Abstract Clear glass has been crucial to the development of modern architecture, with windows and clear glazing being a major catalyst for indoor living. However this dependence on clear glass has also created a dependence on new material, from a specific and limited sand for glass making, as well as high energy use to fire and float form the sand into glass panes. This project aims at exploring a biological alternative to glass panes, as well as developing the scientific results that could point to future work that uses this biosilica for novel materials, both aggregated with construction waste and as a pure material out of diatom blooms. Diatoms are a type of algae that have silica cell walls called frustules, and these frustules form into intricate lacy, opalescant patterns as the colonies of algae grow. Cylindrotheca fusiformis is a marine diatom species that relies on proteins including silaffins for silicic acid polymerization. By modifying the proteins that are responsible for the diatom structure, this project opens up the mechanical properties of diatoms as a material, where structure is responsible for color expression and for potential material attachment and other characteristics for future projects.
Clear glass has been crucial to the development of modern architecture, with windows and clear glazing being a major catalyst for indoor living. However this dependence on clear glass has also created a dependence on new material, from a specific and limited sand for glass making, as well as high energy use to fire and float form the sand into glass panes. This project aims at exploring a biological alternative to glass panes, as well as developing the scientific results that could point to future work that uses this biosilica for novel materials, both aggregated with construction waste and as a pure material out of diatom blooms. Diatoms are a type of algae that have silica cell walls called frustules, and these frustules form into intricate lacy, opalescant patterns as the colonies of algae grow. Cylindrotheca fusiformis is a marine diatom species that relies on proteins including silaffins for silicic acid polymerization. By modifying the proteins that are responsible for the diatom structure, this project opens up the mechanical properties of diatoms as a material, where structure is responsible for color expression and for potential material attachment and other characteristics for future projects.
Project Aims
Protein Modification Modify the Sil1p protein of Cylindrotheca fusiformis through cell-free systems. In this aim, I am interested in observing a change of the cell-free system after Sil1p modification from the original sequence. The modification would aim to either overexpress the polymerizing protein, or supress it and see the potential shifts that emerge from those DNA modifications. These shifts would be tested using florescent tagging.
Sil1p c.fusiformis sequence and protein structure without modification:
reverse translation of sp|Q9SE35|SIL1_CYLFU Silaffin-1 OS=Cylindrotheca fusiformis OX=2853 GN=SIL1 PE=1 SV=1 to a 795 base sequence of most likely codons.
Algae Modification Modify a non-diatom, and non-frustule forming algae to have the Sil1p protein in place of their traditional cell wall proteins. This would be an exciting yet achievable way to observe the protein in an algae structure, without worrying about breaking the diatom cell wall.
OLD AIM 2: Use CRISPR/Cas-9 or the gene gun (most likely CRISPR) to input the modified protein into the T.pseudonona THIS CHANGED AFTER FURTHER DIATOM MODIFICATION RESEARCH.
Color Shift Modify the TpSil1/2 protein of T.pseudonona to refract light with a visibly different hue through structural modification. The overexpression or knockout of this gene can result in more or less silica deposition, resulting in an altered macropore structure, and thus modifying the light scattering by the physical structure of the frustules.
Rubble Attachment Further modifying the diatom structure to act as a bandaid between two pieces of glass, or two pieces of cement rubble. Can diatoms from silica patterns that attach onto surrounding objects/surfaces?
Background
Diatoms are a unique class of algae that produce silica cell walls called frustules. These frustules have shown up in fossil records, indicating a continued presence of diatoms in our environment over millenia.
Experimental Design, Techniques, Tools, and Technology
image of the Sil1p protein structure without any modifications
This model on Alphafold demonstrates that this silaffin protein has a much greater structural confidence than the t.pseudonona silaffin protein did. this demonstrates that this diatom is probably more studied than t.pseudonona.
Modify the Sil1p protein R5 peptide
R5 Peptide portion: SSKKSGSYSGSKGSKRRIL (from Claude and double checked through Googling the peptide)
β’ Poulsen, N., Chesley, P.M. & KrΓΆger, N. (2006). Molecular Genetic Manipulation of the Diatom Thalassiosira pseudonana. Journal of Phycology, 42, 1059β1065.
