Report

N6-methyladenosine RNA modification–mediated cellular metabolism rewiring inhibits viral replication

See allHide authors and affiliations

Science  13 Sep 2019:
Vol. 365, Issue 6458, pp. 1171-1176
DOI: 10.1126/science.aax4468

RNA modification meets immune metabolism

N6-methyladenosine (m6A) RNA modification regulates various cellular functions. Liu et al. found that host cells impair RNA m6A demethylase activity after viral infection, leading to increased m6A and reduced stability of α-ketoglutarate dehydrogenase (OGDH) mRNA. As a result, reduced OGDH decreases the generation of itaconate, thereby inhibiting viral replication. The authors explore the function of OGDH and itaconate in viral infection, provide insights into m6A RNA modification and metabolic reprogramming in modulating virus-host interaction, and suggest potential therapeutic targets for the control of viral infection.

Science, this issue p. 1171

Abstract

Host cell metabolism can be modulated by viral infection, affecting viral survival or clearance. Yet the cellular metabolism rewiring mediated by the N6-methyladenosine (m6A) modification in interactions between virus and host remains largely unknown. Here we report that in response to viral infection, host cells impair the enzymatic activity of the RNA m6A demethylase ALKBH5. This behavior increases the m6A methylation on α-ketoglutarate dehydrogenase (OGDH) messenger RNA (mRNA) to reduce its mRNA stability and protein expression. Reduced OGDH decreases the production of the metabolite itaconate that is required for viral replication. With reduced OGDH and itaconate production in vivo, Alkbh5-deficient mice display innate immune response–independent resistance to viral exposure. Our findings reveal that m6A RNA modification–mediated down-regulation of the OGDH-itaconate pathway reprograms cellular metabolism to inhibit viral replication, proposing potential targets for controlling viral infection.

Cellular metabolism is involved in various biological processes, including interactions between virus and host. Upon recognition or sensing of the invading viruses, host cells actively rewire metabolism to either provide necessary components for viral survival or inhibit viral replication for its clearance. Multiple regulators participate in these changes during virus–host interaction (13). Viruses can also modulate host metabolic pathways to complete their life cycles, evade host immune responses, and create a favorable cellular microenvironment for persistent infection. For example, adenovirus infection changes host cell metabolism by inducing a Warburg-like shift to aerobic glycolysis to synthesize the proteins and nucleic acids required for viral replication (4).

RNA modifications, especially the most common mammalian mRNA modification, N6-methyladenosine (m6A) (5), can modulate gene expression and regulate viral infection (6, 7). For instance, the m6A methyltransferase complex components METTL3 and METTL14 restrict production of the flavivirus Zika, whereas the m6A demethylases ALKBH5 and FTO enhance its production (7).

To investigate the role of m6A RNA modification in host response to viral infection, we detected the m6A levels in virus-infected host cells. The m6A levels in total RNA increased initially and then decreased during infection with the RNA virus vesicular stomatitis virus (VSV) in mouse primary peritoneal macrophages, accompanying the increased and subsequently decreased viral loads (Fig. 1A). We then performed RNA interference–mediated functional screening of the m6A writer complex (METTL3, METTL14, and WTAP) and erasers (ALKBH5 and FTO) and found that knockdown of ALKBH5 most significantly reduced VSV RNA levels (Fig. 1B), a result that was later validated using four independent small interfering RNAs (siRNAs) (fig. S1, A to C). ALKBH5 knockout in the mouse macrophage cell line RAW264.7 by CRISPR-Cas9 also decreased intracellular virus production upon infection of recombinant green fluorescent protein–expressing VSV (GFP-VSV) (fig. S1, D to G).

Fig. 1 ALKBH5 facilitates viral replication both in vitro and in vivo, independent of innate response.

