Extracellular matrix, regional heterogeneity of the aorta, and aortic aneurysm

Aortic aneurysm is an asymptomatic disease with dire outcomes if undiagnosed. Aortic aneurysm rupture is a significant cause of death worldwide. To date, surgical repair or endovascular repair (EVAR) is the only effective treatment for aortic aneurysm, as no pharmacological treatment has been found effective. Aortic aneurysm, a focal dilation of the aorta, can be formed in the thoracic (TAA) or the abdominal (AAA) region; however, our understanding as to what determines the site of aneurysm formation remains quite limited. The extracellular matrix (ECM) is the noncellular component of the aortic wall, that in addition to providing structural support, regulates bioavailability of an array of growth factors and cytokines, thereby influencing cell function and behavior that ultimately determine physiological or pathological remodeling of the aortic wall. Here, we provide an overview of the ECM proteins that have been reported to be involved in aortic aneurysm formation in humans or animal models, and the experimental models for TAA and AAA and the link to ECM manipulations. We also provide a comparative analysis, where data available, between TAA and AAA, and how aberrant ECM proteolysis versus disrupted synthesis may determine the site of aneurysm formation.


Introduction
Aortic aneurysm is morphologically defined as a focal dilation and structural degradation of the aorta, which can occur at different sites along the aortic tree. Pathologically, formation and progression of an aortic aneurysm is driven by dysregulated cellular and acellular events. While thorough reviews on the cellular mechanisms and current medical management of thoracic aortic aneurysm (TAA) 1,2 and abdominal aortic aneurysm (AAA) [2][3][4] have been published elsewhere, this review focuses on the role of the aortic extracellular matrix (ECM) and the proteases and their inhibitors that maintain the integrity and homeostasis of the ECM in the pathogenesis of TAA or AAA in humans and animal models.

Aortic wall structure and the ECM
The aorta is the largest conduit artery in the body. Owing to its extraordinary ability to expand and recoil, the aorta also serves as a reservoir that transforms the highly pressured and pulsatile heart output into a flow of moderate fluctuations. The composition and threedimensional organization of its ECM is critical for optimal physiological functions of the aorta. The aortic wall comprises an intricate arrangement of endothelial cells (ECs), smooth muscle cells (SMCs), fibroblasts (FBs), and ECM proteins in three layers: tunica intima, tunica media, and tunica adventitia (Fig. 1). The tunica intima consists of a single layer of ECs that lines the lumen of the blood vessel, and is anchored to the underlying basement membrane, a highly specialized ECM network consisting primarily of laminin, collagen type IV, fibronectin, perlecan, and heparan sulfate proteoglycans 5 . This basement membrane also plays a pivotal role in signaling events that regulate EC migration, invasion, proliferation, and survival 6 . The basement membrane together with the internal elastic lamina (IEL) serves as an interface between the tunica intima and tunica media. IEL is a thin sheet of elastic fiber with a variable number of fenestrations that facilitate diffusion of molecules through the tunica intima to the tunica media 7 . The tunica media consists of concentric elastic lamellae with circumferentially oriented SMCs. This layer is rich in ECM proteins, such as proteoglycans (PGs), glycoproteins, glycosaminoglycans (GAGs), as well as collagen. The intermittent layers of elastic fibers and SMCs in this layer provide the compliance and recoil properties of the aortic wall 8 . The medial layer of the aortic wall is the most variable of the tunics among different arteries. The differences in ECM composition in this layer account for the main variations in the function and properties of the arterial tree. The external elastic lamina (EEL) forms a boundary between the tunica media and tunica adventitia, the outermost layer of the aortic wall that is rich in collagen, FBs, and elastic fiber, and provides the tensile strength of the aortic wall. This collagen-rich tunica also harbors innervations, lymphatics, and the vasa vasorum, which are responsible for the high susceptibility of the adventitial layer to vascular inflammation.
The ECM is a key constituent of the vascular wall. SMCs and the FBs in the aortic wall synthesize and organize a complex ECM structure that defines the passive mechanical behavior of the large, elastic arteries such as the aorta. Arterial ECM is primarily composed of elastin, collagen, proteoglycans (PGs), and glycoproteins. Elastic fibers and fibrillar collagen, comprising~50% of the dry weight of larger arteries 9 , are the predominant ECM components in the aortic wall and almost entirely define the mechanical properties of the aorta. Elastic fibers provide the expandability and recoil properties, while fibrillar collagens (predominantly collagen type I and type III) are responsible for the tensile strength of the aortic wall to withstand the high pressure of blood pumped by the heart 10 . In addition to their mechanical properties of the aortic wall, elastic fibers and fibrillar collagens can directly interact with various integrins and other proteins to modulate the adhesion, proliferation, and migration of SMCs [11][12][13][14] . GAGs consist of linear chains of repeating disaccharides and are highly negatively charged, which allow them to sequester water. Hydrated GAGs occupy a large volume of the extracellular space Fig. 1 Cross section of the aortic wall and its remodeling during physiological versus pathological remodeling. The tunica media is separated from the tunica intima by the internal elastic lamina, and from the tunica adventitia by the external elastic lamina. Intermittent layers of elastic fiber and smooth muscle cells make up the tunica media, while this region is also rich in proteoglycans and glycoproteins. During physiological remodeling, degradation of ECM proteins by MMPs (and other proteases) is balanced by newly synthesized ECM proteins, while tissue inhibitors of proteases (TIMPs) keep the proteolytic activity of MMPs under check. Excess ECM degradation, or impaired ECM renewal synthesis, along with smooth muscle cell death, lead to adverse ECM remodeling as observed in aortic aneurysm and cushion-compressive forces on cells. Among the four primary classes of GAGs, hyaluronan, chondroitin sulfate, heparan sulfate, and keratan sulfate, the first three are present in the aortic wall 15,16 . PGs are proteins with covalently attached GAG chains that provide mechanical support and mediate signaling responses for cells in the aortic wall. Versican, which is rich in chondroitin sulfate, is among the principal proteoglycans present in the aorta 17 . Among small leucine-rich proteoglycans (SLRP), biglycan and decorin are reported to promote collagen fibrinogenesis in the aorta 18,19 . Thrombospondins, tenascins, and fibronectin are also common glycoproteins present in the aorta 20,21 .

