Main

In the adult mouse myocardium, cardiac fibroblasts constitute a large fraction of noncardiomyocytes (<20%)1, and are largely responsible for the synthesis and degradation of the extracellular matrix components to maintain tissue homeostasis2. Under pathological conditions, including pressure overload, infection or ischemia, increased fibroblast proliferation and activation is associated with fibrosis2. Activated fibroblasts deposit excessive extracellular matrix components into the cardiac interstitium, resulting in detrimental loss of myocardial compliance and decreased cardiac function. In accordance with this, ablation of activated fibroblasts in mice is associated with decreased fibrosis and improved cardiac function in response to pressure overload and myocardial infarction3,4. In pressure overload-induced HF in mice, the development of cardiac fibrosis coincides with the infiltration of CD4+ T cells into the myocardium and is characterized by the transformation of cardiac fibroblasts to myofibroblasts, which deposit excess collagen5,6. CD4+ T cell-deficient mice are protected from development of cardiac fibrosis5,7, and direct interactions between cardiac fibroblasts and IFN-γ-producing T helper 1 (TH1) cells have been shown to promote cardiac fibroblast transforming growth factor beta production, resulting in myofibroblast transformation6. Here, we aim to uncover the bidirectionality of these cellular interactions and determine if cardiac fibroblasts concomitantly promote CD4+ T cell activation, a new capability that is yet to be established.

CD4+ T cells are classically activated by professional antigen-presenting cells (APCs), which efficiently internalize and process antigens that are displayed by MHCII molecules on the cell membrane8. Additional co-stimulatory signals (CD80 or CD86) expressed by activated APCs are required for full activation of CD4+ T cells. Professional APCs include dendritic cells (DCs), macrophages and B cells, which are all present in the heart at lower frequencies relative to cardiac fibroblasts8. Unlike professional APCs, nonprofessional APCs do not constitutively express MHCII molecules, but the expression of MHCII can be induced in neutrophils, mast cells and endothelial cells by various stimuli9. Additionally, human dermal fibroblasts exposed to IFN-γ in vitro express MHCII, and promote recall antigen-dependent T cell responses10. We hypothesized that the cellular interactions between cardiac fibroblasts and CD4+ T cells in the heart imply a new role for cardiac fibroblasts as nonprofessional APCs central to cardiac CD4+ T cell responses and pathology.

Results

IFN-γ induces MHCII expression in cardiac fibroblasts

Nonprofessional APCs do not typically express MHCII unless induced by specific stimuli9. In conjunction with MHCII, either CD80 or CD86 co-stimulatory signals are required for full activation of CD4+ T cells upon engagement of the T cell receptor (TCR). We isolated the left ventricle (LV) from adult wild-type (WT) mice and cultured purified CD31CD45MEFSK4+ cardiac fibroblasts (Fig. 1a). We found that MHCII was not expressed at baseline but was progressively induced over 5 d of IFN-γ stimulation in culture (Fig. 1b,c). Conversely, CD80 was highly expressed at baseline and was refractory to IFN-γ stimulation (Fig. 1d,e). CD86 was neither expressed at baseline, nor induced by IFN-γ (Fig. 1f,g). Expression of MHCII after IFN-γ stimulation in vitro was further demonstrated by immunofluorescence microscopy on cardiac fibroblasts (Fig. 1h). In the heart, the majority of resident cardiac fibroblasts express transcription factor 21 (encoded by Tcf21)11. To delineate cardiac fibroblast expression of MHCII more specifically, we utilized cardiac fibroblast lineage tracing reporter mice (Tcf21iCre/+; R26eGFP) with a tamoxifen-regulated Cre recombinase knocked into one allele of Tcf21 that drives membrane-targeted GFP expression from the Rosa26 locus (R26). GFP+ cells from Tcf21iCre/+; R26eGFP mouse hearts were sorted by fluorescence-activated cell sorting (FACS) and cultured for 72 h in the presence of IFN-γ. Similarly, we found that CD80 was highly expressed at baseline, while MHCII was induced after 72 h of IFN-γ stimulation (Fig. 1i–k). Furthermore, expression of MHCII after IFN-γ stimulation in vitro was demonstrated by immunofluorescence microscopy on GFP+ cardiac fibroblasts derived from Tcf21iCre/+; R26eGFP mice (Fig. 1l). We next evaluated in vivo induction of MHCII and co-stimulatory molecule expression on CD31CD45MEFSK4+ cardiac fibroblasts in the myocardial tissue microenvironment by flow cytometry (Fig. 2a). IFN-γ was administered daily to WT mice for 5 d (Fig. 2b). In vehicle-treated mice, 30–50% of cardiac fibroblasts expressed CD80 on the surface but did not express MHCII or CD86 (Fig. 1c–h). However, IFN-γ treatment induced MHCII surface expression on 28–34% of cardiac fibroblasts, while CD80 and CD86 expression was not altered (Fig. 2c–h). These data demonstrate that IFN-γ induces expression of MHCII in cardiac fibroblasts, and, together with the expression of CD80, suggest that cardiac fibroblasts are capable of functioning as nonprofessional APCs that can induce CD4+ T cell activation.

