A common partitioning strategy for photosynthetic products in evolutionarily distinct phytoplankton species


Author for correspondence:

Kimberly H. Halsey

Tel: +1 541 737 1831

Email: halseyk@science.oregonstate.edu


  • We compare the nutrient-dependent photosynthetic efficiencies of the chlorophyte, Dunaliella tertiolecta, with those of the marine diatom, Thalassiosira weissflogii. Despite considerable evolutionary and physiological differences, these two species appear to use nearly identical growth strategies under a wide range of nutrient limitation.
  • Using a variety of physiological measurements, we find that, for both species and across all growth rates, 75% of the gross photosynthetic electron flow is invested in carbon fixation and only 30% is retained as net carbon accumulation. A majority of gross photosynthesis (70%) is ultimately used as reductant for biosynthetic pathways and for the generation of ATP.
  • In both species, newly formed carbon products exhibit much shorter half-lives at slow growth rates than at fast growth rates. We show that this growth rate dependence is a result of increased polysaccharide storage during the S phase of the cell cycle.
  • We present a model of carbon utilization that incorporates this growth rate-dependent carbon allocation and accurately captures (r2 = 0.94) the observed time-resolved carbon retention. Together, our findings suggest a common photosynthetic optimization strategy in evolutionarily distinct phytoplankton species and contribute towards a systems-level understanding of carbon flow in photoautotrophs.


Throughout the biosphere, photoautotrophs harvest the energy of sunlight to extract electrons from water. Following the initial water-splitting reaction that generates electrons and molecular oxygen (O2), a diversity of pathways consume the products of photosynthesis and alter the yield of biomass produced per O2 evolved. The balance of these pathways is influenced by environmental growth conditions, making an understanding of this regulation essential for the accurate assessment of biospheric net production from remotely detectable properties (e.g. pigment concentration, leaf area index, daily light absorption). These remotely sensed variables are more closely aligned with gross O2 production than with net carbon accumulation. Photosynthate partitioning is also reflected in the biochemical composition of photoautotrophic cells (Wagner et al., 2006; Jakob et al., 2007; Halsey et al., 2011), which is relevant to biotechnological applications, such as the production of biofuels (Nogales et al., 2012). Finally, modern field measurements employ tracers that detect photosynthetic rates ranging from gross O2 production to net carbon accumulation. The interpretation of the relationships between these observed rates requires an understanding of the mechanisms that underlie photosynthate allocation and their environmental dependences.

Total photosynthetic electron flow is quantitatively reflected by gross oxygen production (GPO2). Only a fraction of these electrons are involved in the reduction of CO2 by the Calvin–Benson cycle to form organic compounds. The remaining electrons are largely consumed by pathways operating only in the light that function to generate ATP, protect the photosystems from damage and chemically reduce nutrient molecules. Thus, early in the photosynthetic process, there is a distinction between GPO2 and gross carbon fixation (GPC). The simple organic export product of the Calvin–Benson cycle (i.e. glyceraldahyde-3-phosphate, GAP) likewise may be used in a wide variety of metabolic pathways that generate the basic building blocks of the cell or fuel catabolic reactions supplying ATP and reductant for cellular processes. The balance of these pathways influences the average lifetime of newly formed organic material and defines the fraction of gross photosynthetic electron flow that is retained by the cell for growth. Thus, another distinction exists between GPC and net carbon production (NPC). Finally, only a portion of electron accumulation registered as net oxygen production (NPO2) remains in carbon molecules that have reduction states equivalent to GAP. The remaining electrons are used to create compounds that are more reduced than GAP (e.g. lipids, nucleic acids), resulting in a further distinction between NPO2 and NPC.

In oceanography, phytoplankton photosynthetic efficiency is commonly reported as the rate of O2 or carbon production normalized to chlorophyll concentration or spectral absorption, parameters that can be routinely assessed in the field and through global remote sensing. By contrast, direct assessment of biomass (phytoplankton carbon) is not yet routine (Graff et al., 2012). Furthermore, remote sensing retrievals of chlorophyll and spectral absorption can be related to primary production through field-based measurements of photosynthetic efficiencies, thus achieving global estimates for use in ocean productivity models (Falkowski & Raven, 1997).

Aquatic microalgal (phytoplankton) cultures provide simple systems for the evaluation of the environmental regulation of photosynthetic efficiencies. In an earlier study (Halsey et al., 2010), we described the partitioning of photosynthetic electron flow from GPO2 through to NPC in steady-state, NO3-limited cultures of the marine chlorophyte, Dunaliella tertiolecta. We focused on impacts of nutrient limitation, because the most common condition in the surface ocean mixed layer is macronutrient (N or P) or micronutrient (e.g. Fe) limitation, and because it is often assumed that nutrient stress results in lower photosynthetic efficiencies. Our results showed that chlorophyll-specific GPO2, GPC and NPC were invariant across growth rates, that a majority of gross photosynthesis was not retained as net carbon accumulation, and that the lifetimes of newly formed carbon products increased with growth rate. These findings allowed the quantitative assessment of growth rate-dependent photosynthate allocation, but the results were limited to D. tertiolecta.

