ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2019

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Medicine 1533

Molecular Approaches to ExploreDrug-Target Interactions

RASEL ABDULLAH AL-AMIN

ISSN 1651-6206ISBN 978-91-513-0560-8urn:nbn:se:uu:diva-374329

Dissertation presented at Uppsala University to be publicly examined in Svedbergsalen(B8), Biomedicinskt centrum, Husargatan 3, Uppsala, Friday, 8 March 2019 at 13:15 for thedegree of Doctor of Philosophy (Faculty of Medicine). The examination will be conductedin English. Faculty examiner: Professor Christopher Schofield (Oxford University).

AbstractAl-Amin, R. A. 2019. Molecular Approaches to Explore Drug-Target Interactions. DigitalComprehensive Summaries of Uppsala Dissertations from the Faculty of Medicine 1533.46 pp. Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-513-0560-8.

Improved means to assess the clinical potential of drug candidates can critically influencedevelopment of new therapeutic entities, a central aim in medical life science. Drug discoveryand development relies on construction and selection of small organic compounds or biologicalagents that bind targets of interest. This thesis includes new methodology to investigate targetengagement - that is the tendency for these drugs and drug candidates to bind their intendedtarget molecules versus any off-targets. This is a matter of great importance and current stronginterest in the pharmaceutical industry as well as academically and an important aim forprecision medicine. Paper I describes the target engagement-mediated amplification (TEMA)technique, an accurate, selective and physiological relevant techniques to monitor target bindingby DNA-conjugated low molecular weight drug molecules. The DNA conjugated forms ofthe drugs are uniquely suited to accurately and sensitively reveal the binding characteristicsof drugs directly in relevant tissues. Paper II describes the evaluation of cellular thermalshift assays (CETSA) by multiplex proximity extension assays (PEA), to sensitively measurebinding of drugs to their proper targets and off-targets in minimal samples of cells and tissues,and for many targets and samples in parallel. The technique provides valuable advantagesduring drug development, and potentially also in clinical care. Paper III describes a high-throughput approach to use in situ proximity ligation assays to investigate protein interactionsor modifications along with phenotypic responses to drugs or cytokines. The technique allowsresponses by large numbers of cells to be evaluated by automated microscopy and computer-based analysis. Our approach expands the scope for combined molecular and morphologicalprofiling, offering an information-rich means to profile cellular responses to drugs and otheragents at the single cell level.

Keywords: Drug discovery, target engagement, target engagement-mediated amplification,cellular thermal shift assay, proximity extension assay, in situ PLA, high-content imaging

Rasel Abdullah Al-Amin, Department of Immunology, Genetics and Pathology,Rudbecklaboratoriet, Uppsala University, SE-751 85 Uppsala, Sweden.

© Rasel Abdullah Al-Amin 2019

ISSN 1651-6206ISBN 978-91-513-0560-8urn:nbn:se:uu:diva-374329 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-374329)

Dedication to my Family, Friends and Teachers

“a drug will not work unless it is bound” Paul Ehrlich (Nobel Prize in Medicine 1908)

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I. Al-Amin, R.A., Johansson, L., Landegren, N., Abdurakhmanov, E., Löf, L., Lönn, P., Hammond, M., Klaesson, A., Svensson, R., Blokzijl, A., Söderberg, O., Kamali-Moghaddam, M., Jensen, A.J., Danielson, H., Artursson, P., Lundbäck, L., Landegren, U. Target Engagement-Mediated Amplification for Monitoring Drug-Target Interactions in Situ. (In Preparation for Submission)

II. Al-Amin, R.A., Gallant*, C.J., Lööf*, S., Bacanu, S., Lengqvist, J., Nordlund, P., Landegren, U. Sensitive Measurement of Drug-Target Engagement Using Cellular Thermal Shift Assays with Multiplex Proximity Extension Assay Readout. (Submitted)

III. Lönn, P., Al-Amin, R.A., Heldin, J., Gallini, R., Björkesten, J., Oel-rich, J., Kamali-Moghaddam, M., Landegren, U. High-Throughput In Situ Mapping of Phosphorylated Protein Complexes Across the Cell Cycle and in Response to Drugs. (In Preparation for Submis-sion)

* These authors contributed equally.

Related work by the author

Original Articles

Shaw*, V. Lundin*, E. Petrova, F. Fördős, E. Benson, A. Al–Amin, A. Her-land, A. Blokzijl, B. Högberg, A. I. Teixeira, (2014), Spatial Control of Membrane Receptor Function using Ligand Nano–Calipers, Nature Meth-ods, 11, 841-846.

D. Mokhtari, A. Al-Amin, K. Turpaev, T. Li, O. Idevall-Hagren, J. Li, A. Wuttke, R. G. Fred, P. Ravassard, R. Scharfmann, A. Tengholm, N. Welsh, (2013), Imatinib mesilate-induced phosphatidylinositol 3-kinase signalling and improved survival in insulin-producing cells: role of Src homology 2-containing inositol 5'-phosphatase interaction with c-Abl, Diabetologia, 56(6): 1327-1338. Al-Amin*, R.A., Muthelo*, P., Abdurakhmanov, E., Vincke, C., Danielson, H., Landegren, U. Oligonucleotide-Assisted Construction of Combinatorial Affinity Binders for Diagnostics and Therapeutics. (Manuscript in Prepara-tion)

Review Article Landegren, U., Al-Amin, R.A., Björkesten, J. (2018), A myopic perspective on the future of protein diagnostics. N Biotechnol. 45: 14-18. PMID: 29309916.

* These authors contributed equally.

Contents

Introduction ................................................................................................... 11

Background ................................................................................................... 13The Druggable Genome ............................................................................ 13Drug Discovery and Development; Challenges ....................................... 14

History of Pharmaceuticals .................................................................. 15Small Molecule Drug Discovery ......................................................... 15Strategies for Drug Discovery ............................................................. 16Challenges in Clinical Drug Development .......................................... 18

Drug-Target Interaction and Magnitudes ................................................. 19Methods for Measuring Target Engagement ............................................ 21

Positron Emission Tomography ........................................................... 22Fluorescence- and Bioluminescence Resonance Energy Transfer ...... 22Affinity-Based Chemical Proteomics .................................................. 22Cellular Thermal Shift Assays ............................................................. 23

TGF-β Signaling Kinases ......................................................................... 24Kinases Inhibitors and Profiling Assays ................................................... 25DNA-Encoded Small Molecules .............................................................. 26Chemical Probe and Site Specific Affinity Reporters .............................. 27Molecular Tools Box ................................................................................ 27

Padlock Probes and Rolling Circle Amplification ............................... 27Proximity Ligation-Based Assays ........................................................ 28Proximity Hybridization Chain Reaction ............................................. 29Proximity Extension Assay .................................................................. 30

Present Investigation ..................................................................................... 32Paper I: Target Engagement-Mediated Amplification for Monitoring Drug-Target Interactions in Situ. .............................................................. 32Paper II: Sensitive Measurement of Drug-Target Engagement Using Cellular Thermal Shift Assays with Multiplex Proximity Extension Assay Readout. ....................................................................................... 35Paper III: High-Throughput In Situ Mapping of Phosphorylated Protein Complexes Across the Cell Cycle and in Response to Drugs. ..... 38

Future Perspectives ....................................................................................... 40

Acknowledgement ......................................................................................... 41

References ..................................................................................................... 44

Abbreviations

BRET Bioluminescence resonance energy transfer

CETSA Cellular thermal shift assay

DECL DNA-encoded chemical library

DNA Deoxyribonucleic acid

EFC Enzyme fragment complementation

EGFR Epidermal growth factor receptor

ELISA Enzyme-linked immunosorbent assay

GPCRs G protein-coupled receptors

FRET Fluorescence resonance energy transfer

HTS High-throughput screening

HCS High-content screening

In situ PLA In situ proximity ligation assay

MS Mass spectrometry

PEA Proximity extension assay

PET Positron emission tomography

PPI Protein-protein interaction

PTM Post translational modification

RCA Rolling circle amplification

RNAi Ribonucleic acid interference

SPR Surface plasmon resonance

TEMA Target engagement-mediated amplification

TGF-β Transforming growth factor-β

TKI Tyrosine kinase inhibitor

11

Introduction

Ascertaining that small drug molecules specifically bind their intended tar-gets remains a central problem in biology, medicinal chemistry and medi-cine. The drug discovery and development process is highly complex with a high attrition rate, and taking on average 10-20 years from the identification of a suitable drug candidate until the introduction of a new medicine (Dick-son et al. 2004). Development of effective drug therapies depends on faithful and highly predictive preclinical candidate drug evaluation. Accordingly, there is a demand for very specific and sensitive methods that can take into account the phenotypic diversity of disease in preclinical investigations, before the start of costly clinical trials, and allow elimination of sub-optimal drug candidates selection earlier in the drug discovery process (Bowes et al. 2012; Anastassiadis et al. 2011). This thesis work has focuses on develop-ment and application of methods for evaluating effects by drug binding as well as a contemporary overview about the challenges and opportunities of drug development.

