Arrested in Glass: Actin within Sophisticated Architectures of Biosilica in Sponges

Abstract Actin is a fundamental member of an ancient superfamily of structural intracellular proteins and plays a crucial role in cytoskeleton dynamics, ciliogenesis, phagocytosis, and force generation in both prokaryotes and eukaryotes. It is shown that actin has another function in metazoans: patterning biosilica deposition, a role that has spanned over 500 million years. Species of glass sponges (Hexactinellida) and demosponges (Demospongiae), representatives of the first metazoans, with a broad diversity of skeletal structures with hierarchical architecture unchanged since the late Precambrian, are studied. By etching their skeletons, organic templates dominated by individual F‐actin filaments, including branched fibers and the longest, thickest actin fiber bundles ever reported, are isolated. It is proposed that these actin‐rich filaments are not the primary site of biosilicification, but this highly sophisticated and multi‐scale form of biomineralization in metazoans is ptterned.


Note 2. Isolation of the axial filaments a) for Phalloidin-and immunostaining
Selected spicules (n=45) and skeletal frameworks (n=25) (see Figure 1 and Figures S5-S11, Supporting Information) have been treated in 70% HNO 3 at room temperature (RT) up to 5 times during 72 h for elimination of possible external organic layers. The absence of residual organic matter has been confirmed using fluorescence microscopy as well as Coomassie blue R-250 staining for proteins as well as using X-ray photoelectron spectroscopy (XPS).
Organic-free skeletal structures have been rinsed in dist. H 2 O up to pH 6.5, dried on air at RT and placed on the Nunc™ Permanox™ (Thermo Fisher Scientific, USA) plastic microscope slides (27/75mm) (n=120) in small drops of water. After water evaporation, one drop of 10% HF acid as recognized silica demineralizing reagent was applied to the surface of the spicules, or corresponding fragment of the skeletal framework. In order to prevent HF vapour from entering the environment, the slide has been placed at RT, inside the Plexiglas Petri dish at 10° angle. This positioning of the sample allows the HF-droplet to slowly roll down the surface of the plastic slide and removal of the dissolved silica, releasing the axial filaments during 7-10 h (see Figure S1a, Supporting Information). Following such treatment, the residual axial filaments remain to be strongly fixed to the plastic surface and have been rinsed with water during 2 h, dried on the air, and after that stained with Phalloidin reagents, or with fluorescently labelled ß-actin antibodies (see Note 3, Supporting Information).
It is to note that HF acid should not destroy the peptide bonds and that proteins treated with it display the same biochemical properties as normal. In work by Lou et al. [61] , it was shown that silica bioreplication preserves three-dimensional spheroid structures of human pluripotent stem cells and HepG2 cells. The authors used diluted in buffer hydrofluoric acid for remove silica composites, and obtained spheroid was successfully stained for F-actin using antigens staining. Figure S1. (a) Schematic diagram depicting the extraction of axial filaments from skeletal structures of sponges by a sliding drop method. The effectiveness of this method never approved before is visualized on the images (b) and (c). Light microscopy image (b) represent the case of partially demineralized skeletal framework of Euplectella sp. glass sponge after staining with iFluor 594-Phalloidin, where the residual silica is signed with dotted lines and visualized in the insertion with arrows under higher magnification. The fluorescence image (c) of the same sample shows with strong evidence localization of actin-based axial filaments even being located with residual silica. It should be noted that the remains of the organic matrix, clearly visible in the image (b) in the form of a case-like structure, were not stained with highly specific to actin phalloidin dye, in contrast to the axial filaments (c).

b) for Immunostaining, MS, Raman, SDS-PAGE
Organic-free skeletal structures of selected sponges under study have been prepared as described above, rinsed in dist. H 2 O up to pH 6.5, dried on air at RT and, then placed in 50 ml plastic vessels containing 25 ml of 10%HF (Fluka) solution according to the method by Drum [21] . The vessel was covered, placed under thermostatic conditions at 25°C for time periods between 24 h and 96 h in dependence of the sponge species. Extracts obtained after dissolution of biosilica were dialyzed against deionized water (10 L) on Roth (Germany) membranes with a molecular weight cut-off 14 kDa at 4°C during 72 h. The dialyzed material was stored at 4°C and used for actin identification as described in detail in the notes 5 and 6, Supporting Information.

