Small GTPases Rab8a and Rab11a Are Dispensable for Rhodopsin Transport in Mouse Photoreceptors

Rab11a and Rab8a are ubiquitous small GTPases shown as required for rhodopsin transport in Xenopus laevis and zebrafish photoreceptors by dominant negative (dn) disruption of function. Here, we generated retina-specific Rab11a (retRab11a) and Rab8a (retRab8a) single and double knockout mice to explore the consequences in mouse photoreceptors. Rhodopsin and other outer segment (OS) membrane proteins targeted correctly to OS and electroretinogram (ERG) responses in all three mutant mouse lines were indistinguishable from wild-type (WT). Further, AAV (adeno-associated virus)-mediated expression of dnRab11b in retRab11a-/- retina, or expression of dnRab8b in retRab8a-/- retina did not cause OS protein mislocalization. Finally, a retRab8a-/- retina injected at one month of age with AAVs expressing dnRab11a, dnRab11b, dnRab8b, and dnRab10 (four dn viruses on Rab8a-/- background) and harvested three months later exhibited normal OS protein localization. In contrast to results obtained with dnRab GTPases in Xenopus and zebrafish, mouse Rab11a and Rab8a are dispensable for proper rhodopsin and outer segment membrane protein targeting. Absence of phenotype after expression of four dn Rab GTPases in a Rab8a-/- retina suggests that Rab8b and Rab11b paralogs maybe dispensable as well. Our data thus demonstrate significant interspecies variation in photoreceptor membrane protein and rhodopsin trafficking.


Introduction
Mammalian photoreceptors are polarized neurons, each consisting of an outer segment (OS), connecting cilium (CC), inner segment (IS), nucleus and synaptic terminal. The OS is a modified primary cilium that houses the phototransduction machinery, including the visual pigment rhodopsin, its G protein transducin, the target enzyme cGMP phosphodiesterase 6 (PDE6) and cGMP-gated channel subunits (CNG). Rhodopsin is the most abundant OS membrane protein [1]. Mature murine photoreceptors constantly renew OS membrane (~10% per day) by adding newly synthesized disk membranes at the OS base and removing old disks at the OS tip by phagocytosis [2;3]. As rhodopsin and other OS proteins are synthesized in the IS, photoreceptors require a sophisticated and highly directed trafficking pathway to compensate for OS protein turnover.
Much information concerning rhodopsin trafficking has been elucidated in leopard frog (Rana berlandieri) photoreceptors because of their large size [4;5]. Following biosynthesis and posttranslational modifications at the Golgi, rhodopsin is transported in vesicles from the trans-Golgi network (TGN) to the distal IS membrane where cargo is assembled for intraflagellar transport to the OS [6]. Rab GTPases play critical roles in guiding and docking vesicles to the target membranes. Rab proteins are small (21-25 kD), lipidated, GTP-binding and -hydrolyzing molecules present in diverse organisms from yeast to human [7]. Rab C-terminal halves contain variable regions that end with prenylation motifs signaling protein geranylgeranylation for membrane association [8]. These proteins participate in vesicular transport, docking and fusion of transport vesicles with their targets [7]. Rab8 and Rab11 each exist as two paralogs, a and b [9;10]. Rab11a and Rab8a have been shown to control rhodopsin transport in Xenopus laevis [11;12]. Budding of frog rhodopsin vesicles from TGN was suggested to be regulated by the GTPase Arf4 and its associated GAP, ASAP1 [4;13]. In this model, Rab11a is recruited to the vesicle surface by interacting with rhodopsin and ASAP1 during budding, and Rab11a reciprocally recruits Rab8a by recruiting Rabin 8, a Rab8 GEF (GDP-GTP exchange factor) [13;14]. Docking and fusion of rhodopsin vesicle at distal IS plasma membrane appears dependent on Rab8a, as transgenic expression of a dnRab8a (Rab8aT22N) causes massive rhodopsin-containing vesicle accumulation at the CC base and rapid retinal degeneration in Xenopus larva [11]. Expression of either dnRab11a or Rab11a shRNA has a similar effect in Xenopus retina [12]. Rab8 localization and Rab8 levels are often altered among various retina ciliopathies and are interpreted as a contributing factor of disease in mouse and zebrafish models [15][16][17]. In zebrafish, morpholino knockdown of Rab8a leads to rhodopsin mislocalization and OS shortening [15]. Rab8a has also been shown linked to intraflagellar transport in zebrafish through interacting with Rabaptin5 that interacts with IFT54 (intraflagellar transport protein 54 homolog) [18]. Notably, roles of Rab11 and Rab8 in rhodopsin transport are conserved across species. In Drosophila, Rab11 associates with rhodopsin vesicles and genetic ablation of Rab11 causes cytoplasmic accumulation of rhodopsin with failed rhabdomere morphogenesis [19], an effect that involves myosin V and an actin network [20].