Here, we compare the nutrient-dependent photosynthetic efficiencies of D. tertiolecta with those of the marine diatom, Thalassiosira weissflogii. Diatoms are photosynthetic protists that emerged from a secondary endosymbiotic event and appear in the fossil record to only c. 125 million yr ago. By contrast, chlorophyta are ‘true plants’ from a far older primary endosymbiotic event of > 1.5 billion yr ago (Rasmussen et al., 2008; Parfrey et al., 2011). Diatoms and chlorophyta have different photosynthetic light-harvesting structures and carbon metabolism (e.g. diatoms lack the oxidative pentose phosphate pathway in their chloroplasts) (Wilhelm et al., 2006). Given these deeply rooted evolutionary and physiological differences, there was no a priori reason to assume that similar strategies would be employed by the two groups with respect to photosynthate partitioning under nutrient stress. Nevertheless, we find nearly identical results for our two test species, suggesting that a general cell model of photosynthate flow may apply for a diverse range of phytoplankton taxa. We further investigated the basis for growth rate-dependent carbon lifetimes and propose that this phenomenon is, in part, linked to the photosynthetic properties of specific cell cycle phases. Our findings indicate that a systems-level understanding of photosynthetic metabolism can yield fundamental insights that apply across broad taxonomic categories.

Materials and Methods

Culture conditions for the measurement of photosynthetic efficiency

Thalassiosira weissflogii (Grunow) and Dunaliella tertiolecta (Butcher) were grown at 20°C in nitrate-limited steady-state and axenic chemostats. The growth conditions (24 h of constant light and nutrient limitation) for both species were comparable. The growth medium for T. weissflogii was artificial seawater f/2 medium with 150 μM Na2SiO3 and 80 μM NaNO3 added as the limiting nutrient. The growth medium for D. tertiolecta was f/2 medium with no added Na2SiO3 and 100 μM NaNO3 added as the limiting nutrient. Nitrogen-limited growth rates were established by the rate of dilution of the culture with fresh medium according to the equation

display math

where μ is the specific growth rate (d−1), D is the dilution rate (ml d−1) and V is the culture volume (ml) maintained by siphon tubing (= 0.90 l for measures of gross and net O2 production and light-dependent respiration (LDR), and 0.13 l for 24-h 14C uptake time course). Cultures were aerated by bubbling with filtered air and were provided with constant 24-h light from cool-white fluorescent tubes (General Electric, Fairfield, CT, USA). Growth irradiance was nearly saturating at 220–250 μmol photons m−2 s−1 as measured with a quantum meter (Biospherical Instruments, San Diego, CA, USA, model QSL-100) fitted with a 4π spherical quantum sensor. Following at least seven generations, the cell density was monitored daily. When the cell numbers had been stable for three consecutive days, cultures were considered to be at steady state.

Measurement of photosynthetic efficiency

Phytoplankton photosynthetic efficiency is commonly reported as the rate of O2 or carbon production normalized to chlorophyll concentration or spectral absorption, parameters that can be routinely assessed in the field and through global remote sensing. Photosynthetic efficiencies normalized to chlorophyll a have been reported previously for D. tertiolecta in Halsey et al. (2010). Productivity per unit carbon increases linearly with growth rate (Supporting Information Fig. S2); however, cellular carbon decreases and cellular chlorophyll content increases with the nutrient-limited growth rate (Laws & Bannister, 1980; Halsey et al., 2010).

Triplicate measurements of cell density, diameter and volume were averaged using a Multisizer 3 Coulter counter equipped with a 70-μm aperture (Beckman Coulter, Miami, FL, USA). Chlorophyll concentrations were determined from 6-ml samples filtered onto 25-mm glass fiber filters (Whatman, Buckinghamshire, UK) that were extracted overnight in 90% acetone. The absorptivity of the extract was measured spectrophotometrically and chlorophyll was quantified using the equation of Jeffrey & Humphrey (1975). The spectrally averaged cross-sectional area (ā*; used to calculate absorbed light; Falkowski et al., 1985) was measured using the filter pad method with appropriate path length corrections (Mitchell et al., 2003). All production measurements were normalized to absorbed light according to the equation

display math

where P * is the photosynthetic efficiency, P is the rate of oxygen or carbon production, Ig is the growth irradiance and ā* is the spectrally averaged cross-sectional area.

