Differential gene expression in human abdominal aortic aneurysm and aortic occlusive disease

1Peripheral arterial diseases include a collection of occlusive and aneurysmal diseases affecting arteries outside the heart estimated to affect approximately 5% of adults. The mechanisms responsible for ischemic complications of peripheral artery atherosclerosis are believed to be similar to those implicated in coronary atherosclerosis. In keeping with the perceived similar pathogenesis of coronary and peripheral artery disease most of the drug therapies used in patients with these conditions are the same. The mechanisms involved in abdominal aortic aneurysm (AAA) development and their relationship with atherosclerosis are however controversial. Examination of human AAA biopsies has consistently identified concurrent intimal atherosclerosis and most AAAs have been considered to result from atherosclerosis. Some of the risk factors for atherosclerosis and AAA are however distinct, for example diabetes an important positive risk factor for occlusive artery disease is a negative risk factor for AAA. Dyslipidemia is also believed to be a less important risk factor for AAA than atherosclerotic occlusive disease. There are also a number of similarities between atherosclerosis and AAA, including common risk factors of smoking, hypertension, and male gender. The role of atherosclerosis in AAA therefore is currently controversial.

2Examination of mechanisms involved in AAA has mainly relied on use of animal, particularly rodent, models. The availability of whole genome expression profiling has allowed insight into a range of human pathologies but has been relatively sparingly applied to AAA and aortic atherosclerosis. Two previous studies have examined the whole genome expression profile of human AAA specimen although sample sizes for these studies were relatively small totaling 7 and 10 AAA patients, and no comparison was performed with atherosclerosis. In the current study we examined the whole genome expression profiles of a relatively large number of human aortic biopsies from subjects undergoing surgery to repair AAAs and to treat chronic lower limb ischemic due to aortic atherosclerosis (aortic occlusive disease, AOD). Aortic wall samples obtained from organ donors were used as controls. We aimed to compare the aortic gene expression profile of patients with AAA and AOD to those of older subjects with normal aortic histology.

3The gene expression was initially assessed in the aortic biopsies of 20 patients with small AAAs (mean AAA diameter 53.4±2.3 mm), 29 patients with large AAAs (mean AAA diameter 68.4±14.3 mm), 9 patients with AOD (mean aortic diameter 19.6±2.6 mm), and 10 control organ donors using microarrays (Table 1). AAA patients were less likely to be female compared to controls (small AAA 0%, large AAA 7%, controls 40%; P < 0.05), while AOD patients were younger compared to controls (mean age 61.6±9.3 vs. 68.4±4.5 years, P < 0.05). The validity of microarray findings was further assessed in biopsies obtained from 6 patients with small AAAs, 9 patients with large AAAs, 8 AOD patients, and 6 controls (Table 1). AAA and AOD cases were older compared to controls (mean age small AAAs 69.7±3.8, large AAAs 71.5±7.8, AOD 67.8±4.2, controls 42.7±12.5 years; P < 0.05). Mean maximum AAA diameter was 48.5±5.7 and 65.9±7.3 mm in patients with small and large AAAs, respectively.

4A total of 39,157 transcripts in small AAAs, 39,389 transcripts in large AAAs, 37,904 transcripts in AOD, and 36,725 transcripts in control samples, representing > 80% of the Illumina HumanHT-12 v4 Expression BeadChip reference list, were expressed above background and subjected to differential gene expression analysis. A total of 840 individual genes were differentially expressed (≥ 2.0-fold, adjusted P < 0.05) in small AAAs compared to controls (122 upregulated and 718 downregulated genes; Figure 1a and 1b, respectively). A total of 1,014 individual genes were differentially expressed (≥ 2.0 fold, adjusted P < 0.05) in large AAAs compared to controls (215 upregulated and 799 downregulated genes; Figure 1a and 1b, respectively). 1,765 individual genes were differentially expressed (≥ 2.0-fold, adjusted P < 0.05) in AOD compare to controls (1,677 upregulated and 88 downregulated genes; Figure 1a and 1b, respectively). Full lists of differentially expressed genes are given in Supplemental Table I, Supplemental Table II, and Supplemental Table III for small AAA, large AAA, and AOD, respectively. A preponderance of genes upregulated in small AAAs (100/122; 82%) were also upregulated in large AAAs. Approximately 47% (100/215; Figure 1a) of the genes upregulated in large AAAs were also upregulated in small AAAs. In contrast only one gene (CXCL12) differentially expressed in AOD samples (1/1,765) was also differentially expressed in small (1/840) or large (1/1,014) AAAs (Figure 1a and 1b).

