Modulation of subsets of cardiac B lymphocytes improves cardiac function after acute injury

Pirfenidone reduces cardiac CD19⁺ B lymphocytes following DT-mediated acute myocardial injury.

Given that treatment with pirfenidone did not reduce cardiac myocyte necrosis or apoptosis, we asked whether pirfenidone improved survival by modulating the innate immune response to acute cardiac injury. Accordingly, we performed FACS analysis 4 days after DT injection. The gating strategy for this FACS analysis is shown in Supplemental Figure 1A (supplemental material available online with this article; https://doi.org/10.1172/jci.insight.120137DS1). In preliminary control studies, we determined that treatment with pirfenidone for 1 week in naive WT hearts had no significant effect on the number of CD45⁺ cells/mg tissue (P = 0.53), Ly6G⁺ neutrophils (P = 0.82), Ly6C⁺CD64^lo/– monocytes (P = 0.81), CD64⁺Ly6C^lo/– macrophages (P = 0.82), or CD19⁺ B lymphocytes (P = 0.94; Supplemental Figure 1, B and C).

As shown in Figure 2, there were no significant differences in the DT-injured hearts from mice treated with pirfenidone chow or normal chow with respect to the number of myocardial CD45⁺ cells (P = 0.8, Figure 2A), Ly6G⁺ neutrophils (P = 0.27, Figure 2B), Ly6C⁺CD64^lo/– monocytes (P = 0.15, Figure 2B), and Ly6C^lo/–CD64⁺ macrophages (P = 0.9, Figure 2B). The adult heart macrophage pool consists of resident and recruited cells, the latter of which have been associated with adverse LV remodeling following injury. These subpopulations are largely divided by the expression of CCR2 and MHC-II (13, 14). Therefore, we further characterized the macrophage populations in control and pirfenidone-treated animals. As shown in Figure 2, C and D there was no significant difference in the percentage of MHC-II^hiCCR2^lo (P = 0.43), MHC-II^hiCCR2^hi (P = 0.36), MHC-II^loCCR2^hi (P = 0.21), or MHC-II^loCCR2^lo (P = 0.11) macrophage subsets in the presence and absence of treatment with pirfenidone. Despite the lack of differences in cardiac myeloid populations after damage, we did observe that treatment with pirfenidone resulted in a greater than 3-fold reduction in the percentage of CD19⁺ myocardial B lymphocytes following DT-induced injury (P = 0.02, Figure 2B) when compared with mice that were fed normal chow.

Open in a separate window

Figure 2

Effect of pirfenidone on myocardial inflammation (day 4) after DT treatment.

Mice expressing the diphtheria toxin receptor (DTR) in the myocardium were exposed to diphtheria toxin (DT) and fed either chow enriched with pirfenidone (DTR-PFD) or regular chow (DTR-control). Mice were sacrificed at day 4 after DT injection and the heart was collected for analysis via flow cytometry. (A) Total number of CD45⁺ cells/mg heart tissue (n = 17 control, n = 19 pirfenidone). (B) Leukocyte subsets in the myocardium (percentage of total: CD19⁺, n = 14 control, n = 16 pirfenidone; Ly6g⁺, n = 6/group, Ly6C⁺CD64^lo/–, n = 10 control, n = 12 pirfenidone; CD64⁺Ly6C^lo/–, n = 10 control, n = 12 pirfenidone). (C) Representative FACS analysis of MHC-II and CCR-2 macrophages and monocytes. (D) Macrophage/monocyte subsets in the myocardium as defined by expression of CCR2, low (l) or high (h) and MHC-II expression, low (l) or high (h). Percentage of total, n = 10 control, n = 12 pirfenidone.*P < 0.05. Bars represent the mean, and error bars represent standard deviation. P values were calculated with Student’s t test.

Pirfenidone reduces adverse LV remodeling and cardiac CD19⁺ B lymphocytes after closed-chest ischemia-reperfusion injury.

To determine the effects of pirfenidone in a more pathophysiologically relevant model of cardiac injury, we subjected control mice and pirfenidone-treated mice to closed-chest ischemia-reperfusion (I/R) injury (90 minutes of ischemia), as previously described (15). As shown in Figure 3A, there was no difference (P = 0.68) in the area at risk in control and pirfenidone-treated animals, as determined by the segmental wall motion score index (SWMSI). Treatment with pirfenidone resulted in a significant decrease in the heart weight–to–tibia length ratio (Figure 3, B and C) and LV mass measured by 2-D echocardiography (P = 0.03, Figure 3D) when compared with control mice. Consistent with these findings, we observed a significant (P = 0.02) decrease in LV end-diastolic volumes (Figure 3E) in pirfenidone-treated animals when compared with control animals, whereas there was a trend (P = 0.08) towards improvement in LV ejection fraction (LVEF) in the treatment group (Figure 3F). Figure 3G shows representative Masson’s trichrome staining in control and pirfenidone-treated animals, whereas Figure 3H summarizes the results of the group data. As shown in Figure 3H, there was no significant difference (P = 0.4) in the percentage of myocardium exhibiting collagen staining with Masson’s at 2 weeks, when the pirfenidone chow mice were compared to the mice fed normal chow.

