Journal of Structural Biology 191 (2015) 165–174

Contents lists available at ScienceDirect

Journal of Structural Biology

journal homepage: www.elsevier.com/locate/yjsbi

Electron microscope analyses of the bio-silica basal spicule from the Monorhaphis chuni ⇑ Peter Werner a, , Horst Blumtritt a, Igor Zlotnikov b, Andreas Graff c, Yannicke Dauphin d, Peter Fratzl b a MPI of Microstructure Physics, Weinberg 2, D-06120 Halle (Saale), Germany b MPI of Colloids and Interfaces, Am Mühlenberg 1, D-14476 Potsdam, Germany c Fraunhofer Institute for Mechanics of Materials, 06120 Halle, Germany d Micropaléontologie, UFR TEB, Université P. & M. Curie, 75252 Paris Cedex 05, France article info abstract

Article history: We report on a structural analysis of several basal spicules of the deep-sea silica sponge Monorhaphis Received 3 March 2015 chuni by electron microscope techniques supported by a precise focused ion beam (FIB) target prepara- Received in revised form 16 June 2015 tion. To get a deeper understanding of the spicules length growth, we concentrated our investigation onto Accepted 18 June 2015 the apical segments of two selected spicules with apparently different growth states and studied in detail Available online 19 June 2015 permanent and temporary growth structures in the central compact silica axial cylinder (AC) as well as the structure of the organic axial filament (AF) in its center. The new findings concern the following mor- Keywords: phology features: (i) at the tip we could identify thin silica layers, which overgrow as a tongue-like fea- Biosilica ture the front face of the AC and completely fuse during the subsequent growth state. This basically Biomineralization Silica differs from the radial growth of the surrounding lamellar zone of the spicules made of alternating silica SEM lamellae and organic interlayers. (ii) A newly detected disturbed cylindrical zone in the central region of TEM the AC (diameter about 30 lm) contains vertical and horizontal cavities, channels and agglomerates, STEM which can be interpreted as permanent leftover of a formerly open axial channel, later filled by silica. FIB (iii) The AF consists of a three-dimensional crystal-like arrangement of organic molecules and amorphous Monorhaphis chuni silica surrounding these molecules. Similar to an inorganic crystal, this encased protein crystal is typified by crystallographic directions, lattice planes and surface steps. The h001i growth direction is especially favored, thereby scaffolding the axial cylinders growth and consequently the spicules’ morphology. Ó 2015 Elsevier Inc. All rights reserved.

