DYSFUNCTIONAL AND PRO-INFLAMMATORY REGULATORY T-LYMPHOCYTES ARE ESSENTIAL FOR ADVERSE CARDIAC REMODELING IN ISCHEMIC CARDIOMYOPATHY

Abstract

Background.

Heart failure (HF) is a state of inappropriately sustained inflammation, suggesting the loss of normal immunosuppressive mechanisms. Regulatory T-lymphocytes (Tregs) are considered key suppressors of immune responses; however, their role in HF is unknown. We hypothesized that Tregs are dysfunctional in ischemic cardiomyopathy and HF, and promote immune activation and left ventricular (LV) remodeling.

Methods.

Adult male wild-type (WT) C57BL/6 mice, Foxp3-diptheria toxin receptor(DTR) transgenic mice, and tumor necrosis factor(TNF)α receptor-1(TNFR1)^−/− mice underwent non-reperfused myocardial infarction (MI) to induce HF, or sham operation. LV remodeling was assessed by echocardiography, and histological and molecular phenotyping. Alterations in Treg profile and function were examined by flow cytometry, immunostaining, and in vitro cell assays.

Results.

As compared with WT sham mice, CD4⁺Foxp3⁺ Tregs in WT HF mice robustly expanded in the heart, circulation, spleen, and lymph nodes in a phasic manner after MI, beyond the early phase of wound healing, and exhibited pro-inflammatory Th1-type features with interferon-γ, TNFα, and TNFR1 expression, loss of immunomodulatory capacity, heightened proliferation, and potentiated anti-angiogenic and pro-fibrotic properties. Selective Treg ablation in Foxp3-DTR mice with ischemic cardiomyopathy reversed LV remodeling and dysfunction, alleviating hypertrophy and fibrosis, while suppressing circulating CD4⁺ T-cells and systemic inflammation, and enhancing tissue neovascularization. Importantly, Tregs reconstituted after ablation exhibited restoration of immunosuppressive capacity and normalized TNFR1 expression. Treg dysfunction was also tightly coupled to Treg-endothelial cell contact- and TNFR1-dependent inhibition of angiogenesis, and the mobilization and tissue infiltration of CD34⁺Flk1⁺ circulating angiogenic cells in a CCL5/CCR5-dependent manner. Anti-CD25-mediated Treg depletion in WT mice imparted similar benefits on LV remodeling, CACs, and tissue neovascularization.

Conclusions.

Pro-inflammatory and anti-angiogenic Tregs play an essential pathogenetic role in chronic ischemic HF to promote immune activation and pathological LV remodeling. The restoration of normal Treg function may be a viable approach to therapeutic immunomodulation in this disease.

INTRODUCTION

Myocardial infarction (MI) triggers a heightened inflammatory response, followed by inflammation resolution and wound healing.¹ Ideally, this results in tissue repair and preservation of cardiac function. However, if the infarct is large enough, or if the healing response is insufficient, adverse left ventricular (LV) remodeling and heart failure (HF) develop over time, accompanied by chronic and inappropriately sustained inflammation.² In addition to increased levels of pro-inflammatory cytokines, recent studies have indicated that in chronic ischemic HF, there is robust expansion of both innate (e.g., macrophages and dendritic cells [DCs])^3–5 and adaptive (e.g., T-cells)⁶ immune cells in the heart that are pathologically activated, induce tissue injury, and contribute to adverse remodeling. Hence, ischemic cardiomyopathy can be considered a state of chronic immune system activation.

Persistent immune activation in HF suggests insufficiency of normal immunomodulatory mechanisms. In this regard, regulatory T-cells (Tregs) are viewed as key orchestrators of immune homeostasis and peripheral tolerance.⁷ Suppression of effector T-cell (Teff) responses by Tregs may occur via multiple pathways including the secretion of inhibitory cytokines such as interleukin(IL)-10 and transforming growth factor(TGF)-β; granzyme A- and perforin-mediated Teff cytolysis; Teff metabolic disruption; and contact-dependent suppression of DC-Teff interactions.⁷ Lack of the Treg-specific transcription factor forkhead box protein-3 (Foxp3) has been linked to severe autoimmune disease in both humans⁸ and rodents.⁹ After acute MI, Treg activation is beneficial and limits inflammation and modulates macrophage phenotype, thereby improving cardiac wound healing and remodeling.^{10, 11}

We have recently demonstrated that CD4⁺ T-cells, including Tregs, are globally expanded and activated in chronic murine ischemic cardiomyopathy.⁶ Why there is persistent Teff activation and inflammation despite Treg expansion in ischemic HF is unclear, but may result from defective Treg-mediated immunosuppression. Notably, in autoimmune arthritis, the inflammatory microenvironment alters the phenotype of peripherally-induced Tregs to a pro-inflammatory state that accelerates synovial damage,¹² suggesting that Treg plasticity may underlie self-injury. Treg expansion in ischemic cardiomyopathy may also potentially exacerbate HF via non-immune mechanisms. Specifically, Tregs are anti-angiogenic and induce endothelial cell (EC) apoptosis,^{13, 14} and secrete TGF-β, which promotes extracellular matrix accumulation.¹⁵ In the failing heart, these effects would promote capillary rarefaction and interstitial fibrosis that exacerbate adverse LV remodeling. However, whether Treg dysfunction occurs in chronic HF and the role that Tregs play in disease pathogenesis are unknown. Hence, we tested the hypothesis that Tregs become dysfunctional, pro-inflammatory, and tissue-injurious in chronic ischemic HF, and that this phenotypic switch contributes to sustained inflammation and the progression of LV remodeling.

METHODS

The data and analytic methods pertinent to the findings of this study are available from the corresponding author (S.D.P.) on reasonable request.

Mouse models and in vivo protocols.

All animal studies were approved by the Institutional Animal Care and Use Committee of the University of Alabama at Birmingham and were compliant with the NIH Guide for the Care and Use of Laboratory Animals (DHHS publication No. 85–23, revised 1996). A total of 225 mice were used. C57BL/6 wild-type (WT) and TNFR1^−/− mice were obtained from Jackson Laboratories (stock #000664 and 002818, respectively) and Foxp3-diptheria toxin receptor(DTR) mice from Dr. Alexander Rudensky (Memorial Sloan Kettering Cancer Center). After general anesthesia and mechanical ventilation, 8–10 week-old male mice underwent either sham surgery or left coronary artery ligation to induce large MI and ischemic HF, as previously described.^{3, 5, 6, 16–18}

Based on pilot studies, Treg ablation in Foxp3-DTR mice was induced by i.p. diphtheria toxin (DT; 20 μg/kg/d) at 28 and 29 d after ligation or sham surgery, and again at 38 and 39 d with a dose of 10 μg/kg/d (vehicle control). In some studies, DT administration was accompanied by co-administration of the small molecule CCR5 antagonist maraviroc, 10 mg/kg (PBS-vehicle control) i.p. daily from 28 to 56 d post-MI.¹⁹ For antibody-mediated Treg ablation, C57BL/6 mice received anti-CD25 (or IgG control) (100 μg/mouse) initiated on day 28 post-MI and repeated every 10 d thereafter for a total of 3 doses. We ensured that HF mice exhibited comparable post-MI LV remodeling by echocardiography 25–27 d post-MI prior to randomization to the respective treatment groups.

Echocardiography.

