Proteasome Inhibitors against Cancer: Determining Biology and Finding Novel Compounds

Proteasome Inhibitors

against Cancer:

Determining Biology and

Finding Novel Compounds

Arjan H. H. Mofers

Proteasome Inhibitors

against Cancer: Determining

Biology and Finding Novel

Compounds

Arjan Hubert Hendrik Mofers

Division of Drug Research

Department of Biomedical and Clinical Sciences Linköping University, Sweden

 Arjan Mofers, 2020

Cover art adapted from McGuffee & Elcock, PLoS Comput Biol 6(3): e1000694. The image has been photo edited for artistic effect.

Published articles and figures have been reprinted with the permission of the copyright holder.

Printed in Sweden by LiU-Tryck, Linköping, Sweden, 2020

THESIS FOR DOCTORAL DEGREE (Ph.D.) By

Arjan Hubert Hendrik Mofers

Principal Supervisor

Professor Stig Linder Linköping University

Department of Biomedical and Clinical Sciences Karolinska Institute

Department of Oncology-Pathology

Co-supervisor

Docent Pádraig D’Arcy Linköping University

Department of Biomedical and Clinical Sciences

Faculty Opponent

Professor Sonia Lain Karolinska Institute

Dedicated to my mother, to whom I owe all of my virtues and none

of my vices

Abstract

Cancer continues to be a tremendous burden on society in terms of loss of quality of life and life-years. The ubiquitin proteasome system (UPS), responsible for the bulk of protein turnover in the cell, is considered an attractive target in cancer therapy. The proteolytic subunit of the UPS, the 20S, is targeted by three clinically approved anti-cancer drugs: Bortezomib, Carfilzomib, and Ixazomib. Proteins destined for degradation by the UPS are tagged with small protein ubiquitin. The 19S regulatory cap is responsible for recognition and processing of these ubiquitinated substrates, followed by proteolysis by the 20S. Deubiquitination by proteolytic enzymes of the 19S is an essential step in the progression of substrates to progress into the 20S. Our strategy is to pharmacologically inhibit the deubiquitinating enzymes associated with the 19S with the goal of disrupting the degradation of substrates.

In Paper I of this thesis, we explore the occurrence of resistance to the 19S deubiquitinating enzyme inhibitor b-AP15. We find that minimal resistance to this compound occurs, and the small observed induction of resistance is likely mediated by glutathione. We also find that cells that are slow- or non-cycling are particularly insensitive to the treatment. In Paper II we employ screening methods based on the chemical properties of b-AP15 and find that a small subset of the screened compounds are relatively selective UPS inhibitors. In Paper III we employ a different screening methods, based on the gene expression profiles associated with disruption of the UPS. This screen finds several compounds with previously described UPS inhibitory effects but also compounds with different proposed mechanisms of action. In both Paper II and Paper III we reflect on the chemical structure of the compounds and challenges that come with chemical groups that are potentially promiscuously reactive. In Paper IV we explore the validity of b-AP15 as a treatment for acute lymphoblastic leukemia, a form of cancer for which the efficacy of b-AP15 has not been fully elucidated. This thesis contributes to the body of work supporting 19S inhibition in general, and 19S inhibition with b-AP15 in particular, as a promising modality for the treatment of cancer.

Populärvetenskaplig Sammanfattning

Cancer är en allvarlig sjukdom som drabbar många människor och leder till omfattande dödstal och lidande. Till skillnad från många andra sjukdomar, är cancer en sjukdom som kommer från kroppen själv: Cancerceller är patientens egna celler som har muterat för att växa okontrollerat. Mutationer inträffar hela tiden, men vanligtvis upptäcks dem av kroppen som gör sig av med de muterade cellerna. Om en cell muteras och inte tas bort kan den bilda en tumör som så småningom kan störa organs vitala funktioner och leda till dödsfall.

Det finns många sätt att behandla cancer, inklusive kirurgi, strålning och kemoterapi. Om cancern ännu inte har spridit sig kan den ofta behandlas bra med kirurgi eller strålbehandling. Om cancern har spridit sig från sin ursprungliga plats till andra organ blir den mycket svår att behandla med kirurgi eller strålbehandling. Därför är kemoterapi ett viktigt vapen mot cancer. Kemoterapier kan döda cancerceller på många olika sätt beroende på kemiska föreningars eller kombinationer av kemiska föreningars specifika mekanismer, men vad de har gemensamt är att de måste vara selektiva för cancerceller och lämna så många friska celler som möjligt levande. I denna avhandling studerar vi ubiquitin-proteasomsystemet, som är cellens viktigaste system för återvinning av proteiner. Många av cellens funktioner utförs av proteiner, som måste veckas noggrant för att fungera korrekt. När en cancercell tillverkar för många proteiner kommer inte alla att veckas rätt, vilket leder till en giftig miljö med felveckade proteiner i cellen. Ubiquitin-proteasomsystemet hjälper till att rensa bort dessa felveckade proteiner, men när vi använder kemoterapi med kemiska föreningar som blockerar detta system så får cellerna åter för många felveckade proteiner. Detta har visat sig vara ett effektivt sätt att döda cancerceller. Tre kemoterapeutiska läkemedel, Bortezomib, Karfilzomib och Ixazomib, använder sig av denna mekanism och används i kliniken för att behandla vissa typer av cancer.

Dessa kliniskt använda kemiska föreningar blockerar direkt den del av ubiquitin-proteasomsystemet som bryter ned de felveckade proteinerna. Vi försöker istället utveckla kemiska föreningar som är riktade mot den reglerande delen i ubiquitin-proteasomsystemet, som är ansvarig för att känna igen och bearbeta de felveckade proteinerna innan de bryts ned. Slutmålet är, på

nedbrytningen av felveckade proteiner för att döda cancerceller. Vi har tidigare utvecklat en sådan kemisk förening, benämnd b-AP15. I denna avhandling studerar vi hur celler kan bli resistenta mot b-AP15, vilket är viktig information för att utveckla läkemedlet som sedan ska användas i kliniken. Vi studerar också hur effektivt detta läkemedel är vid akut lymfatisk leukemi, en sjukdom där effekterna av b-AP15 ännu inte är välförstådda. Vi använder slutligen screeningmetoder för att hitta nya kemiska föreningar som kan hämma proteasomsystemet. I screeningprocessen analyseras många föreningar för att se om de leder till en önskad biologisk effekt och det finns många sätt att skapa en initial grupp av föreningar för att söka efter denna effekt. Vi använder två screeningbibliotek i denna avhandling. Ett bibliotek med föreningar som innehåller samma reaktiva kemiska grupp som finns i b-AP15. Detta hjälper oss att studera om den reaktiva kemiska gruppen i allmänhet kan hämma proteasomsystemet oavsett hur resten av strukturen ser ut, eller om den bara är generellt reaktiv och inte specifik för proteasomen. Det andra biblioteket består av föreningar valda baserat på deras genuttryck. Vi använde en online-databas för att hitta vilka gener celler uttrycker som svar på behandling med kända proteasomhämmare och sökte sedan efter andra föreningar som leder till att en liknande uppsättning gener uttrycks men som inte tidigare har beskrivits som proteasomhämmare.

Popular summary

Cancer is a serious disease that affects many people and causes tremendous death and suffering. Unlike many other diseases, cancer is a disease that grows from the body itself: Cancer cells are the patient’s own cells that have mutated to grow out of control. Mutations happen all the time, but usually the body detects them and gets rid of the mutated cells. If a cell becomes mutated and does not get taken out, it can form a tumor that can eventually interfere with the vital function of organs and lead to death.

