66 Drug Discovery World Summer 2014

Medicinal Chemistry

As James Black is quoted as saying: “Themost fruitful basis for the discovery of anew drug is to start with an old drug1.”

Consequently, small-molecule drug discovery usu-ally involves an iterative process of moleculardesign, chemical synthesis, biological assay anddata analysis feeding directly into the next cycle,but this process always needs a chemical startingpoint. For two decades or more, the rate at whichdrugs against new targets are launched has beenrelatively constant but the rate of developing drugsagainst completely new classes of drug target has

been significantly lower2. However, over the sameperiod protein kinases have rapidly become one ofthe most significant classes of drug targets for thepharmaceutical industry, with the global marketfor kinase therapies being about US$15 billion perannum in 2010 and this value is predicted to dou-ble by 20203. Research on protein kinases is nowreported to account for approximately 30% of thedrug discovery programmes in the pharmaceuticalindustry and more than 50% of cancer researchand development3.

Target-focused compound libraries have been a

By Dr Jason Brownand Dr J. Mark

Treherne

Targeted chemical libraries:the keys to unlock theubiquitin system Is novel chemistry the final frontier forubiquitin system drug discovery?

Although the widespread use of target-focused libraries has led to the rapidexpansion of novel chemistry as research tools in drug discovery to exploitmany classes of drug target, the ubiquitin system still remains a largelyuntapped medicinal chemistry opportunity. Protein kinases, on the other hand,have become one of the most important classes of drug targets for thepharmaceutical industry over the last decade, following on from theexploitation of kinase-focused libraries for at least the last two decades. Likeprotein phosphorylation by kinases, protein ubiquitylation regulates manyaspects of cell function and provides a wealth of drug target opportunitiesacross many therapeutic areas including cancer, cardiovascular, metabolism,inflammation, neurodegeneration and infectious diseases. In this article, wepropose that chemical libraries which target the ubiquitin system are themissing keys to unlock the therapeutic potential of ubiquitin system drugdiscovery.

Drug Discovery World Summer 2014 67

Medicinal Chemistry

Figure 1A cartoon representation of ubiquitin (Ub) and ubiquitin-like (Ubl) signalling pathways. Ubiquitin or ubiquitin-likeproteins (red) are attached to substrate proteins (yellow) via a sequential process involving ATP-dependent ‘activation’of the Ub/Ubl by an E1 activating enzyme (orange), transfer to an E2 conjugating enzyme (green) followed byattachment to the substrate protein (yellow) – a process ‘catalysed’ by an E3 Ub/Ubl ligase which may be a singleprotein or a multi-subunit complex (blue). Ub/Ubl may be attached to the substrate as a monomer or a polymer (apolyubiquitin chain). Further a mono-ubiquitylated protein may itself act as a substrate for further polyubiquitylation.Polyubiquitin may take eight – and possibly more – chain linkage forms including K6, K11, K27, K29, K33, K48, K63 orM1. For example, the activated C-terminus of a distal ubiquitin may be linked to the Lysine-6 (K6) epsilon amino groupof a proximal ubiquitin and so on6. It is the deconjugating enzymes (DCEs) that remove Ub/Ubls from substrates andthe DCEs for ubiquitin deconjugation are referred to as deubiquitylases or deubiquitinases and may cleave the entireubiquitin chain from the substrate or perform what is referred to as ‘chain editing’, which involves the cleavage of theubiquitin chain to leave a mono-ubiquitylated substrate, which may then be ubiquitylated resulting in the attachment ofa chain with the same or a different ubiquitin linkage type. The outcome of substrate ubiquitylation is largelydetermined by the interaction of the ubiquitin chain with ubiquitin binding proteins (UBP – which may comprise adomain on a protein or part of a multi-subunit complex; purple) and which UBP the ubiquitylated substrate binds isdetermined by the structure of the ubiquitin chain which is in turn determined by its linkage type9. Intriguingly itappears that in many instances it is the subtype of E2 that determines what ubiquitin chain type is assembled on thesubstrate thus leading to a general working model in which the E3 provides substrate specificity while the E2determines linkage type and thus the destiny and/or altered functionality of the substrate (there are some exceptionsto this)26-29. Thus for example a K48-linked ubiquitin chain leads to proteasomal targeting of the ubiquitylated substratewhile K63 ubiquitylation confers on the substrate ‘non-degradative’ signalling functions. Proteasomal-dependentproteolytic destruction of a K48-linked ubiquitylated substrate occurs in an ATP-dependent manner and results inrelease and recycling of the ubiquitin monomers. In a further level of system regulation, phosphorylation has beendemonstrated to modulate E3 activity, substrate affinity for E3s and, most recently, the functionality of ubiquitin. Referto text for further details

KeyADP: Adenosine diphosphate; AMP: Adenosine monophosphate; ATP: Adenosine triphosphate; DCE: Deconjugating Enzymes (~100 predicted of which~90 deubiquitylases are known*); E1: E1 Activating Enzyme (~9 of which 2 ubiquitin E1s are known*); E2: E2 Conjugating Enzyme (~38 of which~36 ubiquitin E2s are known*); E3: E3 Ligase (700-~1,000 ub/ubl E3s and/or E3 subunits are predicted*); (p): Phosphate; Pi: Inorganic Phosphate; PPi:Pyrophosphate; Ub: Ubiquitin; Ubl: Ubiquitin-like protein (~10 are known*); UBP: Ubiquitin Binding Protein (>20 families described*). *Human proteins

