Lack of CaBP1/Caldendrin or CaBP2 Leads to Altered Ganglion Cell Responses

Visual Abstract


Introduction
CaBPs are neuronal Ca 2ϩ -binding proteins with high homology to calmodulin (CaM; Haeseleer et al., 2000). CaBPs possess four EF-hand Ca 2ϩ -binding domains, but unlike CaM, only three of them are capable of binding Ca 2ϩ . CaM and CaBPs can interact with similar targets, but their differences can underlie distinct forms of regulation. For example, CaM and CaBPs can both bind to voltage-gated Ca v 1 L-type Ca 2ϩ channels, but CaM promotes Ca 2ϩ -dependent inactivation, whereas CaBPs oppose this process and prolong channel opening (Zhou et al., 2004;Cui et al., 2007).
Despite the identification of a large variety of interacting partners, the contributions of CaBP1/CD and CaBP2 to the transmission of light responses through the retina is unknown. In this study, we used CaBP1/CD knockout (KO; Cabp1 -/-) and newly generated CaBP2 KO (Cabp2 -/-) mice to investigate the importance of CaBP1/CD and CaBP2 for normal visual function. We find that both CaBP1/CD and CaBP2 regulate ganglion cell light responses and have different effects on their contrast sensitivity.

Generation of rat anti-CaBP1/CD and rat anti-CaBP2 antibodies
Anti-CaBP1/CD and anti-CaBP2 polyclonal antibodies were raised in rats by subcutaneous immunization with purified S-CaBP1 or S-CaBP2 recombinant proteins mixed with Freund's adjuvant (Cocalico Biologicals, Reamstown, PA). For affinity purification of rat anti-CaBP1/CD or anti-CaBP2 antibodies, purified CaBP1 or CaBP2 was coupled to cyanogen bromide-activated Sepharose (GE Healthcare, Piscataway, NJ) according to the manufacturer's protocol. After loading of a 10-fold dilution of the sera in PBS, the columns were washed with 20 volumes of PBS. The bound antibodies were eluted with 0.1 M glycine buffer, pH 2.5, and dialyzed overnight against PBS.
Generation and genotyping of Cabp2 -/mice All animal procedures were performed in accordance with the University of Washington Animal Care and Use Committee's regulations. A ϳ2.1-kb fragment covering part of exon 2 to part of intron 5 of Cabp2 gene was amplified first from C57Bl/6 genomic DNA by PCR with primer FH939 (5=-GTCGACTAAGTAGCTGAGACCAG AAGAGATCGAAG-3=) that was extended with a SalI restriction site and includes a stop codon in all three open reading frames and primer FH940 (5=-GGTACCAGGAGG GCTCAGTTGCTCACATTA-3=) that was extended with a KpnI site. After sequencing of this 2.1-kb fragment, it was cloned into the targeting vector SalI and KpnI opened between the neomycin phosphotransferase gene and herpes simplex virus thymidine kinase gene. The long arm of ϳ5.2 kb covering the promoter region of the Cabp2 gene upstream the ATG was amplified in two fragments. The upstream fragment of 2 kb was amplified by PCR with primer FH937 (5=-GCGGCCGCTCGTGGTTTCAGGTGCTCTA-CACA-3=) that was extended with a NotI site and primer FH947 (5=-TAAGGTCTTAGAGGGTCTGACAGG-3=) that covers a SpeI restriction site. A 3.2-kb fragment upstream of the CaBP2 initiation codon was amplified by PCR with primer FH938 (5=-ACCCAGGTTTCTGGCCTTATGTCT-3=) that also covers the SpeI restriction site and FH948 (5=-TACCGACTGACTCATGCCTAGGTT-3=) that hybridizes a few bases downstream of the CaBP2 initiation codon. All fragments were cloned in the pCRII-TOPO vector and sequenced. A tdTomato vector (originally a gift from Dr. Roger Tsien, provided by Dr. Rachel Wong) was modified by mutagenesis using QuikChange Lightning Multi Site-Directed Mutagenesis (Agilent Technologies, Santa Clara, CA) to introduce a NheI site after the SV40 polyadenylation site with primer FH1043 (5=-GTATCTTAAGGCGTAGCTAG-CAAGCTTTAATATTTTGTTAAAATTCGC-3=) and delete the internal NcoI site in tdTomato with primer FH1044 (5=-CGTAATGCAGAAGAAGACGATGGGCTGGGAGGCCTCC-3=). tdTomato was then fused to the CaBP2 promoter as a fragment NcoI-NheI and transferred together in the targeting vector as NotI-BglII and BglII-NheI fragments.
The KpnI linearized targeting vector was electroporated into B6/BLU embryonic stem cells. Recombinant clones were selected on medium containing G418. Transfected embryonic stem (ES) cell clones were first screened through PCR analysis. To screen for homologous recombination, we used primers FH 1064 (5=-GGGTCGTTTGTTC GGATCCTCTAGAGTC-3=) located in the neo cassette and FH1139 (5=-TACACAGGCTCACCGAGACATCAT-3=) hybridizing approximately 163 bp downstream of the 3= end of the short arm in the Cabp2 gene and amplifying a fragment of ϳ2.3 kb. A control PCR for the wild-type (WT) gene was made with primers FH1139 and FH1140 (5=-ACCAGGCATGGAGTTGGGTATGAA-3=) hybridizing in intron 2A, 480 bp upstream of the 5= end of the short arm of the Cabp2 gene and amplifying a fragment of ϳ2.75 kb. Targeted disruption of the Cabp2 gene was then confirmed by Southern blot analysis. Ten micrograms of genomic DNA was digested with MfeI and hybridized with a 0.6-kb 5=-end probe located 100 bp upstream of the 5= end of the long arm (Fig. 1). This probe hybridized to a MfeI fragment of ϳ13.4 kb of the WT allele or a MfeI fragment of ϳ9.1 kb if the Cabp2 gene is targeted.
One targeted ES clone was injected into C57BL/6J blastocysts. One 80% male chimera was crossed with C57BL/6J mice, and agouti offspring were genotyped by PCR to verify germline transmission. Confirmation of Cabp2 gene targeting was first performed with primers FH1139, FH1140, and FH1064 as indicated above. For routine genotyping of the offspring, the WT allele was identified with primers FH1214 (5=-CCCTAAGACACCCAG ACAGATGA-3=, located in intron 2A) and FH1218 (5=-GAAGTGTCAGCCAGATGGACAAA-3=, hybridizing in intron 2B) that generate a PCR product of 0.4 kb. The targeted Cabp2 allele was identified with primers FH1218 and FH1384 (5=-TGGAGAGGCTATTCGGCTATGA-3=', lo-cated in the neo cassette) that produces a PCR product of ϳ1.07 kb.

