Mutations in the FAM161A gene were previously identified as the cause for autosomal-recessive retinitis pigmentosa 28. To study the effects of Fam161a dysfunction in vivo, we generated gene-trapped Fam161aGT/GT mice with a disruption of its C-terminal domain essential for protein–protein interactions. We confirmed the absence of the full-length Fam161a protein in the retina of Fam161aGT/GT mice using western blots and showed weak expression of a truncated Fam161a protein by immunohistochemistry. Histological analyses demonstrated that photoreceptor segments were disorganized in young Fam161aGT/GT mice and that the outer retina was completely lost at 6 months of age. Reactive microglia appeared in the outer retina and electroretinography showed an early loss of photoreceptor function in 4-month-old Fam161aGT/GT animals. Light and electron microscopy revealed a remarkable phenotype of a significantly shortened connecting cilium, spread ciliary microtubule doublets and disturbed disk organization in Fam161aGT/GT photoreceptor cells. Co-immunolabeling experiments demonstrated reduced expression and mislocalization of centrin 3 and disturbed targeting of the Fam161a interactors lebercilin and Cep290, which were restricted to the basal body and proximal connecting cilium in Fam161aGT/GT retinas. Moreover, we identified misrouting of the outer segment cargo proteins opsin and rds/peripherin 2 in Fam161aGT/GT mice. In conclusion, our results suggest a critical role for the C-terminal domain of Fam161a for molecular interactions and integrity of the connecting cilium. Fam161a is required for the molecular delivery into the outer segment cilium, a function which is essential for outer segment disk formation and ultimately visual function.
Retinitis pigmentosa [RP (MIM 268000)] is the most prevalent hereditary degeneration of the human retina with an incidence of 1:4000 worldwide. Initial symptoms include night blindness and a gradual constriction of peripheral vision caused by rod photoreceptor death (1,2). Secondary loss of cone photoreceptors may subsequently lead to impairment of central vision and ultimately legal blindness (3). RP can be transmitted in an X-linked, autosomal-dominant or autosomal-recessive inheritance pattern. So far, mutations in at least 50 different genes have been identified as genetic causes for non-syndromic RP (RetNet, https://sph.uth.edu/retnet/). Genetic heterogeneity of RP is also reflected in the broad clinical variability in course and severity of the disease (1).
On the molecular level, proteins encoded by RP genes are involved in a multitude of cellular processes including the phototransduction cascade, cytoskeletal dynamics, regulation of gene transcription and intracellular trafficking of proteins (1,3). Due to the complex organization and high extent of specialization of photoreceptors, these cells are very vulnerable to stress. As a consequence, genetically induced dysfunctions of single photoreceptor gene products often lead to a loss of photoreceptor function and programmed cell death, a common hallmark in the pathogenesis of inherited retinal degenerations (4,5).
We and others have previously identified mutations in the FAM161A gene as a cause for autosomal-recessive retinitis pigmentosa 28 (RP28) (6,7). Homozygosity for a 685C-T transition in FAM161A exon 3 was identified in affected members from an Indian family, converting the codon for arginine 229 into a termination codon in FAM161A exon 3 (p.Arg229X). Other mutations causing frameshifts or premature termination have been identified in RP cohorts of German, Jewish, Palestinian and North American decent (6–9). Notably, except for one mutation in exon 4 (p.Arg596X), all of the identified mutations are located in the large exon 3 of FAM161A.
The orthologous human and mouse FAM161A genes consist of seven exons and the Fam161A protein is present in several tissues with high expression level in the retina (6). Fam161A protein is found in inner photoreceptor segments and co-localizes with ciliary marker proteins in the connecting cilium and the adjacent centriole of photoreceptor cells (10,11). FAM161A has been shown to interact with microtubules in vitro via its C-terminal part and a putative role in microtubule stabilization has been proposed (10,11). The FAM161A protein contains a highly conserved UPF0564 domain which is encoded by exon 3, and the N-terminal part of this domain is crucial for both homo- and heterotypic protein–protein interactions with the ciliopathy proteins Cep290 and lebercilin (Lca5) (11). In vitro data suggest that FAM161A is involved in transport processes across the photoreceptor cilium; however, it is currently unknown whether loss of FAM161A function affects ciliary structure in vivo.
To study the consequences of Fam161a disruption in the retina and to determine its role in photoreceptor cells in vivo, we generated and analyzed homozygous gene trapped Fam161a (Fam161aGT/GT) mice. We found that Fam161aGT/GT mice displayed progressive photoreceptor degeneration, early impairment of retinal signal processing and microglial activation. Further investigations of the histopathology revealed that Fam161aGT/GT retinas displayed a spread and shortened photoreceptor connecting cilium, mislocalization of Fam161a interacting proteins and disturbed outer segment formation. Our findings suggest that Fam161a plays a crucial role in the assembly and function of photoreceptor cilia and highlight the Fam161aGT/GT mouse line as a novel retinal ciliopathy model.