M-mode and 2-D echocardiography were performed under 1–2% inhaled isoflurane anesthesia (with 100% supplemental O2) and continuous ECG monitoring using a VisualSonics Vevo770 High-Resolution System and 30 MHz RMV707B scanhead.^{6, 20, 21} A heated, bench-mounted adjustable rail system was used, with body temperature maintained at 37.0°C ± 0.5°C and heart rate 500 ± 50 bpm. Imaging was performed in the parasternal long-axis (PLAX) and short-axis views. LV chamber volume in end-diastole (EDV) and end-systole (ESV) was determined using the area-length method from PLAX images captured in ECG-gated Kilohertz Visualization (EKV) mode. LV ejection fraction was used to index systolic function and calculated as [(EDV-ESV)/EDV]*100.

Immune cell isolation and flow cytometry.

Live mononuclear cells were isolated from the peripheral blood, heart, spleen, and LNs and processed for flow cytometry as previously described.^{3, 5, 6, 20, 21} All cells were fixed with 1% paraformaldehyde for flow cytometric analysis. Cells were initially stained for membrane markers by incubating with antibodies against CD3-FITC (BD Biosciences), CD4-Qdot 650NC/eVolve 605/PE/Cy7 (eBioscience), and TNFR1 (Santa Cruz Biotechnology, rabbit polyclonal sc7895) for 1 h. Anti-TNFR1 was then conjugated with secondary goat anti-rabbit antibody attached to Alexa Fluor 405 for 30 min. Cells were then permeabilized with 0.5% Tween-20 for 20 min followed by intracellular staining with anti-Foxp3-APC, TNFα-PE, IFNγ-eFluor 450, IL-4-PerCP-eFluor 710, and IL-17-Alexa Fluor 700 (eBioscience) for 45–60 min. All staining was performed on ice to maintain cell morphology. Flow cytometric data was acquired using a BD FACSdiva LSR-II Flow Cytometer and was analyzed using FlowJo software v10.0.6. In select studies, we evaluated the intravascular cell contribution to the cardiac cell isolate. One day post-MI, BV605-conjugated anti-CD45 antibody (3 μg/mouse; Biolegend) was given intravenously, and 3 min later, hearts were harvested and mononuclear cells were isolated either before or after modified Langendorff retrograde coronary perfusion with PBS for 5 mins. The cell isolate was stained with PE/Cy7-conjugated anti-CD45 antibody and the frequency of BV605-CD45⁺ cells among total PE/Cy7-CD45⁺ cells was measured by flow cytometry.

Cytokine analysis.

Peripheral blood was coagulated on ice and centrifuged at 3000g to separate serum. Mouse cytokine bead array for Th1/Th2/Th17 cytokine analysis (BD Biosciences) was used as per the manufacturer’s protocol with results presented as MFI.

Gene expression analysis.

Total mRNA from the LV remote zone (~10 mg) or sorted cells was extracted using Trizol reagent and gene expression was measured as described previously.^{3, 5, 6, 16, 17} Supplemental Table 1 lists the forward and reverse primer sequences used.

Histological analysis.

Fibrosis and capillary density were measured using Masson’s Trichrome and isolectin IB4 staining, respectively, as previously described.^{3, 6, 16, 18} Fibrosis was quantified in the LV border zone in an automated fashion using Metamorph software v6.3r5 (Molecular Devices). Collagen (blue) staining threshold was initially adjusted so that all fibrotic areas were captured, and tissue fibrosis was then determined from 4–6 high-power fields per section, expressed as percent of total cross-sectional area. Myocyte membranes were stained using Alexa Fluor (AF) 555-conjugated wheat-germ agglutinin (WGA) (Invitrogen). Myocyte apoptosis was evaluated using the DeadEnd Fluorometric TUNEL System (Promega).^{3, 18} For cardiac Treg quantitation, LV sections were embedded in OCT compound (Tissue-Tek) and then kept at −80°C. 4 μm LV sections were fixed with ice cold acetone, rehydrated in PBS and stained with anti-mouse CD4 (1:50, GK1.5 cat#14–0041082, secondary anti-rat AF555, 1:100) and FoxP3 (1:50, 5H10L18 cat#700914, secondary anti-rabbit AF488, 1:100). Nuclei were stained with DAPI. CD4⁺Foxp3⁺ cells were counted in 5 high power fields (60×) per LV region (remote, border, and infarct zone). Images were acquired using a Nikon A1 confocal microscope.

BrdU pulse studies.

To assess Treg proliferation in vivo, BrdU (2 mg/mouse i.p.) was injected 2–3 h prior to euthanasia. After euthanasia, cells were harvested and fixed for flow cytometric analysis of BrdU⁺ Tregs, labeled with anti-BrdU-eFluor 450 (eBioscience).

In vitro tube formation assay.

Matrigel (50 μL) was layered in 96-well plates and incubated at 37°C for 30 min followed by plating of mouse coronary ECs (MCECs), 50,000 cells/well in supplemented endothelial basal media. Splenic CD4⁺ T cells were isolated from naïve Foxp3-DTR mice (pre-treated with i.p. thioglycollate 2.5 mL, 5 d prior to harvest), Foxp3-DTR HF mice, and TNFR1^−/− HF mice by negative selection using the Miltenyi CD4⁺ T-cell isolation kit. Subsequently, CD4⁺CD25⁺GFP⁺ Tregs (or CD4⁺CD25⁺ TNFR1^−/− Tregs) and CD4+CD25^–GFP^– Teffs (or CD4⁺CD25^– TNFR1^−/− Teffs) were flow-sorted and added to the wells in triplicate (50,000 cells/well) simultaneously with the plating of MCECs in the wells. After 16 h of co-incubation, tube formation (total length) was measured using Metamorph software. To isolate paracrine effects, MCECs (200,000 cells) were plated on layered Matrigel (200 μL) in 24-well plates and an equivalent number of Tregs/Teffs from naïve Foxp3-DTR mice were plated on 3.0 micron Transwell inserts, with tube formation measured after 16 h. To measure MCEC gene expression, MCECs (200,000 cells) were co-cultured for 16 h in 24-well plates with an equal number of Tregs. Flk1⁺ MCECs were subsequently flow sorted for gene expression analysis; MCEC apoptosis was assessed by anti-annexin V antibody (Invitrogen) staining using flow cytometry.

Treg suppression assay.

Splenocytes were harvested from Foxp3-DTR mice and total CD4⁺ T-cells were isolated using the Miltenyi CD4⁺ T-cell isolation kit. Using FACS, CD4⁺CD25⁺GFP⁺ Tregs and CD4⁺CD25^–GFP^– Teffs were then separated. Teffs were membrane labeled with eFluor 670 dye (eBioscience) using the manufacturer’s instructions and cultured either alone or with Tregs in different ratios for 72 h in the presence of CD3/CD28 coated beads (Miltenyi). CD4⁺ Teff proliferation, in the presence or absence of Tregs, was subsequently determined by dye dilution.

Statistical analysis.

All data are mean ± SD. Group variances were compared using the Brown-Forsythe test, whereas normality was assessed using the D’Agostino–Pearson test. For statistical comparisons of 2 groups, 2-tailed paired t-test or unpaired t-test with equal or unequal variance was used. For comparison of more than 2 groups, 1-way or 2-way ANOVA with either Tukey’s or Bonferroni post-test was used. The specific approaches applied are stated in the figure legends. All analyses were done using GraphPad Prism version 7.0a. A p < 0.05 was considered significant.

