes in future. That would characterize in detail the protein quality control system in the nucleus.

5.1.2. Role in gene expression regulation

The present work links Hsp90 to gene expression regulation at several different levels. First, it strengthens the well-documented stabilizing effect of Hsp90 on transcription (co)factors. We identified a few novel Hsp90 interactors (BCLAF1, Oct1, DEK, NCOR1, HCF-1, etc.) that act as transcription (co)factors and thus affect the expression of individual genes. Second, our data points to more general role of Hsp90 in transcription initiation (via the general transcription factors TAF4 and TAF15), elongation (via components of PAF1 and FACT complexes) and probably pausing (via PAF1 complex). Third, our findings emphasise a strong link between Hsp90 and chromatin remodelling/modification via interactions with related complexes. According to our data SET1, Sin3A, NuRD, CoREST and pBAF (FIGURE 11A) complement the list of chromatin modifiers/remodellers that Hsp90

CHAPTER 1 DISCUSSION AND OUTLOOK

interplays with (Smyd3, trithorax/MLL, RSC complex, Ezh2, etc. [135, 298-300]; FIGURE 5) and strengthens its role in the world of epigenetics. Thus, Hsp90 may contribute to chromatin structure modulation by histone methylation (via SET1 complex [427]), demethylation (via CoREST [428]), deacetylation (via Sin3A, CoREST and NuRD complexes [428-430]) or nucleosome sliding (via pBAF and NuRD complexes [429, 431]). Hsp90 may modulate the assembly, disassembly, stability, recruitment to chromatin or activity of the above-mentioned epigenetic machineries and thus contribute to chromatin dynamics. Regardless the precise mechanism, it is tempting to speculate that these interactions enable the chaperone to coordinate rapid widespread changes in gene transcription according to cellular demands. In support of this hypothesis, an early report in Drosophila documented Hsp90 association with specific chromatin domains engaged in active gene transcription following heat shock, known as heat shock-induced puffs [316]. Recruitment to these loci is apparently crucial for cell survival and recovery.

The precise relationship of Hsp90 with the above-mentioned complexes remains to be investigated. Hsp90 has been so far shown to promote the assembly and disassembly of numerous complexes [208, 277]. By supporting a desired conformation the chaperone enables protein interactions that result in cooperative assembly of large multicomponent complexes with important functions. Disassembly of the so-formed complexes is not of a lesser importance. They must be actively and persistently deconstructed in order to function on a useful time scale and to prevent intracellular crowding effects. However, mechanisms for modulating or terminating the actions of these complexes are not well understood. An intriguing speculation by Echtenkamp and Freeman, 2014 [414] describes molecular chaperones (Hsp90 included) as protein complexes disassembly factors that meet the immediate and specific needs of a cell. This may explain the specific interactions of Hsp90 with numerous protein complexes.

5.1.3. RNA metabolism

The amount of RNA-processing proteins described as Hsp90 interactors in this study supports a strong involvement of Hsp90 in different aspects of RNA metabolism – mRNA splicing, transport, degradation, cleavage and poly-adenylation. The identification of such big number of RNA-processing factors may be due to the fact that they usually operate as part of multi-component complexes that Hsp90 pulls down as a whole. Thus, many of them may appear indirect interactors. It remains to be investigated which of these factors directly interact with the chaperone. Another possibility could be that a client has already bound both

CHAPTER 1 DISCUSSION AND OUTLOOK

Taken together, our efforts re-established Hsp90 role in protein folding/degradation, transcription regulation, DNA repair, and RNA metabolism by uncovering previously known as well as novel interactors from the above-mentioned categories. The failure to re-discover additional well-documented Hsp90 clients (HSF1, p53, Cdk4/6/9, NF-kB, STAT3, DNA replication licencing factor MCM7, etc.) as in the case of many other studies [139, 201, 203, 204, 254] is consistent with the appreciation that proteomics techniques usually detect only the most abundant proteins. The applied protocol could be further improved towards the detection of low-abundant as well as transient substrates by increase of starting material (number of cells) or utilization of protein crosslinking strategies (to avoid interactions dissociation during experimental procedure). A preliminary experiment employing the protein crosslinking reagent DSP was able to uncover a number of co-chaperones that we could not detect in the affinity purification experiment in native conditions - Hop, Aha1, FKBP4, FKBP5, AH receptor-interacting protein (AIP), protein phosphatase 5 (PP5), and suppressor of G2 allele of SKP1 homolog (SUGT1). Optimization of the procedure in future could help us complement the list of novel nuclear Hsp90 clients. In addition, performing an IP using a mutant of Hsp90 that binds substrates but does not allow their dissociation [367] could tremendously boost the identification of novel clients – especially the ones that transiently interact with the chaperone. This strategy, for example, may allow the detection of additional protein kinases and thus expand the short list we uncovered (PRPF4B, TRRAP, PRKDC and CSNK2A2) performing affinity purification of Hsp90 followed by MS analysis. Establishing a dependency of Hsp90 with oncogenic kinases, for example, may offer an opportunity to target the stability of these nuclear kinases via Hsp90i [432].

Remarkably, the current progress in revealing Hsp90 interactome could only be achieved by integration of results obtained by using various strategies [138, 139, 203]. Apart from demonstrating physical relationship with proteins, we provided a list of factors that also functionally interact with the chaperone (demonstrated in a synthetic lethality screen)

CHAPTER 1 DISCUSSION AND OUTLOOK

(TABLE 7). The overlap is not big but one should keep in mind that in the synthetic lethality screen only certain transcription regulators were investigated and not the complete list of Hsp90 physical interactors. Thus, we consider the hits of that list high-confident novel Hsp90 substrates as paired physical and functional interaction studies represent the gold standard for demonstrating an Hsp90 partnership. Their precise relationship with the chaperone remains to be investigated. First, reciprocal immunoprecipitation/affinity purification could strengthen the evidence of physical interaction. Second, one needs to demonstrate that Hsp90 inactivation (achieved by knock-down or pharmacological inhibition, for example) leads to destabilization and degradation of the factors mentioned above – a dependency that defines a client of Hsp90. Furthermore, it will be intriguing to reveal whether the chaperone affects the total nuclear pool of a protein or a subpopulation only. PBRM1, for example, is a key component of the pBAF chromatin remodelling complex and appears both physical nad functional interactor of Hsp90. It is possible that Hsp90 stabilizes the total nuclear pool of PBRM1 or just a subpopulation – PBRM1 that is pBAF complex-engaged, DNA-bound or diffused in nucleoplasm. This question could be addressed by Hsp90 inactivation followed by:

• nuclei fractionation and subsequent Western blotting analysis to detect PBRM1 in total nuclear, nucleoplasmic and chromatin protein extracts • PBRM1 ChIP-seq analysis to test whether Hsp90 is needed for PBRM1 chromatin localization and if so, does is it affect its genome-wide distribution or just isolated loci. • PBRM1 purification to check whether Hsp90 is required for its incorporation into the pBAF complex – failure to pull down additional components of the complex could prove such a dependency

The sequence of experiment described above could successfully be applied to investigate the relationship of Hsp90 with other chromatin remodelling/modifying complexes found as part of its interactome network – SET1A, Sin3A, NuRD, CoREST, etc, as well as transcription regulators – BPTF, DEK, Oct1, NCOR1, etc. Furthermore, PBRM1 and Hsp90 inactivation phenotypes need to be compared in order to strengthen their functional relationship. This could be achieved in the context of gene expression mis-regulation (detected by RNA-seq) or chromatin accessibility changes (investigated by ATAC-seq). Notably, PBRM1 is linked to oncogenic transformation [433]. Thus, it will be worth testing whether the malignant phenotype driven by PBRM1 mutations could be rescued by Hsp90 inhibition (this point will be covered in the next section).

CHAPTER 1 DISCUSSION AND OUTLOOK

5.1.3. Novel opportunities of anti-cancer therapies

Remarkably, a considerable number of known Hsp90 substrates are overexpressed or mutated in malignancies [174]. Thereby, it does not come as a surprise that suppressing the interaction between Hsp90 and client proteins has turned into an attractive approach in cancer therapy. More than 50 clinical trials (finalized or in progress) utilized Hsp90 inhibitors to destabilize oncoproteins by blocking their interaction with the chaperone (https://clinicaltrials.gov) [74, 178, 179]. In that respect, nuclear Hsp90 interactome data obtained in this work can be exploited as a tool to investigate opportunities of additional anti- cancer therapies. Many novel Hsp90 clients that we uncovered are known to drive oncogenic transformation of cells and contribute to malignant phenotype maintenance – pBAF chromatin remodelling complex, DEK transcription regulator, CBLL1 E3 protein ubiquitin ligase and BPTF [293], and others. pBAF chromatin remodelling complex represents an intriguing example in that respect. It is the human analogue of the chromatin remodelling SWI/SNF (SWItch/Sucrose Non-Fermentable) complex in yeast [431]. It is capable of altering the position of nucleosomes along DNA and thereby makes space for loading of RNA pol II and other TFs. Potential Hsp90-pBAF interaction could open another possibility for Hsp90 inhibition mediated oncogenic transformation suppression as the complex components are mutated in almost 20% of all human cancers [434-436]. Malignant transformations driven by pBAF mutations could be targeted through inhibition of Hsp90 resulting in destabilization of this oncogenic driver. Intriguingly, recent reports suggest that pBAF loss and the resulting chromatin structure changes sensitize clear cell renal cell carcinoma (ccRCC) cells to immunotherapy [437, 438]. Both the studies disrupted pBAF function by inactivating the PBRM1 complex component. Notably, we found that Hsp90 not only physically but also functionally interacts with PBRM1. That strongly points towards the assumption that PBRM1 is a genuine client of the chaperone and opens an additional opportunity for PBRM1/pBAF inactivation by employing Hsp90 inhibitors.

An exciting aspect of Hsp90 inhibitors application in cancer treatment is the possibility of combinatorial therapy [439]. Thus, an oncogenic client could be simultaneously targeted to achieve both activity reduction (achieved by a target-specific inhibitor) and destabilization (achieved by Hsp90 inhibition) [440]. HER-2 (human epidermal growth factor receptor 2) protein kinase, for example, is a client of Hsp90 [249, 441]. It is amplified in 20–30% of early-stage breast cancers [442]. Combinatorial therapeutic strategies utilizing NVP-AUY922 (Hsp90 inhibitor) and trastuzumab (HER-2 inhibitor) gave exciting results [443, 444]. In the

CHAPTER 1 DISCUSSION AND OUTLOOK

same fashion, Hsp90i could be combined with specific inactivators of the novel oncogenic clients (pBAF chromatin remodelling complex, DEK transcription regulator, CBLL1 E3 protein ubiquitin ligase and others) we discovered within the current work.

5.1.4. Hsp90 during proteotoxic stress

To obtain insight into the contribution of nuclear Hsp90 in cell survival and recovery upon proteotoxic stress, we investigated its interactome upon exposure of HEK293 cells to heat shock (43 °C for 90 min). We speculate that the surprisingly negligible overlap between biological affinity purification replicates arises from the chaperone titration model [368-370]. It assumes that upon heat shock chaperones are titrated away from their clients by the unfolded/misfolded proteins that massively accumulate. Thus, our results point to a more general role of Hsp90 in protein stabilization upon proteotoxic stress compared to its specific clientele in normal growth conditions. This transformation of Hsp90 function could represent a cellular strategy to ensure survival and recovery. Although not investigated, we expect that similar results will be obtained not only for Hsp90 nuclear pool but also the cytosolic one.

Taken together, our data expand the knowledge of the Hsp90 interactome and PTMs, and provide a further step in our understanding of the Hsp90 chaperone system. Comprehensive knowledge about the roles of Hsp90 in various aspects of cellular physiology would not only improve our understanding of cell biology but might also lead to the design of alternative strategies for treatment of certain diseases, for example, cancer.

5.2. Nuclear Hsp90 PTMs

Molecular chaperone Hsp90 stabilizes a large number of proteins to ensure proper functioning of the cell in normal conditions and its survival and recovery upon proteotoxic stress. Its complex reaction cycle, the vast diversity of clients and dynamic conditions it operates at make its regulation extremely complicated and crucial. In addition to co- chaperones binding, fine-tuning of its function has been achieved by a variety of PTMs. While cytosolic Hsp90 has been extensively studied in that respect its nuclear pool remains poorly investigated. Here we characterized nuclear Hsp90α and Hsp90β isoforms PTMs.

The identified phosphorylated serine residues of nuclear Hsp90α (Ser-231 and Ser-263) and Hsp90β (Ser-255) were already reported for cytosolic Hsp90 by Lees-Miller and Anderson in 1989 [409]. Later on, a considerable number of studies confirmed the observation in different cell lines (phosphosite.org). Detecting these well-established modifications of cytosolic Hsp90 in its nuclear pool shows that they are probably constitutive

CHAPTER 1 DISCUSSION AND OUTLOOK

phosphorylation sites of crucial importance for chaperone general functioning. They have been previously shown to modulate Hsp90 binding to certain clients and contribute to malignant phenotype [80, 84, 98]. Their functional significance for nuclear Hsp90 remains to be investigated. A literature search revealed that Ser-231, Ser-255 and Ser-263 are targeted by casein kinase 2 (CK2) [409, 445] – a client of Hsp90 [446]. Found in nearly every subcellular compartment [447] CK2 could probably phosphorylate Hsp90 in the cytosol (prior to its translocation to the nucleus) or directly act on its nuclear pool.

Phosphorylated Hsp90α (Ser-231 and Ser-263) and Hsp90β (Ser-255) were identified in both Ctl and HS conditions but with different intensities. Phosphorylated at Ser-231 and Ser- 263 of Hsp90α increased in abundance in response to stress, whereas phosphorylation at Ser- 255 of Hsp90β - decreased (Fig. 4). It is not clear whether these are negative consequences of the proteotoxic stress or regulatory changes in favour of the heat shock response. Charged linker domain has been reported to affect temperature sensitivity in yeast cells [34] so it is tempting to speculate that charged linker PTMs state could be important for the HS response in mammalian cells.

Acetylation of Lys-559 in the middle domain of nuclear Hsp90β that we observed has been previously reported in a number of malignant cell lines (phosphosite.org) and probably represents a constitutive modification crucial for both cytosolic and nuclear pool of the chaperone. Here we demonstrate for the first time K559 acetylation of Hsp90β in untransformed cells and clearly demonstrate that it is the nuclear pool affected as well (in contrast to work done so far on the whole-cell pool of the chaperone). No acetylation of nuclear Hsp90α isoform was detected although modified K558 and other residues have been previously described [105]. The impact of K559 acetylation on Hsp90β function has not been investigated so far. However the corresponding K558 residue of Hsp90α has been described in literature [105] as important for interaction with p23 and extracellular secretion of the chaperone in malignant cells. Both Lys-558 and Lys-559 of Hsp90α and Hsp90β, respectively have been also found ubiquitylated (phosphosite.org). One could speculate that ubiquitin ligases and acetylases may compete for these residues to fine-tune Hsp90 function - acetylation of Lys-559 could prevent HSP90β degradation and extend the half-life of HSP90 complexes. A similar interplay between ubiquitination and acetylation has already been described in yeast [448]. If proven, Lys-559 acetylation blockage could potentially stimulate Hsp90 ubiquitination and degradation in cancer cells addicted to the chaperone and represent a novel therapeutic strategy [449, 450]

CHAPTER 1 DISCUSSION AND OUTLOOK

Hsp90 methylation has been poorly studies so far [107-109]. The addition of a methyl group to arginine or lysine residues of proteins has emerged as an important regulator of nuclear processes [451, 452] thus it was intriguing to assess whether nuclear Hsp90 is subjected to methylation. We revealed three novel methylation sites for Hsp90β –Lys-284 (charged linker), Arg-502 and Lys-550 (middle domain) (Fig. 3). Lys-550 was modified in both Ctl and HS conditions with no change in abundance thus probably representing a constitutive modification site. Lys-284 and Arg-502 methylation appeared Ctl and HS specific, respectively. That dynamics could probably be involved in translocation of Hsp90β to the nucleus upon stress and its contribution to the HS response. Further functional analysis may unveil the physiological significance of HSP90β methylation in more detail. In addition, modified Lys-284, Arg-502 and Lys-550 of HSP90β may represent isoform-specific PTMs since we could not detect methylation of nuclear Hsp90α. Most of Hsp90 PTMs described in literature so far are common to both Hsp90α and Hsp90β isoforms. There are few instances, such as phosphorylation at Thr-5 and Thr-7 of Hsp90α in response to DNA damage [89, 90], which are isoform-specific. However, there is no side-by-side comparison done so far and it remains unclear whether the two isoforms strongly differ in their PTMs status and whether those variations could explain the functional differences between them. Our work demonstrates isoform-specific modification of Hsp90 by direct comparison for the first time - Lys-284, Arg-502 and Lys-550 methylation of HSP90β but not Hsp90α. The above methylation sites could also represent cellular compartment-specific modifications since no study identified modification of these residues in cytosolic Hsp90β so far.

Smyd2 is the only methyltransferase known so far to modify Hsp90 [108]. Given its almost exclusively cytosolic localization, it may methylate the chaperone in the cytosol (prior to its translocation to the nucleus). However, it is more likely that a different nucleus-resident methylase is targeting Lys-284, Arg-502 and Lys-550 since not reported in cytosolic Hsp90β so far.

Nuclear Hsp90 PTMs analysis reported here could serve as a starting point for future experiments in the lab for further characterization of nuclear Hsp90. To begin with, the applied protocol could be optimized further in order to improve PTMs detection – increase of starting material, use of additional specific inhibitors to prevent PTMs removal during the experimental procedure, immunoenrichment of certain PTMs to boost their identification, etc. Thus the initial observations could be confirmed and additional PTMs - detected.

CHAPTER 1 DISCUSSION AND OUTLOOK

Functional significance of the identified modifications could be investigated in future by site-directed mutagenesis to reveal the way PTMs fine-tune nuclear Hsp90. Corresponding amino acids could be substituted either by non-modifiable residues to unveil loss-of-function effect or by mimetics to reveal gain-of-function consequences. These type of experiments for example, could further establish the supposedly constitutive nature of Ser-231/255/263 phosphorylation and Lys-559 acetylation by demonstrating their role in clients/co-chaperones binding. Nuclear clients of Hsp90 described by us (section 4.1 of chapter 1) as well as nucleus-resident co-chaperones could be studied in that respect.

Site-directed mutagenesis followed by proteotoxic stress response investigation could reveal the effect of certain PTMs on heat shock response implementation. Given Ser- 231/255/263 phosphorylation, Arg-502 methylation and Lys-284 methylation change in abundance upon HS it will be intriguing to study their contribution in that respect. Furthermore, mutagenesis of Lys-284 and Arg-502 methylation sites followed by subcellular fractionation or microscopy may reveal their effect on Hsp90 nuclear translocation as proposed earlier.

An exciting follow-up in the light of Hsp90 inhibition cancer therapy could be Lys-559 acetylation. Mutagenesis followed by Hsp90 stability assessment could potentially reveal its functional significance as part of a "switch-off" mechanism as discussed earlier – acetylases and ubiquitin ligases may compete for the residue in order to promote stabilization or degradation of Hsp90, respectively. If so, Lys-559 acetylation could be investigated as a cancer therapy target.

The PTMs identified could also be confirmed in other human cell lines. This will be especially important in the case of Lys-559 acetylation since we demonstrated for the first time its existence in untransformed cells. The development of PTM-specific antibodies in future would help confirm important observations –dynamics in response to stress (Ser- 231/255/263 phosphorylation; Arg-502 and Lys-284 methylation), isoform-specificity (Lys- 550, Arg-502 and Lys-284 methylation), effect on Hsp90 subcellular localization (Arg-502 and Lys-284 methylation), cellular compartment specificity (Lys-550, Arg-502 and Lys-284 methylation) etc.

The modifiers responsible for the vast majority of Hsp90 PTMs are still unknown. We could make use of the nuclear Hsp90 interactome revealed by us (section 4.1 of chapter 1) to pick up potential candidates and study their effect on the chaperone. Knock-down

CHAPTER 1 DISCUSSION AND OUTLOOK experiments followed by Hsp90 PTMs analysis could lead to discovery of Hsp90 modifying enzymes. A few enzymes known to target cytosolic Hsp90 (p300 acetylase, casein kinase 2, SMYD-2 methylase) could also be tested to look for an effect on the nuclear pool of the chaperone.

CHAPTER 2 INTRODUCTION

CHAPTER II

Hsp90 chaperones the transcriptional co-factor Host Cell Factor 1 (HCF-1)

1. Introduction

HCF-1 (also known as HCF, C1, HCFC1, VCAF, and CFF) is a chromatin-associated scaffold protein [359] originally discovered and widely characterized as a host-cell factor required for herpes simplex virus (HSV) transcription [360-363]. It is conserved across humans [453, 454], mice [455], nematodes [456], and fruit flies [457] but not in yeast. This suggests that HCF-1 is a highly conserved protein with an important role in the biology of metazoans. Cumulative evidences link the protein to cell cycle regulation at transcriptional level [458-461].

1.1. HCF-1 role in HSV gene expression

Infection by the HSV follows a tightly regulated viral gene transcription program that begins with the expression of the viral immediate early (IE) genes and ultimately results in virus production and lysis of the host cell [360]. HCF-1 contributes to the formation of the so called VP16-induced complex that activates the expression of viral IE genes. VP16 (virion protein 16) is contained within the virus particle of the HSV and released into animal cells upon infection. It is initially bound by HCF-1 in the absence of target DNA and the host-cell factor facilitates its interaction with the cellular transcription factor Oct-1 bound to the HSV IE promoters [360, 361, 363, 462, 463].

1.2. HCF-1 protein structure

HCF-1 contains several distinct regions (FIGURE 26):

• Kelch domain It has been named due to its similarity to the Drosophila protein Kelch [464]. It binds VP16 and is sufficient to stabilize a VP16-induced complex. The Kelch domain is responsible for interaction with TF such as VP16 and thus tethers HCF-1 to chromatin. It is well conserved in HCF homologues, consistent with a critical cellular function [359, 456, 460].

CHAPTER 2 INTRODUCTION

Figure 23. Role of VP16-induced complex in the herpes simplex virus (HSV) lytic cycle. In the lytic mode of HSV infection, VP16 (yellow) is delivered to the host cell, where it associates with HCF-1 (red), leading to VP16-induced complex assembly on HSV immediate-early (IE) promoters with Oct-1 (blue). This initiates a cascade of gene expression that results in virus production and lysis of the host cell (adapted from Wysocka and Herr, 2003 [463]).

• Basic region It is enriched in basic amino acid residues and serves as a platform for diverse protein interactions [463].

• Proteolytic processing domain (PPD)

Comprised of six HCF-1PRO repeats that direct the proteolytic cleavage of the protein by the unusual protease OGT (O-linked β-N-acetylglucosamine transferase) [465].

• Acidic region It is enriched in acidic amino acid residues. It contains a transcriptional activation domain, which stimulates the activity of VP16 [466].

CHAPTER 2 INTRODUCTION

• Fibronectin type 3 (Fn3) repeats The two Fn3 repeats facilitate HCF-1 subunits association [467].

• C-terminal nuclear localization signal (NLS) [468].

Figure 24. HCF-1 protein structure. Structural or sequence elements and functional regions are indicated (taken from Wysocka et al., 2003 [364]).

1.3. HCF-1 polypeptide cleavage

HCF-1 is initially synthesized as a single 300-kDa precursor molecule that migrates to the cellular nucleus. There it is subsequently cleaved within the central proteolytic processing domain (PPD) by the unusual protease OGT [469, 470]. That cleavage generates stable N- and C-terminal (HCF-1N and HCF-1C) subunits that remain non-covalently associated through two pairs of ‘self-association sequences’ (SAS1 and SAS2) [453, 454, 465, 467]. In line with this, the two subunits co-precipitate in pull-down experiments [364] and associate with chromatin as a heterodimer [471]. Although HCF-1 is not known to bind DNA directly, each subunit displays chromatin association activity [359, 472]. On the one hand, the Kelch domain tethers HCF-1N to chromatin by contacting DNA-binding proteins that possess HCF- 1-binding motif (HBM) - D/EHxY tetrapeptide (“x” stands for any residue) [466, 473, 474]. On the other hand, the acidic region and NLS are required for HCF-1C chromatin localization [472].

The functional significance of HCF-1 polypeptide cleavage followed by N- and C- terminal subunits re-association remains unclear. Studies through the years have proposed different theories: • Control of HCF-1 interactions and transcriptional co-factor function. HCF-1 centrally located proteolytic processing domain (PPD) interacts with a series of proteins including the transcriptional coactivator FHL2 [475, 476] - a protein which collaborates with HCF-1 to stimulate the HSV IE genes [477]. As the interaction

CHAPTER 2 INTRODUCTION

determinants for FHL2 lie within the PPD, cleavage of HCF-1 results in loss of FHL2 binding and abolishment of downstream effects. • Separations of HCF-1N and HCF-1C functions. Separate regions of HCF-1 may need to function in unison but still be able to disassociate in a regulated manner to perform different functions. The proteolytic processing may provide the required flexibility and quick handover of substrates, for example. In line with this hypothesis, a study by Julien et al., 2003 demonstrated that HCF-1N subunit is responsible for promoting G1 phase progression, and the HCF-1C subunit is involved in proper M phase progression. HCF-1 proteolytic processing proves necessary to separate and ensure these distinct functions although the mechanism involved has not been elucidated [459]. Perhaps, it is required to fully activate HCF-1 subunits activity similarly to the MLL family of HMT proteins [478, 479]. A study by Luciano and Wilson., 2002 supports this hypothesis demonstrating that HCF-1N determines VP16 promotor targeting whereas HCF-1C facilitates VP16 activity [480]. • Stoichiometric levels of HCF-1N and HCF-1C. HCF-1 subunits may be expressed as a single precursor protein to ensure their stoichiometric levels in the cell, as has been proposed for a number of viral polyproteins [481] . • Nuclear retention of the HCF-1N subunit. HCF-1 is a stable protein [465] and may persist through multiple rounds of cell division. Perhaps HCF-1 subunits association is required to ensure the localization of HCF-1N subunit back to the newly formed nucleus during mitosis. HCF-1N does not possess an NLS sequence and relies on its association with HCF-1C for translocation to the nucleus.

1.4. HCF-1 role in cell cycle

HCF-1 role in cell cycle first became apparent in a study by Goto et al., 1997 through the characterization of a temperature-sensitive hamster cell line called tsBN67 [458, 482]. These cells undergo a reversible cell proliferation arrest after 36 to 48 h at the non-permissive temperature of 40 °C. Interestingly, a single proline-to-serine substitution in the HCF-1 Kelch domain (P134S) was shown to cause the phenotype [458]. This mutation was later on demonstrated to disrupt HCF-1 association with chromatin at the non-permissive temperature [359] suggesting that (i) the Kelch domain of HCF-1 is crucial for interaction with factor which tether it to chromatin, (ii) the disruption of chromatin association is the primary cause of the tsBN67 cells proliferation defect and (iii) HCF-1 is a key regulator of cell cycle [359]. The gene expression profile of arrested tsBN67 cells suggests a G0/G1 arrest [458, 483]

CHAPTER 2 INTRODUCTION

pointing to HCF-1 role in G1 to S phase transition. In addition, a fraction of tsBN67 cells display binucleation which defines a role of HCF-1 in cytokinesis as well [458, 483]. Rescue experiments revealed that although the Kelch domain is sufficient for chromatin association, it is insufficient to overcome the tsBN67 cell proliferation defects - an adjacent “basic” region is required [460], indicating that multiple elements of HCF-1 are involved in promoting cell proliferation.

