The RNA methyltransferase NSUN2 catalyzes 5-methylcytosine (m5C) on IL1B mRNA to promote transcript stability

TL;DR

This study shows that the enzyme NSUN2 adds a specific RNA modification that stabilizes a key inflammation-related gene, potentially offering new ways to treat inflammatory diseases.

Contribution

The study identifies NSUN2-mediated m5C modification as a novel regulator of inflammation through stabilization of IL1B mRNA.

Findings

NSUN2 catalyzes m5C modification on IL1B mRNA, promoting its stability.

Nsun2 deficiency reduces inflammation in systemic and local inflammatory models.

m5C modifications are recognized by YBX1, enhancing IL1B expression and inflammation.

Abstract

Among the plethora of RNA modifications, 5-methylcytosine (m5C) has held substantial roles in some biological processes, yet its impact on inflammation remains largely uncharted. Here, we reported an m5C-related epitranscriptomic regulatory axis in inflammatory pathogenesis. Bacterial lipopolysaccharides (LPS) challenge induces marked elevation in RNA m5C abundance as well as the expression of NOP2/Sun RNA methyltransferase 2 (NSUN2), an essential enzyme catalyzing m5C modification. Nsun2 deficient mice showed significantly reduced inflammatory response in the models of systemic inflammation and local pulpitis compared with wild-type control mice. Mechanistically, NSUN2 installs m5C modifications on interleukin 1 beta (IL1B) mRNA. These modifications are selectively recognized by the m5C reader Y-box binding protein 1 (YBX1), thereby enhancing transcript stability and promoting substantial elevation of IL1B, which might further drive inflammation cascades. Our research highlights a pivotal role of NSUN2-mediated m5C RNA modification in regulating inflammatory responses, offering fresh perspectives for the treatment of inflammatory diseases.

Figures

Full text

Bacterial infections can provoke inflammation, leading to systemic disorders like sepsis or localized pathologies such as pulpitis. Bacterial virulence factors, particularly lipopolysaccharides (LPS), are pivotal in activating the host's immune system and inducing inflammatory responses (1, 2). The hyperactivation of the inflammatory response to infection is frequently characterized by excessive production of cytokines, potentially culminating in a “cytokine storm” phenomenon. This pathological state may lead to secondary cytotoxicity and tissue damage, further exacerbating the progression of inflammation. Notably, interleukin-1 beta (IL1B) emerges as a key proinflammatory cytokine that is significantly upregulated during infections and serves as a driver of cytokine cascades (3). For example, central administration of IL1B has been demonstrated to significantly elevate circulating levels of interleukin 6 (IL6) and serum corticosterone, thereby aggravating inflammation in the central nervous system (4). Therefore, unveiling the underlying mechanisms that regulate the upregulation of cytokines is critical not only for curbing the hyperactivation of inflammatory responses but also for the advancement of therapeutic strategies against inflammation-related diseases in the future.

Recent developments highlight the significance of RNA modifications, such as N6-methyladenosine (m^6^A), in multiple physiological and pathophysiological processes. Among the numerous RNA modifications identified, 5-methylcytosine (m^5^C) is a long-known RNA modification (5) and has gained increasing attention in recent years. m^5^C RNA modification, mainly catalyzed by the DNA methyltransferase member 2 (DNMT2) and NOL1/NOP2/sun (NSUN) methyltransferase family, has been identified to regulate the stability (6), nuclear-cytoplasmic shuttling (7), and translation (8, 9) of mRNA. Existing research recognizes the critical role played by NSUN2, which is the main methyltransferase that can install m^5^C on mRNA (10). Followed by the modification mediated by NSUN2 on mRNA, the specific mRNA m^5^C-binding proteins, such as Y-box binding protein 1 (YBX1) and Aly/REF export factor (ALYREF), can read these modifications, inducing varieties of biological effects (6, 7). Abnormal RNA m^5^C modification has been found in many pathophysiological processes, such as tumor development (11, 12), viral infection (8, 13), autoimmunity disease (14), and neurological abnormalities (15). Despite these findings, the specific roles of m^5^C modifications and their regulatory genes in inflammation, particularly inflammation induced by microbial infections, remain profoundly elusive and warrant further investigation.

In this study, we sought to elucidate the roles and underlying mechanisms of m^5^C methyltransferase NSUN2 in the progression of inflammation. To investigate the association between NSUN2 and inflammation, Nsun2 knockout mice were employed to construct experimental model of local pulp inflammation and LPS-induced systemic inflammation. Utilizing human dental pulp cells (DPCs) as an in vitro model of inflammation, we investigated whether NSUN2 silencing or overexpression would affect the expression of inflammatory factors. Mechanistically, the m^5^C methylation site on target gene was identified by bisulfite treatment combined with Sanger sequencing. Furthermore, we delved into the potential of any putative m^5^C readers acting in concert with NSUN2 to modulate m^5^C modifications on target gene. Generally, we define the m^5^C-modulator NSUN2/YBX1 as a master regulator of IL1B-driven inflammation, offering new therapeutic paradigms.

