Objective: Chronic rejection in cardiac allografts depletes vascular smooth muscle (VSM) α-actin from the coronary arterial smooth muscle bed while promoting its abnormal accumulation in cardiomyocytes and myofibroblasts. The objective was to determine if the newly discovered TEF1, MSY1, Purα and Purβ VSM α-actin transcriptional reprogramming proteins (TRPs) were associated with development of chronic rejection histopathology in accepted murine cardiac allografts. Methods: A mouse heterotopic cardiac transplant model was employed using H2 locus-mismatched mouse strains (DBA/2 or FVB/N to C57BL/6). Recipients were immunosuppressed to promote long-term allograft acceptance and emergence of chronic rejection. Explanted grafts and isolated heart cells were evaluated for changes in the DNA-binding activity and subcellular distribution of VSM α-actin transcriptional regulatory proteins. Results: The DNA-binding activity of all four TRPs was high in the developing mouse ventricle, minimal in adult donor hearts and increased substantially within 30 days after transplantation. Immunohistologic analysis revealed nuclear localization of Purβ and MSY1 particularly in fibrotic areas of the allograft myocardium demonstrating extravascular accumulation of VSM α-actin. Cardiomyocytes isolated from adult, non-transplanted mouse hearts not only exhibited less VSM α-actin expression and lower levels of TRPs compared to isolated cardiac fibroblasts or neonatal cardiomyocytes, but also contained a novel size variant of the MSY1 protein. Conclusion: Accumulation of TRPs in cardiac allografts, particularly within the fibroblast-enriched myocardial interstitium, was consistent with their potential role in VSM α-actin gene reprogramming, fibrosis and dysfunctional remodeling following transplant. These nuclear protein markers could help stage peri-transplant cellular events that precede formation of graft-destructive fibrosis and coronary vasculopathy during chronic rejection.
Time for primary review 35 days.
This article is referred to in the Editorial by W. Briest (pages 492–494) in this issue.
Transplant vascular sclerosis (TVS) is a clinical indicator of allograft chronic rejection and a major obstacle to long-term survival of human heart grafts . While depletion of contractile smooth muscle cells from the medial layer of the blood vessel wall is a well-known feature of TVS [2,3], recent studies in rodents and non-human primates showed that parenchymal cardiomyocytes and stromal cardiac fibroblasts also can become phenotypically remodeled in chronically-rejected heart grafts [4,5]. Vascular smooth muscle α-actin (VSM α-actin) is lost from coronary arterial smooth muscle cells during TVS whereas cardiac allograft remodeling is associated with accumulation of this protein in both stromal and parenchymal cells. Moreover, allograft remodeling occurs early after cardiac transplant well before the emergence of TVS [5–8]. One explanation is that accumulation of VSM α-actin-enriched myofibroblasts  and their extracellular matrix protein products within the allograft interstitium produces mechanical stress on adjacent cardiomyocytes that causes re-expression of fetal contractile protein genes such as VSM α-actin [10,11]. Additionally, perivascular fibrosis localized within the coronary tunica adventitia may promote accelerated TVS in accepted allografts due to ischemic damage to the vasa vasorum and autonomic plexuses [12–15]. Combined interstitial and coronary perivascular fibrosis may therefore represent very early maladaptations to transplant that could influence the microperfusion and long-term viability of accepted cardiac allografts.
In our laboratories, we have identified several DNA-binding proteins that mediate transcriptional activity of the VSM α-actin gene in mouse stromal myofibroblasts [16–19]. Three nuclear proteins (Purα, Purβ and MSY1) bind opposite strands of an essential DNA enhancer within the mouse VSM α-actin gene promoter located about 175 base pairs (bp) upstream from the start of transcription. A macromolecular complex containing these proteins influences enhancer conformation and limits its access to a fourth protein, TEF1, required for VSM α-actin gene activation in stromal fibroblasts. Myofibroblast differentiation induced by TGFβ1 revealed significant changes in VSM α-actin enhancer topology consistent with chromatin relaxation and transcriptional activation . Dynamic interplay between activating and inhibitory enhancer-binding proteins could be required to continuously adjust VSM α-actin promoter output in TGFβ1-activated myofibroblasts and limit accumulation of retractile tissue that could otherwise remodel cardiac allografts to the point of functional disruption.
