Lysine-Targeting Covalent Inhibition of Glutaminase Using Cyanopyrimidines

Abstract

Aberrant metabolic activity is a hallmark of cancer and is recognized as a promising target for cancer therapy. However, covalent inhibitors that irreversibly modulate cancer-associated metabolic states remain largely unexplored. Here, we report the development of covalent inhibitors targeting glutaminase 1 (GLS1), a rate-determining enzyme in the glutamine metabolism. To develop covalent inhibitors of GLS1, we systematically evaluated the reactivity of N-heteroaryl nitriles and identified 2-cyanopyrimidine as a lysine-reactive electrophile suitable for selective covalent protein targeting. A covalent inhibitor bearing a 2-cyanopyrimidine irreversibly binds GLS1 in cancer cells with excellent proteome-wide selectivity. This mode of action effectively suppresses glutamine metabolism and induces death of glutamine-dependent cancer cells, highlighting the therapeutic potential of covalent targeting of GLS1.

Introduction

Cancer cells undergo profound metabolic reprogramming to meet the energetic and biosynthetic demands required for rapid proliferation. , In addition to the well-known Warburg effect, which involves altered glucose metabolism, glutamine metabolism represents another major metabolic pathway that is reprogrammed in cancer. , Glutamine serves as an essential source of carbon and nitrogen for the tricarboxylic acid (TCA) cycle and supports the biosynthesis of lipids, proteins, and nucleotides. Several cancer types, including lung, triple-negative breast, and pancreatic cancers, rely heavily on glutamine for growth and survival, a phenomenon referred to as glutamine addiction. Glutaminase catalyzes the conversion of glutamine into glutamate and plays a central role in glutamine metabolism. Given its critical role in cancer cell metabolism, glutaminase has emerged as an attractive therapeutic target, leading to the development of numerous inhibitors. − However, despite extensive efforts, no glutaminase inhibitors have yet been approved for clinical use.

Covalent drugs have attracted considerable interest in small-molecule drug discovery. Their irreversible mode of action can offer several therapeutic advantages, including potent and durable efficacy, isoform- or mutant-selective inhibition, and the ability to overcome acquired drug resistance. − Cysteine-targeting covalent inhibitors have been successfully developed for the treatment of cancer and autoimmune diseases. − However, the relatively low abundance of reactive cysteine residues in the proteome limits the broader applicability of this strategy. Lysines represent an attractive alternative target for covalent modification owing to their high abundance and frequent occurrence at functional protein sites. − Consequently, lysine-targeting approaches have gained increasing attention in recent years. − Nevertheless, progress in this area has been hindered by the limited availability of aminophilic electrophiles that exhibit both high chemoselectivity and sufficient chemical and metabolic stability under biological conditions.

Here, we report a new covalent inhibitor of glutaminase-1 (GLS1) that modulates the cellular glutamine metabolism. Systematic evaluation of N-heteroaryl nitriles identified 2-cyanopyrimidine as an electrophile with moderate reactivity toward lysine residues. Leveraging this reactivity, we developed a covalent GLS1 inhibitor incorporating 2-cyanopyrimidine that covalently modifies lysine residues within the catalytic pocket of GLS1. The inhibitor induced sustained suppression of glutamate levels and promoted the death of glutamine-dependent cancer cells, highlighting the potential utility of covalently targeting GLS1 for cancer therapy.

