Effect of Coadministration of Neuronal Growth Factors on Neuroglial Differentiation of Bone Marrow–Derived Stem Cells in the Ischemic Retina

Abstract

Purpose.: Brain-derived neurotrophic factor (BDNF) and ciliary neurotrophic factor (CNTF) have limited and transient supportive effects on retinal recovery from ischemia. The aim of this study was to investigate their effect on engrafted adult bone marrow–derived stem cells in a rodent model of anterior ischemic optic neuropathy (rAION).

Methods.: Small cells were isolated from the bone marrow of green fluorescent protein expressing mice by counterflow centrifugal elutriation, depleted of cells expressing lineage markers, and grafted in conjunction with growth factors into the vitreous body of mice with unilateral rAION. Progenitors were mobilized with granulocyte macrophage colony-stimulating factor (GM-CSF) or stem cell factor (SCF). The contralateral eye served as a control.

Results.: At 4 weeks, the quantitative incorporation of donor cells in the injured retina was increased by BDNF (P < 0.01 versus control) and decreased by CNTF (P < 0.01 versus control), with no notable difference at 24 weeks. Both growth factors improved the short-term and long-term qualitative engraftment of cells adopting neural phenotypes in the retinal ganglion cell (RGC) layer and astrocyte phenotypes in the anterior vasculature. The RGC-engrafted cells formed extensions toward the inner nuclear layer. In the presence of growth factors, donor cells migrated to the optic nerve and contributed to repair by gliosis. Mobilization with GM-CSF restricted cell fate to microglia, whereas SCF was associated with limited neuroglial differentiation.

Conclusions.: Both BDNF and CNTF enhance engraftment and neuroglial differentiation of adult bone marrow stem cells in injured retina, with BDNF having an early quantitative and qualitative advantage over CNTF. Mobilization with differentiation factors restricts cell fate in the injured retina.

Introduction

Researchers have designed several neuroprotective strategies in the study of optic nerve injury in rat models. ^1–7 These include the use of brain-derived neurotrophic factor (BDNF), ciliary neurotrophic factor (CNTF), and bFGF. ^2–4,7 This practice is based on the rationale that endogenous growth factors are upregulated in tissues following ischemic and mechanical insult to the optic nerve. ^8–10 The intravitreal administration of a single dose of BDNF before (5–6 days) and after (1–3 days) axotomy was shown to support the early survival of retinal ganglion cells (RGCs). ³ Maximal protection was attained when BDNF was infused 1 day after ischemia. ⁴ Repeated BDNF injections further improved RGC survival ^2–4 and triggered axonal sprouting around the optic disc. However, protection declined to undetectable levels by 6 weeks after therapy. ³ Similar findings were observed for CNTF in models of pressure-induced ischemic injury or intraorbital optic nerve transection, ^2,4 with protection vanishing after 5 to 7 weeks. There was no evident beneficial synergism between BDNF and CNTF. ²  

The limited efficacy of neuroprotective agents to rescue vision ^2–4 has directed attention to the possibility of cell-mediated delivery of growth factors to achieve sustained expression during the remodeling process ^11–13 given that growth factors also function as modulators during the differentiation of various types of stem cells and neural progenitors. ^14–18 Support was provided by in vitro studies showing that BDNF and CNTF enhance the differentiation of neural ^18–21 and retinal ^22,23 progenitors. The aim of the present study was to assess the potential of BDNF and CNTF to direct and enhance the glial and neural differentiation of small primitive cells derived from the bone marrow in situ in a rodent model of anterior ischemic optic neuropathy (rAION) developed by our group. ^24–26  

The pathophysiology of optic nerve stroke in our model is simulated by induction of acute microvascular damage in the retroglobal portion of the nerve using selective argon laser photoactivation of intravenously injected rose bengal with a plastic contact lens. ^24,25 This procedure causes secondary degeneration of the axonal cell bodies (RGCs), while maintaining a patent retinal vasculature so that cellular grafts can be trafficked to the retina. ²⁶ The subset of adult bone marrow cells used in this study is known to possess a capacity for long-term multilineage hematopoietic reconstitution. ^27,28 In the presence of injury, cells can differentiate into liver, ²⁹ the endocrine system and pancreas, ³⁰ and multiple epithelial tissues. ³¹ The cells were isolated by counterflow centrifugal elutriation. ³² The retinal migration of intravitreally or intravenously injected bone marrow stem cells has been recently described. In one study, ²⁶ the cells dispersed throughout the retina but were incorporated mainly in areas of damage and RGC loss, significantly contributing to glial remodeling. In another study, ³³ elutriated cells were induced to differentiate in situ by activation of Flt-1 following intravitreal administration of VEGF. Our data indicate that both BDNF and CNTF impact retinal remodeling by quantitatively enhancing neural and glial cell incorporation. 

