Metabolic reprogramming can contribute to colorectal cancer progression and therapy resistance. Identification of key regulators of colorectal cancer metabolism could provide new approaches to improve treatment and reduce recurrence. Here, we demonstrate a critical role for the COP9 signalosome subunit CSN6 in rewiring nucleotide metabolism in colorectal cancer. Transcriptomic analysis of colorectal cancer patient samples revealed a correlation between CSN6 expression and purine and pyrimidine metabolism. A colitis-associated colorectal cancer model established that Csn6 intestinal conditional deletion decreased tumor development and altered nucleotide metabolism. CSN6 knockdown increased the chemosensitivity of colorectal cancer cells in vitro and in vivo, which could be partially reversed with nucleoside supplementation. Isotope metabolite tracing showed that CSN6 loss reduced de novo nucleotide synthesis. Mechanistically, CSN6 upregulated purine and pyrimidine biosynthesis by increasing expression of PHGDH, a key enzyme in the serine synthesis pathway. CSN6 inhibited β-Trcp–mediated DDX5 polyubiquitination and degradation, which in turn promoted DDX5-mediated PHGDH mRNA stabilization, leading to metabolic reprogramming and colorectal cancer progression. Butyrate treatment decreased CSN6 expression and improved chemotherapy efficacy. These findings unravel the oncogenic role of CSN6 in regulating nucleotide metabolism and chemosensitivity in colorectal cancer.

Significance:

CSN6 deficiency inhibits colorectal cancer development and chemoresistance by downregulating PHGDH to block nucleotide biosynthesis, providing potential therapeutic targets to improve colorectal cancer treatment.

Colorectal cancer is the third most common type of cancer and a leading cause of cancer-derived mortality worldwide (1–3). Surgery is currently the initial treatment for early-stage colorectal cancer patients, whereas most colorectal cancer patients also rely on chemotherapy (4, 5). However, a certain proportion of patients suffer from recurrence, and chemoresistance is becoming a major obstacle that limits treatment success (6–8). Thus, molecules that could serve as potential therapeutic targets would be useful for overcoming treatment resistance and prolonging patient survival.

The dysregulation of metabolic reprogramming has emerged as a key node in cancer hallmarks, which supports tumor cell proliferation, metastasis, and resistance to therapies (8–10). Notably, recent studies have highlighted the cross-talk between nucleotide synthesis and colorectal cancer tumorigenesis (11). The biosynthesis of nucleotides, which generates most carbon units from the serine–glycine–one-carbon pathway and the pentose phosphate pathway, guarantees timely DNA replication and thereby promotes cancer cell proliferation and resistance to chemotherapy (11–14). For example, 5-fluorouracil (5-FU), the chemotherapeutic agent that is most frequently applied in colorectal cancer, can induce an imbalance in the dNTP pool by competing for the active site of dUMP on thymidylate synthetase (TYMS) and thereby inhibits the conversion of dUMP to dTMP (15, 16). Phosphoglycerate dehydrogenase (PHGDH), the vital enzyme of the SGOC pathway, maintains the de novo biosynthesis of both purine and pyrimidine (17–19). Nevertheless, the mechanism underlying cancer nucleotide biosynthesis is not fully understood, which suggests that additional regulators remain to be explored.

The COP9 signalosome (CSN) contains eight core subunits with various sizes that participate in many physiologic processes, including ubiquitination-mediated degradation, cell cycle, signal transduction, DNA damage, and tumorigenesis (20–22). Previous studies have shown that the CSN6 subunit is highly expressed in different cancer types and plays an essential role in proteasome-associated degradation (23). However, due to the embryonic lethal of CSN6 total knockout mice, most of the studies were performed based on cell lines. Recently, cardiac-specific Csn6 inducible knockout (Csn6-iKO) mice were first constructed and used for arrhythmogenic right ventricular dysplasia research (22). CSN6 function on cancer progression in vivo remains unexplored.

In this study, we crossed CSN6 conditional knockout mice and revealed that the loss of CSN6 hindered purine and pyrimidine synthesis. Stable isotope tracing was performed to confirm the CSN6-inducing de novo nucleotide synthesis, and the underlying mechanism was found to involve the activation of PHGDH. We illustrate that CSN6 stimulates nucleotide metabolic programs by mitigating β-Trcp–mediated DEAD (Asp–Glu–Ala–Asp) box helicase 5 (DDX5) polyubiquitination and degradation, which promotes PHGDH mRNA stability, and subsequently leads to poor cancer survival. Furthermore, our results indicate that butyrate might serve as a potential CSN6 antagonist that could be susceptible to chemoresistance therapeutic intervention.

Animals

NU/NU nude mice and C57BL/6 mice were purchased from GemPharmatech Biotechnology Corporation, respectively. ໿To specifically delete CSN6 in the colon, Csn6fl/fl and Lgr5CreERT mice (GemPharmatech Biotechnology Corporation) were interbred as needed to obtain Csn6CKO mice. All mice were crossed at least 5 generations to a C57BL/6 background. Littermates were used as control for experiments. For xenograft studies, female nude mice of 4 weeks of age were used, which were randomly divided to different treatment groups after cell injections. For the patient-derived xenograft (PDX), PDX models were obtained from the Sixth Affiliated Hospital of Sun Yat-sen University. All mice were maintained under pathogen-free conditions. Animal experiments were approved by the Institutional Animal Care and Use Committee of The Sixth Affiliated Hospital of Sun Yat-sen University (IACUC-2020030901). The numbers of mice per experiment are indicated in the figure legends.

PDX models

Patient-derived tumor fragments (3 mm3) were implanted into the skin of female NCG(NOD/ShiLtJGpt-Prkdcem26Cd52Il2rgem26Cd22/Gpt) mice (GemPharmatech Biotechnology Corporation). When tumors reached approximately 40 to 50 mm3, mice were assigned randomly into two groups and treated with control or NCT-503 (40 mg/kg/day). For the butyrate study, mice were treated with control or butyrate (1.8 g/kg) for 3 weeks [intraperitoneal (i.p.), every other day]. For drug combination, when tumors reached approximately 40 to 50 mm3, mice were assigned randomly into four groups and treated with control, butyrate (1.8 g/kg, every other day), 5-FU (40 mg/kg, twice a week), butyrate + 5-FU for 3 weeks (i.p, every other day]. Randomization and single blinding were performed when measuring tumor volume (volume = 0.5 × length × width2) and weight. Finally, all the mice were sacrificed.

Xenograft tumor model in nude mice

1.5 × 106 DLD1 cells stably infected with shCSN6 or scramble were inoculated subcutaneously into the hind flanks of each 4-week-old female BALB/c-nu/nu mouse. Then, the mice were randomly divided into four groups (n = 7 per group): (i) scramble group; (ii) scramble + 5-FU group (40 mg/kg twice weekly via i.p. injection); (iii) shCSN6 group; and (iv) shCSN6 + 5-FU group (40 mg/kg twice weekly via i.p.). When the tumor volume reached 20 to 40 mm3, we initiated 5-FU therapy and continued for 26 days. Then, the mice were sacrificed and tumors were removed. Randomization and single blinding were performed when measuring the tumor volume (volume = 0.5 × length × width2) and weight.

The azoxymethane/dextran sulfate sodium model

Mice were injected i.p. with tamoxifen (100 mg/kg) for 3 consecutive days at the age of 4 to 6 weeks. One week later, mice were used in the azoxymethane (AOM)/dextran sulfate sodium (DSS)-induced colitis-associated colorectal cancer model. AOM (10 mg/kg) was injected to the mice on day 0. Then, DSS was administered 3 times in weekly intervals. Representative images of colorectal tissue were taken by the ENDOQ colonoscopy system. On day 81, all mice were sacrificed.

Cell lines

All the cell lines were obtained from ATCC and were routinely tested for Mycoplasma. DLD1, SW480, and HT-29 cells were cultured with RPMI-1640 supplemented with 10% (v/v) fetal bovine serum (FBS). HEK293T and HCT116 cells were maintained in Dulbecco's modified Eagle's medium with 10% FBS. Cell lines in this study were passaged about 10 times.

Human samples

Paired colorectal cancer and surrounding nontumor tissues were collected from the Department of Surgery at the Sixth Affiliated Hospital of Sun Yat-sen University with patients’ written informed consent and approval. Tumor acquisition and experimental usage were approved by The Sixth Affiliated Hospital of Sun Yat-sen University Review Board (2017ZSLYEC-111 and 2021ZSLYEC-100).

Metabolite pool and isotopomer labeling analysis

For metabolic analysis, cells were incubated with glucose-free RPMI-1640 medium, supplemented with 10% dialyzed FBS (Dialized, Gibco, #30067334) and 11 mmol/L 13C-glucose (Cambridge Isotope Laboratories, CLM-1396) for 24 hours. Metabolites were extracted by 80% ice-cold methanol. Then, cells were detached and centrifuged for 15 minutes at 15,000 rpm. The supernatant was transferred to LC-MS vials, whereas the pellet was used to determine protein levels for normalization of metabolite levels. For in vivo metabolism measurement, mice were starved overnight and then intraperitoneally injected with a 13C-glucose solution (2 g/kg in PBS). A glucometer was used to monitor blood glucose at 15-minute intervals. Mice were then sacrificed 2.5 hours later when blood glucose levels began to decline (24, 25). For targeted measurement, a Dionex UltiMate 3000 LC System (Thermo Scientific) coupled to a Q-Exactive Orbitrap mass spectrometer (Thermo Scientific) operating in negative mode was used as described previously (17). For the calculation of the total carbon contribution, we corrected for naturally occurring isotopes. Metabolite abundances were presented relative to the internal standard and normalized to the protein content.

