To investigate the mechanism by which ZMAT3 promotes growth suppression, we utilized CRISPR/Cas9 to deplete ZMAT3 in HCT116 cells (CRC) using two sgRNAs flanking the p53 response element (RE) in the second intron of ZMAT3. The rationale for deleting the p53RE was based on previous data showing that p53RE of the ZMAT3 gene is critical for ZMAT3 expression (Muys et al., 2021). Deleting the p53RE should, therefore, result in a marked decrease in ZMAT3 expression without disrupting the entire ZMAT3 locus. Although this approach did not completely delete the region spanning the two sgRNAs, a ~57 bp region near sgRNA#1 was deleted, resulting in >75% decrease in ZMAT3 mRNA levels (Figure 1A and B). At the protein level, ZMAT3 was strongly induced only in ZMAT3-WT (wild-type) cells upon treatment with Nutlin, a small molecule that upregulates p53 and its target genes (Figure 1—figure supplement 1A). As expected, upon Nutlin treatment, p53 and its canonical target p21 were induced to similar levels in ZMAT3-WT and isogenic ZMAT3-KO (knockout) cells (Figure 1—figure supplement 1A, Figure 1—figure supplement 2A–B, Figure 1—figure supplement 1—source data 1, Figure 1—figure supplement 2—source data 1). Depletion of ZMAT3 resulted in increased proliferation and clonogenicity, consistent with previous reports (Figure 1C, Figure 1—figure supplement 1B; Bieging-Rolett and Attardi, 2021; Brennan et al., 2024; Muys et al., 2021).

To identify the genes regulated by ZMAT3, we next performed RNA-seq from biological triplicates of ZMAT3-WT and ZMAT3-KO HCT116 cells. As expected, ZMAT3 mRNA levels significantly decreased (~7.5 fold) in ZMAT3-KO cells (Supplementary file 1); the recently identified ZMAT3 target gene MDM4 (Bieging-Rolett et al., 2020) was modestly but significantly upregulated in ZMAT3-KO cells (Supplementary file 1). Transcriptome-wide, upon loss of ZMAT3, 606 genes were significantly up-regulated (adj. p<0.05 and 1.5-fold change) and 552 were down-regulated with a median fold change of 1.76 and 0.55 for the up- and down-regulated genes, respectively (Figure 1D and Supplementary file 1). Because we and others recently reported that ZMAT3 directly regulates alternative splicing (Bieging-Rolett et al., 2020; Muys et al., 2021), we reasoned that ZMAT3-dependent changes in splicing could lead to altered protein levels without altering mRNA levels. We, therefore, performed global quantitative proteomics from ZMAT3-WT and ZMAT3-KO HCT116 cells. At the protein level, 228 proteins were significantly (p<0.05) up-regulated and 108 were down-regulated upon loss of ZMAT3 (Figure 1E; Supplementary file 2).

ZMAT3 directly binds to intronic sequences in pre-mRNAs and can inhibit inclusion of the neighboring exon (Bieging-Rolett et al., 2020; Muys et al., 2021). Depletion of ZMAT3 can, therefore, result in increased target gene expression. Gene set enrichment analysis (GSEA) for the proteins upregulated in ZMAT3-KO cells revealed glycolysis as the most significantly overrepresented biological process (Figure 1F). GSEA analysis from our RNA-seq data showed that glycolysis was among the top 10 overrepresented biological processes for the mRNAs upregulated in ZMAT3-KO cells (Figure 1—figure supplement 1C). Interestingly, the most strongly upregulated protein in ZMAT3-KO cells was the hexokinase HKDC1 (~3.4-fold, p<0.05) (Figure 1E and G and Supplementary file 2). The increase in HKDC1 expression was also observed at the mRNA level; ZMAT3 mRNA downregulation in ZMAT3-KO cells served as positive control (Figure 1H, Figure 1—figure supplement 1D and Supplementary file 1). We, therefore, chose to focus on HKDC1 because it was the most strongly upregulated protein upon loss of ZMAT3 and is directly involved in regulating glucose metabolism and mitochondrial respiration that are cellular processes not previously associated with ZMAT3.

To validate inhibition of HKDC1 expression by ZMAT3, we next performed RT-qPCR from ZMAT3-WT and ZMAT3-KO HCT116 cells. We observed a ~ 4-fold upregulation of HKDC1 mRNA upon ZMAT3 depletion (Figure 2A). This upregulation was further confirmed at the protein level by immunoblotting (Figure 2B, Figure 2—source data 1). Because these experiments were conducted from a single ZMAT3-KO clone, we next analyzed our recently published RNA-seq data (Muys et al., 2021) conducted in biological triplicates from HCT116 cells transfected with a control siRNA (siCTRL) or ZMAT3 siRNAs (SMARTpool of 4 siRNAs). ZMAT3 mRNA was strongly down-regulated, whereas HKDC1 mRNA was modestly but significantly up-regulated (Figure 2—figure supplement 1A and Supplementary file 3), a result that was validated by RT-qPCR (Figure 2C). GSEA for the mRNAs up-regulated upon ZMAT3 knockdown revealed that glycolysis was among the top 10 over-represented biological processes (Figure 2—figure supplement 1B). Additionally, comparison of the RNA-seq data from the ZMAT3-WT vs ZMAT3-KO HCT116 cells and CTRL siRNA vs ZMAT3 siRNA transfected HCT116 cells indicated that 1023 genes were commonly up-regulated (p<0.05 and log₂fold change >0), and 1042 genes were commonly down-regulated (p<0.05 and log₂fold change <0) upon loss of ZMAT3 (Figure 2—figure supplement 1C and D), suggesting that ZMAT3 depletion results in altered expression of thousands of genes. GSEA analysis for the top 500 mRNAs up-regulated upon ZMAT3 knockdown showed that glycolysis was among the top 10 over-represented biological processes (Figure 2—figure supplement 1E). It should be noted that for the comparison of both RNA-seq datasets (ZMAT3-WT vs ZMAT3-KO and siCTRL vs. siZMAT3), we included genes that were consistently up- or down-regulated, without applying a fold change threshold, focusing instead on significantly altered genes (p<0.05) in both datasets. This allowed us to capture broader and more reproducible transcriptomic changes that occur upon ZMAT3 depletion, including modest but significant changes.

