Glucocorticoid receptor dimerization induces MKP1 to protect against TNF-induced inflammation (original) (raw)
GRdim/dim mice are extremely sensitive to TNF and fail to induce MKP1. Whether the TA potential of GR is necessary for its antiinflammatory activity is somewhat controversial. In this study, we investigated whether the induction of GRE genes is necessary for the antiinflammatory functions counteracting TNF lethality. We injected WT control (GRwt/wt) and GRdim/dim mice i.p. with 25 μg TNF (a nonlethal dose for FVB/N mice) and monitored survival and body temperature. Mortality rate was significantly higher and hypothermia more pronounced in GRdim/dim than in GRwt/wt mice (Figure 1, A and B). Because IL-6 level is a good indicator of TNF sensitivity (26, 27), we measured IL-6 protein levels in circulation 0.5 hours after TNF stimulation. The significantly higher IL-6 levels in GRdim/dim mice (Figure 1C) confirmed their hypersensitivity to TNF. Furthermore, H&E staining of ileum samples showed that TNF treatment resulted in more severe intestinal damage in GRdim/dim than in GRwt/wt mice (Supplemental Figure 1A; supplemental material available online with this article; doi:10.1172/JCI60006DS1). These data showed that dimerization of GR was essential for protection against TNF-induced shock, presumably by the induction of antiinflammatory GRE genes by endogenous GCs.
GRdim/dim mice are hypersensitive to TNF lethality and cannot induce MKP1. (A and B) Survival (A) and body temperature (B) of GRwt/wt (n = 7) and GRdim/dim (n = 9) mice after i.p. injection of 25 μg TNF. **P < 0.01, GRdim/dim vs. GRwt/wt. (C) IL-6 concentration in serum 0 and 0.5 hours after injection of 25 μg TNF (n = 5 per group). **P < 0.01, ***P < 0.001 vs. 0 hours; *P < 0.05 as indicated by brackets. (D) Mice were injected i.p. with PBS or 500 μg DEX. 1 hour later, they were euthanized, and livers were harvested for qPCR analysis of Mkp1 and Sgk1 levels (n = 4 per group). (E) Mice were injected i.p. with PBS or 500 μg DEX. 1 hour later, mice were euthanized, and livers were harvested for ELISA of MKP1 protein levels (n = 6 per group). (F) qPCR analysis of Mkp1 mRNA expression in livers of mice treated with 25 μg TNF for 0.5 hours (n = 5 per group). (D–F) *P < 0.05, **P < 0.01, ***P < 0.001 vs. PBS or as indicated by brackets. (A–F) Black bars and symbols, GRwt/wt; white bars and symbols, GRdim/dim.
One of the most potent antiinflammatory genes induced by GR is Mkp1. To evaluate Mkp1 expression in our setting, GRwt/wt and GRdim/dim mice were injected i.p. with 500 μg of a synthetic GC, dexamethasone (DEX), or with PBS. Liver samples were isolated 1 hour after DEX treatment, and mRNA expression of Mkp1 and Sgk1, a well-known GR dimer–dependent TA gene, was measured by quantitative real-time PCR (qPCR). DEX treatment of GRwt/wt mice resulted in a strong induction of Mkp1 and Sgk1 expression in liver; this induction was significantly weaker in GRdim/dim mice (Figure 1D). Because Mkp1 regulation by GCs is not restricted to the transcriptional level, MKP1 protein levels were also measured. MKP1 protein was induced after DEX injection only in GRwt/wt mice (Figure 1E), which indicates that the induction of both Mkp1 gene expression and MKP1 protein relies on GR dimerization. Since the regulation of MKP1 is controversial, we performed ChIP analysis to test whether the GR in GRdim/dim mice is indeed unable to bind to the GRE of the Mkp1 gene. GRwt/wt and GRdim/dim mice were injected i.p. with 500 μg DEX, and liver samples were isolated at time point 0 or after 1 hour of DEX treatment. ChIP analysis was performed on these liver samples, and the isolated DNA was checked for Mkp1 expression. The results showed that DEX treatment of GRwt/wt mice resulted in strong binding of GR to Mkp1 (Supplemental Figure 1, B–E). This binding was substantially reduced in GRdim/dim mice, which indicates that the binding of GR to Mkp1 and subsequent induction of this gene indeed require GR dimerization.
