NSUN2 promotes colorectal cancer progression by stabilizing PHGDH mRNA to promote serine metabolism reprogramming

Abstract

Purpose

Cancer cells rely on serine biosynthesis for growth, but its regulation in colorectal cancer (CRC) remains not well understood. This study identifies the m⁵C methyltransferase NSUN2 (NOP2/Sun domain family, member 2) as a key regulator of serine biosynthesis, revealing a novel mechanism driving CRC progression.

Methods

The expression and prognostic value of NSUN2 were evaluated using bioinformatics analyses and immunohistochemistry (IHC) assays. The effects of NSUN2 on cellular serine biosynthesis, intracellular reactive oxygen species (ROS) levels, and apoptosis levels were analyzed both in vitro and in vivo. Additionally, RNA sequencing, Methylated RNA Immunoprecipitation sequencing (MeRIP-seq), RNA immunoprecipitation (RIP), and RNA stability assays were utilized to screen and validate the association between NSUN2 and phosphoglycerate dehydrogenase (PHGDH).

Results

NSUN2 was found to be highly expressed in CRC and associated with poor patient survival. PHGDH, a direct downstream target of NSUN2, plays a crucial role in NSUN2-mediated serine biosynthesis. Furthermore, inhibition of NSUN2 significantly reduced the intracellular NADH/NAD⁺ and NADPH/NADP⁺ ratios, leading to an increase in ROS levels and apoptosis levels, thereby inhibiting CRC progression. Additionally, NSUN2 enhances PHGDH expression and mRNA stability by binding to the “reader” protein m⁵C-Aly/REF export factor (ALYREF).

Conclusions

This study identified a novel NSUN2/ALYREF/m⁵C-PHGDH axis might be promising therapeutic targets for CRC.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40170-025-00406-1.

Background

Colorectal cancer (CRC) was the fourth leading cause of cancer-related deaths in both men and women under 50 years of age in the late 1990s, but it is now the leading cause in men and the second leading cause in women. The incidence of CRC in young adults (< 55 years) has also been increasing by 1%-2% annually [1]. Although surgery, chemotherapy, and radiotherapy are the main treatment options for CRC, the prognosis for patients remains unfavorable, with a 5-year survival rate of just 13.1% in metastatic CRC cases [2]. Understanding the evolutionary and metastatic mechanisms of CRC is crucial for developing more targeted and effective therapies.

Posttranscriptional modifications in RNA have been shown to be involved in tumor progression [3–5]. 5-methylcytosine (m⁵C), a recently identified RNA modification, can occur in various types of eukaryotic RNAs, including tRNA, rRNA, lncRNA, and mRNA [6], and has been recognized for its role in various cellular processes, such as RNA stability, trafficking, and translational control, thereby influencing cell fate determination and potentially contributing to tumorigenesis. The dysregulation of m⁵C methylation has been increasingly recognized for its intricate association with the carcinogenesis and tumorigenesis of various human cancers, including bladder cancer (BLCA), esophageal carcinoma (ESCA), stomach adenocarcinoma (STAD), and liver hepatocellular carcinoma (LIHC) [3, 7–10] Nonetheless, complete elucidation of whether m⁵C modifications contribute to the development of CRC remains to be determined.

Deregulation of cellular energetics (metabolic reprogramming), a hallmark of cancer, leads tumor cells to rewire metabolic pathways to support rapid proliferation, continuous growth, metastasis, and resistance to therapies [11]. Serine, a critical one-carbon unit donor, is involved in both the methionine and folate cycles, contributing to nucleotide synthesis, methylation reactions, and the generation of NADPH for antioxidant defense [12, 13]. Serine metabolism is activated in various cancer types to support tumor growth and metastasis by providing precursors for macromolecule synthesis (such as proteins, nucleotides, and lipids) and one-carbon units for methylation reactions [14–16]. Despite their important roles in tumorigenesis, the regulatory mechanisms underlying the amino acid metabolic reprogramming of tumor cells remain incompletely elucidated.

Herein, we initially utilized clinical data analysis and supported our findings with both in vivo and in vitro experiments, revealing elevated expression levels of NSUN2 in CRC, which exerts oncogenic effects. Subsequently, methylation sequencing combined with targeted proteomics analysis suggested that NSUN2 mediates metabolic reprogramming in tumor tissues—by regulating the m⁵C methylation of PHGDH, it enhances serine metabolism, leading to increased serine production levels, thereby promoting tumor cell survival and proliferation. Mechanistically, elevated levels of NSUN2 lead to increased m⁵C modifications of PHGDH mRNA, which, upon recognition by the ALYREF reading protein, enhances PHGDH mRNA stability, thereby increasing its protein levels and ultimately activating the downstream serine metabolism pathway. Taken together, we uncovered an intriguing NSUN2/ALYREF/m⁵C-PHGDH signaling axis, thereby bridging the connection between metabolic reprogramming and epigenetic remodeling.

Methods

Patient samples

Tumor samples of CRC patients were obtained from Renji Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. Written informed consent was provided by all patients. The acquisition and use of tumor specimens were approved by the Ethics Committee of Renji Hospital Affiliated to Shanghai Jiao Tong University School of Medicine.

IHC staining and analysis

Tissues were fixed with formalin, embedded in paraffin, and then cut into 3 μm-thick sections. IHC staining was performed on a Leica BOND-MAX (Leica, Germany) and the antibodies used are listed in Table 1.

Table 1.

Antibodies used in the experiment

Antibody	Lot	Company
NSUN2	ab214727	Abcam
PHGDH	14719-1-AP	Proteintech
ALYREF	16690-1-AP	Proteintech
YBX1	ab76149	Abcam
NOP2	10448-1-AP	Proteintech
NSUN5	15449-1-AP	Proteintech
DNMT1	24206-1-AP	Proteintech
Beta-Actin	66009-1-Ig	Proteintech
igG	ab172730	Abcam
m5C	ab214727	Abcam
ABflo® 405-conjugated Goat anti-Rabbit IgG (H + L)	as056	Abclonal
HRP-conjugated Goat anti-Rabbit IgG (H + L)	as014	Abclonal

IHC staining was evaluated based on staining intensity and the proportion of positively stained area. Staining intensity was scored on a scale of 0 to 3, representing no, weak, moderate, and strong staining, respectively. The extent of positively stained cells was graded on a scale from 1 to 4, indicating a positively stained area of < 10%, 10–40%, 40–70%, and > 70%, respectively. The immunoreactivity score, which combines both staining intensity and the proportion of positively stained area, was calculated by two independent pathologists.

