Drosophila R8 photoreceptor cell subtype specification requires hibris

Cell differentiation and cell fate determination in sensory systems are essential for stimulus discrimination and coding of environmental stimuli. Color vision is based on the differential color sensitivity of retinal photoreceptors, however the developmental programs that control photoreceptor cell differentiation and specify color sensitivity are poorly understood. In Drosophila melanogaster, there is evidence that the color sensitivity of different photoreceptors in the compound eye is regulated by inductive signals between cells, but the exact nature of these signals and how they are propagated remains unknown. We conducted a genetic screen to identify additional regulators of this process and identified a novel mutation in the hibris gene, which encodes an irre cell recognition module protein (IRM). These immunoglobulin super family cell adhesion molecules include human KIRREL and nephrin (NPHS1). hibris is expressed dynamically in the developing Drosophila melanogaster eye and loss-of-function mutations give rise to a diverse range of mutant phenotypes including disruption of the specification of R8 photoreceptor cell diversity. We demonstrate that hibris is required within the retina, and that hibris over-expression is sufficient to disrupt normal photoreceptor cell patterning. These findings suggest an additional layer of complexity in the signaling process that produces paired expression of opsin genes in adjacent R7 and R8 photoreceptor cells.

These two main ommatidial subtypes were initially identified based on pale or yellow fluorescence when illuminated with blue light [15,16], with pale (R7p/R8p) expressing Rh3/Rh5, while yellow (R7y/R8y) cell pairs express Rh4/Rh6 (Fig 1) [10,11,17]. This paired expression of opsin genes in adjacent R7 and R8 cells within an individual ommatidium is thought to result from a series of developmental steps. First, a subset of R7 cells stochastically and cell autonomously express spineless (ss, FBgn0003513) which represses Rh3 and induces Rh4 expression [18][19][20][21]. In R7p cells that stochastically fail to express ss and do express Rh3, a signal is initiated that induces the expression of Rh5 in adjacent R8p cells. Extensive studies have identified the genes warts (wts, FBgn0011739), melted (melt, FBgn0023001), members of the hippo (hpo, FBgn0261456) pathway, along with the TGFβ superfamily receptors baboon (babo, FBgn0011300) and thick vein (tkv, FBgn0003726), their respective ligands and numerous transcription factors as components of the induced versus default signal that establishes R7 and R8 photoreceptor cell subtype patterning [8,12,[22][23][24][25][26]. This signal from R7p drives the expression of Rh5 in R8p, and in the absence of a signal from R7y, the default R8y fate and expression of Rh6 occurs. In addition, we have found that the Epidermal growth factor receptor (Egfr, FBgn0003731) and rhomboid (rho, FBgn0004635) are also required for this process [27,28]. Here we undertook a genetic screen to identify additional genes required for this process and show that hibris (hbs, FBgn0029082), an irre Cell Recognition Molecule (IRM) [29], NPHS1 (nephrin, Homo sapiens, HGNC:9801) related member of the Immunoglobulin Super Family (IgSF), is required for the establishment of paired opsin expression in adjacent R7 and R8 photoreceptor cells. We found that hbs is required within the retina for this process, suggesting that it interacts with the network of genes that regulate R7 and R8 photoreceptor cell differentiation.

Isolation and characterization of the a69 mutant
To identify genes required for the induction of Rh5 expression in R8 photoreceptors, we screened approximately 150 homozygous viable eye-expressing enhancer trap lines carrying insertions of the P{etau-lacZ} transposon (FBtp0001352) [30]. This was based on the rationale that genes required for the induction of Rh5 expression would be expressed in the eye, the P {etau-lacZ} transposon has been especially useful in studies of the nervous system, and insertion of this element into loci of interest would provide a convenient means to identify the affected genes [30]. The percentage of Rh5-expressing R8 cells was determined by labeling dissociated ommatidia with antibodies against Rh5 and Rh6. Several mutants with abnormal percentages of Rh5-expressing R8 cells were noted and a69 (FBgn0026612), with the lowest percentage of Rh5 (9%) was further characterized. Immunostaining of both dissociated ommatidia and tissue sections showed that in the a69 enhancer-trap line, Rh5-expressing R8 cells are reduced and most R8 cells have assumed the default fate and express Rh6 (Fig 2A, 2B, 2D and 2E, Table 1). Since mutants lacking R7 cells or having a reduced number of Rh3 expressing R7 cells would also show diminished Rh5 expression, we next examined the R7 cells and found that the percentage of those expressing Rh3 was similar to white 1118 (w 1118 , RRID:BDSC_3605) control flies (42% vs. 47%, Table 1). However, there was a dramatic increase in mispairing between Rh3 expressing R7 cells adjacent to Rh6 expressing R8 cells (Fig 2C and 2F, Table 1) compared to both w 1118 and cinnabar 1 brown 1 controls (cn 1 bw 1 , RRID:BDSC_264) consistent with the idea that the a69 enhancer trap line carries a mutation in a gene required for the induction of Rh5 expression in R8 cells.
