Myeloid mineralocorticoid receptor controls macrophage polarization and cardiovascular hypertrophy and remodeling in mice (original) (raw)
MR controls macrophage polarization. To test the role of MR directly in macrophages, we investigated the effects MR agonists and antagonists on the in vitro phenotype in classical and alternative activation of the macrophages. Isolated peritoneal thioglycolate–elicited macrophages (PEMs) cultured in charcoal/dextan-stripped media (C/D medium), to remove endogenous steroids, responded to aldosterone with an increase in the mRNA of a classical activation marker, Tnfa. This activation was inhibited by the MR antagonist eplerenone but not the GR antagonist RU486 (Figure 1A). The dose response curve was also consistent with MR binding. Aldosterone induced other M1 markers as well (Figure 1B). LPS stimulation of M1 markers was similarly enhanced by aldosterone and blocked by MR antagonists (Figure 1C).
MR controls macrophage polarization. We studied the in vitro effects of MR agonists and antagonists in cultured PEMs. (A) Stimulation of TNF-α expression by aldosterone (Aldo; 10 nM) in steroid-depleted serum is abolished by MR but not GR antagonist. The dose response of Tnfa expression to different concentrations of aldosterone in steroid-depleted medium shows a high-affinity response consistent with MR. (B) Aldosterone (10 nM) causes increases in M1 classically activated proinflammatory genes in steroid-depleted medium. (C) Aldosterone (10 nM) increases the Tnfa response to LPS in steroid-depleted medium that is blocked by eplerenone (Epl; 5 μM). (D) In normal serum (10% FBS), the MR antagonist spironolactone (Spiro; 5 μM for 24 hours) decreases proinflammatory gene expression. (E) Corticosterone increases Tnfa expression in steroid-depleted medium that is blocked by the GR antagonist RU486 (5 μM) but not eplerenone (5 μM). (F) Twenty-four hours of 1 μM RU26752, a specific MR antagonist, increases AMϕ marker expression. (G) The MR antagonist RU26752 (1 μM) increases some antifibrotic (Htra1, a TGF-β inhibitor) and cardioprotective (e.g., Adm) genes and decreases some profibrotic genes (e.g., Pai1). #P < 0.03, †P < 0.038, ‡P < 0.034, §P < 0.025, Benjamini-Hochberg correction; *P < 0.05, **P < 0.01, ***P < 0.001 by 2-way ANOVA with Bonferroni post-tests.
Most reports fail to show any expression of the glucocorticoid-inactivating enzyme 11βHSD2 (19), and we did not detect any 11_β_HSD2 mRNA by PCR in PEMs (data not shown). Therefore, the ligand is often thought to be glucocorticoids, as in the hippocampus (20). When cultured in normal serum and stimulated with LPS, eplerenone suppressed M1 markers, suggesting that MR is occupied and stimulatory when normal serum is present (Figure 1D).
However, when we tested this hypothesis in C/D medium, we found a proinflammatory effect of corticosterone at low concentrations, but this effect was blocked by a GR antagonist and not by an MR antagonist, consistent with previous reports (Figure 1E) (21). This raised the possibility that eplerenone was having an inverse agonist effect, hence suppressing the effect of MR below baseline. However, in C/D medium, eplerenone had no effect on baseline expression of Tnfa and did not block effects below baseline (Figure 1, A and E), which is consistent with occupation by an endogenous steroid in the presence of steroid.
Since classical and alternative activation have been reported to be competing pathways, we determined the effect on AMϕ markers, which showed a stimulation of the majority of the markers tested (Figure 1F). Because of dramatic effects on fibrosis by aldosterone in vivo and the role of macrophages in fibrosis, we also tested the changes in expression of genes involved in fibrosis (Figure 1G). These included the profibrotic Tgfb and plasminogen activator inhibitor–1 (Pai1), with Pai1 being significantly suppressed. An inhibitor of TGF-β, Htra1, was significantly increased, as were two other genes associated with fibrosis, Pdk4 and Cdh2.
To better understand and test the role of MR in myeloid cells, we inactivated the MR gene in a myeloid-specific knockout (MyMRKO). This knockout was produced by crossing floxed MR allele (22) with LysM-Cre (23) and resulted in the near complete inactivation of the gene in isolated PEMs (Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI41080DS1). We then used quantitative PCR (QPCR) to determine the expression changes associated with classical (M1) and alternatively activated macrophages (AMϕ). Similar to the in vitro response to MR antagonists, PEMs showed a reduction in most of the M1 markers and an increase in many AMϕ markers (Figure 2A). Most notably, expression of iNOS and Cd36 was not altered.
