Harnessing microalgae for the biosynthesis of molecular crystals

Cultivation of A.
carterae

A. carterae from the National Center for Marine Algae and Microbiota (CCMP1314) were cultivated using f/2 medium. The cultivation regime was 14 h light/10 h dark illumination. The irradiance was 40 μmol (photons) m⁻² s⁻¹ generated by a white-light emitting diode. The f/2 growth medium was prepared with the following dry salts per liter: 400 mM NaCl, 10 mM KCl, 9 mM CaCl₂·2H₂O, 20 mM MgCl₂·6H₂O, 20 mM MgSO₄·7H₂O and 1.85 mM KBr. The following nutrients were first dissolved separately and then added in final concentrations of 88.3 × 10⁻⁵ M NaNO₃, 3.62 × 10⁻⁵ M NaH₂PO₄·H₂O, 1.06 × 10⁻⁴ M Na₂SiO₃·9H₂O, 2 × 10⁻⁴ M H₃BO₃, 9.89 × 10⁻⁹ M Selene, 1.03 mM Tris (hydroxymethyl). The following trace metals (made separately): 1.18 × 10⁻⁵ M FeCl₃·6H₂O, 1.51 × 10⁻⁵ M Na₂EDTA·2H₂O, 3.93 × 10⁻⁸ M CuSO₄·5H₂O, 2.60 × 10⁻⁸ M Na₂MoO₄·2H₂O, 7.65 × 10⁻⁸ M ZnSO₄·7H₂O, 4.20 × 10⁻⁸ M CoCl₂·6H₂O and 9.10 × 10⁻⁷ M MnCl₂·4H₂O. The following vitamins were also added at final concentrations of 2.96 × 10⁻⁷ M thiamine HCl (vitamin B₁), 2.56 × 10⁻⁹ M biotin (vitamin H) and 4.61 × 10⁻¹⁰ M cyanocobalamin (vitamin B₁₂). Finally, 20 ml of NaHCO₃ 2 mM was filtered (0.22 μm) directly into the prepared medium. Suspension culture flasks (250 ml) were used with a filter screw cap (CELLSTAR).

Intracellular uptake of different N-heterocycles in A.
carterae

For uptake experiments (n = 4 for each compound), the nitrate in the growth medium was replaced with one of the N-heterocyclic compounds (purines, pyrimidines or pteridines) at final concentrations of 40 μM for guanine, hypoxanthine, xanthine, uric acid, guanosine and melamine; 6 μM for pterin, isoxanthopterin, xanthopterin and leucopterin; 0.1 mM for theobromine and 1–20 mM for cytosine and 5-fluorocytosine. To remove any undissolved particles from the growth media, the solutions were placed in an ultrasonic bath and heated to 40–50 °C for 20–30 min. The solutions were then allowed to cool to room temperature for a few hours and passed through a 0.22-μm filter. The concentration of cells was kept constant at 2 × 10⁵ ml⁻¹ and the experiments were carried out in 50-ml suspension culture flasks with a filter screw cap (CELLSTAR) filled with 20 ml of the prepared medium. A UV–vis spectrophotometer (Evolution 220, Thermo Fisher Scientific) was used to measure absorbance of the growth medium with the different nitrogenous compounds in the 230–350 nm range to monitor uptake. Hundreds of cells were screened in every experiment and, in all reported cases, over 90% of the cells grew crystals (Supplementary Fig. 30) as described in the main text. Accumulation of birefringent deposits in live A. carterae cells was monitored using a polarized light microscope (Zeiss AX10) equipped with a Zeiss Axiocam 705 color camera using transmission and polarization modes with ×5, ×10, ×20 and ×50 air objectives and a ×100 oil immersion objective.

The mass of the accumulated compound per cell can be calculated as the mass of the dissolved compound divided by the number of cells. For example, for 20 ml of growth medium with a concentration of 40 μM guanine, the mass of guanine is m_guanine = 1.21 × 10⁻⁴ g. This value divided by the number of cells in the flask, 2 × 10⁵ cells, equates to ~30 pg of guanine per cell. Mojzes et al. showed that A. carterae cells could accumulate as much as 150 pg of guanine. The accumulated mass depends on the concentration of guanine or other N-heterocycle and number of cells per milliliter¹⁰.