DCEE3E2E1

Ub/Ubl

DCE

UBP

signalling or targeting

ATP

AMP+

PPi

ATP ADP + Pi

proteasome complex

(Kinases)ATP ADPDP

–(P)–(P)

(P)–

68 Drug Discovery World Summer 2014

Medicinal Chemistry

key enabling component of the tool kit opening upkinase drug discovery, consisting of collections ofcompounds designed to interact with a family ofrelated kinase targets4. Such libraries are used forscreening against therapeutic targets in order tofind hit compounds that may be further developedinto drugs and/or used as research tools to betterunderstand the underlying biology and its relevanceto pharmacological intervention. The design of tar-geted libraries can use structural information, whenavailable, but can also employ a model that incor-porates sequence and mutagenesis data to predictthe properties of the binding site. The pharmacolo-gy of known ligands of the target can also be usedto inform the development of libraries from chemi-cal scaffolds. However, computational methods forthe selection of molecules for drug discovery areconstantly evolving and new approaches are nowavailable that can assist the medicinal chemist inselecting new compounds for library synthesis.These methods may also incorporate simple calcu-lated properties, for example the so-called ‘Rule of5’5 to ensure libraries stay within sufficiently drug-like areas of chemical space as well as existingstructure activity relationship data describing theinteraction of the chemical matter with the princi-ple biological target, as well as related targets.

Ubiquitylation and phosphorylationhave striking biological parallelsUbiquitylation describes the covalent attachmentof a small 76-amino acid protein, ubiquitin, toother proteins. The ubiquitylation process involvesthree sequential steps, each of which is controlledby a different class of enzyme (Figure 1). In the firststep, a single ubiquitin molecule is ‘activated’ bythe ubiquitin activating enzyme (E1) to which it isconjugated, in an ATP-dependent reaction. In thesecond step, the ubiquitin molecule is transferredfrom E1 to a ubiquitin conjugating enzyme (E2) viaa transthiolation reaction. In the final step, ubiqui-tin is transferred to the protein substrate in aprocess mediated by an E3 ubiquitin ligase, whichprovides a binding platform for ubiquitin-chargedE2 and the substrate. However, in contrast to pro-tein phosphorylation, ubiquitin can not only beattached to a substrate as a monomer (mono-ubiq-uitylation) but may be conjugated in the form of apolyubiquitin chain and these may be assembledinto at least eight different types determined bywhich amine group the ‘activated’ C-terminus of adistal ubiquitin attaches itself to on the proximalubiquitin; through the epsilon amino group of aLysine side chain (K6, K11, K27, K29, K33, K48or K63) or through the alpha amino group of the

M1 residue (there are also reports of branched andmixed linkage chains)6. Further, monoubiquitylat-ed proteins may act as substrates themselves result-ing in them becoming polyubiquitylated (Figure 1).Even greater versatility is provided by further fam-ilies of ubiquitin-like proteins (including NEDD8,SUMO, FAT10 and ISG15), which are alsoattached covalently to proteins via their own dedi-cated E1, E2, E3 pathways. The formation ofpolyubiquitin chains and the function of theseubiquitin-like-modifier proteins make the ubiquitinsystem a more complex and, potentially, more ver-satile intracellular signalling mechanism for phar-macological manipulation than phosphorylation3.As is the case with phosphorylation, ubiquitylationis a reversible process. Deubiquitylases (or deubiq-uitinases; DUBs) catalyse the cleavage of ubiquitinfrom proteins, a process known as deubiquityla-tion. While a DUB deconstructs ubiquitin chains,other enzymes recognise substrate-conjugatedSUMO and NEDD8, for example, and are thusreferred to as desumolase and deneddylaseenzymes respectively. Together these families ofenzymes are known as deconjugating enzymes orDCEs (Figure 1). Interestingly, the total number ofDUBs is comparable with the number of proteinphosphatases but, taken together, the number ofE1-activating enzymes, E2-conjugating enzymesand E3 ligases encoded by the human genome(~700-1,000) exceeds the total number of proteinkinases. The sole function of the ubiquitin systemwas originally thought to be restricted to the regu-lation of protein turnover inside the cell, as attach-ing a ubiquitin chain of a specific linkage type to aprotein directs it to the proteasome for destruc-tion7 (Figure 1). The canonical chain type deter-mining that the protein to which it is attached willbe destined for proteasome-dependent destructionbeing K48-linked however, other forms of ubiqui-tylation (defined by different chain linkages orlinkage to non-lysine residues on substrate pro-teins) are known to occur and these may regulate abroad range of biological processes. However, theroles of some of these modifications remain to befully elucidated6,8. Just like phosphorylation, ubiq-uitylation can also induce conformational changesthat alter the biological function of proteins. Incommon with phosphorylation, many effects ofubiquitylation are mediated by interactions withmodification state-dependent binding proteins: inthis case ‘ubiquitin-binding proteins’. Alternativepolyubiquitin chain linkage types can adopt dis-tinct three-dimensional structures and interactwith different polyubiquitin-binding proteins toregulate specific cellular processes9.

Drug Discovery World Summer 2014 69

Medicinal Chemistry

Protein phosphorylation and ubiquitylationshould not be thought of as two distinct and sepa-rate regulatory systems because interactionsbetween them are critical for the normal control ofmany cell signalling processes. Given the ubiqui-tous nature of protein phosphorylation and ubiq-uitylation inside the cell, understanding the com-plex interplay between these two systems is likelyto become increasingly important in basic research,drug discovery, small molecule screening and selec-tivity profiling.

