Coccolithophores are important calcifying phytoplankton predicted to be impacted by changes in ocean carbonate chemistry caused by the absorption of anthropogenic CO2. However, it is difficult to disentangle the effects of the simultaneously changing carbonate system parameters (CO2, bicarbonate, carbonate and protons) on the physiological responses to elevated CO2.
Here, we adopted a multifactorial approach at constant pH or CO2 whilst varying dissolved inorganic carbon (DIC) to determine physiological and transcriptional responses to individual carbonate system parameters.
We show that Emiliania huxleyi is sensitive to low CO2 (growth and photosynthesis) and low bicarbonate (calcification) as well as low pH beyond a limited tolerance range, but is much less sensitive to elevated CO2 and bicarbonate. Multiple up-regulated genes at low DIC bear the hallmarks of a carbon-concentrating mechanism (CCM) that is responsive to CO2 and bicarbonate but not to pH.
Emiliania huxleyi appears to have evolved mechanisms to respond to limiting rather than elevated CO2. Calcification does not function as a CCM, but is inhibited at low DIC to allow the redistribution of DIC from calcification to photosynthesis. The presented data provides a significant step in understanding how E. huxleyi will respond to changing carbonate chemistry at a cellular level.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Marine photoautotrophic organisms fix c. 55 gigatonnes of carbon yr–1 which is equal to the photosynthetic production by the terrestrial biosphere (Field et al., 1998). Coccolithophores play a major role in the global carbon cycle by contributing c. 1–10% to total organic carbon fixation (Poulton et al., 2007) and providing ballast through the formation of calcite, which enhances organic matter sinking into the deep ocean (Thierstein et al., 1977). The globally most abundant coccolithophore species is Emiliania huxleyi, which has the ability to form blooms up to 8 × 106 km2 (Moore et al., 2012). Despite the global significance of E. huxleyi, there is only a limited understanding of important cellular processes and their response to environmental change.
Under present-day conditions, marine phytoplankton growth is mostly limited by low light availability or by the insufficient supply of inorganic nutrients, such as nitrogen, phosphorus or iron (Sarmiento & Gruber, 2006), while carbon dioxide (CO2) is usually not considered to be limiting. Nevertheless, CO2 diffusion rates are in most cases not high enough to account for the photosynthetic rates seen in the majority of phytoplankton (Falkowski & Raven, 2007). This discrepancy is explained by the action of carbon (or CO2) concentrating mechanisms (CCMs). In algae these are predominantly C3 biophysical mechanisms which link carbonic anhydrases (CAs), dissolved inorganic carbon (DIC) transporters and pH gradients to enhance [CO2] at the active site of Ribulose-1,5-bisphosphate carboxylase oxygenase (RubisCO) (Reinfelder, 2011). It is thought that nearly all marine phytoplankton operate a CCM, although the DIC species used (CO2 and/or bicarbonate ()), its regulation, cellular components, and DIC affinity can vary significantly between species (Giordano et al., 2005). E. huxleyi operates a low-affinity CCM (Rost et al., 2003). Several studies indicate that CO2 is the primary source for photosynthesis, although there are some discrepancies over the importance of , especially at lower CO2 concentrations (Paasche, 1964; Sikes et al., 1980; Nimer & Merrett, 1992; Sekino & Shiraiwa, 1994; Herfort et al., 2002; Rost et al., 2003; Schulz et al., 2007; Bach et al., 2011). In addition to a biophysical mechanism, intracellular calcification has been proposed to act as a CCM by providing protons (H+) as a by-product of calcification to support the dehydration of to CO2 (reviewed in; Paasche, 2001). Although there are some supporting data (Nimer & Merrett, 1992; Buitenhuis et al., 1999), other studies contradict the concept (Paasche, 1964; Herfort et al., 2004; Trimborn et al., 2007; Leonardos et al., 2009).
In the forthcoming centuries, ongoing uptake of anthropogenic atmospheric CO2 into the oceans will continuously change the marine carbonate chemistry – a process known as ocean acidification (Caldeira & Wickett, 2003). Chemically, ocean acidification leads to a strong decrease of the carbonate ion () concentration, a slight increase in  and a strong increase in [CO2] and [H+] (Wolf-Gladrow et al., 1999). These components are thought to affect coccolithophores in varying ways, with  influencing calcite saturation concentrations, [H+] affecting cellular pH homeostasis, [CO2] affecting photosynthesis and  influencing calcification (and photosynthesis). The potential effects of ocean acidification on calcification and photosynthesis by E. huxleyi have been repeatedly reported (reviewed in Riebesell & Tortell, 2011), but the importance of changes in the individual carbonate parameters for the observed responses is still not fully understood.
The present study disentangles the carbonate system to improve our conceptual understanding of the acquisition of DIC and its subsequent use in calcification and photosynthesis. In particular, we address two important questions in E. huxleyi ecophysiology: how sensitive is E. huxleyi to low and elevated components of the carbonate system; and does calcification act as a CCM?
