Calcium enters the outer segment of a vertebrate photoreceptor through a cGMP-gated channel and is extruded via a Na/Ca, K exchanger. We have identified another element in mammalian cones that might help to control cytoplasmic calcium. Reverse transcription-PCR performed on isolated photoreceptors identified mRNA for the SII− splice variant of the type I receptor for inositol 1,4,5-triphosphate (IP3), and Western blots showed that the protein also is expressed in outer segments. Immunocytochemistry showed type I IP3 receptor to be abundant in red-sensitive and green-sensitive cones of the trichromatic monkey retina, but it was negative or weakly expressed in blue-sensitive cones and rods. Similarly, the green-sensitive cones expressed the receptor in dichromatic retina (cat, rabbit, and rat), but the blue-sensitive cones did not. Immunostain was localized to disk and plasma membranes on the cytoplasmic face. To restore sensitivity after a light flash, cytoplasmic cGMP must rise to its basal level, and this requires cytoplasmic calcium to fall. Cessation of calcium release via the IP3 receptor might accelerate this fall and thus explain why the cone recovers much faster than the rod. Furthermore, because its own activity of the IP3 receptor depends partly on cytoplasmic calcium, the receptor might control the set point of cytoplasmic calcium and thus affect cone sensitivity.
Sensitivity of vertebrate photoreceptors is regulated by cytoplasmic Ca2+([Ca2+]i) (Lamb and Torre, 1990; Koch, 1995; McNaughton, 1995; Koutalos and Yau, 1996) (for review, see Yau, 1994). Ca2+ enters the outer segment via cGMP-gated channels, is buffered by calcium binding proteins, and exits via the Na/Ca, K exchanger (Korenbrot, 1995) (for review, see Schnetkamp, 1995a). The outer segment contains an additional store of Ca2+ within membrane saccules (disks) that resemble smooth endoplasmic reticulum (Liebman, 1974; Fain and Schroder, 1985; Nicol et al., 1987; Schnetkamp and Bownds, 1987). Because the latter releases stored Ca2+ via an inositol 1,4,5-triphosphate (IP3) receptor or a ryanodine receptor (Berridge, 1993; Mikoshiba et al., 1994), so might the disks bear such a receptor and provide an additional source of cytoplasmic Ca2+.
Indeed, a light flash to bovine outer segment membranes releases IP3 (Ghalayini and Anderson, 1984; Hayashi and Amakawa, 1985; Brown et al., 1987), and an IP3 receptor has been identified biochemically (Day et al., 1993). Although an antibody against purified brain IP3 receptor did not bind to outer segments (Peng et al., 1991), the IP3 receptor is encoded by at least four genes, each of which might be spliced into several isoforms (Danoff et al., 1991; Nakagawa et al., 1991; Lin, 1995;Nucifora et al., 1995). Therefore, a negative immunoreaction could result simply from a mismatch between isoform and antibody. Here we show by RT-PCR, Western blot, and immunocytochemistry that photoreceptor outer segments express the SII−splice variant of the type I IP3 receptor on their plasma and disk membranes. The receptor is more abundant in red- and green-sensitive cones than in blue-sensitive cones and rods.
MATERIALS AND METHODS
Eyes were enucleated under deep anesthesia from adult rat (Sprague Dawley), rabbit (Dutch-Belted), guinea pig (Dunkin Hartley), cat, and monkey (Macaca mulatta). All procedures complied with federal regulations and University of Pennsylvania policies.
Small pieces of rat retina were incubated in oxygenated Hank’s medium (Life Technologies, Gaithersburg, MD) containing 14.4 U/ml papain, 0.1 gm/ml cysteine, and 0.5 mm EDTA for 10 min at 28°C. After the retina was rinsed with papain-free Hank’s medium containing 0.5% bovine serum albumin, the retina was triturated gently with a wide-bore Pasteur pipette. An aliquot of dissociated cell suspension was diluted with Hank’s medium and dropped on a cover glass coated with concanavalin A (Sasaki and Kaneko, 1996). After 30 min, most cells attached to the coated cover glass, which then was washed with Hank’s medium at least five times to remove loose cells. Isolated photoreceptors were identified by their characteristic morphology (see Fig. 1 C) and sucked into a patch pipette.
