Ongoing angiogenesis in blood vessels of the abdominal aortic aneurysm

Pathogenesis of the abdominal aortic aneurysm has been attributed to neovascularization of the aortic wall. However, it is not clear whether angiogenesis persists in the aneurysm. In sections of aneurysms, we determined the immunohistochemical distributions of the αvβ3 integrin, tenascin and endothelial nitric oxide synthase (eNOS), which are markers respectively, of angiogenesis, matrix remodeling and vasoregulatory function. In addition, we used reverse transcription followed by in situ PCR, to determine the distribution of αv mRNA. All aneurysm specimens exhibited extensive increases of wall vascularization as compared with the control aortic wall, and showed the presence of perivascular inflammatory exudates containing macrophages and lymphocytes. The neovascularization consisted of thick-walled vessels in the media and adventitia, and capillaries in the subintima. The majority of vessels stained positively for the αvβ3 antigen and eNOS. Tenascin was deposited as bands that circumscribed thick-walled vessels. The distribution of av mRNA was extensive and was positive even in those vessels that failed to stain for the αvβ3 protein. No staining was evident in control aortas for the αvβ3 antigen, tenascin or αv mRNA. The upregulation of av mRNA and the αvβ3 integrin in blood vessels surrounded by a matrix expressing tenascin, indicates that angiogenesis is an ongoing process in the mature aortic aneurysm.


Introduction
The pathophysiology of the abdominal aortic aneurysm (AAA), which carries an annual mortality in the USA of ~15,000 (Brophy et al., 1991) is not well understood; and its treatment is restricted to surgical repair with its attendant risks of morbidity and mortality. The pathophysiological understanding of the aneurysm is based largely on biochemical and immunohistochemical studies that indicate aortic wall remodelling and proteolysis of matrix proteins (Tilson et al., 1990). A multicellular inflammatory infiltrate and increased expression of tissue-type and urokinasetype plasminogen activators have also been demonstrated in the aneurysm wall (Koch et al., 1990;Brophy et al., 1991;Schneiderman et al., 1995).
A conspicuous feature of the aneurysm is neovascularization of the aortic wall, which, in contrast to normal, becomes enriched with microvessels (Koch et al., 1990;Tilson et al., 1990;Brophy et al., 1991;Holmes et al., 1995;Schneiderman et al., 1995). Although the significance of the neovascularization remains unclear, Herron et al. (1991) speculated that the new vessels play a sustaining or even a causal role in the pathophysiology of the aneurysm by secreting proteinases that destabilize the aortic matrix. Supportive evidence for this hypothesis comes from immunohistochemical studies in which proteinases such as gelatinase (matrix metalloproteinase 2, MMP-2) (Herron et al., 1991) and collagenase (MMP-1) (Irizarry et al., 1993) have been localized to the neovascular endothelium.
A highly relevant but poorly understood question is whether the neovascularization is an ongoing process in well-developed aneurysms. If neovascularization were ongoing it would signify the presence of active disease in the aneurysm wall and would support the hypothesis that the new vessels are critical in the progression of the aneurysm. We address this question through detection of the αvβ3 integrin in adventitial vessels of aneurysms. The integrin αvβ3 has been proposed to play a critical role in angiogenesis. Brooks et al. have reported that this adhesion molecule is a marker for active angiogenic vessels in man and the chick, and that blocking its activity with a monoclonal antibody suppresses angiogenesis (Brooks et al., 1994).
During angiogenesis, the αvβ3 integrin uses the RGD sequence in the tenth fibronectin (III)-like domain of tenascin to bind to extracellular matrix (ECM) (Joshi et al., 1993;Sriramarao et al., 1993), and tenascin in turn has additional binding sites for matrix proteoglycans and heparin. Expression of αvβ3 is ubiquitous in the embryology and development of the vascular system, and expression of members of the tenascin superfamily is similarly widespread in fetal tissue. In adult tissues, although the αvβ3 integrin may be detectable in low amounts (Suzuki et al., 1993), expression of tenascin isoforms is markedly downmodulated within one month of birth (Saga et al., 1991). Under abnormal circumstances, such as in tumorigenesis and wound healing, αvβ3 expression increases significantly (Suzuki et al., 1987;Felding-Habermann and Cheresh, 1993;Juhasz et al., 1993). Thus, detection of these proteins associated with the neovascularization, as we report in the present work, indicates the presence of active angiogenesis in aneurysms.