Furthermore, as it is known that inflammatory stimuli can also induce Mkp1, most likely via the production of endogenous GCs, we investigated the induction of Mkp1 after TNF treatment. We injected GRwt/wt and GRdim/dim mice i.p. with 25 μg TNF and harvested liver samples after 0.5 hours. In GRwt/wt mice, TNF treatment resulted in the induction of Mkp1, whereas in GRdim/dim mice, this induction was significantly reduced (Figure 1F). These data indicate that GR dimerization is necessary for the induction of Mkp1 by DEX or by endogenous GCs induced by TNF. Taken together, our data demonstrated that dimerization of GR was indispensable for protection against TNF lethality. Furthermore, our findings provided evidence that Mkp1 is a GR dimer–dependent gene that might be involved in this protection against TNF.
Mkp1–/– mice are hypersensitive to TNF lethality. To examine whether MKP1 is indeed a critical player in the protection against TNF-induced shock, we injected Mkp1+/+ and Mkp1–/– mice i.p. with 5 μg TNF, which is not lethal for normal C57BL6/J mice. Mortality rate was higher and hypothermia was more pronounced in Mkp1–/– than in Mkp1+/+ mice (Figure 2, A and B). These findings indicate that MKP1 plays a crucial role in controlling TNF lethality. To investigate the underlying mechanism, Mkp1+/+ and Mkp1–/– mice were injected i.p. with 5 μg TNF, and blood, liver, and ileum samples were obtained 0, 1, and 6 hours after TNF treatment. We measured IL-6 protein in circulation and Il6 gene expression levels in liver. TNF induced high levels of hepatic Il6 mRNA and serum IL-6 protein 6 hours after TNF treatment, especially in Mkp1–/– mice (Figure 2C). Several other cytokines and chemokines in circulation were also higher in Mkp1–/– than in Mkp1+/+ mice, especially 6 hours after TNF treatment (Supplemental Figure 2). Additionally, we tested the expression levels of different TNF-induced proinflammatory genes in liver samples and in intestinal epithelial cells (IECs). Again, Mkp1–/– livers had significantly higher mRNA levels of Ccl5 (encoding CCL5 or RANTES), Timp1 (encoding the MMP inhibitor TIMP1) and Nos2 (encoding iNOS) 6 hours after TNF treatment (Figure 2, D and E). Strikingly, Mkp1–/– mice had significantly higher mRNA levels of these proinflammatory genes in IECs as early as 1 hour after TNF. These results indicated that MKP1 has an antiinflammatory effect that protects against TNF-induced lethal inflammation.
Mkp1–/– mice are hypersensitive to TNF-induced lethality. (A and B) Survival (A) and body temperature (B) of Mkp1+/+ (n = 12) and Mkp1–/– (n = 22) mice after injection of 5 μg TNF. *P < 0.05, ***P < 0.001, Mkp1–/– vs. Mkp1+/+. (C) Serum IL-6 levels and liver Il6 mRNA levels 0, 1, and 6 hours after 5 μg TNF (n = 5 per group). (D and E) Mice were injected i.p. with 5 μg TNF, and 0, 1, and 6 hours later, they were euthanized, and livers (D) and IECs (E) were isolated for qPCR analysis of Ccl5, Timp1 and Nos2 levels (n = 5 per group). (F) Standard H&E and active caspase 3 staining of ileum samples (n = 5 per group). Representative images are shown. The ileum was sampled 0 and 1 hour after injection of 5 μg TNF. The micrograph at the 0-hour time point is representative of both Mkp1+/+ and Mkp1–/– mice. Scale bars: 100 μm. Original magnification, ×40. (C–F) *P < 0.05, **P < 0.01, ***P < 0.001 vs. 0 hours or as indicated by brackets. (G and H) Survival of Mkp1+/+ (G) and Mkp1–/– (H) mice pretreated with 10 mg/kg DEX (squares; n = 8 per group) or solvent (circles; n = 8 [_Mkp1+/+_]; 7 [_Mkp1–/–_]) and injected with 15 μg (G) or 7.5 μg (H) TNF. **P < 0.01, DEX vs. solvent. (A–H) Black bars and symbols, Mkp1+/+; white bars and symbols, Mkp1–/–.