RESULTS

Mice exhibit ischemic cardiomyopathy and chronic HF after coronary ligation.

Mice reliably demonstrated signs of ischemic cardiomyopathy and chronic HF 8 w after non-reperfused MI.^{3, 6} As compared with sham-operated mice, HF mice exhibited marked LV dilatation (increased end-diastolic and end-systolic volume [EDV and ESV]) and depressed LV ejection fraction (EF) by echocardiography (Supplemental Figure 1A), and greater heart, LV, lung, and spleen weight by gravimetry (Supplemental Figure 1B), indicative of cardiac hypertrophy, pulmonary congestion, and splenic remodeling.

Tregs are expanded in chronic ischemic HF.

Using flow cytometry, we assessed the levels of cardiac CD3⁺CD4⁺Foxp3⁺ Tregs in C57BL/6 mice up to 8 w after coronary ligation. Representative flow cytometry gates for the identification of peripheral blood Tregs and cardiac CD3⁺CD4⁺ T-cells are depicted in Supplemental Figure 2. Subsequent identification and quantitation of cardiac Tregs are shown in Figure 1A. Tregs were recruited to the heart within 1 d post-MI (~4-fold increase) in ligated hearts versus sham hearts, and increased dramatically by 3 d (~57-fold). Tregs remained elevated ~3-fold at 7 d, normalized by 2 w, and subsequently re-expanded in chronic HF (8 w post-MI) to levels ~7-fold over sham. By comparison, naïve hearts contained a very small tissue-resident Treg population. At 8 w post-MI, the failing heart also exhibited increased gene expression of several chemokines, chemokine receptors, and adhesion molecules that drive Treg recruitment and activation,^22–27 including CCL22, CXCL5, CXCL10, CXCL13, CX3CL1, CCR7, VCAM-1 and ICAM-1 (and strong trends toward increased CCL19 and CCL21) (Supplemental Figure 3). In blood, both Treg frequency (percent CD4⁺ T-cells) and absolute levels were comparable between sham and MI mice at 1 w post-MI (Figure 1B). However, at later time points, circulating Treg frequency was increased in HF mice, with absolute levels augmented ~2-fold over sham at 8 w post-MI.

Figure 1.

Representative flow cytometry scatter plots and group quantitation for CD3⁺CD4⁺Foxp3⁺ Tregs in the heart (A) and blood (B) from sham and HF mice at the time points indicated. Naïve non-surgical control (NS Ctrl) also shown. (C) Representative low and high magnification confocal images of CD4⁺ (red) and Foxp3⁺ (green) immunostaining in sham and HF hearts (border, remote and scar zones 8 w post-MI) and quantitation of CD4⁺Foxp3⁺ cells. DAPI (blue) was used to label nuclei. (D) Representative flow cytometry scatter plots and quantitation for Tregs in the spleen and lymph nodes (LNs) from mice 8 w after MI or sham operation. (E) Representative flow plots for BrdU in cardiac Tregs (Left) and quantitation of total BrdU⁺ Tregs in the heart, blood, spleen and LNs 8 w after MI or sham operation. (F) LV (remote zone) and splenic gene expression of IL-2 and IL-6, and IL-2 and IL-6 serum levels, in sham and HF mice (8 w post-MI). Statistical comparisons: 2-tailed unpaired t-test at each time point except for A and B (right panel), 2-way ANOVA and Bonferroni post-test performed after (A) or without (B) logarithmic data transformation. N = 4–10/group. *p < 0.05, **p<0.01, ***p<0.001, ****p<0.0001 vs. sham.

To define the contribution of circulating Tregs to the measured cardiac Treg population, we performed intravascular CD45 labeling in vivo immediately prior to cardiac harvest, with or without subsequent ex vivo retrograde coronary perfusion, in mice 24 h after MI. As shown in Supplemental Figure 4, even under these conditions of high tissue inflammation, intravascular leukocytes comprised <5% of the total cardiac mononuclear cell population, and isolation purity was similar between perfused and non-perfused hearts. Moreover, cardiac CD4 and Foxp3 immunostaining revealed a ~4–7-fold increase in CD4⁺Foxp3⁺ cells in the remote, border and infarct (scar) zones of failing hearts as compared with sham (Figure 1C), confirming the flow cytometry results indicative of tissue Treg expansion in HF.

In HF, a pathological cardiosplenic axis contributes to LV remodeling.³ Therefore, we assessed Treg abundance in the spleen and heart-draining mediastinal lymph nodes (LNs) in chronic HF (8 w post-MI). As shown in Figure 1D, both Treg frequency and number in the spleen and LNs were significantly increased in HF mice versus sham. Circulating Treg populations are dynamic and characterized by a balance between proliferation and apoptosis; Treg survival is linked to the availability of IL-2²⁸ and Treg plasticity to IL-6.¹² To evaluate Treg proliferation, we administered BrdU to mice 2–3 h prior to tissue collection. Absolute numbers of BrdU⁺ proliferating Tregs were increased in the heart, blood, spleen, and LNs of HF mice versus sham (Supplemental Figure 5), with increased frequency (% total Tregs) in the heart, blood and spleen (Figure 1E). Moreover, cardiac Tregs in failing hearts exhibited considerably higher proliferative rates than the overall CD45⁺ cardiac cell population (Supplemental Figure 5) and Tregs in sham-operated hearts in turn exhibited greater proliferative rates than Tregs in lymphoid organs and blood. This was accompanied by significantly increased IL-2 expression in the spleen, IL-6 expression in the spleen and heart, and serum levels of IL-6 in HF (Figure 1F).

HF Tregs exhibit pro-inflammatory features and diminished immunomodulatory capacity.

Under suitable conditions, Foxp3⁺ Tregs are capable of expressing Th-1, Th-2 and Th-17 cytokines and exhibiting plasticity towards a pro-inflammatory phenotype.^{12, 29} Accordingly, we measured intracellular tumor necrosis factor-α (TNFα), interferon-γ (IFNγ), IL-4, and IL-17 levels in Tregs by flow cytometry. As shown in Figure 2A, while circulating TNFα⁺ Treg frequency was comparable between sham and HF mice early (1 w) post-MI, at 4 w post-MI and beyond, blood TNFα⁺ Tregs were robustly increased in HF mice. TNFα⁺ Treg frequency was also significantly increased in the heart and spleen, together with a nearly 6-fold increase in total TNFα⁺ Tregs in heart-draining LNs in HF (Supplemental Figure 6A). Additionally, at 1 w post-MI, circulating Tregs increased expression of both Th1 (IFNγ) and Th2 (IL-4) cytokines as compared with sham; however, in chronic HF 8 w post-MI, only IFNγ⁺ Tregs remained increased (~50% higher) in HF mice (Supplemental Figure 6B-C). There were no differences in circulating IL-17⁺ Tregs between the groups at either time point (Supplemental Figure 6D). Taken together, these results indicate that after the healing phase post-MI, Tregs switch to a pro-inflammatory Th1-like phenotype in chronic HF.

Figure 2.