Note 3. Staining with Phalloidins and Immunostaining
Axial filaments were stained with the actin staining kits: Cell Navigator™ F-Actin Labeling Kit *Red Fluorescence*; Cell Navigator™ F-Actin Labeling Kit *Green Fluorescence*; Cell Navigator™ F-Actin Labeling Kit *Blue Fluorescence* provided by AAT Bioquest (USA).
iFluor 594-Phalloidin (red); iFluor™ 488-Phalloidin (green) and iFluor™ 350-Phalloidin (blue) fluorescent phalloidin conjugates are a high-affinity probes for F-actins. Used at nanomolar concentrations, phallotoxins are convenient probes for labelling, identifying and quantitating especially of F-actins. In brief, isolated axial filaments fixed on the Nunc™ Permanox™ (Thermo Fisher Scientific) plastic microscope slides were treated with 100 µL/well of iFluor™ 594-Phalloidin or iFluor™ 488-Phalloidin or iFluor™ 350-Phalloidin working solution and stained at RT. Afterwards the plates were carefully washed 5 times with dist. water over 1 h, dried and observed using light and fluorescent microscopy.
For immunostaining, the samples of both partially demineralized and mineral-free isolated axial filaments were placed on the sample glasses in 30 µL of dH 2 O and were dried overnight.
Then 30 µL of 4% PFA/PBS were added over each sample and the samples were incubated for 20 min in the wet chamber. The PFA solution was removed and the samples were washed with PBS (3 x 5 min, 30 µL/sample). The samples were blocked with 30 µL of 2% of normal goat serum in PBS were added over each sample and the samples were incubated for 30 min in the wet chamber. The primary anti-β-actin antibody (#PA1-16889, Invitrogen) were diluted 1:200 in the buffer containing 3% BSA / 0.02% NaN 3 / PBS according to the manufacture"s protocol. 30 µL of the primary antibody solution were placed over each sample and the glasses were incubated overnight in the wet chamber at +4°C. Then the samples were washed with PBS (3 x 5 min, 30 µL/sample). The secondary antibody (Anti-rabbit IgG (H+L), F(ab')2 Fragment (Alexa Fluor® 488 Conjugate) #4412, Sell Signaling Technology) were diluted in PBS according to the manufacture"s protocol and 30 µL of this solution were placed over each sample and the glasses were incubated for 1 h in the wet chamber at room temperature.
Then the samples were washed with PBS (3 x 5 min, 30 µL/sample). The DAPI-free mounting medium was added over each sample and the the glass slides were covered with glass slips and immediately analyzed using fluorescent AxioScope.A1 (Carl Zeiss, Göttingen, Germany) microscope and with the AxioVision40 V4.8 software (Carl Zeiss Imaging Solutions, Göttingen, Germany).
The samples of the axial filaments under study, which have been treated only with secondary antibodies (#4412, Sell Signaling Technology) have been used as control.
As you can see, there is principal distinction with respect to reagents (i.e. PFA/PBS treatment) and reaction conditions between phalloidin-based staining and immunostaining of axial filaments under study. Due to fundamental differences with regard to the handling of samples, we did not expect the same image after visualization with respect to thickness and density.

Note 4. Stereo, light and fluorescent microscopy
Stereomicroscopy images were taken with a Keyence VHX-5000 digital optical microscope and VH-ZST swing-head zoom lens. Light microscopy and fluorescent images were obtained using a Keyence BZ-9000 fluorescence microscope.

Note 5. SDS-PAGE and western blotting
Axial filament samples isolated from A. setubalense, S. lacustris and C. arcticus were precipitated with ice cold acetone, incubated at -20°C and centrifuged at 10,000 g for 10".
For western blot analysis, proteins were transferred to PVDF membrane using the Trans-Blot TM Turbo system (Bio-Rad, USA) and blocked with TBS-T/4 -5% BSA. Blots were incubated overnight with the IgG anti-ß-actin (for S. lacustris #5125, Cell Singling, Beverly, MA, USA; for A. setubalense -#PA1-16889, Thermo Fisher Scientific, USA) as a primary antibody. Chemiluminescent detection was performed using ChemiDoc XRS imaging system (Biorad, USA) or ECL chemiluminescence system (Thermo Fisher Scientific, USA) according to the manufacturer protocol.
Since there is no information on the sequence of actin from S. lacustris in the Uniprot database, we do not know what molecular mass the actin from this sponge should have. We suggest that actin in this sponge species has a peptide bond that is labile under the conditions of dissolution of spicules, the cleavage of which divides the molecule into two unequal parts: 25 and 15 kDa; protein bands corresponding to precisely these molecular weights are observed in the Western blotting picture (Figure S15a, b, Supporting Information).

