Chemoproteomic Mapping of Glycolytic Targetome in Cancer Cells

doi:10.21203/rs.3.rs-2087840/v1

Download PDF

Article

Chemoproteomic Mapping of Glycolytic Targetome in Cancer Cells

https://doi.org/10.21203/rs.3.rs-2087840/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 15 Jun, 2023

Read the published version in Nature Chemical Biology →

Version 1

posted

You are reading this latest preprint version

Hyperactivated glycolysis is a metabolic hallmark of most cancer cells. Although sporadic information has revealed that glycolytic metabolites possess non-metabolic functions as signaling molecules, it remains largely elusive how these metabolites interact with and functionally regulate their binding targets. Here we introduce a Target Responsive Accessibility Profiling (TRAP) approach that measures ligand binding-induced accessibility changes for target identification through globally labeling reactive proteinaceous lysines. With TRAP, we mapped 913 target candidates and 2,487 interactions for 10 major glycolytic metabolites in a model cancer cell line. The wide targetome depicted by TRAP unveils diverse regulatory modalities of glycolytic metabolites involving direct perturbation of carbohydrate metabolism enzymes, intervention of orphan transcriptional proteins’ interactions, and modulation of targetome-level acetylation. These results deepen our understanding of how glycolysis orchestrates signaling pathways in cancer cells in support of their survival and inspire the exploitation of the glycolytic targetome for cancer therapy development.

Emerging evidence supports the concept that metabolic dysregulation is not only the readouts of diseases but more likely the pathological causes and driving forces of diseases^1,2. In agreement, endogenous metabolites are endowed with newly discovered roles as immune-mediator and epigenetic regulator other than the conserved role as metabolic fuel^3–8. Currently, we are in the exploratory phase of gathering information about the functions and the targets through which the bioactive metabolites exert their regulatory functions in cells, and such knowledge is anticipated to enlighten novel mechanisms of how metabolic dyregulation controls cells, expand the ligandable proteome and pinpoint potential therapeutic targets for metabolism-associated diseases^9–14.

Glycolysis is a well-established hallmark of cancers and is known to confer significant advantages to support cancer cell survival, growth, and invasion over normal populations¹⁵. The metabolic intermediates produced in glycolysis have been substantiated to serve as building blocks for lipogenesis and macromolecule synthesis that are essential for cancer cell growth and survival. Although increasing evidences indicate that the end product of glycolysis, lactate, can direct epigenetic cell metabolism and other signaling pathways^7,16,17, the knowledge of what proteins interact with other glycolytic metabolites and, therefore, how they are functionally regulated by glycolysis remains largely elusive. We thus sought to depict the landscape of glycolytic metabolites-binding proteins, termed herein as glycolytic targetome, because of the hypothetically promiscuous feature of glycolytic metabolites in interacting with proteins.

Chemoproteomics has empowered targetome mapping for individual functional metabolites, among which activity-based proteome profiling (ABPP) has been most frequently employed^18,19. ABPP often relies on the synthesis of metabolite-derivatized probes to mimic the binding behaviors of the native metabolites, followed by identifying the probe-enriched proteome as the binding targets. However, structural modification to the metabolites is likely to change the ligand’s binding behaviors and hardly applies to targetome discovery for multiple metabolites. Therefore, modification-free chemoproteomics approaches that measure ligand-induced stability changes of target proteins in response to proteolytic, thermal, solvent, reagent and other denaturation stresses have been developed ^20–25, and readily applied to ATP^14,26, arginine²⁷ and metabolites of central carbon metabolism⁹. Alternatively, we developed a Target Responsive Accessibility Profiling (TRAP) approach that probes proteome-level accessibility changes induced by ligand engagement through covalent lysine labeling and hence identify the responsive targets for ligands of interest. We first demonstrated that TRAP is complementary to current thermal stability-based target discovery approaches. Then, we used TRAP to profile the targetome for classic glycolytic metabolites and revealed 913 target candidates and 2,487 metabolite-target interactions in the human colorectal cancer cell line HCT116. Validating the deciphered glycolytic targetome led us to discover that the glycolytic metabolites contribute to regulate metabolism, interact with orphan transcriptional protein, modulate post-translational modification (PTM) status and influence pharmacological responses through direct engagement with their targets. The findings uncovered via the TRAP approach renew our understanding of how glycolytic signals impinge upon cancer cells, and provide evidence indicating that glycolysis is programmed to orchestrate diverse cellular signals.

TRAP detects ligand binding-induced accessibility changes on reactive lysines

TRAP is established on a premise that, due to ligand binding and the concomitantly increased steric hindrance, reactive residues in the ligand-binding region of the target proteins are labeled by covalent chemical reagents to different extents (Fig. 1a). Therefore, by quantifying peptides with and without ligand incubation, we can define a TRAP ratio R_{ligand/control} that reflects the altered accessibility of each examined reactive lysine and assign those displaying significant abundance changes as target-responsive peptides (TRPs) and concomitantly identify the ligand-bound target proteins.

We benchmarked the TRAP approach using ribonuclease A (RNase, Fig. 1b) and its ligands―cytidine diphosphate (CDP) and cytidine triphosphate (CTP)―whose affinities were measured by native mass spectrometry (MS) (Fig. 1c) and microscale thermophoresis (Extended Data Fig. 1a). A reductive dimethylation reaction mix consisting of isotope coded-formaldehyde and borane-pyridine complex was employed as the covalent label to detect the accessibility of free lysine ε-amine groups on RNase. The resultant time-resolved intact mass measurement confirmed the retarded reaction kinetic upon ligand engagement (Fig. 1d). Subsequent bottom-up analysis confirmed the peptide bearing labeled K41, located in proximity to the CDP/CTP-binding pocket²⁸_, as a sensitive TRP, since its abundance significantly decreased upon ligand binding. In contrast, peptides bearing lysines located remotely from the binding pocket held constant R_{ligand/control} (Fig. 1e, Extended Data Fig. 1b). Notably, the TRP bearing labeled K41 responded more pronouncedly to CTP than CDP, in line with the different affinities of ligands to the same target. Besides the ligand-binding sites, we asked whether TRAP can detect the global conformational changes of targets elicited by allosteric regulators. PKM2 is a multi-domain enzyme metabolizing phosphoenolpyruvate (PEP) to pyruvate (Pyr), and its allosteric activator fructose 1,6-bisphosphate (FBP) can trigger its tetramerization (Extended Data Fig. 1c)²⁹. With TRAP, we identified not only the exact FBP-binding site K433 that showed the greatest decreased accessibility upon FBP incubation (R_FBP/control=0.07), but also the TRPs carrying K270 and K337 that are located close to the active site (Fig. 1f-g, Extended Data Fig. 1d). Their decreased accessibilities agree with the partially closed conformation of active site following FBP activation²⁹. Moreover, activation-induced intersubunit contact explains the decreased accessibility of K422 located near the C-C’ subunit interface²⁹, suggested by the reduced intensity of the labeled K422-bearing TRP (Fig. 1f).

Benchmarking the TRAP approach for targetome mining from cell milieu

Considering the dynamic conformations of target proteins in crowded cell milieu may differ from that of the purified recombinant proteins, whether target discovery achieved through assessing protein accessibility changes upon ligand binding are amenable to cellular systems must be leveraged. We thus set out to benchmark the TRAP approach in target identification from complex cell milieu and ultimately in mapping the targetome for glycolytic metabolites. We first used TEPP-46, an anticancer reagent known to specifically bind to and activate PKM2 (AC₅₀ value ~ 92 nM), as a model ligand³⁰. We prepared HCT116 cell lysates in nondenaturing condition, labeled the proteome, and applied a label free quantification (LFQ) proteomics workflow to determine the accessibility changes of quantifiable lysine-carrying peptides. Remarkably, the TRP that displayed the most pronouncedly changed accessibility harbors labeled K305 of PKM2 (Fig. 2a-b), indicating that the tightest engagement of TEPP-46 with PKM2 near K305 can be captured by TRAP from complex cell lysate. This TRP can be consistently recognized using recombinant PKM2 (Extended Data Fig. 2a-b) and even across different models including A549 lung carcinoma cells and E. coli (Fig. 2c, Extended Data Fig. 2c-d), verifying the unbiased nature of TRAP towards species for target identification.

Then, we developed a multiplexed-TRAP workflow to further increase the throughput. It requires the division of target cell lysates into aliquots followed by incubation with ligands of interest and the matching solvent, separately (Fig. 2d). After protein-level lysine labeling, the proteome was digested, further labeled with multiplexing reagents such as 6-plex Tandem Mass Tags (TMT), and fractionated followed by MS3 quantitative analysis. As expected, the multiplexed-TRAP analysis also identified K305 of PKM2 whose accessibility to labeling reagents decreased-evidenced by the decreased abundance of labeled K305-carrying peptide།following TEPP-46 treatment (Fig. 2e-f). Moreover, the extent of labeling at K305 dose-dependently decreased in response to the incubated TEPP-46 (Extended Data Fig. 2e-f). In contrast, the non-TRP harboring lysine that has negligible involvement in TEPP-46-PKM2 interaction exhibited constant level of labeling (Fig. 2g, Extended Data Fig. 2g). Altogether, TRAP is poised to be a proteome-wide tool for targetome discovery and can pinpoint the ligand-binding responsive regions with no need of modification made to the ligands of interest.

TRAP identifies thermally stable drug targets

Since TRAP identifies targets based on ligand engagement-induced accessibility changes or conformational alterations, we queried that whether TRAP complements current target discovery approaches such as thermal proteome profiling (TPP) and cellular thermal shift assay (CETSA) that screen target proteins based on binding-induced thermal stability changes^22,23. Interestingly, analysis of the melting behaviors of the human proteome using a representative HEK293T cell meltome dataset indeed revealed the existence of proteins highly resistant to heat denaturation (defined as non-melters) (Fig. 3a, Extended Data Fig. 3a-b)³¹. Categorizing the non-melters based on their biological process terms of Gene Ontology (GO) showed their enrichment in proteasomal proteins (Fig. 3b, Extended Data Fig. 3c), including the essential components of 20S proteasome―proteasome ɑ and β subunits (Fig. 3c). We inferred that the thermally stable nature of these proteins make classic thermal stability-based target discovery approaches inapplicable. To validate this, we focused on the protein 20S proteasome β5 subunit that can be targeted by Bortezomib³², and plotted its melting curves in additional human cell lines by re-processing the Meltome Atlas (Fig. 3d). We confirmed the β5 subunit as a consistent non-melter, and then asked whether such non-melter can be identified as the target for the ligand of interest by TRAP due to its distinct working mechanism from TPP or CETSA. We incubated HCT116 cells with Bortezomib, and performed TRAP analysis. A TRP bearing K91 was identified for Bortezomib from the β5 subunit, in accordance with the crystallography structure illustrating their proximity (Fig. 3e-f). Consequently, this proof-of-principle study allowed us to demonstrate that TRAP complements to current approaches in expanding the identifiable targetome.