We demonstrate in this report that the DNA-binding activities of MSY1, TEF1 and Purβ are enhanced following cardiac transplant. Immunohistologic analysis of cardiac allografts revealed pronounced nuclear localization of Purβ and MSY1 in fibrotic areas of the myocardium, a few parenchymal cells and within the coronary arterial wall. Accumulation of these proteins in the stromal, parenchymal and vascular compartments was consistent with their potential role in VSM α-actin gene reprogramming, fibrosis and remodeling in accepted cardiac allografts. The early accumulation of these regulatory proteins in cardiac allografts indicates their potential utility as staging markers for monitoring cellular changes that precede clinically-detectable TVS.
2.1 Murine heart transplantation protocol
Five-week-old, specific pathogen-free female DBA/2 (H-2d), FVB/N (H-2k) and C57BL/6 (H-2b) mice were purchased from Harlan Sprague–Dawley (Indianapolis, IN) or the Jackson Laboratory (Bar Harbor, ME) and maintained in a barrier facility in accordance with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996). DBA/2 and FVB/N donor hearts were heterotopically-transplanted into the abdomen of C57BL/6 recipient mice using an adaptation of the method described by Corey et al. . Recipients were immunosuppressed with gallium nitrate to promote long-term graft survival and development of chronic rejection .
2.2 Preparation of protein extracts and electrophoretic methods
Native hearts were removed from stage E13 embryos, 6-day-old neonates and 8–10-week-old adult donor mice. Cardiac allografts were removed from C57BL/6 recipients at various post-transplant time-points. Trimmed ventricles from non-transplanted hearts and allografts were individually frozen in liquid nitrogen, ground to fine powder and used to prepare separate protein extracts  for use in electrophoretic mobility shift assays (EMSA) with 32P-labeled oligonucleotide probes specific for the Purα/Purβ, MSY1 and TEF1 enhancer-binding proteins [16,17]. Ventricles from multiple embryonic or neonatal mice were pooled owing to their small size. EMSA reaction mixtures were incubated for 30 min at ambient temperature before electrophoresis on 4–5% non-denaturing polyacrylamide gels. For antibody supershift EMSAs, 2 to 5 μg aliquots of anti-MSY1 and anti-Purβ rabbit polyclonal antibodies were pre-incubated on ice with 5 μg of tissue extract for 1 h prior to performing EMSA reactions. Autoradiographs prepared from EMSA gels were quantified using a laser densitometer and ImageQuant™ image analysis software (Molecular Dynamics, Sunnyvale, CA). Three to five ventricles harvested at each post-transplant time-point plus age- and sex-matched non-transplanted control were evaluated by EMSA. The data was presented as the mean±S.E.M. and tested for statistical significance (P<0.05) by ANOVA.
Western blot analysis was performed using approximately 2–4 μg aliquots of protein extract resolved on 10% polyacrylamide mini-gels and then transferred to nitrocellulose membranes. After overnight blocking at 4 °C, the membrane was incubated with mAb 1A4, an anti-VSM α-actin monoclonal antibody conjugated to horseradish peroxidase (Dako, Carpinteria, CA) at a 1:100 dilution in Tris-buffered saline (TBS) containing 0.05% Tween 20 (TBST). Following a 60–90 min incubation at ambient temperature with gentle mixing, the membrane was washed 2–3 times in TBST and developed using reagents provided in a diaminobenzidine (DAB)–peroxidase Vectastain™ kit (Vector Laboratories, Burlingame, CA).