In mammalian cell culture, the Rab11a-Rab8a cascade also appears to play a general role in primary ciliogenesis by delivering and docking secretory vesicles to the periciliary plasma membrane via interaction with the exocyst, as shown by disruption of Rab8a or Rab11a function by dominant negative protein expression or by RNAi knock-down [21][22][23][24][25]. Activated Rab8a also enters the cilium during ciliogenesis but its transport to the cilium decreases thereafter and the functional significance of Rab8-GTP in cilia is currently unknown [21;24]. In C. elegans, mutations in Rab8 suppresses formation of the membranous fans at the tip of AWB dendrite and constitutively active Rab8 results in shortened cilia, suggesting that Rab8 may inhibit cilium length but promote ciliary membrane genesis [26].
In mammals, however, the role of Rab11 and Rab8 in rhodopsin transport has not been directly tested using molecular genetics. Here we generated Rab11a KO-, Rab8a KO-, as well as Rab11a and Rab8a double KO mouse lines and found that rhodopsin and other major phototransduction proteins traffic correctly. Further, expression of dnRab11b in Rab11a KO retina, or expression of dnRab8b in Rab8a KO retina, does not cause rhodopsin mislocalization. Simultaneously expression of dnRab8b, dnRab10, dnRab11a, and dnRab11b in retina-specific Rab8a-/-retina does not impair OS ciliary targeting of rhodopsin. These results suggest that in contrast to Xenopus and zebrafish, Rab11a and Rab8a, and likely their paralogs as well, are nonessential for targeting rhodopsin to the mouse outer segment. Further, as outer segments develop normally in Rab8a and11a GTPase mutants, Rab11a and Rab8a appear dispensable for photoreceptor ciliogenesis.

Materials and Methods
Animals C57BL/6, flippase (FLP) and EIIa-Cre transgenic mice were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). Transgenic iCre75 were generated at the University of Utah and obtained from Dr. Ching-Kang Chen [27], HRGP-Cre from Dr. Yun Z. Le (Oklahoma Health Science Center, Oklahoma City) [28], and Six3-Cre lines from Dr. Monica Vetter (University of Utah) with permission of Dr. Furuta [29]. Mice were maintained under 12-hour cyclic dark/light conditions in standard T-cages and fed Teklad Global Soy Protein-free Extruded Rodent Diet ad libido. Mice were sacrificed by cervical dislocation. All experimental procedures were approved by the University of Utah Institutional Animal Care and Use Committee (IACUC) and were conducted in compliance with the NIH Guide for Care and Use of Laboratory Animals.

RT-PCR
Total RNA was extracted from retinas of two 5 month old ret Rab11a +/and ret Rab11a -/mice using Trizol reagents (Invitrogen), and reverse transcription (RT) was carried out using Superscript III first strand synthesis kit (Invitrogen) following manufacturer's instructions. Full length coding sequences for Rab11a and Rab11b were PCR amplified with 28 cycles. GAPDH was used as internal control (22 cycles). PCR primers used are:Rab11a-F, Rab11a-R, Rab11b-F, Rab11b-R, GAPDH-F and GAPDH-R (Table 1).

Statistics
Data are presented as mean ± sd where n represents the number of mice analyzed. Statistical comparisons were done using one-way ANOVA for all experimental data. Differences were considered to be statistically significant for p< 0.05.