There are many ways to treat cancer, including surgery, radiation therapy and chemotherapy. When cancer has not spread yet, it can often be treated well with surgery or radiation therapy. When cancer has spread from the original site to other organs it becomes very difficult to treat for a surgeon or radiologist. Therefore, chemotherapy is an important weapon against cancer. Chemotherapy can kill cancer cells in many different ways depending on the specific chemical compound or combination of compounds, but what they have in common is that they must be selective for cancer cells and leave as many healthy cells as possible alive. In this thesis, we studied the ubiquitin proteasome system, which is the most important system in the cell for the recycling of proteins. Many of the functions of the cell are performed by proteins, which must be carefully folded to function properly. When a cancer cell makes too many proteins it cannot fold them all properly, which creates a toxic environment of misfolded proteins in the cell. The ubiquitin proteasome system helps to clear out these misfolded proteins, but when we use chemical compounds that block this system misfolded proteins once again build up. This has shown to be an effective way to kill cancer cells Three drugs, Bortezomib, Carfilzomib and Ixazomib, are such proteasome inhibitors.

These clinically used compounds directly block the part of the ubiquitin proteasome system that degrades the proteins. We instead tried to develop compounds that target the regulatory particle of the ubiquitin proteasome system, which is responsible for recognizing and processing misfolded proteins before they are degraded. The end goal was, similarly to the clinically approved

We previously developed such a compound, termed b-AP15. In this thesis, we studied how cells can become resistant to b-AP15, which is important information for developing the drug to be used in the clinic. We also studied how effective this drug is in acute lymphoblastic leukemia, a disease in which the effects of b-AP15 are not yet well understood. We lastly employed screening methods to find new compounds that can inhibit the proteasome. Screening is the process where many compounds are analyzed to see whether they induce a desired biological effect, and there are many ways to create an initial group of compounds to search for this effect. We used two screening libraries in this thesis. One library contains compounds that contained the same reactive chemical group found in b-AP15. This helped us to study whether this reactive chemical group can in general cause proteasome inhibition no matter what the rest of the structure looks like, or whether it is just generally reactive and not specific to the proteasome. The second library consisted of compounds selected based on the gene expression they induce. We used an online database to find what genes cells express in response to known proteasome inhibitors, and then searched for other compounds that express a similar set of genes but have not necessarily been described as proteasome inhibitors.

List of papers

Following are the papers included in this thesis. They are referred to by their numbers in the text.

I. Analysis of determinants for in vitro resistance to the small

molecule deubiquitinase inhibitor b-AP15.

ARJAN MOFERS, Paola Perego, Karthik Slevaraju, Laura Gatti, Joachim Gullbo, Stig Linder, Pádraig D’Arcy.

PLoS ONE 14(10): e0223807.

II. Cytotoxic unsaturated electrophilic compounds commonly target the ubiquitin proteasome system. Karthik Selvaraju,

ARJAN MOFERS, Paola Pellegrini, Johannes Saomonsson, Alexandra Ahlner, Vivian Morad, Ellin-Kristina Hillert, Belen Espinosa, Elias S.J. Arnér, Lasse Jensen, Jonas Malström, Maria V. Turkina, Pádraig D’Arcy, Michael A. Walters, Maria Sunnerhagen, Stig Linder.

Scientific Reports (2019) 9:9841

III. A screen for novel proteasome inhibitors based on gene expression patterns.

ARJAN MOFERS, Karthik Selvaraju, Johannes Gubat, Pádraig D’Arcy, Stig Linder.

Manuscript

IV. Activation of the unfolded protein response (UPR) in acute lymphocytic leukemia cells by proteasome inhibitors.

Paola Pellegrini, Elena Faustini, Karthik Selvaraju, ARJAN MOFERS, Xiaonan Zhang, Jens Ternerot, Alice Schubert, Stig Linder, Pádraig D’Arcy.

The following publications are not included in this thesis:

Proteasome-associated deubiquitinases and cancer. [Review

Article]

ARJAN MOFERS, Paola Pellegrini, Stig Linder, Pádraig D’Arcy. Cancer Metastasis Rev. 2017; 36(4): 635–653.

MYC is downregulated by a mitochondrial checkpoint mechanism.

Xiaonan Zhang, ARJAN MOFERS, Per Hydbring, Maria Hägg Olofsson, Jing Guo, Stig Linder, Padraig D'Arcy

Oncotarget. 2017; 8:90225-90237.

Proteasome inhibitor b-AP15 induces enhanced proteotoxicity by inhibiting cytoprotective aggresome formation.

Ellin-Kristina Hillert, Slavica Brnjic, Xiaonan Zhang, Magdalena Mazurkiewicz, Amir Ata Saei, ARJAN MOFERS, Karthik Selvaraju, Roman Zubarev, Stig Linder, Pádraig D’Arcy.

Contents

Abbreviations ... 1

Introduction ... 7

The ubiquitin-proteasome system ... 10

26S proteasome ... 13

19S regulatory particle ... 14

20S core particle ... 17

Targeting the UPS in cancer ... 18

DUB inhibition as a treatment of cancer ... 24

Aims ... 29

Results & discussion ... 31

Paper I ... 31 Paper II ... 35 Paper III ... 45 Paper IV... 53 Conclusions ... 59 Paper I ... 59 Paper II ... 59 Paper III ... 59 Paper IV... 59

Considerations for the future ... 61

Acknowledgements ... 63

Abbreviations

15Δ-PGJ2 15-deoxy-Δ12,14-prostaglandin J2

Å Ångström

Akt Protein Kinase B

ALL Acute Lymphoblastic Leukemia

ASP-451 Asparaginase residue 451

ATF Activating Transcription Factor

BCL B Cell Lymphoma (protein)

BCRP Breast Carcinoma Resistance

Protein

BiP Binding Immunoglobin Protein

BL Blocking Loop

BTK Bruton’s Tyrosine Kinase

CHOP C/EBP Homologous Protein

CHX Cycloheximide

CMap Connectivity Map

COMPARE Compared and Expressed as

Correlation Coefficients

CYS-114 Cysteine residue 114

Da Dalton

DUB Deubiquitinating Enzyme

DUSP Dual Specificity Phosphatase

E1 E1 Enzymes

E3 E3 Enzymes

E4 E4 Enzymes

EGFR Epidermal Growth Factor Receptor

eIF Eukaryotic Translation Initiator

Factor

EMA European Medicines Agency

ER Endoplasmic Reticulum

ERAD ER Associated Protein Degradation

FBW7 F-Box and WD Repeat

Domain-Containing 7

FDA Food and Drug Administration

Ftase Farnesyltransferase

GADD34 Growth Arrest and DNA

Damage-Inducible Protein 34

GDP Guanosine Diphosphate

GI50 50% Growth inhibitory

concentration

GO Gene Ontology

GRP78 Glucose Regulation Protein 78

GSEA Gene Set Enrichment Analysis

GTP Guanosine Triphosphate

HDAC Histone Deacetylase

HDM2 Human Double Minute 2

HECT Homologous to the

E6-AP Carboxyl Terminus Domain

HER2 Epidermal Growth Factor Receptor

2

HER4 Epidermal Growth Factor Receptor

4

HIS-435 Histidine residue 435

HMOX1 Heme Oxygenase 1 (Gene)

3

IAP Inhibitor of Apoptosis Protein

IC50 50% Inhibitory Concentration

IC90 90% Inhibitory Concentration

IE Immediate Early

IKK IκB Kinase

IΚB Nuclear Factor of Kappa Light

Polypeptide Gene Enhancer in B-cells Inhibitor

IRE1 Inositol-requiring protein 1

ISRIB Integrated Stress Response

Inhibitor

JAMM JAMM/MPN Domain-Associated

Metallopeptidase

JNK c-JUN N-terminal Kinase

k11 Lysine residue 11 k27 Lysine residue 27 k29 Lysine residue 29 k33 Lysine residue 33 k48 Lysine residue 48 k6 Lysine residue 6 k63 Lysine residue 63 L1000FWD L1000 Fireworks Display

LC-MS Liquid Chromatography - Mass

Spectrometry

LOF Loss of Function

LUBAC Linear Ubiquitin Chain Assembly

Complex

M1 Methionine residue 1

MALDI-TOF Matrix-Assisted Laser

Desorption/Ionization - Time-of-Flight Mass Spectrometer

MCL1 Myeloid Cell Leukemia 1 (Protein)