For CHN analysis, 1-, 2- and 3-ml samples were filtered on precombusted GF/F filters to ensure a linear relationship between carbon or nitrogen and the volume filtered. The y-intercept (filter blank) was subtracted from sample mass values. Total carbon and nitrogen were measured with an Exeter Analytical (North Chelmsford, MA, USA) EA1 elemental analyzer. Net carbon production is the product of μ and cellular carbon. The fraction of gross photosynthetic electron flow used directly for nitrogen and sulfur reductions (DUNS) was calculated using cellular nitrogen and an N : S ratio of 16 : 1.3 (Ho et al., 2003) according to the equation

display math

where the measurement of GPO2 and LDR is described below. These DUNS values were used to calculate GPC according to the equation

display math

Membrane inlet mass spectrometry (MIMS) was used to simultaneously measure real-time net O2 production, gross O2 production and light-dependent consumption of O2 using 18O2 as a tracer, as described in Halsey et al. (2010). A membrane inlet system mounted to a 5-ml water-jacketed incubation chamber, maintained at 20°C and continuously mixed using a magnetic stir bar (c. 8 Hz), was attached to a Prisma QMS-200 (Pfeiffer, Asslar, Germany) quadrupole mass spectrometer with closed ion source and electron multiplier detector recording at mass/charge ratios of 32 (16O2), 36 (18O2) and 40 (Ar). Oxygen signals were calibrated with O2-saturated water and zero-O2 water (+sodium dithionite) and normalized to Ar. We verified that cellular 18O2 consumption matched 16O2 consumption in the dark when the initial fraction of 18O2 was varied from 38 to 90 μM. Thalassiosira weissflogii was concentrated by membrane filtration to 1.1–2.4 chlorophyll a ml−1. The suspension was initially exposed to a 4–6-min dark period, followed by exposure to seven light levels (0–750 μmol quanta m−2 s−1) for 3–5 min each, alternating with 1–2-min dark periods. Measurements were concluded with a 4–6-min dark period. Rates of oxygen production and consumption were calculated by dividing the observed rates by the fraction of 16O2 or 18O2 present during exposure to each light level, thus taking into account changes in isotope dilution throughout the experiment. Photosynthesis–irradiance curves of gross and net O2 production were modeled using nonlinear least-squares regression analysis and a hyperbolic tangent model (Jassby & Platt, 1976) to estimate O2 production at the growth irradiance. Net O2 production rates were converted to carbon units using DUNS values to allow comparison with GPC and NPC. Thus, carbon-based net O2 production was calculated as

display math

Measurements to determine the time dependence of 14C uptake were performed as described previously (Halsey et al., 2011). Briefly, freshly prepared bovine carbonic anhydrase (CA) (Sigma-Aldrich) was added to cultures (5 U ml−1) before inoculation with NaH14CO3 (1 μCi ml−1) to ensure isotopic equilibrium was quickly attained among dissolved inorganic carbon (DIC) species. Sampling for 14C uptake rates commenced 2 min later and continued periodically over 24 h. At each time point, 2-ml samples were withdrawn. Two 50-μl subsamples were added to 50 μl phenethylamine and 900 μl H2O to determine the remaining NaH14CO3 activity. Two 0.5-ml subsamples were transferred to 7-ml scintillation vials and immediately acidified and degassed for 24 h before the measurement of assimilated 14C by liquid scintillation counting. Changes in specific activity caused by dilution were determined using the measured remaining activity values at each time point (DPMt) and the calculated time-weighted average DPM for each time interval according to the equation

display math

where DPMtn and DPMt(n–1) are disintegrations per minute used to calculate the carbon fixation rate at time point tn and the previous time point, Δt is the time interval separating tn and t(n–1) and μ is the specific growth rate. Cultures were maintained at steady-state growth rates throughout the experiment by the adjustment of D and the level of the siphon tube according to changes in V caused by sampling.