For me as a PhD student, this period has been an exciting and interesting journey. I have had the opportunity to meet with highly competent scientists and also learning how to combine entrepreneurship with research. I have been pursuing a project where I have applied molecular genetic techniques in the context of drug development and therapy selection. This thesis work mostly builds upon my training in pharmacy and my interest in drug discov-ery. It establishes the feasibility as well as the value of equipping low mo-lecular weight drugs or drug candidates with conjugated oligonucleotides, while preserving much of their binding characteristics. I have cooperated with the Chemical Biology Center of Sweden to obtain a set of clinical and preclinical kinase inhibitors, modified with clickable residues designed so as not to block target protein binding. I have used oligonucleotides with a com-plementary clickable function to conjugate the modified drugs with DNA strands. These reagents have allowed me to detect binding of the DNA-conjugated drugs in tissue sections, cells in solution, and among proteins printed in arrays, by adding a circularizable oligonucleotide (padlock) probes, followed by local signal amplification via rolling circle amplifica-tion. We refer to this technique as target engagement-mediated amplification (TEMA). I have taken another approach to investigate drug-target engage-ment where I combined the cellular thermal shift assay (CETSA) with multi-

12

plex proximity extension assays (PEA) for highly sensitive measurement of drug-target binding in cells. The CETSA assay has recently been established as a convenient method to evaluate the binding of low molecular weight drugs to proteins directly in cells and tissues by taking advantage of the ef-fects of drug binding on the susceptibility of the proteins to thermal denatur-ation. The generally stabilizing effect on the proteins is commonly measured via protein blots or in an untargeted fashion by mass spectrometry. I realized that it is often of interest to evaluate target binding to a particular set of on- and off-targets, including proteins potentially mediating toxicity reactions, and to be able to do so in clinically relevant samples. I therefore applied multiplex PEA assays to measure thermal denaturation of proteins from a cancer cell line as a model for such analyses. Furthermore, together with my colleagues, I have also actively participated in establishing a high-content image-based in situ proximity ligation assay approach for morphological mapping of phospho-signaling complexes in single cells as a means to screen for effects by drug compounds. This is an attractive technique which ex-pands the scope for morphological profiling, offering a unique, information-rich, unbiased approach to profile complex cellular responses and target deconvolution at the single cell level.

13

Background

The Druggable Genome The human genome incudes an estimated ∼3000 genes that encode what is currently viewed as druggable targets, representing some 15% of all human genes (Hopkins et al. 2002). Progress in human genetics and in cellular and systems biology offers a novel approach for analysis of biological networks to: i) identifying pathways of disease, ii) discovering drug targets and ii) defining biomarkers for monitoring the treatment response. Seven subclasses of genes are in focus as therapeutic targets, including receptors, enzymes, hormones and ion channels as the largest target groups (Bleicher et al. 2003). Currently around 80% of marketed small molecule drugs target enzymes and subfamilies of G protein-coupled receptors (GPCRs) (Bleicher et al. 2003).

Figure 1. Numbers of drug targets and disease modifying proteins, encoded in the human genome. Adapted from Hopkins Nat Rev 1:727, 2002. GPCRs play a crucial role for many functions in the pathology of diseases such as cancer and cardiovascular, endocrine and metabolic disorders). They constitute the largest class of therapeutic targets, corresponding to currently over 30% of marketed drugs. Kinases are the single most frequently targeted component of the druggable genome at some 22%, and the second largest target family for drug discovery (Hopkins et al. 2002). Kinases are involved in intracellular signal transduction, growth, differentiation, and apoptosis in the course of normal cellular functions, but kinases are also involved in the pathobiology of common diseases such as cancer, disorders of the immune

14

system, diabetes, hypertension, rheumatoid arthritis, degenerative diseases, fibrosis, etc (Ferguson et al. 2018; Manning et al. 2002). Key structural ele-ments of the active sites are conserved across all 518 human kinases, and off-target effects on non-target kinases represent common challenges in the development of kinase inhibitor (Anastassiadis et al. 2011). Protein-protein interactions (PPIs) play a central role in most biological processes, and they are frequently dysregulated in diseases. Accordingly, there is an enormous therapeutic potential associated with PPls.

Drug Discovery and Development; Challenges Drug discovery and development is a costly and lengthy process with high failure rates from target identification to final approval as a clinical medicine (Dickson et al. 2004). Despite many new approaches, drugs with a high ther-apeutic index - the ratio between toxic and effective doses - remain difficult and highly expensive to develop - or entirely out of reach for many proteins of potential interest as drug targets. Efficient production of drugs that specif-ically bind the intended target proteins therefore is of central importance for medical progress.

Figure 2. Overview of the main pharmaceutical classes and their targets (Dig-gers et al. 2008). Small molecule drugs that bind specific target protein pockets in and block enzyme active sites. Macrocyclic compounds are an underexplored new class, positioned somewhere between small molecules and biologic drugs, that is typically seen as particularly attractive for modulating protein-protein interactions. Targets for biological drugs need to be present outside or on surfaces of cells. Drugs for these targets are suited to modulate ligands and receptors

15

History of Pharmaceuticals Paracelsus (~1500) identified herbs and plant extracts as medicines. Louis Pasteur (1822-1895) developed a vaccine against rabies. Pelletier & Caven-tou (1826) established the first “modern" pharmaceutical company by isolat-ing quinine from cinchona bark for the treatment of fever. Knorr & Filehne (1884) introduced the first synthetic drug antipyrin (phenazone) and later Bayer (1899) introduced Aspirin (acetylsalicylic acid) in the market. The discovery by Banting & Best (1921) of insulin for treatment of diabetes lead to production of this molecule as a protein drug. The first recombinant anti-body approved for cancer therapy - Rituxan (rituximab) – was developed by Biogen (1997) (Source: www.pharmaphorum.com).

Table 1. The top-ten best selling medicines in 2017 (genengnews.com).

Small Molecule Drug Discovery High-throughput screening permits analysis of large libraries of compounds (Howe et al. 2008), and specifically binding compounds may also be built up from smaller units through fragment-based lead discovery (Erlanson et al. 2004). However, the biophysical assays cannot provide insights in other relevant characteristics such as efficacy, toxicity and membrane permeabi-lity. The lead optimization usually begins with testing of in vitro selectivity and potency using biochemical methods after the early biophysical evolu-tions of hits (Kung et al. 2006). More than 37% of drug discovery has been based on phenotypic assays for validation of target engagement in cells and tissue, providing functional information of compound activity that is useful for follow up with biochemical assays (Cook et al. 2014). In general, lead optimization assays are a way to measure or characterize the ability of chem-ical compounds to influence biological events in biochemical, cellular, or in vivo condition (Vincent et al. 2015). Cellular-based target validations using chemical probes to measures intracellular drug binding involves assessment

16

of the exposure of the drug at the site of action, target engagement, function-al pharmacology and relevant phenotypic effects (Bunnage et al. 2013).

Figure 3. The small molecular drug discovery processes can be roughly divided into four major phases. i) Target identification and validation phases; the process of identifying and fully confirming the direct molecular target or pathway of the diseases. ii) Hit discovery phase; identification of a compound that has the desired activity in a high-throughput compound screen and whose activity is confirmed upon re-testing. iii) Lead identification phases; development of a chemical entity having a defined structure and shown to have a desired effect in a biological assay. iv) Lead optimization and candidate selection phases; the production of a chemical entity with a defined effect in a biological assay and with the goal to develop as a candidate drug (CD) for preclinical development. The work often involves modifica-tions of lead structures allowing these to be transformed into clinically useful drug molecules.

Strategies for Drug Discovery The primary goal in early drug development is to find a suitable candidate drug molecules for the next phase in the clinical development. Currently two different strategies are followed in small molecular drug discovery, target- and phenotype-based approaches. Phenotypic high-content screening is a powerful tool in drug discovery, both to find starting points for potential therapeutics, and to identify the targets of particular compounds and also their mode of action (MoA) (Boutros et al. 2015). Every steps of the drug discovery process should use disease-relevant models, including initial screens.

Figure 4. Strategies for small molecule drug discovery. (Adapted from Terstap-pen et al. 2007).

