SMAD3 deficiency promotes vessel wall remodeling, collagen fiber reorganization and leukocyte infiltration in an inflammatory abdominal aortic aneurysm mouse model

TGF-β signaling plays critical roles in the pathogenesis of aneurysms; however, it is still unclear whether its role is protective or destructive. In this study, we investigate the role of SMAD3 in the pathogenesis of calcium chloride (CaCl2)-induced abdominal aortic aneurysms (AAA) in Smad3−/−, Smad3+/− and Smad3+/+ mice. We find that loss of SMAD3 drastically increases wall thickening of the abdominal aorta. Histological analyses show significant vessel wall remodeling with elastic fiber fragmentation. Remarkably, under polarized light, collagen fibers in the hyperplastic adventitia of Smad3−/− mice show extensive reorganization accompanied by loosely packed thin and radial collagen fibers. The expressions of matrix metalloproteinases including MMP2, MMP9, and MMP12 and infiltration of macrophage/T cells are drastically enhanced in the vascular wall of Smad3−/− mice. We also observe marked increase of NF-κB and ERK1/2 signaling as well as the expression of nuclear Smad2, Smad4 and TGF-β1 in the vessel wall of Smad3−/− mice. In addition, we find that SMAD3 expression is reduced in the dedifferentiated medial smooth muscle-like cells of human AAA patients. These findings provide direct in vivo evidence to support the essential roles of SMAD3 in protecting vessel wall integrity and suppressing inflammation in the pathogenesis of AAAs.


Results
Loss of SMAD3 promotes AAA formation with marked vessel wall remodeling in response to CaCl 2 treatment. To explore the roles of SMAD3 in the pathogenesis of AAA, we analyzed the CaCl 2 -induced AAAs using the Smad3 knockout mice that was generated by replacing the translation initiation site of Smad3 gene with neomycine (Supplemental Fig. S1A) 19 . Western blot (WB) analyses confirm that the expression of SMAD3 in the resulting Smad3 −/− mice is abolished (Supplemental Fig. S1B). CaCl 2 treatment induced drastic overall increase of the abdominal aorta in Smad3 −/− mice compared with the Smad3 +/− and Smad3 +/+ control mice (Fig. 1A, Supplemental Fig. S1C). Before the treatment, the maximum aortic external diameters of Smad3 +/+ , Smad3 +/− , and Smad3 −/− were of similar size at about 0.66 mm. Three weeks after the treatment, the maximum external aortic diameter in Smad3 −/− mice increased to about 2.0 mm and showed an obvious 223% increase when compared with its pretreatment size (Fig. 1B). The aortic sizes of Smad3 +/− and Smad3 +/+ mice showed 36% and 21% increase respectively when compared with their size before the treatment (Fig. 1B). These results indicate that loss of SMAD3 promotes AAA formation in this experimental model. Consistent with the increased aortic size, histopathological analyses revealed apparent changes in the vessel wall architecture of Smad3 −/− mice versus Smad3 +/+ and Smad3 +/− mice: H&E staining showed that there is marked thickening of the wall with outward remodeling: this is distinct from most aneurysms showing thinning and degradation of vessel wall. We also observed the presence of neointima as well as enlarged media and adventitia in the vessel wall of Smad3 −/− mice (Fig. 1C). Despite the existence of neointima, the lumen diameter of the aorta in Smad3 −/− mice increased from 0.87 mm to 1.07 mm (a 24% increase) compared with Smad3 +/+ mice (Fig. 1C, the left bar graph): this is consistent with the overall increase of the abdominal aorta. However, mice exhibited individual variation and not all sections from enlarged aorta had obvious neointima. In addition, the number of cells as counted by the number of nuclei in cross sections of the vessel wall of Smad3 −/− mice increased 380% compared with Smad3 +/+ and Smad3 +/− control mice (Fig. 1C, the right bar graph).
To determine whether cell proliferation contributes to the increased cell number, we performed immunohistochemical (IHC) analyses using antibodies against Ki67, a nuclear cell proliferation marker. The result showed a marked increase in the number of proliferating cells indicated by Ki67 positive nuclei in Smad3 −/− mice compared with Smad3 +/+ mice in response to CaCl 2 treatment (Supplemental Fig. S2). This result suggests that cell proliferation contributes to the overall increase of the aortic size. mice, elastin in the media of the vessel wall of Smad3 −/− mice exhibited extensive degradation three weeks after the surgery ( Fig. 2A).
