|Home | About | Journals | Submit | Contact Us | Français|
Recent findings on the role of fibulin-5 (Fbln5) have provided substantial progress in understanding the molecular mechanism of elastic fiber assembly in vitro. However, little is known about differential roles of fibulins in the elastogenesis of blood vessels. Here, we generated double knockout mice for Fbln5 and Fbln2 (termed DKO) and examined the role of fibulins-2 and -5 in development and injury response of the blood vessel wall.
Fibulin-2 is distinctly located in the subendothelial matrix, whereas fibulin-5 is observed throughout the vessel wall. All of the elastic laminae, including the internal elastic lamina (IEL), were severely disorganized in DKO mice, which was not observed in single knockout mice for Fbln2 or Fbln5. Furthermore, DKO vessels displayed upregulation of vascular adhesion molecules, tissue factor expression and thrombus formation with marked dilation and thinning of the vessel wall after carotid artery ligation-injury.
Fibulin-2 and fibulin-5 cooperatively function to form the IEL during postnatal development by directing the assembly of elastic fibers, and are responsible for maintenance of the adult vessel wall after injury. The DKO mouse will serve as a unique animal model to test the effect of vessel integrity during various pathological insults.
The internal elastic lamina (IEL) is located beneath the endothelium of blood vessels and forms the innermost elastic lamina. The IEL provides elasticity and recoil to the vessel wall, as well as functions as a physical barrier against chemical/mechanical stresses, preventing direct contact of plasma components to smooth muscle cells (SMCs). Several pathologic processes have been shown to bring about disruption of the IEL. For example, increased mechanical forces and shear stress associated with angioplasty, enzymatic activation of matrix degrading enzymes in atherosclerosis, and abdominal aneurysms are all associated with disruption of the IEL and progression of vascular disease 1–3.
Molecular mechanisms of elastic fiber assembly have begun to be explored by the identification of elastic fiber-associated proteins and their biochemical interactions with elastin and/or the microfibrillar scaffold 4. Members of the fibulin family of extracellular matrix (ECM) proteins, particularly fibulin-4 and fibulin-5, play essential roles in elastic fiber development 5–7. We demonstrated that fibulin-5 preferentially binds the monomeric form of elastin, but not polymerized elastin 8. Others have shown that fibulin-5 accelerates the self-aggregation process of elastin, called coacervation 9, and fibulin-5 limits maturation of the coacervated elastin 10. Among five of the known fibulins tested, fibulin-2 and fibulin-5 exhibit the highest binding affinity to elastin 11. Biochemical interaction assays showed that fibulin-2 also binds numerous basement membrane (BM) proteins including nidogen, laminin and fibronectin 12. Recently, we have generated knockout mice for the fibulin-2 gene (Fbln2−/−) and unexpectedly found that Fbln2−/− mice do not display significant alterations in elastic fibers in vivo, despite strong tropoelastin binding in vitro 13. In addition, Fbln2−/− embryonic fibroblasts retained the ability to assemble normal fibers of elastin, fibrillin-1 and fibronectin in vitro. Thus, precise roles of fibulin proteins during the assembly process and how they coordinate elastogenesis in different anatomical locations in vivo remain unknown.
In this study, we analyzed the interaction between fibulin-2 and fibulin-5 in vitro and tested their roles in vascular elastogenesis by generating double knockout mice (DKO) for Fbln2 and the fibulin-5 gene (Fbln5). Finally, we examined the effect of disrupted elastic laminae on vessel remodeling using a carotid artery ligation model in DKO mice.
Detailed information is provided in Supplemental material.
Aortae were harvested following cardiac perfusion with 3% glutaraldehyde in 0.1 M sodium cacodylate (pH 7.4) and prepared for electron microscopic analysis as described in Supplemental material.
Recombinant mouse fibulin-2 and fibulin-5 were produced as previously described 11. Recombinant human tropoelastin was kindly provided by Dr. Joel Rosenbloom (University of Pennsylvania). Detailed methods of solid phase binding assays and surface plasmon resonance assays are provided in Supplemental material.
