In this study we sought to determine which domains in Crb2a contribute to its localization and function in rods. Zebrafish rods express two crb2 paralogs, crb2a and crb2b, whereas crb1 expression was not detected in the retina [6, 8]. Mouse rods also express two crumbs orthologs, crb1 and crb2[8, 25, 26]. The observation that in zebrafish cells in the retina express two crb2 genes but no crb1 suggests that one or both of the crb2 genes has adopted the function of crb1. While mutations in human CRB1 are associated with several severe and early onset retinal degeneration diseases [9, 10] (and reviewed in ), as yet no human disease has been associated with CRB2 mutations. Loss-of-crb1 function in mice also causes retinal defects but they appear less severe than those observed in humans; the OLM is disrupted in the rd8 mouse and inner and outer segments are smaller than normal [28, 29]. We observed no defects in the OLM in any of the crb2a transgenic lines we created; anti-ZO-1 labeling was normal (data not shown) and adherens junctions in transgenic rods were visible as a distinct gap in GFP in the inner segment. As yet, loss-of-crb2 function in mice has not been reported, so its role in mammalian photoreceptors remains unknown.
Loss-of-crb2b function resulted in reduced photoreceptor apical size and we previously reported that loss-of-moe function, a putative negative regulator or Crumbs protein function, resulted in larger than normal outer segments [6, 8, 30]. In this study, overexpression of Crb2aIntraWT, Crb2aIntraΔFBD, Crb2aIntraΔPBD, Crb2aExtra_TM, or Crb2aExtra_Secr resulted in a significant increase in outer segment size and without interfering with normal development of rods. This result supports our hypothesis that Crb2a may be involved in the renewal mechanism in photoreceptors. Overexpression of Crb2aIntraWT, Crb2aIntraΔFBD, Crb2aIntraΔPBD proteins may increase outer segment size by competing for negative regulators of endogenous Crumbs proteins. For example, the FBD in Crb2aIntraΔPBD may compete with endogenous Crb2a for binding to Moe, a suggested negative regulator of Crumbs protein function [6, 8, 30], and, thus, could lead to potentially more activity of endogenous Crb2a and a larger outer segment. It is more difficult to envision a mechanism by which overexpression of Crb2aExtra_TM and Crb2aExtra_Secr increases outer segment size without knowing what molecules interact with the extracellular domain of Crumbs proteins. Transgene expression likely increases outer segment size by increasing outer segment growth rather than decreasing outer segment shedding because at 6 d shedding has yet to begin; we see no RPE phagosomes at 6 d by immunocytochemistry or TEM (AMJ, unpublished observation).
Crb family proteins and several components of the Crumbs complex have a restricted localization just apical to adherens junctions in epithelia or the outer limiting membrane (OLM) in photoreceptors [6, 8, 28, 29, 31–33]. Crb family proteins are found in photoreceptor inner segments immediately apical to the OLM and in Müller glial microvillar processes that project into the inner segment region [8, 32]. In zebrafish rods there is a morphologically distinct region just apical to the OLM that can be recognized as a bulge in the proximal inner segment (see Fig. 1D, arrow). In mouse, Crb2 localized by immunoEM to this region . Our current findings suggest that domains in both the extracellular region and intracellular regions of Crb2a contribute to its proper localization in rods. Only Crb2aFL localization in the myoid region approximated the normal localization of endogenous Crb2a/b proteins, although its localization was expanded and this expansion caused the bulge to expand. The functional significance of this region in photoreceptors is unknown.
Crb2aIntraΔPBD was retained in the inner segment and cell body plasma membrane and very little was found in the outer segment in contrast to Crb2aIntraΔFBD, which localized mostly to the outer segment. These results suggest that the FBD is responsible for retaining Crb2a in the inner segment. Crb2aIntraWT is also found in the outer segment, however, and it retains the FBD (and the PBD). Why? One possible explanation is that proteins brought into the Crb complex by the PBD disrupt or alter the interaction between the FBD and its binding partner and thus Crb2aIntraWT behaves more like Crb2aIntraΔFBD. The PBD could bring PRKCi into the Crb complex, PRKCi could phosphorylate the FBD and thus lower its affinity for Moe, which localizes cortically in the inner segment and cell body , and, thus, Crb2aIntraWT localizes similarly to Crb2aIntraΔFBD and Crb2aIntraΔFBDΔPBD. Drosophila Crumbs has been shown to be a substrate for phosphorylation by aPKC (orthologue of PRKCi) and Crb activity during embryogenesis is regulated by phosphorylation . Finally, the observation that several transgene products localize to the outer segment in the absence of any outer segment 'targeting signal' suggests that the outer segment could be the default localization for proteins that lack cytoskeletal (or extracellular) anchorage. Observations by Baker and colleagues lead them to also suggest that the rod outer segment seems to be the default localization for single-pass transmembrane proteins .