Mass spectrometry analysis

Cells in triplicate with or without treatment were lysed in 30 μL SDT buffer (4% SDS, 100 mmol/L Tris-HCl, 1 mmol/L DTT, pH7.6). UA buffer (200 μL) containing 8 M urea and 150 mmol/L Tris–HCl pH 8.0 was used to remove the detergent, DTT, and other low-molecular-weight components. Then, samples were treated with 100 μL iodoacetamide (50 mmol/L IAA in UA buffer) for 30 minutes in dark. After washing, the suspensions were digested with trypsin (Promega) in 25 mmol/L NH4HCO3 buffer overnight at 37°C. Subsequently, 40 μL NH4HCO3 was added, and centrifugation and acidification were performed. Each sample was separated by an HPLC liquid system Easy nLC with a nanoliter flow rate. The column (Thermo Scientific Acclaim PepMap100, 100 μm × 2 cm, nanoViper C18) was equilibrated with 95% liquid A (0.1% formic acid). Samples were loaded by autosampler to a column, then separated by the analytical column (Thermo Scientific EASY Column, 10 cm, ID75 μm, 3 μm, C18-A2) with a 300 nL/minute flow rate in a linear gradient of buffer B (84% acetonitrile and 0.1% formic acid). Next, samples were ionized and introduced into the Q-Exactive mass spectrometer. In brief, detection was positive ion, and the parent ion scan range was 300 to 1,800 m/z. Survey scans were set as a resolution of 70,000 at m/z 200, and the resolution for HCD spectra was set to 17,500 at m/z 200 and the isolation width was 2 m/z. Finally, the acquired raw data were processed using the Mascot2.2 software for identification and quantitation analyses.

RNA immunoprecipitation

DLD1 cells were collected, and RNA immunoprecipitation was performed as described previously (14), using the Magna RIP RNA-Binding Protein Immunoprecipitation Kit (17-701, Millipore). Briefly, cells were lysed in RIP lysis buffer supplemented with a 0.1% RNase inhibitor and 0.2% protease inhibitor cocktail and then subjected to immunoprecipitation. The immunoprecipitated complex was treated with proteinase K, followed by RNA extraction and quantitation by qRT-PCR.

Statistical analysis

All statistical analyses were performed using SPSS software (RRID:SCR_002865) version 22.0. Student t test and one-way ANOVA test were used to analyze quantitative data between groups. Kaplan–Meier survival analyses were used to compare survival, and the log-rank test was used to generate P values. Data are presented as the means ± SD. A Chi-square test was used to estimate the association between CSN6, DDX5, and PHGDH staining intensities. P < 0.05 was considered statistically significant.

Data availability

Microarray data of 20 colorectal cancer samples have been deposited with the Gene-Expression Omnibus (GSE60697). Colorectal cancer data sets were downloaded from the publicly available GEO databases (177 patients from Series GSE17536 and 373 patients from Series GSE2109). Gene set enrichment analysis (GSEA) was performed by the JAVA program (https://www.gsea-msigdb.org/gsea/index.jsp) using metabolism relative kegg.v7.0.symbols gene set collection and visualized in the Enrichment Map software. The MS data have been deposited to the iProX (https://www.iprox.cn/page/home.html, ID: IPX0005453000).

The nucleotide synthesis pathway is associated with colorectal cancer patient survival and is regulated by CSN6

To explore potential metabolic disorders related to colorectal cancer patient recurrence, we compared the gene-expression profiles of 20 colorectal cancer patient samples, including 7 patients who experienced recurrence-free survival (RFS) within 3 years and 13 patients who experienced RFS for more than 3 years in our previous study and then performed a metabolism relative GSEA. Notably, we found that the pyrimidine and purine metabolism pathways were enriched in colorectal cancer patients who suffered recurrence (Fig. 1A).

Figure 1.

The nucleotide synthesis pathway is associated with colorectal cancer patient survival and regulated by CSN6. A, GSEA of Kyoto Encyclopedia of Genes and Genomes (KEGG) metabolic pathways between 20 colorectal cancer patient samples with or without recurrence. Top 10 KEGG metabolic pathways and PYRIMIDINE METABOLISM pathway among recurrence colorectal cancer patient samples. B, NTP (ATP, GTP, UTP, and CTP) levels in DLD1 cells with or without CSN6 shRNA treatment. Data, means ± SD. ***, P < 0.001, t test. C, Scheme of establishing colitis-associated colorectal cancer model of CSN6 conditional knockout mice. D, Representative colonoscopy images of Lgr5-CreERT2;Csn6fl/fl and Csn6fl/fl mice. Arrows, tumors. E, Representative colon tumor images and hematoxylin and eosin images of Lgr5-CreERT2;Csn6fl/fl and Csn6fl/fl mice. Arrows, tumors. Scale bar, 1 mm. F, Tumor numbers and sizes of Lgr5-CreERT2;Csn6fl/fl and Csn6fl/fl mice were quantitated. *, P < 0.05 by nonparametric test. G, NTP (ATP, CTP, UTP, and GTP) levels in the colonic tumors from Lgr5–CreERT2;Csn6fl/fl and Csn6fl/fl mice (n = 4 per group). Data, means ± SD. *, P < 0.05; **, P < 0.01 by a nonparametric test. H, IMP and UMP levels in colonic tumors from Lgr5-CreERT2;Csn6fl/fl and Csn6fl/fl mice (n = 4 per group). Data, means ± SD. **, P < 0.01 by nonparametric test. I and J, Mice inoculated with shCSN6 or control expressing DLD1 cells were injected with PBS or 5-FU treatment (40 mg/kg twice weekly via i.p. injection). Tumor growth was monitored. Graph, means ±SD. ***, P < 0.001 by the two-way ANOVA test. K, IncuCyte machine was used to evaluate the effects of nucleosides repletion (adenine, guanine, cytosine, and uracil; 100 μmol/L) on cancer cell proliferation upon CSN6 depletion and 5-FU treatment. Data, means ± SD. **, P < 0.01; ***, P < 0.001 by the two-way ANOVA test.

Figure 1.

The nucleotide synthesis pathway is associated with colorectal cancer patient survival and regulated by CSN6. A, GSEA of Kyoto Encyclopedia of Genes and Genomes (KEGG) metabolic pathways between 20 colorectal cancer patient samples with or without recurrence. Top 10 KEGG metabolic pathways and PYRIMIDINE METABOLISM pathway among recurrence colorectal cancer patient samples. B, NTP (ATP, GTP, UTP, and CTP) levels in DLD1 cells with or without CSN6 shRNA treatment. Data, means ± SD. ***, P < 0.001, t test. C, Scheme of establishing colitis-associated colorectal cancer model of CSN6 conditional knockout mice. D, Representative colonoscopy images of Lgr5-CreERT2;Csn6fl/fl and Csn6fl/fl mice. Arrows, tumors. E, Representative colon tumor images and hematoxylin and eosin images of Lgr5-CreERT2;Csn6fl/fl and Csn6fl/fl mice. Arrows, tumors. Scale bar, 1 mm. F, Tumor numbers and sizes of Lgr5-CreERT2;Csn6fl/fl and Csn6fl/fl mice were quantitated. *, P < 0.05 by nonparametric test. G, NTP (ATP, CTP, UTP, and GTP) levels in the colonic tumors from Lgr5–CreERT2;Csn6fl/fl and Csn6fl/fl mice (n = 4 per group). Data, means ± SD. *, P < 0.05; **, P < 0.01 by a nonparametric test. H, IMP and UMP levels in colonic tumors from Lgr5-CreERT2;Csn6fl/fl and Csn6fl/fl mice (n = 4 per group). Data, means ± SD. **, P < 0.01 by nonparametric test. I and J, Mice inoculated with shCSN6 or control expressing DLD1 cells were injected with PBS or 5-FU treatment (40 mg/kg twice weekly via i.p. injection). Tumor growth was monitored. Graph, means ±SD. ***, P < 0.001 by the two-way ANOVA test. K, IncuCyte machine was used to evaluate the effects of nucleosides repletion (adenine, guanine, cytosine, and uracil; 100 μmol/L) on cancer cell proliferation upon CSN6 depletion and 5-FU treatment. Data, means ± SD. **, P < 0.01; ***, P < 0.001 by the two-way ANOVA test.

Close modal

Our previous study found that colorectal cancer patients with poor survival exhibited high CSN6 expression (21). Therefore, we wondered whether there was a connection between nucleotide metabolism and CSN6 expression in colorectal cancer. The GSEA of GSE2109, which contains 373 colorectal cancer samples, revealed that purine and pyrimidine metabolism is positively correlated with high CSN6 expression. This finding was further validated with another data set (GSE17536, which contains 177 colorectal cancer samples; Supplementary Fig. S1A and S1B). We then sought to verify the impact of CSN6 on nucleotide synthesis. As shown in Fig. 1B and Supplementary Fig. S1C, the depletion of CSN6 by short hairpin RNA (shRNA) decreased the abundance of NTPs [adenosine triphosphate (ATP), guanosine triphosphate (GTP), cytidine triphosphate (CTP), and uridine triphosphate (UTP)] in DLD1, HCT116, and SW480 cells. We also found that the CSN6 knocking down cells had lower levels of various intermediates, including inosine monophosphate (IMP) and uridine monophosphate (UMP; Supplementary Fig. S1D). Additionally, nucleoside repletion could partially restore the proliferation of CSN6 knocking down cells (Supplementary Fig. S1E), which suggests that CSN6 regulates nucleotide synthesis in colorectal cancer.

Besides, we first generated CSN6 intestinal conditional knockout (Csn6CKO) mice with CSN6 alleles flanked by loxP sites, which was deleted by Cre recombinase expressed from an Lgr5 promoter (Supplementary Fig. S2A and S2B). Then, AOM and DSS were administered to establish a colitis-associated colorectal cancer model (Fig. 1C). Mice were treated with a single intraperitoneal dose (10 mg/kg of body weight) of AOM followed by three 6-day cycles of oral 2% or 1.5% DSS (Supplementary Fig. S2C). Colonoscopy and histopathologic analysis revealed that Lgr5-CreERT2;Csn6fl/fl mice developed smaller and fewer number of tumors compared with Csn6fl/fl control mice (Fig. 1DF). Consistently, the expression levels of Ki67 were significantly lower in colon histologic images from Lgr5-CreERT2;Csn6fl/fl mice (Supplementary Fig. S2D). Importantly, nucleotide levels in colon tumors of Lgr5-CreERT2;Csn6fl/fl mice were significantly decreased (Fig. 1G and H), further confirming the function of CSN6 on nucleotide metabolism regulation.

