Atherosclerosis is an inflammatory disease of the arterial wall that leads to myocardial infarction and stroke. Regulatory T cells (Tregs) and IL-10 exert significant anti-atherogenic effects in experimental models of atherosclerosis by modulating vascular inflammation. We have previously shown that Mycobacterium bovis BCG killed by extended freeze-drying (EFD BCG) decreases lung and colon inflammation by recruiting IL-10-producing Tregs. Therefore, the aim of this study was to investigate the effect of EFD BCG on the development of atherosclerosis.
We used two strains of atherosclerosis-prone mice: Ldlr−/− (four or six EFD BCG injections) and Apoe−/− (six injections).
In both models, EFD BCG significantly reduced the size of atherosclerotic lesions, increased IL-10 production and reduced the serum levels of pro-inflammatory cytokines (IL-6, IL-13, KC and tumour necrosis factor-α). Shortly after treatment with EFD BCG, the number of plasmacytoid dendritic cells (pDCs) and Foxp3+ Tregs in the draining lymph nodes increased. EFD BCG also led to accumulation of Tregs, but not of pDCs in the spleen, and reduced activity of NF-κB and increased activity of PPAR-γ in both the spleen and vascular tissue of treated mice.
EFD BCG has atheroprotective effects through IL-10 production and Treg expansion. These findings support a novel approach to the prevention and treatment of atherosclerosis.
Atherosclerotic inflammation of the arterial wall is mediated by innate and adaptive immune responses. Although most inflammatory cells in atherosclerotic lesions are macrophages, up to 20% of the cells are T lymphocytes . T-cell activation regulates the initiation and growth of atherosclerotic lesions and might accelerate atherosclerotic plaque destabilization, which leads to plaque disruption and the onset of acute coronary syndrome .
IL-10 has anti-atherogenic effects [2, 3]. T regulatory cells (Tregs) suppress effector T cells through IL-10 production and can inhibit the development of atherosclerosis in mice . Adoptive transfer of Tregs reduces the inflammation in atherosclerotic lesions, whereas transfer of effector CD4+ T lymphocytes aggravates atherosclerosis [4, 5]. Adaptive Tregs, which have been implicated in the regulation of atherogenesis, develop in the periphery in response to antigenic stimulation and can be generated in vitro in the presence of plasmacytoid dendritic cells (pDCs) in an IL-10-dependent manner [6, 7].
These findings have raised the possibility that immunoregulatory pathways can be triggered to reduce inflammation and treat or prevent atherosclerosis. Bacille Calmette–Guérin (BCG) vaccination was shown to protect against the development of atopy in infants, suggesting that BCG can modulate immune responses . Mycobacterium bovis BCG that is killed by extended freeze-drying (EFD BCG) does not contain live bacteria; it was found to reduce inflammation in murine models of asthma and colitis by stimulating IL-10 production and expansion of Tregs [7, 9, 10].
The aim of this study was to determine the effect of EFD BCG treatment on atherogenesis, using two atherosclerosis-prone mouse models.
Production of EFD BCG
The live BCG Pasteur strain 1173P2 was grown in Sauton medium under conditions that are used for vaccine production . The BCG cultures were killed by extended freeze-drying ; this was verified by a lack of growth of 20 mg EFD BCG in Middlebrook 7H11 medium (Difco, BD Biosciences, San Jose, CA, USA).
Animals and treatment protocol
We used two atherosclerosis-prone mouse models: (i) low-density lipoprotein receptor-deficient mice (Ldlr−/−), which develop atherosclerosis when fed a high-fat diet; and (ii) apolipoprotein E-deficient (Apoe−/−) mice, which spontaneously develop atherosclerotic lesions at 20 weeks. Male Ldlr−/− and Apoe−/− mice (both on a C57BL/6J background) were obtained from Jackson Laboratories (Bar Harbor, ME, USA) and from S. Rabot (Jouy en Josas, France), respectively. The mice were housed either in the Animal Experimentation Unit at Centre de Recherche INRA or the animal facilities at Karolinska University Hospital. Ldlr−/− mice received either four (at the age of 6, 10, 14 and 18 weeks) or six (at the age of 6, 10, 14, 18, 22 and 26 weeks) subcutaneous injections of either EFD BCG (100 μg) or phosphate-buffered saline (PBS; 100 μL) in the base of the tail and were killed at the age of 20 or 30 weeks, respectively (Table 1). From 5 weeks of age, Ldlr−/− mice were fed chow diet supplemented with 15% lard and 0.5% cholesterol (Genestil, Rayaucourt, France). Apoe−/− mice received six (at the age of 6, 10, 14, 18, 22 and 26 weeks) subcutaneous injections, were fed standard mouse chow and were killed at 30 weeks (Table 1). All animal experiments were approved by the local ethics committees.