RT-PCR analysis of CaBP1 and CaBP2 transcripts
Total RNAs were isolated from mouse retina using RNeasy kit (Qiagen, Valencia, CA). Total RNA (1 g) was subjected to first-strand cDNA synthesis using Superscript III reverse transcriptase and oligo(dT) in a volume of 20 l according to the manufacturer's protocol (Invitrogen). Short and long CaBP1 were amplified with common forward FH705 (5=-CACCATGGGCAACTGCGTCAAGT CG-3=) and reverse FH706 (5=-TCAGCGAGACATCAT CCGGACAAAC-3=) primers. Caldendrin was amplified with primers FH1089 (5=-ACACCAATCATATCTGCC GTCTCC-3=) and FH1093 (5=-GCGATGGGGAGGAACGG  Figure 1. Targeting of the Cabp2 gene. A, Scheme of the mouse Cabp2 gene with its exons. Arrows above the scheme indicate primers used to clone by PCR Cabp2 genomic fragments. The PGK-DTA and HSV-TK cassettes were included in the targeting vector for negative selection in transfected ES cells. In the targeting vector, the neo cassette (positive selection) replaces exon 1 and exon 2A of the Cabp2 gene. The targeting vector is constructed by using a ϳ5-kb DNA fragment as long arm that extends upstream of the initial ATG and covers the CaBP2 promoter. tdTomato was cloned and fused on the initiation codon of CaBP2. The short arm is a ϳ2.0-kb genomic fragment encompassing exon 2B to intron 5 of the Cabp2 gene. Arrows (FH1064, FH1139, and Fh1140) below the scheme indicate primers used to select targeted Cabp2 allele. The location of MfeI restriction site as well as the probe (probe SB) used for analysis of the targeted allele using Southern blot are also indicated. B, Southern blot analysis of ES cell clones. MfeI-digested genomic DNA isolated from wild-type B6/Blu ES cells (WT) or targeted clones 1, 37, and 40 was analyzed by Southern blot with an external probe as shown in A and shows a fragment of 13.4 kb for the wild-type Cabp2 allele and a fragment of 9.1 kb for the targeted Cabp2 allele. C, PCR-based analysis of mice for Cabp2 targeting. A 2.3-and 2.75-kb PCR product is amplified with primers FH1140 and FH1139 as shown in A for the WT Cabp2 allele and with primers FH1064 and FH1139 for the targeted allele, respectively.
For the analysis of whole-mount retinas, the mouse retinas were fixed for 1 h in 4% paraformaldehyde in PB before dissection, incubated with 5% normal goat serum in PBST buffer overnight at 4°C, and incubated overnight at 4°C with a mix of anti-Syt2 (1:200) or anti-NK3R (1:500) and anti-Ctbp2 (1:1000). After three washes for 15 min in PBST, a mix of Alexa Fluor 568 goat anti-mouse IgG2a (␥2a) and Alexa Fluor 488 goat anti-mouse IgG1 (␥1) or a mix of Alexa Fluor 488 goat anti-mouse IgG and Alexa Fluor 555 goat anti-rabbit was reacted with the sections overnight at 4°C. After three washes in PBST, the retinal whole mounts were mounted with the photoreceptor side down in Prolong antifade reagent and analyzed under a confocal microscope (LSM710, Zeiss, Oberkochen, Germany). Immunofluorescent images were obtained with 63ϫ/1.4, 40ϫ/1.1, or 40ϫ/1.3 NA objectives (Zeiss).
For quantification of the Ctbp2-labeled synaptic ribbons, confocal images of en face views in whole mounts of WT, Cabp1 -/-, and Cabp2 -/retinas were acquired using a 63ϫ/ 1.4 objective at a voxel size of 0.028 ϫ 0.028 ϫ 0.359 m (x-y-z). The sublamina containing the axon terminals of cone OFF bipolar type 1 and type 2 was identified by colabeling with anti-NK3R and anti-Ctbp2. The axon terminals of cone ON bipolar type 6 were visualized by colabeling with anti-syt2 and Ctbp2. Ctbp2-labeled synaptic ribbons were counted in images of six randomly chosen 25-m 2 areas across the retina of four mice for each phenotype. The statistical analysis was done using the two-tailed Student's t-test. Data are expressed as mean Ϯ SE.