Fluorouracil (FU), an analogue of uracil that suppresses thymidine synthesis, leading to the depletion of intracellular dTTP pools, incorporation into RNA, and disruption of DNA biosynthesis, is the first-line chemotherapeutic drug for colorectal cancer (15, 25–27). Indeed, CSN6 knocking down colorectal cancer cells were more sensitive to 5-FU treatment (Supplementary Fig. S2E). To further detect whether the depletion of CSN6 can improve 5-FU efficacy in vivo, a subcutaneous xenograft model was established by inoculating mice with control shRNA or shCSN6 expressing DLD1 cells and treated with 5-FU. We found that there was a remarkable synergistic effect of shCSN6 and 5-FU that led to almost total regression of tumors compared with the single agent and control mice (Fig. 1I and J), as well as with tumor weight (Supplementary Fig. S2F). In agreement, the detection of the in vivo functional contribution of CSN6 to metabolism revealed that CSN6 depletion caused a pronounced decrease in the NTP, IMP, and UMP levels in xenograft tumors (Supplementary Fig. S2G and S2H). CSN6 knockdown–induced chemosensitivity could be partially rescued by nucleoside supplementation in colorectal cancer cells (Fig. 1K), which further proves that abrogation of nucleotide metabolism by CSN6 depletion plays an essential role in chemosensitivity.

De novo nucleotide synthesis is altered by the silencing of CSN6 expression

It has been well characterized that glucose provides carbons for IMP or UMP through ribose-5-phosphate (R5P) from the pentose phosphate pathway and partially through 10-formyl tetrahydrofolate from the serine–glycine–one-carbon pathway (11, 12). Interestingly, CSN6 knocking down had no effect on the ribose-5-phosphate levels (Supplementary Fig. S3A). In contrast, the depletion of CSN6 significantly decreased the total intracellular serine and glycine pools (Supplementary Fig. S3B).

To gain further insights, we used uniformly labeled [U-13C] glucose for stable isotope tracing. As shown in Fig. 2AD and Supplementary Fig. S3C, shCSN6 expression in DLD1 and HCT116 cells resulted in lower serine and glycine biosynthesis rates, which are key components in the de novo purine metabolic pathway but had no effect on R5P or 3-phosphoglycerate (3-PG). Furthermore, the knockdown of CSN6 led to a pronounced decrease in car-aspartate and dihydroorotate production, which are key components in the de novo pyrimidine metabolic pathway (Fig. 2F and G; Supplementary Fig. S3C; ref. 28). As a result, the depletion of CSN6 resulted in altered distributions of IMP and UMP (m + 6 to m + 9; Fig. 2E and H; Supplementary Fig. S3D).

Figure 2.

De novo nucleotide synthesis is altered by the silencing of CSN6 expression. A, Incorporation of carbon atoms from [U-13C] glucose into ribose-5-phosphate (R5P) in control and CSN6-KD (knockdown) DLD1 cells. Data, means ± SD. One-way ANOVA. B, Incorporation of carbon atoms from [U-13C] glucose into 3-phosphoglycerate (3-PG) in control and CSN6-KD DLD1 cells. Data, means ± SD. One-way ANOVA. C, Incorporation of carbon atoms from [U-13C] glucose into serine in control and CSN6-KD DLD1 cells. Data, means ± SD. *, P < 0.001; **, P < 0.01 by one-way ANOVA. D, Incorporation of carbon atoms from [U-13C] glucose into glycine in control and CSN6-KD DLD1 cells. Data, means ± SD. ***, P < 0.001 by one-way ANOVA. E, Fractional contribution of carbon atoms from [U-13C] glucose into the IMP (m + 0 to m + 10) in control and CSN6-KD DLD1 cells. *, P < 0.05; **, P < 0.01; ***, P < 0.001 by one-way ANOVA. F, Incorporation of carbon atoms from [U-13C] glucose into car-aspartate in control and CSN6-KD DLD1 cells. ***, P < 0.001 by one-way ANOVA. G, Incorporation of carbon atoms from [U-13C] glucose into dihydroorotate in control and CSN6-KD DLD1 cells. ***, P < 0.001 by one-way ANOVA. H, Fractional contribution of carbon atoms from [U-13C] glucose into UMP isotopomers (m + 0 to m + 9) in control and CSN6-KD DLD1 cells. Data, means ± SD. **, P < 0.01; ***, P < 0.001 by one-way ANOVA. I, Schematic of in vivo tissue metabolism measurement. J, Incorporation of carbon atoms from [U-13C] glucose into serine and glycine in control tumors (scramble) compared with CSN6 knockdown (shCSN6) DLD1 xenograft tumors (n = 5 per group). Data, means ± SD. **, P < 0.01 by a nonparametric test. ns, no significance.

Figure 2.

De novo nucleotide synthesis is altered by the silencing of CSN6 expression. A, Incorporation of carbon atoms from [U-13C] glucose into ribose-5-phosphate (R5P) in control and CSN6-KD (knockdown) DLD1 cells. Data, means ± SD. One-way ANOVA. B, Incorporation of carbon atoms from [U-13C] glucose into 3-phosphoglycerate (3-PG) in control and CSN6-KD DLD1 cells. Data, means ± SD. One-way ANOVA. C, Incorporation of carbon atoms from [U-13C] glucose into serine in control and CSN6-KD DLD1 cells. Data, means ± SD. *, P < 0.001; **, P < 0.01 by one-way ANOVA. D, Incorporation of carbon atoms from [U-13C] glucose into glycine in control and CSN6-KD DLD1 cells. Data, means ± SD. ***, P < 0.001 by one-way ANOVA. E, Fractional contribution of carbon atoms from [U-13C] glucose into the IMP (m + 0 to m + 10) in control and CSN6-KD DLD1 cells. *, P < 0.05; **, P < 0.01; ***, P < 0.001 by one-way ANOVA. F, Incorporation of carbon atoms from [U-13C] glucose into car-aspartate in control and CSN6-KD DLD1 cells. ***, P < 0.001 by one-way ANOVA. G, Incorporation of carbon atoms from [U-13C] glucose into dihydroorotate in control and CSN6-KD DLD1 cells. ***, P < 0.001 by one-way ANOVA. H, Fractional contribution of carbon atoms from [U-13C] glucose into UMP isotopomers (m + 0 to m + 9) in control and CSN6-KD DLD1 cells. Data, means ± SD. **, P < 0.01; ***, P < 0.001 by one-way ANOVA. I, Schematic of in vivo tissue metabolism measurement. J, Incorporation of carbon atoms from [U-13C] glucose into serine and glycine in control tumors (scramble) compared with CSN6 knockdown (shCSN6) DLD1 xenograft tumors (n = 5 per group). Data, means ± SD. **, P < 0.01 by a nonparametric test. ns, no significance.

Close modal

To investigate the metabolic consequences of CSN6 knockdown in vivo, mice xenografted with shCSN6 or control DLD1 cells were injected with [U-13C] glucose, and metabolites of tumors were measured (Fig. 2I). CSN6 knocking down xenograft tumors and colon tumors of Lgr5-CreERT2;Csn6fl/fl mice exhibited a marked reduction in serine and glycine levels (Supplementary Fig. S3E and S3F). With the glucose tracer, we confirmed that CSN6 knockdown affected serine and glycine biosynthesis in vivo (Fig. 2J). Taken together, the observed isotope labeling patterns demonstrated that the knockdown of CSN6 altered the de novo synthesis of both purine and pyrimidine nucleotides.

The silencing of CSN6 impaired purine and pyrimidine synthesis through PHGDH

PHGDH, a key enzyme in the de novo serine synthesis pathway, plays an essential role in nucleotide metabolism (18). We cultured patient-derived organoids (PDO) in vitro and found that nucleoside repletion could partially restore the organoid formation when knocking down PHGDH (Supplementary Fig. S4A; Supplementary Table S1). Besides, PHGDH knockdown–induced chemosensitivity could be partially rescued by nucleoside supplementation in colorectal cancer cells (Fig. 3A).

Figure 3.

The silencing of CSN6 impaired purine and pyrimidine synthesis through PHGDH. A, IncuCyte machine was used to evaluate the effects of nucleoside repletion (adenine, guanine, cytosine, and uracil; 100 μmol/L) on cancer cell proliferation upon PHGDH depletion and 5-FU treatment. Graph, means ± SD. **, P < 0.01; ***, P < 0.001 by the two-way ANOVA test. B, qRT-PCR and Western blot analysis of PHGDH expression in DLD1 and SW480 cells stably expressing CSN6 shRNA or scramble. Data, means ± SD. ***, P < 0.001 by one-way ANOVA test. C, NTP (ATP, CTP, UTP, and GTP) levels in DLD1 cells with or without PLVX-CSN6 or shPHGDH treatment. Data, means ± SD. *, P < 0.05; **, P < 0.01 by one-way ANOVA. D, Incorporation of carbon atoms from [U-13C] glucose into serine, glycine in DLD1 cells with or without PLVX-CSN6 or shPHGDH treatment. Data, means ± SD. **, P < 0.01 by one-way ANOVA. E, Fractional contribution of carbon atoms from [U-13C] glucose into the UMP in DLD1 cells with or without PLVX-CSN6 or shPHGDH treatment. Data, means ± SD. *, P < 0.05 by one-way ANOVA. F, CSN6–PHGDH axis promotes colorectal cancer tumorigenesis.DLD1 cells with or without shCSN6 or PLVX–PHGDH infection were subcutaneously injected into nude mice (n = 4). Tumors were isolated at the end of experiments. Tumor volume is shown. **, P < 0.01; ***, P < 0.001 by the two-way ANOVA test. G, Proliferation assay of control and PLVX-CSN6 DLD1 cells treated with NCT503 or NCT-503i. Graph represents means ±SD. *, P < 0.05; ***, P < 0.001 by the two-way ANOVA test. H, IHC staining represents the expression levels of CSN6 and PHGDH in indicated PDXs, and fluorescence images represent abundances of IMP and UMP in tissues. I, Impact of NCT-503 on tumor growth in indicated PDX-bearing mice. Data, means ± SD. **, P < 0.01 by the two-way ANOVA test. J, Impact of NCT-503 and 5-FU combination on tumor growth in mice bearing indicated PDXs (case E). Graph, means ± SD. *, P < 0.05 by two-way ANOVA test. K, Impact of NCT-503 and 5-FU combination on tumor weight in mice bearing indicated PDXs (case E). Data, means ± SD. **, P < 0.01; ***, P < 0.001, by one-way ANOVA. ns, no significance.

