Kv2.1/Kv8.2 Channels Regulate Fluid Homeostasis in the Outer Retina

Abstract

Purpose:

Photoreceptor Kv2.1/Kv8.2 voltage-gated potassium channels carry an outward potassium current, helping to set the resting membrane potential and to shape dim light responses. Because potassium flux in the outer retina influences extracellular osmolarity and fluid distribution, we hypothesized that Kv2.1/Kv8.2 channels also contribute to fluid homeostasis in this region of the retina.

Methods:

OCT imaging was performed in Kv8.2 heterozygous (Het) and knockout (KO) mice aged 4–7 weeks under dark- and light-adapted conditions. Light-dark differences in the distance between the external limiting membrane (ELM) and retinal pigment epithelium (RPE) (ΔELM-RPE) were calculated to quantify light-evoked expansion of the subretinal space (SRS). As a secondary outcome, outer nuclear layer (ONL) thickness was also measured under both lighting conditions. Retinal gene expression was assessed by RNA-seq and droplet digital RT-PCR. Retinal protein expression was determined by western blotting and immunolabeling.

Results:

ΔELM-RPE was significantly reduced in Kv8.2 KO mice compared with Het controls, indicating reduced SRS hydration. ONL thickness exhibited a small but significant light-dark change that was different between genotypes. Transcriptomic analyses revealed upregulation of osmosensitive genes, including osmolyte transporters and aquaporins. AQP1 protein expression in photoreceptors increased.

Conclusions:

These findings reveal a previously unrecognized role for Kv2.1/Kv8.2 channels in outer retinal fluid homeostasis and support a model in which photoreceptor potassium efflux contributes to osmotic water movement into the subretinal space.

Introduction

Kv2 voltage-gated potassium channels are widely expressed and conduct delayed rectifier currents in many excitable cells. In photoreceptors, Kv2.1 assembles with the regulatory subunit Kv8.2 to form heterotetrameric channels that carry the potassium current I_Kx.^1,2 I_Kx constitutes the major component of the outward arm of the circulating dark current, contributing to the establishment of the resting membrane potential and shaping dim light responses. Loss of Kv8.2 disrupts I_Kx causing cone dystrophy with supernormal rod responses (CDSRR), also known as KCNV2 retinopathy.³ This inherited retinal disease is characterized by delayed and reduced ERG responses, early loss of central cone vision with macular atrophy, and variable loss of night vision. In mice, loss of Kv8.2 causes delayed and reduced ERG responses characteristic of CDSRR, and rod photoreceptors undergo a slow degeneration.^4–7 It is not known if the photoreceptor degeneration associated with loss of Kv8.2 is a direct result of disrupted photosignaling or if there are additional physiological consequences to altered potassium homeostasis in the outer retina.

Another role for potassium flux in the outer retina besides regulating photoreceptor membrane voltage is to regulate osmotic fluid uptake from the subretinal space (SRS). The subretinal space constitutes a restricted extracellular microenvironment that is the interface between photoreceptors and the RPE. There is substantial net flux of fluid from the neural retina, through the SRS, and across the RPE into the choroidal circulation as this is the route for the removal of excess water produced by the prodigious metabolic rate of photoreceptors.^8–15 The process of fluid absorption from the SRS by the apical RPE is coupled to the concentration of potassium in the SRS through the activity of the Na⁺- K⁺- Cl⁻ cotransporter (NKCC1).^16,17 Unlike mechanisms governing water clearance from the SRS, the processes that facilitate movement of water from photoreceptors into the SRS are less well studied.

Kv2.1/Kv8.2 channels are positioned along the photoreceptor inner segment plasma membrane so that they carry a flow of potassium from photoreceptors into the SRS. Therefore, it is possible that potassium flux through these channels contributes to drawing excess water out of photoreceptors into the SRS. This idea predicts that in Kv8.2 knockout (KO) mice, photoreceptors would retain water, and the SRS would become relatively dehydrated.

