Abstract
Glutamate released onto retinal ON bipolar neurons binds to a metabotropic receptor to activate a heterotrimeric G-protein (Go) that ultimately closes a nonspecific cation channel. Signaling requires the α subunit (Gαo), but its effector is unknown. Because Gαo is transcribed into two splice variants (αo1 and αo2) that differ in the key GTPase domain, the next step in elucidating this pathway was to determine which splice variant carries the signal. Here we show by reverse transcription-PCR and Western blots that retina expresses both splice variants. Furthermore, in situ hybridization and immunostaining on mouse retina deficient in one splice variant or the other show that both αo1 and αo2 are expressed by ON bipolar cells but that αo1 is much more abundant. Finally, electroretinography performed on mice deficient for one splice variant or the other shows that the positive b-wave (response of ON bipolar cells to rod and cone input) requires αo1 but not αo2. Thus, the light response of the ON bipolar cell is probably carried by its strongly expressed splice variant, Gαo1.
- G-protein
- Go splice variants
- splice variant knock-out mouse
- mGluR6
- metabotropic glutamate receptor
- retina
- ERG
At the first visual synapse, glutamate modulates three-quarters of the postsynaptic neurons (ON bipolar cells) via a metabotropic glutamate receptor (mGluR6) (Nomura et al., 1994; Vardi and Morigiwa, 1997; Vardi et al., 1998, 2000) that couples to a heterotrimeric G-protein (Go) (Weng et al., 1997; Nawy, 1999; Dhingra et al., 2000). Signaling requires the α subunit (Gαo), but, the next step in this pathway, which ultimately closes a nonspecific cation channel, is unknown (Nawy and Jahr, 1990; Shiells and Falk, 1990; de la Villa et al., 1995; Euler et al., 1996). To identify an effector for Gαo seems important even beyond the visual system because, although effectors are known for the β/γ subunits of this most abundant brain G-protein (Dolphin, 1998), no effectors have yet been identified for the α subunit.
Gαo is transcribed into two splice variants, αo1 and αo2, which contain identical helical domains near the N terminus (NT) but different GTPase domains near the C terminus (CT) (Goldsmith et al., 1988; Hsu et al., 1990; Strathmann et al., 1990; Tsukamoto et al., 1991; Horn and Latchman, 1993). Because the GTPase domain binds both receptor and effector, it is crucial for coupling (for review, seeNeer, 1994; Clapham, 1996; Gudermann et al., 1997). Therefore, the next step in elucidating this pathway was to determine which splice variant carries the signal. Previous studies that colocalized Gαo with mGluR6 in ON bipolar dendrites could not distinguish between different splice variants because the antibody (Ab) used recognizes both (Vardi, 1998), nor did the αo null mouse, which lacks the ON response, resolve this question because it is missing both splice variants (Jiang et al., 1998; Dhingra et al., 2000). Here, using mice deficient in one splice variant or the other, we show that both are expressed by ON bipolar dendrites but that only Gαo1 matters for the light response.
MATERIALS AND METHODS
Animals were deeply anesthetized by intraperitoneal injection (for rat, 45 mg/kg pentobarbital; for mouse, a mixture of 85 μg/gm ketamine and 13 μg/gm xylazine), and the eyes were enucleated; animals were then killed by anesthetic overdose (three times the initial doses). Animals were treated in compliance with federal regulations and University of Pennsylvania policy. The eye was incised at the ora serata and fixed by immersion in buffered 3 or 4% paraformaldehyde with (for immunocytochemistry) or without (forin situ hybridization) 0.01% glutaraldehyde for 1 hr. It was then rinsed in buffer, soaked overnight in 30% buffered sucrose, and embedded in a mixture of two parts 20% sucrose in phosphate buffer and one part of tissue freezing medium. Radial cryosections were 10–15 μm thick.
