Breeding pairs of C57BL/6 (WT) or IL-27Rα-deficient (KO) mice on a C57BL/6 genetic background were purchased from Jackson Laboratories (Bar Harbor, ME) and maintained under specific pathogen-free conditions in the vivarium at West Virginia University Health Sciences Center. Mice were maintained on a 12-h light/dark cycle and were fed/watered ad libitum. Male and female pups were used for experimental infection. Blood and tissues were collected from mice at the appropriate age by humane procedures.

Escherichia coli strain O1:K1:H7 was obtained from the ATCC (Manassis, VA) and grown in Luria broth from a single colony isolated on Tryptic Soy agar (TSA). To prepare infectious inoculums, the bacteria were enumerated as described previously (21).

Neonatal pups (n=3 per group (control vs. infected) per genotype (WT vs. KO)) at 4 days were inoculated subcutaneously in the scapular region with E. coli O1:K1:H7 using a 28-gauge insulin needle as described previously (20, 21). The bacteria were washed with PBS, centrifuged at 2,000 x g for 5 min, and suspended in a volume of PBS equivalent to an inoculum of 50 μL/mouse. The target inoculum was 10⁶ CFUs per mouse and actual inoculums as determined by standard plate counts, ranged from 4.8 x 10⁵ to 1.6 x 10⁶ CFUs per mouse. Vehicle (PBS)-inoculated pups were identified from E. coli-infected pups using a tail snip. The weights of mice were recorded immediately prior to infection, and at 24 h immediately prior to euthanasia. At 24 h post-infection, mice were harvested and individual mice (E. coli infected and tail-snipped PBS pups) were numbered; all downstream experiments were associated with the same pre-identified mice. Spleens isolated from pups were placed in PBS on ice. Blood was deposited in tubes that contained 5 μL of 500 mM ethylenediamine tetraacetate acid (EDTA) and placed on ice. The bacterial burden in the blood was enumerated by serial dilution and standard plating on TSA.

Four-day old C57BL/6 (n=10) and IL-27Rα-deficient (n=11) pups infected for 24 h were humanely euthanized and the spleens harvested as described above. To obtain sufficient cells for downstream analysis, spleens were pooled based on infection status (control vs. infected) to generate three control and three infected samples for sequencing from each genotype. Single cell suspensions were generated by crushing the spleens through a 40 µM strainer. Cells were placed in complete Dulbecco’s Modified eagle Medium (DMEM) with 30% fetal bovine serum (FBS). The myeloid-enriched fraction of mononuclear cells was isolated using Optiprep™ (Sigma) density gradient centrifugation as described previously and scaled down four-fold to accommodate the smaller number of cells in neonatal spleens (22). The resulting myeloid cells were washed, suspended in complete DMEM with 10% FBS, and counted (1-5x10⁵ cells per sample) before immediately proceeding to processing for ATAC and RNA sequencing.

Spleens were homogenized in TRI Reagent^® (Molecular Research Center, Cincinnati, OH). Using the commercial product protocol, the upper aqueous layer following phase separation was mixed with an equal volume of 75% ethanol and transferred to E.Z.N.A.^® RNA isolation columns (Omega Biotek, Norcross, GA). The manufacturer’s instructions were followed to complete tissue RNA isolation. RNA was isolated from approximately 7.5x10⁴ splenic myeloid cells using the RNeasy^® Plus Mini Kit (Qiagen, Hilden, Germany, Cat# 74134) according to the manufacturer’s instructions. RNA from cells or tissue was quantified using a Qubit Fluorometer and the RNA integrity was analyzed using an Agilent Bioanalyzer 2100 (Agilent technologies, Santa Clara, CA).