As it was recently shown that MKP1 inhibits TNF-induced endothelial barrier dysfunction and apoptosis in HUVECs (28), and since TNF can cause intestinal damage and enterocyte apoptosis (5, 29), we investigated acute cell death in ileum samples of Mkp1+/+ and Mkp1–/– mice using H&E staining. As early as 1 hour after TNF challenge, the toxic effect of TNF was much more pronounced in Mkp1–/– than in Mkp1+/+ mice (Figure 2F). The intestinal damage was mainly characterized by loosening of the lamina propria, erosion of the villi, and loss of goblet cells. The degree of intestinal damage was scored according to a previously published method (30). Intestines of Mkp1–/– mice were significantly more damaged than those of control Mkp1+/+ mice (Figure 2F). Furthermore, ileum samples were stained for active caspase 3 as a marker of apoptosis. In agreement with the H&E staining, we found significantly more cells expressing active caspase 3 in ileum samples of Mkp1–/– mice (Figure 2F). These data indicate that within 1 hour after TNF treatment, Mkp1–/– mouse intestine is severely damaged because of a high rate of apoptosis combined with severe inflammation.
Additionally, to investigate whether GC inhibition of the response to TNF occurs through MKP1, we examined whether pretreating Mkp1–/– mice with endogenous GCs can protect against TNF-induced lethal shock. We injected the mice with 10 mg/kg DEX and 0.5 hours later with LD80 of TNF (15 μg for Mkp1+/+; 7.5 μg for Mkp1–/–). In keeping with previous reports by our group (7, 9), Mkp1+/+ mice were protected against TNF by pretreatment with DEX compared with mice pretreated with diluted methanol solvent (90% vs. 10% survival; Figure 2G). However, pretreatment of Mkp1–/– mice with DEX or solvent failed to protect them against TNF lethality (10% survival; Figure 2H). These data demonstrated that MKP1 was indispensable in the protective action of GCs against TNF.
Prolonged JNK activation in Mkp1–/– and GRdim/dim mice. MKP1 is known to dephosphorylate and hence inactivate MAPKs, particularly JNK and p38, but also ERK (16). As MAPKs are activated by TNF and mediate part of the TNF proinflammatory induction profile (31), we examined the status of activated MAPKs in liver samples of Mkp1+/+ and Mkp1–/– mice. Both mouse groups were injected i.p. with 5 μg TNF and sacrificed at time point 0 or after 10 minutes or 0.5, 1, or 2 hours, and hepatic MAPK levels were examined. We observed rapid induction of phospho-JNK1/2 10 minutes after TNF injection in both Mkp1+/+ and Mkp1–/– mice (Figure 3A). This induction declined rapidly in Mkp1+/+ mice, starting 0.5 hours after TNF injection, when phospho-JNK1/2 levels in Mkp1–/– mice were still increasing (Figure 3A and Supplemental Figure 3). These data indicate that TNF treatment results in prolonged activation of JNK1/2 in Mkp1–/– compared with Mkp1+/+ mice. Levels of phospho-ERK and phospho-p38 were also examined, but no differences between Mkp1+/+ and Mkp1–/– mice were observed (Supplemental Figure 4, A and B).
JNK phosphorylation in liver is higher in Mkp1–/– and GRdim/dim mice. (A) Western blot analysis of phospho-JNK1/2 protein levels in livers of Mkp1+/+ and Mkp1–/– mice. Mice were injected i.p. with 5 μg TNF, and livers were harvested at the indicated times after challenge. The phospho-JNK1/2 bands (46 kDa and 54 kDa) were normalized to the intensities of the total JNK1/2 and actin bands (42 kDa). Black bars, Mkp1+/+; white bars, Mkp1–/–. (B) Western blot analysis of phospho-JNK1/2 protein levels in livers of GRwt/wt and GRdim/dim mice. Mice were treated with 25 μg TNF; 0 and 0.5 hours later, they were euthanized, and livers were obtained for Western blot analysis. Normalized values are also shown. Black bars, GRwt/wt; white bars, GRdim/dim. (A and B) *P < 0.05, **P < 0.01, ***P < 0.001 vs. 0 hours or as indicated by brackets. See complete unedited blots in the supplemental material.
