General morphology of the sensory pits

The body of Stenostomum brevipharyngium is divided into the rostrum, pharynx, and trunk (Fig. 1A), and the sensory pits are positioned in the middle of the rostral area. When examined via light microscopy, they appear as two conspicuous globular cavities occupying the midsection of the head (arrowheads, Fig. 1A). Scanning electron microscopy (SEM) clearly reveals the openings of the cavities to the external environment, which are devoid of the otherwise ubiquitous surface cilia (arrowhead, Fig. 1B, C). In confocal laser microscopy (CLM) the sensory pits can be visualized with DAPI staining of cell nuclei (as empty cavities on the lateral sides of the brain, arrowheads, Fig. 1D), with antibody staining against tyrosinated tubulin (arrowheads, Fig. 1E) that reveals ciliated and nervous structures and with phalloidin staining of actin filaments (arrowheads, Fig. 1F).

Fig. 1

General morphology and localization of sensory pits (arrowheads) in Stenostomum brevipharyngium. A Light microscopy image of a living worm, showing division into rostrum, pharynx, and trunk. B Head region of the worm in the lateral view under scanning electron microscope. C Magnified view of the pit opening. Horizontal optical sections through the anterior region of the worm at the level of ciliated pits stained with DAPI for cell nuclei (D), antibodies against tyrosinated tubulin (E), and phalloidin for F-actin (F). Abbreviations: br brain, g gut, mo mouth opening, np brain neuropile, ph pharynx, pn protonephridium, r rostrum, rm rostral musculature, t trunk, tm trunk musculature

Each of the pits is composed of an internal cavity, which laterally opens through the pit opening to the external environment, and the fundus — a bottom part of the pit (Fig. 2). The opening is equipped with a delicate sphincter muscle (ps, Fig. 2C and D), that likely controls its width. Although the only muscle fibers that are directly associated with the pits are sphincters, there are numerous additional muscles in the areas adjacent to the organs, such as longitudinal, helical, and circular muscles of the rostrum (Fig. 2B–D). It is likely that those muscles are responsible for the frequent changes in the shape and apparent position of the sensory pits that can be observed in living worms. The fundus can be stained with phalloidin and antibodies against tyrosinated tubulin (f, Fig. 2A–C) and does not contain any cell nuclei. Close examination of the tyrosinated tubulin staining reveals continuity between the dense tyrosinated tubulin immunoreactive (tyrTub-IR) projections forming the fundus and neurites that connect the sensory pit with a brain neuropile. The neurites form two bundles, here named nerves of sensory pit 1 and 2 (nsp1 and nsp2, Fig. 2B and D). The nerves are tyrTub-IR and although they follow slightly different paths in the rostrum, they both connect the fundus with the same lateral region of the brain neuropile, just anterior to the root of the main ventrolateral nerve cord (Fig. 2B and D). In several of the fixed specimens, we noticed an apparent secretory discharge from the opening of each of the sensory pits, which was stained with phalloidin and antibodies against serotonin (double arrowheads, Fig. 2A and C). This discharge likely corresponds to the mucus-like substance that has been reported to fill the internal cavities of the pits in Stenostomum leucops [9, 10, 12].

Fig. 2

Details of the fully formed sensory pits. A Optical horizontal section through the broadest area of the pits showing a fundus, pit cavity, and mucus-like discharge (double arrowheads). B Maximum intensity projection showing the nerves of the sensory pit and rostral musculature. C Maximum intensity projection showing sphincter and fundus of the sensory pit and mucus-like discharge (double arrowheads). D The schematic drawing of the sensory pits in the context of the internal anatomy of the head. Cell nuclei stained with DAPI in cyan, a signal from antibodies against tyrosinated tubulin in yellow, F-actin stained with phalloidin in green, a signal from antibodies against serotonin in red. All panels show dorsoventral sections, anterior to the top. Abbreviations: br brain, f fundus, np brain neuropile, nsp nerve of the sensory pit, ps sphincter of the sensory pit, rm rostral musculature, rn rostral nerves

Ultrastructure of the pit

We next examined the ultrastructure of the pit with transmission electron microscopy (TEM). Specifically, we examined sections at the approximate level of the maximal width of the pit (Sect. 1; Fig. 3A and B) and at its ventral extremity (Sect. 2; Fig. 3C and D).