Electron microscopy
Eye cups from WT, Cabp1 -/-, and Cabp2 -/mice of either sex were fixed in 4% glutaraldehyde in 0.1 M sodium cacodylate buffer at 4°C. After four washes in 0.13 M PB, eyecups were dissected and postfixed in 1% buffered osmium tetroxide for 2 h on ice. After another four washes in PB, the retinas were en bloc stained in 1% uranyl acetate overnight at 4°C, dehydrated through a graded series of ethanol and two incubations in propylene oxide, and infiltrated in a 1:1 mixture of propylene oxide:epon araldite and two incubations in epon araldite, before polymerization at 60°C. Seventy-nanometer sections were cut on a Leica UC7 ultra microtome and stained with Reynolds lead citrate. The sections were viewed and images recorded on a JEOL 1230 electron microscope (JEOL, Peabody, MA). Images of bipolar cell terminals were analyzed in the outer part of the inner plexiform layer within 2-5 m from the inner nuclear layer where type 1 and type 2 cone OFF bipolar cell axon terminals are located. Images of photoreceptor synapses with bipolar cells were taken in the outer plexiform layer.

Electrophysiology and visual stimulation
Experiments were conducted on whole-mount retinal preparations taken from dark-adapted WT, Cabp1 -/-, or Cabp2 -/mice (Schwartz et al., 2012). Isolated retina was stored in oxygenated (95% O 2 /5% CO 2 ) Ames medium (Sigma-Aldrich, St. Louis, MO) at ϳ32-34°C. Isolated retinas were flattened onto poly-L-lysine slides (whole mount; Schwartz et al., 2012), placed in a custom microscope, and perfused with Ames solution at a rate of ϳ8 mL/min. Retinal neurons were visualized and targeted for patch-clamp recordings using infrared light (Ͼ950 nm). Voltage-clamp recordings were obtained using pipettes (2-3 M⍀) filled with an intracellular solution containing (in mM): 105 Cs methanesulfonate, 10 tetraethylammonium chloride, 20 HEPES, 10 EGTA, 2 QX-314, 5 Mg-ATP, 0.5 Tris-GTP (ϳ280 mOsm, pH ϳ7.3 with CsOH). Alexa 594 dye (100-200 microM) was included in the intracellular solution to image the ganglion cells post recording as shown in Figs. 8 and 9. Full-field light stimuli (diameter 500 m) were delivered to the tissue from an LED with peak spectral output at 405 nm. Recordings were performed from ventral retina. All ganglion cell recordings shown were recorded at a background light level of ϳ900 R‫/ء‬S cone/s, where cones dominate retinal responses. The statistical analysis was done using the two-tailed unpaired Student's t-test; ‫ء‬p Ͻ 0.05, ‫‪p‬ءء‬ Ͻ 0.01, ‫‪p‬ءءء‬ Ͻ 0.001; error bars indicate SD.