MyMRKO causes alteration similar to that of MR antagonists in macrophages. PEMs were isolate from littermate FC or MyMRKO mice. (A) PEMs isolated from MyMRKO (MRKO) compared with FC mice show increased expression of AMϕ markers and decreased M1 markers. (B) Differences in expression of genes associated with fibrosis. (C) MyMRKO or MR antagonist (1 μM RU26752) alteration of LPS response. (D) Increases in AMϕ markers with IL-4 stimulation. #P < 0.035, †P < 0.031, Benjamini-Hochberg correction; *P < 0.05, **P < 0.01 by 2-tailed Student’s t test for multiple gene comparisons corrected for false discovery rate or 2-way ANOVA with Bonferroni post-tests.
Also similar to the in vitro effects of MR antagonists, the knockout of MR caused parallel changes in genes associated with fibrosis (Figure 2B). A comparison of mRNA levels between MRKO and floxed control (FC) macrophages using Affymetrix expression chips demonstrated that MR regulates multiple macrophage functions linked to wounding responses, including extracellular matrix structure, BMP signaling, and redox control (Supplemental Figure 2), consistent with a broad range of changes involved in tissue modeling.
When stimulated with LPS, MR antagonists inhibited the M1 response as before in WT cells but not in MRKO PEMs (Figure 2C). These results show that the effects of MR antagonists are MR dependent and not due to an off-target effect. MRKO PEMs showed reduced response for some but not all genes, suggesting that there was some compensation with the knockout compared with acute inhibition by MR antagonists.
IL-4, an inducer of one type of AMϕ, synergized with MRKO to display enhanced expression of AMϕ markers (Figure 2D). This result may indicate that the MRKO macrophages are primed to polarize in the alternative activation direction, as was recently reported for stimulation of PPARγ (13). Interestingly, we demonstrate a decrease in Htra1 expression with IL-4 that is counteracted in the MRKO (Figure 2D). This result illustrates that there are important differences between the AMϕ phenotype induced by MR inhibition and that induced by IL-4. We also determined the responses in resident peritoneal macrophages (Supplemental Figure 3). These macrophages had 30%–40% remaining MR mRNA expression, indicating incomplete knockout. Since this was after isolation and culture, these results are comparable to the incomplete knockout of resident macrophage lineage cells in the brain, which do not express LysM in the quiescent state (24). Responses to LPS were largely intact (Supplemental Figure 3A), but there was an increase in the responsiveness to IL-4 for some AMϕ genes (Supplemental Figure 3B).
Myeloid MRKO is protective against cardiac hypertrophy, inflammation, and fibrosis induced by L-NAME/Ang II. To test the hypothesis that MyMRKO would phenocopy the effects of MR antagonists in this in vivo model, we treated MyMRKO and control mice with a combination of a NOS inhibitor, _N_G-nitro-l-arginine methyl ester (L-NAME), and Ang II to generate cardiac and vascular damage and fibrosis (25). The cardiovascular injury in this model is mitigated by treatment with MR antagonists (26). Plasma corticosterone and aldosterone levels were similar in control and MyMRKO animals (Supplemental Table 1).
Cardiac hypertrophy was significantly less pronounced in the MyMRKO (Figure 3A). Importantly, there was also inhibition of the fetal gene program (increases in ANP and BNP and downregulation of α_-MHC_; Figure 3A), indicating an important role for myeloid cells in modulating the cardiomyocyte responses and gene expression.
MyMRKO protects against L-NAME/Ang II–induced cardiac hypertrophy and fibrosis. (A) Heart weight to body weight ratios and gene expression of markers of cardiac hypertrophy in FC and MyMRKO mice with or without (–) L-NAME/Ang II treatment (see Methods). (B) Representative H&E and picrosirius red staining of cross sections of heart samples. Fibrotic tissues stained red. Scale bar: 100 μm. (C) Quantification of interstitial fibrotic areas. (D) Gene expression of markers of cardiac fibrosis. *P < 0.05, **P < 0.01, ***P < 0.001 by 2-way ANOVA Bonferroni post-tests.
The MyMRKO mice showed dramatically less interstitial fibrosis, measured by picrosirius staining, which could have also contributed to the decreased heart size (Figure 3, B and C). This is essentially identical to the reported effects of eplerenone in this system (26). Expression of genes important in fibrosis, such as collagen III, Tgfb, and Pai1, was coordinately suppressed by MyMRKO compared with controls in the treated groups (Figure 3D).
