Increased expression of NUDT21 in inflammatory diseases

To explore the role of NUDT21 in hyperinflammatory conditions, we first analyzed the mRNA expression profiles of various immune-mediated inflammatory diseases (IMIDs) via the GEO database. Elevated levels of NUDT21 mRNA were observed in inflamed tissues from patients with inflammatory bowel disease (IBD) (Fig. 1A), psoriasis (Fig. 1B), rheumatoid arthritis (RA) (Fig. 1C), and sepsis (Fig. 1D, Fig. S1B). There was also a notable increase in expression in systemic lupus erythematosus (SLE) patients (Fig. S1C) and in peritoneal monocyte-derived macrophages from mice subjected to CAR-T-cell therapy-induced cytokine release syndrome (Fig. S1D). Additionally, Nudt21 was upregulated in bone marrow-derived macrophages (BMDMs) after IFNγ/LPS stimulation at both the mRNA and protein levels (Fig. S1E, F). These findings suggest that NUDT21-expressing macrophages may facilitate the progression of inflammation.

Fig. 1

Increased NUDT21 expression in inflammatory diseases and its protective role against colitis in Nudt21-deficient macrophages. mRNA expression levels of NUDT21 in inflamed tissues from patients with inflammatory bowel disease (IBD) (GSE179285) (A), psoriasis (GSE13355) (B), rheumatoid arthritis (RA) and osteoarthritis (OA) (GSE236924) (C) and sepsis (GSE95233) (D). UC-un = uninflamed tissue from ulcerative colitis patients; UC = inflamed tissue from ulcerative colitis patients; CD-un = uninflamed tissue from Crohn’s disease patients; CD = inflamed tissue from Crohn’s disease patients; NN = normal skin from controls; PN = uninvolved skin from psoriatic patients; PP = involved skin from psoriatic patients; RA = rheumatoid arthritis; OA = osteoarthritis. Group sizes: (A) UC-un = 32; UC = 23, CD-un = 121, CD = 47, (B) NN = 64, PN = 58, PP = 58, C Control = 7, RA = 36, OA = 89, D Sepsis Nonsurvivor = 34, Sepsis Survivor = 68. Wild-type (WT) and Nudt21-cKO mice were treated with 3% dextran sulfate sodium (DSS) for 5 days followed by access to regular drinking water for 3 days. Body weight changes (E), representative images of the large intestine (F), colon length measurements (G), representative hematoxylin and eosin (H&E)-stained colon sections (H), and histological scoring (I) from both groups of mice were analyzed on day 8 after DSS treatment. Scale bars: 100 μm in H. The arrows indicate edema (yellow), lymphocytes (black), and neutrophils (green). The error bars represent the standard errors of the means (SEMs). P values were determined by unpaired Student’s t-test or one-way ANOVA

Protective role of Nudt21-deficient macrophages against Colitis

To investigate the in vivo function of Nudt21 in macrophages, we generated myeloid-specific Nudt21-deficient mice (Nudt21fl/fl LysMCre, hereafter referred to as Nudt21-cKO) by crossing Nudt21-floxed mice (Nudt21fl/fl) with lysozyme M-Cre tool mice (LysM-Cre, referred to as WT) (Fig. S2A). RT‒qPCR and immunoblot analyses confirmed the efficient depletion of Nudt21 in bone marrow-derived macrophages (BMDMs) from Nudt21-cKO mice at both the mRNA and protein levels (Fig. S2B, C). Immune cell profiling revealed similar compositions of the spleen and bone marrow between Nudt21-cKO and WT mice (Fig. S2D, E).

To determine the functional impact of Nudt21 deficiency on macrophages during inflammatory disease progression, we subjected WT and Nudt21-cKO mice to a 3% dextran sulfate sodium (DSS) protocol [26,27,28] for five days to induce colitis. Compared with their WT counterparts, Nudt21-cKO mice presented significantly less body weight loss, reduced colon shortening, and milder colon inflammation (Fig. 1E–I) while maintaining comparable myeloid cell counts in both groups (Fig. S2F). These findings suggest that Nudt21 deficiency confers a protective effect against colitis in mice, primarily through alterations in macrophage function. The reduction in colitis severity in Nudt21-cKO mice underscores the critical role of Nudt21 in modulating macrophage-mediated inflammatory responses.