We previously demonstrated that light-dark differences in hydration of the outer retina can be readily assessed noninvasively using an OCT-imaging based metabolic biomarker, in which the distance from the ELM to the RPE is measured under different lighting conditions.¹⁸ The volume of the SRS is regulated by retinal pH such that fluid uptake by the RPE is stimulated in the dark reducing the volume of the SRS and therefore the distance measured from the ELM to RPE. In light adapted animals, pH stimulation and fluid removal slows, so there is a net accumulation of fluid causing the ELM-RPE distance to expand. Therefore, in this study, we quantified paired light-dark ELM-RPE changes in Kv8.2 knockout mice using OCT to measure hydration of the SRS. We also assessed activation of osmosensitive pathways at the transcript and protein levels. We found that Kv8.2 knockout mice exhibit reduced light-evoked SRS expansion, upregulation of osmosensitive genes, and increased AQP1 protein expression. Together, these findings support a role for Kv2.1/Kv8.2-mediated potassium flux in maintaining fluid homeostasis within the outer retinal microenvironment.

Materials and Methods

Animals

Kv8.2 KO (Kcnv2^−/−) mice (MGI:7660814) were maintained in the laboratory. Mice were housed in a central vivarium, maintained on a standard 12/12-hour light/dark cycle, with food and water provided ad libitum in accordance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. Routine genotyping was performed by Transnetyx. Kv8.2 wildtype (WT) and heterozygous (Het) animals were indistinguishable in terms of ERG responses and rate of retinal degeneration, so either WT or Het were used as littermate controls for Kv8.2 knockout (KO) in different experiments (Supplemental Fig. S1–S2). A mix of male and female mice were used in all experiments. For experiments requiring anesthesia, a mixture of ketamine (87.5 mg/kg) and xylazine (2.5 mg/kg) was used with body temperature maintained by keeping animals on a heating pad. Eye drops to dilate the pupils (1% tropicamide) and maintain hydration (Genteal) were applied. All procedures were approved by the University of Iowa IACUC committee and adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

Optical Coherence Tomography (OCT)

Mice were dark-adapted overnight, anesthetized, and the OD eye alternatively imaged either with very dim red lighting (< 1 lux, “dark”) or under room lighting (~800 lux) after one hour of light-adaptation. The animals were allowed to rest for one week to recover from anesthesia and were then imaged again in the opposite lighting condition between postnatal days 33–41 (average age of imaging was 5 weeks). A total of 6 Kv8.2 Het and 11 Kv8.2 KO animals (representing 3 different litters) were used. Images were collected with a Bioptigen spectral-domain imaging system (model #75–10023 B00, Bioptigen, Inc.) as previously described.⁷ The central 100 B scans for each image file were then sent to Wayne State University for image analysis.

A first-pass rigid body registration with RNiftyReg (a function in R) was used to rotate the image and interpolate the signal at each pixel. Next, non-rotational rigid body approaches (at the level of a given row or column of pixels) were applied three times. The 100 images were then visually compared as a final step before averaging. After registration, layer segmentation boundaries were estimated using a machine learning model–based computer program, as described previously.¹⁹ Boundaries were first estimated with a previously described U-net convolutional neural network trained using the “dice loss” function and the Adam optimizer (learning rate = 0.001), with 665 previously labeled images for training and 166 and 356 images for validation and testing, respectively. Model predictions were then postprocessed by applying a shortest-path algorithm. Based on these model-based estimates for laminar borders, final segmentation refinement was performed with an R script, as described previously.¹⁹ Once segmented, inferior and superior retinas (350–624 μm from the optic nerve head on the inferior and superior sides) were each analyzed; starting at 350 μm ensured that our data were analyzed away from the optic nerve head, where the outer retina is relatively uniform in all OCT data. We measured the ONL and ELM-RPE thicknesses from in-house R scripts that objectively extracted layer boundaries obtained after searching the space provided by machine-learning estimates (“seed boundaries”) as above. The ELM and RPE are initially identified by local signal maxima and the R script determines the ELM-RPE thickness by calculating the distance from ELM to the basal side of the RPE at the level of Bruch's membrane.^20,21