Inflammatory infiltration of macrophages was also suppressed, as revealed by histological staining for the macrophage marker Mac2 (Figure 4A) and quantified in Figure 4B. This result indicates an alteration in myeloid trafficking and is corroborated by the decrease in the expression of the macrophage marker F4/80 (Figure 4C). Markers of M1 macrophages were suppressed (Tnfa, RANTES), whereas markers of AMϕ (Fizz1, F13a1) were increased (Figure 4D) in MyMRKO relative to L-NAME/Ang II–treated FC mice. Similar findings are shown for perivascular inflammation and fibrosis (Figure 5).
MyMRKO reduces L-NAME/Ang II–induced interstitial macrophage recruitment in heart. (A) Representative immunofluorescence assay of cross sections of heart samples from FC and MyMRKO mice with or without L-NAME/Ang II treatment. Mac2 antibody was used to detect macrophages. Scale bar: 100 μm. BF, bright field. (B) Quantification of interstitial macrophages stained by Mac2. The results are expressed as macrophage counts per high-power field (×400) under microscope. (C) Gene expression of macrophage marker F4/80. (D) Gene expression of M1 markers (Tnfa and RANTES) and AMF markers (Fizz1 and F13a1). *P < 0.05, **P < 0.01, ***P < 0.001 by 2-way ANOVA Bonferroni post-tests.
MyMRKO reduces perivascular fibrosis and macrophage recruitment induced by L-NAME/Ang II in heart. (A) Representative picrosirius red staining and immunofluorescence assay of cross sections of heart samples from FC and MyMRKO mice with or without L-NAME/Ang II treatment. Fibrotic tissues stained red. Mac2 antibody was used to detect macrophages. Scale bar: 100 μm. (B) Quantification of perivascular fibrotic areas. (C) Quantification of perivascular macrophages stained by Mac2. The results are expressed as macrophage counts per high-power field (×400) under microscope. ***P < 0.001 by 2-way ANOVA Bonferroni post-tests.
MyMRKO reduces vascular hypertrophy, inflammation, and fibrosis. L-NAME/Ang II–induced aortic remodeling, including fibrosis, decreased in the MyMRKO compared with treated FC mice. Picrosirius staining showed decreased fibrotic area in the treated MyMRKO mice compared with controls (Figure 6, A and B). The fibrotic area in treated controls appeared to be more diffuse within the expanded adventitia. Expression of the fibrosis-associated genes collagen III, Tgfb, and Pai1 was suppressed in treated MyMRKO mice (Figure 6C). Similarly, aortic wall thickness was decreased in treated MyMRKO animals (Figure 6D).
MyMRKO protects against L-NAME/Ang II–induced aortic fibrosis and hypertrophy. (A) Representative picrosirius red staining of cross sections of aortas from FC and MyMRKO mice with or without L-NAME/Ang II treatment. Fibrotic tissues stained red. Scale bar: 100 μm. (B) Quantification of aortic fibrotic areas. (C) Gene expression of markers of cardiac fibrosis. (D) Quantification of aortic hypertrophy. **P < 0.01, ***P < 0.001 by 2-way ANOVA Bonferroni post-tests.
There was marked increase in macrophage staining in aortas from treated FC mice but much less from MyMRKO mice (Figure 7, A and B). The increase was largely seen in the adventitia, but there was also some infiltration in the media and intima. The MyMRKO mice exhibited decreased expression of a marker for total (F4/80) (Figure 7C) and classically activated macrophages (Tnfa) and increased expression of a marker for alternative activation (Arg1) in MyMRKO compared with treated controls (Figure 7D).
MyMRKO reduces L-NAME/Ang II–induced aortic macrophage recruitment. (A) Representative immunofluorescence assay of cross sections of aortas from FC and MyMRKO mice with or without L-NAME/Ang II treatment. Mac2 antibody was used to detect macrophages. Scale bar: 100 μm. (B) Quantification of aortic macrophages stained by Mac2. The results are expressed as macrophage counts per high-power field (×400) under microscope. (C) Gene expression of macrophage marker F4/80. (D) Gene expression of M1 marker Tnfa and AMϕ marker Arg1. ***P < 0.001 by 2-way ANOVA Bonferroni post-tests.
These results demonstrate that MR directs macrophage polarization state by promoting classical macrophage activation and repressing alternative activation in a model of cardiac injury. MyMRKO causes a robust AMϕ phenotype and leads to a decrease in the trafficking of the macrophages into the tissue and a marked reduction in M1 cytokines. The AMϕ shift caused by MyMRKO correlates with protection in cardiac and vascular hypertrophy and inflammation.