To determine if ZMAT3 inhibits HKDC1 expression across diverse cell types, we performed RT-qPCR after ZMAT3 knockdown in SW1222 (CRC cells), HCEC-1CT (immortalized human colonic epithelial cells), and HepG2 (liver cancer cells). We observed a significant increase in HKDC1 mRNA levels upon ZMAT3 knockdown in these cell lines (Figure 2D). At the protein level, we observed that HKDC1 levels increased upon ZMAT3 knockdown in HCT116, HepG2, SW1222, and HCEC-1CT cells (Figure 2E, Figure 2—figure supplement 1F-G, Figure 2—source data 1, Figure 2—figure supplement 1—source data 1). To determine whether this regulatory relationship is conserved between human and mouse, we analyzed recently published RNA-seq data from Zmat3 knockout mouse embryonic fibroblasts (MEFs) (Boutelle et al., 2025). Hkdc1 mRNA was significantly up-regulated (~ 6-fold) in Zmat3-KO MEFs (Figure 2F). As expected, we observed a decrease in Zmat3 mRNA, upregulation of the Zmat3-target gene Mdm4 and no changes in Trp53 mRNA levels in Zmat3-KO MEFs (Figure 2F).

In the context of human CRC, analysis of the TCGA colorectal adenocarcinoma (COAD) cohort revealed that HKDC1 mRNA levels were significantly higher in tumors as compared to normal tissues (Figure 2G). Further analysis of the TCGA COAD data showed that, as compared to p53 wild-type tumors, mutant p53 tumors exhibited significantly higher HKDC1 mRNA levels (Figure 2—figure supplement 1H). As expected for the p53-induced gene ZMAT3, and as shown previously in other cancer types (Bieging-Rolett et al., 2020), ZMAT3 mRNA levels were significantly lower in the mutant p53 tumors compared to p53 wild-type tumors (Figure 2—figure supplement 1I). Furthermore, RNA-seq from Trp53 knockout MEFs from the same study (Boutelle et al., 2025) showed significant upregulation (~8.6 fold) of Hkdc1 mRNA, and down-regulation of Trp53 and its target genes Zmat3 and Cdkn1a (p21) (Figure 2H). Collectively, these data suggest that ZMAT3 and HKDC1 mRNA expression levels are negatively correlated within the p53 pathway, and inhibition of HKDC1 expression by ZMAT3 and p53 is conserved between humans and mice.

Previous studies suggest that HKDC1 plays a crucial role in regulating glucose metabolism and proliferation in various cell types (Khan et al., 2022; Zapater et al., 2022). To determine whether ZMAT3 regulates glucose metabolism and/or proliferation by regulating HKDC1 expression, we performed glucose metabolic assays. Since HKDC1 is a hexokinase that catalyzes the phosphorylation of glucose to glucose 6-phosphate upon entry into cells, we measured changes in hexokinase activity upon ZMAT3 depletion. We incubated the cells with the non-catabolic glucose analog 2-deoxy glucose (2-DG) for a short period of time and quantified the conversion of 2-DG to 2-DG6P using a luminescence-based assay. We observed a significant increase in relative 2-DG6P levels in ZMAT3-KO cells compared to WT cells, and interestingly, this increase was reversed upon HKDC1 knockdown (Figure 3A). Furthermore, siRNA-mediated knockdown of ZMAT3 in SW1222 and HepG2 also showed a similar increase in glucose uptake that was reversed upon simultaneous knockdown of ZMAT3 and HKDC1, suggesting that ZMAT3 inhibits glucose uptake, and this effect is HKDC1-dependent (Figure 3A). We next determined if increased glucose uptake upon ZMAT3 knockdown leads to increased glycolysis in cells. To do this, we performed Seahorse assays to measure extracellular acidification rate and calculated the glycolysis proton efflux. It should be noted that the proton efflux rate measured with Seahorse is reflecting the lactate production from glycolysis more than the pyruvate end point of glycolysis that fueled the mitochondria. Thus, we also measured mitochondria activity. Knockdown of ZMAT3 or HKDC1 by siRNAs resulted in modest, but not significant increases in basal glycolysis (Figure 3B).

Recent studies suggest that HKDC1 plays a significant role in regulating mitochondrial respiration, and depletion of HKDC1 results in mitochondrial dysfunction and senescence (Cui et al., 2024; Khan et al., 2022). We, therefore, examined whether ZMAT3 regulates basal mitochondrial respiration by inhibiting HKDC1 expression. To this end, we performed Seahorse assays to measure the oxygen consumption rate (OCR) following knockdown of HKDC1 and/or ZMAT3. Interestingly, ZMAT3 knockdown resulted in a significant increase in basal mitochondrial respiration, and simultaneous ZMAT3 and HKDC1 knockdown rescued this effect (Figure 3C), without altering oxygen consumption from non-mitochondrial respiration (Figure 3—figure supplement 1). These data indicate that ZMAT3 regulates mitochondrial respiration without significantly affecting glycolysis. It is possible that mitochondria in ZMAT3-KO cells oxidize substrates other than those derived from glycolysis. Further studies will be required to determine these mechanisms in detail.

