Rapid RGR-dependent visual pigment recycling is mediated by the RPE and specialized Müller glia

**Background:** In daylight, the demand for 11-cis-retinal (visual chromophore) exceeds the supply from the classical visual cycle mediated by RPE65. The retinal G-protein-coupled receptor (RGR) is a photoisomerase expressed in both the retinal pigment epithelium (RPE) and Müller glia that can convert all-trans-retinal to 11-cis-retinal using light energy. However, the relative contributions of these two cellular pools of RGR to photoreceptor function were unknown. This study aimed to characterize the kinetics of RGR-mediated chromophore production, analyze RGR distribution across species, and determine the roles of RPE and Müller glial RGR in rod and cone function.

**Methods:** The authors generated a conditional Rgr knockout mouse (Rgr^S/S) with a loxP-flanked stop cassette, enabling cell-specific rescue via Cre recombinase. They created five lines on a Gnat1-/- background (to isolate cone responses): wild-type (WT), knockout (KO), RPE-Cre (RGR restored in RPE), MG-Cre (RGR restored in Müller glia), and 2-Cre (RGR restored in both). Tamoxifen induction at 2 months of age for 3 weeks activated Cre. In vivo electroretinography (ERG) measured M-cone b-wave sensitivity during 60-min exposure to 300 cd/m² green background light and subsequent dark adaptation. Rod function was assessed in Gnat1+/+ mice after 30-min background light. Retinoid levels were quantified by HPLC, and retinosome area by two-photon fluorescence lifetime imaging (FLIM). In vitro, the hydrolysis kinetics of RGR* (photoactivated RGR) were measured in bovine RPE microsomes using LC-MS/MS after pronase digestion. Single-cell RNA-seq data from mouse, macaque, and human retinas were analyzed to assess RGR expression in Müller glia.

**Key Results:** In vitro, the 11-cis-retinylidene Schiff base of RGR* hydrolyzed with a half-life of 7.5 seconds at 20°C, approximately 60-fold faster than the hydrolysis of all-trans-retinylidene from photoactivated rhodopsin. In vivo, KO mice showed a ~2-fold greater desensitization of M-cone b-wave sensitivity during background light compared to WT, and dark adaptation was ~5-fold slower (by 30 min). Selective rescue in RPE (RPE-Cre) improved cone sensitivity ~1.9-fold over KO, while rescue in Müller glia (MG-Cre) improved it ~1.4-fold. Combined rescue (2-Cre) restored cone function to WT levels. Rod a-wave recovery after 30-min background light was also suppressed in KO mice. scRNA-seq revealed that only 21.2% of mouse Müller glia, 5.5–8.4% of macaque Müller glia, and 93.6–96.3% of human Müller glia express RGR. RGR+ Müller glia showed upregulation of CRALBP (RLBP1, ~1.4-fold), RDH11 (~1.4-fold), and GLUT1 (SLC2A1, ~1.5-fold) compared to RGR− cells. RGR loss led to elevated LRAT levels and increased retinyl ester storage (retinosome area significantly larger in Rgr^S/S mice under both dark and light conditions).

**Clinical Implications:** This study establishes RGR as a rapid, light-driven isomerase that complements the classical visual cycle, particularly under sustained bright light. The finding that both RPE and Müller glial RGR pools are required for optimal cone function and that RGR accelerates dark adaptation of both rods and cones has implications for understanding retinal diseases involving chromophore deficiency or all-trans-retinal toxicity (e.g., Stargardt disease, age-related macular degeneration). The species differences in Müller glial RGR expression (high in human fovea) suggest that the photic visual cycle is particularly important for human cone-rich vision. The upregulation of LRAT in RGR-deficient mice indicates a compensatory mechanism to prevent all-trans-retinal accumulation, highlighting RGR's role as a retinal sink for bleached chromophore. These insights may guide therapeutic strategies targeting RGR or the photic visual cycle to improve visual function in conditions with impaired chromophore regeneration.