Myeloid cell deficiency of p38γ/p38δ protects against candidiasis and regulates antifungal immunity

p38γ/p38δ regulate Candida albicans‐induced cytokine production

To investigate the role of p38γ/p38δ in C. albicans infection, we assessed inflammatory cytokine and chemokine mRNA levels in response to heat‐killed C. albicans (HK‐Ca) in WT and p38γ/δ^−/− bone marrow‐derived macrophages (BMDM) in vitro. Macrophages are one of the main cells that come in contact with the fungus early after infection in candidiasis (Lionakis et al, 2013). HK‐Ca stimulation in p38γ/δ^−/− BMDM had no effect in the mRNA expression of TNFα and IL‐6, whereas IL‐1β, IL‐10, KC, MIP‐2 and CCL2 mRNA expression was markedly reduced as compared to that in WT BMDM (Fig 1A), indicating the need of these p38 kinases for cytokine and chemokine production in the response to C. albicans.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Cytokine production and ERK1/2 activation is impaired in p38γ/p38δ‐null BMDM in response to TLRs and Dectin‐1 ligands

1. BMDM from WT or p38γ/δ^−/− mice were exposed to HK‐Ca (1 × 10⁶ CFU/ml) for the indicated times. Relative mRNA expression was determined by qPCR for TNFα, IL‐6, IL‐1β, IL‐10, KC, MIP‐2 and CCL2. Results were normalized to β‐actin mRNA expression, and x‐fold induction was calculated relative to WT expression at 0 h. Data show mean ± SEM from one representative experiment of two in triplicate, with similar results. Only significant results are indicated, *P ≤ 0.05 relative to WT BMDM exposed to HK‐Ca, at each time point. Parametric, unpaired t‐test.

2. BMDM from WT or p38γ/δ^−/− mice were stimulated as in (A). Cell lysates (50 μg) were immunoblotted with antibodies to active phosphorylated ERK1/2 (P‐ERK1/2), p38α (P‐p38α) or JNK1/2 (P‐JNK1/2), or to phosphorylated p105 NF‐κB1 (P‐p105). Total protein levels for the above proteins and for TPL2 were also measured as loading controls. Representative immunoblots from three independent experiments are shown.

3. BMDM were stimulated 200 ng/ml Pam3Cys or 100 ng/ml LPS. Cell lysates were immunoblotted with the indicated antibodies. Representative immunoblots from three independent experiments are shown.

4. BMDM were exposed for 1 h to 10 μg/ml Curdlan. Relative mRNA expression was determined by qPCR for IL‐1β. Results were normalized and fold induction calculated as in (A). Data show mean ± SEM from one representative experiment of two in triplicate, with similar results. Only significant results are indicated, *P ≤ 0.05 relative to WT BMDM exposed to Curdlan. Parametric, unpaired t‐test.

5. BMDM were stimulated with 10 μg/ml Curdlan and cell lysated immunoblotted as shown. Representative immunoblots from three independent experiments are shown.

Source data are available online for this figure.

Source Data for Figure 1 [emmm201708485-sup-0002-SDataFig1.pdf]

To assess whether p38γ/p38δ loss altered the signalling response involved in cytokine production, we analysed changes in NF‐κB and MAPK (p38α, ERK1/2, JNK) pathways activation in response to C. albicans in p38γ/δ^−/− and WT BMDM. We found that p38γ/p38δ deletion did not have a significant effect on the activation of p38α and the NF‐κB pathways, although the activation of these pathways was more sustained in WT than in p38γ/δ^−/− BMDM (Fig 1B). In contrast, ERK1/2 phosphorylation was substantially reduced in p38γ/δ‐null BMDM as compared to WT BMDM (Fig 1B). We did not detect significant HK‐Ca‐induced phosphorylation of JNK1/2 either in WT or p38γ/δ^−/− BMDM (Fig 1B).

To determine the mechanism by which ERK1/2 pathway activation is impaired in p38γ/δ^−/− BMDM after C. albicans infection, we examined ERK1/2 phosphorylation in response to specific ligands for each C. albicans‐activated receptor in macrophages. We have previously reported that p38γ/p38δ positively regulate TLR4‐induced ERK1/2 activation and cytokine production by controlling the steady‐state levels of the MKK kinase TPL2 in macrophages (Risco et al, 2012), which is not expressed in p38γ/δ^−/− cells (Fig 1B). p38γ/δ deletion impaired ERK1/2 pathway activation in response to both the TLR2/6 ligand tripalmitoylated lipopeptide (Pam3Cys) and the TLR4 ligand LPS (Fig 1C). Accordingly, the activation of ERK1/2 pathway in response to unmethylated CpG oligonucleotide (ODN, TLR9‐ligand), Imiquimod (TLR7‐ligand) and poly I‐C (TLR3‐ligand), which is mediated by TPL2, was also impaired in p38γ/δ^−/− macrophages (Appendix Fig S1A and B). In contrast, ERK1/2 activation in response to PMA, which is dependent on Raf‐1, another MKK1/2 kinase (Wellbrock et al, 2004), and TPL2‐independent (Beinke & Ley, 2004) was unaffected (Appendix Fig S1B). p105 NF‐κB1, JNK1/2 and p38α were still phosphorylated in response to TLR ligands in p38γ/δ‐null macrophages (Appendix Fig S1C). All these data demonstrate that p38γ/p38δ regulate ERK1/2 activation triggered by TLRs. Nonetheless, we also examined the contribution of Dectin‐1 to cytokine production and ERK1/2 pathway activation in BMDM by stimulating with Curdlan, a purified β‐glucan that mimics fungal stimulation in innate immune cells (Gantner et al, 2003). The stimulation of p38γ/δ‐null macrophages with Curdlan resulted in decreased IL‐1β production (Fig 1D) and decreased MKK1‐ERK1/2 activation as compared to WT BMDM (Fig 1E). p38α, JNK1/2 and p105 NF‐κB1 were phosphorylated in response to Curdlan in WT and p38γ/δ^−/− macrophages (Fig 1E).