Current drugs and compounds inclinical trials that modulate the ubiquitin systemThe proteasome inhibitor Bortezomib (marketedas Velcade®) was the first approved drug to targeta key element of the ubiquitin system. Bortezomibwas approved as a treatment for multiple myelo-ma and is a dipeptidyl boronic acid derivative,which is given intravenously, and binds non-cova-

lently to catalytic subunits of the proteasome toinhibit their chymotrypsin-like activity. The pre-cise molecular basis for its efficacy had been some-what uncertain. However, it is now thought thatits mechanism of action is due, in large part, to theinduction of a selective increased proteotoxicstress load in multiple myeloma cells, whichexpress high levels of misfoldedimmunoglobulins10. There is considerable interestin developing improved less-peptidic inhibitorsthat can be taken orally and a number of suchcompounds are already in clinical trials (Table 1).Although Bortezomib demonstrates the clinicalbenefits of modulating a component of the ubiq-uitin system, the compound is a more ‘classical’protease inhibitor and does not illustrate a mech-anistic proof of concept for modulating eithermembers of the ligase machinery, DUB families orother elements of the ubiquitin cascade.

There are two inhibitors of E1 enzymes in clini-cal trials (Table 2). MLN4924 targets NAE1

References1 Raju, TN. The NobelChronicles. 1988: James WhyteBlack, (b 1924), Gertrude Elion(1918-99), and George HHitchings (1905-98). Lancet,2000. 355(9208): p. 1022.2 Overington, JP, Al-Lazikani, Band Hopkins, AL. How manydrug targets are there? NatRev Drug Discov, 2006. 5(12):p. 993-6.3 Cohen, P and Tcherpakov, M.Will the ubiquitin systemfurnish as many drug targets asprotein kinases? Cell, 2010.143(5): p. 686-93.4 Harris, CJ et al. The designand application of target-focused compound libraries.Comb Chem High ThroughputScreen, 2011. 14(6): p. 521-31.5 Lipinski, CA et al.Experimental andcomputational approaches toestimate solubility andpermeability in drug discoveryand development settings. Adv.Drug Delivery Rev., 1997. 23: p. 3-25.6 Komander, D. The emergingcomplexity of proteinubiquitination. Biochem SocTrans, 2009. 37(Pt 5): p. 937-53.7 Ciechanover, A and Stanhill,A. The complexity ofrecognition of ubiquitinatedsubstrates by the 26Sproteasome. Biochim BiophysActa, 2014. 1843(1): p. 86-96.8 McDowell, GS and Philpott,A. Non-canonicalubiquitylation: Mechanisms andconsequences. Int J BiochemCell Biol, 2013. 45(8): p. 1833-42.9 Husnjak, K and Dikic, I.Ubiquitin-binding proteins:Decoders of ubiquitin-mediated cellular functions.Annu Rev Biochem, 2012. 81:p. 291-322.10 Kubiczkova, L et al.Proteasome inhibitors –molecular basis and currentperspectives in multiplemyeloma. J Cell Mol Med,2014. [Epub ahead of print].

Continued on page 71

COMPANY INHIBITOR DEVELOPMENTSTATUS

INDICATION

Millennium/Takeda Bortezomib/Velcade® (PS-341)

Approved Multiple myeloma

ONYX (Proteolix) Carfilzomib/Kyprolis®(PR171)

Approved Multiple myeloma

ONYX (Proteolix) Oprozomib (Onx 0912/PR047)

Phase I/II Cancer

Nereus Pharmaceuticals Marizomib/SalinosporamideA/NPI-0052

Phase I/II Cancer

Millennium/Takeda Ixazomib (MLN9708) Phase I/II Cancer

Cephalon/Teva Delanzomib (CEP-18770) Phase I Cancer

Table 1: Marketed or clinical stage proteasome inhibitors

COMPANY INHIBITOR TARGET DEVELOPMENTSTATUS

INDICATION

Millennium/Takeda MLN4924 NAE1 Phase I/II Cancer

Millennium/Takeda MLN7243 UAE1 Phase I Cancer

Table 2: Clinical stage inhibitors of E1-activating enzymes

KeyNAE1: NEDD8 Activating Enzyme 1; UAE1: Ubiquitin Activating Enzyme 1

Medicinal Chemistry

(NEDD8 Activating Enzyme 1), the E1 activatingenzyme for the ubiquitin-like protein NEDD811.The key targets of ‘NEDDylation’ are the Cullinscaffold proteins of the Cullin Ring Ligase (CRL)sub-family of E3 ubiquitin ligases. CRLs – in par-ticular the SKP1-Cullin1-F-box sub-family – arethought to be the key sub-family of E3 ligases con-trolling turnover of proteins involved in cellularproliferation12. MLN4924 is currently in Phase I/IItrials. More recently MLN7243, a UAE1(Ubiquitin Activating Enzyme 1) entered Phase Idose escalation studies.

There are currently no inhibitors of E2s in clini-cal trials although there have been reports of asmall molecule E2 inhibitor whose mechanism ofaction appears to involve the stabilisation of theinteraction of ubiquitin with the E2 donor ubiqui-tin-binding site13,14.