Result and Discussion

Biochemical Activity of Sulfonyl Fluoride-based Covalent GLS1 Inhibitors

CB-839 (telaglenastat) is a representative allosteric inhibitor of GLS1. − X-ray crystallographic analysis revealed that CB-839 binds at the dimer–dimer interface of the GLS1 tetramer (PDB: 5HL1), where it stabilizes an inactive conformation of the tetrameric GLS1 complex and suppresses its enzymatic activity. In this complex, Lys320 is positioned in close proximity to CB-839 and forms a cluster of four lysine residues at the tetramer interface (Figure ). Although previous lysinome analyses did not identify Lys320 as a highly reactive lysine residue, − we hypothesized that tetramerization brings four Lys320 residues into close proximity and these lysine residues could be targeted by a covalent ligand based on the molecular scaffold of CB-839. To test this idea, we synthesized covalent ligands 1–5, which bear a lysine-reactive sulfonyl fluoride (SF) group at the terminal aryl moiety of CB-839 (Figure a). In an antiproliferative assay using the breast cancer cell line MDA-MB-231, compounds 1–3 displayed potent antiproliferative activity with IC₅₀ values in the low-nanomolar range, comparable to that of CB-839 (IC₅₀ = 18.6 nM) (Figures a and S1), indicating that installation of the electrophilic warhead does not substantially compromise the inhibitory activity of CB-839 against GLS1. In contrast, compounds 4 and 5 exhibited markedly reduced activity (IC₅₀ > 1000 nM), highlighting the sensitivity of this terminal aromatic region to structural modifications. , We next evaluated the proteome-wide reactivity profiles of compounds 1 and 2 by in-gel fluorescence activity-based protein profiling (ABPP) (Figure b). MDA-MB-231 cells were treated with the corresponding alkynylated probes 6 and 7 (0.25 μM, 2 h), followed by CuAAC-mediated conjugation to TAMRA azide. Probe 6, derived from compound 1, exhibited multiple fluorescent bands, indicative of broad proteome reactivity. Preincubation with CB-839 selectively attenuated a prominent band at approximately 58 kDa, consistent with the molecular weight of GLS1, suggesting covalent engagement of GLS1 by probe 6. Pull-down assay using probe 6 followed by immunoblot analysis with an anti-GLS1 antibody detected a protein band at approximately 58 kDa, which was abolished by CB-839 pretreatment (Figure S2). These results support the covalent modification of GLS1 by probe 6. In competitive in-gel fluorescence ABPP analysis (Figure S3), probes 4 and 5 showed minimal competition with 6 for GLS1 labeling, consistent with their weak antiproliferative activities against MDA-MB-231 cells. Probe 7, bearing a methoxy substituent on the SF-phenyl ring, exhibited slightly improved labeling selectivity relative to that of 6. This improvement is likely attributable to its attenuated electrophilicity, which reduces nonspecific reactivity while maintaining target engagement. Nevertheless, the selectivity achieved with probe 7 remained insufficient. These results prompted us to explore alternative lysine-reactive electrophiles.

1.

X-ray crystal structure of glutaminase 1 (GLS1) tetramer in complex with CB-839 (PDB: 5HL1). The side chain of Lys320 is shown as light-pink sticks.

2.

Evaluation of biochemical activity of probes 1-5 bearing sulfonyl fluoride. (a) Chemical structures of CB-838 and probe 1-5 and their antiproliferative activities (IC₅₀) against MDA-MB-231 cells. Accurate IC₅₀ value could not be determined due to the limited aqueous solubility of compound. (b) In-gel fluorescence-based ABPP analysis of GLS1 labeling of alkynylated probes 6 and 7 (0.25 μM, 2 h) in MDA-MB-231 cells.