Because these phenotypes can be associated with proliferation, we next asked whether HKDC1 regulates proliferation and whether it is an effector of ZMAT3. Live cell proliferation assays and cell counting kit-8 (CCK-8) cell viability assays showed that knocking down HKDC1 in ZMAT3-WT and ZMAT3-KO cells resulted in decreased proliferation, but the effect of HKDC1 knockdown was more pronounced in ZMAT3-KO cells (Figure 3D and E). This observation was unexpected, given that HKDC1 acts downstream of ZMAT3. One possible explanation is that elevated HKDC1 expression in ZMAT3-KO cells increases their reliance on HKDC1 for sustaining proliferation and that HKDC1 may also participate in additional pathways in ZMAT3-KO cells. Consequently, transient knockdown of HKDC1 in ZMAT3-KO cells would have a more pronounced effect on proliferation due to their increased dependency on HKDC1 activity. In contrast, ZMAT3-WT cells, which express lower levels of HKDC1 are less dependent on its function and, therefore, less sensitive to its depletion.

Because ZMAT3 transcription is activated by p53, we next examined whether the p53 pathway inhibits HKDC1 expression. To this end, we performed RNA-seq on HCT116 cells transfected with siCTRL or sip53 (SMARTpool of four siRNAs). Interestingly, HKDC1 mRNA was significantly up-regulated (~2.2 fold) upon p53 knockdown (Figure 4A and B and Supplementary file 4). As expected, p53 knockdown led to strong downregulation of mRNAs encoding p53 and its target genes p21 and ZMAT3 (Figure 4A and B). We validated the observed up-regulation of HKDC1 upon p53 knockdown by RT-qPCR and immunoblotting (Figure 4C and D). In these experiments, mRNA and/or protein levels of p53, p21, and ZMAT3 were markedly decreased upon p53 knockdown (Figure 4C, D, Figure 4—source data 1). At the transcriptome-wide level, ~2850 genes were differentially expressed (p<0.05) upon p53 knockdown (Supplementary file 4 and Figure 4—figure supplement 1A). Intersection of these differentially expressed genes (p<0.05) with those regulated by ZMAT3 identified 351 genes commonly upregulated and 425 genes commonly downregulated genes (Figure 4—figure supplement 1B and C). GSEA revealed enrichment of glycolysis and the p53 pathway for the genes upregulated or downregulated following knockdown of p53 or in ZMAT3-KO, respectively (Figure 4—figure supplement 1D and E).

 Since p53 knockdown resulted in increased HKDC1 expression, we next asked whether increasing p53 expression would lead to downregulation of HKDC1. Indeed, analysis of our RNA-seq data revealed ~40% reduction in HKDC1 mRNA levels following Nutlin treatment of siCTRL-transfected HCT116 cells (Figure 4E and Supplementary file 3). As expected for a direct p53 target gene, p53 activation by Nutlin led to upregulation of ZMAT3 and p21 mRNAs by ~2.5 and ~ 6-fold, respectively, while p53 mRNA remained unchanged as Nutlin specifically stabilizes p53 protein levels (Figure 4E). At the protein level, p53 activation by Nutlin decreased HKDC1 levels and increased ZMAT3 and p21 levels in ZMAT3-WT cells (Figure 4F, Figure 4—source data 1). In ZMAT3-KO cells, basal HKDC1 levels were modestly higher, and Nutlin treatment resulted in a slight decrease (Figure 4F, Figure 4—source data 1). To further examine this regulation, we generated doxycycline-inducible ZMAT3-FLAG-HA expressing HCT116 cells to overexpress ZMAT3. Following 48 hr of doxycycline treatment, ZMAT3-FLAG-HA induction was confirmed by RT-qPCR and immunoblotting (Figure 4G and H, Figure 4—source data 1). Furthermore, knockdown of p53 in these cells resulted in a significant increase in both HKDC1 mRNA and protein levels under untreated conditions. Importantly, upon inducing ZMAT3 levels using doxycycline, the upregulation of HKDC1 mRNA and protein associated with p53 knockdown was no longer observed (Figure 4I and J, Figure 4—source data 1). Collectively, these data demonstrate that the p53/ZMAT3 axis plays a critical role in regulating HKDC1 expression, with ZMAT3 acting downstream of p53 to inhibit HKDC1 expression.

ZMAT3 is an RNA-binding protein that regulates alternative splicing. We, therefore, hypothesized that ZMAT3 directly binds to the HKDC1 pre-mRNA and regulates its splicing. However, in our previously published ZMAT3 PAR-CLIP (Muys et al., 2021), we did not detect binding of ZMAT3 to HKDC1 pre-mRNA (data not shown). Furthermore, in RNA immunoprecipitation (RIP) assays performed in doxycycline-inducible ZMAT3-FLAG-HA HCT116 cells, we did not observe significant enrichment of HKDC1 mRNA in the anti-FLAG RNA-IPs (Figure 5—figure supplement 1A). These data suggest that ZMAT3 does not bind to HKDC1 mRNA or pre-mRNA. Moreover, we analyzed changes in HKDC1 pre-mRNA splicing using rMATS in HCT116 cells by reanalyzing our previously published RNA-seq data from siCTRL- and siZMAT3-transfected cells (Muys et al., 2021). We focused on splicing events with an adj.p <0.05 and a ΔPSI > |0.1| (representing at least a 10% change in splicing). The splicing analysis did not reveal any significant alterations in HKDC1 pre-mRNA splicing upon ZMAT3 knockdown, suggesting that the observed increase in HKDC1 mRNA is not at the level of splicing (data not shown).