(A) Western blot of VSV–G protein (top) and dot blot of m6A levels (250 ng total RNA; middle) in mouse peritoneal macrophages infected with VSV for the indicated times. The m6A level intensity (relative to 0 hours; bottom) was determined using the ImageJ program (n = 3). GAPDH, glyceraldehyde-3-phosphate dehydrogenase; MB, methylene blue. (B) qPCR of VSV RNA in peritoneal macrophages transfected with the indicated siRNAs and then infected with VSV for 8 hours (n = 3). (C) qPCR of VSV RNA in WT and Alkbh5-deficient (Alkbh5−/−) peritoneal macrophages infected with VSV for the indicated times (n = 3). (D) VSV titers by median tissue culture infectious dose (TCID50) assay in supernatants of peritoneal macrophages infected with VSV for the indicated times (n = 3). (E) qPCR of EMCV RNA (left) or HSV-1 abundance (right) in peritoneal macrophages infected with the indicated virus for 8 hours (n = 3). (F) Survival curves of 8-week-old Alkbh5−/− and WT littermate mice after infection with VSV [5 × 107 plaque-forming units (PFU) per gram of body weight] (n = 5). (G) TCID50 determination of VSV loads in organ homogenate supernatants 18 hours after infection with VSV (5 × 107 PFU per gram of body weight) (n = 5). (H) qPCR of VSV RNA replicates in organs from mice, as in (G) (n = 5). (I) qPCR of HSV-1 abundance in organs from Alkbh5−/− and WT littermate mice infected with HSV-1 (1 × 107 PFU per mouse) for 3 days (n = 5). (J) qPCR of Ifnb1 mRNA in peritoneal macrophages infected with the indicated virus for 8 hours (n = 3). SeV, Sendai virus. (K) Western blot of VSV–G protein in peritoneal macrophages from Ifnar1−/−Alkbh5+/+, Ifnar1, and Alkbh5 double-KO mice infected with VSV. (L) Enzyme-linked immunosorbent assay (ELISA) of serum IFN-I from mice, as in (G) (n = 3). (M) ELISA of serum IFN-I from mice infected with HSV-1, as in (I), for 6 hours. (n = 5). All data are mean ± SEM of biologically independent samples. n = number of biological replicates. Data are representative of three independent experiments with similar results [(A) and (K)]. qPCR data are normalized to Actb expression. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; NS, not significant. Kaplan-Meier (F) or two-tailed unpaired Student’s t test [(A) to (E), (G) to (J), (L), and (M)].

Once infected with VSV in vitro, the peritoneal macrophages derived from Alkbh5-deficient mice (8) had lower VSV RNA levels, virus titers (Fig. 1, C and D), and intracellular GFP-VSV levels (fig. S1H) than those of wild-type (WT) littermates. Conversely, ALKBH5 overexpression increased VSV RNA levels (fig. S1, I and J). In addition, Alkbh5 deficiency inhibited VSV infection in other types of mouse cells, including bone marrow–derived macrophages (fig. S2A), primary mouse embryonic fibroblasts (fig. S2B), and alveolar epithelial cells (fig. S2C). ALKBH5 is highly conserved between humans and mice (9). We found that knockdown of ALKBH5 inhibited viral infection in human THP-1 cells (fig. S2, D and E). Alkbh5 deficiency also rendered the cells resistant to the infection of other types of RNA or DNA viruses, including the RNA virus encephalomyocarditis virus (EMCV) and the DNA virus herpes simplex virus type 1 (HSV-1) (Fig. 1E). Furthermore, short-term infection assays, performed when a virus had just completed the entry phase but had not yet started to replicate, showed that Alkbh5 deficiency did not affect viral entry (fig. S2F). There was no difference between WT and Alkbh5-deficient cells in responses to poly(I:C) stimulation (fig. S2G).

Alkbh5-deficient mice showed improved survival rates compared with their WT littermates after viral infection (Fig. 1F). The viral burden was significantly reduced in several organs of Alkbh5-deficient mice (Fig. 1G), with reduced inflammatory cell infiltration and attenuated pathological changes in the lung (fig. S2H). Consistently, Alkbh5-deficient mice had substantially decreased RNA virus VSV replicates in organs and peritoneal cells (Fig. 1H and fig. S2I). Furthermore, Alkbh5 deficiency reduced DNA virus HSV-1 abundance in vivo (Fig. 1I). Taken together, these results indicate that ALKBH5 has a broad effect on facilitating both RNA and DNA virus replication in vitro and in vivo.