Scientific RepoRts | 5:10180 | DOi: 10.1038/srep10180 and spatial distribution confer the mechanical properties (such as strength and stiffness) of the aortic wall 23,24 . We thus examined collagen expression and architecture in the vessel wall. Picrosirius red staining under bright light detected a decrease of collagen staining intensity in the tunica adventitia of the aneurysm in Smad3 −/− mice (Fig. 2Ba). Under polarized light, the architecture including the thickness and organization (orientation and spatial distribution) of collagen fibers was readily identified by the changes of colors and packing density respectively 25,26 (Fig. 2B). As illustrated in panel 2Bb, the tunica adventitia of Smad3 −/− mice contained more green (i.e., thin) collagen fibers than the Smad3 +/+ mice, Also, the collagen in the adventitia of Smad3 −/− mice tended to contain areas of disruption where the fibers were loosely packed. Such areas were seldom seen in Smad3 +/+ mice. Although the predominant orientation of fibers in all of the sections was circumferential, fibers with a radial alignment were seen in central regions of the wall in some sections from Smad3 −/− mice. Thin and radial collagen fibers were only identified in the adventitia of aneurysms.
We used histology scores to quantify the color and organization of collagen fibers (see the Method section for explanation of the scoring system). The frequency distributions of the histological scores for collagen fiber thickness (indicated by color) and organization are shown in Fig. 2Bc. Overall, the color score was significantly lower for Smad3 −/− mice than for Smad3 +/+ mice (P = 0.0004; Wilcoxon rank-sum test) as was the organization score (P < 0.0001; Wilcoxon rank-sum test). The median values for each distribution and their corresponding Inter Quartile Ranges (IQR) were as following: the median(IQR) values for color distribution were 3(4,3): 3(3,2) for Smad3 +/+ : Smad3 −/− ; the median(IQR) values of organization distribution were 5(5,4): 4(4,2) for Smad3 +/+ : Smad3 −/− . Although the median values of color and organization are similar, the color and organization distribution are distinct (Fig. 2Bc). It should be noted that there were no green fibers and loosely packed fibers in Smad3 +/+ mice, and there were no orange fibers in Smad3 −/− mice. Due to the small sample size (n = 3), we did not include data from Smad3 +/− mice; however, their median scores (color = 4, organization = 5) and distributions were similar to those of the wild type mice.
Drastic increase in the expression and gelatinolytic activities of MMPs are detected in the abdominal aorta of Smad3 −/− mice. MMPs are responsible for the destruction of the orderly elastin and collagen network of the aorta 1 . MMP9, one of the prominently expressed MMPs in aortic aneurysms, works in concert with MMP2 to promote aneurysm progression 27 MMP12 plays a direct role in elastin degradation in AAAs 28 . To determine whether MMPs contribute to the disruption of vessel wall integrity in Smad3 −/− mice, we performed IHC using antibodies against MMP2, MMP9 and MMP12. The expression of MMP2, MMP9 and MMP12 was detected in the aorta of all three Smad3 genotype mice after CaCl 2 treatment; their expression was 6.6, 25.5 and 8.6-fold higher in Smad3 −/− mice for MMP2, MMP9 and MMP12, respectively, compared with the control mice (Fig. 3A).
To ensure that the increased expression of MMPs is accompanied by increased MMP gelatinolytic activities that accounts for the degradation of elastin and collagens, we performed in situ gelatinolytic zymography to detect MMP activities. In this assay, we used the OCT sections from abdominal aorta using highly quenched, fluorescein-labeled gelatin as a substrate that releases bright green fluorescence upon proteolytic digestion. Consistent with the increased expression of MMPs shown in Fig. 3A, abdominal aorta from Smad3 −/− mice show higher gelatinolytic MMP activities (Fig. 3B, the bright green fluorescence) than those from Smad3 +/+ mice in both the media and the adventitia of the vessel wall. Such MMP activities are diminished in the presence of the metalloenzyme inhibitor 1,10-phenanthroline ( Fig. 3B). This finding suggests that increased MMP activities contribute to vessel wall structure disruption observed in the aneurysms of Smad3 −/− mice.