Methods

Isolation of Bone Marrow Cells by Counterflow Centrifugal Elutriation

Nucleated bone marrow cells were harvested from femurs and tibias and loaded onto a counterflow elutriation centrifuge (Beckman, Fullerton, CA) equipped with a J-6 rotor. ^26,28,30 The elutriation solution consisted of isotonic saline containing 5.5 mM glucose, 0.26 mM EDTA, and 0.5% (w/v) BSA. Fractions were collected at increasing medium flow rates at a constant speed of the rotor (867 g) in order to isolate the smallest subset of nucleated cells at a flow rate of 25 mL/min (Fr25). Large rotor off (R/O) progenitors were collected after arrest of the elutriation centrifuge. Cells expressing lineage markers were depleted using saturating amounts of antibodies against T cells (CD5), B lymphocytes (B220), granulocytes (GR-1), and macrophages (Mac-1) obtained from hybridoma cell lines (ATCC, Manassas, VA) and commercially available Ter-119 antibodies against erythroid cells (eBiosciences, San Diego, CA). Cells coated with monoclonal antibodies (mAbs) were washed twice with PBS and incubated with sheep anti-rat IgG conjugated to M-450 Dynal magnetic beads (Invitrogen, Oslo, Norway) at a ratio of four beads per cell. Rosetted cells were precipitated by exposure to a magnetic field to collect the supernatant containing lineage-negative (Fr25lin⁻) cells. The efficacy of isolation was assayed by flow cytometry (Vantage SE flow cytometer; Becton Dickinson, Franklin Lakes, NJ), and contamination was determined using primary labeled mAbs against the various lineages. ²⁶  

Animals

The following mice were purchased from The Jackson Laboratory (Bar Harbor, ME): C57Bl/6J (B6, CD45.2), B6.SJL-Ptprc^a Pepc^b/BoyJ (CD45.1), and C57BL/6-TgN(ACTbEGFP)1Osb (GFP, CD45.2). Mice were maintained and handled in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the National Institutes of Health guidelines. Animal protocols were approved by the Institutional Animal Care Committee at Rabin Medical Center, Israel. 

Cell Grafting

Elutriated Fr25lin⁻ cells were grafted via the tail vein (1 × 10⁶ in 200 μL PBS [n = 37]) or intravitreally (2 × 10⁵ in 2 μL PBS [n = 32]). ³³ After penetrating the sclera immediately behind the limbus in a slightly oblique direction, the beveled 30-gauge needle tip of the microsyringe (Hamilton, Reno, NV) was placed intravitreally in the retrolental space of the eye under direct ophthalmoscopic control with an operating microscope (Opmi 6S; Carl Zeiss Microscopy, Göttingen, Germany). Preliminary studies showed that because of the beveled configuration of the needle some vitreous outflow occurs during its placement, greatly reducing the risk of drug reflux. In our hands, 2 μL could be reliably injected intravitreously without reflux during the injection or apparent inflammation of the eye after surgery. 

