Transmission of visual signals at the first retinal synapse is associated with changes in calcium concentration in photoreceptors and bipolar cells. We investigated how loss of plasma membrane Ca2+ ATPase isoform 2 (PMCA2), the calcium transporter isoform with the highest affinity for Ca2+/calmodulin, affects transmission of rod- and cone-mediated responses. PMCA2 expression in the neuroblast layer was observed soon after birth; in the adult, PMCA2 was expressed in inner segments and synaptic terminals of rod photoreceptors, in rod bipolar cells, and in most inner retinal neurons but was absent from cones. To determine the role of PMCA2 in retinal signaling, we compared morphology and light responses of retinas from control mice and deafwaddler dfw2J mice, which lack functional PMCA2 protein. The cytoarchitecture of retinas from control and dfw2J mice was indistinguishable at the light microscope level. Suction electrode recordings revealed no difference in the sensitivity or amplitude of outer segment light responses of control and dfw2J rods. However, rod-mediated ERG b-wave responses in dfw2J mice were ∼45% smaller and significantly slower than those of control mice. Furthermore, recordings from individual rod bipolar cells showed that the sensitivity of transmission at the rod output synapse was reduced by ∼50%. No changes in the amplitude or timing of cone-mediated ERG responses were observed. These results suggest that PMCA2-mediated Ca2+ extrusion modulates the amplitude and timing of the high-sensitivity rod pathway to a much greater extent than that of the cone pathway.
The time course and amplitude of changes in presynaptic and postsynaptic [Ca2+]i are key determinants of the kinetics of neuronal signaling (Barrett and Stevens, 1972; Felmy et al., 2003). For short, action potential-driven stimuli, the time course of the intracellular calcium concentration [Ca2+]i generally reflects the kinetics of Ca2+ influx via voltage-activated Ca2+ channels and intracellular buffering (Meinrenken et al., 2003). However, after high-frequency stimulation or during sustained Ca2+ influx at tonic synapses, release is governed by residual [Ca2+]i and its time course is determined by the properties of presynaptic Ca2+ extrusion (Tank et al., 1995; Zhong et al., 2001).
Plasma membrane Ca2+ extrusion is controlled by two transporter families: Na+/Ca2+ exchangers (NCXs) and plasma membrane Ca2+ ATPases (PMCAs). In general, PMCAs have ∼10-fold higher affinity for Ca2+ than NCXs and therefore regulate baseline [Ca2+]i and [Ca2+]i kinetics during transient depolarizations (Blaustein et al., 1991; Wanaverbecq et al., 2003). The PMCA family consists of four isoforms, which are expressed in a cell type- and tissue-specific manner throughout the brain (Stahl et al., 1992; Stauffer et al., 1995). With exception of cochlear and vestibular structures, which are highly dependent on PMCA isoform 2 (PMCA2) (Street et al., 1998; Dodson and Charalabapoulou, 2001), the significance of these isoforms for neuronal function is mostly unknown.
PMCA2 has ∼10 times higher affinity for Ca2+/calmodulin than the other three PMCA isoforms (apparent Km of 2–4 nm) (Hilfiker et al., 1994) and is the fastest Ca2+ pump activated in response to an increase in [Ca2+]i (Caride et al., 2001; Brini et al., 2003). PMCA2 expression is especially prominent in the brain and in sensory tissues (Dumont et al., 2001; Križaj et al., 2002; Burette et al., 2003; Silverstein and Tempel, 2006), indicating that this isoform could play a significant role in sensory transmission. Defects in the ATP2B2 gene encoding PMCA isoform 2 produce severe neurological phenotypes (Kozel et al., 1998; Street et al., 1998; Takahashi and Kitamura, 1999; Shull, 2000). In the deafwaddler mouse mutant dfw2J, a 2 bp deletion in Atp2b2 produces a frame-shift mutation that causes a reduction in the Atp2b2 transcript and loss of PMCA2 protein (Street et al., 1998; McCullough and Tempel, 2004). Homozygous deafwaddler mice have severe abnormalities in the function of cochlear, vestibular, cerebellar, and sensory brainstem structures resulting in tremor, ataxia, hyperexcitability, and deafness (Noben-Trauth et al., 1997; Street et al., 1998; Konrad-Martin et al., 2001).
Given the prominent PMCA2 expression in retinal neurons (Križaj et al., 2002), its upregulation during retinal development (Rentería et al., 2005) and its key role in development and function of the cochlea and vestibular system (Street et al., 1998), which share a number of signaling elements with the retina (Zenisek et al., 2003; El-Amraoui and Petit, 2005), we hypothesized that PMCA2 modulates the establishment of retinal circuits in the developing retina as well as light responses at mature retinal synapses. To test this hypothesis, we analyzed retinal development and transmission of light-evoked signals in control and dfw2J mice.
Materials and Methods
All studies were conducted in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research and guidelines issued by the Society for Neuroscience. The care and use of animals presented in this study were approved by the Animal Care and Use Committees at the University of Washington and the University of California, San Francisco.
The Atp2b2dfw2J mouse strain (referred to here as dfw2J) arose spontaneously in CBy.A/J-fsn (a substrain of BALB/cByJ) at The Jackson Laboratory (Bar Harbor, ME) (Noben-Trauth et al., 1997). Obtained in 1997 (Street et al., 1998), the dfw2J locus was transferred via serial backcrosses to CBA/CaJ mice acquired directly from The Jackson Laboratory to maintain isogeneity with the commercially available inbred stock. The congenic animals used in this study were from backcross generations N6 (the 18-month-old animals) and N8 (the younger animals). Age-matched littermate controls were used throughout. The L7–green fluorescent protein (GFP)-expressing mice were a gift from Dr. Michisuke Yuzaki (St. Jude Children's Research Hospital, Memphis, TN). All experimental animals used in this study were raised in a 12 h light/dark cycle in the University of Washington animal facility. Mice used for electroretinogram (ERG) recordings were maintained in a 12 h light/dark cycle. Genotyping was performed using allele-specific PCR primers (for details, see http://depts.washington.edu/tempelab/Protocols).