(A) Representative peripheral blood scatter plots and quantitative data for CD3⁺CD4⁺Foxp3⁺TNF⁺ Tregs in the blood, heart and spleen of sham and HF mice (heart and spleen measured 8 w post-operation); n = 4–7/group. (B) Single cell images for HF splenic Tregs expressing Foxp3 (pink) and CD4 (red), exhibiting variable expression of TNFR1 (green) and TNF (yellow). Also shown are DAPI nuclear staining (blue) and brightfield images. (C) Scatter plots and quantitation of cardiac TNFR1⁺ Tregs at various time points post-operation (Top). Frequency of circulating TNFR1⁺ Tregs in sham and HF mice at the time points indicated, and in the spleen and LNs 8 w post-operation (Bottom); n = 4–8/group. (D) In vitro Treg immunosuppression assay using splenic Tregs and Teffs isolated from naïve, sham and HF Foxp3-DTR (untreated) mice. Each experiment was conducted 3–4 times in triplicate using cells isolated from 3–4 mice. Statistical comparisons: 2-tailed unpaired t-test at each time point except for C (upper right panel) and D, 2-way ANOVA and Bonferroni post-test performed after (C) or without (D) logarithmic data transformation. *p < 0.05, **p <0.01, ***p<0.001, ****p<0.0001 vs. sham. ^#p < 0.05, ^###p<0.001 vs. naive.

TNFα exerts divergent effects on cardiac remodeling after MI that are directly referable to its cell surface receptors, TNFα receptor (TNFR) 1 and 2.¹⁶ Whereas TNFR1 exerts pro-inflammatory and pro-hypertrophic effects, TNFR2 is protective and counterbalances these detrimental responses. TNFα also has divergent receptor-dependent effects on Treg function. On one hand, it augments Treg immunomodulatory function via TNFR2.³⁰ On the other hand, in inflamed synovial tissues, TNFα promotes Treg dysfunction by inducing protein phosphatase-1-mediated Foxp3 dephosphorylation via nuclear factor (NF)-κB,³¹ a transcription factor primarily activated by TNFR1 in HF.¹⁶ Therefore, we next investigated whether Treg TNFR expression was altered during the progression of HF.

Tregs preferentially express TNFR2, which is necessary for proper Treg immunosuppressive function.^{32, 33} Consistent with prior studies, Tregs in both HF and sham mice robustly expressed TNFR2⁺ (generally >98% of circulating CD4⁺Foxp3⁺ cells) up to 8 w after MI (Supplemental Figure 6E). TNFR1-deficient Tregs are strongly immunosuppressive.³⁴ However, presumably due to low lymphocytic TNFR1 expression, prior studies have failed to detect TNFR1 on Tregs using fluorochrome-conjugated antibodies.^{32, 33} To improve flow cytometric TNFR1 detection, we used a primary rabbit anti-mouse TNFR1 antibody¹⁶ and a secondary Alexa-fluor 405-conjugated goat anti-rabbit antibody. We stained splenocytes from HF mice for CD4, Foxp3, TNFα and TNFR1, and nuclei with DAPI, and visualized stained cells using imaging flow cytometry (ImageStream). As shown in Figure 2B, we readily identified TNFR1^–TNFα^–, TNFR1⁺TNFα^–, and TNFR1⁺TNFα⁺Tregs.

As compared with sham, TNFR1⁺ Tregs were significantly increased in infarcted hearts by 1 d post-MI, peaking at 3 d, normalized by 2 w, and then re-expanded in chronic HF (8 w post-MI) to ~6-fold higher levels over sham (Figure 2C). This was accompanied by changes in the frequency of TNFR1⁺ Tregs in the heart. In naïve hearts, ~60% of the total Treg population expressed TNFR1; this initially decreased significantly to ~20% at 1 d post-MI, normalized by 3 d post-MI, and increased at 8 w to ~90% of cardiac Tregs (Supplemental Figure 7A). Moreover, TNFR1⁺ Tregs in failing hearts expressed increased levels of both TNFα and IFNγ as compared with TNFR1^– Tregs, in contrast to TNFR1⁺ Tregs from 3 d post-MI hearts, which exhibited the reverse relationship (Supplemental Figure 7B). In the blood, TNFR1⁺ Tregs exhibited biphasic kinetics similar to TNFα⁺ Tregs, with diminished frequency but no change in absolute counts at 1 w post-MI versus sham (Figure 2C, Supplemental Figure 6F), and with augmented frequency and absolute levels (~5-fold increase) in chronic HF. The spleen and LNs in HF mice also contained increased populations of TNFR1⁺ Tregs. Hence in HF, Tregs increase surface TNFR1 expression with greater propensity for a pro-inflammatory phenotype and Treg dysfunction, as compared with TNFR1⁺ Tregs early after MI.

We then measured immunosuppressive capacity of Tregs from sham and HF mice. Using FACS, we isolated splenic CD4⁺CD25⁺eGFP⁺ Tregs and CD4⁺CD25^–eGFP^– Teffs from naïve, sham-operated, and HF Foxp3-DTR transgenic mice (8 w post-MI). Foxp3-DTR mice have human DTR cDNA fused to enhanced green fluorescent protein (eGFP) and an internal ribosome entry site in the 3’ UTR of the Foxp3 gene.⁹ These mice exhibit normal baseline phenotype with all Foxp3⁺Tregs labeled with eGFP. Teffs labeled with eFluor 670 were co-cultured with different ratios of Tregs (1:0, 1:0.25, 1:0.5 and 1:1) for 72 h and Teff proliferation was measured by dye dilution using flow cytometry. As shown in Figure 2D, HF Tregs failed to suppress Teff proliferation as compared with Tregs isolated from either sham-operated or naïve mice. Taken together, these findings establish that Tregs in HF are paradoxically pro-inflammatory and have lost immunomodulatory capacity.

Periodic Treg depletion in HF reverses LV remodeling.

We next performed DT-induced Treg depletion⁹ in Foxp3-DTR sham and HF mice, as illustrated in Supplemental Figure 8A. In pilot experiments, 20 μg/kg DT i.p. resulted in ~60% Treg depletion by 2 d with complete reconstitution by ~7–10 d (Supplemental Figure 8B). At 25 d post-MI, HF mice with comparable LV remodeling by echocardiography (Supplemental Figure 8C) were randomized to receive either DT or PBS over 4 w (sham mice similarly treated). As shown in Figures 3A–3B, pathology, short-axis histological sections, and gravimetry revealed cardiac and LV enlargement in PBS-treated HF mice that was substantially less in DT-treated HF mice. Echocardiography revealed significant LV dilatation (increased EDV and ESV) and dysfunction (decreased EF) in HF-PBS mice as compared with sham (Supplemental Figure 8C), that progressed over the 4-w treatment period (Figures 3C–3D, Supplemental Figure 8D). In contrast, HF-DT mice, with similar degrees of LV remodeling at 4 w post-MI, exhibited improvements in EDV, ESV, and EF (indicative of reverse remodeling) to levels significantly better than HF-PBS mice (Figures 3C–3D, Supplemental Figure 8D). Heart rate (HR) was comparable among all treatment groups. Periodic Treg depletion also reduced circulating levels of several pro-inflammatory (IL-17, IL-2 and IL-6) and pro-fibrotic (IL-6 and IL-4) cytokines in HF mice (Supplemental Figure 8E), but did not impact cardiomyocyte apoptosis in the remote zone of failing hearts (Supplemental Figure 8F). Additionally, both sham-operated Foxp3-DTR mice and WT C57BL/6 mice administered DT exhibited no changes in LV structure or function (Supplemental Figures 8C-D, 9), whereas WT HF mice given DT exhibited progressive LV remodeling and dysfunction (Supplemental Figure 9) similar to PBS-treated Foxp3-DTR HF mice (Figure 3D). These controls established that modulation of LV remodeling in Foxp3-DTR HF mice was specifically related to dysfunctional Treg depletion and the HF state, rather than non-specific effects related to DT or induced immune cell death.