Open in a separate window

Figure 3

Effect of pirfenidone on LV structure and function after closed-chest I/R injury.

Wild-type mice were subjected to 90 minutes of closed-chest ischemia followed by 2 weeks of reperfusion (I/R injury). Mice were fed either chow enriched with pirfenidone or regular chow. (A) Area at risk during closed-chest ischemia as determined by the simplified segmental wall motion score index (SWMSI) at time of ischemia. (B) Representative pictures of hearts harvested from control mice (left) and pirfenidone-treated animals (right) 2 weeks after I/R injury. Scale bar: 1 mm. (C) Gravimetric analysis of hearts harvested from control mice (I/R-control) and pirfenidone-treated animals (I/R-PFD). n = 8 control, n = 7 pirfenidone. (D–F) Echocardiographic assessment of myocardial function at the time of ischemia and 2 weeks after I/R injury. n = 8 I/R-control, n = 7 I/R-PFD. (D) Left ventricular (LV) mass (LVM) by 2-D echocardiography. (E) LV end-diastolic volume (LVEDV). (F) LV ejection fraction (LVEF). (G) Representative trichrome staining of histological sections of hearts from control animals (left panel) and pirfenidone-treated animals (right panel). Original magnification, ×1.25. (H) Quantitative assessment of the percentage trichrome-positive staining, 2 sections analyzed per heart. n = 16 sections I/R-control, n = 14 sections I/R-PFD. *P < 0.05. Bars represent the mean, and error bars represent standard deviation. P values were calculated with Student’s t test.

As the preceding experiments examined the effect of pretreatment with pirfenidone prior to I/R injury, we sought to determine whether the acute administration of pirfenidone was sufficient to influence LV remodeling after I/R injury. We therefore treated mice with intravenous pirfenidone or diluent soon after reperfusion injury, followed by i.p. injections for 1 day and oral pirfenidone in the chow for 14 days. Supplemental Figure 2 shows that the area at risk was not different (P = 0.47) in the control and pirfenidone-treated animals. However, the salient finding shown by Supplemental Figure 2 is that acute administration of pirfenidone significantly (P = 0.015) reduced the heart weight/tibia length ratio (P = 0.015, Supplemental Figure 2B), LV mass by 2-D echocardiography (P = 0.029, Supplemental Figure 2D), and LV end-diastolic volume (P = 0.037, Supplemental Figure 2E), when compared with diluent-treated control mice.

To determine whether the mechanism(s) for the beneficial effects of pirfenidone in the closed-chest I/R injury model was dependent on modulation of the innate immune response to acute cardiac injury, we harvested hearts on day 4 after I/R injury, and analyzed the inflammatory infiltrate via FACS. Consistent with the results observed with the DT-induced injury model, treatment with pirfenidone did not significantly (P = 0.98, Figure 4A) alter the total number of CD45⁺ cells per mg heart, Ly6G⁺ neutrophils (P = 0.44, Figure 4B), Ly6C⁺CD64^lo/– monocytes (P = 0.74, Figure 4B), or CD64⁺Ly6C^lo/– macrophages (P = 0.91, Figure 4B). Moreover, treatment with pirfenidone did not alter the relative prevalence of macrophage subpopulations as defined by levels of expression of CCR2 and MHC-II (Figure 4, C and D). However, we again noticed that treatment with pirfenidone significantly (P = 0.004) reduced the percentage of CD19⁺ B lymphocytes (Figure 4B) in the heart following injury.

Open in a separate window

Figure 4

Effect of pirfenidone on myocardial inflammation (day 4) after I/R injury.