Since inflammation is prominent in the aneurysms, we examined the expression of p65 (aka RELA), the master regulator of cellular proinflammatory responses in the vessel wall of the aneurysms in Smad3 −/− mice (Fig. 4B). Please note that the p65 positive signals were localized in the nuclei. The expression of nuclear p65 showed 1.6-fold and 2.2-fold increase in Smad3 +/− and Smad3 −/− mice, respectively, compared with Smad3 +/+ mice (Fig. 4B). Thus, Smad3 deficiency affects NF-kB expression level in a dosage-dependent manner. ERK1/2 signal activation has been shown to play critical roles in the pathogenesis of aneurysm 29,30 . We, therefore, examined the activation of ERK1/2 signaling in the CaCl 2 -induced AAAs. IHC analyses revealed robust upregulation of phosphorylated ERK1/2 throughout the vessel wall of aneurysms in Smad3 −/− mice (Fig. 4B). The expression of phosphorylated ERK1/2 in Smad3 −/− mice was 2.8-fold and 8.8-fold higher than in Smad3 +/− and Smad3 +/+ mice respectively.
Scientific RepoRts | 5:10180 | DOi: 10.1038/srep10180 SMAD3 deficiency induces the upregulation of SMAD2, SMAD4 and TGF-β1. SMAD2 and SMAD4 share similar structure and overlapping function with SMAD3 6,14,15. We thus examined the expression of SMAD2 and SMAD4 by IHC assays. We found that the expression of SMAD2 and SMAD4 was upregulated in Smad3 −/− mice (Fig. 4C). The activation of SMAD2 in the nucleus as indicated by the phosphorylated SMAD2 also significantly increased in the aneurysms of Smad3 −/− mice using the antibody recognizing phosphorylated SMAD2/3 (Fig. 4C). Smad3 −/− mice do not express SMAD3 (Supplemental Fig. S1B). This finding suggests that the expression of both SMAD2 and SMAD4 is upregulated in Smad3 −/− mice.
SMAD2 and SMAD4 are the intracellular mediators of TGF-β1 signaling. The upregulation of SMAD2 and SMAD4 prompted us to examine the expression level of TGF-β1. Indeed, the expression of TGF-β1 is increased in the vessel wall of CaCl 2 treated Smad3 −/− mice comparing with their wild type controls (Supplemental Fig. S3). Therefore, CaCl 2 treatment induces the upregulation of TGF-β1 that in turn may contribute to the increased expression of SMAD2 and SMAD4 via an autocrine mechanism.  SMAD3 expression is down-regulated in human AAAs. To examine the clinical relevancy of SMAD3 deficiency in the pathogenesis of non-syndromic form of human AAA, we re-examined SMAD3 mRNA expression in our previously published microarray analyses of human aneurysmal and non-aneurysmal abdominal aorta tissue 31 : SMAD3 mRNA expression was significantly reduced in AAA tissue (Fig. 5A).
To access SMAD3 expression patterns in the aneurysms, we performed IHC assays using abdominal aorta tissues from both AAA patients and age-matched controls (Supplemental Table S). In the control aorta, the expression of SMAD3, predominantly in the nuclei rather than in the cytoplasm, was restricted to the medial smooth muscle cells (SMC) (Fig. 5B). The medial SMCs show high expression of ACTA2 (in red color) in non-aneurysmal autopsy control abdominal aorta (Fig. 5C).
As expected, the vessel wall from AAA tissues showed severe medial degeneration with extensive leukocyte infiltration and fibrosis (Fig. 5C, Supplemental Fig. S5): ACTA2 expression was high in vasa vasorum but low or absent in spindle shaped cells in the media of the vessel wall (Fig. 5C), a finding consistent with large AAA. Histological analyses showed high level of SMAD3 expression in inflammatory cells, ganglia cells and endothelial cells for neoangiogenesis throughout the vessel wall of AAA tissues where extensive fibrotic collagen and few medial SMCs were present (Supplemental Fig. S5).
In sections co-stained with ACTA2 and SMAD3 antibodies, SMAD3 IHC signals were less intense compared with ACTA2 IHC signals (Fig. 5C). We found that SMAD3 staining was mostly present in the infiltrated leukocytes in AAA samples (arrowheads in Fig. 5C). Consistent with the reduced SMAD3 expression in the preserved SMC-like cells in AAA sections (Fig. 5B), less SMAD3 staining was detected in AAA tissues in spindle shaped SMCs-like cells where ACTA2 expression was low or absent (Fig. 5C). Taken together, the data indicate that SMAD3 expression is reduced in medial SMC-like cells of the AAA tissues where ACTA2 is significantly downregulated, suggesting that SMAD3 deficiency may be associated with medial SMC degeneration.

