In crayfish, two gastroliths are elaborated during each premolt in the cardiac stomach wall [7]. When fully developed, at ecdysis (Figure 1A), all the gastroliths exhibit the same general morphology. They are semi-spherical, the face on the stomach lumen side (the first synthesized part of the gastrolith) appears flat or slightly concave whereas the other face, in close contact with the calcifying epithelium, is convex (Figure 1B–D; Figure 2A).

Gastrolith cross-sections observed in light microscopy show a dense structure of successive calcified layers of irregular width (Figure 2A–F). By different treatments with acetic acid, we obtained more or less partially decalcified gastroliths. After slight decalcification (Figure 2E–H), the organic matrix is revealed, showing that the layers are parallel to each other and to the cuticle at the beginning of the storage process, then they become concentric, appearing more and more enlarged in correspondence with the regular (almost linear) calcium precipitation occurring in late premolt until ecdysis [9]. Finally, at the end of the process, they develop considerably, covering laterally the first formed layers (Figure 2A, 2G and 2H, arrows). Organic layers are clearly visible when decalcification is a little bit stronger (Figure 2F).

After a few minutes incubation of broken gastroliths in 5% acetic acid, the large-diameter organic fibers, corresponding to chitin-protein fibers, are well visible on surface (Figure 3A–C). When more decalcified, a complex network of fibers appear showing not only the parallel organic layers previously observed in light microscopy but also cross fibers of similar size (Figure 3D and 3E). This transversal striated structure was already visible in some pictures of Figure 2 (notably in Figure 2A).

All these fibers organize a 3-dimensional framework forming pockets of different sizes. These pockets are themselves subdivided by fibers of nanometric diameters resulting in nano-compartments filled with CaCO₃ (Figure 3F and Figure 3G).

Such a pocket-like structure is visible on gastroliths extracted from the stomach of animals in early postmolt, which means after the beginning of a natural decalcification (Figure 4A–E). The decalcification process seems to occur “pocket by pocket” following enzymatic and/or acidic digestion within the stomach.

After extraction, lyophilisation and weighing of the IM and SM fractions, the whole organic matrix was estimated to represent between 3% to 5%, w:w, of a whole gastrolith. The organization and quantitative importance of the matrix network in the general morphology of a gastrolith is in accordance with the sponge-like structure obtained after complete decalcification of a gastrolith. After such treatment, if the volume is reduced because calcium carbonate has been discarded, the general morphology is conserved (Figure 4F–H) in accordance with the presence of a very dense and organized organic network.

When observed in light microscopy or in SEM, the layers appear densely and homogeneously calcified. At higher magnification, the precipitated mineral is present as layers of small spherules (about 100 nm diameter; Figure 4I–K), putatively aligned along and around organic matrix fibers as demonstrated for the outer decapod cuticle [47,48].

The determination of the mineralogical composition of gastroliths was performed by FTIR and XRD analyses. The CaCO₃ FTIR spectrum is characterized by four infrared vibration modes of the free CO₃²⁻ ions at about 1060–1080 cm⁻¹ (ν₁), 870–860 cm⁻¹ (ν₂), 1390–1430 cm⁻¹ (ν₃) and 680–710 cm⁻¹ (ν₄) [49]. The precise absorption bands of each vibration vary depending of the polymorph of CaCO₃. FTIR ACC spectrum is characterized by a slight peak at about 860 cm⁻¹ (ν₂), a broad/double peak at 1000–1070 cm⁻¹ (ν₁) and a double peak at 1380–1470 cm⁻¹ (ν₃).

As for the putative presence of phosphate compounds, the PO₄³⁻ ions infrared vibration modes give rise to two characteristic absorption bands located at 1200–900 cm⁻¹ and 650–550 cm⁻¹[49].

When performed on a Cherax gastrolith cross-section in four different sites, the obtained FTIR spectra demonstrate everywhere the presence of ACC (Figure 5A and B). Similarly, when performed on powdered gastrolith from Cherax quadricarinatus (Figure 5C) or Pacifasctacus leniusculus (Figure 5D), FTIR spectra are characteristic of ACC.

For comparison, calcite spectrum is characterized by a small peak at 710 cm⁻¹, a sharp and high peak at 870 cm⁻¹ and a single peak at 1390 cm⁻¹ (see Figure 6C) [50]. On the other hand, the FTIR spectrum obtained is neither characteristic of aragonite nor characteristic of another biogenic crystalline calcium carbonate polymorph, which have been analyzed previously [50,51,52].

We do not exclude the possibility that the absorption band visible at 1070 cm⁻¹ could be also related to the presence of PO₄³⁻ ions as indicated for the analysis of the giant prawn exoskeleton [49]. Nevertheless, no absorption band is visible in the range 650–550 cm⁻¹.

XRD spectra obtained on gastrolith powder exhibit a typical ACC spectrum with no narrow sharp peaks but a very broad band spreading from 700 to 330° (2-Theta scale) [53,54]. It is to notice that a typical ACP spectrum exhibits similarly two diffuse broad bands as visible for example on the spectrum 1 of Figure 5E.