For histological analysis, mice were killed with carbon dioxide inhalation and immediately perfused intracardially with a mixture of aldehydes (2% paraformaldehyde and 2.5% glutaraldehyde). Eyes were removed, bisected along the vertical meridian, postfixed in osmium tetroxide, and embedded in an Epon–Araldite mixture. Sections of the entire retina were cut at 1 μm thickness and stained with toluidine blue as described previously (LaVail and Battelle, 1975). Tissue sections were chosen in which the rod outer segment (OS) and Müller cell processes crossing the inner plexiform layer (IPL) were continuous in the plane of section, or nearly so, to ensure that the sections were not oblique (Duncan et al., 2003).
The eyes were enucleated, corneas were cut with a razor blade, and the eyecups with the retinas were immersion fixed at room temperature for 0.5 h in 4% (w/v) paraformaldehyde in 0.1 m phosphate buffer, pH 7.4. The retinas were rinsed two times in phosphate buffer and cryoprotected in 30% sucrose overnight at 4°C. Pieces of retinas were mounted in OCT, sectioned vertically at 14 μm thickness on a cryostat, collected on Super-Frost Plus slides (Fisher, Pittsburgh, PA), and stored at −20°C until use. Rabbit polyclonal antibodies NR1, NR2, and NR3 against PMCA1, PMCA2, and PMCA3, respectively, were generated against 13–18 residue peptide sequences at the N terminus of the corresponding rat PMCA (Filoteo et al., 1997) or against an N-terminal sequence of the human PMCA1–PMCA4 (Stauffer et al., 1995). The NR1–NR3 antibodies were a kind gift from Drs. Strehler and Penniston (Mayo Clinic, Rochester, MN) or were purchased from Affinity BioReagents (Golden, CO) and Swiss Antibodies (Swant, Bellinzona, Switzerland) and used at 1:200–1:300. The RIBEYE antibody was a gift from Dr. Thomas Südhof (Texas Southwestern Medical Center, Dallas, TX). Postsynaptic density-95 (PSD-95) and sarco(endo)plasmic reticulum Ca2+ ATPase antibodies were obtained from Chemicon (Temecula, CA), glutamine synthetase antibodies were from BD Biosciences (San Jose, CA), and the protein kinase C antibody was from Santa Cruz Biotechnology (Santa Cruz, CA). Synaptic vesicle protein 2 (SV2) antibody was developed by Kathleen Buckley (Harvard Medical School, Cambridge, MA) and obtained from the Developmental Studies Hybridoma Bank (Iowa City, IA). Fluorescein-conjugated peanut agglutinin (PNA) (Invitrogen, Carlsbad, CA) was used at 1:25.
Cytoplasmic and nuclear proteins were prepared using the NE-PER extraction reagents (Pierce, Rockford, IL). The protein concentrations were determined by BCA protein assay kit (Pierce). Fifteen to 20 μg of extract were resolved on NOVEX-NuPAGE 4–12% BT gels (Invitrogen) and transferred to polyvinylidene difluoride membranes (Invitrogen). After 15–30 min incubation with StartingBlock blocking buffer (Pierce), the membranes were incubated in primary antibody at 4°C overnight. After washing, the blots were then incubated with appropriate anti-peroxidase-conjugated secondary antibodies (1:4000; Amersham Biosciences, Arlington Heights, IL) at room temperature for 2–3 h. Finally, the proteins on the membranes were detected using the ECL Plus chemiluminescence system (Amersham Biosciences). The blots were stripped with Restore Western blot stripping buffer (Pierce) at room temperature for 60 min and then reprobed with different antibodies. Finally, the blots were probed with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as an internal loading control. Each experiment was repeated two to three times with independent samples from different animals.
Quantitative PCR methods
Six-week-old dfw2J and control littermates were dark adapted for at least 30 min before the animals were killed by cervical dislocation. The retinas were dissected from the eyes in RNase free PBS and transferred immediately to RNAlater (Qiagen, Valencia, CA) for at least 30 min on ice. Liquid was removed and retinas were stored at −80°C until additional processing. RNA was isolated using the RNeasy Micro kit (Qiagen) per the instructions of the manufacturer, with a single retina constituting a single sample. Carrier tRNA was not added. There was no significant difference between the yield of normal and mutant retinas (control littermates, 1.99 ± 0.73 μg; dfw2J, 1.81 ± 0.75 μg; p = 0.68).
One microgram of total RNA was incubated with 1 μl (50 pm) of Random Hexamers (Applied Biosystems, Foster City, CA) in 9.5 μl of reaction for 10 min at 70°C, followed by immediate transfer to ice. To each reaction was added 4 μl of 5× first-strand buffer (Clontech, Mountain View, CA), 2 μl of DTT (20 mm), 2 μl of dNTP (10 mm each), 0.5 μl of Rnasin (Promega, Madison, WI), and 1 μl of Powerscript Reverse Transcriptase (Clontech). The reaction (20 μl) was incubated for 90 min at 42°C, followed by 10 min at 72°C. cDNAs were then diluted to 200 μl final volume in 10 mm Tris and 0.1 mm EDTA, pH 8.
Quantitative PCR primer design and validation.