Because ZMAT3 has three C2H2-zinc-finger motifs, it has the potential to bind DNA (Figure 5—figure supplement 1B; Brayer and Segal, 2008; Gosztyla et al., 2024; Nabeel-Shah et al., 2024). A recent study also reported that ZMAT3 and other zinc finger proteins function as dual DNA-RNA binding proteins (DRBPs) (Gosztyla et al., 2024). To determine whether ZMAT3 binds DNA and regulates HKDC1 transcription directly, we performed ZMAT3-FLAG-HA CUT&RUN-seq and ChIP-seq in three biological replicates using anti-HA, anti-FLAG, or anti-ZMAT3 antibodies. However, we did not detect reproducible ZMAT3 binding at the HKDC1 locus or at other genomic regions in HCT116 cells (data not shown).

We, therefore, hypothesized that ZMAT3 inhibits HKDC1 expression by interacting with a specific transcription factor. To identify proteins that interact with ZMAT3, we performed IPs using an anti-FLAG antibody followed by mass spectrometry on whole-cell lysates from untreated or doxycycline-treated ZMAT3-FLAG-HA HCT116 cells (Figure 5A, Figure 5—figure supplement 1C and Figure 5—figure supplement 1—source data 1). After filtering out common contaminants (<10% in CRAPome), this unbiased approach identified 21 ZMAT3-interacting proteins (Figure 5B and Supplementary file 5). As expected, ZMAT3 was the most strongly enriched protein (~36,000 fold). Of note, due to the low abundance of ZMAT3 in one of the control samples, the p-value for ZMAT3 enrichment was not highly significant (p=0.09) (Supplementary file 5).

Interestingly, the transcription factor JUN (c-Jun) was strongly enriched in the ZMAT3-FLAG pulldowns (~8500 fold). JUN is a proto-oncogene previously implicated in glucose metabolism (Belgardt et al., 2010; Xiao et al., 2019b) and p53 function (Scherer et al., 2000; Eferl et al., 2003). To further explore this potential regulatory link, we analyzed publicly available JUN ChIP-seq data from multiple ENCODE cell lines, focusing on JUN binding at the HKDC1 locus. Notably, in three cell lines, we found a JUN ChIP-seq peak containing the consensus JUN-binding motif within HKDC1 intron 1. This peak coincided with ChIP-seq peaks for POLR2A, H3K27Ac, and H3K4Me3 in HCT116 cells (Figure 5C). We next validated the interaction between ZMAT3 and JUN by performing anti-FLAG IPs followed by immunoblotting on whole-cell lysates from untreated or doxycycline-treated ZMAT3-FLAG-HA HCT116 cells (Figure 5D, Figure 5—source data 1). Treatment with DNase or RNase abolished the interaction between ZMAT3 and JUN, indicating that nucleic acids mediate their association (Figure 5—figure supplement 1D, Figure 5—figure supplement 1—source data 1).

Importantly, JUN knockdown resulted in decreased HKDC1 mRNA levels in ZMAT3-WT cells and rescued the elevated HKDC1 mRNA and protein levels observed in ZMAT3-KO cells (Figure 5E and F, Figure 5—source data 1). To determine whether ZMAT3 inhibits JUN binding to the HKDC1 locus, we performed ChIP-qPCR for JUN in ZMAT3-WT and ZMAT3-KO cells. JUN showed significant enrichment at the HKDC1 intron 1 region compared to the IgG control in ZMAT3-WT cells, and this enrichment was further increased in ZMAT3-KO cells, indicating that ZMAT3 inhibits JUN binding to the HKDC1 intron 1 DNA (Figure 5G). We next cloned a ~700 bp DNA fragment encompassing the JUN and POLR2A binding peaks within HKDC1 intron 1 into the pGL4 basic luciferase reporter vector, hereafter referred to as pGL4-HKDC1-intron 1. Knockdown of JUN in HCT116 cells significantly decreased luciferase activity of the HKDC1 reporter (Figure 5H). Conversely, ZMAT3 knockdown led to a marked increase in luciferase activity, which was rescued by simultaneous knockdown of ZMAT3 and JUN (Figure 5H).

Next, to identify additional JUN targets that might be regulated by ZMAT3, we first examined a set of well-characterized JUN target genes (GLS Lukey et al., 2016, SREBF1 Jin et al., 2022, SLC2A1 Zhu et al., 2022, CD36 Banerjee et al., 2025 ,and WEE1 Kappelmann-Fenzl et al., 2019). Using the ChIPAtlas dataset for human JUN and cross-referencing it with JUN peaks from three ENCODE cell lines, we found that only GLS, SREBF1, and SLC2A1 showed consistent JUN binding in all three cell lines (Figure 5—figure supplement 2A–E). We then measured the expression of GLS, SREBF1, and SLC2A1 by RT-qPCR in ZMAT3-WT and ZMAT3-KO cells, with or without JUN knockdown. GLS mRNA was significantly reduced upon JUN knockdown in both ZMAT3-WT and ZMAT3-KO cells but was not upregulated in ZMAT3-KO cells, indicating that GLS is regulated by JUN in HCT116 cells but is not regulated by ZMAT3. In contrast, SREBF1 and SLC2A1 expression remained unchanged upon JUN knockdown (Figure 5—figure supplement 2F–H). These data suggest that JUN targets are highly cell type specific and not all JUN targets are negatively regulated by ZMAT3. To move beyond this candidate approach and identify additional JUN targets potentially regulated by ZMAT3, we intersected the genes upregulated upon ZMAT3 knockout (from our RNA-seq data) with the ChIP-Atlas dataset for human JUN and cross-referenced these with JUN peaks from three ENCODE cell lines. From this analysis, we selected for further analysis the top four candidate genes - LAMA2, VSNL1, SAMD3, and IL6R (Figure 5—figure supplement 3A–D). Like HKDC1, these genes were upregulated in ZMAT3-KO cells, and this upregulation was abolished upon siRNA-mediated JUN knockdown in ZMAT3-KO cells (Figure 5—figure supplement 3E). Moreover, by ChIP-qPCR, we observed increased JUN binding to the JUN peak for these genes in ZMAT3-KO cells as compared to the ZMAT3-WT (Figure 5—figure supplement 3F). These data suggest that the ZMAT3/JUN axis negatively regulates HKDC1 expression and additional JUN target genes.

