Development and Applications of Chimera Platforms for Tyrosine Phosphorylation

Chimeric small molecules that induce post-translational modification (PTM) on a target protein by bringing it into proximity to a PTM-inducing enzyme are furnishing novel modalities to perturb protein function. Despite recent advances, such molecules are unavailable for a critical PTM, tyrosine phosphorylation. Furthermore, the contemporary design paradigm of chimeric molecules, formed by joining a noninhibitory binder of the PTM-inducing enzyme with the binder of the target protein, prohibits the recruitment of most PTM-inducing enzymes as their noninhibitory binders are unavailable. Here, we report two platforms to generate phosphorylation-inducing chimeric small molecules (PHICS) for tyrosine phosphorylation. We generate PHICS from both noninhibitory binders (scantily available, platform 1) and kinase inhibitors (abundantly available, platform 2) using cysteine-based group transfer chemistry. PHICS triggered phosphorylation on tyrosine residues in diverse sequence contexts and target proteins (e.g., membrane-associated, cytosolic) and displayed multiple bioactivities, including the initiation of a growth receptor signaling cascade and the death of drug-resistant cancer cells. These studies provide an approach to induce biologically relevant PTM and lay the foundation for pharmacologic PTM editing (i.e., induction or removal) of target proteins using abundantly available inhibitors of PTM-inducing or -erasing enzymes.


■ INTRODUCTION
Protein phosphorylation is the most frequent and well-studied post-translational modification (PTM), 1−3 and its dysregulation is associated with various pathologies. 4,5In particular, phosphorylated tyrosine (pTyr) is vital 6,7 in many diverse cellular functions and signaling, including highly dynamic processes often controlled using small molecules.The human genome codes 90 tyrosine kinases, 6 107 tyrosine phosphatases, 8 and over hundreds of pTyr-recognition domains (e.g., SH2 domain), 9−11 and the immune system can produce antibodies that selectively recognize pTyr over phosphorylated serine or threonine.These attributes have led to suggestions to consider pTyr a functionally separate PTM class. 12While molecules that inhibit tyrosine phosphorylation exist and constitute one of the largest classes of drugs, 13−15 there are no small molecules that can induce tyrosine phosphorylation on a given target of interest, even though these compounds will be of immense utility in both basic research and biomedicine. 16HICS-induced phosphorylation can rewire cell-signaling, change protein conformations and function, 17 evoke an immune response, 18 induce phase separation, 19,20 and alter protein−protein and protein−RNA/DNA interactions. 21,22us, PHICS can be useful in diverse settings, including basic research and biomedicine.
−25 These PHICS were formed by joining small-molecule ligands of targeted protein and a kinase via a linker.Our previously reported PHICS brought Ser/Thr kinases (AMPK or PKC) in proximity to a target protein, inducing the phosphorylation of the latter even for nonsubstrates (i.e., neo-substrates) of the kinase.However, these PHICS did not recruit endogenous tyrosine kinases or induce a downstream signaling event.Furthermore, these PHICS require noninhibitory kinase binders that are unavailable for most kinases, limiting the recruitment of only two of ≈538 human kinases. 26Beyond phosphorylation, heterobifunctional molecules that can induce or remove other PTMs, including acetylation and glycosylation, by recruiting PTM-inducing or -erasing enzymes are needed. 25,27However, the necessity of noninhibitory ligands for these enzymes has been a major roadblock in the development of such heterobifunctional molecules.−32 Herein, we report tyrosine PHICS that recruit Abelson Kinase (ABL) or Bruton's Tyrosine Kinase (BTK) to induce the phosphorylation of neo-substrates in cells.We report five noninhibitory molecular scaffolds to recruit ABL to a target protein, and PHICS generated from these scaffolds display hallmarks of contemporary heterobifunctional molecules, including hook effect, turnover, and ternary complex formation in cells.Importantly, we showed that ABL PHICS was able to initiate a signaling pathway by inducing phosphorylation of the epidermal growth factor receptor (EGFR).Furthermore, we report a new design paradigm of a heterobifunctional molecule consisting of a BTK inhibitor connected to the target binder through a linker that covalently labels BTK with the target protein binder while releasing the inhibitor from BTK and enabling phosphorylation of the target by BTK.We demonstrated this group transfer chemistry with two commonly employed covalent inhibitor scaffolds (i.e., aliphatic amine and aryl amine).Using BTK or ABL PHICS, we phosphorylated tyrosine residues in a diverse sequence context, including those surrounded by neutral, acidic, or basic residues.Finally, we generated ABL-BTK PHICS using a covalent inhibitor of BTK (i.e., ibrutinib) to induce its ABL-mediated phosphorylation; these PHICS effectively kill ibrutinibresistant cancer cell lines and inhibit the downstream survival signaling of BTK, highlighting the therapeutic potential of this class of molecules.