Reverse transcription-PCR. Retina was homogenized in solution containing 4 m guanidium thiocyanate, 25 mm sodium citrate, 0.5% sarcosyl, and 0.1m 2-mercaptoethanol, and total RNA was isolated by the acid-guanidium phenol–chloroform method (Chomczynski and Sacchi, 1987). The reverse transcription (RT) reaction was performed at 42°C for 50 min with 1–5 μg of total RNA in 20 μl of buffer containing 50 mm Tris-HCl, pH 7.4, 60 mm KCl, 10 mmMgCl2, 1 mm DTT, 1 U/ml RNase inhibitor, 0.5 mm of each dNTP, 500 pmol of random hexamer (or 100 pmol of oligo-dT), and 200 U of SuperScript II Moloney murine leukemia virus reverse transcriptase (Invitrogen, Gaithersburg, MD). PCR reaction was performed in a buffer containing 10 mm Tris, pH 8.3, 50 mm KCl, 2.5 mmMgCl2, 0.4 mm dNTP, 0.2 μm 5′ and 3′ primers, 2 μl of reverse-transcribed cDNA, and 2.5 U of AmpliTaq (PerkinElmer Life Sciences, 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 (PerkinElmer Life Sciences). The sequences of PCR primers (synthesized by Invitrogen) were the same as those used to generate the long probe for in situ hybridization (see below).
In situ hybridization. αo1 and αo2 sequences were amplified by RT-PCR from whole retina. The sequences of PCR primers were as follows: αo1 upstream, 5′-catcctccgaaccagggtc-3′; αo1 downstream, 5′-caagccacagccccggag-3′; αo2 upstream, 5′-catcctccgaaccagggtc-3′; and αo2 downstream, 5′-ggcgatgatgacgtccgt-3′. We also designed a set of probes at the most diverse region of αo (which gave a shorter reaction product): αo1 upstream, 5′-gctcttcgactccatctgt-3′; αo1 downstream, same as for the first set; αo2 upstream, 5′-gacagcatctgcaacaac-3′; and αo2 downstream, same as for the first set. PCR products were subcloned into PCRII vector (Invitrogen) or pBluescript (Stratagene, La Jolla, CA), and the authenticity of the products was verified by direct sequencing.33P-labeled riboprobe was made by in vitro transcription. Briefly, the reaction was performed by incubating the linearized plasmid DNA in 20 μl of solution containing 40 U of RNA polymerase, 10 mm DTT, 20 U of RNase inhibitor, 0.5 mm ATP, GTP, and CTP, 250 μm UTP, and 25 μCi of [α-33P]UTP. After incubating at 37°C for 2 hr, the reaction was treated with RNase-free DNase I and precipitated by LiCl. RNA was further purified by phenol–chloroform extraction, precipitated by ethanol, and finally dissolved in DEPC water.
To test whether the antisense probes cross-hybridized, we applied the sense transcripts of αo1 and αo2 to a nitrocellulose membrane and then tried to hybridize them with the antisense probes. Each antisense hybridized to its own sense but not to the other; therefore, the in situ hybridization probably reflects genuine distribution of αo1 and αo2 mRNAs.
Retina sections were hybridized overnight (∼18 hr) with33P-labeled probes in in situhybridization buffer containing 10% dextran sulfate, 50% formamide, 4× SSC, 0.1% SDS, and 2× Denhardt's solution at 106 cpm/50 μl. After overnight hybridization at 58 or 65°C, slides were washed twice at 10°C above the hybridization temperature in 2× SSC–50% formamide, once at room temperature in 2× SSC, followed by one wash at 10°C above the hybridization temperature in 1× SSC–0.1% SDS. Slides were rinsed in 0.1× SSC, incubated with 300 mm ammonium acetate, dehydrated in ethanol, and finally dipped in photographic emulsion (Kodak NTB-2; Eastman Kodak, Rochester, NY) and exposed for 1–5 weeks.
