We first investigated the expression of Tribolium UV-opsin. To generate a Tribolium UV-opsin RNA probe for whole mount in situ hybridization, we cloned a 502 bp fragment of the predicted Tribolium UV-opsin gene. In Drosophila, the UV-opsin paralogs Rh3 and Rh4 are stochastically distributed in non-overlapping patterns in all central R7 photoreceptor cells (Fig. 1, Fig 2a and 2b) 1415. Similarly, in situ hybridization experiments with an RNA antisense probe against Tribolium UV-opsin in mature pupal and early adult retina detected strong expression in a single photoreceptor cell per ommatidia throughout the entire retinal field (Fig. 2d). Differential interference contrast microscopy at high magnification confirmed that the Tribolium UV-opsin paralog was selectively expressed in all R7-type cells (Fig. 2e). The comparison with Drosophila thus revealed that Tribolium UV-opsin is expressed in a conserved R7-specific manner.

Next we investigated the expression of Tribolium LW-opsin. In Drosophila, the peripheral photoreceptors R1-6 express the LW-opsin paralog Rh1 (Fig. 2e and 2f), and 70% of central R8 cells express the LW-opsin Rh6 16. The remaining 30% of R8 cells express the Drosophila B-opsin paralog Rh5 1718. For whole mount in situ hybridization in Tribolium, we cloned a 972 bp long RT-PCR fragment that includes the entire predicted Tribolium LW-opsin ORF from head specific Tribolium total RNA. Analysis of the expression of the single Tribolium LW-opsin gene revealed uniform transcription in all photoreceptors of the retina (Fig. 2g). Importantly, expression was not only detected in the peripheral photoreceptors, but also in all central R7 and R8 photoreceptors (Fig. 2h). This finding reveals a strongly expanded expression domain of LW-opsin compared to Drosophila, and indicates that R7 cells in Tribolium co-express the LW and UV-opsin-encoding genes.

Like in most insects, ommatidia in the dorsal rim area (DRA) of the Drosophila compound eye exhibit modifications for the detection of polarized light including expression of UV-opsin in both R7 and R8 1920. However, no evidence of regional heterogeneity of opsin expression could be detected in the dorsal region of the Tribolium retina (Fig. 2c and 2g), consistent with the lack of morphological evidence of DR ommatidia in this species 13.

The expression of the Tribolium LW- and UV-opsins accounts for all photoreceptors in the Tribolium retina, and is compatible with a loss of B-opsin in the flour beetle. This finding further cements evidence for color vision reduction in the evolutionary lineage leading to Tribolium (Fig. 3). Consistent with this, the spectral sensitivity of the very closely related mealworm Tenebrio molitor has been reported to peak only in the UV and green wavelength ranges based on electroretinogram measurements 21. Moreover, preliminary PCR screens from Tenebrio molitor genomic DNA resulted in isolation of closely related orthologs of Tribolium castaneum LW- and UV-opsins, while similar attempts to isolate B-opsin orthologs failed, just as in Tribolium castaneum (Bao and Friedrich, unpublished). Hence genome, gene expression and physiological data converge on supporting the loss of B-opsin in Tribolium castaneum.

Our data reveal that the lack of B-opsin in Tribolium correlates with a dramatic expansion of LW-opsin expression. In most endopterygote retinas studied so far, LW-opsins are expressed in the peripheral photoreceptors R1-6 and a fraction of central R7 or R8 cells (Fig. 3). Thus, the expanded LW-opsin we observe in Tribolium is likely to have played an essential role in the evolution of its retinal mosaic. The comparative data further imply that these changes involved the expansion of the peripheral LW-opsin expression domain into R7 and R8.

A most accurate inference of the underlying mutational events underlying the Tribolium opsin expression pattern depends on knowing the organization of the retinal mosaic in the last B-opsin expressing ancestor of Tribolium. Although this is still much a matter of speculation, some conclusions can be drawn from physiological data from other Coleoptera and opsin expression data in other endopterygote orders. Based on electrophysiological studies, for instance, species in the beetle families Carabidae (ground beetles) and Coccinelidae (lady beetles) are equipped with sensitivities in the green, blue and UV 2223. Considering further that B-, LW- and UV-opsins are expressed in R7 and R8 cells in Hymenoptera, Lepidoptera and Diptera, it is most likely that Tribolium evolved from an ancestor in which R8 and R7 cells differentially expressed combinations of at least three opsins, similar to Drosophila (Fig. 3) 51124.