TRAP mapped a landscape of glycolytic targetome in human cancer cells

Glycolytic metabolites are often defined as bioenergetics and biosynthetic precursors for cancer cells, albeit with their non-metabolic roles remaining encrypted. With the benchmarked TRAP approach, we set out to map the glycolytic metabolites targetome in cancer cells in hope to uncover new sensors of glycolytic intermediates, thereby providing novel insights into understanding how glycolysis is programmed to support cancer progression. TRAP analysis of 10 glycolytic metabolites (Fig. 4a) was carried out in two batches using the multiplex-TRAP workflow for the colorectal cancer cell line HCT116 (Fig. 4b). In total, 160,459 peptides mapping to 6,926 unique proteins were identified, and 159,823 peptides belonging to 6,913 proteins were quantified. We first assessed the coverage of lysine-centric labels used in this study, and noted ~ 57.42% of lysine residues carrying the designated modification (Fig. 4c, Extended Data Fig. 4a). Consistently, the labeled fraction of the quantified protein reached ~ 57.34% (Extended Data Fig. 4b-c). Besides the relatively wide coverage³³, distribution plot of the labeled lysine residues further showed that the labeling reaction exploited by TRAP is not preferential towards any secondary structure elements (Fig. 4d, Extended Data Fig. 4d). The unbiased and reactive nature of lysine-centric labeling thus guarantees the discovery coverage of glycolytic metabolite targetome via TRAP. Overall, we initially identified 913 proteins as glycolytic metabolite target candidates and 2,487 interactions in HCT116 cells (Fig. 4e, Extended Data Fig. 4e, Table S1). For a total of 140 previously known metabolite-target interactions for homo sapiens documented in BRENDA that are quantified in this study, TRAP identified 30 known enzyme-metabolite interactions involving 19 proteins (Extended Data Table S2) and delivered a true positive rate of 21.43% comparable with previous stability-based approach⁹, which is expected to be further enhanced by overcoming proteomics undersampling (Extended Data Fig. 4f-g).

Among the specific set of lysines in TRPs, we explored their functional ramifications and found that the responsive sites are more conserved across organisms than all the quantified lysines, indicating their potential functionality (Fig. 4f). Since functional analysis of the identified glycolytic targetome suggested the enrichment in the category of catalytic proteins, we then analyzed the lysines in the TRPs of these enzymes and found that, consistently, these residues are closer to the functional sites of given enzymes compared to those of the quantified peptides, reiterating the functionality of these glycolytic metabolites-responsive lysines (Fig. 4g, Extended Data Fig. 5a). Considering the fact that multiple glycolytic intermediates are well-recognized enzymatic ligands and/or substrates⁹, we asked whether these findings together imply that the assayed glycolytic metabolites are likely to modulate the activity of the enzymes that are assigned as glycolytic targets by TRAP (Fig. 4h).

Glycolytic metabolites as promiscuous regulators of carbohydrate metabolism

Since the enzymatic targets are further grouped to the carbohydrate metabolism pathway at highest frequency via KEGG pathway annotation network analysis (Fig. 5a-b, Extended Data Fig. 5b, Table S3), we focused on these carbohydrate metabolism enzymes and examined whether the TRAP-identified glycolytic metabolites interactions tend to influence the occupancy of the active sites and concomitantly their activities. We define the boundary of an active site detectable by TRAP as 5.69 Å based on the Euclidean distances measured between the detected TRPs from known enzymatic targets that use these examined metabolites as substrates and the corresponding active sites (Extended Data Fig. 5c, Table S4). Therefore, when the distance from TRPs of given metabolites to an enzymatic active site falls within 5.69 Å, altered enzymatic activities may be expected. We chose PKM2 and GAPDH, two promiscuous targets identified by TRAP (Fig. 5c), to test the proposed regulatory model. With the PKM2 activity assay, we identified two novel enzymatic activators, F6P and G6P, besides the classic activator FBP (Fig. 5d, Extended Data Fig. 5d). Such regulatory effect agrees with our observation of the pronouncedly decreased accessibility in PKM2 active site (~ 1.44 Å) following F6P/G6P incubation (Fig. 5c, Extended Data Fig. 5e). In contrast to these activators, we noted 3PG as a unique ligand (Fig. 5c) that harbors a TRP located distantly from the active site of PKM2 (~ 26.78 Å) among the examined glycolytic metabolites. Thus, with surface plasmon resonance (SPR) supporting the binding of 3PG with PKM2 (Fig. 5e), we hypothesized that 3PG may not be able to modulate PKM2 activity (Fig. 5d). Intriguingly, TRAP analysis solicited us to note that a K433-contaning TRP of 3PG is shared by FBP, indicating that 3PG may compete with FBP for binding to the K433-proximal region (Extended Data Fig. 5f). This inference was substantiated by the counteracted PKM2 activation upon co-administrating 3PG with FBP (Fig. 5f). Thus, 3PG hardly alters the enzymatic activity of PKM2 itself yet inhibits the FBP-induced activation. This finding suggests a complex and intrinsic sensory mechanism of cancer cells in precisely tuning cancer glycolysis via the central hub of PKM2, the activation of which may be balanced by the composition ratios of multiple glycolytic intermediates.

The GAPDH enzymatic assay against the discovered ligand candidates surprisingly revealed that pyruvate can bind to and weakly activate GAPDH (Extended Data Fig. 6a). The altered accessibility of lysines located remotely from the active site signifies a plausible allosteric mechanism heretofore unknown for GAPDH (Extended Data Fig. 6b-c). This may inspire the design of novel chemicals targeting this ubiquitous and multi-functional protein that is intimately associated with various pathophysiological diseases including cancer, immune diseases and neurodegeneration⁶.

Glycolytic metabolites modulate druggable enzymes activity

Metabolic enzymes constitute a major class of drug targets. Theoretically, such enzymes’ activities can be modulated by their corresponding endogenous substrates/products/modulators, which may contribute to determine the ultimate activities of these drug-targeting metabolic enzymes. However, this possibility has largely been ignored. We are thus of interest to test whether glycolytic metabolite intermediates can influence the efficacies of drugs acting on the metabolites-binding enzymes. Firstly, screening of the identified glycolytic targetome showed that they are enriched as druggable proteins compared to the quantified whole proteome (Extended Data Fig. 6d). Among these druggable proteins, we found that FBP administration dose-responsively inhibited nicotinamide phosphoribosyltransferase (NAMPT), an essential protein for NAD⁺ biosynthesis in cancer cells³⁴. Moreover, FBP synergized with FK866, the NAMPT inhibitor under clinical investigation, in inhibiting NAMPT activity (Fig. 5g). Mechanistically, TRAP analysis revealed that FBP binding reduced the chemical accessibility of K189 that is located in proximity to FK866 (~ 3.42 Å) (Fig. 5h), and thus suggested that FBP treatment may enhance the affinity of FK866 to NAMPT and lead to the strengthened inhibition. Another example of the regulatory effect of glycolytic metabolite on druggable enzymes is the effect of co-administrating pyruvate―our newly identified GAPDH activator―with dimethyl fumarate (DMF), a clinical agent that combats multiple sclerosis by inhibiting GAPDH⁶. In contrast, pyruvate was found to effectively compromise the DMF-mediated inhibition on GAPDH (Extended Data Fig. 6e-f). Together, these findings present an important concern that glycolytic metabolites abundance may contribute to individually variant responses to therapeutics targeting the exact protein that can be liganded by endogenous metabolites, suggesting that glycolytic homeostasis can be harnessed as adjuvants for current cancer therapeutic options.

Glycolytic metabolites engage transcriptional factors and protein-protein interactions

Besides enzymes, targetome mining allows us to uncover noncanonical proteins as glycolytic targets and thereby interrogate new regulatory outcomes elicited by glycolytic metabolites in cancer cells. Intriguingly, we noted that the glycolytic targetome constitute proteins ascribed to transcriptional activity according to the Human Transcription Factor Database³⁵ (Fig. 6a). TRIM28 remains as an orphan transcriptional regulator without established ligands and here we suggested it as the binding partner for multiple glycolytic metabolites. TRIM28 is a universal nuclear co-repressor for Kruppel-associated box zinc finger proteins (KRAB-ZFPs), and a SUMO and ubiquitin E3 ligase^36,37,38. Its association with tumorigenesis is reported by mechanistic studies³⁹ and supported by consistent upregulation in multiple cancer types and negative association with overall survival (Extended Data Fig. 7a-b). Among the TRAP-identified ligands, lactate changed the chemical accessibility of TRIM28 mostly, indicating its stronger interaction with TRIM28 compared with other assayed glycolytic ligands. We confirmed the proposed lactate-TRIM28 interaction by thermal shift assay in cell lysates (Fig. 6b) and SPR (Fig. 6c). Subsequent TRAP analysis of cellular and recombinant TRIM28 for lactate both led us to note the TRP carrying labeled K337/K340 located in the coiled-coil (CC) domain (Fig. 6d-e, Extended Data Fig. 7c), through which TRIM28 is known to interact with important protein partners bearing the KRAB domain and MAGE homology domain^38,40,41. We confirmed that mutation at K337/K340 weakened the lactate-mutant TRIM28 interaction (Fig. 6c), corroborating the TRAP-assigned binding sites for lactate. These findings solicited us to further explore that whether the accumulated lactate in cancer cells may influence the protein-protein interactions (PPIs) of TRIM28 through binding to the K337/K340-proximal region (Fig. 6f). Thus, we performed co-immunoprecipitation experiments and identified the potential interacting partners of TRIM28 by quantitative proteomics, in hope to verify that lactate incubation would disturb the interactome of TRIM28. We incubated an anti-FLAG antibody with the lysates of FLAG-TRIM28 overexpressed HEK293T cells with and without lactate treatment (Fig. 6f, Extended Data Fig. 7d). As expected, quantitative assessment of the co-immunoprecipitated proteins showed that the TRIM28-interactome was disturbed―with most interacting proteins weakened their binding affinities to TRIM28 following lactate administration, albeit with rare cases that showed enhanced affinity to TRIM28 (Fig. 6g-h, Extended Data Table S5).

Besides TRIM28, we also noticed that lactate affected the chemical accessibility and hence likely the conformations and functions of other transcriptional proteins that execute cancer malignant phenotypes exemplified by YBX1, a core regulator of MEK/ERK signaling-dependent gene expression signatures, and NPM1, a multifunctional phosphoprotein implicated in tumorigenesis and immune escape^42,43. Together with the recent discovery of lactate-derived lactylation on histones and other transcriptional factors⁷, our revisit of lactate as a potential ligand of transcriptional proteins in cancer cells elicits a hypothesis that the overproduced lactate is fundamental for cancer incidence and progression, possibly through perturbing its targets of transcriptional activity.

Direct binding of glycolytic metabolites to target proteins affect acetylation levels

Besides transcriptional regulation that controls downstream genes and protein abundance, PTM serves as another core mechanism of tuning protein conformation and activity. Driven by the question of whether glycolytic metabolites binding affects the PTM levels of their targetome, we retrieved PTM evidence for lysines in TRPs and noted that the metabolite-binding responsive lysines are more likely to carry PTMs compared to lysines in the total quantified proteome (Fig. 7a). Noteworthy, acetylation was enriched for ~ 2 folds for lysines in TRPs (Fig. 7b). To exclude the possibility that the differences in the frequency of acetylated lysines resulted from the heightened percentage of acetylation in the assigned targetome, we compared the frequency of lysine acetylation for lysines of TRPs with those of non-TRPs yet from identical glycolytic targets, and also noted enriched acetylation (Extended Data Fig. 8a). GO analysis showed that these proteins do not distribute in exclusive subcellular location or specific class compared to the quantified proteome (Extended Data Fig. 8b-c). Therefore, we conjecture a regulatory mechanism of glycolytic metabolites on protein lysine acetylation-direct engagement would shelter the acetylation sites and partially block acetylation events.