Discussion
Our study provides strong in vivo evidence supporting the important roles of SMAD3 in protecting vessel wall integrity and suppressing inflammation in the pathogenesis of CaCl 2 -induced AAAs. These results are consistent with the essential roles of TGF-β signaling in inhibiting leukocyte infiltration and in promoting the synthesis of elastin and collagens to increase ECM deposition 32 . Interestingly, our present study using CaCl 2 -induced AAA mouse model is consistent with two recent studies 19,22 using a different SMAD3 knockout mouse line. These three independent studies demonstrate that under stress condition disrupting SMAD3 expression activates SMAD-mediated TGF-β canonical pathway as well as SMAD-independent pathways including ERK1/2 and NF-kB signal pathways. The increased inflammation and the SMAD3 mutation-induced immune defects 19,22 promote vessel wall remodeling and aortic aneurysm formation as shown in the proposed model (Fig. 6).
The differences between the two Smad3 knockout mice should be pointed out: our Smad3 knockout mice were generated by deleting the Smad3 translation initiation site (i.e. SMAD3-null) (Supplemental Fig. S1A-B) and they do not show early sudden deaths; on the other hand, the mice used in the other two studies 20,21 were generated by deletion of the C-terminal activation domain (i.e. SMAD3∆C) and typically Figure 6. The schematic diagram of a hypothesized model showing that SMAD3 is required to protect vessel wall integrity and prevent aneurysm formation under stress conditions. In response to environmental stress (such as Ca ++ , TGF-β1 or AngII), SMAD-dependent canonical and SMAD-independent non-canonical signal pathways are activated in vascular cells including smooth muscle cells with defective SMAD3. The interplay of SMAD2, SMAD4, ERK1/2 and NF-κB signaling leads to increased MMP activities and the expression of proinflammatory genes, The resulting vascular inflammation, vessel wall remodeling (elastin degradation, collagen reorganization), and immune defects 19,22 promote the formation of abdominal aortic aneurysm. die between 1-3 months. Since the SMAD3∆C lacks the TGFBR interaction domain and phosphorylation site for the SMAD3 activation, it is not surprising that loss of SMAD3 and the expression of the inactive SMAD3∆C protein induce aneurysm formation via similar mechanisms.
Regarding the more severe vascular phenotypes observed in SMAD3∆C mice 21 , the expression of the defective SMAD3∆C protein may contribute to the pathology. Given that human AOS is caused by SMAD3 mutation throughout the SMAD3 protein 12,13 , the SMAD3∆C knockout mouse line should be an excellent mouse model for human AOS. Interestingly, Hilhorst-Hofstee et al also reported a small interstitial deletion of chromosome 15, leading to the deletion of SMAD3 in a three-generation family with features of the aneurysms-osteoarthritis syndrome (AOS) 33 . Therefore, our SMAD3 null mouse line may also serve as an AOS mouse model. Although SMAD3 positive cells in AAA patients are likely to be proinflammatory cells by morphology (Fig. 5,5S), further study is needed to determine whether these cells are of dedifferentiated SMCs using a SMC lineage marker. Unfortunately, it is not easy to have access to a number of small AAAs to validate the protective roles of SMAD3 in human aneurysms.
It is interesting to note that SMAD2 activation is increased in aneurysms with SMAD3 deficiency in this study (Fig. 4C) and the report by Dr. Xia's group 21 . The functional roles and regulatory mechanisms of SMAD2/3/4 in gene regulation are dependent on target genes and cell types 16,34 : for example, in mesenchymal progenitor cells, our previous study showed that SMAD3, not SMAD2, interacts with myocardin to modulate SMC gene transcription 35 . However, SMAD2 is now identified as the key regulator to control SMC gene transcription in neural crest derived VSMCs 36 . Interestingly, SMAD2 has been shown to be upregulated in thoracic aortic aneurysms that are of neural crest origins 37 . Indeed the underlying molecular mechanisms of the intricate interplay of TGF-β, SMAD2, SMAD3 and SMAD4 in the pathogenesis of aneurysm remains to be determined. Tissue-specific Smad3 knockout studies should be used to address the functional roles of SMAD3 in each cell type in future studies.