To isolate the gene responsible for the a69 phenotype, the location of the P-element insertion in a69 was determined and found to map to the right arm of the second chromosome at position 60E. To determine whether the P-element in a69 is the cause of the phenotype, P-element excision lines were generated and analyzed. Thirty-five homozygous strains of these excision chromosomes were analyzed by staining dissociated ommatidia with antibodies against Rh5 and Rh6, and all of them (100%) were found to have a low Rh5 percentage, similar to that of a69. Only 1% of excision strains would be expected to retain the mutant phenotype as a result of imprecise excision, thus our inability to revert the mutant phenotype is consistent with the a69 P-element not being responsible for the mutation [31]. Furthermore, mapping via recombination analysis revealed that the a69 mutation is localized to the interval between the purple (pr, FBgn0003141) and curved (c, FBgn0000245) genes in the middle of the second chromosome (Fig 3, S1 Table), far away from the P-element insertion site in a69. From this we conclude that the a69 mutation is not associated with the insertion of the P-element. Thirty-three deficiency lines located in the region between pr and c were tested for a69 complementation (S2 Table). These analyses narrowed the location of the a69 mutation to 51C3-51D1 on the right arm of the second chromosome (Fig 4). The lower portion of Fig 4 shows a diagram of this genomic region, spanning~300 Kb and encompassing 25 known protein coding genes. To identify the gene specifically affected in the a69 mutation, we took two approaches. First, a subset of genes were examined for alterations in expression in the a69 mutant, and second, a large series of complementation studies were performed with alleles of known mutants in the region. cDNAs from 5 genes in the region were obtained and in-situ hybridization of third instar larval eye imaginal discs was performed on cn 1 bw 1 (wild-type) and a69 mutants. Three multiply marked chromosomes (al 1 dpy ov1 b 1 pr 1 c 1 px 1 sp 1 , al 1 dpy ov1 b 1 pr 1 , and b 1 pr 1 c 1 px 1 sp 1 ) were recombined with the w 1118 ; P{etau-lacZ}a69 mutant. After marker identification, recombinant strains were back crossed to the a69 mutant and scored for the percentage of Rh5 expression. The regions of the recombinant chromosomes assumed to be derived from the a69 parental mutant strain are indicated in gray, while the regions assumed to be derived from the multiple marked (wild-type) chromosomes are black. Sixteen recombinant strains were phenotypically wild-type and complemented a69. Four recombinant strains were intermediate and eight strains were mutant and failed to complement a69. The four intermediate strains and one wild type strain, al10, differed from the expected phenotypes and may have resulted from multiple recombination events or exposure of cryptic modifier loci. See S1 Table. Complementation of a69 Recombinant Strains.
https://doi.org/10.1371/journal.pone.0240451.g003 In each case the expression pattern of the gene was not substantially disrupted in a69 mutants, suggesting that the phenotype is not due to the disruption of patterned mRNA expression of these genes in the 3rd instar eye-antennal disc (Fig 5). hibris (hbs) was expressed strongly in the morphogenetic furrow and maintained weakly posteriorly, consistent with a previous report [32]. It was also expressed in the ocellar region and in the developing antenna. parcas (pcs, FBgn0033988) was expressed strongly in the morphogenetic furrow and in the antenna. CG10265 (FBgn0033990) did not appear to be expressed in either the eye or antennal regions. CG7639 (FBgn0033989) appeared to be weakly expressed in the region anterior to the morphogenetic furrow. caskin (ckn, FBgn0033987) was expressed anterior to the furrow and in the antenna.
