Gene Expression Profiles and B-Type Natriuretic Peptide Elevation in Heart Transplantation
More Than a Hemodynamic Marker
Background— B-type natriuretic peptide (BNP) is chronically elevated in heart transplantation and reflects diastolic dysfunction, cardiac allograft vasculopathy, and poor late outcome. This investigation studied peripheral gene expression signatures of elevated BNP concentrations in clinically quiescent heart transplant recipients in an effort to elucidate molecular correlates beyond hemodynamic perturbations.
Methods and Results— We performed gene microarray analysis in peripheral blood mononuclear cells of 28 heart transplant recipients with clinical quiescence (absence of dyspnea or fatigue; normal left ventricular ejection fraction [EF >55%]; ISHLT biopsy score 0 or 1A; and normal hemodynamics [RAP <7 mm Hg, PCWP ≤15 mm Hg, and CI ≥2.5 L/min per m2]). BNP levels were performed using the Triage B-type Natriuretic Peptide test (Biosite Diagnostics Inc, San Diego, Calif) and median BNP concentration was 165 pg/mL. Seventy-eight probes (of 7370) mapped to 54 unique genes were significantly correlated with BNP concentrations (P<0.001). Of these, the strongest correlated genes (P<0.0001) were in the domains of gelsolin (actin cytoskeleton), matrix metallopeptidases (collagen degradation), platelet function, and immune activity (human leukocyte antigen system, heat shock protein, mast cell, and B-cell lineage).
Conclusions— In the clinically quiescent heart transplant recipient, an elevated BNP concentration is associated with molecular patterns that point to ongoing active cardiac structural remodeling, vascular injury, inflammation, and alloimmune processes. Thus, these findings allude to the notion that BNP elevation is not merely a hemodynamic marker but should be considered reflective of integrated processes that determine the balance between active cardiac allograft injury and repair.
B-type natriuretic peptide (BNP), a 32-amino acid neurohormone, secreted predominantly from the cardiac ventricle in response to increased wall stress, is chronically elevated in heart transplant recipients.1 Because an elevation in BNP is associated with cardiopulmonary hemodynamic aberrations, it has been thought to represent diastolic dysfunction or ventriculo-vascular uncoupling of the allograft with its surrounding vasculature.2,3 More recently, investigations have noted that natriuretic peptides in the chronic phase of transplantation are associated with cardiac rejection4 and are predictive of cardiac allograft vasculopathy5,6 and graft loss.6,7
The primary purpose of this investigation included investigation of molecular pathways using peripheral gene expression (GE) patterns that correlate with elevated BNP concentrations in otherwise clinically quiescent heart transplant recipients.
Materials and Methods
Of 42 consecutive heart transplant recipients initially screened, we enrolled 28 consecutive clinically quiescent patients in this study at least 4 weeks after heart transplantation who met the inclusion criteria. Clinical quiescence was defined as the absence of any symptoms of dyspnea or fatigue; normal left ventricular ejection fraction (EF >55%) by echocardiography; no histological evidence of rejection on endomyocardial biopsy (International Society for Heart and Lung Transplantation [ISHLT] biopsy score 0 or 1A)8; and normal hemodynamics (right arterial pressure <7 mm Hg, pulmonary capillary wedge pressure ≤15 mm Hg, and cardiac index ≥2.5 L/min per m2) measured by invasive right heart catheterization. Furthermore, patients with abnormal renal function as defined by a serum creatinine >1.5 mg/dL were excluded because of the possibility of interference with BNP concentrations.8 Clinical data, including immunosuppressive drug regimen and measures of allograft function for each patient encounter, were also collected and analyzed. All patients received tacrolimus and mycophenolate mofetil-based immunosuppression with adjunctive corticosteroids (prednisone). The study was conducted as part of an Institutional Review Board (IRB)-approved protocol. The authors had full access to the data and take full responsibility for their integrity. All authors have read and agree to the manuscript as written.