Fig. 1: Cardiac fibroblasts express MHCII in response to IFN-γ stimulation in vitro.
figure 1

ah, CD31CD45MEFSK4+ primary adult mouse cardiac fibroblasts were sorted from heart digests and cultured in the presence of 100 U ml−1 recombinant mouse IFN-γ for up to 5 d. CD31CD45MEFSK4+ cardiac fibroblasts were analyzed by flow cytometry (≥20,000 target cells acquired; a) to determine surface expression of MHCII (b and c) as well as CD80 (d and e) and CD86 (f and g) co-stimulatory molecules. D, day. n = 5 (control and D3) and n = 3 (D5) independent experiments and cellular preparations. Immunofluorescence staining for MHCII as well as vimentin was performed after 3 d of IFN-γ stimulation. n = 3 independent experiments and cardiac fibroblasts preparations (h). il, Sorted GFP+ cardiac fibroblasts from Tcf21iCre/+; R26eGFP lineage tracing reporter mice were cultured in the presence of 100 U ml−1 recombinant mouse IFN-γ for 72 h. GFP+ cardiac fibroblasts were analyzed by flow cytometry (≥5,000 target cells acquired; i) to determine surface expression of MHCII and CD80 (j and k) and quantified. n = 3 (MHCII) and n = 5 (CD80) independent experiments and cellular preparations. Immunofluorescence staining for MHCII on GFP+ cardiac fibroblasts was performed (l). n = 3 independent experiments and cardiac fibroblasts preparations. Scale bars, 100 μm. Error bars represent the mean ± s.d. *P < 0.05, ***P < 0.001; one-way analysis of variance (ANOVA; c, e and g) and two-tailed unpaired t-test (j and k). NS, not significant.

Source data

Fig. 2: Cardiac fibroblasts express MHCII in response to IFN-γ stimulation in vivo.
figure 2

ah, CD31CD45MEFSK4+ cardiac fibroblasts were directly analyzed by flow cytometry (a) from the LV of WT mice treated with daily intraperitoneal (i.p.) injections of IFN-γ (25,000U) or PBS vehicle and collected on day 5 (b). Surface expression of MHCII (c and d), CD80 (e and f) and CD86 (g and h) was determined (≥5,000 target cells acquired). n = 4 vehicle and n = 5 IFN-γ-treated mice. Error bars represent the mean ± s.d. *P = 0.0159; two-tailed Mann–Whitney test.

Source data

MHCII+ cardiac fibroblasts induce CD4+ T cell activation

We next evaluated the ability of cardiac fibroblasts to activate naïve OTII CD4+ T cells, which express a transgenic TCR specific for chicken ovalbumin (OVA) peptide. CD62LhiCD44lo naïve OTII CD4+ T cells were sorted by FACS (Fig. 3a) and co-cultured with primary LV cardiac fibroblasts stimulated with IFN-γ to induce MHCII expression and loaded with OVA peptide. We identified robust induction of CD4+ T cell proliferation after 72 h of co-culture as determined by CFSE dilution assay (Fig. 3b,d), although not to the same extent as professional antigen-presenting bone marrow-derived dendritic cells (BMDCs; Fig. 3c,d). These responses were restricted by MHCII as MHCII-deficient (MhcII/) cardiac fibroblasts did not induce naïve CD4+ T cell proliferation (Fig. 3b). To exclude possible contamination with professional APCs in culture, purified GFP+ CD45 cardiac fibroblasts from Tcf21iCre/+; R26eGFP mice were sorted by FACS (Fig. 3e) and stimulated with IFN-γ to induce MHCII expression and co-cultured with FACS-sorted CD62LhiCD44lo naïve OTII CD4+ T cells in the presence of OVA peptide. Similarly, we identified robust induction of CD4+ T cell proliferation after 72 h of co-culture as determined by CFSE dilution (Fig. 3f,g). TH1 cells are a major cellular source of endogenous IFN-γ and establish intimate contact with cardiac fibroblasts upon infiltration into the heart6. We examined the possibility that antigen presentation to TH1 cells during this contact triggers further IFN-γ production to support sustained antigen presentation. We co-cultured cardiac fibroblasts and differentiated OTII TH1 cells in the presence of OVA peptide (Fig. 4a). After 24 h, TH1 cells showed marked induction of IFN-γ production, as determined by intracellular cytokine staining (Fig. 4b,c). Moreover, 72-h co-cultures of primary adult cardiac fibroblasts with OTII TH1 cells labeled with CFSE, in the presence of OVA peptide, demonstrated increased MHCII surface expression in cardiac fibroblasts as compared to cardiac fibroblasts alone (Fig. 4d,e), and resulted in OTII TH1 cell, but not WT TH1 cell, proliferation (Fig. 4f,g). These data demonstrate the ability of cardiac fibroblasts in the presence of IFN-γ to present peptide antigens to naïve CD4+ T cells. Additionally, cardiac fibroblasts express MHCII during cardiac fibroblast–TH1 cell contact and promote antigen-specific TH1 cell activation.

Fig. 3: Cardiac fibroblasts induce antigen-specific CD4+ T cell proliferation in vitro in the presence of IFN-γ.
figure 3

a, Naïve CD62LhiCD44lo CD4+ T cells were sorted by FACS from OTII bulk splenocytes and labeled with CFSE (2 μM). bd, Primary adult mouse cardiac fibroblasts and BMDCs derived from WT or MhcII/ mice were cultured in the presence or absence of IFN-γ (100 U ml−1) for 24 h and loaded with OVA peptide (4 μg ml−1) where indicated, 2 h before co-culture with OTII naïve CD4+ T cells (b). After 72 h, CFSE dilution was evaluated by flow cytometry (≥10,000 target cells acquired; c) and quantified (d). n = 3 (CFB OVA), n = 5 (CFB + IFN-γ + OVA; BMDCs + OVA), independent experiments and cell preparations. CFB, cardiac fibroblast. e, Sorted GFP+ mouse cardiac fibroblasts from Tcf21iCre/+; R26eGFP lineage tracing reporter mice were cultured in the presence or absence of IFN-γ (100 U ml−1) for 24 h and loaded with OVA peptide (4 µg ml−1) where indicated, 2 h before co-culture with OTII naïve CD4+ T cells. f,g, After 72 h CFSE dilution was evaluated by flow cytometry (≥5,000 target cells acquired; f) and quantified (g). n = 3 independent experiments and cell preparations. Error bars represent the mean ± s.d. **P < 0.01, ***P < 0.001; one-way ANOVA.

