REEP6 promotes colorectal cancer glycolysis and tumorigenesis through PRMT5-mediated PGAM1 arginine methylation.

1
Introduction
Colorectal cancer (CRC) ranks third in incidence and second in mortality globally, making it the most frequent digestive tract malignancy worldwide1. According to Global Cancer Statistics 2022, more than 1.9 million new cases and 904,000 deaths were estimated to occur in 20222. Metabolic reprogramming has been widely recognized as an emerging biological hallmark of cancer, including CRC3. Numerous studies have been dedicated to developing such cancer therapeutic agents that could target cancer metabolism, and some were advancing to clinical trials4, 5, 6. However, clinical trials that target tumor metabolism for cancer therapy have sometimes shown unsatisfactory efficacy6. For this purpose, it is urgent to illustrate the exact molecular mechanisms that drive CRC metabolic reprogramming and therefore promote CRC tumorigenesis.
Protein arginine methyltransferases (PRMTs) are responsible for catalyzing the transfer of methyl group from S-adenosylmethionine (SAM) to the guanidine nitrogen atoms of specific arginine residues in target proteins, which results in the formation of three types of methylated arginine residue: monomethylarginine (MMA), asymmetric dimethylarginine (ADMA), and symmetric dimethylarginine (SDMA)7. As a type II arginine methyltransferase, protein arginine methyltransferase 5 (PRMT5) catalyzes the synthesis of SDMA on protein substrates8. Recently, PRMT5 has been identified as an oncoprotein in CRC progression due to its PRMT activity that regulate enzymatic activity, subcellular localization, and protein stability of target proteins9,10. For example, Liu et al. reported that PRMT5 methylated SMAD4 at R361, inducing SMAD complex formation and nuclear import9. PRMT5 directly catalyzes the SDMA of ALKBH5 at the R316 residue, which promotes TRIM28-mediated ALKBH5 ubiquitination and degradation10.
Aerobic glycolysis also termed the Warburg effect, means that even in a sufficient oxygen supply, cancer cells preferentially consume glucose via the glycolysis pathway rather than oxidative phosphorylation11,12. It has been reported that aerobic glycolysis could be fueled by the upregulation of many glycolytic enzymes and transporters in cancer cells13,14. Among these, phosphoglycerate mutase 1 (PGAM1), a pivotal metabolic enzyme in the glycolysis pathway, has been a research hotspot for a long time15. The interconversion between 3-phosphoglycerate (3-PG) and 2-phosphoglycerate (2-PG) in glycolysis is critically catalyzed by PGAM116. The aberrantly upregulated protein expression or overactivated enzymatic activity of PGAM1 was observed in various human cancers, including hepatocellular carcinoma (HCC) and non-small-cell lung cancer (NSCLC)17, 18, 19, 20. Increasing studies have confirmed that multiple post-translational modifications (PTMs) of PGAM1 play a critical role in regulating its enzymatic activity21, 22, 23. For example, Tyr26 phosphorylation of PGAM1 enhances and stabilizes H11 phosphorylation, and therefore activates PGAM121. Histone acetyltransferase 1 (HAT1) succinylates PGAM1 at Lys99, increasing its enzymatic activity and promoting proliferation and tumorigenesis in many cancers22. William et al.23 reported that under glucose restriction, NAD+-dependent protein deacetylase Sirt1 deacetylates PGAM1 and then attenuates its catalytic efficiency. However, the identification and the function of other PTMs on PGAM1 remain largely elusive.
Receptor expression enhancing protein 6 (REEP6) belongs to the REEPs family, which was initially identified for its capacity to facilitate the plasma membrane expression of olfactory receptors, G protein-coupled receptors, and taste receptors24. Recent studies have revealed some novel roles of REEP6 in cancer cell biology25, 26, 27. A previous study reported that REEP5 and REEP6 interact with CXCR1 and then enhance IL8-stimulated endocytosis of CXCR1 which mediates cancer cell responses in lung cancer25. Xia and his colleagues26 found that REEP6 suppressed ferroptosis in OSCC cells by maintaining endoplasmic reticulum homeostasis, contributing to OSCC progression. But even so, the role of REEP6 in cancer, especially in CRC, still requires further investigation.
In this report, we found that REEP6 is extremely upregulated in CRC tissues and promotes CRC tumorigenesis through enhancing glycolytic flux. Further analysis revealed that REEP6, acting as a molecular scaffold, consolidates the interaction between PRMT5 and PGAM1. Our results showed that the REEP6-PRMT5-PGAM1 heterotrimeric complex could enhance PRMT5-mediated SDMA modification of PGAM1 at the R40 residue, which dramatically increased PGAM1 enzymatic activity. More importantly, combined treatment with specific shRNA and inhibitors shows synergistic inhibitory efficacy in tumorigenesis and proliferation of CRC cell lines, PDO, and xenograft models. Taken together, our data support that targeting the REEP6–PRMT5–PGAM1 complex may be a promising therapeutic strategy for CRC.