Induction of Optic Neuropathy and Administration of Growth Factors

A rodent model of anterior ischemic optic neuropathy was induced by intravenous injection of 40 μL 2.5 mM rose bengal dye, followed by selective photoactivation above the optic disc using an argon laser beam at 532 nm. Injury was evaluated by counting the RGCs in tissue sections; a 35% to 40% RGC loss was observed as previously reported for this reproducible procedure. ^24,25 One day after induction of ischemic injury, isolated bone marrow cells (2 × 10⁵) were coadministered with 5 μg/2 μL BDNF (n = 13) or CNTF (n = 8) (Serotec, Raleigh, NC) into the vitreous body as described previously.^34–40 Alternatively, the cells were transplanted intravenously, with simultaneous injection of BDNF (n = 12) or CNTF (n = 6) intravitreally. These 39 mice were evaluated at 4 weeks. An additional 12 mice after rAION induction and intravenous (n = 6) or intravitreal (n = 6) bone marrow–derived stem cell transplantation and intravitreal injection of BDNF (n = 6) or CNTF (n = 6) were evaluated at 24 weeks. Thirty-eight mice underwent rAION induction with cell transplantation (16 intravenously, 10 intravitreally, and 12 R/O) without injection of growth factor. We used an established model of intravitreal injection of growth factors behind the limbus in order to prevent mechanical injury to the retina by the procedure itself; studies ^34–36 have shown that the vitreous volume of mice is 20 μL such that injection of 2 μL solution is well tolerated. Furthermore, to refine our technique, we conducted preliminary experiments consisting of trypan blue injection into the vitreous body of euthanized mice using a 30-gauge needle. In all experiments, injury was induced in one eye, and the contralateral eye served as a control. 

Mobilization of Bone Marrow Progenitors

Radiation chimeras expressing green fluorescent protein (GFP) in hematopoietic cells were prepared 3 months before induction of injury. ³³ Briefly, bone marrow cells from mice constitutively expressing GFP were transplanted into 28 wild-type C57/B6 mice. Cells were mobilized with five consecutive intraperitoneal injections of 15 μg/kg granulocyte macrophage colony-stimulating factor (GM-CSF) (Serotec) (n = 16) or neutralizing antibodies against stem cell factor (SCF) (Serotec) (n = 12) on days −3 to +1 of rAION injury (day 0). 

Histology and Immunohistochemistry

Mice were killed at 4 weeks or 24 weeks after cell grafting or at 4 weeks after rAION induction with mobilization of cells in the chimeras, and the tissues were histologically and immunohistochemically analyzed. ⁴¹ The eye bulb was enucleated, fixed in 4% paraformaldehyde overnight, and transferred to 20% sucrose overnight at 4°C. Samples were embedded in frozen optimum temperature cutting compound medium (Tissue-Tek; Sakura, Finerek, CA) and cryosectioned at 6-μm thickness using a Cryotome (Termo Shandon, Cheshire, England). Sections were blocked for 15 minutes at room temperature with PBS containing 2% BSA and 0.5% Triton X-100, followed by several washes. Every 10th slide from each eye was selected for staining for the following tissue markers by overnight incubation at 4°C: GFP (rabbit polyclonal, 1:100; Santa Cruz Biotechnology, Santa Cruz, CA), glial fibrillary acidic protein (GFAP) (mouse polyclonal, 1:250; Millipore, Billerica, MA), vimentin (chicken polyclonal, 1:500; Millipore), CD45 (rat, 1:50; Millipore), CD68 (rat,1:100; Abcam plc, Cambridge, England), and NeuN (mouse, 1:100; Millipore). Primary antibodies were counterstained with FITC-labeled donkey anti-rabbit (1:200; Jackson Immunoresearch, West Grove, PA) and goat anti-rat mAb (1:200; Santa Cruz Biotechnology), AlexaFluor 568–conjugated donkey anti-goat (1:500), and Cy3-conjugated rat anti-goat (1:200, Molecular Probes; Invitrogen, Grand Island, NY). Biotinylated primary antibodies were counterstained with Cy3-conjugated streptavidin (1:400) and Cy5-conjugated streptavidin (1:500), both from Jackson Immunoresearch. 

Microscopy

Images were acquired with an Axioplan 2 fluorescence microscope equipped with an ApoTome using a AxioVision 4.5 software and an LSM confocal laser scanning microscope using LSM 510 software, all from Carl Zeiss Microscopy. To condition the illumination, we used narrow-band excitation with emission filters (rhodamine 540/580 nm, FITC-GFP 485/530 nm, and DAPI-Hoechst 360/455 nm) at ×10 to ×100 magnification. Images were pseudocolored (Photoshop; Adobe Systems, Incorporated, San Jose, CA) and reconstructed from red/blue/green layers. 