Since G3P is the top-ranked glycolytic metabolite that bears most documented acetylated lysines in the TRPs aside from lactate (Fig. 7c), we sought to validate whether G3P engagement may downregulate the acetylation levels of its target proteins. We first identified α-enolase (ENO1) as its potential target, as ENO1 carried the TRPs sensitively responding to G3P incubation (TRAP Ratio R_G3P/control=0.36) (Fig. 7d). The interaction was subsequently confirmed in cell lysates by thermal shift assay revealing G3P-mediated stabilization of ENO1 across the tested temperatures compared to the vehicle (Fig. 7e). An iso-thermal dose-response experiment further corroborated the stabilizing effect (Extended Data Fig. 9a). In agreement with the TRPs of lactate identified from HCT116 cell lysate (Extended Data Fig. 9b), TRAP analysis of the purified ENO1 also led us to note a labeled K330/K335-bearing TRP showing significantly reduced accessibility upon G3P incubation (Fig. 7f). The involvement of K330/K335 in G3P-ENO1 interaction was validated by SPR, as the K330/335E point mutation impaired the binding of G3P to ENO1 (Fig. 7g, Extended Data Fig. 9c). As expected, differential scanning fluorimetry (DSF) also showed that G3P’s protection of WT-ENO1 against heat denaturation was abrogated upon the K330/335E mutation (Extended Data Fig. 9d). As both K330 and K335 are documented acetylation sites, we asked whether G3P binding may inhibit K330/K335 acetylation. We incubated ENO1 with p300, the responsible acetyltransferase⁴⁴, in the presence and absence of G3P, and performed bottom-up analysis on the in vitro acetylated ENO1 (Fig. 7h). As expected, we confirmed that acetylation at K330/K335 was significantly downregulated by G3P incubation among the markedly acetylated lysines following p300 treatment. As such, we corroborated the inference that G3P blocked the acetylation of ENO1-K330/335 by direct engagement and suggested that glycolytic metabolites can prevalently regulate the PTM levels on its targetome.

We are also intrigued by the impact of pyruvate treatment on the acetylation level of its targetome, as pyruvate is another top-ranked metabolite that possibly serves as an acetylation modulator (Fig. 7c). TRAP analysis indicated that pyruvate affected the accessibility of a TRP carrying K27/K36 of histone H3.3. Given the essentialness of histone acetylation in epigenetic regulation, we queried that whether pyruvate can modulate histone acetylation. We in vitro incubated pyruvate with recombinant histone H3.3 and its acetyltransferase p300⁴⁵, and then performed bottom-up analysis to examine the pyruvate-induced changes on histone lysine acetylation. Quantitative analysis confirmed that multiple lysines were acetylated upon the addition of p300, whereas the acetylation on K27 of H3.3 (H3K27)—in agreement with the TRAP-identified responsive sites—was significantly inhibited by pyruvate (Fig. 8a). We next investigated whether the reduced H3K27 acetylation holds true in cells. Upon validating the ability of pyruvate in entering HCT116 cells (Extended Data Fig. 10a-b), we treated the HCT116 cells with trichostatin A (TSA), a mammalian histone deacetylase (HDAC) inhibitor⁴⁶, to upregulate H3K27 acetylation, and then examined whether the presence of pyruvate would abrogate the induced acetylation. In agreement with the in vitro acetylation assay, immunoblotting confirmed that pyruvate administration downregulated the TSA-induced acetylation of H3K27 (Fig. 8b). Projecting forward, as H3K27 acetylation being reported to induce apoptosis in several cancer cell lines and suppress tumor growth^47,48 we asked whether the influence of pyruvate on H3K27 acetylation would affect the pharmacological phenotypes induced by TSA. As a result, we observed that pyruvate treatment mitigated the TSA-induced apoptosis (Fig. 8c) and reduced PARP and cleaved caspase-3 levels accordingly (Fig. 8d), implying that pyruvate ameliorated the TSA-induced HCT116 cell death. From these findings, we bolster an intriguing hypothesis that the accumulation of glycolytic intermediates may represent an intrinsic mechanism of cancer cells preferring glycolysis in resisting various stress-elicited cell death.

Although the knowledge of the metabolites-binding targetome and its translational significance is expanding, the metabolite-protein interaction network remains relatively understudied especially when compared with the extensively characterized protein-protein interactome. Here we introduce a chemical accessibility-based TRAP approach and apply it to map the targetome of glycolytic metabolites in human cancer cell milieu. The uncovered targetome supports the recently revisited roles of metabolites—not solely as metabolic fuels and biosynthetic building blocks—but also as central mediators that intimately regulate carbohydrate metabolism, nuclear transcription, protein PTM dynamics and macromolecular assemblies in cancer cells.

Previously, target discovery of small molecules has been resolved by an arsenal of chemoproteomics methodologies including both the probe-based ABPP approach¹⁹ and derivatization-free approaches that examine ligand binding-induced stability changes in response to thermal stress, chaotropes and proteolysis^20–25. Although powerful, the ABPP approach—frequently relying on ligand-derivatized probes for target enrichment—is limited in throughput and may suffer from inaccuracy owing to the introduced derivatization to the original structure, whereas the derivatization-free approaches are mostly applied to discovering drug targets featuring high-affinity and specific interactions and may not fit for the mining of the low-affinity and promiscuous metabolites-targetome interactions. Alternatively, we developed the TRAP approach to identify the glycolytic targetome based on binding-induced steric hindrance and/or conformational change-that can be captured by altered extents of labeling on reactive lysines and recognized by quantitative proteomics. Indeed, the targetome revealed by TRAP confirmed the promiscuity of glycolytic metabolite-protein interactions and suggested the existence of a co-regulatory network. Such promiscuity is anticipated due to the chemical mimicry among metabolites produced through the same pathway. Meanwhile, the metabolites of analogous structures often display variant affinity and binding topology to the promiscuous target and may thus act differently, for instance, by serving as agonists/antagonists with a spectrum of potency on the same protein―as we have learned from bile acids on farnesoid X receptor (FXR) and tryptophan derivatives on aryl hydrocarbon receptor (AhR)⁴⁹. Although this seems to be a prevalent mechanism for nuclear receptors sensing metabolic signals, our discovery that PKM2—a well-defined rate-limiting glycolytic enzyme—can also detect multiple glycolytic metabolite species and dynamics is intriguing. Together with previous knowledge of FBP as an allosteric PKM2 activator, our findings of F6P and G6P as novel activators while 3PG counteracting FBP-induced activation elicited us to propose PKM2 and possibly other promiscuous targets that constitute convergent hubs by receiving diverse metabolic signals, determining the net effect and ultimately culminating in well-tuned responses in a metabolic context-dependent manner.

The co-existence of counteracting ligands in native systems also sheds light on answering why some potent activators or inhibitors in vitro are less effective in vivo. Considering that the TRAP-assigned glycolytic targets consist of a wealth of validated drug targets, we reasoned that variant glycolytic activity among individuals may modulate the pharmacological outcomes as certain glycolytic metabolites may compete or strengthen drug-target engagement in vivo based on our proof-of-principle validations of the synergistic effect of co-administering FBP with FK866 and the opposite effect on GAPDH activity by co-administering pyruvate with DMF. Interestingly, the previously uncovered lipid-binding proteome also highlights its significant enrichment in drug targets. As such, we believe that monitoring metabolism is essential in determining individual responses to cancer therapeutics whose targets are also liganded by endogenous metabolites, and the manipulation of metabolites-drug crosstalk represents therapeutic opportunity to improve clinical benefits. Indeed, pharmacometabolomics approach has been proposed and exploited to explain the individual variations of responses to drugs, and verified that metabolic biomarkers can be connected with varied responses to drugs, but without clear mechanisms^1,50. We thus proposed here that the pharmacometabolomics approach should be coupled with chemoproteomics-based targetome mining, as exemplified in this study, to explore how metabolic variations determine the individually different responses to therapeutics.

The decoded glycolytic targetome also constitute nuclear proteins of transcriptional activity―the class that is often categorized as unligandable. Thus, validation of lactate as a new ligand to the orphan transcriptional protein TRIM28 provided a glimpse into the broad ligandability of glycolytic targetome and, likewise, the endogenous metabolites constituting a fertile resource for innovative drug design—especially for proteins that do not possess known ligands or lack functional annotations and remain as the dark matter in the human proteome. Specifically, we noted that lactate engagement with TRIM28 can disturb the interactome of TRIM28 in HEK293T cells, weakening the affinity of most protein partners to TRIM28 yet strengthening interactions such as TRIM28 with high mobility group protein B1 (HMGB1) and inosine-5’-monophosphate dehydrogenase 2 (IMPDH2). Intriguingly, TRIM28, besides being a transcription co-repressor, is an E3 ubiquitin ligase and ubiquitinate substrate proteins through PPIs. A recent in vitro ubiquitination assay has determined HMGB1 as a primary degradation target of TRIM28 when assembled with MAGE family proteins, which further enhances the ubiquitin ligase activity of TRIM28⁵¹. As the damage-associated molecular pattern protein HMGB1 is also a critical pro-autophagic regulator⁵², we reasoned that the lactate-induced TRIM28-HMGB1 assembly may facilitate the degradation of HMGB1 and lead to the suppression of autophagy. Since autophagy has been demonstrated as critical to the induction of potent immune responses and sensitivity to CTLA-4 treatment in melanoma patients⁵¹, we bolster a hypothesis that the repression of autophagy of cancer cells and the conferred resistance to immunogenic therapies are connected with glycolysis through the end product, lactate. Such survival advantage conferred by glycolytic metabolites to cancer cells is further supported by the effect of pyruvate in attenuating H3K27 acetylation and concomitantly ameliorating the TSA-induced apoptosis in HCT116 cells. Together, these findings endowed the glycolytic metabolites with new roles as the regulators of nuclear proteins that are key in cell fate determination for conferring survival benefits to cancer cells―this plausibly explains why hyperactivated glycolysis is favored in cancer cells besides supporting energy production and biosynthesis.

In summary, the wide glycolytic targetome identified via TRAP in this study provides a new perspective to perceive how glycolytic metabolites regulate cancer cells through finely tuning its targets. Together with recent emerging studies deciphering the targetome and elucidating novel action modes for functional metabolites like lactate, itaconate and arginine, isolated recording of proteins and metabolites changes at abundance level and with spatial information are insufficient to understand the underlying pathological mechanisms, whereas globally mapping the metabolites-targetome interactions can truly reflect how proteins integrate multiple metabolic signals, respond and orchestrate pathophysiological phenotypes—the knowledge gap between the rich metabolite-protein interactions that are naturally occurring in cells and the scarce interactions that we have collected and validated must be bridged.

