WO2020049130A1

WO2020049130A1 - Methods

Info

Publication number: WO2020049130A1
Application number: PCT/EP2019/073769
Authority: WO
Inventors: Nicholas Carl Kristoffer VALERIE; Brent David George Page; Mikael ALTUN
Original assignee: Valerie Nicholas Carl Kristoffer; Brent David George Page; Altun Mikael
Priority date: 2018-09-05
Filing date: 2019-09-05
Publication date: 2020-03-12
Also published as: GB201814451D0

Abstract

The present invention relates generally to methods for identifying agents which bind to, and modulate the activity of proteins. The invention also relates to related kits and complexes.

Description

METHODS

By binding to one or more protein targets, small molecule therapeutic drugs play important roles across diverse biological processes. The interaction between a drug and its target is often poorly understood and generally cannot be visualized in live cells or entire organisms due to the lack of methods to directly measure drug target engagement in a biological setting. To date, monitoring of target protein engagement inside cells has presented difficult technical hurdles, and presently there are few methods suitable for use with large sets of test agents, and none that describe simultaneous evaluation of direct protein binding (“target engagement”) and the downstream effect of said binding (“phenotypic response”).

Phenotypic-based screening with a small molecule library plays an important role in the drug discovery field. Using such screening approaches, compound libraries, without prior knowledge of their underlying cellular targets, are screened for their ability to elicit a phenotypic response (e.g., modulate downstream gene expression). While this approach can be used to identify bioactive agents (e.g., small molecules) that are able to modulate cellular physiology, determining whether these bioactive agents actually bind directly to the relevant target is a major technical challenge. In addition, small molecules promoting some desirable phenotypic responses may pose in vivo liabilities due to off-target interactions.

Thus, there is a need for reliable methods suitable for the high-throughput, simultaneous evaluation of small molecule-target engagement and the downstream effect, for use in drug discovery and optimization.

Against this background, the present inventors have developed a novel method for determining target engagement in a cellular context whilst simultaneously determining the effect of target engagement on the activity of the target. Unlike previous approaches, the inventors’ method allows identification of direct binding of a bioactive agent to the target, as well as the effect of that direct binding on the function of the target. As described in the accompanying Examples and Figures, in particular the inventors have developed a novel, translatable, and multiparametric cellular target engagement technology termed CeTEAM - Cellular Target Engagement by Accumulation of Mutant. With CeTEAM, a protein of interest with reduced stability is present in a cell. Under basal conditions, this protein with reduced stability is rapidly degraded in cells via the ubiquitin- proteasome pathway or other pathways. Using CeTEAM a bioactive agent for the protein of interest will bind and stabilize the protein by reducing its proteolytic degradation. As a result, the protein of interest with reduced stability will accumulate in cells upon agent binding.

With the assumption that engagement of the protein of interest with reduced stability can be extrapolated to indicate that the wild-type (WT) protein also binds the compound in question, CeTEAM makes it possible to quantify cellular target engagement by any conventionally-used protein detection method. This can be done with single-cell resolution and be combined with other phenotypic readouts to permits multiparametric analyses. CeTEAM represents a truly adaptable technology that can be utilized from early screening applications, to lead discovery and optimization, to preclinical in vivo characterization for inhibitor development programs.

Accordingly, in a first aspect, the invention provides a method for identifying an agent that binds to, and modulates the one or more activity of, a target protein, comprising the steps of:

i. providing an agent to be tested;

ii. providing one or more cell, each cell comprising a target protein and a variant of the target protein which has reduced stability;

iii. contacting the one or more cell with the agent to be tested;

iv. determining the stability of the variant; and

v. determining one or more activity of the target protein.

Thus, the invention can be used to screen a molecule library for molecules which are capable of directly binding to a target protein and modulating its activity and could therefore be useful in drug screening and/or drug development. The invention overcomes many of the problems associated with the prior art, in which the general effect of an agent on a cell can be determined, but it is not clear whether the agent actually binds to and affects the target protein of interest or whether it achieves its effect by binding to other proteins in the cell. By containing both an unstable variant of the target protein, and a functional version of the target protein, in the same cell, the present invention makes it possible to determine both the direct binding of the agent to the target, and its effect on the target simultaneously. Thus, the unstable variant of the target protein acts as a surrogate for binding of an agent to the target protein.

As used herein, the term "agent" refers generally to any synthetic or natural molecule or compound. In some embodiments, the agent is a bioactive agent; by“bioactive” we include the meaning that the agent has an effect on a living organism, tissue, cell and/or protein. Preferably the agent is cell-permeable. The term“agent” and“ligand” may be used interchangeably herein. The agent may bind to the target protein with any affinity, for example with high or low affinity. The agent may compete for binding to the target protein with a physiological protein. Agents for use in the invention are not limited by size or structure.

The term“agent to be tested” as used herein refers to any agent as described above, which is tested in the methods of the invention.

In an embodiment, the agent is one that selectively binds to the target protein. By an agent that“selectively binds” to the target protein, we include the meaning that the agent binds the target protein with a greater affinity than it binds to an unrelated protein. Preferably, the agent binds the target protein with at least 5, or at least 10 or at least 50 times greater affinity than to the unrelated protein. More preferably, the agent binds the target protein with at least 100, or at least 1 ,000, or at least 10,000 times greater affinity than to the unrelated protein. Such binding may be determined or confirmed by methods well known in the art, such as one of the Biacore^® systems, differential scanning fluorimetry (DSF), a cellular thermal shift assay (CETSA), isothermal titration calorimetry and Drug Affinity Responsive Target Stability (DARTS). Binding of the agent to the target protein is also termed“target engagement”.

By“one or more activity of the target protein”, we include any downstream activity or function of the target protein that may be determined using known techniques. Examples of activities are discussed in more detail below.

By“modulates the one or more activity of the target protein, we include the meaning of an agent that reduces the activity of the target protein, as compared to the activity of the target protein in the absence of the agent. Thus, the agent may be an inhibitor or an antagonist of the target protein. Preferably, the agent is one that reduces the one or more activity of the target protein by at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 95%, 99% or 100%. The agent may reduce the activity to an undetectable level, or eliminate the one or more activity.

Additionally, the term “modulates the one or more activity of the target protein also includes the meaning of an agent that enhances the activity of the target protein, as compared to the activity of the target protein in the absence of the agent. Thus, the agent may be an activator, or agonist, of the target protein. Preferably, the agent is one that enhances the one or more activity of the target protein by at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 95%, 99%, 100%, 200%, 300%, 400%, or 500% or more.

By“protein” we include an amino-acid based polymer (i.e. two or more amino acids joined to each other by peptide bonds or modified peptide bonds, such as a peptide or polypeptide. The terms "protein", "peptide" and "polypeptide" may be used interchangeably herein. Polypeptides may contain amino acids other than the 20 natural amino acids, and/or amino acid sequences modified either by natural processes (such as post-translational processing) or by chemical modification, as is known in the art.

The term“target protein” as used herein, refers to a protein which is being assessed in the method of the invention for agent binding and one or more activity. The target protein can therefore be any protein that is present in a cell, and may be in wildtype (WT) form (i.e. as it usually occurs in nature) or may comprise one or more mutations. Typically, the target protein is soluble and not present in inclusion bodies or aggregated.

By“variant of the target protein” we include a form or version of the target protein which differs from the target protein by the presence of one or more mutation. In an embodiment, the target protein and the variant thereof originate from the same species. For example, if the target protein is a human protein, the variant thereof is a variant of the same human protein.

In an embodiment, the variant has at least 60% or more sequence identity to the target protein, such as 70% or more, 80% or more, 90% or more, 91 % or more, 92% or more, 93% or more, 94% or more, 95% or more, 96% or more, 97% or more, 98% or more, or 99% or more. Preferably the sequence and/or function of the agent binding site is conserved between the target and variant of the target.

To determine the percent identity of two amino acid sequences or of two nucleic acid sequences, the sequences may be aligned using methods known in the art. The percent sequence identity between two amino acid sequences or of two nucleic acid sequences may be determined using any suitable computer program, for example the GAP program of the University of Wisconsin Genetic Computing Group and it will be appreciated that percent identity is calculated in relation to polypeptides whose sequence has been aligned optimally. The alignment may alternatively be carried out using the Clustal W program Thompson et al., (1994) Nucleic Acids Res 22, 4673-80), or using EMBOSS Needle (EMBOSS: The European Molecular Biology Open Software Suite (2000) Rice.P. LongdenJ. and Bleasby.A. Trends in Genetics 16, (6) pp276— 277).

Importantly, the variant of the target protein has reduced stability. As discussed further below, that is a key aspect of the invention which enables the method to determine binding of the agent to the target protein. It is well established that the binding of a ligand to a protein can affect the stability of the protein. Generally, when a protein-ligand complex forms it will create a more energetically stable complex that is resistant to thermal and/or proteolytic stress. This is commonly demonstrated with techniques such as CETSA, differential scanning fluorimetry (DSF) and DARTS (B. Lomerick et al. Target identification using drug affinity responsive target stability (DARTS). PNAS, 2009, 106 (51 ) 21984- 21989; Niessen F. H. et al. The use of differential scanning fluorimetry to detect ligand interactions that promote protein stability. Nat. Protoc. 2007, 2 (9), 2212-2221 , and Jafari, R. et al. The cellular thermal shift assay for evaluating drug target interactions in cells. Science, 2013, 9 (9), 2100-2122).

By“variant of the target protein which has reduced stability” we include a variant of the target protein that has reduced stability relative to the target protein, for example when present in a cell. In a preferred embodiment, the variant has reduced stability under physiological conditions. Preferably, the variant is degraded in cellular environments under physiological conditions to a greater extent than the wild type target protein is degraded. In an embodiment, the variant has a reduced ability to form interactions with a chaperone or other protein that would otherwise stabilise the variant.

Methods of assessing protein stability are well known in the art and could be used by those skilled in the art to determine whether a variant had reduced stability relative to the target protein. For example, the variant may exhibit a decrease in thermal stability relative to the target protein as assessed by DSF. Other methods to assess the stability include using Differential Scanning Light Scattering (DSLS) (Senisterra G et al., Assay and Drug Development Technologies, April 2012), nanoDSF, DARTS, CETSA, alpha-screen (Eglen RM et al., Current Chemical Genomics 2008, 1 , 2-10), or similar. For determining the level of the variant, flow cytometry, immunofluorescence, in-cell Western blot may be used. In a further example, levels of the variant could be compared by Western blot to the levels of the wild type protein (as shown in Example 1 ).

The concept of protein stability is well understood in the art. For example, by“stability” we include the ability of a protein to retain its tertiary structure. This can be measured by methods known in the art - for example, changes in fluorescence spectra can be a sensitive indicator of unfolding, either by use of intrinsic tryptophan fluorescence or the use of extrinsic fluorescent probes such as 1-anilino-8-napthaleneulfonate (ANS), for example as implemented in the Thermofluor(TM) method (Mezzasalma et al, J Biomol Screening, 2007, Apr; 12(3) :418-428). Alternatively, proteolytic stability, deuterium/hydrogen exchange measured by mass spectrometry, blue native gels, capillary zone electrophoresis, circular dichroism (CD) spectra and/or light scattering may be used to measure unfolding by loss of signals associated with secondary or tertiary structure.

Protein stability may also be understood to include the ability of a protein to retain its structural conformation and/or its activity when subjected to one or more denaturing conditions, which may be physical and/or chemical manipulations or conditions. Examples of denaturing conditions include exposure to heat, salt, extreme pH, detergents, organic solvents, chaotropic agents.

Thus, the present invention takes advantage of the concept of ligand-induced stabilisation or destabilisation of a protein. As described in more detail below, binding of the agent to the variant either increases or decreases the stability of the variant, and the increased or decreased stability of the variant is indicative of direct binding of the agent to the variant.

In a preferred embodiment, the method further comprises the step of

- determining whether the test agent is one that binds to, and modulates the one or more activity of, the target protein, on the basis of the determinations in steps (iv) and (v).

It will be appreciated that once the stability of the variant has been determined in step (iv) and the one or more activity of the target protein has been determined in step (v), it is possible to determine whether the agent directly binds to the target protein and whether it modulates one or more activity of the target protein. Preferably, the agent is identified as one that binds to, and modulates the one or more activity of, the target protein if:

a) the stability of the variant is increased; and

b) one or more activity of the target protein is modulated.

By“stability of the variant is increased”, we include the meaning that the stability of the variant is increased for example, by 10% or more, 20% or more, 30% or more, 40% or more, 50% or more, 60% or more, 70% or more, 80% or more, 90% or more, 95% or more, 99% or more, or 100% or more, 200% or more, 300% or more, 400% or more, or 500% or more.

Preferably, the agent is identified as an inhibitor of the target protein if:

a) the stability of the variant is increased; and

b) one or more activity of the target protein is decreased and/or inhibited.

Preferably, the agent is identified as an activator of the target protein if:

a) the stability of the variant is increased; and

b) one or more activity of the target protein is increased.

By“one or more activity of the target protein is decreased and/or inhibited”, we include the meaning that the activity of the target protein is decreased and/or inhibited for example, by 10% or more, 20% or more, 30% or more, 40% or more, 50% or more, 60% or more, 70% or more, 80% or more, 90% or more, 95% or more, 99% or more or 100%.

By“one or more activity of the target protein is increased”, we include the meaning that the activity of the target protein is increased for example, by 10% or more, 20% or more, 30% or more, 40% or more, 50% or more, 60% or more, 70% or more, 80% or more, 90% or more, 95% or more, 99% or more, or 100% or more, 200% or more, 300% or more, 400% or more, or 500% or more.

In an embodiment, the method further comprises:

- determining the effect of the agent on one or more property of the one or more cell.

It will be appreciated that in addition to binding to the target protein and modulating its one or more activity, the agent may have other“off-target” effects in the one or more cell. It will be appreciated that this would include any phenotype not associated with the target protein or inhibiting or promoting its activity in cells. For example, the agent may cause, but is not limited to, any of the following responses in the one or more cell DNA damage, cell death (apoptosis, necrosis, etc.), activation or suppression of signalling pathways, cell cycle (DNA content or other), alteration of post-translational modifications (for example phosphorylation, ubiquitination, glycosylation and/or ribosylation.

In a preferred embodiment, the method of the invention is performed in vivo. In other words, the method is performed in a living organism, such as a cell.

The method of the invention may typically comprise a control step, for example, in step (i) rather than providing an agent to be tested a“vehicle control” which contains the solvent or buffer which the agent is dissolved in is used. It will be appreciated that an additional control may involve the use of a proteasome or autophagy inhibitor, or a combination thereof, that will cause an accumulation of the variant protein. This control verifies that the mutation causes a destabilised variant.

As discussed above, increased stability of the variant is indicative of binding of the agent to the variant. Preferably, increased stability of the variant is indicative of binding of the agent to the variant.

Preferably, binding of the agent to the variant increases the stability of the variant. Preferably, binding of the agent increases the stability of the variant.

In an embodiment in which the variant exists in a conformation that cannot form interactions with a chaperone or other protein that would otherwise help stabilise the variant, agent binding to the variant may restore the conformation and interaction with the chaperone or other protein resulting in increased variant stability.

In an embodiment, an agent which binds to the variant protein with high affinity may result in a more thermally stable variant protein compared to an agent which binds to the variant proteins with a lower affinity. Typically, an agent capable of binding to a variant protein may result in the thermal stabilisation of that variant by at least 0.25 or 0.5°C. and preferably at least 1 , 1.5 or 2°C, such as 3°C, 4°C, 5°C, 6°C, 7°C, 8°C, 9°C, or 10°C or more. The term "thermal stability" refers to a quality of a macromolecule (such as a protein) to resist irreversible change in its chemical or physical structure at a high relative temperature. As described in the accompanying Examples, the inventors demonstrated that certain inhibitors stabilized NUDT15 by approximately 10°C (melting temperature shift from 56°C to 64-66°C) (Figure 1c). NUDT15 R139C had a much lower melting temperature compared to the wild type protein; however, it could be stabilized with these inhibitors, again providing a 10°C melting temperature shift (from 46°C to 54-68°C; Figure 1d).

Preferably, binding of the agent to the variant is reversible or irreversible.

By“reversible binding” we include the meaning that the agent binds to the protein of interest with non-covalent interactions such as hydrogen bonds, hydrophobic interactions and/or ionic bonds. An agent that binds reversibly generally can be removed by dilution or dialysis. In an embodiment, binding of the agent to the variant, and stabilisation of the variant, is reversible.

By“irreversible binding” we include that meaning that the agent covalently modifies the protein of interest, and the action of the agent (such as inhibition or activation of the protein of interest) can therefore not be reversed. In an embodiment, binding of the agent to the variant, and stabilisation of the variant, is irreversible.

As described in Example 1 , the accumulation of NUDT15 R139C in the presence of a small molecule inhibitor was reversible, as NUDT15 R139C protein levels returned to basal levels after approximately 72-hours following a 24-hour incubation with inhibitor and drug washout (Figure 1f).

Preferably, binding of the agent to the variant reduces and/or prevents degradation of the variant.

By“reduces and/or prevents degradation of the variant” we include the meaning that the degradation of the variant is decreased following binding of the agent. The degradation of the variant may be decreased by 10% or more, 20% or more, 30% or more, 40% or more, 50% or more, 60% or more, 70% or more, 80% or more, 90% or more, 95% or more, 99% or more, or by 100% (in which case the degradation of the variant may be completely prevented following agent binding).

As demonstrated in Example 1 , expression of NUDT15 R139C in the presence of a control gave little-to-no protein expression by Western blot. However, incubation with an inhibitor caused accumulation of the protein as early as three hours after addition and even higher at 24 hours (~8-fold increase in protein compared to without the inhibitor), while the wild type protein was expressed normally and the level of protein was unaffected by treatment with inhibitor. A longer time-course with the R139C mutant indicated a plateau in accumulation of around 72-hours (=10-fold higher than control; Figure 1f). Thus, cellular NUDT15 R139C accumulated in the presence of a NUDT15 inhibitor.

Preferably the degradation is proteasomal degradation or lysosomal degradation.

It will be appreciated that, in an embodiment, in the absence of an agent which can bind to and stabilise the variant, the variant protein is unstable and aggregates, and/or is targeted to inclusion bodies, and/or is degraded by the cell, for example by proteasomal degradation and/or lysosomal degradation.

By“proteasomal degradation" we include degradation by the ubiquitin (Ub)-proteasome pathway (UPS), in which proteins are marked for degradation by the attachment of ubiquitin or ubiquitin-like proteins. Additional ubiquitins are then added to form a multiubiquitin chain and these polyubiquinated proteins are recognized and degraded by a large, multi-subunit protease complex, called the proteasome (Cooper GM. The Cell: A Molecular Approach. 2nd edition. Sunderland (MA): Sinauer Associates; 2000. Protein Degradation. Available from: https://www.ncbi.nlm.nih.gov/books/NBK9957/).

By “lysosomal degradation” we include the degradation of proteins by lysosomes. Lysosomes are membrane-enclosed organelles that contain an array of digestive enzymes, including several proteases. To be degraded by lysosomal proteolysis, cellular proteins must first be taken up by lysosomes. One pathway for this uptake of cellular proteins, autophagy, involves the formation of autophagosomes in which small areas of cytoplasm or cytoplasmic organelles are enclosed in membranes derived from the endoplasmic reticulum. These vesicles then fuse with lysosomes, and the degradative lysosomal enzymes digest their contents (Cooper GM. The Cell: A Molecular Approach. 2nd edition. Sunderland (MA): Sinauer Associates; 2000. Protein Degradation. Available from: https://www.ncbi.nlm.nih.gov/books/NBK9957/).