Immunocytochemistry. Staining was performed according to a standard protocol: soak in diluent containing 10% normal goat serum, 5% sucrose, and 0.3% Triton X-100 in phosphate buffer; incubate in primary antibodies overnight at 4°C; wash and incubate (3 hr) in anti-rabbit F(ab)2 fragment conjugated to a fluorescent marker; and rinse and mount in Vectashield (Vector Laboratories, Burlingame, CA). To stain a 4% paraformaldehyde-fixed retina for protein kinase C (PKC), we needed to retrieve the antigenicity with 0.5% sodium borohydrate (8 min) and use both Triton X-100 (0.75%) and Tween 20 (0.2%) as detergents. For double staining, incubation was done as for single labeling, with both primary antibodies and both secondary antibodies simultaneously. Some sections were incubated in horseradish peroxidase-conjugated secondary antibody and visualized with 3,3′-diaminobenzidine (DAB) reaction product. For electron microscopy, DAB reaction product was intensified by the gold-substituted silver-intensified peroxidase method. The tissue was then osmicated (1.5% osmium tetroxide, 60 min), stained with 1% uranyl acetate in 70% methanol (60 min), dehydrated in 80, 90, and 100% methanol, cleared in propylene oxide, and embedded in Epon 812. Ultrathin sections were mounted on Formvar-coated slot grids and stained with uranyl acetate.
Antibodies. The antibody against αo1(Ab 1718) was raised in rabbit against the peptide EYPGSNTYED, and the antibody against αo2 (Ab 1715) was prepared against peptide EYTGPSAFTE (both are gifts from Dr. D. Manning, University of Pennsylvania, Philadelphia, PA). In addition, three antibodies against αo that recognize both splice variants on Western blots were used. (1) A polyclonal antibody (Ab 9072) was raised in rabbit against the peptide ANNLRGCGLY located at the C terminus (gift from Dr. D. Manning). This sequence is identical to that of αo1 peptide but is different in one amino acid from the corresponding peptide of αo2 (in which the bolded asparagine is replaced by lysine). (2) A polyclonal antibody was raised in rabbit against a peptide DGISAAKDV located at the N terminus (gift from Kwen-Jen Chang, The Burroughs Wellcome Co., Research Triangle Park, NC) (Chang et al., 1988; Codina et al., 1991). This peptide is identical in both splice variants. (3) A monoclonal antibody (mAb 3073) raised in mouse against the purified bovine protein (Chemicon; Li et al., 1995). The antibody to PKCα was monoclonal raised in mouse (mAb 5; Amersham Biosciences, Little Chalfont, UK).
Selective disruption of Gαo1 and Gαo2 expression. Standard molecular biology techniques, to be reported in detail elsewhere, were used to construct targeting vectors of two types (Rudolph et al., 1993, 1994;Jiang et al., 2002). For disruption of Gαo1, a replacement-type targeting vector was made. It contained a genomic segment of the Gαo gene with exons 6, 7.2, 8.2, 7.1, 8.1, and 9.1, of which exon 7.1 was disrupted by insertion of a neomycin selection cassette, followed by an internal ribosome entry site, and the open reading frame of β-galactosidase. Gαo1-targeted embryonic stem (ES) cells were obtained by selection for G418 resistant clones. For disruption of Gαo2, an insertion-type vector was made with the same genomic segment, but, instead of disrupting exon 7.1, the codon for Cys-255 of exon 7.2 was replaced by a stop codon. In addition, a double neomycin–thymidine kinase selection cassette was placed at the end of the genomic segment to allow for positive and negative selection strategies.
Gαo2-targeted ES cells were generated by the “hit-and-run” procedure in which insertion of the mutated homology and selection cassettes is selected for in the presence of G418, and subsequent excision of the wild-type duplicate with attending loss of both selection cassettes was obtained by negative selection in the presence of FIAU (2-fluoro,2-deoxy-5-iodouracyl-β-d-arabinofuranoside). After germline transmission, the resulting F2 mice (50:50 129Sv/C57BJ/6) either lacked Gαo2 but preserved expression of Gαo1 or lacked Gαo1 but preserved expression of Gαo2 and, in addition, expressing β-galactosidase (M. Jiang and L. Birnbaumer, unpublished observations) (vide infra for differential Go-protein expression data).
