Photoproximity labeling of c-Myc reveals SLK as a cancer-specific co-regulator

Abstract

Transcription factors (TFs) are aspirational therapeutic targets, as their dysregulation drives altered cell states. Yet many disease-relevant TFs are disordered and lack canonical binding pockets, frustrating direct small-molecule inhibition. Indirectly targeting the effector molecules that modulate TF function is a promising, underexplored alternative. Here we report a strategy for capturing cancer-specific protein–protein interactions using context-dependent μMap photoproximity labeling. With an intein-based method for catalyst conjugation in biochemically intact nuclei, we capture unique c-Myc interactomes in healthy and cancerous prostate cells and mine them for druggable vulnerabilities. We identify STE20-like kinase (SLK), a cancer-specific interactor that stabilizes c-Myc, drives epithelial morphology and is essential for tumorigenesis. Mechanistically, SLK phosphorylates c-Myc at serine 329, antagonizing GSK3β-dependent phosphodegron phosphorylation. This interaction is associated with a splicing change promoting nuclear localization of the long SLK isoform. Patient data link this isoform to c-Myc target expression across tumor types; the interaction validates across diverse tissues.

Transcription factors (TFs) remain one of the most compelling yet elusive targets in cancer biology¹. They control gene expression programs that drive cancer progression, yet their disordered structures and lack of binding pockets have rendered direct inhibition challenging^2–4. TF activity depends on dynamic protein–DNA and protein–protein interactions, with these cistromic elements guiding localization and function⁵. These interactions have presented opportunities for therapeutic disruption, although disease specificity is rarely considered, leading to clinical challenges as many TFs are also essential for normal tissue homeostasis^6,7. Furthermore, the oncogenic makeup of any cancer can create a distinct context for TF interactomes that can result in divergent responses to inhibition of a particular protein–protein interaction^8,9. Despite this, context-dependent investigation of oncogenic TFs has yet to be realized.

The oncogenic TF c-Myc controls the expression of thousands of genes associated with growth and is deregulated in more than 40% of cancers¹⁰. This master regulator of oncogenesis is characterized by its complex network of effector proteins that modulate its function¹¹, and pharmacological interrogation of these interactors has been shown to be an effective method for treating c-Myc driven cancers in model systems^10,12–21. To identify co-regulating proteins, landmark studies have investigated the interactome of c-Myc through a variety of methods, including coimmunoprecipitation (co-IP) and BioID proximity labeling^22–25. These methods have led to long lists of putative interactors that have been challenging to apply to the development of therapies as they broadly label transcriptionally active regions of chromatin rather than direct c-Myc regulators. Furthermore, these approaches have never been applied contemporaneously to matched cancer cell lines to identify interactors specific to a single cancer phenotype or subtype.

We recently reported a nanoscale proximity labeling method (μMap) that bears a dramatically reduced labeling radius when compared to existing methods (4 nm versus 100 nm)^26,27. This short radius has been shown to enable proximity labeling experiments that are sensitive to small changes in protein structure and environment^26,28. Furthermore, μMap catalysts can be tracelessly incorporated onto histones using ultrafast split-intein splicing²⁶. Based on these precedents, we sought to translate this method to c-Myc and capture cancer-specific protein interactors. We reasoned that such a strategy would reveal context-dependent interactomes that could be mined to uncover protein–protein interactions modulating c-Myc at aberrant genetic loci unique to cancer cells (Fig. 1a). By integrating nanoscale proximity labeling with disease-specific experimental design, we can highlight co-regulators that would remain undetected using either approach alone.

Fig. 1 |. μMap identifies known c-Myc co-regulatory proteins in HEK293T cells.

a, Cartoon schematic of μMap photoproximity labeling in nucleo: the iridium photocatalyst is conjugated to c-Myc via ultrafast split-intein splicing; after 1 min of blue-light irradiation, diazirine-biotin probes form carbenes that crosslink proteins within ~4 nm, which are enriched and identified by MS. b, Volcano plot of the μMap MS data from HEK293T cells transfected with c-Myc-CfaN, plotting enrichment against significance; the bait (c-Myc) and known c-Myc coregulators are labeled. n = 3 biological replicates, with 2 technical replicates per biological replicate. P values were assessed using a two-sided Welch’s t-test with permutation-based FDR correction (FDR = 0.05, S0 = 0.1) in Perseus, where S0 is a constant that sets a minimum fold-change requirement to the significance test. c, Bar chart comparing number of c-Myc interactors identified by μMap with the co-IP²⁴ and BioID²² datasets. d, Venn diagram comparing proteins that are shared by all three methods, with MAX, EP400, TRRAP and EPC1 falling within the top 10% of μMap-enriched proteins. e,f, μMap enrichment of subunits of the NuA4/TIP60 (e) and the Ada-two-A-containing (ATAC) (f) histone acetyltransferase (HAT) complexes, mapped onto each complex.

Results

Nuclear μMap captures c-Myc interactomes

We initiated our study by generating full length c-Myc constructs bearing a C-terminal intein tag (CfaN) flanked by FLAG and hemagglutinin (HA) epitope tags for analytical convenience. Expression of this construct in human embryonic kidney 293T (HEK293T) cells showed a single band at 75 kDa by western blot that could be stained with FLAG and HA antibodies, suggesting the fidelity of the construct is maintained on expression (Extended Data Fig. 1a,b). Next, we demonstrated that the construct can splice with the CfaC peptides bearing the iridium photocatalyst. Incubation of biochemically intact nuclei with 0.5 μM CfaC-Ir for 40 min at 37 °C led to robust splicing, with approximately 67% conversion of the initial construct (Extended Data Fig. 1b). Further assay optimization identified 1.0 μM as the optimal concentration of CfaC-Ir for maximal labeling efficiency, and irradiation of spliced nuclei led to substantial biotinylation (Extended Data Fig. 1d). Immunofluorescent staining of transfected cells showed clean nuclear localization of the construct, in line with the known function of c-Myc (Extended Data Fig. 1c).

We next validated that our c-Myc construct remained functionally active and promoted the activation of Myc-dependent genes. RNA sequencing (RNA-seq) analysis showed similar changes in the expression of Myc-controlled genes (HK2, SORD, UNG, HNRNPC) when compared to an unmodified c-Myc plasmid, suggesting our construct maintains transcriptional activity (Extended Data Fig. 1e,f).

Confident in the validity of our proximity labeling system, we performed label-free chemoproteomics using data independent analysis to capture the interactome of c-Myc using our μMap method in HEK293T cells. Our first experiment compared the c-Myc interactome to an identical protocol without transfection of the c-Myc-CfaN transgene. Each condition was performed with three biological replicates with two technical replicates per biological replicate to account for any inconsistencies in the protocol. For hit determination, we established cutoffs of log₂(fold change) ≥0.25 and false discovery rate (FDR)-corrected P value <0.05, and hits were found in all positive replicates to be considered. We obtained 3,503 protein IDs with 3,291 IDs being found in >75% of all injections. Compartmental analysis of the identified proteins showed 62% of all IDs were localized to the nucleus, consistent with flow cytometry data from our nuclear isolation protocol (Extended Data Fig. 2a,b). Principal component analysis of all replicates showed a clear difference between untransfected conditions and c-Myc localized labeling (Extended Data Fig. 2c).

We observed strong enrichment of the c-Myc bait protein (log₂(fold change) 7.17, P = 1.78 × 10⁻¹³), in addition to 156 other hits that meet significance (Fig. 1b). We next compared our list of interactors with several other c-Myc interactomes from the literature to look for enrichment of common c-Myc transcriptional co-regulators. We used a published co-IP dataset and a BioID proximity labeling dataset^22,24. These datasets identified 418 and 336 hits, respectively, with 22 genes identified in both datasets. Our method provided a more restricted interactome of 157 hits and showed clear enrichment of 6 of these 22 common interactors, 4 of which were found in our top 10% enriched genes (Fig. 1c,d). These four proteins—MAX, EP400, TRRAP, EPC1—are well characterized c-Myc interactors^{23,25,29–31}. Furthermore, all of our top 40 most enriched genes were annotated interactors of c-Myc on the IntAct and/ or BioGRID databases (Extended Data Fig. 2d). GO analysis of the dataset showed significant enrichment of proteins associated with histone acetylation, transcription co-regulator activity and DNA-binding TF binding (Extended Data Fig. 2e,f). Specific analysis of our interactome showed significant enrichment of the Ada-two-A-containing (ATAC) and NuA4/TIP60 histone acetyltransferase complexes, known functional regulators of c-Myc^30,32 (Fig. 1e,f). We note that these co-IP and BioID datasets were generated in different cell lines, but the Flp-In 293 T-REx cells used in the BioID study are derived from HEK293 cells, providing a relevant like-for-like cellular context comparison with our μMap experiments. Our log2(fold change) threshold of ≥0.25 was applied in conjunction with an FDR-corrected P < 0.05 and a requirement that hits were present in all positive replicates; comparable thresholds were employed in the co-IP and BioID datasets for comparison. We note that data-independent acquisition label-free quantitation (DIA-LFQ) provides more complete and reproducible peptide quantitation than the data-dependent acquisition (DDA)-based approaches used in previous studies, and that the more restricted interactome produced by μMap reflects its nanoscale labeling radius (~4 nm) rather than reduced sensitivity, consistent with the enrichment of known direct co-regulators among our top hits. Finally, to probe the ability of our approach to measure subtle changes to the chromatin bound c-Myc interactome, we treated c-Myc-CfaN-transfected HEK29T cells with vehicle or 1 μM JQ-1 before photolabeling. In this experiment we observed clean enrichment of BRD2/3/4 only in the untreated samples, showing that JQ-1 blocks bromodomain and extra-terminal domain (BET) proteins from binding to chromatin and consequently disrupts the Myc-BET axis³³ (Extended Data Fig. 2g).

c-Myc interactome in prostate cell lines

Androgen receptor (AR) negative prostate cancer is an aggressive form of prostate cancer that no longer responds to frontline treatment options such as androgen deprivation therapy, resulting in poor prognoses³⁴. Expression of androgen receptors in these tumors is often low or null, and growth is promoted by c-Myc, with few treatment options available for intervention^35,36. We questioned whether our proximity labeling approach would be sensitive enough to identify cancer-specific transcriptional coactivators that may lead to c-Myc activation and subsequent growth and metastasis. To identify factors that may contribute to this effect, we compared the c-Myc interactome in three prostate cell lines that represent healthy prostate cells (WPMY-1), AR-negative prostate cancer (PC-3) and AR-positive prostate cancer (LNCaP) (Fig. 2a and Extended Data Fig. 3d). In both AR⁻ (PC-3) and AR⁺ (LNCaP) cell lines, c-Myc is primarily stabilized at the protein level rather than being transcriptionally upregulated, suggesting changes in protein structure or interactome may promote cancer progression through increased c-Myc activation (Extended Data Fig. 3a,b).

Fig. 2 |. μMap identifies a new interaction between SLK and c-Myc.

a, Venn diagram of c-Myc interactors enriched by μMap in healthy prostate cells (WPMY- 1) and metastatic prostate cancer cells (LNCaP, PC-3). b, Scatter plots of all c-Myc interactors enriched by μMap in PC-3 cells against their DepMap dependency scores (https://depmap.org/portal)⁶⁰. SLK is labeled (PC-3-specific interactor; composite score 31.76). c, Filtering with μMap and DepMap reveals SLK as a druggable, PC-3-specific c-Myc interactor. d, Immunofluorescent (IF) staining of SLK in WPMY-1, LNCaP and PC-3 cells. Scale bars: 30 μm (PC-3, WPMY-1), 20 μm (LNCaP). n = 2 biological replicates. e,f, Anti-V5 (c-Myc) immunoprecipitation (IP) (e) followed by SLK blotting (f) in PC-3 cells expressing c-Myc with Myc-box deletions (MB0, MBII, MBIV). n = 2 biological replicates. g, Western blot of c-Myc in WT and SLK-KO PC-3 and WPMY-1 cells (β-actin loading control). n = 2 biological replicates. h, Left: quantification of c-Myc protein (relative to β-actin, from n = 7 replicate western blots). Right: MYC mRNA (RNA-seq) in WT versus SLK-KO PC-3 cells. n = 4 biological replicates for mRNA-seq. i, Cycloheximide-chase quantification of c-Myc half-life in WT versus SLK-KO PC-3 cells. n = 3 biological replicates. In h and i, data are mean ± standard error of the mean (s.e.m.). CPM, counts per million; IB, immunoblot.

Comparison of the three interactomes revealed several common co-regulating proteins such as MAX, MORF4L1, MORF4L2 and CCAR2, suggesting that these interactions are critical for c-Myc function in both healthy and cancerous contexts (Extended Data Fig. 3c,d). We observed significant differences in the enrichment of proteins associated with protein degradation (proteosome, E3 ligase machinery) across the three cell lines, with highest enrichment in WPMY-1 cells and lowest in LNCaP, which correlated with the half-life of c-Myc (Extended Data Fig. 3e,f). As c-Myc degradation has been implicated to be controlled by phosphorylation status^37–40, we compiled all enriched kinases across all cell lines tested and found a panel of unique kinase interactors in each along with several c-Myc associated kinases that were enriched in both cancer lines but not in the healthy prostate line (Supplementary Fig. 3g). These include CDK1, CDK9, CDK11B and PLK1, all of which are critical regulators of cellular growth. To identify cancer-specific co-regulators that modulate cellular phenotypes, we cross-analyzed the enriched proteome with DepMap RNA interference (RNAi) screens using a gene effect score threshold of less than zero, which includes moderate dependencies that may be enhanced by disrupting the c-Myc interaction. We then applied a stringent μMap composite score threshold of more than ten to ensure only the strongest c-Myc interactions were included. In PC-3, we identified a single druggable protein that met our filters, STE20-like kinase (SLK) (Fig. 2b,c).

We further sought to benchmark our μMap dataset in PC-3 cells with BioID in the same cell line (Extended Data Fig. 4a–g). Our results indicate that both BioID and μMap enrich a comparable amount of known c-Myc interactions (Extended Data Fig. 4d). However, the nuclear enrichment and smaller labeling radius ofμMap considerably reduces the amount of total hits from 1,563 proteins in BioID to 519 in μMap (Extended Data Fig. 4d–f). Notably, BioID was unable to identify SLK as a c-Myc interactor in PC-3 cells (Extended Data Fig. 4e).