Fig. 3

Ultrastructure of the sensory pit. A The optical horizontal section showing the approximate area in panel B. B The ultrathin horizontal section through the widest part of the sensory pit and associated structures. C The optical horizontal section showing the approximate area in panel D. D The ultrathin horizontal section through the fundus of the pit and associated structures. E Details of the pit as visible in panel B. F Details of the fundus as visible in panel D. Longitudinal (G) and cross (H) section through the cellular processes forming fundus of the pit. I Cell bodies of the pit-forming cells as visible in panel D. J Schematic reconstruction of the fundus-forming cell. Cell nuclei stained with DAPI in cyan, a signal from antibodies against tyrosinated tubulin in yellow, and F-actin stained with phalloidin in green. The lettered boxes in panels A, C, D, E, and F indicate regions magnified in the corresponding panels. The arrows in panel E indicate muscle fibers of the sphincter. Abbreviations: ap apical portion of the fundus cell, av apical vesicle, ax axon, bl brain lobe, cp cavity of the pit, f fundus, lbv large basal vesicle, lm longitudinal muscle, mt mitochondrion, mv microvilli, np brain neuropile, nsp nerve of the sensory pit, nt neurite-like process, nu nucleus of the fundus cell, oc outer cell, rm rostral musculature, sbv small basal vesicle

Section 1 shows that the internal cavity is completely filled with long, convoluted threads, which can be identified as extensive microvilli originating from the surrounding cells (mv, Fig. 3E–H). Although the opening of the pit is not present in Sect. 1, the sphincter muscle is visible in the cross-section (arrows, Fig. 3E), as expected from the corresponding optical section obtained with CLM (arrows, Fig. 3A).

The fundus of the pit consists of elongated cell processes which were cut longitudinally in Sect. 1 (f, Fig. 3E, G) and transversely or obliquely in Sect. 2 (f, Fig. 3F and H). Those densely arranged processes apically give rise to the aforementioned extensive microvilli (Fig. 3G and H). Additionally, their apical regions are packed with numerous small apical electron-translucent vesicles (av, Fig. 3G, H, and J). Slightly below the apical vesicular region, conspicuous mitochondria are filling the process (mt, Fig. 3G, H, and J). Approximately 1.5 μm below the fundus surface, the cellular processes are becoming considerably thinner and start to extend as neurite-like fibers that form the nerves of the sensory pit (nt and nsp2, Fig. 3B, D–G). These neurites connect on the other side of the nerve to the bodies of neuron-like cells residing in the lateral lobes of the brain. Although we cannot directly trace the connection between particular brain perikarya and cellular processes at the fundus of the pit, we suggest that the cell bodies (with nuclei) of the fundus-forming cells are located in the lateral parts of the brain lobes. Such an arrangement would explain the lack of cell nuclei in the fundus of the pit that we observed with both CLM and TEM. Besides the nucleus, those cell bodies also contain large and small electron-translucent vesicles (lbv and sbv, Fig. 3i and J), grouped together close to the base of the neurite-like projection.

Based on these results, we reconstructed the fundus-forming cells as resembling neurons (Fig. 3J). The cell body, containing the nucleus and basal vesicles, is located in the lateral lobes of the brain and sends out neurite-like projection, which forms the nerves of the sensory pit. Apically, the projection swells, forming a structural unit of the fundus, that harbors mitochondria and apical vesicles, and which also gives rise to the modified microvilli that fill the cavity of the pit. The results of antibody stainings suggest that both neurite-like projections (nt) and apical portions of these cells (ap) are reinforced with tyrosinated tubulin fibers. We could not detect axons emanating from those cells in our sections. However, as antibody staining revealed continuous neural connections between the fundus of the sensory pit and the neuropile, it is likely that the cells also project axons to the neuropile.