■ RESULTS AND DISCUSSION
Identification of ABL Binders for the Generation of BRD4 PHICS.−36 We selected four scaffolds that competitively bind to the myristoyl pocket of ABL: dihydropyrazole (scaffold 1, Figure S1A), thiazole (scaffold 2, Figure S1B), pyrazole (scaffold 3, Figure S1C), 36 and hydantoin (scaffold 4, Figure S2). 35Using molecular docking studies, we identified the solvent-exposed 3-amino group as the linker attachment site for the dihydropyrazole, pyrazole, and thiazole scaffolds (Figure S1A−C, PDB ID: 6NPV and 6NPE). 36For the hydantoin scaffold (PDB ID: 3PYY), 35 linkers were grown from the ortho, meta, or para positions of the solvent-exposed aryl ring (Figure S2).Previous reports for synthesizing hydantoin scaffold 35 involving the Vilsmer−Haack reaction provided poor yield 37 and were sensitive to substitutions on the aryl ring (Figure S3A) preventing exploration of linker growth from -ortho, -meta, or -para positions from aryl ring.We developed a new synthetic route with fewer steps to rapidly access hydantoin scaffold precursors by involving Cu-mediated high-yielding C−N bond coupling 38 between pyrazole and appropriately substituted phenylboronic acids (Figure S3B).By applying this synthetic strategy, we were able to rapidly generate a library of hydantoin-based PHICS with linkers through the -ortho, -meta, and -para positions of the aryl ring.
To assess the efficacy of these scaffolds, we synthesized four different PHICS, each containing one of the four binders of ABL, a linker composed of four poly(ethylene glycol) units, and the small molecule (S)-JQ1 (Figures S3C and S4), which binds our proof-of-concept target bromodomain-containing protein 4 (BRD4). 39With these compounds in hand, we measured PHICS-induced phosphorylation using recombinant ABL and BRD4 proteins.Dihydropyrazole induced the most BRD4 phosphorylation, followed by PHICS based on thiazole, while the pyrazole scaffold (an oxidized form of the dihydropyrazole) induced significantly less phosphorylation (Figure S5).Hydantoin-derived PHICS induced less phosphorylation than the dihydropyrazole-derived PHICS with the meta-substituted analogue showing the highest phosphorylation levels compared to the ortho-and para-substituted analogues (Figures S5 and S6).These studies identified the dihydropyrazole scaffold as the most efficient for inducing BRD4 phosphorylation.