β’ Tesson B, Lerch SJL, Hildebrand M. Characterization of a New Protein Family Associated With the Silica Deposition Vesicle Membrane Enables Genetic Manipulation of Diatom Silica. Sci Rep. 2017 Oct 18;7(1):13457. doi: 10.1038/s41598-017-13613-8. PMID: 29044150; PMCID: PMC5647440.
Alternative project approaches
Another potential project was to create a touch-reactive biofilm that pigmented with pressure. This could use a system along the lines of the cell-free systems that were discussed in the Week 8 lecture. However for this project, I wanted the color to fade away once the pressure was lessened, however that may not be possible with cell-free systems as they seem to be a one-time reaction.
Alternatively to diatoms, it could also be interesting to use mycelium and genetically modify it to be clear. This uses the strength and potential for mycelium as a building material, binding to architectural waste like glass pieces to create a composite material with the tensile strength and growth rate of mycelium. A clear mycelium would entail a melanin gene knockout, and potentially doing so with a more clear mycelium strain to begin with. This method was developed with the help of CRISPR.
Claude partnered protocol suggestion:
Biosilica by Design: Engineering Sil1p Repeat Domains for Tunable Silica Morphology and Structural Color Toward Sustainable Architectural Glass
SECTION 1: ABSTRACT
Clear glass has been crucial to the development of modern architecture, with windows and clear glazing being a major catalyst for indoor living. However, this dependence on clear glass has also created a dependence on new material β from a specific and limited sand for glassmaking, as well as high energy use to fire and float-form the sand into glass panes. This project aims at exploring a biological alternative to silica panes, as well as developing the scientific results that could point to future work that uses this biosilica for a novel, rubble-and-biosilica composite material. Diatoms are a type of algae that have silica cell walls called frustules, and these frustules form into intricate, lacy, opalescent patterns as the colonies of algae grow. Cylindrotheca fusiformis is a model diatom species whose silaffin protein Sil1p, containing tandem R5 repeat domains, drives silicic acid polymerization into structured biosilica. By systematically modifying the R5 repeat architecture of Sil1p β truncating, duplicating, and rearranging these domains β this project investigates how repeat domain structure controls silica precipitation efficiency and nanoscale morphology.
The central hypothesis is that the number and arrangement of R5 repeat domains in Sil1p directly determines the rate, quantity, and structural organization of silica precipitation, with downstream implications for the optical properties of the resulting biosilica material. Three engineered constructs (wild-type Sil1p, Sil1p-ΞR5, and Sil1p-2xR5) will be expressed in a cell-free BL21 DE3 lysate system at Ginkgo Bioworks and assayed using the silicomolybdate colorimetric assay and a full-spectrum absorbance scan on a Spark plate reader. Aim 1 establishes the cell-free expression and functional comparison platform. Aim 2 extends the work into a native diatom expression system to observe changes in frustule morphology directly. Aim 3 envisions the engineering of precise structural color in biosilica β programmable biological stained glass β as a sustainable replacement for energy-intensive architectural glazing.
SECTION 2: PROJECT AIMS
Aim 1 β Experimental Aim
The first aim of my final project is to express duplicated and truncated Sil1p variants in a cell-free BL21 DE3 expression system by utilizing whole plasmid synthesis from Twist Bioscience, cell-free protein synthesis at Ginkgo Bioworks, silicomolybdate colorimetric assay, and full-spectrum absorbance scanning on the Spark plate reader to determine how R5 repeat domain copy number affects silica precipitation efficiency and optical properties of the resulting biosilica.
Aim 2 β Medium-Term Aim
Building on the cell-free expression results of Aim 1, the second aim is to introduce the highest-performing Sil1p variant into a native or surrogate diatom expression system to observe changes in frustule morphology under physiologically relevant conditions. This aim will involve confocal fluorescence microscopy and scanning electron microscopy (SEM) to characterize the three-dimensional nanoarchitecture of frustules produced by engineered versus wild-type diatoms. By correlating R5 repeat domain structure with observable frustule geometry, Aim 2 will establish a quantitative structure-morphology relationship that directly informs rational design of biosilica optical properties. Collaboration with Ginkgo Bioworks on diatom strain engineering and Basecamp Research for mining novel silaffin sequence diversity from environmental diatom metagenomes will extend the design space available for Aim 3.
