Histologic Correlation of Human Neural Retinal Transplantation

Abstract

purpose. To describe the histologic findings of the transplanted eye of a 94-year-old man with neovascular age-related macular degeneration, who 3 years earlier underwent subretinal transplantation of both a fetal neural retinal sheet and a retinal microaggregrate suspension.

methods. Serial sections of the posterior segment of the eye and the transplanted areas were processed and studied by routine histologic techniques, including both light and transmission electron microscopy (TEM). Transplanted areas were also examined for the presence of glial, neuronal, and photoreceptor cell markers by standard immunohistochemical methods.

results. After transplantation in this patient, there was no visual improvement. Light microscopic examination disclosed survival of the transplanted cells in the subretinal space with no evidence of inflammation or rejection. The neural retinal sheet transplant developed a layered configuration. The retinal pigment epithelium (RPE) was absent over much of the posterior pole, including the area of transplantation. TEM examination and immunohistochemical analysis disclosed the presence of neuronal and glial cells within the transplant. A few transplant neuronal cell processes overlying a focus of residual RPE cells were positive for S-antigen, but well-developed photoreceptor outer segments were not present.

conclusions. Long-term survival of transplanted neural retinal tissue can be achieved in human patients without immunosuppression. The lack of photoreceptor development in this patient may be the result of absent or dysfunctional RPE. Nonetheless, the long-term survival of grafted tissue in the human subretinal space in the absence of immunosuppressive treatment is promising for future efforts in the field of neural retinal transplantation.

Avariety of ocular diseases, including age-related macular degeneration (AMD) and retinitis pigmentosa (RP), are characterized by the loss of photoreceptors, resulting in visual loss. In this group of diseases, the inner retinal layers are relatively well preserved despite profound loss of the outer retinal layers. ^{1 2 3} Replacement of retinal tissue by transplantation has been proposed as a potential method of restoring vision in these patients. 

A number of rodent (rat and mouse) models of retinal transplantation have been developed. ^{4 5 6 7 8 9 10 11 12 13 14} These models have shown that retinal tissue transplanted into subretinal space can survive and differentiate normally. 

The encouraging results from animal studies have prompted initial investigations of the feasibility of human transplantation. Kaplan and coworkers ¹⁵ transplanted a sheet of photoreceptor cells from cadaveric human eyes into two patients with RP (no light perception visual acuity), without immunosuppression. Although there was no improvement in visual function during the 12-month follow-up period, there was no clinical evidence of inflammation within the eyes, including at the graft sites. Humayun and coworkers ¹⁶ transplanted microaggregate suspensions of fetal tissue into the subretinal space of eight patients with RP and one patient with AMD. The patient with AMD also received a fetal retinal sheet transplant. This patient died from a cerebrovascular accident 3 years after the transplantation procedure. The transplanted eye was obtained postmortem for microscopic study, and the histologic features are described in this report. 

Methods

Patient History

A 94-year-old white man with bilateral neovascular AMD was recruited to participate in a study to investigate the safety and efficacy of human neural retinal transplantation. This patient developed progressive loss of central vision in each eye in the mid-1980s due to neovascular AMD. The patient developed severe subretinal hemorrhage in the left eye (OS) in June 1993, and vitreous hemorrhage, subretinal hemorrhage, and bullous hemorrhagic retinal detachment of the right eye (OD) in March 1994, resulting in light perception vision OS and no light perception OD. The patient underwent pars plana vitrectomy and evacuation of the subretinal hemorrhage OS in April 1994, but remained bare light perception postoperatively as a result of extensive subretinal fibrosis. Electrical stimulation of the retina OS in September 1994 elicited positive visual sensations, confirming the presence of functional neuronal cells in the inner retina. ¹⁷ In January 1995, the patient volunteered to participate in a pilot study of human fetal neural retinal transplantation. This study adhered to the Recommendations Guiding Medical Doctors in Biomedical Research Involving Human Subjects provided by the Declaration of Helsinki. The details of the experimental protocol and the operative procedure are described in the companion article in this issue. ¹⁶ The retinal tissue used for transplantation in this patient was obtained from a 16-week-old fetus. After a standard three-port pars plana vitrectomy, both the fetal microaggregate suspension (0.2 ml) and the fetal sheet graft (measuring 2 × 2 mm in size) were placed subretinally in an extramacular location. The microaggregate suspension was placed superior to the superotemporal vascular arcade, and the sheet was inserted along the superonasal arcade (Fig. 1A ). Postoperatively, there was no change in subjective visual status, visual acuity, electroretinography (ERG), and scanning laser ophthalmoscopic perimetry compared with baseline. Fluorescein angiography (Fig. 2) 4 months after transplantation disclosed a well-defined area of staining (without diffuse leakage) corresponding to the fetal sheet transplant. The fetal microaggregate area was not as well defined. 

