Proteomic analysis of descending thoracic aorta identifies unique and universal signatures of aneurysm and dissection

Thoracic aortic dissection has continued to be one of the highest causes of morbidity and mortality among the cardiovascular diseases. However, few reliable risk factors are available, with the exception of aortic size. Recent studies, however, have indicated that many descending thoracic aortas will dissect without any evidence of prior aneurysm growth and, of even greater concern, do so at less than the recommended thresholds currently set for surgical intervention.^, These clinical findings beg the question of whether dissections with and without an aneurysm represent two different types of diseases. We found at least some differences in the pathways that govern these two diseases, as indicated by the differential protein expression analysis between the TAAs and TBADs. No proteins, however, were significantly altered comparing TADA to TBAD or TAA. This finding was consistent with the CAM results, which showed that within most subtypes TADA was clustered between TBAD and TAA, with the exception of perhaps subtype S1, for which TADA clustered more closely with TAA than with TBAD. The inability to distinguish TADA from TBAD and TAA might have been because TADAs share a signature with both TAA and TBAD or because TADAs represent a mixture of signatures, such that each aorta behaves more like TAA or TBAD. Thus, given the low sample size, the differential expression and CAM were unable to identify significant distinguishing features. We favored the latter theory because the TADAs tended to cluster with either TAA or TBAD using principle component analysis. In addition, we lacked information regarding whether the TADA samples represented atherosclerotic aneurysms that had dissected or dissected tissue that had degenerated into an aneurysm or pseudoaneurysm.

The results of our analyses have shown that the normal tissue was molecularly different from the pathologic tissue. However, we also sought to determine whether any similarities were present between the different types of pathologic aortas. We found 71 proteins that were similarly differentially expressed between each pathology and normal tissue, representing 66%, 17%, and 21% of the differentially expressed proteins from TADA, TAA, and TBAD, respectively. Clearly, some common features exist among aneurysms at risk of dissection, aneurysms that dissect, and aortas without aneurysms that dissect. A network analysis of these 71 proteins generated one main cluster. THBS1 was at the center of this cluster, with a degree of 15. THBS1 was one of the few upregulated proteins in disease tissue compared with the normal tissue, and prior studies have demonstrated a pathologic role for increased THBS1 expression in both aneurysms and dissections. In the Fbln4^SMKO aneurysm mouse model, THBS1 is upregulated and causes disruption of the actin cytoskeleton and ELN-contractile apparatus. Deletion of this gene prevents aneurysm formation. In humans, THBS1 is upregulated in the aortas and plasma of patients with acute dissection and might play a role in smooth muscle cell apoptosis and macrophage activation. Thus, THBS1 might be a useful diagnostic marker for all types of aortic degeneration. In addition, THBS1 enzyme-linked immunosorbent assay kits are commercially available and have been used by others to easily detect plasma levels in patients, indicating the practicality of using this protein as a biomarker. Further work in animal models is needed to determine whether THBS1 can be used to predict aneurysms at risk of dissection.

The proteome we analyzed represents a homogenate of multiple underlying cell types. Unsupervised deconvolution by CAM provided provisional insight into how the proteome is organized in terms of the molecular or cellular subtypes. When the proteins in our CAM were screened for those with the gene ontology annotation “contractile,” and their expression was compared across the six subtypes, subtypes S2 and S4 demonstrated a clear enrichment for contractile protein expression (Supplementary Fig). Nevertheless, subtype S4 was more abundant in normal tissue, and subtype S2 was more abundant in diseased tissue. Recent studies using single cell–based omics have shown different populations of smooth muscle cells with unique genetic signatures. Subtypes S4 and S2 could represent two such unique populations of smooth muscle cells, and the shift in abundance of these two populations could be a phenotype of aortic disease. Infiltration of different cells, such as immune cells, from the blood also likely occurs. Subtypes S1 and S6 were enhanced in disease tissue compared with the control tissue, and this group was enriched in inflammatory proteins. Although subtype S6 was enriched for markers of neutrophils and was predicted to be upregulated in all three disease states, subtype S1 was marked by monocyte and IgG (eg, B cell) proteins, and its upregulation appeared limited to the cases with aneurysms. This suggests differences could exist in the nature of the inflammatory involvement between TAA and TBAD, favoring unique involvement of monocyte and B cell infiltration in the etiology of aneurysms but not dissections. Alternatively, this could reflect shifting proportions due to the loss of smooth muscle cells without a large amount of additional infiltration or, even, phenotype switching of smooth muscle cells to phagocytic or monocyte-like cells. This will require significant additional study but presents one intriguing novel hypothesis to explain the differences between these two disease presentations in the descending aorta.

