Molecular Vision 2017; 23:90-102
Received 18 August 2016 | Accepted 06 March 2017 | Published 08 March 2017
The first two authors contributed equally to this study.
1Department of Cell Biology and Physiology, The University of North Carolina at Chapel Hill, NC; 2The Neuroscience Center, The University of North Carolina at Chapel Hill, NC; 3The Lineberger Comprehensive Cancer Center, The University of North Carolina at Chapel Hill, NC
Correspondence to: Ellen R. Weiss, The University of North Carolina at Chapel Hill, Department of Cell Biology and Physiology, 5340B MBRB, Chapel Hill, NC 27599; Phone: (919) 966-7683; FAX: (919) 966-6927; email: firstname.lastname@example.org
Dr. Amelia Bachleda is now at: The Institute for Learning and Brain Sciences, University of Washington, Seattle, WA 98195.
Dr. Enheng Dong is now at: The School of Public Health, Xinxiang Medical University, Xinxiang City, Henan Province, China 453003.
Purpose: The mechanisms that trigger retinal degeneration are not well understood, despite the availability of several animal models with different mutations. In the present report, the rd10 mouse, a model for retinitis pigmentosa (RP) that contains a mutation in the gene for PDE6β (Pde6b), is used to evaluate gliosis, as a marker for retinal stress, and cyclic AMP response element binding protein (CREB) phosphorylation, which may be important early in retinal degeneration.
Methods: Wild-type C57Bl6J and rd10 mice raised under cyclic light were examined for changes in gliosis and CREB phosphorylation for approximately 3 weeks beginning at P14 to P17 using immunocytochemistry. Mice raised under normal cyclic light and in complete darkness were also compared for changes in CREB phosphorylation.
Results: Gliosis in rd10 mice raised under cyclic light was apparent at P17, before extensive degeneration of the photoreceptor layer is visible, and increased over time. Phosphorylation of CREB at Ser133 (pCREB) was detected in Müller glia (MG) in the wild-type and rd10 mice. However, at all phases of photoreceptor degeneration, the pCREB levels were lower in the rd10 mice. We also observed extensive migration of MG cell bodies to the outer nuclear layer (ONL) during degeneration. In contrast to the mice raised under cyclic light, the rd10 mice raised in the dark exhibited slower rates of degeneration. When the dark-reared mice were exposed to cyclic light, the photoreceptor layer degenerated within 4 days to approximately one to two rows of nuclei. Interestingly, the pCREB levels in the MG also decreased during this 4-day cyclic light exposure compared to the levels in the rd10 mice raised continuously in the dark.
Conclusions: The results of these studies suggest that photoreceptors communicate directly or indirectly with MG at early stages, inducing gliosis before extensive retinal degeneration is apparent in rd10 mice. Surprisingly, phosphorylation of CREB is downregulated in the MG. These results raise the interesting possibility that Müller glia undergo CREB-mediated transcriptional changes that influence photoreceptor degeneration either positively or negatively. Unlike canine models of RP, no increase in pCREB was observed in photoreceptor cells during this period suggesting possible mechanistic differences in the role of CREB in photoreceptors between these species.
Retinitis pigmentosa (RP) is a class of retinopathies typically characterized by rod photoreceptor degeneration followed by cone degeneration and leads, in most cases, to total blindness. Approximately 4% to 5% of patients with recessive RP have mutations in the genes for PDE6α, and 3% to 4% have mutations in PDE6β, the catalytic subunits of cGMP-phosphodiesterase 6 (PDE6) [1,2]. The study of mouse models with mutations in orthologous proteins provides information on the critical factors that cause RP in humans. In rods, PDE6 is composed of catalytic α and β subunits and two inhibitory γ subunits. Light-activated rhodopsin stimulates the activation of its G protein, transducin (Gt), which activates PDE6 by the binding of the inhibitory PDE6γ subunits to Gtα. The breakdown of cGMP catalyzed by activated PDE6 leads to closure of the cGMP-gated ion channels and hyperpolarization of rod photoreceptors . These events are the initial steps in phototransduction. The rd10 mouse possesses a mutation in the pde6β gene that reduces the level of this enzyme and results in a retinal degeneration phenotype . In rd10 mice, degeneration begins at approximately P17–20 [4-6]. This timing makes it possible to distinguish biochemical and transcriptional events that are involved early in retinal degeneration from those that occur during normal postnatal retinal development.