Source data

Fig. 4: IFN-γ produced by TH1 cells induces MHCII expression by cardiac fibroblasts and promotes antigen-specific CD4+ T cell activation.
figure 4

a, OTII TH1 cells were differentiated in vitro from bulk CD4+ splenocytes, and co-cultured with adult cardiac fibroblasts and OVA peptide (4 μg ml−1). b,c, After 24 h of co-culture TH1 IFN-γ production was determined by intracellular cytokine staining (≥15,000 target cells acquired). n = 3 control and n = 6 OVA, independent experiments and cell preparations. dg, After 72 h of co-culture CD31CD45MEFSK4+ cardiac fibroblasts were analyzed by flow cytometry (≥5,000 target cells acquired) for surface expression of MHCII (n = 3, independent experiments and cell preparations; d and e) and antigen-specific TH1 cell proliferation via CFSE dilution (≥8,000 target cells acquired; f and g). n = 3 independent experiments and cell preparations. Error bars represent the mean ± s.d. *P < 0.05, **P < 0.01, two-tailed Student’s unpaired t-test (c and e) and one-way ANOVA (g). MFI, mean fluorescence intensity. FMO, Fluorescence minus one control.

Source data

Cardiac fibroblasts capture antigens for MHCII presentation

Exogenously delivered OVA peptide can directly bind the peptide groove of cell surface MHCII molecules for presentation to CD4+ T cells (Fig. 2). However, defining attributes of APCs include the capacity to internalize soluble and particulate antigens, process them, load peptides onto MHCII in the cytosol and translocate MHCII–peptide complexes to the cell surface to activate CD4+ T cells. Soluble OVA protein is taken up via macropinocytosis or mannose receptor-mediated endocytosis12, whereas the uptake of relatively larger particles involves phagocytosis, with phagosomes containing the ingested material maturing into acidified phagolysosomes13. To track soluble antigen uptake and processing by cardiac fibroblasts, we used DQ OVA, a self-quenched conjugate of OVA protein that fluoresces upon proteolytic degradation. Strikingly, cardiac fibroblasts efficiently took up and processed DQ OVA, as demonstrated by flow cytometry and immunofluorescence, and, unlike MHCII expression, these functions were independent of IFN-γ stimulation (Fig. 5a,b). Moreover, OVA protein internalized by cardiac fibroblasts, combined with IFN-γ stimulation to promote MHCII expression, resulted in naïve OTII CD4 + T cell proliferation (Fig. 5c,d). We next evaluated phagocytic activity of cardiac fibroblasts using pHrodo green Escherichia coli BioParticles, which become fluorescent in acidified phagolysosomes. A strong fluorescent signal emitted from cardiac fibroblasts was detected by both flow cytometry and immunofluorescence, demonstrating efficient phagocytosis independent of IFN-γ stimulation (Fig. 5e,f). Furthermore, E. coli transformed with a plasmid conferring OVA protein expression were engulfed by cardiac fibroblasts and induced antigen-specific naïve OTII CD4+ T cell proliferation, which was remarkably comparable to that induced by BMDCs (Fig. 5g,h). These data demonstrate several functions of cardiac fibroblasts as APCs, including efficient uptake and processing of soluble and particulate antigens, and presentation of peptide fragments via MHCII to generate CD4+ T cell immune responses.

Fig. 5: Cardiac fibroblasts take up, process and present antigens to CD4+ T cells in vitro.
figure 5

a, CD31CD45MEFSK4+ primary adult mouse cardiac fibroblasts were cultured in the presence of DQ OVA (100 μg ml−1) and IFN-γ (100 U ml−1) overnight to evaluate uptake and intracellular processing of DQ OVA by flow cytometry (≥10,000 target cells acquired). FMO, Fluorescence minus one control. b, Immunofluorescence microscopy in cardiac fibroblasts and BMDCs. n = 3 independent experiments). c,d, WT cardiac fibroblasts and BMDCs were treated with purified OVA protein (100 μg ml−1) overnight and co-cultured with CFSE-labeled naïve OTII CD4+ T cells in the presence of IFN-γ (100 U ml−1) for 72 h to evaluate CD4+ T cell proliferation by flow cytometry (≥5,000 target cells acquired). n = 7 and n = 5 (CFB untreated and OVA + IFN-γ treated) and n = 6 and n = 7 (BMDCs untreated and OVA treated) independent experiments and cell preparations. e, WT cardiac fibroblasts were cultured in the presence of pHrodo Green E. coli bioparticles overnight to evaluate phagocytosis by flow cytometry (≥7,000 target cells acquired). f, Intracellular accumulation of bioparticles in acidified phagolysosomes was further evaluated by immunofluorescence microscopy in cardiac fibroblasts and BMDCs. n = 3 independent experiments. g,h, WT cardiac fibroblasts and BMDCs were cultured overnight in the presence of transformed E. coli particles expressing either chicken OVA or empty vector (EV), and co-cultured with CFSE-labeled naïve OTII CD4+ T cells for 72 h to evaluate CD4+ T cell proliferation by flow cytometry (≥10,000 target cells acquired). n = 11 BMDC EV and n = 8 independent experiments for all other conditions. Scale bars, 100 μm. Error bars represent the mean ± s.d. **P < 0.01, ***P < 0.001; one-way ANOVA test.

