# Source paper archive - Source title: Fluorescence microscopic evidence for intracellular silicon storage pools in diatoms and their partial separation - Author(s): Tobias Reichelt, Julia Breyer, Eike Brunner - Source landing URL: https://www.biorxiv.org/content/10.64898/2026.11.04.764211v1 - PDF URL: https://www.biorxiv.org/content/10.64898/2026.11.04.764211v1.full.pdf - PDF SHA-256: ab84e8d56fbf1cd741ea72f1d02f0874b30cba1538ea8fa5e62ca1ca7544dbcc - Converter: pymupdf4llm 0.0.27 - Limits: Automated PDF text extraction; reading order, equations, tables and formatting may be incomplete or incorrect. Images are omitted, and no OCR was performed. Markup and links are neutralized for safe downloading. Consult the original PDF for the authoritative version. --- [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. # 1 Fluorescence microscopic evidence for intracellular silicon 2 storage pools in diatoms and their partial separation [1 ] 3 4 **Tobias Reichelt**, Chair for Bioanalytical Chemistry, Faculty of Chemistry and Food Chemistry, TU 5 Dresden, Bergstrasse 66, 01062 Dresden, Germany 6 **Julia Breyer**, Chair for Bioanalytical Chemistry, Faculty of Chemistry and Food Chemistry, TU 7 Dresden, Bergstrasse 66, 01062 Dresden, Germany 8 **Eike Brunner** ², Chair for Bioanalytical Chemistry, Faculty of Chemistry and Food Chemistry, TU 9 Dresden, Bergstrasse 66, 01062 Dresden, Germany, eike.brunner@tu-dresden.de 10 11 **Running title** 12 Separation of silicon pools 13 14 **Acknowledgments** 15 Financial support by the German Research Foundation (Deutsche Forschungsgemeinschaft, 16 DFG, Project No. Br 1278/35-1) is gratefully acknowledged. 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 November 3, 2026 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. ## 32 Abstract 33 Diatoms are unicellular algae that occur ubiquitously in almost every salt- and freshwater habitat 34 on earth. Silicic acid is taken up from the environment and stored in intracellular silicon storage 35 pools (SSPs) to form species-specific silica-based cell walls (frustules). During biosilica 36 formation in the silicon deposition vesicle (SDV), the silicon is transported from SSPs into the SDV. 37 The present contribution describes the fluorescence microscopic visualization and 38 quantification of silica solely formed from SSPs under silicon starvation. Labeling with the 39 fluorescence marker PDMPO demonstrates continuing cell growth and cell division without 40 silicon uptake from the environment. The SSPs of _Cyclotella nana_ may account for up to 1/3 of 41 total cellular silicon and intracellular concentrations up to ca. 540 mM, significantly beyond the 42 saturation level (2 mM). 43 The chemical state and localization of SSPs within cells are yet unclear. Here, we report a 44 method for subcellular fractionation of _C. nana_ by density gradient centrifugation. Quantification 45 of cytoplasmic silicon yields 3–10% of the total cellular silicon corresponding to intracellular 46 concentrations of ca. 50–160 mM. Labeling with NMR-active [29] Si shows that this silicon occurs as 47 condensed silica, i.e., silicic acid is converted into condensed silica after uptake. However, the 48 detected intracellular silicon concentrations are lower than the value of ca. 540 mM predicted by 49 the fluorescence experiments. It is thus concluded that the missing SSP silicon is either present 50 in the form of large aggregates or compartments. Alternatively, they may also be bound to cell wall 51 biosilica. 52 53 **Key index words** 54 Silicon storage pools, nuclear magnetic resonance, isotope labeling, silicon starvation, pulse 55 labeling, PDMPO, Molybdenum blue assay, density gradient centrifugation, cytoplasmic silicon, 56 _Cyclotella nana_ 57 58 **Abbreviations** 59 ASW, artificial seawater 60 ASW-Si, silicon-free artificial seawater 61 bSiO 2, biogenic silica 62 CI, confidence interval 63 cryo-FIB, cryogenic focused ion beam 64 EDS, energy-dispersive X-ray spectroscopy 65 HEPES, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid 66 LOQ, limit of quantification 67 MAS, magic angle spinning 68 MBA, molybdenum blue assay 69 ME, mercaptoethanol 70 NMR, nuclear magnetic resonance 71 PDMPO, [2-(4-pyridyl)-5-((4-(2-dimethylaminoethylaminocarbamoyl)methoxy)phenyl)oxazole] 72 PMSF, phenylmethylsulfonyl fluoride 73 RCF, relative centrifugal force 74 SDV, silica deposition vesicle - 1 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 75 SEM, scanning electron microscopy 76 SSPs, silicon storage pools 77 STVs, silicon transport vesicles 78 TEM, transmission electron microscopy 79 tSi, total silicon (amount of substance) 80 tSi cell, total cellular silicon (amount of substance) 81 [tSi cell ], concentration of the tSi cell in the cell volume 82 tSi SSP, total silicon in the silicon storage pools (amount of substance) 83 [tSi SSP ], concentration of the tSi SSP in the cell volume 84 Q [n], notation for the different chemical groups of silicon in silicic acids and solid silica 85 YSZ, yttria-stabilized zirconia 86 87 - 2 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. ## 88 Introduction 89 Diatoms are unicellular microalgae with a pronounced silicon metabolism. They ubiquitously 90 occur in salt- and freshwater habitats and make up a major part of the phytoplankton. Diatoms 91 are responsible for approximately 40% of the oceanic primary production (Treguer & Pondaven, 92 2000; P. Tréguer et al., 1995) corresponding to about 20% of the global carbon fixation (Field et al., 93 1998). Today, about 12,000 diatom species are known (Guiry, 2012). Estimates of the total number 94 range from 20,000 to well over 200,000 species (Guiry, 2012; Malviya et al., 2016; D. G. Mann & 95 Droop, 1996; David G. Mann & Vanormelingen, 2013). For comparison, only about 700 Radiolaria 96 species are known (Biard et al., 2017). This diversity is reflected by a wealth of sometimes bizarre 97 cell wall shapes and patterns. Various geometries and highly ordered structures down to the 98 nanoscale occur. The synthesis mechanisms underlying this wide variety of shapes and the 99 biomolecules involved in biomineralization are still not fully identified and understood. Several 100 polypeptides were isolated from biosilica of diatoms, including silaffins (Kröger et al., 1999, 2001) 101 and silacidins (Wenzl et al., 2008), as well as special long-chain polyamines (Kröger et al., 2000). 102 The latter were found in all diatom species analyzed yet (Sumper et al., 2005) and even in 103 sediments (Annenkov et al., 2018; Bridoux et al., 2011; Bridoux & Ingalls, 2010) as well as the 104 silica-forming _synurales_ species _Synura echinulata_ (Reinke et al., 2025). 105 Diatoms actively take up and accumulate silicon (silicic acid) from their environment 106 (Azam et al., 1974; Lewin & Chen, 1968; Lewin, 1955). This silicon is transformed into biosilica 107 during the formation of their cell wall (biomineralization) in a specialized compartment called 108 silicon deposition vesicle (SDV) (Drum & Pankratz, 1964; Heintze et al., 2020; Reimann et al., 109 1966; Stoermer et al., 1965). Diatoms can store silicic acid in so-called silicon storage pools 110 (SSPs) (Azam et al., 1974; Brunner et al., 2009; Gröger et al., 2008; Kumar et al., 2020; Lewin, 1954; 111 Martin-Jézéquel et al., 2000; Sullivan, 1979; Thamatrakoln & Hildebrand, 2008). However, the 112 intracellular localization and chemical form of SSPs is not yet clear (Martin-Jézéquel et al., 2000). 113 The decisive causes for the enormous evolutionary success of diatoms are still being 114 debated (Cermeño et al., 2015). In any case, their rapid spread in geological terms, which is still 115 ongoing, seriously impacts global silicon cycles. The silicon concentration in the oceans has 116 changed dramatically over the course of Earth's history. More than 550 million years ago, the level 117 was close to saturation, i.e., about 10 – 20 times higher than today (Conley et al., 2017; McCutchin 118 et al., 2025). The decline of oceanic silicon concentration coincides with the emergence of the 119 first silicifying organisms (radiolarians, sponges) in the Phanerozoic era (Siever, 1991; P. J. Tréguer 120 et al., 2021). Diatoms appeared much later during the Triassic period about 240 million years ago 121 (Kooistra & Medlin, 1996; Medlin et al., 1997). The earliest fossil record dates back about 190 122 million years (Sims et al., 2006). During two periods of rapidly increasing biodiversity, first between 123 the Eocene and Oligocene (38 to 32 million years ago) and later in the early Miocene (20 million 124 years ago to the present), the evolutionary and geographical spread of diatoms took its course 125 (Cermeño et al., 2015; Kooistra & Medlin, 1996). Their global success was accompanied by a 126 further decreasing silicic acid concentration in the oceans. A concentration around 600 µM is 127 estimated for time 240 million years ago (Kooistra & Medlin, 1996; Siever, 1991). Today, the 128 oceanic average is approximately 90 µM (P. J. Tréguer et al., 2021). Typical concentrations in 129 surface waters where microalgae grow are about one order of magnitude lower (Cermeño et al., 130 2015) at an average of <10 µM (Siever, 1991). During the growth period or algal blooms, the 131 environmental silicon concentration can even be close to zero which may explain the need for 132 SSPs. - 3 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 133 Lewin investigated the silicic acid uptake of silicon-deficient _Navicula pelliculosa_ cultures 134 after silicon replenishment and observed uptake even in the darkness (non-growing conditions). 135 It was hypothesized that silica is deposited on the surface of the pre-existing cell wall (Lewin, 136 1954). Microscopic observations of the complementary situation showed that cell wall formation 137 (silica deposition on the site of the new cell walls) takes place even in silicon-free culture medium 138 (Coombs et al., 1967). Isotope uptake studies with [31] Si (Azam et al., 1974; Sullivan, 1979) and [68] Ge 139 (Azam et al., 1974; Sullivan, 1979) delivered final evidence for the existence of SSPs. It is 140 concluded that the SSPs of some species may be sufficient to build up a complete cell wall and 141 thus account for up to 50% of the total cellular silicon content (tSi cell ). Examples are 142 _Ditylum brightwellii_ and _Conticribra weissflogii_ (Chisholm et al., 1978; Martin-Jézéquel et al., 143 2000). The amount of silicon stored in SSPs depends on the species, environmental conditions, 144 day time, and cell cycle. Estimated intracellular concentrations range from 19 mM 145 ( _Cylindotheca fusiformis_ ), 58–162 mM ( _N. pelliculosa_ ) (Martin-Jézéquel et al., 2000) up to 300 mM 146 ( _Lithodesmium undulatum_ ), 340 mM ( _Stephanopyxis turris_ ) up to even 70–2000 mM 147 ( _D. brightwellii_ ) (Chisholm et al., 1978). These concentrations significantly exceed the saturation 148 concentration of 2 mM for silicic acid at near-neutral pH (Iler, 1979). It is concluded that SSPs 149 must be stabilized to prevent spontaneous and uncontrolled precipitation in the cell (Azam et al., 150 1974; Sullivan, 1979). Stabilization or complexation of silicic acid by biomolecules and/ or 151 controlled polymerization of silicic acid to oligomers (Annenkov et al., 2008; Azam et al., 1974; S. 152 D. Kinrade et al., 1999; Stephen D. Kinrade et al., 2002; Spinde et al., 2011; Werner & Pirson, 1967), 153 a silica gel (Grachev et al., 2017) or nanoparticles (Brunner et al., 2009; Gröger et al., 2008) have 154 been proposed. Combined transmission electron microscopy (TEM) and energy-dispersive X-ray 155 spectroscopy (EDS) indicated silicon-containing intracellular inclusions in _Synedra acus_ 156 (Grachev et al., 2017). Cytoplasmic vesicles of 30–40 nm size fusing with the silicalemma of the 157 SDV in _Thalassiosira eccentrica_ were observed by TEM and scanning electron microscopy (SEM). 158 It was hypothesized that these vesicles are silicon transport vesicles (STVs) that originate from the 159 Golgi apparatus (Schmid & Schulz, 1979), although there was no evidence for silicon inside that 160 vesicles. Similar observations were made in _S. turris, D. brightwelli_, and _Gomphonema parvulum_ 161 (Dawson, 1973; Li & Volcani, 1985a, 1985b). Consistently high intracellular silicon concentrations 162 were detected in _Cyclotella nana_ (formerly _Thalassiosira pseudonana_ ) by cryogenic focused ion 163 beam (cryo-FIB) in combination with SEM and EDS, leading to the assumption of a homeostatic 164 silicon concentration within the cell. After 12 h silicon starvation, this concentration still amounts 165 to 187 ± 58 mM which even exceeds the value of 140 ± 51 mM determined in cells during 166 exponential growth. The authors report a rather uneven distribution of silicon throughout the cell 167 (Kumar et al., 2020). Solid-state [29] Si NMR investigations revealed that silicic acid is condensed 168 immediately after uptake and converted into silica. This became evident from [29] Si pulse labeling 169 of _C. nana_ and _D. brightwellii_ cultures. Accordingly, the intracellular storage of large quantities of 170 soluble silicic acid (Q [0] /Q [1] ) can be ruled out. Instead, it could be concluded that SSPs consist of 171 silica in a pre-condensed form. One hypothesis is that SSPs consist of stabilized silica particles 172 (Bertermann et al., 2003; Brunner et al., 2009; Gröger et al., 2008). Furthermore hypervalent, five 173 or sixfold coordinated silicon species can at least transiently occur in diatoms as shown by [29] Si 174 NMR studies on _N. pelliculosa_ and _C. nana_ (Brunner et al., 2009; Stephen D. Kinrade et al., 2002). 