C3 and C4 pathways of photosynthetic carbon assimilation in marine diatoms are under genetic, not environmental, control.

Marine diatoms are responsible for up to 20% of global CO2 fixation. Their photosynthetic efficiency is enhanced by concentrating CO2 around Rubisco, diminishing photorespiration, but the mechanism is yet to be resolved. Diatoms have been regarded as C3 photosynthesizers, but recent metabolic labeling and genome sequencing data suggest that they perform C4 photosynthesis. We studied the pathways of photosynthetic carbon assimilation in two diatoms by short-term metabolic 14C labeling. In Thalassiosira weissflogii, both C3 (glycerate-P and triose-P) and C4 (mainly malate) compounds were major initial (2–5 s) products, whereas Thalassiosira pseudonana produced mainly C3 and C6 (hexose-P) compounds. The data provide evidence of C3-C4 intermediate photosynthesis in T. weissflogii, but exclusively C3 photosynthesis in T. pseudonana. The labeling patterns were the same for cells grown at near-ambient (380 μL L−1) and low (100 μL L−1) CO2 concentrations. The lack of environmental modulation of carbon assimilatory pathways was supported in T. pseudonana by measurements of gene transcript and protein abundances of C4-metabolic enzymes (phosphoenolpyruvate carboxylase and phosphoenolpyruvate carboxykinase) and Rubisco. This study suggests that the photosynthetic pathways of diatoms are diverse, and may involve combined CO2-concentrating mechanisms. Furthermore, it emphasizes the requirement for metabolic and functional genetic and enzymic analyses before accepting the presence of C4-metabolic enzymes as evidence for C4 photosynthesis.

Marine planktonic diatoms are responsible for up to 20% of primary production on earth, fixing more than 10 billion tons of inorganic carbon each year (Falkowski and Raven, 2007). Diatoms achieve this, despite CO 2limiting conditions in the oceans, by using CO 2concentrating mechanisms (CCMs) actively to increase the steady-state CO 2 concentration around Rubisco, the principal photosynthetic carboxylase (Giordano et al., 2005;Roberts et al., 2007). By increasing the ratio of CO 2 to O 2 , this diminishes the wasteful process of photorespiration. Despite their great ecological impact, photosynthetic carbon acquisition by diatoms is still poorly understood.
It has generally been held that diatoms have biophysical CCMs, based on transport of inorganic carbon across cellular membranes (Giordano et al., 2005;Roberts et al., 2007). However, evidence has recently emerged of C 4 photosynthesis, a biochemical CCM, in the marine diatom Thalassiosira weissflogii (Reinfelder et al., 2000(Reinfelder et al., , 2004Morel et al., 2002;compare with Johnston et al., 2001;Granum et al., 2005), emphasizing the need for additional photosynthetic labeling experiments. The case for C 4 photosynthesis has been further strengthened by the occurrence of relevant genes in recently sequenced marine phytoplankton genomes, including the diatoms Thalassiosira pseudonana (Armbrust et al., 2004) and Phaeodactylum tricornutum (Montsant et al., 2005) and the green alga Ostreococcus tauri (Derelle et al., 2006). Of particular relevance to this study is the finding that T. pseudonana possesses the enzymic apparatus to operate C 4 photosynthesis of the kind suggested for T. weissflogii (Reinfelder et al., 2000(Reinfelder et al., , 2004Morel et al., 2002), including phosphoenolpyruvate carboxylase (PEPC) and phosphoenolpyruvate carboxykinase (PEPCK). The hypothetical mechanism of unicellular C 4 photosynthesis is a compartmentalized carboxylation-decarboxylation cycle analogous to terrestrial C 4 plants, albeit utilizing different intracellular compartments rather than different specialized cells (Edwards et al., 2004). In the proposed model, PEPC functions as primary carboxylase in the cytoplasm, forming oxaloacetate (C4) from phosphoenolpyruvate (C3) and HCO 3 2 . C4 acids are then transported into the chloroplast (possibly the pyrenoid) and decarboxylated by PEPCK, releasing CO 2 that is refixed by Rubisco. To complete the cycle, C3 acids are transported back to the cytoplasm. Important components of most biophysical CCMs are carbonic anhydrases (CAs), which catalyze the reversible hydration of CO 2 and usually depend on the trace metal zinc for activity (Giordano et al., 2005). A potential advantage of a biochemical versus a biophysical CCM in T. weissflogii is that of economizing on zinc (Reinfelder et al., 2000), which may be limiting or co-limiting in parts of the ocean (Crawford et al., 2003;Franck et al., 2003). However, the presence of cadmiumspecific CAs in T. weissflogii and other diatoms (Lane et al., 2005;Park et al., 2007) further complicates consideration of how zinc influences their CCM.