Cell culture

The human CRC cell lines HCT15, HCT116, SW620 were purchased from the American Type Culture Collection (ATCC). HCT15 cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS), HCT116 cells were cultured in McCoy’s 5 A medium supplemented with 10% FBS, and SW620 cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% FBS. All cell lines were maintained in a humidified incubator at 37 °C with 5% CO2.

Cell transfection

All plasmids were purchased from Genechem (Shanghai, China), and their sequences are listed in Table 2. The knockdown (GV248) vector or overexpression (GV492) vector, along with pMD2.G and psPAX2, were transfected into 293T cells to generate lentiviruses using Lipofectamine 3000 (#L3000150, Invitrogen, USA). CRC cells were then stably transfected with lentiviruses targeting NSUN2 knockdown (sh-NSUN2), NSUN2 overexpression, PHGDH overexpression, ALYREF knockdown, or a negative control, using polybrene (#40804ES76, YEASEN, China).

Table 2.

Primers, ShRNA used in the experiment

Gene	Forwards (5′-3′)	Reverse (5′-3′)
NSUN2	ACAGCCACTGAGTTGGTATCC	TTATGATGAGGCCGCACGTT
PHGDH	CAAGGCACTACGCCTGTACT	TCCCCATCTGACACCAGTGA
ALYREF	CCATGGCCGACAAAATGGAC	CGGGAAGTTGTTTTGGCCTG
Beta-Actin	CATGTACGTTGCTATCCAGGC	CTCCTTAATGTCACGCACGAT
RIP-PHGDH	ACTGGTGTCAGATGGGGAGA	GCGTTCAGCCCAAGAATGTG
PLEKHG2	AGGAAGGGCTGGAGATGGAT	GAGAGGGTGGGAGTGCTAGA
VGF	CCTCTTGGTCATGAAAGC	GGCTCTTTATGCTCAGAG
LAMA5	GGGGTGTCTGTATCGACTG	ACCGCTCCCCAGAGAAGTT
KCTD15	GCGGAGGGAGGAAACAT	TCAGGACGTAGCGGAAAA
C6orf141	AGGAGCCCAACTACCCTTCT	TCCTCAGTCCTCGTGGTCAT
Gene	Forwards (5′-3′)
shNSUN2-1	CCCAAGAATGAACGGCTTCAT
shNSUN2-2	CAGTGGAAGGTAATGACGAAA
shALYREF-1	GCTTGTCACGTCACAGATTGA
shALYREF-2	GCGTAAACAGAGGTGGCATGA
shNOP2	GACGATGCTGATACGGTAGAT
shNSUN5	CTCCGATGATGTAGTTGATTA
shDNMT1	GGAATGGCAGATGCCAACAGC
shYBX1	GTTCAATGTAAGGAACGGATA

Cell protein lysis and western blotting assay

To extract total cellular proteins, cells were lysed in RIPA lysis buffer (#20114ES, YEASEN, China) supplemented with protease inhibitor cocktail (#HY-K0011, MedChemExpress, USA). Western blotting analysis was performed following standard protocols, and target protein signals were detected using the ChemiDoc™ MP Imaging System (BIO-RAD, USA).

qRT-PCR

Total RNA was extracted from cells and tissues using TRIzol reagent (#12183555, Invitrogen, USA). The extracted RNA was then reverse transcribed into cDNA using Hifair^® III 1st Strand cDNA Synthesis Kit (#11139ES60, YEASEN, China). The quantitative PCR mixture was prepared using Hieff^® qPCR SYBR Green Master Mix (#11203ES03, YEASEN, China). The amplified transcript levels of each specific gene were normalized to that of ACTB. Primers used in the Cellular Oncology study were purchased from Sanggon Biotech and are listed in Table 2.

Cell proliferation and colony formation assays

For the cell proliferation assay, control and transfected cells were seeded in a 96-well plate (1 000 cells/well), with five replicates per group. Cell viability was assessed every 24 h. After adding 10 µL of Cell Counting Kit-8 (#C0005, TargetMoI, USA) to each well, the plate was incubated at 37 °C for 2 h. The absorbance at 450 nm was then measured for each sample using a spectrophotometer. For the colony formation assay, 800–1 000 control or transfected cells were plated in a six-well plate and cultured in medium containing 10% FBS for 12 days. Colonies were fixed with 4% paraformaldehyde (#P0099-3 L, Beyotime, China) and stained with Crystal Violet Staining Solution (#C0121, Beyotime, China). The colony number was quantified by counting stained colonies using ImageJ software.

Transwell migration and invasion assay

Cell migration and invasion were assessed using the Transwell assay. A total of 2 × 10^4 cells/well were seeded in 24-wellTranswell insert chambers (#CLS3422, Corning, USA) containing serum-free medium (2% Matrigel was used to pre-coat inserting chambers for invasion assay). FBS (20%) was added to the lower chamber as a chemoattractant. After 48 h, non-migrated cells on the upper surface of the membrane were removed, while the migrated cells on the lower surface were fixed with 4% paraformaldehyde (#P0099-3 L, Beyotime, China) and stained with Crystal Violet Staining Solution (#C0121, Beyotime, China). The stained cells were then observed and photographed using a microscope.

Immunofluorescence

CRC cells were seeded onto glass coverslips in 24-well plates (50,000 cells/well). After washing with PBS (#G4202, Servicebio, China), the cells were fixed with 4% paraformaldehyde (#P0099-3 L, Beyotime, China) for 15 min at room temperature (RT). The slides were then incubated with 0.2% Triton X-100 (#HY-Y1883A, MedChemExpress, USA) for 15 min to permeabilize the cells and blocked with 3% BSA (#HY-D0842, MedChemExpress, USA) for 30 min. After washing three times with PBS containing 0.1% Tween-20 (PBST, #ST825, Beyotime, China), the cells were incubated with primary antibody (Table 1) overnight at 4 °C. The slides were washed three times with PBST and incubated with secondary antibody at RT for 1 h. Next, the slides were washed three times with PBST and incubated with DAPI at room temperature for 15 min. Finally, after three washes, the cells were mounted with Anti-Fade Mounting Medium (#HY-K1042, MedChemExpress, USA) and imaged using the LSM 900 Airyscan confocal laser-scanning microscope (Zeiss, Germany).