Figure 3.

Representative whole hearts and transverse LV trichrome-stained sections (A) and gravimetric data normalized to tibia length (B) from sham and HF Foxp3-DTR mice treated with either DT or PBS vehicle as outlined in Figure S8A. N = 7–14/group. (C) Representative parasternal long-axis 2D echocardiograms at end-diastole and end-systole from Foxp3-DTR mice 8 w post-MI (HF) or sham operation after treatment with DT or PBS. (D) Group data for end-diastolic and end-systolic volume (EDV and ESV) and ejection fraction (EF) from Foxp3-DTR HF mice 4 and 8 w post-MI during 4 w of treatment with either DT or PBS. Statistical comparisons: B, 2-way ANOVA (Bonferroni post-test); D, 2-tailed paired t-test. N = 13–14/group. *p < 0.05, **p<0.01, ***p<0.001.

Periodic Treg depletion in HF reduces cardiac fibrosis and improves neovascularization.

As shown in Figure 4A, cardiac peri-infarct zone interstitial fibrosis (trichrome staining) in HF-PBS mice was markedly increased versus sham-PBS mice. While there were no changes in fibrosis in sham-DT hearts, DT treatment significantly reduced fibrosis in failing hearts. Moreover, remote zone LV gene expression revealed a marked increase in collagen-I and collagen-III mRNA in HF-PBS hearts versus sham, with loss of this upregulation in HF-DT hearts (Figure 4B). To evaluate myocardial neovascularization, we performed isolectin IB4 and WGA staining. HF-PBS mice displayed reduced myocardial capillary:myocyte ratio (Figure 5A), increased myocyte cross-sectional area (Figure 5A) and decreased overall capillary number (Supplemental Figure 10) versus sham mice treated with either PBS or DT (which were comparable). Importantly, Treg depletion in HF-DT mice normalized both capillary:myocyte ratio and capillary number to levels observed in sham hearts, and reduced myocyte size, indicating enhanced neovascularization and alleviation of pathological hypertrophy.

Figure 4.

(A) Representative Masson’s trichrome stains and group quantitation of peri-infarct interstitial fibrosis (% area) in Foxp3-DTR sham and HF hearts treated with either DT or PBS as in Figure 3. (B) LV (remote zone) collagen-I and collagen-III gene expression in the same groups. Statistical comparisons: 2-way ANOVA (Bonferroni post-test). N = 5–6/group; *p < 0.05, **p<0.01, ***p<0.001.

Figure 5.

(A) Representative cardiac isolectin (green) and WGA (red) staining, and quantitation of capillary:myocyte ratio and myocyte cross-sectional area in remote zone LV from sham and HF Foxp3-DTR mice given either DT or PBS vehicle. (B) Representative flow scatter plots for CD34⁺Flk1⁺ CACs and quantitation of circulating levels at 1 and 4 w after initiating DT or PBS as per the protocol in Figure S8A (i.e., 5 and 8 w post-MI or sham operation). (C) Frequency of CD34⁺Flk1⁺ CACs in the heart, spleen, LN and bone marrow (BM) in the same groups at 8 w. (D) Absolute CAC counts in blood (5 and 8 w post-MI), and heart, spleen, LN, and BM of Foxp3-DTR HF mice given DT or PBS at 8 w post-MI. (E) LV gene expression (remote zone) of CCR5 and CCL5 in sham and HF Foxp3-DTR mice treated with either DT or PBS. Statistical comparisons: A-C and E, 2-way ANOVA (Bonferroni post-test); D, 2-tailed unpaired t-test. N = 5–9/group; *p < 0.05, **p<0.01, ***p<0.001.

We next evaluated the effects of Treg depletion on CD34⁺Flk1⁺ circulating angiogenic cells (CACs),³⁵ which are thought to contribute to functional neovascularization.^{35, 36} As shown in Figure 5B, blood CACs were reduced in HF-PBS mice as compared to sham mice, but were expanded ~5–6-fold by in HF-DT mice 1 w after the first DT dose, with maintained robust expansion in HF-DT mice even after Treg reconstitution to normal levels at 8 w post-MI. DT-mediated Treg depletion did not change blood CACs in Foxp3-DTR sham mice, indicating specificity for the HF state. Analysis of tissue mononuclear cells (Figure 5C) similarly revealed a ~2–3-fold increase in CD34⁺Flk1⁺ cell frequency in HF-DT hearts as compared with HF-PBS hearts, without changes in sham hearts. Interestingly, DT treatment also increased CACs in the spleen and mediastinal LNs (which were decreased at baseline in HF) in both sham and HF mice versus their respective PBS treated counterparts, without differences in the bone marrow of any group. Further assessment of absolute CAC number in HF-PBS and HF-DT mice confirmed that periodic Treg depletion expanded CAC populations in the blood, failing heart, and spleen, with a trend toward CAC increase in mediastinal LNs (Figure 5D). Moreover, DT administration in WT animals did not change blood CACs from baseline in sham mice and failed to increase CACs in HF mice (Supplemental Figure 11), excluding a non-specific effect of DT on CACs.

C-C chemokine ligand 5 (CCL5) signaling via CC receptor 5 (CCR5) plays an essential role in the tissue recruitment of CACs.³⁷ As shown in Figure 5E, cardiac CCL5 gene expression (remote zone) was increased nearly 11-fold, and CCR5 expression nearly 3-fold, in HF-DT mice as compared with HF-PBS mice and sham-PBS mice. Cardiac CCL5 expression was also significantly increased in sham mice treated with DT. To define the role of CCR5 in CAC responses, Foxp3-DTR HF mice with comparable post-MI LV remodeling were randomized and treated with the small molecule CCR5 antagonist maraviroc (10 mg/kg) or PBS-vehicle daily during DT administration (Supplemental Figure 12A-B). CAC frequency and absolute number were significantly decreased in the blood at 5 and 8 w post-MI, and in the heart and spleen (absolute number) at 8 w post-MI in HF-DT-maraviroc mice as compared with HF-DT-vehicle mice (Supplemental Figure 12C), without changes in CAC number in the BM or LNs. Reverse LV remodeling was not observed in HF-DT-maraviroc mice (Supplemental Figure 12D) suggesting that CCR5-dependent CAC mobilization to failing hearts contributes importantly to the benefits of periodic Treg depletion. Heart rate (HR) was similar for all treatment groups (Supplemental Figure 12E).

Tregs reconstituted after Treg depletion in HF regain immunomodulatory capacity.

