Targeted Expression of Anoctamin Calcium-Activated Chloride Channels in Rod Photoreceptor Terminals of the Rodent Retina

Abstract

Purpose.: In the vertebrate retina, calcium-activated chloride channels are expressed in photoreceptor synaptic terminals. These channels are involved in the control of transmitter release in the dark. The search for their molecular identity has recently lead to the localization of the protein anoctamin 2 (also TMEM16B) in the outer plexiform layer of the rodent retina. Since both rod and cone photoreceptors have their terminals in this layer, it was not clear which of these express anoctamin 2. Here, we examine rod spherules and cone pedicles for expression of anoctamin 2.

Methods.: Expression of anoctamin genes was studied in the rat eye using RT-PCR. Immunohistochemical experiments were used to localize anoctamins and chloride transporters with their regulatory kinases. Photoreceptor synaptic proteins, as well as the lectins Peanut agglutinin and Griffonia simplicifolia agglutinin, were used to distinguish retinal structures.

Results.: Anoctamin 1, 2, and 10 were found to be expressed in the eye. Anoctamin 2 was expressed as a splice variant that includes exon 15 of the genomic structure. The protein is exclusively expressed in rod terminals and is not present in cone pedicles. Expression is not clustered at the ribbon complex, but spread across the presynaptic membrane where it colocalizes with the plasma membrane calcium pump. The electroneutral chloride transporter NKCC1 is expressed in photoreceptor terminals, together with its regulatory kinases SPAK and OSR1.

Conclusions.: Rod photoreceptor terminals possess the molecular machinery for chloride accumulation and for the generation of calcium-dependent chloride currents conducted through anoctamin 2 channels. We discuss this finding in the framework of the established hypothesis that calcium-activated chloride channels are part of a feedback inhibition mechanism that limits transmitter release in the dark.

Introduction

Studies of the amphibian retina have revealed that transmitter release from synaptic endings of photoreceptors is codetermined by the intracellular chloride concentration ([Cl⁻]_i). In the tiger salamander (Ambystoma tigrinum), chloride ions have a facilitating effect, as the L-type voltage-gated Ca²⁺ channels in rod spherules and cone pedicles require intracellular Cl⁻ for efficient gating activity. ^1,2 Cl⁻ dependence may have a regulatory function, as a decrease of [Cl⁻]_i tends to reduce presynaptic Ca²⁺ influx and to inhibit transmitter release. Cl⁻ levels and Ca²⁺ currents are functionally coupled through Ca²⁺-activated Cl⁻ channels. These channels open upon Ca²⁺ influx and provide a substantial Cl⁻ conductance in the membrane of photoreceptor terminals. ^3–5 In rod spherules, Cl⁻ efflux is triggered when Ca²⁺ channels are activated at the resting membrane voltage in the dark (V_rest ∼ −40 mV) because the chloride equilibrium potential (E_Cl) is near −20 mV. ⁶ The charge transfer resulting from Cl⁻ efflux depolarizes the plasma membrane and may support Ca²⁺ channel activation to some extent. However, the resulting decline of [Cl⁻]_i within the spherule lumen promotes closing of Ca²⁺ channels and, thereby, mediates feedback inhibition of Ca²⁺ entry. The net effect of presynaptic Cl⁻ fluxes on Ca²⁺ entry in rod spherules is not completely understood, but a critical function appears to be the prevention of excessive transmitter release under scotopic conditions. ⁷ In salamander cone pedicles, Cl⁻ effects are less pronounced because V_rest and the chloride equilibrium potential (E_Cl) have similar values (∼ −46 mV ⁸ ). In the absence of a chemo-electrical potential difference for Cl⁻, effects of Ca²⁺-activated Cl⁻ channels on membrane voltage and presynaptic [Cl⁻]_i appear to be small in cones pedicles. Thus, studies of amphibian photoreceptors have revealed a dichotomy between rods and cones with respect to presynaptic Cl⁻ dependent regulation: regulation in rod spherules is more effective than in cone pedicles. 

The search for the molecular identity of photoreceptor Ca²⁺-activated Cl⁻ channels has led to recent reports that the channel proteins anoctamin 1 (ANO 1, alias TMEM16A) and anoctamin 2 (ANO 2, alias TMEM16B) are expressed in the outer plexiform layer (OPL) of the amphibian and mammalian retina. ^4,9,10 Anoctamin proteins can form ion channels, and the isoforms ANO 1 and ANO 2 have been shown to be anion-selective. ^11–13 In salamander photoreceptors, ANO 1 immunoreactivity was detected in terminals of isolated rods and cones. ⁴ Patch-clamp recordings in the same study indicated that the Ca²⁺-sensitivities of Cl⁻ currents were different between rods and cones, suggesting that different Cl⁻ channels operate in the two photoreceptors. In the mouse retina, anoctamin expression was studied in the OPL, but differences between rod spherules and cone pedicles were not examined. ANO 2, a neuron-specific anoctamin isoform, was found to be a constituent of a presynaptic protein cluster, tethered by adaptor proteins to the ribbon structures of the photoreceptor terminals. ⁹ These findings indicate that anoctamin channels may be involved in the formation of presynaptic Ca²⁺-activated Cl⁻ channels in photoreceptors. They provide, however, no information about the specific functions that ANO 1 and ANO 2 serve in rods and cones. 