Adult male mice between 3 to 8 months of age were used in the study. Ligation of the left carotid artery was performed as previously described 14. Detailed methods are provided in Supplemental material.
Histological sections at level 400 (1.4 mm from the ligature) were digitally captured using Leica DM2000 microscope for comparison between different genotypes. Morphometric analysis was performed with Scion NIH IMAGE Software (National Institutes of Health, Frederick, MD).
Five to six carotid arteries, unligated or ligated, were pooled and RNA was prepared as described in supplemental material.
Data were analyzed using 1-way ANOVA with Bonferroni post hoc tests, t-test or χ2 analyses and a p-value less than 0.05 (p<0.05) was considered statistically significant. Bars indicate the means ± SEM unless noted otherwise.
Fbln5−/− mice have been established as a mouse model of systemic elastinopathy, involving the vascular system as a major target organ 5, 7. In adults, Fbln5−/− aortae are elongated and tortuous due to the lack of continuous elastic fibers. However, the defect is not homogeneous throughout the thickness of vessel wall. Disruption of elastic fibers becomes progressively worse toward the adventitia and the IEL is relatively well formed 7. To evaluate the role of fibulin-5 in the development of the IEL, we used electron microscopy (EM) to examine aortae from postnatal day (P) 1 and P120 wild-type and Fbln5−/− mice. In wild-type aortae, elastic fibers at P1 were not yet continuous, indicating that elastic laminae were being organized during the neonatal period in the wild-type aorta (Supplemental Figure IA, please see www.ahajournals.org). Consistent with our previous observations, there was a marked delay in the formation of all elastic laminae in Fbln5−/− mice (Supplemental Figure IB). By P120, however, the surface of the IEL subjacent to the endothelial cells (ECs) was relatively well developed in Fbln5−/− mice, similar to that of wild-type aortae (compare asterisk in Supplemental Figures IC and ID). In contrast, the surface of the IEL adjacent to the first SMC layer and the elastic laminae in the rest of the media remained disrupted, with multiple aggregates of elastin being observed (arrows in Supplemental Figure ID). These findings led us to hypothesize that another molecule with a similar biological activity may have compensated for the absence of fibulin-5 in the formation of the IEL.
Although fibulin-2 interacts with tropoelastin with high affinity in vitro 11, a recent knockout study has shown that fibulin-2 is dispensable for elastic fiber development in vivo 13. We speculated that the loss of fibulin-2 was compensated by fibulin-5, and therefore, that fibulin-2 may be able to compensate for the loss of fibulin-5 in IEL formation. We first tested if fibulin-2 and fibulin-5 physically interact since both fibulins form homodimers in vitro 8, 15. Solid phase binding assays indicated that both fibulin-2 and fibulin-5 strongly interact with immobilized tropoelastin (Figurea 1A, 1B). Although soluble fibulin-2 showed weak binding to immobilized fibulin-5 (Figure 1A), almost no binding was detected between fibulin-2 and fibulin-5 using the reverse experiment (Figure 1B). This was confirmed by surface plasmon resonance assays where the calculated dissociation constant (Kd) between fibulin-2 and fibulin-5 was100 nM when fibulin-5 was immobilized, and 1650 nM when fibulin-2 was immobilized. Although fibulin-2 and fibulin-5 bound tropoelastin equally well in solid phase binding assays, different Kd values were calculated for the two binding interactions by plasmon resonance assays (Supplemental Table 2). The lower Kd value determined for the binding of tropoelastin to fibulin-2 than fibulin-5 was due to the higher dissociation rate constant of the fibulin-2 and tropoelastin interaction. These data from the two different binding assays indicate that fibulin-2 and fibulin-5 are unlikely to interact in vitro.