Regional heterogeneity of the aortic wall
Morphology, structural composition, and mechanical properties of the aortic wall vary in the ascending and along the descending regions 22 to ensure optimal mechanical operation 23 . The tunica intima accounts for <1% of the total wall thickness and remains uniform along the aorta, whereas the thickness of tunica adventitia is negligible in the thoracic aorta but increases considerably along the abdominal aorta 24 . The medial layer exhibits the greatest regional variations and diversitise along the aorta. The thickness of tunica media and the aortic lumen size steadily decrease from the ascending to the descending aorta, such that the wall thickness-to-lumen diameter ratio remains constant throughout the aorta. In the thoracic aorta, elastin occupies a higher proportion of the tunica media compared with that in the abdominal aorta. The amount of elastin decreases with distance from the heart, as less pulse dampening is required in the distal aorta. Meanwhile, the amount of collagen, which provides strength and limits stretch at high pressure, remains approximately constant with distance from the heart 25 . Therefore, the elastin-to-collagen ratio is lower in the abdominal compared with the thoracic aorta. In addition, there is a linear relationship between the number of lamellar units and the tensional forces within the aortic wall, and the number of lamellar units decreases from the thoracic to the abdominal aorta 10 .

Physiological remodeling of the ECM
The ECM composition and integrity are key determinants of the physical characteristics of the aortic wall. Vascular ECM undergoes continuous physiological remodeling, whereby the existing ECM proteins are proteolytically degraded to be replaced by newly synthesized proteins. Several protease families are involved in this process. Matrix metalloproteinases (MMPs) are the most explored ECM-degrading proteinases 26 , while other metalloproteinases such as ADAMs (a disintegrin and metalloproteinases), ADAM-TSs (ADAMs with thrombospondin motifs), or serine/cysteine proteases, including cathepsins and granzymes, can also contribute to this process. MMPs are a family of zinc-containing calcium-dependent endopeptidases that are best known for their ECMdegrading function, although they also contribute to a number of other cellular events 26 . To date, 23 human MMPs have been identified, which are classified into six groups based on their substrate specificities: collagenase, gelatinase, stromelysin, matrilysin, membrane-type MMPs, and others. MMPs are mostly secreted as inactive zymogens and are activated by proteolytic removal of their N-terminal pro-domain. The proteolytic activity of MMPs is tightly regulated in vivo by their endogenous inhibitors, tissue inhibitor of metalloproteinases (TIMPs) 27,28 . Other MMP inhibitors, such as α2-macroglobulin, inhibit MMPs in the plasma and liquid phase by irreversibly removing them through endocytosis 29 . The four members of the TIMP family (TIMP-1, −2, −3, and −4) can collectively inhibit all MMPs and a number of ADAMs and ADAM-TSs 30 through a covalent bond between their N terminus and the catalytic domain of the proteinase in a 1:1 stoichiometric ratio. In addition to inhibiting metalloproteinases, TIMPs play a number of MMP-independent functions [31][32][33][34] . Sustained balance in proteolytic turnover of the ECM proteins is critical in maintaining the structural and functional integrity of the aortic wall. Cathepsins, a family of serine, cysteine, and aspartyl proteases are lysosomal proteases that also possess extracellular functions and can cleave multiple ECM components, including fibronectin, laminin, and elastin, and a number of non-ECM components 35 . Granzymes are serine proteases that contribute to aortic remodeling and have been the subject of a recent review 36 .

Aortic aneurysm
Aortic aneurysm is characterized by a localized irreversible dilatation of the aortic lumen by ≥50% of its original diameter. The dilatation is induced by aberrant and adverse remodeling of the aortic wall, including endothelial damage in the intima, SMC loss and ECM degradation in the media, and ECM degradation in the adventitia. Aortic aneurysm affects all three layers of the blood vessel, while any dilation of the aorta that is limited to the outer wall is considered a false aneurysm or pseudo aneurysm 37 . Aortic aneurysm is asymptomatic, but can progress, gradually leading to aortic dissection or rupture, and as such is associated with critical morbidity and mortality in the aging population and accounts for 1-2% of total death in developed countries 38 . Although TAA and AAA display similar physical appearances, they each demonstrate distinct etiologies, pathological features, epidemiology, and risk factors.

