Both clinical and experimental observations suggest that allograft rejection is a complex process with multiple components that are, at least partially, functionally redundant. Studies using graft recipients deficient in various genes including chemokines, cytokines, and other immune-associated genes frequently produce a phenotype of delayed, but not indefinitely prevented, rejection. Only a small subset of genetic deletions (for example, TCRα or β, MHC I and II, B7-1 and B7-2, and recombinase-activating gene) permit permanent graft acceptance suggesting that rejection is orchestrated by a complex network of interrelated inflammatory and immune responses. To investigate this complex process, we have used oligonucleotide microarrays to generate quantitative mRNA expression profiles following transplantation. Patterns of gene expression were confirmed with real-time PCR data. Hierarchical clustering algorithms clearly differentiated the early and late phases of rejection. Self-organizing maps identified clusters of coordinately regulated genes. Genes up-regulated during the early phase included genes with prior biological functions associated with ischemia, injury, and Ag-independent innate immunity, whereas genes up-regulated in the late phase were enriched for genes associated with adaptive immunity.

Multiple types of evidence indicate that allograft rejection is a complex process involving a diverse array of proinflammatory and immune responses. In clinical transplantation, despite the use of multiple immunosuppressive agents, the incidence of both acute and chronic rejection remains a significant problem (1, 2, 3, 4, 5, 6). The standard requirement for multiple immunosuppressive agents is consistent with the notion of a complex immune response involving multiple components. Studies in animal models are also consistent with this concept. For example, in the murine heterotopic heart transplant model, many studies of knockout strains deficient in various chemokines, cytokines, inflammatory molecules, and receptors have been published showing a similar phenotype of delayed, but not totally prevented, rejection. The common observation in these studies of gene-deficient mice that rejection is delayed, but not indefinitely prevented, supports a model in which multiple different components of the alloimmune response are important, but not always necessary, to induce graft rejection. Only in a few gene-deficient models, such as elimination of all T cells or both B7-1 and B7-2 costimulatory molecules, and probably the deletion of both MHC class I and II, is the graft permanently accepted without rejection (7, 8, 9, 10, 11, 12).

Additional complexity is generated by contributions from both innate and adaptive immunity to the process of allograft rejection. Previous results from our laboratory analyzing proinflammatory responses in graft recipients deficient in T and B lymphocytes have characterized a robust innate immune response that occurs during the first 24 h following transplantation (13). Importantly, the innate response included multiple proinflammatory molecules including chemokines, cytokines, and receptors. These observations suggested that, similar to infectious models, the early phase of rejection consists predominantly of innate immune responses, whereas the late phases of rejection are enriched for components of adaptive immunity. In the current study, our objective was to analyze the early and late phases of rejection to investigate the contributions of the innate and adaptive immune responses.

Although the notion that the alloimmune response is complex is commonly accepted, experimental approaches have typically analyzed a small subset of parameters following a single manipulation such as administration of an immune modulator or deletion of a gene. With the development of DNA microarray technology, the ability to produce databases including expression profiles of large numbers of mRNAs has become practical. In contrast to hypothesis driven approaches that focus on a single mechanism in a network of biological complexity, microarray experiments attempt to globally describe gene expression. With the completion of the human and murine genome sequences, it is becoming feasible to monitor the expression of all genes. Such global approaches have already been successfully applied to simpler organisms such as yeast (14, 15). One obvious limitation of microarray studies is that data is limited to RNA, but not protein, expression. However, by analyzing a large dataset of mRNA expression, previous reports have detected the perturbation of biological responses by detecting changes in gene expression despite the fact that a subset of the genes in the network may be regulated posttranscriptionally. A recent single study in yeast shows that microarray data can corroborate, and even expand, our understanding of metabolic pathways or transcriptional regulation by DNA binding proteins decoded by decades of biochemical studies (16, 17). Interestingly, in this yeast study, only 5% of genes are regulated independently of transcription (17). In our study, we linked physiological outcomes (allograft rejection) with changes in gene expression (determined by microarrays). Our assumption was that the combination of biological function with gene expression data from a kinetic analysis would characterize the innate and adaptive phases of graft rejection.

A major challenge of microarray studies is the meaningful interpretation of the huge databases of expression values. In our studies, hierarchical clustering algorithms were used to distinguish various experimental groups (15). These algorithms successfully differentiated different tissues, different time points, and rejection from the control grafts. In addition, to identify specific subsets of coordinately regulated genes in the alloimmune response, we used self-organizing maps (SOM)4 to detect clusters of genes that were coordinately regulated (18). Our results combine biologic function with gene expression profiles to differentiate the early response, which includes genes associated with ischemia, injury, and components of innate immunity, from the late response, which includes genes enriched for components of adaptive immunity.

Eight- to 12-wk-old male mice, including BALB/cByJ (BALB/c) (H-2d), C57BL/6J (B6) (H-2b), C57BL/6J-Rag-1tm1Mom (B6-Rag) (H-2b) (The Jackson Laboratory, Bar Harbor, ME), and BALB/c-AnNTac-Rag2tm1N12 (BALB/c-Rag) (H-2d) (Taconic Farms, Germantown, NY), were used as donors and recipients in the transplant experiments. As previously described (13), hearts were harvested from donors and immediately transplanted into 8- to 12-wk-old recipients which were anesthetized via i.p. injection with 60 μg/kg of pentobarbital sodium. The donor aorta was attached to the recipient abdominal aorta by end to side anastomosis, and the donor pulmonary artery was attached to the recipient vena cava by end to side anastomosis. All surgical procedures were completed in <60 min from the time that the donor heart was harvested. Donor hearts that did not beat immediately after reperfusion or stopped within 1 day following transplantation were excluded (>95% of all grafts functioned at 1 day following transplantation). The native heart of the recipient was not surgically manipulated and remained functional.

dscDNA was synthesized from RNA samples by means of the SuperScript Choice system (Life Technologies, Rockville, MD) and a T7-(dT) 24 Primer (Genset Oligos, La Jolla, CA). The cDNA was purified using phenol/chloroform extraction with Phase Lock gel (Brinkmann Instruments, Westbury, NY) and concentrated by ethanol precipitation. In vitro transcription was performed to produce biotin-labeled cRNA using a BioArray HighYield RNA Transcript Labeling kit (Enzo Diagnostics, Farmingdale, NY) according to the manufacturer’s instructions. cRNA was linearly amplified ∼40-fold with T7 polymerase using dscDNA that was synthesized. The biotinylated RNA was purified with an RNeasy Mini kit (Qiagen, Valencia, CA), fragmented to 50–200 nt, and then hybridized to an Affymetrix murine array (Mu11kB; Santa Clara, CA), which contains probe sets for 6500 genes and expressed sequence tags. After being washed, the array was stained with streptavidin-PE (Molecular Probes, Eugene, OR), amplified by biotinylated anti-streptavidin (Vector Laboratories, Burlingame, CA), and analyzed on a Hewlett-Packard Genearray scanner (Cupertino, CA).

Array data was analyzed with Microarray Suite 4.0.1 (Affymetrix). A single expression level for each gene was derived from the 20 probe pairs representing each gene, 20 perfectly matched (PM) and mismatched (MM) control probes. The MM probes act as specificity control that allow the direct subtraction of background and cross-hybridization signals. Each array was normalized to a standard of 2500 U/probe set. To determine the quantitative RNA level, the average of the differences (avg diff) representing PM−MM for each gene-specific probe set was calculated. The expression of each probe set was categorized as present (P), marginal (M), or absent (A). Calculations of means and variances were performed with JMP statistical software (SAS Institute, Cary, NC).

Total RNA isolated from murine untransplanted and transplanted allogeneic and syngenic graft heart samples was reverse transcribed using SuperScript II RNase Reverse Transcriptase (Life Technologies, Carlsbad, CA). Specific primer pairs were designed using the Primer Express software (Applied Biosystems, Foster City, CA). Direct detection of the PCR product was monitored by measuring the increase in fluorescence caused by the binding of SYBR Green to dsDNA. Reactions were performed in a MicroAmp Optical 96-well reaction plate (Applied Biosystems) using, for each separate well, 5 μl of cDNA mix, 5 μl of primer, and 10 μl of SYBR Green Master Mix (Applied Biosystems). Each well contained the primer pair for amplification of one of the parameters of interest. The gene-specific PCR products are continuously measured by means of the GeneAmp 5700 Sequence Detection system (Applied Biosystems) during 40 cycles. All experiments were run in duplicate and the same thermal cycling parameters were used. Nontemplate controls and dissociation curves were used to detect primer-dimer conformation and nonspecific amplification. The threshold cycle (CT) of each target product is determined and set in relation to the amplification plot of GAPDH. The CT is the number of PCR cycles required for the fluorescence signal to exceed the detection threshold value. The detection threshold is set to the log linear range of the amplification curve and kept constant (0.05) for all data analysis. With the PCR efficiency of 100%, the difference in CT values (ΔCT) of two genes can be used to calculate the fold difference (fold difference = 2 −(CT1−CTcontrol)= 2−ΔCT). The relative quantitation results are used to determine patterns of graft gene expression change in response to allogeneic and syngeneic transplantation (19, 20).