Generation of a mouse line with gene trap disruption of Fam161a
We generated a mouse line with targeted disruption of the Fam161a gene using a CMMR ES cell clone GT_462E7_5S with the UPA-vector (12) gene trap allele Fam161aGT(462E7)Chmd inserted into exon 3 of the gene (Fig. 1A). The insertion of the gene trap cassette disrupts the highly conserved UPF0564 domain after amino acid position 362 (Fig. 1B, Supplementary Material, Fig. S1). Mice heterozygous and homozygous for the targeted allele were identified by PCR genotyping (Fig. 1C). Mice homozygous for the Fam161aGT allele and hereafter referred to as Fam161aGT/GT mice showed no behavioral abnormalities compared with their heterozygous littermates. RT-PCR to amplify the 5′ and 3′ parts of the Fam161a mRNA did not show major expression differences in wild-type, Fam161a+/GT and Fam161aGT/GT retinas (Fig. 1D). This suggests that it is unlikely that Fam161aGT mRNA is subjected to complete nonsense-mediated decay.
Weak expression of a truncated Fam161a protein in the Fam161aGT/GT retina
Next, we studied Fam161a protein expression. First, we over-expressed a corresponding Fam161a (amino acids 1–363) variant with an EGFP tag in Hek293 cells (Fig. 2A). This truncated EGFP-tagged Fam161 was readily detected by staining with an antibody against an N-terminal epitope of Fam161a and retained its ability to localize to the microtubule cytoskeleton (Fig. 2B). The specificity of this antibody was further confirmed by immunoblotting, as it detected the truncated version as well as the full-length form of Fam161a at the expected sizes of 70 and 80 kDa, respectively (Fig. 2C). Next, we analyzed Fam161a protein expression in the retina by immunoblotting with two different anti-Fam161a antibodies (Fig. 2D, Supplementary Material, Fig. S1). We detected a specific band of 80 kDa in lysates from wild-type retinas which was absent in 1-month-old Fam161aGT/GT animals. However, we did not detect the truncated Fam161a protein in retinal lysates of Fam161aGT/GT mice, probably due to the low expression level of this polypeptide. Nevertheless, we studied Fam161a protein expression by immunohistochemistry in longitudinal retinal sections from 1-month-old wild-type, Fam161a+/GT and Fam161aGT/GT mice. Fam161a immunoreactivity was detected in inner segments of photoreceptors, the outer plexiform layer and the inner plexiform layer of wild-type mice confirming previous findings (10) (Fig. 2E). Retinas of heterozygous Fam161a mutants showed similar immunoreactivity in photoreceptor inner segments, but lower levels in the plexiform layers (Fig. 2E). Fam161aGT/GT mice showed strongly reduced levels of immunoreactivity in all retinal layers. Based on these observations, Fam161aGT/GT mice can be considered to weakly express a truncated form of Fam161a in the inner segment of retinal photoreceptor cells.
Fam161a gene-trapped mice show a severe and progressive degeneration of the outer retina
We then analyzed the morphology of retinal layers in Fam161aGT/GT mice at different ages using histology and spectral domain optical coherence tomography (SD-OCT) imaging. Longitudinal retinal sections of Fam161aGT/GT mice displayed a severe and progressive retinal degeneration (Fig. 3A–G), involving a gradual loss of the outer nuclear layer which was almost absent in 5-month-old Fam161aGT/GT mice (Fig. 3F). Higher magnification microscopy further revealed that Fam161aGT/GT retinas lacked clear photoreceptor inner/outer segment junctions and displayed disorganized photoreceptor outer segments already at 1 month of age (Fig. 3H and I). Thinning of the outer retina was accompanied by disturbance of the structural order and the appearance of pyknotic photoreceptor nuclei (Fig. 3J–N). In later stages, the outer retina had completely disappeared, whereas the inner retina remained largely unaffected (Fig. 3G and N). Progressive retinal degeneration of Fam161aGT/GT mice was also observed using SD-OCT imaging with early absence of the defined layer between inner and outer segments (Supplementary Material, Fig. S2A). Further quantitative morphometric analyses along the nasal/temporal axis revealed that Fam161aGT/GT animals have a significantly thinner outer nuclear layer already at 1 month compared with wild-type littermates (Supplementary Material, Fig. S2B). At 2 months of age, retinal thickness had declined to 50% until more than 90% of the outer nuclear layer were lost after 6 months in Fam161aGT/GT retinas (Supplementary Material, Fig. S2B).