175 The quantitative determination and chemical analysis of SSPs requires their separation 176 from the siliceous cell walls. Established methods are for example the extraction with perchloric 177 acid at 0 °C, trichloroacetic acid at 0 °C, or water at 100 °C (Azam et al., 1974; Sullivan, 1979). 178 These rather harsh methods are, however, likely to change the chemical state of SSPs and to at - 4 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 179 least partly dissolve cell wall silica. Therefore, mechanical cell lysis by bead beating followed by 180 centrifugation offers a suitable alternative for gentle SSPs extraction (Reichelt et al., 2023). 181 Sensitive silicon quantification can then be performed using the established Molybdenum blue 182 assay (MBA) method (Coradin et al., 2004; Koistinen et al., 2020). 183 The present work focuses on the development of a density gradient centrifugation-based 184 method with the aim of separating SSPs in _C. nana._ The original and current scientific name is 185 _Cyclotella nana_ (Alverson 2011, Poulsen 2023). The former name, _Thalassiosira pseudonana,_ is 186 still widely in use. The results of this subcellular fractionation make SSPs accessible for 187 quantification and further chemical analysis. The basic concept is shown in Fig. 1. Moreover, a 188 novel fluorescence microscopic method for SSPs estimation is developed using [2-(4-pyridyl)-5189 ((4-(2-dimethylaminoethylaminocarbamoyl)methoxy)phenyl)oxazole] (PDMPO) for pulse 190 labeling. Finally, cell cultures are enriched with the NMR-active isotope [29] Si. Successful 191 separation yields fractions that are enriched with the SSPs and can be analysed by solid-state [29] Si 192 NMR spectroscopy. 193 194 **Figure 1** . Illustration of the separation process carried out as a parallel experiment to compare 195 cells (top) with extracted cell walls (bottom) as control in order to identify the SSPs in the fractions. ## 196 Materials and methods 197 **Cultivation of** _**C. nana**_ **at standard conditions** 198 _C. nana_ strain CCMP 1335 was cultivated in sterile, sealed 20-liter polycarbonate containers 199 (Nalgene, Rochester, NY, USA). Artificial seawater (ASW) was prepared with 100 µM of silicic acid 200 (Harrison et al., 1980). For isotope labeling, [29] Si enriched silicic acid solution was used instead _._ 201 For silicon starvation and washing, silicon-free ASW (ASW-Si) was prepared accordingly, without 202 the addition of silicic acid. Exclusively plastic labware was used during cultivation. The cultures 203 were maintained at 20 °C in an incubator (RuMed1301, Rubarth Apparate GmbH, Laatzen, 204 Germany) under gentle agitation and constant illumination at 6.000 – 8.000 Lux. The pH was 205 regularly adjusted to 8.0. The cell density was monitored using an automatic cell counter 206 (CellDrop FL, DeNovix, Wilmington, DE, USA) and the silicic acid concentration was measured 207 using the MBA. Harvesting took place during exponential growth and before silicic acid was 208 depleted. This was usually 7-10 days after inoculation at a density not exceeding 0.8∙10 [6] ml [-1] . The 209 cells were concentrated on a cellulose filter (round filter, Rotilabo 15A, 2-3 µM retention range, 210 Carl Roth GmbH, Karlsruhe, Germany) and washed with cold ASW-Si under vacuum. To 211 resuspend the cells, the filter was immersed in a small volume ASW-Si, placed on ice and gently - 5 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 212 shaken for 5 min to gently detach the cells. For silicon starvation the concentrated cell 213 suspension was added to 20 L ASW-Si. For final harvest the suspension was centrifuged (2,500 x g 214 for 5 min at 4 °C), washed with a cold solution of 4.0% sodium chloride and again centrifuged. The 215 pellets were shock frozen and stored at -80 °C. 216 **Isotope labeling with** **[29]** **Si** 217 Isotope labeling was carried out using [29] Si-enriched sodium metasilicate, which was prepared 218 from [29] SiO 2 of 98.2% purity (Cortecnet, Les Ulis, France). Equimolar amounts of SiO 2 and sodium 219 carbonate were mixed in a platinum crucible and heated to 1150 °C in a furnace. For details see 220 the references (Schwarz & Menner, 1924; Wisser et al., 2015). The obtained sodium metasilicate 221 readily dissolves in water and depolymerizes to silicic acid. Insoluble residues were removed by 222 a syringe filter. A yield of 98% was determined with the MBA. 223 **Quantitative estimation of the SSPs by PDMPO labeling** 224 Methanol extraction was proven by McNair et al. to largely eliminate unbound PDMPO and remove 225 most of the photopigments, which lowers the background fluorescence and increases 226 measurement sensitivity (McNair et al., 2015). Only plastic labware was used during cultivation. 227 Two separate media with 1.5 L of ASW were inoculated with _C. nana_ strain CCMP 1335 in sterile 228 and sealed 2.8 L polycarbonate containers (Nalgene, Rochester, NY, USA). A concentration of 229 0.125 µM PDMPO (AAT Bioquest, Pleasanton, CA, USA) was added to one of these batches for full 230 labeling. Cultivation was carried out under constant illumination (6.000 – 8.000 lux) to exclude 231 cell cycle effects. Harvesting took place during the exponential growth phase after 5 days at a cell 232 density of 0.6 - 0.8 · 10 [6 ] ml [-1] . The fully labeled cells were harvested by centrifugation (2,500 x g for 233 5 min at 4°C). The pellet was suspended in 20 ml methanol (HPLC grade) and left for extraction at 234 4°C for at least 24 h. Methanol extraction was proved by McNair et al. to largely eliminate unbound 235 PDMPO and remove most of the photopigments, which lowers the background fluorescence and 236 increases measurement sensitivity (McNair et al., 2015). The unlabeled cells were harvested with 237 a cellulose filter (round filter, Rotilabo 15A, 2-3 µM retention range, Carl Roth GmbH, Karlsruhe, 238 Germany) and washed to completely remove silicic acid. To achieve this requirement, the cell 239 suspension was filtered until less than 100 ml remained above the filter. Then 500 ml of ASW-Si 240 was added for washing and filtered until less than 100 ml remained. The washing step was 241 repeated 6 times. After these steps, the remaining Si concentration is negligible, i.e., it is equal to 242 the trace contamination of silicon in ASW-Si (results section). The unlabeled cells were 243 transferred directly from the filter into fresh ASW-Si (silicon starvation) containing 0.125 µM 244 PDMPO (pulse labeling). After further incubation for 48 h at continuous illumination, the cells 245 were harvested by centrifugation as described before and suspended in methanol. 246 The extracted cells were washed once with methanol and twice with 4% (w/v) sodium 247 chloride solution (2,500 x g for 5 min) before the analysis under the microscope. Aliquots of the 248 sample were pipetted onto a microscope slide and covered with a cover slip. Adding sucrose to 249 the suspensions to increase the viscosity prevented the cells from moving and mechanical 250 damage. Images were taken in the fluorescence channel of the microscope Biozero BZ8000 251 (Keyence, Osaka, Japan) at a wavelength of 360 nm for excitation and 535 nm for emission at 252 20fold magnification and an acquisition time of 10 s. No oversaturation of the detector occurred 253 for these parameters. At least 6 observations were made in z-stack mode. A stack of 6 to 12 254 individual images with a vertical focus distance of 1-2 µm was sufficient to focus on all cells in the 255 layer. The images were calculated to form a resolved image (full focus) using the analysis software - 6 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 256 of the manufacturer (Keyence). The intensities were quantified using the image analysis software 257 _ImageJ_ . Different from the original method introduced by Mc Nair et al., no fluorescence standard 258 was used in the present study. Instead, the PDMPO fluorescence intensity of the pulse labeled 259 cells was directly correlated to the intensity of the fully labeled cells. Additional species were 260 investigated: _C. weissflogii_ strain CCAP 1085/18, _N. pelliculosa_ strain SAG 1050-3 _, Cyclotella_ 261 _cryptica_ strain CCMP 332 _and_ _Biddulphia tridens_ unknown strain (see supporting information, 262 Figs. S2 – S5 and Tab. S1). 263 **Bead beating lysis** 264 The samples, dense cell pellets or biosilica, not exceeding 300 µL in volume, were given in 2.0 ml 265 microtubes. The cold lysis buffer was added to a final volume of 1500 µL. This buffer at pH of 7.5 266 contained 150 mM sodium chloride, 50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid 267 (HEPES), 1 mM mercaptoethanol (ME), and 100 µM phenylmethyl-sulfonyl fluoride (PMSF). ME 268 and PMSF were added shortly before the lysis. Following the addition of 2.2 g yttria-stabilized 269 zirconia (YSZ) beads of 0.1-0.2 mm diameter (Sigmund Lindner GmbH, Warmensteinach, 270 Germany), the microtube was vigorously mixed on a vortex mixer for 90 s with breaks for cooling. 271 The lysate was separated from the beads and kept at 4 °C. (Reichelt et al., 2023). 272 **Density gradient centrifugation** 273 A continuous sucrose gradient was prepared from 10% (w/w) and 60% (w/w) sucrose solutions 274 using a gradient mixer, a peristaltic pump, and a 14G cannula attached to the bottom of a 50 ml 275 centrifuge tube. The solutions were buffered with 20 mM HEPES at pH 7.5. The final gradients were 276 cooled until the lysate was carefully layered on top using a syringe and a cannula. Centrifugation 277 at 4,200 x g (5,000 rpm) and 4°C was performed for 60 min with acceleration and deceleration set 278 to low values. The tube was emptied beginning from the top with a canula and a peristaltic pump 279 and the collected fractions were immediately frozen. The linearity of the density gradient was 280 confirmed by measuring the refractive index of the fractions. 281 **Isopycnic density gradient ultracentrifugation** 282 Ultracentrifugation was performed with unlabeled samples and isotope labeled samples. A pellet 283 of cells or extracted cell walls of 2000 µL volume (ca. 13 x 10 [9] cells) was used for the experiment. 284 The lysis resulted in 10 ml of lysate, which was layered over a stepped density gradient covering a 285 density range of 1.11 - 1.43 g/ml. The seven layers, each of 4 ml volume, consisted of an aqueous 286 solution of iodixanol (OptiPrep, 60% w/v aqueous solution, Serumwerk, Bernburg, Germany) in 287 the upper part and iohexol (Nycodenz, powder, Serumwerk, Bernburg, Germany) in the lower part 288 of the column, or a mixture of the two components. The upper five layers contained 20, 30, 40, 50 289 and 60% (w/v) OptiPrep and sodium chloride at concentrations of 122, 108, 93, 79, 72 and 65 mM 290 to obtain iso-osmolar conditions. The sixth layer consisted of a 1:1 mixture of OptiPrep and 291 Nycodenz at 70% (w/v) at ca. 440 mOsm, and the seventh layer contained 80% (w/v) Nycodenz at 292 ca. 770 mOsm. All solutions were buffered with 20 mM HEPES at pH 7.5. 293 Ultracentrifugation was performed in open-top thin-wall ultra-clear polycarbonate tubes 294 (Beckmann Coulter, Brea, CA, USA) with a volume of 38.5 ml (25x89 mm) using the rotor SW 32Ti 295 (Beckmann Coulter) on the ultracentrifuge Optima XE90 (Beckmann Coulter) at 32,000 rpm and 296 4 °C for 60 min. The relative centrifugal force (RCF) was thus 175,000 x g at the bottom and 297 76,600 x g at the top of the tube. Acceleration and deceleration were set to intermediate values. 298 The tube was emptied beginning from the top using a cannula and a peristaltic pump. The - 7 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 299 fractions were removed in a way that the transition between two density layers was exactly in the 300 middle of the fraction and the particles in the boundary area were completely taken up. Only the 301 first fraction (8 ml) and the last fraction with the pellet (2 ml) not contained a density transition. 302 The fractions were carefully homogenized and samples were taken for the analysis of the total 303 silicon amount (tSi). The fractions were shock-frozen and stored at – 80 °C. Selected fractions 304 were analyzed by NMR spectroscopy. 