Circulating HF Tregs at 8 w post-MI reconstituted after periodic Treg ablation (Figure 6A) displayed decreased TNFR1 expression without change in TNFR2 expression (Figure 6B). Circulating total CD4⁺ T-cells (but not CD8⁺ T-cells) decreased after Treg depletion, suggesting re-establishment of Teff suppression and reduced systemic inflammation (Figure 6C). Indeed, in vitro Teff proliferation assays revealed restoration of immunosuppressive capacity in reconstituted splenic Tregs, which significantly inhibited Teff proliferation as compared with untreated HF splenic Tregs (Figure 6D). Although reconstituted Tregs had less TNFR1 expression, dysfunctional immunomodulation was not specifically related to TNFR1, as splenic Tregs from TNFR1^−/− mice with HF (8 w post-MI) also failed to suppress Teff proliferation, analogous to wild-type HF Tregs (Figure 6D). Notably, TNFR1^−/− HF Teff cells exhibited similar basal proliferation rates (52.9±3.7%) as WT HF Teffs (49.8±24.8%) indicating that TNFR1 loss in Teffs did not directly impact their proliferative capacity. This suggested that alternate mechanisms underlie the loss of Treg immunosuppression. In this regard, a recent elegant study³⁸ demonstrated that inflammatory signals (via Toll-like receptors) metabolically reprogram Tregs towards glycolysis via upregulation of the glucose transporter Glut1, and that Glut1 expression is sufficient to both augment Treg proliferation and diminish Treg suppressive function. As shown in Supplemental Figure 13, circulating Glut1^hiTregs were significantly increased in HF mice versus sham mice. Hence, in addition to pro-inflammatory features, HF Tregs exhibit metabolic features associated with proliferation and reduced immunomodulatory capacity. While TNFR1 expression is a specific biomarker for Treg dysfunction, it appears not to be required for the development of immunosuppressive alterations.

Figure 6.

(A) Frequency of circulating Tregs in sham and HF Foxp3-DTR mice treated with either DT or PBS at 8 w post-MI or sham operation; n = 4–7/group. Frequency of TNFR1⁺ and TNFR2⁺ Tregs (B), and CD4⁺ and CD8⁺ T-cells (C), in HF Foxp3-DTR mice treated with DT or PBS at the same time point; n = 4–7/group. (D) In vitro Treg immunosuppression assay using splenic Tregs and Teffs isolated from untreated HF Foxp3-DTR mice, HF Foxp3-DTR mice treated with DT, and untreated TNFR1^−/− HF mice. All cells were isolated 8 w post-MI, and the data for HF Foxp3-DTR mice are the same as in Figure 2D. Each experiment was conducted 3–4 times in triplicate using cells isolated from 3–4 mice. Statistical comparisons: A and D, 2-way ANOVA (Tukey’s post-test); B and C, 2-tailed unpaired t-test. *p < 0.05, **p <0.01, ***p<0.001 vs HF or as depicted; ^##p <0.01, ^###p<0.001 vs HF TNFR1^-/−.

Antibody-mediated Treg depletion in HF alleviates LV remodeling.

To complement studies of genetically-mediated Treg depletion, and to explore a potentially translatable therapeutic approach, we performed Treg depletion studies using anti-CD25 antibody.¹¹ While Tregs classically express CD25, effector T-cells may also upregulate CD25 upon activation.⁷ Hence, this approach is less specific than the genetic approach, although anti-CD25 remains effective for Treg depletion in mice.³⁹ C57BL/6 HF mice with comparable LV remodeling 4 w post-MI received either anti-CD25 or IgG isotype control every 10 d (Supplemental Figure 14A-B). Notably, anti-CD25 significantly decreased the frequency (%CD45) of CD4⁺Foxp3⁺ Tregs but did not significantly change circulating CD4⁺Foxp3^– T effector cells (Supplemental Figure 14C). As shown in Figure 7A, HF mice treated with IgG exhibited progressively worsening LV dilatation and persistently reduced LVEF. In contrast, anti-CD25 initiated 4 w post-MI prevented progression of LV remodeling over the treatment period, with no increase in EDV or ESV. HR was similar between groups at both time points (Supplemental Figure 14D). As with genetic Treg depletion in Foxp3-DTR HF mice, anti-CD25 in wild-type HF mice significantly increased both capillary:myocyte ratio (Figure 7B) and overall capillary density (Supplemental Figure 14E) as compared with IgG-treated HF mice, and was accompanied by normalization of pro-angiogenic VEGFa and CCR5 gene expression in the heart (Figure 7C). Anti-CD25 also increased CAC frequency and absolute levels in blood at 5 and 8 w post-MI, and in the heart and spleen at 8 w post-MI (Figure 7D). We also observed a significant increase in CAC numbers at 8 w post-MI in BM after anti-CD25 treatment (Figure 7D). Additionally, as compared with IgG-treated HF mice, anti-CD25 HF mice exhibited decreased cardiac expression of the pro-inflammatory mediators IFNγ, IL-17 and TNFR1, and pro-fibrotic factors IL-10 and IL-4 (Supplemental Figure 14F). TGFβ and connective tissue growth factor (CTGF) gene expression was similar between the HF groups. Hence, unlike the acute MI period,^{10, 11} antibody-mediated Treg depletion in ischemic cardiomyopathy ameliorated pathological LV remodeling and inflammation.

Figure 7.

(A) Quantitative group data at the post-MI time points indicated for LV EDV, ESV, and EF in C57BL/6 HF mice treated with either anti-CD25 or isotype IgG control as per the protocol in Figure S14A. Representative cardiac isolectin (green) and WGA (red) staining, and quantitation of capillary:myocyte ratio (B) and LV remote zone gene expression of VEGF-A and CCR5 (C) in hearts from sham and HF mice (8 w post-MI) treated with either isotype IgG or anti-CD25 antibody. (D) Frequency and absolute counts of CD34⁺Flk1⁺ CACs in blood (5 and 8 w post-MI), and in the heart, spleen, BM, and LN at 8 w post-MI in anti-CD25- or IgG-treated HF mice. Statistical comparisons: A, 2-tailed paired t-test; B and C, 1-way ANOVA (Tukey’s post-test); D, 2-tailed unpaired t-test at each time point. N = 4–8/group; *p < 0.05, **p <0.01 vs sham or as depicted.

HF Tregs exert augmented TNFR1- and cell contact-dependent anti-angiogenic effects.

Given that Tregs modulate neovascularization in failing hearts, we directly evaluated anti-angiogenic effects of Tregs using MCEC tube formation assays. Mouse splenic CD4⁺ T-cells were isolated using the Miltenyi CD4 isolation kit, and CD4⁺CD25⁺ Tregs and CD4⁺CD25^– Teffs were FACS sorted. MCECs were cultured alone or with Tregs and/or Teffs for 16 h and tube formation was quantified by image analysis (Figure 8A). Both naïve Tregs and Teffs independently inhibited tube formation as compared with MCECs alone (Figure 8B), but when cultured together, tube formation was not attenuated, suggesting Treg-mediated suppression of the anti-angiogenic effects of Teffs. In contrast, HF CD4⁺ Teffs had no effect on tube formation whereas HF Tregs profoundly inhibited angiogenesis (more than 2-fold versus naïve Tregs) and also tended to inhibit (rather than improve) tube formation in the presence of Teffs. Studies using Transwell inserts established that T-cell-mediated anti-angiogenic effects required cell-to-cell contact with MCECs, whereas Treg-mediated suppression of the anti-angiogenic effects of Teffs did not, and was a paracrine rather than juxtacrine effect (Supplemental Figure 15A).

Figure 8.

(A) Representative brightfield images of tube formation by MCECs cultured alone and in the presence of CD4⁺CD25⁺eGFP⁺(Foxp3⁺) Tregs and CD4⁺CD25^–eGFP^–(Foxp3^–) Teffs isolated from Foxp3-DTR HF mice 8 w post-MI, or CD4⁺CD25⁺ Tregs and CD4⁺CD25^– Teffs from TNFR1^−/− HF mice. (B) Quantitation of total MCEC tube length under the various conditions indicated as detailed in the methods. Each experiment was conducted 3–4 times in triplicate using cells isolated from 3–4 mice. Statistical comparisons were performed using 1-way ANOVA (Tukey’s post-test). *p < 0.05, **p <0.01, ***p<0.001.

