IL-17-dependent Autoimmunity to Collagen Type V in Atherosclerosis

Abstract

Rationale

Considerable evidence shows atherosclerosis to be a chronic inflammatory disease in which immunity to self-antigens contributes to disease progression. We recently identified the collagen V [col(V)] α1(V) chain as a key autoantigen driving the Th17-dependent cellular immunity underlying another chronic inflammatory disease, obliterative bronchiolitis. Since specific induction of α1(V) chains has previously been reported in human atheromas, we postulated involvement of col(V) autoimmunity in atherosclerosis.

Objective

To determine whether col(V) autoimmunity may be involved in the pathogenesis of atherosclerosis.

Methods and Results

Here we demonstrate Th17-dependent anti-col(V) immunity to be characteristic of atherosclerosis in human coronary artery disease (CAD) patients and in apolipoprotein E null (ApoE−/−) atherosclerotic mice. Responses were α1(V)-specific in CAD with variable Th1 pathway involvement. In early atherosclerosis in ApoE−/− mice, anti-col(V) immunity was tempered by an IL-10-dependent mechanism. In support of a causal role for col(V) autoimmunity in the pathogenesis of atherosclerosis, col(V)-sensitization of ApoE−/− mice on a regular chow diet overcame IL-10-mediated inhibition of col(V) autoimmunity, leading to increased atherosclerotic burden in these mice and local accumulation of IL-17 producing cells, particularly in the col(V)-rich adventitia subjacent to the atheromas.

Conclusions

These findings establish col(V) as an autoantigen in human CAD and show col(V) autoimmunity to be a consistent feature in atherosclerosis in humans and mice. Furthermore, data are consistent with a causative role for col(V) in the pathogenesis of atherosclerosis.

Atherosclerosis is the common pathologic process underlying coronary arterial disease (CAD), carotid stenosis, and peripheral arterial disease: the major causes of death and disability in Western societies ¹. Traditional therapy for atherosclerosis has consisted of risk factor modification, including treatment of hypercholesterolemia, hyperglycemia, and hypertension. It was once thought that treatment of these risk factors might eradicate cardiovascular disease by the end of the 20^th century. However, as cardiovascular disease remains the leading cause of death in the Western world, a clear role for new approaches in preventing and treating atherosclerosis is evident ^{2, 3}.

Atherosclerosis is a chronic inflammatory process regulated by a complex interplay of innate and adaptive immune responses. The concept that atherosclerosis is, at least in part, an autoimmune disease has gained considerable support ^{1, 4-6}, and both oxidized low-density lipoproteins (oxLDL) ^{2, 7-9} and heat shock proteins (HSP) ^{10, 11} have been identified as autoantigens involved in the disease process. The type of immune response associated with atherosclerosis includes Th1 cells, and is characterized by increased production of IFN-γ by plaque infiltrating and peripheral blood T cells ^{8-10, 12}. However, the relatively recent discovery of Th17 cells, characterized by IL-17 and IL-22 production, and promotion of several inflammatory autoimmune diseases ^13-15, necessitates considering potential roles for these cells in the autoimmunity of atherosclerosis. Consistent with this possibility, recent studies have found infiltration of both IL-17 and IFN-γ producing T cells within vascular plaques ¹⁶. Additionally, elevated IL-17 cytokine levels have been associated with adverse outcomes in unstable angina and acute myocardial infarction ^{17, 18}.

Collagens are critical components of the extracellular matrix of atherosclerotic plaques, where they can constitute up to 60% of total plaque protein ¹⁹ and stimulate cellular responses central to plaque development ²⁰. Recently, we demonstrated that IL-17-dependent cellular autoimmunity against the α1(V) chain of collagen type V [col(V)] underlies bronchiolitis obliterans syndrome (BOS), the chronic inflammatory/fibro-obliterative process leading to occlusion of small airways and eventual rejection of the majority of human lung transplants ²¹. Pre-transplant col(V)-specific autoimmunity was also identified as a significant risk factor for primary graft dysfunction (PGD), the leading cause of early morbidity and mortality after lung transplantation ^{22, 23}. Identification of the col(V) α1(V) chain as a critical autoantigen in these conditions, together with growing recognition of the autoimmune component underlying atherosclerosis, made it of interest that the α1(V) chain is specifically up-regulated in human atherosclerotic plaques ²⁴.

Although col(V) usually exists as α1(V)₂ α2(V) heterotrimers sequestered in the interiors of fibrils of the highly abundant collagen type I [col(I)] ²⁵, excess α1(V) chains can form aberrant α1(V)₃ homotrimers ²⁶ that are excluded from col(I) fibrils ²⁷. This suggested a model in which homotrimers comprising the excess α1(V) chains of atherosclerotic plaques could present normally cryptic epitopes that, in the inflammatory microenvironment of plaques, initiate autoimmunity. Here, we tested the hypothesis that cellular and/or humoral col(V) autoimmunity is a component of the chronic inflammatory response characterizing atherosclerosis. This hypothesis was tested both with peripheral blood mononuclear cells (PBMCs) from patients with known CAD, necessitating coronary artery bypass grafting (CABG), and in the apolipoprotein E knockout (ApoE−/−) mouse model of atherosclerosis.