Nudt21 loss in macrophages alleviates excessive inflammation

To investigate the role of macrophage-specific Nudt21 deficiency in acute hyperinflammation, we subjected mice to a hemophagocytic lymphohistiocytosis (HLH) animal model induced by intraperitoneal injections of polyI:C for 24 hours followed by LPS [29,30,31]. Compared with their WT counterparts, Nudt21-cKO mice presented significantly improved outcomes, as evidenced by increased survival rates (Fig. 2A) and less pronounced decreases in rectal temperature (Fig. 2B) and body weight (Fig. 2C) after HLH model induction. Additionally, the HLH Nudt21-cKO mice presented lower peripheral white blood cell counts (Fig. 2D), lower spleen injury scores (Fig. 2E), decreased splenic cellularity (Fig. S3A), less splenomegaly (Fig. 2F), and higher peripheral platelet counts (Fig. S3B). Moreover, these mice presented decreased mRNA expression of the proinflammatory cytokines IL6 and IL1β in both spleen and liver tissues, along with reduced plasma levels of IL1β (Fig. 2G, Fig. S3C, D).

Fig. 2

Loss of Nudt21 in Macrophages Alleviates Excessive Inflammation in an HLH Model. A Survival curves of WT and Nudt21-cKO mice in the hemophagocytic lymphohistiocytosis (HLH) model, with polyI:C (10 mg/kg) followed by lipopolysaccharide (LPS) (5 mg/kg) injections (n = 5 per group). Survival curves were analyzed via the log-rank (Mantel‒Cox) test. Analyses conducted 6 hours post-LPS injection included rectal temperature changes (B), body weight changes (C), white blood cell counts in peripheral blood (D), representative H&E-stained spleen sections and spleen injury scores (E) (n = 3 per group), the ratio of spleen weight to body weight (F), and the mRNA expression levels of Il6 and Il1β in spleen tissues (G). P values were determined by unpaired Student’s t test. (H) Survival curves of WT and Nudt21-cKO mice following the indicated treatments (n = 6 per group). Survival curves were analyzed via the log-rank (Mantel‒Cox) test. The error bars represent the standard errors of the means (SEMs)

To assess whether the enhanced survival of Nudt21-cKO mice was attributable primarily to macrophage effects, we depleted macrophages via clodronate liposomes [32,33,34,35]. This intervention reduced mortality in WT mice under the HLH model to a rate comparable with that in Nudt21-cKO mice, highlighting that the absence of Nudt21 in macrophages plays a key role in mitigating acute inflammation (Fig. 2H). Collectively, these findings demonstrate that Nudt21 deficiency in macrophages significantly ameliorates inflammation progression.

Reduced proinflammatory traits in Nudt21-deficient macrophages

To further explore the characteristics of macrophages lacking Nudt21 during acute inflammatory disease progression, we analyzed myeloid cell populations in the spleen in the HLH model. Our findings revealed increases in both the absolute number and percentage of splenic macrophages in Nudt21-cKO mice compared with WT mice (Fig. S3F, G), suggesting increased cell viability. Additionally, a decreased percentage of Annexin V + 7-AAD+ cells was observed in splenic macrophages from Nudt21-cKO mice (Fig. S3H, I), indicating reduced macrophage death.

In terms of proinflammatory cytokines, splenic macrophages from Nudt21-cKO mice presented significantly lower levels of IL6 and TNFα (Fig. 3A, B), as did a reduced percentage of TNFα-positive macrophages (Fig. 3C). Consistently, in the supernatant of the splenic macrophages sorted from the HLH model mice, the cKO group presented lower IL6 and TNFα levels than the WT group did (Fig. 3D, E). Furthermore, these mice presented decreased mRNA levels of Il6, Tnfα, and Il1β (Fig. 3F) and a reduction in the IL12β geometric mean fluorescence intensity (gMFI) in both bone marrow monocytes (BMMs) and peritoneal monocyte-derived macrophages (PMMs) (Fig. S4A, B). Additionally, in the peritoneal cavity of HLH Nudt21-cKO mice, there was a decrease in the proportion of Ly6Chi to Ly6Clow macrophages (Fig. 3G, H). Similar results were observed in the DSS model, where compared with WT intestinal macrophages, Nudt21-cKO intestinal macrophages exhibited reduced production of TNFα and IL6 (Fig. S4C-F).