1. Introduction 2008), toughening and a reduced stiffness (Woesz et al., 2006; Zlotnikov et al., 2013). For both classes a thin organic axial filament Bio-minerals may have a complex mineral matrix structure (AF) with a diameter in the lm range is a typical feature of the spi- with organic interlayers or inclusions. Such composite structures cules’ center. It contains proteins, so-called silicatein (Shimizu are hierarchically organized with building units ranging from the et al., 1998; Müller et al., 2009), which are involved in the enzy- macroscopic to the nanometer scale. As an example, the two matic synthesis of the silica. The contribution of collagen for the classes of siliceous – and hexactinellids – synthesis of silica is also under discussion (Ehrlich et al., 2008; build their skeletons of quite differently sized spicules. In the case Ehrlich, 2010). of demosponges, the mature spicules ultimately consist only of a The corresponding basic growth phenomena, including compact bio-silica cylinder with a central axial channel (Weaver bio-molecular processes as well as the morphologenesis, have been et al., 2010). In contrast, the spicules of hexactinellids are charac- intensively studied and widely understood in recent years (see, terized by an inner axial cylinder of more compact silica and a e.g., Wang et al., 2008, 2012). Nevertheless, several of detailed major outer lamellar structure of silica layers around it, which questions concerning structures and growth processes remain are separated by thin organic interlayers in the range of approxi- open. mately 50 nm. This laminar architecture generates improved The sponge Monorhaphis chuni is a well-known example of the mechanical properties, e.g., a fracture resistance (Miserez et al., hexactinellids. It represents one of the largest bio-silica structures (first described by Schulze, 1904). Its skeleton is built around a ⇑ giant basal spicule (GBS) of silica, which can reach a length up Corresponding author. to 2–3 m and a diameter up to approximately 1 cm. As an E-mail address: [email protected] (P. Werner). http://dx.doi.org/10.1016/j.jsb.2015.06.018 1047-8477/Ó 2015 Elsevier Inc. All rights reserved. 166 P. Werner et al. / Journal of Structural Biology 191 (2015) 165–174 introduction, Fig. 1 represents the basic morphological features of of a crystalline organization within the AFs was received by such a well-developed mature GBS of this , observed by transmission electron microscopy (Garonne et al., 1981). The scanning electron microscopy (SEM) and by light optical micro- existence of crystalline phases in several sponge species was scopy. The present paper deals specifically with structural investi- later proven by X-ray diffraction, too (Croce et al., 2007). gations of such tip segments where the length growth takes place. Therefore, a similar phenomenon could be expected for the The SEM micrograph in Fig. 1a shows the typical silica lamellae, hexactinellids. Indeed, in a recent work the authors could which have a thickness in the range of 2–10 lm. Large, mature show unambiguously by high-resolution transmission GBSs can contain up to several hundreds of such lamellae. The electron microscopy (TEM), energy-dispersive X-ray lamellar outer zone imbeds a compact, non-lamellar, inner axial spectroscopy (EDX) and X-ray diffraction that the AF in cylinder of silica (AC, in the insert below Fig. 1a). Fig. 1b and c show the spicule of M. chuni is actually a hybrid material where corresponding light-microscopic images of the tip region from the silicatein molecules are arranged on a regular such a spicule. These – and most of the following optical micro- body-centered tetragonal lattice encased in mesoporous sil- graphs – were taken in dark-field mode since this technique is ica (Zlotnikov et al., 2014). In which way corresponding more sensitive to inner structural details within the silica rod. results would refer to the formation of the GBS? Thereby Fig. 1b and c clearly show a light-blue cylinder, which cor- (iii) Moreover, even by optical microscopy one easily can detect responds to the AC. In the present sample, it has a diameter of a heavily disturbed cylindrical zone (DCZ) with a diameter of approximately 150–200 lm along the whole investigated segment approximately 20–30 lm (see inside the white rectangle in of the spicule (see cylindrical scheme in Fig. 1c). Fig. 1d). This zone is magnified in Fig. 1e, where the distur- While the outer lamellar zone of the M. chuni GBS has been bances are visible as small bright dots. What is the morphol- studied in detail including its bio-mechanical properties (e.g., ogy of this DCZ and in which way it could be attributed to Miserez et al., 2008; Mayer, 2009), till now no deeper analysis of the GBS growth? the morphology of the AC, e.g., by electron microscopy techniques, was performed up to the knowledge of the authors. Open questions Related to these aspects, the present study is concentrated onto concern, e.g., the following aspects: the morphology and structure of the inner region of the AC, espe- cially at the tip segment of the spicules, where its length growth (i) What is the structure and morphology of the compact proceeds. We show that the disturbed zone in the AC around the AC? Is it formed in a subsequent sintering process from axial filament is likely to be associated with the initially open ‘‘ax- originally grown silica lamellae or is its growth mecha- ial channel’’ that is described in former studies of young spicule nism different from that of the lamellar zone from the tips (see, e.g., Müller et al., 2009). In this tip region, the length beginning? development proceeds and it can be regarded as the active and (ii) A second topic concerns the so-called axial filament in the youngest zone of the spicule. Here also temporary growth struc- AC axis. Inside this zone, along the axis of the spicule, a thin tures may be found, the analysis of which would lead to a better vertical stripe with a diameter of approximately 2 lmis insight in the length growth mechanism. Next, we demonstrate located (marked by an arrow in Fig. 1e) described already that the AF located in the center of the amorphous silica matrix, in the earliest reports of M. chuni (e.g., by Schulze, 1904) has a crystalline quality. Its building blocks consist of organic as an organic axial filament (see also Fig. 1a, insert). molecules imbedded in silica. Its crystallinity shows similarities Concerning its morphology recent studies discuss, on the to inorganic crystals. Furthermore, organic inclusions and precipi- one side, that the silicatein molecules within the AF are tates occur in the AC. arranged as bundles of fibers (Müller et al., 2008). On the For the structural and chemical analysis, we combined SEM, other side, in the case of demosponges, even a first evidence focused ion beam target milling (FIB), transmission electron