To more deeply characterize the ability of the dihydropyrazole scaffold to induce BRD4 phosphorylation, PHICS 1 (Figure 1A) was selected for further studies together with three controls: 2 generated by replacing (S)-JQ1 in 1 with inactive enantiomer (R)-JQ1, 3 formed by replacing the dihydropyrazole in 1 with a pyrazole scaffold, and 4 a dihydropyrazole scaffold with a short linker.In biochemical settings, we confirmed that tyrosine phosphorylation of BRD4 occurred only in the presence of all components of the ternary complex, namely, 1, BRD4, and ABL (Figure 1B).We then assessed the formation of the ternary complex between 1, BRD4-GST, and ABL-His using Amplified Luminescent Proximity Homogenous assay (AlphaScreen), 40 wherein 1, but not 2, showed a "hook effect" which is a hallmark of 3-body equilibrium 28,29 (Figure 1C).Here, as the concentration of compound 1 increases, there is an increase in the population of the ternary complex.However, after a certain concentration, the binary complexes between 1 and ABL and 1 and BRD4 dominate the equilibrium over the ternary complex, resulting in a hookshaped dose curve.Furthermore, binding experiments were performed to investigate cooperativity between the BRD4 and ABL in the presence of PHICS.The cooperativity (α) defined as the ratio of affinity constants (K D ) of binary to the ternary complex 30 suggested negative cooperativity between BRD4 and ABL (Figure S7).
To demonstrate BRD4 phosphorylation in cells using PHICS, we used a construct of BRD4 lacking an intrinsic nuclear localization signal 41 and incubated HEK293T cells transfected with both HA-BRD4 and ABL-FLAG plasmids with bifunctional molecules and controls.Here, HA-based immunoprecipitation (IP) showed significantly higher coimmunoprecipitation of ABL-FLAG with 1 compared to the control compounds (i.e., 2, 3, or 4), suggesting ternary complex formation is specific to 1 (Figure 1D).Probing the immunoprecipitated HA-BRD4 with a pan phosphotyrosine antibody showed significantly higher phosphorylation levels in cells treated with 1 than in the control compounds, confirming that the observed phosphorylation is arising from PHICSmediated ternary complex formation between ABL and BRD4.To confirm that the reversible binding of PHICS to both BRD4 and ABL allows it to exhibit catalytic turnover, we used an ADP-Glo assay 42 to determine the amount of ADP generated per molecule of ABL using active molecule 1 and control 2 (Figure 1E).Here, the amount of ADP generated in the reaction with 1 was 2261 ± 411 nM higher than the amount of ADP generated in the reaction with 2 at a limiting concentration of ABL kinase (30 nM), confirming that PHICS exhibits turnover, like our previously reported Ser/Thr PHICS. 23HICS Can Activate EGFR Signaling.EGFR is a tyrosine kinase receptor that responds to the epidermal growth factor (EGF) binding to initiate a signaling pathway that regulates the growth, proliferation, survival, and differentiation of cells. 43,44pon EGF binding, EGFR oligomerizes and autophosphorylates tyrosine residues, which acts as docking site of proteins with phosphotyrosine-binding SH2 or PTB domains leading to a series of molecular events and cellular response. 45We were interested in determining if ABL PHICS can trigger signaling by phosphorylating relevant EGFR tyrosines (Figure 2A).Some kinases prefer specific residues surrounding the phosphosite, while others do not.Since signaling-relevant EGFR tyrosines are present in diverse sequence environments (i.e., surrounded by acidic, basic, or neutral residues), we were interested in determining if ABL PHICS can phosphorylate these diverse tyrosines and trigger signaling.We constructed a catalytically inactive intracellular domain of the EGFR variant fused to FKBP F36V (iEGFR-FKBP F36V -FLAG) and synthesized PHICS 5 using the FKBP F36V binder AP1867 (Figures 2B and  S8). 46HEK293T cells transfected with iEGFR-FKBP F36V -FLAG and ABL-HA were treated with 5 or dimethyl sulfoxide (DMSO).Immunoblotting of cell lysates and probing with site-specific phosphotyrosine antibodies suggested that ABL PHICS could phosphorylate tyrosines present in diverse sequence environments (Figure 2C) in line with previous reports that ABL lacks sequence motifs. 34,47−52 Similarly, we observed phosphorylation when using catalytically inactive HER2, another receptor tyrosine kinase (Figure S9), pointing to the generality of the designed PHICS.