MCPIP Chemotactic Protein-Induced

MDR1 Multi-Drug Resistance Mutation 1

Met Methionine

MINDY Motif Interacting with

Ub-Containing Novel DUB

MJD Machado-Joseph Disease Protein

Domain Protease

MMP2 Matrix Metalloprotease 2

MOA Mechanism of Action

mTOR Mammalian Target of Rapamycin

NCI60 National Cancer Institute 60

NEDD8 Neuronal Precursor Cell-Expressed

Developmentally Downregulated Protein 8

NFκB Nuclear Factor

Kappa-Light-Chain-Enhancer of Activated B Cells

OUT Ovarian Tumor

Domain-Containing Protease

PAINS Pan-Assay Interference

Compounds

PARP Poly (ADP-Ribose) Polymerase

PBX1 Pre-B-cell Leukemia Transcription

Factor 1

PERK Protein Kinase RNA-like

Endoplasmic Reticulum Kinase

PgP P-Glycoprotein

POH1 Human Pad1 Homologue

polysome Poly Ribosome

PSMA Proteasome Subunit Alpha

PSMB1 Proteasome Subunit Beta

PSMD8 Proteasome Subunit Delta

RING Really Interesting New Gene Finger

Domain

5

RPT Regulatory particle ATPase

SERCA Sarco/Endoplasmic Reticulum

Ca2+ ATPase

SKP2 S-phase Kinase-Associated Protein

2

SMAD4 Mothers against decapentaplegic

homolog 4

suc-LLVY-2R110 Succinyl - Leucine Leucine Valine Tyrosine – 2 Rhodamine 110

TNFα Tumor Necrosis Factor Alpha

TrxR Thioredoxin Reductase

UBA Ubiquitin-Associated Domain

UBL Ubiquitin-Like Domain

Ubp Ubiquitin-Specific Protease

UCH Ubiquitin Carboxyl-Terminal

Hydrolase

UCHL Ubiquitin Carboxyl-Terminal

Hydrolase L

UPR Unfolded Protein Response

UPS Ubiquitin Proteasome System

USP Ubiquitin Specific Peptidase

VHL Von Hippel-Lindau (disease,

protein)

VLX Vivolux

XBP1 X Box-Binding Protein 1

XBP1s Spliced X Box-Binding Protein 1

ZUFSP Zn-Finger and UFSP Domain

7

Introduction

Cancer is one of the greatest challenges in medicine and research. The disease is responsible for over 9 million fatalities annually and one in every six deaths is attributed to cancer (1). It is the most common natural cause of death in developed countries, and the second most in developing countries (2). Cancer can have many causes including internal factors such as inherited and acquired genetic mutations as well as external chemical, biological, or physical agents. No matter what causes cancer to develop, the essence is the same for all: It is the body’s own cells that have stopped working to the benefit of the organism, and are now chasing their own evolutionary success.

Nothing in Biology Makes Sense Except in the Light of Evolution

Dr. Theodosius Dobzhansky

Healthy cells have regulatory systems that keep them from straying from their assigned role. Crucial facets of maintaining the integrity of tissues, such as whether cells can migrate to other tissues and the rate at which cells divide, are under strict control. When genes that drive cells to divide, termed proto-oncogenes, become mutated or are overexpressed they turn into active cancer-causing oncogenes. The cell can antagonize this effect by expressing tumor-suppressor genes that slow cell division back down (3, 4). Even when cell division gets out of hand, there is a limit set to the number of DNA replications a cell can progress through mediated primarily by telomeres. Telomeres are repetitive DNA sequences at the ends of the chromosome arms, and a part of the telomere is lost in every cell division. When telomeres become too short, the cell recognizes that it has reached its replicative limit.

A healthy cell behaves altruistically, and if a critical amount of mutations is detected or if the telomeres have become too short it has three options to avoid entering a cancerous state. One such mechanism is senescence (5-7), which can be described as a dormant state of the cell. Another is apoptosis, the self-killing of the cell by intrinsic pathways (5, 8, 9). In the final line of defense there is the body’s immune system, which can detect and kill cancer cells. Cells that are progressing towards becoming cancerous express small fractions of mutated proteins on their cell surface, termed tumor-specific antigens. These antigens are foreign to the immune system, which triggers immune cells to recognize them as dangerous and kill the cancer cells (10, 11). For the propagation of cancer cells, they undergo changes to support their accelerated metabolism. Cancer cells alter their metabolic profile including a switch to glucose-dominant metabolism (the so-called Warburg effect (12)), and they stimulate nearby blood vessels to expand towards the tumor to create a more nutrient-rich environment (13, 14). All these aspects have together been termed “the hallmarks of cancer” (Fig. 1) (15, 16), as they describe common traits that all cancer cells acquire (n.b. induction of angiogenesis is naturally specific to solid tumors). It should be noted that these processes are complex and fluid – not all processes are active or required at the same time and they do not necessarily occur sequentially.

9

These mechanisms provide a threshold for the development of cancer: A single mutated gene is generally not sufficient for a cell to circumvent its intrinsic control mechanism and cell killing by the immune system. Consequently, several “hits” (mutations) are required to overcome these mechanisms. To become cancerous cells must divide in an uncontrolled manner. This means that growth is accelerated (oncogenes are active), that inhibition of growth is deactivated (tumor suppressors are attenuated), and that telomeres are restored (predominantly through induction of telomer lengthening enzyme telomerase). When this accelerated growth is achieved, cells must avoid entering a senescent state as well as disabling apoptotic pathways. To avoid being killed by the immune system, cancer cells acquire mutations that prevent the presentation of tumor antigens on their cell surfaces. Cancers are found to be driven by an average of 4 genetic mutations, but more or fewer mutations are needed as drivers depending on the type of cancer (17). Although only a relatively small amount of mutations is strictly required for cancer to develop, due to loss of control mechanisms cancer cells are typically characterized by a much greater amount of mutations. While not required for cancer to propagate, these mutations can improve the evolutionary strength of the cells in which they occur by providing them with numerous pro-carcinogenic traits.

Cancer thus exploits the fundamental logic of evolution unlike any other illness. If we, as a species, are the ultimate product of Darwinian selection, then so, too, is this incredible disease that lurks inside us.

Dr. Siddhartha Mukherjee, The Emperor of All Maladies: A Biography of Cancer

A cancer type that is studied in more detail in this thesis is acute lymphoblastic leukemia (ALL). ALL is a predominantly pediatric disease with an incidence peak between 2 and 5 years of age and a second peak in incidence occurring around 50 years of age and later (18). Disease outcome is decidedly poorer for the older age cohort of patients, with young patients achieving long-term remission in around 80% of cases and older patients in around 10-40% of cases (18, 19). ALL is a cancer of both T-cell and B-cell lymphoid lines, with the B-cell background being more common (20). Various non-random genetic translocations are found to lie at the basis of ALL. The most common translocation is t(12;21), found in around 25% of cases (21, 22). This translocation generates a fusion gene of genes TEL and AML1, which inhibits the differentiation of progenitor cells and apoptosis

(23). Other translocations include t(1;19) occurring in around 5% of cases producing fusion gene TCF3-PBX1, t(9;22) occurring in around 1.6% of cases producing fusion gene BCR-ABL1, and t(4;11) occurring in around 1.6% of cases producing fusion gene MLL:AF4 (24-29). These translocations should be viewed as initiators of ALL, but they are not by themselves sufficient to generate leukemia (19). In older ALL patients, survival is poor and there is an urgent need for novel treatment modalities. While long-term survival for pediatric ALL patients approaches 90% in standard-risk patients, survival following relapse remains poor and standard-of-care treatment is genotoxic which poses a significant risk for developing cancer later in life (30).

