Sustained increase in intraocular pressure represents a major risk factor for eye disease, yet the cellular mechanisms of pressure transduction in the posterior eye are essentially unknown. Here we show that the mouse retina expresses mRNA and protein for the polymodal transient receptor potential vanilloid 4 (TRPV4) cation channel known to mediate osmotransduction and mechanotransduction. TRPV4 antibodies labeled perikarya, axons, and dendrites of retinal ganglion cells (RGCs) and intensely immunostained the optic nerve head. Müller glial cells, but not retinal astrocytes or microglia, also expressed TRPV4 immunoreactivity. The selective TRPV4 agonists 4α-PDD and GSK1016790A elevated [Ca2+]i in dissociated RGCs in a dose-dependent manner, whereas the TRPV1 agonist capsaicin had no effect on [Ca2+]RGC. Exposure to hypotonic stimulation evoked robust increases in [Ca2+]RGC. RGC responses to TRPV4-selective agonists and hypotonic stimulation were absent in Ca2+-free saline and were antagonized by the nonselective TRP channel antagonists Ruthenium Red and gadolinium, but were unaffected by the TRPV1 antagonist capsazepine. TRPV4-selective agonists increased the spiking frequency recorded from intact retinas recorded with multielectrode arrays. Sustained exposure to TRPV4 agonists evoked dose-dependent apoptosis of RGCs. Our results demonstrate functional TRPV4 expression in RGCs and suggest that its activation mediates response to membrane stretch leading to elevated [Ca2+]i and augmented excitability. Excessive Ca2+ influx through TRPV4 predisposes RGCs to activation of Ca2+-dependent proapoptotic signaling pathways, indicating that TRPV4 is a component of the response mechanism to pathological elevations of intraocular pressure.
Cells of multicellular organisms experience mechanical stimuli that range from the direct mechanical impact of pulling and stretching to changes in osmotic and hydrostatic pressure (Wang and Thampatty, 2006; Bourque, 2008). Mechanical stretch or pressure activates ion channels in the plasma membrane (Loukin et al., 2010a,b), resulting in depolarization, increased intracellular Ca2+ concentration ([Ca2+]i) (Zabel et al., 1996; Wu and Davis, 2001), and changes in gene expression, cell shape, and cytoskeletal organization (Naruse et al., 1998; Thodeti et al., 2009).
Recent studies have established that members of the transient receptor potential (TRP) superfamily transduce visual, chemical, thermal, mechanical, painful, and osmotic stimuli into Ca2+ fluxes (for review, see Liedtke and Kim, 2005; Kung, 2005; Christensen and Corey, 2007; Sharif-Naeini et al., 2008). We are particularly interested in the possibility that the vertebrate retina, which is exposed to systemic blood pressure, hydrostatic pressure from the CSF, and intrinsic intraocular pressure (IOP), contains one or more pressure-sensitive TRP channels. Pathological elevations in IOP or systemic pressure represent primary risk factors for glaucoma, a group of inherited optic neuropathies characterized by apoptotic loss of retinal ganglion cells (RGCs), degeneration of the optic nerve, and progressive loss of visual fields (Quigley, 2005; Whitmore et al., 2005). The cellular pathophysiology of glaucoma is not well understood, in part because the mechanisms that couple the mechanical stimulus (ΔIOP) to cellular signal transduction remain to be characterized.
In eukaryotic cells, the TRP vanilloid 4 (TRPV4) channel (GenBank accession number NM_022017) represents a polymodal mechanism that transduces osmotic pressure, shear force stimuli, mechanical stretch, and moderate warmth (∼27–37°C) into cation influx with a preference for Ca2+ (PCa/PNa ∼6) (Liedtke et al., 2000; Güler et al., 2002; O'Neil and Heller, 2005). The channel may be activated by intracellular mediators such as arachidonic acid and cytochrome P450-dependent formation of 5,6-epoxyeicosatrienoic acid (Vriens et al., 2004, but see Loukin et al., 2009). In neural tissues, TRPV4 expression has been localized to sensory neurons in dorsal root and trigeminal ganglia, inner ear hair cells, Merkel cells, hippocampal and hypothalamic neurons, and astrocytes (Liedtke et al., 2000; Reiter et al., 2006; Benfenati et al., 2007; Shibasaki et al., 2007; Alessandri-Haber et al., 2009). Mice lacking TRPV4 have defects in noxious mechanosensation and pressure sensation (Liedtke and Friedman, 2003; Suzuki et al., 2003), whereas the functional ortholog of TRPV4 in the worm Caenorhabditis elegans, OSM-9, mediates osmotic, mechanical, and chemical avoidance (Liedtke et al., 2003). Together, these findings suggest that TRPV4 represents an evolutionarily conserved element of the neural response to mechanical stimulation.
We report here that TRPV4 is expressed in RGC somata and optic axon fibers. Transient activation of TRPV4 induced Ca2+ influx and increased the excitability of ganglion cells, whereas sustained activation resulted in RGC apoptosis. The prominent expression of TRPV4-immunoreactivity (IR) at the optic nerve head (ONH), and the role of TRPV4 in gating Ca2+ entry and RGC firing, implicate this channel in the retinal remodeling that occurs during chronic increases in IOP.
Materials and Methods
C57BL/6J mice of either sex were obtained from commercial suppliers, whereas B6.Cg-Tg(Thy1-CFP)23Jrs/J (hereafter, Thy1:CFP) animals or retinal sections were a kind gift from Dr. Ning Tian (University of Utah, Salt Lake City, UT). The animals were maintained in the University animal quarters on a 12 h light/dark cycle. Mice were fed lab chow and water ad libitum. Mice were killed before isolation and dissociation of retinas (Duncan et al., 2006). Animal handling and anesthetic procedures were approved by University Institutional Animal Care committees and conform to National Institutes of Health guidelines.
Eyes were enucleated, and their corneas and lenses dissected away. The remaining posterior pole of the eye was fixed by immersion for 1 h at room temperature in freshly prepared 4% paraformaldehyde in 0.1 m phosphate buffer, pH 7.2, then washed 3 × 10 min in pH 7.2 PBS. Fixed eye tissue was then immersed for 12–16 h in 30% sucrose at 4°C, embedded in OCT (Ted Pella), cryostat sectioned at 16 μm, and mounted on Superfrost slides (Fisher Scientific). After drying for 1 h at 37°C, slides were stored frozen (−80°C) until used. Dissociated cells were isolated with papain (7 U/ml; Worthington), plated on concanavalin A-coated coverslips, fixed in 4% paraformaldehyde in 0.1 m phosphate buffer, pH 7.2, for 30 min, washed 3 × 10 min in pH 7.2 PBS, and stored at 4°C.
