ALKBH5 is required to initiate emergency granulopoiesis in vivo
Given the crucial role of neutrophil generation in antibacterial host defense, we investigated whether ALKBH5 might influence granulopoiesis in vivo by performing cecal ligation and puncture (CLP), a well-established polymicrobial sepsis model that can induce severe systemic infection and is closely associated with clinical sepsis, on Alkbh5-deficient mice and their wild-type (WT) littermates. As expected, we observed a reduction in the percentage and numbers of total neutrophils (CD11b+ Ly6G+) in bone marrow of WT mice upon CLP-induced sepsis (Fig. 1A and Supplementary Fig. 1A, B). Emergency granulopoiesis is characteristically featured with reduced bone marrow mature neutrophils (CD11b+ Ly6Ghigh) paralleled by increased bone marrow immature neutrophils (CD11b+ Ly6Glow) in relative percentage and absolute number12,20. Consistently, WT mice undergoing sepsis exhibited an increase in immature neutrophils and a significant decline in mature neutrophils in bone marrow when compared with unchallenged WT mice (Fig. 1B-D), representing the typical hallmarks of emergency granulopoiesis. In the physiological steady state, there was no difference in the percentage and numbers of bone marrow neutrophils between Alkbh5-deficient mice and WT littermates (Fig. 1B-D), suggesting that ALKBH5 has no impact on normal granulopoiesis. However, a significantly reduced immature neutrophils in bone marrow was observed in Alkbh5-deficient mice undergoing sepsis, since 12h after CLP, the early stage of bacterial infection (Fig. 1B, C). These data indicated the severely defective emergency granulopoiesis results from ALKBH5 deletion.
G-CSF and interleukin-3 (IL-3) are two major growth factors that stimulate normal granulopoiesis and emergency granulopoiesis during infection12,15,48. Interestingly, the levels of G-CSF and IL-3 were comparable in plasma (Fig. 1E), bone marrow (Fig. 1F), and peritoneal lavage fluid (Fig. 1G) of Alkbh5-deficient mice and WT littermates in the steady state and during sepsis, which excluded the possibility that the reduced ability to launch emergency granulopoiesis was due to altered production of both growth factors in Alkbh5-deficient mice. Altogether, ALKBH5 is required to induce emergency granulopoiesis in host defense response to bacterial infection.
ALKBH5 promotes neutrophil mobilization in response to bacterial infection
In response to infection, pool of neutrophils that mostly reside in the bone marrow under basal condition can be mobilized into the blood, thus providing a mechanism to rapidly deliver neutrophils to the sites of infection for host defense14,29. Interestingly, Alkbh5-deficient mice retained a higher percentage of leftover mature neutrophils in the bone marrow during sepsis (Fig. 1B, D), suggesting a defect in the egress of neutrophils from bone marrow to blood in Alkbh5-deficient mice. FACS analysis revealed a dramatic increase in the release of mature neutrophils from bone marrow into peripheral blood of WT mice during sepsis than steady state, yet this increase was far less profound in Alkbh5-deficient mice (Fig. 2A, B and Supplementary Fig. 2A). Moreover, the percentage and absolute counts of blood immature neutrophils and blood total neutrophils were markedly decreased in Alkbh5-deficient mice than WT littermates, once given CLP for 12h, the early stage of sepsis (Fig. 2A, C, D). Accordingly, Alkbh5-deficient mice displayed a substantially reduced mature, immature, and total neutrophils in the infected site, peritoneal cavity, during sepsis (Fig. 2E-G and Supplementary Fig. 3A).
The Chemokine C-X-C motif chemokine ligand 12 (CXCL12, also known as SDF-1) is a master retention signal for neutrophil storage in the bone marrow and disruption of CXCL12 leads to neutrophil mobilization into blood49. Yet Alkbh5-deficient mice produced comparable levels of CXCL12 in the bone marrow compared to WT mice in the steady state and undergoing sepsis (Fig. 2H). Similarly, loss of ALKBH5 did not alter the CXCL12 levels in plasma or peritoneal lavage fluid of unchallenged or CLP mice (Fig. 2I). Therefore, we excluded the possibility of disrupting the CXCL12 retention signal involved in the impaired neutrophil mobilization in Alkbh5-deficient sepsis mice.