The principal glial cells in the retina are the Müller glia (MG), which support the survival and function of the neuronal population through various mechanisms, including playing a protective role in response to retinopathic insults . Thus, MG are highly sensitive to genetic and environmental stress in neurons and to physical damage (e.g., diabetic retinopathy, proliferative retinopathies, retinitis pigmentosa, and retinal detachment) [7-10]. These conditions result in disruption of multiple functions of the MG, including K+ homeostasis in the extracellular environment, ammonia detoxification, and glutamate recycling. MG also undergo reactive gliosis, manifested as increased expression of intermediate filaments, such as glial fibrillary acidic protein (GFAP) and vimentin, hypertrophy, and the secretion of cytokines and neurotrophic factors .
Cyclic AMP response element binding protein (CREB) is a ubiquitous nuclear factor that assembles protein complexes to initiate gene transcription when phosphorylated on Ser133 (pCREB) . CREB is known to play a protective role against degeneration in the central nervous system . In several canine RP models with photoreceptor-specific mutations in genes, a dramatic upregulation of pCREB is observed in photoreceptor cells during degeneration . Therefore, we evaluated the phosphorylation of CREB on Ser133 (pCREB) during photoreceptor degeneration in the rd10 mice to determine whether CREB might be activated and could play a role in either inhibiting or enhancing retinal degeneration. In contrast to the results reported in similar canine models of RP, we did not observe pCREB in the photoreceptor cells in the rd10 mice. The difference between canine and mouse models for RP illustrates species variation, although the mutations are in the same proteins, possibly due to different rates of degeneration or mechanistic differences in the photoreceptor degeneration process. Defining the mechanisms behind these species-specific differences may contribute to a better understanding of RP.
In the wild-type and rd10 retinas, we detected pCREB consistently in the inner nuclear and ganglion cell layers. However, in the rd10 retinas, the pCREB levels were lower in the MG compared to the MG in the wild-type retinas. The lower pCREB levels were observed before and during the migration of the MG cell bodies toward the outer nuclear layer (ONL), a known response of gliotic MG in the rd10 retina . The levels of pCREB in the MG were also found to be influenced by light in the rd10 mouse. Because light is an exacerbating factor for retinal degeneration in this model, these data suggest the intriguing possibility that transcriptional regulation by CREB in MG is affected by light and influences the progress of retinal degeneration.
The antibodies used in this study were purchased from companies listed in Table 1 (primary antibodies) and Table 2 (secondary antibodies). Anti-pCREB was purchased from Cell Signaling (Danvers, MA) as either an unconjugated antibody (anti-pCREB; catalog #9198) or as a conjugate to Alexa Fluor 488 (anti-pCREB- Alexa Fluor 488 conjugate; catalog #9187; see Table 1). Two antibodies were directly conjugated to florescent dyes in our laboratory as follows. The antibody for SOX9 (#AB5535; Millipore; Billerica, MA) was conjugated to CF555 by diluting it 1:2 with the Mix-n-Stain CF555 antibody labeling kit (Biotium, Inc.; Hayward, CA) according to the manufacturer’s directions. Similarly, Iba1 (#019–19741; Wako Chemicals USA, Inc.; Richmond, VA) was conjugated to CF488A by diluting it 1:2 using the Mix-n-Stain CF488A kit from the same company. Hoechst33258 was purchased from ThermoFisher Scientific Inc. (Pittsburgh, PA). ProLong Gold antifade mounting media containing 4',6-diamidino-2-phenylindole (DAPI) was purchased from ThermoFisher Scientific Inc. λ-phosphatase was purchased from New England Biolabs (Ipswich, MA).
Wild-type and rd10 mice on a C57BL/6J background were obtained from Jackson Laboratories (Bar Harbor, MA). The age-matched wild-type and rd10 mice used for the experiments were raised either under a normal 12 h:12 h light-dark cycle or in total darkness as described in the Results. Mice raised in cyclic light were reared on the same shelf to match the light exposure in the cages as closely as possible.