For fluorescence immunocytochemistry, cryostat sections were thawed and washed in PBS, then placed for 30 min to 1 h in blocking solution (10 ml of PBS, 30 μl of Triton X-100, 100 mg of bovine serum albumin, 100 μl of 10% w/v Na azide solution). Primary and secondary antibodies were diluted in blocking solution and applied for 2 and 1 h, respectively, at room temperature, with three intervening washes in PBS. After a final set of washes in PBS, the sections were covered in VectaShield (Vector Laboratories).
Terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end labeling assay.
Retinas were embedded, cryosectioned, and processed for terminal deoxynucleotidyl transferase-mediated biotinylated UTP nick end labeling (TUNEL) following the protocol by Gavrieli et al. (1992). After 30 min rehydration in 70% ethanol, 2× 5 min rinses in PBS and 1% Triton-X in 1% citrate buffer, pH 7.3, the slides were subjected to a final rinse in PBS. Slides were then incubated for 30 min in the reaction buffer (30 mm Trizma-HCl, 140 mm Na+ cacodylate, 1.0 mm CaCl2, 0.2% Triton X-100, pH 7.2). Positive controls were treated with DNase I (Roche Diagnostics) (10 U/ml) in the reaction buffer. Next, slides were treated for 1 h with 0.03 U/μl terminal transferase and 4 μm biotin-16-dUTP (Roche Diagnostics). The reaction was terminated in 30 mm Na+ citrate, 300 mm NaCl, and 0.2% Triton X-100 in PBS (5 min), and rinsed in PBS (2×5 min), exposed to 1% bovine serum albumin in PBS for 20 min, and rinsed. As a negative control, the retinal sections were processed omitting the incubation step with terminal deoxynucleotidyl transferase during DNA labeling. Apoptotic cells were visualized with a confocal microscope (LSM510; Zeiss) and categorized by size. We found that our fixation protocol resulted in ∼1 μm shrinkage in the diameter of dissociated Brn3a-IR cells in retinal sections. Therefore, the TUNEL analysis compared cells with a diameter of >6 μm (presumed RGCs) to photoreceptor cells with diameters of 3–5 μm (the identity of photoreceptors was confirmed by immunostaining for mouse cone arrestin (a gift from Dr. Wolfgang Baehr, University of Utah, Salt Lake City, UT) and rhodopsin (Santa Cruz Biotechnology). A cell was deemed to be TUNEL positive if the fluorescent signal completely filled the cell body.
Image acquisition and processing
Sections were examined by confocal microscopy (LSM 510, Zeiss; or PM 800, Nikon). Digital images were acquired separately from each laser channel, then recombined. Files were further processed with deconvolution software (AutoQuant Imaging). Adjustments of contrast and intensity were made in Photoshop (Adobe); any such adjustments were made uniformly to the entire image.
We essentially followed the procedure outlined in previous publications (Liedtke and Friedman, 2003; Phan et al., 2009). In brief, homogenized retinal tissue was subjected to protein preparation and denaturation, and proteins (10 μg per lane) were separated on an SDS gel, then immunoblotted to polyvinylidene difluoride membranes. After blocking with milk powder, TRPV4 was immunodetected using the above antibody (Phan et al., 2009) and peroxidase-based chemoluminescence.
Primary and secondary antibodies
Table 1 shows the primary and secondary antibodies used in this study. Multiple commercially available anti-TRPV4 antibodies (ACC-034, Alomone Labs; LS-A8583, MBL International Corporation; OSR00136W, Affinity Bioreagents; and ab63003, Abcam) were tested by immunostaining and Western blots in wild-type and TRPV4-null mice. However, the only antibody that proved completely satisfactory was LS-C94498 (Lifespan Biosciences), based on the following criteria: (1) The antibody stained Western blot bands of 85 and 106 kDa from retinal tissue harvested from wild-type mice but not from TRPV4-null mice (see Fig. 1); (2) it colocalized with green fluorescent protein (GFP) in RGCs from transgenic animals in which GFP was driven by the Trpv4 gene promoter (see Fig. 2C); and (3) it detected protein bands of appropriate molecular mass from HEK293 cells transfected with Trpv4 cDNA, whereas the bands were missing in immunoblots from nonexpressing cells (Fig. 1). All TRPV4-IR data reported in the present study were obtained with the anti-TRPV4 antibody LS-C94498. No immunostaining was observed when the primary antibody was omitted.
Total RNA from retina was extracted with Trizol, and total RNA (2 μg of total RNA used for first-strand synthesis with oligo-dT and primers) was converted to cDNA using the SuperScript III First-Strand Synthesis kit from Invitrogen. PCR products were amplified in a thermocycler (Veriti, ABI) with nucleic acid stain (SYBR Green, ABI) reagents according to the manufacturer's instructions. Amplification of PCR products was measured by fluorescence associated with binding of double-stranded DNA to SYBR Green in the reaction mixture. After an initial denaturation step of 50°C for 2 min and 95°C for 10 min, PCR was repeated for 40 cycles at 95°C for 15 s, 58°C for 30 s, and 72°C for 30 s. After amplification, the ratio of gene-of-interest mRNA to a housekeeping gene, glyceraldehyde-3-phosphate dehydrogenase (Gapdh), was calculated for each sample. Five to 10 independent biological replicates (retinas) were used at each age.
Generation of a bacterial artificial chromosome transgenic mouse line with a fluorescent reporter driven by the Trpv4 promoter
A mouse bacterial artificial chromosome (BAC) harboring the Trpv4 gene was modeled by “recombineering” (Copeland et al., 2001), so that the copepod-GFP coding region was placed directly after the ATG start codon of mouse Trpv4. In addition, this BAC was engineered to not harbor exons 10–14, so that a functional ion channel could not arise from the transgene. Engineered DNA served as a transgenesis template, which was constructed following standard procedures (Zhao et al., 2008). Of four resulting transgenic lines, the one used revealed robust fluorescence in trigeminal ganglion sensory neurons (not illustrated), both in acute sections and acute dissociations, colocalizing with TRPV4 immunolabeling. This line was outcrossed for five or more generations before analysis of retinas.