Culture conditions and measurement of cell cycle-dependent carbon metabolism

This project began with a comparison of photosynthetic efficiencies in the green alga, D. tertiolecta, and the diatom, T. weissflogii. In an initial attempt to investigate linkages between these productivity properties and temporal carbon allocation, we attempted to synchronize the growth of these two strains with respect to their cell cycles. For both strains, synchronization strategies included cyclostat growth (nitrogen-limited chemostat with a light : dark cycle as demonstrated for Prochlorococcus PCC9511; Bruyant et al., 2001), growth with cell cycle inhibitors (demonstrated for Cylindrotheca fusiformis; Claquin et al., 2004) and semi-continuous culturing using nitrogen-replete media and light : dark cycles. None of these strategies yielded synchronized growth in our test strains. Therefore, we used the closely related strain Dunaliella bioculata that has been shown previously to be capable of synchronous growth (Marano et al., 1978). Dunaliella bioculata was maintained in semi-continuous, axenic culture for 14 d in nutrient-replete Erdshreiber's medium at 20°C on a 8.7 h : 15.3 h light : dark cycle. Growth irradiance was 250 μmol photons m−2 s−1. These growth conditions resulted in a division rate of 1 d−1 (μ = 0.69 d−1). Cell density was measured using a Coulter counter and maintained between 4 × 104 and 3.6 × 105 cells ml−1 for at least seven generations. For cell cycle analysis, 0.5 ml of culture was fixed with 2 μl of 0.05% glutaraldehyde and incubated at room temperature for 15 min before freezing in liquid nitrogen. Samples were stored at −80°C until analysis. Cellular DNA was stained with 2.5 μl of SYBR Green I (Invitrogen) for 30 min before measurement using a flow cytometer (Influx; Becton Dickinson, Franklin Lakes, NJ, USA). Ten thousand events were collected and the fraction of cells in G1, S and G2-M cell cycle phases were determined using the M-cycle.

For short-term (20-min) 14C uptake incubations, 5-ml culture samples of D. bioculata were diluted with 8 ml of Erdshreiber's medium and inoculated with 5 μCi NaH14CO3. The inoculated sample was aliquoted into 12-ml scintillation vials (1 ml each) and placed in a photosynthetron (CHPT Mfg Inc., Georgetown, DE, USA) for 20 min at 20°C at 11 light levels (0–750 μmol quanta m−2 s−1 from two tungsten projector bulbs) (Lewis & Smith, 1983). Source light was attenuated with neutral density filters and measured with a quantum meter with a 4π sensor. Following incubation, samples were acidified and degassed for 24 h. The total activity of NaH14CO3 added was determined by adding 50 μl of inoculated sample to 50 μl of phenethylamine and 900 μl of H2O. Photosynthesis–irradiance curves were modeled using nonlinear least-squares regression analysis and a hyperbolic tangent model to estimate * and alpha.

For 14C-pulse labeling studies, samples (3 × 20 ml) of synchronized cultures were pulse labeled for 20 min with 5 μCi NaH14CO3 at 20°C at the growth irradiance. Samples were filtered (GF/F), rinsed with 3 × 10 ml of unlabelled Erdshreiber's medium and frozen at −20°C for later extraction, fractionation and quantification of labeled biochemical pools. Biochemical fractionation of polysaccharides followed the procedures of Smith & Geider (1985) and Smith et al. (1997). Filters were dried and extracted for 24 h in 1.5 ml of 2 : 1 chloroform–methanol (v/v) at −20°C. The extract was filtered through a GF/F filter and the filters were rinsed twice with 1.5 ml of chloroform–methanol. The filter was extracted in 5% trichloroacetic acid (TCA) for 1 h at 80°C. The extract was filtered through a new GF/F filter and both filters were rinsed twice with 1.5 ml of 5% TCA to recover polysaccharides (and some nucleic acids; however, nucleic acids are <5% of total cellular carbon (Williams & Laurens, 2010) and were not detected in the ‘hot TCA-soluble’ fraction (Morris et al., 1974), and are therefore an insignificant fraction of the polysaccharide signal).

Development of a cell model of carbon utilization

Cellular carbon allocation and mobilization were modeled by considering that newly fixed carbon is allocated to two pools with different half-lives: (1) a short-half-life pool representing GAP that is rapidly mobilized for respiration and biosynthetic purposes; and (2) a long-half-life pool representing GAP that is allocated to a storage polysaccharide. Our data indicate that, during the S phase of division, newly formed GAP is allocated to the long-half-life pool. For the remainder of the cell cycle, newly formed GAP is allocated to the short-half-life pool. Thus, in a population of cells growing at the same rate, but randomized in terms of their cell cycles, newly formed GAP is allocated to each pool as a function of the fraction of the population undergoing division at any point in time. A fast-growing population has a greater fraction of dividing cells than a slow-growing population. Therefore, in a fast-growing population, a greater fraction of GAP is allocated to a storage polysaccharide form compared to a slow-growing population.

Carbon allocation to the two pools occurs simultaneously with CO2 fixation, and the rate of CO2 fixation to GAP remains constant. Carbon fixation and mobilization were modeled in discrete 1-min time steps. The parameters used for our runs of the model are shown in Table 1.