Mice raised in cyclic light were euthanized under ambient white light, and dark-raised animals were euthanized under red light (LED with a peak wavelength at 660 nm ± 20 nm, LEDtronics, Inc., Torrance, CA) by cervical dislocation following procedures in compliance with the Institutional Animal Care and Use Committee at the University of North Carolina at Chapel Hill and in adherence to the ARVO Statement for Use of Animals in Research. Euthanasia was performed between 13:00 and 15:00 to avoid any influence of the time of day. The eyes were enucleated and incubated in 4% paraformaldehyde (PFA) in PBS (1X: 137 mM NaCl, 2.7 mM KCl, 1 mM Na2HPO4, 1.5 mM KH2PO4, pH 7.2) for 1 h, followed by removal of the anterior segment and lens in HEPES-Ringer buffer containing 10 mM HEPES, pH 7.5, 120 mM sodium chloride, 0.5 mM potassium chloride, 0.2 mM calcium chloride, 0.2 mM magnesium chloride, 0.1 mM EDTA, 10 mM glucose, and 1 mM dithiothreitol (DTT). The eyecups were incubated with 4% paraformaldehyde (PFA) in PBS overnight at 4 °C, washed five times with PBS, and equilibrated sequentially in 10%, 20%, and 30% sucrose in PBS, followed by embedding in optimum cutting temperature compound (OCT) and freezing at −80 °C. The eyecups were cryosectioned at 12-μm thickness at −20 °C and stored at −80 °C until used for immunocytochemistry. For all imaging and quantification in Figure 1, Figure 2, Figure 3, Figure 4, and Figure 5 (with the exception of the image of the P34 mice shown in Figure 4B), the regions of the retina adjacent to the optic nerve were selected to ensure reproducibility across experiments.
For GFAP and SOX2 costaining (Figure 1), the cryostat sections were blocked in PBS containing 10% goat serum and 1.0% Triton X-100 for 2 h. The sections were incubated overnight with a polyclonal antibody against GFAP at 1:500 and a monoclonal antibody against SOX2 at 1:100 in PBS containing 5% goat serum and 0.1% Triton-X-100 at 4 °C. After three washes in PBS, the samples were incubated with Alexa Fluor 488 goat anti-rabbit immunoglobulin (IgG; 1:2,000) and Alexa Fluor 546 goat anti-mouse IgG (1:1,000) for 1 h at room temperature in the same buffer as the primary antibody incubation. After two washes in PBS, the samples were incubated for 5 min in Hoechst33258 at 1:10,000 to stain the nuclei, followed by an additional three washes in PBS. A Zeiss LSM710 microscope (Carl Zeiss Microscopy, Thornwood, NY) was used to collect Z-stacks at 20X, which were processed as maximum projections using the computer program, Adobe Photoshop (Adobe Systems Incorporated, San José, CA).
For immunostaining with CREB, pCREB, SOX9, and Iba1 antibodies (Figure 2, Figure 3, Figure 4, and Figure 5), the slides were placed at room temperature overnight to enhance the attachment of the sections to the slides. These slides were incubated in PBS for 5 min, incubated in blocking buffer (PBS containing 0.5% Triton X-100 and 5% normal goat serum) for 1–2 h and then incubated overnight at 4 °C in PBS containing 0.5% Triton X-100 and primary antibody at concentrations indicated in the figure legends. After washing five times with PBS, the sections were stained with the anti-pCREB antibody 87G3 and the anti-CREB antibody 86B10 followed by incubating with secondary antibodies, Alexa Fluor 555 goat anti-rabbit IgG and Alexa Fluor 488 goat anti-mouse IgG at 1:1,000 at room temperature for 2 h (Figure 2B). The secondary antibody step was eliminated when the sections were costained for pCREB and SOX9 (Figure 3 and Figure 5) because the antibodies were directly conjugated to fluorophores 488A (anti-pCREB-Alexa Fluor 488 conjugate from Cell Signaling; see Table 1) and CF555 (by us, using the Mix-n-Stain CF555 antibody labeling kit from Biotium; see above), respectively. The sections were then rinsed five times in PBS and covered with ProLong Gold antifade mounting media containing DAPI to stain the nuclei. Images were collected using either an Olympus FV1000 (Olympus America, Center Valley, PA) or a Zeiss 880 confocal microscope. Under the Olympus confocal microscope, the samples were excited with the 405 nm diode laser, the 488 nm spectral line of the Argon ion laser, and the 543 nm helium-neon laser to acquire fluorescence. A series of z-stacks was acquired with an Olympus PLAPON 60x/1.42 objective in sequential mode to avoid bleed-through (Figure 2B and Figure 5). Under the Zeiss confocal microscope, the samples were excited with a 405 diode laser, the 488 nm spectral line of an Argon-ion laser, and a 561 nm-Helium Neon laser to acquire fluorescence. Z-stacks were acquired with a Zeiss Plan-Apo 63x oil/1.4 objective in sequential mode (Figure 3 and Figure 4).
