Amorphous calcium carbonate biomineralization in the earthworm’s calciferous gland: Pathways to the formation of crystalline phases

Journal of Structural Biology 162 (2008) 422–435

Contents lists available at ScienceDirect

Journal of Structural Biology journal homepage: www.elsevier.com/locate/yjsbi

Amorphous calcium carbonate biomineralization in the earthworm’s calciferous gland: Pathways to the formation of crystalline phases L. Gago-Duport a, M.J.I. Briones b,*, J.B. Rodríguez c, B. Covelo d a

Dept. Geociencias Marinas. Universidad de Vigo, 36310 Vigo, Spain Dept. Ecología y Biología Animal, Universidad de Vigo, 36310 Vigo, Spain c Servicio de Microscopía Electrónica, CACTI. Universidad de Vigo, 36310 Vigo, Spain d Servicio de difracción de Rayos X, CACTI. Universidad de Vigo, 36310 Vigo, Spain b

a r t i c l e

i n f o

Article history: Received 16 July 2007 Received in revised form 22 February 2008 Accepted 25 February 2008 Available online 7 March 2008

Keywords: Calciferous gland Biomineralization Calcium carbonate Earthworm HREM X-ray diffraction FT-IR spectroscopy

a b s t r a c t In this study, we investigated the microstructural transformations that take place during carbonate formation in the earthworm’s calciferous gland by analysing the evolution from the precursor fluid of the solid phases (spherulites) to the final carbonate concretions released by the gland. Results from HREM and electron diffraction showed that the spherulithic deposits merely consisted of ACC partially transformed to vaterite. Furthermore, comparisons of the diffraction spectra and microstructural analyses allowed the identification of the transition sequences to more stable carbonates. And thus, transformations of ACC to calcite were observed on the surfaces of these amorphous globular aggregates as their smooth characteristic surface became rougher with time. This transition path was not unique, and the presence of aragonite, as an intermediate phase, has also been found. In this particular case, the transition process followed a completely different pathway with the crystallization starting in the centre of the sphere and progressively extending to the periphery, leading to the formation of radial aggregates. In situ experiments performed on the freshly extracted precursor fluid and analysed by FT-IR spectroscopy showed that ACC is the main constituent and is probably stabilised by macromolecules such as proteins and sugars. Furthermore, the Debye–Scherrer diffraction experiments showed that the carbonate phase present in this fluid remains stable as ACC for more than a week. All these features are indicative of this entire process being biologically controlled by the earthworms. The analysis of the amorphous structure factor of this ACC indicates that these transformations are preceded by short-range order modifications of the amorphous precursor phase. Ó 2008 Elsevier Inc. All rights reserved.

1. Introduction Calcium carbonate formation in the earthworms’ calciferous glands is a remarkable case of biomineralization. Furthermore, although the particular mechanism involved in the production of calcium carbonate by earthworms (Annelida, Oligochaeta) remains unknown, the catalytic role of carbonic anhydrase has been anticipated (Clark, 1957). The biological functionality of these glands in the family Lumbricidae is also intriguing. From the early work by Darwin (1881), several hypotheses have been proposed, namely pH buffering of the blood and the ingested plant material, respiratory functions, egg formation or simply spurious mineralization (for example, see reviews by Robertson (1936) and Piearce (1972). More recently, studies have mainly focused on the morphological characterisation of the solid calcareous concretions (see also Canti and Piearce, 2003). However, little is known about the * Corresponding author. E-mail address: [email protected] (M.J.I. Briones). 1047-8477/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.jsb.2008.02.007

crystallization process itself and, in particular, in relation to those aspects concerning the nucleation mechanism and the sequence of the polymorphic carbonate phases. The formation of the stable crystalline polymorphs of calcium carbonate, aragonite and calcite is a common process in biological systems, but the less stable crystalline phases, such as vaterite and hydrocalcite (Gauldie et al., 1997; Estroff et al., 2003), can also be present in some extent. More recently, biomineralization studies (Raz et al., 2002, 2003; Weiss et al., 2002; Addadi et al., 2003; Politi et al., 2004) have shown that several organisms from different phyla produce amorphous calcium carbonate (ACC). Despite the fact of this phase being thermodynamically unstable and occurring relatively infrequently in inorganic systems, it seems to play a crucial role in biological mineralization by acting as a precursor phase of more stable polymorphs. Its high solubility also provides a source of calcium carbonate, which can be promptly stabilised or disestablished when necessary and according to the metabolic and functional requirements of the organisms. As a general rule, ACC is only stable during the lifetime of the animal, and its

L. Gago-Duport et al. / Journal of Structural Biology 162 (2008) 422–435

characterisation is difficult because it usually requires performing in situ experiments. Its amorphous nature also makes the structural characterisation difficult, and the application of more complex analytical techniques are necessary to provide enough information on the short-range order. Recent studies reporting variations in the pair distribution function of ACC (Becker et al., 2003; LeviKalisman et al., 2000; Lam et al., 2007) suggest that this phase can rapidly modify its local order through changes in the number of atoms surrounding the calcium ion, and consequently, can play an important role in the formation of a particular crystalline phase. In this study, we investigated the microstructural transformations that take place during the carbonate formation inside the calciferous gland of the earthworm species Lumbricus friendi Cognetti. First, we identified the presence of ACC by FT-IR, and then we carefully followed the different evolution stages of calcium carbonate in the precursor fluid prior to the formation of the crystalline phases by performing in situ XRD experiments. This was achieved by using a goniometer with Debye–Scherrer geometry and quantifying the changes in the amorphous structure factor with time, as an indication of the modifications occurring in the local ordering. Finally, we analysed the solid concretions from both morphological and microstructural points of view by means of FE-SEM and HRTEM. To our knowledge, these are the first direct structural investigations of the different phases involved in the biomineralization process occurring in the earthworm’s calciferous gland at the atomic level. This information is of great relevance, in particular in the early stages where the recognition of different phases based on purely morphological considerations is very complicated. This is due to the fact that a great number of carbonate polymorphs happen to grow in such a way under determined conditions so that identical forms are produced.