305 **Alkaline treatment** 306 For the alkaline treatment, the silicon concentration in the sample (solution or suspension) 307 should not significantly exceed 2 mM. Otherwise, it should be diluted before. 0.2 M sodium 308 hydroxide solution was added to the sample with a volume at least equal to the volume of the 309 sample. The mixture was heated to 95 °C for 15 min. After cooling, the same volume of 0.2 M 310 hydrochloric acid was added to the mixture for neutralisation. The tube is filled with water to the 311 desired volume and dilution (Koistinen et al., 2020). 312 **Molybdenum blue assay** 313 An aliquot of the alkaline treated sample was transferred to a 2.0 ml microtube. For the standard 314 addition, a series of 5 aliquots was taken. 4 aliquots were mixed with a silicic acid standard 315 solution (BerndKraft, AnalytiChem GmbH, Duisburg, Germany) to a final added standard 316 concentrations of 5, 10, 15 and 20 µM (based on the final volume of 1800 µL). The microtubes 317 were filled with ultrapure water to a volume of 1620 µl. 60 µL of a solution of 81 mM ammonium 318 heptamolybdate and 1.8 M sulfuric acid was added. After mixing and equilibration for at least 319 3 min, 60 µL of 1 M oxalic acid solution was added, followed by mixing and 1 min equilibration. 320 Finally, 60 µL of 100 mM ascorbic acid solution was added. Measurements were performed after 321 30 min. If precipitation occurred or particles were present, the microtube was centrifuged at 322 20,000 x g for 3 min. The supernatant was transferred to a cuvette (10 mm, polystyrene, VWR, 323 Radnor, PA, USA) for photometric measurement (Cary 60, Agilent, Santa Clara, CA, USA). The 324 background of a blank sample was subtracted. (Koistinen et al., 2020). The silicic acid 325 concentration in the cultivation media was quantified accordingly. The trace contamination of 326 ASW-Si was determined using diluted silicic acid standards with final added concentrations of 327 0.2, 0.4, 0.6 and 0.8 µM. Multiple blanks in separate cuvettes (n=16) were measured to determine 328 the limit of quantification (LOQ) of 0.17 µM. The absorbance was detected for 5 s (average time) 329 at a wavelength of 810 nm. 330 **Solid-state** **[29]** **Si NMR spectroscopy** 331 The [29] Si-enriched samples were investigated without drying. The fractions (4 ml) obtained after 332 ultracentrifugation were suspended in cold lysis buffer (8 ml) to reduce the solution density, 333 followed by centrifugation at 4,200 x g and 4 °C for 40 min. Centrifugation was repeated with the 334 supernatant for 90 min, to recover few amounts of additional material. The wet pellets of the 335 washed fractions and the wet pellets of the whole cells were resuspended with a few drops of the 336 buffer. The fractions belonging to the gradient densities of 1.21 g/ml and 1.26 g/ml were 337 completely combined to obtain sufficient sample material for the measurement. The suspension 338 particles were transferred to a rotor insert (disposable insert for 4 mm magic angle spinning (MAS) 339 rotor, 25 µL nominal volume, Bruker BioSpin, Karlsruhe, Germany) by repeated centrifugation. To 340 this end, a 200 µl pipette tip was cut on the narrow end to tightly fit inside the orifice of the rotor 341 insert. Both were connected with a rubber tubing cut short. The wide end of the tip was cut so that - 8 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 342 the whole construction fit in a 2 ml Eppendorf tube. The insert and the tip were filled with up to 80 343 µL of the suspension and were centrifuged for 10 min starting at 1,000 x g and slowly rising the 344 RCF to 20,000 x g. The supernatant was carefully removed and additional suspension was added. 345 The step was repeated until the particles of the suspension were completely transferred. In case 346 the suspension spilled out it helped to reduce the volume in the tip. The sample was tightly 347 compressed with the screw cap of the insert and given into a 4 mm MAS rotor for solid-state NMR 348 spectroscopy. Spectra were recorded by direct [29] Si excitation at a sample spinning rate of 10 kHz 349 within 1–4 days according to 250–1000 scans with a recycle delay of 360 seconds on a BRUKER 350 Avance 300 MHz spectrometer using a 4 mm double resonance {H,X} MAS NMR probe (Bruker 351 BioSpin, Karlsruhe, Germany). 352 **Determination of the average radius, volume and silicon amount of** _**C. nana**_ 353 Cultures of _C. nana_ in the exponential growth phase were concentrated by centrifugation and 354 washed several times with ASW-Si to remove dissolved silicic acid. Aggregation must be avoided 355 during centrifugation and resuspension, since aggregates distort cell counting. The cell density 356 (n > 16) was determined on the automated cell counter. The suspension was alkali-treated and tSi 357 was determined using the MBA and the standard addition method (n = 7). The determined average 358 silicon content of a cell in exponential growth was 138 ± 5 fmol (95% confidence interval (CI), 359 n = 3). For volume determination, the cell size was determined from microscopic images using 360 _ImageJ_ software. The average radius amounted to 2440 ± 46 nm (95% CI, microscopically 361 determined as the average over n = 414 cells). The mean volume of a cell in exponential growth 362 was 88.1 ± 3.9 fL (95% CI, microscopically determined as the average over n = 204 cells) assuming 363 a cylindrical cell shape. ## 364 Results 365 **Quantification of SSPs by PDMPO pulse labeling** 366 In order to visualize and estimate the silicon amount in SSPs of _C. nana_, the following indirect 367 approach was applied (cf. Fig. 2, top). Cells in the exponential growth phase are transferred into 368 silicon-free artificial seawater (ASW-Si) and pulse labeled with PDMPO. After 48 h of incubation, 369 fluorescence microscopy reveals new biosilica formed after pulse labeling whereas previously 370 formed biosilica is not labeled because PDMPO is only incorporated during biogenesis/ silica 371 precipitation (Shimizu et al., 2001). Since, however, no external silicon is available, biosilica 372 synthesis can only occur at the expense of SSPs present before pulse labelling and the 373 fluorescence intensity thus reflects this SSP-derived biosilica. 374 Remarkably, PDMPO pulse labeling results in detectable fluorescence for more than 99% 375 of the cells (Fig. S1) after 48 h in silicon-free medium (Fig. 2b). Cell division is recognizable by the 376 presence of labeled valve pairs. For quantification, a cell culture in the exponential growth phase 377 was fully PDMPO labeled in Si-containing artificial seawater (ASW) and used as reference (Fig. 378 2a). - 9 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 379 380 **Figure 2** . Top: Scheme illustrating the experiment for quantifying the SSPs based on fluorescence 381 intensities of PDMPO in cellular biosilica after full labeling (left) and pulse labeling in silicon-free 382 medium (right, ASW-Si: Si-free ASW). Bottom: Microscopic images show the fluorescence of 383 PDMPO in the biosilica of methanol extracted cells of _C. nana_ after full labeling (a) and pulse 384 labeling in silicon-free medium (b). 385 The cells were extracted with methanol prior to microscopic imaging to remove PDMPO 386 that is not incorporated into biosilica (McNair et al., 2015). The fluorescence intensities were 387 determined using the software _ImageJ_ . Cell counting was performed manually. To estimate the 388 SSPs, the quotient of the mean fluorescence intensities of the pulse labeled cells (Si starved) and 389 the fully labeled cells (exponential growth) was calculated. The obtained result of 34 ± 6% thus 390 represents the average fraction of SSP-silicon relative to the total silicon amount of a cell (tSi cell ) 391 in exponential growth. Six images with an average of 90 cells were evaluated from each sample. 392 Note that the experimental error (95% CI) represents the average fluorescence intensity variation 393 of a cell in these images (n = 6), it does not reflect the variability within the cell culture. Also note 394 that the applied fluorescence technique would only detect the entire amount of SSPs after their 395 complete utilization by the cells. To come close to this situation, we have chosen 48 h growth 396 period for the experiments and observed fast growth only within 24 h, at maximum 48 h after 397 transfer to ASW-Si. Afterwards, cell growth slows down severely or stops completely. 398 The described estimation of SSPs in _C. nana_ relies on the observation that PDMPO 399 incorporation into cell wall biosilica, also referred to as biogenic silica (bSiO 2 ), occurs with 400 constant molar bSiO 2 : PDMPO ratio - independent of the external silicic acid concentration in the 401 medium. The fluorescence intensity can thus be used as a quantitative biosilica marker (Leblanc - 10 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 402 & Hutchins, 2005; McNair et al., 2015). However, there is also preliminary work indicating an 403 increasing molar bSiO2 : PDMPO ratio if the external silicon concentration falls below 3 µM 404 (McNair et al., 2015). This would lead to an overestimation of SSPs and the determined value of 405 34% should then be considered as an upper bound. However, the observation of an increasing 406 bSiO 2 : PDMPO ratio for very low Si concentrations (< 3 µM) suffers from an inherent methodical 407 problem of the [32] Si-labeling experiment as discussed below in detail (discussion section). 408 In our experiments, a fluorescence standard - the fully PDMPO-labeled cells - were used 409 as reference. Therefore, we obtain the dimensionless fraction representing the SSPs in relation to 410 the total cellular Si amount tSi cell (see above). In addition, this experimental setup ensures that 411 species-specific variability of the molar bSiO2 : PDMPO incorporation ratio has no effect. From 412 this fraction, the Si concentration of SSPs [tSi SSP ] = 540 ± 100 mM and the corresponding cellular 413 Si amount of tSi SSP = 47 ± 8 fmol/cell were determined using the measured total cellular Si amount 414 tSi cell = 138 ± 5 fmol and a cell volume of 88.1 ± 3.9 fL (Tab. 1) (materials and methods). Four 415 additional diatom species have been analyzed with this fluorescence method, their estimated 416 SSPs range from ca. 4% to 33% of tSi cell (Tab. S1 and Fig. S2 – S5). 417 ASW-Si was tested for silicon contaminations using the MBA and standard addition. A 418 trace concentration of 0.26 ± 0.2 µM Si was found. The LOQ was determined to be 0.17 µM 419 (materials and methods). In order to minimize the influence of that contamination, the cell 420 suspensions were adjusted to a high density before silicon starvation (ca. 1.0 · 10 [6] cells/ml for 421 C. nana). As a result, the contamination would account for ca. 0.26 fmol silicon per cell. 422 Compared to the tSi cell of ca. 138 fmol (Tab. 1), the influence of the contamination is negligible. 423 In this context, it should be mentioned that the physical dissolution of pre-existing 424 biosilica is considered as a potential undesired source for silicon release. As far as we know, there 425 is no data available on how fast this process occurs on living cells. We considered dead cells to 426 be a potential pathway for silicon release. A low dead cell count of < 1% was thus ensured before 427 the starvation experiment. 428 It should be emphasized, that the obtained value of 34% is considered as an upper bound 429 of the SSPs proportion on tSi cell . An important issue is the uncertainty about the precise ratio of 430 bSiO 2 : PDMPO in the biosilica synthesized during silicon starvation, as discussed above and in 431 detail in the discussion section. Furthermore, it is not certain that SSPs are indeed stainable with 432 PDMPO in the exponentially growing cells (control), since the chemical state of SSPs is still not 433 fully clear. In that case, the resulting SSPs would be reduced to 25% of tSi cell . 434 435 **Table 1** . _C. nana_ cell parameters: Average cell volume, total cellular silicon amount tSi cell and 436 concentration [tSi cell ] determined using the Molybdenum blue assay. The silicon amount tSi SSP and 437 concentration [tSi SSP ] of SSPs was estimated based on the described fluorescence method. The 439 |experimental error is given as 95% CI. volume|88 ± 4 fL| |---|---| |volume|88 ± 4 fL| |tSicell|138 ± 5 fmol| |[tSicell]|1570 ± 90 mM| |tSiSSP|47± 8fmol| |[tSiSSP]|540± 100 mM| - 11 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 440 **Density gradient centrifugation** 441 A continuous sucrose density gradient was used to fractionate exponentially grown cells after 442 bead beating cell lysis. For control, extracted cell walls were treated and analysed in parallel. 