Methods

Human Subjects

Immunological monitoring was performed on blood samples from patients with end stage CAD awaiting CABG. Subject consent was obtained using human subjects committee-approved, written, informed consent procedures at the University of Wisconsin Hospital and Clinics. Blood was collected and processed as described previously ²⁸. PBMC from healthy volunteers served as controls (Online Table I, Demographic & Risk Factor Analyses).

Mice

ApoE−/− mice (strain B6.129P2-Apoe^tmi/Unc/J) were obtained from Jackson Labs, and C57BL/6 control and DTH recipient mice from Harlan Sprague-Dawley Inc. Some mice at 8 weeks of age were shifted from regular chow (8604, Harlan Teklad) to a high fat diet (TD.88137, Harlan Teklad), while some remained on regular chow. At 13 weeks, all mice were immunized subcutaneously in the inguinal region with 1.5 limits of flocculation (lf) tetanus toxoid and diphtheria (TT/DT) pediatrics vaccine (Sanofi-Aventis Pasteur). After 15 weeks, spleens, serum, hearts and aortas were collected and processed. CB17.SCID mice were purchased (Harlan Sprague-Dawley Inc.) or bred locally (Gnotobiotics, UW-Madison). For Col(V) sensitization, mice were i.v. injected every two weeks with 50 μg native bovine col(V) (a generous gift from ImmuneWorks, Inc., Indianapolis, IN) for a total of 8 injections while on regular chow. Experiments were performed in accordance with the National Institutes of Health and U.S. Department of Agriculture guidelines, after approval by the University of Wisconsin Institutional Animal Care and Use Committee.

Trans-vivo Delayed-Type Hypersensitivity Assay

TV-DTH was performed by co-transfer of human PBMCs and antigens into the footpads of SCID mice as previously described ²⁹, or by injection of 8-10×10⁶ mouse splenocytes into footpads of naïve B6 recipients as previously described ³⁰. Inactivated 5 μg EBV (Viral Antigens Inc.) and 0.25 lf TT/DT (Sanofi-Aventis Pasteur) were used as positive control recall antigens for human and mouse cells, respectively. Bovine col(V), human col(V), and separated bovine α1(V) and α2(V) chains were all obtained from D. Brand, University of Tennessee (Memphis, TN, USA), and were prepared from placenta, essentially as described elsewhere ³¹. Col(V) used in all panels of Fig. 1 was human, with the exception of the native col(V) and separate α1(V) and α2(V) chains used in panel B, which were all bovine. Col(V) used in all mouse studies was bovine. Human col(I) (BD Pharmingen), bovine col(II) (Southern Biotech), and human col(IV) (Sigma) were purchased. Collagens were tested at 5 μg/injection (human) and 20 μg/injection (mouse). Cytokines were neutralized by co-injection of 10 μg antibodies against human IFN- γ, IL-17, TNF-α, IL-1β, IL-2, IL-4 (all eBiosciences), or IL-22 (R&D Systems), or antibodies directed against mouse IFN- γ, IL-17, TGF-β (all three from BD Pharmingen), TNF-α, IL-1β (both eBiosciences), or IL-10 (R&D Systems). In some TV-DTH assays, human PBMC were incubated with CD4, CD14, or CD8 microbeads (Miltenyi) and were depleted using autoMACS (Miltenyi). Fab fragments of anti-HLA-DR (mAb L243) and anti-HLA class I (mAb W6/32) were generated using a Fab micro-preparation kit (Pierce) and used at 1 ug/injection for neutralization experiments.

Figure 1. Patients with coronary artery disease have an autoimmune response to col(V).

A, TV-DTH responses by PBMC from age-matched controls or end-stage CAD patients. Responses to EBV recall antigen (positive control), col(V), col(II), or col(I), shown as individual data points, were averaged from duplicate tests. Horizontal bars denote group means. The p value of comparison between CAD and control swelling responses to col(V) is shown. B, Dose-response in TV-DTH assay of PBMC from n=3 CAD patients and n=2 controls to col(V), or to separate α1(V) and α2(V) chains. C, TV-DTH responses to col(V) by whole PBMC, CD4 T cell-depleted PBMC, CD8 T cell-depleted, and CD14 monocyte depleted PBMC for n=4 CAD patients. Data points display mean and standard deviation for n=1-3 depletion experiments per patient. p value shown compares col(V) responses between CD4 T cell depletion or CD14 monocyte depleted and whole PBMC. D, CAD patients (n=3-5) whose TV-DTH responses to col(V) were > 75mm-3 were selected. PBMC were co-injected with col(V) and either IgG isotype control or neutralizing antibodies to IL-2, IL-4, IFN-γ, IL-17, IL-22 (n=4), IL-1β, or TNF-α. Horizontal bars represent the mean value for each group. Responses were compared with IgG isotype control injection. *p≤0.05; **p≤0.01, ***p≤0.0002.

