Cyclic AMP Prevents Retraction of Axon Terminals in Photoreceptors Prepared for Transplantation: An In Vitro Study

Abstract

purpose. Cell transplantation has emerged as a possible remedy for degeneration and injury in the central nervous system (CNS). In the retina, photoreceptor transplantation is a potential treatment for retinal degenerative disease. Graft survival has been well documented, but evidence of functional recovery is lacking. A major obstacle to recovery of vision is lack of synapse formation between grafted photoreceptors and host bipolar and horizontal cells. A prior study demonstrated that photoreceptors prepared for transplantation undergo rapid morphologic changes, including retraction of axon terminals toward their cell bodies, away from potential synaptic partners, a phenomenon that may interfere with graft–host synaptic interaction after transplantation. In this study, prevention of retraction of photoreceptor axon terminals was possible by pharmacological intervention.

methods. Photoreceptor sheets, prepared by vibratome sectioning, and full-thickness retinas, harvested from adult porcine eyes, were maintained in culture and treated with either the cyclic adenosine monophosphate analogue 8-(4-chlorophenylthio)-cyclic 3′,5′-adenosine monophosphate (CPT-cAMP), or forskolin, an adenylyl cyclase stimulant, for up to 48 hours.

results. Both CPT-cAMP and forskolin treatments successfully blocked retraction of photoreceptor axon terminals. This effect was not due to cell toxicity and was reversed after removal of treatment, indicating its specificity.

conclusions. Pharmacological manipulation of photoreceptor axonal plasticity may improve graft–host synaptic interaction after subretinal photoreceptor cell transplantation.

Whether adult, fetal, or stem cell populations are used in transplantation into the CNS, restoration of function depends in part on the formation of synapses between the graft and host nervous tissue. Even though transplants may mediate a trophic effect that can preserve residual host function, re-establishing normal activity demands reconstruction of synaptic networks. Photoreceptor cell transplantation is a potential technique for restoration of vision in patients with retinal degenerative diseases, such as retinitis pigmentosa and age-related macular degeneration, in which there is substantial loss of rod and cone cells. However, the scarcity of neural process outgrowth from transplanted photoreceptors as well as the absence of definite synaptic contacts with the appropriate host second-order neurons is a recurring result in all transplantation paradigms, regardless of the type of preparation or age of donor tissue. ^{1 2 3 4 5}  

The apparent lack of synaptic regeneration by transplanted photoreceptors contrasts with the growth of processes by both rod and cone photoreceptors and by other cells of the retina after isolation in culture, ^{6 7 8} in experimental retinal detachment and reattachment, ^{9 10} and during retinal degeneration in both animal models ^{11 12 13 14 15} and human disease. ¹⁶ In these cases, second-order neurons extend elaborate dendrites, and rod and cone cells grow neurites and form numerous presynaptic varicosities. To investigate whether photoreceptors prepared as grafts for retinal transplantation retain the ability to extend neural processes, we developed an in vitro system for the maintenance of adult porcine photoreceptor sheets and intact retinas in culture. We found that photoreceptors do not extend but instead retract their axons and terminals toward the cell body within hours of being detached from the retinal pigment epithelium (RPE), ¹⁷ similar to what occurs shortly after retinal detachment or photoreceptor cell isolation. ^{8 9 10} Retraction may represent an impediment to establishing graft–host synaptic contacts. 

Cyclic 3′,5′-adenosine monophosphate (cAMP) has been shown to abolish pharmacologically induced neurite retraction in neuronal cell lines ¹⁸ and growth factor–induced growth cone collapse as well as neurite retraction in primary neuronal cultures. ¹⁹ Furthermore, cAMP, through its downstream effector protein kinase A (PKA), blocks myelin and the myelin-associated glycoprotein inhibition of neurite outgrowth in cultured rat retinal ganglion cells, dorsal root ganglion neurons, cerebellar neurons, and raphespinal projection neurons. ²⁰ In the present study, we increased intracellular levels of the second-messenger cAMP to pharmacologically manipulate photoreceptor axonal plasticity. Retraction was prevented whether cAMP was applied exogenously, using the membrane-permeable cAMP analogue 8-(4-chlorophenylthio)-cyclic 3′,5′-adenosine monophosphate (CPT-cAMP) or increased by activation of endogenous adenylyl cyclase with forskolin. These effects were reversible and did not affect cell viability. These data have been presented in abstract form (Khodair MA, et al. IOVS 2001;42:ARVO Abstract 4192). 