Reverse transcription-PCR (RT-PCR)
Total RNA from both whole retina and photoreceptors was isolated by acid guanidium and phenol-chloroform extraction (Chomzynski and Sacchi, 1987). The reverse transcription (RT) reaction was performed at 42°C for 50 min with 1–5 μg of total RNA in a 20 μl buffer containing (in mm) 50 Tris-HCl, pH 7.4, 60 KCl, 10 MgCl2, 1 DTT, and 0.5 of each dNTP plus 1 U/ml RNase inhibitor, 500 pmol random hexamer or 100 pmol of oligo dT, and 200 U of Super II M-MLV reverse transcriptase (Life Technologies). PCR reaction was performed in a buffer containing (in mm) 10 Tris, pH 8.3, 50 KCl, 2.5 MgCl2, and 0.4 dNTP plus 0.2 μm 5′ and 3′ primers, 2 μl of reverse-transcribed cDNA, and 2.5 U of AmpliTaq (Perkin-Elmer, Branchburg, NJ). Thirty cycles (94°C for 1 min, 52°C for 1 min, and 72°C for 2 min) were performed on a programmable thermocycler (Perkin-Elmer). The sequences of PCR primers (synthesized by Life Technologies) designed to amplify the SII region of type I IP3 receptor included the upstream primer 5′GAGCTGTCTGTGCTCGTG3′ and downstream primer 5′GTCGATGACCAGATTGGAG3′.
Isolating outer segment proteins for Western blot
Outer segment proteins were isolated by a protocol described previously (Panico et al., 1990). Briefly, retina was vortexed in 51% sucrose in MOPS buffer [(in mm) 20 MOPS, 2 MgCl2, 100 KCl, 0.1 EDTA, 1 DTT, and 0.1 PMSF plus 0.7 μg/ml aprotinin, 0.7 μg/ml leupeptin, 0.7 μg/ml pepstatin A, and 0.7 μg/ml benzamidine], layered with more MOPS buffer, and spun for 30 min at 27,000 × g. Outer segments floating at the interface were collected, diluted with MOPS buffer, and spun again. The pellet was resuspended in 38% sucrose in MOPS buffer and passed three times through an 18 gauge needle. The preparation was layered again with MOPS buffer and spun. The orange material at the interface (now mostly outer segments) then was diluted with MOPS buffer, spun again, and saved for analysis.
Protein samples (5–10 μg/μl) were dissolved in SDS loading buffer and separated by 8% SDS-PAGE (Laemmli, 1970). Proteins were transferred to a nitrocellulose membrane, incubated with primary antibody against type I IP3 receptor (1:500 to 1:1000 dilution) for 2 hr at room temperature, washed, incubated 2 hr with 1:2000 dilution of horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch, West Grove, PA), washed, and detected by chemiluminescence (Amersham, Arlington Heights, IL).
Posterior eyecups from cat, monkey, rat, and rabbit were fixed with 4% paraformaldehyde and 0.01% glutaraldehyde in phosphate buffer (PB), pH 7.3, at room temperature for 1 hr and cryoprotected overnight in PB containing 30% sucrose.
Light microscopy. Retina was embedded in Tissue Freezing Medium (Triangle Biomedical Sciences, Durham, NC) and sectioned radially at 10 μm in the cryostat. Sections were preincubated in PB containing 10% normal goat serum and 0.3% Triton X-100 for 1 hr and then in the same solution containing primary antibody (diluted 1:200 to 1:1000) overnight at 4°C. After being rinsed, the sections were incubated in goat anti-rabbit F(ab′)2 conjugated to a fluorescent dye for 2 hr at room temperature, mounted in Vectashield mounting medium (Vector Laboratories, Burlingame, CA), and coverslipped.
Double labeling. For type I IP3 receptor and blue-sensitive opsin (both antibodies raised in rabbit) the sections were incubated in antibody against type I IP3 receptor, rinsed and incubated in goat anti-rabbit F(ab′)2 conjugated to HRP, developed with 3,3′-diaminobenzidine (DAB) and 0.1% hydrogen peroxide, treated with glycine buffer, pH 2.2, for 5 min to elute antibodies (DAB reaction product remains), incubated in antiserum against blue opsin, and incubated with goat anti-rabbit F(ab′)2 conjugated to Cy3. For type I IP3receptor and red- and green-sensitive opsin (both antibodies raised in rabbit) the sections were incubated sequentially in antibody against type I IP3 receptor, an excess of goat anti-rabbit Fab′ fragments conjugated to FITC (to cover all rabbit epitopes), antiserum against red/green opsin, and goat anti-rabbit F(ab′)2conjugated to rhodamine. Control experiments were similar except that the antiserum against red/green opsin was omitted. Under regular fluorescent intensity the immunoreactivity of type I IP3receptor was detected only with the FITC filter set, indicating that the second secondary antibody did not react with the first primary antibody.