Venous blood samples obtained from enrolled patients were processed on the day of scheduled surveillance biopsies. BNP measurements were performed using whole blood (5 mL) collected into tubes containing potassium EDTA (1 mg/mL blood). The Triage B-type Natriuretic Peptide test (Biosite Diagnostics Inc, San Diego, Calif) was used for this measurement. The Triage BNP test is a fluorescence immunoassay for the quantitative determination of BNP in whole blood and plasma specimens with a turnaround time of 15 minutes and a coefficient of variability of 15%.1,6 Peripheral blood mononuclear cells (PBMCs) were isolated from 8 mL venous blood using density gradient centrifugation (CPT tubes; Becton-Dickinson). Samples were frozen in lysis buffer (RLT; Qiagen) within 2 hours of phlebotomy. Total RNA was isolated from each sample (RNeasy; Qiagen) and assessed spectrophotometrically (Spectromax). Two micrograms of RNA from each sample were converted to cDNA, which were then used as templates for in vitro transcription with Cy3-dCTP to generate Cy3-labeled RNA.
Standard techniques were used to obtain biopsy samples, which were graded by local pathologists as well as by 3 independent (“central”) pathologists blinded to clinical information. All samples were obtained >4 weeks after transplant, transfusion, or rejection therapy. Absence of rejection was defined as ISHLT 1A or 0 grades, confirmed by a panel of blinded central pathologists.9
A custom microarray was designed using sequences from subtracted, suppressed libraries (25 000 sequences) of stimulated and resting human leukocytes (PCR Select; Clontech) and also identified using publicly available databases; 24 000 50-mer oligonucleotide (Sigma) gene probes (8000 probes in triplicate) representing 7370 genes were spotted on a custom microarray (Telechem). Microarrays were imaged on confocal laser scanners (Agilent) and data were extracted (GenePix 3.0; Axon Instruments), corrected for background and normalized. Gene probes were excluded from analysis if: expression values were available for only ≤10 samples (ie, gene was not detectable in a sufficient number of samples); or the standard deviation of expression values across the samples was <0.05 with expression values expressed as log_base10 (ratio[sample/control]) This condition was introduced to base correlation analysis on sufficient variance; or average signal-to-noise ratio associated sample was <5 (empirical threshold). This left 5927 gene probes for analysis. Probes mapping to the same gene transcript were not averaged. A global median normalization was applied to each hybridization sample.
Microarray Data Analysis
Correlation analysis was performed using Pearson correlation coefficient and rank-based Spearman correlation coefficient. Correlation coefficients (r) and associated P values were computed and averaged for Pearson and Spearman methods. Hierarchical clustering was performed using the pair-wise Pearson correlation coefficients computed from the expression values in the 28 samples as a distance measure. Only gene probes significantly correlated with BNP concentrations were subject to clustering. The average linkage method as implemented in the program OC by Geoffrey Barton was used to generate expression-based dendrograms of gene probes.10
Estimation of the False Discovery Rate
Gene expression–BNP concentration correlation results in experimental data were compared with randomized BNP concentrations to estimate the proportion of false-positives among the genes identified with P below a certain cutoff value. BNP concentrations were randomly reassigned among the 28 samples and correlation analysis was applied and the process repeated 10 times. The number of probes below a given P value threshold obtained from the randomization trials was then compared with the number of significant genes in the unscrambled data set at the same P value threshold to estimate the proportion of false-positives. As an added precaution for robustness, we implemented the Benjamini and Hochberg frequentist methodology11 for detection of the false discovery rate.
The demographic characteristics of the 28 patients enrolled included a mean recipient age of 58±10 years; 68% were men, 96% were white, and it was 16±3 weeks after transplant. The donor characteristics included an average age of 31±6 years; 67% were women and 96% were white. Based on the inclusion criteria of normal hemodynamics, the cohort had a right atrial pressure of 6±1 mm Hg, pulmonary capillary pressure of 11±2 mm Hg and cardiac index of 3.2±0.5 L/min per m2. The mean serum creatinine was 1.2±0.1 mg/dL (median, 1.0 mg/dL). The average tacrolimus trough concentration was 12.2±4 ng/mL, whereas prednisone daily dose was 18.9±6 mg. Mycophenolate mofetil concentrations were not measured and the average daily dose was 2.4±0.3 grams. These 28 samples on biopsy included 20 with grade 0 and 8 with grade IA pathology as confirmed by the centralized pathologists.