Materials and Methods

Experimental Animals

Twenty-eight eyes from 14 adult male and female Yorkshire pigs, 3 to 5 months old and weighing 25 to 55 kg (Animal Biotech Industries, Danbora, PA), served as donors of retinal tissue. Animals were kept on a 12-hour light–dark cycle and fed porcine chow, ad libitum. They were killed at 9:00 AM by anesthetizing with telazol (7 mg/kg; 100 mg/mL), xylazine (2.2 mg/kg; 100 mg/mL), and atropine (0.02 mg/kg; 0.54 mg/mL) administered intramuscularly, followed by an overdose of pentobarbital sodium, 1 mL/4.5 kg, administered intravenously. Experimental procedures adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the University of Medicine and Dentistry of New Jersey-New Jersey Medical School Institutional Animal Care and Use Committee. 

Preparation and Culture of Photoreceptor Sheets and Full-Thickness Retinas

Photoreceptor sheets were prepared by vibratome sectioning with a method modified from Huang et al. ²¹ and previously described. ¹⁷ Briefly, retinas still attached to the underlying RPE, choroid and sclera, were cut from the central and midperipheral regions of the eyecup with a trephine. Tissue was kept at 4°C at all times in Eagle’s minimum essential medium (MEM, 10370-021; Invitrogen-Gibco-Life Technologies Inc., Rockville, MD), pH 7.4, supplemented with 0.292 mg/mL glutamine, 10% (vol/vol) fetal calf serum, 10 μg/mL porcine insulin, 5.5 mM d-glucose, 1 mM pyruvate, 0.1 mM taurine, 2.0 mM ascorbic acid, 100 U/mL penicillin, 100 μg/mL streptomycin, and 250 ng/mL amphotericin B, and aerated with humidified 5% CO₂ and 95% O₂. To obtain photoreceptor sheets, we gently detached the neural retina from the underlying tissues and mounted it on a sterile gelatin block with the photoreceptor outer segments (OSs) oriented downward. Inner retinal layers were removed sequentially by vibratome sectioning. Sectioning was terminated at the level at which one to two cells of the inner nuclear layer (INL) remained, to ensure the intactness of the photoreceptor terminals. The sectioned retina and an underlying gelatin layer were then placed on a glass slide, covered by another 100-μm-thick gelatin layer, incubated for 30 to 60 seconds at 37°C to allow the gelatin layers to melt, and finally placed on ice to resolidify. A photoreceptor sheet embedded in a gelatin sandwich, with a total thickness of approximately 250 to 300 μm, was thus produced. 

For full-thickness preparations, retinas were embedded in gelatin sandwiches according to the procedure just described, but without sectioning the retina. 

Photoreceptor sheets and full-thickness retinas were bathed in fortified MEM (pH 7.4), with 0.1, 0.5, or 1 mM CPT-cAMP (C3912; Sigma-Aldrich, St. Louis, MO) or forskolin (F-6886; Sigma-Aldrich) dissolved in 100% dimethyl sulfoxide (DMSO) and then diluted with medium to obtain final concentrations of 0.1, 1, or 10 μM in 0.1% DMSO. Specimens were incubated for <45 minutes, 24 hours, or 48 hours at 37°C in a humidified mixture of 5% CO₂ and 95% O₂. For comparison of photoreceptor sheets and full-thickness retinas, one eye of a donor pig was used to obtain sheets, and the contralateral eye was used to obtain retinal preparations. Control samples were photoreceptor and retinal preparations obtained from the same retinal areas as in the treatment groups, but incubated in fortified MEM only or with 0.1% DMSO. Full-thickness retinas fixed immediately after detachment from the RPE served as overall control specimens for both treated and untreated groups. 