17

The target-based approach: The first step in the target-based drug discov-ery process is identification and validation of a drug target, once the target is fully confirmed. In the hit discovery-phase suitable compounds for target binding are sought. The discovery of small molecule hits for lead optimiza-tion usually begins with testing of in vitro selectivity and potency using bio-chemical methods and further validation after biophysical evolutions of hits (Kung et al. 2006). Biochemical assay are used to characterize selected lead compounds, and can serve to confirm a highly specific physical interaction with a target of interest. The final step and goal of the drug discovery phase is to develop a candidate drug (CD) for preclinical development. This often involves modifications of the lead molecules with the goal in to find a CD for pre-clinical development. The target-based drug discovery process can roughly be divided into five major phases:

Figure 5. Overview of the target-based drug discovery approach. (Illustration modified from drug target review issue 3, 2018; www.drugtargetreview.com).

Phenotype-based approach: Phenotypic-based screen is an attractive alter-native to target-based efforts as it can yield novel opportunities to target disease-relevant pathways, and for target deconvolution of screening hits. The phenotypic approaches provide functional read-out of effects by com-pound, and are useful for follow up of target-based biochemical assays. The endpoint measurement of high-content screening (HCS) assays and kinetic measurement in live cell represent the most commonly used phenotypic ap-proaches in oncology drug discovery programs. Phenotypic screening pro-vides an alternative way to improve predictivity of compound activity and toxicity in the early stages of drug development, and an efficient way to identify novel targets. This strategy requires a strong target deconvolution pipeline, in order to allow compounds to progress through the regulatory process.

Figure 6. A schematic illustration of the phenotypic drug discovery approach. (modified from drug target review issue 3, 2018).

18

Challenges in Clinical Drug Development The vast majority of drug candidates do not make it to the market, and only one compound in nine that are tested in the clinic proves successful (Cook et al. 2014). The clinical trial phase I is carried out by experimenting in healthy volunteers in order to document pharmaceutical safety. This is followed by phase II patient trials in order to define a safe and effective dose. Phase III trials are used to confirm efficacy and safety of the candidate drug in larger, diverse patient populations.

Figure 7. The processes of clinical drug development. In phase I trials, drugs are tested in small numbers of volunteers (20-80) for the first time to evaluate their safety, determine a safe dosage range and identify side effects. Phase II trials are performed in larger groups of people (100-300) to observe effectiveness and to fur-ther evaluate the safety within a diverse population. In Phase III trials, the treatment is given to large groups of people (1,000-3,000) to confirm effectiveness, monitor side effects, and compare to commonly used treatments. Finally the post market analyses delineate additional information including the drug's risks, benefits, and optimal use.

Drug development programs often fail for reason of safety or efficacy during the clinical phases I-III and 51% of these failures are due to lack of efficacy, while safety concerns accounted for 16% of failures (Morgan et al. 2012; Cook et al. 2014). Drug safety relies on the selectivity profile of the drugs, where compounds with poor selectivity may lead to unwanted side effect (Smyth et al. 2009; Zhang et al. 2009). The drug selectivity profile can be directly correlated to in vivo drug efficacy and toxicity effects (Simon et al. 2013). The efficacy of drugs depends on how well the compounds can modulate the primary target molecule, a process referred to as target en-gagement (Bunnage et al. 2013; Simon et al. 2013). A potential lack of ap-propriate physiologically relevant model systems frequently leads to subop-timal candidate drug selection during the preclinical phase in drug discovery (Vincent et al. 2015).

19

Figure 8. Cumulative failure rate of projects initiated for drug development. (Bio, Biomedtracker, Amplicon Clinical Development Success Rates 2006-2015)

Determning the target engagement of a drug is essential to link exposure at the site of action to pharmacological effect, and early proof of target en-gagement at the site of action can lead to lower failure rates during clinical phase 2 studies in drug development (Bunnage et al. 2013; Simon et al. 2013). This illustrates the requirement for improved, highly specific and sensitive methods to investigate target engagement of new therapeutic enti-ties before the start of costly clinical trials, in order to eliminate sub-optimal drug candidates during earlier phases of drug discovery (Bowes et al. 2012; Anastassiadis et al. 2011).

Drug-Target Interaction and Magnitudes The rate at which a drug associates with and dissociates from its target plays a pivotal role for its clinical effectiveness. Many studies show the potential of considering binding kinetics, and particularly residence time for drug efficacy, safety and duration of action and to differentiate medicines. Direct binding assays mostly investigate the off-rate of test compounds, defined as the residual time (residence time, t1/2 = 1/ Koff-rate) of the interaction with on- and off-targets and Kd kinetics (dissociation constant, Kd = off-rate/on-rate) of the molecule (Pan et al. 2013; Copeland et al. 2007). Good correlation has been demonstrated between experimental and simulated drug-target resi-dence times. Such simulations can help scientists better predict the capacity of a small molecule to remain bound in a complex, thereby supporting priori-tization of molecules for synthesis. Biophysical assays, including SPR, LC-MS, NMR, ITC, DSF and CETSA, are more consistent and suitable for high-throughput screening to investigate interactions between compounds and their target proteins. Structure-based ligand binding assays (NMR or X-ray crystallography) are applied for structure based drug design in order to un-derstand the molecular details of ligand-target interactions (Holdgate et al. 2010). Biosensor based assays are commonly used to determine real-time binding kinetics of drug molecules to their targets.

20

Table 2: Assays commonly used for investigating binding kinetics in drug-target interactions.

Figure 9: Drug-target interaction kinetics calculation and residual time.

Biochemical methods are applied to characterize selected leads compounds, and can serve to improve several aspects by confirming a specific physical interaction with a target of interest. Simple and useful biochemical ligand binding assays are available that build on measurement of properties such as fluorescence intensity (FI), FRET & TR-FRET (time-resolved fluorescence resonance energy-transfer), BRET (bioluminescence resonance energy trans-fer), or fluorescence polarization (FP). But the limited sensitivity of such assays can be a challenge due to interference from background auto-fluorescence, pH and ion concentration, variable solubility of reagents in buffer, stability and aggregation of proteins, as well as short comings of in-strumentation (Ma et al. 2008). Radioligand-binding assays are quite robust to investigate drug-target in vivo but costly, and there are significant issues regarding the handling of radioactive materials.

21

Table 3. Summary of pros and cons of some methods used to investigate drug-target interaction.

Methods for Measuring Target Engagement In small molecule drug discovery, target engagement (TE) is defined as the ability of a ligand to interact with its proper target biomolecule(s). Two commonly used methods for measuring target engagement are positron emission tomography (PET) imaging and activity-based MS proteomics using small-molecule probes, called activity-based probes (ABPs) by bio-orthogonal reactions such as ‘click chemistry or via biotin. Other common cell-based techniques are FRET, BRET and enzyme fragment complementa-tion (EFC). These techniques enable detection of close proximity, but the compound and the target both need to be tagged. The label-free cellular thermal shift assay (CETSA) serves to monitor target engagement and to

Methods Advantages Limitations FRET (TR-FRET) & BRET

- Robust - Close proximity detection

- Requires labeling of both compound and target - Limited sensitivity, hin-dered by auto-fluorescence interfer-ence - Limited to interactions over very short distances (5-10 nm)

Affinity-based chem-ical proteomics

Direct determination - Requires resynthesis of original compounds - Requires information about SAR (structure-activity relationship) for labels

Cellular thermal shift assay (CETSA)

Monitor the binding of underivatized mole-cules to endogenous proteins

- Not all proteins show a change in thermal stability upon ligand binding. - Single cell resolution cannot be achieved.

22

evaluate cell permeability of compounds. The technique requires no specific equipment other than what is normally present in a biochemistry lab.

Positron Emission Tomography Positron emission tomography (PET) is used for measuring target engage-ment by investigating drug distribution in vivo across the whole body but with limited spatial resolution. PET can provide quantitative measurements of target occupancy, informing estimates of the relation between fractional target occupancy and efficacy. The technique requires labeling compounds of interest with positron-emitting radionuclides, creating radioactive tracers that are injected in the body. The tracers are subjected to distribution, me-tabolism and accumulation in the target tissue. The location of the tracer can then be followed through the body through the emission signal.

Fluorescence- and Bioluminescence Resonance Energy Transfer Fluorescence resonance energy transfer (FRET or time-resolved FRET) and bioluminescence resonance energy transfer (BRET) can be used to study the interaction between a protein and a ligand or a compound. For FRET and BRET both targets and ligands must be labeled with fluorophores and/or luminescence donors. Due to the requirement for proximity between the protein and the compound or ligand, FRET and BRET are good tools for monitoring target engagement and other intra-molecular protein interactions. FRET depends on energy transfer between fluorophores, and can provide both spatial and temporal information of the compound-protein interaction. BRET works in a similar way as FRET, with the difference that it is depend-ent on the transfer of energy from a luminescence donor to a fluorescent acceptor.

Affinity-Based Chemical Proteomics Affinity-based chemoproteomics and (2D-) thermal proteome profiling are two orthogonal methods that enable identification of target proteins and demonstration of target engagement. The bio-orthogonal chemistry-based modular probe strategy opens new avenues to prove cellular target engage-ment and to identify cellular localization.