We obtained the following alleles of hbs: hbs 66 (FBal0239852), hbs 361 (FBal0130217), hbs 459 , (FBal0130216), hbs 1130 (obtained from M. Baylies) and hbs 2593 (FBal0130218). With one exception, all of these alleles fail to complement a69 ( Table 2). Furthermore, hbs 361 homozygotes and heteroallelic combinations of all alleles show a substantial decrease in the proportion of Rh5 expressing R8 photoreceptor cells. With a few exceptions, viable combinations of these alleles over deficiencies in the region show the same complementation pattern as the a69 mutant (S3 Table).
Exon sequencing of the hbs gene failed to identify unique polymorphisms in the a69 mutant that were absent in phenotypically wild type control strains. Nonetheless, given that the gene spans over 30 Kb including 24 Kb in the first intron, it seems likely that a mutation within a regulatory region of the hbs gene may be responsible for the hypomorphic a69 phenotype. Based on the failure of complementation of a69 by all but one allele of hbs, and the finding that all eleven heteroallelic combinations of four known alleles of hbs also display the a69 phenotype ( Table 2), we believe the data is consistent with a69 being a hbs allele, hbs a69 . The two discrepancies to this conclusion, 1) complementation between a69 and hbs 459 , and 2) differences in complementation patterns of deficiencies (6 crosses out of 47, S3 Table) are consistent with intragenic (interallelic) complementation. This pattern of complex complementation, which has been described for numerous genes in Drosophila, is particularly common with alleles of intermediate phenotypic effects (e.g. hypomorphic alleles like a69), and may arise from differences in genetic background or complementation between alleles having defects in different functional regions of the gene [33][34][35][36]. One or more of these mechanisms are likely to underlie the complex complementation pattern observed in our experiments. Despite these discrepancies, the results with existing, molecularly characterized alleles of hbs clearly demonstrate that the hbs gene is required for the differentiation of R7 and R8 photoreceptor cells and the regulation of Rh5 and Rh6 opsin expression.
Because hbs a69 is a hypomorph, subsequent genetic experiments were performed with the hbs 66 allele (FBal0239852), which has a stronger phenotype, is characterized at the molecular level and is available on the P{ry +t7.2 = neoFRT}42D chromosome [37].

hibris is expressed in the developing third instar eye imaginal disc
Consistent with in situ hybridization analyses (Fig 5) and previous studies [38], we find that the hbs protein is expressed in the developing third instar eye imaginal disc in the morphogenetic furrow. The protein is found in preclusters of photoreceptor cells emerging from the morphogenetic furrow and ultimately in all photoreceptor cells (Fig 6A and 6B). hbs is expressed immediately anterior to, within, and posterior to the region of early senseless (sens, FBgn0002573) expression, which is expressed in R8 cells just posterior to the morphogenetic furrow (Fig 6A-6C).

hibris is required in the retina for R7 and R8 cell differentiation
To assess the function of hbs in Rh5 and Rh6 expression in R7 and R8 photoreceptor cell patterning, we examined hbs 66 mosaic flies. We used the ey-FLP driver to generate homozygous mutant clones in the retina and optic lobes of animals that were heterozygous for hbs 66 . We used a cell autonomous lethal mutation to generate large homozygous mutant clones and eliminate homozygous wildtype tissue, as described [39]. Fig 7A shows that loss of hbs in the retina and optic lobe leads to a dramatic decrease in Rh5 expression and mispairing of Rh3 and Rh6 in adjacent R7 and R8 cells of individual ommatidia. This is in contrast to Rh3, Rh5 and Rh6 expression in a similarly FRT recombined clone of a wild type chromosome (Fig 7B,  S1 Fig).