It will be appreciated that, as proof of concept, poor protein expression of the variant may be rescued by proteasome or lysosome inhibitors. As demonstrated in Example 1 , the addition of proteasome inhibitor MG-132 led to accumulation of the unstable variant protein, NUDT15 R139C. Preferably, the variant is unstable relative to the target protein under physiological conditions. Such conditions may include physiological temperatures and cell culture conditions.

Preferably, the target protein is a functional protein. It will be appreciated that in order to determine the one or more activity of the target protein, the function of that target protein must be capable of being assessed.

The variant may be non-functional or functional. It will be appreciated that the integrity of the agent-binding site is preserved between the variant and the target protein. In an embodiment, the variant is non-functional, but the agent-binding site is maintained.

As discussed above, the variant comprises one or more mutation in its polypeptide sequence which results in reduced stability of the variant. The variant may contain one or more mutation, such as at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 1 1 , 12, 13, 14 ,15 16, 17, 18, 19, or 20 or more mutations relative to the target protein. It will be appreciated that a mutation of an amino acid in the sequence of the target protein results in a variant form. It will be appreciated that the integrity of the agent binding site is maintained in the variant and is unaffected by any mutations. Such mutations may also be termed “destabilising mutations” herein. Such mutations may lead to increased ubiquitination, decreased thermal stability and mislocalisation.

Preferably, the one or more mutation comprises a substitution, deletion and/or an addition to the polypeptide sequence.

It will be appreciated that several point mutations may be introduced so long as the agent binding site is conserved between the target and variant of the target. Insertion mutations could be introduced that would result in a frame shift that may destabilize the protein. Alternatively, mutations that disrupt introns may be introduced so that the introns are incorrectly excised which may result in an unstable protein.

As can be seen in the accompanying Examples, the variant of the target protein NUDT15 comprises one amino acid substitution (R139C). The variant of the target protein MTH1 comprises the G48E substitution; and the variant of the target protein PARP1 comprises one amino acid substitution (L713F). All of these mutations render the respective variant proteins unstable relative to the unmutated target protein. Those skilled in the art will be capable of generating and isolating variant forms of a target protein in which stability is reduced, using methods known in the art.

Preferably, the mutation is generated by performing any one of the group comprising: in silico characterization and mutagenesis (for example site-directed mutagenesis or random mutagenesis). It will be appreciated that if the structure of a protein of interest is known (for example a crystal structure or NMR structure), in silico analyses can be performed to assess the effects of amino acid substitutions on protein stability (from changes to thermodynamic potential, DQ). Alternatively, to screen for destabilizing mutations in protein(s) of interest if structural information is not available, random mutagenesis can be performed within the open reading frame (ORF) of the gene of interest. This can be performed with screening of thermounstable mutations with bacterial cultures, similarly as described by Asial et al., Nature Comms, 2013. A similar methodology, which may be used to screen for CeTEAM candidates, was developed by Banaszynski et al., Ceil, 2006. Here, random mutagenesis was performed within the ligand binding domain (LBD) before fusing to eGFP in a mammalian expression vector. The authors then screened for clones that fluoresced upon ligand exposure, then lost fluorescence following ligand removal and once again tested for increased fluorescence in the presence of ligand by FACS. Clones that met these criteria were then sequenced to identify destabilizing mutations that could be stabilized by binding of a ligand. These LBDs were then fused to other genes of interest and used to tightly regulate expression of the protein of interest dependent on the ligand being added to the cells.

In one embodiment, the variant comprises a destabilising domain. By“destabilising domain” we include a protein sequence which is capable of conferring instability on to another protein to which it is attached.

The term“destabilising domain” may also be known as a“degron”,“degradation motif and “degradation signal”, and these terms are used interchangeably herein.

By“destabilising domain” we include a protein sequence that is sufficient for recognition and degradation by a proteolytic apparatus. The activity of a destabilising domain may be transferable, in other words, the fusion of such sequences confers instability (i.e. a shorter half-life) on otherwise longer-lived proteins.

In some embodiments, the destabilising domain is capable of being ubiquitylated, including polyubiquitylated. In some embodiments, sequences or structures within the destabilising domain bind directly to the ubiquitin-ligase complex or its associated factors. In some embodiments, sequences or structures within the destabilising domain initiate unfolding and insertion into the proteasome. Destabilising domains can be conditional and activated or inhibited by temperature, small molecules, light, or the expression of another protein, all of which are known in the art.

Destabilising domains can consist of protein sequences that act as targets for ubiquitylation^{25 27}. Destabilising domains can regulate intrinsic protein turnover and their fusion to small molecules has created a new class of promising therapeutics, known as (PROTACs²⁸·²⁹). Cryptic destabilising domains may be exposed upon heating or partial unfolding of the particular domain³⁰, thus permitting ubiquitylation of key lysine residues³¹; therefore, it is likely that many destabilising missense mutations cause rapid proteolytic degradation in a similar fashion³². As such, fusing destabilising domains sequences to the termini of a stable target protein results in conditional depletion in cells^{25 27 30}·³¹. Like traditional unstable ligand binding domains (LBDs), degron fusion proteins (e.g., FKBP12- L106P:POI¹⁹, C-b-gakPOI²⁵·³³, Arg-DHFR:POI³⁰·³³ or R-ARD1 :POI22, among others¹⁶·³¹), can be stabilized in the presence of a ligand bound to the LBD¹⁹·²¹·³⁴ or protein itself¹⁶, which has also proven useful in the study of protein functions³⁵ and increasing tunability of LBD-based biosensors¹⁶.

Destabilising domains are known in the art and are described in Wandless et al., (2006) Cell 126, 995-1004. Wandless et al describe the fusion of a protein of interest to a ligand binding domain that is engineered to be unstable, and thus degraded, in the absence of its ligand. Binding of the ligand to the destabilizing domain stabilizes the fusion protein and shields it from degradation, thus restoring function to the protein of interest.

Examples of destabilising domains include, but are not limited to, those in Table 1 :

Table 1 : Examples of destabilising domains

Examples of destabilising motifs include, but are not limited to, those in Table 2:

Table 2: Examples of destabilising motifs

Further examples of destabilising domains for use in the present invention are described in Tables 3-6 and the accompanying Table legends. In a preferred embodiment, the variant comprises one or more destabilising domain selected from Tables 3-6. It will be appreciated that attachment of a destabilising domain to a target protein can confer instability and therefore produce a variant of the target protein which has reduced stability. The degradation domain can be genetically engineered to reside within the target protein, and/or at the N- and/or C-terminus of the target protein. Preferably the agent does not bind to the destabilising domain. It will be appreciated that in order to identify an agent which directly binds to the target protein, it must bind to the variant itself which subsequently results in its stabilisation. Where a destabilising domain is present, the destabilising domain targets the variant for degradation by the proteasome - without wishing to be bound by theory, the inventors believe that when the variant interacts with the proteasomal machinery, it must be unwound from its three-dimensional (3D) structure in order to be processed by the proteasome; binding of an agent stabilises the 3D structure of the variant making it more resistant to degradation. Thus, agent binding may prevent and/or reduce destabilising domain-dependent proteasomal targeting. Preferably, binding of the agent to the variant is not affected by the mutation and/or the destabilising domain. It will be appreciated that this allows for the identification of agents that modulate one or more activity of the target protein by binding directly to the target protein, and avoids any influence of the agent binding elsewhere, such as to the destabilising domain. As it will be appreciated, there may be subtle differences in agent binding affinity between the variant and target protein.

As discussed above, it is preferred that the target protein and the variant are identical apart from the introduction of one or more destabilising mutations and/or the presence of one or more destabilising domain.

Preferably, the one or more cell is selected from the group comprising: a mammalian cell, a non-mammalian cell, a primary cell, a cell line, a cell within a model organism, and/or a cell within a xenograft.

Examples of a mammalian cells include, but are not limited to, mouse, rat, hamster, rabbit, cow, monkey, dog or other species).

Examples of non-mammalian cells include, but are not limited to, plant, yeast, fungus and bacteria ceils. Specific examples include but are not limited to BY-2 cells (tobacco plant) Schneider 2 (S2) cells (drosophila), Sf9, Sf21 (army worm), High five (cabbage looper)

A6 (Xenopus kidney epithelial), AB9 (zebrafish), INVSd (S. cerevisiae).

The one or more cell may be a cell line or a primary cell which is taken, for example, directly from a human or an animal. In an embodiment the cell is from a patient. It will be appreciated that it may be possible to perform the method of the invention in one or more cell taken from a patient and compare with the results from one or more cell taken from a healthy subject. Alternatively, it may be possible to perform the method of the invention in one or more healthy cell and compare with the results from one or more malignant or diseased cell from the same subject.

Where the one or more cell originates from a cell line, the cell line may be“finite” or “continuous”. A“finite cell line” contains normal cells which usually divide only a limited number of times before losing their ability to proliferate, which is a genetically determined event known as senescence. By“continuous cell line” we include a cell line that has become immortal through a process of transformation, which can occur spontaneously or can be chemically or virally induced. When a finite cell line undergoes transformation and acquires the ability to divide indefinitely, it becomes a continuous cell line. Examples of cell lines that could be used in the methods of the invention include, but are not limited to 2780AD, 293, 3T6, A549, A9, AtT-20, BALB/3T3, BHK-21 , BHL-100, BT, Caco-2, Chang, CHO-K1 , Clone 9, Clone M-3, COS-1 , COS-3, COS-7 , CRFK, CV-1 , D-17, Daudi, GH1 , GH3, H9, HaK, HCT-116, HCT1 16 3-6, HCT-15, HeLa, HEp-2, HL-60, HT-1080, HT-29, HUVEC, 1-10, IM-9, JEG-2, Jensen, Jurkat, K-562, KB, KG-1 , L2, LLC-WRC 256, McCoy, MCF7, MOLT-4, Raji, U-2 OS, U-937, WI-38, WISH, XC, Y-1.

By“model organism" (or“animal model”), we include an organism as a model of human anatomy and physiology. Examples of model organism include but are not limited to non human primates such as a macaque or a marmoset, mouse, rat or other rodent, rabbit, monkey, dog, zebrafish, xenopus, medaka, fruit fly (Drosophila), C. Elegans.

By“xenograft” we include a graft or tissue taken from a donor of one species and grafted into a recipient of another species. Approaches for doing so are well known to those skilled in the art.

In a preferred embodiment, the one or more cell is an intact cell. By“intact” we include the meaning of a cell that is not damaged or impaired in any way.

Preferably, step (iii) comprises conditions permitting binding of the agent to the target protein and to the variant of the target protein.

It will be appreciated that the one or more cell is subjected to physical conditions that will allow the agent to bind to the variant of the target protein, for example, physiologic temperature, physiological pH and/or physiological salt concentrations. Such conditions are known in the art.

Preferably, step (iv) comprises quantitative and/or qualitative analysis of the amount and/or concentration of the variant.

Methods for quantitating proteins in a given sample for total protein content, and for single protein content are well known in the art and include ELISA, Western blot analysis, immunofluorescence, flow cytometry, luminescence or other reporter, HPLC, absorbance (for total protein), immunoprecipitation and mass spectrometry. Preferably, step (iv) comprises determining accumulation of the variant.

It will be appreciated that since binding of the agent to the variant modulates degradation of the variant, measuring accumulation of the variant can be indicative of agent binding. Methods for determining protein accumulation are well known in the art and include any conventional protein detection technique such as an ELISA, mass spectrometry and western blot analysis, as demonstrated in Example 1.

As demonstrated in Example 1 , expression of an unstable variant, NUDT15 R139C, in the presence of a control gave little-to-no protein expression by Western blot. However, incubation with an inhibitor caused accumulation of the protein as early as 3 hours after addition and even higher at 24 hours (=8-fold increase in protein compared to without the inhibitor), while the wild type protein was expressed normally and was unaffected by treatment with inhibitor. A longer time-course with the R139C mutant indicated a plateau in accumulation of around 72 hours («10-fold higher than control; Figure 1f). Thus, cellular NUDT15 R139C accumulated in the presence of a NUDT15 inhibitor.

As will be appreciated, detecting protein accumulation can be performed at single-cell resolution and at high throughput, dependent on the instrumentation used by the skilled person (as demonstrated in Example 1 ). As described in more detail below, using cells expressing NUDT15 R139C, the inventors set up a 96-well imaging plate and exposed the cells to serial dilutions of an inhibitor for 72 hours and quantified by high-throughput immunofluorescence microscopy (Figure 2a). This resulted in a dose-dependent increase in levels of NUDT15 R139C, indicating that this method is sensitive enough to detect a wide range of inhibitor concentrations at single cell resolution (Figure 2b and c).

Preferably, step (v) comprises determining one or more activity of the target protein by measuring one or more of: modification of the target protein, including but not limited to post-translational modification; modification of a substrate of the target protein; expression of the target protein; expression of a substrate of the target protein; localisation of the target protein; localisation of a substrate of the target protein; expression of one or more genes downstream of the target protein; expression of one or more genes downstream of a substrate of the target protein; repression of one or more gene downstream of the target protein; repression of one or more gene downstream of a substrate of the target protein; morphology of the one or more cell (for example due to cell cycle arrest); the interaction of the target protein with one or more known or unknown interaction partners; modulation of target protein mRNA or protein levels; genomic and/or epigenetic regulation; sensitisation or resistance of the target protein or cell to a further agent; other phenotypic markers.

It will be appreciated that any activity or function of the target protein may be determined in the methods of the invention. The selection of the appropriate assays to be used to determine the one or more activity of the target protein will be determined by the function of target protein, and will be apparent to those skilled in the art.

As shown in the Examples, as proof of principle, given that NUDT15 hydrolyzes 6- thioguanine triphosphates in cells, the depletion of which potentiates thioguanine toxicity, NUDT15 inhibitors were combined with a low-dose of 6-thioguanine (200 nM) 3 hours after adding an inhibitor, in order to measure phenotypic readouts (Figure 2a). The incorporation of 6-thioguanine into genomic DNA results in a prolonged G2 cell cycle delay caused by futile mismatch repair cycling and ATR-Chk1 activation. Accordingly, the inventors measured DNA damage, through the use of yH2A.X, a ubiquitous marker of DNA damage, and by staining nuclear DNA with Hoechst 33342 to determine the cell cycle (Figure 2d, e, and f). Titration of inhibitor alone increased variant NUDT15 levels in a dose-dependent manner, indicating agent binding (target engagement), but there were no differences in yH2A.X intensity or DNA content compared to the control. However, in the presence of 200 nM thioguanine, there was the same dose-dependent increase in NUDT15 variant levels but there was also an increase in gH2A.C intensity and G2/M cell cycle content. Visualization of all three parameters simultaneously emphasizes the correlation of NUDT15 target engagement with thiopurine potentiation (Figure 2g).

Accordingly, the method of the invention allows multiplexing with phenotypic markers and the direct correlation of target engagement with phenotypic response.

Preferably, step (v) is performed using one or more method selected from the group comprising: fluorescence microscopy, flow cytometry, fluorescence polarization, fluorescence spectroscopy, luminescence spectroscopy, automated microscopy, automated image analysis, imaging of a whole animal or organism, Western blot; and PCR.

Preferably step (v) is performed using one or more method selected from the group comprising: transient transfection of a vector construct, stable transfection of a vector construct, fluorescence resonance energy transfer, bio-luminescence resonance energy transfer, immunofluorescence, immunohistochemistry, protein-fragment complementation assays, enzyme-fragment complementation assays, expression of a chimeric protein, tagging of an expressed protein or peptide with a fluorescent protein, epitope tagging, labelling of a reagent or cellular state with a quantum dot, production of an optically detectable reaction product, binding of an optically detectable probe, subcellular localization of an optically detectable signal or probe, immunofluorescence, flow cytometry, luminescence or other reporter, HPLC, absorbance (for total protein), and immunoprecipitation.

It will be appreciated that the method step of determining the effect of the agent on one or more property of the one or more cell may also be performed using any of the above methods.

It will be appreciated that step (v) may be performed using any technique available to the skilled person which allows the activity of the target protein to be determined and/or measured. For example, quantitative analysis of protein and/or genomic material levels utilizing various experimental means encompassing for example, antibody-based technologies, may be used.

Preferably, the variant is capable of detection. By“capable of detection” we include that the variant can be detected for example by: antibody-based techniques, fluorescence, luminescence, catalysis, signalling, gene transcription and/or protein expression.

Detection may be based on affinity binding between the variant protein and a detection agent, for example an antibody, antibody fragment or affibody. Preferably, in an embodiment wherein the variant is endogenous to the one or more cell, the variant protein may be detected using antibodies, monoclonal or polyclonal, directed directly to the endogenous variant protein.

Preferably, the variant further comprises one or more detectable moiety.

By“detectable moiety” we include any molecule that can be attached to the variant to render the variant detectable by an instrument or method.

For example, the variant may be detectably labelled so as to facilitate detection of the variant and consequently the effect of the agent on the stability of the variant. Examples of suitable labels include a peptide label, a nucleic acid label (Kerr et al (1993) JACS vol. 1 15, p. 2529-2531 ; and Brenner & Lerner (1992) Proc. Natl. Acad. Sci. USA vol. 89, p. 5381-5383), a chemical label (Ohlmeyer et al (1993) Proc. Natl. Acad. Sci. USA vol. 90, p. 109222-10926; and Maclean et al (1997) Proc. Natl. Acad. Sci. USA vol. 94, p. 2805- 2810); a fluorescent label (Yamashita & Weinstock (SmithKIine Beecham Corporation), W095/32425 (1995); and Sebestyen et al (1993) Pept. Proc. Eur. Pept. Symp. 22nd 1992, p. 63-64), or a radio frequency tag (Nicolaou et al (1995) Angew. Chem. Int. Ed. Engl. vol. 34, p. 2289-2291 ; and Moran et al (1995) JACS vol. 117, p. 10787-10788).

Preferably, the detectable moiety comprises a fluorescent molecule, a chemiluminescent molecule, a bioluminescent molecule, a radioactive molecule, an epitope tag, a polymerase, a transcription factor, an enzyme, a signalling protein, and/or a functional protein.

“Fluorescent molecules” are known in the art and examples include fluorescein and its derivatives, fluorochrome, rhodamine and its derivatives, Green Fluorescent Protein (GFP), dansyl, umbelliferone etc. In such conjugates, the variant of the method of the invention can be prepared by methods known to the person skilled in the art.

“Chemiluminescent molecules” such as luminol and the dioxetanes, or a bioluminescent label such as luciferase and luciferin may be used in the method of the invention.

Suitable“radioactive molecules” include technetium-99m or iodine-123 for scintigraphic studies. Others may be selected from the group consisting of: iodine-124; iodine-125; iodine-126; iodine-131 ; iodine-133; indium-1 11 ; indium-1 13m, fluorine-18; fluorine-19; carbon-1 1 ; carbon-13; copper-64; nitrogen-13; nitrogen-15; oxygen-15; oxygen-17; arsenic-72; gadolinium; manganese; iron; deuterium; tritium; yttrium-86; zirconium-89; bromine-77, gallium-67; gallium-68, ruthenium-95, ruthenium-97, ruthenium-103, ruthenium-105, mercury-107, rhenium-99m, rhenium-101 , rhenium-105, scandium-47. The radioactive molecule may be incorporated in the variant in known ways. For example, the variant may be biosynthesised or synthesised by chemical amino acid synthesis using suitable amino acid precursors involving, for example, fluorine-19 in place of hydrogen. Labels such as 99mTc, 1231, 186Rh, 188Rh and 1111n can, for example, be attached via cysteine residues in the variant. Yttrium-90 can be attached via a lysine residue. The IODOGEN method (Fraker et al (1978) Biochem. Biophys. Res. Comm. 80, 49-57) can be used to incorporate iodine-123. The reference “Monoclonal Antibodies in Immunoscintigraphy”, (J.F. Chatal, CRC Press, 1989) describes other methods in detail. “Epitope tags”, also known as“affinity tags”, are well known in the art and include, for example, Fc tag, BirA tag, maltose-binding protein tag, GST tag, HA tag, histidine tag, V5 tag, T7 tag, FLAG tag or any short protein sequence to which a specific antibody is available, thioredoxin and maltose binding protein. Tags are preferably between 1-100 amino acids in length, preferably between 1-70, 2-50, 1-30 or 1 -20 amino acids in length. More preferably, tags can be 3, 4, 5, 6, 7, 8, 9 or 10 amino acids in length.