Fig. 1. Morphology and cylindrical structure of the tip region of the basal spicule of Monorhaphis chuni observed by SEM (a) and by light microscopy (b–e). (a) The glass spicule consists of a major, lamellar silica shell and a non-lamellar silica core region, the axial cylinder (AC), schematically shown in the lower insert. (b and c) Light- microscopic dark-field images reveal this compact AC. It has a diameter of about 150–200 lm as schematically marked in the magnified image (c) as a small white cylinder AC. Although it consists of silica too, this inner cylinder has a different light-scattering behavior than the outer lamellar zone. In dark-field micrograph (d) it appears in light- blue on a dark background marked by an oval. (d and e) Inside the AC, a disturbed cylindrical zone exists with a diameter of approximately 20–30 lm marked by a box. It contains numerous discrete inhomogeneities (probably pores and precipitates) visible as bright dots in (e). In the center of the disturbed zone, a vertical filament with a diameter of approximately 2 lm is visible (marked by an arrow) that corresponds to the organic axial filament (AF). P. Werner et al. / Journal of Structural Biology 191 (2015) 165–174 167 microscopy (TEM), scanning-TEM (STEM), and element mapping 2.2. SEM investigation and FIB preparation by energy-dispersive X-ray spectroscopy (EDX). A number of subsequent sections were prepared by FIB milling 2. Material and methods for SEM inspection of the as-grown spicule tip morphology. We used an FEI Nova Nanolab 600 SEM/FIB operating at 2–5 keV 2.1. Selection of specimen material electron-beam energy and 30 keV and 5 keV ion-beam energy, respectively. For the following TEM/STEM investigations, FIB In this paper we compare the tip structure of two basal spi- lamellae were prepared that fulfill the following properties: (i) cules of M. chuni, which we assume to represent two stages in lamellae thickness < 200 nm, (ii) constant thickness within a lat- their axial growth process. We selected them from a set of four eral sample region of approximately 5 lm, (iii) no disturbing spicules collected in 1989 from the sea near New Caledonia at preparation damage and artifacts, (iv) well-defined regions of the a depth of approximately 2000 m (Weaver et al., 2010). spicule, especially of the AF with a targeting accuracy of <1 lm. Although this number of specimens does not allow for signifi- In general, two kinds of FIB sections were fabricated: (i) perpendic- cant statistics, qualitative differences exist. Tip segments with ular to the AF axis (‘‘cross-section’’) and (ii) along the AF axis (‘‘lon- a total length of approximately 20 cm were available from both gitudinal section’’), respectively. spicules. The first one, which is referred as spicule A, has a For comparison, conventional ion-thinned TEM samples were base diameter of approximately 2 mm. Sample B represents a also prepared from the basal regions at different distances away thinner spicule with a smaller base diameter approx. 1.3 mm. from the spicules tip to follow the morphology of the AC Besides their different diameters, the two spicules differ in downwards. their tip morphology, as demonstrated in SEM micrographs pre- sented in Fig. 2. Related to the reports we assume that both 2.3. TEM/STEM investigation spicules were alive when they were collected by a deep sea net. TEM techniques have been applied to typify the morphology, At the front segment of spicule A, some packets of silica the fine structure, as well as the elemental composition of the lamellae are partially peeled off (Fig. 2a). At the upper end of organic AF and the surrounding silica matrix. We combined con- the tip a short protruding segment of the AC with a naturally ventional TEM with STEM. Strong scattering features (e.g., of higher beveled/inclined front plane (Fig. 2a and b) is obvious. Near mass density) appear as dark image structures in TEM micro- the center of the AC, the position of the AF is indicated by an graphs, but as bright details in STEM micrographs, respectively. arrow in Fig. 2c. The advantage of the STEM imaging technique is that the corre- The corresponding SEM images of spicule B clearly show the sponding image feature’s intensity correlates with the mass den- outer lamellar morphology (Fig. 2d–f). Here the front plane of the sity of the specimen and is not adulterated by electron AC does not show a flat beveled face but a tongue-like structure diffraction phenomena. of silica. The gap between the tongue segment (T) and the under- For TEM and STEM investigations, we used a Jeol JEM 4010 lying cylinder AC is not caused by a crack but is a result of the (accelerating voltage U = 200–400 kV) and a FEI TITAN 80/300 growth process. This is supported by the fact that the adjacent sur- (U = 300 kV) working in a low-dosage mode to avoid strong radia- faces are not smooth, but show a roughness formed by small silica tion damage caused by the electron beam. The TEM/STEM images agglomerates as shown below in Section 3. Please note that the were recorded by 2k-CCD cameras. To improve the vertical flats (*) in the center of Fig. 2f are caused by mechanical ‘signal-to-noise ratio’ of the recorded CCD-images a carful image damage. processing (such as Wiener-filtering) was partly applied.

Fig. 2. SEM micrographs of the tip regions of spicules A and B with protruding AC and the surrounding lamellar zones. (a–c) Spicule A. The compact AC has a smooth, inclined front face without any obvious gaps or cracks. For further details, see Fig. 3. (d–f) Spicule B. The AC has a tongue-like apical end of silica, an open growth slit and longitudinal cracks; due to previous mechanical damage, nearly the entire half of the axial cylinder’s tip segment is missing; conf. Figs. 1 and 4. 168 P. Werner et al. / Journal of Structural Biology 191 (2015) 165–174