Aim 3 β Visionary Aim
By precisely programming the nanoarchitecture of diatom frustules through silaffin repeat domain engineering, Aim 3 envisions the creation of living stained glass β biosilica panels with genetically encoded, tunable structural color, produced sustainably through biological self-assembly rather than energy-intensive sand smelting and float glass manufacturing. Just as the Gothic cathedral captured light through colored glass to transform interior space, engineered diatom biosilica would capture and diffract light through photonic crystal nanostructures encoded in protein sequence β creating architectural glazing materials that are simultaneously functional, beautiful, carbon-neutral, and endlessly customizable. In partnership with BioFabricate and Mycoworks, this vision points toward a future where buildings grow their own windows.
SECTION 3: BACKGROUND
Literature Context
KrΓΆger et al. (1999) first isolated and characterized the silaffin proteins from Cylindrotheca fusiformis, identifying Sil1p and its tandem R5 repeat domains as the primary drivers of rapid, in vitro silica precipitation from silicic acid precursors. The R5 peptide (SSKKSGSYSGSKGSKRRIL) was shown to be both necessary and sufficient for silica polymerization, establishing the minimal functional unit of biosilicification. Sumper and Brunner (2008) subsequently demonstrated that the number and spacing of polyamine-modified lysine residues within repeat domains directly controls the geometry of silica nanostructure formation, with longer repeat arrays producing larger, more ordered silica particles. Together, these studies establish a clear mechanistic link between silaffin primary sequence and silica nanoarchitecture β yet the full design space of R5 repeat domain engineering for tunable material properties remains largely unexplored. This project addresses that gap by systematically varying R5 copy number and testing the functional consequences using a scalable, automated cell-free platform.
SECTION 4: EXPERIMENTAL DESIGN
Detailed Workflow
Step 1 β Codon Optimization and Construct Design (Week 1)
Design three expression constructs encoding: (1) wild-type Sil1p (UniProt O74824), (2) Sil1p-ΞR5 with all five R5 repeat units removed, and (3) Sil1p-2xR5 with the R5 repeat array duplicated from 5 to 10 copies. All constructs include an N-terminal T7 promoter, Shine-Dalgarno sequence, His6-tag, and T7 terminator. Codon-optimize sequences for E. coli BL21 DE3 expression using Twist’s integrated codon optimization tool.
Machine: None (computational design)
Expected result: Three validated construct sequences ready for synthesis ordering
Timeline: Days 1β3
Step 2 β Twist Bioscience DNA Order (Week 1)
Submit all three constructs as whole plasmid synthesis orders to Twist Bioscience using the pTwist-T7 backbone. Select clonal gene synthesis for each construct.
Machine: None (online order)
Expected result: Sequence-verified plasmid DNA delivered within 7β10 business days
Timeline: Days 3β4
Step 3 β Plasmid Receipt and Quality Check (Week 2)
Upon receipt, resuspend lyophilized plasmid DNA per Twist instructions. Verify by Sanger sequencing and gel electrophoresis on a 1% agarose gel.
Machine: ATC Thermal Cycler (for Sanger PCR), gel electrophoresis system
Plate: 96-Armadillo-PCR-AB2396X for PCR setup
Expected result: All three plasmids confirmed sequence-correct
Timeline: Days 10β12
Step 4 β Cell-Free Reaction Setup (Week 2β3)
Using the Echo525 acoustic liquid handler, transfer plasmid DNA (250 ng each) and Ginkgo BL21 DE3 cell-free master mix into a 384-well Echo PP plate in triplicate for each construct plus a no-template negative control. Total reaction volume: 5 Β΅L per well.
Machine: Echo525
Plate: 384-well Echo PP
Expected result: Consistent, low-volume transfers with <5% CV across replicates
Timeline: Day 14
Step 5 β Plate Sealing and Cell-Free Expression (Week 3)
Seal the 384-well plate with the Plateloc using A4s breathable seal to allow gas exchange during expression. Incubate in the Inheco Plate Incubator at 37Β°C for 4 hours.
Expected result: Protein expression occurs in all construct wells; negative control shows no expression
Timeline: Day 14, hours 0β4
Step 6 β TMOS Silica Precipitation Reaction (Week 3)
Transfer 2 Β΅L of each cell-free reaction to a fresh 384 Greiner black-well clear-bottom plate using the Bravo-384 plate stamp. Using the Multiflo dispenser, add 3 Β΅L of freshly hydrolyzed tetramethyl orthosilicate (TMOS, 1M in 1mM HCl) to each well to initiate silica precipitation.
