Interactions between C8orf37 and FAM161A, Two Ciliary Proteins Essential for Photoreceptor Survival

Mutations in C8orf37 cause Bardet-Biedl syndrome (BBS), retinitis pigmentosa (RP), and cone–rod dystrophy (CRD), all manifest in photoreceptor degeneration. Little is known about which proteins C8orf37 interacts with to contribute to photoreceptor survival. To determine the proteins that potentially interact with C8orf37, we carried out a yeast two-hybrid (Y2H) screen using C8orf37 as a bait. FAM161A, a microtubule-binding protein localized at the photoreceptor cilium required for photoreceptor survival, was identified as one of the preys. Double immunofluorescence staining and proximity ligation assay (PLA) of marmoset retinal sections showed that C8orf37 was enriched and was co-localized with FAM161A at the ciliary base of photoreceptors. Epitope-tagged C8orf37 and FAM161A, expressed in HEK293 cells, were also found to be co-localized by double immunofluorescence staining and PLA. Furthermore, interaction domain mapping assays identified that the N-terminal region of C8orf37 and amino acid residues 341-517 within the PFAM UPF0564 domain of FAM161A were critical for C8orf37–FAM161A interaction. These data suggest that the two photoreceptor survival proteins, C8orf37 and FAM161A, interact with each other which may contribute to photoreceptor health.


Introduction
Mutations in nearly 300 genes have been identified as causes of retinal degeneration. Abnormal C8orf37 have been found in patients with Bardet-Biedl syndrome (BBS) [1,2]. BBS (MIM209900) is characterized by photoreceptor degeneration, obesity, digit anomalies, genito-urinary anomalies, and cognitive deficits. This autosomal recessive disorder is caused by mutations in over 20 genes that are implicated in the function of primary cilia. Mutations in C8orf37 can also cause non-syndromic retinal degeneration including conerod dystrophy (CRD, MIM 120,970, cone-rod dystrophy 16 (CORD16) [3][4][5]) and retinitis pigmentosa (RP, MIM 268,000, retinitis pigmentosa 64 (RP64) [3,[6][7][8][9]). Retinitis pigmentosa consists of a group of blinding diseases that are genetically heterogeneous. Amongst all inherited retinal degenerative diseases, RP is the most common form and is characterized by an initial loss of rod photoreceptors followed by loss of cone photoreceptors. In cone-rod dystrophy, cone photoreceptors are predominantly affected. Loss of vision in these diseases is usually progressive and severe, with no effective treatment.
Human C8orf37 is a ubiquitously expressed protein with 207 amino acid residues and has no identified structural domains. How C8orf37 protein contributes to photoreceptor survival is unknown. Previous research suggests that C8orf37 may be involved in the function of primary cilia. C8orf37 immunoreactivity is found at the ciliary base in hTert-RPE1 cells as well as the ciliary base and inner segments of mouse photoreceptors [6].