Particular detectable moieties that may be used in the present invention, include but are not limited to: green fluorescent protein (GFP) and derivatives thereof ((E)GFP; yellow fluorescent protein (YFP); Cyan Fluorescent Protein (CFP)), dsRed, Myc tag, E tag, FLAG tag, Glu-Glu tag, GST tag, HA tag, His tag, HSV tag, luciferase and derivatives thereof (Akaluc), MBP, nanoLuciferase, protein C tag, S tag, T7 tag, V5 tag, VSV-g tag, avidin/streptavidin/strep tag, thioredoxin, His-patch thioredoxin, b-galactosidase, chloramphenicol acetyltransferase, cellulose binding domains (CBDs), chitin binding domain, staphylococcal protein A, streptococcal protein G, neo, hyg, pac, zeo, gpt, ble, dhfr, hpt and npt II.

The detectable moiety may comprise a detectable enzyme such as peroxidase, alkaline phosphatase, beta-D-galactosidase, glucose oxidase, glucose amylase, carbonic anhydrase, acetylcholinesterase, lysozyme, malate dehydrogenase or glucose 6- phosphate dehydrogenase.

The variant may be detected by the enzymatic activity of a tag e.g. where the enzymatic activity results in the production of a detectable signal. For example, fusion tags that possess enzymatic activity such as green fluorescent protein, horseradish peroxidase (HRP), luciferase and glutathione-S-transferase.

The detectable moiety may comprise a directly detectable label (such as a fluorophore, a radioactive molecule, a contrast agent, or a luminescent label); or an indirectly detectable label (such as an enzyme, an enzyme substrate, an antibody, an antibody fragment, an antigen, a hapten, a ligand, an affinity molecule, a chromogenic substrate, a protein, a peptide, a nucleic acid, a carbohydrate and a lipid).

In some embodiments, the detectable moiety may be fused to the variant of the target protein. For a detectable moiety to be fused to the variant, it is generally transcribed and translated with the variant protein as a single molecule. Thus, antibodies which bind to the target protein and which may be labelled with HRP etc allow the variant to be detected but are not considered to be fused to the variant protein. Short tags are preferred, to allow proteins of interest to maintain a native-like conformation. Further, C-terminal tags are preferred, although N-terminal His tags are also used. It will be appreciated that a detection step involving the use of a tag fused to a target protein can only be used where the target protein is derived from a recombinant expression system. Therefore, generally this detection method will not be used in an embodiment wherein the variant protein is for example endogenous to the one or more cell.

A further aspect of the invention provides a polynucleotide encoding a variant as defined above, wherein the variant and the detectable moiety are fused. It will be appreciated that this aspect of the invention also provides a vector comprising a polynucleotide encoding a variant as defined above fused to a detectable moiety.

Preferably, the method further comprises the step of:

- determining whether the agent is one that binds to the target protein.

It will be appreciated that it will be necessary to confirm direct binding of the agent to the target protein (i.e. target engagement). This step can be performed using any technique known in the art for determining target engagement, such as protein-protein binding, or drug-protein binding.

Preferably, the step of determining whether the agent is one that binds to the target protein comprises one or more method selected from the group comprising: a cellular thermal shift assay (CETSA), differential scanning fluorimetry (DSF), a protease stability assay, for example Drug Affinity Responsive Target Stability (DARTS), an oxidation rate assay, such as Stability of Proteins from Rates of Oxidation (SPROX), an enzymatic activity assay, a binding assay, for example a Stability of Unpurified Proteins from Rates of H/D Exchange (SUPREX), a radioligand displacement assay or a fluorescence polarization assay.

CETSA involves treatment of cells with a compound of interest, heating to denature and precipitate proteins, cell lysis, and the separation of cell debris and aggregates from the soluble protein fraction. Whereas unbound proteins denature and precipitate at elevated temperatures, ligand-bound proteins remain in solution, the stabilised protein in the soluble fraction can then be detected (Jafari, 2014, The cellular thermal shift assay for evaluating drug target interactions in cells; Molina, 2013, Monitoring Drug Target Engagement in Cells and Tissues Using the Cellular Thermal Shift Assay). Differential scanning fluorimetry (DSF), is a thermal-denaturation assay that measures the thermal stability of a target protein and a subsequent increase in protein melting temperature upon binding of an agent to the protein. The thermal stability change is measured by performing a thermal denaturation curve in the presence of a fluorescent dye (Niesen, 2007, The use of differential scanning fluorimetry to detect ligand interactions that promote protein stability).

Alternatively, a drug affinity responsive target stability (DARTS) assay could be used. DARTS relies on the phenomena that proteins are more stable when bound to a ligand, which makes them less susceptible to proteolysis. Samples are mixed with a small molecule or control to identify protein-small molecule interactions. These samples are then subjected to limited proteolysis and compared by gel electrophoresis and quantitative mass spectrometry. Protein targets are identified as those proteins that display increased protease resistance in the presence of the small molecule (Lomenick, 2009, Target identification using drug affinity responsive target stability (DARTS).

SPROX (stability of proteins from rates of oxidation) assesses protein and protein-ligand stability by using hydrogen peroxide in combination with varying concentrations of chemical denaturant followed by mass spectrometry. The level of oxidation is coupled to the denaturant concentration, and this dependency, in turn, is used to determine the folding free energy of a given protein. The stability of the oxidation reaction permits analysis of proteins or protein-ligand stability in isolation or in more complex protein mixtures (West, 2008, Thermodynamic analysis of protein stability and ligand binding using a chemical modification- and mass spectrometry-based strategy).

SUPREX (stability of unpurified proteins from rates of H/D exchange) measures the stability of proteins in a rapid, high-throughput manner by utilizing hydrogen/deuterium (H/D) exchange followed by matrix-assisted laser desorption/ionization mass spectrometry. E. coli expressing the proteins of interest are pelleted, lysed, treated with deuterated exchange buffer containing a range of guanidinium monochloride concentrations prior to incubation in MALDI matrix solution and subsequent mass spectrometry analysis. More stable proteins require higher guanidinium monochloride concentrations to increase deuterium exchange. It is possible to analyse the stabilities (or effect on stability) of several different proteins, variants of a single protein and proteins bound to ligands with this methodology (Ghaemmaghami, 2000, A quantitative^' , high- throughput screen for protein stability). Preferably, the agent is selected from the group comprising: a small molecule, an antibody, a peptide, a peptidomimetic, a natural product, a carbohydrate, a nucleic acid and an aptamer.

In an embodiment, the agent to be tested may be a small molecule. The term“small molecule” includes small organic molecules. Suitable small molecules may be identified by methods such as screening large libraries of compounds (Beck-Sickinger & Weber (2001 ) Combinational Strategies in Biology and Chemistry (John Wiley & Sons, Chichester, Sussex); by structure-activity relationship by nuclear magnetic resonance (Shuker et al (1996)“Discovering high-affinity ligands for proteins: SAR by NMR. Science 274: 1531-1534); encoded self-assembling chemical libraries Melkko et al (2004) “Encoded self-assembling chemical libraries.” Nature Biotechnol. 22: 568-574); DNA- templated chemistry (Gartner et al (2004)“DNA-templated organic synthesis and selection of a library of macrocycles. Science 305: 1601-1605); dynamic combinatorial chemistry (Ramstrom & Lehn (2002)“Drug discovery by dynamic combinatorial libraries.” Nature Rev. Drug Discov. 1 : 26-36); tethering (Arkin & Wells (2004)“Small-molecule inhibitors of protein-protein interactions: progressing towards the dream. Nature Rev. Drug Discov. 3: 301-317); and speed screen (Muckenschnabel et al (2004)“SpeedScreen: label-free liquid chromatography-mass spectrometry-based high-throughput screening for the discovery of orphan protein ligands.” Anal. Biochem. 324: 241-249). The benefits of most small organic molecule binders include their ease of manufacture, lack of immunogenicity, tissue distribution properties, chemical modification strategies and oral bioavailability. Small molecules with molecular weights of less than 5000 daltons are preferred, for example less than 400, 3000, 2000, or 1000 daltons, or less than 500 daltons.

In an embodiment, the agent to be tested may be an antibody. The term“antibody” or “antibody molecule” as used herein throughout the specification includes but is not limited to polyclonal, monoclonal, chimeric, single chain, Fab fragments and fragments produced by a Fab expression library. Such fragments include fragments of whole antibodies which retain their binding activity for a target substance, Fv, F(ab’) and F(ab')2 fragments, as well as single chain antibodies (scFv), fusion proteins and other synthetic proteins which comprise the antigen-binding site of the antibody. The term also includes antibody-like molecules which may be produced using phage-display techniques or other random selection techniques for molecules which bind to the specified polypeptide or to particular regions of it. Thus, the term antibody includes all molecules which contain a structure, preferably a peptide structure, which is part of the recognition site (i.e. the part of the antibody that binds or combines with the epitope or antigen) of a natural antibody. Furthermore, the antibodies and fragments thereof may be humanised antibodies, which are well known in the art.

By“ScFv molecules” we mean molecules wherein the VH and VL partner domains are linked via a flexible oligopeptide. Engineered antibodies, such as ScFv antibodies, can be made using the techniques and approaches long known in the art. The advantages of using antibody fragments, rather than whole antibodies, are several-fold. The smaller size of the fragments may lead to improved pharmacological properties, such as better penetration to the target site. Effector functions of whole antibodies, such as complement binding, are removed. Fab, Fv, ScFv and dAb antibody fragments can all be expressed in and secreted from E. coli, thus allowing the facile production of large amounts of the fragments. Whole antibodies, and F(ab')2 fragments are“bivalent”. By "bivalent” we mean that the antibodies and F(ab')2 fragments have two antigen combining sites. In contrast, Fab, Fv, ScFv and dAb fragments are usually monovalent, having only one antigen combining site.

It is possible however that the ScFv may be monovalent, divalent, trivalent or tetravalent. The ScFv may be a diabody, tribody, or a tetrabody. The two or more VH and VL partner domains in a divalent, trivalent or tetravalent or diabody, tribody, or a tetrabody may be different. In such a situation, an ScFv agent may comprise more than 2 or more than 3, for example 4 different VH and VL domains.

Antibodies may be produced by techniques known in the art, for example by immunisation with the appropriate (glyco)polypeptide or portion thereof, or by using a phage display library.

In one embodiment, the antibody is a polyclonal antibody.

If polyclonal antibodies are desired, a selected mammal (e.g., mouse, rabbit, goat, horse, etc) is immunised with an immunogenic polypeptide bearing a desired epitope(s), optionally haptenised to another polypeptide. Depending on the host species, various adjuvants may be used to increase immunological response. Such adjuvants include, but are not limited to, Freund’s, mineral gels such as aluminium hydroxide, and surface active substances such as lysolecithin, pluronic polyols, polyanions, peptides, oil emulsions, keyhole limpet hemocyanin, and dinitrophenol. Serum from the immunised animal is collected and treated according to known procedures. If serum containing polyclonal antibodies to the desired epitope contains antibodies to other antigens, the polyclonal antibodies can be purified by immunoaffinity chromatography. Techniques for producing and processing polyclonal antisera are well known in the art.

In one embodiment, the antibody is a monoclonal antibody.

Monoclonal antibodies directed against entire polypeptides or particular epitopes thereof can also be readily produced by one skilled in the art. The general methodology for making monoclonal antibodies by hybridomas is well known in the art. Immortal antibody- producing cell lines can be created by cell fusion, and also by other techniques such as direct transformation of B lymphocytes with oncogenic DNA, or transfection with Epstein- Barr virus. Panels of monoclonal antibodies produced against the polypeptides listed above can be screened for various properties; i.e., for isotype and epitope affinity. Monoclonal antibodies may be prepared using any of the well-known techniques which provides for the production of antibody molecules by continuous cell lines in culture.

It is preferred if the antibody is a monoclonal antibody. In some circumstances, particularly if the antibody is to be administered repeatedly to a human patient, it is preferred if the monoclonal antibody is a human monoclonal antibody or a humanised monoclonal antibody, which are suitable for administration to humans without engendering an immune response by the human against the administered immunoglobulin. Suitably prepared non- human antibodies can be“humanised” in known ways, for example by inserting the CDR regions of mouse antibodies into the framework of human antibodies. Humanised antibodies can be made using the techniques and approaches described in Verhoeyen et al (1988) Science, 239, 1534-1536, and in Kettleborough et al, (1991 ) Protein Engineering, I4(7), 773-783. In some instances, Fv framework residues of the human immunoglobulin are replaced by corresponding non-human residues. In general, the humanised antibody will contain variable domains in which all or most of the CDR regions correspond to those of a non-human immunoglobulin, and framework regions which are substantially or completely those of a human immunoglobulin consensus sequence.

Completely human antibodies may be produced using recombinant technologies. Typically, large libraries comprising billions of different antibodies are used. In contrast to the previous technologies employing chimerisation or humanisation of e.g. murine antibodies this technology does not rely on immunisation of animals to generate the specific antibody. Instead the recombinant libraries comprise a huge number of pre-made antibody variants wherein it is likely that the library will have at least one antibody specific for any antigen. Thus, using such libraries, an existing antibody having the desired binding characteristics can be identified.

It is appreciated that when the antibody is for administration to a non-human individual, the antibody may have been specifically designed/produced for the intended recipient species.

WO 98/32845 and Soderlind et al (2000) Nature BioTechnol. 18: 852-856, describe technology for the generation of variability in antibody libraries. Antibody fragments derived from this library all have the same framework regions and only differ in their CDRs. Since the framework regions are of germline sequence the immunogenicity of antibodies derived from the library, or similar libraries produced using the same technology, are expected to be particularly low (Soderlind et al, 2000). This property is of great value for therapeutic antibodies, reducing the risk that the patient forms antibodies to the administered antibody, thereby reducing risks for allergic reactions, the occurrence of blocking antibodies, and allowing a long plasma half-life of the antibody. Thus, when developing therapeutic antibodies to be used in humans, modern recombinant library technology (Soderlind et al, 2001 , Comb. Chem. & High Throughput Screen. 4: 409-416) is now used in preference to the earlier hybridoma technology.

By antibodies we also include heavy-chain antibodies structurally derived from camelidae antibodies, such as Nanobodies® (Ablynx). These are antibody-derived therapeutic proteins that contain the structural and functional properties of naturally-occurring heavy- chain antibodies. The Nanobody® technology was developed following the discovery that camelidae (camels and llamas) possess fully functional antibodies that lack light chains. These heavy-chain antibodies contain a single variable domain (VHH) and two constant domains (CH2 and CH3). The cloned and isolated VHH domain is a perfectly stable polypeptide harbouring the full antigen-binding capacity of the original heavy-chain antibody. These VHH domains with their unique structural and functional properties form the basis of Nanobodies®. They combine the advantages of conventional antibodies (high target specificity, high target affinity and low inherent toxicity) with important features of small molecule drugs (the ability to inhibit enzymes and access receptor clefts). Furthermore, they are stable, have the potential to be administered by means other than injection, are easier to manufacture, and can be humanised. (See, for example US 5,840,526; US 5,874,541 ; US 6,005,079, US 6.765,087; EP 1 589 107; WO 97/34103; WO97/49805; US 5,800,988; US 5,874, 541 and US 6,015,695). In another embodiment, the agent to be tested may be a peptide. Suitable peptides may be identified by methods such as phage display of peptide libraries (Scott & Smith (1990) “Searching for peptide ligands with an epitope library.” Science 249: 386-390; Felici et al (1995) “Peptide and protein display on the surface of filamentous bacteriophage.” Biotechnol. Annu. Rev. 1 : 149-183); and Collins et al (2001 )“Cosmix-plexing: a novel recombinatorial approach for evolutionary selection from combinatorial libraries.” J. Biotechnol. 74: 317-338); including in vivo panning (Pasqualini et al (1997)“av inte.g.rins as receptors for tumor targeting by circulating ligands. Nature Biotechnol. 15: 542-546), and solid-phase parallel synthesis (Frank (2002) “The SPOT-synthesis technique. Synthetic peptide arrays on membrane supports — principles and applications.” J. Immunol. Methods 267: 13-26; and Pinilla et al (2003)“Advances in the use of synthetic combinatorial chemistry: mixture-based libraries.” Nature Med. 9: 118-122). The dissociation constants of peptides are typically in the micromolar range, although avidity can be improved by multimerization (Terskikh et al (1997)“Peptabody”: a new type of high avidity binding protein. Proc. Natl Acad. Sci. USA 94, 1663-1668; and Wrighton et al (1997) “Increased potency of an erythropoietin peptide mimetic through covalent dimerization. Nature Biotechnol. 15, 1261-1265).

In another embodiment, the agent to be tested may be a peptidomimetic. The term “peptidomimetic” refers to a compound that mimics the conformation and desirable features of a particular peptide as a therapeutic agent, but that avoids the undesirable features. For example, morphine is a compound which can be orally administered, and which is a peptidomimetic of the peptide endorphin.

Peptidomimetics are small molecules that can bind to proteins by mimicking certain structural aspects of peptides and proteins. They are used extensively in science and medicine as agonists and antagonists of protein and peptide ligands of cellular and other receptors, and as substrates and substrate analogues for enzymes. Some examples are morphine alkaloids (naturally-occurring endorphin analogues), penicillins (semi-synthetic), and HIV protease inhibitors (synthetic). Such compounds have structural features that mimic a peptide or a protein and as such are recognised and bound by other proteins. Binding the peptidomimetic either induces the binding protein to carry out the normal function caused by such binding (agonist) or disrupts such function (antagonist, inhibitor).

Peptidomimetics that are non-peptide in nature can be designed and synthesised by standard organic chemical methods. Peptidomimetics that are non-peptide in nature can be even more advantageous in therapeutic use, in the resistance to degradation, in permeability and in possible oral administration.

In another embodiment, the agent to be tested may be a nucleic acid. By“nucleic acid” we include the meaning of both DNA and RNA, single or double stranded, synthetic or natural.

In still another embodiment, the agent to be tested may be an aptamer, i.e. a single- stranded DNA molecule that folds into a specific ligand-binding structure. Aptamers typically have dissociation constants in the micromolar to the subnanomolar range.

Preferably, the target protein is selected from the group comprising: an enzyme, a signalling protein, a receptor, a transcription factor, a ribozyme; and a scaffold protein.

Target proteins may include transferase, oxidoreductase, hydrolase, ligase, and isomerase, along with kinases, phosphatases, carboxylases, phosphodiesterases, dehydrogenases, oxidases, peroxidases, proteases, metalloproteins, cytoplasmic proteins and nuclear localization proteins. Target proteins may also include signalling proteins that govern basic cellular activities and coordinate cell actions.

Preferably the target protein is a human protein.

The method of any of the preceding claims wherein the target protein and/or variant is introduced exogenously to the one or more cell.

In an embodiment, both the target protein and the variant are exogenous. Preferably, the variant protein is exogenous.

