Contribution of increased ISG15, ISGylation and deregulated type I IFN signaling in Usp18 mutant mice during the course of bacterial infections.

Host genetics has a key role in susceptibility to Salmonella Typhimurium infection. We previously used N-ethyl-N-nitrosourea (ENU) mutagenesis to identify a loss-of-function mutation within the gene ubiquitin-specific peptidase 18 (Usp18(Ity9)), which confers increased susceptibility to Salmonella Typhimurium. USP18 functions to regulate type I interferon (IFN) signaling and as a protease to remove ISG15 from substrate proteins. Usp18(Ity9) mice are susceptible to infection with Salmonella Typhimurium and have increased expression and function of ISG15, but Usp18(Ity9) mice lacking Isg15 do not show improved survival with Salmonella challenge. Type I IFN signaling is increased in Usp18(Ity9) mice and inhibition of type I IFN signaling is associated with improved survival in mutant mice. Hyperactivation of type I IFN signaling leads to increased IL-10, deregulated expression of autophagy markers and elevated interleukin (IL)-1β and IL-17. Furthermore, Usp18(Ity9) mice are more susceptible to infection with Mycobacterium tuberculosis, have increased bacterial load in the lung and spleen, elevated inflammatory cytokines and more severe lung pathology. These findings demonstrate that regulation of type I IFN signaling is the predominant mechanism affecting the susceptibility of Usp18(Ity9) mice to Salmonella infection and that hyperactivation of signaling leads to increased IL-10, deregulation of autophagic markers and increased proinflammatory cytokine production.

INTRODUCTION

Salmonella enterica are facultative, intracellular, gram-negative enterobacteria that cause a range of enteric diseases in mammalian hosts. The human restricted serovars Salmonella enterica Typhi and Paratyphi are the causal agents of Typhoid fever, affecting more than 27 million people worldwide and resulting in greater than 200 000 deaths each year through contaminated food and drinking water (1). In contrast, infection with the Salmonella enterica serovar Typhimurium results in a self-limiting gastroenteritis in humans, but in mice, is an established model of fatal systemic disease, whereby infection leads to dissemination of the bacteria to the spleen and liver and activation of both innate and adaptive immune responses. Typhoid fever remains an important global health issue due to geographic spread as a result of foreign travel to areas of endemicity, including Africa and Asia.

The outcome and severity of infection with Salmonella is dependent on several parameters including microbial virulence factors, environment, immune status and host genetics. Numerous quantitative trait loci influencing microbial pathogenicity have been discovered from the inherent differential susceptibility of inbred mouse strains (2). However, to identify additional gene candidates important during Salmonella infection, we have used a large-scale N-ethyl-N-nitrosurea (ENU) mutagenesis screen. Our ENU screen has previously identified a loss of function mutation within the gene ubiquitin specific peptidase 18 (Usp18), which confers increased susceptibility to Salmonella Typhimurium (3). We have previously shown that the decreased survival in mice that carry a point mutation in Usp18 results from increased bacterial load in the spleen and liver, an increased inflammatory response and increased type I IFN signaling through STAT1 activation (3). USP18 functions both to regulate the type I IFN signaling pathway, and independently, as a protease to remove ISG15 adducts from substrate proteins (4, 5). However, the contribution of hyperactivation of type I IFN signaling and the host ISGylation pathway to the susceptibility of Usp18 mice has not been fully characterized.

Although the role of type I IFN in the host response to viral infection is well established, its role during bacterial infection is more controversial with activities that are both favorable and detrimental for the host (reviewed in (6, 7)). In addition to providing a protective role during infection, the production of type I IFN is also associated with suppression of the innate immune response through mechanisms that include decreasing the antibacterial production or function of IFN-γ (8), promoting the generation of IL-10-producing regulatory T cells (9) and decreasing the recruitment of leukocytes to the site of infection (10). Thus, an increase in type I IFN signaling can lead to increased susceptibility of the host to bacterial infection (3, 11).

Autophagy, a cellular degradation pathway, has emerged as a key component of the innate immune system through recognition and elimination of intracellular bacteria (reviewed in (12)). Upon entry into the cell, Salmonella reside within specialized vesicles called Salmonella containing vacuoles (SCV). Damage to the SCV membrane results in escape of the Salmonella to the cytosol and accumulation of polyubiquitinated proteins on the surface of the bacteria (13). Autophagy has been shown to limit the growth of Salmonella Typhimurium through association between ubiquitinated proteins and autophagic cargo receptors, including p62 (SQSTM1), NDP52 and optineurin (OPTN) (14–17). Recognition of the bacteria by these receptors facilitates interaction with the autophagy-related gene 8/microtubule-associated protein 1 light chain 3 (ATG8/LC3) family of proteins at the phagophore, resulting in lysosomal fusion and degradation of the bacteria. Autophagy is regulated by both host and pathogen-derived immune signals. Several cytokines have been reported to play an important role in either upregulating (IFNγ, TNFα, IL-1β) or inhibiting (IL-4, IL13, IL-10) autophagy (18). However, deregulation of autophagy as a consequence of increased cytokine production during Salmonella infection has not been described.