For each transcript of interest, quantitative PCR (qPCR) primers were designed using Primer3 software (Rozen and Skaletsky, 2000). Design criteria included the following: amplicons were chosen to be 50–150 bp in length, primers were designed with melting temperature between 60 and 64°C, guanine–cytosine base content was between 35 and 65%, and the complementarity between and within primers was stringently minimized (Table 1). To verify that each primer set produced a single product of the correct size, a pilot qPCR experiment was run, using the same conditions as were used in all subsequent experiments. This consisted of a 25 μl reaction, containing 12.5 μl of 2× iQ SybR Green Supermix (Bio-Rad, Hercules, CA), 300 nm each primer (final concentration), and cDNA template (from adult mouse brain, from equivalent of 5 ng starting total RNA). The cycling parameters (for amplification plus melting curve) were as follows: 95°C for 3 min (activate enzyme), 40 repeats of 95°C at 30 s, 60°C at 30 s (amplification), 95°C at 1 min, 55°C × 1 min (premelt curve), and 90 repeats of 10 s each starting at 55°C and incrementing 0.5°C per step (melt curve). The melt curves for each primer set were inspected to ensure that only a single peak was present. A small amount of the product was also run on a 2.5% agarose gel to verify product size. All primer efficiencies were verified to be 100 ± 5% via standard curve.
Reference gene selection.
To compare expression levels between samples, we normalized gene-of-interest expression levels to the geometric mean of multiple control genes, offering a far more robust normalization than can be achieved using a single reference gene such as α-actin or GAPDH (Vandesompele et al., 2002). Briefly, reference gene selection is performed as follows. For all of the samples to be assayed in a given study, the expression of a series of candidate reference genes (Table 1) is quantified via qPCR. These expression data are entered into the geNORM software (http://medgen.ugent.be/~jvdesomp/genorm/). This Microsoft (Seattle, WA) Excel-based algorithm evaluates all of the candidate genes and ranks them in order of stability across the sample set. Use of the two most stable reference genes was sufficient (using additional genes added little to the overall stability of the normalization scheme).
qPCR quantification and normalization.
qPCR was performed using the reaction conditions described in the validation section above, for both the genes of interest and reference genes. Within a single qPCR experiment, each data point consisted of the average threshold cycle number from two experimental replicates (same biological sample, same reaction mixture). For each data point, normalized PMCA gene expression was calculated by the formula: pmcai/r = 2npi/(2nr1 × 2nr2)1/2, where pmcai is the target gene copy number (for PMCA1, PMCA2, PMCA3, or PMCA4), r is the geometric mean of the two reference gene copy numbers, npi is the threshold cycle number for PMCAi, and nri is the threshold cycle number for reference gene ri. Final reported values were based on three independent qPCR runs from three animals of each genotype (for details, see Results).
Full-field ERGs were recorded from dfw2J and control littermates as described previously (Bok et al., 2002). Briefly, mice were dark adapted overnight and anesthetized with ketamine (87 mg/kg) plus xylazine (13 mg/kg), pupils were dilated with 2.5% phenylephrine and 1% atropine in dim red light, and mice were kept on a warming blanket. Contact lens electrodes for mice (Bayer et al., 1999) were placed on the corneal surfaces bilaterally with 1% methylcellulose, and silver wire reference and ground electrodes were placed subcutaneously in the nose and tail, respectively. Using a UTAS-E 3000 Visual Diagnostic System (LKC Technologies, Gaithersburg, MD) and beginning below ERG threshold, stimuli were presented in order of increasing luminance from −4.6 to +0.4 log cd s m−2, and 3–15 responses at each intensity were computer averaged. If we assume that a Ganzfeld stimulus of 1 scotopic (scot) cd s m−2 produces 500 photoisomerizations per mouse rod (Rh*), between estimates in the literature of 100 (Hetling and Pepperberg, 1999) and 1500 (Pennesi et al., 1998), these stimulus intensities ranged from 0.012 to 5 × 106 Rh*; thus, 100 Rh* was produced by approximately −0.5 log scot cd s m−2. Interstimulus intervals ranged from 5 s at the lowest intensities to 120 s at the highest intensities. The amplitude and time-to-peak of the a- and b-waves were measured. Below a-wave threshold, b-wave amplitudes were measured from baseline to peak; at intensities in which measurable a-waves were present, b-wave amplitudes were measured from the a-wave trough to the b-wave peak.
A subset of dfw2J and control mice were prepared for ERG analysis using infrared goggles in the absence of any dim red light to record the scotopic threshold response (STR) after overnight adaptation and exposed to stimuli presented in order of increasing luminance from −6.6 log cd s m−2. In addition, responses to a single flash of +2.4 log cd s m−2 were recorded from a subset of dfw2J and control mice. Mice were exposed to a background light of 30 cd s m−2 for 10 min before recording photopic responses to stimuli presented at a rate of 2 Hz at 0.4 log cd s m−2. Because photopic a-waves are negligible in mice (Peachey et al., 1993), the amplitude and time-to-peak of the photopic b-waves were measured.
Outer segment currents from single rod photoreceptors were recorded using suction electrodes (Baylor et al., 1979; Field and Rieke, 2002). Mice were dark adapted for 12–15 h, and the retinas were isolated under infrared illumination (>950 nm). Brief flashes were delivered from a light-emitting diode with peak output at 470 nm. Calibrated flash strengths were converted to photoisomerizations (Rh*) assuming a collecting area of 0.5 μm2. Cells were continuously superfused with bicarbonate-buffered Ames' solution (Sigma, St. Louis, MO) warmed to 36–37°C.
