Antigenic stimulation in the simian model of HIV infection yields dilated cardiomyopathy through effects of TNF[alpha]

Objective: To investigate a role for endogenous myocardial cytokine production in the development of HIV-associated cardiomyopathy.

Design: Cardiomyopathy is a late-stage sequela of HIV infection. Although pathogenesis of this condition in HIV infection is poorly defined, inflammatory cytokines are recognized for their detrimental effects on myocardial structure and function. HIV infection is characterized by chronic immune activation and inflammatory cytokine dysregulation. As the myocardium itself is a rich potential source of inflammatory cytokines, HIV-mediated cytokine dysregulation may be an important contributor to development of HIV cardiomyopathy. An antigenic stimulation protocol conducted in the simian immunodeficiency virus (SIV) model of HIV infection was used to study the effects of endogenous cytokine production on myocardial structure and function.

Methods: Twenty-six rhesus monkeys were assigned to treatment groups for a 35-day study. Animals were SIV-infected; SIV-infected and treated with killed Mycobacterium avium complex bacteria (MAC); SIV-infected, MAC-treated, and given the TNFα antagonist etanercept; or uninfected and MAC-treated. All animals were subjected to weekly echocardiographic studies. Hearts were collected for further evaluation at euthanasia.

Results: SIV-infected, MAC-treated animals developed significant systolic dysfunction [left ventricular ejection fraction (LVEF) decline of 19 ± 2%] and ventricular chamber dilatation [left ventricular end-diastolic diameter (LVEDD) increase of 26 ± 6%] not seen in other groups. Concurrent treatment with etanercept prevented development of these changes, implicating a causative role of myocardial TNFα.

Conclusions: SIV-infected animals develop exaggerated myocardial pathology on stimulation with the ubiquitous environmental agent MAC. These responses are TNFα-dependent and may play a significant role in the development of cardiomyopathy in HIV infection.

Introduction

Ventricular dysfunction and dilated cardiomyopathy are well documented sequelae of late-stage HIV infection ^[1-4]. Factors influencing the development of myocardial pathology in HIV infection are at present poorly defined. Hypothesized mechanisms have included cytokine-induced effects, tissue damage resulting from myocarditis, drug-induced cardiotoxicities, and effects of viral proteins ^[5-11]. Although current evidence suggests that highly active antiretroviral therapy (HAART) has reduced the incidence of clinically significant cardiac disease among HIV-infected people, such treatment is available to only a minority of those in need ^[12,13]. In addition, HIV infection provides a defined venue for exploring the role of chronic immune activation and host inflammatory response in the development and progression of myocardial injury, features which have relevance beyond the context of HIV-infection itself.

The myocardial tissue environment is a rich potential source of inflammatory cytokines, with both cardiomyocytes and local noncardiomyocyte cell populations competent to produce a variety of inflammatory mediators, and heart tissue capable of generating as much or more TNFα per gram of tissue in response to endotoxin stimulation as liver or spleen ^[14-17]. Nonmyocyte populations comprise up to 70% of the total cellular constituency of the myocardium and consist of a mixed assemblage of cell types, including substantial populations of dendritic cells and macrophages ^[14,18-21]. This represents a volatile setting in the context of HIV infection, which is intrinsically characterized by chronic immune activation and inflammatory cytokine dysregulation ^[22,23]. Inflammatory cytokine-induced myocardial dysfunction is a well documented phenomenon in multiple experimental models, is a significant contributor to hemodynamic compromise in sepsis, and may play a contributory role in development and progression of heart failure regardless of initiating etiology ^[24-29]. Mechanisms of cytokine-induced contractile dysfunction are complex and multifactorial ^[24].

The study of the pathogenesis of HIV-associated cardiomyopathy (HIVCM) in humans is limited by many factors. Identifying the earliest time points of development of myocardial pathology and placing them within the natural history of HIV infection is generally not possible; complex, variable, and potentially toxic medication regimens are routinely used, and there may be substantial variation in environment and lifestyle among patients. The simian immunodeficiency virus (SIV) model of HIV infection is well established, has been extensively characterized, and provides a strong context for study of HIVCM in that dilated cardiomyopathy and histologic myocardial lesions similar to those documented in HIV infection are frequently seen in chronically SIV-infected rhesus monkeys, implying a shared pathogenesis ^[18,30].