Results

Nsun2 reduction effectively mitigates the systemic inflammation in vivo

To investigate the role of RNA m^5^C modification in inflammation, we initially measured the levels of m^5^C in total RNA extracted from mouse lung tissues following LPS intraperitoneal injection. Strikingly, LPS treatment resulted in a global increase in m^5^C abundance within the total RNA of inflammatory mouse lung tissues (Fig. 1A). To delve deeper into the underlying mechanism behind this substantial elevation in m^5^C abundance, we systematically profiled the gene expression of RNA m^5^C-associated methyltransferase and demethylase expressions in the pulmonary inflammatory tissues in three common mice systemic inflammation models: LPS treatment (GSE217695 (16)), cecal microbiome treatment (GSE239388 (17)), as well as cecal ligation and puncture treatment (CLP) (GSE179554 (18)) from the Gene Expression Omnibus (GEO) database. The heatmap analysis demonstrated a markedly elevated expression of m^5^C writers NOP2 nucleolar protein (Nop2) and Nsun2 in the context of pulmonary inflammation in mice as compared to the control cohort (Fig. 1, B and C). Given that NSUN2 has been recognized as the predominant mRNA m^5^C methyltransferase (19), we then focused on NSUN2. Collectively, these results showed an increase in m^5^C abundance and NSUN2 expression during inflammation, indicating that m^5^C modification mediated by NSUN2 might play a role in the inflammatory process.Figure 1**Nsun2 reduction decelerates the advancement of systemic inflammation in mice.**A, the m^5^C methylation level in total RNA of mouse lung tissues administrated with LPS was determined by m^5^C dot blot assay. Methylene blue (MB) staining was set as an internal reference for sample loading. The histogram summarized the ratios of the dot intensities of m^5^C versus methylene blue. Data are means ± SD, n = 6. B and C, published RNA-seq data derived from other researchers was downloaded from the GEO database. We reanalyzed the expression profiles of m^5^C modification-associated genes in lung tissues from mice with systemic inflammation. These systemic inflammation mouse models were conducted by three common methods: LPS treatment (GSE217695), cecal microbiome (CM) treatment (GSE239388), as well as cecal ligation and puncture treatment (CLP) (GSE179554). B, heatmap represents normalized gene expression (red, high expression; blue, low expression). C, the histogram summarized the relative expression of Nsun2. Data are means ± SD, n = 3 or 4. D, schema for the strategy used to generate the Nsun2 knockout mice via the CRISPR/Cas9 technology. E, PCR identification of genotypes of C57BL/6N wild-type (WT) mice and Nsun2 mutant (heterozygous) mice. F, determine the protein expression of NSUN2 in lung tissues of WT or Nsun2^+/−^ mice by IHC. The scale bar represents 100 and 50 μm (high magnification). G, timeline for LPS-induced systemic inflammation as well as body weight measurements and behavioral assessment (15 mice per group). H, weight changes of Nsun2^+/−^ and control WT mice before injection (0 h) and 16, 18, 20, 22, and 24 h after LPS or PBS injection. The significance of the weight change differences between WT + LPS group and Nsun2^+/−^ + LPS group was analyzed and plot on the weight change polyline of Nsun2^+/−^ + LPS group. Weight change (%) = (weight at a specific time-weight at 0 h)/weight at 0 h × 100% (15 mice per group). I, the statistical analysis of the M-CASS scores was measured at 24 h after LPS administration (n = 15). J, representative H&E staining of mouse lung sections from each group. The degree of lung damage was measured via the lung injury scoring system (n = 15). The scale bar represents 100 and 50 μm (high magnification). K, quantitative real-time PCR analysis of the mRNA levels of Il1b, Il6, Cxcl10, and Ccl2 in lung tissues. Gapdh served as a loading control. Data are means ± SD, n = 15. Trdmt1, tRNA aspartic acid methyltransferase 1; Tet, tet methylcytosine dioxygenase; Alkbh, alkB homolog 1. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. CCL2, C-C motif chemokine ligand 2; CXCL10, C-X-C motif chemokine ligand 10; GEO, Gene Expression Omnibus; IHC, immunohistochemical; IL1B, interleukin 1 beta; LPS, lipopolysaccharides; M-CASS, mouse clinical assessment score for sepsis; m5C, 5-methylcytosine; NSUN, NOP2/Sun RNA methyltransferase.

To further evaluate the role of NSUN2 in inflammation, Nsun2 knockout mice were employed. The whole-body Nsun2 knockout mice were engineered using the CRISPR/Cas9 technology, which introduced a mutation comprising the deletion of 1684 base pairs within the exon 4 of the Nsun2 gene and its flanking regions (Fig. 1, D and E). However, we found that the homozygous knockout mice (Nsun2^−/−^) experienced embryonic lethality, while the heterozygous knockout mice (Nsun2^+/−^) were viable and reached adulthood. Given this developmental constraint, we selected Nsun2^+/−^ mice in the following research. Compared to their wild-type (WT) littermates, Nsun2^+/−^ mice exhibit no significant differences in terms of body weight and stature (Fig. S1, A–C). Immunohistochemical (IHC) results substantiated a reduction in NSUN2 protein levels in lung tissues of Nsun2^+/−^ mice compared with wild-type (WT) littermates, further validating a successful knockdown of Nsun2 (Fig. 1F).

To further assess NSUN2’s functional role in inflammatory responses, age-matched Nsun2^+/−^ mice and WT controls were employed to develop LPS-induced systemic inflammation sustained over a 24-h period (Fig. 1G). Notably, compared with WT controls, a milder systemic inflammatory response was observed in Nsun2^+/−^ mice after LPS administration, evidenced by less pronounced weight loss (Fig. 1H, WT + LPS group versus Nsun2^+/−^ + LPS group) as well as lower clinical scores (mouse clinical assessment score for sepsis, mouse clinical assessment score for sepsis (M-CASS), (20)) (Fig. 1I, WT + LPS group versus Nsun2^+/−^ + LPS group). Histological analysis further exhibited that LPS administration elicited conspicuous lung injury in mice, characterized by augmented alveolar wall thickness, pervasive interstitial inflammatory cell infiltration and alveolar collapse (Fig. 1J, WT + PBS group versus WT + LPS group). Strikingly, in comparison with WT controls, Nsun2 reduction led to a less severe disruption of the lung tissues in response to LPS treatment (Fig. 1J, WT + LPS group versus Nsun2^+/−^ + LPS group). Further investigation into the molecular underpinnings of this phenomenon involved the assessment of proinflammatory cytokines (IL1B and IL6) and chemokines (C-X-C motif chemokine ligand 10 [CXCL10] and C-C motif chemokine ligand 2, [CCL2]) in lung tissues, which have been established to be persistently induced during experimental systemic inflammation (17, 21). As expected, the mRNA levels of Il1b, Il6, Cxcl10, and Ccl2 in lung tissues were markedly elevated after LPS treatment, while Nsun2 deficiency effectively attenuated LPS-induced upregulation of these cytokines (Fig. 1K). Together, these results suggest that NSUN2 deficiency attenuates inflammatory cascades, indicating its potential critical positive role in inflammation.