The traits that make a cell cancerous, also represent opportunities for the treatment of the disease. Cancer treatment must necessarily exploit the differences between healthy and cancerous cells, lest the treatment is as toxic to the patient as it is to the tumor. Various strategies that exploit the hallmarks of cancer exist. Examples are mitotic inhibitors and genotoxic agents which are dramatically more toxic to cells with aberrant cell cycling, inhibitors of angiogenesis that attenuate the generation of novel blood vessels to supply the metabolically overactive cancer cells, and retraining the immune system to again recognize and kill cancer cells. This thesis concerns the use of proteasome inhibitors in the treatment of cancer, including but not limited to ALL. The ubiquitin proteasome system (UPS) is the primary system of the cell for the degradation and recycling of proteins. The targeting of the UPS affects not only the metabolism of proteins but also the way the cell regulates its response to its environment. The effect of proteasome inhibition is thus multi-factorial. Before further assessment of the role of proteasome inhibitors, it is first outlined how this complex system functions under normal conditions.

The ubiquitin-proteasome system

Maintaining control over the turnover of proteins is of crucial importance to the normal working of the cell (31). The UPS is responsible for regulating the turnover of various important factors that control processes including the cell cycle (32), apoptosis (33), and cell differentiation (34). Loss of proteasome activity can cause defects in these pathways. The UPS further plays a crucial role in protein homeostasis by clearing misfolded proteins from the cell, preventing proteotoxic stress (35, 36).

11

Figure 2: Top: Schematic representation of the ubiquitination of substrates by E1, E2, E3, and E4 enzymes. Bottom: Ubiquitinated proteins are recognized by the proteasome and degraded to small peptides and free ubiquitin.

Fundamentally, the UPS consists of a sequence of enzymatic reactions that selectively marks proteins with ubiquitin, a small protein of 8.6 kDa. Addition of ubiquitin to substrate proteins functions as a tag that determines the faith of the substrate. Tags that signal for the destruction of the protein are recognized by the 26S, the proteolytic complex at the center of the UPS. The process of tagging proteins with ubiquitin occurs in three sequential steps by specific classes of enzymes (Fig. 2).

o E1 enzymes (E1) are responsible for the activation of ubiquitin which is a two-step process. In the presence of ATP the E1 catalyze acyl-adenylation of the C-terminus of ubiquitin. The active cysteine residue of E1 then attacks the adenylated C-terminus resulting in a high-energy thioester bond intermediate. In humans only two enzymes are known to belong to the E1 class.

o E2 enzymes (E2) bind the E1-ubiquitin complex, which places it in the appropriate position to catalyze the trans-thioesterrification of ubiquitin to the active cysteine residue of E2. Often described as merely a passive intermediate in the ubiquitination process, E2 can in fact also play a role in substrate recognition (37). In humans, around 40 enzymes are known to belong to E2.

AMP ATP Ub Ub Ub Ub Ub Ub Ub Ub + E1 E2 E3 E3 E1 E2 Ub Ub Ub Ub Ub Ub Ub Ub Proteasome Ub E4

o E3 enzymes (E3) interact with the E2-ubiquitin complex to catalyze the generation of an isopeptide bond between a lysine residue of the substrate and the C-terminal carboxyl group of ubiquitin. A sub-group of E3, also termed E4, are responsible for the prolongation of the ubiquitin tag into polyubiquitin chains. In humans, at least 600 enzymes are described to belong to E3 (38).

The scope of enzymes belonging to each group is indicative of the role of the enzyme in substrate selection. With only two E1s and around 40 E2s, the effect that these enzymes can have on substrate selection is limited. E3s are a far more diverse class of enzymes, consisting of three subfamilies based on sequence homology: RING, HECT, and RING-between-RING ubiquitin ligases. Hybrids between these families have also been described (39). This diversity in E3s allows for individual E3s to show relatively high substrate selectivity. By regulating this considerable network of E3s, the cell exerts control over which substrates become ubiquitinated.

Ubiquitination has roles outside of the UPS, and ubiquitination does not necessarily signal for proteolysis. It can code for various signals depending on how many ubiquitin tags are present on a protein and, if a polyubiquitin chain exists, how the chain is linked. Tagging of proteins with one ubiquitin tag (monoubiquitination) or multiple individual ubiquitin tags on different residues of the substrate protein (multi-monoubiquitination) is best described to regulate in protein-protein interaction and activation or deactivation of proteins (40, 41). More recent studies indicate that monoubiquitination and multi-monoubiquitination can also code for destruction by the 26S proteasome (38). The faith of polyubiquitinated protein, describing substrates tagged with at least one elongated chain of multiple ubiquitin proteins, further depends by which lysine residue the ubiquitin tags are linked. There are seven potential lysine residues present in ubiquitin through which polyubiquitin chains can be linked. Lysine-11 (K11), lysine-27 (K27) lysine-29 (K29), and lysine-48 (K48) linked polyubiquitin chains are the best described to code for proteolysis by the proteasome. The other lysine linkages, lysine-6 (K6), lysine-33 (K33), and lysine-63 (K63), play major roles in protein localization and trafficking as well as protein activity (42). One special case must be noted of the complex termed linear ubiquitin chain assembly complex (LUBAC). To knowledge, this complex is uniquely capable of forming linear ubiquitin chains by linking ubiquitin tags not by any of the lysine residues, but instead by the N-terminal methionine-1 (M1) (43). LUBAC and linear ubiquitin are best studied in the regulation of immune response by mediating activity of the TNFα/NFκB pathway (44, 45).

13

Ubiquitination is not a terminal process and can be reversed. A class of isopeptidases known as deubiquitinating enzymes (DUBs) break the isopeptide bond between either ubiquitin and the substrate, or between two ubiquitin proteins in a polyubiquitin chain. A total of 102 DUBs have been described for humans, subclassified into eight subfamilies (46, 47): Ub-specific protease (USP), Ub carboxy-terminal hydrolase (UCH), ovarian tumor domain-containing protease (OUT) Machado-Joseph disease protein domain protease (MJD), JAMM/MPN domain-associated metallopeptidase (JAMM), motif interacting with Ub-containing novel DUB (MINDY)(48), chemotactic protein-induced protein (MCPIP)(49), and Zn-finger and UFSP domain protein (ZUFSP)(50-53). The USP subfamily is the most diverse class of DUBs with 54 described members. An interesting aspect of this subfamily is that they commonly possess protein-protein interaction domains, which gives them considerable substrate specificity (54, 55). Conversely, USPs show almost no preference towards the characteristics of the ubiquitin chain (56, 57). Other DUBs, such as most JAMM and MINDY family DUBs, have little-to-no substrate selectivity but instead are specific for ubiquitin chain linkage types (57-59). Overall, the DUBeome represents a diverse and complex group of proteins which in turn represents a large amount of the complexity of the faith of protein substrates.

26S proteasome

Proteins that have been tagged with appropriate ubiquitin code are transported to the 26S proteasome for ATP-dependent destruction. The 26S proteasome consists of two sub-units: The regulatory 19S cap, and the proteolytic 20S core particle (Fig. 3). One 20S core particle can be flanked by one or two 19S caps. Other complexes can cap the 20S, with various functions in the cell (60, 61). These types of complexes are outside the scope of this thesis.

Figure 3: Schematic representation of a polyubiquitinated substrate progressing to degradation by the 26S proteasome. Indicated are the three 19S DUBs: USP14 bound to RPN1, UCHL5 bound to RPN13, and POH1.

19S regulatory particle

The 19S regulatory particle functions as a gatekeeper for the proteolytic 20S core particle, allowing only appropriately ubiquitin tagged and processed proteins to progress to degradation. The 19S itself consists of two distinguishable structures termed the “lid” and the “base” (Fig. 3). The lid consists of 9 regulatory particle non-ATPase subunits (RPN): RPN3, RPN5-9, RPN11, RPN12, and RPN15. The base consists of four RPNs and six regulatory particle ATPases (RPTs): RPN1, RPN2, RPN10, RPN13, and RPT1-RPT6. The 19S lid, located most distal from the 20S core particle, is primarily responsible for substrate deubiquitination. The 19S base, located between the 19S lid and the 20S core particle, is primarily responsible for the unfolding of the substrate and the opening of the 20S core particle to allow for substrate translocation into the catalytic core.