Unexpectedly, thalidomide has been found tobind and alter the substrate specificity of an E3ligase substrate binding adaptor called cereblon15

that is important for limb outgrowth and theexpression of a fibroblast growth factor duringembryonic development16. This finding explainswhy thalidomide, which was originally prescribedas a sedative and then as a treatment for nausea,caused severe developmental defects in unbornchildren. Nevertheless, thalidomide is still used forthe treatment of numerous conditions, includingleprosy, skin sores and cancer. Indeed Celgene hasa number of thalidomide (IMiD®) analogues inclinical trials across a number of applicationsincluding multiple myeloma and acute myeloidleukaemia. Discovering the molecular mechanismof the compound’s devastating side effects fivedecades after its first use facilitates both a betterunderstanding of the mechanism of action of thisfamily of molecules but also an intriguing aspectof ubiquitin system biology – namely the ability ofa small molecule to modulate E3 substrate speci-ficity, a phenomena previously identified inplants17, and that could help to inform oneapproach to small molecule targeting of E3 ligas-es. Thalidomide was not originally designed totarget an E3 ligase but demonstrates that orallybioavailable ligase modulators are chemically pos-sible. A number of inhibitors or modulators of E3ligases are in clinical trials (Table 3) and severalare under pre-clinical investigation including asmall molecule activator of the E3 ligaseParkin18,19.

Although the above list is significantly less thanfor kinase-targeted drugs, pre-clinical drug dis-covery interest and activity in the ubiquitin fieldis increasing.

www.elrig.org

Manchester, UKCentral Convention Centre

2nd+3rd Sept ‘14ELRIG

ELRIG Drug Discovery 2014

#ELRIGDD14

FREE TO

ATTEND

Register NOW

Plenary Keynote : Susan Galbraith, Global Head of Oncology, iMed, AstraZeneca

“Genetic drivers of cancer growth and resistance mechanisms”

“A fantastic programme …” Steve Rees, Astra Zeneca

3 SLAS Short Courses : Monday 1st September• Introduction to Laboratory Automation

• High Content Screening: Instrumentation, Assay Development, Screening, Image and Data Analysis• Establishing Cell-Based Assays and Implementing 3D Culture Models for screening and drug testing

We have a great programme lined up for you … including …

Extensive Trade Exhibition • Innovation Zone • Careers Zone • Media Zone • Dragon’s Den• Networking Sessions • Posters • Prizes • Seminars

Session Session Keynote Presentation

Oncology Drug Discovery Ollie KallioniemiFIMM

Systems medicine in acutemyeloid leukemia ( AML )and beyond

Innovations in Screeningand Assay Development

Lorenz MayrAstra Zeneca

High-ThroughputScreening – Challenges &Opportunities

Phenotypic Discovery &Cellular Imaging

Daniela GabrielNovartis

Phenotypic Assays in DrugDiscovery

Core Technology Forums

Target ValidationApproaches in DrugDiscovery

Stefan OffermannsMax Planck Institute

Regulation of metabolismand immune functionsthrough metabolite GPCRs

Chemical Innovation in LeadDiscovery

Dario NeriETH Zurich

Targeting tumours witharmed antibodies and withsmall targeted cytotoxics

Engineered Proteins forTherapy

Lutz JermutusMedImmune

Novel targets and antibodyadd-ons

Neuroscience Drug Targets Eric KarranAlzheimer’s Research UK

Drug development inAlzheimer’s disease: dark-ness before the dawn?

Tues

day

2nd

Sep

tW

edne

sday

3rd

Sep

t

Is a fully integrated sample management capability theFuture?Innovative Automation – What can we automate next?Safety & Metabolism AssaysComplex, High Content & Biologics Data

FREE TO ATTEND! – Register now:http://elrig.org/portfolio/drug-discovery-2014/

Each of these is a full-day session, so take a look at our website for the full programme

70 Drug Discovery World Summer 2014

Drug Discovery World Summer 2014 71

Medicinal Chemistry

Ubiquitylation/phosphorylation cross-talkIntriguingly the close interplay and cross-talkbetween ubiquitylation and phosphorylation(Figure 1) enables a further approach to controllingubiquitylation pathways, that being via modulationof the phosphorylation of ubiquitylation ligasemachinery, ubiquitin ligase substrates or even ubiq-uitin itself. Indeed, substrate phosphorylation iscritical to the recognition of a number of substratesby E3 ligase substrate binding adaptors20 and like-wise E3 ligases may be activated in a phosphoryla-tion dependent manner; for example, IRAK-catal-ysed phosphorylation of the E3 Pellino21 and Srcfamily kinase (SFK) phosphorylation of the E3

TRAF622. Indeed, this cross-talk operates in bothdirections with TRAF6 ubiquitylating SFKs. A fur-ther level of complexity and demonstration of theinterplay between ubiquitylation and phosphoryla-tion is exemplified by the recent intriguing discov-ery that ubiquitin itself may be phosphorylated andplay a role in E3 ligase activation. Thus activationof the E3 ligase parkin is dependent upon bothphosphorylation of a ubiquitin-like domain on theE3 ligase itself and interaction with phosphorylatedubiquitin; both parkin and the ubiquitin with whichit interacts are phosphorylated by PINK123-25.These examples underline the fact that post-transla-tional modifications, such as ubiquitylation andphosphorylation, may display significant cross-talk

