Cytotoxic CD8+ T lymphocytes are recruited and activated in the ischemic heart tissue after MI
First, we characterized the kinetics of inflammatory cell recruitment within the injured myocardium. We used an experimental model of acute MI in C57BL/6J mice induced by permanent coronary artery ligation and we analyzed cell suspensions of digested hearts at different time points after the onset of ischemia by flow cytometry. We found that CD3+CD8+ T (Fig.1a-c) and CD3+CD4+ T lymphocytes accumulated in the injured myocardium as early as day 1 and peaked at day 3 after MI. Infiltration of CD4+ T cell was 2-fold higher than that of CD8+ T cells (Fig. 1d). Immunohistological analyses confirmed the increased accumulation of CD3+ CD8+ T lymphocytes in both peri-infarct and infarct areas after MI compared to sham-operated animals (Fig. 1c & Supplementary Fig. 1 & Supplementary Fig. 2). At day 1, infiltrating CD8+ T cells were mostly naive and the proportion of CD44+CCR7high central memory and CD44+CCR7low effector memory CD8+ T cells increased over time (Fig. 1e & Supplementary Fig. 3). Flow cytometry analyses showed that the proportion of CD8+ T cells expressing CD69 and CD107a increased during the first week after MI in the heart (Fig. 1f), as well as in draining mediastinal lymph nodes (Supplementary Fig. 4). Local activation and degranulation of recruited cytotoxic CD8+ T cells were confirmed by the detection of Granzyme B within the ischemic heart tissue at day 1 after MI at both protein (Fig. 1g) and mRNA levels (Fig. 1h). Immunofluorescence staining showed colocalization of Granzyme B and CD8+ T cells. Of note, Granzyme B was mainly detected in the cytoplasm of T cells in non-ischemic areas, whereas it was detected around CD8+ T cells in ischemic areas, suggesting active CD8+ T cell degranulation in the infarcted myocardium. (Fig. 1g & Supplementary Fig. 5).
Previous studies suggest that CD4+ T cells may orchestrate myeloid and lymphoid cell recruitment 15,16. We quantified T subsets at day 1 and Day 3 after MI in mice treated with anti-CD4 depleting monoclonal antibody or isotype control. At day 1, CD8+ T cell number was decreased in blood (Supplementary Fig. 6a), but increased in spleen after CD4+ T cell depletion (Supplementary Fig. 6b-c). At day 3, we observed that CD4+ T cell depletion (Fig. 1i) led to a 50 % reduction in infiltrating CD8+ T cells in heart tissue (P<0.01) (Fig. 1j-k), indicating that CD4+ T cells partly controlled CD8+ T cell mobilization from spleen into peripheral blood and ultimately their infiltration into the infarcted heart.
These findings indicate that circulating CD8+ T cells are recruited into the myocardium following MI, are activated and release Granzyme B, suggesting a potential role of CD8+ T cell-mediated immune response in this setting.
CD8+ T lymphocyte depletion prevents adverse ventricular remodeling and improves cardiac function after acute MI in mice.
To directly assess the role of CD8+ T lymphocytes in cardiac remodeling after acute MI, we depleted CD8+ T lymphocytes using a CD8-specific monoclonal antibody (CD8 mAb) 17 one hour after coronary ligation. CD8 mAb treatment rapidly depleted CD8+ T cells (>98%), within 6 hours after mAb injection (Supplementary Fig. 7) in the peripheral blood (Fig. 2a & Supplementary Fig. 8), in the cardiac tissue (Fig. 2b), and in the spleen (Supplementary Fig. 9). Full CD8+ T cell depletion was confirmed by immunofluorescence in ischemic heart tissue (Fig. 2c & Supplementary Fig.2). CD8+ T cell depletion did not significantly impact survival following MI (at day 21, 76% in control group versus 92% in CD8 depleted group, P=0.28, data not shown). We next assessed cardiac function by echocardiography 9 and 28 days after MI. CD8 mAb–induced T cell depletion led to smaller end-systolic (P<0.01) and end-diastolic left ventricular volumes (P<0.05) (Fig. 2d & Supplementary Fig. 10), and to a significant improvement of left ventricular fractional shortening (Fig. 2d & Supplementary Fig. 10) compared to isotype-treated control mice at both time points. At day 28, left ventricular pressures were measured using an intracardiac probe, confirming that CD8 depletion preserved both diastolic and systolic LV functions (Fig. 2e). Of interest, left ventricular myocardial contractility was also improved by CD8 T cell depletion (P <0.001) (Fig. 2f).