HCF-1 primarily exhibits its effect on cell cycle through interactions with members of the E2F family of transcription factors – proteins that control human cell proliferation by either repressing or activating the transcription of genes required for cell-cycle progression, particularly the S phase [484-486]. HCF-1 associates with both activator (E2F1 and E2F3a) and repressor (E2F4) E2F proteins in a cell-cycle selective manner. During the G1-to-S phase transition, HCF-1 displays coactivator properties – it binds the E2F1 activator and recruits the mixed-lineage leukemia (MLL) and Set-1 histone H3 lysine 4 methyltransferases to E2F responsive promoters to induce histone methylation and transcriptional activation. During the early G1, HCF-1 displays repressor properties – it binds the E2F4 repressor and recruits Sin3A complex to induce histone deacetylation and transcriptional repression [487].

Though a key player in cell cycle regulation, HCF-1 levels and nuclear localization are known not to be cell-cycle regulated [487].

1.5. HCF-1 interaction partners

HCF-1 is now recognized as an essential transcriptional co-factor with global impact on gene transcription and cell cycle progression via interactions with multiple cellular TFs, co- factors, and chromatin modification/remodeling components. It has been shown to interact with several well-characterized transcription factors such as Sp1, ZNF143, THAP11 (also known as Ronin), GABP, LZIP/Luman, Zhangfei, FOXO3 and YY1 [471, 474, 488-494]. Other chromatin-binding factors such as protein phosphatase PP1 [495], programmed cell death regulator PDCD2 [496] and the deubiquitinating enzyme BAP1 [491] are also among HCF-1 interactors. Through the years, HCF-1 has also been associated with a variety of histone-modifying activities that facilitate activation or repression of transcription. Mixed- lineage leukemia (MLL) and Set1 histone H3 lysine 4 methyltransferases (H3K4 HMTs), Sin3 HDAC, MOF HAT, PHF8 HDM, and others have been shown to interact with HCF-1 to control the expression of both host cell and viral genes [362, 364, 497-503].

CHAPTER 2 INTRODUCTION

A recent study by Alfonso-Dunn et al., 2017 [504] investigated HCF-1 interactome on a global scale for the first time. The analysis done in the human lung fibroblast cell line MRC-5 demonstrated that the protein not only interacts with a variety of TFs and chromatin modification/remodeling machineries but with components of the mediator complex, subunits of RNA polymerase II and elongation factors. Thus, HCF-1 role in transcription regulation was expanded to include specific molecular links to transcription elongation.

CHAPTER 2 AIMS

2. Aims

The investigation of nuclear Hsp90 physical (section 4.1 of chapter 1) and functional (section 4.3 of chapter 1) interactors done by us revealed that HCF-1 is one of the factors that interact with the chaperone both physically and functionally. We could also demonstrate that Hsp90 co-immunoprecipitates with components of SET1 histone methylation and SIN3A histone deacetylation complexes (FIGURE 11) which are known HCF-1 interactors [364]. These observations inspired us to ask whether there is an interplay between Hsp90 and HCF- 1 in the nucleus – a co-operation that has not been reported so far. Therefore, in this part of my thesis, I investigated the collaboration between Hsp90 and HCF-1 in the nucleus in order to reveal its functional significance. Such efforts would enhance our understanding of the following aspects:

• HCF-1-mediated control of gene expression. One could speculate that due to its complex structure and elaborate activity of a factor bridging TF and proteins with chromatin modifying/ remodelling activities HCF-1 necessitates chaperone action and control. The Hsp90 molecular chaperone could provide the conformational stability and flexibility required. • Hsp90-mediated control of gene expression. Hsp90-HCF-1 collaboration may prove an example of chromatin-associated action and transcription regulation role of the chaperone Hsp90 considered to be mostly cytosolic.

CHAPTER 2 MATERIALS AND METHODS

3. Materials and Methods

3.1. Cell Culture and Cell Lines

HEK293 and K562 cells were grown as already described in section 3.1 of chapter 1. Wild type HA36CB1 embryonic stem cells, derived from mixed 129-C57Bl/6 background blastocysts were kindly provided by Prof. Dr. Dirk Schübeler (Friedrich Miescher Institute for Biomedical Research, Basel, Switzerland). Cells were grown on 0.2 % gelatine-coated plates (Invitrogen, #S006100) in DMEM (Sigma Aldrich, #D5671) supplemented with 15 % Fetal Bovine Serum (Invitrogen), Nonessential Amino Acids (Millipore, #TMS-00-C), 1 mM L-glutamine (Gibco, #25030), LIF (in-house produced), 1 mM Sodium Pyruvate (Gibco, #11360-070) and 100 μM b-mercaptoethanol (Gibco, #31350-010).

3.2. siRNA-mediated protein knockdown

Hsp90 alpha, Hsp90 beta and HCF-1 siRNAs were purchased from Dharmacon and used at 3 nmol final concentration. Transfection was performed using RNAiMAX transfection reagent (Thermo Fisher Scientific, #13778075) following manufacturer’s protocol. HEK cells were plated 16 hours before and collected 48 hours after treatment. Protein knockdown efficiency was evaluated by western blotting.

3.3. ChIP-qPCR and ChIP-seq

Chromatin immunoprecipitation was performed using wild-type K562, wild-type HEK or recombinant HA36CB1 ES cells [505]. The last were generated by recombinase-mediated cassette exchange (RMCE) to express either biotin-tagged Hsp90 or biotin-tagged GFP as previously done by Lienert et al., 2011 [506]. Briefly cells were maintained in growth medium supplemented with 25 μg/ml hygromycin (Roche, #10843555001) for 10 days. 0.8 x 105 cells were plated in a 12-well plate and transfected with Lipofectamine 3000 (Life Technologies, #L3000-015) according to manufacturer`s protocol. Positive clones were selected by adding 3 μM ganciclovir (Sigma, #G2536) into growing medium 72 hours post transfection and stable cell lines were generated. All the buffers used except for EB buffer (Qiagen, #19086) were supplemented with the following protease inhibitors: 3 µg/ml aprotinin, 10 µg/ml leupeptin, 1 µg/ml pepstatin, 0.1 mM PMSF.

3.3.1. Preparation of chromatin

Twenty million K562, HEK or ES cells were used per condition. They fixed using 1 % methanol-free formaldehyde (Thermo Scientific, #28906) in IMDM (K562), DMEM (HEK)

CHAPTER 2 MATERIALS AND METHODS

or PBS (ES cells) for 10 min at room temperature (end-over-end rotation). The cross-linking reaction was stopped by adding glycine to a ﬁnal concentration of 125 mM and incubated for 5 min at room temperature (end-over-end rotation). Cells were spun down at 750 xg for 5 min at 4 and washed twice with cold PBS.

℃Chromatin preparation for K652 and ES cells was performed according to Arrigoni et al., 2016 [507]. Pellet was resuspended in 1 ml of Farnham Buffer (5 mM PIPES pH 8, 85 mM KCl, 0.5 % Igepal) and sonicated in 1 ml Covaris tubes using Covaris S220. The following settings were used: Peak Power=75, Duty Factor=2, Cycles/Burst=200, Time=2 min (K562) or 1 min (ES cells). Nuclei were transferred to a fresh microcentrifuge tube and spun down at 1000 xg for 5 min. Nuclear pellet was washed twice with Farnham buffer and resuspended in shearing buffer (10 mM Tris-HCl pH 8, 0.1 % SDS, 1 mM EDTA). Material was sonicated in 1 ml Covaris tubes using the following settings: Peak Power=140, Duty Factor=5, Cycles/Burst=200, Time=25-30 min (K562) or 12 min (ES cells). Obtained chromatin was spun down at 20,000 xg for 10 min at 4 to remove debris. Supernatant was transferred to a fresh protein low binding tube. A small ℃aliquot of sheared chromatin was processed in order to assess sonication efﬁciency and chromatin yield prior to IP. For that purpose, chromatin was treated with proteinase K and RNase A and decrosslinked overnight at 70 . DNA was purified by phenol-chloroform-isoamyl alcohol extraction method. DNA concentration℃ was determined using Qubit following manufacturer’s protocol. Chromatin shearing efficiency was assessed on a 2 % agarose gel. The optimal size range of DNA for ChIP-seq analysis should be between 200 and 600 base pairs. Prior to IP chromatin was diluted 1:1 with lysis buffer 3 to achieve a final SDS concentration of 0.05%.

Chromatin preparation for HEK cells was performed according to Lodhi and Tulin, 2011 [508] with minor modifications. Cell pellet was resuspended in 2 ml of cell lysis buffer (5 mM PIPES pH 8, 85 mM KCl, 1 % NP-40) and incubated on ice for 15 min. Cell suspension was passed through an 23G needle (5x times) and spun down at 1000 xg for 5 min. Nuclei were washed twice with cell lysis buffer and resuspended in 600 μl nuclei lysis buffer (50 mM Tris-HCl pH 8, 1 % SDS, 10 mM EDTA). Material was sonicated for 15 min with Bioruptor sonicator (30 sec on/30 sec off, high power). Obtained chromatin was spun down at 20,000 xg for 10 min at 4 to remove debris. Supernatant was recovered and a small aliquot of sheared chromatin was℃ processed in order to assess sonication efﬁciency and chromatin yield. Prior to IP chromatin was diluted 1:10 with ChIP dilution buffer (1.1 % Triton-X, 1.2

CHAPTER 2 MATERIALS AND METHODS

mM EDTA, 16.7 mM Tris, 167 mM NaCl) and precleared for 1 hour at 4 using Protein A

agarose beads (Roche, # 11134515001). ℃ 3.3.2 Chromatin Immunoprecipitation

Chromatin Immunoprecipitation from K562 and HEK cells was performed as previously described by Lee et al., 2006 [509]. Briefly, Protein A or G magnetic beads were washed twice with blocking solution (0.5 % BSA in PBS) using a magnetic rack and incubated with 5ug of antibody in block solution for 6 hours at 4 (end-over-end rotation). Bead-antibody

complex was then washed twice with block solution.℃ 200ug of chromatin was added and samples were incubated overnight at 4 (end-over-end rotation). An aliquot of chromatin

was saved as input DNA. Beads were washed℃ once with wash buffer 1 (20 mM Tris-HCl pH 8, 150 mM NaCl, 2mM EDTA, 0.1% SDS, 1%Triton X-100), wash buffer 2 (20 mM Tris- HCl pH 8, 500 mM NaCl, 2 mM EDTA, 0.1 % SDS, 1 %Triton X-100) and wash buffer 3 (10 mM Tris-HCl pH 8, 250 nM LiCl, 2 mM EDTA, 1% NP40) and TE. DNA was eluted by addition of 200 μl elution buffer (50 mM Tris-HCl, pH 8.0, 10 mM EDTA and 0.5 % – 1 % SDS) to beads and heating at 65 °C for 45 min under vigorous shaking. Reverse crosslinking was performed overnight at 65 °C in a laboratory oven for at least 6 hours. Samples were then treated with RNase and proteinase K and the DNA was purified by Phenol–chloroform– isoamyl alcohol extraction. DNA pellet was resuspended in 50 μl of EB buffer. Immunoprecipitated DNA was either used for qPCR or sequenced by the in-house next generation sequencing facility.

Chromatin Immunoprecipitation from HA36CB1 ES cells was performed as previously described by Baubec et al., 2013 [505] .

3.3.4. ChIP-qPCR

Real-time quantitative PCR of ChIP DNA was performed using Absolute QPCR Mix, SYBR Green, ROX kit (Thermo Fisher, #AB-1163/A) according to manufacturer’s protocol. Raw data was analysed and enrichment of the immunoprecipitated DNA at the corresponding loci expressed as percentage relative to the input DNA.

3.3.5. ChIP-seq data analysis (done by Barbara Hummel, MPI-IE, Freiburg)

ChIP-seq reads were aligned to the human genome hg19 using Bowtie 2. Only uniquely mapping reads with up to two mismatches within the entire length of the read were considered for further analysis. The resulting reads were extended to fragment size as defined

CHAPTER 2 MATERIALS AND METHODS

by paired-end sequencing and were normalized to total reads aligned (reads per million, rpm). Peak detection was done with MACS (model- based analysis of ChIP-Seq) version 1.4.2 using default parameters. Gene annotations and transcript start site information for human genes were from taken from Ensembl release 72.

3.4. Chem-seq

Twenty million K562 cells were used per condition. Chromatin was obtained as described in section 3.3.1 of chapter 2 and diluted 1:10 with ChIP buffer (Triton X‐100 1%, Sodium, Deoxycholate 0.1%, EDTA 1mM). It was spun down at 16 000 xg for 10 min and precleared for 2 hours at 4 using Protein A agarose beads. Meanwhile PU-beads (kindly provided by Chiosis Lab, Memorial℃ Sloan Kettering Cancer Center, New York, USA) were blocked with 0.5 % BSA supplemented with 1mg/ml yeast tRNA (Thermo Fisher, #15401- 029) for 2 hours and eventually added to chromatin. Following ON incubation at 4 (end- over-end rotation) beads were washed three times with 1X ChIP buffer and eluted with℃ 200ul elution buffer by heating at 65 °C for 45 min under vigorous shaking. Released DNA was process further as already described in section 3.3.2 of chapter 2 and subjected to sequencing.

Cellular fractionation, SDS-PAGE and Western blotting were performed as previously described in section 3 of chapter 1.

100

CHAPTER 2 RESULTS

4. Results

4.1. HCF-1 and Hsp90 share multiple common interactors

The affinity purification of nuclear Hsp90 followed by MS analysis of its binding partners (section 4.1 of chapter 1) revealed that HCF-1 physically interacts with the chaperone. At the same time, in a synthetic lethality screen we were also able to show that simultaneous deficiency of both the factors results in cell death (section 4.3 of chapter 1), thus implying a functional relationship between Hsp90 and HCF-1. Given the demonstrated physical and functional association, we wanted to know whether Hsp90 and HCF-1 share common interactors. For that purpose the publically available HCF-1 interactome published in 2017 [504] was compared to the nuclear Hsp90 interactome network revealed by us (section 4.1 of chapter 1). 14 common binding partners were identified (TABLE 8). Furthermore, a detailed HCF-1 literature search revealed additional shared interactors - transcription factors YY1 [471, 491] and Sp1 [364, 488], molecular chaperone Hsp70 [364], protein phosphatase 1 (PP1) [495], histone lysine demethylase PHF8 [510], and histone deacetylase Sin3A complex components (HDAC1/2, Sin3a, RBBP4/7, SAP45) [364]. Overall, this suggests that Hsp90 may support HCF-1 in establishing diverse interaction and points towards a general role of the chaperone in HCF-1 mode of action.

It is worth mentioning that the two chromatin modifying complexes that co-precipitated with nuclear Hsp90 (section 4.1 of chapter 1) – Set1 and Sin3A, have been shown to interact with HCF-1 in a cell-cycle specific manner [364, 487]. Furthermore, in a study by Wysocka et al., 2003 HCF-1 was shown to co-purify not only with components of the aforementioned complexes but Hsp90 and Hsp70 as well [364]. However, this lead was not a focus of their work and was not investigated.

An additional piece of data from a study by Taipale et al., 2014 came to strengthen our hypothesis of Hsp90/HCF-1 collaboration. In that work, the NudCD3 co-chaperone of Hsp90 was found to associate with HCF-1. The biological roles of the NudC family of co- chaperones are largely unknown [511]. NudCD3 has been proposed to exhibit a passive chaperone activity [512] – a feature that other client recruiters share. Thus, it could well be that NudCD3 binds HCF-1 and transfers it to Hsp90 for conformational stabilization. Furthermore, Taipale’s study established NUDCD3 co-chaperone as specific for the β- propeller Kelch domain and demonstrated that Hsp90 is able to bind Kelch domains. Thus, the Kelch domain of HCF-1 could be the one that facilitates interaction with the chaperone.

101

CHAPTER 2 RESULTS

In summary, Hsp90 and HCF-1 interaction networks comparison showed strong evidence to support potential interplay between the two factors.

Protein Symbol Protein Name Description Transcription Elongation CDK9 Cyclin-dependent kinase 9 P-TEF-b Complex PAF1 RNA polymerase II-associated factor 1 homolog PAF1 Complex CDC73 Parafibromin PAF1 Complex CTR9 RNA polymerase-associated protein CTR9 homolog PAF1 Complex SUPT16H FACT complex subunit SPT16 FACT Complex SSRP1 FACT complex subunit SSRP1 FACT Complex RNA Polymerase II subunits and Transcription Initiation factors POLR2A DNA-directed RNA polymerase II subunit RPB1 RNAPII subunit POLR2E DNA-directed RNA polymerases I, II, and III subunit RPABC1 RNAPII subunit TAF15 TATA-binding protein-associated factor 2N Transcription Initiation factor Chromatin Modification/Protein Modification UDP-N-acetylglucosamine--peptide N-acetylglucosaminyltransferase OGT O-Linked N-Acetylglucosamine Transferase 110 kDa subunit Set1A-B/COMPASS Complex/also present DPY-30 Protein dpy-30 homolog in other hCOMPASS like Complexes Set1A-B/COMPASS Complex/also present ASH2 Set1/Ash2 histone methyltransferase complex subunit ASH2 in other hCOMPASS like Complexes Set1A-B/COMPASS Complex/also present RBBP5 Retinoblastoma-binding protein 5 in other hCOMPASS like Complexes WDR82 WD repeat-containing protein 82 Set1A-B/COMPASS Complex Table 8. Proteins interacting with both Hsp90 and HCF-1. HCF-1 binding partners identified by Alfonso-Dunn et al., 2017 [504] were compared to nuclear Hsp90 interactome uncovered in section 4.1 of chapter 1.

4.2. HCF-1 is a client of Hsp90

Next we wanted to test whether HCF-1 is a novel client of Hsp90. A widely appreciated method of functional Hsp90-profiling is the use of Hsp90 inhibitors such as NVP-AUY922, which causes disruption of Hsp90-client association. As a consequence, clients become conformationaly unstable and are degraded [513].

We were able to demonstrate that Hsp90 inhibition in HEK cells leads to decrease in HCF-1 protein level in total nuclear extracts, nucleoplasm and chromatin (FIGURE 25). At the same time HCF-1 mRNA level was not affected. Notably, both HCF-1 precursor polypeptide and mature short subunits seemed to be affected, suggesting that Hsp90 provides stability to the precursor molecule as well. In support to this hypothesis, HCF-1 peptides identified by MS upon Hsp90 affinity purification (section 4.1 of chapter 1) originated from both N- and C-terminal regions. These results show that HCF-1 is indeed a client of Hsp90.

102

CHAPTER 2 RESULTS

Figure 25. HCF-1 is a client of Hsp90. Hsp90 pharmacological inactivation induced by NVP- AUY922 inhibitor leads to HCF-1 protein level decrease in total nuclear extracts (TNE), nucleoplasm and chromatin. SnRNP70 and H3 served as loading controls.

4.3. Hsp90 and HCF-1 co-localize at chromatin

In an attempt to investigate in details Hsp90-HCF-1 interplay and figure out whether it happens at chromatin level, we wanted to compare the genome-wide chromatin occupancy of both the factors. To reveal Hsp90 location at chromatin we employed different strategies: • ChIP-seq analysis in HEK and K562 cells - pull down of Hsp90-bound loci using different antibodies against endogenous Hsp90. • Chem-seq analysis in K562 cells - pull down of Hsp90-bound loci using agarose beads coupled to an inhibitor of the chaperone (PU-H71, a kind gift of Gabriela Chiosis) [207, 514, 515]. • ChIP-seq analysis in mouse embryonic stem cells – expression of a recombinant biotin- tagged Hsp90 and pull down of Hsp90-bound loci by Streptavidin beads affinity purification.

Unfortunately, we were not able to detect significant enrichment of Hsp90-bound loci with any method employed. Although frustrating, this does not come as a surprise due to Hsp90 low chromatin abundance, its notoriously unstable and transient interactions and importantly – its indirect binding to chromatin. Despite the finding that Hsp90 localizes at many coding and noncoding genes in Drosophila [293, 516], its genome-wide occupancy in mammalian cells still remains elusive. Indeed, to date there is just a single study that provides insights into Hsp90 localization at chromatin in mammalian cells [317]. Thus, given the issues encountered, we decided to use that publically available Hsp90 ChIP-seq dataset [317] and

103

CHAPTER 2 RESULTS

Figure 26. Hsp90 and HCF-1 co-localize at chromatin. (A) Overlap between publicly available Hsp90 and HCF-1 ChIP-seq peaks. (B) Overlaping peaks genomic location. (C) A screenshot example of peaks co-localization. (D) ChIP-seq datasets and RNA-seq upon Hsp90i heatmaps.

104

CHAPTER 2 RESULTS

compare it with HCF-1 ChIP-seq data by Deplus et al., 2013 [517]. The computational analysis of publicly available Hsp90 and HCF-1 ChIP-seq datasets revealed co-localization of these factors (FIGURE 26A). Around 30 % of Hsp90 peaks were found at HCF-1-occupied loci and most of them located at gene promoters (FIGURE 26B). An example is shown in FIGURE 26C.

Since HCF-1 does not bind DNA directly but gets recruited by a set of transcription factors, we wanted to know whether Hsp90 co-operates with any of them. For that purpose we compared publically available ChIP-seq dataset of selected HCF-1 interacting transcription factors - YY1, Sp1, E2F4 and ZNF143, with Hsp90. Notably, we were able to show co-localization of Hsp90, HCF-1 and the TFs mentioned above at particular chromatin loci shown as cluster 1 in FIGURE 26D.

4.4. HCF-1 chromatin binding and effect on gene expression is Hsp90-dependent

After demonstrating that Hsp90 and HCF-1 co-localize at chromatin we wanted to test whether Hsp90 depletion affects HCF-1 association with chromatin and its downstream effects. For that purpose we chose a locus that showed both Hsp90 and HCF-1 binding (PPRC1) and a control locus that is occupied by HCF-1 only (TRAPPC13). ChIP-qPCR data show that HCF-1 binding to the PPRC1 locus decreases upon Hsp90 inhibition. The control locus binding is not affected consistent with lack of Hsp90 association (FIGURE 27A).

To reveal the functional consequences of loss of HCF-1 binding upon Hsp90 inactivation, we performed RNA-seq analysis upon Hsp90 inhibition in K562 cells. Consistent with decrease of HCF-1 association, the PPRC1 locus showed decrease in transcription (FIGURE 27B) suggesting that HCF-1-mediated effect on its expression is Hsp90 dependent. Investigation of the RNA-seq data upon Hsp90 inhibition on global scale revealed that biological processes GO enrichment analysis associates the mis-regulated genes to cell proliferation and regulation of cell cycle. Motif analysis showed that the affected genes are predicted to possess binding sites for transcription factors such as E2Fs, Sp1, YY1, ZNF143 – all known to act in collaboration with HCF-1. Furthermore, Sp1 and YY1 were identified as Hsp90 interactors (section 4.1 of chapter 1). Finally, around 30 % of the genes that were misregulated upon Hsp90 inhibition proved to be HCF-1-bound (FIGURE 26D).

Overall, these results suggest that HCF-1 binding to chromatin and subsequent effect on gene expression are Hsp90-dependent and imply a role of nuclear Hsp90 in the modulation of chromatin dynamics via its interaction with HCF-1.

105

CHAPTER 2 RESULTS

Figure 27. HCF-1 association to chromatin and expression of corresponding genes is Hsp90- dependent. (A) HCF-1 ChIP-qPCR of the PPRC1 and TRAPPC13 loci in the absence (blue bar) or presence (grey bar) of Hsp90 inhibitor. Enrichment is calculated as percentage of the input. (B) PPRC1 expression level in the absence (blue) or presence (red) of Hsp90 inhibitor.

106

CHAPTER 2 DISCUSSION AND OUTLOOK

5. Discussion and Outlook

HCF-1 is a conserved and highly abundant transcripttional co-factor with established roles in cell-cycle specific and viral gene expression. It has been shown to act as a scaffold protein mediating the interaction between numerous transcription factors (GABP, Sp1, YY1, E2Fs, THAP11, etc.) and chromatin modifying/remodelling machineries (Sin3A HDAC, Set1 HMT, MOF HAT, PHF8 HDM, etc.) [461]. Here, we demonstrate that HCF-1 performs its gene expression regulation role supported by the molecular chaperone Hsp90. We could show that these two factors interact both physically and functionally and co-localize at chromatin. Furthermore, pharmacological inhibition of Hsp90 seems to destabilize HCF-1 and reduce its chromatin occupancy. To our excitement, a very recent report by Savitski et al., 2018 [406] confirms HCF-1 physical interaction with Hsp90 in MCF7, Jurkat and T-cells.

HCF-1 dependency on Hsp90 could be explained by its complicated structure that requires additional stabilization. In addition, it acts as a bridging platform between TF and proteins with chromatin modifying/ remodelling activities. It is likely that the structural complexity of such a platform necessitates chaperone action and control. Thus, it may require the assistance of a chaperone to ensure conformational stability and flexibility, appropriate association between N- and C-terminal subunits or correct chromatin targeting, for example.

HCF-1 precursor full-length polypeptide (HCF-1FL) as well as its N- and C-terminal subunits seems to be equally affected by Hsp90 inhibition. This could be potentially explained by three non-exclusive posibilities: (i) Hsp90 associates with HCF-1FL in the cellular cytosol and Hsp90i abolishes HCF-1FL nuclear entry thus leading to decrease in the nuclear level of all HCF-1 species, (ii) Hsp90 associates with HCF-1FL in the nucleus; Hsp90i leads to destabilization and degradation of HCF-1FL and reduces subunits level as a side effect or (iii) Hsp90 interacts with HCF-1FL in the nucleus prior to its proteolytic cleavage and subsequently remains associated with the generated subunits.