NSUN2 deficiency alleviates the pulpitis in vivo

The observed modulation of inflammation by NSUN2 in the context of LPS-induced systemic inflammation provides a compelling foundation for further exploration of its role in inflammatory conditions. Given the established link between NSUN2 and inflammatory responses, our investigation turned to pulpitis, a disease characterized by localized inflammation in dental pulp due to the invasion of oral bacterial. DPCs, the primary constituents of dental pulp, are capable of recognizing bacteria and their metabolites via surface pattern recognition receptors such as TLRs and NODs, playing indispensable roles in regulating the innate immune response (22). In addition, owing to their tractability in culture and transfection, DPCs have been selected as an ideal cellular model for our study. The m^5^C dot blot assay results revealed that LPS treatment resulted in a global increase in m^5^C abundance in DPCs’ RNA as well (Fig. 2A). Subsequently, we examined the expression of RNA m^5^C-associated methyltransferases in DPCs with or without LPS treatment. NSUN2 mRNA expression significantly increased after LPS treatment (Fig. 2B). Western blot analysis confirmed concomitant augment of NSUN2 protein expression in DPCs after LPS treatment (Fig. 2C). To further characterize the role of NSUN2 in pulpitis, we collected the normal and inflamed human dental pulp samples. Dental pulp samples of pulpitis showed conspicuously increased expression of NSUN2 at the protein level and mRNA level, analyzed by IHC staining (Fig. 2, D and E) and real-time quantitative reverse transcription PCR (RT-qPCR) (Fig. 2F), respectively, which indicate that NSUN2 might play a crucial role in the progression of pulpitis.Figure 2**Nsun2 reduction effectively mitigates pulpitis in vivo.**A, the m^5^C methylation level in total RNA of DPCs treated with LPS was determined by m^5^C dot blot assay. Methylene blue staining was set as an internal reference for sample loading. The histogram summarized the ratios of the dot intensities of m^5^C versus methylene blue. Data are means ± SD, n = 5. B and C, total RNA and protein were extracted from DPCs treated with LPS. B, RT-PCR analyzed the mRNA expression of all known mammalian m^5^C RNA methyltransferases (NSUN1-NSUN7 and DNMT2) and inflammatory cytokines (IL-1β and IL-6). β-actin served as a loading control. Data are means ± SD, n = 3. C, the protein expression of NSUN2 was analyzed by western blot, and GAPDH served as a loading control. Data are means ± SD, n = 5. D, the schematic representation on the right exhibited the pathological characteristics of pulpitis resulting from dental caries. Representative IHC staining of NSUN2 expression in normal (n = 6) and pulpitis (n = 5) teeth was shown on the left. The H&E stained images below showed typical normal pulp and inflamed pulp. P: pulp tissue; D: dentine. The scale bar represents 50 μm. E, the expression of NSUN2 in dental pulp detected by IHC was quantified as the average optical density (AOD) using ImageJ software and IHC Toolbox plugin. F, measurement of NSUN2, IL1B and IL6 mRNA level analyzed by RT-qPCR in normal (n = 7) and pulpitis (n = 8) teeth. β-actin served as a loading control. G, determine the protein expression of NSUN2 in dental pulp tissues of WT or Nsun2^+/−^ mice by IHC. The scale bar represents 100 and 50 μm (high magnification). H, the illustration delineated the construction of a mouse model for pulpitis. I, the micro-CT scanning of the pulp exposures in both the unexposed pulp group (anesthesia-only control group) and the pulp-exposed group. Arrows indicate the sites of pulp exposure. The scale bar represents 400 μm. J and K, HE staining (J) and quantification of the inflammatory cells infiltrated in pulp tissue (K) in Nsun2^+/−^ and WT mice from control, 0 and 12-h post exposure. Data are means ± SD, n = 10 (representing bilateral first molars from five mice per group). The scale bar represents 400 and 50 μm (high magnification). PTAU, pulp tissue area unit. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. DPC, dental pulp cell; IL-6, interleukin 6; IL1B, interleukin 1 beta; LPS, lipopolysaccharides; m5C, 5-methylcytosine; micro-CT, micro-computed tomographic; NSUN, NOP2/Sun RNA methyltransferase; RT-PCR, semiquantitative reverse transcription PCR; RT-qPCR, real-time quantitative reverse transcription PCR.

To further assess the in vivo role of NSUN2 in pulpitis, an experimental pulpitis model was established in Nsun2^+/−^ and WT mice through pulp exposure, with pathological changes analyzed at 0 and 12 h following the procedure (Fig. 2, G–I). As revealed by HE staining, in both the control normal dental pulp (without pulp exposure) and the 0-h post exposure groups, odontoblasts orderly and tightly surrounded the dentin, with well-organized collagen fibers in the extracellular matrix and a near-complete absence of inflammatory cell infiltration (Fig. 2J, control group and 0-h group). However, at 12-h post exposure, significant infiltration of inflammatory cells and vascular extravasation emerged in the dental pulp area (Figs. 2J, 12-h group). Notably, in comparison to the WT group, the dental pulp of the Nsun2^+/−^ group exhibited significantly less inflammatory damage, characterized by the more attenuated disruption and disarray of odontoblasts, alongside a more localized area of inflammatory edema with a substantially diminished infiltration of inflammatory cells as well as less pronounced tissue necrosis (Fig. 2, J and K, WT 12-h group versus Nsun2^+/−^ 12-h group). Collectively, these data suggest NSUN2 knockdown exerts an inhibitory effect on the development of pulpitis in vivo.

NSUN2 promotes the expression of inflammatory cytokines and chemokines at the mRNA level in DPCs

Mirroring systemic inflammatory responses, the progression of pulpitis developed from caries is usually mechanistically accompanied by the accumulation of inflammatory cytokines such as IL1B, IL6, and chemokines such as CXCL10, CCL2 within affected tissues, which orchestrates immune cell recruitment and promotes tissue destruction (23). To elucidate the proinflammatory effect of NSUN2, the expression of these cytokines and chemokines of DPCs overexpressing NSUN2 was analyzed by semiquantitative reverse transcription PCR (RT-PCR). As expected, NSUN2 overexpression alone significantly promoted the expression of IL1B and IL6 as well as CXCL10 and CCL2 at the mRNA level (Fig. 3, A–C). LPS markedly enhanced the mRNA expression of IL1B, IL6, CXCL10, and CCL2 in DPCs (Fig. 3, D, E and G). Notably, the combination of NSUN2 overexpression and LPS treatment further amplified the mRNA induction of IL1B, IL6, CXCL10, and CCL2 compared to LPS alone (Fig. 3, D–F), suggesting that NSUN2 potentiates LPS-driven inflammatory responses. However, NSUN2 silencing attenuated the capacity of LPS in inducing the expression of these critical cytokines and chemokines (Fig. 3, G and H). Together, these results demonstrated that the induction of these cytokines and chemokines by LPS is, at least in part, in an NSUN2-dependent manner.Figure 3**NSUN2 is responsible for the upregulation of inflammatory cytokines and chemokines in DPCs.**A–C, DPCs were transfected with either NSUN2-3 × Flag or empty vector (control) plasmids. The transcriptional levels of IL1B and IL6 (A), as well as CXCL10 and CCL2 (B), were detected by RT-PCR. β-actin served as a loading control. Data are means ± SD, n = 3 or 4. C, western blot confirmed the overexpression efficiency of NSUN2. The anti-Flag antibody was used to detect the Flag-tagged fusion protein. GAPDH served as a loading control. D–F, DPCs were treated with LPS or PBS combined with transfection with either NSUN2-3 × Flag or empty vector (control) plasmids. The transcriptional levels of IL6 and IL1B (D), as well as CCL2 and CXCL10 (E), were detected by RT-PCR. β-actin served as a loading control. Data are means ± SD, n = 3. F, western blot confirmed the overexpression efficiency of NSUN2. The anti-Flag antibody was used to detect the Flag-tagged fusion protein. GAPDH served as a loading control. G and H, DPCs were treated with LPS or PBS combined with transfection with either specific siRNAs targeting NSUN2 (siN2-1#, siN2-2#) or negative control siRNA (siNC). G, RT-PCR analyzed the mRNA levels of inflammatory factors (IL1B, IL6, CXCL10, and CCL2) in DPCs. The accompanying histograms on the right summarized the expression levels of IL1B, IL6, CXCL10, and CCL2 upon NSUN2 silencing in LPS stimulated DPCs. β-actin served as a loading control. Data are means ± SD, n = 3 or 4. H, knockdown efficiency of siNSUN2 (siN2-1# and siN2-2#) was confirmed by western blot. GAPDH served as a loading control. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. CCL2, C-C motif chemokine ligand 2; CXCL10, C-X-C motif chemokine ligand 10; DPC, dental pulp cell; IL1B, interleukin 1 beta; LPS, lipopolysaccharides; m5C, 5-methylcytosine; NSUN, NOP2/Sun RNA methyltransferase; RT-PCR, semiquantitative reverse transcription PCR; siRNA, small interfering RNA.

