Abstract
Dilated cardiomyopathy is a severe pathology of the heart with poorly understood etiology. Disruption of the gene encoding the negative immunoregulatory receptor PD-1 in BALB/c mice, but not in BALB/c RAG-2−/− mice, caused dilated cardiomyopathy with severely impaired contraction and sudden death by congestive heart failure. Affected hearts showed diffuse deposition of immunoglobulin G (IgG) on the surface of cardiomyocytes. All of the affected PD-1−/− mice exhibited high-titer circulating IgG autoantibodies reactive to a 33-kilodalton protein expressed specifically on the surface of cardiomyocytes. These results indicate that PD-1 may be an important factor contributing to the prevention of autoimmune diseases.
Dilated cardiomyopathy is a chronic disorder of the heart muscle characterized by a poorly contractile and dilated ventricle. The diagnosis is based primarily on clinical criteria and on the exclusion of identifiable underlying causes (1). Consequently, patients with dilated cardiomyopathy represent a heterogeneous group affected to varying degrees by genetic, viral, immunological, and environmental factors (2). This has complicated identification of the underlying pathogenic mechanisms operative in this disease. Involvement of an immune mechanism in patients with dilated cardiomyopathy is still controversial (3), although a certain fraction of patients do possess antibodies against self antigens (4–6). We examined the potential contribution of an autoimmune pathway in a spontaneous model of dilated cardiomyopathy.
When bred on the BALB/c background (7), PD-1−/− mice started to die as early as 5 weeks of age (Fig. 1A). By 30 weeks, two-thirds of PD-1−/− and 10% of PD-1+/− mice had died, whereas all of the PD-1+/+ controls survived. In contrast, premature death was not observed in either BALB/c PD-1−/−–RAG2−/− or B6–PD-1−/− mice (8). Diseased PD-1−/− mice exhibited protrusion of eyeballs a few weeks preceding death, and autopsy examination revealed that all diseased mice exhibited massively enlarged hearts (Fig. 1B) and varying degrees of hepatomegaly. Collectively, these features suggested that the cause of death was congestive heart failure. Histological examination (9) revealed that the right ventricular walls of PD-1−/− mice were thinner than those of the control mice, and that both ventricles were dilated about twofold in diameter (Fig. 1C). Sporadic fibrotic reaction, including increased cellularity and interstitial fibrosis with scar formation (Fig. 1C, right), was also observed, although the ventricular walls appeared otherwise grossly normal, without any apparent infiltration of mononuclear cells. Electron microscopic examination (9), however, revealed the scattered degeneration of cardiomyocytes with disarrayed and disrupted myofilaments and irregularly shaped mitochondria throughout the ventricular walls [Fig. 1D and Web fig. 1 (10)].
Development of dilated cardiomyopathy in BALB/c–PD-1−/− mice leading to heart failure. (A) Survival curves for PD-1+/+, PD-1+/−, PD- 1−/−, and PD-1−/−RAG-2−/− mice. (B) Representative images of hearts from PD-1+/+ (left) and PD-1−/−(right) mice. (C) Hematoxylin and eosin staining of the mid-transverse section of hearts from PD-1+/+ (+/+) (left) and PD-1−/− (−/−) mice (middle) shown at the same magnification. LV, left ventricle; RV, right ventricle. (Right) The ventricular region of PD-1−/− mice (−/−) (original magnification, ×40). Arrows indicate cellular proliferation, consisting mainly of fibroblasts. (D) Electron micrographs of hearts from PD-1+/+(+/+) and PD-1−/− (−/−) mice. Arrowheads indicate the degenerated cardiac myocyte apposed by the relatively intact ones. MF and Mc indicate myofilaments and mitochondoria, respectively. Original magnification, ×8000.
To evaluate heart function, we performed transthoracic echocardiography (11) on diseased, as well as PD-1+/+ control, mice. Ventricular cavities of PD-1−/− mice, in particular those of the right ventricles, appeared greatly dilated, and their wall thickness was markedly reduced as compared with those of PD-1+/+ mice (Fig. 2A, “2D”). The movement of cardiac walls of the left ventricle and interventricular septum was decreased at both diastole (LVDd) and systole (LVDs) in PD-1−/− mice (Fig. 2A, “M-mode”), and the ventricular fractional shortening (11), a measure of systolic function, was greatly reduced, from 71.9% (PD-1+/+) to 14.9% (PD-1−/−) (P < 0.005) (Fig. 2B). These findings indicated that the pump function of the dilated hearts of PD-1−/− mice was severely impaired in a manner consistent with dilated cardiomyopathy (12).
