We retrieved public gene expression data and full clinical annotations of Cervical Squamous Cell Carcinoma and Endocervical Adenocarcinoma (CESC) from The Cancer Genome Atlas (TCGA) (https://portal.gdc.cancer.gov/). This study enrolled 297 patients with cervical cancer, including 245 with squamous cell carcinoma of the cervix, 48 with adenocarcinoma of the cervix, and 4 patients with other pathological types. The baseline data is shown in Supplementary Table 1. The transcripts per million (TPM) format was log2 transformed and processed for analysis using R version 4.2.2 (Lucent Technologies, Murray Hill, NJ, USA). Using the function “surv_categorize” of R package “survminer” (v0.4.9), the expression levels of NAT10 and SLC7A5 mRNA were divided into “high” or “low” groups. Survival analysis was carried out for patients with CESC using the expression data of tumors with “survminer” and the R package “survival” (v3.4.0), with ordinate representing survival rate and abscissa representing survival time. The p values were adjusted for multiple testing using the Benjamini–Hochberg method.

A cervical cancer tissue microarray (HUteS136Su01, Shanghai Outdo Biotech Co, China) was dewaxed in xylene for 20 minutes; then hydrated in different gradients of ethanol for 5 minutes each. Following this, it was subjected to high-pressure fixation with sodium citrate for 5 min followed by blocking with 5% bovine serum albumin (BSA, CCS30014.01, MRC, Cincinnati, OH, USA) for 1 hour. Slides were dried, and NAT10 monoclonal antibody (1:100, SC-271770, Santa Cruz, Dalas, TX, USA) was incubated at 4 °C overnight. Slides were incubated with the mouse secondary antibody (1:5000, SA00001-1, Proteintech, Chicago, IL, USA) at room temperature for 0.5 h. After incubation, DAB (DA1010, Beijing Solaibao Technology Co., LTD, Beijing, China) was added for 3–5 minutes to lightly counterstain with hematoxylin. Staining intensity was measured using H-score method as described previously [13]. In short, scoring was based on percentage and frequency of positive staining cells as follows: 0 indicates no positive staining cells; 1–100 indicates 1–100% of cells stained. Staining intensity was scored as follows: 0 represents no staining, 1 represents weak positive staining, 2 represents medium positive staining, and 3 representes strong positive staining.

SiHa and CaSki human cervical cancer cells were purchased from Procell Company (Wuhan, China). C33A and HeLa human cervical cancer cells were kindly donated by Professor Zhang Bingzhong. The cell lines were cultured in Dulbecco’s Modified Eagle Medium (DMEM; Gibco, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS; Procell) and 5% CO₂. All cell lines were validated by short tandem repeat (STR) analysis, and mycoplasma testing was negative.

The Cell Counting Kit-8 (CCK-8) assay (Dojindo, Kumamoto, Japan) was performed to assess proliferative potential. Cells (3 $\times$ 10³) were seeded in 96-well plates and cultured for 0, 24, 48, 72, and 96 h. Absorbance was recorded at 450 nm using microplate reader (Thermo Fisher Scientific, Waltham, MA, USA) after 2-h incubating with 10 µL CCK-8. Three independent experiments were performed.

Lentiviral vectors containing short hairpin RNA (shRNA) and the NAT10 coding sequence were purchased from Genechem (Shanghai, China). ShRNA sequence for NAT10 was: AGGGCCCTCCTTTCCTATAAG. HeLa and SiHa cells were infected with lentivirus. Small interfering RNA (siRNA; KeScience, Shanghai, China) was utilized to knock out NAT10 in cells using Lipofectamine 3000 (Invitrogen, Waltham, MA, USA). The siRNA sequence was: GCUGUGGUGUUAUAAGAAATT.

Cell monolayer was scratched using a pipette tip, and wound width at 0, 24 h was captured using the DMC4500 digital microscope (Leica, Wetzlar, Germany). Migration distance was evaluated by ImageJ software (version 1.53t, National Institutes of Health [NIH], Bethesda, Rockville, MD, USA).