Wild-type mice were subjected to 90 minutes closed-chest ischemia followed by reperfusion (I/R injury). Mice were fed either chow enriched with pirfenidone (PFD) or regular chow. Mice were sacrificed at day 4 after I/R injury and the heart was collected for analysis via flow cytometry. n = 8 I/R-control, n = 4 I/R-PFD. (A) Total number of CD45⁺ cells/mg heart tissue. (B) Leukocyte subsets in the myocardium (percentage of total: CD19⁺, Ly6g⁺, Ly6C⁺CD64^lo/–, CD64⁺Ly6C^lo; n = 4/group). (C) Representative FACS analysis of MHC-II and CCR-2 macrophages and monocytes. (D) Macrophage/monocyte subsets in the myocardium as defined by expression of CCR 2, low (l) or high (h) and MHC-II expression, low (l) or high (h). Percentage of total, n = 4/group.***P < 0.001. Bars represent the mean, and error bars represent standard deviation. P values were calculated with Student’s t test.

Pirfenidone modulates subsets of cardiac B lymphocytes following cardiac injury.

To further interrogate the effects of pirfenidone on cardiac B lymphocytes, we first analyzed the surface antigens of cardiac B cells using known markers of B lymphocytic subtypes. The gating strategy that was used is illustrated in Supplemental Figure 3 and staining of splenic B lymphocytes is provided for reference in Supplemental Figure 4. As shown in Figure 5, the majority of CD19⁺ cardiac B cells in the uninjured adult heart were CD19⁺CD11b^–. Further characterization of this population revealed that they were IgD⁺ and IgM^lo/–CD21^–CD23^–CD11c^– (Supplemental Figure 3). Approximately 10% of the resident cardiac B cells in uninjured hearts were CD19⁺CD11b⁺. This population of cells was predominately CD5^–IgM⁺, which is characteristic of B1b lymphocytes, whereas a smaller percentage was CD5⁺IgM⁺, which is characteristic of B1a lymphocytes (16) (Figure 5A).

Open in a separate window

Figure 5

Characterization of subsets of myocardial B lymphocytes at baseline and after DT-induced injury and I/R injury.

(A) Analysis of subsets of myocardial CD19⁺ B lymphocytes in naive hearts (n = 4). (B) Mice expressing the diphtheria toxin receptor (DTR) in the myocardium were exposed to diphtheria toxin (DT) and fed either regular chow (control, gray bars) or chow enriched with pirfenidone (PFD, white bars). Mice were sacrificed at day 4 after DT injection and the heart was collected for analysis of myocardial CD19⁺ B lymphocytes via flow cytometry (n = 4 control, n = 3 pirfenidone). (C) Wild-type mice were subjected to 90 minutes closed-chest ischemia followed by reperfusion (I/R injury). Mice were fed either regular chow (control, gray bars) or chow enriched with pirfenidone (PFD, white bars). Mice were sacrificed at day 4 after I/R injury and the heart was collected for analysis of myocardial CD19⁺ B lymphocytes via flow cytometry (n = 8 controls, n = 5 pirfenidone). *P < 0.05, **P < 0.01, ***P < 0.001 versus control; ^†P < 0.001 versus naive hearts. Bars represent the mean, and error bars represent standard deviation. P values were calculated with 2-way ANOVA followed by Tukey’s test for multiple comparisons.

We next analyzed B cell profiles in the presence or absence of pirfenidone in the DT-induced injury model and the closed-chest I/R injury model. As shown in Figure 5, in both models of cardiac injury, the relative ratio of CD19⁺CD11b^– cells to B1 cells on day 4 after cardiac injury remained unchanged relative to naive hearts. In the DT-injured hearts (day 4) there was a small but nonsignificant (Figure 5B) increase in the number of CD19⁺ cells/mg of cardiac tissue (P = 0.09) and CD19⁺CD11b⁺ cells/mg of cardiac tissue (P = 0.14), but no change in the number of CD19⁺CD11b^– cells/mg of cardiac tissue (P = 0.96) when compared with naive hearts. In contrast, there was a significant 2-fold increase in CD19⁺ (P < 0.001) and CD19⁺CD11b^– (P < 0.001) cells at the same time point following I/R injury (Figure 5C). In the DT-induced injury cardiac myocyte ablation model, the pirfenidone-treated animals had significantly fewer (P < 0.01) CD19⁺ cells/mg of tissue, with a significant (P = 0.04) reduction in the subset of CD19⁺CD11b^– cells (Figure 5B). We observed very similar changes in the closed-chest I/R injury model, in which pirfenidone treatment resulted in a significant (P < 0.001) reduction in CD19⁺ lymphocytes on day 4 after I/R, as well as a significant (P < 0.001) reduction in the subset of CD19⁺CD11b^– B lymphocytes (Figure 5C).

The cardioprotective effect of pirfenidone is mediated by B lymphocytes.