Retinal slice preparation and whole-cell recording
Current responses of rod bipolar cells were recorded using whole-cell, voltage-clamp recordings in retinal slices prepared as described previously (Armstrong-Gold and Rieke, 2003). The internal solution contained the following (in mm): 125 K-aspartate, 10 KCl, 10 HEPES, 5 NMG-HEDTA, 0.5 CaCl2, 1 ATP-Mg, and 0.2 GTP-Mg, pH 7.2 (osmolarity was 275–285 mOsm). Slices were continuously superfused with Ames' solution warmed to 36–37°C. Light delivery and calibration were similar to that for the rod recordings. Rod bipolar cells were identified based on a soma location in the outermost part of the inner nuclear layer (INL) and characteristic light responses.
Most ERG responses from dfw2J and control mice were compared using a Student's t test. Values are presented as mean ± SEM unless otherwise noted. p values <0.05 were considered significant.
To test for a role of PMCA2 in retinal development or in the transmission of rod- and cone-mediated signals across the retina, we measured the expression pattern of PMCA2 and compared the histology and light responses of retinas lacking PMCA2 with normal retinas. As described below, absence of PMCA2 did not produce clear histological changes but did alter the transmission of rod-mediated signals.
PMCA2 is abundant in the retina but is not required for normal development
Levels of PMCA2 expression in control and dfw2J retinas
The retinal homogenates from homozygous dfw2J mice and control mice were analyzed by Western blotting with a polyclonal PMCA2-specific antibody and by reverse transcription (RT)-PCR. In the control, the antibody recognized a single band of ∼140 kDa (Strehler and Zacharias, 2001; Križaj et al., 2002) (Fig. 1). The dfw2J mice are functionally null because of a frame shift in the coding sequence (Street et al., 1998; McCullough and Tempel, 2004). Accordingly, little or no PMCA2 protein was detected in the retinal homogenate derived from the dfw2J homozygotes (Fig. 1).
Previous studies suggested that all four PMCA isoforms are expressed in the vertebrate retina (Križaj et al., 2002, 2004; Rentería et al., 2005). Quantitative real-time RT-PCR was performed to assess the abundance of PMCA transcripts in the mouse retina. The cDNA values for all transcripts were normalized to the geometric mean of the expression levels of two reference genes: ACTG (actin) and YWHAZ (tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, ζ polypeptide). As shown in Figure 2, PMCA1 accounted for ∼80% of the PMCA transcript in the mouse retina and PMCA2 for ∼20%; PMCA3 and PMCA4 were expressed at much lower levels.
Loss of PMCA2 could trigger compensatory upregulation of other PMCA isoforms (Guerini et al., 1999; Brini et al., 2003; Wood et al., 2004). To determine whether transcripts encoding PMCA 1, PMCA3, and/or PMCA4 are upregulated in dfw2J retinas, qPCR was performed in parallel on dfw2J retinas. Small, statistically insignificant changes in PMCA1, PMCA3, and PMCA4 expression were observed in dfw2J retinas (Fig. 2). The PMCA2 transcript detected in dfw2J retinas is attributable to incomplete degradation (McCullough and Tempel, 2004).
These results show that PMCA2 is the second most abundant PMCA2 isoform in the mouse retina. The absence of compensatory upregulation of PMCA1, PMCA3, and PMCA4 isoforms suggests that a dfw2J phenotype is likely associated with the loss of PMCA2.
PMCA2 is expressed in inner and outer retina
PMCA2 immunolocalization in control (littermate) mice revealed that PMCA2 is found in both inner and outer retina as well as in the retinal pigment epithelium (Fig. 3). No PMCA2 signal was detected in dfw2J retinas immunostained with PMCA2 antibodies (Fig. 3D). As shown previously (Križaj et al., 2002), PMCA2 was strongly expressed in the inner retina of the mouse. PMCA2 immunoreactivity was detected in the majority of amacrine perikarya in the adult mouse. The immunoreactivity observed in two brightly stained bands corresponds to cell processes of cholinergic amacrine cells (Rentería et al., 2005). At higher confocal gain, PMCA2 immunoreactivity was detected in rod inner segments (ISs) (Fig. 3B, arrowheads), in the outer plexiform layer (OPL) (Fig. 3C, arrows), in the inner nuclear layer, and in the retinal pigment epithelium (Fig. 3A); PMCA2 signal was absent from rod OSs (Fig. 3B,C). The IS labeling was observed with two different anti-PMCA2 antibodies, raised against rat and human epitopes. PMCA2 localization to ISs was not described in our previous report (Križaj et al., 2002) probably because the PMCA2 signal in the IS is much weaker than that in the inner retina and the signal is better seen in retinas fixed for shorter time periods (Rentería et al., 2005).
To determine whether PMCA2 is expressed early in development of the mouse retina, we examined retinal sections from postnatal day 1 (P1) to P35 mice. At P1, PMCA2 was found in the IPL with a moderate expression in the neuroblast layer (NBL) (Fig. 3). No signal was observed in the absence of the primary antibody (Fig. 3A). The IPL signal remained prominent throughout development. At P5, PMCA2 immunoreactivity was detected in the developing OPL in addition to the ganglion cell layer and the NBL. From P7 up to adult, there was a gradual retraction in PMCA2 immunoreactivity from the outer nuclear layer (ONL) and INL into both plexiform layers (Fig. 3A), with a moderate PMCA2 signal visible at higher confocal gains throughout the INL. These results indicate that, in the mouse retina, PMCA2 is already widely expressed at birth in both cell bodies and cell processes; during maturation, PMCA2 expression becomes mainly confined to both synaptic layers.