Evaluation of Electrophilic Reactivity of N-Heteroaryl Nitriles

N-Heteroaryl nitriles are well established as cysteine-reactive electrophiles. , These aromatic nitriles undergo nucleophilic addition by thiols to form thioimidates, and their reactivity can be tuned by varying the number and position of nitrogen atoms within the heteroaromatic ring. − In contrast to their well-documented cysteine reactivity, the application of N-heteroaryl nitriles as lysine-reactive electrophiles remains largely unexplored. Recently, an alkyl nitrile has recently been shown to function as a lysine-reactive eletrophile, , raising the possibility that the electron-deficient N-heteroaryl nitriles could also serve as aminophilic warheads. We therefore synthesized a series of N-heteroaryl nitriles 8-12 and systematically evaluated their aminophilic reactivity (Table and Figure ). These probes contain distinct N-heteroaryl rings and an alkyne handle for subsequent conjugation to reporter tags. We first evaluated the reactivity of 2-cyanopyrimidine 10 toward amines using N-butylamine in aqueous buffer (100 mM phosphate buffer, pH 7.4) (Figure a). UPLC analysis showed no detectable adduct formation after 48 h (Figure S4). This low reactivity is primarily attributable to the protonation of n-butylamine under neutral aqueous conditions, which substantially reduces its nucleophilicity. However, under the basic aqueous conditions (100 mM borate buffer, pH 10.0), 10 gradually underwent hydrolysis to the corresponding amide (Figure S5). We therefore changed the reaction conditions to a CH₃CN solution buffered with NEt₃/AcOH. UPLC analysis revealed that the peak corresponding to 10 gradually decreased in a time-dependent manner (Figure b), accompanied by the appearance of a new peak that was identified as amidine 13 by LC–MS (m/z = 330.19) and NMR analyses (Figure S6). The second-order rate constant (k _amine, M^–1 s^–1) for this reaction was determined to be 2.86 × 10^–5 by curve-fitting analysis (Table and Figure c), which is approximately 4-fold lower than that of sulfonyl fluoride 14 (k _amine = 11.0 × 10^–5 M^–1 s^–1) (Figure S7) under the same reaction conditions. Amidine 13 derived remained stable for more than 48 h without hydrolysis under neutral aqueous conditions (Figure S8), suggesting that the reaction of 2-cyanopyrimidine with lysine residues proceeds in an irreversible manner. This behavior contrasts with that of the corresponding thioimidate, which undergoes gradual hydrolysis over several hours under the same conditions (vide infra). The reactivity of 6-cyanopyrimidine 11 toward amines (k _amine = 2.73 × 10^–5 M^–1 s^–1) was essentially identical to that of 10, whereas 2-cyanopyrazine 12 exhibited approximately 30% lower reactivity (k _amine = 1.98 × 10^–5 M^–1 s^–1) (Figure S7). In contrast, cyanopyridines 8 and 9 displayed negligible reactivity toward amines (k _amine < 0.1 × 10^–5 M^–1 s^–1) under these conditions (Figure S7). We next assessed the thiol reactivity of N-heteroaryl nitrile probes 8-12 under neutral aqueous conditions (100 mM phosphate buffer, pH 7.4) (Table and Figure ). UPLC analysis showed that 10 unexpectedly yielded multiple products upon reaction with GSH (Figure d). LC–MS analysis identified the products as GSH adduct 15, the corresponding hydrolyzed amide 16, and a peptide bond–cleaved product 17 (Figure a). Time-course analysis revealed that the peak corresponding to 10 gradually decreased over 6 h, accompanied by the formation of 17 as the major product (Figure d). Probe 10 did not react with oxidized glutathione (GSSG), suggesting that it has no detectable reactivity toward the amino group of GSH under neutral aqueous conditions (Figure S9). Based on these data, we estimated the consumption rate (k _GSH, s^–1) of 10 in the reaction with GSH to be 17.5 × 10^–5 s^–1 (Figure S10). Table summarizes the consumption rates (k _GSH, s^–1)

1. Reactivity Profile of Probes 8-12 and 14 toward Amine and Thiols.

3.

Evaluation of reactivity of probe 10 toward amine and thiols. (a) Reaction schemes of 10 with n-butylamine and GSH. (b) UPLC analysis of reaction of n-butylamine and 10. Probe 10 (100 mM) was incubated with n-butylamine (250 mM) in MeCN containing triethylamine (1 M), acetic acid (1 M) at 37 °C. I.S. means internal standard (1-pyrenemethanol). Detection wavelength = 254 nm. (c) Time course of the reaction of probe 10 with n-butylamine monitored by UPLC analysis. Data are presented as mean ± SD from three independent experiments. The data represents mean ± SD of three independent experiments. (d) UPLC analysis of the reaction of 10 with GSH. Probe 10 (1 mM) was incubated with GSH (10 mM) in 100 mM potassium phosphate buffer (pH 7.4) containing 10% acetonitrile at 37 °C. I.S. means the internal standard (1-pyrenemethanol). Detection wavelength = 254 nm. (e) Reaction scheme of 10 with thiol 18. (f) UPLC analysis of the reaction of 10 with thiol 18. Probe 10 (1 mM) was incubated with 18 (10 mM) in 100 mM potassium phosphate buffer (pH 7.4) containing 10% acetonitrile at 37 °C. I.S. means internal standard (1-pyrenemethanol). Detection wavelength = 254 nm. (g) ¹H NMR spectral change of 10 upon incubation with thiol 18. 10 (1 mM) was incubated with 18 (10 mM) in 100 mM sodium deuterium phosphate buffer (pD 7.5) containing 10% acetonitrile at 37 °C. (h) Time-dependent change of probe 10 to thioimidate 19 upon reaction with thiol 18 analyzed by ¹H NMR analysis.