According to our data, HCF-1 and Hsp90 co-localize at chromatin demonstrating that their interplay does not happen at the level of nuclear import-related but rather at chromatin- associated activities. As for the second scenario, the scarce data on HCF-1FL function [477] makes it hard to be justified. It is known that the precursor molecule is unstable and is the least abundant of all HCF-1 species [454]. It is still contradictory whether and to which extent it localizes to chromatin and exerts any transcription regulation effects. On the other hand, HCF-1 subunits have confirmed chromatin localization and well-established role in gene

107

CHAPTER 2 DISCUSSION AND OUTLOOK

expression. Given that we could show HCF-1 and Hsp90 co-localization at chromatin and clear evidence that this co-localization is required for proper HCF-1-mediated regulation of gene expression, we suggest the third scenario as our model - Hsp90 interacts with HCF-1FL in the nucleus prior to its proteolytic cleavage and subsequently remains associated with the generated subunits. Molecular chaperones are generally defined as proteins that interact with, stabilize or help another protein to acquire its functionally active conformation, without being present in its final structure [4, 5]. However, there are exceptions demonstrating a more long- lasting relationship between chaperones and their clients. Hsp90, for instance, is required not only for telomerase protein conformational stabilization but remains associated with the enzyme during its reaction cycle to ensure sufficient activity [225-227]. Another example demonstrates that after inducing conformational changes that facilitate hormone binding later on, Hsp90 remains bound to estrogen receptors until they enter the nucleus [63]. Hsp90/HCF- 1 collaboration could prove to be yet another exception of the general definition. It could well be that the chaperone initially binds to HCF-1FL to support a conformation that would favour proteolytic cleavage as in the case of Picornavirus capsid precursor poly-protein (P1) [518]. Later on, Hsp90 could remain associated to the subunits that are non-covalently attached to ensure stability and flexibility. Intriguingly, the viral P1 polypeptide is not the only example to involve Hsp90 in scenarios of polypeptide proteolytic cleavage. Both the Trithorax and MLL histone methylation enzymes undergo proteolytic cleavage and function as heterodimers of the generated subunits are clients of the chaperone [304]. It will be intriguing to determine Hsp90, HCF-1FL, HCF-1N and HCF-1C genome-wide distribution by ChIP-seq analysis. That will bring clarity on which exactly is the HCF-1 species that Hsp90 interacts with at chromatin level. Furthermore, HCF-1 pulldown upon Hsp90i could answer key questions: (i) whether Hsp90 supports the interaction between HCF-1N and HCF-C (inability to detect HCF-1N upon HCF-1C immunoprecipitation and vice versa could prove the point) and (ii) whether Hsp90 is required for the interaction of HCF- with histone modifying/remodelling complexes (inability to detect components of corresponding complexes upon immunoprecipitation could prove the point).

According to published data [59], Hsp90 can bind Kelch domains. Intriguingly, a single proline-to-serine substitution in the same domain (P134S) causes cell proliferation arrest in tsBN67 cells when grown at 40 °C [458] (for details see section 1.5 of chapter 2). It is tempting to speculate that at physiological temperature Hsp90 binding to the mutated Kelch domain buffers the associated phenotypic variation and cells proliferate normally. When

108

CHAPTER 2 DISCUSSION AND OUTLOOK

shifted to 40 °C they experience heat shock that as described previously causes chaperones titration away from clients. Thus, the Kelch domain mutation is not buffered anymore which causes specific phenotype – conformational change in the domain that leads to inability to bind chromatin, mis-regulation of gene expression and eventually cell cycle arrest [359]. It will be curious to test whether Hsp90i at permissive temperature causes the same phenotype as growth at 40 °C. Hsp90 and HCF-1 inactivation phenotypes could be further compared in the context of cell cycle progression. As HCF-1 inhibition causes block in G1 to S phase transition and appearance of binucleated [458, 483], Hsp90i phenotype needs to be investigated in that respect. That could be achieved in a FACS experiment upon staining a particular cell proliferation marker (Ki67, for example) to detect the number of cells that have successfully entered S phase. Binucleation could be examined microscopically.

As a conclusion, our data proposes a model in which nuclear Hsp90 supports the abundant transcriptional co-factor HCF-1 in its transcriptional regulation activity. Hsp90 does so in two ways: (i) maintenance of HCF-1FL conformational state that favours proteolytic cleavage and (ii) stabilization of the resulting N- and C-terminal subunits heterodimer. These findings shed more light on HCF-1 mode of action. Furthermore, they provide an example of chromatin-associated action and transcription regulation role of Hsp90 – a chaperone that is traditionally considered to be cytosolic. Notably, the established HCF-1/ Hsp90 collaboration may prove a new opportunity for anti-cancer therapy - HCF-1 is overexpressed in certain tumors. Notably, HCF-1 is one of the 11 genes defined as “death-from-cancer” signature identified in a study of ~ 1000 cancer patients and Hsp90i could oppose the associated malignant transformation [519].

109

CONCLUDING REMARKS

Molecular chaperones are defined as factors that help other proteins to acquire functionally active conformation [4, 5]. Most of the chaperones bind nascent polypeptides that emerge directly from the ribosome and thus take part in the initial steps of protein folding. In addition, chaperones strongly contribute to cell survival and recovery upon proteotoxic stress - they carry out refolding of misfolded proteins, solubilization of aggregates and target terminally damaged proteins to degradation. Importantly, since abnormal proteostasis is often a cause or consequence of different disease states in human such as Alzheimer’s, Huntington’s, Parkinson’s and cancer, chaperones are found misregulated in several disorders [18-20]. The complex chaperone network (or the “chaperome”) in human is encoded by 332 genes [20]. Among them Hsp90 is a special chaperone. It does not take part in the early stage of protein folding but rather stabilizes late folding intermediates - partially folded and conformationally labile structures [62]. Furthermore, it does not bind every substrate sharing the above-mentioned characteristics but a restricted set of proteins collectively called clients - transcription factors, kinases, ubiquitin ligases, metabolic enzymes and others. It functions as part of a machinery involving numerous regulatory partners known as co-chaperones. They regulate the function of Hsp90 in different ways such as modulation of its ATPase activity and conformational cycle as well as recruitment of specific clients [110]. Notably, many Hsp90 clients (p53, c-Myc, v-Src, BCR-Abl, etc.) contribute to oncogenic transformation and maintenance of malignant phenotypes [174]. Thus, the chaperone has turned into an attractive target in cancer therapy.

Hsp90 comprises around 1 %–2 % of total protein in unstressed cells, and is one of the most abundant cellular proteins [26, 27]. The chaperone is traditionally viewed as a cytoplasmic factor but recent reports suggest important roles in nuclear events [280]. The poorly investigated nuclear Hsp90 pool has increasingly attracted attention and its comprehensive characterization will provide a further step in our understanding of the Hsp90 chaperone system.

In the first part of my thesis we characterized nuclear Hsp90 interactome and PTMs for the first time. Our data revealed that Hsp90 forms a complex network of interactors from the following main categories: (i) chaperones and co-chaperones, (ii) components of the protein degradation machinery, (iii) DNA repair proteins, (iv) RNA metabolism factors and (v) transcription regulators. We not only re-established previously known interactions but more

110

CONCLUDING REMARKS

importantly, we were able to identify a large number of potential novel clients. That demonstrated the advantage of the applied experimental approach towards revealing the interactions of nuclear Hsp90 exclusively (FIGURE 7). It will be intriguing to reveal which of the newly-discovered binding partners are direct interactors and thus genuine clients of the chaperone. To our excitement, we not only identified novel clients but we likely uncovered the first nucleus-specific co-chaperone – FKBP3. Further evidence is needed to confirm the nature of FKBP3 as a co-chaperone. Overall, Hsp90 interactome data provides a base for more detailed mechanistic and functional studies of nuclear Hsp90 – knowledge that would not only expand our understanding of nuclear biology and Hsp90 chaperone system, but might also lead to the design of alternative strategies for cancer treatment.

We were able to position the chaperone as part of the nuclear PQC network by revealing its interaction with components of the protein folding and degradation machineries. According to our data, Hsp90 interacts with several other chaperones (Hsp70, NPM1 and NPM3), co-chaperones (Cdc37 and PPP1CA), TRiC/CCT chaperonin, proteasome components and E3 protein ubiquitin ligases in the nucleus. These collaborations may ensure proteins welfare in the highly-dynamic nuclear environment. Apart from PQC our findings link Hsp90 to an important nuclear process, transcription regulation. The chaperone interacts with POLR2A and POLR2E subunits of RNA Pol II, components of the basic transcriptional machinery (such as TAF4 and TAF15), transcription elongation factors (components of PAF1 and FACT complexes), chromatin remodelling/modifying complexes (pBAF, Set1, Sin3A, CoREST, NuRD) and transcription factors/co-factors (HCF-1, Oct1, BCLAF1, DEK, etc.). Several of these physical interactors appear to also functionally interact with the chaperone according to the synthetic lethality screen that we performed (such as HCF-1, HDAC2, WDR82, PBRM1, BPTF, etc). All these leads promise to bring exciting details on Hsp90 role in gene expression regulation and chromatin biology. It will be interesting to investigate whether and how the chaperone contributes to rapid transcription fine-tuning in response to environmental stimuli. Following this line, the next exciting milestone to be reached will be uncovering Hsp90 genome-wide distribution in normal growth and stress conditions.

Various PTMs are known to fine-tune Hsp90 function [75, 76]. Within this work, we characterized nuclear Hsp90 PTMs in normal growth and stress conditions. Notably, this was the first investigation of Hsp90 PTMs upon proteotoxic stress.We could detect previously known phosphorylation and acetylation sites and reveal three novel methylation sites. In addition, we found out that proteotoxic stress changes Hsp90 methylation status since – R502

111

CONCLUDING REMARKS

appears upon stress only, K284 is modified exclusively in normal growth conditions and K550 is present regardless of the investigated conditions. Notably, the PTMs data obtained in this work can be exploited not only as a tool to investigate Hsp90 regulation mechanisms (ATPase activity, interaction with clients, subcellular distribution, etc.) but also opportunities of additional anti-cancer therapies. As already demonstrated [77, 81, 96, 99-101] altered modification status of Hsp90 could affect drug binding and confer cancer cells sensitivity to Hsp90i. Thus, Modulation of certain Hsp90 PTMs states could sensitize cancer cells to Hsp90 inhibitors.

In the second part of the thesis we selected an example from our nuclear interactome analysis of Hsp90 to confirm chromatin-associated action and transcription regulation role of the chaperone. HCF-1 is an abundant transcription regulator that recruits various histone remodelling/modifying complexes to chromatin – SET1A, Sin3A, NSL, etc. It is mostly known for its critical role in cell cycle progression. We were able to demonstrate that HCF-1 is a novel client of Hsp90 and both the factors co-localize at chromatin. Hsp90i significantly reduced nuclear HCF-1 protein level and binding at particular chromatin loci. These findings shed more light on HCF-1 mode of action and implied a role of nuclear Hsp90 in the modulation of chromatin dynamics via HCF-1. It will be intriguing to test how Hsp90i affects HCF-1 cell cycle-related function and whether HCF-1-overexpressing cancers could be targeted by inhibitors of Hsp90.

Part of this work was included in a manuscript submitted for publication in February, 2019 to the Molecular Cell journal (Antonova et al., 2019; see Appendix).

112

REFERENCES

1. Lodish H, B.A., Matsudaira P, Kaiser CA, Krieger M, Scott MP, Zipurksy SL, Darnell J Protein Structure and Function, in Molecular Cell Biology (5th ed.), N.Y.W.F.a. Company., Editor. 2004: https://www.ncbi.nlm.nih.gov/books/NBK21475/. 2. Kiefhaber, T., et al., Protein aggregation in vitro and in vivo: a quantitative model of the kinetic competition between folding and aggregation. Biotechnology (N Y), 1991. 9(9): p. 825-9. 3. Kim, Y.E., et al., Molecular chaperone functions in protein folding and proteostasis. Annu Rev Biochem, 2013. 82: p. 323-55. 4. Ellis, R.J., S.M. van der Vies, and S.M. Hemmingsen, The molecular chaperone concept. Biochem Soc Symp, 1989. 55: p. 145-53. 5. Hartl, F.U., Molecular chaperones in cellular protein folding. Nature, 1996. 381(6583): p. 571- 9. 6. Wegele, H., L. Muller, and J. Buchner, Hsp70 and Hsp90--a relay team for protein folding. Rev Physiol Biochem Pharmacol, 2004. 151: p. 1-44. 7. Vabulas, R.M., et al., Protein Folding in the Cytoplasm and the Heat Shock Response. Vol. 2. 2010. a004390. 8. Beissinger, M. and J. Buchner, How chaperones fold proteins. Biol Chem, 1998. 379(3): p. 245-59. 9. Kumar, C.M.S., S.C. Mande, and G. Mahajan, Multiple chaperonins in bacteria—novel functions and non-canonical behaviors. Cell Stress & Chaperones, 2015. 20(4): p. 555-574. 10. Morimoto, R.I., Cells in stress: transcriptional activation of heat shock genes. Science, 1993. 259(5100): p. 1409-10. 11. Stefani, M., Protein misfolding and aggregation: new examples in medicine and biology of the dark side of the protein world. Biochimica et Biophysica Acta (BBA) - Molecular Basis of Disease, 2004. 1739(1): p. 5-25. 12. Morimoto, R.I., Proteotoxic stress and inducible chaperone networks in neurodegenerative disease and aging. Genes Dev, 2008. 22(11): p. 1427-38. 13. Dai, C., S. Dai, and J. Cao, Proteotoxic stress of cancer: implication of the heat-shock response in oncogenesis. Journal of cellular physiology, 2012. 227(8): p. 2982-2987. 14. Ritossa, F., Discovery of the heat shock response. Cell Stress Chaperones, 1996. 1(2): p. 97-8. 15. Ritossa, F., A new puffing pattern induced by temperature shock and DNP in drosophila. Experientia, 1962. 18(12): p. 571-573. 16. Richter, K., M. Haslbeck, and J. Buchner, The heat shock response: life on the verge of death. Mol Cell, 2010. 40(2): p. 253-66. 17. Velichko, A.K., et al., Mechanisms of heat shock response in mammals. Cell Mol Life Sci, 2013. 70(22): p. 4229-41. 18. Brehme, M. and C. Voisine, Model systems of protein-misfolding diseases reveal chaperone modifiers of proteotoxicity. Disease Models & Mechanisms, 2016. 9(8): p. 823-838. 19. Hadizadeh Esfahani, A., et al., A systematic atlas of chaperome deregulation topologies across the human cancer landscape. PLOS Computational Biology, 2018. 14(1): p. e1005890. 20. Brehme, M., et al., A chaperome subnetwork safeguards proteostasis in aging and neurodegenerative disease. Cell Rep, 2014. 9(3): p. 1135-50. 21. Jeng, W., et al., Molecular chaperones: guardians of the proteome in normal and disease states. F1000Res, 2015. 4. 22. Balchin, D., M. Hayer-Hartl, and F.U. Hartl, In vivo aspects of protein folding and quality control. Science, 2016. 353(6294): p. aac4354. 23. Compton, J.L. and B.J. McCarthy, Induction of the Drosophila heat shock response in isolated polytene nuclei. Cell, 1978. 14(1): p. 191-201.

113

REFERENCES

24. Farrelly, F.W. and D.B. Finkelstein, Complete sequence of the heat shock-inducible HSP90 gene of Saccharomyces cerevisiae. J Biol Chem, 1984. 259(9): p. 5745-51. 25. Oppermann, H., W. Levinson, and J.M. Bishop, A cellular protein that associates with the transforming protein of Rous sarcoma virus is also a heat-shock protein. Proceedings of the National Academy of Sciences of the United States of America, 1981. 78(2): p. 1067-1071. 26. Lai, B.T., et al., Quantitation and intracellular localization of the 85K heat shock protein by using monoclonal and polyclonal antibodies. Mol Cell Biol, 1984. 4(12): p. 2802-10. 27. Welch, W.J. and J.R. Feramisco, Purification of the major mammalian heat shock proteins. J Biol Chem, 1982. 257(24): p. 14949-59. 28. Borkovich, K.A., et al., hsp82 is an essential protein that is required in higher concentrations for growth of cells at higher temperatures. Mol Cell Biol, 1989. 9(9): p. 3919-30. 29. Cutforth, T. and G.M. Rubin, Mutations in Hsp83 and cdc37 impair signaling by the sevenless receptor tyrosine kinase in Drosophila. Cell, 1994. 77(7): p. 1027-36. 30. Bardwell, J.C. and E.A. Craig, Ancient heat shock gene is dispensable. J Bacteriol, 1988. 170(7): p. 2977-83. 31. Spence, J., A. Cegielska, and C. Georgopoulos, Role of Escherichia coli heat shock proteins DnaK and HtpG (C62.5) in response to nutritional deprivation. J Bacteriol, 1990. 172(12): p. 7157-66. 32. Wayne, N. and D.N. Bolon, Dimerization of Hsp90 is required for in vivo function. Design and analysis of monomers and dimers. J Biol Chem, 2007. 282(48): p. 35386-95. 33. Prodromou, C., et al., Identification and structural characterization of the ATP/ADP-binding site in the Hsp90 molecular chaperone. Cell, 1997. 90(1): p. 65-75. 34. Jahn, M., et al., The charged linker of the molecular chaperone Hsp90 modulates domain contacts and biological function. Proc Natl Acad Sci U S A, 2014. 111(50): p. 17881-6. 35. Hainzl, O., et al., The charged linker region is an important regulator of Hsp90 function. J Biol Chem, 2009. 284(34): p. 22559-67. 36. Tsutsumi, S., et al., Charged linker sequence modulates eukaryotic heat shock protein 90 (Hsp90) chaperone activity. Proc Natl Acad Sci U S A, 2012. 109(8): p. 2937-42. 37. Harris, S.F., A.K. Shiau, and D.A. Agard, The crystal structure of the carboxy-terminal dimerization domain of htpG, the Escherichia coli Hsp90, reveals a potential substrate binding site. Structure, 2004. 12(6): p. 1087-97. 38. Buchner, J., Hsp90 & Co. - a holding for folding. Trends Biochem Sci, 1999. 24(4): p. 136-41. 39. Dutta, R. and M. Inouye, GHKL, an emergent ATPase/kinase superfamily. Trends Biochem Sci, 2000. 25(1): p. 24-8. 40. Pearl, L.H., Review: The HSP90 molecular chaperone—an enigmatic ATPase. Biopolymers, 2016. 105(8): p. 594-607. 41. Obermann, W.M., et al., In vivo function of Hsp90 is dependent on ATP binding and ATP hydrolysis. J Cell Biol, 1998. 143(4): p. 901-10. 42. Peng, X., et al., Heat shock protein 90 stabilization of ErbB2 expression is disrupted by ATP depletion in myocytes. J Biol Chem, 2005. 280(13): p. 13148-52. 43. Scheibel, T., et al., ATP-binding properties of human Hsp90. J Biol Chem, 1997. 272(30): p. 18608-13. 44. McLaughlin, S.H., H.W. Smith, and S.E. Jackson, Stimulation of the weak ATPase activity of human hsp90 by a client protein. J Mol Biol, 2002. 315(4): p. 787-98. 45. Shiau, A.K., et al., Structural Analysis of E. coli hsp90 reveals dramatic nucleotide-dependent conformational rearrangements. Cell, 2006. 127(2): p. 329-40. 46. Dollins, D.E., et al., Structures of GRP94-nucleotide complexes reveal mechanistic differences between the hsp90 chaperones. Mol Cell, 2007. 28(1): p. 41-56. 47. Hellenkamp, B., et al., Multidomain structure and correlated dynamics determined by self- consistent FRET networks. Nat Methods, 2017. 14(2): p. 174-180.

114

REFERENCES

48. Hessling, M., K. Richter, and J. Buchner, Dissection of the ATP-induced conformational cycle of the molecular chaperone Hsp90. Nat Struct Mol Biol, 2009. 16(3): p. 287-93. 49. Cunningham, C.N., K.A. Krukenberg, and D.A. Agard, Intra- and intermonomer interactions are required to synergistically facilitate ATP hydrolysis in Hsp90. J Biol Chem, 2008. 283(30): p. 21170-8. 50. Zierer, B.K., et al., Importance of cycle timing for the function of the molecular chaperone Hsp90. Nat Struct Mol Biol, 2016. 23(11): p. 1020-1028. 51. Moran Luengo, T., et al., Hsp90 Breaks the Deadlock of the Hsp70 Chaperone System. Mol Cell, 2018. 70(3): p. 545-552.e9. 52. Whitesell, L., et al., Inhibition of heat shock protein HSP90-pp60v-src heteroprotein complex formation by benzoquinone ansamycins: essential role for stress proteins in oncogenic transformation. Proc Natl Acad Sci U S A, 1994. 91(18): p. 8324-8. 53. Ding, G., et al., Regulation of Ubiquitin-like with Plant Homeodomain and RING Finger Domain 1 (UHRF1) Protein Stability by Heat Shock Protein 90 Chaperone Machinery. J Biol Chem, 2016. 291(38): p. 20125-35. 54. Galigniana, M.D., et al., Retrograde transport of the glucocorticoid receptor in neurites requires dynamic assembly of complexes with the protein chaperone hsp90 and is linked to the CHIP component of the machinery for proteasomal degradation. Molecular Brain Research, 2004. 123(1): p. 27-36. 55. Wang, X., et al., Geldanamycin-induced PCNA degradation in isolated Hsp90 complex from cancer cells. Cancer Invest, 2010. 28(6): p. 635-41. 56. Brugge, J.S., E. Erikson, and R.L. Erikson, The specific interaction of the Rous sarcoma virus transforming protein, pp60src, with two cellular proteins. Cell, 1981. 25(2): p. 363-72. 57. Sanchez, E.R., et al., Evidence that the 90-kDa phosphoprotein associated with the untransformed L-cell glucocorticoid receptor is a murine heat shock protein. J Biol Chem, 1985. 260(23): p. 12398-401. 58. Schuh, S., et al., A 90,000-dalton binding protein common to both steroid receptors and the Rous sarcoma virus transforming protein, pp60v-src. J Biol Chem, 1985. 260(26): p. 14292-6. 59. Taipale, M., et al., A quantitative chaperone interaction network reveals the architecture of cellular protein homeostasis pathways. Cell, 2014. 158(2): p. 434-448. 60. Taipale, M., D.F. Jarosz, and S. Lindquist, HSP90 at the hub of protein homeostasis: emerging mechanistic insights. Nat Rev Mol Cell Biol, 2010. 11(7): p. 515-28. 61. Schopf, F.H., M.M. Biebl, and J. Buchner, The HSP90 chaperone machinery. Nat Rev Mol Cell Biol, 2017. 18(6): p. 345-360. 62. Jakob, U., et al., Transient interaction of Hsp90 with early unfolding intermediates of citrate synthase. Implications for heat shock in vivo. J Biol Chem, 1995. 270(13): p. 7288-94. 63. Echeverria, P.C. and D. Picard, Molecular chaperones, essential partners of steroid hormone receptors for activity and mobility. Biochimica et Biophysica Acta (BBA) - Molecular Cell Research, 2010. 1803(6): p. 641-649. 64. Schlatter, H., et al., A novel function for the 90 kDa heat-shock protein (Hsp90): facilitating nuclear export of 60 S ribosomal subunits. Biochemical Journal, 2002. 362(Pt 3): p. 675-684. 65. Boulon, S., et al., HSP90 and its R2TP/Prefoldin-like cochaperone are involved in the cytoplasmic assembly of RNA polymerase II. Mol Cell, 2010. 39(6): p. 912-924. 66. Fang, Q., et al., HSP90 regulates DNA repair via the interaction between XRCC1 and DNA polymerase beta. Nat Commun, 2014. 5: p. 5513. 67. Xu, W., et al., Surface charge and hydrophobicity determine ErbB2 binding to the Hsp90 chaperone complex. Nat Struct Mol Biol, 2005. 12(2): p. 120-6. 68. Taipale, M., et al., Quantitative analysis of HSP90-client interactions reveals principles of substrate recognition. Cell, 2012. 150(5): p. 987-1001.

115

REFERENCES

69. Nathan, D.F., M.H. Vos, and S. Lindquist, In vivo functions of the Saccharomyces cerevisiae Hsp90 chaperone. Proceedings of the National Academy of Sciences, 1997. 94(24): p. 12949- 12956. 70. Rutherford, S.L. and S. Lindquist, Hsp90 as a capacitor for morphological evolution. Nature, 1998. 396(6709): p. 336-42. 71. Hummel, B., et al., The evolutionary capacitor HSP90 buffers the regulatory effects of mammalian endogenous retroviruses. Nat Struct Mol Biol, 2017. 24(3): p. 234-242. 72. Prodromou, C., Regulatory Mechanisms of Hsp90. Biochem Mol Biol J, 2017. 3(1): p. 2. 73. Voellmy, R. and F. Boellmann, Chaperone regulation of the heat shock protein response. Adv Exp Med Biol, 2007. 594: p. 89-99. 74. Neckers, L., et al., Methods to validate Hsp90 inhibitor specificity, to identify off-target effects, and to rethink approaches for further clinical development. Cell Stress Chaperones, 2018. 75. Prodromou, C., Mechanisms of Hsp90 regulation. Biochemical Journal, 2016. 473(16): p. 2439-2452. 76. Scroggins, B.T. and L. Neckers, Post-translational modification of heat-shock protein 90: impact on chaperone function. Expert Opinion on Drug Discovery, 2007. 2(10): p. 1403-1414. 77. Mollapour, M., et al., Casein kinase 2 phosphorylation of Hsp90 threonine 22 modulates chaperone function and drug sensitivity. Oncotarget, 2011. 2(5): p. 407-17. 78. Mollapour, M. and L. Neckers, Post-translational modifications of Hsp90 and their contributions to chaperone regulation. Biochimica et Biophysica Acta, 2012. 1823(3): p. 648- 655. 79. Mimnaugh, E.G., et al., Possible role for serine/threonine phosphorylation in the regulation of the heteroprotein complex between the hsp90 stress protein and the pp60v-src tyrosine kinase. J Biol Chem, 1995. 270(48): p. 28654-9. 80. Ogiso, H., et al., Phosphorylation analysis of 90 kDa heat shock protein within the cytosolic arylhydrocarbon receptor complex. Biochemistry, 2004. 43(49): p. 15510-9. 81. Wang, X., et al., The regulatory mechanism of Hsp90α secretion and its function in tumor malignancy. Proceedings of the National Academy of Sciences of the United States of America, 2009. 106(50): p. 21288-21293. 82. Mollapour, M., et al., Threonine 22 phosphorylation attenuates Hsp90 interaction with cochaperones and affects its chaperone activity. Mol Cell, 2011. 41(6): p. 672-81. 83. Barati, M.T., et al., A Proteomic Screen Identified Stress-Induced Chaperone Proteins as Targets of Akt Phosphorylation in Mesangial Cells. Journal of proteome research, 2006. 5(7): p. 1636-1646. 84. Old, W.M., et al., Functional Proteomics Identifies Targets of Phosphorylation by B-Raf Signaling in Melanoma. Molecular cell, 2009. 34(1): p. 115-131. 85. Duval, M., et al., Src-mediated Phosphorylation of Hsp90 in Response to Vascular Endothelial Growth Factor (VEGF) Is Required for VEGF Receptor-2 Signaling to Endothelial NO Synthase. Molecular Biology of the Cell, 2007. 18(11): p. 4659-4668. 86. Ciccia, A. and S.J. Elledge, The DNA Damage Response: Making it safe to play with knives. Molecular cell, 2010. 40(2): p. 179-204. 87. Bennetzen, M.V., et al., Site-specific phosphorylation dynamics of the nuclear proteome during the DNA damage response. Mol Cell Proteomics, 2010. 9(6): p. 1314-23. 88. Matsuoka, S., et al., ATM and ATR substrate analysis reveals extensive protein networks responsive to DNA damage. Science, 2007. 316(5828): p. 1160-6. 89. Quanz, M., et al., Heat Shock Protein 90α (Hsp90α) Is Phosphorylated in Response to DNA Damage and Accumulates in Repair Foci. The Journal of Biological Chemistry, 2012. 287(12): p. 8803-8815.