Abdominal aortic aneurysm (AAA)
AAA represents dilation of the aorta in any portion of the infra-diaphragmatic aorta, including the most common site of AAA formation, the infrarenal aorta. Depending on the degree of dilatation, AAA severity is classified as small (<55 mm) or large (≥55 mm). In the latter case, the patient will be considered eligible for surgical repair 39 . AAA is the 12-15 th leading cause of death in the aged population, mostly males, over the age of 65 in many developed countries 4 . Almost 1-2% of all 65-year-old males have AAA, whereas 0.5% of women at 70 years of age are affected with AAA 40,41 . The overall mortality due to AAA rupture has been reported to be 65-85% 38 . However, AAA-related mortalities in both male and female populations are declining in developed countries 42 , with the United States and the United Kingdom showing the fastest decline in AAA mortalities in male (6.7-6.2% per year) and female (3.9-4% per year) populations. This was found to closely correlate with the decline in smoking prevalence in these countries. On the other hand, in many European countries, including Romania and Hungary, AAA mortalities have increased by 1.7-2.7% for males and 1-3.5% for females annually 42 .
AAA is a complex multifactorial disease that is affected by both environmental and genetic factors. AAA incidences depend on age and sex, with a higher prevalence in males over the age of 55 compared with females of the same age range (6:1) 43 . However, AAA appears more detrimental in females, as they usually experience a higher risk of a small aneurysm rupture and less eligibility for standard surgical repair, because of the more complicated aneurysm morphology 44,45 . Other risk factors include smoking, a family history of aneurysm 46 , the presence of other diseases such as coronary heart disease, hyperlipidemia and atherosclerosis, hypertension, as well as acute or chronic infection 4,38 . Tobacco smokers have 3-4-fold greater risk of developing AAA than nonsmokers 47,48 . The link between hypertension and AAA is paradoxical, since hypertension has been shown to increase the risk of AAA by 30-40%; however, antihypertensive medication further increased the risk by 70-80% 47 . More recently, a randomized placebo-controlled trial reported that antihypertensive medications did not impact AAA growth, despite effective lowering of BP 49 . Interestingly, patients with diabetes mellitus have shown decreased prevalence or progression of AAA 50 , which could be linked to the effects of antidiabetic medications. Hypercholesterolemia, increased body mass index, lack of exercise, and cardiovascular diseases were found to be associative factors for higher incidences of AAAs in Swedish male population 51 . Although atherosclerosis shows a close association with AAA, multiple studies have demonstrated the lack of a causal role for atherosclerosis in the initiation of AAA 52-54 .

Thoracic aortic aneurysm (TAA)
TAA represents dilation of one or more aortic segments above the diaphragm, including the aortic root, ascending aorta, aortic arch, and descending aorta. Approximately 60% of TAA occurs in the aortic root or the ascending aorta, 10% in the aortic arch, 40% in the descending aorta, and 10% in the thoraco-abdominal aorta 55 . About 13% of patients having aortic aneurysm develop multiple aneurysms, and~20% of them can have a large TAA along with an AAA 56 . The prevalence of TAA is much less than that of AAA 57,58 , but the prognosis of TAA is more dire, and the rate of rupture-associated mortality is 2-3 times higher than that for AAA patients. While AAA is more commonly found in elders, TAA can occur at a young age due to the strong hereditary influence 59 .
Similar to AAA, no causal relationship has been found between atherosclerosis and TAA 60 , and it is relatively less associated with ascending TAA 55 . TAA often results from genetic disorders, and is now referred to as syndromic or familial TAA. In a study of 135 non-syndromic TAA patients without Marfan syndrome (MFS), almost 19% of the patients were found to have a family history of TAA and developed the disease in relatively a younger age 61 . Moreover, patients with a family history of TAA were found to have a faster aneurysm growth rate (0.22 cm/year) compared to patients with sporadic TAA (0.03 cm/year) or patients with MFS (0.10 cm/year) 61 . The specific genetic mutations associated with TAA are discussed later in this review. TAA has also been linked to inflammation and inflammatory diseases. Syphilis was once considered as one of the most common causes of ascending TAA 55 , and although extremely rare nowadays, recent cases of syphilis aortitis and TAA have been reported 62 . Aortic arteritis, which includes Takayasu's arteritis and giant-cell arteritis, refers to a group of chronic inflammatory diseases of unknown etiologies, and is associated with a high prevalence of TAA 63,64 .
Recent clinical studies with large scale populations have identified fluoroquinolones, a commonly prescribed class of antibiotics as a risk factor for aortic aneurysm 65,66 . Fluoroquinolone usage elevated the risk of aortic aneurysm or dissection by 2-3 folds compared with the nontreatment group 67,68 , and by 2-fold compared with other antibiotics such as amoxicillin 69 . Fluoroquinolones are believed to affect the synthesis and structural integrity of collagen in the aortic wall 68 , or their chelating properties could affect collagen cross-linking and metalloprotease activities 70,71 . In a mouse model of AAA, ciprofloxacin (a fluoroquinolone) treatment significantly reduced lysyl oxidase (LOX) expression and activity, and increased MMP expression and activity 72 . However, the exact biological mechanism of action for this class of antibiotics requires further investigation.