Generation of Cabp2 -/mice
Two splice variants of CaBP2, S-CaBP2, and L-CaBP2 exist and differ by the absence or presence of exon 2A ( Fig. 1A; Seidenbecher et al., 1998;Haeseleer et al., 2000;Sokal et al., 2000). To generate Cabp2 -/mice, we designed a targeting vector to replace exon 1 and exon 2a of the Cabp2 gene with tdTomato and a PGK-neo cassette as shown in Figure 1A. Stop codons were also introduced in all three open reading frames of exon 2B to prevent expression of any CaBP2 variants. PCR screening of ES cells identified three targeted ES cell clones for which targeting of the Cabp2 allele was further verified using Southern blot (Fig. 1B). Injection of targeted ES clone 37 produced one male chimera that gave germline transmission. Targeting of the Cabp2 gene was confirmed in the offspring by PCR on genomic DNA (Fig. 1C). Cabp2 -/mice were viable and did not show any apparent physiological deficits or breeding problems.

Analysis of CaBP1/CD and CaBP2 expression in KO mice
The generation of Cabp1 -/mice was described by Kim et al. (2014). As a first step in the characterization of Cabp1 -/mice, we verified the absence of CaBP1/CD expression in the retina. Because exon 1 of caldendrin and exon 1 of S-and L-CaBP1 were removed in Cabp1 -/mice, all three splice variants were targeted. The absence of transcripts for all three splice variants was confirmed by reverse-transcription (RT)-PCR using retinal RNA of Cabp1 -/mice ( Fig. 2A), whereas all three variants were observed in Cabp1 ϩ/ϩ mice. As a positive control, PCR was performed with primers designed to detect GAPDH and showed transcripts in both Cabp1 ϩ/ϩ and Cabp1 -/mice. The absence of CaBP1/CD proteins in the retina of Cabp1 -/mice ( Fig. 2B) was confirmed using immunohistochemistry. Staining was not detected in Cabp1 -/mice with anti-CaBP1/CD antibodies recognizing all three splice variants.
We also verified the absence of CaBP2 expression in the newly generated Cabp2 -/mice. Using RT-PCR, transcripts were not detected for CaBP2 in Cabp2 -/retina ( Fig. 2C) that showed transcripts for the GAPDH positive control. Using immunohistochemistry, we tested that CaBP2 proteins are not present in Cabp2 -/mice. Unfortunately, because of the high homology between CaBP1/CD and CaBP2 proteins, our anti-CaBP2 antibodies recognize both CaBP1/CD and CaBP2. As a consequence, signals were still observed in the inner nuclear layer of Cabp2 -/mouse retina immunolabeled with anti-CaBP2 (Fig. 2D). This residual signal was confirmed to be due to cross-reactivity of anti-CaBP2 antibodies with CaBP1/CD proteins as demonstrated by the absence of signals in Cabp1 -/-/Cabp2 -/double KO retinas immunolabeled with anti-CaBP2 (Fig. 2D). Therefore, specific labeling of CaBP2 proteins can be visualized with our anti-CaBP2 antibodies in Cabp1 -/mouse retina. Altogether, these results confirmed that CaBP1/CD and CaBP2 are not expressed in Cabp1 -/and Cabp2 -/mice.
To determine whether the absence of CaBP1/CD triggers structural changes of the retina, we analyzed the morphology of the retinal cells lacking CaBP1/CD using antibodies against these marker proteins. NK3R and Syt2 immunolabeling showed normal bipolar cell morphology and normal thickness of the INL and IPL in Cabp1 -/compared with Cabp1 ϩ/ϩ retina (Fig. 3D, F). Calretinin immunostaining demonstrated normal stratification of the IPL of Cabp1 -/retina (Fig. 3E). To investigate whether lack of CaBP1/CD results in synaptic alterations, we analyzed retinas colabeled with anti-Syt2 or anti-NK3R and anti-ribeye/CtBP2 that labels synaptic ribbons in both the IPL and the outer plexiform layer (OPL). We analyzed the synapses in both plexiform layers because CaBP1/CD is expressed throughout the cone bipolar cells. Normal horseshoe-shaped and punctuate staining of the synaptic ribbons was observed in the IPL (Fig. 4A) and OPL (Fig.  4C) of Cabp1 ϩ/ϩ and Cabp1 -/mouse retina. We quantified the synaptic ribbons labeled with anti-Ctbp2 in 25m 2 fields of OPL and IPL of WT and Cabp1 -/mice. The density of synaptic ribbons in WT and Cabp1 -/mice did not differ significantly (Table 2).
Synaptic terminals were analyzed at higher magnification using transmission electron microscopy to look for more subtle changes. Cone synapse morphology was normal, showing invagination of a pair of horizontal cells and a cone ON bipolar cell (Fig. 4D). Normal membraneanchored synaptic ribbons and tethered vesicles were observed in both IPL and OPL of Cabp1 ϩ/ϩ and Cabp1 -/retina (Fig. 4B, D). These results suggest that CaBP1/CD is not required for normal development or maintenance of overall retinal morphology.

