Molluscs were among the first organisms on earth that were able to produce highly organized calcium carbonate composite materials with unique structural features and remarkable materials properties 12. Even today, the mollusc shell surprises us with new concepts for understanding biomineralization processes 3.

Chitin, a linear homopolymer consisting of β-(1–4)-linked N-acetyl-D-glucosamine subunits, plays an important role in mollusc shell formation. The presence of chitin in mollusc shell matrices is well documented in the literature 456. Recently, it has been demonstrated that chitin fulfils various structural tasks in the formation of larval shells of the bivalve mollusc Mytilus galloprovincialis 7. In the adult stage, the fibres of chitin and certain crystallographic axes of aragonite are aligned in mollusc nacre 89. Based on a cryo-TEM study, Levi-Kalisman and colleagues suggested that chitin is the ordered component of the decalcified nacre matrix, whereas a silk-like protein gel environs the chitin sheets 10. Mineralizing proteins are either attached to the core sheets or distributed within the silk like protein gel. Thus, the chitin and the silk together form a regular lamellar structure. Subsequently, certain mollusc shell protein fractions induce aragonite formation within this lamellar β-chitin and silk framework 11. The involvement of a transient amorphous mineral precursor phase in the formation of aragonite biominerals is currently discussed based on the presence of such a phase in calcite forming sea urchins and aragonite forming mollusc larvae 3121314. These observations raised new questions regarding the role of structural biopolymers such as chitin in the formation of shell microtextures 15. As recently discovered, the calcitic prismatic layer of the bivalve mollusc Atrina rigida (Pteriomorphia, Pinnidae) is the result of a structural interplay between chitin and the mineral phases 16.

Chitin synthases are transmembrane glycosyltransferases that are responsible for the enzymatic synthesis of chitin 17. The representatives of these enzymes in molluscs contain a N-terminal myosin motor head domain 18. For many years, the chitin synthases have been studied mainly in fungi 1920212223. Chitin synthases are localized either in the plasma membrane or in so-called chitosomes. Chitosomes are intracellular membrane vesicles that host the chitin synthase or its zymogenic precursor form and may contain preformed chitin in their lumen during vesicle transport prior to their fusion with the cytoplasmic membrane 2425.

Despite its strong ecological impact, the structure of the much more complex insect chitin synthase 26, which is closely related to the C-terminus of the mollusc enzyme 18, has attracted significant research interest in recent years 27. Mutagenesis experiments and RNAi approaches demonstrated that chitin does not only fulfil a structural role in the arthropod exoskeleton. In fact, it guides the development of invertebrates such as Drosophila, Tribolium, and Caenorhabditis 282930.

There are several options in order to interfere chemically with the biosynthesis of chitin 3132. One prominent example are small-molecule inhibitor drugs such as nucleosid-peptides that structurally imitate the UDP-activated chitin precursor substrate, UDP-N-acetylglucosamine (UDP-GlcNAc) and thus inhibit the chitin synthases of fungi and insects in a competitive manner 333435. Polyoxins, first time described in 1965, and Nikkomycins, first described in 1976, belong to this class of inhibitors that are produced by certain strains of Streptomyces, such as S. tendae and S. cacaoi 3637. Remarkably, these nucleosid antibiotics did not interfere with protein synthesis (reviewed by 31). Nevertheless, a broad profile of inhibitory potency, ranging from K_i = 0.02 μM to K_i = 3.5 μM, was reported for chitin synthases across different eukaryotic phyla in vitro 35. The commercially available chitin synthase inhibitor NikkomycinZ is well suited for non-invasive cell biological and biochemical studies in vivo: It is cell permeable due to active transport mechanisms that are employed by eukaryotic cells for the uptake of dipeptides. The fact that this substance is built from unusual amino acids counts for its resistance against intracellular proteolytic degradation 383940.

In the early embryonic stages of mollusc development, differentiated ectodermal cells form the shell field, a tissue that excretes the larval shell, as reviewed by Kniprath 41. In Mytilus galloprovincialis the shell field invaginates and forms a 'glandular' canal 42. The internal lumen of this "shell gland" is closed some hours after fertilization and the outmost cells start to secrete an organic shell cover, the periostracum, underneath a glycocalyx that originates from neighbouring cell microvilli. The "shell gland" everts while forming the larval mantle epithelium and the first mineralized shell 4344, see 45 for review). The shell of this trochophora larval stage, the prodissoconch I, is enlarged until the entire embryo is covered, and the organism metamorphoses into the motile veliger larva. In bivalves, the subsequently formed shell grows as well in thickness as concentrically at the hinge and at the shell edges, thus forming the prodissoconch II (for explanation, see also Fig. 1 of Ref. 14). Metamorphosis into the adult stage is indicated by the development of the foot tissue (pediveliger stage), and a sharp transition in microtexture between the remaining larval shell and the newly formed adult dissoconch 4647. During the prodissoconch II stage, the shell hinge is elongated in a straight manner. The interlocking hinge teeth are enlarged, and new teeth are formed in a row from the centre towards the edge. Remarkably, the functionality of the hinge apparatus is maintained throughout this complex developmental process 474849.