SLK interacts with c-Myc only in PC-3 cells

SLK is involved in focal adhesion and microtubule organization, but very little research has been dedicated to its role in cancer, and it has never been associated physically or genetically with c-Myc^41–43. We assessed the expression levels of SLK in all three cell lines by western blotting, showing similar levels of the protein with no increase observed in PC-3 cells that would account for differential enrichment (Extended Data Fig. 5a). Immunofluorescent staining of SLK confirmed the expression levels in all cell lines yet showed an altered distribution only in PC-3 cells, which exhibited a substantial increase in nuclear SLK (Fig. 2d). Co-IP experiments showed a robust Myc–SLK interaction in PC-3 but not in LNCaP or WPMY-1 cells (Extended Data Fig. 5b). In addition, recovery of Myc protein after immunoprecipitation of SLK in PC-3 cells was strengthened following nuclease treatment, suggesting the interaction between SLK and c-Myc is not dependent on DNA scaffolding (Extended Data Fig. 5c). MolBoolean analysis of the two proteins also showed significant colocalization within the nucleus that is unique to PC-3 cells (Extended Data Fig. 5d,e). Co-IP with Myc-box deletions showed this interaction was dependent on both MB0 and MBIV, with MB0 deletion producing the greatest loss of SLK recovery and MBIV deletion producing a reproducible but smaller reduction (Fig. 2e,f). MBII deletion did not consistently reduce SLK recovery across biological replicates and is not considered a primary binding determinant. As systematic deletions of SLK regions led to poor protein stability, we employed diazirine footprinting as an alternative strategy to map the SLK–c-Myc binding interface on recombinant proteins. This analysis identified several intrinsically disordered regions (IDRa–e) and the C-terminal coiled-coil domain of SLK as sites of c-Myc engagement in vitro (Extended Data Fig. 6a–d).

SLK controls c-Myc stability in PC-3

We next investigated the role SLK plays in c-Myc biology. CRISPR-knockout (KO) of SLK in PC-3 cells had a notable effect on c-Myc protein levels, reducing c-Myc to ~20% of wild-type (WT) PC-3 cells (Fig. 2g,h and Extended Data Fig. 5f). Similarly, small interfering RNA (siRNA) knockdown of SLK in PC-3 cells led to a ~45% reduction in c-Myc protein levels (Extended Data Fig. 7a,b). RNA-seq analysis showed this change was predominantly at the protein level rather than reflecting a loss of messenger RNA (mRNA), which was confirmed by cycloheximide chase showing a 2.3-fold reduction in c-Myc half-life in SLK-KO PC-3 cells (138 min to 60 min) (Fig. 2h,i). The modest 1.5-fold decrease in MYC mRNA on SLK-KO is probably an indirect consequence of sustained reduction in c-Myc protein. The global reduction in c-Myc target gene activity suppresses positive regulators of the MYC locus, such as CCHC-type zinc finger nucleic acid binding protein (CNBP) and NME1/2, which directly activate MYC transcription at the NHEIII1 promoter element^44,45 (Extended Data Fig. 5g). Together, these data suggest that c-Myc protein destabilization by SLK-KO is a primary effect, while dampened MYC transcription is likely a secondary, downstream effect.

We considered whether this change in c-Myc stability could be enabled via phosphorylation by SLK, consistent with the established role of other c-Myc-associated kinases in regulating its turnover. To test this, we combined recombinant SLK and c-Myc in the presence of ATP and analyzed the reaction by western blot. We observed robust serine/threonine phosphorylation of c-Myc under these conditions, alongside SLK autophosphorylation, while blotting for the known c-Myc phosphodegron sites pS62 and pT58 yielded no signal (Fig. 3a). Phosphoproteomic analysis of recombinant c-Myc identified serine 329 as the predominant SLK phosphorylation site, with the corresponding phosphopeptide consistent with the known substrate preferences of SLK⁴⁶ (Fig. 3b–d). This site has previously been shown to stabilize c-Myc via PIM1/2 kinases⁴⁷. To validate the functional significance of this modification, we performed cycloheximide-chase experiments with the alanine substitution mutant S329A and the phosphomimetic S329D. In PC-3 cells, S329A reduced c-Myc half-life while S329D extended it, demonstrating that phosphorylation at S329 is sufficient to modulate c-Myc stability in a manner consistent with our hypothesis (Fig. 3e).

Fig. 3 |. SLK phosphorylates c-Myc at serine 329 in vitro.

a, In vitro kinase assay of recombinant SLK and c-Myc (30 min), blotted for phospho-serine/threonine and for c-Myc phosphodegron sites Ser62 and Thr58. n = 3 biological replicates. b, Left: schematic of the in vitro kinase-assay workflow adapted for MS. Right: MS detection of c-Myc phosphosites, with Ser329 detected in both input and phosphopeptide-enriched samples. n = 3 biological replicates. c, Representative mass spectrum of the c-Myc phosphopeptide bearing p-Ser329 from a +ATP, phosphopeptide-enriched sample. d, Amino acid sequence of c-Myc surrounding Ser329, aligned to the reported SLK substrate motif (lysine/arginine at the +2 position)⁴⁶. e, Cycloheximide-chase quantification of transgenic c-Myc-V5 half-life in PC-3 cells expressing WT, S329A or S329D c-Myc. n = 3 biological replicates. f, In vitro kinase assay of c-Myc with GSK3β ± SLK, blotted for Thr58 phosphorylation. n = 2 biological replicates. g, PLA of c-Myc and GSK3β in WT and SLK-KO PC-3 cells, with quantification of PLA signal per nucleus. Eight images across n = 3 biological replicates. Scale bar: 20 μm. Statistical significance was assessed using a two-tailed unpaired Welch’s t-test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. In b and g, data are mean ± s.e.m. DDA, data-dependent acquisition.

We next examined whether SLK influences c-Myc stability through the canonical T58 phosphodegron, which is targeted by GSK3β to promote proteasomal degradation. Combining recombinant SLK, GSK3β and c-Myc in vitro showed that GSK3β alone produced strong T58 phosphorylation, while SLK alone produced no signal (Fig. 3f). When catalytic amounts of both GSK3β and SLK were combined with c-Myc, T58 phosphorylation was reduced compared with GSK3β alone, and this reduction was further enhanced when SLK was preincubated with c-Myc before GSK3β addition (Fig. 3f). In vitro enzyme-linked immunosorbent assays (ELISAs) showed that both SLK and GSK3β bound c-Myc with similar affinities (20–30 nM), consistent with a competitive mechanism that partially but not fully antagonizes GSK3β-mediated phosphorylation (Extended Data Fig. 7c). To validate this in a cellular context, we performed proximity ligation assays (PLAs) between c-Myc and GSK3β in WT and SLK-KO PC-3 cells, observing a significant increase in the GSK3β–c-Myc interaction on SLK loss, confirming that SLK competes with GSK3β for c-Myc binding in cells (Fig. 3g).

Impact of SLK on cellular phenotype

We next investigated the consequences of SLK loss on cellular phenotype. RNA-seq of WT and SLK-KO PC-3 cells revealed a clear reduction in c-Myc target gene expression on SLK-KO (Fig. 4a,b). To establish that these transcriptional changes are causally linked to c-Myc rather than reflecting general SLK-dependent signaling, we re-expressed exogenous c-Myc in the SLK-KO PC-3 cells via lentiviral transduction, which rescued c-Myc target gene transcription (Fig. 4a,b). To determine whether the reduction in c-Myc target gene expression reflected altered genomic occupancy of c-Myc, we performed chromatin immunoprecipitation with sequencing (ChIP–seq) for c-Myc and RNA Pol II in WT and SLK-KO PC-3 cells. This analysis revealed that c-Myc and RNA Pol II occupancy at c-Myc target genes was largely unchanged on SLK loss, indicating that the transcriptional changes are driven by reduced c-Myc protein levels and activity rather than redistribution of c-Myc across chromatin (Extended Data Fig. 7d,e).

Fig. 4 |. SLK-KO in PC-3 cells reduces migration and proliferation in a c-Myc-dependent manner.

a,b, mRNA-seq of WT, SLK-KO and SLK-KO + MYC (Rescue) PC-3 cells. a, c-Myc target gene expression across the three conditions. n = 4 biological replicates per group. b, Gene set enrichment (HALLMARK_MYC_TARGETS_V1) for the two comparisons to SLK-KO. c, Brightfield images of WT, SLK-KO and Rescue PC-3 cells. Scale bar: 50 μm. n = 2 technical replicates. d, Six-day growth curves measured with the ATP-based CellTiter-Glo 2.0 reagent. n = 6 technical replicates per timepoint. e, Doubling times calculated from the log phase (days 0–3) of the curves in d. n = 6 technical replicates per group. f, Cell-cycle distribution by propidium iodide staining for WT, SLK-KO and Rescue PC-3 cells. n = 3 biological replicates. Statistical significance was assessed using a two-way ANOVA with Tukey’s multiple comparisons test (3 families, 3 comparisons per family, α = 0.05). In d–f, data are mean ± s.e.m. g, The 24-h scratch (wound-healing) assays of PC-3 cells: WT, SLK-KO, and WT treated with vehicle (DMSO) or an SLK inhibitor (SLK/STK10-IN-1, 10 μM). Scale bars: 300 μm. n = 3 technical replicates per group. h, Flank xenografts of WT and SLK-KO PC-3 cells; tumor diameters and endpoint tumor weights. n = 9 biological replicates; line at the median is shown. Statistical significance was assessed using a twotailed unpaired Student’s t-test (P < 0.0001). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. GFP, green fluorescent protein; NES, normalized enrichment score; BF, brightfield; P_adj, adjusted P.

Visual inspection of SLK-KO cells showed a progressive loss of cellular integrity that was similarly rescued by c-Myc re-expression (Fig. 4c). ATP-based proliferation assays showed diminished cell division in SLK-KO cells, with doubling times increasing from 31 h to 54 h in a c-Myc-dependent manner (Fig. 4d,e). Cell-cycle analysis using propidium iodide revealed a loss of S phase cells and a corresponding G2 accumulation that was also c-Myc-dependent (Fig. 4f and Extended Data Fig. 8d). A scratch assay to simulate wound healing showed a similar trend, with SLK-KO and pharmacological inhibition both repressing wound closure (Fig. 4g). Proliferation, cell cycle and wound healing were only minimally affected by SLK-KO in WPMY-1 cells, suggesting the oncogenic role of SLK is cancer-specific (Extended Data Fig. 8a–e). These data are consistent with previous reports of transcriptional silencing of MYC using siRNA in PC-3 cells, suggesting these observed phenotypes arise through Myc depletion⁴⁸. Furthermore, xenograft models using WT and SLK-KO PC-3 cells showed complete inhibition of tumorigenesis in SLK-KO cells, demonstrating the importance of SLK for cancer growth in vivo (Fig. 4h).

To better understand these phenotypes, we compared the proteomes and phosphoproteomes of the WT and SLK-KO cells. We observed considerable remodeling of the cellular proteome (Extended Data Fig. 9a,b). Global proteomics demonstrated a significant loss of epithelial markers (epithelial cell adhesion molecule, epidermal growth factor receptor) and proteins associated with growth and chromatin regulation (BRD4, MKI67, AURKB, MCM complex, CDCA8, PLK1, CTNNB1) (Extended Data Fig. 9c,d). Concurrently, we observe an increase in proteins associated with autophagy (SQSTM1, MAP1LC3A, ATG3, ATG5, ATG12, ATG16L1, GABARAPL2, WIPI2, OPTN) and the interferon response (MX1, MX2, IFIT1, IFIT2, IFIT3, IFIT5, IFI44, IFI44L, IFI16, ISG20) (Extended Data Fig. 9f,h). We validated a selection of the most enriched proteins by western blot, immunofluorescence or flow cytometry, confirming the accuracy of our analysis (Supplementary Fig. 9e,g,i–l). The phosphoproteome of the WT PC-3 cells were enriched in phosphosites related to the cell cycle and chromatin organization, whereas SLK-KO led to enrichment of mRNA metabolism, splicing and GTP signaling (Extended Data Fig. 10a,b). In addition, Myc and epidermal growth factor receptor activity were the most differentially enriched pathways in WT PC-3 cells, consistent with SLK-KO repressing c-Myc based transcription (Extended Data Fig. 10c). ATM was the most active kinase in SLK-KO based on substrate analysis, suggesting a loss of chromatin integrity and DNA damage (Extended Data Fig. 10e). The Myc phosphodegron (T58, S62) associated with its proteasomal degradation was enriched in the SLK-KO cell line, supporting our observations that SLK binding inhibits Myc phosphorylation at T58 and prevents degradation of Myc by the proteasome (Extended Data Fig. 10d). In addition to a loss of Myc expression, we observed striking morphological changes to the SLK deficient PC-3 cells over time (Extended Data Fig. 10i). Quantification of nuclear size at early passages showed an increase in SLK-KO compared with WT PC-3 (Extended Data Fig. 10h). In agreement with our analysis of the phosphoproteome, we observed an increase in phospho-ATM (Ser1981) in SLK-KO, a marker for the DNA damage response (Extended Data Fig. 10f). Quantitative PCR with reverse transcription (RT–qPCR) analysis of markers of the epithelial to mesenchymal transition (SLUG, ZEB1, ZEB2) showed positive regulators of epithelial-mesenchymal transition are transcriptionally upregulated after SLK-KO (Extended Data Fig. 10g). Taken together, these data indicate a mechanism where SLK-KO leads to loss of Myc stabilization, driving an epithelial to mesenchymal transition, subsequent loss of chromatin integrity, inhibition of autophagy and activation of the interferon response.

SLK splicing is unique in PC-3 cells

While our data strongly support a mechanism in which nuclear SLK binds and stabilizes c-Myc via phosphorylation and/or antagonism of GSK3β-mediated turnover, the basis for the selectivity of this interaction across cell lines remained unclear. Two different splice isoforms are known for SLK that differ through inclusion of exon 13, leading to a change of 31 amino acids that is thought to affect protein dimerization^49–51. ESRP1 and ESRP2 promote exon 13 inclusion, generating the long form (SLK-L)^50,52,53, and RBFOX2 promotes exon 13 skipping, generating the short form (SLK-S)^50,54. Some evidence has suggested that isoform selectivity leads to altered proliferation in cancer^54,55 and cellular localization⁵⁶. Recent research shows that nuclear localization of SLK is dependent on the inclusion of exon 13, which contains a nuclear localization sequence (NLS)⁵⁶. To assess whether the change in SLK localization and c-Myc-binding may be attributed to differential expression of SLK isoforms, we performed splicing analysis on each cell line to assess altered splice isoforms in the SLK transcript. These data showed inclusion of exon 13 of SLK only in PC-3 cells, which correlated with increased expression of ESRP1/2 and reduced expression of RBFOX2 (Fig. 5a–c). We then evaluated publicly available RNA-seq data for expression of the two SLK isoforms in cancer and healthy samples^57,58. In prostate cancer samples, the SLK-L isoform represents ~75% of SLK transcripts, but in normal prostate samples SLK-S is the dominant transcript (Fig. 5d). This trend held true when including all samples and tissue types in the analysis, suggesting the SLK-L isoform is oncogenic (Fig. 5e). Furthermore, expression of SLK-L, not SLK-S, correlates with expression of c-Myc target genes in prostate cancer, supporting a mechanism where this isoform drives oncogenesis by increasing c-Myc activity (Fig. 5f and Extended Data Fig. 8f–g). SLK dependency does not appear to be unique to AR-negative prostate cancer cell lines (Extended Data Fig. 8h–i).