2. Materials and methods 2.1. Collection of the ‘milky fluid’ and solid concretions Earthworm specimens obtained from our laboratory cultures were dissected in deionised water to obtain both the ‘milky fluid’ (precursor fluid) (MF) and the mineral concretions secreted by the glands. The precursor fluid was collected using a plastic syringe and placed in sealed vials to perform X-ray diffraction (XRD) and Fourier Transform-Infrared spectroscopy (FT-IR) determinations. Two different fluid samples were obtained on the basis of their distinct density and their location: (i) low density milky fluid (LDMF) collected from the oesophageal pouches and (ii) high density milky fluid (HMDF) from the oesophagus lumen. In addition, the solid concretions (SC) stored in the oesophageal pouches were also carefully picked up using tweezers. Both liquid and solid samples were analysed separately using different technical procedures. 2.2. In situ analyses of the MF by FT-IR and XRD The aggregation state of the calcium carbonate present in the two types of MF samples was characterised by means of two different techniques: X-ray scattering and FT-IR. In the case of XRD, the absence of Bragg reflections in the scattering patterns was considered to be diagnostic of an amorphous state of the samples (Klug and Alexander, 1954; Cullity, 1976). In this case, the diffraction profile is characterised by only the presence of a diffuse scattering arising from the short-range ordering in the sample. FT-IR was performed on a Nicolet 6700 with a resolution of 0.5 cm 1. Prior to FT-IR analyses, samples were pressed into KBr pellets. The FT-IR vibration spectra were used as a precise fingerprint of the amorphous aggregation state of the samples based

423

on the specific variations in the vibration spectra of the ACC compared to those of the crystalline carbonate phases (Beniash et al., 1997; Politi et al., 2004). The presence of a broad band at 1084 cm 1 corresponding to the m1 mode of the carbonate oxy-anion is of particular importance since this vibration mode is connected to the orientation disorder of the CO3 @ and is not FT-IR active in the crystalline phases of calcium carbonate. The simultaneous presence of the m1 (1084 cm 1), m2 (866 cm 1) and m3 (1420– 1470 cm 1) bands together with the absence of the m4 mode (714 cm 1) are, therefore, the diagnostic features employed in the identification of ACC. FT-IR analyses were also applied to identify the presence of organic macromolecules in both the MF and SC concretions by analysing the decarbonated samples. In this case, the presence of the 1655 cm 1 band characteristic of amide groups was attributed to proteins, whereas the bands at 1100, 1560 and 1630 cm 1 are indicative of the presence of sugars and carboxyl groups. Furthermore, in order to investigate the stability of the amorphous aggregation state of the MF we performed time-elapsed XRD experiments. Samples of freshly extracted MF were injected into Lindemann capillaries, hermetically sealed and immediately placed in the goniometer. A Debye–Scherrer diffractometer equipped with a Coupled Charge Device (CCD) detector (Bruker SMART-CCD 1000) and Mo Ka radiation (k = 0.071073 nm) was employed to obtain the diffraction patterns. Experiments were carried out at 2h = 7°–90° to give a maximum value of the momentum transfer Qmax of 1.2 nm 1. Readings were taken at the start of the experiment and after 1 h intervals during the first day (24 h) and then daily for a week, with 10 min counting time for each individual spectrum. Raw data from the CCD detector were converted to lineal spectra by radial averaging of the Debye–Scherrer rings derived from the CCD frames. The GADDS software package from Socabim was used for this purpose. Additionally, the spectrum of the empty sample container (i.e. the Lindemann capillary) was obtained separately under the same conditions and its contribution to the scattering (corrected for absorption) was subtracted from the sample. An application of the Rietveld general procedure, specifically developed for microstructural analysis of the amorphous and disordered phases, was used to model the profile broadening. This method allows for the determination of the maximum correlation length of the scattering domains by approximating the amorphous phase to a nano-crystalline solid (i.e. when the long-range order is lost). Modelling the amorphous structure factor was achieved under the assumption that its structure would be approximated from that of its crystalline polymorphs (when available). Initially, fits were obtained using vaterite, aragonite and calcite as structural models. The best result, in the least square sense, was obtained by refining calcite in the R-3m super-group instead of the usual R-3c and thus, reducing the c-axis length by half. With this procedure, the CO3 @ groups were allowed to rotate freely and to become equivalent. However, although this model is useful in characterising the maximum length of the coherent diffracting domains, it is limited when trying to describe the local atomic ordering of the amorphous structure. This is the consequence of the fitting model being constrained to crystallographic structural parameters (i.e. only crystallographic positions are permitted). For this reason, the modelling of the amorphous stages is only based on the coherence losses, which are attributable to microstructural effects (microstrain and very small ‘crystalline’ sizes). Therefore, the Reverse Monte Carlo Method (RMC) was employed to characterise the diffuse scattering of both the completely amorphous samples and the partially crystallized ones. In RMC, the atomic coordinates for a certain set of ions are non-periodically arranged in a simulation box and varied randomly. The optimal configuration is achieved when the total calculated powder structure factor S(Q)

424

L. Gago-Duport et al. / Journal of Structural Biology 162 (2008) 422–435

fits well with the experimental one, according to the least-squares minimization. By using this fitted configuration it is possible to calculate both the partial and total pair correlations functions G(r) without performing the inverse Fourier transformation, and consequently, there is no need for using high Qmax values in the diffraction spectrum. RMCPOW, a crystalline version of RMC, was used for separate modelling of both the Bragg and diffuse scattering in those XRD patterns where the sample was partially crystalline (Mellergard and McGreevy, 1999, 2000). For the RMC simulations, we used a box equivalent to 3  3  1 (c-the long axis) single cells of the calcite structure. Normalisation and reduction of the raw XRD data to electronic units was performed using the RAD2003 code (Petkov, 1989) followed by application of the minimum interatomic distances and coordination constraints. The smallest distance for two atoms to be allowed to get close to each other was based on the structure of the crystalline carbonates and on previous data of molten carbonates (Kohara et al., 1999). The nearest distances of 0.37, 0.1, 0.2, 0.24 and 0.2 nm were used for the CAC, CAO, OAO, CaACa and OACa bonds, respectively. In order to model the CO3 @ triangular group the coordination constraints consisted of each carbon atom being coordinated with three oxygen atoms and every oxygen atom with two other oxygen atoms. 2.3. Mesoscale and microstructural analyses of the SC Morphological and compositional characteristics of the mineral phases involved in the calcareous secretions were analysed by field emission scanning electron microscopy (FE-SEM). Image analyses were performed using a JEOL JSM-6700F electron microscope. Elemental analyses and mapping of the samples were performed by EDS (Oxford Inca Energy 300). Transmission electron microscopy (TEM) and selected area electron diffraction (SAED) were used to structurally characterise the carbonate phase involved in the SC. Furthermore, the lattice fringes images obtained from the high resolution-TEM (HREM) were used to analyse any possible structural heterogeneities induced by the coexistence of different carbonate phases. A JEOL-2010 microscope with an accelerating voltage of 200 kV was used in the acquisition of both the HREM images and SAED patterns. Voltage, centre alignment, correction of the objective astigmatism and focusing were all performed on an adjacent patch of the carbon film. The area of interest was then returned to the focal point under the beam. Preparation of the SC samples for TEM was performed according to Rodriguez-Navarro et al. (2007). Briefly, fragments of the SC were firstly powdered and washed in ethanol. Then, one drop of this suspension was collected on a carbon-coated holey film, air-dried and mounted onto a copper grid for TEM measurements. Because mechanical treatments could induce some phase transitions in the carbonate samples (Criado and Trillo, 1975), no mechanical grinding was applied during the sample manipulation and instead, disaggregation was achieved by sonication for 20 min. The moderate voltage values employed (200 Kv) were considered to be adequate for preventing any crystallization or lattice distortion derived from the radiation beam. Previous studies suggest that this kind of induced re-crystallization usually takes place at voltage values in the order of 400 kV (e.g. Ness et al., 1990; Nassif et al., 2005). This aspect was further checked by performing repeated analyses of the same sample, and no observable phase transformations attributable to beam radiation were detected. Indexing and zone axis determinations of the single crystalline SAED patterns and 2D-FFT diagrams derived from the lattice fringes images were performed with the EMS software (Stadelmann, 1987). In order to obtain a quantitative determination of the crystalline and amorphous contribution in those samples with spherulithic morphology and to enable direct comparisons with usual diffractometers, the powder SAED diagrams were trans-

