The number of investigations of angiogenesis in the eye has increased significantly in recent years due to findings that neovascularization of the retina and choroid plays a central role in the development of a number of major blinding diseases. These include AMD as well as diabetic retinopathy, polypoidal choroidal vasculopathy, myopic choroidal neovascularization, neovascular glaucoma, retinopathy of prematurity, and ocular tumorigenesis (all reviewed in 
). Since a seroepidemiologic clinical study by us demonstrated an apparent association between HCMV infection and neovascular AMD 
, we used an experimental C57BL/6 mouse model of CNV to test the hypothesis that systemic MCMV infection will contribute to the severity of CNV. It has not escaped our attention that mouse strain-dependent factors might play a factor in CNV development during systemic MCMV infection since macrophages from C57BL/6 mice (a prototypical Th1 mouse strain) and macrophages from BALB/c mice (a prototypical Th2 mouse strain) exhibit distinct M1- or M2-dominant responses 
. Nonetheless, our results collectively showed that systemic MCMV infection of C57BL/6 mice did indeed result in more severe CNV, and, more importantly, chronically infected mice showed the greatest severity of CNV. Although MCMV DNA sequences could not be detected within choroidal tissues of chronically infected animals, splenic macrophages collected from chronically infected animals produced increased amounts of transcripts to several pro-inflammatory and pro-angiogenic cytokines including VEGF. That MCMV infection of mouse macrophages will modulate a pro-angiogenic M2 phenotype that included significant stimulation of VEGF production was shown directly by in vitro studies using a mouse macrophage cell line of C57BL/6 origin. Further evidence that virus infection induced stimulation of VEGF production both in vivo and in vitro was provided by ganciclovir treatment studies that demonstrated sensitivity of VEGF production to the antiviral both in vivo and in vitro. Thus, our findings are novel with respect to chronic eye disease since they provide for the first time new data that suggests that chronic cytomegalovirus infection can contribute to the pathogenesis of wet AMD, possibly via activation of macrophages towards a pro-angiogenic phenotype and stimulation of VEGF production. While we have not yet demonstrated in our model direct visualization of MCMV-infected, VEGF-producing macrophages associated with areas of CNV, several observations would argue that this is a likely occurrence. Firstly, we 
and others 
have shown previously that macrophages are essential for development of CNV. Secondly, we have shown previously in the context of MCMV retinitis that IC-21 macrophages infected with a β-galactsidase-expressing LacZ recombinant MCMV will travel to ocular tissues of C57BL/6 mice following tail vein injection 
. Finally, we show herein that MCMV infection of IC-21 macrophages stimulates VEGF production, a stimulation that is also observed in splenic macrophages collected from chronically infected mice with severe CNV. Future in vivo immunostaining studies will directly address this important issue.
The concept that infectious agents might contribute to the pathogenesis of vascular diseases has become an intense and controversial area of investigation. Two major hypotheses have emerged. One hypothesis proposes that vascular disease is caused by direct infection of the target tissue 
, while the second hypothesis proposes a bystander effect caused by infection at a distant tissue 
. In atherosclerosis, direct infection of the atheromatous plaque by Chlamydia pneumoniae
has been suggested as a stimulus for recruitment of inflammatory cells. Arguing against this hypothesis, however, are antibiotic treatment trials designed to suppress Chlamydia pneumoniae
infection and failing to demonstrate a measurable clinical effect on preventing myocardial infarction or other sequelae 
. On the other hand, patients with chronic periodontal infection and inflammation have provided evidence suggesting that chronic infection at a distant site may play a role in vascular disease. In this patient population, infection by a variety of different organisms appeared to lead to more severe vascular disease 
. Since in our study, MCMV-specific DNA sequences could not be detected in choroidal tissues of eyes with the most severe choroidal neovascularization, we propose a similar bystander hypothesis for the role of HCMV infection in chroroidal neovascularization of the eye.