RNA-seq

Two retinas from individual mice were pooled per sample; retinas from 3 Kv8.2 WT and 5 Kv8.2 KO mice at 1 month of age were collected. Total RNA was isolated from retinas collected into RNAlater using the RNeasy Mini Kit (Qiagen) according to the manufacturer’s protocol. Library preparation and sequencing were performed by Novogene. Sequencing reads were adapter-trimmed using TrimGalore and quantified against the mouse transcriptome (GENCODE Mus musculus GRCm38) using Salmon (v1.10.2). Transcript-level abundances were summarized to gene-level counts using tximport. Differential gene expression analysis was performed in R using DESeq2 following filtering of low-count genes. Based on exploratory analysis of sample relationships, two KO samples were identified as potential outliers. Differential expression analyses were performed both including and excluding these samples. This comparison changed the statistical significance of three differentially expressed genes (Hsp1b, Akr1b1, and Slc12a5) but did not change the overall interpretation of the dataset or the genes selected for downstream validation. The results shown in Figure 4 are from the filtered dataset (3 WT and 3 KO samples). RNA-seq data have been deposited in the Gene Expression Omnibus (GEO) under accession number GSE327286. Separate RNA preparations from Kv8.2 WT and KO mice at 3.5 months of age were used for reverse transcriptase (RT) droplet digital PCR, with probe sets for mouse Slc6a6 (Mm.PT.58.11922589) and Nfat5 (Mm.PT.58.7500558) purchased from IDT, as previously described.²²

Figure 4. Upregulation of osmosensitive genes in Kv8.2 KO retina.

(A) Volcano plot showing differential gene expression from RNA-seq analysis comparing Kv8.2 KO and WT retina at four weeks of age. Each point represents an individual gene plotted by log2(FC) versus −log10(adjusted p-value). Genes with ≥20% change in expression are highlighted (orange, upregulated; green, downregulated), whereas genes with <20% change are shown in yellow. (B) Curated list of genes reported to be regulated by osmolarity in other tissues, grouped by functional category. (C) Fold changes for the genes shown in (B), bars are log2(FC) ± SEM. Asterisks denote statistically significant differences between Kv8.2 KO and WT retina. (D) Droplet digital RT-PCR validation of Slc6a6 and Nfat5 transcript abundance.

Western blotting

Two retinas from individual mice were pooled per sample; retinas from 3 Kv8.2 Het and 3 Kv8.2 KO mice at 2 months of age were collected. Retinas were homogenized in 300 μl lysis buffer (1% CHAPS, PBS, protease inhibitor cocktail), clarified by centrifugation at 5000 × g for 5 minutes and the supernatant mixed with 4x LDS sample buffer and 10x DTT, without heat denaturation. Samples (20 μl) were separated on 4–20% TBX gels and transferred to PVDF for blotting. Revert 700 staining was used for lane normalization followed by incubation in Licor blocker. Primary antibodies were anti-AQP1 (Abcam, Cat# ab168387, RRID:AB_2810992) used at 1:4000 and anti-AQP4 (Millipore Cat# AB3594, RRID:AB_91530) used at 1:5000, secondary antibody was anti-rabbit: DyLight800 (Rockland, Cat# 5151P) used at 1:15000. Blots were imaged using a LI-COR Odyssey and band intensity analyzed with Image Studio (v5) software.

Immunolabeling

Posterior eyecups from 3 Kv8.2 Het and 3 Kv8.2 KO mice, at 2 months of age, were dissected, fixed in 4% paraformaldehyde, cryoprotected in 30% sucrose, and embedded in tissue freezing media. Sets of Kv8.2 Het and KO retina sections were collected onto the same slides to ensure they were treated identically. After blocking with 10% normal goat serum containing 0.5% triton X-100, sections were incubated overnight at 4°C with anti-AQP1 (Abcam, Cat# ab168387, RRID:AB_2810992) used at 1:2000. After rinsing with PBS, sections were incubated with secondary antibody (anti-rabbit IgG: Alexa 647 (ThermoFisher, Cat #A-21235, RRID:AB_2535804) used at 1:500, mixed with Hoechst 33342, for two hours at room temperature. Images were collected using a THUNDER Imager 3D Tissue Fully automated upright research microscope Leica DM6 B equipped with a Leica DFC9000 GT camera. Image analysis, including THUNDER computational clearing and adjusting intensity levels for display, was performed using LASx software. Image acquisition and display settings were held constant for each set of Kv8.2 Het and KO pair. Signal intensity in the inner segment region was measured using FIJI.