To analyze gene expression during an in vivo alloimmune response, we performed murine heterotopic cardiac transplants in an allogeneic (BALB/c→B6) strain combination that has a complete MHC class I and II mismatch. As previously reported, these grafts are rejected at ∼8 days (not shown). RNA was prepared from untransplantated control donor strain hearts, untransplanted recipient strain lymph nodes, and grafts at day 1 and 7 following transplantation. RNA was then analyzed with oligonucleotide microarrays. Total fluorescence for each array was normalized, absolute difference values were calculated, and each gene was determined to be A, M, or P by Microarray Suite algorithms. All genes that lacked a present call in at least one sample were masked from subsequent analyses. To determine the reproducibility of array data, we analyzed duplicate samples from control RNA from untransplanted hearts. Scatter plots showed a strong linear correlation (r = 0.98) of positively expressed genes (Fig. 1,a). Analysis of the distribution of the ratios of control values showed a variance of only 0.09. As expected, a scatter plot of lymph node vs control heart showed increased dispersion (r = 0.29) with a variance of 18.61 (Fig. 1,b), due to differential gene expression in the heart and lymph node tissues. To determine the modulation of gene expression following transplantation, we analyzed the ratios of the days 1 and 7 graft heart vs untransplanted control heart RNA. The scatter plot of day 1 vs control showed increased dispersion (r = 0.87), which was confirmed by analysis of the distribution of the ratios that had an increased variance of 1.49 (Fig. 1,c). The scatter plot of day 7 vs control showed further increased dispersion (r = 0.60), which correlated with a greater variance of 13.83 (Fig. 1 d). These results demonstrate an increase in differentially expressed genes during the rejection process.

FIGURE 1.

Scatterplot of expression profiles. RNA was harvested from duplicate control hearts, lymph nodes, and allogeneic grafts from days 1 and 7, and was hybridized to the Mu11kB DNA microarray; the average difference values were calculated for each probe set by Microarray Suite software. For each plot, the ordinate is control heart (sample 1). The abscissas are control heart (sample 2; a), lymph node (b), day 1 allograft (c), and day 7 allograft (d). An absolute call was determined for each probe set as P (black), M (black), or A (gray). Correlation coefficients were calculated using values that were determined to be present in at least one sample.

FIGURE 1.

Scatterplot of expression profiles. RNA was harvested from duplicate control hearts, lymph nodes, and allogeneic grafts from days 1 and 7, and was hybridized to the Mu11kB DNA microarray; the average difference values were calculated for each probe set by Microarray Suite software. For each plot, the ordinate is control heart (sample 1). The abscissas are control heart (sample 2; a), lymph node (b), day 1 allograft (c), and day 7 allograft (d). An absolute call was determined for each probe set as P (black), M (black), or A (gray). Correlation coefficients were calculated using values that were determined to be present in at least one sample.

Close modal

Of a total of 6519 genes analyzed by the microarrays, 974 were present in control hearts (Fig. 2). Of these 974 genes expressed at baseline, only 593 were detected in grafts 1 day following transplantation; however, 288 genes that were not detected in the control samples were present at day 1. At day 7 following transplantation, 495 of the genes that were expressed in the control sample were still detected, plus 97 genes, that were in the control but absent at day 1, were re-expressed. Also, 122 of the new genes expressed at day 1 were still detected at day 7, plus 212 additional genes not previously detected were present. Thus, almost half (479 of 974) of the genes constitutively expressed in control hearts were lost to detection during the rejection process, whereas a total of 400 new genes were detected due to the response to injury and the induction of inflammation.

FIGURE 2.

Flow diagram of expressed genes that were present following transplantation. A total of 6519 genes were analyzed in control hearts days 1 and 7 following transplantation and were categorized as P or absent/marginal (A).

FIGURE 2.

Flow diagram of expressed genes that were present following transplantation. A total of 6519 genes were analyzed in control hearts days 1 and 7 following transplantation and were categorized as P or absent/marginal (A).

Close modal

To obtain an overview of the kinetics of allograft rejection, we analyzed absolute difference gene expression at days 1 and 7 compared with control samples by hierarchical clustering algorithms which perform pairwise calculations for each gene using Pearson correlation coefficients (Fig. 3,a). As expected, the two control samples showed minimal dissimilarity. The days 1 and 7 groups were moderately dissimilar, but these posttransplant groups were most dissimilar from the untransplanted control samples. Next, we analyzed the genes most highly up-regulated at day 1 following transplantation compared with control by two approaches, fold change and absolute change in expression (Table I). Fold change analysis is less sensitive for detection of genes expressed constitutively, whereas absolute change is less sensitive for detection of genes expressed at low levels; therefore, the two approaches can detect different subsets of differentially regulated genes. First, using fold increase in expression, we detected metallothionein, calpactin I H chain, major excreted protein, and T lymphoma oncogene. Second, using absolute increase in expression we detected IFN β, serglycin, EN-7, osteopontin, Mac-2, Eta-1, β tubulin, fibronectin, gly96, mbh1. Of note, data analysis by these two approaches detected different subsets of genes; however, both approaches identified genes known to be involved in response to injury or stress. Importantly, at day 1 there was a conspicuous lack of immune genes with known associations with lymphocytes supporting the hypothesis that the early phase is predominantly lymphocyte independent. However, at day 7 following transplantation, numerous immune related genes were detected (Table I). Based on fold increase, highly up-regulated genes included invariant chain, MHC class II H chain, C1q C-chain, C1q α-chain, β2-microglobulin, class I H-2Dk, and C1q B-chain. In addition, based on an absolute increase in expression, we detected Mac-2 and TCR β-chain. Taken together, these results suggest that the early response at day 1 following transplantation consists predominantly of the innate immune response to injury and stress, whereas the later response at day 7 includes many components of adaptive immunity.

FIGURE 3.

Dendrogram of gene expression following transplantation. Agglomerative hierarchical clustering algorithms were used. x-axis distance is proportional to the dissimilarity between groups. a, Dendrogram of control hearts and days 1 and 7 graft hearts. b, Dendrogram of control hearts, control lymph nodes, and days 1 and 7 graft hearts. c, Dendrogram of allogeneic graft, alymphoid graft, allogeneic native, and control hearts.

FIGURE 3.

Dendrogram of gene expression following transplantation. Agglomerative hierarchical clustering algorithms were used. x-axis distance is proportional to the dissimilarity between groups. a, Dendrogram of control hearts and days 1 and 7 graft hearts. b, Dendrogram of control hearts, control lymph nodes, and days 1 and 7 graft hearts. c, Dendrogram of allogeneic graft, alymphoid graft, allogeneic native, and control hearts.

Close modal
Table I.

Genes up-regulated following transplantationa

IdentificationDescriptionFold ChangeAbsolute Difference
Day 1/allogeneic graft V00835 Metallothionein-I 9.4  
 L20315 MPS1 6.4  
 D00472 Cofilin 4.6  
 D10024 Calpactin 1 H chain 4.8  
 X06086 Major excreted protein (MEP) 4.6  
 X52634 tlm 4.3  
 V00755 IFNβ (type 1)  1958 
 X16133 Serglycin  1818 
 X81627 24p3  1037 
 X53247 EN-7  992 
 X51834 Osteopontin  802 
 X16834 Mac-2 Ag  581 
 X04663 β-tubulin (isotype Mβ 5)  486 
 M18194 Fibronectin  423 
 X67644 gly96  371 
 X54511 Myc basic motif homologue-1 (mbh1)  361 
     
Day 7/allogeneic graft X00496 Ia-associated invariant chain (li) 84.5  
 V00832 Ia Ag (H chain) from MHC 39.8  
 X66295 C1q C-chain 13.4  
 X58861 C1Q α-chain 10.2  
 X14425 Profilin 8.1  
 X01838 β2-microglobulin 8.0  
 X14425 Profilin 7.4  
 ET61563 H-2 class I histocompatibility Ag D-k α 7.3  
 M22531 C1q B chain 7.0  
 M21495 Cytoskeletal γ-actin 6.9  
 X00246 Set 1 repetitive element for a class I MHC 6.7  
 M17327 Murine leukemia virus modified polytropic provirus DNA 6.4  
 D00472 Cofilin 6.0  
 V00755 IFNβ (type 1)  3149 
 X16133 Serglycin  2588 
 X53247 EN-7  2564 
 X16834 Mac-2 Ag  2559 
 M26053 TCR germline β-chain  1496 
     
Lymph node X00496 Ia-associated invariant chain (li) 84.5  
 V00832 Ia Ag (H chain) from MHC 39.8  
 M17327 Murine leukemia virus modified polytropic provirus DNA 6.4  
 Y09010 Serine/threonine kinase 2.7  
 M29325 L1Md-9 repetitive sequence 2.5  
     