The work described here includes short-term photosynthetic labeling studies on the marine diatoms T. weissflogii and T. pseudonana, and measurements in the latter of relevant gene transcripts and proteins (specific primers and antisera were only obtained for T. pseudonana). This study compared cells grown at near-ambient (380 mL L 21 ; similar to the present atmospheric level) and low (100 mL L 21 ; lower than last glacial maximum) air-equilibrium CO 2 concentrations, since earlier work suggested that C 4 photosynthesis is induced by low CO 2 in T. weissflogii (Reinfelder et al., 2000(Reinfelder et al., , 2004Morel et al., 2002). It is known that growth in media equilibrated with CO 2 at below the present atmospheric level increases the photosynthetic affinity for inorganic carbon in, for example, T. pseudonana (Fielding et al., 1998) and the freshwater green alga Chlamydomonas reinhardtii (Vance and Spalding, 2005).

RESULTS
Short-term photosynthetic 14 C labeling was studied in T. weissflogii and T. pseudonana grown at near-ambient (380 mL L 21 ) or low (100 mL L 21 ) CO 2 concentration. In T. weissflogii, both C3 (glycerate-P and triose-P) and C4 (mainly malate) compounds were major initial products, with respectively approximately 45% and 30% of label after 2 s (Fig. 1). The fraction of these early products then decreased rapidly, while that of C6 (hexose-P) compounds increased reciprocally (from 15% to 60% within 30 s). The sigmoid shape of the C6 labeling curve is consistent with transient C4 labeling. Growth of the diatom at different CO 2 concentrations (380 or 100 mL L 21 ) had no significant effect on the distribution of short-term labeled products (Fig. 2).
The results indicate that a combination of glycerate 3-P (formed by Rubisco) and malate (derived from oxaloacetate formed by PEPC) are formed as primary products and sugar-P as secondary products by C 3 -C 4 intermediate photosynthesis in T. weissflogii.
In T. pseudonana, the fraction of 14 C-labeled C3 compounds was similar to that in T. weissflogii (40% after 2 s, and then rapidly decreasing), but there was very little label in malate (C4) and no label detected in Asp (C4) at any time studied (Fig. 1). Most of the remaining label was incorporated into C6 compounds, which increased from 40% (after 2 s) to 65% within 30 s. The hyperbolic shape of the C6 labeling curve is consistent with the lack of C4 labeling. As with T. weissflogii, there was no significant effect of the growth CO 2 concentration (380 or 100 mL L 21 ) on the short-term labeling pattern in T. pseudonana (Fig. 2). The results indicate that glycerate 3-P is formed (by Rubisco) as primary product and sugar-P as secondary products by exclusive C 3 photosynthesis in T. pseudonana.

Carbon Fixation in Diatoms
Plant Physiol. Vol. 145,2007 Growth CO 2 concentration (380 or 100 mL L 21 ) had small or negligible effects on transcripts of the C 4 -metabolic genes PEPC1, PEPC2, and PEPCK, or Rubisco large subunit (RBCL), in T. pseudonana (Fig. 3). Transcription of RBCL, but neither PEPCs nor PEPCK, was strongly enhanced (approximately 30-fold) at the start of the light period compared to the start of the dark period. In accordance with the gene transcripts, CO 2 concentration had no significant effect on protein abundances of PEPC1, PEPC2, PEPCK, and RBCL ( Fig. 4), nor did the transcripts or proteins respond to transient changes in CO 2 concentration (data not shown). In contrast, transcription of the PCOC gene for the P-subunit of Gly decarboxylase (GDCP) was highly up-regulated (3-to 6-fold) by low CO 2 concentration (Fig. 3). The results indicate that the putative C 4 -photosynthetic carboxylases and decarboxylase in T. pseudonana are not influenced by inorganic carbon.