Co-localization of C8orf37 and FAM161A in the Marmoset Retina
A yeast two-hybrid screen of human retinal library, using full length human C8orf37 as a bait, identified 28 potential interactions ( Table 1). Two of these genes have been implicated in retinal degeneration. Mutations in transcription factor CRX are known to cause autosomal dominant cone-rod dystrophy type 2, Leber congenital amaurosis type 7 and autosomal dominant retinitis pigmentosa "https://web.sph.uth.edu/RetNet/disease.htm (accessed on 1 October 2022)". Interestingly, the prey FAM161A is a microtubule-binding protein enriched at the ciliary base of photoreceptors [12,13] and has been implicated in retinitis pigmentosa 28 [10,11]. The Y2H screening captured a fragment of FAM161A that corresponded to aa 317-553, which lies within its PFAM UPF0564 domain. To evaluate whether C8orf37 and FAM161A proteins in photoreceptors were co-localized, we first carried out double immunofluorescence staining of common marmoset retinal sections using rabbit anti-C8orf37 or rabbit anti-FAM161A with mouse anti-acetylated α-tubulin, an axoneme marker. We chose the marmoset retina for evaluating their co-localization because the commercially available antibodies were raised against human proteins, and thus more likely to recognize primate antigens than those of the mouse.
C8orf37 protein has been previously reported to be highly enriched at the ciliary base at the junction between the outer and inner segments of photoreceptors [6]. In this study, we used an anti-C8orf37 antibody with immunofluorescence pattern identical to turboGFP fluorescence in HEK293 cells transfected with a C8orf37-turboGFP fusion construct (Supplementary Figure S1A-C). This antibody also recognized the same band as anti-turboGFP on Western blots of C8orf37-turboGFP-transfected cell lysate, supporting antibody specificity (Supplementary Figure S1D). C8orf37 immunoreactivity was widely distributed in the neural retina with strong immunofluorescence reactivity in the outer nuclear layer (ONL), inner nuclear layer (INL), and ganglion cell layer (GCL) ( Figure 1A,C). Anti-acetylated α-tubulin staining was used to identify photoreceptor axonemes ( Figure 1B,E,H,K). C8orf37 immunoreactivity was enriched at the inner segment, including the ciliary base ( Figure 1D,F, arrows) and along the length of the photoreceptor axoneme ( Figure 1F, arrowheads). Similarly, FAM161A immunoreactivity was also widely distributed in the neural retina ( Figure 1G,I,J,L). As reported in mouse photoreceptors [12,13], FAM161A immunoreactivity was also enriched in the inner segment and highly enriched at the ciliary base ( Figure 1L, arrows). To evaluate whether C8orf37 and FAM161A were co-localized, we performed proximity ligation assays by conjugating anti-C8orf37 and anti-FAM161A antibodies with Duolink™ In Situ Probemaker PLUS probes and Duolink™ In Situ Probemaker MINUS probes (Sigma Aldrich, Burlington, MA, USA), respectively. When both C8orf37-conjugated PLUS probes and FAM161A-conjugated MINUS probes were used, PLA puncta were found throughout the neural retina (Figure 2A,D,G,J). When PLA was performed with either C8orf37-PLUS or FAM161A-MINUS probes alone, PLA puncta were significantly reduced ( Figure 2B,C,E,F,H,I,K,L). We quantified the PLA puncta, comparing their numbers between C8orf37/FAM161A double probe assays and C8orf37-only and FAM161A-only single probe assays ( Figure 3). The numbers of PLA puncta in every retinal layer were higher when both probes were used than when single probes were used (ANOVA followed by post-hoc Tukey HSD, Figure 3D). There was no statistically significant difference between the C8orf37-only and FAM161A-only single probe control assays ( Figure 3B,C). Together, these data indicate that C8orf37 and FAM161A immunoreactivity was co-localized in the marmoset retina, including photoreceptors. The sections were counterstained with DAPI (blue). Images (A-F) show C8orf37 and acetylated α-tubulin double staining. C8orf37 immunoreactivity was observed in the inner segment layer, outer nuclear layer, inner nuclear layer, and ganglion cell layer. C8orf37 immunoreactivity puncta was also observed at the base of acetylated α-tubulin-labeled cilia (arrows in F), as well as along the cilia (arrowheads). Images (G-L) show FAM161A and acetylated α-tubulin double staining. FAM161A immunoreactivity was widely distributed in the retina. Its puncta were also enriched at the base of cilia (arrows in L) and along the cilia. Abbreviations: GCL, ganglion cell layer; INL, inner nuclear layer; IPL, inner plexiform layer; IS, inner segment; ONL, outer nuclear layer; OPL, outer plexiform layer; OS, outer segment. To evaluate whether C8orf37 and FAM161A were co-localized, we performed proximity ligation assays by conjugating anti-C8orf37 and anti-FAM161A antibodies with Du-olink™ In Situ Probemaker PLUS probes and Duolink™ In Situ Probemaker MINUS probes (Sigma Aldrich, Burlington, MA, USA), respectively. When both C8orf37- The sections were counterstained with DAPI (blue). Images (A-F) show C8orf37 and acetylated α-tubulin double staining. C8orf37 immunoreactivity was observed in the inner segment layer, outer nuclear layer, inner nuclear layer, and ganglion cell layer. C8orf37 immunoreactivity puncta was also observed at the base of acetylated α-tubulin-labeled cilia (arrows in F), as well as along the cilia (arrowheads). Images (G-L) show FAM161A and acetylated α-tubulin double staining. FAM161A immunoreactivity was widely distributed in the retina. Its puncta were also enriched at the base of cilia (arrows in L) and along the cilia. Abbreviations: GCL, ganglion cell layer; INL, inner nuclear layer; IPL, inner plexiform layer; IS, inner segment; ONL, outer nuclear layer; OPL, outer plexiform layer; OS, outer segment. Scale bar: 20 µm for (A-C) and (G-I); 6 µm for (D-F) and (J-L); 3 µm for insets in (F) and (L).
ing their numbers between C8orf37/FAM161A double probe assays and C8orf37-only and FAM161A-only single probe assays ( Figure 3). The numbers of PLA puncta in every retinal layer were higher when both probes were used than when single probes were used (ANOVA followed by post-hoc Tukey HSD, Figure 3D). There was no statistically significant difference between the C8orf37-only and FAM161A-only single probe control assays ( Figure 3B,C). Together, these data indicate that C8orf37 and FAM161A immunoreactivity was co-localized in the marmoset retina, including photoreceptors.