Histology

The patient died from a cerebrovascular accident 35 months after transplantation, and the eyes were obtained 190 minutes postmortem by the Iowa Lions Eye Bank, fixed in 4% neutral-buffered formaldehyde, and sent to the Eye Pathology Laboratory of the Wilmer Ophthalmological Institute. The left eye was opened horizontally by removal of the inferior cap. Gross internal examination disclosed an extensive disciform scar with many areas of retinal pigment epithelium (RPE) atrophy and retinal thinning and some areas of hyperpigmentation. The areas of transplantation were not readily apparent but could be identified by comparison to previous clinical photographs. The posterior pole with the optic nerve head, macula, and transplant areas was removed, dehydrated, embedded in paraffin, and serially sectioned. Multiple sections were placed on each slide, and the slides were alternately stained with hematoxylin–eosin and periodic acid–Schiff, with every third slide unstained. Each section was examined using a microscope with a micrometer. 

Transmission Electron Microscopy

Two unstained 8-μm-thick paraffin-embedded sections from the area of fetal sheet transplantation were retrieved from the slide and reprocessed for transmission electron microscopy according to a previously reported technique. ¹⁸  

Immunohistochemistry

Immunohistochemical staining was performed according to the streptavidin peroxidase method previously described by Lutty and coworkers, ¹⁹ using the Vectastain avidin-biotin-complex–horseradish peroxidase (ABC–HRP) kit (Vector Laboratories, Burlingame, CA). Briefly, 8-μm-thick unstained sections were deparaffinized with Xylene for 10 minutes (two times), rehydrated with serial ethanol washes, treated with hydrogen peroxide for 30 minutes, and blocked with 10% goat serum and the ABC–HRP blocking kit. 

Primary antibodies were obtained from a variety of sources. Sheep anti-serum against bovine opsin was provided courtesy of David Papermaster (University of Texas Health Science Center, San Antonio, TX) and was used at a dilution of 1:2000. Rabbit anti-γ-aminoGABA antibody was generously provided by Ruben Adler (Wilmer Eye Institute, Johns Hopkins University, Baltimore, MD) and was used at a titer of 1:10,000. Rabbit anti-human synaptophysin (DAKO, Carpinteria, CA) was used at a 1:1000 dilution. Rabbit anti-S-antigen was a gift from Igal Gery (National Eye Institute, Bethesda, MD) and used at a titer of 1:1000. Rabbit antiserum against S-100 protein (ICN Biochemicals, Aurora, IL) was used at 1:1000. Rabbit anti–glial fibrillary acidic protein (GFAP; DAKO) and rabbit anti–neuron-specific enolase (NSE; DAKO) were both used at dilutions of 1:50. All dilutions were performed with phosphate-buffered saline (PBS). 

Histologic sections were incubated with the various primary antibodies (at the concentrations described above) for 20 hours at 4°C. Secondary antibodies (biotinylated rabbit anti-sheep IgG, goat anti-rabbit IgG, or goat anti-mouse IgG; Kirkegaard and Perry Labs, Gaithersburg, MD) were prepared by preadsorption with human serum for 30 minutes at 37°C (1 part antibody, 9 parts serum), followed by a further 1:50 dilution. After washing, sections were incubated for 30 minutes at room temperature with secondary antibodies diluted 1:500. Sections were then incubated with streptavidin labeled with peroxide in PBS (1:500 dilution; Kierkegaard and Perry Labs). Finally, the ABC-peroxidase complex was detected using 3-amino-9-ethylcarbazol as the substrate. Some sections were counterstained with Harris’ hematoxylin. 