Heart X04405 Myoglobin 0.2  
 U27315 Adenine nucleotide translocase-1 0.2  
 X01756 Cytochrome c 0.3  
IdentificationDescriptionFold ChangeAbsolute Difference
Day 1/allogeneic graft V00835 Metallothionein-I 9.4  
 L20315 MPS1 6.4  
 D00472 Cofilin 4.6  
 D10024 Calpactin 1 H chain 4.8  
 X06086 Major excreted protein (MEP) 4.6  
 X52634 tlm 4.3  
 V00755 IFNβ (type 1)  1958 
 X16133 Serglycin  1818 
 X81627 24p3  1037 
 X53247 EN-7  992 
 X51834 Osteopontin  802 
 X16834 Mac-2 Ag  581 
 X04663 β-tubulin (isotype Mβ 5)  486 
 M18194 Fibronectin  423 
 X67644 gly96  371 
 X54511 Myc basic motif homologue-1 (mbh1)  361 
     
Day 7/allogeneic graft X00496 Ia-associated invariant chain (li) 84.5  
 V00832 Ia Ag (H chain) from MHC 39.8  
 X66295 C1q C-chain 13.4  
 X58861 C1Q α-chain 10.2  
 X14425 Profilin 8.1  
 X01838 β2-microglobulin 8.0  
 X14425 Profilin 7.4  
 ET61563 H-2 class I histocompatibility Ag D-k α 7.3  
 M22531 C1q B chain 7.0  
 M21495 Cytoskeletal γ-actin 6.9  
 X00246 Set 1 repetitive element for a class I MHC 6.7  
 M17327 Murine leukemia virus modified polytropic provirus DNA 6.4  
 D00472 Cofilin 6.0  
 V00755 IFNβ (type 1)  3149 
 X16133 Serglycin  2588 
 X53247 EN-7  2564 
 X16834 Mac-2 Ag  2559 
 M26053 TCR germline β-chain  1496 
     
Lymph node X00496 Ia-associated invariant chain (li) 84.5  
 V00832 Ia Ag (H chain) from MHC 39.8  
 M17327 Murine leukemia virus modified polytropic provirus DNA 6.4  
 Y09010 Serine/threonine kinase 2.7  
 M29325 L1Md-9 repetitive sequence 2.5  
     
Heart X04405 Myoglobin 0.2  
 U27315 Adenine nucleotide translocase-1 0.2  
 X01756 Cytochrome c 0.3  
a

Differential gene expression via Microarray analysis is shown for allogeneic graft day 1, allogeneic graft day 7, lymph node, and heart. Relative gene expression is calculated by two methods: fold change and absolute difference.

Microarray data was independently verified by quantitative real-time PCR. Duplicate allogeneic and syngeneic murine cardiac transplants were performed. Primers were created from genes identified in Table I as up-regulated by fold increase and by absolute increase (Mac-2, serglycin, 24p3, C1q subunits, and β2-microglobulin). Overall, the expression patterns for allogeneic graft samples as measured by these two independent methods were similar (Fig. 4 and Table I). In both the allogeneic graft microarray and PCR data, 24p3 is highly expressed at day 1 but not in day 7,serglycin and Mac-2 are highly expressed in both allogeneic days 1 and 7 and C1q subunits and β2-microglobulin are highly expressed at day 7.

FIGURE 4.

Independent verification of microarray quantitation. Relative mRNA levels of 24p3, Mac-2, C1q subunits, serglycin, and β2-microglobulin (b2-M) were measured with quantitative real-time PCR in repeat murine allogeneic and syngeneic cardiac transplants. Each data point was normalized to mRNA expression of the untransplanted control heart. Normalization of the data to control hearts allows for comparison of the PCR results to the microarray fold change and absolute difference data. In the allogeneic transplants, quantitation with both methods gave similar patterns of gene expression.

FIGURE 4.

Independent verification of microarray quantitation. Relative mRNA levels of 24p3, Mac-2, C1q subunits, serglycin, and β2-microglobulin (b2-M) were measured with quantitative real-time PCR in repeat murine allogeneic and syngeneic cardiac transplants. Each data point was normalized to mRNA expression of the untransplanted control heart. Normalization of the data to control hearts allows for comparison of the PCR results to the microarray fold change and absolute difference data. In the allogeneic transplants, quantitation with both methods gave similar patterns of gene expression.

Close modal

To differentiate between injury-induced genes and those expressed in response to an immune response, repeat syngeneic transplants were performed and the graft mRNA analyzed by real-time PCR (Fig. 4). Consistent with previously described gene functions, the syngeneic day 1 data suggest that 24p3, Mac-2, and serglycin are genes likely expressed in response to the injury associated with transplantation. 24p3 is a known acute phase protein and Mac-2 is a galactose binding lectin and a monocyte/macrophage differentiation and activation marker. (21) Serglycin, the major proteoglycan of endothelial cells, (22) is up-regulated in response to inflammatory mediators TNF-α and IL-1α (23). β2-microglobulin, present on all nucleated cells as a subunit of the class I MHC, is up-regulated on syngeneic day 1 only. Consistent with our observations, β2-microglobulin is up-regulated by inflammatory mediators IFNα, IFNβ, IFN-γ, and TNF-α (24) which, in the syngeneic graft, are likely generated due to the tissue injury of transplantation. In the allogeneic graft, β2-microglobulin is further up-regulated at day 7 due to rejection. The C1q subunits are not up-regulated in the syngeneic PCR data suggesting that C1q is not involved in the injury or stress response to transplantation.

To test the kinetics of rejection in this model, we repeated the hierarchical clustering algorithm with inclusion of our control lymph node sample (Fig. 3,b). These results clearly show association between the day 7 allograft and lymph node samples, whereas the day 1 sample associates more closely with the untransplanted control heart samples. To confirm that our data differentiated heart and lymph node genes, we analyzed expression ratios identifying invariant chain, MHC class II, Igκ chain (two different fragments) as the most highly expressed relative to heart (Table I). Conversely, genes most highly expressed in hearts compared with lymph nodes included myoglobin, adenine nucleotide translocase-1, and cytochrome c (Table I). These results are consistent with our prior information of gene expression in lymph node and heart tissues.

To evaluate the overall contribution of innate immunity to the early response, we determined the dissimilarity among transcriptional profiles of samples from alymphoid (BALB-recombinase-activating gene (RAG)→B6-RAG) graft hearts, allogeneic (BALB/c→B6) graft hearts, native hearts, and untransplanted control hearts using a hierarchical clustering algorithm (Fig. 3 c). The alymphoid group lacks functional T and B lymphocytes due to the deficiency of RAG. Our results show that at day 1 following transplantation, the alymphoid and allogeneic grafts have only a small amount of dissimilarity indicating that the early response is, at least in part, independent of adaptive immunity. At day 7 following transplantation, the profile of the alymphoid group is similar to control hearts suggesting that in the absence of an alloimmune stimulus, the inflammatory response resolves. In contrast, the allogeneic group at day 7 following transplantation is highly dissimilar from all other experimental groups consistent with the generation of an adaptive response.

Based on our results showing modulated expression of substantial numbers of genes 1 day following transplantation, we next analyzed expression 6 h following allogeneic transplantation. To detect clusters of genes with similar patterns of expression in the various experimental groups, we applied SOM to microarray data from samples from allogeneic grafts at 6 h, 1 day, and 7 days; in addition, we included control data from alymphoid grafts and untransplanted control hearts (Fig. 5 and Table II). Using a heuristic approach to optimize clustering and minimize SD, we tested numerous geometries, numbers of epochs, and normalization parameters. Using a 3 × 4 geometry, 100 epochs, and a row variation threshold of 3, we generated 12 clusters (0–11) that differentiated gene expression in terms of the kinetics of rejection. A comparison of clusters 2 and 8 support the specificity of the analysis based on comparison of clustered genes with prior biological information. Cluster 2 contains a cluster of 37 genes with low expression in lymph node, but high expression in all heart samples. Genes included in cluster 2 include myosin H chain, myosin L chain 2, cytochrome c oxidase, myoglobin, myosin L chain, troponin I, α cardiac actin (Table II). In contrast, cluster 8, which has high expression in lymph node, includes CD3δ, T11, mb-1, Ig L chain, H2-M, p56-tck, Thy-1, and TCRβ among others. These results are consistent with previous studies of heart and lymph node tissue.

FIGURE 5.