Statistical Analysis

OCT biomarker data, ELM-RPE and ONL were analyzed using a linear mixed model with fixed effects of genotype, condition, side, and all interactions, and a random intercept for each mouse nested within genotype (biomarker ~ genotype x condition x side + (1| genotype:mouse)). For ΔELM-RPE and ΔONL, the lighting condition is already encoded in the outcome so a mixed model with fixed effects of biomarker, side, and their interaction (biomarker ~ genotype x side + (1 | genotype:mouse)) was used. For all models, denominator degrees of freedom were estimated using the Satterthwaite approximation. Estimated marginal means and contrasts were computed and visualized in R (v4.4.1) using Quarto (v1.5.45). Complete statistical outputs, including model summaries, estimated marginal means, and contrasts (Tables S1A–S4C), are available in Supplemental File S1. For RNA-seq, detailed analysis scripts and parameters are provided in Supplemental File S2. Western blot band intensities and immunolabeling signal intensities were quantified from independent biological samples (with technical replicates averaged) and compared between genotypes using unpaired two-sample t-tests with Welch’s correction. Generative AI (ChatGPT-5, OpenAI) was used to assist with R scripting.

Results

Reduced light-evoked expansion of the subretinal space in Kv8.2 KO mice

We used OCT to measure the thickness of the outer retina between the ELM and RPE bands along with the thickness of the ONL (Fig. 1A). Using animals that were light-adapted for one hour, we found that ELM-RPE thickness was reduced in Kv8.2 KO mice compared to Het littermate controls (Fig. 1B). The reduction was greater in inferior retina (mean difference of 6.23 μm, 95% CI [4.95, 7.51], p < 0.001) than in superior retina (mean difference of 4.63 μm, 95% CI [3.35, 5.91], p < 0.001; Table S1D) (Fig. 1C). ONL thickness in these images was reduced slightly in Kv8.2 KO mice compared with Hets (12% in inferior retina and 8% in superior retina in the light-adapted state), indicating retinal degeneration had begun with only the loss of approximately one row of photoreceptor nuclei (Supplemental Fig. S2, Table S3C).

Figure 1. Reduced ELM–RPE thickness in light-adapted Kv8.2 KO retina.

(A) Representative OCT image showing inferior and superior retinal regions analyzed relative to the optic nerve head (orange rectangles). (B) Representative region-of-interest showing ELM-RPE distance in Kv8.2 Het versus KO mice. Dashed lines indicate the boundaries used to measure ELM-RPE thickness. (C) Quantification of ELM-RPE thickness in inferior and superior retina. Bars are mean ± 95% CI for Kv8.2 Het (blue) and KO (red). Asterisks indicate statistically significant differences between genotypes (***p < 0.001).

To isolate the dynamics of SRS volume, we quantified light-evoked swelling as the difference between light- and dark-adapted measurements within each animal (ΔELM-RPE). Using this paired metric, Kv8.2 KO mice exhibited significantly reduced light-evoked expansion of the outer retina compared with Het controls (Fig. 2A). The magnitude of the defect was slightly greater in inferior retina (mean difference 2.75 μm, 95% CI [1.38, 4.11], p < 0.001) than in superior retina (mean difference 2.20 μm, 95% CI [0.84, 3.56], p < 0.01; Table S2C). These results indicate that although light-evoked ELM–RPE expansion remains present in the absence of Kv8.2, the magnitude is attenuated, consistent with impaired regulation of subretinal fluid dynamics.

Figure 2. Reduced light-evoked expansion of the SRS in Kv8.2 KO retina.