The outer walls of the pit (lateral to the edges of the fundus) are formed by the outer cells that also have extensive microvilli (oc, Fig. 3E, F), but otherwise do not resemble the fundus cells. A nucleus is the only internal organelle that can be unequivocally identified within those extremely thin cells. Although it is difficult to ascertain, the fiber of the sphincter muscle seems to be also associated with the outer cell (arrows, Fig. 3E).

Altogether our data suggest that the sensory pit of S. brevipharyngium is mostly composed of only two cell types — the outer cells, and the fundus-forming cells (that are partially positioned in the brain and also form the nerves of the sensory pit). Importantly, neither of those cell types contains elements of the ciliary apparatus and there are no cilia within the inner cavity of the pit, as evident from antibody staining and TEM sections.

Formation of the pits during asexual development

The asexually reproducing individuals of S. brvipharyngium can be distinguished in cultures by longer trunks and an inconspicuous transverse furrow that bisects the trunk at the site of the future fission plane. We used Hoechst, phalloidin, and tyrosinated-tubulin staining to visualize the internal structures of the developing zooid and to stage the temporal sequence of the fission process. At the earliest stage 1 (Fig. 4A), two cell accumulations on each side of the gut and a thin tyrTub-IR commissure connecting the lateral nerve cords constitute the earliest anlage of the future head. At stage 2 (Fig. 4B), the future neuropile thickens via the addition of further neurites, the gut becomes considerably constricted, and the remodeling of splanchnic musculature in the future rostral region becomes evident. In the following stage 3 (Fig. 4C), the brain gains a bilobed shape with the neuropile already resembling its final form. The rudiment of the newly formed zooid already adopts the shape reminiscent of the final head, while the rostrum and pharynx are still undergoing development. At stage 4 (Fig. 4D) the rostral and pharyngeal musculature is established and the general appearance of the zooid rostrum resembles that of a fully formed worm, although it is proportionally smaller. Stage 4 is directly followed by fission between the maternal and new zooids.

Fig. 4

Formation of the pits during asexual development, as visualized with antibodies against tyrosinated tubulin (yellow, A–D), phalloidin staining for F-actin (green, A′–D′), and Hoechst staining for cell nuclei (cyan). Schematic drawings of the pit formation (A″–D″), scale bars 10 μm. Abbreviations: br brain, f fundus, gr reorganizing gut tissue, lnc longitudinal nerve cord, np brain neuropile, nsp nerve of the sensory pit, ph pharynx, ps sphincter of the sensory pit, rm rostral musculature, rn rostral nerves, sm splanchnic musculature

The earliest rudiments of the sensory pits can be already observed at stage 1 (arrowheads Fig. 4A), when some of the epidermal cells at the level of the division furrow form small depressions that show strong tyrosinated tubulin immunoreactivity and weaker staining with phalloidin. The short tyrTub-IR projections fan out from these depressions towards developing brain rudiments. At stage 2, the depressions become slightly deeper, the tyrTub-IR projections extend further towards the forming brain neuropile and the phalloidin staining at the prospective fundus area becomes stronger (Fig. 4B). At stage 3, the pits already acquire the form of large empty cavities with a clearly visible fundus, while some of the tyrTub-IR projections reach the neuropile and thus establish the nerves of the sensory pits (Fig. 4C). Already at this stage, there are no cell nuclei associated with the fundus of the pit. Finally, at stage 4, the sphincters of the pit opening become evident in phalloidin staining (Fig. 4D) and the pits take their final form with a contractile opening, inner cavity, fundus, and nerves.

Formation of the pits during head regeneration

Head regeneration in S. brevipharyngium takes about 4 days to complete. The first indication of the formation of sensory pits can be observed at 48 h post-amputation (hpa). At this regeneration stage, the brain rudiment with two lobes and the main commissure of the neuropile is already formed (Fig. 5A). The rudiments of the pits are visible as small invaginations on the surface of the regenerating head that could be stained with phalloidin and antibodies against tyrosinated tubulin (arrowheads, Fig. 5A). Some fine tyrTub-IR projections extend from these invaginations, penetrating the brain rudiment. At 72hpa, the sensory pits are already divided into an internal cavity and fundus that can be visualized with phalloidin and tyrosinated tubulin staining (Fig. 5B). At this stage of head regeneration, the brain neuropile has already reached its final shape and the thick tyrTub-IR nerves of sensory pits can be detected, connecting the neuropile with the pits (nsp, Fig. 5B). At 96hpa, when most of the head structures are already fully regenerated, the sphincter of the sensory pit opening can be visualized by phalloidin staining (Fig. 5C), thus constituting the last regenerating structure of the organ.

