Transcript level coordination of carbon pathways during silicon starvation‐induced lipid accumulation in the diatom Thalassiosira pseudonana

Summary Diatoms are one of the most productive and successful photosynthetic taxa on Earth and possess attributes such as rapid growth rates and production of lipids, making them candidate sources of renewable fuels. Despite their significance, few details of the mechanisms used to regulate growth and carbon metabolism are currently known, hindering metabolic engineering approaches to enhance productivity. To characterize the transcript level component of metabolic regulation, genome‐wide changes in transcript abundance were documented in the model diatom Thalassiosira pseudonana on a time‐course of silicon starvation. Growth, cell cycle progression, chloroplast replication, fatty acid composition, pigmentation, and photosynthetic parameters were characterized alongside lipid accumulation. Extensive coordination of large suites of genes was observed, highlighting the existence of clusters of coregulated genes as a key feature of global gene regulation in T. pseudonana. The identity of key enzymes for carbon metabolic pathway inputs (photosynthesis) and outputs (growth and storage) reveals these clusters are organized to synchronize these processes. Coordinated transcript level responses to silicon starvation are probably driven by signals linked to cell cycle progression and shifts in photophysiology. A mechanistic understanding of how this is accomplished will aid efforts to engineer metabolism for development of algal‐derived biofuels.


Fig. S1
Comparison of microarray and RNA-Seq data.      Methods S1 Detailed methods.

Fig. S1
Comparison of microarray and RNA-Seq data. Data show frequency distribution of Pearson correlation coefficients calculated using core signal fluorescence intensities (from the microarray) and the DESeq2 normalized counts (from RNA-Seq) for 10,945 gene models.  Hennon et al. (2015). Genes shown in blue are a putative silicon sensing mechanism (Fig. S3). Genes shown in green are the cytosolic isoform of ACCase and a coexpressed chloroplast-localized keto-acyl-reductase of fatty acid biosynthesis.  protein ID (Table S2).    C18 and C20 to C24 fatty acids was performed relative to known concentrations and Flame Ionization Detection (FID) peak areas of C13 and C19 FAME internal standards, respectively. The response factor for each standard compound (in GLC 68A and 68D) of interest was applied to the respective sample peak areas to calculate the final concentration of each peak of interest.

Cell counts and cell cycle analysis
The fatty acid methylation reaction (i.e. transesterification) efficiency was determined from the C15 FA internal standard added to each lyophilized algal pellet. Each sample was analyzed in duplicate.

Pigment analysis
Culture samples (10 ml) were collected and immediately filtered through GF/F Whatman filters (25 mm) and stored at -80°C until their analysis by HPLC. Filters were extracted in 90% HPLCgrade acetone and ultrasonicated in an ice bath for 10 min, stored at -20°C for 24 h and prefiltered through a 0.45 μm Whatman nylon Puradisk filter before injection. Afterwards, pigments were separated using a reverse phase C-8 column following the method of Zapata et al. (2000). External pigment standards were used for system calibration (DHI, Denmark).
Samples were analyzed on a Waters 600 HPLC system, equipped with a Thermo Separation Products AS3000 sampler, a TSP Spectra 100 variable wavelength (fixed at 440 nm), a Waters 996 diode array (scanning 330-800 nm) and a Waters 470 scanning fluorescence detector. Data were collected and analyzed using the Waters Millennium 32 software package. Each sample was analyzed in duplicate.

RNA isolation
For experiments Si-#3, Si-#9, and Si-#10, 750 ml of culture was removed, treated with cycloheximide (final concentration, 20 μg ml -1 ) and harvested by filtration. RNA isolated from 6 time points (0, 4, 8, 12, 18, and 24 h). Cells were pelleted and stored at -80°C prior to total RNA isolation (Hildebrand & Dahlin, 2000). RNA from experiment Si-#3 was processed for hybridization to an Affymetrix GeneChip whole genome tiling array while RNA from experiments Si-#9 and Si-#10 were processed for Illumina-based RNA-Seq. were then scanned with the GeneChip® Scanner, to generate the probe cell intensity data files.