Light-dark difference in ELM-RPE thickness (ΔELM-RPE) measured in inferior and superior retina. Points represent individual animals. Floating bars are mean ± 95% CI for Het (blue) and KO (red). Asterisks indicate statistically significant differences between genotypes (**p < 0.01; ***p < 0.001).

We also observed a genotype-dependent light-dark difference in ONL thickness, expressed as ΔONL (Fig. 3). Using this paired metric, Kv8.2 KO showed a small expansion in ONL thickness compared to Het controls. The genotype difference was larger in the inferior (mean difference 2.06 μm, 95% CI [0.53, 3.59], p = 0.0108) than in the superior (mean difference 1.76 μm, 95% CI [0.23, 3.28], p = 0.0264; Table S4C). This secondary finding indicates altered light-dependent changes in ONL thickness in Kv8.2 KO mice reciprocal to the reduced light-evoked expansion of the SRS.

Figure 3. Increased light-evoked changes in the ONL in Kv8.2 KO retina.

Light-dark difference in ONL thickness (ΔONL) measured in inferior and superior retina. Points represent individual animals. Floating bars are mean ± 95% CI for Het (blue) and KO (red). Asterisks indicate statistically significant differences between genotypes (*p < 0.05).

Retinal molecular changes consistent with osmotic imbalance in Kv8.2 KO retina

We then assessed retinal gene expression by RNA-seq. At four weeks of age, 590 genes were upregulated and 335 genes were downregulated (adjp < 0.01) in Kv8.2 KO compared with Kv8.2 WT retina (Fig. 4A). Among the upregulated genes were markers of Müller glial activation, while many downregulated genes were associated with oxidative phosphorylation.

Cellular adaptation to osmotic imbalance is mediated in part by the tonicity-responsive transcription factor NFAT5.^23–25. Because NFAT5 signaling has not been examined in the retina, we curated a set of genes reported to be regulated by NFAT5 or tonicity (Fig. 4B; Supplementary Table S5).^23–33 Most of these genes were upregulated in Kv8.2 KO retina, although the magnitude of change was modest (20–50%; Fig. 4B,C). We used droplet digital RT-PCR to independently quantify expression of two genes: Slc6a6, which encodes the transporter for taurine, one of the most abundant osmolytes in the retina, and Nfat5. Slc6a6 expression was increased in Kv8.2 KO retina, whereas Nfat5 expression was unchanged, consistent with the RNA-seq results (Fig. 4D).

We next examined aquaporins, which mediate water transport across cell membranes. Both Aqp1, abundant in photoreceptors, and Aqp4, abundant in Müller glia, were upregulated in the RNA-seq dataset. At the protein level, Western blotting showed a twofold increase in AQP1 in Kv8.2 KO, but no significant change in AQP4 after normalization to total protein loading (mean difference in Aqp1 was 0.71 arbitrary units, 95% CI [0.15, 1.26], p = 0.0284 and mean difference in Aqp4 was 0.23, 95% CI [−0.58, 1.01], p = 0.1118; Fig. 5A). A similar increase in labeling intensity for AQP1 in Kv8.2 KO photoreceptor inner segments was also observed by immunolabeling (mean difference 0.77 arbitrary units, 95% CI [0.50, 1.03], p = 0.0065; Fig. 5B). Together, these findings indicate that loss of Kv8.2 is accompanied by increased expression of water channels that could facilitate water movement from photoreceptors.

Figure 5. Increased aquaporin protein expression in Kv8.2 KO retina.

(A) Representative western blots and quantification of AQP1 and AQP4 protein levels in retina from Kv8.2 Het and KO mice. Band intensity was normalized to total protein loading and plotted in arbitrary units (a.u.). (B) Immunolabeling of the outer retina with AQP1. Yellow triangles mark presumed red blood cells expressing AQP1. Abbreviations are OS, outer segment, IS, inner segment, ONL, outer nuclear layer, and OPL, outer plexiform layer. Normalized signal intensity per area in the inner segments are plotted in arbitrary units (a.u.). Bars represent mean ± SEM. Asterisks denote statistically significant differences between genotypes (*p < 0.05).