Note 6. LC-MS/MS analysis
To identify proteins separated using SDS-PAGE, bands were manually excised from gels and   Figure S15). In these samples 4 silicatein peptides were also found.
Characteristics of these peptides and representative spectra are presented in Supporting Information, Table 5 and Figure S16 and S17.
In the case of M. chuni, M. sp and C. arcticus 14, 15 and 21 actin peptides were identified utilizing label-free LC-MS/MS approach without SDS-PAGE separation, respectively.
Sequences and results of these analyses are presented in Supporting Information, Tables 2, 3 and 4. Representative spectra are presented in Figure S16, Supporting Information.

Note 7. Scanning electron microscopy
The samples were fixed in a sample holder and covered with carbon, or with a gold layer for 1 min using an Edwards S150B sputter coater. The samples were then placed in an ESEM XL 30 Philips or LEO DSM 982 Gemini scanning electron microscope.

Note 8. Ultramicrotomy and transmission electron microscopy
Dry axial filaments of glass sponges, C. arcticus and M. chuni were placed in ethanol 30% for one day at RT. They were subsequently dehydrated in ethanol series at RT and stored in 100% ethanol at 4°C. Then they were cut into short fragments using micro scissors. The samples

Note 9. High-resolution transmission electron microscopy (HRTEM)
TEM analyses were performed using a FEI Tecnai F30-G2 with Super-Twin lens (Thermo Fisher, Eindhoven, NL) with a field emission gun at an acceleration voltage of 300 kV. The point resolution amounts to 2.0 Å, and the information limit to about 1.2 Å. The microscope is equipped with a wide-angle slow scan CCD camera (MultiScan, 2k×2k pixels; Gatan Inc., Pleasanton, CA, USA).
Raman spectra of the reference proteins and of axial filaments extracted from glass sponges M. chuni and from C. arcticus were recorded with a Raman spectrometer (RamanRxn1™, Kaiser Optical Systems Inc., Ann Arbor, USA) coupled to an upright microscope (DM2500 P, Leica Microsystems GmbH, Wetzlar, Germany). The excitation of Raman scattering was obtained with a diode laser with wavelength 785 nm, which was propagated to the microscope with a 62.5 µm optical fiber and focused on the samples by means of a 20x/0.45 microscope objective, leading to a focal spot of about 35 µm. The Raman signal was collected in reflection configuration and propagated to the spectrograph using an optical fiber with 100 µm core. The spectral resolution is 4 cm -1 . Raman spectra were punctually recorded, using a laser power of about 150 mW. Raman spectra of single axial filaments isolated from Asconema setubalense and Petrosia fuciformis placed on glass substrates were acquired using a confocal Raman microscope (Alpha 300S, WITec GmbH, Ulm, Germany) coupled to a Raman spectrometer UHTS 300S) and using a laser excitation at 780 nm with TEM00 quality (TA Pro, Toptica Photonics AG, Gräfelfing, Germany). A 100x magnification objective with NA=0.9 was used to focalize the excitation and collect the Raman signal in reflection configuration. Raman spectra were punctually recorded on single fibres, using a laser power of about 30 mW.
An integration time of 2 s was used in both measurements and several spectra were averaged in order to improve the signal-to-noise ratio. The fluorescence background was removed with a multi-point linear baseline using the software GRAMS/AI (Thermo Fisher Scientific, USA Inc, Waltham, MA, USA).