To validate the specificity of the anti-pCREB antibody for the phosphorylated form of this transcription factor (Figure 2A), slides containing wild-type mouse sections were preincubated for 10 min in PBS containing 0.3% Triton-X-100 and 1% bovine serum albumin (BSA). The slides were rinsed three times in PBS followed by incubation with or without 3,000 units of λ-phosphatase for 2 h at 37 °C in 250 μl buffer supplied by the manufacturer. The slides were subsequently rinsed five times in PBS and stained for pCREB with anti-pCREB-Alexa Fluor 488 conjugate. Z-stacks were acquired on a Zeiss microscope as described above.
For quantification of pCREB staining in the MG compared to other cells in the wild-type and rd10 retinas (Figure 3 and Figure 5), individual z-stack images were selected and converted to RGB TIFF files. The pixel density of an area of pCREB staining (green) in 20 randomly selected MG cells (identified by SOX9 staining; red) and 20 randomly selected pCREB-positive cells in the inner nuclear layer (INL) that were not MG (non-Müller glia; NMG) was quantified using Fiji (NIH Image, NIH). The ratio of the pCREB staining intensity between the MG and NMG cells (“MG/NMG”) was obtained and analyzed using Microsoft Excel (Microsoft, Redmond, WA) and Prism 6 (GraphPad Software, Inc., San Diego, CA). Prism 6 was also used to calculate the standard error of the mean (SEM) and one-way analysis of variance (ANOVA) using Sidak’s multiple comparisons test as measures of statistical significance where noted in the figure legends.
The onset of photoreceptor degeneration in the rd10 mouse ranges from 17 to 20 days after birth, depending on the study and location in the retina [4-6]. Although a mutation in the pde6b gene is the underlying cause, the mechanism that triggers degeneration is unknown. Changes in the retinal gene expression profile in these mice include genes involved in cell survival, apoptosis, and inflammation . Upregulation of GFAP, an intermediate filament protein, has been reported in light-damaged retinas within 24 h  and in retinas from rd10 mice at later stages in degeneration during widespread photoreceptor cell death (P28; ), but the timing of early GFAP expression in rd10 mice has not been analyzed. To determine when MG sense stress from photoreceptors, we examined GFAP expression in retina sections from rd10 mice of different ages. In the experiment shown in Figure 1, Müller glia ranging from P14 to P28 were costained with an antibody against GFAP and an antibody that recognizes SOX2, a Müller glial and amacrine cell marker . In the wild-type P28 retinas, GFAP staining was observed in the MG end feet and in the astrocytes located in the ganglion cell and nerve fiber layers (Figure 1A) as described previously . In some sections, GFAP staining was also observed in the outer plexiform layer (OPL), which has been described previously, although it is not clear what this staining represents [17,18]. These observations were consistent with reports that, unlike astrocytes, MG do not express significant levels of GFAP under homeostatic conditions [7,19,20]. The MG in the rd10 retinas at P14 exhibit no difference in GFAP staining compared to the MG in the wild-type retinas (Figure 1F compared to Figure 1C). In contrast, at P17, the typical longitudinal streaks that represent GFAP upregulation associated with gliosis were observed (white arrows in Figure 1I). This is the earliest time at which gliosis has been detected in the rd10 mouse. By P22, and at later stages of retinal degeneration, GFAP staining in longitudinal streaks was abundant in the rd10 retinas (Figure 1L,O,R). These results demonstrate that upregulation of GFAP expression occurs at early stages of degeneration before severe loss of photoreceptors, suggesting early communication from the photoreceptors to the MG in response to stress. Migration of MG cells into the ONL was also observed at P22, as shown by the staining of MG with the anti-SOX2 antibody (white arrowheads; Figure 1J,K).