SOM. Clusters of genes with similar patterns of expression in the experimental groups were identified using SOM. Experimental groups distributed on the x-axis include control heart 1 (♦), allogeneic graft 6 h following transplantation (), allogeneic graft 1 day following transplantation (▵), allogeneic graft 7 day following transplantation (★), alymphoid graft (▦), and control lymph node ( ). The y-axis is relative expression, which is autoscaled to enhance visualization and is thus variable among the 12 profiles. The mean expression is depicted as the symbols with error bars indicating 2 SD. The number of genes in each cluster is 23 (c0), 18 (c1), 37 (c2), 50 (c3), 26 (c4), 20 (c5), 30 (c6), 47 (c7), 46 (c8), 22 (c9), 24 (c10), and 29 (c11). The algorithm was initiated with 3 × 4 geometry using 100 epochs. Based on multiple heuristic observations, increased numbers of nodes produced clusters with low numbers of genes, whereas decreased numbers of nodes produced larger SD. Increasing the number of epochs (=500) did not produce detectable changes in the clusters or SD.

FIGURE 5.

SOM. Clusters of genes with similar patterns of expression in the experimental groups were identified using SOM. Experimental groups distributed on the x-axis include control heart 1 (♦), allogeneic graft 6 h following transplantation (), allogeneic graft 1 day following transplantation (▵), allogeneic graft 7 day following transplantation (★), alymphoid graft (▦), and control lymph node ( ). The y-axis is relative expression, which is autoscaled to enhance visualization and is thus variable among the 12 profiles. The mean expression is depicted as the symbols with error bars indicating 2 SD. The number of genes in each cluster is 23 (c0), 18 (c1), 37 (c2), 50 (c3), 26 (c4), 20 (c5), 30 (c6), 47 (c7), 46 (c8), 22 (c9), 24 (c10), and 29 (c11). The algorithm was initiated with 3 × 4 geometry using 100 epochs. Based on multiple heuristic observations, increased numbers of nodes produced clusters with low numbers of genes, whereas decreased numbers of nodes produced larger SD. Increasing the number of epochs (=500) did not produce detectable changes in the clusters or SD.

Close modal
Table II.

Gene clusters selected by SOMa

ClusterIdentifierGene Description
X62742 Ma 
 W11156 IP-30 precursor 
 X16874 C1q B-chain 
 U34277 PAF acetylhydrolase 
 M86736 Acrograinin 
 X16834 Mac-2 Ag 
 X56603 Calcium-binding protein 
 X58861 C1q α-chain 
 L20315 MPS1 
 ET62967 C1q C chain precursor 
 X66295 C1q C chain 
 M22531 C1q B chain 
 U27106 Clathrin-associated AP-2 complex AP50 subunit 
   
M18194 Fibronectin (FN) 
 X51834 Osteopontin 
 D00466 Apolipoprotein E 
 U08020 Collagen pro-α-1 type I chain 
 D12907 HSP47 
 X04017 Cysteine-rich glycoprotein SPARC 
 X52046 Collagen α-1 
 X52886 Cathepsin D 
 X58251 Pro-α-2(I) collagen 
 X59047 MD3 
 X65582 α-2 collagen VI 
 X66532 L14 lectin 
 X80638 RhoC 
   
M73741 α-B2-crystallin 
 M76601 α cardiac myosin H chain 
 M91602 Myosin L chain 2 
 U08439 Cytochrome c oxidase subunit VIaH 
 V00835 Metallothionein-I 
 X04405 Myoglobin 
 X67685 Ventricular alkali myosin L chain 
 Z22784 Troponin I 
 M15501 α-cardiac actin 
 M32599 Glyceraldehyde-3-phosphate dehydrogenase 
 M32599 Glyceraldehyde-3-phosphate dehydrogenase 
 V00722 β-1-globin 
 X03767 Adult cardiac muscle β-actin 
 X52940 Cytochrome c oxidase VIIc 
 Y07708 NADH dehydrogenase 
   
X63349 Tyrosine-related protein-2 
 U30840 Voltage dependent anion channel 1 
 X03233 Muscle creatine kinase 
 X61600 β-enolase 
 X01756 Cytochrome c 
 X51905 Lactate dehydrogenase-B 
 X51905 Lactate dehydrogenase-B 
 X53929 Decorin 
 Z38015 DMR-N9 and DM-PK gene 
 Z49204 NADP transhydrogenase 
 L01062 ATP synthase α subunit 
 M36084 Cystolic malate dehydrogenase 
 U27315 Adenine nucleotide translocase-1 
 X15963 Cytochrome c oxidase subunit Va 
 X58486 Cytochrome c oxidase subunit VIIa 
   
X00619 TCR (helper hybridoma 2B4) 
 V01527 Major histocompatibility Ag I-A-β 
 X00958 I-E (β-b) 
 K01923 MHC class II H2-IA α 
 X91144 P-selectin glycoprotein ligand 1 
 U22031 20S proteasome subunit Lmp7 
 X53247 EN-7 
 Z16078 CD53 gene exon 7 
 V00832 Ia Ag from MHC 
(Table continues  
ClusterIdentifierGene Description
X62742 Ma 
 W11156 IP-30 precursor 
 X16874 C1q B-chain 
 U34277 PAF acetylhydrolase 
 M86736 Acrograinin 
 X16834 Mac-2 Ag 
 X56603 Calcium-binding protein 
 X58861 C1q α-chain 
 L20315 MPS1 
 ET62967 C1q C chain precursor 
 X66295 C1q C chain 
 M22531 C1q B chain 
 U27106 Clathrin-associated AP-2 complex AP50 subunit 
   
M18194 Fibronectin (FN) 
 X51834 Osteopontin 
 D00466 Apolipoprotein E 
 U08020 Collagen pro-α-1 type I chain 
 D12907 HSP47 
 X04017 Cysteine-rich glycoprotein SPARC 
 X52046 Collagen α-1 
 X52886 Cathepsin D 
 X58251 Pro-α-2(I) collagen 
 X59047 MD3 
 X65582 α-2 collagen VI 
 X66532 L14 lectin 
 X80638 RhoC 
   
M73741 α-B2-crystallin 
 M76601 α cardiac myosin H chain 
 M91602 Myosin L chain 2 
 U08439 Cytochrome c oxidase subunit VIaH 
 V00835 Metallothionein-I 
 X04405 Myoglobin 
 X67685 Ventricular alkali myosin L chain 
 Z22784 Troponin I 
 M15501 α-cardiac actin 
 M32599 Glyceraldehyde-3-phosphate dehydrogenase 
 M32599 Glyceraldehyde-3-phosphate dehydrogenase 
 V00722 β-1-globin 
 X03767 Adult cardiac muscle β-actin 
 X52940 Cytochrome c oxidase VIIc 
 Y07708 NADH dehydrogenase 
   
X63349 Tyrosine-related protein-2 
 U30840 Voltage dependent anion channel 1 
 X03233 Muscle creatine kinase 
 X61600 β-enolase 
 X01756 Cytochrome c 
 X51905 Lactate dehydrogenase-B 
 X51905 Lactate dehydrogenase-B 
 X53929 Decorin 
 Z38015 DMR-N9 and DM-PK gene 
 Z49204 NADP transhydrogenase 
 L01062 ATP synthase α subunit 
 M36084 Cystolic malate dehydrogenase 
 U27315 Adenine nucleotide translocase-1 
 X15963 Cytochrome c oxidase subunit Va 
 X58486 Cytochrome c oxidase subunit VIIa 
   
X00619 TCR (helper hybridoma 2B4) 
 V01527 Major histocompatibility Ag I-A-β 
 X00958 I-E (β-b) 
 K01923 MHC class II H2-IA α 
 X91144 P-selectin glycoprotein ligand 1 
 U22031 20S proteasome subunit Lmp7 
 X53247 EN-7 
 Z16078 CD53 gene exon 7 
 V00832 Ia Ag from MHC 
(Table continues  
a

All known genes from the 12 clusters are shown with cluster number, identifier, and gene description. All expressed sequence tags were eliminated for simplicity.

Additional clusters differentiated genes expressed during the early and late phases of rejection. Cluster 0 includes a cluster of 23 genes highly up-regulated at day 7 following transplantation, but not highly expressed in lymph nodes. This cluster includes several complement genes, acrogranin, Mac-2, and IFN-γ-inducible protein (IP)-30 precursor. Cluster 4 includes 26 genes also up-regulated at day 7, but in addition, these genes are highly expressed in lymph nodes. Cluster 4 includes TCR, I-Aβ, I-Eβ, I-Aα, P-selectin, 20S proteasome, CD53, and calmodulin. These results differentiate genes highly expressed at day 7 based on high or low expression in lymph nodes and delineate two components of the late phase of rejection. In contrast, cluster 11 identified a cluster of 29 genes that are highly up-regulated at 6 h following transplantation. Genes up-regulated at this early time point include IL-1R, macrophage-inflammatory protein (MIP)-2, IL-6, haptoglobin, MIP-1β, IκBα, CD14, and metalloproteinase-3 tissue inhibitor, among others. Consistent with the expression profile, these genes have been associated with the acute phase response and early proinflammatory responses, but in general are not major components of adaptive immunity.