5We performed an enrichment analysis of differentially expressed genes with particular attention to Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways to assess the biological contexts of these genes. KEGG annotations included 93% (114/122) of upregulated genes and 92% (661/718) of downregulated genes in small AAAs; and 91% (196/215) of upregulated genes and 95% (756/799) of downregulated genes in large AAA samples. The KEGG analysis revealed marked upregulation of genes related to immune responses in both small and large AAA. These pathways included cytokine-cytokine receptor interaction (hsa04060; P = 1.15*10−14 for small AAA, and P = 2.52*10−17 for large AAA), chemokine signaling pathway (hsa04062; P = 4.50*10−11 for small AAA, and P = 3.53*10−12 for large AAA), and T-cell receptor signaling pathway (hsa04660; P = 4.07*10−05 for small AAA, and P = 2.40*10−06 for large AAA) as shown in Table 2. Metabolic pathways (hsa01100) and oxidative phosphorylation (hsa00190) were found among the top 10 most significant KEGG categories enriched by downregulated genes with P = 2.30*10−22 (metabolic pathways) and P = 4.25*10−20 (oxidative phosphorylation) for small AAA; and P = 1.05*10−24 (metabolic pathways) and P = 8.60*10−20 (oxidative phosphorylation) for large AAA (Table 2).

6Considering the marked enrichment of immune response categories in small and large AAA, we further examined a number of genes related to the immune response (Table 3). A majority of genes upregulated in small AAA were also upregulated in large AAA. Examples include genes coding for: pro-inflammatory cytokines IL1B (5.08-fold, P = 0.002 for small AAA; 3.85-fold, P = 0.001 for large AAA), IL6 (6.28-fold, P = 0.004 for small AAA; 6.48-fold, P = 0.002 for large AAA), and IL8 (4.49-fold, P = 0.005 for small AAA; 3.65-fold, P = 0.004 for large AAA); chemokines CXCL2 (3.35-fold, P = 0.004 for small AAA; 2.73-fold, P = 0.005 for large AAA), CXCL13 (2.21-fold, P = 0.013 for small AAA; 3.09-fold, P = 0.002 for large AAA), CCL3L3 (3.14-fold, P = 0.039 for small AAA; 2.59-fold, P = 0.041 for large AAA), and CCL4L2 (2.48-fold, P = 0.021 for small AAA; 2.12-fold, P = 0.023 for large AAA); early T-cell and B-cell activation cluster of differentiation (CD) antigens CD69 (2.67-fold, P = 0.039 for small AAA; 3.45-fold, P = 0.004 for large AAA) and CD19 (2.79-fold, P = 0.026 for small AAA; 4.45-fold, P = 0.002 for large AAA), respectively; and the natural killer (NK) cells associated protein NKTR (2.37-fold, P = 0.002 for small AAA; 2.66-fold, P = 0.001 for large AAA). An important negative regulator of T-cells responses, the CTLA4 gene, was exclusively upregulated in small AAA (2.01-fold, P = 0.018). On the other hand, there was increased expression of genes related to T-cell and B-cell responses exclusively in large AAA, including alpha chain of the T-cell co-receptor CD8A (2.24-fold, P = 0.012); the CD79A/CD79B membrane-bound dimer immunoglobulin in B-cells (2.76-fold, P = 0.005/3.43-fold, P = 0.004); the receptors for Fc fragment of IgG in B-cells FCRL2 (2.53-fold, P = 0.002), FCRL5 (2.01-fold, P = 0.001), and FCRLA (3.15-fold, P = 0.009). Three genes (CTLA4, NKTR, and CD8A) were selected for further validation with quantitative real-time PCR.

792% (1,550/1,677) of upregulated genes and 90% (79/88) of downregulated genes in AOD samples were annotated in the KEGG pathways. The pathways enriched in AOD samples were distinct to those identified in AAA samples. In AOD, the most significant KEGG categories representing upregulated genes were metabolic pathways (hsa01100; P = 1.49*10−43) and oxidative phosphorylation (hsa00190; P = 5.68*10−35). Downregulated genes primarily fell into the immune response categories such as cytokine-cytokine receptor interaction (hsa04060; P = 5.69−07) and chemokine signaling pathways (hsa04062; P = 8.64*10−06) as seen in Table 4.