Given that USP18 suppresses type I IFN signaling and that the mutation in Usp18 mice lies within the IFNAR2-binding region of USP18 (3, 4), as well as the importance of type I IFN signaling in bacterial infection, we sought to determine if the IFNAR regulatory function of USP18 is compromised in Usp18 mice and whether hyperactivation of type I IFN signaling contributes to the pathogenesis of infection. In this study, we show that Usp18 mice succumb to Salmonella infection due to an inability to regulate type I IFN signaling, which in turn results in increased IL-10 production and deregulation of autophagy cargo receptors. Furthermore, this influences the production of IL-1β and the amount of IL-17 produced from CD4+ T cells. These data demonstrate for the first time a link between type I IFN and autophagy in Salmonella infected mice.

RESULTS

The genetic background significantly affects the expressivity of the Usp18Ity9 mutation

USP18 has two known functions including the deconjugation of the ubiquitin-like modifier protein ISG15 from target proteins and the inhibition of type I IFN-induced JAK/STAT activation (4, 5, 19). To address the relative importance of these two functions in Usp18 mice, we first transferred the Usp18 allele from the mixed genetic background of C57BL/6 × 129S1 × DBA/2J (3) to homogenous 129S1, DBA/2J or C57BL/6J inbred strains. This was particularly important to minimize the confounding effects of the mixed background on the expression of the Usp18 phenotype during infection and to have the mutation on the appropriate background for further evaluation of susceptibility to other important human pathogens. The genetic background significantly affected viability. Usp18 mutant mice were found in the expected frequencies in 129S1 and DBA/2J, whereas homozygous animals showed perinatal lethality in the C57BL/6J genetic background (data not shown). These findings are consistent with perinatal lethality observed in the C57BL/6J background for Usp18 mice (20).

The transfer of the Usp18 mutation to 129S1 or DBA/2J background resulted in decreased survival to Salmonella Typhimurium infection although the susceptibility was delayed in the 129S1 background compared to mice on a DBA/2J or a mixed genetic background (Fig. 1A, left and right panel respectively, and (3)). Consistent with our earlier findings in original mixed-strain background mice, 129S1.Cg-Usp18 and D2.Cg-Usp18 mice had significantly increased bacterial load in the spleen and liver post-infection compared with wild-type mice (Fig. 1B, left and right panel respectively ). In addition, 129S1.Cg-Usp18 and D2.Cg-Usp18 mice had elevated levels of serum IL-6 (Fig. 1C, left and right panel respectively), indicative of a systemic pro-inflammatory response. However, in contrast to Usp18 mice on a mixed background, which showed a transient decrease in IFN-γ production following Salmonella infection, serum IFN-γ levels were not significantly different in 129S1.Cg-Usp18 and D2.Cg-Usp18 mice compared to wild-type controls (Fig. 1D, left and right panel respectively). Moreover, the transient decrease in IFN-γ production observed in mixed background mutant mice was not seen at days 1 and 5 post-infection in 129S1.Cg-Usp18 mice compared to control (data not shown).

Figure 1

Genetic background affects the allelic expression of 129S1.Cg-Usp18 and D2.Cg-Usp18

Mice were infected intravenously with 1.5 X 104 CFUs Salmonella Typhimurium isolate Keller and (A) survival was monitored for 24 days (129S1.Cg-Usp18, left panel; Log-rank (Mantel-Cox) p <0.001) or 14 days (D2.Cg-Usp18, right panel; Log-rank (Mantel-Cox) p <0.007); Usp18 (n=4), Usp18 (n=7), (B) bacterial load was measured in the spleen and liver at 8 days post-infection (p.i.) (129S1.Cg-Usp18, left panel; *** p < 0.0001) or 5 days p.i. (D2.Cg-Usp18, right panel; ** p = 0.0015 and *** p = 0.0002) (Usp18, dark circles and Usp18, open circles), and (C–D) cytokines were measured at 8 days p.i. (129S1.Cg-Usp18, left panel) or 5 days p.i. (D2.Cg-Usp18, right panel) in the serum by ELISA in Usp18 (black, n=5), Usp18 (white, n=5); * p = 0.01.

Taken together, these results demonstrate that the increased susceptibility of Ity9 mutant mice is 100% penetrant when the mutation is transferred to a homogenous 129S1 or DBA/2J genetic background, albeit with variable expressivity that could be explained by the inherent degree of susceptibility of the background strains, the DBA/2J presenting an intermediate phenotype and the 129S1 being highly resistant to infection (21). The susceptibility is paralleled by an increase in bacterial load at systemic sites of infection and an increase in pro-inflammatory cytokines, collectively contributing to the clinical phenotype. The difference in the IFN-γ response in the 129S1 and DBA/2J congenic mice suggests that strain-specific modifier genes, most likely contributed from the C57BL/6J background, may be involved in the regulation of IFN-γ levels in Usp18 mice.