2.3 Cardiac fibroblast and cardiomyocyte isolation
Ventricles were removed under sterile conditions from 10 6-day-old DBA/2 pups or from single 8–10-week-old adult DBA/2 mice, placed in cold sterile calcium-free medium (CFM), minced into approximately 2 mm cubes and treated with 1 mg/ml collagenase (type CLS 2, Worthington Biochemical Corp., Lakewood, NJ) as described previously . Tissue fragments retained on 105 μm nylon mesh (Tetko, Briarcliff Manor, NY) were placed in CFM containing 1% bovine serum albumin (BSA) and 100 μM CaCl2, teased apart with sterile forceps and dispersed by trituration using a wide-bore pipette. Tissue fragments were removed and the cell suspension was adjusted to 500 μM CaCl2 and centrifuged over a 5 ml cushion of Dulbecco's modified Eagle's medium (DMEM, Gibco/BRL) containing 6% BSA. The resulting cardiomyocyte-enriched pellet was resuspended in DMEM containing 10% fetal bovine serum (FBS, Gibco/BRL) and seeded on 100 mm collagen-coated tissue culture plates for 3 h at 37 °C in a CO2 incubator to remove non-myocyte cells. Non-adhered cardiomyocytes were collected and stored at −80 °C for EMSA and Western blot analysis. For cardiac fibroblast isolations, ventricles were removed as noted above, minced and incubated in Hank's buffer containing trypsin (0.1 mg/ml) and collagenase (50 units/ml) for 10 consecutive 10 min treatment periods at 37 °C . Cells from each digestion period were pooled, resuspended in Ham's F12:DMEM (1:1) containing 10% FBS and seeded in standard culture dishes for 8 h. Non-adherent debris was discarded and the attached fibroblasts were maintained in Ham's F12:DMEM/10% FBS, scraped into ice-cold PBS, pelleted and stored at −80 °C for EMSA and Western blot analysis.
2.4 Immunohistochemical methods
Four-micrometer-thick sections from groups of three allografts, isografts or non-transplanted DBA/2 hearts were deparaffinized, rehydrated, blocked with 2% goat serum, 0.1% BSA and 0.05% Tween 20 in PBS (blocking solution) and incubated for 90 min at 37 °C with the anti-MSY1, anti-Purα and anti-Purβ rabbit polyclonal antibodies described in the text at a concentration of 2 μg/ml in blocking solution . After washing, sections were incubated for 30 min with a goat anti-rabbit secondary antibody and a Vectastain™ protocol was followed for color development using a DAB peroxidase kit (Vector Laboratories). In cases where peroxidase-conjugated mAb 1A4 was used, color development was performed after the primary antibody incubation. Double-labeling was performed by incubating sections with a second primary antibody after completing the staining protocol with the first primary antibody and washing the sections in TBS, pH 7.4. To generate color differences in each reaction product, a nickel solution was added to the second color development reagent which produced a blue–black precipitate that could be clearly distinguished from the red–brown reaction product obtained using the first developing solution without nickel ions. Sections were viewed using ×12.5 and ×25 Leitz objectives and images recorded on color print film. Appropriate controls were performed including omitting primary antibodies as well as including peptide inhibitors for blocking antibody binding.
Re-activation of the developmentally-repressed fetal VSM α-actin gene in adult ventricle has been reported in certain cardiomyopathies [10,11]. DNA-binding proteins that control fetal VSM α-actin expression in the normally-developing heart might also be responsible for its dysfunctional reprogramming in chronically-rejected cardiac allografts. Therefore, we compared VSM α-actin promoter-binding activity in extracts prepared from DBA/2 mouse ventricles at day 13 of fetal development and from 6-day-old neonatal and 8–10-week-old adult mice. Fig. 1 shows that the DNA-binding activities of the TEF1, MSY1 and Purα/Purβ transcriptional reprogramming proteins (TRPs) initially were high in the embryonic ventricle but decreased during development in parallel with a 90% reduction in the level of fetal VSM α-actin expression. While Purα/Purβ and MSY1 were detected in adult ventricle, there was 84% and 50% less, respectively, compared to embryonic ventricles. TEF1 activity was very low in the adult ventricle. Competition studies using excess unlabeled DNA probe eliminated binding thereby confirming sequence specificity of the protein:DNA interactions (Fig. 1c).
A potential role for TRPs in chronic rejection pathobiology was suggested by the observation that the DNA-binding activities of TEF1, MSY1 and Purα/Purβ all were significantly elevated following cardiac transplant (Fig. 2). Increased TRP DNA-binding activity was noted in both DBA/2 and FVB/N cardiac allografts as early as 7 days after transplant (data not shown), reached a peak within 30 days after transplant and remained elevated for at least 60 days post-transplant implying that accumulation of TRPs was an early and sustained response to transplant. Notably, the increase in TRP DNA-binding activity occurred in the same time frame previously noted for myocardial VSM α-actin gene re-activation in this model .