The presynaptic and postsynaptic localization in the OPL was studied in colocalization experiments. The PDZ (postsynaptic density-95/Discs large/zona occludens-1) domain-containing protein PSD-95 is a marker for photoreceptor terminals in the mouse retina. As illustrated in Figure 4B, PMCA2 overlapped with PSD-95-immunoreactive structures in the OPL, suggesting that it is expressed in synaptic terminals of rods. To determine whether PMCA2 is also localized to cone synapses, retinas were double immunostained with the FITC-conjugated cone marker PNA. PNA labeled cone outer segments as well as a row of OPL puncta corresponding to cone pedicles (Fig. 4A). PMCA2 staining was specifically excluded from PNA-positive structures (Fig. 1A,Aii). The localization of PMCA2 to rod spherules and its absence from pedicles is consistent with a presynaptic role for PMCA2 at rod, but not cone, synapses.
Double staining with the rod bipolar cell marker protein kinase C suggested that PMCA2 is expressed in cell bodies and dendrites of rod bipolar cells (Fig. 4D, arrows). PMCA2 immunoreactivity was localized to the dendritic region of rod bipolar cells and in bipolar perikarya (Fig. 4C,D). Immunostaining of retinas from transgenic mice that express high levels of GFP in rod bipolar cells (Tomomura et al., 2001) additionally suggested postsynaptic localization of PMCA2 to rod bipolar cells (Fig. 4C, arrowheads). A subset of INL neurons that did not express GFP was PMCA2 immunoreactive (Fig. 4C, right panel, arrow), suggesting possible localization to cone bipolar neurons. Finally, PMCA2 antibodies labeled the majority of neurons at the proximal edge of the INL, corresponding to amacrine cells (Fig. 3).
To determine PMCA2 expression in radial glia that envelop the synaptic terminals of rods and cones (Sarantis and Mobbs, 1992) and regulate development (Rich et al., 1995) and visual responses of retinal neurons (Newman, 2004), retinas were immunostained with the Müller cell marker glutamine synthetase. PMCA2 partially colocalized with glutamine synthetase at the level of the outer limiting membrane (Fig. 4E), in the OPL and INL (Figs. 4E,F), suggesting that PMCA-mediated Ca2+ extrusion regulates Müller cell Ca2+ homeostasis. Together, these data suggest that PMCA2 is expressed in the OPL at presynaptic and postsynaptic locations comprising the rod pathway and glial processes but is excluded from cones.
To determine whether the changes in kinetics of visual transmission in dfw2J mice (see below) were associated with altered expression of synaptic markers, we examined the expression of the several presynaptic proteins in dfw2J mice as well as that of PMCA1, the major PMCA isoform expressed in rod and cone terminals (Križaj et al., 2002; Rentería et al., 2005). Confocal analysis of PMCA1, the synaptic ribbon protein RIBEYE, and the synaptic marker SV2 expression found no major differences in the pattern of immunostaining for these proteins (data not shown). Furthermore, the expression of rod bipolar cell marker protein kinase C, horizontal cell marker calbindin, GABAergic amacrine cell marker GAD-65, and the Müller cell marker glutamine synthetase was not qualitatively different from controls (data not shown).
Retinas from dfw2J mice are anatomically normal
Given the prominent PMCA2 expression in the retina (Figs. 1⇑⇑–4) and massive degeneration of cochlear and vestibular structures in PMCA2-deficient mice (Street et al., 1998; Dodson and Charabalapoulou, 2001), it was of interest to determine whether retinal anatomy was compromised in dfw2J mice. No differences were apparent in length or density of rod OS, ONL thickness, ONL cell count, or the combined thickness of the OS and IS layers in retinal sections from P65 dfw2J mice and their littermate controls. No degradation in the inner retina was apparent at the light microscopic level (n = 5) (Fig. 5). No evidence of cellular degeneration was found in any dfw2J mice compared with littermate controls examined at up to 18 months of age (data not shown).
The results summarized in Figures 1⇑⇑⇑–5 indicate that PMCA2 is expressed in several locations in the retina but is not required for normal development and maintenance of retinal layers. We next tested for a role of PMCA2 in the generation and transmission of light responses in the retina.
Loss of PMCA2 alters transmission of rod-mediated signals
We used ERGs and single-cell recordings to evaluate the importance of PMCA2 for the transmission of rod-mediated signals through the retina. As described below, the kinetics and sensitivity of scotopic (i.e., rod-mediated) ERGs were altered in mice lacking PMCA2. Responses of single rod outer segments were not significantly altered, whereas those of rod bipolar cells were. Together, these results indicate a role of PMCA2 in the normal transmission of signals from rod outer segments to rod bipolar cells.
Scotopic ERG responses in dfw2J mice are reduced and delayed
ERGs represent the summed activity of rods and downstream retinal cells and thus provide an effective means to identify deficits in signaling. Figure 6shows representative scotopic ERG recordings for control and dfw2J mice. The STR is a cornea-negative waveform that originates in the inner retina (Sieving and Nino, 1988; Robson and Frishman, 1995; Saszik et al., 2002) and is the highest sensitivity component of the ERG. Compared with control (−5.1 ± 0.2 log cd/s/m−2 (n = 12 eyes, 6 mice), the dfw2J−/− STR threshold was elevated (−4.4 ± 0.04 log cd s m−2 (Fig. 6, arrows) (n = 8 eyes, 4 mice; p = 0.001). Thus, loss of PMCA2 reduced inner retinal sensitivity at rod threshold intensities.
Signaling in the outer retina was assayed from the a-wave, which reflects primarily signals in rod photoreceptors (Granit, 1947; Penn and Hagins, 1969; Lyubarsky and Pugh, 1996), and the b-wave, which at low light levels primarily reflects the integrated responses of rod bipolar cells (Masu et al., 1995; Robson and Frishman, 1995; Green and Kapousta-Bruneau, 1999).
