The key role of the surface membrane in why gastropod nacre grows in towers

Abstract

The nacre of gastropod molluscs is intriguingly stacked in towers. It is covered by a surface membrane, which protects the growing nacre surface from damage when the animal withdraws into its shell. The surface membrane is supplied by vesicles that adhere to it on its mantle side and secretes interlamellar membranes from the nacre side. Nacre tablets rapidly grow in height and later expand sideways; the part of the tablet formed during this initial growth phase is here called the core. During initial growth, the tips of the cores remain permanently submerged within the surface membrane. The interlamellar membranes, which otherwise separate the nacre tablet lamellae, do not extend across cores, which are aligned in stacked tablets forming the tower axis, and thus towers of nacre tablets are continuous along the central axis. We hypothesize that in gastropod nacre growth core formation precedes that of the interlamellar membrane. Once the core is complete, a new interlamellar membrane, which covers the area of the tablet outside the core, detaches from the surface membrane. In this way, the tower-like growth of gastropod nacre becomes comprehensible.

Nacre is by far the most intensively studied non-human organo–mineral biocomposite. It has a high proportion, ≈5%, of organic matter (proteins and polysaccharides; ref 1), the mineral fraction being exclusively in the form of aragonite. Jackson et al. (2) estimated that its work of fracture is 3,000 times higher than that of inorganic aragonite, although later estimates reduce this figure considerably (see the review in ref. 3). Its superior biomechanical properties, together with its interest to the pearl industry and its possible biomedical uses (see e.g., ref. 4), make nacre the subject of many biomimetic studies. An ultimate aim of such work is to mimic nacre in the laboratory, following the biological principles used by molluscs to produce such a biomaterial (5). It is sine qua non for this objective to have a complete understanding of the mechanisms involved in nacre growth.

Nacre is secreted only by the molluscan classes Gastropoda, Bivalvia, Cephalopoda, and, to a minor extent, Tryblidiida. It has a lamellar structure consisting of alternating tablets of aragonite 300–500 nm thick and 5–15 μm wide and organic interlamellar membranes ≈30 nm thick, which have a core of β-chitin surrounded by acidic proteins (6). It is now clear that the sequence of nacre formation involves the secretion of interlamellar membranes (7) separated by a liquid rich in silk fibroin (5); only subsequently is the liquid replaced with mineral (7–9). This pattern is the same for the bivalves and gastropods, and it is likely so too for the other nacre-secreting molluscs, although this is yet to be determined. There are, however, structural differences between bivalve and gastropod nacre. In the former group, the interlamellar membranes are secreted with just the liquid-filled extrapallial space between them and the cells of the mantle epithelium, and mineralization within the membranes proceeds in a step-like manner (7, 9). The nacre thus produced is said to have a terraced arrangement (Fig. 1A).

Fig. 1.

Bivalve and gastropod nacre growth compared. (A) Oblique view of the terraced nacre of the bivalve P. margaritifera. (B) Oblique view of the towered nacre of the gastropod Perotrochus caledonicus.

In gastropods, however, the biomineralization compartment of nacre is enclosed by a surface membrane first reported by Nakahara (8) in Monodonta and Haliotis. Since its discovery, its existence went unremarked, until Cartwright and Checa (10) realized that it is widespread in nacre-secreting gastropods and that the interlamellar membranes must necessarily detach from it. The surface membrane acts as a protective seal, which prevents the organic compounds and minerals involved in nacre growth from being lost to the external environment when the soft body of the gastropod withdraws into its shell, something evidently not necessary with bivalves. Below the surface membrane, many parallel interlamellar membranes with tablets growing between them can be found. These tablets are typically stacked in towers (Fig. 1B), with the smaller, more recently begun tablets found at the top. Although the nacre of gastropods, in particular that of the abalone, i.e., the genus Haliotis, has been intensively studied, there are still many pieces to be assembled in the puzzle. One, perhaps key piece, is the surface membrane, key both because it is intimately related to the other components of nacre and because the mineral ions and organic molecules for nacre growth are necessarily introduced into the biomineralization compartment through it. Its ultrastructure, growth, and secretional activity have never been elucidated.

This work is dedicated to determining the relationship of the surface membrane to the interlamellar membranes and mineral tablets. Our conclusions shed light not only on the dynamics of gastropod nacre growth but also bear on the present debate about whether superimposed nacre tablets nucleate and grow onto the organic interlamellar matrix or, alternatively, whether there is crystallographic continuity between them across the interlamellar membranes.

Results

Surface Membrane.