2
Materials and methods
2.1
Human specimens and cell culture
All samples were collected from CRC patients with signed informed consent at our hospital. Formalin-fixed paraffin-embedded tissue microarrays (TMA) were constructed by Servicebio (Wuhan, China) including 48 cases of paired normal mucosa, CRC primary, and liver metastatic tumors, and 80 cases of CRC tissues and distal normal mucosa. The Ethics Committee of our hospital approved this study.
The cell lines DLD-1, RKO, and human embryonic kidney 293T (HEK-293T) were purchased from the Cell Bank of Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). All cells were cultured in the recommended medium (Gibco, USA) supplemented with 10% fetal bovine serum (FBS; Gibco, USA) and 1% penicillin/streptomycin (Gibco, USA).

2.2
Quantitative real-time PCR (qRT-PCR)
TRIzol reagent (Invitrogen, USA) was applied to extract the total RNA of cell lines and tissues according to the manufacturer's protocols. Total RNA was reversely transcribed to cDNA using HiScript RT Mix (Vazyme, Jiangsu, China). Then, the expression of mRNA was analyzed using the SYBR Premix Ex Taq Kit (TaKaRa Biotechnology, Dalian, China) with the Applied Biosystems 7500 Sequence Detection System using relative primers for amplification. The relative gene expression was determined by the StepOne platform (version 2.3), while the primer sequences for qRT-PCR can be referenced in Supporting Information Table S1.

2.3
Immunofluorescence (IF)
Cells were cultured in the confocal dish for 48 h and then washed with PBS three times. The cells were fixed with an immunostaining fixative overnight at 4 °C. Next, cells were blocked with blocking buffer and then incubated with primary antibody overnight at 4 °C. The cells were incubated with a secondary antibody for 60 min in the dark and then counterstained with 4′,6-diamidino-2-phenylindole (DAPI) as a nuclear indicator and imaged with a confocal laser-scanning microscope. The primary antibodies used are listed in Supporting Information Table S2.

2.4
Cell proliferation assays
The Cell Counting Kit-8 (Beyotime Biotechnology, Shanghai, China) and colony formation assays were performed as described previously28.

2.5
Extracellular acidification rate (ECAR)
Seahorse XF Glycolysis Stress Test Kit (Agilent Technologies) was used for ECAR measurement by Seahorse Bioscience XF96 extracellular flux analyzer (Seahorse Bioscience). Briefly, CRC cells were seeded into the Seahorse plates and incubated overnight at 37 °C. The next day, the monolayer cells were washed with Seahorse buffer, and baseline concentration was measured. Then, glucose, oligomycin, and 2-deoxy-d-glucose were sequentially added for ECAR measurement. Data was analyzed by Seahorse XF96 Wave software.

2.6
Glucose uptake assay and lactate production assay
Quantification of glucose uptake and lactate production was performed with the Glucose Uptake Assay Kit (ab136955, Abcam) and l-Lactate Assay Kit (ab65331, Abcam), in accordance with the manufacturer's protocols as previously described29.

2.7
2-Phosphoglycerate (2-PG) and 3-phosphoglycerate (3-PG) assay
The measurement of 2-PG and 3-PG was conducted as previously described30.

2.8
Animal models
The subcutaneous xenograft model was developed using 6-week-old male BALB/c nude mice. The mice were injected with 1 × 106 DLD-1 cells (stably transfected with the indicated lentiviruses) into the bilateral groin separately. Xenograft size was recorded at 3-day intervals, and the volume was derived from equation L × W2/2 (L: length, W: width). After 21 days, the animals were humanely euthanized to retrieve xenografts for statistical analysis. For combined treatment, the mice were intraperitoneally administered with PGMI-004A (MCE, 100 mg/kg) and/or GSK3326595 (MCE, 100 mg/kg) every 3 days. And the mice were euthanized on Day 30.
For the AOM/DSS-induced CRC model31,32, four-week-old male BALB/c mice were injected into the tail veins with different AAVs as indicated. All mice were once administered azoxymethane (AOM, 10 mg/kg body weight, Sigma–Aldrich) intraperitoneally, then received 2.5% dextran sodium sulfate (DSS, MP) in drinking water for 7 days (1 week after azoxymethane injection), subsequently given normal water for 14 days. The DSS-normal water cycle was repeated in three rounds. Mice were euthanized at the end of the study to harvest their colon. All animal experiments were approved by the Committee on the Ethics of Animal Experiments of our university.

2.9
Statistical analysis
Statistical analyses were performed using GraphPad Prism 10.0 (GraphPad Software, San Diego, CA, USA) and IBM SPSS Statistics 26.0 (IBM Corp., Armonk, NY, USA). All experiments were independently repeated at least three times. Quantitative data are presented as mean ± standard deviation (SD). Statistical significance was determined using two-tailed Student's t-tests, Chi-square test, and Kaplan-Meier log-rank test. P < 0.05 was considered statistically significant.