Flow Cytometry Detection of Serum Anti-col(V) Antibodies

Serum anti-col(V) antibodies were detected using streptavidin-coated beads (Polyscience) bound to biotinylated bovine col(V) (ImmuneWorks, Inc), followed by incubation with anti-human or antimouse PE-conjugated IgG antibodies (Sigma) and analysis on a FACSCalibur cytofluorograph (BD Biosciences). Additional details are in the Online Data Supplement.

Histological Analysis

Mouse aortic roots and arches were frozen in OCT (Tissue-Tek) or embedded in paraffin, and sectioned at a thickness of 5 μm over 50 μm intervals for both standard H/E preparations and immunohistochemistry. Atherosclerotic plaque burden in the aortic root, composition of atherosclerotic plaques, and immunohistochemical staining were performed as described in the expanded Methods section, available in the online data supplement.

Statistical Analysis

Data was analyzed using GraphPad Prism 5 software (GraphPad Software, La Jolla, CA). The non-parametric Mann-Whitney test was used to analyze all TV-DTH and antibody data. Unpaired t tests were used for all other analyses.

Results

Th17 and Th1/17 Cellular Immunity to Col (V) in Human Coronary Artery Disease (CAD)

To examine the possibility of anti-col(V) cellular autoimmunity as a component of human atherosclerosis, we tested for cellular anti-col(V) immunity in the PBMC of patients with end-stage CAD requiring CABG. Col(V)-specific cell-mediated immunity was assessed using the trans-vivo delayed-type hypersensitivity (TV-DTH) assay. Figure 1A shows responses to recall antigen [Epstein Barr Virus (EBV)], col(V), and collagen types II and I [col(II) and col(I), respectively] by PBMC of end stage CAD patients and of healthy normal subjects. Both groups of subjects had similar positive responses to the recall antigen and negative responses to col(II), an irrelevant collagen not found in the cardiovascular system, as well as to col(I), a collagen present in large quantities in atherosclerotic lesions ³². However, the CAD patients had a significantly higher cellular immune response to col(V) (p<0.0001 vs. controls) (Fig. 1A). To determine whether this cell-mediated immune response was directed against one or both col(V) chains, separate α1(V) and α2(V) chains were tested via the TV-DTH assay. As shown in Figure 1B, the cellular reactivity of the CAD patients mapped to the α1(V) chain, with little-to-no response directed against the α2(V) chain. TV-DTH responses from control patients’ PBMC remained low against intact col(V) and the separate col(V) chains (Fig. 1B).

To determine whether CD4⁺ or CD8⁺ T cells, as well as monocytes, mediate the col(V)-specific TV-DTH response, PBMC from CAD patients (n=4) were depleted of CD4⁺ or CD8⁺ T cells, or CD14+ monocytes prior to co-injection with col(V). Depletion of CD4⁺ T cells abolished the col(V) swelling response, whereas CD8⁺ T cell depletion had no significant effect, indicating the importance of CD4⁺ T cells in the response (Fig. 1C). This col(V) swelling response was also inhibited in the presence of neutralizing HLA-DR Fab fragments, but was not inhibited in the presence of anti-HLA class I Fab fragments, thus further supporting a role for CD4⁺ T cells in this col(V) autoimmune response (Online Figure I). Depletion of CD14⁺ monocytes also abolished the col(V) swelling response indicating that these cells are critical APCs in the col(V) response (Fig. 1C). When PBMC were co-injected with cytokine-neutralizing antibodies, the strong swelling response to col(V) was not affected by isotype control IgG, anti-human IL-2, or anti-human IL-4. However, the col(V) swelling response was significantly blocked by addition of anti-human IFN- γ, IL-17, IL-22, IL-1β, and TNF-α, with dependence on IFN-γ being relatively variable (Fig. 1D). When plasma samples from CAD patients and healthy controls were analyzed for the presence of anti-col(V) antibody, significant differences were not observed (Online Figure II).