We confirmed the specificity of Curdlan as a Dectin‐1 ligand by examining ERK1/2 activation in MyD88^−/−, Dectin1^−/− BMDM and also using the Syk inhibitor PRT062607 (Spurgeon et al, 2013) in WT BMDM. TLRs signal via the adaptor molecule MyD88, whereas Dectin‐1 signalling is mediated by the recruitment of the tyrosine kinase Syk (Underhill, 2007; Reid et al, 2009; Takeuchi & Akira, 2010). Curdlan, HK‐Ca and LPS (used as control) induced ERK1/2 phosphorylation in WT macrophages. The activation of ERK1/2 in MyD88^−/− BMDM was blocked in response to LPS, whereas in response to Curdlan was not affected and in response to HK‐Ca was partially impaired, compared to WT BMDM (Appendix Fig S1D). Furthermore, in Dectin1^−/− BMDM, ERK1/2 activation by Curdlan was blocked, by HK‐Ca was partially inhibited and by LPS was not affected (Appendix Fig S1E). In WT macrophages pre‐treated with the compound PRT062607, the phosphorylation of ERK1/2 was significantly inhibited in response to Curdlan and was not affected in response to LPS (Appendix Fig S1F). These results confirm that, in macrophages, the activation of ERK1/2 induced by Curdlan was mediated by Dectin‐1‐Syk signalling and not by TLRs, whereas ERK1/2 activation in response to HK‐Ca was also partially induced via TLR‐MyD88 signalling.

TAK1‐IKKβ‐TPL2 activation is essential for Dectin‐1 signalling in macrophages

Since TPL2 was not expressed in p38γ/δ^−/− cells (Fig 1B and E), our results suggest that the kinase TPL2 regulates ERK1/2 activation triggered by Dectin‐1. To investigate this in more depth, we used TPL2^−/− BMDM. In these cells, ERK1/2 activation was abolished in response to LPS, Curdlan, HK‐Ca and Zymosan, which activates Dectin‐1 and TLR2 in macrophages (Brown et al, 2003; Fig 2A), confirming that TPL2 mediates ERK1/2 activation not only downstream of TLR but also downstream of the receptor Dectin‐1. Consistently, we found that pharmacological blockade of TPL2 by the compound C34 (Green et al, 2007; Wu et al, 2009; Appendix Fig S2A) and of MKK1 by the inhibitor PD184352 (Bain et al, 2007) inhibited Curdlan and HK‐Ca‐induced ERK1/2 activation in WT BMDM (Fig 2B and Appendix Fig S2B). C34 also reduced IL‐1β mRNA production in Curdlan‐ and HK‐Ca‐activated WT BMDM, but not in p38γ/δ^−/− BMDM (Fig 2C and Appendix Fig S2C), supporting the important role of p38γ/p38δ‐TPL2 in Dectin‐1 signalling. This is the first time that a role for TPL2 in ERK1/2 activation triggered by Dectin‐1 signalling is described.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. ERK1/2 activation is mediated by TPL2 in Dectin‐1 signalling

1. BMDM from TPL2^+/+ or TPL2^−/− mice were stimulated with 1 × 10⁶ CFU/ml HK‐Ca, 10 μg/ml Curdlan or 50 μg/ml Zymosan for 1 h, or with 100 ng/ml LPS for 30 min. Cell lysates were immunoblotted with the indicated antibodies. Representative immunoblots are shown. Bands from three experiments were quantified using the Odyssey infrared imaging system (bottom), and data show mean ± SEM from two experiments in duplicate. ***P ≤ 0.001. Parametric, unpaired t‐test.

2. WT BMDM were incubated for 1 h with or without 5 μM C34 or 2 μM PD184352, and then stimulated for 1 h with 10 μg/ml Curdlan. Representative immunoblots from two independent experiments are shown.

3. BMDM were incubated for 1 h with DMSO or with 5 μM C34, and then exposed for 1 h to 10 μg/ml Curdlan. Relative mRNA expression was determined by qPCR for IL‐1β. Results were normalized and fold induction calculated. Data show mean ± SEM from one representative experiment of two in triplicate, with similar results. ns, not significant, *P ≤ 0.05 relative to WT BMDM exposed to Curdlan. Parametric, unpaired t‐test.