For each immunohistochemical marker, positive and negative controls were also performed. Positive controls included paraffin sections through the posterior poles of the globes of an 81-year-old man with neovascular AMD (but with preservation of most photoreceptor outer segments) and a 73-year-old man with a normal retina by microscopic examination. Negative controls were performed (on the same patients as the positive controls) for each immunohistochemical marker by substituting PBS for the primary antibody or by substituting nonimmune rabbit IgG (0.4 μg/ml). The transplant sections were not used as negative controls because of the limited supply of available sections. 

TdT-dUTP Terminal Nick-End Labeling

In situ detection of apoptotic cells was performed using the TACS Blue Label Detection kit (Trevigen, Gaithersburg, MD) according to the manufacturer’s protocol. Briefly, sections were deparaffinized, dehydrated, and rehydrated. The sections were then treated with proteinase K to increase tissue permeability, followed by 2% H₂O₂ to quench the endogenous peroxides. Sections were then in situ-labeled with a dNTP mix and terminal deoxynucleotidyl transferase in the presence of MnCl₂ at 37°C. 

The reaction was stopped with stop buffer. Streptavidin–HRP conjugate was then added to the tissue. The positive signal was visualized by blue label and counterstained with red Counterstain C. 

Results

Light Microscopy

Posterior Segment, Left Eye.

Examination of 1517 serial sections of the posterior pole of the left eye disclosed an extensive two-component disciform scar measuring up to 15.2 mm in diameter and up to 1.1 mm in thickness. The scar was composed of a thicker subretinal dense fibrocellular component and a thinner, but variably thick, sub–RPE fibrovascular component. A few areas of calcification and bone formation were present in the sub–RPE component. The two components were separated by a slit-like structure lined by hyperplastic RPE and a variably thick layer of basal laminar deposit. There were numerous defects of various sizes in this slit-like structure (RPE tears) where the two components of the scar were continuous. In addition, there were nine discontinuities of Bruch’s membrane, and in seven of these, blood vessels extended from the choroid into the intra–Bruch’s membrane component of the scar. There was total loss of the photoreceptor layer and partial loss of the outer plexiform and inner nuclear layers over the disciform scar. Apparent contraction of the disciform scar was associated with redundancy of Bruch’s membrane. A thin, hypocellular epiretinal membrane with moderate wrinkling of the internal limiting lamina was present over the macula. A moderate lymphocytic infiltrate was present in the choroid, and in some areas of the sub-RPE component of the scar. 

Fetal Sheet Transplant.

A 2.0-mm segment of donor retina was present external to the host retina with a thin membrane located at the graft-host interface (Fig. 1B) . This membrane was not continuous, and there was apparent contact between donor and recipient cells at these discontinuities (Fig. 1C) . No donor ganglion cells were present, but the transplanted cells were organized in a layered configuration in some areas. In other areas, however, the cells were scattered and more loosely arranged. Also, in some locations the transplanted tissue adopted a “rolled-up” configuration. The inner nuclear layer was relatively intact and was apparently continuous with the recipient’s inner nuclear layer at the nasal and temporal margins of the transplant (Fig. 3) . In one area there was a focus of neurons with cytoplasmic extensions (possible photoreceptors) overlying a small patch of residual RPE cells; this area was studied by immunohistochemical techniques (S-antigen stain, Fig. 4 ). However, the RPE was largely absent over much of the macular and perimacular regions, including the areas of transplantation (Fig. 1B) . There was no inflammation in the recipient or fetal retina or in the subjacent choroid. 

Fetal Microaggregate Transplant.

A 1.3-mm by up to 75-μm focus of loosely compacted donor cells was present between the recipient retina and Bruch’s membrane. The RPE was absent in this region. An apparent membrane was present at the inner aspect of the transplanted tissue (Fig. 1D) . No definite rosette formation was present, although some cells were arranged in rosette-like configurations. No definite photoreceptors were present. There was no inflammation in the recipient or fetal retina or in the subjacent choroid. 