Dilation and impaired movement of hearts from PD-1−/- mice as revealed by transthoracic echocardiography. (A) Echocardiographic analyses of heart from PD-1+/+ (upper) and PD-1−/− (lower) mice. Representative images of 2D (left) and M-mode (right) analyses are shown (11). Bars in M-mode images indicate left ventricular end-diastolic (left) and end-systolic (right) dimensions. IVS, interventricular septum; LV, left ventricle; RV, right ventricle; LVPW, left ventricular posterior wall. (B) Indices of heart movement based on the echocardiographic analysis of PD-1+/+ (n = 5) and PD-1−/−(n = 4) mice. LVDd, left ventricular end-diastolic dimension; LVDs, left ventricular end-systolic dimension; and %FS, percent fraction shortening. Statistical significance (P value) was evaluated by Student's t test.
Because all PD-1−/− mice bred on the BALB/c–RAG-2−/− background remained healthy (Fig. 1A), we attributed development of the heart disease in BALB/c–PD-1−/− mice to the functions of T and/or B lymphocytes. Consistent with this notion, preliminary studies revealed that disease could be successfully transferred into RAG-2−/− mice with spleen or bone marrow cells from diseased mice. In these experiments, recipient mice that developed cardiomyopathy again exhibited high-titer autoantibodies of equivalent specificity to those observed in the original animals (13). We therefore examined PD-1−/− mice for signs of heart-specific immune reactions. Immunofluorescent analysis (9) revealed the linear deposition of immunoglobulin G (IgG), but little IgM, together with C3 complement surrounding the cardiomyocytes in the affected PD-1−/− hearts, whereas no significant IgG deposition was detected in PD-1+/+ hearts (Fig. 3A). The isotype of the deposited IgG was predominantly IgG1 (14). IgG deposition was observed diffusely throughout the entire cardiac wall regardless of the presence of tissue damage. Little IgG deposition could be detected in other organs, including renal glomeruli (14). Immunoelectron microscopic observation (15) indicated that virtually all myocytes in the affected PD-1−/− hearts were surrounded by a large number of the IgG-specific immunogold particles, which were located both on the surface plasma membrane and in the extracellular matrix. In contrast, no immunogold particles were detected around the wild-type cardiomyocytes (Fig. 3B). These results suggest that IgG deposition may result from antibody binding to heart-specific surface antigens.
(A) Deposition of IgG and C3 complement in the affected hearts of PD-1−/− mice. Heart sections from PD-1+/+ (+/+) or PD-1−/− mice (−/−) were directly stained with FITC-labeled anti-mouse IgM (left), anti-mouse IgG (middle panels), or anti-C3 complement (right). Actin was counterstained with rhodamine-phalloidin (middle right). Bars, 100 μm. (B) Ultrastructural localization of the mouse IgG deposition around cardiomyocytes. M, myocytes. Arrow, plasma membrane; arrowhead, IgG deposition on plasma membrane. Bars, 1 μm.
To confirm the presence of an autoimmune reaction against the heart, we examined mice for the presence of autoantibodies specific for heart tissue (16). Sera from all of the diseased PD-1−/− mice (17 out of 17) exhibited high-titer IgG reactive to a 33-kD protein in the normal heart extract (Fig. 4A). In contrast, autoantibodies were not detected in the sera from age-matched PD-1+/+ mice (0 out of 7). Similarly, sera from PD-1−/− mice without macroscopic cardiomegaly did not exhibit comparable reactivity to the 33-kD protein (0 out of 21), except at very high concentrations. Reactivity was also not detected in B6 or B6-PD-1−/− mice (14). Thus, the presence of high-titer IgG autoantibodies specific for the 33-kD protein in the heart correlated strongly with the clinical manifestation of dilated cardiomyopathy. The 33-kD autoantigen appeared to be specifically expressed in the heart tissue, because it was not detected in other tissues such as liver, kidney, or skeletal muscle in sera from BALB/c–PD-1−/−mice with the disease (Fig. 4B).