Immunoelectron microscopy. Radial vibratome sections (50–100 μm) were immunostained as described above except that Triton X-100 was omitted. Sections were incubated with HRP-conjugated secondary antibody, developed with DAB and hydrogen peroxide, and intensified by gold-substitution silver-intensification (Johnson and Vardi, 1998). Sections were osmicated with osmium tetroxide (2%; 1 hr), stained with uranyl acetate (1%, 1 hr), dehydrated in ethanol (70–100%), cleared in propylene oxide, and embedded in Epon 812. Ultrathin sections (70–90 nm) were stained with uranyl acetate and lead citrate and viewed with a transmission electron microscope (JEOL 1200EX).
We used three different polyclonal antibodies against type I IP3 receptor (all raised in rabbit). The first was raised against the C-terminal peptide (amino acid 2731–2749; from Dr. S. K. Joseph, Thomas Jefferson University, Philadelphia, PA). The specificity of this antibody was established (Mignery et al., 1989;Joseph and Samanta, 1993; Joseph et al., 1995). The second (M) was raised against a fusion protein containing amino acids 4466–5723 of type I IP3 receptor (Lin, 1995). The third (3′α) was raised against a fusion protein containing amino acids 7761–8027 (Lin, 1995). M and 3′α antibodies were obtained from Dr. William Agnew, Johns Hopkins University (Baltimore, MD). We also used a monoclonal antibody to type III IP3 receptor (Transduction Laboratories, Lexington, KY). Rabbit polyclonal antibodies against blue and red/green opsins were obtained from Dr. Jeremy Nathans, Johns Hopkins University (Baltimore, MD).
Photoreceptors express a splice variant of the type I IP3 receptor
To determine whether the transcript of the type I IP3receptor is expressed in photoreceptors, we performed RT-PCR. PCR primers were designed to flank the SII splicing region (Fig.1 A). In whole retina, RT-PCR amplified two major DNA products with distinct molecular sizes: 545 and 429 bp (Fig. 1 B). Direct DNA sequencing from these two bands showed that the larger product contained the exons A, B, and C in the SII region (SII+), whereas the smaller one lacked any of these exons (SII−). In isolated photoreceptors (Fig. 1 C), RT-PCR amplified only the smaller splice variant (Fig. 1 B) with a DNA sequence identical to the known sequence of the splice variant lacking the A, B, and C exons (SII−). This experiment was repeated three times with identical results.
To see if type I IP3 receptor is translated in photoreceptors, we prepared Western blots from whole retina (rat and cat) and from outer segments (cat) and probed them with an antibody against the C terminus. A prominent band at ∼220 kDa, corresponding to the approximate molecular weight of type I IP3 receptor, was detected in all blots (Fig. 1 D). Sometimes an additional band at ∼130 kDa was labeled also; this is probably a degradation product (Joseph et al., 1995), but it could be a cross-reaction with a different protein. In the outer segments the major band of the expected molecular weight was prominent, and the smaller degradation product was always negligible. This suggests that the SII+ splice variant of type I IP3receptor is more likely to degrade.
Type I IP3 receptor is expressed strongly in cone outer segments
In all species tested (monkey, cat, rat, guinea pig, and rabbit), stain for type I IP3 receptor (against the C terminus) was strong in cone outer segments but very weak in the outer plexiform layer, the inner part of the inner nuclear layer, and the ganglion cell layer (Fig. 2 A). No stain was observed in the inner plexiform layer. Rod outer segments were slightly positive (especially in rat), but this was evident only in semithin and ultrathin sections (see below).
To test whether staining was specific for type I IP3receptor, we applied two additional antibodies prepared against different domains of the rat receptor. Both antibodies gave distinct staining of cone outer segments, but the background was high. Control sections incubated with the preimmune serum were negative (Fig.2 B, shown only for antibody “M”). A monoclonal antibody against type III IP3 receptor was negative for all retinal cells (tested in rat; data not shown).