Crb2aExtra_TM also localizes to the outer segment and this finding is different from that observed in fly photoreceptors where an equivalent construct localized to the stalk membrane . We also found that Crb2aExtra_TM expression in the outer segment is very different from Crb2aIntraWT and Crb2aIntraΔFBD; Crb2aExtra_TM protein forms fine stripes that are concentrated on one side of the outer segment, perhaps near the axoneme. We have no explanation of why Crb2aExtra_TM is more concentrated on one side of the outer segment other then suggesting that since Crb2aExtra_TM is a much larger protein than Crb2aIntraWT, Crb2aIntraΔFBD and Crb2aIntraΔFBDΔFBD and, consequently, it would be less likely to diffuse freely in the disk membrane.
The localization of Crb2aExtra_Secr is intriguing. It is found in the cell body and in and around the inner segment of Crb2aExtra_Secr rods as well as around the inner segments of neighboring cones. It is possible that the secreted Crb2a extracellular domain is trapped by an unknown receptor located on cone inner segments or Müller processes in the region. Functionally, overexpression of Crb2aExtra_Secr in rods led to a small increase in the size of the outer segment and a modest increase in the area of the myoid region, which is opposite to the effect observed in Drosophila where overexpression of this construct shortened the stalk . One possible explanation could be that most Crb2aExtra_Secr protein is sequestered away from the site where it could interfere with normal Crb2a signaling at the base of the myoid in rods; most Crb2aExtra_Secr protein localizes to the region near rod ellipsoids and cone inner segments.
While we cannot exclude entirely the possibility that differences in protein folding between the different constructs contribute to differences in localization, we think it is unlikely for the following reasons. First, given that the extracellular domain is identical in the three constructs that retain that domain (Crb2aFL, Crb2aExtra_TM, and Crb2aExtra_Secr) it seems unlikely that these proteins would fold differently in the ER. It is also unlikely that the presence of the intracellular domain would alter folding kinetics in the ER but we cannot exclude the possibility that it could affect time spent in the ER. Two, it seems unlikely that the small intracellular domain (37 amino acids at the longest) is subject to folding, given what is known about other FERM-binding domains and PDZ-binding domains. Third, the zs-4 antibody, which only recognizes the native (i.e. folded) extracellular domain of Crb2a, recognizes all constructs that retain the extracellular domain, suggesting that the extracellular domain is folded properly.
Overexpression of Crb2aFL had the greatest effect on rod morphology and dramatically increased the width and area of the inner segment myoid region. Similarly, overexpression of full-length Crb had the greatest effect on stalk length in Drosophila photoreceptors . It has been suggested that the inner segment of vertebrate photoreceptors may be a homologous structure to the stalk region in ommatidial photoreceptors in insects, as both lie in between the sensory compartment (outer segment in vertebrates and rhabdomere in insects) and the cell body . The mechanism of inner segment enlargement in Crb2aFL-expressing rods remains unclear. It seems unlikely that it is due to enlargement of the endoplasmic reticulum as two other constructs (Crb2aExtra_TM and Crb2aExtra_Secr) that are identical in the extracellular domain and expressed at similar levels do not produce such an enlargement (Additional file 6). It also seems that it is unlikely to be due to enlargement of the Golgi apparatus for similar reasons and we did not observe an enlarged Golgi apparatus in Crb2aFLexpressing rods (Additional file 3). We also observed that overexpression of Crb2aFL resulted in the appearance of fine processes emerging from the inner segment myoid. Interestingly, we also observed ectopic processes in the inner segment myoids of rods that lack Moe function, the FERM protein shown to bind the FBD of Crumbs proteins, and which was suggested to act as a negative regulator of Crumbs proteins [8, 36]. The molecular origin of these processes remains unclear.