To obtain the elemental composition with a high lateral resolu- labeled by I–II–III). Two of them (I and II) are almost completely tion and element sensitivity, STEM imaging was combined with fused together with the exception of both lateral edges. The third, EDX-mapping (TITAN equipped with a ‘Super-X’ detector). Here, half-grown top layer (III) is still separated by a sub-lm wide inter- we chiefly measured the local two-dimensional-distribution of car- nal growth slit (arrow in Fig. 4e). bon, oxygen and silicon, but also considered other elements (e.g., To analyze the entire horizontal slit through the full length of Na, Cl, Mg). The two-dimensional EDX maps have been produced the tongue, a longitudinal FIB section was prepared through the (512 512 points) with a probe diameter of approximately tongue (Fig. 4a, dark central region, horizontal width approx. 0.2 nm applying a specific sensitive counting technique to prevent 55 lm). The corresponding magnified SEM image in Fig. 4c reveals significant radiation damage. that the slit between layers II and III extends over 50 lm from the As it will be demonstrated below, the AF is characterized by a spicules’ center to the tip of the tongue. Perpendicular to the slit, crystalline structure. Therefore, we investigated the AF and its circular growth slits within the DCZ are visible. Also in this case, crystal structure in different crystallographic orientations. the inner surfaces consist of silica particles; a very distinct example Accordant TEM/STEM micrographs of the cross-section will be is shown in Fig. 4d with particles of 50–100 nm diameter. referred in the following as a [001] viewing direction. Images from Fig. 4g presents a FIB cut through the central area of the tongue. longitudinal sections of the AF will be indicated as a [100] orienta- The SEM micrograph shows the front segment of the axial channel, tion, and 45° inclined samples as a [101] direction, respectively. which contains a rod-like structure, most probably a short piece of the AF marked by an arrow. Obviously, the axial channel is not yet completely filled up by the silica, like it was found in case of spi- 3. Results cule A. In slightly deeper segments of spicule B, but also in spicule A, the AF is completely embedded in the silica matrix (see also 3.1. Growth structures in the axial cylinder (SEM, FIB, TEM) Fig. 6a). We assume that – as a general growth phenomenon – the open axial channels mostly exist in the uppermost tip segment, We concentrated our investigations onto the uppermost tip seg- which represents the active and youngest zone of the growing spi- ments of the AC of spicules A and B within a range of 100 lmas cule. During subsequent growth, the open channels are filled by sil- shown in the SEM images in Fig. 2. On the AC’s inclined frontal ica. The filling process seems to start from the outer wall of the end of spicule A, a smooth surface is observable (Fig. 2c). The formerly open axial channel layer by layer towards the axial fila- assumed exit point of the axial filament seems to be covered by ment. Therefore, the open ‘‘axial channel’’ is filled by the over- a sub-lm thin layer showing only a faint subsurface secondary growth of silica just a few lm behind the tip of the spicule. electron contrast (arrow in Fig. 2c). During a successive removal Comparing our optical investigations with these SEM observa- of a several lm thin layer of silica, it was discovered that the AF tions, we find that the silica-filled axial channel directly correlates was closely embedded in the surrounding silica without any indi- to the disturbed cylindrical zone visible in light microscopy (Fig. 1d cation of an empty axial channel around it (for details see and e), diameter here approx. 20 lm). A magnified optical Supplemental data, Fig. A1). From this region at the AF, bright-field image of this zone is shown in a longitudinal cross-section and longitudinal section TEM samples were finally cross-section (Fig. 5a) and highlighted by the schematic DCZ cylin- prepared by FIB (see Fig. 6a and Supplemental data, Fig. A1). der. The small visible features might be caused by light-scattering During these thinnings processes along the AF axis down to a pores, gaps and precipitates. To test this hypothesis, several depth of 30–40 lm no obvious wide gaps, cavities or other struc- FIB-sections were prepared from these regions of spicule B and tures in the lm-scale were detected in the direct surrounding of investigated by SEM and TEM. the AF. In corresponding FIB-cut cross-section samples, circular stria- A different morphological situation was found at the tip region tions with small radial cavities are observed. An example is shown of the spicule B (overview Fig. 2d–f). As shown in Fig. 3a–c, the in Fig. 5b as an SEM image, where two ring-like features are visible front end of the AC is not a flat face but characterized by a at a distance of 3.5 lm and 5 lm from the center. Optical images of tongue-like structure of silica, which extends from the shorter side this zone already show circular inhomogeneities in the silica towards the most protruding edge of the AC. This tongue is sepa- (Fig. 5c). The structure of the matrix of this inner zone is character- rated from the underlying AC by a narrow gap approximately ized by nanopores within a size range of 20–200 nm. In STEM 1 lm wide. SEM micrographs of the regions marked in Fig. 3b show images, such nanopores appear as dark spots. An example is given at higher magnification that these surfaces possess a texture of in Fig. 5d, which shows their arrangement along circles in the sub-micrometer silica particles (see Fig. 3e and f) indicating them cross-section image, the organic AF being visible in the upper bor- as younger, actively growing surfaces (Pisera, 2003). These spheri- der of the image as a dark circular segment. Note, that the organic cal particles are the well-known primary building units generated AF is also visible as a vertical line in Fig. 5a within the ‘‘disturbed’’ in the basic silica synthesis. Directly on the surfaces the particles cylindrical zone. are loosely packed; during the successive maturation of the silica Fig. 5e schematically represents a summary of the silica struc- the pores between the particles disappear by a sintering process ture of the spicule’s central region: the AC (£ 150 lm) includes (‘‘biosintering’’). the disturbed cylindrical zone (£ 20–30 lm) with larger pores Further growth structures could be identified by SEM: Firstly, at the outer regions and nanopores within the whole cylinder. the surfaces of the tongue and of the underlying AC show in their The AF (£ 2 lm) is indicated by a yellow dot. It is very likely that center rough, ring-like structures, which are shown in Fig. 3d and e the disturbed cylindrical zone corresponds to the previously open, (see also Fig. 4f). These circular regions, having diameters of but now silica-filled AC described earlier. approximately 30 lm, obviously correspond to the disturbed cylin- drical zone mentioned above (Fig. 1e). This feature will be dis- 3.2. The structure of the axial filament (AF) cussed in more detail later in this paper. Secondly, for a deeper analysis of the internal growth structures The morphology and the detailed structure of the organic AF several compact FIB sections were prepared at selected locations of were studied by conventional TEM and by STEM techniques. As the silica tongue. One section close to the outermost tip of the ton- previously mentioned, corresponding micrographs are typified by gue clearly revealed that the tongue is actually composed of three a nearly reversed image contrast. This has to be taken into account layers, each approximately up to 3 lm thick (Figs. 3f and 4e, layers for the analysis of the following figures that show images taken P. Werner et al. / Journal of Structural Biology 191 (2015) 165–174 169