By“introduced exogenously” we include the meaning that the variant and/or target protein is not native to the cell, and therefore originates from outside the one or more cell and is subsequently introduced. For example, the one or more cell may be transformed with an exogenous nucleic acid, where the exogenous nucleic acid codes for the polypeptide of interest.“Exogenous nucleic acid” means a nucleic acid sequence that is not native to the one or more cell. For example, the target protein may be recombinantly expressed e.g. may be expressed from a plasmid which has been introduced into a cell. If exogenous, the target protein, or variant of the target protein may be made by recombinant DNA technology. Suitable techniques for cloning, manipulation, modification and expression of nucleic acids, and purification of expressed proteins, are well known in the art and are described for example in Sambrook et al (2001 )“Molecular Cloning, a Laboratory Manual”, 3rd edition, Sambrook et al (eds), Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, USA. Alternatively, the target protein, or variant of the target protein, may be made using protein chemistry techniques for example using partial proteolysis (either exolytically or endolytically), or by de novo synthesis.

If introduced into the one or more cell exogenously, the variant and/or target protein can be inducibly or constitutively expressed in the one or more cell.

The method of any one of the preceding claims wherein the target protein and/or variant is endogenous to the one or more cell.

By“endogenous” we include the meaning that the target protein and/or variant is native to, and/or originates from within, the one or more cell. In other words, the target protein and/or variant is naturally occurring in the one or more cell.

In an embodiment, the one or more cell provided in step (ii) comprises an endogenous target protein and an exogenous variant which is thus introduced into the one or more cell. As can be seen in the accompanying Examples, NUDT15 is endogenous to HCT1 16 3-6 cells and NUDT15 R139C is exogenous and is introduced into the HCT116 3-6 cell using a lentivirus (Figure 1 ).

In an alternative embodiment, the one or more cell provided in step (ii) comprises an endogenous variant and an exogenous target protein which is thus introduced into the one or more cell. For example, the one or more cell may contain a protein comprising a naturally-occurring single nucleotide polymorphism (SNP), or missense mutation which results in an endogenous variant. In this embodiment, the target protein is introduced exogenously. It will be appreciated by the skilled person that a cell which comprises an unstable variant of a target protein of interest could be used in the methods of the invention.

In some embodiments, the one or more cell is transiently and/or stably transformed or transfected with vector(s) (e.g., encoding target and/or variant proteins and/or fusions thereof, etc.). In some embodiments, transgenic organisms are generated that code for the necessary components (e.g., encoding target and/or variant proteins and/or fusions thereof, etc.) to carry out the methods described herein. In other embodiments, vectors are introduced into whole organisms.

Preferably, step (iv) does not comprise one or more of a cellular thermal shift assay (CETSA); Drug Affinity Responsive Target Stability (DARTS) assay; HSP90 Inhibitor Stability Assay (HIPStA) and thermal proteome profiling (TPP).

In an embodiment, step (iv) does not comprise heating above the physiological temperature of the one or more cell. By“heating above the physiological temperature” we include the meaning that step (iv) does not comprise heating the one or more cell by more than 5°C above its physiological temperature, such as by 6°C, 7°C, 8°C, 9°C, 10°C, 15°C, 20°C or 30°C above physiological temperature. For example, if the one or more cell is a human cell, step (iv) does not comprise heating the human cell by more than 5°C of the physiological temperature of a human cell (i.e. 37°C).

Preferably, the method is a high-throughput method. By a“high-throughput” method we include the automation of a method such that large scale data collection and repetition becomes feasible.

Preferably, the method is automated. By an“automated” method, we include a method operated largely by automatic equipment. For example, the method may comprise the use of an integrated robot system consisting of one or more robots which transport assay- microplates from station to station for the automated dispending of assay components (e.g. sample and reagent addition), mixing, incubation, and readout and/or detection.

In an embodiment, any of steps (i) to (v) further comprise the use of robotics, data processing/control software, liquid handling devices, microtiter plates and/or sensitive detectors in order to be high-throughput.

Since microtiter plates have up to 1536 wells, hundreds of thousands of agents can be screened against hundreds of targets and variants thereof using available automated screening facilities. Agent-variant complexes showing stabilisation over control as read out by fluorescence or luminescence and showing modulation of target protein activity can then be easily identified.

Accordingly, the method of the invention is amenable to high through-put screening (HTS). This can be carried out using automated systems for microtiter plate assays, for example using microtiter plates with 96 wells, 384 wells or 1536 wells. Such methods and technology are known in the art.

The method of any of any of the preceding claims wherein any of steps (i) to (v) are performed in a microtiter plate.

The method of any one of the preceding claims, wherein the one or more cell of step (ii) additionally comprises a further target protein and a variant of the further target protein.

In an embodiment, more than one target and/or variant protein is analysed in the method of the invention and particularly at least 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more, such as 20, 30, 40, 50, 100, or more target and/or variant proteins may be analysed using the method of the invention. Accordingly, a plurality of target and variant proteins may be used in the method of the invention.

It will be appreciated that by comprising a further target protein and a variant of the further target protein in the one or more cell, following contacting the one or more cell with the agent to be tested, it will be possible to determine the stability of multiple variants and the one or more activity of multiple target proteins simultaneously.

It will be appreciated that the method of invention allows the selectivity of an agent to be determined. Without being bound by theory, the inventors hypothesise that the variant of a target protein, for which the agent had increased selectivity, would accumulate (through agent binding) to a greater extent than alternative variants of alternative target proteins present in the one or more cell. Therefore, using the method of the invention it would be possible to determine which is the preferred target protein for a given agent. For example, if the one or more cell contained target proteins from the same family, and variants thereof, the method of the invention could be used to identify family members that are preferentially stabilised, and therefore that the agent has increased selectivity for.

It is appreciated that the method of the invention can be used to identify agents that may be useful in combating a particular disease or condition.

The methods may comprise the further step of synthesising and/or purifying the identified agent. The methods may further comprise the step of formulating the agent into a pharmaceutically acceptable composition. Accordingly, the invention also includes a method of making a pharmaceutical composition comprising the step of mixing the agent identified using the methods described above with a pharmaceutically acceptable carrier.

Agents may also be subjected to other tests, for example toxicology or metabolism tests, as is well known to those skilled in the art.

A complex comprising:

i. a variant of a target protein as defined in any preceding claim; and ii. an agent as defined in any preceding claim;

wherein binding of the agent to the variant stabilises the variant.

Preferably, the complex is capable of detection.

Preferably, the complex further comprises a detectable moiety.

A further aspect of the invention provides a kit comprising:

i. a target protein as defined in any previous aspect of the invention; and ii. a variant of the target protein which has reduced stability as defined in any previous aspect of the invention.

Preferably, the kit further comprises one or more agent to be tested. It will be appreciated that the agent to be tested may be provided in a library of agents, such as a small molecule library, also known as a compound library. Small molecule libraries are well known in the art and are commercially available. For example, LOPAC®1280 (Sigma Aldrich) contains 1 ,280 pharmacologically active compounds. Commercially available compound libraries are also available from ChemDiv Inc. It will be appreciated that the library may comprise FDA approved drugs.

Preferably, the agent binds to, and modulates the one or more activity of the target protein.

The kit may further include one or more additional components which find use in practicing certain embodiments of the invention, including but not limited to enzyme substrates, cell growth media, buffers, a vector containing the variant, or one or more cell expressing the variant.

In an embodiment, the kit comprises a destabilising domain which may be introduced into the target protein of interest, in order to generate an unstable variant of the target protein. As described above, the genetically engineered attachment of a destabilising domain or degron to a target protein can confer instability and therefore produce an unstable variant of the target protein. The degradation domain can be genetically engineered to reside within the target protein, or at either the N- or C-terminus of the target protein. Accordingly, the destabilising domain may be provided in the form of a nucleic acid construct, or provided in a host cell which is capable of expressing the destabilising domain, which could subsequently be isolated from said host cell.

The kit may further contain positive and negative controls relevant to the protein of interest. For example, the kit may contain an epitope control which may be used to ensure the binding of the agent is stabilising the variant and not any other portion, such as a detectable moiety.

In addition to the above components, the subject kits may further include instructions for practicing the subject methods and, optionally, intact cells for use with the in-cell embodiments. These instructions may be present in the subject kit in a variety of forms, one or more of which may be present in the kit. One form in which these instructions may be present is as printed information on a suitable medium or substrate, e.g., a piece or pieces of paper on which the information is printed, in the packaging of the kit, in a package insert etc. Yet another means would be a computer readable medium, e.g., diskette, CD, etc., on which the information has been recorded. Yet another means that may be present is a website address which may be used via the internet to access the information at a removed site. Any convenient means may be present in the kits.

A further aspect of the invention provides use of one or more cell, each cell comprising a target protein and a variant of the target protein which has reduced stability, for identifying an agent that binds to, and modulates the one or more activity, of the target protein.

Preferably, the use comprises a method as defined in the first aspect of the invention.

Preferably, the agent, the target protein and/or the variant are as defined in any one of the preceding aspects.

A further aspect of the invention provides a method, complex or kit substantially as described herein, with reference to the accompanying description, examples and drawings. All of the documents referred to herein are incorporated herein, in their entirety, by reference. The listing or discussion of an apparently prior published document in this specification should not necessarily be taken as an acknowledgement that the document is part of the state of the art or is common general knowledge.

Preferred, non-limiting examples which embody certain aspects of the invention will now be described, with reference to the following figures:

Figure 1 : NUDT15 inhibitors reversibly bind and stabilize NUDT15 R139C in cells. a, In silico energy calculations for NUDT 15 mutants.

The structures were constructed from the crystallographic structure (PDB ID: 5BON). The coordinates for the A and B chains were used to construct a dimeric structure. Crystallographic water molecules were included in the structure based on the Dowser algorithm. Mutants were constructed based on this structure using the Mutator module of VMD. The protein structures were solvated in an orthorhombic periodic water box with a 10 A padding.

The protein was described using the CHARMM36 force field. Water molecules were described using the TIP3P model. A 2 ns equilibration MD simulation was performed with the atoms of the proteins restrained to their crystallographic positions. Subsequently, 20 ns MD simulations were performed for each structure. The coordinates of the Mg(ll) ions were restrained to their crystallographic positions in all simulations. The simulation was performed using Langevin dynamics with a time step of 2 fs. Electrostatic interactions were described with a Particle Mesh Ewald scheme with a grid spacing of 1 A. A 14 A cutoff was used for the Lennard-Jones interactions. All molecular dynamics simulations were performed using NAMD 2.11. b, Compounds 1 , 2 and 3 were tested for their ability to inhibit hydrolysis of dGTP by purified, recombinant human NUDT15 by enzyme-coupled malachite green assay.

16 nM NUDT15 was added to 100 mM dGTP in assay buffer composed of 100 mM Tris- acetate pH 8.0, 40 mM NaCI, 10 mM MgAc and 1 mM dithiothreitol (DTT) for 30 min in the presence of 200 nM pyrophosphatase and malachite green reagent (ammonium molybdate). Absorbance was read at 630 nm emission. c, Compounds 1 , 2 and 3 were tested for their ability to stabilize purified, recombinant wild- type (WT) NUDT15 by differential scanning fluorimetry (DSF). 4 mM purified, recombinant WT NUDT15 was added to buffer containing 100 mM Tris- Acetate, pH 7.5, 40 mM NaCI, 10 mM Mg-Acetate and 5x Sypro Orange in the presence of DMSO or 10 mM NUDT15 inhibitor. Samples were heated at 1“C/minute and analyzed on a Bio-Rad RT-qPCR machine. The data shown are the negative derivative of relative fluorescence units (RFU) over temperature (-5RFU/5T) and the minima are indicative of the melting temperature (T_m). d, Compounds 1 , 2 and 3 were tested for their ability to stabilize purified, recombinant missense mutant R139C NUDT15 by differential scanning fluorimetry (DSF).

4 mM purified, recombinant NUDT15 R139C was added to buffer containing 100 mM Tris- Acetate, pH 7.5, 40 mM NaCI, 10 mM Mg-Acetate and 5x Sypro Orange in the presence of DMSO or 10 pM NUDT15 inhibitor. Samples were heated at 1 °C/minute and analyzed on a Bio-Rad RT-qPCR machine. The data shown are the negative derivative of relative fluorescence units (RFU) over temperature (-6RFU/6T) and the minima are indicative of the melting temperature (TM). e, The 3xHA-NUDT15 R139C fusion protein.

NUDT15 R139C was cloned into the pENTR4-N-3xHA entry vector and subsequently transferred to the plNDUCER20, Tet-On lentiviral expression vector for cellular experiments. f, 5 pM Cmpd 3 was tested for its ability to induce accumulation of WT and R139C NUDT15 in cells up to 24 hours.

HCT1 16 3-6 cells transduced with plnducer20-3xHA-NUDT15 WT or R139C were incubated with 1 pg/mL doxycycline for 24 hours, then treated with DMSO (1 :1000 v/v) or

5 pM Cmpd 3 in DMSO (1 :1000 v/v) for the times indicated. The cells were then harvested and lysed, prior to Western blotting. The blots were probed with an anti-HA-tag antibody, anti-NUDT15 antibody and beta-actin. The black arrow indicates 3xHA-NUDT15 and the grey arrow indicates endogenous, WT NUDT15. Quantification of band intensity (relative to DMSO-treated samples) is plotted to the right (WT- blue, R139C - red). g, Kinetics of 3xHA-NUDT15 R139C accumulation in cells following a drug washout.

Cells were treated as in d, but for up to 72 hours. A second sample was treated with 5 pM Cmpd 3 for 24 hours, washed with PBS and then supplemented with fresh culture medium for an additional 72 hours. Protein quantification is displayed on the right with an arrow indicating when drug was removed from the washout sample. Figure 2: Cellular target engagement can be assessed with NUDT15 R139C at single cell resolution and multiplexed with phenotypic markers. a, Experimental set-up to confirm NUDT15 inhibitor cellular target engagement and potentiation of thioguanine. b, Representative micrograph of cells expressing 3xHA-NUDT15 R139C in the absence or presence of 33 mM Cmpd 3

HCT1 16 3-6 cells expressing 3xHA-NUDT15 R139C were incubated with serial dilutions of Cmpd 3 for 72 hours in a black-walled 96-well plate. Cells were then fixed with 4% paraformaldehyde, permeabilized with Triton X-100, and probed with anti-HA tag antibody overnight. Anti-mouse AlexaFluor 647 secondary antibody was then used to label HA signal before counterstaining cell nuclei with DAPI. Scale bar is equivalent to 50 pm. c, Dose-dependent effects of Cmpd 3 on 3xHA-NUDT15 R139C HA staining intensity.

Quantified from the experiment in b using Cell Profiler software. HA intensity is graphed as log2 scale ns - not significant; ^**** - p < 0.0001 ; Kruskal-Wallis test d, Quantification of 3xHA-NUDT15 R139C signal following dose-dependent Cmpd 3 incubation in the absence or presence of 200 nM thioguanine.

As part of the same experiment as b and c, cells treated with serial dilutions of Cmpd 3 were incubated with DMSO or 200 nM thioguanine. 3 hours following addition of NUDT 15 inhibitors, DMSO or thioguanine was added for an additional 72 hours prior to fixation and staining with anti-HA antibody. Log2 scale is used for the y-axis. ns - not significant; ^**^** - p < 0.0001 ; Kruskal-Wallis test e, Simultaneous quantification of yH2A.X in HCT1 16 3-6 3xHA-NUDT15 R139C cells treated with serial dilutions of Cmpd 3 in the absence or presence of 200 nM thioguanine.

As part of the same experiment in b, c, d, yH2A.X signal was detected with an anti-yH2A.X antibody and subsequent staining with anti-rabbit AlexaFluor 488 secondary antibody. Signal was again quantified with Cell Profiler and data are plotted as log2. ns - not significant; *^*** - p < 0.0001 ; Kruskal-Wallis test f, Simultaneous DAPI intensity quantification for determination of cell cycle in the presence of Cmpd 3 or Cmpd 3 and 200 nM thioguanine. DAPI signal was quantified with Cell Profiler and plotted as a histogram to create cell cycle profiles. g, Depiction of target engagement (HA signal), DNA damage (yH2A.X) and cell cycle (Hoechst) on a per-cell basis.

Log2(HA intensity) is plotted on the y-axis, DNA content (Hoechst) is plotted on the x-axis and size of the point represents DNA damage (yH2A.X). Cells treated with DMSO and 200 nM thioguanine are shown in dark grey and cells treated with 33 mM Cmpd 3 and 200 nM thioguanine are plotted light grey.

Figure 3: CeTEAM permits evaluation of cellular target engagement at single-cell resolution with phenotypic multiplexing. a, The principle of CeTEAM. Cells of interest express a wild-type, endogenous protein of interest (Target) and an exogenous unstable mutant of the target protein of interest (^mtTarget) fused to a protein tag (Tag) or an exogenous Target fused to a degron (^De9Target) and protein tag (Tag). Due to structural instability of the ^mtTarget or ubiquitination of the ^DegTarget, the exogenous Target is degraded by the cell and protein expression is low. In the presence of a ligand that binds the Target and ^mtTarget/^DegTarget, stabilization of the ^mtTarget/^De9Target causes its accumulation in the cell, which can be monitored with a protein-tag or specific antibody towards the tag. Binding of the ligand to the endogenous Target (or off-target interactions) also causes phenotypic alterations that can be measured, permiting simultaneous analysis of cellular target engagement and phenotypic responses due to a ligand in an individual cell.

Figure 4: CeTEAM can evaluate inhibitor libraries and distinguish on- and off-target effects to streamline lead identification. a, Cells treated with serial dilutions of Cmpd 3 and DMSO or 200 nM thioguanine for 72 hours, followed by staining for HA, yH2A.X and DNA content with Hoechst dye. HCT1 16 3-6 3xHA-NUDT15 R139C cells were treated as in Figure 2 and prepared for immunofluorescence microscopy as before. Cells were stained with anti-HA tag (+ anti- mouse AlexaFluor 647 secondary), anti-yH2A.X (+ anti-rabbit AlexaFluor 488 secondary) and Hoechst 33342. ns - not significant; ^* - p < 0.05; ^** - p < 0.01 ; ^*** - p < 0.001 ; ^**** - p < 0.0001 ; Kruskal-Wallis test compared to DMSO control b, Cells treated with serial dilutions of Cmpd 2 and DMSO or 200 nM thioguanine for 72 hours, followed by staining for HA, yH2A.X and DNA content with Hoechst dye. c, Cells treated with serial dilutions of Cmpd 1 and DMSO or 200 nM thioguanine for 72 hours, followed by staining for HA, yH2A.X and DNA content with Hoechst dye. d, Cells treated with serial dilutions of Cmpd 4 and DMSO or 200 nM thioguanine for 72 hours, followed by staining for HA, /H2A.X and DNA content with Hoechst dye. Figure 5: CeTEAM can be utilized to assess target engagement in vivo. a, A schematic describing target engagement evaluation in zebrafish embryos.

HA-NUDT 15 WT or R139C cells loaded with Dil stain were injected into zebrafish embryos. The cells were exposed to doxycycline for 24 hours followed by addition of DMSO or 20 mM Cmpd 3 for another 48 hours. The embryos were then fixed and cryosectioned, prior to staining with anti-HA antibody and imaging on a confocal microscope. Zebrafish images courtesy of Lizzy Griffiths. b, Representative images from the experiment described in a.