HCMV is a common β-herpesvirus that persists for the life of its host following primary infection. While chronic HCMV infection of healthy, immunologically normal persons was initially thought to have no significant disease consequence, chronic HCMV infection has now been associated with a growing number of long-term diseases that include the vascular disease atherosclerosis, restenosis following angioplasty, transplant vascular sclerosis associated with chronic allograft rejection of solid organ grafts (reviewed in 
), and possibly tumor formation (reviewed in 
). Evidence for a link between HCMV and vascular disease was first provided by Melnick, DeBakey, and coworkers 
when virus antigen was detected within arterial tissues from carotid artery plaques obtained from patients with atherosclerosis. Since this fundamental observation of ~20 years ago, however, it has been difficult to determine the precise mechanisms by which HCMV might participate in the pathophysiology of vascular disease because the etiologies of chronic diseases are complex and multifactorial. Nonetheless, seropositive HCMV persons are two to three-times more likely to develop coronary artery disease when compared with HCMV seronegative patients 
. In support of this association are recent findings that 76% of patients with ischemic heart disease have detectable HCMV DNA within their vascular tissues 
, and up to 53% of carotid artery atherosclerotic lesions are positive for HCMV DNA 
. A number of animal studies have also provided compelling evidence that cytomegalovirus plays an important role in the pathophysiology of atherosclerosis, including several studies that have demonstrated more severe atherosclerosis in apoE −/− mice following systemic MCMV infection 
While an association has been recognized between cytomegalovirus infection and atherosclerosis, the strongest association of cytomegalovirus in vascular disease is with the development of restenosis and transplant vascular sclerosis. Several clinical studies have shown that HCMV infection is involved in accelerating both acute and chronic graft failure in all types of solid organ transplants by promoting vascular disease associated with rejection 
, probably by virus originating from the vasculature of transplanted organs from HCMV seropositive donors 
. For example, HCMV infection was shown to double the 5-year rate of graft failure in cardiac allograft recipients due to accelerated transplant vascular sclerosis 
. Similarly, kidney transplant allograft survival was decreased in asymptomatic HCMV-infected recipients during the first 100 days after transplantation when compared with recipient patients who had no evidence for HCMV infection, an outcome suggesting that HCMV infection, even when asymptomatic, has a negative impact on graft survival 
. These clinical findings have been supported by a number of rat models of heart, kidney, lung, and small bowel transplantation in which infection with rat cytomegalovirus (RCMV) significantly decreased the mean time to graft failure while concomitantly increasing the degree of vasculopathy within the allograft tissue 
Neovascularization is a complex, multi-step process of angiogenesis that rapidly takes place in response to inflammation and tissue injury, and involves many cell types, cytokines, chemokines, and proteases that work in concert to form new blood vessels from existing blood vessels. In brief (reviewed in 
), angiogenesis is initiated by release of pro-angiogenic factors from activated endothelial cells and tissue-resident macrophages, followed by removal of pericytes that surround the existing blood vessels. This results in the breakdown of the basement membrane of the existing blood vessel wall through activation of several proteases including matrix metalloproteinases (MMPs). The release of extracellular remodeling proteins during continued degradation of the blood vessel wall leads to the release of growth factors that promote endothelial cell migration toward the angiogenic stimulus and ultimately mediates endothelial cell proliferation that drives the formation of neotubules. These neotubules in turn release additional growth factors such as platelet-derived growth factor (PDGF) that recruit vascular smooth muscle cells and pericytes that stabilize the newly formed blood vessel. Importantly, pro-angiogenic M2 macrophages have been shown recently to act as bridging cells that promote the fusion of neotubules into one continuous blood vessel 
. Cytomegalovirus infection could therefore enhance neovascularization at various stages of angiogenesis through a number of direct and indirect mechanisms.
Monocytes are the primary target in vivo for HCMV (and MCMV and RCMV) infection 
. They serve as a site for virus latency and persistence 
, and help to disseminate virus throughout the host including the vasculature. When virus-infected monocytes enter the vasculature, they mature, and during the maturation process to become macrophages, they initiate an activation program that also serves to stimulate virus replication 
. In this manner, infected macrophages may disseminate virus to other cells of the vasculature that are involved in angiogenesis and vascular disease. These include endothelial cells, smooth muscle cells, pericytes, and fibroblasts 
. Given this complexity, the precise temporal relationship between virus infection of individual cell types and disease pathogenesis remains obscure and difficult to determine. Nonetheless, it is known that HCMV infection of endothelial cells induces the expression of adhesion molecules ICAM-1 and VCAM-1 
that serve to magnify transendothelial cell migration of inflammatory cells including monocytes. These monocytes become resident macrophages that promote angiogenesis by secretion of VEGF and other pro-angiogenic factors such as IL-6 
During virus replication, the HCMV-encoded chemokine receptor US28 also plays a prominent yet multifaceted role in angiogenesis. Firstly, US28 has been shown to stimulate VEGF production directly by induction of COX-2 via activation of the NF-κB pathway 
. Secondly, this HCMV-encoded chemokine receptor promotes the migration of macrophages in response to the CX3CL1 chemokine Fractalkine 
, a function that may help to attract additional HCMV-infected macrophages to areas of inflammation and thereby amplify angiogenesis. Thirdly, US28 also promotes the migration of vascular smooth muscle cells 
, but does so by binding to CC-chemokines and not Fractalkine 
. Thus, US28 appears to stimulate the migration of both macrophages and vascular smooth muscle cells, but in a ligand-dependent manner. Whereas US28-induced migration of macrophages takes place after ligation with Fractalkine, but not CC-chemokines, US28-induced migration of vascular smooth muscle cells is mediated by binding to CC-chemokines, but not Fractalkine. Since HCMV-encoded US28 apparently plays multiple roles in promoting angiogenesis, we postulate the same is true for M33, the MCMV homologue of US28 
. Ongoing studies are therefore oriented toward testing the hypothesis that MCMV-encoded M33 plays significant roles in the pathophysiology and increased severity of CNV during chronic MCMV infection.