Collectively, these data suggest that ZMAT3 plays a crucial role in regulating mitochondrial respiration and cell proliferation by suppressing HKDC1 transcription. We propose a model in which, in ZMAT3-WT cells, p53 drives ZMAT3 transcription, and ZMAT3 protein binds to JUN, thereby inhibiting JUN binding to the HKDC1 locus and repressing HKDC1 transcription. This repression maintains controlled mitochondrial respiration and proliferation. In the absence of ZMAT3, increased JUN binding at the HKDC1 locus leads to elevated HKDC1 expression, enhanced mitochondrial respiration, and increased proliferation (Figure 6).

HCEC-1CT, HepG2, HCT116, and HEK293T cells were purchased from the American Type Culture Collection (ATCC), and SW1222 cells were purchased from https://cancertools.org/. Cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM, Thermo Fisher Scientific, Catalog no. 11995065) supplemented with 10% fetal bovine serum (FBS; Thermo Fisher Scientific, Catalog no. 10082147) and 100 U/mL of penicillin and 0.1 mg/mL of streptomycin (Thermo Fisher Scientific, Catalog no. 15070063). Cultures were incubated at 37 °C in a humidified atmosphere containing 5% CO₂. All cell lines were regularly tested for mycoplasma using the Venor GeM Mycoplasma Detection Kit (Millipore Sigma, Catalog no. MP0025-1KT). All cell lines were authenticated by STR profiling.

To generate ZMAT3-KO clones, we employed the PiggyBac CRISPR/Cas9 system developed from the Zhang lab (Shalem et al., 2014). Two sgRNAs flanking the p53RE within ZMAT3 intron 2 were designed and individually cloned in pENTR221 vector. These constructs were electroporated into 1×10⁶ parental HCT116 cells using the Amaxa Cell Line Nucleofector Kit (Lonza, Catalog no. VCA-1005), together with pT3.5-FLAG-Cas9, pCDNA-pB7, and pBSB-CG-LUC-GFP-(puro)(cre+) vectors. After two days, cells were treated with 2 µg/mL puromycin (Thermo Fisher Scientific, Catalog no. A1113803) for three days. Following puromycin selection, single cells were seeded into 96-well plates to isolate ZMAT3-WT and ZMAT3-KO clones. Clones were expanded for three weeks and transferred to 24-well plates. Total RNA was isolated from each well, and ZMAT3 expression was measured by RT-qPCR and normalized to GAPDH. Genomic DNA was extracted from individual clones showing a strong reduction of ZMAT3 expression, and the genomic region flanking the p53RE of ZMAT3 was PCR-amplified and verified by Sanger sequencing.

The pLVX-Puro-Tet-One vector from TaKaRa (Catalog no. 631849) was used as a backbone to construct an expression construct encoding ZMAT3-3xFLAG-2xHA. The resulting plasmids were transformed into E. coli DH5α cells (Thermo Fisher Scientific, Catalog no. 18265017), and purified using the Monarch Plasmid Miniprep Kit (NEB, Catalog no. T1010L). Lentiviral particles were produced in 3×10⁵ HEK293T cells after co-transfection of 1 µg of plasmid DNA and a third-generation lentiviral packaging system using Lipofectamine 2000 (Thermo Fisher Scientific Catalog no. 11668027). HCT116 cells were transduced at a multiplicity of infection (MOI) of ~1 and, after 2 days, were selected with 2 μg/mL puromycin (Thermo Fisher Scientific, Catalog no. A1113803) for 1 week.

For luciferase assays, a 706 bp genomic region encompassing the JUN-binding peak within the HKDC1 locus was cloned into the pGL4-Basic vector (Promega, Catalog no. E6651), generating the pGL4-HKDC1-intron 1 reporter. To do this, a gene fragment corresponding to chr10:69,222,339–69,223,043 (hg38) was synthesized by Twist Bioscience and included restriction sites for KpnI (NEB, Catalog no. R3142) and XhoI (NEB, Catalog no. R0146S) at 5′ and 3′ ends, respectively. Both the pGL4-Basic vector and insert DNA fragment were digested with these enzymes and purified using the Monarch DNA Gel Extraction Kit (NEB, Catalog no. T1020S) or QIAquick PCR purification kit (Qiagen, Catalog no. 28106), and ligated using T4 DNA ligase (NEB, Catalog no. M0202S) to generate the final pGL4-HKDC1-intron 1 luciferase reporter.