Calcium imaging was performed on acutely isolated retinal cells, as described previously (Szikra et al., 2009). In brief, dissociated retinal cells were plated on concanavalin A-coated (0.2 mg/ml; Sigma) coverslips, loaded with fura-2 AM (1–5 μm; Invitrogen) for 15–30 min and washed for 10 min in dye-free L-15 medium. Cells were viewed with Nikon Ti inverted or 600EF upright microscopes using 20× 0.95 numerical aperture (NA), 40× 0.85 NA, or 40× 1.25 NA objective lenses. Excitation for 340 and 380 nm filters (Chroma and Semrock) was provided by a 150 W Xenon arc lamp (DG4, Sutter Instruments). Fluorescence emission was high-pass filtered at 510 nm and captured with cooled digital CCD cameras (HQ2, Photometrics). Data acquisition and F340/F380 ratio calculations were performed by NIS Elements software. Fluorescence imaging was performed on regions of interest (ROIs) encompassing the RGC perikaryon, typically at 3 × 3 binning. Background fluorescence was measured in similarly sized ROIs in neighboring areas devoid of cells. After sequential image acquisition (0.167–0.5 Hz) of cell fluorescence at 340/380 nm, the background was subtracted. Calibration of free [Ca2+]i was performed in vivo using 10 μm ionomycin and 10 mm Ca2+ or 0 Ca2+/3 mm EGTA. The apparent free [Ca2+]i was determined from the equation [Ca2+]i = ((R − Rmin)/(Rmax − R)) × (F380max/F380min) × Kd, where R is the ratio of emission intensity at 510 nm evoked by 340 nm excitation versus emission intensity at 510 nm evoked by 380 nm excitation, Rmin is the ratio at zero free Ca2+, Rmax is the ratio at saturating Ca2+, and the dissociation constant Kd for Ca2+-fura 2 at room temperature was taken to be 224 nm (Neher, 1995). Glutamate (100 μm) was added at the beginning of each experiment to control for neuronal health, type, and responsiveness. DMSO, the solvent for the indicator dye, did not induce any responses in RGCs (data not shown). Previous studies using Mn2+ quenching showed that 95% of the fura-2 fluorescence emanates from the cytosol in the cells (Szikra et al., 2009). Experiments were conducted at room temperature, and encompassed stimulation with glutamate, TRPV4, and TRPV1 agonists and antagonists.
RGCs in short-term culture were identified initially by morphology and perikaryal size (7–15 μm). In a subset of experiments, RGCs isolated from Thy1:CFP retinas were used and identified by intrinsic fluorescence (Raymond et al., 2009). Alternatively, test neurons isolated from wild-type retinas were confirmed as RGCs by immunocytochemistry. In each experiment, a 100 μm glutamate stimulus was used to confirm that the visually identified putative RGCs expressed ionotropic glutamate receptors. Presumed RGCs responded to 30 mm KCl with rapid high-amplitude increases in [Ca2+]i, indicating that the cells were healthy and maintained their excitability. In a subset of experiments, cells recorded during stimulation with hypotonic saline or TRPV4 agonists were fixed and immunostained with TRPV4 and Brn3a antibodies.
Solutions and reagents
The isotonic superfusing saline contained the following (in mm): 133 NaCl, 2.5 KCl, 2 CaCl2, 1.5 MgCl2, 1.25 NaH2PO4, 10 HEPES hemisodium salt, 10 glucose, 1 pyruvic acid, 1 lactic acid, and 0.5 glutathione. In Ca2+-free solutions, no external Ca2+ was added, and the saline was supplemented with 1 mm EGTA. The osmolarity and pH of each external solution was measured before each experiment. pH was adjusted to 7.4 with NaOH. Osmolarity was measured with a vapor-pressure osmometer (VAPRO); for control saline, osmolarity was 280 mOsm. For experiments involving hypotonic stimulation, the isotonic ringer contained 132 mm mannitol and the NaCl concentration was reduced to 57.5 mm. Hypoosmotic solutions were prepared by reducing the final concentration of mannitol to 44.5 mm without changing the ionic composition. In a subset of experiments using hypotonic stimulation, cells were coloaded with fura-2 AM + calcein AM (1 μm; Invitrogen). Calcein fluorescence was elicited using 490 nm excitation filters. Because calcein fluorescence is Ca2+ independent and volume dependent, it was used as a measure of changes in the cell volume.
Unless otherwise indicated, the salts and reagents were purchased from Sigma. Ruthenium Red and capsaicin were purchased from Ascent Scientific. Capsaicin and capsazepine were also purchased from Tocris. 4α-PDD was obtained from LC Laboratories.
Multielectrode array recordings and data analysis
Multielectrode array (MEA) recordings were performed using multielectrode, extracellular recording chambers (MEA1060 system, MultiChannel Systems) consisting of an array of 60 planar electrodes, each 10 μm in diameter, spaced 100 μm apart for a total array size of 700 μm2, as described previously (Rentería et al., 2006). Briefly, young adult C57BL6J mice were killed after 1 h of dark adaptation, and an eye was removed under dim red illumination. The retina was dissected and placed with the ganglion cell side against the recording electrodes using a piece of nitrocellulose paper as support. The retina was typically placed on the array within a millimeter of the optic nerve head, and a manipulator (Cell Micro Controls) was used to hold the tissue down with slight pressure. Retinas were perfused at room temperature for 30 min and at 30°C for another 30 min before recordings were started, the temperature was maintained at 30°C thereafter, and spontaneous spiking in the dark was recorded. The perfusion saline was either Ames' medium or Ringer's modified solution as follows (in mm): 124 NaCl, 2.5 KCl, 2 CaCl2, 2 MgCl2, 1.25 NaH2PO4, 26 NaHCO3, and 22.2 glucose; the pH was maintained at 7.3–7.4 by bubbling with 95% O2/5% CO2 mixed gas. Voltage signals sampled at 50 kHz were bandpass filtered at 100 Hz to 3 kHz, and waveforms that crossed a negative voltage threshold (set at −5.5 SDs of the mean noise) were recorded to disk for off-line analysis. Under these conditions, nearly all recorded cells are likely to be RGCs (Rentería et al., 2006).
Spike trains for each RGC were determined by spike sorting based on clustering in principal component space using software (Offline Sorter, Plexon). Not every detected waveform was assigned to a unit; obvious automatic sorting errors were corrected for each cluster manually (Rentería et al., 2006). Time stamps of the action potentials of each sorted unit were used to generate peristimulus time histograms (30 s bins). The average spike rates during the 3 min period before and after drug application were determined for every recorded cell and expressed as the percentage change in firing in the presence of drug. This normalization allowed for comparison of cells with varied initial spike rates. The very few cells that had zero rates before application and spiking during drug application were considered to have increased by 100%.
Data are expressed as the mean ± SEM, with the number of cells, slides, or animals indicated by N. Cell diameter data are expressed as the mean ± SD. Statistical comparisons between two treatments in the same cell were determined using the t test; unless indicated otherwise, comparisons between different groups were evaluated by the Mann–Whitney test or the Wilcoxon signed rank test. Data obtained in multielectrode array studies were analyzed with the Wilcoxon signed rank test. A value of p < 0.05 was considered statistically significant.
TRPV4 mRNA and protein in the mouse retina
RT-PCR was performed using total RNA extracted from adult (postnatal day 90) mouse retinas. Gel analysis revealed PCR products at the appropriate size of 174 bp. The same transcript was amplified in kidney tissue, known to express Trpv4 (Strotmann et al., 2000, Liedtke et al., 2000; Güler et al., 2002) (Fig. 1A).