Electroretinographic recordings. The experimental apparatus, methods of light stimulation and quantification, electroretinographic (ERG) recording, and cone signal isolation have been described in detail previously (Lyubarsky et al., 1999, 2002). Briefly, a mouse was dark adapted for 2 hr, and then, under dim red light, it was deeply anesthetized by intraperitoneally injecting ketamine (20 μg/gm) plus xylazine (8 μg/gm). The animal was immobilized in a holder, the pupils were dilated with 1% tropicamide, and the eyes were protected with a drop of methylcellulose. A platinum recording electrode contacted both corneas, and a tungsten reference electrode was inserted subcutaneously on the forehead. The animal in its holder was then placed inside a light-proof Faraday cage, and light stimuli were delivered through several ports. Stimulus intensity and spectral composition were controlled with neutral density and bandpass interference filters. Light intensities were calibrated and converted to estimated numbers of photoisomerization per photoreceptor (R*) as described previously (Lyubarsky et al., 1999,2000).
RESULTS
Both splice variants of Gαo (αo1 and αo2) are expressed in retina
To determine which transcripts are expressed, we first performed RT-PCR on rat whole retina. Both transcripts were amplified: the bands were at the expected sizes (530 for αo1 and 510 for αo2), and the PCR products were cut at the expected positions by the appropriate restriction enzyme (ClaI applied to αo1 andHindIII applied to αo2) (Fig.1A,B). In several experiments in which the PCR products were sequenced, there was good agreement with published sequences for hamster αo1 or αo2sequences.
Retina expresses both splice variants of αo. A, Primers for amplifying αo1 and αo2 mRNA; restriction enzymes cutting sites are also indicated. B, mRNA from rat retina was reverse transcribed and amplified using the specific primers shown in A. In lanes 2 and4, the PCR products were cut with restriction enzymes.C, Dot blots show that αo1 antibody does not recognize the αo2 peptide (αo2pep) and that αo2 antibody does not recognize the αo1peptide (αo1 pep). Neither antibody reacted with bovine serum albumin (BSA). D, Western blots of whole rat retina show, for αo1, a single prominent band at ∼43 kDa and, for αo2, a prominent band at ∼40 kDa plus a weak band at ∼50 kDa.
To determine whether both proteins are expressed, we used antibodies specific for each splice variant. The specificity of the antibody was tested by dot blots: the antibody for αo1reacted with the αo1 peptide but not with the αo2 peptide, and the antibody for αo2 reacted with the αo2 peptide but not with the αo1 peptide (Fig. 1C). By SDS-PAGE of retinal homogenates followed by Western blotting, the αo1 antibody detected a prominent band at ∼43 kDa, and the antibody for αo2 detected a prominent band at ∼40 kDa; both agree with their known molecular weights of 40 and 39 kDa (Fig. 1D). The antibody for αo2 detected an additional band at ∼50 kDa. We conclude that the retina expresses RNA transcripts and proteins of both splice variants of αo.
αo1 is expressed by rod bipolar cells
First, we tested the distribution of αo1mRNA by in situ hybridization. Antisense riboprobe for αo1 transcript applied to sections of fixed rat retina showed strong labeling in the inner nuclear layer (INL) and ganglion cell layer. Labeling was strongest in the upper tiers of the INL, the location of rod bipolar cell somas. Labeling in the outer nuclear layer (ONL) and in photoreceptor inner segments was weak and was not significantly different from background obtained with sense probes for αo1 (Fig.2A,B). A shorter, more specific probe gave less signal but the same labeling pattern. Also, the same probes applied to mouse retina gave weaker but similar labeling.