Microarray hybridization and processing
Initial data analysis was performed as described in Shrestha et al. (2012). Microarray data are presented in Table S2 Normalized and mapped probe intensities were then subjected to a two-pronged test for differential expression which included a false discovery rate (FDR) corrected student t-test statistic along with a log fold change cutoff of +/-1. Two tailed unpaired t-tests were performed between time-points using the standard function within Microsoft Excel. The FDR correction was performed via the method of Benjamini & Hochberg (1995) implemented by the R statistical software package. Fold changes were calculated as the difference of any time-point from the initial 0hr time-point. Genes were considered to be differentially expressed if they had an FDR corrected p-value < .05 and a fold change greater or less than ± 1.  Table S2.

RNA-Seq sequencing and processing
Accessibility Transcriptomic data are available at NCBI's Gene Expression Omnibus under GEO series accessions GSE75651 (microarray) and GSE75460 (RNA-Seq).

Photosynthesis vs irradiance
Photosynthetic characteristics were obtained from photosynthesis-irradiance (P-I) measurements using a modified 14 C-bicarbonate incorporation technique (Lewis & Smith, 1983;Arrigo et al., 1999). P-I incubations were carried out in a photosynthetron incubator at 24 irradiances ranging from 0 to 830 μmol photons m -2 s -1 . Illumination was provided by two 150W tungsten-halogen lamps and adjusted within each vial chamber by neutral density filters. Total photosynthetically available radiation (PAR, 400-700 nm) within each illumination chamber was measured using a Biospherical Instrument QSL 101 sensor. P-I experiments were carried out at 18°C.
For each P-I curve, a 60 ml sample of culture was spiked with 0.06 mCi NaH 14 CO 3 to obtain a final activity of 0.001 mCi mL -1 . The spiked sample was distributed in 2 ml aliquots to 7 ml glass scintillation vials. Twenty-four of these vials were placed in the individually illuminated chambers within the incubator and incubated for 1 h. Two vials (time-zero samples) were acidified immediately with 200 μl of 20% HCl and placed in a hood during the 24 h experiment.
Total activity in the samples was determined by adding 100 μL of sample at 0 h into two scintillation vials containing 100 μl of 1M NaOH. Five milliliters of liquid scintillation cocktail (ECOLUME TM ) was then added.
After incubation, the samples were acidified with 200 μL of 20% HCl and placed in a hood for 24 h to drive off unincorporated inorganic radioisotope. After 24 h ventilation, all samples received 5 ml of ECOLUME TM liquid scintillation cocktail and were vigorously shaken.
Carbon uptake, normalized by Chl a, was calculated from radioisotope incorporation.
The photosynthetic response was modeled by curve fitting as suggested by Platt et al. (1975): Where P B is Photosynthesis per unit biomass (Chl a) in units mgC mgChla -1 h -1 , P max B is maximum rate of photosynthesis per unit of biomass, α is the initial slope in units of [mgC mgChla -1 h -1 (μmol photons m -2 s -1 ) -1 ] and E is irradiance in units of μmol photons m -2 s -1 .

Fast repetition rate fluorescence
The photochemical quantum yield of photosystem II (F v /F m ) and the functional absorption cross section (σ PSII ) of photosystem II (PSII) was measured using a Chelsea Fastracka Fast Repetition Rate Fluorometer, programmed to deliver single turnover saturation of PSII from 100 flashlets of 1.1 μs at 1 μs intervals. Samples were dark acclimated for >30 min, after which single turnover fluorescence induction curves were measured over a range of background light levels.
Photosynthetic parameters (F v /F m and σ PSII ) were estimated by fitting standard models to data to determine values of F o (initial fluorescence), F m (maximal fluorescence), F v /F m ( = (F m -F o )/F m ), and σ PSII (Kolber et al., 1998).