To evaluate the effect of PHICS 5 on downstream EGFR signaling, we used a previously described luciferase-based reporter gene assay with serum response element (SRE). 53ere, HEK293T cells were cotransfected with iEGFR-FKBP F36V -FLAG and pGL4.33[luc2P/SRE/Hygro]vectors, treated with PHICS 5, or controls (DMSO or 4); the activation of EGFR signaling was quantified using Promega's ONE-Glo Luciferase Assay System.PHICS 5 induced a 2-fold higher luciferase activity than DMSO or ABL binder 4, indicating activation of EGFR by 5 (Figure 2D).These studies confirm that cytoplasmic ABL can be recruited to membranelocalized targets to induce functionally relevant tyrosine phosphorylations flanked by various amino-acid sequences and further confirms the broad range of phosphorylating motifs available to ABL kinase. 54,55HICS Can Be Developed from a Covalent Binder of the Target Protein.While covalency at the kinase end allows phosphorylation of multiple molecules of the target protein per molecule of PHICS owing to turnover, the use of a covalent binder of the target protein will block such turnover by PHICS and may dramatically reduce the efficacy of PHICS to induce phosphorylation.We were interested in determining whether efficacious PHICS can be generated using a covalent binder of the target protein.For this purpose, several bifunctional molecules were designed by connecting hydantoin or dihydropyrazole binders of ABL to chloroalkane linkers that covalently bind HaloTag (Figures 3A,B and S10). 56To confirm the cell permeability of these compounds and their ability to efficiently label HaloTag, we used a competition assay wherein HEK293T cells stably expressing HaloTag were treated with compounds 7, 9, 11, or 12 and controls (10 and 4), and the lysates were subsequently labeled with tetramethylrhodamine (TMR)-HaloTag, a fluorescent probe. 57e found that the lysates from cells pretreated with HaloTag-PHICS were not labeled by TMR-HaloTag, whereas controltreated lysates (10 and 4) showed efficient TMR-labeling (Figure S11), indicating that HaloTag-PHICS are cellpermeable and efficiently label the HaloTag protein.To assess the relative ability of these covalent PHICS to induce tyrosine phosphorylation, we transiently expressed the ABL-HA construct in HEK293T cells stably expressing the HaloTag protein and treated these cells with various covalent PHICS (compounds 7, 9, 11, and 12) and controls (compounds 6, 8, 10, and 4).We observed significant tyrosine phosphorylation on HaloTag in the presence of active PHICS compared with control-treated samples (Figure 3C).These results demonstrate that irreversible binding is tolerated by both the kinase and the target to generate a functional PHICS, opening avenues for the use of covalent inhibitors, a burgeoning class of chemical matter. 58HICS Can Be Developed Using Kinase Inhibitors.All of the reported PHICS were developed using noninhibitory allosteric binders, which are rare, though kinase inhibitors are plentiful and form an important class of therapeutic. 59A PHICS can be designed 60−62 from a kinase inhibitor in which a nucleophile (e.g., cysteine) near the inhibitor-binding pocket can initiate a group transfer chemistry 63−71 resulting in appending the target protein binder to the kinase while releasing the inhibitor from the kinase's active site (Figure 4A).The kinase binder in such cysteine-triggered PHICS can be derived from abundantly available acrylamide-based inhibitors, where the amide nitrogen is often an aryl amine 72 (13, Figure 4B) or less frequently aliphatic amine 73 (14, Figure 4B). 58,74sing these BTK inhibitor scaffolds, we synthesized BRD4 PHICS 17 (aryl amine scaffold, Figure 4B) and 19 (aliphatic amine scaffold, Figure 4B), wherein the inhibitor is connected to BRD4 binder JQ1 via a cleavable methacrylamide linker (Figure 4D and S12).We also synthesized inactive controls (18 and 20, Figure 4D) for BRD4 using (R)-JQ1 and tested whether 17 or 19 can rewire the BTK's specificity and induce the phosphorylation of BRD4.Indeed, we observed much higher BRD4 phosphorylation in the presence of 17 or 19 (Figure 4D) than with inactive controls (18 or 20) in cells (Figure 4E) .Using the same assessments of ternary complex formation as previously, we observed significantly higher levels of coimmunoprecipitated BTK-FLAG and higher levels of BRD4 phosphorylation in samples treated with 17 or 19 compared to samples treated with inactive controls, 18 or 20 (Figure 4E).Next, we generated a BTK C481S variant that cannot undergo group transfer chemistry and evaluated its ability to phosphorylate BRD4 in the presence of PHICS.We did not observe any significant BRD4 phosphorylation in the presence of 17 or 19, even though BTK C481S coimmunoprecipitated with BRD4, suggesting ternary complex formation (Figure S13A).