Table 1. Values and descriptions of the model parameters
ParameterDescriptionUnits Dunaliella tertiolecta Thalassiosira weissflogii
r F Common ratio for the ‘fast-burn’ (short-half-life) carbon poolDimensionless0.2000.300
r S Common ratio for the ‘slow-burn’ (long-half-life) carbon poolDimensionless0.9940.997
m Slope of line describing growth rate-dependent carbon allocationmmol C h−10.01350.0145
b Offset of line describing growth rate-dependent carbon allocationmmol C−0.006−0.0055
GPCAbsorption-normalized rate of gross carbon fixationmmol C (mmol photons)−10.03690.0365
NPCAbsorption-normalized rate of net carbon accumulationmmol C (mmol photons)−10.01320.0100
Doubling timeRelated to specific growth rate (μ) by: inline imageh13.9, 33.3, 138.616.6, 33.3, 83.2

Here, we first show CO2 fixation and mobilization by one pool. At any given point in time (t), the realized rate of carbon production measured by 14C uptake is the sum of: GAP produced at t, GAP produced at – 1 less the carbon that was oxidized between t – 1 and t, and GAP produced at – 2 less the carbon that was oxidized between t – 2 and t, etc. This cascading process is the sum of a geometric series (S) as shown in Eqn 1

display math(Eqn 1)

where n is the number of steps in the series and r is the common ratio that is the carbon available for utilization between successive steps, and is between zero and one with oxidation rates increasing inversely with r. It should be noted that the half-life (τ, in min) is related to r by:

display math

Equation 1 can be rewritten:

display math(Eqn 2)

Measurement of 14C uptake at step n gives the average of S (Save):

display math(Eqn 3)

Equation (Eqn 3) was converted to a time series with observations at time (t) in minutes and, in our model, carbon utilization initiates 1.5 min after 14C-labeled sodium bicarbonate is added to the cell suspension, thus yielding:

display math(Eqn 4)

Our data indicate that newly fixed carbon is allocated to two carbon pools with different half-lives. The amount of newly fixed carbon allocated to each pool is a linear function of growth rate and is directly proportional to log10 of the culture doubling time (in hours):

display math(Eqn 5)

where CF is the fraction of newly fixed carbon allocated to the carbon pool oxidized at a ‘fast-burn’ rate (short half-life) and the parameters m and b are taken from Table 1. The amount of newly fixed carbon allocated to the carbon pool oxidized at a ‘slow-burn’ rate (long half-life) is:

display math(Eqn 6)

where (GPC – NPC) is the ‘transient carbon pool’, which is the total amount of carbon available for oxidation. The apparent rate of carbon fixation for the ‘fast-burn’ pool (TF) at any time is given by:

display math(Eqn 7)

where rF is the ‘fast-burn’ ratio (r) of carbon oxidation and is related to the turnover time; conversely, the apparent rate of carbon fixation for the ‘slow-burn’ pool (TS) at any time is given by:

display math(Eqn 8)

where rS is the ‘slow-burn’ ratio of carbon oxidation. Therefore, P (the realized rate of carbon production measured by 14C uptake) is:

display math(Eqn 9)

Results and Discussion

Photosynthetic efficiencies in evolutionarily distant species

Steady-state, nitrogen-limited growth rates for T. weissflogii ranged from 0.2 to 1.0 d−1, which is only slightly less than the 0.12–1.2 d−1 range used earlier for D. tertiolecta (Halsey et al., 2010). Thalassiosira weissflogii chlorophyll content (chlorophyll a : C) varied in direct proportion to the growth rate (r2 = 0.99; Table 2), implying precisely tuned, highly efficient photosynthesis at all levels of NO3 limitation. This same linear relationship is observed in T. weissflogii grown under NH4 or PO4 limitation (Laws & Bannister, 1980). Qualitatively similar responses are observed among chlorophytes, haptophytes and other diatoms (Laws & Wong, 1978; Herzig & Falkowski, 1989; Halsey et al., 2010). However, quantitative comparison of photosynthesis between evolutionarily distant groups requires an accounting for taxonomic differences in light-harvesting pigments. We therefore calculated photosynthetic light utilization efficiencies by normalizing oxygen and carbon production to spectral absorption for T. weissflogii and D. tertiolecta (data for D. tertiolecta were re-analyzed from Halsey et al., 2010).

Table 2. Steady-state characteristics of nitrate-limited chemostat cultures of Thalassiosira weissflogii (values in parentheses are SE)a
μ (d−1)Cells ml−1 (×105)Chlorophyll (Chl) a per cell (pg Chla)Chl : C (μg μg−1) × 10−3Particulate organic carbon per cell (pg C)C : N (μg μg−1)Cell volume (μm3)
  1. a

    Note that steady-state characteristics of nitrate-limited chemostat cultures of Dunaliella tertiolecta have been reported previously in Halsey et al. (2010).