In the present study, an acute SIV infection model employing recurrent antigenic stimulation with whole, heat-killed Mycobacterium avium complex (MAC) bacteria was used to evaluate patterns of early SIV-associated myocardial injury. Pathogenesis studies focused on early time points postinfection allow evaluation of changes when viral loads are high, sufficient immune competence is retained to prevent opportunistic infections, and effects of viral replication or viral determinants can be examined without complication by effects of differential disease progression. MAC bacteria are environmentally ubiquitous and disseminated MAC infections are among the most common opportunistic infections encountered in AIDS patients ^[31-33]. The MAC stimulation protocol applied represents an experimentally induced augmentation of normal environmental antigenic stimulation, using an organism with a high degree of clinical relevance to late-stage HIV and SIV infection.

Methods

Animals and study design

Twenty-six male rhesus macaques (Macaca mulatta) aged 2-4 years were housed at the New England Primate Research Center (NEPRC) in a biolevel 3 animal-containment facility in accordance with standards of the Association for Assessment and Accreditation of Laboratory Animal Care and Harvard Medical School's Animal Care and Use Committee. Prior to initiation of experimental protocols, all animals tested negative for infection with simian retrovirus type D, SIV, simian T-lymphotropic virus-1, and herpes B virus. Animals were divided into four cohorts for study over a 35-day period. Eighteen animals were subjected to infection with uncloned, pathogenic SIVmac251 [25 ng p27 antigen in 1 ml sterile phosphate buffered saline (PBS), i.v.] at day 0. Of these, 12 animals received four doses of heat-killed MAC bacteria (10⁷ CFU-equivalents in 1 ml sterile PBS per administration, i.v.) with the first dose given at day 7, then repeated on days 12, 14, and 21. Of the 12 SIV-infected, MAC-treated animals, four were treated with the TNFα antagonist etanercept (Enbrel; Amgen, Thousand Oaks, California, USA) at 0.4 mg/kg, i.m. every other day for the duration of the study. A control group of eight SIV-uninfected animals was administered MAC treatments as described on the same schedule as the SIV-infected, MAC-treated animals. Blood drawn from all animals before virus inoculation at day 0, and weekly thereafter was evaluated for circulating cytokine and chemokine levels, peripheral T-cell subset composition, and plasma viral load evaluation as appropriate. Echocardiograms were performed weekly, including prior to SIV inoculation on day 0, and prior to MAC treatments on days 7, 14, and 21. All echocardiographic studies were conducted under ketamine sedation (10-15 mg/kg, i.m.). All SIV-infected animals and four of the eight uninfected, MAC-treated animals were euthanized at the end of the study (30 mg/kg pentobarbital sodium, i.v., followed by 2 mEq/kg potassium chloride, i.v.). Complete gross and microscopic post-mortem examinations were performed on all euthanized animals. Hearts were removed with myocardial tissue snap frozen and collected into formalin within 15 min of confirmation of death.

Mycobacterium avium complex inocula

The MAC isolate used (no. 88415) was derived from a case of spontaneous disseminated mycobacterial disease in a rhesus monkey with simian AIDS, and retains the ability to generate disseminated infections in SIV-infected animals when not inactivated ^[34]. Frozen stocks of the isolate were grown out in Middlebrook 7H9 broth at 37°C and 5% CO₂ for 8 days, titered, diluted to 10⁷ CFU/ml in PBS, and killed by immersion in boiling water for 2 min. One milliliter of the heat-treated preparation was administered intravenously per animal per MAC treatment. The inoculum used produced no systemic signs of illness in treated animals as determined by clinical veterinary staff.

Echocardiography

M-mode and two-dimensional echocardiograms were performed using a Hewlett Packard Image Point HX ultrasound machine and 5 MHz transducer. Standard parasternal long-axis and short-axis views as well as standard two-chamber and four-chamber apical views were obtained and recorded on VHS tape for later analysis using ImageView DCR (Nova Microsonics, Mahwah, New Jersey, USA) and modified American Society of Echocardiography volumetric analysis and regional and global wall motion scores. Heart rate (HR) was recorded using lead II of standard ECG. Analyses were conducted blinded to animal group allocations.