116

REFERENCES

90. Solier, S., et al., Heat shock protein 90alpha (HSP90alpha), a substrate and chaperone of DNA-PK necessary for the apoptotic response. Proc Natl Acad Sci U S A, 2012. 109(32): p. 12866-72. 91. Legagneux, V., M. Morange, and O. Bensaude, Heat shock increases turnover of 90 kDa heat shock protein phosphate groups in HeLa cells. FEBS Letters, 1991. 291(2): p. 359-362. 92. Mollapour, M., S. Tsutsumi, and L. Neckers, Hsp90 phosphorylation, Wee1 and the cell cycle. Cell Cycle, 2010. 9(12): p. 2310-2316. 93. Lei, H., et al., Protein Kinase A-dependent Translocation of Hsp90α Impairs Endothelial Nitric- oxide Synthase Activity in High Glucose and Diabetes. Journal of Biological Chemistry, 2007. 282(13): p. 9364-9371. 94. Cheng, C.F., et al., Transforming growth factor alpha (TGFalpha)-stimulated secretion of HSP90alpha: using the receptor LRP-1/CD91 to promote human skin cell migration against a TGFbeta-rich environment during wound healing. Mol Cell Biol, 2008. 28(10): p. 3344-58. 95. Tsutsumi, S., et al., A small molecule cell-impermeant Hsp90 antagonist inhibits tumor cell motility and invasion. Oncogene, 2008. 27(17): p. 2478-87. 96. Mollapour, M., et al., Swe1Wee1-dependent tyrosine phosphorylation of Hsp90 regulates distinct facets of chaperone function. Mol Cell, 2010. 37(3): p. 333-43. 97. Lees-Miller, S.P. and C.W. Anderson, Two human 90-kDa heat shock proteins are phosphorylated in vivo at conserved serines that are phosphorylated in vitro by casein kinase II. J Biol Chem, 1989. 264(5): p. 2431-7. 98. Kurokawa, M., et al., Inhibition of apoptosome formation by suppression of Hsp90beta phosphorylation in tyrosine kinase-induced leukemias. Mol Cell Biol, 2008. 28(17): p. 5494- 506. 99. Beebe, K., et al., Posttranslational modification and conformational state of heat shock protein 90 differentially affect binding of chemically diverse small molecule inhibitors. Oncotarget, 2013. 4(7): p. 1065-74. 100. Mollapour, M., S. Tsutsumi, and L. Neckers, Hsp90 phosphorylation, Wee1 and the cell cycle. Cell Cycle, 2010. 9(12): p. 2310-6. 101. Woodford, M.R., et al., Mps1 Mediated Phosphorylation of Hsp90 Confers Renal Cell Carcinoma Sensitivity and Selectivity to Hsp90 Inhibitors. Cell Reports. 14(4): p. 872-884. 102. Woodford, M.R., et al., Impact of Posttranslational Modifications on the Anticancer Activity of Hsp90 Inhibitors. Adv Cancer Res, 2016. 129: p. 31-50. 103. Yu, X., et al., Modulation of p53, ErbB1, ErbB2, and Raf-1 expression in lung cancer cells by depsipeptide FR901228. J Natl Cancer Inst, 2002. 94(7): p. 504-13. 104. Scroggins, B.T., et al., An acetylation site in the middle domain of Hsp90 regulates chaperone function. Mol Cell, 2007. 25(1): p. 151-9. 105. Yang, Y., et al., Role of acetylation and extracellular location of heat shock protein 90alpha in tumor cell invasion. Cancer Res, 2008. 68(12): p. 4833-42. 106. Ai, J., et al., HDAC6 regulates androgen receptor hypersensitivity and nuclear localization via modulating Hsp90 acetylation in castration-resistant prostate cancer. Mol Endocrinol, 2009. 23(12): p. 1963-72. 107. Abu-Farha, M., et al., Proteomic analyses of the SMYD family interactomes identify HSP90 as a novel target for SMYD2. J Mol Cell Biol, 2011. 3(5): p. 301-8. 108. Hamamoto, R., et al., SMYD2-dependent HSP90 methylation promotes cancer cell proliferation by regulating the chaperone complex formation. Cancer Lett, 2014. 351(1): p. 126-33. 109. Donlin, L.T., et al., Smyd2 controls cytoplasmic lysine methylation of Hsp90 and myofilament organization. Genes Dev, 2012. 26(2): p. 114-9. 110. Caplan, A.J., What is a co-chaperone? Cell Stress & Chaperones, 2003. 8(2): p. 105-107. 111. Li, J., J. Soroka, and J. Buchner, The Hsp90 chaperone machinery: conformational dynamics and regulation by co-chaperones. Biochim Biophys Acta, 2012. 1823(3): p. 624-35.

117

REFERENCES

112. Li, J., K. Richter, and J. Buchner, Mixed Hsp90-cochaperone complexes are important for the progression of the reaction cycle. Nat Struct Mol Biol, 2011. 18(1): p. 61-6. 113. Blatch, G.L. and M. Lassle, The tetratricopeptide repeat: a structural motif mediating protein- protein interactions. Bioessays, 1999. 21(11): p. 932-9. 114. Shirasu, K., et al., A novel class of eukaryotic zinc-binding proteins is required for disease resistance signaling in barley and development in C. elegans. Cell, 1999. 99(4): p. 355-66. 115. Ratzke, C., et al., From a Ratchet Mechanism to Random Fluctuations Evolution of Hsp90's Mechanochemical Cycle. Journal of Molecular Biology, 2012. 423(3): p. 462-471. 116. Zuehlke, A.D., et al., An Hsp90 co-chaperone protein in yeast is functionally replaced by site- specific posttranslational modification in humans. Nature Communications, 2017. 8: p. 15328. 117. Richter, K., et al., Sti1 is a non-competitive inhibitor of the Hsp90 ATPase. Binding prevents the N-terminal dimerization reaction during the atpase cycle. J Biol Chem, 2003. 278(12): p. 10328-33. 118. Wegele, H., et al., Substrate transfer from the chaperone Hsp70 to Hsp90. J Mol Biol, 2006. 356(3): p. 802-11. 119. Alvira, S., et al., Structural characterization of the substrate transfer mechanism in Hsp70/Hsp90 folding machinery mediated by Hop. Nat Commun, 2014. 5: p. 5484. 120. Beraldo, F.H., et al., Stress-inducible phosphoprotein 1 has unique cochaperone activity during development and regulates cellular response to ischemia via the prion protein. Faseb j, 2013. 27(9): p. 3594-607. 121. Roe, S.M., et al., The Mechanism of Hsp90 regulation by the protein kinase-specific cochaperone p50(cdc37). Cell, 2004. 116(1): p. 87-98. 122. Kimura, Y., et al., Cdc37 is a molecular chaperone with specific functions in signal transduction. Genes Dev, 1997. 11(14): p. 1775-85. 123. Wandinger, S.K., et al., The phosphatase Ppt1 is a dedicated regulator of the molecular chaperone Hsp90. Embo j, 2006. 25(2): p. 367-76. 124. Vaughan, C.K., et al., Hsp90-dependent activation of protein kinases is regulated by chaperone-targeted dephosphorylation of Cdc37. Mol Cell, 2008. 31(6): p. 886-95. 125. Soroka, J., et al., Conformational switching of the molecular chaperone Hsp90 via regulated phosphorylation. Mol Cell, 2012. 45(4): p. 517-28. 126. Kang, C.B., et al., FKBP family proteins: immunophilins with versatile biological functions. Neurosignals, 2008. 16(4): p. 318-25. 127. Bose, S., et al., Chaperone function of Hsp90-associated proteins. Science, 1996. 274(5293): p. 1715-7. 128. Panaretou, B., et al., Activation of the ATPase activity of hsp90 by the stress-regulated cochaperone aha1. Mol Cell, 2002. 10(6): p. 1307-18. 129. Retzlaff, M., et al., Asymmetric activation of the hsp90 dimer by its cochaperone aha1. Mol Cell, 2010. 37(3): p. 344-54. 130. Li, J., et al., Integration of the accelerator Aha1 in the Hsp90 co-chaperone cycle. Nat Struct Mol Biol, 2013. 20(3): p. 326-31. 131. Lorenz, O.R., et al., Modulation of the Hsp90 chaperone cycle by a stringent client protein. Mol Cell, 2014. 53(6): p. 941-53. 132. Richter, K., S. Walter, and J. Buchner, The Co-chaperone Sba1 connects the ATPase reaction of Hsp90 to the progression of the chaperone cycle. J Mol Biol, 2004. 342(5): p. 1403-13. 133. Johnson, J.L. and D.O. Toft, A novel chaperone complex for steroid receptors involving heat shock proteins, immunophilins, and p23. J Biol Chem, 1994. 269(40): p. 24989-93. 134. Weikl, T., K. Abelmann, and J. Buchner, An unstructured C-terminal region of the Hsp90 co- chaperone p23 is important for its chaperone function. J Mol Biol, 1999. 293(3): p. 685-91.

118

REFERENCES

135. Echtenkamp, F.J., et al., Hsp90 and p23 Molecular Chaperones Control Chromatin Architecture by Maintaining the Functional Pool of the RSC Chromatin Remodeler. Mol Cell, 2016. 64(5): p. 888-899. 136. Zelin, E., et al., The p23 molecular chaperone and GCN5 acetylase jointly modulate protein- DNA dynamics and open chromatin status. Mol Cell, 2012. 48(3): p. 459-70. 137. Echtenkamp, F.J., et al., Global functional map of the p23 molecular chaperone reveals an extensive cellular network. Mol Cell, 2011. 43(2): p. 229-41. 138. Zhao, R., et al., Navigating the chaperone network: an integrative map of physical and genetic interactions mediated by the hsp90 chaperone. Cell, 2005. 120(5): p. 715-27. 139. Te, J., et al., Novel subunits of the mammalian Hsp90 signal transduction chaperone. J Proteome Res, 2007. 6(5): p. 1963-73. 140. Boulon, S., E. Bertrand, and B. Pradet-Balade, HSP90 and the R2TP co-chaperone complex: building multi-protein machineries essential for cell growth and gene expression. RNA Biol, 2012. 9(2): p. 148-54. 141. Kakihara, Y. and W.A. Houry, The R2TP complex: discovery and functions. Biochim Biophys Acta, 2012. 1823(1): p. 101-7. 142. Eckert, K., et al., The Pih1-Tah1 Cochaperone Complex Inhibits Hsp90 Molecular Chaperone ATPase Activity. The Journal of Biological Chemistry, 2010. 285(41): p. 31304-31312. 143. Siligardi, G., et al., Regulation of Hsp90 ATPase activity by the co-chaperone Cdc37p/p50cdc37. J Biol Chem, 2002. 277(23): p. 20151-9. 144. Prodromou, C., et al., Regulation of Hsp90 ATPase activity by tetratricopeptide repeat (TPR)- domain co-chaperones. The EMBO Journal, 1999. 18(3): p. 754-762. 145. Rivera-Calzada, A., et al., The Structure of the R2TP Complex Defines a Platform for Recruiting Diverse Client Proteins to the HSP90 Molecular Chaperone System. Structure. 25(7): p. 1145- 1152.e4. 146. Connell, P., et al., The co-chaperone CHIP regulates protein triage decisions mediated by heat-shock proteins. Nat Cell Biol, 2001. 3(1): p. 93-6. 147. Picard, D., Chaperoning steroid hormone action. Trends Endocrinol Metab, 2006. 17(6): p. 229-35. 148. Echeverria, P.C., P.A. Briand, and D. Picard, A Remodeled Hsp90 Molecular Chaperone Ensemble with the Novel Cochaperone Aarsd1 Is Required for Muscle Differentiation. Mol Cell Biol, 2016. 36(8): p. 1310-21. 149. Sreedhar, A.S., et al., Hsp90 isoforms: functions, expression and clinical importance. FEBS Lett, 2004. 562(1-3): p. 11-5. 150. Grad, I., et al., The molecular chaperone Hsp90alpha is required for meiotic progression of spermatocytes beyond pachytene in the mouse. PLoS One, 2010. 5(12): p. e15770. 151. Gruppi, C.M., Z.F. Zakeri, and D.J. Wolgemuth, Stage and lineage-regulated expression of two hsp90 transcripts during mouse germ cell differentiation and embryogenesis. Mol Reprod Dev, 1991. 28(3): p. 209-17. 152. Gruppi, C.M. and D.J. Wolgemuth, HSP86 and HSP84 exhibit cellular specificity of expression and co-precipitate with an HSP70 family member in the murine testis. Dev Genet, 1993. 14(2): p. 119-26. 153. Krone, P.H. and J.B. Sass, HSP 90 alpha and HSP 90 beta genes are present in the zebrafish and are differentially regulated in developing embryos. Biochem Biophys Res Commun, 1994. 204(2): p. 746-52. 154. Meng, X., et al., Cloning of chicken hsp90 beta: the only vertebrate hsp90 insensitive to heat shock. Biochem Biophys Res Commun, 1993. 190(2): p. 630-6. 155. Taherian, A., P.H. Krone, and N. Ovsenek, A comparison of Hsp90alpha and Hsp90beta interactions with cochaperones and substrates. Biochem Cell Biol, 2008. 86(1): p. 37-45. 156. Bouchier-Hayes, L., et al., Characterization of cytoplasmic caspase-2 activation by induced proximity. Mol Cell, 2009. 35(6): p. 830-40.

119

REFERENCES

157. Eustace, B.K., et al., Functional proteomic screens reveal an essential extracellular role for hsp90 alpha in cancer cell invasiveness. Nat Cell Biol, 2004. 6(6): p. 507-14. 158. Voss, A.K., T. Thomas, and P. Gruss, Mice lacking HSP90beta fail to develop a placental labyrinth. Development, 2000. 127(1): p. 1. 159. Kajiwara, C., et al., Spermatogenesis arrest caused by conditional deletion of Hsp90alpha in adult mice. Biol Open, 2012. 1(10): p. 977-82. 160. Ichiyanagi, T., et al., HSP90α plays an important role in piRNA biogenesis and retrotransposon repression in mouse. Nucleic Acids Research, 2014. 42(19): p. 11903-11911. 161. Shiu, R.P., J. Pouyssegur, and I. Pastan, Glucose depletion accounts for the induction of two transformation-sensitive membrane proteinsin Rous sarcoma virus-transformed chick embryo fibroblasts. Proc Natl Acad Sci U S A, 1977. 74(9): p. 3840-4. 162. Sorger, P.K. and H.R. Pelham, The glucose-regulated protein grp94 is related to heat shock protein hsp90. J Mol Biol, 1987. 194(2): p. 341-4. 163. Ansa-Addo, E.A., et al., Clients and Oncogenic Roles of Molecular Chaperone gp96/grp94. Curr Top Med Chem, 2016. 16(25): p. 2765-78. 164. Marzec, M., D. Eletto, and Y. Argon, GRP94: An HSP90-like protein specialized for protein folding and quality control in the endoplasmic reticulum. Biochim Biophys Acta, 2012. 1823(3): p. 774-87. 165. Rosenbaum, M., et al., MZB1 is a GRP94 cochaperone that enables proper immunoglobulin heavy chain biosynthesis upon ER stress. Genes Dev, 2014. 28(11): p. 1165-78. 166. Liu, B., et al., Folding of Toll-like receptors by the HSP90 paralogue gp96 requires a substrate- specific cochaperone. Nat Commun, 2010. 1: p. 79. 167. Song, H.Y., et al., Identification of a protein with homology to hsp90 that binds the type 1 tumor necrosis factor receptor. J Biol Chem, 1995. 270(8): p. 3574-81. 168. Altieri, D.C., et al., TRAP-1, the mitochondrial Hsp90. Biochim Biophys Acta, 2012. 1823(3): p. 767-73. 169. Felts, S.J., et al., The hsp90-related protein TRAP1 is a mitochondrial protein with distinct functional properties. J Biol Chem, 2000. 275(5): p. 3305-12. 170. Masgras, I., et al., The Chaperone TRAP1 As a Modulator of the Mitochondrial Adaptations in Cancer Cells. Front Oncol, 2017. 7: p. 58. 171. Lettini, G., et al., TRAP1: a viable therapeutic target for future cancer treatments? Expert Opin Ther Targets, 2017. 21(8): p. 805-815. 172. Seo, Y.H., Organelle-specific Hsp90 inhibitors. Arch Pharm Res, 2015. 38(9): p. 1582-90. 173. Hanahan, D. and Robert A. Weinberg, Hallmarks of Cancer: The Next Generation. Cell. 144(5): p. 646-674. 174. Vartholomaiou, E., P.C. Echeverria, and D. Picard, Unusual Suspects in the Twilight Zone Between the Hsp90 Interactome and Carcinogenesis. Adv Cancer Res, 2016. 129: p. 1-30. 175. Falsone, S.F., et al., Oncogenic mutations reduce the stability of SRC kinase. J Mol Biol, 2004. 344(1): p. 281-91. 176. Pick, E., et al., High HSP90 expression is associated with decreased survival in breast cancer. Cancer Res, 2007. 67(7): p. 2932-7. 177. Alexandrova, E.M., et al., Improving survival by exploiting tumour dependence on stabilized mutant p53 for treatment. Nature, 2015. 523(7560): p. 352-6. 178. Trepel, J., et al., Targeting the dynamic HSP90 complex in cancer. Nat Rev Cancer, 2010. 10(8): p. 537-49. 179. Yuno, A., et al., Clinical Evaluation and Biomarker Profiling of Hsp90 Inhibitors. Methods Mol Biol, 2018. 1709: p. 423-441. 180. Chiosis, G. and L. Neckers, Tumor selectivity of Hsp90 inhibitors: the explanation remains elusive. ACS Chem Biol, 2006. 1(5): p. 279-84. 181. Kamal, A., et al., A high-affinity conformation of Hsp90 confers tumour selectivity on Hsp90 inhibitors. Nature, 2003. 425(6956): p. 407-10.

120

REFERENCES

182. Rodina, A., et al., The epichaperome is an integrated chaperome network that facilitates tumour survival. Nature, 2016. 538(7625): p. 397-401. 183. Soga, S., S. Akinaga, and Y. Shiotsu, Hsp90 inhibitors as anti-cancer agents, from basic discoveries to clinical development. Curr Pharm Des, 2013. 19(3): p. 366-76. 184. Bhat, R., S.R. Tummalapalli, and D.P. Rotella, Progress in the discovery and development of heat shock protein 90 (Hsp90) inhibitors. J Med Chem, 2014. 57(21): p. 8718-28. 185. Sidera, K. and E. Patsavoudi, HSP90 inhibitors: current development and potential in cancer therapy. Recent Pat Anticancer Drug Discov, 2014. 9(1): p. 1-20. 186. Chiosis, G., et al., Development of a purine-scaffold novel class of Hsp90 binders that inhibit the proliferation of cancer cells and induce the degradation of Her2 tyrosine kinase. Bioorg Med Chem, 2002. 10(11): p. 3555-64. 187. Garg, G., A. Khandelwal, and B.S. Blagg, Anticancer Inhibitors of Hsp90 Function: Beyond the Usual Suspects. Adv Cancer Res, 2016. 129: p. 51-88. 188. Marcu, M.G., et al., The heat shock protein 90 antagonist novobiocin interacts with a previously unrecognized ATP-binding domain in the carboxyl terminus of the chaperone. J Biol Chem, 2000. 275(47): p. 37181-6. 189. Whitesell, L. and S.L. Lindquist, HSP90 and the chaperoning of cancer. Nat Rev Cancer, 2005. 5(10): p. 761-72. 190. Zismanov, V., L. Drucker, and M. Gottfried, Combined inhibition of Hsp90 and the proteasome affects NSCLC proteostasis and attenuates cell migration. Anticancer Drugs, 2014. 25(9): p. 998-1006. 191. Cavanaugh, A., et al., Combined inhibition of heat shock proteins 90 and 70 leads to simultaneous degradation of the oncogenic signaling proteins involved in muscle invasive bladder cancer. Oncotarget, 2015. 6(37): p. 39821-39838. 192. Chakraborty, S.N., et al., Combination of JAK2 and HSP90 inhibitors: an effective therapeutic option in drug-resistant chronic myelogenous leukemia. Genes & Cancer, 2016. 7(5-6): p. 201-208. 193. Iwai, A., et al., Combined inhibition of Wee1 and Hsp90 activates intrinsic apoptosis in cancer cells. Cell Cycle, 2012. 11(19): p. 3649-55. 194. Krämer, O.H., S. Mahboobi, and A. Sellmer, Drugging the HDAC6–HSP90 interplay in malignant cells. Trends in Pharmacological Sciences. 35(10): p. 501-509. 195. Mellacheruvu, D., et al., The CRAPome: a Contaminant Repository for Affinity Purification Mass Spectrometry Data. Nature methods, 2013. 10(8): p. 730-736. 196. Prodromou, C. and L.H. Pearl, Structure and functional relationships of Hsp90. Curr Cancer Drug Targets, 2003. 3(5): p. 301-23. 197. Lipsich, L.A., J.R. Cutt, and J.S. Brugge, Association of the transforming proteins of Rous, Fujinami, and Y73 avian sarcoma viruses with the same two cellular proteins. Mol Cell Biol, 1982. 2(7): p. 875-80. 198. Joab, I., et al., Common non-hormone binding component in non-transformed chick oviduct receptors of four steroid hormones. Nature, 1984. 308(5962): p. 850-3. 199. Sreedhar, A.S., C. Soti, and P. Csermely, Inhibition of Hsp90: a new strategy for inhibiting protein kinases. Biochim Biophys Acta, 2004. 1697(1-2): p. 233-42. 200. Millson, S.H., et al., A two-hybrid screen of the yeast proteome for Hsp90 interactors uncovers a novel Hsp90 chaperone requirement in the activity of a stress-activated mitogen- activated protein kinase, Slt2p (Mpk1p). Eukaryot Cell, 2005. 4(5): p. 849-60. 201. Falsone, S.F., et al., A proteomic snapshot of the human heat shock protein 90 interactome. FEBS Lett, 2005. 579(28): p. 6350-4. 202. Skarra, D.V., et al., Label-free quantitative proteomics and SAINT analysis enable interactome mapping for the human Ser/Thr protein phosphatase 5. Proteomics, 2011. 11(8): p. 1508-16. 203. Tsaytler, P.A., et al., Novel Hsp90 partners discovered using complementary proteomic approaches. Cell Stress Chaperones, 2009. 14(6): p. 629-38.

121

REFERENCES

204. Gano, J.J. and J.A. Simon, A proteomic investigation of ligand-dependent HSP90 complexes reveals CHORDC1 as a novel ADP-dependent HSP90-interacting protein. Mol Cell Proteomics, 2010. 9(2): p. 255-70. 205. Gopinath, R.K., et al., The Hsp90-dependent proteome is conserved and enriched for hub proteins with high levels of protein-protein connectivity. Genome Biol Evol, 2014. 6(10): p. 2851-65. 206. Gong, Y., et al., An atlas of chaperone-protein interactions in Saccharomyces cerevisiae: implications to protein folding pathways in the cell. Mol Syst Biol, 2009. 5: p. 275. 207. Moulick, K., et al., Affinity-based proteomics reveal cancer-specific networks coordinated by Hsp90. Nat Chem Biol, 2011. 7(11): p. 818-26. 208. Makhnevych, T. and W.A. Houry, The role of Hsp90 in protein complex assembly. Biochim Biophys Acta, 2012. 1823(3): p. 674-82. 209. Gonzales, F.A., et al., Characterization of Saccharomyces cerevisiae Nop17p, a novel Nop58p- interacting protein that is involved in Pre-rRNA processing. J Mol Biol, 2005. 346(2): p. 437- 55. 210. Zhao, R., et al., Molecular chaperone Hsp90 stabilizes Pih1/Nop17 to maintain R2TP complex activity that regulates snoRNA accumulation. J Cell Biol, 2008. 180(3): p. 563-78. 211. Granato, D.C., et al., Nop53p, an essential nucleolar protein that interacts with Nop17p and Nip7p, is required for pre-rRNA processing in Saccharomyces cerevisiae. Febs j, 2005. 272(17): p. 4450-63. 212. Boulon, S., et al., The Hsp90 chaperone controls the biogenesis of L7Ae RNPs through conserved machinery. J Cell Biol, 2008. 180(3): p. 579-95. 213. Cloutier, P., et al., R2TP/Prefoldin-like component RUVBL1/RUVBL2 directly interacts with ZNHIT2 to regulate assembly of U5 small nuclear ribonucleoprotein. Nature Communications, 2017. 8: p. 15615. 214. Bizarro, J., et al., NUFIP and the HSP90/R2TP chaperone bind the SMN complex and facilitate assembly of U4-specific proteins. Nucleic Acids Res, 2015. 43(18): p. 8973-89. 215. Malinova, A., et al., Assembly of the U5 snRNP component PRPF8 is controlled by the HSP90/R2TP chaperones. J Cell Biol, 2017. 216(6): p. 1579-1596. 216. Iwasaki, S., et al., Hsc70/Hsp90 chaperone machinery mediates ATP-dependent RISC loading of small RNA duplexes. Mol Cell, 2010. 39(2): p. 292-9. 217. Johnston, M., et al., HSP90 protein stabilizes unloaded argonaute complexes and microscopic P-bodies in human cells. Mol Biol Cell, 2010. 21(9): p. 1462-9. 218. Miyoshi, T., et al., A direct role for Hsp90 in pre-RISC formation in Drosophila. Nat Struct Mol Biol, 2010. 17(8): p. 1024-6. 219. Pare, J.M., et al., Hsp90 regulates the function of argonaute 2 and its recruitment to stress granules and P-bodies. Mol Biol Cell, 2009. 20(14): p. 3273-84. 220. Suzuki, Y., et al., The Hsp90 inhibitor geldanamycin abrogates colocalization of eIF4E and eIF4E-transporter into stress granules and association of eIF4E with eIF4G. J Biol Chem, 2009. 284(51): p. 35597-604. 221. Takai, H., et al., Tel2 structure and function in the Hsp90-dependent maturation of mTOR and ATR complexes. Genes Dev, 2010. 24(18): p. 2019-30. 222. Izumi, N., et al., Heat shock protein 90 regulates phosphatidylinositol 3-kinase-related protein kinase family proteins together with the RUVBL1/2 and Tel2-containing co-factor complex. Cancer Sci, 2012. 103(1): p. 50-7. 223. Rodrigo-Brenni, M.C., et al., Sgt1p and Skp1p modulate the assembly and turnover of CBF3 complexes required for proper kinetochore function. Mol Biol Cell, 2004. 15(7): p. 3366-78. 224. Lingelbach, L.B. and K.B. Kaplan, The interaction between Sgt1p and Skp1p is regulated by HSP90 chaperones and is required for proper CBF3 assembly. Mol Cell Biol, 2004. 24(20): p. 8938-50.