0.201.29 (0.11)0.57 (0.15)3.3 (0.8)174 (1.9)15.0 (1.3)414 (13)
0.501.53 (0.11)1.10 (0.18)10.8 (1.3)106 (17)9.5 (1.5)361 (11)
1.01.41 (0.08)1.32 (0.14)23.8 (2.0)55.8 (6.5)4.3 (0.2)274 (10)

Absorption-specific GPO2, GPC and NPC are invariant across growth rates and are remarkably consistent between the two species (Fig. 1). Light utilization efficiencies for gross oxygen evolution are tightly constrained between 0.047 and 0.050 mmol (mmol photon)−1. For both species, c. 25% of GPO2 is immediately consumed by LDR pathways and for reduction of nitrogen and sulfur (Fig. 1, light and dark gray areas, respectively). Thus, 75% of GPO2 is invested in gross carbon fixation (GPC). For both species, only c. 30% of GPO2 is ultimately retained as net carbon production (NPC) at all growth rates (Fig. 1, green areas). The remaining c. 45% of photosynthetic production represents a transient carbon pool that is soon consumed for either mitochondrial ATP production (O2 consuming) (yellow areas in Fig. 1) or catabolism that provides electrons for carbon reduction (blue areas in Fig. 1). In both species, these reductive pathways are greater at faster growth rates (Fig. 1). Thalassiosira weissflogii mobilizes a larger fraction of its transient carbon pool for mitochondrial respiration, whereas, in D. tertiolecta, the reductive pathways dominate (compare blue and yellow wedges between species in Fig. 1). The overall cellular reduction states of green algae and diatoms increase with growth rate (Smith & Geider, 1985; Jakob et al., 2007; Halsey et al., 2011), a phenomenon that is reflected by a higher fraction of lipid and protein carbon in fast-growing D. tertiolecta relative to slow-growing cells (Halsey et al., 2011).

Figure 1.

Absorption-normalized photosynthetic efficiencies in Dunaliella tertiolecta and Thalassiosira weissflogii across a wide range of nitrate-limited specific growth rates. Measurements of gross O2 production (GPO2), light-dependent respiration (LDR), net O2 production (NPO2) and net carbon production (NPC) yielded data symbols and regression lines. Error bars are SE of measurements made from triplicate chemostat cultures. Colored sections show the allocation of photosynthetic electron flow to different metabolic sinks. Data for D. tertiolecta were reanalyzed from Halsey et al. (2010).

Our findings for T. weissflogii and D. tertiolecta highlight multiple significant features of photosynthesis in microalgae. First, highly comparable photosynthetic efficiencies were observed in evolutionarily distinct species when differences in light-harvesting pigment compositions were taken into account (i.e. if T. weissflogii data were normalized to chlorophyll a alone, the apparent photosynthetic efficiencies would be 24% higher than chlorophyll a-normalized data for D. tertiolecta; Fig. S1). Second, steady-state nutrient limitation is not synonymous with diminished photosynthetic efficiencies. Third, the observed GPO2/NPC ratios of 3.3 and 3.5 provide a baseline for comparison with values assessed in the field. This ratio is critical to the assessment of ecosystem carbon cycling from incubation-free oxygen isotope measurements. A canonical value of 2.7 is typically used in these assessments to interconvert GPO2 and NPC (Marra, 2001). However, in practice, this ratio appears to vary between at least 1.5 and 4 (Wagner et al., 2006; Jakob et al., 2007; Langner et al., 2009), with some studies suggesting an even broader range (e.g. Luz & Barkan, 2009; but note that the recent study of Nicholson et al. (2012) suggests that these extremes may reflect an assessment error). If diverse phytoplankton behave similarly to T. weissflogii and D. tertiolecta, explanations other than nutrient limitation might be needed to explain deviations in GPO2/NPC.

Our investigation also leads to the conclusion that different arrangements of biochemical pathways and energetic allocation strategies can result in similar growth efficiencies. Photosynthetic light reactions give an ATP/NADPH ratio of c. 3 : 2, similar to the consumption ratio of the Calvin–Benson cycle. However, this energetic stoichiometry does not capture the overall cellular cost for net carbon accumulation. The oxygen-consuming pathways, LDR and mitochondrial respiration, together constitute 42–59% of GPO2 (light gray and yellow wedges in Fig. 1), consistent with reports that respiration can account for 21–89% of GPC (Geider, 1992; Langdon, 1993) and 41–69% of GPO2 (Kunath et al., 2012). Another 13–29% of GPO2 is used to reduce nitrogen, sulfur and organic compounds (Fig. 1 dark gray and blue wedges). Overall, we find investment ratios for mitochondrial respiration, nonrespiratory carbon reduction and NPC of 1.4 : 1.0 : 1.0 for D. tertiolecta and 2.1 : 0.5 : 1.0 for T. weissflogii. These ratios reflect the relatively high lipid content of D. tertiolecta, a characteristic that makes this species attractive for biodiesel production (Williams & Laurens, 2010). Viewed from the perspective of their relative investment ratios, these two species are distinct in their photosynthate investment strategies, despite their striking similarity in photosynthetic efficiencies (Fig. 1). Differences in photosynthate investments between species are also found among cryptophytes and cyanobacteria (Kunath et al., 2012) and diatoms (Su et al., 2012).

