Basement membrane collagen IV deficiency promotes abdominal aortic aneurysm formation

Abdominal aortic aneurysm (AAA) is a complex disease which is incompletely accounted for. Basement membrane (BM) Collagen IV (COL4A1/A2) is abundant in the artery wall, and several lines of evidence indicate a protective role of baseline COL4A1/A2 in AAA development. Using Col4a1/a2 hemizygous knockout mice (Col4a1/a2+/−, 129Svj background) we show that partial Col4a1/a2 deficiency augmented AAA formation. Although unchallenged aortas were morphometrically and biomechanically unaffected by genotype, explorative proteomic analyses of aortas revealed a clear reduction in BM components and contractile vascular smooth muscle cell (VSMC) proteins, suggesting a central effect of the BM in maintaining VSMCs in the contractile phenotype. These findings were translated to human arteries by showing that COL4A1/A2 correlated to BM proteins and VSMC markers in non-lesioned internal mammary arteries obtained from coronary artery bypass procedures. Moreover, in human AAA tissue, MYH11 (VSMC marker) was depleted in areas of reduced COL4 as assessed by immunohistochemistry. Finally, circulating COL4A1 degradation fragments correlated with AAA progression in the largest Danish AAA cohort, suggesting COL4A1/A2 proteolysis to be an important feature of AAA formation. In sum, we identify COL4A1/A2 as a critical regulator of VSMC phenotype and a protective factor in AAA formation.

www.nature.com/scientificreports/ -apoptosis 2 . Aside from age and male sex, the strongest risk factors for AAA are smoking and a family history of the disease, while, hyperglycemia and diabetes mellitus are negatively associated with AAA 3,4 . As for most tissues, arteries comprise two types of ECM: Interstitial matrix and basement membrane (BM). The artery BM is located at the base of the endothelial lining and throughout the medial layer, where it encapsulates VSMCs 5,6 . The principal component of the artery BM is Collagen IV hemi-trimers consisting of two ⍺1 (COL4A1) chains and one ⍺2 (COL4A2) chain. Unlike fibrillar collagens, Collagen IV hemi-trimers cross-link to form a meshwork, which along with laminins assemble into a scaffold for additional BM constituents to bind. As such, the artery BM serves as an adhesive substrate for vascular cells to anchor, and at the same time directly (by integrin binding) or indirectly (by regulating diffusion of macromolecules to and from cells) maintains cellular homeostasis 7 .
COL4A1 and COL4A2 are encoded by their homonymous genes, which are organized head-to-head on opposite strands and separated by a 126-bp bidirectional promoter on chromosome 13 (13q34) 7 . The locus harboring COL4A1/A2 is a highly replicated susceptibility locus for coronary-and peripheral artery disease, respectively 8,9 , and the relevance of the BM in artery biology is evident from rare familial syndromes caused by missense mutations in COL4A1. Aneurysms of carotid and cerebral arteries are common features of these syndromes, collectively termed hereditary angiopathy with nephropathy, aneurysms, and muscle cramps (HANAC syndrome) 10,11 , underlining the importance of COL4A1/A2 integrity in the artery wall. Moreover, we previously reported that non-lesioned arteries from diabetic patients, which display relative protection from AAA development (of note, the relative contribution of hyperglycemia per se 4 and metformin intake 12 to AAA protection observed in diabetics remains elusive) have a distinct elevation in BM proteins lead by a 40% increase in COL4A1/A2-an observation in line with reports on increased of COL4A1/A2 in diabetic vascular organoids 13 and with the well-known accumulation of BM in small vessels 14 .
Based on the above observations, we hypothesize that the level of COL4A1/A2 in arteries (as a consequence of genetic or environmental factors) is a critical determinant for AAA formation.
The objective of this study was to test the hypothesis that COL4A1/A2 is a protective factor in AAA disease combining a novel mouse model of Col4a1/a2 deficiency with proteomic analysis of non-lesioned artery and AAA tissue from human patients.