Viewed together, the above results suggest that the cardioprotective effects of pirfenidone may be mediated, at least in part, by cardiac B lymphocytes. To test this hypothesis, we depleted native B cells using a commercially available anti-CD20 antibody, and then repeated the closed-chest I/R studies in the presence and absence of pirfenidone. Given that the mice were instrumented 7 days prior to I/R injury in the closed-chest model, we injected the mice with anti-CD20 antibody at the time of instrumentation. The efficiency of anti-CD20 antibody B cell depletion was assessed 8 days after anti-CD20 injection (i.e., at the time that I/R injury is performed) and on day 12 after injection (i.e., day 4 after I/R). As shown in Supplemental Figure 5, A and B, a single injection of anti-CD20 antibody ablated over 99% of splenic B cells and cardiac B cells by day 8. We noticed that the number of cardiac B cells began to recover on day 4 after I/R injury (i.e., day 12 after injection). However, the total number of B cells remained less than 10% that of untreated controls (Supplemental Figure 5C). Notably, while B cell depletion after I/R injury significantly reduced (P < 0.01) the number of CD19⁺CD11b^– cells, it did not significantly (P = 0.2) reduce the total number of CD19⁺CD11b⁺ cells (Supplemental Figure 5C). However, when we examined CD19⁺CD11b⁺ subsets, treatment with the anti-CD20 antibody did significantly (P = 0.05) reduce the number of B1b lymphocytes, whereas it had no effect on the number of cardiac B1a lymphocytes (P = 0.58, Supplemental Figure 5C). There was no significant difference in the heart weight–to–tibia length ratio (P = 0.75, Figure 6, B and C), LV mass (P = 0.21, Figure 6D), LV end-diastolic volume (P = 0.40, Figure 6E), LVEF (P = 0.15, Figure 6F), or percentage of trichrome-positive myocardium (P = 0.43, Figure 6, G and H) between the B cell–depleted mice and B cell–depleted + pirfenidone-treated mice 2 weeks after I/R injury. Thus, the cardioprotective effects of pirfenidone following I/R injury were B cell dependent.

Open in a separate window

Figure 6

Effect of B cell depletion on pirfenidone cardioprotective effect after I/R injury.

Wild-type mice were B cell depleted via injection of anti-CD20 antibody. Seven days after injection of anti-CD20 antibody, B cell–depleted mice were subjected to 90 minutes of closed-chest ischemia followed by reperfusion (I/R injury). Mice were fed either chow enriched with pirfenidone (anti-CD20 PFD) or regular chow (anti-CD20 control). (A) Area at risk during closed chest-ischemia as determined by the simplified segmental wall motion score index (SWMSI) at time of ischemia. (B) Representative pictures of hearts harvested from anti-CD20–treated mice (left) and anti-CD20 + pirfenidone–treated animals (right). Scale bar: 1 mm. (C) Gravimetric analysis of hearts harvested from pirfenidone-treated animals (anti-CD20 PFD) and untreated controls (anti-CD20 control), n = 7/group. (D–F) Echocardiographic assessment of myocardial function at the time of ischemia and 2 weeks after I/R injury, n = 7/group. Data from non–anti-CD20–treated animals already reported in Figure 3, D–F are shown again for comparison only (control). (D) Left ventricular (LV) mass (LVM) by 2-D echocardiography. (E) LV end-diastolic volume (LVEDV). (F) LV ejection fraction (LVEF). (G) Representative trichrome staining of histological sections of hearts from control animals (left panel) and pirfenidone-treated animals (right panel). Original magnification, ×1.25. (H) Quantitative assessment of the percentage of trichrome-positive staining, 2 sections analyzed per heart, n = 10 sections/group. Bars represent the mean, and error bars represent standard deviation. P values were calculated with Student’s t test.

Myocardial B lymphocytes in health and disease.