Fig. 3

Reduced proinflammatory traits in Nudt21-deficient monocyte-derived macrophages. A Histogram illustrating the geometric mean fluorescence intensity (gMFI) of IL6 and quantification of the gMFI in splenic macrophages from WT and Nudt21-cKO mice in the HLH model (6 h, n = 5 per group). B Histogram displaying the gMFI of TNFα and quantification of the gMFI in splenic macrophages from WT and Nudt21-cKO mice in the HLH model (6 h, n = 3 per group). C Percentages of TNFα-positive macrophages in the spleens of WT and Nudt21-cKO mice in the HLH model (6 h, n = 3 per group). ELISA results showing IL6 (D) and TNFα (E) levels in the supernatants of splenic macrophages from HLH-WT and HLH-cKO mice 6 hours after LPS injection. F RT‒qPCR analysis of Il6, Tnfα, and Il1β in bone marrow monocytes from WT and Nudt21-cKO mice in the HLH model (6 h, n = 4 per group). Percentages of Ly6Clo cells among monocyte-derived macrophages (G) and the ratio of Ly6Chi to Ly6Clo monocyte-derived macrophages (H) in the peritoneal cavities of WT and Nudt21-cKO mice in the HLH model (2 h, n = 4 per group). Histograms showing the gMFI of IL6 (I) and TNFα (J), along with the quantification (K) of bone marrow-derived macrophages (BMDMs) from WT and Nudt21-cKO mice primed with 50 ng/ml IFNγ overnight and 50 ng/ml LPS for 8 hours (n = 3 per group). L RT‒qPCR analysis of Il6, Tnfα, and Il1β in BMDMs from WT and Nudt21-cKO mice primed with 50 ng/ml IFNγ overnight and 50 ng/ml LPS for 8 hours (n = 3 per group). The error bars represent the standard errors of the means (SEMs). P values were determined by unpaired Student’s t-test

On the basis of these in vivo findings, we stimulated BMDMs with IFNγ/LPS and assessed cytokine levels and cell viability. Flow cytometry analysis revealed reduced expression levels of IL6 and TNFα in Nudt21-cKO BMDMs (Fig. 3I–K). RT‒qPCR analysis confirmed that the mRNA levels of Il6, Tnfα, and Il1β were lower in Nudt21-cKO BMDMs than in their WT counterparts (Fig. 3L). Additionally, the viability of Nudt21-cKO BMDMs was greater than that of WT BMDMs (Fig. S3J), which is consistent with the in vivo disease model results.

These findings underscore the critical role of Nudt21 in modulating macrophage proinflammatory activation, with Nudt21 deficiency leading to significant reductions in inflammatory responses and enhanced macrophage viability across multiple experimental models.

Enhanced autophagy in Nudt21-ablated macrophages in response to inflammatory activation

To investigate the molecular mechanism underlying the anti-inflammatory characteristics of Nudt21-deficient macrophages, we conducted high-throughput transcriptome sequencing (RNA-seq) on BMDMs from WT and Nudt21-cKO mice, both with and without IFNγ/LPS stimulation (RNA-seq, >3.0 × 108 reads, 40 G raw data/sample). This analysis identified 2,707 differentially expressed genes, with 88% (2,373 genes) upregulated and 12% (334 genes) downregulated (Fig. 4A, Fig. S5A, B). Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses revealed that autophagy-related pathways were the most enriched (Fig. 4B, Fig. S5C, D). Compared with those from WT mice, BMMs from HLH Nudt21-cKO mice presented increased expression of key autophagy regulators, such as Map1lc3b, Ulk2, Wipi2, and Atg12 (Fig. 4D). To explore whether Nudt21 regulates autophagy-related gene expression at the posttranscriptional level, we conducted RT‒qPCR analyses in WT and cKO BMDMs with or without the transcriptional inhibitors actinomycin D (ActD) and α-amanitin. Actinomycin D broadly inhibits transcription by preventing RNA polymerase movement [36, 37], whereas α-amanitin specifically targets RNA polymerase II, inhibiting mRNA synthesis [38]. Notably, even in the presence of these inhibitors, the mRNA levels of Map1lc3b, Ulk2, Wipi2, and Atg12 remained elevated in the cKO group (Fig. 4E). These results strongly suggest that Nudt21 plays a critical role in modulating the expression of these autophagy-related genes via posttranscriptional mechanisms.