Materials and methods
Mouse procedures. Generation of Col4a1/a2 +/− mice (129/Svj) (a kind gift from Ernst Pöschl, University of East Anglia, Norwich, UK) have been described previously 15 . Col4a1/a2 +/− mice were crossed with Col4a1/a2 +/+ littermates to generate Col4a1/a2 +/− and Col4a1/a2 +/+ offspring at a 1:1 ratio for experimental procedures. Primers used for genotyping were 5′CAT CGC CTT CTA TCG CCT TCT TGA CG, 5′CTG GCT CAG AGC GTC TTG , and 5′ACC ACC TCT GAG TTT CTG GA resulting in 351 bp and 610 bp amplicons for WT and KO alleles, respectively. Mice used for general characterization (12 weeks of age) and wire myography experiments (9-11 weeks of age) were euthanized by CO 2 /O 2 inhalation and tissues were either snap frozen in liquid nitrogen or immersed in 4% formaldehyde in phosphate-buffered saline (PBS) for 24 h and then transferred to PBS (0.05% sodium azide) at 4 °C. Exclusion criteria set before experiments were death, or euthanasia of mice showing pre-defined humane endpoints. No mice were excluded from the presented experiments except for three Col4a1/a2 +/+ mice that died in the first two days after PPE surgery due to post-surgery related issues reducing the Col4a1/a2 +/+ group to n = 14. No data were excluded from analysis. Mouse group allocation was random, concealed, and all experimental procedures and analyses were performed blinded to the observers. The study is reported in accordance with ARRIVE guidelines. All procedures were performed in accordance with protocols approved by the local authorities (the Danish Animal Experiments Inspectorate) and conform to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes. PPE model of AAA .  week-old mice were anaesthetized (ketaminol 100 mg/kg, xylazine 10 mg/ kg) and the baseline infrarenal aortic lumen diameter during systole was determined by ultrasonography using the LOGIQ e portable ultrasound imaging system (GE healthcare) with a 22 MHz central frequency probe (L10-22-RS, GE healthcare). Mice underwent laparotomy and the infrarenal aorta was isolated. The baseline infrarenal outer aortic diameter during systole was determined from video recordings captured using a Leica Camera attached to the microscope. Then the infrarenal aorta was occluded using silk sutures and aortotomy was introduced allowing infusion of porcine pancreatic elastase (PPE) (1.5 units/ml, Sigma Aldrich) via polyethylene catheter expanding the aorta to twice the baseline diameter for 5 min, followed by flushing of the aorta and abdomen cavity with physiological saline. The aortotomy was closed with a suture, blood flow was restored, and the laparotomy was closed. At day 7 and 14 post-surgery, the maximal infrarenal aortic lumen diameter during systole was determined by ultrasonography in mice anesthetized by isoflurane. At day 14, the infrarenal aorta was isolated in anaesthetized mice and a video of the aorta was recorded to determine the maximal infrarenal outer aortic diameter during systole. Mice were sacrificed, aortas were isolated and fixed in 4% formaldehyde overnight, then changed to PBS (0.05% sodium azide).
Wire myography. Abdominal aortas were removed from mice euthanized by CO 2 /O 2 inhalation, dissected free of perivascular adipose tissue, transferred to ice cold physiological salt solution (PSS, 115 mM NaCl, 25 mM NaHCO 3 , 2.5 mM K 2 HPO 4 , 1.2 mM MgSO 4 , 5.5 mM glucose, 10 mM HEPES, and 1.3 mM CaCl 2 ) and mounted on 200 µm pins in a Danish Myo Technology multi wire myograph system (Danish Myo Technology, Søften, Denmark). After mounting, segments were allowed to rest at 37 °C for 30 min in calcium free PSS supplemented with 1 mM EGTA and 3 µM SNP. Chambers were continuously aerated with 5% CO 2 in air. Then, segments were stretched in steps of 200 µm, every 3 min, to an internal diameter and wall tension corresponding to 150 mmHg (20 kPa) according to the Law of Laplace. Following, tension was reduced to 100 mmHg (13.3 kPa), buffer changed to PSS, and segments allowed to rest for 30 min before smooth muscle viability was tested by addition of 60 mM K + or 10 µM phenylephrine, and endothelial cell viability by addition of increasing concentrations of acetylcholine. Finally, stretch was released to 50% and segments were fixed in 4% formaldehyde overnight, before embedding in paraffin. Segments were cut perpendicularly to the length of the aorta, and wall and media areas determined on Masson trichrome stained sections. Wall stress and strain was calculated, and β-values determined as reported previously 15 on basis of wall thicknesses estimated from the morphometry. Method adapted from Mulvany et al. 16 . qPCR. RNA was extracted from snap frozen tissue using NORGEN Total RNA Purification Kit (37500, Norgen Biotek Corp.) and cDNA was synthesized by High-Capacity cDNA Reverse Transcription Kit (4368814, Thermo Fisher Scientific). qPCR was performed using SYBR™ Green PCR Master Mix (4364344, Thermo Fisher Scientific) and the ViiA7 Real-Time PCR system (Thermo Fisher Scientific). Primers used were: Col4a1: 5′GGC CCT TCA TTA GCA GGT GT and 5′GTG AGG ACC AAC CGT TAG GG; Col4a2: 5′CCC GGA TCT GTA CAA GGG TG and 5′CGC CTT TTG AGA TTA CGC CG; Gapdh: 5′GGC TGC CCA GAA CAT CAT and 5′CGG ACA CAT TGG GGG TAG . Gapdh was used as reference gene.