122

REFERENCES

225. Holt, S.E., et al., Functional requirement of p23 and Hsp90 in telomerase complexes. Genes Dev, 1999. 13(7): p. 817-26. 226. Forsythe, H.L., et al., Stable association of hsp90 and p23, but Not hsp70, with active human telomerase. J Biol Chem, 2001. 276(19): p. 15571-4. 227. Keppler, B.R., A.T. Grady, and M.B. Jarstfer, The biochemical role of the heat shock protein 90 chaperone complex in establishing human telomerase activity. J Biol Chem, 2006. 281(29): p. 19840-8. 228. Lee, J.H. and I.K. Chung, Curcumin inhibits nuclear localization of telomerase by dissociating the Hsp90 co-chaperone p23 from hTERT. Cancer Lett, 2010. 290(1): p. 76-86. 229. DeZwaan, D.C. and B.C. Freeman, HSP90 manages the ends. Trends Biochem Sci, 2010. 35(7): p. 384-91. 230. Imai, J., et al., The molecular chaperone Hsp90 plays a role in the assembly and maintenance of the 26S proteasome. Embo j, 2003. 22(14): p. 3557-67. 231. Yamano, T., et al., Hsp90-mediated assembly of the 26 S proteasome is involved in major histocompatibility complex class I antigen processing. J Biol Chem, 2008. 283(42): p. 28060-5. 232. Young, J.C., N.J. Hoogenraad, and F.U. Hartl, Molecular chaperones Hsp90 and Hsp70 deliver preproteins to the mitochondrial import receptor Tom70. Cell, 2003. 112(1): p. 41-50. 233. Haverinen, M., et al., Heat shock protein 90 and the nuclear transport of progesterone receptor. Cell Stress Chaperones, 2001. 6(3): p. 256-62. 234. Gallo, L.I., et al., Differential recruitment of tetratricorpeptide repeat domain immunophilins to the mineralocorticoid receptor influences both heat-shock protein 90-dependent retrotransport and hormone-dependent transcriptional activity. Biochemistry, 2007. 46(49): p. 14044-57. 235. Zhang, X., A.F. Clark, and T. Yorio, Heat shock protein 90 is an essential molecular chaperone for nuclear transport of glucocorticoid receptor beta. Invest Ophthalmol Vis Sci, 2006. 47(2): p. 700-8. 236. Galigniana, M.D., et al., Heat shock protein 90-dependent (geldanamycin-inhibited) movement of the glucocorticoid receptor through the cytoplasm to the nucleus requires intact cytoskeleton. Mol Endocrinol, 1998. 12(12): p. 1903-13. 237. Chen, C.Y. and W.E. Balch, The Hsp90 chaperone complex regulates GDI-dependent Rab recycling. Mol Biol Cell, 2006. 17(8): p. 3494-507. 238. Lotz, G.P., et al., A novel HSP90 chaperone complex regulates intracellular vesicle transport. J Cell Sci, 2008. 121(Pt 5): p. 717-23. 239. Pennisi, R., P. Ascenzi, and A. di Masi, Hsp90: A New Player in DNA Repair? Biomolecules, 2015. 5(4): p. 2589-2618. 240. Sharma, K., et al., Quantitative proteomics reveals that Hsp90 inhibition preferentially targets kinases and the DNA damage response. Mol Cell Proteomics, 2012. 11(3): p. M111.014654. 241. McClellan, A.J., et al., Diverse cellular functions of the Hsp90 molecular chaperone uncovered using systems approaches. Cell, 2007. 131(1): p. 121-35. 242. Rappsilber, J., et al., A generic strategy to analyze the spatial organization of multi-protein complexes by cross-linking and mass spectrometry. Anal Chem, 2000. 72(2): p. 267-75. 243. Vasilescu, J., X. Guo, and J. Kast, Identification of protein-protein interactions using in vivo cross-linking and mass spectrometry. Proteomics, 2004. 4(12): p. 3845-54. 244. Rexin, M., W. Busch, and U. Gehring, Chemical cross-linking of heteromeric glucocorticoid receptors. Biochemistry, 1988. 27(15): p. 5593-601. 245. Renoir, J.M., et al., The non-DNA-binding heterooligomeric form of mammalian steroid hormone receptors contains a hsp90-bound 59-kilodalton protein. J Biol Chem, 1990. 265(18): p. 10740-5. 246. Rexin, M., W. Busch, and U. Gehring, Protein components of the nonactivated glucocorticoid receptor. J Biol Chem, 1991. 266(36): p. 24601-5.

123

REFERENCES

247. Song, S., S. Kole, and M. Bernier, A chemical cross-linking method for the analysis of binding partners of heat shock protein-90 in intact cells. Biotechniques, 2012. 0(0): p. 1-7. 248. Mahony, D., D.A. Parry, and E. Lees, Active cdk6 complexes are predominantly nuclear and represent only a minority of the cdk6 in T cells. Oncogene, 1998. 16(5): p. 603-11. 249. Citri, A., et al., Hsp90 recognizes a common surface on client kinases. J Biol Chem, 2006. 281(20): p. 14361-9. 250. Ding, G., et al., Regulation of Ubiquitin-like with Plant Homeodomain and RING Finger Domain 1 (UHRF1) Protein Stability by Heat Shock Protein 90 Chaperone Machinery. The Journal of Biological Chemistry, 2016. 291(38): p. 20125-20135. 251. Li, Z., et al., HECTD3 Mediates an HSP90-Dependent Degradation Pathway for Protein Kinase Clients. Cell Rep, 2017. 19(12): p. 2515-2528. 252. Fierro-Monti, I., et al., Dynamic impacts of the inhibition of the molecular chaperone Hsp90 on the T-cell proteome have implications for anti-cancer therapy. PLoS One, 2013. 8(11): p. e80425. 253. Hartson, S.D. and R.L. Matts, Approaches for defining the Hsp90-dependent proteome. Biochim Biophys Acta, 2012. 1823(3): p. 656-67. 254. Falsone, S.F., et al., A proteomic approach towards the Hsp90-dependent ubiquitinylated proteome. Proteomics, 2007. 7(14): p. 2375-83. 255. Quadroni, M., A. Potts, and P. Waridel, Hsp90 inhibition induces both protein-specific and global changes in the ubiquitinome. J Proteomics, 2015. 120: p. 215-29. 256. Xu, G., J.S. Paige, and S.R. Jaffrey, Global analysis of lysine ubiquitination by ubiquitin remnant immunoaffinity profiling. Nat Biotechnol, 2010. 28(8): p. 868-73. 257. Kim, W., et al., Systematic and quantitative assessment of the ubiquitin-modified proteome. Mol Cell, 2011. 44(2): p. 325-40. 258. Berbers, G.A., et al., Localization and quantitation of hsp84 in mammalian cells. Exp Cell Res, 1988. 177(2): p. 257-71. 259. Arrigo, A.P., S. Fakan, and A. Tissieres, Localization of the heat shock-induced proteins in Drosophila melanogaster tissue culture cells. Dev Biol, 1980. 78(1): p. 86-103. 260. Pekki, A.K., Different immunoelectron microscopic locations of progesterone receptor and HSP90 in chick oviduct epithelial cells. J Histochem Cytochem, 1991. 39(8): p. 1095-101. 261. Perdew, G.H., et al., Localization and characterization of the 86- and 84-kDa heat shock proteins in Hepa 1c1c7 cells. Exp Cell Res, 1993. 209(2): p. 350-6. 262. Akner, G., et al., Evidence for reversible, non-microtubule and non-microfilament-dependent nuclear translocation of hsp90 after heat shock in human fibroblasts. Eur J Cell Biol, 1992. 58(2): p. 356-64. 263. Collier, N.C. and M.J. Schlesinger, The dynamic state of heat shock proteins in chicken embryo fibroblasts. J Cell Biol, 1986. 103(4): p. 1495-507. 264. Langer, T., S. Rosmus, and H. Fasold, Intracellular localization of the 90 kDA heat shock protein (HSP90alpha) determined by expression of a EGFP-HSP90alpha-fusion protein in unstressed and heat stressed 3T3 cells. Cell Biol Int, 2003. 27(1): p. 47-52. 265. Biggiogera, M., et al., Localization of heat shock proteins in mouse male germ cells: an immunoelectron microscopical study. Exp Cell Res, 1996. 229(1): p. 77-85. 266. Vázquez-Nin, G.H., et al., Immunoelectron microscope localization of Mr 90000 heat shock protein and Mr 70000 heat shock cognate protein in the salivary glands of Chironomus thummi. Chromosoma, 1992. 102(1): p. 50-59. 267. Passinen, S., et al., The C-terminal half of Hsp90 is responsible for its cytoplasmic localization. Eur J Biochem, 2001. 268(20): p. 5337-42. 268. Longshaw, V.M., et al., Nuclear translocation of the Hsp70/Hsp90 organizing protein mSTI1 is regulated by cell cycle kinases. J Cell Sci, 2004. 117(Pt 5): p. 701-10. 269. Pratt, W.B. and D.O. Toft, Steroid Receptor Interactions with Heat Shock Protein and Immunophilin Chaperones. Endocrine Reviews, 1997. 18(3): p. 306-360.

124

REFERENCES

270. Bennesch, M.A., et al., LSD1 engages a corepressor complex for the activation of the estrogen receptor α by estrogen and cAMP. Nucleic Acids Research, 2016. 44(18): p. 8655- 8670. 271. Segnitz, B. and U. Gehring, The function of steroid hormone receptors is inhibited by the hsp90-specific compound geldanamycin. J Biol Chem, 1997. 272(30): p. 18694-701. 272. Veldscholte, J., et al., Anti-androgens and the mutated androgen receptor of LNCaP cells: differential effects on binding affinity, heat-shock protein interaction, and transcription activation. Biochemistry, 1992. 31(8): p. 2393-2399. 273. Vanaja, D.K., et al., Effect of geldanamycin on androgen receptor function and stability. Cell Stress & Chaperones, 2002. 7(1): p. 55-64. 274. Marivoet, S., et al., Interaction of the 90-kDa heat shock protein with native and in vitro translated androgen receptor and receptor fragments. Mol Cell Endocrinol, 1992. 88(1-3): p. 165-74. 275. Knoblauch, R. and M.J. Garabedian, Role for Hsp90-associated cochaperone p23 in estrogen receptor signal transduction. Mol Cell Biol, 1999. 19(5): p. 3748-59. 276. Picard, D., et al., Reduced levels of hsp90 compromise steroid receptor action in vivo. Nature, 1990. 348(6297): p. 166-8. 277. Freeman, B.C. and K.R. Yamamoto, Disassembly of transcriptional regulatory complexes by molecular chaperones. Science, 2002. 296(5576): p. 2232-5. 278. Kirschke, E., et al., Glucocorticoid receptor function regulated by coordinated action of the Hsp90 and Hsp70 chaperone cycles. Cell, 2014. 157(7): p. 1685-97. 279. Echeverria, P.C., et al., Nuclear import of the glucocorticoid receptor-hsp90 complex through the nuclear pore complex is mediated by its interaction with Nup62 and importin beta. Mol Cell Biol, 2009. 29(17): p. 4788-97. 280. Calderwood, S.K. and L. Neckers, Chapter Four - Hsp90 in Cancer: Transcriptional Roles in the Nucleus, in Advances in Cancer Research, I. Jennifer and W. Luke, Editors. 2016, Academic Press. p. 89-106. 281. Sawarkar, R. and R. Paro, [email protected]: an emerging hub of a networker. Trends in Cell Biology, 2013. 23(4): p. 193-201. 282. Erlejman, A.G., et al., Regulatory role of the 90-kDa-heat-shock protein (Hsp90) and associated factors on gene expression. Biochim Biophys Acta, 2014. 1839(2): p. 71-87. 283. Khurana, N. and S. Bhattacharyya, Hsp90, the concertmaster: tuning transcription. Front Oncol, 2015. 5: p. 100. 284. Ihle, J.N., STATs: Signal Transducers and Activators of Transcription. Cell. 84(3): p. 331-334. 285. Shah, M., et al., Interactions of STAT3 with caveolin-1 and heat shock protein 90 in plasma membrane raft and cytosolic complexes. Preservation of cytokine signaling during fever. J Biol Chem, 2002. 277(47): p. 45662-9. 286. Sato, N., et al., Involvement of heat-shock protein 90 in the interleukin-6-mediated signaling pathway through STAT3. Biochem Biophys Res Commun, 2003. 300(4): p. 847-52. 287. Chen, G., P. Cao, and D.V. Goeddel, TNF-Induced Recruitment and Activation of the IKK Complex Require Cdc37 and Hsp90. Molecular Cell. 9(2): p. 401-410. 288. Cvoro, A., et al., Distinct Roles of Unliganded and Liganded Estrogen Receptors in Transcriptional Repression. Molecular Cell. 21(4): p. 555-564. 289. Paul, I., et al., The ubiquitin ligase CHIP regulates c-Myc stability and transcriptional activity. Oncogene, 2013. 32(10): p. 1284-95. 290. Conacci-Sorrell, M., L. McFerrin, and R.N. Eisenman, An overview of MYC and its interactome. Cold Spring Harb Perspect Med, 2014. 4(1): p. a014357. 291. Carystinos, G.D., et al., Unexpected induction of the human connexin 43 promoter by the ras signaling pathway is mediated by a novel putative promoter sequence. Mol Pharmacol, 2003. 63(4): p. 821-31.

125

REFERENCES

292. Boulon, S., et al., HSP90 and Its R2TP/Prefoldin-like Cochaperone Are Involved in the Cytoplasmic Assembly of RNA Polymerase II. Molecular Cell, 2010. 39(6): p. 912-924. 293. Sawarkar, R., C. Sievers, and R. Paro, Hsp90 globally targets paused RNA polymerase to regulate gene expression in response to environmental stimuli. Cell, 2012. 149(4): p. 807-18. 294. Csermely, P. and C.R. Kahn, The 90-kDa heat shock protein (hsp-90) possesses an ATP binding site and autophosphorylating activity. J Biol Chem, 1991. 266(8): p. 4943-50. 295. Schnaider, T., et al., Interactions of Hsp90 with histones and related peptides. Life Sci, 1999. 65(22): p. 2417-26. 296. Csermely, P., et al., The 90 kDa heat shock protein (hsp90) induces the condensation of the chromatin structure. Biochem Biophys Res Commun, 1994. 202(3): p. 1657-63. 297. Isaacs, J.S., Chapter Five - Hsp90 as a “Chaperone” of the Epigenome: Insights and Opportunities for Cancer Therapy, in Advances in Cancer Research, I. Jennifer and W. Luke, Editors. 2016, Academic Press. p. 107-140. 298. Brown, M.A., et al., C-terminal domain of SMYD3 serves as a unique HSP90-regulated motif in oncogenesis. Oncotarget, 2015. 6(6): p. 4005-4019. 299. Tariq, M., et al., Trithorax requires Hsp90 for maintenance of active chromatin at sites of gene expression. Proceedings of the National Academy of Sciences of the United States of America, 2009. 106(4): p. 1157-1162. 300. Fiskus, W., et al., Panobinostat treatment depletes EZH2 and DNMT1 levels and enhances decitabine mediated de-repression of JunB and loss of survival of human acute leukemia cells. Cancer biology & therapy, 2009. 8(10): p. 939-950. 301. Milne, T.A., et al., MLL targets SET domain methyltransferase activity to Hox gene promoters. Mol Cell, 2002. 10(5): p. 1107-17. 302. Cao, R., et al., Role of histone H3 lysine 27 methylation in Polycomb-group silencing. Science, 2002. 298(5595): p. 1039-43. 303. Sollars, V., et al., Evidence for an epigenetic mechanism by which Hsp90 acts as a capacitor for morphological evolution. Nat Genet, 2003. 33(1): p. 70-4. 304. Tariq, M., et al., Trithorax requires Hsp90 for maintenance of active chromatin at sites of gene expression. Proc Natl Acad Sci U S A, 2009. 106(4): p. 1157-62. 305. Chen, Y., et al., Identification of mixed lineage leukemia 1(MLL1) protein as a coactivator of heat shock factor 1(HSF1) protein in response to heat shock protein 90 (HSP90) inhibition. J Biol Chem, 2014. 289(27): p. 18914-27. 306. Li, L., et al., Widespread rearrangement of 3D chromatin organization underlies polycomb- mediated stress-induced silencing. Mol Cell, 2015. 58(2): p. 216-31. 307. Abu-Farha, M., et al., The tale of two domains: proteomics and genomics analysis of SMYD2, a new histone methyltransferase. Mol Cell Proteomics, 2008. 7(3): p. 560-72. 308. Hamamoto, R., et al., SMYD3 encodes a histone methyltransferase involved in the proliferation of cancer cells. Nat Cell Biol, 2004. 6(8): p. 731-40. 309. Koepp, D.M., Cell cycle regulation by protein degradation. Methods Mol Biol, 2014. 1170: p. 61-73. 310. Desterro, J.M., M.S. Rodriguez, and R.T. Hay, Regulation of transcription factors by protein degradation. Cell Mol Life Sci, 2000. 57(8-9): p. 1207-19. 311. Chen, B., D. Zhong, and A. Monteiro, Comparative genomics and evolution of the HSP90 family of genes across all kingdoms of organisms. BMC Genomics, 2006. 7: p. 156. 312. Kunisawa, J. and N. Shastri, Hsp90α Chaperones Large C-Terminally Extended Proteolytic Intermediates in the MHC Class I Antigen Processing Pathway. Immunity. 24(5): p. 523-534. 313. Li, W., et al., Extracellular heat shock protein-90alpha: linking hypoxia to skin cell motility and wound healing. Embo j, 2007. 26(5): p. 1221-33. 314. Metchat, A., et al., Mammalian heat shock factor 1 is essential for oocyte meiosis and directly regulates Hsp90alpha expression. J Biol Chem, 2009. 284(14): p. 9521-8.

126

REFERENCES

315. Chadli, A., S.J. Felts, and D.O. Toft, GCUNC45 is the first Hsp90 co-chaperone to show alpha/beta isoform specificity. J Biol Chem, 2008. 283(15): p. 9509-12. 316. Morcillo, G., et al., HSP90 associates with specific heat shock puffs (hsr omega) in polytene chromosomes of Drosophila and Chironomus. Chromosoma, 1993. 102(9): p. 648-59. 317. Greer, C.B., et al., Histone deacetylases positively regulate transcription through the elongation machinery. Cell reports, 2015. 13(7): p. 1444-1455. 318. Margueron, R., et al., Ezh1 and Ezh2 maintain repressive chromatin through different mechanisms. Mol Cell, 2008. 32(4): p. 503-18. 319. Hauri, S., et al., A High-Density Map for Navigating the Human Polycomb Complexome. Cell Rep, 2016. 17(2): p. 583-595. 320. Dignam, J.D., R.M. Lebovitz, and R.G. Roeder, Accurate transcription initiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei. Nucleic Acids Res, 1983. 11(5): p. 1475-89. 321. Ong, S.E., et al., Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics. Mol Cell Proteomics, 2002. 1(5): p. 376-86. 322. Povlsen, L.K., et al., Systems-wide analysis of ubiquitylation dynamics reveals a key role for PAF15 ubiquitylation in DNA-damage bypass. Nat Cell Biol, 2012. 14(10): p. 1089-1098. 323. Udeshi, N.D., et al., Large-scale identification of ubiquitination sites by mass spectrometry. Nat Protoc, 2013. 8(10): p. 1950-60. 324. Scuoppo, C., et al., A tumour suppressor network relying on the polyamine-hypusine axis. Nature, 2012. 487(7406): p. 244-8. 325. Thomas, P. and T.G. Smart, HEK293 cell line: A vehicle for the expression of recombinant proteins. Journal of Pharmacological and Toxicological Methods, 2005. 51(3): p. 187-200. 326. Pratt, W.B. and D.O. Toft, Steroid receptor interactions with heat shock protein and immunophilin chaperones. Endocr Rev, 1997. 18(3): p. 306-60. 327. Leach, K.L., et al., Molybdate inhibition of glucocorticoid receptor inactivation and transformation. J Biol Chem, 1979. 254(23): p. 11884-90. 328. Ashraf, W., et al., The epigenetic integrator UHRF1: on the road to become a universal biomarker for cancer. Oncotarget, 2017. 8(31): p. 51946-51962. 329. Wandinger, S.K., et al., The phosphatase Ppt1 is a dedicated regulator of the molecular chaperone Hsp90. The EMBO Journal, 2006. 25(2): p. 367-376. 330. Jin, Y.J. and S.J. Burakoff, The 25-kDa FK506-binding protein is localized in the nucleus and associates with casein kinase II and nucleolin. Proc Natl Acad Sci U S A, 1993. 90(16): p. 7769- 73. 331. Yao, Y.L., et al., FKBPs in chromatin modification and cancer. Curr Opin Pharmacol, 2011. 11(4): p. 301-7. 332. Yang, W.M., Y.L. Yao, and E. Seto, The FK506-binding protein 25 functionally associates with histone deacetylases and with transcription factor YY1. Embo j, 2001. 20(17): p. 4814-25. 333. Riviere, S., A. Menez, and A. Galat, On the localization of FKBP25 in T-lymphocytes. FEBS Lett, 1993. 315(3): p. 247-51. 334. Eitoku, M., et al., Histone chaperones: 30 years from isolation to elucidation of the mechanisms of nucleosome assembly and disassembly. Cell Mol Life Sci, 2008. 65(3): p. 414- 44. 335. Szklarczyk, D., et al., The STRING database in 2017: quality-controlled protein-protein association networks, made broadly accessible. Nucleic Acids Res, 2017. 45(D1): p. D362- d368. 336. Roh, S.H., et al., Contribution of the Type II Chaperonin, TRiC/CCT, to Oncogenesis. Int J Mol Sci, 2015. 16(11): p. 26706-20. 337. Louder, R.K., et al., Structure of promoter-bound TFIID and model of human pre-initiation complex assembly. Nature, 2016. 531(7596): p. 604-9.

127

REFERENCES

338. Chen, F., et al., PAF1, a molecular regulator of promoter-proximal pausing by RNA Polymerase II. Cell, 2015. 162(5): p. 1003-1015. 339. Van Oss, S.B., C.E. Cucinotta, and K.M. Arndt, Emerging Insights into the Roles of the Paf1 Complex in Gene Regulation. Trends Biochem Sci, 2017. 42(10): p. 788-798. 340. Belotserkovskaya, R. and D. Reinberg, Facts about FACT and transcript elongation through chromatin. Curr Opin Genet Dev, 2004. 14(2): p. 139-46. 341. Sanden, C. and U. Gullberg, The DEK oncoprotein and its emerging roles in gene regulation. Leukemia, 2015. 29(8): p. 1632-6. 342. von Lindern, M., et al., The translocation (6;9), associated with a specific subtype of acute myeloid leukemia, results in the fusion of two genes, dek and can, and the expression of a chimeric, leukemia-specific dek-can mRNA. Mol Cell Biol, 1992. 12(4): p. 1687-97. 343. Li, H., et al., Molecular basis for site-specific read-out of histone H3K4me3 by the BPTF PHD finger of NURF. Nature, 2006. 442(7098): p. 91-5. 344. Jones, M.H., N. Hamana, and M. Shimane, Identification and Characterization of BPTF, a Novel Bromodomain Transcription Factor. Genomics, 2000. 63(1): p. 35-39. 345. Wang, P. and T. Jin, Oct-1 functions as a sensor for metabolic and stress signals. Islets, 2010. 2(1): p. 46-8. 346. Pance, A., Oct-1, to go or not to go? That is the PolII question. Biochim Biophys Acta, 2016. 1859(6): p. 820-4. 347. Boubriak, II, et al., Stress-induced release of Oct-1 from the nuclear envelope is mediated by JNK phosphorylation of lamin B1. PLoS One, 2017. 12(5): p. e0177990. 348. Tantin, D., et al., The octamer binding transcription factor Oct-1 is a stress sensor. Cancer Res, 2005. 65(23): p. 10750-8. 349. Williams, S.R., et al., Smith-Magenis syndrome results in disruption of CLOCK gene transcription and reveals an integral role for RAI1 in the maintenance of circadian rhythmicity. American Journal of Human Genetics, 2012. 90(6): p. 941-949. 350. Kasof, G.M., L. Goyal, and E. White, Btf, a novel death-promoting transcriptional repressor that interacts with Bcl-2-related proteins. Mol Cell Biol, 1999. 19(6): p. 4390-404. 351. Horlein, A.J., et al., Ligand-independent repression by the thyroid hormone receptor mediated by a nuclear receptor co-repressor. Nature, 1995. 377(6548): p. 397-404. 352. Wang, J., et al., ETO, fusion partner in t(8;21) acute myeloid leukemia, represses transcription by interaction with the human N-CoR/mSin3/HDAC1 complex. Proceedings of the National Academy of Sciences of the United States of America, 1998. 95(18): p. 10860-10865. 353. Doyon, Y., et al., Structural and functional conservation of the NuA4 histone acetyltransferase complex from yeast to humans. Mol Cell Biol, 2004. 24(5): p. 1884-96. 354. Hayakawa, T., et al., MRG15 binds directly to PALB2 and stimulates homology-directed repair of chromosomal breaks. J Cell Sci, 2010. 123(Pt 7): p. 1124-30. 355. Yochum, G.S. and D.E. Ayer, Role for the Mortality Factors MORF4, MRGX, and MRG15 in Transcriptional Repression via Associations with Pf1, mSin3A, and Transducin-Like Enhancer of Split. Molecular and Cellular Biology, 2002. 22(22): p. 7868-7876. 356. Kelly, R.D. and S.M. Cowley, The physiological roles of histone deacetylase (HDAC) 1 and 2: complex co-stars with multiple leading parts. Biochem Soc Trans, 2013. 41(3): p. 741-9. 357. Black, A.R., J.D. Black, and J. Azizkhan-Clifford, Sp1 and kruppel-like factor family of transcription factors in cell growth regulation and cancer. J Cell Physiol, 2001. 188(2): p. 143- 60. 358. Wierstra, I., Sp1: emerging roles--beyond constitutive activation of TATA-less housekeeping genes. Biochem Biophys Res Commun, 2008. 372(1): p. 1-13. 359. Wysocka, J., P.T. Reilly, and W. Herr, Loss of HCF-1-chromatin association precedes temperature-induced growth arrest of tsBN67 cells. Mol Cell Biol, 2001. 21(11): p. 3820-9. 360. Herr, W., The herpes simplex virus VP16-induced complex: mechanisms of combinatorial transcriptional regulation. Cold Spring Harb Symp Quant Biol, 1998. 63: p. 599-607.