The robust activation of ERK1/2 signaling in the present work is consistent with a series of recent studies in which blocking Smad-mediated canonical TGF-β signal transduction in different mouse aneurysm models invariably boosts the activation of the ERK1/2-mediated non-canonical TGF-β pathway 20,21,29,30 . The upregulation of TGF-β1 and ERK1/2 signaling may also account for the observed proliferation in the vessel wall 38 (Fig. 1, Supplemental Fig. S3).These studies emphasize the complexity of TGF-β signaling pathways in the pathogenesis of aortic aneurysms. Further studies are needed to explore the molecular mechanisms of the complex yet delicate balance of SMADs, ERK1/2, and NF-kB signaling in the pathogenesis of aneurysms.
Since SMAD3 is the major intracellular mediator of the TGF-β1 signal pathway, the results of this study appear to support the protective roles of TGF-β1 in AAA described by Mallat's group 10 . However, the underlying molecular mechanisms may be different: in our study, we observed the increase of TGF-β1 in the vessel wall of CaCl 2 -induced AAA in Smad3 −/− mice. When interpreting molecular mechanisms caused by gene mutation, one should take into account that genetic mutations induce a series of adaptive changes in the organisms; thus the functional role of a mutated gene in genetic diseases may be significantly different from the gene's function under normal development. Different Smad3 mutations may elicit different alterations and adaptations in gene interaction network and signal pathway activation. Although three studies using two Smad3 knockout mouse lines reveal the protective roles of SMAD3 via similar mechanisms, the functional roles of defective SMAD3 in human aneurysms appear to be more complex. For example, aneurysms from patients with SMAD3 mutations paradoxically display increased collagen expression along with increased SMAD2 phosphorylation and increased SMAD3 protein 13 . It is very likely that the upregulation of SMAD proteins and collagen is part of the adaptive response of SMAD3 mutations. The whole exome or whole genome sequence data in a large enough sample set may reveal potential association of rare variants in the SMAD3 gene with AAA patients.
Finally, we would like to highlight that we are able to observe the unique architectural features of collagen fibers using the simple Picrosirius red staining combined with polarized light microscopy 25,26 (Fig. 2B). This result is consistent with the study using electronic microscope observed by Dr. Tan's group 20 . Aortic function and mechanical properties are influenced by the amount, organization, thickness, and density of collagen fibers 24,39 . Changes in collagen fiber architecture caused by molecular changes such as SMAD3 deficiency thus play important roles in the pathogenesis of AAA. Although collagen amount in AAAs are commonly examined by Picrosirius red staining, the important architectural properties of collagen fibers observable under polarized light are seldom accessed. Our finding of thin, disorganized fibers in Smad3 −/− mice is consistent with suggested roles of SMAD3 in structural integrity of the vessel wall. We believe that this histological approach could provide important insight into the pathophysiology of AAA.
In conclusion, this study reveals essential roles of SMAD3 in protecting the vessel wall integrity and suppressing inflammation. Our study suggests that the balance and interplay of NF-kB, ERK1/2, and SMAD2/4 signaling pathways have important roles in the pathogenesis of AAAs. Given the complexity of TGFβ signaling in aneurysm pathogenesis 6,32 , and aortic heterogeneity in embryonic origin and local hemodynamic stress 40,41 , caution should be taken when considering SMAD3 or TGF-β signaling pathways as the pharmacological target. Since there is a subgroup of human AAAs showing significant wall thickening and tremendous inflammatory cell infiltration, our CaCl 2 -treated Smad3 −/− mice can be used as a new valuable pro-inflammatory AAA mouse model. Methods Animal model. The animal protocols used in this study were approved by the Animal Investigation Committee at Wayne State University. The animal procedures performed conform the NIH guidelines (Guide for the care and use of laboratory animals). Smad3 +/− mice 19 were maintained at C57BL6 genetic background. Smad3 +/− mice are fertile and phenotypically indistinguishable from the wild type mice.