RNA m⁵C Dot blot assay

Total RNA was extracted from CRC cells, and mRNA was isolated and denatured at 65 °C for 3 min. Different amounts of mRNA (100, 200, or 400 ng) were then loaded onto a nitrocellulose membrane (#41105339, Merck, Germany). The membranes were UV-crosslinked for 5 minutes, followed by blocking with 5% nonfat dry milk in PBST for 2 h. The membranes were incubated with an m⁵C antibody (#ab214727, Abcam, USA) overnight at 4 °C. After washing, the membranes were incubated with HRP-conjugated goat anti-mouse IgG (#as014, Proteintech, USA) for 1 h at RT. Chemiluminescent detection was performed using a chemiluminescence system (Bio-Rad, USA). Subsequently, the membranes were stained with 0.02% methylene blue (#A610622, Sangon Biotech, China) to verify RNA loading.

RIP assay

Total RNA was isolated from CRC cells and treated with DNase I (#11284932001, Merck, Germany) to remove any contaminating DNA. The RNA was fragmented by sonication for 10 s in an ice-cold bath. RIP was then performed using the BeyoRIP™ RIP Assay Kit (#P1801S, Beyotime, China) following the manufacturer’s protocol. The RNA from both experimental and control groups, as well as the Input group, was collected for subsequent qRT-PCR analysis. Primer sequences used for this study are listed in Table 2.

mRNA stability assay

Stable transfected CRC cell lines were treated with actinomycin D (10 µg/ml) (#HY-17559, MedChemExpress, USA) for the indicated time points (0, 2, 4, or 6 h). Following treatment, the cells were harvested, and total RNA was extracted using TRIzol reagent (#12183555, Invitrogen, USA). The isolated mRNA was reverse transcribed, and the expression levels of target genes were quantified by qRT-PCR.

Luciferase reporter assay

For mutant reporter plasmid, the specific thymine (T) were replaced by guanine (G) (5’UTR mutation:51G; CDS mutations:1446G, 1580G, 1686G), respectively. Pre-treated CRC cells were seeded into 24-well plate followed by co-transfection of 0.5 µg of wild-type or mutated PHGDH reporter plasmids and 25 ng pRL-TK plasmids (renilla luciferase reporter vector) using Lipofectamine 3000 (#L3000150, Invitrogen, USA). Following transfection for 24–36 h, cells were lysed, and luciferase activities were measured using the Dual-Glo Luciferase system (Promega, USA) with the normalization to pRL-TK. All experiments were performed in triplicate to ensure statistical reliability.

NADH/NAD + Ratio, NADPH/NADP + Ratio quantification

The intracellular NADH/NAD + ratio and NADPH/NADP + ratio was quantified using the NAD+/NADH Assay Kit (Beyotime, S0175) and the NADP+/NADPH Assay Kit (Beyotime, S0179) following the manufacturer’s guidelines. The absorbance of NAD+, NADH, NADP + and NADPH was measured at a wavelength of 450 nm using a spectrophotometer.

ROS detection

Cells were seeded in 6-well plates and then incubated with DCFH-DA (#S0035M, Beyotime, China) according to the manufacturer’s instructions. Intracellular ROS levels were measured using a flow cytometer (BD Biosciences, USA), and the data were analyzed with FlowJo software (FlowJo 10.4, LLC, USA).

Cell apoptosis assay

Apoptotic cell proportions were assessed using an Annexin V-PE/7-AAD Cell Apoptosis Detection Kit (#G1512, Servicebio, China) according to the manufacturer’s instructions. Following the indicated treatments, CRC cells were analyzed using a flow cytometer (BD Biosciences, USA) and the data were processed with FlowJo software (FlowJo 10.4, LLC, USA).

Measurement of serine and glycine levels

Cells were cultured in six-well plates with serine- and glycine-free DMEM, supplemented with 10% dialyzed FBS. Serine and glycine concentrations in the cell lysates were subsequently measured using the DL-Serine Assay Kit (#ab211100, Abcam, USA) and the Glycine Assay Kit (#ab241027, Abcam, USA), according to the manufacturer’s protocols. Fluorescent intensity was quantified using a Varioskan LUX Multimode Microplate Reader (Invitrogen, USA) with excitation/emission wavelengths set at 535/587 nm.

Measurement of intracellular amino acid content

1 × 10^7 stable transfected SW620 cells were washed three times with PBS, then scraped into a 2 ml centrifuge tube using a cell scraper. The centrifuge tube was quickly placed in liquid nitrogen for 30 s. LC–MS/MS analysis to measure the intracellular amino acid content was performed by Biotree Biomedical technology (Shanghai, China). Metabolites were extracted from SW620 cells using extract solution (acetonitrile: methanol = 1:1, containing isotopically-labeled internal standard mixture). Standard solutions were prepared. An Agilent 1290 Infinity II series ultra-high-performance liquid chromatography (UHPLC) System (Agilent Technologies), equipped with a Waters ACQUITY UPLC BEH Amide column, was used to carry out UHPLC separation. An Agilent 6460 triple quadrupole mass spectrometer (Agilent Technologies), equipped with an AJS electrospray ionization (AJS-ESI) interface, was used for mass spectrometry. Agilent MassHunter Work Station Software (B.08.00, Agilent Technologies) was employed for data acquisition and quantification of target amino acids. Final amino acid concentrations were normalized to total cell count.

Metabolic flux analysis with ¹³C isotopic tracers

5 × 10^6 SW620 cells were washed with PBS twice and then cultured in a glucose-freemedium supplemented with 10% dialyzed FBS and 25mM U-[¹³C]-glucose (HY-B0389A, MedChemExpress, USA) for 24 h. The cells were then washed twice with cold PBS, resuspended in pre-chilled 80% methanol (#M116125, Aladdin, China) and 20% ddH2O for metabolite extraction. LC–MS/MS analysis to measure the abundance of serine and glycine with U-[¹³C]-glucose carbon incorporation was performed by LipidALL Technologies (Changzhou, China).