Quantitative Assessment of Cell Incorporation in the Injured Retina

To estimate the number of labeled cells in each eye, six to 15 nonoverlapping fields of view were assessed in every 10th slide at ×20 magnification. Each slide included three consecutive cryosections. The incorporation of transplanted cells was calculated by dividing the number of donor cells (GFP⁺) by the number of cells staining positive for tissue markers in the relevant layer. The total number of donor cells was calculated by multiplying the number of cells per section times the number of slides (n = 10) and correcting for the spherical shape of the retina using the following equation ³³ : πr² /4r² = π/4. The homing efficiency was calculated by dividing the total number of GFP⁺ cells by the number of infused cells; the results were expressed as percentages. ²⁶  

Statistical Analysis

All data are expressed as means (SEs). The results were statistically analyzed using SPSS for Windows version 15.0.1 (SPSS, Inc., Chicago, IL). Two-way within-subject ANOVA and between-subject ANOVA were used with repeated measures. P < 0.05 was considered significant. Because of the small sample size, we applied the nonparametric Mann-Whitney test for two independent samples. 

Results

Effect of Growth Factors on Incorporation of Fr25lin⁻ Cells by the Injured Retina

Simultaneous inoculation of Fr25lin⁻ GFP⁺ cells and BDNF into the vitreous body 1 day after rAION induction resulted in incorporation of the cells by the RGC layer accompanied by expression of nuclear NeuN (Fig. 1A) and cytosolic neurofilament (Fig. 1B). Other cells seeded in the anterior vascular layer and expressed GFAP (Fig. 1C) and vimentin (Fig. 1D). Notably, green GFP and red GFAP antibodies that colocalized in the cytosol appear yellow (Figs. 1B–D), although in some cases one of the markers was compartmentalized and dominant in different regions of the cell (GFP > neurofilament) (Fig. 1B). The presence of extensions of astrocytes and endothelium in the anterior vasculature (Fig. 1C) indicated that the NeuN⁺ RGC-incorporated cells identified by the cytosolic GFP expressed in the RGC layer were a true-positive finding. 

Figure 1

 

Qualitative and quantitative incorporation of Fr25lin⁻ cells in the injured retina: the short-term impact of BDNF. High-resolution analysis of Fr25lin⁻ GFP⁺ cell incorporation in the injured retina of syngeneic recipients (C57BL/6) following intravitreal inoculation in conjunction with BDNF. Images were reconstructed by superposition of pseudocolored RGB layers (scale bars: 20 μm). The cells are located within the RGC layer, independent from the injection site. (A) The GFP⁺ cells incorporated in the RGC layer with pyramidal morphology express nuclear NeuN and extend dendritic processes (white arrows). (B) The RGC-incorporated GFP⁺ cells express neurofilament (NFL) extending toward the deeper retinal layer (yellow arrows). (C) The GFP⁺ cells located in the anterior vascular layer adopt astrocyte morphology and express cytosolic GFAP. Colocalization of cytosolic GFP and GFAP antibodies pseudostained in red results in conversion to yellow (yellow arrows). (D) The GFP⁺ cells incorporated in the RGC layer (white arrows) are distinct from GFP⁺ astrocytes expressing vimentin (yellow arrows). (E) Positive identification of donor cells incorporated in the RGC layer by a genomic marker in male to female transplants shows five X⁺Y⁺ and three X⁻Y⁺ among 15 RGCs (white arrows indicate donor cells X⁺Y⁺), along with incorporation of male donor cells in the inner nuclear layer (INL). (F) Quantitative incorporation of intravitreally injected donor cells at 4 and 24 weeks after rAION injury in controls without growth factor injection (none) and under the local influence of BDNF (n = 13 [nine mice at 4 weeks and four mice at 24 weeks]) and CNTF (n = 8 [three mice at 4 weeks and five mice at 24 weeks]). Data are expressed as percentages of 2 × 10⁵ cells inoculated intravitreally.

Because there is reduced GFP expression (as a reporter protein) during differentiation, ⁴¹ particularly when cells adopt neuronal phenotypes, ⁴² we analyzed a genomic marker in sex-mismatched transplants. Donor cells in the uninjured retinas identified by the male Y chromosome disclosed a normal diploid genotype (Fig. 1E). The apparent correlation between seeding site and donor cell expression of markers characteristic of the particular layer suggested in situ induction of donor cell differentiation. 