Cellular and Humoral Immunity to Col(V) in a Mouse Model of Atherosclerosis

To further explore the extent to which col(V) autoimmunity accompanies and drives atherosclerosis, we assayed for possible col(V) autoimmunity in the apolipoprotein E-deficient (ApoE−/−) mouse model of atherosclerosis ³³. As in human atherosclerosis, early and advanced lesions in ApoE−/− mice are infiltrated by macrophages and CD4⁺ T cells, which contribute to plaque pathogenesis ^{34, 35}, reflecting an immune component apparently conserved between the two species. ApoE−/− mice and control wild type B6 mice were fed a high fat (Western) diet for 15 weeks and all mice were immunized with tetanus and diphtheria toxoid (TT/DT) prior to experimentation. Splenocytes obtained from ApoE−/− mice displayed a TV-DTH col(V) response significantly higher than that of B6 controls (p=<0.0001) (Fig. 2A), and similar in magnitude to the positive control TT/DT-specific response. The response was collagen type-specific, in that DTH responses to col(I), col(II), and to the major basement membrane collagen col(IV) (Fig. 2A), were all significantly lower in comparison to the ApoE−/− col(V) response [p<0.005 for col(I), col(II), and col(IV)]. When splenocytes from col(V) “high –responder” ApoE−/− mice were co-injected with cytokine-specific antibodies, TV-DTH responses to col(V) were blocked by neutralization of murine IFN-γ, IL-17, and IL-1β (Fig. 2B), consistent with dependence on both the Th17 and Th1 pathways, as in most human CAD patients tested (Fig. 1D). However, unlike the human CAD patients, TV-DTH to col(V) was blocked by TNF-α neutralizing antibodies in only 2/7 ApoE−/− mice tested (Fig. 2B). Analysis of serum samples for col(V)-specific antibodies revealed that ApoE−/− mice fed a Western diet had significantly higher levels of anti-col(V) antibodies compared to similarly fed B6 controls (p=0.004, Fig. 2C), indicating involvement of B cells as well as T cells in col(V) autoimmunity in this model.

Figure 2. ApoE−/− mice on a high fat diet have T and B cell autoimmunity to col(V).

ApoE−/− and B6 control mice were fed a high fat diet 15 weeks prior to experimentation. All mice were immunized with TT/DT 2 weeks prior to assay. A, TV-DTH measured T cell responses by splenocytes from ApoE−/− (n=9-21) or B6 control (n=15-21) mice. Responses are against TT/DT, col(V), col(II), col(I), and col(IV). Horizontal bars denote group means. Statistical significance is in comparison to the ApoE−/− col(V) swelling response. B, ApoE−/− mice with a swelling response > 50mm-3 to col(V) (n=7 mice) were selected for analysis. TV-DTH responses were compared between splenocytes injected with col(V) alone or in the presence of neutralizing antibodies to IFN-γ, IL-17, IL-1β, or TNF-α. Statistical significance is in comparison to swelling responses obtained with col(V) alone. C, Serum samples were analyzed for col(V)-specific antibodies (n=10 mice per group). Statistical significance in is comparision to antibody MFI obtained from ApoE−/− mice on a high fat diet. Horizontal bars represent the mean value for each group. *p≤0.05, **p≤0.005, and ***p≤0.0001.

Col(V) is Expressed in Atherosclerotic Lesions of ApoE−/− Mice on a High Fat Diet

Aortic root and arch sections collected from ApoE−/− and control B6 mice fed a Western diet for 15 weeks were analyzed for the presence of atherosclerotic lesions and col(V) content. Representative H&E staining of the aortic sinus displayed the prototypical lesion development normally seen in ApoE−/− mice fed a high fat diet (Fig. 3A), with an absence of lesion development in B6 controls (Fig 3B). As expected, ApoE−/− mice had a significantly higher total lesion area (Fig. 3G) and percentage of vessel occlusion due to plaque burden (Fig. 3H) than did B6 controls, and macrophages were distributed throughout plaques (Fig. 3E) as were lipids (Fig. 3F). Importantly, analysis of advanced lesions from ApoE−/− aortae showed col(V) to be strongly expressed in intimal and medial layer regions of the lesion (Fig. 3C), consistent with a mechanism whereby local intralesional col(V) might contribute to generating the types of anti-col(V) immune responses reported here. In contrast, B6 control aorta lacked detectable col(V) expression in intimal and medial layers (Fig. 3D). Some col(V) expression was detected in B6 aortae, exclusively in vessel adventitia, with broader and more intense col(V) staining detected in the adventitia of ApoE−/− aortae (Figs. 3C and D). It has been suggested previously (Burlingham et al JCI, Ooshima, Summers and Wilkes) that increased expression of col(V) in an inflammatory setting is accompanied by generation of abnormal col(V) forms, likely to be involved in triggering anti-col(V) responses. Generation of such forms may thus accompany the increased col(V) expression in ApoE−/− atheromas (Fig. C), thereby contributing to induction of anti-col(V) responses.

Figure 3. Col(V) is strongly expressed in advanced atherosclerotic plaques from ApoE−/− mice fed a high fat diet.

A, C, E, F: ApoE−/− on a high fat diet. B, D: B6 control on a high fat diet. A & B, Representative cross sections of H&E stained aortic sinuses. C & D, Representative cross sections of Col (V) stained aortic arch sections. The hole within the plaque seen in C is a cholesterol cleft resulting from dissolving out of cholesterol crystals during paraffin processing. E, Representative cross section of CD68+ macrophage stained plaques in aortic sinus. F, Representative cross section of lipid staining using Oil Red O of a plaque in the aortic sinus (P=Plaque; M=Media). G, Quantitative analysis of total lesion area in aortic root sections. H, Quantitative analysis of percent vessel occlusion from plaque burden was performed on aortic roots. All samples collected from mice maintained on a high fat diet for 15 weeks. Each data point represents the mean value obtained from 3 serial aortic sinus sections for each mouse; horizontal bars represent the mean value for each group. Mann-Whitney p value is shown.