Bipolar cells strongly stain for the αo1 splice variant. A, B,In situ hybridization (rat). Antisense probe gave labeling in the INL and in the ganglion cell layer (GCL). Strongest labeling is in the upper tier of the INL, the location of rod bipolar somas. Sense probe gave low background. Left panels, Differential interference contrast shows the retinal layers of the section in A.PR, Photoreceptors. C–E, Immunostaining with the αo1-specific antibody visualized with 3,3′-diaminobenzidine (rat). C, Staining is strong in bipolar dendrites (Den) in OPL and weaker in bipolar somas (BP) in the INL and in the IPL. D, Staining with the preimmune serum was negative. E, Staining with the antibody preabsorbed with the αo1peptide was also negative. F–H, Immunostaining with the αo1-specific antibody visualized with FITC (mouse).F, In the OPL and INL, staining in wild-type mouse resembles that in rat. In the IPL, note the thin bands of higher intensity (black arrows), a thicker band of lower intensity (brackets), and a band devoid of staining (white arrow). G, H, No staining above general background was observed when the same antibody was applied, respectively, to an αo1 or αo1+2 null mouse. ko, Knock-out;wt, wild type.
Next, we tested the distribution of αo1protein. An antibody specific for this splice variant, applied to both rat and mouse retinas, showed staining similar to that obtained previously for αo (Vardi et al., 1993; Vardi, 1998). Stained somas were numerous in the upper tiers of the INL, with dendrites reaching high in the outer plexiform layer (OPL) (Fig. 2C,F). Electron microscopy showed the stain in dendrites that invaginate the rod terminal (data not shown). Thus, rod bipolar somas and dendrites stain for αo1. The bipolar axons crossing the INL were stained for αo1, but, as they cross the inner plexiform layer (IPL), they were unstained. The IPL was also stained but more weakly than OPL. By confocal microscopy, the IPL showed two intense strata at ∼25 and 55% of IPL depth, a lack of stain between ∼25 and 35%, and weak stain elsewhere (0% is the interface between INL and IPL). Photoreceptors were unstained. Staining pattern for αo1 was specific because (1) retinal sections from rat incubated with preimmune serum or with antibody preabsorbed with the αo1 peptide were negative (Fig.2D,E), and (2) retinal sections from an αo1 null mouse and an αo null mouse were negative (Fig.2G,H).
αo1 is expressed by all types of ON cone bipolar cell
Because rod bipolar cells and ON cone bipolar cells differ in response kinetics, components of their transduction cascades might also differ (Berntson and Taylor, 2000; Wu et al., 2000). Therefore, we asked whether rod and cone bipolar cells express the same splice variant of Go. Double immunostaining showed that all ON bipolar cells express αo1. First, rod bipolar cells identified with antibody for PKC stained for αo1, but additional somas positive for αo1 were negative for PKC; these are cone bipolar cells (Fig. 3, left). Similarly, rod bipolar dendritic terminals projecting high in the OPL stained for both PKC and αo1, but processes at midlevel in the OPL, in which cone bipolar dendrites terminate, stained only for αo1 (Fig. 3, left). Second, we labeled all ON bipolar cells using a monoclonal antibody for Go. This antibody is suitable because it stains exactly the same set of cells as the antibody against the C terminus, which was shown to label all ON bipolar cells (Vardi, 1998). In this double staining, all somas stained by the monoclonal antibody also stained for αo1 (Fig. 3, right). Interestingly, stain for αo and αo1 completely colocalized, even in the inner plexiform layer, suggesting that in retina all the αo-positive cells express at least the αo1 splice variant.
All ON cone bipolar cells stain for αo1 (rat). Left, All rod bipolar somas (stained for PKC) also stained for αo1 (double arrowhead), but some somas stained only for αo1(long arrow). In the OPL (bracketed), rod bipolar dendrites stained for both PKC and αo1(arrowhead), but some dendrites stained only for αo1 (short arrow). These probably belong to ON cone bipolar cells. Right, All ON bipolar somas were identified by staining with the monoclonal antibody for αo. All of these somas also stained for αo1. Also, in the IPL, the two staining patterns were the same, indicating that the retina expresses primarily the αo1 splice variant.
ON bipolar cells and certain amacrine cells express low levels of αo2
A specific antisense probe for αo2 mRNA distributed similarly to that of αo1: strong labeling in the INL, somewhat weaker in the ganglion cell layer, and none in the outer nuclear layer and photoreceptor inner segments (Fig.4A). A shorter, more specific probe distributed similarly, but less intensely, whereas the sense probe distributed randomly and rather weakly (Fig.4B).