ALKBH5 remodels transcriptional landscape in neutrophils to enable their generation and mobilization
Because of the unchanged levels of growth factors and retention signal but divergence in neutrophil generation and mobilization, we speculated that ALKBH5 might intrinsically program the gene expression in neutrophils. Then, we performed RNA-seq on bone marrow neutrophils isolated from Alkbh5-deficient mice and WT littermates given CLP. RNA-seq analysis showed that loss of ALKBH5 resulted in significant upregulation of 748 genes and downregulation of 371 genes in neutrophils (Fig. 3A). Gene Ontology enrichment analysis of the significantly differentially expressed genes (DEGs) indicated that leukocyte proliferation was one of the most significantly enriched biological processes in Alkbh5-deficient bone marrow neutrophils compared to WT cells (Fig. 3B, up). Many significantly DEGs also encompassed KEGG pathways related to neutrophils, including cytokine-cytokine receptor interaction and hematopoietic cell lineage (Fig. 3B, bottom). Notably, loss of ALKBH5 substantially downregulated transcript levels of several receptors of growth factor or cytokine that are critical for neutrophil generation and mobilization, such as Csf3r20,50,51 and Cx3cr116 (Fig. 3A, C). Conversely, ALKBH5 deletion increased mRNA levels of negative regulators of neutrophil generation, such as Socs2 (Fig. 3C). In addition, the transcript levels of two markers of mouse neutrophils, Ly6g and Itgam (also known as Cd11b), were reduced in neutrophils from Alkbh5-deficient sepsis mice (Fig. 3C), further confirming the reality of the data analysis.
As ALKBH5 is highly conserved between mouse and human52, we then confirmed the ALKBH5-driven transcriptional landscape in human neutrophils by using in vitro model of differentiated HL-60 neutrophil-like (dHL-60) cells with Escherichia coli (E.coli) infection. RNA-seq of ALKBH5-deficient and WT dHL-60 cells showed that ALKBH5 deletion significantly upregulated 1255 genes while downregulated 1035 genes in bacteria-infected human neutrophils (Fig. 3D). Most of genes that significantly differentially expressed in ALKBH5-deficient dHL-60 cells compared with WT cells were enriched in biological processes associated with neutrophil-mediated immune response, or in pathways related to granulopoiesis and neutrophil mobilization (Fig. 3E, F), which encompassed transcriptional signatures as similar as in bone marrow neutrophils from Alkbh5-deficient sepsis mice (Fig. 3B, C). These results further highlighted the critical role of ALKBH5 in endowing neutrophil generation and mobilization.
Based on the same KEGG pathways which were top enriched, we went to identify the downstream targets of ALKBH5 by integrative analysis of significantly DEGs that were overlapped in mouse and human neutrophils (Fig. 3G). Among four overlapped genes, the particularly interested one was Csf3r, whose transcript level was significantly downregulated by ALKBH5 deletion (Fig. 3A, D). Csf3r encodes G-CSFR, the central driver of granulopoiesis and neutrophil mobilization20,29,31,51. We then determined whether defective neutrophil generation and mobilization in Alkbh5-deficient mice resulted from reduced G-CSFR. FACS analysis showed that protein levels of G-CSFR were indeed downregulated on the cell-surface of bone marrow neutrophils from Alkbh5-deficient sepsis mice than WT littermates (Fig. 3H). Consistently, significantly decreased cell-surface G-CSFR levels were observed on ALKBH5-deficient dHL-60 human neutrophils upon bacterial infection (Fig. 3I). Together, ALKBH5 induces transcriptional program in both mouse and human neutrophils to intrinsically drive emergency granulopoiesis and neutrophil mobilization.
ALKBH5 upregulates cell-surface G-CSFR expression and strengthens its downstream signaling in neutrophils
G-CSFR is required for neutrophil responses by driving granulopoiesis and neutrophil mobilization, especially in antibacterial defense20,29,53. We therefore tested the effect of ALKBH5 on G-CSFR expression and its downstream signaling. FACS analysis revealed that expression of intracellular total G-CSFR protein was substantially reduced in ALKBH5-deficient dHL-60 cells during E.coli infection (Fig. 4A), which was consistent with our observation about declined G-CSFR protein on cell-surface of neutrophils (Fig. 3H, I). Upon LPS stimulation, ALKBH5-deficient neutrophils also exhibited significantly reduced protein levels of cell-surface and intracellular G-CSFR (Fig. 4B, C). These findings suggested that G-CSFR represents the downstream target of ALKBH5 and its protein expression can be upregulated by ALKBH5.
