Foxp3⁺ CD4⁺ T Cells Improve Healing After Myocardial Infarction by Modulating Monocyte/Macrophage Differentiation

Mice between 8 and 12 weeks of age were used for all experiments. C57BL/6J mice were purchased from Harlan Laboratories. Mice expressing the DTX under control of the Foxp3 promotor, such as depletion of regulatory T cell mice,¹⁶ namely B6.129(Cg)-Foxp3^tm3Ayr/J (Foxp3^DTR mice), and respective wild-type controls from Jackson Laboratory were used here for pharmacological ablation of T_reg cells. Mice underwent left coronary artery ligation as described previously.⁵ The study conformed to the regulations for animal experimentation and was approved by the local government. Briefly, mice were anesthetized with isoflurane. After intubation, thoracotomy was performed and MI induced by ligation of the proximal part of the left coronary artery. Buprenorphine was administered for analgesia after surgery. For sham operation, thoracotomy was performed without ligating the coronary artery. In depletion experiments, 500 ng DTX per mouse were administered intraperitoneally on day 2 and day 1 before MI induction. To prevent a rebound of T_reg cells, 250 ng DTX per mouse were additionally injected on day 2 and day 4 after MI. For antibody-mediated T_reg-cell depletion, 1 mg of anti-CD25 monoclonal antibody (clone PC61) was injected. For expansion of T_reg cells, mice were treated with 300 µg of a superagonistic anti-CD28 monoclonal antibody (clone D665).

At autopsy, large amounts of blood clots around the heart and within the thoracic cavity in combination with perforation of the infarcted wall indicated rupture.

Echocardiography was performed on a Toshiba Aplio system with a 15-MHz ultrasound probe with the mice under light anesthesia with isoflurane (1.5 vol%).¹⁷

Cardiac scar tissue was digested with collagenase type 2 and protease type XIV (Sigma-Aldrich, Munich, Germany) as described previously.¹⁸ Staining protocols are specified in the Online Data Supplement.

Cryosections (5 µm) were fixed with 4% formaldehyde in PBS for 10 minutes, permeabilized with 0.5% dodecyltrimethylammonium chloride in PBS (all chemicals from Sigma-Aldrich). After blocking, Foxp3 was detected using the antimouse Foxp3 antibody clone (FJK-16s; eBioscience, Frankfurt, Germany). The staining protocol is specified in the Online Data Supplement.

Quantitative real-time polymerase chain reaction was performed (iCycler from Bio-Rad, München) with commercial TaqMan probes (Life Technologies, Darmstadt, Germany). Target gene mRNA levels were normalized to glyceraldehyde 3-phosphate dehydrogenase. For real-time reverse transcriptase–polymerase chain reaction of sorted cells, mRNA was amplified before complementary DNA synthesis using the C&E Trinucleotide Pico Kit (AmpTec, Hamburg, Germany) according to the manufacturer’s protocol.

For sorting of T-cell subsets and monocytic cells, single-cell suspensions were prepared from spleens as well as inguinal, cervical, and axillary lymph nodes. Monocytic cells were defined as CD11b⁺Ly-6G⁻CD11c⁻. T_reg cells were defined as CD4⁺CD25^high, and conventional T cells as CD4⁺CD25⁻. Cells (2×10⁵ per well) were seeded in 96-well flat-bottom plates (greiner bio-one, Frickenhausen, Germany). In coculture experiments, 1×10⁵ cells per cell type were used. All cell cultures were supplemented with 25 IU/mL recombinant interleukin (IL)-2 (Proleukin; Novartis, Basel, Germany) and, where denoted, anti-TGFβ antibody (10 µg/mL; clone 1D11; R&D, Wiesbaden, Germany) or recombinant cytokines. CD28-SA (clone D665; AbD Serotec, Raleigh, NC) or isotype control immunoglobulin (clone MOPC-21; Bio-X-Cell, West Lebanon, NH) was coated on pan mouse immunoglobulin G dynabeads (Invitrogen, Darmstadt, Germany) and added to cell cultures where indicated.

