Trace‐Element Incorporation into Intracellular Pools Uncovers Calcium‐Pathways in a Coccolithophore

Abstract Many organisms form minerals from precursor phases that crystallize under strict biological control. The dynamic intracellular processes of formation, transport, and deposition of these precursor phases are challenging to identify. An unusual situation is recently revealed for the calcifying alga Emiliania huxleyi, as the cells contain a compartment filled with a concentrated Ca and P phase but the final calcite crystals, which are nucleated in a different compartment, are P‐free. Thus, the connection of the Ca–P‐rich pool to the mineralization process remains unclear. Here, pulse‐chase experiments are used with Sr to label the Ca–P‐rich phase in E. huxleyi cells, and cryo X‐ray absorption spectroscopy and analytical transmission electron microscopy to follow the Sr within cells. It is found that Sr is first found in the Ca–P‐rich phase and then becomes incorporated into the calcite. This demonstrates that the calcium used by the cells to build calcite originates from the Ca–P‐rich pool.

does not produce coccoliths, were grown in artificial seawater medium Aquil at 18 °C and a 12/12 h light/dark cycle, as described before. [13] The concentration of nitrate in the medium was 0.2 mM and of phosphate 10 µM. The standard medium contained 10 mM CaCl 2 and 60 M SrCl 2 . Cells devoid of extracellular coccoliths were obtained by adding 1 / 50 volume of 0.5 M EDTA pH = 8.0 to the culture. This treatment dissolved extracellular calcite, but the calcite of intracellular coccoliths remained undissolved. Cultures of cells growing at low-Ca medium (0.1 mM CaCl 2 ) were grown for more than 6 months, these cultures continued to divide whereas coccolith formation was ceased.

Sample preparation for cryoXANES
For pulse-chase experiments 50 ml of cell culture were placed at 14 °C during the dark period prior to the experiment. The entire pulse-chase experiment was conducted at 14 °C since the low temperature slowed down the calcification process while keeping the same percentage of fully calcifying cells after 24 h. At the beginning of the pulse the cells were exposed to light and the appropriate volumes of 1 M CaCl 2 and 100 mM SrCl 2 were added to the culture. With the exposure to light the cells resumed coccolith formation, which at the end of the dark phase is ceased. After 50 minutes, the cells were pelleted at 5000 g, 14 °C, for ten minutes.
49.8 ml of the medium were decanted, and 49.8 ml Sr-free medium was added to the remaining 0.2 ml of culture containing the cells. The cells were re-suspended and left at continuous light conditions. In order to freeze a sample 10 ml cell culture were removed and pelleted at 5000 g, 4 °C, for ten minutes. 9.5 ml of the supernatant were decanted. The 0.5 ml medium containing the cells was filtered using an Isopore TM membrane filters, 100 nm pore size (Merck Millipore, Germany). The membrane filter with the cells was blotted dry and immediately frozen in liquid nitrogen.

cryoXANES measurements
The frozen filters were attached to a frozen copper sample holder using two metals rings screwed to the holder. The XANES spectra of the Sr L-edge and P K-edge were acquired at the LUCIA beamline at the SOLEIL Synchrotron Light Source (Gif-sur-Yvette, France) by scanning the X-ray beam energy from 1900-2200 eV in 1.0 eV steps before the Sr L 3 -edge and between the Sr L 3 -edge and L 2 -edge. The Sr absorption edges were scanned with 0.2 eV steps, and the entire P K-edge was scanned with 2 eV steps. All samples were measured under vacuum and at cryo-conditions. The synchrotron ring energy was 2.75 GeV and the current was up to 400 mA. The X-rays energy was selected by a double crystal Si (111) monochromator, which was calibrated using Ti foil and setting the edge at 4966 eV. The beam size on the sample was approximately 0.5 × 2 mm 2 . In order to improve the signal-tonoise ratios, 3 scans of the full Sr L 3,2 -edge spectra were averaged. 10 scans were averaged for the Sr L 3 -edge spectra.

Data analysis and fitting
For each Baseline subtraction, normalization, and data processing and analysis were performed using the Athena software package4.0. Normalization was performed using a linear pre-edge function between 25 eV and 5 eV below E o (set to 1940 eV) and a linear postedge function between 10 eV and 60 eV above E 0 . Linear combination fitting was performed on spectra using two standards, Sr-doped ACP and calcite. were placed on a copper grid with a holey carbon film. High-angle annular dark-field electron imaging (HAADF) and EDX spectroscopy were performed on a Tecnai G2 F20 X-Twin microscope equipped with a field-emission electron source (operated at 200 kV), a Fishione HAADF detector at 330 mm camera length for image acquisition in the STEM mode and an EDAX Genesis X-ray analyzer. Figure S1. Sr L 3,2 -edge and P K-edge XANES spectra of the same samples as in Figure 1c.

Supplementary Figures
The spectra were normalized to the phosphorus edge white-line as an approximation for cellular material in the sample. The two pulse samples, at the beginning of the pulse (0 h) and just before washing (1 h), show that Sr is not incorporated into the cells during this period. Figure S2. Light microscopy images of E. huxleyi cells during a pulse-chase experiment.
The cells were growing in low-Ca medium before the pulse. The indicated time is after the start of the one hour pulse. Each sample was imaged in bright field and with crossed polarizers to detect the birefringence of the calcite crystals. Red arrows indicate cells with detectable birefringence. Figure S3. Sr L 3,2 -edge and P K-edge XANES spectra of the same samples as in Figure   1d. The spectra (except for the 4 h) were normalized to the phosphorus edge 'white-line' as an approximation for cellular material in the sample. The undulating background, which is of the same magnitude as the Sr signal, is due to the uneven sample geometry. Figure S4. The different components of the linear combination fitting analysis for the spectra shown in Figure 1e. The raw data of the Sr L 3 -edge spectra are shown, as well as the resulting fits and the relative components of the Sr-ACP fraction and the Sr-calcite fraction. Two orange lines indicate the data range that was considered for achieving the best fit. See also Table S1.  For each sample the scaling factor ('scaleby') used for the fitting is showed. For each of the two reference spectra the weight given in the fit and calculated error is shown, as well as the difference applied to the E0 value and its error.