Co-Localization of Epitope-Tagged C8orf37 and FAM161A in Transfected HEK293 Cells
Since antibodies may recognize nonspecific antigens in immunostaining, some of the PLA puncta detected in the marmoset retina might have arisen from non-specific binding of anti-C8orf37 and anti-FAM161A. Thus, we sought to confirm the interaction with epitope-tagged C8orf37 and FAM161A heterologously expressed in HEK293 cells using epitope-specific antibodies. We obtained turboGFP-tagged human C8orf37 and c-Myctagged human FAM161A expression constructs from Origene Technologies (Rockville, MD, USA), pCMV-C8orf37-tGFP and pCMV-FAM161A-c-Myc. HEK293 cells were cotransfected with these epitope-tagged constructs. GFP fluorescence and immunofluorescence staining with c-Myc antibody indicated efficient transfection for both constructs ( Figure 4). In double transfected cells, overlapping immunofluorescence was observed ( Figure 4A-C). We then carried out proximity ligation assays with antibodies against tur-boGFP and c-Myc. PLA puncta were frequently observed in double-transfected cells

Co-Localization of Epitope-Tagged C8orf37 and FAM161A in Transfected HEK293 Cells
Since antibodies may recognize nonspecific antigens in immunostaining, some of the PLA puncta detected in the marmoset retina might have arisen from non-specific binding of anti-C8orf37 and anti-FAM161A. Thus, we sought to confirm the interaction with epitopetagged C8orf37 and FAM161A heterologously expressed in HEK293 cells using epitopespecific antibodies. We obtained turboGFP-tagged human C8orf37 and c-Myc-tagged human FAM161A expression constructs from Origene Technologies (Rockville, MD, USA), pCMV-C8orf37-tGFP and pCMV-FAM161A-c-Myc. HEK293 cells were co-transfected with these epitope-tagged constructs. GFP fluorescence and immunofluorescence staining with c-Myc antibody indicated efficient transfection for both constructs (Figure 4). In double transfected cells, overlapping immunofluorescence was observed ( Figure 4A-C). We then carried out proximity ligation assays with antibodies against turboGFP and c-Myc. PLA puncta were frequently observed in double-transfected cells ( Figure 4D). By contrast, the numbers of PLA-positive puncta observed in cells transfected with pCMV-C8orf37-tGFP alone, in cells transfected with pCMV-FAM161A-c-Myc alone, or in non-transfected cells were significantly reduced ( Figure 4E-H). These results suggested that overexpressed C8orf37 and FAM161A can be co-localized.
( Figure 4D). By contrast, the numbers of PLA-positive puncta observed in cells transfected with pCMV-C8orf37-tGFP alone, in cells transfected with pCMV-FAM161A-c-Myc alone, or in non-transfected cells were significantly reduced ( Figure 4E-H). These results suggested that overexpressed C8orf37 and FAM161A can be co-localized.