ApoE−/− Mice Fed Regular Chow Have an IL-10-regulated Anti-col(V) Autoimmune Response

To determine the degree to which col(V) autoimmunity in ApoE−/− mice was diet-dependent, and thus limited to mice with relatively advanced atherosclerosis on a high fat diet, we assayed for a cellular immune response to col(V) in ApoE−/− mice maintained on regular chow. Splenocytes from ApoE−/− mice fed regular chow had a significantly lower response to col(V) compared to ApoE−/− mice fed a Western diet, with both groups displaying a low response to col(I) (Fig. 4A). To determine whether a low col(V) response in ApoE−/− mice on regular chow was due to absence of col(V)-specific T effector (T_E) cells, or to regulation of an incipient response to this autoantigen, splenocytes from ApoE−/− mice fed regular chow were co-injected in the TV-DTH assay with col(V) or col(I), with or without neutralizing antibodies to TGF- β or IL-10. Whereas TGF-β neutralization only slightly increased the swelling response, IL-10 neutralization significantly uncovered a specific anti-col(V) response, similar in strength to the positive control swelling response to TT/DT (Fig. 4B). B6 age-matched controls maintained on regular chow displayed no DTH swelling response to col(V) either with or without addition of IL-10 or TGF- β neutralizing antibody (data not shown). These data indicate that splenocytes from ApoE−/− mice maintained on regular chow contain col(V)-specific T_E cells; however, these cells are being controlled by IL-10-producing cells.

Figure 4. ApoE−/− mice on a regular chow diet have an IL-10 regulated autoimmune response to col (V).

ApoE−/− mice were fed either a high fat diet or fed a regular chow diet for 15 weeks prior to collection of splenocytes for the TV-DTH assay. A) TV-DTH responses to col(V) (ApoE−/− high fat n=27; ApoE regular chow n=14) and col(I) (ApoE−/− high fat n=9; ApoE regular chow n=14). B)ApoE−/− mice fed a regular chow diet were immunized to TT/DT two weeks prior to splenocyte collection. TV-DTH responses to recall TT/DT, col(V) or col(I) alone, or col(V) or col(I) in the presence of neutralizing antibodies to TGF-b or IL-10. n=14 for assays done with col(V), n=3-9 for assays done with col(I). Data shown as mean +/− standard deviation.

Sensitization to Col(V) Breaks Tolerance and Results in Exacerbated Disease

To determine a possible causal relationship between col(V) autoimmunity and atherosclerosis, we sought to break tolerance via multiple exposures to col(V). ApoE−/− mice maintained on regular chow were treated biweekly with 8 × 50 μg intravenous (i.v.) doses of col(V), a route that we have found to be immunogenic in the ApoE−/− mice. After 15 weeks, splenocytes, serum, and aortic roots were collected and analyzed for col(V) reactivity and disease pathology. Splenocytes showed an increased TV-DTH response to col(V) in the absence of IL-10-mediated inhibition (Online Figure III), consistent with intensified anti-col(V) cellular immunity. Serum analysis confirmed that col(V)-treated ApoE−/− mice were sensitized, as indicated by significantly higher titers of anti-col(V) antibodies compared to untreated ApoE−/− controls (p=0.0007; Fig. 5C), reminiscent of the elevated levels of anti-col(V) antibodies seen in ApoE−/− mice on a high fat diet (Fig. 2C). When aortic root sections were stained for C3d deposition, plaque areas of the ApoE−/− mice sensitized to col(V) had much brighter staining compared to untreated ApoE−/− controls (Fig. 5A – bottom panels), consistent with increased IgG deposition from a heightened humoral response.

Figure 5. Sensitization to col(V) results in exacerbated atherosclerotic disease.

ApoE−/− mice were i.v. treated with 8 doses of 50 μg col(V) at biweekly intervals or were left untreated; all were fed regular chow for 15 weeks. A, Representative H&E stained (40×), CD68 macrophage stained (200X), lipid stained (Oil Red O - 200X), and complement C3d stained (200X) aortic root sections. P-plaque, M-media, and A-adventitia. B, Quantitative analysis of total plaque area was performed on aortic roots taken at age 22 weeks. Each data point represents the mean value of total lesion area from 3 serial aortic sinus sections (n=10 mice per group). C, Serum samples were analyzed for col(V)-specific antibodies (n=10 in each group). Horizontal bars represent mean value for each group. Mann-Whitney p value for comparison is shown.