Source data

Cardiac fibroblast MHCII expression modulates pathology

Because myocardial IFN-γ is upregulated in response to transverse aortic constriction (TAC), a well-established model of HF6,14, and Chagas disease15,16, we investigated the induction of MHCII on cardiac fibroblasts in vivo, in response to these distinct models of cardiac inflammation. We analyzed the surface expression of MHCII and CD80 on CD31CD45MEFSK4+ cardiac fibroblasts by flow cytometry (Fig. 6a). Four weeks of TAC in WT mice resulted in increased expression of both MHCII and CD80 on cardiac fibroblasts relative to Sham control mice (Fig. 6b–e). Similarly, WT mice infected with Trypanosoma cruzi and collected after 19 d, at the peak of parasitemia, demonstrated even greater MHCII induction on cardiac fibroblasts relative to mock infected controls. However, CD80 expression did not change (Extended Data Fig. 1). Expression of MHCII in vivo by cardiac fibroblasts after TAC in LV tissue sections was further demonstrated by immunofluorescence microscopy on alpha smooth muscle actin (α-SMA)-positive cells in WT mice (Fig. 6f), and in GFP+ cells traced from Tcf21iCre/+; R26eGFP reporter mice (Fig. 6g). Additionally, direct contact can be identified between CD4+ T cells and αSMA+ cardiac fibroblasts cells within LV fibrotic lesions in WT TAC mice (Fig. 6h). We next generated Tcf21iCre/+MhcIIfl/fl mice expressing tamoxifen-inducible Cre recombinase driven by Tcf21 expression to conditionally delete the H2Ab1, also known as Ia2, gene (MhcII) in cardiac fibroblasts (Fig. 7a). After tamoxifen treatment of Tcf21iCre/+MhcIIfl/fl mice, MhcII recombination in sorted CD31CD45MEFSK4+ cardiac fibroblasts was detected by PCR, and occurred in a cell-specific manner, as CD31CD45+ cardiac leukocytes did not undergo MhcII recombination (Fig. 7b and Extended Data Fig. 2). Furthermore, CD31CD45MEFSK4+ cardiac fibroblasts cultured with 4-hydroxytamoxifen showed 3.5-fold reduction in MHCII protein expression after 3 d of IFN-γ stimulation (Fig. 7h–l), while CD45+CD11b+CD11c+ BMDCs did not show a reduction in MHCII in response to 4-hydroxytamoxifen treatment (Fig. 7e,f). Tcf21iCre/+MhcIIfl/fl mice were treated with tamoxifen by intraperitoneal injection before TAC surgery and maintained on tamoxifen chow for 4 weeks until collection (Fig. 7g). MHCII protein surface expression on myeloid cells in the mediastinal lymph nodes (mLNs) draining the heart (Fig. 7h), as well as in cardiac leukocytes and cardiac endothelial cells, was similar across the treatment groups, with only observed decreased MHCII protein surface expression in cardiac fibroblasts from Tcf21iCre/+MhcIIfl/fl mice treated with tamoxifen (Extended Data Fig. 3). Vehicle-treated and tamoxifen-treated (cardiac fibroblast MHCII-deficient) mice showed similar LV CD4+ T cell infiltration (Fig. 7i,j) and similar CD4+ T cell activation in the mLNs (Extended Data Fig. 4); however, only tamoxifen-treated mice had decreased cardiac fibroblasts compared to vehicle-treated mice in the onset of TAC (Extended Data Fig. 5), and an amelioration of adverse cardiac remodeling, including perivascular fibrosis (Fig. 7k,l) and cardiomyocyte hypertrophy (Extended Data Fig. 4) compared to the vehicle treatment group. Additionally, cardiac systolic function remained preserved in tamoxifen-treated mice in contrast to the vehicle treatment group as demonstrated by preserved fractional shortening (Fig. 7m,n and Table 1). These results demonstrate that IFN-γ-associated cardiac inflammation promotes MHCII expression in cardiac fibroblasts, and that MHCII in cardiac fibroblasts is central to cardiac remodeling and dysfunction in response to TAC.

Fig. 6: Cardiac fibroblasts express MHCII in response to cardiac pressure overload.
figure 6

ae, TAC surgery was performed on mice and collected after 4 weeks. a, CD31CD45MEFSK4+ cardiac fibroblasts from the LV of WT mice were directly analyzed by flow cytometry (≥5,000 target cells acquired) to determine surface expression of MHCII (b and c) and CD80 (d and e). n = 3 sham-treated and TAC-treated mice. f,g, Immunofluorescence staining of LV tissue sections was performed to detect MHCII on α-SMA+ cells in WT mice (white arrows; f) and MHCII on GFP+ cells in Tcf21iCre/+; R26eGFP lineage tracing reporter mice (white arrows; g). Scale bars, 100 μm. h, Immunofluorescence staining of WT LV tissue sections was performed to evaluate colocalization of CD4+ cells and α-SMA+ cells. Scale bars, 25 μm. Images are representative of n = 3 mice (f and h) and n = 5 mice (g). Error bars represent the mean ± s.d. *P < 0.05, **P < 0.01; two-tailed unpaired t-test.

Source data

Fig. 7: Cardiac fibroblast MHCII expression is required for cardiac dysfunction to develop in response to pressure overload.
figure 7