Figure 3.

The silencing of CSN6 impaired purine and pyrimidine synthesis through PHGDH. A, IncuCyte machine was used to evaluate the effects of nucleoside repletion (adenine, guanine, cytosine, and uracil; 100 μmol/L) on cancer cell proliferation upon PHGDH depletion and 5-FU treatment. Graph, means ± SD. **, P < 0.01; ***, P < 0.001 by the two-way ANOVA test. B, qRT-PCR and Western blot analysis of PHGDH expression in DLD1 and SW480 cells stably expressing CSN6 shRNA or scramble. Data, means ± SD. ***, P < 0.001 by one-way ANOVA test. C, NTP (ATP, CTP, UTP, and GTP) levels in DLD1 cells with or without PLVX-CSN6 or shPHGDH treatment. Data, means ± SD. *, P < 0.05; **, P < 0.01 by one-way ANOVA. D, Incorporation of carbon atoms from [U-13C] glucose into serine, glycine in DLD1 cells with or without PLVX-CSN6 or shPHGDH treatment. Data, means ± SD. **, P < 0.01 by one-way ANOVA. E, Fractional contribution of carbon atoms from [U-13C] glucose into the UMP in DLD1 cells with or without PLVX-CSN6 or shPHGDH treatment. Data, means ± SD. *, P < 0.05 by one-way ANOVA. F, CSN6–PHGDH axis promotes colorectal cancer tumorigenesis.DLD1 cells with or without shCSN6 or PLVX–PHGDH infection were subcutaneously injected into nude mice (n = 4). Tumors were isolated at the end of experiments. Tumor volume is shown. **, P < 0.01; ***, P < 0.001 by the two-way ANOVA test. G, Proliferation assay of control and PLVX-CSN6 DLD1 cells treated with NCT503 or NCT-503i. Graph represents means ±SD. *, P < 0.05; ***, P < 0.001 by the two-way ANOVA test. H, IHC staining represents the expression levels of CSN6 and PHGDH in indicated PDXs, and fluorescence images represent abundances of IMP and UMP in tissues. I, Impact of NCT-503 on tumor growth in indicated PDX-bearing mice. Data, means ± SD. **, P < 0.01 by the two-way ANOVA test. J, Impact of NCT-503 and 5-FU combination on tumor growth in mice bearing indicated PDXs (case E). Graph, means ± SD. *, P < 0.05 by two-way ANOVA test. K, Impact of NCT-503 and 5-FU combination on tumor weight in mice bearing indicated PDXs (case E). Data, means ± SD. **, P < 0.01; ***, P < 0.001, by one-way ANOVA. ns, no significance.

Close modal

It was confirmed that CSN6 depletion decreased the expression of PHGDH in different colorectal cancer cell lines (Fig. 3B; Supplementary Fig. S4B), while reintroducing PLVX-CSN6 could rescue PHGDH gene expression (Supplementary Fig. S4C). Supplementation of serine in mediums significantly rescued cancer cell growth, which was suppressed by CSN6 knockdown (Supplementary Fig. S4D).

It is well reported that the inhibition of PHGDH induces alterations in both purine and pyrimidine metabolism (17, 18). Here, ectopic expression of CSN6 promoted the NTPs levels in cells, concomitant with an increase of the incorporation of [U-13C] glucose into serine, glycine, and UMP biosynthesis. Although PHGDH depletion notably abrogated these effects (Fig. 3CE; Supplementary Fig. S4E). Moreover, the silencing of PHGDH was effective in abrogating CSN6-induced colorectal cancer cell proliferation and colony formation ability (Supplementary Fig. S4F and S4G; ref. 29). In vivo mice models confirmed the roles of the CSN6–PHGDH axis in colorectal cancer tumorigenesis (Fig. 3F; Supplementary Fig. S4H and S4I).

Consistently, NCT-503 treatment, a PHGDH inhibitor, caused significant inhibition of CSN6-overexpressed colorectal cancer cell growth (Fig. 3G; Supplementary Fig. S5A). Then, we collected fresh colorectal cancer tumor samples to establish PDXs. The expression of PHGDH and the abundance of UMP and IMP were higher in CSN6-high expressed colorectal cancer samples (Fig. 3H). Then, we used these PDXs for drug efficacy testing. Significantly, NCT-503 treatment decreased PDX tumor volume and tumor weight, which has a higher level of CSN6 expression (Fig. 3I; Supplementary Fig. S5B; Supplementary Table S1). By contrast, administration of NCT-503 had no significant impact on PDX tumor growth from colorectal cancer with low CSN6 expression (Fig. 3I; Supplementary Fig. S5C and Supplementary Table S1). These findings were confirmed by the other two cases (Supplementary Fig. S5D and S5E; Supplementary Table S1). Moreover, combination therapy with 5-FU and NCT-503 showed the lowest tumor volume and tumor weight compared with 5-FU alone or NCT-503 alone (Fig. 3J and K). CSN6 knockdown-induced chemosensitivity was partially rescued by nucleoside supplementation in colorectal cancer cells, while overexpressing PHGDH at the same time, the effect would be partially abrogated (Supplementary Fig. S5F). These findings support our hypothesis that CSN6 affects nucleotide synthesis via PHGDH.

CSN6 promotes PHGDH expression through DDX5

We subsequently investigated the mechanism underlying the CSN6-mediated upregulation of PHGDH. A mass spectrometry analysis of CSN6 immunoprecipitates revealed that DDX5 was a novel binding protein for CSN6 (Supplementary Table S2). Then, we found a positive correlation between DDX5 and PHGDH expression in colorectal cancer by analyzing TCGA database (Supplementary Fig. S6A). So, we proposed that DDX5 is the intermediator between CSN6 and PHGDH. Coimmunoprecipitation assays confirmed the interaction between CSN6 and DDX5 (Fig. 4A; Supplementary Fig. S6B). The mapping of the CSN6- and DDX5-binding regions revealed that the N-terminal region of CSN6 (aa 1–174), which was mostly occupied by the MPN domain, was necessary and sufficient for its interaction with DDX5 (Supplementary Fig. S6C), whereas the C-terminal region of DDX5 (aa 301–614) exhibited higher affinity binding to CSN6 (Supplementary Fig. S6D). We then sought to examine whether CSN6 regulated DDX5. Interestingly, DDX5 protein, but not DDX5 mRNA, was markedly decreased after the shRNA-mediated knockdown of CSN6 (Fig. 4B; Supplementary Fig. S6E), which suggests that CSN6 induces the posttranscriptional accumulation of DDX5. Also, overexpression of CSN6 can affect the steady-state expression level of DDX5 (Supplementary Fig. S6F). Rescue experiments further confirm these findings (Supplementary Fig. S6G). Furthermore, in colon cancer patients, high CSN6 expression is positively correlated with high levels of DDX5 and PHGDH protein (Supplementary Fig. S6H).

Figure 4.

CSN6 promotes PHGDH expression through DDX5. A, Lysates of DLD1 cells were immunoprecipitated with anti-CSN6 or anti-DDX5 antibodies and immunoblotted with the indicated antibodies. B, DLD1, SW480, HCT116, and HT-29 cells were treated with CSN6 shRNA or scramble. Cell lysates were immunoblotted with indicated antibodies. C, qRT-PCR and Western blot analysis of PHGDH expression in control and shDDX5-expressing colorectal cancer cells. For qRT-PCR, data, means ± SD. ***, P < 0.001 by one-way ANOVA. D, IncuCyte machine was used to evaluate the effects of nucleoside repletion on cancer cell proliferation upon DDX5 depletion and 5-FU treatment. Graph, means ± SD. **, P < 0.01; ***, P < 0.001 by the two-way ANOVA test. E, RIP assays for RT-PCR quantification of PHGDH binding to endogenous DDX5 in DLD1 cells. IgG served as a control. Data, means ± SD. **, P < 0.01 by one-way ANOVA. F, Exogenous RNA pulldown assays validate the specific interaction between DDX5 and PHGDH in DLD1 cells. G, Quantification of the half-life of PHGDH (% mRNA remaining) over time after actinomycin D (10 μg/mL) addition in shDDX5 with or without PLVX-DDX5–treated DLD1 cells. Graph, means ±SD. ***, P < 0.001 by the two-way ANOVA test. H, HCT116 cells treated with scramble or shCSN6 were transfected with Flag-DDX5 or control. Cell lysates were then immunoblotted with indicated antibodies. I, NTP (ATP, UTP, GTP, and CTP) levels in cells with or without PLVX-CSN6 or shDDX5 treatment. Data, means ±SD. **, P < 0.01; ***, P < 0.001 by one-way ANOVA. J, The CSN6–DDX5 axis promotes colorectal cancer tumorigenesis. DLD1 cells with or without shCSN6 or PLVX-DDX5 infection were subcutaneously injected into nude mice (n = 4). Tumors were isolated at the end of the experiments. Tumor volume is shown. **, P < 0.01; ***, P < 0.001 by the two-way ANOVA test. K, The CSN6–DDX5–PHGDH axis promotes colorectal cancer tumorigenesis. CSN6 knockdown through the injection of shRNA lentivirus strongly inhibits the subcutaneous tumor growth of nude mice, which could be rescued by stable overexpression of PLVX–DDX5. Furthermore, NCT-503 treatment could abrogate these effects. Tumor volume is shown. ***, P < 0.001 by the two-way ANOVA test. L,In vivo nude mice models showed the impacts of the CSN6–DDX5–PHGDH axis on nucleotide biogenesis. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 4.