The surface membrane extends between the adoral and apical boundaries of the nacreous layer, usually bounded by the external spherulithic layer and an internal aragonitic lamellar layer of uncertain microstructure (Fig. 2A). In Gibbula and Monodonta, at least, its mantle-side surface is dotted with bodies adhering to it (Fig. 2 A and F). In transmission electron microscopy (TEM) sections these structures are seen to be hollow (Figs. 2 B and C and 3C) and thus may be called vesicles. They vary in shape from spherical, when they are just touching the surface membrane, to strongly compressed, when they are partially or wholly integrated into the surface membrane (Fig. 2 B and C). Their walls are electron-dense and have a mean thickness of 10–15 nm (Fig. 2B Inset).

Fig. 2.

SEM views of the nacre of G. pennanti. (A) View of the internal surface of the shell, close to the aperture. The nacre compartment is overlain by the surface membrane, visible as the darker area, which has cracked and curled upon contraction during preparation. (Inset) View of the mantle side of the surface membrane with vesicles. (B) TEM section of decalcified nacre with vesicles adhering to the mantle side of the surface layer. (Inset) Bilayered appearance of the wall of one such vesicle. (C) As in B, with some vesicles apparently in the process of being incorporated into the surface membrane. (D) Nacre-side view of the surface membrane, with the last-formed interlamellar membrane adhering. The surface membrane can be differentiated by its smooth aspect. Arrows indicate two tablet cores semidetached from the surface membrane. (E and F) Transverse views of the surface membrane and of the underlying interlamellar membrane and tablets in formation. (E) The last-formed interlamellar membrane and the surface membrane meet in the apical direction (Right of photograph). (F) The surface membrane (with vesicles) has partly been torn off, which has exposed the last core. il, internal lamellar layer; ilm, interlamellar membranes; n, nacre; os, outer spherulithic layer; sm, surface membrane.    

Fig. 3.

TEM sections that have partly penetrated the axes of nacre towers of M. labio in formation show crystal continuity and/or the absence of interlamellar membranes (arrows). (C and E Insets) Details. (B and F) Details of A and E, respectively. (D and E) White areas are where tablets have partly disappeared during sample preparation. ilm, interlamellar membranes; sm, surface membrane; v, vesicles.

In section, the surface membrane has a mean thickness of 100 nm, markedly thicker than the underlying interlamellar membranes (≈30 nm) (Figs. 2 B, C, E, and F, 3 C and D, 4A). In fracture and TEM sections, it has a homogeneous appearance.

Fig. 4.

Nacre of G. umbilicalis. (A) TEM section approximately through the axis of a decalcified sample. The interlamellar membranes are seen as diffuse organic membranes or disappear entirely at the axial zone. (Inset) Detail of two such interlamellar membranes. (B) Polished section through the axis of a nacre tower (BSE mode). (Inset) The organic membranes become diffuse and tend to be convex upward. (C and D) Fixed and decalcified polished section of nacre. The axis of the tower is marked by aligned holes with coarsened rims within the interlamellar membranes (arrows). (C) Succession of ≈10 such holes that are perfectly aligned. Their mean diameter is ≈150 nm. (D) Similar case in which a hole has been sectioned. ilm, interlamellar membranes; sm, surface membrane.

Relationship Between the Surface Membrane and Other Nacre Components.

Examination of the nacre side of the surface membrane reveals that the interlamellar membranes develop in close contact with the surface membrane. The difference between them is evident from their fibrous nature compared with the smoothness of the surface membrane (Fig. 2D). In section, it is notable that the surface membrane generally intercepts the last-formed interlamellar membranes at a shallow angle in the apical direction (Fig. 2E; see also ref. 10, Fig. 4 g and h).

Scanning electron microscopy (SEM) observations of the nacre side of the surface membrane also reveal the existence of tablet cores growing between the last-formed interlamellar membrane and the surface membrane and partly encased within it (Fig. 2 D and E). Where tablets have been torn off upon contraction of the membranes during sample preparation, the scars remaining can also be seen (Fig. 2D). The width of these cores is estimated to be between 100 and 200 nm (Figs. 2 E and F, and 3 A, C, and D). The topographical relationship observed in SEM samples is so recurrent that the possibility that this is artifactual can be excluded. The relationship is further demonstrated by TEM, which reveals that the tip of a growing tablet (i.e., the last 50–70 nm) is directly embedded within the surface membrane (Fig. 3 A, C, and D).

