Reproductively active adult specimens of H. asinina were collected from Heron Island Reef, Great Barrier Reef, Australia (23°27′S; 151°55′E). Detailed protocols for maintenance of collected specimens, spawning, fertilizations, and larval culture were followed in accordance with Williams et al. (2009a). In brief, eggs and sperm from at least three males and three females were collected and fertilized on the night of the natural spawning cycle (Counihan et al., 2001). Embryos and larvae were cultured in 300 mm diameter larval culture chambers with flow-through 10 μm filtered sea water (FSW) at 25°C ± 0.5 until the larvae matured to the competent state (96 hpf) (Jackson et al., 2005). The culture chambers were treated with 20 μg/L rifampicin for 2 h daily to prevent bacterial outbreaks.

Because variable metamorphic responses to specific NOS inhibitors have been reported in C. fornicata (Pechenik et al., 2007) and Capitella teleta (Annelida) (Biggers et al., 2011), we elected to test several different NOS inhibitors and NO donors to overcome a potential incompatibility between the pharmacological agents and H. asinina veligers (Table 1). As NOS inhibitors, we used L-nitroarginine-methyl-ester (L-NAME) (Sigma), aminoguanidine hemisulfate (AGH) (Sapphire Bioscience), and S-methylisothiourea sulfate (SMIS) (Sapphire Bioscience). To examine the effects of increased NO on metamorphosis, we used S-nitroso-N-acetyl-penicillamine (SNAP) (Sapphire Bioscience) as a direct NO donor. L-Arginine (Sigma), which is the essential substrate of NOS enzymatic activity, was also applied as an NO donor, since it theoretically increases the internal availability of NO (Stuehr, 2004). All chemicals used here have been shown to affect the rate of metamorphosis in other marine invertebrates (Froggett and Leise, 1999; Bishop and Brandhorst, 2001; Bishop et al., 2001, 2008; Comes et al., 2007; Pechenik et al., 2007; Biggers et al., 2011; Zhang et al., 2012; Romero et al., 2013).

Stock solutions of 0.5 M L-NAME, 0.5 M AGH, 0.5 M SMIS, and 0.5 M L-Arginine were prepared in 0.22 μm FSW, stored at 4°C, and diluted to final experimental concentrations just prior to the experiments. For SNAP, a stock solution of 0.1 M was prepared in dimethyl sulfoxide (DMSO) immediately before the experiment and diluted to final concentrations just prior to use. To maintain a steady concentration of NO delivered by SNAP during the metamorphosis assay, the solution was renewed every 6 h (Froggett and Leise, 1999). The final concentrations of each chemical used in the experiments are listed in Table 1.

Metamorphosis assays were initiated at competency (96 hpf) (Figure 1) and performed in 6-well 35-mm diameter sterile polycarbonate tissue culture dishes with 10 ml of 0.22 μm FSW per well. The following controls and treatments were included in experiments with each pharmacological agent: (1) FSW only (negative control), (2) FSW with living articulated red coralline algae (CA), Amphiroa ephedraea shards covering approximately 25% of the bottom of the well (positive control), and (3) FSW containing a pharmacological agent either with or without live CA shards (treatments). The live CA used for the experiment were collected from Heron Island reef flat just prior to experimentation, washed in FSW, cleaned of any epiphytes growing on the surface and placed in the wells for immediate use (Williams et al., 2008). Initiation of juvenile shell growth at 24 hpi was used as a definitive indicator of metamorphosis (Williams et al., 2008).

To investigate temporal expressional patterns of NOS and HSP90 genes in H. asinina, total ribonucleic acid (RNA) samples were collected during embryonic, larval, and post-larval development (Figure 1). Embryonic and larval samples were collected from the culture chambers described above. To collect the post-larval samples of H. asinina, competent veligers (96 hpf) were exposed to live CA in 12 cm petri dishes with live CA shards. We only collected those individuals crawling on live CA to ensure a clear discrimination between still-swimming larvae and post-larvae that had initiated settlement and metamorphosis.

For all RNA samples, ~1000 individuals were collected for each sampling point, and were preserved in TRI reagent (Sigma) at −80°C prior to extraction. RNA extraction and complementary DNA (cDNA) synthesis followed methods described in Williams et al. (2009a). For the assessment of genomic DNA contamination, no-RT control samples were prepared from the 0.5 μg DNase-treated RNA for the all RNA samples and tested by quantitative RT-PCR (qRT-PCR). All synthesized cDNA and no-RT control samples were stored at −20°C.