Monitoring culture growth

To quantify the cell density of the cultures, the number of A. carterae cells was correlated to the absorbance of the chloroplasts. A UV–vis spectrophotometer (Evolution 220, Thermo Fisher Scientific) was used to measure absorbance of the culture at the 600–700 nm range. The number of cells was counted using a Brand Bürker Counting Chamber (Fisher Scientific) with the Zeiss AX10 microscope equipped with a Zeiss Axiocam 705 color camera and ×10 objective. We found the intensity of the absorbance at 673 nm was correlated with the number of cells according to the following equation: log(cells ml⁻¹) = (log(A_673 nm) + 6.05) / 0.92, similar to the correlation found by Hotos et al.⁵³.

Extraction of accumulated N-heterocyclic crystals for imaging and structural analysis

Once a culture had sufficient time to accumulate crystalline deposits, the microalgae were collected by centrifugation. After centrifugation at 10,000g for 5 min at 4 °C, the supernatant was discarded, and the cell pellet was washed with double distilled water (DDW) to remove salts originating from the growth medium and begin cell lysis. The number and volume of DDW washes (generally two to three) may change according to the solubility of the crystalline material and the observed impurities present in the final samples (as viewed by transmission electron microscopy (TEM)). To prevent enzymatic activity that may damage the crystalline inclusions, 1 µl 100× protease inhibitor was added to 5 ml of washed cells. Cell lysis was completed by probe sonication on ice (40% amplitude, 10 s on/12 s off cycle for 2 min ‘on’ time). The lysate was purified using a differential centrifugation at 10,000g for 3 min at 4 °C. The pellet was suspended in ca. 150–200 µl DDW and incubated for 3 h with 200 µl with α-amylase from Bacillus sp., at 75 mg ml⁻¹ (Sigma Aldrich, Cat. no. 9000-90-2) at room temperature to digest the starch granules produced by the cells. Following another centrifugation and resuspension, the sample was loaded on a sucrose gradient. The sucrose gradient was composed of three 300-µl density layers: 60, 40, 20 w/w% sucrose solutions. Gradient centrifugation was carried out at 2,000g for 30 min at 4 °C and the resulting pellet was mostly the pure crystalline deposits. Improved separation of the crystals from the starch granules was achieved using a second sucrose gradient, composed of three 300-µl density layers: 65, 60, 30 w/w% sucrose solutions. Gradient centrifugation was carried out at 800g for 40 min at 4 °C and resulted in distinct separation of starch granules (60% layer) from the crystals (bottom of 65 w/w% layer). The resulting pellet was mostly the pure crystalline deposits. Finally, one to two washes with ca. 150–200 µL DDW were performed with centrifugation at 20,000g for 3 min at 4 °C between every wash. This pelleted the extracted crystalline deposits. In general, wherever possible the protocol was performed on ice with chilled DDW, whereas the enzyme incubation was conducted at room temperature.

A Tecnai T12 G² TWIN TEM operating at 120 kV was used for the imaging of the extracted deposits. Images and ED patterns were recorded using a Gatan 794 MultiScan CCD camera. ED was analyzed using Gatan Digital Micrograph software (GMS v.1.4.5) within the DIFPack module. Images and ED patterns were recorded with consideration of potential beam damage to the sample, thus appropriate illumination conditions (spot size) were used to avoid it.

Control of guanine and xanthine spherulite sizes using cerulenin

A stock solution of cerulenin (Sigma Aldrich, Cat. no. 17397-89-6) in dimethylsulfoxide at a concentration of 40 mg ml⁻¹ was prepared and stored at −20 °C. In 50-ml suspension culture flasks, stationary stage cells at a concentration of 2 × 10⁵ ml⁻¹ were incubated for 1 h with 60 μM of cerulenin in a f/2 medium without a nitrogen source. The cells were then gently pelleted by centrifugation at 4,000g for 2 min at room temperature and the f/2 medium was replaced with one containing guanine or xanthine as a nitrogen source at concentrations between 5 and 40 μM and 60 μM of cerulenin. The cells were then allowed to accumulate crystals for times between 15 min and 240 min (Supplementary Fig. 24). For optical microscopy, the accumulation was stopped by a second medium replacement to a medium not containing the nitrogen source. Accumulated spherulites were then extracted according to the protocol above.