formed into lineal spectra and converted to angular values of Cu Ka radiation (k = 0.15405 nm). All of these transformations were done using the process-diffraction code (Lábár, 2000), which enables a fine refinement of the SAED centre, the correction of the astigmatism and the full indexing of the patterns obtained. Prior to the Rietveld refinements, the line diagrams were corrected against the instrumental background emitted by the sample holder (i.e. scattering from the empty microscope camera and the carbon grid). Rietveld analyses were also employed in the quantitative phase analysis of the lineal patterns. All calculations were performed with the Fullprof-2000 software (Rodríguez-Carvajal et al., 1987). Refinements were achieved in two steps: firstly, the cell, the profile parameters and the detector zero point were carefully refined using the Le Bail method (Le Bail, 1992) ; secondly, the scale factor of every crystalline phase was refined following the conventional Rietveld procedure. Electron scattering factors, instead of X-ray factors, were used in the calculations of the amplitudes. The refinements were performed within the following space groups and cell parameters (in nm): orthorhombic Pnma, (a = 0.5741, b = 0.4961, c = 0.7967) for aragonite; hexagonal P63/mmc (a = 0.4130, b = 0.8490) for vaterite and rhombohedral R-3c (a = 0.4990, c = 1.7061) for calcite. The hexagonal description of the rhombohedral calcite cell was employed in the generation of the Miller indexes. This procedure, based on Rietveld treatment, allowed for the obtaining of the relative contributions (wt%) of the crystalline phases involved in the diffraction pattern of the individual spherulites. In order to give a quantitative evaluation of the amorphous content in each sample, an additional analysis was done separately in which the Polymer Crystalline Index (PCI) was calculated using the Ruland-Vonk method (Ruland, 1961; Polizzi et al., 1990). The crystalline ratio was calculated according to the expression PCI = 1 (If Ic/Ic), where If is the integrated intensity of the full pattern after subtraction of the instrumental background and Ic is the sum of the integrated intensities for all crystalline phases, as determined by the Rietveld analysis. Thus, very low values of PCI are indicative of the sample being close to an amorphous state. 3. Results and discussion The earthworm’s calciferous gland in the genus Lumbricus consists of a pair of oesophageal pouches located in segment X, which open to the oesophagus lumen, and a pair of glandular enlargements in segments XI and XII (Fig. 1A). The secretory cells of the glandular region produce a concentrated suspension of calcium carbonate (CaCO3), commonly named as ‘milky fluid’ (MF) (Fig. 1B), which passes forward to the oesophageal pouches where it precipitates as concretions of calcium carbonate (SC), which are then released to the gut and finally into the soil (Fig. 1C). 3.1. FT-IR analyses of the MF and SC The FT-IR spectra of four samples of the precursor fluid (three LDMF plus one HDMF samples) and of one SC sample are shown in Fig. 2A. The three first spectra correspond to the three LDMF samples, whereas the fourth and the fifth ones at the top of this figure represent the HDMF and SC, respectively. It can be seen that, in addition to the peak variations due to the vibration of water molecules at 1641 and 3350 cm 1, there are remarkable changes in the region around 1600–1300 cm 1 where the calcium carbonate shows its strongest IR active vibration of the CO3 2 ion (m3) at its asymmetrical in-plane stretch. In the three represented spectra of the LDMF, the intensity peak around 1420 cm 1 is relatively broad when compared to the typical band found in the crystalline carbonate. Furthermore, this carbonate band of the LDMF appears to be partially convoluted with some other peaks, probably due to

L. Gago-Duport et al. / Journal of Structural Biology 162 (2008) 422–435

425

Fig. 1. The calciferous gland of Lumbricus friendi (50). (A) View of the two pairs of glandular enlargements in segments XI and XII and the anterior oesophageal pouches in segment X; (B) precursor fluid, (C) the calcium carbonate granules (SEM micrograph).

the organic matrix (see below). In addition, a sharp peak appearing at 1395 cm 1 can be assigned to the asymmetrical stretching vibration band of the bicarbonate HCO3 group (Nickolov et al., 2003).

Fig. 2. FT-IR spectra of: (A) three LDMF (bottom), one HDMF and one SC (top) samples collected from the calciferous gland; (B) the 1200–600 cm 1 region showing the m2, m4 and m1 carbonate bands for LDMF, HDMF and SC; (C) LDMF sample (i) and the decarbonated one (ii) to evidence the organic component.

426

L. Gago-Duport et al. / Journal of Structural Biology 162 (2008) 422–435

At these early stages the solution is probably slightly supersaturated in calcium carbonate due to the low concentration of the CO3 @ oxy-anion, and thus, the broadening of the carbonate asymmetrical band could probably reflect the coexistence of ion pairs and hydrated carbonate ions. With increasing carbonate concentration, this band shifts to lower wave numbers and increases its intensity. The presence of a precipitate of ACC is always marked by the appearance of a band at 866 cm 1 originated from the (m2) carbonate out-of plane bending and the absence of the (m4) vibration mode (Fig. 2B), together with the appearance of a broad band centred at 1081 cm 1, which can be assigned to the symmetrical stretching mode (m1) of the carbonate ion, although organic components could also contribute to this peak. This band is not IR active in crystalline carbonates, and its presence is the consequence of the loss of symmetry due to the presence of the amorphous state of the carbonate phase in the sample (Beniash et al., 1997). Unlike the LDMF samples, in the spectrum of the HDMF (Fig. 2A), both the m3 and m2 carbonate bands considerably increased in intensity, even though the HCO3 peak at 1395 cm 1 was still present. The m3 maximum forms a wide band centred at 1470 cm 1, but a very small peak corresponding to the m4 vibration mode is also observable. The high m2:m4 intensity ratio (Im2/ Im4 = 6.65) and the presence of the m1 band suggest that the solid phase in the sample is almost ACC, although the splitting of the m4 band into two values at 710 cm 1 and 745 cm 1 is consistent with a partial transformation to vaterite (Dupont et al., 1997). The top spectrum (Fig. 2A), corresponding to the SC sample, shows the typical features of the crystalline calcite, with a sharp