Protein Reactivity of N-Heteroaryl Nitriles

We next evaluated the proteome-wide reactivity of N-heteroaryl nitrile probes 8-12 by in-gel fluorescence ABPP analysis (Figure a,b). Incubation of MDA-MB-231 cell lysates with the probes (100 μM, 3 h) revealed that cyanopyrimidine probes 10-12 exhibited substantial reactivity toward cellular proteins, although their overall reactivity was noticeably lower than that of sulfonyl fluoride 14. In contrast, cyanopyridine probes 8 and 9 showed negligible proteome reactivity. This trend was consistent with their solution reactivity (Table ). Pretreatment of the lysates with N-hydroxysuccinimide acetate (NHS-Ac) or N-ethylmaleimide markedly reduced the intensity of the fluorescent bands generated by probes 10-12, suggesting that these N-heteroaryl nitriles predominantly target lysine and cysteine residues in cellular proteins (Figure c,d). To further confirm lysine labeling by N-heteroaryl nitriles, we performed labeling experiments using recombinant maltose-binding protein (MBP), which contain no cysteine residues. In-gel fluorescence ABPP analysis revealed that probes 10-12 labeled MBP, and the order of labeling efficiency mirrored that observed in lysate labeling experiments (Figure e). Pretreatment of MBP with NHS-Ac markedly reduced the fluorescent band corresponding to MBP, suggesting that the probes predominantly reacted with lysine residues in the protein. Interestingly, MBP labeling by 10 was greatly enhanced by GSH in a concentration-dependent manner (0–1 mM) (Figure e,f). These data suggest that the thioimidate adduct of 10 undergoes intermolecular thioimidate transfer to lysine residues of MBP. Such S-to-N imidate transfer has been reported previously in peptide synthesis and protein bioconjugation studies. −

4.

Protein reactivity profile of N-heteroaryl nitriles. (a) In-gel fluorescence-based ABPP analysis of probes 8-12 and 14 in MDA-MB-231 cell lysate. The lysate was treated with the probes (100 μM) for 3 h. (b) Quantification of proteome reactivity of probes shown in (a). (c) Competitive in-gel fluorescence-based ABPP analysis of 10-12 and 14 with Ac-NHS (5 mM) and NEM (1 mM). The lysate was pretreated with Ac-NHS (5 mM) or NEM (1 mM) for 1 h and subsequently treated with probes (100 μM) for 3 h. (d) Quantification of proteome reaction shown in (c). (e) In-gel fluorescence-based ABPP analysis of MBP (10 μM) labeling with 10, 11 and 12 (100 μM, 1 h). The reaction was conducted with or without preincubation with Ac-NHS (5 mM, 1 h) or in the presence of GSH (0–1 mM). (f) Quantification of MBP labeling shown in (e) (lower panel). (g) Waterfall plot of proteins enriched by 10 or 14 determined by chemical proteomics analysis. MDA-MB-231 cells were treated with 10 (10 μM) or 14 (5 μM) for 3 h. Highly enriched proteins were defined as log₂[abundance ratio­(probe/DMSO)] > 2. (h) A Venn diagram illustrating the inclusion relationships of the hit proteins among probes 10 and 14 in Figure g. (i) Volcano plot comparing the proteins enriched by 10 and 14. Highly enriched proteins log₂[abundance ratio­(probe/probe)] > 2 or < −2, p-value < 0.05] were defined as unique targets of each probe.

We next performed MS-based chemoproteomic analysis to evaluate the proteome-wide reactivity profile of 2-cyanopyrimidine 10. MDA-MB-231 cells were treated with 10 (10 μM) or the SF probe 14 (5 μM) for 3 h. Following cell lysis, labeled proteins were conjugated to biotin–azide via CuAAC, enriched by streptavidin affinity purification, and identified by LC–MS/MS using label-free quantification (LFQ). Probe-enriched hits were defined as proteins with log₂(probe/DMSO) > 2 and p < 0.05 (n = 3). The numbers of protein hits identified for 10 and 14 were 306 and 498, respectively, of which 30 were uniquely enriched by 2-cyanopyrimidine 10 (Figure g,h and Table S1). In a direct probe-to-probe comparison (log₂(10/14) > 2, p < 0.05), seven proteins exhibited a marked preference for 2-cyanopyrimidine over the SF probe among the commonly enriched proteins (Figure i and Table S2). These results indicate that 10 exhibits a proteome-wide selectivity profile distinct from that of 14.