In chemoproteomics endogenous target proteins using tagged compounds from cell extracts are subjected to affinity enrichment, followed by analysis via LC-MS/MS. By varying the dose of the tagged compounds in the binding experiments it is possible to determine compound affinity (Kd

app) profiles and dose-dependent engagement (EC50s) of cellular targets (Bantscheff et al. 2012). Using affinity-based chemical proteomics it is possible to demon-

23

strate direct or indirect target engagement by purification of protein-compound complexes. The compounds are chemically modified with a han-dle, which most often is an alkyne or an azide group. After the compound has bound its target, some form of a tag is attached to the compound, typical-ly using click-chemistry. This enables affinity purification of the bound pro-tein-compound complexes. Protein identification can be done by tryptic di-gestion and mass spectrometry. The technique requires synthesis of the com-pound with a chemical handle and an established structure-activity relation-ship. (2D-) Thermal proteome profiling reveals altered thermal stability of cellular target proteins having bound a compound (Becher et al. 2016). The recently developed CETSA-MS technique allows multiplexed quantitative prote-omics profiling for target deconvolution of phenotypic screening hits, off-target profiling, correlation with off- target activity (repurposing) and inves-tigation of causes of adverse events (safety).

Cellular Thermal Shift Assays The cellular thermal shift assays (CETSA) is a direct biophysical assay for target engagement in cells. The method measures changes in thermal stabil-ity of a target protein upon drug binding. The technology relies on the prin-ciple that binding of a ligand to a protein often induces a change in thermal stability and thereby a shift in its melting curve and the melting temperature (Tm) of the protein that the ligand binds to. In CETSA, a cell or tissue sam-ple that is treated with a drug or control is divided in aliquots and these are heated to different temperatures. Any aggregated proteins are then precipi-tated and removed after centrifugation. Remaining soluble proteins of inter-est are quantified by protein detection methods such as immunoblotting or mass spectroscopy. The proteins that escape precipitation correspond to the correctly folded, non-denatured proteins. The amount of bound compound can be determined for each treatment temperature and the melting tempera-ture, Tm, can be calculated by plotting the soluble protein against the tem-perature. A major advantage of this method is that it is label-free, in that no modification of the compound is required and the technique therefore does not risk disrupting binding to on- or off-targets. CETSA can also be used to monitor downstream efficacy, toxicity, acquired drug resistance mecha-nisms, drug transport, metabolism, overexpression and mutations of the tar-get protein. In principle it may also be used to report on protein-protein in-teractions, protein-nucleic acids interactions, protein-metabolite interactions and protein-membrane interactions. CETSA does however have some limita-tions; not all proteins are amenable to thermal shift studies and not all com-pounds that bind to targets produce thermal stabilization. Some proteins, often-small ones, sometimes does not aggregate within the anticipated tem-perature range.

24

Table 4: Comparison of the available methods for detection of target-engagement in cells. Adapted from Schurmann Cell Chem. Biol. 2016.

TGF-β Signaling Kinases The TGF-β ligand binds and activates cell surface type I and type II TGF-β receptors, which in turn phosphorylate cytoplasmic Smad2 and Smad3 at their C-termini, causing them to accumulate in the nucleus and interact with Smad4 (Massagué et al. 2012; Heldin et al. 2012). Downstream Smad2 phosphorylations and interactions are known to be regulated by multiple kinases. Through further modifications and interactions, the Smad complex-es ultimately orchestrate gene expression at specific promoter/enhancer DNA sequences. The signaling is terminated via protein relocation, dephosphorylation and/or proteasomal degradation of the transcriptional complexes. The pathway controls several important cellular functions during embryonic development and in the adult organism, including cellular prolif-eration, differentiation, migration, and apoptosis (Massagué et al. 2012; Heldin et al. 2012). TGF-β is often dysregulated in tumors, such as for ex-ample in cancers of the breast, colon, and pancreas, thereby losing its growth

25

suppressor functions and instead promoting cancer progression (Tang et al. 2003; Massagué et al. 2008; Heldin et al. 2012). Multiple kinases have been shown to operate these sites including CDKs, GSK3b, and MAPKs (Matsu-zaki et al. 2004; Alarcón et al. 2009; Fuentealba et al. 2007).

Figure 10. Therapeutic areas for kinase inhibitors. (Adapted from Ferguson et al. 2018).

Kinases Inhibitors and Profiling Assays A key aspect of efforts to rationally design safer drugs involves determining binding profiles across a wide range of potential on- and off-targets and tak-ing this into account when designing novel and improved compounds. This is particularly important for protein-families that possess conserved folds and have numerous members such as kinases. The binding modes of kinase inhibitors generally include the following categories: Type I competitive inhibitors compete with ATP for binding at the DGF motif in active state kinases. Type II inhibitors bind kinases in inactive states and Type-III inhibi-tors are non-ATP site inhibitors (Davis et al. 2011; Lui et al. 2006). A num-ber of commercial biochemical kinase-selectivity profiling assays are availa-

26

ble, employing several assay mechanisms: i) substrate depletion assays (PKLight, Kinase Glo, Caliper etc), ii) substrate-ligand direct binding assays (BIAcore, HitHunter, KinomeScan etc.) and iii) complex product generation assays (LanthaScreen, Alpha-Screen, ELISA, IMAP etc.) (Ma et al. 2008). Several approaches are in common use for kinase inhibitor profiling; for instance inhibitor compounds may be immobilized on a solid phase, fol-lowed by binding of labeled kinases, allowing displacement curves with unliganded inhibitors to be recorded and used to profile compound selectivi-ty (Fabian et al. 2005; Bantscheff et al. 2007). It requires a considerable effort to develop a biochemical competition binding assay for selectivity profiling of compounds across the kinome for screening purposes.

Table 5: Comparison of common kinase profiling platforms.

DNA-Encoded Small Molecules A theoretical paper by Sydney Brenner and Richard Lerner proposed an approach to improve the search for drug candidates by tagging libraries of compounds with DNA strands, serving to encode the identity of the individ-ual compounds (Brenner et al. 1992). The authors pointed out that this mechanism offers several potential advantages, including the possibility to amplify DNA tags from bound substances for improved sensitivity of detec-tion, and to accurately and conveniently decode the tags via DNA sequenc-ing. In this manner very large substance libraries, suitably modified with DNA, can be screened in parallel for target protein binding, followed by identification of compounds that may serve as lead compounds for further screens with or without added DNA strands. DNA-encoded compound li-braries (DECL) have come of age and are now in practical use to improve the search for candidate drugs by analyzing DNA barcodes on small mole-cules that bind target proteins (Mullard et al. 2016; Melkko et al. 2004). In addition, McGregor and colleagues developed an interaction-dependent PCR procedure for identifying ligand-target pairs in solution phase (McGregor et

27

al. 2010). Recent advances in this field include the preparation and screening of collections of billions of compounds, and cheap, robust screens of these billions of compounds in a single vessel (Clark et al. 2009; Gura et al. 2015). This development is led by a number of successful companies; Nuevolution, Praecis, Ensemble, Vipergen, Philochem, Lexicon, X-Chem, HitGen, Nurix, Dice Molecules and Forma. The attached DNA barcodes are unique for each compound, permitting efficient screening (Mullard et al. 2016; Gura et al. 2015).

Chemical Probes and Site-specific Affinity Reporters Studies using advanced SAR and X-ray co-crystal structure analysis point to the feasibility of chemically modifying lead compounds without losing sig-nificant binding affinity for a target molecule. Click chemistry is a suitable orthogonal chemistry to use in complex biological systems. Click-probe approaches enable investigation of cellular localization and binding of unla-beled drugs at high resolution in individual cells. Because of the modularity of this approach it is possible to simultaneously address several fundamental questions in drug discovery, such as probe localization at high spatial resolu-tion, direct target engagement measurement, and target identification. A broad spectrum of clickable reporters can be applied in fluorescence- or ra-dioactivity-based readouts (e.g. super-resolution), spectroscopy-based readouts (e.g. Raman) and tomography-based readouts (e.g. PET). Small molecule-DNA conjugation reactions by click chemistry are very chemo-selective, easy to perform and versatile with high yielding synthesis (Kolb et al. 2001).

Molecular Tools Box Padlock Probes and Rolling Circle Amplification Padlock probes are oligonucleotide reagents that are designed to have se-quences complementary to a target nucleotide sequence at their 3’and 5’ends such that these ends are brought next to each other and can be joined by liga-tion upon target recognition. The reaction converts the probes to DNA cir-cles and the probes can be used for detection of target nucleic acid sequences in a sample. If the padlock probes do not exactly match their target sequenc-es, ligation of the ends of the padlock probe and the formation of a circular structure is inhibited. After hybridization and circularization of padlock probes, the reacted probes can be amplified by rolling circle amplification (RCA). This results in micrometer-sized DNA-clusters of repeated comple-

28

ments of the probe sequences, serving as easily detectable indicators for the recognition of a target molecule.