ANO 1 and ANO 2 display functional differences when expressed in cell lines. Their Ca²⁺ sensitivities differ 10-fold; ANO 2 requires greater than 1 μM free Ca²⁺ to open, ^14,15 while ANO 1 channels open at 0.05 to 0.3 μM Ca²⁺, depending on the splice variant of the protein. ¹⁶ Considering the rod–cone dichotomy described above, it is important to know which anoctamin isoforms operates in rods and which in cones. Here, we demonstrate that ANO 2 is exclusively expressed in rod spherules of the rat retina where it occurs as a retina-specific splice variant. ANO 1 protein is expressed in small blood vessels of the inner retina and in the retinal pigment epithelial cells, but we find no evidence for its expression in photoreceptors. We also confirm that the cation-driven chloride transporter NKCC1 is present in photoreceptor terminals. The transporter is coexpressed with its regulatory kinases STE20/SPS1-related proline/alanine-rich kinase (SPAK) and oxidative stress-responsive kinase 1 (OSR1), which are needed to drive Cl⁻ accumulation inside the terminals. Our findings suggest that ANO 2 channels contribute to Ca²⁺-dependent Cl⁻ currents in rod spherules, and that NKCC1-mediated Cl⁻ uptake supports Cl⁻ currents in these structures. Together, these proteins codetermine presynaptic Cl⁻ dynamics and control transmitter release in rod photoreceptors. In contrast, Ca²⁺-dependent Cl⁻ currents in cone pedicles appear not to be conducted by anoctamin channels. 

Methods

Animals and Tissue Preparation

Eyes were prepared from 24-week-old Wistar rats. Animals were anesthetized by isofluran inhalation and killed by decapitation. All experiments were performed in accordance with the Animal Protection Law, the guidelines and permissions of Heidelberg University, and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Eyes were excised, washed in PBS (130 mM NaCl, 8.1 mM Na₂HPO₄, 1.9 mM NaH₂PO₄, pH 7.4) and fixed in PBS containing 4% (wt/vol) paraformaldehyde (PFA) for 30 minutes on ice. After 3 × 15 minutes wash steps in PBS, the tissue was dehydrated with 10% saccharose (in PBS) for 2.5 hours followed by 30% saccharose over night at 4° C. Eyes were then embedded in Jung Tissue Freezing Medium (Leica Microsystems, Wetzlar, Germany) and stored at −20° C. Cryosections (20 μm) were prepared in a cryotome (CM 3050 S; Leica Microsystems) at −23° C and mounted on gelatine-coated SuperFrost glass slides (Fisher Scientific, Houston, TX), air dried, and stored at −20° C. 

Antisera and Lectins

To obtain reliable and selective immunosignals for ANO 1 and ANO 2 proteins, we generated polyclonal, isoform-specific antisera in guinea pigs. Immunogenic agents were fusion proteins containing the C-terminus of ANO 1, and the cytosolic domain connecting transmembrane segment 2 and 3 in ANO 2. ¹⁷ The specificity of these antisera was previously documented in nasal epithelia where ANO 1 antiserum labels microvilli of various epithelia, while ANO 2 antiserum specifically labels cilia in olfactory sensory neurons and microvilli in vomeronasal sensory neurons. ^17,18 For Western blot analysis, a polyclonal peptide antibody against ANO 2 was used, which was also described before. ¹⁷ The antiserum against ANO 10 (C-terminus of ANO 10 encoding amino acids 605 to 688 [GenBank XM_236774]) was generated in the same way. The reactivity and specificity of the purified antibodies were demonstrated by several independent methods. ¹⁷ The antiserum directed against plasma membrane calcium ATPase (PMCA) was obtained from Thermo Scientific (cat. MA3-914, clone 5F10, dilution 1:80; Waltham, MA) and was previously characterized in retina. ^19,20 The CtBP2/RIBEYE antiserum (cat. 612044, dilution 1:80; BD Biosciences, Heidelberg, Germany; http://antibodyregistry.org/AB_399431) has previously been tested in various tissues, including retina. ²¹ The monoclonal NKCC1 antiserum T4 was obtained from the Iowa Developmental Studies Hybridoma Bank (dilution 1:20; Iowa City, IA; http://antibodyregistry.org/AB_528406) and was previously tested for cross reactions in the retinae of wild-type and NKCC^−/− mice. ²² Affinity-purified polyclonal goat antisera raised against SPAK (raised against the human HA-tagged full-length protein; cat. AB-173, dilution 1:50; Kinasource, Dundee, UK) and OSR1 (raised against a 10–25 aa immunizing antigenic peptide mapping between aa residues 475 and 525 of human origin; sc-67721, 1:50; Santa Cruz Biotechnology, Santa Cruz, CA) have previously been tested on olfactory epithelium. ²³ Vimentin antiserum was kindly provided by Harald Herrmann-Lerdon, PhD (DKFZ, Heidelberg, Germany), ²⁴ and used at a dilution of 1:50. Secondary antibodies were: Alexa Fluor 488 goat antiguinea pig (A11073; Molecular Probes, Darmstadt, Germany); Alexa Fluor 568 goat antiguinea pig (A11075; Molecular Probes); Alexa Fluor 568 goat antimouse (A11004; Molecular Probes); Alexa Fluor 488 goat antimouse (A21121; Molecular Probes); Alexa Fluor 568 donkey antisheep (A21099; Molecular Probes); Alexa Fluor 568 donkey antigoat (A11057; Molecular Probes); Alexa Fluor 488 donkey antirabbit (A21206; Molecular Probes); Alexa Fluor 405 goat antimouse (A31553; Molecular Probes). Peanut agglutinin (PNA) a lectin that specifically stains cone photoreceptors, ²⁵ was used at a concentration of 100 μg/mL, conjugated with Alexa Fluor 568 (L-32458; Molecular Probes) or with Alexa Fluor 488 (L-21409; Molecular Probes). Griffonia simplicifolia lectin, a marker for endothelial cells in retinal blood vessels, ^26,27 was used as TRITC conjugate (L5264-2MG; Sigma-Aldrich, Taufkirchen, Germany) at 500 μg/ml. 