3
Results
3.1
REEP6 is significantly overexpressed in CRC tissues
To elucidate the expression pattern of REEP6 in CRC tissues, we analyzed the mRNA level of REEP6 in The Cancer Genome Atlas (TCGA) and GEO (GSE156451 and GSE50760) databases. Compared to the normal tissues, REEP6 was significantly elevated in CRC tissues (Fig. 1A–C), moreover, CRC liver metastasis (CRLM) tissues had a more increased expression (Fig. 1C). Clinical Proteomic Tumor Analysis Consortium (CPTAC) datasets proved that REEP6 was highly expressed in colon cancer tissues than noncancerous tissues (Fig. 1D). The above result was further validated by our cohort 1 incorporating 200 cases of CRC patients (Fig. 1E). Statistical analyses of clinicopathological parameters in cohort 1 exhibited that REEP6 expression was positively correlated with T classification, tumor size, lymph node metastasis and TNM stage (Fig. 1F). Tissue microarray IHC staining of 48 CRLM patients in cohorts 2 and 80 cases of CRC patients (referred as cohorts 3) demonstrated that the protein levels of REEP6 were dramatically upregulated in CRLM and CRC tissues compared to adjacent normal tissues (Fig. 1G–J). Kaplan-Meier survival curve of cohort 3 showed that CRC patients with increased REEP6 expression had a worse overall survival (Fig. 1K). Furthermore, the analysis of the relation between REEP6 expression and clinicopathological factors in cohort 3 revealed that elevated REEP6 expression was positively associated with tumor size, T classification, TNM stage, lymph node metastasis, and distant metastasis (Supporting Information Table S3). Western blotting assays were conducted in 18 cases of CRLM patients, namely cohort 4. It was also observed that REEP6 expression was significantly enhanced in CRLM tissues compared to primary CRC tissues, and paired para-carcinoma tissues had the lowest REEP6 expression (Fig. 1L). Taken together, these results suggest that REEP6 is increased and correlated with a poor prognosis in CRC.

3.2
REEP6 facilitates glycolysis and tumorigenesis in CRC
To explore the dominant biological processes of CRC progression driven by REEP6, gene set enrichment analysis (GSEA) was performed in the TCGA database. The results exhibited a marked enrichment of the glycolysis pathway (Fig. 2A and Supporting Information Fig. S1A). Compared with NCM460, REEP6 was significantly upregulated in CRC cell lines, and moreover, DLD-1 and RKO cells had the highest expression levels (Fig. S1B and S1C). To verify whether abnormal glycolysis is mediated by REEP6, glucose uptake and lactate production were measured in CRC cell lines in which REEP6 expression was depleted or overexpression by a lentivirus-mediated transfection system. As shown in Fig. 2B and Fig. S1D, REEP6 deletion dramatically attenuated the glucose uptake and lactate production in DLD-1 and RKO cells, while ectopic REEP6 expression led to the opposite effects. Likewise, similar results were further validated by extracellular acidification rate (ECAR) assay, an indicator of overall glycolytic flux. The glycolytic rate and glycolytic capacity were impaired by knocking down REEP6 in DLD-1 and RKO cells and enhanced when REEP6 was overexpressed (Fig. 2C and D, Fig. S1E and S1F). Collectively, these findings suggest that REEP6 promotes glycolysis in CRC cells.
Accumulating evidence has shown that aerobic glycolysis prominently fuels tumorigenesis and proliferation33,34. In consequence, we investigate whether REEP6 affects tumorigenesis and proliferation of CRC cells in vitro and in vivo. As expected, silencing REEP6 strikingly undermined the cell viability and colony formation capacity of DLD-1 and RKO cells, and upregulating REEP6 yielded an increased proliferation ability (Fig. S1G–S1I). The tumor-promoting role of REEP6 was further confirmed by CRC patient-derived organoid (PDO) models. Organoids’ number and size were both reduced by lentivirus-mediated REEP6 depletion. In contrast, organoids stably overexpressing REEP6 markedly strengthened their growth ability (Fig. 2E–G).
Next, subcutaneous xenograft mouse models were established to assess the function of REEP6 in CRC tumorigenesis in vivo. REEP6 knocking down or overexpressing DLD-1 cells and their corresponding control cells were inoculated into nude mice. Consistent with the above in vitro findings, DLD-1 cells stably depleted REEP6 developed tumors more slowly, with smaller tumor volumes and weights. While opposite outcomes were observed in DLD-1 cells with excessive expression of REEP6 (Fig. 2H and I). Recombinant adeno-associated virus serotypes 9 (AAV9) were excessively applied to intestinal disease study, for its relatively highly efficient gene transduction potential to intestinal epithelial cells31,35,36. AAV-REEP6 knockdown (AAV-REEP6KD), AAV-REEP6 overexpression (AAV-REEP6OE), and their corresponding control AAV (AAVshNC or AAVvector) were injected into 4-week-old male BALB/c mice through the tail vein. To confirm intestinal transfection efficiency, qRT-PCR and western blotting assays were employed to detect REEP6 expressions in colons (Supporting Information Fig. S2A and S2B). Azoxymethane and dextran sodium sulfate (AOM/DSS) models representing inflammation-induced tumorigenesis in CRC were combined with the AAV system to evaluate REEP6 in CRC tumorigenesis (Fig. 2J). AAV-REEP6KD mice displayed notably fewer tumor numbers and smaller tumor sizes than their AAVshNC mice (Fig. 2K and L). Moreover, increased survival rates were also observed in AAV-REEP6KD mice rather than AAVshNC mice (Fig. 2M). Consistently, AAV-REEP6KD animals displayed a dramatic deficiency in tumor cell proliferation and a remarked increase in tumor cell apoptosis characterized by decreased proportions of Ki-67 staining and elevated proportions of TUNEL staining (Fig. S2C–S2E). Moreover, AAV-REEP6OE mice exhibited opposite phenotypes in the abovementioned observational indicators compared to corresponding control mice (Fig. 2K–M, Fig. S2C–S2E). Together, these data illustrate that REEP6 facilitates CRC tumorigenesis in vivo.

