Journal of Pathology J Pathol 2000; 192: 234±242. DOI: 10.1002 /1096-9896(2000)9999 : 99993.0.CO;2-9
Expression of angiogenic factor thymidine phosphorylase and angiogenesis in human atherosclerosis Joseph J. Boyle1*, Beverley Wilson2, Roy Bicknell3, Steven Harrower4, Peter L. Weissberg1 and Tai-Ping Fan5 1
Department of Medicine, Cambridge University, Cambridge, UK Department of Histopathology, Cambridge University, Cambridge, UK 3 Imperial Cancer Research Fund Molecular Angiogenesis Laboratory, Oxford University, Oxford, UK 4 Department of Histopathology, Victoria In®rmary, Glasgow, UK 5 Department of Pharmacology, Cambridge University, Cambridge, UK 2
* Correspondence to: Dr J. J. Boyle, Unit of Cardiovascular Medicine, Addenbrookes Centre for Clinical Investigation, Box 110, Addenbrooke's Hospital, Cambridge, CB2 2QQ, UK. E-mail: [email protected]
Received: 3 February 2000 Accepted: 2 May 2000 Published online: 19 July 2000
Abstract In atherosclerosis, leukocyte migration into the plaque is thought to occur across the endothelium of the arterial lumen. However, intraplaque microvessels have been noted. While the signi®cance of, and stimuli for these are uncertain, it seems possible that they may represent a second portal of entry for leukocytes into the plaque. This study performed a basic characterization of intraplaque microvessels and tested the hypothesis that the novel angiogenic factor thymidine phosphorylase (TP) is expressed in atherosclerosis. Immunocytochemistry was performed on aortic and coronary plaques and morphometry on coronary plaques. In plaques from both sites, macrophages, foam cells, and giant cells were immunoreactive for the angiogenic factors TP and vascular endothelial growth factor. Venule-like intraplaque microvessels expressed endothelial leukocyte adhesion molecules HLA-DR and ICAM-1, in contrast to the endothelium overlying the plaque. In coronary plaques, there was a correlation between severity of stenosis and plaque microvascular density. These results are consistent with a role for plaque macrophage/foam cell TP in stimulating plaque angiogenesis. While attention has focused on dysfunction of the endothelium overlying the plaque, microvascular endothelium may also represent a portal of entry for leukocytes into established plaques. Copyright # 2000 John Wiley & Sons, Ltd. Keywords:
atherosclerosis; angiogenesis; in¯ammation; thymidine phosphorylase
Introduction In the main paradigm of atherogenesis, leukocyte migration occurs across the luminal endothelium overlying the atherosclerotic plaque , yet small vessels measuring tens of micrometres across and lined by endothelium have been noted within the lesion . Normal arteries have adventitial vasa vasorum but are devoid of microvasculature within the media or intima. Therefore any small vessels or `intraplaque microvessels' within plaques imply that angiogenesis must have occurred at some point and intraplaque vessels must by this de®nition represent `new' vessels (`neovessels'). However, the characteristics of these new vessels and possible stimuli for their development have received little attention. The lack of interest in these vessels is surprising, since angiogenesis and microvascular endothelial activation have undoubtedly important pathogenetic roles in chronic in¯ammation, wound healing, and malignancy, where either the therapeutic promotion or inhibition of angiogenesis may be useful . Possible stimuli for plaque angiogenesis include plaque hypoxia or leukocyte-derived cytokines. It is feasible that in¯ammatory cells within the plaque may stimulate angiogenesis. Vascular endothelial growth factor (VEGF), an endothelial cell mitogen in vitro and angiogenic factor in vivo, is induced by hypoxia and Copyright # 2000 John Wiley & Sons, Ltd.
in¯ammatory cytokines [4±7] and has been detected in plaques . Another angiogenic factor, not previously studied in atherosclerosis, is thymidine phosphorylase (TP, also known as platelet-derived endothelial cell growth factor), which was recently characterized as a tumour angiogenic factor by us . Like VEGF, TP is induced by hypoxia  and is found in neoplasms and rheumatoid synovium, where angiogenesis is pathogenetically important [7,9,11], or in association with chronic in¯ammation in gastric ulcer healing  or psoriasis . However, TP only stimulates endothelial cell migration, but not proliferation . We hypothesized that plaque in¯ammatory cells may induce neovascularization by producing angiogenic factors. We therefore examined human atherosclerotic lesions to determine the production and localization of the angiogenic factors VEGF and TP.
Materials and methods Autopsy atherosclerotic plaques Coronary plaques (n=89) were collected as 2±4 plaques from each of 20 consecutive autopsies performed or seen by one pathologist (JJB) as part of a previously published study . There were plaques from eight patients with fatal coronary thrombosis;
Expression of TP and angiogenesis in human atherosclerosis
four where death was considered due to ischaemic heart disease but without identi®able thrombosis; and eight patients where death was unrelated to vascular disease. Hearts were removed during hospital autopsies and the coronary arteries examined with transverse cuts at approximately 5 mm intervals. The ratio of lumen cross-sectional area to total arterial area of each 5 mm segment was estimated by visual inspection. The 2±4 plaques of lowest estimated lumen to total area ratio were sampled from each heart. If only two or three segments bore plaques, only these were sampled. Plaques were ®xed in 10% neutral buffered formaldehyde or Bouin's ®xative at room temperature for 24±48 h, conventionally processed, wax-embedded, and sectioned at 3±5 mm. To classify lesions, one section of each plaque was examined after haematoxylin and eosin (H&E) staining. Additional sections were immunostained for the angiogenic factors TP and VEGF; for the activation marker HLA-DR; for endothelial markers Factor VIII-related antigen (FVIIIRA), formerly known as vonWillebrand factor (vWF) and CD34; and for macrophage marker CD68.
Fresh atherosclerotic plaques Samples of super¯uous atherosclerotic artery wall from ten aortas were collected at peripheral arterial bypass and aortic aneurysm repair surgery, and divided in equal portions. Half of each sample was snap-frozen within 10 min of removal from the patient. The other half was simultaneously ®xed and decalci®ed for up to 48 h in 10% formalin with 1.0 mM ethylene diamine tetraacetic acid (EDTA) before conventional processing and microtomy. Cryostat sections were used for immunocytochemistry (ICC) of ®xation-sensitive epitopes (ICAM-1). Paraf®n sections were used for ICC for TP and VEGF, which were resistant to ®xation.
Analysis and numerical methods Morphometry was performed on plaques immunostained with FVIIIRA (vWF). Images of vWFimmunostained sections were taken with a Nikon Optiphot microscope and charge coupled device three-colour video camera interfaced via a video grabber board to an Apple Macintosh microcomputer running the program `Digital Image' (Graphics Unlimited, Cambridge, UK). Images of the whole plaque were captured with a r2 objective and saved digitally. The area of densest plaque neovascularization was identi®ed in each plaque and a r20 objective image of this area was captured and saved digitally. The digitally stored images were analysed on an Apple Macintosh desktop microcomputer running the shareware (free) program `NIH-Image 1.60' (from the National Institutes of Health, Bethesda, USA; internet site ftp.zippy.nimh.nih.gov). This was used to measure areas de®ned by manually drawing around vessels. The ratio of lumen area to total arterial area (L/T area) was then calculated as an estimate of the severity of stenosis, as we have previously described . A low Copyright # 2000 John Wiley & Sons, Ltd.
L/T ratio represents a tight stenosis. This, rather than plaque area, was calculated to compensate for variation in the size of the original coronary artery arising from sampling of plaques from anywhere in the extramural coronary tree. The r20 microvascular area fraction was calculated as the total area of microvessel lumina divided by the total image area, for the most densely neovascularized area of each plaque. This area was chosen for consistency between plaques and because measures of the average or total microvascular density are prone to skewing by acellular areas of the plaque such as the necrotic core. Thus, advanced plaques with large necrotic cores might have high angiogenesis but severely underestimated average microvessel density. The area fraction of plaque microvessels was measured to eliminate the argument that larger plaques would contain more microvessels by chance. The microvascular area measured in each plaque was then correlated with the L/T ratio and Spearman's rank correlation coef®cient was used to assess the strength of this correlation non-parametrically. To provide additional correlation of intraplaque microvessels with lesion severity and provide an overall estimate of the range of pathology dealt with in this study, H&E-stained plaques were classi®ed according to Stary et al.  as follows: I, initial lesion (diffuse intimal thickening); II, fatty streak; III, small pools of extracellular lipid; IV, ®brolipid plaque with a welldeveloped lipid core and ®brous cap; V, formation of new ®brous tissue; VI, complicated plaque. Plaques were scored as containing intraplaque microvessels or not. Whether intraplaque microvessels were associated with the Stary classi®cation was then tested with x2 analysis.
Immunocytochemistry The following mouse monoclonal antibodies were applied to plaque sections: CD34, Dako clone QBEND10; CD31, Dako clone JC70; HLA-DR, Dako clone TAL.1B5; HLA-D invariant region, Cymbus Biosciences clone CBL IQU9; ICAM-1, Serotec clone 83H10; TP/PDECGF, clone PGF.44C 15; VEGF, R&D Systems clone MAB293. The anti-TP antibodies and anti-VEGF antibodies used have been previously validated for use on paraf®n-embedded tissue [9,17]. Endothelial antigen vWF (FVIIIRA) was detected with a rabbit polyclonal antibody (Dako). These were routinely detected with an avidin biotin complex immunoperoxidase method, with 3,3k-diaminobenzidine (DAB) substrate for visualization. In double-labelling experiments, the weaker antigen (anti-HLA-DR) was ®rst visualized with ABC-DAB enhanced with Ni/Co; then the second antibody (antivWF) was applied and visualized with a standard alkaline phosphatase±anti-alkaline-phosphatase (APAAP) technique and either Fast Red or Fast Blue substrate. Species and isotype-matched irrelevant monoclonal antibodies were used as negative controls. J Pathol 2000; 192: 234±242.
Results General features Autopsy coronary plaques
The sections studied included all severities of atherosclerosis from diffuse intimal thickening to the formation of a central necrotic core and plaque rupture. Endothelial-lined intraplaque vascular channels were identi®ed in all types of plaque other than those representing only diffuse intimal thickening. Intraplaque microvessels were more conspicuous in more advanced plaques. These impressions were formally
J. J. Boyle et al.
tested morphometrically and by comparison to Stary classi®cation (below). The intraplaque microvessels identi®ed were usually thin-walled channels a few tens of micrometres across and often contained red blood cells, indicating that they were not lymphatics (Figure 1). The plaque microvessels were identi®ed at the plaque base and also more super®cially in the shoulder region and ®brous cap (Figure 1). Occasionally leukocytes were identi®ed adherent to plaque microvascular endothelium and in some plaques, clusters of leukocytes could be identi®ed around microvascular channels (Figure 1).
Figure 1. (a) Low-power view of a coronary plaque. Blood-containing intraplaque microvessels are identi®ed (H&E; r20). This pattern was seen in 58 Stary IV plaques and 78 Stary V plaques. (b) High-power view of an advanced coronary plaque. Blood-®lled intraplaque microvessels with an adjacent accumulation of mononuclear cells. (H&E; r40). (c) An intraplaque microvessel in the wall of an aortic atherosclerotic aneurysm showing strong endothelial immunoreactivity for ICAM-1, and with an adjacent accumulation of mononuclear cells (Serotec clone 83H10; APAAP)
Figure 2. TP immunoreactivity in coronary plaques. (a±c) Super®cial cap. (a) Coronary plaque: numerous macrophages are strongly immunoreactive for TP (clone GP44F; ABC/DAB; haematoxylin counterstain; Nomarski r40). (b) High-power view of a. Detail of plaque foam cells expressing TP. (c) Coronary plaque: step section, macrophages are strongly immunoreactive for CD68 (macrophage marker) (clone GP44F; ABC/DAB; haematoxylin counterstain; r400). (d±f) Plaque base. (d) Necrotic core and associated macrophages are immunoreactive for TP (clone GP44F; ABC/DAB; haematoxylin counterstain; Nomarski r40). (e) Negative control shows no immunostaining (isotype-matched antibody, ABC/DAB; haematoxylin counterstain; Nomarski r40). (f) Necrotic core and associated macrophages are immunoreactive for CD68 con®rming macrophage origin (CD68, ABC/DAB; haematoxylin counterstain; Nomarski r40) Copyright # 2000 John Wiley & Sons, Ltd.
J Pathol 2000; 192: 234±242.
Expression of TP and angiogenesis in human atherosclerosis
The plaque new vessels were always lined by endothelium, but a smooth muscle cell coat was never identi®ed in their walls. Fresh aortic aneurysms
The aortic aneurysm walls studied were all very advanced complicated plaques (Stary VI) with thin or broken media and were ulcerated, devoid of luminal endothelium. Abnormal thin-walled vascular channels were identi®ed in the plaques (Figure 1).
Angiogenic factors Thymidine phosphorylase (TP) immunoreactivity was assessed in autopsy coronary plaques and surgical
aortic plaques. Both sites showed a similar pattern of TP immunoreactivity in foam cells, macrophages, and necrotic debris (Figure 2). The mural thrombus of aneurysms was TP-immunoreactive (not shown). Contemporaneous negative controls using isotype and concentration-matched antibody showed no staining (Figure 2). The identity of macrophages was con®rmed with CD68 immunostaining (Figure 2). Vascular endothelial growth factor (VEGF) immunoreactivity was assessed in autopsy coronary and surgical aortic plaques. In both sites, VEGF immunoreactivity was exhibited by mononuclear cells, foam cells, and the necrotic debris in the lesion core. In the (surgical) aneurysms, the mural thrombus was VEGF-immunoreactive (Figure 3). Plaque endothelial
Figure 3. VEGF immunoreactivity in aortic and coronary plaques. Left-hand side: step sections from representative abdominal aortic aneurysm plaque (n=5 subjects). Right-hand side: step sections from representative advanced coronary plaque (n=5 subjects). Top row to bottom row: staining for VEGF, negative control, and CD38 macrophage marker. Numerous cells in the aortic plaque and coronary plaque stain positively for VEGF but negative controls on adjacent step sections show no staining. Comparison of staining in coronary plaque shows that VEGF-positive cells in coronary plaque are also CD68-positive, indicating that they are macrophages Copyright # 2000 John Wiley & Sons, Ltd.
J Pathol 2000; 192: 234±242.
Figure 4. Scattergram of plaque microvessel density and severity of stenosis: each point represents a plaque. The x-axis shows the ratio of lumen area to total area bounded by adventitia for the corresponding plaque; this is inversely related to the severity of stenosis. The y-axis shows the microvascular area fraction in the most densely neovascularized r20 optical ®eld in the plaque. There is a correlation; Spearman's rho=0.59
cells and smooth muscle cells were more weakly VEGF-immunoreactive. Plaque collagen matrix was weakly VEGF-immunoreactive. Contemporaneous negative controls using isotype and concentrationmatched antibody showed no staining (Figure 3).
Morphometry of intraplaque vessels There was a signi®cant negative correlation between the lumen to total area ratio and the microvascular area of the most densely neovascularized area of each plaque [Spearman's r (rho)=0.59; 99% con®dence interval 0.165±0.7647; Figure 4]. That is, for a given overall size of coronary artery, the smaller the lumen area, the higher the maximal density of plaque microvessels. This con®rms the initial impression that intraplaque microvessels are more pronounced in more advanced plaques. Then whether plaque intraplaque microvessels were associated with advanced plaques, as assessed with
J. J. Boyle et al.
Figure 5. Presence of intraplaque microvessels compared with Stary classi®cation. y-axis: number of plaques; x-axis: Stary classi®cation (see the Materials and methods section); ®lled histograms: plaques without intraplaque microvessels; shaded histograms: plaques with intraplaque microvessels. Advanced plaques contained more intraplaque microvessels than less advanced plaques (x2, p