Importantly, aortic roots of col(V)-sensitized ApoE−/− mice showed markedly exacerbated atherosclerotic burden compared with untreated ApoE−/− mice (Figs. 5A and B). Thus, in support of a causal role for col(V) autoimmunity in the pathogenesis of atherosclerosis, sensitization to col(V) was sufficient to increase atherosclerotic burden in ApoE−/− mice on normal chow. Additionally, although plaques of both sensitized animals and controls had macrophages and lipids distributed throughout (Fig. 5A), immunostaining with antibodies to CD3 revealed that, unlike untreated ApoE−/− controls, col(V)-sensitized ApoE−/− mice had large numbers of T cells infiltrating the root area (Fig. 6A). Col(V)-sensitized ApoE−/− mice also had significantly higher numbers of infiltrating T cells within the adventitia underlying the plaque, as well as in the plaques themselves, compared to untreated ApoE−/− controls, even when normalizing for the lesser plaque area of the latter (Fig. 6B). In fact, plaques of col(V)-sensitized ApoE−/− mice seemed to have cellularity that was somewhat high for murine atherosclerotic lesions. We believe this to reflect the cellular nature of the anti-col(V) immune response, which is accentuated in lesions of col(V)-sensitized ApoE−/− mice Co-staining the aortic root sections with anti-CD68 (green – left panel; Fig. 7) or anti-CD3 (green-right panel; Fig. 7) along with anti-IL-17 (red) and DAPI (blue) revealed that plaque- and adventitia-infiltrating macrophages and T cells in the col(V) sensitized ApoE−/− mice were expressing IL-17 (Figure 7, yellow color), whereas T cells in untreated ApoE−/− control aortic root sections remained negative for IL-17 production (data not shown). We could also detect IL-17-expressing cells (Figure 7, red color) that were neither macrophages nor CD3⁺ T cells throughout the adventitia and plaque area; expression that was markedly more prominent in col(V) sensitized ApoE−/− mice than in untreated ApoE−/− controls (data not shown).

Figure 6. Sensitization to col(V) results in exacerbated atherosclerotic disease.

ApoE−/− mice were i.v. treated with 8 doses of 50 μg col(V) at biweekly intervals or were left untreated; all were fed regular chow for 15 weeks. A, CD3 T cell staining from aortic root sections. Top panels - 100X view of plaque area and associated adventia. Center panels - plaque area at 400X. Bottom panels – underlying adventitia area at 400X. P-plaque, M-media, and A-adventitia. B, Number of infiltrating CD3+ T cells in the adventitial area underlying the plaque (left panel) and in the plaque itself (right panel) calculated from serial aortic root sections (n=5). Horizontal bars represent mean value for each group. Mann-Whitney p value for comparison is shown.

Figure 7. IL-17-producing macrophages and T cells can be found in atherosclerotic plaques and adventitia of col(V)-sensitized ApoE−/− mice.

Representative immunofluorescent staining of plaques in aortic root sections from col(V) i.v. treated ApoE−/− mice maintained on a regular chow. Left panels were stained for CD68+ macrophages (green), DAPI for nuclei (blue), and IL-17 (red). Right panels were stained for CD3+ T cells (green) DAPI for nuclei (blue), and IL-17 (red). P=plaque; M=media; and A=adventitia.

Discussion

Although high plasma LDL cholesterol and other hemodynamic factors are clearly involved in atherogenesis, atherosclerosis is a chronic inflammatory disease with an important immune component occurring at each stage of the disease ^{1, 2, 5, 36, 37}. It has become apparent that the pro-inflammatory process that affects plaque growth and destabilization, and which also perpetuates and amplifies the inflammatory state, involves autoimmune responses that develop in the lesion against local self-antigens. To date, the self-antigens have comprised HSP 60/65 ^{10, 11} and oxLDL ^{2, 7-9}. Here we demonstrate col(V) to be an autoantigen characteristic of atherosclerosis in humans and mice.

Previous studies demonstrated anti-col(V) autoimmunity to play a causal role in lung transplant rejection in rat and to be a critical component of BOS, the chronic inflammatory/fibro-obliterative process causal in rejection of most human lung transplants ^{21, 38}. Col(V) is a minor fibrillar collagen, broadly distributed in tissues as α1(V)₂α2(V) heterotrimers buried in the interior of fibrils of the much more abundant col(I) ²⁵. Because of this sequestration, col(V) has the potential to contain epitopes that are normally immunologically masked ²⁵. The ease of immunohistochemical detection of normally masked col(V) epitopes in rejected lung transplants ^{21, 38} and the finding that anti-col(V) immunity in BOS is directed exclusively against α1(V) chains ²¹ suggested a model in which aberrant α1(V)₃ homotrimers, excluded from the interior of col(I) fibrils ²⁷, might be involved in neoepitope formation and establishment of anti-col(V) immunity. In the context of this model, it was of great interest that α1(V) chains, capable of forming homotrimers, are specifically up-regulated in human atheromas ²⁴. Taken together, the above observations prompted the current study into the possibility that col(V) may serve as an autoantigen in the immune responses that underlie significant pathogenesis in atherosclerosis. The finding of α1(V)-specific immune responses in CAD patients supports a possible role for α1(V)₃ homotrimers in neoepitope formation, although it is also possible that epitopes on sequestered α1(V)₂α2(V) heterotrimers are exposed via fragmentation of the extracellular matrix (ECM) by matrix metalloproteinases (MMPs), highly active in both BOS ³⁸ and atheromas ⁹, and that α1(V), but not α2(V) epitopes, induce autoimmunity. We also have not formally excluded the possibility that some col(V) neoepitopes may be created via oxidative modification by reactive oxygen species (ROS), although all anti-col(V) responses described in the current report were demonstrated using non-oxidized col(V).