Type I IP3 receptor is not detected in S cones
We noticed in monkey retina that the antibody to type I IP3 receptor failed to stain some cone outer segments (Fig.3 A, arrows). Because the blue cone comprises only 5–10% of all cones, we surmised that it was unstained. To test this, we sequentially stained sections from monkey retina for type I IP3 receptor and blue-sensitive opsin. All cone outer segments negative for type I IP3 receptor were strongly positive for blue-sensitive opsin (Fig. 3 A,B, three experiments). We also double-labeled retina for the IP3 receptor and red- and green-sensitive opsin. All labeled cones (i.e., red and green cones) were positive for type I IP3 receptor (Fig. 3 C,D, two experiments). The blue cone in cat, rat, and rabbit (Fig.3 E–G) was also negative for type I IP3receptor.
Ultrastructural localization of type I IP3 receptor
By light microscopy, stain for type I IP3 receptor was strong in cone outer segments but was barely detectable in rod outer segments. To determine whether this difference merely reflected the greater thickness of the cone or whether it represented a denser expression of receptor, we examined semithin sections cut parallel to the long axis of photoreceptor outer segments. Even in these ∼0.5-μm-thick sections, one-half the thickness of a rod outer segment, cones stained much more strongly than rods (Fig.4 A). Furthermore, at the electron microscope level, gold particles representing immunostaining were denser in cone outer segments than in rods (Fig.5 A,B). Stronger cone staining might occur if the disks communicate with the extracellular space, as they do in amphibians (Laties and Liebman, 1970), for this might render them more accessible to the antibody. However, disks in mammalian cones commonly are closed, as in rod (Cohen, 1970; Anderson and Fisher, 1976;Rodieck, 1988). Therefore, in cone and rod, access of antibody to the IP3 receptor may be similar. When equal access was assured by treating the tissue with a high concentration of detergent (0.5%), cone staining compared with rods was even more pronounced. Therefore, greater cone staining probably reflects their stronger expression of type I IP3 receptor.
Most immunostain was localized to the cytoplasmic face of the disk and plasma membrane (Fig.6 A). This was easiest to see where disks were swollen, because the disk lumen then could be distinguished clearly from the cytoplasmic space (Fig.6 B). Staining of the cytoplasmic face matches the topology of the receptor, because its C terminus (target of the antibody) is thought to be cytoplasmic (Mikoshiba et al., 1994). In rods the sparse staining might be nonspecific; however, because the gold particles were located only on the cytoplasmic face of the membrane, they probably represent genuine, although weak, expression of IP3 receptor.
Thin sections also revealed stain over the rod cilium that connects the inner and outer segment (see Fig. 4 A). Within the cilium, stain was concentrated along the microtubules (see Fig.4 B). Staining also was observed in the cone cilium (see Fig. 4 A), but microtubules were not discerned easily.
We provide strong evidence that the type I IP3receptor is expressed in photoreceptor outer segments, especially in red- and green-sensitive cones: (1) mRNA of a particular splice variant (SII−) was amplified from isolated photoreceptors; (2) a single protein band with the expected molecular weight was demonstrated by Western blot of the outer segments; (3) strong staining for the receptor was detected with three different antibodies, whereas controls (preimmune serum) were negative; (4) antibody against the C terminus was localized to the cytoplasmic face of the disk and the plasma membrane, in accordance with the known receptor topology (Mikoshiba et al., 1994).
The finding of IP3 receptor on the outer face of the disk supports previous findings that IP3 can release Ca2+ from internal stores (Parker et al., 1986;Schnetkamp and Szerencsei, 1993; Schnetkamp, 1995b). Because IP3 usually is associated with smooth endoplasmic reticulum, localization of the type I IP3 receptor on theplasma membrane may seem surprising. However, IP3 receptor also localizes to plasma membrane in olfactory cilia, mast cells, and T-lymphocytes (Kuno and Gardner, 1987; Penner et al., 1988; Cunningham et al., 1993). This site can admit Ca2+ from the extracellular space, where the concentration (∼3 mm) is apparently the same as in the disk lumen (for review, see Schnetkamp, 1989). In mammals, because disk surface area is greater than plasma membrane surface area, the disks probably provide most of the IP3-mediated Ca2+ influx.