Fig. 5 |. PC-3 cells express the long splice isoform of SLK, the dominant isoform in prostate cancer.

a, Schematic of the two SLK splice isoforms, SLK-L and SLK-S, differing by inclusion of exon 13 (93 bp); exon 13 inclusion is regulated by ESRP1/2 and skipping by RBFOX2. b, RT–qPCR of total SLK (exon 8–11 primers) and SLK-L (exon 13 primers) in WPMY-1, LNCaP and PC-3 cells. n = 3 for glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and total SLK (one outlier excluded in the LNCaP and WPMY-1 GAPDH groups); n = 12 technical replicates for SLK-L. c, Total RNA-seq reads for SLK exon 13, RBFOX2, ESRP1 and ESRP2 in WPMY-1, LNCaP and PC-3 cells. n = 3 biological replicates. Statistical significance was assessed using a one-way ANOVA with Tukey’s multiple comparisons test (1 family, 3 comparisons, α = 0.05); asterisks denote adjusted P values. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. In b and c, data are mean ± s.e.m. d–f, Analysis of RNA-seq data from the TCGA TARGET GTEX study (UCSC Xena, xenabrowser.net)^57,58. d, Relative SLK-S versus SLK-L isoform abundance in normal prostate (GTEX) and prostate adenocarcinoma (TCGA). e, SLK-S versus SLK-L isoform abundance across all normal (GTEX, n = 7,862) and cancer (TCGA, n = 10,535) samples; line at the median is shown. f, Correlation of c-Myc target gene expression with SLK-L versus SLK-S isoform expression in prostate adenocarcinoma (TCGA, n = 496).

Generality of the c-Myc–SLK interaction

Although our bioinformatic analysis suggests that the regulatory axis described here is operative in human cancers broadly, we sought to validate that the SLK–c-Myc interaction extends beyond the prostate cancer setting and can be predicted from existing datasets. We correlated cancer cell lines with DepMap dependency on SLK against enrichment of hallmark c-Myc target gene signatures (gene set enrichment analysis) and selected a subset for experimental validation spanning multiple tumor types: glioblastoma (DBTRG-05MG), pancreatic ductal adenocarcinoma (Panc 08.13, PACADD-161) and non-small cell lung cancer (CAL-12T, NCI-H1573) (Fig. 6a). All lines except NCI-H1573 showed enrichment of c-Myc target genes and SLK dependency by DepMap. c-Myc immunoprecipitation followed by SLK blotting confirmed a physical interaction in all lines except NCI-H1573, and immunofluorescence analysis showed that all lines positive for co-IP exhibited nuclear SLK localization, consistent with our observations in the prostate cell line panel (Fig. 6b,c).

Fig. 6 |. The interaction between SLK and c-Myc occurs in other cancer cell lines.

a, DepMap analysis of all cancer cell lines for c-Myc target gene expression versus SLK-KO sensitivity; the five SLK-sensitive lines selected for testing are indicated. b, Anti-c-Myc co-IP followed by SLK blotting in glioblastoma (GBM), pancreatic ductal adenocarcinoma (PDAC) and non-small cell lung cancer (NSCLC) cell lines. n = 2 biological replicates. c, Immunofluorescent staining of SLK in the five cancer cell lines. Scale bar: 30 μm. n = 2 biological replicates.

Discussion

In summary, we developed a photocatalytic proximity labeling strategy to uncover druggable, cancer-specific interactions at c-Myc. The short-range diazirine activation mechanism enables precise interactomic profiling, which can be integrated with DepMap to identify drivers of oncogenic signaling. Using this approach, we identified a splice variant of SLK that localizes to the nucleus, interacts with c-Myc and stabilizes it at the protein level in PC-3 cells, a model of AR⁻ prostate cancer. We demonstrate that this stabilization is mechanistically driven by SLK-mediated phosphorylation of c-Myc at serine 329, which antagonizes GSK3β-dependent phosphorylation of the T58 phosphodegron and thereby stabilizes the c-Myc protein. This represents a previously uncharacterized mechanism of c-Myc stabilization and identifies S329 phosphorylation as a potential node for therapeutic intervention. Genetic knockout of SLK suppresses c-Myc target gene expression, diminishes proliferation and migration in vitro, and abolishes tumor formation in a xenograft model in a c-Myc-dependent manner, indicating that SLK inhibition or degradation may represent a viable therapeutic strategy. These effects are selective, as SLK-KO minimally affects healthy prostate cells.

The cancer selectivity of the SLK–c-Myc interaction appears to be associated with differential splicing and localization rather than changes in total SLK expression, with the long isoform SLK-L overrepresented in cancer cells due to altered expression of the splicing regulators ESRP1/2 and RBFOX2. Inclusion of exon 13 in SLK-L encodes a nuclear localization sequence that redirects SLK from the cytoplasm to the nucleus, possibly facilitating its interaction with c-Myc. As total SLK expression at the RNA and protein levels remains unchanged between normal and malignant tissue, this regulatory axis would be entirely undetectable without high-resolution subcellular technologies such as μMap, underscoring the critical importance of context-dependent, spatially resolved interactomics for uncovering hidden oncogenic mechanisms.

The μMap approach is uniquely suited to this discovery. Performing the equivalent experiment with BioID proximity labeling of c-Myc in PC-3 cells recovered a substantially larger and less spatially restricted interactome, with greater representation of proteins associated with broadly active chromatin regions. Critically, SLK was not significantly enriched in the BioID PC-3 dataset, demonstrating that this interaction would have been missed by conventional proximity labeling and underscoring the importance of labeling radius as a determinant of interactome selectivity. The ~4-nm labeling radius of μMap, compared with the effective ~100-nm radius of BioID, the instantaneous nature of photoactivation versus hours-long biotinylation and the in nucleo execution of μMap collectively account for the higher spatial specificity observed. These considerations should guide the selection of proximity labeling strategies in future TF interactome studies, depending on whether the primary goal is comprehensive interactome coverage or precise identification of direct proximal contacts. In addition, future experiments using endogenous CRISPR-based tagging of c-Myc would provide an important orthogonal validation to perform proximity labeling at physiological expression levels.

While our data indicate that nuclear localization of SLK is possibly required for its interaction with c-Myc, the precise molecular determinants that govern this interaction remain incompletely understood. The exon 13-encoded region differentiates SLK-L from SLK-S not only through the introduction of a nuclear localization sequence, but also through predicted changes to the dimerization interface of the kinase domain, raising the possibility that isoform-specific structural differences beyond nuclear access contribute to selective c-Myc engagement. Moreover, it is possible that cell-context-specific SLK interactions may facilitate nuclear localization and c-Myc binding. It is also unclear whether the two isoforms exert opposing or simply distinct effects on oncogenic transcription, or whether SLK-S plays an independent regulatory role in normal tissue that is subverted on isoform switching in cancer. Future work will focus on a comprehensive comparison of the two SLK isoforms, including assessment of how each isoform differentially regulates the oncogenic transcriptional program. While it is possible that SLK binding alone is sufficient to stabilize c-Myc, future investigations will aim to clarify the requirement of SLK catalytic activity for c-Myc stabilization by rescuing c-Myc half-life in SLK-KO cells with WT SLK or a kinase-dead SLK mutant. In addition, a full structural characterization of SLK–c-Myc binding remains an important future goal, but the disordered nature of both the c-Myc transactivation domain and the SLK intrinsically disordered regions presents a major challenge for conventional approaches. We anticipate that this deeper mechanistic understanding will not only clarify the biology of this new regulatory axis but also inform the design of isoform-selective therapeutic strategies targeting the SLK-L–c-Myc interaction.

The context-dependent kinase interactomes revealed by μMap also provide a resource for investigating c-Myc stabilization mechanisms in other cancer settings. In LNCaP cells, CAMKIIγ is specifically enriched in the μMap dataset relative to WPMY-1, and CAMKIIγ-mediated phosphorylation of c-Myc at S62 is a known stabilizing modification⁵⁹. We propose CAMKIIγ as a candidate regulator of c-Myc stability in LNCaP cells, consistent with the elevated c-Myc half-life observed in this line despite the absence of the SLK–c-Myc interaction, and note that the μMap datasets generated here represent a starting point for systematic investigation of cell-type-specific c-Myc regulatory kinases.

Analysis of samples from patients revealed that the SLK-L isoform is not specific to prostate cancer and is broadly co-opted across cancer types and correlates with c-Myc target gene expression, supporting a model in which SLK stabilizes c-Myc to promote oncogenic transformation beyond the prostate setting (Supplementary Fig. 8j,k). Consistent with this, we validated the SLK–c-Myc interaction across a panel of cancer cell lines spanning glioblastoma, pancreatic and lung cancers, demonstrating that nuclear SLK localization and c-Myc co-IP can be prospectively predicted from DepMap dependency and c-Myc target gene enrichment data. This predictive framework suggests that the approach described here is not only capable of discovering new oncogenic interactions but can also guide their rational validation across diverse cancer contexts.

Our findings underscore the power of cell-context-specific interactomics to illuminate hidden regulatory networks that drive disease. By integrating nanoscale proximity labeling with disease-informed experimental design, we reveal oncogenic co-regulators inaccessible to conventional methods. The identification of nuclear localized SLK as a cancer-selective c-Myc stabilizer, validated across multiple tumor types, positions SLK localization as a candidate predictive biomarker for tumors driven by this mechanism. Beyond c-Myc, this work establishes μMap as a generalizable framework for mapping transcriptional dependencies and decoding the protein networks that underlie malignancy. We anticipate that this approach will accelerate the discovery of therapeutic vulnerabilities across diverse cancer types and transform how context-dependent signaling is studied in complex cellular environments.

Online content

Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41589-026-02284-0.

Methods

General considerations

CfaC-Ir was synthesized as previously described in ref. 26. The diazirine-biotin probes used in μMap are commercially available (MCE, cat. no. HY-154801). All primers listed in Supplementary Table 1 are custom DNA oligos purchased from Integrated DNA Technologies. Tris-buffered saline with Tween (TBST) was purchased from Boston BioProducts (cat. no. IBB-180X). iBright Prestained Protein ladder was purchased from Thermo Scientific (cat. no. LC5615). Water was purified using a Millipore Milli-Q Integral Water Purification System.

Cell culture

HEK293T cells were a gift from the MacMillan group (Princeton University), DBTRG-05MG cells were a gift from the Janiszewska group (University of Colorado Anschutz) and other cell lines were purchased from the American Type Culture Collection (ATCC) or the Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures GmbH (DSMZ). All cells were cultured in a humidified incubator at 37 °C with 5% CO₂. HEK293T, WPMY-1 (ATCC, cat. no. CRL-2854) and CAL-12T cells (DSMZ, cat. no. ACC 443) were maintained in DMEM (Gibco, cat. no. 11995-065). LNCaP (ATCC, cat. no. CRL-1740), Panc 08.13 (ATCC, cat. no. CRL-2551), PACADD-161 (DSMZ, cat. no. ACC 746) and NCI-H1573 cells (ATCC, cat. no. CRL-5877) were maintained in RPMI 1640 (Corning, cat. no. 10-041-CV). PC-3 (ATCC, cat. no. CRL-1435) were maintained in DMEM/F12, GlutaMAX (Gibco, cat. no. 10565-018). All basal media were supplemented with 10% v/v FBS (R&D Systems, cat. no. S11150) and 100 U ml⁻¹ of penicillin 100 μg ml⁻¹ of streptomycin (Gibco, cat. no. 15140-122) to make the complete growth medium in which cells were cultured.

Western blot

Cell lysates were quantified using Pierce bicinchoninic acid (BCA) protein assay kit (Thermo Scientific, cat. no. A55864), normalized to the same concentration and 4× Laemmli Sample Buffer (BioRad, cat. no. 1610747) containing 2-mercaptoethanol was added to reach a 1× final concentration. Samples were run on a precast Invitrogen NuPAGE 4–12% bis-tris acrylamide gel (Invitrogen cat. no. NW04125BOX) and transferred to nitrocellulose membrane (Thermo Scientific, cat. no. 88018). Revert 520 Total Protein Stain (LICOR, cat. no. 926-10011) was used to stain membranes for total protein according to the manufacturer’s instructions. Membranes were then blocked with 5% w/v milk or 3% bovine serum albumin (BSA) in TBST (25 mM tris, 150 mM NaCl, 0.1% v/v Tween-20, pH 7.7) for 1 h at room temperature. Membranes were incubated with primary antibodies at the stated dilutions (Supplementary Table 2) overnight with rotation at 4 °C. After 3 × 5-min washes with TBST, secondary antibodies were applied for 1 h at room temperature (Supplementary Table 2), before imaging on a Li-Cor Odyssey imager (LICOR; IRDye secondaries) or an ImageQuant LAS 500 (GE Healthcare; HRP secondaries). Quantification of western blots was performed in Fiji.

Immunofluorescence

Cells were gathered with trypsin, washed with PBS, plated on sterile polylysine-treated coverslips, then incubated overnight at 37 °C. Cells were then fixed with 4% paraformaldehyde diluted in PBS for 10 min at room temperature. After washing 3× with PBS, cells were permeabilized with 75% ethanol diluted in PBS overnight at 4 °C. The coverslips were then blocked with 1% BSA in PBS for 1 h at room temperature. Coverslips were then incubated with primary antibody at the stated dilutions (Supplementary Table 2) in 1% BSA in PBS overnight at 4 °C. Samples were then washed 3× with PBS and incubated with secondary antibody diluted 1:1,000 for 1 h at room temperature. Slides were then washed three times with PBS and mounted to slides using ProLong Gold Antifade Mountant with DNA Stain 4,6-diamidino-2-phenylindole (DAPI) (Invitrogen, cat. no. P36935). Images were acquired on the FLUOVIEW FV3000 confocal microscope (Olympus).