Note 11. Latrunculin B inhibition
Gemmules isolated from S. lacustris were carefully placed on the sterile glass slides. The hatching of the gemmules was performed by placing them on glass slides in 60 ml of tap water in plastic Petri dishes at RT. During the whole experiment all slides were constantly covered with water, and all containers were covered to prevent the loss of water through the evaporation. First hatching was observed after 72h. The microscope slides containing one or more hatched gemmules were prepared and used for actin inhibition experiments (Figure 5 and Figure  In total of 72 used gemmules, 64 could be observed to hatch. Of the 22 hatched gemmules treated with the inhibitor, none one has been observed to grow with spicules. All hatched gemmules, treated with DMSO or only tap water respectively, were observed to contain spicules during the growing phase.
Latrunculin B, actually originally discovered in sponges, inhibits actin polymerization. This Latrunculin is highly specific and does not kill or stop cell function. Cell viability during and after treatment with Latrunculin is well known and recently has, for example, been confirmed by Durak, et al. [47] and by by Ayscough et al [62] . The evidences that the inhibition of F-actin by Latrunculin is reversible has been shown in vertebrate cultured cells [63] and in corals [64] .
In our experiments with Latrunculin we used protocols worked out on other aquatic organisms.
This informed our selection of concentration and duration of exposure, so they did not affect the physiology of individual cells or the viability of the whole organism. The duration of the experiments was due to the well-studied feature of the development of S. lacustris sponges after their release from the gemmules and the time required for the development of spicules under normal conditions. Recovery experiments were not necessary since cells were observed to be live and not lysed daily via light microscopy, also visible in SEM (Figure 5).

Note 12. Germanium experiment
We have used the same gemmules from S. lacustris as described in the note 11, Supporting Information. To analyse the influence of germanium (Ge) on the morphology of axial filaments in control experiment, slides (n=17) with gemmules were treated with 2 ml of tap water each containing 0.35 mM of Na 2 SiO 3 (Sigma-Aldrich). For the comparative experiment, sufficient quantities of GeO 2 (Sigma-Aldrich) were added to yield Ge/Si molar ratios of 0.5.
Each day these two-culture media were correspondingly replaced with fresh one. The impact of germanium on sponge spiculogenesis was assessed every 24 h by the light and fluorescence microscopy (Keyence BZ 9000, Osaka, Japan) within a course of 8 days.

Note 13. Axial filaments and sclerocytes
Sclerocytes are recognized as cells, in which the siliceos spicules appear. The cytoplasm of an active sclerocyte is filled with abundant small clear vesicles, mitochondria and numerous microtubes  .The spicules are secreted intracellularly within a special vacuole, the membrane of which was termed a silicalemma. The silicalemma is different from the cell membrane [66] . The silicalemma is often very close to the plasma membrane, but no obvious connections between the plasma membrane and silicalemma have been observed. As suggested by Simpson [68] , the silicalemma seems to be an invagination (in pocketing) of the cell membrane and therefore, the axial filament and silica occur in an elongate, extracellular pocket. Therefore, silicalemma appears to form an elongate vacuole in the cytoplasm with no direct connections to the Golgi complex, endoplasmic reticulum, or other membrane system. Spiculogenesis begins with the synthesis of the axial filament, which is then mineralized by silica deposition, however, in such a way that the space around it remains free of mineral phase (see Figure S2, Supporting Information). The deposition of silica is due to the secretory activity of the silicalemma, which enlarges as additional silica is deposited. Thus, the silica closest to the axial filament is the oldest (denser) and that at the periphery is youngest (least dense) [68] .
After the formation of the spicule, the sclerocyte either slides off the newly formed megascleres, or it is transported into the extracellular space. Sclerocyte cells at the end of spiculogenesis do not undergo apoptosis or death. The same cell can start the synthesis of a next spicule [69] . A micro tubular material, more or less organized, is often seen surrounding the axial filament at the first stages of spicule development [70] . It is not known whether these microtubules are present only during the early assembly of the filament. All the current evidence indicates that the axial filament is formed by active sclerocytes during the early stages of spiculogenesis. Axial filament remains an active and important participant in the development of spicules. The axial filament may have as its major function the establishment of the overall morphology of the developing spicule both microscleres and megascleres. Up to now there is only one hypothesis of axial filament origin, proposed by Lévi [71] , further elaborated by Simpson [68] : axial filament forms from the fusion of cytoplasmic vacuoles, developed by invagination of the cell membrane.
Pottu-Boumendil [53] and Simpson et al. [50] described the appearance of hexagonal axial filament in demosponges, surrounding with microtubules in the sclerocyte cytoplasm at a very early stage of spiculogenesis. According to their TEM micrographs, immediately after this, the axial filament is surrounded by a layer of silicalemma adjacent to its surface before development of a vacuole. This observation allows us to propose the second hypothesis of the formation of axial filament in sclerocytes of sponges, namely, its development due to the condensation of extracellular F-actin nano-fibrils.          image, b). Such radial actin pattern upon radially symmetric growth of up to 8 µm long actin bundles remains unique. Occurrence of morphologically similar Factin aster-like structures has been recently reported in early neuronal development [57] . SEM image (c) of the mechanically disrupted oxyasters shows broken siliceous rays with corresponding axial channels, in which axial filaments have been initially located.