Preparing human artery tissue for proteomics. Non-lesioned internal mammary arteries (IMAs) from coronary artery bypass graft (CABG) procedures were obtained from the Odense Artery Biobank. Immediately after tissue harvest, IMAs were systematically dissected and stored at − 80 °C until liquid chromatography tandem-mass spectrometry (LC-MS/MS)-analysis as previously described 5 . All donors had given written consent, and the study was performed in accordance with protocols approved by the local ethics committee (S-20100044) and conformed to the principles of the Declaration of Helsinki.
Targeted MRM analysis on tissue samples was adapted from previously described procedure 18 . Heavy isotopelabelled standard peptides (JPT Peptide Technologies) were added to each sample aiming at a 0.1-10 peak area ratio between each endogenous peptide assayed and its corresponding heavy peptide. Each sample (1 µg endogenous tryptic peptides) was run on an Easy-nLC II NanoLC system using a C18 trapping column (length 2 cm; internal diameter 100 µm) for desalting and a C18 analytical column (length 10 cm, internal diameter 75 µm) for peptide separation (Thermo Scientific). Peptides were eluted with a three-step 45 min gradient of 0.1% formic acid in ACN at a flow rate of 300 nl/min. Peptides were ionized using a Nanospray Flex ion source (Thermo Scientific) and analyzed on a TSQ Vantage triple quadruple mass spectrometer (Thermo Scientific) in MRM mode. MRM raw files were processed using Pinpoint 1.3 (Thermo Scientific). The peak area ratio between endogenous and heavy isotope labelled spiked peptide was used for data analysis. Peak area ratios for Col4a1 (target peptide: GYP*GAP*GLR, P* indicates hydroxyproline) and Col4a2 (target peptides: GLDGFQGPSGPR, IAVQPGTLGPQGR, SVSIGYLLVK) were normalized to Gapdh (target peptides: GAAQNIIPASTGAAK and LISWYDNEYGYSNR) within the same sample (Fig. 1e).
For explorative proteomics of mouse aortas, samples (4 µg tryptic peptides per sample) were randomly labelled with 10-plex tandem mass tags (TMT, Thermo Scientific) and combined into 2 × 3 samples, high-pH fractionated, and analyzed by nano-LC-MS/MS virtually as previously described 17 .
All Eclipse raw data files were processed and quantified using Proteome Discoverer version 2.4 (Thermo Scientific) as previously described 17 .
Proteomic analysis of human IMAs was based on formalin-fixed paraffin-embedded sections of IMAs obtained from CABG surgery. For this study, we analyzed the proteomic dataset of the 30 non-diabetic patients. Please refer to the original article for details on patients and proteomic analysis 5 .
All donors had given written consent, and the study was performed in accordance with protocols approved by the local ethics committee (S-20100044) and conformed to the principles of the Declaration of Helsinki.
