Possible roles of Wnt in the shell growth of the pond snail Lymnaea stagnalis

Abstract

Although the mechanisms of molluscan shell growth have been studied using mathematical models, little is known about the molecular basis underpinning shell morphogenesis. Here, we performed Wnt activation experiments to elucidate the potential roles of Wnt signaling in the shell growth of Lymnaea stagnalis. In general, we observed following three types of shell malformations in both dose- and developmental stage-dependent manners: (i) cap-shaped shell, (ii) cap-shaped shell with hydropic soft tissues, and (iii) compressed shell with a smaller number of coiling. We analyzed the morphologies of these malformed shells using the growing tube model, revealing that the compressed malformations show significantly larger values for T (torsion), with no significant changes in the values for the remaining parameters E (expansion) and C (curvature). We also found that cap-shaped malformations have significantly larger values for E, suggesting that the effects of BIO on shell formation may change during growth. Since the changes in T and/or E parameter values can greatly alter the shell morphologies from a planispiral or a cap-shaped one to various three-dimensional helices, changes in shell developmental processes possibly controlled by Wnt signaling may account for at least a part of the evolution of diverse shell forms in molluscs.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-74794-7.

Introduction

The Phylum Mollusca is spectacularly diverse; having acquired hard skeletons during the Cambrian, they have left behind a rich and continuous fossil record¹. Over time, molluscs have evolved various shell morphologies based on the same principle of spiral accretionary growth, and Raup’s trailblazing logarithmic coiling model^2,3 has since inspired many mathematical models that aim to describe shell growth^4–11. Among them, the growing tube model¹² (Fig. 1) maps relative growth onto a local coordinate system standardized by aperture size, thereby reasonably encapsulating the actual process, and is considered a good bridge between theory and reality. Since the growing tube model incrementally traces three parameters, i.e., expansion rate (hereafter “expansion”: E), standardized curvature (“curvature”: C), and standardized torsion (“torsion”: T) of the aperture, their biological correlates would help illuminate the mechanisms of shell growth. These correlates remained largely unknown until Dpp, or Decapentaplegic, was discovered to be expressed only in the right-hand side of the shell gland in the dextral strain of the pond snail Lymnaea stagnalis¹³. Dpp, a protein, acts as a morphogen that confers positional identity to cells in Drosophila via a concentration gradient¹⁴, and its homologs BMP2 and BMP4 participate in bone formation in vertebrates¹⁵. Shell malformations with cone-like elongation rather than the standard coiled form were observed when Dpp signaling was inhibited at the trochophore and veliger stages when shell formation starts¹⁶. Comparisons of dpp expression patterns between the right and left sides of the mantle in the dextral and sinistral strains of L. stagnalis and in a limpet species revealed that higher dpp expression levels correspond to a particular direction of shell coiling¹⁷. These results demonstrated that Dpp controls shell coiling, and, by extension, the curvature (C) parameter of the growing tube model¹⁸. However, the molecular factors underlying changes to the other parameters, namely expansion (E) and torsion (T), are yet to be explored.

Fig. 1.

The growing tube model of molluscan shell growth. The growing tube model was developed by Okamoto¹² as a model to analyze the theoretical morphologies of molluscan shells and the processes of shell growth. In this model, shell growth in a 3-dimensional space is described based on differential geometry, especially the Frenet-Serret formulas (e.g., Kreyszig¹⁹) (see Okamoto¹² and Noshita²⁰ for details). The model considers the growth of shells as the growth of a tube in the form of serial accretion of a generating curve, the last one of which being the aperture of the shell. The shell growth during a small growth period (ε) is determined using three parameters: expansion rate (E), standardized curvature (C) and standardized torsion (T). In the study, the values of three parameters are assumed to stay constant during the whole growth, which results in isometrically coiled shells. Each parameter is approximately described in this figure which is accurately defined when , where r_s and r_s+ε are the radii of the generating curve at growth stage s and s + ε, respectively, θ_{s, s+ε} is the angle between the two generating curves at growth stage s and s + ε, and φ_{s, s+ε} is the angle of rotation of the generating curves between the growth stages s and s + ε using the maximum growth point (MGP) as the reference. Note that E is redefined as in Noshita²⁰, corresponding to the logarithm of the original E in Okamoto¹². Q_s : center of the generating curve at growth stage s, Q_s+ε : center of the generating curve after ε microperiod, r_s : radius of the generating curve at growth stage s, r_s+ε : radius after ε microperiod,εr_s : length moved after a microperiod ε, MGP_s : the reference point with the highest growth rate at growth stage s, MGP_s+ε : reference point homologous to MGP_s after ε microperiod, θ_{s, s+ε} (radians) : Angle at which the generating curve has moved from its original position with respect to O_s after epsilon time, φ_{s, s+ε} : angle of rotation from MGP_s to MGP_s+ε.