Nude mouse xenograft model

Four-week-old male BALB/c nude mice were purchased from Gem Pharmatech (Nanjing, China). SW620 cells stably transfected with the indicated virus (4 × 10^6 cells per mouse) were inoculated subcutaneously on the right ventral side of the mice. Tumor volume was measured every four days. After approximately three weeks, the mice were euthanized, and the tumors were surgically excised, weighed, fixed, and embedded for IHC. Animal care and procedures were conducted in compliance with all relevant ethical guidelines and approved by the Experimental Animal Welfare and Ethics Committee of Renji Hospital, Affiliated to Shanghai Jiao Tong University School of Medicine.

RNA-seq

RNA was isolated from HCT15 cells stably transfected with either the shNSUN2 lentivirus or the control using Trizol reagent (#12183555, Invitrogen, USA). The RNA samples were then sent to Novelbio (Shanghai, China) for sequencing using the Affymetrix Human U133 Plus 2.0 Array.

Bioinformatics analysis

The CRC datasets were downloaded from the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) databases, with the GEO database including GSE21510, GSE44076, GSE73360, and GSE226129. Bioinformatics analysis and visualization were conducted using R 4.1.0 (http://cran.r-project.org/).

Statistical analysis

For continuous variables, depending on whether the variables were normally distributed, Studentʼs t test or Wilcoxon rank-sum test was used as appropriate for intergroup comparisons between two groups and the Analysis of variance (ANOVA) or Kruskal–Wallis rank-sum test was used for intergroup comparisons among multiple groups. For categorical variables, the Chi-square test or Fisher exact test was used for intergroup comparisons. The Kaplan-Meier method was used to estimate OS, and the log-rank test was used to evaluate differences between survival curves. Prism 9.0 (GraphPad, La Jolla, CA) and R 4.1.0 (http://cran.r-project.org/) were used for statistical analysis. P < 0.05 was considered to indicate statistical significance.

Result

Elevated NSUN2 expression correlates with poor prognosis of patients with CRC

Accumulating studies have shown that m⁵C modification and its regulators play pivotal roles in the initiation and progression of a wide range of cancers [3, 9, 10]. To elucidate the functional roles of m⁵C modification in CRC, we first examined the m⁵C RNA levels in CRC tissues and paired normal intestinal mucosa. We found that the m⁵C modification level was significantly higher in CRC tissues via a dot blot assay (Fig. 1A). Similarly, IHC results from the CRC tissues demonstrated an elevated level of m⁵C modification in tumor tissues compared to adjacent normal tissues (Fig. 1B). Subsequently, we evaluated the expression levels of major m⁵C methyltransferases in the CRC datasets GSE21510, GSE73360, GSE44076, and the CRC cohort from TCGA database. Among m⁵C methyltransferases analyzed, only NOP2, NSUN2, NSUN5, DNMT1and NSUN5 exhibited consistently significant differential expression between CRC tissues and their paired adjacent normal tissues in all examined datasets (Fig. S1A). We then overexpressed these methyltransferase genes in CRC cell lines and found that NSUN2 overexpression most prominently promoted CRC cell proliferation (Fig. S1B-C). Therefore, NSUN2 was selected for further investigation. In freshly collected CRC tissue samples, NSUN2 was observed to be overexpressed at both mRNA and protein levels in tumor tissues compared to adjacent normal tissues (Fig. 1C-D). IHC analysis further confirmed the elevated expression of NSUN2 in tumor tissues (Fig. 1E). Kaplan-Meier survival analysis demonstrated that CRC patients with high NSUN2 expression had significantly worse overall survival (OS) than those with low expression (Fig. 1F and Table 3).

Fig. 1.

Elevated NSUN2 expression correlates with poor prognosis of patients with CRC. (A) m⁵C dot blot assay was used to detect the m⁵C levels of mRNA extracted from CRC tissues compared with adjacent normal tissues respectively in CRC patients. (B) Representative IHC staining images for m⁵C protein in human CRC specimens (scale bars = 250 μm and 50 μm), demonstrating its differential expression between tumor (n = 56) and adjacent normal (n = 56) tissue. (C) Western blot analysis showed the NSUN2 protein levels in paired samples of CRC tumors and corresponding normal tissues. (D) qRT-PCR assay showed the NSUN2 mRNA levels in paired samples of CRC tumors and corresponding normal tissues. (E) Representative IHC staining images for NSUN2 protein in human CRC specimens (scale bars = 250 μm and 50 μm), demonstrating its differential expression between tumor (n = 56) and adjacent normal (n = 56) tissue. (F) Kaplan-Meier analysis of OS in the Renji Hospital cohort (n = 180), comparing CRC patients with high (n = 84) versus low (n = 96) tumor NSUN2 expression

Table 3.

Correlation between NSUN2 expression and different clinical characteristics

Characteristics	n = 180	High (%) (n = 84)	Low (%) (n = 96)	p
Age				0.523
< 60	86 (47.78%)	38 (45.24%)	48 (50.00%)
≥ 60	94 (52.22%)	46 (54.76%)	48 (50.00%)
Gender				0.630
Male	102 (56.67%)	46 (54.76%)	56 (58.33%)
Female	78 (43.33%)	38 (45.24%)	40 (41.67%)
T stage				0.031
T1-T2	48 (26.67%)	16 (19.05%)	32 (33.33%)
T3-T4	132 (73.33%)	68 (80.95%)	64 (66.67%)
N stage				0.140
N0	111 (61.67%)	47 (55.95%)	64 (66.67%)
N1 + N2	69 (38.33%)	37 (44.05%)	32 (33.33%)
M stage				0.013
M0	161 (89.44%)	70 (83.33%)	91 (94.79%)
M1	19 (10.56%)	14 (16.67%)	5 (5.21%)
TNM stage				0.130
I-II	105 (61.67%)	44 (52.38%)	61 (63.54%)
III-IV	75 (38.33%)	40 (47.62%)	35 (36.46%)