Bipolar cells and stratum 1 of the IPL weakly express the αo2 splice variant. A,B, In situ hybridization (rat). Antisense probes specific for αo2 message hybridized strongly in the INL and weakly in the ganglion cell layer (GCL). Sense probes gave weak, diffuse background. Left, Differential interference contrast for retinal section inA. C–F, Immunostaining of mouse retina, visualized with FITC. D–F were captured using the same confocal laser intensity and gain parameters. C, Antibody against the CT peptide of αo1 applied to the αo1 null retina faintly stained the OPL, bipolar somas (BP), and stratum 1 of the IPL (brackets). D, Antibody against the NT peptide applied to the αo1 null retina gave similar staining: faint in the OPL, bipolar somas, and puncta throughout the IPL, plus more intense staining in stratum 1. E, NT antibody applied to αo knock-out (KO) retina gave weak diffuse background staining. F, NT antibody applied to αo2 knock-out retina gave strong labeling, resembling that obtained with this and other anti-αo antibodies in wild-type retina.
To test the distribution pattern of the αo2protein was difficult. The antibody for αo2capriciously stained cone outer segments in rat (but not in mouse), and it also stained Muller cells. The staining of the Muller cell was clearly unspecific because this cell also stained with the preimmune serum applied to wild type and with the antibody applied to αo or αo2 null mice. Thus, the antibody for αo2, under our staining conditions, was unsuitable for immunocytochemistry. Instead, we used three antibodies shown by Western blots to recognize both αo1 and αo2 and applied them to retina of the αo1 null mouse. The monoclonal antibody gave no staining. However, the other two antibodies that recognize both splice variants, one against the CT and one against the NT, both gave faint staining in the OPL, bipolar somas, and stratum 1 of the IPL (Fig. 4C,D). To distinguish this stain from background, we first imaged staining from the αo null retina. This general background served as a reference at the confocal microscope (Fig. 4E). Then, using the same laser intensity and gain parameters, we imaged staining from the αo1 null retina. For both CT and NT antibodies, staining was stronger than in αo null retina, and the pattern was repeatable (Fig. 4, compare E with C, D). Both antibodies applied to the αo2 null retina gave identical staining as in the wild type (Fig. 4F). Thus, both by in situ hybridization and by immunocytochemistry, bipolar cells weakly express αo2.
Next, to determine whether the stained bipolar cells represent only a subset of bipolar cells, we stained the αo1null retina for PKC and anti-αo (C terminus). All cells that stained for PKC also stained for αo, but ∼8% of the cells (13 of 75 in one animal; 1 of 100 in the other) that stained for αo were unstained for PKC (Fig.5). The difference between the two animals could be attributed to weaker expression in cone bipolar cells or to regional differences. The cells that are αo2 positive (but PKC negative) were not identified. We can rule out the possibility that they are OFF bipolar cells because it has been established (using the same antibody as in this experiment) that αo is absent from OFF cone bipolar cells (Vardi, 1998). Thus, αo2 is expressed in rod bipolar cells and in at least some ON cone bipolar cells.
Rod bipolar and some cone bipolar cells stain for Gαo2. Rod bipolar cells were identified by staining for PKC; staining for αo2 was achieved by staining an αo1 null mouse with anti-Gαo-CT. All bipolar somas stained for PKC also stained for αo2(double arrowhead), but some somas were stained only for αo2 (arrow).