In the few instances when towers are fortuitously sectioned exactly through their central axis, the interlamellar membrane possesses a fuzzy appearance or is totally absent. The disappearance of the interlamellar membrane at the very axes of the towers is evident in some exceptional TEM views (Fig. 3). The partial dissolution sometimes produced during sample preparation (Fig. 3 D and E) does not affect the tower axis area. TEM views of decalcified towers of Gibbula umbilicalis show too that the interlamellar membranes are missing at the very axes of the towers, across a maximal width of ≈100 nm, or are replaced by a fuzzy band of organic matter with a different orientation (Fig. 4A). SEM observation in back-scattered electron (BSE) mode of polished axial sections of nacre towers of the same species, in which we can safely assume that membranes have not been disturbed during sample preparation, manifests that the same effect may take place across tens of tablets in a tower (Fig. 4B). The fuzzy band, when present, usually curves slightly toward the top of the tower. When the same samples are decalcified with methanolic solution, the axes of the towers are marked by a succession of holes (≈150 nm wide) with coarsened rims, sometimes traversed by organic threads (Fig. 4 C and D). The regularity and persistence of such structures exclude the possibility that they are artifacts caused by dissolution. Treatment with 2% EDTA preferentially removes the calcified organic-rich components: the interlamellar organic membranes, the lateral boundaries between tablets and, interestingly, the parts of the tablets coinciding with the axes of towers [supporting information (SI) Fig. S1].

High-resolution TEM observations show that the interface between two superimposed tablets is fully crystalline at the axis (Fig. 5A). Observation of lattice fringes and fast Fourier transform (FFT) analysis of small areas provide additional evidence of this crystalline character. The patterns obtained, although indicative of nonuniform orientation (Fig. 5B), are comparable with those obtained within the interior of nacre tablets, which are composed of nanodomains with variable orientations (X. Li, personal communication).

Fig. 5.

High-resolution TEM study of the axial zones of nacre tablets of M. labio. (A) The contact between the last two tablets is fully crystalline and consists of a single crystal domain. (B) The different areas at and around the contact between the last two tablets are crystalline, as shown by the FFT patterns. ilm, position of interlamellar membranes.

Discussion

Our results, together with those of Nakahara (8, 9, 11), demonstrate that the surface membrane is present in representatives of at least three (Haliotidae, Trochidae, and Turbinidae) of the six nacre-secreting families of gastropods (12), all grouped within the Vetigastropoda (13). We may hence consider the surface membrane a basic element of gastropod nacre. Although its growth dynamics is not yet totally elucidated, the surface membrane seems to form by addition of organic vesicles to its mantle-side surface (Fig. 2B), which gradually integrate into it (Fig. 2C). We hypothesize that this is a means by which components necessary for the production of organic and crystalline structures within the nacre compartment are transported. These observations are compatible with others that show that the surface layers of the tablets are of amorphous calcium carbonate (14) and that mineral may be precipitated intracellularly before being transported to the mineralization site where it is remodeled (15), but the exact sequence of events in mineralization is a matter for further study.

Based on our results, we reach four main conclusions. (i) Interlamellar membranes detach from the nacre side of the surface membrane. From the topographic relationships between structures, Cartwright and Checa (10) concluded that interlamellar membranes form at the nacre side of the surface membrane and detach from it in an apical direction. Our observations support this view (Fig. 2 D and E). In this way, the balance between the components acquired via vesicle addition and those lost because of the formation of interlamellar membranes remains steady so that the surface membrane maintains a constant thickness.

(ii) Nacre tablets begin growing within the surface membrane. It is known that tablets first acquire their maximal height and subsequently expand sideways until they impinge on each other (16). Their initial growth in height keeps pace with the separation of an interlamellar membrane from the surface membrane so that they stay in contact with the rest of the tower below them, at the same time that their tips remain permanently submerged within the surface membrane (Fig. 3 A, C, and D). Nakahara (8) and Mutvei (17) concluded, based on TEM and etching techniques, respectively, that nacre tablets have an organic-rich core. In our EDTA-treated samples the cores of tablets etch preferentially (Fig. S1), which is also indicative of their organic-rich composition. This is comprehensible because during their initial growth within the surface membrane, the tablets should incorporate organic material from the surface membrane.

(iii) The interlamellar membrane disappears completely or continues only as a diffuse band at the very core of the crystalline tablets (Fig. 3 and Fig. 4 A–D). Interlamellar membranes are electron-dense and sometimes appear bi- or trilayered (see also refs. 7, 8, and 11 and Figs. 2C and 3D); their aspect as a fuzzy band in samples cut exactly through the tower axis (Figs. 3 and 4) is thus quite distinct (Fig. 4A Inset).

(iv) At the location of the core, the interface between growing tablets is fully crystalline. This observation (Fig. 5) does not totally preclude the existence of the interlamellar membrane at this position because Rousseau et al. (18) showed that the mineralized interlamellar membranes of the pearl oyster Pinctada margaritifera are partly nanocrystalline. In our case, FFT patterns (Fig. 5B) are identical to those obtained within the interior of nacre tablets, being characterized by a nanodomain-like ultrastructure.