NSUN2 catalyzes m5C methylation of IL1B mRNA and is essential for its mRNA stability

NSUN2 has been recognized as an RNA m^5^C methyltransferase and can modulate RNA stability through catalyzing m^5^C modification (24). In the present study, NSUN2 silencing resulted in a marked reduction whereas its overexpression induced a substantial increase in the m^5^C modification levels of total cellular RNA of DPCs (Fig. 4, A and B), which validated the RNA m^5^C methyltransferase role of NSUN2. To mechanistically link NSUN2-mediated methylation to inflammatory factor expression, we first identified the m^5^C sites on the target mRNAs utilizing the bisulfite treatment that converted the unmethylated C base to T base (Fig. 4C). As a result, through bisulfite treatment combined with single-clone sequencing, an m^5^C modification site within the mRNA of IL1B was identified, specifically located at the 86th cytosine (C86) base after the stop codon (Fig. 4D). Moreover, we quantified the methylation frequency among sequenced IL1B mRNA clones. Bisulfite sequencing revealed methylation at C86 in 69.4% of clones (34/49) across four independent experiments (range: 44.4%–89.5%) (Table S2). The overall data support the dynamic methylation at C86. Notably, NSUN2 overexpression significantly increased the m^5^C methylation frequency at this locus (Fig. 4E), indicating that NSUN2 positively regulates m^5^C deposition at C86. To further validate the effect of this m^5^C methylation site on mRNA expression, we cloned the 3′ untranslated region (UTR) of IL1B containing this site into the pEGFP-C1 plasmid downstream the GFP expression gene (IL1B-m^5^C-wt) and performed site-directed mutagenesis at this m^5^C site (IL1B-m^5^C-mt, Fig. 4F). Transfection of these two plasmids into cells was followed by RT-qPCR analysis to evaluate the transcriptional levels of the GFP-IL1B 3′ UTR fusion gene. Our results revealed that the mutation at the m^5^C site in the 3′ UTR of IL1B resulted in a significant reduction in the expression of the GFP transcript, indicating the importance of this modification site in mRNA expression (Fig. 4G). Subsequently, cells transfected with these two plasmids were treated with actinomycin D to inhibit transcription, followed by the quantification of the remaining mRNAs transcribed from these two plasmids to analyze their half-life and stability. The results showed that the GFP-IL1B 3′ UTR mRNA transcribed from “IL1B-m^5^C-mt” showed significantly decreased stability, with shortened half-life, compared with mRNA transcribed from “IL1B-m^5^C-wt” (Fig. 4H), indicating the crucial role of this m^5^C site at C86 in maintaining mRNA stability.Figure 4**NSUN2 catalyzes the m^5^C modification of IL1B mRNA and stabilizes its mRNA.**A and B, dot blot analysis indicated the global m^5^C level in DPCs transfected with siNSUN2 and siNC (A) as well as NSUN2-3 × Flag and control vector plasmids (B). Methylene blue staining was set as a loading control. The histogram summarized the ratios of the dot intensities of m^5^C versus methylene blue. Data are means ± SD, n = 2 or 4. C, schematic diagram of m^5^C modification sites on mRNA analyzed by bisulfite treatment and sequencing. Total RNA sample from DPCs was treated with bisulfite conversion reagent which converted unmethylated cytosines (C) into uracil (U) while the methylated cytosines remained unchanged as C. Then the RNA samples were reverse transcribed, and the genes of interest were amplified followed by Sanger-sequencing to detect methylated C bases. D, an m^5^C modification site on IL1B mRNA was validated by Sanger sequencing, located at the 86th cytosine base following the stop codon. Representative data from a single clone were shown. E, bisulfite sequencing analysis shows the methylation frequency at the 86th cytosine of IL1B mRNA in control and NSUN2-overexpressing DPCs. Data are means ± SD, n = 4. F, diagram showing the m^5^C site near the stop codon in the IL1B 3′ UTR, along with the construction of the plasmids containing either the wild-type (IL1B-m^5^C-wt) or mutant (IL1B-m^5^C-mt) m^5^C site. G, after transfecting HEK 293 cells with the mutant plasmids and the wild-type control plasmids mentioned above, the relative expression of GFP was measured using RT-qPCR. The histogram shows the results of the statistical analysis of the four experiments. To eliminate discrepancies caused by transfection, the transcriptional level of Neo^r^ gene in the vector plasmid served as a loading control. H, RT-qPCR analysis of the stability of GFP-IL1B 3′ UTR mRNA transcribed from IL1B-m^5^C-wt and IL1B-m^5^C-mt expression plasmids. *β-*actin served as a loading control. Data are means ± SD, n = 4. I, diagram of the expression plasmids for wild-type NSUN2 (NSUN2-WT) and a catalytically inactive mutant variant (NSUN2-DM). J, RT-qPCR analysis of the IL1B mRNA level in wild-type and mutant NSUN2 overexpressed DPCs. Empty vector was used as a negative control. β-actin served as a loading control. Data are means ± SD, n = 3. K, western blot confirmed the overexpression efficiency of NSUN2. The anti-Flag antibody was used to detect the Flag-tagged fusion protein. GAPDH served as a loading control. L, m^5^C RIP-PCR analyzed the m^5^C modification on IL1B mRNA after NSUN2 knockdown. M, RT-qPCR analysis of the mRNA stability of IL1B in DPCs transfected with siNC or siNSUN2. β-actin served as a loading control. Data are means ± SD, n = 4. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. DPC, dental pulp cell; IL1B, interleukin 1 beta; m5C, 5-methylcytosine; NSUN, NOP2/Sun RNA methyltransferase; RT-qPCR, real-time quantitative reverse transcription PCR.

Given the elevated NSUN2 expression and m^5^C abundance upon LPS-treatment (Fig. 1, A–C, and Fig. 2, A–F) and its essential roles in LPS-induced expression of inflammatory factors (Fig. 3), we next hypothesized that NSUN2 functions as a critical molecular amplifier of inflammatory signaling through catalyzing the m^5^C modification of IL1B mRNA. To test this hypothesis, we next constructed expression plasmids for both wild-type NSUN2 (NSUN2-WT) and a catalytically inactive mutant variant (NSUN2-DM) (Fig. 4I, the latter harboring dual point mutations at both the releasing site (cysteine 271) and the catalytic site (cysteine 321) (7, 25). The results revealed that the overexpression of NSUN2-WT, but not the NSUN2-DM, led to an increase in IL1B mRNA expression (Fig. 4, J and K). Meanwhile, utilizing m^5^C-IP and PCR techniques, we have conclusively demonstrated that NSUN2 knockdown significantly diminishes the enrichment of m^5^C methylation on IL1B mRNA (Fig. 4L), underscoring the essential role of NSUN2 in the m^5^C modification in IL1B mRNA. We then delved into the mechanism by which reduced NSUN2 downregulates IL1B mRNA. According to mRNA stability assay, depletion of NSUN2 leads to decreased IL1B mRNA stability (Fig. 4M). Collectively, these results suggest that NSUN2 facilitates the m^5^C methylation modification of IL1B mRNA, a process essential for maintaining its mRNA stability.