CD8 mAb–induced improvement in cardiac function was associated with abrogation of adverse LV remodeling. Infarct size (Fig. 2g & Supplementary Fig. 11) (P<0.001) and interstitial fibrosis (P <0.05) assessed by collagen content, were reduced in CD8 mAb treated mice compared to isotype-treated control animals, at day 21 post-MI (Fig. 2h). CD8 T cell depletion also decreased collagen synthesis as revealed by the reduction of Col1a1 and Col1a3 mRNA levels (Supplementary Fig. 12). Such protective effect of CD8 depletion was maintained at day 56 following MI (Supplementary Fig. 13). In summary, these results show that systemic CD8+ T cell depletion significantly reduces post-ischemic heart injury, prevents adverse ventricular remodeling and improves cardiac function after acute MI.
CD8+ T cells pathogenic activity requires TCR engagement
To assess the putative role of antigen recognition by CD8+ T cells, we used OT-I mice, in which the majority of CD8+ T cells exclusively recognize an irrelevant ovalbumin-derived peptide via their TCR. In a first set of experiments, coronary artery ligation was performed in male OT-I mice and one hour later, animals were injected either with an isotype control or an anti-CD8 depleting antibody (Fig. 3a). In this setting, CD8 T cell depletion (Fig. 3b) did not impact mortality (data not shown) and infarct size at day 21 post-MI (Fig. 3c & 3d). To further substantiate the role of TCR-mediated pathogenic activity of CD8+ T cells, we injected Rag1-/- mice with CD8+ T cell-depleted splenocytes, re-supplemented with wild-type or OT-I CD8+ T lymphocytes (Fig. 3e). Survival at day 21 was not statistically different between groups despite a trend toward a better survival in OT-I CD8+ T cells supplemented group (Fig. 3f). Animals re-supplemented with OT-I CD8+ T cells displayed less cardiac damage with a reduction in the infarct size (Fig. 3g) (P=0.054) and a better cardiac function (Fig. 3h) (P<0.01) than animals re-supplemented with WT CD8+ T cells.
Finally, we employed a third approach to address the importance of antigen-specific response of CD8+ T cell using CMy-mOva mice. CMy-mOva mice is a transgenic mouse line that expresses cardiac myocyte restricted membrane-bound ovalbumin 18 that can be recognized by OT-I CD8+ T cells. Three days before MI, CMy-mOva mice were injected either with wild-type or OT-I purified CD8+ T lymphocytes (Fig. 3i). The injection of OT-I CD8+ T lymphocytes enhanced Granzyme B mRNA content in the ischemic heart 2 days after MI when compared to control group (Fig. 3j). In addition, the injection of OT-I CD8+ T lymphocytes increased mortality rate (85% versus 40%, P<0.01) (Fig. 3k) and infarct size among rare survivors (Fig. 3l) when compared to animals receiving wild-type CD8+ T cells.
CD8+ T lymphocyte depletion reduces cardiomyocyte apoptosis and pro-inflammatory responses after acute MI
We next assessed the potential mechanisms involved in CD8+ T cell-mediated effects on cardiac remodeling and function. We first assessed the effect of CD8+ T cell deficiency on other actors of the inflammatory reaction. As shown in supplementary fig. 14-19, CD8+ T cell depletion had no impact on the number of CD4+ T cells, B cells, NK, NKT, classical monocytes, neutrophils, macrophages and dendritic cells into the ischemic heart tissue.
Cytotoxic activity of CD8+ T cells in the context of cancer 19 or virus infection 20 is mainly mediated by the release of Perforin/Granzyme B 21. Of note, Granzyme B colocalized with apoptotic cells in the ischemic heart tissue (Supplementary Fig. 20). CD8+ T cell depletion induced a significant reduction of Granzyme B content in the ischemic heart tissue at both mRNA (Fig. 4a) and protein levels (Fig. 4b). Such decrease in cardiac Granzyme B content was associated with a significant reduction of TUNEL+ apoptotic myocardial cells (Fig. 4c & Supplementary Fig. 21) and a reduction of infarct area 3 days after MI (Supplementary Fig. 22). Treatment with CD8 mAb also reduced local pro-inflammatory cytokine levels, at day 7 after MI. Of note, Tnf-a,Il-1b,Il-6 mRNA levels were significantly lower (P<0.05) in infarct hearts of CD8 depleted-mice compared to the control group (Fig. 4d). In addition, we found a marked decrease of Mmp9 gene expression (Fig. 4e) and a substantially lower metalloproteinase activity (Fig. 4f) in the heart of mice treated with anti-CD8 depleting antibody. Such alteration in the inflammatory landscape without any difference in the number of infiltrating leukocyte subsets suggests a mAb CD8-mediated immune phenotypic switch toward an anti-inflammatory profile. As such, cardiac macrophages displayed a reparative anti-inflammatory signature as revealed by the reduction of Il-1b, Tnf-a and iNOS mRNA levels in macrophages of anti-CD8 treated mice (Supplementary Fig. 23). On the same note, the number of reparative macrophage expressing CD206 was increased at day 7 in the heart of CD8 depleted animals (Supplementary Fig. 24).
Global Granzyme B deficiency limits cardiac damage after acute MI.
These findings prompted us to investigate the direct cytotoxic role of Granzyme B in post-ischemic cardiac remodeling. First, MI was induced in C57bl6 wild type and Granzyme B-deficient (GzmB-/-) adult mice. Granzyme B deficiency was confirmed by immunostaining in the spleen of GzmB-/- mice as well as in the heart (Supplementary fig. 25). Following acute MI, CD8+ T cell infiltration was observed in the ischemic heart of C57bl6 and GzmB-/- mice (Supplementary fig. 26). A significant reduction of TUNEL+ apoptotic cells was found within the injured myocardium (P<0.001) (Fig. 4g) as well as a local reduction of Il-1b, Il-6, Tnf-a and Mmp9 mRNA levels (P<0.01) in GzmB-/- mice compared to WT control group (Fig. 4h & Supplementary fig. 27). Finally, at day 21 following MI, the infarct size was markedly reduced in Granzyme B deficient animals (-55%, P<0.05) (Fig. 4i), and overall survival trended toward improvement (88% vs 64%, P=0.12) (Supplementary fig. 28). These experiments suggest that Granzyme B per se may have direct cytotoxic activity on cardiomyocytes. To test this hypothesis, mouse cardiomyocytes were co-cultured in vitro with purified wild-type or GzmB-/- splenic CD8+ T cells. After 24 hours, T cells were removed and cardiomyocyte apoptosis was monitored during an additional 24-hour period using caspase-3 fluorescent dye. CD8+ T cell activation was achieved using dynabeads mouse T activator CD3/CD28 (Supplementary fig. 29). Pre-incubation with activated CD8+ T cells did not increase cardiomyocyte apoptosis when compared to pre-incubation with non-activated CD8+ T cells at low concentrations (Cardiomyocyte/CD8 ratio 1/1 and 1/3) (Fig. 4j). However, at higher concentrations (ratio 1/5 and 1/10) activated CD8+ T cell strongly promote cardiomyocyte apoptosis measured as Caspase 3/7+ cells (Fig. 4j). Cytotoxicity of CD8+ T cells was abolished in case of Granzyme B deficiency (Fig. 4k). We also hypothesized that the effect of CD8 T lymphocytes expressing Granzyme B exceeds a simple cytotoxic action and could impair cardiomyocyte function. For this purpose, cardiomyocytes isolated from adult C57Bl/6J mice were co-cultured overnight with purified wild-type or GzmB-/- spleen CD8+ T cells at low concentration, i.e. 1/3 ratio. Cardiomyocyte contractility was evaluated using computer-assisted sarcomere shortening measurements. Interestingly, decreased sarcomere shortening was observed in cardiomyocytes co-cultured with activated wild type CD8+ T cells, compared with cardiomyocytes co-cultured with control wild type CD8+ T cells or activated GzmB-/- CD8+ T lymphocytes. (Fig. 4l). Altogether, our results suggest that low number of CD8+ T cells curb cardiomyocyte contractility, but that high number precipitates cardiomyocyte death.