Proximity ligation assay (PLA)

Round coverslips were placed in a 12-well plate and sterilized with 70% ethanol. Wells were then washed with H₂O and treated with polylysine (Sigma Aldrich, cat. no. P4707) for 10 min, followed by 2 washes with sterile H₂O and allowed to air dry for 2 h. Cells were then plated at 2 × 10⁵ cells per well and allowed to adhere overnight. The next morning, the media was aspirated, the wells were washed 2× with PBS and fixed for 10 min with 4% paraformaldehyde in PBS at room temp. Cells were then washed 3× with PBS and permeabilized with 70% ethanol in PBS overnight at 4 °C. Slides were washed three times with Dulbecco’s PBS (DPBS). The PLA was carried out as recommended by the manufacturer using the Duolink in Situ Detection Reagents Red kit mouse/rabbit (Sigma Aldrich, cat. nos. DUO92008, DUO92002 and DUO92004). In brief, slides were blocked for 1 h at 37 °C with blocking buffer. Slides were then incubated with c-Myc (Abcam, cat. no. ab32072) and GSK3β (Abcam, cat. no. ab93926) primary antibodies (Supplementary Table 2) diluted 1:100 in antibody binding buffer at 4 °C overnight. Single stain negative controls were included for each primary antibody. Slides were washed with wash buffer A and then incubated with PLA probe plus donkey anti-rabbit IgG and PLA probe minus donkey anti-mouse IgG antibodies for 1 h at 37 °C. Samples were washed with wash buffer A and incubated for 30 min at 37 °C with Duolink ligase reaction solution. Slides were washed with wash buffer A and incubated for 100 min with Duolink amplification solution. Slides were washed with wash buffer B, mounted with Duolink in situ mounting media with DAPI and sealed with clear nail polish. Images were acquired on the FLUOVIEW FV3000 confocal microscope (Olympus). Cells were located using the DAPI channel to eliminate bias for the PLA signal. Z stack images were taken using 0.5-μm z-spacing between four total images. Maximum projections were quantified using CellProfiler nuclear speckle counting software. GraphPad Prism was used to evaluate statistical significance and visualize the data.

mRNA-seq

PC-3 (WT), PC-3 (SLK-KO) and PC-3 (SLK-KO + c-Myc: ‘Rescue’) cells were prepared according to the standard Plasmisaurus RNA-seq submission guidelines. Briefly, cells were seeded in six-well plates to reach 80% confluency the following day, then incubated overnight at 37 °C with 5% CO₂. The next day, cells were gathered with trypsin, washed with PBS, counted and 5 × 10⁵ viable cells per biological replicate were pelleted and resuspended in 100 μl of 1× Zymo DNA/RNA shield (Zymo, cat. no. R1100-50). Samples were shipped to Plasmidsaurus at room temperature.

The library preparation and sequencing was performed at Plasmidsaurus. Briefly, Plasmidsaurus RNA-Seq utilizes Illumina sequencing and a 3′ end counting approach, and total RNA is isolated using a bead-based extraction approach. For library preparation, mRNA is converted into complementary DNA (cDNA) via reverse transcription and second-strand synthesis using a poly(dT)VN primer, followed by tagmentation, library indexing and amplification. Sequencing is single end (~90-base pair (bp) reads) and 3′ end counting is used to capture differential gene expression. Differential expression is generated using edgeR v.4.0.16 with filtering for low-expressed genes with edgeR::filterByExpr with default values.

Total RNA-seq

WPMY-1, LNCaP and PC-3 cells were seeded in six-well plates to reach 60–75% confluency the following day, then incubated overnight at 37 °C with 5% CO₂. For the transfected PC-3 cell conditions, the media was replaced with fresh media and PC-3 cells were transfected with either c-Myc-CfaN (0.75 μg) or c-Myc-V5-6xHis (0.75 μg) using the Lipofectamine 3000 Transfection Reagent (Thermo Scientific, cat. no. L3000150), following the manufacturer’s instructions. The next day, cells were gathered with trypsin, washed with PBS and RNAs were extracted from 1 × 10⁶ cells using the Rneasy plus mini kit (Qiagen, cat. no. 74134) according to the manufacturer’s instructions. Total RNA was submitted to The Herbert Wertheim UF-Scripps Institute for Biomedical Innovation & Technology—Genomics Core (RRID SCR_017827), where it was quantified using a Qubit 2.0 Fluorometer (Invitrogen) and evaluated on an Agilent 4200 TapeStation (Agilent Technologies) for quality assessment. All RNA samples had RNA Integrity Number of 10.0 and were used for total RNA-seq library preparation.

RNase-free working environment was maintained, and RNase-free tips, tubes and plates were utilized. Next, 1 μg of total RNA per sample was depleted of ribosomal RNA using probes provided in the NEBNext ribosomal RNA (rRNA) Depletion Kit v2 (cat. no. E7405, NEB) according to the manufacturer’s recommendations. The library preparation from the rRNA-depleted RNA was conducted according to the NEBNext Ultra II Directional RNA kit (cat. no. E7760, NEB). Briefly, the RNA samples were chemically fragmented in a buffer containing divalent cations by heating to 94 °C for 10 min. The fragmented RNA was random hexamer primed and reverse transcribed to generate the first strand of cDNA. The second strand was synthesized after removing the RNA template and incorporating deoxyuridine triphosphate in place of deoxythymidine triphosphate to maintain strand specificity. Fragmented double-stranded cDNA was then end repaired and adenylated at their 3′ ends. A corresponding ‘T’ nucleotide on the adaptors was used for ligating the adaptor sequences to the cDNA. The adaptor ligated DNA was purified using magnetic beads and PCR amplified (using 11 cycles) to incorporate a unique barcode and to generate final libraries. The libraries were purified using magnetic beads to remove any remaining primers and adaptors.

The final libraries were validated on an Agilent 4200 TapeStation (Agilent Technologies), normalized to 4 nM, pooled equally and loaded onto an Illumina NextSeq 2000 P3 300-cycle flow cell (cat. no. 20040561, Illumina) at 750-pM final concentration and sequenced using 2 × 155-bp paired-end chemistry. On average, we generated 86 million reads pass filter per sample. Raw and processed data files were uploaded to the National Center for Biotechnology Information Gene Expression Omnibus (GSE307813).

Analysis of RNA-seq data

Splicing analysis.

RNA-seq data were processed using the nf-core/rnasplice pipeline (version 1.0.4, https://doi.org/10.5281/zenodo.8424632)⁶¹ on the HiPerGator high-performance computing cluster with Nextflow (version 25.04.4) and Singularity (version 3.10.4). Raw FASTQ files were quality-checked with FastQC (version 0.12.1) and trimmed using TrimGalore (version 0.6.7, cutadapt version 3.4). Reads were aligned to the ENSEMBL human genome GRCh38 release 111 (Homo_sapiens.GRCh38.dna_sm.primary_assembly.fa) using STAR (version 2.7.9a) with the corresponding STAR index. Gene annotations were derived from the GTF file (Homo_sapiens.GRCh38.111.gtf). FeatureCounts (subread version 2.0.1) was used to quantify read counts, followed by differential expression analysis with edgeR (R version 4.0.3). Event-based differential splicing analysis was performed using rMATS (version 4.1.2).

Analysis of Myc target gene expression after transfection of PC-3 cells.

Version 3.12.0 of the nf-core/rnaseq pipeline (https://doi.org/10.5281/zenodo.7998767)⁶² was run on HiPerGator with Nextflow (v.25.04.4) and Singularity (v.3.10.4). All default settings were used except the following changes: the option for alignment by STAR was used, along with RSEM for quantification and reads were aligned to the human GRCh38 genome. Downstream analysis was performed with DESeq2 (v.1.34.0) to identify differentially expressed genes.

Analysis of publicly available RNA-seq data

The analyses of publicly available RNA-seq data in this paper are based on data generated by The Cancer Genome Atlas (TCGA) Research Network, the Therapeutically Applicable Research to Generate Effective Treatments (TARGET) initiative and The Genotype-Tissue Expression (GTEx) Project (Acknowledgements section). The data from the TCGA TARGET GTEx study were obtained from the UCSC Toil RNAseq Recompute Compendium^57,58, and further analyzed and plotted in GraphPad Prism v.10. The human gene set: DANG_MYC_TARGETS_UP was used to identify Myc target genes in the analyses⁶³.

μMap photoproximity labeling

All μMap experiments were performed with three biological replicates per group (with 2–4× 15-cm plates per replicate).

Transfection.

Cells were seeded in 15-cm plates to reach 60–75% confluency the following day, then incubated overnight at 37 °C with 5% CO₂. The media was replaced with fresh media, and all plates were transfected with either c-Myc-V5-6xHis (control) or c-Myc-CfaN (μMap) using the Lipofectamine 3000 Transfection Reagent (Thermo Scientific, cat. no. L3000150) and following the manufacturer’s instructions. Then 18–24 h post-transfection, the cells were collected using trypsin, washed once with PBS and pelleted in 15-ml conical vials before snap-freezing on dry ice and storage at −80 °C.

μMap photoproximity labeling.

The next day, cell pellets were thawed and resuspended in 0.2 ml of Nuclear Isolation Buffer per 1 × 10⁶ cells (Nuclear Isolation Buffer: 45 mM KCl, 0.1 mM EDTA, 6.25 mM MgCl₂, 12.5 mM Tris-HCl (pH 7.5), 375 mM sucrose, 0.125% NP-40 and 1× halt protease inhibitor cocktail). Cell suspensions were incubated on ice for exactly 7 min, then centrifuged at 600g for 10 min at 4 °C to isolate crude nuclei. The supernatant was discarded, and nuclei were washed twice with PBS, centrifuging at 400g for 3 min at 4 °C. The nuclei were then treated with 1.0 μM CfaC-Ir diluted in PBS and incubated in a 37 °C water bath for 40 min with gentle shaking every 10 min while being protected from light with foil. Initial optimization experiments (Supplementary Fig. 1b) used a CfaC-Ir concentration of 0.5 μM CfaC-Ir, but all subsequent μMap experiments used CfaC-Ir at 1.0 μM after further assay optimization showed maximal labeling efficiency with 1.0 μM. The nuclei were pelleted by centrifugation at 400g for 3 min at 4 °C, then washed 3× with PBS. Nuclei were resuspended in 1 ml of 250 μM diazirine-PEG3-biotin conjugate and irradiated for 1 min at 450 nm with 100% light intensity in the Photoreactor m2 (Acceled Bio). Nuclear pellets were then washed 2× with PBS, centrifuging at 400g for 3 min. Nuclear pellets were lysed in LB3 buffer (1 mM EDTA, 0.5 mM EGTA, 10 mM Tris-HCl (pH 7.5), 100 mM NaCl, 0.1% Na-Deoxycholate, 0.5% N-lauroyl sarcosine, 1× HALT protease inhibitor cocktail) and sonicated using the Diagenode Bioruptor Sonicator (HIGH, 14 cycles at 4 °C, 30 s ON, 30 s OFF). Lysates were then centrifuged at 17,000g for 20 min at 4 °C. The supernatant was extracted, and protein concentrations were determined using the BCA assay and normalized to 1 mg ml⁻¹ in 1 ml of total volume using LB3 buffer. Lysates were incubated overnight at 4 °C with 100 μl of prewashed Streptavidin Mag Sepharose beads (Cytiva, cat. no. 28985799) per 1 mg lysate with end-over-end rotation.

Washing.

Beads were washed 3 × 5 min with 1% SDS in PBS (rotating), then 3 × 5 min with 1 M NaCl in PBS (rotating), followed by 3 × 5 min with 10% EtOH in PBS (rotating). The beads were then transferred to new Lo-bind tubes, and washed 3× with PBS and 3× with 100 mM ammonium bicarbonate. The beads were then resuspended in 0.5 ml of 3 M urea in PBS, and 25 μl of dithiothreitol (DTT) (200 mM in 25 mM ammonium bicarbonate) was added to each sample to reduce proteins. Samples were incubated at 55 °C for 30 min with shaking at 700 rpm. Next, 30 μl of iodoacetamide (IAA) (500 mM in 25 mM ammonium bicarbonate) was added to each sample and incubated at 30 min at room temperature in the dark to alkylate proteins. Beads were then washed 3× with PBS and 6× with 50 mM ammonium bicarbonate, then transferred to new Lo-bind tubes. To elute peptides, the beads were resuspended in 40 μl of 50 mM ammonium bicarbonate with 1.2 μl of MS-grade trypsin (resuspended in 1 mg ml⁻¹ in 50 mM acetic acid). Samples were incubated overnight at 37 °C with end-over-end rotation, followed by another addition of 0.8 μl of trypsin for 1 h at 37 °C the next morning. Supernatants were collected and split into technical replicates of 20 μl each before storage at −80 °C.

Label-free proteomics.

Peptide digests were acidified with trifluoroacetic acid (TFA) to 0.1% (v:v) and desalted using 2 μg capacity Zip-Tips (Millipore) according to manufacturer instructions. Following drying under vacuum, peptides were resolubilized in 0.1% formic acid to a final concentration of 100 ng μl⁻¹. Samples were analyzed on a nanoElute (plug-in v.2.1.60.0; Bruker) coupled to a Bruker TimsTOF Pro 2 mass spectrometer, equipped with a CaptiveSpray source and a 20-μm zero dead volume Sprayer. Peptides (corresponding to 100 ng) were loaded onto a Thermo Fisher PepMap Neo C18 trap column (300 mm × 5 mm, 5-mm particle size) and then separated on a PepSep Series reverse-phase C18 column (15 cm × 150 μm, 1.5-μm particle size) from Bruker. The column temperature was maintained at 50 °C using an integrated Bruker Column Toaster. The column was equilibrated using 4 column volumes before loading samples in 100% buffer A (99.9% Fisher Optima liquid chromatography with mass spectrometry (LC-MS) water, 0.1% formic acid), with both steps performed at 800 bar. The trap column was equilibrated at 201.6 bar. Samples were separated at 500 nl min⁻¹ using a linear gradient from 2% to 35% buffer B (99.9% Fisher Optima LC-MS acetonitrile, 0.1% formic acid) over 20.0 min before ramping to 95% buffer B (in 0.5 min) and sustained at 95% buffer B for 4.5 min (total separation method time 25.0 min). The Bruker TimsTOF Pro 2 was operated in DIA-PASEF mode using Tims Control v.5.0.2. Settings for the MS method were as follows: mass range 100 to 1,700 mass/charge (m/z), 1/K0 Start 0.6 V cm² end 1.4 V s⁻¹ cm², trapped ion mobility spectrometry (TIMS) ramp and accumulation time 75 ms, capillary voltage 1,700 V, dry gas 3 l min⁻¹, dry temperature 200 °C, DIA-PASEF settings: 18 tandem MS (MS/MS) scans (50 m/z windows, 0.21 1/K0 windows, total cycle time 0.74), mass range 300 to 1,200 and collision-induced dissociation (CID) collision energy 20 eV (at 0.60, 1/K0) to 65 eV (at 1.60, 1/K0). The analysis was performed at The Herbert Wertheim UF-Scripps Institute for Biomedical Innovation & Technology, Mass Spectrometry and Proteomics Core Facility (RRID SCR_023576).

Label-free proteomics data analysis.