Degeneration of the Fam161aGT/GT retina is related to microglial reactivity
As microglial activation is a hallmark of inherited retinal degeneration, the morphology and spatial distribution of retinal microglial cells were examined in Fam161aGT/GT mice. The morphology of Iba1-positive microglial cells in wild-type retinas is characterized by small cell bodies and long ramified processes (Fig. 4A), with the cells being distributed in a highly ordered array throughout the plexiform layers (Fig. 4B). In 2-month-old Fam161aGT/GT mice, microglia became bloated with shortened protrusions and a significant number of cells migrated to the outer retina (Fig. 4A and B). Reactive microglia then increasingly accumulated in the subretinal space of older Fam161aGT/GT mice until the territorial microglial network was completely disrupted in 5-month-old animals (Fig. 4A and B). To specify this glial reactivity in relation to cell death, we performed expression profiling using qRT-PCR. The retinal apoptosis marker Caspase 8 (Casp8) was significantly elevated in Fam161aGT/GT retinas from 4 weeks and showed a peak at 5 months (Supplementary Material, Fig. S3A). The microglia markers activated microglia whey acidic protein (Amwap) and Cd68 indicated initiation of microglial reactivity between 2 and 3 weeks of age with maximum levels in 2-month-old Fam161aGT/GT retinas (Supplementary Material, Fig. S3B and C). The Müller glia cell and astrocyte marker Gfap was also significantly up-regulated at all time points and showed a maximum of expression between 6 and 8 weeks after birth (Supplementary Material, Fig. S3D). These data suggest that initiation of immune activation is tightly associated with the onset of photoreceptor degeneration in Fam161aGT/GT retinas.
Fam161a-associated retinal degeneration leads to an early loss of visual function
To test the impact of the observed photoreceptor degeneration on retinal function, we recorded scotopic and photopic electroretinograms (ERGs) from Fam161aGT/GT mice and age-matched controls. Fam161aGT/GT mice exhibited abnormal scotopic ERG responses already at 1 month of age, and at 4 months, ERG responses were almost undetectable (Fig. 5A). We then analyzed the residual visual function in 1-month-old Fam161aGT/GT mice in detail (Fig. 5B–E). The dark-adapted ERG response waveforms (Fig. 5B) of Fam161aGT/GT animals showed lower a- and b-wave amplitudes (Fig. 5C) and prolonged implicit times (Fig. 5D) for all flash intensities. However, the elevated b/a-wave amplitude ratio (Fig. 5E) indicates a photoreceptor-dominated degeneration. Light-adapted ERGs showed a less profound amplitude reduction (Fig. 5F and G) and prolonged implicit times (Fig. 5H) in Fam161aGT/GT animals compared with control mice.
Spreading of the connecting cilium and aberrant disk organization in Fam161aGT/GT photoreceptor cells
Fam161a has been previously localized to the cilium of rod photoreceptor cells (10,11). Here, co-staining of FAM161A in the human retina with the cone marker peanut agglutinin revealed FAM161A localization not only in cilia of rods, but also of cone photoreceptor cells (Supplementary Material, Fig. S4). We therefore focused on the analysis of photoreceptor cilia in Fam161aGT/GT mice applying transmission electron microscopy (TEM). Comparative TEM analysis of Fam161aGT/GT and wild-type photoreceptor cells revealed the following striking morphological changes in Fam161aGT/GT mice (Fig. 6): (i) the connecting cilium was significantly reduced in length (WT: ∼ 1.5 µm; Fam161aGT/GT: ∼ 1.1 µm), (ii) the characteristic microtubule doublets of the connecting cilium were spread in the proximal half thereby, (iii) the outer segment base was markedly dilated (Fig. 6B), and (iv) the membrane disk stacks in the outer segment were distorted, often arranged perpendicular to the disk stack orientation in wild-type photoreceptors (Fig. 6A). In contrast, the structural arrangements of the basal body and the adjacent centriole of the cilium were not altered and no ultrastructural changes were obvious in the inner segment and at the synapses of Fam161aGT/GT photoreceptor cells (Fig. 6 and data not shown). These structural Fam161aGT/GT phenotypes were identified as early as 4 weeks of age. At that time point, alterations in ciliary structures were observed in 95% of all photoreceptors while the disk phenotype was present only in 60% of outer segments. Later at 10 weeks, all photoreceptor cells in Fam161aGT/GT mice showed defective cilia and almost all photoreceptor outer segments were altered.