It is known that during bilaterian body axis formation, Dpp regulates the dorsoventral axis formation, while Wnt regulates the anteroposterior axis formation^21,22. Wnt, like Dpp, is a signal transduction factor that controls cell fate specification, proliferation, polarity, and movements during animal development^23–26. Niehrs²⁷ argues for the existence of a Cartesian coordinate system that orchestrates body axis patterning and contributes to both the invariance and diversity of body forms, which is maintained by perpendicular Wnt and BMP gradients and is conserved across bilaterians. From this, we first hypothesize that a coordinate system exists for the theoretical developmental space, i.e., Dpp and Wnt correspond to independent axes within the developmental space. Then, as several studies on the relationship between Wnt and shell formation are known^28–34, from there, we examined the relationships between Wnt and shell growth processes using chemicals that modulate the Wnt signaling pathway. Although the presence of a coordinate system remains obscure, we found distinct malformations in the shells of trochophores and veligers that were treated with the Wnt activator BIO. Here we detail the morphological features of those malformations and report the results of parameter fitting analyses performed on these malformed shells using the growing tube model based on Noshita’s method²⁰. We argue that Wnt is likely involved in controlling shell growth along the torsion (T) and expansion (E) parameters of the growing tube model.

Results

Wnt expression in mantle transcriptome

We searched for Wnt homologs in the transcriptome data obtained from the mantle tissues of three dextral L. stagnalis individuals³⁵, and identified a total of 11 paralogs (Wnt1, Wnt2, Wnt4, Wnt5, Wnt6, Wnt7, Wnt9, Wnt10, Wnt11, Wnt16 and WntA), which represent all the Wnt subfamilies except Wnt3 and Wnt8 (Supplementary Figs. S1, S2, S3, Tables S1, S2 and S3).

Wnt activation using BIO

Embryos at the trochophore and veliger stages were used for chemical treatments. We performed Wnt inhibition experiments using IWR-1, but observed no malformations (Table S6). Then, we performed Wnt activation experiments using BIO (6-Bromoindirubin-3’-oxime), and observed numerous malformations, detailed below. Table 1; Fig. 2 show the frequencies of normal and malformed individuals that survived 8–10 days after the outset of the treatments. We classified the observed forms into three categories: normal (hatched and before-hatching individuals with normal morphology), malformed (Cap, Hyd, Com, see below for details), abnormal (individuals with malformed soft bodies due to developmental abnormalities, e.g., individuals that stopped growing at the trochophore stage, or are unusually small in size). The malformed individuals were distinguished into three types based on their shell shape and main body condition (Fig. 3): Cap-shaped malformations (Cap), i.e. those with cap-shaped shells, as seen in limpets, and normal main bodies (Fig. 3b, e, h); Hydropic malformations (Hyd), i.e. hydropic bodies with a swollen foot or mantle with a cap-shaped shell (Fig. 3f, i); and Compressed malformations (Com), i.e. those with tightly coiled shells with fewer rotations than usual (Fig. 3c, j).

Table 1.

Phenotypes of the embryos treated with BIO.

Treatment stage	Concentration	Number	Normal　(hatched)	Normal　(hatched) (%)	Normal　(before hatching)	Normal　(before hatching) (%)	Malformed　(Cap)	Malformed　(Cap) (%)	Malformed　(Hyd)
Trochophore	0 µM	314	310	98.7	2	0.6	0	0.0	0
	0.25 µM	58	45	77.6	11	19.0	0	0.0	0
	0.5 µM	173	0	0.0	17	9.8	16	9.2	47
	1 µM	201	0	0.0	18	9.0	7	3.5	33
	2 µM	139	0	0.0	4	2.9	0	0.0	17
	3 µM	54	0	0.0	0	0.0	0	0.0	6
	4 µM	18	0	0.0	0	0.0	0	0.0	0
	5 µM	73	0	0.0	0	0.0	0	0.0	0
Veliger	0 µM	384	381	99.2	1	0.3	0	0.0	0
	0.25 µM	27	27	100.0	0	0.0	0	0.0	0
	0.5 µM	119	68	57.1	48	40.3	1	0.8	0
	1 µM	239	48	20.1	145	60.7	22	9.2	2
	2 µM	137	23	16.8	61	44.5	14	10.2	7
	3 µM	131	25	19.1	26	19.8	10	7.6	8
	4 µM	36	10	27.8	6	16.7	0	0.0	0
	5 µM	65	9	13.8	8	12.3	1	1.5	0
	10 µM	73	9	12.3	20	27.4	0	0.0	0
	20 µM	9	0	0.0	0	0.0	0	0.0	0

Treatment stage	Concentration	Malformed　(Hyd) (%)	Malformed　(Com)	Malformed　(Com) (%)	Abnormal	Abnormal (%)	Dead	Dead 　(%)
Trochophore	0 µM	0.0	0	0.0	0	0.0	2	0.6
	0.25 µM	0.0	0	0.0	0	0.0	2	3.4
	0.5 µM	27.2	0	0.0	8	4.6	85	49.1
	1 µM	16.4	0	0.0	17	8.5	126	62.7
	2 µM	12.2	0	0.0	1	0.7	117	84.2
	3 µM	11.1	0	0.0	0	0.0	48	88.9
	4 µM	0.0	0	0.0	0	0.0	18	100.0
	5 µM	0.0	0	0.0	4	5.5	69	94.5
Veliger	0 µM	0.0	0	0.0	0	0.0	2	0.5
	0.25 µM	0.0	0	0.0	0	0.0	0	0.0
	0.5 µM	0.0	1	0.8	0	0.0	1	0.8
	1 µM	0.8	9	3.8	5	2.1	8	3.3
	2 µM	5.1	0	0.0	5	3.6	27	19.7
	3 µM	6.1	7	5.3	7	5.3	48	36.6
	4 µM	0.0	0	0.0	0	0.0	20	55.6
	5 µM	0.0	15	23.1	0	0.0	32	49.2
	10 µM	0.0	8	11.0	0	0.0	36	49.3
	20 µM	0.0	0	0.0	0	0.0	9	100.0

Fig. 2.