Covalent Targeting of GLS1 with N-Heteroaryl Nitriles

Encouraged by the unique lysine reactivity profile of N-heteroaryl nitriles, we synthesized covalent ligands 20-24 and evaluated their target selectivity in MDA-MB-231 cells (Figure a). Treatment of cells with these ligands (0.1 μM, 2 h) yielded a prominent fluorescence band at ∼50 kDa in in-gel ABPP analysis (Figures b and S15). Among the ligands, 22 bearing a 2-cyanopyrimidine warhead exhibited the strongest fluorescence signal. Pretreatment of the cells with CB-839 markedly reduced the band intensity (Figure c), indicating selective labeling of GLS1 by 22. Notably, 22 showed substantially reduced off-target labeling compared with the SF-based ligands 6 and 7 (Figure S15). The high target selectivity of 22 may be attributed to the intrinsic reactivity profile of the 2-cyanopyrimidine warhead, including its lower amine reactivity (Table ) and negligible tyrosine reactivity compared to sulfonyl fluoride (Figure S16). In concentration-dependent labeling experiments, 22 predominantly labeled GLS1 at concentrations below 0.25 μM in MDA-MB-231 cells, whereas higher concentrations gave rise to an additional prominent band at ∼75 kDa (Figure S17). However, pretreatment with CB-839 or the nonalkynylated ligand 25 had negligible effects on the intensity of this band (Figure c), suggesting that it corresponds to an off-target protein uniquely labeled by the alkynylated ligand 22. In time-dependent labeling experiments, 22 (0.1 μM) labeled GLS1 nearly to completion within 2 h in MDA-MB-231 cells (Figures d and S18). We next performed competitive chemoproteomic analysis using the alkynylated probe 22 to further interrogate the target selectivity of 25 in MDA-MB-231 cells. Analysis identified GLS1 as the sole hit protein (abundance ratio >2 and p < 0.05, n = 3) of 25, indicating its high target selectivity (Figure e). In an in vitro labeling experiment, treatment of recombinant GLS1 with 22 yielded a fluorescent band corresponding to the labeled protein in in-gel fluorescence ABPP analysis, whereas no fluorescent band was detected upon treatment with probe 10, supporting ligand-directed labeling of GLS1 by 22 (Figure f). We next determined the labeling sites of GLS1 by LC–MS/MS analysis. Contrary to our initial expectation that 25 would target Lys320 within the CB-839 binding site, 25 labeled Lys292 and Lys354, both located proximal to the catalytic pocket of GLS1 (Figures g and S19). One possible mechanism underlying this multisite lysine labeling may involve intramolecular imidate transfer from a transiently formed (thio)­imidate intermediate with Cys287 and/or the catalytic Ser286.

5.

Covalent targeting GLS1 by N-heteroaryl probes 20-25. (a) Chemical structures of probes 20-25. (b) In-gel fluorescence-based ABPP analysis of protein reactivity of 20-24 (0.1 μM, 2 h) in MDA-MB-231 cells. (c) Competitive in-gel fluorescence-based ABPP analysis of 22 with 25 and CB-839 (0.25 μM, 2 h). MDA-MB-231 cells were pretreated with 25 or CB-839 (0.25 μM, 2 h) and subsequently treated with 22 (0.25 μM, 2 h). The fluorescence band marked with an asterisk corresponds to an off-target protein uniquely labeled by alkynylated probe 22. (d) Time course of GLS1 labeling by 22 monitored by in-gel fluorescence-based ABPP. MDA-MB-231 cells were treated with 22 (100 nM) for 0–4 h. The data represents mean ± SD of three independent experiments. (e) Volcano plot of proteins identified in the probe competition experiment between 22 and 25. MDA-MB-231 cells were pretreated with 25 (0.25 μM, 2 h) or DMSO, followed by treatment with 22 (0.25 μM, 2 h). Highly enriched proteins log₂[abundance ratio (DMSO → 22/25 → 22)] > 2, p-value < 0.05], were defined as unique targets of 25. (f) In-gel fluorescence ABPP analysis of labeling of recombinant human GLS1 (1 μM) with 22 and 10 (0.5 μM, 2 h). (g) X-ray crystal structure of the active site of GLS 1 (PDB: 5HL1). The lysine residues (Lys 292 and Lys 354) labeled by 22 are shown as orange sticks. of the N-heteroaryl nitriles. The results revealed that the relative order of thiol reactivity closely mirrored that observed for amine reactivity. To more precisely evaluate the thiol reactivity of N-heteroaryl nitriles, we next employed a structurally simple thiol 18 as a reactant (Figure e). UPLC analysis showed rapid formation of thioimidate 19 within 30 min, and the ratio of 10 to 19 remained nearly constant over 6 h, suggesting that thioimidate formation is reversible under neutral aqueous conditions (Figure f). To verify this hypothesis, we monitored the reaction of 10 (1 mM) with 18 (10 mM) by ¹H NMR analysis in 10% CD₃CN/100 mM deuterated phosphate buffer (pD 7.5) (Figures g,h and S11). The data showed that the addition reaction proceeded smoothly to afford thioimidate 19 and nearly reached equilibrium after 2 h, whereas hydrolysis of 19 to the corresponding amide was slower than the formation of 19. Based on these data, we estimated the binding constant (K _a, M^–1) of 10 with thiol 18 to be 0.58 × 10³ M^–1. Table summarizes the binding constants (K _a, M^–1) of the N-heteroaryl nitriles with 18 (Table and Figures S11–14). We found that the relative order of K _a closely mirrored that of k _GSH, suggesting that the consumption of N-heteroaryl nitriles in the reaction with GSH is primarily governed by their intrinsic thiol reactivity.