We also uncovered 42 differentially expressed proteins that could distinguish TAA from TBAD. These proteins were enriched in many canonical genes associated with aneurysms and dissections.^, Only eight proteins were upregulated in the TBAD cohort compared with the TAA group, and seven of these eight genes formed the second largest cluster in the network of all dysregulated proteins. The central node of this cluster was fibrillin 1, and the network was enriched in many processes linked to dissection, including collagen, integrin, and TGFβ signaling.^, These findings highlight the delicate balance governed by TGFβ signaling in maintaining the exact amount of extracellular matrix (ECM) content. Upregulation of TFGβ signaling can lead to both ECM destruction and deposition. Our findings suggest that aortas that dissect without aneurysm formation might have fragile media owing to excessive ECM composition and that aneurysms will be marked by a paucity of at least certain ECM components. These findings are consistent with those from previous studies, which have shown that patients with thoracic aortic dissection tend to have increased collagen content and patients with aneurysms tend to have a loss of ELN and collagen. Atherosclerosis can also be an important initiator of aneurysm formation. The largest cluster in the network of differentially expressed proteins contained 12 proteins, all upregulated in TAA compared with TBAD. Of these 12 proteins, 7 were apolipoproteins. Previous studies have demonstrated upregulation of apolipoproteins in thoracic aortic aneurysm samples compared with normal aortas. This finding was also consistent with the CAM results, because subtype S1, which is enriched in cholesterol and inflammatory proteins, was enhanced in TAAs compared with TBADs. This could signify a shift toward more inflammatory cells in aneurysm tissue compared with dissected tissue, which is known to occur after TGFβ-mediated metalloproteinase destruction of ECM. Finally, the third largest cluster included hemoglobin proteins, which have also been shown to be upregulated in proteomic studies of abdominal and thoracic aortas. Similarly, subtype 3 in the CAM profile, which was enriched in red blood cell markers, was enhanced in aneurysmal tissue. The significance of this trend is unclear but might represent mural thrombus, a stress response and upregulation of fetal proteins, a response to hypoxia locally, or contamination from the destruction of red blood cells in the turbulent flow that occurs in aneurysms. The utility of using these targets to predict acute dissection can be elucidated from experimental serum proteomics in animal models of dissection with and without aneurysms and prospective proteomic analysis of patients with and without aneurysms.

The present study had some limitations. First, we were unable to obtain significant background information such as the comorbidities of the donor control patients. Furthermore, the proportion of white patients was greater in the control group than in the disease groups (P = .03, χ² analysis), although we did not observe significant differences in the differential protein analysis when race was added as a covariate (data not shown). Also, the donor control patients were slightly younger than the patients with aortic disease, although the difference was not statistically significant (P = .18, analysis of variance). In addition, although no significant sex differences were present between all groups in our study (P = .78, χ² analysis), men were overrepresented. As such, additional studies with female patients will be important to uncover potential sex differences. Finally, for the most part, TBAD and TADA represent aortic dissections that do not degenerate to aneurysms and those that do degenerate to aneurysms, respectively. However, we did not have sufficient clinical information to fully rule out a sample in the TADA group representing an aneurysm that had subsequently dissected. We also failed to detect significant differences between TADA and TBAD and TADA and TAA. This was likely in part due to the heterogeneous nature of TADA patients and the low sample size. Nevertheless, we have uncovered significant candidate proteins that could be used as biomarkers for degenerative diseases of the descending aorta and the risk of acute dissection pending serum validation.

We have presented a novel comparison of proteomic profiles from normal human descending thoracic aorta and tissue with aneurysmal growth, dissection, and aneurysm with dissection. Although most of the currently available proteomic profiles of the human aorta are from the ascending thoracic and abdominal segments, we have added profiles from the descending thoracic aorta. Because patients with diseases of the descending aorta have different treatment and diagnostic algorithms compared with diseases of other segments, we sought to share some of the protein signatures that could help explain these clinical differences. We were able to identify THBS1 as potential biomarker of degenerative diseases of the aorta such as aneurysms and dissections. A comparison between patients with TAA and TBAD highlighted the crucial balance of TGFβ signaling in ECM deposition and destruction and provided a potential mechanism for how this pathway could be responsible for two different types of pathology that can lead to the same clinical outcome, namely dissection.

Conception and design: LS, DG, DM, DH, AA, SP

Analysis and interpretation: LS, AO, DG, DM, DT, RH, DH, YW, AA, SP

Data collection: DG, DM, AA, SP

Writing the article: LS, SP

Critical revision of the article: LS, AO, DG, DM, DT, RH, DH, YW, AA, SP

Final approval of the article: LS, AO, DG, DM, DT, RH, DH, YW, AA, SP

Statistical analysis: LS, SP

Obtained funding: DM, DH, AA, SP

Overall responsibility: SP

We are grateful for the support of the Cedars-Sinai Proteomics and Metabolomics Core Facility, including Rakhi Pandey and Ron Holewinski, for technical support in MS data acquisition. We are also deeply grateful to the Genomic and Proteomic Architecture of Atherosclerosis project leadership for access to the control thoracic aortic tissue and the UTHealth SCCOR biorepository for collection, storage, and release of the aortic aneurysm and dissection specimens.