Recent studies have shown that several discrete populations of leukocytes reside in the normal adult murine myocardium. Antigen-presenting mononuclear cells (CD11b⁺CD11c⁺F4/80⁺ and MHCII⁺) represent the most prominent population, followed by B cells, monocytes, and T cells (20). Although there has been tremendous recent interest in the role of macrophages, dendritic cells, and T cells in the heart, comparatively less is known about the role of B cells. Very little is known about the composition of myocardial B cells in naive hearts. Ramos et al. have shown that the naive adult murine myocardium harbors 2 populations of B220⁺ (CD45R) lymphocytes, a larger population of IgM^hiIgD^lo cells, and a smaller population of IgM^loIgD^hi cells (21); however, neither the ontology nor functional role of these 2 populations of lymphocytes was characterized in this study. The existing literature suggests that B lymphocytes play an important role in chronic LV remodeling following cardiac injury, although the specific B cell–mediated mechanisms have not been identified completely. Three different groups have studied the role of myocardial B lymphocytes following acute coronary ligation (MI). Both Yan et al. (22) and Zouggari et al. (23) reported that the number of myocardial CD19⁺ cells increased after MI. Zouggari showed that CD19⁺IgD⁺IgM^lo B lymphocytes influx into the infarcted myocardium and contribute to adverse LV remodeling by recruiting Ly6C⁺ monocytes from the bone marrow through a CCL7-dependent mechanism (23). In contrast, a recent study found that CD19⁺ cells are recruited preferentially to fat-associated lymphoid clusters in pericardial adipose tissue, with minimal detection of CD19⁺ cells in the myocardium after MI (24). Using a nonsurgical model of cardiac injury, Cordero-Reyes et al. showed that adverse cardiac remodeling was attenuated by antibody-dependent depletion of B cells in WT mice (25). Although the mechanisms of action of B lymphocyte–mediated myocardial damage have not been fully elucidated, there are several mechanisms that have been proposed, including dysregulation of B cell subpopulations (B1 vs. B2 vs. regulatory B cells), deposition of IgM or IgG antibodies, as well as B cell–mediated recruitment of Ly6C⁺ monocytes (23). In this regard, it is interesting to note that Zhang et al. showed that natural IgMs play a role in myocardial I/R injury, although they did not specifically study myocardial B lymphocytes (26).

The results of the present study both confirm and expand on prior studies that have described myocardial B lymphocytes and suggested that B lymphocytes play an important role in modulating adverse cardiac remodeling following cardiac injury. Here we show for the first time to our knowledge that in naive murine hearts the majority of B lymphocytes are CD19⁺CD11b^–CD21^–CD23^–CD11c^–IgD⁺IgM^lo/–, whereas a minority of myocardial B lymphocytes are CD19⁺CD11b⁺. In this latter group, a smaller fraction express CD5, which is characteristic of B1a lymphocytes, and a larger fraction is CD5^–, which is characteristic of B1b lymphocytes (16). Although the phenotype of the CD19⁺CD11b^– cells that we describe overlaps the phenotypic description of the B cells that accumulated in the myocardium after infarction in the report by Zouggari et al. (i.e., CD19⁺IgD⁺IgM^lo), it bears emphasis that the precise ontogeny of the CD19⁺CD11b^–CD21^–CD23^–CD11c^–IgD⁺IgM^lo/– cells described herein is not known. To our knowledge, the phenotype of these cells does not match the profile of any previously described B cell subset (27). B cells can be broadly characterized into B1 and B2 lymphocytes, which are classically considered innate and adaptive, respectively. Whereas B1a lymphocyte development occurs primarily during fetal and perinatal life, B1b and B2 lymphocyte production continues throughout adult life. The observation that these cells express relatively high levels of IgD, which is characteristic of B2 cells, and low levels of IgM, which are characteristically expressed by B1 cells, suggests that these cells are not B1 cells (23). Of note, several authors have reported that a distinct population of mature CD19⁺CD21^–CD23^– B cells accumulate in aging mice (28). However, these age-associated B cells are CD11c⁺ and express the transcription factor T-bet, whereas we did not detect CD11c expression via FACS, nor gene expression of T-bet by RNA sequencing (data not shown). Our results further expand on the existing literature on myocardial B cells also by showing that unique subsets of B lymphocytes increase in the heart following I/R injury (Figure 5), and that different subsets of myocardial B lymphocytes (CD19⁺CD11b⁺ and CD19⁺CD11b^– lymphocytes) are activated differentially following cardiac injury, in a process that is modulated by pirfenidone (Supplemental Figure 6). Although we did not observe a statistically significant increase in CD19⁺ lymphocytes in the heart after DT-mediated cardiac injury, the overall trends were similar to those observed with I/R injury. The quantitative differences in B cell response may be secondary to differences in the degree of tissue injury and/or differences in B lymphocyte kinetics in the 2 different injury models.