Aneurysm induction. AAAs were induced with CaCl 2 using age matched male Smad3 −/− , Smad3 +/− and Smad3 +/+ at 8-12 weeks of age according to a well-established CaCl 2 -infusion protocol 42,43 . After anesthesia using 2% Avertin intraperitoneally (0.25 mg/kg body weight), the abdominal aorta between the renal artery and the iliac bifurcation was exposed by blunt dissection. The aorta was photographed using the Nikon SMZ1500 stereo microscope (Nikon Instruments Inc.). Next, a piece of gauze soaked with 0.5 M CaCl 2 was applied uniformly along the abdominal aorta from the left renal artery to the bifurcation for 10 minutes. The aorta and the abdominal cavity were then rinsed with 0.85% sodium chloride three times before the incision was closed. The non-steroidal anti-inflammatory agent, carprofen (at 5 mg/kg in 0.5 ml of normal saline) was then administered to minimize the pain caused by the procedure. All mice were allowed to recover under a warm blanket and closely monitored until they regained consciousness. Mice were then placed into a clean cage and were kept under standard housing condition.
Aorta collection and size measurement. Three weeks after surgery, the mice were anesthetized intraperitoneally using 2% Avertin (0.25 mg/kg). The abdominal aorta was then exposed for photography using the Nikon SMZ1500 stereo microscope. After cardiac perfusion using 4% paraformaldehyde, the abdominal aorta from the left renal branch to the bifurcation was removed, embedded in optimal cutting temperature (OCT) medium (Tissue-Tek® O.C.T. Compound, product code #4583) and sectioned at a thickness of 6 μm.
Aorta size determination. The abdominal aorta filled with blood in live mice reflects aorta size more closely than aorta harvested postmortem. Digital images of the abdominal aorta both before and three weeks after CaCl 2 treatment were thus taken and used to determine the external aortic diameter for each mouse. The diameter of the aorta is the average of three measurements of the maximum external aortic diameter. The diameter of the aorta at the end of CaCl 2 treatment was expressed as a percentage increase from its respective initial diameter.

Histology and immunohistochemical (IHC) analyses.
The histology and IHC staining as described before 44,45 were performed using OCT sections at comparable locations to the left renal artery along the abdominal aorta from Smad3 +/+ , Smad3 +/− , or Smad3 −/− mice. Sections were stained with hemotoxylin-eosin (H&E), Verhoeff-Van Gieson (VVG; Elastin staining kit, Polysciences, Inc. Cat. #25089) or Picrosirius Red (Collagen Staining kit from Polysciences, Inc. Cat. #24901) following the manufacturer's instructions. Digital images were obtained using Leica DM4000B light microscope (Leica Microsystems, Inc.). Digital images of H&E stained sections were used for morphometric analyses. The cell numbers in each section were counted by the nuclei using the Image-Pro Analyzer software (Media Cybernetics). The aortic lumen perimeter was determined by NIH Image J software (http://rsb.info.nih. gov/ij/) 46 .
For IHC analyses, the primary antibodies are the following: rabbit anti-Ki67 (a marker for cell proliferation, Abcam, ab15580), rabbit anti-MMP2 (Abcam, ab37150), rabbit anti-MMP9 (Abcam, ab38898), rabbit anti-MMP12 (Abcam, ab39876), rabbit anti-CD3 (Abcam, ab5690), rat anti-F4/80 (Abcam, ab6640), rabbit anti-p65 (aka RELA, one essential component of the NF-κB complex, Santa Cruz, sc-109), goat anti-SMAD2 (Santa Cruz, sc-6200), goat anti-pSMAD2/3 (Santa Cruz, sc-11769), goat anti-SMAD4 (Santa Cruz, sc-1909), and goat anti p-ERK1/2 (Santa Cruz, sc-16982). The primary antibodies were detected by the Avidin-Biotin Complex (ABC) staining systems (VECTASTAIN Elite ABC Kit, Vector Laboratories, Inc.). Sections were counterstained with Hematoxylin to visualize nuclei. Digital images of all mouse sections were obtained using the Leica DM4000B light microscope. Please note that there is no IHC signal in the lumen of the aorta because the aortas were perfused before being harvested. Threshold gated positive signal on IHC images was detected on the whole aorta vessel wall including the intima, media and adventitia layers. The antigen positive signals over the whole aorta cross section area was quantified using the integrative optical density (IOD) function in the Image-Pro Analyzer software (Media Cybernetics, Bethesda, USA).