Transmission Electron Microscopy

Examination of two sections through the fetal sheet transplant disclosed a zone of transplanted cells between the recipient retina and choroid. In some areas the transplant and the recipient retinas were separated by a fibrocellular membrane composed of Müller cell processes, cellular debris, and collagen (10-nm fiber diameter). Preservation of ultrastructural detail was generally poor, thus impairing the characterization of constituent cells. Several round cells (possible donor neural cells) and a few larger oval-to-round cells (possible Müller cells) were present within the donor retina. Other cell types could not be clearly identified. No definite photoreceptor inner and outer segments were identified (illustrations not shown). 

Immunohistochemical Analysis

Unstained sections through both areas of transplantation (fetal sheet and microaggregate) were studied by a variety of immunostaining techniques (Table 1) . Some transplanted cells did stain positively for GFAP, S-100, GABA, NSE, and synaptophysin, indicating the presence of both glial and neuronal cells within the transplant (Fig. 4A 4B 4C 4D) . The donor tissue stained weaker than the host cells with S-100 but clearly stronger than the negative control (not shown). A few possible GABA+ and synaptophysin+ cell processes were noted to extend between the transplant and the host retina (Figs. 4A 4B) . The cells identified as possible primitive photoreceptors by light microscopy were weakly positive for S-antigen (Fig. 4C) . The opsin stain was negative in the transplant, but this stain was performed on a different transplant area that did not contain intact RPE. TdT-dUTP terminal nick-end labeling (TUNEL labeling) was negative in sections through both transplanted areas (illustration not shown). 

Discussion

Long-term survival of fetal neural retinal allografts transplanted into the subretinal space without immunosuppression has been demonstrated by a variety of investigators in a number of animal models. ^{9 10 12 20 21} Human xenografts, on the other hand, require the use of systemic cyclosporine to remain viable when transplanted into rats. ²² The excellent survival of allografts is believed to be a reflection of the limited immunologic privilege of the subretinal space. This privilege is believed to be primarily the result of the outer blood–retina barrier formed by tight junctions between the RPE. In patients with AMD, this barrier may be compromised, thereby raising the specter of rejection and the question of need for immunosuppression. 

In this report, we describe a patient with AMD who underwent transplantation of both a microaggregate suspension and an intact sheet of fetal neural retina into the subretinal space. The patient died 3 years after the procedure. Clinically, the patient’s eye showed no evidence of rejection. Fluorescein angiography did not show evidence of diffuse leakage (only staining), suggesting that the transplant was not causing a significant local breakdown of the blood–retina barrier (a finding that would be expected if there was an inflammatory reaction by the host). The transplanted tissue was still present in the subretinal space after 3 years, with no evidence of inflammatory cells in the vicinity of the graft to suggest active cell-mediated rejection. Furthermore, no apoptosis was evident by TUNEL labeling. However, the percentage of cells that survived cannot be determined. Certainly, it is possible that some cells were lost over the course of 3 years as a result of inflammation (including rejection), apoptosis, or other cell-death mechanisms, which were not active at the time of the patient’s death. Nonetheless, the survival of a significant amount of transplanted tissue without the use of systemic immunosuppression is a promising finding. The lack of inflammation and rejection of the transplanted tissue may be a reflection of the relatively low immunogenicity of fetal tissue. In addition, neuronal cells, particularly photoreceptors, do not express detectable levels of major histocompatibility complex (MHC) class I molecules. ^{23 24 25} This feature further reduces the likelihood of inciting an immune reaction. Of note, however, Larsson and coworkers ²³ found an upregulation of MHC class I antigen expression in allogeneic retinal transplants in rats but no change in expression in syngeneic transplants. However, this difference in expression did not correlate with rejection or decreased survival of the transplant. Thus, these investigators concluded that allogeneic transplants into the subretinal space are recognized as “non-self” but some modification of the immune response occurs that prevents rejection. 