Blood pressure response in MyMRKO mice. While the protection by eplerenone in this experimental model is reported to be independent of blood pressure lowering (25), hypertension is well established as a stimulus for cardiac and vascular hypertrophy and remodeling. It was therefore important to determine whether the MyMRKO mice had altered blood pressure. Since blood pressure has a diurnal rhythm and is affected by measurement conditions, we used continuous telemetry monitoring to assess unencumbered blood pressure throughout the day.
At baseline, the MyMRKO animals had no differences in blood pressure compared with FC animals (Figure 8). When the diet was changed to a high-salt diet, there was a small but significant increase in light-period blood pressure in the MyMRKO animals. Addition of L-NAME further increased the difference and particular led to a decrease in the magnitude of the circadian rhythm. These changes in circadian rhythm were most pronounced in the pulse pressure.
MyMRKO causes an increased blood pressure and heart rate response to L-NAME/Ang II. Twenty-four-hour averages of blood pressure and heart rate data collected by radiotelemetry under each condition of the model show a significant increase in daytime systolic pressure, pulse pressure, and heart rate in MyMRKO versus FC mice.
When Ang II was added, blood pressure differences between knockout and control mice were reduced, but pressures were still slightly higher in the MyMRKO mice. The most striking finding was that MyMRKO mice showed increased pulse pressure and a near complete loss of pulse pressure circadian variation.
Interaction of MR with PPARγ. The effects of MR antagonism and MRKO are similar to those of the thiazolidinediones (TZDs), PPARγ agonists, in their ability to mitigate cardiovascular inflammation and fibrosis through their action on macrophages (18, 27). TZDs also enhance some aspects of alternative activation in macrophages, suggesting a potential common mechanism (13, 28). In macrophages from myeloid PPARγ–knockout (MyPGKO) mice, the changes in expression of many of the genes associated with macrophage polarization were opposite those in MyMRKO macrophages, as assessed by QPCR (Figure 9A). Although MR antagonists have previously been shown to regulate PPARγ in some systems (6), MR deletion did not affect PPARγ expression. However, since effects were similar, we could not rule out alteration in PPARγ activity. PPARγ deletion did increase MR expression, raising the possibility that changes in MR expression contributed to the proinflammatory phenotype. We therefore chose several markers to determine whether PPARγ and MR acted upstream or downstream of each other or whether they altered macrophage polarization by parallel pathways. We tested the action of MR antagonists in PPARγ-null macrophages and the PPARγ agonist pioglitazone in MR-null macrophages (Figure 9, B and C).
MR opposes PPARγ in macrophage polarization. (A) Hierarchical clustering of M1 and AMϕ marker gene expression measured by real-time PCR from primary peritoneal macrophages demonstrates significant overlap between PPARγ activation and MyMRKO. (B) Deletion of MR enhanced the AMϕ-polarizing effects of 10 μM pioglitazone (Pio). (C) MyPGKO macrophages cultured in C/D medium showed increases in M1 markers that were enhanced by addition of 10 nM aldosterone. (D) MyMRKO and MyPGKO respectively enhance and oppose IL-4 stimulation of the AMϕ marker Ccl7. MRFC, FC for MyMRKO; PGFC, FC for MyPGKO; PGKO, MyPGKO; Cort, corticosterone. *P < 0.05, **P < 0.01, ***P < 0.001 by 2-way ANOVA Bonferroni post-tests.
PPARγ agonists were effective in MR-null cells. In fact, MRKO and TZD can be synergistic. Some genes such as Ym1 that showed a poor response to pioglitazone in control cells showed a markedly enhanced response in MRKO cells. Conversely, MR agonists were effective in PPARγ-null cells at inducing the M1 marker TNF-α and suppressing the AMϕ marker Arg1 (Figure 9C). Similarly, opposing actions on IL-4 induced expression of Ccl7 were seen (Figure 9D) These results show that they act by parallel but interacting pathways and that modification of multiple pathways may give enhanced responses.