CaBP2 localization in bipolar cells of the retina
To determine the consequences of CaBP2 deficiency, we first determined the distribution of CaBP2 in the mouse inner retina. Because our anti-CaBP2 antibody reacts with both CaBP1/CD and CaBP2 (Fig. 2D), and because there are no retina morphological changes in Cabp1 -/-, we chose to study CaBP2 localization using Cabp1 -/mouse retina. Anti-CaBP2-labeled axons ramified in two distinct layers of the IPL, likely corresponding to ON and OFF cone bipolar cells (Fig. 2D). To determine the bipolar cell types that express CaBP2, we used antibodies against selective markers. Double-labeling experiments with antibodies against CaBP2 and CaBP5 ( Fig.  5A-C) demonstrated that CaBP2 does not localize in bipolar cells expressing CaBP5, indicating that CaBP2 is not expressed in type 3 OFF, type 5 ON, and rod bipolar cells Ghosh et al., 2004). CaBP2 is also not expressed in anti-calsenilin-labeled type 4 OFF cone bipolar cells (Fig. 5D-F; Haverkamp et al., 2008). Gustducin-GFP mice express GFP in type 7 ON cone bipolar cells (Lin and Masland, 2005). Although our anti-CaBP2 antibody would cross-react with CaBP1/CD in gustducin-GFP retinas, none of the GFP-labeled cells in these mice were labeled with anti-CaBP2 (Fig. 5G-I). We can thus conclude that type 7 ON bipolar cells do not express CaBP1 and CaBP2 proteins.
To visualize the sublaminae of the inner plexiform layer, we labeled the mouse retina with an antibody to calretinin that displays a laminar pattern in three layers of the IPL separating sublaminae 1/2, 2/3, and 3/4 (Wässle, 2004; Fig. 6C) . Colabeling with anti-CaBP2 and anti-calretinin revealed that CaBP2 localizes in cells with axons in sublamina 1 and sublamina 3/4 (Fig. 6C). A partial overlap of the distribution of CaBP2 with NK3R indicated that either type 1 or type 2 OFF bipolar cells express CaBP2 (Fig.  6A). Costaining with anti-Syt2 and anti-CaBP2 antibodies showed colocalization of Syt 2 and CaBP2 in sublaminae 3/4 but not in sublamina 1 (Fig. 6B). This result indicates that CaBP2 is expressed in type 6 ON bipolar cells. We can also conclude from Syt 2 and NK3R labeling experiments that type 1 OFF bipolar cells express CaBP2 (Table 1). Although CaBP2 showed colocalization with Syt2 in sublamina 3/4, additional CaBP2-labeled cells appeared to have their axon arborization extending in sublamina 3/4. Because of the lack of other markers for these bipolar cells, it was not possible to identify these CaBP2-expressing cells.

Analysis of retinal morphology in Cabp2 -/mice
To determine whether CaBP2 is necessary for normal retinal development and structure, we analyzed the morphology of the CaBP2-deficient cells using antibodies against Syt2 and NK3R to look at type 6 ON and type 1 OFF bipolar cells. Overall morphology of these bipolar cells and thickness of the retinal layers did not differ noticeably between Cabp2 ϩ/ϩ and Cabp2 -/retinas (Fig.  6). The Syt2-labeled axons of type 6 ON bipolar cells in the inner IPL and NK3R-labeled axons of OFF bipolar cells in the outer IPL showed normal morphology (Fig. 7). No difference was observed in the OPL between the Syt2labeled dendrites of Cabp2 ϩ/ϩ and Cabp2 -/retinas. There were also no apparent changes in the number of synapses visualized with anti-Ctbp2 in Cabp2 -/mice (Table 2). Normal synaptic ribbons and tethered vesicles were observed in both IPL and OPL of Cabp2 ϩ/ϩ and Cabp2 -/retinas using electron microscopy (Fig. 7B, D). Cone bipolar cell and horizontal cell invaginations appeared to make a normal synapse with cone photoreceptors (Fig. 7D).