Enzyme-linked immunosorbent assays

Commercial enzyme-linked immunosorbent assay (ELISA) kits for soluble TNF receptors 1 and 2 (sTNFR1 and sTNFR2), monocyte chemoattractant protein-1 (MCP-1), and interleukin-18 (IL-18; R&D Systems, Minneapolis, Minnesota, USA) were used on batched samples of previously frozen plasma according to manufacturers' instructions. Evaluation of sTNFR2 levels in animals treated with etanercept was not possible because of cross-reactivity of the assay with etanercept itself.

Western blot

Membrane fractions prepared using differential centrifugation of homogenates of snap frozen myocardial apex and semitendinosus muscle collected at necropsy were subjected to electrophoretic separation by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and electrotransferred onto polyvinylidene fluoride (PVDF) membrane (Immobilon-P^SQ; Millipore, Bedford, Massachusetts, USA). Nonspecific protein binding was blocked and membranes were probed with primary antibody (anti-TNFα, AB-210-NA and clone 28410, R&D; anti-iNOS, clone 54; BD Biosciences, San Jose, California, USA). Blots were washed and incubated with secondary antibody conjugated to horseradish peroxidase (A5420; Sigma Aldrich, Milwaukee, Wisconsin, USA; sc-2005; Santa Cruz Biotechnology, Santa Cruz, California, USA). Immunoreactive proteins were detected by chemiluminescence (PerkinElmer, Boston, Massachusetts, USA) and exposed to X-ray film (Hyperfilm ECL; Amersham Pharmacia Biotech, Piscataway, New Jersey, USA). Densitometric analysis of bands was carried out using a Personal Densitometer SI and Image QuanNT Software (Molecular Dynamics, Sunnyvale, California, USA).

Progression of simian immunodeficiency virus infection

Plasma viral loads were determined by quantitative reverse transcription PCR, as described previously ^[35]. T-cell subsets and total lymphocyte counts were monitored weekly throughout the study period.

Histologic examination and immunohistochemistry

Left and right ventricular free wall, interventricular septum, left and right atria, and aortic outflow tract were histologically examined for each animal. Immunostaining was performed as previously described, using an avidin-biotin complex method with diaminobenzidine (DAB; DakoCyomation, Carpineteria, California, USA) as chromogen ^[18]. Tissues were evaluated using antibodies specific for CD3 (A0452; DakoCytomation), cleaved caspase 3 (9661L; Cell Signaling Technology, Beverly, Massachusetts, USA), SIV nef [clone KK75, donor Dr K. Kent and Ms C. Arnold, from NIBSC Centre for AIDS Reagents supported by EU Programme the European Vaccine Against AIDS (EVA) contract (QLKZ-CT-1999-00609) and the UK Medical Research Council], HIVp24/SIV p27 (clone 183-H12-5C, obtained through NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH: from Dr Bruce Chesebro and Kathy Wehrly), rhesus cytomegalovirus immediate-early (IE)1 protein (polyclonal, provided by Dr Peter Barry, University of California at Davis), and adenovirus (clone 20/11; Chemicon International, Temecula, California, USA) ^[36].

Tissue scoring

Myocardial tissue was scored in a blinded fashion for lymphocytic infiltration using a grading schema applied to sections immunohistochemically labeled for CD3, as previously described ^[18].

Statistical analysis

Significance of differences between groups was determined using one-way ANOVA and Kruskall-Wallis ANOVA on ranks as appropriate, with posthoc pairwise comparison by Holm-Sidak or Dunn's method, respectively. Linear regressions were conducted to determine significance of correlations. All analyses were conducted with commercially available software (SigmaStat 3.1; Systat Software, Richmond California, USA). Probability values of P less than 0.05 were interpreted as significant.