Under the influence of CNTF, the site of donor cell incorporation was related to the expression of the characteristic phenotypic markers: cells incorporated in the RGC cells were positive for NeuN (Fig. 2C), and cells positioned in the anterior vasculature were positive for vimentin (Fig. 2D). In addition, the presence of donor cells expressing the panhematopoietic marker CD45 was associated with the injured vasculature, sites that also attract microglia of host origin (GFP⁻ CD45⁺) (Fig. 2E). The identification of RGC-engrafted donor cells that adopted a pyramidal cell morphology was not biased by juxtaposed immune cells involved in the removal of debris following injury (Fig. 2F). The negative staining for CD45 of both NeuN⁺ and vimentin/GFAP–positive cells ruled out the possibility that donor-derived GFP⁺ myeloid cells fused with injured retinal cells. 

Figure 2

 

Impact of growth factors on quantitative incorporation of Fr25lin⁻ cells at 4 weeks. Fractional distribution of cells in the injured retina at 4 weeks without growth factor injection (none) and after intravitreal injection of BDNF or CNTF, as determined by expression of neural markers in the RGC layer, astrocyte markers in the vascular layers, and CD45, an immunohematopoietic marker. (A) Without growth factor, more than 95% of the cells were glial. With BDNF, 35% expressed neuronal markers and with CNTF, 64%; rates for glial marker were 49% and 10%, respectively, and for inflammatory marker, 16% and 10%, respectively. (B) Absolute numbers of differentiated cells adopting neural and glial phenotypes under the influence of BDNF and CNTF, expressed as percentages of 2 × 10⁵ donor Fr25lin⁻ cells inoculated into the vitreous body. Neuronal marker expressed in 1991 cells in BDNF mice and in 531 cells in CNTF mice; the numbers for glial marker were 2788 and 216, respectively. (C–F) Representative images at 4 weeks after intravitreal transplantation of Fr25lin⁻ GFP⁺ cells in conjunction with CNTF (scale bars: 20 μm) demonstrate the following: (C) Incorporation in the RGC layer and expression of NeuN (white arrows). (D) Cytosolic vimentin converging with GFP⁺ in astrocytes located in the anterior vasculature (yellow arrows). (E) Cells adjacent to the injured layers expressed the panhematopoietic marker CD45. (F) Identification of RGC-engrafted GFP⁺ donor cells (white arrows) is not biased by clearance of debris by CD45⁺ donor and host immune cells and microglia.

To determine the quantitative effects of the growth factors, engraftment was calculated as the percentage of 2 × 10⁵ Fr25lin⁻ cells inoculated into the vitreous body or the percentage of 1 × 10⁶ cells injected intravenously. ³³ The results for intravenous and intravitreal cell transplantation with intravitreal injection of growth factors are summarized in Table 1. At 4 weeks, the addition of BDNF was associated with early integration of Fr25lin⁻ cells in the retina (P < 0.01 versus control eyes), whereas CNTF was associated with a decrease in quantitative donor cell incorporation (P < 0.01 versus control) (Fig. 1F). Analysis performed at 24 weeks (Table 2) showed an even distribution under the different experimental conditions, indicating that the growth factors had little influence on durable engraftment in the retina. 

Table 1

 

Cell Incorporation at 4 Weeks Following Exogenous Bone Marrow Transplantation With Intravitreal BNDF or CNTF Injection

Cell Incorporation at 4 Weeks,n = 39	Intravenous Transplantation, n = 18	Intravitreal Transplantation, n = 21
rAION with BDNF, n = 25	1979 (324)	5690 (3132)
rAION with CNTF, n = 14	2973 (1909)	831 (822)

 Data are expressed as means (SEs).

Table 2

 

Cell Incorporation at 24 Weeks Following Exogenous Bone Marrow Transplantation With Intravitreal BNDF or CNTF Injection

Cell Incorporation at 24 Weeks, n = 12	Intravenous Transplantation (1 × 106 Cells/ Retina), n = 6	Intravitreal Transplantation (2 × 105 Cells/ Retina), n = 6
rAION with BDNF, n = 6	2842 (677)	2937 (763)
rAION with CNTF , n = 6	2915 (838)	2516 (371)

 Data are expressed as means (SEs).