Several clusters, including 5, 6, 9, and 10, include genes up-regulated in both the allogeneic and alymphoid groups. Genes in these clusters include many genes that have been associated with stress, injury, and wounding, but conspicuously lack genes commonly associated with adaptive immunity. For example, genes included in these clusters include cofilin, profilin, ferritin, calmodulin, secreted protein acidic-rich cysteine, Gla, major excreted protein, GPI, vimentin, heat shock protein (HSP)60, integrin β subunit, junD, serum amyloid A 3, IFNβ, serglycin, calpactin I HSP70, and lactate dehydrogenase A among others. Interestingly, as shown in cluster 3, there is also a subset of genes down-regulated following transplantation in both the allogeneic and alymphoid recipients suggesting that these genes are down-regulated in response to stress or injury. Taken together, the SOMs have differentiated clusters of genes with distinct patterns of expression that identify distinct phases of the process of alloimmunity.

Most studies of transplantation have analyzed a small subset of parameters, often a single gene product, in attempts to understand the pathogenesis of rejection. For example, in murine heterotopic heart transplantation, a large number of studies analyzing knockout strains, which were selected based on prior data suggesting they would be important in the alloimmune response, have been studied. Interestingly, many of the knockout strains studied, including strains deficient in cytokines, chemokines, receptors, adhesion molecules and transcription factors, produce a similar outcome: rejection is delayed, but not indefinitely prevented. These observations suggest that the alloimmune response includes multiple components that are functionally overlapping and capable of compensating, at least partially, for gene deficiencies. Based on these arguments, we reasoned that a global analysis of gene expression during allograft rejection using oligonucleotide microarrays could provide important insights into the alloimmune response.

A previous study of cardiac allograft rejection using DNA microarrays reported that the majority of highly inducible genes, which were analyzed at day 5 following transplantation, were IFN-inducible (25). We previously reported that serum levels of IFN-γ surged at day 5, but had already decreased by day 7 (13). Increased levels of serum IFN-γ at day 5 are consistent with the identification of IFN-γ-inducible genes in the graft; however, decreasing levels of serum IFN-γ at day 7, which corresponds to the time of graft rejection, suggest that IFN-γ is not a crucial component of the rejection process. This interpretation is confirmed by the development of acute rejection in IFN-γ deficient allograft recipients (26). Thus, multiple IFN-γ-independent components of alloimmunity, which remain poorly characterized, are necessary to promote rejection.

In our study, the SOM identified clusters of up-regulated genes as early as 6 h after transplantation. SOM analysis differentiated clusters of genes up-regulated 6 h following transplantation based on different levels of expression during late phases of rejection and in lymph node tissue. For example, cluster 11 contains 29 genes highly up-regulated at 6 h, but not up-regulated in any of the other experimental groups. Cluster 11 includes IL-1R, IL-6, and haptoglobin, which have all been associated with the acute phase response (27, 28, 29, 30). In addition, cluster 11 includes MIP-2, a chemokine known to be induced by ischemia, MIP-1β, an early component of the inflammatory response, and CD14, an important component in Toll-like receptor 4 signaling during innate immune responses (31, 32, 33). Also included was the immediate early gene 3CH134, which has been shown to be up-regulated by ischemia (34, 35). Up-regulation of 3CH134 at the 6-h time point is consistent with its known expression as an immediate early factor and inducibility by ischemia. Of note, two independent chip probes hybridized to the IκBα gene, which has been previously shown to be up-regulated following transplantation (36) supporting the reproducibility of the analysis.

The SOM also identified genes up-regulated at both 6 h and 7 days following transplantation. For example, cluster 4 contains 26 genes up-regulated at both early and late times following transplantation (and in lymph node tissue). Genes in this cluster include MHC class II (identified by four different probes) and TCR, genes necessary for rejection (7, 9), P-selectin, a gene important in cell trafficking and rejection (37), CD53, a pan-leukocyte marker that has been shown to activate lymphocytes (38), and Lmp7, a component of the 20S proteasome important for Ag processing (39). The mean level of expression of these genes was greater at day 7 than at 6 h, which is consistent with the association of these genes with Ag-specific responses.

The third important pattern of expression includes genes up-regulated at day 7 following transplantation, but not in any of the other groups. Because allograft rejection occurs at ∼7.8 days in our experimental protocol, these genes are highly up-regulated immediately before the time of rejection. Genes included in cluster 0 are IP-30 (40), Mac-2, a macrophage cell surface protein that binds extracellular matrix (41, 42), monopolar spindle 1, AP50, a component of the CTLA-4 signaling complex (43, 44), and C1q (identified by four different probes), which is a C-type lectin member of the collectin family important in innate immunity and up-regulated in the serum following renal transplantation (45, 46). Our results suggest that complement may be important in allogeneic rejections.

Clusters 1, 5, and 10 identified genes that were up-regulated in all groups, including alymphoid grafts, following transplantation suggesting induction by Ag nonspecific mechanisms such as ischemia, injury, or stress. Consistent with this notion, cluster 1 includes fibronectin (47, 48); HSP47, a collagen-binding chaperone induced by stress and transformation (49, 50); SPARC, a matricellular protein that responds to injury (51); collagen components (identified by four different probes) important in wound injury (52); the serine protease cathepsin D (53); and L14, a S-type lectin (54). Clusters 5 and 10, which include calmodulin, shown to be induced by ischemia (55); serum amyloid A 3, a component of the acute phase response (56); IFNβ, a factor that accelerates rejection (57); serglycin, induced by TNF-α and IL-1α (23); calpactin I sulfated glycoprotein (Sgp) 1, previously identified in Sertoli cells and preimplantation embryos, but without a clear association with transplantation (58, 59); and HSP70, a gene up-regulated following transplantation and shown to be protective against ischemic injury (60, 61), are consistent with the concept that these genes function in Ag-independent responses. Interestingly, we also identified genes (cluster 3) down-regulated in all groups following transplantation. Of note, both cyclophilin and FK506 binding protein were up-regulated posttransplantation.

Cluster 8 identified 46 genes that are highly expressed in lymph nodes, but not hearts, and cluster 2 detected 37 genes highly expressed in hearts, but not lymph nodes. Neither of these clusters showed differential expression following transplantation; however, the specific genes identified are consistent with previous biological characterizations of heart and lymph node tissues. Genes highly expressed in hearts (cluster 2) include B2-crystallin, myosin H chain, myosin L chain 2, cytochrome c oxidase, myoglobin, ventricular alkali myosin L chain, troponin I, α-cardiac actin, G3PDH (two probes), cardiac α-actin, cytochrome c oxidase VIIc, and NADH dehydrogenase. In contrast, genes highly expressed in lymph nodes (cluster 8) include CD3δ, T11, mb-1, Ig L chain, H2-M, p56-tck, Thy-1, TCRβ, IgH (two probes), T cell-specific transcription factor, and Igκ (three probes). These observations strongly support the power of SOM to cluster genes according to biological functions.

Microarray studies are, in general, limited by the probes present on the array. The arrays used in our study consist of probes for only a portion of the mouse genome. Specific genes that may prove to be important in graft rejection may be absent from the array used in this study. Our results then are a representation of the genes involved in transplantation.

Our results demonstrate the dynamic and complex character of alloimmunity. Our kinetic analysis using hierarchical clustering dendrograms differentiated the rejection process into two broad phases: the early (innate) and late (adaptive) phases. The identification of specific genes in each phase indicates that the innate and adaptive phases are composed of multiple components. For example, based on correlations with prior understanding of gene function, our analysis demonstrates that the innate phase includes complement, stress, ischemia, and injury components. Similarly, the adaptive phase includes markers for CD4, CD8, and B cell responses. And it is not unreasonable that additional components are operative during rejection. Taken together, these observations pose an important question: what are the minimal components necessary to promote rejection? Identification of these components would suggest definitive diagnostic criteria of rejection and essential targets for therapeutic intervention. Traditional reductionist approaches have attempted to identify (with limited success) a single or small number of genes essential for the rejection process. Our results suggest the possibility that the in vivo alloimmune response includes multiple components that are regulated in a modular system or network. A profound understanding of the complex biological response of alloimmunity will likely require the integration of reductionist studies and global analyses.

Table 2A.