443 Centrifugation was performed for 60 min at 4,200 x g and 4°C. With that parameters separation is 444 predominantly based on particle size. The total amount of silicon tSi in the obtained fractions was 445 determined after alkaline treatment with MBA using the standard addition method (n = 5) 446 (materials and methods). Fig. 3 shows the tSi-values of the individual fractions in percent. These 447 proportions refer to tSi of the entire sample (tSi sample ), i.e., the sum of all fractions. Since tSi sample is 448 just a multiple of the average silicon amount of a cell tSi cell, the proportion of the fractions on 449 tSi sample is equal to their proportion on tSi cell . 450 451 **Figure 3** . Top, left: Photograph of the sucrose density gradient with cell lysate after centrifugation. 452 Diagrams showing the tSi-values of the different fractions for the cells (a) and the extracted cell 453 walls (b) as fractions of tSi cell . c) Difference between the curves for cells and extracted cell walls. 454 This difference represents an estimate for the cytoplasmic silicon, which is assumed to be part of 455 the SSPs. * Upper bound radius for pure silica particles (2.20 g/ml) calculated with the Stokes’ 456 law. Larger silica particles – which may be part of the SSPs as well – will be found in the pellet (Fig. 457 S6 and App. S1). Error bars: 95% CI. 458 Extracted cell walls after bead beating treatment (Fig. 3b) behave pronouncedly different 459 from the cell lysate. They almost completely sediment in the pellet (95.4%) and only 2.1% are 460 found in in the fraction directly above the pellet. Lysed cells (Fig. 3a) exhibit significantly higher 461 silicon contents especially in the fractions beyond a density of ca. 1.18 g/ml. This increase starts 462 below the dark visible band containing the chloroplasts (Fig. 3, top left). The reference sample 463 (extracted cell walls) indicates that only limited amounts of cell wall fragments are found in the - 12 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 464 fractions on top of the pellet. This approach of using extracted cell walls as control has already 465 been applied in our previous study (Reichelt et al., 2023). In the present study, it indicates once 466 again that silicon-containing particles (any combination of silicon with organic material of 467 unknown constitution and chemical form) are present in the cytoplasm. By calculating the 468 difference between the quantities found in the gradient of the cell lysate on the one side and the 469 extracted cell walls (control) on the other side, it is concluded that these silicon-containing 470 particles make up 10 ± 0.7% of tSi cell in the lysed cells. The conclusion that those silicon 471 containing particles are likely originating from the cytoplasm and not the cell walls, follows from 472 the fact that cell wall fragments (control) are sedimented almost completely and are missing in 473 the gradient above the pellet, whereas the cell lysate shows high silicon proportions in the 474 gradient. 475 **Isopycnic density gradient ultracentrifugation** 476 To further improve the separation of SSPs, ultracentrifugation was performed with media of high 477 density. In order to obtain (near)isopycnic condition, high RCF was applied. In this case separation 478 is based solely on the particle density. A stepped gradient was used consisting of isoosmolar 479 OptiPrep (iodixanol) solutions up to 1.32 g/ml density. Nycodenz (iohexol) was then used for 480 densities of 1.37 and 1.43 g/ml at hyperosmolar conditions. Centrifugation was performed at 481 175,000 x g and 4°C for 1 h (materials and methods). According to calculations, isopycnic 482 conditions apply to cytoplasmic particles with a radius exceeding ca. 30-50 nm (Fig. S7). Such 483 particles reach equilibrium (buoyant density) after 60 min centrifugation. Silica has a density well 484 above the gradient (ρ biosilica ≈ 2.2 g/ml). Consequently, there is no equilibrium state in the gradient 485 for pure silica particles. Calculations show that silica particles reach the pellet within the runtime 486 of 60 min, if their radius is greater than 40 nm. Below that threshold value, the particles remain in 487 the gradient (Fig. S7). The tSi-values of the various fractions were determined by MBA after alkaline 488 treatment using standard addition (n = 5) for calibration (materials and methods). After 489 ultracentrifugation of lysed cells (Fig. 4a), 92.0% of the silicon was found in the pellet and 8.0 ± 490 0.4% in the fractions of the gradient. A significant amount of tSi representing 5.8 ± 0.4% of tSi cell 491 was found in the upper half of the gradient in the density range of 1.01–1.27 g/ml (Fig. 4a), whereas 492 the control of extracted cell walls only contained 0.48 ± 0.05% in this range (Fig. 4b). 493 Consequently, almost no cell wall fragments are located in this density range and the observed 494 difference of 5.3 ± 0.4% (relative to tSi cell ) thus represents silicon species being part of the SSPs 495 (Fig. 4c). The silicon found at densities higher than 1.27 g/ml, in the density range of 1.32–1.43 496 g/ml, mainly consists of cell wall fragments, as shown by the control. The concentrations are 497 significantly lower than in the fractions above. The presumed SSPs from the cytoplasm are 498 therefore clearly separated from the cell walls, that are predominantly found in the pellet. - 13 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 499 500 **Figure 4** . Top: Photographs of the density gradients after ultracentrifugation and diagrams 501 showing the corresponding tSi in the lysed cell fractions from cultures in exponential growth (a) 502 and the control experiment with extracted cell walls (b). tSi-values are fractions of tSi cell . 503 c) Difference between the curves for lysed cells and extracted cell walls providing an estimate for 504 cytoplasmic silicon which is assumed to be part of the SSPs (Fig. S7 and App. S1). Error bars: 95% 505 CI. 506 The highest contents of the cytoplasmic silicon are found at densities of 1.21–1.27 g/ml. It 507 is concluded based on calculations that the silicon-containing SSP constituents located in that 508 fractions are in equilibrium state and have a density very close or equal to the density of the 509 gradient in this region. This follows from the fact that particles of a higher density, such as pure 510 silica, would need to have very small radii of approximately 10–16 nm to reach this region of the 511 gradient (Fig. S7). However, such particles would have been detected in the sucrose gradient of 512 the previous investigation, namely in the uppermost fraction with the lowest density (Fig. S6). As 513 already shown, there is no cytoplasmic silicon found in this fraction (Fig. 3c) and consequently 514 such particles do not occur in a pure and free form. 515 **29** **Si NMR spectroscopy** 516 For [29] Si solid-state NMR analysis, [29] Si-enrichment was performed by cultivation of _C. nana_ in a 517 growth medium with [29] Si-enriched silicic acid under standard conditions. After the harvest of the 518 exponentially grown cells, half of the culture was transferred to a silicon-free medium and 519 incubated for 24 h to obtain silicon starved cells (materials and methods). 520 - 14 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 521 522 **Figure 5** . Solid-state [29] Si MAS NMR spectrum of exponentially grown whole cells (b) with the 523 characteristic Q [n] group chemical shift regions. For comparison, the liquid-state [29] Si NMR 524 spectrum of a 4 mM silicic acid solution at pH 9 is inserted (a) showing the signals of the soluble 525 Si species, monosilicic acid (Q [0] ) and disilicic acid (Q [1] ). These soluble species are not detected in 526 whole cells (region between the dotted lines). c) The intensities of the detected Q [n] group signals 527 (n = 2,3,4) depend on the degree of silicic acid polycondensation (Belton et al., 2012; McCutchin 528 et al., 2025). 529 The solid-state [29] Si MAS NMR spectrum of the exponentially grown whole cells (Fig. 5b) 530 shows the characteristic signals of biosilica representing Q [n] groups (Fig. 5c). The exponent n 531 indicates the degree of polycondensation: n = 0 for monosilicic acid Si(OH) 4 and n = 4 for 532 completely polycondensed silica SiO 2 . In the spectrum, the Q [4] signal occurs with the highest 533 intensity, followed by Q [3] . The group Q² is hardly detectable. This is in line with the signals typically 534 observed in biosilica (Belton et al., 2012; S. Mann et al., 1983). Q [1] and Q [0] do not appear in the 535 solid-state NMR spectrum (Fig. 5b). However, Q [1] and Q [0] are the predominant signals in the liquid 536 state NMR spectrum of a synthetic silicic acid solution (Fig. 5a) and represent the soluble species, 537 disilicic acid (Q [1] ) and monosilicic acid (Q [0] ). The detection limit of [29] Si MAS NMR spectroscopy 538 derived from the spectra in Figure 6 b corresponds to a silicic acid (monomer or dimer) 539 concentration of about 1 mM. That means, the presence of significant amounts of silicic acid in 540 whole cells can be excluded. It is concluded that silicic acid is rapidly transformed after uptake, 541 probably to polycondensed species. That means, silicic acid, either in a stabilized or a covalently 542 bound form does not represent large parts of the SSPs in _C. nana_ as already observed previously 543 (Gröger et al., 2008). 544 The totally labeled cells were lysed and fractionated by ultracentrifugation as described 545 before (materials and methods). The cytoplasmic silicon content in the density range of 546 1.01 – 1.27 g/ml was determined to be 2.8 ± 0.6% of tSi cell at exponential growth (Fig.6a) and 547 3.7 ± 0.3% of tSi cell after silicon starvation (Fig. 6d). Both fully labeled samples came from the 548 same batch and are therefore comparable. 549 550 - 15 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 551 552 **Figure 6** . Left: Silicon contents of fractions obtained after density gradient ultra-centrifugation 553 (a,d). Right: Solid-state [29] Si MAS NMR spectra of whole cells (c,f) and combined fractions (b,e) 554 between 1.21 and 1.27 g/ml (dark green bars). Top: Exponentially grown cells (a,b,c). Bottom: 555 Silicon starved cells (d,e,f). Error bars: 95% CI. 556 These values are lower compared to the previous separation by ultracentrifugation with 557 unlabeled exponentially growing cells (5.3 ± 0.4% of tSi cell ). Note that the experimental errors 558 shown in the graphs refer to the silicon quantification method and were derived from the linear 559 regression. These errors do not include the variation of the sample variability and separation 560 protocol. The origin of the large difference is probably the variability between the two different 561 cultivation batches, as diatom cultivation is a complex process and less controlable compared to 562 the separation procedure. This hypothesis is qualitatively supported by the fact that the 563 distribution of silicon over the fractions is similar between the two labeled samples (Fig. 6a and 564 Fig. 6d). Nonetheless, the highest silicon concentrations in the gradient are again observed 565 between 1.21 and 1.27 g/ml as already described above (dark green bars in Fig. 6a,d). These two 566 fractions were combined for NMR measurements of the silicon-containing cytoplasmic 567 constituents (Fig. 6b,e). Spectra of the whole cells were also recorded (Fig. 6c,f) and the integrals - 16 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 568 of Q [4] and Q [3] were obtained by deconvolution using the software _dmfit_ . The Q [4] /Q [3] ratios are 569 provided in the figures. Obviously, the cytoplasmic silica species exhibit a significantly lower 570 degree of silica polycondensation than whole cells which are dominated by cell wall silica. This is 571 in line with results obtained on various species including _C. nana_ that already pointed towards 572 pre-condensed silica species in the cytoplasm or the SSPs of diatoms (Bertermann et al., 2003; 573 Gröger et al., 2008). 574 Comparison of the results for the cultures in exponential growth and after silicon starvation shows 575 no significant differences in the NMR spectra and silicon contents (Fig. 6) except for slightly 576 narrower signals, i.e., better spectral resolution for the sample in exponantial growth and slightly 577 higher cytoplasmic silicon content in silicon starved cells. This implies that the cytoplasmic 578 silicon is not depleted during silicon starvation in agreement with cryoFIB-SEM-EDS studies on 579 _C. nana_ (Kumar et al., 2020). On the other hand, the above-described PDMPO labeling 580 experiments clearly show that SSPs are used for biosilica formation during silicon starvation. This 581 should result in a corresponding depletion of the SSPs. It is thus concluded that the clearly 582 observed silica biosynthesis during silicon starvation mainly uses up other SSPs than the species 583 in the cytoplasma separated by our centrifugation experiments. Consequently, there are likely two 584 different silicon pools located in the cell. Based on our observations, we hypothesize the 585 existence of two different silicon pools within the cell. First, the cytoplasmic part of the SSPs. It 586 was detected and characterized through successful separation and following analysis. 