We next compared TRPV4 protein expression in tissues derived from wild-type and TRPV4-null mice (see Materials and Methods). Consistent with RT-PCR, immunoblots from wild-type mouse retinas showed a primary band of 85 kDa and a secondary band at 105 kDa, which probably correspond to unglycosylated and N-glycosylated forms of TRPV4, respectively, which is in keeping with previous reports (Fig. 1B) (Liedtke and Friedman, 2003; Benfenati et al., 2007; Hartmannsgruber et al., 2007). Both bands were absent in retinal tissue derived from TRPV4-null mice (Fig. 1B). The TRPV4 antibody was validated in Western blots from HEK293 cells transfected with Trpv4 cDNA (Fig. 1C) and in trigeminal ganglion tissue known to robustly express Trpv4 (Liedtke et al., 2000; Liedtke and Friedman, 2003). As an additional test of the TRPV4 antibody, we immunostained the kidney, choroid plexus, and nodose ganglion tissues in which TRPV4 is expressed (Liedtke et al., 2000; Strotmann et al., 2000; Brierley et al., 2008). The patterns of TRPV4 expression in these tissues were identical to those identified in previous reports (data not shown). For example, the somata of neurons within the nodose ganglion (Fig. 1Di) were strongly TRPV4 immunoreactive, whereas the nodose tissue from a TRPV4-null mouse was unlabeled by the TRPV4 antibody (Fig. 1Dii).
TRPV4 immunoreactivity in the mouse retina
TRPV4-IR in retinal ganglion cells
In wild-type mouse retinas, TRPV4-IR was punctate and distributed throughout the inner retina (Fig. 2A). A dense array of TRPV4-IR puncta was seen within cell bodies located in the ganglion cell layer (gcl). TRPV4-IR was distributed at a lower density throughout the inner plexiform layer (ipl), but only very sparse TRPV4-IR was located within the inner nuclear layer (inl). This immunostaining pattern was absent in retinas of TRPV4-null mice (Fig. 2B). The apparent immunostaining of the outer plexiform layer (opl) is nonspecific since it was identical in wild-type and TRPV4-null retinas (Fig. 2A,B). Within the outer nuclear layer (onl) consisting primarily of photoreceptor nuclei, TRPV4-IR was seen in vertical processes of Müller glial cells (see also Fig. 2M for Müller cell signals in the ipl).
We also analyzed a transgenic mouse in which GFP was driven by the entire promoter of the mouse Trpv4 gene. The 50 kDa Trpv4 gene was centered within the 200 kDa GFP insert present in a BAC (see Materials and Methods). Using an antibody against GFP, retinal immunostaining was confined to cell bodies within the ganglion cell layer, although additional staining was noted in retinal blood vessels, in keeping with well established vascular-endothelial TRPV4 expression (Hartmannsgruber et al., 2007; Mendoza et al., 2010). When the retina was stained with anti-GFP, the marker was detected in cell bodies of the ganglion cell layer (Fig. 2C). Confirming that reporter gene expression was faithful to the targeted gene, Trpv4, GFP-IR was absent in wild-type mouse retina (not illustrated).
To probe whether TRPV4-IR is associated primarily or exclusively with ganglion cells, we immunostained the retina for Brn3a (also called POU4f1), a transcription factor found almost exclusively in developing (Liu et al., 2000) and mature ganglion cells (Nadal-Nicolás et al., 2009). We found that TRPV4-IR colocalized with that of Brn3a in ganglion cell perikarya and dendrites (Fig. 2D, arrow). Prominent additional TRPV4-IR was found within ganglion cell axon bundles in the optic fiber layer (ofl) (Fig. 2D), and also was abundant within the optic nerve head (Fig. 2H) and the laminar region of the optic nerve (Fig. 2I). The nearly universal distribution of Brn3a-IR in mouse retinal ganglion cells, together with the finding that Brn3a-IR and TRPV4-IR almost invariably colocalize, indicates that TRPV4 protein is distributed among a large fraction of mouse retinal ganglion cells.
A second marker for ganglion cells was provided by a construct in which GFP was linked to the promoter of Thy1, a ganglion cell-specific protein (see Materials and Methods). GFP-IR and TRPV4-IR colocalize in perikarya and dendrites of RGCs (Fig. 2E) and in displaced ganglion cell perikarya (Fig. 2L). Finally, immunostaining with an SMI-32 antibody that labels intermediate filaments in somata and axonal processes of wide-field αRGCs colocalized with TRPV4-IR processes in the ofl entering the optic nerve head (Fig. 2K).
TRPV4-IR is weak or absent in retinal amacrine cells
Approximately 50% of the cells within the ganglion cell layer in rodent retinas are displaced amacrine neurons rather than ganglion cells, whereas a subset of cells in the proximal inl belong to displaced RGCs (Perry and Walker, 1980; Jeon et al., 1998). To probe possible colocalization of TRPV4-IR with amacrine cells, we immunostained the retina with an antibody against choline acetyltransferase, a marker for starburst amacrine cells (Fig. 2G), and with glutamic acid decarboxylase 65 (GAD-65), a marker for GABAergic amacrine cells (Figs. 2F, 3A) that constitute a substantial fraction of displaced amacrine cells in the RGC layer (May et al., 2007). These tests revealed no colocalization of TRPV4-IR with ChAT-IR cells. Moreover, no colocalization with the dopaminergic amacrine cell marker tyrosine hydroxylase was observed (data not shown). A total of 4.6% of cells that labeled with the GABAergic marker GAD-65 stained with the TRPV4 antibody; however, the intensity of TRPV4-IR signals in this subset of cells was always significantly weaker compared with Brn3a-IR cells. Occasionally, a Brn3a-IR or SMI-32-IR cell body was detected in the inl. These perikarya were always TRPV4 immunopositive (Fig. 2L). Collectively, these data indicate that TRPV4-IR is associated with mouse retinal ganglion cells and may be localized to a small subset of GABAergic amacrine cells. Within the mouse ganglion cell, TRPV4-IR is associated principally with the perikaryon and the axon, but at least some ganglion cell dendrites also express TRPV4 (Fig. 2D,F, arrow).
TRPV4-IR in retinal glial cells
Detailed inspection of TRPV4-IR in the inner and outer retina revealed radial immunopositive processes that correspond to Müller macroglia (Fig. 2M). To assess TRPV4 expression in astrocytes, sections from the retina and at the glial lamina at the optic nerve head were colabeled with the glial fibrillar acidic protein (GFAP) antibody. No colocalization was observed, as TRPV4 and GFAP signals in the retina and the glial lamina appeared to label two distinct populations (Fig. 2I,J). Likewise, costaining with the microglia-specific Mac-1/CD11b antibody showed no overlap with TRPV4 (Fig. 2M) but showed staining in the radial processes emanating from the inner limiting membrane that may correspond to Müller macroglia. These data indicate that in the wild-type mouse retina TRPV4 is excluded from astrocytes and microglia but is expressed in Müller cells.