Machine: Bravo-384, Multiflo
Plate: 384 Greiner black-well clear-bottom
Expected result: Visible silica precipitation in WT and 2xR5 wells within 5β10 minutes; reduced precipitation expected in ΞR5 wells
Timeline: Day 14, hour 5
Step 7 β Mixing and Incubation (Week 3)
Shake the precipitation plate on the BioshakeD3000 at 1,200 rpm for 20 minutes at room temperature to ensure complete silica polymerization.
Machine: BioshakeD3000
Plate: 384 Greiner black-well clear-bottom
Expected result: Complete silica polymerization; visible white precipitate in active wells
Timeline: Day 14, hours 5β5.5
Step 8 β Centrifugation to Pellet Silica (Week 3)
Centrifuge the precipitation plate at 3,000 Γ g for 10 minutes in the HiG Centrifuge to pellet silica particles.
Machine: HiG Centrifuge
Plate: 384 Greiner black-well clear-bottom
Expected result: Silica pellet visible at well bottom; clear supernatant containing residual free silicic acid
Timeline: Day 14, hour 6
Step 9 β Supernatant Transfer for Colorimetric Assay (Week 3)
Use the Bravo-384 to transfer 4 Β΅L of supernatant from each well into a fresh 384-flat-corning-3640 plate for silicomolybdate assay. Retain the pellet plate for spectrum scanning in Step 11.
Machine: Bravo-384
Plate: 384-flat-corning-3640
Expected result: Clean supernatant transfer without disturbing silica pellets
Timeline: Day 14, hour 6.5
Step 10 β Silicomolybdate Colorimetric Assay (Week 3)
Using the Tempest bulk dispenser, add 1 Β΅L of silicomolybdate reagent (ammonium molybdate in sulfuric acid) to each supernatant well. Incubate 10 minutes at room temperature. Read absorbance at 810 nm on the Spark Plate Reader. Lower absorbance in the supernatant = more silica precipitated by the protein.
Machine: Tempest, Spark Plate Reader
Plate: 384-flat-corning-3640
Expected result: WT and 2xR5 wells show lower A810 than ΞR5 and no-template control, indicating greater silica precipitation
Timeline: Day 14, hours 7β8
Step 11 β Full Spectrum Absorbance Scan of Silica Pellets (Week 3)
Resuspend silica pellets in 5 Β΅L ultrapure water by pipetting. Run a full-spectrum absorbance scan from 400β800 nm on the Spark Plate Reader to capture any optical differences between silica nanoparticles produced by different Sil1p variants.
Machine: Spark Plate Reader
Plate: 384 Greiner black-well clear-bottom
Expected result: Spectral differences between WT, ΞR5, and 2xR5 silica particles; potential blue shift in 2xR5 particles if larger particle size shifts photonic scattering
Timeline: Day 14, hour 8.5
Step 12 β Expression Confirmation: SDS-PAGE (Validation A, Week 3)
Collect 5 Β΅L of cell-free reaction from each construct well before TMOS addition. Load onto SDS-PAGE gel alongside a His-tag protein ladder. Run at 200V for 35 minutes. Stain with Coomassie blue and image.
Step 13 β qPCR Expression Verification (Week 3)
As an orthogonal expression check, extract total RNA from cell-free reactions and run qPCR using primers flanking the His6-tag sequence to confirm transcript levels across all three constructs are comparable.
Machine: CFX Opus qPCR machine
Plate: 96-Armadillo-PCR-AB2396X
Expected result: Similar Ct values across all three constructs, confirming equivalent transcription; any differences in protein yield are post-transcriptional
Timeline: Day 15
Step 14 β Data Analysis and Construct Ranking (Week 4)
Compile colorimetric assay data, full-spectrum scans, SDS-PAGE results, and qPCR data. Calculate silica precipitation efficiency for each construct as: % silica precipitated = (A810 no-template β A810 construct) / A810 no-template Γ 100. Rank constructs by precipitation efficiency and spectral shift magnitude.