Bar graph showing the ratios of malformed and normal individuals based on the experimental results shown in Table 1. Normal (hatched): Light yellow, Normal (before hatching): Yellow, Cap: Blue, Hyd: Red, Com: Green, Dead: Light gray, Abnormal: Dark gray. The X-axis shows the concentration, and the Y-axis shows the ratios. The number above each bar indicates the number of individuals analyzed.

Fig. 3.

Three types of malformations (Cap, Hyd, Com) obtained from the Wnt signaling activation experiments using BIO and CT scanning. Cap: Cap-shaped malformation, Hyd: Hydropic malformation, Com: Compressed malformation. All individuals were photographed eight days after the outset of the chemical treatment. Experiments for e, h and i were performed at a concentration of 1 µM. Experiments for f and j were performed at a concentration of 2 µM and 5 µM, respectively. a-c Dorsal (right) and ventral (left) views of 3D images reconstructed after CT scanning for each malformation treated with chemicals from the veliger stage. d-f Individuals treated with BIO at the trochophore stage. g-j Individuals treated with BIO at the veliger stage. a, d, g Control (DMSO 0.1%). b, e, h Cap (Cap-shaped malformation). f, i Hyd (Hydropic malformation). c, j Com (Compressed malformation). f and i: lateral view; the others: dorsal view. Scale bar indicates 100 μm.

Cap has two distinguishing features: the relative sizes of the mantle and the shell, and the shell shape. In normal embryos, the mantle extends along the inner surface of the shell and rarely grows beyond the edge of the shell. In Cap embryos, however, the mantle generally protrudes from the edge, as the mantle edge is larger than that of the shell (Fig. 3h). The shell shape of Cap embryos is also entirely different from that of the control embryos, with them being about one-third the size of those of the control embryos, and most of them showing a cap-like form. Moreover, their shells are barely coiled (Fig. 3b, e, h), with most only possessing about a quarter of a rotation, or about half of a rotation in larger individuals, whereas the shells of the control embryos already reach one and a half coils. The cap-shaped shell is also characterized by an abrupt and remarkable expansion near the aperture, resulting in an unusually pointed apex.

Soft body conditions also differ significantly between Hyd and Cap embryos. Some of the soft tissues of Hyd individuals are hydropic, with most individuals possessing a swollen foot or mantle (Fig. 3f, i). Over half of the Hyd individuals had a severely swollen mantle. In addition, many Hyd embryos have smaller shells than Cap embryos of the same age. Many Hyd embryos stopped developing in the middle of the veliger stage. Some embryos lived for over a month, but neither Cap nor Hyd embryos hatched.

Com is characterized by the number of coils and the shape of the shell (Fig. 3c, j). The coils in the shells of control embryos reach about two rotations at the time of hatching (Fig. 3a, d, g), while those of the Com embryos show only one and a half rotations and are slightly smaller. The shells of Com embryos are also more tightly coiled than those of the controls and appear to be compressed along the coiling axis.

Cap morphologies were observed among individuals treated with 0.5 and 1 µM BIO at the trochophore stage, and among individuals treated with 1, 2, and 3 µM BIO at the veliger stage. Hyd morphologies were observed among individuals treated with 0.5, 1, 2, and 3 µM BIO at the trochophore stage, and among individuals treated with 1 and 3 µM BIO at the veliger stage. Com morphologies were observed among individuals treated with 5 and 10 µM BIO at the veliger stage).

A Fisher’s exact test shows that the differences between the frequencies of Cap, Hyd, or Com individuals and those of normal individuals are significant at all concentrations, except for the case of Cap individuals with 0.5µM BIO added and Hyd individuals with 1 µM BIO added during the veliger stage (Table S7).

In this treatment, the growth rate slowed as the concentration of BIO increased. While all control embryos hatched eight days after the start of the experiments at the trochophore stage, embryos treated with 1–3 µM BIO did not hatch and were smaller in size. Similarly, while all control embryos hatched eight days after the start of the experiments at the veliger stage, only a few embryos treated with 1–3 µM BIO hatched, and almost all BIO-treated embryos experienced poor growth.