DISCUSSION
In T. weissflogii, the results of metabolic 14 C labeling are consistent with a combination of glycerate 3-P (formed by Rubisco) and malate (derived from oxaloacetate formed by PEPC) as primary products of photosynthesis and sugar-P as secondary products (Fig. 1). This labeling pattern resembles that of a C 3 -C 4 intermediate flowering plant such as Flaveria linearis (Monson et al., 1986), rather than a pure C 4 plant, in which malate is the dominant initial product and C3 compounds appear as intermediates (Hatch and Slack, 1966). Interestingly, there was little label in Asp (C4), which is a significant or dominant initial product in C 4 flowering plants with PEPCK as their decarboxylase (Wingler et al., 1999). Although C 4 -metabolic gene transcripts or proteins in T. weissflogii were not analyzed in this study (due to the lack of specific primers and antisera), significant PEPC and PEPCK activities have previously been measured in this diatom (Reinfelder et al., 2000).
While the labeling pattern in T. weissflogii could be explained by C 3 -C 4 intermediate photosynthesis, another explanation to consider is nonphotosynthetic C 4 metabolism. An alternative function of PEPC in diatoms is anaplerotic production of C4 skeletons for nitrogen assimilation; the rate of light-independent carbon fixation in Skeletonema costatum, a close relative of Thalassiosira spp. (Kaczmarska et al., 2006), agrees well with the computed anaplerotic requirement (Granum and Myklestad, 1999). Differences in the diel pattern of NO 3 2 assimilation between T. pseudonana and T. weissflogii (Needoba and Harrison, 2004) suggest that the anaplerotic rate is higher in the latter during the photoperiod, resulting in higher fixation of inorganic carbon into C4 compounds. However, the extensive labeling of malate compared to anaplerotic requirements and negligible labeling of other organic and amino acids suggest that the C4 fixation is mainly photosynthetic. The nonphotosynthetic roles of PEPCK are yet to be fully resolved in plants and are made more enigmatic in diatoms by its apparent chloroplastic location (Cabello-Pasini et al., 2001;Granum et al., 2005), but it too may be involved in nitrogen metabolism (Delgado-Alvarado et al., 2007). A recent study of PEPC and PEPCK activities in P. tricornutum indicated that both enzymes are strictly anaplerotic (Cassar and Laws, 2007).
In C 4 photosynthesis with HCO 3 2 entering the cells, HCO 3 2 fixed by PEPC in the cytoplasm, and CO 2 released by PEPCK in the chloroplast, there is no overt role for CAs (Reinfelder et al., 2000(Reinfelder et al., , 2004Morel et al., 2002;Granum et al., 2005). However, the partial C 3 photosynthesis demonstrated in this study suggests the involvement of a parallel biophysical mechanism in T. weissflogii utilizing CAs. Facilitated or energized Figure 2. Distribution of 14 C label into C 3 pathway products (glycerate-P, triose-P, and hexose-P) and C 4 pathway products (malate and Asp) after 2 and 5 s in T. weissflogii and T. pseudonana grown at near-ambient (380 mL L 21 ) or low (100 mL L 21 ) CO 2 . Data are means 6 SE of three separate determinations. uptake of inorganic carbon would require both a CA converting HCO 3 2 to CO 2 in the pyrenoid and a CA converting CO 2 to HCO 3 2 in some more peripheral compartment (Giordano et al., 2005;Roberts et al., 2007). Such involvement of CAs in the biophysical component of a combined CCM means less economy in the use of zinc than is the case with exclusive C 4 photosynthesis (Reinfelder et al., 2000).