We then investigated whether Alkbh5 deficiency inhibits viral replication through enhancing the innate immune response. Fluorescence-activated cell sorting analysis indicated normal development of immune cells in Alkbh5-deficient mice (fig. S3, A to G). RNA sequencing (RNA-seq) showed that mRNA expression of innate immune genes remained unchanged or even decreased in Alkbh5-deficient macrophages infected with VSV (fig. S4A). Quantitative polymerase chain reaction (qPCR) verification also indicated decreased mRNA levels of type I interferons (IFN-I; e.g., Ifnb1 and Ifna4), tumor necrosis factor–α (Tnf-a), and interleukin-6 (Il-6) in Alkbh5-deficient macrophages upon VSV infection (fig. S4, B to D), as well as reduced Ifnb1 expression in Alkbh5-deficient macrophages upon infection with other types of RNA or DNA viruses (Fig. 1J). Knockdown of ALKBH5 inhibited virus-induced IFN-I expression in human THP-1 cells (fig. S4E). Alkbh5 deficiency impaired viral infection–initiated innate signaling (fig. S4F) and IRF3 dimerization (fig. S4G). Conversely, ALKBH5 overexpression increased Ifnb1 expression upon viral infection (fig. S4H). These data indicate that ALKBH5 promotes viral replication by not suppressing the innate immune response.

Furthermore, we generated Alkbh5 and IFN-I receptor (Ifnar1) double-knockout mice and confirmed that Alkbh5 deficiency still suppressed viral replication even without IFN-I signaling (Fig. 1K). Alkbh5 deficiency also reduced or had no effect on the expression of interferon-stimulated genes (ISGs) and innate cytokines (fig. S4I). Lower levels of serum IFN-β, IFN-α, TNF-α, and IL-6 (Fig. 1L and fig. S4J) and decreased mRNA expression of innate cytokines and ISGs in organs and peritoneal cells (fig. S5, A to G) were detected in Alkbh5-deficient mice after RNA virus infection. In vivo deficiency of Alkbh5 also reduced serum IFN-β and IFN-α production in mice exposed to DNA virus HSV-1 (Fig. 1M). Therefore, the innate immune response, which is impaired rather than bolstered by Alkbh5 deficiency, is not responsible for ALKBH5-mediated promotion of viral replication.

We next explored the physiological importance of ALKBH5 in the virus–host interaction. There was no change in ALKBH5 protein expression and subcellular localization after viral infection (Fig. 2A and fig. S6A). However, viral infection decreased cellular RNA m6A demethylation activity to a level comparable to that resulting from Alkbh5 deficiency (fig. S6, B and C). We wondered whether viral infection may impair the enzymatic activity of ALKBH5. The RNA m6A demethylation activity of ALKBH5 purified from virus-infected cells was decreased (Fig. 2B). Furthermore, overexpression of WT ALKBH5, but not the catalytic inactive mutant H205→A (H205A) (8, 10), rescued viral replication in Alkbh5-deficient macrophages (Fig. 2C). Therefore, viral infection impairs the m6A demethylation activity of ALKBH5.

Fig. 2 Host cells impair ALKBH5-mediated RNA m6A demethylation to inhibit viral replication.

(A) Western blot of ALKBH5 in mouse peritoneal macrophages infected with VSV. (B) m6A RNA demethylation activity of ALKBH5 purified from mammalian cells infected with VSV (n = 4). (C) qPCR of VSV RNA in Alkbh5−/− macrophages transfected with WT ALKBH5 (ALKBH5-WT) or catalytic inactive mutant H205A (ALKBH5-H205A) for 24 hours, then infected with VSV for 8 hours (n = 3). (D) Dot blot of the m6A levels in total RNA from Alkbh5−/− macrophages transfected with the indicated vectors for 24 hours. (E) qPCR of VSV RNA in Alkbh5−/− macrophages, as in (D), with VSV infection for 8 hours (n = 5). (F) Cell-free biochemistry determination of m6A demethylation activity of the same amount of recombinant WT ALKBH5, H205A, and R107Q mutants purified from mammalian cells (n = 5). All data are mean ± SEM of biologically independent samples. n = number of biological replicates. Three independent experiments were performed for all blot assays [(A) and (D)]. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; NS, not significant. Two-tailed unpaired Student’s t test [(B), (C), (E), and (F)]. A, Ala; D, Asp; H, His; K, Lys; Q, Gln; R, Arg.