Cell culture supernatants were analyzed for presence of osteopontin by ELISA (R&D) as well as tumor necrosis factor (TNF), interferon-γ, IL-6, IL-2, IL-4, IL-13, and IL-17 by Cytometric Bead Array (BD Biosciences, Heidelberg, Germany) according to the manufacturer’s protocol. The latter results were analyzed with FCAP Array Version 2.0 (Soft Flow, Duesseldorf, Germany).

After blocking, nitrocellulose membranes were incubated overnight in presence of rabbit antirat collagen type I (CL50141AP; Cedarlane, Burlington) and rabbit antirat collagen type III (CL50341AP; Cedarlane). The membrane was developed using an enhanced chemiluminescent detection system. After membrane stripping, total protein amount per lane was assessed by detection of glyceraldehyde 3-phosphate dehydrogenase.

All data are presented as mean values per group and SEM. For comparison of 2 groups, an unpaired t test or, if a t test was not suitable, a Wilcoxon signed rank-sum test was performed. For multiple comparisons, 2-way ANOVA was used. Survival is shown as Kaplan–Meier curve, and data were analyzed by a log-rank test. Variance in a group was assessed using the χ² test. Differences were considered as statistically significant at P<0.05. Data analysis was performed using GraphPad Prism 4.03 (GraphPad Software Inc, San Diego, CA).

To monitor T_reg-cell activation after MI, we analyzed heart-draining lymph nodes. Here, the activation markers CD25 (Figure 1A) and Helios (Online Figure I) were upregulated in T_reg cells 7 days post-MI compared with sham-operated animals. As previously reported by us, increased frequencies of T_reg cells were found in mLNs on days 3 and 7 post-MI as compared with sham-operated animals analyzed on day 7 after surgery (Figure 1B).⁵ Remarkably, compared with sham-operated mice, the percentage of Foxp3⁺ cells among CD4⁺ T cells stayed elevated for ≤56 days (Figure 1B). Absolute cell numbers of T_reg cells were also increased peaking on day 7 after MI (Figure 1B). In addition to T_reg-cell expansion kinetics in heart-draining lymph nodes, we evaluated T_reg cell infiltration into the healing infarct. Compared with sham-operated mice, both frequencies and absolute numbers of T_reg cells were increased post-MI (Figure 1C and 1D). Moreover, using immunofluorescence microscopy, single Foxp3⁺ cells were detected in the infarct border and within the remote myocardium (Figure 1E; Online Figure II).

Hence, we could show that T_reg cells become activated in response to MI, followed by T_reg-cell expansion in heart-draining lymph nodes and successive infiltration into the healing infarct.

To evaluate the influence of T_reg cells on wound healing after MI, we used Foxp3^DTR mice to ablate Foxp3-expressing cells specifically that transgenically express the human DTR. Two days after DTX administration, Foxp3⁺ cells were effectively depleted from blood, mLNs, and hearts as compared with DTX-treated wild-type littermates (Figure 2A). Survival was not significantly different between the 2 groups, but infarct sizes in T_reg-cell–depleted mice were significantly increased as compared with the control group. Furthermore, T_reg-cell ablation resulted in a tendency toward a more pronounced left ventricular dilation in line with the impaired cardiac function evaluated by apical fractional shortening on both day 3 and day 7 post-MI (Figure 2C; Online Tables I and II). Consistently, Foxp3^DTR mice showed increased lung weight/body weight ratios underlining the compromised cardiac function in these mice (Figure 2D). With respect to procollagen synthesis, expression of matrix metalloproteinases, and tissue inhibitors of matrix metalloproteinases, no differences were identifiable between the groups (Online Figure III). In accordance with this observation, there were no differences regarding the frequencies of left ventricular ruptures (data not shown). T_reg-cell ablation, thus, resulted in poorer clinical outcome after MI without, however, directly affecting collagen turnover and scar formation.