Fig. 3. Growth structures of the tip region of the spicule B investigated by SEM. (a) Tip of the AC still cladded by silica lamellae; the axial position of the AF is marked by *. (b) Enlarged view of the silica tongue of the AC with positions of details (d–f) marked by arrows. (c) Schematic of the AC tip shown as a cross-section including tongue T and the disturbed cylindrical zone (DCZ, dashed region around the AF) are indicated. (d–f) Enlarged views of the rough growth surfaces of the tongue and the underlying AC characterized by sintered silica particles. (e) Ring segments at the edge of the DCZ. with both techniques. In the TEM image in Fig. 6a, the AF in the organic/silica matrix of the AF, respectively. However, before dis- center is represented as light gray, whereas the ambient silica cussing their chemical nature, a general structural feature of the appears in darker gray due to its higher mass density. We regard AF matrix should be discussed. this cross-section image as a [001] projection (viewing direction) The AFs of all investigated spicules is characterized by a peri- of the filament. Inside the filament as well as in the interface odic lattice structure. As an example, Fig. 7a shows a TEM image, two different kinds of small inclusions were detected with approx- which corresponds to a h010i viewing direction perpendicular to imate sizes between 20 nm and 100 nm. Fig. 6b shows a corre- the filament axis. The lattice system is nearly rectangular and sponding TEM image of a longitudinal section, which we refer to shows lattice planes with a distance of 7.0 ± 0.2 nm, as magnified as the [100] direction. The inclusions of the first kind are larger in Fig. 8b. The nearly perfect cubic structure is slightly distorted, and dark, indicating a higher mass density in comparison to the creating an angle of 86°. The lattice is typified by a strong period- AF and the silica matrix. Such inclusions occur inside the AF as well icity, which is well demonstrated by diffractograms (by ‘‘fast-Fou as at the interface. As an example, two corresponding particles are rier-transformation’’, or FFT) generated from such high-resolution marked as ‘A’ inside and as ‘B’ in the interface in Fig. 6b. The inho- micrographs (Fig. 7c). Fig. 7d shows the enlarged lattice, where mogeneous radial distribution of such incorporated particles the bright spots are caused by structures of lower mass. It is very should be mentioned. The second type of inclusions was also iden- likely that they correspond to the position of silicatein molecules. tified inside the AF. In the STEM Fig. 6c, small dark and bright In correspondence to former investigations (Zlotnikov et al., agglomerates are visible. The bright ones are particles of higher 2014), the lattice size amounts to a = 9.8 ± 0.2 nm and mass density; the small dark ones of lower density than the c = 10.8 ± 0.2 nm. As determined by the herein mentioned X-ray 170 P. Werner et al. / Journal of Structural Biology 191 (2015) 165–174

Fig. 4. Internal growth structures of spicule B visible by SEM. (a) Overview of the tongue structure after a section in the central region was milled by FIB; this section appears as a dark rectangular contrast in the center. (b) Schematic of the growth features of the spicule’s tip shown as a cross-section; letters refer to the charts. (c) Enlarged view of the large radial gap according to the inserted box in (a); the AF is located at the left side. The large gap is cut by circular-running gaps. (d) The inner surfaces of these gaps show still non-fused silica spheres in the 50–100 nm diameter range. (e) Cross-section of the tongue composed of three sub-layers (I, II, III). (f) View onto the circular growth structures occurring on the tongue (upper part) and the underlying part of the AC (after mechanical cross-section fracture). (g) FIB section of the partly empty axial channel (nearly longitudinal). In its center residual parts of the AF are visible.

Fig. 5. Disturbed cylindrical zone (DCZ) within the spicule B. This axial zone corresponds to an initially open axial channel, now filled with silica during the further spicule growth. (a) Optical micrograph of this zone revealing its diameter of approximately 20 lm (see inserted cylinder). Small agglomerates/pores are visible in this longitudinal section. (b) SEM image of a compact FIB cross-section of this inner zone. The circular defective lines correspond to precipitates, micropores and gaps. The dashed line represents the radius position at approximately 5 lm from the center. (c) Optical dark-field micrograph of a pre-thinned FIB lamella with cylindrical striations caused by additional small mass-inhomogeneities in the silica matrix. (d) STEM image of the same lamella, attributes the cylindrical striations to small pores and larger pores arranged on rings. The dark circle segment on top of charts (c) and (d) corresponds to the AF. (e) Scheme of the AC including the inner DCZ (£ 20–30 lm) and the organic AF (yellow dot, £ 2 lm). diffraction measurements, the arrangement of the silicatein net- organic three-dimensional silicatein lattice embedded in an amor- work represents a body-centered cubic lattice, which is slightly phous silica matrix. tetragonal distorted. The dark matrix around the bright spots cor- STEM techniques allow imaging of the morphology and, simul- responds to silica synthesized by the protein as indicated by the taneously, a mapping of the chemical composition by EDX at high following EDX analysis. We conclude that the AF consists of an spatial resolution (<5 nm). We have applied this combination to P. Werner et al. / Journal of Structural Biology 191 (2015) 165–174 171

Fig. 6. Micrographs of the morphology of the AF (spicule A) embedded in the silica of the disturbed zone of the axial cylinder. (a) TEM cross-section image, where the AF with its organic content appears as bright and the surrounding silica as dark gray. Note the dark particles inside and at the interface of the filament. (b) TEM micrograph of a longitudinal section showing a high density of particles as well (marked, e.g., A, B). (c) STEM micrograph of two types of precipitates in the matrix of the AF visible as dark (low mass density) and bright (higher density).