4. WT BMDM were incubated for 1 h with or without 5 μM C34 or 10 μM BI605906 and then stimulated with Curdlan as in (B). Representative immunoblots from four independent experiments are shown.

5. WT BMDM were incubated for 1 h with DMSO or with 2 or 5 μM NG25 and then stimulated with HK‐Ca, Curdlan or LPS as in (A). Representative immunoblots from two independent experiments are shown.

6. WT BMDM were incubated for 1 h in the absence or the presence of 5 μM C34 or 10 μM BI605906, and then stimulated for 1 h with 1 × 10⁶ CFU/ml HK‐Ca. Cell lysates were immunoblotted as indicated. Representative blots from two independent experiments are shown.

7. Schematic representation of the Dectin‐1 signalling pathways involved in ERK1/2 activation, which is controlled by TAK1‐IKK‐TPL2 in HK‐Ca and Curdlan‐stimulated macrophages. The activation of the TAK1 complex (TAB 1‐TAK1‐TAB 2/3) and of the IKK pathway (Cohen, 2014) might be mediated by TRAF6, which binds to the CARD9/BCL‐10/MALT1 complex downstream of Syk (Geijtenbeek & Gringhuis, 2009). TLR stimulation by HK‐Ca also triggers the activation of TAK1‐IKK‐TPL2 via MyD88. p38γ and p38δ regulate TPL2 steady‐state levels, which is in a complex with ABIN‐2 and p105 (Gantke et al, 2011).

Source data are available online for this figure.

Source Data for Figure 2 [emmm201708485-sup-0003-SDataFig2.pdf]

TPL2 activation is regulated by IKKβ in TLR‐stimulated macrophages (Gantke et al, 2011; Roget et al, 2012; Ben‐Addi et al, 2014). Moreover, the kinase TAK1 is required for the activation of IKKβ in the canonical IKK complex leading to TPL2‐MKK1‐ERK1/2 activation (Cohen, 2014). Therefore, to investigate whether the TAK1‐IKK signalling pathway was involved in the activation of ERK1/2 induced by Dectin‐1, we treated macrophages with the highly selective ΙΚΚβ inhibitor BI605906 (Clark et al, 2011) or with the potent TAK1 inhibitor, NG25 (Dzamko et al, 2012). BMDM pre‐treatment with BI605906 or NG25 blocked Curdlan from inducing ERK1/2 activation (Fig 2D and E). Accordingly, BI605906 and NG25 also impaired ERK1/2 phosphorylation in response to HK‐Ca (Fig 2E and F). NG25 also inhibited Curdlan‐ and HK‐Ca‐induced phosphorylation of p105 NF‐κB1 and in response to LPS the ERK1/2 and p105 NF‐κB1 phosphorylation (Fig 2E). All our data together indicate that p38γ/p38δ positively modulate TPL2 levels and that TAK1‐IKKβ‐TPL2 regulate MKK1‐ERK1/2 activation in both Dectin‐1 and TLR pathways in macrophages (Fig 2G).

Additionally, we studied the implication of TPL2 in C. albicans response in human peripheral blood mononuclear cells (PBMCs)‐derived monocytes. In these cells, Dectin‐1 is an important receptor for the immune sensing of C. albicans, since incubation with laminarin, a specific Dectin‐1 inhibitor, largely decreases cytokine production in response to the fungus (Toth et al, 2013). Monocytes were pre‐treated with the TPL2 inhibitor C34 or the p38α/β inhibitor SB203580, as control, and then stimulated with HK‐Ca. As expected, HK‐Ca‐induced ERK1/2 phosphorylation was reduced to the basal levels by pre‐incubation with C34 (Appendix Fig S2D). HK‐Ca‐induced expression of TNFα, IL‐6 and IL‐10 was impaired by C34, whereas SB203580 only partially blocked IL‐10 production (Appendix Fig S2E), supporting the conclusion that in vitro, C. albicans‐induced cytokine production is dependent on TPL2 activity also in human monocytes.