Based on reports that CREB activity can perform an antiapoptotic function in retinas undergoing stress , pCREB levels were compared in wild-type (Figure 3A) and rd10 (Figure 3B) mouse retinas over approximately 3 weeks beginning at P17, a period during which progressively greater degeneration of the photoreceptor layer was visible, based on the thickness of the ONL (Figure 3B). An antibody to SOX9, a transcription factor expressed exclusively in MG in the adult retina , was used to localize the MG cell bodies. The MG in the wild-type retinas were colabeled for SOX9 and pCREB, confirming that CREB is phosphorylated in MG. The wild-type and rd10 retinas were examined at P17 when GFAP staining first became apparent in rd10 mice (Figure 1I). pCREB also colocalized with SOX9 in the MG in the rd10 mouse retinas. Based on quantitative analysis (Figure 3C), the levels of pCREB appeared to be similar between MG and non-Müller glia (NMG) in the wild-type retinas. However, in the rd10 retinas, pCREB was statistically significantly reduced in the MG compared with the NMG. When the ratio of staining for pCREB in MG/NMG was compared for the retinas from the two mouse lines, it is clear that the pCREB levels were statistically significantly lower in the rd10 mice compared with the wild-type mice until later stages of degeneration (approximately P34) when the levels of pCREB in rd10 mice appeared to increase toward wild-type levels (Figure 3D).
Interestingly, pCREB increased statistically significantly in photoreceptors in the rcd1 canine model at midstages of degeneration . In the present studies, faint pCREB staining was occasionally, but not consistently, observed in the ONL at later stages of degeneration (P34) in the rd10 mice, when the ONL was composed of one to two rows of nuclei. These cells are not likely to be MG based on their shape and lack of SOX9 staining (data not shown). According to previous reports describing the time course of photoreceptor degeneration in rd10 mice, these cells are likely to be cones [5,6]. At P25 (Figure 4B), the MG (stained with SOX9; red) were observed to enter the ONL in the rd10 retinas. They appeared to form a partial barrier between the single layer of photoreceptor nuclei and the region where the RPE and the choroid are located (P34; Figure 4B). Recently, microglia have been identified in patients with RP and in mouse models for RP [2,23,24]. Therefore, we stained sections with an antibody to Iba1 (green) to determine whether microglia are located in the ONL. In Figure 4B, the ONL cells are identified as MG rather than microglia based on the presence of SOX9. Microglia cell bodies can be found throughout the inner nuclear and outer plexiform layers (white arrows) in the wild-type (Figure 4A) and rd10 (Figure 4B) retinas, based on Iba1 staining. Some small extensions of microglia processes were visible adjacent to the MG in the ONL at P34.
It has been reported that rd10 mice undergo more rapid degeneration when they are raised under normal cyclic light compared with mice raised in the dark [4,25]. Similarly, several inherited retinopathies in humans are exacerbated by light . Therefore, we examined the levels of pCREB and the thickness of the ONL in the dark-reared rd10 mice at P42 and P67 (Figure 5). Compared with the cyclic light-reared rd10 mice at P34 (Figure 3B and Figure 4B), the ONL at P42 in the dark-reared mice was significantly thicker (Figure 5A; P42), similar to P28 or earlier in rd10 mice raised in cyclic light (Figure 3B). The ONL persisted even at P67 with at least four to five rows of nuclei. In contrast, when the 42-day-old dark-reared mice were placed in ambient cyclic light for 4 days (Figure 5A; P42+4L), there was a rapid degeneration such that the ONL was reduced to one to two rows of nuclei. Interestingly, in these mice, MG that did not express pCREB were detected (Figure 5A; P42+L, white arrows), and the pCREB levels were statistically significantly lower in these mice than in the dark-adapted animals (Figure 5B). These observations are consistent with the results in Figure 3 and indicate that light exposure, rather than simply age, appears to play an important role in the levels of transcription mediated by pCREB and in disease progression. In addition, during the 4-day light exposure, the MG cell bodies became disorganized and began to move into the ONL. In contrast, the MG in the dark-reared animals at P67 were still located in the INL (Figure 5A). These intriguing differences demonstrate the critical importance of light to transcriptional regulation in MG and the overall phenotype of the rd10 mouse.