Plasma samples from the VIVA screening trial. Baseline heparin plasma samples from a randomized population-based screening trial for AAA, PAD, and hypertension in > 50,000 men, 65-74 years of age, in the Mid Region of Denmark 19 were taken consecutively from 476 AAA patients at the time of diagnosis and from 200 age-matched controls (without AAA or PAD) (all 676 samples were incuded in this study). AAA was defined as a maximal abdominal aortic diameter ≥ 30 mm, and PAD was defined as an ankle-brachial index (ABI) lower than 0.90 or > 1.4. AAA cases among first-degree relatives, smoking status, coexisting diabetes mellitus, hypertension, and use of β-blockers, angiotensin-converting enzyme (ACE) inhibitors, and statins were recorded, as were body mass index (BMI), diastolic-and systolic blood pressure. Maximal anterior-posterior diameter of the infrarenal aorta was measured in the peak of the systole between inner edges of the aorta. The interobserver variation of aortic diameter measurements was 1.52 mm 20 . Growth rates of small AAA in patients kept under ns ns Heart/body (mg/g) Col4a1 mRNA a * ****  Supplementary Fig. 1). (h, i) Weight (h) and relative heart mass (i) of baseline Col4a1/a2 +/− mice and Col4a1/a2 +/+ littermates. Data are presented as mean ± SEM. *p ≤ 0.05, ***p ≤ 0.001, ****p ≤ 0.0001, ns = non-significant. Unpaired t-test was used in a-d and g-i. Although both sexes are plotted, no intersex comparisons were performed. f = female, m = male. n per group = 8-12. www.nature.com/scientificreports/ surveillance were calculated by individual linear regression analysis, using all observations. Blood samples were centrifuged at 3000×g for 12 min, aliquoted, and stored at -80 °C until analysis. Informed consent was obtained from all subjects before participation, and the study was approved by the Local Ethics Committee of the Mid Region of Denmark, and the data protection agency, and conformed to the principles of the Declaration of Helsinki.
ELISA measurements of PRO-C4 and C4M. Markers for COL4A1/A2 formation (PRO-C4) 21 and MMP-2/9-mediated degradation (C4M) 22 were assessed by ELISA developed by Nordic Bioscience, Herlev, Denmark. Briefly, 96-well streptavidin coated plates were coated with the appropriate biotinylated synthetic peptides and incubated for 30 min at 20 °C. The plates were washed five times with washing buffer (20 mmol/L Tris, 50 mmol/L NaCl, Ph 7.2). Twenty µL standard peptide or pre-diluted heparin plasma sample was added to appropriate wells, followed by incubation with peroxidase-conjugated specific monoclonal antibodies for 1 h at 20 °C. Finally, the plates were washed five times with washing buffer and tetramethylbenzidine (TMB) (cat.438OH, Kem-En-Tec Diagnostics, Taastrup, Denmark) was added, and the plates were incubated for 15 min at 20 °C in darkness. The TMB reaction was stopped by adding 0.18 M H 2 SO 4 and measured at 450 nm with 650 nm as reference and concentrations read from a calibration curve. All samples were measured within the detection range of the assays.
Analysis of GWAS data. The GWAS metadata analyzed in this study have been described elsewhere 23 .
In brief, a meta-analysis of six independent cohorts, combining 4972 AAA cases and 99,858 controls, was conducted using the METAL software package. Each cohort underwent imputation which was filtered by quality score (Q > 0.9) and only SNPs with minor allele frequencies > 0.05 were included in the final analysis, which consisted of 5,363,770 SNPs assessed in all participants. In the present study, we tested 1667 SNPs in the 557 kb COL4A1/A2 locus (here defined as the COL4A1 and COL4A2 genes (357 kb) flanked by 100 kb border regions) for association with AAA.
Statistical analyses. Two-group comparisons were performed by Student's t-tests and presented as mean ± standard error of the mean (SEM). Comparison of two-group XY graphs were performed by two-way ANOVA and presented as mean ± SEM. Explorative proteomics data were analyzed by Student's t-test for each protein and subsequent false discovery rate (fdr) correction for multiple testing, and GO enrichment analysis was performed using default settings of the DAVID Bioinformatics Resources 24,25 . Replicas (n) are specified on figures or in figure legends whenever data are not plotted as individual data points. A fixed value of p ≤ 0.05 was the criterion for reliable differences between groups, and the significance level symbols used in the figures are: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001, while non-significant comparisons are shown as "ns". AAA GWAS metadata was analyzed using the METAL software package 26 and was weighted using the effective sample size [N eff = 4/(1/N cases + 1/N controls )] for each cohort.