Experimental models of aortic aneurysm
Understanding the cellular and molecular event during the initiation, maturation, and rupture of human aortic aneurysm holds the key to identifying molecule(s) that are critical in the pathogenesis of this disease. However, aneurysmal aorta specimens from patients are only available at the advanced stages of the disease and therefore are not suitable for identifying the initiating factors. As such, a number of experimental models of aortic aneurysm have been developed and utilized on various genetically modified mice to understand the molecular mechanism(s) underlying aortic aneurysm ( Table 1). Most of these experimental models share some of the features of human aortic aneurysm, although they all fall short of exactly mimicking the full clinical characteristics of this disease in humans 3,73 . Nevertheless, these models have helped provide insight into understanding the disease pathogenesis and can be utilized for the possible discovery of novel therapies for aortic aneurysm.
(i) Hyperlipidemic model The hyperlipidemic model of aortic aneurysm utilizes subcutaneous infusion of a high dose of angiotensin II (Ang II, 1.5 mg/kg/d) in hyperlipidemic mice, namely mice lacking apolipoprotein E (Apoe −/− ) or low-density lipoprotein receptor (Ldlr −/− ) receiving a high-fat diet 73,74 . Long-term high-fat diet (HFD) feeding induces aortic inflammation and atherosclerosis in these mice 75 .
This model is associated with medial phagocyte accumulation, medial dissection, elastic network degradation, and inflammatory cell infiltration 76 . The hyperlipidemic models of AAA share some features of obesity-induced vascular inflammation, atherosclerosis, and aortic aneurysm in humans.

(ii) Angiotensin II infusion model
Subcutaneous infusion of Ang II, in the absence of hyperlipidemia, is commonly used in mice and rats to induce cardiovascular remodeling 77 . This relatively easy and minimally invasive surgical model mimics some features of human aortic aneurysm, including ECM remodeling, SMC activation, and vascular inflammation 73 . When used in combination with β-aminopropionitrile (BAPN), an inhibitor of collagen cross-linking enzyme LOX, Ang II infusion can result in the formation of TAA or AAA, or aortic dissection 78,79 . Simultaneous use of BAPN and a TGFβ inhibitor has also been reported to induce AAA in mice 80,81 . These aneurysm models display an aggressive development of intraluminal thrombus, elastin fragmentation, and influx of a range of inflammatory cells to the aortic wall 81 . Being a time-dependent progressive model, Ang II-induced aneurysm can be utilized for preventive and/or therapeutic studies. Intriguingly, Ang II infusion alone, in the absence of BAPN or hyperlipidemia, triggered AAA in Timp3 −/− mice 82 , and TAA in Mmp2 −/− mice 83 . As discussed in later sections, these two genetic models exhibit the importance of ECM-regulatory proteins in sustaining the homeostasis of aortic wall structure and the regional susceptibility of the aorta to aneurysm formation.

(iii) Intraluminal elastase model
Intraluminal elastase perfusion model introduces porcine pancreatic elastase (PPE, 0.45 U/ml) into the infrarenal aortic lumen (at 100 mmHg, 5 min), which results in AAA formation within 2 weeks 84 . This model combines the mechanical strain with the enzymatic degradation of the elastic network in the aortic wall by PPE to induce injury, inflammation, and ultimately aneurysm 85 . This model is technically challenging as it requires the infrarenal aorta to be surgically separated from the inferior vena cava, and the lumbar and the juxtarenal arteries and the aortic bifurcation need to be temporarily ligated before the infrarenal aorta is punctured and infused with PPE 3 . As such, this model can only be used to induce AAA and not TAA since the close proximity of the thoracic aorta to the lungs and the heart drastically reduces the margin of error.
(iv) Periadventitial elastase model Periadventitial elastase model utilizes PPE to trigger degradation of the aortic elastic network by exposing the adventitia to elastase 86 . In this model, PPE (15 or 30 U/ mL) is directly applied onto the adventitial surface of the aorta for 5 to 10 min to induce AAA 86 or TAA 87,88 , respectively. This model is markedly less invasive than the intraluminal elastase model as it does not require aortic puncture nor pressure-induced mechanical damage. In addition, the adventitial elastase model can be used to induce AAA 86 or TAA 87,88 . Although intraluminal and adventitial elastase models result in morphologically distinct aortic aneurysms in mice and thus represent specific subtypes of this disease in humans 89 , these aneurysms do not display thrombus, atherosclerosis, or rupture, which are classical features of human aortic aneurysm. A recent study showed that a combination of oral administration of BAPN and periadventitial elastase application induced a chronic, advanced-stage AAA with persistent growth, thrombus, and spontaneous rupture 80 . This represents a promising animal model to better understand the pathogenesis of AAA and to test promising therapeutic compounds.
(v) Calcium chloride or calcium phosphate model This model requires application of concentrated calcium chloride (CaCl 2 ) 90 or calcium phosphate (CaCl 2 and PBS) 85 solutions onto the adventitial layer of the aorta to induce calcification, elastin degradation, and subsequently aneurysm formation in 4-8 weeks. This model is sitespecific and can be used to induce AAA and TAA 83 . Of note, CaCl 2 -induced TAA can be associated with a high operative mortality rate due to pulmonary complications upon exposure to CaCl 2 . In addition, the severity of aortic aneurysm in this model can be relatively mild.