Cryo-SEM—sample preparation and imaging

A concentrated pellet of A. carterae cells (centrifuged 800g, 3 min) was sandwiched between two aluminum disks (25 μm thick) and cryo-immobilized in a high-pressure freezing device (EM ICE, Leica). The frozen samples were then mounted on a holder under liquid nitrogen in a specialized loading station (EM VCM, Leica) and transferred under cryogenic conditions (EM VCT500, Leica) to a sample preparation freeze fracture device (EM ACE900, Leica). There, samples are held under vacuum at a temperature of −120 °C. Samples are fractured by hitting the top disc carrier with a tungsten knife at a speed of 150 mm s⁻¹, exposing a clean fracture plane that can be imaged. The samples were then etched for 6 min at −110 °C to uncover additional structural details of the sample by controlled evaporation of water. Finally, the samples were coated with 3 nm of PtC. Samples were imaged using a HRSEM Gemini 300 SEM (Zeiss with SmartSEM software) by secondary electron in-lens detector while maintaining an operating temperature of −120 °C.

Raman microscopy—measurements and analysis

Depending on the cell density, the cells contained in 200–600 µl of culture medium were collected by a gentle centrifugation (3,000g, 30 s). The volume of the supernatant was reduced to approximately 15 µl and the pellet was resuspended by gentle shaking. A 5 μl aliquot of the concentrated cell suspension was mixed with 5 μl of 1% low-melting agarose dissolved in artificial seawater, homogenized gently and spread onto a quartz microscope slide, covered with a quartz coverslip (diameter: 20 mm; thickness: 0.18 µm) and sealed with CoverGrip (Biotium). Immobilization of cells before Raman mapping using low-melting agarose was necessary due to the ability of A. carterae to actively move using their flagella. Raman mapping of the cells was performed using a confocal Raman microscope (WITec alpha300 RSA, Oxford Instruments–WITec) equipped with a ×60 water-immersion objective UPlanSApo, numerical aperture 1.2 (Olympus). For Raman excitation, 532 nm and 830 nm lasers with powers of 20 mW and 40 mW at the focal plane, respectively, were used. Raman mapping was performed with a scanning step of 200 nm in both directions and integration time of 100 ms per voxel.

Before measurements with 532 nm excitation, the autofluorescence of the photosynthetic apparatus was removed by a low-power wide-area photobleaching applying the procedure described previously^10,54. Excitation at 532 nm allowed high-quality Raman spectra not only of crystalline inclusions, but also of other biomolecules present in cells, for example, lipid droplets, starch grains, mitochondria and cell walls. However, this excitation was not suitable for monitoring pteridines exhibiting strong and persistent autofluorescence. In addition, we observed that crystals of more soluble substances (for example, cytosine or uric acid) often dissolved gradually when excited at 532 nm, probably due to slight local heating by the excitation beam. To avoid problems with any autofluorescence and thermally induced crystal dissolution during longer mapping, we used 830 nm excitation. This excitation allowed us to obtain consistent Raman spectra of all studied compounds within the cells without observable cell photobleaching. In cases where it was possible to use both 532 and 830 nm excitations, the results led to the same conclusions. On average, 10–15 cells from each cell culture were measured in at least three independent replicates to ensure that conclusions regarding the occurrence of crystalline structures under given conditions were sufficiently statistically significant. Reference spectra of all pure compounds were obtained from commercial powders deposited directly on a quartz microscope slide or from microcrystals formed by evaporation from their solutions in deionized water. The reference spectra were obtained with the same excitation and detection conditions used for measuring the cells. In such a case, dry objective ×50 EC Epiplan-Neofluar, numerical aperture 0.55 (Zeiss) was used.