Immunohistochemistry

Glass slides with cryosections were transferred to room temperature and air dried. Sections were postfixed with 4% PFA (in PBS) for 5 minutes, washed 3 × 5 min in PBS, and incubated for 1 hour in CT solution (PBS containing 5% vol/vol ChemiBlocker [cat. 2170; Merck Millipore, Billerica, MA] and 0.5% wt/vol Triton X-100). Primary antisera were then applied in CT solution for 2.5 hours, followed by 3 × 5 min washing in PBS. Secondary antisera were then applied for 90 minutes in C solution (CT solution without Triton X-100), together with lectin solution if indicated. Sections were then washed 3 × 5 minutes in PBS, nuclei were stained with 0.3 μM 4,6-Diamidin-2-phenylindol in PBS (DAPI; Life Technologies, Darmstadt, Germany), and washed 3 × 5 min in PBS. Finally, sections were embedded in nonfluorescent mounting medium (Aqua-Poly/ Mount, cat. 18606; Polyscience, Eppelheim, Germany) under a cover slip. Sections were stored at 4° C in the dark. Fluorescence images were obtained with a confocal microscope (Nikon Eclipse 90i/C1; Nikon, Düsseldorf, Germany) and analyzed with Nikon software (NIS elements, version 4.0; Nikon). Optical sectioning was 3.2 to 3.5 μm in individual confocal images. The thickness of image composites representing Z-stacks is specified in the figure legends. 

Expression Analysis by RT-PCR

Eyes and olfactory epithelia were dissected from several-week-old wistar rats, homogenized using a glass pistil, and mRNA was extracted using the Magnetic mRNA Isolation Kit (New England BioLabs, Ipswich, MA). cDNA was synthesized using 200 ng mRNA, random hexamer primers, and RevertAid Premium Reverse Transcriptase (Thermo Scientific). PCR amplification was performed on 0.2 ng single-stranded cDNA with 2U Taq DNA polymerase (Axon, Kaiserslautern, Germany) using the primer pairs listed in the Table. To compare cDNA sequences between retinal and olfactory ANO 2 channels, we used RT-PCR primers that overlap rat exon 15 (corresponding to human exon 13¹⁶) and were designed to indicate the presence (segc-plus; forward: AACGTTCCCAGGAACACT) or absence (segc-minus; forward; GAAGAAGAACACTCCCG) of segment c (reverse: CAATGGAGAACGTCAGGGC). Cycling conditions were 94° C for 3 minutes, 94° C for 30 seconds, 60° C for 20 seconds, 72° C for 30 seconds for 30 to 34 cycles, respectively, and 72° C for 8 minutes. Controls were performed on the cloned olfactory and retinal ANO 2 splice variants. The bands were resolved by gel electrophoresis and were verified by sequencing. 

Table

 