Continued

ClusterIdentifierGene Description
 X61432 Calmodulin 
 ET61563 H-2 class I histocompatibility Ag 
 X00246 Set 1 repetitive element for class I MHC 
 X00496 Ia-associated invariant chain (li) 
 X01838 β-microglobulin β2 
 X65553 poly(A) binding protein 
 X65553 poly(A) binding protein 
 X84037 E-selectin ligand-1 
   
D00472 Cofilin and flanks 
 X14425 Profilin 
 L39879 Ferritin L-subunit 
 M12481 Cytoplasmic β-actin 
 M21495 Cytoskeletal γ-actin 
 M76131 Elongation factor 2 
 X05021 Homology to yeast L29 ribosomal protein 
 X51528 Transplantation Ag P198 
 X52803 Cyclophilin 
 X61432 Calmodulin 
 X73729 Ribosomal protein S8 
 X74856 Ribosomal protein L28 
 X74856 Ribosomal protein L28 
 X75313 GB-like 
 X75313 GB-like 
 X81987 TAX responsive element binding protein 107 
 X81987 TAX responsive element binding protein 107 
 Y0025 J1 protein, yeast ribosomal protein L3 homolog 
   
V00719 Liver α-amylase 
 D00613 Matrix Gla protein (MGP) 
 U13705 Plasma glutathione peroxidase 
 X04017 Cysteine-rich glycoprotein SPARC 
 X06086 MEP 
 X02520 Lactate dehydrogenase A4 isoenzyme 
 D21252 OSF-3 
 D38379 Pyrubate kinase M 
 D78645 Adult brain glucose-regulated protein 78 
 ET61037 TI-225 
 L09104 Glucose phosphate isomerase 
 M25244 Pre-B cell P2B/LAMP-1 
 X06086 MEP 
 X51438 Vimentin 
 X52634 tlm oncogene 
 X53333 Triosephosphate isomerase 
 X53584 HSP60 
 X60203 FK506-binding protein 
 X61433 Sodium/potassium ATPase β subunit 
 X77952 (CD1) endoglin 
 X97047 M2-type pyruvate kinase 
 Y00094 ras-related YPT1 protein 
 Y00769 Integrin β subunit 
   
M13945 pim-1 protein kinase 
 X81627 24p3 
 X67644 gly96 
 X15591 CTLA-2-α 
 J04953 Gelsolin 
 U07159 Medium-chain acyl-CoA dehydrogenase 
 X13297 Vascular smooth muscle α-actin 
 L23108 CD36 Ag 
 X04972 Managanese superoxide dismutase 
 X61576 Connexin 43 
 X62940 TSC-22 
 X82067 Thiol-specific antioxidant 
 X83933 Ryanodine receptor type 2 
 Z22866 Skelemin 
 Z31109 T-ZAP 
 U01310 BC1 scRNA 
 X01756 Cytochrome c gene 
(Table continues  
ClusterIdentifierGene Description
 X61432 Calmodulin 
 ET61563 H-2 class I histocompatibility Ag 
 X00246 Set 1 repetitive element for class I MHC 
 X00496 Ia-associated invariant chain (li) 
 X01838 β-microglobulin β2 
 X65553 poly(A) binding protein 
 X65553 poly(A) binding protein 
 X84037 E-selectin ligand-1 
   
D00472 Cofilin and flanks 
 X14425 Profilin 
 L39879 Ferritin L-subunit 
 M12481 Cytoplasmic β-actin 
 M21495 Cytoskeletal γ-actin 
 M76131 Elongation factor 2 
 X05021 Homology to yeast L29 ribosomal protein 
 X51528 Transplantation Ag P198 
 X52803 Cyclophilin 
 X61432 Calmodulin 
 X73729 Ribosomal protein S8 
 X74856 Ribosomal protein L28 
 X74856 Ribosomal protein L28 
 X75313 GB-like 
 X75313 GB-like 
 X81987 TAX responsive element binding protein 107 
 X81987 TAX responsive element binding protein 107 
 Y0025 J1 protein, yeast ribosomal protein L3 homolog 
   
V00719 Liver α-amylase 
 D00613 Matrix Gla protein (MGP) 
 U13705 Plasma glutathione peroxidase 
 X04017 Cysteine-rich glycoprotein SPARC 
 X06086 MEP 
 X02520 Lactate dehydrogenase A4 isoenzyme 
 D21252 OSF-3 
 D38379 Pyrubate kinase M 
 D78645 Adult brain glucose-regulated protein 78 
 ET61037 TI-225 
 L09104 Glucose phosphate isomerase 
 M25244 Pre-B cell P2B/LAMP-1 
 X06086 MEP 
 X51438 Vimentin 
 X52634 tlm oncogene 
 X53333 Triosephosphate isomerase 
 X53584 HSP60 
 X60203 FK506-binding protein 
 X61433 Sodium/potassium ATPase β subunit 
 X77952 (CD1) endoglin 
 X97047 M2-type pyruvate kinase 
 Y00094 ras-related YPT1 protein 
 Y00769 Integrin β subunit 
   
M13945 pim-1 protein kinase 
 X81627 24p3 
 X67644 gly96 
 X15591 CTLA-2-α 
 J04953 Gelsolin 
 U07159 Medium-chain acyl-CoA dehydrogenase 
 X13297 Vascular smooth muscle α-actin 
 L23108 CD36 Ag 
 X04972 Managanese superoxide dismutase 
 X61576 Connexin 43 
 X62940 TSC-22 
 X82067 Thiol-specific antioxidant 
 X83933 Ryanodine receptor type 2 
 Z22866 Skelemin 
 Z31109 T-ZAP 
 U01310 BC1 scRNA 
 X01756 Cytochrome c gene 
(Table continues  
Table 2B.

Continued

ClusterIdentifierGene Description
 X60203 FK506-binding protein 
 X68193 Nucleoside diphosphate kinase B 
 X73359 mAES-1 
 X82067 Thiol-specific antioxidant 
 Z31049 T-ZAP 
 Z50159 Sui1 
   
X02339 T3 δ chain of T3 TCR glycoprotein 
 X06143 T11 protein 
 X13450 B lymphocyte lineage restricted mb-1 
 ET61206 Ig L chain (Fab 17/9) 
 U35323 H2-M 
 X03533 Tyrosine protein kinase p56-tck 
 X03151 Thy-1 Ag 
 M26053 TCR germline β-chain 
 L24372 Clara cell secretory protein 
 Z19543 h2-calponin 
 J00475 IgH chain gene, DJC region 
 ET62206 anti-digoxin Ig H chain 
 ET62538 Glia-derived neurotrophic growth factor β 
 Y09010 Serine/threonine kinase 
 X61385 T-cell specific transcription factor 
 ET62762 Anti-von Willebrand factor Ab NMC-4 κ chain 
 V00802 κ-Ig (C region) 
 ET62056 Ig rearranged κ chain 
 M80423 Castaneus IgK chain 
 M21285 Stearoyl-CoA desaturase 
 M17327 Endogenous murine leukemia virus modified polytropic provirus DNA 
   
J05277 Hexokinase 
 X12507 Protein synthesis initiation factor 4All 
 X15358 junD 
 X56135 Prothymosin α 
 X60289 Ribosomal protein S24 
 X75895 Ribosomal protein L36 
 X80699 L26 
 X97982 poly(C)-binding protein 
 Z83368 RPS3a 
 Z85979 Histone H3.3A 
   
10 X03479 Serum amyloid A 3 
 V00755 IFNβ (type 1) 
 X16133 Serglycin 
 D10024 Calpactin I H chain 
 U39192 Heparin-binding epidermal growth factor-like growth factor 
 L36611 Protein synthesis initiation factor 4A 
 M13445 α-tubulin isotype M-α-1 
 U27340 Sgp1 
 U73744 HSP70 
 W12941 Similar to IP 1-8U 
 X03040 Initiation factor eIF-4A 
 X59379 Amyloid β precursor (protease nexin II) 
 X70847 Adenine nucleotide translocase 
 Y00309 Lactate dehydrogenase-A 
   
11 X59769 type II IL-1R 
 X53798 MIP2 
 X83601 PTX3 
 X54542 IL-6 
 M96827 ob/ob haptoglobin 
 X61800 C/EBP δ 
 X62600 C/EBP β 
 M35590 MIP-1β 
 U36277 I-κ B α chainκ 
 M63244 Amino levulinate synthase 
 X13333 CD14 
 Z30970 Metalloproteinase-3 tissue inhibitor 
 M63245 Amino levulinate synthase 
 U36277 I-κ B α chainκ 
 X13605 Replacement variant histone H3.3 
 X57277 rac1 
 X61940 Growth factor-inducible immediate early gene 
 Z11870 Cellular nucleic acid binding protein 
ClusterIdentifierGene Description
 X60203 FK506-binding protein 
 X68193 Nucleoside diphosphate kinase B 
 X73359 mAES-1 
 X82067 Thiol-specific antioxidant 
 Z31049 T-ZAP 
 Z50159 Sui1 
   
X02339 T3 δ chain of T3 TCR glycoprotein 
 X06143 T11 protein 
 X13450 B lymphocyte lineage restricted mb-1 
 ET61206 Ig L chain (Fab 17/9) 
 U35323 H2-M 
 X03533 Tyrosine protein kinase p56-tck 
 X03151 Thy-1 Ag 
 M26053 TCR germline β-chain 
 L24372 Clara cell secretory protein 
 Z19543 h2-calponin 
 J00475 IgH chain gene, DJC region 
 ET62206 anti-digoxin Ig H chain 
 ET62538 Glia-derived neurotrophic growth factor β 
 Y09010 Serine/threonine kinase 
 X61385 T-cell specific transcription factor 
 ET62762 Anti-von Willebrand factor Ab NMC-4 κ chain 
 V00802 κ-Ig (C region) 
 ET62056 Ig rearranged κ chain 
 M80423 Castaneus IgK chain 
 M21285 Stearoyl-CoA desaturase 
 M17327 Endogenous murine leukemia virus modified polytropic provirus DNA 
   