Next, we performed target engagement and labeling studies designed to support the mechanism outlined in Figure 4A.Using compounds that have an alkyne handle in place of JQ1 (15 and 16, Figure 4C), we confirmed that the scaffolds of 17 or 19 (Figure 4B) covalently engaged BTK in cells.Briefly,  HEK293T cells transiently expressing BTK were treated with alkyne-containing compounds 15 and 16 for 4 h.After the cells were washed with phosphate-buffered saline (PBS), the Cucatalyzed click reaction was performed on the lysates using sulfo-Cy5.5azide; BTK labeling was confirmed via in-gel fluorescence (Figure 5A).To demonstrate that the BTK inhibitor scaffold is released from the ATP-binding pocket, we used a reporter assay based on Bioluminescence Resonance Energy Transfer (BRET) 75 between a nanoluciferase (nano-Luc) and a fluorophore probe that binds to the ATP pocket; a higher inhibitor occupancy in this ATP pocket will prevent the binding of the tracer and lower the BRET signal 76 (Figure 5B).As expected, 13 and 14 exhibited higher occupancy, indicating blockage of the ATP pocket, while their derivatives with cleavable methacrylamide linkers (alkynes 15 and 16, PHICS 17 and 19) did not affect the binding of tracer at the working concentrations of PHICS (Figure 5C,D).Agreeing with these findings, we observed higher autophosphorylation of BTK with PHICS compounds 17 and 19 compared to that of the parent inhibitors 13 and 14 (Figure S13B).These studies demonstrate that PHICS can be generated from cysteinetargeting covalent inhibitors, which are available for many kinases 58,77 with diverse cellular and tissue localization, sequence preferences, and successful clinical outcomes.
PHICS Induce Death of Drug-Resistant Cancer Cells.After validating two classes of Tyr-PHICS (ABL-and BTKderived), we explored the functional capabilities of bifunctional molecules consisting of binders of both kinases since these kinases are often overexpressed in cancer cells, and we have reported that Ser/Thr PHICS can induce inhibitory phosphorylation on BTK. 24We designed compound 21, connecting the reversible binder of ABL with the covalent binder of BTK via a PEG3 linker (Figures 6A and S14) and used this PHICS along with a mixture of the separate binders (22 and 4) for the treatment of HEK293T cells cotransfected with ABL-HA and BTK-FLAG plasmids.Western blotting with a pan antiphosphotyrosine antibody revealed increased levels of BTK phosphorylation in the presence of 21 compared to the control (mixture of binders 22 and 4) (Figure 6B).Surprisingly, the level of ABL phosphorylation remained unchanged.This phosphorylation outcome motivated us to evaluate BTK-ABL bifunctional 21 in cancer cell lines that depend on ABL or BTK, particularly those that were resistant to the known BTK-targeting drug, ibrutinib. 78When tested in BTK-dependent ibrutinib-resistant Mino and Raji cell lines, 21 reduced their viability with an EC 50 of 1.7 and 4.9 μM, respectively (Figure 6C). 78In contrast, 21 did not impact the viability of BCR-ABL dependent K562 cells (Figure S15), in agreement with our expectation based on the unaffected phosphorylation of ABL in HEK293T-based studies (Figure 6B).We noted that neither the mixture of individual binders (22 and 4) nor ibrutinib induced the same effect of cell viability in BTK-dependent and ibrutinib-resistant cancer cells as compound 21.These results with ibrutinib are especially interesting, as this BTK inhibitor is an approved therapeutic agent for several B-cell cancers.However, the emergence of resistance has resulted in a need for next-generation therapeutics; 79 the promising activity of BTK-ABL PHICS in ibrutinib-resistant cell lines opens such an alternative therapeutic strategy.