Statistical analysis of C4M-and PRO-C4-levels in heparin plasma samples from the VIVA screening trial: Dichotomous variables were expressed as proportions and compared by chi-square tests. One sample Kolmogorov-Smirnov tests and normal probability plots (not shown) were used to determine whether continuous variables were normally distributed, and compared between dichotomous variables by 2-tailed unpaired Student's t-tests and Pearson's correlation analyses regarding other continuous variables. Baseline variables associated with AAA and COL4A1/A2 fragments with both p-values < 10% were considered potential confounders. The COL4A1/A2 fragments were tested as independent predictors of AAA by logistic regression analysis, without and with adjusting for the identified potential confounders: Smoking, hypertension, previous ischemic event in the brain, heart or legs, use of statins, and diastolic blood pressure. The associations between COL4A1/A2 fragments were correlated to maximal aortic diameter and AAA growth rate by univariate and partial Pearson's correlation analysis without and with adjusting for the aforementioned potential AAA confounders, respectively, supplemented with baseline AAA diameter regarding growth rate. Finally, the biomarkers were tested for independent association with need for surgical evaluation and death of any cause by Cox regression analyses without and with adjusting for the aforementioned potential AAA confounders, respectively.

Col4a1/a2 hemizygosity augments experimental AAA development. To test for causation
between reduced Col4a1/a2 and AAA, we induced AAA in male Col4a1/a2 +/− and Col4a1/a2 +/+ mice by the intraluminal porcine pancreatic elastase (PPE) infusion technique 28 . We did not detect differences in outer diameter by direct measurement before, during and after PPE surgery indicating no effect of Col4a1/a2 deficiency on mechanical deformation (Fig. 2a). However, 14 days post-surgery, Col4a1/a2 +/− mice had a modest 17% increase in maximal systolic infrarenal aortic inner (lumen) diameter measured by ultrasonography (Fig. 2b,c and Supplementary Fig. 2a) and a 31% increase in outer aortic diameter relative to diameters before surgery (Fig. 2d, Supplementary Fig. 2b). No difference in body weight was observed between groups (Fig. 2f). Our observation that Col4a1/a2 deficiency does not affect biomechanical properties of the aorta was verified in wire myography experiments performed on unchallenged abdominal aortas from an independent mouse cohort. This experiment showed that resting wall tension (Fig. 2g), structural rigidity (Fig. 2h) and β-values ( Supplementary  Fig. 2c), a measure proportional to Youngs modulus of the arterial wall, were not affected by genotype. Furthermore, the wire myography data indicate that the function of VSMCs under these experimental conditions were   (Fig. 2i) or a contractile agonist (Fig. 2j) were similar.
To translate the observations from unchallenged mouse aortas to non-lesioned human arteries, we analyzed a proteome dataset of human internal mammary arteries (IMAs) (n = 30) previously published from our group 5 ,   For (d, e), the red horizontal lines show the corrected (fdr = 0.05) and uncorrected (p = 0.05) level of significance, respectively. Data in (b) and (d, e) were analyzed by unpaired t-test for each protein followed by fdr correction for multiple testing. www.nature.com/scientificreports/ and found significant correlations between COL4A1/A2 and BM components as well as VSMC markers (Table 1 and Supplemetary Fig. 3).
The human COL4A1/A2 locus harbors alleles with moderate association to AAA risk. True disease-associated genetic risk variants may be overlooked as a consequence of stringent statistical criteria used in GWAS. Instead, we here used a hypothesis-driven approach based on the largest available AAA GWAS metadata 23 testing only the 1667 single nucleotide polymorphisms (SNPs) in the 557 kb COL4A1/A2 locus for association with AAA. We identified a 45 kb region of high linkage disequilibrium in COL4A2 intron 4 harboring multiple SNPs moderately associated with AAA [lead SNP (rs7320498) p = 6.5·10 -5 , OR (A-allele) = 0.887 (95% CI 0.838-0.939)] (Supplementary Fig. 4).