NSUN2 enhances the tumorigenesis and progression of CRC

To further determine whether NSUN2 functions as a tumor-promoting factor in CRC, we constructed two NSUN2-knockdown (shNSUN2-1 and shNSUN2-2) CRC cell lines in SW620 and HCT116 with shRNAs (Fig. 2A, B), while overexpressed NSUN2 using NSUN2 overexpression wild-type vector (NSUN2-WT) in HCT15 cells and determined the efficiency by western blotting (Fig. 2C). CCK8 proliferation and colony formation assays were used to explore the regulatory effect of NSUN2 on the proliferation abilities of CRC cells. Figure 2D-F shows that depletion of NSUN2 inhibited cell proliferation of CRC cells while overexpressing NSUN2 showed the opposite effects (Fig. 2G-H). In addition, reintroducing NSUN2 reversed the compromised growth ability of CRC cells following NSUN2 knockdown (Fig. S2A-C). Additionally, there was a significant decrease in the number of migrating and invading cells following NSUN2 knockdown (Fig. 2I-J). In contrast, the upregulation of NSUN2 resulted in increased cell migration and invasion abilities (Fig. 2K). Rescue experiments indicated that NSUN2 plays a key role in promoting the migration and invasion of CRC cells, further confirming its oncogenic role (Fig. S2D). We further evaluated the effect of NSUN2 on CRC tumorigenesis in vivo. Stably infected SW620 cells were inoculated into the flanks of nude mice to observe the effect of NSUN2 on xenograft tumor growth. Compared to the control group, the NSUN2 knockdown group exhibited significantly reduced tumor sizes and weights, as well as lower levels of the proliferation marker Ki67, effects that were reversed upon reintroduction of NSUN2 (Fig. S2E-H). Taken together, these results indicate that NSUN2 promotes CRC growth both in vitro and in vivo.

Fig. 2.

NSUN2 enhances the tumorigenesis and progression of CRC. A, B) After infection of SW620 (A) and HCT116 (B) cells with shRNAs, the expression of NSUN2 was analsyzed by Western blot (left) and qRT-PCR (right) assays. C) After infection of HCT15 cells with NSUN2-WT plasmids, the expression of NSUN2 was analyzed by Western blot (left) and qRT-PCR (right) assays. D-E) After knocking down NSUN2 in SW620 (D) and HCT116 (E) cells, cell proliferation ability was measured by the CCK8 assay. F) After knocking down NSUN2 in SW620 and HCT116 cells, cell colony formation ability was measured by the colony formation experiments. G) Cell proliferation ability was assessed using the CCK8 assay following NSUN2 overexpression in HCT15 cells. H) After overexpressing NSUN2 in HCT15 cells, cell proliferation ability was assessed by colony formation assay. I-J) After knocking down NSUN2 in SW620 (I) and HCT116 (J) cells, cell migration and invasion were determined by transwell assays. scale bars = 100 μm. K) After overexpressing NSUN2 in HCT115 cells, cell migration and invasion were determined by transwell assays. scale bars = 100 μm

NSUN2 promotes PHGDH expression in CRC

To uncover the functional downstream effectors and signaling pathways perturbed by NSUN2-mediated m⁵C modification, we performed RNA-seq on NSUN2 stably knockdown HCT15 cells and combined it with MeRIP-seq data from the GEO public database (GSE226129) to evaluate potential targets (Fig. 3A and Fig. S3A-C). To validate more specific downstream target genes, we used different cell lines for verification. The results indicated that only PHGDH, a rate-limiting enzyme in the de novo serine synthesis pathway, showed the most significant reductions in both mRNA and protein expression levels following NSUN2 knockdown (Fig. 3B-D). As an oncogene, PHGDH protein and mRNA expression was significantly increased in CRC tissues compared with normal tissues (Fig. 3E-G). The Kaplan-Meier survival curve showed that PHGDH expression may be a predictor of CRC tumorigenesis (Fig. 3H). To evaluate the clinical relevance of NSUN2 and PHGDH, we analyzed the TCGA database as well as the collected matched fresh frozen primary CRC tissues and adjacent normal mucosa. The results demonstrated tumour tissues with high NSUN2 expression had higher levels of PHGDH (Fig. 3I, J). In addition, we performed IHC staining on CRC samples. The results demonstrated a significant positive correlation between NSUN2 and PHGDH (Fig. 3K, L).

Fig. 3.

NSUN2 promotes PHGDH expression in CRC. (A) Venn diagram of RNA-seq in NSUN2-stably knockdown HCT15 cells and combined it with MeRIP-seq data from the GEO public database (GSE226129) to evaluate potential targets. Differential gene analysis between samples was carried out and screened by fold-change and P value. (B) PHGDH, PLEKHG2, VGF, LAMA5, KCTD15 and C6orf141 mRNA expression in SW620 cells and HCT116 cells with NSUN2 knockdown were detected by qRT-PCR. (C) Analyzing protein expression levels of PHGDH in SW620 and HCT116 cells after infection with shNSUN2-1 and shNSUN2-2. (D) Analyzing mRNA expression levels of PHGDH in SW620 and HCT116 cells after infection with shNSUN2-1 and shNSUN2-2. E, F) PHGDH was highly expressed in tumor tissues compared with adjacent normal tissues from GSE21510 (E) and TCGA (F) databases. G) Representative IHC staining images for PHGDH protein in human CRC specimens (scale bars = 250 μm and 50 μm), demonstrating its differential expression between tumor (n = 56) and adjacent normal (n = 56) tissue. H) Kaplan-Meier analysis of OS in the Renji Hospital cohort (n = 180), comparing CRC patients with high (n = 96) versus low (n = 84) tumor PHGDH expression. I) NSUN2 were positively correlated with the expression of PHGDH at mRNA levels in TCGA databases. J) Tumor tissues and normal tissues from CRC patients were collected, and Western blot was used to analyze the levels of the NSUN2 and PHGDH proteins. K) Representative IHC staining for NSUN2 and PHGDH from CRC tissue (scale bars = 250 μm and 50 μm). Tumor 1# is representative of a patient with NSUN2-low CRC. Tumor 2# is representative of a patient with NSUN2-high CRC. L) Correlation of NSUN2 and PHGDH staining in human CRC tissue samples (n = 180). NSUN2 and PHGDH show a positive correlation. ns, non-significant

NSUN2 promotes malignant phenotypes of CRC cells via PHGDH both in vitro and in vivo

To uncover the function of the NSUN2-PHGDH axis, we then designed rescue functional assays. PHGDH overexpression restored the inhibitory effects of NSUN2 knockdown on CRC cell proliferation and colony formation (Fig. 4A-C). SW620 cells treated with or without shNSUN2 and PHGDH-WT were inoculated into the flanks of nude mice. The results indicated that NSUN2 knockdown inhibited tumor growth, reduced tumor weight, and decreased the expression levels of both Ki67 and PHGDH, while PHGDH overexpression reversed these effects (Fig. 4D-J). Altogether, PHGDH is a key target of NSUN2 that promotes CRC malignancy.

