Choroidal neovascularization (CNV) is a known cause of age-related macular degeneration (ARMD). Moreover, the most common cause of blindness in the elderly in advanced countries is ARMD with CNV. It has recently been shown that bone marrow cells (BMCs) can differentiate into various cell lineages in vitro and in vivo. Adults maintain a reservoir of hematopoietic stem cells included in BMCs that can enter the circulation to reach various organs in need of regeneration. It has recently been reported that endothelial progenitor cells (EPCs) included in BMCs are associated with neovascularization. We examine the role of BMCs in CNV using a model of CNV in adult mice. Using methods consisting of fractionated irradiation (6.0 Gy × 2) followed by bone marrow transplantation (BMT), adult mice were engrafted with whole BMCs isolated from transgenic mice expressing enhanced green fluorescent protein (EGFP). Three months after BMT, we confirmed that the hematopoietic cells in the recipients had been completely replaced with donor cells. We then carried out laser photocoagulation to induce CNV in chimeric mice (donor cells >95%). Two weeks after the laser photocoagulation, by which time CNV had occurred, immunohistochemical examination was carried out. The vascular wall cells of the CNV expressed both EGFP and CD31. These findings indicate that newly developed blood vessels in the CNV are derived from the BMCs and suggest that the inhibition of EPC mobilization from the bone marrow to the eyes could be a new approach to the fundamental treatment of CNV in ARMD.
Recently, bone marrow cells (BMCs) have been used as a source of several kinds of mature cells in research for regenerative medicine. It has been reported that BMCs differentiate into various types of cells, including hepatocytes [1, 2], epithelial cells  (in the stomach, esophagus, small intestine, large intestine, and bronchus), cardiac muscle , and skeletal muscle . It has also been reported that BMCs differentiate into neural cells and astrocytes in vitro [6, 7], and also into astrocytes in vivo when BMCs are transplanted into the normal [8, 9] or ischemic  brain. Moreover, the intravenous injection of BMCs into mice has been shown to induce neural differentiation in the brain [11–13]. BMCs have the capacity to differentiate into myelin-forming cells in vivo and to repair demyelinated spinal cord axons . We also have demonstrated that BMCs differentiate into retinal neural cells in the injured rat retina . Adult BMCs, which contain hematopoietic stem cells (HSCs), have the ability of self-renewal and provide individuals with mature hematopoietic cells throughout their life. Therefore, BMCs are utilized in the treatment of bone marrow failure states, including hematological malignancies (leukemia and lymphoma). Recently, it has been reported that endothelial progenitor cells (EPCs) are derived from the bone marrow [16, 17], and that EPCs can differentiate into endothelial cells of blood vessels in the case of myocardial ischemia and hindlimb ischemic models [18–21]. Thus, bone marrow-derived EPCs significantly contribute to blood vessel formation [22, 23]. A recent report has shown that retinal neovascularization is provided by HSCs . Choroidal neovascularization (CNV) is a known cause of age-related macular degeneration (ARMD); ARMD is the most common cause of new blindness in the elderly in advanced countries. It has been reported that the mechanisms underlying CNV are completely different from mechanisms underlying retinal neovascularization ; newly formed blood vessels in CNV have been shown to be provided by the existing choroidal vessels through a broken Bruch's membrane [25, 26]. In this report, we use an experimental mouse model of CNV induced by krypton laser photocoagulation [27, 28] and clarify whether BMCs contribute to newly generated blood vessels in CNV.
Materials and Methods
Inbred male C57BL/6J (B6:H-2kb) mice were purchased from CLEA Japan (Osaka, Japan; http://www.clea-japan.co.jp). Enhanced green fluorescence protein (EGFP) transgenic mice (C57BL/6 background, kindly provided by Dr. Okabe, University of Osaka) were bred in our animal center . The strain carries EGFP driven by chicken β-actin promoter and cytomegalovirus intermediate-early enhancer. Most cell types in this animal express GFP. All mice used were 8–10 weeks of age and treated according to the Association for Research in Vision and Ophthalmology Resolution on the Use of Animals in Research. Each mouse was anesthetized by intramuscular ketamin injection before all surgical procedures.
Bone Marrow Transplantation (BMT)
BMCs were flushed from the femoral and tibial bones of the male EGFP mice and then suspended in phosphate-buffered saline (PBS; pH 7.2). The BMCs were then filtered through 70-μm nylon mesh (Becton Dickinson Labware; Franklin Lakes, NJ; http://www.bd.com), washed, and adjusted to 1.0 × 108 cells/ml with saline (n = 10).