COL4 is degraded during AAA development and areas with sparse COL4 are depleted in MYH11. To evaluate whether COL4 could be involved in maintaining VSMC phenotype integrity in human AAA lesions, we assessed human AAA specimens (n = 5) obtained from open repair surgery by immunohistochemistry. The density of COL4 staining was reduced, patchy and disorganized as compared to non-lesioned aortas obtained from CABG (n = 5), and areas with sparse COL4 had negligible MYH11-staining ( Fig. 4 and Supplementary Fig. 5).
Circulating COL4A1 degradation fragments were independently associated to AAA . Our observations from human AAA specimens suggested that COL4A1/A2 degradation is part of the AAA pathology. To investigate whether degradation of COL4A1/A2 in AAA lesions results in release of circulating COL4A1/ A2 fragments in AAA patients, we quantified two well-characterized COL4A1/A2 fragments in plasma samples from a nested case-control cohort of the VIVA screening trial (476 AAA patients and 200 age-matched controls) 19 (Supplementary Table 1). The MMP-2/9-generated COL4A1 degradation product, C4M, was increased in AAA patients and independently associated with AAA, whereas the COL4A1/A2 formation marker, PRO-C4 was not significantly altered ( Table 2). Both markers showed univariate relation to maximal aortic diameters, but only C4M after adjustment for confounding factors (Table 3). Neither marker was related to AAA growth rate or need for surgical repair, whereas both were related to mortality after adjusting for confounders (Supplementary Table 2).

Discussion
In the present study, we tested the hypothesis that BM proteins COL4A1/A2 have a role in AAA protection. We used hemizygous Col4a1/a2 +/− mice having a 50% reduction in aorta Col4a1/a2, and found that these mice had increased formation of experimentally-induced AAA. Although homozygous Col4a1/a2 KO (Col4a1/a2 −/− ) mouse embryos do not survive intrauterine life, it is noteworthy that these embryos show structural deficiency of BM in blood vessels, which appear dilated, and the presence of internal bleedings indicate rupture of blood vessels 27 . These observations support our findings from adult Col4a1/a2 +/− mice.
To address underlying causes of the AAA phenotype, we investigated proteomic consequences of reduced Col4a1/a2 in both unchallenged aortas and AAA lesions. Interestingly, the level of several BM components was reduced in Col4a1/a2 +/− aortas implying that the COL4A1/A2 scaffold is a limiting factor for BM maturation. Indeed, early in vitro studies proposed that COL4A1/A2 is required for the BM assembly process 30,31 , and while it appears that COL4A1/A2 is not required for BM assembly during early embryonic development, it is indispensable for the integrity of the BM at later embryonic stages 27 .  Supplementary Fig. 4).  [32][33][34][35][36] and VSMCs are assumed to be the principal source of vascular BM. In vitro studies have shown that cultivating VSMCs on BM-rich substrates causes upregulation of myofilaments resulting in a more contractile VSMC phenotype [36][37][38][39][40] . The present study is the first to support these findings in vivo, by demonstrating the dependency of VSMC differentiation on an adequate level of BM. We observed a clear reduction in VSMC differentiation markers in Col4a1/a2 +/− aortas in mice, and in human IMAs, COL4A1/A2 correlated with VSMC markers. Importantly, IMAs are musculo-elastic arteries and have different dimensions, properties and embryological origin as compared to abdominal aortas. Therefore, the use of IMAs rather than abdominal aortas is a limitation to this study.  The BM has been proposed to maintain VSMCs in a differentiated state by sustaining an environmental niche and simultaneously providing an adhesive substrate for VSMCs to anchor 6 . Depriving VSMCs from their niche will not only deny them of direct signals from the BM, but concomitantly increase exposure to structural components of the interstitial matrix having a different composition of diffusible factors all affecting VSMC modulation 7 . Ceteris paribus, it is conceivable that a shift away from a fully differentiated contractile VSMC phenotype in the normal aorta and throughout disease progression, because of a reduction in COL4A1/A2, could have major implications for AAA development. Phenotypic switching of VSMCs is a central process in AAA pathology [41][42][43] , and modulated VSMCs secrete both proteolytic enzymes that degrade ECM components and cytokines promoting vascular inflammation 44 , which may accelerate AAA formation. This notion is compatible with mutations of ACTA2 or MYH11, which display severe vascular consequences including patent ductus arteriosus and familial thoracic aortic dissection [45][46][47][48][49][50][51][52] .