First-strand cDNA was synthesized using the iScript™ cDNA synthesis kit (Bio-Rad, Hercules, CA). Quantitative PCR reactions included cDNA diluted four-fold, gene-specific TaqMan^® primer probe sets (Applied Biosystems^®, Foster City, CA), and iQ™ Supermix (Bio-Rad, Hercules, CA). Cycling was performed in triplicate using a StepOnePlus™ Real-Time detection system (Applied Biosystems^®, Foster City, CA). Gene-specific amplification was normalized to GAPDH as an internal reference gene and expressed as log₂ relative gene expression compared to control spleen using the formula 2^-ΔΔCt.

Library preparation and sequencing were performed at the Genomics Core Facility of West Virginia University and Marshall University, respectively. RNA-Seq libraries were built using KAPA mRNA HyperPrep Kit (KAPA Biosystems, Wilmington MA) with 100-500 ng of total RNA according to manufacturer’s recommendation. The libraries were quantified using a Qubit Fluorometer, and the quality determined by Bioanalyzer High Sensitivity DNA Analysis (Agilent, Santa Clara, CA). Libraries were sequenced at the Genomics Core Facility at Marshall University with an Illumina NextSeq2000 (Illumina, San Diego, CA).

ATAC libraries were prepared from approximately 50 to 75 thousand of mononuclear cells using a Tagment DNA Enzyme Kit from Illumina (Cat # 20034198) by following the Omni-ATAC protocol (23). The ATAC libraries were run on 2% agarose gel and gel band corresponding to ~200–600 bp was cut to elute DNA fragments using the MinElute^® Gel Extraction Kit (Qiagen), subjected to sequencing at the Genomics Core Facility at Marshall University with Illumina Nextseq2000.

RNA-Seq data analysis followed our previous work (24). Pair-end RNA-Seq read were mapped to the mouse genome (mm10) by subread v2.0.1 with default parameters (25). The number of RNA-Seq fragments mapped to transcripts was summarized at gene level by featureCounts (26). Gene transcription level was quantified by RPKM (27). Differentially expressed genes were predicted by EdgeR3 (FDR < 0.05 and fold change > 2) (28). In this analysis, we required genes to be expressed at least in one of the compared conditions (RPKM>3 and log2 (count per million)>0). Meanwhile, we excluded genes exhibiting a substantial variation (maximal/minimal > 2) in expression across samples from the same condition. Gene Set Enrichment Analysis against Reactome pathways, Gene Ontology (biological processes), and Hallmark gene sets from MSigDB (29) were done by GSEA (30). Gene ontology enrichment analysis on biological processes (BP3) was done with the online DAVID bioinformatics resource (31), where non-redundant hits were visualized as bubble plots by REVIGO (32).

For ATAC-seq data analysis, sequencing reads was aligned to mouse genome (mm10) by Bowtie2 (33). Visualization of reads density across selected gene loci was done by IGV (34). ATAC-seq read enriched regions were identified by MACS3 (q-value < 0.001) (35). Differentially accessible regions (DARs) were predicted by EdgeR3 (FDR < 0.05 and fold change > 1.5) (28). Target genes of DARs were predicted by the online tool GREAT (36).

Integrative analysis comparing E. coli-induced expression changes between WT spleen and macrophages isolated from the spleen (Compromised transcriptome response to E. coli in macrophages in the absence of IL-27Ra section) was done with GSEA. In this analysis, we focused on genes identified as up-regulated or down-regulated by E. coli infection in WT macrophages. The GSEA evaluates their overall ranks on a red-to-blue spectrum which represents E. coli-induced expression changes in the spleen (red: higher expression after infection; blue: lower expression after infection). An overall rank to the left side of the spectrum indicates an upregulation in the spleen, while an overall rank to the right side indicated an overall down-regulation.

Weight, bacterial burden, and gene expression data (Mean ± SEM) were analyzed with two-way ANOVAs for each given dependent measure using Prism 8 (GraphPad, San Diego, CA). The threshold for statistical significance was set at alpha = 0.05.