p38γ/p38δ deletion protects from Candida albicans infection

Since inflammation is central to candidiasis, we aimed to define the role of p38γ/p38δ in the host defence against disseminated candidiasis as mirrored by an intravenous challenge model of C. albicans infection. This is an established model of systemic candidiasis, in which the kidneys are the primary target organs; mice develop renal failure and septic shock, and this recapitulates the progressive sepsis seen in humans during severe clinical cases (Spellberg et al, 2005). We first compared the survival of p38γ/p38δ‐deficient (p38γ/δ^−/−) mice and control WT mice to fungal infection. We observed that the lack of p38γ/p38δ caused a remarkable protection to infection (Fig 3A). To determine whether the effect in survival was due to p38γ/p38δ in myeloid cells, we also analysed mice with myeloid cell‐specific p38γ/p38δ deletion (LysM‐p38γ/δ^−/−; Zur et al, 2015). In agreement with the observation in p38γ/δ^−/− mice, the survival of LysM‐p38γ/δ^−/− mice was significantly increased compared with control mice after C. albicans infection (Fig 3A). The effect of p38γ/p38δ deletion on mouse survival was probably due to the decreased load of C. albicans in the organs; accordingly, the fungal burden in p38γ/δ^−/− and LysM‐p38γ/δ^−/− kidneys was significantly lower than in WT control mice (Fig 3B and Appendix Fig S3). When we analysed C. albicans dissemination after intravenous injection, we found that the fungus was rapidly cleared from the bloodstream in all genotypes (Appendix Fig S3). Although the clearance in WT mice bloodstream was markedly slower than in p38γ/δ^−/− and LysM‐p38γ/δ^−/− mice, the fungal burden in spleen, liver and brain was similar in all mice (Appendix Fig S3). In kidney, fungal burden was similar in all mice at day 1, whereas in p38γ/δ^−/− and LysM‐p38γ/δ^−/− mice, the fungal burden was slightly lower at day 2 and significantly lower at day 3 of C. albicans injection than in WT mice (Appendix Fig S3). Histological analysis of kidney sections revealed that at day 3 post‐infection WT mice showed evident growth of C. albicans forming hyphae, whereas kidneys from p38γ/δ^−/− and LysM‐p38γ/δ^−/− mice displayed nearly undetectable hyphae formation (Fig 3C). These results indicate that p38γ/p38δ, particularly in myeloid cells, increase resistance to C. albicans infection.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. p38γ/p38δ deletion decreases Candida albicans infection in mice

• A, B WT, p38γ/δ^−/− and LysM‐p38γ/δ^−/− mice were infected with 1 × 10⁵ CFU C. albicans. (A) Survival monitored as indicated. Data are presented as a Kaplan–Meier plot from two independent experiments (n = 20 mice per genotype). Two‐way ANOVA using GraphPad Prism software. (B) Kidney fungal burden at day 3 post‐infection. Data are expressed as CFU/g kidney (mean ± SEM). Each symbol represents an individual mouse. ns, not significant, *P ≤ 0.05, **P ≤ 0.01 and ***P ≤ 0.001 relative to WT mice. Parametric, unpaired t‐test.

• C. Representative PAS‐haematoxylin staining of kidney sections from mice at day 3 post‐infection. Bottom panels are high magnification of the area marked by a dotted square in the top panels. Scale bars are 100 μm.

• D, E TPL2^+/+ and TPL2^−/− mice were infected with C. albicans as in (A). (D) Death was monitored. Data are a summary of two independent experiments (n = 12 mice per genotype). (E) Kidney fungal burden at day 3 post‐infection with 1 × 10⁵ CFU. Each symbol represents an individual mouse. Data are expressed as CFU/g kidney (mean ± SEM). *P ≤ 0.05 relative to WT mice. Two‐way ANOVA using GraphPad Prism software.

Since p38γ/p38δ regulate TPL2 protein levels in macrophages, we then assessed the role of TPL2 in the response to C. albicans in vivo. Loss of TPL2 did not protect against fungal infection, as TPL2^+/+ and TPL2^−/− mice survival was similar (Fig 3D). We also found that TPL2 deletion led to a small increase in fungal burden in the kidney, when compared with TPL2^+/+ kidney (Fig 3E). Histological analysis of kidney sections confirmed these results; both WT and TPL2^−/− mice showed extensive growth of C. albicans forming hyphae (Fig 3C). These data show that TPL2 deletion does not protect against C. albicans infection and suggest that in vivo p38γ/p38δ diminish fungal infection independently of TPL2.

p38γ/p38δ deletion decreased the inflammatory response against Candida albicans infection

To address the possibility that increased survival of p38γ/δ‐deficient mice to C. albicans infection could be due to effects on the inflammatory response, we quantified the levels of pro‐inflammatory cytokines at early time points of infection. p38γ/δ^−/− mice showed a significant reduction of TNFα and IFNγ serum levels compared to WT mice (Appendix Fig S4A). IL‐1β production was slightly reduced in p38γ/δ^−/− mice compared to control (Appendix Fig S4A). Accordingly, in kidney, TNFα, IL6 and IL‐1β mRNA levels were significantly reduced in p38γ/δ^−/− and LysM‐p38γ/δ^−/− mice at day 1 post‐infection, compared to WT mice (Fig 4A). The reduction in cytokine mRNA production was more pronounced in LysM‐p38γ/δ^−/− than in p38γ/δ^−/− mice. In TPL2^−/− mice infected with C. albicans, the production of cytokines in the kidney was not reduced compared to TPL2^+/+ mice (Appendix Fig S4B). IL6, TNFα, and IL‐1β mRNA levels in TPL2^−/− kidney were significantly higher than in TPL2^+/+ mice (Appendix Fig S4B) indicating that in vivo TPL2 does not mediate cytokine production downstream of p38γ/p38δ in response to C. albicans.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4. Reduced inflammation in p38γ/δ^−/− mice in response to Candida albicans infection

1. WT, p38γ/δ^−/− and LysM‐p38γ/δ^−/− mice were intravenously infected with 1 × 10⁵ CFU C. albicans and at days 1 and 3 post‐infection, relative TNFα, IL‐6 and IL‐1β mRNA expression in the kidney was determined by qPCR and normalized to β‐actin mRNA. Each symbol represents an individual mouse. Figure shows mean ± SEM (n = 5–8). ns, not significant, *P ≤ 0.05, **P ≤ 0.01 relative to WT mice. Parametric, unpaired t‐test.