In a previous study, we investigated the chitinous matrix of larval shells of the marine bivalve mollusc Mytilus galloprovincialis 7. The way how the chitin is distributed throughout the organic framework of the larval shell, how its distribution and structure change with the time-course of larval development, and the reflection of typical functional elements of larval shells such as the hinge teeth, the bivalve symmetry, and the radial and tangential shell growth front within the observed chitin pattern suggest an integrative functional role for chitin synthesis in biomineral composite formation, from the subcellular to the organism scale. These observations motivated us to study the biochemical process of chitin synthesis in relation to the formation of the larval shell. Here, we present the first polarized light microscopic and scanning electron microscopic data from shells of the marine bivalve mollusc Mytilus galloprovincialis that were cultivated in the presence of the chitin synthase inhibitor NikkomycinZ.

By using NikkomycinZ in vivo as a small-molecule inhibitor drug we demonstrated that even a partial inhibition of chitin synthesis interferes dramatically with the biosynthesis of larval mollusc shells at various hierarchical levels. The presented data suggest that chitin formation may provide one of the regulatory targets of "biologically controlled" 50 mineralization, thus forming a multi-link between the constantly developing mollusc shell and the mantle epithelial cells. This might lead to a novel understanding of the role of the epithelial cells that, at one and the same time, guide larval development, secrete the shell precursor components, and monitor the dynamics of the various shell formation interfaces.

The primary aim of this study was to analyze the impact of chitin synthesis on mollusc shell formation. If the cellular regulation of chitin synthesis is related to mollusc shell formation, the partial inhibition of enzymatic chitin synthesis by a competitive enzyme inhibitor such as NikkomycinZ is supposed to induce structural alterations in the shell. Since also mollusc larvae of the marine bivalve species Mytilus galloprovincialis express the gene for a myosin chitin synthase specifically in the shell forming tissue 18, and since larval shells of this species consist of considerable amounts of chitin 7, these larvae provide an ideal system for studying possible effects of NikkomycinZ on the development of mollusc shells. Especially effects related to the transformation of amorphous calcium carbonate (ACC) into aragonite can be observed more clearly during larval shell development 14, while providing the option to extrapolate the results for a better understanding of the adult shell formation process 3. The study of larval shell formation processes is of fundamental interest since both, the mineralogy and the structural organization of the larval shell, are evolutionary highly conserved among diverse bivalve taxa. The fact that a "classification of larvae by hinges results in a classification closely parallel to a classification of adults" as recognized by Rees in 1950 48 indicates that the development and biomineralization of the hinge (provinculum) is of special evolutionary interest. The results presented in this study indicate that NikkomycinZ predominantly interferes with the mineralization process at the hinge region, suggesting that the mollusc myosin chitin synthase may have evolved as an effective regulatory element between cell differentiation and tissue mineralization.

Chitin synthesizing enzymes are complex transmembrane proteins that, unlike other glycosyltransferases 52, have an evolutionary highly conserved amino acid motif in their intracellular active site, which is identical in unicellular fungi as well as in more complex organisms (arthropods, molluscs), and which is well-known to accept UDP-GlcNAc as the only substrate 27. Conceivably, chitin synthases are highly sensitive for the structural UDP-GlcNAc-analogue NikkomycinZ that inhibits chitin synthesis in a competitive manner and is therefore applied as a pesticide against fungi and arthropods 32. This drug was also used to test the biological impact of the insect chitin synthase and found to be lethal for first instar larvae at their transition to second instar at concentrations of > 1 μM 26. It is self-evident that NikkomycinZ can be applied for structural studies on mollusc shell formation in vivo in only sublethal concentrations. We determined a suitable working concentration of NikkomycinZ in the range of 5 μM to 10 μM for larvae of Mytilus galloprovincialis. It is important for the interpretation of the obtained results to keep in mind that chitin synthesis is not supposed to be completely inhibited in that concentrations range. The exact degree of enzyme inhibition can hardly be estimated in vivo, as the quantity and enzymatic activity of mollusc chitin synthases could not be determined, and the exact rate of NikkomycinZ uptake into the larval tissue by mollusc peptide transport systems is unknown. Nevertheless, a phenomenological comparison between larvae cultured in the absence and presence of NikkomycinZ revealed that even a partial inhibition of chitin synthesis at 5 μM – 10 μM inhibitor induced drastic structural alterations in the larval shells. Taking the limited stability of NikkomycinZ in the culture medium (sea water) into account 53, it can be assumed that the effective working concentration of NikkomycinZ continuously decreased from the start value at the time of sea water replacement to the next sea water exchange, allowing the animals to compensate temporarily the detrimental effects on cell metabolism. In fact, the calnexin and calreticulin pathways 5455 provide eukaryotic cells with effective mechanisms to circumvent detrimental effects that an activated sugar analogue such as NikkomycinZ might have on the posttranslational modification of glycoproteins 56. Still, it has to be taken into consideration that certain extracellular biomineralization proteins, which are typically glycosylated (see 57 for review), might have been affected by NikkomycinZ as well, and therefore may contribute to certain effects on shell formation observed in this study. Furthermore, an altered level of chitin deposition could well interfere with intracellular signal transduction pathways that regulate the expression of biomineralization genes.