Reverse transfections were performed using Lipofectamine RNAiMAX Transfection Reagent (Thermo Fisher Scientific Catalog, no.13778075) and Opti-MEM (Thermo Fisher Scientific, Catalog no. 31985062) in HCT116, HCEC-1CT, HepG2, and SW1222 cells according to the manufacturer’s protocol. The final concentration of siRNAs was 20 nM. For RT-qPCR and immunoblotting, two rounds of transfection were performed: the second transfection was performed 48 hr after the first transfection and cells were harvested after 72 hr. HKDC1 siRNA transfections were conducted for only one round following the first transfection with siCTRL or siZMAT3. Negative Control siRNA (Qiagen, Catalog no. 1027281) was used as a control. The following SMARTpool siRNAs were used: siZMAT3 (Horizon Discovery, Catalog no. L-017382-00-0005), siJUN (Horizon Discovery, Catalog no. L-003268-00-0005), and sip53 (Horizon Discovery, Catalog no. L-003329-00-0005). For metabolic assays, cells were transfected with siRNAs targeting more than one gene (e.g. siHKDC1 and siZMAT3) at 20 nM each. HCT116 cells expressing ZMAT3-FLAG-HA were transfected for 48 hr, reseeded, and treated with 2 μg/mL doxycycline to induce ZMAT3-FLAG-HA expression.

HCT116 cells were transfected with siRNAs for 48 hr, and then 1×10⁵ cells were reseeded in 24-well plates for luciferase assays. The following day, cells were co-transfected with 250 ng of pGL4 or pGL4-HKDC1-intron 1 and 25 ng of pRL-TK (Promega, Catalog no. E2231), together with 20 nM of either AllStars Negative Control siRNA, siJUN or siZMAT3. Lipofectamine 2000 transfection reagent (Thermo Fisher Scientific, Catalog no. 11668027) was used for the co-transfections, according to the manufacturer’s protocol. After 2 days, firefly and Renilla luciferase activities were measured using the Dual-Luciferase Reporter Assay System (Promega, Catalog no. E1910) on an EnSight Multimode plate reader (PerkinElmer). Firefly luminescence values were normalized to Renilla luminescence to account for transfection efficiency.

Cells were washed with 1 x DPBS (Thermo Fisher Scientific, Catalog no. 14190250) and total RNA was isolated using 500 μL of TRIzol Reagent (Thermo Fisher Scientific, Catalog no. 15596018). To prepare cDNA, 500 ng of total RNA was reverse-transcribed using the iScript Reverse Transcription Supermix (Bio-Rad, Catalog no. 1708841). Quantitative PCR (qPCR) was performed using 2.5 μL of diluted cDNA combined with 5 μL of 2 x FastStart Universal SYBR Green Master (Rox) (Millipore Sigma, Catalog no. 4913914001) and 0.5 μM (final concentration) of each primer in a 10 μL total reaction volume on a StepOnePlus Real-Time PCR machine (Applied Biosystems). Fold change was calculated using the 2^-ΔΔCt method, with GAPDH mRNA serving as housekeeping control.

For immunoblotting, cells were lysed in 1 mL of RIPA buffer (Thermo Fisher Scientific, Catalog no. 89901). Lysates were sonicated three times for 5 s each at 50% amplitude using a VirTis VIRSONIC 100 sonicator and centrifuged at 16,000 × g for 10 min at 4 °C. The supernatant was collected, and protein concentration was determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Catalog no. 23225). For SDS-PAGE, 20–50 µg of total protein was loaded per lane and transferred to a PVDF membrane using a Bio-Rad semi-dry transfer apparatus. Membranes were blocked for 1 hr with TBST (Tris-Buffered Saline: 19.98 mM Tris, 136 mM NaCl, and 0.05% Tween, pH 7.4) containing 5% skim milk and then incubated with the primary antibody overnight at 4 °C. The following primary antibodies were used: anti-FLAG (1:1000 dilution; Sigma, Catalog no. F1804), anti-p53 (DO-1) (1:1000 dilution; Santa Cruz Biotechnology, Catalog no. sc-126), anti-ZMAT3 (1:500 dilution; Santa Cruz Biotechnology, Catalog no. sc-398712), anti-JUN (1:1000 dilution; Cell Signaling, Catalog no. 9165 S), and anti-HKDC1 (Proteintech, Catalog no: 25874–1-AP). Anti-GAPDH antibody (1:6000 dilution; Cell Signaling, 5174 S) was used for loading control. After incubation with HRP-conjugated secondary antibody (1:5000 dilution) for 1 hr at room temperature, the immunoblot was developed using the ECL Prime Western Blotting Detection Reagent (Fisher Scientific, Catalog no. RPN2232). Band intensities were quantified using ImageJ software and normalized to GAPDH.

One thousand ZMAT3-WT and ZMAT3-KO HCT116 cells were seeded into 6-well plates. After 12 days, colonies were fixed with ice-cold methanol for 15 min and stained with 0.5% crystal violet (prepared in 10% methanol) for 15 min. Images were captured, and colony coverage area was quantified using ImageJ software.

For proliferation assays, 1000 cells were seeded per well in a 96-well plate. Cells were incubated in an Incucyte S3 Live-Cell Analysis System (Sartorius) and imaged every 6 hr for 5 days. Images were analyzed using the manufacturer’s software to determine percent confluence over time.