These four findings above can be understood coherently with the following model. A new interlamellar membrane does not penetrate through the very core of the tablet because core formation precedes that of the interlamellar membrane; when it detaches from the surface membrane, the tip of the tablet core is already present. The fuzzy band sometimes observed may be explained by competition between the mineral and the organic molecules leading to cocrystallization. The sequence of events is thus as follows: (i) formation of an organic-rich tablet core, which grows rapidly in height with its tip embedded in the surface membrane (Fig. 6A); (ii) cessation of tablet growth in height and simultaneous detachment of another interlamellar membrane; organic compounds may adhere to the tip of the tablet within the surface membrane (Fig. 6B); (iii) formation of a new tablet core (Fig. 6C).

Fig. 6.

Scheme for the formation of incipient nacre tablets. (A) The tablet core grows rapidly in height with its tip immersed within the surface membrane. An organic-rich core is formed as the growing tablet absorbs components of the surface membrane. (B) At the same time that vertical crystal growth ceases, a new interlamellar membrane is secreted at the nacre side of the surface membrane. During this time interval, organic material may precipitate on top of the tablet core. (C) Growth of a new tablet commences.

The basic model for gastropod nacre growth of Nakahara (8, 9), which states that interlamellar membranes and the compartments they produce are formed in advance of the nacre tablets, thus has to be modified insofar as each interlamellar membrane is formed after the core of the tablet, which thus becomes a distinct part within the tablet, but before the tablet begins to expand laterally. The above model implies that nacre tablets are crystallographically connected at their cores. The existence of a fuzzy organic band of material in between should not represent an obstacle for crystallographic continuity, as shown by our analysis of the lattice fringes. This would explain why nacre tablets stacked along a single tower retain their overall crystallographic orientation (19 and unpublished data). In more detail, nacre tablets are composed of many twinned crystals (17), and we present additional TEM evidence (Fig. S2) that multiple crystal orientations are inherited by newborn tablets.

Cartwright and Checa (10) hypothesized that the different stacking patterns of gastropods and bivalves could be related to the sizes and densities of the nanopores they observed in the interlamellar membranes. The present work shows that the difference is not merely quantitative because the existence of the surface membrane and its associated effect on nacre growth strongly promote vertical stacking in gastropods.

In implying the crystallographic continuity of the cores of tablets, our hypothesis bears implications on the present debate of whether tablets communicate across lamellae or not. This debate began with Weiner and Traub (20), who found that the fiber axis of the chitin and silk protein forming the interlamellar matrix are perpendicular to each other and aligned with the a- and b-axes of the aragonite tablets, respectively. They proposed that the mineral phase grows epitaxially onto the protein chains of the organic matrix; this is the heteroepitaxial theory. Schäffer et al. (21) recognized the existence of many pores, several tens of nanometers across, in the intercrystalline matrix of abalone nacre, which they showed to be permeable to ions. They suggested that some of the pores allow tablets to grow from one layer to the next, without the need for a new nucleation event; this is the mineral-bridge theory. From then on, the heteroepitaxial and mineral-bridge theories have been two conflicting views on how nacre tablets relate to each other (see the review in ref. 10). The evidence we have presented here does not rule out either theory, but it does show that, for gastropod nacre, there is a third way: tablets are connected at the tower axes. This connection, in turn, explains the tower-like growth of gastropod nacre.

Materials and Methods

SEM.

Shells of living specimens of G. umbilicalis, Gibbula pennanti, Monodonta sp., and Calliostoma zyzyphinus were fixed with 2.5% glutaraldehyde in a 0.1 M cacodylate buffer. Samples were usually observed intact after CO₂ critical point drying. Some polished sections were decalcified according to two different protocols: (i) 2% EDTA for 2–3 min; (ii) fixation of the organic matrix with a mixture of 2.5% glutaraldehyde and 2% formaldehyde and further demineralization with a methanolic solution (3:1:6) in a gel medium (protocol by Á. Hernández-Hernández, unpublished data). This procedure preserves the finest details of the organic membranes (note, e.g., the nanopores in the interlamellar membranes in Fig. 4 C and D).

TEM.

Resin-embedded specimens of G. umbilicalis were completely decalcified with 2% EDTA and prepared with an ultramicrotome in the standard way. We also had access to original material of Monodonta labio, and Haliotis rufescens of the late H. Nakahara. The samples were prepared in Meikai University by M. Kakei according to the protocol described in ref. 9. Only samples of M. labio rendered significant results.

We used a Leo Gemini 1530 field-emission SEM and a Philips CM20 TEM of the Centro de Instrumentación Científica, University of Granada. High-resolution TEM analysis was carried out in a Jeol 2200FS at the Centro de Investigação em Materiais Cerâmicos e Compositos, University of Aveiro.