a, Conditional deletion of MHCII (Ia2 gene) in cardiac fibroblasts was induced by administering tamoxifen to Tcf21iCre/+MhcIIfl/fl mice. b, CD31CD45MEFSK4+ cardiac fibroblasts and CD31CD45+ leukocytes were FACS sorted from the LV of vehicle-treated and tamoxifen (TMX)-treated Tcf21iCre/+MhcIIfl/fl mice and genomic DNA was probed for the intact (P1 and P2 primers) and recombined (P1 and P3 primers) MhcII alleles by PCR using specific primers (blue arrows) that detect the recombined (MhcII excision) or the intact (non-recombined) allele under the different conditions. Excision was only observed in cardiac fibroblasts from tamoxifen-treated mice with the P1–P3 primers. GAPDH is shown as a loading control. Gel represents DNA from cells sorted from n = 3 mice (pooled hearts). c,d, CD31CD45MEFSK4+ cardiac fibroblasts from Tcf21iCre/+MhcIIfl/fl mice were cultured overnight in the presence of 4-hydroxytamoxifen (4OH-TMX; 1 μM), then stimulated with IFN-γ (100 U ml−1) for 72 h to determine surface expression of MHCII by flow cytometry (≥2,000 target cells acquired). n = 4 vehicle and n = 3 for 4HO-TMX, independent experiments and cell preparations. e,f, CD45+CD11b+CD11c+ BMDCs from Tcf21iCre/+MhcIIfl/fl mice were cultured overnight in the presence of 4-hydroxytamoxifen (1 μM), then stimulated with lipopolysaccharides (10 ng ml−1) for 72 h to determine surface expression of MHCII by flow cytometry (≥12,000 target cells acquired). n = 5 independent experiments and cell preparations. g, Tcf21iCre/+MhcIIfl/fl mice were treated with tamoxifen (75 mg per kg body weight) by five daily i.p. injections before TAC surgery and maintained on tamoxifen citrate chow for 4 weeks after surgery to decrease MHCII expression in cardiac fibroblasts, or with vehicle (mice with MHCII-sufficient cardiac fibroblasts). h, MHCII expression on CD11b+ myeloid cells in the mLNs was determined by flow cytometry. i,j, Immunohistochemistry of frozen LV sections was used to determine LV CD4+ T cell infiltration, scanning the entire LV section area. Representative images of n = 3 hearts (vehicle and TMX sham), n = 4 (TAC vehicle) and n = 7 (TAC TMX). k,l, Picrosirius red staining of fixed LV tissue sections was used to determine perivascular fibrosis. Representative images of n = 3 hearts (vehicle and TMX sham), n = 4 (TAC vehicle) and n = 7 (TAC TMX). m,n, Transthoracic echocardiography was used to measure LV fractional shortening (FS). Scale bars, 100 μm. Each data point represents an individual animal in h, j, l and n. Error bars represent the mean ± s.d. *P < 0.05; two-tailed unpaired t-test (d) and one-way ANOVA (h, j, l and n).

Source data

Table 1 Characterization of left ventricular function by echocardiography 4 weeks after surgery in Tcf21iCre/+MhcIIfl/fl

Discussion

We report for the first time, to our knowledge, that cardiac fibroblasts function as APCs and contribute to cardiac fibrosis and dysfunction through IFN-γ-induced MHCII. We find that cardiac fibroblasts: (1) express MHCII in vitro and in vivo in response to IFN-γ as well as in two different experimental models of cardiac disease; (2) express the requisite co-stimulatory signal CD80 necessary for CD4+ T cell activation; (3) efficiently capture and process extracellular soluble and particulate antigens for presentation; and (4) induce naïve CD4+ T cell and TH1 cell activation in an MHCII-dependent and antigen-dependent manner. Moreover, we identify a central role for cardiac fibroblast MHCII expression in cardiac remodeling and dysfunction in an experimental model of HF that is highly dependent on IFN-γ+ TH1 cell immune responses.

It has been known for decades that stromal cells, including dermal fibroblasts, can express MHCII in response to IFN-γ in vitro17,18; therefore, this finding in cardiac fibroblasts could be perhaps unsurprising. The key message of our study emerges from the relevance of cardiac fibroblasts as APCs in the context of cardiac inflammation and pathology. Our in vitro results demonstrate that MHCII expressed on cardiac fibroblasts, loaded with a peptide antigen, can induce naïve CD4+ T cell proliferation, a capability considered exclusive to mature DCs8. This serves as decisive validation of the functionality of cardiac fibroblasts as APCs. The quality of TCR interactions with the peptide–MHCII complex is determinant of the strength of CD4+ T cell activation, with higher-affinity interactions generating a more potent response19. In our in vitro studies, we utilize OVA-specific naïve OTII CD4+ T cells against their optimal OVA peptide ligand, thus it is possible that cardiac fibroblasts have a diminished ability to activate CD4+ T cells when presenting suboptimal cognate antigens. However, given the confinement of cardiac fibroblasts to the heart it is unlikely that they establish contact with naïve CD4+ T cells in vivo, but rather with effector CD4+ T cells, which are more sensitive to triggering through the TCR. We have reported strong TCR engagement of CD4+ T cells within the mouse heart20 and induction of T helper responses in the mLNs during HF development6,21. Here, we find that IFN-γ treatment and two cardiac pathologies associated with elevated IFN-γ show elevated expression of MHCII on cardiac fibroblasts. We demonstrate the ability of cardiac fibroblasts to present peptide antigen to TH1 cells in vitro and propose a role in generating recall responses from TH1 cells infiltrated in the heart through cardiac fibroblasts expressing MHCII and the involvement of this axis in cardiac fibrosis and dysfunction.

We identify baseline expression of the co-stimulatory molecule CD80 in >90% of the cells in vitro, and in approximately 20–50% of cardiac fibroblasts in vivo, and detect increased expression in response to cardiac pressure overload. This variability may arise due to heterogeneity of cardiac fibroblasts in the myocardium at varying stages of maturity22, in comparison to synchronized cells in culture. A variety of stimuli have also been shown to increase CD80 expression in APCs, including CD40 ligand, IFN-α, granulocyte–macrophage colony-stimulating factor (GM-CSF), Toll-like receptor ligands and tumor-necrosis factor, which may become expressed in the heart during TAC23,24,25. In addition to direct antigen presentation by cardiac fibroblasts, CD80 expression may have further relevance in global enhancement of CD4+ T cell activation in the heart via trans-co-stimulation, by augmenting co-stimulatory signals delivered to CD4+ T cells already engaged with an APC26. CD80 and CD86 are thought to be interchangeable co-stimulators, and we find that cardiac fibroblasts constitutively express CD80 but not CD86. In contrast to cardiac fibroblasts, CD86 is abundantly expressed on DCs and is thought to play a prominent role in initial T cell activation27. CD80 expression is more slowly induced following DC activation and functional studies suggest that CD80 is a more potent CD28 ligand due its higher relative affinity for CD28 (ref. 24). Our data demonstrate that in the presence of IFN-γ, MHCII and CD80 are available to engage with the TCR and CD28 in CD4+ T cells, respectively, and induce T cell proliferation.