■ CONCLUSIONS
To address the lack of technologies for the facile editing of PTMs (i.e., addition or removal) on a given protein of interest, we report a new class of bifunctional molecules that induce functionally relevant tyrosine phosphorylation by recruiting an ABL or BTK.PHICS triggered activation of EGFR, which belongs to the receptor tyrosine kinase superfamily that includes the insulin receptor.A PHICS that similarly activates insulin receptor signaling may furnish an orally available, smallmolecule substitute for the insulin that requires injection and may offer an alternative therapeutic modality for patients with insulin resistance.Furthermore, PHICS induced the death of drug-resistant cancer cells, offering a potentially novel and alternative therapeutic modality to kinase inhibitors against which resistance has developed.The CRISPR-Cas system 80,81 has furnished technologies for facile editing of the DNA or RNA, but methods for facile editing of post-translational modifications (i.e., addition or removal) on a given protein of interest is still challenging.We report an approach to induce phosphorylation on the target protein by generating PHICS that utilizes kinase inhibitors and cysteine-triggered group transfer chemistries.Since inhibitors for several PTM-inducing or -removing enzymes are available, and group transfer chemistry can be implemented using other nucleophilic residues (e.g., lysine, tyrosine, methionine); 64,82,83 these studies lay the foundation for pharmacologic editing of PTMs on proteins.Overall, these studies further highlight the power of bifunctional molecules to endow neo-functions to proteins in cells with value in basic research and medicine.

Figure 1 .
Figure 1.(A) Structures of ABL-BRD4 PHICS 1 and control compounds 2−4.(B) PHICS-induced phosphorylation of BRD4 by ABL in vitro.(C) PHICS-induced ternary complex formation between BRD4 and ABL observed by the AlphaScreen assay.(D) PHICS-induced phosphorylation of BRD4 by ABL in HEK293T cells; coimmunoprecipitation (Co-IP) of ABL with BRD4 in the presence of PHICS.(E) ADP-Glo assay for BRD4 phosphorylation by ABL in the presence of 1 and its inactive isomer 2.

Figure 2 .
Figure 2. (A) Schematic representation of the ternary complex between ABL, PHICS, and EGFR.(B) Structure of PHICS 5 used for the phosphorylation of iEGFR-FKBP F36V with ABL.(C) Diverse sequence environments of Tyr phosphorylated by ABL in cells in the presence of 5 were detected using antibodies specific for phospho-EGFR.(D) Induction of luciferase-based reporter gene with serum response element by PHICS 5, but not by DMSO or ABL binder 4.

Figure 5 .
Figure 5. (A) Demonstration of covalent labeling of BTK in cells by 15 and 16 at 1 μM via in-gel fluorescence.(B) Schematic representation of nanoBRET assay to determine ATP pocket occupancy.(C, D) NanoBRET assay for known BTK inhibitors 13 (C) and 14 (D), and their methacrylamide derivatives.Arrow indicates the working concentration of PHICS.

Figure 6 .
Figure 6.(A) Structures of ABL-BTK bifunctional molecule 21 and BTK binder 22. (B) PHICS-induced phosphorylation of BTK by ABL in HEK293T cells.(C) Effect of PHICS 21 on the viability of Mino and Raji cells.