2. Kidney cells from 0‐, 1‐ and 3‐day C. albicans‐treated WT, p38γ/δ^−/− and LysM‐p38γ/δ^−/− mice as in (A) were stained with anti‐CD45, ‐Ly6G and ‐F4/80 antibodies and positive cells analysed by flow cytometry. CD45⁺ cells were gated and ‐F4/80⁺ and ‐Ly6G⁺ cells analysed by flow cytometry. Representative profiles are shown. Each symbol represents an individual mouse (two to three independent experiment). Figure shows mean ± SEM (n = 5–14 mice/condition), ns, not significant; *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, relative to WT kidney cells, at each time point. Parametric, unpaired t‐test.

3. Mice were intraperitoneally infected with 5 × 10⁶ CFU C. albicans, and at day 1 post‐infection, peritoneal cells were stained and analysed as in (B). Representative profiles are shown. Each symbol represents an individual mouse. Figure shows mean ± SEM (n = 4), ns not significant; *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, relative to WT kidney cells. Parametric, unpaired t‐test.

4. Mice were infected with C. albicans as in (A), relative MIP‐2, KC and CCL2 mRNA expression in the kidney was determined by qPCR as in (A). Each symbol represents an individual mouse. Figure shows mean ± SEM (n = 3–5). ns, not significant, *P ≤ 0.05, **P ≤ 0.01 relative to WT mice. Parametric, unpaired t‐test.

We also analysed, by flow cytometry, different leucocyte cell types infiltrating the infected kidney of WT, p38γ/δ^−/− and LysM‐p38γ/δ^−/− mice, at days 0, 1 and 3 after C. albicans infection. The number of total renal‐infiltrating leucocytes (CD45⁺ cells) increased with the infection and was significantly lower in p38γ/δ^−/− and LysM‐p38γ/δ^−/− mice than in WT at day 3 post‐infection (Fig 4B). Next, we determined the recruitment of major leucocyte types involved in inflammation: macrophages, neutrophils and T cells. We found that C. albicans infection caused an increase in the amount of F4/80⁺ macrophages and Ly6G⁺ neutrophils in all genotypes after infection (Fig 4B). Macrophages and neutrophils are a major fraction of the total CD45⁺ population after C. albicans infection (Fig 4B). The amount of macrophages and neutrophils accumulated in p38γ/δ^−/− and LysM‐p38γ/δ^−/− kidney 3 days after C. albicans infection was significantly lower than in WT, although the effect in neutrophil recruitment was more pronounced than in macrophages recruitment (Fig 4B). The reduction in macrophages and neutrophils recruitment was more evident in LysM‐p38γ/δ^−/− than in p38γ/δ^−/− mice. CD4⁺, but not CD8⁺, T lymphocyte amount increased after C. albicans infection; however, they remain a minor percentage of the total leucocyte population and the difference in T‐cell accumulation was minimal between WT, p38γ/δ^−/− and LysM‐p38γ/δ^−/− mice (Appendix Fig S4C). We further confirmed the decrease in F4/80⁺ macrophages and Ly6G⁺ neutrophils recruitment using an intraperitoneal C. albicans model (Netea et al, 1999). As expected, we found a significant reduction in peritoneal neutrophil recruitment in p38γ/δ^−/− and LysM‐p38γ/δ^−/− mice compared to control; however, the recruitment of F4/80⁺ macrophages was clearly reduced only in LysM‐p38γ/δ^−/− peritoneum (Fig 4C).

The decrease in leucocyte recruitment in p38γ/δ^−/− and LysM‐p38γ/δ^−/− kidneys suggests reduced chemokine expression. Thus, CCL2, MIP‐2 and KC mRNA levels were clearly lower in the kidney of p38γ/δ^−/− and LysM‐p38γ/δ^−/− mice than in WT animals at day 1 post‐infection (Fig 4D). Together, these findings suggest that p38γ/p38δ are involved in the early inflammatory response to C. albicans by modulating cytokine and chemokine production and the recruitment of leucocytes into the C. albicans‐infected kidney. We then treated WT and p38γ/δ^−/− mice with ibuprofen, which is a commonly used antiinflammatory compound (Vilaplana et al, 2013). Ibuprofen treatment reduced kidney C. albicans load and neutrophil recruitment in WT mice to similar levels than the loss of p38γ/p38δ (Appendix Fig S5A and B). In p38γ/δ^−/− mice, the treatment with ibuprofen did not cause a major effect in fungal burden or the recruitment of neutrophils after C. albicans infection (Appendix Fig S5A and B). When we examined the effect of ibuprofen on the survival of C. albicans‐infected WT and p38γ/δ^−/− mice, we found that at early times the antiinflammatory compound caused a marked protection and an increase in the survival of WT mice (Appendix Fig S5C). Ibuprofen, however, did not affect p38γ/δ^−/− mice survival (Appendix Fig S5C). These results support that p38γ/p38δ are involved in modulating an early deleterious inflammatory response to C. albicans.