Fig. 7. (a) TEM micrograph of a longitudinal section (h010i viewing direction) showing a lattice structure of the AF. (b) The lattice planes have a distance of approximately 7.0 nm. The corresponding crystallographic lattice directions are inserted. The strong lattice periodicity is confirmed by the corresponding diffractogram (c). Herein the four {101} reflections are marked by circles. (d) Magnified TEM image of the lattice structure with the inserted unit cell projection (a = 9.8 nm, c = 10.8 nm). The bright spots correspond to the location of lower mass, most probably silicatein molecules embedded in silica. get a better understanding of the fine crystal lattice structure. An region. The lattice of the organic crystal is characterized by the example of the analysis of a longitudinal section is given in the {200} planes having a distance of 4.9 ± 0.2 nm. The interfacial Supplement data Fig. A2. steps show heights between 10 nm and 30 nm and often consist To typify the perfection of the three-dimensional crystal lattice, of flat {100} and {110} planes. we have prepared TEM lamellae related to other crystallographic In general, it can be stated that the existence of the crystal lat- directions, too. An example is given in Supplement data Fig. A3 tice of the AF determines its rectangular or octahedral shape and representing the [101] orientation where the viewing direction the faceting of the interfaces. is approximately 45° inclined to the axis of the filament. The figure also includes an electron-diffraction pattern demonstrating the 3.3. Inclusions in the AF perfection of the lattice by sharp diffraction reflection up to the 5th order. As mentioned in the preceding paragraph, the AF as well as the We also analyzed the interfaces between the organic filament adjacent interface to the axial cylinder of the silica matrix contain and the ordinary silica surrounding it by visualizing its structure various inclusions or precipitates. We took care that such features as well as its chemical composition. As mentioned previously, the were not generated as artifacts due to our TEM sample preparation AF is typified by a nearly quadratic or rectangular cross-section, (ion-milling without liquids). Since we find such inclusions in often facetted at the rounded corners, as demonstrated, e.g., in numerous samples of spicules without open slits and channels, Fig. 8a. Given the crystalline nature of the AF, the boundaries can we assume that these inclusions are presumably generated during be described by a h100i and by a h010i orientation of the the spicule growth and not by infiltration of seawater in extinct body-centered tetragonal lattice, respectively. This is indicated in spicules. the TEM micrograph in Fig. 8b including crystallographic direc- As mentioned above, we find in the AF beside large darker fea- tions. The interfaces themselves have a pronounced faceted struc- tures (Fig. 6) small precipitates with lower mass density. In the ture. The corners of the rectangular cross-section are typified by TEM images they do not show a specific structure and seem to longer flat planes (vertical and horizontal), whereas the h100i side be homogeneous and amorphous. We have analyzed the chemical planes are stepped. Fig. 8c shows this situation in more detail. It nature of the inclusions by EDX element-mapping. An example is correlates to the lower right region of Fig. 10b. In this TEM image given in Fig. 9. It represents a longitudinal section of the interface the plain silica surrounding the AF appears as the lower dark region between the AF and the surrounding silica cylinder. The 172 P. Werner et al. / Journal of Structural Biology 191 (2015) 165–174

Fig. 8. TEM/STEM micrographs of the h001i cross-section of the AF. (a) SEM image shows the rectangular character of the axial filament (dark) embedded in silica (light gray). (b) The TEM overview demonstrates the faceted structure of the interface with correlated lattice orientations. (c) Magnified TEM image of the lower right of chart (b) shows the crossed {200} lattice planes. The interface is smooth and contains lattice correlated interfacial steps.

STEM dark-field image (Fig. 9a) shows the filament matrix in a preparation by FIB we are able to get a detailed view inside darker intensity than the more dense silica matrix (right side). In the structure of the basal silica spicule of the sponge M. chuni. both regions, dark spots of different sizes correspond to inclusions. We have specifically investigated the tip regions of two spicules An EDX element map of the same sample regions is presented in (A and B) apparently corresponding to different growth stages. Fig. 9b, where ‘‘green’’ corresponds to silicon and ‘‘red’’ to carbon, Spicule A corresponds to a temporarily terminated growth state, respectively. The dark spots in the filament matrix contain highly whereas spicule B typifies an active growth state, including concentrated carbon. Therefore we assume that they are organic temporary growth features. Combining the results from the inclusions within the silicatein lattice. The dark spots in the silica analysis of the inner axial cylinder AC of both spicules, we matrix do not contain carbon; they are most likely nanopores. can receive a better understanding especially of the spicule’s The larger kind of precipitates at the interface and inside the AF length growth. Whereas the outer lamellar zones of the spicules (A and B in Fig. 6b) do not include carbon. An EDX mapping of the have a width growth by alternating wrapping of a thicker silica elemental distribution is given in Fig. 10. Chart (a) shows a STEM layer and a thin organic layer, the analyzed AC morphology image of the interface region in a longitudinal section. The EDX indicates a dominant axial growth process of silica components. map in Fig. 10b reveals that carbon (red) is only present in the axial Here the AF plays a guiding role for the straight vertical growth cylinder, whereas silicon (green) dominates the surrounding silica. of the AC. Concerning our new findings we propose the follow- In the filament, the Si concentration is much lower (Fig. 10b). ing model: Because of the chosen lower STEM/EDX-mapping magnification, the lattice structure is not resolved in this case. In the (i) The front region of the AC is formed by a successive growth Supplemental data Fig. A4, however, the corresponding magnified of silica layers in the thickness range of some lm. This over- lattice image is shown. Fig. 10c demonstrates that the large parti- growth can occur as a tongue-like silica structure spreading cles at the interface consist of sodium (yellow) and chlorine (pink). over the whole diameter of the AC. In the beginning of this Due to the measured 1:1 element ratio, we conclude that the par- overgrowth, there remains a gap between the tongue and ticles are most likely precipitated NaCl salt. Since they show partly the AC, which is closed in the following by a deposition of an extension into the AF lattice we assume that they were formed silica. Larger tongues, as observed for spicule B, themselves during the AF growth at it surface before enclosed by the silica consist of several thin silica layers with internal growth slits matrix. as demonstrated in Fig. 4e. The plate-like layers grow by fur- ther deposition of silica. At the end of such a process cycle, a 4. Discussion new compact AC front region is formed as represented by spicule B. At the same time a new silica layer is formed, Combining optical microscopy, different methods of electron thereby performing the step-like axial growth of the spicule. microscopy (SEM, TEM, STEM) and a dedicated sample We assume that the system of gaps and openings allows a P. Werner et al. / Journal of Structural Biology 191 (2015) 165–174 173