Transwell assays were performed with 8 µm chambers (Falcon; Corning Inc., Corning, NY, USA). 5 $\times$ 10⁴ cells were suspended in DMEM containing 2% FBS and added to upper chambers. Lower chambers were filled with 20% FBS-DMEM. Remodelin (Selleck Chemicals, Shanghai, China) or dimethyl sulfoxide (DMSO; Sigma, St. Louis, MO, USA) was added to the upper chambers. For invasion assay, each insert was coated with 50 µL diluted Matrigel (BD Biosciences, Franklin Lakes, NJ, USA). After a 48-h incubation, unmigrated cells on inner surface of upper chambers were removed with cotton swabs. Upper chambers were fixed in 4% paraformaldehyde (Biosharp, Hefei, China) and stained with 0.1% methylene blue (Solarbio, Beijing, China) for 30 min. Images were captured and analyzed using ImageJ software (NIH). Three independent experiments were performed.

Total cell proteins were extracted, resolved by sodium dodecyl polyacrylamide gel electrophoresis (EpiZyme, Cambridge, MA, USA), and transferred electrophoretically to polyvinylidene fluoride (PVDF) membranes (MilliporeSigma, Burlington, USA). Then PVDF membranes were blocked in rapid sealing fluid (NCM Biotech, Suzhou, China), cut by predicted molecular weight, and incubated overnight at 4 °C with NAT10 monoclonal antibody or polyclonal antibodies against SCL7A5 (1:1000, 28670-1-AP; Proteintech, Chicago, IL, USA), vimentin (1:1000, 10366-1-AP; Proteintech), N-cadherin (1:1000, YT2988; Immunoway, Plano, TX, USA), GAPDH (1:20,000; AP0063; Bioworld, Minneapolis, MN, USA), and tubulin (1:30,000, AP0064; Bioworld, Irving, TX, USA). After washing, membranes were incubated at room temperature for 120 min with horseradish peroxidase (HRP)-conjugated goat anti-rabbit secondary antibody (1:5000, AP0063; Bioworld) or HRP-conjugated anti-mouse secondary antibody (1:5000, SA00001-1; Proteintech). The Tanon 5200 Chemiluminescent Imaging System (ChampChemi 610; Sage Creation, Beijing, China) was used for immunoblot analysis. The original images of the western blot are presented as Supplementary Materials.

Total RNA was extracted using TRIzol reagent (Thermo Fisher Scientific) and reverse transcribed to cDNA using PrimeScript RT Master Mix (Yeasen Biotech, Shanghai, China). Quantitative PCR (qPCR) was performed using cDNA as the template and SYBR Premix Ex Taq (Yeasen Biotech). Data were normalized to GAPDH expression. The primers for qPCR (Generay Biotech Co., Shanghai, China) were as follows: GAPDH (forward): GGAGCGAGATCCCTCCAAAAT, GAPDH (reverse): GGCTGTTGTCATACTTCTCATGG; NAT10 (forward): ATAGCAGCCACAAACATTCGC, NAT10 (reverse): ACACACATGCCGAAGGTATTG; SCL7A5 (forward): CCGTGAACTGCTACAGCGT, SCL7A5 (reverse): CTTCCCGATCTGGACGAAGC.

We employed the phenol-chloroform method to isolate and purify the total RNA from the Siha ShNAT10 vector and its OE-NAT10 cell line. RNA-seq of the Siha ShNAT10 vector and its OE-NAT10 cell line was conducted by Guangzhou Epibiotek Co., Ltd., China. Prepare Epi™ mRNA Capture Beads (Epibiotek, cat. no. R2020-96) by balancing them at room temperature for 0.5 h. One µg RNA and Beads Binding Buffer were added to the beads, then heated at 65 °C for 5 minutes and 25 °C for another 5 minutes. After magnetically clarified with the beads, the supernatant was discarded. Tris Buffer was added and heated at 80 °C for 2 minutes. We re-added Beads Binding Buffer and repeated the heating process at 65 °C and 25 °C. After the reaction was over, the mixture was magnetically clarified, and supernatant was discarded. Beads were washed with Beads Wash Buffer, and finally, Elution Buffer was added, incubated at 95 °C for 5 minutes. Supernatant was collected, which contains the mRNA elution solution. The Epi™ mRNA Library Fast kit (Epibiotek, #R1810, Guangzhou, Chia) was used to prepare the library. The library quality with the Bioptic Qsep100 Analyzer was checked to ensure the size distribution matches theoretical expectations. Library preparation was done using the VAHTS Stranded mRNA-seq Library Prep Kit for Illumina V2 (Vazyme Biotech, Nanjing, China, NR612-02). Reads were aligned to the GRCh38 human genome using Hisat2 aligner with the “rna-strandness RF” parameter. Feature counts were utilized to calculate reads mapped to genome, and DEGSeq R-package (version 1.44.0) was applied for differential gene expression analysis.