p38γ/p38δ deficiency increases macrophages and neutrophils antifungal activity

Macrophages and neutrophils regulate bacterial and fungal infections by phagocytosis and killing mechanisms (Vonk et al, 2002; Nicola et al, 2008; Lionakis et al, 2013). The production of highly reactive nitrogen and oxygen species (RNS and ROS) is one of the main mechanisms used by phagocytes to control fungal infection and is essential for C. albicans killing (Nicola et al, 2008; Naglik et al, 2014). Our data suggest that C. albicans can be eliminated more efficiently in p38γ/δ‐null mice than in WT controls. We then analysed whether or not the p38γ/p38δ deletion affected the ability of macrophages to express inducible nitric oxide synthase (iNOS), which generates nitric oxide (NO) from arginine and oxygen and is expressed after the activation of phagocytic cells. iNOS mRNA expression was upregulated upon HK‐Ca stimulation in both WT and p38γ/δ^−/− BMDM; however, the levels in p38γ/δ^−/− cells were markedly higher than in WT (Fig 5A). We also found that the loss of p38γ/p38δ significantly enhanced the production of reactive oxygen species (ROS) by C. albicans‐stimulated macrophages (Fig 5B). We then performed an ex vivo killing assay, co‐culturing C. albicans with BMDM, to assess whether the enhanced iNOS and ROS levels in p38γ/δ^−/− cells correlated with an increase in their Candida‐killing potency. Fungal phagocytosis by BMDM was similar in both genotypes (Appendix Fig S6A), indicating that the recognition of C. albicans is not affected in p38γ/δ^−/− cells. Nonetheless, p38γ/δ^−/− macrophages were significantly more efficient in the killing of live C. albicans than WT cells (Fig 5C). The ROS production and the fungal‐killing capacity of p38γ/δ^−/− neutrophils were also markedly higher compared to WT control neutrophils (Fig 5D and E), further indicating that p38γ/p38δ negatively impact the ability of phagocytic cells to clear the C. albicans, which is important for mice survival in the candidiasis model.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5. p38γ/p38δ regulate antifungal activity against Candida albicans

• A. BMDM from WT or p38γ/δ^−/− mice were exposed for the indicated times to 1 × 10⁶ CFU/ml HK‐Ca. Relative iNOS mRNA expression was determined by qPCR. Data show mean ± SEM from one representative experiment of two in triplicate, with similar results. *P ≤ 0.05, ***P ≤ 0.001 relative to WT BMDM. Parametric, unpaired t‐test.

• B. ROS production in BMDM co‐cultured with 1 × 10⁶ CFU C. albicans. Experiments were performed in triplicate. RLU, relative light unit.

• C. Candida albicans killing by BMDM. The results are represented as percentage of killing (the percentage of killed fungal cells among the phagocytosed fungus). The C. albicans/BMDM ratio was 1:10. Values are mean ± SEM (n = 6); ***P ≤ 0.001 relative to WT cells. Parametric, unpaired t‐test.

• D. ROS production in neutrophils isolated from blood of WT or p38γ/δ^−/− mice after infection. Neutrophils were stimulated with 1.5 × 10⁶/ml HK‐Ca. Experiments were performed in triplicate.

• E. Candicidal activity of neutrophils determined as in (C). Values are mean ± SEM (n = 4); *P ≤ 0.05 relative to WT cells. Parametric, unpaired t‐test.

• F, G ROS production without (F) or with (G) re‐stimulation with C. albicans in intraperitoneal immune cell infiltrates of WT or p38γ/δ^−/− mice 1 day after intraperitoneal C. albicans infection (5 × 10⁶ CFU, n = 4 mice per group).

• H. WT and p38γ/δ^−/− mice were intravenously infected with 1 × 10⁵ CFU C. albicans. Relative iNOS mRNA expression in the kidney was determined by qPCR and normalized to β‐actin mRNA. Each symbol represents an individual mouse. Figure shows mean ± SEM (n = 3–5). Only significant results are indicated, *P ≤ 0.05, relative to WT mice. Parametric, unpaired t‐test.

• I. WT and p38γ/δ^−/− mice were infected with 1 × 10⁵ CFU C. albicans and treated with 200 mg/kg body weight per day of N‐acetylcysteine (NAC) from SIGMA or with the same volume of the vehicle PBS for 5 days [WT + CA + NAC (n = 8); p38γ/δ^−/− + CA + NAC (n = 8)]. Control groups of WT and p38γ/δ^−/− mice treated with 200 mg/kg body weight per day of NAC were included to check its toxicity [WT + NAC (n = 10); p38γ/δ^−/− + NAC (n = 10)]. Control groups of WT and p38γ/δ^−/− mice infected with C. albicans were also included for comparison [WT + CA (n = 8); p38γ/δ^−/− + CA (n = 8)]. Survival was monitored as indicated. Data are presented as a Kaplan–Meier plot. ns, not significant, *P ≤ 0.05; **P ≤ 0.01. Two‐way ANOVA using GraphPad Prism software.