Fig. 4.

NSUN2 promotes malignant phenotypes of CRC cells via PHGDH both in vitro and in vivo

A-B) CCK8 assays were conducted to evaluate the effect of the NSUN2-PHGDH axis on the growth of SW620 (A) and HCT116 (B) cells. C) Colony formation assays were conducted to evaluate the effect of the NSUN2-PHGDH axis on the growth of SW620 and HCT116 cells. D-I) SW620 cells were stably infected with the indicated plasmids and subcutaneously injected into BALB/c-nude mice (n = 5 per group). About 3 weeks after injection, xenografts were removed. Representative images of xenografts were shown (D, G). Tumor weight (E, H) and tumor volume (F, I) were statistically analyzed and compared. J) IHC staining was implemented to detect the expression levels of NSUN2, PHGDH and Ki67 using tumor tissues harvested from xenograft model mice (scale bars = 50 μm). Quantification of staining intensity was shown (below)

Methylation by NSUN2 stabilizes PHGDH mRNA in an ALYREF-dependent manner

Next, we explored whether the oncogenic function of NSUN2 depends on its m⁵C methyltransferase activity. An enzymatically inactive double-dead mutant of NSUN2 was generated by introducing point mutations at cysteine residues 271 and 321 (Fig. S4A). Successful transfection of this mutant was confirmed through immunofluorescence and m⁵C dot blot assays (Fig. 5A-B and Fig. S4B). To enhance the credibility of our study, we treated CRC cells stably transfected with the target plasmids using cycloleucine, a widely used inhibitor of methionine adenosyltransferase 2 (MAT II) that reduces mRNA methylation by depleting methyl sources. The results of qRT-PCR and western blot assays showed that silencing NSUN2 decreased PHGDH mRNA and protein levels in CRC cells, an effect reversible by overexpressing NSUN2-WT but not NSUN2-DM, while cycloleucine reduced PHGDH mRNA and protein levels in SW620 and HCT116 cells in a concentration-dependent manner (Fig. 5C-D and Fig. S4C-D). Additionally, following knockdown of other differentially expressed m⁵C methyltransferases in CRC (as identified in Fig. S1A), Western blot analysis revealed no significant alterations in PHGDH expression levels across the knockdown conditions of these selected methyltransferases (Fig. S4E). Overexpression of NSUN2-WT and NSUN2-DM in NSUN2-silenced cells revealed that only the NSUN2-WT could rescue CRC cell proliferation, whereas the NSUN2-DM failed to do so. Furthermore, cycloleucine significantly inhibited the proliferation of CRC cells (Fig. S4F-G). Notably, MeRIP-seq data revealed that NSUN2 knockout significantly reduced m⁵C peak enrichment on the PHGDH gene (Fig. 5E). RIP assays demonstrated effective enrichment of PHGDH mRNA by the m⁵C antibody (Fig. S4H). Silencing NSUN2 decreased the m⁵C levels of PHGDH mRNA in CRC cells, an effect that could be reversed by overexpressing NSUN2-WT but not NSUN2-DM. Additionally, Cycloleucine reduced PHGDH mRNA m⁵C levels in SW620 and HCT116 cells in a concentration-dependent manner (Fig. 5F and Fig. S4I). To investigate how NSUN2-mediated m⁵C regulates PHGDH mRNA expression, we assessed the stability of PHGDH mRNA in SW620 and HCT116 cells using actinomycin D (10 µg/ml). RNA was extracted at 0, 2, 4, 6 h post-treatment and analyzed by qRT-PCR. Higher m⁵C levels of PHGDH mRNA correlated with slower degradation and greater stability (Fig. 5G and Fig. S4J), indicating that NSUN2 enhances PHGDH mRNA stability via m⁵C methyltransferase activity. Differential motif analysis of MeRIP-seq data demonstrated that WCWUC was the top-ranked m⁵C motif in HCT15 cells (p = 1.3e-24), while HCWUC predominated in NSUN2-knockdown cells (p = 1.8e-16) (Fig. 5H). These results are consistent with the previous findings reported by Lin et al. [17]. Based on the motif analysis that NSUN2-regulated m⁵C modification regions are mainly at the 5’UTR and CDS of PHGDH mRNA, we designed and constructed PHGDH-Mut luciferase reporters by replacing the specific adenosine (T) in m⁵C motif with thymine (G) based on PHGDH-WT luciferase reporter (Fig. 5I). The results of the dual-luciferase assay indicated that NSUN2 could not promote the luciferase activity of the reporter construct bearing PHGDH with mutations (Fig. 5J).

Fig. 5.

Methylation by NSUN2 stabilizes PHGDH mRNA in an ALYREF-dependent manner. (A) Immunofluorescence staining was performed to detect the NSUN2-WT or NSUN2-DM overexpression in stably transfected SW620 cells. The red color represents NSUN2 staining; the blue color represents DAPI staining. scale bars = 50 μm. (B) m⁵C dot blot assay was used to detect the m⁵C levels of mRNA extracted from stably transfected SW620 cells. C-D) Western blot (C) and qRT-PCR (D) results demonstrate that NSUN2 and PHGDH expression exhibit a significant correlation in SW620 cells treated with either different plasmids or varying concentrations of Cycloleucine. Cycloleucine was used at three concentrations: 10mM (low), 20mM (medium), and 40mM (high). E) IGV plots of m⁵C peaks in m⁵C immunoprecipitation sample (relative to input sample) at PHGDH mRNA in shNSUN2 cells (green peak) related to the control cells (blue peak). F) The results of the RIP assay in SW620 cells showed the relative m⁵C enrichment of PHGDH mRNA for each group, normalized to the IgG control. Cycloleucine was used at three concentrations: 10mM (low), 20mM (medium), and 40mM (high). G) The curve of PHGDH mRNA remaining versus time after 10 µg/mL actinomycin D treatment in SW620 cells after treatment with the indicated plasmids or 20mM cycloleucine. H) Top enriched motifs within m⁵C peaks identified in HCT5 cells with NSUN2 knowdown and scramble. I) The structure of PHGDH-WT or PHGDH-mutation (5’UTR mutation:51G; CDS mutations:1446G, 1580G, 1686G) luciferase reports were designed according to the result of MeRIP-seq. J) Relative luciferase activity of PHGDH-WT or PHGDH-mutation reporter vectors in NSUN2 knowdown cells transfected with Ctrl plasmids or NSUN2-WT plasmids. ns, non-significant