Because GRdim/dim mice were very sensitive to TNF and had weaker induction of MKP1, it is conceivable that these mice also display stronger activation of JNK1/2 upon TNF challenge. Therefore, we evaluated phospho-JNK1/2 levels 0.5 hours after TNF challenge in GRwt/wt and GRdim/dim mice. Interestingly, activation of JNK1/2 was stronger in GRdim/dim mice (Figure 3B), as was observed in Mkp1–/– mice. These observations indicated that activation of JNK1/2 is stronger in Mkp1–/– and GRdim/dim mice.
JNK2, not JNK1, is an essential mediator of TNF-induced lethality. Because phospho-JNK1/2 levels were significantly higher in livers of Mkp1–/– and GRdim/dim mice than in their WT controls, we studied the response of JNK-deficient mice to TNF. WT control, Jnk1–/–, and Jnk2–/– mice were injected i.p. with 10 μg TNF (an LD50 dose for C57BL/6 mice), and survival and body temperature were monitored. Although Jnk1–/– mice showed no change in sensitivity to TNF, Jnk2–/– mice were significantly more protected against TNF than control mice, as shown by the significantly lower mortality rate and less severe hypothermia (Figure 4, A and B). These findings indicated that JNK2, but not JNK1, is an essential mediator of TNF-induced lethal shock.
Jnk2–/– mice are resistant to TNF-induced shock. (A) Survival of control (black squares; n = 19), Jnk2–/– (white squares; n = 12), and Jnk1–/– (gray circles; n = 6) mice after i.p. injection of 10 μg TNF. **P < 0.01 vs. control. (B) Body temperature of control (n = 12), Jnk2–/– (n = 12), and Jnk1–/– (n = 6) mice after injection of 10 μg TNF. *P < 0.05. (C) Serum IL-6 levels and liver Il6 mRNA levels 0, 1, and 6 hours after challenge with 10 μg TNF (n = 5 per group). (D and E) Mice were injected i.p. with 10 μg TNF; 0, 1, and 6 hours later, they were euthanized, and livers (D) and IECs (E) were obtained for qPCR analysis of Ccl5, Timp1, and Nos2 levels (n = 5 per group). (F) Standard H&E and active caspase 3 staining of ileum samples (n = 5 per group). Representative images are shown. The ileum was sampled 0 and 1 hours after TNF injection. The micrograph at 0 hours is representative of both control and Jnk2–/– mice. Scale bars: 100 μm. Original magnification, ×40. (C–F) *P < 0.05, **P < 0.01, #P < 0.001 vs. 0 hours or as indicated by brackets. (G) Relative permeability 8 hours after injection of 10 μg TNF or PBS. **P < 0.01. (B–G) Black bars and symbols, control; white bars and symbols, Jnk2–/–; gray bars and symbols, Jnk1–/–.
Furthermore, control and Jnk2–/– mice were injected i.p. with 10 μg TNF, and blood, liver, and ileum samples were harvested 0, 1, and 6 hours later. IL-6 protein and Il6 mRNA levels, cytokine and chemokine levels in circulation, and proinflammatory gene expression levels in liver clearly demonstrated that Jnk2–/– mice were protected against TNF-induced inflammation, especially at the later 6-hour time point (Figure 4, C and D, and Supplemental Figure 5). Together, these data indicated that JNK2 has a proinflammatory role.
Moreover, measuring proinflammatory gene expression levels in IECs and stainings of ileum samples with H&E and for active caspase 3 showed that inflammatory state, tissue damage, and acute cell death 1 hour after TNF injection was much more pronounced in the intestine of Jnk2+/+ than in Jnk2–/– mice (Figure 4, E and F). As these characteristics were absent in Jnk2–/– mice, we hypothesized that JNK2 might be the mediator of TNF-induced intestinal permeability. We therefore injected Jnk2+/+ and Jnk2–/– mice with 10 μg TNF, followed 3 hours later by oral administration of 25 mg/ml FITC-dextran. Blood samples were collected 8 hours after TNF challenge, and plasma was tested for FITC signal. Jnk2+/+ mice showed a stronger signal than Jnk2–/– mice (Figure 4G), which indicates that TNF induced more intestinal permeability in the Jnk2–/– mouse. Taken together, these observations indicated that the effects of TNF on the epithelium (i.e., induction of inflammation and apoptosis) started very early, by 1 hour after TNF. Increased inflammation in liver and circulation seemed to be secondary to intestinal damage, as shown by the defect in intestinal permeability. Additionally, our findings provided evidence that JNK2 is a critical mediator of these TNF effects.