No HA signal was detected in embryos injected with 3xHA-R139C cells and only exposed to DMSO. Figure 6: CeTEAM with MTH1 G48E a, The V5-MTH1 G48E fusion protein

MTH1 G48E was cloned into the pENTR4-N-V5 entry vector and subsequently transferred to the plNDUCER20, Tet-On lentiviral expression vector for cellular experiments b, Assessing stabilization of V5-MTH1 G48E with MTH1 inhibitors, Cmpds 5, 6, 7 and 8. U-2 OS cells transduced with plNDUCER20-V5-MTH1 G48E were induced with doxycycline for 24 hours, followed by treatment with 1 mM of the indicated inhibitors for an additional 16 hours. c, Timecourse of MTH1 G48E stabilization, compared with proteasome inhibition with MG- 132 A single clone (#6) of U-2 OS cells transduced with plNDUCER20-V5-MTH1 G48E were induced with doxycycline for 24 hours, followed by treatment with 1 pM MTH1 compound 6 for the times indicated. MG-132 was added at 5 pM for 8 hours. d, Population dynamics of MTH1 G48E-expressing cells treated with MTH1 inhibitors A single clone (#6) of U-2 OS cells transduced with plNDUCER20-V5-MTH1 G48E were induced with doxycycline for 24 hours, followed by treatment with 1 mM MTH1 compound 5 or compound 6 for an additional 16 hours. Cells were fixed in paraformaldehyde and stained with anti-V5 primary antibody, followed by anti-mouse AlexaFluor647 secondary, and then analyzed by flow cytometry. e, Single-cell multiparametric analysis of MTH1 inhibitor target engagement and phenotypic changes by flow cytometry

A single clone (#6) of U-2 OS cells transduced with plNDUCER20-V5-MTH1 G48E were induced with doxycycline for 24 hours, followed by treatment with 1 mM of MTH1 compound 5 or compound 6 for an additional 16 hours. Cells were fixed in paraformaldehyde and stained with anti-V5 primary and anti-histone H3 Ser10 antibody, followed by fluorescent secondary antibodies. Cells were also counterstained for DNA content with Hoechst 33342 and then analyzed by flow cytometry. The left column depicts Hoechst signal (DNA content; linear) on the x-axis and V5 signal (stabilized MTH1 G48E; linear) on the y-axis, and the right column depicts V5 signal (stabilized MTH1 G48E; linear) on the x-axis and histone H3 Ser10 (logarithmic) on the y-axis. f, Quantification of percent histone H3 Ser10⁺ and V5⁺ cells from g Quantification of the data presented in panel g.

Figure 7: Exploring the PARP1 mutant, L713F, with CeTEAM a, The PARP1 L713F-eGFP fusion protein

PARP1 L713F was cloned into the pENTR1a-C-eGFP entry vector and subsequently transferred to the plNDUCER20, Tet-On lentiviral expression vector for cellular experiments. b, PARP1 L713F-GFP, but not WT-PARP1-GFP, is stabilized by PARP1/2 inhibitor, olaparib

U-2 OS cells transduced with plNDUCER20-PARP1 L713F-GFP were induced with doxycycline for 48 hours, followed by treatment with DMSO or 5 mM of olaparib (Ola) for 24 hours. MG-132 was added at 5 pM for 16 hours. U-2 OS cells transduced with plNDUCER20-PARP1 WT-GFP were treated in an identical manner. c, Several clinical PARP inhibitors selectively stabilize PARP1-L713F-GFP and induce replication stress, but not iniparib U-2 OS cells transduced with plNDUCER20-PARP1 L713F-GFP were induced with doxycycline for 24 hours, followed by treatment with DMSO or the indicated inhibitors and doses for an additional 48 hours. U-2 OS cells transduced with plNDUCER20-PARP1 WT- GFP were treated in an identical manner. Cells were harvested, prepared for western blotting and probed with the indicated antibodies. d, Stabilization of PARP1-L713F-GFP, but also induction of replication stress, is time- and dose-dependent in the presence of olaparib

A single-cell clone (clone #6) of U-2 OS cells transduced with plNDUCER20-PARP1 L713F-GFP were induced with doxycycline for 24 hours, followed by treatment with DMSO or the indicated doses and time of olaparib. Cells were harvested, prepared for western blotting and probed with the indicated antibodies. e, Dose-dependent stabilization of PARP1-L713F-GFP with different PARP inhibitors in live cells

A single-cell clone (clone #5) of U-2 OS cells transduced with plNDUCER20-PARP1 L713F-GFP were induced with doxycycline for 24 hours, followed by treatment with DMSO or the indicated doses of PARP inhibitors for an additional 24 hours. Twenty minutes prior to imaging by high-throughput microscopy, all samples were incubated with Hoechst 33342 DNA stain (1 pg/mL). Median GFP fluorescence of each cell population is shown. Data is from two independent experiments. f, Dose-dependent increase in median DNA content with different PARP inhibitors in live cells

From the same experiment shown in e. Median GFP fluorescence of each cell population is shown. g, Ranking of PARP inhibitors by reported PARP1 in vitro IC50 (lowest to greatest) and PARP: DNA trapping ability (best to least) h, Visualizing simultaneous dose-dependent PARP1 target engagement and PARP trapping in live cells

Median GFP fluorescence signal (y-axis) and median DNA content (x-axis) plotted for each treatment group from experiment performed in e and f. -/+ doxycycline (DOX) is represented in grey tones; increasing dose is indicated by blue gradient. Points are designated as“non-PARP trapping” (left of dotted line) and“PARP-trapping” (right of dotted line). i, Single-cell population dynamics of simultaneous PARP1 target engagement and PARP trapping in live cells

Depiction of GFP fluorescence (target engagement) vs DNA content of single cells in the absence and presence of PARP inhibitors (same experiments as e, f and h). Black dots represent DMSO-treated cells and grey dots represent cells treated with the specified PARP inhibitor.

Figure 8: NUDT15 R139C is expressed but undergoes proteasomal degradation in cells. a, HCT116 cells overexpressing HA-WT or HA-R139C NUDT15 after 72 hours of doxycycline induction were treated with 5 mM MG-132 for 3, 6, 9, 12, or 24 hours. Gray arrows indicate HA-tagged expression constructs and the black arrows endogenous NUDT15 (WT). p53 was blotted as a control for MG-132 treatment. b, Representative DSF melting curve demonstrating the stability of NUDT15 WT (blue) and the R139C mutant (red) in the absence or presence of the reducing agent, TCEP (filled or empty circles, respectively). Relative fluorescence units (RFU) represent fluorescence of SYPRO Orange at 570 nm.

Figure 9: GFP-MTH1-nMyc degron is stabilized by Cmpd 6 in cells.

a, Design of the lentiviral expression cassette containing GFP-MTH1-nMyc degron and an independently translated iRFP670 to act as a transduction control.

A second multiple cloning site (MCS; Eurofins GeneStrand) was inserted into pENTRIa by Sall/Notl ligation to generate pENTR2x. An IRES sequence (Eurofins GeneStrand) was then inserted by Ncol/EcoRI digestion and ligation to generate pENTR2x-IRES. Subsequently, iRFP670 was inserted prior to the IRES sequence by subcloning with BamHI/EcoRI and eGFPnMyc degron (corresponding to the peptide: LEKEKLQARQQQLLKKIEHARTC, Eurofins GeneStrand) was inserted following the IRES sequence with Ncol/Xbal to generate pENTR2xiRFP670-IRES-eGFP-nMyc degron. The entry vector was finalized by subcloning wild-type MTH1 (p18, NM_002452.3) by Sall/Notl to make pENTR2x-iRFP670-IRES-eGFP-MTH1-nMyc degron. All insertions were then validated by sequencing. The ORF was then shuttled into the doxycycline-inducible lentiviral expression vector, pCW57.1 (Addgene #41393, a kind gift from David Root), with LR Clonase II (Life Technologies) to generate pCW57.1 -IRFP670-IRES-eGFP-MTH1- nMyc degron. b, Experimental design to test GFP-MTH1-nMyc degron stabilization by MTH1 inhibitor, Cmpd 6. U-2 OS cells transduced with pCW57.1 -iRFP670-IRES-eGFP-MTH1-nMyc degron were pretreated with 1 pg/mL doxycycline for 24 hours, then incubated with DMSO (1 :1000 v/v) or 1 mM Cmpd 6 in DMSO (1 :1000 v/v) for an additional 48 hours. Doxycycline was again added 24 hours prior to harvesting (total exposure time was 72 hours). Cells without doxycycline exposure were included as a negative control. c, Western blot of U-2 OS pCW57.1-iRFP670-IRES-eGFP-MTH1-nMyc degron cells incubated with Cmpd 6. The cells were treated as in Figure 1b, then harvested and lysed, prior to Western blotting. The blots were probed with an anti-GFP antibody (Santa Cruz). Anti-PARP1 staining was used as a loading control (Santa Cruz). d, Quantification of GFP signals from Figure 1c. Western blot band intensities were quantified by Image Studio Software (Li-COR Biosciences)and GFP signal was divided by PARP1 signal for each sample to generate relative GFP intensity. The values were then normalized to cells incubated with doxycycline alone.

Table 3: C-terminal degrons targeted by the CRL2 complex.

C-terminal peptide degron sequences found in various human proteins resulting in their degradation by the CRL2 E3 ubiquitin ligase complex. The asterisks (^*) represent the stop codon for the respective protein.

Table 4: C-terminal degrons targeted by the CRL4A complex.

C-terminal peptide degron sequences found in various human proteins resulting in their degradation by the CRL4A E3 ubiquitin ligase complex. The asterisks (*) represent the stop codon for the respective protein.

Table 5: C-terminal degrons targeted by the CRL4B complex.

C-terminal peptide degron sequences found in various human proteins resulting in their degradation by the CRL4B E3 ubiquitin ligase complex. The asterisks (^*) represent the stop codon for the respective protein.

Table 6: C-terminal degrons targeted by the CRL4A and CRL4B complexes.

C-terminal peptide degron sequences found in various human proteins resulting in their degradation by the CRL4A and CRL4B E3 ubiquitin ligase complexes. The asterisks (^*) represent the stop codon for the respective protein. Example 1:

Scalable, portable and multiparametric cellular target engagement with Cellular

Target Engagement by Accumulation of Mutant (CeTEAM)

As proof of principle, NUDT15 (also referred to as MTH2) (Valerie et al., Cancer Res. 2016 Sep 15;76(18):5501-11 ), NUDT1 (also referred to as MTH1 ) (Mur et al Hum Mutat. 2018 Jun 13.) and PARP1 (Rank et al., Nucleic Acids Research, 2016, Vol. 44, No. 21 ; Langelier et al., Science. 2012 May 1 1 ;336(6082):728-32; Miranda et al., Biochem Biophys Res Commun. 1995 Jul 17;212(2):317-25) and their known mutants were used. The NUDT15 R139C mutation destabilizes the protein so that it is degraded rapidly in cellular environments. As demonstrated in Example 1 , the R139C mutant was less thermally stable than NUDT15 WT.

To determine whether NUDT15 R139C is degraded by the proteasome, we performed a time course experiment with the proteasome inhibitor, MG-132. After 72 hours of doxycycline induced overexpression of NUDT 15 WT or R139C, doxycycline was removed. MG-132 (5 mmol/L) was added for the indicated time points starting with the longest treatment and all samples were harvested at the same time point. While inhibition of the proteasome did not lead to increased accumulation of overexpressed NUDT 15 WT protein, treatment with MG-132 led to a clear and distinct accumulation of R139C in a time- dependent manner (Fig. 8a.) Thus, while the NUDT15 R139C mutant is expressed normally in cells, the protein is rapidly degraded.

HA-tagged WT or R139C NUDT15 were overexpressed using doxycycline-inducible expression constructs in HCT1 16 cells, carrying endogenous WT NUDT15. Overexpression of the HA-tagged proteins was assessed with an anti-HA or anti-NUDT15 antibody. When analyzing protein levels, overexpressed NUDT15 WT was robustly induced upon doxycycline addition, but the overexpressed R139C mutant was barely detectable (Fig. 8b). Protein stability was investigated in vitro using a DSF assay and compared nonreducing and reducing conditions. NUDT15 WT had a melting point (Tm) of 59°C and 58°C with or without TCEP, respectively. This was much higher than the NUDT15 R139C mutant, which had a melting point of 48°C without TCEP and 51 °C with TCEP (Figure 8b). Abstract

Recent developments of cellular target engagement technologies already impacted the ways in which we identify and evaluate promising drug candidates. With an expanding repertoire of available techniques, drug-discovery scientists can determine if their molecules reach their intended targets in cells and tissues with unprecedented accuracy. Within this context, we present CeTEAM, Cellular Target Engagement by Accumulation of Mutants, as a technology to simultaneously evaluate drug-target interactions and phenotypic responses in cells. Relying on the stabilization of engineered variants of a protein of interest, CeTEAM can be employed for applications from high-throughput screening, to mechanism of action studies, lead optimization and in vivo efficacy studies. CeTEAM is a portable, scalable and multi-parametric target engagement technology with the potential to expedite drug discovery and development programs.

Introduction

Confirming cellular target engagement has become a determining factor in the value of probe molecules and the efficacy of clinically used drugs^{1 3}. Until recently, detecting and quantifying cellular target engagement in a straightforward and comprehensive manner had been a major hurdle within drug development. With the advent of the cellular thermal shift assay (CETSA)⁴·⁵, drug discovery scientists, both in academic or industry settings, can evaluate drug-target interactions with relative ease. From this stemmed thermal proteome profiling (TPP), which can also be used to identify unknown targets for established drugs⁶, as well as help to understand selectivity of a given ligand⁷·⁸. Indeed, the successes and limitations of CETSA and TPP have inspired the recent development of several additional target engagement technologies; including methods to quantify target engagement in single cells and in v/Vo^{9 15}.

Modulating cellular protein stability also has applications beyond thermal shift assays. Within the biosensors field, unstable protein variants are utilized to detect the presence of specific chemicals, metabolites or even drug molecules¹⁶. In a similar fashion, modifying a protein of interest (POI) with an engineered/unstable ligand binding domain (LBDs) has been used to rapidly and conditionally regulate protein levels in a tunable, reversible manner, which is particularly valuable when investigating protein function^{17 22}. In this setting, a POI is ectopically expressed as a fusion protein attached to an unstable LBD. This fusion protein is inherently unstable and will be proteolytically degraded in cells unless the LBD is stabilized by binding to specific ligands²³. Beyond engineering unstable LBDs, modifying protein stability is also routinely done to develop biochemical tools and model systems. This includes directed mutagenesis to increase protein stability for protein crystallography applications²⁴, as well as, mutagenesis to replicate missense mutations that are underlying causes for a number of human diseases, (i.e. cystic fibrosis). Pharmaceutical approaches to stabilize functional proteins with targeted ligands, also known as pharmacochaperone therapy, is also an active field of drug discovery research.

An additional consideration in modulating protein degradation comes in the form of degrons, which consist of N- or C-terminal protein fragments that act as targets for ubiquitylation^{25 27}. Degrons regulate intrinsic protein turnover and their fusion to small molecules has created a new class of promising therapeutics, known as PROTACs^28,29. Cryptic degrons may be exposed upon heating or partial unfolding of the particular domain³⁰, thus permitting ubiquitylation of key lysine residues³¹; in-turn, it is likely that many destabilizing missense mutations cause rapid proteolytic degradation in a similar fashion³². As such, fusing degron sequences to the termini of stable POIs results in conditional depletion in cells^{25 27,30}’³¹. Like traditional unstable LBDs, POI-degron fusion proteins (e.g., FKBP12-L106P:POI¹⁹, C-b-gakPOI^25,33, Arg-DHFR:POI^30,33 or R- ARD1 :POI²², among others^16,31), can be stabilized in the presence of a ligand bound to the I__BD ^{19,21 ,34} _or pop⁶, which has also proven useful in the study of protein functions³⁵ and increasing tunability of LBD-based biosensors¹⁶. Despite the many innovations propelled by modulating protein stability in live cells, an underappreciated application may be in measuring intracellular target engagement of small drug-like molecules.

In this light, we present CeTEAM - Cellular Target Engagement by Accumulation of Mutants - as a novel, translatable, and multiparametric cellular target engagement technology. With CeTEAM, a POI is mutated or coupled to a degron sequence to induce its degradation in cells and fused to a peptide tag. The mutations or degron appendages are constructed in a way that should preserve key aspects of the protein active sites, and should maintain functional capacity. However, under basal cellular conditions, this unstable fusion protein is rapidly degraded in cells via the ubiquitin-proteasome system or other pathways. With CeTEAM, a small molecule that binds to the POI, will stabilize the fusion protein to impede its proteolytic degradation. As a result, the tagged mutant POI will accumulate in cells upon inhibitor binding.

With the assumption that engagement of the tagged mutant POI can be extrapolated to indicate that the endogenous wild-type (WT) protein also binds the compound in question, CeTEAM makes it possible to quantify cellular target engagement by any conventional protein detection method. This can be done with single-cell resolution and can be combined with other phenotypic readouts to permit multiparametric analyses. CeTEAM represents a truly adaptable technology that can be utilized starting from early screening applications, to inhibitor discovery and optimization, and all the way to preclinical in vivo characterization for inhibitor development programs.

Results

NUDT15 and thiopurine metabolism as a model system for CeTEAM

NUDT15 (also referred to as MTH2) is a NUDIX hydrolase that is a key factor within the metabolism of thiopurines (6-thioguanine, 6-mercaptopurine and azathioprine), hydrolyzing the active form of these drugs (6-thio-dGTP) to reduce their efficacy. The NUDT15 R139C mutant has been thoroughly studied for its role in thiopurine sensitivity³⁶ ⁴⁰ and is found within a growing list of populations. Despite showing strong clinical phenotypes with respect to thiopurine drug sensitivity, patients with the R139C mutation have not shown other adverse effects⁴¹.

Somewhat surprisingly, NUDT15 R139C maintains its activity towards thioguanine triphosphate species; however, this mutation destabilizes the protein so that it is degraded rapidly in cellular environments, however, its expression can be rescued by proteasome inhibition³⁷. Thus, it would appear that the active site of NUDT15 R139C remains intact despite the point mutation within a proximal alpha-helix⁴². While crystallographic methods have been attempted and would be of further benefit, it is suspected that the impaired stability of NUDT15 R139C is impeding protein crystallography efforts. This was further studied computationally using FEP simulations for the NUDT15 R139 mutations generated previously (Figure 1a)³⁷. This model further supports previous hypotheses suggesting that the cysteine residue cannot make the inter-helical ionic interaction that is present between R139 and D132 in NUDT15 WT and that this contributes to NUDT15 R139C instability. Furthermore, and in-line with enzymatic activity experiments, no major differences are observed between WT and R139C active sites following FEP simulations.

In an effort to find drug-like molecules that would synergize with thiopurines, we sought to identify small molecule NUDT15 inhibitors. Thus, we performed a high-throughput screen for NUDT15 inhibitors using a malachite green reporter assay, as was conducted previously⁴³. Hit compounds from the screen were optimized for biochemical NUDT15 inhibition and then evaluated for synergistic effects with 6-thioguanine in HL60 cells using the resazurin cell viability assay. Several of our top compounds showed promising activity in combination with 6-thioguanine, so we turned to target engagement techniques, including the cellular thermal shift assay (CETSA), to ensure the inhibitors were engaging NUDT15 in cells. Using CETSA, we could show effective stabilization of NUDT15 with our lead small molecule inhibitors; however, due to a poor anti-body signal and the relatively low-throughput nature of CETSA, we sought alternatives to show NUDT15 engagement.

NUDT15 R139C is stabilized in vitro and in cells with small molecule NUDT15 inhibitors

Stemming from our work with the NUDT15 R139C³⁷, we considered the possibility that this unstable protein could be stabilized by targeted small molecule inhibitors. Thus, we first performed differential scanning fluorimetry (DSF) using recombinant WT and R139C NUDT15 and inhibitors that were developed in-house. Two of the molecules were potent inhibitors in the NUDT15 malachite green biochemical assay, while the third had no inhibitory activity (Figure 1b). Inhibitors 2 and 3 stabilized WT NUDT15 by approximately 10°C (melting temperature (T_m) shift from 56 °C to 64-66 °C), while the inactive inhibitor 1 gave a nearly identical response to protein treated with DMSO alone (Figure 1c). NUDT15 R139C had a much lower basal melting temperature compared to the WT protein, as previously³⁷; however, it could also be stabilized with inhibitors 2 and 3, again providing a 10 °C T_m shift (from 46 °C with DMSO to 54-58°C with inhibitors 2 and 3; Figure 1d). Therefore, the unstable NUDT15 R139C mutant could be stabilized in vitro by NUDT15 inhibitors.