Continued from page 69

11 Nawrocki, ST et al.MLN4924: A novel first-in-classinhibitor of NEDD8-activatingenzyme for cancer therapy.Expert Opin Investig Drugs,2012. 21(10): p. 1563-73.12 Zhou, W, Wei, W and Sun, Y.Genetically engineered mousemodels for functional studiesof SKP1-CUL1-F-box-protein(SCF) E3 ubiquitin ligases. CellResearch, 2013. 23(5): p. 599-619.13 Ceccarelli, DF et al. Anallosteric inhibitor of thehuman cdc34 ubiquitin-conjugating enzyme. Cell, 2011.145(7): p. 1075-87.14 Huang, H et al. E2 enzymeinhibition by stabilization of alow-affinity interface withubiquitin. Nat Chem Biol,2014. 10(2): p. 156-63.15 Licht, JD, Shortt, J andJohnstone, R. From anecdoteto targeted therapy: Thecurious case of thalidomide inmultiple myeloma. Cancer Cell,2014. 25(1): p. 9-11.16 Ito, T et al. Identification ofa primary target ofthalidomide teratogenicity.Science, 2010. 327(5971): p. 1345-50.17 Tan, X and Zheng, N.Hormone signaling throughprotein destruction: A lessonfrom plants. Am J PhysiolEndocrinol Metab, 2009.296(2): p. E223-7.18 Johnston, JA. UbiquitinDrug Discovery andDiagnostics Conference –targeting E3 ligases. IDrugs,2010. 13(10): p. 695-7.19 Regnstrom, K et al. Labelfree fragment screening usingsurface plasmon resonance asa tool for fragment finding –analyzing parkin, a difficultCNS target. PLoS One, 2013.8(7): p. e66879.20 Evrard-Todeschi, N et al.Structure of the complexbetween phosphorylatedsubstrates and the SCF beta-TrCP ubiquitin ligase receptor:A combined NMR, molecularmodeling, and dockingapproach. J Chem Inf Model,2008. 48(12): p. 2350-61.

Continued on page 72

KeyCereblon: E3 ubiquitin ligase substrate binding adaptor; HDM2: Human Double Minute 2 Homologue; IAP: inhibitor of apoptosis protein; XIAP: X-linked inhibitor of apoptosis protein

COMPANY MODULATOR TARGET DEVELOPMENTSTATUS

INDICATION

Celgene Thalidomide(Thalidomid®)

Cereblon Approved Multiple myeloma

Celgene Pomalidomide(Pomalyst®)

Cereblon Approved Multiple myeloma

Celgene Lenalidomide(Revlimid®)

Cereblon Approved Mantle celllymphoma

AegeraTherapeutics

AEG 35156(antisenseoligonucleotide)

XIAP Phase II Cancer

Astellas Pharma YM155 Survivin promoter Phase I/II Cancer

Novartis LCL161 IAP Phase I/II Cancer

Tetralogics Pharma Birinapant/(TL-32711)

IAP Phase I/II Cancer

Roche Nutlin(RO5045337/RG7112)

HDM2 Phase I Cancer

Johnson & Johnson Serdemetan (JNJ-26854165)

HDM2 Phase I Cancer

Genentech/Roche GDC-0152 IAP Phase I Cancer

AscentaTherapeutics

AT-406 IAP Phase I Cancer

AegeraTherapeutics

AEG40826/HGS1029

IAP Phase I Cancer

Table 3: Clinical stage inhibitors of E3 ubiquitin ligases

72 Drug Discovery World Summer 2014

Medicinal Chemistry

and interdependency and thus benefit from beingstudied through an integrated cross-specialtyapproach.

The opportunity for chemical librariestargeting the ubiquitin system The parallels between the biology of protein phos-phorylation and ubiquitylation, as well as theirexploitation for the development of drugs to treatdiseases have been well described previously byPhilip Cohen and colleagues3. In summary, bothbiological regulatory mechanisms were identifiedmany years ago but interest in targeting them sys-tematically for drug discovery only really started totake off in earnest over the last two decades. Thefirst compounds inhibiting components of thesesystems entered clinical trials at around the sametime: Bortezomib in 1997 and Gleevec (the firstkinase inhibitor) in 1998. Gleevec was approved in2001 and overtook Bortezomib, which wasapproved later in 2003. Both Gleevec andBortezomib have subsequently proved to be of sig-nificant clinical benefit in the treatment of cancer.

However, kinase inhibitors have subsequentlyshot ahead. Following on from the development ofGleevec, about 25 drugs targeting protein kinaseshave been clinically approved for use, mostly incancer, whereas only one other drug (Kyprolis®;Carfilzomib) targeting the ubiquitin system hasbeen approved since Bortezomib; both of thesemolecules target catalytic subunits of the protea-some. In addition, kinase inhibitors currentlyundergoing clinical trials also outnumber theinhibitors of the ubiquitin system by more than 10to one3. As pointed out by Philip Cohen, a key fac-tor driving the kinase field forward at such a rapidpace is the ease with which targeted chemicallibraries can be synthesised and exploited to devel-op inhibitors of many protein kinases from thesame subfamily3. This observation probablyexplains the stark disparity between the two differ-ent classes of drug target.

Although E3 ubiquitin ligases outnumber pro-tein kinases, medicinal chemists have still notdeveloped a widely available targeted libraryapproach for identifying inhibitors of E3s. To date,chemists have mostly focused on disrupting theinteraction between E3 ligases and their substrates,interactions which are likely to be specific to par-ticular E3 ligase-substrate pairs. One explanationof this discrepancy is that finding compounds todisrupt the interface of two proteins can be intrin-sically more difficult to achieve – and possibly lessgenerically applicable across other members of theprotein family – than searching for small molecules

that block enzyme catalytic activity, particularlywhere one of the substrates is common to allenzyme sub-family members as with ATP in thecase of kinases.