3.3
REEP6 interacts with PGAM1 and enhances PGAM1 enzyme activity
Next, to uncover the underlying molecular mechanism by which REEP6 promotes glycolysis and tumorigenesis in CRC, co-immunoprecipitation (co-IP) coupled with protein mass spectrometry (MS) analyses found that the abundance of PGAM1 was prominent in REEP6-interacting proteins (Fig. 3A and Supporting Information Fig. S3A). Furthermore, PGAM1, known as a glycolytic enzyme in cancer metabolism, has been reported to be upregulated and enzymatically activated by many mechanisms20,21. The interaction between REEP6 and PGAM1 was further validated by endogenous and exogenous co-IP assays in DLD-1, RKO, and HEK-293T cells (Fig. 3B and C). Immunofluorescence (IF) analyses confirmed the cytoplasmic co-localization between REEP6 and PGAM1 in DLD-1 and RKO cells (Fig. 3D). Recombinant full-length GST-tagged REEP6 was generated for Glutathione S-transferase (GST) pull-down assay. The results showed that REEP6 is directly bound to PGAM1 (Fig. 3E). A series of truncated mutants of REEP6 and PGAM1 was constructed to depict the interacting domains (Fig. 3F). The results demonstrated that the C-terminal domain (CTD: residues 109-211) of REEP6 physically interacted with the truncated mutant 3 (TM3: residues 169-254) of PGAM1 (Fig. 3G and H). Moreover, it has been reported that PGAM1 interacts with ACTA2 to promote metastasis, which is not dependent on its metabolic functions37. We therefore investigate whether REEP6 interacts with PGAM1 through the ACTA2-binding motif independent of its metabolic activity. The results showed that overexpression of REEP6 in DLD-1 and RKO cells did not affect the binding between PGAM1 and ACTA2 (Fig. S3B). Furthermore, we constructed PGAM1 Δ201-210 deletions, which were verified to lose interaction with ATCA2, to examine its association with REEP637. The results showed that PGAM1 Δ201–210 deletions still retained the interaction with REEP6 (Fig. S3C). These results collectively suggest that REEP6 interacts with PGAM1 independent of the ACTA2-binding motif.
We next wondered whether REEP6 modulates PGAM1 expression levels and enzymatic activity. qRT-PCR and western blotting showed that REEP6 failed to regulate the mRNA and protein levels of PGAM1 (Fig. S3D–S3F). However, the enzymatic activity of PGAM1 could be effectively crippled via stably introducing shREEP6 to DLD-1 and RKO cells and augmented when REEP6 was ectopically expressed (Fig. 4A and Supporting Information Fig. S3G). Moreover, the levels of PGAM1 enzymatic reaction substrate 3-PG and product 2-PG were measured by ELISA assays. Our data suggest that the depletion of REEP6 markedly reduced the levels of 2-PG and increased the levels of 3-PG in DLD-1 and RKO cells, and vice versa (Fig. 4B and Fig. S3G). Then, we preliminarily explored the expression level of PGAM1 in CRC patients of cohorts 1 and 2. As the results shown in Fig. S3H, there were no significant differences in the expression of PGAM1 mRNA between the tumor tissues and the adjacent normal tissues in cohort 1. Simultaneously, the IHC staining of CRC tissue microarrays in cohort 2 showed that the protein level of PGAM1 did not show significant differences either (Fig. S3I and S3J). These results suggest that PGAM1 exhibits a certain level of expression in CRC, which is functionally significant despite not being highly overexpressed.
A small molecule inhibitor (PGMI-004A) and an enzymatically inactive mutant (PGAM1-H186R) of PGAM1 were introduced to confirm the role of PGAM1 enzyme activity in REEP6-mediated abnormal glycolysis30,37. Boosted glucose uptake, lactate production, glycolytic rate, and glycolytic capacity induced by REEP6 overexpression could be abolished by PGMI-004A treatment in DLD-1 and RKO cells (Fig. 4C and D, Supporting Information Fig. S4A and S4B). Moreover, treatment with PGMI-004A rescued the promoted PGAM1 enzymatic activity, increased 2-PG levels, and reduced 3-PG levels caused by REEP6 ectopic expression in DLD-1 and RKO cells (Fig. 4E and F, Fig. S4C). In consonance with these observations, decreased glucose uptake, lactate production, glycolytic rate and glycolytic capacity caused by REEP6 depletion could be rescued by the introduction of lentivirus-mediated wild-type PGAM1 expression but PGAM1-H186R failed (Fig. 4G–I, Fig. S4D and S4E). Furthermore, REEP6 knockdown impaired PGAM1 enzymatic activity, decreased 2-PG levels, and increased 3-PG levels in DLD-1 and RKO cells, which could be abolished by wild-type PGAM1, but PGAM1-H186R had no significant effect (Fig. 4J and K, Fig. S4F and S4G). These data together suggest that REEP6 interacts with PGAM1 and promotes PGAM1 enzyme activity. More importantly, PGAM1, serving as a mutase in glycolysis, mediates the process of REEP6-enhanced glycolysis.