Results

Mycobacterial antigenic stimulation yields biventricular dilatation and myocardial dysfunction in simian immunodeficiency virus-infected rhesus monkeys which is preventable by TNFα blockade

Significant biventricular chamber dilatation developed among SIV-infected, MAC-treated animals (SIV + MAC group) by day 35 as evidenced by increased right and left ventricular end-diastolic and end-systolic diameters relative to baseline (P < 0.001; Table 1). Chamber diameters of MAC-treated, uninfected controls (MAC group) also showed significant changes relative to baseline (P < 0.05), but these were mild and consistent with dilatation only in the right ventricle (Table 1). Significant changes in chamber diameter were absent in SIV-infected animals not treated with MAC (SIV group; Table 1).

Table 1

Significant decreases in systolic function developed in SIV + MAC animals over the course of the study period as identified through declines in right and left ventricular fractional shortening and in left ventricular ejection fraction relative to baseline values (P < 0.05 right ventricular change; P < 0.001 left ventricular changes; Table 1; Fig. 1c, d). Such changes were absent in animals of the SIV group, and changes in the MAC group though significant, were mild and limited to declines in left ventricular fractional shortening (P < 0.05; Table 1). In keeping with the structural and functional changes, which occurred in members of the SIV + MAC group, a mild but statistically significant (P < 0.05) increase in HR also developed in these animals, which did not occur in other groups (Table 1). Strikingly, treatment of SIV-infected, MAC-treated animals with the TNFα antagonist etanercept was completely protective against development of both chamber dilatation and systolic dysfunction (Table 1).

Fig. 1

Different kinetics characterized the observed alterations in ventricular chamber size and onset of systolic dysfunction (Fig. 1a and b). Indicators of ventricular chamber size, such as left ventricular end-diastolic diameter (LVEDD), showed an abrupt increase in the SIV + MAC group at the final examined time point (Fig. 1a), whereas indicators of systolic dysfunction developed gradually, becoming prominent from day 21 onward (Fig. 1b).

Significant differences in percentage change in LVEDD from baseline were present among the test groups over the study period (P < 0.001 overall, P < 0.01 at day 35; Fig. 1a). Percentage changes in LVEDD relative to baseline at day 35 differed significantly between the SIV + MAC group and the uninfected MAC group (P < 0.001; Fig. 1e). Significant differences in percentage changes in left ventricular ejection fraction (LVEF) from baseline were also present among test groups (P < 0.001 overall, P < 0.001 at day 28; Fig. 1b). Percentage changes in LVEF relative to baseline at day 28 differed significantly between the SIV + MAC group and each of the other three groups (P < 0.001 between SIV + MAC and the SIV-infected, MAC-treated, etanercept-treated group, P < 0.001 between SIV + MAC and the SIV group, and P < 0.05 between SIV + MAC and the MAC group). Percentage changes in LVEF for the MAC group at day 28 also differed significantly from percentage changes in LVEF for the SIV-infected, MAC-treated, etanercept-treated group (P < 0.05), a product of the combined mild decrease in LVEF for the MAC group and the mild increase in LVEF for the etanercept-treated group.

Mycobacterial antigenic stimulation significantly increases myocardial TNFα levels in simian immunodeficiency virus-infected animals, an effect prevented by etanercept treatment

Levels of myocardial TNFα as evaluated by densitometric analysis of western blots were significantly higher in hearts from animals of the SIV + MAC group than in hearts from animals of the SIV group (P < 0.05; Fig. 2a). Levels of TNFα were also significantly higher in cardiac muscle than in skeletal muscle for animals of the SIV + MAC group (P < 0.05), suggesting a tissue-specific effect of the mycobacterial antigenic stimulation rather than a generalized systemic increase in TNFα production (Fig. 2b). In contrast, levels of myocardial TNFα in SIV-infected, MAC-treated, etanercept-treated animals remained low, approximating those in skeletal muscle (Fig. 2b).