Based on this premise, it is reasonable to assume that the UV-opsin expression in R7 is a conserved aspect of the Tribolium retina, whereas the loss of B-opsin expression and correlated expansion of LW-opsin expression is derived. This latter change may have occurred through one of several possible trajectories (Fig. 4). First, the loss of B-opsin expression could have been initiated through mutations that abrogate transcriptional activation or expression of a functional protein prior to or independent of changes in the ancestral expression domain of LW-opsin (Fig. 4a). However, this would create a retinal mosaic with partially light-insensitive photoreceptor cells. Considering the importance of flight in Tribolium (see below), the ensuing deterioration of spatial vision would likely be a devastating fitness penalty. This suggests that the molecular evolution of B-opsin loss was correlated with, or contingent upon, compensatory changes in the regulation of LW-opsin, a second possibility (Fig. 4b).

Several mechanisms exist regarding the nature of such cis-regulatory changes in the B- and LW-opsin genes. Assuming heterogeneity of LW-opsin and B-opsin expression among ancestral R8 cells, for instance, correlated mutations in the cis-regulatory regions of both genes may have led to a gradual increase in the fraction of LW-opsin expressing cells that eventually culminated in the complete replacement of B-opsin expressing cells. This scenario describes a model of gradual replacement of blue versus green-sensitive photoreceptors in which color vision evolves as a quantitative trait (Fig. 4b). Alternatively, cis-regulatory change may have resulted in immediate compensatory loss of B-opsin and expansion of LW-opsin expression. This type of change could have been induced by separate mutations in the cis-regulatory regions of both genes or, as data from Drosophila suggest (see below), through mutational change in the cis-regulation of only one opsin pending pre-existing mechanisms of mutual repression. Both trajectories require at least one additional regulatory change that would allow further expansion of LW-opsin expression into R7 (Fig. 4b).

A third scenario is that a single mutation led to the immediate expansion of the LW-opsin expression into all photoreceptor cells. The consequence would be that of co-expression of LW-opsin with B-opsin in a fraction of R8 cells, and with UV-opsin in R7 cells (Fig. 4c). Such a retinal mosaic would thus be one step away from that in extant Tribolium by mutational disintegration of the activating elements in the ancestral B-opsin cis-regulatory region. This pathway seems the most probable for three reasons: (1) requiring only a single mutation to generate the final LW-opsin pattern is most parsimonious, (2) the co-expression of LW- and UV opsins in the R7 cells of Tribolium is consistent with this scenario, and (3) the latter fact as well as data from other species (see below) demonstrates that opsin co-expressing photoreceptors are viable and functional. Moreover, a strikingly similar scenario exists in Drosophila, where a single point mutation in the LW-opsin promoter, Rh6, or in the transcription factor, Orthodenticle (Otd) leads to expansion of Rh6 (and co-expression of opsins) into the majority of peripheral photoreceptors and many R7 central photoreceptors, but not into the B-opsin-containing R8 cells 25.

Molecular genetic analyses in Drosophila have unraveled several critical mechanisms of cis-regulatory control of opsin expression, some of which are likely conserved 512. Those particularly relevant to the discussion of the evolution of the Tribolium retina involve studies on the regulation of the LW-opsin-encoding gene, Rh6. Such studies have demonstrated that at least four independent transcriptional repression processes normally restrict Drosophila Rh6 to 70% central R8 photoreceptors. In R8 cells, the mutually exclusive B- versus LW-opsin expression in 30 vs 70% R8 cells occurs through a bistable loop of both cell signaling and transcriptional regulation 26. Failure to properly initiate this loop leads to a retina that expresses all B-opsin or all LW-opsin genes in R8 cells. Interestingly, LW-opsin expression is the "default" R8 opsin state, whereas B-opsin requires induction. Thus, this regulatory mechanism could reflect that the loss of B-opsin observed in Tribolium is secondary to a loss in this induction process. This regulatory mechanism also suggests an inherent capacity of the photoreceptor mosaic in higher flies to evolve along the gradual replacement model considered for Tribolium above. Second, Drosophila express two LW-opsins: Rh1 in peripheral non-color sensitive photoreceptors, R1-R6, and Rh6 in 70% R8 cells. Co-expression of both LW-opsins is specifically prevented in peripheral photoreceptors due to cell-specific repression of Rh6 25. A third mechanism of repression for Rh6 occurs in all R7 cells, where it is repressed by the R7-specific transcription factor, Prospero (Pros) 27. Finally, Rh6 expression in DRA ommatidia (both R7 and R8 cells) is prevented through the selective inactivation of the R8-specific, Rh6 activating transcription factor Senseless (Sens) in the DRA 28. Thus, the release of analogous LW-opsin repression mechanisms could have led to the expansion of LW-opsin expression as postulated for the evolution of Tribolium.