Loss of ISG15 had no impact on early susceptibility of Usp18Ity9 mice to Salmonella infection although it mediated delayed susceptibility in wild-type mice

To determine the contribution of ISG15 to the Usp18 phenotype, we examined ISG15 expression and function in vivo and in vitro. 129S1.Cg-Usp18mice have increased spleen Isg15 mRNA expression both prior to infection and following Salmonella challenge (Fig. 2A) and circulating ISG15 post-infection (Fig. 2B). Such increased expression is likely due to the enhanced type I IFN signaling in Usp18 mice since ISG15 is a well-known IFN-inducible gene. To determine whether the increase in Isg15 mRNA expression correlated to an increase in ISG15 conjugation to other proteins, we isolated BMDM from 129S1 wild-type and 129S1.Cg-Usp18 mutant mice and examined protein ISGylation. We found that ISG15-conjugation was significantly increased in mutant cells, under basal conditions and in cells stimulated with LPS (Fig. 2C) compared to wild-type BMDM, suggesting that the loss of functional USP18 in Usp18 mutant mice contributes to a decrease in deISGylation. To determine if the Usp18 mutation affected the enzymatic function of human USP18, we have used USP18 mutant constructed with the corresponding mutation in the human sequence (L365F) to transfect HEK293T cells. The levels of USP18 protein expression were similar in wild-type and USP18L365F transfected cells. As observed in Ity9 mutant mice, there was a reduction in deISGylation in HEK293T cells transfected with USP18L365F compared to wild-type USP18. Levels of ISGylation in USP18L365F transfected cells were comparable to those detected in the absence of USP18 (Fig. 2D).

Figure 2

The susceptibility of 129S1.Cg-Usp18 mice to Salmonella infection is not due to increase ISG15 expression and function

(A–B) Usp18 (black, n=3) and 129S1.Cg-Usp18 (white, n=3) mice were infected intravenously with 103 CFUs Salmonella Typhimurium for 8 days and (A) RNA was extracted from spleen for qRT-PCR (*p = 0.017, **p = 0.0035) and (B) ISG15 was measured in the serum by ELISA (*** p < 0.0001) (C) Western blot analysis of bone marrow derived macrophages stimulated with LPS for 18 h. (D) HEK293T cells were transfected as indicated and immunoblotting was performed. Total protein staining with Ponceau S was used to confirm equal loading. (E) Survival of mice infected with Salmonella Typhimurium for 14 days; Usp18 (n=8), Usp18(n=12), Usp18 (n=10), Usp18 (n=8), Usp18 (n=8), Usp18 (n=7); Log-rank (Mantel-Cox) p <0.0001 and (E–F) bacterial load in the spleen and liver was measured at (F) 4 days p.i. in Usp18 mice (left panel, Usp18, dark circles and Usp18) and (G) 8 days p.i. in Usp18 (right panel, Usp18, dark circles and Usp18, open circles).

To examine the impact of increased ISG15 and ISGylation in vivo, we generated Usp18/Isg15 double-deficient mice by intercrossing Usp18 mice and Isg15 knockout mice. We showed that the Isg15 genotypes (+/+, +/− or −/−) had no impact on survival to infection in Usp18 mice (Fig. 2E). Correspondingly, there was no difference in spleen and liver bacterial load 4 days post-infection between Usp18 and Usp18 mice (Fig. 2F). Interestingly, Usp18 mice lacking Isg15 showed a slight decrease in survival later during infection (Fig. 2E) and this was paralleled by an increase in bacterial load of 2.6-fold in the spleen and 4.5-fold in the liver at 8 days post-infection (Fig 2G). These results demonstrate that increased Isg15 and ISGylation are not primarily responsible for the susceptibility phenotype observed in Usp18 mice, although loss of Isg15 does mediate susceptibility and increased bacterial load later during Salmonella infection in wild-type Usp18 mice.

Blocking type I IFN receptor signaling improves survival of Usp18Ity9 mice to infection with Salmonella

Independent of its isopeptidase activity, USP18 is a negative regulator of type I IFN signaling through binding to the IFNAR2 receptor and blocking interaction between JAK and the type I IFN receptor (4). The mutation in Usp18 mice lies within the IFNAR2-binding region of USP18 suggesting that it may interfere with its regulatory function. We have previously shown that Usp18 mice on a mixed background have increased levels of Ifnb transcript and increased STAT1 phosphorylation downstream of the receptor (3). Similarly, 129S1.Cg-Usp18 mice showed increased basal Ifnb mRNA expression (Fig. 3A) and increased STAT1 activation following Salmonella infection (Fig. 3B). Increased STAT1 activation was also observed in HeLa cells transfected with human USP18L365F following stimulation with IFNα (Fig. 3C). Together, these results indicate that in addition to enzyme inactivation, this single amino acid mutation of USP18 leads to loss of the inhibitory function of USP18 in type I IFN signaling.