peak at 1420 cm 1; however, the splitting of the m4 band at 710 cm 1 into two additional small peaks could also be indicative of the presence of small amounts of other polymorphs, possibly vaterite and/or aragonite in combination with calcite. The IR is not a very sensitive technique to detect small amounts of crystalline polymorphs and therefore, these aspects were investigated further using HREM data. Furthermore, in order to obtain additional information regarding the organic fraction, we obtained the spectrum of the demineralised sample of the LDMF by acidification with diluted CLH ((ii) in Fig. 2C). The comparison with the freshly extracted MF sample ((i) in Fig. 2C) shows an intense absorption band in the region of 1660– 1000 cm 1. The presence of proteins is characterised by the two amide signatures, i.e. 1654 cm 1 of the amide I and 1553 cm 1 of the amide II (Rautaray et al., 2003), and the broad intense band at 1100 cm 1 corresponds to sugars (Aizenberg et al., 2003). In addition, the main signatures of carbonate at 866 cm 1 (m2), 1081 cm 1 (m1), 1420 cm 1 (m3) and the strong reflection at 1395 cm 1 have all disappeared. 3.2. Mesoscale characterisation of the solid phases The morphological characterisation of the SC using FE-SEM showed that nearly all the morphologies detected could be classified into spherulithic, mulberry-like aggregates and rhombohedral crystals (Politi et al., 2004). The spherulithic deposits mainly consisted of ovoid to spherical particles varying from 1 to 5 lm in diameter (Fig. 3A). These globules showed either smooth or rough

Fig. 3. Main morphological types of the calcium carbonate phases found in the calcareous concretions: (A) spherulithic deposits (the two different subtypes of spherulites described in the text, smooth and rough, are denoted as ‘sm’ and ‘rg’, respectively); (B) one mulberry-like aggregate; (C) detail of the surface of the mulberry aggregates at high magnification; (D) surface of the rhombohedral calcite crystals showing their characteristic growth and twins.

L. Gago-Duport et al. / Journal of Structural Biology 162 (2008) 422–435

surfaces (sm and rg in Fig. 3A), although the reason for these different types of surfaces remains unclear (Yamamoto et al., 1985; Jada and Pefferkorn, 2000). Coalescence between two or more spheres was occasionally observed leading to the formation of larger aggregates with flat faces, and possibly to the ultimate formation of rhombohedral calcite crystals. Furthermore, intermediate transformation stages leading to the incipient formation of rhombohedral crystals were also observed from both the globular aggregates and the mulberry-like forms. Occasionally, some disc-like shapes were also present. The mulberry-like aggregates (Fig. 3B) were irregular in shape, with rough surfaces and, at higher magnification, they were found to be formed by a great density of nucleation islands (Fig. 3C). In addition, the transformation inside the mulberry aggregates into a single crystal calcite rhombus was often observed. Finally, isolated rhombohedral calcite crystals showing irregular sizes ranging between 10 and 50 lm and exhibiting an important number of growth steps and twins were also present (Fig. 3D). 3.3. Microstructural analyses of the SC The microstructural analyses of the SC by means of TEM, HREM, SAED and FFT, allowed a better characterisation of the different calcium carbonate structures involved in each morphological type and provided very valuable information on the aggregation state at the micro and nanoscale levels. The TEM analyses (Fig. 4A) showed the existence of distinct evolutionary stages of the spherulithic transformation from ACC to calcite single crystals, ranging from either smooth or rough spherulites, through intermediate morphologies where rhombohedral calcite crystals grow on the surface of the spherulite, to the final calcite single crystals. The SAED patterns of the smooth spherulites (Fig. 4B) even at high magnification did not display the presence of any crystalline phase. This fact was also confirmed by their diffused diffraction pattern. In contrast, the rough spherulites (Fig. 4C) showed a central diffuse ring derived from the amorphous component together with crystalline

427

spots from a polycrystalline aggregate with the characteristic d-spacing of the vaterite lattice, i.e. (10 0)  0.35, (10 1)  0.33, (1 0 2)  0.27 and (10 4)  0.18 nm. The presence of calcite in the SAED patterns is clearly marked by the appearance of the characteristic sharp reflections at (1 0 4) = 0.303 and (11 0) = 0.249 nm. Furthermore, in order to quantitatively determine the contribution of every phase involved in the different types of spherulites, SAED diagrams corresponding to individual particles were transformed into their corresponding lineal diffraction spectra by performing the radial integration of the ED diffraction rings (Fig. 5A). The quantitative phase analyses of the crystalline phases was performed using the Rietveld method (Fig. 5B), and the ratio of the integrated intensities of the amorphous to crystalline phases was calculated and used in the determination of the crystallinity index following the Ruland method (Table 1). These results (Fig. 5A) showed the total absence of Bragg’s reflections in the upper spectrum ((i) in Fig. 5A) which is a strong indication of the smooth spherulites being mainly constituted of ACC. The diffraction profile in this pattern is dominated by diffuse scatterings and has the characteristic profile shape of the ACC, with a broad band centred at 0.31 nm and three small shoulders at higher diffraction angles. In the following steps of the spherulite transformation ((ii) and (iii) in Fig. 5A), only reflections corresponding to vaterite were found yet with an important amount of the amorphous phase, which is responsible for the characteristic background observed in the figure. In the next diagram ((iv) in Fig. 5A), calcite is mostly present although some vaterite might co-exist, demonstrated by the appearance of a hump at the right side of the (1 0 4)c reflection that should correspond to the (1 0 2)v reflection. In the last transitional step ((v) in Fig. 5A), only the calcite peaks are detected. Microstructural aspects of spherulites obtained by means of HREM analyses (Fig. 6) also confirmed the presence of an initial ACC phase but also a variable degree of transformation of this ACC to vaterite and/or calcite. Fig. 6A shows an amorphous region with some minor amounts of crystalline phases in the

Fig. 4. (A) TEM images showing several steps of the morphological variations of the spherulites during their transformation from ACC to vaterite and finally to calcite. Representative SAED patterns of each situation are shown in the inset figures. The d-spacings of 0.35 nm, 0.27 nm and 0.18 correspond to (1 0 0), (1 0 2) and (1 1 2)/(1 0 4) vaterite lattice planes and the d-spacings of 0.30 and 0.25 nm to (1 0 4) and (1 1 0) of calcite. (B and C) Detail of the smooth (B) and rough (C) interfaces with the corresponding SAED patterns. Note that smooth spherulites are almost amorphous, whereas the rough interfaces are characterised by the presence of vaterite.