The light response requires αo1 but not αo2
To test which splice variant is crucial for the ON bipolar cell light response, we examined the electroretinogram in mice lacking either αo1 or αo2. Three different lighting conditions were used: scotopic (dark adapted plus dim stimulus), saturated (dark adapted plus saturating stimulus), and photopic (light adapted plus strong stimulus). Under all conditions, mice lacking αo1 also lacked the b-wave, which is generated by the ON bipolar cells (Robson and Frishman, 1995) (Fig. 6, middle column; two animals, four eyes). This lack of b-wave does not reflect lack of photoreceptor activity because the a-wave, which originates in suppression of the photoreceptor dark current (Hagins et al., 1970; Hood and Birch, 1993; Breton et al., 1994; Lyubarsky and Pugh, 1996; Pugh et al., 1998), was present. Under the “saturated” condition, αo1 null mice gave a reduced, but still profound, a-wave (78 ± 34 μV). Under photopic conditions, in which the negative wave is likely a combination of a-wave and OFF bipolar response, the negative wave was as large or greater than in the wild type. This response pattern resembled that obtained when both splice variants were eliminated (αo null) (Dhingra et al., 2000). Light responses in mice lacking αo2 were indistinguishable from wild type (Fig.6, compare left with right columns; three animals, six eyes). The average peaks for Gαo2null mouse versus wild type were as follows (respectively, in microvolts): rod-generated b-wave, 249 ± 89 versus 220 ± 72; saturated a-wave, 283 ± 51 versus 319 ± 97; and cone-generated b-wave, 146 ± 49 versus 109 ± 32.
Rod- and cone-driven b-waves are present in the electroretinogram of Gαo2 null mouse but absent from the Gαo1 null mouse. A, Animals dark adapted for 2 hr were stimulated with dim flashes that elicited a rod-driven, corneal-positive b-wave in the wild-type mouse and in Gαo2 null but not in the Gαo1 null. Flash intensities estimated as photoisomerizations (R*) per rod (Φ), and number of responses (n) averaged for each trace were as follows: Φ = 9; n = 18. B, Dark-adapted animals were stimulated with an intense flash isomerizing ∼0.1% of the rhodopsin. This elicited in wild-type and Gαo2 null mice a negative a-wave (shading), followed by a positive b-wave. The Gαo1 null mice showed an a-wave (although somewhat reduced) but no b-wave. Φ = 105;n = 4. C, Mice were adapted to a background (white light, 9100 R* rod−1 sec−1) that completely suppressed the rod cGMP-activated current. Rods were then stimulated with an intense white flash that isomerized ∼0.4% of the M-cone pigment and 0.03% of the UV-cone pigment. All of the animals showed a cone-driven a-wave. A typical cone-driven b-wave (positive-going response with superimposed oscillations, peaking ∼70–90 msec after the flash) was present in wild-type and αo2 null mice but was absent in the Gαo1 null animals. For all of the records, n = 16. KO, Knock-out;WT, wild type.
DISCUSSION
We showed by several techniques (RT-PCR, Western blotting,in situ hybridization, and immunostaining with antibodies directed against different domains) that both splice variants of Gαo, Gαo1 and Gαo2, are expressed in retinal ON bipolar cells. However, Gαo1 is much more abundant and is essential for the bipolar cell light response. Mice lacking the Gαo1 splice variant are devoid of the scotopic and photopic b-waves, whereas mice lacking the Gαo2 splice variant show a normal ERG. We noticed that the rod a-wave in the αo1 null mouse was reduced relative to the wild type. The difference is not clear and seems insignificant because in the complete Gαo null, both a-waves were within normal range. Nevertheless, the lower amplitude does not compromise the conclusion that the response of the bipolar cell is absent because this range of photoresponses should have elicited a b-wave ON response. Thus, the ON responses in rod and all ON cone bipolar cells depend only on Go1.
Importance of identifying the crucial splice variant in the mGluR6 cascade
The identification of αo1 as the critical splice variant permitted its use (in constitutively active form) as bait in a yeast two-hybrid assay. One strong interactor encodes the γ subunit of the photoreceptor phosphodiesterase (PDE6). However, available probes for PDE-γ did not localize it to ON bipolar cells, and therefore it is unlikely to be the effector in the ON response (Nawy, 1999). Another interactor encodes Ret-RGS1 (regulator of G-protein signaling), and this protein is expressed in dendritic tips of at least one type of ON cone bipolar cell (Dhingra et al., 2001). Ret-RGS is unlikely to be the actual effector for Gαo1, although it may assist. Nevertheless, these preliminary observations suggest the value of having narrowed down the search for effectors for the specific splice variant of αo.