The raw mass data were further analyzed by DIA-NN (v.1.8.1). Parameters set as follows: trypsin/P digestion, missed cleavages: 3, maximum number of variable modifications: 3, N-term M excision, Ox(M), Ac(N-term) and C carbamidomethylation. Peptide length range was 7–30, precursor charge range 1–4, m/z range 300–1,800 and fragment ion range 200–1,800. Mass accuracy and MS accuracy were both set to 10. The following settings on the algorithm were checked: ‘Use isotopologues’, ‘MBR’, ‘No shared spectra’ and ‘Heuristic protein inference’. Precursor FDR was set to 1% with threads as 16. A spectral library was generated via DIA-NN from all known human proteins (In-Silico spectral library). The resulting matrix. pg file was opened in Perseus (v.2.0.7.0). Intensities were inputted as ‘main’, and the rest of the descriptors were categorical. Data were then transformed (log₂). Data were annotated by treatment. Missing values were imputed with Perseus default settings (0.3 width, 1.8 downshift). Normalization was performed via median subtraction. Following this process, we employed cutoffs of log₂(fold change) ≥0.25 and FDR-corrected P < 0.05, and hits were found in all positive replicates to be considered. The resulting volcano plots were plotted in GraphPad Prism v.10 for final figures.

BioID

All BioID experiments were performed with three biological replicates per group (with 2× 15-cm plates per biological replicate). BioID experiments were based on the protocol from ref. 22 with the following modifications. PC-3 cells were seeded in 15-cm plates to reach 60–75% confluency the following day, then incubated overnight at 37 °C with 5% CO₂. The next morning, the media was replaced with complete medium containing 50 μM biotin (ChemScene, cat. no. CS-2719) and 1 μM MG-132 (MCE, cat. no. HY-13259). The cells were then immediately transfected with either the c-Myc-V5-BirA* or the control HA-BirA* plasmids. Then 2.5 μM doxycycline (MCE, cat. no. HY-N0565) was also added to the HA-BirA* group, as its expression is controlled by a tet-operator. Cells were incubated for 24 h and gathered with trypsin. Cells were then washed 3× with 50 ml of PBS before flash freezing. Cell pellets were lysed in 10 ml of modified RIPA buffer (1% NP-40, 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 0.1% SDS, 0.5% sodium deoxycholate (DOC), 1× halt protease inhibitor cocktail), with 500 U Pierce Universal Nuclease for Cell Lysis (Thermo Scientific, cat. no. 88702). Lysate was rotated for 1 h at 4 °C, and centrifuged at 17,000g for 30 min at 4 °C. The supernatant was extracted, and protein concentrations were determined using the BCA assay and normalized to 1 mg ml⁻¹ in 1-ml total volume using modified RIPA buffer. Lysates were incubated overnight at 4 °C with 100 μl of prewashed Streptavidin Mag Sepharose beads (Cytiva, cat. no. 28985799) per 1 mg of lysate with end-over-end rotation. The next day, the beads were washed and peptides eluted as described above for μMap photoproximity labeling. Similarly, label-free proteomics and data analysis were performed exactly as described above for μMap.

Global proteomics

WT PC-3 cells and SLK-KO PC-3 cells were seeded in 10-cm plates to reach 80–100% confluency the following day, then incubated overnight at 37 °C with 5% CO₂. Each cell line was plated in three 10-cm plates, for three biological replicates per group. The next morning, the cells were gathered using trypsin, washed once with PBS and pelleted in 15-ml conical vials before lysing with SDS lysis buffer (50 mM HEPES, pH 8.5, 1% SDS, 1× halt protease inhibitor cocktail). Then 20 μg of protein was taken for cleanup and digestion. Proteins were cleaned up and digested using Sera-Mag Carboxylate SpeedBeads (Cytiva Life Sciences: E7 cat. no. 45152105050250 and E3 cat. no. 65152105050250) following the manufacturer’s protocol. Digested peptides were split into two technical replicates per sample. Label-free proteomics and subsequent data analysis were performed as described in the ‘Label-free proteomics data analysis’ section.

Phosphoproteomics

Transfection of c-Myc-V5-6xHis.

WT PC-3 cells and SLK-KO PC-3 cells were seeded in 10-cm plates to reach 60–75% confluency the following day, then incubated overnight at 37 °C with 5% CO₂. Each cell line was plated in five 10-cm plates for five biological replicates per group. The next morning, the media was replaced with fresh media, and all plates were transfected with c-Myc-V5-6xHis (1 μg) using the Lipofectamine 3000 Transfection Reagent (Thermo Scientific, cat. no. L3000150), following the manufacturer’s instructions.

Cell lysis, protein precipitation and digestion.

Phosphopeptides were enriched as previously described in ref. 64. Briefly, 18–24 h post-transfection, the cells were gathered using trypsin, washed once with PBS and pelleted in 15 ml conical vials before lysing with lysis buffer (200 mM HEPES, pH 8.5 and 8 M urea with 1× Halt Protease and Phosphatase Inhibitor Single-Use Cocktail (Thermo Scientific, cat. no. 1861280)). The lysates were then passed through a 21-gauge needle 10 times. Protein concentrations were determined by the Pierce BCA protein assay kit (Thermo Scientific, cat. no. A55864), and all samples were normalized to 1 mg ml⁻¹. Disulfide bond reduction was performed by adding 5 mM tris-(2-carboxyethyl)-phosphine to each sample and incubating for 15 min at room temperature. Alkylation was performed by then adding 10 mM IAA to the samples and incubating for 30 min in the dark. Excess IAA was quenched by adding 10 mM DTT to the samples and incubating for 15 min at room temperature. Then 400 μl of 100% methanol was added to 100 μg of each sample (0.1 ml) in a 1.5-ml microcentrifuge tube, and the samples were vortexed for 5 s. Then 100 μl of 100% chloroform was added to each sample, and the samples were vortexed for 5 s. Next 300 μl of water was added to each sample, and the samples were vortexed for 5 s. Samples were centrifuged for 1 min at 14,000g. The aqueous and organic phases were removed, isolating the protein disk, which was then washed with 400 μl of 100% methanol and centrifuged at 21,000g for 2 min at room temperature. The supernatant was removed and samples were resuspended in 70 μl of 200 mM HEPES, pH 8.5. The samples were digested overnight at 37 °C with trypsin (1 μg per sample, Thermo Scientific cat. no. 90057).

TMT labeling.

Here 30 μl of acetonitrile was added to the digested peptides, and the samples were vortexed. TMT10plex Mass Tags (Thermo Scientific, cat. no. 90111) were prepared according to the manufacturer’s instructions; samples were labeled with 200 μg of tandem mass tag (TMT) reagent, which was incubated for 1 h at room temperature with the peptides. The mass tags added to each sample were recorded. The reaction was quenched by adding 0.66 μl of 50% hydroxylamine in water (final concentration of 0.3% hydroxylamine v/v). The TMT-labeled samples were pooled together at a 1:1 ratio, and the volume was evaporated down to ~30% of the total volume by vacuum centrifuge. The sample was then desalted with Pierce Peptide Desalting Spin Columns (Thermo Scientific, cat. no. 89852) according to the manufacturer’s instructions.

Phosphopeptide enrichment.

Phosphopeptides were enriched using the Pierce High-Select Fe-NTA Phosphopeptide Enrichment Kit (Thermo Scientific, cat. no. A32992) according to the manufacturer’s instructions.

TMT quantitative proteomics and MS.

Peptide digest samples corresponding to the enriched phosphoproteome, and the flow-through proteome following the enrichment were dried under vacuum and subsequently acidified using 100 ml of 1% TFA (pH < 3). The samples were then desalted using NuTip Carbon tips from Glygen (Glygen Corp.) in the case of phosphopeptides, or 2 μg capacity ZipTips (Millipore), according to the manufacturer’s instructions, respectively. Then they were dried again under vacuum. Peptides resolubilized in 5 ml of 0.1% TFA were online eluted into a Fusion Tribrid mass spectrometer (Thermo Scientific) from an EASY PepMap RSLC C18 column (2 μm, 100 Å, 75 μm × 50 cm, Thermo Scientific), using a gradient of 5–25% solvent B (80:20 acetonitrile:water, 0.1% formic acid) in 180 min, followed by 25–44% solvent B in 60 min, 44–80% solvent B in 0.1 min, a 5-min hold of 80% solvent B, a return to 5% solvent B in 0.1 min and finally a 10-min hold of 5% solvent B. The gradient was then extended for the purpose of cleaning the column by increasing solvent B to 100% in 3 min, a 100% solvent B hold for 10 min, a return to 5% solvent B in 3 min, a 5% solvent B fold for 3 min, an increase of solvent B to 100% in 3 min, a 100% solvent B hold for 10 min, a return to 5% solvent B in 3 min and a 5% solvent B hold for 3 min and, finally, another increase to 100% solvent B in 3 min and a hold of 100% solvent B for 10 min. All flow rates were 250 nl min⁻¹ delivered using a Thermo Vanquish Neo ultrahigh-performance LC nano system (Thermo Scientific). Solvent A consisted of water and 0.1% formic acid. Ions were created at 2.1 kV using an EASY Spray source (Thermo Scientific) held at 50 °C. A synchronous precursor selection-MS3 MS method was selected based on the work in ref. 65. Scans were conducted between 380 m/z and 2,000 m/z at a resolution of 120,000 for MS1 in the Orbitrap mass analyzer at an automatic gain control (AGC) target of 4 × 10⁵ and a maximum injection of 50 ms. CID was then performed in the linear ion trap of peptide monoisotopic ions with charge 2–8 above an intensity threshold of 5 × 10³, using a quadrupole isolation of 0.7 m/z and a CID energy of 35%. The ion trap AGC target was set to 1.0 × 10⁴ with a maximum injection time of 50 ms. Dynamic exclusion duration was set at 60 s and ions were excluded after one time within the ±10-ppm mass tolerance window. The top 10 MS2 ions in the ion trap between 400 m/z and 1,200 m/z were then chosen for higher-energy C-trap dissociation at 65% energy. Detection occurred in the Orbitrap at a resolution of 60,000, an AGC target of 1 × 10⁵ and an injection time of 120 ms (MS3). All scan events occurred within a 3-s specified cycle time. The analysis was performed at The Herbert Wertheim UF-Scripps Institute for Biomedical Innovation & Technology, Mass Spectrometry and Proteomics Core Facility (RRID SCR_023576).

Data analysis with FragPipe.

Raw data were converted to mzmL files using MSConvert, following the online tutorial (https://fragpipe.nesvilab.org/docs/tutorial_convert.html). Converted data were then processed using FragPipe using the TMT10-MS3 workflow with indicated 229.16293 modification for TMT label adjusted for the N terminus. Samples were filtered by ‘Reverse’ and ‘Potential contaminant’ and were categorized by treatment. A FASTA file of UniProt Homo sapiens proteome was included as the reference. The protein abundance file was processed through Perseus, and proteins were filtered by peptide-spectrum matching (PSM) > 1. Following this process, a volcano plot was generated in Perseus (section on ‘Statistics’). The resulting volcano plots were plotted and analyzed following the procedures listed in the label-free proteomic workflow.

Data analysis with MaxQuant.

The analysis with MaxQuant was executed as previously described in ref. 66. The Phospho(STY) sites file was loaded into Perseus. Rows were filtered based on ‘Reverse’, ‘Potential contaminant’ and localization probability >0.75. Data were then transformed (log₂). The data table was lengthened using ‘Expand site table’. Samples were annotated by treatment using categorical annotation. Rows were filtered by rows containing >50% valid values in each group. Imputation was performed using Perseus default settings. Samples were normalized by median subtraction. A volcano plot was generated using a t-test for statistical significance. The resulting volcano plots were plotted in GraphPad Prism v.10 for final figures. Kinase activity was predicted using RoKAI App (https://rokai.io/). Spectra were visualized using PDV (https://github.com/wenbostar/PDV?tab=readme-ov-file).

In vitro kinase assay for western blot

Recombinant proteins (Supplementary Table 3) were thawed on ice. Proteins were combined in PCR strip tubes and filled to 10 μl with water. +ATP reactions were started by adding 10 μl of Kinase Buffer with 2× ATP/MgCl₂ (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM DTT, 20 mM MgCl₂ (2×), 0.2 mM ATP (2×)) and –ATP control reactions were started by adding 10 μl of Kinase Buffer with 2× MgCl₂ (20 mM Tris-HCl, p 7.5, 150 mM NaCl, 1 mM DTT, 20 mM MgCl₂ (2×)). Reactions were incubated for 30 min at 30 °C on a thermocycler. Reactions were terminated by adding 6.7 μl of 4× Laemmli Sample Buffer and boiling for 5 min at 95 °C on a thermocycler. Western blotting was then performed to detect phosphorylation of the target proteins.

In vitro kinase assay adapted for analysis by MS

Kinase reactions with recombinant c-Myc and SLK.

All kinase reactions were performed in triplicate. Here 1.8 μg of recombinant c-Myc (Raybiotech, cat. no. 230–00580) and 0.2 μg of SLK (Sino Biological, cat. no. S11–10G) were combined in PCR strip tubes and filled to 20 μl with water. The +ATP reactions were started by adding 20 μl of Kinase Buffer with 2× ATP/MgCl₂ (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM DTT, 20 mM MgCl₂ (2×), 0.2 mM ATP (2×)) and –ATP control reactions were started by adding 20 μl of Kinase Buffer with 2× MgCl₂ (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM DTT, 20 mM MgCl₂ (2×)). Reactions were incubated for 1 h at 30 °C on a thermocycler. Reactions were terminated by heat inactivating at 65 °C for 20 min, then cooled at room temperature for 5 min. Then, 4.44 μl of DTT (100 mM) was added and the samples were incubated for 30 min at 56 °C. Next, 1.87 μl of IAA (550 mM) was added and the samples were incubated at room temperature in the dark for 30 min. Proteins were cleaned up and digested using Sera-Mag Carboxylate SpeedBeads (Cytiva Life Sciences: E7 cat. no. 45152105050250 and E3 cat. no. 65152105050250) following the manufacturer’s protocol.

Phosphopeptide enrichment.

Digested peptides were then split into thirds: one-third of the solution was saved as input, and the remaining two-thirds were enriched for phosphopeptides using the Pierce High-Select Fe-NTA Phosphopeptide Enrichment Kit (Thermo Scientific, cat. no. A32992) according to the manufacturer’s instructions.

Quantitative proteomics and MS.