428

L. Gago-Duport et al. / Journal of Structural Biology 162 (2008) 422–435 Table 1 XRD-Rietveld quantitative phase analysis of spherulites DRX-pattern in (Fig. 5A)

Calcite (wt%)a

Vaterite (wt%)a

Crystallinity indexb

(i) (ii) (iii) (iv) (v)

—

— 97.49 (3) 94.10 (3) 10.45 (3) —

0.01 0.31 0.56 0.73 0.81

2.51 (2) 5.90 (2) 89.55 (3) 100

a (wt%) are referred to the total amount of crystalline phases. Esd’s are in parentheses. b From the Ruland-Vonk method.

Fig. 5. (A) XRD lines obtained by radial integration of the ED diffraction rings for individual spherulites with several degrees of transformation to calcite. Identified phases in each pattern together with the d-spacing and lattice indexes are indicated. The more intense reflections from the vaterite and calcite are marked by dotted lines. (B) Typical output plots resulting from the Rietveld procedure applied to spectra (iii) and (v) of (A). Note that the X-axes in the diagrams have been converted to CuKa radiation values to provide a direct comparison with the usual XRD powder patterns.

form of small crystalline clusters (Ø  5 nm) and marked by the presence of lattice fringes. The FFT analyses of the small area enclosed by a circle gives d-spacing values of 0.27 and 0.18 nm with angle of 28° between them. These features are consistent with the (1 0 2) and (1 1 2) lattice planes of the vaterite viewed along the [ 2 0 1] zone axis. In all spherulites analysed, vaterite

clusters were poorly crystallized and randomly oriented in between the amorphous regions. On some occasions, the presence of calcite together with vaterite was also observed (Fig. 6B). In this case, both phases formed a mosaic of crystalline nanoparticles with an average size of 5–10 nm and separated by amorphous domains. The 2D-FFT analyses of the whole micrograph (top inset) resulted in a powder-like pattern consisting of spots of the (1 0 2)v and (1 1 0)c lattice fringes and corresponding to the observed nanocrystalline domains with different orientations. Oscillating lattice fringes and edge dislocations were commonly observed for vaterite nanocrystals. The SAED patterns of these areas (bottom inset) showed the presence of Oblique Texture Electron Diagrams (OTED) (Drits and Tchoubar, 1990), a typical feature of the lamellar phases with a tubostratic disorder of the rotational type. In sum, all these aspects exposed above strongly suggest that ACC is the first carbonate phase that precipitates during the formation of the spherulites from the precursor fluid (MF). Upon aging, this initial amorphous phase evolves to more stable polymorphs according to the sequence ACC ? vaterite ? calcite (polycrystalline) ? calcite (single crystal). Aragonite was never detected in any of the spherulites (smooth and rough) investigated. The random distribution of the crystalline clusters of vaterite in relation to both size and crystallographic orientation appears to be indicative of a stochastic nucleation process of this metastable phase occurring in the amorphous matrix. On the other hand, the transformation from ACC or vaterite to calcite preferably takes place at the spherulite surfaces (Fig. 6B), suggesting that this transformation is mediated by the solvent, following an Ostwald ripening process. Interestingly, a completely different crystallisation behaviour was observed for the disc-like shaped particles. In this type of particles, the radial distributions of the crystal from the disc centre are evident (Fig. 7A). A more detailed picture of the aggregation characteristics of these structures was performed by means of HREM and 2D-FFT analyses (Fig. 7B), which revealed that the outer surface of these particles is clearly amorphous, whereas towards the centre of the disc the structure is almost crystalline as demonstrated by its well-developed lattice fringes. Measured d-space of these lattice fringes gave a value of 4.22 Å, which is consistent with the aragonite lattice plane (0 1 1). In the case of the mulberry-like forms, there was an important amorphous component also present, but mixed regions, in which ACC and aragonite coexist, were frequently observed too. These aragonite crystals appear to be arranged in the form of single-crystal nano-domains with an average size of 30 nm, and they were embedded by an amorphous matrix. In most cases, aragonite exhibited single-crystal electron diffraction patterns (Fig. 8) although some distortion of the lattice fringes was observed, probably as a result of the microstrain in the aragonite orthorhombic cell. This feature was also appreciable in the FFT (inset) where the appearance of double spots are associated with the most intense reflections of the [10 0]A zone axis. Sometimes, instead of the formation of large crystalline domains, a spread of nanocrystals was observed in those areas

L. Gago-Duport et al. / Journal of Structural Biology 162 (2008) 422–435

429

Fig. 6. (A) ACC region of the spherulite with some minor contributions of crystalline phases in the form of small clusters (inset circle: SAED diagram of a vaterite single crystal represented along the [ 2 0 1] zone axis). (B) Area of the spherulite showing vaterite nano-crystals partially transformed to calcite (calcite domains tend to be oriented with lattice planes parallel to the spherulite’s edge and the regions between vaterite domains are separated by amorphous areas). Note that vaterite exhibits oscillating lattice planes, which are characteristic of disordered lamellar phases. The SAED diagram shown in the inset at the bottom of this figure corresponds to a single crystal of vaterite viewed along [0 1 0] showing these features. Note the formation of oblique texture patterns (OTED) marked by arrows, which are indicative of a rotational disorder. The top inset shows the FFT analysis of the whole micrograph revealing a powder-like pattern.

where aragonite was starting to transform to calcite (Fig. 9). The aragonite domain in the figure is oriented along the [11 1] zone axis (top inset), and it is surrounded by a region where the transformation to calcite has been nearly accomplished. 3.4. Temporal stability of ACC in the MF and preliminary structural characterisation One important question that needs to be addressed here is whether the ACC is inorganically precipitated or whether it becomes stabilised by the organism. One possible way of resolving this question could be to determine the maximum length of time for ACC to remain stable without being transformed to any of the later crystalline phases. It is well known that the inorganic synthe-

sis of ACC results in a highly unstable product with an estimated induction time on the order of a few minutes before transforming to a crystalline phase (Sawada, 1997; Loste et al., 2003; Andreassen, 2005). In contrast, if the formation of ACC is induced biologically, it can remain stable for several days (Beniash et al., 1997; Lowenstam and Weiner, 1989; Faazt et al., 2004). In the case of the earthworms, the analyses made with FT-IR showed the presence of an important organic component that could play an essential role in the stabilisation of this ACC initially formed as a result of carbonation reactions. Therefore, in order to get a deeper knowledge of the transformations of the MF, we performed time-elapsed XRD experiments, and the scattering patterns obtained were corrected against the background of the Lindemann capillary and represented as a