Müller glia are critical for the proper development, structure, and function of the mature retina [7,10]. In response to genetic and environmental insults to the retina, MG initiate a stress response (gliosis), in which intermediate filaments, such as GFAP and vimentin, are upregulated . In the present study, we report that increased expression of GFAP in MG occurs as early as P17 in rd10 mice, which is before photoreceptor loss becomes severe. This result indicates for the first time that early communication of stress signals occurs between photoreceptors and MG and introduces the question of how interactions between photoreceptors and MG influence the degeneration process. Although it is not clear what signals are transmitted, the upregulation of GFAP in MG during retinal degeneration is part of the remodeling of the retina that occurs alongside neuroplastic changes during degeneration from multiple causes . During remodeling, as described by Marc and Strettoi , MG cell bodies move toward the ONL. Gliosis is upregulated at this time. We observed MG migrating into the ONL at P22-P25 (Figure 1 and Figure 4). In some parts of the ONL, these cells appear to fill the spaces left by degenerating photoreceptors (P34; Figure 4). This activity is likely to be preliminary to the formation of a glial seal, which is thought to be triggered by cone loss and occurs later in the remodeling process in mice undergoing RP .
CREB is one of the transcription factors implicated in neuronal survival, as well as playing a key role in memory formation
and plasticity in the central nervous system . In the present study, we observed a statistically significant reduction in phosphorylation of CREB in the MG during degeneration
in the rd10 mice. Reduced pCREB also coincided with the movement of MG cell bodies toward the ONL in these experiments. There are approximately
4,000 target sites for CREB in the human genome 
Dark-reared rd10 retinas were observed to degenerate more slowly than those raised under cyclic light as previously reported . Exposure of a dark-reared 42-day-old mouse to ambient cyclic light for 4 days resulted in rapid degeneration, with an ONL thickness that was approximately equivalent to P28 in light-reared mouse retinas (Figure 5). Therefore, this change in degeneration kinetics is directly correlated with exposure to light. The prevailing theory in humans and mice with PDE mutations (such as the rd10 mouse) is that a continuous flow of cations through the cGMP-gated ion channels directly induces retinal degeneration . However, if that were the case, one might expect that dark-reared wild-type mice would exhibit accelerated retinal degeneration. In fact, just the opposite appears to be true (Figure 5) [4, 25]. It has also been reported that rod Gtα knockout (Gnat1−/−) mice display slow degeneration, although this mouse should mimic the dark-adapted retina where cGMP levels are high and the ion channels are maintained in an open state. Therefore, the cause of retinal degeneration in humans and animals with defective PDE6 activity is likely to be significantly more complex.
Finally, previous work in canine animals with mutations in the PDE6β gene (rcd1), and in other canine genetic models of RP, such as rcd2, erd, prcd, and T4RHO, demonstrated that phosphorylated CREB is dramatically upregulated in photoreceptors with no obvious indication of changes in pCREB in the INL . Surprisingly, we did not observe a similar result in the rd10 mice. In some sections, a barely detectable amount of pCREB appeared to localize to a few photoreceptor cell nuclei at P34, when the ONL was reduced to one or two nuclear layers (data not shown). Based on previous reports, these photoreceptors are likely to be cones . The reason for this apparent difference between canine and mouse models is unclear. Although the naturally occurring rcd1 canine model of RP the pde6β gene , the time course of normal and pathogenic development in canine retinas is different from that in mice. It takes approximately 60 days for the canine retina to mature , whereas maturation of the mouse retina is considered complete at approximately P30 [39,40]. Abnormal development, characterized by disrupted formation of rod outer segments, is apparent in the canine rcd1 retina at approximately P13, and degeneration begins at approximately P25 . The loss of rods is complete at 1 year. It is possible that the effect of the mutations in canines may occur at an earlier phase in retinal development, thus altering transcriptional networks in ways that are different from that in rd10 mice. In addition, canine retinas produce melatonin whereas C57Bl6 mice do not [41,42]. A role for circadian rhythm in these differences cannot absolutely be ruled out, although we were careful to perform our experiments at the same time of the day to eliminate potential differences caused by photoentrainment. Whether the lack of pCREB in photoreceptor cells is also true in other RP mouse models, other animal models, or patients with retinal degeneration remains to be determined.
NIH grants R01EY012224; R01EY022341. Imaging was supported by the Confocal and Multiphoton Imaging Core of NINDS Center Grant P30NS045892 and by the Hooker Imaging Core, Department of Cell Biology and Physiology. We thank Vladimir Ghukasyan, PhD (The Neuroscience Center) and Robert Currin, PhD (Department of Cell Biology and Physiology) for advice and assistance with confocal imaging. We thank David Courson, PhD and Richard Cheney, PhD for helpful discussions.