The a-wave was reduced in amplitude and slowed slightly in the absence of PMCA2. In response to flashes producing ∼105 Rh*/rod, the a-wave in dfw2J mice was ∼25% smaller than control (Fig. 7A) (455 ± 15 μV for 22 control eyes vs 336 ± 25 μV for 24 dfw2J eyes; p < 0.0001). These differences were first apparent for intensities producing ∼100 Rh*/rod (Fig. 7C). The time-to-peak for the maximum scotopic a-wave was slightly slowed (Fig. 7B) (6.4 ± 0.1 ms in control mice vs 6.8 ± 0.08 ms in dfw2J mice; p = 0.023). Thus, flashes that saturated individual rods (see below) reduced the amplitude and slowed the kinetics of the summed rod response.
Loss of PMCA2 produced larger changes in the b-wave than a-wave. In response to flashes producing ∼103 Rh*/rod, b-waves from dfw2J mice were reduced by >45% compared with control (Fig. 7D) (818 ± 30 μV in 26 control eyes, 445 ± 32 μV in 26 dfw2J eyes; p < 0.0001). The reduction in b-wave amplitude was observed across a wide range of intensities (Fig. 7C), including intensities in which the rod a-wave responses were little affected. Thus, at −4.6 log cd s m−2 (∼0.01 Rh*/rod), the b-wave amplitude in control animals was 31 ± 2 μV, whereas dfw2J response was 17 ± 2 μV, a 45% reduction (Fig. 7C). The time-to-peak of the maximum scotopic b-wave in dfw2J mutants was also significantly delayed (Fig. 7E) (95 ± 2 ms in control vs 106 ± 2 ms in dfw2J; p < 0.0001). The delay in the dfw2J b-wave was still significant in responses matched for equal a-wave amplitudes (data not shown) and was apparent across a wide range of intensities (Fig. 7F).
At intermediate intensities, the reduction in b-wave amplitude and increased time-to-peak can likely be attributed to reduction in the summed rod bipolar responses, although, at low intensities, the b-wave is contaminated by the STR and, at high intensities, reflects activity of both rod and cone bipolar cells. The larger changes in amplitude and kinetics of the b-wave compared with the a-wave in dfw2J mice suggest that PMCA2 plays a role in regulating the transmission of rod signals to rod bipolar cells. To test this suggestion directly, we recorded responses of individual rod outer segments and rod bipolar cells.
The photosensitivity of dfw2J rods is normal
We used suction electrode recordings to determine whether the effect of PMCA2 on the summed rod responses indicated by the ERG measurements was indicative of the behavior of individual rod outer segments. As described below, these experiments indicated that rod outer segment light responses were near normal in the absence of PMCA2.
Figure 8, A and B, superimposes average responses to flashes producing between 2 and 200 Rh* for a single control (Fig. 8A) and dfw2J (Fig. 8B) rod. The qualitative features of the responses in the two types of rods were similar. Figure 8C plots response amplitude against flash strength for 22 control and 25 dfw2J rods. The half-saturating flash strength was 9.9 ± 0.4 Rh* for control rods and 10.5 ± 0.4 Rh* for dfw2J rods (mean ± SEM). The kinetics of responses to flashes producing 1–5 Rh* were also similar (Fig. 8D). Responses to these flashes reached peak in 206 ± 8 ms in control rods and 218 ± 10 ms in dfw2J rods. Finally, both types of rods had similar dark currents: 14.5 ± 0.5 pA in both control and dfw2J rods. These results indicate that individual dfw2J rod outer segments show normal or near-normal sensitivity to dim light and indicate that the phototransduction process in individual rods is not modulated by PMCA2. This differs from the conclusion reached from the ERGs, which indicated a reduction in the amplitude of the summed rod response; we return to this difference in Discussion.
Rod bipolar cell sensitivity in dfw2J mice is reduced
To test for an effect of PMCA2 on signal transfer downstream from the rod outer segment, we recorded light responses of voltage-clamped rod bipolar cells in retinal slices. The sensitivity of the rod bipolar cells was reduced in dfw2J mice, indicating a role for PMCA2 in transmission of signals to the rod bipolar cell.
Light responses of rod bipolar cells from control and dfw2J mice were qualitatively similar (Fig. 9A,B). Responses of dfw2J rod bipolar cells, however, were ∼50% less sensitive than control rod bipolar cells (Fig. 9C); the half-saturating flash strength was 3.7 ± 0.2 Rh*/rod for dfw2J rod bipolars (mean ± SEM; n = 11) and 2.4 ± 0.1 Rh*/rod for control rod bipolars (n = 61). Half-saturating flash strengths for rod bipolar cells from littermate dfw2J control mice (2.6 ± 0.1 Rh*/rod; n = 18) and C57BL/6 mice (2.4 ± 0.1 Rh*/rod; n = 43) were similar; hence, responses from both genetic backgrounds were combined as controls. The kinetics of responses producing <1 Rh*/rod were similar in control and dfw2J rod bipolar cells (Fig. 9D); the time-to-peak was 157 ± 5 ms in dfw2J rod bipolar cells and 158 ± 4 ms in controls. Saturating flashes in the rod bipolar (producing ∼20 Rh*/rod) had a time-to-peak of 101 ± 3 ms in control and 104 ± 8 ms in mutants (mean ± SEM). The ∼5 ms difference in time-to-peak of the b-wave observed at comparable light levels would not be detected using this technique.
Both ERG and single-cell recordings indicate a greater disruption in signaling downstream of the rods than in the rods themselves. The single-cell recordings indicate that at least part of the differences observed in the ERGs can be attributed to a deficit in the transfer of signals from rod outer segments to rod bipolar cells in the absence of PMCA2.
Loss of PMCA2 does not alter transmission of cone-mediated signals
Results from experiments performed in scotopic conditions indicated that PMCA2 modulates signaling between rods and rod bipolar cells. As described below, loss of PMCA2 has little or no effect on cone-mediated ERGs.