Both input and phosphopeptide-enriched peptide digests were acidified with TFA to 1% and LC-MS/MS analysis was carried out using an Orbitrap Fusion Tribrid mass spectrometer following 2-μg capacity ZipTip (Millipore) C18 sample cleanup, according to the manufacturer’s instructions. Peptides were eluted from an EASY PepMapTM RSLC C18 column (2 μm, 100 Å, 75 μm × 50 cm, Thermo Scientific) into the mass spectrometer using a gradient of 5–25% solvent B (80:20 acetonitrile:water, 0.1% formic acid) in 45 min, followed by 25–44% solvent B in 15 min, 44–80% solvent B in 0.10 min, a 10-min hold of 80% solvent B, a return to 5% solvent B in 3 min and finally with another 3-min hold of 5% solvent B. The gradient was then extended for the purpose of cleaning the column by increasing solvent B to 98% in 3 min, a 98% solvent B hold for 10 min, a return to 5% solvent B in 3 min, a 5% solvent B fold for 3 min, an increase of solvent B to 98% in 3 min, a 98% solvent B hold for 10 min, a return to 5% solvent B in 3 min and a 5% solvent B hold for 3 min and, finally, another increase to 98% solvent B in 3 min and a hold of 98% solvent B for 10 min. All flow rates were 250 nl min⁻¹ delivered using a Vanquish Neo nano LC system (Thermo Fisher Scientific). Solvent A consisted of 0.1% formic acid. Ions were created with an EASY Spray source (Thermo Scientific) held at 50 °C using a voltage of 2.2 kV. Data-dependent scanning was performed by the Xcalibur v.4.0.27.10 software using a survey scan at 120,000 resolution in the Orbitrap analyzer scanning 200–2,000 m/z followed by higher-energy collisional dissociation MS/MS at a normalized collision energy of 30% of the most intense ions at maximum speed, at an AGC of 1.0 × 10⁴. Precursor ions were selected by the monoisotopic precursor selection setting to peptide and MS/MS was performed on charged species of 2–8 charges at a resolution of 30,000. Dynamic exclusion was set to exclude ions once within a 25-s window. All scan events occurred within a 2-s specified cycle time. The analysis was performed at The Herbert Wertheim UF-Scripps Institute for Biomedical Innovation & Technology, Mass Spectrometry and Proteomics Core Facility (RRID SCR_023576).

Data analysis with FragPipe.

Raw data were converted to mzmL files using MSConvert, following the online tutorial (https://fragpipe.nesvilab.org/docs/tutorial_convert.html). Converted data were then processed using FragPipe. A new FASTA file with just MYC and SLK proteins included was used to analyze the data with the built in LFQ-phospho workflow was performed with default settings in FragPipe v.23.1 (precursor mass tolerance −20–20 ppm, protein digestion ENZYMATIC stricttrypsin, peptide length 7–50, peptide mass range (Da) 500–5,000), variable modifications checked: M, [^, STY). Reported intensities from the combined_site_STY_79.9663 output were plotted in Prism v.10, and mass spectra were visualized in PDV.

Diazirine footprinting

Carbene labeling was performed as described previously in ref. 67. Briefly, recombinant SLK and c-Myc proteins (Supplementary Table 3) were prepared in 20 mM Tris-HCl (pH 7.5) containing 150 mM NaCl. SLK (0.4 μM, 1.34 μg) alone or SLK (0.4 μM, 1.34 μg) supplemented with c-Myc (0.5 μM, 0.40 μg) were transferred to separate 1.5-ml microcentrifuge tubes. Sodium 4-(3-(trifluoromethyl)-3H-diazirin-3-yl)benzoate was added from a 100 mM aqueous stock solution to a final concentration of 10 mM. The reaction volume was adjusted to 20 μl using Tris/NaCl buffer. Samples were equilibrated at 30 °C for 30 min, rapidly frozen in liquid nitrogen (77 K) and irradiated with ultraviolet light for 10 min. Following irradiation, proteins were reduced with DTT (10 mM final concentration) for 30 min at room temperature and subsequently alkylated with IAA (15 mM final concentration) for 30 min in the dark. Excess IAA was quenched by addition of DTT to a final concentration of 20 mM and incubation for 15 min at room temperature. Proteins were cleaned up and digested using Sera-Mag Carboxylate Speed-Beads (Cytiva Life Sciences: E7 cat. no. 45152105050250 and E3 cat. no. 65152105050250) following the manufacturer’s protocol. Digested peptides were analyzed with a Fusion Tribrid mass spectrometer and subsequent data analysis were performed as in FragPipe v.23.1 with a custom variable modification of +202.0242 Da corresponding to diazirine-derived labeling. Label-free quantification was performed using the FragPipe DDA workflow. Three independent biological replicates were analyzed for each condition (SLK-only and SLK + c-Myc). For each peptide, the labeling ratio was calculated as the intensity (Int) of the labeled species divided by the total peptide intensity (labeled plus unlabeled species): $\text{labeling ratio}={\text{Int}}_{\text{labeled}}/({\text{Int}}_{\text{labeled}}+{\text{Int}}_{\text{unlabeled}})$. Labeling ratios were averaged across biological replicates and compared between the SLK-only and SLK + c-Myc conditions. A decrease in labeling ratio in the presence of c-Myc was interpreted as protection of the corresponding residue resulting from SLK–c-Myc interaction.

ChIP

WT and SLK-KO PC-3 cells were grown to 80% confluence. Forty million cells were cross-linked in volume of 22 ml of complete media with addition of 37% formaldehyde for final concentration of 1% formaldehyde for 8 min at 37 °C with intermittent agitation. Cross-linking was quenched by the addition of 1.2 ml of 2.5 M glycine for 5 min at room temperature. Cells were pelleted at 500g for 5 min at 4 °C and washed twice with PBS supplemented with protease inhibitors and aliquoted (10 million cells per aliquot) in 1.5-ml Bioruptor TPX tubes (C30010010-300) and frozen at −80 °C. The following day, cell pellets were lysed with 300 μl of ice-cold lysis buffer (50 mM Tris-HCl pH 7.4, 1% SDS, 0.25% DOC) supplemented with fresh protease inhibitors for 10 min on ice. Chromatin was fragmented by sonication using a Bioruptor sonicator (cat. no. B01020014) at high power in a 4 °C water bath with 40 cycles of 30 s on and 30 s off, targeting fragment sizes of 200–1,000 bp as confirmed by 1% agarose gel electrophoresis. Lysates were diluted fourfold with ChIP dilution buffer (50 mM Tris-HCl pH 7.4, 0.1% SDS, 150 mM NaCl, 1.84% Triton X-100) and clarified by centrifugation at 17,000g for 10 min at 4 °C. Supernatants were further diluted with an additional 1.8 ml of ChIP dilution buffer with protease inhibitors to a final volume of 3 ml. A 10–20 μl aliquot of each lysate was reserved as whole-cell extract input control for later analysis. For immunoprecipitation, 1.5 ml of diluted lysate (approximately 5 million cell equivalents) was incubated with 4 μg of antibody overnight at 4 °C on a rotator. The following antibodies were used: anti-MYC (Abcam, cat. no. AB32072), anti-POL2 (Abcam, cat. no. AB26721) and anti-IgG (Cell Signaling, cat. no. 2729S) as negative control. Antibody–chromatin complexes were captured by incubation with prewashed Protein G Dynabeads (Invitrogen) at a concentration of 1 mg per sample for 2 h at 4 °C. Beads were washed sequentially with ice-cold RIPA buffers of increasing stringency: RIPA 150 buffer (0.1% SDS, 0.1% DOC, 1% Triton X-100, 1 mM EDTA, 10 mM Tris-HCl pH 8, 150 mM NaCl) twice, RIPA 500 buffer (0.1% SDS, 0.1% DOC, 1% Triton X-100, 1 mM EDTA, 10 mM Tris-HCl pH 8, 500 mM NaCl) twice, LiCl wash buffer (10 mM Tris-HCl pH 8, 250 mM LiCl, 0.5% Triton X-100, 0.5% DOC) twice and finally with 10 mM Tris-HCl pH 8) twice. Antibody–chromatin complexes were eluted in 100 μl of elution buffer (10 mM Tris-HCl pH 8, 0.1% SDS, 150 mM NaCl, 5 mM DTT) for 1 h at 65 °C with shaking at 600 rpm. RNA was digested with 1 μl of RNase (Roche 11 119 915 001) at 37 °C for 30 min. DNA–protein crosslinks were reversed by incubation with 2 μl of Proteinase K (NEB P8107S) at 65 °C overnight with shaking at 700 rpm. DNA was purified using AMPure XP beads (Beckman Coulter, no. A63881) at 1.8× sample volume, followed by two 70% ethanol washes and elution in 40 μl of molecular grade water. Quality control and sequencing library preparation were performed by the Herbert Wertheim UF-Scripps Genomics Core Facility.

ChIP–seq library preparation and sequencing

Sequencing libraries were prepared using NEBNext Ultra II DNA Library Prep Kit for Illumina (cat. no. E7645S) from purified ChIP DNA following manufacturer protocols. Libraries were sequenced on an Illumina NextSeq500 instrument to generate 50-bp paired-end reads with a minimum sequencing depth of 30 million reads per sample for TFs (MYC and POL2).

ChIP–seq analysis

ChIP–seq libraries for POL2, MYC and IgG controls were generated for two cell lines (WT and SLK-KO) in biological triplicate. Raw paired-end FASTQ files were processed using the nf-core/chipseq pipeline v.2.1.0 (revision 76e2382) executed under Nextflow v.24.10.4 with the singularity profile on the University of Florida HiPerGator HPC cluster. The pipeline was run with default settings using the GRCh38.p14 primary assembly as the reference genome, the GENCODE v.46 (chr_patch_hapl_scaff) annotation GTF, the ENCODE hg38 blacklist v.2 for filtering, a read length of 50 bp and–narrow_peak peak-calling mode. In brief, raw reads were quality-trimmed with Trim Galore v.0.6.7 (Cutadapt v.3.4), assessed with FastQC v.0.12.1 and aligned to GRCh38.p14 with BWA-MEM v.0.7.18-r1243. Aligned reads were sorted and indexed with SAMtools v.1.2/v1.15.1, duplicates were marked with Picard v.3.2.0 and library complexity was estimated with Preseq v.3.1.1 and phantompeak-qualtools v.1.2.2. Per-sample coverage tracks (bigWig) were generated with BEDTools genomecov v.2.30.0 and UCSC bedGraphToBigWig v.445. Per-replicate narrow peaks were called with MACS3 v.3.0.1 against the matched IgG input, and per-antibody consensus peak sets and a sample-by-peak boolean matrix were generated by the MACS3_CONSENSUS module (Python v.3.10.0, R v.4.1.1). Peaks were quantified with featureCounts (Subread v.2.0.1) and annotated with HOMER annotatePeaks.pl v.4.11. Quality control was assessed with deepTools v.3.5.5 (computeMatrix, plotProfile, plotHeatmap, plotFingerprint), Picard CollectMultipleMetrics v.3.2.0 and aggregated into a MultiQC report (sed v.4.7; multiqc). Canonical MYC target genes were obtained from the Molecular Signatures Database (MSigDB) Hallmark collection, specifically the HALLMARK_MYC_TARGETS_V1 gene set (200 genes; MSigDB v.2024.1.Hs). Gene-level coordinates for each symbol were retrieved from the ENSEMBL REST API (assembly GRCh38) using a custom Python v.3.10.12 script, and a 4-kb promoter interval was defined as the transcription start site ±2 kb on each gene’s coding strand. The resulting BED file (HALLMARK_MYC_TARGETS_V1_promoters_hg38. bed) was used as the regions of interest for all downstream overlap and signal-profiling analyses. Per-replicate POL2 and MYC bigWig coverage tracks output by nf-core/chipseq were averaged per antibody and cell line using deepTools v.3.5.5 bigwigAverage to yield merged tracks ({WT,SLK-KO}_{POL2,MYC}_avg.bw). Signal matrices centered on each MYC promoter midpoint (transcription start site reference point, −1 kb to +1 kb, 25-bp bins,–skipZeros) were computed with deepTools v.3.5.2 computeMatrix reference point. Sample-specific and combined matrices were used to generate per-cell-line heatmaps with plotHeatmap. All plots were exported to PDF and 150-dpi PNG files.

Cycloheximide-chase assay

Cells were seeded in six-well plates to reach 80–100% confluency the following day, then incubated overnight at 37 °C with 5% CO₂. The media was then removed and replaced with media containing cycloheximide (50 μg ml⁻¹). At post-treatment collection timepoints, cells were washed with 1 ml of PBS, then lysed directly in the well with ice-cold lysis buffer (M-PER (Thermo Scientific, no. 78501), 1× protease inhibitor (Thermo Scientific no. 1861279), 500 U of Pierce Universal Nuclease for Cell Lysis (Thermo Scientific, no. 88702)). Western blotting was then performed to quantify the amount of the target protein remaining at each timepoint.

Native co-IP

Cells were grown in 15-cm plates until 80–100% confluent. The cells were washed 2–3× with ice-cold PBS, then lysed directly in the plate with EBC buffer (50 mM Tris-HCl, 120 mM NaCl, 0.1 mM EDTA, 0.5% NP-40, 10% glycerol, 1× protease inhibitor cocktail) and collected via scraping. After lysing for 30 min on ice, the lysates were sonicated on the Diagenode Bioruptor Sonicator (LOW, 5 cycles at 4 °C, 7 s ON, 5 s OFF). Samples were then centrifuged (12,000g, 15 min, 4 °C), the supernatant was gathered and all lysates were normalized to 1 mg ml⁻¹. Then 1–5 μg of the primary antibody and the isotype control antibody were conjugated to Pierce Protein A Magnetic Beads (Thermo Scientific, no. 88845) according to the manufacturer’s instructions. Next, lysates were then incubated with the prepared protein A beads overnight at 4 °C with rotation. Beads were washed 4× with wash buffer (50 mM Tris-HCl, 150 mM NaCl, 5 mM EDTA, 0.1% Tween-20) and transferred to new tubes. Proteins were eluted by boiling beads at 100 °C for 5 min in 1× Laemmli sample buffer diluted in water. Target proteins were detected by western blot, using a Clean-Blot immunoprecipitation Detection Reagent (Thermo Scientific, no. 21230) to limit background associated with heavy and light chains ofthe antibodies. For co-IP experiments with nuclease treatment, 500 U of Pierce Universal Nuclease for Cell Lysis (Thermo Scientific, no. 88702) was added to the lysate.

RT–qPCR

Here 1 × 10⁶ cells were collected with trypsin, washed with PBS and RNAs were extracted using TRIzol reagent (Invitrogen, cat. no. 15596026) according to the manufacturer’s instructions. The RNA samples were treated with DNase (Invitrogen, cat. no. AM1907) to remove genomic DNA. The RNA concentrations were measured using Qubit RNA BR Assay Kit (Thermo Scientific, cat. no. Q10210) and normalized to the same concentration. cDNAs were synthesized with The SuperScript III First-Strand Synthesis System (Invitrogen, cat. no. 18080051) according to the manufacturer’s protocol with random hexamers. The cDNA concentration was measured with Qubit single-stranded DNA Assay Kit (Thermo Scientific, cat. no. Q10212) and normalized to 5 ng μl⁻¹. Next, 10 ng of cDNA was loaded in real-time qPCR reactions, which were performed in triplicate using the Power SYBR Green PCR Master Mix (Thermo Scientific cat. no. 4367659) and specified primer pairs (Table 1). Quantification of gene expression was measured using the 5 Real-Time PCR System.