Possible function of Go2 in ON bipolar cells
If only Go1 is crucial for the ON response, what is the function of Go2? We propose several possibilities. (1) Gαo2 may contribute to fine-tuning of the mGluR6 cascade in certain ON bipolar cells. The ON bipolar cells are classified into rod bipolar cells and approximately five different types of cone bipolar cells that differ by their morphology (Cohen and Sterling, 1990; Boycott and Wässle, 1991;Euler and Wässle, 1995), and they are thought to divide the range of temporal frequencies by responding with different kinetics (Sterling, 1998; Freed, 2000a,b; Masland, 2001; Roska and Werblin, 2001). All ON bipolar cells express the same receptor (mGluR6) (Masu et al., 1995; Vardi and Morigiwa, 1997; Vardi et al., 2000), and the same G-protein α subunit (αo1) and γ subunit (Gγ13) (Huang et al., 2000). Because Gαo2 is expressed only in a subtype of the ON bipolar cells, it may differentially contribute to shaping response kinetics. (2) Gαo2 could couple different receptors to different biochemical cascades. Gαo1 and Gαo2 differ in the C-terminal domain that is generally involved in specific interaction with the receptor and the effector (Conklin et al., 1996). An example for effector specificity is given by Helisoma neurons, in which αo2, but not αo1, inhibits Ca2+ channels (Man-Son-Hing et al., 1992). An example for receptor specificity is given by rat pituitary GH3 cells, in which Go1 inhibits the Ca2+ channel by coupling muscarinic receptor, whereas Go2 inhibits this channel by coupling somatostatin receptors (Kleuss et al., 1991; Chen and Clarke, 1996; Degtiar et al., 1997). In rat retina, rod bipolar cells and a subset of cone bipolar cells express, in addition to mGluR6, the metabotropic glutamate receptors mGluR1 and mGluR5 (Koulen et al., 1997). Therefore, it is possible that Go2 couples these receptors to a second-messenger cascade. Because these receptors typically regulate intracellular [Ca2+] (via PLC pathway) (Abe et al., 1992) and because intracellular [Ca2+] is critical for adaptation in rod bipolar cells (Shiells and Falk, 1999; Nawy, 2000; Berntson and Taylor, 2000), Go2 may contribute to adaptation. (3) Go2 may be involved in some aspects of development. Expression patterns of Go1 and Go2 during development suggest that it mediates several processes in development (Strittmatter et al., 1990; Rouot et al., 1992; Duc-Goiran et al., 1999). Thus, similar function may apply also in retina. It is important, however, to note that whatever the function in development of Go is, it is not crucial for gross morphology because all of the αo knock-outs (Gαo−/−, Gαo1−/−, and Gαo2−/−) have normal brain and retinal morphology (Valenzuela et al., 1997; Jiang et al., 1998; Dhingra et al., 2000; this study).
Footnotes
↵* A.D., M.J., and T.-L.W. contributed equally to this work.
This work was supported by National Eye Institute Grants EY11105 (N.V.), EY00828 (P.S.), and EY02660 (E.N.P.), the Research to Prevent Blindness Foundation (E.N.P.), and National Institutes of Health Grant DK19318 (L.B.). We thank Yi-Jun Shi and Sally Shrom for excellent technical assistance. We also thank David Manning for donating the antibodies for Gαo (C terminus), Gαo1, and Gαo2 and Kwen-Jen Chang for donating the antibody for Gαo (N terminus).
Correspondence should be addressed to Dr. Anuradha Dhingra, 123, Anat-Chem Building, Department of Neuroscience, University of Pennsylvania, Philadelphia, PA 19104. E-mail:annu{at}retina.anatomy.upenn.edu.
T.-L. Wang's present address: Howard Hughes Medical Institute and Johns Hopkins Oncology Center, The Johns Hopkins University, Baltimore, MD 21231.