The transplanted tissue in this patient demonstrated some features suggestive of primitive retinal development. In some areas of the fetal sheet transplant, the cells were organized in a layered configuration. The component cell types of the transplanted tissue were characterized by a variety of immunohistochemical stains. Antibody stains for GABA, ^{26 27} an important inhibitory neurotransmitter in retinal neuronal cells, and synaptophysin, ²⁸ an N-glycosylated integral membrane protein of synaptic vesicles, were positive, thus confirming the presence of neuronal cells within the transplant. In addition, the presence of synaptophsyin suggests that the transplanted cells have the potential to form synaptic connections. Moreover GABA-positive and synaptophysin-positive neuronal cell processes were present between the host and the transplant tissues. This suggests that there was some degree of attempted integration (at least structural) of the transplant into the recipient retina. However, it was not possible to determine whether synapses formed between host and transplant neurons. 

Glial cells were also present within the transplant. S-100 ²⁹ is a calcium-binding protein, which is present in the cell bodies and processes of retinal astrocytes and Müller cells. GFAP ²⁹ is a structural protein expressed in the processes of retinal astrocytes and also in Müller cell processes after retinal injury or damage. Retinal astrocytes are believed to migrate to the retina from the optic nerve head, whereas Müller cells arise from intrinsic retinal cells. ²⁹ Transplanted tissue from both the microaggregate and intact sheet regions stained positively for both S-100 and GFAP. Although within the transplant retinal astrocytes and Müller cells could not be distinguished with certainty by light microscopy or immunohistochemical stains, possible Müller cells were identified by electron microscopy. These Müller cells could have differentiated from the transplanted fetal cells or migrated from the host retina. 

Definite photoreceptor outer segments could not be identified by light or electron microscopy. There was, however, a single small focus of cells that stained positively for S-antigen. S-antigen ^{30 31 32 33} is a photoreceptor cell–specific protein that is present predominantly in rod outer segments. Unfortunately, additional sections were not available through this small area to look for ultrastructural features of photoreceptor differentiation. Nonetheless, the S-antigen staining suggests that some transplanted cells with possible primitive photoreceptor differentiation remained. Of note, these remaining possible photoreceptors were located internal to a small focus of residual RPE cells. It has been well established that functional RPE cells are crucial for the maintenance and survival of photoreceptors. The RPE was absent over much of the posterior poles in both eyes of this patient (fellow eye examined microscopically but not described in this report), including the areas of transplantation in the left eye. Residual hypertrophic RPE cells were present in some areas of the posterior pole, but they were often arranged in tubuloacinar configurations and other structures incompatible with normal function. The lack of functional RPE is a potential cause for the paucity of photoreceptor cells within the transplant. 

The RPE may have been lost for a variety of reasons. Some RPE loss may be a reflection of the patient’s advanced AMD. Certainly, RPE atrophy is an important feature of AMD. This hypothesis is supported by the observation that the RPE loss was present in both eyes. In addition, RPE cells may been lost during the evacuation of submacular hemorrhage in the left eye 9 months before the transplantation procedure. Finally, some RPE loss may have occurred as a direct result of transplantation procedure. Although a very slow infusion of fluid accompanied the insertion of fetal tissue, direct trauma from the fluid wave or the insertion forceps (in the case of the sheet transplant) may have disrupted the RPE. All three mechanisms likely contributed to RPE loss, but the relative contributions of each mechanism cannot be precisely determined. 

In some areas, in both the microaggregate and the sheet transplants, a membrane was evident between the internal surface of the transplanted tissue and the recipient retina. The membrane was of variable thickness and composed of Müller cell processes, cellular debris, and collagen. The relative contributions of the recipient and the donor to the formation of this membrane are unclear. This structure may be similar to the “Müller cell barrier” that has been previously described in some animal models of retinal transplantation. ¹² This barrier, although not a continuous structure, has been postulated by some investigators to be an impediment to the formation of connections between host and transplant neurons. Methods aimed at reducing this “barrier” and improving graft–host integration are areas of active investigation in animal retinal transplantation. However, despite the membrane at the graft–host interface, neuronal processes were present between the transplant and the recipient retina in the patient described in this report. This observation is promising for the prospects of developing functional connections between the host retina and transplanted retinal tissue. 