In some analyses, some peaks, more or less high, can be referred to calcite (Figure 5E). Figure 5F shows two XRD spectra obtained from gastroliths of the same crayfish species. One is typical of ACC whereas the other exhibits a mix of ACC and calcite. The presence of a natural mix of amorphous and crystallline CaCO₃ has previously been reported [49,55].

Another explanation has been obtained from observations of gastrolith cross-sections in polarized light microscopy. Figure 6 shows pictures obtained in bright-field light microscopy (Figure 6A1 and B1) and polarized light microscopy (Figure 6A2 and B2) on gastrolith cross-sections. The surface of the gastrolith appears everywhere opaque except for birefringent areas corresponding to crystalline part (Figure 6A2 and 6B2, arrows). Light microscopy showed that these zones are connected to the surface (Figure 6A1 and 6B1, arrows) and correspond to artificial fractures within which aqueous solution penetrated the gastrolith. This produced a local solubilization of ACC followed by a secondary recrystallization in calcite, the most stable CaCO₃ polymorph, as demonstrated by XRD and FTIR spectroscopy (Figure 6C and 6D). Such an experimental transformation may occur during the extraction procedure or during the NaOCl washing step, which has the purpose of eliminating putative organic contaminants (originating from the stomach epithelium for example). A similar experimental secondary crystallization has been previously observed for ACC storage deposits elaborated by a terrestrial amphipod, Orchestia cavimana [56].

Among the molecules of the organic matrix of crayfish gastrolith that are considered as mainly composed of chitin-protein fibers, proteins and proteoglycans have received the most attention so far [13,14,15,16,17,18,19,20,21,22,23,24,25,28].

In this work, we performed an analysis of monosaccharides from the acetic acid-insoluble fraction of matrix gastrolith (IM; obtained as described in the Experimental section) originating from three crayfish species by high performance anion-exchange chromatography after TFA hydrolysis.

The amounts of the detected monosaccharides are reported in Table 1A and classified by their order of abundance in Cherax quadricarinatus.

For all the analyzed crayfish species, the IM matrix exhibits a high level of glucosamine, from about 65 to 72.5% of the total sugar content (Table 1C), in accordance to the presence of large chitin fibers, which cannot be solubilized and depolymerized during the acetic acid decalcification process, nor during the NaOH treatment. The high level of glucosamine in the IM matrix is in accordance with the FTIR spectrum obtained from the C. quadricarinatus IM fraction (Figure 7A). This spectrum is highly similar to the spectrum obtained by using commercial crustacean chitin (from the king crab; Sigma™), whereas the soluble matrix FTIR spectrum (Figure 7B) obtained from C. quadricarinatus is a little bit different, even though some absorption peaks could be attributed also to the presence of chitin. Notably, if the glucosamine level is about 72.4% for Cherax in the IM fraction (Table 1C), it remains not negligible in the SM fraction (about 35.8%) [28]. This could be due to the presence of very small chitin fibers (see Figure 3F and G) subdividing the large pocket-like compartments into micro-units framed by larger chitin fibers. Some of these nanometric diameter fibers could be depolymerized and released during the decalcification process, remaining in the supernatant after centrifugation of the decalcifying solution.

Besides glucosamine, we detected other monosaccharides, each one representing less that 10% of the total sugar content of the IM matrix (Table 1B). The order of abundance of these other sugars is not exactly the same from one crayfish to another even though xylose, arabinose, glucose and mannose are always the most abundant (from about 3 to 9.5%; Table 1C). It is notable that glucuronic acid was not detected in C. quadricarinatus, Parastacidae, and that galacturonic acid and rhamnose were only found in O. limosus, Cambaridae.

The presence of the poorly abundant monosaccharides cannot be really explained in all cases, for example in the case of arabinose, glucose and fucose. They could be free or part of glycosyl groups linked to proteins, the precise function of which remains to be determined.

Nevertheless, the presence of xylose and mannose may alternatively be related to the presence of proteoglycans [28]. Indeed, chondroitin sulfate, dermatan sulfate and keratan sulfate are proteoglycans composed of a glycosaminoglycan (GAG) bound to a protein core via a xylose sugar. On the other hand, keratan sulfate II (KSII) is composed of a glycosaminoglycan (GAG) bound to a protein core via a mannose sugar [57,58].

How the biogenic ACC polymorph can be induced and stabilized over time is a long-standing and disputable question that has received recent attention. Stabilization could be due to stabilizing molecules or to molecules able to inhibit the transformation of ACC to another more stable crystalline polymorph. It was suggested that specialized macromolecules (acidic proteins, phosphoproteins, sulfated glycoproteins, polysaccharides) or ions such as magnesium or phosphate could contribute to the ACC determination/stabilization [18,20,46,59,60,61,62,63,64,65,66,67].

The gastrolith matrix protein GAP 65, with putative phosphorylation sites, was shown to in vitro inhibit calcium carbonate crystallisation [18]. More recently, Bentov et al. [13] demonstrated the ability of an isoluble matrix fraction to in vitro induce ACC. This ability was attributed to a doublet of phosphorylated and calcium-binding polypeptides migrating at about 70-75 kDa in SDS-PAGE. Nevertheless, these proteins were not purified and tested alone for their capacity to induce ACC.