When the differences in the amount of cell incorporation between the growth factors were analyzed (Fig. 1F), BDNF was found to have a stronger effect than CNTF on early engraftment of donor cells (Table 1) but a lesser effect (P < 0.01) on cell conversion to neural phenotypes (Fig. 2A). The effect of BDNF also lasted longer than that of CNTF in terms of both neuronal and glial differentiation (Fig. 2B). 

Long-term Consequences of Growth Factors on Cell Engraftment

By contrast to the transient neuroprotective effect of growth factors on endogenous tissue reported in the literature, ^2–4 their initial impact on the small bone marrow–derived cells was sustained over an extended period. Despite the significant lineage-specific variations revealed at 30 days, the quantitative effect of BDNF and CNTF leveled off after 24 weeks (Fig. 1F and Table 2). At this point, the fractions of donor cells expressing neural markers were similar, and only immune phenotypes were elevated following treatment with BDNF (Fig. 3A). Qualitatively, both factors induced a substantial increase in the absolute number of cells that adopted a neural phenotype in the injured retina (Fig. 3B) compared with controls. The GFP⁺⁻ grafted cells expressing NeuN filled some of the gaps caused by the disruption of the retinal architecture, still evident at 24 weeks, and improved the confluence of this layer (Fig. 3C). Likewise, donor cells adopted the morphologic and phenotypic characteristics of astrocytes and considerably contributed to restoration of the cellularity of the anterior vascular layer (Fig. 3D). 

Figure 3

 

Long-term consequences of growth factors on cell incorporation in the injured retina. (A) Fractional distribution of retina-incorporated cells at 24 weeks after intravitreal inoculation of Fr25lin⁻ GFP⁺ cells in conjunction with growth factors. Neural marker was identified in 38% of engrafted cells in both groups; glial marker in 38% and 58%, respectively; and inflammatory marker in 23% and 4%, respectively. (B) Absolute rates of differentiated cells, as determined by expression of neural markers in the RGC layer, astrocyte markers in vascular layers. and CD45 (immune) adopting neural (1101 vs 956), glial (1101 vs 1459), and inflammatory (734 vs 100) phenotypes under the influence of BDNF and CNTF, respectively (NS). (C) Engraftment of donor cells in the RGC layer and expression of nuclear NeuN (white arrows) along astrocytes (yellow arrows [scale bar: 20 μm]). (D) Cells expressing vimentin in the anterior vascular layer (yellow arrows [scale bar: 20 μm]). (E) The RGC-engrafted cells communicate with cells within the inner nuclear layer (INL) (scale bars: 20 μm).

The grafted cells also incorporated in the proximal optic nerve (Fig. 4A). The dominant process of gliosis as a mechanism of durable neural tissue repair was supported by our analysis of the optic disc (Fig. 4B) and optic nerve (Fig. 4C), which showed a substantial differentiation of donor cells to the astrocyte phenotype, as indicated by positive staining for GFAP (optic nerve) and vimentin (optic disc). 

Figure 4

 

Gliosis of the injured optic nerve. (A) Cross-section of the optic nerve and central retina showing staining for protein kinase C (PKC) at 24 weeks after intravitreal cell grafting (scale bar: 60 μm), as well as demarcation of the location of B and C. Robust incorporation of GFP⁺ donor cells with astrocyte phenotypes at the site of optic disk injury. (B) Expression of GFAP in the anterior vasculature (scale bar: 40 μm). (C) Expression of vimentin in optic nerve astrocytes (scale bar: 20 μm).

Mobilization of Bone Marrow Progenitors

Mobilization with GM-CSF induced the growth of retinal microglia (Fig. 5A), identified by expression of CD68 (Fig. 5B), with no significant engraftment of cells expressing astrocyte or neural markers at 30 days. By contrast, mobilization with SCF resulted in scarce incorporation of bone marrow–derived astrocytes (Fig. 5C), with few cells expressing neural markers and considerably lesser appearance of bone marrow–derived (GFP⁺) microglia compared with BDNF. 