CSN6 promotes PHGDH expression through DDX5. A, Lysates of DLD1 cells were immunoprecipitated with anti-CSN6 or anti-DDX5 antibodies and immunoblotted with the indicated antibodies. B, DLD1, SW480, HCT116, and HT-29 cells were treated with CSN6 shRNA or scramble. Cell lysates were immunoblotted with indicated antibodies. C, qRT-PCR and Western blot analysis of PHGDH expression in control and shDDX5-expressing colorectal cancer cells. For qRT-PCR, data, means ± SD. ***, P < 0.001 by one-way ANOVA. D, IncuCyte machine was used to evaluate the effects of nucleoside repletion on cancer cell proliferation upon DDX5 depletion and 5-FU treatment. Graph, means ± SD. **, P < 0.01; ***, P < 0.001 by the two-way ANOVA test. E, RIP assays for RT-PCR quantification of PHGDH binding to endogenous DDX5 in DLD1 cells. IgG served as a control. Data, means ± SD. **, P < 0.01 by one-way ANOVA. F, Exogenous RNA pulldown assays validate the specific interaction between DDX5 and PHGDH in DLD1 cells. G, Quantification of the half-life of PHGDH (% mRNA remaining) over time after actinomycin D (10 μg/mL) addition in shDDX5 with or without PLVX-DDX5–treated DLD1 cells. Graph, means ±SD. ***, P < 0.001 by the two-way ANOVA test. H, HCT116 cells treated with scramble or shCSN6 were transfected with Flag-DDX5 or control. Cell lysates were then immunoblotted with indicated antibodies. I, NTP (ATP, UTP, GTP, and CTP) levels in cells with or without PLVX-CSN6 or shDDX5 treatment. Data, means ±SD. **, P < 0.01; ***, P < 0.001 by one-way ANOVA. J, The CSN6–DDX5 axis promotes colorectal cancer tumorigenesis. DLD1 cells with or without shCSN6 or PLVX-DDX5 infection were subcutaneously injected into nude mice (n = 4). Tumors were isolated at the end of the experiments. Tumor volume is shown. **, P < 0.01; ***, P < 0.001 by the two-way ANOVA test. K, The CSN6–DDX5–PHGDH axis promotes colorectal cancer tumorigenesis. CSN6 knockdown through the injection of shRNA lentivirus strongly inhibits the subcutaneous tumor growth of nude mice, which could be rescued by stable overexpression of PLVX–DDX5. Furthermore, NCT-503 treatment could abrogate these effects. Tumor volume is shown. ***, P < 0.001 by the two-way ANOVA test. L,In vivo nude mice models showed the impacts of the CSN6–DDX5–PHGDH axis on nucleotide biogenesis. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

We next characterized the relevance of DDX5 and PHGDH. As expected, the knockdown of DDX5 decreased PHGDH expression (Fig. 4C; Supplementary Fig. S7A). Nucleoside repletion could partially restore the organoid formation of DDX5 knocking down (Supplementary Fig. S7B). In addition, DDX5 knockdown–induced chemosensitivity could be partially rescued by nucleoside supplementation in colorectal cancer cells (Fig. 4D). DDX5, one of RNA-binding proteins (RBP), has functions of RNA binding and splicing, rRNA processing, and mRNA degradation (30–32). RNA immunoprecipitation and RNA pulldown assays revealed that DDX5 can bind to PHGDH mRNA (Fig. 4E and F). Moreover, we treated cells with actinomycin D after the shRNA-mediated depletion of DDX5 and found that the knockdown of DDX5 increased PHGDH mRNA turnover (Fig. 4G; Supplementary Fig. S7C).

Notably, the enforced expression of DDX5 abolished the effect of CSN6 depletion on PHGDH dysregulation (Fig. 4H), suggesting CSN6-mediated PHGDH upregulation is DDX5 dependent. Analysis of metabolite abundances revealed that DDX5 inhibition by shRNA decreased the CSN6-induced nucleotide accumulation (Fig. 4I). Consistent with the PHGDH result, [U-13C] glucose analysis displayed knockdown of DDX5 hindered the CSN6 induced serine, glycine, and UMP biosynthesis (Supplementary Fig. S7D–S7F). The depletion of DDX5 clearly reduced the promotive effect of CSN6 on cell growth and colony formation ability (Supplementary Fig. S7G and S7H). In vivo mice models further demonstrated that the CSN6–DDX5 axis promoted colorectal cancer tumorigenesis (Fig. 4J and Supplementary Fig. S7I). Besides, we found that both tumor volume and weight in nude mice were significantly reduced after CSN6 knockdown, which could be rescued by stable overexpression of PLVX–DDX5. Furthermore, NCT-503 treatment could abrogate these effects (Fig. 4K; Supplementary Fig. S7J). Consistently, depletion of CSN6 significantly decreased nucleotide abundances and overexpression DDX5 could restore it, finally repressed by PHGDH inhibition (Fig. 4L). Taken together, these data indicate that the CSN6-mediated induction of PHGDH expression and metabolic reprogramming relies on DDX5.

CSN6 inhibits the ubiquitin–proteasome–mediated degradation of DDX5

Previous studies have shown that CSN6 mainly serves as a regulator of the degradation of cancer-related proteins through the ubiquitin–proteasome system (23, 33, 34). We found that CSN6 regulated DDX5 protein but not mRNA (Fig. 4B; Supplementary Fig. S6E) and thus postulated that CSN6 might elevate DDX5 expression by preventing the proteasome-mediated degradation of DDX5. To test this hypothesis, we treated cells with the proteasome inhibitor MG132 and lysosome inhibitor chloroquine (CQ). MG132 restored the shCSN6-induced reduction in DDX5 expression, whereas CQ had no effect (Fig. 5A; Supplementary Fig. S8A). Importantly, the DDX5 protein level gradually decreased in the control cells after cycloheximide (CHX) treatment in a time-dependent manner, whereas a more notable degradation of DDX5 was observed in the CSN6 knockdown cells (Fig. 5B). We also found that the amount of polyubiquitinated DDX5 was markedly decreased in cells with ectopic CSN6 expression (Fig. 5C; Supplementary Fig. S8B). Conversely, the polyubiquitination level of DDX5 was increased in shCSN6-expressing HCT116 cells (Fig. 5D). By screening different K sites, K48 was able to facilitate DDX5 ubiquitination, which usually leads to protein degradation (Supplementary Fig. S8C).

Figure 5.

CSN6 stabilizes DDX5 protein by decreasing β-Trcp–mediated ubiquitination. A, Cell lysates were prepared from DLD1 cells expressing CSN6 shRNA or scrambled shRNA that had been treated with or without MG132 for 6 hours and then immunoblotted with indicated antibodies. B, DDX5 turnover rate of CSN6 knocking down cells. DLD1 cells stably expressing CSN6 shRNA or a control shRNA were treated with cycloheximide (100 μg/mL) for indicated times. The turnover of DDX5 is indicated graphically. C, CSN6 decreased DDX5 ubiquitination levels. D, HCT116 cells infected with indicated CSN6-shRNA were treated with MG132 for 6 hours before harvesting. Cells were lysed in guanidine-HCl containing buffer, and cell lysates were then pulled down with nickel beads and immunoblotted with indicated antibodies. E, Cells were transfected with indicated plasmids or siRNA and immunoblotted with indicated antibodies. F, Cells were transfected with indicated plasmids. Cell lysates were immunoprecipitated with anti-Flag or anti-Myc beads and then immunoblotted with indicated antibodies. G, HCT116 cells were transfected with Flag-β-Trcp or vector and then treated with cycloheximide (100 μg/mL) for indicated times. The turnover of DDX5 is indicated graphically. H, β-Trcp enhanced the ubiquitination level of DDX5. I, HCT116 cells transfected with the siβ-Trcp were treated with MG132 for 6 hours before harvesting. Cells were lysed in guanidine-HCl containing buffer, and cell lysates were then pulled down with nickel beads and immunoblotted with indicated antibodies. J, HEK293T cells were cotransfected with indicated plasmids. Cells were then treated with MG132 6 hours prior to harvesting. Polyubiquitinated DDX5 was immunoprecipitated with nickel beads and immunoblotted with an anti-DDX5 antibody.

Figure 5.

CSN6 stabilizes DDX5 protein by decreasing β-Trcp–mediated ubiquitination. A, Cell lysates were prepared from DLD1 cells expressing CSN6 shRNA or scrambled shRNA that had been treated with or without MG132 for 6 hours and then immunoblotted with indicated antibodies. B, DDX5 turnover rate of CSN6 knocking down cells. DLD1 cells stably expressing CSN6 shRNA or a control shRNA were treated with cycloheximide (100 μg/mL) for indicated times. The turnover of DDX5 is indicated graphically. C, CSN6 decreased DDX5 ubiquitination levels. D, HCT116 cells infected with indicated CSN6-shRNA were treated with MG132 for 6 hours before harvesting. Cells were lysed in guanidine-HCl containing buffer, and cell lysates were then pulled down with nickel beads and immunoblotted with indicated antibodies. E, Cells were transfected with indicated plasmids or siRNA and immunoblotted with indicated antibodies. F, Cells were transfected with indicated plasmids. Cell lysates were immunoprecipitated with anti-Flag or anti-Myc beads and then immunoblotted with indicated antibodies. G, HCT116 cells were transfected with Flag-β-Trcp or vector and then treated with cycloheximide (100 μg/mL) for indicated times. The turnover of DDX5 is indicated graphically. H, β-Trcp enhanced the ubiquitination level of DDX5. I, HCT116 cells transfected with the siβ-Trcp were treated with MG132 for 6 hours before harvesting. Cells were lysed in guanidine-HCl containing buffer, and cell lysates were then pulled down with nickel beads and immunoblotted with indicated antibodies. J, HEK293T cells were cotransfected with indicated plasmids. Cells were then treated with MG132 6 hours prior to harvesting. Polyubiquitinated DDX5 was immunoprecipitated with nickel beads and immunoblotted with an anti-DDX5 antibody.