The m5C reader YBX1 is highly expressed in pulpitis and promotes the expression of inflammatory cytokines and chemokines in DPCs

The biological consequences of RNA m^5^C modification are conducted by m^5^C readers that specifically recognize and bind to the modified sites, leading to the subsequent regulation of RNA expression. Among the three known mammalian RNA m^5^C readers, YBX1 has been identified to maintain the stability of its targeted mRNA (6) while ALYREF mediates mRNA nuclear-cytoplasmic shutting (7) and YTHDF2 mainly modulates rRNA (26). Given the NSUN2-dependent stabilization of IL1B mRNA uncovered in our study, we sought to identify which of the aforementioned readers participates in the recognition of m^5^C sites in inflammation. We found that the mRNA level of YBX1 was remarkably elevated in LPS-treated DPCs, whereas the expression of the other two readers (ALYREF and YTHDF2) remained largely unchanged (Fig. 5A). IHC staining showed a significantly higher level of YBX1 in pulpitis compared with normal pulp samples (Fig. 5, B and C). Consistently, the mRNA level of YBX1 was substantially upregulated in pulpitis (Figs. 2F and 5D) as well. These results proved the increase of YBX1 in pulpitis, indicating its potential participation in the inflammatory process.Figure 5**YBX1, acting as an m^5^C reader, collaborates with NSUN2 to facilitate the expression of inflammatory cytokines in DPCs.**A, total RNA was purified from DPCs stimulated with LPS. RT-PCR analyzed the mRNA expression of the m^5^C reader (YBX1, YTHDF2, and ALYREF) and inflammatory cytokines (IL-1β and IL-6). β-actin served as a loading control. Data are means ± SD, n = 3. B, representative IHC staining of YBX1 expression in normal (n = 6) and pulpitis (n = 5) teeth. The scale bars represent 50 μm. P: pulp tissue; D: dentine. C, the expression of YBX1 in dental pulp detected by IHC was quantified as the average optical density (AOD) using ImageJ software and IHC Toolbox plugin. D, measurement of YBX1 mRNA expression analyzed by RT-qPCR in normal (n = 7) and pulpitis (n = 8) teeth. β-actin served as a loading control. E–G, DPCs were transfected with either Myc-tagged YBX1 or empty vector (control) plasmids. The transcriptional levels of IL1B and IL6 (E), as well as CXCL10 (F) and CCL2 (G), were detected by RT-PCR. β-actin served as a loading control. Data are means ± SD, n = 3 or 4. H, western blot confirmed the overexpression efficiency of YBX1. The anti-Myc antibody was used to detect the Myc-tagged fusion protein. GAPDH served as a loading control. I–K, DPCs were treated with LPS or PBS after transfection with either siYBX1 or siNC. I, knockdown efficiency of siYBX1 was confirmed by western blot. GAPDH served as a loading control. J and K, RT-PCR was employed to illustrate the effect of YBX1 knockdown on the mRNA levels of IL1B and CXCL10 (J) as well as IL6 and CCL2 (K) in DPCs. β-actin served as a loading control. Data are means ± SD, n = 3 or 4. L–N The interaction between YBX1 and IL1B transcripts were analyzed by RNA immunoprecipitation (RIP) using anti-YBX1 antibody, with normal rabbit IgG as the isotype control. L and N, immunoprecipitated RNA was analyzed by RT-PCR. L, endogenous IL1B transcripts bound by YBX1. N, exogenous GFP-IL1B transcripts bound by YBX1. The histogram below showed the percentage of GFP-ILB mRNA co-immunoprecipitated with YBX1 relative to Input levels. Data are means ± SD, n = 3. M and N, immunoprecipitated YBX1 protein was confirmed by western blot. O, the mRNA half-life of IL1B in DPCs transfected with siYBX1. siNC was used as a negative control and β-actin served as a loading control. Data are means ± SD, n = 4. P–R, DPCs with or without NSUN2 overexpression were transfected with siYBX1 or siNC. The expression levels of IL1B and CCL2 (P), as well as IL6 and CXCL10 (Q), were detected by RT-PCR. β-actin served as a loading control. Data are means ± SD, n = 3 or 4. R, the overexpression of NSUN2 and the knockdown efficiency of siYBX1 were analyzed by western blot. GAPDH served as a loading control. S, RIP-PCR analysis of the interaction between YBX1 and IL1B mRNA in NSUN2 knockdown and control DPCs. T, schema of the mechanism by which NSUN2 collaborating with YBX1 promotes the progression of dental pulp inflammation. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. ALYREF, Aly/REF export factor; CCL2, C-C motif chemokine ligand 2; CXCL10, C-X-C motif chemokine ligand 10; DPCs, dental pulp cells; IL1B, interleukin 1 beta; LPS, lipopolysaccharides; m5C, 5-methylcytosine; NSUN, NOP2/Sun RNA methyltransferase; RT-PCR, semiquantitative reverse transcription PCR; RT-qPCR, real-time quantitative reverse transcription PCR; YBX1, Y-box binding protein 1.

YBX1 recognizes the m5C modification on IL1B mRNA and stabilizes its mRNA via an NSUN2-dependent manner

To further explore the role of YBX1 in inflammation, we detected the mRNA level of inflammatory cytokines IL1B and IL6 (Fig. 5E), as well as chemokines CXCL10 and CCL2 (Fig. 5, F and G), following YBX1 overexpression in DPCs (Fig. 5H). Indeed, YBX1 significantly increased the expression of these factors related to inflammation. Meanwhile, in cells with YBX1 silencing, the effect of LPS in inducing the expression of IL1B, IL6, CXCL10, and CCL2 was remarkably diminished (Fig. 5, I–K). Overall, these results confirmed that YBX1 participated in the inflammatory process by positively modulating the expression of inflammatory cytokines and chemokines.

Subsequently, we sought to explore the role of YBX1, acting as an m^5^C reader, in NSUN2-mediated m^5^C modification on IL1B mRNA. Given its function as an RNA binding protein, YBX1 is capable of binding to mRNA and mediating m^5^C-regulated mRNA stability (27). An anti-YBX1 antibody was used to immunoprecipitate YBX1-bound mRNAs from human DPCs. The results showed that YBX1 protein can bind to the mRNA of IL1B (Fig. 5, L and M). To determine whether the binding of YBX1 protein to IL1B mRNA is dependent on the m^5^C modification at C86, we performed further RNA immunoprecipitation (RIP) assay. Cells transfected with “IL1B-m^5^C-wt” and “IL1B-m^5^C-mt” expression plasmids were collected, followed by co-immunoprecipitation with anti-YBX1 antibody. The results showed that the anti-YBX1 antibody could pull down both the “IL1B-m^5^C-wt” and “IL1B-m^5^C-mt” transcripts, yet the mutant RNA showed sharply diminished YBX1 binding (Fig. 5N), indicating that the identified m^5^C site is critical for YBX1 recognition. Mechanistic analysis revealed that YBX1 depletion accelerated the decay of IL1B mRNA, mirroring the NSUN2 knockdown phenotype (Fig. 5O). The above findings lead to a speculation that YBX1 and NSUN2 cooperatively stabilize IL1B mRNA through m^5^C modifications. To validate this hypothesis, we conducted experiments to assess this synergistic effect. The results demonstrated that while NSUN2 overexpression enhanced the mRNA expression levels of various inflammatory cytokines and chemokines (IL1B, CCL2, IL6, and CXCL10) in DPCs, concomitant YBX1 silencing abrogated this effect (Fig. 5, P–R). This suggests a YBX1-dependent manner of NSUN2 in regulating the mRNA of the above factors. Furthermore, the YBX1-RIP-PCR assay demonstrated that NSUN2 ablation dramatically reduced the binding between YBX1 and IL1B mRNA, further confirming the NSUN2-dependent regulation of IL1B by YBX1 (Fig. 5S). These results collectively revealed a potential hierarchical regulatory axis wherein NSUN2 catalytically establishes m^5^C modifications on IL1B mRNA, which are subsequently recognized and stabilized by the m^5^C reader YBX1 (Fig. 5T).