To further refine the spatial requirement for hbs in R7 and R8 photoreceptor cell differentiation and opsin gene expression we also generated mutant clones of hbs 66 with ey3.5-FLP [40]. ey3.5-FLP is a modified form of ey-FLP that efficiently induces clone formation in the third instar larval eye imaginal disc, but not in the lamina or medulla [40]. Retina specific clones generated with ey3.5-FLP also show a loss of Rh5 expression along with increased mispairing of Rh3 and Rh6 (Fig 7C), as compared to an FRT recombined clone of a wild type chromosome (Fig 7D, S1 Fig). These results indicate that hbs is required in the retina for normal R7 and R8 photoreceptor cell differentiation and opsin gene expression.  Overexpression of hibris is sufficient to disrupt R7 and R8 cell differentiation To determine whether ectopic expression of hbs is sufficient to induce the expression of Rh5 in R8 photoreceptor cells, we over-expressed hbs using the GAL4-UAS system [41] and the P {GAL4-ninaE.GMR} driver (FBtp0001315), which drives transcription in the developing eye in all cell types posterior to the morphogenetic furrow [42,43]. Fig 8A shows that overexpression of hbs leads to a large increase in Rh5% expression, demonstrating that hbs is sufficient to induce Rh5 expression in many, but not all R8 photoreceptor cells. This occurs with a modest 10% decrease in Rh3-expressing R7 cells ( Table 1). To test whether this effect results from hbs acting on Rh4 expressing R7 cells to inappropriately induce Rh5 expression or from hbs acting directly on R8 cells, we overexpressed hbs in a sevenless (sev) mutant background that lacks R7 photoreceptor cells. Fig 8B shows that removal of R7 cells leads to a~20% reduction in the number of Rh5 expressing R8 cells, but still significantly more Rh5 expression than is seen in sev mutants alone ( Table 1). These results suggest that the ability of overexpressed hbs to induce Rh5 expression in R8 cells is at least partially independent of the R7 photoreceptor cells and that hbs may act directly on, or in R8 cells.

Stocks and genetics
Stocks were maintained in humidified incubators on cornmeal/molasses/agar media or standard cornmeal food with malt, and transferred on a rotating basis every three weeks as described [44][45][46]. D. melanogaster strains were obtained from individual laboratories or the Bloomington Drosophila Stock Center (BDSC). Genotypes were constructed using conventional genetic techniques, dominant markers and appropriate balancer chromosomes [45,47].  [48]. Heterozygous strains carrying new transposon insertions (~1900) were screened for β-galactosidase expression in whole dissected adult heads [49]. Lines showing expression in the eye (retina and/or optic lobe) (323) were crossed to generate homozygous viable stocks. The collection was not retained long term and has since been discarded.  Table 1.
Third instar eye-antennal discs were dissected in 1X PBS with 0.1% triton X-100 (PBT), fixed for 20 min at room temperature in 4% paraformaldehyde in 1X PBS, and washed three times for 5 min at room temperature with PBT. Discs were blocked with PBT supplemented with 1mg/mL BSA and 5% NGS for 1hr at room temperature before being incubated with primary antibodies diluted (as below) in PBT. After three 10 min washes, discs were incubated with secondary antibodies diluted 1:200 in PBT (see below) and once again washed three times for 10 min before being mounted in PermaFluor (ThermoFisher). Images were obtained on a Nikon A1R Confocal microscope and are constructed from a series of z-stacks as a maximum intensity projection. Primary antibodies used: guinea pig polyclonal antisenseless (1:1000, [51]), rabbit polyclonal anti-hibris (1:400, AS-14, RRID:AB_2568633, [52]). Secondary antibodies used: goat anti-guinea pig conjugated to Alexa Fluor 568 and goat anti-rabbit conjugated to Alexa Fluor 488 (ThermoFisher A11075 and A11008, respectively).

Statistical analysis
Comparisons of the proportions (percentages) of opsin expression in different genetic backgrounds were performed with a z-score and are shown in Tables 1 and 2 legends for Fig 7, S1-S3 Tables and S1 Fig [53]. The z-score was calculated using the equation: z ¼ r 2 À r 1 ½ � À 1 2 1=n 1 þ 1=n 2 ð Þ ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffiffi r avg q avg 1=n 1 þ 1=n 2 ð Þ q p 1 and p 2 = proportions of marker expression in each of the two different genotypes under comparison. n 1 and n 2 = number of ommatidia counted for each genotype. p avg = average proportion for both genotypes combined. q avg = 1-p avg . The significance of the difference between the two proportions was determined from the normal distribution as a one-or twotailed test. The 95% confidence interval of a proportion was calculated using the Wilson procedure without continuity correction [54,55] using VasarStats [56].