3.4
REEP6 strengthens the PRMT5-PGAM1 complex and promotes PRMT5 catalyzing symmetric dimethylation of PGAM1
Considering many researchers reported that the enzymatic activity of PGAM1 could be regulated by multiple kinds of post-translational modifications (PTMs), we presumed that REEP6 induced the increase in PGAM1 activity may be mediated by a type of PTM21, 22, 23. By interrogating REEP6 potentially interacting proteins, PRMT5 as a classical arginine methyltransferase is ranked third based on protein abundance, which aroused our interest (Figure 3, Figure 5A). Furthermore, co-IP and GST pull-down assays confirmed the interaction between REEP6 and PRMT5 in CRC cells and HEK-293T cells (Fig. 5B, Supporting Information Fig. S5A and S5B). Confocal microscopy analysis of DLD-1 and RKO cells observed the co-localization between REEP6 and PRMT5 mainly in the cytoplasm (Fig. 5C). Indeed, we also found that REEP6 could not affect the mRNA and protein levels of PRMT5 (Fig. S5C–S5F). Moreover, PRMT5 did not regulate REEP6 or PGAM1 mRNA and protein expression (Fig. S5G–S5J). Next, we wondered whether REEP6 could act as a molecular scaffold to bridge and consolidate PRMT5 and PGAM1 interaction. Quantified co-IP experiments confirmed the interaction between PRMT5 and PGAM1, and the formation of this complex could be drastically cemented via REEP6 overexpression, whereas REEP6 depletion weakened it (Fig. 5D and E, Supporting Information Fig. S6A–S6C). Molecular domain mapping showed that the N-terminal domain (NTD: residues 1–43) of REEP6 and the β-barrel domain (D3: residues 420–637) located in PRMT5 mediated the interaction between REEP6 and PRMT5 (Fig. 5F–H). The Rossman fold domain (D2: residues 292–420) of PRMT5 and the truncated mutant 1 (TM1: residues 1–84) of PGAM1 were responsible for the binding of PRMT5 and PGAM1 (Fig. S6D and S6E). Furthermore, GST pull-down assays demonstrated direct physical interaction between PRMT5 and PGAM1 (Fig. S6F). These data suggest that REEP6-PRMT5-PGAM1 forms a heterotrimeric complex in CRC.
We next examined whether PRMT5 plays an arginine methyltransferase role in the symmetric dimethylation of PGAM1 and whether REEP6 enhanced this catalytic reaction. Our results exhibited that the SDMA level of PGAM1 was prominently suppressed through PRMT5 inhibitor GSK3326595 treatment in a dose- and time-dependent manner in DLD-1 and RKO cells (Fig. 6A and B, Fig. S6G–S6L). More importantly, this effect was reinforced by REEP6 overexpression, while abated by REEP6 depletion (Fig. 6A and B, Fig. S6G–S6L). Genetical inhibition of PRMT5 notably reduced the SDMA modification of PGAM1 and genetical introduction of PRMT5 facilitated PGAM1 methylation (Fig. 6C–F and Supporting Information Fig. S7A–S7D). On this basis, REEP6 overexpression strengthened and REEP6 knockdown weakened the SDMA of PGAM1 mediated by PRMT5 in DLD-1 and RKO cells (Fig. 6C–F and Fig. S7A–S7D). Similarly, the enzyme inactive mutant of PRMT5 (PRMT5-G367A/R368A) failed to catalyze the SDMA modification of PGAM1 compared to wild-type PRMT5 (Fig. 6G and Fig. S7E–S7G). Moreover, the promotion effect of REEP6 on PGAM1 methylation catalyzed by wild-type PRMT5 was also observed whereas it was difficult to detect the SDMA of PGAM1 mediated by PRMT5-G367A/R368A (Fig. 6G and Fig. S7E–S7G). Through an in vitro methylation assay, we also validated that the enzymatic activity of PRMT5 was essential for SDMA modification of PGAM1, and REEP6 could intensify this enzymatic reaction (Fig. 6H). Next, to investigate whether PRMT5 catalyzes SDMA modification on REEP6, co-IP assays were performed in HEK-293T cells. As shown in Fig. S7H, no SDMA signals were detected on REEP6 proteins when PRMT5 was ectopically expressed. In conclusion, these data indicate that REEP6 strengthens the formation of PRMT5–PGAM1 complex and further promotes PRMT5 catalyzing PGAM1 symmetric dimethylation.