Parameter fitting analysis for malformed individuals using the growing tube model

We performed CT scanning on six Com and seven control individuals obtained from the same clutch as Com, and five Cap and four control individuals obtained from the same clutch as Cap, and reconstructed their respective three-dimensional structures. We then reconstructed the internal spaces as spirally wound tubes inside the shells using the 3D images, based on Monnet et al.’s method³⁶. We further measured the thickness and central curve (we regard the trace of it as the growth trajectory of the shell) of the tubes from their 3D images (Fig. 4, S4, S5) using the “auto-skeleton” function in Avizo 3D 2021.2 (Thermo Fisher Scientific). Comparisons between the 3D images of the control and Com individuals indicate that Com shells are smaller and possess fewer coils (Fig. 4, S4). The aperture of the control shell expands radially perpendicular to the direction of shell growth, while the aperture of Com shell extends parallel to the growth axis. Comparisons of the reconstructed growth trajectories between the Com and control shells confirm that the Com shells are less coiled than the latter, resulting in a more compressed appearance. For the 13 reconstructed growth trajectories, the best fit theoretical growth trajectories and the parameter values of the growing tube model were estimated based on Noshita’s previous work²⁰. The growth trajectories estimated using the growing tube model (red curves) are generally consistent with those reconstructed from the 3D images of the tubes inside the shells (blue curves). The estimated parameter values are shown in Fig. 5a, b and c and Table S8. Wilcoxon rank sum tests indicated no significant difference between Com and control for parameters E and C (p > 0.05; Fig. 5a, b). On the other hand, the malformed Com individuals showed significantly higher T values than the controls (p = 0.0012) (Fig. 5c). Comparisons between the 3D reconstructions of the control and that of Cap individuals reveal that Cap shells are about one-third the size of the control shells (Fig. S5). Comparisons of the growth trajectories revealed that the Cap shells are barely and slightly dextrally coiled. Due to technical difficulties, out of the three parameters in the growing tube model, only the E parameter could be estimated from the 3D reconstructions of Cap shells. The parameter values for control shells were estimated using growth trajectories of the same length as Cap shells, measured from the apex. Cap shells were found to have significantly larger E values (p = 0.0159; Fig. 5d and Table S9).

Fig. 4.

Results of parameter fitting analysis for Com and control shells. 3D reconstructions (colored in yellow or green) and the inferred shell growth trajectories (white curves) for a control shell (yellow: upper panel left) and a Com shell (green: lower panel left). The middle figures in the upper and lower panels show the growth trajectory (blue) and the growth trajectory fitted using the growing tube model (red) for the control and Com shells, respectively. Shells with color gradients in the upper and lower right panels indicate shell morphology obtained by simulation based on the degree of coiling and the parameter values estimated from the theoretical growth trajectory of the control and Com shells, respectively. All images are dorsal views. Scale bar indicates 100 μm.

Fig. 5.

Box-and-whisker plots for the three parameters (expansion (E), standardized curvature (C), and standardized torsion (T)) from the parameter fitting analysis using the growing tube model and schematic representation of the growing tube model. a, b, c Box-and-whisker plots for the three parameters calculated from the growth trajectories of six Com individuals and seven control individuals assessed based on reconstructed three-dimensional representation. See Tables S8, S9 for data for each individual. Results of the Wilcoxon rank-sum test show that Com differs significantly (p = 0.0012) from control individuals in parameter T (asterisk). d Box-and-whisker plot for the E parameter calculated from the growth trajectories of five Cap individuals and four control individuals assessed based on reconstructed three-dimensional images. Results of the Wilcoxon rank-sum test show that Cap differs significantly (p = 0.0159) from control individuals in parameter E (asterisk).

RT-qPCR of the candidate readout genes of the Wnt signaling pathway in L. stagnalis

Figure 6 shows the results of our qPCR analysis of Wnt candidate readout genes in BIO-treated and control samples (Fig. 6, Table S4). BIO-treatment experiments were performed twice independently, and the resulting samples are designated here as Dataset-A and Dataset-B. Each was prepared using samples from four BIO-treated and four control individuals. For BIO-treated samples, RNA was extracted only from individuals that produced the expected shell malformation. The relative expression levels for Pangolin and Frizzled2, represented by the ΔCт values standardized by EF1ɑ expression levels, differ between Dataset-A and Dataset-B to some extent, but within each dataset, the gene expression levels of either Pangolin of Frizzled2 do not differ between the BIO-treated and the control samples (Fig. 6, Table S4).

Fig. 6.

Results of RT-qPCR on the Wnt candidate readout genes of L. stagnalis. Black bars indicate standard errors, where applicable. Wilcoxon rank-sum tests revealed no significant differences in the gene expression levels (shown in ΔCт values) between the control and BIO-treated samples in all the datasets analyzed. p : p-value, n.s.: not significant.

Discussion

BIO (6-Bromoindirubin-3’-oxime), the Wnt activator used here and in a previous study³⁷, acts on the canonical Wnt signaling pathway which depends on β-catenin in the cell³⁸. In the absence of Wnt, β-catenin is degraded by a protein complex called the destruction complex³⁹. Wnt signaling activates target genes by degrading the destruction complex and causing β-catenin to accumulate. BIO activates Wnt signaling by inhibiting GSK-3β, a component of the destruction complex. Thus, the Wnt paralogs of the canonical pathway, namely, Wnt1, Wnt2, and Wnt10, likely caused the malformations identified as Wnt ligand on the mantle transcriptome data in this study(Fig. S3). IWR-1, an inhibitor of Wnt, acts also on the canonical Wnt signaling pathway, but no phenotypic effects or lethality were observed at all concentrations tested in this study. IWR-1 inhibits Wnt signaling by stabilizing Axin proteins, also components of the destruction complex⁴⁰. One possibility is that IWR-1 has low penetration abilities and/or affinities for target regions in non-mammals. For example, in Xenopus and Danio rerio, Wnt inhibitors, including IWR-1, did not work, failing to reproduce the phenotypes observed in Wnt gene mutants⁴¹ (and Tetsuhiro Kudoh, personal communication, 2023).