The observation that specific T_reg-cell depletion leads to increased infarct sizes and impaired cardiac function prompted us to focus on the leukocyte influx into the infarct zone. Both monocyte subsets and neutrophils have been demonstrated to influence cardiac wound healing after MI. Using fluorescence-activated cell sorting analyses, neutrophils were defined as Ly-6G⁺CD11b⁺. Compared with control mice, neutrophil numbers were significantly increased in the infarct zone of T_reg-cell–depleted animals (Figure 3A). Furthermore, we discriminated between monocyte subsets based on lymphocyte antigen 6C (Ly-6C) surface expression. In comparison with control mice, the proportion of inflammatory Ly-6C^high cells among CD11b⁺F4-80⁻Ly-6G⁻ monocytes was elevated in the infarct zone of T_reg-cell–ablated mice (Figure 3B).

Because T_reg cells are capable of regulating inflammatory reactions, we evaluated macrophage polarization in T_reg-cell–sufficient and T_reg-cell–ablated mice. Expression of prototypic markers for inflammatory M1 macrophage polarization was assessed in CD11b⁺Ly-6G⁻ monocytes/macrophages sorted from the healing infarct 5 days post-MI. Compared with DTX-treated wild-type mice, monocytic cells sorted from Foxp3^DTR mice showed significantly higher mRNA expression of inducible nitric oxide (NO) synthase, but no difference regarding TNFα mRNA synthesis (Figure 3C). Moreover, compared with wild-type littermates, mRNA expression of M2-associated anti-inflammatory IL-10 and transforming growth factor β1 (TGFβ1) in line with mRNA synthesis of wound-stabilizing osteopontin and transglutaminase factor XIII (FXIII) was downregulated in monocytic cells sorted from the healing infarct of T_reg-cell–ablated mice (Figure 3C).

In addition to characterizing the myeloid cell compartment, both T-cell infiltration into the infarct zone and expression of T-cell mediators were assessed. Compared with wild-type littermates, absolute numbers of CD4⁺ and CD8⁺ T cells were increased in the healing infarct of T_reg-cell–ablated mice 5 days post-MI (Figure 3D). Analysis of interferon-γ and TNFα mRNA synthesis in bulk scar tissue homogenates (Figure 3D) and increased frequencies of interferon-γ and TNFα-positive cells among CD4⁺ and CD8⁺ T cells in heart-draining lymph nodes (Figure 3E) suggest increased production of these proinflammatory mediators by heart-infiltrating T cells in T_reg-cell–depleted mice.

Therefore, the poor clinical outcome of T_reg-cell–depleted mice was associated with impaired M2-like differentiation of cardiac macrophages and pronounced infiltration of the heart by proinflammatory T cells.

To confirm the phenotype of deteriorated clinical outcome in Foxp3^DTR mice post-MI, we used an additional approach of T_reg-cell ablation. The T_reg-cell compartment was depleted before MI induction by administration of a monoclonal rat antimouse CD25 antibody 8 days before MI. In anti-CD25 antibody–treated animals, T_reg cells were nearly completely absent from peripheral blood on the day of MI induction and still significantly reduced in mLNs on day 7 post-MI, whereas treatment with an isotype-matched control antibody of irrelevant specificity (isotype control immunoglobulin) did not provoke T_reg-cell depletion (Figure 4A).

Antibody-mediated T_reg-cell depletion led to a significantly higher mortality as compared with control mice (Figure 4B). By day 7, survival was 55.9% in control mice and 25% in T_reg-cell–ablated animals (Figure 4B). Regarding frequencies of left ventricular ruptures, no differences could be found between treatment groups. Thus, T_reg-cell–depleted mice died presumably because of a higher incidence of cardiac failure as indicated by greater echocardiographic expansion of the left ventricular area on day 7 after MI (Figure 4C; Online Table III). Congruently with the Foxp3^DTR model, the infarct zone of the anti-CD25 antibody–treated mice showed increased numbers of neutrophils 7 days after MI as compared with the control group, in line with higher frequencies of proinflammatory Ly-6C^high cells among monocytes (Figure 4D).

Thus, both in Foxp3^DTR mice as well as in anti-CD25 antibody–treated mice, T_reg-cell depletion was associated with enhanced recruitment of neutrophils, and Ly-6C^high monocytes into the infarcted myocardium, impaired cardiac function, and anti-CD25 antibody treatment even increased mortality after MI.

Therapeutic T_reg-cell activation after MI induction results in enhanced recruitment of T_reg cells into the infarct zone and improves scar tissue formation and survival.