128

REFERENCES

361. Gerster, T. and R.G. Roeder, A herpesvirus trans-activating protein interacts with transcription factor OTF-1 and other cellular proteins. Proc Natl Acad Sci U S A, 1988. 85(17): p. 6347-51. 362. Katan, M., et al., Characterization of a cellular factor which interacts functionally with Oct-1 in the assembly of a multicomponent transcription complex. Nucleic Acids Res, 1990. 18(23): p. 6871-80. 363. Xiao, P. and J.P. Capone, A cellular factor binds to the herpes simplex virus type 1 transactivator Vmw65 and is required for Vmw65-dependent protein-DNA complex assembly with Oct-1. Mol Cell Biol, 1990. 10(9): p. 4974-7. 364. Wysocka, J., et al., Human Sin3 deacetylase and trithorax-related Set1/Ash2 histone H3-K4 methyltransferase are tethered together selectively by the cell-proliferation factor HCF-1. Genes Dev, 2003. 17(7): p. 896-911. 365. Glaab, E., et al., EnrichNet: network-based gene set enrichment analysis. Bioinformatics, 2012. 28(18): p. i451-i457. 366. Moroianu, J., G. Blobel, and A. Radu, Nuclear protein import: Ran-GTP dissociates the karyopherin alphabeta heterodimer by displacing alpha from an overlapping binding site on beta. Proceedings of the National Academy of Sciences, 1996. 93(14): p. 7059-7062. 367. Prince, T.L., et al., Client Proteins and Small Molecule Inhibitors Display Distinct Binding Preferences for Constitutive and Stress-Induced HSP90 Isoforms and Their Conformationally Restricted Mutants. PLoS One, 2015. 10(10): p. e0141786. 368. Bukau, B., Regulation of the Escherichia coli heat-shock response. Mol Microbiol, 1993. 9(4): p. 671-80. 369. Craig, E.A. and C.A. Gross, Is hsp70 the cellular thermometer? Trends Biochem Sci, 1991. 16(4): p. 135-40. 370. Straus, D., W. Walter, and C.A. Gross, DnaK, DnaJ, and GrpE heat shock proteins negatively regulate heat shock gene expression by controlling the synthesis and stability of sigma 32. Genes Dev, 1990. 4(12a): p. 2202-9. 371. Anckar, J. and L. Sistonen, Regulation of HSF1 Function in the Heat Stress Response: Implications in Aging and Disease. Annual Review of Biochemistry, 2011. 80(1): p. 1089- 1115. 372. Zheng, X., et al., Dynamic control of Hsf1 during heat shock by a chaperone switch and phosphorylation. Elife, 2016. 5. 373. Matsumoto, K., et al., Hsp90 is involved in the formation of P-bodies and stress granules. Biochem Biophys Res Commun, 2011. 407(4): p. 720-4. 374. Legagneux, V., M. Morange, and O. Bensaude, Heat shock increases turnover of 90 kDa heat shock protein phosphate groups in HeLa cells. FEBS Lett, 1991. 291(2): p. 359-62. 375. Adams, J. and M. Kauffman, Development of the proteasome inhibitor Velcade (Bortezomib). Cancer Invest, 2004. 22(2): p. 304-11. 376. House, D.W. FDA clears Velcade label expansion. 2014. 377. Katerina, S. and P. Evangelia, HSP90 Inhibitors: Current Development and Potential in Cancer Therapy. Recent Patents on Anti-Cancer Drug Discovery, 2014. 9(1): p. 1-20. 378. Jensen, M.R., et al., NVP-AUY922: a small molecule HSP90 inhibitor with potent antitumor activity in preclinical breast cancer models. Breast Cancer Research, 2008. 10(2): p. R33. 379. Gaspar, N., et al., Mechanistic evaluation of the novel HSP90 inhibitor NVP-AUY922 in adult and pediatric glioblastoma. Molecular cancer therapeutics, 2010. 9(5): p. 1219-1233. 380. Okui, T., et al., Antitumor effect of novel HSP90 inhibitor NVP-AUY922 against oral squamous cell carcinoma. Anticancer Res, 2011. 31(4): p. 1197-204. 381. Komander, D., The emerging complexity of protein ubiquitination. Biochem Soc Trans, 2009. 37(Pt 5): p. 937-53. 382. Xu, G., et al., Vulnerability of newly synthesized proteins to proteostasis stress. Journal of Cell Science, 2016. 129(9): p. 1892-1901.

129

REFERENCES

383. Aprile-Garcia, F., et al., Nascent-protein ubiquitination is required for heat shock-induced gene downregulation in human cells. Nat Struct Mol Biol, 2019. 26(2): p. 137-146. 384. Nielsen, S.V., et al., Protein Quality Control in the Nucleus. Biomolecules, 2014. 4(3): p. 646- 661. 385. Park, S.H., et al., PolyQ proteins interfere with nuclear degradation of cytosolic proteins by sequestering the Sis1p chaperone. Cell, 2013. 154(1): p. 134-45. 386. Wagstaff, K.M., et al., Ivermectin is a specific inhibitor of importin alpha/beta-mediated nuclear import able to inhibit replication of HIV-1 and dengue virus. Biochem J, 2012. 443(3): p. 851-6. 387. Gorgoni, B., et al., Poly(A)-binding proteins are functionally distinct and have essential roles during vertebrate development. Proc Natl Acad Sci U S A, 2011. 108(19): p. 7844-9. 388. Wu, Z., A. Moghaddas Gholami, and B. Kuster, Systematic Identification of the HSP90 Regulated Proteome. Molecular & Cellular Proteomics : MCP, 2012. 11(6): p. M111.016675. 389. Pathan, M., et al., FunRich: An open access standalone functional enrichment and interaction network analysis tool. PROTEOMICS, 2015. 15(15): p. 2597-2601. 390. Nakai, A. and T. Ishikawa, Cell cycle transition under stress conditions controlled by vertebrate heat shock factors. Embo j, 2001. 20(11): p. 2885-95. 391. Komatsu, M., et al., Loss of autophagy in the central nervous system causes neurodegeneration in mice. Nature, 2006. 441(7095): p. 880-4. 392. Qing, G., P. Yan, and G. Xiao, Hsp90 inhibition results in autophagy-mediated proteasome- independent degradation of IkappaB kinase (IKK). Cell Res, 2006. 16(11): p. 895-901. 393. Johnston, J.A., C.L. Ward, and R.R. Kopito, Aggresomes: a cellular response to misfolded proteins. J Cell Biol, 1998. 143(7): p. 1883-98. 394. Markossian, K.A. and B.I. Kurganov, Protein folding, misfolding, and aggregation. Formation of inclusion bodies and aggresomes. Biochemistry (Mosc), 2004. 69(9): p. 971-84. 395. Tyedmers, J., A. Mogk, and B. Bukau, Cellular strategies for controlling protein aggregation. Nat Rev Mol Cell Biol, 2010. 11(11): p. 777-88. 396. Bennett, E.J., et al., Global impairment of the ubiquitin-proteasome system by nuclear or cytoplasmic protein aggregates precedes inclusion body formation. Mol Cell, 2005. 17(3): p. 351-65. 397. Park, J., et al., Misfolded polypeptides are selectively recognized and transported toward aggresomes by a CED complex. Nat Commun, 2017. 8: p. 15730. 398. Scior, A., K. Juenemann, and J. Kirstein, Cellular strategies to cope with protein aggregation. Essays Biochem, 2016. 60(2): p. 153-161. 399. Nivon, M., et al., NF-kappaB regulates protein quality control after heat stress through modulation of the BAG3-HspB8 complex. J Cell Sci, 2012. 125(Pt 5): p. 1141-51. 400. Zaarur, N., et al., Triggering aggresome formation. Dissecting aggresome-targeting and aggregation signals in synphilin 1. J Biol Chem, 2008. 283(41): p. 27575-84. 401. Kolodziejska, K.E., et al., Regulation of inducible nitric oxide synthase by aggresome formation. Proc Natl Acad Sci U S A, 2005. 102(13): p. 4854-9. 402. Kawaguchi, Y., et al., The deacetylase HDAC6 regulates aggresome formation and cell viability in response to misfolded protein stress. Cell, 2003. 115(6): p. 727-38. 403. Salemi, L.M., et al., Aggresome formation is regulated by RanBPM through an interaction with HDAC6. Biol Open, 2014. 3(6): p. 418-30. 404. Wigley, W.C., et al., Dynamic association of proteasomal machinery with the centrosome. J Cell Biol, 1999. 145(3): p. 481-90. 405. Zhang, X. and S.B. Qian, Chaperone-mediated hierarchical control in targeting misfolded proteins to aggresomes. Mol Biol Cell, 2011. 22(18): p. 3277-88. 406. Savitski, M.M., et al., Multiplexed Proteome Dynamics Profiling Reveals Mechanisms Controlling Protein Homeostasis. Cell, 2018. 173(1): p. 260-274.e25.

130

REFERENCES

407. Huebner, A.R., et al., Deubiquitylation of Protein Cargo Is Not an Essential Step in Exosome Formation. Mol Cell Proteomics, 2016. 15(5): p. 1556-71. 408. Nijman, S.M.B., Synthetic lethality: General principles, utility and detection using genetic screens in human cells. Febs Letters, 2011. 585(1): p. 1-6. 409. Lees-Miller, S.P. and C.W. Anderson, The human double-stranded DNA-activated protein kinase phosphorylates the 90-kDa heat-shock protein, hsp90 alpha at two NH2-terminal threonine residues. Journal of Biological Chemistry, 1989. 264(29): p. 17275-17280. 410. Chen, Y., et al., Mascot-derived False Positive Peptide Identifications Revealed by Manual Analysis of Tandem Mass Spectra. Journal of proteome research, 2009. 8(6): p. 3141-3147. 411. Paik, W.K., D.C. Paik, and S. Kim, Historical review: the field of protein methylation. Trends Biochem Sci, 2007. 32(3): p. 146-52. 412. Dahlberg, J.E., E. Lund, and E.B. Goodwin, Nuclear translation: what is the evidence? Rna, 2003. 9(1): p. 1-8. 413. Pederson, T., The persistent plausibility of protein synthesis in the nucleus: process, palimpsest or pitfall? Curr Opin Cell Biol, 2013. 25(4): p. 520-1. 414. Echtenkamp, F.J. and B.C. Freeman, Molecular chaperone-mediated nuclear protein dynamics. Curr Protein Pept Sci, 2014. 15(3): p. 216-24. 415. Shibata, Y. and R.I. Morimoto, How the nucleus copes with proteotoxic stress. Current biology : CB, 2014. 24(10): p. R463-R474. 416. Jones, R.D. and R.G. Gardner, Protein quality control in the nucleus. Curr Opin Cell Biol, 2016. 40: p. 81-89. 417. Enam, C., et al., Protein Quality Control Degradation in the Nucleus. Annu Rev Biochem, 2018. 87: p. 725-749. 418. Medard, G., et al., Optimized chemical proteomics assay for kinase inhibitor profiling. J Proteome Res, 2015. 14(3): p. 1574-86. 419. Yam, A.Y., et al., Defining the TRiC/CCT interactome links chaperonin function to stabilization of newly-made proteins with complex topologies. Nature structural & molecular biology, 2008. 15(12): p. 1255-1262. 420. Kasembeli, M., et al., Modulation of STAT3 Folding and Function by TRiC/CCT Chaperonin. PLOS Biology, 2014. 12(4): p. e1001844. 421. Huang, R., et al., New insights into the functions and localization of nuclear CCT protein complex in K562 leukemia cells. Proteomics Clin Appl, 2012. 6(9-10): p. 467-75. 422. Frehlick, L.J., J.M. Eirin-Lopez, and J. Ausio, New insights into the nucleophosmin/nucleoplasmin family of nuclear chaperones. Bioessays, 2007. 29(1): p. 49- 59. 423. Freeman, B.C., et al., Analysis of molecular chaperone activities using in vitro and in vivo approaches. Methods Mol Biol, 2000. 99: p. 393-419. 424. Horiuchi, K., et al., Identification of Wilms' tumor 1-associating protein complex and its role in alternative splicing and the cell cycle. J Biol Chem, 2013. 288(46): p. 33292-302. 425. Horiuchi, K., et al., Wilms' tumor 1-associating protein regulates G2/M transition through stabilization of cyclin A2 mRNA. Proc Natl Acad Sci U S A, 2006. 103(46): p. 17278-83. 426. Figueroa, A., et al., Novel roles of hakai in cell proliferation and oncogenesis. Mol Biol Cell, 2009. 20(15): p. 3533-42. 427. Dehe, P.M. and V. Geli, The multiple faces of Set1. Biochem Cell Biol, 2006. 84(4): p. 536-48. 428. Lakowski, B., I. Roelens, and S. Jacob, CoREST-like complexes regulate chromatin modification and neuronal gene expression. J Mol Neurosci, 2006. 29(3): p. 227-39. 429. Torchy, M.P., A. Hamiche, and B.P. Klaholz, Structure and function insights into the NuRD chromatin remodeling complex. Cell Mol Life Sci, 2015. 72(13): p. 2491-507. 430. Kadamb, R., et al., Sin3: Insight into its transcription regulatory functions. European Journal of Cell Biology, 2013. 92(8): p. 237-246.

131

REFERENCES

431. Brownlee, P.M., C. Meisenberg, and J.A. Downs, The SWI/SNF chromatin remodelling complex: Its role in maintaining genome stability and preventing tumourigenesis. DNA Repair (Amst), 2015. 32: p. 127-33. 432. Pitts, T.M., et al., Targeting nuclear kinases in cancer: development of cell cycle kinase inhibitors. Pharmacol Ther, 2014. 142(2): p. 258-69. 433. Shain, A.H. and J.R. Pollack, The spectrum of SWI/SNF mutations, ubiquitous in human cancers. PLoS One, 2013. 8(1): p. e55119. 434. Hodges, C., J.G. Kirkland, and G.R. Crabtree, The many roles of BAF (mSWI/SNF) and PBAF complexes in cancer. Cold Spring Harbor perspectives in medicine, 2016. 6(8): p. a026930. 435. St Pierre, R. and C. Kadoch, Mammalian SWI/SNF complexes in cancer: emerging therapeutic opportunities. Curr Opin Genet Dev, 2017. 42: p. 56-67. 436. Wang, X., J.R. Haswell, and C.W. Roberts, Molecular pathways: SWI/SNF (BAF) complexes are frequently mutated in cancer--mechanisms and potential therapeutic insights. Clin Cancer Res, 2014. 20(1): p. 21-7. 437. Miao, D., et al., Genomic correlates of response to immune checkpoint therapies in clear cell renal cell carcinoma. Science, 2018. 359(6377): p. 801-806. 438. Pan, D., et al., A major chromatin regulator determines resistance of tumor cells to T cell- mediated killing. Science, 2018. 359(6377): p. 770-775. 439. Al-Lazikani, B., U. Banerji, and P. Workman, Combinatorial drug therapy for cancer in the post-genomic era. Nat Biotechnol, 2012. 30(7): p. 679-92. 440. Hwang, M., L. Moretti, and B. Lu, HSP90 inhibitors: multi-targeted antitumor effects and novel combinatorial therapeutic approaches in cancer therapy. Curr Med Chem, 2009. 16(24): p. 3081-92. 441. Xu, W., et al., Sensitivity of Mature ErbB2 to Geldanamycin Is Conferred by Its Kinase Domain and Is Mediated by the Chaperone Protein Hsp90. Journal of Biological Chemistry, 2001. 276(5): p. 3702-3708. 442. Le, X.F., F. Pruefer, and R.C. Bast, Jr., HER2-targeting antibodies modulate the cyclin- dependent kinase inhibitor p27Kip1 via multiple signaling pathways. Cell Cycle, 2005. 4(1): p. 87-95. 443. Wainberg, Z.A., et al., Inhibition of HSP90 with AUY922 induces synergy in HER2-amplified trastuzumab-resistant breast and gastric cancer. Mol Cancer Ther, 2013. 12(4): p. 509-19. 444. Bailey, T.A., et al., Mechanisms of Trastuzumab resistance in ErbB2-driven breast cancer and newer opportunities to overcome therapy resistance. Journal of Carcinogenesis, 2011. 10: p. 28. 445. Franchin, C., et al., Quantitative analysis of a phosphoproteome readily altered by the protein kinase CK2 inhibitor quinalizarin in HEK-293T cells. Biochim Biophys Acta, 2015. 1854(6): p. 609-23. 446. Miyata, Y. and I. Yahara, The 90-kDa heat shock protein, HSP90, binds and protects casein kinase II from self-aggregation and enhances its kinase activity. Journal of Biological Chemistry, 1992. 267(10): p. 7042-7047. 447. Faust, M. and M. Montenarh, Subcellular localization of protein kinase CK2. A key to its function? Cell Tissue Res, 2000. 301(3): p. 329-40. 448. Pang, C.N.I., E. Gasteiger, and M.R. Wilkins, Identification of arginine- and lysine-methylation in the proteome of Saccharomyces cerevisiae and its functional implications. BMC Genomics, 2010. 11: p. 92-92. 449. Blank, M., et al., Enhanced Ubiquitinylation of Heat Shock Protein 90 as a Potential Mechanism for Mitotic Cell Death in Cancer Cells Induced with Hypericin. Cancer Research, 2003. 63(23): p. 8241. 450. Murtagh, J., H. Lu, and E.L. Schwartz, Taxotere-Induced Inhibition of Human Endothelial Cell Migration Is a Result of Heat Shock Protein 90 Degradation. Cancer Research, 2006. 66(16): p. 8192.

132

REFERENCES

451. Martin, C. and Y. Zhang, The diverse functions of histone lysine methylation. Nat Rev Mol Cell Biol, 2005. 6(11): p. 838-49. 452. McBride, A.E. and P.A. Silver, State of the arg: protein methylation at arginine comes of age. Cell, 2001. 106(1): p. 5-8. 453. Kristie, T.M., et al., The cellular C1 factor of the herpes simplex virus enhancer complex is a family of polypeptides. J Biol Chem, 1995. 270(9): p. 4387-94. 454. Wilson, A.C., et al., The VP16 accessory protein HCF is a family of polypeptides processed from a large precursor protein. Cell, 1993. 74(1): p. 115-25. 455. Kristie, T.M., The Mouse Homologue of the Human Transcription Factor C1 (Host Cell Factor): CONSERVATION OF FORMS AND FUNCTION. Journal of Biological Chemistry, 1997. 272(42): p. 26749-26755. 456. LaBoissiere, S., S. Walker, and P. O'Hare, Concerted activity of host cell factor subregions in promoting stable VP16 complex assembly and preventing interference by the acidic activation domain. Mol Cell Biol, 1997. 17(12): p. 7108-18. 457. Rose, R.E., et al., The CRE-binding protein dCREB-A is required for Drosophila embryonic development. Genetics, 1997. 146(2): p. 595-606. 458. Goto, H., et al., A single-point mutation in HCF causes temperature-sensitive cell-cycle arrest and disrupts VP16 function. Genes Dev, 1997. 11(6): p. 726-37. 459. Julien, E. and W. Herr, Proteolytic processing is necessary to separate and ensure proper cell growth and cytokinesis functions of HCF-1. Embo j, 2003. 22(10): p. 2360-9. 460. Wilson, A.C., et al., VP16 targets an amino-terminal domain of HCF involved in cell cycle progression. Mol Cell Biol, 1997. 17(10): p. 6139-46. 461. Zargar, Z. and S. Tyagi, Role of host cell factor-1 in cell cycle regulation. Transcription, 2012. 3(4): p. 187-92. 462. O'Hare, P., The virion transactivator of herpes simplex virus. Seminars in Virology, 1993. 4(3): p. 145-155. 463. Wysocka, J. and W. Herr, The herpes simplex virus VP16-induced complex: the makings of a regulatory switch. Trends Biochem Sci, 2003. 28(6): p. 294-304. 464. Adams, J., R. Kelso, and L. Cooley, The kelch repeat superfamily of proteins: propellers of cell function. Trends Cell Biol, 2000. 10(1): p. 17-24. 465. Wilson, A.C., M.G. Peterson, and W. Herr, The HCF repeat is an unusual proteolytic cleavage signal. Genes Dev, 1995. 9(20): p. 2445-58. 466. Luciano, R.L. and A.C. Wilson, HCF-1 functions as a coactivator for the zinc finger protein Krox20. J Biol Chem, 2003. 278(51): p. 51116-24. 467. Wilson, A.C., et al., HCF-1 amino- and carboxy-terminal subunit association through two separate sets of interaction modules: involvement of fibronectin type 3 repeats. Mol Cell Biol, 2000. 20(18): p. 6721-30. 468. La Boissière, S., T. Hughes, and P. O'Hare, HCF-dependent nuclear import of VP16. The EMBO Journal, 1999. 18(2): p. 480-489. 469. Bhuiyan, T., et al., Distinct OGT-Binding Sites Promote HCF-1 Cleavage. PLoS ONE, 2015. 10(8): p. e0136636. 470. Capotosti, F., et al., O-GlcNAc transferase catalyzes site-specific proteolysis of HCF-1. Cell, 2011. 144(3): p. 376-88. 471. Michaud, J., et al., HCFC1 is a common component of active human CpG-island promoters and coincides with ZNF143, THAP11, YY1, and GABP transcription factor occupancy. Genome Res, 2013. 23(6): p. 907-16. 472. Julien, E. and W. Herr, A switch in mitotic histone H4 lysine 20 methylation status is linked to M phase defects upon loss of HCF-1. Mol Cell, 2004. 14(6): p. 713-25. 473. Freiman, R.N. and W. Herr, Viral mimicry: common mode of association with HCF by VP16 and the cellular protein LZIP. Genes Dev, 1997. 11(23): p. 3122-7.

133

REFERENCES

474. Lu, R., et al., The herpesvirus transactivator VP16 mimics a human basic domain leucine zipper protein, luman, in its interaction with HCF. J Virol, 1998. 72(8): p. 6291-7. 475. Johannessen, M., et al., The multifunctional roles of the four-and-a-half-LIM only protein FHL2. Cell Mol Life Sci, 2006. 63(3): p. 268-84. 476. Morlon, A. and P. Sassone-Corsi, The LIM-only protein FHL2 is a serum-inducible transcriptional coactivator of AP-1. Proc Natl Acad Sci U S A, 2003. 100(7): p. 3977-82. 477. Vogel, J.L. and T.M. Kristie, Site-specific proteolysis of the transcriptional coactivator HCF-1 can regulate its interaction with protein cofactors. Proc Natl Acad Sci U S A, 2006. 103(18): p. 6817-22. 478. Hsieh, J.J.D., et al., Proteolytic Cleavage of MLL Generates a Complex of N- and C-Terminal Fragments That Confers Protein Stability and Subnuclear Localization. Molecular and Cellular Biology, 2003. 23(1): p. 186-194. 479. Yokoyama, A., et al., Leukemia proto-oncoprotein MLL is proteolytically processed into 2 fragments with opposite transcriptional properties. Blood, 2002. 100(10): p. 3710-8. 480. Luciano, R.L. and A.C. Wilson, An activation domain in the C-terminal subunit of HCF-1 is important for transactivation by VP16 and LZIP. Proceedings of the National Academy of Sciences of the United States of America, 2002. 99(21): p. 13403-13408. 481. Krausslich, H.G. and E. Wimmer, Viral proteinases. Annu Rev Biochem, 1988. 57: p. 701-54. 482. Nishimoto, T. and C. Basilico, Analysis of a method for selecting temperature-sensitive mutants of BHK cells. Somatic Cell Genet, 1978. 4(3): p. 323-40. 483. Reilly, P.T. and W. Herr, Spontaneous reversion of tsBN67 cell proliferation and cytokinesis defects in the absence of HCF-1 function. Exp Cell Res, 2002. 277(1): p. 119-30. 484. Trimarchi, J.M. and J.A. Lees, Sibling rivalry in the E2F family. Nat Rev Mol Cell Biol, 2002. 3(1): p. 11-20. 485. Dimova, D.K. and N.J. Dyson, The E2F transcriptional network: old acquaintances with new faces. Oncogene, 2005. 24(17): p. 2810-26. 486. Blais, A. and B.D. Dynlacht, Hitting their targets: an emerging picture of E2F and cell cycle control. Curr Opin Genet Dev, 2004. 14(5): p. 527-32. 487. Tyagi, S., et al., E2F activation of S phase promoters via association with HCF-1 and the MLL family of histone H3K4 methyltransferases. Mol Cell, 2007. 27(1): p. 107-19. 488. Gunther, M., M. Laithier, and O. Brison, A set of proteins interacting with transcription factor Sp1 identified in a two-hybrid screening. Mol Cell Biochem, 2000. 210(1-2): p. 131-42. 489. Vogel, J.L. and T.M. Kristie, The novel coactivator C1 (HCF) coordinates multiprotein enhancer formation and mediates transcription activation by GABP. Embo j, 2000. 19(4): p. 683-90. 490. Parker, J.B., et al., Host Cell Factor-1 Recruitment to E2F-bound and Cell Cycle Control Genes is Mediated by THAP11 and ZNF143. Cell reports, 2014. 9(3): p. 967-982. 491. Yu, H., et al., The ubiquitin carboxyl hydrolase BAP1 forms a ternary complex with YY1 and HCF-1 and is a critical regulator of gene expression. Mol Cell Biol, 2010. 30(21): p. 5071-85. 492. Dejosez, M., et al., Ronin/Hcf-1 binds to a hyperconserved enhancer element and regulates genes involved in the growth of embryonic stem cells. Genes Dev, 2010. 24(14): p. 1479-84. 493. Luciano, R.L. and A.C. Wilson, An activation domain in the C-terminal subunit of HCF-1 is important for transactivation by VP16 and LZIP. Proc Natl Acad Sci U S A, 2002. 99(21): p. 13403-8. 494. Lu, R. and V. Misra, Zhangfei: a second cellular protein interacts with herpes simplex virus accessory factor HCF in a manner similar to Luman and VP16. Nucleic Acids Res, 2000. 28(12): p. 2446-54. 495. Ajuh, P.M., et al., Association of a protein phosphatase 1 activity with the human factor C1 (HCF) complex. Nucleic Acids Res, 2000. 28(3): p. 678-86. 496. Scarr, R.B. and P.A. Sharp, PDCD2 is a negative regulator of HCF-1 (C1). Oncogene, 2002. 21(34): p. 5245-54.

134

REFERENCES

497. Dou, Y., et al., Physical association and coordinate function of the H3 K4 methyltransferase MLL1 and the H4 K16 acetyltransferase MOF. Cell, 2005. 121(6): p. 873-85. 498. Smith, E.R., et al., A human protein complex homologous to the Drosophila MSL complex is responsible for the majority of histone H4 acetylation at lysine 16. Mol Cell Biol, 2005. 25(21): p. 9175-88. 499. Yokoyama, A., et al., Leukemia proto-oncoprotein MLL forms a SET1-like histone methyltransferase complex with menin to regulate Hox gene expression. Mol Cell Biol, 2004. 24(13): p. 5639-49. 500. Narayanan, A., W.T. Ruyechan, and T.M. Kristie, The coactivator host cell factor-1 mediates Set1 and MLL1 H3K4 trimethylation at herpesvirus immediate early promoters for initiation of infection. Proc Natl Acad Sci U S A, 2007. 104(26): p. 10835-40. 501. Kristie, T.M., Dynamic modulation of HSV chromatin drives initiation of infection and provides targets for epigenetic therapies. Virology, 2015. 0: p. 555-561. 502. Liu, W., et al., PHF8 Mediates Histone H4 Lysine 20 Demethylation Events Involved in Cell Cycle Progression. Nature, 2010. 466(7305): p. 508-512. 503. Vogel, J.L. and T.M. Kristie, The dynamics of HCF-1 modulation of herpes simplex virus chromatin during initiation of infection. Viruses, 2013. 5(5): p. 1272-91. 504. Alfonso-Dunn, R., et al., Transcriptional Elongation of HSV Immediate Early Genes by the Super Elongation Complex Drives Lytic Infection and Reactivation from Latency. Cell Host & Microbe. 21(4): p. 507-517.e5. 505. Baubec, T., et al., Methylation-dependent and -independent genomic targeting principles of the MBD protein family. Cell, 2013. 153(2): p. 480-92. 506. Lienert, F., et al., Identification of genetic elements that autonomously determine DNA methylation states. Nature Genetics, 2011. 43: p. 1091. 507. Arrigoni, L., et al., Standardizing chromatin research: a simple and universal method for ChIP- seq. Nucleic Acids Res, 2016. 44(7): p. e67. 508. Lodhi, N. and A.V. Tulin, PARP1 genomics: chromatin immunoprecipitation approach using anti-PARP1 antibody (ChIP and ChIP-seq). Methods Mol Biol, 2011. 780: p. 191-208. 509. Lee, T.I., S.E. Johnstone, and R.A. Young, Chromatin immunoprecipitation and microarray- based analysis of protein location. Nat Protoc, 2006. 1(2): p. 729-48. 510. Liu, W., et al., PHF8 mediates histone H4 lysine 20 demethylation events involved in cell cycle progression. Nature, 2010. 466(7305): p. 508-12. 511. Zheng, M., et al., Structural features and chaperone activity of the NudC protein family. Journal of molecular biology, 2011. 409(5): p. 722-741. 512. Zhou, T., et al., A mammalian NudC-like protein essential for dynein stability and cell viability. Proceedings of the National Academy of Sciences of the United States of America, 2006. 103(24): p. 9039-9044. 513. Schneider, C., et al., Pharmacologic shifting of a balance between protein refolding and degradation mediated by Hsp90. Proc Natl Acad Sci U S A, 1996. 93(25): p. 14536-41. 514. Ambati, S.R., et al., Pre-clinical efficacy of PU-H71, a novel HSP90 inhibitor, alone and in combination with bortezomib in Ewing sarcoma. Mol Oncol, 2014. 8(2): p. 323-36. 515. Anders, L., et al., Genome-wide localization of small molecules. Nat Biotechnol, 2014. 32(1): p. 92-6. 516. Yoveva, A. and R. Sawarkar, Chromatin Immunoprecipitation (ChIP) of Heat Shock Protein 90 (Hsp90). Methods Mol Biol, 2018. 1709: p. 221-231. 517. Deplus, R., et al., TET2 and TET3 regulate GlcNAcylation and H3K4 methylation through OGT and SET1/COMPASS. Embo j, 2013. 32(5): p. 645-55. 518. Geller, R., S. Taguwa, and J. Frydman, Broad action of Hsp90 as a host chaperone required for viral replication. Biochimica et Biophysica Acta, 2012. 1823(3): p. 698-706.