Our qPCR analysis indicated unchanged Pangolin and Frizzled2 expression levels following BIO treatment. In Drosophila, it is known that Pangolin (DTcf) is upregulated (e.g., Brunner et al.⁴²), while Frizzled2 is downregulated (e.g., Cadigan et al.⁴³), when the canonical Wnt signaling pathway is activated. However, there is no guarantee that those two genes are used as Wnt readout genes in molluscs because no real molluscan “model organisms” exist, and the Wnt signaling pathway in molluscs awaits characterization. Since BIO was used as a Wnt activating drug in cuttlefish, having demonstrated results that do not contradict BIO’s being a Wnt activating drug, we interpret our qPCR analysis results as strongly suggesting that Pangolin and Frizzled2 do not represent Wnt readout genes in L. stagnalis. Further studies are needed to understand the components of, and their functions in, the molluscan Wnt signaling pathway, including downstream readout genes.

It is worth considering whether the malformations we observed were formed by the direct and specific effects of potential Wnt signal modulation within the mantle, or by other non-specific effects, such as delayed development caused by chemical treatments. Shimizu et al.¹⁶ treated L. stagnalis embryos with rapamycin solution, which inhibits cell growth and proliferation⁴⁴, and found that while the soft bodies showed abnormalities due to delayed development, the shells grew normally. Therefore, abnormally delayed development does not always lead to shell malformation. Com malformations may also attest to the separation between abnormal body development and shell formation. Com individuals hatched at almost the same time as the controls, indicating no difference in body morphologies except for the shell, which is interpreted to have changed because of Wnt activation. Since Wnt is known to be involved in antero-posterior axis formation, one may argue that the malformations merely reflect disturbances to body axis formation, as in the non-coiled shell phenotypes observed in Biomphalaria glabrata through nodal inhibition⁴⁴. However, unlike in the study on nodal inhibition, in which embryos were treated with the inhibitor at the one-cell stage, embryos with shell malformations in our study were treated with the Wnt activator at or after the trochophore stage (Table 1), when the antero-posterior axis had already been established. We therefore believe Wnt is likely involved in embryogenesis and, more specifically, in shell formation, in L. stagnalis.

The causal differences for the three types of malformations (Cap, Hyd, Com) observed (Fig. 3) may be related to differences in both BIO doses and the developmental stages when chemical treatment began. Among embryos treated with BIO at the trochophore stage, both Cap and Hyd were observed at low chemical concentrations (0.5 and 1 µM), while only Hyd were observed at higher concentrations (2 and 3 µM) (Fig. 2; Table 1). This suggests that Hyd represents a severer condition than Cap from a higher dose of BIO, which is manifest in the fact that Hyd individuals suffer from abnormality not only in their shells, as in Cap individuals, but also in their soft tissues in the form of hydropic bodies. Among embryos treated with BIO at the veliger stage, the survival rates after treatment apparently increased, and the frequencies of both Cap and Hyd decreased, compared to those treated at the trochophore stage (Fig. 2; Table 1). Moreover, Com, which were not seen among embryos treated at the trochophore stage, was observed (Fig. 2; Table 1). As discussed above, Com individuals hatched at almost the same time as the control ones, showing what is likely only a superficial abnormality in shell morphology. These observations collectively suggest that the possible role of Wnt pathways changes during development in this species from a more general one, in controlling the development of various tissues at the trochophore stage, to a more specific one, mainly controlling shell growth.

Both Cap and Hyd are characterized by a cap-shaped shell (Fig. 3). This phenotype somewhat resembles that observed by Shimizu et al.¹⁶ in Lymnaea embryos treated with a Dpp inhibitor in that both show no coiling. However, several notable differences exist. First, the form seen by Shimizu et al.¹⁶ is stretched out like a cone, while the shells observed in this study are not, and are instead slightly spirally coiled (Fig. 3h). Second, the malformed shells in Shimizu et al.’s study¹⁶ expanded their aperture as the control shells did during growth, then grew straight without coiling, unlike the controls. On the other hand, the shell aperture of the Cap and Hyd types in this study gradually became bent as the shell grew (Fig. 3e, h). These observations are consistent with the premise that the role of Dpp in shell growth is related to the control of the parameter C in the growing tube model, while the abnormality in Cap and Hyd shells is related to parameter E. Also, our results differ from Shimizu et al.’s¹⁶ in that we observed higher percentages of malformations. For example, in Shimizu et al.¹⁶, in the experiments where malformations occur at the highest frequency, the percentage of malformations was about 0.7% (for embryos treated at veliger stage with 1 µM). Meanwhile, the corresponding value in this study is about 10% for Cap (for embryos treated at veliger stage with 2 µM), 27% for Hyd (for embryos treated at trochophore stage with 0.5 µM), and 23% for Com (for embryos treated at veliger stage with 5 µM) (Table 1; Fig. 2). Thus, the malformations observed by Wnt activation differ from those obtained by Dpp inhibition. Other studies on the molecular bases of shell growth include Baynes et al.⁴⁵, who observed a high percentage of banana-shaped, open-coiled shell malformation in the pond snail Biomphalaria glabrata after exposing them to a 5α-reductase inhibitor. Notably, shell malformations occurred at a very high percentage, reaching 89% in one experimental batch, and the mortality rates were about the same as the controls. These malformations are open-coiled, thus likely have significantly altered parameter C values. Since those malformed shells have not been subjected to quantitative morphological analyses, such as the parameter fitting analysis employed in this study, it remains to be clarified how 5α-reductase is involved and is related to such factors as Dpp and Wnt, in shell growth.