Notably, TNFR1^−/− HF Tregs and Teffs exhibited no anti-angiogenic effects whatsoever and failed to inhibit tube formation. Moreover, co-culture of TNFR1^−/− HF Tregs and Teffs together with MCECs resulted in significant pro-angiogenic effects and enhancement of tube formation (Figure 8B). In separate experiments, MCECs were cultured alone or with sham, HF, and TNFR1^−/− HF Tregs for 16 h and then flow sorted for gene expression analysis (Supplemental Figure 15B). As compared with control MCECs, sham and HF Tregs suppressed MCEC pro-angiogenic VEGFa expression without changing VEGFb, VEGFR1, VEGFR2, and CCR5. Sham Tregs also increased MCEC CCL5 expression, which did not occur with HF Treg co-culture. In contrast, TNFR1^−/− HF Treg co-culture significantly increased MCEC expression of VEGFa, VEGFb, VEGFR1 and CCL5. Moreover, annexin V staining revealed <1% frequency of annexin V⁺ apoptotic MCECs, indicating that Treg-mediated angiogenic effects resulted from changes in the MCEC transcriptome rather than induction of MCEC apoptotic cell death. Taken together, these data indicate that in addition to the modulation of CACs, dysfunctional Tregs in HF exert robust anti-angiogenic effects that require both Treg-EC contact and TNFR1 signaling.

DISCUSSION

In this study, we demonstrate several novel findings. First, in ischemic cardiomyopathy there is global expansion and proliferation of broadly dysfunctional Tregs that express pro-inflammatory Th1 cytokines and TNFR1, have diminished immunomodulatory capabilities, increased Glut1 expression, and potentiated anti-angiogenic and pro-fibrotic properties. Tregs expand and change their phenotype in a phasic manner after MI, beyond the early phase of wound healing that accompanies the long-term progression of HF. Moreover, in the failing heart, there is a favorable chemokine milieu for T-cell recruitment and expansion. Second, dysfunctional Tregs are essential for adverse cardiac remodeling, as selective Treg ablation in ischemic cardiomyopathy reversed LV remodeling and dysfunction, alleviating hypertrophy, fibrosis, and systemic inflammation, while enhancing tissue neovascularization. Importantly, as indicated by the Treg ablation studies, the pathological Treg phenotype is not fixed but is dynamic and reversible, as Tregs reconstituted after depletion exhibited restoration of immunosuppressive capacity and normalized TNFR1 expression. Third, Treg dysfunction is tightly linked to neovascularization in the failing heart, via both cell contact- and TNFR1-dependent anti-angiogenic effects, potentially related to modulated expression of pro-angiogenic mediators, and the mobilization and tissue infiltration of CACs, in part related to CCL5/CCR5 signaling. The results suggest that direct effects of heart-localized Tregs on cardiac ECs contribute to capillary rarefaction in HF. Taken together, these findings establish an essential role for dysfunctional Tregs in the pathogenesis of chronic ischemic HF, and suggests that restoration of Treg function may be a fruitful approach to therapeutic immunomodulation.

Prior work has established a beneficial role for Tregs in cardiac remodeling early after MI.^{11, 40, 41} Here, we assessed the in vivo role of Tregs in chronic ischemic HF. Unlike the acute MI period when the phasic activation and resolution of tissue inflammation are required for wound healing, ischemic cardiomyopathy is a state of inappropriately sustained inflammation that is tissue-injurious.² Prior work has shown that chronic inflammation in HF is characterized by augmentation of both innate and adaptive immune responses,^1–6 with local and systemic expansion of macrophages, DCs, and CD4⁺ effector and memory T-cells, and the elaboration of several pro-inflammatory protein mediators that have detrimental effects on the failing heart.

Broad immune activation in HF suggests compromise of intrinsic immunoregulatory networks. Tregs comprise an important immunomodulatory T-cell subset that expresses the transcription factor Foxp3,⁹ and constitutes 5–10% of circulating CD4⁺ T-cells.²⁹ They may be thymus-derived (‘natural’ Tregs) or peripherally induced from naïve CD4⁺ T-cells (‘induced’ Tregs) in response to specific microenvironmental cues.⁴² As Treg loss-of-function results in chronic autoimmune disease,⁹ these cells play critical roles in immunosuppression that theoretically should impart beneficial effects in pro-inflammatory states such as HF. Indeed, the adoptive transfer of normally-functioning Tregs ameliorates adverse remodeling induced by angiotensin II ⁴³ and acute non-reperfused MI in mice.^{10, 11, 41} Nonetheless, how endogenous Tregs respond to persistent inflammation in chronic HF post-MI and their role in LV remodeling is completely unknown.

Our studies establish for the first time that Tregs re-expand in chronic HF after MI, but unlike in the acute MI period, they are dysfunctional without immunosuppressive capacity, and induce cardiac pathology. Moreover, Tregs exhibit heightened proliferation, suggesting that in addition to peripheral conversion in lymphoid tissues and subsequent tissue recruitment, local renewal contributes importantly to cardiac Treg expansion. Interestingly, as compared to Tregs in blood and lymphoid tissue, Tregs in the heart exhibit greater proliferative rates even in the absence of cardiac injury, implicating local renewal as particularly important for Treg persistence the heart. Tregs are considered to achieve immunosuppression and peripheral tolerance primarily through inhibitory effects on effector T-cell function,⁷ either directly via inhibitory cytokine secretion, cytolysis, and metabolic disruption, and indirectly by inhibition of DC function. As we have previously shown that there is heightened systemic expansion of DCs, and CD4⁺ and CD8⁺ T-cells in ischemic cardiomyopathy,^{3, 6} progressive Treg dysfunction may represent a central mechanism for dysregulated innate and adaptive immune activation. Interestingly, normalization of Treg immunosuppressive function following circumscribed Treg depletion was accompanied by reduced circulating levels of CD4⁺ T-cells, but not CD8⁺ T-cells, suggesting that in ischemic HF, CD4⁺ T-cell populations are more closely regulated by Tregs, and by extension, DCs.

Tregs are classically considered stable in their functional behavior and phenotype.³¹ Emerging evidence indicates, however, that Tregs are dynamic, with an orchestrated balance between cell division and apoptosis, and phenotypic modulation in different tissue niches.²⁸ IL-2 is a key homeostatic regulator, and suppresses Treg apoptosis while enhancing Treg proliferation.²⁸ Inflammatory cytokines such as IL-6 promote Treg plasticity and antagonize immunosuppressive function,^{12, 31} whereas TNFα, in the presence of IL-2, boosts Treg proliferation via TNFR2 signaling.³⁰ As HF is characterized by increased levels of several pro-inflammatory cytokines, including TNFα, IL-6, and IL-2,² findings also consistent with our current and prior studies,^{3, 6, 17} it follows that the microenvironment in the failing heart would be conducive to both Treg expansion and phenotypic instability. Beyond pro-inflammatory cytokines, we have shown local upregulation of chemokines, chemokine receptors, and adhesion molecules that promote Treg trafficking, recruitment, and activation in HF (including CCL22, CXCL5, CXCL10, CXCL13, CX3CL1, CCR7, VCAM-1 and ICAM-1),^22–27 as well as augmented Treg Glut1 expression, which promotes glycolytic reprogramming, proliferation and immunomodulatory dysfunction.³⁸ Hence, multiple factors likely provide a platform for Treg dysfunction in HF.