Fig. 9. Determination of the element distribution in the interfacial region of the AF/AC by STEM/EDX mapping seen perpendicular to the spicule axis. The STEM image (a) shows inclusions appearing as dark spots in both regions. The EDX map (b) shows the distribution of silicon (green) and carbon (red), respectively, demonstrating the different nature of inclusions. The shape of the AF and the AC is schematically indicated in the upper part of the figure.

Fig. 10. EDX analysis of the precipitates at the axial filament/silica cylinder matrix marked by AF and si, respectively. (a) STEM image of precipitates (b) by EDX mapping the distribution of carbon (red) and silicon (green) is represented. (c) Corresponding EDX map demonstrating that the particles consist of sodium.

contact to the sponge tissue and to the water, and thereby uppermost silica layer, the cylindrical zone is partly empty. the flow of reaction components involved in the growth We attribute it to the open ‘‘axial channel’’ described in for- process. mer investigations of specific spicules (Müller et al., 2008, (ii) Building blocks for this silica matrix are small spherical silica 2009) with the AF free-standing in its center. We assume agglomerates with a size ranging from 50 to 100 nm that are that here at the axial growth front, the AF is enclosed, visible on growing inner surfaces and on the outer front e.g., by cell tissue (‘‘silicalemma’’, Pisera, 2003). During faces (Fig. 4d–f). Nanoscopic pores between them remain the further length growth the layer around the AF is no even after the further fusion process (‘‘Biosintering’’, e.g., longer needed and the open axial cylinder is filled by silica. by dehydration), as they were observed by TEM as concen- Our SEM observations (Figs. 3e, and 4g) indicate that the tric rings with a fluctuating density (Fig. 5d). open cylinder is closed layer-by-layer starting from (iii) In the center of the AC exists a cylindrical zone where the the outer wall of the axial channel. The silica filling occurs silica matrix is considerably disturbed (DCZ). This newly not so dense and ordered as in the outer zone of the AC and described phenomenon can be observed already by optical generates numerous irregularities with a lm-size. In the microscopy throughout the full length of the spicule. A early phase of this concentric filling-up, circular growth characteristic feature of this zone (diameter approximately gaps exist, as displayed at the wide FIB section (Fig. 4c 30 lm) is its aperiodic tree-ring structure (Fig. 4f). It can be and d). Summarizing this discussion, the DCZ is interpreted well observed at the surfaces of the growth gaps and of the as a permanent left-over of a formerly open axial tongue (Figs. 3d, and 4f). Only within some lm in the channel. 174 P. Werner et al. / Journal of Structural Biology 191 (2015) 165–174