We optimized the NikkomycinZ treatment of the mollusc larvae in order to maintain a survival rate as documented in Fig. 2. Due to the fact that also chitin synthesis was not intended to be completely inhibited, the culture may have been kept under conditions that allow proteins glycosylation to occur above the critical threshold. We observed that even those organisms were extremely vital that were only partly covered with a shell. Therefore, we concluded that chitin synthesis is the much more NikkomycinZ sensitive process. Schlüter dealt with the same problem many years ago 58, inspired by the observations of Holst and colleagues that Nikkomycin was predominantly lethal for arthropod larvae in the course of moulting 59. In-between the moulting cycles, glycoproteins must be equally important for the survival of the larvae. Obviously, the synthesis of glycoproteins was significantly less affected in these invertebrate animals. Schlüter observed Nikkomycin induced ultrastructural defects in the procuticular region in moulting beetle larvae. Thus, the enzymatic synthesis of chitin appears to be a rather specific target of the Nikkomycins.

One of the general problems of experiments with mollusc larvae cultures is that a 100% survival rate can never be achieved, due to the fact that with proceeding development, a genetic defect of any individual might become lethal or may at least lead to an abnormal phenotype. Therefore, the really significant effect of NikkomycinZ on mollusc larvae populations was that after a certain period of incubation time, none of the animals had a "healthy" phenotype such as observed in populations that have not been in contact with the drug. Malformed animals were not always completely detected as such, mainly due to the low resolution of binocular microscopy, and due to the high motility of the individual larvae and the movements of velar tissue and cilia (see also Additional Files 5, 6, 7, 8 and 12, 13, 14, 15). Therefore the numbers of abnormalities within a NikkomycinZ treated population may in reality be higher than estimated (see Fig. 2 and Fig. 3 for details). Certain malformations were detectable in a definite way only by using scanning electron microscopy, which was not applicable for an in vivo screening.

We regarded it important to study morphological effects of the NikkomycinZ treatment by video microscopy. This method allowed us to demonstrate the viability of the organism along with the induced morphological alterations. Our primary aim was to exclude those individuals from the analysis that were affected by NikkomycinZ in any respect other than shell formation. On the other hand, it is conceivable that malformed shells will also interfere with the regular development of a bivalved organism. Therefore, the criteria as described in the results section were defined in order to group the phenotypes observed only in the motile organisms. In later experimental stages, shells of lethargic organisms were also taken into consideration, if a certain morphological effect was exposed there more clearly.

During the time-course of this study, it was recognized that certain categories of malformations were visualized by video microscopy only in a later developmental stage (after a longer treatment with NikkomycinZ), whereas the same features were observed by SEM comparably earlier (after short treatment with NikkomycinZ). This phenomenon becomes especially clear in Figure 5, when comparing the respective age of specimens in SEM and video microscopy images. Undoubtedly, this is one of the limitations of the in vivo approach on shell formation. Nevertheless, it can be speculated that the number of individual shells affected by NikkomycinZ was even higher than suggested by the data presented in Figure 3.

The interpretation of the observed phenomena is a tightrope walk: At first glance, our results suggest a direct impact of NikkomycinZ on chitin synthesis, despite taking some general effects on protein glycosylation 56 into account. However, as cellular regulation pathways are non-linear in general, which may apply for chitin synthesis in particular, one should keep in mind that effects are manifold and various effects may combine. Based on these considerations, several options remain for interpreting the observed shell morphologies in NikkomycinZ treated Mytilus galloprovincialis larval populations: The list of possible effects during chitin synthesis includes various interference patterns such as 1) with chitin fibril formation, 2) with the quantity of chitin in relation to the secreted mineral, 3) with local differences in the quantity of chitin deposition, 4) complex regulatory effects up to interference with genetic feedback-loops and signalling pathways of shell mineralization including shell proteins. This points towards the biggest problem of an in vivo approach in understanding the role of chitin synthesis in the regulatory network of mollusc shell formation: The function of many mollusc shell proteins still remains unknown 57, which is partly due to the fact that their activity depends on concerted interactions 60. Provided that artefacts (e.g. induced by drying of SEM samples) can be excluded, a possible explanation for some very rarely observed shell malformations in NikkomycinZ treated larva as shown in Fig. 13 could be that one or more additional, though unknown key players in biomineralization are mutated. The combined effects may then lead to uncontrolled self-assembly and mineralization and the organism may try to compensate the detrimental effects at different hierarchical control levels (for example, compare Fig. 13a, and 13b).