Fig. 4

Enhanced Autophagy in Nudt21-Ablated Macrophages Following Inflammatory Activation. A Volcano plot showing DEGs in Nudt21-cKO BMDMs compared with WT BMDMs after treatment with IFNγ (50 ng/ml) overnight, followed by LPS (50 ng/ml) stimulation for 4 hours. Genes with a fold change greater than 1.5 and P < 0.05 are highlighted; upregulated genes are shown in red, and downregulated genes are shown in blue. B Gene Ontology (GO) enrichment analysis of transcripts upregulated in Nudt21-cKO BMDMs relative to WT controls after treatment with IFNγ (50 ng/ml) overnight and LPS (50 ng/ml) for 4 hours. C Heatmap displaying the expression profiles of genes associated with autophagy pathways in WT and Nudt21-cKO BMDMs following treatment with IFNγ (50 ng/ml) overnight and LPS (50 ng/ml) for 4 hours. D RT‒qPCR analysis of autophagy-related gene expression in bone marrow monocytes from WT and Nudt21-cKO mice in the HLH model (6 h, n = 6 per group). E RT‒qPCR analysis of the mRNA levels of autophagy-related genes in BMDMs from WT and Nudt21-cKO mice following treatment with IFNγ (50 ng/mL) overnight and LPS (50 ng/mL) for 4 hours, with or without 5 μM actinomycin D (ActD) or 5 μg/mL α-amanitin. Representative transmission electron microscopy (TEM) images (F) and quantification of autolysosomes (G) in BMDMs primed with 50 ng/ml IFNγ overnight and 50 ng/ml LPS for 4 hours. Scale bar: 500 nm. Images represent macrophages pooled from 2–3 mice per genotype and condition, with at least 6 images analyzed per genotype. H Immunoblot analysis showing the LC3 protein levels in WT and Nudt21-cKO BMDMs following overnight stimulation with IFNγ (50 ng/mL) and LPS (50 ng/mL) for the indicated durations. I Fluorescence microscopy images of LC3B and LAMP1 staining, along with DAPI counterstaining, in WT and Nudt21-cKO BMDMs primed with 50 ng/ml IFNγ overnight and 50 ng/ml LPS for 2 hours. Scale bar in the zoomed image: 2 µm. Captions represent at least two independent experiments. J Quantification of colocalized puncta of LC3 and LAMP1 across 50 cells per group. The error bars represent the standard errors of the means (SEMs). P values were determined by unpaired Student’s t-test

Autophagy plays a crucial role in cellular homeostasis by orchestrating the degradation and recycling of cellular components [39, 40]. Dysfunctional autophagy in macrophages can disrupt the regulation of inflammatory cytokine production, exacerbating inflammatory diseases such as sepsis, inflammatory bowel disease, and atherosclerosis [41,42,43]. In our study, we assessed autophagic activity in Nudt21-cKO BMDMs upon IFNγ/LPS activation via a monodansylcadaverine (MDC) assay, which revealed an increase in the number of autophagic vacuoles (Fig. S5E). Transmission electron microscopy (TEM) analysis confirmed a significant increase in the number of autolysosomes in Nudt21-cKO BMDMs compared with WT controls (Fig. 4F, G). Additionally, immunoblotting revealed an elevated LC3II/LC3-I ratio in Nudt21-cKO BMDMs (Fig. 4H), indicating enhanced conversion of cellular LC3-I to the lipidated LC3-II form. Immunofluorescence staining further revealed pronounced colocalization of LC3 puncta with lysosomes in Nudt21-cKO BMDMs, unlike in the WT group (Fig. 4I, J).

These findings collectively indicate significantly enhanced autophagic activity in Nudt21-deficient macrophages, highlighting the vital role of Nudt21 in regulating autophagy during macrophage activation.