587 Cytoplasmic SSPs are characterized by low density (<1.27 g/ml) and are not depleted during 588 silicon starvation (homeostatic). In addition, we postulate the existence of a high-density fraction 589 of SSPs. They cannot be separated using the employed centrifugation techniques because they 590 sediment together with cell wall fragments due to their high density (>1.43 g/mL). That means, we 591 hypothesize the existence of high-density SSPs based on two observations: (i) The significant 592 difference between the results of the PDMPO labeling experiment (SSPs = 34% of tSi cell ) and the 593 quantification of cytoplasmic silicon (SSPs = 3–10% of tSi cell ). Comparing these results, the high594 density SSPs could potentially account for up to about 30% of tSi cell . (ii) Cytoplasmic SSPs are not 595 depleted during silicon-starvation although cell growth continues. It is tempting to speculate that 596 these high-density SSPs represent a distinct cellular (sub)structure. Nevertheless, alternative 597 explanations are also possible. The cells could mobilize silica species which are tightly bound to 598 the frustules or even mobilize cell-wall biosilica for the synthesis of new frustules. Such 599 hypothetical processes could be subject of future studies. ## 600 Discussion 601 The quantification of the described fluorescence imaging experiments using PDMPO pulse 602 labeling under silicon starvation unequivocally confirms the existence of substantial amounts of 603 silicon stored in SSPs. The use of PDMPO as a proxy for the production of biogenic silica bSiO 2 604 provides a relatively simple implementation compared to the expensive and time consuming but 605 very accurate radioisotope labeling with [32] Si (McNair et al., 2015; Shipe & Brzezinski, 1999). The 606 fluorescence-based method presented here relies on studies of cell cultures grown for a limited 607 time under external silicon starvation. Freshly formed silica must then necessarily be synthesized 608 at the expense of intracellular SSPs. Based on previous work (McNair et al., 2015), the 609 fluorescence intensity of silica-incorporated PDMPO is quantified after pulse labeling. 610 Quantification of this freshly synthesized silica allows to estimate the Si content of the SSPs. To 611 this end, the measured fluorescence intensity from cells under silicon limitation is normalized by - 17 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 612 the intensity measured for cells in ASW without silicon deficiency. The ratio between the 613 respective fluorescence intensities is a measure for the Si-content in SSPs relative to the total 614 cellular silicon content tSi cell at exponential growth. Species-specific variability of the molar bSiO2 615 : PDMPO incorporation ratio is ruled out by that experimental setup. Five diatom species were 616 examined by this method under laboratory conditions (Tab. S1 and Fig. S1 – S5). For _C. nana_, this 617 method predicts that the SSPs contain up to ca. 34% of tSi cell . 618 It is important to note that this quantification relies on the assumption that the bSiO2 : 619 PDMPO ratio in the formed biosilica does not depend on the external Si concentration in the 620 growth medium. Molar ratios (bSiO₂ : PDMPO) of 2800 ± 700 (Leblanc & Hutchins, 2005) and 621 2916 ± 708 (McNair et al., 2015) were reported when incorporating PDMPO into biosilica under 622 laboratory conditions. In field studies with a large number of samples that contained whole 623 diatom communities, this relation was confirmed to apply with only slight variations (Leblanc & 624 Hutchins, 2005; McNair et al., 2015). Although, it should be noted that PDMPO incorporation 625 varied by a factor of three among diatom groups (McNair et al., 2015). However, the correlation 626 was only shown to be valid at silicic acid concentrations beyond 3 µM in the growth medium. 627 Below 3 µM, an increasing amount of PDMPO is apparently incorporated into biosilica (McNair et 628 al., 2015) and the incorporation of PDMPO correlates inversely with the silicic acid concentration 629 (Durkin et al., 2013). McNair et al. determined a ratio bSiO 2 : PDMPO = 912.6 x [Si(OH) 4 ] for silicic 630 acid concentrations below 3 µ M. This behavior is explained by the change in cell wall 631 composition. Silicon deficiency leads to a higher concentration of biosilica-incorporated organic 632 molecules. It is assumed that PDMPO replaces some of the native organic molecules in the 633 biosilica thus leading to higher concentration of incorporated PDMPO (McNair et al., 2015). 634 However, the studies of McNair et al. compared the incorporation of PDMPO to the 635 incorporation of [32] Si, a very sensitive radiomarker (Shipe & Brzezinski, 1999). In short, this method 636 relies on the quantification of biogenic silica based on the measurement of [32] Si uptake from the 637 growth medium, i.e., bSiO 2 ~ [32] Si incorporation (Krause et al., 2011; McNair et al., 2015). This 638 works very well at high external silicic acid concentrations. But at very low concentrations, pre 639 existing SSPs in the cells are increasingly mobilized and used for biogenic silica formation in 640 addition to the uptake from the surrounding medium. That means, bSiO 2 is then not reflected 641 solely by the incorporated [32] Si but increasingly due to unlabeled silicon originating from mobilized 642 SSPs, i.e., bSiO 2 ~ [32] Si incorporation + unlabeled Si from SSPs. The radioisotope [32] Si will thus only 643 represent the continuously decreasing, small biosilica fraction formed from external silicic acid 644 taken up under silicon limitation. Given the considerable size of SSPs, mobilization of silicon from 645 SSPs may well explain the apparent increase of the bSiO2 : PDMPO ratio found by the [32] Si-based 646 method at low external Si concentrations. In contrast, sufficient amounts of silicon are available 647 from the environment at higher external silicic acid concentrations and the SSPs are then not 648 extensively used for biosilica synthesis. 649 In our experiments using PDMPO labeling instead of [32] Si, the silicon mobilized from the 650 SSPs will become PDMPO-labeled because the SSP-silicon is transported into the SDV like silicon 651 freshly taken up from the medium and then precipitated whereby PDMPO is incorporated. In 652 summary, the silicon amount of 34% in SSPs relative to tSi cell determined by the fluorescence653 based method is definitely an upper bound, but probably even a realistic estimation for the silicon 654 content of SSPs. Given a total cellular Si content of 138 fmol and an average cell volume of 88 fL, 655 this would correspond to a intracellular concentration of ca. 540 mM – far beyond the solubility 656 limit of 2 mM at near-neutral pH (Iler, 1979). - 18 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 657 The described separation experiments based on cell lysis by bead beating and 658 subsequent density gradient centrifugation succeeded in separating cytoplasmic silicon. Typical 659 values represent ca. 3 – 10% of tSi cell . This corresponds to intracellular concentrations of 50 – 660 160 mM which also strongly exceeds the solubility limit. However, it is likely that this cytoplasmic 661 fraction only represents part of the SSPs. Furthermore, isopycnic density gradient 662 ultracentrifugation allowed to successfully separate cytoplasmic Si-species of _C. nana_ for the 663 first time. The cytoplasmic fraction with a density of 1.01 – 1.27 g/ml shows silicon contents 664 representing about 3 – 5% of tSi cell corresponding to an intracellular concentration of 50 – 80 mM. 665 A maximum is located in the range of 1.21 – 1.27 g/ml. The content of cytoplasmic silicon is stable 666 and appears to be relatively independent of the culture conditions, as shown by a comparison of 667 cells in exponential growth with silicon-starved cells. The spectra of the cytoplasmic silicon 668 fraction are clearly assigned to polycondensed silica species. However, the degree of silica 669 polycondensation is significantly lower than that of the corresponding whole cells which 670 predominantly consist of cell wall biosilica. Consequently, the cytoplasmic silicon consists of 671 silica in pre-condensed form. The presence of such silica was already shown in _C. nana_ and other 672 species (Bertermann et al., 2003; Gröger et al., 2008). Here, it could be investigated for the first 673 time in separated form. Note that the presence of significant amounts of siliceous cell walls or 674 their fragments in the cytoplasmic fractions is unlikely for several reasons. First, this is indicated 675 by the sedimentation of the control sample consisting of extracted cell walls. On the other hand, 676 silica fragments produced by cell lysis may sediment differently than the control. However, such 677 biosilica fragments cannot be the dominant part of the detected Si in cytoplasmic fractions 678 because such fragments would exhibit the same degree of silica polycondensation, i.e., the same 679 Q [4] /Q [3] ratio as the cell walls. This is contradicted by the observations of [29] Si MAS NMR 680 spectroscopy (see Figure 6) as well as by previous studies on frozen whole cells from 681 synchronized cell cultures (Gröger et al. 2008). This underlines the successful separation of SSPs. 682 Moreover, similar concentrations of intracellular silicon were also reported based on non-invasive 683 techniques like cryoFIB-SEM-EDS (Kumar et al., 2020) which further supports our results. 684 A remarkable observation is that the silicon content and the NMR spectra of the 685 cytoplasmic silicon remain relatively unchanged when comparing exponentially growing cells 686 with silicon-starved cells. In the first place, this contradicts the assumption that the SSPs are 687 depleted or at least reduced during silicon starvation. On the other hand, the intense growth 688 under silicon starvation observed by the PDMPO labeling experiment clearly demonstrates the 689 extensive use of SSPs and supports their depletion. SSPs could account for up to 1/3 of the tSi cell 690 and thus be significantly larger than previously assumed. This contradiction is resolved by 691 assuming that the synthesis of the cell wall biosilica occurs at the expense of a source other than 692 cytoplasmic silicon. It is concluded that SSPs are consisting of two separate pools: The smaller 693 cytoplasmic low-density fraction (2.8 – 5.3% of tSi cell and <1.27 g/ml) which appears to be stable 694 or homeostatic and a larger high-density fraction (<30% of tSi cell and >1.43 g/ml) which likely 695 represents the actual storage pool. 696 In order to separate the high-density fraction of the SSPs, the density gradient 697 centrifugation at lower RCF (rate-zonal) is a suitable technique, because separation is mainly 698 based on particle size (Fig. 3). A calculation of the particle migration in the sucrose gradient using 699 the applied parameters (materials and methods) shows that silica particles (2.20 g/ml) are still 700 located in the gradient above the pellet if their radius is below 300 nm (Fig. S6 and App. S1). This 701 is more than a tenth of the average radius of _C. nana_ (2440 ± 46 nm). It should be noted that the 702 calculation does not take into account that the particles can deviate from spherical shape, and - 19 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 703 that the silica particles might have a lower density due to attached organic material or 704 compartimentalization. Both factors would further increase the calculated threshold value for 705 their effective radius. Taking into account the small cell size, it is quite unlikely that such large 706 particles, aggregates or compartments represent the SSPs or a precursor for cell wall formation. 707 Instead, our observations rather point towards the attachment of significant parts of SSPs to the 708 cell walls. It should be noted that a similar hypothesis was already put forward by Lewin in 1954. 709 It was based on the observation that a silicon-starved _N. pelliculosa_ culture took up larger 710 amounts of silicic acid under light deprivation. During that period, the cell and valve counts 711 remained unchanged. It was concluded from this observation that no growth took place and 712 silicic acid might be accumulation by attachment, i.e., thickening of the pre-existing siliceous cell 713 walls (Lewin, 1954). Further research should therefore focus on investigating a possible 714 attachment of SSPs and the pre-existing cell walls. ## 715 Conclusion 716 The knowledge on SSPs in _C. nana_ gained by the present study as well as previous work is 717 summarized in Tab. 2. The cytoplasmic silicon content determined here by separation 718 experiments is 3–10% of tSi cell (50-160 mM) and agrees well with the previously determined value 719 of 9% of tSi cell (Reichelt et al., 2023). The resulting cytoplasmic Si concentration of 158 ± 14 mM 720 agrees well with the intracellular silicon concentration of 140 mM determined by cryo-FIB-SEM 721 EDS experiments (Kumar et al., 2020). This cytoplasmic silicon concentration is, however, 722 significantly smaller than the value of 34% of tSi cell estimated for the SSPs by PDMPO labeling. 