Neonates exhibit elevated levels of IL-27 in the spleen and blood at resting state relative to adults, and these levels continue to rise during infection peaking at 24 h (7, 20). IL-27Rα^-/- (KO) mice have an allelic exchange of a neomycin resistance cassette for exon 2 in the C57BL/6 background and fail to produce a functional receptor protein ( Supplementary Figure 1 ). These mice exhibit superior control of bacteria and improved outcomes during infection (20). To understand how the transcriptome may be influenced by IL-27 and further identify a transcriptional signature associated with improved host response, we infected WT or IL-27Rα^-/- neonatal mice at day 4 of life as described previously ( Figure 1A ). These mice were euthanized at day 1 post-infection, a time that corresponds with peak IL-27 levels and a critical period for control of bacteria (20). Phenotypic changes are observed in our neonatal sepsis model that are consistent with an increase in IL-27 gene expression during systemic infection ( Figure 1B ). Expression of the IL-27p28 subunit is considered rate limiting for assembly of the bioactive cytokine; EBI3 expression is more constitutive and also not exclusive to IL-27. IL-27Rα^-/- (KO) neonates that fail to respond to IL-27 exhibit improved maintenance of weight compared to WT neonates ( Figure 1C ). Relative to KO pups, WT infected pups exhibited a 16.96% decrease in weight compared to their uninfected WT littermates 1 day-post-infection; this decrease in weight between infected and uninfected KO pups was not observed ( Figure 1C ). This is further consistent with differences in pathogen burden ( Figure 1D ). Infected WT pups demonstrated a profound increase in bacterial burdens in the blood 1-day post-infection. Infected KO pups exhibited significantly fewer CFUs in the blood than infected WT pups ( Figure 1C ).

To understand the impact of IL-27 on regulation of the neonatal host response to infection, we profiled the transcriptional response in uninfected control and E. coli O1:K1:H7-infected neonates at 1-day post-infection. This period is consistent with peak serum IL-27 levels in our model of neonatal sepsis (20). The expression profiles in infected pups show clear separation according to principal component analysis (PCA) ( Figure 2A ). Although the magnitude and abundance of genes that are differentially increased in expression during infection is greater, there is also a clear set of genes that are differentially decreased in expression ( Figure 2B ; Supplementary Table 1 ). According to Gene Set Enrichment Analysis (GSEA) against Reactome (37), the most up-regulated gene set pathways are associated with inflammation, cytokine signaling, and G protein coupled receptor ligand binding and signaling ( Figure 2C ). The results are not unexpected as an intense inflammatory response ensues during neonatal sepsis that frequently manifests as a “cytokine storm” and a potentially pathological level of inflammation. The top ten individual genes with the greatest magnitude of increased expression in one top gene set “cytokine signaling in immune system” are all cytokines and chemokines ( Figure 2D ). Among these cxcl2, ccl4, ccl3, and ccl2 are all chemotactic factors involved with recruitment of neutrophils and monocytes to sites of inflammation. Although speculative, it seems likely that IL-10, an anti-inflammatory factory, is increased in expression as a product of a robust inflammatory response and part of a negative feedback inflammatory loop ( Figure 2D ). Similarly, the top up-regulated biological processes identified by gene ontology analysis aligned with response to bacterial and viral infection, cytokine signaling, and inflammatory response ( Figure 2E ). Many of the top ten individual genes (i.e. cxcl2, csf3, ccl3, il1a, ccl2, il10, and tnf) with the greatest magnitude of increased regulation in expression in the biological process “cellular response to molecular of bacteria origin” ( Figure 2F ) overlap with leading genes from GSEA analysis on “cytokine signaling in immune system” ( Figure 2D ). Similar to IL-10 above, cis-aconitate decarboxylase (acod1) is a negative regulator (38, 39) of inflammatory responses and may be upregulated to counterbalance the robust inflammatory response to bacteria ( Figure 2F ). However, acod1 has also been shown to contribute to antimicrobial activity of macrophages by catalyzing itaconic acid production (39). TNFAIP3-interacting protein 3 (TNIP3) also inhibits NF-κB-dependent gene expression in response to LPS (40), most likely a host attempt to regulate prolonged inflammatory signaling.