We and others have demonstrated intimate contact between cardiac fibroblasts and immune cells, including TH1 cells and macrophages, in the context of cardiac dysfunction due to pressure overload or age6,28. While these reports focused on the unidirectional signals exerted by immune cells on cardiac fibroblasts and highlighted the requirement for intimate cellular contact to develop cardiac fibrosis, the bidirectionality of this interaction remained unexplored. We now demonstrate that IFN-γ-secreting TH1 cells induce MHCII on cardiac fibroblasts, which subsequently induce further IFN-γ production by TH1 cells, propagating a positive feedback loop that likely exacerbates TH1 cell activation and myofibroblast transformation. Observations on the proximity of MHCII+ cardiac fibroblasts and CD4+ T cells in the TAC hearts, together with our findings in mice lacking MHCII in cardiac fibroblasts having decreased fibrosis, are in support of this loop. We demonstrate the ability of cardiac fibroblasts to capture extracellular antigens and process them into peptide antigens loaded onto MHCII to induce CD4+ T cell activation. Cardiac fibroblasts are nonprofessional phagocytes capable of engulfing apoptotic cells to assist with debris clearance after myocardial ischemia29,30. Our results demonstrate that cardiac fibroblasts internalize both soluble and particulate material, process and load peptides onto MHCII and induce CD4+ T cell proliferation. We additionally report that tamoxifen-treated (cardiac fibroblast MHCII-deficient) Tcf21iCre/+MhcIIfl/fl mice have intact effector CD4+ T cell expansion in the mLNs and CD4+ LV T cell recruitment in response to cardiac pressure overload, as compared to vehicle-treated (MHCII-sufficient) mice. Yet, the lack of cardiac fibroblast MHCII results in decreased cardiac fibroblasts and is sufficient to prevent adverse cardiac remodeling and dysfunction. These results are in line with a central role for DCs in antigen presentation and TH1 cell expansion in the mLNs in the context of heart inflammation as we reported6,20,31, and as demonstrated by DC ablation being effective in preventing cardiac fibrosis and dysfunction32. We recently reported that DCs present cardiac neoantigens to CD4+ T cells, and that interrupting this presentation prevents TCR engagement by CD4+ T cells in the heart and ameliorates cardiac dysfunction20. Because cardiac fibroblast MHCII-deficient mice have intact MHCII expression in myeloid cells, the observed similar number of effector T cells in the lymph nodes, devoid of cardiac fibroblasts, is not surprising. The similar T cell numbers observed in the heart in the presence or absence of MHCII may suggest continuous cardiotropism that is not altered by cardiac fibroblast MHCII expression, as well as a T cell recall response in situ induced by cardiac DCs expressing MHCII that additionally takes place. It is also possible that DCs and cardiac fibroblasts participate in trans-co-stimulation augmenting signals delivered to CD4+ T cells already engaged with an APC and that this is altered in the absence of cardiac fibroblast MHCII, something worth addressing in future studies. While our in vitro results demonstrate cardiac fibroblast induction of T cell proliferation and IFN-γ production through MHCII, we have not directly tested T cell proliferation in the heart in vivo. Nevertheless, our in vivo data demonstrate that cardiac fibroblast expression of MHCII is required for cardiac fibrosis and systolic dysfunction, which are both T cell dependent5,6,7,20. Our new findings endorse the intriguing possibility of treating HF, the leading cause of death in the United States33, with targeted therapies to the cardiac fibroblast that specifically modulate cardiac CD4+ T cell responses without impairing systemic CD4+ T cell activation by DCs, whiich could have undesired immunosuppressive side effects. A detailed investigation of the dynamics of cell-specific MHCII engagement of the TCR in the mLNs and in the heart will identify when such targeted therapies may be more effective.

We acknowledge some limitations in this study. For instance, while we clearly demonstrate that IFN-γ induces MHCII expression of the vast majority of cardiac fibroblasts in vitro, a lower frequency of cardiac fibroblasts express MHCII in vivo in the three experimental models we use (TAC, Chagas or IFN-γ injections) compared to purified cardiac fibroblasts in vitro. It is possible that, in vivo, in the onset of TAC, other factors besides IFN-γ contribute to MHCII expression, and that, given the plasticity of cardiac fibroblasts, specific cardiac fibroblast populations are responsible for antigen presentation. Our data showing that conditional deletion of MHCII in cardiac fibroblasts results in decreased cardiac fibroblast presence in TAC suggest an additional role for cardiac fibroblast MHCII in fibroblast proliferation, which is critical for cardiac fibrosis in pathological conditions that include cardiac pressure overload. Future studies will investigate which fibroblast subsets express MHCII and how MHCII regulates fibroblast function and proliferation. These results are in line with recently published single-cell RNA-sequencing data demonstrating enhanced expression of MHCII in mice 5 weeks after TAC and variable expression in different fibroblast clusters identified in the human cardiac samples obtained from individuals with dilated cardiomyopathy34,35. Noteworthy, recent RNA-sequencing data in individuals with dilated cardiomyopathy before and after left ventricular assist device implantation demonstrate lower expression of cardiac fibroblast HLA-DR in individuals after left ventricular assist device support, in line with our findings that MHCII expression in cardiac fibroblasts contributes to systolic function35. It is also plausible that other APCs sequentially dominate this response during the progression of adverse cardiac remodeling after TAC. These APCs may include lymph node fibroblastic reticular cells, which have been reported to express MHCII in response to IFN-γ36,37. Whether this is the case in mLNs in this context is unknown. For instance, CD4+ T cell engagement of antigen presented by DCs or fibroblastic reticular cells may happen early on in the mLNs, before T cell infiltration in the heart occurs, and continues once T cells infiltrate the heart, coexisting with cardiac fibroblast presentation to intramyocardial CD4+ T cells and cardiac fibroblast proliferation and transformation. The data presented here are limited to one time point after cardiac pressure overload has been induced and future studies will be necessary to determine the dynamics of the cell-specific role of MHCII. This is something worth studying in the immediate future that may have consequences to understand immune tolerance in the heart, which has not been investigated herein. The mechanisms of peptide loading of MHCII in cardiac fibroblasts, and whether these are shared with professional APCs, are also an ongoing research direction we are pursuing. Additionally, we only used male mice in this study, and, given that there are sex-specific differences in HF, it is possible that sex-specific differences also exist in the way cardiac fibroblasts sustain cardiac T cell immune responses. Despite these limitations, our study unveils a new role for cardiac fibroblasts as central contributors to cardiac inflammation and adverse remodeling through MHCII that may have potential implications when targeting inflammation in heart disease (Extended Data Fig. 6).