Immunohistochemistry

At the end of incubation, specimens were fixed in 4% paraformaldehyde in 0.125 M phosphate buffer (PB; pH 7.4) overnight at room temperature, rinsed in phosphate-buffered saline (PBS), re-embedded in gelatin, and returned to the fixative for at least an additional 24 hours at 4°C. Fixed, gelatin-embedded specimens were rinsed in PBS, mounted on a tissue cutter (Sorvall, Newtown, CT), chopped into 100-μm-thick sections, and immunolabeled as previously described. ^{8 17} Rabbit polyclonal anti-synaptophysin (1:100; Dako Corp., Carpinteria, CA), which localizes synaptic vesicles, was used to evaluate morphologic changes in photoreceptor terminals. The mouse monoclonal antibody specific for rod opsin (4D2; 1:25; kindly provided by Robert Molday, University of British Columbia, Vancouver, BC, Canada) was used to identify rod photoreceptor membranes. Goat anti-rabbit-FITC and goat anti-mouse-tetramethylrhodamine isothiocyanate (TRITC; 1:35) were used as secondary antibodies. Control sections were processed without primary antibodies. 

Morphometric Analysis

One-micrometer-thick optical sections of photoreceptor sheets and full-thickness retinas were obtained with a laser scanning confocal microscope equipped with an argon/krypton laser and a 63×, 1.4 numerical aperture (N/A) oil-immersion objective lens (LSM-410; Carl Zeiss Meditec, Thornwood, NY). Brightness and contrast were set to obtain unsaturated images. Parameters were maintained throughout a single experiment. Using normalized settings, we were able to detect changes in labeling patterns between specimens. Adjustments in brightness and contrast were performed after image analysis, exclusively for presentation purposes. 

Photoreceptor axonal and terminal retraction, indicated by the area of synaptophysin labeling in the ONL, was analyzed using NIH Image 1.62 software (http://rsb.info.nih.gov/nih-image; developed by Wayne Rasband and provided in the public domain by the National Institutes of Health, Bethesda, MD). Images were opened in the fluorescein channel and a standard threshold was assigned to all images of the same experiment to eliminate background fluorescence. The total area of fluorescein labeling was measured within a rectangular area, outlining the ONL, as previously described. ¹⁷ Measurements were in square pixels, scaled at 6 pixels/μm. Data were collected from nine sections per specimen. Statistical analysis was performed with the unpaired t-test and one-way analysis of variance (ANOVA), with Tukey’s post hoc test for all pair-wise multiple comparisons, when appropriate. 

Evaluation of Photoreceptor Viability in Culture

Photoreceptor viability was assessed by the lactate dehydrogenase (LDH) release assay. Photoreceptor sheets and full-thickness retinal preparations were incubated at 37°C in supplemented Eagle’s MEM (as described earlier) without phenol red (51200-038; Invitrogen-Gibco-Life Technologies, Inc.) only, or with either 1 mM CPT-cAMP or 10 μM forskolin for <45 minutes, 24 hours, and 48 hours. Specimens treated with 100 μM ouabain were incubated under similar conditions and served as the positive control. Tissue-free aliquots were centrifuged at 250g for 4 minutes to remove cellular debris. Supernatant was processed with an LDH in vitro toxicology assay kit (TOX-7; Sigma-Aldrich). Colorimetric absorbances were obtained with a spectrophotometer (Du series 60; Beckman Instruments Inc., Fullerton, CA) at a 490-nm wavelength. Medium-only represented the blank solution, and medium reacted according to the LDH assay kit instructions served as the background. 

Results

Effect of CPT-cAMP on Photoreceptor Axon Terminal Retraction

Two types of grafts were examined: sheets of photoreceptor cells prepared by vibratome sectioning and full-thickness retinas. CPT-cAMP was chosen for treatment of the preparations because it is a C-8 substituted cAMP analogue that is more efficacious in promoting long-term survival and neurite outgrowth than N⁶-substituted analogues in neuronal cell culture. ²² In addition, CPT-cAMP is highly membrane permeable because of its lipophilic C-8-substituent and unlike other analogues, such as 8-Br-cAMP, it can be hydrolyzed by cAMP-specific phosphodiesterase (PDE). A potential disadvantage of CPT-cAMP is that it is a competitive inhibitor of the cGMP-specific phosphodiesterase (PDE5a), ²³ which may elevate intracellular cGMP levels. However, this may have had a minimal effect on our experiments, because the predominant PDE in photoreceptors is PDE6. 