Taken together, it is likely that an inadequate level of COL4A1/A2 renders VSMCs more prone to dedifferentiating at baseline and throughout AAA progression leading to the observed effect of AAA formation. This idea requires further clarification beyond the scope of the present study.
A limitation to our study is that the effect of Col4a1/a2 hemizygosity in mice may not be confined to the artery wall. This also applies to the modest association signal identified by our hypothesis-driven testing of the COL4A1/ A2 locus. Effects in all tissues comprising epithelia, peripheral nerves, fat-and muscle-cells, which are known to have a COL4A1/A2-containing BM cannot be excluded 53,54 . It is noteworthy, however, that both HANAC patients and mice carrying a missense mutation of Col4a1 first and foremost display a severe vascular phenotype 10,11,55 . This observation points to vascular tissue, as particularly susceptible to dysfunctional COL4A1/A2.
The histological analyses presented in this study show a COL4 distribution in advanced human AAA lesions compatible with increased degradation, in line with a previous report 56 , and furthermore show that MYH11positive VSMCs primarily remain in areas with dense COL4 staining. Although, the direction of causality (COL4 degradation leading to MYH11 reduction, or VSMC modulation leading to COL4 degradation) or the presence of a common underlying factor cannot be dissected from the simple analyses presented here, these observations are compatible with the notion that COL4A1/A2 reduction may be involved in VSMC dedifferentiation. Indeed, despite variation in the mouse AAA-proteomic data, Myh11 remained nominally reduced suggesting that the effect on VSMCs persists throughout lesion progression. A limitation to the presented histological analysis is that the ascending aorta was used for comparison. The ascending aorta is different from the abdominal aorta in several ways (dimensions, local hemodynamics and embryological origin). Moreover, the histological analyses were based on a pan-COL4 antibody, which does not distinguish COL4A1-COL4A2 from COL4A3-A6. Although this is a limitation to the analysis, unpublished observations from our group show that the COL4A1/A2 chains predominate in vascular tissues, while COL4A3-A6 are scarce if at all measurable. COL4A1/A2 degradation as a feature of AAA is further supported by our observation that C4M levels are augmented in plasma from AAA-patients. The COL4A1 neoepitope is well-characterized, and known to be formed after MMP-2/9-mediated degradation 57 . Plasma C4M was found to be independent of all relevant risk factors and in addition significantly correlated to maximal abdominal aortic diameter in the VIVA screening trial. This finding is supported by a previous report, where a non-specified Collagen IV epitope was elevated in AAA patients 56 . Interestingly, the COL4A1/A2 formation marker, PRO-C4 was not associated with either AAAdiagnosis or maximal aortic diameter further underlining MMP-2/9-dependent degradation as the principal mechanism resulting in elevated COL4A1 fragment plasma levels. Both tissue and plasma investigations suggest that the artery COL4A1/A2 amount may diminish during development of AAA, an aspect that may further accelerate disease development. Interestingly, several COL4A1/A2 degradation products have been shown to be biologically active. Arresten and Canstatin (degradation products of COL4A1 and -A2, respectively) have anti-angiogenic activity 7 , but whether these peptides play a significant role in AAA pathogenesis remains elusive. Recently, AXT107-a peptide derived from sequences of both COL4A1 and COL4A5 58 was proposed as a drug displaying strong anti-permeability effects on several vessel types 58,59 -an effect which could also be relevant in the context of AAA 60 .
In conclusion, we provide evidence that reduced baseline artery COL4A1/A2 content increases AAA formation, and that this effect is possibly mediated by mechanisms involving dedifferentiation of VSMCs. Moreover, we Table 3. Univariate and partial correlation analyses of the plasma C4M and PRO-C4 levels potential associations with maximal aortic diameter and aneurysmal growth rate. *Adjusted for smoking, hypertension, previous ischemic event, statins, and diastolic blood pressure. **Adjusted as above and baseline maximal AAA diameter. show that COL4 degradation is part of the AAA pathology and that COL4A1 degradation fragments are released into the circulation. This may assist future diagnostic and prognostic evaluation of AAA patients.

Data availability
The data underlying this article will be shared on reasonable request to the corresponding author.