723 **Table 2** . Overview of the methods applied in this investigation and in other studies with the 724 quantitative estimate obtained for the SSPs of _C. nana_ in exponential growth. The experimental 725 error is given as 95% CI. * Calculated according to Stokes’ law assuming pure silica particles 726 (Fig. S6-S7 and App. S1). [†] Recalculated with the data from the present study (App. S1). [‡] Same 727 cultivation batch. [a] (Reichelt et al., 2023) [ b ] (Martin-Jézéquel et al., 2000) [ c ] (Thamatrakoln & 729 |Hildebrand, 2008) d (Kumar et method|- t al., 2020). characteristic / information gain|[SSP] / % of tSi|[SSP] / mM| |---|---|---|---| |**method**|**characteristic / informationgain**|**[SSP] / % of tSicell**|**[SSP] / mM**| |PDMPOpulse labeling|mobilized during silicon starvation|34.3± 5.7|538± 96| |density gradient centrifugation|particle radius < 300 nm *|10.1± 0.7|158± 14| |isopycnic ultracentrifugation|density < 1.27 g/ml|5.3± 0.4<br>2.8± 0.6‡<br>3.7± 0.3 ‡|83± 7<br>44± 9<br>58 ± 6 (Si starvation)| |differential centrifugation a|particle radius < 200 nm *|9|141†| |extraction at 95 °Cb, c|water extractable|0.5 – 3.5<br>1.3|- <br>-| |cryoFIB-SEM-EDS d|located in the intracellular space|- <br>-|140<br>190(Si starvation)| 730 Solid-state NMR investigations show that the cytoplasmic silica has a lower polycondensation 731 degree than whole cells, confirming previous reports about the existence of pre-condensed silica 732 species in SSPs. Under silicon starvation, the separated cytoplasmic silicon remains practically 733 unchanged compared to cultures without silicon starvation and the [29] Si MAS NMR spectra do also 734 not significantly change. Together with the higher amount of SSPs predicted by the fluorescence735 based method (34% of tSi cell ), it seems likely that the cytoplasmic low-density silicon fraction 736 analyzed here only represents a smaller part of the SSPs. The assumption that the presumably 737 much larger fraction of the SSPs consist of large particles, compartments or aggregates seems - 20 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 738 unlikely in light of particle size calculations based on Stokes’ law. The preferred hypothesis for the 739 location of that missing inseparable high-density part of the SSPs is a connection or association 740 to the cell walls. 741 **Author contributions** 742 **Eike Brunner** : funding acquisition (lead); project administration (lead); conceptualization (equal); 743 supervision (equal); writing - original draft preparation (equal); writing - review and editing (equal). 744 **Tobias** **Reichelt** : methodology (lead); formal analysis (lead); visualization (lead); 745 conceptualization (equal); supervision (equal); writing - original draft preparation (equal); writing 746 - review and editing (equal); investigation (equal); validation (equal). **Julia Breyer** : investigation 747 (equal); validation (equal). All authors have read and agreed to the published version of the 748 manuscript. 749 **Acknowledgments** 750 Financial support by the German Research Foundation (Deutsche Forschungsgemeinschaft, 751 DFG, Project No. Br 1278/35-1) is gratefully acknowledged. The authors gratefully acknowledge 752 Prof. Dr. Thomas Henle (chair of food chemistry – Dresden university of technology) for the 753 possibility to use the ultracentrifuge and Kristina Seliverstova for the instruction and support on 754 that instrument, Dr. Silvia Paasch for supervision and support during NMR experiments, 755 Christiane Leudolph for support in the laboratory work, Yousef Pousti for performing many of the 756 PDMPO labeling experiments and Dr. Susanne Machill for helpful discussions. Figures 1 and 2 757 were prepared with the software “biorender”. 758 **Data Availability Statement** 759 The data presented in this study are available on request from the corresponding author. 760 **Conflicts of interest disclosure** 761 The authors declare no conflict of interest. # 762 References 763 Alverson, A. J., Beszteri, B., Julius, M. L., & Theriot, E. C. (2011). The model marine 764 diatom Thalassiosira pseudonana likely descended from a freshwater ancestor in 765 the genus Cyclotella. _BMC evolutionary biology_, _11_ (1), 125 766 Annenkov, V. V., Danilovtseva, E. N., Likhoshway, Y. V., Patwardhan, S. V., & Perry, C. C. 767 (2008). Controlled stabilisation of silicic acid below pH 9 using poly(1 768 vinylimidazole). _Journal of Materials Chemistry_, _18_ (5), 553–559. 769 https://doi.org/10.1039/B716367N 770 Annenkov, V. V., Verkhozina, O. N., Zelinskiy, S. N., Shishlyannikova, T. A., Bridoux, M. C., 771 & Danilovtseva, E. N. (2018). Unusual Polyamines from Baikalian Diatoms. 772 _ChemistrySelect_, _3_ (33), 9708–9713. https://doi.org/10.1002/slct.201802032 773 Azam, F., Hemmingsen, B. B., & Volcani, B. E. (1974). Role of silicon in diatom 774 metabolism. V. Silicic acid transport and metabolism in the heterotrophic diatom 775 Nitzschia alba. _Archives of Microbiology_, _97_ (2), 103–114. 776 https://doi.org/10.1007/BF00403050 - 21 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 777 Belton, D. J., Deschaume, O., & Perry, C. C. (2012). An overview of the fundamentals of 778 the chemistry of silica with relevance to biosilicification and technological 779 advances. _The FEBS Journal_, _279_ (10), 1710–1720. https://doi.org/10.1111/j.1742 780 4658.2012.08531.x 781 Bertermann, R., Kröger, N., & Tacke, R. (2003). Solid-state 29Si MAS NMR studies of 782 diatoms: Structural characterization of biosilica deposits. _Analytical and_ 783 _Bioanalytical Chemistry_, _375_ (5), 630–634. https://doi.org/10.1007/s00216-003 784 1769-5 785 Biard, T., Bigeard, E., Audic, S., Poulain, J., Gutierrez-Rodriguez, A., Pesant, S., 786 Stemmann, L., & Not, F. (2017). Biogeography and diversity of Collodaria 787 (Radiolaria) in the global ocean. _The ISME Journal_, _11_ (6), 1331–1344. 788 https://doi.org/10.1038/ismej.2017.12 789 Bridoux, M. C., Annenkov, V. V., Menzel, H., Keil, R. G., & Ingalls, A. E. (2011). A new liquid 790 chromatography/electrospray ionization mass spectrometry method for the 791 analysis of underivatized aliphatic long-chain polyamines: Application to diatom 792 rich sediments. _Rapid Communications in Mass Spectrometry_, _25_ (7), 877–888. 793 https://doi.org/10.1002/rcm.4931 794 Bridoux, M. C., & Ingalls, A. E. (2010). Structural identification of long-chain polyamines 795 associated with diatom biosilica in a Southern Ocean sediment core. 796 _Geochimica Et Cosmochimica Acta_, _74_ (14), 4044–4057. 797 https://doi.org/10.1016/j.gca.2010.04.010 798 Brunner, E., Gröger, C., Lutz, K., Richthammer, P., Spinde, K., & Sumper, M. (2009). 799 Analytical studies of silica biomineralization: Towards an understanding of silica 800 processing by diatoms. _Applied Microbiology and Biotechnology_, _84_ (4), 607–616. 801 https://doi.org/10.1007/s00253-009-2140-3 802 Cermeño, P., Falkowski, P. G., Romero, O. E., Schaller, M. F., & Vallina, S. M. (2015). 803 Continental erosion and the Cenozoic rise of marine diatoms. _Proceedings of the_ 804 _National Academy of Sciences of the United States of America_, _112_ (14), 4239– 805 4244. https://doi.org/10.1073/pnas.1412883112 806 Chisholm, S. W., Azam, F., & Eppley, R. W. (1978). Silicic acid incorporation in marine 807 diatoms on light:dark cycles: Use as an assay for phased cell division. _Limnology_ 808 _and Oceanography_, _23_ (3), 518–529. https://doi.org/10.4319/lo.1978.23.3.0518 809 Conley, D. J., Frings, P. J., Fontorbe, G., Clymans, W., Stadmark, J., Hendry, K. R., 810 Marron, A. O., & La Rocha, C. L. de (2017). Biosilicification drives a decline of 811 dissolved Si in the oceans through geologic time. _Frontiers in Marine Science_, _4_, 812 Article 397, 273520. https://doi.org/10.3389/fmars.2017.00397 813 Coombs, J., Darley, W. M., Holm-Hansen, O., & Volcani, B. E. (1967). Studies on the 814 biochemistry and fine structure of silica shell formation in diatoms. Chemical 815 composition of Navicula pelliculosa during silicon-starvation synchrony. _Plant_ 816 _Physiology_, _42_ (11), 1601–1606. https://doi.org/10.1104/pp.42.11.1601 - 22 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 817 Coradin, T., Eglin, D., & Livage, J. (2004). The silicomolybdic acid spectrophotometric 818 method and its application to silicate/biopolymer interaction studies. _Journal of_ 819 _Spectroscopy_, _18_ (4), 567–576. https://doi.org/10.1155/2004/356207 820 Dawson, P. A. (1973). Observations on the structure of some forms of Gomphonema 821 parvulum Kütz. III. Frustule formation. _Journal of Phycology_, _9_ (4), 353–365. 822 https://doi.org/10.1111/j.1529-8817.1973.tb04108.x 823 Drum, R. W., & Pankratz, H. S. (1964). Post mitotic fine structure of Gomphonema 824 parvulum. _Journal of Ultrastructure Research_, _10_, 217–223. 825 https://doi.org/10.1016/s0022-5320(64)80006-x 826 Durkin, C. A., Bender, S. J., Chan, K. Y. K., Gaessner, K., Grünbaum, D., & Armbrust, E. V. 827 (2013). Silicic acid supplied to coastal diatom communities influences cellular 828 silicification and the potential export of carbon. _Limnology and Oceanography_, 829 _58_ (5), 1707–1726. https://doi.org/10.4319/lo.2013.58.5.1707 830 Field, C. B., Behrenfeld, M. J., Randerson, J. T., & Falkowski, P. (1998). Primary production 831 of the biosphere: Integrating terrestrial and oceanic components. _Science_, 832 _281_ (5374), 237–240. https://doi.org/10.1126/science.281.5374.237 833 Grachev, M. A., Bedoshvili, Y. D., Gerasimov, E. Y., Zaikovskii, V. I., Gneusheva, K. V., & 834 Likhoshway, Y. V. (2017). Silica-containing inclusions in the cytoplasm of diatom 835 Synedra acus. _Doklady. Biochemistry and Biophysics_, _472_ (1), 44–48. 836 https://doi.org/10.1134/S1607672917010124 837 Gröger, C., Sumper, M., & Brunner, E. (2008). Silicon uptake and metabolism of the 838 marine diatom Thalassiosira pseudonana: Solid-state 29Si NMR and 839 fluorescence microscopic studies. _Journal of Structural Biology_, _161_ (1), 55–63. 840 https://doi.org/10.1016/j.jsb.2007.09.010 841 Guiry, M. D. (2012). How many species of algae are there? _Journal of Phycology_, _48_ (5), 842 1057–1063. https://doi.org/10.1111/j.1529-8817.2012.01222.x 843 Harrison, P. J., Waters, R. E., & Taylor, F. J. R. (1980). A broad spectrum artificial sea water 844 medium for coastal and open ocean phytoplankton. _Journal of Phycology_, _16_ (1), 845 28–35. https://doi.org/10.1111/j.0022-3646.1980.00028.x 846 Heintze, C., Formanek, P., Pohl, D., Hauptstein, J., Rellinghaus, B., & Kröger, N. (2020). 847 An intimate view into the silica deposition vesicles of diatoms. _BMC Materials_, 848 _2_ (1), 11. https://doi.org/10.1186/s42833-020-00017-8 849 Iler, K. R. (1979). The Chemistry of Silica. _Solubility, Polymerization, Colloid and Surface_ 850 _Properties and Biochemistry of Silica_ . 851 https://cir.nii.ac.jp/crid/1571135649987057536 852 Kinrade, S. D., Del Nin, J. W., Schach, A. S., Sloan, T. A., Wilson, K. L., & Knight, C. T. 853 (1999). Stable five- and six-coordinated silicate anions in aqueous solution. 854 _Science (New York, N.Y.)_, _285_ (5433), 1542–1545. 855 https://doi.org/10.1126/science.285.5433.1542 856 Kinrade, S. D., Gillson, A.-M. E., & Knight, C. T. G. (2002). Silicon-29 NMR evidence of a 857 transient hexavalent silicon complex in the diatom Navicula pelliculosa. _Journal_ - 23 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 858 _of the Chemical Society, Dalton Transactions_ (3), 307–309. 859 https://doi.org/10.1039/B105379P 860 Koistinen, J., Sjöblom, M., & Spilling, K. (2020). Determining dissolved and biogenic 861 silica. _Methods in Molecular Biology (Clifton, N.J.)_, _1980_, 95–101. 862 https://doi.org/10.1007/7651_2018_129 863 Kooistra, W. H., & Medlin, L. K. (1996). Evolution of the diatoms (Bacillariophyta). IV. A 864 reconstruction of their age from small subunit rRNA coding regions and the fossil 865 record. _Molecular Phylogenetics and Evolution_, _6_ (3), 391–407. 866 https://doi.org/10.1006/mpev.1996.0088 867 Krause, J. W., Brzezinski, M. A., & Jones, J. L. (2011). Application of low-level beta 868 counting of 32Si for the measurement of silica production rates in aquatic 869 environments. _Marine Chemistry_, _127_ (1-4), 40–47. 870 https://doi.org/10.1016/j.marchem.2011.07.001 871 Kröger, N., Deutzmann, R., Bergsdorf, C., & Sumper, M. (2000). Species-specific 872 polyamines from diatoms control silica morphology. _Proceedings of the National_ 873 _Academy of Sciences of the United States of America_, _97_ (26), 14133–14138. 874 https://doi.org/10.1073/pnas.260496497 875 Kröger, N., Deutzmann, R., & Sumper, M. (1999). Polycationic peptides from diatom 876 biosilica that direct silica nanosphere formation. _Science_, _286_ (5442), 1129–1132. 