For the patient cohort, median BNP concentration was 165 pg/mL (mean 255±32 [SEM] pg/mL). Women displayed higher BNP concentrations than men (medians 191 versus 156 pg/mL, P=0.012). There were no correlations of BNP concentration with renal function or hemodynamics in this specially selected cohort.
BNP and Gene Expression Profiles
Of the 5927 probes used, 3850 mapped to 2863 unique annotated genes and 2077 did not map to any presently characterized human transcript sequences in the RefSeq database. A total of 78 gene probes were significantly correlated with BNP concentrations at P<0.001. At this level of significance, the false discovery rate (FDR) correcting for multiple testing was estimated at 10%. An average of only 8 probes were found to be <0.001 of the P value threshold in the randomization trials. The FDR by the method of Benjamini and Hochberg was estimated at <7.6%.
These 78 probes mapped to 54 uniquely characterized genes (existing gene symbol) and 19 probes were associated sequences not mapping to currently annotated genes; 48 probes were significantly negatively correlated with BNP concentrations (ie, increased gene expression was associated with decreased BNP concentrations), whereas 30 probes were positively correlated with BNP. These 78 gene probes are depicted in the dendrogram with their annotations in the Figure.
Specific Gene Correlates of BNP
We reviewed all uniquely characterized genes (n =54) and categorized them into domains of cellular remodeling (those encoding proteins involved with cellular structure), vascular injury and repair (those encoding proteins involved with platelet and endothelial function), and alloimmune inflammatory interactions (those genes encoding proteins involved with cellular and humoral immune processes). Additional genes involved in stem cell mobilization pathways, antiviral activity, and apoptosis were also described. These genes and their ascribed functions (n=25) are detailed in the Table. We did not display those significant genes with a nonspecific function. Several genes (n =29) encoding cell signaling proteins and G proteins fell into this category. Examples of these genes include PP (inorganic pyrophosphatase), HNRPDL (heterogeneous nuclear ribonucleoprotein D-like), PDCL3 (phosducin-like 3), ARHGEF3 (Rho guanine nucleotide exchange factor [GEF] 3), NCL (nucleolin), ELMO2 (engulfment and cell motility 2), HIPK2 (homeodomain interacting protein kinase 2), MPP1 (membrane protein, palmitoylated 1), and CCNF (cyclin F).
Actin cytoskeleton genes (gelsolin) and matrix metallopeptidase genes (MMP8 and MMP9) were strongly upregulated in the presence of elevated BNP concentrations and these genes denote collagen turnover and remodeling at a myocyte, interstitial and vascular level. Gelsolin was increased 25.2-fold (r value 0.74), MMP9 increased 10.5-fold (r value 0.68), and MMP8 increased 6.1-fold (r value 0.67). Genes involved in vascular injury were strongly represented and these included the domains representing platelet function (platelet factor-4, thromboxane-2, PAI-1), adhesion cell molecules, and integrins (platelet glycoprotein 2B-3A). Platelet factor-4 expression was increased 17.4-fold (r value 0.72). Alloimmune activity related genes were both positively (Histamine Receptor H2 involved in mast cell regulation) and negatively (B cell genes that play a role in T cell differentiation; human leukocyte antigen system genes) correlated with BNP concentrations.