135

REFERENCES

519. Glinsky, G.V., O. Berezovska, and A.B. Glinskii, Microarray analysis identifies a death-from- cancer signature predicting therapy failure in patients with multiple types of cancer. J Clin Invest, 2005. 115(6): p. 1503-21.

136

APPENDIX

Part of the present work was included in a manuscript submitted for publication in February, 2019 to the Molecular Cell journal (Antonova et al., 2019):

A Systematic Analysis of Nuclear Heat-shock Protein 90 Identifies a Metazoan- Specific Regulatory Module

Aneliya Antonova1,4,7, Barbara Hummel1,7, Desiree M. Redhaber2,3,4, Erik C. Hansen1, Ashkan Khavaran1,5, Kathrin Gundel1, Megan Schneck1, Jan Mitschke3, Gerhard Mittler1, Cornelius Miething2,3,6 and Ritwick Sawarkar1,*

1Max Planck Institute of Immunobiology and Epigenetics, Freiburg, Germany. 2German Consortium for Translational Cancer Research (DKTK), Partner Site Freiburg and German Cancer Research Center (DKFZ), Heidelberg, Germany. 3Department of Hematology, Oncology and Stem Cell Transplantation, Faculty of Medicine, Medical Center - University of Freiburg, Freiburg, Germany. 4Faculty of Biology, University of Freiburg, Freiburg, Germany. 5Faculty of Medicine, University of Freiburg, Freiburg, Germany. 6BIOSS Centre for Biological Signalling Studies, University of Freiburg, Freiburg, Germany 7Equal Contribution * Correspondence: [email protected]

Molecular chaperones such as heat-shock proteins (HSPs) help in protein folding and complex assembly processes. Their role in cytosol has been very well elucidated. Chaperones are also present in the nucleus, a compartment where proteins enter after being fully folded in the cytosol, raising an important question about chaperone function in this compartment. We have performed a systematic analysis of nuclear heat-shock protein 90 to identify regulatory functions of this chaperone in the nucleus. Combining physical and genetic interactomes with a cancer co-expression screen allowed us to define 'core functional interactors' of nuclear HSP90 consisting of five proteins. Using transcriptional studies we identified Host Cell Factor C1 (HCFC1) as a metazoan- specific transcriptional regulator that depends on HSP90 for its stability in the nucleus. We found that HSP90 is required for optimal activity of HCFC1 in transcription. Thus our study provides the first global insight into the function of nuclear HSP90.

Introduction proteins, known as 'clients'. There is no common sequence, structure or function Molecular chaperones are proteins that among the diverse HSP90 clients. Rather, stabilize, or help other non-native proteins to exposure of hydrophobic residues over a long acquire a native conformation, but are stretch as well as general conformational themselves not present in the final functional stability appears to define HSP90 clients [3]. structure [1]. In conjunction with the protein Metastable proteins like steroid receptors and degradation machinery, molecular chaperones signaling kinases exemplify HSP90 clients. influence half-lives of proteins controlling The N-terminal ATPase domain of HSP90 protein homeostasis or proteostasis. Initially powers the chaperoning process, while the C- defined as 'heat-shock proteins (HSPs)' owing terminal domain is required for dimerization. to their stress-responsiveness [2], chaperones A suite of co-chaperones including p23, are classified based on their structure and AHA1, CDC37 drive the conformation cycle function into various families such as HSP70 of the chaperone and/ or deliver specific and HSP90. While HSP70 binds and helps clients to HSP90 [4]. HSP90 chaperone early events in polypeptide folding, HSP90 system also helps in assembly and interacts with a limited set of partially folded

137 APPENDIX

disassembly of protein complexes with the regulates transcription at the sites of gene help of specific co-chaperones [5, 6]. expression is not fully known. By directly The unique role of HSP90 in the folding interacting with chromatin-associated proteins landscape of cellular proteome coupled with HSP90 may help nucleosome disassembly, its high levels in unstressed cells confer upon regulate transcriptional pausing/ elongation HSP90 unexpected functions in evolution and checkpoint [26]. HSP90 and p23 control the cancer. A partial depletion of HSP90 activity residence time of proteins bound to chromatin in yeast, flies, plants and fish leads to an by evicting protein complexes off chromatin increased phenotypic variation. An interesting [28]. Importantly, high nuclear but not interpretation of this observation is that the cytosolic, HSP90 has been shown to correlate chaperone acts as a 'buffer' of genetic variation with poor prognosis in a long-term study of a and an 'evolutionary capacitor' [7]. Even if cohort of patients with non-small cell lung HSP90 levels are high in normal cells, they cancer (NSCLC) [29]. Given that many further increase in several cancers [8-10]. cancers are driven by transcriptional Small-molecule inhibitors of HSP90's ATPase abnormalities, it is highly likely that activity have shown promising results in oncogenic transcription programs are clinical trials against a variety of cancers [11, stabilized by HSP90's chaperone activity in 12], causally implicating HSP90 in cancer the nucleus [30, 31]. Moreover both growth/ metastasis. The precise mechanism by transcription and HSP90 have been which HSP90 supports tumor proliferation independently targeted in cancer therapeutics. remains ill-defined, but likely involves Understanding mechanisms by which HSP90 chaperoning a number of oncogenic drivers regulates transcription may pave a way for a [13]. Thus HSP90 represents a key chaperone rational synergistic combination of cancer system that is conserved from bacteria to therapeutics targeting the chaperone and human, essential for viability in eukaryotes, transcription. Hence the complete realm of and relevant for evolution and cancer. An HSP90 activity in the nucleus must be understanding of the breadth of HSP90 thoroughly investigated. For this purpose, we function in cellular context requires exhaustive need to define the spectrum of clients and co- definition of its clients and co-chaperones. chaperones of nuclear HSP90 in mammalian Systems-wide studies in yeast have outlined cells. Published interaction studies have been the genetic and physical interactome of HSP90 performed in cytosol or total cell extract [17- [14-16], illuminating the different cellular 22, 32], wherein nuclear interactions are either processes HSP90 contributes to. Similar lost or outnumbered by large amount of global studies in metazoan are technically cytosolic HSP90 pool. While physical challenging, and have recently been reported, interactomes of chaperones generate lists of widening our understanding of HSP90 several hundred proteins, genetic interactions functions [3, 17-22]. can be used to prioritize clients that are Human HSP90 is encoded by two paralogs - functionally important. Given the importance HSP90α and β in addition to ER-specific of HSP90 in cancers, harnessing gene GRP94 and mitochondria-specific TRAP1. expression data available for numerous tumor Early studies indicated that about 2-3 [23, 24] samples would allow us to further narrow HSP90 resides in the nucleus. Unlike the ER down on nuclear HSP90 interactors with and mitochondria, nucleus imports fully potential relevance to oncogenesis. folded proteins and thus presents a unique opportunity to study the function of Results chaperones in proteostasis after the initial In this study we systematically analyzed the folding of proteins is complete. We and others interactions of nuclear HSP90 in human cells have shown that HSP90 binds chromatin and by employing three orthogonal and influences gene expression [6, 22, 25-27]. The independent approaches: (i) a chemical- molecular mechanism by which HSP90 genetic screen to quantitate genetic

138 APPENDIX

interactions of HSP90 with transcriptional annotation as 'chromatin/ transcriptional regulators, (ii) affinity purification of nuclear regulation' (Supp. Table 1). Each of the 668 HSP90 to identify direct physical interactors genes was targeted using 4 independent in the nucleus, and (iii) a computational screen shRNA to reduce the influence of off-target to identify transcriptional regulators that are effects on CGIS (Fig. 1B). Furthermore we co-expressed with HSP90 in cancer patient used doxycycline-inducible shRNA library datasets. The individual approaches are (Supp. Fig. 1) to minimize long-term described below. consequences of knocking down critical genes. A chemical-genetic screen for interactors of We performed the chemical-genetic screen human HSP90 in the nucleus using HSP90i on K562 cells in two replicates that showed high reproducibility (Supp. Fig. The proliferation of human erythroleukemic 2A). The shRNA targeting luciferase had K562 cell line is sensitive to HSP90 inhibition CGIS of ~0 (Fig. 1C, D, Supp. Fig. 2B) as using small molecules that target the ATPase expected for the negative control. Independent activity of the chaperone [33]. HSP90 is likely shRNA targeting the same gene showed very required for folding/ stabilization of some similar results confirming our data (Fig. 1D). proteins that are critical for cell proliferation. Thus our chemical genetic screen yielded Thus shRNA-mediated depletion of such robust CGIS for all 668 tested genes. With a proteins is expected to make cells more cut-off of |CGIS| >1.5, we identified 203 sensitive to HSP90 inhibition, analogous to genes that genetically interact with human genetic synthetic lethality. Based on this idea, HSP90 in the context of K562 cells: 103 genes we performed a chemical-genetic screen to with positive CGIS and 100 with negative identify genes which when knocked-down, CGIS (Supp. Table 2). Diverse proteins enhance/ reduce the sensitivity of K562 cells known to interact physically, functionally or to HSP90 inhibitor (hereafter referred to as genetically with HSP90 in yeast and flies were HSP90i) (Fig. 1A). As HSP90i, we employed recovered in the chemical-genetic screen in NVP-AUY922 – a drug that is in phase II human K562 cells validating our approach clinical trials against several cancers [34, 35]. (Fig. 1E). We identified key members of the We leveraged the approach of barcoded- Polycomb/ trithorax group (EZH2, JARID2, shRNA pooling [36] in which pools of shRNA MLL2/4), histone deacetylases targeting hundreds of genes are used to (HDAC4/5/6/7/11) and SWI/ SNF chromatin transduce cell population (Fig. 1A). The remodeler family (SMARCC2/B1/D1) that proportion of cells carrying each shRNA in a have all been genetically shown to interact mixed population is quantified by deep with HSP90 in flies and yeast [14-16, 37]. We sequencing of barcodes. Thus the difference in also found histone H3 Lys 9 methylation shRNA frequency in populations exposed to (H3K9me) machinery (EHMT2, CBX5, vehicle compared to HSP90i will represent the ZFP57) that we had recently linked to HSP90's quantitative effect of HSP90 inhibition on role as an evolutionary capacitor [38]. proliferation of cells expressing an individual Surprisingly 5 out of 6 members of the shRNA, allowing us to infer the 'chemical- nucleosome remodeling and deacetylase genetic interaction score' or CGIS of the target complex (NuRD) were found in the screen - gene. A negative CGIS of a gene indicates that CHD4, MTA1/3, MBD3, GATAD2B and cells depleted of the corresponding protein RBBP4 (Fig. 1E). Thus our chemical-genetic were more sensitive to HSP90i than control screen identified several novel chromatin cells, and thus the gene shows synthetic regulators (e.g. HCFC1, BRD4), transcription lethality with HSP90i. To focus on the factors (including TWIST1, PAX6), ubiquitin function of nuclear HSP90 in gene expression, ligases (such as FBXO17, UBE2E1) and we employed a library of pooled shRNA kinases (CDK2 and CDK8; Supp. Table 2, targeting 668 genes with 'Kyoto Encyclopedia Fig. 1D and E) as genetic interactors of of Genes and Genomes' (KEGG) pathway

139 APPENDIX

HSP90. Many of the genes identified in the would not have been identified as HSP90 chemical-genetic screen such as NuRD interactors in previous global screens in yeast components and HCFC1 are not encoded in [14-16]. the genome of Saccharomyces cerevisiae, and

Figure 1. Chemical-genetic interactome of HSP90. (A) Outline of the experimental approach. K562 erythroleukemia cell line (black circles) was retrovirally infected to introduce barcoded shRNA library and selected for blasticidin- resistant cells (green circles). shRNA expression was induced (red circles) by the addition of doxycycline (dox). HSP90i (NVP-AUY922) was added as shown. Black circles with ┼ sign indicate dead cells. 4 days after dox and/ or HSP90i treatment, cells were harvested from each pool and barcode frequency was determined by deep sequencing to calculate Chemical-Genetic Interaction Score (CGIS; see Methods). (B) Representative western blot analysis of DNMT3A expression in K562 cells showing the effect of individual shRNAs targeting DNMT3A or Renilla. Tubulin is used for normalization. (C) Waterfall plot showing CGIS of all 668 genes screened in the experiment on the Y-axis plotted against the gene-rank from lowest to highest CGIS. Threshold used to shortlist HSP90 interactors are indicated with horizontal blue lines. (D) Z-scores of individual shRNA (red boxes) targeting 20 genes with the lowest CGIS identified in the screen are shown. The grey histogram depicts normalized distribution of Z-score of all candidates in the screen. Control shRNAs (see Methods) are highlighted as blue boxes. Red vertical lines indicate 66th and 98th percentile in Z-scores on either sides of zero. (E) Genes identified with significant positive (blue) or negative (pink) CGIS from (C) are shown in a network derived using Cytoscape. Genes encoding components of a complex or proteins in the same functional pathway are shown together. Lines connecting individual genes indicate interactions known in literature.

140 APPENDIX

Physical interaction network of nuclear cytoplasm known to be HSP90 clients HSP90 in human cells (MAPK6, B-RAF, c-SRC, SYK, FYN, PKM, PKD1, etc) were not identified in our AP-MS Genetic interactions are functionally important (Supp. Table 3). Instead 9 out of 10 kinases for cell proliferation but can be attributed to that we identified were all known to have indirect relationship between the two nuclear functions (Supp. Table 3). The interacting genes. Moreover genetic resulting complex network of HSP90 binding interaction approaches have a limited capacity partners comprised proteins falling into the on their own to provide molecular insights into following main categories: (i) chaperones and the mechanism linking the two interacting co-chaperones, (ii) components of the protein genes. In the case of HSP90, genetic degradation machinery, (iii) DNA repair interactors need not necessarily be HSP90 proteins, (iv) RNA metabolism factors and (v) clients. Physical protein-protein interactome transcription regulators. We recovered several can provide a complementary approach by proteins known to be interacting with HSP90 confirming a direct contact between which included CDC37 and HSP70 co- genetically interacting proteins. Only 2-3% of chaperones, casein kinase, STUB1 ubiquitin total cellular HSP90 is in the nucleus [23, 24], ligase, POL II subunits and pausing/ and hence previous attempts to find HSP90 elongation proteins [26] as well as proteins interactors in total cell extracts have yielded involved in DNA repair [32]. We confirmed mostly cytosolic proteins [17-21]. We sought HSP90 binding to TRIM28/ KAP1 as we have to identify proteins that are physically recently shown [38]. Focusing on transcription interacting with nuclear HSP90 by affinity regulators only, we shortlisted 102 proteins as purification coupled to mass-spectrometry high-confidence interactors of nuclear HSP90 (AP-MS) after subcellular fractionation of (Supp. Table 4, Fig. 2D, sup. Fig. 3B). Of human cells (Fig. 2A). Given that HSP90α these, 6 were already reported to be HSP90 binds clients with higher affinity than HSP90β interactors in the database maintained by [39], we focused on the former for AP-MS. Didier Picard We used HEK293 Flp-In T-Rex cells [40] (www.picard.ch/HSP90Int/index.php). expressing HSP90α N-terminally tagged to Interestingly several members of two distinct Strep-tag. Tetracycline inducible production chromatin remodeling complexes SWI/ SNF ensured levels of tagged HSP90 lower than the and NuRD were identified. A related SWI/ endogenous protein (Fig. 2B). We used SNF complex called RSC in yeast has been nuclear extracts of induced cells to perform recently shown to require HSP90/ p23 for its StrepTactin-based AP-MS in four independent optimal function [28], and our data suggest the biological replicates. Cells expressing RFP conservation of this interaction in human. All instead of HSP90α acted as a negative control. the subunits of NuRD complex were pulled The proteins identified in HSP90α and RFP down along with nuclear HSP90 (Fig. 2E), AP-MS were statistically analyzed to identify strongly suggesting that HSP90 is likely interactors of nuclear HSP90α (Supp. Table required for the formation or function of 3). Since cytosolic HSP90 is much more NuRD complex, concordant with the results of abundant than the nuclear counterpart, we the chemical genetic screen (Fig. 1E). Thus made sure that cytosol did not contaminate the the physical interactome of nuclear HSP90 nuclear extracts used to identify HSP90 revealed a few proteins known to bind HSP90, interactome. First, purity of nuclear extracts and identified several novel interactions was confirmed using Western blotting prior to (Supp. Table 4). Since the follow-up AP-MS (Fig. 2C). Second, gene ontology validation and mechanistic studies can only be analysis of proteins identified in AP-MS performed with a small number of interactors, confirmed that mostly nuclear proteins were we shortlisted those nuclear HSP90 enriched, ruling out cytosolic contamination interactor(s) which are likely relevant in (Supp. Fig. 3A). Third, kinases present in the

141 APPENDIX

HSP90’s role in cancer proliferation as detailed below.

Figure 2. Physical interactome of nuclear HSP90α in human cells. (A) Schematic overview of the experimental protocol. (B) Expression of recombinant HSP90α upon tetracycline induction in HEK293 cell line demonstrated by western blot of cell extracts. GAPDH is used for normalization. (C) Purity of cellular fractions as demonstrated by western blot using snRNP70 as a nuclear- and GAPDH as a cytosolic marker. (D) Visualization of affinity purified HSP90-bound complexes by SDS-PAGE and silver staining. RFP-expressing HEK293 cells were used as a negative control. The blue arrows indicate the expected position of the tagged proteins. (E) Transcription regulators identified as physical interactors of HSP90α are shown in a network derived using Cytoscape. Genes encoding components of a complex or proteins in the same functional pathway are shown together. Lines connecting individual genes indicate interactions known in literature.

142 APPENDIX

Genes co-expressing with HSP90 in cancer NuRD and SWI/ SNF chromatin remodeling tissues complex, HDACs and HOX genes. In a 10-year longitudinal study of non-small Core functional interactome of nuclear lung cancer patients, high nuclear HSP90 HSP90 levels but not cytosolic HSP90 levels correlated with poor prognosis [29]. Nuclear While individual hits obtained in each of the HSP90 in tumor cells may stabilize critical three datasets - genetic interactors, physical transcriptional regulators and sustain the interactors and cancer co-expression gene set - oncogenic transcriptional program fueling will be a significant resource for future cancer [30, 31]. Thus, it is likely that tumors studies, we wanted to focus on proteins that with HSP90 over-expression also show were identified in all three dataset. A pair-wise upregulation of HSP90-dependent comparison of the three datasets showed a transcriptional regulators that drive cell high degree of overlap (Fig. 3D) despite the proliferation. We sought to test this fact that the three approaches were completely hypothesis. Since large-scale quantitative data independent of each other. Five proteins were on protein abundance in cancer and normal common to all three approaches – ACTL6, cells are not available, we resorted to curated CHD4, HCFC1, RBBP4 and SMARCC2, gene expression data in The Cancer Genome constituting the 'core functional interactome of Atlas (TCGA) (http://cancergenome.nih.gov/). nuclear HSP90' that interact with HSP90 both We focused on datasets from 17 cancers genetically and physically and are co- encompassing 7080 patients and 710 healthy expressed with HSP90 in tumors (Fig. 3E). controls in total. We confirmed that HSP90 is SMARCC2 and ACTL6 are part of BAF transcriptionally upregulated in 12 out of the complex from the SWI/ SNF chromatin 17 types of cancers compared to the remodeller family (Fig. 3F). The SWI/ SNF corresponding healthy tissues (Fig. 3A), as has complex has been functionally linked to been shown before [9, 10]. The physical HSP90 in independent studies in yeast [14] interactors of nuclear HSP90 identified in this validating the conclusion of 'core functional study (Fig. 2E, Supp. Table 4) are interactome of nuclear HSP90'. RBBP4 and significantly upregulated in most of the cancer CHD4 are part of the NuRD complex which is types where HSP90 levels are higher (Fig. absent in budding yeast and has so far not 3B). This observation motivated us to perform been connected with HSP90. Almost all an unbiased search for transcription regulators members of the NuRD complex were whose expression correlates with the identified in one or more of the three datasets expression of HSP90 across all 17 cancer- described here (Fig. 3F). Finally metazoan- types. We constructed an expression- specific Host cell factor C1 (HCFC1) was the correlation matrix using the TCGA dataset to only core functional interactor that showed identify 67 transcription regulator genes negative genetic interaction with HSP90 (Fig. whose expression is similar to the expression 1D and 3E). HCFC1, initially discovered as a of both HSP90α and HSP90β across all the host factor that supports transcription of viral cancer types studied (Supp. Table 5; Fig. 3C). genes is a protein that is recruited to chromatin These co-expressed genes were not via a variety of TF and in turn acts as a upregulated in all cancers, ruling out generic platform to recruit chromatin modifying/ markers of cell proliferation. Even if these 67 remodelling complexes such as Sin3A HDAC co-expressed genes were not selected based on complex and MLL/ SET histone functional relation with HSP90, 37% of these methyltransferase complex [41]. Several co-expressed genes were independently members of HCFC1 complex were recovered identified to interact either genetically or in one/ more screens here (Fig. 3F) suggesting physically with HSP90 in our study (Fig. 3D). a close molecular partnership between HSP90 These include members of Polycomb group, and HCFC1. Thus using a combination of orthogonal and independent approaches, we

143 APPENDIX identified 5 proteins belonging to three to identify those core interactors that require different transcriptional complexes as core HSP90 for their gene-regulatory role at functional interactors of nuclear HSP90 (Fig. chromatin. 3E and F). To validate our findings in more detail at a mechanistic level, we further sought

Figure 3. Core functional interactors of nuclear HSP90. (A) Normalized expression levels of HSP90 in human RNA-seq datasets from The Cancer Genome Atlas (TCGA). Normal tissue is compared to primary tumors in 17 different cancer types. Statistical significance determined by Student’s t-test against normal tissue is indicated by asterisks. The acronyms of cancer types are in the standard form of TCGA (see Methods). (B) Gene Set Analysis (GSA) of gene expression levels in primary tumor samples compared to normal tissue samples based on RNA-seq datasets from TCGA. Gene sets used are HSP90 interactors identified in this study and permutations of genes that do not interact with HSP90 (Others). Heatmaps indicate significance of up or down-regulation of primary tumor versus normal tissue gene expression as ∆GSA values. Maximal significance of upregulation is indicated by '+1' (p value = 0), while maximal significance of downregulation is indicated by ‘-1’ (p value = 0). (C) Heatmap showing the hierarchical

144 APPENDIX

clustering of 668 transcriptional regulators (Supp. Table 1) based on the median score of their expression across 17 different cancer types. Median score was calculated as indicated in the formula. Median PT and median NT denote median expression of a gene in primary tumor (PT) and normal tissue (NT) samples. The cluster containing HSP90AA1 and HSP90AB1 is highlighted with a black box. (D) Scheme showing the overlap between genetic interactors, physical interactors and genes co-expressed with HSP90 in different cancer types. (E) Five genes were identified in all three screens. (F) Three complexes (BAF, NuRD and HCFC1) are depicted with their respective subunits. Color scheme of the subunit names represent the number of screens in which the subunit was identified. Alternative subunits are shown in lighter-colored ovals.

Gene regulatory function of HSP90 expression data to ask if components of interactors HCFC1 complex are co-expressed with HSP90 in cancers. We used HCFC1 If HSP90 is required for the function or interactome that we manually curated (Supp. stability of an interactor X at chromatin, two Table 7) based on published data [41, 43-45, predictions can be made: first, HSP90 and the 47-50], and found that cancers that show high interactor X should share genomic binding HSP90 expression also express higher amount sites as evidenced by co-occupancy, and of HCFC1 interactors (Fig. 4C). This is further second, HSP90 inhibition should lead to loss illustrated when we look at the raw expression of function of the interactor causing data of individual genes across various cancers misregulation specifically of the genes (Fig. 4D, Supp. Fig. 4). It is noteworthy that targeted by the interactor X. We tested both most of the HCFC1-interacting proteins were these predictions in an unbiased manner. We not used in the TCGA screen described in Fig. utilized ChIP-seq data of HSP90 in two human 3. Thus several independent analyses lead us cell lines – K562 (Supp. Table 6) and BT474 to suggest that nuclear HSP90 interacts with [25], to identify DNA sequence motifs the metazoan-specific HCFC1-regulatory enriched in regions bound by HSP90. In module. Curiously, HCFC1 is one of the 11 addition we analyzed promoter sequences of genes termed as 'death-from-cancer' signature genes that are misregulated upon HSP90 identified in a longitudinal study of ~1000 inhibition using RNA-seq data of 6 different cancer patients [51]. Tumors that have high cell lines of human and mouse origin (either transcript levels of HCFC1 along with 10 published [38] or generated in this study). other genes show poor therapeutic outcome Bioinformatic analyses of HSP90 binding sites irrespective of the cancer type, reminiscent of and promoter sequence of misregulated genes the finding that non-small cell lung cancer identified 14 transcription factor (TF) motifs with high nuclear HSP90 levels show poor (Fig. 4A). 7 of these 14 TF are known to prognosis [29]. recruit HCFC1 to chromatin (Fig. 3F) suggesting that HSP90 binds and chaperones HCFC1 regulatory complex at chromatin. HSP90 stabilizes nuclear HCFC1 Indeed ChIP-seq analyses confirmed that We tested the possibility that nuclear HSP90 binding sites of HSP90 overlap with binding chaperones and stabilizes HCFC1 against of HCFC1 as well as TF known to recruit misfolding and degradation. HCFC1 is a large HCFC1 to chromatin [42-46] (Fig. 4B). A set protein of ~300kDa and is proteolytically of promoters not targeted by HSP90 acted as a processed into N- and C-terminal fragments of negative control, and showed no binding of ~150kDa each [52]. These two forms of HCFC1 (Fig. 4B), confirming that HSP90 and HCFC1 bind chromatin at the same promoters HCFC1 co-occupy genomic binding sites. tethered by various Having defined HCFC1 as a nuclear target of HSP90 in human, we re-analyzed the TCGA

145 APPENDIX

Figure 4. Physical and functional interplay between HSP90 and HCFC1 at chromatin. (A) Scheme to identify motifs underlying HSP90`s function at chromatin. Promoter regions of genes misregulated upon HSP90 inhibition as well as HSP90 peak regions were used to identify enriched TF binding motifs. Names in bold indicate TF known to recruit HCFC1 to chromatin (Fig. 3F). Cell types used are indicated. (B) ChIP-seq analysis for HSP90, HCFC1 and genes known to recruit HCFC1. Heatmaps visualize the occupancy of HSP90, HCFC1, ZNF143, THAP11, GAPB, YY1, E2F1, E2F4 and E2F6 in a 1000bp region around HSP90 peak summits as well as random regions in HSP90 unbound promoter regions. HSP90 ChIP-seq was done in K562 cells, all other data are from HeLa cells. Color-scaled intensities are in units of reads per million (rpm). (C) Gene Set Analysis (GSA) of gene expression levels in primary tumor samples compared to normal tissue samples based on RNA-seq datasets from TCGA. Gene sets used are HSP90 interactors identified in this study and permutations of genes that do not interact with HSP90 (Others). Heatmaps indicate significance of up or down-regulation of primary tumor versus normal tissue gene expression as ∆GSA values. Maximal significance of upregulation is indicated by '+1' (p value = 0), while maximal significance of downregulation is indicated by ‘-1’ (p value = 0). (D) Normalized expression levels of HSP90, HCFC1 and genes known to recruit HCFC1 in samples of normal tissue in comparison with primary tumors in 4 indicated cancer types. Source data from TCGA (see Methods).