Physiological tests of TRPV4 function in mouse retina
Identification of RGCs in short-term culture
In retinal neurons isolated into short-term culture (see Materials and Methods) GABAergic neurons, identified by GAD65-IR, were invariably distinct from cells showing TRPV4-IR (Fig. 3Ai–iii). It was also apparent that TRPV4-IR cells were larger than those immunostained by GAD-65 (Fig. 3Aiii,B). Another test of ganglion cell identity in cultured neurons was immunoreactivity to an anti-Brn3a antibody (Fig. 4Ai). Brn3a-IR cells were invariably immunopositive for TRPV4, whereas photoreceptor and presumed amacrine perikarya were not (Figs. 2⇑–4).
We extended measures of cell diameters to Brna3a-IR and Thy1:CFP neurons in culture and compared their dimensions to those of Brna3a-IR neurons in retinal slices; the data are summarized in Figure 3B. Diameters of dissociated Brn3a-IR cells were 9.1 ± 1.5 μm (N = 74; mean ± SD), not significantly different from 10.2 ± 1.8 μm (N = 58) in dissociated Thy1:CFP cells and 10.3 ± 2.0 μm in Brn3a-IR cells in retinal sections (N = 201) (p > 0.05, Dunn's multiple-comparisons test) (Fig. 3B). The distribution of cell diameters confirms that the vast majority of ganglion cells have diameters ≥8 μm, whereas GABAergic amacrines are almost invariably <8 μm. Moreover, the size distributions of ganglion cell diameters based on the different criteria illustrated in Figure 3B are coextensive. Thus, our selection of a neuron for physiological study initially was made on the basis of cell diameter. In a few cases, we recorded Ca2+ signals from cyan fluorescent protein (CFP)-positive RGCs isolated from Thy1:CFP retinas, and in other experiments we confirmed the identity of TRPV4 agonist-responding cells by immunostaining the test cells for TRPV4-IR and Brn3a-IR following drug exposure.
TRPV4 agonists induce an increase in [Ca2+]RGC
Once a putative RGC was selected for experimentation, we first tested the ability of glutamate to evoke an increase in [Ca2+]i. RGCs have ionotropic glutamate receptors that result in ganglion cell depolarization and calcium entry, so a vigorous response to glutamate is a good initial test of RGC viability. To evaluate functional expression of TRPV4 channels in the retina, we measured calcium concentrations in dissociated RGCs before and after TRPV4 agonist application, using ratiometric Ca2+ imaging of RGCs loaded with Fura-2 AM.
On average, glutamate (100 μm) evoked a [Ca2+]RGC increase of 490 ± 25 nm (mean ± SE) over a baseline of 45 ± 2 nm (N = 216; p < 0.0001, Wilcoxon matched-pairs signed ranks test). Endogenous TRPV4 channels in dissociated neurons were activated with the following two selective TRPV4 activators: the synthetic phorbol ester 4α-phorbol 12-myristate13-acetate (didecanoate; 4α-PDD), which directly binds to the transmembrane region of the protein (Watanabe et al., 2002; Vriens et al., 2007; Loukin et al., 2010a); and the recently characterized high-affinity agonist GSK1016790A (hereafter GSK; EC50 ∼1–10 nm) (Thorneloe et al., 2008; Willette et al., 2008). Figure 4B illustrates a simultaneous recording from two putative RGCs that responded to glutamate and the two TRPV4 agonists (GSK, 25 nm; 4α-PDD, 30 μm). At 25 and 100 nm, GSK elicited robust [Ca2+]i increases in 40.5 and 47.2% of glutamate-responding cells (N = 10 and 4 experiments), respectively (Figs. 4, 5, 6). In the continued presence of TRPV4 activators, [Ca2+]i levels typically exhibited a decline, consistent with desensitization of the channel (Fig. 4B).
Mean [Ca2+]RGC elevation induced by 25 nm GSK was 381 ± 41 nm over the [Ca2+]RGC baseline (Fig. 4G) (N = 23; p < 0.0001, Wilcoxon matched-pairs signed-ranks test), whereas 100 nm GSK elevated [Ca2+]i by 600 ± 210 nm (N = 9; p < 0.004, Wilcoxon matched-pairs signed-ranks test). In 15 putative RGCs, 4α-PDD (30 μm) increased the 340/380 ratio from 0.464 ± 0.029 to 1.430 ± 0.097 (p < 0.0001, paired t test) (Fig. 4B,E). When cells that had been stimulated with TRPV4 activators were subsequently immunostained for TRPV4 and Brn3a, the two markers labeled responding cells (Fig. 4D, arrowhead) but not GSK-insensitive cells. The average diameter of cells that responded to TRPV4 agonists was 8.9 ± 1.0 μm, not significantly different from diameters of Thy1:CFP and Brn3a-IR dissociated cells (p > 0.05) (Figs. 3B, 4H). The very substantial overlap of these two populations, responders and immunoreactive cells, provides more evidence that at least the great majority of cells selected for physiological study were retinal ganglion cells. Our data would also suggest that a subset of large-diameter RGCs may not express TRPV4 (Fig. 3B) or respond to TRPV4 agonists (Fig. 4G).
Ruthenium Red and lanthanides are nonselective TRP channel blockers that also antagonize activation of TRPV4 channels (Liedtke et al., 2000; Strotmann et al., 2000; Watanabe et al., 2002). [Ca2+]i responses to GSK and 4α-PDD were suppressed by Ruthenium Red (10 μm; N = 85) (Fig. 4D,E,G). The antagonist decreased the amplitude of GSK-evoked [Ca2+]i responses from 341 ± 41 nm (N = 23) to 139 ± 22 nm (Fig. 4G) (N = 16; p = 0.0001, Mann–Whitney test) and reduced the percentage of cells responding to GSK from 40.5 ± 6.7 (N = 10) to 16.9 ± 4.2% (N = 4; p = 0.026, unpaired t test). In the phorbol ester-insensitive cell depicted in Figure 4E, Ruthenium Red by itself lowered baseline [Ca2+]i (arrowhead), indicating suppression of tonic Ca2+ influx through TRP-like channels. On average, Ruthenium Red decreased baseline [Ca2+]i from 50 ± 3 nm to 47 ± 4 nm (N = 133; p = 0.0003, Wilcoxon matched-pairs signed-ranks test).
The rise in intracellular Ca2+ induced by exposure to GSK depends on extracellular Ca2+. Figure 4F illustrates that a test cell shows normal increases in [Ca2+]i in responses to GSK in 2 mm [Ca2+]o, but this increase was abolished in zero extracellular [Ca2+]o (N = 8; p = 0.78, paired t test). The mean fluorescence changes induced by glutamate or GSK in normal versus zero extracellular Ca2+ are summarized in the inset to Figure 4F.