Figure 3

Blocking type I IFN signaling partially rescues the susceptibility of 129S1.Cg-Usp18 mice to Salmonella infection

(A–B) Usp18 (black, n=3) and 129S1.Cg-Usp18 (white, n=3) mice were infected intravenously with 1.5 X 104 CFUs Salmonella Typhimurium for 8 days and (A) RNA was extracted from spleen for qRT-PCR (*p = 0.015) (B) Western blot analysis was performed on spleen tissue lysates; n=3 mice/genotype (C) Hela cells were transduced with retrovirus for expressing the wild type and mutant USP18 as indicated, stimulated with IFNα (3000 U/mL, 15 min) and immunoblotting was performed. Total protein staining with Ponceau S was used to confirm equal loading. (D–E) Usp18 mice were given MAR1-5A3 or isotype control 1 day prior to infection with Salmonella Typhimurium and (D) survival was monitored for 22 days p.i.; n=5 mice/treatment; Log-rank (Mantel-Cox) p <0.005 and (E) bacterial load was measured in the spleen and liver in at 8 days p.i. (IgG, dark circles and MAR1-5A3, open circles; **p = 0.001 and 0.005 for spleen and liver, respectively).

To determine the impact of type I IFN signaling on the survival of 129S1.Cg-Usp18 mice, we treated mice with the IFNα/β receptor 1 (IFNAR1)-specific MAR1-5A3 monoclonal antibody one day prior to infection with S. Typhimurium. This antibody has been shown to potently inhibit type I IFN receptor signaling in mouse models of infection (22). Usp18 mice that were pretreated with the MAR1-5A3 antibody showed improved survival following Salmonella infection compared to mice that received an isotype control (Fig. 3D). The MAR1-5A3 treated mice also had significantly reduced bacterial loads in the spleen and liver (Fig. 3E). Taken together, these results suggest that loss of the regulation of the type I IFN signaling pathway that is normally imparted by USP18 contributes to the increase in susceptibility of mutant mice to Salmonella infection.

Usp18Ity9 mice have increased IL-10 and deregulation of the levels of autophagy substrates

Given that type I IFN has been shown to induce IL-10 in a STAT1-dependent manner, (23), we examined whether the increase in type I IFN signaling in Usp18 mice affected IL-10 production. At 8 days post-infection, 129S1.Cg-Usp18 mice had elevated Il10 transcript in the spleen (Fig. 4A) and an increase in circulating IL-10 (Fig. 4B). Consistent with the finding that STAT3 is activated downstream of the IL-10 receptor (24), we also observed that infected Usp18 mice had elevated levels of phosphorylated STAT3 in the spleen (Fig. 4C).

Figure 4

129S1.Cg-Usp18 have increased IL-10 and deregulated expression of autophagy markers

Mice were infected intravenously with Salmonella Typhimurium for 8 days and (A) RNA was extracted from spleen for qRT-PCR; (**p = 0.001), (B) IL-10 was measured in the serum by ELISA (*p = 0.019), and (C–F) Western blot analysis was performed on spleen tissue lysates using antibodies for STAT3 or p-STAT3 (C), p62 (D), OPTN (E) and LC3 (F). (G–H) RNA was extracted from spleen for qRT-PCR (*p = 0.05); (I) BMDM were infected with heat-killed Salmonella (MOI = 50), stained with CM-H2DCFDA and analyzed by FACS. Data is represented as mean fluorescence intensity fold changes (*p = 0.01); (J) IL-10 was measured in the serum of MAR1-5A3 or IgG control treated mice at 8 days p.i. by ELISA (***p = 0.0005); (K) Western blot analysis of OPTN was performed on spleen tissue lysates from MAR1-5A3 or IgG control treated mice at 8 days p.i. β-actin was used as a loading control.

Several studies have demonstrated a role for autophagy in innate immunity to Salmonella infection (14, 15). Since IL-10 is an inhibitor of autophagy (25), we next asked whether the increase in IL-10 in Usp18 mice affected the expression of autophagy markers in vivo. We collected lysates from spleen tissues of Usp18mice and wild-type controls and assessed the levels of the autophagy cargo receptor, p62 (SQSTM1), which has been shown to accumulate in vivo in conditions where autophagy is repressed (26). Indeed, Usp18mice showed increased accumulation of p62 following infection compared to wild-type controls (Fig. 4D), and this was not a consequence of increased p62 transcript (Fig. 4G). To further study the impact of Usp18 on autophagic markers, we measured LC3 conversion in the spleen of wild-type and mutant mice during infection. LC3 is a ubiquitin-like protein that undergoes phosphatidylethanolamine modification to facilitate association with Salmonella during infection (27). The conversion of LC3 (LC3I) to its lipidated form (LC3II) is correlated with the formation of autophagosomes. We found that the intensity of the LC3I and LC3II bands was decreased relative to β-actin in the Usp18 mice in comparison to wild-type mice before and during Salmonella infection, although to a greater extent following bacterial infection (Fig. 4E). During Salmonella infection, an additional autophagy receptor, optineurin (OPTN), is recruited to ubiquitylated bacteria in the cytosol and, following phosphorylation by TBK1, binds LC3 to bring Salmonella to the autophagosome (16). Therefore, we assessed the levels of OPTN in Usp18 mice to further define the impact of the mutation on autophagy during Salmonella infection. We found that Usp18 mice had decreased levels of Optn transcript and OPTN protein after infection (Fig. 4F and H).