Fig. 5

Formation of the pits during head regeneration, as visualized with antibodies against tyrosinated tubulin (yellow, A–C), phalloidin staining for F-actin (green, A′–C′), and Hoechst staining for cell nuclei (cyan). Schematic drawings of the pit formation during regeneration (A″–C″), scale bars 10 μm. Abbreviations: br brain, f fundus, lnc longitudinal nerve cord, np brain neuropile, nsp nerve of the sensory pit, ph pharynx, ps sphincter of the sensory pit, rm rostral musculature, rn rostral nerves

Gene search

To study gene expression in the sensory pits of S. brevipharyngium we first needed to generate a reference transcriptome for this species. We pooled together thousands of worms at different stages of asexual development and regeneration, generated the cDNA library, and sequenced it using the Illumina next-generation sequencing method (see the “Methods” section for details). The raw reads were de novo assembled using an established pipeline [47] into the reference transcriptome of S. brevipharyngium (see method section for details), which contained 35,979 unique transcripts. We searched this transcriptome using sequences of the six transcription factors (TFs): otx, pax4/6, dach, tll, emx, and svp, which in the nemertean Lineus ruber have been shown to be expressed in the developing cerebral organs [42]. The BLAST search identified 10 unique sequences that were confirmed as putative orthologs of those six candidate genes by reciprocal BLASTP search against the NCBI database of protein sequences (Additional file 1: Table S1). The identity of each of those TFs has been further confirmed by the phylogenetic analysis of the protein sequence (Additional file 2: Figs. S1–S5). Four out of those six TFs (otx, pax4/6, tll, and emx) have been duplicated in S. brevipharyngium, resulting in a total of ten gene targets for expression analysis.

To test whether the sensory pits of Stenostomum might have a photoreceptive function, we also searched the S. brevipharyngium transcriptome and published transcriptomes of S. leucops and Paracatenula sp. for opsin gene homologs. Despite the use of different opsin gene sequence queries (see the “Methods” section for details), we were unable to detect opsin homologs in any of the catenulid transcriptomes. The closest BLAST hits always corresponded to other G-protein coupled receptors (e.g., neuropeptide Y receptor, 5-HT receptor, octopamine receptor, tachykinin receptors, dopamine receptor D2, neuropeptide F receptor, and muscarinic acetylcholine receptor (Additional file 1: Table S1)), indicating that opsin may have been lost in the examined catenulids.

Gene expression

To gain insights into the molecular specification of the sensory pits, we examined the expression of the ten identified orthologs of the candidate transcription factors (TFs), otx, pax4/6, dach, tll, emx, and svp. For all of the genes, with the exception of svp, we performed colorimetric in situ hybridization (CISH) on the worms in an asexual phase. The genes were expressed in multiple cell types/tissues: in the pharynx (genes: otxA, otxB, pax4/6A, pax4/6B, tllB, emxA, and emxB; red arrowheads, Fig. 6B, E, H, I, K, L, Q–X), in the brain (genes: otxB, pax4/6A, pax4/6B, and emxB; double white arrowheads, Fig. 6B–I, W, X), in the longitudinal nerve cords (genes: otxB, pax4/6A and pax4/6B; white arrows, Fig. 6B, E–H), in the gut tissue (genes: tllA, emxA, and emxB; blue arrowheads, Fig. 6N–P, T–V, X), in the additional distinct pharyngeal domains (gene: emxB, double red arrowheads, Fig. 6V–X), around the mouth opening (gene: emxB, green arrowhead, Fig. 6V) and in individual cells spread throughout the worm body without any clear pattern (genes: otxA, dach, and tllB; magenta arrowheads, Fig. 6A, K, L, Q, R). However, from all of the tested TFs, only pax4/6A showed expression in the sensory pits, both in the fully formed and in the developing organs (black arrowheads, Fig. 6E–G).