Altered distribution of ciliary molecules in spread photoreceptor cilia of Fam161aGT/GT mice
Next, we studied whether spreading of microtubule doublets in the connecting cilium and structural disruptions of the outer segments were associated with molecular changes in the ciliary compartment of Fam161aGT/GT photoreceptor cells. We first analyzed the localization of the truncated Fam161a protein in the ciliary compartment. In wild-type mice, Fam161a was present in the connecting cilium, the basal body and the adjacent centriole (Fig. 7A and B) confirming our previous studies (6,10), a report from rat retina (11), and our data for human rod and cone photoreceptor cells (Supplementary Material, Fig. S4). Fam161a co-localized with centrin3, a marker for the adjacent centriole, the basal body and the connecting cilium and with acetylated-tubulin, a marker for stabilized microtubules of cilia and centrioles (Fig. 7A and B, left panels). All ciliary marker stains confirmed the shortening of the connecting cilium in Fam161aGT/GT photoreceptor cells (Fig. 7). Co-labeling for Fam161a and centrin3 or acetylated-tubulin revealed that the truncated Fam161a protein was restricted to the basal body and the adjacent centriole of the photoreceptor cilium (Fig. 7A and B, right panels). However, Fam161a staining was absent in the connecting cilium of Fam161aGT/GT photoreceptor cells. In contrast, acetylated-tubulin remained in all compartments of the cilium, namely the axoneme, the shortened connecting cilium, the basal body and the adjacent centriole of the Fam161aGT/GT photoreceptor cilium (Fig. 7C–E). Centrin3 also remained at the two centrioles of the cilium but was significantly reduced in the connecting cilium of Fam161aGT/GT photoreceptors (Fig. 7A, B, D, E). The retraction of centrin3 to the proximal part of the connecting cilium was more evident in double labeling experiments for centrin3 and glutamylated-tubulin, both markers for the adjacent centriole, the basal body and the connecting cilium in wild-type mice (Fig. 7F). In contrast to glutamylated tubulin, which was stained all along the shortened connecting cilium, centrin3 was found only at its base in Fam161aGT/GT photoreceptor cells (Fig. 7F).
Next, we tested whether the molecular composition of the photoreceptor axoneme, which projects from the tip of the connecting cilium into the photoreceptor outer segment, was altered in Fam161aGT/GT photoreceptor cells. We performed immunolabeling of Rp1 and BBS5, two molecules which have been previously described as molecular components of the axoneme in photoreceptor outer segments (14,15). Immunofluorescence analysis revealed that both proteins accumulated at the tip of the connecting cilium of Fam161aGT/GT photoreceptors (Fig. 7G–I).
To specify our results on the disruption of the connecting cilium at higher resolution and by molecular means, we applied immunoelectron microscopy. We labeled centrin3 and acetylated tubulin on ultrathin LR White sections through wild-type and Fam161aGT/GT photoreceptor cells (Supplementary Material, Fig. S5). Our analysis confirmed that centrin3, which is normally present along the entire connecting cilium of photoreceptor cells (Supplementary Material, Fig. S5A) (16), was restricted to the lower part of the connecting cilium in Fam161aGT/GT photoreceptor cells (Supplementary Material, Fig. S5B and C). In the apical region, where the microtubule doublets spread, no centrin3 was detected (Supplementary Material, Fig. S5B and C). In contrast, acetylated tubulin was also present in the spread microtubules projecting into the remains of the outer segments of Fam161aGT/GT photoreceptors (Supplementary Material, Fig. S5D–F). Quantification of the silver-enhanced immunogold labeling of centrin3 revealed that the total density of centrin3 per cilium was significantly reduced (Supplementary Material, Fig. S5G). Furthermore, our quantitative analysis demonstrates that the spatial distribution of centrin3 over the ciliary compartments was severely altered in Fam161aGT/GT cilia (Supplementary Material, Fig. S5H), confirming our immunofluorescence data.
Fam161a-interacting partners are mislocalized in Fam161aGT/GT photoreceptor cells
A recent study demonstrated the interaction of Fam161a with the ciliopathy proteins lebercilin (Lca5) and Cep290 (11). The interaction of both proteins is mediated by the C-terminal domain mostly deleted in Fam161aGT/GT mice. To analyze the consequences of Fam161a deletion/truncation for the distribution of these disease proteins, we double-labeled specimens for the ciliary marker centrin3 and lebercilin or Cep290, respectively. In wild-type photoreceptors, centrin3 and lebercilin co-localized in the basal body and the connecting cilium (Fig. 8A, left panel). Lebercilin was also present in the axoneme of the outer segment (Fig. 8A, left panel). In contrast, lebercilin was absent from the axoneme and the connecting cilium in Fam161aGT/GT photoreceptors (Fig. 8A, right panel). Its co-localization with centrin3 was restricted to the basal body of Fam161aGT/GT cilia (Fig. 8A, right panel). Cep290 also co-localized with centrin3 in the connecting cilium of wild-type mice (Fig. 8B, left panel). In Fam161aGT/GT photoreceptor cells, Cep290 exclusively localized to the proximal connecting cilium and the basal body (Fig. 8B, right panel). The fact that both interacting partners of Fam161a mislocalized in the photoreceptor cilia of Fam161aGT/GT mice (Fig. 8C) indicates that interaction with Fam161a is essential for their correct position in the connecting cilium of photoreceptor cells.