We next sought to determine if NUDT15 R139C could be similarly stabilized in cells. To do this, we used HCT116 3-6 cells expressing an HA-tagged NUDT15 R139C (HA- NUDT15 R139C, Figure 1e) or WT protein under a doxycycline-inducible promoter, as was previously described³⁷, and treated the cells with 5 mM inhibitor 3 for up to 24 hours (Figure 1f). As expected, expression of the HA-NUDT15 R139C fusion gene in the presence of DMSO gave little-to-no protein expression by western blot. Incubation with inhibitor 3 caused accumulation of the HA-NUDT15 R139C as early as 3 hours after addition, with an approximately 8-fold increase in protein level after the 24-hour treatment (compared to DMSO treated controls). As a control, HA-NUDT15 WT protein was well- expressed in HCT1 16 3-6 cells upon doxycycline exposure and the expression levels of this protein were unaffected by inhibitor treatment. A longer time-course with the R139C mutant indicated maximal accumulation occurred around 72 hours (=10-fold higher than DMSO control; Figure 1g). This effect was reversible, as HA-NUDT15 R139C protein levels returned to basal levels after approximately 72 hours following a 24-hour incubation with inhibitor 3 and drug washout. Thus, cellular NUDT15 R139C protein is temporally and reversibly accumulated in the presence of a NUDT15 inhibitor. CeTEAM is a target engagement platform delivering unparalleled insight

It was then apparent that monitoring the accumulation of NUDT15 R139C could be a method to track cellular target engagement of ligands directed towards NUDT15 and potentially other protein targets. As we are simply quantifying protein accumulation, it should be possible to detect target engagement with any conventional protein detection technique, down to single-cell resolution and at a throughput that is only limited by instrumentation available to the researcher. Using the same HCT1 16 3-6 HA-NUDT15 R139C cells, we set up a 96-well imaging plate and exposed the cells to serial dilutions of inhibitor 3 (up to 33 mM) for 72 hours and quantified HA signal by high-throughput immunofluorescence microscopy (Figure 2a). This resulted in a dose-dependent increase in average cellular HA intensity, indicating that this method is sensitive enough to detect a wide range of inhibitor concentrations at single cell resolution (Figure 2b and c). Noticeably, the cellular responses are heterogeneous in nature, which may reflect the heterogeneity of the lentiviral transduction or compound penetrance into individual cells, among other possibilities.

Another key advantage of this approach is that cellular target engagement is determined at physiological temperatures and culture conditions. This presents a unique opportunity to directly relate target engagement to phenotypic responses - an aspect that is not addressed or not possible with other available cellular target engagement methods. Given that NUDT 15 hydrolyzes 6-thioguanine triphosphates in cells and its depletion potentiates thioguanine toxicity³⁷, we sought to combine NUDT15 inhibitors with a low-dose of 6- thioguanine (200 nM) 3 hours after adding inhibitor 3 - with the goal of simultaneously analyzing target engagement and other phenotypic readouts (Figure 2a). The incorporation of 6-thioguanine into genomic DNA results in a prolonged G2 cell cycle delay caused by futile mismatch repair cycling and ATR-Chk1 activation³⁷'^{44 48}. Thus, in addition to probing for HA signal intensity, we also stained for yH2A.X, a ubiquitous marker of DNA damage, and counterstained nuclear DNA with Hoechst 33342 to determine the cell cycle (Figure 2d, e, and f). Titration of NUDT15 inhibitor 3 alone increased the HA signal in a dose-dependent manner, indicating target engagement, and there were no differences in yH2A.X intensity or DNA content compared to the DMSO control. However, in the presence of 200 nM thioguanine, the same dose-dependent shift in HA signal appeared, but there was also an increase in gH2A.C intensity and G2/M cell cycle content. Visualization of all three parameters simultaneously emphasizes the correlation of NUDT15 target engagement with 6-thioguanine potentiation (Figure 2g). CeTEAM gives unparalleled power compared to other target engagement assays. The keys to this technique are 1 ) the expression of an exogenous, unstable mutant protein of interest fused to a protein tag (or WT target protein fused to degron sequence), 2) the concurrent expression of endogenous, WT protein in the cell and 3) preserved active site integrity between the variant and WT protein (Figure 3). In the absence of a stabilizing ligand, the variant is degraded by the cell but can be rescued by stabilization with a ligand. Thus, the variant acts as a surrogate to measure cellular target engagement and is not appreciably active in the cell. The assumption that is made is that the inhibitor will interact with both the endogenous target protein and the variant, creating both a target engagement response ( via variant binding) and a measurable phenotypic response from binding the endogenous protein or other unintended targets. As a consequence, it is possible to generate information regarding context-dependent on- and off-target activities.

We next tested other NUDT15 inhibitors with a range of biochemical potencies under the same conditions, including one that had no inhibitory activity, compound 1 ; potent NUDT 15 inhibitors, compounds 2 and 3; and one compound that was active but was expected to behave similarly to 6-thioguanine in cells, compound 4 (Figure 4). As before, compound 3 gave a titratable target engagement response by HA signal and a similar pattern in gH2A.C intensity and cell cycle profile with the addition of 200 nM 6-thioguanine (Figure 4a) Compound 2 gave essentially the same profile under all conditions but at smaller overall magnitude, which is likely reflective of its weaker inhibitory activity in vitro (Figure 4b). In line with biochemical characterizations, treatment with up to 33 mM of compound 1 gave no noticeable differences in HA accumulation, yH2A.X intensity or cell cycle alterations in the absence or presence of 200 nM 6-thioguanine, (Figure 4c). Lastly, compound 4 produced a profile distinct from the other three. HA, gH2A.C, and G2/M cells were significantly elevated with compound alone, in a dose-dependent manner, and the addition of 6-thioguanine had no further effect (Figure 4d). Thus, this compound binds to HA-NUDT15 R139C and simultaneously elicits thiopurine-like toxicity, suggesting a dichotomous role of this molecule in cells. Thus, an added benefit of CeTEAM is the ability to discern non-responders and on- versus off-target effects of particular compounds or larger libraries of potential inhibitors.

Preclinical evaluation by in vivo target engagement with CeTEAM

One particular downside of CETSA and many other techniques is that the tissue or xenograft needs to be excised and processed, or measured with specialized equipment in order to assess cellular target engagement in animal models^5,14’¹⁵. Additionally, while it is possible to assess target engagement in vivo with CETSA, the sample analysis requires isolation of a population of cells to be used, where one loses valuable spatial information. Thus, information pertaining to drug penetrance or how tumor heterogeneity and microenvironment affects target engagement can only be crudely preserved with such methods, if at all. In contrast, with CeTEAM, it is possible to preserve spatial topology with intact tissues and live animals, allowing for unprecedented understanding of how drugs interact with their targets within intact tissues.

As CeTEAM does not require highly specialized equipment and can be performed under physiological conditions, we sought to evaluate the application of this technology in zebrafish xenografts (housed at 33°C) using the HCT 1 16 3-6 HA-NUDT 15 WT and R139C cells and examining by cryosectioning and confocal microscopy (Figure 5a). Following implantation of the dil-stained cells in zebrafish embryos, induction with doxycycline gave a strong HA signal with the HA-NUDT15 WT cells, but no discernable signal in the R139C cells (Figure 5b). Upon treatment of 20 mM compound 3, the HA signal appeared in the HA-NUDT15 R139C xenografted cells, albeit at lower intensity than HA-NUDT15 WT cells. No HA signal was detected in HA-NUDT15 R139C samples without NUDT15 inhibitors. Therefore, CeTEAM is highly translatable and can be utilized from lead identification to preclinical evaluation stage of drug/probe development programs.

CeTEAM for evaluating inhibitors of the proposed anticancer target, MTH1 (NUDT1)

To get an idea of how generalizable CeTEAM can be as a target engagement technology, we began experimenting with other, relevant cancer drug targets, as well as other fusion tags. Inhibitors targeting MTH1 , a key player in the sanitation of the oxidized nucleotide pool, were initially promising⁴⁹, but the lack of activity seen with recently synthesized and biochemically-potent inhibitors from a number of independent research groups has cast doubt on its value as a cancer target^{50 52}. Missense mutations to MTH1 were recently identified in patients with familial colon cancer⁵³. One of them, G48E, is a prime CeTEAM candidate, as the protein product is clearly destabilized in comparison to WT MTH1 in vitro. MTH1 G48E maintained its activity towards canonical MTH1 substrates, demonstrated a strong decrease in thermal stability by DSF and was inhibited and stabilized by small molecule MTH1 inhibitors in vitro⁵³. In addition, when WT and G48E MTH1 were fused to a V5 epitope tag in U-2 OS cells (Figure 6a), the V5-MTH1 G48E protein expression was significantly lower than V5-MTH1 WT⁵³. Treatment of these cells with the proteasome inhibitor MG-132 induced a temporal increase in V5 signal up to 9 hours after treatment, suggesting that V5-MTH1 G458E was also being degraded by ubiquitin-proteasome pathways, in a similar manner to what was seen with the NUDT15 R139C mutant^37,53. Thus, to evaluate this mutant as a candidate for CeTEAM, the V5-MTH1 G48E-expressing cells were treated with several structurally diverse MTH1 inhibitors for 24 hours at 1 mM (Figure 6b). As controls, we had cells treated with doxycycline alone. As expected, the doxycycline-treated cells had a poorly visible band corresponding to V5-MTH1 G48E when probed for V5 or MTH1 by Western blot, but cells treated with inhibitors 5, 6, 7 or 8 contained substantially higher levels, which was indicative of inhibitor binding and slowed turnover of the mutant species. This accumulation increased temporally (up to 24 hours) and was comparable to treatment with the proteasome inhibitor, MG-132 (Figure 6c). Thus, V5-MTH1 G48E can be utilized for CeTEAM evaluation of MTH1 inhibitors in cells.

Among the current controversy surrounding MTH1 inhibitors, is the potential influence of off-target interactions seen in cancer cell-killing MTH1 inhibitors⁵⁴. MTH1 inhibitors with anti-tumor activity are reported to cause mitotic arrest and apoptotic cell death^{54 56}. Therefore, in addition to monitoring V5-G48E target engagement, we probed for histone H3 phosphorylated at Ser10, which is a marker of mitotic cells, as well as yH2A.X, a ubiquitous marker of DNA damage. Of the four compounds tested, only MTH1 inhibitor 5 gave substantial increases in phosphorylated histone H3 and yH2A.X (Figure 6b), whereas the others had no discernable effects on these markers.

To gain insight on how this would look at single-cell resolution, clonal populations from the original V5-MTH1 G48E transductants were expanded. This was necessary in order to eliminate superficial variation within the cell populations. These clonal populations were then treated with MTH1 inhibitor 5 or 6, as before, but probed with anti-V5 and anti- phosphorylated histone H3 antibodies, as well as Hoechst 33342 DNA stain, before analysis by flow cytometry (Figure 6d, e, f). Addition of the MTH1 inhibitors noticeably increased the median V5 signal of V5-MTH1 G48E cells (Figure 6f), but simultaneous visualization of target engagement with phosphorylated histone H3 or DNA content provided clear evidence that engagement of compound 5 correlated with increased phospho-HH3 and G2/M DNA content on a per-cell basis (Figure 6e, f). That is to say that cells containing this inhibitor tend to arrest in mitosis. On the other hand, cells treated with MTH1 inhibitor 6 displayed increases in V5 target engagement signal indiscriminately throughout the cell cycle, without changing the overall distribution as compared to DMSO- treated cells, and had no tendency to increase phospho-HH3. Thus, although all MTH1 inhibitors tested are extremely potent in vitro and stabilize the V5-G48E mutant, CeTEAM can easily discern that their effects on cell populations vary tremendously. Visualizing PARP1 target engagement and trapping with CeTEAM

To venture outside of the NUDIX hydrolase family, we then turned our attention to PARP1 and PARP inhibitors, which represent one of the most successful targeted therapies for cancer treatment, particularly BRCA-defective breast and ovarian cancers^{57 59}. The interest in PARP biology and PARP-targeted therapies underscores its relevance as a cancer therapy and in general as a drug target. Several PARP inhibitors are clinically available, such as olaparib, but the field was almost completely shuttered by the failure of iniparib (BSI 201 ), which was later found to not inhibit PARP at all⁶⁰. Iniparib was later confirmed to not bind PARP1 in cells using CETSA and served as a proof of concept in the validation and utility of CETSA⁵.

Many naturally occurring and synthetic missense mutations of PARP1 have been described and characterized, both in vitro and in cells⁶¹. The PARP1 L713F mutation was identified as a gain-of-function mutation by random mutagenesis and is constitutively active, even in the absence of DNA, although, this activity is much less than that seen with WT PARP1 stimulated with H2O2^{62 65}. Interestingly, the L713F mutation is also destabilizing and the expression of the protein is noticeably less than other overexpressed PARP mutants, even when fused to eGFP^{63 65}. Rank et al. also demonstrated that the effects of PARP1 L713F expression are completely rescued by treatment with the PARP inhibitor, ABT-888 (veliparib), including NAD+ depletion and cell death⁶⁵. This indicates that the integrity of the L713F active site likely remains intact - a hypothesis supported by the fact that L713F was originally purified with immobilized 3-aminobenzamide (3-AB), a PARP inhibitor⁶². Therefore, PARP1 L713F is a provoking and clinically relevant candidate for CeTEAM analysis.

Thus, we evaluated PARP1 L713F fused to eGFP in U-2 OS osteosarcoma cells driven with a doxycycline-inducible promoter as a potential system for CeTEAM (Figure 7a). Initially, we incubated these cells, as well as ones expressing WT PARP1 -GFP, with DMSO, olaparib, or MG-132 (Figure 7b). We could confirm the previous report that L713F expressed poorly but also saw that exposure to olaparib increased the amount of mutant fusion protein. Interestingly, MG-132 did not appreciably affect mutant accumulation, which may reflect that PARP1 can be degraded by both the proteasome and autophagy systems⁶⁶·⁶⁷. A variety of known PARP1 inhibitors, including olaparib (AZD2281 ), talazoparib (BMN 673), niraparib (MK-4827) and iniparib were then tested in the same cell line (Figure 7c). All inhibitors, except for iniparib, caused robust accumulation of L713F- GFP (but had no effect on WT expression), indicating binding and stabilization of the protein surrogate. Lack of response with iniparib is in line with previous studies showing iniparib does not bind PARP1⁵. Stabilization of PARP1 L713F-GFP by olaparib was also time- and dose-dependent (Figure 7d), as it was for NUDT 15 and MTH1.

Clinical PARP inhibitors are known to trap PARP1/2 to DNA and increase DNA replication stress as part of their effective mechanism of action⁶⁸·⁶⁹. Therefore, we also blotted for a marker of replication stress (phosphorylated Chk1 ) in the same samples and could see that all of these inhibitors increased the levels of phospho-Chk1 (even more so in the PARP1 L713F cells; Figure 7c). Replication stress was also time- and dose-dependent, where stabilization of L713F-GFP occurs prior to Chk1 phosphorylation (Figure 7d).

A unique advantage of L713F-GFP is the possibility of tracking cellular target engagement by GFP fluorescence in live cells. We then utilized high-content, live cell microscopy following incubation of cells with a dose response of the PARP inhibitors used previously (along with veliparib and 3-aminobenzamide) and tracked cell cycle with cell permeable Hoechst 33342 and nuclear PARP1 L713F-GFP fluorescence (Figure 7e-i). Median GFP saturation curves demonstrated that most of the inhibitors had target engagement ECso values on the order of 50 nM or lower with talazoparib being the most potent, 3-AB the least potent (EC₅₀ ~ 200 mM) and iniparib having no effect on GFP signal (Figure 7e). The median DNA content, which is used here to indicate PARP trapping and PARP-induced replication stress, gave a gradient when comparing all of the PARP inhibitors (Figure 7f). These trends followed the reported relative biochemical IC₅₀ values and PARP trapping ability for each inhibitor very closely (Figure 7g).

Indeed, when visualizing both parameters, dose-dependently for each PARP inhibitor on the population level, clear trends are readily apparent that relate target engagement potency with PARP trapping ability (Figure 7h). Talazoparib is extremely potent for both target engagement and trapping ability, while olaparib and niraparib are less potent but still effectively trap. Interestingly, veliparib is just as potent as olaparib or niraparib, but trapping is non-linear and is only noticeable at concentrations above 400 nM. 3-AB, while far less potent than the clinically used PARP inhibitors, was still effective at stabilizing PARP1 L713F-GFP, but resulted in a G1 block, as has been reported previously⁷⁰. Once again, iniparib affected neither GFP signal nor DNA content, supporting previously claims that it does not bind to PARP1 in cells. Viewing these factors at the single cell level further highlights the diversity of available PARP inhibitors and demonstrates the true power of CeTEAM to relate cellular target engagement to induced phenotypic changes (Figure 7i). Discussion

We have presented CeTEAM as a viable technology to simultaneously detect cellular target engagement and phenotypic feedback following exposure of cells to target-specific ligands. CeTEAM works by identifying ligands that impair the rapid turnover of variant target proteins that contain destabilizing missense mutations or are fused to degron domains. Within this method, the measurement of tagged variant accumulation in response to a ligand occurs under physiological conditions. To our knowledge, this is the first cellular target engagement approach that directly integrates the detection of phenotypic responses with the detection of ligand binding. Using other approaches, it may be possible to derive information pertaining to ligand binding under physiological conditions (e.g., with CETSA⁷¹), but the CeTEAM method provides this information on a per-cell basis; akin to multiplexing antibody or fluorophore detection.

The mutations that proved to be useful in developing CeTEAM for NUDT15, MTH1 , and PARP1 arose from very different study analyses. The NUDT15 R139C mutation was recognized as a causative link to thiopurine sensitivity from genome-wide association studies; MTH1 G48E was identified from mutational screening, where a proposed link between MTH1 germline mutations and hereditary forms of colorectal cancer were investigated; and PARP1 L713F was identified in an attempt to study PARP1 structure- function relationships. It is hoped that this new use for loss-of-stability mutations will spark further interest in this field and will help to identify mutations that can be used within the CeTEAM framework.

In addition to mutation origin, we have demonstrated the successful use of CeTEAM with a variety of proteins and fusion tags. While NUDT15 and MTH1 belong to the same protein family, they have distinct substrates, functions and structures; in addition, their functionality in CeTEAM was exemplified using two different epitope tags. PARP1 , meanwhile, is a much larger protein with DNA binding capabilities, and was fused to eGFP to generate a PARP1 CeTEAM system capable of live cell detection. In each case, the treatment of cells with ligands had no effect on variant mRNA transcript levels, which supports the notion that accumulation of variant protein is purely a result of binding and stabilization rather than increased transcription. However, the applicability of different fusion tags is not ubiquitous, as the fusion of eGFP or nanoLuc to NUDT15 R139C or MTH1 G48E resulted in complete stabilization of these variants in the absence of inhibitors. The size and stability of a fusion tag can influence the stability of the fusion protein; therefore, the nature, and extent of destabilization of missense mutations, or the strength of the degron, will likely dictate the choice of fusion tag⁷². The variety of available fusion tags imparts the potential to have cell lines expressing multiple tagged variant proteins and run CeTEAM for multiple targets simultaneously in a single population of cells. This could be especially useful for exploring selectivity of a series of inhibitors between two (or more) cellular targets and could be done in a high-throughput manner.