Proximity Ligation-Based Assays Proximity ligation assay: The proximity ligation assay (PLA) uses two or more affinity reagents, such as aptamers or more commonly antibodies, to detect a target protein, protein modifications or protein complex (Fredriks-son et al. 2002). Upon binding by the reagents to their targets, the attached oligonucleotides are brought in proximity and can hybridize jointly to an added connector oligonucleotide, allowing the conjugated oligonucleotides to be joined by enzymatic ligation creating unique reporter DNA molecules that can be detected using methods such as quantitative real-time PCR or DNA sequencing. This requirement for recognition by two and sometimes three or more affinity reagents in order to generate a signal, in combination with the opportunity for amplification of reporter DNA molecules encoding the identities of the detected molecules, jointly serve to maximize signal to noise and allow the study of protein concentrations over wide concentration range down to very low limit of detection. Several recent publications from our lab and others have established that PLA presents advantages for anal-yses of large numbers of target proteins in very small sample aliquots, com-pared to other method for sensitive analysis of proteins (Fredriksson et al. 2007; Darmanis et al. 2011). The technology has been shown to offer high sensitivity of protein detection, in some cases more than 100-fold improved over the more commonly used sandwich ELISA.

Figure 11: Characteristics of proximity assays by enzymatic ligation.

In situ PLA: In in situ PLA (isPLA) the basic PLA architecture has been modified for analytical applications that allow visualization of localization of proteins or modifications of proteins and endogenous protein-protein interac-tions by microscopy (Söderberg et al. 2006). In this modified version of PLA pairwise binding of antibodies carrying two distinct oligonucleotides serves

29

to template the formation of circular DNA strands via two DNA ligation reactions. The circular DNA in turn templates localized rolling-circle ampli-fication (RCA) reactions by a DNA polymerase, resulting in the production of single-stranded RCA products that are easily detected for localized visual-ization and digital recording of results.

Table 6: Comparison of protein detection immunoassays.

UnFold: The UnFold-technology is the most recent PLA generation. UnFold relies on the same principle as the earlier isPLA technique. The difference between the earlier isPLA and UnFold is the structure of the conjugated probes. In the UnFold method, at least one of the proximity probes has a hairpin structure. This structure prevents the functional element of the two probes from interacting when they are added to a sample. After washes the proximity probes can be “unfolded” by e.g. a cleavage reaction. The two proximity probes are then allowed to interact with each other in such a way that one serves as a template for ligase-mediated circularization of an oligo-nucleotide hybridized to the other reagent. This DNA circle can serve as template for amplification and detection, thus becoming a detectable indica-tor for the target molecule or target interaction. This version offers higher sensitivity compared to the earlier isPLA technique.

Proximity Hybridization Chain Reaction The Proximity Hybridization Chain Reaction (ProxHRC) is an enzyme-free alternative to of isPLA. In ProxHCR, each member of a pair of proximity probes is a protein-binding reagent connected to an oligonucleotide with a

30

hairpin structure. When an activator oligonucleotide is added, this opens up a stem in one of the hairpins, which in turn opens the hairpin oligonucleotide on the second proximity probe if this has bound in proximity. When a fluo-rescence labeled HCR oligonucleotide is added, an enzyme-independent linear amplification reaction will ensue. This proximity detection reaction offers lower assay cost compared to the RCA based methods isPLA and UnFold.

Proximity Extension Assay The proximity extension assay (PEA) is another variant of the proximity techniques, useful for measuring proteins in solution phase. In PEA DNA strands conjugated to pairs of antibodies brought in proximity by binding the same target protein can hybridize to each other, initiating a polymerization reaction upon addition of a DNA polymerase (Assarsson et al. 2014). This specific reaction creates a unique DNA molecule upon dual recognition of a target protein molecule. The products of the extension reactions can be quan-tified and detected using quantitative real-time PCR or DNA sequencing to record the incorporated DNA tag sequences identifying the antibodies, with-out any need for washes or separations. PEA has proven a highly sensitive, specific method for protein detection, capable of analyzing sets of 96 pro-teins and controls in small aliquots of samples (Assarsson et al. 2014). Mul-tiplex PEA allows for analysis of large numbers of target proteins in aliquots of as little as 1 μl of plasma or tissue lysate samples (Lundberg et al. 2011) or even single cells (Darmanis et al. 2016). The sensitivity of proximity ex-tension technology affords protein measurement at levels that in certain cas-es surpass the more commonly used sandwich ELISA by a factor of 100-fold.

31

Purpose and Aims of the Thesis

Drug development is exceedingly costly, and late failures incur much of the costs without resulting in a marketable product (David et al. 2014). The pur-pose of the current investigation has been to explore new opportunities to assess, both at early and later stages of drug development, how a new drug candidate interacts with its intended target protein, and with other proteins, including ones known to mediate toxic complications, and how to screen drugs for their effects on functional states of cells. The specific aims were:

• to develop a scalable preclinical in vitro platform for profiling the

selectivity of drugs in order to predict their efficacy and safety.

• to accurately and sensitively measure binding of drugs to their prop-er targets and to off-targets in tissue preparations.

• to studies molecular and morphological responses to drugs and other

agents at the single cell level.

32

Present Investigation

Paper I: Target Engagement-Mediated Amplification for Monitoring Drug-Target Interactions in Situ.

Background and Aim of the Study Methods are needed to confirm specific binding by drugs to their proper targets directly and with high selectivity as well as to determine the correct localization of a candidate drug interacting with its target in relevant clinical specimen during drug discovery. An important aim of this project was to decrease drug attrition rates and to improve the quality of novel drugs, there-by contributing to reduced costs of drug discovery. A central objective of the study was to demonstrate the use of small molecule inhibitors for applica-tions with RCA to characterize on- and off-target binding in cells, tissues and protein arrays. The approach, or variants thereof, may be utilized for high-throughput analysis and selectivity profiling of small molecules, and for in situ localization and visualisation of small molecule-target protein interactions for testing drug efficacy and safety.

Methods In this proof-of-concept study, low molecular weight pharmaceutical com-pounds, modified through attachment of DNA strands were allowed to inter-act with their targets and the sites of interaction were visualized by micros-copy or flow cytometry. The localization of physical interaction reaction was amplified very specifically with circularized oligonucleotide molecules that served to template localized RCA reactions. The procedure generated strong, discrete and quantifiable amplified signals for each bound low molecular weight-DNA conjugated, revealing the localization of drug binding. We established this target engagement-mediated amplification (TEMA) method using kinase inhibitors with conjugated oligonucleotides to measure specific drug-protein in thousands of proteins in arrays, or directly in cell prepara-tions and tissue sections. We obtained very sensitive and specific results for in situ drug-target interaction detection, where DNA-linked drug-target in-teraction was detected via the RCA mechanism within cell lines and tissue sections.

33

We also established a variant of the proximity ligation assay (PLA) for effi-cient identification of molecules specifically binding proteins. In this assay, antibodies specific for an intended target protein of the drug are added to a sample together with the oligonucleotide-conjugated drug. Upon joint bind-ing to the target the two conjugated oligonucleotides are used to template the formation of a circular DNA strand via two DNA ligation reactions. The DNA circle then serves to template a local amplification reaction. Using this mechanism, we could focus the analysis of drug binding to a target protein of interest via this proximity ligation-based target engagement-mediated amplification (proxTEMA).

Figure 12. Overview of TEMA methods. In TEMA, low molecular weight com-pounds with conjugated oligonucleotides bind their target proteins in protein arrays or among fixated cells or tissues sections or in blood cells in suspension. After washes the localization of drug molecules is visualized via the addition of padlock probes. These DNA probes form DNA circles bound to the conjugated oligonucleo-tides, which first serve as templates for ligation reactions, and then prime localized RCA reactions, whose products are visualized via fluorescent oligonucleotide probes. The RCA products that form in TEMA reactions can be visualized, analyzed and digitally quantified using a microarray scanner, fluorescence microscope or via flow cytometry to evaluate drug-target interactions in the investigated materials.