We isolated a single NOS gene from H. asinina (HasNOS) by degenerate PCR. The degenerate primers were designed from the alignment of NOS derived amino acid sequence from the following species: Lehmannia valentiana (BAF73722), Aplysia californica 1 and 2 (AAK83069 and AAK92211, respectively), Branchiostoma floridae (AAQ02989), Nematostella vectensis (XP001631503), Sepia officinalis A and B (AAS93626 and AAS93627, respectively), and Lottia gigantia (JGI223312) (Figure 2). The degenerate forward (DegF1) and reverse (DegR1) primers were designed from conserved amino acid sequences of CNHIKY and CPADWVW, respectively (Figure 2). The latter amino acid sequence contains a part of the tetrahydrobiopterin (BH₄) binding site, which is a highly conserved region of metazoan NOS (Andreakis et al., 2011). The DegF1 and DegR1 sequences were 5′-TGYAAYCAYATHARTAY-3′ and 5′-CCANACCCARTCNSCNGGRCA-3′, respectively. A touch-down PCR profile was used: 94°C for 2 min, 5 cycles at 94°C for 30 s, 55 to 49°C (2°C increment for every 5 cycles) for 30 s, and 70°C for 3 min, 25 cycles at 94°C for 30 s, 47°C for 30 s, and 70°C for 3 min, with a final extension of 72°C for 10 min. Each PCR reaction comprised 1x reaction buffer (Promega), 0.2 mM dNTP, 1 U Taq polymerase (New England Biolab), and 1 μM of each primer in a total volume of 20 μL.

The degenerate PCR products were separated and visualized by 1.5 % agarose TAE gel electrophoresis. Products of the expected size (approximately 500 bp) were excised and gel-purified by a silica suspension method (Boyle and Lew, 1995). The purified PCR products were then cloned using the pGEM-T Easy Vector System 1, following the manufacturer's recommendations (Promega). Successfully transformed recombinant colonies were picked and directly added to a second PCR mixture with a final concentration of 1x reaction buffer (Promega), 0.5 mM MgCl₂, 0.2 mM dNTP, 1.5 U Taq polymerase (New England Biolab), and 0.25 μM of both M13 forward and reverse primers to amplify inserts. This second PCR reaction used a profile of 94°C for 5 min, 35 cycles of 94°C for 30 s, 55°C for 1 min, and 72°C for 1 min, with a final extension of 72°C for 10 min. Products were then separated by 1.5 % agarose TAE gel electrophoresis and those of correct size were purified using the silica suspension method (Boyle and Lew, 1995) as described above. The purified PCR products were Sanger-sequenced using reactions containing 1 μL of Big Dye Terminator mix v 3.1 (Applied Biosystems), 1x Big Dye Terminator reaction buffer, 3.2 pmol of primer, and 6–10 ng in a final volume of 10 μL, as recommended by the Australian Genome Research Facility (AGRF). The sequenced products were purified by magnesium sulfate precipitation in accordance with the AGRF protocol and submitted to the AGRF Brisbane node (Queensland, Australia). The resulting sequences were compared to previously-deposited orthologs in the National Center for Biotechnology Information (NCBI) by tBLASTx queries.

