Regnase-1 Deficiency Restrains Klebsiella pneumoniae Infection by Regulation of a Type I Interferon Response

ABSTRACT Excessive inflammation can cause tissue damage and autoimmunity, sometimes accompanied by severe morbidity or mortality. Numerous negative feedback mechanisms exist to prevent unchecked inflammation, but this restraint may come at the cost of suboptimal infection control. Regnase-1 (MCPIP1), a feedback regulator of IL-17 and LPS signaling, binds and degrades target mRNAs. Consequently, Reg1 deficiency exacerbates autoimmunity in multiple models. However, the role of Reg1 in bacterial immunity remains poorly defined. Here, we show that mice deficient in Reg1 are resistant to Klebsiella pneumoniae (KP). Reg1 deficiency did not accelerate bacterial eradication. Rather, Reg1-deficient alveolar macrophages had elevated Ifnb1 and enrichment of type I IFN genes. Blockade of IFNR during KP infection reversed disease improvement. Reg1 did not impact Ifnb1 stability directly, but Irf7 expression was affected. Thus, Reg1 suppresses type I IFN signaling restricting resistance to KP, suggesting that Reg1 could potentially be a target in severe bacterial infections.

macrophages from KP-infected mice showed striking enrichment of type I IFN-associated gene pathways during Reg1 deficiency. Consistent with this, blockade of IFNR signaling reversed the protective effects of Reg1 deficiency. Mechanistically, Reg1 regulated the stability of mRNA encoding IRF7, an upstream regulator of type 1 IFN gene expression by TLRs.

RESULTS
Reg1-deficient mice are resistant to KP lung infection. A complete absence of Reg1 is extremely deleterious, as Reg1 2/2 mice exhibit severe inflammation and severely shortened life spans (9). In contrast, Reg1 1/2 mice (sometimes termed here Reg1-deficient) have a normal life span without peripheral organ inflammation. Nonetheless, Reg1 1/2 mice show enhanced pro-inflammatory responses in multiple IL-17-driven autoinflammatory model settings (11,15,16). To determine whether Reg1 haploinsufficiency influences immunity to KP infection, Reg1 1/2 and Reg1 1/1 littermate controls were infected with KP by oropharyngeal aspiration. Survival and parameters of health including weight loss were monitored over time. Indeed, infection-induced survival and weight loss were significantly ameliorated in Reg1-deficient mice compared with Reg1 1/1 littermate controls ( Fig. 1A and B). Unexpectedly, this improvement in survival was not accompanied by statistically significant differences in lung bacterial burdens (Fig. 1C). Moreover, there were no changes in bacterial dissemination into the spleen between groups at any measured time point, even though Reg1 1/2 mice had lower expression of Reg1 (Zc3h12a) than controls (Fig. 1D, E). Furthermore, histologic analysis of infected lung tissues between Reg1 1/1 and Reg1 1/2 animals, showed that Reg1 1/2 mice developed smaller foci of pneumonia compared with controls, with particularly reduced levels of parenchymal inflammation (Fig. 1F, G). These data suggest that factors other than bacterial burden underlie the improved survival of Reg1-deficient animals.
Reg1 functions in both hematopoietic and non-hematopoietic compartments during KP pneumonia. Reg1 has been shown to act in multiple cell types, including T cells, macrophages, and epithelial cells (7,23). Of relevance to KP, Reg1 can restrict signaling by both TLR4 and IL-17. Whereas TLR4 acts predominantly on hematopoietic cells (especially macrophages), IL-17 predominantly mediates signaling in epithelial and/or mesenchymal target tissues, demonstrated in many settings including KPinfected lung (22). As a first step to define the essential compartments in which Reg1 functions to limit immunity to pulmonary KP infection, we used an adoptive transfer approach. Femoral bone marrow (BM) from Reg1 1/2 (CD45.2) or WT (CD45.1) mice were transferred into reciprocal irradiated Reg1 1/2 or WT recipients. After 8 weeks, engraftment was confirmed by flow cytometry (data not shown). Successfully reconstituted mice were infected with KP by oropharyngeal aspiration and followed up to 7 days. As expected, WT mice that received WT BM cells were susceptible to KP, whereas Reg1 1/2 mice receiving Reg1 1/2 BM were more resistant. As shown, Reg1 1/2 or WT mice that received Reg1 1/2 BM cells were resistant to KP, with almost 80% survival at day 7 postinfection. Additionally, Reg1 1/2 mice that received WT BM cells were resistant to KP, showing 60% survival at day 7 ( Fig. 2A). Thus, Reg1 appears to act in multiple compartments.