Ligand-evoked [Ca2+]i increases in mouse RGCs can be augmented by secondary contributions from voltage-operated Ca2+ channels (Hartwick et al., 2008). We therefore investigated whether the response to 25 nm GSK was affected by suppression of voltage-operated Ca2+ entry. In the presence of the nonselective voltage-gated calcium channel antagonist cadmium (Cd2+; 100 μm), high KCl (30 mm)-evoked [Ca2+]i increases were blocked (N = 5; data not shown). Nonetheless, GSK elevated [Ca2+]i to 355 ± 110 nm in Cd2+-containing saline (N = 6), a value not significantly different from responses to the agonist alone (p > 0.89, Mann–Whitney test). Cd2+ had no significant effect on the percentage of GSK-responding cells (31.3 ± 5.2%; p > 0.36, unpaired t test).
A representation of glutamate- and GSK-evoked [Ca2+]i elevations in a presumed RGC is represented graphically in Figure 5. This cell was proximal to a perikaryon from a presumed rod photoreceptor (Fig. 5A). A 60 s exposure to a saturating concentration of the neurotransmitter (100 μm) and sustained (5 min) exposure to GSK (25 nm) evoked an increase in global free [Ca2+]i across the RGC cytosol, with superimposed local [Ca2+]i elevations. The response to the TRPV4 agonist desensitized during continued exposure (Fig. 5J,K) in contrast to responses to sustained [Ca2+]i elevations that were observed during 3–5 min exposures to glutamate. No changes in [Ca2+]i were observed in the simultaneously recorded rod perikaryon.
Hypotonic stimuli evoke changes in cytosolic [Ca2+]i
TRPV4 was originally identified as a plasma membrane channel activated by hypotonic cell swelling (Liedtke et al., 2000; Strotmann et al., 2000; Watanabe et al., 2002; Vriens et al., 2004; Loukin et al., 2010a). A decrease in extracellular osmolarity induces Ca2+ influx through TRPV4 channels (Güler et al., 2002; Liedtke and Friedman, 2003; Raoux et al., 2007; Phan et al., 2009). To determine whether hypotonicity modulates [Ca2+]RGCs, cells were exposed to saline solutions with osmolarity reduced from 280 to 190 mOsm. As illustrated in Figure 6A for an RGC loaded with the Ca2+-insensitive cell-volume indicator dye calcein AM, a reduction in osmolarity of the superfusing saline from 280 to 192 mOsm saline evoked sustained swelling of the cell. The resulting increase in cell volume was detected as a decrease in the intensity of calcein fluorescence (Fig. 6A, green trace). In contrast to hypotonic stimuli, no change in intracellular volume was observed during exposure to 100 μm glutamate.
Cell swelling was accompanied by tonicity-dependent elevations in [Ca2+]i. The 190–195 mOsm saline evoked an average [Ca2+]i increase of 439 ± 59 nm (N = 48; p < 0.0001) (Fig. 6) in 72 ± 7% of putative RGCs. During continued stretch, [Ca2+]i levels gradually returned to control levels, indicating desensitization of the mechanosensing mechanism (Fig. 6). The diameters of cells that responded to hypotonic stimulation with [Ca2+]i increases (8.7 ± 1 μm; N = 126) were not significantly different from diameters of dissociated Brn3a-immunopositive cells (9.1 ± 1.5 μm; p > 0.05; N = 74) (Fig. 6E). Swelling-mediated [Ca2+]i signals also were measured in the presence of gadolinium (Gd3+), a universal antagonist of Ca2+-permeable TRP channels (including TRPV4) (Gustin et al., 1988; Becker et al., 2005). Gd3+ (100 μm) reduced, from 72.1 ± 6.7% (N = 10) to 26.0 ± 9.5% (N = 3 slides), the percentage of RGCs that responded to hypotonic stimulation with a [Ca2+]i increase (p < 0.02, unpaired t test, Welch corrected) but had no effect on cell swelling, as indicated by unchanged calcein responses (Fig. 6A). Overall, the amplitude of hypotonic stretch-induced [Ca2+]i increases was reduced by Gd3+ to 261 ± 42 nm (N = 17; p < 0.05, Mann–Whitney test). Hypotonicity-induced [Ca2+]i responses were suppressed by Ruthenium Red to 206 ± 21 nm (N = 22; p < 0.0001, Mann–Whitney test) (Fig. 6C,D), and the percentage of RGCs responding to hypotonic stimulation in the presence of the antagonist decreased from 72.1 ± 6.7% (N = 10) to 16.9 ± 4.2% (N = 3 slides; p < 0.005, unpaired t test, Welch corrected).
We asked whether hypotonically induced [Ca2+]i elevations are mediated by TRPV4. [Ca2+]i increases induced by TRPV4 agonists and membrane stretch typically desensitized with a time constant of several tens of seconds. Stimulation with GSK at the asymptotic end of the desensitized response elicited little or no change in [Ca2+]i (169.9 ± 66.2 nm), whereas significant [Ca2+]i increases (524.0 ± 63.8 nm) were measured following washout with isotonic saline (Fig. 6G) (N = 14; p < 0.002, Wilcoxon matched-pairs signed-ranks test). Of 84.4 ± 4.6% putative RGCs that responded to hypotonic stimuli in this experiment, 2.0 ± 0.7% exhibited a response to GSK during membrane stretch, significantly less than the 51.7 ± 7% that responded to GSK alone (N = 3 slides, 14 cells; p < 0.05, paired t test). To confirm that hypotonic cell swelling evoked Ca2+ influx from the extracellular space, cells were challenged with Ca2+-free saline. RGCs responding to hypotonic stimulation in 2 mm Ca2+-containing saline with [Ca2+]i elevations displayed no change in [Ca2+]i in Ca2+-free hypotonic saline (N = 0/14) (Fig. 6D). Thus, stretch-induced [Ca2+]i elevations in RGCs primarily depend on entry of extracellular Ca2+ rather than on release from intracellular stores. Together, these data indicate that RGCs express an osmosensitive channel that exhibits the pharmacological profile of TRPV4.
GSK-responding RGCs constituted a subset of the total hypotonicity-sensitive cell cohort, possibly indicating the expression of other volume increase-sensitive channels within RGCs. We therefore tested for functional expression of TRPV1, another vanilloid TRP channel that has been associated with pressure-sensitive [Ca2+]i responses in RGCs (Sappington et al., 2009). Under our experimental conditions, the TRPV1 agonist capsaicin (5–100 μm) did not induce a rise in [Ca2+]i in acutely isolated mouse RGCs (N > 110). Moreover, the selective TRPV1 antagonist capsazepine (5 μm) had no effect on the amplitude of hypotonic stretch-evoked increases (487 ± 59 nm in controls; N = 48 vs 429 ± 48 in the presence of capsazepine; N = 29; p = 0.39, Mann–Whitney test) (Fig. 6C) or the percentage of cells responding to hypotonic stimulation (N = 2; 62.0 ± 14.9%; p = 0.62, unpaired t test). These results argue against a significant role for TRPV1 signals in stretch-evoked Ca2+ signaling in mouse RGCs.