To determine cell viability, cells were incubated with Cell Counting Kit-8 (CCK-8; Dojindo, Kumamoto, Japan) for 4 hr, and absorbance at 450 nm was measured using an Envision microplate reader (PerkinElmer).

siCTRL, siZMAT3, and siHKDC1 were transfected into HCT116, SW1222, and HepG2 cells seeded in poly-L-lysine–coated white 96-well plates with opaque bottoms (Costar) and incubated at 37 °C for 24 hr. After the incubation, the growth medium was removed, and cells were washed twice with PBS to eliminate residual glucose. Glucose uptake was measured using the Glucose Uptake-Glo Assay Kit (Promega, Catalog no. J1341) according to the manufacturer’s instructions. Uptake was initiated by adding 1 mM 2-deoxyglucose and incubating for 10 min at 37 °C, after which luminescence was recorded using an Envision Multimode Plate Reader (PerkinElmer).

First, HCT116 cells were reverse-transfected with siCTRL or siZMAT3 for 48 hr. After 48 hr, a second round of transfection was conducted in 24-well Seahorse plates (Agilent Technologies Inc, Catalog no. 100777–004,) at a concentration of 3x10⁴ per 200 µl per well. After 48 hr cells were washed twice with 500 mL of XF DMEM Medium, pH 7.4 (Agilent Technologies Inc, Catalog no. 103575–10) containing 1 mM pyruvate, 2 mM of glutamine, and 10 mM of glucose (Agilent Technologies Inc, Catalog no. 103578–100, 103579–100, 103577–100). Eventually, cells were incubated for 45 min in a non-CO2 incubator prior to the assays. Meanwhile, drugs from the Mito Stress Test Kit (Agilent Technologies Inc, Catalog no. 103015–10) and Glycolytic Rate Assay Kit (Agilent Technologies Inc, Catalog no. 103344–100) were prepared in XF DMEM Medium, pH 7.4 containing 1 mM pyruvate, 2 mM of glutamine and 10 mM of glucose. For the Mito Stress Test Kit, the working drug solutions concentration were Oligomycin 1.5 µM, FCCP 1.0 µM and Rot/AA 0.5 µM. For the glycolytic Rate Assay Kit, the working drug solutions concentration were Rot/AA 0.5 µM and 2-DG 50 mM. After the 45 min incubation, the plates were loaded into the Seahorse Analyzer and the commercial protocols for the drugs distribution were used. Immediately after the assays, media was removed from wells and cells frozen for later protein concentration measurement. Protein concentrations were measured with Pierce BCA Protein Assay kit (Thermo Fisher Scientific, Catalog no. 23225) and used to normalize Seahorse results.

GSEA was performed using the MSigDB Hallmark gene sets from the Molecular Signature database (https://www.gsea-msigdb.org/gsea/msigdb/human/annotate.jsp).

ZMAT3-FLAG-HA coimmunoprecipitations were performed using anti-FLAG M2-coated magnetic beads (Sigma-Aldrich, Catalog no. M8823). Approximately 5×10⁷ doxycycline-treated or untreated HCT116 ZMAT3-FLAG-HA cells were lysed in IP lysis buffer (10 mM Tris/Cl pH 7.5, 150 mM NaCl, 0.5 mM EDTA, 0.5% NP40) supplemented with 1 mM PMSF and complete protease inhibitor cocktail (Roche). Lysates were incubated for 30 min at 4 °C with gentle mixing and clarified by centrifugation at 16,000 × g for 10 min at 4 °C. Equal amounts of clarified lysates were incubated with prewashed M2 beads overnight at 4 °C with constant rotation. Beads were magnetically separated from the unbound material and washed four times with IP wash buffer (10 mM Tris/Cl pH 7.5, 150 mM NaCl, 0.05% NP40, 0.5 mM EDTA). Bound proteins were eluted with FLAG elution buffer containing 125 µg/mL 3x FLAG peptide (Sigma-Aldrich, Catalog no. F4799). Equal volumes of eluates were boiled at 100 °C for 5 min in Laemmli sample buffer and then centrifuged at 16,000 × g for 5 min at room temperature. Total cell lysate was used as input, and proteins were detected by immunoblotting. For IP-mass spectrometry, bead-bound samples were processed directly after washing the beads, without the elution step.

ZMAT3-FLAG-HA RNA-immunoprecipitations were performed using anti-FLAG M2-coated magnetic beads (Sigma-Aldrich Catalog no. M8823), as described above for the co-IP assays. After the IPs, beads were magnetically separated from the unbound material and washed four times with IP Wash buffer. Bound RNAs were extracted directly from the beads using TRIzol Reagent (Thermo Fisher Scientific, Catalog no. 15596018) following the manufacturer’s protocol.

ChIP-qPCR was performed using the ChIP-IT Express Kit (Active Motif, Catalog no. 53008) following the manufacturer’s instructions. Briefly, 5×10⁷ ZMAT3-WT and ZMAT3-KO HCT116 cells grown in 15 cm plates were cross-linked with 1% formaldehyde, scraped, lysed, and sheared. Chromatin fragment size was verified on a 1% agarose gel. Chromatin was immunoprecipitated overnight at 4 °C using 1 μg of anti-JUN antibody or IgG isotype control. The IP material was washed, eluted, and reverse crosslinked overnight at 65 °C. ChIP DNA was purified using the QIAquick PCR Purification Kit (Qiagen, Catalog no. 28104) and analyzed by qPCR. ChIP-qPCR primers were designed based on the genomic regions harboring JUN-binding peaks in HKDC1, LAMA2, VSNL1, SAMD3, and IL6R (Figure 5C, Figure 5—figure supplement 2A-D).