Generation of ROS by NADPH oxidase is important for the induction of autophagy, LC3 recruitment to phagosomes and restriction of intracellular replication of Salmonella Typhimurium (28). Consistent with our finding that Usp18 mice have deregulated autophagy marker expression following Salmonella infection in vivo, we also observed that Usp18 macrophages have decreased ROS production after exposure to heat-killed Salmonella (Fig. 4I). Together, these data suggest that the diminished levels of ROS and OPTN result in a failure of autophagy to proceed, thus leading to an accumulation of p62 and lack of LC3 conversion.

To further demonstrate that type I IFN is important in the control of IL-10 and autophagy during Salmonella infection in Usp18 mutant mice, we investigated whether inhibition of Type I IFN in Usp18 mice would affect IL-10 levels in circulation. We found that the increase in IL-10 observed in Salmonella-infected mice given the IgG control was diminished in mice treated with the type I IFN neutralizing antibody (Fig. 4J). Moreover, inhibition of type I IFN in mutant mice was also sufficient to restore levels of OPTN following Salmonella infection (Fig. 4K). Together, these data support our hypothesis that IL-10 and autophagic marker levels are affected by the levels of type I IFN in Usp18 mice.

Usp18Ity9 mice have increased IL-1β and an elevated Th17 response

IL-1β is a pro-inflammatory cytokine important for innate immunity to Salmonella infection, but in excess, can result in endotoxemia (29). Autophagy regulates pro-IL-1β production and an accumulation of cellular p62 due to deficient autophagy can promote activation of NF-κB and subsequently induce pro-IL-1β (30). Therefore, we evaluated the impact of the Usp18 mutation on IL-1β expression. We found that Il1b transcript and pro-IL-1β protein levels were increased in Usp18 at 8 days post-infection (Fig. 5A–B). In addition, explanted splenocytes from Salmonella Typhimurium-infected mice produced more IL-1β compared to control mice (Fig. 5C). Since IL-1β release can lead to an increase in IL-23 secretion, (31) and together, potently induce the secretion of IL-17 by Th17 cells (32), we next investigated the levels of these cytokines in Usp18 mice after Salmonella infection. We found that mutant mice had elevated levels of IL-23 and IL-17 transcript (Fig. 5D–E) and increased production of IL-17 from CD4+ T cells (Fig. 5F) showing that high IL-1β levels in Usp18 mutant mice induced a Th17 response.

Figure 5

129S1.Cg-Usp18 have increased IL-1β and an elevated Th17 response

Mice were infected intravenously with 1.5 X 104 CFUs Salmonella Typhimurium for 8 days and (A) RNA was extracted from spleen for qRT-PCR (** p = 0.009), (B) Western blot analysis was performed on spleen tissue lysate using an antibody for IL-1β, (C) Splenocytes were harvested and cultured for 24 h prior to measuring IL-1β in the supernatant by ELISA, (D–E) RNA was extracted from spleen for qRT-PCR of (D) IL-23 (*p = 0.02, **p = 0.006) and (E) IL-17 (**p = 0.001), (F) Flow cytometry of intracellular IL-17 in CD4+ T cells from the spleen (*p = 0.016). n=3–5 mice/genotype from 2 experiments.

Usp18Ity9 mice are susceptible to Mycobacterium tuberculosis infection

To further investigate the role of USP18 during bacterial infection, we tested whether the Ity9 mutation would impact the susceptibility to another Gram-negative bacteria (Citrobacter rodentii) and mycobacteria (Mycobacterium bovis BCG and Mycobacterium Tuberculosis). Usp18 were not susceptible to C. rodentium as measured by survival analysis (data not shown) and bacterial shedding (data not shown) and to M. bovis BCG (data not shown). In contrast, we found that Usp18 mice showed significantly increased susceptibility to infection compared to both wild-type littermates and DBA/2J mice following aerosol infection with the highly virulent M. tuberculosis H37Rv strain (Fig. 6A). Decreased survival correlated with significantly higher bacterial burden in both the lung and spleen of M. tuberculosis infected Usp18 mice (Fig. 6B). At necropsy 6 weeks post-infection, Usp18 mutant mice showed large infected foci with extensive necrotic centre and increased lymphohistiocytic inflammatory cell infiltration in the lung compared to littermate control mice with a greater percentage of the lung affected by inflammation (Supplemental Fig. 1, left and middle panels). Moreover, Zeihl-Neelsen staining of acid-fast bacilli revealed significantly more Mycobacterium tuberculosis bacteria in the lungs of mutant mice compared to wild-type controls (Supplemental Fig. 1, right panels). As observed during the course of Salmonella infection, Usp18mice showed elevated levels of circulating ISG15 and several cytokines including IFNγ, TNFα, and IL-10 (Fig. 6D). In addition, Usp18mice also showed increased lung mRNA expression of Il17, Il1b and Isg15, with no difference in the mRNA expression of Il10 between wild-type and mutant mice (Fig. 6E). Together, these observations suggest that Usp18 mutant mice are more susceptible to mycobacterial infection as a result of increased bacterial load and excessive inflammatory response.