Close modal

We subsequently determined which E3 ligase was responsible for DDX5 degradation. After screening several possible E3 ubiquitin ligases that have been verified by the interaction with CSN6 and regulated by CSN6, we found that β-Trcp, an E3 ligase, could regulate DDX5 at the steady-state level (Fig. 5E; Supplementary Fig. S8D). Immunoprecipitation assays revealed that β-Trcp and DDX5 interacted with each other (Fig. 5F). In addition, the overexpression of β-Trcp increased the turnover rate of DDX5 and enhanced DDX5 ubiquitylation (Fig. 5G and H), whereas β-Trcp knockdown reduced the ubiquitination level of DDX5 (Fig. 5I). Also, these processes were indicated based on K48 link ubiquitination (Supplementary Fig. S8E).

To further confirm that CSN6 hinders β-Trcp–dependent DDX5 degradation, CSN6 was cotransfected with β-Trcp in HCT116 cells. Our results showed that β-Trcp abolished the effect of CSN6 on both the DDX5 protein and ubiquitination levels (Fig. 5J; Supplementary Fig. S8F). Together, these data support the notion that CSN6 stabilizes DDX5 protein by inhibiting ubiquitin–proteasome-mediated protein degradation and indicate that β-Trcp serves as the E3 ligase in this process.

The CSN6–DDX5–PHGDH axis promotes tumorigenesis and is associated with poor colorectal cancer patient prognosis

We then determined the in vivo functional contribution of CSN6 to DDX5 and PHGDH expression. Consistent with the above-described results, the expression levels of DDX5 protein, but not mRNA, and the mRNA levels of PHGDH were decreased after CSN6 depletion in xenograft tumors (Supplementary Fig. S9A and S9B). Lgr5-CreERT2;Csn6fl/fl mice also displayed decreased protein levels of DDX5 and PHGDH (Fig. 6A). IHC staining indicated that signal intensities of DDX5 and PHGDH were decreased in shCSN6 tumors (Fig. 6B).

Figure 6.

The CSN6–DDX5–PHGDH axis promotes tumorigenesis and is associated with poor colorectal cancer patient prognosis. A, Western blot analysis of the indicated proteins in colonic tumors from Csn6fl/fl and Lgr5-CreERT2;Csn6fl/fl mice (n = 3 per group). B, Paraffin-embedded tumor sections derived from the DLD1 xenografts were stained with CSN6, DDX5, and PHGDH antibodies. Scale bar, 100 μm. C, Representative IHC staining of CSN6, DDX5, and PHGDH in serial sections of colon cancer patient samples. Case 1 is representative of a colorectal cancer patient with high CSN6 expression, and case 2 is representative of a patient with low CSN6 expression. The samples were derived from 267 colon cancer cases. Scale bar, 100 μm. D, Percentages of samples showing the relationship among CSN6, DDX5, and PHGDH. P < 0.001 by a Chi-square test. E, Kaplan–Meier curves of the overall survival of patients with different expression levels of CSN6, DDX5, and PHGDH. *, P < 0.05; **, P < 0.01; ***, P < 0.001 by the log-rank test.

Figure 6.

The CSN6–DDX5–PHGDH axis promotes tumorigenesis and is associated with poor colorectal cancer patient prognosis. A, Western blot analysis of the indicated proteins in colonic tumors from Csn6fl/fl and Lgr5-CreERT2;Csn6fl/fl mice (n = 3 per group). B, Paraffin-embedded tumor sections derived from the DLD1 xenografts were stained with CSN6, DDX5, and PHGDH antibodies. Scale bar, 100 μm. C, Representative IHC staining of CSN6, DDX5, and PHGDH in serial sections of colon cancer patient samples. Case 1 is representative of a colorectal cancer patient with high CSN6 expression, and case 2 is representative of a patient with low CSN6 expression. The samples were derived from 267 colon cancer cases. Scale bar, 100 μm. D, Percentages of samples showing the relationship among CSN6, DDX5, and PHGDH. P < 0.001 by a Chi-square test. E, Kaplan–Meier curves of the overall survival of patients with different expression levels of CSN6, DDX5, and PHGDH. *, P < 0.05; **, P < 0.01; ***, P < 0.001 by the log-rank test.

Close modal

We further evaluated the clinical relevance of the above-described findings by analyzing the expression of all three proteins using a tissue microarray with 267 colorectal cancer tissue specimens (Supplementary Table S3). Correlation studies showed that these three proteins were positively correlated (Fig. 6C and D). In addition, a Kaplan–Meier analysis indicated that patients with high expression levels of CSN6, DDX5, and PHGDH tended to exhibit worse overall survival (Fig. 6E). Taken together, all of these results further demonstrate that CSN6, DDX5, and PHGDH are interconnected in colorectal cancer.

Butyrate is a CSN6 antagonist and improves the antitumor therapy efficacy

Butyrate, a short-chain fatty acid, is produced by the bacterial fermentation of undigested dietary fibers in the lumen (35, 36). Recent clinical studies have revealed that individuals with advanced colon cancer tend to have lower abundances of butyrate-producing bacteria and short-chain fatty acids (SCFA) than normal individuals (37). Through an analysis of patient samples, we determined that butanoate metabolism was negatively correlated with colorectal cancer relapse (Supplementary Fig. S10A). In agreement with this finding, we cultured PDOs in vitro and found that butyrate treatment significantly inhibited organoid growth (Supplementary Fig. S10B). Indeed, colorectal cancer cells incubated with butyrate were more sensitive to 5-FU treatment (Supplementary Fig. S10C). Because these results indicated that CSN6 plays a critical role in patient survival, we wondered whether butyrate might serve as a potential CSN6 antagonist for colorectal cancer treatment.

As shown in Supplementary Fig. S10D, butyrate treatment led to decreased expression of CSN6 and its downstream targets, and these effects included reductions in DDX5 protein levels and PHGDH mRNA and protein levels. Interestingly, another HDAC inhibitor, SAHA, had a similar effect (Supplementary Fig. S10E), suggesting that butyrate may function on CSN6 depending on its HDAC inhibitor activity. Consistent with the effects of CSN6 depletion, we also revealed that the butyrate treatment decreased NTP levels (Supplementary Fig. S10F). A tracing analysis using uniformly labeled [U-13C] glucose further demonstrated that butyrate treatment reduced the incorporation of 13C from [U-13C] glucose into serine, glycine, and car-aspartate (Supplementary Fig. S10G–S10I). To further confirm this hypothesis, we incubated cells with butyrate and then reexpressed CSN6. Notably, CSN6 partly rescued the butyrate-induced inhibition of DDX5 and PHGDH expression (Supplementary Fig. S10J).

To investigate the therapeutic potential of butyrate, we implanted fresh colorectal cancer patient-derived tumor samples into immunodeficient mice (Supplementary Table S4). CSN6 expression levels were detected in different tumor masses (Supplementary Fig. S10K). Interestingly, butyrate treatment in the established CSN6-high PDX tumors (cases 1 and 2) hindered tumor progression. By contrast, administration of butyrate had no significant influence on the CSN6-low PDX tumor (cases 3 and 4) growth (Supplementary Figs. S10 L and S11A). IHC analysis demonstrated that butyrate impaired CSN6, DDX5, and PHGDH expression in CSN6-high tumors, while it had no significant impact in CSN6-low tumors (Supplementary Fig. S11B). We next assessed whether butyrate could improve chemotherapy. Tumor-bearing mice were treated with 5-FU alone or together with butyrate. Combination therapy with 5-FU and butyrate displayed more promising tumor regression, including the lowest tumor volume and tumor weight (Supplementary Fig. S11C and S11D).

Collectively, our data illustrated that butyrate, as a potential CSN6 antagonist, in combination with 5-FU shows increased tumor-suppressive effects, indicating possible implications for future cancer therapy.

Here, we report that CSN6 induces the expression of PHGDH, thereby potentiating nucleotide synthesis and colorectal cancer recurrence. Chemotherapy is the standard strategy used for patients with colorectal cancer, but drug resistance remains a major hurdle to treatment (4). In agreement with a previous study (38), we noted that nucleotide metabolism promotes chemoresistance, as demonstrated based on clinical data. Metabolic reprogramming is one of the defining hallmarks of cancer responsible for providing high energy levels required to support cancer cell proliferation and survival (14, 39). Metabolic changes induced by oncogenic drivers of cancer contribute to abnormal cellular bioenergetics and are attractive targets for cancer therapy (39).

Nucleotide metabolic processes are essential for DNA and RNA synthesis, and analogues of nucleotide precursors have been identified as an important class of antitumor agents. Nevertheless, the identification of additional targets for therapeutic applications remains a challenge, and thus, a more in-depth exploration of the mechanisms underlying the regulation of the nucleotide synthesis pathway is needed. Recent studies have revealed that PHGDH, which catalyzes the vital step in de novo serine biosynthesis, is required for the maintenance of purine and pyrimidine synthesis by supporting central carbon metabolism (11, 18). In addition, the knockdown of PHGDH also impairs pyrimidine metabolism by hindering DHODH expression and activity (17). We confirmed that the depletion of CSN6 decreased PHGDH expression and thereby led to alterations in both de novo purine and pyrimidine biosynthesis, and these findings provide a foundation for the exploration of additional regulators of nucleotide synthesis.