Figure 5

 

Cell incorporation in the injured retina following bone marrow mobilization. (A, B) Mobilization with GM-CSF results in ample appearance of microglia in the injured retina (scale bar: 30 μm) (A) identified by expression of specific markers (scale bar: 20 μm) (B). (C) Mobilization by SCF neutralization results in neogenesis of vimentin-positive astrocytes from bone marrow cells (scale bar: 20 μm).

Discussion

The introduction of bone marrow–derived stem cells into injured retina contributes to the revascularization of surviving cells through neogenesis of endothelium and astrocytes ^26,33 or neogenesis of RGCs and other ganglion components of the retina. The present study shows that local enrichment of the ocular environment with growth factors affects the engraftment of the injected stem cells. Specifically, cell engraftment was increased at 30 days with the addition of BDNF compared with uninjured control eyes and increased, but less than with BDNF, with the addition of CNTF. However, neither factor had a long-term quantitative effect. In addition, both growth factors had an early inductive effect on the development of neural and immune phenotypes. Concerns of a biased interpretation of the data as resulting from the fusion of donor immune cells with residual injured host cells were eliminated by the demonstration of diploid genotypes and distinct patterns of CD45 immunohematopoietic cell distribution. ^30,41 The detection of significant numbers of donor cells at 24 weeks after transplantation emphasized the durability of the engraftment. 

Local and systemic administration of hematopoietic progenitors promotes cerebral and retinal recovery from injury, ^43,44 with variable efficacy compared with retinal progenitors. ⁴⁵ The capacity of the grafted cells to incorporate in various layers of the injured retina and adopt morphologic and phenotypic features of particular lineages is consistent with their reported role in retinal remodeling: the assumption that this process is a result of in situ priming and induction of differentiation ^29–31 is supported by findings that cell engraftment in the retina can be attained only in the presence of injury, in response to strong signals emanating from the wounded stroma and hypocellular structures. ^26,33 Accordingly, irradiation, which induces competitive chemotaxis between the bone marrow and retina, is followed by the competitive distribution of cells between these units of reconstitution. The Fr25lin⁻ cells, which are part of the hematopoiesis compartment, engage in hematopoietic activity 2 to 3 months after transplantation. ^27,28,32 Hematopoiesis is the last differentiation trait assessed so far in the liver, pancreas, and retina ^29–31,33 and may be the final differentiation trait inhibited in this cell subset in order to maintain its role in the continuous process of hematopoiesis by the bone marrow. 

Studies have shown that the smallest fraction of bone marrow cells isolated by elutriation and depleted of lineage markers (Fr25lin⁻) incorporates in the injured retina ²⁶ and that better engraftment is attained following intravitreal infusion than with site-specific chemotaxis of these cells from the peripheral blood. ³³ However, the migration of locally grafted cells from the vitreous body across the membrane is a rate-limiting step of their incorporation into the retina. ⁴⁶ We previously observed cell migration in a model of retinal injury, ²⁶ but it did not exclude the possibility of cell trafficking through the patent local retinal vascular plexus. Cell migration from the vitreous body is not restricted to the retinal surface, as indicated by the detection of significant gliosis in the optic nerve. Recovery of the optic nerve is important because the cellular reconstitution of the RGC layer and its functional incorporation require that axonal extensions reach the relay ganglia and establish continuity of neural conduction. In the present study, the RGC-engrafted donor cells expressing neuronal markers were seen to extend and establish connections with the inner nuclear layer and glial cells of host and donor origin. ^26,33 Connections were detected (Fig. 3E) but not confirmed to be functional. If functional, the potential of bone marrow stem cell transplantation to repair impaired retinas is enormous. 

To obviate direct graft inoculation, we attempted to mobilize the bone marrow progenitors. The mobilization of endogenous precursors to support retinal recovery from injury might have significant clinical implications for human anterior ischemic optic neuropathy. Similar to our earlier study ³³ using VEGF for induction, we found that mobilization with GM-CSF resulted in myeloid induction and conversion of the mobilized progenitors into microglia. Mobilization with SCF resulted in the appearance of bone marrow–derived astrocytes and occasional engraftment in the RGC layer. Overall, injection of purified stem cells into the vitreous body yielded the best quantitative and qualitative engraftment and repair of the injured retina. 