RNA-seq was performed in biological triplicates from HCT116 ZMAT3-WT and ZMAT3-KO cells. Total RNA was isolated using the RNeasy Plus Mini Kit (Qiagen, Catalog no. 74134) following the manufacturer’s instructions. Libraries were prepared using the Illumina Stranded mRNA Ligation Library Kit with 450 ng of total RNA as the input for mRNA capture using oligo(dT)-coated magnetic beads. The captured mRNA was fragmented and reverse-transcribed using random primers to synthesize first-strand cDNA, followed by second-strand synthesis. The resulting double-stranded cDNA was subjected to standard Illumina library preparation, including end repair, adapter ligation, and PCR amplification, to generate sequencing-ready libraries. The final purified libraries were quantified by qPCR prior to cluster generation and paired-end sequencing on the NovaSeq 6000 platform using an SP 200-cycle kit. Demultiplexing and conversion of binary base call (BCL) files to FASTQ format were performed using Illumina bcl2fastq v2.20. Sequencing reads were trimmed to remove adapters and low-quality bases using Cutadapt (v1.18). Trimmed reads were aligned to the human reference genome (hg38) using the STAR aligner (v2.7.0f) with the two-pass alignment option and GENCODE annotation (v30). Gene and transcript quantification was performed using RSEM (v1.3.1) based on GENCODE annotation.

For total protein identification, cell pellets were suspended in 8 M urea buffer supplemented with protease and phosphatase inhibitors (Roche). All samples were transferred to 2 mL TissueLyser tubes containing 5 mm steel balls and kept on ice. Cells were lysed in a TissueLyser (Qiagen) for 2×2 min, with chilling at −20 °C for 2–3 min between cycles. Lysates were centrifuged at 12,500 × g for 15 min at 4 °C, and supernatants were transferred to new tubes. Protein concentrations were measured using the BCA assay (Thermo Fisher Scientific). For downstream processing, 200 µg of protein from each sample was reduced with 10 mM DTT at 56 °C for 1 hr and alkylated with 20 mM iodoacetamide at room temperature for 30 min in the dark. Following alkylation, samples were diluted fourfold with 50 mM triethylammonium bicarbonate (TEAB) to reduce the urea concentration to 2 M and digested with trypsin (substrate:enzyme = 40:1) at 37 °C overnight. Digested peptides were desalted using C18 columns and lyophilized. Peptide concentrations were measured using the colorimetric BCA peptide assay (Thermo Fisher Scientific). For TMT labeling, 100 µg of digested peptides was labeled with TMTpro 16-plex reagent at room temperature in the dark for 1 hr. Reactions were stopped by the addition of 5% hydroxylamine and incubation at room temperature in the dark for 15 min. Following labeling, peptide samples were pooled and lyophilized. The pooled samples were reconstituted in 0.1% TFA and fractionated using the high-pH reverse-phase peptide fractionation kit (Thermo Fisher Scientific) with nine elution buffers containing 0.1% triethylamine and 10%, 12.5%, 15%, 17.5%, 20%, 22.5%, 25%, 50%, and 75% acetonitrile, respectively. Fractions were lyophilized separately.

For ZMAT3-FLAG interacting protein identification, IP samples were solution digested with trypsin using S traps (Protifi), following the manufacturer’s instructions. Briefly, proteins were denatured in 5% SDS, 50 mM triethylammonium bicarbonate (TEAB) pH 8.5. They were next reduced with 5 mM Tris(2-carboxyethyl)phosphine (TCEP) and alkylated with 20 mM iodoacetamide. The proteins were acidified to a final concentration of 2.5% phosphoric acid and diluted into 100 mM TEAB pH 7.55 in 90% methanol. They were loaded onto the S-traps, washed four times with 100 mM TEAB pH 7.55 in 90% methanol, and digested with trypsin overnight at 37 °C. Peptides were eluted from the S-trap using 50 mM TEAB pH 8.5; 0.2% formic acid in water; and 50% acetonitrile in water. The elutions were pooled and dried by lyophilization.

Dried peptides were resuspended in 5% acetonitrile, 0.05% TFA in water for mass spectrometry analysis on an Orbitrap Exploris 480 (Thermo) mass spectrometer. The peptides were separated on a 75 µm × 15 cm, 3 µm Acclaim PepMap reverse phase column (Thermo) at 300 nl/min using an UltiMate 3000 RSLCnano HPLC (Thermo) and eluted directly into the mass spectrometer. For analysis, parent full-scan mass spectra acquired at 120,000 FWHM resolution and product ion spectra at 45,000 resolution with a 0.7 m/z isolation window. Proteome Discoverer 3.0 (Thermo) was used to search the data against the human database from Uniprot using SequestHT and with INFERYS rescoring. The search was limited to tryptic peptides, with maximally two missed cleavages allowed. Cysteine carbamidomethylation and TMT pro modification of lysine and peptide N-termini were set as a fixed modification, with methionine oxidation as a variable modification. The precursor mass tolerance was 10 ppm, and the fragment mass tolerance was 0.02 Da. The Percolator node was used to score and rank peptide matches using a 1% false discovery rate. TMT quantitation was performed using the Reporter Ions Quantifier nodes with correction of the values for lot-specific TMT reagent isotopic impurities.

Gene expression and clinical data for COAD were obtained from The Cancer Genome Atlas (TCGA) using the TCGAbiolinks package (v2.16.0) in R (v4.0.0). The GDC query function was used to query samples by setting the data category to ‘gene expression,’ data type to ‘gene expression quantification,’ platform to ‘Illumina HiSeq,’ file type to ‘normalized results,’ and experimental strategy to ‘RNA-seq.’ Data were then downloaded using GDC download, followed by GDC prepare to obtain normalized gene expression values.

Statistical analysis for all data was performed using data from at least three independent replicates. The difference between two groups were determined using a two-tailed Student’s t-test, while comparisons involving multiple groups were analyzed using two-way ANOVA.