To examine whether the observed differences in ROS production in vitro were also seen in vivo in immune cells that are recruited to the inflammatory sites, we infected intraperitoneally WT and p38γ/δ^−/− mice with C. albicans and 24 h later measured ROS production both without (Fig 5F) or with (Fig 5G) re‐stimulation in vitro with C. albicans. Cells that infiltrate the intraperitoneal cavity of p38γ/δ^−/− mice produced significantly more ROS than cells from WT mice (Fig 5F and G). Additionally, the induction of iNOS mRNA production in the kidney of C. albicans‐infected mice was markedly higher in p38γ/δ^−/− than in WT mice (Fig 5H). All these data suggest that the protective effect of p38γ/δ deficiency against fungal infection is due to increased killing of C. albicans.

To investigate whether the increased levels of ROS observed in p38γ/δ^−/− BMDM contribute to antifungal activity of these cells against C. albicans, we treated them with the antioxidant compound N‐acetylcysteine (NAC; Victor et al, 2003). NAC reduced both ROS levels and C. albicans killing in vitro (Appendix Fig S6B and C). Moreover, we found that in vivo the antioxidant compound decreased the survival of p38γ/δ^−/−‐infected mice to similar levels than those in WT‐infected mice (Fig 5I). These results indicate that the increase in ROS production observed in p38γ/δ^−/− mice is important for the antifungal activity and protection against C. albicans infection.

Pharmacological inhibition of p38γ/p38δ ameliorates Candida albicans infection

Based on our data, we hypothesized that p38γ/p38δ inhibition in vivo might improve the outcome of C. albicans infection and provide evidence for a new therapeutic approach. We performed experiments treating mice with the p38MAPK inhibitor BIRB796. Since BIRB796 inhibits all p38 isoforms (Kuma et al, 2005), we also used as control the compound SB203580, which only blocks p38α/p38β activity (Kuma et al, 2005; Bain et al, 2007), so we could determine which effect was caused by p38γ/p38δ or by p38α/p38β inhibition. We first checked whether these compounds inhibit p38MAPKs in the kidney of infected mice. Both compounds worked equally well in vivo as shown by the reduction in TNFα mRNA expression in the kidney of C. albicans‐infected mice (Appendix Fig S7A) and inhibited p38α, as shown by the loss of its phosphorylation (Appendix Fig S7B). Moreover, BIRB796, but not SB203580, was able to decrease the phosphorylation of p38γ and p38δ induced after C. albicans infection (Appendix Fig S7C). Notably, BIRB796 treatment significantly reduced kidney fungal load at day 3 (Fig 6A), but not at day 1 (Appendix Fig S7D) in WT mice. This fungal burden reduction was similar to that observed in p38γ/δ^−/− mice (Appendix Fig S7E). BIRB796 treatment did not affect C. albicans load in p38γ/δ^−/− mice (Appendix Fig S7E). SB2013580 did not affect C. albicans growth in the kidney compared to control mice treated with the vehicle DMSO (Fig 6A). Consequently, BIRB796 treatment led to a higher fungal‐induced iNOS mRNA levels compared to controls (Fig 6B). Treatment with the inhibitor also decreased the recruitment of neutrophils to the kidney (Fig 6C). We confirmed this using the intraperitoneal C. albicans infection model and found that only the treatment with BIRB796 led to a significant reduction in peritoneal neutrophil recruitment compared to control mice treated with either DMSO or SB203580 (Fig 6D). Thus, p38γ/p38δ inhibition enhances antifungal immunity and might protect from C. albicans sepsis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6. Treatment with p38γ/p38δ inhibitor shows antifungal effects in vivo

1. WT mice were intravenously injected with 1 × 10⁵ CFU of Candida albicans and treated with 10 mg BIRB796 or SB203580 per kg body weight per day, or with the same volume of the vehicle DMSO. Kidney fungal load was determined 3 days after infection. ns, not significant, *P ≤ 0.05 (n = 5 mice/condition). Each symbol represents an individual mouse. Parametric, unpaired t‐test.

2. Mice were treated as in (A) and iNOS mRNA levels in the kidney measured 3 days after infection by qPCR. Each symbol represents an individual mouse. Figure shows mean ± SEM (n = 5 mice/condition). ns, not significant, *P ≤ 0.05. Parametric, unpaired t‐test.

3. Neutrophil infiltration in the kidney of infected WT mice, treated with BIRB796 or SB203580 inhibitor as in (A) was determined by flow cytometry. Each symbol represents an individual mouse. Figure shows mean ± SEM (n = 6 mice/condition), ns not significant; ***P ≤ 0.001. Parametric, unpaired t‐test.

4. Mice were intraperitoneally infected with 5 × 10⁶ CFU C. albicans and treated with 10 mg/kg body weight per day BIRB796 or SB203580, or with the same volume of the vehicle DMSO. Neutrophil infiltration in the peritoneum was measured by flow cytometry at day 1 post‐infection. Each symbol represents an individual mouse. Figure shows mean ± SEM (n = 4 mice/condition), ns, not significant; *P ≤ 0.05, **P ≤ 0.01. Parametric, unpaired t‐test.