Because the lack of T_reg cells aggravated clinical outcome, we focused on T_reg-cell expansion to improve wound healing post-MI. T_reg-cell activation was accomplished in a therapeutic fashion, that is, after MI. The peak of physiological T_reg-cell response was found to be between day 3 and day 7 after MI induction (Figure 1B). For therapeutic T_reg expansion, we used a superagonistic anti-CD28 antibody (CD28-SA). The CD28-SA–mediated T_reg-cell expansion reaches its full effect at the earliest 2 days after a single administration.¹⁵ Thus, CD28-SA was administered on day 2 after MI induction to support the physiological expansion of T_reg cells in response to MI. By day 3 after CD28 administration, T_reg-cell frequencies among CD4⁺ T cells in blood and heart-draining lymph nodes were >2-fold elevated in CD28-SA–treated animals (Figure 5A).

In comparison with control animals, CD28-SA–treated animals had significantly improved survival owing to fewer left ventricular ruptures during the first 7 days after MI (Figure 5B and 5C). After CD28-SA treatment, 5% of infarcted animals had left ventricular ruptures. In control animals, 23% of infarcted mice died from cardiac rupture, which is in accordance with former studies.^19,20

By day 56, survival was 47.1% in isotype control immunoglobulin–treated mice and 76.6% in CD28-SA–treated animals. In surviving mice, echocardiography on day 7 and day 56 revealed no statistically significant differences between the treatment groups (Online Table IV). Better survival of mice with large MI after CD28-SA treatment presumably explains the observation that no attenuation of adverse remodeling was found in the CD28-SA group compared with isotype control immunoglobulin–treated mice.

Having observed that therapeutic T_reg-cell activation resulted in decreased rates of heart ruptures, we focused on scar tissue formation and the extracellular collagen matrix (ECM). On day 5 post-MI, CD28-SA–treated mice showed, compared with control antibody–treated animals, increased mRNA expression of both procollagen α-1 (I) and procollagen α-1 (III) that constitute integral components of the reparative scar (Figure 5D). Elevated expression of both collagen I and collagen III in CD28-SA–treated mice was confirmed at the protein level (Figure 5E). Consistently, mesenchymal vimentin and α-smooth muscle actin mRNA levels were upregulated in the healing myocardium 5 days post-MI, but not mRNA synthesis of von Willebrand factor, reflecting increased myofibroblast numbers rather than accelerated angiogenesis after CD28-SA treatment (Online Figure IVA and IVB). Increased myofibroblast numbers were confirmed by α-smooth muscle actin immunohistochemistry staining (Online Figure IVC). Moreover, compared with control mice, CD28-treated mice showed increased mRNA expression of both tissue inhibitors of matrix metalloproteinases 1 and 2, endogenous tissue inhibitors of matrix metalloproteinases preventing ECM degradation (Online Figure IVD).

To better understand how CD28-SA–activated T_reg cells improve scar tissue formation, we evaluated T_reg-cell infiltration into the healing myocardium. In CD28-SA–treated animals, the frequency of bulk CD4⁺ T cells among CD45⁺ leukocytes was slightly reduced on day 5 post-MI as compared with control animals (Figure 6A). However, the proportion of T_reg cells was ≈50% in CD28-SA–treated mice as compared with ≈25% in isotype control immunoglobulin–treated animals (Figure 6B and 6C). Absolute T_reg-cell numbers were increased in CD28-SA–treated mice on day 5 post-MI (Figure 6D). The increased prevalence of Foxp3⁺ cells in the scar tissue of CD28-SA–treated mice could be confirmed by immunohistochemistry stainings (Figure 6E), whereas CD28-SA treatment in the absence of tissue injury did not lead to T_reg-cell accumulation in the heart (Figure 6F).

Collectively, therapeutic T_reg-cell activation leads to increased recruitment of T_reg cells to the infarcted myocardium, enhances de novo extracellular matrix formation in the infarct zone, and consequently reduces ventricular rupture and mortality after MI.