Primer Pairs Used for RT-PCR Experiments

TMEM16a forward	CCATCAGAGCCAAAGACATCGG
TMEM16a reverse	GGGCTGTGGGACTGTGGTTGTTA
TMEM16b forward	GCCCTCCTCAACAATGTCATCGAA
TMEM16b reverse	GATCTGGTCGCTGATGTCTGTCGG
TMEM16c forward	GAACATTCATGGGCTGATGGACGA
TMEM16c reverse	CGGAGGCCCTTTGGTATATCTG
TMEM16d forward	AGCACCGCTTCTGGCCTTACTG
TMEM16d reverse	TCTCTCCTCATCCGGTCCCTTAGA
TMEM16e forward	CGTGGGATATCCTGGAAGGTACAC
TMEM16e reverse	TGTGGTTGGGAAAGTCAGCTATCA
TMEM16f forward	GGGCAGGCTGGGATTGTTCTATGA
TMEM16f reverse	ATCCGGAATTGCGTACGAAATG
TMEM16g forward	TGAACTTCTACGCCTCGCCTGT
TMEM16g reverse	GCGCGTCCAGCTGTAGTAGAC
TMEM16h forward	GATCAGACAAGCCGGGATGTT
TMEM16h reverse	CATCTCTTTCAGGCGGTCCA
TMEM16j forward	CGCCCACTTCTCCTCGCTCATT
TMEM16j reverse	TTCTTTCTGCACGGGCCATAAT
TMEM16k forward	TCGCCTCGCTCTTCTACATTGC
TMEM16k reverse	CTGGATGTGGCGTGGTTTATCAG
NKCC1 forward	CGAATTATTGGAGCCATTACAGT
NKCC1 reverse	ACATCTGGAAAGCTGGGTAGATA
SPAK forward	ATGGCGACTGGGAGTGGAGT
SPAK reverse	TTGCTGGGATTTGGGAGTGAG
OXSR1 forward	GGGAGTGGAGCGACGATGAGT
OXSR1 reverse	AGAAGGGCGGAGGCAAGAAAA
PMCA1 forward	GCCATGGAGATGCAGCCTCTGA
PMCA1 reverse	ATTTGGCGGGATGGCTTCTGG
CtBP2 forward	AAGGCACGCGGGTACAAAGC
CtBP2 reverse	CCTGTGATTGCTCGGCGGAT

Western Blot Analysis

Eyes were dissected from several-week-old wistar rats, freezed in liquid nitrogen and stored at −80° C. Frozen tissue was homogenized in SDS-buffer (313 mM Tris, pH 6.8; 10% SDS; 0.05% Bromphenolblau; 50% Glycerol; 0.1 M Dithiothreitol [DTT]), boiled at 100° C for 10 minutes and centrifuged at 12,000g for 20 minutes. Proteins in the supernatant were separated by SDS-PAGE on 10% gels. After separation, proteins were electrophoretically blotted on polyvinylidene difluoride (PVDF) membranes. Membranes were then blocked in 5% milkpowder (in PBS/1% Triton X−100) for 1 hour and afterwards incubated with primary antibodies for at least 4 hours. After 3 times of washing with PBS, the secondary antibody was added and incubated for 1.5 hours. Blots were washed again and the chemiluminescence was detected via ECL Plus Western Blotting Detection Reagents (GE Healthcare, München, Germany). 

Results

Expression of Anoctamin Genes in the Rat Retina

The family of anoctamin proteins has 10 members, currently identified as anoctamin 1 (ANO 1, alias TMEM16A) to anoctamin 10 (ANO 10, alias TMEM16K). ²⁸ A semiquantitative RT-PCR analysis on cDNA obtained from retina yielded signals for retinal marker messages that encode plasma membrane Ca²⁺-ATPase (PMCA1) and C-terminal binding protein 2 (CtBP2) (Fig. 1A) as well as three anoctamin signals, ANO 1, ANO 2, and ANO 10 (Fig. 1B). The strongest RT-PCR signal was obtained with ANO 2. Ca²⁺-activated Cl⁻ channels are often coexpressed with Cl⁻ transporters that mediate Cl⁻ accumulation. For example, the Na⁺-K⁺-2Cl⁻ cotransporter NKCC1 and its activating kinases OSR1 and SPAK are coexpressed with Ca²⁺-activated Cl⁻ channels in olfactory receptor neurons ²³ and somatosensory neurons. ²⁹ We, therefore, looked for their expression in retina and detected all three messages (Fig. 1C). A previous study demonstrated that ANO 2 exists in two splice forms, one, the olfactory variant, lacking four amino acids encoded by exon 15. ¹⁵ We checked for the occurrence of the two splice forms in the eye, using primers that reported the presence or absence of exon 15. The results confirmed that retinal ANO 2 contains exon 15 and olfactory ANO 2 does not (Fig. 1D). These data show that only three members of the anoctamin family are expressed on RNA level in the rat eye: ANO 1, ANO 2, and ANO 10. 

Figure 1

 

Transcription of anoctamin genes in rat eye. (A) RT-PCR detection of two retinal marker proteins in cDNA prepared from rat eye tissue: plasma membrane calcium ATPase (PMCA) and C-terminal binding protein 2 (CtBP2). (B) RT-PCR detection of anoctamin isoforms ANO 1, 2, and 10 in cDNA from rat eye tissue. Anoctamin proteins ANO 1 to 10 correspond to TMEM16A to K in alternative nomenclature. ⁶⁰ (C) RT-PCR detection of NKCC1 and its activating kinases SPAK and OSR1 in cDNA from rat eye tissue. (D) RT-PCR detection of ANO 2 splice variants that either include (+) or exclude (-) exon 15. The suitability of primer pairs was confirmed in HEK 293 cells transfected with either of the two splice variant (ctrl). cDNA from rat eye contained exon 15, while cDNA from rat olfactory epithelium (OE) did not. (E) Western blots obtained from rat eye tissue demonstrate the specificity of the guinea pig antisera for the anoctamin antigenes of ANO 1, ANO 2, or ANO 10, which were used for immunization. (F) Immunocytochemical staining with the guinea pig antisera of HEK 293 cells transfected with the respective anoctamin protein, carrying a C-terminal YFP label. Scale bar: 10 μm.