Our analyses based on a previous study²⁰ indicated that Com and normal individuals differ only in their T parameter values, with it being significantly larger in malformed individuals (Figs. 4 and 5c). Since these malformations were caused by Wnt activation, we believe Wnt is likely involved in the regulation of the T parameter in L. stagnalis shell growth, though its exact control mechanisms over shell twisting remain opaque. Shimizu et al.¹⁶ argued that since Dpp is involved in cell proliferation, the asymmetric expression of Dpp on the left and right sides of the mantle may induce different rates of cell proliferation along the aperture, resulting in an asymmetric shell growth pattern, thereby regulating the C parameter. However, the effect of Dpp on cell proliferation cannot be directly linked to changes in the T parameter. Likewise, since Wnt is also a signaling factor involved in cell proliferation, it is difficult to relate its role directly to the control of the T parameter; a mediating hypothesis would be necessary to bridge the gap and explain its actions. Regarding the mechanisms for the shell twisting, at least two hypotheses can be considered here: one is Chirat et al.’s⁴⁶ “mechanical model” that relates mechanical stress on the mantle to shell morphogenesis, and the other may be called the “directional cell division” hypothesis, which posits that the cell proliferation at the mantle edges occurs directionally, like in the spiral cleavage of the 4-cell stage blastomeres. An example of directional cell proliferation in gastropod mantle edges was recently reported by Johnson et al.⁴⁷, demonstrating that the cells on the dorsal edge of the mantle are at an oblique angle relative to the coiling axis. Theoretical links between the growing tube model and the aperture maps indicate that only the C parameter is related to the gradient of shell growth in the aperture map¹⁸, and the T parameter may be regulated by a different mechanism from the one underlying parameter C. Further studies on Wnt in this species are warranted to identify the paralog(s) involved in shell formation and to reveal the spatial distributions and the roles of the relevant Wnt paralog(s) in the mantle tissues.

In the growing tube model, the shell becomes more twisted when the parameter T is large. When the parameter T is small, the shell is shifted from a helical shell to a flat-coiled shell (Fig. 7). In other words, T indicates the degree to which the shell deviates from being flat-coiled (e.g., as Okamoto¹² stated, if T = 0 the tube is planispiral). Our parameter fitting analysis indicated that Com individuals formed by Wnt activation have a greater T value, and we expect the inhibition of Wnt signaling to change the shell to a flat-coiled-like morphology. Therefore, changes in Wnt signaling could have played a role in the morphological evolution from a three-dimensionally-coiled shell to a flat-coiled one, or vice versa. Members of Hygrophila, the pulmonated superorder that includes L. stagnalis, possess shells of various shapes, including helical, flat-coiled, and cap-shaped, and phylogenetic relationships suggest that each shell morphology has evolved independently many times. Wnt is very unlikely to be solely responsible for the evolution of the flat-coiled morphology, but changes to its expression levels or patterns during specific developmental stages in ancestral species likely contributed to the diverse morphology of Hygrophila. Future developmental studies on the causal relationships between Wnt and shell growth in Hygrophila, as well as in other gastropods, would help elucidate the mechanisms underlying morphological evolution in gastropods.

Fig. 7.

Theoretical diversity of the shells produced by the growing tube model represented by variations in the three parameters: expansion (E), standardized curvature (C), and standardized torsion (T). Parameter values for the theoretical morphology shown in the center are E = 0.096, C = 1.506, and T = 0.727.

Parameter fitting analyses indicated a significant difference in T parameter values between Com and control shells. On the other hand, Cap shells had a higher E parameter value than the control shells (Fig. 7). Given that the morphologies of Cap (and Hyd) and Com shells differ from those of the control shells in different ways, and that Com shells were observed in the experiments beginning at the veliger stage, but not in those at the trochophore stage, we may infer that the functions of Wnt change across the trochophore and veliger stages as discussed above. Simulations using the growing tube model help highlight the morphological differences between Cap and Com shells (Supplementary Fig. S6). In this simulation, we tested if the Cap shell morphology can be produced by simply adjusting the T parameter values of control shells (code available at https://github.com/noshita/growing_tube_model_estimation), and confirmed that it was impossible (Fig. S6). We also heuristically changed the parameter values, and discovered that a morphology very similar to Cap can be reproduced when E = 0.4, C = 1.0, and T = 0.3 (Fig. S7). In this case, the value of E almost coincides with the ones theoretically estimated from our 3D reconstructions of Cap shells. The simulation results suggest that variations in T values controlled by Wnt to produce Com shells are insufficient to produce Cap and Hyd shells, and additional controls on the E parameter are needed. Cap and Hyd are typically produced when embryos are treated at the trochophore stage. Since the most prominent feature of Cap and Hyd appears to be their body size, which is about a third of that of the controls, one major role of Wnt at the trochophore stage could be controlling the growth rate of the whole body. If so, changes in growth rates could explain the morphological differences between Cap (or Hyd) and control as argued by Urdy et al.^8,9 and Rice⁶.