Altered light responses of ganglion cells in Cabp1 -/and Cabp2 -/mice
To determine whether CaBP1/CD and CaBP2 are important for retinal function, we measured ganglion cell responses to full-field light stimuli. We selected three well-characterized mouse ganglion cell types: ON-alpha, OFF transient (OFF T), and OFF sustained (OFF S). Extensive work on these alpha-like ganglion cells has identified type-specific response properties generated from distinct circuit mechanisms (Pang et al., 2003;Murphy and Rieke, 2006;van Wyk et al., 2009;Schwartz et al., 2012). Recordings were performed at a background of 900 R‫/ء‬Scone/s, where cones dominate retinal responses.
Because CaBP2 is expressed in type 6 bipolar cells, and because ON alpha ganglion cells receive most excitatory synaptic input from this bipolar cell type (Morgan et al., 2011;Schwartz et al., 2012), we chose ON alphas as the primary candidate for studying perturbation in ganglion cell function in Cabp2 -/retinas (Fig. 8). Responses to a 100% contrast light flash were significantly smaller in mice lacking CaBP2 (Fig. 8C, D).
We characterized the response kinetics and contrast sensitivity using linear-nonlinear models computed from excitatory synaptic currents in response to a random noise stimulus (Chichilnisky, 2001;Kim and Rieke, 2001;Rieke, 2001). This model provides a relatively simple description of how continuous light inputs are transformed into cellular responses and provides an effective means of characterizing contrast-dependent changes in the amplitude and kinetics of the light response of a cell. The model consists of a linear filter that accounts for the time course of the response and a time-independent or "static" nonlinearity that converts the filtered stimulus into responses (excitatory synaptic currents; Fig. 8).
The linear filters of ON alpha ganglion cells were slightly slower in Cabp2 -/retinas compared with control retinas (Fig. 8F, G). The nonlinearities were strikingly different between the ON alpha ganglion cells of Cabp2 ϩ/ϩ and Cabp2 -/retinas (Fig. 8H, I).
The response range of the responses-i.e., the difference between the maximum and minimum excitatory synaptic current evoked by the random noise stimulus-was significantly smaller for ON alpha ganglion cells in Cabp2 -/retinas compared with control cells (Fig. 8H, I).  Because CaBP1/CD is expressed in OFF bipolar cells, we analyzed responses of OFF S and OFF T ganglion cells in Cabp1 -/retinas. Differences in lamination of OFF S and OFF T ganglion cells in the IPL suggest that they derive their excitatory inputs from different OFF bipolar cells. Responses to a light flash were larger and slower for OFF T ganglion cells compared with control retinas. Responses of OFF S ganglion cells in Cabp1 -/retinas to a light flash exhibited no significant difference in amplitude with respect to control retinas (Fig. 9A, B). The response kinetics were slower for OFF S and OFF T ganglion cells in Cabp1 -/retinas compared with controls (Fig. 9A, B).
Nonlinearities were strikingly different between the OFF alpha ganglion cells from Cabp1 -/and control retinas (Fig. 9A, B). The change in the nonlinearity was opposite for OFF S and OFF T ganglion cells, with a smaller response range for Cabp1 -/-OFF S ganglion cells and larger response range for Cabp1 -/-OFF T ganglion cells. Changes in excitatory synaptic currents affect the spike output of the ganglion cells. Indeed, spike responses of ON alpha ganglion cell to a 100% contrast step were weaker in Cabp2 -/retinas than in control retinas (Fig. 10). Similarly, OFF T ganglion cells exhibited a stronger spike response in Cabp1 -/than in control retinas (Fig. 10).  Figures 4 and 6, from four mice for each phenotype were analyzed by confocal microscopy. Ctbp2-labeled ribbons were counted in images of six areas of 25 m 2 for each mouse in the OPL. Ctbp2-labeled puncta were also quantified in the NK3R-labeled sublamina containing type 1 and 2 OFF bipolar cell axon terminals or Syt2-labeled sublaminae of the inner IPL containing the type 6 ON bipolar cells. Numbers represent mean Ϯ SE (n ϭ 24). Statistical comparisons with WT mice were performed using the Student's t-test. We also compared excitatory synaptic currents of ON alpha ganglion cells in Cabp1 -/retinas and OFF T ganglion cells in Cabp2 -/retinas with corresponding control cells (Fig. 11). Responses of these cells could be affected, since CaBP1/CD is expressed in amacrine cells and CaBP2 is expressed in type 1 OFF bipolar cells. Interestingly, the response kinetics of the ON alpha ganglion cells were slower, but the response range was unchanged or slightly increased in Cabp1 -/retinas, unlike in Cabp2 -/retinas (Fig. 11). The response kinetics of the OFF T ganglion cells and the response range of responses did not change significantly in Cabp2 -/retinas (Fig. 11).