Surprisingly, less effort has been devoted todeveloping compounds that disrupt the interac-tions between E2-conjugating enzymes and E3 lig-ases or at targeting E2s directly. E2-E3 interactionsare usually relatively weak and may therefore berelatively easy to disrupt. Moreover, compoundsthat disturb the interaction between an E2-conju-gating enzyme and an E3 ligase could, in principle,exert their effects by binding to the E2, the E3, theE2-E3 interface or via an allosteric mechanism,creating the potential to identify three or four typesof inhibitors from a single screen. There are ~40E2-conjugating enzymes encoded by the humangenome; therefore, on average, each E2 might bepredicted to interact productively with ~15 E3 lig-ases although the reality is likely to be more com-plicated than this with certain E2s having distinctroles26-29 whose interactions with E3s may be reg-ulated as much by spatial and temporal factors asbinding affinities. Compounds that disrupt E2-E3interactions by binding specifically to the E3 ligasecould be identified by counter screening withanother E3 that also forms a productive interactionwith the same E2. Focusing efforts on large fami-lies of E3 ligases may lead to the development ofchemical libraries with the capability of disruptingmany E2-E3 interactions. By analogy with kinases,perhaps the key to developing inhibitors of specif-ic E2-E3 interactions is to find compounds thatbind to small hydrophobic pockets on the E2 or E3located proximal to the E2-E3 interface itself or toidentify allosteric inhibitors that disrupt the E2-E3interaction by inducing long-range conformationalchanges. The determination of such three-dimen-sional structures of E2-ubiquitin/E3 ligase com-plexes will be critical to such efforts. For example,the solving of the structure of the E3 ligase CBL-B/E2~ubiquitin complex may not only help toinform such approaches but has also revealed howthe phosphorylation of Tyrosine-363 in the E3induces a structural event that enhances catalyticefficiency by ~200-fold, providing a further exam-ple of the close interplay between these two keysignalling systems30. The CBL family of E3 ligasesattenuate non-receptor and receptor tyrosinekinase signalling by ubiquitylating and therebydirecting these kinases for degradation through theendocytic or proteasomal pathway31.

Another area where more effort will probably befruitful is the design and generation of chemicallibraries to target the large and diverse families of

Continued from page 71

21 Ordureau, A et al. TheIRAK-catalysed activation ofthe E3 ligase function ofPellino isoforms induces theLys63-linked polyubiquitinationof IRAK1. Biochem J, 2008.409(1): p. 43-52.22 Liu, A et al. TRAF6 proteincouples Toll-like receptor 4signaling to Src family kinaseactivation and opening ofparacellular pathway in humanlung microvascular endothelia.J Biol Chem, 2012. 287(20): p. 16132-45.23 Kazlauskaite, A et al. Parkinis activated by PINK1-dependent phosphorylation ofubiquitin at Ser65. Biochem J,2014. 460(1): p. 127-39.24 Koyano, F et al. Ubiquitin isphosphorylated by PINK1 toactivate parkin. Nature, 2014.510(7503): p. 162-6.25 Shaw, GS. Switching onubiquitylation byphosphorylating a ubiquitousactivator. Biochem J, 2014.460(3): p. e1-3.26 Christensen, DE, Brzovic,PS and Klevit, RE. E2-BRCA1RING interactions dictatesynthesis of mono- or specificpolyubiquitin chain linkages.Nat Struct Mol Biol, 2007.14(10): p. 941-8.27 Wenzel, DM, Stoll, KE andKlevit, RE. E2s: Structurallyeconomical and functionallyreplete. Biochem J, 2010.433(1): p. 31-42.28 Pruneda, JN et al. Ubiquitinin motion: Structural studies ofthe ubiquitin-conjugatingenzyme approximatelyubiquitin conjugate.Biochemistry, 2011. 50(10): p. 1624-33.29 Metzger, MB et al. RING-type E3 ligases: Mastermanipulators of E2 ubiquitin-conjugating enzymes andubiquitination. Biochim BiophysActa, 2014. 1843(1): p. 47-60.30 Dou, H et al. Essentiality ofa non-RING element inpriming donor ubiquitin forcatalysis by a monomeric E3.Nat Struct Mol Biol, 2013.20(8): p. 982-6.

Continued on page 74

Drug Discovery World Summer 2014 73

Medicinal Chemistry

DUBs (Figure 2). Deubiquitylases are attractingincreasing attention as drug targets across manytherapeutic areas, including cancer32-34 and infec-tious diseases35. Inhibitors of many DUBs arealready in pre-clinical development. Learning fromexperience with kinase-targeted libraries hastaught us that compounds developed as inhibitorsof one protein kinase commonly turn out to inhib-it other protein kinases, sometimes with evengreater potency, and can kick start completely newdrug discovery projects. Therefore, developingchemical libraries that target DUBs is likely to yieldsimilar surprises and generate drug leads targetinga number of these enzymes. Iterative hit-to-leadmedicinal chemistry optimisation after such libraryscreening and hit identification is commonly sup-ported by selectivity profiling. Such activitiesinform us which chemical series deliver therequired balance between target affinity and selec-tivity versus enzymes of the same family (eg relat-ed DUBs; see Figure 2).