(vi) Decellularized aortic xenograft model
This model involves surgical interspecies transplantation of a decellularized aorta segment, for instance from guinea pig to rat 91,92 . The underlying principle of this model is to differentiate the impact of the cellular component versus the ECM. In the absence of the histocompatibility complex and protease inhibitors that are synthesized by the cellular compartment of the aorta, the decellularized aortic xenograft becomes the target of immune reaction and proteolytic degradation of the ECM and progressive aortic dilation 92,93 . Similar to human AAA, this model develops intraluminal thrombus and aortic aneurysm and rupture 92 .

(vii) Large animal models
Large animal models of TAA and AAA are less diverse and less prevalently used compared to the rodent models, and have been proven useful for preclinical testing of endovascular devices or surgical procedures, rather than aneurysm pathogenesis. Intraluminal PPE perfusion has also been used in dogs, which, similar to the rodent model, leads to disruption of the medial elastic lamellae and aneurysm formation 94 . However, this model requires high concentrations of elastase and aneurysm formation is inconsistent. To avoid these issues, intra-aortic elastase and collagenase perfusion combined with aortic balloon angioplasty has been used 95 . In a pig model of TAA, intraadventitial injection of collagenase with periadventitial application of crystalline CaCl 2 resulted in elastic lamellar degradation, decreased collagen content, and TAA formation after 3 weeks 96 .
In summary, although each experimental model represents select characteristics of aortic aneurysm in the context of associated comorbidity(ies), and therefore can be useful in understanding the pathology of this disease. However, the interpretation of the findings should be exclusively based on the model used since contradictory results can be observed when the same animals are subjected to different models. For example, Mmp12-deficiency in mice suppresses CaCl 2 -induced AAA formation 97 , but does not affect elastase-induced AAA formation 84 .

ECM-regulatory proteins in aortic aneurysm
Aortic aneurysm involves dramatic changes in the structural components of the vasculature, due to SMC dysfunction, imbalance between proteinases and their inhibitors, and abnormal synthesis, accumulation, or degradation of ECM proteins 3 . Among different classes of proteases, metalloproteinases (MMPs, ADAMs, and ADAM-TSs) 83,88,98,99 , and cathepsins 100,101 are important factors for pathogenesis of aneurysm, because of their involvement in cleavage of different ECM and non-ECM proteins. The availability of genetically modified mouse models has paved the way to understanding the role of the ECM and its remodeling, the importance of protease and their inhibitors, and the contribution of different proteoglycans and glycoproteins in aortic aneurysm.

Matrix metalloproteinases
Multiple MMPs contribute to physiological aortic remodeling and their increased expression are well documented in both human and mouse aortic aneurysms 102 . Under physiological conditions, vascular SMCs (VSMC) are the major source of MMPs production in the aorta, which is amplified under inflammatory conditions leading to augmented expression of MMPs 103 . Inflammatory cells, such as neutrophils and macrophages, also contribute to MMP production which further promote ECM degradation and aneurysm formation. Neutrophils can produce MMP8 and MMP9 104 Studies on MMP-deficient mice have provided us with important insight into the role of specific MMPs in different models of aortic aneurysm. Mmp2-deficient mice showed reduced aortic aneurysm formations in CaCl 2 -induced TAA and AAA models 83,108 . However, in response to Ang II infusion, Mmp2-deficient mice developed severe TAA but not AAA 83 . The absence of MMP9 in mice attenuated AAA formations 84 ; however, the protection was reversed by introducing wild-type macrophages in Mmp9 −/− mice, but not in Mmp2 −/− mice, highlighting the role of macrophage MMP9 and stromal MMP2 in the pathogenesis of AAA 108 . Even in the absence of MMP2 or MMP9, MT1-MMP can be a key regulatory factor to induce aortic aneurysm by promoting macrophage-dependent elastolytic activity in CaCl 2 -induced AAA 109 . Apoe −/− /Mmp3 −/− mice showed reduced aortic aneurysm formation despite persistence of atherosclerotic plaques compared with Apoe −/− mice 110 . Loss of MMP12, also known as macrophage metalloelastase, attenuated aneurysm formation in CaCl 2 -induced AAA model 97 .
Although there is significant evidence that MMPs are upregulated in AAA and TAA, and MMP-deficiency often results in protection against aneurysm formation, MMP inhibition may not be an effective approach in treating aortic aneurysm. Consistently, broad-spectrum MMP inhibitors, such as doxycycline, were reported to have promising effects in treating small AAAs 111 . However, recent randomized trials reported no changes in AAA growth or in reducing the need for AAA repair or time to repair in patients receiving doxycycline 112 .