Here we demonstrate that CAD patients have cellular autoimmune responses to col(V) in the absence of responses against the much more abundant fibrillar collagens, col(I) and col(II), or against α2(V) chains. Similarly, atherosclerotic ApoE−/− mice on a high fat diet had highly significant and specific cellular and humoral immune responses to col(V), whereas B6 controls lacked such responses. As in human CAD, ApoE−/− anti-col(V) responses were highly specific, with no significant responses detected against the major collagen types I, II, and IV. Moreover, advanced lesions in ApoE−/− mice on a high fat diet showed strong, readily detectable col(V) staining, demonstrating the auto-antigen to be present in the mouse lesions, as previously shown for human plaques ²⁴. Thus, T cell and monocyte-dependent anti-col(V)-specific cellular autoimmunity is a feature not only of human CAD, but also of atherosclerotic-prone mice, suggesting col(V) autoimmunity to be a consistent feature of atherosclerosis. Ooshima ²⁴ has suggested that the increased α1(V) chains of human atheromas are likely produced by intralesional smooth muscle cells. As smooth muscle cells also appear in the atheromas of ApoE−/− mice ³⁹, such cells are reasonable candidates as the source of col(V) that elicits autoimmune reponses in this system as well. Humoral autoimmunity may be more prevalent in emerging than in late-stage atherosclerosis, possibly accounting for the discrepancy between human (assayed at end-stage CAD) and mouse (assayed at relatively early stages of atherosclerosis) anti-col(V) antibody results.

Our previous analysis of patients awaiting lung transplantation for end-stage pulmonary disease resulted in the surprising finding that the majority of patients with idiopathic pulmonary fibrosis (IPF) were strongly positive for Th17-dependent col(V)-specific autoimmunity. This finding was unique to IPF, as COPD, CF, and AAD patients rarely had pre-transplant cellular immunity to col(V) ²³. Thus, it is of great interest that IPF patients who develop CAD have a significantly more rapid onset of CAD-related morbidity and mortality compared with the clinical course of CAD in COPD subjects ⁴⁰. These findings are thus also supportive of a role for col(V) autoimmunity in CAD.

In the present study, human CAD anti-col(V) cellular immune responses were dependent on CD4⁺ T cells, IL-17, and IL-22, the latter a cytokine secreted by Th17 CD4⁺ T cells and by Th22 CD4⁺ T cells, which can play roles in chronic inflammation ⁴¹. Also, IL-17 induces rapid production of TNF- α and IL-1β in human monocytes, a cell type known to be a critical promoter of the immunoinflammatory response in atherosclerosis ⁴², and to be required for some anti-col(V) TV-DTH responses ^{21, 23}, and which is apparently required for anti-col(V) TV-DTH responses in CAD patients as well (Fig. 1C). In previous studies of lung transplant patients with BOS or PGD, anti-col(V) cellular responses required IL-17, TNF-α and IL-1β, but not IFN-γ, suggesting dependence upon the Th17, but not the Th1 pathway ^{21, 23}. While 1/5 CAD patients in the present study had such a cytokine pattern, some degree of dependence of the anti-col(V) cellular response on IFN-γ was found in the other 4 patients tested. In ApoE−/− mice, anti-col(V)-specific cellular responses were also dependent on both IL-17 and IFN-γ, with more variable involvement of IL-1β and TNF-α than in human CAD. CAD has previously been shown to be associated with increased plasma levels of IL-17 ^{17, 18} and of IL-6 and TGF-β ^{17, 43}, two cytokines known to work in concert to promote Th17 cell differentiation. The finding of IL-17-producing cells in the plaques and especially in the subjacent adventitia of col(V) sensitized mice is consistent with a potential role for Th17 cells in the progression of atherosclerosis. However, the relative importance to anti-col(V) cellular responses of IL-17 secreted in the plaque itself or secreted by cells in lymphoid organs remains to be determined. In addition, dependence of anti-col(V) cellular immunity on IFN-γ in some CAD patients and in most ApoE−/− mice assayed suggests that col(V) autoimmunity in atherosclerosis may involve elements of the Th1 pathway as well. Involvement of elements from both Th17 (IL-17, IL-22) and Th1 (IFN- γ) cells, as well as the monocyte/macrophage product IL-1 β, is consistent with previous observations that both pathways are capable of inducing autoimmunity and that Th17 and Th1 pathways are not always completely separable; as some T cell clones infiltrating human atherosclerotic lesions are capable of producing both IL-17 and IFN- γ ¹⁶. Thus the relative importance of elements of the Th1, Th17, and/or Th22 pathways in promoting atherosclerosis remains to be elucidated.