430

L. Gago-Duport et al. / Journal of Structural Biology 162 (2008) 422–435

Fig. 7. (A) Low magnification TEM micrograph of a disc-like shape particle showing its internal radial texture. (B) HREM image of the same particle showing the distribution of the amorphous and crystalline zones. The 0.422 nm d-spacing measured in the lattice fringes is consistent with the aragonite lattice plane (0 1 1). The FTT analyses of the three regions are shown in the insets.

function of Q (the momentum transfer Q = 4p sin h/k, in nm 1). The lack of Bragg peaks (Fig. 10) demonstrated that the sample remained amorphous during the whole experimental period (8 days). Crystallization was finally induced by heating the sample at 60 °C overnight and resulted in the calcite formation, which confirmed that calcium carbonate was previously present in an amorphous state. This crystallization was, however, only partial as denoted by the important amount of amorphous background. Following this, the crystallization progressed with a decrease in the intensity of the characteristic diffuse scattering of the amorphous component, which clearly indicates that the formation of calcite results from recrystallization of the ACC. It is interesting to note that the observed inflections in the shape of the scattering profile of the amorphous phase (Fig. 10)

can be easily related to the clusters of Bragg peaks observed in crystalline carbonates. This is highlighted in the figure, where the three main oscillations in the profile of the ACC patterns centred at d = 0.33, 0.20 and 0.13 nm can be compared to the corresponding main groups of the Bragg reflections in the crystallized calcite. Furthermore, in order to characterise the microscopic aggregation state of the amorphous phase, we performed Rietveld analyses so that the maximum length of the coherent diffracting domains of the amorphous phase could be estimated. The integral breadth was calculated using the Scherrer expression (Klug and Alexander, 1954) and by refining the full scattering profile obtained from the Rietveld method (Le Bail, 1995). The instrumental resolution of the diffractometer was firstly calibrated using corundum as a standard and modelled according to the Cagliotti formula (Caglioti

L. Gago-Duport et al. / Journal of Structural Biology 162 (2008) 422–435

Fig. 8. Aragonite single-crystal nanodomains. The microscope beam is parallel to the [1 0 0] zone axis. Note the double spots in the FTT, which are indicative of misalignments between the different domains. Indexing was done in the Pnma space group (a = 0.5741, b = 0.4961, c = 0.7967 nm).

et al., 1958), which describes the angular resolution as a function of the diffraction angle. The best fit was obtained using the space group R-3m instead of R-3c and consequently, the parameter c that was employed was half of the calcite one (c = 0.856 nm 1). This structural model is in agreement with the results obtained with the FT-IR analyses where the presence of the m1 band is associated with a reduction in the symmetry imposed by the free rotation of the carbonate group in the disordered phase. The comparison of the experimental and calculated profiles obtained with the Rietveld analysis is shown in Fig. 11A, and includes the reflections

431

corresponding to the initial structural model. The maximum length of the coherent scattering domains obtained through this procedure provided a value of 0.64 nm, and confirmed that the sample is fully amorphous. The only ordered diffracting entities were small clusters with diameters ranging from the first to the second coordination shells. RMC modelling was further used to analyse the diffuse scattering of the ACC diffraction pattern. The RMCPOW code enables the deconvolution of the intensities of the Bragg reflections from those derived from the diffuse scattering that correspond to atoms in non-crystallographic positions, and thus allows for the calculation of the total G(r) and partial Gi-j(r) pair distribution functions in the real space. Data from the previous Rietveld analyses were used as starting parameters to generate a 3  3  1 supercell, which contained a fixed number of Cartesian coordinates of atomic positions. The results are shown in Fig. 11B, where (i), (ii) and (iii) correspond to the three XRD patterns displayed in Fig. 10, i.e. completely amorphous (i), start of crystallization (ii) and later stages of crystallization (iii). The pair distribution functions, obtained from the final atomic configuration using the RCM, were compared to those of the crystalline calcite (Martin et al., 2006). Although the characteristic correlation lengths of the calcite were present in all diagrams (Fig. 12), there were appreciable variations in shape. Thus, the total G(r) functions (Fig. 12A) exhibited a general tendency to broaden their characteristic correlation lengths as the sample became progressively amorphous, and in the case of ACC, there was a decrease in the characteristic correlation lengths greater than 0.37 nm. Furthermore, the partial G(r) functions showed that, in the amorphous phases, the bond distances of CaAO (i.e. 0.25 nm) and that of CaAC (i.e. 0.33 nm), which are typical of the crystalline calcite, are broadened and split into two maxima (Fig. 12B and C). Similarly, some other precise correlation lengths of the crystalline phase (Fig. 12B) tended either to merge into one single broad peak (e.g. 0.46 nm) or to disappear (e.g. 0.63 nm) in the amorphous state. These results, although preliminary, are in agreement with previous findings (Levi-Kalisman et al., 2000; Raymond et al., 2007) that suggest that the short-range order observed in the ACC is not unique and involves structural differences accounting for the various mechanisms involved in the formation and stabilisation of the different crystalline phases. Further studies using higher resolution

Fig. 9. Detailed region of a mulberry-like aggregate showing the coexistence of aragonite, calcite and ACC domains. The FFT analyses of the enclosed areas (circles) are displayed in two insets: the top inset shows aragonite viewed along the [1 1 1] zone axis and the bottom inset shows calcite viewed along the [1 2 0] axis zone.

432

L. Gago-Duport et al. / Journal of Structural Biology 162 (2008) 422–435

Fig. 10. Time-elapsed XRD experiments of the MF samples: evolution of the diffuse scattering intensity versus Q (the momentum transfer) showing no crystallization after 8 days (i.e. four upper spectra) and the induced crystallization. The Debye–Scherrer rings representing every aggregation state are shown at the top.