For normal morphology, retinas fixed immediately after removal from the eyecup (Fig. 1A)were immunolabeled with anti-opsin and anti-synaptophysin antibodies, to delineate photoreceptor cell membranes and synaptic vesicles, respectively. Opsin labeling was confined primarily to photoreceptor OSs, which were relatively rectilinear. The outer nuclear layer (ONL) was composed of five to seven layers of uniformly spaced photoreceptor nuclei and showed no signs of pyknosis. Synaptophysin labeling was detected in both plexiform layers. In the outer plexiform layer (OPL), however, synaptophysin labeling was confined to the sclerad half of the layer where photoreceptor terminals are located. Photoreceptor axons and the ONL were mostly devoid of synaptophysin labeling. 

Photoreceptor sheets and full-thickness retinas maintained at 37°C for up to 48 hours in culture, with or without CPT-cAMP, were fixed after various incubation times and immunolabeled as described earlier. In agreement with our previous findings, ¹⁷ significant morphologic changes were seen in photoreceptor sheets and full-thickness retinal preparations. Opsin labeling, normally largely restricted to membranes of photoreceptor OSs, was observed to varying degrees in the inner segments (ISs), cell bodies, and terminals of rod cells (Figs. 1B 1E) . The extent of opsin spread seemed to increase with longer incubation times. Synaptic vesicle (synaptophysin) labeling was present in the OPL, as expected, but also in the ONL (Figs. 1B 1E) . Labeling in the ONL represents retraction of axons and terminals toward the cell bodies and has been described in experimental retinal detachment ^{9 24} and in porcine photoreceptor sheets. ¹⁷ The retraction of synaptic terminals created areas devoid of synaptic vesicle label in the OPL, most noticeably in photoreceptor sheets (Fig. 1B) . In contrast, specimens maintained under the same culture conditions but treated with CPT-cAMP showed almost no labeling in the ONL and almost continuous labeling in the OPL (Figs. 1C 1F) . Opsin labeling, however, remained dispersed (Fig. 1D) . 

The change in synaptophysin localization in untreated specimens began by 45 minutes and was most pronounced by 48 hours in culture (Figs. 1G 1H) . Treatment with 0.1, 0.5, or 1 mM CPT-cAMP significantly reduced ONL synaptophysin labeling in photoreceptor sheets and full-thickness retinas to only 7% to 11% of that observed in untreated specimens after 24 to 48 hours in culture (Figs. 1G 1H) . Furthermore, in photoreceptor sheets, CPT-cAMP treatment resulted in no significant difference in synaptophysin labeling of the ONL, compared with control specimens for up to 24 hours, indicating a complete block of retraction. It was not until 48 hours that these specimens showed a small (1- to 1.5-fold) but significant increase in synaptophysin ONL labeling compared with the control samples (Fig. 1G) . In the full-thickness retinal preparations, however, there was a slight (1- to 1.5-fold) but significant increase in synaptophysin labeling as early as 45 minutes in culture, compared with retinal control specimens, indicating a less complete blockage of retraction (Fig. 1H) . The effect of CPT-cAMP on blockage of retraction did not show a dose-dependent variation in either photoreceptor sheets or full-thickness retinas, suggesting that saturation of the response was reached with the lowest dose. 

Reversibility of the Effects of CPT-cAMP

To determine whether the blockage of photoreceptor axon terminal retraction by CPT-cAMP is caused by toxicity or irreversible stabilization of the cytoskeleton, photoreceptors were examined after drug removal. Photoreceptor sheets were maintained in medium alone or with 0.1 mM CPT-cAMP, the lowest effective concentration, for 48 hours at 37°C. In a group of specimens, CPT-cAMP was removed and replaced by CPT-cAMP-free medium after 45 minutes or 24 hours of incubation with the drug. After an additional 24 hours of incubation in cAMP-free culture medium, specimens were fixed. For untreated and CPT-cAMP-treated photoreceptor sheets, medium was simply renewed. Results were compared with retinas fixed immediately after removal from the eyecup. Untreated photoreceptor sheets showed an increase in ONL synaptophysin labeling, whereas in CPT-cAMP-treated photoreceptor sheets, labeling of the ONL remained at very low levels throughout the entire course of the experiment (Fig. 2) . Thus, consistent with previous results, CPT-cAMP treatment dramatically reduced axon terminal retraction in photoreceptor sheets. In preparations from which CPT-cAMP had been removed, a rapid and significant increase in ONL synaptophysin labeling was seen within 24 hours of CPT-cAMP removal, reaching 46% to 58% of that in untreated specimens (Fig. 2) . This increase in labeling occurred regardless of whether the duration of CPT-cAMP treatment was 45 minutes or 24 hours. Thus, retraction of photoreceptor axon terminals toward their cell bodies resumed after removal of CPT-cAMP treatment, indicating that the block of the retraction is reversible. 