Total RNA (5 µg) was heated at 75 °C for 5 min and then cooled to 0 °C for 1 min. Then the membranes were loaded with Amersham Hybond-N+ (0.22 µm, 66485; Pall Co., Port Washington, NY, USA) and crosslinked twice with ultraviolet light. After blocking membranes in 5% skim milk (BD Biosciences, Franklin Lakes, NJ, USA), they were incubated overnight at 4 °C with anti-ac4C antibody (1:1000, ab252215; Abcam, Cambridge, UK). Membranes were washed with Tris-buffered saline with Tween 20 (NCM Biotech) and incubated at room temperature with anti-rabbit secondary antibody (1:5000, AP0063; Bioworld). Proteins were detected by incubating the membranes with a chemiluminescent HRP substrate (NCM Biotech), followed by staining with methylene blue buffer (Solarbio) for 30 min.

We used the prediction of N4-acetylcytidine (ac4C) modification sites in mRNA (PACES) tool to analyze potential acetylation sites of SLC7A5 mRNA coding sequence (http://rnanut.net/paces).

For the ac4C RNA immunoprecipitation assay, cells were lysed using NP40 (Beyotime, Jiangsu, China) at 4 °C overnight. Twenty µL of solution was stored at –80 °C. Next, 200 µL lysate was mixed with anti-ac4C antibody (1:50, ab252215; Abcam), and another 200 µL lysate was mixed with normal rabbit IgG (1:50, 30000-0-AP; Proteintech) at 4 °C for 4 h. Protein A/G beads (MedChemExpress, Monmouth Junction, NJ, USA) were incubated with lysate at 4 °C overnight. Next morning, the beads were washed with buffer. Finally, the RNA was extracted and tested by qPCR.

Actinomycin D (a transcription inhibitor) was used to evaluate RNA stability. Actinomycin D at a concentration of 200 nM was added to culture medium and co-cultured with cells as the experimental group, whereas cells cultured in complete medium served as control group. At 0, 6, and 12 h after actinomycin D treatment, cells were collected for extracting RNA. Total RNA was used for qPCR, and the level of SLC7A5 was detected at each time point.

BALB/c nude mice have a mutation in the FOXn1 gene, which leads to the absence or degeneration of the thymus and the lack of mature T cells, making them able to accept foreign transplanted cells without producing an immune rejection response. This characteristic makes BALB/c nude mice an ideal model for studying tumor growth and metastasis. BALB/c nude mouse tumor models can retain the morphological and genetic characteristics of the primary tumor, which is of great significance for studying the biological characteristics of tumors and tumor therapy. Therefore, we selected BALB/c nude mice as a xenograft model to explore the anti-tumor effects of Remodelin in vivo. All animal experiments followed rules of the Ethics Committee on Animal Experiments of Zhujiang Hospital (Guangzhou, China). Female mice (BALB/c, 5–6 weeks old, 16–19 g; Gongsimingzi, Zhaoqing, China) were housed under pathogen-free conditions. Animal experiments were performed as previously reported [31]. Briefly, 12 mice were subcutaneously injected in the right axillary fossa with SiHa cells (2 $\times$ 10⁶) in 100 µL phosphate-buffered saline. Ten mice developed tumors. Mice were randomized into two groups: remodelin (5 mg/kg) (Selleck) or DMSO (Sigma) vehicle (control group) administered intraperitoneally every 3 days. We measured the size of the subcutaneous transplant tumors with calipers every other day. Tumor volume (mm³) = 0.5 $\times$ long diameter $\times$ short diameter². All mice were anesthetized by intraperitoneal injection of 80 mg/kg pentobarbitone (Cat# DORMINAL 20%; Macklin, Shanghai, China. The stock concentration is 200 mg/ml, and is diluted with saline to a concentration of 16 mg/ml, and administered at 5 ml/kg to give a dose of 80 mg/kg) after almost 3 weeks of treatment and sacrificed by cervical dislocation after being photographed. Their neoplasms were dissected.