Additional direct and indirect mechanisms by which cytomegalovirus might contribute to angiogenesis and vascular disease are suggested by other studies. Examples include studies that have shown that HCMV infection induces a reduction of endothelial nitric oxide synthase activity commonly observed during cardiovascular disease 
; RCMV induces the stimulation of a number of proteases including MMPs that are involved in degradation of the basement membrane required during the angiogenesis process 
; HCMV induces an upregulation of a number of cellular chemokines including macrophage inflammatory protein 1 alpha (MIP1-α), MIP1-β, RANTES, and IL-2 that play critical roles in angiogenesis and development of vascular disease 
; and HCMV infection of coronary artery smooth muscle cells stimulates VEGF expression 
Since angiogenesis in health and disease is a process of great complexity that offers a number of mechanisms by which cytomegalovirus infection of multiple cell types might serve as a stimulatory cofactor in the development of more severe choroidal neovascularization, we elected to focus our study on a possible role for macrophages during chronic systemic MCMV infection. Macrophages can be either pro-angiogenic or anti-angiogenic depending on their polarization phenotype 
that is regulated by the cytokine patterns encountered by macrophages within the resident tissue milieu 
. Classically activated macrophages, or M1 macrophages, exhibit an anti-angiogenic phenotype and produce high amounts of IL-12, IL-23, IL-6, and TNF-α, but low amounts of IL-10 
. In comparison, alternatively activated macrophages, or M2 macrophages, exhibit a pro-angiogenic phenotype and produce high amounts of IL-10, but low amounts of pro-inflammatory cytokines such as IL-6 and TNF-α 
. Moreover, M1 macrophages inhibit angiogenesis by inducing a cell-death program in endothelial cells, whereas M2 macrophages promote angiogenesis by stimulating production and release of pro-angiogenic factors such as VEGF that encourage endothelial tip cell formation 
. In this regard, Fantin and coworkers 
have recently made the extraordinary observation that M2 macrophages may also play a critical role during formation of new blood vessels by serving as bridge cells to properly position and fuse neotubules into one continuous blood vessel, possibly via activation of the DII4-a ligand and expression of Notch receptors 
. Thus, cytomegalovirus infection of monocytes and macrophages may influence angiogenesis-related activities by several possible mechanisms. For example, HCMV infection of monocytes appears to influence the polarization phenotype of the activated macrophage by modulating in a selective manner many M1/M2-associated factors 
, thereby inducing angiogenesis through stimulation of VEGF production and other angiogenic factors. Importantly, MCMV-infected IC-21 mouse macrophages exhibited a pro-angiogenic M2 phenotype in our studies. Alternatively, HCMV infection could conceivably have a detrimental on the normal angiogenic process by promoting inflammation. HCMV infection of endothelial cells may also enhance the stability of newly formed blood vessels through stimulation and release of several cytokines and growth factors including the Notch 2 receptor 
. We therefore postulate that chronic MCMV infection results in more severe choroidal neovascularization in our study by driving monocytes toward a M2 macrophage phenotype that favors angiogenesis through stimulation and release of pro-angiogenic factors that includes VEGF. It has not escaped our attention, however, that chronic MCMV infection might also cause more severe choroidal neovascularization by direct or indirect mechanisms associated with endothelial cell infection, a focus of future studies.