Our findings suggest that B cells are required for the cardioprotective effects of pirfenidone after acute tissue injury and that B cell depletion does not mimic the cardioprotective effects of pirfenidone following tissue injury in vivo. Although these results are seemingly incongruent at first blush, they suggest a more nuanced model supported by our transcriptional profiling studies in vivo (Supplemental Figure 6), wherein pirfenidone exerts its beneficial effects by modulating the immune response of B cells to tissue injury. Based on the results of our informatics analysis we hypothesized that activation of TLR signaling in response to the release of the cytosolic contents from necrotic cells (i.e., DAMPS) might activate B cells through a pirfenidone-sensitive process. We found that NCEs were sufficient to upregulate the costimulatory molecule CD86 in B lymphocytes isolated from cultures of mixed inflammatory cells, whereas NCEs were not sufficient to upregulate CD86 in purified B cell preparations. In contrast, LPS, a classic TLR4 agonist, induced upregulation of CD86 in CD19⁺ cells in PDICs as well as in purified B cell preparations. These findings are consistent with prior reports showing that the B lymphocyte response to TLR agonists is context dependent (29), and suggest that B lymphocyte activation in response to NCE is, at least in part, non–cell autonomous. Importantly, we observed that pirfenidone significantly attenuated the number CD19⁺CD86⁺ lymphocytes in PDIC cultures stimulated with NCEs or LPS. Consistent with our in vitro studies, we observed that there was a significant decrease in CD19⁺CD86^hi lymphocytes in the hearts of DT-injured and I/R-injured mice treated with pirfenidone when compared with untreated controls (Figure 7, E and F). Our in vitro results with TIRAP^–/– cells (Figure 7) suggest that activation of B cells by NCEs is TIRAP dependent. However, LPS-induced activation of B cells in vitro was attenuated but not abrogated in TIRAP^–/– mice. This is consistent with prior observations that LPS signaling can activate multiple intracellular pathways in B cells (30).

Our work has some important limitations. First, we did not specifically compare the effect of pirfenidone and anti-CD20 treatment. Second, we did not measure the effect of pirfenidone treatment in IgG-treated animals. In addition, we did not determine which actions of B cells accentuate adverse remodeling following acute myocardial injury. Further work will be needed to detail the cell-autonomous versus non–cell-autonomous role of TIRAP during B cell activation in the context of the acute inflammatory response to sterile injury, to understand if pirfenidone-mediated modulation of B cells is dependent on the modulation of other signaling pathways, including TGF-β signaling, and to define the mechanistic basis of B cell contribution to adverse cardiac remodeling.

Although the scope of these studies was not intended to define the molecular mechanisms of pirfenidone, our findings are consistent with prior reports that have shown that pirfenidone inhibits TLR signaling through the MAPK signaling cascade (31), as well as studies that show that pirfenidone inhibits MAPK signaling in the heart (10, 12). Importantly, this is the first study to our knowledge that demonstrates a direct cellular target that mediates the beneficial effects of pirfenidone in vivo.

In our work we found that pirfenidone-dependent immunomodulation of the cardiac inflammatory infiltrate improved cardiac function after acute injury. Surprisingly, treatment with pirfenidone was not associated with a reduction in troponin on day 4 after administration of DT (Figure 1B) nor a reduction in infarct size after I/R injury, as assessed by trichrome staining (Figure 3, G and H). This finding suggests that cardiac B lymphocytes might modulate cardiac remodeling after injury. Further work will be needed to understand the significance of this intriguing observation.

Pirfenidone treatment

In order to administer pirfenidone, mice were fed powdered chow mixed with pirfenidone (eNovation Chemicals, catalog D404655) at 0.5% by weight or powdered chow alone. This dosage has been commonly used in rodent models and has been suggested to result in plasma concentrations similar to those observed in humans treated with pirfenidone (36). Unless specifically noted, mice were switched to a diet enriched with pirfenidone 3 days prior to DT-induced injury or I/R injury, and maintained on this diet throughout the duration of the experiments. In parallel studies, we fed mice normal chow prior to I/R injury and then administered pirfenidone acutely at the time of cardiac injury. For these latter studies, mice were injected i.p. with diluent (200 μl PBS) or i.p. with pirfenidone (200 μl of pirfenidone, 5 mg/ml in PBS) after reperfusion and the morning after I/R injury (2 injections total). They were switched to pirfenidone-enriched diet or control diet the evening of I/R injury and maintained on the same diet until completion of the experiment.

B cell depletion

For the B cell depletion studies, mice were injected with 100 μg of anti-CD20 antibody (Biolegend, clone SA271G2, catalog 152104) through the jugular vein. The injected antibody was diluted in PBS, pH 7.2, containing no preservative (endotoxin level < 0.01 U/μg of protein) and filtered using a 0.2-μm filter. Immunodepletion was assessed at different time points, as described in the results section.