19S base α ring α ring β rings 19S lid 19S 20S DIstal chain trimming

Base cleaving USP14 UCHL5 Ub Ub UbUbUbUb Ub UbUbUbUb RPN1 RPN13 POH1

15

Recognition of substrate protein is initiated by binding to the two ubiquitin receptors present in the 19S base, RPN10 and RPN13. Substrate recognition can further be mediated by proteins containing ubiquitin-like (UBL) and ubiquitin-associated (UBA) domains. UBL/UBA proteins interact with the proteasome through their UBL domains, and their UBA domain interacts with the ubiquitin tags of substrate proteins, allowing them to influence substrate recognition and transport to the proteasome (62-64). The 20S has a maximum internal width of around 53Å, but the entrance is as narrow as 13Å (65). This necessitates the deubiquitination and unfolding of most proteins to enter the 20S. Deubiquitination at the proteasome is mediated by three 19S associated DUBS. POH1/RPN11 is an integral subunit of the 19S lid, while UCHL5/UCH37 and USP14/Ubp6 are facultatively associated with the proteasome (66-70). Once deubiquitinated, unfolding of substrates and the opening of the 20S gate is facilitated by RPT1-RPT6 of the 19S base (71).

POH1 is a zinc-dependent JAMM family DUB. Besides being a DUB, POH1 is a structural element of the 19S, and knock-down of this protein prevents correct 26S assembly (72). POH1 is also essential as an enzyme, as a catalytically muted from of POH1 inhibits the UPS and is lethal to cells (73). The role of POH1 at the 19S is the deubiquitination of substrate proteins for entry into the 20S (68). POH1 is likely essential because it cleaves ubiquitin chains at their base, completely clearing the substrate of ubiquitin. Opposed to this action, 19S-associated cysteine DUBs UCHL5 and USP14 trim ubiquitin chains distally (74). Based on these findings it is likely that absent or dysfunctional POH1 completely blocks the ability of the 19S to clear proteins of ubiquitin, and thus prevents the progression of substrates into the 20S.

UCHL5 contains a C-terminal extension which acts as an active-site cross-over loop, which facilitates auto-inhibition of the free form of this enzyme. This autoinhibition is relieved when the C-terminal region of UCHL5 binds to the C-terminal region of RPN13 (Fig. 3) (75, 76). UCHL5 engages in disassembly of K48 polyubiquitinated substrates from the distal end of the chain (77). UCHL5 is found to be upregulated in various malignancies and its overexpression is associated with worse clinical outcome, underlining its importance in the maintenance of proteostasis in cancer cells (78-82).

Figure 4: Molecular structure of USP14 with ubiquitin bound (yellow). Thumb (aquamarine), palm (green), and finger (red) regions are color-coded. Blocking loops 1 (pink) and 2 (orange) are also indicated, here in the "open" position to facilitate ubiquitin binding. Magnifier: Close-up of the catalytic triad of USP14 (n.b. orientation is adjusted for visibility of residues). Source file: 2AYO (https://www.rcsb.org/structure/2AYO).

Similarly to UCHL5, USP14 is auto-inhibiting to avoid aberrant deubiquitinating activity. In USP14 this catalytic muting is facilitated by two blocking loops, termed BL1 and BL2. Crystal structures have determined that USP14 associating with the 19S in itself is insufficient to reveal the catalytic pocket, and that binding of ubiquitin to USP14 is also required (83). USP14 contains a UBL domain, which allows it to associate with RPN1 at the base of the 19S (Fig. 3). USP14 is normally only found to be associated with between 10-20% of the 26S proteasomes, but in contrast with UCHL5 this number is highly inducible by the accumulation of ubiquitinated proteins (84). This is likely driven by USP14 association to the 19S stimulating multiple proteasome activities and enhancing substrate degradation (85, 86). In addition to its 9 kDa UBL domain, USP14 consists of a 45 kDa catalytic domain. The catalytic domain is described as a right-handed finger-palm-thumb structure, in which the open surface from the finger domain to the thumb domain is predicted to bind ubiquitin (Fig. 4). The palm domain contains several surface loops, including BL1 and BL2

17

located in the palm region. The catalytic triad is natively in sufficient proximity to each other to be catalytically active, underlining the importance of the blocking loops to regulate activity (83).

Holistically, USP14 and UCHL5 likely serve two major roles: 1) The recycling of ubiquitin by hydrolyzing polyubiquitin chains from the distal end down to monoubiquitin (74), and 2) controlling the efficiency of the 26S by stimulating degradation (87). The exact way USP14 and UCHL5 influence the efficiency of substrate degradation is unknown. Small molecule USP14 inhibitors have been proposed to both enhance and attenuate substrate degradation, which is likely highly dependent on the precise manner in which the compounds interact with USP14 (88-90). The yeast homolog of USP14, Ubp6, has been shown to non-catalytically inhibit substrate deubiquitination by RPN11 (POH1 in humans) leading to a delay in degradation (91), emphasizing that USP14 is a complex mediator and that targeting specific regions of the protein is likely to lead to dramatically different biological effects. The interplay between USP14 and UCHL5 is an ongoing topic of investigation. It has been shown that knock-down of either enzyme individually can be rescued by the activity of the other enzyme (87). Conversely to this notion of redundancy, catalytically muted forms of the enzymes have shown to lead to the stabilization of different substrates (92). Overall the function of the 19S associated DUBs, particularly the cysteine DUBs, is poorly understood and more insight is necessary to progress UPS-based therapies.

20S core particle

The 20S core particle is a symmetrical cylindrical complex consisting of four stacked rings (Fig 3). The outermost two rings of the 20S are in direct contact with the base of the 19S. These rings are termed α rings and consist of seven distinct α subunits, α1-α7, each. In their native conformation, the α rings close off the 20S which prevents ubiquitous proteolysis of improperly processed substrates (93). The opening of the α rings is mediated by the ATPases of the 19S base and occurs concurrently with the unfolding of the substrate (94-96). The central two rings of the 20S are β rings and consist of seven distinct subunits, termed β1-β7, each. Three of the β subunits are catalytically active: β1 has caspase-like activity, β2 has trypsin-like activity and β5 has chymotrypsin-like activity (97, 98). Through the combined activity of these proteolytic enzymes the substrate is cleaved to peptides with a size ranging from 2 to 35 amino acids (99). The final digestion of these peptides to single amino acids, ready for re-use in protein synthesis or metabolism, is facilitated by free oligopeptidases in the cytosol (100).

Figure 5: Schematic representation of the three pathways involved in the unfolded protein response. ERAD: ER-associated protein degradation.

Targeting the UPS in cancer

While inhibitors of the UPS have been in clinical use for over 16 years (101), the manner in which they induce apoptosis is still not fully understood. Accumulation of misfolded proteins is expected to play a major role, which occurs when the UPS is inhibited and misfolded proteins are no longer degraded leading to a state of proteotoxic stress (102). Misfolded protein accumulation leads to endoplasmic reticulum (ER) stress and induction of the unfolded protein response (UPR) (Fig. 5). The UPR is mediated by three sensors that become activated when they come into contact with misfolded proteins: Inositol-requiring protein 1 (IRE1), protein kinase RNA-like endoplasmic reticulum kinase (PERK) and activating transcription factor 6 (ATF6) (103). Glucose regulation protein 78 (GRP78) / binding immunoglobin protein (BiP) is an ER chaperone that interacts with the three ER stress sensors to facilitate the recognition of misfolded proteins (104, 105). Each of these sensors has distinct downstream pathways. Activated IRE1 has RNAse activity and interacts with the mRNA of X box-binding protein 1 (XBP1) as well as functioning as a general endoribonuclease (106-108). The endoribonuclease activity of IER1 leads

ATF6 PERK IRE1

XBP1

ER chaperones Amino acid biosynthesisRedox Autophagy ER chaperones ERAD Lipid synthesis Protein translation mRNA ATF6 ATF6 ATF4 ATF4 XBP1s XBP1s eIF2α mRNA Misfolded protein

19

transcription factor form of XBP1, also termed spliced XBP1 (XBP1s). XBP1s in turn acts as a transcription factor to induce expression of ER stress response genes.