Posttranslational modifications (PTMs) are known to regulate enzyme activity in biological processes, including cell response to viral infection. For instance, viral infection can induce the methylation of STAT1 and IRF3 to increase their transcriptional activities, which enhances antiviral innate immunity and promotes elimination of invading virus (11, 12). We performed mass spectrometry (MS) analysis to identify the potential PTM changes of ALKBH5. Upon viral infection, two residues (R107 and D109) were demethylated and two residues (K345 and R349) were methylated on the ALKBH5 protein (table S1). We constructed ALKBH5 mutants by replacing MS-identified methylated R with Q, K with A, and D with A, respectively (fig. S6D). When these mutants were overexpressed in Alkbh5-deficient macrophages, only the R107Q mutant did not remove the RNA m6A modification (Fig. 2D) and could not rescue viral replication (Fig. 2E), similar to the catalytic inactive mutant H205A. These results indicate that R107 demethylation may impair ALKBH5’s m6A demethylation activity. We then used a cell-free biochemical system to compare the m6A demethylation activity of recombinant WT ALKBH5, the R107Q mutant, and the control catalytically inactive mutant H205A and found that both the H205A and R107Q mutants exhibited impaired m6A demethylation activity (Fig. 2F). Therefore, in response to viral infection, host cells actively induce R107 demethylation on the ALKBH5 protein to impair its enzymatic activity, which leads to attenuated RNA m6A demethylation in virus-infected cells for inhibiting viral replication.

To identify the critical downstream target(s) of ALKBH5, we performed KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway enrichment analysis of the RNA-seq data and found that metabolic pathways were most substantially affected in Alkbh5-deficient macrophages upon viral infection (fig. S7A). RNA-seq revealed pronounced changes of many metabolism-related genes in Alkbh5-deficient macrophages (fig. S7B). Moreover, metabolomics profiling showed that Alkbh5 deficiency led to metabolite profile perturbations in vivo (fig. S7C).

We then validated the expression of the nine metabolic genes that were most substantially down- or up-regulated in Alkbh5-deficient macrophages after viral infection (fig. S8, A to H). Among those, Ogdh was the most markedly down-regulated gene (Fig. 3A and fig. S7B). Alkbh5 deficiency or viral infection significantly down-regulated OGDH expression (Fig. 3B). OGDH expression was also down-regulated after infection with different types of RNA or DNA viruses but remained unchanged in response to the stimuli with Toll-like receptor (TLR) ligands (Fig. 3C and fig. S8, I and J). Moreover, OGDH expression has been reported to be suppressed after human pandemic virus infection (13). Functional screening of the top nine metabolic genes showed that OGDH knockdown most robustly inhibited viral replication, accompanied by a decrease in Ifnb1 expression (Fig. 3D and fig. S9, A and B), phenocopying Alkbh5 deficiency. Knockdown of other metabolic genes had no such effect, except that SCGB1A1 knockdown inhibited viral replication less potently than did OGDH knockdown (fig. S9A). The other candidates not involved in the tricarboxylic acid (TCA) cycle were not cross-regulated with OGDH in this process. Furthermore, OGDH knockdown reduced viral replication and IFN-I expression in human THP-1 cells (fig. S9, C to E). Therefore, suppression of metabolic enzyme OGDH expression is a general mechanism by which host cells inhibit viral replication.

Fig. 3 ALKBH5 promotes viral replication by up-regulating the OGDH-itaconate metabolic pathway.