Im m unohistochem istry
Aortas were obtained from 17 consecutive patients undergoing elective resection of AAA. All patients presented with abdominal aortic aneurysms of 5 cm in diameter size or greater. None of the patients suffered from rupture of AAA at the time of surgery. The sections of the AAA wall were reviewed macroscopically. Normal control aortic tissues were from 6 organ donors and 1 aorto-occlusive patient (ages average 38 yr; 4 male; 3 female). All tissues were fixed in buffered formalin (pH 7.4) for 16-24 h.
Histological sectioning and immunohistochemistry were processed as in previous work (Fu et al., 1994). Briefly, aortic paraffin embedded tissue sections (4 µm thick) were dewaxed in xylene and absolute ethanol, inhibited for endogenous peroxidase activity in 0.5% hydrogen peroxide in absolute methanol (20 min), and blocked with 0.2% BSA to prevent nonspecific binding. The anti-αvβ3 antibody R838 (1:100) and the anti-tenascin antibodies, EH5B01 (1:50) and TN2 (1:200), were applied to the sections overnight at 4 o C. The anti-eNOS antibodies (1:2000) were applied to slides following a microwaving step used to enhance antigenicity. Following PBS washes, the sections were exposed to the appropriate secondary antibody for 1 h at room temperature. After PBS washes the sections were incubated for 10 min at room temperature with either DAB solution in PBS containing 0.01% hydrogen peroxide, or NBT/BCIP solution, depending on whether the secondary antibody was respectively, HRP or alkaline phosphatase linked. The sections were finally washed in running tap water, counterstained with hematoxylin for DAB or Nuclear Fast Red for NBT/BCIP, dehydrated in a reversed alcohol-xylene series, and mounted for semi-quantitative morphometric analysis, microscopy and photography.

Sem i-quantitative m orphom etric analysis of m icrovessels
Quantitation of microvessel staining by anti-eNOS was carried out by an observer blinded to the source of each specimen, using three criteria: (1) overall impression of staining at 100X magnification, graded on 1 to 4 scale (1 = minimal, 2 = easily detectable, 3 = moderate, 4 = extensive), (2) the number of capillaries identified in the most densely positive 100 × field, and (3) the number of endothelial cells identified in the most densely positive 400 × field.

Reverse transcription (RT) in situ PCR
For in situ PCR, we modified Nuovo's method (Nouvo, 1994) and used the EZ rTth RNA PCR kit as follows: after dewaxing and air drying, all sections were digested in 2 mg/ml pepsin in 0.01 N HCl (45 min, room temperature). The pepsin was inactivated by 1 min washes in ultrapure water and absolute ethanol, then the sections were air dried. As a positive control, amplification of tissue genomic DNA was performed by carrying slides directly on PCR cycles without digestion by DNAse I. For detection, the sections were digested by RNase-free DNase I (400 U/ml in digestive buffer) overnight at 37 o C. To control for nonspecific binding and background staining, after digestion of DNAse I, the slides were further digested by RNase H (50 U/ml in the digestive buffer) for 2 hours at room temperature.
For RT and in situ PCR processes, 50 µl of the reagent mixure (1 × EZ buffer, 200 µM of dNTP, 15 µM of -11-dUTP, 2.5 mM of Mn(AOc)2 solution, 0.45 µM/each of αv or β3 5' and 3' primers, 2.5 U/50 µl of rTth DNA Polymerase, and 60 U RNasin) was added to each section and sealed by the In Situ PCR assembly tool (Perkin Elmer Corp., Newark. CT). After incubation for 25 min at 62 o C for reverse transcription, the PCR cycles were instituted in an in situ PCR Cycler (Ampligen PCR 1000, Perkin Elmer Corp., Newark, CT) under modified conditions (10 cycles: 94 C for 60 s, 65 C for 60 s and 72 C for 120 s).