In gastroliths, as putative molecules bound to chitin involved in the ACC induction process upstream the ACC stabilization effect of low-molecular weight components [24,25], we may suspect proteins with post-translational modifications such as phosphorylations, sulfations or glycosylations. Proteoglycans can also be considered as another category of gastrolith matrix components, bound or not to chitin, putatively involved in the ACC determination.

The Australian red claw crayfish, Cherax quadricarinatus, were bred in a farm located on the campus of the University of Chile, Santiago de Chile. Gastroliths were extracted from animals at ecdysis (moment when the animals shed their old cuticle). The American crayfish, Orconectes limosus, and the Californian crayfish, Pacifasctacus leniusculus, were fished in the Parc du Morvan and in Burgundy rivers, France, then kept in freshwater tanks in the University of Burgundy until ecdysis. Gastroliths of the Louisianan crayfish, Procambarus clarkii, were obtained from animals originating from the region of Bordeaux, France, and were a gift from Dr. J.P. Delbecque from the “Centre de Neurosciences Intégratives et Cognitives”, University of Bordeaux, France.

Gastroliths were immersed in 10% (v/v) NaOCl overnight at 4 °C under gentle stirring to remove superficial organic contaminants then thoroughly rinsed with distilled water and air-dried at 37 °C for 1 hour.

For XRD or IR spectroscopy analyses, gastroliths were ground to a powder and kept at room temperature and dry air until used.

For sugar and matrix analyses, powder was decalcified during 48 h in cold acetic acid (10% v/v) at 4 °C. The solution was then centrifuged 30 min at 3893 g at 4 °C. The pellet, corresponding to the acetic acid-insoluble matrix (IM), was rinsed 10x with 50 ml MilliQ water (Millipore^®), freeze-dried and weighed. The supernatant containing the acetic acid-soluble matrix (SM) was filtered (5 µm cut-off) and concentrated with an Amicon ultrafiltration system on a Millipore^® membrane (YM10; 10-kDa cut-off). The solution (approximately 5 mL) was extensively dialyzed against MilliQ water (3 days, several water changes) before being freeze-dried and weighed.

For chemical treatment, pieces of freshly broken gastroliths were incubated in acetic acid 10% or 5% (30 sec), depending of the level of decalcification expected for examining the structure on the surface or deeper. Then, they were carefully rinsed with distilled water and air-dried before microscopic observations. Full gastroliths were completely and slowly decalcified with 2.5% acetic acid to avoid desorganizing the organic network.

For light microscopy observations, gastroliths were embedded in araldite and cross-sections of gastroliths (50-µm thickness) were performed. They were observed using a Nikon Eclipse E200 microscope or a Nikon SMZ800 binocular stereomicroscope.

For SEM observations, gastroliths were slightly broken in a mortar with a pestle. Some pieces were observed without any further treatment, others were decalcified with 1M EDTA for 2 min for a slight surface treatment. For a more intensive decalcification, samples were incubated 30 min in 10% acetic acid. After chemical treatment, samples were air-dried and carbon covered. They were observed using a Hitachi TM-100 scanning electron microscope or a JEOL 6400 F scanning electron microscope (for high magnifications).

Observations in polarized light were performed using a Zeiss Axiophot microscope.

X-ray diffraction spectra were obtained from powdered gastroliths using a Siemens Buker D500 X-ray diffractometer. CaCO₃ calcite (calcite D, ref 00-024-0027, Siemens) peaks were used as a reference to determine the amorphous to calcite transformation.

Infrared spectroscopy was performed on powdered samples by direct ATR transmission according to the method previously described [52,68]. IR absorption spectra were obtained using a Fourier transform infrared (FTIR) spectrometer (Bruker-Vector 22) in the range 2000–200 cm⁻¹, 2 cm⁻¹ resolution and accumulation of 32 scans.

For sugar analysis, we used lyophilized IM matrix samples from 3 different species. Each of these samples corresponds to a pool of six gastroliths ground in powder and decalcified as described above (see 3.2). The lyophilized samples were hydrolyzed in trifluoroacetic acid (TFA, 2 M) at 105 °C for 4 h. Samples were evaporated to dryness before being resuspended with NaOH 20 mM. The acidic, neutral, and amino sugar contents of the hydrolysates were determined by high-performance anion exchange-pulsed amperometric detection (HPAE-PAD) by using a CarboPac PA100 column (Dionex P/N043055), according to the Dionex instructions. Non-hydrolyzed samples were analyzed similarly, in order to detect free monosaccharides in the samples. The results obtained (means±standard deviations) for each species come from three analyses performed with three different pools of gastroliths. Blank experiments were performed using BSA diluted in 10% acetic acid and submitted to all the steps of the extraction procedure. It is to notice that this technique does not allow the quantification of sialic acids, which are destroyed during hydrolysis with TFA, nor the detection of iduronic acid because of the lack of available standards. All the analyses have been performed in the same lab, UPSP PROXISS, AgroSupDijon, Dijon, and by the same person to avoid possible experimental discrepancy.