Fertilized eggs (Eggsiao, Suzhou, China) were developed for 10 days. Then they were laid on their side and the shell were punctured on the air bag side (the top side). After sucking the air from the airbag, the chicken chorioallantoic membranes (CAMs) were dropped. Next, 100 µL cell suspension containing 1,000,000 cells was inoculated per CAM. Tumors were collected after incubation at 37 °C in an incubator for 6 days.

Data were analyzed using GraphPad Prism 9 software (GraphPad Software Inc., San Diego, CA, USA). Data are presented as the mean $\pm$ standard deviation. t-test was used to compare differences between two groups, if data followed a normal distribution. For more than three groups, one- or two-way analysis of variance followed by Bonferroni’s post-hoc tests were used. If the data did not follow a normal distribution or were categorical, non-parametric tests such as the Wilcoxon rank-sum test or chi-square test were applied. To ensure statistical power, each experiment was repeated three times. Survival curves were analyzed using the Kaplan–Meier method. Statistical significance was set at p 0.05.

We found that NAT10 mRNA high expression was closely correlated with shorter disease-free survival (DFS) and overall survival (OS) of patients with CESC in TCGA (Fig. 1A,B). Clinical characteristics of patients were shown Supplementary Table 1. Furthermore, association between NAT10 mRNA expression and OS of patients with cervical adenocarcinoma was assessed, and high expression of NAT10 mRNA was correlated with shorter OS of patients with cervical adenocarcinoma (Supplementary Fig. 1). To further explore NAT10 expression and its association with patient prognosis, a commercial cervical cancer tissue chip was used to detect NAT10 protein levels by immunohistochemistry (IHC). The chip included 110 cervical cancer tissues and 25 adjacent non-cervical cancer tissues. Results showed NAT10 protein expression in nuclei and cytoplasm of cervical cancer cells (Fig. 1C), which was significantly higher in cervical cancer tissues than in normal ones (Fig. 1C,D). Kaplan–Meier curves revealed that patients with high NAT10 expression had lower OS and DFS (Fig. 1E,F), suggesting that NAT10 might predict poor prognosis. Interestingly, we found that NAT10 was overexpressed in older patients compared with young patients (Table 1), consistent with evidence showing involvement NAT10’s involvement in age-related disorders [32]. However, we found that NAT10 was not correlated with lymph node metastasis, pathological stage, or pathological grade (Table 1).

Fig. 1.

NAT10 is overexpressed in cervical cancer. (A) Effect of NAT10 expression level on DFS of patients with cervical cancer based on TCGA dataset. (B) Effect of NAT10 expression level on OS of patients with cervical cancer based on TCGA dataset. (C) Representative IHC images of cervical microarrays showing NAT10 expression in cervical tumors and corresponding normal tissues. Scale bar for the images is 20$\times$, 100 µm. (D) NAT10 expression in cervical cancer and adjacent tissues. (E) Effect of NAT10 on OS of patients with cervical cancer. (F) Effect of NAT10 on DFS of patients with cervical cancer. TGCA, the cancer genome atlas; IHC, immunohistochemistry; TPM, transcripts per million; OS, overall survival; DFS, disease-free survival; NAT10, N-acetyltransferase 10. Note: *p 0.05.

To evaluate NAT10’s role in cervical cancer, four cervical cancer cell lines (SiHa, CaSki, C33A, and HeLa) were used to detect NAT10 expression. NAT10 was detected in all of these cell lines and was overexpressed in Caski and SiHa cells (Fig. 2A). ShNAT10 and SiNAT10 were used to repress NAT10 expression in CaSki and SiHa cells, respectively. NAT10 expression was significantly reduced after NAT10 knockdown (Fig. 2B,C). The proliferative ability of SiHa and CaSki cells was decreased after silencing NAT10 (Fig. 2D). Wound healing assays showed that NAT10 downregulation slowed wound healing in these cells (Fig. 2E). Silencing of NAT10 also decreased the migratory and invasive abilities of SiHa and CaSki cells (Fig. 2F,G). Subsequently, we found that NAT10 knockdown attenuated cervical cancer cell proliferation in vivo in CAMs (Fig. 2H,I).