Fig. 2

Plasma sTNFR2 and IL-18 levels at day 14 correlate significantly with changes in left ventricular chamber diameter whereas baseline levels of IL-18 correlate with development of systolic dysfunction

In SIV-infected groups, plasma levels of both sTNFR2 and IL-18 became significantly elevated (P < 0.001) relative to baseline by day 14, corresponding with the occurrence of peak viremia (Fig. 3a and b). Percentage changes in LVEDD at day 35 correlated significantly with sTNFR2 levels at every time point from day 14 onward (day 14: P = 0.005, R = 0.578; day 21: P = 0.006, R = 0.567; day 28: P = 0.028, R = 0.479; day 35: P = 0. 018, R = 0.499) such that higher levels of sTNFR2 were associated with greater left ventricular chamber enlargement at the final measured time point (Fig. 3c). Elevations in plasma IL-18 at day 14 also correlated significantly with increased LVEDD at day 35 (P < 0.05, R = 0.407) such that higher levels were associated with greater left ventricular chamber enlargement at the final measured time point (Fig. 3d). In contrast, baseline IL-18 levels correlated negatively with development of systolic dysfunction as measured by percentage change in LVEF at day 35 (P < 0.05, R = 0.445), such that higher levels of IL-18 at day 0 were associated with greater preservation of systolic function at the terminal time point (Fig. 3e).

Fig. 3

Neither plasma sTNFR1 nor plasma MCP-1 showed significant correlations with echocardiographic parameters at any time point.

Inducible nitric oxide synthase levels, lymphocytic infiltration, plasma viral load, myocardial simian immunodeficiency virus-infected cell burden, and peripheral CD4 T-cell counts show no significant correlations with measures of ventricular chamber diameter or systolic function

The cardiodepressant effects of nitric oxide serve as one of the central effector mechanisms by which inflammatory cytokines impact cardiac function ^[24]. Myocardial inducible nitric oxide synthase (iNOS) levels were higher in all SIV-infected groups relative to the uninfected MAC control group, and this difference was statistically significant between the MAC group and the SIV + MAC group (P < 0.05); however, myocardial iNOS levels did not differ significantly among SIV-infected groups (Fig. 4a) and there were no significant correlations between iNOS levels and any examined measure of either ventricular chamber diameter or systolic function.

Fig. 4

Histologically, mild inflammatory infiltrates were present in multiple hearts, though cardiomyocyte necrosis was not a prominent feature in any. Quantitation of lymphocytic infiltrates based on rule-based scoring of CD3-labeled tissue sections ^[18] demonstrated no significant differences between groups (P > 0.1) and no significant correlations with any measure of systolic function or ventricular chamber diameter.

Myocardial infected cell burden, assessed by quantitation of positive signal in sections immunohistochemically labeled for SIV nef and SIV p27 gag proteins, demonstrated infected cells within hearts of six of the 18 SIV-infected animals. In each case, SIV-infected cells were rare (one to three per animal), exclusively interstitial in location, and morphologically consistent with lymphocytes or macrophages (Fig. 4c). All hearts were negative for adenovirus and cytomegalovirus by immunohistochemistry. Numbers of myocardial SIV-infected cells did not correlate significantly with any measure of systolic dysfunction or ventricular dilatation.

Plasma viral load in SIV-infected groups also did not correlate significantly with any measure of ventricular chamber diameter or systolic function. Viral loads within the SIV-infected, MAC-treated, etanercept-treated group, which developed no structural or functional disorders, were significantly higher (P < 0.05) than those in the SIV + MAC group, which developed significant adverse structural and functional changes (Fig. 4b).

CD4 T-cell counts differed significantly among groups overall (P < 0.001), as all three SIV-infected groups developed significant decreases relative to the uninfected MAC group over the course of the study (P < 0.01). Counts among the SIV-infected groups did not significantly differ, however, and there were no significant correlations between CD4 T-cell counts and any evaluated echocardiographic parameter (Fig. 4d).

Immunohistochemical staining of tissue sections for cleaved caspase 3 as an early indicator of apoptosis revealed infrequent staining within inflammatory cell clusters, and strong appropriate staining on positive control sections of ileum, but no staining of cardiomyocytes in any section and no correlations with any echocardiographic parameter.