Machine: None (computational analysis)
Expected result: Clear ranking of Sil1p-2xR5 > WT Sil1p > Sil1p-ΞR5 in silica precipitation efficiency
Timeline: Days 16β18
Step 15 β Iteration and Construct Refinement (Week 4β5)
Based on results, design a second round of constructs if needed β e.g., Sil1p-3xR5, Sil1p with randomized repeat spacing, or Sil1p with non-native repeat sequences. Order from Twist Bioscience and re-enter the cell-free pipeline. This iteration loop establishes the quantitative R5 dose-response relationship that forms the foundation for Aim 2 diatom expression work.
Cell-free protein synthesis (CFPS) is an in vitro method for producing proteins directly from DNA templates without the use of living cells. The system consists of a cell lysate β in this project, BL21 DE3 lysate prepared at Ginkgo Bioworks β combined with a master mix containing ribosomes, amino acids, energy regeneration components, and RNA polymerase. When a plasmid encoding a T7 promoter-driven gene is added, T7 RNA polymerase transcribes the gene into mRNA, which is then translated by ribosomes present in the lysate into protein. CFPS is particularly powerful for this project because it allows multiple silaffin variants to be expressed and assayed in parallel in a 384-well format within a single day, without the need for bacterial transformation, overnight culture, or IPTG induction β dramatically accelerating the design-build-test cycle for protein engineering.
2. Silicomolybdate Colorimetric Assay
The silicomolybdate assay, also known as the molybdenum blue assay, is a well-established colorimetric method for quantifying free silicic acid (Si(OH)β) in solution. In the presence of ammonium molybdate under acidic conditions, free silicic acid forms a yellow silicomolybdate complex; upon reduction with ascorbic acid or other reducing agents, this complex turns an intense blue color with peak absorbance at 810 nm. In this project, the assay is applied to the supernatant after silica precipitation β the more silica the Sil1p variant has precipitated from solution, the less free silicic acid remains, and therefore the lower the A810 reading. This indirect measurement elegantly reports on the silica-precipitating activity of each Sil1p variant in a format fully compatible with automated 384-well plate reading on the Spark platform, enabling quantitative comparison of precipitation efficiency across all three constructs simultaneously.
SECTION 6: PROJECT VALIDATION
10a β Validation Choice
Two complementary validation experiments are planned for this project, to be performed based on available lab access and timeline. Validation A (SDS-PAGE) directly confirms that all three Sil1p variants are being produced as proteins of the expected size in the cell-free system, ruling out expression failure as a confound. Validation B (silicomolybdate pilot assay) directly confirms that the wild-type Sil1p is functionally active in precipitating silica from TMOS, and that the ΞR5 truncation reduces this activity β establishing the functional assay and the expected directionality of results before the full 384-well campaign is run.
10b β Validation Protocols
Validation A: SDS-PAGE Expression Confirmation
Collect 5 Β΅L of cell-free reaction from each of the three Sil1p constructs and the no-template control after 4 hours of expression.
Add 5 Β΅L of 2Γ Laemmli SDS sample buffer to each sample.
Heat samples at 95Β°C for 5 minutes using the ATC Thermal Cycler.
Load 8 Β΅L of each sample onto a 4β20% gradient SDS-PAGE gel alongside a His-tag protein molecular weight ladder.
Run electrophoresis at 200V for 35 minutes in Tris-glycine SDS running buffer.
Stain gel with InstantBlue Coomassie stain for 30 minutes.
Confirm absence of bands in no-template control lane.
Validation B: Silicomolybdate Pilot Assay
Set up cell-free reactions for WT Sil1p, Sil1p-ΞR5, and no-template control in triplicate in a 96-well plate (25 Β΅L reactions).
Incubate at 37Β°C for 4 hours in the Inheco Plate Incubator.
Add 5 Β΅L of freshly hydrolyzed TMOS (1M in 1mM HCl) to each well.
Shake on BioshakeD3000 at 1,200 rpm for 20 minutes at room temperature.
Centrifuge at 3,000 Γ g for 10 minutes in the HiG Centrifuge to pellet silica.
Transfer 20 Β΅L of supernatant to a fresh 96-well flat-bottom plate.
Add 5 Β΅L of silicomolybdate reagent (0.026M ammonium molybdate in 0.1M HβSOβ) to each well.
Incubate 10 minutes at room temperature.
Add 5 Β΅L of reducing solution (0.1M ascorbic acid) if molybdenum blue endpoint is desired.