FigureS11. Visualization of axial filaments isolated from Euplectella aspergillum.
Hierarchical and complex skeletal architecture of E. aspergillum glass sponge (a, b, c, d, e) remains to be the classical research object for materials scientists. HF treatment leads to isolation of axial filaments (light microscopy images f and h) from the skeletal framework after 72 h. In spite of fragility, some of axial filaments still resemble the structural motives with geometries (h) known for intact skeleton (e). Specific iFluor™ 594-Phalloidin staining identifies their actin nature (g, i).       Table 5, Supporting Information). Retention time (RT), precursor mass (m/z), charge (z) and carbamidomethyl / oxidation modifications are presented. Figure S18. Structural similarity between F-actin bundles and nanofibers of axial filament of Monorhapis chuni glass sponge. TEM images (a, d) of the bundles of F-actin standard observed previously [54] show high similarity to that obtained by us for selected nanofibers of M. chuni axial filament (b, e). The magnification shows double helices typical for actin filaments (c, f). where the fibril length amounts to 1.2 µm and the width varies between 90 nm to 140 nm. The FFT in Figure S19b, Supporting Information taken from the red frame area of Figure S19a indicates different large periodicities typical for actin such as 39.8 nm and 35.7 nm, see [59][60][61] . This value is due to a 28/13 symmetry, or a helical repeat of 28 subunits in 13 turns of the 59 Å. Spacing of 81.2 nm is close to the value that Poole et al. [58] registered at 77 nm. We propose this spacing corresponds to the first order reflection of the actin spacings around 39 nm according to 77nm/2 giving 38.5nm. The typical reflection of 14 nm for actin was also detected, (see e.g. [63,64] ), corresponding to the distance between the centres of neighbouring actin filaments. Figure S19c shows a bundle of actin filaments about 800 nm thick, subdivided to 140 nm thin sub-fibrils. The FFT of the fibrils of the red marked area of (c) show reflections corresponding to spacings of 19.3 nm, and 39.3 nm in one direction corresponding to a first and second order reflection. In the other direction we observed spacings of 133 nm, 68 nm and 35.3 nm, which represent the first through third order reflections of actin. The two directions are visualized by arrows. e) HRTEM analysis of the surface of an individual fibril of M. chuni axial filament (see also Figures S18 and S22, Supporting Information) shows structural features typical for mono-filament-bound cross-links observed previously from actins of diverse origin [65] . Figure S20. Structural similarity between F-actin bundles and nanofibers of Monorhapis chuni axial filament. TEM images (a, c) of the bundles of F-actin standard observed previously [58] show high similarity as compared to that obtained by us for selected nanofibers of M. chuni axial filament (b, d).  Table 7, Supporting Information). .1 nm is fitting to the c-axis value of (001) [66] whereas 3.4 nm corresponds either to (003) or (103) and 2.4 nm to (201) or (202) of actin filaments see [63,67] . As reported previously [68] , the spacing at 2.4 nm is strongly related to the actin helix.  [60,61,64,67] , respectively) (see Table 7, Supporting Information).    control (a, b). The fluorescence microscopy images (b, d) strongly confirm that formation of the malformed spicules (c) is attributed to the bifurcation (d, arrows) of the actin-based (Figure 1) axial filament as comparatively represented on the schematic presentation of actin filaments trajectories (e, f). In contrast to other fibrillar proteins (i.e. collagen), structural branching is attributed to the one of characteristic features of the actin filament growth in diverse organisms (see for review [69] ).         were identified with 12% and 19% sequence coverage, respectively.  * Elongation factor-1 alpha is an actin binding protein.