J05277 Hexokinase 
 X12507 Protein synthesis initiation factor 4All 
 X15358 junD 
 X56135 Prothymosin α 
 X60289 Ribosomal protein S24 
 X75895 Ribosomal protein L36 
 X80699 L26 
 X97982 poly(C)-binding protein 
 Z83368 RPS3a 
 Z85979 Histone H3.3A 
   
10 X03479 Serum amyloid A 3 
 V00755 IFNβ (type 1) 
 X16133 Serglycin 
 D10024 Calpactin I H chain 
 U39192 Heparin-binding epidermal growth factor-like growth factor 
 L36611 Protein synthesis initiation factor 4A 
 M13445 α-tubulin isotype M-α-1 
 U27340 Sgp1 
 U73744 HSP70 
 W12941 Similar to IP 1-8U 
 X03040 Initiation factor eIF-4A 
 X59379 Amyloid β precursor (protease nexin II) 
 X70847 Adenine nucleotide translocase 
 Y00309 Lactate dehydrogenase-A 
   
11 X59769 type II IL-1R 
 X53798 MIP2 
 X83601 PTX3 
 X54542 IL-6 
 M96827 ob/ob haptoglobin 
 X61800 C/EBP δ 
 X62600 C/EBP β 
 M35590 MIP-1β 
 U36277 I-κ B α chainκ 
 M63244 Amino levulinate synthase 
 X13333 CD14 
 Z30970 Metalloproteinase-3 tissue inhibitor 
 M63245 Amino levulinate synthase 
 U36277 I-κ B α chainκ 
 X13605 Replacement variant histone H3.3 
 X57277 rac1 
 X61940 Growth factor-inducible immediate early gene 
 Z11870 Cellular nucleic acid binding protein 

We thank Jim Lederer for the control lymph node data and Walter Zybko and Min Xu for technical support.

1

This work was supported by an American Heart Association Established Investigator Award, an Arthritis Foundation Research Award, and National Institutes of Health Grant RO1 AI44085 (to D.L.P.).

4

Abbreviations used in this paper: SOM, self-organizing map; PM, perfectly matched; MM, mismatched; P, present; M, marginal; A, active; CT, cycle threshold; MIP, macrophage-inflammatory protein; RAG, recombinase-activating gene; IP, IFN-γ-inducible protein; HSP, heat shock protein; Sgp, sulfated glycoprotein.