TRPV4 agonists modulate RGC firing
Given that TRPV4 activation elevates [Ca2+] in RGCs, we asked whether the depolarizing cation influx through the TRPV4 channel modulates RGC excitability. Intact retinas were isolated from the eye and positioned on MEAs, which allow recording of spontaneous spikes from many RGCs simultaneously (Rentería et al., 2006). Perfusion with the TRPV4 agonists 4α-PDD (5 μm) or GSK (100 or 300 nm) evoked a transient increase in the frequency of spontaneous firing in a subset of recorded RGCs. The spike rates of three RGCs, in which typical increases in spiking after GSK application occurred, are shown in Figure 7A. For all recorded cells, the average firing rate during the first 3 min after drug application was compared with the preapplication frequency (4α-PDD: N = 160 RGCs from three retinas; GSK: N = 115 RGCs from 4 retinas). Figure 7, B and C, shows the changes in cell firing evoked by each TRPV4 agonist as a percentage of the preapplication firing frequency. Each agonist evoked increased spike rates of twofold or more in a large percentage of the recorded RGCs (Fig. 7B,C, arrows) (p < 0.00001 for 4α-PDD; p < 0.02 for GSK; Wilcoxon signed-rank test). The increase in spike rate during drug application was transient, exhibiting significant adaptation in the continued presence of drug (Fig. 7A). Displaced amacrine cells either do not spike (Zhou and Fain, 1996) or express few action potentials in response to depolarizing current injection (Ozaita et al., 2004). This suggests that the majority of cells responding to TRPV4 agonists were RGCs, although some nonresponders (5–10%/retina) may have been displaced amacrine cells. These experiments indicate that TRPV4 activation transiently increased RGC excitability and augmented RGC output from the retina.
Sustained activation of TRPV4 channels is cytotoxic for RGCs
We next considered whether extended stimulation with a TRPV4 agonist impacts the survival of acutely isolated RGCs. As shown previously (Otori et al., 1998; Hartwick et al., 2008), 1 h incubation with the AMPA/KA receptor agonist kainate (10 μm) increased the number of TUNEL-positive cells by 68 ± 7% (N = 305; p < 0.0001, Mann–Whitney) (Fig. 8B,D), whereas RGCs exposed only to control saline supplemented with L15 showed little TUNEL signal (Fig. 8A,D). Exposure to GSK1016790A induced significant cell death, which was confined to cells with large somata (>6 μm, arrowheads). The 25 nm GSK induced TUNEL-positive signals in 33 ± 7% (N = 1523; p = 0.0002), and 100 nm GSK induced TUNEL labeling in 67 ± 13% of presumed RGCs (N = 917; p < 0.0005). GSK had relatively little effect on signals in presumed photoreceptor perikarya (cell diameters, 3–5 μm; arrows).
This study provides anatomical, molecular, and physiological evidence that the polymodal pressure-sensitive TRPV4 ion channel participates in the transduction of hypotonic stimuli and contributes to Ca2+-dependent intracellular signaling and membrane excitability of mammalian RGCs. First, we have shown that the wild-type mouse retina expresses TRPV4 mRNA and protein. Second, we show that the TRPV4 protein is localized primarily in RGC somata and axons. Third, using Ca2+ imaging and MEA recording, we directly demonstrate that TRPV4 channels contribute to transmembrane Ca2+ flux and spike firing rates of RGCs. Finally, we show that pharmacological characteristics of hypotonicity-induced [Ca2+]i increases in RGCs match the known pharmacology of TRPV4 and that sustained activation of this subtype of TRPV channels initiates cell death pathways in RGCs.
In the mouse retina, almost every cell identified with RGC-specific markers (Thy1:CFP; Thy1:GFP; Brn3a; SMI-32) was immunopositive for TRPV4, indicating that all identified RGC perikarya express TRPV4. TRPV4-IR was particularly prominent within RGC axons in the nerve fiber layer and the optic nerve head, both of which represent initial targets for deleterious effects of increased IOP. TRPV4 staining was excluded from amacrine, astrocytic, and microglial processes in the retina but was apparent in Müller cell processes. Our demonstration of the localization of the channels to RGCs suggests that retinal output neurons are capable of transducing mechanical, thermal, and/or osmotic stimuli.
Functional role for TRPV4 in the retina
Cell stretch induced by hypotonic swelling caused a marked increase in [Ca2+]i of RGCs. Because the resting membrane potential of RGCs is within the operational range for inactivating and persistent Na+ conductances (∼ −60 mV) (Kim and Rieke, 2003; Hayashida and Ishida, 2004; Margolis et al., 2010), we postulate that sustained depolarization through stretch-sensitive cation-permeable channels will contribute to the excitability and dynamic range of RGC responses (Shibasaki et al., 2007) by countering, in a significant subset of RGCs, the tonic inhibition mediated by GABAergic and glycinergic amacrines. Consistent with this hypothesis, TRPV4 activators 4α-PDD and GSK1016790A elevated [Ca2+]i and caused a >100-fold increase in the frequency of spontaneous RGC spiking. While TRPV4 activation may also have influenced the increase in spiking through reciprocal neuroglial interactions (Newman, 2001), the comparable time courses of agonist-evoked [Ca2+]i signals and spike firing argues for a predominant action on RGCs themselves. This finding may be relevant to entoptic physiology as it implicates TRPV4 channels in the perception of “pressure phosphenes” believed to originate in RGCs (Grüsser et al., 1989).
The effects of pharmacological and hypotonic stimulation observed in RGCs are consistent with properties of heterologously expressed and endogenous TRPV4 channels (Liedtke et al., 2000; Strotmann et al., 2000; Becker et al., 2005; Hartmannsgruber et al., 2007). Plasma membrane ion channels activated by an increase in cell volume were reported to be identical to TRPV4 agonist-sensitive channels with respect to Ca2+ permeability, rectification behavior, and permeability sequence of monovalent cations (Watanabe et al., 2002; Phan et al., 2009). While all Brn3a-IR and Thy1:CFP-positive cells labeled with TRPV4 antibodies, only a subset of dissociated putative RGCs was activated by hypotonic stimulation (∼72%) or TRPV4 agonists (∼16.7–95.8%). We attribute the variability in stretch- and agonist-induced responses to the mechanical trauma associated with the dissociation protocol, which may have compromised a necessary step in TRPV4 activation. It is possible that, as suggested for other tissues (Raoux et al., 2007; Sharif-Naeini et al., 2008; Alessandri-Haber et al., 2009), RGC plasma membranes have other volume increase-sensitive channels such as TRPC1, TRPV1, TRPV2, TREK-1, 2, or TRAAK (TWIK-related arachidonic acid-stimulated potassium channel) (Maingret et al., 1999; Reyes et al., 2000; Križaj, 2005; Leonelli et al., 2009). In particular, TRPV1 was recently implicated in mediating [Ca2+]i increases and pressure-induced apoptotic cell death in cultured rat RGCs (Sappington et al., 2009). While no isoforms of TRPV1 are known to be affected by hypotonic stimulation (Liedtke et al., 2000; Strotmann et al., 2000; Loukin et al., 2009) and stimulation with specific TRPV1 agonists/antagonists failed to induce changes in [Ca2+]i in acutely dissociated mouse RGCs under isotonic or hypotonic conditions, the difference between our study and that of Sappington et al. (2009) can be attributed in part to the dramatic upregulation of Trpv1 gene expression in cultured RGCs. We propose that TRPV4 is a necessary sensing/transducer protein that participates in the RGC response to volume increase, whereas TRPV1 may contribute to RGC excitability as well, once upregulated under certain culture conditions.