Effect of Adenylyl Cyclase Activity on Photoreceptor Axon Terminal Retraction

To confirm that the inhibitory effect on retraction of photoreceptor axon terminals is specific to cAMP treatment, we increased the photoreceptor intracellular cAMP levels by the stimulation of endogenous adenylyl cyclase, using forskolin. Photoreceptor sheets were incubated at 37°C in medium treated with 0.1, 1, or 10 μM forskolin dissolved in 0.1% dimethyl sulfoxide (DMSO) or in medium with 0.1% DMSO alone (the control), for up to 48 hours. Results were compared with those in retinas fixed immediately after removal from the eyecup. A significant increase of synaptophysin labeling in the ONL was detected in untreated specimens within 45 minutes of incubation (Fig. 3) , but specimens treated with 1 μM forskolin showed no significant increase in labeling. In specimens treated with either 0.1 or 10 μM forskolin, however, there was a small (less than one-fold) but significant increase in labeling compared with control specimens (Fig. 3) . By 24 hours in culture, significantly larger amounts of synaptophysin labeling were seen in all forskolin-treated compared with control specimens (Fig. 3) ; however, the amount of ONL labeling was only between 11% and 14% of that in untreated specimens. This pattern persisted for 48 hours in culture, where untreated specimens showed dramatically more synaptophysin labeling in the ONL than did treated specimens (Fig. 3) . Thus, forskolin treatment successfully reduced synaptic terminal retraction in photoreceptor sheets for up to 48 hours. In specimens receiving the largest dose of forskolin (10 μM) the ONL synaptophysin labeling was significantly greater (approximately two-fold) than in specimens treated with lower doses of forskolin (Fig. 3) . The less effective inhibition of axon terminal retraction by 10 μM forskolin applied for 48 hours may indicate a desensitization of adenylyl cyclase at high forskolin concentrations. ²⁵  

Cyclic AMP and Photoreceptor Viability

Increases in cAMP can be associated with cell death ^{26 27} ; 10 μM forskolin, specifically, has been shown to increase cell death of isolated rod cells in culture. ²⁸ Cell death was evaluated in photoreceptor sheets as a function of LDH released into the medium. Photoreceptor sheets were incubated at 37°C for 48 hours, with medium alone or in medium containing 1 mM CPT-cAMP or 10 μM forskolin, the highest concentrations used in our experiments. As a positive control, photoreceptor sheets were maintained in medium containing 100 μM ouabain. In all specimens, spectrophotometric absorbances showed a statistically nonsignificant increase during the first 24 hours of incubation (Fig. 4) . Levels plateaued at 48 hours, with one exception. Medium obtained from cultures of photoreceptor sheets treated with ouabain showed a sharp and significant increase in colorimetric absorbance after 48 hours of incubation (Fig. 4) . Therefore, ouabain caused increased cell death, whereas treatment of photoreceptor sheets with either CPT-cAMP or forskolin had no significant effect on LDH levels for up to 48 hours. 

Discussion

Effect of Elevation of Intracellular cAMP on Axon Terminal Retraction

Adult porcine retina prepared for transplantation, whether as full-thickness retina or a photoreceptor sheet, shows a redistribution of synaptophysin labeling from the OPL deep into the ONL. ¹⁷ This change in labeling most likely represents retraction of photoreceptor axon terminals toward their cell bodies, a process that also occurs in experimental retinal detachment. ^{9 24}  

We have suggested that detachment of photoreceptors or neural retina from RPE during graft preparation is the stimulus for photoreceptor axon terminal retraction. ¹⁷ This perturbation results in spreading depression, a massive wave of depolarization known to follow metabolic, ischemic, or traumatic injuries to neuronal tissue, ²⁹ that presumably results in a myriad of changes in cellular homeostasis, including alterations in ionic channel conductance, signal transduction pathways, gene expression, and protein turnover. 

