Streptoneury is independent from ontogenetic torsion in the caenogastropod snail Marisa cornuarietis


 A hallmark in snails’ anatomy is the conspicuous crossing of the pleurovisceral nerve cords present in all but the most derived gastropod clades. This feature is called streptoneury and hitherto near-universally believed to derive from the process of torsion which is, ontogenetically, visible by a 180° rotation of the visceral sac relative to the cephalopodium, being also responsible for the formation of a cranially bent gut and the location of gills in a mantle cavity that opens to the anterior. However, the mechanical link between the ontogenetic rotation of the visceropallium and streptoneury has never been demonstrated directly. After suppressing ontogenetic torsion in the freshwater apple snail Marisa cornuarietis, we could show in a 3D reconstruction based on serial sectioning that the nervous system of the non-torted snail almost identically mirrored the classical organization of a normally developed individual and showed all features of streptoneury in this species. Furthermore, confocal laser scanning microscopy revealed the pleurovisceral cords not to be fully shaped after completion of ontogenetic torsion. We therefore conclude that, ontogenetically, and potentially also phylogenetically, torsion is not an implicit prerequisite for streptoneury, thereby fundamentally challenging a century-old ‘certainty’ in molluscan developmental biology and evolution.

The origin of the various body plans of molluscs has been a matter of speculation for a long time. More than 130 years ago, Lankester 1 proposed the anatomy of a 'schematic mollusc' which later has been adopted as a blueprint for the ʽancestral mollusc´s' body plan2,3 that has formed the basis for generations of textbooks. Today, there is almost unanimous agreement that the supposedly single-shelled unsegmented 4 ancestral mollusc´s mantle cavity, or bilaterally positioned mantle cavities, together with the ctenidia (gills) and the anus were confined to the posterior end of the body, and the pleurovisceral nerve cords did not cross (Fig.   1A). Present-day snails show an anterior position of the mantle cavity, ctenidia and anus, and feature a crossing of the pleurovisceral nerves, the so-called streptoneury, in all but the most derived extant gastropod classes. This is explained by a hypothetical phylogenetic process, called ʽtorsion', that rotated all anatomical components of the visceropallium by 180° relative to the cephalopodium 2,5,6 . To harmonize this theory with embryological observations of gastropod early development from the early 1900s Garstang 7 proposed an evolutionary saltation by a macromutation that altered the embryonic development of an ancient, pregastropod mollusc and gave rise to the ʽontogenetic torsion' still preserved in the developmental programme of today´s gastropods. According to this theory, the ontogenetic torsion is represented by the clearly visible 180° counterclockwise rotation of the visceropallium, directly resulting in the establishment of streptoneury, together with a U-shaped gut and an anteriorly positioned mantle cavity in extant snails 4,8 . However, more than a decade ago this causal relation, as well as the view on torsion as a uniform process have been questioned 9 based on embryological observations in the vetigastropod Haliotis kamtschatkana 10 as not all components of the visceropallium were found to rotate synchronously in this species. In consequence, Page 9 proposed the formation of the gastropod´s anterior mantle cavity from only a single cavity on the right of, originally, a bilateral set of mantle cavities (ʽasymmetry hypothesis'). Nevertheless, also this hypothesis involves a rotation of visceropallial structures by 90°-180° which, according to figure 6 in that publication, supposedly leads to streptoneury. However, it has never been directly demonstrated that ontogenetic torsion mechanically twists the pleurovisceral nerves and thus leads to streptoneury, because nerve cords that join the posterior ganglia in adult snails could not be visualized in gastropod embryos or larvae 11 . Even though several serotonin-, FMRFamide-and catecholamine-immunoreactive cells have been found in early embryonic and larval stages of snails they seem to 'disappear' later and are replaced by an 'adult' nervous system after a short time of coexistence and cooperation [11][12][13][14] . Thus, despite of its implied conclusiveness, the proof for a causal linkage of ontogenetic torsion and streptoneury in the adult gastropod has not yet been provided. In the caenogastropod apple snail, Marisa cornuarietis (Linnaeus, 1758), the classical anatomy of the nervous system, including streptoneury, is secondarily modified (Fig. 1B). In the apple snail family, Ampullariidae, the right pallial ganglion, called subintestinal ganglion after torsion, has moved to the anterior and fused with the right pleural ganglion, thus disguising the crossing of the pleurovisceral cords to some extent. Nevertheless, the presence of the supraintestinal nerve transversely crossing the longitudinal midline body axis clearly demonstrates streptoneury also in ampullariids. On the left side of the body, a secondary nerve, the left zygosis, connects the left pleural ganglion with the supraintestinal ganglion which, in turn, is connected to the osphradial ganglion which develops at a later embryonic stage by proliferation of the tissue below the osphradium itself. The anatomy of the nervous system of M. cornuarietis has been described in detail by Demian and Yousif 15 (Fig. 1C).
To clarify whether, in M. cornuarietis, streptoneury is a direct mechanical consequence of ontogenetic torsion we used two approaches. First, we investigated whether the pleurovisceral cords are already present prior to the rotation of the visceropallium, so that they have a chance to be twisted during this process or whether they are formed for the first time after ontogenetic torsion. Second, we used a methodology to block ontogenetic torsion during the embryonic development of M. cornuarietis and 3D reconstructed the nerve system of a non-torted individual after completed embryonic development. The latter was possible by chemically preventing the TGF-β cytokine-dependent 16