Table 1. Reduction in atherosclerotic burden after treatment with EFD BCG
Histological analysis and immunostaining of the atherosclerotic lesions
The heart and ascending aorta were dissected and preserved for immunohistochemistry and lesion analysis. Cryosection and lesion analyses were performed as previously described . Briefly, hearts were serially sectioned from the proximal 1 mm of the aortic root using a cryostat. Haematoxylin- and Oil Red O-stained sections were used to evaluate lesion size. Lesion size was determined in eight haematoxylin- and Oil Red O-stained sections acquired at 100-μm intervals over a 1-mm segment of the aortic root. For each section, images were captured, and the surface areas of the lesion and of the lumen of the entire vessel were measured.
Lipid accumulation was determined in en face aortic arch preparations from mice using Sudan IV staining. Briefly, dissected arches were fixed in 4% neutral-buffered formalin. Samples were then cut longitudinally, splayed, pinned and stained with Sudan IV (red colour). The overall area of all the plaques in a given aortic arch was calculated as a percentage of the total surface area of the arch (excluding branching vessels).
Serum samples were collected at the end of the experiment. Levels of cytokines were measured using the Bio-Plex Cytokine Assay (Bio-Rad, Marnes La Coquette, France) as previously described . Levels of transforming growth factor-beta (TGF-β) were measured by enzyme-linked immunosorbent assay (CliniScience, avenue Georges Clémenceau, Nanterre, France). Analysis of lipid profiles and antibodies against oxidized LDL (oxLDL), as well as the side effects of EFD BCG, is described in the Supporting information.
Detection of transcription factors
Proteins were extracted from spleens of Ldlr−/− and Apoe−/− mice and resolved using 7.5% SDS-PAGE. Proteins were transferred to a nitrocellulose membrane and probed with antibodies against Foxp3, T-bet, GATA-3, RORγt, RXRα, p-RXRα, STAT-1, phosphorylated STAT-1 (p-STAT-1), STAT-4, p-STAT-4, STAT-5b, p-STAT-5b, STAT-6 and p-STAT-6 (Santa Cruz Biotechnology, Dallas, TX, USA). β-actin (Ac-15 Abcam, Cambridge, UK) was used as an internal control. Specific bands were investigated using scanning analysis as previously described .
Nuclear proteins were extracted from spleen or vascular tissue homogenates and processed with nuclear factor-kappaB subunit p65 (NF-κBp65), peroxisome proliferator-activated receptor-gamma and peroxisome proliferator-activated receptor-alpha (PPAR-γ and PPAR-α) or SP-1 transcription factor using TransAMTM transcription factor assay kits (Active Motif, Carlsbad, CA, USA) according to the manufacturer's instructions. Statistical analysis was not performed for vascular tissue homogenates because transcription factors were determined in pooled extracts (two groups of three vascular tissues).
Flow cytometric analysis
Tregs and dendritic cells (DCs) were identified in the inguinal draining lymph nodes (iDLNs) (4 days after the single injection) or spleens (at the end of the experiment) of Ldlr−/− mice. Cells were stained with antibodies against CD4, Foxp3, inducible co-stimulator (ICOS), CD11c, B220, ICOS ligand (ICOS-L; BD Biosciences) and PDCA-1 (Miltenyi Biotec, Paris, France). All samples were analysed on a FACSCalibur flow cytometer using CellQuest software (BD Biosciences).
Dendritic cells were isolated from iDLNs of C57Bl/6 mice 4 days after one EFD BCG injection. Following cell sorting, pDCs (CD11clow B220high) and conventional DCs (cDCs; CD11chigh B220neg) were cultured alone or co-cultured with naïve T cells (CD4+CD45RBhigh) purified from spleens of untreated mice as described in the Supporting information. Levels of IL-10, IL-12p40 and interferon (IFN)-γ were measured in the culture supernatants using Bio-Plex cytokine assay kits (Bio-Rad).
Data are expressed as mean ± SD. The Instat package from GraphPad Software (San Diego, CA, USA) was used to analyse the data using the t-test with Welch's correction.