In addition to axon terminal retraction, mislocalization of opsin was seen in these specimens. Opsin labeling, which is normally concentrated in the rod OS, was observed along the entire photoreceptor cell membrane after 24 hours in culture. This phenomenon has also been described previously in isolated photoreceptor cells in vitro ²⁸ and in vivo, in experimental retinal detachment, ³⁰ and in diseased retina. ^{13 14 16 31} The underlying mechanism for mislocalization of normal opsin is unknown but may involve disruption of membrane targeting or intersegmental fusion of OSs and ISs, leading to loss of the barriers that mediate compartmentalization of opsin molecules. ^{32 33}  

We hypothesize that retraction of photoreceptor axon terminals toward their cell bodies represents an obstacle to graft–host synaptic connectivity after transplantation into the subretinal space. Indeed, one of the major problems encountered in retinal transplantation is the absence or paucity of graft–host synaptic interactions. In contrast, the presence of opsin in axonal membranes is unlikely to prevent synaptogenesis, because adult rod cells in culture with mislocalized opsin can form synapses, ^{6 28} and developing rod cells can contain opsin labeling throughout their membranes while the outer synaptic layer is created. ³⁴ Opsin labeling is also found throughout developing photoreceptors in animal models of retinal degeneration in which synapses are formed before degeneration is severe. ^{11 31}  

Applying CPT-cAMP to photoreceptors prepared as a sheet for transplantation led to approximately 93% reduction of retraction for up to 48 hours. CPT-cAMP also blocked retraction in full-thickness retinas. However, full-thickness preparations were slightly less inhibited than the photoreceptor sheets after 45 minutes and 24 hours of incubation. A less effective blockage in the intact neural retinas may have occurred because of the reduced accessibility of the cAMP analogue to the photoreceptors. Alternatively, CPT-cAMP may induce other retinal neuronal or glial cell types to release factors that counteract the effect of the cAMP analogue on photoreceptors. Forskolin was less effective than CPT-cAMP, which blocked retraction completely at both 45 minutes and 24 hours, whereas forskolin allowed a small but significant increase in labeling in the ONL at these times and after 48 hours. The complex interactions that control adenylyl cyclase activity may underlie the relatively less effective blocking of retraction by forskolin. In photoreceptors, for instance, adenylyl cyclase activity is downregulated by dopamine and light, ³⁵ whereas its expression may be controlled by circadian rhythm. ³⁶ It is possible that detachment itself reduces adenylyl cyclase expression as it does for other photoreceptor proteins. ³⁷ Variations in adenylyl cyclase protein levels affect the extent of cAMP production. Inhibition of retraction by either drug, however, was not due to toxicity or cell death, as shown by both the ability to reverse the blockage of retraction and by LDH release assays. 

Role of cAMP in Inhibition of Retraction

Axonal retraction is dependent on changes in the cytoskeleton. Cyclic AMP may modulate actin cytoskeletal dynamics through interactions of its downstream effector, cAMP-dependent PKA, with a member of the Rho subfamily of small guanosine-triphosphatases (GTPases), RhoA, which has been shown to inhibit process outgrowth and promote neurite retraction in vitro, in neuronal cell lines and primary neuronal cultures, and in vivo. ^{38 39 40 41} Inactivation of RhoA by ADP-ribosylation using the Clostridium botulinum C3 exoenzyme promotes neurite outgrowth in vitro ^{38 39 40} and in vivo ^{40 41} even on complex inhibitory substrates. ⁴⁰ Furthermore, inactivation of the Rho-associated kinase (ROK) also promotes neurite outgrowth. ⁴¹ Cyclic AMP-activated PKA phosphorylates RhoA decreasing its guanine-nucleotide exchange, possibly by increasing its affinity to guanine-nucleotide dissociation inhibitors (GDIs). GDI-bound RhoA dissociates from the membrane and translocates to the cytosol, where it remains inactive. ⁴² In addition, phosphorylation of RhoA reduces its affinity to its effectors (e.g., ROK). ⁴²  