(A) RNA-seq identification of Ogdh as one of the most significantly down-regulated genes in Alkbh5−/− macrophages infected with VSV for 8 hours. Two independent biological replicates. (B) qPCR (top) (n = 3) and Western blot (bottom) of OGDH in Alkbh5−/− peritoneal macrophages infected with VSV. (C) Kinetic analysis of OGDH protein expression in peritoneal macrophages exposed to the indicated RNA or DNA viruses (8 hours; left) or TLR ligands (9 hours; right). OGDH protein intensity (relative to the level at 0 hours) was determined using ImageJ (n = 5). (D) qPCR of VSV RNA (left) and Ifnb1 mRNA (right) in macrophages transfected with OGDH siRNAs and then infected with VSV (n = 3). (E) qPCR of VSV RNA in macrophages pretreated with SP or CPI-613 and infected with VSV for 8 hours (n = 5). DMSO, dimethyl sulfoxide. (F) qPCR of VSV RNA in Alkbh5−/− macrophages transfected with OGDH for 24 hours, followed by VSV infection (n = 3). (G) qPCR of VSV RNA in macrophages cotransfected with WT ALKBH5 and OGDH siRNA for 24 hours, followed by VSV infection (n = 3). (H) Liquid chromatography–tandem MS (LC-MS/MS) analysis of cellular metabolite levels in macrophages transfected with OGDH siRNA, followed by VSV infection (n = 5). FA, fumarate; MA, malate; AA, cis-aconitate; ITA, itaconate; ICA, isocitrate; α-KG, α-ketoglutarate; SC, succinate–coenzyme A; ASP, aspartate. (I) Schematic illustration of the OGDH-itaconate metabolic pathway. (J) qPCR of VSV RNA in peritoneal macrophages pretreated with DMI, followed by VSV infection (n = 3). (K) qPCR of VSV RNA in peritoneal macrophages pretreated with DMI or OI (62.5 μM), followed by VSV infection (n = 3). (L) qPCR of VSV RNA in macrophages transfected with OGDH siRNA, then pretreated with OI followed by VSV infection (n = 3). (M and N) qPCR of Ogdh mRNA (M) and metabolomics analysis of itaconate abundance (N) in lungs from Alkbh5−/− and WT littermate mice infected with VSV (n = 5). (O) VSV loads by TCID50 assay in organ homogenate supernatants from Alkbh5−/− and WT littermate mice pre-injected intraperitoneally with OI (50 mg/kg) and 2 hours later infected with VSV for 18 hours (n = 5). (P) qPCR of VSV RNA in organs from mice, as in (O) (n = 5). (Q) ELISA of serum IFN-I from mice, as in (O) (n = 5). All data are mean ± SEM of biologically independent samples. n = number of biological replicates. Five independent experiments were performed for all blot assays [(B) and (C)]. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; NS, not significant. Two-tailed unpaired Student’s t test [(B) to (H) and (J) to (Q)].

OGDH is the first rate-limiting enzyme in the TCA cycle, yet there has been no report on the function of OGDH in viral infection. Pharmacological treatment of WT macrophages with two inhibitors of OGDH, succinyl phosphonate (SP) and CPI-613 (14), dose-dependently inhibited viral replication (Fig. 3E). OGDH overexpression rescued viral replication in Alkbh5-deficient macrophages (Fig. 3F), whereas ALKBH5 overexpression in OGDH-silenced cells could not rescue viral replication (Fig. 3G). Thus, ALKBH5 promotes viral replication metabolically by up-regulating OGDH.

Targeted metabolomics analyses showed that OGDH silencing changed the levels of several important metabolic intermediates in virus-infected macrophages (Fig. 3H). Aspartate, which is well known to promote viral replication (3), was one of the reduced intermediates, consistent with previous work showing that OGDH silencing depletes aspartate (15). Itaconate was one of the most significantly down-regulated intermediates in OGDH-silenced macrophages upon viral infection (Fig. 3H). Itaconate is converted from cis-aconitate, an intermediate of the TCA cycle, which was also decreased in OGDH-silenced macrophages (Fig. 3, H and I). As the most abundant metabolite in lipopolysaccharide-treated macrophages, itaconate exerts anti-inflammatory effects by modulating mitochondrial respiration and metabolic remodeling or by KEAP1 alkylation to activate NRF2 (16, 17). However, the role of itaconate in regulating viral replication remains unknown. Treatment of Alkbh5-deficient macrophages with the cell-permeable itaconate derivative dimethyl itaconate (DMI) (16) dose-dependently rescued viral replication (Fig. 3J). Another cell-permeable itaconate derivative, 4-octyl itaconate (OI) (17), also potently rescued viral replication (Fig. 3K). Consistently, OI rescued viral replication in OGDH-silenced macrophages and promoted viral replication in a dose-dependent manner (Fig. 3L).