Photopic (cone-mediated) responses were recorded in response to single flashes after 30 min light adaptation at 30 cd s m−2. The photopic a-wave, caused by cone photoresponses, is very small in mice because of the small number of cones; thus, b-wave responses were measured at 2 Hz stimulation. Representative traces from control and dfw2J mice are shown in Figure 10A. Figure 10B shows that the amplitudes of photopic b-wave responses in dfw2Jmice were not significantly different from control littermates (110.1 ± 4.4 μV in controls, n = 26 eyes vs 102.1 ± 5.5 μV in dfw2J mice, n = 36; p = 0.29). The b-wave time-to-peak was also similar in control and dfw2J mice at 61.9 ± 1 ms (n = 26 eyes) vs 63.9 ± 1 ms in dfw2J mice (n = 36 eyes; p = 0.20) (Fig. 10C). These results suggest that the absence of PMCA2 has a greater effect on scotopic than photopic responses.
PMCA isoform 2, a Ca2+ transporter with high affinity for Ca2+, is strongly expressed in all vertebrate retinas (Križaj et al., 2002, 2004; Rentería et al., 2005). This study shows that amplitude and timing of scotopic light-evoked responses is compromised in deafwaddler2J (dfw2J) mouse retinas that lack PMCA2. In contrast, the photopic cone-dominated pathway was relatively unaffected. These results suggest that high-affinity Ca2+ extrusion modulates transmission of light-evoked responses in the scotopic visual pathway to a much greater extent than the photopic pathway.
The role of PMCA2 in photoreceptor signaling
Photoreceptors in dfw2J mice were anatomically indistinguishable from those of their littermate controls. The thickness of the outer and inner nuclear layers and the length of rod outer segments in dfw2J mice and controls were nearly identical. Direct recordings from isolated rod outer segments revealed that sensitivity and kinetics of light responses in dfw2J animals were normal. This result was consistent with absence of PMCA2 immunoreactivity from dfw2J outer segments and suggested that PMCA2 does not regulate photoreceptor development or phototransduction.
As opposed to the recordings from single cells, the amplitude of the scotopic a-wave, generally taken to represent integrated light-evoked activity of rods, was reduced by ∼25% in dfw2J retinas. A similar effect was reported by Jiang et al. (1996), who found that a-wave amplitudes in PLCβ4 knock-out mice were decreased with no change in the number of rods, rod outer segment length, or photoresponses of individual rods. At least four mechanisms could contribute to the reduction of the a-wave in retinas with normal rod photoresponses. (1) Trivially, the decrease in the a-wave amplitude in dfw2Js could be attributable to an increase in extracellular conductivity in dfw2J retinas. This mechanism would predict an equivalent decrease in the amplitude of photopic b-waves in dfw2J animals, which was not observed. (2) A-waves could have decreased because of intervention of an attenuating filter provided by a cataractous lens. However, rodent lens expresses only PMCA1 (Nabekura et al., 2001), and no cataracts were observed during external examination of dfw2J eyes. (3) Mutant retinas could contain patches of dysfunctional rods. Although no obvious patches were identified by light microscopic analysis of dfw2J retinal anatomy or by immunofluorescence, this possibility cannot be excluded. (4) PMCA2 could modulate the a-wave amplitude via posttransduction signaling pathway either in the inner segment and/or postsynaptically. Pharmacological manipulation of glutamate, dopamine, and GABA receptors has been shown to reduce dark-adapted a-wave amplitudes in frog, cat, rat, and primate retinas (Jaffe et al., 1985; Xu et al., 1991; Jamison et al., 2001; Szikra and Witkovsky, 2001; Möller and Eysteinsson, 2003; Robson et al., 2003), suggesting that the a-wave might have been compromised indirectly by signaling deficits in the inner segment, the inner retina, or altered feedback from postsynaptic cells. However, it is noteworthy that the a-wave in nob mice is normal despite total loss of rod bipolar cell responses (Pardue et al., 1998).
PMCA2 modulates synaptic transmission at rod synapses
Loss of PMCA2 alters signaling from rods to rod bipolar cells. Recordings from individual dfw2J rod bipolar cells showed that the sensitivity of light-evoked synaptic transmission was reduced by ∼50%, consistent with the reduction in the amplitude of the scotopic b-wave in dfw2J animals. Decreased b-wave responses were observed at light intensities that evoked no discernible a-wave responses (Fig. 7C). This loss of sensitivity could indirectly control the timing of visual signals, as seen in the 11 ms increase in time-to-peak of the dfw2J scotopic b-wave. On the assumption that loss of PMCA2 during development did not have an indirect effect on rod and rod bipolar synaptogenesis, our data indicate that PMCA2-mediated Ca2+ extrusion represents a post-phototransduction mechanism that increases the gain of the scotopic pathway.
PMCA2 is expressed both presynaptically in rod terminals and postsynaptically in bipolar cells. Hence, glutamate release and/or mGluR6 signaling could have been affected in dfw2J retinas. Although the kinetics of exocytosis and synaptic depression appear to be similar in rods and cones (Kreft et al., 2003; Rabl et al., 2006), the neurotransmission at cone synapses is significantly faster (Schnapf and Copenhagen, 1982; Cadetti et al., 2005). This may be attributable to different mechanisms of [Ca2+]i homeostasis in rods and cones (Križaj et al., 2003; Heidelberger et al., 2005). Loss of high-affinity Ca2+ extrusion would slow kinetics of Ca2+ extrusion and glutamate release, causing an increase in synaptic delays. A non-mutually exclusive alternative mechanism would be an increase in steady-state [Ca2+]i in dark-adapted rod bipolar cells attributable to Ca2+ influx through mGluR6-gated transduction channels. The transduction channels in depolarizing bipolar cells are Ca2+-permeable and desensitize in the presence of elevated [Ca2+]i (Shiells and Falk, 1999; Berntson et al., 2004; Nawy, 2004). Elevated [Ca2+]i would, in turn, “adapt” rod bipolar responses to light stimulation, accounting for the loss of sensitivity observed in ERG and single-cell recordings. In preliminary experiments using BAPTA-filled rod bipolar cells depolarized to +60 mV, we saw little change in the sensitivity of light responses (A. Sampath and F. Rieke, unpublished observations), suggesting that changes in [Ca2+]i play a minor role in mouse rod bipolar function. This would suggest a more prominent role for the presynaptic component of PMCA2 action at the rod synapse. However, it is likely that the role of PMCA2 is more significant in light-adapted cells with elevated [Ca2+]i (Berntson et al., 2004).