Disintegrin and metalloproteinases (ADAMs, ADAM-TSs)
A number of ADAM-TSs, including ADAM-TS4, ADAM-TS5, and ADAM-TS7, have been reported to be involved in TAA in humans and in mouse models 98,113,114 . Mice expressing truncated ADAM-TS5 (lacking the catalytic domain) showed increased aortic dilation in Ang II-induced TAA compared to wild-type mice 114 . ADAM-TS4 deficiency reduced aortic dilation, aneurysm formation, dissection and aortic rupture in a mouse model of sporadic aortic aneurysm and dissection (AAD) induced by HFD and Ang II infusion 98 . However, ADAM-TS1 overexpressing Apoe −/− mice subjected to HFD and Ang II infusion did not show elevated aneurysm response compared to Apoe −/− mice 115 .
Compared with ADAM-TSs, the role of ADAMs in aortic aneurysm is less explored. The most studied ADAM, ADAM17, was found to be critically involved in aortic aneurysm development in both humans and mice 88 . SMC-or EC-specific deletion of ADAM17 protected the aorta from elastase-induced TAA formation 88 , whereas ADAM17 deficiency in SMCs provided protection against AAA triggered by Ang II + BAPN treatment 116 . ADAM10 is an important regulator for vascular remodelling and inhibition of ADAM10 mRNA with miR-103a prevented AAA formation in Apoe −/− mice following HFD and Ang II treatment 117 .

Cathepsins
Cathepsins are a group of serine, aspartic, and cysteine proteases, which contribute to vascular remodelling. Several cathepsins, including cathepsins A, D, S, K, and L are reported to be involved or upregulated in humans aortic aneurysmal tissues 100,101,118 . Cathepsin K and S are potent elastases that are produced by SMCs and macrophages in the aorta under inflammatory milieus. Deficiency of cathepsin S attenuated Ang II-induced AAA formation in Apoe −/− mice 119 . Cathepsin K has been reported to contribute to CD4 + T-cell proliferation and SMC apoptosis since its deficiency attenuated the AAA in the intraluminal elastase infusion model in mice 120 . Cathepsin L knockout mice exhibited reduced lesions, in vitro macrophage and T-cell transmigration, and angiogenic responses upon intraluminal elastase-induced AAA development 121 .

Other proteases
There are several other proteases that are important in aortic aneurysm formation. Chymase and plasminogen activators are serine proteases that are involved in activation of other proteases, including MMPs. Chymases, secreted mostly by mast cells, activate MMP9 in human AAAs 122 . In addition to converting plasminogen into plasmin, plasminogen activators can directly degrade ECM components, activate collagenases, and degranulate neutrophils to release MMPs and elastases 123,124 . Deficiency of urokinase-plasminogen activator, but not tissueplasminogen activator suppressed aortic aneurysm formation in hypercholesteremic Apoe −/− mice 125 . Plasminogen deficiency also attenuated CaCl 2 -induced AAA by targeting macrophage migration and MMP9 activation 126 . Alternatively, local overexpression of plasminogen activator inhibitor-1 (PAI-1) prevented aneurysm development and rupture in mice 127 . Granzyme B is a proapoptotic serine protease that is abundantly expressed in advanced human atherosclerotic lesions and may contribute to plaque instability 128 . Granzyme B contributes to AAA through an extracellular, perforinindependent mechanism involving ECM cleavage. The absence of Granzyme B decreased AAA formation and increase survival of Apoe −/− mice upon Ang II infusion 129 .

Tissue inhibitors of metalloproteinase (TIMP)
TIMPs regulate the proteolytic activities in the vascular wall by directly binding to and inhibiting different groups of metalloproteinases, including MMPs, ADAMs, and ADAM-TSs. Although there is a high degree of structural homology among the four members of the TIMP family, they exhibit varying substrate affinities, inhibitory efficiencies, and differential responses to pathological stimuli 130 . TIMP1 can inhibit most members of the MMP family except the membrane-type MMPs (MT-MMPs) and MMP19, and few ADAMs, such as ADAM10 130 . TIMP1 expression is reduced in aneurysmal tissues, and the absence of TIMP1 resulted in severe aneurysm in the intraluminal elastase-perfused AAA 131 and CaCl 2induced TAA mouse models 132 . Local overexpression of TIMP1 prevented aortic aneurysm degeneration and rupture in a rat model 133 . TIMP2 which acts as an inhibitor for many MMPs 134 and ADAM12 135 , also plays an important role in MMP2 activation in coordination with MT1-MMP 136 . TIMP2 deficiency resulted in protection against CaCl 2 -induced AAA in mice 137 , perhaps by inhibiting MMP2 activation. TIMP3 has the broadest range of substrates among TIMPs which includes most of the MMPs, many ADAMs (ADAM −10, −12, −17, −19, and −33) and ADAMTSs (ADAMTS-1, −2, −4, and −5) 28 . Moreover, TIMP3 is distinct from other TIMPs in being ECM-bound 138 . Increased expression of TIMP3 in aneurysmal aortic tissue suggests its protective role against the disease 139 , as the absence of TIMP3 promoted aortic dilation and aneurysm in Ang II-induced (nonhyperlipidemic) AAA 82 . TIMP4 can inhibit a number of MMPs and ADAM28 140 . Interestingly, the N-terminal domain of TIMP4, but not the full-length protein, can inhibit ADAM17 141 . The role of TIMP4 in aortic aneurysm is less explored. Increased TIMP4 expression has been reported in hyperhomocysteinemia-associated aortic aneurysm in humans and mice 142 .