Subcelluar Localization of Retinal Anoctamin Proteins

To examine the expression of anoctamin proteins in the retina, we raised isoform-specific antisera directed against ANO 1, ANO 2, and ANO 10 in guinea pigs. The antisera stained single bands at the expected sizes of ANO 1 (115 kDa) and ANO 2 (110 kDa) in Western blots from rat eyes (Fig. 1E). The ANO 10 signal ran somewhat higher than predicted (100 kDa instead 77 kDa). In HEK 293 cells transfected with any of the three anoctamin proteins, each protein tagged with a C-terminal yellow fluorescent protein (YFP), the guinea pig antisera specifically labeled transfected cells (Fig. 1F). No cross-staining was observed, demonstrating the isoform-specificity of each antiserum. To localize retinal anoctamin protein expression sites on a cellular level, we used the guinea pig antisera for immunohistochemistry on retinal cryosections. Our ANO 1 antiserum labeled tubular structures running mostly horizontally along the upper and lower margin of the inner nuclear layer (INL; Figs. 2A, 2B). These structures are localized distinctly underneath the OPL. The most prominent presynaptic structure in the OPL is the ribbon complex that can be labeled with antisera raised against CtBP2, a protein that corresponds to the B-domain of RIBEYE, the main ribbon-forming protein. ³⁰ The ANO 1–positive structures are located below the CtBP2-positive layer, and were identified as retinal blood vessels of the capillary layer ³¹ by costaining with G. simplicifolia lectin (Fig. 2D). Furthermore, a prominent hexagonal ANO 1 immunosignal was discernible in the layer of the pigment epithelium (PE; Fig. 2C). The shape of this signal indicates that it originates from the apical junctional complex, which forms the interface between the PE and the photoreceptor outer segments. ³² In the photoreceptors themselves, no ANO 1 signal was detectable on any cellular level. 

Figure 2

 

ANO1 is not expressed in photoreceptors. (A) ANO 1-positive structures in the INL of the rat retina. Visualization of the ribbon complex in photoreceptor synapses with CtBP2. Arrows point to blood vessels. (B) ANO 1-positive structures at higher magnification. Costaining with CtBP2 antiserum shows the localization of ANO 1–positive structures relative to photoreceptor synapses. ANO 1 and CtBP2 are expressed in two distinct strata. (C) ANO 1 immunosignals in the RPE and INL. Arrow points to basolateral membranes of the PE. (D) Costaining with G. simplicifolia lectin identifies ANO 1–positive structures as blood vessels. Blue nuclear stain: DAPI. Scale bar: 15 μm (A–C), 5 μm (D). ONL, outer nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer; Images (B) and (C) are Z-stacks over 17 μm and 15 μm, respectively.

The immunosignal of ANO 2 was prominent in the OPL, the site of photoreceptor terminals. ANO 2 signals colocalized with the plasma membrane Ca²⁺ ATPase, a protein that is preferentially expressed in the presynaptic membrane of rod spherules in the OPL ²⁰ (Figs. 3A–C). We identified no retinal expression site of ANO 2 outside the OPL, including the PE. Double-stainings of ANO 2 protein with a CtBP2 antiserum revealed that ANO 2 is expressed in close proximity to the ribbon complex (Fig. 4A). However, the ANO 2 immunosignal was not as clustered as the CtBP2 signal, but appeared spread over the surrounding presynaptic microstructures (Fig. 4B), comparable to the PMCA-immunosignals shown in Figure 3C. This expression pattern suggests that the channels are not restricted to the ribbon complex itself, but that they are present in the surrounding presynaptic membrane of the photoreceptor terminals. 

Figure 3

 

ANO 2 is expressed in photoreceptor synaptic terminals. (A) Bright-field image of rat retina cryosection. (B) ANO 2 is exclusively expressed in the OPL together with PMCA. (C) Complete colocalization of ANO 2 and PMCA in the OPL. Blue nuclear stain: DAPI. Scale bar: 25 μm (A, B), 5 μM (C). OS/IS, outer and inner segments. Image (C) is a Z-stack over 5 μm.

Figure 4

 

Nonclustered ANO 2 expression in presynaptic structures. (A) Comparison of ANO 2 and CtBP2 immunosignals reveals colocalization in the same stratum of the OPL. Scale bar: 10 μm. (B) Higher magnification shows that ANO 2 is not clustered with CtBP2, but is also present in the surrounding presynaptic structures. Scale bar: 10 μm. Blue nuclear stain: DAPI. Image (B) is a Z-stack over 16 μm.

No ANO 10 immunosignal was detected in the retina. ANO 10 expression was either too weak to be detected with our antiserum or restricted to extra retinal tissues that were not examined here. 