Discussion
Ca 2ϩ -binding proteins modulate the cellular activities of enzymes, channels, and structural proteins in response to changes in intracellular [Ca 2ϩ ]. Here, we determined whether the lack of CaBP1/CD or CaBP2, neuronal calcium binding proteins expressed in distinct retinal secondary neurons, results in altered transmission of light responses through the retina. We found that not only the kinetics but also the amplitude and response range of the light responses in ganglion cells are affected in the absence of CaBP1/CD or CaBP2, whereas the overall morphology of the CaBP1/CD-or CaBP2-deficient retina and synapses is not affected. These findings give a first insight into the importance of CaBP1/CD and CaBP2 for normal visual function.

CaBP2 is expressed in cone ON and OFF bipolar cells
Our immunohistochemistry results showed that CaBP2 is expressed in type 1 cone OFF bipolar and type 6 cone ON bipolar cells. Type 1 cone ON bipolar cells are the first retinal neurons to date shown to express two different CaBPs, CaBP2, and CaBP1/CD (Figs. 3 and 6;Haverkamp et al., 2003a), since CaBPs were previously shown to be expressed in distinct neurons Haverkamp et al., 2003a;Ghosh et al., 2004). This raises the question of whether CaBP1/CD and CaBP2 might modulate different targets and/or have a different effect on the same target in type 1 cone OFF bipolar cells. In addition to type 1 and type 6 cone bipolar cells, other cells of the inner nuclear layers in sublamina 3/4 appeared to be immunostained with anti-CaBP2 antibody. Although we cannot exclude that type 8 and type 9 bipolar cells might express CaBP2, the axons of those cell types have been described to stratify in S4, S5 (Ghosh et al., 2004) rather than S3,S4 as observed for the CaBP2labeled cells. Still, type 8 bipolar cells appeared to stratify broadly with the Syt2-labeled type 6 bipolar cells (Dunn and Wong, 2012). These unidentified CaBP2-labeled cells looked rather similar to the CB3,4 cells previously visualized using gene gun labeling and displaying a broad axonal arbor that extends through layers 3 and 4 (Pignatelli and Strettoi, 2004).

Retinal morphology of Cabp1 -/and Cabp2 -/mice is normal
Immunohistochemistry analysis of Cabp1 -/and Cabp2 -/mouse retinas using specific markers showed no changes in the cell, dendritic, and axon morphologies of cells deficient for CaBP1/CD or CaBP2. Analysis of synapses by immunostaining with Ctbp2 or electron microscopy did not reveal changes in the morphology or density of synaptic ribbons or tethered synaptic vesicles. This suggests that CaBP1/CD and CaBP2 are not essential for the development and maintenance of a normal synapse or the docking of synaptic vesicles. Similar results were obtained for CaBP5 KO mice. As for Cabp1 -/and Cabp2 -/mouse retinas, no significant changes in the number of synaptic vesicles or docked synaptic vesicles was observed in Cabp5 -/mouse retina (Rieke et al., 2008). However, CaBP5 was shown to stimulate synaptic vesicle exocytosis (Sokal and Haeseleer, 2011). It is thus possible that CaBP1/CD and CaBP2 play a role in neurotransmitter release by regulating events beyond vesicle docking.