Taken together, our results contribute to the growing field of cardio-immunology and advance the intriguing possibility that diverse subsets of fibroblasts abundantly dispersed within other organs perform similar functions, serving as sentinel cells that sense local insults and directly boost adaptive immune responses.

Methods

Mice

Mice were bred and maintained under pathogen-free conditions at Tufts University animal facilities and treated in compliance with the Guide for the Care and Use of Laboratory Animals (National Academy of Science). C57BL/6 WT, OTII, MhcII-floxed mice (all purchased from Jackson Laboratories) and Tcf21-cre mice (provided by J. Davis, University of Washington) were bred and maintained in-house and euthanized at 4–14 weeks of age for tissue collection. Tcf21iCre/+; R26eGFP mice were bred and maintained under pathogen-free conditions at the University of Washington, with all animal experimentation approved by the University of Washington Institutional Animal Care and Use Committee. All animal studies were approved by the Tufts University Institutional Animal Care and Use Committee.

Adult cardiac fibroblasts preparation

Heart ventricles from adult WT and Tcf21iCre/+; R26eGFP mice (4–8 weeks old) were excised, minced and digested at 37 °C for 30 min with agitation in a mixture of 0.25% trypsin and 40 μg ml−1 liberase TL (Roche). Digested tissue was then centrifuged at 300g for 5 min and the pellet resuspended in fibroblast growth medium (Lonza) and plated on 0.1% gelatin-coated plates for 2 h at 37 °C with 5% CO2. Unattached cells were discarded, and adherent cells were further cultured. The digested heart cell suspension was filtered through a 40-µm cell strainer and stained with BV711-conjugated anti-CD45.2 (104), PE-conjugated anti-CD31 (MEC13.3), APC-conjugated anti-feeder cells (mEF-SK4), as outlined below for quantitative flow cytometry and CD31CD45MEFSK4+ cells, or GFP+ cells when using Tcf21iCre/+; R26eGFP lineage tracing mice, were sorted on a BD FACSAria and plated on 0.1% gelatin-coated plates at 37 °C with 5% CO2. Once confluent, cardiac fibroblasts were detached with trypsin and either used directly in experiments or passaged one more time to use in experiments. Confirmation of purity was confirmed using these markers, or GFP, in every experiment.

Dendritic cell preparation

Bone marrow was flushed from femurs and tibias of WT mice. Red blood cells were lysed with Tris ammonium chloride buffer (Roche) and cells were cultured in complete RPMI medium with recombinant GM-CSF (15 ng ml−1; PeproTech) for 7 d, with additional medium added on days 3 and 5. CD11c+ DC purity was >85% by FACS analysis.

CD4+ T cell preparation

CD4+ cells were isolated from spleen cell suspensions of OTII mice by positive selection with CD4 microbeads (Miltenyi Biotec) and further FACS sorted based on the expression of CD62L and CD44. Naïve CD4+ T cells were differentiated into TH1 cells by stimulation with plate-bound anti-CD3 (5 µg ml−1) and soluble anti-CD28 (1 µg ml−1) in the presence of interleukin (IL)-12 (0.01 µg ml−1), IL-2 (25 U ml−1) and anti-IL-4 (0.5 µg ml−1). On day 3 of stimulation, TH1 cultures were split at a ratio of 1:1 with fresh medium containing IL-2 (25 U ml−1). All cytokines and blocking antibodies were purchased from PeproTech. Differentiated T cells were collected a day later and immediately used for experiments. TH1 cell generation was confirmed by IFN-γ production upon PMA/ionomycin stimulation. Cells were labeled with 2 µM CFDA-SE tracer (Thermo Fisher) according to the manufacturer’s specifications.

Proliferation assay

In a 24-well plate, cardiac fibroblasts or BMDCs (100,000 cells per well) were allowed to adhere and were either pulsed for 4 h with 4 µg ml−1 OVA 323-339 (AnaSpec), or incubated overnight with 100 μg ml−1 OVA protein (Sigma), 100 μg ml−1 DQ OVA (Thermo Fisher) or OVA transformed E. coli (MOI 10). Cells were washed twice with PBS and co-cultured for 3 d with 1 million CFDA-SE-labeled OTII CD4 + T cells. For FACS sorting of purified CD44loCD62Lhi naïve CD4+ T cells, bulk splenocytes from OTII mice were stained with FITC-conjugated anti-CD4 (GK1.5), PE-conjugated anti-CD62L (MEL-14), and APC-conjugated anti-CD44 (IM7) as outlined below for FACS sorting on a BD FACSAria (BD Biosciences).