While T. pseudonana possesses an enzymic complement that could permit C 4 photosynthesis, the combined metabolic 14 C labeling, gene transcript, and protein measurements (Figs. 1-4) indicate solely C 3 photosynthetic biochemistry. Although posttranscriptional regulation of PEPC and PEPCK cannot be ruled out, regulatory phosphorylation domains characteristic for flowering plant enzymes are absent in the diatom enzymes (Granum et al., 2005). In neither diatom is the pathway(s) of inorganic carbon assimilation significantly altered by the CO 2 concentration used for growth, although there are clearly other acclimatory responses of both species (Fielding et al., 1998;Reinfelder et al., 2000Reinfelder et al., , 2004Burkhardt et al., 2001;Morel et al., 2002;Wilhelm et al., 2006;Roberts et al., 2007). Taken together, the evidence suggests that T. pseudonana acclimates to low inorganic carbon concentration by increasing its affinity for inorganic carbon using a biophysical CCM. Inorganic carbon depletion experiments with T. pseudonana and T. weissflogii showed similar growth rates as functions of inorganic carbon concentration (Clark and Flynn, 2000). Hence, any differences in the CCMs of these diatoms are not reflected in their growth rates.
Our results indicate significant rates of photorespiration in both T. weissflogii and T. pseudonana, and suggest that glycolate 2-P is metabolized by some PCOC, which is still not completely resolved in diatoms (Wilhelm et al., 2006). Transcription of GDCP was highly upregulated by low CO 2 concentration in T. pseudonana (Fig. 3), in accordance with previous evidence of high light induction of the Gly decarboxylase T-subunit in T. pseudonana (Parker and Armbrust, 2005) and T. weissflogii (Parker et al., 2004). Early PCOC intermediates were more extensively labeled in T. pseudonana than in T. weissflogii (Fig. 1). These data suggest that the combined biochemical and biophysical CCM in T. weissflogii is more effective in suppressing Rubisco oxygenase activity than is the exclusively biophysical CCM of T. pseudonana.
The work reported here highlights the hazards of assuming that the pathway(s) of photosynthetic carbon assimilation is consistent throughout a genus of diatoms. Interestingly, a recent molecular phylogenetic study showed that the genus Thalassiosira is paraphyletic, with T. weissflogii and T. pseudonana rather distantly related within the Thalassiosira plus Skeletonema clade (Kaczmarska et al., 2006). Variation in photosynthetic pathways is already recognized in many genera of flowering plants (Monson et al., 1986), and therefore caution should be exercised when inferring metabolic Figure 3. Gene transcripts (mRNA) of PEPC1, PEPC2, PEPCK, RBCL, and GDCP (normalized to Act1) at the start of the light period (9 AM; 09:00) and dark period (9 PM; 21:00) in T. pseudonana grown at near-ambient (380 mL L 21 ) or low (100 mL L 21 ) CO 2 . Data are means 6 SE of three separate determinations. Figure 4. Immunoblots with antisera against PEPC1, PEPC2, PEPCK, and RBCL proteins from T. pseudonana grown at near-ambient (380 mL L 21 ) or low (100 mL L 21 ) CO 2 (full-length blots in Supplemental Fig. S3). function in vivo from genomic data (Armbrust et al., 2004;Montsant et al., 2005;Derelle et al., 2006). This study suggests that diatoms utilize a combination of complementary CCMs, which may confer plasticity in acclimating to the changing atmospheric (and sea surface) CO 2 concentrations and variations in the availability of other resources (Giordano et al., 2005).