Figure 6

Usp18are more susceptible to M. tuberculosis infection

Mice were infected by aerosol with 100 CFUs of H37Rv and (A) survival was monitored for 150 days p.i. (p < 0.001), n=11 mice/genotype from 2 experiments (B) bacterial load was measured in the lung and spleen (Usp18, dark circles and Usp18, open circles) at 42 days p.i. (***p = 0.005 and 0.001 for lung and spleen, respectively), (C) cytokines were measured in the serum by ELISA at day 42 p.i. n=6–8 mice/genotype and (D) RNA was extracted from lung for qRT-PCR n=3 mice/genotype (*p = 0.05, **p = 0.002).

DISCUSSION

In the current paper we examined the contribution of the deISGylating and Type I IFN regulatory functions of Usp18 during Salmonella infection. Our in vitro and in vivo data are consistent with those from Usp18 knockout mice where LPS-stimulated macrophages have increased ISGylation compared to wild-type mice (33). The robust expression of ISG15 and ISGylation in response to Salmonella infection due to the loss of USP18 activity does not appear to play a major role in susceptibility to infection of Usp18 mutant mice. Loss of Isg15 in Usp18 mutant mice does not impact on the in vivo susceptibility, suggesting that deregulation of the ISGylation pathway is not the predominant mechanism of susceptibility in Usp18 mice. The secretion of ISG15 from various immune cells suggests that in addition to its role as a ubiquitin-like molecule, ISG15 acts as a cytokine that synergizes with IL-12 to increase IFNγ and provide protective immunity to infection (34–36). Although the expression and secretion of IFN-γ in Usp18 mutant mice was not affected by the high levels of ISG15, it is possible that the loss of ISG15 in its secreted form in the double Isg15 and Usp18 deficient mice may explain, in part, the inability to improve survival to infection. Indeed, Isg15deficient mice carrying a wild-type allele at Usp18 showed increased susceptibility later during infection. Late susceptibility of Isg15 deficient mice to Mycobacterium tuberculosis infection has been also reported (35). Identification of ISG15 substrate proteins, including those involved in adaptive immune responses, will provide insight on the mechanism of susceptibility later during bacterial infection.

On the other hand, our results indicate that Usp18 mice treated with a neutralizing antibody to the type I IFN receptor have improved survival and decreased bacterial load after Salmonella challenge. We have previously reported that signaling through the type I IFN pathway is deleterious to the host during Salmonella infection (3, 11). In addition, IFNAR−/− mice are more resistant to infection with Salmonella Typhimurium as a result of increased macrophage necroptosis thereby permitting evasion of the host response (37). Type I IFNs have been shown to be detrimental to the host during bacterial infection through a number of mechanisms including chemokine production, leukocyte recruitment, T cell responses and host cell apoptosis, among others (reviewed in (6), suggesting that the mechanisms underlying the action of type I IFN are complex.

In Usp18 mutant mice, the proinflammatory immune response to infection becomes amplified and dysregulated as shown by excessive production of the cytokines IL-1β and IL-6 These cytokines most likely act in synergy with other cytokines (IFNγ and TNF) that were also upregulated during infection, to cause septic shock, tissue damage, and death. In parallel, we showed that Usp18 mice have elevated systemic IL-10. IL-10 is an anti-inflammatory cytokine that prevents damage to the host. IL-10 works in opposition to IL-6, which also signals through STAT3 (38). The activation of IL-10 signaling does not appear to repress the expression of proinflammatory genes in Usp18 mutant mice although decreased ROS production in vitro was observed, which is consistent with studies showing that IL-10 inhibits ROS production in LPS-stimulated macrophages and neutrophils (39, 40). The attenuated oxidative burst activity in Usp18 mutant mice may well explain the higher bacterial load observed in the spleens and livers of these animals.