Mkp1–/–Jnk2–/– and GRdim/dimJnk2–/– mice are less sensitive to TNF. Since Mkp1–/– mice were very sensitive to the in vivo effects of TNF, and since MKP1 also dephosphorylated JNK1/2 in the TNF model, we wondered whether the sensitivity of Mkp1–/– mice to TNF is due to overactive JNK kinases. To test this hypothesis, we generated Mkp1–/–Jnk2–/– mice and studied their sensitivity to TNF. Control Mkp1+/+Jnk2+/+, Mkp1–/–, and Mkp1–/–Jnk2–/– mice were injected i.p. with 5 μg TNF (which is lethal for Mkp1–/– mice), and mortality and body temperature were monitored. The sensitivity of Mkp1–/–Jnk2–/– mice to TNF was intermediate between that of Mkp1+/+Jnk2+/+ and Mkp1–/– mice (Figure 5, A and B), which suggests that Jnk2 rescues TNF sensitivity, at least in part, in Mkp1–/– mice. The intermediate sensitivity of Mkp1–/–Jnk2–/– mice was confirmed by measuring IL-6 protein in circulation and Il6 mRNA in liver (Figure 5C). Furthermore, H&E staining of ileum samples 1 hour after TNF injection showed that intestinal damage was comparable in Mkp1+/+Jnk2+/+ and Mkp1–/–Jnk2–/– mice, but much more pronounced in Mkp1–/– mice (Figure 5D). These observations indicate that Mkp1–/–Jnk2–/– mice are much less sensitive to TNF than are Mkp1–/– mice, which indicates that the sensitivity of Mkp1–/– mice can be partly rescued by specifically inhibiting JNK2. We therefore conclude that JNK2 is one of the critical players responsible for the increased sensitivity of Mkp1–/– mice to TNF.
Mkp1–/–Jnk2–/– mice show reduced sensitivity to TNF. (A) Survival of Mkp1+/+Jnk2+/+ (black squares; n = 22), Mkp1–/– (gray circles; n = 34), and Mkp1–/–Jnk2–/– (white squares; n = 24) mice after i.p. injection of 5 μg TNF. *P < 0.05, ***P < 0.001 vs. control; **P < 0.01, Mkp1–/–Jnk2–/– vs. Mkp1–/–. (B) Body temperature of mice as in A. **P < 0.01, ***P < 0.001. (C) Serum IL-6 concentrations and liver Il6 mRNA levels 0, 1, and 6 hours after challenge with 5 μg TNF (n = 5 per group). (D) Standard H&E staining of ileum samples (n = 5 per group). The ileum was sampled 0 and 1 hour after i.p. injection of 5 μg TNF. The micrograph at 0 hour is representative of all genotypes. Scale bars: 100 μm. Original magnification, ×40. (C and D) *P < 0.05, **P < 0.01, ***P < 0.001 vs. 0 hours or as indicated by brackets. (B–D) Black bars, Mkp1+/+Jnk2+/+; white bars, Mkp1–/–Jnk2–/–; gray bars, Mkp1–/–. (E and F) Survival (E) and body temperature (F) of GRwt/wt_Jnk2+/+_ (black bars and symbols; n = 7), GRdim/dim (gray bars and gray circles; n = 5), and GRdim/dim_Jnk2–/–_ (white bars and white squares; n = 3) mice after i.p. injection of 20 μg TNF. *P < 0.05, **P < 0.01, ***P < 0.001.
Furthermore, we assessed the response of GRdim/dim_Jnk2–/–_ mice to shock. Control GRwt/wt_Jnk2+/+, GRdim/dim, and GRdim/dim_Jnk2–/– mice were injected i.p. with 20 μg TNF (which is lethal for GRdim/dim mice with a mixed C57BL/6-FvB background), and mortality and body temperature were monitored. The sensitivity of GRdim/dim_Jnk2–/–_ mice was intermediate between those of GRwt/wt_Jnk2+/+_ and GRdim/dim mice (Figure 5, E and F). These results suggest that JNK2, at least in part, rescues the sensitivity of GRdim/dim mice to TNF, as it did in Mkp1–/–Jnk2–/– mice. Taken together, these data indicate that dimerized GR protects against TNF-induced shock by induction of MKP1 and subsequent inhibition of JNK2.