We first isolated a HasNOS fragment of 210 bp, sufficient to design a primer set for quantitative real-time PCR (qRT-PCR). To extend the length of the HasNOS sequence, as necessary to produce an adequate length RNA probe for whole mount in situ hybridization analysis, we then designed a set of degenerate primers from AA sequences of AWRNAPRCIGRIQW (DegF2) and ALGSKAYP (DegR2) in aforementioned species, except Lottia gigantia (Figure 2). These amino acid sequences are located in the functional heme and flavin mononucleotide (FMN) binding sites, respectively. Andreakis et al. (2011) found both binding sites in all metazoan NOS orthologs analyzed, showing the highly conserved nature of these two functional sites. The sequences of DegF2 and DegR2 were 5′-CCNMGNTGYATHGGNMGNATHCARTG-3′ and 5′-GRTANGCNCKNSWNCCNARNCC-3′, respectively. This target region is located in the NOS oxygenase domain and encompasses the previously identified HasNOS nucleotide sequence. A touch-down PCR cycle was run as above with modified annealing temperatures from 54 to 46°C (2°C increment for every 5 cycles) and 44°C for the following cycle. The obtained PCR product was used to run a nested-PCR with the degenerate forward primer and a gene-specific reverse primer (5′-CCCGAAGACGCTGCTCGTTCTCC-3′) designed from the fragment of HasNOS nucleotide sequence isolated by the first attempt of isolation. This overlap in primer sets ensured that the two sequenced regions were contiguous. Nested PCRs used a profile of 94°C for 5 min, 35 cycles at 94°C for 30 s, 48°C for 30 s, and 72°C for 1 min, and 72°C for 10 min. Resulting products of the expected size were purified and directly submitted to AGRF for Sanger sequencing. The obtained sequence was again confirmed as a fragment of NOS gene by tBLASTx.

The full sequence of the H. asinina HSP90 gene was previously obtained and named as HasHSP90A (Gunter and Degnan, 2007) [GenBank: EF621884].

To analyze the transcriptional profiles of HasNOS and HasHSP90A genes during H. asinina development (embryonic, larval, and post-larval stages) (Figure 1), we performed quantitative real-time PCR (qRT-PCR) as described in Williams and Degnan (2009) with following modifications. About 3.75 ng cDNA template of each sample was mixed in a 15 μL reaction mix comprising cDNA, SYBR Green Master mix (Roche), and 0.17–0.34 μM primer on a Light Cycler 480 (Roche). The following gene-specific primers were designed manually to meet criteria of >45% GC content and >60°C primer melting temperature: HasNOS forward 5′-TGGGTTTGGACGTCCGGAAGAGC-3′ and reverse 5′-CCCGAAGACGCTGCTCGTTCTCC-3′ and HasHSP90A forward 5′-GTTACCTCCCCTTGCTGTATTGTCAC-3′ and reverse 5′-TTGTCAGCATCTGCCTTCTCCTTC-3′.

To normalize the level of transcription for obtaining relative gene expression values, we used the geometric mean of two reference genes - Has-nascent polypeptide- associated complex alpha polypeptide (NACA) (forward 5′-TGTCGCAAGCCAACGTTTCA-3′ and reverse 5′- GACAGCATGTTCAGCACTGGT-3′) and Has-ubiquitin (forward 5′-TGGCAAGCAGTTGGAAGATGGT-3′ and reverse 5′-CAGTTGTACTTGGAGGCCAGGAT-3′). These reference genes were previously chosen by Williams et al. (2009a) from microarray expression data by selecting for transcripts showing the least variable expression across larval and post-larval developmental stages. Their stability within our experimental sample set was confirmed using Genorm software (Vandesompele et al., 2002).

The following qRT-PCR reaction parameters were used: initial denaturation 95°C for 10 min (ramp rate 4.4°C/s), and 40–50 cycles of 95°C for 5 s (ramp rate 4.4°C/s), 58°C for 10 s (ramp rate 2.2°C/s), and 72°C for 20 s (ramp rate 4.4°C/s). Melt curve data acquisition was from 55–95°C with continuous measurement (acquisition/°C = 5). All samples were run in triplicate. The purity of PCR product was confirmed by the presence of only a single peak in the resultant temperature melt curve. For each primer pair, a standard curve was generated to calculate the efficiency of qRT-PCR using a dilution series from the calibrator sample, which was a mixture of 4 μl of all undiluted cDNA samples. In addition to the developmental stages cDNAs, a no-template (H₂O) control and the calibrator sample were included for each qRT-PCR run and for each primer pair. The efficiencies of each primer pair and the cycle threshold of each sample were calculated by the second derivative method using Roche Light Cycler 480 software program. Relative expression ratios were calculated as described in Williams and Degnan (2009).