The ciliary localization of the IFT complex b transport module is not altered in Fam161aGT/GT photoreceptor cells
The assembly and maintenance of cilia requires intraflagellar transport (IFT), a process mediated by molecular motors and IFT proteins organized into supramolecular protein complexes (17). Previous studies showed that IFT proteins are associated with photoreceptor cilia (18,19). This prompted us to also examine whether IFT proteins were affected in the cilium of Fam161aGT/GT photoreceptor cells. However, immunostainings of IFT20, IFT57 and IFT88 revealed that none of these IFT molecules was mislocalized in Fam161aGT/GT retinal photoreceptor cilia (Fig. 8D–G).
Spreading of Fam161aGT/GT photoreceptor cilia impairs transport to outer segment proteins
We next investigated the functional consequences of the ciliary phenotype observed in Fam161aGT/GT photoreceptor cells. Therefore, we stained cryosections through the retina for the outer segment proteins opsin and rds/peripherin2. Most opsin is normally localized within photoreceptor outer segments of wild-type mice (Fig. 9A). In contrast, Fam161aGT/GT retinal opsin was found in the inner segments, the outer nuclear layer and at photoreceptor synapses (Fig. 9B). The higher resolution of anti-opsin immunoelectron microscopy then revealed the presence of opsin in the cytoplasmic and internal membranes of the inner segment of photoreceptor cells (Fig. 9C), the cytoplasmic membrane facing the perinuclear cytoplasm of rod nuclei (Fig. 9D), and in the synaptic terminals of rod cells (Fig. 9E). At the same developmental stage, the amount of rds/peripherin2 was reduced to ∼48% in photoreceptor outer segments (Fig. 9F and G). Furthermore, a mislocalization of rds/peripherin2 to synapses of the outer plexiform layer was obvious in Fam161aGT/GT retinas (Fig. 9E and G).
We and others have earlier reported mutations in the FAM161A gene as one genetic cause for retinitis pigmentosa (6–9). In vitro biochemical studies have indicated that Fam161a may contribute to the function of the photoreceptor cilium (10,11). To gain insights into the molecular pathomechanism underlying RP28, we continued with the functional characterization of Fam161a in vivo using a novel Fam161a gene trapped mouse line.
Photoreceptor cell death and microglial activation represent common pathological features in several retinal degeneration models (4,20,21). Fam161aGT/GT retinas showed a progressive loss of photoreceptor cells and had severely impaired visual function. In the Fam161aGT/GT retina, microglia also changed their morphology and migrated to the outer retina. Reactive microglial cells can secrete neurotoxic substances which then accelerate photoreceptor death and influence the function of retinal pigment epithelial cells (22–24). Transcript profiling showed that cell death in Fam161aGT/GT retinas was coincident with the induction of microglial activation. Notably, the peak of gliosis was reached between 1 and 2 months of age, which corresponds to the period with the steepest rate of cell loss and the time with most decline of retinal function. We therefore conclude that glial reactivity parallels retinal degeneration in the Fam161aGT/GT retina.
The overt accumulation of opsin in photoreceptor inner segments, around photoreceptor nuclei and at photoreceptor synapses in Fam161aGT/GT mice may trigger photoreceptor cell death. The same phenomenon was observed in Rp1-deficient or Lca5Gt/GT gene trapped mice, two ciliopathy models in which defective ciliary transport also leads to the accumulation of opsin in the outer nuclear layer (14,25). Moreover, rd16 mice, which express a truncated version of the ciliary protein Cep290 in the retina, display abnormal outer segment structure, retention of outer segment proteins and rapid degeneration (26,27). Photoreceptors may tolerate this to some extent but mislocalization of truncated rhodopsin to membranes of the inner segment favors cell death (28). A recent study raised some concerns that photoreceptor death in several RP mouse models may be caused by proteasomal insufficiency due to misfolded proteins (29). In this study, we did not observe accumulation of ubiquitinylated proteins in Fam161aGT/GT retinas (data not shown) and thus exclude proteasomal dysfunction as a trigger of cell death.
In Fam161aGT/GT mice, the layered retinal architecture is preserved similarly to other retinal ciliopathy models (14,25,30). However, our analyses revealed a significant reduction of connecting cilium length and spreading of microtubule doublets and thereby distortion of outer segment disks at early stages of degeneration. These results provide in vivo evidence for an essential role of Fam161a in the structural composition, maintenance and function of the connecting cilium of photoreceptor cells. Our analyses showed that the gene-trapped N-terminal part of Fam161a targeted to microtubules and to the basal body, but not to the connecting cilium of photoreceptor cells. Interestingly, acetylation and glutamylation of microtubules in the cilium were not reduced, indicating that microtubules were not destabilized in Fam161aGT/GT mice. This is in line with in vitro studies linking the N-terminal part of Fam161a with microtubule binding and stabilization (10,11).