Results
Using confocal laser scanning microscopy (CLSM), we could visualize all prominent ganglia  [ Figure 3]

Discussion
Our results lead to the following conclusions and proposals: (1) Previous failures to visualize pleurovisceral nerve cords in pre-torsion stages of snail larvae are likely not to be attributed to inadequate staining techniques, but rather to the absence of them in these stages. Despite its high resolution and potential to visualize single neurons, also the CLSM method used here did not provide any evidence for the presence of pleurovisceral nerve kamtchatkana, the first two neurites of a neuron that was formed before ontogenetic torsion and appeared to delineate the trajectory of the future pleurovisceral nerve cords did not cross over during torsion because this neuron´s soma was shifted in the same direction as the rotating visceropallium 14 . Nevertheless, also in that case a full crossing of the pleurovisceral connectives occurred secondarily at later stages in ontogeny. In the pulmonate snail, Lymnaea stagnalis, the pathways of embryonic neurites seem to exhibit streptoneury and, later, also detorsion (a feature typical for Pulmonata) but these structures did not appear to join the ganglia of the future adult nerve system, ceased to express immunoreactivity and disappeared after hatching 12,13 . We therefore must conclude that the establishment of streptoneury in the post-embryonic nerve system is a secondary process and, at most, triggered by the early neurons´ pioneer pathways and their signals which may act as guidance for the neuronal growth of the developing adult nerve system 11 .
(2) The development of streptoneury is independent of whether or not the parietal ganglia have changed places during ontogenetic torsion, at least in M. cornuarietis. Plausibly, any rotation of the visceropallium may mechanically relocate structures. This is confirmed by our study as the observable 90° movement of the visceropallium in the non-torted individual 19 (Fig 4C), which most likely is due to the increasing weight of the growing shell that pulls down the left side of the body went along with a shift of the position of at least 3 posterior ganglia. On the one hand, the osphradial ganglion that develops by proliferation directly below the osphradium was shifted ventrally (Fig. 3D) and, on the other hand, the arrangement of the supra-and subintestinal ganglia (which are considered homologous to the parietal ganglia; cf. Fig. 1A-B) relative to one another was tilted in the same direction and came to lie at almost the same height (Fig. 3E). Even though torsion likely is not a uniform process 9,17 and thus may result in gradual interspecific variation in the degree to which different organs are displaced 9 , we do not question the general possibility that the right and the left parietal ganglion change places in the course of the rotation of the visceropallium. However, in the non-torted (torsionally suppressed) M. cornuarietis individual, we assume that the two parietal ganglia have not been interchanged because the visceropallium does not show any horizontal movement (which is supposed to be the driving process behind this exchange). Although the sub-and supraintestinal ganglia of the non-torted individual almost certainly originate from the right and the left parietal ganglia, respectively -and not, vice versa, from the left and the right parietal ganglia (as traditionally assumed in torted snails) -, the key criterion of streptoneury, i.e. the transversal crossing of the midline axis by the supra-and subintestinal nerves (as shown in Fig. 1B: right, C), was nevertheless established also in the nontorted snail. Our findings allow the explanation that, in torted Marisa snails, the pleurovisceral connectives, which are formed after ontogenetic torsion, grow crosswise towards the opposite ganglia across the central axis. The same occurs when ontogenetic torsion has been blocked: also in this case, the pleurovisceral connectives connect the opposite ganglia in a crosswise manner. This suggests that, independent of a rotation of the visceropallium, the ultimately relevant information for the formation of the adult nervous system, including streptoneury, is established at the earliest at a point in time after the ontogenetic torsion has taken place in torted individuals.
(3) We propose that, in M. cornuarietis, streptoneury is determined not concurrently with but rather subsequent to -or even independent of -the rotation of the visceropallium; potentially by