CRISPR-KO cell lines

Cloning into pSpCas9(BB)-2A-GFP.

The design of guide sequences targeting human SLK (Table 1) was done using the CRISPick web tool by the Genetic Perturbation Platform at the Broad Institute. Guide RNA sequences were cloned into pSpCas9(BB)-2A-GFP (PX458) (Addgene, cat. no. 48138)⁶⁸. The PX458 vector was digested with BbsI overnight at 37 °C, followed by gel purification of the digested plasmid. Then 100 μM of forward and reverse guide oligos were phosphorylated with T4 polynucleotide kinase (NEB, cat. no. M0201S) and annealed in a thermocycler (30 min 37 °C, 5 min 95 °C, ramp down 4 °C at 4 °C min⁻¹). The annealed guide oligos were diluted 20× in nuclease-free water and ligated into the BbsI-digested PX458 vector at 16 °C overnight. NEB 5-alpha Competent E. coli (NEB, cat. no. C2987H) were transformed with the ligated product to obtain the PX458 plasmids containing guide RNA sequences targeting human SLK (PX458-SLK).

Transfection of target cells.

Cells were seeded in 10-cm plates to reach 60–75% confluency the following day, then incubated overnight at 37 °C with 5% CO₂. The media was replaced with fresh media, and all plates were transfected with using the Lipofectamine 3000 Transfection Reagent (Thermo Scientific, cat. no. L3000150) and following the manufacturer’s instructions. Then, 18–24 h post-transfection, the cells were collected using trypsin, washed once with PBS and resuspended in fluorescence-activated cell sorting (FACS) buffer (PBS with 1% BSA).

Single-cell sorting and validation of knockout clones.

GFP positive cells were sorted using the BD FACSAria III instrument, and 1 cell was seeded into each well of a 96-well plate containing complete growth media. The clones were incubated at 37 °C with 5% CO₂ until the clonal populations expanded enough to be replated in 10-cm plates. Knockout of the target gene was then validated by western blotting.

Lentiviral transduction for rescue experiments with c-Myc-V5-FLAG-P2A-GFP

Low passage PC-3 SLK-KO cells were counted and seeded at 1 × 10⁵ cells per well in a 12-well plate to reach 30–50% confluency the following day, then incubated overnight at 37 °C with 5% CO₂. Lentivirus (pLV[Exp]-Bleo-EF1A>hMYC[NM_002467.6]*/3xGS/V5/3xGS/FLAG/P2A/EGFP, VectorBuilder, cat. no. VB251202-1300drg) was thawed on ice and diluted in complete media to reach an approximate multiplicity of infection of 10. Cells were washed with PBS and the media was replaced with 400 μl of the lentivirus-containing media, then cells were incubated overnight at 37 °C with 5% CO_2. The following day, the lentivirus-containing media was removed and replaced with complete media. Cells were again incubated overnight at 37 °C with 5% CO_2. The day after that, cells were gathered with trypsin and washed once with PBS before being diluted in FACS buffer for bulk cell sorting. GFP positive cells were sorted using the BD FACSAria III instrument, and all positive cells were plated in a single well of a 12-well plate in complete growth media. The clones were incubated at 37 °C with 5% CO₂ until the clonal populations expanded enough to be replated in 10-cm plates (~1–2 weeks). Complete media was replaced every 3 days. Selection of these cells with Zeocin was not required or performed.

MolBoolean assay

The MolBoolean assay (Atlas Antibodies, cat. no. MolB00001) was executed according to the manufacturer’s instructions with a mouse monoclonal antibody targeting c-Myc and a rabbit polyclonal antibody targeting SLK (Supplementary Table 2). Briefly, cells were gathered with trypsin, washed with PBS, plated on polylysine-treated coverslips, then incubated overnight at 37 °C. Cells were then fixed with 3.7% paraformaldehyde diluted in PBS for 15 min at room temperature. After washing 3× with PBS, cells were permeabilized with 0.2% Triton X-100 in 1× TBS for 15 min at room temperature. The cells were washed with TBST 2× for 2 min at room temperature with rocking. The cells were incubated in blocking solution for 1 h at room temperature, rinsed with TBST and primary antibodies were incubated on the cells overnight at 4 °C. Cells were washed with TBST 3× for 3 min at room temperature, and Proximity probes A and B were diluted 1:80 with Complete Diluent and incubated on the cells for 30 min at 37 °C. After washing with high salt buffer (0.85 M NaCl in 1× TBS) supplemented with 0.05% Tween-20 (HSBT) 1× for 3 min, and TBST 2× for 3 min, Circle oligos were hybridized to the Proximity probes by incubating the cells with Circle oligos diluted in 1× buffer A with 1× additive for 30 min at 37 °C. The cells were washed with TBST 1× for 3 min at room temperature, then the cells were incubated with 1× Nickase enzyme diluted in 1× buffer B for 1 h at 37 °C. The cells were washed with TBST 3× for 3 min at room temperature, then Tag oligos and additive were diluted in TBS to 1× concentration and incubated on the cells for 30 min at 37 °C. The Tag oligo solution was decanted off and, without washing, 1× ligase enzyme with 1× additive diluted in 1× buffer A was incubated on the cells for 30 min at 37 °C. After washing with HSBT 1× for 3 min and TBST 1× for 3 min, Rolling-circle amplification was initiated by adding 1× polymerase enzyme diluted in 1× buffer C to the cells and incubating for 90 min at 37 °C. The cells were then washed with TBST 2× for 5 min and 1× detection oligos diluted in 1× buffer D were incubated on the cells for 30 min at 37 °C. Finally, cells were washed with HSB (1×, 5 min), TBS (1×, 5 min) and 0.2× TBST (2×, 3 min), then mounted to slides using ProLong Gold Antifade Mountant with DNA Stain DAPI (Invitrogen, cat. no. P36935). Images were acquired on the FLUOVIEW FV3000 confocal microscope (Olympus). Analysis of the data was done with CellProfiler as described in the manufacturer’s instructions.

Cell proliferation assay

For this, 2.5 × 10³ cells were seeded into 96-well plates in 100 μl of complete growth media. At desired timepoints postseeding, 100 μl of CellTiter-Glo v.2.0 (Promega, cat. no. G9242) was added to each well. Luminescence was measured with the Tecan i-control infinite M1000Pro microplate reader. The results were analyzed and plotted in GraphPad Prism v.10.

Scratch wound assay

Cells were seeded in 12-well plates to reach 90–100% confluency the following day, then incubated overnight at 37 °C with 5% CO₂. The next morning, a scratch wound was made using a multichannel pipette equipped with P200 pipette tips. The cells were then washed 2× with PBS, and complete media was added to the wells. The rate of wound closure in each well was monitored via live cell imaging. While the cells were incubated at 37 °C with 5% CO₂, brightfield images were taken every 30 min for 24 h on the Nikon Eclipse Ti2 microscope.

Cell-cycle analysis

Cells were seeded in six-well plates to reach 80–100% confluency the following day, then incubated overnight at 37 °C with 5% CO₂. The next morning, 1 × 10⁶ cells per replicate were gathered using trypsin, washed once with PBS and pelleted in 15-ml conical vials. Cells were fixed by slowly adding ice-cold 70% ethanol and incubating at 4 °C for 1 h. The cells were washed 2× with PBS and resuspended in 0.5 ml of FxCycle PI/RNase Staining Solution (Thermo Scientific, cat. no. F10797). Samples were incubated for 15–30 min at room temperature in the dark. The samples were then analyzed on the BD LSRII flow cytometer. The results were analyzed and plotted using FlowJo and GraphPad Prism v.10.

siRNA knockdown

WT PC-3 cells were seeded in six-well plates to reach 20–30% confluency the following day, then incubated overnight at 37 °C with 5% CO₂. The next morning, Accell Human SLK (9748) siRNA (Horizon Discovery, cat. no. A-003850-15-0050) and Accell Non-targeting siRNA no. 1 (Horizon Discovery, cat. no. D-001910-01-20) were prepared according to the manufacturer’s protocol. Briefly, siRNAs were diluted to a working concentration in 1× siRNA buffer (Horizon Discovery, cat. no. B-002000-UB-100), then diluted in Opti-MEM to 250 nM (10×). The DharmaFECT 2 Transfection Reagent (Horizon Discovery, cat. no. T-2002-02) was then prepared in Opti-MEM at 6.7 μl per well of a six-well plate. siRNA was added to the DharmaFECT solution 1:1 v:v and the mixture was incubated for 20 min at room temperature. Finally, the polyplexes were diluted fivefold with antibiotic-free complete medium to reach a final siRNA concentration of 25 nM. The complete media was then removed from the wells and 3.33 ml of 25 nM siRNA was added to each well. Cells were then incubated at 37 °C with 5% CO₂, and media was replaced every 2 days. For timepoints greater than 5 days, cells were passaged 1:3 on the fifth day in culture to prevent overgrowth and cell death. At the desired timepoints, cells were washed 3× with PBS and lysed directly in the well with ice-cold lysis buffer (M-PER (Thermo Scientific, cat. no. 78501), 1× protease inhibitor (Thermo Scientific cat. no. 1861279), 500 U of Pierce Universal Nuclease for Cell Lysis (Thermo Scientific, cat. no. 88702)). Western blotting was then performed to quantify the amount of the target protein remaining at each timepoint.

In vivo tumor growth

All animal experimental procedures were performed in accordance with the guidelines of the UF-Scripps Institutional Animal Care and Use Committee (protocol no. 18-031-03). Eight-week-old male Nod Scid Gamma mice (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ, JAX: 005557, The Jackson Laboratory) were used in this study. All animals were maintained in pathogen-free conditions with an ambient temperature between 20 °C and 26 °C, 30-70% humidity (on average, 50%) and a 12 h/12 h light/dark cycle. To initiate the xenograft tumors, the animals were subcutaneously injected bilaterally with 1.2 × 10⁶ PC-3 WT or PC-3 SLK-KO cells in DMEM (Corning) with 50% Matrigel (Corning). Tumor diameter was measured twice a week with a caliper. The experiment was terminated when the first animal reached the maximum bilateral tumor size (1.5 cm).

c-Myc binding assay

Biotin-PEG3-NHS (ChemScene, cat. no. CS-0106527) was resuspended in dimethylsulfoxide to a stock concentration of 100 mM, then diluted 800-fold in PBS to reach a concentration of 125 μM. Recombinant c-Myc protein (Supplementary Table 3) was resuspended in water at a stock concentration of 2.5 μM and 60 μl of c-Myc protein was mixed with 60 μl of 125 μM biotin-PEG3-NHS and incubated for 1 h at room temperature with shaking at 400 rpm. Excess biotin was removed with 7K MWCO Zeba Spin Desalting Columns (Thermo Scientific, cat. no. 89882) pre-equilibrated 3× with PBS. Biotinylated c-Myc protein was then diluted fivefold with PBS to reach a final concentration of 250 nM. A Pierce Streptavidin Coated Clear 384-Well Plate with Super-Block Blocking Buffer (Thermo Scientific, cat. no. 15405) was then coated with 10 μl of biotinylated c-Myc protein or PBS (background control wells) per well, covered and incubated at room temperature with gentle rocking. Unbound c-Myc was washed off 3 × 5 min with wash buffer (150 mM NaCl, 25 mM Tris-HCl pH 7.5, 0.1% BSA, 0.05% Tween-20). All wells were blocked with 100 μl of 5% gelatin from cold water fish skin (Sigma, cat. no. G7041-100G) in wash buffer at 4 °C, covered and then left overnight with gentle rocking. The next morning, the plates were washed 3× for 5 min with wash buffer. Recombinant glutathione-S-transferase-tagged GSK3β and SLK proteins (Supplementary Table 3) were diluted to 250 nM and serially diluted 1:2 6 times. Both +MYC and background control wells were coated with 10 μl of the SLK or GSK3β serial dilutions, and each curve was run in triplicate with a blank. The plate was covered and incubated for 1 h at room temperature with gentle rocking. The wells were then rinsed 2× with 75 μl of enhanced wash buffer (150 mM NaCl, 25 mM Tris-HCl pH 7.5, 0.1% BSA, 0.3% Tween-20), and washed 8 × 5 min with 100 μl of enhanced wash buffer. The glutathione-S-transferase Antibody [HRP] mAb (Supplementary Table 2) was diluted 1:2,500 in PBS and 100 μl of the detection antibody was added to the wells. The plate was covered and incubated for 1 h at room temperature with gentle rocking. The wells were washed 3× for 5 min with wash buffer, then 45 μl of the 1-Step Slow TMB-ELISA (Thermo Scientific, cat. no. 34024) reagent was added to the wells. After 30 min at room temperature, the reaction was stopped with 2 M sulfuric acid and 450-nm absorbance was measured on the Tecan i-control infinite M1000Pro microplate reader.

Flow cytometry

WT PC-3 cells and SLK-KO PC-3 cells were incubated in 10 ml of calcium-free and magnesium-free DPBS in 15-cm culture dishes for 15 min. Cells were then scraped, collected and pelleted in 1.5-ml microcentrifuge tubes. The cell pellets were washed twice with 1 ml of ice-cold DPBS and resuspended in 1% BSA in DPBS to a final concentration of 1 × 10⁶ cells per ml. Primary antibody (1:1,000 dilution) was added to the cell suspension and incubated for 30 min at 4 °C. Following incubation, cells were pelleted, the supernatant was removed and the cells were washed twice with 1 ml of ice-cold DPBS. Cells were then resuspended in Goat Anti-Mice IgG Alexa Fluor 647 secondary antibody (1:1,000 dilution in 1% BSA in DPBS) and incubated for 1 h at 4 °C. After incubation, cells were washed 3× with 0.5 ml of 5% BSA in DPBS and finally resuspended in 0.5 ml of DPBS for flow cytometry analysis. Data were acquired using the BD LSRII flow cytometer and analyzed with FlowJo software.

Statistics

Two-sided unpaired t-tests were used to assess statistical significance when comparing two groups, and ordinary one-way analysis of variance (ANOVA) was used when comparing three or more groups. Data distributions were assumed to be normal. Half-life was determined by fitting the data to a one-phase decay nonlinear curve. Doubling times were determined by identifying the log phase of the growth curve, transforming the data for that portion of the curve to natural log, then fitting a simple linear regression to the natural log-transformed curve; doubling time (DT) was calculated with the following equation: $\mathrm{DT}=\frac{\ln (2)}{\text{slope}}$. All statistical analyses, curve fitting and Pearson r calculations were performed with GraphPad Prism v.10. The error bars in the figures represent the standard error of the mean unless stated otherwise. For volcano plots of proteomics data, the FDR curve matrices were generated with Perseus using FDR = 0.05 and default settings (t-test, number of randomizations: 250, S0 = 0.1), and fold change and P values were calculated with Perseus. For each experiment, at least three technical and/or biological replicates were used unless otherwise specified in the figure caption.