Because monocytic cells are key players in wound healing post-MI, we focused on monocyte recruitment to the infarct zone. Regarding absolute cell numbers and monocyte subset composition, defined on the basis of Ly-6C expression levels, we could not detect significant differences between groups (data not shown). However, having observed that T_reg-cell ablation leads to M1-like macrophage polarization, we also assessed the monocyte/macrophage activation state in the healing myocardium of CD28-SA–treated mice. In scar tissue homogenates of CD28-SA–treated mice, M2-induced IL-13, IL-10, and TGFβ1 mRNA expressions were upregulated on day 5 post-MI as compared with isotype control immunoglobulin–treated mice (Figure 7A). Increased expression of prototypical M2-associated arginase I and the ECM-bracing M2 effectors, osteopontin and FXIII, in bulk scar tissue homogenates further indicated that M2 cells are generated in these hearts (Figure 7B).^21,22

To directly follow expression of M2 markers in monocytes/macrophages after CD28-SA treatment, we sorted CD11b⁺Ly-6G⁻ monocytic cells from the scar tissue and evaluated mRNA expression of both M1 and M2 marker genes. Compared with isotype control immunoglobulin–treated mice, M2-associated expression of arginase I and CD206 was significantly upregulated in monocytes/macrophages sorted from the scar of CD28-SA–treated animals (Figure 7C). Consistently, the M2 mediators TGFβ1 and IL-10 were also upregulated in these cells, whereas M1-associated TNFα (Figure 7D), but not IL-1β or IL-6 (Online Figure V), was downregulated.

Conclusively, therapeutic CD28-SA treatment leads to M2-like macrophage differentiation in the healing myocardium and expression of mediators such as osteopontin and FXIII, factors well known to contribute to myocardial healing.

Given that T_reg-cell numbers were elevated in the infarct zone of CD28-SA–treated mice, we speculated that T_reg cells might modulate expression of M2-associated genes in monocytic cells. To assess how T_reg cells modulate monocyte/macrophage differentiation in CD28-SA–treated mice, we set up an in vitro cell culture system with primary CD4⁺ T cells and monocytic cells sorted from secondary lympoid organs of naïve animals (Online Figures VI and VII). Monocultures of sorted T_reg cells, conventional T cells, or monocytes showed comparable minor secretion of osteopontin in the presence of CD28-SA and IL-2 after 4 days of incubation (Figure 8A). However, coculture of monocytes and T_reg cells in the presence of CD28-SA and IL-2 resulted in a dramatic increase in osteopontin secretion, whereas an unspecific isotype control immunoglobulin and IL-2 induced only mild osteopontin release (Figure 8A). Because T_reg-cell numbers are, compared with monocyte/macrophage numbers, relatively small, we titrated T_reg cells in the coculture system. Even at a T_reg cell/macrophage ratio of 1:25, osteopontin release was still higher as compared with monocytes cultivated in complete absence of T_reg cells (Figure 8B).

To test whether T_reg-cell–derived soluble mediators modulate osteopontin secretion by monocytes and to show that osteopontin is, indeed, secreted by the monocytes in our cocultures, we activated T_reg cells for 4 days by CD28-SA and IL-2 stimulation in vitro and subsequently cultured monocytes with a volume ratio of T_reg-cell supernatant to fresh culture medium containing purified monocytes of 1:1. Monocytes incubated with supernatant from T_reg cells that had been cultivated in the presence of isotype control immunoglobulin and IL-2 were used as reference. After 3 days of incubation, monocytic cells had released high amounts of osteopontin only in response to mediators in the supernatant of CD28-SA–activated T_reg cells (Figure 8C). Moreover, osteopontin release from monocytic cells was accompanied by CD206 and arginase I mRNA upregulation indicating M2-like macrophage polarization (Figure 8C).

To identify the T_reg-cell–derived mediators that might modulate the monocyte phenotype and function in our test system, we analyzed cytokine secretion into the supernatants of T_reg cells after CD28-SA stimulation for 4 days in vitro. Compared with T_reg cells cultivated in the presence of isotype control immunoglobulin and IL-2, stimulation with CD28-SA and IL-2 resulted in a modest release of TGFβ1 as well as strong secretion of both IL-10 and IL-13, but no release of IL-2 or IL-4 (Figure 8D).