Since the Pur and MSY1 proteins mediate VSM α-actin transcriptional repression , we reasoned that these TRPs could play an essential role in silencing fetal smooth muscle actin gene expression in the injured myocardium and restoring normal cellular function during post-transplant surgical recovery. On the other hand, chronic accumulation of Pur and MSY1 proteins in allografts may be indicative of excessive myofibroblast involvement that could eventually compromise graft function and survival. Peptide polyclonal antibodies for Purα, Purβ and MSY1 previously developed for biochemical analysis of VSM α-actin transcriptional regulatory complexes in mouse stromal fibroblasts  were employed for immunohistologic analysis of allografts, isografts and non-transplanted donor hearts. Recipient mice were immunosuppressed with gallium nitrate to promote graft acceptance and development of chronic rejection histopathology. Of the three antibodies employed in the analysis, the Purβ-specific polyclonal antibody (anti-Pβ302/324) consistently showed the most intense reactivity with cardiac allografts (Fig. 3). Particularly striking was an intense anti-Pβ302/324 nuclear localization within the interstitial and coronary perivascular regions where cardiac fibroblasts, connective tissue and VSM α-actin-positive cardiomyocytes were observed in accepted allografts. Purβ-positive cell nuclei also were detected in the allograft coronary arterial wall and proximal to VSM α-actin-positive myofibroblasts (Figs. 3d and 4c, respectively). Similar to the distribution of the anti-Pβ302/324 antibody, strongly-reactive MSY1-positive (anti-M85/110) cell nuclei also were evident in coronary perivascular and interstitial regions (Fig. 4a). In contrast, very few allograft cell nuclei reacted with a Purα isoform-specific antibody (anti-Pα291/313, data not shown). TRP-positive cell nuclei also were not detectable in syngeneic isograft control hearts analyzed 30 days post-transplant (Fig. 4d).
TRP antibody reactivity was detected in both stromal and parenchymal cells although staining in the latter was very diffuse and fewer TRP-positive cardiomyocyte nuclei were observed compared to cardiac fibroblast nuclei (Fig. 4). Some CD4- or Mac1-positive inflammatory cells were observed in accepted cardiac allografts although they were confined to microvascular channels that did not react with any of the TRP antibodies (data not shown). To confirm the specificity of TRP:antibody interactions noted in immunohistologic preparations, EMSA reactions were performed using donor heart and allograft extracts that had been pre-incubated with Purβ or MSY1-specific antibodies. Each antibody substantially retarded the mobility of DNA:protein complexes containing the corresponding TRP (Fig. 5). Supportive of the immunohistologic findings, the supershift study also revealed that Purβ was the functionally predominant Pur protein isoform detected in the mouse ventricle both before and after transplant since the entire DNA:Pur protein complex was size-shifted by the anti-Pβ302/324 antibody. Three Purβ supershifted complexes that were specifically enriched in allografts may represent novel heteromeric TRP complexes (Carlini et al., submitted for publication; Subramanian et al., work in progress).
To investigate possible cell type-specific patterns of TRP DNA-binding activity, we analyzed extracts prepared from purified cardiomyocytes and cardiac fibroblasts. We reasoned that exposure to serum growth factors and cytokines during cell isolation would mimic cellular injury associated with tissue inflammatory events following allotransplantation. As shown in Fig. 6a, cardiomyocytes dispersed into medium containing 10% serum demonstrated less Pur protein and TEF1 DNA-binding activity compared to isolated cardiac fibroblasts. Cardiomyocytes isolated from donor hearts remained differentiated and exhibited a normal rod-shaped appearance (data not shown). Moreover, actin expression was not reprogrammed in isolated cardiomyocytes because they contained little VSM α-actin whereas isolated cardiac fibroblasts were highly enriched for this protein (Fig. 6b). Reproducible differences between cardiomyocytes and fibroblasts also were noted with respect to the molecular size of MSY1:DNA complexes (Fig. 6a). Binding of both the low and high mobility forms of MSY1 to DNA was sequence specific and fully competed by titration with unlabeled DNA containing a consensus MSY1 binding site (data not shown). In extracts prepared from intact, non-transplanted ventricle, a MSY1-specific antibody efficiently supershifted the fibroblast-type, low mobility MSY1 complex but not the high mobility cardiomyocyte-type complex (refer to Fig. 5). However, binding of high mobility cardiomyocyte MSY1 to DNA was enhanced when the low mobility MSY1 form was sequestered into antibody-supershifted complexes. Interestingly, isolated neonatal cardiomyocytes contained a low mobility MSY1:DNA complex that was indistinguishable from the fibroblast form (Fig. 7a). Since cardiac fibroblasts and neonatal cardiomyocytes both expressed VSM α-actin protein (Fig. 6b), the low mobility MSY1:DNA complex may be permissive for VSM α-actin transcriptional activation. Finally, a functional link between the size of MSY1:DNA complexes and allograft remodeling was suggested by observation that reprogrammed, VSM α-actin-positive cardiomyocytes isolated from chronically-rejected allografts were enriched for the low mobility MSY1:DNA complex (Fig. 7b,c). Additionally, in contrast to extracts from non-transplanted ventricles, both the high and low mobility MSY1:DNA complexes present in allograft extracts were noticeably supershifted by the MSY1 antibody (Fig. 5).