Rab11a Retina Conditional Knockout Mice ( ret Rab11a -/-)
We generated ret Rab11a -/conditional knockouts using gene-trapped ES cell clones. In the Rab11a ES cell line, a gene trap (GT) cassette was inserted into intron 1 and exons 2 and 3 are flanked by loxP sites (Fig 1A). Mating with a transgenic FLP mouse line removed the GT cassette thereby generating the conditional allele. Further crossings with a Cre recombinase line under the control of a universal promoter (EIIA-Cre) [32] excised exons 2 and 3 and generated the KO allele ( Fig 1A). As exons 1 and 4 are in-frame, a truncated Rab11a protein may be expressed but is predicted to be inactive and most likely degraded. We verified the gene targeting in ES cells and animals using PCR (Fig 1B). Germline deletion of Rab11a is embryonically lethal as live pups were never born (in >20 litters), as published [35]. ret Rab11a -/were generated by crossing Rab11a fl/fl with transgenic Six3-Cre mice which express Cre in retina starting at embryonic day 9.5 (E9.5). PCR analysis confirmed the deletion of exons 2 and 3 in retina but not in tail samples as expected (Fig 1C). At the RNA level, Rab11b mRNA expression was not upregulated In the ret Rab11a -/retina ( Fig 1D). Immunolocalization of Rab11a occurs predominantly in the inner segment of WT retina but is absent in the cKO retina (Fig 1E), confirming retina-specific deletion of Rab11a in ret Rab11a -/mice. Antibody directed against Rab11b revealed localization in WT retina similar to Rab11a (Fig 1E).
Previous work in frog photoreceptor cells [4;5;47-50] suggested that Rab8a and Rab11a interact directly with rhodopsin [12;14] and are components of the rhodopsin ciliary targeting complex which includes ASAP1 (ArfGAP with SH3 Domain, Ankyrin Repeat And PH Domain 1), Rab11a, Rab11 FIP3 (Rab11 family interacting protein 3), Rabin8 (GEF of Rab8) and Rab8a. During ciliary targeting, activation of Rab8a is thought to promote docking of rhodopsin-laden vesicles and their fusion to the plasma membrane at the periciliary ridge membrane. This model was supported by the observation of massive rhodopsin vesicle accumulation at CC base and faster photoreceptor degeneration in dnRab8a transgenic Xenopus and similar but milder rhodopsin mislocalization and photoreceptor degeneration phenotype in Rab11a shRNA or dnRab11a transgenic Xenopus [11;12]. Our finding that Rab11a and Rab8a, and likely their paralogs as well, are dispensable for rhodopsin OS targeting, photoreceptor ciliogenesis, and disk morphogenesis in mouse photoreceptors therefore is unexpected, but based on solid experimentation and mouse genetics.
There are several possible explanations for these interspecies discrepancies. First, the morphology of frog and mouse photoreceptors may contribute to the development of distinct rhodopsin ciliary targeting pathways. The frog rod outer segment has a much larger OS volume than mouse (~40x) and roughly 10 times more rhodopsin must be synthesized and delivered in frog rods than in mouse (700 vs. 80 molecules rhodopsin trafficking through the connecting cilium/second) [6]. It is possible that Rab11a and Rab8a are required for efficient rhodopsin transport in frog, while mouse photoreceptor cells may be able to tolerate the absence of Rab8a and Rab11a. Supporting this hypothesis, endogenous Rab8a, Rab11a or transgenic Rab8a-GFP, Rab11a-GFP are concentrated at the ciliary base as bright spots in frog photoreceptors [11;12;49], while in mouse Rab8a and Rab11a are dispersed throughout IS (Figs 1D and 7C). The absence of phenotype in Rab8a and Rab11a knockouts suggests that there may be other Rab proteins that are able to mediate rhodopsin vesicle docking and fusion (Rab redundancy). Redundancy does not originate from Rab11b and Rab8b/10, as dnRab11a in Rab11a KO and dnRab8b/10 in Rab8a KO do not cause rhodopsin mislocalization. Expression of dnRab proteins should largely, if not completely, disrupt the activity of corresponding endogenous Rab GTPase, as these constructs are effective in in vitro experiments [36-40;45] and they are expressed efficiently together with reporter genes in our high titer AAV injection (Figs 3B, 7D and 9). In vitro primary ciliogenesis assays reveal that, the phenotypes of Rab11a or Rab8a knockdowns by RNA interference only reduce the percentage of cells with intact cilia or reduce ciliary lengths [21;22;24;25]. This observation argues either that knockdowns were incomplete or that unidentified redundant Rabs partially compensate the loss of Rab11a or Rab8a in ciliary membrane trafficking. Currently, the Rab protein family comprises over 60 members (7), few of which have been investigated thoroughly using molecular genetics. To date, only mutations in Rab28 have been identified to cause human retina disease (recessive cone-rod dystrophy 18) [51;52], and defective geranylgeranylation of Rab27 was shown to be linked to choroideremia [53]. Preliminary results with a Rab28 germline knockout show that rhodopsin and OS proteins traffic normally (GY and WB, unpublished).