Test animals.
We used the caenogastropod species Marisa cornuarietis (Linnaeus 1758) [Ampullariidae], because previous research has enabled us to block the process of ontogenetic torsion by high concentrations of PtCl2 using the protocols of Osterauer et al. 17 and Marschner et al. 18 . The rearing conditions of the lab stock culture were the same as in their work (17,18,20). weeks in ScaleB4 32 , which was also used as mounting medium during microscopy. The overview image shown in Fig. 2B (left) was generated using maximum intensity projection.
The structure of the stained areas was retraced in Amira 5.2.1 to generate a 3D model of the nervous system, shown in Figure 2B (right).
Blocking ontogenetic torsion. Egg clutches of M. cornuarietis were scraped off the aquarium wall with a razorblade and extracted from the clutch with pipette tips, as in Osterauer et al. 17,34 .
Treatment with PtCl2 was executed according to Osterauer et al. 17 . In that same paper as well as in the present work, glass Petri dishes and filtered tap water from the aquaria for raising the lab stock culture were used and the medium was exchanged daily; however, different concentrations were necessary in the present study. PtCl2 concentrations (400 μg/l) were higher than in Osterauer et al. 17 , as it was the aim to obtain non-torted individuals as reliably as possible. The individual used for serial sectioning was approximately 1 mm long after 14 days post-fertilization (dpf). Note that PtCl2, apart from having the desired effect of interfering with torsion also decelerates growth 35 .
Histological fixation and embedding. After 14 dpf, the individual used in this experiment was near the maximal size that would still allow 3D reconstruction using the ultramicrotome approach at our disposal. to dissolve the internal shell the snail was known to produce after PtCl2 treatment (18). The ascending ethanol series was then completed with two steps of 10 minutes in 90% and 100%, before the sample was passed through increasing concentrations of Spurr's embedding resin in acetone (3:1, 1:1, 1:3, 100% Spurr). All steps were performed at room temperature, unless stated otherwise. To centre the probe in the resin block, a thin 'ground-layer' of resin was prepolymerised (4.5 hours at 70 °C) in the embedding moulds. The sample was then oriented in the moulds to produce transversal slices from anterior to posterior in fresh pure resin (formulated for hard blocks) and was then polymerized at 70 °C for 10 hr.

Series sectioning and digitalization.
Serial sections of 600 nm thickness were cut with a Jumbo-Histo-diamond knife on a Leica Ultracut Microtome, collected on glass slides and then stained with Toluidine blue for 30 s at 60 °C on a hot plate.
The image series was generated at a Zeiss Axioplan microscope, equipped with a Nikon D7100 camera, using a 10x Plan-Neofluar objective and Helicon Remote 3.6.2.w software. After BWconversion the images were aligned in FIJI 33 using TrakEM2 37 first by a rigid alignment followed by a second alignment using the 'Elastic Stack Alignment' plugin 38 . The stack was exported to Visualization of the final meshes for the image plates was done in Amira 6.0.