In summary, this report illustrates that long-term survival of fetal neural retinal tissue transplanted into the human subretinal space can be achieved without immunosuppression even in the absence of an intact blood–retina barrier. Although the transplant did not develop normal retinal architecture, the cells in the sheet transplant did develop and organize into layers. In addition, the transplanted tissue gave rise to both glial and neuronal cells. Well-differentiated photoreceptor outer segments did not develop in the transplanted tissue in this patient, but this may be because of the absence of subjacent RPE cells. The lack of RPE cells in this patient may be a result of previous subretinal surgery, AMD, the transplantation procedure itself, or a combination of all three processes. A fibrocellular membrane was evident between the transplant and the recipient retina and may be similar to the Müller cell “barrier” observed in animal models of retinal transplantation. Despite this membrane, apparent neuronal processes were present between the graft and the recipient retina. Future efforts should be directed toward enhancing graft–host integration and improving the techniques of transplantation (to allow more retinal cells to be transplanted while minimizing damage to existing host tissues). Nonetheless, the long-term survival of grafted tissue in the human subretinal space in the absence of immunosuppressive treatment is promising for future efforts in the field of neural retinal transplantation. 

 Supported by the Grousbeck Family Foundation, Menlo Park, California; the International Order of Odd Fellows, Winston–Salem, North Carolina; and Macula Foundation, New York, New York.

 Submitted for publication March 25, 1999; revised November 10, 1999 and March 20, 2000; accepted March 31, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

(A) Appearance of posterior pole of left eye 4 months after transplantation. An extensive disciform scar (asterisk) occupies much of the macula. The retinal sheet transplant is well demarcated (boxed area 1). The area of the microaggregate transplant (boxed area 2) is not well-defined, but the retinotomy site (arrowhead) is evident. (B, C) Appearance of fetal sheet transplant. (B) Donor cells are arranged in a layered configuration (below asterisk). An epiretinal membrane is present along the host internal limiting lamina (arrow). An apparent membrane (arrowhead) is also present between the donor and recipient (asterisk) retinal layers. (C) There are large gaps (large arrows) in this membrane (arrowheads), allowing direct contact between donor and recipient cells. The RPE is largely absent at the transplant site. (D) Appearance of fetal microaggregate transplant. Donor cells (asterisk) are loosely arranged external to a membrane at the graft–host interface (arrowhead). Periodic acid–Schiff; magnification, (B, C) ×140; (D) ×350. Scale bar, 40 μm.

Figure 2.

 

Fluorescein angiogram of left eye 4 months after transplantation. (A) Early phase (53 seconds); (B) late phase (448 seconds). There is staining of the retinal sheet transplant (arrowheads) and the wall of an overlying retinal vessel (arrow).

Figure 3.

 

Fetal sheet transplant. Large arrows define the nasal and temporal borders of the transplant. Small arrows indicate the interface between the graft and host retina. The membrane at the interface is not well seen at this magnification and with this stain (hematoxylin and eosin; magnification, ×25).

Figure 4.

 

Immunohistochemical markers, fetal sheet transplant. Graft–host interface membrane indicated by arrowheads. Negative controls (primary antibody omitted) shown in insets for (A), (B), (C). (A) GABA-positive processes (arrows) extend between donor and recipient layers through gaps in the membrane at the graft–host interface (arrowhead). (B) Synaptophysin-positive processes (arrows) extend between donor and recipient layers through gaps in the membrane at the graft–host interface (arrowhead). (C) S-antigen. A small focus of positive staining (arrow) is present overlying a few residual RPE cells. (D) GFAP. Positive brown staining is present in the graft. Radial glial fibers are observed (arrow). H, host; T, transplant. Scale bar, 20μ m.

Table 1.

 

Immunohistochemistry Studies

	Positive Control	Transplant	Recipient
GFAP	++	++	++
S-100	++	+	++
Synaptophsyin	++	++	++
GABA	++	++	++
NSE	++	++	++
Opsin	++	−	−
S-antigen	++	+	−

 −, negative; +, weakly positive; ++, moderately positive; +++, markedly positive.

The authors thank Sam Williams and the Rochester Eye Bank.