Mice

All mice (13–18 weeks old female) were housed in specific pathogen‐free conditions in the CNB‐CSIC animal house, and all animal procedures were performed in accordance with national and EU guidelines, with the approval of the Centro Nacional de Biotecnología Animal Ethics Committee, CSIC and Comunidad de Madrid (Reference: CAM PROEX 316/15).

C57BL/6J TPL2^+/+ and C57BL/6J TPL2^−/− littermates were produced from heterozygous mice (Rodriguez et al, 2008). C57BL/6J WT, p38γ/δ^flox/flox, p38γ/δ^−/− and LysM‐Cre^+/−‐p38γ/δ^flox/flox (called here LysM‐Crep38γ/δ^−/−) mice have been described (Risco et al, 2012; Zur et al, 2015). Double knockout mice have been used instead of single knockout mice because each of these p38MAPKs normally compensates the loss of the other in many biological processes (Escós et al, 2016). WT and p38γ/δ^flox/flox mice have been used in all the in vivo experiments as control, with similar results. WT, LysM‐Cre^+/− and p38γ/δ^flox/flox mice were used as control when cell recruitment was analysed in the intraperitoneal C. albicans model. The recruitment of F4/80⁺ and Ly6G⁺ cells in this model was similar in all mouse lines (Appendix Fig S8). Mouse genotypes were determined by PCR. All strains were backcrossed onto the C57BL/6 strain for at least nine generations.

Antibody

The description of all the antibodies and the dilution used in this study is provided in Appendix Table S1.

Candida albicans infection

Candida albicans (strain SC5314) was grown on YPD agar plates at 30°C for 48 h. Eight‐ to 12‐week‐old female mice were infected intravenously with 1 × 10⁵ colony‐forming units (CFU) of C. albicans and monitored daily for weight and survival. Kidney fungal burden was determined at indicated times post‐infection by plating the kidney homogenates in serial dilutions on YPD agar plates. After 48 h, CFU were counted. Kidneys from infected mice were fixed in 4% formalin and embedded in paraffin. Serial sections were examined microscopically after staining with periodic acid Schiff (PAS) and haematoxylin–eosin.

Bone marrow‐derived macrophages (BMDM) and stimulation

BMDM were isolated and cultured as described in Appendix Supplementary Methods (Risco et al, 2012). BMDM were stimulated with: 1 × 10⁶/ml HK‐Ca, 10 μg/ml Zymosan, 5 μg/ml Imiquimod‐R837, 200 ng/ml Pam3Cys, 200 ng/ml Poly I‐C, 250 ng/ml ODN‐1668 (InvivoGen); 10 μg/ml Curdlan (a water‐insoluble β‐1,3 polysaccharide from Alcaligenes faecalis), 100 ng/ml LPS, 100 ng/ml PMA (Sigma). Where indicated, cells were pre‐treated with DMSO, SB203580, BIRB0796, C34; PD184352; PRT062607; BI605906; Sorafenib (Bay 43‐9006) or NG25. Cells were lysed as described in Appendix Supplementary Methods. For mRNA expression analysis, cells were lysed with NZYol (NZYtech) and RNA extracted using a standard protocol with chloroform–isopropanol–ethanol.

Neutrophil isolation

Neutrophils were obtained from adult WT and p38γ/δ^−/− mice blood followed by hypotonic red blood cell lysis as described in Appendix Supplementary Methods.

Isolation and stimulation of human monocytes

This protocol is described in Appendix Supplementary Methods. Informed consent was obtained from all healthy volunteers. The experiments conformed to the principles set out in the WMA Declaration of Helsinki and the Department of Health and Human Services Belmont Report.

Phagocytosis and killing of Candida albicans by macrophages and neutrophils

BMDM and neutrophils were obtained, and phagocytosis and killing were performed using the method described earlier (Vonk et al, 2012) at a Candida/BMDM or Candida/neutrophils ratio of 1:10.

Measurement of reactive oxygen species production

Production of ROS was measured with an assay using luminol as the probe in real time over 220 min. 75,000 neutrophils or 500,000 BMDM or peritoneal cells were plated in 200 μl culture medium (0.05% FBS in HBSS) on a 96‐well sterile luminometer plate (Costar, Corning, NY). Cells were stimulated or not as indicated, and L‐012 (Wako Chemicals, Osaka, Japan) was incorporated to the medium (7.75 μg/well final concentration) at the beginning of the stimulation. Chemiluminescence was measured at 1‐min intervals and expressed as relative light units (RLU).

Statistical analysis

In vitro experiments have been performed at least twice with three independent replicates per experiment. For the analysis of mouse survival, production of inflammatory molecules, cell recruitment and CFU in the kidney, the size of the groups was established according to the Spanish ethical legislation for animal experiments. At least 4–5 mice per group were used. Differences in mouse survival were analysed by two‐way ANOVA using GraphPad Prism software. Other data were analysed using Student's t‐test. In all cases, P‐values < 0.05 were considered significant. Data are shown as mean ± SEM. The exact n and P‐values are given in Appendix Tables S2 and S3.