Important Findings In this work, we developed an accurate, selective and highly specific tech-nique to monitor drug-target binding interactions. We applied molecular genetic approaches previously developed and used in our lab, in an entirely new context to solve important problems during drug development. The new TEMA technique is useful for defining molecular targets of new therapeutic entities in relevant tissues and among arrayed proteins. Thereby TEMA may prove valuable for improving candidate selection via drug target engagement investigation during the lead optimization stage in drug discovery, with pos-

34

sible applications also in personalized medicine. The significances of TEMA are as follows:

1. TEMA enables on- and off-target identification and screening among larges sets of proteins spotted in arrays.

2. The method permits kinase selectivity profiling, measuring binding kinet-ics and screening for competitors.

3. The localization of drug binding can be demonstrated in situ in cells and pathological tissue.

4. Target engagement can also be investigated in patient materials to study on- and off-target binding.

35

Paper II: Sensitive Measurement of Drug-Target Engagement Using Cellular Thermal Shift Assay with Multiplex Proximity Extension Assay Readout.

Background and Aim of the Study When exposed to increased temperatures natively folded proteins denature, unfold and eventually aggregate and precipitate. However, a drug binding to a protein can have the effect to stabilize the protein against thermal denatura-tion, allowing it to melt or be denatured only at a higher temperature, record-ed as a thermal shift (Huang 2013; Vedadi et al. 2006). The cellular thermal shift assay (CETSA) is used to measure drug-target engagement in situ, in e.g. pathological clinical specimen. The method can also permit evaluation of drug-target engagement in living systems (Molina et al. 2013). The ability of PEA to use very low amounts of sample and its suitability for most types of biological samples is promising for CETSA applications, where the amounts of sample that are available may be insufficient for mass spec-trometry (MS) analysis (Huber et al. 2015; Savitski et al. 2014). In this paper our approach was therefore to combine CETSA with multiplex proximity extension assays (PEA), and we compared the PEA results with mass spectrometry data for the same samples subjected to drug-induced thermal shift. The approach can meet the needs for accurate and sensitive measurement of drug binding to their proper targets and to off-targets in biological systems.

Methods The overall basic steps of the assay are drug incubation, heat treatment at variable temperatures, centrifugation to remove precipitated protein, and measurement of protein concentrations remaining in solution. The cells were treated with a drug or vehicle control and aliquoted into PCR tubes, followed by heating to one of ten temperatures in a gradient PCR machine to de-naturate increasing proportions of protein at higher temperatures. The sam-ples are then lysed by three cycles of freeze thawing in liquid nitrogen, fol-lowed by centrifugation to remove the precipitated fraction. Materials re-maining in the soluble fractions or supernatants were analyzed with PEA and also with MS to measure specific protein concentrations. Model CETSA experiments were performed in the human cancer cell line K562, treated with kinases inhibitors or vehicle. The CETSA-PEA analysis allowed ther-mal shift assays in as little as 5000 cells, and exhibited good correlation with the results from MS, but MS required 10-fold more cells for analysis com-pared to PEA.

36

Figure 1. Schematic illustration of the cellular thermal shift assay (CETSA) readout with proximity extension assay (PEA) methodology. Representation of work flows; drug incubation, heat treatment, centrifugation and protein detection via PEA. (A) Cell lysates were incubated either with or without drugs and aliquoted into PCR tubes. (B) The treated aliquots were incubated at either of ten different temper-atures in a gradient PCR machine, followed by removal of the precipitated protein fraction by centrifugation. (C) The supernatants were analyzed by multiplex PEA detection with realtime PCR read-out using a 96.96 Dynamic ArrayTM Integrated Fluidic Circuit (IFC) on a Biomark HD system (Fluidigm). (D) The raw data from the realtime PCR are log2 Ct values. Signals were normalized over background (ΔΔCt) and plots were generated using an in-house script developed in ‘R’.

Important Findings In summary, we present a CETSA-PEA methodology that presents ad-vantages for monitoring thermal stability of sets of proteins as an effect of drug treatment in minimal amounts of samples. In short this work estab-lished the following results:

1. CETSA-PEA allows for sensitive multiplex measurement of drug-target engagement using small numbers of cells.

37

2. The results of CETSA-PEA correlates well with CETSA-MS detection in a model system, allowing target engagement to be assessed for a set of ki-nase inhibitors.

3. The work supports the usefulness of CETSA-PEA for measurements of target engagement in clinical samples for target-driven drug discovery and therapy selection.

The multiplex CETSA-PEA technique allows many targets and samples to be investigated in parallel while using small amounts of sample, thus provid-ing important advantages during drug development and potentially also in clinical care.

38

Paper III: High-Throughput In Situ Mapping of Phosphorylated Protein Complexes Across the Cell Cycle and in Response to Drugs.

Background and Aim of the Study Phenotypic high-content screening is a powerful tools in drug discovery, both to find starting points for potential therapeutics, and to identify targets and mode of action (MoA) of compounds of interest (Boutros et al. 2015). Posttranslational modifications (PTMs) and protein-protein interactions are dynamic events that regulate protein activities and cellular processes. TGF-β, which is an important signaling cytokine that regulates cell growth, differ-entiation, migration and death, was used as a model system together with the HaCAT cell line. Downstream Smad2 phosphorylations and interactions that are knows to be regulated by multiple kinases were investigated. The in situ proximity ligation assay (isPLA) technology offers means to study protein modifications and co-localization within 40 nm in situ. These cellular re-sponses can reflect ongoing signaling activities and are clinically relevant as they can reveal pathway-specific changes in disease or effects of targeted therapy. The molecular analyses were complemented by morphological analyses for the same individual cells, using algorithms for image analysis. The procedure provided robust quantitative profiling data of dynamic effects by compounds both for cell populations and at the single cell level in a man-ner that has been difficult to monitor at high-throughput.

Our main aim in this paper was to establish a high content high-throughput, semi-automated microscopy system for in situ proximity ligation assays (isPLA).

Methods In order to screen for effects on cellular signaling and on cellular morpholo-gy in basic research, in screening campaigns for small molecules or in clini-cal routine, we established a semi-automated high-content microscopy sys-tem using in situ proximity ligation assays (isPLA). The assay was imple-mented in microtiter wells with a scanning microscope and high-performance computer-based image analysis using the CellProfiler software, plotting the result using R Studio. We demonstrate specific protein phos-phorylations and interactions of cellular signaling by investigating TGF-β responsive Smad2 linker phosphorylations and complex formations over time and across millions of individual cells. We digitally recorded in situ PLA products along with morphological features of individual cells in rela-tion to e.g. local cell crowding conditions and cell cycle progression via DNA content and nuclear size measurements. The assay allowed us to screen for temporal effects of stimulation, and for consequences of treatment with a modest library of drugs.

39

Important Findings A key component of phenotypic screen based discovery programs is the identification of effects on signaling pathways by screening hits. Here, we have established a semi-automated isPLA protocol to delineate and detail the life cycles and dynamics of endogenous linker-phosphorylated Smad2 at the single cell level. We also investigated effects of treatment with specific compounds by screening a library of phosphatase inhibitors having known modes of action (MoA) by targeting phosphorylation and with effects on complex formation. Our approach expands the scope of morphological pro-filing and offers a unique, information-rich, and largely unbiased approach to profile complex cellular responses and target deconvolution at the single cell level.

40

Future Perspectives

In this thesis I present target engagement-mediated amplification (TEMA) as an emerging technique to investigate target engagement. The technique al-lows convenient detection of binding by drug molecules via conjugated DNA strands, as a means to measure target specificity, in situ availability, and stability of binding to intended and unintended target molecules proteo-me-wide. A better prediction of clinical outcome will decrease the risk of failure due to lack of efficacy in phase III, which has been pointed out as the major and most expensive reason behind the high failure rates in drug devel-opment. TEMA can play an important role during drug discovery where it can be used to establish a link between target occupancy and pharmacologi-cal effect. Further technological developments will be explored with the aim to define a path to commercialization of the technology via product devel-opment, perhaps at an existing spinout company or through a new dedicated spinout focusing on analytic challenges in drug development. I also demonstrate the CETSA-PEA method for evaluating target engage-ment in cell extracts. The approach is particularly promising as a means to evaluate drug effects on limited material such as what may be obtained through fine needle biopsy from patients with solid tumors where only a few hundred cells may be available for analysis, disqualifying MS as an analytic method. In future work we aim to benchmark the CETSA-PEA panel by screening the Prestwick library of 1280 approved drugs in cell lines. The multiplex CETSA-PEA technique allows convenient analyses of targeted sets of proteins in many small samples aliquots, rendering the technique suitable for broad application during drug development and potentially also for therapy selection in routine clinical care. Finally, together with my colleagues I established a semi-automated, high-throughput isPLA approach to analyze specific cellular responses to treat-ment with cytokines and with drugs in individual cells. We show that the assays can be used to screen for signaling dynamics across millions of indi-vidual cells and to efficiently screen compound libraries in a 96-well format. We extended the possibility to use this high-throughput isPLA approach to screen for compound effects on specific molecular events in disease-derived primary cells to evaluate new and established treatment regimes.