Nudt21 modulates PolyA site selection and represses the mRNA stability of key autophagy genes

Nudt21 specifically recognizes UGUA motifs to facilitate the use of distal polyadenylation sites, and its absence leads to 3’ UTR shortening in mRNAs [21, 44, 45]. To investigate the effect of Nudt21 deletion on the 3’UTR length of mRNAs in activated macrophages, we utilized the DAPARS algorithm [46] to analyze deep RNA sequencing data from WT and Nudt21-cKO BMDMs. This analysis revealed that approximately 1,800 genes underwent Percent Distal PolyA Site Usage Index (PDUI) changes, 90% of which exhibited significant 3’UTR shortening (Fig. 5A). This finding underscores Nudt21’s conserved role in promoting 3’UTR lengthening across various cell types.

Fig. 5

Nudt21 Directs PolyA Site Selection and Represses the mRNA Stability of Autophagy-Related Genes. A Histogram summarizing the number of genes associated with changes in the percent distal polyA site usage index (PDUI) in WT and Nudt21-cKO BMDMs primed with 50 ng/ml IFNγ overnight and 50 ng/ml LPS for the indicated durations. B Venn diagram showing the overlap of genes with changes in both 3’UTR length and mRNA expression levels in Nudt21-cKO BMDMs primed with 50 ng/ml IFNγ overnight and 50 ng/ml LPS for 4 hours. C Histogram of Gene Ontology (GO) enrichment analysis identifying autophagy-related pathways among the 354 genes with changes in PDUI and mRNA levels in Nudt21-cKO BMDMs compared with WT controls. D List of autophagy-related genes from the overlapping set shown in B. E RNA immunoprecipitation (RIP)-qPCR analysis showing the binding of Nudt21 to the mRNAs of Map1lc3b and Ulk2 in WT BMDMs primed with 50 ng/ml IFNγ overnight and 50 ng/ml LPS for 4 hours. F RNA-seq reads across the Map1lc3b and Ulk2 loci in WT (blue) and Nudt21-cKO (red) BMDMs primed with 50 ng/ml IFNγ overnight and 50 ng/ml LPS for 4 hours. G, H RT‒qPCR analysis of distal polyA site usage in the 3’UTRs of Map1lc3b (G) and Ulk2 (H) in WT and Nudt21-cKO BMDMs primed with 50 ng/ml IFNγ overnight and 50 ng/ml LPS for the indicated durations. I, J RT‒qPCR analysis of Map1lc3b (J) and Ulk2 (K) mRNA decay in WT and Nudt21-cKO BMDMs at multiple time points (1 hour and 2 hours) after 5 μM actinomycin D (ActD) treatment (n = 3 per group). The error bars represent the standard errors of the means (SEMs). P values were determined by unpaired Student’s t-test

Given that Nudt21 loss might affect gene expression through mRNA 3′-UTR shortening [20], we further explored its impact on gene expression in activated macrophages. We identified 354 genes in Nudt21-cKO BMDMs whose PDUI and mRNA expression levels were altered by overlapping PUDI shortening and mRNA changes (Fig. 5B). GO analysis revealed that autophagy-related pathways were predominantly affected (Fig. 5C, Fig. S5F, G).

In terms of autophagy-related genes (Fig. 5D), RNA immunoprecipitation of IFNγ/LPS-treated WT BMDMs with an Nudt21 antibody confirmed the direct binding of Nudt21 to the mRNAs of Map1lc3b and Ulk2 but not Wipi2 (Fig. 5E). Map1lc3b (LC3B) and Ulk2 are essential for autophagosome formation and initiation, with LC3B involved in the expansion and closure of the autophagosome membrane and Ulk2 acting as an upstream initiator of autophagy by phosphorylating various downstream targets [47]. Significant 3’UTR shortening in Map1lc3b and Ulk2 was observed in Nudt21-cKO BMDMs (Fig. 5F), as validated by RT‒qPCR via primers targeting both proximal and distal PAS (Fig. 5G, H). 3’ UTRs contain destabilizing elements, such as AU-rich elements (AREs), GU-rich elements (GREs), and PUF protein-binding sites, which regulate mRNA stability through interactions with RNA-binding proteins (RBPs) and miRNAs [16, 20, 48]. The inclusion or exclusion of these elements due to 3’ UTR-APA can substantially affect transcript stability. To investigate whether 3’ UTR shortening influences mRNA stability, we measured the decay rates of the Map1lc3b (NM_026160) and Ulk2 (NM_013881) mRNAs at different time points following actinomycin D treatment. RT‒qPCR analysis revealed delayed mRNA decay in Nudt21-cKO BMDMs compared with WT BMDMs, whereas Wipi2 mRNA decay rates remained unaffected (Fig. 5I, J, Fig. S5H).