EFD BCG reduces atherosclerosis and inflammation in hypercholesterolaemic mice
We investigated the effect of EFD BCG on atherogenesis in Ldlr−/− and Apoe−/− mice. Independent of age, mouse strain and number of injections, EFD BCG treatment reduced lesion cross-sectional area in the aortic roots and decreased lipid lesion area in the en face aortic preparation compared with treatment with PBS (Table 1 and Fig. 1). Immunohistochemical staining showed less infiltration of MOMA2 + macrophages in atherosclerotic lesions in EFD BCG- compared with PBS-treated Ldlr−/− (Fig. 2a, b) and Apoe−/− mice (77.8 ± 5.6 vs. 42.6 ± 7.4; P <0.0001). RNA expression of macrophage- and T cell-specific markers was decreased in aortae of EFG BCG-treated LDLR−/−mice (Fig. 2c).
Serum levels of IL-10 increased 4-fold in Ldlr−/− mice and 8-fold in Apoe−/− mice after EFD BCG compared with PBS treatment (Fig. 2d and Fig. S1A). In addition, treatment with EFD BCG significantly lowered serum levels of IL-1β, IL-13, KC, MIP-1β and tumour necrosis factor-α, and increased the level of IFN-γ in both mouse strains. IL-6 and IL-17 levels were decreased in Apoe−/−, but not in Ldlr−/− mice (Fig. 2d and Fig. S1).
EFD BCG modulates the activation of key transcription factors in spleen and vascular tissue of treated mice
NF-κB is a transcriptional factor that is associated with the activation of inflammatory responses in atherosclerotic lesions . NF-κBp65-binding activity decreased in nuclear extracts from vascular tissues and spleens of Ldlr−/− mice after EFD BCG treatment compared with control treatment with PBS (Fig. 3a,b). PPAR-γ signalling inhibits the production of inflammatory cytokines . Unphosphorylated retinoid X receptor α (RXRα) forms a heterodimer with PPAR-γ to reduce inflammation . EFD BCG increased PPAR-γ binding activity in vascular tissues and spleens and inhibited RXRα phosphorylation in spleens of Ldlr−/− mice (Fig. 3a–c). Moreover, splenic nuclear extracts from EFD BCG-treated Ldlr−/− and Apoe−/− mice had higher levels of nuclear SP-1 (Fig. 3d and Fig. S1B), which is an important component of IL-10-mediated immunoregulation .
We analysed the effect of EFD BCG on transcription factors that regulate T-cell commitment. EFD BCG treatment was associated with increased T-bet and Foxp3 and decreased GATA3 and RORγt levels in spleens of Ldlr−/− and Apoe−/− mice (Fig. 3e and Fig. S1D). In Ldlr−/− mice, EFD BCG treatment correlated with lower levels of phosphorylation of STAT-1, STAT-4 and STAT-6, but a higher level of phosphorylation of STAT-5b (Fig. 3f). Thus, our data indicate that EFD BCG initiates immunoregulatory responses by activating SP-1 and by increasing the levels of Foxp3+ and p-STAT-5b in mice. Our findings also suggest that EFD BCG has anti-inflammatory effects, as evidenced by impaired NF-κB activity, increased PPAR-γ activity and decreased phosphorylation of RXRα, STAT-1, STAT-4 and STAT-6.
EFD BCG treatment increases the number of pDCs and Tregs
Various subsets of DCs contribute to the expansion and differentiation of Tregs . We analysed DC subsets and Tregs in Ldlr−/− mice (i) in iDLNs, 4 days after the first injection of either EFD BCG or PBS, and (ii) in spleens, at the end of the experiment (30 weeks) (Fig. 4). More pDCs (CD11clowB220hi) were detected in the iDLNs from EFD BCG- compared with PBS-treated mice (Fig. 4a). We observed that almost 100% of these pDCs expressed PDCA-1, a marker of mouse pDCs . By contrast, fewer cDCs (CD11chiB220neg) were observed in the iDLNs after injection of EFD BCG (Fig. 4a).
In parallel to the increased population of pDCs in the iDLNs, the percentage and the absolute number of CD4+Foxp3+ Tregs in the iDLNs of EFD BCG-treated Ldlr−/− mice were higher compared with control PBS-treated animals (Fig. 4c,e). The absolute number of CD4+Foxp3+ cells in spleens of Ldlr−/− mice was increased after EFD BCG treatment (Fig. 4f); this finding was confirmed in Apoe−/− mice (Fig. S1C). It should be noted that the number of pDCs was not increased in spleens of EFD BCG-treated Ldlr−/− mice compared with control animals (13.1 ± 0.9 × 104 vs. 12.2 ± 0.6 × 104, respectively), suggesting that pDCs recruited to the iDLNs after EFD BCG injection polarize naïve CD4 T cells towards Tregs without migrating and expanding in the periphery.