Activated PERK phosphorylates eukaryotic translation initiator factor 2 (eIF2) α. Together with eIF2β and eIF2γ, eIF2α forms the eIF2 complex, which is responsible for initiating RNA translation by presenting initiator tRNA to the ribosome in a GTP-dependent manner (Fig. 6). Non-phosphorylated eIF2 receives GTP from guanine nucleotide exchange factor eIF2B to facilitate this process. When eIF2 is phosphorylated on its eIF2α subunit, the eIF2 complex greatly increases its binding affinity to eIF2B while simultaneously inhibiting the ability of eIF2B to exchange GDP for GTP. This functionally blocks the capacity of eIF2 to initiate translation (110). Conversely, the translation of certain specific mRNA products, containing specific motifs, is enhanced. Activating transcription factor 4 (ATF4) is preferentially translated by phosphorylated eIF2, leading to greatly elevated levels of ATF4 under this condition (111). Similarly to XBP1s, ATF4 is a transcription factor that activates the transcription of ER stress response genes.

Figure 6: Schematic representation of initiation of translation by eIF2, and inhibition of general translation by phosphorylation of eIF2.

Met eIF2B GDP eIF2B GDP General translation inhibition, initiation of tranlation of speci c mRNA products eIF2 eIF2 eIF2B eIF2 GDP eIF2 GTP eIF2 Met Association with ribosome and initiation of translation Initiatior tRNA GTP

ATF6 is transported from the ER to the Golgi when activated. Here the protein is cleaved, and the cytosolic fraction acts as the third transcription factor of the UPR which can activate transcription of ER stress response genes. In a holistic view, the UPR functions to decrease the misfolded protein load by inhibiting overall transcription (through XBP1s) and translation (through phosphorylation of eIF2α), as well as to increase the capacity of the ER to process misfolded proteins through induction of chaperones by transcription factors XBP1s, ATF4, and ATF6 (103). Under normal conditions, the UPR maintains proteostasis through these pathways. Under conditions of prolonged or severe ER stress the pathway also functions as an apoptotic signal by inducing pro-apoptotic factors such as CHOP (112, 113), which regulates BCL-2 family proteins (114, 115). Cancer cells are particularly liable to experience proteotoxic stress owing to their aberrant overall protein expression through mTOR and other pro-growth pathways (116, 117).

Another form of stress induced by UPS inhibitors is oxidative stress. Oxidation of proteins is a common and intended occurrence in the ER, as many proteins owe their tertiary structure to oxidized disulfide bonds (118). Treatment with first-in-class clinically approved proteasome inhibitor bortezomib (Velcade®, originally PS-341) is associated with mitochondrial depolarization, generation of reactive oxygen species (ROS) and mitochondrial adaptations that eventually drive bortezomib resistance (119, 120). The consequential apoptosis was inhibited by antioxidants but not by caspase inhibitors, indicating that ROS and not the induction of the mitochondrial apoptotic pathways drove apoptosis (119). Inhibition of the UPS through targeting of the 19S is similarly shown to induce oxidative stress and apoptosis which could be rescued by antioxidant treatment (121). The method by which 19S inhibition leads to ROS generation has been theorized to be due to misfolded proteins disrupting the mitochondrial membrane integrity as well as altered regulation of essential mitochondrial proteins (122, 123).

In addition to proteotoxic and oxidative stress, cytotoxicity of UPS inhibition is mediated by its influence on crucial signaling pathways. The UPS not only degrades misfolded proteins but is also leveraged to tightly regulate the half-life of specific proteins to activate and deactivate signaling pathways (124). Indeed, bortezomib was originally developed to study the effects of NFκB inhibition (125). Myriad other cellular processes and pathways have been elucidated to rely on the UPS for their proper functioning, including the cell cycle and DNA damage repair through factors such as p53 (126, 127) and stress pathways including NFκB (125), JNK (128), Akt (protein kinase B) (129), and heat shock factor (HSF)(130).

21

UPS. A summary of known mutations or dysregulations of members of the UPS that are known to drive malignancy is given in Table 1.

Table 1: Summary of UPS subunits involved in the formation of cancer. Where known, the biological effect of the alteration and specific cancer subtypes in which the alteration is involved are indicated. Adapted from Mofers et al (131).

UPS subunit Mutation or

dysregulation Malignancy Reference

E3 HDM2 Various (132-137) Overexpression, loss of tumor suppressor function through p53 FBW7 Leukemia, cholangiocarcinoma, gastrointestinal, and endometrial cancer (138-147)

Mutant, loss of tumor suppressor function through cyclin E, MYC, JUN, and Notch

SKP2 Colorectal, breast,

biliary tract, and prostate cancer. NSCLC

(148-155)

Mutant, loss of tumor suppressor function through p27

VHL Lung cancer, clear-cell carcinoma, VHL disease

(156)

Mutant, loss of tumor suppressor function through HIF

DUB USP1 Fanconi anemia

(leukemia risk factor) (157)

Mutations in FANCD2 DNA repair pathways

DUB (cont.) Stabilizes HDM2, facilitates malignant metabolic profile through fatty acid synthetase activation

USP4 Adenocarcinoma,

breast cancer (159, 160)

Interactions with retinoblastoma

protein, SMAD4, and β-catenin

USP8 - (161)

Regulates expression of EGFR

USP9x Leukemia, myeloma,

lymphoma, and pancreatic cancer (41, 162, 163) Stabilizes β-catenin, SMAD4, and BCL1 family protein MCL1 USP15 Glioblastoma (164) Stabilizes SMAD4 USP18 Leukemia (165) USP19 - (166) Stabilize anti-apoptotic regulators c-IAP1 and c-IAP2

USP28

Stabilizes c-MYC

- (167)

USP7, USP2a,

23

DUB (cont) Stabilize p53

19S USP14 Lung

adenocarcinoma (168, 169)

Stabilization of various regulators including IκB and β-catenin

POH1/RPN11 - (170)

Stabilization of c-JUN

Other Human papilloma

virus onco-protein Cervical, head-and-neck cancer (171, 172)

Viral particle that mimics E3 activity, altering stability of various substrates including p53 and c-MYC

DUB inhibition as a treatment of cancer

It is evident that the UPS influences myriad processes and pathways in the cell through the degradation of numerous substrates. Indeed, various known mutant members of the UPS are known to drive cancers through enhanced stabilization or degradation of crucial signaling proteins (Table 1). Given this long-standing knowledge, it is perhaps surprising that clinical UPS inhibitors have exclusively been “non-specific” UPS inhibitors, i.e. inhibitors of overall protein degradation by targeting the 26S. The lack of clinically approved “specific” UPS inhibitors, i.e. inhibitors of specific DUBs leading to stabilization of specific substrates, has certainly not been due to lack of effort (for reviews see (173, 174)). With the exception of the family of metalloproteases, DUBS are predicted to be cysteine enzymes which are in principle highly druggable, but the development of DUB inhibitors has encountered considerable challenges. Inhibition of cysteines is not inherently difficult from a pharmacological perspective, but the development of a drug that can bind into a DUB’s pocket selectively while still having drug-like properties has so far been elusive. A second problem occurs in the screening for novel DUB inhibitors, which typically happens in screens that are sensitive to non-selective redox or alkylating activity of the DUBs. For screening assays to functioning without these non-specific reactions a reducing environment is often required, which can in turn be to the great detriment to the activity of candidate compounds (175). Lastly, the function of many DUBs is not yet well understood. There is a great deal of unelucidated complexity to how the DUBeome interacts with substrates and how it is affected by the environment in the cell. DUB activity is known to be mediated by allosteric factors, post-translational modifications, association with E2s and E3s as well as other co-factors (176, 177). A better understanding of this complexity is expected to provide increased opportunities in targeting DUBs in clinical setting.