Discussion

We provide first-time, non-invasive metabolic biomarker evidence that loss of Kv8.2 is associated with altered fluid homeostasis in the outer retina and molecular evidence of an osmotic shift in the microenvironment. Specifically, we found that in the absence of Kv2.1/Kv8.2 channels there was i) reduced light-dependent SRS volume increase, ii) a light-dependent expansion of the outer nuclear layer, iii) upregulation of osmosensitive genes, and iv) increased expression of aquaporin-1 in photoreceptors.

The ELM-RPE biomarker is well validated and reflects an integrated physiological response because the hydration changes reported by this metric are regulated by photoreceptor photosignaling and mitochondrial activity.¹⁹ However, the basis of the light-dark difference in ONL thickness that we observed in Kv8.2 KO mice is not as well understood. Prior work using acetazolamide has shown that acute perturbation of retinal fluid handling can reduce ONL thickness in light-adapted retina, supporting a link between ONL thickness and hydration.³⁴ In the present study, ΔONL increased in Kv8.2 KO while ΔELM-RPE decreased, raising the possibility that relative expansion of the ONL reflects impaired redistribution of fluid from photoreceptors into the SRS, resulting in water retention within photoreceptors. Alternatively, the expansion could reflect structural remodeling or fluid accumulation in Muller glia. Additional studies will be required to define the physiological basis of this phenomenon.

In the retina, coupling of potassium flux to water movement is already well established in Müller glia, where inwardly rectifying Kir4.1 channels and AQP4 coordinate ion and fluid transport.³⁵ By analogy, a similar relationship may exist in photoreceptors, where Kv2.1/Kv8.2 channels and AQP1 are positioned at the inner segment plasma membrane adjacent to the SRS. In this context, potassium efflux through Kv2.1/Kv8.2 channels–best known for shaping the photoreceptor light response–could contribute to local osmotic gradients that facilitate water movement from photoreceptors into the SRS for subsequent clearance across the RPE (Fig. 6).

Figure 6. Model linking Kv2.1/Kv8.2 mediated potassium flux to osmotic regulation in the outer retina.

Schematic comparing control (left) and Kv8.2 KO (right) retina. The SRS is bounded by photoreceptors, the RPE, and Müller glial processes at the ELM. In control retina, potassium efflux through Kv2.1/Kv8.2 channels (orange arrows) contributes to an osmotic gradient that facilitates water movement (blue arrows) from photoreceptors into the SRS, followed by clearance across the RPE. In Kv8.2 KO retina, reduced potassium efflux is proposed to diminish this osmotic driving force, resulting in retention of water within photoreceptors and reduced SRS volume.

Tonicity signaling in the retina remains poorly understood. In other parts of the body, osmolyte transporters and aquaporins frequently work together to modulate fluctuations in cellular volume. That makes the increased expression of the transporter for taurine (Slc6a6), the most abundant osmolyte in the retina, and the increased expression of aquaporin-1 notable. To unravel the functional consequences of these changes, two limitations of the molecular studies should be addressed. First, our use of bulk retina does not allow us to identify the cellular source of the response. Photoreceptors are implicated as the source of the response simply because they are the most abundant cell type in the retina, express Kv2.1/Kv8.2 channels, and have increased aquaporin-1 expression in the Kv8.2 KO. However, Muller glia and the RPE are important regulators of retinal fluid homeostasis and may also contribute adaptive responses to altered fluid and ion homeostasis initiated by loss of Kv8.2 in photoreceptors. Second, regulation of osmolyte transporters and aquaporins occurs through mechanisms in addition to gene expression, including trafficking and post-translational modification.^36–40 Consequently, the changes reported here should not be interpreted as direct evidence for activation of a specific tonicity signaling pathway, but rather as evidence that loss of Kv8.2 is accompanied by a coordinated molecular response involving osmolyte and water homeostasis.

In summary, the convergence of structural and molecular changes supports the concept that potassium efflux through Kv2.1/Kv8.2 channels contributes to regulation of fluid balance in the outer retina. Disruption of this mechanism may represent an underappreciated component of KCNV2 retinopathy and could have broader relevance for retinal diseases in which ion fluxes are altered.