While it is premature to draw final conclusions on the trajectory of cis-regulatory changes that generated the retinal mosaic of Tribolium castaneum, data and implications identify Tribolium as unique model to study compensatory gene regulatory mechanisms associated with gene loss. Progress in this area can be made by studying the cis-regulatory control of Tribolium LW-opsin in comparison to that of Drosophila Rh6 and Rh1, and by exploring the conservation of opsin expression in a wider range of darkling beetles, as well as in distantly related Coleoptera.

Across orders, the "one receptor rule" is important for sensory receptor expression 329. The only two exceptions to this rule in insect photoreceptors reported thus far occur in butterfly species that have regionalized photoreceptors that co-express two opsins 3031. Thus, Tribolium is the first insect example in which a specific photoreceptor subtype co-expresses two different opsins across the entire retina. Opsin co-expression is more common in vertebrates than invertebrates 32. In the Siberian hamster and the pouched mouse, for instance, dual opsin-expressing cone cells are uniformly spread throughout the retina similar to Tribolium 33. Intriguingly, vertebrate species with exceptionally high frequencies of opsin co-expressing photoreceptors are either nocturnal or fossorial 3234. This evolutionary parallel raises the possibility that opsin co-expression is of advantage under conditions where brightness sensitivity is critical. This bodes well with the low light ecological niche of Tribolium since all postembryonic stages of Tribolium show strong negative phototaxis 3536.

Both vertebrate and insect opsin co-expressing photoreceptors have been found sensitive to a light spectrum that equals the summation of the single sensitivities of the expressed opsins 303738. While the excitation characteristics of the Tribolium R7 photoreceptors await to be analyzed by electrophysiological means, it is reasonable to assume that these cells are sensitive to a broad range of light from the green into UV wavelengths. What then may have caused the evolution of this unusual trait? For this question to be answered, it is necessary to further explore the functional consequences of opsin co-expression and, secondly, the state from which the Tribolium patterns originated.

Modeling data suggest that optimal single opsin-containing photoreceptors exhibit higher photon capture rate than dual opsin-expressing photoreceptors 6. However, physiological measurements and ecological correlations point to a stronger impact of extended wavelength sensitivity for enhancing brightness detection. In house fly, for instance, the LW-opsin-expressing peripheral photoreceptors R1-6 contain a UV sensitizing pigment 39, which leads to an expanded wavelength sensitivity similar to that assumed for the Tribolium R7 cells. Importantly, this supplemental UV-sensitizing pigment has been shown to significantly increase overall brightness sensitivity in the peripheral photoreceptors of the house fly. The effect is considered to be particularly significant under low light conditions 39. Interestingly, Tribolium castaneum does disperse by flight under conditions of stress or exposure to strong light 40. The conservation of flight capability indicates that flight has maintained fitness importance throughout Tribolium evolution. Moreover, adults of the meal worm Tenebrio molitor are negatively phototactic like Tribolium but become positively phototactic after extended times of desiccation 41, suggesting that this closely related species to Tribolium behaves as a facultative cryptozoic species. Thus, the Tribolium visual system may have been molded by long-term cryptozoic adaptation.

From which functional ancestral state did the broadband sensitivity of Tribolium R7 cells evolve? While the function of the Tribolium R7 photoreceptors may have changed to extreme broadband photoreceptors to boost the overall photon capture capacity of the retina, enhanced brightness sensitivity is not incompatible with maintenance of color detection capacity. In theory, excitation differences between LW-opsin expressing R8 and LW/UV-opsin-coexpressing R7 cells may still discriminate UV from green patterns to facilitate dichromatic vision in Tribolium. The former may have evolved without completely sacrificing UV versus LW discrimination ability. Indeed, structural and comparative evidence suggest that the Tribolium central photoreceptor cells R7 and R8 represent ancestral color detectors. The cellular architecture of the Tribolium ommatidia is very similar to that in mealworm Tenebrio molitor. In both species, the six peripheral photoreceptors form a ring of rhabdomeres while the inner R7 and R8 photoreceptors contribute a separate rhabdom stem 1342. Importantly, Tenebrio ommatidia exercise pupil-like brightness adaptation through volume changes of the primary pigment cells which sit on top of the photoreceptor bundle 42. In the light-adapted state, the primary pigment cells are expanded such that only the central R7 and R8 cells are exposed to the light path. In the dark-adapted state, the primary pigment cells further contract to additionally expose the peripheral R1-6 to light. This mechanism suggests specialization or division of brightness detection versus color discrimination tasks between inner and outer photoreceptors, similar to Drosophila. While this interpretation remains to be experimentally tested, the fact that the Tenebrio light adaptation mechanism is conserved in the more distantly related darkling species Zophobas atratus strongly suggests that the Tribolium situation evolved from a central photoreceptor-based color discriminating system 42.