Loss of IL-17RA in pulmonary epithelial cells renders mice highly susceptible to KP infection (22). Accordingly, given the potent impact of Reg1 on IL-17 signaling seen in prior studies (11,12,15,24), we crossed Reg1 fl/fl to surfactant protein C (Sftpc) Cre mice in order to delete Reg1 conditionally in distal pulmonary epithelial cells (distal bronchi and alveoli). As shown, mice were modestly more resistant to KP than controls, though the improvement survival was much less profound than seen in Reg1 1/2 mice (Fig. 2B), which is consistent with the BM chimera data showing contributions from both the hematopoietic and non-hematopoietic compartments contribute upon KP infection.
Reg1 deficiency does not influence immune cell recruitment or proliferation during KP infection. Based on these data, it is evident that a Reg1 deficiency in the hematopoietic compartment provides a survival advantage to the host. We saw increased percentage of alveolar macrophages at 72-h postinfection (Fig. 3A). However, there were no changes in the absolute numbers of recruited myeloid cells between Reg1 1/2 and Loss of Regnase-1 Promotes Resilience to Klebsiella pneumoniae ® control mice at 48-h and 72-h postinfection (Fig. 3B). Consistent with this, expression of myeloid-recruiting chemokines such as Cxcl1, Cxcl5, and Ccl2 were similar among groups, despite the fact that these are all known transcriptional targets of Reg1 in other settings ( Fig. S1A) (8,25). There were also no differences in cellular proliferation and macrophage bacterial killing between Reg1 1/2 and control mice ( Fig. 3B and C). It is known that Reg1 controls B cell homeostasis and that complete Reg1 deletion in B cells increases antibody secretion (26). However, IgM and IgA levels in BALF and lung tissue were comparable between Reg1 1/2 and control mice (Fig. 3D). Taken together, these data suggest that differences in survival during KP infection in Reg1 1/2 mice is not explained by altered recruitment of myeloid cells, cellular proliferation, macrophage killing capacity, or a B cell antibody response. Trevejo-Nuñez et al.

®
Resistance to KP caused by Reg1 deficiency is linked to increased interferon signature. To understand the mechanisms by which Reg1 deficiency promotes survival, we performed transcriptomic profiling of purified alveolar macrophages from Reg1 1/2 and littermate control lungs at 24-h post-KP infection. This time point was chosen to define the early events that are operative during the initial stages of infection. There were 796 differentially expressed genes, of which 237 were upregulated and 559 were downregulated. Gene set enrichment analysis (GSEA) revealed major changes in type I IFN-related responses (IFN-a), oxidative phosphorylation, and IFN-g response as the top three enriched gene sets, with a normalized enrichment score (NES) ,2.5 for the type I IFN group ( Fig. 4A and B).
We next validated these pathways by functional analysis. Consistent with the RNASeq data, type I IFN (IFN-b) expression was higher in lung homogenates of Reg1deficient mice compared with controls at 24-h and 48-h postinfection (Fig. 5A). Because IFN-b can be produced by macrophages during Gram-negative pneumonia (27), we stimulated Reg1 1/2 or control BMDMs with heat-killed KP and assessed Ifnb1 mRNA. Indeed, Ifnb1 was elevated in Reg1-deficient macrophages compared to controls as early as 2-h postinfection (Fig. 5B). In contrast, IFN-g expression was similarly expressed between Reg1 1/2 and Reg1 1/1 mice after KP infection (Fig. 5C). Similarly, oxidative phosphorylation, of sorted alveolar macrophages assessed by mitochondrial respiration, was not different between Reg1 1/2 mice and controls (Fig. S1B,C). Collectively, these data indicate that Reg1 deficiency induces IFN-b upon KP infection.