A caveat regarding the feasibility of CeTEAM as a target engagement platform with broad applicability is the ability to identify destabilizing mutations or suitable degron fusions that will enable accumulation in the presence of a ligand. While not applicable to degron-based CeTEAM approaches, destabilizing missense mutations are often identified serendipitously due to their linkage as causative agents responsible for particular diseases^{73 74} or by structure-functional studies, but most proteins, including many viewed as attractive drug targets, have no published accounts of destabilizing mutations. This being said, several methodologies already exist that can mitigate this issue.

With increasing computing power and machine learning capacities, in silico computational platforms can now predict mutation-induced stability changes with increasing speed and accuracy⁷³·⁷⁵. While we have used these methods to predict stability changes in NUDT15 protein after identifying relevant mutations for CeTEAM analyses, it is suspected that these computational methods could also be used a priori for proteins of interest with known structures.

Strategies to modulate protein stability to facilitate protein crystallography have existed for many years. Although in most cases the intent is to increase stability, one can use these same principles to identify destabilizing mutations. To screen for destabilizing mutations in protein(s) of interest (even without structural information available), random mutagenesis within the open reading frame (ORF) of the gene of interest can be performed. This can be accomplished by screening for destabilizing mutations with bacterial cultures, similarly as described by Asial et a!.²⁴. Instead of sequencing ORFs whose protein was still soluble above 37°C, sequencing of ORFs that aggregated at 37°C or below would be of interest to identify destabilizing mutations.

A similar methodology in eukaryotic cells, which would likely be more practical for screening of CeTEAM candidates, was developed by Banaszynski et al.¹⁵. Here, random mutagenesis was performed within the ligand binding domain (LBD) before fusing to eGFP in a mammalian expression vector. The authors then screened for clones that fluoresced upon ligand exposure, then lost fluorescence following ligand removal and once again tested for increased fluorescence in the presence of ligand by consecutive cell sortings by FACS. Clones that met these criteria were then sequenced to identify destabilizing mutations that could be stabilized by binding of a ligand. These LBDs were then fused to other genes of interest and used to tightly regulate expression of the protein of interest dependent on the ligand being added to the cells.

A further consideration with CeTEAM is that in some cases the variant and WT POI may have different affinities for small molecule inhibitors. In which case, it would still be required to look at WT target engagement with other techniques. Thus, instead of replacing existing thermal stability methodologies (CETSA and TPP), we believe that CeTEAM could be a valuable addition to this field and would complement these techniques to rapidly identify, optimize and study the effects of small molecule inhibitors for disease-relevant targets.

While identifying appropriate tagged mutants for CeTEAM requires more initial development than other target engagement techniques, such as CETSA, the flexibility, portability and translatable power of this technique should justify such investments. The nature of the technology is streamlined in such a way that the same biological tools can be employed in each stage of the drug discovery and development process; only requiring a cell line with a tagged, variant POI that is stabilized upon inhibitor binding. Starting with high-throughput screening campaigns, accumulation of the protein tag is quantified using an applicable detection method, depending on the nature of the protein tag. Moving forward, hit compounds from the high-throughput screen can be quickly and inexpensively analyzed in dose-response experiments, where target engagement can also be multiplexed with other phenotypic markers that are related to the target of interest to provide high-content data regarding potential lead compounds. Once lead agents are identified, in vivo studies can provide unprecedented information about preclinical candidates. Xenograft studies can assess target engagement and phenotypic responses in intact tissues or following excision and processing. Another exciting possibility is whole- animal drug distribution studies using CeTEAM transgenic animal models containing fusions to fluorescent or bioluminescent reporters. Thus, CeTEAM can facilitate drug discovery and development from early high-throughput screening campaigns, to mechanism of action studies, to lead discovery and even preclinical in vivo testing.

References

1. Arrowsmith, C.H. et al. The promise and peril of chemical probes. Nature

Chemical Biology 11 , 536-541 (2015).

2. Bunnage, M.E., Gilbert, A.M., Jones, L.H. & Hett, E.C. Know your target, know your molecule. Nature Chemical Biology 11 , 368-372 (2015).

3. Bunnage, M.E., Chekler, E.L.P. & Jones, L.H. Target validation using chemical probes. Nature Chemical Biology 9, 195-199 (2013). 4. Jafari, R. et al. The cellular thermal shift assay for evaluating drug target interactions in cells. Nature Protocols 9, 2100-2122 (2014).

5. Molina, D.M. et al. Monitoring Drug Target Engagement in Cells and Tissues Using the Cellular Thermal Shift Assay. Science 341 , 84-87 (2013).

6. Almqvist, H. et al. CETSA screening identifies known and novel thymidylate

synthase inhibitors and slow intracellular activation of 5-fluorouracil. Nat Commun 7, 1 1040 (2016).

7. Franken, H. et al. Thermal proteome profiling for unbiased identification of direct and indirect drug targets using multiplexed quantitative mass spectrometry. Nat Protoc 10, 1567-93 (2015).

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

9. Carney, B. et al. Target engagement imaging of PARP inhibitors in small-cell lung cancer. Nat Commun 9, 176 (2018).

10. Li, G. et al. An activity-dependent proximity ligation platform for spatially resolved quantification of active enzymes in single cells. Nat Commun 8, 1775 (2017).

1 1. Ohoka, N. et al. In Vivo Knockdown of Pathogenic Proteins via Specific and

Nongenetic Inhibitor of Apoptosis Protein (lAP)-dependent Protein Erasers (SNIPERs). J Biol Chem 292, 4556-4570 (2017).

12. Vasta, J.D. et al. Quantitative, Wide-Spectrum Kinase Profiling in Live Cells for Assessing the Effect of Cellular ATP on Target Engagement. Cell Chem Biol (2017).

13. Chessum, N.E.A. et al. Demonstrating In-Cell Target Engagement using a Pirin Protein Degradation Probe (CCT367766). J Med Chem (2017).

14. Dubach, J.M. et al. Quantitating drug-target engagement in single cells in vitro and in vivo. Nat Chem Biol 13, 168-173 (2017).

15. Dubach, J.M. et al. In vivo imaging of specific drug-target binding at subcellular resolution. Nature Communications 5, 3946 (2014).

16. Feng, J. et al. A general strategy to construct small molecule biosensors in

eukaryotes. Elite 4(2015).

17. Iwamoto, M., Bjdrklund, T., Lundberg, C., Kirik, D. & Wandless, T.J. A general chemical method to regulate protein stability in the mammalian central nervous system. Chem Biol 17, 981-8 (2010).

18. Banaszynski, L.A., Sellmyer, M.A., Contag, C.H., Wandless, T.J. & Thorne, S.H.

Chemical control of protein stability and function in living mice. Nat Med 14, 1 123- 7 (2008). 9. Banaszynski, L.A., Chen, L.C., Maynard-Smith, L.A., Ooi, A.G. & Wandless, T.J.

A rapid, reversible, and tunable method to regulate protein function in living cells using synthetic small molecules. Cell 126, 995-1004 (2006).

0. Stankunas, K. et al. Conditional protein alleles using knockin mice and a chemical inducer of dimerization. Mol Cell 12, 1615-24 (2003).

1. Levy, F., Johnston, J.A. & Varshavsky, A. Analysis of a conditional degradation signal in yeast and mammalian cells. Eur J Biochem 259, 244-52 (1999).

2. Park, E.C., Finley, D. & Szostak, J.W. A strategy for the generation of conditional mutations by protein destabilization. Proc Natl Acad Sci U S A 89, 1249-52 (1992).

3. Egeler, E.L., Urner, L.M., Rakhit, R„ Liu, C.W. & Wandless, T.J. Ligand- switchable substrates for a ubiquitin-proteasome system. J Biol Chem 286, 31328-36 (201 1 ).

24. Asial, I. et al. Engineering protein thermostability using a generic activity- independent biophysical screen inside the cell. Nature Communications 4, 2901 (2013).

25. Bachmair, A., Finley, D. & Varshavsky, A. In vivo half-life of a protein is a function of its amino-terminal residue. Science 234, 179-86 (1986).

26. Koren, I. et al. The Eukaryotic Proteome Is Shaped by E3 Ubiquitin Ligases

Targeting C-Terminal Degrons. Cell 173, 1622-1635.e14 (2018).

27. Lin, H.C. et al. C-Terminal End-Directed Protein Elimination by CRL2 Ubiquitin Ligases. Mol Cell 70, 602-613.e3 (2018).

28. Bondeson, D.P. et al. Catalytic in vivo protein knockdown by small-molecule

PROTACs. Nat Chem Biol 11, 611-7 (2015).

29. Sakamoto, K.M. et al. Protacs: chimeric molecules that target proteins to the

Skp1 -Cullin-F box complex for ubiquitination and degradation. Proc Natl Acad Sci U S A 98, 8554-9 (2001 ).

30. Dohmen, R.J., Wu, P. & Varshavsky, A. Heat-inducible degron: a method for constructing temperature-sensitive mutants. Science 263, 1273-6 (1994).

31. Varshavsky, A. The N-end rule. Cell 69, 725-35 (1992).

32. Varshavsky, A. The N-end rule pathway and regulation by proteolysis. Protein Sci

20, 1298-345 (201 1 ).

33. Bachmair, A. & Varshavsky, A. The degradation signal in a short-lived protein.

Cell 56, 1019-32 (1989).

34. Johnston, J.A., Johnson, E.S., Waller, P.R. & Varshavsky, A. Methotrexate

inhibits proteolysis of dihydrofolate reductase by the N-end rule pathway. J Biol Chem 270, 8172-8 (1995). 35. Banaszynski, L.A. & Wandless, T.J. Conditional control of protein function. Chem Biol 13, 1 1 -21 (2006).

36. Moriyama, T. et al. NUDT15 polymorphisms alter thiopurine metabolism and hematopoietic toxicity. Nature Genetics 48, 367-373 (2016).

37. Valerie, N.C.K. et al. NUDT15 Hydrolyzes 6-Thio-DeoxyGTP to Mediate the Anticancer Efficacy of 6-Thioguanine. Cancer Research 76, 5501-551 1 (2016).

38. Kakuta, Y. et al. NUDT 15 R139C causes thiopurine-induced early severe hair loss and leukopenia in Japanese patients with IBD. The Pharmacogenomics Journal (2015).

39. Yang, J.J. et al. Inherited NUDT15 Variant Is a Genetic Determinant of

Mercaptopurine Intolerance in Children With Acute Lymphoblastic Leukemia. Journal of Clinical Oncology 33, 1235-1242 (2015).

40. Yang, S.-K. et al. A common missense variant in NUDT15 confers susceptibility to thiopurine-induced leukopenia. Nature Genetics 46, 1017-1020 (2014).

41. Moriyama, T. et al. NUDT15 polymorphisms alter thiopurine metabolism and hematopoietic toxicity. Nat Genet advance online publication(2016).

42. Carter, M. et al. Crystal structure, biochemical and cellular activities demonstrate separate functions of MTH1 and MTH2. Nature Communications 6, 7871 (2015).

43. Page, B.D.G. et al. Targeted NUDT5 inhibitors block hormone signaling in breast cancer cells. Nat Commun 9, 250 (2018).

44. Hawn, M.T. et al. Evidence for a Connection between the Mismatch Repair System and the G2 Cell Cycle Checkpoint. Cancer Research 55, 3721-3725 (1995).

45. Swann, P.F. et al. Role of postreplicative DNA mismatch repair in the cytotoxic action of thioguanine. Science 273, 1109-1 1 (1996).

46. Yan, T., Berry, S.E., Desai, A.B. & Kinsella, T.J. DNA Mismatch Repair (MMR) Mediates 6-Thioguanine Genotoxicity by Introducing Single-strand Breaks to Signal a G2-M Arrest in MMR-proficient RKO Cells. Clinical Cancer Research 9, 2327-2334 (2003).

47. Yamane, K., Taylor, K. & Kinsella, T.J. Mismatch repair-mediated G2/M arrest by 6-thioguanine involves the ATR-Chk1 pathway. Biochemical and Biophysical Research Communications 318, 297-302 (2004).

48. Yan, T. et al. CHK1 and CHK2 are differentially involved in mismatch repair- mediated 6-thioguanine-induced cell cycle checkpoint responses. Molecular Cancer Therapeutics 3, 1 147-1157 (2004).

49. Gad, H. et al. MTH1 inhibition eradicates cancer by preventing sanitation of the dNTP pool. Nature 508, 215-221 (2014). 50. Ellermann, M. et al. Novel Class of Potent and Cellularly Active Inhibitors

Devalidates MTH1 as Broad-Spectrum Cancer Target. ACS Chem Biol 12, 1986- 1992 (2017).

51. Petrocchi, A. et al. Identification of potent and selective MTH1 inhibitors. Bioorg Med Chem Lett 26, 1503-1507 (2016).

52. Kettle, J.G. et al. Potent and Selective Inhibitors of MTH1 Probe Its Role in

Cancer Cell Survival. J Med Chem 59, 2346-61 (2016).

53. Mur, P. et al. Germline variation in the oxidative DNA repair genes NUDT1 and OGG1 is not associated with hereditary colorectal cancer or polyposis. Hum Mutat (2018).

54. Kawamura, T. et al. Proteomic profiling of small-molecule inhibitors reveals

dispensability of MTH1 for cancer cell survival. Sci Rep 6, 26521 (2016).

55. Pudelko, L. et al. Glioblastoma and glioblastoma stem cells are dependent on functional MTH1. Oncotarget 8, 84671-84684 (2017).

56. Gad, H. et al. MTH1 inhibition eradicates cancer by preventing sanitation of the dNTP pool. Nature 508, 215-221 (2014).

57. Fong, P.C. et al. Inhibition of poly(ADP-ribose) polymerase in tumors from BRCA mutation carriers. N Engl J Med 361, 123-34 (2009).

58. Bryant, H.E. et al. Specific killing of BRCA2-deficient tumours with inhibitors of poly(ADP-ribose) polymerase. Nature 434, 913-7 (2005).

59. Farmer, H. et al. Targeting the DNA repair defect in BRCA mutant cells as a

therapeutic strategy. Nature 434, 917-21 (2005).

60. Mateo, J., Ong, M., Tan, D.S.P., Gonzalez, M.A. & de Bono, J.S. Appraising

iniparib, the PARP inhibitor that never was— what must we learn? Nature

Reviews Clinical Oncology 10, 688-696 (2013).

61. Pettitt, S. J. et al. Genome-wide and high-density CRISPR-Cas9 screens identify point mutations in PARP1 causing PARP inhibitor resistance. Nat Commun 9, 1849 (2018).

62. Miranda, E.A., Dantzer, F., O'Farrell, M., de Murcia, G. & de Murcia, J.M.

Characterisation of a gain-of-function mutant of poly(ADP-ribose) polymerase. Biochem Biophys Res Commun 212, 317-25 (1995).

63. Langelier, M.F., Planck, J.L., Roy, S. & Pascal, J.M. Structural basis for DNA damage-dependent poly(ADP-ribosyl)ation by human PARP-1. Science 336, 728- 32 (2012).

64. Dawicki-McKenna, J.M. et al. PARP-1 Activation Requires Local Unfolding of an Autoinhibitory Domain. Mol Cell 60, 755-768 (2015). 5. Rank, L. et al. Analyzing structure-function relationships of artificial and cancer- associated PARP1 variants by reconstituting TALEN-generated HeLa PARP1 knock-out cells. Nucleic Acids Res 44, 10386-10405 (2016).

66. Zhang, Y„ Liao, X.H., Xie, H.Y., Shao, Z.M. & Li, D.Q. RBR-type E3 ubiquitin ligase RNF144A targets PARP1 for ubiquitin-dependent degradation and regulates PARP inhibitor sensitivity in breast cancer cells. Oncotarget 8, 94505- 94518 (2017).

67. Wang, Y. et al. Gemcitabine induces poly (ADP-ribose) polymerase-1 (PARP-1 ) degradation through autophagy in pancreatic cancer. PLoS One 9, e109076 (2014).

68. Murai, J. et al. Trapping of PARP1 and PARP2 by Clinical PARP Inhibitors.

Cancer Res 72, 5588-99 (2012).

69. Maya-Mendoza, A. et al. High speed of fork progression induces DNA replication stress and genomic instability. Nature 559, 279-284 (2018).

70. De Blasio, A. et al. Differentiative pathway activated by 3-aminobenzamide, an inhibitor of PARP, in human osteosarcoma MG-63 cells. FEBS Lett 579, 615-20 (2005).

71. Seashore-Ludlow, B. & Lundback, T. Early Perspective Microplate Applications of the Cellular Thermal Shift Assay (CETSA). Journal of Biomolecular Screening, 10870571 16659256 (2016).

72. Kimple, M.E., Brill, A.L. & Pasker, R.L. Overview of affinity tags for protein

purification. Curr Protoc Protein Sci 73, Unit 9.9 (2013).

73. Stefl, S., Nishi, H., Petukh, M., Panchenko, A.R. & Alexov, E. Molecular

mechanisms of disease-causing missense mutations. J Mol Biol 425, 3919-36 (2013).

74. Redler, R.L., Das, J., Diaz, J.R. & Dokholyan, N.V. Protein Destabilization as a Common Factor in Diverse Inherited Disorders. J Mol Evol 82, 11-6 (2016).

75. Yin, S., Ding, F. & Dokholyan, N.V. Eris: an automated estimator of protein

stability. Nat Methods 4, 466-7 (2007). Example 2; Further experimental data.

Degron fusion proteins and the CeTEAM technology

Abstract

Novel cellular target engagement technologies have impacted the ways in which we identify and evaluate promising drug candidates. Our technique - CeTEAM (Cellular Target Engagement by Accumulation of Mutants) - can simultaneously evaluate drug- target interactions and phenotypic responses in cells. This approach was predicated upon the stabilization and subsequent accumulation of an unstable missense mutant as a drug biosensor for specific proteins of interest. Here, we extend this methodology by way of degron fusion proteins and demonstrate their ability to further simplify and streamline the application of CeTEAM technology for ligand discovery and development projects.

Introduction

Confirming cellular target engagement has become a determining factor in the value of probe molecules and the efficacy of clinically used drugs (1-3). Until recently, detecting and quantifying cellular target engagement in a straightforward and comprehensive manner had been a major hurdle within drug development. With the advent of the cellular thermal shift assay (CETSA) (4,5), drug discovery scientists, both in academic or industry settings, can evaluate drug-target interactions with relative ease. From this stemmed thermal proteome profiling (TPP), which can also be used to identify unknown targets for established drugs6, as well as help to understand selectivity of a given ligand (7,8). Indeed, the successes and limitations of CETSA and TPP have inspired the recent development of several additional target engagement technologies; including methods to quantify target engagement in single cells and in vivo (9-15).

Modulating cellular protein stability also has applications beyond thermal shift assays. Within the biosensors field, unstable protein variants are utilized to detect the presence of specific chemicals, metabolites or even drug molecules (16). In a similar fashion, modifying a protein of interest (POI) with an engineered/unstable ligand binding domain (LBDs) has been used to rapidly and conditionally regulate protein levels in a tunable, reversible manner, which is particularly valuable when investigating protein function (17- 22). In this setting, a POI is ectopically expressed as a fusion protein attached to an unstable LBD. This fusion protein is inherently unstable and will be proteolytically degraded in cells unless the LBD is stabilized by binding to specific ligands (23). An additional consideration in modulating protein degradation comes in the form of degrons, which consist of N- or C-terminal protein fragments that act as targets for ubiquitylation (24-26).