Kinetics and amplitude of ganglion cell responses are affected in Cabp1 -/and Cabp2 -/mice
We investigated the light responses of alpha ganglion cells to determine the importance of CaBP1/CD and CaBP2 for proper transmission of light responses across the retina. Compared with responses of ganglion cells from WT retina, we observed a ϳ50% increase in the amplitude of the excitatory postsynaptic currents of OFF T ganglion cells of Cabp1 -/mice and a ϳ60% decrease of the excitatory synaptic currents in ON-alpha ganglion cells of Cabp2 -/mice. Because CaBP1/CD and CaBP2 are expressed throughout bipolar cells, it is possible that these altered responses originate from a defect in the processing of the light signal received from cones. Such a defect at the cone-to-cone bipolar cell synapse should affect the electroretinograms of Cabp1 -/and Cabp2 -/mice. However, the time to peak and amplitude of b-waves of electroretinograms of Cabp1 -/and Cabp2 -/mice did not differ significantly from those of WT mice (data not shown); this suggests that alterations in the transmission of the light signal likely occur downstream of the bipolar cell dendrites.
Interestingly, the two OFF alpha ganglion cell types displayed similar changes in kinetics but opposite changes in response range in Cabp1 -/mice. The opposite changes in response range could be explained if the bipolar cells that provide input to OFF S and OFF T ganglion cells express different targets for CaBP1/CD. Alternatively, CaBP1 may differently shape the synaptic input to these two types of OFF ganglions cells.
The absence of CaBP2 resulted in a decrease in the response range of the ON alpha ganglion cell response. The absence of CaBP1/CD slowed response kinetics but did not change the response range of the excitatory synaptic currents of the ON alpha ganglion cells. Because CaBP1 is not found in cone ON bipolar cells, the effect of CaBP1 deficiency on ON alpha ganglion cell excitatory inputs could result from alterations in responses of amacrine cells that provide presynaptic inhibition on type 6 bipolar cells and shape the kinetics of excitatory inputs of ON alpha ganglion cells. The complex interplay between ON and OFF pathways means that other explanations are possible as well.
Although CaBP2 deficiency results in decreased amplitude of the excitatory synaptic currents in ON alpha ganglion cells (Fig. 8), the lack of CaBP1/CD causes an increase of OFF ganglion cell responses (Fig. 9). This opposite effect on the amplitude of ganglion cell responses suggests that CaBPs may have different targets in different cell types or opposing effects on similar targets. Prominent targets of CaBPs, including CaBP1 and CaBP2, are the voltage-gated Ca 2ϩ channels Haeseleer et al., 2004;Zhou et al., 2004;Cui et al., 2007;Rieke et al., 2008;Schrauwen et al., 2012). Voltagegated Ca 2ϩ channels are directly involved in vesicle exocytosis and are a possible target for CaBP1/CD and CaBP2 in bipolar cells and amacrine cells (Pan et al., 2001;Habermann et al., 2003;Singer and Diamond, 2003;Balakrishnan et al., 2015). Specifically for CaBP1, it was shown that CaBP1 has opposite effects depending on the type of voltage-gated Ca 2ϩ channels it modulates Zhou et al., 2004;Cui et al., 2007). Because CaBPs can both inhibit or activate voltage-gated calcium channels, it might concomitantly decrease or increase synaptic vesicle release. Using electrophysiological recordings, cone bipolar cells have been shown to express diverse Ca v channels, including Ca v 1 (L-type) and Ca v 3 (T-type; Kaneko et al., 1989;Pan, 2000;Pan et al., 2001). T-type Ca 2ϩ channels have not been shown to be modulated by calmodulin or calmodulin-like CaBPs. Although CaBP1 can interact directly with L-type and P/Q type calcium channels Zhou et al., 2004;Cui et al., 2007), it could also modulate Ca 2ϩ channel activity by protein kinase pathways. Indeed, in vitro experiments have shown that CaBP1 and CaBP2 can modulate calmodulin kinase II activity  and calmodulin kinase II promotes the activity of T-type voltage-gated calcium channels (Welsby et al., 2003). The variety of voltage-gated calcium channels expressed in the inner retinal neurons and their specific regulation by CaBPs might thus account for the distinct effect of CaBP1/CD and CaBP2 deficiency on the transmission of light responses to ganglion cells.
CaBPs have also been shown to interact with other proteins playing an important role in neurotransmitter release. For example, CaBP5 interacts with Munc18-1, a component of the synaptic vesicle cycle. CaBP5 was suggested to have a role in synaptic vesicle priming by binding to a specific domain of Munc18-1 necessary for its interaction with the soluble N-ethylmaleimide-sensitive   fusion protein attachment protein receptors (SNARE) complex. We found here that deficiency of CaBP1/CD also results in slower response kinetics than in WT animals. If CaBP1/CD were to have similar effects on synaptic vesicles priming, it could affect the kinetics of the light responses.
In conclusion, the phenotype of Cabp1 -/and Cabp2 -/mice gives us a first insight into the important role of CaBP1/CD and CaBP2 for normal transmission of the light responses through retinal circuits. Although the effects of CaBP1/CD and CaBP2 deficiency might result from a change of modulation of voltage-gated calcium channels, other cellular processes might also be altered. Future experiments investigating the interacting partners for CaBP1/CD and CaBP2 in bipolar cells will help to explain the visual phenotypes in the CaBP KO mice and the cellular functions of these proteins in the retina.