Splenic macrophages collected from chronically infected mice with the most severe choroidal neovascularization in our study showed significant increases in the amounts of transcripts to MMP-9 and COX-2, two proteins known to be involved in angiogenesis 
. The most dramatic increase in transcript level, however, was observed for that of VEGF, a critical pro-angiogenic factor. This observation was confirmed in a second independent animal study by us that demonstrated an even greater increase in VEGF transcript production in splenic macrophages collected from chronically infected animals. One interpretation of these reproducible findings is that when chronically infected monocytes are recruited to choroidal sites of laser-induced damage, their activation programs are initiated and oriented toward the pro-angiogenic M2 phenotype. Since they are also chronically infected with MCMV, this activation program stimulates virus replication, an event that leads to enhanced production and secretion of several pro-angiogenic factors including VEGF. Inoculation of cultures of human foreskin fibroblasts or cultures of coronary artery smooth muscle cells with HCMV has been shown to result in stimulation of functionally active VEGF production 
. It is therefore not surprising in the present study that MCMV infection of cultures of IC-21 mouse macrophages significantly stimulated production of VEGF mRNA and VEGF protein. Additional observations made during immunostaining studies also demonstrated that VEGF is indeed produced in high amounts by MCMV-infected IC-21 mouse macrophages, especially those in the early stages of cytopathology during plaque formation.
Of particular interest, however, was the additional observation that monolayer cells too early to be infected with MCMV (given the low multiplicity of infection used) were also VEGF-positive, an observation suggesting the attractive hypothesis that uninfected bystander macrophages might also be stimulated by adjacent MCMV-infected macrophages to produce enhanced amounts of VEGF during virus infection. Thus, MCMV infection of resident macrophages of tissues of the lung and spleen, and even bone marrow cells, could conceivably contribute to macrophage activation during chronic infection. MCMV-infected bone marrow cells, especially stromal cells, could favor a pro-angiogenic microenvironment that induces bystander activation during development within the marrow since stromal cells serve as a substrate upon which monocytes are induced to differentiate 
. In addition, due to their high vascularity, both lung and spleen experience high monocyte traffic, and chronic MCMV infection of these tissues could induce bystander activation. We therefore postulate that chronic MCMV infection of monocytes and macrophages distant from the eye serves as an important mechanism for macrophage activation of the M2 phenotype that would contribute to the pro-angiogenic microenvironment of the choroidal tissues of the eye.
Treatment with ganciclovir, a potent inhibitor of active cytomegalovirus replication and HCMV disease in the clinical setting 
, has been shown to delay the time to development of allograft rejection in heart transplant recipients 
, a finding that underscores the importance for active HCMV replication in acceleration of vascular disease. Additional studies using experimental rat transplant models have provided similar data showing that ganciclovir therapy also reduced or prevented RCMV-associated acceleration of tissue rejection when compared with RCMV-infected animals not treated with the antiviral 
. Since we hypothesize that MCMV infection of macrophages plays a central role in amplifying the severity of experimental choroidal neovascularization in mice by stimulation of pro-angiogenic factors including VEGF, we used a similar antiviral approach to demonstrate that VEGF-specific transcript production by splenic macrophages collected from chronically infected mice was indeed ganciclovir-sensitive. This outcome strongly supports the need for active MCMV virus replication in stimulation of production of pro-angiogenic factors such as VEGF that is required for increased severity of choroidal neovascularization. These in vivo findings were duplicated and extended in culture using ganciclovir-treated, MCMV-infected monolayers of IC-21 mouse macrophages, and in a relatively dose-dependent manner.
In summary, the findings reported herein using an experimental mouse model of CNV serve to clarify our previous seroepidemiologic clinical study in which a significant association was identified between high titers of anti-HCMV IgG and development of neovascular AMD 
. The presence of high anti-HCMV titers may indicate a subset of patients who harbor a greater total body burden of chronic HCMV infection, or who have experienced a recent, significant reactivation event. In either case, we hypothesize that the blood load of circulating HCMV-infected monocytes would be exceptionally high in this subset of patients. Upon recruitment to sites of drusen formation in patients who manifest the dry form of AMD, HCMV-infected monocytes would mature into tissue-resident macrophages with active virus replication, become polarized toward the pro-angiogenic M2 phenotype, and become a major source for production of a number of pro-angiogenic factors including VEGF that would amplify choroidal neovascularization associated with the wet form of AMD. We therefore believe that HCMV infection should be considered as a heretofore unrecognized risk factor for development of neovascular AMD. If true, subsets of patients who harbor a low virus load of HCMV would be predicted to experience decreased onset and progression of choroidal neovascularization, an occurrence that would impact their clinical outcome in terms of time of onset of visual loss and degree of visual loss. It is therefore possible that antiviral treatment might be effective in suppressing choroidal neovascularization associated with wet AMD in a fashion similar to that for suppression of allograft rejection in heart transplant recipients. Future studies will be oriented toward this investigation.