ANO 2 Expression Is Restricted to Rod Spherules

The OPL contains synaptic structures of rod and cone photoreceptors. To distinguish between them, we used the fluo-labeled lectin PNA, which specifically binds to surface glycoproteins of cone photoreceptors. ²⁵ Figure 5A shows PNA signals highlighting the inner segments as well as the pedicles of cone photoreceptors in rat retina. Costaining with ANO 2 displays both signals in contiguous layers within the OPL (Fig. 5B). At higher resolution, mutually exclusive expression patterns are discernible (Figs. 5C, 5D). The PNA signals occur in regions of the OPL, which are ANO 2 negative. These regions probably represent the cone pedicles, which are surrounded by a meshwork of ANO 2 positive rod spherules. This staining pattern was not only observed in rat retina. In mouse retina, PNA signals were also located in a layer underneath the ANO 2 signal, suggesting separate expression in rod and cone terminals (Figs. 5E, 5F). As in rat, ANO 1 was also absent from the OPL in mouse retina and could only be detected in blood vessels (Fig. 5G). To obtain a clearer image of the synaptic colocalization of ANO 2 in rat rod and cone terminals, we triple-labeled retinal cryosections for ANO 2, CtBP2, and PNA. Two distinct layers of ribbon structures are visible in these images (Figs. 6A–C). At the bottom layer of the OPL, ribbon structures were associated with PNA, but devoid of ANO 2. In the upper OPL regions, ribbon structures were surrounded by ANO 2 immunosignals, but were devoid of PNA. Taken together, these data suggest that ANO 2 channels are only present in rod spherules and not in cone pedicles. 

Figure 5

 

ANO 2 is expressed in rod spherules, but not in cone pedicles. (A) Visualization of cone photoreceptors in the rat retina with fluorescence-labeled PNA shows OS/IS, as well as cone pedicles in the OPL. Arrows point to the OS/IS and a pedicle of a single cone photoreceptor. (B) ANO 2 immunosignals are restricted to the OPL. Scale bar: 15 μm. Blue nuclear stain: DAPI. (C, D) Higher magnification reveals that PNA-positive regions in the images are ANO 2-negative, indicating that ANO 2 is absent from cone pedicles. The surrounding ANO 2-positive regions are rod spherules. Scale bar: 15 μm. (E, F) Costaining of ANO 2 and PNA in mouse retina illustrates the segregation of OPL layers containing rod spherules (green, ANO 2-positive) and cone pedicles (red, PNA-positive). Scale bar: 10 μm. (G) ANO 1 immunosignals originating from blood capillaries in the mouse retina. Scale bar: 15 μm. Images in (C, D, G) are Z-stacks over 17 μm.

Figure 6

 

Stratification of ANO 2 signals in the OPL. (A–C) Cone pedicles are labeled with PNA and reside in a stratum distinct from the ANO 2-positive stratum of rod spherules. All photoreceptor terminals are visualized through CtBP2 immunosignals. Scale bar: 15 μm. Z-stack over 8 μm.

Expression of the Cl⁻ Transporter NKCC1 in Photoreceptor Terminals

Staining for the Na⁺-K⁺-2Cl⁻ cotransporter NKCC1 produced a punctate pattern in the OPL together with a filamentous pattern in the inner retina. The latter signal corresponds to the macroglial Müller cells, as demonstrated by colocalization with vimentin, a Müller-cell marker in the mammalian retina ³³ (Fig. 7A). The punctate NKCC1 pattern in the OPL did not match the vimentin signal (Fig. 7B), but colocalized with CtBP2 (Fig. 7C), indicating that the NKCC1 density is particularly high at the ribbon synapses of the photoreceptors. NKCC1 immunosignals appeared indeed more clustered than ANO 2 (Fig. 7D), suggesting an accumulation of NKCC1 protein close to the ribbon complex. OPL immunosignals were also obtained with two other antisera (C-14 and S763B ²³ ) directed against NKCC1 (data not shown). In contrast to the granular appearance of NKCC1 immunosignals, the protein kinases that regulate NKCC1 activity, SPAK and OSR1, were found largely colocalized with ANO 2 in the OPL without pronounced clustering (Figs. 7E, 7F). NKCC1 expression appeared to be uniformly distributed in the OPL, and no segregation into rod and cone terminals was discernible. In accordance with earlier studies from adult mammals, ^22,34 our results demonstrate the presence of the chloride accumulation machinery in the OPL, consisting of NKCC1, SPAK, and OSR1 in photoreceptor terminals where it may support the Ca²⁺-induced Cl⁻ currents characterized previously. ^1,3,35  

Figure 7

 

Expression of Cl⁻ transporters in photoreceptor terminals. (A) NKCC1 expression shows a punctate pattern in the OPL and can also be detected in Müller glial cells, which are colabeled with a vimentin antiserum. (B) The punctate NKCC1 pattern in the OPL does not originate from Müller cells, as it is not colocalized with vimentin. (C) NKCC1 and CtBP2 colocalize at the ribbon complex in photoreceptor terminals. (D) A triple staining for NKCC1, CtBP2, and ANO 2 illustrates that NKCC1 appears more clustered around the ribbon complex than ANO 2. (E, F) The protein kinases SPAK and OSR1 that regulate NKCC1 activity are colocalized with ANO 2 in the OPL. Blue nuclear stain: DAPI. Scale bar: 25 μm (A, D–F), 10 μm (B, C). Images in (A, B, C, and D) are Z-stacks over 16 μm, 16 μm, 17 μm, and 19 μm, respectively.