To be effective, targeted libraries need to deliv-er useful Structure Activity Relationship (SAR)data to the medicinal chemist. Such libraries alsoprovide opportunities for the identification ofpotent tool molecules to probe target tractabilityand mechanism of action. This offers significantscope for impacting early stage biology whereresearch efforts may be hampered by inadequatetools to elucidate the biology and, at the sametime, kick start hit-to-lead optimisation.

Currently any one of a number of hit identifica-tion methods may identify a good number ofinteresting looking molecules which then gothrough a rigorous triage process to ensure onlythe most tractable are actually resourced in thelaboratory. Evaluation of a wider range of chem-ical hits though targeted libraries, coupled withstringent design criteria, will enable a more thor-ough evaluation of the accessible chemical space,

USP15

0 200 400 600 800 1000 1200 1400 1600

981 aa

USP4 963 aa

USP11 963 aa

MIT

MIT

PH

PH

PH

USP32 1604 aa

USP6 1406 aa

USP8 1118 aa

USP19 1318 aa

USP21 565 aa

USP2 605 aa

USP31 1352 aa

USP43 1124 aa

USP50* 339 aa

USP39* 565 aa

USP20 914 aa

USP33 942 aa

USP16 823 aa

USP45 814 aa

USP3 520 aa

USP51 711 aa

USP22 525 aa

USP27X 438 aa

USP10 798 aa

USP17 530 aa

USP36 1121 aa

USP42 1325 aa

USP44 712 aa

USP49 688 aa

USP12 370 aa

USP46 366 aa

USP35 1017 aa

USP38 1042 aa

USP37 979 aa

USP26 913 aa

USP29 922 aa

USP13 863 aa

USP5 858 aa

CYLD 956 aa

USP48 1035 aa

USP14 494 aa

USP47 1375 aa

USP40 1235 aa

USP7 1102 aa

USP30 517 aa

USP18 372 aa

USP41 358 aa

USP25 1087 aa

USP28 1077 aa

USP34 3546 aa

USP24 2620 aa

USP9X 2547 aa

USP9Y 2555 aa

USP1 785 aa

USP52* 1202 aa

USPL1 1092 aa

USP53* 1073 aa

USP54* 1684 aa

Z

Z

M

B

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL UBL UBL UBL UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBL

UBLUBL

D D D

D

D

D

D D

D D

D

WD40

CAAX

CS CS

Rhod

CAP CAP CAP

TBC

MATH

Exo

Z

Z

Z

Z

Z

Z

Z

Z

Z

Z

UBL

+1000/+1000

+600/+400

+700/+300

+400

+700/+300

USPFamily

Key

MIT

PH

M

B

UBL

D

WD40

CAAX

Z

***

USP catalytic domain

PH

ZnF-UBP

ZnF-MYND

non-Zn binding ZnF motif

B-box

Ub-like

DUSP

MIT domain

UBA

UIM

Transmembrane

EF_hand, not Ca2+ binding

EF_hand, Ca2+ binding

Ataxin-2 C

WD40 ß Propeller

Rhodanese, Exonuclease, TBC-RABGAP, CAP-Gly, MATH domain, CS domain

CAAX-box

node supported by a bootstrap value of >50%

predicted to be inactive

hypothetical protein

UCHL1 223 aa

UCHL3 230 aa

UCHL5 329 aa

BAP1 729 aa

ATXN3 376 aa

ATXN3L 355 aa

JOSD1 202 aa

JOSD2 188 aa

OTUD4 1113 aa

HIN1L** 443 aa

YOD1 348 aa

OTUD1 481 aa

OTUD5 571 aa

OTUD3 398 aa

OTUD6A 288 aa

OTUD6B 293 aa

OTUB1 271 aa

OTUB2 234 aa

VCPIP 1222 aa

TRABID 708 aa

A20 790 aa

Cezanne2

FAM105*

926 aa

356 aa

Cezanne

OTULIN

843 aa

352 aa

CSN5 334 aa

POH1 310 aa

BRCC3 316 aa

MPND 471 aa

MYSM1 828 aa

EIF3H* 352 aa

CSN6* 327 aa

PSMD7* 324 aa

EIF3F* 357 aa

AMSH 424 aa

AMSH-LP 436 aa

PRPF8* 2335 aa

UBL

UBL

T

S

N N N

A

A

A A A A A A A

PFU

ANK

MIT

1 2 3

4 5 6 7

MIT

SW

C

UCH

Josephin

OTU

JAMM/MPN+

Family

Family

Family

Family

Key

MIT

UBL

T

S

C

N

A

PFU

ANK

SW

***

UCH

Josephin

OTU

catalytic domains

JAMM

Ub-like

MIT domain

UBA

UIM

ZnF-A20

NZF-type ZnF

ZnF-C2H2

PFU

ANK-UBD

SWIRM domain

SANT domain

Tudor domain-related

MitMem_reg domain

PRPF8 domains

node supported by a bootstrap value of >50%

predicted to be inactive

hypothetical protein

Figure 2A set of phylogenetic trees representing the families of

Deconjugating Enzymes (DCEs) constructed fromcomparing sequence homologies of the de-ubiquitylating

catalytic domains of the enzymes. Further enzymearchitecture is indicated as predicted from domain

(CDD, HHsearch, Pfam and SMART) and structure (PDB)databases. These trees and the key are adapted from

http://ubiquigent.com/services/dubprofiler_compound_screening_service/usp_family/ and presented with the

permission of Professor Michael Clague and ProfessorSylvie Urbé at the Institute of Translational Medicine, and