Recent cell culture data indicated that Fam161a is integrated into a network of ciliary disease proteins present in the transition zone of cilia (11). The transition zone is localized between the basal body and the axoneme and its overlapping protein modules play a crucial role in the regulation of ciliary transport (31,32). This strategic zone is homolog to the connecting cilium of photoreceptor cells (33,34). The ciliary phenotype of Fam161aGT/GT mice could therefore result from the disruption of these protein modules. The truncated version of Fam161a lacks the C-terminal interaction sites for Cep290 and lebercilin and thereby their connection to Fam161a is probably disturbed. Indeed, both proteins do not correctly localize to the connecting cilium in Fam161aGT/GT mice, indicating that their interaction with Fam161a is crucial for their correct ciliary position. Cep290 normally tethers the transition zone microtubule pairs to the membrane (35) and interacts with the Bardet-Biedl Syndrome protein MKKS (36). The disruption of this connection may thus underly the spreading of microtubule doublets in the connecting cilium in Fam161aGT/GT mice. Fam161aGT/GT photoreceptors were also devoid of lebercilin in the connecting cilium. Lebercilin and Fam161a are associated with the IFT module, which is essential for assembly and maintenance of cilia and flagella (25,37). However, like the absence of lebercilin (25), Fam161a inactivation did not affect the ciliary localization of IFT proteins, indicating that both proteins do not serve as core components of IFT particles. Nevertheless, the molecular interplay between Fam161a, Cep290 and lebercilin may provide a structural framework for correct organization of the connecting cilium and delivery of cargo molecules to outer segments.
Our data from Fam161aGT/GT photoreceptor cells also indicate a widening of the outer segment basis, where novel membrane disks containing all components of the visual cycle are generated (38,39). These outer segment alterations suggest that Fam161a may participate in disk neogenesis. Since we did not find Fam161a at the rims of the outer segments, we exclude its direct involvement in disk stacking. A comparable phenotype has been described in the Prph2Rd2 mouse, where the disruption of the disk rim protein peripherin-2 causes defective disk organization (40). In accordance, our data show a reduction of Prph2 in Fam161aGT/GT photoreceptor outer segments, indicating that Fam161a may participate in the delivery of peripherin-2 to the site of disk formation.
In conclusion, we have established the Fam161aGT/GT mouse as a novel retinal degeneration model with a remarkable phenotype of a shortened connecting cilium and spread ciliary microtubule doublets. Mislocalization of essential transition zone components, misrouting of outer segment cargo and disturbed disk organization in Fam161aGT/GT mice implicate that Fam161a is crucial in the structural composition, maintenance and function of the connecting cilium. The Fam161aGT/GT mouse line may be very useful to reveal further insights into Fam161a containing protein modules and into photoreceptor ciliary biology in general.
MATERIALS AND METHODS
Mice were kept in an air-conditioned environment on a 12-h light-dark schedule at 22°C, and had free access to food and water. All procedures complied with the German Law on Animal Protection and the Institute for Laboratory Animal Research Guide for the Care and Use of Laboratory Animals, 2011. When necessary, mice were anesthetized by intraperitoneal or subcutaneous injection of ketamine (150 mg/kg) and xylazine (10 mg/kg). Pupils were dilated using 1% tropicamide and 2.5% phenylephrine.
Human donor eyes were obtained from the Department of Ophthalmology, University Medical Center Mainz, Germany. The guidelines to the declaration of Helsinki were followed.
Fam161a gene trapped mouse
The stem cell gene trap clone CMHD-GT_462E7_5S was obtained from the Canadian Mouse Mutant Repository. ES cell DNA was screened for correct gene trap insertion at the Fam161a locus by Sanger sequencing. ES cells were microinjected into C57BL/6J blastocysts and transferred to pseudopregnant female CD1 mice to generate chimeras. Germline transmission of the Fam161a gene trap allele was checked by genotyping and gene-trap positive males were bred with C57BL/6J females. The subsequent progeny was intercrossed to obtain homozygous Fam161aGT/GT mice. Genotyping was performed by PCR with DNA isolated from tail tips using the primers GT1-F (5′-TCCCTGAAGCCCTTTAAGTTC-3′), GT2-R (5′-AAAGCATTGGGCATTTCAGA-3′), GT5-F (5′-CGGACAGACACAGATAAGTTGC-3′) and GT6-R (5′-AAAGCATTGGGCATTTCAGA-3′). Primer pairs GT1-F/GT2-R (485 bp fragment) and GT5-F/GT6-R (929 bp fragment) were used to detect the gene-trapped sequence. The wild-type sequence was detected using the primers GT1-F/GT6-R (385 bp fragment). Prior to breeding, founder animals were tested negative for the rd1, rd8 and rd10 alleles as described earlier (13,41,42).