Data were analyzed using WITec Project SIX Plus v.6.2 software (Oxford Instruments–WITec) by implementing the following steps: advanced cosmic ray removal, background subtraction, cropping of the spectral edges affected by detector margins and spectral demixing with the True Component Analysis tool (Oxford Instruments–WITec). Raman chemical maps were constructed from the spatial distribution of the demixed spectral components and typical spectra of the compounds within cells were compared with the references.

Structure determination of biogenic leucopterin and xanthine crystals with 3D ED and PXRD

For extracted biogenic leucopterin crystals, 3D ED data were recorded at room temperature on an ELDICO ED-1 electron diffractometer. The device was equipped with a LaB₆ source operating at an acceleration voltage of 160 kV (λ = 0.02851 Å). For the data collection and initial data assessment, the software Eldix was used. Crystals were mapped and centered in scanning TEM imaging mode using a single diode brightfield detector and 3D ED data were collected in diffraction mode (parallel beam, beam size of approximately 800 nm) with a hybrid-pixel detector (Dectris QUADRO). 3D ED data collection was performed under continuous rotation with an angular step of 0.5° per frame and an exposure time of 0.5 s per frame. The electron dose was 0.01 el Å⁻² s⁻¹ and each measurement lasted 120 s per crystal for a rotation range from −60° to +60°.

For PXRD, extracted samples of biogenic leucopterin crystals and biogenic xanthine crystals were compacted into 0.3-mm diameter borosilicate glass capillaries. PXRD data were recorded on the Materials Science/PXRD beamline at the SESAME synchrotron using radiation with wavelength 0.82491 Å. For each measurement, the sample was placed at a distance of 740.4 mm from a Pilatus 300 K area detector of 172 μm pixel size. The area detector covered 6.4° and was used to collect PXRD data from 2θ = 4.3° to 28° for xanthine and from 2θ = 7.3° to 31° for leucopterin (four frames) with an exposure time collected per detector frame of 1,800 s to 2,000 s, with 7 s between frames. The PXRD data were recorded in transmission mode at room temperature. A NIST (640 f) Si standard was used to calibrate the instrument, and the lattice parameter of Si was used to determine the exact wavelength during the measurements. The detector was set to collect a diffraction image every 6°, and the collected images were then processed to extract the merged diffraction pattern through an Image J scripting mode. Crystal structure determination of biogenic leucopterin was based on a multi-technique strategy combining analysis of 3D ED and powder XRD data, augmented by periodic DFT-D calculations. Accurate unit cell parameters were determined by profile fitting of powder XRD data using the Le Bail technique in the program GSAS. Structure solution from 3D-ED data was carried out using the direct-space genetic-algorithm technique in the program EAGER, followed by structure refinement from 3D-ED data using SHELX⁵⁵. Structure validation included periodic DFT-D geometry optimization of the refined structure using CASTEP⁵⁶, which led to only minor atomic displacements. Structural analysis of biogenic xanthine focused on powder XRD data recorded at ambient temperature, which was observed to match the powder XRD pattern corresponding to the known crystal structure of synthetic xanthine³⁵. Accurate unit cell parameters for the biogenic sample determined from powder XRD data by profile fitting (using the Le Bail technique) are in close quantitative agreement with those reported (at ambient temperature) for synthetic xanthine³⁵, supporting the conclusion that the biogenic and synthetic materials represent the same crystalline phase of xanthine with the same crystal structure. Additional details of the analysis are discussed in Supplementary Text.

Single-crystal XRD of synthetic and biogenic cytosine crystals

Biogenic cytosine crystals

A single-crystal XRD experiment was conducted on a sample of A. carterae cells 24 h after transfer to a growth medium containing cytosine as the sole nitrogen source. The algae were concentrated by centrifugation (3,500 rpm, 3 min) in an Eppendorf tube, and a small droplet of the concentrated sample was mounted onto a CrystalCap ALS HT cryo-loop for data collection at 100 K. Data were collected using a Rigaku XtaLAB Synergy-S single-crystal X-ray diffractometer equipped with a HyPix-Arc 100° detector and a standard Cu Kα X-ray radiation source (λ = 1.5148 Å). Data analysis was carried out using the CrysAlis PRO software package (v.1.171.43.134a, released 2024). This setup enabled the identification of diffraction spots from internal crystalline structures formed within the algae. It is worth noting that extensive efforts were made to characterize the internal crystal formed within the algae using various complementary techniques, but only the single-crystal XRD method at 100 K yielded reliable results.