ACKNOWLEDGEMENTS

This research was supported by the National Natural Science Foundation of China (grants 82173783 to H.Y., grants 81930109, 81720108032 to H.H.), the Natural Science Foundation of Jiangsu Province (BK20220088), the National Key Research and Development Program of China (2021YFA1301300), the Fundamental Research Funds for the Central Universities (2632022YC03), the Overseas Expertise Introduction Project for Discipline Innovation (G20582017001) and Sanming Project of Medicine in Shenzhen (SZSM201801060). We thank Prof. Yibei Xiao from China Pharmaceutical University, Dr. Qing Yu from Harvard University and Dr. Baozhen Shan and Wenting Li from PEAKS Studio for useful discussions.

AUTHOR CONTRIBUTIONS

H.H., H.Y. and G.W. conceived the project. Y.T., N.W., H.Z., C.S., H.Y. and H.H. designed the experiments. Y.T., N.W., H.Z., H.Y. and M.D. performed the proteomics experiments. Y.T., H.Z. and C.S. performed the flow cytometry and western blotting experiments, Q.B. performed the SPR experiments, K.Z. and S.C. carried out protein site-mutation and purification experiments, H. H. conducted the conservation analysis, N.W., Y.T., H.Y. and H. H. analyzed the experimental data. N.W., H.Y. and H.H. wrote the paper with input from co-authors.

COMPETING INTERESTS

The authors declare no competing interests.

MATERIALS & CORRESPONDENCE

Correspondence and material requests should be addressed to [email protected] and [email protected] (lead contact).

Wishart D. S. (2016). Emerging applications of metabolomics in drug discovery and precision medicine. Nature Rev. Drug Discov. 15, 473–484.
Guijas, C., Montenegro-Burke, J. R., Warth, B., Spilker, M. E., & Siuzdak, G. (2018). Metabolomics activity screening for identifying metabolites that modulate phenotype. Nat. Biotechnol. 36, 316–320.
Zasłona, Z., & O'Neill, L. (2020). Cytokine-like roles for metabolites in immunity. Mol. Cell 78, 814–823.
Runtsch, M. C., Angiari, S., Hooftman, A., Wadhwa, R., Zhang, Y., Zheng, Y., Spina, J. S., Ruzek, M. C., Argiriadi, M. A., McGettrick, A. F., Mendez, R. S., Zotta, A., Peace, C. G., Walsh, A., Chirillo, R., Hams, E., Fallon, P. G., Jayamaran, R., Dua, K., Brown, A. C., et al. (2022). Itaconate and itaconate derivatives target JAK1 to suppress alternative activation of macrophages. Cell Metab. 34, 487–501.e8.
Bunse, L., Pusch, S., Bunse, T., Sahm, F., Sanghvi, K., Friedrich, M., Alansary, D., Sonner, J. K., Green, E., Deumelandt, K., Kilian, M., Neftel, C., Uhlig, S., Kessler, T., von Landenberg, A., Berghoff, A. S., Marsh, K., Steadman, M., Zhu, D., Nicolay, B., et al. (2018). Suppression of antitumor T cell immunity by the oncometabolite (R)-2-hydroxyglutarate. Nat. Med. 24, 1192–1203.
Kornberg, M. D., Bhargava, P., Kim, P. M., Putluri, V., Snowman, A. M., Putluri, N., Calabresi, P. A., & Snyder, S. H. (2018). Dimethyl fumarate targets GAPDH and aerobic glycolysis to modulate immunity. Science 360, 449–453.
Zhang, D., Tang, Z., Huang, H., Zhou, G., Cui, C., Weng, Y., Liu, W., Kim, S., Lee, S., Perez-Neut, M., Ding, J., Czyz, D., Hu, R., Ye, Z., He, M., Zheng, Y. G., Shuman, H. A., Dai, L., Ren, B., Roeder, R. G., et al. (2019). Metabolic regulation of gene expression by histone lactylation. Nature 574, 575–580.
Humphries F, Shmuel-Galia L, Ketelut-Carneiro N, Li S, Wang B, Nemmara VV, Wilson R, Jiang Z, Khalighinejad F, Muneeruddin K, Shaffer SA, Dutta R, Ionete C, Pesiridis S, Yang S, Thompson PR, Fitzgerald KA. (2020). Succination inactivates gasdermin D and blocks pyroptosis. Science 369, 1633–1637.
Piazza, I., Kochanowski, K., Cappelletti, V., Fuhrer, T., Noor, E., Sauer, U., & Picotti, P. (2018). A Map of protein-metabolite interactions reveals principles of chemical communication. Cell 172, 358–372.e23.
Niphakis, M. J., Lum, K. M., Cognetta, A. B., 3rd, Correia, B. E., Ichu, T. A., Olucha, J., Brown, S. J., Kundu, S., Piscitelli, F., Rosen, H., & Cravatt, B. F. (2015). A global map of lipid-binding proteins and their ligandability in cells. Cell 161, 1668–1680.
Hulce, J. J., Cognetta, A. B., Niphakis, M. J., Tully, S. E., & Cravatt, B. F. (2013). Proteome-wide mapping of cholesterol-interacting proteins in mammalian cells. Nat. methods 10, 259–264.
Bollong, M. J., Lee, G., Coukos, J. S., Yun, H., Zambaldo, C., Chang, J. W., Chin, E. N., Ahmad, I., Chatterjee, A. K., Lairson, L. L., Schultz, P. G., & Moellering, R. E. (2018). A metabolite-derived protein modification integrates glycolysis with KEAP1-NRF2 signalling. Nature 562, 600–604.
Qin, W., Qin, K., Zhang, Y., Jia, W., Chen, Y., Cheng, B., Peng, L., Chen, N., Liu, Y., Zhou, W., Wang, Y. L., Chen, X., & Wang, C. (2019). S-glycosylation-based cysteine profiling reveals regulation of glycolysis by itaconate. Nat. Chem. Biol. 15, 983–991.
Sridharan, S., Kurzawa, N., Werner, T., Günthner, I., Helm, D., Huber, W., Bantscheff, M., & Savitski, M. M. (2019). Proteome-wide solubility and thermal stability profiling reveals distinct regulatory roles for ATP. Nat. Commun. 10, 1155.
Pavlova, N. N., Zhu, J., & Thompson, C. B. (2022). The hallmarks of cancer metabolism: Still emerging. Cell Metab. 34, 355–377.
Caielli, S., Cardenas, J., de Jesus, A. A., Baisch, J., Walters, L., Blanck, J. P., Balasubramanian, P., Stagnar, C., Ohouo, M., Hong, S., Nassi, L., Stewart, K., Fuller, J., Gu, J., Banchereau, J. F., Wright, T., Goldbach-Mansky, R., & Pascual, V. (2021). Erythroid mitochondrial retention triggers myeloid-dependent type I interferon in human SLE. Cell 184, 4464–4479.e19.
Wan, N., Wang, N., Yu, S., Zhang, H., Tang, S., Wang, D., Lu, W., Li, H., Delafield, D. G., Kong, Y., Wang, X., Shao, C., Lv, L., Wang, G., Tan, R., Wang, N., Hao, H., & Ye, H. (2022). Cyclic immonium ion of lactyllysine reveals widespread lactylation in the human proteome. Nat. Methods 19, 854–864.
Qin, W., Yang, F., & Wang, C. (2020). Chemoproteomic profiling of protein-metabolite interactions. Curr. Opin. Chem. Biol. 54, 28–36.
Weerapana, E., Speers, A. E., & Cravatt, B. F. (2007). Tandem orthogonal proteolysis-activity-based protein profiling (TOP-ABPP)--a general method for mapping sites of probe modification in proteomes. Nat. Protoc. 2, 1414–1425.
Cappelletti, V., Hauser, T., Piazza, I., Pepelnjak, M., Malinovska, L., Fuhrer, T., Li, Y., Dörig, C., Boersema, P., Gillet, L., Grossbach, J., Dugourd, A., Saez-Rodriguez, J., Beyer, A., Zamboni, N., Caflisch, A., de Souza, N., & Picotti, P. (2021). Dynamic 3D proteomes reveal protein functional alterations at high resolution in situ. Cell 184, 545–559.e22.
Lomenick, B., Hao, R., Jonai, N., Chin, R. M., Aghajan, M., Warburton, S., Wang, J., Wu, R. P., Gomez, F., Loo, J. A., Wohlschlegel, J. A., Vondriska, T. M., Pelletier, J., Herschman, H. R., Clardy, J., Clarke, C. F., & Huang, J. (2009). Target identification using drug affinity responsive target stability (DARTS). Proc. Natl. Acad. Sci. USA 106, 21984–21989.
Savitski, M. M., Reinhard, F. B., Franken, H., Werner, T., Savitski, M. F., Eberhard, D., Martinez Molina, D., Jafari, R., Dovega, R. B., Klaeger, S., Kuster, B., Nordlund, P., Bantscheff, M., & Drewes, G. (2014). Tracking cancer drugs in living cells by thermal profiling of the proteome. Science 346, 1255784.
Martinez Molina, D., Jafari, R., Ignatushchenko, M., Seki, T., Larsson, E. A., Dan, C., Sreekumar, L., Cao, Y., & Nordlund, P. (2013). Monitoring drug target engagement in cells and tissues using the cellular thermal shift assay. Science 341, 84–87.
Zhang, X., Wang, Q., Li, Y., Ruan, C., Wang, S., Hu, L., & Ye, M. (2020). Solvent-induced protein precipitation for drug target discovery on the proteomic scale. Anal. Chem. 92, 1363–1371.
West, G. M., Tang, L., & Fitzgerald, M. C. (2008). Thermodynamic analysis of protein stability and ligand binding using a chemical modification- and mass spectrometry-based strategy. Anal. Chem. 80, 4175–4185.
Reinhard, F. B., Eberhard, D., Werner, T., Franken, H., Childs, D., Doce, C., Savitski, M. F., Huber, W., Bantscheff, M., Savitski, M. M., & Drewes, G. (2015). Thermal proteome profiling monitors ligand interactions with cellular membrane proteins. Nat. Methods 12, 1129–1131.
Geiger, R., Rieckmann, J. C., Wolf, T., Basso, C., Feng, Y., Fuhrer, T., Kogadeeva, M., Picotti, P., Meissner, F., Mann, M., Zamboni, N., Sallusto, F., & Lanzavecchia, A. (2016). L-arginine modulates T cell metabolism and enhances survival and anti-tumor activity. Cell 167, 829–842.e13.
Raines R. T. (1998). Ribonuclease A. Chem. Rev. 98, 1045–1066.
Dombrauckas, J. D., Santarsiero, B. D., & Mesecar, A. D. (2005). Structural basis for tumor pyruvate kinase M2 allosteric regulation and catalysis. Biochemistry 44, 9417–9429.
Anastasiou, D., Yu, Y., Israelsen, W. J., Jiang, J. K., Boxer, M. B., Hong, B. S., Tempel, W., Dimov, S., Shen, M., Jha, A., Yang, H., Mattaini, K. R., Metallo, C. M., Fiske, B. P., Courtney, K. D., Malstrom, S., Khan, T. M., Kung, C., Skoumbourdis, A. P., Veith, H., et al. (2012). Pyruvate kinase M2 activators promote tetramer formation and suppress tumorigenesis. Nat. Chem. Biol. 8, 839–847.
Jarzab, A., Kurzawa, N., Hopf, T., Moerch, M., Zecha, J., Leijten, N., Bian, Y., Musiol, E., Maschberger, M., Stoehr, G., Becher, I., Daly, C., Samaras, P., Mergner, J., Spanier, B., Angelov, A., Werner, T., Bantscheff, M., Wilhelm, M., Klingenspor, M., et al. (2020). Meltome atlas-thermal proteome stability across the tree of life. Nat. Methods 17, 495–503.
Schrader, J., Henneberg, F., Mata, R. A., Tittmann, K., Schneider, T. R., Stark, H., Bourenkov, G., & Chari, A. (2016). The inhibition mechanism of human 20S proteasomes enables next-generation inhibitor design. Science 353, 594–598.
Hacker, S. M., Backus, K. M., Lazear, M. R., Forli, S., Correia, B. E., & Cravatt, B. F. (2017). Global profiling of lysine reactivity and ligandability in the human proteome. Nat. Chem. 9, 1181–1190.
Khan, J. A., Tao, X., & Tong, L. (2006). Molecular basis for the inhibition of human NMPRTase, a novel target for anticancer agents. Nat. Struct. Mol. Biol. 13, 582–588.
Lambert, S. A., Jolma, A., Campitelli, L. F., Das, P. K., Yin, Y., Albu, M., Chen, X., Taipale, J., Hughes, T. R., & Weirauch, M. T. (2018). The human transcription factors. Cell 172, 650–665.
Kim, W. J., Wittner, B. S., Amzallag, A., Brannigan, B. W., Ting, D. T., Ramaswamy, S., Maheswaran, S., & Haber, D. A. (2015). The WTX tumor suppressor interacts with the transcriptional corepressor TRIM28. J. Biol. Chem. 290, 14381–14390.
Li, M., Xu, X., Chang, C. W., & Liu, Y. (2020). TRIM28 functions as the SUMO E3 ligase for PCNA in prevention of transcription induced DNA breaks. Proc. Natl. Acad. Sci. USA. 117, 23588–23596.
Doyle, J. M., Gao, J., Wang, J., Yang, M., & Potts, P. R. (2010). MAGE-RING protein complexes comprise a family of E3 ubiquitin ligases. Mol. Cell 39, 963–974.
Addison, J. B., Koontz, C., Fugett, J. H., Creighton, C. J., Chen, D., Farrugia, M. K., Padon, R. R., Voronkova, M. A., McLaughlin, S. L., Livengood, R. H., Lin, C. C., Ruppert, J. M., Pugacheva, E. N., & Ivanov, A. V. (2015). KAP1 promotes proliferation and metastatic progression of breast cancer cells. Cancer Res. 75, 344–355.
Iyengar, S., & Farnham, P. J. (2011). KAP1 protein: an enigmatic master regulator of the genome. J. Biol. Chem. 286, 26267–26276.
Pineda, C. T., Ramanathan, S., Fon Tacer, K., Weon, J. L., Potts, M. B., Ou, Y. H., White, M. A., & Potts, P. R. (2015). Degradation of AMPK by a cancer-specific ubiquitin ligase. Cell 160, 715–728.
Jürchott, K., Kuban, R. J., Krech, T., Blüthgen, N., Stein, U., Walther, W., Friese, C., Kiełbasa, S. M., Ungethüm, U., Lund, P., Knösel, T., Kemmner, W., Morkel, M., Fritzmann, J., Schlag, P. M., Birchmeier, W., Krueger, T., Sperling, S., Sers, C., Royer, H. D., et al. (2010). Identification of Y-box binding protein 1 as a core regulator of MEK/ERK pathway-dependent gene signatures in colorectal cancer cells. PLoS Genet. 6, e1001231.
Qin, G., Wang, X., Ye, S., Li, Y., Chen, M., Wang, S., Qin, T., Zhang, C., Li, Y., Long, Q., Hu, H., Shi, D., Li, J., Zhang, K., Zhai, Q., Tang, Y., Kang, T., Lan, P., Xie, F., Lu, J., et al. (2020). NPM1 upregulates the transcription of PD-L1 and suppresses T cell activity in triple-negative breast cancer. Nat. Commun. 11, 1669.
Arito, M., Nagai, K., Ooka, S., Sato, T., Takakuwa, Y., Kurokawa, M. S., Sase, T., Okamoto, K., Suematsu, N., & Kato, T. (2015). Altered acetylation of proteins in patients with rheumatoid arthritis revealed by acetyl-proteomics. Clin. Exp. Rheumatol. 33, 877–886.
Das, C., Lucia, M. S., Hansen, K. C., & Tyler, J. K. (2009). CBP/p300-mediated acetylation of histone H3 on lysine 56. Nature 459, 113–117.
Williams R. J. (2001). Trichostatin A, an inhibitor of histone deacetylase, inhibits hypoxia-induced angiogenesis. Expert Opin. Investig. Drugs 10, 1571–1573.
Huang, J. P., & Ling, K. (2017). EZH2 and histone deacetylase inhibitors induce apoptosis in triple negative breast cancer cells by differentially increasing H3 Lys²⁷ acetylation in the BIM gene promoter and enhancers. Oncol. Lett. 14, 5735–5742.
Liu, Y., Tong, Y., Yang, X., Li, F., Zheng, L., Liu, W., Wu, J., Ou, R., Zhang, G., Hu, M., Liu, Z., & Lu, L. (2016). Novel histone deacetylase inhibitors derived from Magnolia officinalis significantly enhance TRAIL-induced apoptosis in non-small cell lung cancer. Pharmacol. Res. 111, 113–125.
Zheng, X., Cai, X., & Hao, H. (2022). Emerging targetome and signalome landscape of gut microbial metabolites. Cell Metab. 34, 35–58.
Delzenne, N. M., & Bindels, L. B. (2018). Gut microbiota in 2017: Contribution of gut microbiota-host cooperation to drug efficacy. Nat. Rev. Gastroenterol. Hepatol. 15, 69–70.
Shukla, S. A., Bachireddy, P., Schilling, B., Galonska, C., Zhan, Q., Bango, C., Langer, R., Lee, P. C., Gusenleitner, D., Keskin, D. B., Babadi, M., Mohammad, A., Gnirke, A., Clement, K., Cartun, Z. J., Van Allen, E. M., Miao, D., Huang, Y., Snyder, A., Merghoub, T., et al. (2018). Cancer-germline antigen expression discriminates clinical outcome to CTLA-4 blockade. Cell 173, 624–633.e8.
Tang, D., Kang, R., Livesey, K. M., Cheh, C. W., Farkas, A., Loughran, P., Hoppe, G., Bianchi, M. E., Tracey, K. J., Zeh, H. J., 3rd, & Lotze, M. T. (2010). Endogenous HMGB1 regulates autophagy. J. Cell Biol. 190, 881–892.