Institute of Integrative Biology, University of LiverpoolUnited Kingdom

KeyPH: Pleckstrin Homology Domain; ZnF-UBP: Ubiquitin Carboxyl-

terminal Hydrolase-like Zinc Finger domain; ZnF-MYND: [Myeloid,Nervy and DEAF-1]-like Zinc Finger domain; B-box: B-box-type zinc

finger domain; DUSP: Domain in Ubiquitin-Specific Proteases;MITdomain: Microtubule Interacting and Trafficking molecule domain;

UBA: Ubiquitin Associated Domain; UIM: Ubiquitin-Interacting Motif;EF_hand: EF_hand, calcium binding motif; PFU: PLAA family UbiquitinBinding Domain; ANK-UBD: Ankyrin-repeat Ubiquitin Binding Domain;

MittMem_regdomain: Involved in maintenance of mitochondrialstructure and function

74 Drug Discovery World Summer 2014

Medicinal Chemistry

which would be expected to impact positively onthe final quality of candidates being produced.

A key element of the fully integrated approach todrug discovery is the rapid cycle time, whereby bio-logical assays are integrated with the targetedlibraries. Consequently, integrated platforms, andassay kits combining biology and chemistry, willenable a range of chemical structures and corre-sponding biological activity data to potentiallytransform ubiquitin system drug discovery.However, the keys to unlocking the potential ofubiquitin system drug discovery are the introduc-tion of chemical libraries targeting DUBs and otherenzyme classes of the ubiquitin system. AsUbiquigent is based in the Sir James Black Centrewe are now following his advice and, even if thereare few existing ‘old drugs’ to start from, we areplanning to fill this chemical void with newlibraries as ‘the most fruitful basis for the discoveryof new drugs’ targeting the ubiquitin system.Ultimately, the acid test for our prediction of tar-geted chemical libraries being the keys to unlockubiquitin system drug discovery will be whendevelopment candidates targeting the ubiquitinsystem approach or even overtake those targetingprotein kinases. DDW

Dr Jason Brown co-founded Ubiquigent in 2009 asits Managing and Scientific Director in collabora-tion with the University of Dundee, the MedicalResearch Council and Stemgent Inc. Before start-ing Ubiquigent he was part of a biotech investmentand operations group and involved in supporting amolecular diagnostics, kinase drug discovery andvarious other drug discovery-focused service com-panies as well as evaluating investment opportuni-ties. Prior to this he built and ran a kinase-focusedassay development and drug discovery servicefacility for Upstate Biotechnology, a leadingprovider of cell signalling research products andservices which grew out of a close collaborationwith Sir Philip Cohen and his MRC Protein

Phosphorylation Unit (now the MRC ProteinPhosphorylation and Ubiquitylation Unit whereUbiquigent is based). Jason received his MPhil andDPhil from the University of Cambridge in associ-ation with Parke-Davis/Warner-Lambert (Pfizer),during which he identified a voltage-dependent cal-cium channel subunit as the molecular target of theblockbuster epilepsy and neuropathic pain drugsNeurontin and Lyrica. After his DPhil Jasonworked in and subsequently ran an assay develop-ment group for Parke-Davis.

Dr Mark Treherne has been actively involved in thebiopharmaceutical industry for more than 25 yearsand previously led the neurodegeneration researchgroup at Pfizer’s research facility in Sandwich,where he initiated a number kinase drug discoveryprojects with Sir Philip Cohen in Dundee. In 1997,he co-founded Cambridge Drug Discovery asChief Executive, leading the company’s subsequentacquisition by BioFocus plc, where he becameCommercial Director and drove significant growthof the profitable services business, again in collab-oration with Sir Philip Cohen. BioFocus pioneeredthe commercial exploitation of kinase-focusedlibraries. Dr Treherne joined Ubiquigent asChairman in 2013 to help unlock the potential ofubiquitin system drug discovery from the Sir JamesBlack Centre in Dundee. He has a BSc inPhysiology and Pharmacology from the UniversitySt Andrews and a PhD in Pharmacology from theUniversity of Cambridge.

Continued from page 72

31 Mohapatra, B et al. Proteintyrosine kinase regulation byubiquitination: Critical roles ofCbl-family ubiquitin ligases.Biochim Biophys Acta, 2013.1833(1): p. 122-39.32 Lim, KH and Baek, KH.Deubiquitinating enzymes astherapeutic targets in cancer.Current PharmaceuticalDesign, 2013. 19(22): p. 4039-52.33 Jacq, X et al.Deubiquitylating enzymes andDNA damage responsepathways. Cell BiochemBiophys, 2013. 67(1): p. 25-43.34 Fraile, JM et al.Deubiquitinases in cancer:New functions and therapeuticoptions. Oncogene, 2012.31(19): p. 2373-88.35 Nanduri, B et al.Deubiquitinating enzymes aspromising drug targets forinfectious diseases. CurrentPharmaceutical Design, 2013.19(18): p. 3234-47.

BioTek Instruments, Inc 20

Cell Signaling Technology, Inc 6

Cellular Dynamics International 35

eBioscience, Inc 4

ELRIG 70

ForteBio, Inc 3

Horizon Discovery Ltd 36

HTStec Ltd 33

IDEXX BioResearch 28

IntelliCyt Corporation 8

Molecular Devices LLC 56

Seahorse Bioscience, Inc 42

Select Biosciences Ltd IBC

Singulex, Inc 13

Tecan Schweiz AG IFC

Thermo Fisher Scientific 27

Waters Corporation OBC

ADVERTISEMENT INDEX