It has been reported that CSN6 is involved in a wide range of regulatory processes, but most of the studies were performed in vitro. Here, we first generated CSN6 intestinal conditional knockout (Csn6CKO) mice and combined uniformly carbon13-labeled glucose ([U-13C] glucose), demonstrating that the knockdown of CSN6 altered isotopomer distributions of the nucleotides in vivo. Our results provide insights into the consequences of abnormal CSN6 expression on nucleotide metabolism. Because COP9 signalosome complex contains eight subunits, it would be interesting to determine whether other subunits have this capability in the future.

DDX5 is an ATP-dependent RNA helicase that is frequently overexpressed in various tumors and correlated with poor prognosis in cancer patients (40–42). Here, we uncovered that DDX5 protein can be polyubiquitinated and degraded by the E3 ligase β-Trcp and that this ligase can be regulated by CSN6. As a multifunctional protein, DDX5 participates in a number of biological processes, including RNA binding (43), ribosome biogenesis (44), and can act as a coregulator of multiple cancer-associated transcription factors (41). Besides, DDX5 could interact with NMD/SMD components to trigger mRNA degradation (45, 46). In macrophages, DDX5 has been reported to promote MSR1 expression by stabilizing MSR1 mRNA (30). Recent studies also revealed that DDX5 could bind to and stabilize CCNB2 mRNA (47). Moreover, DDX5 could stabilize mRNA via interactions with other RBPs, such as ILF3 (47). In accordance with these findings, we demonstrated that DDX5 bound to PHGDH mRNA and stimulated its expression by suppressing mRNA degradation in colorectal cancer.

Butyrate, a by-product of the bacterial fermentation of undigested dietary fibers in the intestine, has been proposed as an antitumor agent (48). Here, we confirmed that butyrate treatment can enhance the sensitivity of colorectal cancer cells to 5-FU. A metabolite analysis identified butyrate as an inhibitor of CSN6 that can lead to the disruption of DDX5 and PHGDH expression and nucleotide reprogramming. Butyrate could modify cellular functions either by preventing histone deacetylase activity and thereby affecting gene transcription or through the activation of G-protein–coupled receptors such as GPR109a (49, 50). We found that another HDAC inhibitor, SAHA, could function similarly to butyrate, suggesting that butyrate negatively regulated CSN6 depending on its inhibition of histone deacetylase activity. Further studies are needed to elucidate the mechanisms underlying the butyrate-mediated inhibition of CSN6 and the resulting obstruction of nucleotide metabolism. Moreover, given that butyrate treatment enhanced 5-FU chemosensitivity, the therapeutic effect of butyrate needs to be further explored for colorectal cancer patients' therapy in the clinic.

Lekun Fang reports grants from National Key R&D Program of China (2021YFF0702600), the National Natural Science Foundation of China (82222056 and 82111530099), the Guangdong Special Young Talent Plan of Scientific and Technological Innovation (2019TQ05Y510), the the Natural Science Foundation of Guangdong (2022A1515012316), the Guangdong International Joint Research Program (2020A0505100027), and the National Key Clinical Discipline during the conduct of the study. No disclosures were reported by the other authors.

S. Zou: Data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. B. Qin: Investigation, methodology. Z. Yang: Software, methodology. W. Wang: Methodology. J. Zhang: Methodology. Y. Zhang: Methodology. M. Meng: Methodology. J. Feng: Methodology. Y. Xie: Methodology. L. Fang: Methodology. L. Xiao: Methodology. P. Zhang: Methodology. X. Meng: Methodology. H. Choi: Methodology. W. Wen: Methodology. Q. Pan: Software. B. Ghesquière: Software, methodology. P. Lan: Resources, supervision. M.-H. Lee: Conceptualization, resources, supervision, project administration. L. Fang: Conceptualization, resources, data curation, software, supervision, funding acquisition, writing–original draft, project administration, writing–review and editing.

This research was supported by the National Key R&D Program of China (2021YFF0702600 to L.K. Fang), the National Natural Science Foundation of China (82222056 and 82111530099 to L.K. Fang), Guangdong Special Young Talent Plan of Scientific and Technological Innovation (2019TQ05Y510 to L.K. Fang), the Natural Science Foundation of Guangdong (2022A1515012316 to L.K. Fang), the Guangdong International Joint Research Program (2020A0505100027 to L.K. Fang), and National Key Clinical Discipline (L.K. Fang).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).