Discussion

In the present study, we delved into the role of RNA m^5^C modifications during inflammatory processes, revealing that m^5^C modification served as a positive determinant in the regulation of inflammatory responses. Utilizing Nsun2 heterozygous knockout mice, coupled with an LPS-induced systemic inflammation model and a pulpitis model initiated by dental pulp exposure, we discovered that NSUN2 reduction significantly alleviated inflammatory damage. Mechanistically, we uncovered that the m^5^C writer NSUN2 catalyzed the m^5^C modification on IL1B mRNA, which was further recognized by the m^5^C reader YBX1, thereby enhancing the mRNA stability of IL1B as well as its expression at the transcriptional level. This intricate interplay between NSUN2 and YBX1 underscores the critical roles of m^5^C regulators play in the posttranscriptional regulation of gene expression, ultimately influencing the inflammatory responses through the modulation of cytokine levels. This revelation provides intriguing insights into how m^5^C and its regulatory proteins orchestrate inflammation.

Recently, the emerging field of chemical modification of RNA, also termed “epitranscriptome”, has revolutionized our understanding of posttranscriptional regulation due to their widely spread as well as indispensable roles in the regulation of gene expression. The most extensive and in-depth research has been on m^6^A. For instance, researchers have found that the m^6^A reader IGF2 mRNA-binding protein 2 (IGF2BP2) participated in promoting cytokine-induced kidney autoimmunity through the recognition of m^6^A on CCAAT/enhancer binding proteins (C/EBPs), thereby triggering IL-17-mediated inflammatory response (28). Moreover, in mice with bacterial-related systemic inflammation, the m^6^A demethylase alkB homolog 5 (Alkbh5) mRNA level significantly decreased in the spleen, kidney, and brain, and Alkbh5-deficient mice exhibited more severe inflammatory responses (29). In contrast to the well-characterized m^6^A machinery, the biological significance of m^5^C in infection-associated inflammation remains poorly understood. Although NSUN2 has been reported to be a positive regulator of sterile vascular endothelial inflammation via methylating ICAM-1 mRNA (9), and modulate the function of group 3 innate lymphoid cells (ILC3) in autoimmune diseases (30), the role of m^5^C in bacterial-driven inflammatory responses remains largely unknown. Our study revealed the elevated abundance of RNA m^5^C modification and the upregulation of NSUN2 in bacterial infection, in which NSUN2 acts as a key proinflammatory factor.

Bacterial infectious diseases have been identified as a major global heath burden, accounting for substantial morbidity and mortality worldwide. Although antibiotics can suppress bacterial proliferation to some extent, effective therapeutic strategies for managing infection-induced local and systemic inflammatory responses remain elusive. And the current main treatment method to control sustained inflammation in pulpitis is irreversibly removing pulp tissues, which weaken the treated teeth due to the loss of physiological and defensive functions of the pulp and make teeth more susceptible to fracture. Notably, pathogen-triggered cytokine upregulation frequently initiates excessive inflammatory cascades, exacerbating tissue damage (31). Therefore, curbing cytokine upregulation during infections is critical for mitigating inflammation-induced tissue injury and preserving tissue integrity. In this study, we reported the m^5^C modification mediated by NSUN2 and recognized by YBX1 on the core proinflammatory cytokine-IL1B, which enhanced the stability of its mRNA. Given the elevated NSUN2 and YBX1 expression in inflammatory tissues, this epitranscriptomic regulatory mechanism might help explain the upregulation of IL1B in inflammatory microenvironment. Meanwhile, our results demonstrated Nsun2 deficient mice exhibited milder systemic and local inflammation in response to similar bacterial-related stimulus compared with WT control mice. Our research provides a new potential target, centered around NSUN2/IL1B, for inflammation treatment and preliminarily provided supports to verify the validity.

Anticytokine therapy, a therapeutic approach that involves inhibiting or neutralizing the activity of inflammatory cytokines, has shown great potential in the treatment of a variety of diseases (32, 33). For instance, in the management of coronavirus disease 2019 (COVID-19), the cytokine storm is an essential issue. Research has shown that cytokine inhibitors such as Anakinra, targeting IL-1, can effectively attenuate the cytokine storm in patients with severe and critical COVID-19 (34). However, given the crucial roles of appropriate concentrations of cytokines in activating immune cells, treatment targeting cytokines might induce excessive immunosuppression and increase the risk of infections caused by other pathogenic microorganisms (35). Meanwhile, the limitations of targeting a single cytokine may lead to an inability to suppress the inflammatory cascade triggered by other types of inflammatory factors. Our study provides a potential strategy to circumvent these risks. Firstly, the dynamic and reversible character of m^5^C modification, along with its multisite nature and the involvement of multiple regulatory genes, collectively ensure that targeting a single modification site or regulatory axis will not result in complete loss of downstream genes. Second, the ubiquitous presence of m^5^C modifications implies that suppressing the function of upstream regulatory genes mediating m^5^C modifications may lead to moderate downregulation in the expression of multiple inflammation-related genes.

Experimental procedures

Clinical specimen collection

The study was approved by the Ethics Committee of Hospital of Stomatology, Wuhan University (LUNSHENZI 2022A14), in accordance with the principles of the Declaration of Helsinki (36). Written informed consent was obtained from every enrolled participant. The collected teeth were categorized into the healthy pulp group and the pulpitis group based on clinical assessment and diagnosis. Part of the samples was fixed in 4% paraformaldehyde (PFA), then demineralized in a 10% ethylenediaminetetraacetic acid (EDTA) solution, followed by dehydration and paraffin embedding. Pulp tissues were collected from the remaining samples and preserved in RNA later solution (Ambion), followed by total RNA extraction.

Cell culture

Human DPCs were separated and cultivated employing an enzymatic method previously described (37). In brief, dental pulp tissue was gently isolated from the pulp chamber and canals, and then washed with phosphate-buffered saline (PBS, Servicebio) containing 2% antibiotic–antimycotic (Gibco). After being minced into fragments, the tissue was digested in a solution with 3 mg/ml collagenase type I (Sigma-Aldrich) and 3 mmol/L CaCl_2_ for 20 min at 37 °C. Following digestion, the tissue was cultured in minimum essential medium alpha modification (α-MEM, Gibco) supplemented with 20% fetal bovine serum (Gibco) and 2% antibiotic–antimycotic (Gibco) at 37 °C with 5% CO_2_. All experiments were performed with cells after the third passage (P3).