In this study, we assumed isometric growth, or the invariant nature of each of the growth parameters during growth. In reality, shell growth can be allometric, but Noshita’s theoretical estimations of growth trajectories, which assume isometric growth, explains the actual growth trajectories of shells well (Fig. 4, S4). However, since the growth trajectories of shell regions that were formed very early on could not be reconstructed by the method employed in this study, we cannot rule out possible changes in the mode of shell growth in this species. If allometric growth or changes to the mode of growth is indeed confirmed, then we may need to use a model that accommodates such changes, as exercised by Urdy et al.^8,9.

In this context, Rice’s argument⁶ based on a theoretical model involving an “aperture map”, which represents another type of model developed for gastropod shell forms¹⁸, may be important. This model shows that it is developmentally easy to derive a slightly coiled limpet shell from a high-spired ancestor, but difficult to produce a fully conical limpet, a fact interpreted to have arisen from the decoupling of shell production rates from the growth rate of the animal within the shell⁶. Noshita et al.¹⁸ further argued, by combining the aperture map (or growth vector) model and the growing tube model, that an asymmetric aperture map, or a gradient of shell growth rate at different positions along the aperture, can be produced by only the C parameter value of the growing tube model, and that at a given C parameter value, if the value of E² + T² remains constant, then a slightly-coiled cap-shaped shell can be produced from a spirally coiled shell simply by decreasing the value of T and increasing the value of E without changing the profile of the aperture map¹⁸. To produce a fully conical limpet shell, meanwhile, the value of C needs to be changed. Those arguments appear to mirror our findings that Com and Cap were produced simply by changing the concentrations of the Wnt activator, and the developmental stages at which our experiments begin. Since the T value for Cap could not be assessed in our study, we cannot see if E² + T² remained the same between Com and Cap. Future improvements to Noshita’s method²⁰ will allow us to analyze the morphology of cap-shaped shells by using not only growth trajectories but also surface structures, such as growth lines, as reference²⁰.

Since multiple factors, apart from Wnt, Dpp, and 5α-reductase, are likely involved in shell growth, it is necessary to verify the roles of these proteins in shell growth through further quantitative comparisons. This study paved the way for the quantitative analysis of the effects of proteins on shell growth to understand the still enigmatic mechanisms of molluscan shell growth and evolution. Changes in the level and/or timing of Wnt expression could have altered the orientation of shell growth, which may have played a vital role in the evolution of diverse shell morphologies in Hygrophila and possibly in many other groups of gastropods.

Materials and methods

Animals

In this study, a dextral strain of the pond snail L. stagnalis was used. Adult individuals are reared in tap water in the laboratory at around 20 °C. They lay eggs in capsules which are coated with jelly throughout the year. Embryos hatch after 10 days and start to reproduce after around 3 months at 25 °C. Eggs were collected and separated from the jelly by rolling the capsules on a sheet of paper (COMFORT service towel, NIPPON PAPER CRECIA) by tweezers and incubated at 23–25 °C in sterilized tap water using six-well dishes (BD Bioscience, Franklin Lakes).

Identification and expression analysis of Wnt paralogs

The sequences of Wnt paralogs in L. stagnalis and their expression levels in the mantle tissues were assessed using the transcriptome data of Ishikawa et al.³⁵ (for details in this section, see Supplementary Information). In the identification of the Wnt homologs in L. stagnalis, reciprocal blast searches against the genome and transcriptome data of three molluscan species (Lottia gigantea, Crassostrea gigas, and Pinctada fucata), and Pfam domain searches were performed. The paralog identities were confirmed by phylogenetic analysis using the maximum likelihood using MEGA7.0 after alignment with clustalW for the Wnt paralogs identified in L. stagnalis and the sequences of seven other animal species (Lottia gigantea, Crassostrea gigas, Pinctada fucata, Euprymna scolopes, Capitella teleta, Drosophila melanogaster, Homo sapiens) obtained from GenBank.

Drug treatment

To investigate the function of Wnt signaling in shell formation in this species, we exposed the embryos to the chemical inhibitor IWR-1 (inhibitor of Wnt response 1, (I0161, Sigma-Aldrich)) and activator BIO (6-bro- mumoindirubin-3’-oxime, (B1686, Sigma-Aldrich)). Embryos used in this study were reared at 18 °C. The trochophore stage embryos at 120–144 h after one-cell stage and the veliger stage embryos at 168–240 h after one-cell stage were used in the drug treatments. In the inhibition experiments, embryos were incubated in sterilized tap water containing 0.1% (v/v) DMSO and IWR-1 at concentrations of 0, 10, 50, or 100 µM. In the activation experiments using BIO, embryos were incubated in sterilized tap water containing 0.1% (v/v) DMSO and BIO at concentrations of 0, 0.25, 0.5, 1, 2, 3, 4, 5, 10, or 20 µM. They were kept in the solutions in the dark at 23–25 °C for 8–10 days, which is the amount of time control embryos take to hatch.