Discussion
Here we describe the isolation and characterization of a novel allele of the D. melanogaster gene hibris, an evolutionarily conserved NPHS1 (nephrin) related IgSF member [58]. We show that hibris is required for the coordinated expression of opsin genes in adjacent R7 and R8 photoreceptor cells within the compound eye. Orthologues of this gene have been identified in many species, and numerous paralogues within species play diverse roles in organ system development and function [59]. Within the context of R7 and R8 photoreceptor cell differentiation and the regulation of opsin gene expression in the retinal mosaic, hbs is both required and partially sufficient for the expression of Rh5 in R8 photoreceptor cells.
hbs is required in the retina for the induction of Rh5 expression based upon our experiments making homozygous mutant clones with ey3.5-FLP (Fig 7). Subsequent studies will be required to place hbs within the large network of genes involved in R7 and R8 photoreceptor cell differentiation and the inductive signal that is thought to coordinate opsin gene expression in adjacent R7 and R8 photoreceptor cells.
Traditionally, inductive processes are thought to occur between tissues or cells in which there is an inducer and a responder. Inductive signals are also often defined as instructive or permissive [63]. In the presence of an instructive interaction (i.e. from a R7p cell), the responder (R8) develops in a certain way (as a R8p cell expressing Rh5). By contrast, in the absence of the instructive interaction (R7y or R7 cells absent, e.g. sev mutants), the responder (R8) does not develop in a certain way, i.e. does not become R8p expressing Rh5, but rather becomes R8y and expresses Rh6 instead as a default fate (with some exceptions [11]). If hbs played a formal instructive role in regulating the expression of Rh5 in R8 photoreceptor cells, then we would expect that its expression throughout the retina (GMR-Gal4; UAS-hbs) would lead to expression of Rh5 in all R8 photoreceptor cells even in the absence of R7 cells (Fig 8B). Although all Rh5 is not expressed in all R8 cells in this experiment, it is far higher than in sev mutants alone [10,11,27,28]. This demonstrates that hbs driven expression of Rh5 in R8 photoreceptor cells is partially R7 cell independent and suggests that hbs may act on, or in R8 cells and play an instructive role in this process.
Alternatively, as a potentially permissive regulator of R8 photoreceptor cell differentiation, hbs may play a role in establishing the architecture of the developing eye. Perhaps loss of hbs disrupts cellular contacts that mediate signaling between R7 and R8. There is ample evidence for disruption of cone and pigment cell differentiation and eye roughening in hbs mutants [64,65]. Furthermore, hbs and its binding partner roughest (rst) are known to have effects on axon guidance and synapse formation in the optic lobes [66][67][68][69]. Perhaps these interactions between retinal cells other than the R7 and R8 photoreceptor cells are responsible for some aspect of inductive signaling and expression of Rh5 in R8p. Finally, perhaps the loss of Rh5 expression in the hbs mutant eye reflects an inability to respond to the inductive signal, a loss of competence [70]. We previously suggested that rhomboid (rho, FBgn000463) and the Epidermal growth factor receptor (Egfr, FBgn0003731) may play a role in establishing competence of the R8 cell [27]. Thus, hbs could potentially play a permissive role in R7 and R8 differentiation.
Subsequent analysis of the role of hbs in R7 and R8 photoreceptor cell differentiation will require further identification of its specific interaction partners in this system in the retina, as well as the temporal requirement for its involvement in R7 and R8 cell differentiation. Ample resources are available including mutant strains [71], RNAi transgenics [72], and temporal and spatial mis-expression tools [73][74][75][76][77]. Despite these technical resources, defining the precise role of hbs in R7 and R8 differentiation will likely yield a complex system, reflecting coregulation of the IRM proteins [78], involvement of large complexes associated with scaffolding proteins [79], functional or genetic redundancy, compensation [80] and feedback.