Alternatively, Rab8a /11a GTPases and their paralogs, may not be required at all in mouse retina for rhodopsin ciliary targeting and photoreceptor ciliogenesis. Because rhodopsin itself is a receptor for cytoplasmic dynein 1 (through binding to dynein light chain Tctex1) [54], rhodopsin vesicles have the potential of being transported from the trans-Golgi network (TGN) to the basal body by dynein motors independent of any Rab guidance since the photoreceptor microtubule track is built with the minus end anchored at the basal body, also called the microtubule organizing center [55]. Rhodopsin vesicles dissociated from the dynein complex must recruit additional proteins for plasma membrane docking and fusion to the periciliary ridge membrane. Mouse Rab8a and likely Rab8b are dispensable for docking, because Rab8a KO and expression of dnRab8b in Rab8 KO, did not cause rhodopsin accumulation in the IS, an observation consistent with ultrastructural data from Rab8a/8b double KO mouse [46]. Further, the only detectable phenotype in Rab8a/8b germline double KO mice is the disruption of apical membrane protein transport specifically in intestinal epithelial cells. Ciliogenesis in all tested organs (including retina photoreceptors) was normal [29;46], indicating that in contrast to numerous in vitro data, Rab8 is not essential for general ciliary trafficking in mouse. It will be important to generate conditional Rab11a deletion in other mouse ciliated tissues (olfactory bulb, kidney) to examine whether Rab11a is generally required for ciliogenesis and ciliary trafficking in cells other than photoreceptors.
Evidence for a direct role of Rab8a or Rab11a in rhodopsin transport in frog was solely based on overexpression of dominant-negative proteins or shRNA knockdown, but not by gene knockout [11;12]. As different Rabs could share the same Rab GDP-GTP exchange factor [56], GDP-locked dominant-negative Rab11a and Rab8a may target other Rabs through sequestering the shared GEF(s). It will be necessary to verify the rhodopsin mistrafficking phenotype in Rab8a and Rab11a loss of function Xenopus lines by deletion with CRISPR/CAS9. Similarly, it will be important to generate Rab11a/11b double KO mice to clarify unambiguously that both Rab11a and 11b are nonessential for rhodopsin OS targeting.
Our data demonstrate that Rab11a and Rab8a are dispensable for rhodopsin transport to the mouse OS. However, a role for Rab8a and Rab11a cannot be entirely excluded, as other Rab proteins may substitute (Rab redundancy), Other membrane trafficking events may depend on Rab8a and Rab11a in mouse photoreceptors but no obvious phenotype could be observed. Rab11a has been implicated in a variety of cellular trafficking pathways with well-established roles in recycling endosome-associated membrane trafficking [9;57]. Rab8a is involved in a broad membrane trafficking pathways including endocytosis-based membrane recycling and exocytosis at the cell edge [10]. Given the protein localization pattern of Rab11 and Rab8 in IS and OPL (Figs 1 and 7), we speculate that Rab11 and Rab8 may function in endosome-related protein sorting/recycling events in the the inner segment and synaptic IS and OPL regions of mouse photoreceptors.