Gravimetric and histological analysis

Rosa26-DT^Mlc2v-Cre and littermate control WT mice were euthanized 14 days after DT injection, and the hearts were removed and weighed to determine the heart weight/tibia length ratio. WT mice were euthanized 14 days after I/R injury and the hearts were removed and weighed to determine the heart weight/tibia length ratio. Hearts were processed, paraffin-embedded, and stained with hematoxylin and eosin and Masson’s trichrome, as described previously (37). TUNEL staining was performed with the Millipore S7200 ApopTag Peroxidase In Situ Oligo Ligation Apoptosis Detection Kit according to the manufacturer’s instructions.

Echocardiographic studies

Cell culture

PDICs were harvested from the peritoneal cavity of thioglycolate-stimulated WT and TIRAP^–/– mice. PDIC cultures contain an admixture of inflammatory cells, including CD19⁺ B cells (39). TIRAP-deficient mice were a gift from Ruslan Medzhitov (Yale University, New Haven, Connecticut, USA). PDICs were harvested on day 4 after i.p. injection of 1 ml of thioglycolate medium (Millipore). Mice were euthanized, and the peritoneal cavity was washed 3–4 times with 3 ml of DMEM containing 10% FCS. A 3-ml syringe with an 18-gauge needle was used to inject the peritoneal cavity and retrieve the medium containing PDICs. Harvested cells were centrifuged at 250 g for 5 minutes, resuspended in 5 ml of media, and then filtered through a 40-μm strainer. Cells were plated at approximately 1.5 million cells/ml in 12- or 24-well plates. Primary spleen-derived B lymphocytes were isolated with the MagniSort Mouse B cell Enrichment Kit (Invitrogen) according to manufacturer’s instructions and plated in 96-well plates at 1 million cells/ml. The purity of isolated B cells was checked by flow cytometry and was greater than 85% in all reported experiments. Both PDICs and primary B lymphocytes were cultured in DMEM with 10% FCS, 1 mM sodium pyruvate, penicillin/ streptomycin (GIBCO, 1×), 2 mM L-glutamine, 10 mM Hepes, and 55 μM 2-mercaptoethanol.

We used LPS-EB ultrapure LPS (InvivoGen) or NCEs to stimulate the cell cultures. For stimulation with LPS, LPS was added at 100 ng/ml at the time of plating. NCEs were prepared from H9c2 cells, in a similar manner to the protocol described for mouse heart and mouse liver extracts (18). For stimulation, NCEs were added at a final concentration of 10 μg/ml at the time the cells were plated. LPS and NCE stimulation was performed in the presence and absence of pirfenidone for 24 hours. Pirfenidone was solubilized (10 mg/ml) in DMEM by 2 rounds of gentle heating in a water bath at 65°C for 5 minutes, followed by vortexing for 20 seconds. The pirfenidone solution was stored at 4°C and used within 24 hours of preparation. Pirfenidone was added at a final concentration of 150 ng/ml. For FACS analysis, both adherent and nonadherent cultured cells were collected. Nonadherent cells were collected by pipetting up and down in the culture media prior to media collection and a single rapid wash with Cell Stripper (Corning), and then gently centrifuged at 250 g for 3 minutes. To collect the adherent fraction, cells were incubated for 30 minutes at 37°C with Cell Stripper and then detached by gentle pipetting followed by mechanical detachment with a cell lifter. The adherent cells were mixed with the nonadherent cells, spun down at 250 g for 3 minutes, and resuspended in 300 μl of FACS buffer for further processing.

Flow cytometry

For flow cytometry experiments, mice were euthanized in a CO₂ chamber and the hearts were perfused with cold PBS, carefully dissected from extracardiac tissue under a stereo microscope, finely minced, suspended in 3 ml of DMEM, and then digested with 120 U of DNAse (Sigma-Aldrich), 180 U of hyaluronidase (Sigma-Aldrich), and 1,350 U of collagenase (Sigma-Aldrich) for 60 minutes at 37°C. The digested material was filtered through 40-μm filters and pelleted by centrifugation (250 g for 3 minutes at 4°C) in HBSS supplemented with 2% FCS plus 0.2% bovine serum albumin (BSA). Red blood cells were lysed in ACK lysis buffer (Invitrogen) for 15 minutes on ice and the remaining cells were resuspended in 300 μl of FACS buffer (PBS with 2% FCS and 2 mM EDTA). Cultured cells were collected for FACS analysis as described above and resuspended in 300 μl of FACS buffer. Samples were labeled with fluorescently conjugated antibodies. A list of all antibodies used including brand, clone number, and fluorophores used are provided in Supplemental Table 3. Cells were stained for 45 minutes on ice, and washed in FACS buffer prior to analysis. All antibodies from BioLegend were used at 0.2 μl per 300 μl sample, while antibodies from BD Biosciences were used at 0.4 μl per 300 μl sample. FACS was performed on Becton Dickinson analyzers (LSRII, Canto, X20, or Fortessa). Compensation controls were generated using UltraComp eBeads (Invitrogen) and verified on single-color control samples obtained by staining primary splenocytes. Gating strategies are summarized in Supplemental Figures 1, 3, and 4. For flow cytometry experiments that involved the DT injury model, we collected serum via retro-orbital puncture immediately prior to terminal sacrifice for quantification of troponin. In pilot studies we observed that mice with a serum troponin below 1 ng/ml did not have appreciable tissue injury; therefore, mice with serum troponin below 1 ng/ml were excluded from further analysis. Cell sorting was performed using a FACSAria instrument (Becton Dickinson) at the Washington University Department of Pathology Flow Cytometry and Sorting Core.