Linking transient carbon lifetimes to the cell cycle

For both D. tertiolecta and T. weissflogii, the transient carbon pool is 60–63% of gross carbon fixation (i.e. the area between GPc and NPc in Fig. 1). By conducting time-resolved measurements of 14C uptake, Halsey et al. (2011) showed that the average lifetime of this transient carbon increases significantly with increasing growth rate in D. tertiolecta. We find nearly identical results for T. weissflogii (Fig. 3). To explain these observations, we hypothesized that carbon is allocated to different end products during different phases of the cell cycle. Specifically, we envisioned the growth phase (G1) as a period when newly formed GAP is rapidly mobilized for ATP production and carbon reduction (or direct reductant utilization; see Halsey et al., 2011), whereas the division phase of the cell cycle (S, G2 and M) is a period when a significant fraction of GAP is first converted to a temporary storage polysaccharide (thus imparting a longer average lifetime). Given that the duration of the division phase is constant and independent of growth rate (Burbage & Binder, 2007; Matsumura et al., 2010), it follows that a rapidly growing population will have a greater fraction of cells in the division phase than a slower growing population, thus yielding longer average transient carbon lifetimes when growth rates increase. Because cell cycles of D. tertiolecta and T. weissflogii are not readily synchronized in culture, we tested this hypothesis using the related species, Dunaliella bioculata, which is easily synchronized (Marano et al., 1978).

For synchronized D. bioculata cultures growing at one division per day, the S phase spanned a 6-h period that overlapped with the G2-M phase, which spanned another 6-h period. The 20-min 14C uptake was similar for G2-M and G1 phases, but was notably higher during S phase (Fig. 2a). Furthermore, polysaccharide production roughly doubled during S phase compared with the other cell cycle stages and resulted in a rapid drop in cellular N : C stoichiometry (Fig. 2b). This storage carbon was later mobilized during the G2-M phase, thereby restoring the N : C ratio (Fig. 2b). These results implicate the S phase specifically as the period of enhanced GAP allocation to a storage pool. More broadly, we interpret our results as suggesting that the S phase is a period of lower energy requirements in which new photosynthate can be stored for the subsequent, energy-demanding cell division event.

Figure 2.

Cell dynamics of Dunaliella bioculata during synchronized growth (μ = 0.69 d−1) across a complete cell cycle. Cells were synchronized using an 8.7 h : 15.3 h light : dark cycle (black bar). Data points are every 3 h over 24 h. (a) Cell number (X) doubles during G2-M phase (green X) and 14C-based photosynthesis–irradiance parameters peak during S phase. Error bars are SE of the estimate for P* (closed triangles) or alpha (open triangles). (b) Polysaccharide synthesis (closed squares) peaks during S phase. Polysaccharide metabolism strongly impacts the ratio of N and C (open circles).

Earlier investigations of gene expression during the cell cycle showed that synthesis and degradation genes for storage polysaccharides are up-regulated just before S phase and G2-M phase, respectively (Mohr et al., 2010). Taken together with our cell cycle-resolved data on carbon pools, these gene expression results imply a temporal disjunction between gene expression and protein activity.

Our results for D. bioculata provide a mechanistic explanation for the observed growth rate dependence of 14C uptake data. Drawing from these results, we developed a cell model of carbon utilization for D. tertiolecta and T. weissflogii (see the 'Materials and Methods' section for details). In the model, newly fixed carbon is divided into a pool of short-lived (‘fast-burn’) GAP pathways that dominate during G1 and G2-M phases (note that, in diatoms, inorganic carbon can pass through a C4 form which may provide a small fraction of carbon backbones) and a longer lived (‘slow-burn’) pool, representing the temporary polysaccharide storage of S phase. The allocation of carbon between these two pools was described as a function of growth rate (i.e. fraction of the population in division). The application of this model to 14C uptake data collected across 24-h time courses for D. tertiolecta and T. weissflogii yields an excellent description (r2 = 0.94 for both species, Fig. S3) of the observed photosynthetic efficiencies across all growth rates and incubation time scales (Figs 3, S3, S4). Moreover, this performance is achieved with model parameters that are very similar for the two species, where optimized half-lives for the ‘fast-burn’ and ‘slow-burn’ carbon pools were 26 s and 1.9 h, respectively for D. tertiolecta and 34 s and 3.8 h, respectively for T. weissflogii. These parameters again highlight the remarkable similarity of photosynthetic and catabolic processes in these two evolutionarily distinct species.