As previously discussed, there are three DUBs associated with the proteasome which makes them particularly unique constituents of the DUBeome. Inhibition of POH1 has received considerable interest in recent years, largely driven by the discovery of zinc chelator and specific POH1 inhibitor Capzimin (178). This compound has shown effectiveness in killing various cell lines including those resistant to bortezomib, but given the essentiality of POH1 the future clinical development of Capzimin and other POH1 inhibitors is in sore need of tolerability data. Our laboratory was first to describe an inhibitor of USP14 and UCLH5, b-AP15 (89, 90). This chalcone-like compound contains two α,β-unsaturated ketones (Fig. 7, Appendix I), which generates two electrophilic “warhead” β-carbons which can engage in Michael addition with nucleophilic groups. The

25

nucleophilic cysteine groups present in USP14 and UCHL5 are the preferential target of b-AP15. As previously discussed, exposed cysteine groups are by no means unique to these two proteins, which argues that selectivity of the compound cannot be driven solely by electrophilicity. It is expected that steric effects mediate the specificity of b-AP15 for the catalytic binding pockets, followed by Michael addition facilitated by the proximity between b-AP15 and the catalytic cysteine residues. b-AP15 has shown to induce accumulation of ubiquitinated proteins without affecting the activity of the 20S, consistent with the expected biological effect of preventing the processing of ubiquitinated proteins by the 19S (89, 90, 121, 173, 179). Even at supra-pharmacological concentrations, b-AP15 did not inhibit general DUB activity (90), indicating that the compound is a relatively selective inhibitor even in the homological DUBeome. A b-AP15 related compound termed 2C was found to interact with USP1 and USP33 with higher affinity than with USP14 (180), which may also be the case for b-AP15. These DUBs are however rare and their inhibition may not be of great consequence to the biological effect of the drugs. Of particular

interest is the effectiveness of b-AP15 against multiple myeloma (MM) (89), a cancer of plasma B cells characterized by high protein expression owing to its lineage’s native role in antibody secretion. Bortezomib was found to be particularly effective in treating MM, likely owed at least in part to the high baseline protein synthesis levels (181). Treatment against cancer inevitably encounters what is perhaps its greatest enemy: Treatment resistance. As effective as any treatment may initially be, if even a minute portion of the population of cancerous cells acquire a trait that helps it to survives survive treatment that newly resistant population can regrow and create a resistant form of cancer. To understand why a subpopulation of a tumor may survive treatment, it is

Figure 7: Top: Generic structure of α,β-unsaturated ketone, with α-carbon and electrophilic β-carbon indicated. Bottom: Structure of b-AP15 with one set of α-carbon and electrophilic β-carbon indicated. Electron-drawing effect by one of the nitro side-groups is also indicated.

e -α

β α β

important to appreciate the heterogeneity of tumors. Most cancers are driven by a relatively small set of genetic mutations (17), but due to genomic and chromosomal instability and other factors mutation rates in cancer cells are high. These high mutation rates result in subpopulations within the tumor that become competitors in the evolutionary arms race (182). Although bortezomib is often an effective drug, patients invariably acquire resistance to this treatment necessitating alternative therapies (183). b-AP15 overcomes resistance to bortezomib in MM cell lines (89, 184), making it a particularly attractive molecule to fill this role. It is thus of great importance for the clinical toolkit against MM and other cancers to develop novel drugs that overcome resistance to bortezomib. We have previously found that cancerous cells derived from a patient resistant to bortezomib are not cross-resistant to b-AP15 (185, 186). Relatively little information is available about the acquisition of resistance by cancer cells to b-AP15, which is the topic of Paper I.

Although b-AP15 and its lead-optimized form VLX1570 have shown promising pre-clinical results, the compounds suffer from poor solubility. To find a compound with better solubility, and potentially also with potent biological effect at lower doses, we engaged in two screening projects to find such new compounds. These two screening projects are described in Paper II and Paper III. These two papers take different approaches to screening. Paper II follows a traditional concept for screening by using a UPS reporter cell line to discover a biological effect in a relatively large set of around 5000 compounds containing the α,β-unsaturated carbonyl motif. Paper III instead aims to find novel compounds by comparing gene expression patterns of candidate compounds to gene expression patterns caused by known proteasome inhibitors and knock-out of crucial UPS subunits. Here, too, we pay special mind to the presence of α,β-unsaturated carbonyl motifs in screened compounds. This structure is a relatively common feature in natural compounds, representing around one-sixth of the space (187). Natural compounds have seen success in recent decades with around half of New Chemical Entries consisting of either natural compounds or direct derivatives thereof (188). In the space of UPS inhibition, naturally occurring electrophilic epoxyketone Epoxomicin is a selective proteasome inhibitor. Its derivative Carfilzomib is an FDA/EMA approved drug against relapsed and refractory MM (189). In the broader cancer pharmacology space, Afatinib, Osimertinib and Ibrutinib are clinically approved compounds with α,β-unsaturated carbonyl motifs that are selective kinase inhibitors targeting epidermal growth factor receptor 1, 2 and 4 (EGFR, HER2 and HER4, Afatinib, Osimertibin) or Bruton’s tyrosine kinase (BTK, Ibrutinib) (190-194). Evidently, electrophilic

27

Yet, as a class, compounds such as enones are considered highly problematic as they are capable of interacting with a broad range of targets and are thus classified as Pan Assay Interference compoundS (PAINS) (195). There are unfortunately no definitive tools available that can a priori predict whether a compound is promiscuous or specific, and this must instead be achieved experimentally by meticulous counter screening. As previously described, ALL patients stand to benefit from the development of UPS inhibition as a treatment modality for this disease. Bortezomib in addition to standard of care regimen carried acceptable toxicity and was found to be of benefit to sub-groups of high-risk relapsed ALL patients (196). This paves the way for additional studies on the use of bortezomib against ALL, perhaps even as a first-line treatment for non-relapsed patients, to avoid the drawbacks of genotoxic therapy. In in vitro models, ALL was found to develop resistance to bortezomib through upregulation of β subunits of the 20S, which is a similar mechanism to that found in MM cells but ALL cells had a greater propensity to develop this phenotype (197). Given the promising results of proteasome inhibitors in the initial treatment of ALL, contrasted with the likelihood of ALL patients developing resistance to bortezomib, there is a niche here for 19S inhibition to be an effective treatment that overcomes bortezomib resistance. Paper IV explores this niche by assessing the biological response of ALL cells to VLX1570 as well as the treatment of ALL cells in the zebrafish model.

In God We Trust, others must provide data.

29

Aims

The overall aim of this thesis is to better understand known 19S inhibitor b-AP15/VLX1570, and to develop novel compounds that can similarly inhibit the UPS. The specific aims of the papers described in this thesis are:

Paper I: To elucidate to what extend cancer cells can develop resistance to

b-AP15, and what the biological processes driving this resistance are.

Paper II: To lever the α,β-unsaturated structure of b-AP15/VLX1570 to

screen for novel compounds with the capacity to selectively inhibit the UPS.

Paper III: To lever the gene expression patterns resulting from UPS

inhibition to screen for novel compounds with the capacity to selectively inhibit the UPS.

Paper IV: To elucidate the activity of VLX1570 in ALL cells based on their

ER stress response and whether VLX1570 is an effective treatment of ALL cells in an in vivo model.

31

Results & Discussion

Paper I

b-AP15 is a candidate for clinical development as an anti-cancer drug, but there is limited information on the grade of resistance acquired against this drug nor the mechanisms which facilitate this resistance. Paper I addresses the question of the grade of resistance cancer cells can develop against b-AP15, and which mechanisms lie at the foundation of this resistance. To answer our research aims we first characterized the proliferation characteristics of the cells under b-AP15 treatment. To determine the level of acquired resistance, we performed a two-step experiment where cells were treated with b-AP15, allowed to recover, and treated with b-AP15 again. By clonogenic assay it was determined that the number of viable clones did not increase between the first and second treatment bout, indicating no strong resistance is acquired by the cells.