Reg1 functions primarily by promoting endonucleolytic decay of target mRNA transcripts (28). Based on the observed increased Ifnb1 expression in Reg-deficient BMDMs, we hypothesized that Reg1 deficiency may result in prolonged stabilization of Ifnb1 mRNA or regulatory factors upstream of Ifnb1 regulation. To test this, Reg1 1/2 and control BMDMs were stimulated with LPS to prime expression of genes in the IFN pathway. Cells were treated with actinomycin D (Act D) to block further transcription, and the half-life of candidate target transcripts was assessed over a 2-h time course. Although the intrinsic mRNA stability of Ifnb1 or other mRNAs was not detectably altered in Reg1-deficient cells, the half-life of Irf7 mRNA considerably increased in the setting of Reg1 deficiency (Fig. 5D), potentially explaining the increased IFN-b seen in Reg-1 deficient mice upon KP pneumonia. Loss of Regnase-1 Promotes Resilience to Klebsiella pneumoniae ® To determine whether type I IFN accounted for the prolonged survival in Reg1 1/2 mice, we administered anti-IFNR1 Abs or isotype controls at days 0 and 2 post-KP infection and followed mice for 7 days. Strikingly, 100% of Reg1 1/2 mice that received isotype control Abs survived, while 60% of Reg1 1/2 mice that received anti-IFNR1 succumbed (Fig. 6). Therefore, the protection afforded by Reg1-deficiency requires type I IFN signaling.
Increased Ifnb1 correlates with enhanced anti-inflammatory response in Reg1deficient mice. IFN-b has the capacity to exert anti-inflammatory effects in the host, reported in several disease models. This effect has been attributed in part to type I IFN induction of IL-10, which dampens expression of transcription of cytokines such as Tnfa, Cxcl1, and Il6 (29, 30). Based on this, we assessed IL-10 in lung homogenates of KP-infected mice. There was a trend to increased IL-10 at 72-h postinfection in Reg1 1/2 mice compared with control mice, though this did not reach statistical significance (Fig. 7A). TNF-a was not affected Reg1 1/2 compared with controls (Fig. 7B). We then asked whether apoptosis was affected in Reg1 1/2 mice compared with controls. As shown, cleaved caspase 3 expression in total lung homogenates was increased in Reg1 1/2 mice compared with controls ( Fig. 7C and D). These data suggest that increased IFN-b arising from Reg1 deficiency may confer an anti-inflammatory advantage to Reg1 1/2 mice sufficient to prolong survival upon KP infection.

DISCUSSION
Reg1 controls the inflammatory response by degrading many individual inflammatory cytokine mRNAs, including Il6, Il1b, and Il12b, thus tempering the overall inflammatory milieu (9). Perhaps even more significantly, Reg1 degrades transcripts encoding Loss of Regnase-1 Promotes Resilience to Klebsiella pneumoniae ® inflammatory transcription factors (TF) such as Nfkbiz, which encodes the non-canonical TF Ik Bj , and therefore Reg1 can indirectly affect all Ik Bj -inducible genes (11). Thus, it is not unexpected that a complete knockout of Reg1 gene (Zc3h12a) is fatal due to severe autoinflammation. Reg1 1/2 mice, on the other hand, appear to have a normal life span, fertility, and do not exhibit peripheral baseline inflammation. However, Reg1 1/2 mice still have an enhanced inflammatory response, particularly when the disease model relies heavily on the IL-17R pathway (11,15,24). In this KP infection model, we were surprised that levels of inflammatory cytokines and chemokines were similar between Reg1-deficient mice and controls. Because these mice are haploinsufficient for the Reg1 gene (Zc3h12a), and KP is sensed by TLR4, one potential explanation is the remaining Reg1 levels may be enough to compensate for the cytoplasmic MyD88-driven pathway, but not the endosomal TRIF-dependent pathway, thus leading to increased type I IFN expression and protection from KP. Given that most pro-inflammatory genes known to be driven by Reg1 were unchanged in lung tissues regardless of Reg1 deficiency, and there was a clear increase in the percentage of alveolar macrophage (AM) recruitment, we performed RNAseq in AMs cells in order to evaluate macrophage-intrinsic activities underlying the Reg1-deficient phenotype. This approach has the caveat that contributions from other cells such as interstitial macrophages, inflammatory monocytes, or other immune cells may be missed. In this regard, CCR2 1 inflammatory monocytes recruited during KP infection are known to enhance IL-17 production from ILC3 cells leading to KP eradication (31,32). However, this mechanism has been shown to be important with clinical strains of KP and not with the serotype used here (ATCC 43816). Future studies will focus on additional cell types to determine more broadly where Reg1 is operative in this setting.