Fig. 2.

Knockout of NAT10 expression impaired proliferative and metastatic potentials of cervical cancer cells. (A) Expression of NAT10 in four cervical cancer cell lines. (B) Knockdown efficiency of NAT10 determined by Western blotting. (C) Validation of NAT10 mRNA knockdown by qPCR (n = 3). (D) NAT10 silence impaired proliferative potential of SiHa (n = 4) and CaSki cells (n = 3). (E) NAT10 silence inhibited wound healing of SiHa (n = 3) and CaSki cells (n = 3). Scale bar = 200 µm. (F) Cell migratory capacity was reduced after NAT10 knockdown) (n = 3). Scale bar = 200 µm. (G) Knockdown of NAT10 expression reduced the invasion capability of SiHa (n = 3) and CaSki cells (n =3). Scale bar = 200 µm. (H) Representative images of tumors in CAMs after knockdown of NAT10 expression in SiHa cells. (I) Knockout of NAT10 expression in SiHa cells inhibited the proliferation of tumors in CAMs (n = 3–5). NAT10, N-acetyltransferase 10; ShNC, negative control short hairpin RNA; ShNAT10, short hairpin RNA targeting NAT10; SiNC, negative control small interfering RNA; SiNAT10, small interfering RNA targeting NAT10; OD, optical density; CAMs, chicken chorioallantoic membranes; qPCR, quantitative PCR. Note: *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001.

We explored the role of NAT10 by overexpressing NAT10 in a SiHa NAT10-knockdown cell line (ShNAT10 cell line) using the lentivirus method. NAT10 expression in the SiHaShNAT10 overexpression (OE) cell line was significantly increased after transfection with OE lentivirus (Fig. 3A,B). Cell proliferative ability was significantly enhanced after NAT10 overexpression (Fig. 3C). Moreover, overexpression of NAT10 promoted wound healing in SiHa cells (Fig. 3D). NAT10 overexpression promoted SiHa cell metastasis and invasion in transwell assays (Fig. 3E,F). We also overexpressed NAT10 in HeLa cervical cancer cells. Similarly, NAT10 expression in the HeLa-NAT10 OE cell line was increased after transfection with OE lentivirus (Fig. 3A,B). Together, NAT10 overexpression promoted proliferation, wound healing, invasion, and metastasis abilities of HeLa cells (Fig. 3C–F), indicating that NAT10 is an oncogene in cervical cancer.

Fig. 3.

Upregulation of NAT10 expression in SiHa and HeLa cells promoted proliferative and metastatic potentials of cervical cancer cells. (A) NAT10 protein overexpression as confirmed by Western blotting. (B) Determination of NAT10 mRNA overexpression by qPCR (n = 3). (C) Promotion of cell proliferation after overexpressing NAT10 in SiHa (n = 5) and HeLa cells (n = 3). (D) Effect of NAT10 overexpression on migration of SiHa (n = 3, scale bar = 200 µm) and HeLa cells (n = 3, scale bar = 200 µm) was evaluated using a wound healing assay. (E) Migration capability of SiHa and HeLa cells overexpressing NAT10 as evaluated by transwell assays (n = 3, scale bar = 200 µm). (F) Invasion capability of SiHa and HeLa cells overexpressing NAT10 as evaluated by transwell assays (n = 3, scale bar = 200 µm). NAT10, N-acetyltransferase 10; OE-NAT10, NAT10 overexpression; OD, optical density. Note: *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001.

NAT10 acts as an acetyltransferase that can catalyze ac4C modification on mRNA, thereby improving mRNA stability and translation efficiency [13, 17, 33]. We performed an acetylation dot blot assay to investigate whether NAT10 exerts its biological effects by acetylating mRNA in cervical cancer cells. Inhibiting NAT10 in SiHa cells significantly reduced the mRNA ac4C acetylation level (Fig. 4F), consistent with previous reports [13, 17]. Increased ac4C acetylation level was detected when NAT10 was overexpressed in SiHa cells (Fig. 4F).