To determine the specificity of αv primers, we sampled aliquots of the supernatant on tissue sections immediately following the RT in situ PCR procedure, with or without treating the sections with DNase I. The aliquots were loaded on to 1% agarose gel and Southern blotted by αv or β3 probes. These probes (kindly provided by Dr. D. Shinar) were synthesized and labeled with biotin-11-dUTP by PCR amplification of rat αv (RAV 611) or β3 (RIB 494/3) cDN A, using oligonucleotide prim ers (show n as above) (Shinar et al., 1993). The membrane was developed using Streptavidine-biotin-alkalinephosphatase and NBT/BCIP. The sections were chilled at 4 o C for 20 min, and blocked by 0.2% BSA in 0.1 SSC solution at 45 o C for 10 min. Then the slides were incubated in goat anti-digoxigenin antibody conjugated with alkaline phosphatase for 30 min at room temperature. After 3 washes of Tris buffer (0.1 M, pH 7.5), the sections were treated for 10 min at room temperature with NBT (2.5 µl/ml)/BCIP (2.5 µl/ml) in Tris-buffered CaCl2 solution (pH 9.5). The development was stopped by washing in running tap water. The sections were counterstained with Nuclear Fast Red, dehydrated and mounted for viewing.

R esults M icrovessels detected by H&E and anti-eNO S
Histological sections of aortic aneurysms by H&E staining showed the presence of several hemorrhagic regions, large numbers of inflammatory cells and numerous microvessels ( Figure 1). The inflammatory cells consisted mostly of macrophages and lymphocytes that were associated largely with the microvessels in the adventitia. In addition, some sub-intimal and medial regions showed the presence of neutrophils. By contrast, no inflammatory infiltrates or matrix abnormalities were evident in control aortas ( Figure 1D). Anti-eNOS immunohistochemical staining was employed for two reasons. First, detection of the intracellular eNOS enzyme in AAA and control tissue sections confirmed antigenic integrity of the preserved specimens. Second, anti-eNOS staining served as a marker for microvessel quantitation since the antibody detects the endothelial layer of both small and large microvessels.
In normal aortic wall, positive eNOS staining was seen in the endothelial cell layer of microvessels located in the adventitial vasa vasorum. The number of vessels seen per 100 × field was few, varying from none to 6. Intact medial smooth muscle cell nuclei were noted in most sections and medial elastin lamellae, as expected, were preserved in all control sections ( Figure 2B). In AAA sections positive staining was again seen in the endothelial lining of microvessels. These vessels, however, were numerous compared to controls. Of the 10-30 vessels present per high power field (100 ×), more than half were thin-walled capillaries of diameter 10 µm and the remainders were thick-walled microvessels of diameter 25 µm and wall thickness 10 µm (Figures 2A and C). The capillaries were localized to both the subintimal region and the adventitia, while the thick-walled vessels occurred almost exclusively in the adventitia. Also noted in many sections were dense inflammatory infiltrates surrounding both thick-walled microvessels as well as capillaries. These cells, based on their morphology on H&E staining, were mostly macrophages and lymphocytes. Lastly, elastic lamellae, in contrast to controls, were markedly disrupted and depleted. The results of the anti-eNOS morphometric analysis are shown in Figure 3. The differences between AAA and controls were striking. Mean capillaries were increased 15-fold per 100 × field (P 0.001), and mean endothelial cells 8-fold per 400 × field (P 0.001) in 17 AAA specimens by comparison to 7 controls. The general increase in staining for eNOS was highly remarkable in the AAA, although endothelial cells from control patients had approximately equivalent staining reactions on a per cell basis.

Anti-α v β 3 , anti-tenascin, and m RNA α v studies
To determine the specificity of αv primers, we sampled aliquots of the supernatant on tissue sections immediately following the RT in situ PCR procedure, with or without treating the sections with DNase I. Then we blotted the aliquots using our αv probe. As shown in Figure 4, in non-aneurysm tissues a band was detectable only in the absence of DNase I treatment, which is expected because the probe can The aortic wall distribution of the αvβ3 antigen is shown in Figure 5. In aneurysmal aortas, an overwhelming majority of microvessels, including capillaries, stained positively for the αvβ3 antigen. However, no staining occurred over the aortic endothelium, the medial and adventitial stroma, and in most large microvessels. A large number of macrophages, particularly those associated with microvessels, also stained positively. No staining was evident in either control aortas or in aneurysms in which R838 was replaced with non-immune rabbit serum as the primary antibody (negative control).