Three-dimensional reconstruction of L. stagnalis shell morphologies

CT scans of BIO-treated malformed embryos and control embryos were used for the three-dimensional reconstructions of shell morphologies. To obtain morphological data, specimens were scanned with the micro-XCT instrument (Scan Xmate-DF160TSS105, ScanXmate-A-CF160TSS650, Comscantecno Co.) installed in JAMSTEC (Yokosuka, Japan), using an X-ray tube voltage of 80–90 kV and a current of 33–105 µA. The slice thickness of the embryos was 1.3–3.6 μm. To reconstruct the three-dimensional structure, segmentation was performed using the software Avizo 3D 2021.2 (Thermo Fisher Scientific). Based on the methodology used by Monnet et al.³⁶, we measured tube thickness and the locus of the centroid of each tube slice using the “auto-skeleton” function in Avizo to acquire the parameter values for the growing tube model from the image data. The locus of the centroid of each tube slice will hereafter be regarded as the growth trajectory of the shell. The trajectories branched toward the edges of the aperture, obvious technical artifacts which may arise at every tube slice, were excluded manually. Since we could not estimate the growth trajectory of the shell up to the vicinity of the aperture based on the internal volume of the interior shell data alone, we obtained the growth trajectory for the entire shell interior by superimposing multiple slices of approximately the same shape as the aperture on the data of the aperture interior (see Supplementary Information Fig. S8 for details). The obtained shell growth trajectory data were calculated as parameters for the growing tube model based on Noshita’s work²⁰. The code for estimating the parameters of the growing tube model has been made available on GitHub (https://github.com/noshita/growing_tube_model_estimation).

RT-qPCR of the candidate readout genes of the Wnt signaling pathway in L. stagnalis

RT-qPCR was performed to analyze how the expression levels of potential Wnt readout genes are affected by the treatment of the embryos of L. stagnalis with BIO. Since no known Wnt readout gene exists in molluscs, we searched for the readout genes of the Wnt signaling pathway of Drosophila summarized in “the Wnt homepage” (https://web.stanford.edu/group/nusselab/cgi-bin/wnt/). After searches for the homologs of the potential readout genes in molluscs and in L. stagnalis in particular, using the mantle transcriptome data of L. stagnalis (Ishikawa et al.³⁵), two genes, namely, Pangolin and Frizzled2, remained as candidates as Wnt readout genes in L. stagnalis. Pangolin was reported as readout gene in Drosophila by Brunner et al.⁴², and Frizzled2 by Cadigan et al.⁴³.

Total RNA was extracted from four individuals of L. stagnalis following the protocol of Isowa et al.⁴⁸, for each of the BIO-treated (and with the phenotype of shell malformation) and the control samples. The experiments of BIO treatment and subsequent RNA extractions were performed twice independently to serve biological duplicates, which are named Dataset-A and Dataset-B (see Table S4 for details). Complementary DNA was prepared, and quantitative PCR was performed using the StepOne Real-time PCR system (Applied Biosystems, Foster City, USA), based on the protocol of Ishikawa et al.³⁵. Two sets of primers (20–22 nucleotides in length; Table S5) were designed for each of Pangolin and Frizzled2 genes, using Primer3Plus (https://primer3plus.com/cgi-bin/dev/primer3plus.cgi; last accessed May 3, 2024). A pair of primers designed by Young et al.⁴⁹ for L. stagnalis was used to amplify an EF1ɑ gene sequence as the endogenous control. For each combination of the templates (control and BIO-treated of Dataset-A and Dataset-B) and the primer sets (Lst Pangolin-1-F and Lst Pangolin-1-R (Pangolin-1), Lst Pangolin-2-F and Lst Pangolin-2-R (Pangolin-2), Lst Frizzled2-1-F and Lst Frizzled2-1-R (Frizzled2-1), Lst Frizzled2-2-F and Lst Frizzled2-2-R (Frizzled2-2), and the primer set for EF1ɑ) (Table S5), three to four technical replicates were included. The qPCR consisted of 95 °C for 30 s; 40 cycles of 95 °C for 15 s, 56 °C for 30 s, and the gene expression levels were quantified using the comparative CT method (Livak and Schmittgen⁵⁰).

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

S.O., K.S., H.S. and K.E. conceived and designed the experiments. S.O., K.N., K.K. and A.I. performed the experiments. S.O. and K.E. contributed material for the study. S.O., K.N., K.K. and K.E. analyzed data. S.O., H.S. and K.E. wrote the paper.

Data availability

Transcriptome data from this study are available in the DDBJ Sequence Read Archive (DRA) under accession numbers DRA005517 and DRA006373. All other datasets used in this study are available from the corresponding authors on request. The 3D reconstructions of the different types of shell morphologies (Control, Cap, Hyd, and Com) are made publicly available on MorphoSource and are accessible via (https://www.morphosource.org/projects/000636914).

Declarations

Competing interests

The authors declare no competing interests.