Mature pDCs can induce generation of Tregs by expressing high levels of ICOS-L , whereas ICOS has been implicated in the reduction of atherosclerosis . We detected more ICOS-L-expressing pDCs and a higher number of CD4+Foxp3+ICOS+ cells in the iDLNs of EFD BCG- than of PBS-treated Ldlr−/− mice (Fig. 4b,d).
pDCs polarize naïve T cells towards Foxp3+ Tregs in an IL-10-dependent manner
To investigate the functional activity of pDCs and cDCs in the polarization of naïve T cells towards Tregs, we purified both DC subsets from the iDLNs of C57BL/6J mice 4 days after one injection of EFD BCG. After 24 h, the supernatant of cultured pDCs contained predominantly IL-10, whereas cDCs secreted mainly IL-12p40 (Fig. 5a). When pDCs were co-cultured (for 72 h) with sorted naïve T cells, the concentration of IL-10 in the supernatant was significantly increased as compared to pDCs cultured alone (Fig. 5b). By contrast, after co-culturing cDCs with sorted naïve T cells, secretion of IFN-γ but not of IL-10 was increased (Fig. 5b).
Furthermore, the number of CD4+Foxp3+ cells was increased when naïve T cells were co-cultured with pDCs isolated from EFD BCG-treated mice (Fig. 5c). The effect was abrogated by anti-IL-10, but not by anti-TGF-β antibodies (Fig. 5c). By contrast, cDCs failed to generate Tregs (Fig. 5c). We therefore suggest that EFD BCG causes pDCs to produce IL-10. pDCs secreting IL-10, in turn, are able to polarize naïve T cells towards IL-10-producing Tregs.
Lipid profiles and antibody titres against oxLDL after EFD BCG treatment
Serum cholesterol, triglycerides, lipoprotein profiles and titres of antibodies against serum oxLDL were unchanged after EFD BCG treatment in Ldlr−/− and Apoe−/− mice (Figs S2–S4).
Multiple injections of EFD BCG do not induce measurable side effects
We assessed the effects of EFD BCG on the defensive host immune response. Despite its immunoregulatory effects, EFD BCG did not modify the protection that was conferred by vaccines against Mycobacterium tuberculosis in guinea pigs or against Neisseria meningitidis in mice [infections that induce T helper (Th)1- and Th2-mediated immune responses, respectively] (Fig. S5A,B). Moreover, EFD BCG did not exacerbate M. tuberculosis infection in guinea pigs (Fig. S5A).
We have recently shown that a single injection of EFD BCG does not induce sensitization to purified protein derivative, fever or lymph node enlargement . We evaluated the side effects of repeated treatment with EFD BCG in OF1 outbred Swiss mice, which are often used in toxicological studies due to high population diversity. After 14 consecutive injections with various doses of EFD BCG (0.01, 0.1, 1 mg), no changes in rectal temperature or body weight were observed, compared with PBS-treated mice (Fig. S6A,B). Thus, EFD BCG does not cause any measurable adverse effects.
It has been demonstrated that treatment with a dual PPAR agonist increases atherosclerosis in Apoe−/− mice . We measured the expression of two PPAR isoforms (α and γ) in spleens from OF1 Swiss mice that received increasing doses of EFD BCG for 14 consecutive days. Compared with PBS-treated mice, PPAR-γ expression increased after EFD BCG injection in a dose-dependent manner, whereas PPAR-α levels remained unchanged (Fig. S6C).
The present findings provide evidence that EFD BCG injections lead to reduction in atherosclerosis in hypercholesterolaemic mice via elevation of IL-10 levels and expansion of Foxp3+ Tregs. IL-10 and Tregs are known to control inflammation during atherogenesis [3, 4, 21]. Tolerogenic modulation of DCs by IL-10 treatment attenuates systemic inflammation and reduces the atherosclerotic plaque burden in hypercholesterolaemic mice . We previously demonstrated that EFD BCG treatment regulates lung and colon inflammation by increasing pDC recruitment and expansion of IL-10-producing Tregs [7, 9, 10]. Moreover, depletion of either pDCs or Tregs, as well as anti-IL-10 injections, abrogated the protective effect of EFD BCG in these studies. Consistent with our previous findings, the atheroprotective effect of EFD BCG in the present study was associated with the early recruitment of pDCs to iDLNs, which was followed by the expansion of Foxp3+ Tregs in the spleens and vascular tissues of treated mice. Increased serum levels of IL-10 in EFD BCG-treated mice were accompanied by increased expression of transcription factors that govern the activation of IL-10 (SP-1 and p-STAT5b) in the spleen. Taken together, the results of the present study confirm and extend recent findings that induction of IL-10 and Tregs protects against atherosclerosis [21, 23]. However, in contrast to antigen-specific regulatory protection induced either by nasal immunization with cholera toxin subunit B-conjugated LDL peptide  or by the apolipoprotein B peptide vaccine , EFD BCG-induced protection appears to be independent of an atherosclerosis-specific antigen.