Recipient mice (n = 10; C57BL/6J) were irradiated twice (6.0 Gy × 2, 4-hour interval on the day before BMT) from a cesium 137 source (Gamma cell 40 Exactor, Nordion International Inc.; Kanata, Ontario, Canada; http://www.mds.nordion.com), and whole BMCs (3.0 × 107 cells/0.3 ml) from EGFP mice were injected into the tail vein.
Induction and Analysis of CNV
Three months after BMT, laser photocoagulation was performed on the eyes of chimeric mice (donor cells >95%) (10 mice, 3 spots on one eye of each mouse = 30 spots) using krypton laser photocoagulation (wavelength: 647 nm; spot size: 100 μm; power: 0.1 W, duration: 0.1 second; Coherent; 920k; http://www.cohr.com) through a contact lens for mice (Unicon; Osaka, Japan). All the laser spots (n = 30) perforated Bruch's membranes and induced CNV. Two weeks after laser photocoagulation, one eye on each of five mice was enucleated and embedded in optimal cutting temperature compound (Miles; Elkhart, IN) after adjustment of the horizontal plane parallel to the cutting plane, and 10 μm frozen sections including CNV were made in a cryostat. The specimens, which had been fixed with 4% paraformaldehyde, were stained with phycoerythrin (PE)-conjugated anti-CD31 (1:10; Caltag Labs; Burlingame, CA; http://www.caltag.com). The stained specimens were observed using a confocal microscope (Fluoview, Olympus; Tokyo, Japan; http://www.olympus.com). Some frozen serial sections (10 μm) were histochemically stained with biotinylated Griffonia simplicifolia lectin B4 (GSA; Vector Laboratories; Burlingame, CA; http://www.vectorlabs.com), which selectively binds to vascular cells. Slides were incubated in methanol/H2O2 for 10 minutes at 4°C, washed with 0.05 mol/L Tris-buffered saline (TBS, pH 7.6), and incubated for 2 hours at room temperature with biotinylated GSA. After rinsing with 0.05 mol/L TBS, slides were incubated with avidin coupled with alkaline phosphatase (Vector Laboratories) for 45 minutes at room temperature. After being washed for 10 minutes with 0.05 mol/L TBS, slides were incubated with simple stain AEC (Nichirei; Tokyo, Japan; http://www.nichirei.com) to produce a red reaction product that is distinguishable from melanin.
The other mice (n = 5) were perfused with 1.0 ml PBS containing 50 mg/ml tetramethylrhodamine isothiocyanate (TRITC)-conjugated dextran (160,000 Da average molecular weight; Sigma; St. Louis, MO; http://www.sigmaaldrich.com), which was administered through the left ventricle to perfuse the ocular vessels. Immediately after perfusion, one eye from each of the five mice was removed and fixed for 1 hour in 10% phosphate-buffered formalin. The cornea and lens were removed and the entire retina was carefully dissected and removed from the eyecups. Next, choroids were dissected and mounted flat with mounting media (Aquamount, BDH Laboratories; Poole, UK; http://www.bdh.com) for examination using a confocal microscope (n = 15 spots).
Surface Marker Analysis in Peripheral Blood
Surface markers of peripheral blood in the recipient mice were analyzed. After washing twice with PBS, the cells were suspended in 10 ml of PBS containing 1% fetal bovine serum and placed on 10 ml of Lympholyte-Mammal density solution (1.0860 g/ml, Cedarlane Laboratories Ltd.; Hornby, Ontario, Canada; http://www.cedarlanelabs.com). After centrifugation for 30 minutes at 2,000 rpm at room temperature, these mononuclear cells were collected from the defined layer at the interface. To detect donor-derived or residual recipient-derived cells, these cells were stained with PE-conjugated monoclonal antibodies against B220, CD3, Mac-1, and Gr-1 (PharMingen; San Diego, CA; http://www.bdbioscience.com/pharmingen) to analyze the cell surface phenotypes. These cells were analyzed by fluorescent-activated cell sorting FACScan® (Becton Dickinson).