We propose that cAMP-mediated phosphorylation downregulates the RhoA signaling pathway and thus prevents retraction. Other GTPases of the Rho family, Rac and Cdc42, also regulate the actin cytoskeleton, but in contrast to RhoA, they serve to enhance lamellipodial and filopodial process extension, respectively, and thereby counteract the effects of RhoA. ^{39 43} Inhibition of RhoA, due to the increased cAMP in our specimens, coupled with the unopposed activity of Rac and Cdc42, may serve to shift the neuronal terminal toward a process of outgrowth. Indeed, some neuritic processes have been observed in full-thickness specimens, which provide some cellular substrate for process growth, after 48 hours of forskolin treatment (Khodair MA, Townes-Anderson E, personal observations, 2001). 

Cyclic AMP-mediated phosphorylation also influences other cytoskeletal proteins. PKA can directly phosphorylate and inactivate myosin light-chain kinase, thus preventing phosphorylation of the regulatory subunit of myosin light chain. This inactivation reduces myosin II ATPase and motor activity and therefore may decrease the retraction-mediating contractile forces of actomyosin. PKA also phosphorylates microtubule-associated proteins (MAPs) which play a key role in regulation of microtubule dynamics, neuronal morphogenesis, and plasticity. ⁴⁴ MAP2c and MAP5, which are expressed in the developing brain during periods of neurite growth and synaptogenesis, are present in adult photoreceptors. ⁴⁴ PKA phosphorylates MAP2c on specific serine residues located in the tubulin-binding domain and adjacent proline-rich regulatory region. ^{45 46} These site-specific phosphorylations promote neuritogenesis, ⁴⁵ possibly by causing MAP2c to localize to the actin cytoskeleton. ⁴⁶  

Although cAMP through its downstream effector PKA can play a pivotal role in regulation of both the actin and microtubule components of the cytoskeleton, it may also influence axonal plasticity through its effects on gene expression and protein synthesis. The relative importance of changes in gene expression would be expected to increase with time. 

In summary, experiments with CPT-cAMP have demonstrated that early morphologic changes in the axons of adult photoreceptors that occur during preparation of grafts and possibly after transplantation into the subretinal space can be prevented by pharmacological manipulation. This step may be important in the evolution of successful retinal transplantation techniques, since these changes are thought to interfere with the ability of the grafted photoreceptors to establish connectivity with host neurons. Similar effects of cAMP may be expected in transplants of other differentiated nerve cells. 

 Supported by National Eye Institute Grant EY12031 (ET-A); the F. M. Kirby Foundation (ET-A, MAZ); Research to Prevent Blindness, Inc.; NJ Lions Eye Research Foundation; and Eye Institute of New Jersey (MZ).

 Submitted for publication May 22, 2004; revised September 30, 2004; accepted October 5, 2004.