Both BDNF and CNTF have a relatively short lifetime in vivo. ^47,48 As shown herein and by others, ^14,18,49 the primary activity of BDNF in the setting of rAION appears to be related to early priming of the grafted adult bone marrow–derived stem cells and neural progenitors for integration and differentiation into neural phenotypes. Similar findings were reported for hippocampus-derived neural progenitors grafted in neonatal eyes. ⁵⁰ Furthermore, BDNF, nerve growth factor, and bFGF have all been shown to prime bone marrow–derived mesenchymal stromal cells (MSCs) for differentiation and incorporation in murine models of retinal degeneration. ⁴⁵ This action has the added benefit of improving the survival of residual retinal cells after injury ^{13–15,49–52} because BDNF has been shown to be secreted by grafted MSCs ⁵³ and by umbilical cord blood cells (UCBs), ⁵⁴ thereby serving as a local neuroprotective factor. Other investigators have shown that the microenvironment in optic nerve crush is enriched by UCB-derived MSC secretion of TGF-β1, neutrotrophin 3, BDNF, and CNTF, ⁵⁵ as well as by bone marrow–derived secretion of bFGF and CNTF. ⁵⁶ However, it is questionable if this mechanism is operative in grafted Fr25lin⁻ cells because a previous study ²⁶ found no evidence of their production of TGF-β, FGF-2, or TNF-α, and only enhanced levels of insulin-like growth factor 1 correlated with an improved outcome. 

Ciliary neurotrophic factor has several pleiotropic effects. ⁵⁷ It induces gliosis, ^58,59 supports axonal regeneration, ⁶⁰ and enhances neural progenitor differentiation. ⁶¹ By contrast to BDNF, it did not augment quantitative engraftment of intravitreally injected Fr25lin⁻ cells; rather, it reduced their early incorporation. Nevertheless, CNTF induced the neural differentiation of the cells that migrated from the vitreous body to the injured retina. The CNTF-induced neural and glial differentiation ²⁶ may depend on the effective concentration of CNTF to which the cells are exposed ⁶¹ : high concentrations apparently enhance early neuronal differentiation of cells, whereas low concentrations induce the formation of glial phenotypes. Furthermore, the relative and absolute increase in glial lineages is consistent with proliferation of glia originating from the donor cells. ⁶² The increase in absolute numbers of engrafted donor cells, confirmed by cell-dividing antibody, and the significant contribution to the cellularity of the optic nerve might evolve from indirect support of neural and oligodendrocyte differentiation by astrocytes. ^16,48,63  

In summary, small primitive cells derived from the adult bone marrow and injected intravitreally in a model of rAION incorporate in the injured retina and adopt morphologic and phenotypic features of ganglion cells and astrocytes. The presence of injury appears to be a mandatory prerequisite for activation of the mitotically and functionally quiescent subset of Fr25lin⁻ cells. Brain-derived neurotrophic factor enhances both the quantitative and qualitative neuroglial differentiation of the cells, whereas CNTF augments only their qualitative conversion. The early inductive activity of these growth factors has long-term consequences on retinal remodeling, with retention of predominant gliosis-type repair of the neural tissues, including the retina and optic nerve. These data provide evidence that autologous bone marrow cells in conjunction with neural growth factors may have important therapeutic potential provided that the communication of the engrafted cells with the inner retinal neurons and their projection to the central visual system have a functional component. 

Acknowledgments

Supported by grants from the Leah and Edward M. Frankel Trust, The Zanvyl and Isabelle Krieger Fund, Grants 1371/08 and 1189/12 from the Israel Science Foundation (NG-C), Chief Scientist Eldor-Metzner Clinician Scientist Award 3-3741 (NG-C), Chief Scientist Ministry of Health and Lirot Foundation Award 3-4538 from the Israel Ministry of Health (NG-C), the North American Neuro-Ophthalmology Society, the Claire and Amédée Maratier Fund at Tel Aviv University (NG-C, B-CRA-L), the Walter Friendliest and Herman Shudder Fund at Tel Aviv University (NG-C), the Lions at Tel Aviv University (B-CRA-L), and the Mazritzky Fund at Tel Aviv University (TS). 

Disclosure: N. Goldenberg-Cohen, None; B.-C.R. Avraham-Lubin, None; T. Sadikov, None; N. Askenasy, None