Cell culture and chemicals

Human HCT 116 cells were purchased from Cell Bank/Stem Cell Bank, Chinese Academy of Sciences. Human HEK293T and A549 cells were obtained from American Type Culture Collection. All cells were grown at 37 ℃ with 5% CO₂. HCT116 cells were cultured in McCoy’s 5A Medium, HEK293T cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM), and A549 cells were cultured in RPMI 1640 medium purchased from Gibco (Grand Island, NY, USA). All media were supplemented with 10% fetal bovine serum (FBS) (Biological Industries, Kibbutz, Israel), 100 unit/mL penicillin and 100 μg/mL streptomycin (Thermo Fisher Scientific, San Jose, CA, USA). LC-MS grade water and acetonitrile (ACN) were obtained from Merck (Darmstadt, Germany). All chemicals were purchased from Sigma Aldrich (St Louis, MO, USA) unless otherwise specified.

TRAP workflow for purified/recombinant proteins

Model proteins including purified bovine RNase A and human recombinant PKM2 were incubated with their corresponding ligands or solvent for 1 hr at room temperature. The proteins were labeled with 2 μL of 0.5% CD₂O and 10 μL of 10 mM Borane Pyridine Complex (BPC) (J&K Scientific Ltd., Beijing, China, cat. no. 121499) for 30 min, and the reaction was quenched by the addition of NH₄HCO₃ to a final concentration of 50 mM and incubation for 20 min. The resultant mixtures were filtered with 10 kDa molecular weight cut-off (Merck, Darmstadt, Germany) by centrifugation at 12,000 g for 15 min. Then, the enriched proteins were denatured by 8 M urea for 30 min followed by incubation with 5 mM DTT at 56 ℃ for 30 min and subsequent incubation with IAM at room temperature in dark for 30 min. DTT was added again to neutralize the excess IAM for another 10 min. The sample was then diluted with 25 mM NH₄HCO₃ to make the concentration of urea less than 1 M, and overnight digested by trypsin in an enzyme/protein ratio of 1:40 (wt/wt) at 37 ℃. The digestion was quenched by formic acid (FA) followed by desalting with C₁₈ Ziptip (Waters, Milford, MA, USA). The desalted digest was dried and stored until further LC-MS/MS analysis.

TRAP workflow for human cancer cell line proteome

Cell lysates preparation

Cells were washed with cold phosphate buffered saline (PBS) for three times, scraped and centrifuged at 1,000 rpm for 5 min. The cell pellets were resuspended in mammalian protein extraction reagent (M-PER, Pierce/Thermo Fisher Scientific, cat. no. 78503) containing protease inhibitor (ApexBio Technology, Houston, TX, USA, cat. no. K1007) and phosphatase inhibitor (ApexBio Technology, cat. no. K1013), and lysed on ice for 30 min. The supernatants were collected by centrifugation at 18,000 g for 10 min. Protein concentration was determined by bicinchoninic acid (BCA) Protein Assay (Beyotime Biotechnology, Beijing, China, cat. no. P0011). The cell lysates were then diluted with M-PER lysis to 3 μg/μL and incubated with given metabolites or solvents for 1 hr at 25 ℃. Specifically, the glycolytic metabolites were administered at dosages detailed as follows: 200 μM D-Fructose 1,6-bisphosphate (FBP), 100 μM D-fructose 6-phosphate (F6P), 100 μM D-glucose 6-phosphate (G6P), 30 μM D-ribose 5-phosphate (R5P), 30 μM D/L-glyceraldehyde-3-phosphate (G3P), 10 μM D-2-phosphoglycerate (2PG), 10 μM D-3-phosphoglycerate (3PG), 5 μM phosphoenolpyruvate (PEP), 300 μM sodium pyruvate (Pyr), 2 mM sodium L-lactate (Lac). 2PG was purchased from Shanghai yuanye Bio-Technology Co., Ltd. The dosages of these metabolites were set based on their intracellular concentrations as previously reported^52,53.