3.5
REEP6 increased PGAM1 enzymatic activity is mediated by arginine methyltransferase PRMT5
We further investigated whether PRMT5 serving as an arginine methyltransferase could regulate enhanced PGAM1 enzymatic activity mediated by REEP6. Pharmacological or genetic elimination of PRMT5 obviously reversed the elevated glucose uptake, lactate production, glycolytic rate, and glycolytic capacity caused by ectopic expression of REEP6 in CRC cell lines (Supporting Information Fig. S8A–S8H). What's more, improved PGAM1 enzymatic activity, elevated 2-PG levels, and declined 3-PG levels mediated by REEP6 could be partially abolished by PRMT5 inhibition (Supporting Information Fig. S9A–S9D). Consistent with these results, overexpression of PRMT5-WT rescued the impaired glucose uptake, lactate production, glycolytic rate, and glycolytic capacity mediated by stable depletion of REEP6 in DLD-1 and RKO cells, while catalytic inactive mutant PRMT5-G367A/R368A was incapable of doing it (Fig. S9E–S9H). Similarly, PRMT5-WT but not PRMT5-G367A/R368A rescued the weakened PGAM1 enzymatic activity, reduced 2-PG levels, and increased 3-PG levels caused by REEP6 knockdown in DLD-1 and RKO cells (Fig. S9I and S9J). In summary, the catalytic activity of PRMT5 as an arginine methyltransferase is essential for increasing PGAM1 enzymatic activity mediated by REEP6.

3.6
REEP6 enhances PRMT5 catalyzing the symmetric dimethylation of PGAM1 at R40
To identify the exact arginine residues of PGAM1 SDMA modification, GPS-MSP and PRmePred were applied to predict the candidate arginine residues in PGAM1 (Supporting Information Fig. S10A and S10B)38, 39, 40. Then, wild-type PGAM1 (PGAM1-WT) and its mutants in which specific arginine was substituted by lysine (termed R40K, R180K, R191K, and R240K) were generated. In vitro methylation assay illustrated that PGAM1-R40K mutant clearly abolished methylation in PGAM1 mediated by recombinant PRMT5 and REEP6 failed to strengthen the SDMA level in PGAM1-R40K mutant (Fig. 7A). Then, PGAM1-WT and four mutants were transiently transfected to HEK-293T cells. Similar results were also observed in the in vivo methylation assay (Supporting Information Fig. S11A). Subsequently, we further verified whether the residue R40 of PGAM1 was dimethylated symmetrically in CRC cell lines and in vitro. The result showed that REEP6 obviously facilitated the SDMA level of PGAM1-WT which is catalyzed by PRMT5, whereas PGAM1-R40K completely eliminated it (Fig. 7B, Fig. S11B and S11C). More importantly, we also demonstrated that the enzymatic activity of PRMT5 is indispensable for the methylation of PGAM1 at R40 (Fig. 7C and Fig. S11D). Collectively, these data indicated that PRMT5 dimethylates symmetrically PGAM1 at R40, and REEP6 dramatically enhanced this catalytic reaction.
Considering REEP6-PRMT5-PGAM1 heterotrimeric complex promoted glycolysis in CRC cells, we detected whether combined inhibition of this ternary complex could yield a better therapeutic effect in CRC. GSK3326595 and PGMI-004A both effectively suppressed CRC cell lines and PDO proliferation in concentration-dependent effects, and the IC50 values were calculated (Fig. S11E and S11F). Next, our results showed that combined targeting REEP6-PRMT5-PGAM1 prominently impeded the cell proliferation and colony formation capacities of DLD-1 and RKO cells compared to single or two treatments (Fig. 7D and Fig. S11G–S11I). A similar phenomenon with an inhibitory organoid number and size was also observed in CRC PDO models (Fig. 7E and F). DLD-1 cell-derived xenograft models with stably depleted REEP6 were also applied to test the synergistic effect of this combined treatment. Compared with single or two treatments with shREEP6, GSK3326595, or PGMI-004A, combined three treatments with shREEP6, GSK3326595, and PGMI-004A dramatically inhibited xenograft growth (Fig. 7G–I). Collectively, the above data suggest that the combined targeting of REEP6–PRMT5–PGAM1 complex possesses anti-tumor potential in CRC.