41

Acknowledgement

This thesis work was mainly carried out at the Department of Immunology, Genetics and Pathology (IGP) at Uppsala University. Financial support was provided by a grant (2012-5852) from the Swedish Research Council’s spe-cial call for future therapy.

I am very grateful and would like to thank to my faculty opponent Prof. Christopher Schofield to get the opportunity to discussion my thesis and the evaluation of my work. I sincerely appreciate to my thesis committee board Prof. Amelie Eriksson Karlström, Associate Prof. Mårten Fryknäs, Associate Prof. Sara Mangsbo, Associate Prof. Marcel den Hoed and Associate Prof. Daniel Globisch for the time that you have taken to read my thesis. My deepest gratitude to my supervisor Ulf Landegren and greatly appreciat-ed all the supports that you have given me. I would also like to give big thanks to my co-supervisors Per Artursson and Andries Blokzijl for the sup-ports during my Ph.D. period.

Claes Wadelius, Christina Magnusson, Helene Norlin, Ulrica Bergström, Tuulikki Simu and the administrative stuffs at IGP, thanks for making my journey smooth with your kind and warm help. Thanks to the Men-tor4Research program by Royal Swedish Academy of Engineering Sciences (IVA) for giving me the amazing opportunity. Big thanks to my mentor Hel-ena S. for being my mentor and exciting journey.

I want to thank all my present and former colleagues in the MolTools and (MolDia)-groups for all the support and help that I have got all these years. It was a pleasant time that I have spent with you in this creative and inspiring group. Thanks Erik U, for your everlasting smile on your face for me and being a good mentor. Mats N, it was fantastic experience with you been part of MolTools, although I have not been closely working with you. I would also like to thank to Ola S, for sharing your great experience in situ PLA. Thank Masood for initial all your help for introducing in the MolTool-group and I am always grateful. Joakim, thanks for you’re to help me. I sincerely appreciate and thank to those without whom it would be difficult to make the MolTool-group and lab run properly. Elin E, Johanna, Christina C, thanks for doing everything to organize and create this great scientific environment.

42

Johan O, my deepest gratitude for your great patience and all the help re-garding my computers, Bengali conversion and maintenance of databases. Peter, I will always remember your help for my thesis, our interesting con-versations and all the best for your new journey. Caroline, thanks for all the support and help for data analysis and results evaluation in the CETSA pro-jects. Maria thanks you for all the nice scientific discussions and special gift for Rayan. Liza, thank you for the beautiful times in the lab and all the ap-preciation. Malte, thank your for being a friend, ever smiley face☺. Axel, thank you for helping in my tough time. Carl-Magnus, thank you for the important discussions about my real life and science. Karin G, you been always the best person to talk with. Linda, your enthusiasm and encourage of workout; very inspiring. Lei, an inspiring face of all time, thanks for your endless stamina and enthusiasm to devote yourself in science. Johan B, thanks for your inspiring and motivational discussion. Pathau, my office mate, thanks for all the lunch and dinners that we had together especially for tolerate the spiciness of Bengali food :). Marcus, always nice to talk with you and thanks for helping for CETSA project. Thanks David H, for all the interesting discussion we had especially; cultures. Felipe, thanks for your positive energy and all the jokes, and thank you for directing spex. Radiosa, thank you for bringing color in the lab. Hongxing, thanks for your warmth and sharing your interesting views. Ryoyo, Thanks for your love to science and help for data analysis. Ehsan, all the best for your future research work. Johan H, thanks for sharing inspiring knowledge and all the best talk we had. Di, your smiling face. Alireza, all interesting chats. Tonge, thanks for the fun times we had together and your everlasting smile. Annika, your warmth and positive energy, Anna E, thank for the interesting discussions and good time together. Sophie, thanks for all the interesting chats and your kindness, posi-tive attitude and all the well wishes for Rayan. Reddy, you motivation of hard work. Claudia, nice to meet at your lab in Stockholm. Sathis, welcome to Moltools group. Björn, thank you for all the nice talks about science. Ra-chel, thanks for your great enthusiasm. Spyros, a shining star in science and a great person, thanks for your kind willingness to help all the time. Jun-hong, thanks for sharing our office together. Dorothea, always nice to meet you. Takao, thanks for the interesting discussion about science. Johan V, thanks for introducing me in this inspiring and friendly lab. Agata, it was a great experience to discuss about molecular cell biology with you and An-dries together. Lotta W, thanks for all your help. I want to express my grati-tude to all my former colleagues in the MolTools-group for making our lab into such a great place to be; Camilla, Anja, Rongqin, Tomas (good time in Uppsala), Elin FS, Gucci, Anne-Li and Lotte M, Elin L, Thomasz, Tagrid, Hanan, Leisa, Samaneh, Mikaela, Yajun.

43

I also greatly appreciate and show my utmost gratitude to all my collabora-tors, Elder A, Lars J, Annika J, Thomas L, Nils L, Ronald S, Richard S, Sara L, and Per N, for great collaborations. Thanks to all my friends, here and all around the world, especially Ashik & Tareq. Thanks to the Bangladeshi community in Sweden (especially Stock-holm & Uppsala) and all of my friends for their encouragements and sup-ports during my PhD studies. I want to express my gratitude to all who supported me during these years has not mentioned and in different ways contributed to this work. Thanks Father for giving me the opportunity to do whatever I wanted. Reza vai thank you for your kind heart and inspiration all the time. Nasrin Apu thanks for being a mother instead of a sister. Thanks to Jahangir Chacha for being a guide in my whole life and giving me the instructions and show me a better way to reach my goal in life. My mother, grandparents and father-in-law, I really missing you all rest of my life. Thanks to all my brothers, brother-in-law, sisters, sisters-in-law and all of my family members for believing in me. My very special gratitude to wife Shahnaz for your love, supports in my tuff time and continuous believing on me over the years. Rayan (my son), you are my love, daylight and inspiration.

44

References

Anastassiadis, T. et al. Comprehensive assay of kinase catalytic activity reveals features of kinase inhibitor selectivity. Nat. Biotech. 29, 1039-1046 (2011).

Alarcón, C. et al. Nuclear CDKs drive Smad transcriptional activation and turnover in BMP and TGF-beta pathways. Cell 139, 757-69 (2009).

Assarsson, E. et al. Homogenous 96-plex PEA immunoassay exhibiting high sensi-tivity, specificity, and excellent scalability. PLoS One 9(4), e95192 (2014).

Bantscheff, M. et al. Quantitative chemical proteomics reveals mechanisms of action of clinical ABL kinase inhibitors. Nat. Biotech. 25, 1035-1044 (2007).

Bleicher, K. H. et al. Hit and lead generation: beyond high-throughput screening. Nat. Rev. Drg. Disco. 2, 369-378 (2003).

Bowes, J. et al. Reducing safety-related drug attrition: the use of in vitro pharmaco-logical profiling. Nat. Rev. Drg. Disco. 11, 909-922 (2012).

Brandt, B. et al. Small kinase assay panels can provide a measure of selectivity. Bioorg. & Med. Chem. Lett. 19, 5861-5861 (2009).

Brenner, S. et al. Encoded combinatorial chemistry. Proc. Natl. Acad. Sci. 89, 5381-5383 (1992).

Bunnage, M. E. et al. Target validation using chemical probe. Nat. Chem. Biol. 9, 195-199 (2013).

Carpenter, A. E. et al. CellProfiler: image analysis software for identifying and quantifying cell phenotypes. Genome Biol. 7(10), R100 (2006).

Casper, D. et al. Comparison of p38 kinase inhibitor binding affinity and kinetics using Biacore. Biacore Journal 3, 4-7 (2003).

Clark, M. A. et al. Design, synthesis and selection of DNA-encoded small-molecule libraries. Nat. Chem. Biol. 5, 647-654 (2009).

Cook, D. et al. Lessons learned from the fate of AstraZeneca’s drug pipeline: a five-dimensional framework. Nat. Rev. Drg. Disco. 13, 419-431 (2014).

Copeland R. A. et al. Drug–target residence time and its implications for lead opti-mization. Nat. Rev. Drg. Disco. 5, 730-739 (2006).

Crothers, D. M. et al. The influence of polyvalency on the binding properties of antibodies. Immunochemistry 9, 341-357 (1972).

Chan, A. I. et al. Novel selection methods for DNA-encoded chemical libraries. Curr. Opin. Chem. Biol. 26, 55-61 (2015).

Clark, M. A. et al. Selecting chemicals: the emerging utility of DNA-encoded li-brarie. Curr. Opin. Chem. Biol. 14, 396-403 (2010).

Darmanis, S. et al. Simultaneous Multiplexed Measurement of RNA and Proteins in Single Cells. Cell Reports 14(2), 380-389 (2016).

Davis, M. I. et al. Comprehensive analysis of kinase inhibitor selectivity. Nat. Bio-tech. 29, 1046-1052 (2011).