Building upon the paradigm of reader protein-mediated RNA methylation effects, we systematically evaluated two well-characterized m⁵C readers (YBX1 and ALYREF) in PHGDH regulation. Strikingly, only ALYREF depletion (not YBX1 knockdown) substantially diminished PHGDH abundance (Fig. S5A-B). Western blot analysis showed that silencing ALYREF inhibited the rescue effect of NSUN2-WT expression on NSUN2-KD cells (Fig. S5C). ALYREF, also known as Aly/REF export factor, is known to interact with modified nucleotides, such as m⁵C (5-methylcytosine), in RNA [18]. To investigate ALYREF’s regulatory role in CRC cell proliferation, we performed CCK8 assays and colony formation experiments. As shown in Fig. S5D-E, ALYREF knockdown significantly inhibited the proliferative capacity of CRC cells. RIP assays using ALYREF antibodies demonstrated that ALYREF binds to PHGDH mRNA (Fig. S5F), and this interaction was significantly weakened in NSUN2-silenced cells (Fig. S5G), suggesting that ALYREF may act as a m⁵C reader for PHGDH. Further studies revealed that knockdown of ALYREF reduced PHGDH mRNA stability. (Fig. S5H-I). Collectively, these findings suggest that NSUN2 can upregulate PHGDH expression in CRC cells by enhancing the stability of PHGDH mRNA in an ALYREF-dependent manner.

NUSN2 induces PHGDH-mediated Serine metabolism to promote CRC tumorigenesis

Given that PHGDH is the first rate-limiting enzyme for de novo serine synthesis [12, 16], we next aimed to evaluate whether m⁵C modification of PHGDH affects serine metabolism using a high-throughput targeted amino acid quantification assay. The assay indicated decreased concentrations of metabolites such as Glycine and L-Serine in the NSUN2 knockdown group (Fig. 6A). Serine metabolism is tightly linked to folate and methionine cycles, generating various metabolites that are crucial for tumor cell survival and proliferation [16]. To explore whether NSUN2 regulates serine metabolism in tumor cells via PHGDH, we demonstrated that overexpressing PHGDH in NSUN2 knockdown SW620 cells effectively rescued the significant reductions in serine and glycine pools (Fig. 6B, C). To further confirm these observations, we used U-[¹³C]-glucose to trace the de novo synthesis of serine and glycine. We observed that the incorporation of ¹³C from glucose into serine and glycine was markedly diminished by downregulating expression of NSUN2, which could be rescued by overexpressing PHGDH (Fig. 6D, E). These data indicate that NSUN2 inhibits serine synthesis by downregulating PHGDH. Serine metabolism can also support redox homeostasis. NADPH and NADH, crucial cellular reductants, maintain the redox homeostasis necessary for tumor cells to manage the high levels of ROS produced during rapid proliferation. In this study, we demonstrated reduced NADH/NAD⁺ and NADPH/NADP⁺ ratios in NSUN2 knockdown cells, which were restored by PHGDH overexpression as expected (Fig. 6F, G). Cellular levels of ROS were significantly increased in NSUN2 knockdown SW620 cells, and the re-introduction of PHGDH reversed this effect (Fig. 6H). In addition, the apoptotic rate induced by NSUN2 knockdown was also suppressed by PHGDH overexpression (Fig. 6I). Collectively, these findings suggest that NSUN2/ALYREF/m⁵C-PHGDH signaling axis plays a crucial role in the aggressiveness of CRC (Fig. 7).

Fig. 6.

NUSN2 induces PHGDH-mediated serine metabolism to promote CRC tumorigenesis. A) High-throughput targeted amino acid quantification assay was conducted between the NSUN2 KD and scramble groups in SW620. Metabolomics analysis was conducted using non-targeted LC-MS/MS. B, C) Intracellular pool levels of serine and glycine in SW620 cells and HCT116 cells after treatment with the indicated plasmids. D) Schematic of U-[¹³C]-glucose incorporation into serine and glycine in cells. E) Incorporation of U-[¹³C]-glucose carbon into serine and glycine in SW620 cells after treatment with the indicated plasmids. F, G) Measurement of NADH/NAD+ (F) and NADPH/NADP+ (G) levels in SW620 and HCT116 cells after treatment with the indicated plasmids. H) Intracellular ROS level in SW620 cells after treatment with the indicated plasmids. I) Percentage of apoptotic cells in SW620 cells after treatment with the indicated plasmids

Fig. 7.

Schematic illustration shows that NSUN2 stabilizes PHGDH mRNA, thereby enhancing its translation and increasing PHGDH protein levels. The elevated PHGDH protein promotes serine metabolism and inhibits apoptosis, ultimately driving CRC progression

Discussion

Abnormal m⁵C modification has been implicated in a variety of diseases, including cancer, inflammation [19–21], neurodevelopmental disorders [22], intellectual disabilities [23], infertility [24], and mitochondrial dysfunction [25]. In 2022, Chen et al. [26] demonstrated that the novel m⁵C clusters, comprising 18 m⁵C regulators and a scoring system, accurately reflect the distinct prognostic signature, clinicopathological characteristics, immunological phenotypes, and stratifying therapeutic opportunities in CRC. Among the 18 m⁵C regulators, NSUN2 is a well-characterized RNA methyltransferase responsible for m⁵C modification of tRNA and mRNA [3, 7–10, 27]. Multiple studies indicate that NSUN2 promotes the progression of various cancers through m⁵C modification of mRNAs such as SKIL, GRB2, FABP5, p57Kip2, PFAS, TIAM2, and HDGF, each linked to specific cancers including CRC, esophageal squamous cell carcinoma, osteosarcoma, gastric cancer, retinoblastoma, pancreatic cancer, and bladder cancer respectively [3–5, 8, 28–30].