VSM α-actin transcriptional reprogramming contributes to tissue remodeling of accepted cardiac allografts in the mouse . While sufficient in the embryonic and perinatal heart, the VSM α-actin polypeptide is less adapted to the physiologic needs of the adult myocardium where replacement by a striated muscle-specific actin during postnatal development results in more efficient cardiac contractility . In this report, we showed that abnormal reprogramming of the VSM α-actin gene in cardiac allografts was accompanied by enhanced DNA-binding activity of its cognate transcriptional regulatory proteins. Our data further indicated that cardiac fibroblasts in chronically-rejected hearts may exhibit a greater degree of VSM α-actin gene reprogramming compared to cardiomyocytes. Pronounced nuclear localization of TRPs was most consistently observed in the allograft interstitial and perivascular compartments. Moreover, on an equivalent protein mass basis, DNA-binding activity of TRPs in isolated cardiac fibroblasts exceeded that seen in dispersed cardiomyocytes. Taken together these observations suggested that VSM α-actin gene reprogramming may first emerge in cardiac fibroblasts early after transplant with subsequent involvement of parenchymal cardiomyocytes at later stages. Nuclear import of specialized regulatory proteins such as Purβ and MSY1 may become significantly elevated in allograft myofibroblasts because VSM α-actin transcription in these cells typically is robust yet transient. Accordingly, Purβ and MSY1 may be essential for silencing high-level VSM α-actin transcription upon completion of graft surgical healing. In contrast, reprogramming of the developmentally-silenced fetal VSM α-actin gene in highly-differentiated, inherently less-adaptable cardiomyocytes most likely requires extensive changes in chromatin structure. Once activated, cardiomyocytes may be less adept at bringing about VSM α-actin gene silencing and recovery of cellular homeostasis. Supportive of this idea, allograft cardiomyocytes up-regulated VSM α-actin gene expression yet they accumulated very little nuclear Pur and MSY1 repressor protein required to limit gene output. Of course there could be timing issues involved in the chronically-rejected heart such that fibroblasts respond earlier and acquire more repressor protein than cardiomyocytes. In this regard, accumulation of cardiac myofibroblasts and fibrotic tissue in accepted allografts could eventually influence cardiomyocyte phenotype (Fig. 8). In this hypothetical model, excessive connective tissue impairs graft microperfusion, restricts myocardial contractility and stress-activates fetal VSM α-actin expression in cardiomyocytes that may further compromise graft function. Recent histologic analysis of 11 different microscopic fields from trichrome- and VSM α-actin antibody-stained allograft tissue sections revealed more pronounced fibrosis in the left ventricular endocardium, accounting for nearly 19% of the total cross-sectional area (unpublished observations). In support of a possible link between fibrosis and myocardial actin gene reprogramming, this same region was shown in our earlier study to be preferentially enriched for VSM α-actin-positive cardiomyocytes .