Metabolic 14 C Labeling
To optimize photosynthetic 14 C fixation, metabolic labeling experiments were conducted at mid-light period when circadian-regulated photosynthetic capacity is at a maximum (Putt and Prézelin, 1988). Triplicate cultures were harvested on membrane filters (0.45-mm pore), resuspended in 600 mL NaHCO 3 -free Aquil (rather than the less ecologically relevant sorbitol buffer [Reinfelder et al., 2000[Reinfelder et al., , 2004) to give densities of 2.5 to 5.0 3 10 8 cells mL 21 for T. pseudonana and 2.5 to 5.0 3 10 7 cells mL 21 for T. weissflogii, and incubated at 200 mmol quanta m 22 s 21 at 22°C. Metabolic labeling was initiated by adding 600 mL Aquil containing 2.4 mM NaH 14 CO 3 (1.0 MBq), and terminated after 2 to 30 s by adding 2.4 mL of 100°C Milli-Q water. Hence, during labeling the cells were exposed to an inorganic carbon concentration (1.2 mM) and pH intermediate between those of the two growth media. The inorganic carbon system in the labeling solution was at, or very close to, chemical and isotopic equilibrium. Cells were extracted at 100°C for 30 min, acidified (pH # 2.0) for 24 h to remove inorganic 14 C, freeze-dried, and resuspended in 2.0 mL of Milli-Q water. After centrifugation, the supernatant was freeze-dried to a final volume of 300 mL, and a 30-mL portion was assayed for radioactivity using a Tri-Carb 3100TR liquid scintillation analyzer with QuantaSmart Version 1.31 software (Packard). The 14 C signal retained in the acid-stable products constituted ,0.5% of the total 14 C available from NaH 14 CO 3 . The efficiency of the termination method was examined in experiments with T. pseudonana. When the labeling reaction was terminated immediately after the addition of NaH 14 CO 3 , the total acid-stable 14 C signal was only 1% of that after 5-s reaction, and the samples showed no discernable peaks in HPLC analysis. After cell extraction and centrifugation, 91% 6 5% (n 5 47) of all acid-stable radioactivity was recovered in the supernatant. An ANOVA test showed that neither labeling time nor cell species had any effect on the proportion of radioactivity recovered, indicating that the 14 C signal retained in the cell debris came from residual extract. To separate unresolved peaks in HPLC analysis, half of each extracted sample was treated with calf intestinal alkaline phosphatase (Sigma-Aldrich) in 50 mM Tris-HCl, pH 9.8, for 16 h. Native and dephosphorylated samples were analyzed separately on a Gynotek HPLC using a CoreGel 64H interaction column (Transgenomic) with 8 mM H 2 SO 4 as the mobile phase and a flow rate of 0.6 mL min 21 . Scintillant (Ultima-Flo M cocktail; Packard) was mixed with the column eluate at a rate of 3 mL min 21 , and radioactivity was measured using a flow scintillation detector (Packard 150TR). HPLC results were visualized and analyzed with Chromeleon Version 6.60 SP1a Build 1449 software (Dionex). Peak identification was based on the co-retention of a series of 14 C-labeled organic and amino acid and sugar standards, using both 14 C and UV detection (sugars give no UV signal). Individual 14 C peaks were computed as percentages of the total combined peaks detected by HPLC.

Gene Transcript Analysis
Gene-specific primers for PEPC1, PEPC2, PEPCK, RBCL, GDCP, and the housekeeping gene actin 1 (Act1) were designed (Supplemental Table S2) using genomic sequence data for T. pseudonana (http://genome.jgi-psf.org/ Thaps3/). Triplicate cultures of T. pseudonana were harvested at the start of the light period (9 AM) and dark period (9 PM) by centrifugation or gentle filtration and flash frozen. Cellular genomic DNA and total RNA were isolated with the DNeasy and the RNeasy plant mini kit (Qiagen), respectively. RNA was treated with RNase-free DNase (Qiagen) and tested for DNA contamination by PCR. One to 10 mg of RNA was reverse transcribed to cDNA using 150 units of BioScript RNase H Minus (Bioline) with 1 mg oligo(dT) 15 primers (Promega), 0.1 mmol dNTP mix (Bioline), and 40 units of rRNasin RNase inhibitor (Promega). The cDNA was purified with a QIAquick PCR purification kit (Qiagen). cDNA and genomic DNA standards were amplified by real-time PCR using gene-specific primers and Power SYBR Green PCR master mix (Applied Biosystems) with an ABI Prism 7700 detection system (Applied Biosystems) run at 50°C for 2 min, 95°C for 10 min, and 40 cycles of 95°C for 15 s and 60°C for 1 min. cDNA was quantified by DNA standard curves, and normalized to Act1. PCR products were resolved by PAGE and stained with ethidium bromide.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Table S1. Cell size measurements.