Possible function of type I IP3 receptor in red- and green-sensitive cones
Ca2+ plays a key role in terminating the light response and adaptation (Fig. 7). When light via the rhodopsin cascade reduces cGMP, Ca2+influx through the cGMP-gated channel decreases and, as extrusion continues, cytoplasmic Ca2+ declines. Lower Ca2+ (1) activates rhodopsin kinase, (2) inhibits phosphodiesterase, (3) activates guanylyl cyclase, and (4) increases the affinity of the cGMP-gated channel for its ligand. All of these effects help to terminate the light response and adapt the photoreceptor (i.e., reduce gain and restore sensitivity to a stronger light). A quantitative model of the rod response does not require an additional mechanism to modulate Ca2+ (Lamb and Pugh, 1992; Lyubarsky and Pugh, 1996; Nikonov et al., 1998). However, the cone recovers faster than the rod and is less sensitive. Conceivably, the abundant IP3 receptor on cone disk membranes might contribute to these response properties as we now explain.
The IP3 receptor might provide a positive feedback loop (Fig. 7). This would accelerate the fall of cytoplasmic Ca2+ after a light flash or its rise after a dark flash. Phospholipase C (PLC), the enzyme that produces IP3, is present in cones [Ferreira and Pak (1994), but see Peng et al. (1997) and discussion below] where it might be stimulated constitutively by cGMP and Ca2+(Ghalayini and Anderson, 1987; Rhee and Bae, 1997; Haque et al., 1998). Therefore, light ONset, by reducing cGMP and Ca2+, would inhibit PLC. This would reduce IP3 and thus the release of intradisk Ca2+. The IP3ligand binding of the receptor is increased by Ca2+up to ∼200 nm but is reduced above this level (Bezprozvanny et al., 1991; Li et al., 1995; Patel and Taylor, 1995;López-Colomè and Lee, 1996; Kaznacheyeva et al., 1998). In darkness, cytoplasmic Ca2+ is near this optimum for IP3 binding (Korenbrot, 1995), so the fall in Ca2+ after light stimulation would reduce IP3 binding and accelerate the fall in Ca2+. At light OFFset, Ca2+influx via the cGMP-gated channel rises. This would activate IP3 binding and accelerate the rise of cytoplasmic Ca2+. This loop for accelerating the rise of Ca2+ would cease as Ca2+ rises beyond the optimal concentration for IP3 binding.
Two points might seem inconsistent with the model. First, although biochemistry suggests an IP3 signaling system in purified photoreceptor outer segments, physiology finds no such effect on the light response of intact cells. However, most physiology has focused on rods in which the pathway is minor. Second, although light on isolated disk membranes increases IP3, our model shows light decreasing IP3. However, both PLC and the IP3receptor depend critically on the Ca2+ level, so the decisive test requires an intact cell.
It remains unclear which isoform of PLC is expressed by cones. Ferreira and Pak (1994) identified PLC-β4, but Peng et al. (1997) did not concur. The issue matters because PLC-β is activated by a member of the Gq family, whereas other PLC isoforms are activated differently, for example, by a different G-protein (Gh), a tyrosine kinase, or a lipid-derived second messenger (for review, see Rhee and Bae, 1997).
If the IP3 receptor accelerates the cone response, why is it absent from the blue-sensitive cone? Possibly the blue-sensitive cone expresses a different isoform; alternatively, the blue-sensitive cone and the rod both express the IP3 receptor at very low levels. Vision mediated by the blue cone does share several features with rod vision. For example, both have a longer integration time and a higher sensitivity than vision mediated by red- and green-sensitive cones (Brindley et al., 1966; Mollon and Polden, 1977a,b; Zrenner and Gouras, 1979, 1981; Williams et al., 1981; Nelson, 1985). Conceivably, the extra loop for rapidly driving cytoplasmic Ca2+through larger excursions is reduced or absent because it would ill serve a slower, more sensitive response.
This work was supported by National Institutes of Health Grants EY11105-0 and EY08124. We thank Drs. Paul Liebman, Edward Pugh, Robert Smith, William Agnew, and Suresh K. Joseph for insightful discussions. We also thank Yi-Jun Shi and Tina Geueke for excellent technical assistance and John Demb and Madeleine Johnson for reading this manuscript. Antibodies are kind gifts from Drs. Suresh K. Joseph, William Agnew, Dan Wu, and Jeremy Nathans.
Correspondence should be addressed to Dr. Noga Vardi, Department of Neuroscience, University of Pennsylvania, 122B Anatomy/Chemistry Building, Philadelphia, PA 19104.
Dr. Wang’s present address: Department of Pathology, Johns Hopkins University, School of Medicine, Baltimore, MD 21205.