What is the in vivo role of the ALKBH5-OGDH-itaconate axis? In vivo Alkbh5 deficiency down-regulated Ogdh expression and itaconate production in mice exposed to virus (Fig. 3, M and N). Notably, administration of itaconate into Alkbh5-deficient mice sufficiently rescued the viral burden and replicates (Fig. 3, O and P) without affecting the mRNA and protein expression of OGDH in vivo (fig. S9, F and G). An increased serum IFN-I level was observed in mice after administration of itaconate (Fig. 3Q). These results demonstrate that the ALKBH5-OGDH-itaconate pathway promotes viral replication in an IFN-independent manner.

We next sought to determine whether increased m6A RNA methylation triggered by Alkbh5 deficiency directly contributes to the down-regulation of OGDH expression. We performed transcriptome-wide m6A methylation profiling of WT and Alkbh5-deficient macrophages by m6A sequencing (m6A-seq), which showed the consensus m6A motif with typical m6A peak distribution features (Fig. 4A and fig. S10, A to D). There was no up-regulation of m6A signal on VSV RNA in Alkbh5-deficient macrophages (fig. S10E), indicating that ALKBH5 promotes viral replication by not targeting m6A modification of viral RNA. METTL3 deletion is reported to reduce the level of m6A on IFNB1 mRNA to increase its expression in foreskin fibroblasts (18). However, our m6A-seq analysis showed only marginal enrichment and no significant difference of m6A signal on Ifnb1 mRNA between WT and Alkbh5-deficient macrophages (fig. S10F), further confirming an IFN-independent mechanism in ALKBH5-promoted viral replication.

Fig. 4 ALKBH5 directly erases m6A modification on Ogdh mRNA to increase its stability and expression.

(A) Consensus motifs and P value of m6A peaks identified by HOMER from m6A-seq analysis. Two independent biological replicates. (B) m6A abundance on Ogdh mRNA as in (A). Red, Alkbh5−/− IP; blue, Alkbh5+/+ IP; gray, input. The y axis represents the normalized m6A signal along the gene. (C) m6A enrichment of Ogdh mRNA in the indicated macrophages by m6A-RIP-qPCR (n = 3). Results are presented relative to those obtained with immunoglobulin G (IgG). Gapdh, m6A negative control; Myc peak, m6A positive control. (D) qPCR of Ogdh mRNA in Alkbh5−/− macrophages transfected with the indicated vectors for 24 hours, then infected with VSV (n = 4). (E) Piled reads of Ogdh mRNA from ALKBH5-iCLIP-seq. Blue box, ALKBH5-iCLIP-seq peaks; orange, overlapped regions between high-confidence iCLIP-seq peaks and m6A-seq peaks. The y axis represents the normalized iCLIP signal along the gene. (F) Ogdh mRNA degradation in peritoneal macrophages treated with actinomycin D for the indicated times (n = 4). The residual RNAs were normalized to 0 hours. (G) qPCR of Ogdh mRNAs, as in (F) (n = 4). (H) Ogdh mRNA degradation in the indicated YTHDF2-silenced macrophages treated with actinomycin D for the indicated times (n = 3). The residual RNAs were normalized to 0 hours. (I) qPCR of VSV RNA in the indicated YTHDF2-silenced macrophages infected with VSV (n = 3). All data are mean ± SEM of biologically independent samples. n = number of biological replicates. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; NS, not significant. Two-tailed unpaired Student’s t test [(C), (D), and (F) to (I)].

m6A-seq revealed highly enriched and specific m6A peaks on Ogdh mRNA, which were substantially increased in Alkbh5-deficient macrophages (Fig. 4B). m6A–RNA immunoprecipitation (RIP) combined with qPCR confirmed that Ogdh mRNA exhibits the m6A modification, and the m6A levels on Ogdh mRNA were further accumulated upon Alkbh5 deficiency or during viral infection (Fig. 4C). Overexpression of WT ALKBH5, but not H205A or R107Q mutants, rescued Ogdh expression in Alkbh5-deficient macrophages (Fig. 4D). Therefore, impaired m6A demethylation activity of ALKBH5 induced by viral infection is responsible for the down-regulation of OGDH expression.