146 APPENDIX

TF [43]. Upon HSP90 inhibition, protein 0.01, absolute fold change > 2) [38]. levels of uncleaved HCFC1 as well as its N- Surprisingly mainly downregulated genes and C-terminal fragments drastically reduced, were targeted by HCFC1 at their promoters, without much effect on HCFC1 transcript but not upregulated genes (Fig. 5B and C, levels (Fig. 5A, Supp. Fig. 5A). Next we Supp. Fig. 5B). HCFC1 both activates and assessed if HSP90 inhibition causes represses gene expression by recruiting either misregulation of HCFC1 targets, as can be histone methylase complex or HDAC complex expected from decrease in chromatin- respectively. Thus it appears that HSP90 is associated HCFC1. Acute HSP90 inhibition in required for nuclear forms of HCFC1 mouse ESC causes upregulation of 2727 genes complexes that control gene activation. and down-regulation of 1539 genes (P-value <

Figure 5. HSP90 stabilizes HCFC1. (A) HSP90 inhibition leads to HCFC1 protein level decrease in total nuclear extracts, nucleoplasm and chromatin fractions in HEK293 cells. snRNP70 and H3 served as loading controls. (B) Metaplots of HCFC1 binding around transcriptional start sites (TSS) of downregulated (purple) and upregulated (red) genes in mouse embryonic stem cells (mESC) upon HSP90 inhibition. Y-axes indicate normalized ChIP-seq occupancy in reads per million (rpm). (C) Genome browser tracks of two representative genes: NOP2 in K562 cells (upper panel) and Cdc6 in mESCs (lower panel). ChIP-seq occupancy of HCFC1 is shown in cyan. Expression levels are visualized in control conditions (blue) and upon HSP90 inhibition (red). The vertical scale indicates normalized read density in rpm.

147 APPENDIX

Stabilization of HCFC1 by HSP90 is confirming the requirement of HSP90 for required for cellular physiology optimal HCFC1 function. In the third model, we studied the ability of HCFC1 to activate To further test the role of chaperoning of VP16-dependent transcription of immediate HCFC1 function by HSP90 in a cellular early (IE) promoter of Herpes Simplex virus context, we employed three independent (HSV) [54]. We found that the activity of models. In the first model we asked if HSP90- HSV-IE promoter is reduced by treatment HCFC1 interaction is important for with HSP90 inhibitor (Fig. 6C) suggesting that proliferation of K562 cells. Analogous to HSP90 may be required by one/ more proteins chemical-genetic screen (Fig. 1), we in the HSV-IE promoter activity. Interestingly quantified the effect of HSP90i on K562 cells we found that overexpressing full length with or without knock-down of HCFC1. HCFC1 partially rescues the effect of HSP90 Control knockdown of renilla luciferase did inhibition on HSV transcription, providing the not sensitize cells to HSP90 inhibition. evidence that HSP90 inhibition decreases HCFC1 depletion on the other hand HSV-IE activity via HCFC1 destabilization exacerbated the effect of HSP90i on cell (fig. 6C). Hence HCFC1 requires HSP90 for viability (Fig. 6A). In the second model, we activating transcription from viral promoters, capitalized on a known spontaneous point in line with the observation that cellular mutation in the Kelch domain of HCFC1 - promoters targeted by HCFC1 are also P134S – originally isolated from the hamster downregulated upon HSP90 inhibition (Fig. cell line BHK-21 [53]. This model allowed us 5B and C). Thus using three independent to compare the effect of HSP90 inhibitor on cellular models, we show that HSP90-HCFC1 two isogenic cell lines differing only in one regulatory nexus discovered in this study is residue of HCFC1. While HSP90 inhibition vital for transcriptional maintenance of only mildly affected the growth rate of wild- cellular as well as viral genes. type cells, mutant cells were significantly sensitive to HSP90 inhibition (Fig. 6B),

Figure 6. Functional relevance of HSP90-HCFC1 interaction assessed in independent cellular models. (A) Effect of HCFC1 knockdown on growth rate of K562 cells in presence and absence of HSP90i. Inset: Western blot analysis of HCFC1 upon knockdown with shRNA directed against HCFC1. For normalization snRNP70 was used. (B) Effect of a point mutation P134S in the Kelch domain of HCFC1 sensitizes cells to HSP90 inhibition. Two isogenic BHK-21 cell lines differing in one residue of HCFC1 (mutant and wild-type) were exposed to HSP90 inhibition and effect on growth was determined. The statistical significance was calculated using one-tailed T-test. (C) Effect of HSP90 inhibitor and

148 APPENDIX

HCFC1-overexpression on VP16-induced transcription of viral promoter-reporter constructs. Normalized values of mRNA abundance of the reporter as evaluated by RT-qPCR are shown.

Discussion Proteostatic demands of the nucleus are HSP90i in cancer treatment [57]. The physical different from that of the cytosol or the ER interactome of nuclear HSP90 identified where nascent polypeptides acquire mature multiple proteins that are components of conformation. Folded proteins enter the individual complexes such as NuRD. It is nucleus to encounter a unique challenge of likely that HSP90 is critical for the assembly dealing with chromatin polymers. and/ or the function of these complexes in the Additionally several transcription regulators nucleus. Additionally, genes regulated in a have disordered domains and likely require manner similar to HSP90 in cancers identified chaperones to assemble into complexes or in this study suggest shared upstream bind to chromatin to perform specific tasks. regulators, which drive cancer-specific Our understanding of nuclear protein quality transcriptional programs. Mechanisms control still requires global and robust underlying the activation of such regulators in identification of chaperone clients in this cancers will be a therapeutically relevant area cellular compartment. HSP90 is a specialized of future studies. chaperone that binds near native or natively Protein Quality Control mechanism controlled folded proteins, either to perform final folding by HSP90 works in unison with co- event or to keep the proteins in a metastable chaperones that hand over specific groups of configuration to bind a ligand. While HSP90's clients to HSP90 as well as E3 ubiquitin cytosolic clients and functions have been ligases that degrade misfolded clients. studied in detail in the last decade, this study Besides Hsp70, we found kinase-specific co- provides the first global analysis of the chaperone CDC37 in the nuclear interactome nuclear pool of the chaperone in human cells. of HSP90, along with multiple kinases. By employing three orthogonal and Additionally proteins such as FKBP3 in the independent screens, we identified and interactome may facilitate HSP90's chaperone validated a core functional interactome of activity, similar to other members of the nuclear HSP90 to define the breadth of its FKBP/ immunophilin family. A recent study function. found that NUDCD3 co-chaperone While we focused on proteins identified in all specifically binds Kelch domains of proteins three screens, the individual screens detailed [18]. Given that HCFC1 has a Kelch domain, in our work will be a rich resource for several and HSP90i-sensitive point mutation P134S important aspects of HSP90's nuclear falls in this domain may suggest that HSP90 function. For example, genetic interactors binds HCFC1 via NUDCD3-Kelch exemplify genes which when knocked-down interaction. When HSP90 activity is render cancer cells resistant or sensitive to compromised, clients such as HCFC1 are HSP90 inhibitors. Expression of such genes in ubiquitinated and degraded. Ubiquitin ligases tumors treated with HSP90i may predict such as CHIP and HECTD3 are integral part therapeutic responsiveness among patients. In of HSP90 quality control mechanism. We this regard, it is noteworthy that we identified found several ubiquitin ligases in our nuclear SWI/ SNF complex in multiple screens. interactome, and one/ more of these may be Mutations in SWI/ SNF components have responsible for ubiquitinating clients to been recently linked to tumor sensitivity to constitute the nuclear protein quality control immune therapy [55, 56] suggesting HSP90 module operated by HSP90. inhibitors may be synergistically combined How does HSP90 select its clients in the with immune checkpoint blockade. Similarly, nucleus where all the proteins have already HDACs recovered in our screens have already finished initial aspects of folding and proved to be successful when combined with maturation? HCFC-1 undergoes proteolytic

149 APPENDIX processing in the nucleus as a part of its validated the findings using one metazoan- functional maturation, in a manner similar to specific client protein. Trithorax/ MLL as well as Picornavirus capsid precursor poly-protein (P1). Curiously, Author contributions all three proteins have been shown to be R.S. conceived the project; A.A., B.H., targets of HSP90 chaperoning (this study, [37, D.M.R., G.M., C.M. and R.S. designed the 58], indicating that there might be a common study; A.A., D.S.R., E.C.H., A.K. and K.G. quality control feature associated with performed experiments and interpreted the proteolytic maturation of proteins that results; B.H., M.S., J.M., G.M., and C.M. requires HSP90. Additionally, we recently performed computational analyses, R.S. wrote reported the identification of TRIM28 in the manuscript with input from all the context of SETDB1 and KRAB Zinc finger authors. proteins as a regulatory module that requires HSP90 [38], which can be recapitulated in our screens here (Fig. 2E, Supp. Table 3 and 4). Competing interests TRIM28 and HCFC1, are both metazoan- The authors declare no competing financial specific, and act as a bridging platform interests. between TF and proteins with chromatin- modifying activity. It is likely that structural Data Availability complexity of such a platform necessitates chaperone action and control. The molecular All the deep-sequencing data reported in this determinants by which HSP90 controls these study are deposited in GEO and are available platforms will be an exciting area of future under accession number GSE126151. The studies. data will be available for public upon HCFC1 was named after its function as a host acceptance of the manuscript. The reviewers factor necessary for activation of viral can access the data, go to immediate early promoters, via binding viral https://www.ncbi.nlm.nih.gov/geo/query/acc.c protein 16 (VP16). The finding that HCFC1 gi?acc=GSE126151. critically requires HSP90 for its stability Enter token mduloqmqdpcdzyf into the box. suggests that viral transcription and propagation in host cells will be sensitive to Acknowledgements HSP90 inhibition. Thus HSP90 constitutes an We would like to thank Winship Herr important host protein required for viral (University of Lausanne, Switzerland), Prof. transcription as well as replication [59, 60] Takeharu Nishimoto (Kyushu University, and HSP90 inhibitors could be used as broad Japan) and Christos A. Panagiotidis (Aristotle antiviral agents [61]. Interestingly viral University of Thessaloniki, Greece) for transcription was exquisitely sensitive to low sharing cell lines, protocols and plasmids. nanomolar concentrations of HSP90 inhibitor This work was financially supported by the (Fig. 6C), despite the fact that cellular HSP90 Max Planck Society, German Research concentrations is in the micromolar range Foundation (DFG) through the collaborative [62]. This could be attributed to low nuclear research center CRC992 (Medical concentrations or to specific post-translational Epigenetics), through the research grant SA modifications/ complexes of HSP90 that drive 3190 and through Germany's Excellence viral transcription. A similar idea has been Strategy (CIBSS – EXC-2189 – Project ID proposed to explain why cancer cells are more 390939984) to R.S. C.M. acknowledges sensitive to HSP90 inhibitor more than their financial support from the German Cancer normal counterpart. Thus our study combined Consortium (DKTK). the strengths of three orthogonal approaches to decipher the function of nuclear HSP90 and

150 APPENDIX

1. Hartl, F.U., Molecular chaperones in cellular genetic interactions mediated by the hsp90 protein folding. Nature, 1996. 381(6583): p. chaperone. Cell, 2005. 120(5): p. 715-27. 571-9. 15. McClellan, A.J., et al., Diverse cellular 2. Richter, K., M. Haslbeck, and J. Buchner, The functions of the Hsp90 molecular chaperone heat shock response: life on the verge of uncovered using systems approaches. Cell, death. Mol Cell, 2010. 40(2): p. 253-66. 2007. 131(1): p. 121-35. 3. Taipale, M., et al., Quantitative analysis of 16. Millson, S.H., et al., A two-hybrid screen of HSP90-client interactions reveals principles of the yeast proteome for Hsp90 interactors substrate recognition. Cell, 2012. 150(5): p. uncovers a novel Hsp90 chaperone 987-1001. requirement in the activity of a stress- 4. Li, J., J. Soroka, and J. Buchner, The Hsp90 activated mitogen-activated protein kinase, chaperone machinery: conformational Slt2p (Mpk1p). Eukaryot Cell, 2005. 4(5): p. dynamics and regulation by co-chaperones. 849-60. Biochim Biophys Acta, 2012. 1823(3): p. 624- 17. Falsone, S.F., et al., A proteomic snapshot of 35. the human heat shock protein 90 5. Makhnevych, T. and W.A. Houry, The role of interactome. FEBS Lett, 2005. 579(28): p. Hsp90 in protein complex assembly. Biochim 6350-4. Biophys Acta, 2012. 1823(3): p. 674-82. 18. Taipale, M., et al., A quantitative chaperone 6. Freeman, B.C. and K.R. Yamamoto, interaction network reveals the architecture Disassembly of transcriptional regulatory of cellular protein homeostasis pathways. complexes by molecular chaperones. Science, Cell, 2014. 158(2): p. 434-448. 2002. 296(5576): p. 2232-5. 19. Falsone, S.F., et al., A proteomic approach 7. Rutherford, S.L. and S. Lindquist, Hsp90 as a towards the Hsp90-dependent capacitor for morphological evolution. ubiquitinylated proteome. Proteomics, 2007. Nature, 1998. 396(6709): p. 336-42. 7(14): p. 2375-83. 8. Pick, E., et al., High HSP90 expression is 20. Gano, J.J. and J.A. Simon, A proteomic associated with decreased survival in breast investigation of ligand-dependent HSP90 cancer. Cancer Res, 2007. 67(7): p. 2932-7. complexes reveals CHORDC1 as a novel ADP- 9. Hadizadeh Esfahani, A., et al., A systematic dependent HSP90-interacting protein. Mol atlas of chaperome deregulation topologies Cell Proteomics, 2010. 9(2): p. 255-70. across the human cancer landscape. PLOS 21. Tsaytler, P.A., et al., Novel Hsp90 partners Computational Biology, 2018. 14(1): p. discovered using complementary proteomic e1005890. approaches. Cell Stress Chaperones, 2009. 10. Whitesell, L. and S.L. Lindquist, HSP90 and 14(6): p. 629-38. the chaperoning of cancer. Nat Rev Cancer, 22. Moulick, K., et al., Affinity-based proteomics 2005. 5(10): p. 761-72. reveal cancer-specific networks coordinated 11. Trepel, J., et al., Targeting the dynamic by Hsp90. Nat Chem Biol, 2011. 7(11): p. HSP90 complex in cancer. Nat Rev Cancer, 818-26. 2010. 10(8): p. 537-49. 23. Berbers, G.A., et al., Localization and 12. Canonici, A., et al., The HSP90 inhibitor NVP- quantitation of hsp84 in mammalian cells. AUY922 inhibits growth of HER2 positive and Exp Cell Res, 1988. 177(2): p. 257-71. trastuzumab-resistant breast cancer cells. 24. Perdew, G.H., et al., Localization and Invest New Drugs, 2018. 36(4): p. 581-589. characterization of the 86- and 84-kDa heat 13. Vartholomaiou, E., P.C. Echeverria, and D. shock proteins in Hepa 1c1c7 cells. Exp Cell Picard, Unusual Suspects in the Twilight Zone Res, 1993. 209(2): p. 350-6. Between the Hsp90 Interactome and 25. Greer, C.B., et al., Histone deacetylases Carcinogenesis. Adv Cancer Res, 2016. 129: positively regulate transcription through the p. 1-30. elongation machinery. Cell reports, 2015. 14. Zhao, R., et al., Navigating the chaperone 13(7): p. 1444-1455. network: an integrative map of physical and

151 APPENDIX

26. Sawarkar, R., C. Sievers, and R. Paro, Hsp90 36. Scuoppo, C., et al., A tumour suppressor globally targets paused RNA polymerase to network relying on the polyamine-hypusine regulate gene expression in response to axis. Nature, 2012. 487(7406): p. 244-8. environmental stimuli. Cell, 2012. 149(4): p. 37. Tariq, M., et al., Trithorax requires Hsp90 for 807-18. maintenance of active chromatin at sites of 27. Cvoro, A., et al., Distinct Roles of Unliganded gene expression. Proc Natl Acad Sci U S A, and Liganded Estrogen Receptors in 2009. 106(4): p. 1157-62. Transcriptional Repression. Molecular Cell, 38. Hummel, B., et al., The evolutionary 2006. 21(4): p. 555-564. capacitor HSP90 buffers the regulatory 28. Echtenkamp, F.J., et al., Hsp90 and p23 effects of mammalian endogenous Molecular Chaperones Control Chromatin retroviruses. Nat Struct Mol Biol, 2017. 24(3): Architecture by Maintaining the Functional p. 234-242. Pool of the RSC Chromatin Remodeler. Mol 39. Prince, T.L., et al., Client Proteins and Small Cell, 2016. 64(5): p. 888-899. Molecule Inhibitors Display Distinct Binding 29. SU, J.-M., et al., Nuclear Accumulation of Preferences for Constitutive and Stress- Heat-shock Protein 90 Is Associated with Induced HSP90 Isoforms and Their Poor Survival and Metastasis in Patients with Conformationally Restricted Mutants. PLoS Non-small Cell Lung Cancer. Anticancer One, 2015. 10(10): p. e0141786. Research, 2016. 36(5): p. 2197-2203. 40. Hauri, S., et al., A High-Density Map for 30. Isaacs, J.S., Hsp90 as a “Chaperone” of the Navigating the Human Polycomb Epigenome: Insights and Opportunities for Complexome. Cell Rep, 2016. 17(2): p. 583- Cancer Therapy, in Advances in Cancer 595. Research, I. Jennifer and W. Luke, Editors. 41. Wysocka, J., et al., Human Sin3 deacetylase 2016, Academic Press. p. 107-140. and trithorax-related Set1/Ash2 histone H3- 31. Calderwood, S.K. and L. Neckers, Hsp90 in K4 methyltransferase are tethered together Cancer: Transcriptional Roles in the Nucleus, selectively by the cell-proliferation factor in Advances in Cancer Research, I. Jennifer HCF-1. Genes Dev, 2003. 17(7): p. 896-911. and W. Luke, Editors. 2016, Academic Press. 42. Vogel, J.L. and T.M. Kristie, The novel p. 89-106. coactivator C1 (HCF) coordinates 32. Sharma, K., et al., Quantitative proteomics multiprotein enhancer formation and reveals that Hsp90 inhibition preferentially mediates transcription activation by GABP. targets kinases and the DNA damage Embo j, 2000. 19(4): p. 683-90. response. Mol Cell Proteomics, 2012. 11(3): 43. Michaud, J., et al., HCFC1 is a common p. M111.014654. component of active human CpG-island 33. Wang, R., et al., The Hsp90 inhibitor SNX- promoters and coincides with ZNF143, 2112, induces apoptosis in multidrug THAP11, YY1, and GABP transcription factor resistant K562/ADR cells through suppression occupancy. Genome Res, 2013. 23(6): p. 907- of Akt/NF-kappaB and disruption of 16. mitochondria-dependent pathways. Chem 44. Parker, J.B., et al., Host Cell Factor-1 Biol Interact, 2013. 205(1): p. 1-10. Recruitment to E2F-bound and Cell Cycle 34. Brough, P.A., et al., 4,5-diarylisoxazole Hsp90 Control Genes is Mediated by THAP11 and chaperone inhibitors: potential therapeutic ZNF143. Cell reports, 2014. 9(3): p. 967-982. agents for the treatment of cancer. J Med 45. Yu, H., et al., The ubiquitin carboxyl Chem, 2008. 51(2): p. 196-218. hydrolase BAP1 forms a ternary complex 35. Eccles, S.A., et al., NVP-AUY922: a novel heat with YY1 and HCF-1 and is a critical regulator shock protein 90 inhibitor active against of gene expression. Mol Cell Biol, 2010. xenograft tumor growth, angiogenesis, and 30(21): p. 5071-85. metastasis. Cancer Res, 2008. 68(8): p. 2850- 46. Zargar, Z. and S. Tyagi, Role of host cell 60. factor-1 in cell cycle regulation. Transcription, 2012. 3(4): p. 187-92.

152 APPENDIX

47. Alfonso-Dunn, R., et al., Transcriptional 55. Miao, D., et al., Genomic correlates of Elongation of HSV Immediate Early Genes by response to immune checkpoint therapies in the Super Elongation Complex Drives Lytic clear cell renal cell carcinoma. Science, 2018. Infection and Reactivation from Latency. Cell 359(6377): p. 801-806. Host & Microbe. 21(4): p. 507-517.e5. 56. Pan, D., et al., A major chromatin regulator 48. Liu, W., et al., PHF8 Mediates Histone H4 determines resistance of tumor cells to T cell- Lysine 20 Demethylation Events Involved in mediated killing. Science, 2018. 359(6377): Cell Cycle Progression. Nature, 2010. p. 770-775. 466(7305): p. 508-512. 57. Krämer, O.H., S. Mahboobi, and A. Sellmer, 49. Deplus, R., et al., TET2 and TET3 regulate Drugging the HDAC6–HSP90 GlcNAcylation and H3K4 methylation interplay in malignant cells. Trends in through OGT and SET1/COMPASS. Embo j, Pharmacological Sciences. 35(10): p. 501- 2013. 32(5): p. 645-55. 509. 50. Ajuh, P.M., et al., Association of a protein 58. Geller, R., et al., Evolutionary constraints on phosphatase 1 activity with the human factor chaperone-mediated folding provide an C1 (HCF) complex. Nucleic Acids Res, 2000. antiviral approach refractory to development 28(3): p. 678-86. of drug resistance. Genes & development, 51. Glinsky, G.V., O. Berezovska, and A.B. 2007. 21(2): p. 195-205. Glinskii, Microarray analysis identifies a 59. Geller, R., S. Taguwa, and J. Frydman, Broad death-from-cancer signature predicting action of Hsp90 as a host chaperone required therapy failure in patients with multiple for viral replication. Biochimica et Biophysica types of cancer. The Journal of clinical Acta, 2012. 1823(3): p. 698-706. investigation, 2005. 115(6): p. 1503-1521. 60. Wang, Y., et al., Heat-shock protein 90alpha 52. Capotosti, F., et al., O-GlcNAc transferase is involved in maintaining the stability of catalyzes site-specific proteolysis of HCF-1. VP16 and VP16-mediated transactivation of Cell, 2011. 144(3): p. 376-88. alpha genes from herpes simplex virus-1. Mol 53. Goto, H., et al., A single-point mutation in Med, 2018. 24(1): p. 65. HCF causes temperature-sensitive cell-cycle 61. Wang, Y., et al., HSP90: a promising broad- arrest and disrupts VP16 function. Genes spectrum antiviral drug target. Archives of Dev, 1997. 11(6): p. 726-37. Virology, 2017. 162(11): p. 3269-3282. 54. Wysocka, J. and W. Herr, The herpes simplex 62. Kundrat, L. and L. Regan, Balance between virus VP16-induced complex: the makings of folding and degradation for Hsp90- a regulatory switch. Trends Biochem Sci, dependent client proteins: a key role for 2003. 28(6): p. 294-304. CHIP. Biochemistry, 2010. 49(35): p. 7428- 7438.

153 APPENDIX

Supplementary Figure 1. Control experiments for the chemical genetic screen in K562 using HSP90i. (A) Efficiency of doxycycline-mediated shRNA induction in each pool as assessed by DsRED positive cells. (B) Distribution of multiplexed samples (a total of 30) as determined by deep sequencing and barcode demultiplexing. Each color indicates a specific barcode, the percentage of relative contribution to the sequencing run is shown. (C) Average shRNA read numbers in logarithmic scale for indicated conditions. The asterisk indicates the subpool containing the negative control hairpin Renilla.713.

154 APPENDIX

Supplementary Figure 2. Reproducibility of the chemical-genetic screen in K562 using HSP90i. (A) Correlation plot showing the log10 read numbers for two biological replicates of shRNA representation for the indicated condition. Pearson correlation coefficient is indicated. (B) Z-scores of individual shRNA (red boxes) targeting 20 genes with the highest CGIS identified in the screen are shown. The grey histogram depicts normalized distribution of Z-score of all candidates in the screen. Control shRNAs (see Methods) are highlighted as blue boxes. Red vertical lines indicate 66th and 98th percentile in Z-scores on either sides of zero.

155 APPENDIX

Supplementary Figure 3. Physical interactome of nuclear HSP90α in human cells. (A) Cellular compartment gene onthology enrichment analysis of HSP90α-proteins demonstrates a lack of cytosolic contamination. (B) Venn diagram showing the overlap of proteins identified as physical interactors of HSP90α in four independent replicates. The groups of proteins found simultaneously in at least 3 biological replicates are outlined in grey.

156 APPENDIX

Supplementary Figure 4. Normalized expression levels in samples of normal tissue in comparison with primary tumors in 13 different cancer types for HSP90, HCFC1 and HCFC1-recruiting factors. Source data from TCGA (see Methods).

157 APPENDIX

Supplementary Figure 5. (A) Genome browser tracks of HCFC1 gene expression in HEK293 cells. Expression levels are visualized in control conditions (blue) and upon HSP90 inhibition (red). The vertical scale indicates normalized read density in rpkm. (B) BETA analysis (see Methods) for activating/repressive function prediction of HCFC1 target genes in mouse embryonic stem cells. Genes are arranged by the rank of their regulatory score. Genes that are upregulated (red) or downregulated (purple) upon HSP90 inhibition are represented with solid lines. Dotted line represents the nondifferentially expressed genes. A Kolmogorov-Smirnov test is used to calculate the significance of the UP (red) or DOWN (purple) group distribution against the background group. P-value is indicated on the graph.

158