Degrons regulate intrinsic protein turnover and their fusion to small molecules has created a new class of promising therapeutics, known as PROTACs (27,28). Cryptic degrons may be exposed upon heating or partial unfolding of the particular domain (29), thus permitting ubiquitylation of key lysine residues (30); in turn, it is likely that many destabilizing missense mutations cause rapid proteolytic degradation in a similar fashion (31 ). As such, fusing degron sequences to the termini of stable POIs results in conditional depletion in cells (24-26,29,30). Like traditional unstable LBDs, POIdegron fusion proteins (e.g., FKBP12-L106P:POI (19), C-b-gaLPOI (24,32), Arg-DHFR:POI (29,32) or RARD1 :POI (22), among others (16,30)), can be stabilized in the presence of a ligand bound to the LBD (19,21 ,33) or POI (16), which has also proven useful in the study of protein functions (34) and increasing tunability of LBD-based biosensors (16). Despite the many innovations propelled by modulating protein stability in live cells, an underappreciated application may be in measuring intracellular target engagement of small drug-like molecules.

Here, we complement the advent of CeTEAM by destabilizing missense mutations and demonstrate that a wild-type MTH1-nMyc degron fusion is stabilized by an MTH1 inhibitor. This study provides further demonstration that wild-type POI-degron fusions can be stabilized by ligands binding to the POI, and their application in the CeTEAM technology.

Results

MTH1 fused to a degron from nMyc is stabilized by an MTH1 inhibitor

Degrons are generally short, lysine-containing peptide sequences typically found at the N- or C-termini of proteins that dictate ubiquitin-dependent protein degradation and turnover (24,25). They are portable sequences that offer selective, tunable degradation of proteins and can be conditionally“hidden” by structural realignments (e.g., upon binding of a ligand), resulting in their inactivation (16,19,33).

Our experience of applying CeTEAM with unstable missense mutants, indicated that a wild-type POI fused to a degron motif can be stabilized by a ligand bound to the wild-type POI. The increased rate of proteolytic degradation due to the degron can be counteracted by stabilizing the POI with a ligand, thus shifting the half-life of the fusion protein and causing its apparent accumulation in the cellular environment. To further demonstrate this, we designed a novel entry vector cassette that simplifies the testing of degron fusions in cells and is based off of a similar approach used by Elledge and colleagues to identify novel C-terminal degron sequences (Figure 9a)(25). The all-in- one cassette features an iRFP670 fluorescent protein uncoupled from expression of the eGFP-POI-degron fusion protein by an IRES sequence, therefore facilitating normalization of the degron fusion stability. We then decided to evaluate stabilization of wild-type MTH1 by MTH1 inhibitors in U-2 OS cells with a recently described C-terminal degron sequence for nMyc (LEKEKLQARQQQLLKKIEHARTC) that routes proteolytic degradation via the E3 adaptor, TRPC4AP, and the Cullin4 complex (Figure 9a, b) (25,35). When induced with doxycycline, the GFP-MTH1-nMyc degron was detectable by western blot. Interestingly, incubation of these cells with Cmpd 6 resulted in a marked accumulation of the GFP-MTH1 -degron fusion, suggesting that binding of the inhibitor to wild-type MTH1 protects the fusion protein from degradation via the nMyc degron (Figure 9c, d).

Discussion

We have described that unstable missense variants of proteins can be stabilized in cells by their cognate small molecule inhibitors and that their accumulation can be utilized as drug biosensors to understand drug binding and mechanism of action - referred to as the CeTEAM platform.

As discussed in Example 1 , the CeTEAM platform was used to evaluate NUDT15, MTH1 , and PARP1 inhibitors in cells, which made use of destabilizing missense variants.

Destabilizing missense mutations are often discovered serendipitously as causative agents in human diseases or in structure-function surveys (36,37), so for most of the proteome, including attractive drug targets, there is not yet published accounts of such variants. There are, however, established and emerging approaches that can used instead - such as in silico modelling predictions (36,38) and random mutagenesis screens, both in vitro and in cells (19,39), although such approaches are associated with time investments to identify and validate novel mutants.

There are other reasons why an alternative to using missense mutants for CeTEAM could improve the system further. For one, the mutation may affect the structure of the ligand binding site. While these effects may be subtle, they could have implications on ligand binding and the sensitivity with which the unstable variant can be used as surrogate drug biosensor for the wild-type protein. This possibility would require further structural evaluation and comparison of ligand binding to the wild-type protein. Furthermore, despite their appreciatively low functional levels in cells, unstable missense variants could potentially possess ulterior or dominant negative functions that are not inherent to the wild- type protein - such is the case with the constitutive, DNA independent activity of PARP1 L713F (40-44), which could be difficult to identify and potentially affect the sensitivity of the CeTEAM system.

Here, we have demonstrated an attractive modification for CeTEAM - that a POI-degron fusion protein can be stabilized in cells by binding of a cognate ligand to the POI. In this instance, we utilized wild-type, p18 MTH1 fused to GFP and an nMyc degron motif; however, there are likely to be numerous degron sequences that would work equally well and their efficacy as degraders would be POI- and cell-type-specific. We envisage a scenario where mining and evaluation of curated degron sequences from larger surveys would generate a library of degraders varying in their“degradation strength”. For example, a handful of C-terminal degron sequences that function via the CRL E3 adaptors have recently been described, among many others (Tables 3-6) (25). In this way, selection of degrons could be tailored to the stabilities of individual POIs/tags and account for variation of expression patterns in different cell types.

By and large, the degron aspect for modulating protein stability could further improve the CeTEAM system. Since this would employ wild-type proteins, any effects on protein structure and function of missense variants, as well as the time and monetary investment to identify novel destabilizing mutations, would no longer apply. Thus, these advantages should further improve the application of CeTEAM technology to a larger POI cohort and facilitate establishment of the technique for ligand discovery and development projects in laboratories around the world.

References (Example 2)

1. Arrowsmith, C.H. et al. The promise and peril of chemical probes. Nature Chemical Biology 11 , 536-541 (2015).

3. Bunnage, M.E., Chekler, E.L.P. & Jones, L.H. Target validation using chemical probes. Nature Chemical Biology 9, 195-199 (2013).

4. Jafari, R. et al. The cellular thermal shift assay for evaluating drug target interactions in cells. Nature Protocols 9, 2100-2122 (2014). . Molina, D.M. et al. Monitoring Drug Target Engagement in Cells and Tissues Using the Cellular Thermal Shift Assay. Science 341 , 84-87 (2013).

. Almqvist, H. et al. CETSA screening identifies known and novel thymidylate synthase inhibitors and slow intracellular activation of 5-fluorouracil. Nat Commun 7, 1 1040 (2016).

. Franken, H. et al. Thermal proteome profiling for unbiased identification of direct and indirect drug targets using multiplexed quantitative mass spectrometry. Nat Protoc 10, 1567-93 (2015).

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

. Carney, B. et al. Target engagement imaging of PARP inhibitors in small-cell lung cancer. Nat Commun 9, 176 (2018).

1 1. Ohoka, N. et al. In Vivo Knockdown of Pathogenic Proteins via Specific and Nongenetic Inhibitor of Apoptosis Protein (lAP)-dependent Protein Erasers (SNIPERs). J Biol Chem 292, 4556-4570 (2017).

16. Feng, J. et al. A general strategy to construct small molecule biosensors in eukaryotes. Eiife 4(2015).

17. Iwamoto, M., Bjorklund, T., Lundberg, C., Kirik, D. & Wandless, T.J. A general chemical method to regulate protein stability in the mammalian central nervous system. Chem Biol 17, 981 -8 (2010).

Chemical control of protein stability and function in living mice. Nat Med 14, 1 123- 7 (2008).

19. Banaszynski, L.A., Chen, L.C., Maynard-Smith, L.A., Ooi, A.G. & Wandless, T.J. A rapid, reversible, and tunable method to regulate protein function in living cells using synthetic small molecules. Cell 126, 995-1004 (2006). 20. Stankunas, K. et al. Conditional protein alleles using knockin mice and a chemical inducer of dimerization. Mol Cell 12, 1615-24 (2003).

21. Levy, F., Johnston, J.A. & Varshavsky, A. Analysis of a conditional degradation signal in yeast and mammalian cells. Eur J Biochem 259, 244-52 (1999).

22. Park, E.C., Finley, D. & Szostak, J.W. A strategy for the generation of conditional mutations by protein destabilization. Proc Natl Acad Sci U S A 89, 1249-52 (1992).

23. Egeler, E.L., Urner, L.M., Rakhit, R., Liu, C.W. & Wandless, T.J. Ligand-switchable substrates for a ubiquitin-proteasome system. J Biol Chem 286, 31328-36 (2011 ).

24. Bachmair, A., Finley, D. & Varshavsky, A. In vivo half-life of a protein is a function of its amino-terminal residue. Science 234, 179-86 (1986).

25. Koren, I. et al. The Eukaryotic Proteome Is Shaped by E3 Ubiquitin Ligases Targeting CTerminal Degrons. Cell 173, 1622-1635.e14 (2018).

26. Lin, H.C. et al. C-Terminal End-Directed Protein Elimination by CRL2 Ubiquitin Ligases. Mol Cell 70, 602-613.e3 (2018).

27. Bondeson, D.P. et al. Catalytic in vivo protein knockdown by small-molecule PROTACs. Nat Chem Biol 11 , 61 1-7 (2015).

28. Sakamoto, K.M. et al. Protacs: chimeric molecules that target proteins to the Skp1- Cullin-F box complex for ubiquitination and degradation. Proc Natl Acad Sci U S A 98, 8554-9 (2001 ).

29. Dohmen, R.J., Wu, P. & Varshavsky, A. Heat-inducible degron: a method for constructing temperature-sensitive mutants. Science 263, 1273-6 (1994).

30. Varshavsky, A. The N-end rule. Cell 69, 725-35 (1992).

31. Varshavsky, A. The N-end rule pathway and regulation by proteolysis. Protein Sci

20, 1298-345 (201 1 ).

32. Bachmair, A. & Varshavsky, A. The degradation signal in a short-lived protein. Cell 56, 1019-32 (1989).

33. Johnston, J.A., Johnson, E.S., Waller, P.R. & Varshavsky, A. Methotrexate inhibits proteolysis of dihydrofolate reductase by the N-end rule pathway. J Biol Chem 270, 8172-8 (1995).

34. Banaszynski, L.A. & Wandless, T.J. Conditional control of protein function. Chem Biol 13, 1 1-21 (2006).

35. Choi, S.H., Wright, J.B., Gerber, S.A. & Cole, M.D. Myc protein is stabilized by suppression of a novel E3 ligase complex in cancer cells. Genes Dev 24, 1236-41 (2010).

36. Stefl, S., Nishi, H., Petukh, M., Panchenko, A.R. & Alexov, E. Molecular mechanisms of disease-causing missense mutations. J Mol Biol 425, 3919-36 (2013). 37. Redler, R.L., Das, J., Diaz, J.R. & Dokholyan, N.V. Protein Destabilization as a Common Factor in Diverse Inherited Disorders. J Mol Evol 82, 11-6 (2016).

38. Yin, S., Ding, F. & Dokholyan, N.V. Eris: an automated estimator of protein stability.

Nat Methods 4, 466-7 (2007).

39. Asial, I. et al. Engineering protein thermostability using a generic activity- independent biophysical screen inside the cell. Nature Communications 4, 2901 (2013).

40. Pettitt, S.J. et al. Genome-wide and high-density CRISPR-Cas9 screens identify point mutations in PARP1 causing PARP inhibitor resistance. Nat Commun e, 1849 (2018).

41. Miranda, E.A., Dantzer, F., O'Farrell, M., de Murcia, G. & de Murcia, J.M.

42. Langelier, M.F., Planck, J.L., Roy, S. & Pascal, J.M. Structural basis for DNA damagedependent poly(ADP-ribosyl)ation by human PARP-1. Science 336, 728-

32 (2012).

43. Dawicki-McKenna, J.M. et al. PARP-1 Activation Requires Local Unfolding of an Autoinhibitory Domain. Mol Cell 60, 755-768 (2015).

44. Rank, L. et al. Analyzing structure-function relationships of artificial and cancerassociated PARP1 variants by reconstituting TALEN-generated HeLa

PARP1 knock-out cells. Nucleic Acids Res 44, 10386-10405 (2016).

Table 3

Table 4

Table 5

Table 6

Claims

1. A method for identifying an agent that binds to, and modulates one or more activity of, a target protein, comprising the steps of:

i. providing an agent to be tested;

iii. contacting the one or more cell with the agent to be tested;

iv. determining the stability of the variant; and

v. determining one or more activity of the target protein.

2. The method of Claim 1 further comprising the step of:

determining whether the test agent is one that binds to, and modulates one or more activity of, the target protein, on the basis of the determinations in steps (iv) and (v).

3. The method of any of Claims 1 and 2 wherein the agent is identified as one that binds to, and modulates one or more activity of, the target protein if:

a) the stability of the variant is modulated; and

b) one or more activity of the target protein is modulated.

4. The method of any of Claims 1-3 wherein the agent is identified as an inhibitor of the target protein if:

a) the stability of the variant is increased; and

b) one or more activity of the target protein is decreased and/or inhibited.

5. The method of any of Claims 1 -3 wherein the agent is identified as an activator of the target protein if:

a) the stability of the variant is increased; and

b) one or more activity of the target protein is increased.

6. The method of Claim 1 further comprising:

7. The method of any of the preceding claims wherein increased stability of the variant is indicative of binding of the agent to the variant.

8. The method of any of the preceding claims wherein binding of the agent to the variant increases the stability of the variant.

9. The method of any of the preceding claims wherein binding of the agent to the variant is reversible or irreversible.

10. The method of any of the preceding claims wherein binding of the agent to the variant reduces and/or prevents degradation of the variant.

1 1. The method of Claim 10 wherein the degradation is proteasomal degradation or lysosomal degradation.

12. The method of any of the preceding claims wherein the variant is unstable relative to the target protein under physiological conditions.

13. The method of any of the preceding claims wherein the target protein is a functional protein.

14. The method of any of the preceding claims wherein the variant comprises one or more mutation in its polypeptide sequence which results in reduced stability of the variant.

15. The method of Claim 14 wherein the one or more mutation comprises a substitution, deletion and/or an addition to the polypeptide sequence.

16. The method of any of Claims 1-15 wherein the variant comprises a destabilising domain.

17. The method of Claim 16 wherein the agent does not bind to the destabilising domain.

18. The method of any of Claims 14-17 wherein binding of the agent to the variant is not affected by the mutation and/or the destabilising domain.

19. The method of any of the preceding claims wherein the one or more cell is selected from the group comprising: a mammalian cell, a non-mammalian cell, a primary cell, a cell line, a cell within a model organism, and/or a cell within a xenograft.

20. The method of any of the preceding claims wherein step (iii) comprises conditions permitting binding of the agent to the target protein and to the variant of the target protein.

21. The method of any of the preceding claims wherein step (iv) comprises quantitative and/or qualitative analysis of the amount and/or concentration of the variant.

22. The method of any of the preceding claims wherein step (iv) comprises determining accumulation of the variant.

23. The method of any of the preceding claims wherein step (v) comprises determining one or more activity of the target protein by measuring one or more of: modification of the target protein, including but not limited to post-translational modification; modification of a substrate of the target protein; expression of the target protein; expression of a substrate of the target protein; localisation of the target protein; localisation of a substrate of the target protein; expression of one or more genes downstream of the target protein; expression of one or more genes downstream of a substrate of the target protein; repression of one or more gene downstream of the target protein; repression of one or more gene downstream of a substrate of the target protein; morphology of the one or more cell (for example due to cell cycle arrest); the interaction of the target protein with one or more known or unknown interaction partners; modulation of target protein mRNA or protein levels; genomic and/or epigenetic regulation; sensitisation or resistance of the target protein or cell to a further agent; other phenotypic markers.

24. The method of any of the preceding claims wherein step (v) is performed using one or more method selected from the group comprising: fluorescence microscopy, flow cytometry, fluorescence polarization, fluorescence spectroscopy, luminescence spectroscopy, automated microscopy, automated image analysis, imaging of a whole animal or organism, Western blot; and PCR.

25. The method of any of the preceding claims wherein step (v) is performed using one or more method selected from the group comprising: transient transfection of a vector construct, stable transfection of a vector construct, fluorescence resonance energy transfer, bio-luminescence resonance energy transfer, immunofluorescence, immunohistochemistry, protein-fragment complementation assays, enzyme-fragment complementation assays, expression of a chimeric protein, tagging of an expressed protein or peptide with a fluorescent protein, epitope tagging, labelling of a reagent or cellular state with a quantum dot, production of an optically detectable reaction product, binding of an optically detectable probe, and subcellular localization of an optically detectable signal or probe.

26. The method of any of the preceding claims wherein the variant is capable of detection.

27. The method of any of the preceding claims wherein the variant further comprises one or more detectable moiety.

28. The method of Claim 27 wherein the detectable moiety comprises a fluorescent molecule, a chemiluminescent molecule, a bioluminescent molecule, a radioactive molecule, an epitope tag, a polymerase, a transcription factor, an enzyme, a signalling protein, and/or a functional protein.

29. The method of any of the preceding claims further comprising the step of:

determining whether the agent is one that binds to the target protein.

30. The method of Claim 29 wherein the step of determining whether the agent is one that binds to the target protein comprises one or more method selected from the group comprising: a cellular thermal shift assay (CETSA), differential scanning fluorimetry (DSF), a protease stability assay, for example Drug Affinity Responsive Target Stability (DARTS), an oxidation rate assay, such as Stability of Proteins from Rates of Oxidation (SPROX), an enzymatic activity assay, a binding assay, for example a Stability of Unpurified Proteins from Rates of H/D Exchange (SUPREX), a radioligand displacement assay or a fluorescence polarization assay.

31. The method of any of the preceding claims wherein the agent is selected from the group comprising: a small molecule, an antibody, a peptide, a peptidomimetic, a natural product, a carbohydrate, a nucleic acid and an aptamer.

32. The method of any of the preceding claims wherein the target protein is selected from the group comprising: an enzyme, a signalling protein, a receptor, a transcription factor, a ribozyme; and a scaffold protein.

33. The method of any of the preceding claims wherein the target protein and/or variant is introduced exogenously to the one or more cell.

34. The method of any of the preceding claims wherein the target protein and/or variant is endogenous to the one or more cell.

35. The method of any of Claims 1-34 wherein the method is a high-throughput method.

36. The method of any of the preceding claims wherein the method is automated.

37. The method of any of any of the preceding claims wherein any of steps (i) to (v) are performed in a microtiter plate.

38. The method of any of the preceding claims, wherein the one or more cell of step (ii) additionally comprises a further target protein and a variant of the further target protein.

39. A complex comprising:

wherein binding of the agent to the variant stabilises or destabilises the variant.

40. The complex of Claim 39 wherein the complex is capable of detection.

41. The complex of any of Claims 39-40 further comprising a detectable moiety.

42. A kit comprising:

i. a target protein as defined in any preceding claim; and

ii. a variant of the target protein which has reduced stability as defined in any preceding claim.

43. The kit of Claim 42 further comprising one or more agent to be tested.

44. The kit of Claim 43 wherein the agent binds to, and modulates the one or more activity of the target protein.

45. Use of one or more cell, each cell comprising a target protein and a variant of the target protein which has reduced stability, for identifying an agent that binds to, and modulates the one or more activity of the target protein.

46. A use according to Claim 44 wherein the use comprises a method as defined in any of Claims 1-38.

47. A use according to of Claims 45-46 wherein the agent, the target protein and/or the variant are as defined in any one Claims 1-38.

48. A method, complex or kit substantially as described herein, with reference to the accompanying description, examples and drawings.