The role of type I interferons (IFN-a/IFN-b) in bacterial infections is not fully defined, especially when compared with its extensively-studied roles in viral infections. Even among bacterial pathogens, the response to type I IFN differs depending on site of  Loss of Regnase-1 Promotes Resilience to Klebsiella pneumoniae ® infection and pathogen characteristics. For instance, type I IFN is detrimental in animal models of Staphylococcus aureus and Listeria monocytogenes (33,34). On the other hand, there is a protective response in models of Legionella pneumophilia and Streptococcus pyogenes (35). In the case of Klebsiella pneumoniae, IFN-b signaling in NK cells enhances secretion of IFN-g, decreasing bacterial burden (36). In this study, we did not see differences in recruitment of NK cells or enhanced IFN-g secretion by NK cells in Reg1 1/2 mice compared with controls. However, the enhanced mortality in control WT mice is significantly decreased in Reg1 1/2 mice, which we propose is due to increased type I IFN that modulates the inflammatory response enough to facilitate bacterial eradication and pneumonia resolution, and this model is supported by histological evidence as well as the observation that anti-IFNR1 blockade reverses the protection provided by Reg1 deficiency.
IFN-b can lead to many complex events in bacterial infections, including increased apoptosis, macrophage efferocytosis and resolution of infection in models of Escherichia coli pneumonia and peritonitis, associated with enhanced IL-10 secretion (27). IL-10 mediates many anti-inflammatory effects and regulates metabolic reprograming of macrophages (37). Related to this, interstitial macrophages have immunoregulatory properties by secreting IL-10 upon LPS and CPG-DNA stimulation in models of asthma (38). Although there was only a modest trend of increased IL-10 in these settings, there was a clear increase in cleaved Caspase-3 in Reg1-deficient lungs, correlating with increased IFN-b and resistance to KP pneumonia. Concomitantly, transcriptomic profiles revealed several IFN-stimulated genes (ISGs) implicated in regulating apoptosis, including Ifit2, Ifit3, and Ifitm3. Intriguingly, IFIT2 is an RBP that enhances apoptosis (39,40), though its precise role in the context of Reg1 immunoregulation is as yet unknown. Thus, we speculate one mechanism underlying these results is through actions of type I IFN mediating increases in apoptosis, an important step for resolution of pneumonia (41).
Because IFN-b levels were increased in Reg1-deficient mice upon KP infection, we initially expected that Reg1 deficiency would result in enhanced stability of the Ifnb1 transcript. However, our data instead indicate that the impact on Ifnb1 appears to be indirect via control of its upstream regulator Irf7 mRNA, opening a new facet of how Reg1 influences immune responses. A deficiency in a related RBP, Regnase-3 (Zc3h12c) also leads to increased IFN (type I and II) signaling, and IRF7 can transcriptionally control expression of Regnase-3 (42). However, Reg3 expression was not altered in AMs based on the RNAseq data or in total lung by qPCR, and thus we believe this axis is not a major driver of the effects seen here (Fig. S2D).
The type III interferons have emerged in recent years as key regulators of antiviral and antifungal immunity (43). It has been shown that IFN-lambda (L) increases lung epithelial permeability and facilitates bacterial transmigration in animal models infected with KP (KP ST258), an effect that is counteracted by IL-22 (44). In this model we did not see increases in IFN-L despite increase of type I IFN and in vitro stability of IRF-7, though determining whether this axis of interferon activity contributes in any way will require further study (Fig. S2C). Surprisingly, although IL-22 and IL-17 signaling are well described to play important roles in KP eradication (20)(21)(22), their expression was also not altered by Reg1 deficiency (Fig. S2A,B). This would be consistent with activities of Reg1 being downstream of these or other cytokines rather than upstream of their production, though further analyses will be required to prove that point definitively.