Together, these results demonstrate that Nudt21 represses the mRNA stability of the critical autophagy regulators Map1lc3b and Ulk2 by directing the selection of their 3’UTR polyA sites, contributing to the regulation of autophagy and inflammatory responses in macrophages.

Nudt21 impedes autophagy’s role in reducing inflammatory cytokine production

Recent research highlights macrophage autophagy as a pivotal anti-inflammatory mechanism across various inflammatory diseases [40, 42, 43, 49]. Given these findings, we hypothesized that the absence of Nudt21 might enhance autophagic flux in macrophages, thereby reducing inflammatory responses. To test this hypothesis, we treated BMDMs with the lysosomal inhibitor bafilomycin A1 (BafA1) during IFNγ/LPS stimulation. The flow cytometry results revealed that BafA1 attenuated the differences in the IL6 and TNFα protein levels between the wild-type and Nudt21-cKO BMDMs (Fig. 6A, B). This result was corroborated by ELISA, which revealed similar cytokine levels in both groups upon BafA1 treatment (Fig. 6C, D). RT‒qPCR analysis revealed that the differences in Il6 and Tnfα mRNA levels between the two groups were significantly reduced after BafA1 treatment (Fig. 6E, F).

Fig. 6

Nudt21 Impedes Autophagy’s Role in Reducing Inflammatory Cytokine Production. A, B Histograms displaying cytokine levels in BMDMs from WT and Nudt21-cKO mice primed with 50 ng/ml IFNγ overnight and 50 ng/ml LPS for 8 hours (n = 3 per group): (A) IL6 gMFI. (B) TNFα gMFI. ELISA results showing the IL6 (C) and TNFα (D) levels in the supernatants of the cultures primed with 50 ng/ml IFNγ overnight and 50 ng/ml LPS for 12 hours, with or without 100 nM BafA1. (n = 5 per group). RT‒qPCR analysis of cytokine mRNA levels in BMDMs primed with 50 ng/ml IFNγ overnight and 50 ng/ml LPS for 8 hours: (E) IL6 mRNA levels (n = 3 per group). (F) TNFα mRNA levels (n = 3 per group). G Immunoblot analysis of phosphorylated IKKα/β, total IKKα, total IKKβ, phosphorylated NF-κB p65 (p-p65), total p65, p62, LC3, and NUDT21 levels in WT and Nudt21-cKO BMDMs primed with 50 ng/ml IFNγ overnight and 50 ng/ml LPS, with or without 100 nM BafA1, for the indicated durations. β-ACTIN was used as the loading control. The error bars represent the standard errors of the means (SEMs). P values were determined by unpaired Student’s t-test

Considering the established connection between autophagy and the NF-κB signaling pathway [43, 50,51,52,53], we further examined changes in NF-κB activation. Compared with WT BMDMs, Nudt21-cKO BMDMs presented decreased phosphorylation levels of NF-κB. Additionally, the expression level of p62 (Sqstm1), a canonical autophagy adaptor, increased over time in WT BMDMs but decreased in Nudt21-cKO BMDMs. Following BafA1 treatment, there was a significant increase in NF-κB phosphorylation in Nudt21-cKO BMDMs, similar to the changes observed in p62 and LC3B levels (Fig. 6G). Moreover, we detected reduced phosphorylation and total levels of both IKKα and IKKβ in the Nudt21-cKO group, indicating impaired activation of these kinases. Interestingly, when the autophagy pathway was inhibited by BafA1, we observed a marked increase in IKKα levels in the cKO group, whereas IKKβ levels remained unchanged (Fig. 6G). These results suggest that Nudt21 deficiency may enhance the autophagic degradation of IKKα, which in turn could disrupt the IKKα‒NF-κB p65 axis, potentially leading to altered macrophage inflammatory responses.

Taken together, these findings suggest a critical role for Nudt21 in impeding autophagic flux in macrophages and its subsequent augmentation of inflammatory responses. This regulatory mechanism underscores the importance of Nudt21 in controlling inflammatory diseases, highlighting its potential as a therapeutic target for conditions characterized by dysregulated autophagy and inflammation.