ECM proteins that are central to aneurysm formation
Elastin Elastin is one of the principle ECM molecules synthesized by SMCs in the aorta in response to mechanical stress or pressure, and is formed as a complex of elastin protein assembled on a microfibril platform. Elastic fibers provide the aortic wall with a unique ability to expand and recoil that is essential for optimal blood perfusion throughout the body. Fragmentation of elastic fibers is a common feature of aortic aneurysm independent from the initiating factors, in both humans and animal models. Deletion of the elastin gene (Eln) in mice resulted in an uncontrolled proliferation of SMCs and perinatal death due to high blood pressure and tortuous, stenotic arteries. While elastin haploinsufficiency (Eln +/− ) improved survival, it resulted in supravalvular aortic stenosis 143 .

Collagen
Fibrillar collagens type I and type III account for 80-90% of the total collagen present in the aorta, while collagens type IV, V, VI, and VII represent the remaining fraction of collagens. Collagen turnover is critical for vessel wall repair and regeneration. In addition to its contribution to the vascular structure and tensile strength, it can regulate cell proliferation through interacting with integrins 14 . In humans, higher levels of collagens type I/III and collagen cross-linking are reported in aneurysmal aortas 144 , as increased collagen can enhance arterial stiffness and susceptibility to dissection and rupture. On the other hand, decreased collagen content and cross-linking can weaken the aortic wall, leading to aneurysm formation and/or aortic dissection 144 . The disparity in collagen content might reflect different phases of aortic remodeling, with fibrosis occurring at the late phase of inflammation during vessel repair. This also highlights the importance of sustaining a balance in collagen content for optimal aortic structure and function. Deletion of the first intron of Col1α1 gene resulted in an age-dependent aortic dissection and rupture due to gradual decrease in collagen content 145 . Haploinsufficiency of Col3α1 caused aortic dissection in mice and a pathology similar to the vascular-type Ehlers-Danlos syndrome (VEDS) 146 . Appropriate cross-linking of collagen monomers by LOX is critical for collagen fibril formation 147 , as inactivation of LOX, through mutations or gene deletion, can lead to TAA, cardiovascular dysfunction, and perinatal death in mice [148][149][150] .

Proteoglycans/glycoproteins
Proteoglycans and glycoproteins comprise the nonfibrillar fraction of the ECM, which fills the extracellular space that is not occupied by the fibrillar ECM, and interacts with various molecules (growth factors, cytokines, etc.) to mediate their sequestration within the ECM. Several proteoglycans and glycoproteins have been reported to be upregulated during aortic inflammation and aneurysms 151 . The proteoglycans in the aortic wall mainly include large proteoglycans, such as versicans and aggrecans, or SLRP such as decorin, lumican, biglycan, etc 152 . These proteins are also involved in developing the pericellular matrix and promoting proliferation and migration of VSMC 153 . Versican is present in the intimal and medial layers of the aorta and is mostly expressed at the external part of the medial layer during vascular inflammation 17,154 . TAA and dissections are reported to have massive aggrecan and versican accumulation 155 ; however, the levels of cleaved versican (V0 isoform) are decreased 17 . Biglycan is expressed throughout the aortic wall and regulates collagen fibrillogenesis 156 . Biglycan knockout mice develop spontaneous aortic dissection and rupture 18 . Similar to biglycan, decorin promotes collagen assembly and can be cleaved by Granzyme B, which has been associated with aortic aneurysm formation in mice 128 . Moreover, recombinant decorin fusion protein expression in mice attenuates Ang II-induced AAA formation and rupture 19 .
Thrombospondins, a family of secreted glycoproteins with antiangiogenic functions, are highly expressed in the aortic wall under inflammatory conditions. Thrombospondin-1 is reported to negatively regulate cell adhesion, migration, proliferation, and angiogenesis 157 . SMC-specific deletion of thrombospondin-1 attenuates TAA formation by restoring the disruption of the elastic lamina-SMC connections and preserving the mechanotransduction in mice and humans 21 . Thrombospondin-4 regulates vascular inflammation and atherogenesis 158 , and its deficiency can lead to aortic aneurysm formation following 3 weeks of Ang II infusion 159 . Fibronectin is a prevalent glycoprotein in the aortic wall, and its expression has been reported to be elevated in aneurysmal aorta from patients with bicuspid aortic valve and tricuspid aortic valve stenosis 160 . Recent studies have reported that impaired splicing of fibronectin is associated with TAA formation 161 . Tenascins are glycoproteins highly expressed in the aorta. Loss of tenascin C led to AAD in a combined CaCl 2 and Ang II infusion model in mice, which was linked to enhanced inflammation, impaired TGFβ signaling, and collagen synthesis 162 .