Synthetic cytosine crystals

Single crystals of cytosine monohydrate were grown from a saturated solution of cytosine in DDW left uncovered overnight. A colorless crystal with the shape of a rectangular plate (0.531 × 0.226 × 0.084 mm³) of C₄H₇N₃O₂ was mounted on a CrystalCap ALS HT cryo-loop mount for data collection at 293 K on a Rigaku XtalLAB Synergy-S single-crystal X-ray diffractometer, which includes an HyPix-Arc 100° detector and a standard detector and a standard Mo Kα x-ray radiation source (λ = 0.71073 Å). Unit cell dimensions, space group assignment, data reduction and finalization were done using the CrysAlis PRO software package (v.1.171.43.134a, released 2024). A total of 6,366 reflections were collected, of which 1,931 were used after merging by SHELXL⁵⁵ according to the crystal class and based on Friedel pair equivalency for structure solution. Gaussian correction was done using a multifaceted crystal model and empirical absorption correction was done using spherical harmonics (SCALE3 ABSPACK—an Oxford Diffraction program v.1.0.4; Oxford Diffraction Ltd., 2005). The structure was solved in the monoclinic P2₁/c space group (no. 14) by SHELXT using intrinsic phasing and refined by SHELXL using a full-matrix least-squares technique.

Calculation of the refractive index for the crystal structure of xanthine

DFT calculations were performed using v.24 of the CASTEP software⁵⁶. The XRD crystal structure was used as the starting point for a geometry optimization, allowing all unit cell and internal coordinates to vary. The PBE exchange-correlation functional⁵⁷ with Tkatchenko−Scheffler dispersion corrections⁵⁸ was employed together with Vanderbilt Ultrasoft pseudopotentials⁵⁹ in CASTEP’s on-the-fly generated implementation. Calculations used a 700 eV plane wave cutoff energy and the Brillouin Zone was sampled at a density of 0.05 Å⁻¹. The dielectric function and hence the RI was obtained in a single-particle formulation. The electronic states were obtained from DFT using the parameters described above. The OptaDOS code⁴⁷ was used to calculate the dielectric function. The calculations were carefully checked to ensure converge with Brillouin zone sampling and the number of unoccupied states. The dielectric function was obtained in the direction perpendicular to the molecular plane and as an average within the molecular plane.

MD and FDTD

MD simulations were carried out using HOOMD-blue (v.2.9.4)⁶⁰ with the Langevin integrator (kT = 0.05) and polydisperse 12–0 softcore pair-potential using the PolydisperseMD v.2.9 plugin⁶¹. Particle sizes d_i were sampled randomly from a log-normal distribution, with mean

and

where µ₁ and s₁ are the mean particle diameter and polydispersity. The d_i values outside range (100 nm, 1,300 nm) were rounded to the nearest boundary to avoid the chance of extreme particle sizes. Simulations were initiated with random placement of particles in a 20 × 20 × 20 μm³ box with periodic boundary conditions, and relaxed with Gaussian softcore repulsive potential, then squeezed to a final 5 × 5 × 5 μm³ volume during 7.5 × 10⁵ time steps under the 12–0 pair-potential. Particle coordinates and diameters were then imported to Lumerical FDTD (Ansys), v.2024-R1.3 and the optical properties were set to either constant n or spatially varying spherically symmetric RI with diagonal permittivity tensor

for isotropic and birefringent particles, respectively, where n, n_t and n_r are the averaged, tangential and radial RI. In the FDTD simulation setup, periodic boundary conditions in the x- and y-directions, a plane wave source and a reflectance monitor above particles in the z-direction were used to obtain reflectance values comparable to an integrating sphere measurement (Supplementary Fig. 31). A filling fraction of 65% was chosen to represent packed extracted particles in air. The recorded reflectance was integrated in Lumerical over the 5 × 5 μm² monitor area and normalized with respect to the source power. The slow optical axis (tangential direction) was oriented parallel to the particle surface, consistent with the radial π-stacking determined by ED.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.