8Inspection of genes within KEGG metabolic categories revealed upregulation of genes related to mitochondrial biogenesis that included many members of all five mitochondrial respiratory and ATP production complexes (Table 5). We also demonstrated the upregulation of numerous genes involved in glycolysis/gluconeogenesis, fatty acid metabolism, and propanoate metabolism (Table 5). Several complement system genes were overexpressed such as C1S (2.08-fold, P = 0.044), CD46 (2.00-fold, P = 0.004), CFH (2.81-fold, P = 0.006), CFI (2.02-fold, P = 0.004), C1QBP (2.10-fold, P = 0.003), C1RL (2.08-fold, P = 0.003), and C7 (3.46-fold, P = 0.037). One important mitochondrial regulatory gene called mitochondrial uncoupling protein 2 (UCP2) was downregulated in AOD samples (−3.73-fold, P = 0.014), (Table 5). The UCP2 gene was selected for further validation with quantitative real-time PCR.

9Expression levels of a number of genes were validated in biopsies from the six patients with small AAAs, nine patients with large AAAs, eight patients with AOD and six controls (Table 1). The relative expression of CTLA4 was significantly increased in small AAA but not large AAA biopsies compared to normal controls (Table 6). Median relative expression of CTLA4 was also markedly increased in small AAA versus large AAA biopsies (202.161 vs. 65.239, P = 0.041). The relative expression of CD8A was significantly increased in large AAA but not small AAA biopsies compared to normal controls (Table 6), with the median relative expression of CD8A in large AAA ~2-fold higher and approaching significance compared to small AAA (109.489 vs 61.004; P = 0.065). Correspondingly, immunohistochemical examination of AAA tissue demonstrated marked presence of CD8-positive T lymphocytes within the adventitia and outer media of a large (82mm) AAA (Figure 2A) with limited immunostaining for CTLA4 (Figure 2C). In contrast there was relatively sparse presence of CD8-positive T cells within tissue from a small (55mm) AAA (Figure 2B) but marked CTLA4 staining around tertiary lymphoid tissue at the media-adventia border (Figure 2D).

10Median relative expression of NKTR was significantly greater in AAA biopsies compared to normal controls with approximately 2- (P = 0.041) and 3- (P = 0.015) fold increase for small and large AAAs, respectively (Table 6). Median relative expression of NKTR was also markedly increased in large AAA compared to small AAA biopsies (1.186 vs. 0.651, P = 0.040).

11The relative expression of UCP2 was markedly decreased in AOD samples compared to normal controls (Table 6). In subjects with AOD the UCP2 mRNA expression was more than 3-fold lower compared to the control aortic biopsies (0.739 vs. 2.306, P = 0.017).

12Human AAA biopsies have been consistently shown to have intimal atheroma and historically AAAs have been considered to result from atherosclerosis. However the mechanisms involved in AAA development and its relationship with atherosclerosis remain controversial. We determined the global gene expression profiles within biopsies of AAA and AOD using microarrays. The main finding from this study was that the genes differentially expressed in AAA and AOD were almost completely distinct, supporting the concept of separate developmental mechanisms underlying the pathogenesis of each disease. With this in mind, global gene expression profiles associated with AAA and AOD will be further discussed separately.