Aortic aneurysm associated with mutations in ECMregulatory genes in humans
Over the last few decades, clinical studies have revealed a growing connection between mutations in ECM proteins and aortic aneurysm in humans ( Table 2). Many of these mutations are well known, as they underlie the heritable syndromic diseases such as MFS, Loeys-Dietz syndrome (LDS), and Ehlers-Danlos syndrome (EDS); others are mostly responsible for non-syndromic aneurysm-related disorders 163,164 . MFS is one of the most extensively studied autosomal dominant disorder of the connective tissues. MFS affects the skeletal, ocular, and cardiovascular systems and is often associated with the development of TAA 165 . The incidence frequency of MFS is around 1:5,000 and it is implicated in 3-5% of all aortic dissections 166 . MFS is the result of mutations of the fibrilin-1 gene (Fbn1). Fibrillin-1 is a large, ECM glycoprotein that serves as a structural component of calciumbinding microfibrils 167 that is central to elastic fiber assembly. Mutation of Fbn1 impairs elastic fiber structure and promotes dilatation of the aortic root and the ascending aorta in the affected patients, causing aortic dissection and hemorrhage which can lead to sudden deaths 165,167 . Fibrillin-2 (Fbn2) is mostly associated with aortic development, and plays a major role during early morphogenesis in directing elastic fiber assembly and is found preferentially in elastic tissues, such as the cartilage, the tunica medial layer of the aorta and along the bronchial tree 168 . Mutations in Fbn2 cause congenital contractual arachnodactyly, an autosomal dominant syndrome associated with aortic root dilation and aneurysm 169 . Fibrillin proteins exhibit marked structural homology to latent TGFβ binding protein, and are believed to influence TGFβ signaling in MFS 163,170 . However, the importance of TGFβ signaling in aortic aneurysm is independently recognized in another human aortopathy, LDS. LDS is an autosomal dominant disorder characterized by aggressive aortic root and TAAs where multiple family members are responsible for different types of LDS 171 . All five LDS variations (LDS1-5) are characterized by respective mutations of the following genes, Tgfbr1, Tgfbr2, Smad3, Tgfb2, and Tgfb3. The mutations present in Tgfbr1 and Tgfbr2 are principally missense and mostly occur in the kinase domains of the receptors 172 .
VEDS is an autosomal dominant trait, which results from heterogenous mutation of type III procollagen gene (Col3α1) leading to frequent arterial dissections with TAA. VEDS is one of the most prevalent vasculopathies with an incidence rate of 1:5,000-20,000 in developed countries 173 . In the aorta, collagen supports and maintains the tensile strength and stiffness of the aorta for which collagen maturation and cross-linking are important. For proper assembly of collagen alpha mono-trimers into triple-helix collagen, it is important to have the

Lox1
Loss-of-function missense mutation TAA and dissection 149,150

Tgfβr1
Heterozygous mutations Loeys-Dietz syndrome type 1 190 Tgfβr2 Heterozygous mutations Loeys-Dietz syndrome type 2 190,191 repetitive presence of Gly-X-Y amino acid sequences in all three alpha chains. In VEDS, however, the glycine is replaced by other amino acids, which disrupts the formation of collagen alpha trimers and subsequently, intact collagen fibrils 174 . Individuals with VEDS are at risk of arterial rupture, aneurysm and dissection, uterine rupture during pregnancy, and gastrointestinal perforation or rupture which could cause sudden death 164 . Mutations in collagen genes can also induce aortic dysfunction.
Mutations of Col1α1 and Col1α2 are reported in patients with osteogenesis imperfecta with rare aortic aneurysm 175 . However, Col1α1 mutation has been associated with dissections in medium-sized arteries and Col1α2 mutation with borderline aortic root enlargement with regurgitation in patients 175,176 . Collagen IV is a major component of the basement membrane in various tissues including the aorta, and mutation of Col4α1 gene has been linked to hereditary angiopathies, including nephropathy and aneurysm 177 . Mutation of Col4α5 is associated with X-linked Alport syndrome and ascending aortic and AAA and dissection 178 . A missense mutation in the Lox gene has been linked to TAA and aortic dissection because of insufficient elastin and collagen cross-linking in the aortic wall 149 .
Elastin is a key structural component in cardiovascular development, vascular elasticity and structural integrity of the aorta. Decreased elastin in humans is associated with development of an autosomal dominant disease, supravalvular aortic stenosis (SVAS). SVAS results from loss-of-function mutations in the elastin gene and is associated with aortic stenosis, hypertension, and cardiac failure 179 .

Conclusion/future prospects
In conclusion, aortic aneurysm remains a serious health concern as its rupture can cause significant morbidity and mortality. It is well acknowledged that TAA is a distinct disease from AAA and should be treated as such.
Research to date suggests that given the regional heterogeneity of the aortic structure and ECM, events that disrupt ECM synthesis or ECM protein assembly could underlie the formation of aneurysm in the thoracic area, whereas factors that trigger enhanced proteolytic degradation of the ECM contribute to AAA formation. This is consistent with the notion that genetic disorders that disrupt collagen or elastin production (or assembly) are associated mostly with TAA and to a smaller extent with AAA. While studies on genetically modified mice have revealed a number of potential molecular and cellular mechanisms that may contribute to TAA or AAA, the lack of availability of an effective pharmacological treatment indicates that we may not yet have the full picture of all the factors that contribute to aortic aneurysm. Identifying the initiating factors in aneurysm formation is key in developing a treatment strategy, since at later stages, to repair the severe damage on the aortic wall if possible, would require effective replenishment of ECM-producing cells and ensuring optimal assembly and organization of the newly synthesized ECM proteins.