877 https://doi.org/10.1126/science.286.5442.1129 878 Kröger, N., Deutzmann, R., & Sumper, M. (2001). Silica-precipitating peptides from 879 _The_ diatoms. The chemical structure of silaffin-A from Cylindrotheca fusiformis. 880 _Journal of Biological Chemistry_, _276_ (28), 26066–26070. 881 https://doi.org/10.1074/jbc.M102093200 882 Kumar, S., Rechav, K., Kaplan-Ashiri, I., & Gal, A. (2020). Imaging and quantifying 883 homeostatic levels of intracellular silicon in diatoms. _Science Advances_, _6_ (42). 884 https://doi.org/10.1126/sciadv.aaz7554 885 Leblanc, K., & Hutchins, D. A. (2005). New applications of a biogenic silica deposition 886 fluorophore in the study of oceanic diatoms. _Limnology and Oceanography:_ 887 _Methods_, _3_ (10), 462–476. https://doi.org/10.4319/lom.2005.3.462 888 Lewin, J., & Chen, C. H. (1968). Silicon metabolism in diatoms. VI. Silicic acid uptake by 889 a colorless diatom, Nitzschia alba Lewin and Lewin. _Journal of Phycology_, _4_ (2), 890 161–166. https://doi.org/10.1111/j.1529-8817.1968.tb04691.x 891 Lewin, J. C. (1954). Silicon metabolism in diatoms. I. Evidence for the role of reduced 892 sulfur compounds in silicon utilization. _The Journal of General Physiology_, _37_ (5), 893 589–599. https://doi.org/10.1085/jgp.37.5.589 894 Lewin, J. C. (1955). Silicon metabolism in diatoms. III. Respiration and silicon uptake in 895 Navicula pelliculosa. _The Journal of General Physiology_, _39_ (1), 1–10. 896 https://doi.org/10.1085/jgp.39.1.1 897 Li, C.-W., & Volcani, B. E. (1985a). Studies on the biochemistry and fine structure of 898 silica shell formation in diatoms. _Protoplasma_, _124_ (1-2), 10–29. 899 https://doi.org/10.1007/BF01279720 - 24 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 900 Li, C.-W., & Volcani, B. E. (1985b). Studies on the biochemistry and fine structure of 901 silica shell formation in diatoms. _Protoplasma_, _124_ (3), 147–156. 902 https://doi.org/10.1007/BF01290765 903 Malviya, S., Scalco, E., Audic, S., Vincent, F., Veluchamy, A., Poulain, J., Wincker, P., 904 Iudicone, D., Vargas, C. de, Bittner, L., Zingone, A., & Bowler, C. (2016). Insights 905 into global diatom distribution and diversity in the world's ocean. _Proceedings of_ 906 _the National Academy of Sciences of the United States of America_, _113_ (11), 907 E1516-25. https://doi.org/10.1073/pnas.1509523113 908 Mann, D. G., & Droop, S. J. M. (1996). 3. Biodiversity, biogeography and conservation of 909 diatoms. _Hydrobiologia_, _336_ (1-3), 19–32. https://doi.org/10.1007/BF00010816 910 Mann, D. G., & Vanormelingen, P. (2013). An inordinate fondness? The number, 911 distributions, and origins of diatom species. _The Journal of Eukaryotic_ 912 _Microbiology_, _60_ (4), 414–420. https://doi.org/10.1111/jeu.12047 913 Mann, S., Perry, C. C., Williams, R. J. P., Fyfe, C. A., Gobbi, G. C., & Kennedy, G. J. (1983). 914 The characterisation of the nature of silica in biological systems. _Journal of the_ 915 _Chemical Society, Chemical Communications_ (4), 168. 916 https://doi.org/10.1039/c39830000168 917 Martin-Jézéquel, V., Hildebrand, M., & Brzezinski, M. A. (2000). Silicon metabolism in 918 diatoms: Implications for growth. _Journal of Phycology_, _36_ (5), 821–840. 919 https://doi.org/10.1046/j.1529-8817.2000.00019.x 920 McCutchin, C. A., Edgar, K. J., Chen, C.-L., & Dove, P. M. (2025). Silica 921 biomacromolecule interactions: Toward a mechanistic understanding of 922 silicification. _Biomacromolecules_, _26_ (1), 43–84. 923 https://doi.org/10.1021/acs.biomac.4c00674 924 McNair, H. M., Brzezinski, M. A., & Krause, J. W. (2015). Quantifying diatom silicification 925 with the fluorescent dye, PDMPO. _Limnology and Oceanography: Methods_, 926 _13_ (10), 587–599. https://doi.org/10.1002/lom3.10049 927 Medlin, L. K., Kooistra, W. H., Sims, P., & Wellbrock, U. (1997). Is the origin of the diatoms 928 related to the end-Permian mass extinction? _Nova Hedwigia_, _65_ (1-4), 1–11. 929 https://doi.org/10.1127/nova.hedwigia/65/1997/1 930 Reichelt, T., Bode, T., Jordan, P.-F., & Brunner, E. (2023). Towards the chemical analysis of 931 diatoms’ silicon storage pools: A differential centrifugation-based separation 932 approach. _Minerals_, _13_ (5), 653. https://doi.org/10.3390/min13050653 933 Poulsen, N., & Kröger, N. (2023). Thalassiosira pseudonana (Cyclotella nana)(Hustedt) 934 Hasle et Heimdal (Bacillariophyceae): A genetically tractable model organism for 935 studying diatom biology, including biological silica formation. _Journal of_ 936 _Phycology_, 59(5), 809-817. 937 Reimann, B. E., Lewin, J. C., & Volcani, B. E. (1966). Studies on the biochemistry and fine 938 structure of silica shell formation in diatoms. II. The structure of the cell wall of 939 Navicula pelliculosa (Bréb.) Hilse. _Journal of Phycology_, _2_ (2), 74–84. 940 https://doi.org/10.1111/j.1529-8817.1966.tb04597.x - 25 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 941 Reinke, O., Machill, S., & Brunner, E. (2025). Polyamines of unique structure are 942 integrated in Synura echinulata biosilica. _Analytical and Bioanalytical Chemistry_, 943 _417_ (16), 3675–3684. https://doi.org/10.1007/s00216-025-05891-3 944 Schmid, A.-M. M., & Schulz, D. (1979). Wall morphogenesis in diatoms: Deposition of 945 silica by cytoplasmic vesicles. _Protoplasma_, _100_ (3-4), 267–288. 946 https://doi.org/10.1007/BF01279316 947 Schwarz, R., & Menner, E. (1924). Zur Kenntnis der Kieselsäuren (I.). _Berichte Der_ 948 _Deutschen Chemischen Gesellschaft (A and B Series)_, _57_ (8), 1477–1481. 949 https://doi.org/10.1002/cber.19240570850 950 Shimizu, K., Del Amo, Y., Brzezinski, M. A., Stucky, G. D., & Morse, D. E. (2001). A novel 951 fluorescent silica tracer for biological silicification studies. _Chemistry & Biology_, 952 _8_ (11), 1051–1060. https://doi.org/10.1016/S1074-5521(01)00072-2 953 Shipe, R. F., & Brzezinski, M. A. (1999). A study of Si deposition synchrony in Rhizosolenia 954 (Bacillariophyceae) mats using novel 32 Si autoradiographic method. _Journal of_ 955 _Phycology_, _35_ (5), 995–1004. https://doi.org/10.1046/j.1529-8817.1999.3550995.x 956 Siever, R. (1991). Silica in the oceans; biological-geochemical interplay. 957 https://eurekamag.com/research/020/017/020017116.php?srsltid=AfmBOoqvAk 958 mzJhLFjO9NXPy7Z-UkykOBwnpp05lKL3NsOm6rMgoqRFR6 959 Sims, P. A., Mann, D. G., & Medlin, L. K. (2006). Evolution of the diatoms: insights from 960 fossil, biological and molecular data. _Phycologia_, _45_ (4), 361–402. 961 https://doi.org/10.2216/05-22.1 962 Spinde, K., Pachis, K., Antonakaki, I., Paasch, S., Brunner, E., & Demadis, K. D. (2011). 963 Influence of polyamines and related macromolecules on silicic acid 964 polycondensation: Relevance to “Soluble Silicon Pools”? _Chemistry of Materials_, 965 _23_ (21), 4676–4687. https://doi.org/10.1021/cm201988g 966 Stoermer, E. F., Pankratz, H. S., & Bowen, C. C. (1965). Fine structure of the diatom 967 Amphipleura pellucida. II. Cytoplasmic fine structure and frustule formation. 968 _American Journal of Botany_, _52_ (10), 1067–1078. https://doi.org/10.1002/j.1537 969 2197.1965.tb07286.x 970 Sullivan, C. W. (1979). Diatom mineralization of silicic acid. IV. Kinetics of soluble Si pool 971 formation in exponentially growing and synchronized Navicula pelliculosa. 972 _Journal of Phycology_, _15_ (2), 210–216. https://doi.org/10.1111/j.1529 973 8817.1979.tb02986.x 974 Sumper, M., Brunner, E., & Lehmann, G. (2005). Biomineralization in diatoms: 975 Characterization of novel polyamines associated with silica. _FEBS Letters_, 976 _579_ (17), 3765–3769. https://doi.org/10.1016/j.febslet.2005.06.001 977 Thamatrakoln, K., & Hildebrand, M. (2008). Silicon uptake in diatoms revisited: A model 978 for saturable and nonsaturable uptake kinetics and the role of silicon 979 transporters. _Plant Physiology_, _146_ (3), 1397–1407. 980 https://doi.org/10.1104/pp.107.107094 981 Treguer, P., & Pondaven, P. (2000). Global change. Silica control of carbon dioxide. 982 _Nature_, _406_ (6794), 358–359. https://doi.org/10.1038/35019236 - 26 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 983 Tréguer, P., Nelson, D. M., van Bennekom, A. J., Demaster, D. J., Leynaert, A., & 984 Quéguiner, B. (1995). The silica balance in the world ocean: A reestimate. 985 _Science_, _268_ (5209), 375–379. https://doi.org/10.1126/science.268.5209.375 986 Tréguer, P. J., Sutton, J. N., Brzezinski, M., Charette, M. A., Devries, T., Dutkiewicz, S., 987 Ehlert, C., Hawkings, J., Leynaert, A., Liu, S. M., Llopis Monferrer, N., López 988 Acosta, M., Maldonado, M., Rahman, S., Ran, L., & Rouxel, O. (2021). Reviews 989 and syntheses: The biogeochemical cycle of silicon in the modern ocean. 990 _Biogeosciences_, _18_ (4), 1269–1289. https://doi.org/10.5194/bg-18-1269-2021 991 Wenzl, S., Hett, R., Richthammer, P., & Sumper, M. (2008). Silacidins: Highly acidic 992 phosphopeptides from diatom shells assist in silica precipitation in vitro. 993 _Angewandte Chemie (International Ed. In English)_, _47_ (9), 1729–1732. 994 https://doi.org/10.1002/anie.200704994 995 Werner, D., & Pirson, A. (1967). Über reversible Speicherung von Kieselsäure in 996 Cyclotella cryptica. _Archives of Microbiology_, _57_ (1), 43–50. 997 https://doi.org/10.1007/BF00405766 998 Wisser, D., Brückner, S. I., Wisser, F. M., Althoff-Ospelt, G., Getzschmann, J., Kaskel, S., 999 & Brunner, E. (2015). (1)H-(13)C-(29)Si triple resonance and REDOR solid-state 1000 NMR-A tool to study interactions between biosilica and organic molecules in 1001 diatom cell walls. _Solid State Nuclear Magnetic Resonance_, _66-67_, 33–39. 1002 https://doi.org/10.1016/j.ssnmr.2014.12.007 ## 1003 Figures and Tables 1004 **Figure 1** . Illustration of the separation process carried out as a parallel experiment to compare 1005 cells (top) with extracted cell walls (bottom) as control in order to identify the SSPs in the fractions. 1006 **Figure 2** . Top: Scheme illustrating the experiment for quantifying the SSPs based on fluorescence 1007 intensities of PDMPO in cellular biosilica after full labeling (left) and pulse labeling in silicon-free 1008 medium (right, ASW-Si: Si-free ASW). Bottom: Microscopic images show the fluorescence of 1009 PDMPO in the biosilica of methanol extracted cells of _C. nana_ after full labeling (a) and pulse 1010 labeling in silicon-free medium (b). 1011 **Figure 3** . Top, left: Photograph of the sucrose density gradient with cell lysate after centrifugation. 1012 Diagrams showing the tSi-values of the different fractions for the cells (a) and the extracted cell 1013 walls (b) as fractions of tSi cell . c) Difference between the curves for cells and extracted cell walls. 1014 This difference represents an estimate for the cytoplasmic silicon, which is assumed to be part of 1015 the SSPs. * Upper bound radius for pure silica particles (2.20 g/ml) calculated with the Stokes’ 1016 law. Larger silica particles – which may be part of the SSPs as well – will be found in the pellet (Fig. 1017 S6 and App. S1). Error bars: 95% CI. 1018 **Figure 4** . Top: Photographs of the density gradients after ultracentrifugation and diagrams 1019 showing the corresponding tSi in the lysed cell fractions from cultures in exponential growth (a) 1020 and the control experiment with extracted cell walls (b). tSi-values are fractions of tSi cell . 1021 c) Difference between the curves for lysed cells and extracted cell walls providing an estimate for 1022 cytoplasmic silicon which is assumed to be part of the SSPs (Fig. S7 and App. S1). Error bars: 95% 1023 CI. - 27 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 1024 **Figure 5** . Solid-state [29] Si MAS NMR spectrum of exponentially grown whole cells (b) with the 1025 characteristic Q [n] group chemical shift regions. For comparison, the liquid-state [29] Si NMR 1026 spectrum of a 4 mM silicic acid solution at pH 9 is inserted (a) showing the signals of the soluble 1027 Si species, monosilicic acid (Q [0] ) and disilicic acid (Q [1] ). These soluble species are not detected in 1028 whole cells (region between the dotted lines). c) The intensities of the detected Q [n] group signals 1029 (n = 2,3,4) depend on the degree of silicic acid polycondensation (Belton et al., 2012; McCutchin 1030 et al., 2025). 