13A number of human and animal studies previously suggested a major role of immune responses in AAA pathogenesis and the array findings for cytokines/chemokines are in line with previous quantitative data. Our results substantiate those findings, and more importantly, suggest also that small AAAs (≤ 55 mm) and large AAAs (> 55 mm) may have some distinct immunological characteristics. The most notable finding related to immunity was that the key T-cell regulatory gene called cytotoxic T-lymphocyte-associated protein 4 (CTLA4 or CD152) was exclusively found to be upregulated in small AAAs. Moreover, our data suggests excessive infiltration of CD8+ T-cells in biopsies from large AAAs evidenced by the exclusive overexpression of the gene coding for the CD8A molecule and increased number of CD8+ T-cells found in these specimens. The upregulation of CTLA4 and CD8A in small AAA and large AAA, respectively, was confirmed by qRT-PCR and led us to hypothesize that downregulation of CTLA4 may represent a key mechanism in AAA progression which promotes excessive T-cell-driven immune responses in large AAAs. This hypothesis is supported by our finding of marked downregulation of the CTLA4 mRNA and the protein product in large AAAs compared to small AAAs as determined by qRT-PCR and immunohistochemistry, respectively. CTLA4 is a cell surface molecule that can down-modulate and terminate T-cell adaptive immunity. The CD8A molecule is a constituent of the CD8 antigen found on most cytotoxic T lymphocytes and acts as a co-receptor with the T-cell receptor (TCR). Previous immunohistochemical studies of human AAA have provided evidence that CD8+ T-cells infiltrating the aortic wall express cytotoxic mediators such as perforins, which generate membrane damage. Other studies have shown that immunosuppressive agents directed exclusively at T-cells are capable of reducing the expansion of experimental AAA and reduce vascular inflammation and aortic wall content of several inflammatory cell types including cytotoxic T lymphocytes in human AAA. Taken together, the manipulation of CTLA4 levels may represent a novel therapeutic strategy for AAAs but this need examining in other contexts e.g. animal models. The use of CTLA4-based biologic agents such as Abatacept (Orencia), a novel fusion protein of the Fc region of the immunoglobulin IgG1 and CTLA4 designed to down-regulate T-cells activity, is already under investigation as a potential therapy for other immunopathologies such as active rheumatoid arthritis. Findings from the present study additionally demonstrate the upregulation of genes expressed by NK cells within large AAAs, such as natural killer cell group 7 sequence (NKG7) and natural killer-tumor recognition sequence (NKTR) genes; both involved in the TCR-independent innate immune responses. The latter was assessed and validated to be upregulated in large AAAs by qRT-PCR. It is possible that similar mechanisms involving the upregulation of innate immunity occur at some point between the late and end-stage of AAA. Findings suggest marked upregulation of the NKTR gene in large AAA compared to small AAA biopsies which is in accord with this hypothesis.

14Genes differentially expressed in AOD versus control samples were particularly related to the metabolic and oxidative phosphorylation pathways which accounted for ~12% of all upregulated genes. In particular, we determined the excessive transcription of genes involved in mitochondrial biogenesis affecting numerous members of all four mitochondrial respiratory complexes as well as genes responsible for ATP synthesis in the mitochondrial complex V. Other pathways enriched included those related to Parkinson's disease. Genes involved in this category belonging to the mitochondrial complex I were upregulated in AOD, contrasting the actual disruption of the mitochondrial complex I in Parkinson's disease itself. We also demonstrated the upregulation of a range of genes involved in glycolysis, gluconeogenesis, fatty acid metabolism, and propanoate metabolism. Since the number of overlapping genes in these metabolic pathways is considerable, it would be reasonable to suggest that overexpression of these genes may simply represent a metabolic maladaptation to the excessive mitochondrial biogenesis, as members of these bioenergetic pathways are tightly coordinated transcriptionally. Upregulation of genes related to glucose and fatty acids metabolism is highly relevant to AOD. The role of an impaired glucose tolerance and fatty acids metabolism in atherosclerosis is long-documented. Current evidence suggests that the ataxia telangiectasia mutated (ATM)/mammalian target of rapamycin (mTOR) signaling axis may play an important role in suppression of macrophage transformation into foam cells, thus limiting the formation of atherosclerotic plaques. Interestingly, the old malaria drug Chloroquine (Aralen), a lysosomotropic agent that activates ATM signaling, was found to be capable of decreasing atherosclerosis and improving metabolic phenotype in mice. The hypothesis that Chloroquine may also reduce atherosclerosis in humans is currently under investigation. The mechanisms responsible for upregulation of genes involved in mitochondrial biogenesis are not evident in the current study; however, we did identify downregulation of the UCP2 gene in AOD samples. The UCP2 protein belongs to a group of mitochondrial transporters that create proton leaks across the inner mitochondrial membrane, thus uncoupling respiration from ATP synthesis. Previous studies provide evidence that uncoupling by UCP2 downregulates the production of reactive oxygen species (ROS) in endothelial and smooth muscle cells, whereas the UCP2 deficiency is associated with enhanced ROS production in the endothelium of the aorta. Excessive ROS production is known to contribute to the oxidative damage of mitochondria that needs to be repaired to maintain intact cellular mitochondrial content. This process is accomplished via mitophagy (autophagy), a selective elimination of malfunctioning mitochondria, which must be balanced by mitochondrial biogenesis to meet tissue energy requirements. It is possible that UCP2 deficiency may, in fact, indirectly lead to increased mitochondrial biogenesis, explaining the upregulation of genes related to the metabolic and oxidative phosphorylation pathways in AOD. Pathways related to inflammation were less evident amongst differentially expressed genes in AOD samples. We did demonstrate the upregulation of a number of complement regulatory and effectors genes, such as complement regulatory protein CD46, complement factor H (CFH), and complement factor I (CFI) suggesting potential involvement of the classical complement activation pathway in AOD. The overexpression of the complement component 7 (C7) gene, a constituent of the membrane attack complex (MAC), may indicate the culmination of the alternative complement activation pathway in AOD. Findings are in line with previous studies that have suggested an important role for the complement system in mediating tissue injury and atherosclerosis after oxidative stress.