The immune system has evolved to balance the vital effects of anti-microbial effector functions with the potential for causing collateral tissue damage. In this regard, the activation of every immune signaling pathway is accompanied by negative feedback signaling events that restrain inflammation (45,46). In selected conditions, however, it may be clinically beneficial to allow more fulminant inflammation in order to treat a life-threatening condition. Indeed, checkpoint inhibitor blockade for cancer therapy is built on exploiting this concept (47). By analogy, releasing inflammatory "brakes" may be useful in the context of severe infections such as bacterial pneumonia, and the present data suggest that Reg1 could, in principle at least, be one such target.

MATERIALS AND METHODS
Mice. Reg1 1/2 and Reg1 1/1 (Zc3h12a) littermates on the C57BL/6 background were cohoused and used for all experiments. Reg1(Zc3h12a) fl/fl mice are under material transfer agreement (MTA) from University of Central Florida (Orlando, FL, USA). Sftpc Cre mice and CD45.1 mice were from The Jackson Laboratory. Mice were 8 to 12-weeks old and both sexes were used. All animal studies were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Pittsburgh.
Bacterial infections and anti-IFNR1 treatment. KP ATCC strain 43816 was grown in tryptic soy broth to reach early log phase. 1-2 Â 10 3 CFU/mouse in PBS was administered by deep oropharyngeal aspiration. Where indicated, tamoxifen (TAM) was administered i.p. at 75 mg/kg dissolved in corn oil for 5 days and then rested for 5 days prior to induction of infection. Reg1 1/2 mice were treated with anti-IFNR1 antibodies or isotype control by i.p. injection (Bio-Xcell MAR1-5A3, 250 mg/mouse, administered on day 0 and day 2 p.i.). Mice were followed for 7 days.
Bacterial burden, mRNA, and protein analysis. Tissues were homogenized in PBS and CFU levels were assessed by serial dilution plating. Lung tissues were homogenized in TRIzol (Invitrogen) and subjected to qPCR with SybrGreen probes. Threshold cycle (C T ) values were normalized to Gapdh. ELISA kits were from eBioscience (Thermo Fisher Scientific) and R&D Systems. Abs used in Western blots were cleaved caspase-3, total caspase 3 (Cell Signaling) and beta-actin (Abcam).
RNA sequencing and analysis. Samples from 12 individual mice (uninfected Reg1 1/1 n = 3, uninfected Reg1 1/2 n = 3, KP-infected Reg1 1/1 n = 4, KP-infected Reg1 1/2 n = 4) were used. Mice were infected with KP by oropharyngeal aspiration. After 24 h, alveolar macrophages were stained as described above and sorted by flow cytometry with a purity of 99%. Cells were placed into RLT-plus (Qiagen, Valencia, CA) and total RNA extracted using RNeasy MiniKits (Qiagen). RNA was quantitated using Nanodrop and integrity determined with a total RNA Nano Chip (Agilent Technologies). Singlestranded total RNA-seq libraries were sequenced with an Illumina Nextseq500 sequencer with a depth of 25 million reads per sample (75 bp single-end) at the University of Pittsburgh Health Sciences Sequencing Core. Fastq files with high quality reads (phred score .30) were uploaded to the CLC Genomics Workbench (Qiagen) and reads aligned to the mouse reference genome. Transcript counts and differential expression analyses were carried out using the CLC Genomics Workbench. RNAseq data were deposited to Sequence Read Archive (SRA), BioProject ID: PRJNA789160.
Statistical analysis. Data were analyzed on Prism (GraphPad). Data were analyzed by log-rank, oneway analysis of variance (ANOVA), Student's t test, and post hoc tests were used as indicated. Each symbol represents one mouse unless indicated.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. ACKNOWLEDGMENTS G.T.N. was supported by NIH grants HL135476, HL154231, AI153549. K.C. was supported by HL137709. S.L.G. was supported by AI147383. We thank B.M. Coleman for