Since the IEL continues to develop after birth, we speculated that fibulin-2 might compensate for the absence of fibulin-5 in establishing the IEL in Fbln5−/− mice. We first examined if there was a compensatory upregulation of fibulin-2 in the Fbln5−/− aorta by Western blot analysis (Figure 2A). The bands corresponding to fibulin-2 (arrowhead and asterisk), which were absent in Fbln2−/− (also see Supplemental Figure I for deletion of Fbln2), were modestly upregulated in Fbln5−/− aortae compared to wild-type aortae (Figures 2Aa). Interestingly, fibulin-5 was significantly upregulated in Fbln2−/− aortae compared to wild-type (Figures 2Ab, 2Ac). To further test the hypothesis that fibulin-2 compensates for fibulin-5 in the formation of IEL, we generated double knockout mice for Fbln2 and Fbln5.
In agreement with recently reported findings 13, Fbln2−/− mice were healthy and fertile and indistinguishable from wild-type littermates. DKO mice were viable and exhibited general elastic fiber defects as seen in Fbln5−/− mice. Sudden death or premature death was not observed among DKO mice and neurological defects were not detected. Histological analysis with Hart’s staining showed short, severely disrupted elastic fibers in the dermis, lungs and aorta as was described for Fbln5−/− mice (data not shown). No obvious difference was seen between Fbln5−/− and DKO tissues at the light microscopic level (data not shown).
To confirm the absence of fibulin-2 and fibulin-5 in the DKO aorta, immunohistochemistry was performed using anti-fibulin-2 and anti-fibulin-5 antibodies (Figure 2B). Consistent with previous data 11, 16, fibulin-2 localized most strongly to the IEL and fibulin-5 was observed throughout the aortic wall of adult wild-type mice with intense staining (Figures 2Ba, 2Be). Localization of fibulin-2 and fibulin-5 was not altered in single Fbln5−/− and Fbln2−/− mice, respectively (Figures 2Bc, 2Bf), and both proteins were absent from DKO (Figures 2Bd, 2Bh). Taken together, these data indicate that protein localization of fibulin-2 and fibulin-5 are independent of each other in vivo.
We next examined the sub-lumenal region of the aorta at the ultrastructural level. In the wild-type aorta, a solid IEL was formed under the EC layer and a small extracellular region was seen between IEL and ECs (Figure 3A). The basement membrane of the ECs was tight to the endothelium and situated close to the underlying IEL. The Fbln2−/− IEL was indistinguishable from wild-type (Figure 3B). In the Fbln5−/− aorta, the surface of the IEL adjacent to the ECs was solid, however, small disruptions of the IEL were observed on the SMC side of the IEL (arrowheads in Figure 3C). In contrast to wild-type aortae, the BM in Fbln5−/− animals was separated from the endothelium and the subendothelial region was wider (arrows in Figure 3C). In DKO aortae, we observed two remarkable abnormalities. First, the IEL was markedly disrupted with core of the elastic lamina being severely thinner than that from either the wild-type or single knockout animals (asterisk in Figure 3D), suggesting that fibulins-2 and -5 aid in the assembly of IEL. Second, the subendothelial region was increased significantly and the BM was clearly visible and separated from the endothelium and underlying IEL (arrows in Figure 3D). Two possibilities can be suggested from this observation: 1) changes in the subendothelial ECM occur after the formation of the IEL and BM which influence the eventual organization of these structures and the association of the EC with the underlying matrix, and/or 2) lack of fibulins-2 and -5 in subendothelial matrix affects the stabilization of EC-ECM interactions, since fibulin-2 binds numerous BM proteins and mouse fibulins-2 and -5 each contain a RGD motif that mediates RGD-dependent integrin binding.