Fig. 4.

Exploration of the impact of NAT10 on downstream factors. (A) Transcriptomic comparison was carried out to analyze the dysregulated genes between the vector control and OE-NAT10 (NAT10 overexpressing) SiHa cells. (B) Representative pathway analysis terms according to Gene Ontology analysis. (C) Representative pathway analysis terms according to Kyoto Encyclopedia of Genes and Genome analysis. (D) Scatter plot of TPM values of NAT10 and SLC7A5 mRNA in TCGA-CESC data. Spearman’s correlation analysis showed the association between NAT10 mRNA and SLC7A5 mRNA expression levels (log2 [TPM + 1]) (R = 0.26, p = 4.3 $\times$ 10^-6). (E) Prediction conserved acetylation sites in the SLC7A5 coding sequence in PACES tools (http://rnanut.net/paces/). (F) Detection of ac4C acetylation on mRNA of cervical cancer cells using an ac4C dot blot assay. NAT10, N-acetyltransferase 10; SLC7A5, solute carrier family 7 member 5; TPM, transcripts per million; TCGA, the cancer genome atlas; CESC, cervical squamous cell carcinoma; ac4C, N4-acetylcytidine; PACES, prediction of N4-acetylcytidine (ac4C) modification sites in mRNA.

To decipher mechanisms by which NAT10 affects cervical cancer cell progression, RNA-seq was performed on NAT10-overexpressing SiHa and control cells. The sequencing results showed that 1146 genes were upregulated and 1097 genes were downregulated after NAT10 overexpression in SiHa cells (Fig. 4A). The top 10 significantly enriched gene ontology terms are summarized in Fig. 4B. The results showed that NAT10 was related to several biological processes such as “the C21-steroid hormone metabolic process”, “progesterone metabolic process”, “homophilic cell adhesion via plasma membrane adhesion molecules”, and “carboxylic acid transport”. Kyoto encyclopedia of genes and genomes pathway enrichment analysis indicated that the differentially expressed genes were significantly enriched in “ATP-binding cassette transporters”, “neuroactive ligand-receptor interactions”, and “protein digestion and absorption” (Fig. 4C).

SLC7A5 upregulation is correlated with increased activity in the carboxylic acid transport and organic anion transport pathways [23]. Based on data from TCGA website, NAT10 mRNA expression was positively correlated with SLC7A5 mRNA expression (Fig. 4D). NAT10 can reportedly function by acetylating target mRNA [34]. Therefore, we consulted an acetylation site prediction website, which showed that there are acetylation sites at nucleotides 120 and 134 of SLC7A5 mRNA, with a specificity of 99% (Fig. 4E). Consequently, we speculate that SLC7A5 mRNA is a potential downstream target of NAT10.

By analyzing the public database, SLC7A5 mRNA was highly expressed in cervical cancer tissues (Fig. 5A). Overexpression of SLC7A5 was associated with a low OS rate in cervical cancer (Fig. 5B), both in squamous cell carcinoma (Fig. 5C) and adenocarcinoma (Fig. 5D). Moreover, patients with high NAT10 and SLC7A5 mRNA expression had worse OS than those with low levels according to bioinformatics analysis (Fig. 5E), both in squamous cell carcinoma (Fig. 5F) and adenocarcinoma (Fig. 5G). We also found that samples with overexpressed NAT10 and SLC7A5 had a higher degree of M2 macrophage infiltration and lower degree of M1 macrophages (Fig. 5H). Knockdown of SLC7A5 (Fig. 5I,J) in SiHa cells significantly reduced wound healing (Fig. 5K), migration, and invasion abilities (Fig. 5L).

Fig. 5.