Biochemical Activity of 2-Cyanopyrimidine-based Covalent GLS1 Inhibitor

We next evaluated the biological activities of the covalent ligands bearing a 2-cyanopyrimidine warhead. Ligands 22 and 25 exhibited potent antiproliferative activity against MDA-MB-231 cells (IC₅₀ = 28 and 54 nM, respectively) (Figures a and S20), comparable to that of CB-839 (IC₅₀ = 18.6 nM). In contrast, 25 as well as 10 exhibited only weak antiproliferative activity against GLS1-independent HeLa cells (IC₅₀ = 55 μM and >100 μM, respectively) (Figure S21). Interestingly, their activities were markedly lower than that of the broadly acting irreversible glutamine antagonist 6-diazo-5-oxo-l-norleucine (DON) (IC₅₀ = 3.7 μM), − suggesting that either the high GLS1 selectivity or the lower intrinsic reactivity of 2-cyanopyrimidines, or both, may minimize nonselective cytotoxicity while retaining GLS1-dependent antiproliferative activity of 25. To more directly assess the GLS1 inhibitory activity of 25, we measured glutamate levels in the cell culture medium by fluorescence assay (Figure b). Treatment of MDA-MB-231 cells with 25 or CB-839 (0.25 μM, 3 h) induced a marked decrease in glutamate levels (F/F _DMSO = 0.28 and 0.24, respectively). Subsequent washout of CB-839 from the culture medium resulted in a significant recovery of glutamate production (F/F _DMSO = 0.46) after 3 h, whereas 25 maintained suppression of glutamate production (F/F _DMSO = 0.15) following washout. These results are consistent with sustained inhibition of glutamine metabolism through covalent engagement of GLS1. Finally, we evaluated the effects of 25 on glutamine metabolism and downstream metabolic pathways by targeted metabolomics analysis (Figure c). MDA-MB-231 cells were treated with 25 (0.25 μM) for 6 h, after which metabolites were extracted with cold methanol and subjected to LC–MS analysis. The data showed that 25 induced marked reductions in l-glutamate and α-ketoglutarate levels, indicating suppression of glutamine metabolism. Levels of tricarboxylic acid (TCA) cycle intermediates, including succinate and fumarate, were also significantly decreased, consistent with disruption of the glutamate–TCA cycle axis. Furthermore, the intracellular levels of aspartate and 2-hydroxyglutarate, downstream metabolites of the TCA cycle, were likewise reduced. These results demonstrate that covalent inhibition of GLS1 by 25 induces widespread metabolic perturbations downstream of glutamine metabolism.

6.

Evaluation of biochemical activities of covalent ligand 25. (a) Antiproliferative activity of 25 against MDA-MB-231 cells (n = 3, mean ± s.d). (b) Relative glutamate level in culture medium upon treatment of MDA-MB-231 cells with 25 or CB-839 (0.25 μM, 3 h). In washout experiment, the cells were treated with 25 or CB-839 (0.25 μM, 3 h), followed by incubation for 3 h in fresh culture medium. Data were normalized to the DMSO control. (c) Targeted metabolomics analysis of MDA-MB-231 cells treated with 25 (0.25 μM, 6 h) (n = 3, mean ± s.d). Relative peak intensities were normalized to internal standards (gabapentin for positive ion mode and ¹³C-labeled sucrose for negative ion mode). P values were calculated using a two-tailed Student’s t-test.