To determine if mislocalization or altered expression of BM proteins is involved in alteration of subendothelial ultrastructure, we examined BM proteins in DKO mutants. Using immunofluorescence staining, the EC layer was visualized with CD31 (Figure 4A, B). Lamininγ1 staining was observed uniformly throughout the vessel wall in the wild-type aorta (Figure 4C). In contrast, the staining was increased at the lumenal surface of the aorta in the DKO mouse (Figure 4D). Collagen IV staining was slightly increased in the DKO aorta at the lumenal surface when compared to the wild-type aorta (Figures 4E and 4F). Next, we examined if the structural changes of IEL led to activation of ECs by staining for the adhesion molecules, ICAM-1 and VCAM-1. These molecules are known to be upregualted in atherosclerosis and other pathological insults 17. As shown in Figure 5G and 5H, ICAM-1 staining was more intense in DKO compared to wild-type vessels, suggesting that DKO ECs were affected by the disrupted contact with the BM. VCAM-1 staining was unchanged in DKO aorta (data not shown). We then asked if the altered composition of ECM proteins affected differentiation of vascular SMCs. Staining for SM myosin heavy chain and α-SM actin, a marker of late and early differentiated SMCs, respectively, was indistinguishable between wild-type and DKO aortae, although the alignment of SMCs was disrupted and the number of lamellar units was increased in DKO aorta (Supplemental Figure III).
Finally, we determined whether a compromised IEL in addition to disrupted medial elastic laminae would further affect the response to vascular injury and/or vessel remodeling in vivo by employing a carotid artery ligation-induced injury model. The left carotid artery was ligated proximal to the bifurcation and maintained for 28 days, a time-point when neointima formation has become most prominent 18. Serial transverse sections were analyzed from 1.0 mm proximal to the ligature (designated as level 0) to 1.9 mm (level 900), and morphometric analysis was performed at level 400. Elastin staining of unmanipulated vessels from all genotypes did not reveal any differences in a vessel diameter (Supplemental Figure IV). Upon injury, wild-type and Fbln2−/− arteries showed little neointima and the elastic laminae exhibited a typical undulating structure (Figures 5Aa, Ab, Ba, Bb). Consistent with previous observations 19, Fbln5−/− vessels developed a severe neointima (Figure 5Ac). In contrast, despite a modest neointima being formed, DKO vessels developed a severe, organized thrombus that occupied an abnormally enlarged lumen (Figure 5Ad). Elastic laminae in both Fbln5−/− and DKO vessels were distended as indicated by the absence of undulations (Figures 5Bc, Bd). Whereas both wild-type and Fbln2−/− vessels underwent constrictive (negative) remodeling after 28 days (Figure 5C, WT and Fbln2−/−), Fbln5−/− vessels showed minimal negative remodeling (Figure 5C, Fbln5−/−) and the IEL perimeter was only marginally decreased. Remarkably, the DKO vessels showed an enlarged lumen (outward remodeling), greatly exceeding the original perimeter of unmanipulated vessels (Figure 5C, DKO). Comparisons of medial wall thickness among genotypes revealed that the media was extremely thin in vessels from DKO mice after injury (Figure 5D). The remodeling of post-injured carotid arteries assessed following perfusion fixation showed similar results to those obtained with immersion fixation only (Supplemental Figure V). These data indicate that DKO vessels were unable to undergo injury-induced vascular remodeling.
When the intima/media ratio was compared, DKO vessels showed less neointima but a marked increase in thrombus formation compared with wild-type vessels (Figure 5F, marked in red). Seven of eight vessels from DKO animals developed thrombus that occupied more than 50% of the lumen, whereas only one of six vessels from Fbln5−/− mice developed mixed a lesion consisting of thrombus and neointima (P = 0.026, χ2, Supplemental Figures VIA and VI3B). One DKO vessel developed severe neointima as Fbln5−/− vessels, but the vessel diameter was even more increased in the DKO vessel compared to Fbln5−/− vessels (Supplemental Figures VIC and VID).