RT in situ PCR of tissue sections of aneurysms revealed co-localization of the αv mRNA with the αvβ3 antigen detected by immunohistochemistry ( Figure 5). mRNA staining was much more extensive than antigen staining. Almost all vessels, irrespective of diameter or location, most macrophages, and some lymphocytes, stained positively for αv mRNA. We studied β3 mRNA in one aneurysm specimen and found a distribution for β3 mRNA that closely resembled that of the αv mRNA (not shown). PCR products of αvβ3 integrin were not evident in either non-vascularized regions of aneurysms or in nonaneurysmal control aortas ( Figure 5). These results prove that both the transcription and the expression of the αvβ3 integrin are upregulated in aneurysm microvessels. The distribution of tenascin staining around microvessels was closely linked to that of αvβ3. Tenascin staining typically was seen circumscribing thick-walled adventitial microvessels ( Figure  6). Tenascin was also evident in subintimal regions and in areas of medial fibrosis.

D iscussion
Although neovascularization of the aneurysm wall has been previously reported (Koch et al., 1990;Tilson Figure 2. Anti-eNOS immunohistochemical staining (A) AAA wall at 40 × magnification. Most apparent are numerous eNOS-staining microvessels (arrow) outlined in brown stain. They are pictured infiltrating the AAA wall. Greater than 20 distinct lumens are identifiable. There is disarray of the normal regular elastic lamellar architecture (in comparison with B). Although the photomicrograph depicts the expected location for the adventitia/media interface, it is undiscernable. The arrow marks a single capillary, whose endothelial cell is better seen under higher magnification. (see panel C) (B) Non-aneurysmal aortic wall at 40 × magnification. Intact aortic media is free of microvessel invasion and hence dark brown eNOS staining is absent. The adventitia/media interface is clearly visible to the near left (arrow) and loose adventitial tissue (A) is seen to the far left. Long continuous strands of elastic tissue appear translucent, forming a regular orderly pattern within the media (M). Mutiple smooth muscle cells with blue staining nuclei are seen within the aortic media interspersed between elastic lamellae. (C) 100 × magnification of boxed portion of panel A. An endothelial cell (large arrow) outlines a microvessel, staining dark brown with anti-eNOS antibody. Intraluminal translucent red and white blood cells are easily seen at this magnification. Cell nuclei are stained blue but are poorly visualized in endothelial cells that stain positive for eNOS. The tissue stroma contains a single mononuclear inflammatory cell (small arrow). Brophy et al., 1991;Holmes et al., 1995;Schneiderman et al., 1995) the literature does not clarify whether new vessels appear early or late in the disease, or whether the vasculature undergoes continuous development and remodelling as the aneurysm enlarges. Our findings using an anti-eNOS Ab indicate that all layers of the aneurysm wall are vascularized and display characteristics typical of angiogenesis. Most extensive were thick-walled microvessels that displayed outer-medial to adventitial distribution, and thin-walled capillaries that invested the sub-intima. It is possible therefore, that vessels proliferating in the outer wall also developed transwall capillarization as part of a global neovascularization of the aneurysm. Two angiogenesis markers, namely the αvβ3 integrin and tenascin, were widely detectable in these blood vessels. We conclude from these findings, that vascularization of the aneurysm is extensive and that new blood vessels form in the aneurysm wall as a result of ongoing angiogenesis.
Evidence for angiogenesis was particularly conspicuous in the immunohistochemistry and the RT in situ PCR studies. The αvβ3 integrin, a vascular integrin of the cytoadhesin family (Albelda and Buck, 1990), is located in both the luminal and abluminal surfaces of endothelial cells as well in vascular smooth muscle cells, and has wide-ranging functions including cell adhesion and cell spreading (Albelda and Buck, 1990;Damjanovich et al., 1992;Joshi et al., 1993) and vascular permeability regulation (Tsukada et al., 1995). We confirmed that in thick-walled microvessels, both the endothelial lining as well as the smooth muscle layers stained positively for the integrin. Recently, integrin αvβ3 has been shown to be essential for angiogenesis and a mAb that blocks binding of ligands to αvβ3, promoted tumor regression of angiogenic blood vessels in chick embryos (Brooks et al., 1994). The authors suggest that inhibition of ligand binding to αvβ3 suppresses neovascularization and selectively promotes apoptosis of vascular cells. Interestingly, αvβ3 expressing blood vessels in aneurysms of the present study, were typically associated with macrophages that also expressed both the αvβ3 protein and αv mRNA. The αvβ3 integrin mediates removal of apoptotic cells by macrophages (Brooks et al., 1994;Flora and Gregory, 1994). Since apoptosis co-exists with growing tissue (Brooks et al., 1994), the presence of αvβ3 expressing macrophages in the vicinity of αvβ3 expressing vessels, further signifies that these locations were focal areas of active tissue growth and cell death.