Proteome labeling and preparation for the label-free quantification-TRAP workflow

Cell lysates (150 μL, 3 μg/μL) were labeled by the addition of 6 μL of 1% CD₂O and 90 μL of 10 mM BPC for 30 min. The reaction was quenched by incubating with 50 mM NH₄HCO₃for 20 min. Then, methanol, chloroform and water were added to the labeled lysate in order according to a ratio of 4:1:3:1 by volume followed by centrifugation at 12,000 rpm for 10 min to precipitate the proteome. The flaky precipitate was washed twice with methanol and re-solubilized by 8 M urea in 25 mM ammonium bicarbonate solution. Proteins were then reduced by 10 mM DTT by incubation at 56 ºC for 30 min, and alkylated by 40 mM IAM at 25 ºC for 20 min in dark. Additional DTT was added and allowed to react with excess IAM at room temperature for 10 min. The mixture was added with 25 mM NH₄HCO₃to dilute urea to 1 M followed by overnight digestion with sequencing-grade trypsin (Promega, Madison, WI, USA) at an enzyme/protein ratio of 1:40 (wt/wt) at 37 ºC. The digestion was quenched by FA addition to pH=3. The mixture was desalted with C₁₈ SepPak cartridges (Waters) and evaporated to dryness with vacuum centrifuge (Thermo Fisher Scientific). Samples were stored at -80 ºC prior to analysis.

Sample preparation for TMT-based multiplexed TRAP

Cell lysates were labeled and precipitated using the same protocol as the label-free quantification-TRAP workflow. The precipitate was resuspended in 8 M urea solution (in 50 mM Tris-HCl containing 10 mM EDTA, pH 8.0) followed by DTT reduction and IAM alkylation. To this was added LysC (Signalchem, cat. no. L585-31N-05) in an enzyme/substrate ratio of 1:400 (wt/wt), followed by incubation at 25 ºC for 4 h. The mixture was diluted with 25 mM NH₄HCO₃so that the final concentration of urea is less than 2 M, and then sequencing grade trypsin was added in an enzyme/substrate ratio of 1:50 for overnight digestion at 37 ºC. After quenching by FA, the mixture was desalted with reversed phase SPE cartridges (Waters) and dried with vacuum centrifuge. The sample was resuspended in 300 μL of 50 mM HEPES (pH 8.5) with the concentration of peptides determined by Pierce quantitative colorimetric peptide assay (Thermo Scientific, cat. no. 23275). A 60 μg peptide aliquot of each sample was labeled with TMT reagent for 1.5 hr according to the manufacturer’s instructions. The reaction was quenched by incubating with 10 μL 5% hydroxylamine for 15 min. Aliquots labeled with the 6-plex TMT reagents were combined and acidified with FA followed by evaporation to dryness. The labeled proteome is then desalted again by SPE cartridges and evaporated to dryness. The proteome was dissolved in HPLC phase A buffer (10 mM ammonium formate containing 5% ACN) and then injected into the sample loop of the UPLC system (Acuity, Waters) for fractionation. Phase B consists of ACN with 20% 10 mM ammonium formate aqueous buffer. The flow rate was set as 0.2 mL/min and the gradient was as follows: 0-5 min, 1% B; 5-79 min, 1-50% B; 79-81 min, 50-100% B; 81-98 min, 100% B; 98-100 min, 100-1% B; 100-120 min, 1% B. The effluent was collected every 1.5 min. Every 12 fractions were set as a cycle, and each fraction was combined with the fractions collected in the following cycles. The lyophilized fractions were dissolved in 60 μL of 0.1% FA followed by desalting with C₁₈ Ziptips and storage at -80 ºC prior to analysis.

Mass spectrometry

Intact mass measurement

RNase A (100 μM) and CDP/CTP (1 mM) was both dissolved in 25 mM ammonium acetate buffer and incubated for 30 min, and native MS measurement of the formed holo-complex was conducted on a TripleTOF 5600 system (SCIEX, Framingham, MA, USA) by direct infusion. The instrument was set to acquire over the m/z range of 100−2000 Da for TOF-MS scan.

For intact mass measurement of the dimethylated RNase A with and without ligand incubation, labeled RNase A was desalted by 3 kDa MWCO and analyzed on a C₄ column (4.6×150 mm, 3 µm, 300 Å, Sepax Technologies, Newark, DE, USA) on an LC-30 HPLC system (Shimadzu, Kyoto, Japan). The mobile phase consisting of 0.1% FA in water (phase A) and 0.1% formic acid in ACN (phase B) was delivered at a flow rate of 0.4 mL/min using a 15 min gradient program. The eluent was then introduced via ESI ion source into the TripleTOF 5600 system (AB SCIEX, Framingham, MA, USA) for mass measurement. Q-TOF analyzer was set to scan over the m/z range of 100-2000. The spectra were combined by summing across the chromatographic peak of labeled RNase A and deconvoluted using the SCIEX BioPharma View Software.

Label-free proteomic quantification

Data used for label-free quantification was acquired on a nanoACQUITY UPLC system coupled to SYNAPT G2-Si mass spectrometer (Waters). A C₁₈ trapping column (Waters Acquity UPLC M-Class, 0.18×20 mm, 5 μm, 100 Å) and a HSS T3analytical column (Waters Acquity UPLC M-Class, 75 μm×150 mm, 1.8 μm, 100 Å) were employed. Mobile phases A and B consist of 0.1% FA in water and 0.1% FA in ACN, respectively. A 60 min and 120 min length gradient of 1-40 % ACN at a flow rate of 300 nL/min was used for separation of recombinant protein digests and cell lysates samples, respectively. MS scan range was set to m/z 350-1500 with a scan time of 0.2 s, and MS/MS scan range was set to m/z 50-2000 using data-dependent acquisition (DDA). The top 10 abundant precursors were subjected to MS/MS fragmentation with a ramp CE set between low energy (14-19 eV) and elevated energy (60–90 eV) using a scan time of 0.15 s per function.

TMT-based MS3-level multiplexed quantification

Data used for the multiplexed TRAP workflow was collected on an Orbitrap Fusion Lumos mass spectrometer equipped with an EASY-nano LC 1200 liquid chromatography system (Thermo Fisher Scientific). Mobile phase A consisting of 0.1% FA in water and B consisting of 0.1% FA in ACN-H₂O (8:2 by volume) were delivered at a flow rate of 300 nL/min. The 75 μm capillary column was packed with 35 cm of Accucore 150 resin (2.6 μm, 150 Å, Thermo Fisher Scientific). Peptides were analyzed using a 150 min chromatography gradient from 0%-50% phase B during 5-79 min. For MS data acquisition, MS1 spectra were collected at the m/z range of 375-1500 at a resolution of 120,000 in the Orbitrap with a maximum injection time of 50 ms or a maximum automated gain control (AGC) value of 4e⁵. For MS2 acquisition, fragmentation was conducted by collision-induced dissociation with a normalized collision energy (NCE) at 35. MS2 spectra were collected at the mass range of 400-1200 in ion trap with a maximum AGC of 1e⁴ or a maximum injection time of 50 ms. For accurate quantification, MS3 were conducted for TMT reporter ion quantification by high energy collision-induced dissociation (HCD) with NCE at 65. The MS3 spectra were collected over the mass range of 100-500 at a resolution of 50,000 with the maximum injection time set at 10⁵ ms and AGC target value at 1e⁵.

Proteomic data analysis and bioinformatics

Protein identification and quantification

The acquired label-free DDA data and TMT-MS3 data were searched against the Homo sapiens UniProt database (version 2018) using PEAKS Studio 8.5. Due to the necessary lysine labeling step employed by TRAP, we allowed up to two missed cleavages and semi-specific tryptic digestion. Carboxyamidomethylation on cysteines (+57.02 Da) was selected as fixed modification, and methionine oxidation (+15.99 Da), CD₂O-mediated dimethylation (+32.06 Da) and mono-methylation (+16.03 Da) on lysines were set as variable modifications. For label-free quantification data, precursor mass tolerance was set to 20 ppm, and fragment mass tolerance was set to 0.1 Da. For TMT-based MS3 quantification data, precursor mass tolerance was set to 10 ppm, MS2 fragment mass tolerance was set to 0.6 Da, and MS3 fragment mass tolerance was set at 0.02 Da. The identified proteins were filtered with 1% FDR and the quantified proteins must include at least one unique peptide. Regarding label-free quantification, we set 50 ppm mass tolerance and 3 min retention time shift tolerance for peptide alignment.

Classification of quantified peptides and proteins

We assessed potential glycolytic targetome by removing proteins associated with keratin firstly from the pool of quantified proteins, and only considered the quantified peptides which belong to the following three types as TRP candidates that are described as follows: a. peptides contain K residues that carry dimethylation or methylation and are not located at the C-terminus; b. peptides contain K, and the K does not carry dimethylation or methylation modification and are located at the C-terminus; c. peptides may not contain K, but the amino acid next to the N-terminal residue is K. Accessibility of peptides that can fit into the above types can be probed by TRAP, and are thus considered for TRP screening.

Determination of TRPs for drugs

As for target screening based on label-free quantification data, the screening standard of TRPs was set as peptides displaying p value < 0.01 and TRAP RatioR_drug/control > 2 or < 0.5 in the presence of the assayed drugs relative to the control. For drug targets discovery made based on the TMT-aided multiplexed TRAP data, a more stringent cutoff was utilized for TRP screening: peptides displaying p value < 0.0001 and TRAP RatioR_{drug/control)} > 2 or < 0.5 in the presence of ligands relative to the control.

Determination of TRPs for metabolites

For quantitative screening of the TRPs for the metabolites of interest, we first classified all quantified peptides into two categories, loose and compact. The loose category refers to the TRPs that become more chemically accessible to TRAP labeling reagents CD₂O after incubation with given metabolites. The increased chemical accessibility is assigned based on the increased abundance of peptides that carry dimethylated K (type a) or decreased abundance of peptides belonging to type b and c. Conversely, the compact category refers to the peptides that become less accessible to TRAP labeling after the administration of given metabolites. The reduced chemical accessibility is judged by the decreased abundance of type a peptides or increased abundance of type b and c peptides.

We set the standard of TRP screening for metabolites as peptides that display significant abundance changes with q value < 0.03 (q values are used to adjust for multiple testing using the Benjamini-Hochberg method to control the FDR at the cut-off level of 0.03) and the TRAP Ratio R_{treated/control} > 1.5 or < 0.67 for compact peptides while R_{treated/control} > 2 or < 0.5 for loose peptides. We posit that loose peptides reflect indirect binding events or conformational changes induced by direct binding, so more stringent criteria was given for this category.

Quality assessment of TRAP results

In order to assess the quality of our results, we estimated true positive rate generated by the TRAP approach by modifying previous assessment method⁹. We collected the known interactions from the BRENDA repository and set the species as homo sapiens. Among all potential enzyme-metabolite interactions, 140 known enzyme-metabolite interactions for homo sapiens in BRENDA were retrieved. Our TRAP results detected 30 known enzyme-metabolite interactions that are classified as true positive hits, which confer a true positive rate as 21.43% (calculated by 30/140).

Volcano plot of the TRAP-identified targetome

In the volcano plot, each point accords to a protein that is represented by a peptide selected based on a scoring system. The score of each quantified peptide for given proteins is obtained by consideration of both the TRAP ratio R_{treated/control} and p/q value of the peptide abundances between samples with and without given ligands shown as follows.

TRAP Score = -Log₁₀(p value/q value) * Abs[log₂(Ratio_{treated/control})]

After scoring, the peptide with the maximum TRAP score was selected to represent the given protein.