It has been shown that positive remodeling is associated with structural changes of the media and adventitia, including medial and adventitial breakdown, together with plaque components in a rabbit vascular atherothrombosis model 20. Therefore, we evaluated the changes in the adventitia in all genotypes. Whereas no difference was detected in adventitia thickness among unmanipulated vessels (data not shown), adventitia area was significantly increased in injured DKO vessels. The ratio between adventitia to total vessel area, however, was unchanged in DKO. In contrast, the ratio was significantly increased in the injured Fbln5−/− vessels (Figure 5E).
To gain insight into the pathological changes that lead to thrombus formation in DKO vessels, we harvested wild-type and DKO vessels at 2 days and 7 days after ligation and examined the expression of vascular adhesion molecules by immunostaining (Figure 5G). PECAM-1 was downregulated in both wild-type and DKO vessels 2 days after injury compared to contralateral vessels (Figures 5Ga–5Gd). On day 7, PECAM-1 expression was regained in ECs of wild-type injured vessels but to a lesser extent in DKO vessels. PECAM-1 was also detected in the forming thrombus in DKO vessels (Figure 5Gf arrows). ICAM-1 was not detected in wild-type injured or contralateral vessels (Supplemental Figures VIIa, VIIc, VIIe and VIIg), however, expression was observed in ECs of injured DKO vessels and contralateral vessels at 2 days (Supplemental Figures VIIb, VIId, VIIf, and VIIh). VCAM-1 was upregulated in the DKO vessels at 2 days after injury and the expression was much stronger and extended to the medial layers at 7 days (Figures 5Gj, 5Gn) compared to wild-type injured vessels (Figure 5Gi and 5Gm). We finally asked if the DKO injured vessels affected the expression of tissue factor (TF), which is a key molecule involved in extrinsic coagulation pathway and shown to mediate arterial injury-induced thrombosis 21. qPCR analysis of uninjured carotid arteries from DKO mice showed significantly lower expression of TF transcripts compared to wild-type arteries, whereas von Willebrand factor (vWF), a key molecule in the intrinsic coagulation pathway, showed similar expression (Supplemental Figure VIII). Interestingly, however, carotid arteries harvested at 2 days post-injury from DKO mice showed upregulation of TF compared to the wild-type injured arteries. Transcripts for vWF were comparable between wild-type and DKO vessels after injury. Taken together, these data indicate that IEL disruption has a profound effect on the activation of vascular cells after arterial injury, leading to a permissive environment for thrombus formation.
Elastic fibers are formed by the assembly of tropoelastin monomers onto a microfibrillar scaffold and subsequently crosslinked to form an insoluble elastin polymer 22, 23. Whereas Fbln5 expression is detected throughout the vessel wall during embryogenesis 24, Fbln2 is expressed only in the SMC layers in mid-gestation. No expression for Fbln2 is detectable in ECs until approximately E18. However, fibulin-2 becomes prominently localized to the BM region of ECs in the early postnatal period, which coincides with a period of active elastic fiber assembly 16.
In adult Fbln5−/− mice, a comparable amount of assembly was seen on the lumenal side of the IEL compared to age-match wild-type mice. However, the surface of the IEL adjacent to SMCs and the medial and external elastic lamina (EEL) never assembled properly. This suggests that the mechanism of IEL formation is distinct from other elastic laminae, and this mechanism is maintained even in the absence of fibulin-5. A dramatic disruption of the IEL in the DKO aorta clearly demonstrates that fibulin-2 and fibulin-5 cooperatively function to form IEL during development. Although we have not examined whether fibulin-2 regulates coacervation or maturation of tropoelastin, it is likely that fibulin-2 has a similar molecular function as fibulin-5 and that fibulin-2 can compensate for fibulin-5 and facilitate assembly of the IEL when fibulin-5 is absent. Taken together, it implies that a tissue-specific elastogenesis mechanism involving different members of fibulins may exist in vivo.