SLC7A5 is a carcinogenic factor in cervical cancer. (A) SLC7A5 mRNA was overexpressed in cervical cancer compared with normal cervical tissue. (B) Overexpression of SLC7A5 mRNA corelated with poor OS in cervical cancer. (C) Overexpression of SLC7A5 mRNA corelated with poor OS in squamous cell carcinoma of the cervix. (D) Overexpression of SLC7A5 mRNA corelated with poor OS in cervical adenocarcinoma. (E) Patients with high NAT10 and SLC7A5 mRNA expression had worse OS than those with low levels according to bioinformatics analysis. (F) Patients with high NAT10 and SLC7A5 mRNA expression in cervical squamous cell carcinoma had worse OS than those with low levels according to bioinformatics analysis. (G) Patients with high NAT10 and SLC7A5 mRNA expression in cervical adenocarcinoma had worse OS than those with low levels according to bioinformatics analysis. (H) Differences in abundance of 10 immune cell types enriched in the TCGA-CESC dataset based on quanTIseq algorithm; blue indicates the “low NAT10 & low SLC7A5” group, red indicates the “high NAT10 & high SLC7A5” group; the horizontal axis indicates the 10 immune cell types, and the vertical axis the abundance of immune cell infiltration. (I) Validation of SLC7A5 mRNA knockdown level by qPCR in SiHa cells (n = 3). (J) Knockdown efficiency of SLC7A5 determined by Western blotting in SiHa cells. (K) Knockdown of SLC7A5 expression inhibited the wound healing of SiHa cells (n = 3, scale bar = 200 µm). (L) Cell migratory and invasive capacities of SiHa cells were reduced after SLC7A5 knockdown (n = 3, scale bar = 200 µm). OS, overall survival; NAT10, N-acetyltransferase 10; SLC7A5, solute carrier family 7 member 5; TCGA-CESC, the cancer genome atlas cervical squamous cell carcinoma and endocervical adenocarcinoma; qPCR, quantitative PCR; ns, not significant. Note: *p 0.05; **p 0.01; ***p 0.001, ****p 0.0001.

Next, qPCR was performed for verification purposes. SLC7A5 was significantly decreased after knockdown of NAT10 but was significantly increased after overexpression of NAT10 (Fig. 6A). Western blot analysis showed similar results (Fig. 6B). The acetylated RNA immunoprecipitation-qPCR results further showed that NAT10 downregulation in cervical cancer cell lines significantly decreased acetylation level of SLC7A5 mRNA (Fig. 6C). The results of the mRNA stability assay showed that knocking down NAT10 expression reduced SLC7A5 mRNA stability (Fig. 6D).

Fig. 6.

NAT10 affected the biological behavior of cervical cancer cells by ac4C acetylation of SLC7A5 mRNA. (A) Validation of the mRNA level of SLC7A5 by qPCR after regulating NAT10 expression (n = 3). (B) Detection of SLC7A5 protein levels by western blotting after regulating NAT10 expression in cervical cancer cells. (C) Detection of SLC7A5 mRNA level by acetylated RNA immunoprecipitation-qPCR after knocking down NAT10 in SiHa cells (n = 3). (D) Reducing NAT10 expression in SiHa decreased the stability of SLC7A5 mRNA (n = 3). (E) Determination of the knockout efficiency of SLC7A5 mRNA by qPCR in SiHa OE NAT10 cell lines (n = 3). (F) Determination of the knockout efficiency of SLC7A5 by western blotting in SiHa OE NAT10 cell lines. (G) The cell migration capability, which increased by NAT10 overexpression, decreased to a certain extent after SLC7A5 knockdown, as determined by wound healing assays (n = 3, scale bar = 200 µm). (H) The cell migration/invasion capability, which increased after NAT10 overexpression, decreased to a certain extent after knocking down SLC7A5 by transwell assays (n = 3, scale bar = 200 µm). SLC7A5, Solute carrier family 7 member 5; ac4C, N4-acetylcytidine; NAT10, N-acetyltransferase 10; OE, overexpression; ShNC, Negative control short hairpin RNA; ShNAT10, short hairpin RNA targeting NAT10; SiSLC, small interfering RNA targeting SLC7A5; qPCR, quantitative PCR. Note: *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001.