Fig. 6

Colorimetric RNA in situ hybridization of the studied transcription factors in the worms in the asexual phase. For each panel, the name of the hybridized gene is indicated in the box above. Sensory pits are marked with a dotted circle. The expression was detected in the following morphological structures: pharynx (red arrowheads), brain (double white arrowheads), longitudinal nerve cords (white arrows), gut tissue (blue arrowheads), distinct pharyngeal domains (double red arrowheads), around the mouth opening (green arrowhead), single cells spread throughout worm body (magenta arrowheads) and in the sensory pits (black arrowheads). Worms are mounted dorsoventrally in all panels with the exception of the laterally mounted animal in panel U. Scale bars 20 μm

Next, we studied the expression of the genes svp and pax4/6A with RNA hybridization in situ chain reaction (HCR), which allows visualization of the gene expression at the cellular level. The gene svp is expressed only in two neurons residing close to the posterior part of the neuropile (Fig. 7A and B). In the head region, the signal from pax4/6A could be detected in the fundus of the sensory pits, in the brain, including the areas in which the cell bodies of the fundus-forming cells are likely located, and in the cells that form sensory pits at the different stages of their development in the asexual zooids (Fig. 7D–F).

Fig. 7

Expression of the genes svp and Pax4/6 as visualized with the RNA in situ hybridization chain reaction. For each panel, the name of the hybridized gene is indicated in the bottom left corner. Gene expression in the sensory pits is marked with a dotted circle

Ancestral state reconstruction

Finally, to test whether the sensory pits of Stenostomum are homologous to the similarly positioned ciliated pits of other flatworms we performed ancestral state reconstruction. First, we inferred the phylogeny of flatworms using four molecular markers (18S, 28S, COI, and ITS-5.8S) that are available for a wide range of platyhelminths including multiple catenulid species and other groups where those organs have been reported (Additional file 1: Table S2). Our species selection aimed at 1. covering evenly all major clades of flatworms, 2. including representatives of the groups for which sensory organs were reported, and 3. choosing species for which a maximum number of molecular markers was available.

The obtained maximum likelihood tree (Additional file 2: Fig. S6) has high support values for major clades of flatworms and shows a similar topology to the already published transcriptome-based phylogenies of platyhelminths [1, 2]. For instance, we recovered the sequential branching of Catenulida, Macrostomorpha, and (Polycladida + Prorhynchida) from the remaining flatworms, the monophyly of Neodermata and Adiaphanida (Additional file 2: Fig. S6). We also recovered the internal topologies of important clades congruent with the published phylogenies of those groups — e.g., (Stenostomidae + (Catenulidae + Paracatenulidae)) within catenulids [3, 11, 48], (Haplopharyngidae + (Macrostomidae + Dolichomicrostomida)) within Macrostomorpha [49], and (Maricola + (Geoplanoidea + Planarioidea)) within triclads [50,51,52]. A few differences from the topologies of the transcriptome-based phylogenies (e.g., position of bothrioplanids, proseriates, and rhabdocoels), are likely the effect of insufficient phylogenetic signal in the limited number of molecular markers. However, they should not have profound effects on overall ancestral state reconstruction, especially at the base of the tree.

Based on the available literature we scored the presence or absence of the sensory/ciliated pits in different species included in our analysis and performed ancestral state reconstruction using stochastic character mapping (see the “Methods” section for details). Our results clearly indicate that sensory/ciliated pits evolved at least six times independently within flatworms — in stenostomids, microstomids, prorhynchids, bothrioplanids, pseudostomids, and geoplanoids (Fig. 8), refuting the hypothesis of their reciprocal homology.

Fig. 8

The ancestral state reconstruction of the presence of the sensory pit-like organs in flatworms. The tree topology was inferred based on a maximum-likelihood analysis of concatenated 18S, 28S, ITS-5.8S, and COI datasets (Additional file 2: Fig. S6). The pie charts show the likelihood of the character states at each node. Schematic drawings of the chosen representative flatworms show the position of their sensory pit-like organs (in red) in relation to the body outline (light gray) and gut (dark gray)