(iv) Larger precipitates, identified as NaCl, occur at the AF/AC Appendix A. Supplementary data interface. Additionally organic agglomerates were quite fre- quently found in the AF, but rare in the silica matrix of the Supplementary data associated with this article can be found, in AC. the online version, at http://dx.doi.org/10.1016/j.jsb.2015.06.018. (v) The silicatein-containing AF plays a dominating role for the length development process of the spicule, which is widely References accepted today. Our investigations allow a more detailed view on this process. On the one side, it locally governs Croce, G., Viterbo, A., Milanesio, M., Amenitsch, H., 2007. A mesoporous pattern created by nature in spicules from Tethya aurantium sponge. Biophys. J. 92, 288– the enzymatically driven synthesis of silica. On the other 292. side, nonetheless, the AF scaffolds the growing silica AC. Ehrlich, H., 2010. Biological Materials of Marine Origin – Invertebrates. Springer, The composite lattice structure of periodically arranged pro- New York. tein molecules surrounded by silica could be important in Ehrlich, H., Heinemann, S., Heinemann, Ch., Simon, P., Bazhenov, V.V., Shapkin, N.P., Born, R., Tabachnick, K.R., Hanke, Th., Worch, H., 2008. Nanostructural providing both sufficient stiffness and a much lower fragility organization of naturally occurring composites—Part I: silica–collagen-based than silica alone. As we demonstrated in an earlier work, the biocomposites. J. Nanomater. 2008. http://dx.doi.org/10.1155/2008/623838 AF has a crystal-like structure (Zlotnikov et al., 2014) with (Article ID 623838). Garonne, R., Simpson, T.L., Pottu-Boumendil, J., 1981. Ultrastructure and deposition three-dimensionally arranged silicatein molecules embed- of silica in sponges. In: Simpson, T.L., Volcani, B.E. (Eds.), Silicon and Siliceous ded in a mesoporous silica matrix. The TEM observations Structures in Biological Systems. Springer, New York, pp. 495–525. reveal the existence of lattice planes, crystallographic direc- Kropf, K., 2008. (Ph.D. thesis). University Mainz, Inst. of Physical Chemistry. Mayer, G., 2009. Role of biosilica in materials science: lessons from silicious tions and lattice steps at the surfaces/interfaces that are sim- biological systems for structural composites. In: Müller, W.E.G., Grachev, M.A. ilar to the crystallographic features of an inorganic crystal. (Eds.), Biosilica in Evolution. Marine Molecular Biotechnology 47. Springer Publ, However, up to now there exists no convincing model of that pp. 277–293. Miserez, A., Weaver, J.C., Thurner, P.J., Aizenberg, J., Dauphin, Y., Fratzl, P., Morse, growth process, which would explain the favored growth of D.E., Zok, F.W., 2008. Effects of laminate architecture on fracture resistance of the AF along the axial direction. sponge biosilica: lessons from nature. Adv. Funct. Mater. 18, 1–8. (vi) Concerning the silica production for the AC growth, the AF Müller, W.E.G., Wang, X., Kropf, K., Ushijima, H., Geurtsen, W., Eckert, C., Tahir, M.N., Tremel, W., Boreiko, A., Schloßmacher, U., Li, J., Schröder, H.C., 2008. Bioorganic/ can only take part in the synthesis as long as the silicatein inorganic hybrid composition of sponge spicules: matrix of the giant spicules molecules are not enfolded by the surrounding silica. and of the Comitalia of the deep sea Monorhaphis. J. Struct. Biol. Therefore another source of the silica formation has to be 161, 188–203. taken into consideration (Pisera, 2003). Very likely, the first Müller, W.E.G., Wang, X., Burghard, Z., Bill, J., Krasko, A., Boreiko, A., Schloßmacher, U., Schröder, H.C., Wiens, M., 2009. Bio-sintering processes in hexactinellid inner organic interlayer around the AC, which was reported sponges: fusion of bio-silica in giant basal spicules from Monorhaphis chuni.J. to contain collagen and proteins (designated as ‘‘axial barrel’’ Struct. Biol. 168, 548–561. in: Müller et al., 2008; Kropf, 2008) could be the main source Pisera, A., 2003. Some aspects of silica deposition in lithistid desmas. Microsc. Res. Tech. 62, 312–326. for the silica synthesis. Schulze, F.E., 1904. Hexactinellida. Wissenschaftliche Ergebnisse der Deutschen Tiefsee-Expedition, vol. 4. Fischer-Verlag, Jena, pp. 112–121. 5. Conclusion Shimizu, K., Cha, J., Stucky, G.D., Morse, D.E., 1998. Silicatein alpha: cathepsin L-like protein in sponge biosilica. Proc. Natl. Acad. Sci. U.S.A. 95, 6234–6238. Wang, X., Boreiko, A., Schloßmacher, U., Brandt, D., Schröder, H.C., Li, J., Kaandorp, Our investigation of basal spicules of the hexactinellid M. chuni J.A., Götz, H., Duschner, H., Müller, W.E.G., 2008. Axial growth of hexactinellid yields that the non-lamellar AC is the result of a growth mecha- spicules: Formation of cone-like structural units in the giant basal spicules of the hexactinellid Monorhaphis. J. Struct. Biol. 164, 270–280. nism that is, from the beginning, different from that one effective Wang, X., Schröder, H.C., Wang, K., Kaandorp, J.A., Müller, W.E.G., 2012. Genetic, for the surrounding lamellar zone. It is characterized by an axial, biological and structural hierarchies during sponge spicule formation: from soft layered growth mechanism guided by the AF. The crystalline meso- sol–gels to solid 3D silica composite structures. Soft Matter 8, 9501–9518. Weaver, J.C., Milliron, G.W., Allen, P., Miserez, A., Rawal, A., Garay, J., Thurner, P.T., porous silica/silicatein lattice structure of the AF is assumed to Seto, J., Mayzel, B., Friesen, L.J.F., Chmelka, B.F., Fratzl, P., Aizenberg, J., Dauphin, enable its perfect rectilinear growth and, thereby, the formation Y., Kisailus, D., Morse, D.E., 2010. Unifying design strategies in demosponge and of such long silica spicules. hexactinellid skeletal systems. J. Adhesion 86, 72–95. Beside the new results, there are still unanswered questions. Woesz, A., Weaver, J.C., Kazanici, M., Dauphin, Y., Aizenberg, J., Morse, D.E., Fratzl, P., 2006. Micromechanical properties of biological silica in skeletons of deep-sea These concern a further investigation, e.g., of the specific crystal sponges. J. Mater. Res. 21, 2066–2078. growth mechanism of the one-dimensional elongated AF under Zlotnikov, I., Shilo, D., Dauphin, Y., Blumtritt, H., Werner, P., Zolotoyabkov, E., Fratzl, deep-sea conditions. Under debate is also the question as to what P., 2013. In situ elastic modulus measurements of ultrathin protein-rich organic layers in biosilica: towards deeper understanding of superior resistance to is triggering the formation of the first organic interlayer at this fracture of biocomposites. RSC Adv. 3, 5798–5802. specific radius (approx. 100 lm) and the interaction between it Zlotnikov, I., Werner, P., Blumtritt, H., Graff, A., Dauphin, Y., Zolotoyabko, E., Fratzl, and the AF to receive the scaffolding growth. P., 2014. A perfectly periodic three-dimensional protein/silica mesoporous structure produced by an organism. Adv. Mater. 26, 1682–1687.

Acknowledgments

The authors would like to thank F. Natalio for fruitful discus- sions, S. Hopfe for the preparation of special TEM cross-section samples, and T. Goffin for editing the manuscript.