Collagen fiber structural organization assessment. Picrosirius red-stained sections were viewed with the bright field and polarized light. This combination allows structural properties of the collagen fibers to be assessed 26,47 . Specifically, the color changes as fiber thickness increases: from green to yellow to orange. Additionally, the optical properties of the birefringent fibers can be used to examine their orientation by rotating the section on the microscope stage and observing the angle at which the fibers become dark when viewed with linearly polarized light 25 .
The color and organization of collagen in the tunica adventitia of each section was graded on a 1 to 5 scale in a blinded fashion (by Dr. Peter Whittaker). For collagen fiber color, the scale was; 1 -predominantly green , 2 -mainly green with some orange and yellow, 3 -a mixture of all three colors, 4 -mainly orange and yellow with some green, 5 -predominantly orange. For collagen fiber organization, the scale was: 1 -extensive areas of disrupted and loosely packed around most of the circumference and abnormal structure (the presence of radially aligned fibers), 2 -large areas of collagen disruption (>50% of the vessel circumference), 3 -focal areas of collagen disruption (10-50% of the circumference), 4 -small areas of collagen disruption (<10% of the circumference), and 5 -densely packed collagen around the entire vascular circumference. The histological scores of collagen thickness (color) and structural organization (packing density) are given as medians with their inter quartile ranges (IQR) in the Result section and were compared using the Wilcoxon rank-sum test.
in situ gelatinolytic zymograph. Gelatinolytic activity was demonstrated in OCT sections using DQ-gelatin as a substrate using EnzChek® Gelatinase/Collagenase Assay Kit following manufacturer's instruction (Invitrogen, E12055). Cryostat sections of aorta were air-dried for 10 min. DQ-gelatin was dissolved in water at a concentration of 1 mg/ml. The stock was further diluted at 1:40 ratio using in 1x reaction buffer. About 200 μl of the diluted solution was put on top of the sections and covered with a coverslip. After 2-hour incubation at room temperature, slides were mounted with Vectashield with DAPI (Vectorlabs) and examined under the Leica DM4000B microscope. Specificity of the gelatinolytic activity due to MMP activities was determined with the presence of the metalloenzyme inhibitor 1,10phenanthroline (1 mM). The inhibitor added at a concentration of 1 mM along with the gelatin substrate was able to reduce the gelatinolytic activity.
Studies with human tissue samples. The research carried out was in accordance with the Declaration of Helsinki. The collection of the human tissues was approved by the Institutional Review Board of Wayne State University, Detroit, Michigan, USA and all subjects provided their informed and written consent to participate in this study. Aortic wall tissue specimens were collected from patients undergoing AAA repair operations at the Harper University Hospital, Detroit, Michigan. Non-aneurysmal infrarenal aortic samples were collected at autopsies 41 . Samples were stored in RNAlater for RNA isolations (Ambion), and in phosphate-buffered formalin (and embedded in paraffin) for histological and immunohistochemical analyses.
We used two microarray platforms to generate global mRNA expression profiles for both aneurysmal and non-aneurysmal human abdominal aorta. The details on these studies have been described previously 31 . The microarray data can be obtained at the Gene Expression Omnibus (GEO) database (Series# GSE7084; http://www.ncbi.nlm.nih.gov/geo/). A Box-and-Whiskers plot of SMAD3 mRNA levels for AAA patients and infrarenal aorta tissues from non-aneurysmal autopsy based on microarray expression data (as described previously 31 ) is shown in Fig. 5A.
Human aortic tissues from three AAA patients and age-matched non AAA patients (Supplementary Table S) were used for the IHC assay according to a protocol described previously 48 . Formalin-fixed paraffin sections (5-μm) were deparaffinized followed by antigen retrieval using microwave treatment in 0.01 M sodium citrate solution 13 and incubated with SMAD3 primary antibody (Abcam ab28379, 1:100 dilution) or ACTA2 primary antibody (Abcam, ab5694) on an automatic immunostainer (Autostainer, DAKO, Carpinteria, CA). A biotinylated secondary antibody with avidin-biotin peroxidase amplification from DAKO was used and the signal detected using diaminobenzidine (DAB) as a chromogen. Images in the vessel wall from each sample were taken using an Olympus BX45TF microscope.

Statistical analysis.
Unless specified, all values in this study are expressed as mean±standard errors of the mean (SEM). Differences between two different groups were evaluated by unpaired Student t test using Prism software (Graphpad). Differences with p values <0.05 were considered statistically significant.