15The current study has a number of limitations. The available sample size was relatively small especially for AOD although larger than any previously published microarrays using human aortic biopsies. In view of this limitation we sought to validate important findings in further groups of AAA and AOD biopsies. The assessment of independent samples helps to minimize the possibility that selection biases adversely affect the generalizability of the findings. Furthermore, we were able to obtain the aortic biopsies from advanced stage disease including AAAs measuring at least 50mm and patients needing aortic bypass for peripheral ischemia. Our findings therefore are difficult to relate to early stage vascular disease. The majority of available AAA and AOD biopsies were obtained from men limiting the relevance of our findings to AAA and AOD in women. Finally, the AOD patients were either younger (discovery group) or older (validation group) than their respective controls. In both particular comparisons of cases with controls, however, an important gene regulating mitochondrial biogenesis was found to be downregulated in AOD, suggesting that our findings are descriptive of the disease rather than age differences between cases and controls.

16In conclusion, the current results demonstrate distinct gene expression profiles of AAA and AOD. The AAA transcriptome highlights an important mechanism controlling immune responses as opposed to the mitochondrial biogenesis pathways being predominantly associated with AOD pathology.

17A detailed description of all methods is presented in the online data supplement.

18Full thickness abdominal aortic specimens were obtained from 49 patients undergoing open surgery to treat AAA, 9 patients with AOD undergoing surgery to treat chronic lower limb ischemia, and 10 control individuals (discovery group, Table 1). Maximum aortic diameter was 50-55mm (defined as small AAA) and 56-120mm (defined as large AAA) in 20 and 29 AAA subjects, respectively. Control full thickness abdominal aortic samples were obtained during kidney transplant (all from heart-beating, brain-dead donors). The study was approved by the ethics committees of the Townsville Hospital, James Cook University, Royal Brisbane and Women's Hospital, Medical Faculty at the Technical University Dresden, and Leiden University Medical Center and the protocol conformed to ethical guidelines of Declaration of Helsinki. All patients gave written informed consent.

19Microarrays were performed using the Illumina HumanHT-12v4 Expression BeadChip® platform as previously described. The microarray data can be obtained at the Gene Expression Omnibus (GEO) database repository (GSE57691; http://www.ncbi.nlm.nih.gov/geo/info/linking.html).

20The raw data matrix extracted from BeadStudio was uploaded into the GeneSpring GX version 11.5.1 (Agilent Technologies Pty Ltd) software for downstream analysis as previously described. We sought to identify genes with a 2-fold differential expression within the aortas of patients with small AAA, large AAA or AOD compared to donor abdominal aortas based on corrected p value < 0.05 by the Benjamini Hochberg false discovery rate (FDR) method and determined by non-parametric Mann–Whitney U test. Genes showing ≥ 2-fold differences in expression between groups were considered to be significantly differentially expressed.

21Using total RNA obtained from a further 29 patients with small AAAs (n = 6), large AAAs (n = 9) and AOD (n = 8), and another 6 organ donor biopsies (validation group, Table 1) we assessed the validity of microarrays findings using qRT-PCR as described previously. Three genes associated with AAA (CTLA4, cytotoxic T-lymphocyte-associated protein 4; NKTR, natural killer cell triggering receptor; CD8A, CD8A molecule), and the uncoupling protein 2 gene (UCP2) associated with AOD were chosen for further assessment. Mann–Whitney U test was performed to identify differences in expression levels of selected genes between small AAA, or large AAA, or AOD and control biopsies. Data are reported as median and interquartile range. All computations were performed using the SPSS statistical package v.17.0.2. Statistical significance was defined at the conventional 5% level.

22CD8 and CTLA4 proteins were immunohistochemically detected on 5-μm thick paraffin sections of AAA tissue taken from the maximally dilated aneurysm body as previously described.