Having observed a CD28-SA–induced release of IL-10, IL-13, and TGFβ1 from T_reg cells in vitro, we tried to induce osteopontin secretion from monocytes as indicator for M2 polarization by cultivation in the presence of the aforementioned cytokines. Supplementation with TGFβ1 alone did not influence osteopontin release from monocytes, but simultaneous presence of TGFβ1 and IL-10 induced osteopontin secretion, which could be further increased by the addition of IL-13. Neutralization of TGFβ1 dramatically restrained IL-10– and IL-13–driven osteopontin release from the monocytes (Figure 8E).

Conclusively, TGFβ1, IL-13, and IL-10 produced by CD28-SA–activated T_reg cells synergized in inducing M2-like differentiation and subsequent osteopontin release from monocytes/macrophages in vitro.

To address this hypothesis, we ablated T_reg cells before MI induction. T_reg-cell ablation in Foxp3^DTR mice resulted in an increased infarct size and, thus, left ventricular dilation that was consistently associated with impaired cardiac function. The phenotype of impaired outcome in the absence of T_reg cells could be confirmed in a model of anti-CD25 antibody–mediated T_reg-cell depletion. Anti-CD25 antibody–treated mice showed a significantly increased left ventricular dilation and, most strikingly, an impaired survival, suggesting that these mice succumbed predominantly to heart failure. This observation is in accordance with the phenotype of infarcted CCR5-knockout mice in which the attenuated recruitment of T_reg cells, in line with a decreased influx of other cell types, correlates with adverse remodeling and cardiac deterioration.²³

Neither of the 2 T_reg-cell depletion models showed a predisposition to develop left ventricular ruptures. This observation is seemingly at odds with the finding that CD28-SA–treated mice exhibited a reduced incidence of left ventricular ruptures. However, CD28-SA treatment increased the expression of procollagens as well as tissue inhibitors of matrix metalloproteinases 1 and 2, indicating that activated T_reg cells and M2 cells augmented or even accelerated scar tissue construction that likely prevented left ventricular ruptures in these hearts. In contrast, T_reg-cell depletion did not restrain mRNA synthesis of scar-forming collagens or collagenolytic enzymes compared with control mice implying that T_reg-cell deficiency and M1 differentiation do not impair myofibroblast function during scar tissue formation. However, as a limitation of this study, we cannot exclude that DTX- and anti-CD25 antibody–mediated ablation of T_reg cells might have differentially influenced the prevalence of fatal cardiac arrhythmias. We assume that the differences in survival observed between the 2 models of induced T_reg-cell deficiency are because of off-target or indirect effects that both DTX injections²⁴ and anti-CD25 antibody treatments²⁵ are known to be associated with to some degree.

In both models, the healing myocardium of T_reg-cell–depleted mice harbored increased numbers of inflammatory myeloid cells, that is, neutrophils, Ly-6C^high monocytes, and M1-polarized macrophages. The observation of enhanced recruitment of myeloid cells to inflammatory sites is in accordance with other models of wound healing and inflammation. In a mouse model of healing after burn injury, the lack of T_reg cells correlated with an increased influx of innate immune cells into the lesion.¹⁰ Moreover, T_reg cells modulate chemokine expression during inflammation influencing myeloid cell infiltration.²⁶ In addition to an increased proportion of inflammatory myeloid cells, both conventional CD4⁺ and CD8⁺ T cells showed a pronounced accumulation in the infarct zone of T_reg-cell–depleted Foxp3^DTR mice, contributing to the increased synthesis of TNFα and interferon-γ that both are individually capable of inducing M1 macrophage polarization.²⁷

