P23H rhodopsin accumulation causes transient disruptions to synaptic protein levels in rod photoreceptors in a model of retinitis pigmentosa

Abstract

Rod photoreceptor neurons in the retina detect scotopic light through the visual pigment rhodopsin (Rho) in their outer segment (OS). Efficient Rho trafficking to the OS through the inner rod compartments is critical for long-term rod health. However, given the importance of protein trafficking to the OS, little is known about the trafficking of rod synaptic proteins. Furthermore, the subcellular impact of Rho mislocalization on rod synapses (i.e. 'spherules') has not been investigated. In this study, we used super-resolution and electron microscopies, along with proteomics, to perform a subcellular analysis of Rho synaptic mislocalization in P23H-Rho-RFP mutant mice. We discovered that mutant P23H-Rho-RFP protein mislocalized in distinct accumulations within the spherule cytoplasm, which we confirmed with adeno-associated virus overexpression. Additionally, we found specific synaptic protein abundance differences in P23H-Rho-RFP mice. Interestingly, in P23H knock-in mice with no RFP tag, we detected no synaptic protein abundance changes. In rd10 mutant rods, Rho mislocalized along the spherule plasma membrane, and there were synaptic protein abundance differences at postnatal day 20. Our findings demonstrate that some rod photoreceptor synaptic proteins are sensitive to Rho mislocalization.

Keywords: Mislocalization; Photoreceptors; Retina; Rhodopsin; Super-resolution; Synapse.

Fig. 1.

Mutant P23H-hRho-RFP protein is mislocalized within the cytoplasm of rod photoreceptor presynaptic spherules. (A) Diagram depicting the layers of the mouse outer retina (RPE, retinal pigment epithelium; OS, outer segment; IS, inner segment; ONL, outer nuclear layer; OPL, outer plexiform layer) and the two types of rod spherules (R1, top; R2, bottom). Spherule illustrations were based on Li et al. (2016) under the terms of the CC-BY 4.0 license. (B,C) Confocal z-projections of WT-RFP/+ and P23H-RFP/+ retinal cryosections at postnatal day (P)30 (B) and P90 (C). RFP fluorescence is magenta, and sections were co-stained with wheat germ agglutinin (WGA) to label OS membranes (green) and DAPI to label nuclei (blue). White arrows indicate mislocalized RFP. (D) Structured illumination microscopy (SIM) z-projections with 3D deconvolution of the OPL from a P30 wild-type (WT) retina. In the images, RIBEYE (yellow)- and PSD95 (cyan)-immunolabeled rod spherules are aligned in the OPL above a cone pedicle (white arrowhead). No 1D4 Rho labeling (magenta) is present in the WT OPL. In single-spherule examples, the RIBEYE⁺ ribbons are horseshoe-shaped structures. (E,F) SIM images of the OPL in a P23H-RFP/+ retina at P30 with the same immunolabeling as in D. Accumulations of 1D4 immunolabeling in the OPL were localized near the synaptic ribbons (white arrows). (F) Single-spherule SIM examples in P23H-RFP/+ P30 retinas. 1D4⁺ Rho accumulations are localized in the cytoplasm of the spherules, typically above the ribbon. In the second SIM+3D deconvolution image, a gap in the accumulated 1D4⁺ fluorescence is indicated by a white asterisk. The far-right example is a R2-type mutant spherule with 1D4 fluorescence that surrounds the nucleus (magenta arrows) and extends into the spherule cytoplasm. (G) SIM images of P23H-RFP/+ retinas at P90 with the same immunolabeling as in D-F. White arrows indicate 1D4⁺ OPL accumulations. In single-spherule examples, cytoplasmic 1D4⁺ accumulated puncta surround gaps in fluorescence (white asterisk). (H,I) SIM z-projection images of P30 WT (H) and P30 P23H-RFP/+ (I) retinas immunolabeled for Sec61β (ER marker, magenta), PSD95 (cyan) and bassoon (Bsn; yellow). In magnified images, the PSD95 rod spherule border is annotated in select rod spherules to demonstrate Sec61β⁺ ER fluorescence within individual WT and P23H-RFP/+ rod spherules (white arrowheads). In the P23H-RFP/+ image in I, ER accumulations are labeled with yellow arrows. All experiments were repeated in triplicate (N=3 mice per timepoint, per experiment).

Fig. 2.