RNA isolation and RT-PCR
Total RNA was extracted from murine retinas using the RNeasy Micro Kit (Qiagen, Hilden, Germany). cDNA synthesis was performed with the RevertAidTM H Minus First Strand cDNA Synthesis Kit (Fermentas, St-Leon-Roth, Germany). RT-PCR was performed with 50 ng cDNA to amplify intron-spanning fragments between Fam161a exons 2–3 (329 bp) and exons 4–5 (267 bp) using primers Ex2-3_F (5′-TGGCATTCGAGAGCACTATG-3′), Ex2-3_R (5′-CCCGGGAGTCTTTTCCTCTA-3′), Ex4-5_F (5′-AGGAGTGAAAAGGCCAGGAT-3′) and Ex4-5_R (5′-TTTCTCCAGTGTGCAGCTTCT-3′). A 292 bp fragment of β-actin was amplified as the reference transcript with primers β-actin_F (5′-ACCCACACTGTGCCCATCTA-3′) and β-actin_R (5′-CGGAACCGCTCATTGCC-3′). PCRs were carried out using the Qiagen Taq Core kit (Qiagen) and standard PCR conditions.
Quantitative real-time RT-PCR
Amplifications of 50 ng cDNA were performed on the ABI7900HT system (Applied Biosystems) in 10 µl reactions containing 1xTaqMan Universal PCR Master Mix (Applied Biosystems), 200 nm of primers and 0.25 µl of dual-labeled probe (Roche ProbeLibrary, Roche Applied Science). Measurements were performed in triplicates and results were analyzed with an ABI sequence detector software version 2.3 using the ΔΔCt method for relative quantification.
Cloning and recombinant protein expression
For expression in Hek293 cells, EGFP-mFAM161A amino acids 1–363 were subcloned from full-length EGFP-mFAM161A plasmid (10) using primer pair 5′-AAG CTT TGA TGG CTT CGC CGC ACC-3′ and 5′-GTC GAC TTA CGC AGC TGG CCT GTA AAT AAA-3′; the sequence was verified by direct sequencing.
Mouse retinal tissue or Hek293 cells were homogenized in cold RIPA buffer using a TissueLyser LT (Qiagen) and protein concentrations were determined by Bradford assay (Roti-quant, Roth, Karlsruhe, Germany). Fifty micrograms of total proteins were separated by SDS–PAGE on 10% gels and subjected to western blotting. Membranes were incubated with a primary antibody against Fam161a as described previously (10) (Fam161a ab1), a commercial anti-Fam161a antibody (ab115810, Abcam, Cambridge, UK) (Fam161a ab2) and anti-Actin antibody (sc-1616, Santa Cruz Biotechnology, Santa Cruz, CA, USA). Blots were visualized with chemiluminescence signals using secondary IgG-HRP antibodies (Santa Cruz Biotechnology).
Spectral-domain optical coherence tomography (SD-OCT)
SD-OCT was performed on a Spectralis HRA + OCT device (Heidelberg Engineering GmbH, Dossenheim, Germany). Mice were placed on a custom-made mounting platform for SD-OCT measurements (λ = 870 nm; acquisition speed, 40 000 A-scans per second; average images per scan, 24). SD-OCT volume scans of 61 B-scans with 70 µm distance between B-scans (human dimension) were performed on the upper quadrant of the globe for all eyes; this corresponds to 23.33 µm distance between B-scans for murine eyes.
Mouse ERGs were performed as described previously (43). Statistically significant difference between groups was calculated by one-way ANOVA, followed by multiple t-test using the Holm–Sidak method. All analyses and plotting were processed with R 2.15.2 and ggplot 0.9.2.