1
Almond, P. S., A. Matas, K. Gillingham, D. L. Dunn, W. D. Payne, P. Gores, R. Gruessner, J. S. Najarian.
1993
. Risk factors for chronic rejection in renal allograft recipients.
Transplantation
55
:
752
2
Cecka, J..
1999
.
Clinical Transplants 1998
UCLA Immunogenetics Center, Los Angeles.
3
Cecka, J..
2000
. The UNOS scientific renal transplant registry-2000.
Clin. Transpl.
:
1
4
Gulanikar, A. C., A. S. MacDonald, U. Sungurtekin, P. Belitsky.
1992
. The incidence and impact of early rejection episodes on graft outcome in recipients of first cadaver kidney transplants.
Transplantation
53
:
323
5
Hariharan, S., C. P. Johnson, B. A. Bresnahan, S. E. Taranto, M. J. McIntosh, D. Stablein.
2000
. Improved graft survival after renal transplantation in the United States, 1988 to 1996.
N. Engl. J. Med.
342
:
605
6
Lindholm, A., S. Ohlman, D. Albrechtsen, G. Tufveson, H. Persson, N. H. Persson.
1993
. The impact of acute rejection episodes on long-term graft function and outcome in 1347 primary renal transplants treated by 3 cyclosporine regimens.
Transplantation
56
:
307
7
Exner, B. G., X. Que, Y. M. Mueller, M. A. Domenick, M. Neipp, S. T. Ildstad.
1999
. αβ TCR+ T cells play a nonredundant role in the rejection of heart allografts in mice.
Surgery
126
:
121
8
Szot, G. L., P. Zhou, A. H. Sharpe, G. He, O. Kim, K. A. Newell, J. A. Bluestone, J. R. Thistlethwaite, Jr.
2000
. Absence of host B7 expression is sufficient for long-term murine vascularized heart allograft survival.
Transplantation
69
:
904
9
Sun, H., Y. Wakizaka, A. S. Rao, F. Pan, J. Madariaga, I. Y. Park, S. Celli, J. J. Fung, T. E. Starzl, L. A. Valdivia.
1996
. Use of MHC class I or II “knock out” mice to delineate the role of these molecules in acceptance/rejection of xenografts.
Transplant. Proc.
28
:
732
10
Shimizu, K., U. Schonbeck, F. Mach, P. Libby, R. N. Mitchell.
2000
. Host CD40 ligand deficiency induces long-term allograft survival and donor-specific tolerance in mouse cardiac transplantation but does not prevent graft arteriosclerosis.
J. Immunol.
165
:
3506
11
Qian, S., F. Fu, Y. Li, L. Lu, A. S. Rao, T. E. Starzl, A. W. Thomson, J. J. Fung.
1996
. Impact of donor MHC class I or class II antigen deficiency on first- and second-set rejection of mouse heart or liver allografts.
Immunology
88
:
124
12
Mandelbrot, D. A., Y. Furukawa, A. J. McAdam, S. I. Alexander, P. Libby, R. N. Mitchell, A. H. Sharpe.
1999
. Expression of B7 molecules in recipient, not donor, mice determines the survival of cardiac allografts.
J. Immunol.
163
:
3753
13
He, H., J. R. Stone, D. L. Perkins.
2002
. Analysis of robust innate immune response following transplantation in the absence of adaptive immunity.
Transplantation
73
:
853
14
Spellman, P. T., G. Sherlock, M. Q. Zhang, V. R. Iyer, K. Anders, M. B. Eisen, P. O. Brown, D. Botstein, B. Futcher.
1998
. Comprehensive identification of cell cycle-regulated genes of the yeast Saccharomyces cerevisiae by microarray hybridization.
Mol. Biol. Cell
9
:
3273
15
Eisen, M. B., P. T. Spellman, P. O. Brown, D. Botstein.
1998
. Cluster analysis and display of genome-wide expression patterns.
Proc. Natl. Acad. Sci. USA
95
:
14863
16
Ren, B., F. Robert, J. J. Wyrick, O. Aparicio, E. G. Jennings, I. Simon, J. Zeitlinger, J. Schreiber, N. Hannett, E. Kanin, et al
2000
. Genome-wide location and function of DNA binding proteins.
Science
290
:
2306
17
Ideker, T., V. Thorsson, J. A. Ranish, R. Christmas, J. Buhler, J. K. Eng, R. Bumgarner, D. R. Goodlett, R. Aebersold, L. Hood.
2001
. Integrated genomic and proteomic analyses of a systematically perturbed metabolic network.
Science
292
:
929
18
Tamayo, P., D. Slonim, J. Mesirov, Q. Zhu, S. Kitareewan, E. Dmitrovsky, E. S. Lander, T. R. Golub.
1999
. Interpreting patterns of gene expression with self-organizing maps: methods and application to hematopoietic differentiation.
Proc. Natl. Acad. Sci. USA
96
:
2907
19
Gibson, U. E., C. A. Heid, P. M. Williams.
1996
. A novel method for real time quantitative RT-PCR.
Genome Res.
6
:
995
20
Heid, C. A., J. Stevens, K. J. Livak, P. M. Williams.
1996
. Real time quantitative PCR.
Genome Res.
6
:
986
21
Reichert, F., S. Rotshenker.
1999
. Galectin-3/MAC-2 in experimental allergic encephalomyelitis.
Exp. Neurol.
160
:
508
22
Schick, B. P., J. F. Gradowski, J. D. San Antonio.
2001
. Synthesis, secretion, and subcellular localization of serglycin proteoglycan in human endothelial cells.
Blood
97
:
449
23
Kulseth, M. A., S. O. Kolset, T. Ranheim.
1999
. Stimulation of serglycin and CD44 mRNA expression in endothelial cells exposed to TNF-α and IL-1α.
Biochim. Biophys. Acta
1428
:
225
24
Boehm, U., T. Klamp, M. Groot, J. C. Howard.
1997
. Cellular responses to interferon-γ.
Annu. Rev. Immunol.
15
:
749
25
Saiura, A., C. Mataki, T. Murakami, M. Umetani, Y. Wada, T. Kohro, H. Aburatani, Y. Harihara, T. Hamakubo, T. Yamaguchi, et al
2001
. A comparison of gene expression in murine cardiac allografts and isografts by means DNA microarray analysis.
Transplantation
72
:
320
26
Ring, G. H., S. Saleem, Z. Dai, A. T. Hassan, B. T. Konieczny, F. K. Baddoura, F. G. Lakkis.
1999
. Interferon-γ is necessary for initiating the acute rejection of major histocompatibility complex class II-disparate skin allografts.
Transplantation
67
:
1362
27
Zahedi, K., M. F. Seldin, M. Rits, R. A. Ezekowitz, A. S. Whitehead.
1991
. Mouse IL-1 receptor antagonist protein: molecular characterization, gene mapping, and expression of mRNA in vitro and in vivo.
J. Immunol.
146
:
4228
28
Dinarello, C. A..
1984
. Interleukin 1 as mediator of the acute-phase response.
Surv. Immunol. Res.
3
:
29
29
Hooper, D. C., C. J. Steer, C. A. Dinarello, A. C. Peacock.
1981
. Haptoglobin and albumin synthesis in isolated rat hepatocytes: response to potential mediators of the acute-phase reaction.
Biochim. Biophys. Acta
653
:
118
30
Heinrich, P. C., J. V. Castell, T. Andus.
1990
. Interleukin-6 and the acute phase response.
Biochem. J.
265
:
621
31
Chow, J. C., D. W. Young, D. T. Golenbock, W. J. Christ, F. Gusovsky.
1999
. Toll-like receptor-4 mediates lipopolysaccharide-induced signal transduction.
J. Biol. Chem.
274
:
10689
32
Chandrasekar, B., J. B. Smith, G. L. Freeman.
2001
. Ischemia-reperfusion of rat myocardium activates nuclear factor-κB and induces neutrophil infiltration via lipopolysaccharide-induced CXC chemokine.
Circulation
103
:
2296
33
Adams, D. H., S. Hubscher, J. Fear, J. Johnston, S. Shaw, S. Afford.
1996
. Hepatic expression of macrophage inflammatory protein-1α and macrophage inflammatory protein-1β after liver transplantation.
Transplantation
61
:
817
34
Takano, S., H. Fukuyama, M. Fukumoto, K. Hirashimizu, T. Higuchi, J. Takenawa, H. Nakayama, J. Kimura, J. Fujita.
1995
. Induction of CL100 protein tyrosine phosphatase following transient forebrain ischemia in the rat brain.
J. Cereb. Blood Flow Metab.
15
:
33
35
Charles, C. H., A. S. Abler, L. F. Lau.
1992
. cDNA sequence of a growth factor-inducible immediate early gene and characterization of its encoded protein.
Oncogene
7
:
187
36
Csizmadia, V., W. Gao, S. A. Hancock, J. B. Rottman, Z. Wu, L. A. Turka, U. Siebenlist, W. W. Hancock.
2001
. Differential NF-κB and IκB gene expression during development of cardiac allograft rejection versus CD154 monoclonal antibody-induced tolerance.
Transplantation
71
:
835
37
Brandt, M., G. Derner, K. Boeke, M. L. Phillips, G. Steinhoff, A. Haverich.
1997
. Anti-rejection prophylaxis by blocking selectin dependent cell adhesion after rat allogeneic and xenogeneic lung transplantation.
Eur. J. Cardiothorac. Surg.
12
:
781
38
Rasmussen, A. M., H. K. Blomhoff, T. Stokke, V. Horejsi, E. B. Smeland.
1994
. Cross-linking of CD53 promotes activation of resting human B lymphocytes.
J. Immunol.
153
:
4997
39
Arnold, D., J. Driscoll, M. Androlewicz, E. Hughes, P. Cresswell, T. Spies.
1992
. Proteasome subunits encoded in the MHC are not generally required for the processing of peptides bound by MHC class I molecules.
Nature
360
:
171
40
Luster, A. D., R. L. Weinshank, R. Feinman, J. V. Ravetch.
1988
. Molecular and biochemical characterization of a novel γ-interferon-inducible protein.
J. Biol. Chem.
263
:
12036
41
Sasaki, T., C. Brakebusch, J. Engel, R. Timpl.
1998
. Mac-2 binding protein is a cell-adhesive protein of the extracellular matrix which self-assembles into ring-like structures and binds β1 integrins, collagens and fibronectin.
EMBO J.
17
:
1606
42
Koths, K., E. Taylor, R. Halenbeck, C. Casipit, A. Wang.
1993
. Cloning and characterization of a human Mac-2-binding protein, a new member of the superfamily defined by the macrophage scavenger receptor cysteine-rich domain.
J. Biol. Chem.
268
:
14245
43
Chuang, E., M. L. Alegre, C. S. Duckett, P. J. Noel, M. G. Vander Heiden, C. B. Thompson.
1997
. Interaction of CTLA-4 with the clathrin-associated protein AP50 results in ligand-independent endocytosis that limits cell surface expression.
J. Immunol.
159
:
144
44
Zhang, Y., J. P. Allison.
1997
. Interaction of CTLA-4 with AP50, a clathrin-coated pit adaptor protein.
Proc. Natl. Acad. Sci. USA
94
:
9273
45
Tenner, A. J..
1998
. C1q receptors: regulating specific functions of phagocytic cells.
Immunobiology
199
:
250
46
Scullion, M., G. Balint, K. Whaley.
1979
. Evaluation of the C1q solid-phase binding assay for immune complexes: a clinical and laboratory study.
J. Clin. Lab. Immunol.
2
:
15
47
Thompson, P. N., E. Cho, F. A. Blumenstock, D. M. Shah, T. M. Saba.
1992
. Rebound elevation of fibronectin after tissue injury and ischemia: role of fibronectin synthesis.
Am. J. Physiol.
263
:
G437
48
Ishiwata, T., T. Aida, M. Yokoyama, G. Asano.
1994
. Fibronectin biosynthesis in endothelial regeneration after intimal injury.
Exp. Mol. Pathol.
60
:
1
49
Hirayoshi, K., H. Kudo, H. Takechi, A. Nakai, A. Iwamatsu, K. M. Yamada, K. Nagata.
1991
. HSP47: a tissue-specific, transformation-sensitive, collagen-binding heat shock protein of chicken embryo fibroblasts.
Mol. Cell. Biol.
11
:
4036
50
Nakai, A., M. Satoh, K. Hirayoshi, K. Nagata.
1992
. Involvement of the stress protein HSP47 in procollagen processing in the endoplasmic reticulum.
J. Cell Biol.
117
:
903
51
Bradshaw, A. D., E. H. Sage.
2001
. SPARC, a matricellular protein that functions in cellular differentiation and tissue response to injury.
J. Clin. Invest.
107
:
1049
52
Whittaker, P..
1998
. Collagen organization in wound healing after myocardial injury.
Basic Res. Cardiol.
93
:
23
53
Caughey, G. H., E. H. Zerweck, P. Vanderslice.
1991
. Structure, chromosomal assignment, and deduced amino acid sequence of a human gene for mast cell chymase.
J. Biol. Chem.
266
:
12956
54
Poirier, F., E. J. Robertson.
1993
. Normal development of mice carrying a null mutation in the gene encoding the L14 S-type lectin.
Development
119
:
1229
55
Zalewska, T., B. Zablocka, K. Domanska-Janik.
1996
. Changes of Ca2+/calmodulin-dependent protein kinase-II after transient ischemia in gerbil hippocampus.
Acta. Neurobiol. Exp.
56
:
41
56
Mitchell, T. I., J. J. Jeffrey, R. D. Palmiter, C. E. Brinckerhoff.
1993
. The acute phase reactant serum amyloid A (SAA3) is a novel substrate for degradation by the metalloproteinases collagenase and stromelysin.
Biochim. Biophys. Acta
1156
:
245
57
Slater, A. D., J. B. Klein, G. Sonnenfeld, L. L. Ogden, II, L. A. Gray, Jr.
1992
. The effects of interferon-α/β in a model of rat heart transplantation.
J. Heart Lung Transplant
11
:
975
58
Cao, Q. P., W. R. Crain.
1995
. Expression of SGP-1 mRNA in preimplantation mouse embryos.
Dev. Genet.
17
:
263
59
Morales, C. R., M. el-Alfy, Q. Zhao, S. Igdoura.
1995
. Molecular role of sulfated glycoprotein-1 (SGP-1/prosaposin) in Sertoli cells.
Histol. Histopathol.
10
:
1023
60
Mehta, N. K., M. Carroll, D. E. Sykes, Z. Tan, J. Bergsland, J. Canty, Jr, J. N. Bhayana, E. L. Hoover, T. A. Salerno.
1997
. Heat shock protein 70 expression in native and heterotopically transplanted rat hearts.
J. Surg. Res.
70
:
151
61
Jayakumar, J., K. Suzuki, M. Khan, R. T. Smolenski, A. Farrell, N. Latif, O. Raisky, H. Abunasra, I. A. Sammut, B. Murtuza, et al
2000
. Gene therapy for myocardial protection: transfection of donor hearts with heat shock protein 70 gene protects cardiac function against ischemia-reperfusion injury.
Circulation
102
:
SIII302