N-Terminal Region of C8orf37 Interacted with FAM161A
The C-terminal half of C8orf37 is highly conserved across multiple species [6]. Mutations within this region, C8orf37 (R177W) [2,6] and C8orf37 (Q182R) [6], are implicated in Bardet-Biedl syndrome (numbering 21) and retinitis pigmentosa (numbering 64), respectively. Thus, we hypothesized that this region may interact with FAM161A. One-by-one Y2H assays were performed to test this hypothesis. Coding sequences for full-length human C8orf37 (C8orf37-FL), full-length C8orf37 (R177W), and C8orf37 (Q182R), as well as C8orf37 fragment corresponding to residues 69-207 (C8orf37 (aa 69-207)) were used as bait fused to the LexA DNA binding domain and a FAM161A fragment corresponding to aa 317-549 was fused to the GAL4 activation domain. While the yeast transformed with an empty vector and neither C8orf37 nor FAM161A vectors grew in selective growth media, SMAD/SMURF positive control grew in stringent selective growth media lacking leucine, tryptophan, and histidine treated with 1mM 3-amino-1,2,4-triazole (-L -W -H +1mM 3AT) ( Figure 5B). As expected, yeast transformed with FAM161A (aa 317-549) and C8orf37-FL grew under stringent selective media. Surprisingly, yeast transformed with FAM161A and either C8orf37 (R177W) or C8orf37 (Q182R) mutants grew under the same conditions as well, indicating that these mutations did not affect interactions between C8orf37 and FAM161A. Further, C8orf37 (aa 69-207)/FAM161A (aa 317-549) did not grow under any stringent conditions. These results indicated that the C-terminal aa 69-207 residues of C8orf37 did not interact with FAM161A. Thus, we tested whether the N-terminal region of C8orf37 (aa 1-75) interacted with FAM161A likewise, with full-length C8orf37 as a positive control. Again, while the yeast transformed with an empty vector and neither full-length C8orf37 nor FAM161A (aa317-549) grew in selective growth media, full-length C8orf37/FAM161A positive control did grow. Yeast transformed with C8orf37 (aa 1-75) and the GAL4 activation domain empty vector grew under selective growth media, suggesting self-activation by aa 1-75 of C8orf37. However, growth of C8orf37 (aa 1-75)/FAM161A (aa 317-549) double-transformed yeast was much more robust in -L-W-H medium treated with both 1 mM 3AT and 5 mM 3AT ( Figure 5C). These results indicated that aa 1-75 of C8orf37 can interact with FAM161A.

Discussion
The precise function of C8orf37 in photoreceptor survival is not clear. C8orf37 protein was shown to be present at the base of the primary cilium in hTert-RPE1 cells [6]. In the mouse retina, C8orf37 was also found at the ciliary base and along the ciliary rootlet of photoreceptors [6]. C8orf37 knock-down in zebrafish results in defective formation of the Kupffer's vesicle, a ciliated developmental organ involved in establishing left-right asymmetry, as well as defects in retrograde melanosome transport [1]. These data suggest that C8orf37 may be involved in the function of the primary cilium. Ciliary expression of C8orf37 was contested in a recent publication [15]. Shariff, et al. [15] propose that C8orf37 is distributed throughout photoreceptor cells excluding the outer segments, as they found that GFP fluorescence in photoreceptors expressing EGFP-C8orf37 fusion protein was widely distributed throughout photoreceptors. Their data showed that overexpressed C8orf37 was localized to photoreceptor cilia in addition to the inner segments and cell bodies. Our results from the marmoset retina showed that C8orf37 was enriched at the ciliary base, present along the axonemes and often co-localized with microtubules in the inner segments of photoreceptors. In addition, C8orf37 was widely expressed throughout the rest of the neural retina, including the photoreceptor cell bodies in the ONL as well as cells in the INL and GCL. The data from Y2H assays, double immunofluorescence staining, and proximity ligation assays strongly supported that C8orf37 interacted with FAM161A. This interaction was not limited to photoreceptor cilia. It was also present in other cell layers of the neural retina. In mouse retina, FAM161A protein is localized at the basal bodies, connecting cilia and inner segments of photoreceptors as well as the OPL, the IPL and the GCL [12,13]. This study showed that, in the marmoset retina, the FAM161A expression pattern was similar to that of the mouse; however, it was also expressed in the ONL and INL. FAM161A associates with microtubules in COS-7, HeLa, hTERT-RPE1, and ARPE-19 cells [12,13]. In the marmoset retina, our data suggested that FAM161A co-localized with microtubules, as well. Thus, C8orf37-FAM161A interaction may play a role in microtubule function. Expression and interaction of C8orf37 and FAM161A in cells other than photoreceptors suggest that they function in other cells, in addition to photoreceptors. Our finding that C8orf37 interacts with FAM161A is consistent with the idea that C8orf37 and FAM161A function, at least in part, as ciliary proteins in photoreceptors.
C8orf37 knockout mice manifest photoreceptor degeneration [15]. The photoreceptors in these mice exhibit defective organization of outer segment discs. Interestingly, outer segment disc defects are often found in rodent models with mutations in outer segment proteins or ciliary proteins, such as peripherin/rds [16,17], GARPs and the beta-subunit of rod cGMP-gated channels [18], protocadherin 21 [19], RP1 [20], RPGR-interacting protein [21], and prominin 1 [22]. Photoreceptors in FAM161A mutant mice exhibit shortened cilia with malformed discs in the outer segment [23]. Disorganization of the outer segment discs in C8orf37 knockout mice is consistent with the hypothesis that C8orf37 plays a role in photoreceptor cilia or outer segments.
Alignment of C8orf37 proteins from several species indicates that~20 residues from the N-terminus and aa 100-207 of the C-terminal half are highly conserved (Figure 7). In addition to splicing and nonsense mutations that may abolish the function of C8orf37, the Q182R mutation is found in RP64 [6] while the R177W mutation is found in CORD16 [6] and BBS21 [2]. These residues are conserved in human, macaque, mouse, and Xenopus, as well as zebrafish C8orf37 orthologues (Figure 7), suggestive of their importance in C8orf37 protein function. Our Y2H assay results indicated that these mutations did not disrupt C8orf37-FAM161A interactions. In fact, our results indicated that the C-terminal aa 69-207 region of C8orf37 was not able to interact with FAM161A. Therefore, the pathogenicity of these two mutations is likely not dependent upon C8orf37-FAM161A interaction. Although no pathogenic mutations have been identified affecting the first 75 amino acid residues at the N-terminal end of C8orf37, our direct 1-by-1 Y2H assays revealed that the N-terminal aa 1-75 region on C8orf37 was sufficient and required for its interaction with FAM161A.