In our model of infection, IL-1β is increased at both the transcript and protein levels in the spleen of infected mice, which contrasts with the observation that IFN-β is able to limit pro-IL-1β availability and IL-1β maturation (23) and that elevated type I IFN inhibits M. tuberculosis-induced IL-1β mRNA expression in macrophages (41). Moreover, type I IFN was shown to inhibit production of IL-1β from myeloid cells in vivo (42), resulting in a loss of IL-1β-mediated control of bacterial burden (43). This discrepancy may be attributed to an increase in p62 post-infection in Usp18 mice since accumulation of p62 and subsequent activation of NF-κB has been shown to increase IL-1β (30). We do observe an increase in IL-23 and IL-17 in the spleen Usp18 mutant mice which is consistent with reports of an IL-1β-dependent increase in IL-23 resulting in enhanced IL-17 secretion from T cells (31, 32). USP18 was recently shown to play an important role in adaptive immunity by controlling Th17 cell differentiation (44). USP18 knock out T cells were shown to be deficient in Th17 differentiation, which contrasts with our observation that Usp18 mutant mice had elevated levels of Il17 transcript and increased production of IL-17 from CD4+ T cells during Salmonella infection. This apparent discrepancy could be explained by the fact that the Usp18 mutation does not affect the peptidase domain of USP18 and Liu et al. reported that the enzymatic activity of USP18 was important for controlling Th17 differentiation. The exact mechanism underlying increased IL-1β and IL-17 in Usp18 mutant mice is not fully understood, however, we suggest that inhibition of autophagy by type I IFN signaling may be one of the mechanisms. This hypothesis is based on the reported observations that inhibition of autophagy promotes inflammasome activity and increases IL-1β production (45) and that IL-17 is increased in M. tuberculosis infected mice deficient in autophagy (46) and a prolonged Th17 response contributes to the pathogenesis of the infection (47). Autophagic degradation of intracellular bacteria is an important host defense mechanism during infection. Recently, invading Salmonella were shown to induce a transient amino acid starvation as a result of membrane damage leading to the induction of autophagy and host protection (48). Yet, Salmonella effector molecules including SseL, a deubiquitinase, and SopD2, a regulator of SCV integrity, contribute to the inhibition of autophagy during infection (49, 50), suggesting that bacterial evasion of the host autophagic response is key to bacterial replication. Low levels of autophagy lead to accumulation of p62 and decreased LC3II (51). Thus, the increased p62 and decreased LC3II observed in Usp18 mice in vivo raises the possibility that autophagy may be impaired in Usp18 mutant mice although a more comprehensive evaluation of autophagy in vitro will be necessary to draw a final conclusion.

The role of type I IFN in the regulation of mycobacteria infections is highlighted by recent studies showing that IFNAR−/− mice are more resistant to M. tuberculosis infection and that patients with active tuberculosis have a characteristic IFN-inducible gene signature (52, 53). An increase in the type I IFN response was also reported in lepromatous leprosy skin lesions when compared to self-healing tuberculoid lesions caused by Mycobacterium leprae in vivo (54). We show in the current paper that Usp18 mice are more susceptible to Mycobacterium tuberculosis infection and have increased bacterial load in lung and spleen, elevated inflammatory cytokine production and more severe lung pathology. As shown during Salmonella infection and in contrast with previous studies (35), the increases in circulating ISG15 and IFNγ do not appear to be protective during infection in Usp18 mice. This finding demonstrates that USP18 plays a broad role during infection with intracellular bacteria, additional studies will provide further insight into the mechanism of susceptibility in Usp18mice during mycobacterial disease.

In summary, this work describes the biological significance of type I IFN signaling in the survival of Usp18 mice following Salmonella infection and outlines a model of susceptibility resulting from increase in proinflammatory cytokine production as a consequence of hyperactivation of type I IFN signaling, and increased IL-10 production, leading to deregulated expression of IL-1β and autophagy markers that results in increased bacterial burden and septic shock in Salmonella-infected Usp18 mice. Our studies reveal that a mutation in human USP18 corresponding to the Usp18 mutation also affects the enzymatic and regulatory functions of USP18, suggesting that the findings presented here may be relevant to the function of human USP18 during infection.

MATERIALS AND METHODS

Ethics Statement

All animal experiments were performed under guidelines specified by the Canadian Council on Animal Care. The animal use protocol was approved by the McGill University Animal Care Committee (protocol #5797).

Mice

The Usp18 mutation was originally identified on a mixed C57BL/6 × 129S1/SvImJ × DBA/2J genetic background (37.5%, 37.5%, 25%, respectively). Backcrossing from the original mixed background was accomplished by nine generations of inbreeding to the 129S1 or DBA/2J strains (Jackson Laboratories), resulting in mice that are >99.8% 129S1 or DBA/2J. Isg15 knockout (B6.129P2-Isg15/J; Jackson Laboratories) were crossed with Usp18 or Usp18 mice to generate F1 mice that were heterozygous for Isg15 and Usp18 or the wild-type allele. These mice were intercrossed to generate mice that were homozygous for Usp18 or the wild-type allele and selected for wild-type Slc11a1. Mice were bred at the Goodman Cancer Research Centre Animal Facility.

In vivo Salmonella infections

Mice between 7–12 weeks of age were infected intravenously with Salmonella Typhimurium strain Keller, as described by us previously (3). Mice were infected in the caudal vein and monitored twice daily for survival. Alternatively, spleens and livers were collected, homogenized in saline and CFUs were determined by plating of serial dilutions on trypticase soy agar plates.