Granzyme B-deficient CD8+ T lymphocytes fail to affect cardiac remodeling and function after acute MI
To further substantiate the role of CD8+ T cell-derived Granzyme B, we injected Rag1-/- mice either with CD8+ T cell-depleted splenocytes, CD8+ T-depleted splenocytes re-supplemented with wild-type or GzmB-/- CD8+ T lymphocytes (Fig. 5a). The purity of the CD8+ T lymphocytes is shown in Supplementary Fig. 30. We first verified that re-supplementation with wild-type or GzmB-/- CD8+ T lymphocytes significantly increased CD8+ T cell numbers in spleens and hearts of Rag1-/- mice compared to mice injected with CD8+ T cell-depleted splenocytes only (Supplementary Fig. 31).
We then examined the consequences of Granzyme B deficiency in CD8+ T lymphocytes on post-ischemic cardiac remodeling. Transfer of wild-type CD8+ T cells into Rag1-/- mice reduced survival (Fig. 5 a-b) and left ventricular shortening fraction (Fig. 5e) (p<0.05) after MI compared to the transfer of CD8-depleted splenocytes. In our re-supplementation experiment, we observed a significant negative correlation between wild-type CD8+ T cell number and cardiac function (Fig. 5f). This pathogenic effect on mortality and LV systolic function was abrogated after re-supplementation with GzmB-/- CD8+ T lymphocytes (Fig. 5b-e). CD8+ T cell supplementation also increased infarct size (P=0.04; Fig. 5c) and collagen content (Fig. 5d), which was prevented by re-supplementation with GzmB-/- CD8+ T lymphocytes (Fig. 5c-d).
CD8+ T lymphocyte depletion is protective in a pig model of coronary ischemia /reperfusion
To confirm the pathogenic role of CD8+ T cells in myocardial infarction and to substantiate the therapeutic interest of CD8-depleting antibody, we used a model of cardiac ischemia-reperfusion in pigs. To achieve CD8+ T depletion , we used an IgG2a mouse anti-swine monoclonal anti-CD8 antibody (clone 76-2-11) with known in vitro22 and in vivo activity against porcine CD8+ T cells 23,24. As shown in supplementary Fig. 32, pig anti-CD8 mAb treatment was efficient but induced delayed CD8+ T cell depletion when compared to mouse anti-CD8 mAb. Indeed, full CD8 depletion was obtained at day 3 after pig anti-CD8 mAb injection whereas complete CD8+ T cell depletion was obtained as early as 6 hours after mAb injection in mice (Supplementary Fig. 6). Based on this observation, we designed a protocol (Supplementary Fig. 33) including one control group and 2 CD8 depleted groups receiving anti-CD8 antibody at 2 different time points to obtain either High depletion (>95% depletion at day 1 after MI) or low depletion (60% depletion at day 1 after MI) (Fig. 6a-b). As previously validated by our group 25, coronary occlusion was maintained during 40 minutes (Supplementary Fig. 33) leading to transmural myocardial infarct. At day 14, we observed no difference in the infarct size between low CD8 depletion and control groups but we found a significant reduction of infarct size in high CD8 depletion group (-60% vs control, p<0.01) (Fig. 6c). Finally, both High and Low CD8 depletion improved significantly LV systolic function but the beneficial impact of CD8 depletion was more important in High depletion group (Fig. 6d & Supplementary Fig. 34), confirming the pathogenic role of CD8+ T cells in reperfused acute MI in a large animal model.
Granzyme B and CD8+ T cells in human MI
In human heart biopsies obtained from acute MI patients (Supplementary Table 1), we detected CD8+ T cell infiltration in the ischemic heart tissue at day 3 (Fig. 6e) and day 8 after MI (Fig. 6f). Granzyme B positive cells were mainly detected in the infarct area within the first week of MI, but predominated in the peri-infarct region after day 7 (Fig 6g & 6h).
Finally, we addressed the relevance of these findings to the human disease by assessing the relationship between circulating Granzyme B levels and clinical outcomes among those 1046 patients (Supplementary Table 2) who contributed to a serum bank in FAST-MI, a nationwide cohort of consecutive adults with ST-segment-elevation or non-ST-segment-elevation MI hospitalized at intensive unit care with symptom onset ≤48 hours, in 213 centers representing 76% of French centers managing acute MI patient (NCT01237418). Interestingly, we found that acute MI patients with high circulating levels of Granzyme B (>median 8.9 pg/mL) at their admission were at higher risk of death after one year of follow-up compared to patients with low levels even after adjustment for several multivariable risk factors (Supplementary Table 3) (adjusted hazard ratio, HR=2.26, 95% CI=1.22-4.18, p=0.009) (Fig. 6i).