1.
Shen
Y
,
Wang
X
,
Lu
J
,
Salfenmoser
M
,
Wirsik
NM
,
Schleussner
N
, et al
.
Reduction of liver metastasis stiffness improves response to bevacizumab in metastatic colorectal cancer
.
Cancer Cell
2020
;
37
:
800
17
.
2.
Bray
F
,
Ferlay
J
,
Soerjomataram
I
,
Siegel
RL
,
Torre
LA
,
Jemal
A
.
Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries
.
CA Cancer J Clin
2018
;
68
:
394
424
.
3.
Fang
L
,
Yang
Z
,
Zhang
M
,
Meng
M
,
Feng
J
,
Chen
C
.
Clinical characteristics and survival analysis of colorectal cancer in China: a retrospective cohort study with 13,328 patients from southern China
.
Gastroenterol Rep
2021
;
9
:
571
82
.
4.
Schmoll
HJ
,
Van Cutsem
E
,
Stein
A
,
Valentini
V
,
Glimelius
B
,
Haustermans
K
, et al
.
ESMO consensus guidelines for management of patients with colon and rectal cancer: a personalized approach to clinical decision making
.
Ann Oncol
2012
;
23
:
2479
516
.
5.
Fotheringham
S
,
Mozolowski
GA
,
Murray
EMA
,
Kerr
DJ
.
Challenges and solutions in patient treatment strategies for stage II colon cancer
.
Gastroenterol Rep
2019
;
7
:
151
61
.
6.
Zhou
Y
,
Wan
G
,
Spizzo
R
,
Ivan
C
,
Mathur
R
,
Hu
X
, et al
.
miR-203 induces oxaliplatin resistance in colorectal cancer cells by negatively regulating ATM kinase
.
Mol Oncol
2014
;
8
:
83
92
.
7.
Rasmussen
MH
,
Lyskjaer
I
,
Jersie-Christensen
RR
,
Tarpgaard
LS
,
Primdal-Bengtson
B
,
Nielsen
MM
, et al
.
miR-625–3p regulates oxaliplatin resistance by targeting MAP2K6-p38 signalling in human colorectal adenocarcinoma cells
.
Nat Commun
2016
;
7
:
12436
.
8.
Reina-Campos
M
,
Linares
JF
,
Duran
A
,
Cordes
T
,
L'Hermitte
A
,
Badur
MG
, et al
.
Increased serine and one-carbon pathway metabolism by PKClambda/iota deficiency promotes neuroendocrine prostate cancer
.
Cancer Cell
2019
;
35
:
385
400
.
9.
Ogiwara
H
,
Takahashi
K
,
Sasaki
M
,
Kuroda
T
,
Yoshida
H
,
Watanabe
R
, et al
.
Targeting the vulnerability of glutathione metabolism in ARID1A-deficient cancers
.
Cancer Cell
2019
;
35
:
177
90
.
10.
Augert
A
,
Mathsyaraja
H
,
Ibrahim
AH
,
Freie
B
,
Geuenich
MJ
,
Cheng
P-F
, et al
.
MAX functions as a tumor suppressor and rewires metabolism in small cell lung cancer
.
Cancer Cell
2020
;
38
:
97
114
.
11.
Newman
AC
,
Maddocks
ODK
.
One-carbon metabolism in cancer
.
Br J Cancer
2017
;
116
:
1499
504
.
12.
Ducker
GS
,
Chen
L
,
Morscher
RJ
,
Ghergurovich
JM
,
Esposito
M
,
Teng
X
, et al
.
Reversal of cytosolic one-carbon flux compensates for loss of the mitochondrial folate pathway
.
Cell Metab
2016
;
23
:
1140
53
.
13.
Halbrook
CJ
,
Pontious
C
,
Kovalenko
I
,
Lapienyte
L
,
Dreyer
S
,
Lee
HJ
, et al
.
Macrophage-released pyrimidines inhibit gemcitabine therapy in pancreatic cancer
.
Cell Metab
2019
;
29
:
1390
9
.
14.
Li
K
,
Wu
J-l
Qin
B
,
Fan
Z
,
Tang
Q
,
Lu
W
, et al
.
ILF3 is a substrate of SPOP for regulating serine biosynthesis in colorectal cancer
.
Cell Res
2020
;
30
:
163
-
178
.
15.
Intuyod
K
,
Saavedra-Garcia
P
,
Zona
S
,
Lai
CF
,
Jiramongkol
Y
,
Vaeteewoottacharn
K
, et al
.
FOXM1 modulates 5-fluorouracil sensitivity in cholangiocarcinoma through thymidylate synthase (TYMS): implications of FOXM1-TYMS axis uncoupling in 5-FU resistance
.
Cell Death Dis
2018
;
9
:
1185
.
16.
Yao
J
,
Huang
A
,
Zheng
X
,
Liu
T
,
Lin
Z
,
Zhang
S
, et al
.
53BP1 loss induces chemoresistance of colorectal cancer cells to 5-fluorouracil by inhibiting the ATM-CHK2-P53 pathway
.
J Cancer Res Clin Oncol
2017
;
143
:
419
31
.
17.
Vandekeere
S
,
Dubois
C
,
Kalucka
J
,
Sullivan
MR
,
Garcia-Caballero
M
,
Goveia
J
, et al
.
Serine synthesis via PHGDH is essential for heme production in endothelial cells
.
Cell Metab
2018
;
28
:
573
87
.
18.
Reid
MA
,
Allen
AE
,
Liu
S
,
Liberti
MV
,
Liu
P
,
Liu
X
, et al
.
Serine synthesis through PHGDH coordinates nucleotide levels by maintaining central carbon metabolism
.
Nat Commun
2018
;
9
:
5442
.
19.
DeNicola
GM
,
Chen
PH
,
Mullarky
E
,
Sudderth
JA
,
Hu
Z
,
Wu
D
, et al
.
NRF2 regulates serine biosynthesis in non-small cell lung cancer
.
Nat Genet
2015
;
47
:
1475
81
.
20.
Chen
J
,
Shin
JH
,
Zhao
R
,
Phan
L
,
Wang
H
,
Xue
Y
, et al
.
CSN6 drives carcinogenesis by positively regulating Myc stability
.
Nat Commun
2014
;
5
:
5384
.
21.
Fang
L
,
Lu
W
,
Choi
HH
,
Yeung
SC
,
Tung
JY
,
Hsiao
CD
, et al
.
ERK2-dependent phosphorylation of CSN6 is critical in colorectal cancer development
.
Cancer Cell
2015
;
28
:
183
97
.
22.
Liang
Y
,
Lyon
RC
,
Pellman
J
,
Bradford
WH
,
Lange
S
,
Bogomolovas
J
, et al
.
Desmosomal COP9 regulates proteome degradation in arrhythmogenic right ventricular dysplasia/cardiomyopathy
.
J Clin Invest
2021
;
131
:
e137689
.
23.
Hou
J
,
Deng
Q
,
Zhou
J
,
Zou
J
,
Zhang
Y
,
Tan
P
, et al
.
CSN6 controls the proliferation and metastasis of glioblastoma by CHIP-mediated degradation of EGFR
.
Oncogene
2017
;
36
:
1134
44
.
24.
Yuan
M
,
Kremer
DM
,
Huang
H
,
Breitkopf
SB
,
Ben-Sahra
I
,
Manning
BD
, et al
.
Ex vivo and in vivo stable isotope labelling of central carbon metabolism and related pathways with analysis by LC–MS
.
Nat Protoc
2019
;
14
:
313
30
.
25.
Ou
J
,
Peng
Y
,
Yang
W
,
Zhang
Y
,
Hao
J
,
Li
F
, et al
.
ABHD5 blunts the sensitivity of colorectal cancer to fluorouracil via promoting autophagic uracil yield
.
Nat Commun
2019
;
10
:
1078
.
26.
Vodenkova
S
,
Buchler
T
,
Cervena
K
,
Veskrnova
V
,
Vodicka
P
,
Vymetalkova
V
.
5-fluorouracil and other fluoropyrimidines in colorectal cancer: Past, present and future
.
Pharmacol Ther
2019
:
107447
.
27.
Qian
Z
,
Li
M
,
Wang
R
,
Xiao
Q
,
Wang
J
,
Li
M
, et al
.
Knockdown of CABYR-a/b increases chemosensitivity of human non-small cell lung cancer cells through inactivation of Akt
.
Mol Cancer Res
2014
;
12
:
335
47
.
28.
Mathur
D
,
Stratikopoulos
E
,
Ozturk
S
,
Steinbach
N
,
Pegno
S
,
Schoenfeld
S
, et al
.
PTEN regulates glutamine flux to pyrimidine synthesis and sensitivity to dihydroorotate dehydrogenase inhibition
.
Cancer Discov
2017
;
7
:
380
90
.
29.
Pacold
ME
,
Brimacombe
KR
,
Chan
SH
,
Rohde
JM
,
Lewis
CA
,
Swier
LJ
, et al
.
A PHGDH inhibitor reveals coordination of serine synthesis and one-carbon unit fate
.
Nat Chem Biol
2016
;
12
:
452
8
.
30.
Zhao
W
,
Wang
Z
,
Sun
Z
,
He
Y
,
Jian
D
,
Hu
X
, et al
.
RNA helicase DDX5 participates in oxLDL-induced macrophage scavenger receptor 1 expression by suppressing mRNA degradation
.
Exp Cell Res
2018
;
366
:
114
20
.
31.
Tago
K
,
Funakoshi-Tago
M
,
Itoh
H
,
Furukawa
Y
,
Kikuchi
J
,
Kato
T
, et al
.
Arf tumor suppressor disrupts the oncogenic positive feedback loop including c-Myc and DDX5
.
Oncogene
2015
;
34
:
314
22
.
32.
Li
H
,
Lai
P
,
Jia
J
,
Song
Y
,
Xia
Q
,
Huang
K
, et al
.
RNA helicase DDX5 inhibits reprogramming to pluripotency by miRNA-based repression of RYBP and its PRC1-dependent and -independent functions
.
Cell Stem Cell
2017
;
20
:
462
77
.e6.
33.
Zhao
R
,
Yeung
SC
,
Chen
J
,
Iwakuma
T
,
Su
CH
,
Chen
B
, et al
.
Subunit 6 of the COP9 signalosome promotes tumorigenesis in mice through stabilization of MDM2 and is upregulated in human cancers
.
J Clin Invest
2011
;
121
:
851
65
.
34.
Xue
Y
,
Chen
J
,
Choi
HH
,
Phan
L
,
Chou
PC
,
Zhao
R
, et al
.
HER2-Akt signaling in regulating COP9 signalsome subunit 6 and p53
.
Cell Cycle
2012
;
11
:
4181
90
.
35.
Schulthess
J
,
Pandey
S
,
Capitani
M
,
Rue-Albrecht
KC
,
Arnold
I
,
Franchini
F
, et al
.
The short chain fatty acid butyrate imprints an antimicrobial program in macrophages
.
Immunity
2019
;
50
:
432
45
.
36.
Singh
N
,
Gurav
A
,
Sivaprakasam
S
,
Brady
E
,
Padia
R
,
Shi
H
, et al
.
Activation of Gpr109a, receptor for niacin and the commensal metabolite butyrate, suppresses colonic inflammation and carcinogenesis
.
Immunity
2014
;
40
:
128
39
.
37.
Chen
HM
,
Yu
YN
,
Wang
JL
,
Lin
YW
,
Kong
X
,
Yang
CQ
, et al
.
Decreased dietary fiber intake and structural alteration of gut microbiota in patients with advanced colorectal adenoma
.
Am J Clin Nutr
2013
;
97
:
1044
52
.
38.
Parker
WB
.
Enzymology of purine and pyrimidine antimetabolites used in the treatment of cancer
.
Chem Rev
2009
;
109
:
2880
93
.
39.
Martinez-Outschoorn
UE
,
Peiris-Pages
M
,
Pestell
RG
,
Sotgia
F
,
Lisanti
MP
.
Cancer metabolism: a therapeutic perspective
.
Nat Rev Clin Oncol
2017
;
14
:
11
31
.
40.
Sarkar
M
,
Khare
V
,
Guturi
KK
,
Das
N
,
Ghosh
MK
.
The DEAD box protein p68: a crucial regulator of AKT/FOXO3a signaling axis in oncogenesis
.
Oncogene
2015
;
34
:
5843
56
.
41.
Wang
Z
,
Luo
Z
,
Zhou
L
,
Li
X
,
Jiang
T
,
Fu
E
.
DDX5 promotes proliferation and tumorigenesis of non-small-cell lung cancer cells by activating beta-catenin signaling pathway
.
Cancer Sci
2015
;
106
:
1303
12
.
42.
Zhang
H
,
Zhang
Y
,
Zhu
X
,
Chen
C
,
Zhang
C
,
Xia
Y
, et al
.
DEAD box protein 5 inhibits liver tumorigenesis by stimulating autophagy via interaction with p62/SQSTM1
.
Hepatology
2019
;
69
:
1046
63
.
43.
Huang
Y
,
Liu
ZR
.
The ATPase, RNA unwinding, and RNA binding activities of recombinant p68 RNA helicase
.
J Biol Chem
2002
;
277
:
12810
5
.
44.
Saporita
AJ
,
Chang
HC
,
Winkeler
CL
,
Apicelli
AJ
,
Kladney
RD
,
Wang
J
, et al
.
RNA helicase DDX5 is a p53-independent target of ARF that participates in ribosome biogenesis
.
Cancer Res
2011
;
71
:
6708
17
.
45.
Geissler
V
,
Altmeyer
S
,
Stein
B
,
Uhlmann-Schiffler
H
,
Stahl
H
.
The RNA helicase Ddx5/p68 binds to hUpf3 and enhances NMD of Ddx17/p72 and Smg5 mRNA
.
Nucleic Acids Res
2013
;
41
:
7875
88
.
46.
Moon
BS
,
Bai
J
,
Cai
M
,
Liu
C
,
Shi
J
,
Lu
W
.
Kruppel-like factor 4-dependent Staufen1-mediated mRNA decay regulates cortical neurogenesis
.
Nat Commun
2018
;
9
:
401
.
47.
Legrand
JMD
,
Chan
AL
,
La
HM
,
Rossello
FJ
,
Anko
ML
,
Fuller-Pace
FV
, et al
.
DDX5 plays essential transcriptional and post-transcriptional roles in the maintenance and function of spermatogonia
.
Nat Commun
2019
;
10
:
2278
.
48.
He
Y
,
Fu
L
,
Li
Y
,
Wang
W
,
Gong
M
,
Zhang
J
, et al
.
Gut microbial metabolites facilitate anticancer therapy efficacy by modulating cytotoxic CD8(+) T cell immunity
.
Cell Metab
2021
;
33
:
988
-
1000
.
49.
Sivaprakasam
S
,
Prasad
PD
,
Singh
N
.
Benefits of short-chain fatty acids and their receptors in inflammation and carcinogenesis
.
Pharmacol Ther
2016
;
164
:
144
51
.
50.
Thangaraju
M
,
Cresci
GA
,
Liu
K
,
Ananth
S
,
Gnanaprakasam
JP
,
Browning
DD
, et al
.
GPR109A is a G-protein-coupled receptor for the bacterial fermentation product butyrate and functions as a tumor suppressor in colon
.
Cancer Res
2009
;
69
:
2826
32
.