This investigation has correlated, for the first time to our knowledge, gene expression pathways associated with increased expression of BNP in the clinically quiescent phase of heart transplantation. Contrary to common perception that BNP is a neurohormone principally responsive only to hemodynamic perturbations, we have confirmed that an elevation of this peptide during normal hemodynamics and in the absence of histological allograft rejection, reflects upregulation of molecular pathways representing ongoing inflammation, alloimmune activation, cardiac structural remodeling, and vascular perturbations. These observations lend credence to the notion that BNP elevations in the quiescent phase after heart transplantation are reflective of ongoing allograft injury and remodeling at activity levels not discernable with clinical techniques. This may explain observations from studies that have found this biomarker to possess prognostic predictive power for the development of cardiac allograft vasculopathy and allograft failure.5–7
The strong association of the gelsolin gene and matrix metallopeptidase genes with an elevated BNP point to the important regulatory role of this natriuretic peptide in cardiac structural remodeling. The gelsolin gene family encodes a number of actin-binding proteins that are thought to function in the cytoplasm by severing, capping, nucleating, or bundling actin filaments.12 Others have demonstrated that the binding step of collagen phagocytosis is facilitated by Ca(2+)-dependent, gelsolin-mediated severing of actin filaments and that phosphatidylinositol-4,5-bisphosphate regulation of gelsolin promotes the actin assembly required for internalization of collagen fibrils.13 Matrix metallopeptidases (MMPs) and their inhibitors regulate the cardiac extracellular matrix by controlling fibrillar collagen. Specifically, MMP8 and MMP9 have been shown to be selectively increased in transplanted hearts as early as 2 weeks after transplantation and correlate with an increase in connective tissue in the allograft.14 MMP9 activity has been found to reflect increased T cell alloreactivity,15 whereas other studies have pointed to a vascular role for this proteolytic enzyme as an effector molecule of oxidant-mediated coronary vasomotor dysfunction.16 It has also been described that systemic activation of MMP2 and MMP9 in donors with intracerebral hemorrhage and subsequent heightened expression of these peptidases in the allograft are associated with the development of allograft vasculopathy.17 Thus, BNP elevation serves as a surrogate for gelsolin and MMP activity, which represent ongoing extracellular matrix and vascular remodeling.
Vascular injury is sentinel to the development of cardiac allograft vasculopathy and this entity is the strongest limitation to long-term survival in heart transplantation.18 The genesis of cardiac allograft vasculopathy is determined by both alloimmune and nonimmunological factors but inflammation is central in its development.19 Thrombosis and inflammation are intertwined in the development of vascular disease. Platelet factor 4 (PF-4), a member of the CXC subfamily of chemokines, is secreted in high amounts by activated platelets. PF-4 specifically binds to human polymorphonuclear granulocytes, but requires tumor necrosis factor alpha-α as a costimulus for the induction of effector functions in vascular cells.20 It has been demonstrated that long-term survivors of heart transplantation are characterized by activation of platelets, which contain increased levels of soluble CD-40 and release enhanced cytokines when stimulated.21 Animal studies have shown that lack of plasminogen activator inhibitor-1 expression in donor tissue greatly exaggerates the extent of vascular intimal proliferation after allogeneic transplantation.22 Similarly, integrins are cell surface adhesion receptors that mediate cell–cell and cell–extracellular matrix interaction with several ligands, including fibrinogen, fibronectin, thrombospondin, and prothrombin. Myocardial ischemic injury after cardiac transplantation is associated with upregulation of integrins, tissue factor, and activation of the MMP induction system, which may contribute to the subsequent development of allograft remodeling and vasculopathy.23 Thrombospondin-1, a matrix glycoprotein, inhibits angiogenesis and facilitates smooth muscle cell proliferation. Increased levels of thrombospondin-1 in human cardiac allografts may alter vascular responses to angiogenic growth factors by inhibiting angiogenesis and promoting smooth muscle cell proliferation characteristic of cardiac allograft vasculopathy.24 Our investigation suggests that BNP elevation is indicative of upregulation of pathways in thrombosis, inflammation, and vascular remodeling in heart transplantation. Importantly, the strong expression of genes regulating platelet function points to the need for anti-platelet therapeutic studies in heart transplantation.
Although we enrolled a selective subset of patients without overt rejection as evidenced by normal allograft function and absence of histological rejection, our investigation correlated genes in the mast cell function domain with BNP elevation. Human mast cells, by elaborating various cytokines, chemokines, and proinflammatory mediators play a complex role in inflammatory disorders and have been linked with cardiac rejection.25 However, in post-transplant hearts during ongoing acute rejection, the mast cells are phenotypically different from nonrejecting hearts and isografts.25 Others have shown that mast cells are not necessarily associated with rejection but underlie enhanced inflammation, neovascularization, and fibrosis during cardiac allograft arteriosclerosis.26 Thus, in our study, it is more likely that the correlation of mast cell genes reflects ongoing inflammation rather than a signature for allograft rejection.