REACTOME pathways were less intuitive in IL-27 biology ( Supplementary Figures 2A, B ). Elastic fibers are scaffolded and crosslinked to form an insoluble polymer that provides tissue elasticity and resistance to mechanical strain (41). Defects in elastic fiber formation result in a variety of tissue pathologies (41). This is the first report of an influence of IL-27 on this gene set pathway. Genes with functional roles in transcriptional regulation by small RNAs may have implications in host control of the inflammatory response, a consistent theme in the infected transcriptome ( Supplementary Figure 2A ). The expression of these genes may be negatively regulated by the persistent and increasing bacterial burden. Biological processes identified by gene ontology enrichment analysis downregulated during infection show significant enrichment in genes that align with DNA replication and cell cycle pathways ( Supplementary Figures 2C, D ). Although speculative, cellular damage amidst a strong inflammatory response and high bacterial burden may oppose cellular replication pathways.

A prior report by Li, et al. (42) explored changes to the liver transcriptome during E. coli-induced adult murine sepsis. Genes that were differentially expressed during infection compared to control included sets that were both increased and decreased ( Supplementary Figure 3A ). Not unexpectedly, biological processes associated with innate immune function and host response to infection were up-regulated through GSEA ( Supplementary Figure 3B ). Down-regulated biological processes were enriched for those involved in central metabolism ( Supplementary Figure 3C ). This finding is also not surprising given the role of the liver in energy metabolism. Further analysis revealed that 222 up-regulated genes and 35 down-regulated genes in liver exhibited consistent expression changes in our study in spleen ( Supplementary Figure 3D ). The shared up-regulated genes reveal a common emphasis in biological processes associated with innate and adaptive immune response, wound healing, cytokine production, cellular adhesion, and signal transduction to name a few ( Supplementary Figure 3E ).

IL-27Rα^-/- mice do not express a functional receptor and cannot respond to IL-27. The change in global gene expression with IL27Rα depletion compared to WT in the absence of infection is not dramatic and reveals only a limited number of differentially expressed genes ( Figure 3A ; Supplementary Table 1 ). Of importance at the gene set level, GSEA analysis revealed a subset of genes involved in inflammatory response and cholesterol homeostasis are preferentially downregulated in the absence of the IL27Rα ( Figures 3B, C ). Overall, it is not clear at this time what the connection may be with the downregulated inflammatory response genes and outcomes reported during infection, but some may be part of a reprogramming of the host response to be less reactive. Regulation of cholesterol homeostasis by IL-27 may have important metabolic implications that are relevant to neonatal physiology. That expression of these genes was lower in IL-27Rα-infected pups may suggest a link with other metabolic pathways and improved maintenance of weight observed during infection. The low-density lipoprotein receptor (ldlr, Figure 3C ) that can potentiate inflammatory responses (43) and is increased in expression in WT pups, is also interrelated with cholesterol homeostasis (44). These data describing transcriptional regulation by IL-27 signaling in the absence of infection are novel findings.

Since the number of IL-27Rα-dependent genes was marginal at base level and we observed phenotypic differences in neonates during infection, we further explored the transcriptional response during systemic E. coli infection. There were 634 genes that were differentially expressed between WT and KO neonates during infection; 271 were significantly upregulated and 363 were downregulated with IL-27Rα deficiency ( Figure 4A ; Supplementary Table 1 ). According to GSEA, the most preferentially downregulated pathways and biological processes in the KO pups are associated with cytokine and chemokine signaling ( Figure 4B ) and response to cytokines or bacteria ( Figure 4C ). Collectively, this highlights a less inflammatory environment that is not expected given the anti-inflammatory/suppressive activity of IL-27. However, it is consistent with a lower bacterial burden ( Figure 1D ). This suggests that improved bacterial clearance in the absence of IL-27 signaling does not require a heightened inflammatory state, and instead, there is a direct relationship between IL-27 signaling and bacterial killing. The most preferentially upregulated pathways and biological processes in the KO pups are less intuitively involved in host response to infection with a central theme of DNA replication, maintenance, and repair ( Supplementary Figures 4A, B ). This may be reflective of overall improved physiology in KO pups that is disrupted in WT animals.