Dickson, M. et al. Key factors in the rising cost of new drug discovery and devel-opment. Nat. Rev. Drg. Disco. 3, 417-429 (2004).

Diggers, E. M. et al. The exploration of macrocycles for drug discovery--an under-exploited structural class. Nat. Rev. Drg. Disco. 7, 608-624 (2008).

45

Fabian, M. A. et al. A small molecule-kinase interaction map for clinical kinase inhibitors. Nat. Biotech. 23(3), 329-336 (2005).

Ferguson, F.M., et al. Kinase inhibitors: the road ahead. Nat. Rev. Drg. Disco. 17 (5): 353-377 (2018).

Franken, H. et al. Thermal proteome profiling for unbiased identification of direct and indirect drug targets using multiplexed quantitative mass spectrometry. Nat. Protocol. 10, 1567-1593 (2015).

Fredriksson, S. et al. Protein detection using proximity-dependent DNA ligation assays. Nat. Biotech. 20, 473-477 (2002).

Fuentealba, L.C. et al. Integrating Patterning Signals: Wnt/GSK3 Regulates the Duration of the BMP/Smad1 Signal. Cell 131, 980-993 (2007).

Garner, A.L. and Janda, K.D. Protein-protein interactions and cancer: targeting the central dogma. Curr. Top. Med. Chem. 11, 258-80 (2011).

Gullberg, M. et al. Cytokine detection by antibody-based proximity ligation. Proc. Natl. Acad. Sci. U S A 101, 8420-8424 (2004).

Gura, T. et al. DNA helps build molecular libraries for drug testing. Science 350, 1139-1140 (2015).

Goodnow, R. A. et al. DNA-encoded chemistry: enabling the deeper sampling of chemical space. Nat. Rev. Drg. Disco. 16,131-147 (2017).

Holdgate, G. A. et al. Affinity-based, biophysical methods to detect and analyze ligand binding to recombinant proteins: Matching high information content with high throughput. J. of Structural Biol. 172, 142-157 (2010).

Hopkins, A. L. et al. The druggable genome. Nat. Rev. Drg. Disco. 1, 727-730 (2002).

Heitner, T. R. and Hansen. N. J. Streamlining hit discovery and optimization with a yoctoliter scale DNA reactor. Expert Opin. Drg. Disco. 4, 1201-1213 (2009).

Hann, M. M. Molecular obesity, potency and other addictions in drug discovery. Med. Chem. Comm. 2, 349-355 (2011).

Jafari, R. et al. The cellular thermal shift assay for evaluating drug target interactions in cells. Nat. Protocol 9, 2100-2122 (2014).

Kolb, H. C. et al. Click Chemistry: Diverse Chemical Function from a Few Good Reactions. Angew. Chem. Int. Ed. 40, 2004-2021 (2001).

Kung, C. et al. Selective Kinase Inhibition by Exploiting Differential Pathway Sen-sitivity. Nat. Chem. Biol. 13, 399-407 (2006).

Li, Z. et al. Design and Synthesis of Minimalist Terminal Alkyne-Containing Diazir-ine Photo-Crosslinkers and Their Incorporation into Kinase Inhibitors for Cell- and tissue-Based Proteome Profiling. Angew. Chem. Int. Ed. 52, 1-7 (2013).

Lönn, P. and Landegren, U. Close Encounters - Probing Proximal Proteins in Live or Fixed Cells. Trends Biochem. Sci. 42(7), 504-515 (2017).

Liu, Y. et al. Rational design of inhibitors that bind to inactive kinase confor-mations. Nat. Chem. Biol. 2, 358-364 (2006).

Lundberg, M. et al. Homogeneous antibody-based proximity extension assays pro-vide sensitive and specific detection of low-abundant proteins in human blood. Nucleic Acids Research 39 (15), e102 (2011).

Leeson, P. D. and Springthorpe, B. The Influence of Drug-Like Concepts on Deci-sion-Making in Medicinal Chemistry. Nat. Rev. Drg. Disco. 6, 881-890 (2007).

Ma, H. et al. The challenge of selecting protein kinase assays for lead discovery optimization. Expert Opin. Drg. Disco. 3, 607-621 (2008).

Manning, G. et al. The Protein Kinase Complement of the Human Genome. Science 298, 1912-1934 (2002).

Martinez Molina, D. et al. Monitoring drug target engagement in cells and tissues using the cellular thermal shift assay. Science 341, 84-87 (2013).

46

Martinez Molina, D. et al. The Cellular Thermal Shift Assay: A Novel Biophysical Assay for In Situ Drug Target Engagement and Mechanistic Biomarker Studies. Annu. Rev. Pharmacol. Toxicol. 56, 141-161 (2015).

Matsuura, I. et al. Cyclin-dependent kinases regulate the antiproliferative function of Smads. Nature 430, 226-31 (2004).

Melkko, S. et al. Lead discovery by DNA-encoded chemical libraries. Drg. Disco. Tod. 12, 465-471 (2007).

Morgan, P. et al. Can the flow of medicines be improved? Fundamental pharmaco-kinetic and pharmacological principles toward improving Phase II survival. Drg. Disco. Tod. 17, 419-424 (2012).

Mullard, A. et al. DNA tags help the hunt for drugs. Nature 530, 367-369 (2016). Massagué, J. TGFβ signalling in context. Nat Rev Mol Cell Biol 13, 616–630 (2012). Massagué, J. TGFbeta in Cancer. Cell 134, 215–230 (2008). Overington, J. P. et al. How Many Drug Targets Are There? Nat. Rev. Drg. Disco. 5,

993-996 (2006). Obach, R. S. et al. Trend analysis of a database of intravenous pharmacokinetic

parameters in humans for 670 drug compounds. Drug Metab. Dispos. 36, 1385-1405 (2008).

Pan, A.C. et al. Molecular determinants of drug-receptor binding kinetics. Drg. Disco. Tod. 18, 667-673 (2013).

Picotti, P. et al. Selected reaction monitoring–based proteomics: workfows, poten-tial, pitfalls and future directions. Nat. Methods 9, 555-566 (2012).

Schallmeiner, E. et al. Sensitive protein detection via triple-binder proximity ligation assays. Nat. Methods 4, 135-137 (2007).

Schurmann et al., Small-Molecule Target Engagement in Cells. Cell Chem. Biol. (2016).

Simon, G. M. et al. Determining target engagement in living systems. Nat. Chem. Biol. 9, 200-205 (2013).

Smyth, L. A. et al. Measuring and interpreting the selectivity of protein kinase inhib-itors. The J. of Chem. Biol. 2, 131-151 (2009).

Söderberg, O. et al. Direct observation of individual endogenous protein complexes in situ by proximity ligation. Nat. Methods 3, 995-1000 (2006).

Savitski, M. M. et al. Tracking cancer drugs in living cells by thermal profiling of the proteome. Science 346, 1255784 (2014).

Smith, R. D. et al. Biophysical Limits of Protein-Ligand Binding. J. Chem. Inf. Mod. 52 (8), 2098-2106 (2012).

Shi, B. et al. Recent advances on the encoding and selection methods of DNA- en-coded chemical library. Bioorg. & Med. Chem. Lett. 27, 361-369 (2017).

Vincent, F. et al. Developing predictive assays: The phenotypic screening “rule of 3”. Sci. Transl. Med. 7, 293ps15 (2015).

Wrenn, S. J. and Harbury, P. B. Chemical evolution as a tool for molecular discov-ery. Annu. Rev. Biochem. 76, 331-349 (2007).

Waring, M. J. Lipophilicity in Drug Discovery. Exp. Opin Drug Disc 5, 235-248 (2010).

Walters, W. P. et al. What Do Medicinal Chemists Actually Make? A 50-Year Ret-rospective. J. Med. Chem. 54, 6405-6416 (2011).

Zhang, J. et al. Targeting cancer with small molecule kinase inhibitors. Nat. Rev. Cancer 9, 28-38 (2009).

Zimmermann, G and Neri, D. DNA-encoded chemical libraries: foundations and applications in lead discovery. Drg. Disco, Tod. 21(11), 1828 (2016).

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Medicine 1533

Editor: The Dean of the Faculty of Medicine

A doctoral dissertation from the Faculty of Medicine, UppsalaUniversity, is usually a summary of a number of papers. A fewcopies of the complete dissertation are kept at major Swedishresearch libraries, while the summary alone is distributedinternationally through the series Digital ComprehensiveSummaries of Uppsala Dissertations from the Faculty ofMedicine. (Prior to January, 2005, the series was publishedunder the title “Comprehensive Summaries of UppsalaDissertations from the Faculty of Medicine”.)

Distribution: publications.uu.seurn:nbn:se:uu:diva-374329

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2019