To determine the acquisition of resistance, it is crucial to analyze cells on an individual level. To this goal, we performed live-cell imaging on proteasome inhibition reporter cells (MelJuSo UbG76V_{-YFP) as they were}

treated with b-AP15. The observation was made that the vast majority of cells accumulated green signal, indicating their proteasome function is inhibited before cell death occurred. There is no consistent lag time from the point where cells accumulate green signal and the time of subsequent cell death, and there is also no clear correlation between time from last cell division and accumulation of ubiquitinated proteins. However, in the rare events that cells survived the treatment at all, the time from last division was invariably exceptionally long. This finding led us to conclude that while the stage in the cell cycle does not affect the sensitivity of cells to b-AP15, cells that are not- or barely progressing through the cell cycle are uniquely capable of withstanding b-AP15 treatment.

Bortezomib resistant tumors are known to overexpress 20S β subunits or express mutant forms of the β subunits (198, 199). A possible manner of resistance to b-AP15 is the expression of mutant forms or overexpression of one or more enzymes involved in the UPS. Using a probe for 20S catalytic activity we find reduced enzymatic activity, indicating that acquired resistance to bortezomib in this instance leads to a persistently less active β subunit. Such alterations in enzyme activity were absent in b-AP15 resistant cells, where neither the enzymatic activity of the 20S or deubiquitinating enzymes were affected by the resistant status. No adaptations in terms of enzymatic activity are thus found to facilitate resistance to b-AP15.

Drug transporters can facilitate treatment resistance by pumping drugs out of the cell or by preventing drugs from being transported into the cell (200). In this study VLX1570, a lead-optimized variant of b-AP15, was found to be present inside cells at equal rates in resistant and wild type cells. We assessed the role of drug transport by correlating expression of ABC transporters and solute carriers, two families of proteins crucially involved in the transport of various drugs (201), to sensitivity to b-AP15. Sensitivity to b-AP15 of NCI60 cell lines was compared to the expression of ABC

transporters and solute carriers in these cells. No correlation was found between the overexpression of any of the transporters and decreased sensitivity to b-AP15. The overexpression of ABCC1 led to an increase in sensitivity. Cells resistant to doxorubicin, overexpressing MDR1/PgP, and cells resistant to mitoxantrone, overexpressing breast carcinoma resistance protein (BCRP), were not found to be resistant to b-AP15. Overall drug transporters were not found to play a role in resistance to b-AP15.

Enzymes not directly related to the UPS can play important roles in protein homeostasis, and the glutathione pathway plays a major role in detoxification and is known to play a major part in resistance to a broad range of anti-cancer drugs (202). We assessed whether b-AP15 treatment induces cross-resistance to compounds melphalan and doxorubicin, two compounds for which glutathione detoxification is a well-established driver of resistance. Cells resistant to AP15 show a decreased sensitivity to b-AP15 of around 2 fold, while the same cells acquire around 3 fold resistance to melphalan and doxorubicin without being desensitized to these compounds.

To further determine the role of glutathione in this cross-resistance, Buthionine Sulphoximine (BSO) was used in combination with previously described drug treatments. As an inhibitor of γ-glutamylcysteine synthetase, BSO effectively inhibits the biosynthesis of glutathione leading

33

even more sensitive than wild type cells. It was further found that glutathione levels were elevated at baseline in resistant cells compared to wild type cells. When assessing the available levels of glutathione, depletion occurs in resistant- as well as wild type cells even at high concentrations of b-AP15, but resistant cells are better capable of maintaining their glutathione pool. Cells resistant to platinum drugs, characterized by an increase in glutathione metabolism, were found to be slightly more resistant to b-AP15 than wild type cells. From these results, it was taken that glutathione is an important player in the acquisition of resistance to b-AP15.

35

Paper II

b-AP15 has shown to be a promising candidate for the treatment of MM and other cancer types, but it continues to suffer from relatively poor solubility. In paper II, we screened a library of 4896 compounds containing Michael acceptor groups to achieve a better understanding of the features of such compounds. Namely, we were interested in what fraction of such compounds could be classified as proteasome inhibitors, and how selective this effect may be as part of the polypharmacological profile. Given the contentious nature of the subject, we assembled a multidisciplinary team involving structural biologists and medicinal chemists to provide an as-complete-as-possible analysis of the workings of these compounds.

Compounds were initially screened for their potential as proteasome inhibitors with the previously described MelJuSo UbG76V_{-YFP cells by}

live-cell imaging. Of 4896 compounds, 141 were cytotoxic at a screening concentration of 5 µM. 59 of these cytotoxic compounds induced accumulation of Ub-YFP at 5 µM, of which 28 also showed accumulation of ubiquitinated proteins by western blotting. A final selection of 10 compounds remained which were cytotoxic and caused accumulation of ubiquitinated proteins by western blot, and at least 50% of cells were positive for the Ub reporter in the live-cell imaging experiment.

As addressed in the introduction, PAINS are compounds which are predicted to be promiscuous and have a high likelihood of showing up in screens as false-positives according to a chemical classification system. Out of the 10 screening hits, 4 compounds were determined to be potential PAINS by the FAFDrugs4 algorithm. Out of the 4 eneamides in the 10 hits, 3 were classified as PAINS. Eneamides have received increased attention by drug developers in recent years in the form of syrbactins, a class of natural compounds that can inhibit the UPS and overcomes bortezomib resistance. (203, 204) The remaining 6 compounds were all enones, of which 1 was classified as a PAIN. In the library of 4896 compounds, enones

were proportionally overrepresented as proteasome inhibitors with only one compound flagged as a PAIN. This finding was somewhat surprising, given that enones are classified as one of the “worst offenders” of assay interference through their potential of cross-linking compounds through covalent bonding (205). Nevertheless, caution is warranted when developing such compounds. Therefore a range of experiments to validate these compounds as proteasome inhibitors without major other modes of operation was conducted.

UPS inhibition manifests primarily as the accumulation of polyubiquitinated proteins. This coincides with increased expression in markers for proteotoxic stress such as inducible chaperone heat shock protein (HSP) 70B’, and with increased expression in a marker for oxidative stress through detoxification enzyme heme oxygenase (HO-1). p21 plays a pivotal role in cell cycle progression and is tightly regulated by the UPS to avoid cell cycle distress (206). As such, p21 is a classical UPS substrate that serves as an additional marker for UPS inhibition. Generally, increases in HSP70B’, HO-1 and p21 are observed with hit compound treatment, but individual discrepancies are observed. In certain instances accumulation of ubiquitinated proteins is observed without induction of HSP70B’, HO-1 or p21. This may indicate that the UPS is impaired but not to the degree that it caused significant proteotoxic and oxidative stress and degradation of critically important regulators such as p21 still occur. Another common observation is that accumulation of HO-1 is often observed only at the lowest screening concentration of 5 µM but is expressed lower or absent at the higher screening concentration of 10 µM. An additional method to determine how dominant UPS inhibition is in the polypharmacology of these drugs is to correlate the incidence of UPS inhibition to the incidence of cell death. We analyzed the live-cell imaging experiment using MelJuSo UbG76V-_{YFP reporter cells by tracing individual}

cells for the occurrence of YFP signal subsequent cell death. Compounds for which UPS inhibition is a dominant part of the polypharmacology are expected to induce accumulation of YFP signal and these cells subsequently die, while a high proportion of cells that did not accumulate YFP signal should survive treatment. If many cells die without accumulation of YFP signal this would indicate that another dominant pharmacological effect is responsible for inducing cell death. If cells show accumulation of YFP signal but do not die, the UPS inhibition was not sufficiently strong to induce cell death. A significant benefit of this method is that it facilitates drawing such correlations on the level of individual cells, rather than on a population of cells. Compounds CB686, CB742 and CB383 all show excellent curves for YFP signal at the population level and show