Secondary structure analysis

UniProt identifiers of all quantified proteins were matched with PDB accession number from Protein Data Bank (PDB) (http://www.rcsb.org/pdb/search/searchModels.do) and only entries with >90% sequence identity were retrieved for analysis. Secondary structure information of the compiled protein pool was downloaded from the DSSP (https://swift.cmbi.umcn.nl/gv/dssp) database and Python script was used to extract the secondary structure for each quantified lysine residue⁵⁴. The extracted secondary structure classes were classified into four categories, namely helix (DSSP classes H, G, I), sheet (DSSP classes B, E), loop (T, S) and no structure (“ ”).

Conservation analysis

To estimate the sequence conservation of the obtained Lysine sites from TRPs, we calculated the Lysine sequence identity across 11 representative vertebrate species (Human, Rhesus monkey, Mouse, Rat, Cow, Dog, Opossum, Chicken, Frog, Zebrafish, and Fugu) using an in-house perl script. Specifically, we downloaded multiple amino acid sequence alignment of coding sequence (CDS) region across 100 species (multiz100way) from UCSC Genome Browser⁵⁵. For each gene, its CDS region alignment across the selected 11 representative vertebrate species was further extracted. For each lysine from TRPs, its sequence conservation was estimated using the percentage of sequence conservation across 11 representative vertebrate species. To further examine whether the obtained TRP lysine sites were more conserved than random expectation, we calculated the sequence conservation of all quantified Lysine sites to estimate the background Lysine conservation, and then use Kolmogorov-Smirnov (KS) test to evaluate the statistical significance of excessive sequence conservation of the obtained TRP Lysine residues. The KS test p <0.05 was considered as significant.

Measurements of Euclidean distances

For functional sites analysis, PyMOL-Python scripts was used to measure the Euclidean distances between the atoms of lysine in TRPs and any atoms of annotated ligands (such as substrates, cofactors, products) in Å for enzymes assigned as glycolytic metabolites targets by TRAP for those with available structures retrieved from PDB files. The minimum distance was recorded to represent each ligand-target pair. Further, if the minimum distance is less than 10 Å, the lysine is categorized as functional and otherwise as unfunctional.

The active site boundary detectable by TRAP is defined based on the median of the minimum distances measured between the TRPs of known enzymatic targets that use the examined metabolites as substrates and their corresponding active sites.

For evaluation of the metabolites’ influence on given enzymes’ activities, the minimum distance between all TRPs from the identified targets involved in the carbohydrate metabolism pathway and their corresponding active sites were measured and shown in Fig. 5c.

GO and KEGG pathway analysis

All the TRAP-identified target proteins were annotated to non-overlapping GO molecule function (MF) terms. The MF classification was performed using the functional annotation tool of PANTHER (http://pantherdb.org/). GO MF terms include Transcription regulator activity (GO: 0140110), Catalytic activity (GO:0003824), Transporter activity (GO:0005215), Molecular transducer activity (GO:0060089), Translation regulator activity (GO:0045182), Structural molecule activity (GO:0005198), Molecular function regulator (GO:0098772), Binding (GO:0005488). If a protein belongs to more than one class, they are categorized based on the above order to prioritize. Proteins that do not match any of the above GO terms were sorted into the category “Uncategorized”.

Specifically, for proteins sorted in the “Catalytic activity” category, ClueGO app included in Cytoscape (version 3.7.1) was employed to perform the KEGG pathway annotation network analysis. A setting of group p value <0.001 and inclusion of at least three genes in each group was used for filtering.

GO biological process (BP) analysis of the druggable glycolytic targetome was performed by BiNGO app included in Cytoscape with a setting of group p value <1e-15.

DrugBank analysis for ligandability classification

The DrugBank database (v. 5.1.3 released on 2019-04-02; group "all_ target_polypeptide_ids") were downloaded and used to classify the glycolytic targetome into DrugBank (ligandable) and non-DrugBank (unligandable) proteins.

Analysis of lysine PTM

The PTM information of lysines from all quantified proteins was collected from the iPTMnet database (https://research.bioinformatics.udel.edu/iptmnet). Glycolytic metabolite-target interaction is considered to potentially affect the lysine-bearing PTM status if a given lysine in TRPs is reported to carry PTM.

Co-immunoprecipitation experiment

Sample preparation

Cultured HEK293T cells were first transfected with plasmids encoding FLAG-TRIM28 or hygro-Negative Control vector using Polyjet transfection reagents for 16 hr at 37 °C following the manufacturer’s instructions. After transfection, the cells were treated with 10 mM sodium lactate or the matching solvent for 24 hr. Then, the cells were harvested and lysed in 1% NP-40 lysis buffer (P0013F, Beyotime Biotechnology, Shanghai, China) with protease inhibitor cocktail and phosphatase inhibitor cocktail. The supernatants were collected by centrifugation at 18,000 g for 10 min. Protein concentrations were determined by BCA assay. For co-immunoprecipitation (co-IP) experiments, approximately 1 mg of protein lysates were incubated overnight at 4°C with anti-FLAG antibody. The lysates were then incubated with Protein A/G agarose beads (Invitrogen, cat. no. 15918014) for 4 hr at 4°C. The beads-bound proteins were then eluted with SDS loading buffer, heated at 95°C for 10 min and then collected for immunoblotting. For analysis of the enriched interactome, on beads-digestion and reductive dimethylated-based labeling were performed. Briefly, the enriched proteins were washed with PBS for three times followed by incubation with urea, DTT, IAA and trypsin for overnight digestion. The resultant mixtures were centrifuged at 10,000 rpm, and the supernatants were collected and desalted with C₁₈ SepPak cartridges. The digested peptides were re-solubilized in 100 mM TEAB buffer (ThermoFisher, cat. no. 90114) and labeled by the addition of 4 μL 4% CH₂O (light-label group) (Sigma, cat. no. 47608-250ML-F) or CD₂O (heavy-label group) and 4 μL 0.6 M NaBH₄CN (Sigma, cat. no. 452882) for 1 hr at room temperature. The dimethylation reaction was quenched by with the addition of 16 μL 1% ammonia for 15 min followed by further termination with 4 μL FA. Then, the light and heavy-labeled samples were combined, evaporated, and reconstituted in 0.1% FA aqueous solution followed by desalting with C₁₈ZipTips. The labeled interactome samples were then evaporated to dryness and subjected to proteomic analysis as previously described.

Data acquisition and analysis:

The acquired light and heavy-dimethylated labeled proteomics data were searched against the Homo sapiens UniProt database (version 2021.09.15, entry 20413) using PEAKS Studio 8.5. Up to three missed cleavages and semi-specific tryptic digestion were allowed. Carboxyamidomethylation on cysteines (+57.02 Da) was selected as fixed modification, and methionine oxidation (+15.99 Da), CD₂O-mediated dimethylation (+32.06 Da) and CH₂O-mediated methylation (+28.03 Da) on lysines were set as variable modifications. For dimethylation labeling-based quantitative proteomics data, precursor mass tolerance was set to 10 ppm, and fragment mass tolerance was set to 0.02 Da. The interacting proteins were screened following the criteria: the protein must have at least 2 unique peptides and can only be detected in the TRIM28-overexpression group (detactable in all replicates) rather than the vector control group. Then, the abundances of the TRIM28-interacting proteins were normalized to the abundance of TRIM28 in each group to obtain the Relative Affinity. Affinity Ratio is further obtained by comparing the Relative Affinity of each TRIM28-interacting proteins with and without lactate. The protein-protein network was built by Cytoscape and the annotated known PPIs were retrieved from databases including STRING, BioGRID, PrePPI, and IID (Integrated Interactions Database). The colors of the nodes represent the changed binding affinity to TRIM28 following lactate treatment, and the colors of the frames represent the frequency of retrieving the identified PPIs in this study from the four databases.

Immunoblotting

Cells were lysed in RIPA lysis buffer supplemented with protease inhibitor cocktail. The protein concentrations were then determined by the BCA assay. The lysates were diluted by 4× XT Sample Buffer, heated to 95 ℃ for 5 min, cooled and separated by 8%-12% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). After being transferred onto polyvinylidene difluoride (PVDF) membranes, proteins were blocked using 5% non-fat dry milk in Tris-buffered saline with 0.1% Tween 20 detergent (TBST) and incubated with primary antibodies at 4 °C overnight. After being washed five times with TBST, the membranes were incubated with HRP-conjugated secondary antibodies for 1 hr at 37 ℃. The immunoblotted bands were detected by the addition of horseradish peroxidase (HRP) substrate and captured on a ChemDoc XRS+ System (Bio-Rad). The primary antibodies used in this study include antibodies against β-tubulin (AP0064, Bioworld Technology), TRIM28 (, 4123, Cell Signaling Technology), ENO1 (3810, Cell Signaling Technology), α-tubulin (11224-1-AP, Proteintech), FLAG tag for WB (8146, Cell Signaling Technology), FLAG tag for immunoprecipitation (14793, Cell Signaling Technology), Histone H3 (4499, Cell Signaling Technology, Acetylated H3K27 (PTM-116RM, PTM Biolabs), cleaved PARP (5625, Cell Signaling Technology), cleaved caspase-3 (9661, Cell Signaling Technology).

Subcloning and mutagenesis

The synthetic genes encoding wild type (WT) human ENO1 (UniProt: P06733) and PKM2 (UniProt: P14618) with an N-terminal hexahistidine purification (His6) tag were subcloned into the GV296 vector (Genechem, Shanghai, China). The synthetic genes encoding WT human TRIM28 with an N-terminal maltose binding protein (MBP) affinity tag (UniProt: Q13263) were subcloned to the pMAL-p2X vector (New England Biolabs, Beverly, MA, USA). The ENO1 K330E and TRIM28 K337E/K340E double mutants were obtained by site-directed mutagenesis. PCR primer sequences used for ENO1 and TRIM28 mutagenesis are listed as below:

Human ENO1 K330E	F: GATCGCCGAGGCCGTGAACGAGAAGTCCTG
	R: ACGGCCTCGGCGATCCTCTTTGGGTTGGTC
Human TRIM28 K335E	F: GAACGAGGAGTCCTGCAACTGCCTCCTGCT R: CAGGACTCCTCGTTCACGGCCTTGGCGATC
Human TRIM28 K337E/K340E	F: GAGATCCAGGAGCACCAGGAGCACATTCTG R: CTGGTGCTCCTGGATCTCGGTCATGGTCCAGTGCTGCC

Recombinant protein expression and purification

All His-tagged recombinant protein expression and purification steps are generally as follows: plasmids were transformed into BL21(DE3) E. coli cells and the transformed bacteria were selected on LB plate containing 50 µg/mL kanamycin. The isolated colonies were grown in LB medium at 37 ℃ and 220 rpm until the OD₆₀₀of culture arrive 0.8. The target protein expression was induced with 0.5 mM isopropyl β-D-thiogalactoside (IPTG) by shaking overnight at 16 ℃. Cells were harvested by centrifugation followed by resuspension in ultrapure water containing protease inhibitor cocktail, and then lysed by sonication. Lysates were centrifuged for 20 min at 18,000 g at 4 ℃ and the resultant supernatants were discarded. The His-tagged target proteins were then purified with a His-tag Protein Purification Kit (Beyotime Biotechnology, cat. no. P2226). After purification, proteins were concentrated by filter with a 10 kDa cut-off and the purity was evaluated by SDS-PAGE.