Discussion

Our data demonstrate the transcription of ANO 1, ANO 2, and ANO 10 in rat eye tissue and the localization of ANO 1 and ANO 2 protein in the retina. It was recently shown that ANO 2 is expressed in the mouse OPL where it colocalized at the ribbon synapse with various adaptor proteins. ⁹ Following this lead, we have found that ANO 2 is expressed specifically in rod terminals, and that it is absent from cone terminals. In the light of substantial differences between rods and cones with respect to the handling of Ca²⁺ and Cl⁻, insights that were obtained mainly from studies of nonmammalian photoreceptors, our data suggest that ANO 2 channels serve a particular role for scotopic vision. 

Differential Expression of Anoctamin Channels in the Rat Retina

We raised antisera against ANO 1 and ANO 2 in guinea pigs and tested them extensively for cross reactivity. ¹⁷ Our results from the rat retina confirm the presence of ANO 2 in the OPL, but they do not match previously published data of ANO 1 expression in the retina. We found ANO 1 immunosignals associated exclusively with blood capillaries and with the PE. Signals were colocalized with an endothelial marker and not with the Müller cell marker vimentin. This suggests that ANO 1 is expressed in the capillary wall and not in perivascular end feet of the Müller cell. ³⁶ We could not detect any ANO 1 signals in photoreceptors, while two recent studies demonstrated ANO 1 immunoreactivity in the retinae of tiger salamander ⁴ and mouse. ¹⁰ In both studies, the same antiserum was used (ab53212, a rabbit polyclonal antiserum; Abcam, Cambridge, UK), and similar expression patterns were reported. ANO 1 stained the OPL, GFAP-positive Müller cells in the INL, as well as some cells in the ganglion cell layer. ANO 1 staining also colocalized with the synaptic marker Synaptic Vesicle Protein 2 (SV2) in terminals of isolated salamander rod and cone photoreceptors. ⁴ The discrepancy between our data and these published results may be attributed to differences between species (salamander versus rat) or between antibody properties and staining methods, including procedures for antigen retrieval. In our hands, the Abcam antiserum produced no specific signals in rat retina without antigen retrieval. After such treatment (1% SDS or 5 M urea), we observed a weak signal in the OPL and in the retinal capillaries (data not shown). Published data from the mouse retina, obtained with this antiserum after boiling cryosections in 10 mM citrate or 10 mM Tris for antigen retrieval, produced signals in the OPL, in scattered cells of the inner retina, but not in blood vessels. ¹⁰ In contrast, the guinea pig antiserum used in our study produced the same ANO 1 staining pattern (blood vessels, PE, but not the OPL) with and without antigen retrieval. It is conceivable that the ANO 1 expression density in photoreceptor terminals is low and difficult to detect. In that case, we may assume that both rods and cones possess ANO 1 channels in their terminals, but that only rod photoreceptors express, in addition, ANO 2 channels at high density. This notion would be consistent with our finding that the retina contains significantly less ANO 1 mRNA than ANO 2 mRNA (Fig. 1B). It would also match earlier evidence for the presence of more than one kind of chloride channels in rods, ⁷ and would be consistent with the well established presence of Ca²⁺-activated chloride currents in cones. ^3,35,37,38 The question of ANO 1 expression in photoreceptor terminals remains to be settled by further experimentation, while the expression of ANO 2 in rod terminals is confirmed. 

Our data demonstrate the presence of a special splice variant of ANO 2 in rod photoreceptors and its targeted expression in the synaptic terminals. At present the functional significance of the alternative splicing between the olfactory and the visual system is not understood. Exon 15 (corresponding to exon 13 in human ANO 1¹⁶), the segment present in photoreceptors but not in olfactory receptor neurons, encodes the amino acid sequence ERSQ within the intracellular loop that connects transmembrane regions 2 and 3 of the ANO 2 channel. The corresponding segment in ANO 1 has been shown to contribute to voltage dependence and Ca²⁺ sensitivity of channel gating. ^16,39 In particular, the segment is necessary for the high Ca²⁺-sensitivity of ANO 1. The ERSQ sequence in the photoreceptor ANO 2 protein may play a similar role. Transmitter release from rod photoreceptor terminals occurs at significantly lower Ca²⁺ concentrations than in other synapses. And the ANO 2 splice variant that includes exon 15 may be an adaptation to these conditions. 