Development of circulating autoantibodies against the heart tissue in diseased PD-1−/− mice. (A) IgG autoantibodies were examined against the normal heart extract with sera from BALB/c–PD-1+/+ mice, BALB/c–PD-1−/− mice with dilated hearts, and BALB/c–PD-1−/− mice without the disease. Normal BALB/c heart extract was subjected to electrophoresis in 12% SDS-PAGE, blotted onto nitrocellulose filters, and probed with sera (1:300) followed by incubation with anti-mouse IgG (13). Representative results for each group are indicated. Arrowhead, 33-kD protein. (B) Immunoreactivity of sera from age-matched BALB/c–PD-1+/+ and BALB/c–PD-1−/− mice was examined against tissue extracts of heart, liver, kidney, and skeletal muscle as in (A). (C) Isotypes of serum autoantibodies specific for the heart 33-kD antigen in PD-1−/− mice. The experiments were done as in (A), except that anti-mouse IgG was used as the secondary antibody. (D) Elution of autoantibodies from the affected hearts. Heart extracts from normal BALB/c and affected PD-1−/−mice were precipitated with protein G–beads; the precipitates were eluted with 0.1 M citric acid (pH 3.0), followed immediately by neutralization (3 M NaCl, 1.5 M glycine, pH 8.9). Eluate from the PD-1−/− mouse contained abundant IgG relative to that from the BALB/c mouse. Immunoreactivity of the eluates was examined against normal heart extracts as in (A) along with serum from the same PD-1−/− mouse.
Occasionally, we observed low but measurable titers of autoantibodies against the 33-kD protein in apparently healthy PD-1−/− mice, which developed the disease within about a month. The results also suggest that autoantibodies against the 33-kD protein may be a cause of the disease rather than the result of tissue damage. No other common autoantibodies, including antibodies to double-stranded DNA, were detected in the sera of PD-1−/−mice. The autoantibody was predominantly of the IgG1 subclass in all affected PD-1−/− mice (7 out of 7) (Fig. 4C), consistent with the immunostaining analysis of the heart. IgG deposited in the affected heart tissue was collected by immunoprecipitation from tissue extract with protein G–beads, and by acid elution. The eluted IgG reacted to the specific 33-kD protein, which appeared to be identical to that detected in sera from diseased PD-1−/−mice (Fig. 4D). These results present direct evidence that circulating autoantibodies are specifically associated with heart tissue in diseased mice. To characterize the properties of the 33-kD antigen recognized by autoantibodies, we purified the antibodies bound to the 33-kD protein and used them for staining isolated cardiomyocytes (17). Autoantibodies specific for the 33-kD protein stained the surface of cardiomyocytes with a dotted pattern that, to some extent, resembled the staining pattern observed with wheat-germ agglutinin, which binds to the transverse (T) tube (18) [Web fig. 2 (10)]. Because there are no known surface proteins of 33 kD on cardiac muscle, it is premature to conclude that the 33-kD antigen is associated with the T tube. However, it is reasonable to speculate that this protein is involved in cardiac function and that inhibition of such functional molecule by antibody binding might cause dysfunction of cardiac muscle.
Engagement of the PD-1 receptor with its membrane-bound ligand (PD-L1) of the B7 family has recently been shown to inhibit the proliferation of anti-CD3–stimulated T cells as well as anti-IgM–stimulated B cells (14, 19), indicating that PD-1 is a negative immuno-regulator of activated lymphocytes. Unlike other B7-related members, PD-L1 is detected not only in antigen-presenting cells, but also constitutively in the heart and kidney (14, 19). Thus, it is possible that cells in the heart can directly down-regulate the proliferation of autoreactive lymphocytes, ensuring a final safeguard against autoreactivity within peripheral tissues (20). Because B6–PD-1−/− mice show distinct autoimmune symptoms from those seen in BALB/c–PD-1−/− mice (8), the present results reinforce the notion that dysfunction of PD-1 may underlie distinct types of autoimmune diseases, which are dependent on other background genetic factors.
Dilated cardiomyopathy is a progressive and life-threatening disease, and little effective therapy is currently available. The results presented here raise the possibility that some forms of cardiomyopathy may have an autoimmune basis, and identification of possible autoantigen(s) may open new therapeutic approaches for this significant disease.
↵* To whom correspondence should be addressed. E-mail: honjo{at}mfour.med.kyoto-u.ac.jp