P23H-Rho-RFP mislocalization does not cause ultrastructural defects in rod synaptic ribbons. (A) Transmission electron microscopy (TEM) images of WT rod spherules at P30. Middle image is the annotated version of the left image (cyan, spherule plasma membrane; yellow, synaptic ribbon). In the right image, ER-like membranes are wrapped around the mitochondrion. (B) TEM image of a P23H-RFP/+ rod spherule at P30. ER-like membrane stacks are localized near the mitochondrion and extend into the cytoplasm. (C) Additional TEM images of P30 P23H-RFP/+ rod spherules with mitochondria-associated ER-like membrane stacks. (D) TEM image of a P30 P23H-RFP/+ R2 rod spherule. (E) Top: ribbon height measurement graph from P30 TEM images. For WT, n=53 ribbons, N=3 mice; for P23H-RFP/+, n=68 ribbons, N=3 mice. Bottom: ribbon-associated synaptic vesicle (SV) count graph from P30 TEM images. For WT, n=46 ribbons, N=3 mice; for P23H-RFP/+, n=65 ribbons, N=3 mice. There are no significant differences (P≥0.05) based on an unpaired two-tailed t-test. (F) TEM images of P90 WT spherules. (G) TEM image of a P90 P23H-RFP/+ spherule with expanded ER. (H,I) Magnified examples of ER-mitochondria contact sites (white arrows) in P23H-RFP/+ spherules at P30 (H) and P90 (I). ER membranes are traced in gold, the mitochondrial membranes are traced in magenta, and the plasma membranes are traced in cyan. Throughout, magenta asterisks indicate mitochondria, green arrowheads indicate endocytosed vesicles, orange arrowheads indicate ER-like membranes, and orange arrows indicate ER-like membranes. Experiments were replicated in triplicate.

Fig. 3.

Adeno-associated virus (AAV) overexpression of P23H-Rho and R135L-Rho in WT rods causes synaptic mislocalization. (A-D) Confocal images of retinal sections from WT mice transduced with AAVs for rod-specific expression of the following Rho fusions: WT-hRho-RFP (A), P23H-hRho-RFP (B), WT-hRho-GFP (C) and R135L-hRho-GFP (D). Images on the left side show Rho fusion localization (magenta) in transduced rods with DAPI (blue) labeling. Right images show co-labeling with WGA (green). Yellow arrows indicate mutant Rho fusion OPL mislocalization. (E,F) AAV-infected retinal sections with rods overexpressing WT-hRho-RFP (E) or WT-hRho-EGFP (F) (magenta, yellow arrows). (G,H) SIM z-projection images of the OPL from 3-μm retinal cryosections from the same AAV conditions as in A-D. Rho fusion fluorescence is magenta, and PSD95 immunolabeling (cyan) was used to identify rod spherules. Single-spherule examples of each AAV-driven Rho fusion are shown below with PSD95 co-labeling (left) and the Rho-RFP/EGFP signal only (right). In G, white arrows indicate P23H-hRho-RFP cytoplasmic accumulated puncta. Dashed gray lines outline the PSD95⁺ plasma membrane of the magnified spherules. In H, R135L-hRho-GFP mislocalizes at the spherule plasma membrane (white arrowheads) and internally (yellow arrowheads). (I,J) SIM z-projection images from thin 1-μm sections of retinas from the same AAV conditions as in A-D. These sections were co-immunolabeled for 1D4 (magenta) and RIBEYE (yellow). White arrows indicate mutant Rho fusion accumulations. Single-spherule examples from each AAV condition are enlarged below. (K) SIM images of WT-hRho-RFP overloaded rod spherules. 1D4⁺ WT-hRho-RFP (magenta) was localized along the plasma membrane of the axon and spherule (white arrows) and colocalized with the RIBEYE synaptic ribbons (yellow arrowheads). In the far-right example, 1D4⁺ WT-hRho-RFP appears to bud off from the presynaptic spherule (magenta arrow). (L) SIM images of WT-hRho-GFP AAV overloaded rod spherules. WT-hRho-GFP colocalizes with the PSD95⁺ rod spherule plasma membrane (white arrowheads) and localizes at the rod axon plasma membrane (white arrows). All experiments were repeated in triplicate (N=3 mice per AAV).

Fig. 4.