T_reg-cell–ablated Foxp3^DTR mice exhibited a significantly increased infarct size and deteriorated cardiac function. From the mechanistic point of view, escalated TNFα synthesis in combination with a disturbed M2 differentiation likely provokes the maladaptive phenotype. TNFα is capable of depressing cardiac contractile function and elicits cardiomyocyte apoptosis as well as left ventricular dilation.^28,29 The detrimental impact of this cytokine on postinfarction remodeling is further underscored in TNFα-knockout mice that are characterized by a preservation of cardiac function.³⁰ Macrophages in the infarct zone of T_reg-cell–depleted mice further showed an increased synthesis of inducible NO synthase that critically compromises cardiac function after MI. Accordingly, previous reports showed that pharmacological inhibition or genetic inducible NO synthase deficiency in mice leads to infarct size reduction and ameliorated remodeling.^31–34 NO-dependant cytotoxicity is mediated by NO-derived radicals and an inactivation of sulfur/iron-centered enzymes that are crucially involved in metabolic pathways.^35,36 Enlarged infarct size in Foxp3^DTR mice, thus, likely arises partially from a secondary loss of cardiomyocytes.³⁷ However, although we could not detect differential expression of collagenases that are critically involved in ECM turnover, we cannot definitely rule out an escalated overall enzyme activity in the hearts of Foxp3^DTR mice, for instance, because of matrix metalloproteinase activation by reactive nitrogen and oxygen species that may have further aggravated infarct expansion.³⁸ Consistent with inducible NO synthase upregulation, monocytic cells showed an attenuated M2 polarization, which has been shown by unrelated experimental studies to be associated with aggravated left ventricular dilation and postinfarction cardiac dysfunction.^20,39 Reduced synthesis of wound-stabilizing osteopontin and FXIII as well as inflammation-resolving TGFβ1 likely further deteriorated the outcome in T_reg-cell–ablated mice.^40–42

Because T_reg depletion showed pronounced effects during the phase of myocardial healing, we next studied whether the obvious beneficial effect that T_reg cells physiologically exert on postinfarct wound healing could be further increased by therapeutic T_reg-cell activation with a CD28-SA after MI. This approach is similar to a recently published report stating that administration of CD28-SA in rats beneficially influenced cardiac remodeling.⁴³ In our mouse model, CD28-SA–treated animals had, compared with isotype control immunoglobulin–treated mice, significantly higher T_reg-cell numbers in the healing myocardium 5 days after MI. This strengthens our interpretation that the therapeutic effect we observed after CD28-SA treatment was, indeed, because of T_reg-cell activation and not a result of reduced T-cell egress from secondary lymphoid organs as has also been observed after CD28-SA treatment in vivo.⁴⁴

Herein, we demonstrate for the first time that therapeutic T_reg-cell activation/expansion by CD28-SA treatment enhances M2-like monocyte differentiation. We think that regulation of monocyte differentiation by T_reg cells certainly is not the only mechanism but constitutes a crucial contribution to the beneficial effects of therapeutic and most likely also physiological T_reg-cell activation after MI.

Alternatively, activated M2-like macrophages have well known anti-inflammatory characteristics and are an integral component of wound-healing processes.⁴⁵ Consistently, experimental modulation of macrophage polarization toward an M2 state has been previously demonstrated to improve wound healing post-MI.⁴⁶

Alternative macrophage activation is initiated in response to IL-4 or IL-13 and other stimuli such as IL-10 or TGFβ inducing an M2-like phenotype.²⁷ We show here that T_reg cells produced little TGFβ1 but high amounts of IL-10 and IL-13 in vitro on CD28-SA activation, capacitating the cells to induce M2-like macrophage differentiation. In vivo, the accumulation of T_reg cells in the hearts of CD28-SA–treated mice was associated with elevated amounts of TGFβ1, IL-10, and IL-13 within the cardiac scar tissue, suggesting that the activated T_reg cells also induced an M2-polarizing milieu locally within the heart (see model in Online Figure VIII). Production of IL-10 and TGFβ1 by M2-like cells themselves might have further contributed to the M2-creating milieu in situ. However, we cannot exclude T_reg-cell–mediated contact-dependent mechanisms that may also contribute to monocyte/macrophage polarization.

TGFβ is considered to drive collagen deposition by myofibroblasts.^45,47 In CD28-SA–treated mice, both activated T_reg cells and M2 macrophages likely contribute to increased TGFβ1 levels in the healing myocardium as compared with control animals. Moreover, IL-13, which was strongly induced by CD28-SA treatment, synergizes with TGFβ1 in promoting collagen synthesis in myofibroblasts.^48,49 Consistently, before the completion of healing 5 days post-MI, the infarct zone of CD28-SA–treated mice showed an increased amount of collagen, indicating accelerated scar tissue formation.