To determine spatial localization of HasNOS and HasHSP90A during larval development, H. asinina veligers were sampled at 48, 72, and 96 hpf (Figure 1); 48 and 72 hpf are precompetent larvae, and 96 hpf are competent larvae. Fixation and storage of H. asinina larvae for WMISH were carried out as described in Williams and Degnan (2009). Dioxygenin (DIG)-labeled antisense RNA probes for HasNOS and HasHSP90A were synthesized from purified PCR products that were amplified from a mixture of embryonic and larval cDNA collected as described above, using 10× DIG labeled mix (Roche) following the manufacturer's instructions. The HasNOS probe was transcribed from a 702 bp fragment of nucleotide sequence identified above. The probe for HasHSP90A was transcribed exactly as described in Gunter and Degnan (2007). The procedure of WMISH to examine the spatial gene expression was performed as described in Williams and Degnan (2009).

Data collected from metamorphosis assays were analyzed by One-Way analysis of variance (ANOVA) with treatment as a factor. Significant differences among treatments were detected by Tukey's HSD post-hoc testing. Prior to ANOVA, all data were arcsine-transformed to improve the normal distribution of samples. Levene's test was performed to ensure homogeneity of variance among treatments. All statistical analyses were performed in R (R Foundation for Statistical Computing). An alpha value of 0.05 was used to determine a significant difference (Zar, 1984).

Pharmacological experiments to test the role of NO in regulating settlement and metamorphosis in the tropical abalone Haliotis asinina were performed using both NOS inhibitors and NO donors (Table 1). Because competent veligers of H. asinina require the presence of coralline algae (CA) to induce metamorphosis (Williams et al., 2008), pharmacological experiments were conducted both with and without live CA. This allowed us to examine whether application of a pharmacological agent alone is sufficient to induce metamorphosis or sufficient to inhibit the inductive capability of live CA. In the absence of live CA that is already known to be an effective environmental inductive cue for metamorphosis in H. asinina, none of the pharmacological treatments—neither inhibitors nor donors—were able to induce metamorphosis on their own (Figures 3, 4).

In the presence of live CA, however, the application of NOS inhibitors gave results that are strikingly discordant with any molluscan studies published so far. All of the three NOS inhibitors examined—L-NAME, AGH, and SMIS (Table 1)—suppressed the mean percentage of larval metamorphosis in the presence of live CA, and did so in a concentration-dependent manner (Figure 3). The mean percent metamorphosis was significantly less than the positive control in the 0.01 and 0.05 mM L-NAME (Figure 2A), 0.25 mM AGH (Figure 2B), and 0.1 mM SMIS (Figure 2C) treatments in the presence of the live CA. In contrast, in the presence of live CA, neither of the NO donors—we tried both SNAP and L-arginine (Table 1)—significantly affected the mean percentage of larval metamorphosis in comparison with the positive controls (Figure 4). We do not believe that our ability to detect an enhanced rate of metamorphosis in the presence of NO donors was constrained by a ceiling effect, because our positive CA controls achieved only 50–70% metamorphosis in these experiments. In the SNAP experiment, the mean percentage of larval metamorphosis in the solvent control, DMSO, also did not significantly differ from the positive control (Figure 4A).

As an initial step to analyze the expression of NOS and HSP90 in H. asinina, we used degenerate PCR to isolate the partial NOS gene sequence, which was used to design specific primers for qRT-PCR and to synthesize specific probes for in situ hybridization analyses (see below).

Degenerate PCR yielded a putative 702 bp HasNOS product, which was confirmed to be a partial NOS gene sequence by tBLASTx against NCBI databases. The tBLASTx analysis showed very high sequence similarity to other molluscan orthologs, including Stramonita haemastoma (FR667655, E-value: 1e–115), Lehmannia valentiana (AB333805, E-value: 3e–110), and Crassostrea virginica (GQ844865, E-value: 7e–110). The Conserved Domain Database (CDD) confirmed that the isolated HasNOS was part of the NOS oxygenase domain, which was the intended target region of the degenerate primers. This partial Haliotis asinina NOS sequence has been named HasNOS and is available at NCBI [GenBank: KC571824].

Temporal expression profiles of HasNOS and HasHSP90A through embryonic, larval, and post-larval development (Figure 1) were measured by qRT-PCR (Figure 5). Both genes are expressed in all developmental stages examined. HasNOS is low from the egg until 24 hpf (Figure 5A). At 48 hpf, expression increases and then is maintained relatively stable, except for a sharp decrease at 84 hpf. The highest expression level among larvae is found at 144 hpf. Expression of HasNOS in post-larvae (that have initiated metamorphosis) shows more fluctuation in comparison with larvae of the same age that have not initiated metamorphosis. During the early post-larval stages (1–8 hpi), the expression level gradually declines, but sharply increases by 12 hpi, at which time it shows the highest expression level observed among samples induced to metamorphose. The expression pattern for 24 and 48 hpi is similar to that for the corresponding larval samples of 120 and 144 hpf.