Human embryonic kidney (HEK) 293 cells were cultured in Dulbecco’s modified Eagle medium (DMEM, HyClone) supplemented with 10% fetal bovine serum and 1% antibiotic-antimycotic. HEK 293 cells were obtained and authenticated as previously described (38). All cells were tested negative for mycoplasma contamination.

RNA extraction, semiquantitative reverse transcription PCR and real-time quantitative reverse transcription PCR

Total RNA was purified by Total RNA Miniprep Kit (Axygen) and quantified by NanoDrop 2000 (Thermo Fisher Scientific). Isolated RNA was processed with DNase I (Thermo Fisher Scientific) to eliminate potential DNA contamination and was reversely transcribed using Maxima H Minus cDNA Synthesis Master Mix (Thermo Fisher Scientific). Semiquantitative PCR and quantitative real-time PCR were performed with Green Taq Mix (Vazyme) and ChamQ Universal SYBR qPCR Master Mix (Vazyme), respectively. Primers involved in this section were listed in Table S1. Image J was used to quantify the relative transcript level.

Western blot

Total cellular protein was lysed with 2 × SDS sample buffer, and then denatured at 95 °C for 3 min. The protein samples were then separated by 4 to 12% SDS-PAGE gel (GenScript) and transferred to nitrocellulose (NC) membranes (Pall Corporation). After blocked in 5% defatted milk, the membranes were incubated with the following antibodies: rabbit monoclonal anti-NSUN2 (#ab259941, 1:2000, Abcam), rabbit monoclonal anti-YBX1 (#ab76149, 1:2000, Abcam), mouse monoclonal anti-YBX1 (#TA806283, 1:1000, Origene Technologies), rabbit anti-FLAG antibody (#20543, 1:2000, Proteintech), rabbit anti-MYC antibody (#10828, 1:2000, Proteintech), and mouse anti-GAPDH (#sc-47724, 1:1000, Santa Cruz Biotechnology). Image J was used to quantify the relative protein expression.

Plasmids and transfection

Plasmid NSUN2-3 × FLAG containing a full length of human NSUN2 open reading frame (ORF) was amplified from DPCs using the primers 5′ GGCTATGGGGCGGCGGTCG 3′ and 5′ TCACCGGGGTGGATGGACC 3′. After being added with the restriction sites, the resulting PCR product was subcloned into the p3 × FLAG-CMV-14 vector at EcoRⅠ and KpnⅠ sites. For stable overexpression of NSUN2, the ORF containing the 3 × FLAG tag was then cloned into EcoRⅠ and SpeⅠ sites of pLVX-IRES-Puro plasmid. The mutant NSUN2 plasmid (NSUN2-DM) was generated by introducing point mutations into WT plasmid (pLVX-NSUN2-3 × FLAG-Puro). The releasing (cystine 271) and catalytic (cystine 321) sites were mutated by overlapping PCR with the following primers: 5′ ATGTGATGTCCCTTACAGTGGAGACGGCA 3′ and 5′ TGCCGTCTCCACTGTAAGGGACATCACAT 3′ for cystine 271 mutation, and 5′ GATGGTGTATTCCACGTATTCACTAAACCCTA 3′ and 5′ TAGGGTTTAGTGAATACGTGGAATACACCATC 3′ for cystine 321 mutation.

To validate the m^5^C modification on IL1B mRNA, IL1B 3′ UTR containing the methylation site (cytosine 86 downstream from the stop codon) was inserted into the pEGFP-C1 vector (named IL1B-m^5^C-wt) using the following primers: 5′ TAAAGAGAGCTGTACCCAGAGAGTCC 3' and 5′ CTTCAGTGAAGTTTATTTCAGAACCA 3'. Utilizing IL1B-m^5^C-wt as a template, IL1B-m^5^C-mt with the cytosine (C, methylated site) mutated to adenine (A) was generated by overlapping PCR with the following primers: 5′ AAAGGTTTTTGAGTAAGGCTATAGCCTGGACTT 3′ and 5′ AAGTCCAGGCTATAGCCTTACTCAAAAACCTTT 3′.

Plasmid expressing Myc-tagged YBX1 was obtained from Dr Thomas Tuschl (Addgene plasmid#19878) as a gift (39).

DPCs were transfected with plasmid by employing Lipofectamine 3000 (Invitrogen). Briefly, the transfection reagent was prepared according to the manufacturer’s protocol and added to the plates as soon as the DPCs seed into plates at a density of 50%–70%.

The lentiviral expression plasmids of NSUN2-3 × FLAG or empty control vector was cotransfected with psPAX2 and pMD2.G into HEK 293T cells in the presence of Lipofectamine 2000 (Invitrogen) to produce lentiviral particles.

RNAi and transfection

Small interfering RNAs (siRNAs) targeting human NSUN2 and YBX1 were synthesized by Sangon Biotech. The sequences of NSUN2 siRNA-1 and siRNA-2 were 5′ CACGUGUUCACUAAACCCUAU 3′ (siNSUN2-1) and 5′ GAGAUCCUCUUCUAUGAUC 3′ (siNSUN2-2), respectively. The sequence of anti-YBX1 siRNA was 5′ GGUCAUCGCAACGAAGGUU 3′. Nonspecific (NS) siRNA was used as a control. DPCs were transfected with siRNA in the presence of Lipofectamine 3000 (Invitrogen) by the manufacturer’s instructions.

Mice

Nsun2^+/−^ mice were derived from C57BL/6N mice employing CRISPR/Cas-mediated genome engineering scheme (Mice Nsun2 Gene ID: 28114) in Cyagen Biotechnology Co, Ltd The wild-type and Nsun2 gene trap alleles were distinguished by polymerase chain reaction (PCR) with One Step Mouse Genotyping Kit (Vazyme). Primers used for PCR were as follows: CAAACTCAGAGAGACAATCCCCTC (F1) and CCTCTAATAGTCACCTTCCCTCAC (R1), CAAACTCAGAGAGACAATCCCCTC (F2) and ATTAATCTCTGTGTTGGCACTGAC (R2).

Induction of LPS-induced systematic inflammatory response and experimental pulpitis in mice

LPS-induced systemic inflammation was conducted in Nsun2^+/−^ and WT mice (aged 8–10 weeks, 15 mice in each group) following i.p. injection with 20 mg/kg in a total volume of 100 μl of LPS (Sigma-Aldrich). Control mice received an equivalent volume of phosphate-buffered saline (PBS; Hyclone). Subsequently, the mice were weighed and scored according to the M-CASS system at 2-h intervals and were sacrificed after 24 h (20). These animal experiments were conducted under guidelines approved by the institutional Animal Ethics Committee, Hospital of Stomatology, Wuhan University (protocol code S07922070D).