Read absorbance at 810 nm on the Spark Plate Reader.
Calculate % silica precipitated relative to no-template control.
Confirm WT Sil1p shows significantly lower A810 than ΞR5 and no-template control (expected: >40% reduction).
10c β Techniques Used
SDS-PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis) separates proteins by molecular weight under denaturing conditions, allowing direct visualization of whether each Sil1p variant has been produced at the correct size and in sufficient quantity by the cell-free system. The silicomolybdate assay is a spectrophotometric technique that quantifies free silicic acid in solution through formation of a chromogenic molybdate complex, providing an indirect but highly sensitive measure of silica precipitation activity. Cell-free protein synthesis leverages the transcription and translation machinery of bacterial lysates to produce proteins from plasmid DNA templates in vitro, enabling rapid, parallel expression of multiple variants without live organism handling. Together, these three techniques β SDS-PAGE, colorimetric spectrophotometry, and cell-free expression β form an integrated validation pipeline that confirms both the production and the functional activity of each engineered Sil1p variant before proceeding to the full high-throughput 384-well campaign.
10d β Hypothetical Data
Hypothetical Silicomolybdate Assay Results
Construct
Mean A810 (supernatant)
SD
% Silica Precipitated
No-template control
0.920
0.012
0% (baseline)
Sil1p-ΞR5
0.810
0.034
12%
WT Sil1p
0.485
0.041
47%
Sil1p-2xR5
0.203
0.028
78%
Interpretation: Higher A810 in the supernatant = more free silicic acid remaining = less silica precipitated. The hypothetical data shows a clear dose-response: duplicating R5 repeats nearly doubles silica precipitation efficiency compared to wild-type, while removing R5 repeats reduces precipitation to near-baseline levels.
Interpretation: The 2xR5 construct produces silica particles with elevated absorbance particularly in the 400β500 nm (blue/violet) range, consistent with smaller, more uniform nanoparticles that scatter shorter wavelengths preferentially β a potential precursor to structural color effects that Aim 3 will engineer more precisely.
Troubleshooting
The most likely technical challenge is low or absent silaffin protein expression in the cell-free system, which could result from poor codon optimization, mRNA secondary structure inhibiting translation, or issues with the His6-tag affecting protein folding. If expression is not confirmed by SDS-PAGE, the first corrective step would be to re-optimize the 5’ UTR sequence using the Salis Lab RBS Calculator and reorder a revised construct from Twist Bioscience before repeating the cell-free expression. A second potential challenge is non-specific silica precipitation in the no-template control wells β TMOS is chemically reactive and will self-polymerize under certain pH conditions independently of silaffin; this can be mitigated by carefully controlling TMOS hydrolysis conditions (1mM HCl, room temperature, freshly prepared) and including a TMOS-only control well in every plate. A third limitation is that cell-free silica precipitation may not faithfully recapitulate the in vivo post-translational modifications of native Sil1p β including phosphorylation and polyamine modifications on lysine residues β which are known to enhance biosilicification activity; this is an inherent constraint of the cell-free platform that will need to be addressed in Aim 2 when the work moves into a native diatom expression system. Finally, the full-spectrum scan for structural color differences is exploratory and may not yield interpretable optical signals at the scale of cell-free silica nanoparticles β if this is the case, dynamic light scattering (DLS) would be pursued as an alternative to characterize particle size distributions.
KrΓΆger, N., Deutzmann, R., & Sumper, M. (1999). Polycationic peptides from diatom biosilica that direct silica nanosphere formation. Science, 286(5442), 1129β1132.
Sumper, M., & Brunner, E. (2008). Silica biomineralisation in diatoms: the model organism Thalassiosira pseudonana. ChemBioChem, 9(8), 1187β1194.
KrΓΆger, N., Lorenz, S., Brunner, E., & Sumper, M. (2002). Self-assembly of highly phosphorylated silaffins and their function in biosilica morphogenesis. Science, 298(5593), 584β586.
Poulsen, N., Sumper, M., & KrΓΆger, N. (2003). Biosilica formation in diatoms: characterization of native silaffin-2 and its role in silica morphogenesis. PNAS, 100(21), 12075β12080.
Lechner, C. C., & Becker, C. F. (2015). Silaffins in silica biomineralization and biomimetic silica precipitation. Marine Drugs, 13(8), 5297β5333.