Taken together, there is circumstantial evidence for both costs and benefits to opsin co-expression in Tribolium R7 cells. Thus, it will be necessary to elucidate the relative significance of these aspects for the biology of Tribolium to understand the evolutionary forces that generated the Tribolium retinal mosaic. Importantly, while dispersal by flight under conditions of stress seems to represent the likeliest condition under which brightness sensitivity may be critical for Tribolium, observations in the laboratory suggest little dependence of food and mating partner detection on vision 35. Nonetheless, the cryptozoic habits of Tribolium speak for an increased selection pressure on brightness sensitivity and relaxation of the selection pressure on color discrimination needs.

The most parsimonious trajectory of cis-regulatory evolution that leads to the retinal mosaic of extant Tribolium is mechanistically linked to both the emergence of UV-/LW-opsin co-expressing R7 cells and the loss of B-opsin expression in R8 cells (Fig. 4c). The potential adaptive advantage of UV-/LW-opsin co-expression in Tribolium R7 cells generates the question: could the loss of B-opsin have been influenced not only by relaxation of color vision needs but also enforced by an adaptive effect of increased brightness sensitivity? Lack of blue sensitivity is reliably documented in only a few cases of insects, including the crepuscular cockroach Periplaneta americana, and the owlfly Ascalaphus macarionius, a diurnal predator 4344. The diverse behavioral patterns of these species, however, have provided little information regarding the potential causes leading to reduced color discrimination 2. Remarkably, in vertebrates, opsin dual-expressing photoreceptors as well as single LW-sensitive photoreceptors are increased in species with cryptozoic life histories (for review see 34). Keeping in mind that brightness detection is primarily based on LW-opsins in both vertebrates and insects, the parallel increase of LW-opsin expressing cells in cryptozoic vertebrates and Tribolium is further suggestive of an adaptive trend. The replacement of B-opsin with LW-opsin expressing cells during Tribolium evolution may thus have been adaptive. However, this does not enforce the conclusion that the loss of Tribolium B-opsin involved positive selection. Either gradual or complete expansion of the LW-opsin followed by decay of B-opsin expression may have occurred as neutral or permissively deleterious changes in color vision lost fitness importance, while brightness sensitivity remained under purifying selection in the lineage leading to Tribolium. It is not clear whether the similarity in the opsin repertoires of cryptozoic insect or vertebrate retinas reflects convergent preferential conservation of photoreceptors essential for brightness detection or adaptive enrichment within a short time window of cryptozoic adaptation 34. Extended comparative data may provide further insights. For instance, conservation of specific UV-opsin expression but not LW- and UV-opsin co-expression in R7 cells in related darkling beetles would indicate a persisting need of detecting UV light patterns but point against a strong advantage of brightness sensitivity increase through LW- and UV co-expression. Conserved co-expression of UV with LW-opsin in R7, on the other hand, would be most reasonably correlated with a fitness advantage of maximizing brightness sensitivity.

Pearl pBac(3xP3-EGFP)af T. castaneum were reared on 5% yeast supplemented whole wheat flour at 31°C in constant darkness. y¹w⁶⁷Drosophila melanogaster lines were maintained on standard cornmeal-molasses food at room temperature.

The putative coding regions of Tribolium castaneum LW-opsin and UV-opsin were identified by TBLASTX search with the Drosophila Rh1 and Rh3 gene sequences to the Tribolium castaneum Georgia GA2 genome assembly (release 8/17/2005), accessed through the Human Genome Sequencing Center server in the Department of Genetics at Baylor University 47. Similar searches with the remaining Drosophila opsins, Rh2, Rh4-Rh6, failed to identify additional Tribolium opsin-encoding genes.

A 502 bp fragment of the predicted Tribolium UV-opsin gene was amplified from Tribolium total genomic DNA by standard PCR using the specific primers 5'-CCCCAAAGACGAACTCATTC-3' and 5'-ATGGTGTACCCCCAAACAAA-3'. This primer combination produces a PCR amplicon that extends from the first putative open reading frame (ORF) containing exon of the Tribolium UV-opsin gene to the second ORF-containing exon bridging a 55 bp intron. 972 bp of the predicted Tribolium LW-opsin ORF were amplified by nested RT-PCR using outer primer pair TcRhR-3F (5'-TCACCGTCGTGGACAAAGTC-3') and TcRhR-3R (5'-TTCTCGTCGGTGGTAACGTTC-3'), and inner primer pair TcRhR-4F (5'-GCCCATTGGTACCAGTTCC-3') and TcRhR-4R (5'-AGCCGTGGAGGTGGTGTC-3'). Total RNA was extracted from adult Tribolium head with the RNAqueous kit (Ambion). The RETROscript kit (Ambion) was used to generate cDNA through reverse transcription initiated with random decamers. The resulting PCR fragments were cloned into pGEM^®-T Vector (Promega). Multiple identical sequences were obtained for each paralog. The sequences are accessible under accession numbers [GenBank: DQ029113] and [GenBank: EU164547].

95% pupal Drosophila whole-mounted retinas were dissected at room temperature (RT) in PBT (1 × PBS + 0.1% Triton X100, pH 7.2), fixed in PLP (1 × PBS, 4% paraformaldehyde, 0.075 M lysine, 0.01 M sodium periodate, 0.05% saponin) 48, and washed 3 × 10 min with PBT. Retinas were transferred to Signal-iT FX (Invitrogen) for 30 min, RT, before incubation with mouse anti-Rh1 (1:100, DHSB), mouse anti-Rh3 (1:5, gift from S.G. Britt), and rabbit anti-Rh4 (1:250, gift from C.S. Zuker) overnight at 4°C in BNTS (1 × PBT, 1.5 M NaCl, 0.1% BSA, 0.05% saponin). Samples were washed 3 × 20 min with PBT, incubated for 2 hrs at RT with AlexaFluor-488, -555, or -648-conjugated secondary antibodies (1:500, Molecular Probes), and washed thrice again with PBT. Retinas were mounted in Prolong Gold antifade-reagent (Invitrogen), and imaged 24 hr later using an Axioplan IIie equipped with the Zeiss ApoTome deconvolution module. Digital images were processed with Axiovision 4.5 (Zeiss) and AdobePhotoshop 7.0 software.

Whole mount in situ hybridization of Tribolium pupal retinae was carried out following previously published protocols with minor modifications 4950. Digoxygenin-labeled antisense RNA probes were prepared with DIG-UTP RNA Labeling Mix (Roche). Pupal retinas were dissected in ice cold PBS, and fixed in 9.25% formaldehyde, 50 mM EGTA ph8, and 10% Triton X in PBT at RT for 1 h. Tissues were then washed one time in PBS, three times in PBT, and equilibrated for hybridization with a 10 min incubation in 50% PBT+50% Hyb-A-RNA (hybridization buffer A) at RT, and 1 h at 55°C in 100% Hyb-A-RNA (50% formamide, 4 × SSC, 250 μg/ml tRNA, 0.1% Tween-20, 1 × Denhardt's solution, 5% dextran sulfate) with 500 μg/ml freshly denatured herring sperm DNA. Hybridization was carried out for 72 h in Hyb-A-RNA with RNA probe at approximately 0.5 μg/ml concentration at 55°C. Subsequently, pupal retinas were washed 3 × 30 min in Hyb-B-RNA (50% formamide, 4 × SSC, 0.1% Tween-20) at 55°C followed by a 24 hrs final washing step. Next, tissues were equilibrated in 5 × SSC at RT for 5 min and another 5 min in maleic acid buffer (0.1 M maleic acid, 0.15 M NaCl, 0.1% Tween-20). Tissue was blocked for 2 hrs at RT in maleic acid buffer supplemented with Roche in situ blocking buffer at 1:10 dilution (Roche). Anti-DIG antibody (Roche) was added at a 1:5000 dilution and tissues were incubated 2 hrs at RT on a shaker. Following antibody incubation, pupal eye discs were washed 3 × 1 hrs. Tissues were then transferred into detection buffer (0.1 M Tris-HCl pH 9.5, 0.1 M NaCl, 50 mM MgCl₂) for 5 min at RT. Color detection was initiated by adding 4.5 μl NBT (50 mg/ml) and 3.5 μl BCIP (50 mg/ml) as substrate. Color reactions were developed for 1 to 2 hrs and stopped by replacing staining buffer with PBT. Labeled retinas were washed with EtOH, rehydrated, and cleared in 70% glycerol in PBS before mounting. DIC brightfield images were recorded through a Zeiss Axioscop coupled to a SPORT RT camera (Diagnostic Instruments). Contrast and brightness were adjusted in Photoshop CS3.