Upon binding to G-CSF, G-CSFR activates a signaling cascade of the downstream pathways such as JAK-STAT to induce neutrophil generation and egress from bone marrow into blood17,20,21. Noticeably, bone marrow neutrophils from Alkbh5-deficient sepsis mice exhibited defective phosphorylation of JAK2 and STAT3 (Fig. 4D). Besides, ALKBH5-deficient dHL-60 cells also displayed less response to G-CSF stimulation, having lower phosphorylation of JAK2 and STAT3 (Fig. 4E). These findings implied that ALKBH5 directly upregulates the protein expression of G-CSFR to increase its cell-surface level on neutrophils so as to mediate G-CSFR downstream JAK-STAT signaling, contributing to emergency granulopoiesis and neutrophil mobilization.
ALKBH5-mediated m6A demethylation enhances the mRNA stability of G-CSFR
Next, we explored the mechanism through which ALKBH5 upregulates G-CSFR expression. m6A enzymes modulate gene expression by altering m6A modification on its target mRNA to regulate RNA metabolic process, such as the stability and translation of mRNA35,36,54−56. RNA-seq data showed that ALKBH5 deletion significantly declined transcript levels of CSF3R in both mouse and human neutrophils upon bacterial infection (Fig. 3A, D). Consistently, qRT-PCR verified that Csf3r mRNA levels were indeed downregulated in bone marrow neutrophils from Alkbh5-deficient mice than WT littermates undergoing sepsis (Fig. 5A). Moreover, deletion of ALKBH5 significantly decreased the levels of CSF3R mRNAs in dHL-60 human neutrophils upon bacterial infection and LPS stimulation (Fig. 5B).
To address whether ALKBH5 could regulate the CSF3R mRNA levels via m6A RNA modification-dependent manner, we analyzed our previous data of transcriptome-wide m6A methylation profiling (m6A-seq) that performed on neutrophils, which exhibited a strong correlation between two biological replicates (Supplementary Fig. 4A). Notably, m6A-seq analysis revealed highly enriched and specific m6A peaks on CSF3R mRNAs in bacteria-infected neutrophils (Fig. 5C). Furthermore, m6A-RNA immunoprecipitation (RIP) combined with qRT-PCR verified that CSF3R mRNAs indeed contained m6A modification, and the m6A levels on CSF3R mRNAs were significantly accumulated in ALKBH5-deficient neutrophils than in WT cells (Fig. 5D). Therefore, CSF3R transcripts are m6A targets directly modulated by ALKBH5 in neutrophils. ALKBH5 acts as m6A demethylase and modulates gene expression by controlling m6A modification-mediated mRNA degradation57–59. Next, we performed RNA decay assay and confirmed that ALKBH5 deficiency markedly promoted the degradation of CSF3R mRNAs in neutrophils after transcription inhibition (Fig. 5E). Altogether, ALKBH5 specifically removes m6A methylation on CSF3R mRNAs to delay their mRNA decay, consequently increasing the mRNA levels and protein expression of G-CSFR in neutrophils.
Bacterial infection downregulates G-CSFR expression by inhibiting the binding of ALKBH5 to CSF3R mRNAs in neutrophils
Bacterial infection can trigger septic shock by desensitizing neutrophil response to G-CSF and impairing granulopoiesis60,61. So, we determined the effect of bacterial infection on G-CSFR expression in our infection model. As speculated, not only the cell-surface protein levels (Fig. 6A, B) but also the intracellular total protein expression (Fig. 6C, D) of G-CSFR were significantly decreased in WT neutrophils upon E.coli infection and LPS stimulation. Furthermore, qRT-PCR showed that E.coli infection and LPS stimulation decreased the CSF3R mRNA levels in WT neutrophils (Fig. 6E, F). Via RIP-qPCR assay, we confirmed that ALKBH5 protein was associated with high enrichment and direct binding of its target, CSF3R mRNAs, in neutrophils. Interestingly, the binding of ALKBH5 to CSF3R mRNAs was reduced during bacterial infection than that in the physiological state (Fig. 6G). The data indicated the possible pathological significance of disassociation or less binding of ALKBH5 to CSF3R mRNAs for insufficient G-CSFR expression in neutrophils, which makes neutrophils insensitive to G-CSF, resulting in disabled emergency granulopoiesis and neutrophil mobilization in severe infections.