For HasHSP90A, the highest expression is seen at 10.5 hpf (Figure 5B). Expression then consistently declines through larval development until 97 hpf, followed by gradual increases from 100 to 108 hpf. After 108 hpf, the expression level decreases quite dramatically by 120 hpf, at which time it shows the lowest expression level observed among larval samples. Interestingly, the fluctuation of HasHSP90A expression pattern during post-larval development is comparable to that of HasNOS expression. The abundance of HasHSP90A transcript gradually decreases during the early post-larval stages and then increases sharply by 12 hpi, followed by fluctuations around 24–48 hpi.

Whole mount in situ hybridization (WMISH) was performed to examine spatial expression of HasNOS and HasHSP90A in precompetent (48 and 72 hpf) and competent (96 hpf) veligers (Figure 1). At 48 hpf, HasNOS is localized to the larval retractor muscle and to a tight cluster of unidentified cells in the right lateral side of the mantle tissue (Figures 6A,B). This unidentified right mantle expression is maintained in 72 and 96 hpf larvae (Figures 6C–F). Additionally at this time, HasNOS expression appears in the cells aligned along the lateral edge of the foot (Figures 6C–F). By 96 hpf, expression in the foot has expanded across the pedal sole (Figure 6F). Unlike previous studies of spatial expression in other molluscan species (Thavaradhara and Leise, 2001; Pechenik et al., 2007; Bishop et al., 2008), no HasNOS expression is detectable in the larval central nervous system (CNS). Schematic representation of HasNOS localizations is shown in Figure 6G.

HasHSP90A is clearly localized to mantle tissue in 48 hpf larvae (Figures 7A,B), but by 72 and 96 hpf, transcripts are difficult to detect above background (Figures 7C,D). These results are concordant with the transcriptional abundance indicated by the qRT-PCR assay, in that HasHSP90A continues to decline through larval development (Figure 4B). Schematic representation of HasHSP90A localizations is shown in Figure 7E.

Past studies so far have reported that NO functions as a negative (repressive) regulator of metamorphosis in several molluscs, namely I. obsoleta, C. fornicata, P. sibogae, and A. willowi (Froggett and Leise, 1999; Pechenik et al., 2007; Bishop et al., 2008; Romero et al., 2013). Here we reveal a first time contrasting situation, in which NO does not have a repressive function, but rather appears to assist in the initiation of larval metamorphosis in the abalone Haliotis asinina. In the context of the generalist-specialist larvae hypothesis (Bishop et al., 2006; Hodin, 2006; Romero et al., 2013), our results support the hypothesis for the modulatory role of NO at the initiation of metamorphosis in specialist larvae. However, in contrast to larvae of P. sibogae or A. willowi, in which NO repressively modulates the initiation of metamorphosis, NO in H. asinina appears to inductively modulate the initiation of metamorphosis.

Specifically, all NOS inhibitors examined in the current study—L-NAME, AGH, and SMIS—significantly suppressed metamorphosis in H. asinina in the presence of live CA (Figure 3), indicating that NO is a facilitator of metamorphic induction in this species. Williams et al. (2009b) showed that live CA induces a significantly higher mean percentage of H. asinina larval metamorphosis than does dead (bleached) CA, implying that the biomolecules derived from live CA play a crucial role in the induction of metamorphosis. We propose therefore that the application of NOS inhibitors impedes the signaling required to modulate the biomolecules derived from live CA, resulting in significant suppression of larval metamorphosis.