Figure 3.

Growth rate-dependent changes in lifetimes of newly fixed carbon (C) cause 14C-based production to decline over time. Comparison of data (symbols) collected from 14C time courses using nitrogen (N)-limited steady-state cultures grown under constant light, and a new model of C utilization describing C fixation and allocation to two pools with different mobilization rates (lines). Allocation of newly fixed C is a function of growth rate, with a greater fraction of C allocated to a temporary storage polysaccharide (long-half-life pool) in faster growing populations. (a) Data points (from Halsey et al., 2011) and lines are for Dunaliella tertiolecta grown at 1.2 d−1 (green triangles and line), 0.5 d−1 (red squares and line) and 0.12 d−1 (blue circles and line). (b) Thalassiosira weissflogii grown at 1.0 d−1 (green triangles and line), 0.5 d−1 (red squares and line) and 0.20 d−1 (blue circles and line). Data points are averages of duplicate measurements from two independent experiments. Bars show the range of values for each point. For all growth rates of D. tertiolecta, the model was implemented using optimized half-life parameters of 26 s for C allocated to the ‘fast-burn’ pool and 1.9 h for C allocated to the ‘slow-burn’ polysaccharide pool. For all growth rates of T. weissflogii, the optimized half-life parameters of the short- and long-half-life pools were 34 s and 3.8 h, respectively.

This investigation aimed to identify the mechanisms underlying growth rate-dependent carbon metabolism in nutrient-limited, steady-state cultures grown under constant light. For this, we relied on information learned using a different species grown under different conditions (D. bioculata grown in a light–dark cycle and nutrient-replete conditions). Despite these differences, the strong correlation between the model and laboratory data suggests that changes in carbon metabolism across the cell cycle are key factors causing average lifetimes in newly fixed carbon to vary with nutrient-limited growth rate. Future studies using cultures synchronized under nutrient-limited growth conditions could reveal additional complexities involving metabolic shifts associated with cell cycle phases.

Generalities for photoautotrophy

Gross photosynthesis is related to net oxygen evolution and organic carbon production through the partitioning of photosynthate among diverse metabolic pathways. The results presented here indicate that this partitioning can be highly conserved between phytoplankton taxonomic groups and independent of the steady-state growth rate. Additional studies are required to generalize our findings to other phytoplankton groups and nutrient conditions. For example, cyanobacteria (Kunath et al., 2012) or highly motile picoeukaryotic species (Goulbourne & Greenberg, 1980; Raven & Richardson, 1984; Kaneda & Furuya, 1986) may have sufficiently different ATP : reductant requirements to significantly alter GPO2 : GPC : NPC ratios from those reported here. Furthermore, growth under high macronutrient levels and limiting iron supply appears to result in over-expressed photosynthetic pigments and lower absorption-normalized photosynthetic efficiencies relative to macronutrient-limited growth (Schrader et al., 2011). Nevertheless, the similarities in photosynthate partitioning and absorption-normalized photosynthetic efficiencies observed during our study suggest that a common optimization solution for photoautotrophic growth has evolved for at least two evolutionarily distinct phytoplankton species. Deeper evaluation of this energetic optimization is clearly warranted.

For both D. tertiolecta and T. weissflogii, we show that the measured 14C uptake rates register gross carbon fixation only during very short incubations (minute time scale) and then approach net carbon fixation with increasing incubation duration. Short-term (c. 30 min to 2 h) field measurements of photosynthesis–irradiance relationships are commonly used to parameterize global satellite-based models of ocean production. Our results indicate that the correct interpretation of these data requires consideration of the incubation duration and a knowledge of the population growth rates. Conversely, it may be possible to infer information about phytoplankton growth rates in the field by conducting 14C uptake measurements over a range of incubation times. Our results also indicate that growth rate-dependent differences in transient carbon lifetimes are linked to cell cycle-related changes in photosynthate storage and mobilization, thus contributing new insight towards a systems-level understanding of cellular carbon flow.


We thank Ihadira Lopez deSearch for assistance with experiments on T. weissflogii. This work was funded through a grant from the National Science Foundation, Biological Oceanography (NSF-OCE 1057244).