To establish the pulpitis model in mice, 8 to 10 weeks old Nsun2^+/−^ and WT counterparts were employed. After anesthesia, the pulp of the bilateral maxillary first molars was exposed with #one-fourth dental round burs and endodontic hand files (0.15-mm diameter tip, 2% taper, 21 mm) on the occlusal surface. The exposed pulp was then left open to the oral environment for 0 and 12 h (Each group consists of five mice, and 2 M in each mouse were used for treatment and evaluation. Total 10 M in each group). Mice that were anesthetized only without pulp exposure served as controls. After removal of surrounding soft tissue, Maxillae were collected and fixed with 4% paraformaldehyde (PFA, Servicebio) for 48 h followed by microcomputed tomographic (micro-CT) analysis. These animal experiments were conducted under guidelines approved by the institutional Animal Ethics Committee, Hospital of Stomatology, Wuhan University (protocol code S0792201010).

Lung histology and inflammatory injury score analysis

The mouse lungs were fixed in 4% PFA, embedded in paraffin, sliced in 4 μm-thickness and then stained with hematoxylin-eosin (H&E). The morphologic changes of lung injury were measured based on a designated scoring system published by the American Thoracic Society (40). Five independent variables are included in the scoring system: neutrophils in the alveolar space, neutrophils in the interstitial space, hyaline membranes, proteinaceous debris filling the airspaces, and alveolar septal thickening.

Micro-CT analysis

The mouse maxillae were subjected to imaging using a micro-CT scanner (PerkinElmer). The exposed preparation cavity was evaluated using the Simple Viewer software from PerkinElmer.

Histological analysis of mouse pulpitis samples

Following the micro-CT analysis, the maxillae containing teeth were rinsed with PBS and then decalcified in 10% EDTA for 6 weeks. Subsequently, the samples were dehydrated, cleared in xylene, paraffin-embedded, and sectioned at 4 μm for histological analysis. The analysis of the HE staining results was conducted in accordance with the method described in a previous research (41). The results were presented as the number of infiltrating inflammatory cells per pulp tissue area unit (PTAU).

Immunohistochemistry staining

IHC was conducted using a Dako EnVision FLEX kit according to the manufacturer’s instructions. In short, histologic sections were deparaffinized with xylene and rehydrated using a series of graded ethanol followed by washing with double distilled water. The sections were then disposed with antigen retrieval in Target Retrieval Solution (Dako). Endogenous peroxidase was blocked with Peroxidase-Blocking Reagent (Dako). Later, the sections were incubated with rabbit monoclonal anti-NSUN2 (#ab259941, Abcam) or rabbit monoclonal anti-YBX1 (#ab76149, Abcam) at 4 °C overnight, followed by incubation with secondary antibody FLEX/HRP from Dako at room temperature for 30 min. The staining process was carried out with DAB (Dako). The average optical density of staining was quantified using ImageJ software along with IHC Toolbox plugin.

RNA m5C dot blot assay

Equal amounts of RNA from different groups were denatured by heating at 65 °C for 5 min, and were then spotted onto the nitrocellulose membranes (Pall Corporation). After ultra-violet (UV) cross-linking, the membrane was stained with 0.1% methylene blue staining buffer, and was then blocked with 5% skim milk followed by incubation with the primary mouse monoclonal anti-m^5^C antibody (#ab10805, 1:1000, Abcam) overnight at 4 °C. After incubated with HRP conjugated anti-mouse IgG antibody, the membrane was visualized with the enhanced chemiluminescence (ECL, Advansta) system. Image J was used to quantify the relative m^5^C level.

m5C-RNA immunoprecipitation

m^5^C-RNA immunoprecipitation (m^5^C-RIP) assay was carried out according to a previously reported m^6^A RIP protocol with some modifications (42). Briefly, total RNA was purified and subjected to DNase I (Invitrogen) to avoid DNA contaminations. RNA was then fragmented by incubating at 95 °C for 25 s in fragmentation buffer, the reaction was stopped using 0.5 M EDTA (Invitrogen), followed by standard ethanol precipitation. After resuspension in RNase-free water (Thermo Fisher Scientific), fragmented RNA was incubated overnight at 4 °C with the premixture of m^5^C antibody (#ab10805, Abcam) and Pierce Protein A Magnetic Beads (Thermo Fisher Scientific). Then bound RNAs were treated with proteinase K, and were further precipitated overnight with sodium acetate (Invitrogen,) and glycogen (Thermo Fisher Scientific).

RNA-stability assay

Human DPCs were transfected with siRNAs against NSUN2 and YBX1 or controls in 6-well plates. After siRNA transfection, the cells were treated with actinomycin D or dimethyl sulfoxide and collected at the indicated times (0, 2, and 4 h). Total RNA was then isolated by TRIzol Reagent (Invitrogen) and analyzed by RT-qPCR. β-actin was used as an internal control. The mRNA half-lives time were analyzed using GraphPad Prism software.

Bisulfite conversion of RNA

RNA fragmentation and bisulfite conversion were performed as previously described (43). The bisulfite conversion of RNA was carried out by applying EZ RNA Methylation Kit (Zymo Research), which converted nonmethylated cytosines into uracil while methylated cytosines remained unchanged during the treatment. Before conversion, isolated RNA from DPCs was treated with DNase I (Invitrogen) to remove genomic DNA. Afterward, input RNA was converted with a unique bisulfite conversion reagent following manufacturer's instructions. Bisulfite-converted RNA was then resuspended in RNase-free water and used for subsequent cDNA synthesis.

Sanger sequencing of PCR products

To validate the methylated sites of m^5^C on target genes, bisulfite-converted RNA was reversely transcribed using Maxima H Minus cDNA Synthesis Master Mix (Thermo Fisher Scientific). The cDNA was then used in PCR amplification with ZymoTaq DNA polymerase (Zymo Research) in accordance with the manufacturer’s protocol. A specific pair of primer for IL1B used in PCR detection was as follows: 5′ GGGTTTGTGTGTGTGTGTGTGGTG 3′ and 5′ CCCTTTCCCTCCTCTCTCTCTCTCTC 3′. Purified PCR fragments were then directly cloned using 5 minTA/Blunt-Zero cloning kit (Vazyme), and individual clones were randomly picked for Sanger sequencing.

RNA immunoprecipitation

RIP assay was performed using a Dynabeads Protein A immunoprecipitation kit (Thermo Fisher Scientific,) as previously described (44). Briefly, monolayer cultured DPCs were treated with ultra-violet (UV) irradiation and lysed with cellular lysis buffer (Thermo Fisher Scientific). Total cell lysates were incubated with protein A beads bound with anti-YBX1 antibody (#ab76149, Abcam) overnight at 4 °C with rotation. Immunoprecipitated proteins were collected with 2 × SDS sample buffer and analyzed by Western blot. RNAs associated with YBX1 were precipitated in anhydrous ethanol with the help of glycogen (Thermo Fisher Scientific), and analyzed by RT-PCR with primers amplifying IL1B.

Statistical analysis

The individual data point in the bar graphs represents “n” independent biological replicates, as indicated in the figure legends. Two-tailed t test was applied for two-group statistical comparisons and one-way ANOVA was used to analyze the statistically significant differences of multiple groups.

Data availability

The gene expression data in the GEO database analyzed in this study can be accessed on the GEO website. All other data needed to evaluate the conclusions in the article are present in the paper and/or the Supporting Information.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.