techniques are needed to provide a more detailed analysis of the local ordering of the amorphous calcium carbonate. 3.5. Suggested mechanism for the formation and storage of calcium carbonate by the earthworms Based on the analyses of both the SC and MF extracted from the earthworm’s calciferous gland, we can conclude that the formation of the carbonate phases in these organisms follows a complicated transformation sequence, starting from the initial liquid precursor (MF), followed by the formation of ACC, and then evolving to different crystalline phases (namely, vaterite, aragonite and finally calcite). In addition, the information provided by the Debye–Scherrer diffraction analyses of the MF clearly supports the idea of the MF being amorphous in its nature and remarkably stable in this aggregation state. All these results clearly confirm that ACC is the first carbonate phase involved in the mineralization process carried out by the earthworm’s calciferous gland. The presence of macromolecules in the precursor fluid (in our case amines and glycosylated products) could play a critical role in the stabilisation of this ACC by the worms, as revealed by the FT-IR analyses of the MF. This, together with the results from the XRD experiments showing that a long induction time is needed for its crystallization to more sta-

ble polymorphs, suggest that the formation and stabilisation of the ACC is biologically controlled by these organisms. After the precipitation of ACC, crystallisation takes place in the globular aggregates with the formation of vaterite, as revealed by the ED diagrams showing the coexistence of both phases in many sperulites. Further transformations to the more stable calcite phase occur at the smooth spherulite surface, which becomes rougher with time due to the formation of minute calcite crystals. Recrystallisation of the calcite crystals is the next step in this process and, in many cases, this occurs by aggregation of several spherules. This resembles an Ostwald-ripening process where the transition is driven by the variations in the free energy surface of the crystal, and it is the result of differences in the solubility of the bigger calcite rhombohedra and the amorphous globules. Although this is likely to be the most plausible transformation path, the formation of aragonite, as an intermediate phase, has also been recorded. In this case, the HREM analyses of the disc-like shapes revealed that the transformation into aragonite starts in the centre of the disc while the outer part remains amorphous until the crystallisation progresses to the periphery and forms the typical radial aggregates. Although the previous results clearly demonstrate that ACC is produced under the organism’s control and remains stable,

L. Gago-Duport et al. / Journal of Structural Biology 162 (2008) 422–435

433

Fig. 11. (A) Rietveld fit of the amorphous phase used to determine the maximum size of the coherent scattering domains of the MF samples following the Scherrer method. The reflections of the starting structural model (dotted areas) were based on the R-3m space group and calculated according to the instrument resolution. (B) RMC fits of those samples corresponding to ACC and various crystallization stages and showing the deconvolution between total and diffuse scattering.

434

L. Gago-Duport et al. / Journal of Structural Biology 162 (2008) 422–435

whether it is a random process resulting from inorganic constrains. This is an important point that must be further elucidated in order to fully understand the biological functionality of ACC in earthworms. 4. Conclusions The results from this study have clearly shown that the formation of crystalline polymorphs of calcium carbonate by earthworms is preceded by the precipitation of ACC and consequently, the process involved seems to be a true biomineralization process controlled by the organism. The high solubility of ACC is likely to be critical for temporal storage of calcium carbonate ions, which could then get dissolved and used elsewhere according to the organism’s requirements. Stabilisation of ACC is later achieved thanks to the presence of macromolecular constituents in the organic matrix, which could also be selectively used to promote the elimination of the carbonate excess by nuclei induction of the crystalline phases. Next, the successive appearance of a series of crystalline polymorphs follows, in general terms, the Ostwald steps rule (i.e. precipitation proceeds from the lesser stable to the more stable polymorph), although two different transformation paths have been observed: (i) ACC ? vaterite ? calcite and (ii) ACC ? aragonite ? calcite. The final formation of the macroscopic rhombohedra observed in the SC could presumably be attributed to a secondary Ostwald-ripening process through solvent–mediated transformations. From this point of view, mineralization in the oesophageal glands of the earthworms follows a similar path to the ACC precipitation and stabilisation observed in other carbonate mineralizing organisms (e.g. Raz et al., 2002; Politi et al., 2004), and the main difference, but also the most intriguing aspect, is that in the case of earthworms the final resulting SC does not seem to have any biological function since they get finally excreted and released into the soil. Further work is needed to fully elucidate whether the formation of a particular polymorph is also controlled by the organism or is kinetically driven by inorganic processes. Acknowledgments We are in debt to Ms. E. López for performing the FT-IR analyses of the samples. Thanks are also due to Prof. S. Weiner for valuable comments that greatly improved the quality of the manuscript and to three anonymous referees for their constructive discussions. References

Fig. 12. Pair distribution function resulting from RMC modelling of the diffuse scattering of the XRD patterns of the previous figure and compared to the perfect crystalline calcite: (A) total pair distribution function; (B) and (C) partial pair distribution functions for CaAO and CaAC, respectively.

it remains unclear whether the secondary crystallization of a particular polymorph is also directed by the organism or

Addadi, L., Raz, S., Weiner, S., 2003. Taking advantage of disorder: amorphous calcium carbonate and its roles in biomineralization. Adv. Mat. 15, 959–970. Aizenberg, J., Weiner, S., Addadi, L., 2003. Coexistence of amorphous and crystalline calcium carbonate in skeletal tissue. Connect. Tissue Res. 44, 20–25. Andreassen, J.P., 2005. Formation mechanism and morphology in precipitation of vaterite nano-aggregation or crystal growth. J. Cryst. Growth 274 (1–2), 256– 264. Becker, A., Bismayer, U., Epple, M., Fabritius, H., Hasse, B., Shi, J., Ziegler, A., 2003. Structural characterisation of X-ray amorphous calcium carbonate (ACC) in sternal deposits of the crustacean Porcellio scaber. Dalton Trans. 4, 551–555. Beniash, E., Aizenberg, J., Addadi, L., Weiner, S., 1997. Amorphous calcium carbonate transforms into calcite during sea urchin larval spicule growth. Proc. R. Soc. Lond. B 264, 461–465. Caglioti, G., Paoletti, A., Ricci, F., 1958. Choice of collimators for a crystal spectrometer for neutron diffraction. Nucl. Instrum. Methods 3, 223–228. Canti, M.G., Piearce, T.G., 2003. Morphology and dynamics of calcium carbonate granules produced by different earthworm species. Pedobiologia 47, 511–521. Clark, A.M., 1957. The distribution of carbonic anhydrase in the earthworm and snail. Aust. J. Sci. 19, 205–207. Criado, J.M., Trillo, J.M., 1975. Effects of mechanical grinding on the texture and structure of calcium carbonate. J. Chem. Soc., Faraday Trans. 71, 961–966. Cullity, B.D., 1976. X-ray Diffraction. Addison-Wesley Publishing, Reading, MA. Darwin, C., 1881. The formation of Vegetable Mould, through the Action of Worms, with Observations on their Habits. John Murray, London.