We summarized the possible consequences of NikkomycinZ in the shell formation process of molluscs in a schematic model (Fig. 14). In mollusc mantle cells, NikkomycinZ unequivocally inhibits chitin synthesis, whereas the extent of inhibition of glycosyltransferases involved in posttranslational protein modification is currently unknown. The proper glycosylation of secreted proteins is subject to cellular control mechanisms. However, biomineralization is an extremely fine-tuned process that is highly sensitive to the glycosylation pattern of the proteins involved in crystal nucleation and growth. Thus the NikkomycinZ induced shell malformations must not purely be associated with chitin synthesis inhibition. A putative mechanical feedback loop mediated by the myosin coupling of the chitin synthase with cytoskeletal signal transduction could explain the complexity of observed effects induced by NikkomycinZ, such as differences in intensity and quality of effects (survival and malformation rates, mineral texture, shell solubility etc.) observed in larvae of various age. At the present stage it cannot be excluded that the chitin synthase acts in an indirect way by serving "simply" as a polymer matrix supplier for biomineralization. Even then it may have an important function for the mechanical integrity of the shell (Fig. 12) as well as for the maintenance of the composite material's properties. As shown here, the mineral becomes more soluble in case of chitin depletion (Fig. 10). This observation is consistent with the rapid disappearance of "dead shells" from the culture, or at least the decrease in shell size, as often observed even in living animals (Fig. 4, Fig. 11). The observed increase in crystallinity (Fig. 11) may have something to do with the interplay between chitin and the acidic mollusc shell proteins, which are suggested to stabilize the amorphous mineral fraction 16.

Our observations presented here may also contribute to an implementation of mechanical signal transduction into mollusc shell formation concepts. The mechanical functionality of the hinge gets lost whenever the formation of hinge teeth (see Fig. 9) is prevented. This might influence the formation of the valves as well, if one assumes a mechanical feedback mechanism that is responsible for the regulation of shell formation. If forces during valve closure are responsible for a correct valve formation, a direct correlation with mineral deposition should be guaranteed. An asymmetry of the valves (Fig. 4b) or an irregularly shaped shell edge (Fig. 5a,b, Fig. 6b,d), hinge line (Fig. 4a, Fig. 5c, Fig. 11e), or hinge teeth (Fig. 9d,h) will inevitably prevent a correct force application. Consequently, this mechanical feedback at the organism's level will interfere with stimulating the correct mechanically induced cell answer such as deposition of multiple organic and inorganic shell precursor components that are necessary for correct valve structuring. As the mollusc myosin chitin synthase strongly interferes with the actin cytoskeleton (unpublished results), multiple bottom-up effects are expected to be induced once the central coordination gets lost. Thus, the chitin synthase inhibitor NikkomycinZ has obviously the ability to interfere directly with one of the central players of this biomechanical coordination process.

Our continued research on larval shell biomineralization in Mytilus galloprovincialis aims to address the issue of chitin synthesis and structural modifications in the context of regulatory cascades that coordinate shell formation and the development of the shell forming tissue within the whole body plan in vivo.

The small-molecule drug NikkomycinZ, a competitive chitin synthase inhibitor, was used at low concentrations (5–10 μM) for investigating the impact of chitin synthesis on mollusc larval shell formation in vivo. Being aware of the tremendous uncertainties that any in vivo approach bears, we observed dramatic structural alterations in mollusc larval shells by using polarized light video microscopy for the in vivo investigations and scanning electron microscopy imaging of extracted shell material prepared from NikkomycinZ treated Mytilus galloprovincialis larvae of various age. Provided that NikkomycinZ mainly affects chitin synthesis, our data obtained from different shell locations suggest that the mollusc chitin synthase fulfils an important enzymatic role in the coordinated formation of larval bivalve shells. It can be speculated that chitin synthesis bears the potential to contribute via signal transduction pathways to the implementation of hierarchical patterns into chitin mineral-composites such as prismatic, nacre, and crossed-lamellar shell types.

The author(s) declare that they have no competing interests.

Both authors contributed equally to the experimental work. VS participated in the design of the study, and drafted the manuscript. IMW proposed the main idea, designed and coordinated the study, and drafted the manuscript. All authors read and approved the final manuscript.