Changes in cardiac gene expression after pig-to-primate orthotopic xenotransplantation

Abstract

Background: Gene profiling methods have been widely useful for delineating changes in gene expression as an approach for gaining insight into the mechanism of rejection or disease pathology. Herein, we use gene profiling to compare changes in gene expression associated with different orthotopic cardiac xenotransplantation (OCXTx) outcomes and to identify potential effects of OCXTx on cardiac physiology.

Methods: We used the Affymetrix GeneChip Porcine Genomic Array to characterize three types of orthotopic cardiac xenograft outcomes: 1) rejected hearts that underwent delayed xenograft rejection (DXR); 2) survivor hearts in which the xenograft was not rejected and recipient death was due to model complications; and 3) hearts which failed to provide sufficient circulatory support within the first 48 h of transplant, termed "perioperative cardiac xenograft dysfunction" (PCXD). Gene expression in each group was compared to control, not transplanted pig hearts, and changes in gene expression > 3 standard deviations (±3SD) from the control samples were analyzed. A bioinformatics analysis was used to identify enrichments in genes involved in Kyoto Encyclopedia of Genes and Genomes pathways and gene ontogeny molecular functions. Changes in gene expression were confirmed by quantitative RT-PCR.

Results: The ±3SD data set contained 260 probes, which minimally exhibited a 3.5-fold change in gene expression compared to control pig hearts. Hierarchical cluster analysis segregated rejected, survivor and PCXD samples, indicating a unique change in gene expression for each group. All transplant outcomes shared a set of 21 probes with similarly altered expression, which were indicative of ongoing myocardial inflammation and injury. Some outcome-specific changes in gene expression were identified. Bioinformatics analysis detected an enrichment of genes involved in protein, carbohydrate and branched amino acid metabolism, extracellular matrix-receptor interactions, focal adhesion, and cell communication.

Conclusions: This is the first genome wide assessment of changes in cardiac gene expression after OCXTx. Hierarchical cluster analysis indicates a unique gene profile for each transplant outcome but additional samples will be required to define the unique classifier probe sets. Quantitative RT-PCR confirmed that all transplants exhibited strong evidence of ongoing inflammation and myocardial injury consistent with the effects of cytokines and vascular antibody-mediated inflammation. This was also consistent with bioinformatic analysis suggesting ongoing tissue repair in survivor and PCXD samples. Bioinformatics analysis suggests for the first time that xenotransplantation may affect cardiac metabolism in survivor and rejected samples. This study highlights the potential utility of molecular analysis to monitor xenograft function, to identify new molecular markers and to understand processes, which may contribute to DXR.

Fig. 1.

Histology and overall profiles for cardiac gene expression. (A–C) Haematoxylin and eosin staining of representative sections from rejected (A), survivor (B), and PCXD (C) tissues. (A) Sample rejected on POD 25 showing coagulative necrosis. (B) Sample explanted on POD 57 with minimal evidence of delayed xenograft rejection. (C) Sample explanted on day 1 with minimal histological evidence of injury. (D–F) Scatter plots of gene array data comparing the average log2 intensity scores for each probe that passed the exclusion test. (D) Rejected samples. (E) Survivor samples. (F) PCXD samples. The Pearson correlation coefficients were 0.9399, 0.9659, and 0.9737 for rejected, survivor and PCXD samples, respectively. POD, post-operative day; PCXD, perioperative cardiac xenograft dysfunction.

Fig. 2.

Changes in cardiac gene expression. (A) Venn diagram showing the distribution of probes with ±3 standard deviation (SD) changes in expression compared to controls. The total number of probes (n) is shown for each group. Values presented in each sector of the Venn diagram represent the number of probes unique to, or shared between groups. The values in parenthesis (21, 28 and 42) for group-specific probes are the number of probes, which maintain group specificity at ±3SD and all other levels of variance. (B) Hierarchical cluster analysis of all 260 probes illustrated in the Venn diagram. Clustering of genes and arrays performed using a Pearson (or “centered”) correlation. The asterisk marks the major nodal point for each group. From these points in the dendrogram, the Pearson correlation was obtained as a measure of intra-group variation in gene expression. (C–E) Cluster analysis of probes with ±3SD changes in expression compared to controls for each experimental group. (C) Rejected samples. (D) Survivors samples. (E) PCXD samples. For each panel (C–D), the first two columns are the control samples and the remaining columns are the individual samples from each group. Cluster analysis was performed on both genes and sample groups. Color coding is red for log2DIFF(FC) > 0, green for log2DIFF(FC) < 0, and black for log2DIFF(FC) = 0. PCXD, perioperative cardiac xenograft dysfunction. Log2DIFF(FC), log2 difference fold change.

Fig. 3.

Quantitative real-time RT-PCR validation of gene array results. (A) Quantitative RT-PCR analysis of 21 common probes showing similar changes in expression in each of the study groups. (B) Quantitative RT-PCR for 10 of 31 probes, which retained potential diagnostic value for distinguishing between the study groups. Graphs show the log2DIFF(FC) for each probe. All values were normalized to porcine β-actin expression, which did not vary significantly between the groups (data not shown). Results are from quantitative real-time RT-PCR analysis of Rej-9, Surv-34 and PCXD01 RNA. Rejected (Rej-9) black bars, survivor (Surv-34) gray bars, PCXD (PCXD01) white bars. PCXD, perioperative cardiac xenograft dysfunction. Log2DIFF(FC), log2 difference fold change.

Fig. 4.

Summary of Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis. (A) Rejected samples. (B) Survivor samples. (C) PCXD samples. Graphs indicate the KEGG pathway names and the number of probes with decreased (filled) and increased (opened) expression, relative to controls. Annotation for KEGG pathway and gene ontogeny (GO) molecular function enrichment was based on Tsai et al. [27]. All probes with significant changes in expression and within ±1 standard deviation variance from the control samples were used to create a gene list. The gene background was defined as the total set of probes, which passed the present/absent exclusion test for each group (Table 2). All KEGG pathways and GO molecular functions selected from DAVID Bioinformatics had EASE P-values < 0.05. PCXD, perioperative cardiac xenograft dysfunction.