1031 **Figure 6** . Left: Silicon contents of fractions obtained after density gradient ultra-centrifugation 1032 (a,d). Right: Solid-state [29] Si MAS NMR spectra of whole cells (c,f) and combined fractions (b,e) 1033 between 1.21 and 1.27 g/ml (dark green bars). Top: Exponentially grown cells (a,b,c). Bottom: 1034 Silicon starved cells (d,e,f). Error bars: 95% CI. 1035 **Table 1** . _C. nana_ cell parameters: Average cell volume, total cellular silicon amount tSi cell and 1036 concentration [tSi cell ] determined using the Molybdenum blue assay. The silicon amount tSi SSP and 1037 concentration [tSi SSP ] of SSPs was estimated based on the described fluorescence method. The 1038 experimental error is given as 95% CI. 1039 **Table 2** . Overview of the methods applied in this investigation and in other studies with the 1040 quantitative estimate obtained for the SSPs of _C. nana_ in exponential growth. The experimental 1041 error is given as 95% CI. * Calculated according to Stokes’ law assuming pure silica particles 1042 (Fig. S6-S7 and App. S1). [†] Recalculated with the data from the present study (App. S1). [‡] Same 1043 cultivation batch. [a] (Reichelt et al., 2023) [ b ] (Martin-Jézéquel et al., 2000) [ c ] (Thamatrakoln & 1044 Hildebrand, 2008) [d ] (Kumar et al., 2020). ## 1045 Figures and tables in the supporting information (short caption) 1046 **Figure S1** . PDMPO labeling of _Cyclotella nana_ . 1047 **Figure S2** . PDMPO labeling of _Conticribra weissflogii_ . 1048 **Figure S3** . PDMPO labeling of _Cyclotella cryptica_ . 1049 **Figure S4** . PDMPO labeling of _Navicula pelliculosa_ . 1050 **Figure S5** . PDMPO labeling of _Biddulphia tridens_ . 1051 **Figure S6** . Particle migration in the sucrose gradient. 1052 **Figure S7** . Particle migration in the combined OptiPrep and Nycodenz gradient. 1053 **Table S1** . SSPs of various diatom species estimated by PDMPO labeling. 1054 **Appendix S1** . Calculation of the particle migration with Stokes’ law. ## 1055 Figures and tables in the supporting information (entire caption) 1056 **Figure S1** . PDMPO pulse labeling of _Cyclotella nana in_ silicon-free artificial seawater (ASW-Si). 1057 The identic microscopic images are showing a silicon-starved culture of _C. nana_ in the bright field 1058 channel (left) and the fluorescence channel (right). The fluorescence results from PDMPO 1059 incorporation and indicates growth of the cell wall on the expense of the silicon storage pools - 28 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 1060 (SSPs). The cells were pulse labeled with 0.125 µM of PDMPO for 48 h. Note that only one cell of 1061 the 151 cells in the image remained unlabeled and has not produced fresh biosilica. 1062 **Figure S2** . PDMPO labeling of _Conticribra weissflogii_ . _C. weissflogii_ was totally labeled with 1063 PDMPO at exponential growth (left) and pulse-labeled with PDMPO during silicon starvation in 1064 ASW-Si (right). 1065 **Figure S3** . PDMPO labeling of _Cyclotella cryptica_ . _C. cryptica_ was totally labeled with PDMPO at 1066 exponential growth (left) and pulse-labeled with PDMPO during silicon starvation in ASW-Si 1067 (right). 1068 **Figure S4** . PDMPO labeling of _Navicula pelliculosa_ . _N. pelliculosa_ was totally labeled with 1069 PDMPO at exponential growth (left) and pulse-labeled with PDMPO during silicon starvation in 1070 ASW-Si (right). 1071 **Figure S5** . PDMPO labeling of _Biddulphia tridens_ . _B. tridens_ was totally labeled with PDMPO at 1072 exponential growth (left) and pulse-labeled with PDMPO during silicon starvation in ASW-Si 1073 (right). 1074 **Figure S6** . Particle migration in the sucrose gradient. The theoretical particle migration over time 1075 is plotted for particles with a density of 1.25 g/ml (top left) and 2.20 g/ml (top right) in the 1076 continuous gradient at 4.200 x g. The location of a particle at the end of the 60 min centrifugation 1077 run is shown as a function of the radius (bottom). The different particle densities are related to the 1078 cytoplasm (color) or biosilica and silica particles (black). 1079 **Figure S7** . Particle migration in the combined OptiPrep and Nycodenz gradient. The theoretical 1080 particle migration over time is plotted for particles with a density of 1.25 g/ml (top left) and 1081 2.20 g/ml (top right) in the stepped gradient at 175,000 x g. The location of a particle at the end of 1082 the 60 min centrifugation run is shown as a function of the radius (bottom). The different particle 1083 densities are related to the cytoplasm (color) or biosilica and silica particles (black). 1084 **Table S1** . The SSPs of various diatom species were estimated by combined PDMPO pulse labeling 1085 and silicon starvation in ASW-Si. The values are given as a fraction of the average total silicon 1086 amount of a cell (tSi cell ). Constant light (48 h) was applied to exclude cell cycle effects. The 1087 experiment was reproduced for _C. nana_ with results of 35.1 ± 4.8% and 33.5 ± 3.3%. All other 1088 experiments were performed once. The experimental error is given as 95% confidence interval. 1089 **Appendix S1** . Calculation of the particle migration with Stokes’ law. 1090 - 29 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. # **Fluorescence microscopic evidence for intracellular silicon** **storage pools in diatoms and their partial separation [1]** **(supporting information)** **Tobias Reichelt**, Chair for Bioanalytical Chemistry, Faculty of Chemistry and Food Chemistry, TU Dresden, Bergstrasse 66, 01062 Dresden, Germany **Julia Breyer**, Chair for Bioanalytical Chemistry, Faculty of Chemistry and Food Chemistry, TU Dresden, Bergstrasse 66, 01062 Dresden, Germany **Eike Brunner** ², Chair for Bioanalytical Chemistry, Faculty of Chemistry and Food Chemistry, TU Dresden, Bergstrasse 66, 01062 Dresden, Germany, eike.brunner@tu-dresden.de **Running title** Separation of silicon pools **Acknowledgments** Financial support by the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG, Project No. Br 1278/35-1) is gratefully acknowledged. 1 November 3, 2026 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. **Figure S1** . PDMPO pulse labeling of _Cyclotella nana in_ silicon-free artificial seawater (ASW-Si). The identic microscopic images are showing a silicon-starved culture of _C. nana_ in the bright field channel (left) and the fluorescence channel (right). The fluorescence results from PDMPO incorporation and indicates growth of the cell wall on the expense of the silicon storage pools (SSPs). The cells were pulse labeled with 0.125 µM of PDMPO for 48 h. Note that only one cell of the 151 cells in the image remained unlabeled and has not produced fresh biosilica. **Table S1** . The SSPs of various diatom species were estimated by combined PDMPO pulse labeling and silicon starvation in ASW-Si. The values are given as a fraction of the average total silicon amount of a cell (tSi cell ). Constant light (48 h) was applied to exclude cell cycle effects. The experiment was reproduced for _C. nana_ with results of 35.1 ± 4.8% and 33.5 ± 3.3%. All other experiments were performed once. The experimental error is given as 95% confidence interval. |species(strain)|SSPs / % of tSi<br>cell| |---|---| |_Cyclotella nana_ (CCMP 1335)<br>_Conticribra weissflogii_ (CCAP 1085/18)<br>_Cyclotella cryptica_ (CCMP 332)<br>_Navicula pelliculosa_ (SAG 1050-3)<br>_Biddulphia tridens_ (unknown)|34.3± 5.8<br>32.5± 4.5<br>21.6± 2.6<br>5.5± 1.1<br>4.1± 1.8| [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. **Figure S2** . PDMPO labeling of _Conticribra weissflogii_ . _C. weissflogii_ was totally labeled with PDMPO at exponential growth (left) and pulse-labeled with PDMPO during silicon starvation in ASW-Si (right). **Figure S3** . PDMPO labeling of _Cyclotella cryptica_ . _C. cryptica_ was totally labeled with PDMPO at exponential growth (left) and pulse-labeled with PDMPO during silicon starvation in ASW-Si (right). **Figure S4** . PDMPO labeling of _Navicula pelliculosa_ . _N. pelliculosa_ was totally labeled with PDMPO at exponential growth (left) and pulse-labeled with PDMPO during silicon starvation in ASW-Si (right). [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. **Figure S5** . PDMPO labeling of _Biddulphia tridens_ . _B. tridens_ was totally labeled with PDMPO at exponential growth (left) and pulse-labeled with PDMPO during silicon starvation in ASW-Si (right). **Figure S6** . Particle migration in the sucrose gradient. The theoretical particle migration over time is plotted for particles with a density of 1.25 g/ml (top left) and 2.20 g/ml (top right) in the continuous gradient at 4.200 x g. The location of a particle at the end of the 60 min centrifugation run is shown as a function of the radius (bottom). The different particle densities are related to the cytoplasm (color) or biosilica and silica particles (black). [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. **Figure S7** . Particle migration in the combined OptiPrep and Nycodenz gradient. The theoretical particle migration over time is plotted for particles with a density of 1.25 g/ml (top left) and 2.20 g/ml (top right) in the stepped gradient at 175,000 x g. The location of a particle at the end of the 60 min centrifugation run is shown as a function of the radius (bottom). The different particle densities are related to the cytoplasm (color) or biosilica and silica particles (black). [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. **Appendix S1** . Calculation of the particle migration with Stokes’ law. **𝑅** **𝑝** **𝑝** **𝑝** **𝑝** **𝑓** **𝑓** f - The movement of a particle is calculated with the law of Stokes. The speed of a particle _ν_ _p_ is obtained with the following expression. **𝑅** **𝑝** **𝑝** **𝑝** **𝑝** **𝑓** **𝑓** f = **𝑅** **𝑝** **𝑝** **𝑝** **𝑝** **𝑓** **𝑓** 𝑝𝑝 𝑣𝑣 f 𝑅 � 𝑝 𝑝 𝑝 𝑝 𝑓 𝑓 � 2 𝑟𝑟 [2] **𝑅** 𝑅𝑅 𝜌𝜌 **𝑝** **𝑝** **𝑝** **𝑝** − 𝜌𝜌 **𝑓** **𝑓** 𝑓𝑓 f 𝑝 𝑝 𝑝 𝑓 𝑓 � (1) **𝑅** **𝑝** **𝑝** **𝑝** **𝑝** **𝑓** **𝑓** 𝑓𝑓 9 𝜂𝜂 f **𝑅** **𝑝** **𝑝** **𝑝** **𝑝** **𝑓** **𝑓** Density gradient centrifugation: A continuous gradient of sucrose was used. Therefore, each movement of a particle along the gradient involves a slight change of three parameters in equation (1) which are the density of the medium _ρ_ _Fluid_, the viscosity _η_ and the rotational centrifugal force _RCF_ . Since this correlation is a differential function, the calculation is done with incremental time steps of 1 s. After each step, the three parameters are recalculated with the new position in the gradient. The parameters are expressed as functions of the distance to the rotor centre. The temperature dependent viscosity of the sucrose solution is calculated with the empirical equations presented by Chenlo et al. (Chenlo et al., 2002). The density increase in the gradient is linear due to the use of a gradient mixer. That fact was confirmed by measurements of the refractive index. The upper bound for the particle radius of 300 nm given in Fig. S6 (bottom) is the threshold over which a silica particle with a density of 2.20 g/ml is expected to reach the pellet. Isopycnic density gradient ultracentrifugation: A stepped gradient combining OptiPrep and Nycodenz was used. Therefore, _ρ_ _Fluid_ and _η_ change stepwise in equation (1). The densities were obtained from the data sheets of the manufacturer (Serumwerk, Bernburg, Germany). The viscosities of Nycodenz and OptiPrep at 4°C were estimated using the Andrade equation. Therefore, reported viscosities of OptiPrep and Nycodenz at 20°C and 37°C were used (Eivindvik & Sjogren, 1995). Diferential centrifugation:f The threshold of the particle radius was calculated with the parameters applied in the cited study (Reichelt et al., 2023), which were a _RCF_ of 500 x g, a runtime of 5 min, a path of 20 mm, a density of the medium of 1.01 g/ml and a density of the particles of 2.20 g/ml. The threshold value represents the radius above which a particle is certain to reach the bottom of the tube during centrifugation. Accordingly, particles with radii below this threshold value may remain in the supernatant. It should be noted that at the beginning of the centrifugation the particles are randomly distributed in the solution resulting in random path lengths to the bottom of the tube. **𝑅** **𝑝** **𝑝** **𝑝** **𝑝** **𝑓** **𝑓** f - The SSPs of _C. nana_ separated by differential centrifugation were estimated to represent 9% of tSi cell (Reichelt et al., 2023). Using the values determined in the present study for tSi cell = 138 fmol and V cell = 88 fL, a recalculated intracellular concentration for the SSPs of 141 mM is obtained. That differs from the value reported in the cited study due to improved techniques. [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.04.764211; this version posted November 5, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.04.764211) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. ## References Chenlo, F., Moreira, R., Pereira, G., & Ampudia, A. (2002). Viscosities of aqueous solutions of sucrose and sodium chloride of interest in osmotic dehydration processes. _Journal of_ _Food Engineering_, _54_ (4), 347–352. https://doi.org/10.1016/s0260-8774(01)00221-7 Eivindvik, K., & Sjogren, C. E. (1995). Physicochemical properties of iodixanol. _Acta Radiologica._ _Supplementum_, _399_ (399_suppl), 32–38. https://doi.org/10.1177/0284185195036s39904 Reichelt, T., Bode, T., Jordan, P. F., & Brunner, E. (2023). Towards the Chemical Analysis of Diatoms’ Silicon Storage Pools: A Differential Centrifugation-Based Separation Approach. _Minerals_, _13_ (5), 653. https://doi.org/10.3390/min13050653