In addition to enhancing IL-10 signalling, EFD BCG robustly inhibited the inflammatory state by reduction of NF-κBp65 activation, augmentation of PPAR-γ activities and reduction of RXRα phosphorylation. EFD BCG decreased systemic levels of inflammatory cytokines, except IFN-γ, and reduced local accumulation of macrophages in plaques.
The expression profile of transcription factors and cytokines reflected downregulation of Th2 and Th17 differentiation by EFD BCG, as evidenced by lower serum levels of IL-13 and IL-17, as well as by decreased levels of GATA3, p-STAT-6 and RORγt in the spleen. The pattern of regulation of Th1 differentiation by EFD BCG was less apparent. In spleens, EFD BCG triggered an increase in serum levels of IFN-γ, the signature cytokine of Th1 cells, and enhanced levels of T-bet, a critical transcription factor for Th1 development . On the other hand, EFD BCG prevented the phosphorylation of STAT-4, another important factor that positively regulates the development of Th1 cells  and impairs Treg development . The role of co-expression of IFN-γ and IL-10 remains unclear with regard to immunoregulation and atherosclerosis. IFN-γ secreted by IL-10 + type 1 regulatory T cells (Tr1) has been associated with an immunoregulatory profile in several studies [27-29]. Interestingly, it was suggested that IFN-γ produced by bone marrow-derived cells may delay atherosclerosis progression . In addition, it has been proposed that IFN-γ may increase indoleamine 2,3 dioxygenase (IDO) activity in smooth muscle cells and confer vascular resistance against inflammation at the early stages of atherogenesis . Indeed, we recently demonstrated that metabolites of tryptophan downstream of IDO have anti-inflammatory and anti-atherosclerotic properties .
A nucleoprotein derived from measles virus has been shown to reduce atherosclerosis in mice  by triggering the development of suppressive Tregs that secreted IL-10 upon in vitro stimulation. However, administration of the measles viral nucleoprotein in mice reduces IL-4 and IFN-γ production in T cells, suggesting that such treatment might be immunosuppressive . By contrast, efficient immune competence of the host was sustained by EFD BCG treatment, as demonstrated by the unaltered immune defence against M. tuberculosis (Th1-mediated protection) and N. meningitidis (Th2-mediated protection) in mice. The maintenance of immune competence against infections after EFD BCG treatment suggests that this could be a tolerable therapeutic agent in humans.
Finally, we observed that the activation of PPAR-γ expression by EFD BCG is not associated with increased PPAR-α expression. Indeed, it has been found that the combination of PPAR-γ agonists (used as insulin sensitizers in diabetic patients) and PPAR-α agonists (used to treat dyslipidaemia) enhances atherosclerosis in Apoe−/− mice  and induces major adverse cardiovascular events in humans . The fact that EFD BCG is a selective inducer of PPAR-γ, but not of PPAR-α, supports its potential safety in clinical use.
In conclusion, the findings of the present study demonstrate that EFD BCG exerts atheroprotective effects in two distinct mouse models. EFD BCG elicited simultaneously (i) an immunoregulatory effect through IL-10 production and expansion of Tregs, (ii) an inhibition of NF-κB activation and (iii) an increase in PPAR-γ expression without altering PPAR-α levels. EFD BCG had no measurable side effects in this study, and efficient immune competence was sustained in the host. These findings suggest that EFD BCG may provide a novel strategy for the prevention and treatment of atherosclerosis, which could be introduced into the clinic relatively rapidly.
We thank Dr Geneviève Milon for helpful discussions; Dr Michel Huerre for assistance with the histological studies, Dr Mohammad Abolhassani for Western blot analysis and Drs Jean-Michel Alonso and Marek Szatanik for conducting the studies with vaccines against N. meningitidis.
This study was supported by Immunotherapix, Institut Pasteur, INRA, Oséo, la Mairie de Paris (‘Research in Paris’), the Swedish Research Council, Swedish Heart-Lung Foundation, Stockholm County Council/ALF, the Foundation for Strategic Research, Karolinska University Hospital, the Leducq Foundation and the European Union (AtheroRemo project).
ML and GM are shareholders of Immunotherapix, a start-up company of the Institut Pasteur that is dedicated to EFD BCG development.
Conflict of interest statement
None of the other authors has any conflict of interests to declare.