Transcriptional profiling

We performed transcriptional profiling of cardiac lymphocytes that were isolated from naive and DTR-injured hearts in the presence and absence of pirfenidone. We FACS sorted CD19⁺CD11b⁺ and CD19⁺CD11b^– cardiac lymphocytes from naive, DT-injured, and DT + pirfenidone hearts directly into lysis buffer for RNA isolation. A total of 600 to 12,000 cells were sorted from each heart. RNA was extracted using the QuickRNA MicroPrep (Zymo Research), according to the manufacturer’s instructions. RNA was eluted in 11 μl and the entire volume was submitted for whole-transcriptome analysis. Myocardial B lymphocytes were sorted from 8 naive animals, 6 DT mice, and 3 DT + pirfenidone-treated mice. Total RNA obtained from sorted B cell populations was selected for polyadenylated RNA and converted to RNA-sequencing libraries using the SMARTer v2 kit from Clontech. Single-end, 50-bp reads were obtained on an Illumina HiSeq 3000 and aligned to the Illumina iGenomes GRCm38_Ensembl release of the mouse transcriptome using Tophat2.1 (40), yielding an average of 1.8 × 10⁷ aligned reads per sample. Gene-level quantification was performed using HTSeq 2 (41). mRNAs were included in downstream analyses if present in all biological replicates in any of the 6 experimental groups (CD19⁺CD11b⁺ and CD19⁺CD11b^– from WT, DT, and DT + pirfenidone) at an abundance of at least 1 read per million. A total of 12,097 mRNAs were obtained from this filtering procedure. In order to compare gene expression respectively in CD19⁺CD11b⁺ cells and CD19⁺CD11b^– cells across untreated, DT-injured, and DT-injured + pirfenidone conditions, gene expression data were corrected for variability among experimental batches and a principal components analysis across all 12,097 mRNAs for the 17 sequenced samples was performed. This highlighted 5 samples (1 CD19⁺CD11b⁺ DT heart, 2 CD19⁺CD11b^– naive hearts, and 1 CD19⁺CD11b^– DT heart) that were strong outliers being outside the 95% confidence interval (2 standard deviations) for their group and that were removed from further analysis. The data are available in the NCBI’s Gene Expression Omnibus repository (GEO {"type":"entrez-geo","attrs":{"text":"GSE112984","term_id":"112984"}}GSE112984). The limma-voom procedure was used to identify mRNAs with differential expression between experimental conditions (42). KEGG pathway analysis was performed on mRNAs with greater than 2-fold change in expression level between specific experimental groups using the appropriate functional annotation module within the 2017 release of the NIH online resource DAVID (43) and accepting an FDR ≤ 0.05 (Benjamini-Hochberg method) to identify pathways enriched between the assigned conditions. KEGG pathways within the following 3 categories were analyzed: signal transduction (3.2), signaling molecules and interactions (3.3), and immune system (5.1).

Statistics

Data are expressed as mean ± standard deviation. Two-tailed Student’s t test was used for pairwise comparisons between 2 groups, 1-way ANOVA with Tukey’s correction for multiple post hoc comparisons was used to compare multiple experimental groups, while 2-way ANOVA with Tukey’s correction for multiple post hoc comparisons was used to compare across multiple experimental groups with multiple conditions in each group. The Gehan-Breslow-Wilcoxon method was used to compare Kaplan-Meier curves. A P value less than 0.05 was considered statistically significant. The specific statistical test used in each experiment is indicated in the figure captions. All calculations were made using GraphPad Prism version 7.04.

This research was partially funded through NIH grant R01HL107594-06. We thank Erica Lantelme and the Washington University Flow Cytometry & Fluorescence Activated Cell Sorting Core for assistance provided with flow cytometric analysis and cell sorting. Luigi Adamo was supported by grant T32 HL007081 from the NIH.