It is interesting to note that DKO vessels do not develop spontaneous aneurysms despite a severe developmental defect of the IEL and medial elastic laminae. We observed upregulation of major BM proteins, including laminin and collagen IV in DKO vessels. Since laminin is shown to attenuate EC response to shear stress, such as nuclear translocation of NF-kB and activation of C-Jun NH2-terminal kinase (JNK) 25, 26, the changes in subendothelial matrix composition in DKO vessels may influence EC stability in non-injured conditions. After injury, however, DKO vessels develop severe thrombus with thinning of the medial wall and a marked enlargement of the vessel diameter. Unlike a wire withdrawal injury model, ligation-induced injury causes stasis of blood flow without directly damaging the ECs. However, stasis and hypoxia in ligated vessels can induce fibrin deposition onto ECs through accumulation of inflammatory cells and activated platelets 27. We observed upregulation of VCAM-1 and ICAM-1 from 2 days after the ligation in DKO vessels, suggesting that attachment of ECs onto an intact IEL provides protection and stabilization of ECs during vascular injury. In addition, an increase in the TF transcripts in DKO arteries upon injury is compatible with the thrombotic phenotype in DKO mice. Since SMC-derived TF has recently been shown to be critical for thrombus formation after arterial injury, a compromised IEL may further facilitate production and interaction of TF with plasma components and lead to the activation of the coagulation cascade.
Morphological changes of the IEL in DKO vessels are much more severe than those in Fbln5−/− vessels. A previous report suggests that damage involving the EEL can be a more potent stimulus for neointima formation than a lesion only involving the IEL 1. MMP upregulation was shown to correlate with the extent of loss of elastic content and architecture in Marfan mouse models 28. Our present study highlights that the IEL also plays a role in determining vessel integrity during injury by providing structural stability to the vessel wall.
In injured DKO vessels, the total adventitial area was significantly increased compared with wild-type or Fbln2−/− vessels, however, DKO vessels failed to maintain a normal diameter. It has been reported that in the angiotensin II-induced ApoE−/− aorta, the intact vessel is surrounded by remodeled adventitia, whereas a break in medial layers is accompanied by a thin adventitial layer, suggesting a protective role of adventitia in vessel remodeling 29. In the current study, the ratio between adventitia and total vessel area was significantly increased in Fbln5−/− vessels compared to DKO vessels, and Fbln5−/− vessels showed less abnormal remodeling compared to DKO vessels. Thus, adequate adventitial thickening in the presence of an intact IEL may be critical for maintaining vessel remodeling during injury.
We have previously proposed two mutually compatible mechanisms for accelerated neointima formation in Fbln5−/− vessels after injury 19. One is due to a developmental defect of the elastic laminae and the inability to correctly assemble elastic fibers within the neointima, and the other is due to the lack of an inhibitory effect of fibulin-5 on proliferation and migration of SMCs. In vitro data by others also suggests that fibulin-5 inhibits SMC recruitment and EC proliferation 30, 31. Fibulin-2, on the other hand, was proposed to increase SMC migration by facilitating the versican-hyaluronan/fibulin-2 complex in vitro 32. Our present study, together with a recent study on the fibulin-2 null mouse 13, indicates that loss of fibulin-2 alone does not cause any SMC defects during development, and Fbln2−/− vessels do not show any abnormal response to ligation-induced injury. Interestingly, however, DKO vessels develop much less neointima compared with Fbln5−/− vessels despite much more severely disrupted elastic laminae. This indicates the possibility that the proliferative response of activated Fbln5−/− SMCs may be mediated in part by fibulin-2 or it may require the presence of fibulin-2. Further investigation will be necessary to clarify these possibilities.
Source of funding
This work was supported in part by NIH grants HL071157 (H.Y), GM55625 (M-L.C.), American Heart Association South Central Affiliate grant 0855200F (H.Y.), and the Canadian Institutes of Health Research grant MOP86713 (E.C.D.). E.C.D. is a Canada Research Chair.
We thank Nadine Korah for assistance with electron microscopy, and R. Ann Word for critical reading of the manuscript.