L. Gago-Duport et al. / Journal of Structural Biology 162 (2008) 422–435 Drits, V.A., Tchoubar, C., 1990. X-ray Diffraction by Disordered Lamellar Structures. Springer, Berlin. Dupont, L., Portemer, F., Figlarz, M., 1997. Synthesis and study of a well crystallized CaCO3 vaterite showing a new habitus. J. Mater. Chem. 7, 797–800. Estroff, L.A., Leiserowitz, L., Addadi, L., Weiner, S., Hamilton, A.D., 2003. Characterization of an organic hydrogel: a cryo-TEM and X-ray diffraction study. Org. Biomol. Chem. 2, 137–141. Faazt, M., Gröhn, F., Wegner, G., 2004. Amorphous calcium carbonate synthesis and potential intermediate in biomineralization. Adv. Mater. 36, 996–1000. Gauldie, R.W., Sharma, S.K., Volk, E., 1997. Micro-Raman spectral study of vaterite and aragonite otoliths of the coho salmon, Oncorhynchus kisutch. Comp. Biochem. Physiol. A Comp. Physiol. 118, 753–757. Jada, A., Pefferkorn, E., 2000. Smooth and rough spherical calcium carbonate particles. J. Mater. Sci. Lett. 19, 2077–2079. Klug, H.P., Alexander, L.E., 1954. X-ray Diffraction Procedures for Polycrystalline and Amorphous Materials. Wiley, New York. Kohara, S., Suzuyab, K., Ohnob, H., 1999. A reverse Monte Carlo study of molten lithium carbonate. Plasmas Ions 2, 79–83. Lábár, J.L., 2000. Process diffraction: a computer program to process electron diffraction patterns from polycrystalline or amorphous samples. In: Frank, L., Ciampor, F., (Eds.), Proc. EUREM 12, vol. III, Brno (Czech Republic), pp. 379–380. Lam, R.S.K., Charnock, J.M., Lenniec, A., Meldrum, F.C., 2007. Synthesis-dependant structural variations in amorphous calcium carbonate. Cryst. Eng. Commun. 9, 1226–1236. Le Bail, A. 1992. Extracting structure factors from powder diffraction data by iterative full pattern profile fitting. In: Prince, E., Stalick, J.K., (Eds.), Accuracy in Powder Diffraction II, Special Publication 846, National Institute of Standards and Technology, Gaithersburg, Maryland, USA, p. 213. Le Bail, A., 1995. Modelling the silica glass structure by Rietveld method. J. NonCryst. Solids 183, 39–42. Levi-Kalisman, Y., Raz, S., Weiner, S., Addadi, L., Sagi, I., 2000. X-Ray absorption spectroscopy studies on the structure of a biogenic ‘‘amorphous” calcium carbonate phase. J. Chem. Soc., Dalton Trans. 21, 3977–3982. Loste, E., Wilson, R.M., Seshadri, R., Meldrum, F., 2003. The role of magnesium in stabilising amorphous calcium carbonate and controlling calcite morphologies. J. Cryst. Growth 254, 206–218. Lowenstam, H.A., Weiner, S., 1989. On Biomineralization. Oxford University Press, New York. Martin, P., Spagnoli, D., Marmier, A., Parker, C., Sayle, D.C., Watson, G., 2006. Application of molecular dynamics DL_POLY codes to interfaces of inorganic materials. Mol. Simulat. 32 (12–13), 1079–1093. Mellergard, A., McGreevy, R.L., 1999. Reverse Monte Carlo modelling of neutron powder diffraction data. Acta Crystallogr., Sect. A: Found Crystallogr. 55, 783. Mellergard, A., McGreevy, R.L., 2000. Recent developments of the RMCPOW method for structural modelling. Chem. Phys. 261, 267–274. Nassif, N., Pinna, N., Gehrke, N., Antonietti, M., Jäger, C., Cölfen, H., 2005. Amorphous layer around aragonite platelets in nacre. Proc. Natl. Acad. Sci. USA 102, 12653– 12655.

435

Ness, S.E., Haywick, D.W., Cuff, C., 1990. Overcoming the problems of highresolution transmission electron microscopy of biogenic aragonite. Mineral. Mag. 54, 589–592. Nickolov, Zh.S., Ozcan, O., Miller, J.D., 2003. FTIR analysis of water structure and its significance in the flotation of sodium carbonate and sodium bicarbonate salts. Colloids Surf. A: Physicochem. Eng. Aspects 224, 231–239. Petkov, V., 1989. RAD, a program for analysis of X-ray diffraction data from amorphous materials for personal computers. J. Appl. Cryst. 22, 387. Piearce, T.G., 1972. The calcium relations of selected Lumbricidae. J. Anim. Ecol. 41, 167–188. Politi, Y., Klein, E., Arad, T., Weiner, S., Addadi, L., 2004. Sea urchin spine calcite forms via a transient amorphous calcium carbonate phase. Science 306, 1161– 1164. Polizzi, S., Fagherazzi, G., Beneditti, A., Battagliarin, M., Asano, T., 1990. A fitting method for the determination of crystallinity by means of X-ray diffraction. J. Appl. Cryst. 23, 359–365. Raymond, S.K., Lam, J.M., Charnock, A.L., Fiona, C.M., 2007. Synthesis-dependant structural variations in amorphous calcium carbonate. Cryst. Eng. Commun. 9, 1226–1236. Raz, S., Testeniere, O., Hecker, A., Weiner, W., Luquet, G., 2002. Stable amorphous calcium carbonate is the main component of the calcium storage structures of the crustacean Orchestia cavimana. Biol. Bull. 203, 269–274. Raz, S., Hamilton, P.C., Wilt, F.H., Weiner, S., Addadi, L., 2003. The transient phase of amorphous calcium carbonate in sea urchin larval spicules: the involvement of proteins and magnesium ions in its formation and stabilization. Adv. Funct. Mat. 13, 480–486. Robertson, J.D., 1936. The function of the calciferous glands of earthworms. J. Exp. Biol. 13, 279–297. Rodríguez-Carvajal, J., Anne, M., Pannetier, J., 1987. A system for time-resolved data analysis (Powder Diffraction Pattern). IIL Internal Report 87R014T. Institut Max Von Paul Langevin, VCH, Grenoble. Rodriguez-Navarro, C., Jimenez-Lopez, C., Rodriguez-Navarro, A., Gonzalez-Muñoz, M.T., Rodriguez-Gallego, M., 2007. Bacterially mediated mineralization of vaterite. Geochim. Cosmochim. Acta 71, 1197–1213. Ruland, W., 1961. X-ray determination of crystallinity and diffuse disorder scattering. Acta Cryst. 14, 1180–1185. Sawada, K., 1997. The mechanisms of crystallization and transformation of calcium carbonates. Pure Appl. Chem. 69, 921–928. Stadelmann, P.A., 1987. EMS—a software package for electron diffraction analysis and HREM image simulation in materials science. Ultramicroscopy 21, 131– 145. Weiss, I.M., Tuross, N., Addadi, L., Weiner, S., 2002. Mollusc larval shell formation: amorphous calcium carbonate is a precursor phase for aragonite. J. Exp. Zool. 293, 478–491. Yamamoto, H., Sakae, T., Schäfer, H., 1985. Analysis of vaterite microspherolith deposits on a pure cholesterol gallstone by X-ray diffraction, X-ray microanalysis and infrared absorption techniques. Virchows. Arch. [Pathol. Anat.] 405, 463–471.