3.7
SDMA of PGAM1 at R40 promotes glycolysis and tumorigenesis in CRC
Next, we investigated whether PGAM1 SDMA affects its enzymatic activity, glycolysis, and tumorigenesis in CRC. Firstly, DLD-1 and RKO cells expressing PGAM1-WT but not PGAM1-R40K mutant showed stronger ability in glucose uptake, lactate production, glycolytic rate, and glycolytic capacity (Fig. 8A and B, Supporting Information Fig. S12A and S12B). Moreover, compared to PGAM1-WT, undermined PGAM1 enzymatic activity, reduced 2-PG levels, and increased 3-PG levels were detected in cell lines with PGAM1-R40K ectopic expression (Fig. 8C, Fig. S12C and S12D). In addition, REEP6/PRMT5 overexpression in CRC cell lines which is expressing PGAM1-WT significantly promoted glycolysis and enzymatic activity of PGAM1, while PGAM1-R40K impaired this effect (Fig. 8A–C, Fig. S12A–S12D). Next, the reason why methylated PGAM1 enhanced its activity as a glycolytic enzyme was further explored. The current consensus is that the enzymatic activity of PGAM1 is derived from the phosphorylation of histidine 11 (H11)41,42. Therefore, we speculated that SDMA of PGAM1 at R40 might change the three-dimensional conformation of PGAM1, rendering subsequent enhanced H11 phosphorylation. AlphaFold v3 and PyMOL2.6.0 were employed to investigate the potential mechanism by which symmetric dimethylarginine at R40 (R40me2s) enhances phosphorylation of histidine 11 (H11) in PGAM1 (Fig. S12E) 43. As shown in Fig. 8D, R40me2s induces a conformational deflection in the C-terminal α-helix (residues 232–254) of PGAM1, resulting in a displacement of 14.1 Å and a bending angle of 26.9°. This structural rearrangement enlarges the entrance of the H11 phosphorylation pocket, which is formed by two α-helices (residues 30–49 and 232–254) (Fig. 8E). The increased cavity volume enhances the contact area and binding opportunities between PGAM1 and its upstream kinase. Furthermore, the extended side chain of R40me2s stabilizes the pocket entrance. Therefore, we speculate that the alteration in cavity volume caused by the R40me2s enhances the phosphorylation of H11 in PGAM1. In order to verify our speculation, we transfected a series of PGAM1 point mutants into CRC cells. The results showed that compared with wild-type PGAM1, mutating R40 to lysine (PGAM1-R40K) abolished SDMA signals and significantly weakened phosphorylation signals at H11 (Fig. S12F). Consistent with this, PGAM1-R40K and H11A mutants showed a remarked deficiency in enzymatic activity (Fig. 8F and Fig. S12G).
Then, we further evaluated the impact of PGAM1 SDMA on CRC tumorigenesis and proliferation in vitro and in vivo. PGAM1-WT but not PGAM1-R40K exhibited pronouncedly promoted effects on cell viability and colony formation capacity in DLD-1 and RKO cells (Supporting Information Figs. S12H, S12I, and S13A, S13B). Moreover, PGAM1-WT drastically enhanced cell viability and colony formation capacity in the presence of REEP6/PRMT5 ectopic expression, but not so much in the PGAM1-R40K mutant (Figs. S12H, S12I, and S13A, S13B). Subsequently, CRC PDO models and subcutaneous xenograft models further confirmed the above in vitro results (Fig. S13C and S13D, Fig. 8G and H). The AAV gene delivery system combined with AOM/DSS models was employed to validate the effect of PGAM1 SDMA on CRC tumorigenesis. Mice from the AAV-PGAM1WT group had elevated tumor numbers and enhanced tumor volumes than their AAVvector mice, whereas AAV-PGAM1R40K prominently whittled this effect down (Fig. 8I and J). What's more, the AAV-PGAM1WT group rather than AAVvector mice, showed a crippled survival rate, and PGAM1-R40K mutant dramatically improved it (Fig. 8K). Meanwhile, we also found that proliferation potential which is characterized by Ki67 positive nuclei staining was increased, while apoptosis proportions quantified by the intensity of TUNEL staining were diminished in the AAV-PGAM1WT group rather than their AAVvector or AAV-PGAM1R40K mice (Fig. S13E–S13G). More than that, GSK3326595 treatment could obviously impede AOM/DSS-induced tumorigenesis in AAV-PGAM1WT, but the effect was weaker in AAV-PGAM1R40K mice (Fig. 8I–K, Fig. S13E–S13G).