Collagen production and array is also crucially affected by osteopontin, which was significantly upregulated in the scar tissue of CD28-SA–treated mice. The healing myocardium of osteopontin-knockout mice has been shown to exhibit disarrayed and decreased collagen deposition, implying an essential role of osteopontin in matrix assembly and organization.^40,50 Therefore, elevated osteopontin levels in the healing myocardium likely contribute to wound-stabilizing scar tissue formation in CD28-SA–treated mice.

In other situations of tissue trauma such as nonischemic skeletal muscle damage, formation of a collagenous scar can be avoided and full tissue recovery is possible. Under such circumstances, T_reg cells have also been shown to influence the healing process beneficially, in this case by suppressing osteopontin production in favor of myotrophic factors such as amphiregulin with the latter being produced by muscle-infiltrating T_reg cells themselves.⁵¹

Because, after MI, osteopontin favors the formation of a robust collagenous scar, we set up an in vitro system to assess the contribution of monocytic cells and T_reg cells to the amount of cardiac osteopontin in vivo. T_reg cells stimulated with a CD28-SA in vitro secreted TGFβ1, IL-10, and IL-13, which, in combination, provoked strong osteopontin release from monocytic cells along with M2 differentiation. TGFβ neutralization dramatically restrained osteopontin release from monocytic cells, which is in line with the observation that a lack of TGFβ receptor engagement inhibits M2 polarization.⁵² However, TGFβ receptor signaling alone was not sufficient to elicit full osteopontin secretion, showing that TGFβ is required to render the monocytic cells responsive to T_reg-cell–derived IL-10 and IL-13 driving osteopontin expression.

Because monocytic cells are the predominant leukocyte fraction in the scar tissue 5 days after MI,⁵³ a large proportion of osteopontin within the healing myocardium of CD28-SA–treated mice might be derived from monocytes/macrophages. Nevertheless, we cannot exclude the contribution of other cell types to the observed increase in osteopontin expression.⁵⁴

Apart from osteopontin, M2-associated transglutaminase, FXIII, also crucially improves scar tissue integrity by promoting cross-linking of ECM components.⁵⁵ FXIII deficiency in mice and low expression in humans correlate with a high incidence of cardiac rupture after MI.^41,56 Gain of FXIII function by intravenous FXIII administration in mice results in both increased collagen fiber density as well as decreased rates of heart ruptures.⁵⁶ Thus, increased FXIII expression in the myocardium of CD28-SA–treated mice likely also contributes to improved scar tissue formation (Online Figure VIII).

The concept of manipulating the T_reg-cell compartment to alleviate inflammatory disorders in human patients has been proposed years ago. Recently, a human CD28-SA was successfully tested in a phase I clinical trial provoking a significant increase of the T_reg-cell signature mediator IL-10 in the plasma after treatment.⁵⁷ Therefore, administration of CD28-SA may become an eligible treatment modality to improve postinfarction healing in the clinical arena. Moreover, other unrelated strategies have been developed to selectively expand T_reg cells in vivo. For example, recombinant human IL-2 is used at low dosages to selectively expand T_reg cells compared with conventional T cells in human patients.⁵⁸ Complexing IL-2 with an anti-IL-2 monoclonal antibody currently used only experimentally probably holds even more clinical promise than low-dose IL-2 treatment because it also increases the suppressive activity of T_reg cells on a per-cell basis.⁵⁹

In conclusion, we have shown that, on the one hand, T_reg cells beneficially regulate wound healing and thus improve clinical outcome mostly by attenuating inflammation within the healing myocardium. On the other hand, therapeutic activation of T_reg cells especially improves the replacement of necrotic tissue by a stable collagenous scar and thus prevents left ventricular dilation and rupture. This potentially implicates high clinical relevance because modulation of both T_reg-cell compartment and monocyte differentiation was induced days after MI induction, resulting in improved survival (also see scheme in Online Figure VIII). Thus, there might be a therapeutic window of days to weeks in humans to prevent expansion of the myocardial infarct zone, left ventricular rupture post-MI, and detrimental progressive remodeling of viable myocardium by such an approach.