Yeast Two-Hybrid
The Y2H assays were performed at Hybrigenics Services SAS (Evry, France) on a fee-for-service basis. Coding sequence of full-length human C8orf37 (NM_177965.3) was cloned into pB27 derived from the pBTM116 vector [41] with the LexA DNA binding domain (DBD) as a LexA-bait fusion. The bait library was constructed from a human retina cDNA library cloned into the pP6 vector. In total, 50.5 million clones were screened.

Immunofluorescence Staining
Common marmoset eye samples were obtained at the Central Institute for Experimental Animals (CIEA), Kawasaki, Kanagawa, Japan. All handling of animals was in accordance to the guidelines published by the Institute for Laboratory Animal Research (Guide for the Care and Use of Laboratory Animals) and the US Public Health Service (Public Health Service Policy on Humane Care and Use of Laboratory Animals). All animal experiments at CIEA were approved by the Institutional Animal Care and Use Committee of the Central Institute for Experimental Animals (CIEA approval number: 18031A, 19033A).

Proximity Ligation Assay
For proximity ligation assay on marmoset retinal sections, rabbit antibodies against C8orf37 and FAM161A were conjugated with the Duolink In Situ Probemaker Plus (Sigma-Aldrich; cat#DUO92009-1KT) and the Duolink In Situ Probemaker Minus (Sigma-Aldrich; cat#DUO92010-1KT) kits, respectively. Proximity ligation assay was performed using the Duolink In Situ Detection Reagents Red kit (Sigma-Aldrich, cat# DUO92008) according to the manufacturer's recommendation. A Zeiss confocal microscope system (Zeiss LSM 780) was used to obtain images. To quantify PLA signals, images were exported to ImageJ "https://imagej.nih.gov/ij/ (accessed on 1 September 2015)". Each image was divided into multiple regions corresponding to the photoreceptor outer segments (OS), photoreceptor inner segments (IS), outer nuclear layer (ONL), outer plexiform layer (OPL), inner nuclear layer (INL), inner plexiform layer (IPL), and ganglion cell layer (GCL). Within each layer, the total number of PLA puncta was manually counted in ImageJ and normalized to a 100 µm retina.
For proximity ligation assay of epitope-tagged human C8orf37 and human FAM161A, transfected cells were fixed with 4% paraformaldehyde 48 h post-transfection. Rabbit anti-c-Myc and mouse anti-turboGFP (tGFP) (Origene Technologies, cat#TA150041, 1:500) primary antibodies were incubated with the slides in a procedure identical to immunofluorescence staining. After washing, proximity ligation assays (PLA) were performed using the Duolink In Situ Red Starter Kit Mouse/Rabbit (Sigma-Aldrich; Cat#DUO92101) according to the manufacturer's recommendations. Confocal images were obtained, as above. To quantify PLA puncta all images were exported to ImageJ for processing. The ImageJ manual thresholding function was used to adjust each image to eliminate background fluorescence in the PLA color channel. Following thresholding, PLA puncta were automatically counted using the "Analyze Particles" function in ImageJ with the minimum particle size set at 0.1 µm 2 . For each image, the number of PLA puncta was normalized to the number of nuclei visible. The experiments were performed in triplicate with the average number of PLA puncta per cell within each replicate consisting of results from five images.