Semi-thin (1 µm) sections for light microscopy were prepared along the nasal-temporal plane of the retina and stained with Richardson's stain as described previously (44). Immunohistochemistry was performed on 10 µm thick horizontal retinal cryo-sections prepared from eyes embedded in optimal cutting temperature (OCT) compound (Hartenstein, Wurzburg, Germany). For subcellular analysis, eyes were cryofixed in melting isopentane and cryosectioned as described elsewhere (45). The following antibodies were used: rabbit anti-Iba1 (Wako Chemicals, Neuss, Germany), rabbit anti-Fam161a ab1, mouse monoclonal anti-Rhodopsin (Invitrogen), mouse anti-Prph2 (5H2; provided by M. Biehl, Munich), rabbit anti-IFT20 and rabbit anti-IFT57 (provided by G. Pazour, Worcester, MA, USA), rabbit anti-IFT88 and rabbit anti-lebercilin (Proteintech), mouse anti-Cep290 (3G4, provided by Chen and Shou), rabbit anti-BBS5 (provided by M. Nachury, Stanford, CA, USA), chicken anti-Rp1 (provided by E. Pierce, Boston, MA, USA), rabbit anti-glial fibrillary acidic protein (Sigma), mouse anti-acetylated α-tubulin (Sigma), mouse monoclonal anti-centrin3 (16), rabbit polyclonal anti-centrin3 antibodies and fluorescein-labelled lectin peanut agglutinin. Samples were analyzed with a Leica DM 6000 B microscope (Leica Microsystems, Bensheim, Germany) and images were processed with Adobe Photoshop CS (Adobe Systems, San Jose, CA, USA).
For retinal whole mounts, eyes were enucleated and fixed in 4% formaldehyde for 4 h before dissection. Isolated retinas were permeabilized with a solution of 25% Triton X-100 and 25% Tween 20 in 1x PBS followed by blocking in BLOTTO solution for 1 h (1% milk powder and 0.01% in Triton X-100 in 1x PBS). Retinas were incubated overnight with rabbit anti-Iba1 (1:500) in antibody solution (2% BSA, 0.02% NaN3 0.1% Triton X-100 in 1x PBS) at 4°C followed by 1 h secondary antibody incubation. Whole mounts and cross-sections were mounted in Dako fluorescent mounting medium (DakoCytomation GmbH, Hamburg, Germany) and imaged on an Axioskop2 MOT Plus Apotome microscope (Carl Zeiss).
Fixation for conventional electron microscopy
Enucleated eyes were fixed in 2.5% glutaraldehyde in 0.1 m cacodylate buffer containing 0.1 m sucrose for 1.5 h at 4°C. After 30 min fixation, the cornea and lens were removed and the eye cups were fixed for additional 1 h. Eye cups were then washed with 0.1 m cacodylate buffer containing 0.1 m sucrose for 30 min. Subsequently, eye cups were fixed with 2% osmium tetroxide (OsO4) in 0.1 m cacodylate buffer containing 0.1 m sucrose for 1 h at room temperature, followed by dehydration in ethanol (30–100%) and embedding in Renlam® M-1 resin (Serva Electrophoresis, Heidelberg, Germany).
Fixation and postembedding labeling for immunoelectron microscopy
Mouse retinas were fixed 0.1% glutaraldehyde and 3% paraformaldehyde in 0.1 m phosphate buffer (pH 7.4). Fixed tissue was dehydrated stepwise in 30–98% ethanol, embedded in LR White hard (Science Services, Munich, Germany), polymerized at 4°C under ultraviolet (UV) light for 48–60 h and postembedding immunolabeling was performed as described previously (46). In brief, ultrathin LR White sections were incubated with mouse anti-acetylated α-tubulin (Sigma), mouse monoclonal anti-centrin3 (16) or mouse monoclonal anti-opsin (clone K16-155) antibodies. A secondary goat anti-mouse Fab conjugated to Nanogold™ (Nanoprobes, Stony Brook, NY, USA) was applied in IgG-gold buffer. After postfixation in 2% glutaraldehyde for 10 min, the Nanogold™ labeling was silver-enhanced for 25 min.
Ultrathin sectioning and transmission electron microscopy
Ultrathin sections were made using a Reichert Ultracut S ultramicrotome (Leica), collected on Formvar-coated copper or nickel grids and counterstained with 2% uranyl acetate in 50% ethanol aund aq. 2% lead citrate. Ultrathin sections were analysed in a Tecnai 12 BioTwin transmission electron microscope (FEI, Eindhoven, The Netherlands). Images were obtained with a CCD camera (charge-coupled-device camera; SIS MegaView3; Surface Imaging Systems, Herzogenrath, Germany) and processed with Adobe Photoshop CS (Adobe Systems).
Funds were provided by the Deutsche Forschungsgemeinschaft (DFG, LA1203/8-1 to T.L., GRK 1044 to U.W.); the ProRetina Stiftung (to T.L., U.W.); the FAUN-Stiftung (to U.W.); European Community FP7/2009/241955 (SYSCILIA) (to U.W.) and FP7/2009/242013 (TREATRUSH) (to U.W.); BMBF, grant 0314106 (HOPE2) (to U.W.); the ERA-Net for Research on Rare Diseases (EUR-USH) (to K.N.W.), and the Hans und Marlies Stock-Stiftung (to T.L.).
The authors thank the members of the Langmann lab, Wolfrum lab, Tamm lab and Stoehr lab for excellent technical assistance. We thank Bernd Kirchhof for helpful discussions and support.
Conflict of Interest statement. None declared.