From the experiments, two important results were, first, αv mRNA was undetectable in control aortas but strongly detectable in vessels of the aneurysm wall; second, some aneurysm microvessels failed to stain for αvβ3 although they stained positively for αv mRNA. These findings indicate that increased transcription of the αv gene is a significant feature of aneurysm microvessels. Increased transcription of the αv and β3 genes are co-associated (Shinar et al., 1993), hence, transcription of β3 mRNA and β3 expression also increased in these aneurysms. This is indirectly indicated in these experiments, in the increased staining with a polyclonal antibody that recognizes the αvβ3  dimer but not the individual subunits of the integrin (Suzuki et al., 1986). In one case, we determined the distribution of β3 mRNA by Rt in situ PCR and found this distribution to be similar to that of αv mRNA (data not shown). The mismatch between negative avb3 integrin staining and positive av mRNA expression in some areas indicates that increased αvβ3 expression did not always follow increased gene transcription. This may reconcile the apparent inconsistency between our findings and those recently reported by Cheuk and Cheng (2004) in which no significant difference in RNA transcripts for αv and β3 were ob-served in AAA vs. control homogenates. Hence, it is possible that increased αvβ3 expression is restricted to specific regions of active angiogenesis. The immunohistochemical evidence for blood vessel-associated tenascin expression also supports our proposal that angiogenesis was active in these aneurysms. Tenascin is a large oligomeric glycoprotein that appears transiently in the extracellular matrix during tissue modelling and tissue repair (Tremle et al., 1994). Two domains of tenascin, both of which contain the integrin-recognizing arg-gly-asp (RGD) tripeptide sequence, support endothelial cell adhesion  (Joshi et al., 1993). One of these, the third fibronectin-III domain, also supports cell spreading (Sriramarao et al., 1993). The involvement of endothelial integrins and in particular, the αvβ3 integrin is indicated in that endothelial adhesion to tenascin is RGD dependent (Joshi et al., 1993) and is partly inhibited by anti-αvβ3 antibodies (Sriramarao et al., 1993). The ability of endothelial cells to adhere to tenascin (Joshi et al., 1993) suggests that endothelial tenascin receptors play a supportive role for establishing blood vessel structure during angiogenesis. The circumscribed perivascular tenascin expression seen here, is consistent with this notion and suggests that that tenascin provides a cylindrical template for vessel tube formation. Addition of tenascin to a fibronectin-based substratum upregulates the synthesis of three important matrix metalloproteinases in fibroblasts: collagenase (MMP-1), stromelysin (MMP-3), and gelatinase-B (MMP-9) (Newman et al., 1994;Newman et al., 1994). We have reported that these three matrix metalloproteinases are present in significantly higher levels in aneurysms than in control aortic specimens (Newman et al., 1994). Tenascin may promote aneurysm progression by inducing these metalloproteinases in endothelial cells of new blood vessels as well as in mesenchymal cells.
Using anti-eNOS Ab directed against endothelial cells, we observed marked increases in neo-vessel number in the aneurysm wall. Although no conclusions can be made concerning the possible role of eNOS or nitric oxide in aneurysm formation from our study, Kuhlencordt et al. have suggested that alterations in eNOS expression play a role in the formation of AAA (2001). Furthermore, we have shown that the nitric oxide by-product, nitrite, is capable of invoking deleterious changes in connective tissue proteins such as collagen and elastin (Paik et al., 2001). Since levels of nitrogen oxide gases are high in tobacco smoke and cigarette smoking is a strong environmental risk factor for AAA disease (Lederle et al., 2003), future studies will be directed at clarifying the role of nitric oxide and its by-products in the formation of AAA.
Finally, our finding that angiogenesis is active in the mature aortic aneurysm, suggests that a potential intervention site exists for pharmacologically reducing the rate of aneurysm enlargement or preventing the disease in genetically susceptible individuals. The observation by Brooks et al. (1994) that the angiogenesis in tumorigenesis can be arrested by antibody to αvβ3 suggests that anti-angiogenic therapy might be considered among other novel interventions that are presently under investigation, based on recent developments in our understanding of aneurysm pathogenesis. Roosevelt Hospital Center, with NIH grant R01 HL64334-4 to MDT.