Site-directed mutagenesis

The S329A and S329D point mutations were executed in the c-Myc-V5-6xHis plasmid (Addgene, cat. no. 176049) using the Q5 Site-Directed Mutagenesis Kit (NEB, cat. no. E0552S) with position specific forward and reverse primers encoding the desired codon switch (Table 1). The protocol was performed exactly as described in the manufacturer’s instructions, and mutations were verified by full plasmid sequencing (Plasmidsaurus).

Cloning

Plasmid sequences were verified by full plasmid sequencing (Plasmidsaurus). DNA sequences and corresponding amino acid sequences for the encoded proteins are provided below for each construct.

c-Myc-HA-CfaN-FLAG was cloned into a pcDNA3.1 vector backbone using Gibson Assembly (New England Biolabs) following the manufacturer’s instructions. The encoded construct was under the control of a cytomegalovirus promoter.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Extended Data

Extended Data Fig. 1 |. Validation of the c-Myc-CfaN transgene.

a) Schematic of iridium-photocatalyst conjugation onto c-Myc via ultrafast split-intein splicing. b) Western blot (HA and FLAG) of nuclear lysates from wild-type or c-Myc-CfaN-transfected HEK293T cells, ± CfaC-Ir treatment (0.5 μM, 37 °C, 40 min). n = 1. c) Immunofluorescent staining of the c-Myc-CfaN transgene in LNCaP cells. Scale bar = 20 μm (100×), 5 μm (200×). n = 1. d) Total biotin labeling by μMap with each component omitted (c-Myc-CfaN, CfaC-Ir, Bt-Dz probe, or 450 nm light); the 1.0 μM CfaC-Ir condition was used in subsequent experiments. n = 1. e) Total RNA-sequencing heatmap of HALLMARK_MYC_TARGETS_V1/V2 genes in WT PC-3 cells and PC-3 cells transfected with c-Myc-V5-6xHis (control) or c-Myc-CfaN (μMap). Statistical significance was assessed using a two-way ANOVA with Dunnett’s multiple comparisons test (1 family, 2 comparisons, α = 0.05) versus WT PC-3; adjusted P-values are shown for the four plotted genes. n = 3 biological replicates; box plots show the range with a line at the mean. f) Correlation of Myc target gene expression after transfection of c-Myc-V5-6xHis versus c-Myc-CfaN, for the genes in the E heatmap.

Extended Data Fig. 2 |. μMap captures known c-Myc interactions in HEK293T cells.

a) (Left) Volcano plot of all proteins in the μMap mass spectrometry data from HEK293T cells, with nuclear proteins colored navy. (Right) Zoomed view of the significantly enriched c-Myc-proximal proteins. n = 3 biological replicates, with 2 technical replicates per biological replicate. P-values were assessed using a two-sided Welch’s t-test with permutation-based FDR correction (FDR = 0.05, S0 = 0.1) in Perseus. b) Flow cytometry of the nuclear isolation, reporting the fraction of biochemically intact nuclei. n = 1. c) Principal component analysis (PCA) of the HEK293T μMap mass spectrometry data. d) Annotation of the top 40 μMap-enriched proteins against the IntAct and BioGRID c-Myc interactor databases. e, f) Gene ontology analysis (molecular functions and cellular components) of the significantly enriched μMap proteins. g) (Left) Cartoon describing rationale for treatment with JQ1 prior to labeling. (Right) Volcano plot of μMap data from cells treated with JQ1 (1 μM, 37 °C, 3 h), with BET proteins BRD2/3/4 indicated. n = 3 biological replicates, with 2 technical replicates per biological replicate. P-values were assessed as described in A.

Extended Data Fig. 3 |. μMap identifies cancer-specific c-Myc interaction.

a) Western blot of endogenous c-Myc in WPMY-1 (healthy), LNCaP, and PC-3 cells. n = 1. b) Total RNA-sequencing of MYC mRNA in WPMY-1, LNCaP, and PC-3 cells. n = 3 biological replicates; data are mean ± SEM. c) Venn diagram of c-Myc interactors identified by μMap across the three cell lines (24 shared). d) Volcano plots of all proteins in the μMap mass spectrometry data from WPMY-1 (left), LNCaP (middle), and PC-3 (right) cells, with MYC and MAX labeled in red. e) Gene ontology analysis of significantly enriched proteins related to proteasomal degradation in each cell line. For d and e, n = 3 biological replicates, with 2 technical replicates per biological replicate; P-values were assessed using a two-sided Welch’s t-test with permutation-based FDR correction (FDR = 0.05, S0 = 0.1) in Perseus. f) Cycloheximide (CHX) chase assays of c-Myc in each cell line (exogenous c-Myc transfected into WPMY-1, where endogenous c-Myc is undetectable). n = 2 biological replicates per timepoint; data are mean ± SEM. g) Analysis of significantly enriched proteins with kinase activity in each cell line. P-values were assessed as described in d and e.

Extended Data Fig. 4 |. A comparison of BioID and μMap proximity labeling methods in PC-3 cells.

a) Schematics of the c-Myc-BirA* fusion and BirA* control plasmids used for BioID in PC-3 cells. b, c) Western blots of BioID construct expression and biotinylation of proximal proteins (50 μM biotin, 24 h). n = 1. d) Total protein IDs from BioID and μMap, with the fraction annotated as c-Myc interactors in BioGRID ( ≥ 2 references). e) Volcano plot of all proteins in the BioID mass spectrometry data from PC-3 cells. f) Scatter plot of c-Myc interactors shared by BioID and μMap (130/243, 54% BioGRID-validated); MYC, MAX, and the ten most μMap-enriched proteins are labeled. For e and f, P-values were assessed using a two-sided Welch’s t-test with permutation-based FDR correction (FDR = 0.05, S0 = 0.1) in Perseus. g) Gene ontology analysis of hits unique to BioID (molecular function).

Extended Data Fig. 5 |. SLK interacts with c-Myc in PC-3 cells specifically.

a) Western blot of endogenous SLK across the three cell lines. n = 1. b) Anti-SLK co-immunoprecipitation followed by c-Myc blotting in WPMY-1, LNCaP, and PC-3 cells. n = 2 biological replicates. c) Anti-SLK co-immunoprecipitation followed by c-Myc blotting in PC-3 cells, ± universal nuclease. n = 1. d) Immunofluorescent staining of c-Myc (green) and SLK (magenta) with rolling-circle amplification (MolBoolean™) in WPMY-1 and PC-3 cells. Scale bar = 10 μm. n = 2 biological replicates per cell line. e) Quantification of c-Myc–SLK colocalization from the MolBoolean assay in D; N = total nuclei analyzed across three images, n = 2 biological replicates. P-values were assessed using a one-way ANOVA with Tukey’s multiple comparisons test (1 family, 6 comparisons, α = 0.05). f) Western blot confirming CRISPR knockout of SLK in WPMY-1 and PC-3 cells. n = 1. g) mRNA expression by RNA-seq of CNBP, NME1, and NME2 in WT and SLK-KO PC-3 cells. n = 4 biological replicates per cell line; data are mean ± SEM. P-values were assessed using a two-way ANOVA with Šídák’s multiple comparisons test (1 family, 3 comparisons, α = 0.05), PC-3 WT vs SLK-KO within each row. Related to Fig. 4a, b. ****adjusted P < 0.0001.

Extended Data Fig. 6 |. Diazirine footprinting identifies c-Myc binding sites in SLK.

a, b) Cartoon schematics of the diazirine-footprinting concept and workflow. c) The degree of diazirine labeling per region of SLK for SLK alone (navy) and SLK + MYC (gray), with the fold change (SLK alone / SLK + MYC) plotted beneath; ‘X’ denotes no labeling in SLK + MYC, and candidate c-Myc binding sites are highlighted in red. n = 3 biological replicates; data are mean + SEM. d) Diagrams of SLK and c-Myc summarizing the reciprocal binding sites and the c-Myc Ser329 phosphorylation site. Illustrations in b created in BioRender; Tong, F. https://biorender.com/a23j248cv0goy (2026).

Extended Data Fig. 7 |. siRNA knockdown of SLK, c-Myc binding assay, and ChIP-sequencing.

a) Western blot of c-Myc following siRNA knockdown of SLK in PC-3 cells (four- to eight-day post-transfection time course). n = 3 biological replicates. b) Quantification of the replicate western blots in A (c-Myc relative to β-actin). n = 3 biological replicates; data are mean ± SEM. c) c-Myc binding curves for recombinant GSK3β (top) and SLK (middle), fitted with a four-parameter variable-slope model in Prism 10. (bottom) Table of fitted Bmax and EC50 values. n = 5 technical replicates (GSK3β), n = 3 technical replicates (SLK); data are mean ± SEM. d) ChIP-seq metagene profiles of c-Myc and RNA Pol II occupancy at HALLMARK_MYC_TARGETS_V1 genes in WT and SLK-KO PC-3 cells. Average of n = 3 biological replicates per group. e) ChIP-seq gene tracks for a representative c-Myc target gene (ODC1), UCSC Genome Browser. Each track is the average of n = 3 biological replicates.

Extended Data Fig. 8 |. Phenotypic analysis after SLK-KO, and TCGA RNA-sequencing data.

a, b) Four-day proliferation of WT and SLK-KO PC-3 (a) and WPMY-1 (b) cells (ATP-based CellTiter-Glo 2.0). c) Doubling times calculated from the curves in A and B. n = 6 technical replicates; bar plots are mean ± SEM. d) Cell-cycle distribution by propidium iodide staining in WT and SLK-KO PC-3 and WPMY-1 cells. n = 3 biological replicates. e) 24-hour scratch (wound-healing) assays of WPMY-1 cells: WT, SLK-KO, and WT + SLK inhibitor (SLK/STK10-IN-1, 10 μM). Scale bar = 200 μm. n = 3 technical replicates. f-k) RNA-sequencing data from the TCGA TARGET GTEX study (UCSC Xena, xenabrowser.net). f) Correlation of SLK-L (red) and SLK-S (blue) with Myc target genes EIF4E and GNL3 in prostate cancer (n = 496). g) SLK-L vs SLK-S expression in prostate cancer; median shown. h) Correlation of SLK-L (red) and SLK-S (blue) with AR expression. i) SLK dependency in prostate cancer cell lines by AR status (n = 2 AR-negative lines); mean ± SD across all cell lines (n = 1,208) shown. j, k) Correlation of SLK-L expression with Myc target gene expression across all samples (n = 19,131) and TCGA cancer samples (n = 10,535).

Extended Data Fig. 9 |. Global proteomics in WT and SLK-KO PC-3 cells.

a–d, f, h) Mass spectrometry comparison of the global proteome of WT and SLK-KO PC-3 cells. n = 3 biological replicates, with 2 technical replicates per biological replicate. P-values were assessed using a two-sided Welch’s t-test with permutation-based FDR correction (FDR = 0.05, S0 = 0.1) in Perseus. Bar plots in c, d, f, and h are mean ± SEM. b) Gene ontology analysis of canonical pathways down- (blue) and up-regulated (red) in the SLK-KO PC-3 proteome. c) Proteome levels of DNA-replication (MCM2-7), proliferation (Ki-67, β-catenin, CDCA8), and mitotic (BRD4, AURKB, PLK1) proteins in WT vs SLK-KO PC-3 cells. d) Proteome levels of the epithelial cell-surface proteins EGFR and EpCAM in WT vs SLK-KO PC-3 cells. e) Flow cytometry of EpCAM expression in WT vs SLK-KO PC-3 cells. f) Proteome levels of autophagy regulators (p62/SQSTM1, OPTN, Atg3-LC3, Atg12-5-16L1, WIPI2, GABARAPL2) in WT vs SLK-KO PC-3 cells. g) Immunofluorescent staining of LC3 A/B in WT vs SLK-KO PC-3 cells, with quantification (right; n = 26 WT, n = 28 SLK-KO cells). Scale bar = 20 μm. n = 2 technical replicates; data are mean ± SD. h) Proteome levels of interferon-stimulated, dsDNA-sensing genes in WT vs SLK-KO PC-3 cells. i–k) Western blots of MX1, OAS2, and LC3 A/B in WT vs SLK-KO PC-3 cells. n = 1. l) Time course of LC3 A/B and p62/SQSTM1 protein in WT vs SLK-KO PC-3 cells. n = 1.

Extended Data Fig. 10 |. Phosphoproteomics in WT and SLK-KO PC-3 cells.

a–c) Gene ontology analysis of the WT vs SLK-KO PC-3 phosphoproteomes. d) (Left) Volcano plot of the phosphoproteome in WT vs SLK-KO PC-3 cells. (Right) Identified peptide from c-Myc phosphodegron motif (F-E-L-L-P-pT58-P-P-L-pS62-P-S) in WT vs SLK-KO cells. n = 5 biological replicates per group; P-values were assessed using a two-sided Welch’s t-test with permutation-based FDR correction (FDR = 0.01, S0 = 0.1) in Perseus. e) Predicted kinase activity (RoKAI App, https://rokai.io/, ≥3 substrates) in WT vs SLK-KO cells, with ATM indicated; raw data analyzed in MaxQuant. P-values were calculated using a two-tailed Z-test. f) Western blot of phospho-ATM (p-Ser1981) in WT vs SLK-KO PC-3 cells. n = 1. g) RT-qPCR of epithelial (LAD1, CDH1) and mesenchymal (ZEB1, SNAI2) markers in WT vs SLK-KO PC-3 cells. n = 3 technical replicates per primer set. h) Nuclear area (DAPI) in WT vs SLK-KO PC-3 cells. n = 30 nuclei across n = 2 technical replicates. Data in G and H are mean ± SD. i) Brightfield images of WT (top) and SLK-KO (bottom) PC-3 cells over time. Scale bar = 50 μm. n = 1.

Supplementary Material

The online version contains supplementary material available at https://doi.org/10.1038/s41589-026-02284-0.

Data availability

All relevant data are included in the paper and Supplementary Information. MS data files have been uploaded to the MassIVE proteomics database (MSV000099411). RNA-seq data files have been uploaded to the Gene Expression Omnibus database (GSE331167). The analyses of publicly available RNA-seq data in this paper are based on data generated by the TCGA Research Network at https://www.cancer.gov/tcga, the Therapeutically Applicable Research to Generate Effective Treatments (TARGET) initiative (phs000218) available at https://www.cancer.gov/ccg/research/genome-sequencing/target and The GTEx Project (supported by the Common Fund of the Office of the Director of the National Institutes of Health, and by NCI, NHGRI, NHLBI, NIDA, NIMH and NINDS). The data used for the analyses described in this paper were obtained from the UCSC Toil RNAseq Recompute Compendium^57,58. Source data are provided with this paper.