Quantitative flow cytometry

Flow cytometry was performed to analyze cell surface protein expression. The data were acquired on a BD LSR II flow cytometer (BD Biosciences) and analyzed using FlowJo software. LV digestion was performed by collagenase type II (0.895 mg ml−1) for 30 min at 37 °C. Cells were then stained with the following antibodies: FITC-conjugated anti-CD4 (GK1.5), APC-conjugated anti-CD4 (RM4-5), BV711-conjugated anti-CD45.2 (104), APC-conjugated anti-CD44 (IM7), PE-conjugated anti-CD62L (MEL-14), Alexa Fluor 488-conjugated anti-I-A/I-E, BV421-conjugated anti-CD80, PerCP/Cy5.5-conjugated anti-CD86 and PE-conjugated anti-CD31 (MEC13.3). All antibodies were purchased from BioLegend. APC-conjugated anti-feeder cells (mEF-SK4) were purchased from Miltenyi Biotec. Cells were surface stained by incubation with the relevant antibodies diluted in PBS + 2% FBS for 20 min at 4 °C, followed by two washes with PBS 2% FBS. When intracellular staining of signature cytokines was performed, cell suspensions were incubated overnight at 37 °C in the presence of 0.1% brefeldin A (BioLegend) and 0.1% monensin (BioLegend). After the incubation, surface staining was performed as indicated above, followed by cell fixation for 20 min at room temperature (RT) with fixation buffer (BioLegend). Upon fixation and washing with PBS 2% FBS, cell suspension was permeabilized with 1× Perm Wash buffer (BioLegend), and stained intracellularly for 30 min at RT in the dark. Absolute cell numbers were quantified using Precision Count Beads (BioLegend).

Immunofluorescence

Cardiac fibroblasts were cultured on 0.1% gelatin-coated glass coverslips in a six-well plate until confluent and treated daily with 100 U ml−1 mouse IFN-γ (PeproTech) for 3 d. At the end of each treatment, cell layers were washed twice with PBS and fixed in 3% paraformaldehyde. Nonspecific binding was prevented by incubation with PBS containing 10% goat serum (Jackson ImmunoResearch) for 1 h. Cells were incubated with primary antibodies against vimentin (Abcam, ab20346) and MHCII (Invitrogen, 14-5321-82) at a 1:250 dilution. The cells were then incubated at 4 °C overnight and washed three times with PBS. As controls, parallel coverslips were incubated with no primary antibody. Alexa Fluor 568-conjugated goat anti-mouse (Invitrogen, A-11004) and Alexa Fluor 488-conjugated goat anti-rat (Invitrogen, A48262) at a 1:500 dilution were used as secondary antibodies. Visualization was performed with a Nikon Ti inverted fluorescence microscope.

Acute T. cruzi Infection

Eight-week-old WT mice were infected with 20,000 T. cruzi trypomastigotes (Colombiana strain, discrete typing unit TcI, the most widely distributed in Colombia). Trypomastigotes were collected from infected Vero cells (American Type Culture Collection), collected from culture supernatant, purified by differential centrifugation, and resuspended in PBS before intraperitoneal injection in mice. Mice were euthanized at the peak of parasitemia, at 19 d after infection.

Mouse model of transverse aortic constriction

Pressure overload was induced by minimally invasive TAC surgery to constrict the transverse aorta (26 G, 0.51 mm outer diameter) of randomized 8- to 10-week-old male mice to induce HF as previously described38. Sham-operated mice underwent the same procedure but without aortic constriction. Four weeks after surgery, mice were euthanized, and tissue was collected for further analysis.

Tamoxifen treatment

To induce Cre recombinase activity, Tcf21iCre/+MhcIIfl/fl mice were treated with tamoxifen (20 mg ml−1; Millipore Sigma, T5648) dissolved in sunflower oil using 10% ethanol for 5 consecutive daily i.p. injections at 75 mg per kg body weight. After TAC surgery, mice were maintained on a diet containing 400 mg per kg body weight tamoxifen citrate (Envigo, TD.55125) for 4 weeks until collection.

In vivo echocardiography

In vivo transthoracic echocardiography was assessed 1 d before tissue collection. Mice were lightly sedated at 1–2% isoflurane in medical oxygen (0.7 l min−1), on a heated stage in the supine position as previously described38. Heart rates and respiratory rates were continuously monitored via the stage electrodes. Depilatory cream (Nair) was applied to the chest to remove fur and ultrasonic gel was applied to the 22–55-MHz echocardiography transducer (MS550D; Vevo 2100, FUJIFILM VisualSonics) to obtain parasternal short-axis views of the LV in M-mode with a target heart rate of 400–500 b.p.m. LV parameters and heart rate (Table 1) were measured by averaging values obtained from five cardiac cycles. All analyses were performed blindly using Vevo 2100 software (v3.1.1, FUJIFILM VisualSonics).

Histology

Immunohistochemistry, and immunofluorescence of cardiac tissue were performed in midpapillary LV tissue, cut into 5-μm sections. Frozen LV sections were incubated with primary antibody against mouse CD4 (BioLegend, clone GK1.5) for 1 h (1:500 dilution) followed by incubation with goat anti-rat biotinylated secondary antibody (1:300 dilution; Jackson ImmunoResearch, 112-065-062). Sections were incubated with Streptavidin-HRP (DAKO, K0675) and visualized using AEC Substrate-Chromogen. CD4+ T cells were quantified by manually scanning the entire LV tissue section and counting stained cells per section. LV sections from tissue fixed in 10% formalin and embedded in paraffin were stained with FITC-conjugated wheat germ agglutinin (Millipore Sigma, L4895) at 5 μg ml−1 to determine cardiomyocyte cross-sectional area by tracing the outline of at least 25 myocytes per section with NIS-Elements software. Antibodies to α-SMA (Sigma, A2547; 1:250 dilution), and to CD4 (BioLegend, 100401; 1:25 dilution), and MHCII (Invitrogen, 14-5321-82; 1:250 dilution), were used in immunofluorescence staining of LV tissue of WT and Tcf21iCre/+; R26eGFP mice subjected to sham and TAC. Fixed sections were stained with Picrosirius red to determine the percentage fibrotic area quantified using the National Institutes of Health (NIH) ImageJ software. All analyses were performed blindly.

Statistics

Data are presented as the mean ± s.d. Statistical analyses were done by Student’s unpaired t-test (two-tailed) or nonparametric Mann–Whitney test (two-tailed) to adjust for nonequal Gaussian distribution when comparing two groups. Multiple-group comparisons were performed by one-way ANOVA and Tukey’s post hoc test, where indicated, using GraphPad Prism software. Differences were considered statistically significant at P ≤ 0.05.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.