Conclusion

In covalent drug discovery, nitriles have attracted considerable attention because of their thiol reactivity, which enables covalent adduct formation with cysteine residues in proteins. As a result, nitrile-containing motifs have been widely employed in cysteine-targeted covalent inhibitors. − In contrast, the potential of nitriles as lysine-reactive warheads remains largely unexplored. In this study, we systematically investigated the reactivity profiles of N-heteroaryl nitriles to assess their potential for lysine-targeted covalent modification. Among the compounds examined, 2-cyanopyrimidine emerged as a promising electrophile, exhibiting sufficient aminophilicity to form stable amidine adducts both in vitro and under cellular conditions. We further elucidated several unique features of cyanopyrimidine reactivity, including structure-dependent aminophilicity, the reversible formation of thioimidates characterized by low binding constants, and thiol-assisted enhancement of amine reactivity. Although 2-cyanopyrimidine retains appreciable thiol reactivity, the lability of the resulting thioimidate adducts is expected to minimize persistent cysteine modification. This unique balance of thiol reversibility and amine reactivity may ultimately facilitate lysine-selective covalent targeting in cellular environments. Our study also revealed that the electrophilic reactivity of nitriles largely depends on the adjacent heteroaromatic scaffold. This finding suggests that selecting an appropriate heteroaryl nitrile is important not only for the development of covalent ligands with high target selectivity but also for the design of noncovalent ligands containing an inert aromatic nitrile.

GLS1 catalyzes the rate-limiting step of glutamine metabolism and has been recognized as an attractive therapeutic target in oncology. − Early efforts to develop glutaminase inhibitors relied primarily on substrate-based strategies, leading to the identification of reactive glutamine analogs such as DON. However, these orthosteric inhibitors exhibited substantial cytotoxicity, likely owing to their nonselective inhibition of a broad range of glutamine-utilizing enzymes (Figure S21). − More recently, Ruan et al. reported Michael acceptor-based covalent ligands for GLS1 derived from the CB-839 scaffold. However, their chemoselectivity and proteome-wide target selectivity remain largely unexplored. In this study, we introduced 2-cyanopyrimidine as an aminophilic electrophile for the covalent targeting of GLS1. Chemical proteomic analyses demonstrated that 22 exhibits the high target selectivity for GLS1 in living cells. These favorable properties likely arise from the moderate amine reactivity of 2-cyanopyrimidine and its minimal structural perturbation of the CB-839 scaffold. Owing to its irreversible binding to GLS1, 25 achieved sustained GLS1 inhibition while maintaining antiproliferative activity comparable to that of CB-839. We speculate that covalent modification of Lys292 and Lys354 by 25 proceeds through intramolecular transfer of an activated (thio)­imidate intermediate generated on nearby cysteine and/or serine residues. This unique reaction mode of cyanopyrimidines may provide a promising strategy for the selective covalent targeting of low-reactivity lysine residues that are inaccessible to conventional lysine-reactive electrophiles.

In addition to the well-established role of GLS1 in cancer cell metabolism, recent studies have shown that GLS1 activity is also linked to the survival of senescent cells. As a result, pharmacological inhibition of GLS1 has attracted increasing attention as a potential senolytic strategy for the selective elimination of senescent cells. , We anticipate that the targeted covalent GLS1 inhibitors could provide an effective therapeutic approach for cancer through sustained metabolic suppression while expanding the therapeutic potential of GLS1 inhibitors as senolytic agents.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.6c00924.

• Chemical proteomic data set for LFQ experiments using probe 10 and 14 (XLSX)

• Chemical proteomic data set for LFQ experiments using probe 22 and 25 (XLSX)

• Data set for targeted metabolomics analysis using probe 25 (XLSX)

• Additional tables and figures as described in the text, experimental procedures, synthetic procedures, and spectral data for the target compounds and the new intermediates (PDF)

A.O. and S.U. conceived the project and supervised the research. X.X. designed and performed the majority of experiments and analyzed the data. H.O. and S.I. synthesized compounds, and performed chemical, biochemical, and cellular experiments, and analyzed the data. N.Z., Y.M. and K.Y. conducted plasmid construction and protein expression. E.M.-S. and K.K. performed mass spectrometry–based proteomics experiments. A.O. and S.U. interpreted the results, prepared the figures, and wrote the manuscript.

The authors declare no competing financial interest.