Importantly, the failure of NO donors examined in this study—SNAP and L-arginine—to induce metamorphosis in the absence of live CA indicates that elevated NO alone cannot act as an inductive agent (Figure 4). This suggests that the activation of at least one other morphogenetic pathway by inductive live CA is required for successful metamorphosis of H. asinina. More than 25 years ago, Baxter and Morse (1987) proposed the existence of two independent metamorphic pathways in the red abalone, Haliotis rufescens. One of these, which constitutes adenylate cyclase cascades coupled with gamma-aminobutyric acid (GABA) receptors, they named the morphogenetic pathway. The other, which constitutes chemosensory lysine receptors regulated by G protein signaling complexes, they named the regulatory pathway. While the sole activation of the morphogenetic pathway by externally-applied GABA can directly induce metamorphosis of H. rufescens larvae, stimulation of the regulatory pathway alone fails to do so (Baxter and Morse, 1987). Remarkably, however, the regulatory pathway can significantly amplify the effect of metamorphic induction by sub-optimal concentrations of GABA, suggesting that the actual role of the regulatory pathway is to increase the sensitivity of the morphogenetic pathway (Baxter and Morse, 1987). The fact that the application of NO donors alone is unable to induce metamorphosis in H. asinina suggests that a similar system of dual metamorphic pathways may exist in H. asinina to that reported in H. rufescens.

Temporal gene expression profiles of HasNOS and HasHSP90A in post-larvae exposed to live CA are similar to each other (Figure 5), tentatively supporting the Bishop and Brandhorst (2003) hypothesis of NOS-HSP90 interaction at the initiation of settlement and metamorphosis. However, this support is not compelling and does not extend to expression profiles during attainment of competence, nor to spatial localization of transcripts. During larval development, the HasNOS expression has a sharp decrease to 84 hpf, but then increases again to 96 hpf, by which time all H. asinina larvae have achieved competency (Figure 5). As discussed above, NO may be playing a critical role in processing signals from inductive biomolecules to regulate the initiation of metamorphosis in H. asinina; therefore, we hypothesize that this increased expression around the time of acquiring competency is a molecular indication that larvae are developmentally mature enough to be able to respond to environmental cues that trigger the critical pelagic-benthic transition. The localization of HasNOS transcripts to the larval foot by 72 hpf (Figures 6C,D) and their expansion across the pedal sole by 96 hpf (Figures 6E,F) also supports this hypothesis (discussed further below). Intriguingly, both HasNOS and HasHSP90A expression show a second spike of up-regulation at 12 hpi (Figure 4), suggesting perhaps a different—and currently unknown—role of NO during post-larval development.

In other molluscs, in which NO acts as a negative regulator of settlement and metamorphosis, NOS expression data is reported only from the mud snail I. obsoleta (Hens et al., 2006) and the slipper shell snail C. fornicata (Taris et al., 2009). In I. obsoleta, the down-regulation of NOS soon after initiation of metamorphosis (Hens et al., 2006) was considered to be consistent with the hypothesis that endogenous NO is necessary for retention of the larval state (Froggett and Leise, 1999). In C. fornicata, NOS expression was reported to gradually increase through larval development, again consistent with a negative regulatory role of NO (Taris et al., 2009). However, NOS expression also then steadily increased during the first 6 h post metamorphosis, and this was attributed to handling stress (Taris et al., 2009). In H. asinina, our data also show some fluctuation in HasNOS expression soon after induction (Figure 5A), and we conclude that proper interpretation of these fluctuations will require additional expression analyses in conjunction with detailed descriptions of the cellular and morphological reorganization during metamorphosis of each species.

Past studies have provided evidence that the foot of gastropod veliger larvae plays a significant role in the induction of metamorphosis, as it is a potential source of both chemo- and mechano-receptors (Chia and Koss, 1988, 1989; Arkett et al., 1989; Jackson et al., 2005; Stewart et al., 2011). In H. asinina, the localization of Has-tft1, which is thought to be involved in the recognition of inductive cues, was detected in the anterior part of the larval foot (and the cephalic tentacle) in competent larvae (Jackson et al., 2005). More recently, Has-GABA_AR-β transcripts were reported localized to the lateral edge of the larval foot, and the application of an anti-GABA receptor protein antibody significantly reduced percent settlement induced by GABA, providing further evidence for the larval foot as a sensory structure to modulate the induction of metamorphosis in competent larvae of H. asinina (Stewart et al., 2011).