Discussion

In the presented model, recurrent antigenic stimulation with heat-killed, opportunistic mycobacteria yielded significant myocardial contractile dysfunction and ventricular chamber dilatation in SIV-infected animals during acute infection, which was not seen with SIV-infection alone and was mild to absent in uninfected animals treated with the same stimulation protocol. Although the rapidity of induction of cardiac dysfunction in this model differs from that seen in natural disease, the findings suggest a fundamental hyperresponsivity of hearts from SIV-infected animals to antigenic stimulation. The cardiac changes induced by antigenic stimulation in this acute SIV-infection model reached magnitudes similar to those previously identified in studies of dilated cardiomyopathy in chronic SIV infection ^[30], at the same time excluding many potential confounders which complicate interpretation in late-stage infection. Concurrent etanercept treatment protected SIV-infected, MAC-treated animals from development of pathologic myocardial changes, suggesting a critical role of TNFα in induction of these changes. Although it is not possible to control for variation in host immune response using a small number of test subjects, these findings suggest that endogenous production of TNFα upon antigenic stimulation in the context of SIV infection is sufficient to induce significant cardiac dysfunction and implies exaggerated myocardial production of and/or responsiveness to TNFα by SIV-infected animals. In addition, the striking difference in myocardial TNFα levels between SIV-infected, MAC-treated animals which received etanercept and those which did not implies an important role of TNFα autoinduction in generation of the observed elevations in SIV-infected, MAC-treated animals ^[37]. This finding suggests that even mild hyperresponsivity in the initial TNFα response might be rapidly enhanced through positive feedback loops.

As further evidence of the importance of activation of the TNF system in evolution of the observed changes, plasma sTNFR2 levels from day 14 onward correlated significantly with extent of LVEDD change at day 35, predicting chamber dilatation with high sensitivity. IL-18 levels at day 14 also correlated positively with extent of LVEDD change, but as plasma IL-18 and sTNFR2 values correlated closely with one another (day 14: P < 0.001) the relationship between IL-18 and LVEDD may have simply reflected a connection between myocardial remodeling and the overall systemic inflammatory response. Nevertheless, the finding that higher levels of both sTNFR2 and IL-18 at day 14 correlated with increased percentage changes in LVEDD at day 35 suggests that the degree of activation of the inflammatory response around the period of peak viremia, during which viral replication is largely uncontrolled and early components of the innate immune response are activated, may directly contribute to the extent of subsequent structural change. This suggests that immune activation, well recognized as a major contributor to progression of SIV and HIV infection ^[38], may also play an important role in myocardial end-organ damage. In contrast, whereas elevated levels of IL-18 have been associated with myocardial contractile dysfunction ^[27,39,40], lower levels of IL-18 at baseline were associated with more severe declines in LVEF at later time points, suggesting that preinfection innate immune activation state may also play an important role in divergent postinfection myocardial effects.

In the present model of HIVCM pathogenesis, no evidence was found for the role played by plasma viral load, myocardial infected cell burden, peripheral CD4 T-cell count, myocardial lymphocytic infiltration, cardiomyocyte apoptosis, or cardiomyocyte necrosis. Whereas iNOS has been implicated in the impairment of myocardial function in inflammatory contexts ^[24], myocardial iNOS levels did not differ significantly among SIV-infected groups in this model. Likewise, although HIV gp120 protein has been demonstrated to be negatively inotropic in vitro ^[9] and high viral loads might therefore be expected to be associated with alterations of contractile function, no correlations were identified between plasma viral load and any examined measure of ventricular chamber diameter or systolic function.

Prominent myocardial lesion developed over the 35-day study period in SIV-infected, MAC-treated animals, but not in uninfected MAC-treated controls or animals which were SIV-infected alone. Treatment with the TNFα blocking agent etanercept proved fully protective from development of myocardial lesions, indicating that abnormal TNFα responses to the administered mycobacterial stimulus played a critical role in development of the observed changes. These findings suggest that aberrant inflammatory cytokine responsiveness to antigenic stimulation in the context of SIV or HIV infection may play an important role in the pathogenesis of cardiomyopathy. As myocardium-specific abnormalities in the cytokine response to antigenic stimulation appear central to the cardiac changes documented, defining the mechanism by which this occurs will be critical to understanding their ultimate causation.

Acknowledgements

The authors wish to thank Carol Stolarski and Laurie Machen for technical support. This work conducted with support of NIH grants HL075836, RR00168, K01 RR024120 and T32 RR007.