Interaction of MR with GR. MRKO has an effect similar to that of GR agonists in that they both produce a phenotype resembling IL-4–induced M2 polarization (9). Therefore, we investigated whether MRKO could modulate the GR response, including induction of the AMϕ phenotype. Since a majority of MR’s evolution occurred prior to the existence of physiologic aldosterone, this cellular role is likely a conserved ancestral function, observable in cell types such as neurons, cardiomyocytes, and macrophages (29, 30). Transcriptional changes caused by MRKO overlapped with the glucocorticoid response (Figure 10A and Supplemental Figures 4 and 5). However, the majority of genes regulated by glucocorticoids were unaffected by MRKO, whereas about half of the MRKO-altered genes were also regulated by glucocorticoids. By cluster analysis, we separated the different responses to corticosterone in FC and MyMRKO macrophages according to their correspondence to genes where MRKO mimicked or enhanced the response to corticosterone (cluster A); genes where MRKO and corticosterone caused suppression of expression (cluster B); and genes where corticosterone opposed the inductions caused by MRKO (cluster C) (Figure 10B). Cluster A contained many AMϕ markers. The synergy between MRKO and corticosterone in inducing enhanced expression of AMϕ markers Ym1 and F13a1 was confirmed by QPCR (Figure 10C). Also in cluster A are the proteases Htra1 (a TGF-β inhibitor) and Serpine2, which showed marked synergy and is a relative of Pai1 but has not been linked to cardiovascular disease (Figure 10C). Cluster B included genes where MRKO and corticosterone synergized to repress the expression (Cdh1, IL27a) or where MRKO and corticosterone in FC cells suppressed the expression but addition of corticosterone to the MRKO cells did not result in further suppression (Il1b and Clec2, a proinflammatory prothrombotic C-type lectin) (Figure 10D). Cluster C had a number of different genes — including formyl-peptide receptor 2, involved in chemotaxis, and adrenomedullin (Adm), which has been linked to atherosclerosis protection — that are involved in recruitment of macrophages and may contribute to the protective effects of the knockout.
MR cooperates with glucocorticoid signaling in macrophages. (A) Affymetrix analysis of peritoneal macrophages treated with 1 μM corticosterone for 24 hours shows a minority of genes regulated by MR and GR activation. (B) MyMRKO can enhance the induction of AMϕ by corticosterone (cluster A), mimic the repression of proinflammatory factors by corticosterone (cluster B), or have the same effect as corticosterone but an not an additional effect with combination (cluster C). Representative genes are listed to the right of each cluster. (C) MyMRKO synergized with corticosterone to induce genes important in AMϕ macrophage polarization (Ym1 and F13a1) and enhance the TGF-β inhibitor Htra1 and thrombin inhibitor Serpine2. (D) MyMRKO synergized with corticosterone to repress TGF-β target E-cadherin (Cdh1) and IL-27 receptor. In addition, MyMRKO abolished additional repression of Il1b and the proinflammatory C-type lectin Clec2 by corticosterone. *P < 0.05, **P < 0.01, ***P < 0.001 by 2-way ANOVA Bonferroni post-tests.
Cluster analysis of M1- and AMϕ-related genes evaluated by QPCR for the 4 treatments clearly separated the two groups. It also showed that there were marked differences in the AMϕ-associated genes between the different treatments and separated the genes into those that are in agreement (which contained the most often used AMϕ markers) and genes where there are marked differences (Figure 11A). Consistent with the previously described changes, M1 markers were increased by PGKO and decreased by MRKO, corticosterone, and IL-4. Opposite effects were seen with the commonly used AMϕ marker Arg1. Other AMϕ markers, such as Ym1 and Fizz1, showed concordance between corticosterone, MRKO, and IL-4, although the magnitude of changes could be very different. Other genes showed marked differences and point out that the AMϕ are different depending on the treatment. Some marked differences were in the cadherins, where MRKO and corticosterone had effects opposite to IL-4. Therefore, there is a system of opposing and cooperating nuclear transcription factors that are capable of modifying the polarization phenotype of macrophages, thus modifying the response to injury and disease in the cardiovascular system. Further delineation of the phenotype similarities and differences of these subtypes may reveal the particular function of this spectrum of macrophage polarization (Figure 11B).
Nuclear receptor network regulating macrophage polarization. (A) Heatmap of hierarchical cluster analysis of gene expression altered by MyPGKO, IL-4, MyMRKO, and corticosterone treatment in macrophages. Nuclear receptor activation or inactivation can dramatically alter the expression patterns of genes associated with macrophage polarization. Classical activation markers (M1) are segregated from AMϕ-associated genes. There are two major subtypes of AMϕ gene patterns: one where the expression in the different treatments is concordant (AMϕ1); and one where there are a number of genes where there are substantial differences (AMϕ2). (B) Summary of macrophage activation and cardiovascular consequences.