Further analysis indicated that genes upregulated by E. coli infection were generally downregulated post-infection in the absence of IL-27Rα with a similar observation made for genes downregulated by E. coli infection ( Figure 4D ). We confirmed the observations by visually inspecting gene expression changes prior to and after infection with E. coli in the WT and KO; 363 genes failed to increase, and 271 genes failed to decrease in the KO group following infection ( Figure 4E ). Among the genes that fail to increase during infection in the absence of IL-27 signaling are chemokines, chemokine receptors, and interleukins ( Figure 4F ; Supplementary Figure 4C ). This further supports a less inflammatory environment in the infected KO compared to WT neonates. Our data further suggests that this is the product of a transcriptional reprogramming that occurs in WT but not KO neonates during infection. Expression of genes for notable transcriptional activators such as Stat-2, Batf, XBP1, and CEBP-δ that regulate host inflammatory responses are not increased in the spleen of IL-27Rα-deficient pups ( Figure 4G ). Among TFs that failed to down-regulated their expression levels ( Figure 4G ), E2f1 is well-characterized for its functions in regulating cell cycle (45). Collectively these data demonstrate an improved response to infection that is not reliant upon heightened levels of inflammation as have been shown with other mouse models of chronic tuberculosis (46, 47). This offers new promise to the potential of antagonizing IL-27 as a host-directed therapy for neonates without concern for driving inflammatory responses that are pathological to host tissues.

Improved control of bacteria was observed in infected KO pups, and we have previously demonstrated that IL-27 impairs macrophage control of a variety of bacteria, including E. coli (7, 20, 48–51). We therefore isolated the myeloid fraction of cells enriched with macrophages from the spleens of WT and IL-27Rα KO neonatal pups challenged with E. coli. Gene expression data analysis with RNA sequencing identified dozens of differentially expressed genes in response to E. coli infection in WT mice ( Figure 5A ; Supplementary Table 1 ). We aligned the differentially expressed genes defined from the macrophage population to the E. coli-induced expression changes at the spleen tissue level using GSEA and observed consistent changes in expression ( Figure 5B ). As expected, E. coli infection upregulated IL-10 signaling in macrophages, though the level of significance was as modest as FDR q = 0.096 ( Figure 5C ). Interestingly, compared to WT mice, the expression changes induced by E. coli infection were generally smaller in the KO mice ( Figure 5D ), suggesting that in the absence of IL-27Rα the transcriptional program of macrophages became less responsive to E. coli infection.

To understand potential epigenetic regulatory mechanisms, we extended the analysis to profile chromatin accessibility in the same sets of macrophage-enriched cells for which RNA sequencing was performed. The analysis identified genes associated with multiple genomic regions that showed an increase or decrease in chromatin accessibility in response to E. coli. Examples include s100a9 and mrc1 ( Supplementary Figure 5A ). Proteins in the S100 family are among the most abundant cytoplasmic proteins in myeloid-differentiated cells that are released during activation of phagocytes and known to play a significant role in the pathogenesis of sepsis (52–54), while MRC1 is a C-type lectin receptor and expression is generally considered a marker of a more M2-like or regulatory macrophage polarization (55). As expected, genes associated with regulatory regions that increased in accessibility were up-regulated in expression level ( Supplementary Figure 5B ), while those associated with decrease in accessibility were down-regulated ( Supplementary Figure 5B ). Consistent with expression changes, in the absence of IL-27Rα, E. coli-induced changes in chromatin accessibility were generally smaller than in the WT mice ( Supplementary Figure 5C ).