Unlike the presynaptic Ca²⁺ channels, which are closely attached to the ribbon synapse, ⁴⁰ ANO 2 expression appears broadly distributed across the rod terminal membrane. This uniform pattern resembles the expression of PMCA, the Ca²⁺ pump that extrudes Ca²⁺ from the terminal lumen. ²⁰ PMCA has a high Ca²⁺ affinity and appears to keep the intraterminal free Ca²⁺ concentration within the range of 0.3 to 2 μM, ^41,42 which matches the operating range of ANO 2. This may indicate that most ANO 2 channels in the terminal are controlled by the bulk Ca²⁺ concentration and not by local Ca²⁺ peaks near the ribbon complex. Chloride effects on transmitter release are, therefore, expected to be slow, reflecting the Ca²⁺ dynamics in the lumen of the synaptic terminal. The temporal properties of Cl⁻ accumulation are the second major factor that determines how rapidly Cl⁻ effects may develop. The speed of Cl⁻ uptake into the terminals has not yet been measured. But earlier immunohistochemical and pharmacologic evidence indicate that photoreceptor terminals use NKCC1 to accumulate intracellular Cl⁻. ^22,43 Moreover, a recent study demonstrated that NKCC1 within the OPL is phosphorylated upon illumination of the retina. ⁴⁴ The electroneutral Na⁺-K⁺-2Cl⁻ cotransporter is dynamically regulated by a phosphorylation/dephosphorylation equilibrium. ⁴⁵ Three threonine residues in the C-terminal region of the NKCC1 protein serve as targets for phosphorylation by the kinases SPAK and OSR1. Our data demonstrate that both kinases are colocalized with NKCC1 in the OPL. Activation of NKCC1 tends to raise the intraterminal [Cl⁻]_i, providing the elevated Cl⁻ levels (30–50 mM⁶) that support Cl⁻ efflux through Ca²⁺-activated Cl⁻ channels. Rod photoreceptor terminals, thus, possess the entire molecular machinery that is needed for Cl⁻ accumulation and for the generation of Ca²⁺-dependent Cl⁻ currents carried by efflux of Cl⁻ ions. An interesting finding is that the amount of phosphorylated NKCC1 in the OPL increases upon light adaptation and decreases in the light-adapted retina. ⁴⁴ The significance of this light-controlled phosphorylation is not yet understood, mainly because its time course has not yet been established. But the NKCC1-phosphorylation data suggest that intraterminal [Cl⁻]_i is higher in light than in darkness. 

A Possible Role for ANO 2 Channels in the Regulation of Transmitter Release From Rod Photoreceptors

The synaptic transfer of light-induced signals from rod photoreceptors to horizontal cells and bipolar cells in the OPL displays linear transmission characteristics, ⁴⁶ a property that enables a proportional response of the inner retina to small changes of light intensity. ⁴⁷ Apparently, this linearity results from a near-linear dependence of glutamate release on the membrane voltage of dark-adapted rod photoreceptors within the response range between −40 mV and −55 mV. ⁴⁸ The dynamics of presynaptic Ca²⁺ signaling, which involve both L-type Ca²⁺ channels ^47,49 and Ca²⁺ release from intracellular stores, ^50,51 are not completely understood. But transmitter vesicles with a near-linear Ca²⁺-dependence within the Ca²⁺ range of 0.5 to 3 μM are present and may preserve linearity between membrane voltage and exocytosis. ⁵² Ca²⁺-activated Cl⁻ channels with a relatively broad range of Ca²⁺ sensitivities (half-maximal activation requires ∼ 0.6 μM Ca²⁺ in salamander rods ⁴ and 1–3 μM Ca²⁺ in heterologously expressed ANO 2 ^14,15 ) are active in the dark-adapted photoreceptor when the free Ca²⁺ concentration in rod terminals exceeds 1 μM. ^20,41,53,54 As demonstrated by the Thoreson group, the Ca²⁺-induced Cl⁻ conductance mediates a decline of [Cl⁻]_i in the rod terminals and, hence, inhibits voltage-gated L-type Ca²⁺-channels, limiting presynaptic Ca²⁺-influx in the dark. ^1,6,7,55 ANO 2 channels are good candidates for this function as they are specifically expressed in rod terminals. Any depolarization in excess of the resting voltage of −40 mV causes substantial activation of L-type Ca²⁺ channels and, consequently, increased Ca²⁺ influx and transmitter release. The feedback inhibition mediated by ANO 2 counteracts such a depolarization as decreasing Cl⁻ levels progressively suppress Ca²⁺ influx. In this way, ANO 2 channels appear to contribute to the control of transmitter release in the dark, supporting other inhibitory factors like Ca²⁺ channel inactivation and Ca²⁺ depletion in the synaptic cleft. ⁵⁶ These control mechanisms may help to explain the observation that [Ca²⁺]_i is lower in rod terminals than in cone terminals, ^57,58 and that vesicle release in the dark is slower in rods than in cones. ^58,59 At increasing light intensity, when Ca²⁺ channels are switched off by hyperpolarization, the phosphorylation of NKCC1 may reinforce Cl⁻ uptake and may help to refill the rod terminal with Cl⁻, so that the feedback mechanism is operative again upon return to the dark. 

Thus, ANO 2 channels and NKCC1 transporters appear to play a role in defining the response characteristics of rod photoreceptors on the level of transmitter discharge. As there is evidence that more than one type of Cl⁻ channel operate in rod photoreceptors, ⁷ it will be exciting to see which specific alterations can be discovered in the performance of the rod system of mice with ablated Ano 2 genes. ¹⁰ Examination of these animals may reveal the significance of ANO 2 channels for scotopic vision. 

Acknowledgments

The authors thank Frank Müller for helpful discussions. They also thank Kerstin Vocke for providing the anoctamin expression plasmids. 

Supported by Deutsche Forschungsgemeinschaft Grants MO 1384/2 (FM) and FR 937/15 (SF). 

Disclosure: K. Dauner, None; C. Möbus, None; S. Frings, None; F. Möhrlen, None