Finally, we acknowledge several limitations of this work. First, the patient number is small; however, we chose to develop a highly selected cohort of clinically quiescent patients with normal renal function within the first year of transplant and in this context represents a reasonable sample size. Second, we correlated peripheral blood gene expression patterns with BNP concentrations and did not evaluate intragraft events. Therefore, it should be recognized that overexpression or underexpression of a specific gene in peripheral blood should not necessarily be construed to describe the direction of that pathway within the allograft. Third, we used a custom leukocyte microarray and it is entirely possible that we might have missed some significant pathways if a whole genome approach had been undertaken. However, the fact that a number of genes that correlated with BNP concentrations all mapped to the same domain of molecular pathways increases the likelihood of the robustness of our findings. We concede that independent validation studies should help confirm these findings.
In the clinically quiescent heart transplant recipient, an elevated BNP concentration is associated with molecular patterns that point to ongoing to active cardiac structural remodeling, vascular injury, inflammation, and alloimmune processes. Thus, these findings allude to the notion that BNP elevation is not merely a hemodynamic marker but should be considered reflective of integrated processes that determine the balance between active cardiac allograft injury and repair.
The authors express their gratitude to Steve Gottlieb, MD, for his critical appraisal of this manuscript.
Sources of Funding
This study was supported in part by a research grant from XDx Inc. M.R.M. has received research grants from XDx Inc.
M.R.M. is a consultant for XDx, Inc, and D.W., J.G.W., J.P., and D.T. are or were employees of XDx Inc.
Presented at the American Heart Association Scientific Sessions, Dallas, Tex, November 13–16, 2005.
Hervas I, Arnau MA, Almenar L, Perez-Pastor JL, Chirivella M, Osca J, Bello P, Osa A, Marti JF, Vera F, Mateo A. Ventricular natriuretic peptide (BNP) in heart transplantation: BNP correlation with endomyocardial biopsy, laboratory and hemodynamic measures. Lab Invest. 2004; 84: 138–145.
Chenevier-Gobeaux C, Claessens YE, Voyer S, Desmoulins D, Ekindjian OG. Influence of renal function on N-terminal pro-brain natriuretic peptide (NT-proBNP) in patients admitted for dyspnoea in the Emergency Department: comparison with brain natriuretic peptide (BNP). Clin Chim Acta. 2005; 361: 167–175.
Billingham ME, Cary NR, Hammond ME, Kemnitz J, Marboe C, McCallister HA, Snovar DC, Winters GL, Zerbe A. A working formulation for the standardization of nomenclature in the diagnosis of heart and lung rejection: Heart Rejection Study Group. The International Society for Heart Transplantation. J Heart Transplant. 1990; 9: 587–593.
Barton, G.J. (1993, 2002) “OC - A cluster analysis program”, University of Dundee, Scotland, UK; www.compbio.dundee.ac.uk/downloads/oc.
Benjamini, Y. and Hochberg, Y. Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society. 1995; Series B 57: 289–300.
Arora PD, Chan MW, Anderson RA, Janmey PA, McCulloch CA. Separate functions of gelsolin mediate sequential steps of collagen phagocytosis. Mol Biol Cell. 2005; 16: 5175–5190.
Schupp DJ, Huck BP, Sykora J, Flechtenmacher C, Gorenflo M, Koch A, Sack FU, Haass M, Katus HA, Ulmer HE, Hagl S, Otto HF, Schnabel PA. Right ventricular expression of extracellular matrix proteins, matrix-metalloproteinases, and their inhibitors over a period of 3 years after heart transplantation. Virchows Arch. 2006; 448: 184–194.
Egi K, Conrad NE, Kwan J, Schulze C, Schulz R, Wildhirt SM. Inhibition of inducible nitric oxide synthase and superoxide production reduces matrix metalloproteinase-9 activity and restores coronary vasomotor function in rat cardiac allografts. Eur J Cardiothorac Surg. 2004; 26: 262–269.
Yamani MH, Starling RC, Cook DJ, Tuzcu EM, Abdo A, Paul P, Powell K, Ratliff NB, Yu Y, McCarthy PM, Young JB. Donor spontaneous intracerebral hemorrhage is associated with systemic activation of matrix metalloproteinase-2 and matrix metalloproteinase-9 and subsequent development of coronary vasculopathy in the heart transplant recipient. Circulation. 2003; 108: 1724–1728.
Zhao XM, Hu Y, Miller GG, Mitchell RN, Libby P. Association of thrombospondin-1 and cardiac allograft vasculopathy in human cardiac allografts. Circulation. 2001; 103: 525–531.