We speculated that NAT10 exerts its biological function by acetylating SLC7A5 mRNA. To further explore SLC7A5’s role, we silenced SLC7A5 in cervical cancer NAT10overexpressing stable cell lines (Fig. 6E,F). The results showed that after downregulating SLC7A5, the wound healing ability, which increased by NAT10 overexpression, decreased to a certain extent (Fig. 6G). Migration capacity of cervical cancer cells was also reduced to some extent following SLC7A5 downregulation (Fig. 6H).

Remodelin is a NAT10 inhibitor [35, 36]. Thus, we used remodelin to determine if NAT10 could be used as a therapeutic target for cervical cancer. CCK-8 assays revealed that remodelin significantly attenuated SiHa, CaSki, and C33A cell proliferation (Fig. 7A). Remodelin significantly weakened the wound healing ability of Siha, CaSki, and C33A cells compared to controls (Fig. 7B–D). Remodelin decreased migratory (Fig. 7E–G) and invasive abilities of SiHa, CaSki, and C33A cells (Fig. 7H–J). These results further confirmed that NAT10 is a potential target for cervical cancer, and that remodelin is a promising therapeutic drug that might attenuate cervical cancer progression.

Fig. 7.

Remodelin, inhibitor of NAT10, impaired proliferative and mestatic potentials of cervical cancer cells. (A) Remodelin impaired proliferation of cervical cancer cells (SiHa, CaSki, and C33A) as shown by CCK-8 assays (n = 3). (B) Remodelin impaired the migration capability of SiHa cells by wound healing assays (n = 3, scale bar = 200 µm). (C) Remodelin impaired the migration capability of CaSki cells as shown by wound healing assays (n = 3, scale bar = 200 µm). (D) Remodelin impaired the migration capability of C33A cells as shown by wound healing assays (n = 3, scale bar = 200 µm). (E) Remodelin impaired the migration capability of SiHa cells as shown by transwell assays (n = 3, scale bar = 200 µm). (F) Remodelin impaired the migration capability of CaSki cells as shown by transwell assays (n = 3, scale bar = 200 µm). (G) Remodelin impaired the migration capability of C33A cells as shown by transwell assays (n = 3, scale bar = 200 µm). (H) Remodelin impaired the invasion capability of SiHa cells as shown by transwell assays (n = 3, scale bar = 200 µm). (I) Remodelin impaired the invasion capability of CaSki cells as shown by transwell assays (n = 3, scale bar = 200 µm). (J) Remodelin impaired the invasion capability of C33A cells as shown by transwell assays (n = 3, scale bar = 200 µm). CCK-8, cell counting kit-8; NAT10, N-acetyltransferase 10. Note: *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001, ns, not significant.

We found that remodelin attenuated growth and reduced the weight of cervical tumors (Fig. 8A–C). IHC showed that Ki67 expression (marker of tumor proliferation) was reduced after remodelin treatment (Fig. 8F). Western blotting showed that remodelin significantly decreased N-cadherin and vimentin levels (Fig. 8E), indicating that remodelin inhibits the epithelial–mesenchymal transition in cervical cancer tumors in vivo. More importantly, treatment with remodelin had little effect on organ damage in mice (Fig. 8G), indicating its safety. SLC7A5 expression was reduced after remodelin treatment in vivo by qPCR and IHC (Fig. 8D,F).

Fig. 8.

Remodelin prevented cervical cancer growth in vivo. (A) Tumor images of a xenograft nude mouse model subcutaneously injected with SiHa cells and treated with remodelin or NC. (B) Weight of tumors in the remodelin and NC groups at the experiment endpoint. (C) Growth rate of the tumors with or without remodelin treatment. (D) Validation of SLC7A5 mRNA levels in samples with or without remodelin treatment. (E) Expression of vimentin and N-cadherin after treatment with remodelin or NC as assessed by Western blotting. (F) Representative images for H&E and IHC staining of the Ki-67 (marker of proliferation) and SLC7A5 in nude mouse sections. Scale bar = 20$\times$, 100 µm. (G) Representative H&E images of lung/liver/spleen/kidney from mice treated with remodelin or from the NC group. Scale bar is 20$\times$, 100 µm. H&E, hematoxylin and eosin; IHC, immunohistochemistry; SLC7A5, solute carrier family 7 member 5; NC, negative control; Note: n = 5, *p 0.05, ***p 0.001, ****p 0.0001.