Defective suppressive function of circulating Tregs has also been reported in a small study of human HF,⁴⁴ consistent with our results. However, that study also reported reduced Treg frequency, diverging from our observation of Treg expansion in murine ischemic HF. The reasons for this divergence are unclear, but may be related to several factors including heterogeneity of the study population (which included subjects both with and without prior MI); the impact of co-morbidities (e.g., hypertension, diabetes) on Tregs; and differences in HF disease stage. Importantly, absolute blood Treg number in humans was not quantified. Hence, further studies are needed to resolve the Treg profile associated with human HF.

Increased cellular TNFα, IFNγ, and TNFR1 expression accompanied the loss of Treg immunosuppressive function in chronic HF, consistent with a pro-inflammatory Treg transition. While TNFα can increase proliferation and activation of mouse Tregs via TNFR2,³⁰ it can also induce the opposite effects via TNFR1 in mice with type I diabetes and via canonical NF-κB signaling in human Tregs.^{31, 45} In ischemic HF, NF-κB is predominantly activated by TNFR1, which, in contrast to TNFR2, exacerbates pathological LV remodeling.^{16, 17} As HF Tregs exhibited increased TNFR1 and TNFα expression without changes in TNFR2, augmented TNFR1 signaling and greater intracellular crosstalk between TNFR1 and TNFR2⁴⁶ likely contribute to Treg dysfunction. Our studies with TNFR1 deficient HF Tregs established this to be the case, but with a primary impact of TNFR1 on Treg inhibition of angiogenesis rather than Treg immunosuppressive function.

Modulation of angiogenesis by Tregs is context specific. The tumor microenvironment recruits Tregs that promote immune tolerance and angiogenesis.⁴⁷ In contrast, following vascular ischemic injury, Tregs robustly inhibit neovascularization by suppressing pro-angiogenic macrophages and CD3⁺ T-cells.¹⁴ Moreover, via direct cell contact, Tregs inhibit EC angiogenesis and tube formation in vitro.¹³ Here, we show that HF Tregs exhibit potent direct antiangiogenic effects mediated by cell contact with cardiac ECs (as would occur in failing myocardium) and TNFR1 (which is upregulated in HF Tregs). Importantly, dysfunctional Tregs also appear to suppress the mobilization of CACs. In this regard, chronic HF is characterized by reduced numbers of CACs,⁴⁸ a phenomenon thought to contribute to deficient endothelial repair and neovascularization and shown to predict long-term cardiovascular mortality.⁴⁹ We have shown, to our knowledge for the first time, that circulating CACs, diminished in HF, are vigorously expanded upon the pulsed ablation and functional restoration of Tregs, with repletion of splenic and lymphoid reservoirs, increased CAC mobilization, and augmented CAC recruitment to the heart. Hence, the improvement of myocardial neovascularization after Treg depletion in HF likely reflects the combined effects of both diminished direct inhibitory Treg-EC intrinsic interactions and cardiac CAC expansion. The mechanisms linking Tregs to CAC recruitment were dependent, at least in part, on modulation of the CCR5/CCL5 axis, previously shown to be critical for CAC mobilization and trafficking during wound healing.³⁷

A key finding of our study is that periodic genetic Treg depletion in ischemic cardiomyopathy reverses both pathological LV remodeling and the altered immunomodulatory Treg phenotype. At the chamber level, selective Treg ablation resulted in smaller LV volume, less pathological hypertrophy, and improved systolic function; in human HF, these are important surrogate markers for improved clinical outcomes. Moreover, Treg depletion resulted in improved myocardial capillary density as discussed above, and a significant reduction in collagen-I and collagen-III expression and interstitial fibrosis, which has been independently linked to mortality risk in humans.⁵⁰ This may reflect in part the suppression of circulating CD4⁺ T-cells, as we have shown cardiac pro-fibrotic effects of splenic CD4⁺ T-cells from mice with HF in vivo after adoptive transfer.⁶ The benefits of Treg depletion appeared to be independent of effects on cardiomyocyte apoptosis in the failing heart. Notably, we previously demonstrated that HF macrophages directly depress cardiomyocyte contractility upon cell-to-cell contact;⁵ this raises the possibility that Tregs may exert analogous effects, although this was not specifically examined. Alleviation of cardiac remodeling in chronic HF was also achieved with anti-CD25 mediated Treg depletion, suggesting a possible approach to clinical translation.

Importantly, periodic Treg depletion normalized Treg immunosuppressive function and diminished circulating Teffs and systemic inflammation. This suggests that the immune system can endogenously recalibrate in chronic HF, and that circumscribed targeting of dysfunctional immune cell subsets may be an effective anti-inflammatory strategy. Presumably, ablation of pathological Tregs triggered the induction of new Treg populations with intact function and antigenic tolerance that then suppressed ongoing injurious Teff responses to ultimately repair remodeled myocardium. These results are diametrically opposite to the observed detrimental effects of Treg depletion on wound healing and remodeling in the acute MI period,¹¹ highlighting important differences in immune cell activation and inflammatory profiles during the acute and chronic phases of post-MI remodeling.

In summary, ischemic cardiomyopathy in mice is characterized by diffuse expansion and proliferation of Tregs with pro-inflammatory features, deficient immunosuppressive capacity, and heightened anti-angiogenic effects. Dysfunctional Tregs are required for adverse LV remodeling at the tissue and chamber level, with prominent pathological effects on hypertrophy, fibrosis, and tissue neovascularization, the last related to both TNFR1-dependent direct effects on ECs and Treg-mediated regulation, in part through CCL5/CCR5, of CAC mobilization and recruitment. Importantly, the immunocompromised Treg phenotype in HF exhibits plasticity and normalizes upon selective depletion and subsequent reconstitution with new Tregs. These findings identify dysfunctional Tregs as an appealing therapeutic target for the resolution of chronic inflammation in ischemic cardiomyopathy and HF.

CLINICAL PERSPECTIVE

What is new?

• Ischemic cardiomyopathy in mice is characterized by diffuse expansion and proliferation of dysfunctional regulatory T-cells (Tregs) with pro-inflammatory features, deficient immunosuppressive capacity, and heightened anti-angiogenic effects.

• Dysfunctional Tregs are required for adverse LV remodeling, and impart pathological effects on hypertrophy, fibrosis, and tissue neovascularization.

• Treg-dependent effects on neovascularization in heart failure (HF) is related to both direct anti-angiogenic effects on endothelial cells and the mobilization and recruitment of circulating angiogenic cells, in part related to CCL5/CCR5 signaling.

• The immunocompromised Treg phenotype in HF exhibits plasticity and normalizes upon selective depletion and subsequent reconstitution with new Tregs.

What are the clinical implications?

• To date, there have been no large scale immunomodulatory or anti-inflammatory therapies for chronic HF successfully translated into clinical practice.

• As Tregs are key immunomodulatory immune cells, Treg dysfunction may underlie the persistent and chronic inflammation seen in HF.

• Restoration of proper Treg function may be a fruitful approach to therapeutic immunomodulation in this disease.