Photoreceptor and synaptic protein abundance changes are found in P23H-RFP/+ retinas with tandem mass tag-mass spectrometry (TMT-MS) proteomics. (A) Diagram of the front view of a rod spherule (top) and a magnified view of the active zone and the synaptic cleft (bottom) to highlight the approximate localizations of rod synaptic proteins, including key trans-synaptic protein complexes. (B) Volcano plot of TMT-MS relative peptide abundances for select photoreceptor and synaptic proteins (see Materials and Methods) in P30 P23H-RFP/+ versus WT retinas. x-axis, log₂ fold change values with a significance threshold of 0.2; y-axis, P-values (−log10) with a significance threshold of 1.5. Green points represent protein targets above the thresholds and magenta points represent targets below the thresholds. Annotated protein names are based on FASTA gene names. (C) Peptide abundance graph of select log2 peptide values from the P30 TMT-MS data in Fig. S3B. WT (N=3, open circles) were superimposed with P23H-RFP/+ values (N=4, gray diamonds). Magenta asterisks indicate significant downregulation; green asterisks indicate significant upregulation. (D) TMT-MS volcano plot comparing relative peptide abundances in P90 P23H-RFP/+ (N=4) and WT (N=3) retinas for the same protein list and plot parameters as in B. (E) Linear scale graph of select peptide values from the P90 TMT-MS data in Fig. S3B with the same formatting as in C. *P<0.055, **P<0.01, ***P<0.001 (unpaired two-tailed t-test).

Fig. 5.

Quantitative confocal imaging analysis of synaptic protein abundance changes in P23H-RFP/+ mice. (A,B) Example confocal z-projections from WT (N=3 mice) and P23H-RFP/+ (N=3 P30 mice, N=4 P90 mice) retinal cryosections at P30 (A) and P90 (B) focused on regions of the lower/proximal ONL and OPL with dystrophin (Dmd; cyan) and Bsn (yellow) immunolabeling and DAPI counterstaining (blue). RFP fluorescence (magenta) was detectable only in the ONL and OPL of the P23H-RFP/+ sections. White arrowheads indicate cone synapses. (C,D) Confocal z-projections for WT and P23H-RFP/+ retinal cryosections at P30 (C) and P90 (D) with ELFN1 immunolabeling (green) and DAPI nuclear staining (blue). Yellow arrows indicate mislocalized strings of ELFN1. (E) Time course plot of Dmd (blue) and Bsn (yellow) normalized puncta intensity measurements from P23H-RFP/+ retinas. Raw data image examples for timepoints P14 (N=3 mice, each group), P60 (N=3 WT mice, N=5 P23H-RFP/+ mice), P180 (N=3 mice, each group) and P365 (N=3 mice, each group) are provided in Fig. S4E. (F) Time course plot of DAPI⁺ photoreceptor nuclei per column counted from both WT and P23H-RFP/+ in confocal images analyzed throughout the puncta analyses. Two-phase decay curve fits were added in GraphPad Prism. (G) Graph of normalized ELFN1 OPL intensities between WT and P23H-RFP/+ retinas at P30 and P90 (N=3 mice, each group at each age). (H) Example confocal images from P23H-RFP/+ retinas depicting RFP fluorescence (magenta) colocalized with either Dmd (top, green) or Bsn (bottom, green) immunolabeling. In magnified views, Dmd and Bsn intensities are heat map pseudocolored and superimposed with the RFP signal as white outlines. White arrows indicate RFP⁺ Dmd/Bsn colocalizations; yellow arrows indicate RFP⁻ Dmd/Bsn examples. (I) Graph of averaged puncta intensity values from the P90 P23H-RFP/+ RFP colocalization assay for Dmd (cyan) and Bsn (yellow). Bars, mean values; error bars, s.d. Significance determined using unpaired two-tailed t-tests. Experiments were repeated in triplicate. **P<0.01.

Fig. 6.