For the expression of TRIM28 constructs, cultures were supplemented with LB medium containing 50 µg/mL ampicillin and 0.5% glucose. Protein expression was induced at OD₆₀₀ = 0.8 with 0.2 mM IPTG by shaking at 100 rpm overnight at 16 ℃. To purify the MBP-tagged TRIM28, cells were resuspended in lysis buffer (50 mM Tris pH 7.4, 50mM NaCl, 5 mM DTT, 1:10,000 (v/v) benzonase solution, 1×protease inhibitor cocktail and lysed by sonication. The lysate was clarified by centrifugation. The supernatant was transferred to amylose resin (New England Biolabs) and incubated overnight at 4 ℃. The protein was purified using a New England Biolabs kit according to the manufacturers’ protocol. The bound fusion protein was eluted with 10 mM maltose (New England Biolabs). The yield of the fusion proteins was evaluated by separation on SDS-PAGE and visualization via Coomassie blue staining.

PKM2 activity assay

In vitro activity of PKM2 was performed using luminescent Kinase-Glo Plus reagent (Promega, cat. no. V3773) and used according to the manufacturer’s instructions. In brief, human recombinant PKM2 was diluted in the assay buffer (50 mM Tris HCl, 100 mM KCl, 10 mM MgCl₂, PH 7.4) and incubated with each metabolite/drug/vehicle at 25 ℃ for 40 min. Substrate solution was prepared by mixing ADP at 200 μM and PEP at 200 μM with the assay buffer. Then, a 50 μL aliquot of substrate solution was added to 50 μL recombinant PKM2 solution, and allowed to react for 10 min at room temperature in 96-well plates. The luminescence response of each sample was measured due to the formation of ATP at 37 ℃ in a Gen5 platform (BioTek). The measurements were fitted by a nonlinear fitting algorithm (log [agonist] vs. response-variable slope with 4 parameters) in Prism 8.0.1 (GraphPad, San Diego, CA, USA).

GAPDH activity assay

In vitro GAPDH enzymatic activity assay was performed in 96-well plates at room temperature by measuring the reduced NAD⁺ level. The recombinant GAPDH (Abnova, Taiwan, China, cat. no. P4547) was first diluted in 10 mM sodium pyrophosphate buffer (pH 8.5) to 30 U/mg. A 100 μL of reaction mixture containing 20 mM sodium arsenate, 1 mM NAD⁺ and 2.88 mM G3P was readily added to GAPDH. The NAD⁺ level was examined by measuring the absorbance at 340 nm every 20 s for 20 min. Reaction rate was calculated by curve fitting of the time course measurements by linear regression.

NAMPT activity assay

In vitro activity of NAMPT was measured using a NAMPT colorimetric assay kit (CycLex, MBL International, cat. no. CY 1251V2) in 96-well plates by the one-step method. In brief, recombinant NAMPT was diluted in NAMPT assay buffer and subsequently incubated with solvent/metabolites/FK866 for 30 min at 30 ℃. The reaction was initiated by the addition of 60 μL One-Step Assay Mixture to each well. NAMPT activity was measured photometrically in absorbance at 450 nm for 60 min with a 1 min-interval.

Thermal shift assay and isothermal dose-response assay

For thermal shift assay conducted using cell lysate samples, cultured HCT116 cells were harvested and washed with PBS. The washed cells were diluted in 0.1% NP-40 lysis buffer supplemented with 1× protease inhibitor cocktail and 1× phosphatase inhibitor cocktail. The cell suspensions were freeze-thawed five times using liquid nitrogen and passed through a 27” gauge needle five times. The samples were snap-frozen in liquid nitrogen for > 1 min and placed onto a heating block set at 25 °C until 60% of the confluent was thawed. Then, the samples were subjected to centrifugation at 20,000 g for 20 min at 4 °C. The supernatant was collected and divided into two aliquots, with one aliquot being treated with given glycolytic metabolites and another aliquot with the solvent as the control. After 60 min-incubation at room temperature, each sample were further divided into 8 aliquots and heated individually at their designated temperature for 3 min in a 96-well thermal cycler followed by cooling at room temperature for 3 min. The heated lysates were centrifuged at 20,000 g for 20 min at 4 °C in order to separate the soluble fractions from precipitates. Each resultant soluble fraction of the proteome was transferred to a new low-adsorption 1.5 mL microtube and analyzed by SDS-PAGE followed by western blotting for target proteins.

For the cell lysate-level isothermal dose-response (ITDR) experiments, G3P were serially diluted to generate a 9 points dose–response curve. HCT116 cells lysate were incubated with the assayed metabolite of serial concentrations (at least eight concentrations) and one vehicle as control in 1.5 mL low absorption microtubes for 1 hr at room temperature. Then, the samples were aliquoted to 120 μL and transferred into 200 μL PCR microtubes followed by heating at designated temperatures for 3 min in a 96-well thermal cycler and subsequent cooling for 3 min at room temperature. Then, isolation of the soluble fractions and immunoblotting of given target protein are repeated as described for thermal shift assays.

For cell lysate-level thermal shift assay, the band intensities detected at increasing temperatures were normalized to that of the lowest temperature, and the Boltzmann sigmoid equation was fitted using GraphPad Prism. For the ITDR experiments, the band intensities at increasing doses were normalized to the intensity at the highest concentration of the assayed metabolites and analyzed by the saturation binding curve function in Prism.

Surface plasmon resonance analysis

SPR analysis was conducted on a Biacore T200 system (GE Healthcare, Sweden). Target protein was diluted in 10 mM sodium acetate and immobilized via the amine coupling method on a CM5 sensor chip. Metabolites were dissolved in H₂O and diluted to a serial concentration with running buffer (PBS with 0.05% tween 20). Then, the metabolites were injected through the reference and active channels at a flow rate of 30 μL/min. The association and dissociation times were both set at 60 s. The affinity fitting was carried out on a Biacore T200 evaluation software by global fitting via a steady-state affinity model to obtain the equilibrium dissociation constant K_D.

Protein thermal stability analysis by nanoDSF

Both the WT-ENO1 (0.5 μg/μL) and the mutant-ENO1 (0.5 μg/μL) were incubated with 2 mM G3P for 30 minutes firstly. Samples were filled within the nanoDSF capillaries (n=3), and subsequently loaded into the Prometheus NT.48 device (NanoTemper Technologies, Germany). Samples were heated from 20 °C to 95 °C with a slope of 1 °C/min and the unfolding transition temperatures were automatically identified by the PR. ThermControl software (NanoTemper Technologies). Raw data was exported for plotting the thermal stability curve in GraphPad Prism.

In vitro lysine acetylation (Kac) assay

The in vitro Kac assay was performed by following literature⁵⁶. Briefly, the human recombinant ENO1 and Histone 3.3 (10 μg) were diluted with the reaction buffer (50 mM Tris-HCl, pH 8.0, 100 nM trichostatin A, 0.1 mM EDTA, 1 mM DTT and protease inhibitor cocktail) and incubated with G3P and pyruvate, respectively, for 30 min. Recombinant Histone 3.3 was purchased from New England Biolab (cat. no. M2507S).Then, the mixtures were incubated with 2 mM Acetyl-CoA (Shanghai yuanye Bio-Technology Co., Ltd, cat. no. Y34045) and 0.55 μg of the recombinant human p300 protein (Active Motif, cat. no. 81158) at 30 °C for 45 min for in vitro Kac. The Kac process was terminated by adding 8 M urea, and subsequently digestion was performed as previously described. The digested proteins were desalted with C₁₈ SepPak cartridges, evaporated to dryness and subjected to bottom-up analysis.

Flow cytometric analysis of cell apoptosis

Cell apoptotic rate was examined using an Annexin-V/FITC Kit (BD Biosciences, San Jose, CA, USA). Cells were pretreated with pyruvate for 4 hr and then treated with TSA for 24 hr. The cells were harvested, washed with PBS, resuspended in 1× Annexin V binding buffer, and stained with annexin V and PI for 15 min at room temperature in darkness. The rates of apoptotic cells were determined using an Accuri C6 flow cytometer (BD Biosciences).

Statistical analysis
Statistical analysis was performed in GraphPad Prism. All data represent mean ± SEM (n = 3 or 5 per group as indicated in legend). The statistical significance of differences between two groups was determined using Student’s t-test (unpaired, two-tailed) unless otherwise specified, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

References

52. Chen, W. W., Freinkman, E., Wang, T., Birsoy, K., & Sabatini, D. M. (2016). Absolute quantification of matrix metabolites reveals the dynamics of mitochondrial metabolism. Cell 166, 1324–1337.e11.

53. Miyo, M., Konno, M., Nishida, N., Sueda, T., Noguchi, K., Matsui, H., Colvin, H., Kawamoto, K., Koseki, J., Haraguchi, N., Nishimura, J., Hata, T., Gotoh, N., Matsuda, F., Satoh, T., Mizushima, T., Shimizu, H., Doki, Y., Mori, M., & Ishii, H. (2016). Metabolic adaptation to nutritional stress in human colorectal cancer. Sci. Rep. 6, 38415.

54. Feng, Y., De Franceschi, G., Kahraman, A., Soste, M., Melnik, A., Boersema, P. J., de Laureto, P. P., Nikolaev, Y., Oliveira, A. P., & Picotti, P. (2014). Global analysis of protein structural changes in complex proteomes. Nat. Biotechnol. 32, 1036–1044.

55. Haeussler, M., Zweig, A. S., Tyner, C., Speir, M. L., Rosenbloom, K. R., Raney, B. J., Lee, C. M., Lee, B. T., Hinrichs, A. S., Gonzalez, J. N., Gibson, D., Diekhans, M., Clawson, H., Casper, J., Barber, G. P., Haussler, D., Kuhn, R. M., & Kent, W. J. (2019). The UCSC Genome Browser database: 2019 update. Nucleic Acids Res. 47, D853–D858.

56. Huang, H., Tang, S., Ji, M., Tang, Z., Shimada, M., Liu, X., Qi, S., Locasale, J. W., Roeder, R. G., Zhao, Y., & Li, X. (2018). p300-mediated lysine 2-hydroxyisobutyrylation regulates glycolysis. Mol. Cell 70, 663–678.e6.

There is NO Competing Interest.

Download PDF

Journal Publication

published 15 Jun, 2023

Read the published version in Nature Chemical Biology →

Version 1

posted

You are reading this latest preprint version

Chemoproteomic Mapping of Glycolytic Targetome in Cancer Cells

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Results

TRAP detects ligand binding-induced accessibility changes on reactive lysines

Benchmarking the TRAP approach for targetome mining from cell milieu

TRAP identifies thermally stable drug targets

TRAP mapped a landscape of glycolytic targetome in human cancer cells

Glycolytic metabolites as promiscuous regulators of carbohydrate metabolism

Glycolytic metabolites modulate druggable enzymes activity

Glycolytic metabolites engage transcriptional factors and protein-protein interactions

Direct binding of glycolytic metabolites to target proteins affect acetylation levels

Discussion

Declarations

References

methods

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1