We then used individual nucleotide resolution UV cross-linking and immunoprecipitation–based sequencing (iCLIP-seq) to map the targeting transcripts and binding sites of ALKBH5 and confirmed that ALKBH5 directly targeted the Ogdh transcript in macrophages (Fig. 4E and fig. S10G). Combining ALKBH5-iCLIP-seq with m6A-seq data, we identified overlapping peaks on Ogdh mRNA and found that the binding of ALKBH5 to Ogdh mRNA decreased upon viral infection (Fig. 4E). Other mRNAs, such as the mRNA of Got2 that plays a metabolic role in viral infection (3), were both targets for ALKBH5 and substrates for m6A demethylation. Therefore, ALKBH5-mediated metabolic rewiring via demethylation of Ogdh and other transcripts is important in regulating viral replication.

m6A RNA modification can regulate gene expression by modulating the splicing, stability, and translation of mRNA (5, 6). Our m6A-seq data did not show any significant difference in the alternative splicing pattern of Ogdh mRNA between WT and Alkbh5-deficient macrophages (fig. S10H). RNA decay assays showed that the stability and expression of Ogdh mRNA were markedly decreased in Alkbh5-deficient macrophages (Fig. 4, F and G). m6A RNA modification can promote mRNA degradation and be specifically recognized by a family of reader proteins, among which YTHDF2 is the major m6A reader responsible for the decay of m6A-modified mRNA (19, 20). Silencing of YTHDF2 delayed Ogdh mRNA degradation and rescued viral replication in Alkbh5-deficient macrophages (Fig. 4, H and I). Taken together, our findings indicate that ALKBH5 directly erases m6A RNA methylation on Ogdh mRNA to delay its decay mediated via YTHDF2, resulting in increased OGDH protein expression to metabolically promote viral replication.

We have elucidated that host cells actively respond to viral infection by impairing m6A demethylation activity of ALKBH5, leading to increased mRNA decay and reduced protein expression of the metabolic enzyme OGDH and consequently reprogramming the cellular metabolic state to confer IFN-I–independent host resistance to viral infection. Decreased OGDH expression limits the TCA cycle and reduces itaconate production that is required for viral replication, thus restricting viral infection in host cells (fig. S11). Our findings on the cross-talk of m6A RNA modification and metabolic reprogramming via the ALKBH5-OGDH-itaconate pathway provide mechanistic insight into the cellular defense against invading pathogens through the innate response–independent metabolism rewiring. Our ability to identify the cellular metabolites in host–virus interactions will facilitate an improved understanding of the viral infection process and help identify potential targets for intervention. Our study reveals that OGDH and itaconate promote viral replication in an innate immunity–independent manner, indicating that the OGDH-itaconate metabolic response may be targeted to control viral infectious diseases.

Supplementary Materials

science.sciencemag.org/content/365/6458/1171/suppl/DC1

Materials and Methods

Figs. S1 to S11

Tables S1 and S2

References (2130)

References and Notes

Acknowledgments: Funding: This work is supported by grants from the National Natural Science Foundation of China (81788101) and the Chinese Academy of Medical Sciences (CAMS) Innovation Fund for Medical Sciences (2016-12M-1-003). Author contributions: X.C. and Y.L. designed the experiments. Y.L., Y.Y., J.Y., P.L., L.L., and H.X. performed the experiments. Z.L. conducted bioinformatics analysis. Y.N. provided ALKBH5-deficient mice. X.C. and Y.L. analyzed data and wrote the paper. X.C. supervised research, coordination, and strategy. Competing interests: The authors declare no competing interests. Data and materials availability: All data are available in the manuscript or the supplementary materials. RNA-seq, m6A-seq, and iCLIP-seq raw data have been deposited in the NCBI Gene Expression Omnibus database under accession numbers GSE127739, GSE127732, and GSE134754, respectively.

Stay Connected to Science

Navigate This Article