ELISA

Serum was obtained from the blood of infected mice and cytokines were assayed by ELISA (eBioscience) according to the manufacturers directions.

RNA extraction and quantitative RT-PCR

Total RNA was isolated from mouse tissue using the TRIzol reagent (Invitrogen Life Technologies). First-strand cDNA was generated using MMLV-RT (Invitrogen) and random oligonucleotides as primers. Quantitative PCR was performed in duplicate for each transcript using SYBR® green qPCR master mix (Applied Biosystems) on a StepOnePlus apparatus (Applied Biosystems). The Ct values for the genes of interest were normalized to the housekeeping gene TATA-binding protein (TBP). The relative expression of the gene was calculated as 2−ΔΔCt.

Bone marrow-derived macrophages

Femurs were collected from 8–12 week old mice and bone marrow was extracted by flushing the femurs with RPMI using a 25-G needle. A single-cell suspension was obtained by passage through a 25-G needle and RBCs lysed for 5 min using a commercial RBC Lysis Buffer (Sigma-Aldrich). Cells were resuspended in complete medium (RPMI supplemented with 10% heat-inactivated FBS, 2 mM glutamine, 100 μg/mL Penstrep and 30% L929 conditioned medium as a source of murine CSF. L929 supernatant was replenished every 2 days prior to cell counting and plating.

Immunoblotting

Protein lysates were prepared using the CellLytic-M reagent (Sigma-Aldrich) according to the manufacturers directions. Proteins were quantified by the Bradford method (Bio-Rad) and Western blotting was carried out using 25–50 μg of protein resolved by SDS-PAGE, transferred to PVDF membrane and immunoblotted as indicated. β-actin was used to assess protein loading.

Plasmids and transfection

The wild type USP18 expression constructs were described previously (55). USP18L365F mutant constructs were generated by site directed mutagenesis. ISG15 isopeptidase activity and type I IFN inhibitor activity of wild type and mutant USP18 were examined using an ex vivo ISGylation system in 293T cells and in HeLa cells, respectively, as described previously (55, 56).

MAR1-5A3 treatment

Mice were pretreated with 2 mg of either MAR1-5A3 or mouse IgG1 isotype control antibody (Leinco Technologies, Inc.) 1 day prior to infection with 1000 CFUs of Salmonella Typhimurium.

Flow cytometry

Spleens were harvested from 8–12 week old mice, macerated and passaged through a 70 μM cell strainer. RBC lysis was performed using ACK lysing buffer and cells were enumerated and plated for ELISA or for flow cytometry. Intracellular staining of IL-17 (eBioscience) was performed on prepared splenocytes (10 × 106 cells) using CytoFix/CytoPerm (BD Biosciences). Briefly, cells were stimulated in vitro (4 h) with PMA (50 ng/mL; Sigma-Aldrich) and Ionomycin (500 ng/mL; Sigma-Aldrich) and intracellular transport was inhibited using GolgiStop (BD Biosciences). Cells were stained with antibodies to CD4, CD3, B220 and IL-17 and gating was performed to include CD3+CD4+B220−IL-17+ cells. All samples were analysed using a FACSCanto (BD Biosciences) with FlowJo software (Tree Star).

ROS measurements

BMDM were plated in non-tissue-culture treated six-well dishes and infected with heat-killed Salmonella (MOI = 50, 6 h). Complete DMEM was aspirated, cells were washed with PBS and incubated with CM-H2DCFDA (10 μM, 30 min; Invitrogen) in serum-free medium. Cells were washed with warm PBS, then removed from the well using cold PBS containing 10 mM EDTA, pelleted at 1200 r.p.m., resuspended in cold PBS containing 1% FBS and analysed using a FACSCanto (BD Biosciences) with FlowJo software (Tree Star). Mean fluorescence intensity values were calculated as fold change over uninfected cells.

In vivo Mycobacterium tuberculosis infections

Mycobacterium tuberculosis H37Rv was grown at 37°C in Middlebrook 7H9 medium (Difco Laboratories) containing 0.05% Tween-20 (Sigma-Aldrich) and 10% albumin-dextrase-catalase (ADC) supplement (Becton Dickson and Co.). Bacteria were delivered by aerosol using an inhalation exposure system (In-Tox Products) and infectious dose was confirmed by enumeration of bacteria within the lungs of control mice at 24 hours post-infection. Mice were euthanized at 6 weeks post-infection, organs were homogenized in PBS and bacterial burden was determined by serial dilution on Middlebrook 7H10 agar (Difco Laboratories) plates supplemented with OADC enrichment (Becton Dickson and Co.) and BacTac Panta Plus (Becton Dickson and Co.). Serum was collected for ELISA and tissues were either fixed in buffered formalin prior to immunohistochemical analysis or stored in RNA later (Ambion). For histology, representative slides were assessed by a pathologist and scored for degree of inflammation.

Statistical Analyses

Results are expressed as means ± s.e.m. Data were analysed using a two-tailed Student’s t-test using the GraphPad Prism statistical program. P values of less than 0.05 were considered significant.