Analysis of Chimerism
To determine if BMCs contribute to CNV, we prepared complete bone marrow chimeric mice . C57BL/J recipient mice were lethally irradiated (6.0 Gy × 2 with a 4-hour interval the day before BMT), and whole BMCs from EGFP mice were transplanted intravenously via the tail vein. Three months after BMT, we examined the chimerism of recipients using a FACS scan (Fig. 1). More than 95% of the hematopoietic cells in the peripheral blood and hematolymphoid organs such as the spleen (data not shown) were positive for green fluorescence (donor derived). Thus, we confirmed the complete chimerism of the hematopoietic cells in the recipient mice.
Generation of Blood Vessels in CNV from Donor BMCs after BMT
After establishing complete chimerism, the mice were treated with laser photocoagulation to induce CNV of the ARMD model in one eye per mouse (three spots per eye). One eye from each of 10 mice was prepared in this experiment, and CNV was induced in all of the laser spots (n = 30). Two weeks after laser photocoagulation, one eye from each of 5 mice was enucleated under anesthesia and frozen sections were prepared. These frozen sections were stained with a PE-conjugated anti-CD31 antibody and vascular cell-selective lectin [31, 32], which is known as a specific marker for endothelial cells of blood vessels in CNV. A minimum of 30 sections per choroid, with sampling on both sides of the optic nerve, were examined. In the sections stained with vascular cell-selective lectin, Bruch's membrane was broken by krypton laser photocoagulation, and new blood vessels were generated in the CNVs (Fig. 2).
To examine the pattern of vascular development in CNV, one eye from each of the other five mice was perfused with red fluorescent-labeled dextran (n = 5) and was observed in the flat-mount preparation of choroids. The CNV in the laser-photocoagulated lesions of the mounted choroids was observed using confocal microscopy. EGFP-positive cells existed along the blood vessels, and the shapes of those cells were compatible with endothelial cells; EGFP-positive cells existed along the red-fluorescence-positive lumen of the CNV (n = 15 spots; Fig. 3). As shown in Figure 4, the CNVs were stained with PE-conjugated anti-CD31, and over 70% of EGFP-positive cells, which were in the CNVs, were also positive for CD31 (n = 15 spots). These results suggest that EGFP-positive donor BMCs differentiate into CD31-positive endothelial cells, and that these endothelial cells constitute functional blood vessels.
It has been reported that EPCs in the peripheral blood are derived from BMCs [16, 17] and that EPCs contribute to neovascularization in various organs [18–23]. In the present study, we have shown that some of the newly generated blood vessels in the CNV are derived from BMCs. CNV develops when the Bruch's membrane is broken because of aging, trauma, and other reasons . Although it has been reported that newly generated blood vessels and endothelial cells in CNV are derived from the existing choroidal vessels in animal models produced by laser photocoagulation , the origins of endothelial cells in CNV have not been identified. However, in the present study, we have confirmed that some of the endothelial cells in CNV are derived from BMCs.
One possible criticism of the results shown in Figure 4 is that white blood cells may be contributing to the green signal at the blood vessel. However, we emphasize that the GFP-positive cells in Figure 4 were not influenced by white blood cells, since EGFP-positive cells existed along the blood vessels and the shapes of these cells were compatible with the endothelial cells in Figure 3 perfused by TRITC-conjugated dextran.
Generally, fibrovascular membranes exist around the periphery of CNV . It has been reported that fibrovascular membranes with CNV produce vascular endothelial growth factor (VEGF), which accelerates neovascularization [25, 33–36]. On the other hand, it has been reported that VEGF augments the mobilization of EPCs from the bone marrow into the peripheral blood and promotes neovascularization [37–39]. On the basis of these reports, we predicted that EPCs would be mobilized from BMCs to choroids, and fibrovascular membranes expressing VEGF would help differentiate the EPCs into CNV.
To our knowledge, this is the first report to demonstrate the differentiation of newly generated blood vessels in CNV from BMCs. CNV is known as a cause of ARMD and is the most common cause of new blindness in the elderly in advanced countries. No fundamental treatment for CNV in patients with ARMD has yet been identified. CNV is generally treated with laser photocoagulation or vitreous surgery, but these procedures do not permit the recovery of visual acuity. Therefore, there has been a need to clarify not only the pathophysiology but also the mechanisms of ARMD with CNV in order to develop a fundamental treatment. The present study strongly suggests that the inhibition of EPC mobilization from the bone marrow into the eyes could be a new method for the fundamental treatment of ARMD with CNV.
We thank professor M. Okabe (Osaka University) for the donation of EGFP-transgenic mice. We also thank Hilary Eastwick-Field and K. Ando for the preparation of this manuscript.