 Disclosure: M.A. Khodair, None; M.A. Zarbin, None; E. Townes-Anderson, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Intact retinas or photoreceptor sheets, in medium alone or with CPT-cAMP, double-labeled for synaptophysin (green) and rod opsin (red). (A) Control retina fixed immediately after removal from the eyecup. OS, outer segments; IS, inner segments; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer; and NFL, nerve fiber layer. Opsin labeling was primarily confined to the rod OSs (arrowheads), which appeared well formed. Synaptophysin labeling highlighted presynaptic terminals in the IPL and rod and cone terminals in the OPL. Some synaptophysin labeling was also present in the ISs (arrows). (B) Photoreceptor sheets with no treatment after 48 hours in vitro labeled for synaptophysin (top left), opsin (top right), and double labeled (bottom). Synaptophysin labeling was present in the OPL and ONL (arrows). Individual terminals were seen retracting toward their cell bodies, creating gaps in OPL labeling (large arrowheads). Opsin labeling was present in the OS, cell body, and terminal of rod cells (small arrowheads). Synaptophysin and opsin colocalized in both the retracting and stable photoreceptor terminals (yellow). OSs were distorted. (C) Photoreceptor sheet treated with 0.1 mM CPT-cAMP for 48 hours. Synaptophysin labeling remained confined to the OPL (arrowheads), indicating the inhibition of axon terminal retraction. Opsin labeling remained dispersed. Arrows: OSs. (D) Higher magnification of CPT-cAMP-treated photoreceptor sheet. Left: opsin labeling (rhodamine channel) delineated the entire rod photoreceptor cell membrane, including the terminals. Middle: synaptophysin-labeled terminals (fluorescein channel) remained in the OPL. Right: opsin and synaptophysin (both channels) colocalized in the terminals. (E) Full-thickness retina without treatment for 48 hours. As in the untreated photoreceptor sheet, synaptophysin labeling spread into the ONL and opsin throughout the rod cells. (F) Retina with 0.1 mM CPT-cAMP treatment for 48 hours showed little movement of synaptic labeling (arrows) but substantial opsin labeling in the ONL (arrowheads). Scale bar: (A, E) 25 μm; (D) 10 μm. (G, H) Treatment with 0.1, 0.5, or 1 mM CPT-cAMP. (G) Photoreceptor sheets receiving no treatment showed a significant increase in ONL labeling compared with the control by 45 minutes in vitro. The area of labeling in the ONL increased approximately 10-fold by 24 hours and was slightly increased again after 48 hours in culture. For treated specimens, there was no significant difference in the area of ONL synaptophysin labeling among treatment groups or in the control group for up to 24 hours. By 48 hours, treatment groups showed small but significant increases in labeling compared with the control, but no difference compared with each other. Labeling in the ONL of treated specimens was only 7% to 11% of that in untreated specimens. (H) Full-thickness retinas showed a similar pattern. Untreated retinas had a significant increase in ONL labeling compared with control and treatment groups. Treated specimens showed only a small, but significant, increase in labeling when compared with control specimens at all time points. * P < 0.001; mean ± SEM; n = 10 eyes of five animals.

Figure 2.

 

Removal of CPT-cAMP. (A) Photoreceptor sheets that received no treatment showed the expected increase of synaptophysin labeling in the ONL at 45 minutes, 24 hours, and 48 hours. In CPT-cAMP-treated photoreceptor sheets, synaptophysin labeling in the ONL remained at very low levels throughout the experiment, and became significantly different from the controls only after 48 hours in culture. Removal of CPT-cAMP was performed for two groups: one after 45 minutes and one after 24 hours. Sheets were then incubated for an additional 24 hours. A fivefold increase in labeling occurred after 24 hours. The increase in labeling occurred regardless of the duration of treatment and was significant compared with controls and CPT-cAMP-treated sheets. *P < 0.001; mean ± SEM; n = 6 eyes of three animals. (B) Photoreceptor sheets maintained for 48 hours in medium alone (left), with 0.1 mM CPT-cAMP (middle) or with CPT-cAMP for 24 hours and then with medium alone for 24 hours (right). Synaptophysin labeling moved into the ONL in untreated photoreceptor sheets, as expected (arrowheads). Labeling remained in the OPL in sheets with CPT-cAMP treatment. In specimens in which treatment was reversed, labeling occurred in the ONL (arrows). Thus, retraction of terminals toward their cell bodies appears to have resumed. Scale bar, 25 μm.

Figure 3.

 

Forskolin treatment. Within 45 minutes, untreated photoreceptor sheets showed a significant (approximately onefold) increase in synaptophysin labeling in the ONL, compared with the control. After 24 and 48 hours, specimens showed a 24-fold increase in ONL synaptophysin labeling compared with the control. One micromolar forskolin completely prevented the spread of synaptophysin labeling into the ONL whereas 0.1- and 10-μM forskolin treatments showed a small but significant increase in ONL labeling compared with the control at 45 minutes. After 24 hours, all forskolin-treated specimens showed a significant increase in ONL labeling compared with the control, but only 10% to 11% of that in untreated specimens. After 48 hours, all specimens still showed significant ONL labeling compared with the control but less labeling than that in untreated specimens. Lower concentrations of forskolin were more effective at reducing labeling. * P < 0.05; mean ± SEM; n = 6 eyes of three animals.

Figure 4.

 

LDH assay of cell viability. Cell-free aliquots of medium were collected from cultures of photoreceptor sheets maintained in medium only or treated with 1 mM CPT-cAMP (A), 1 μM forskolin (B), or 100 μM ouabain for up to 48 hours. The only significant increase in colorimetric absorbance occurred with ouabain treatment after 48 hours. *P < 0.05; n = 6 eyes of three animals.