Consistent with a role in sensory transduction, TRPV4 is expressed in sensory neurons of dorsal root ganglia, in the trigeminal ganglion, and in hair cells (Liedtke et al., 2000; Alessandri-Haber et al., 2009). A major TRPV4 role in healthy RGCs may be to function as an osmoreceptor (Liedtke et al., 2003; Liedtke and Kim, 2005). The RGC cytosolic volume should be susceptible to local changes in tonicity that occur during the light response or retinal waves (Huang and Karwoski, 1992; Dmitriev et al., 1999). Swelling-induced increases in [Ca2+]i, but not swelling itself, were suppressed by Gd3+ and Ruthenium Red, two compounds that inhibit osmosensory transduction (Liedtke et al., 2003; Bourque, 2008). Activation of endogenous TRPV4 channels has been implicated in other mechanotransduction events including shear stress in vascular endothelia (Hartmannsgruber et al., 2007), stretch-induced integrin signaling (Thodeti et al., 2009), shear stress-mediated relaxation of endothelial cells (Mendoza et al., 2010), and viscous load-coupled ciliary activity in epithelial cells (Andrade et al., 2005).
Implications for glaucoma
TRPV4 localization to the NFL, ONH and the proximal optic nerve together with proapoptotic effects of sustained TRPV4 channel stimulation appear to implicate this channel in the initiation and progression of glaucomatous remodeling. The pressure-induced activation range of TRPV4 (10–30 mmHg) (Loukin et al., 2010b) matches the sustained IOP elevations in chronic glaucoma that can span a range of 10 to several tens of mmHg over the control value of 10–15 mmHg (Bonomi et al., 2001; Quigley, 2005; Whitmore et al., 2005). Increases in hydrostatic or intraocular pressure are correlated with RGC death in cell cultures (Tezel and Wax, 2000; Agar et al., 2006), isolated retinas (Resta et al., 2007), the acute rat glaucoma model (Morrison et al., 1997), the chronic DBA/2J mouse model (John et al., 1998; Libby et al., 2005b), and human glaucoma patients (Bonomi et al., 2001; Gordon et al., 2002), whereas lowering of the IOP slows the progression of axonal loss at all stages of glaucomatous degeneration (Quigley, 2005). Rapid and severe IOP rises (such as occur in acute angle closure glaucoma) (Saw et al., 2003) may cause RGC loss within hours (Naskar et al., 2002) by compromising RGC function at their axons (Quigley, 1983) and their cell bodies (Libby et al., 2005a; Agar et al., 2006; Liu et al., 2007).
Our calcium imaging and immunolocalization experiments suggest that RGC perikarya and axons could be targeted by changes in mechanical and osmotic pressure. At the level of the perikaryon, [Ca2+]RGC increases induced by membrane stretch and TRPV4 agonists had maximal amplitudes comparable to levels evoked by intense stimulation of ionotropic glutamate receptors. Consistent with the hypothesis that RGCs represent the first responder to an acute increase in IOP is the finding that the cornea-positive arm of the scotopic threshold response, the component of the visual ERG response that is the most sensitive to increased IOP, has a significant contribution from RGCs (Bui and Fortune, 2004; Kong et al., 2009). While retinal astrocytes and microglia were not immunolabeled by the TRPV4 antibody, it remains to be determined whether retinal glial TRPV4 expression is affected by sustained IOP increases, a possibility based on TRPV4 expression in cortical astrocytes (Benfenati et al., 2007) and the sensitivity of astrocyte Ca2+ homeostasis to increases in hydrostatic pressure (Mandal et al., 2010).
While the Trpv4 gene plays a critical function in regulation of systemic tonicity in mammals (Liedtke et al., 2000; Bourque, 2008; McHugh et al., 2010), inappropriate activation of TRPV4 in rodents and canines produces an acute circulatory collapse associated with edema, pulmonary hypertension, endothelial injury, ischemia, and/or cell death (Willette et al., 2008). We report that sustained exposure to TRPV4 agonists compromises the viability of mouse RGCs by triggering the apoptotic process, consistent with the observation that even low levels of elevated [Ca2+]i are toxic for RGCs if sustained over an extended period of time (Hartwick et al., 2008). RGCs express the PAR-2 receptor that has been implicated in sensitization of TRPV4 to mechanical stimuli (Luo et al., 2005; Grant et al., 2007). Moreover, gain-of-function TRPV4 mutations cause a range of cellular problems that include axonal neuropathy and suppression of growth (Camacho et al., 2010; Loukin et al., 2010b; Zimoń et al., 2010). Thus, antagonizing excessive TRPV4 activation may be protective against apoptosis in RGCs stressed by sustained mechanical and/or osmotic stimulation.
Together, our data indicate the presence, in mammalian RGCs, of a novel background cation-permeable channel that confers mechanical/pressure sensitivity to cells that convey light-evoked signals to visual centers in the brain. The TRPV4 mechanism in RGCs represents a prime molecular target for severe blinding diseases such as diabetic retinopathy and glaucoma.
The work was supported by the National Institutes of Health (Grants T32DC008553, RO1EY13870, and P30EY014800), the International Retina Research Foundation, the Richard H. Chartrand Foundation, The Foundation Fighting Blindness, the Neuroscience Program at the University of Utah, and the Moran TIGER award. The research was also supported by unrestricted grants from Research to Prevent Blindness to the Moran Eye Institute at the University of Utah. We thank Dr. Ning Tian (University of Utah) for the gift of Thy1:CFP mice and Thy1:GFP sections; and Dr. Tünde Molnar and Ms. Carolyn Groves for help with data analysis.
- Correspondence should be addressed to David Križaj, Department of Ophthalmology & Visual Sciences, John A. Moran Eye Center, University of Utah School of Medicine, Salt Lake City, UT 84132.