ELFN1 protein distribution is altered in P23H-RFP/+ retinas. (A,B) Confocal z-projection images of P30 WT (A; N=3 mice) and P30 P23H-RFP/+ (B; N=3 mice) retinal cryosections immunolabeled for ELFN1 (green) and mGluR6 (magenta) and counterstained for DAPI (blue). (C) Magnified confocal images with the same labeling focused on the OPL. Staining, acquisition and intensity settings were matched throughout A-C between the P30 WT and P30 P23H-RFP/+ sections. (D) Normalized intensity graph (left) based on layer-specific intensity measurements for ELFN1 (green) and mGluR6 (magenta). Values were aggregated from replicate WT versus P23H-RFP/+ experiments. On the right, ratios of ELFN1 intensity ONL/OPL intensities are plotted. (E) Example RNAScope SIM z-projections images for Elfn1 mRNA (green) in P30 WT (N=3 mice) and P30 P23H-RFP/+ (N=3 mice) retinas, alongside example SIM images probed for POLR2A (positive control) and dapB (negative control) mRNAs (P30 WT control images are in Fig. S5C). Sections were co-immunolabeled for centrin (magenta) to label connecting cilia and counterstained with DAPI (blue). Yellow dashed lines indicate the IS:ONL and ONL:OPL boundaries based on the DAPI staining. (F) Confocal images of P30 WT and P23H/+ (right) retinas labelled for 1D4 (magenta), RIBEYE (green), and DAPI (blue) after antigen retrieval demasking. Yellow arrows indicate mislocalized Rho in the P23H/+ ONL. (G) Graphed normalized puncta intensities for dystrophin (cyan), Bsn (yellow) and ELFN1 (green) from P30 WT and P23H/+ OPLs. Bars, mean values; error bars, s.d. Significance determined using unpaired two-tailed t-tests. Experiments were repeated in triplicate. ns, not significant; *P<0.055, ***P<0.001.

Fig. 7.

Analysis of Rho mislocalization and presynaptic protein levels in rd10 RP mutant retinas. (A) SIM images of the OPL in WT and rd10 mice at P16 and P20. Retinas were immunolabeled for 4D2 (Rho, magenta), RIBEYE (yellow) and PSD95 (cyan). In the WT OPL, the 4D2 signal is not consistently above background levels. (B) Magnified single-rod spherule SIM images from P16 rd10 retinas with the same labeling as in A. (C) Confocal z-projections of cryosections from Rho^GFP/+ (left) and rd10; Rho^GFP/+ (right) retinas depicting Rho-GFP (GFP, green) localization. Sections were counterstained with DAPI (blue). Grayscale images are GFP only with increased brightness to demonstrate Rho-GFP OPL mislocalization in rd10; Rho^GFP/+ retinas (white arrows). (D) SIM image of a P16 rd10; Rho^GFP/+ retinal cryosection immunolabeled for RIBEYE (magenta) to demonstrate mislocalized Rho-GFP overlap with RIBEYE⁺ ribbons. (E) SIM images of thin resin sections of P16 rd10; Rho^GFP/+ retinas co-immunolabeled with a GFP nanobody (NbGFP-A647, magenta) and for RIBEYE (yellow). NbGFP-A647 labeling was specific for Rho-GFP because cone ribbons are present (white arrowheads) had no NbGFP-A647 staining. In magnified views of rd10; Rho^GFP/+ spherules, Rho-GFP is localized in the OPL along the spherule plasma membranes (white arrows). Small Rho-GFP puncta were observed in the OPL extracellular space as if detached from the spherule membrane (green arrowheads). (F) Graph of aggregated normalized puncta intensities for dystrophin (cyan), Bsn (yellow) and ELFN1 (purple) from P16 and P20 WT (open circles) and rd10 (filled circles) retinas. Raw image examples are in Fig. S6G. Experiments were repeated in triplicate. Significance determined using unpaired two-tailed t-test (*P<0.05,) (G) Raw confocal image examples of WT and rd10 retinas at P16 and P20 used in F.

Fig. 8.

Diagram comparing Rho mislocalization pattern differences in mutant rod spherules. The spherule plasma membrane is highlighted in cyan. In WT spherules, ER (gold) wraps around the mitochondrion (tan) in the cytoplasmic space above the synaptic ribbon (yellow) and the synaptic cleft containing the postsynaptic neurites (ON-type bipolar cells, orange; horizontal cells, black). In approximately one-quarter of P23H-RFP/+ rod spherules, mutant Rho (magenta dots) accumulates in expanded ER (gold) in the cytoplasmic space . In rd10 rod spherules or WT spherules with overexpressed WT-Rho fusion protein or mutant R135L-hRho-GFP protein, Rho (magenta) mislocalizes along the spherule plasma membrane (cyan) such that the ER (gold) is putatively unaffected.