Association of TRP-positive cells with regions of perivascular fibrosis additionally implicated coronary adventitial myofibroblasts in the pathobiology of chronic rejection (Fig. 8). In this regard, the appearance of nuclear TRPs in murine cardiac allografts preceded development of coronary arteriosclerosis lesions by about 4 to 5 weeks [6,7]. TRPs may therefore be useful staging markers for early graft dysfunction that precedes clinically-detectable transplant vascular sclerosis. Adventitial fibroblasts recently have been shown to be key participants in arterial remodeling and neovascularization associated with the development of native coronary arteriosclerosis [12,14,15]. Additional studies will be required to determine if TRP-positive cells in the coronary arterial wall of allografts in fact correspond to reprogrammed adventitial myofibroblasts. Since loss of VSM α-actin microfilaments has been linked with increased cellular motility, an alternative possibility is that TRP-positive cells in the vessel wall are α-actin-depleted smooth muscle cells that have migrated from their native position in the tunica media . We observed that coronary arterial Purβ and MSY1 nuclear staining was most evident in 30 day allografts at a time when expression of VSM α-actin polypeptide in medial smooth muscle cells was significantly reduced relative to donor hearts. Further supporting a possible smooth muscle origin for TRP-positive cells in the coronary bed, preliminary evidence from studies employing mouse aortic allografts indicates that medial smooth muscle cells acquire pronounced Purβ nuclear reactivity well before neointimal lesion formation (Subramanian et al., work in progress). Comparative analysis of heterotopic aorta and cardiac allograft models currently is underway to investigate the origin and fate of TRP-positive cell types. In this regard, the aorta allograft model provides an enlarged view of vessel wall histology not easily revealed in smaller caliber mouse coronary arteries.
Cardiac fibroblasts contained a high molecular weight MSY1:DNA complex whereas cardiomyocytes were enriched for a low molecular weight complex but acquired the larger form when reprogrammed to express VSM α-actin following transplant. We propose that accumulation of high molecular weight MSY1 may permit VSM α-actin transcription by modifying chromatin structure around the VSM α-actin enhancer. For example, large MSY1 complexes in injury-activated myofibroblasts or stressed cardiomyocytes may additionally contain transcriptional activators such as TEF1 and/or serum response factor [16,17]. Previous reports indicate that MSY1 forms multimers  as well as heteromeric complexes with both TEF1 and Pur proteins [19,28]. On the other hand, studies of the human MSY1 homolog, YB-1, have shown that the carboxyl terminus is proteolytically removed prior to nuclear importation suggesting that large, unprocessed MSY1 might be sequestered from the VSM α-actin promoter in intact cells and thus unable to repress transcription [29,30]. Further investigation of MSY1 post-translational processing and subcellular compartmentalization is required in the setting of cardiac transplantation. We interpret the inability of a MSY1-specific antibody to supershift small, cardiomyocyte-type MSY1:DNA complexes in non-transplanted hearts as evidence for epitope masking due to proteolytic cleavage or tight compaction with the DNA probe. Compaction of the VSM α-actin enhancer by proteolytically-processed MSY1 may explain why this gene is transcriptionally inactive in the adult, non-transplanted heart. In support of this idea, preliminary sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and immunoblot analysis clearly revealed two MSY1 size variants in donor mouse hearts and that the ratio of the two forms changed following allotransplant (Subramanian et al., work in progress). We plan to analyze cytokine-stimulated cardiomyocytes to examine dynamic relationships between VSM α-actin gene transcription and MSY1 proteolytic processing. Detailed study of MSY1 function in the transplanted heart also is warranted because of its role in suppressing interferon-γ-mediated activation of class II major histocompatibility complex (MHC) genes . MHC class II proteins participate in the generation of the T-cell repertoire and are required for donor antigen presentation to recipient inflammatory cells. Recent reports using the mouse heart and carotid artery transplant models indicate that repression of donor-specific MHC class II gene expression is beneficial to allograft survival [3,32]. A central role for MSY1 in the management of the wound healing response in chronically-rejected cardiac allografts also is indicated by its participation in the transcriptional regulation of genes encoding type I collagens , matrix metalloproteinase-2  and GM-CSF .
In summary, we have presented evidence showing for the first time that the DNA-binding activity of TRPs responsible for regulation of VSM α-actin gene expression is markedly influenced by cardiac allotransplantation. The data suggest that cardiac fibroblasts may be most susceptible to gene reprogramming with subsequent involvement of parenchymal cardiomyocytes as elevated myofibroblast activity and fibrosis restrict graft microperfusion and function. The accumulation of TRPs in accepted allografts reflects a state of wound healing, while beneficial in the short term for repair of surgical injury, may have serious implications for graft survival when left chronically unchecked.
This work was supported by RO1 grants HL 60876 (ARS), HL 54281 (RJK) and HL 40150 (CGO) from the National Institutes of Health. Dr Subramanian was supported by NIH NRSA HL 09970 and Dr Kelm was supported by a grant from the American Heart Association/Northland Affiliate, 9930343Z.