The increase in time-to-peak responses in scotopic b-waves of dfw2J retinas was highly significant at high intensities of illumination. At light levels comparable with those used in retinal slice recordings b-wave delays were reduced to ∼5 ms, which is within the SE of kinetic measurements in rod bipolar cell recordings. Consequently, the kinetics of rod bipolar responses in dfw2J retinas was not significantly different from the controls.
In contrast to scotopic responses, no statistically significant effect on the amplitude and kinetics of light-evoked responses was observed in the cone-dominated pathway. Cones do not express PMCA2, suggesting the cone pathway is dominated by the less sensitive PMCA1 mechanism (Križaj et al., 2002; Haverkamp et al., 2003). A subset of GFP-negative INL cells in retinas from L7–GFP transgenic mice was immunoreactive for PMCA2, indicating that PMCA2 could be localized to cone bipolar cells (Tomomura et al., 2001). However, our observation that photopic ERG is unaffected in PMCA2-deficient mice, together with lack of PMCA2 expression in cones, suggests that this isoform does not play a major role in the cone pathway. This is the first report of PMCA expression in retinal glia. At the moment, it is not known what role PMCAs play in Müller cell function or whether glial processes regulate rod or cone neurotransmission.
PMCA2 plays a different role in synaptogenesis of visual and auditory tissues
Cell populations in the mammalian retina express specific PMCA isoforms at defined developmental stages (Križaj et al., 2002; Rentería et al., 2005). PMCA2 was expressed throughout the early postnatal mouse retina, including the neuroblast precursors of photoreceptor cells. However, unlike the pronounced cellular degeneration observed in the cochlea, vestibular, and cerebellar structures of PMCA2-deficient mice (Street et al., 1998; Kozel et al., 1998, 2002; Dodson and Charabalapoulou, 2001), no obvious morphological phenotype was observed in adult dfw2J retinas up to 2 years of age. Moreover, a qualitative analysis revealed no correlation between the eye size and ERG potentials in dfw2J mice. These results suggest that upregulated activation of other Ca2+ extrusion and/or intracellular sequestration mechanisms may be able to compensate for the absence of PMCA2 during development of the retina and the eye. Because the loss of the retinal PMCA2 gene is functionally selective and does not lead to detectable changes in mRNA levels of other three PMCA isoforms (Fig. 2), it is likely that visual deficits observed in dfw2J mice occur as a result of loss of specific regulatory properties of PMCA2 and not secondary effects attributable to abnormalities in number or type of retinal neurons or compensatory expression of PMCA1, PMCA3, or PMCA4 genes.
Our results suggest that catalytic and/or modulatory properties of PMCA2 are relevant for regulating the sensitivity and kinetics of light responses in retinal pathways. PMCA1, the isoform mainly expressed in ribbon-containing cells (Križaj et al., 2002), is by far the most abundant retinal PMCA isoform (Fig. 2), required for the “housekeeping” Ca2+ clearance in tonically depolarized photoreceptors and cone bipolar cells. PMCA2 is likely to dominate Ca2+ clearance in amacrine and ganglion cells, as well as to increase the sensitivity of signaling in rods and rod bipolar cells. Although we did not directly measure function in inner retinal neurons from dfw2J animals, the elevated STR threshold responses suggest that activation of amacrine cells and/or ganglion cells is also compromised in dfw2J retinas.
In conclusion, we quantified the expression of all four PMCA isoforms in the mouse retina, characterized the expression of PMCA2 during development, and determined its functional role in modulating light-evoked signaling in retinal circuits and visual cortex. Our results suggest that high-affinity Ca2+ extrusion uniquely provided by PMCA2 modulates the rod–cone balance by selectively increasing the dynamic range and timing of rod synapses.
This work was supported by National Institutes of Health Grants EY13870 (D.K.), EY00415 (J.L.D.), EY01869 (D.R.C.), EY01919 (H.Y.), EY11850 (F.R.), DC02739 (B.L.T.), and EY02162 (D.K., J.L.D., D.R.C.), That Man May See (J.L.D., H.Y., D.R.C., D.K.), a career development award from Research to Prevent Blindness (J.L.D.), a career development award from The Foundation Fighting Blindness (J.L.D.), the Bernard A. Newcomb Macular Degeneration Fund (J.L.D., G.N.), the Howard Hughes Medical Institute (G.J.M., F.R.), and an unrestricted grant from Research to Prevent Blindness to the University of California, San Francisco Department of Ophthalmology. D.K. is a recipient of the James S. Adams Scholar Award from Research to Prevent Blindness. We are grateful to Edwin Dumlao, Doug Yasumura, and Linda Robinson for technical assistance. We thank Dr. Laura Frishman for her comments on this manuscript.
- Correspondence should be addressed to David Križaj, Department of Ophthalmology, Beckman Vision Center, Room K-131, University of California, San Francisco School of Medicine, 10 Koret Way, San Francisco, CA 94143-0730. Email: