In order to characterize the presence of putative photoreceptor cells in the sea urchin larva we first consulted the transcriptomic expression of S. purpuratus opsins. After analyzing the publicly available RNAseq data coming from a survey of 10 embryonic stages (Tu et al., 2014) we concluded that, of the nine genes encoding opsins found in the genome, the Go opsin Sp-opsin3.2 (SPU027633) and the echinopsin Sp-opsin2 (SPU003451) are the only opsin genes expressed at significant levels. Starting from the late gastrula stage (48 h), these two genes show expression levels reaching the value of about 100 transcripts per embryo at the early pluteus stage (72 h), when neurons start to differentiate (for gene expression profiling see Supplementary Figure 1). Next, successful amplification of the Go opsin Sp-opsin3.2 was carried out, and the corresponding antisense riboprobe was used to localize the cells of interest. Unfortunately, various attempts in the amplification with different set of primers of the “echinopsin” Sp-opsin2 did not give any result.

Here, both RNA fluorescence (Figures 1A–C; 3 days post fertilization: dpf larvae, early four armed larvae) and chromogenic (Figures 1D–F; 4dpf, four armed larvae) whole mount in situ hybridization (WMISH) revealed that the Go opsin Sp-opsin3.2 is expressed in two cells arranged bilaterally adjacent the apical organ—i.e., a portion of the epithelium that form the oral hood that is considered to act as central nervous system of the larva (Byrne et al., 2007)—and at the base of the left and right anterolateral arms (for a schematic view of the 4 armed pluteus in which we included the terminology used in this work, see Figure 2). The expression of this gene in such a small number of cells is consistent with the above mentioned low levels of expression observed from the transcriptomic data. In order to identify the position of these Sp-opsin3.2 positive cells with respect to the ciliary band (the distinct thickening of ciliated epidermis that outlines the oral field and traces the edges of the four larval arms), cilia were labeled by immunohistochemistry with anti-acetylated α-tubulin after Sp-opsin3.2 WMISH. As shown by both in situ techniques (Figure 1), the main body of these cells appear to be located just orally to the thick epidermal band of the ciliated cells (see Figure 2 for schematic representation). The Sp-opsin3.2 positive cells are suggestive of the presence of a photoreception system in the sea urchin larvae.

A key difference between visual and non-visual photoreception system is the presence of shading structures, generally in the form of pigment cells, in proximity of light perceiving cells. Therefore, S. purpuratus larvae were observed under the light microscope at 4, 6, and 8 arm stages to detect for the presence of observable pigments that can be organized to act as shading for the described Sp-opsin3.2 positive cells. The only pigmented cells found in the vicinity of these cells were the granulated pigment cells, a particular population of red colored cells of dendritic morphology and immune role that are distributed all over the body (Ho et al., 2016). We therefore decided to explore the presence of screening pigments in the vicinities of the Go Sp-opsin3.2 opsin-positive cells by means of TEM.

Shading pigments involved in directional photoreception, which can be located both in the opsin positive cells or adjacently, are easily recognized in TEM as a group of black-solid dots in the cytoplasm (e.g., Marshall and Hodgson, 1990; Leys and Degnan, 2001). For our TEM analysis, three larvae were fixed, and transversal sections of 50–70 nm were made in the apical region in search of shading pigments (Figure 3A). Of them, micrographs corresponding to different sections of the apical organ (Figures 3D,E) and the bases of the left (Figures 3B,C) and right (Figures 3F,G) anterolateral arms were selected. Interestingly, none of the cells embedded in the apical organ nor in the area of the ciliary band exhibited observable shading pigments. In particular, the regions of the ectoderm in between the apical organ and either left or right anterolateral arms (encircled in Figures 3A,C,F) where the Sp-opsin3.2 positive cells are located (see also schematics of Figure 2) are void of shading pigment granules. These findings suggest that the Sp-opsin3.2 positive cells are not involved in directional photoreception. Serial multiplex immunogold labeling experiments are needed to better characterize the morphology of the encountered Sp-opsin3.2 positive cells.

Due to limitations of WMISH efficiency on late developmental stages and the availability of a specific antibody against the sea urchin rhabdomeric opsin Sp-Opsin4 (Ullrich-Lüter et al., 2011), we decided to use an immunohistochemical approach to explore the opsin toolkit of the premetamorphic larva. During late larval development (second week of development and thereafter), a portion of the coelom and the overlying ectoderm get in contact and form the imaginal adult rudiment (Smith et al., 2008; Heyland and Hodin, 2014; for a schematic view see drawings in Figure 4). This rudiment represents the presumptive juvenile that grows from the left side of the larva (for a schematic, see Figure 4A). In order to analyze the spatiotemporal expression of the rhabdomeric opsin Sp-Opsin4, we tested its presence in time series of 3, 4, 5, 6, and 7 days (4 armed pluteus), 16d (6 armed pluteus, contact flattened stage), 17d (6 armed pluteus, 5-fold mesoderm stage), 18d (8 armed pluteus, 5-fold ectoderm stage), 19d (8 armed pluteus, primary podia stage), 21d (8 armed pluteus, primary podia-folded stage), and 23d (8 armed pluteus, primary podia-touching stage) post fertilization (for staging of the echinopluteus see also Smith et al., 2008; Heyland and Hodin, 2014). These experiments suggested the absence of expression of the rhabdomeric opsin Sp-Opsin4 prior to the tube feet formation in any part of the larva. No protein expression was found either in sensu stricto larval structures until the 5-fold mesoderm stages (17dpf; Figures 4B,B′) with our method. Larvae started to exhibit Sp-Opsin4 positivity in conspicuous clusters of cells on the vestibular floor at pentagonal disc stage that would give rise to the tube feet disc during 5-fold ectoderm stage, a stage in which the ectoderm and the primordia of the five podia begin to push through the floor of the vestibular ectoderm (day 18; Figures 4C,C′). At this point, the interior of the 5 incipient podia are spherical in shape or shorter than wide. We also detected Sp-Opsin4 positive cells later on, in the tube feet disc during advanced rudiment stage, when the primary podia are taller than they are wide, but the podia are not yet folding in toward one another (day 20–21; Figures 4D,D′). At tube-foot protrusion stage (day 21–45), Sp-Opsin4 positive cells were detected both in disc (Figure 4E) and basal (Figure 4F) photoreceptors of the tube feet. These data indicate that the rhabdomeric opsin Sp-opsin4 may not regulate the photoreception mechanism of the larva, but only of the juvenile, where it appears to be involved in negative phototaxis (Ullrich-Lüter et al., 2011). For a schematic view on the different rudimental stages, see Figures 4B′–F′.

Phylogenetic analyses indicate the presence of at least seven opsins in the last common ancestor of Bilateria (Ramirez et al., 2016), thus suggesting that light reception had many roles very early in animal evolution. These opsins, together with present-day animal opsins, have been classified into four groups: (i) tetraopsins (Go-opsins, RGR/retinochrome opsins and neuropsins), (ii) xenopsins, (iii) Gq-opsins (including canonical and non-canonical r-opsins as well as “chaopsins”), and (iv) c-opsins, i.e., canonical c-opsins and bathyopsins (Ramirez et al., 2016). While the canonical c- and r-opsin groups have been extensively studied, little is known about the Go opsin group included in the tetraopsin clade (Gühmann et al., 2015).

In support of the ancient origin of the Go opsins, cells expressing this class of opsins have been localized in diverse animal clades, thereby suggesting the presence of this opsin group before the protostome-deuterostome split. Examples of Go-opsins are found in the ciliary cells of the eyes of the adult scallop Patinopecten yesooensis (Kojima et al., 1997), in the gastrula of the brachiopod Terebratalia transversa (Passamaneck and Martindale, 2013), in the rhabdomeric adult eye of the polychaete Platynereis dumerilii (Gühmann et al., 2015) as well as in the photoreceptor system here described. In the amphioxus Branchiostoma belcheri, a Go-opsin has been demonstrated by an in vitro analysis (Koyanagi et al., 2002) but, to our best knowledge, this is the first report in which the spatial expression of a Go-opsin has been described in a deuterostome larva.

In marine invertebrates, the expression of opsins in non-visual photoreceptors has been documented in the apical organ of planktonic larvae of protostome and deuterostome lineages (e.g., Arendt et al., 2004 and herein). A shared feature of these apical organs—regions specified by conserved developmental patterning mechanisms (Marlow et al., 2014)—is the presence of multiple sensory cells connected to the nervous system, which regulates ciliary beating and the vertical position of the animal in the water column (Tosches et al., 2014). In vertebrates, a population of non-directional retinal ganglion cells (the intrinsically photosensitive photoreceptive retinal ganglion cells: ipRGCs), are critical in relaying light information to the brain in order to control circadian photo-entrainment, pupillary light reflex, and sleep (Provencio et al., 1998; Schmidt et al., 2011).

Our discovery of non-directional photoreceptors in the pluteus of S. purpuratus suggests that these cells may also have a role in controlling the vertical position of the larva in the water column, which may be used for monitoring the time of day or the depth (Nilsson, 2013). This adjustment is likely to be achieved by modulating the length and frequency of ciliary arrests, as proposed for this and other marine larvae (e.g., Wada et al., 1997; Maldonado et al., 2003; Braubach et al., 2006; Jékely et al., 2008). The use of non-directional photoreceptors in vertebrates for tasks such as the regulation of nocturnal-diurnal behaviors (Provencio et al., 1998; Schmidt et al., 2011) could represent the retention of such chronobiological role.

The absence of shading pigments in the region where the Sp-Opsin3.2 is expressed strongly suggests that these opsin-positive cells lack directional sensitivity, but whether this represents a plesiomorphic character or a secondary loss is not immediately clear. Directional photoreception for phototaxis, with shading pigment near the site of opsin expression, is believed to have evolved from non-directional photoreception where screening pigment is not needed (Nilsson, 2009, 2013). Directional photoreceptors are typically bilaterally paired organs (Brandenburger et al., 1973; Arendt and Wittbrodt, 2001; Braun et al., 2015), whereas non-directional photoreceptors are often un-paired median structures (Mano and Fukada, 2007; Van Gelder, 2008). Our finding of paired non-directional photoreceptors represents an interesting intermediate.

The bilateral arrangement of photoreceptor cells is typically associated with helical swimming behaviors in the pluteus and other marine invertebrate larvae (Lacalli et al., 1990; reviewed in Jékely, 2009). Bilaterally paired photoreceptors may seem redundant for non-directional photoreception, and without shading pigment they do not have the directionality required for phototaxis.

It is possible that the shading pigment associated with these opsin-positive cells might have been lost during evolution to increase transparency or reduce energy expenditure. The lack of shading pigments may have been favored by selection to allow a better camouflage against predators (Nilsson, 1996). Consequently, the bilateral arrangement of these opsin positive cells may be primitive, and the lack of screening pigment a consequence of an adaptive transition from a directional to a non-directional role. Alternatively, it is possible that the pluteus have retained the non-directional photoreceptors of “Urbilateria,” an ancestor that may have had both directional and non-directional photoreceptors (Arendt and Wittbrodt, 2001). The bilateral arrangement of these non-directional photoreceptors would have been the result of developmental constrains associated with the bilateral symmetry, or maybe profitable for increasing the robustness and sensitivity of the photoreceptor system.

To better understand when a possible switch occurred (i.e., whether Go-opsins originally mediated a non-directional task in the dipleurula larvae of the Ambulacraria stem group, or if an association with screening pigments was lost secondarily in the Echinodermata crown group) a further comparison of the photoreceptor systems of different dipleurula-type larvae is required.

The fact that the two bilateral photoreceptors connected to the apical organ of the pluteus larva use a Go-opsin, while r-opsins are present in similar structures of nearly all other larvae, results remarkable. One possible explanation to why putative homologous paired photoreceptors express distinct opsins in different Bilateria clades could be that Urbilateria had bilaterally paired photoreceptors with r-opsin, c-opsins and Go-opsins serving different functions (Feuda et al., 2012; Ramirez et al., 2016). This variety of functions can be ascribed to the need of different spectral or temporal properties, as well as to different roles in the transducing the light signal. Losses would then account for the fact that echinoid larvae seem to have only a Go-opsin, most other protostomes only a r-opsin, and vertebrates c- and r-opsins. Cell duplication and subsequent specialization must also be assumed for vertebrates.

The most plausible role of the Go photoreceptors described in this study is the regulation of vertical movement of the larva during photoperiodic transitions (Jékely et al., 2008; Mason and Cohen, 2012). Such a unimodal system could resemble the earliest photoreceptor mechanism in the first marine larvae. If this is the case, study of this system could provide clues as how the first planktonic animals perceived light cues. It remains possible that other opsins are present at the same larval stage that have not been identified.

Because the main locomotory organ of the pluteus is the ciliary band, it would be informative to know whether the Sp-Opsin3.2 positive cells are connected to the ciliary band via the nervous system, which has been described as “a network of cells that span the blastocoel and connect nearly all parts of the larva” (Ryberg and Lundgren, 1977). Previous studies of the nervous system of the pluteus of Strongylocentrotus droebrachiensis (Burke, 1978), a closely related species, report the presence of serotonergic neurons in the area of the apical organ, located between the cells homologous to the Go-opsin expressing cells of S. purpuratus. This serotonergic system is suggested to be involved in the regulation of the ciliary band activity in the pluteus (Gustafson et al., 1972; Burke, 1978; Yaguchi and Katow, 2003) and also in many other marine larvae (e.g., Mackie et al., 1969; Beiras and Widdows, 1995; Pires and Woollacott, 1997; Kuang and Goldberg, 2001). The topology of the SpOpsin3.2 expressing cells in the proximity of serotonergic neurons lead us to hypothesize that Go expressing cells may be involved in locomotory control, probably in the activation or excitation of the ciliary band to position the animal in the upper photic zone. Knockout experiments of this opsin coupled with behavioral experiments could be used to test this hypothesis.

Adult S. purpuratus were obtained from San Diego Bay at 25–30 m in depth (San Diego, CA, USA) and housed in 12°C circulating seawater aquaria at the Stazione Zoologica Anton Dohrn, Italy. Spawning was induced by intracoelomic injection of 0.5M KCl. Embryos/larvae were cultured in Mediterranean filtered seawater (mesh pore size: 0.2 mm) diluted in de-ionized water (final salinity: 32.5‰) and kept at 15°C on a 12/12 h light/dark cycle. From 3 days onwards, larvae were fed with a mixed diet of Isochrysis galbana [~2,000 cells mL⁻¹] and Rhodomonas sp. [~2,000 cells mL⁻¹]. All larval cultures were maintained at a decreasing with age concentration from 5 to 1 pluteus mL⁻¹, mixed by gentle rotary stirring and washed every other day. Larval washes were made by inverted filtration (mesh size: 100 μM).

Contig sequence for Sp-opsin3.2 was identified in the genome (ref. code: SPU027633) and transcriptome (ref. code: WHL22.338995) data sets. A 1,175 bp transcript was amplified by PCR with the cloning primers Sp-opsin3.2-F (5′-CCACTCATTTCGTGCGGATT-3′) and Sp-opsin3.2-R (5′-CTCTAGTGATGACGGGCGAT-3′) from cDNA prepared with a Bio-Rad iScript synthesis kit, ligated into pGEMT-easy vector (Promega), and transformed into Top10 chemically competent E. coli (Invitrogen). Clone fragments were verified by Sanger sequencing prior to riboprobe generation. DIG-labeled antisense and sense (negative control) RNA probes were generated from plasmid DNA with T7- and SP6-RNA polymerases (Roche) respectively, and purify with mini Quick Spim Columns (Roche).

Strongylocentrotus purpuratus larvae were collected at early pluteus stage (3dpf), fixed overnight at 4°C in 4% paraformaldehyde/0.1M MOPS pH 7, 0.5M NaCl, washed thoroughly in MOPS buffer, and stored in 70% ethanol until use. Whole mount fluorescent in situ hybridization (FISH) was performed as described in Andrikou et al. (2013). Immunohistochemistry coupled to WMISH was performed by incubating the larvae with anti-acetylated α-tubulin antibody (Sigma-Aldrich T6793, St Louis, MO, USA) in a dilution 1:250 together with the anti-DIG antibody; the secondary antibody was a goat anti-mouse IgG-Alexa 488 (Invitrogen, CA, USA) diluted 1: 1000.

Four-armed S. purpuratus larvae were fixed at 4dpf as explained above. Single probe chromogenic in situ hybridization on whole mount fixed embryos was performed as previously described by Ransick (2004) with the following changes: (i) all washes were carried out in TBST (0.2M Tris pH 7.5, 0.15M NaCl, 0.1% Tween-20); (ii) hybridization was performed over-night at 60°C; (iii) 1X SSC and 0.1X SSC washes were performed at 60°C; (iv) Anti-Digoxigenin-AP, Fab fragments (Roche) were diluted 1: 2000.

Larvae were fixed in 4% paraformaldehyde in PBS pH 7.4 containing 0.5M NaCl for 30 min at room temperature. Late 6 and 8 armed larvae (days 14–23) were post-treated 2 min with pure cold MetOH in order to partially remove membrane lipids and facilitate antibody penetration. After five 5 min rinses in phosphate buffered saline (PBS), samples were washed thoroughly in PBS/0.1% Tween-20 (PBST). Following incubations were carried out on an orbital shaker. The first blocking step was performed with 4% heat-inactivated Normal Goat Serum (NGS) in PBST for 1 h, prior to incubating specimens with primary antibodies anti-Sp-opsin4 1:50 [1.21 mg mL⁻¹] (Ullrich-Lüter et al., 2011), anti-1E11 1:100 [~10.00 mg mL⁻¹] (monoclonal antibody that recognize S. purpuratus synaptotagmin B and is used as “pan-neural” marker; Nakajima et al., 2004), and anti-acetylated α-tubulin (Sigma T6793) 1:250—in PBST overnight at 4°C. After five washes in PBST, a second blocking step was performed as described above prior to incubating specimens with secondary antibodies—goat anti-rabbit IgG-Alexa 488 and goat anti-mouse IgG-Alexa 647—diluted 1: 1,000 in blocking buffer (4% NGS in PBST) at 4°C overnight. All specimens were washed thoroughly in PBS and then counterstained with DAPI (1 μg/mL in PBS) for nuclear labeling. For Sp-opsin4 antibodies, controls were carried out using their respective rabbit pre-immune sera. For commercial antibodies, control experiments were run in parallel by omitting primary antibodies.

Strongylocentrotus purpuratus plutei were first fixed in modified Karnovsky solution (2.5% glutaraldehyde, 2% paraformaldehyde, and 3% sucrose in 0.1 M phosphate buffer pH 7.4 containing 0.5M NaCl) for 1 h at room temperature. After several rinses in PBS, samples were post fixed in 1% osmium tetraoxide in distilled water 1 h at 7°C, and dehydrated in a series of ethanol (30/50/70/96/100) and infiltrated and embedded in EPON (Agar 100). Samples were kept at 60°C for 48 h to allow polymerization. Thin sections (50–70 nm) were cut with a diamond knife with a Leica EM UC7 ultramicrotome and mounted on pioloform coated copper grids.

Light microscopic images were taken using a Zeiss M1 Axio Imager microscope. Confocal acquisition was performed on a Zeiss LSM 510 Meta confocal microscope. TEM acquisitions were performed on a 120 kV JEOL 1400 plus microscope with a bottom mounted CMOS camera. Figure plates were made with Illustrator CS6 (Adobe). Brightness/contrast and color balance adjustments were always applied to the whole image and not to parts.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.