To verify these findings, colony formation, transwell invasion, and tumor sphere formation assays in our study demonstrated that NSUN2 knockdown significantly suppressed the proliferation and invasion of CRC cells. In accordance with the in vitro results, a subcutaneous tumor model was established, and the results indicated that NSUN2 knockdown significantly inhibited subcutaneous tumor growth. By using cycloleucine to reduce mRNA methylation through depletion of methyl sources, we found that the enhanced growth of CRC induced by NSUN2 overexpression was suppressed, confirming that NSUN2-mediated tumor progression relies on its m5C methyltransferase activity. Collectively, these in vitro and in vivo findings underscore the critical role of NSUN2 in the development of CRC.

To investigate the mechanisms underlying NSUN2-mediated regulation of CRC progression, we integrated RNA-seq, m⁵C MeRIP-seq (GSE226129 dataset), and untargeted metabolomics. Our results demonstrate that PHGDH acts as an m⁵C-dependent downstream target of NSUN2. PHGDH serves as the rate-limiting enzyme in the serine biosynthesis pathway, catalyzing the conversion of 3-phosphoglycerate to 3-phosphohydroxypyruvate, and plays a central role in maintaining cellular serine and redox homeostasis [31]. Recent studies have revealed that PHGDH contributes to tumorigenesis and progression in multiple cancers through canonical metabolic pathways or non-metabolic atypical pathways—including CRC, bladder, and breast cancers [32–34].

In recent years, research has increasingly focused on the mechanisms underlying interactions between metabolic reprogramming and modification in cancer, as well as the full extent of their profound impact on tumorigenesis through abnormal crosstalk. Mounting evidence demonstrates that dysregulated crosstalk between metabolic reprogramming and post-translational modifications (PTMs) drives tumorigenesis, wherein PHGDH emerges as a critical regulatory node. Specifically, PRMT1-mediated methylation at R236 enhances PHGDH enzymatic activity to promote de novo serine biosynthesis in hepatocellular carcinoma, while in CRC, the Cullin 4 A-based E3 ubiquitin ligase complex induces PHGDH activation via K146 monoubiquitination, consequently upregulating serine synthesis. These paradigm-shifting findings collectively establish PTM-dependent PHGDH regulation as a conserved oncogenic mechanism across malignancies, illuminating targetable vulnerabilities in cancer metabolism [35, 36]. Nevertheless, pre-translational regulation of PHGDH—particularly epigenetic modifications—remains largely unexplored.

Herein, we demonstrate that NSUN2 enhances PHGDH expression by elevating m⁵C methylation on PHGDH mRNA and thereby increasing its mRNA stability. Furthermore, NSUN2 knockdown significantly reduced PHGDH expression, consequently suppressing serine metabolic processes in CRC. Typically, m⁵C-modified RNA is recognized by reader proteins, which allow it to exert functions such as enhancing RNA stability in an m⁵C-dependent manner. Notably, the upstream regulatory mechanisms of PHGDH are not limited to NSUN2-mediated RNA m5C modification. For instance, zinc finger protein 146 (ZNF146) directly regulates PHGDH expression at the transcriptional level, while mitogen-activated protein kinase 13 (MAPK13) catalyzes the phosphorylation of PHGDH at serine 371 (Ser371) [37, 38]. This phosphorylation triggers PHGDH protein degradation via the chaperone-mediated autophagy (CMA) pathway. These findings indicate that PHGDH is regulated by diverse and complex mechanisms. However, our study highlights the critical role of NSUN2-mediated RNA m5C modification in regulating PHGDH during CRC progression.

ALYREF, a well-known m⁵C-binding reader protein, enhances the mRNA stability of EGFR, which in turn activates pathways such as pSTAT3, contributing to tumor progression in hepatocellular carcinoma (HCC) [39]. Additionally, m⁵C has been demonstrated to promote mRNA export, coordinately regulated by its methyltransferase NSUN2 and its binding partner ALYREF [18]. RIP experiments confirmed that ALYREF binds to the target gene PHGDH. Furthermore, knockdown of ALYREF in tumor cells validated its regulation of PHGDH mRNA stability and protein levels, and correspondingly demonstrated significant suppression of CRC cell proliferation. This regulatory role was further substantiated through tumor xenograft experiments, confirming ALYREF’s essential function in tumor development.

Overall, in this study, we determined the elevated expression levels of the m⁵C methyltransferase NSUN2 and its oncogenic role in CRC, along with the underlying genetic and molecular mechanisms. NSUN2-induced metabolic reprogramming involves modulating the expression of PHGDH in a ALYREF-m⁵C-dependent manner, which results in enhanced serine metabolism in CRC cells. Furthermore, through the analysis of clinical samples and bioinformatics data, we further confirmed the correlation between the expression levels of NSUN2 and PHGDH in clinical tissues and established their clinical relevance to downstream regulated serine metabolism in the progression and prognosis of CRC. Taken together, we uncovered an NSUN2/ALYREF/m⁵C-PHGDH signaling axis, offering valuable insights into the pathogenesis and identification of epigenetic–metabolic targets in CRC.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

All authors read and approved the final manuscript. Conceptualization, LH, GTY and ZYH; methodology, LH, GTY and ZYH; investigation, LH and ZYH; writing, LH and GTY; resources, TSB and LY; formal analysis, LH; data curation, ZYH and LY; visualization, WTF; supervision, LHP; funding acquisition, ZM.

Funding

This work was supported by the National Natural Science Foundation of China (No. 81873555) and the Shanghai Shenkang Clinical Development Center (No. SHDC2020CR5006).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval and consent to participate

The study for human tumour samples complied with the ethical requirements of The Ethics Committee of Renji Hospital, School of Medicine, Shanghai Jiao Tong University. The study protocol of animal experiments was approved by the Animal Ethics Committee of Renji Hospital, School of Medicine, Shanghai Jiao Tong University.

Informed consent

We declare that this study obtained written informed consent from all involved individuals.