Although ApoE−/−mice can develop atherosclerosis even in the absence of T cells ^{44, 45}, it has also been clearly shown that, in the context of an intact immune system, T cells are key in regulating the pathogenesis of atherosclerotic plaques ^{42, 46}. Dysfunction or reduction of T_R cells and resulting imbalance of the T_R:T_E ratio has been considered an important cause of autoimmunity ¹⁸, and may contribute to plaque inflammation and development in atherosclerosis ⁴⁷. ApoE−/− mice fed regular chow develop atherosclerotic lesions, albeit at a much slower rate than when fed a high fat diet. In the present study, we’ve shown that ApoE−/− mice maintained on regular chow have IL-10-mediated regulation of anti-col(V) immunity. We hypothesize that this is due to modulation of T_E cells involved in col(V) autoimmunity by IL-10-produced by monocytes/macrophages, and/or T_R cells in early stages of plaque formation. IL-10 deficiency in ApoE−/− mice is known to promote early atherosclerotic lesion formation, characterized by increased infiltration of activated T cells and production of pro-inflammatory cytokines⁴⁸. Furthermore, adoptive transfer of IL-10-producing T_R (Tr1) cells can down-regulate atherosclerosis-associated pathogenic immune responses, leading to decreased plaque development and inflammation in ApoE−/− mice⁴⁹. We speculate that the IL-10-mediated regulation of col(V)-specific autoimmunity seen in ApoE−/− mice on regular chow contributes to slowed atherogenesis, which can be overcome by increased hypercholesterolemia due to high fat diet, or by sensitization of ApoE−/− mice to col(V), in each case resulting in chronic inflammation.

An exciting aspect of the identification of pro-atherogenic autoantigens is the potential for specific immunomodulatory therapeutic approaches centered on such autoantigens in attempts to prevent and treat disease. Animal studies have provided proof-of-concept results, indicating that it is possible to ameliorate atherosclerotic burden by use of vaccines and via mucosal immunization against oxLDL and HSPs, and clinical testing for other vaccines and immunomodulatory treatments for atherosclerosis are expected in the near term ^{1, 2, 5, 50}. Furthermore, induction of tolerance to col(V) has already been shown to prevent lung transplant acute rejection in a rat model ^{51, 52}. Here, we not only demonstrate an association between col(V) autoimmunity and atherosclerosis in humans and mice, but provide evidence that col(V) autoimmunity can play a causative role in the pathogenesis of atherosclerosis, as sensitization of ApoE−/− mice on normal chow to col(V) is sufficient to increase atherosclerotic burden. Thus, the observations presented herein may represent a first step toward applying immune tolerance approaches to prevent/ameliorate atherosclerosis. As col(V) is a fibrillar collagen, another important area for future study is a possible role for col(V) autoimmunity in fibrous cap thinning and subsequent plaque rupture leading to thrombo-embolic events and acute end organ ischemia, including myocardial infarction and stroke.

Novelty and Significance.

What is known?

• Atherosclerotic lesions contain many cells of both innate and adaptive immunity Autoimmunity to self-antigens contributes to chronic inflammation and progression of atherosclerosis.

• The α1(V) chain form of collagen is abundant in human atherosclerotic plaques.

What new information does this article contribute?

• Collagen type V is a self-antigen that is present in atherosclerotic lesions of human coronary arteries and ApoE deficient mice, providing an association with development of the disease.

• Sensitizing mice to collagen V increases atherosclerotic lesion size demonstrating a causative role for collagen V autoimmunity in atherosclerosis.

Chronic inflammation affects the growth and composition of atherosclerotic lesions potentially through autoimmune responses that develop against self-antigens. We hypothesized collagen type V to be a candidate self-antigen, as it has previously been shown to be specifically upregulated in human atherosclerotic lesions. Furthermore anti-collagen type V immune responses have been associated previously with other disease processes in humans and rodents. This study provides evidence that anti-collagen type V autoimmunity is present in atherosclerosis in both humans and mice, and exerts a causal role in the formation of experimental lesions. These data represent a first step towards developing immune tolerance approaches to reducing atherosclerosis.

Non-standard Abbreviations and Acronyms

oxLDL

oxidized low density lipoprotein

HSP

heat shock protein

Col(V)

collagen type V

Col(I)

collagen type I

Col(II)

collagen type II

BOS

bronchiolitis obliterans syndrome

PBMC

peripheral blood mononuclear cells

CAD

coronary artery disease

CABG

coronary artery bypass grafting

TV-DTH

trans-vivo delayed-type hypersensitivity

HDL

high density lipoprotein

ApoE−/−

apolipoprotein E null

TT/DT

tetanus and diphtheria toxoid

T_R

T regulatory

T_E

T effector

ROS

reactive oxygen species

PGD

Primary Graft Dysfunction