4
Discussion
The aberrant expression of REEP6 was reported in many cancers and its cancer-promoting role was illustrated by some researchers25, 26, 27. But REEP6 initially acts as a receptor expression-enhancing protein, the expression pattern and the new function of cancer glycolysis remain largely unknown in CRC. Here, our study reported that REEP6 is significantly elevated in CRC tissues compared to normal tissues. More importantly, CRC patients with high REEP6 expression indicated worse clinicopathological features and a poorer overall survival rate in our cohorts. Our results confirmed that REEP6 fueled the cell proliferation and tumorigenesis in CRC cells which was caused by the enhanced glycolytic flux. Even so, the exact mechanism implicated in the upregulation of REEP6 in CRC cells needs further exploration. The integration of nanotechnology (including lipid nanoparticles, polymer nanoparticles, and bioinspired vectors) combined with RNA interference (RNAi) has emerged as an innovative approach for cancer targeting therapy44, 45, 46. Given the aberrant expression and the cancer-promoting role of REEP6, it will be a promising avenue for CRC therapeutic strategy that the utilization of nanoparticles delivered siRNA targeting REEP6.
In this study, we demonstrated that REEP6 serving as a molecular scaffold contributed to the formation of the PRMT5–PGAM1 complex. Furthermore, the formation of REEP6–PRMT5–PGAM1 ternary complex promoted PRMT5-mediated SDMA of PGAM1 at R40, which dramatically increased PGAM1 enzymatic activity in glycolysis. PRMT5 has been reported as an essential protein arginine methyltransferase involved in multiple non-metabolic pathways during CRC progression, which implicates several cellular processes such as epithelial-mesenchymal transition (EMT), ferroptosis, and histone modification9,10,47, 48, 49. Zhu et al.47 reported that the inhibition of PRMT5 enhances irinotecan sensitivity, inducing the release of cytosolic double-stranded DNA, then activating the cGAS–STING signaling pathway, thereby enhancing anti-tumor immunotherapy for CRC patients with microsatellite stable state. This study emphasizes the novel role of PRMT5 as a DNA damage repair-related epigenetic gene in enhancing chemotherapy and immunotherapy sensitivity. In addition, PRMT5 has been demonstrated to regulate the AKT/mTOR signaling pathway through arginine methylation of AKT1, which participates in colorectal cancer liver metastasis48. Moreover, PRMT5 directly catalyzes symmetric demethylation of AlkB homologue 5 (ALKBH5) at Arg316 or inhibits ALKBH5 transcription by mediating symmetric dimethylation of histone residues (H4R3me2s and H3R8me2s). These actions aggravate CRC progression by facilitating immune evasion or attenuating ferroptosis, respectively10,49. It is necessary to recognize that these non-metabolic pathways may have synergistic or independent effects with our study on the metabolic axis. The therapeutic strategy of inhibiting PRMT5 activity may simultaneously target metabolic and non-metabolic pathways in CRC patients.
Most importantly, our report found that combined treatment targeting the REEP6–PRMT5–PGAM1 heterotrimeric complex showed a synergistic effect on anti-tumor tactics. For suppressing the enzymatic activity of PGAM1 in glycolysis, a series of inhibitors, including PGMI-004A, HK99, and KH3 have been developed in cancer treatment30,50, 51, 52. Among these, PGMI-004A, a potent PGAM1 inhibitor with an IC50 of 13.1 μmol/L, showed remarkable anti-tumor potential in xenograft mice in vivo, and in various human malignant tumor cells in vitro30. Next, as a potent and highly specific small-molecule inhibitor, GSK3326595 targeting the catalytic site of PRMT5 was also applied in the combined treatment39,53. GSK3326595 has shown a markedly antitumor and a favorable prognostic efficiency against some solid tumors and hematological malignant tumors in phase I/II clinical trials (NCT04676516, NCT03614728, and NCT02783300). Our report indicated that compared with single or two treatments with shREEP6, GSK3326595, or PGMI-004A, combined three treatments with shREEP6, GSK3326595, and PGMI-004A obviously impeded the growth and tumorigenesis in CRC cell lines, CRC PDO models, and xenograft models. This result paves the way for the combination targeting therapies that may provide a more promising strategy for CRC treatment. However, it should be noted that further validation through animal studies and subsequent clinical trials is essential for this combination therapy strategy.

5
Conclusions
Our study identified that REEP6 is significantly upregulated in CRC tissues and correlated with a poor prognosis in CRC patients. REEP6 promotes aerobic glycolysis and tumorigenesis in CRC both in vitro and in vivo. Mechanistically, REEP6 strengthens the PRMT5–PGAM1 complex and promotes PRMT5-mediated SDMA of PGAM1 at R40 residues. The methylated PGAM1 dramatically enhances its enzymatic activity and therefore boosts glycolytic flux in CRC cells. More than that, combined treatment targeting the REEP6–PRMT5–PGAM1 complex exhibit synergistic anti-tumor efficacy in CRC, which may present a novel and efficacious therapeutic strategy for CRC treatment.

Author contributions
Yueming Sun and Tuo Wang generated the hypothesis and designed the experiments. Tuo Wang, Dongsheng Zhang, Chi Jin, and Hengjie Xu performed experiments. Tuo Wang performed animal experiments. Tuo Wang interpreted the data. Tuo Wang wrote the manuscript. Yueming Sun supervised the overall research, secured funding, and interpreted results. The author(s) read and approved the final manuscript.

Ethics statement
Human tissue study is approved by the committee on the ethics of The First School of Clinical Medicine, Nanjing Medical University and all patients endorsed informed consent. All animal experiments are conducted with the approval of the Committee on the Ethics of Animal Experiments of Nanjing Medical University. The ethical number of the animal experiment is IACUC-2403048.

Conflicts of interest
The authors declare no conflicts of interest.