Spatial analysis of HasNOS gene expression (Figure 6) also provides evidence for a sensory role of the larval foot in H. asinina metamorphosis. HasNOS transcripts are not localized to the foot of 48 hpf non-competent veligers; however, localization is apparent in the lateral edge of the foot by 72 hpf and expands across the pedal sole by 96 hpf, by which time all larvae are competent to settle (Jackson et al., 2005). At the same times, there is no indication of HasNOS expression in any components of the larval CNS (Figure 6), regardless of whether these components are larval-specific or are retained in the adult CNS post-metamorphosis. Together, these data on NOS gene expression, combined with larval metamorphosis assays, strongly suggest that NO modulates the metamorphic induction of H. asinina larvae via the larval foot that is putatively functioning as a chemosensory structure to receive and process external settlement cues. This result differs from previous reports of NOS localization in the larval CNS (Thavaradhara and Leise, 2001; Pechenik et al., 2007; Bishop et al., 2008), and we propose below a relationship between these contrasting spatial expression patterns and contrasting regulatory roles for NO in molluscan metamorphosis.

A particularly significant outcome of this study is the unequivocal demonstration of NO as an essential molecule for the induction of H. asinina metamorphosis. This is a contrasting situation to all previously published studies on NO in molluscs. Based upon the limited comparative data available for NOS transcript and/or protein localization in gastropod larvae of multiple species, we present here a hypothesis to explain this contrasting function of NO in molluscan metamorphosis. Including the current study, the role of NO in molluscan metamorphosis has demonstrated three discrete patterns so far: (a) a negative regulatory role, whereby metamorphosis is induced solely by application of a NOS inhibitor, as reported in I. obsoleta, C. fornicata, and A. willowi (Froggett and Leise, 1999; Pechenik et al., 2007; Romero et al., 2013); (b) a negative modulatory role, whereby induction of metamorphosis is enhanced by the application of a NOS inhibitor in combination with an inductive substrate, as reported in P. sibogae and A. willowi (Bishop et al., 2008; Romero et al., 2013); and (c) an inductive role, whereby metamorphosis is suppressed by the application of a NOS inhibitor, as reported here for H. asinina. Very interestingly, these patterns are in accordance with variations in spatial NOS or NOS expression in these species.

Spatial expression data for the NOS protein or gene are available for I. obsoleta (Thavaradhara and Leise, 2001), C. fornicata (Pechenik et al., 2007), P. sibogae (Bishop et al., 2008), and H. asinina (this study). Both I. obsoleta and C. fornicata express NOS in the ASO (Thavaradhara and Leise, 2001; Pechenik et al., 2007), which as previously mentioned is a larval-specific sensory structure thought to function mainly to detect settlement cues (Croll and Dickinson, 2004; Croll, 2009). In I. obsoleta, the ASO is completely degraded by programmed cell death by 72 hpi (Gifondorwa and Leise, 2006). Although the fate of the ASO during and immediately after the metamorphosis of C. fornicata has not yet been reported, NOS expression is no longer detected in older veligers (9 days pre-competent and 23 days competent veligers), suggesting a mitigated function of the ASO as the veligers age (Pechenik et al., 2007). In P. sibogae, Bishop et al. (2008) found NOS localized to the cerebral ganglia, pedal ganglia, peripheral nervous systems around pedal ganglia, and the putative sensory cells of the foot. Unlike the ASO, the cerebral and pedal ganglia of P. sibogae become a part of the adult CNS after metamorphosis (Croll and Dickinson, 2004). Although the ASO of P. sibogae is a sensory structure that seems to receive inductive cues and modulate the initiation of settlement and metamorphosis (Hadfield et al., 2000), NOS localization is undetectable in the ASO in this species (Bishop et al., 2008).

In contrast to I. obsoleta, C. fornicata, and P. sibogae, NOS is localized to neither the larval nervous structure nor the CNS in H. asinina competent veligers, but instead to the putative peripheral sensory system of the foot. Therefore, we postulate that the means by which NO exerts its effect on metamorphosis of molluscan larvae may be directly related to the localization of NOS gene or protein with following patterns: (1) when NOS expression is mainly localized to larval-specific nervous structures that have a sensory role in the CNS, NO acts as a negative regulator; (2) when NOS expression is localized to parts of the CNS that are not larval-specific structures, NO acts as a negative modulator; and (3) when NOS expression is localized to peripheral sensory systems, NO acts as a facilitator of metamorphic induction. This hypothesis will be explicitly testable as more, taxonomically diverse studies on NOS regulation and expression in molluscan larval metamorphosis become available.

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.