Materials and Methods
Conceptual background of the experiments
The marine carbonate system is defined by the concentrations of CO2, , , pCO2, total alkalinity (TA), DIC (i.e. combined CO2, and ), and pH ([H+]; Zeebe & Wolf-Gladrow, 2001). The physiologically relevant parameters of the carbonate system are CO2, , and H+, as only these can be perceived by a cell. They are connected to each other in the equilibrium reaction:
As no other parameters of physiological relevance other than CO2, , and H+ were changed in the experiments (e.g. light or temperature), it is assumed that only changing concentrations of these particular parameters can induce physiological or genetic responses. CO2, , and H+ are closely codependent (Eqn (Eqn 1)) and any change in the concentration of one will lead to changes in the others. Nevertheless, it is possible to keep one of the four parameters constant while changing the other three. We made use of this feature and performed three experiments where we kept either [CO2] or [H+] constant between treatments ([H+] was kept constant at two different concentrations). The constant carbonate system parameter within an experiment can be excluded from being responsible for the observed physiological or genetic response (Buitenhuis et al., 1999). Note that we chose to focus on CO2 and H+, as previous work points towards a primary importance of these particular parameters for E. huxleyi physiology (Schulz et al., 2007; Bach et al., 2011).
Experimental design and basic setup
Three experiments were conducted to test the physiological and molecular responses of Emiliania huxleyi (Lohmann) Hay and Mohler to changes in individual carbonate chemistry parameters. DIC was varied in all experiments, while either pHf (8.34 or 7.74 on free scale) or CO2 (16 μmol kg−1) was kept constant. In all experiments, cells of E. huxleyi (strain B92/11) were grown in monoclonal dilute batch cultures (LaRoche et al., 2010) at 15°C and 150 μmol m−2 s−1 incident photon flux density under a 16 : 8 h, light: dark cycle. The growth medium was artificial seawater prepared as described in Kester et al. (1967) but without the addition of NaHCO3, which was added in a later step (see the following section). Artificial seawater was enriched with c. 64 μmol kg−1 nitrate, 4 μmol kg−1 phosphate, f/8 concentrations of a trace metal and vitamin mixture (Guillard & Ryther, 1962), 10 nmol kg−1 of SeO2, and 2 ml kg−1 of natural North Sea water. Concentrations of nitrate and phosphate were measured according to Hansen & Koroleff (1999). The nutrient-enriched artificial seawater was sterile-filtered into polycarbonate bottles where the carbonate chemistry was manipulated. After taking samples for carbonate chemistry measurements (see the following section), the artificial seawater was divided carefully into three 2.3 l polycarbonate bottles before inoculation. Before inoculation, E. huxleyi cells were acclimated to exponential growth and carbonate chemistry conditions for at least seven generations. Approximate cell densities ranged from 50 to 300 cells ml−1 at inoculation and 40 000–100 000 cells ml−1 at sampling (see description of sampling later).
Carbonate chemistry manipulation and determination
In all experiments, target DIC concentrations were adjusted by adding calculated amounts of NaHCO3 or Na2CO3 (see Bach et al., 2012 for details). In the constant-CO2 experiment, CO2 was set to a constant concentration of c. 16 (± 2) μmol kg−1 through additions of calculated amounts of HCl (3.571 molar). In the constant-pH experiments, pH was adjusted to 7.74 (± 0.004) or 8.34 (± 0.008) by adding 2 mmol kg−1 of 2-[-4-(2-hydroxyethyl)-1-piperazinyl]-ethanesulfonic acid (HEPES, adjusted to target pHf levels).
Carbonate chemistry in the constant-CO2 experiment was determined by measuring TA and pHf, while in both constant-pH experiments it was determined from pHf and DIC. Carbonate chemistry samples were taken at the beginning and the end of the experiments. Samples for TA were filtered (0.7 μm), poisoned with saturated HgCl2 solution (0.5‰ final concentration) and stored at 4°C until measured (Dickson et al., 2003). TA values higher than 4700 μmol kg−1 were outside the range that can be accurately determined with the applied method and therefore diluted with double deionized water as described in Bach et al. (2012).
Samples for DIC were sterile-filtered (0.2 μm) by gentle pressure into 4 ml borosilicate bottles, made air-tight without headspace and subsequently measured as described in Stoll et al. (2001). DIC samples lower than 1000 or higher than 3000 μmol kg−1 could not be reliably measured with the applied method and were therefore either diluted or concentrated (see Bach et al., 2011, 2012).
Samples for pHf were measured potentiometrically at 15°C with separate glass and reference electrodes (METROHM) calibrated with reference seawater, certified for TA and DIC (supplied by Prof. A. Dickson, La Jolla, CA, USA; see Bach et al., 2011, 2012 for details).
Carbonate chemistry parameters that were not directly measured were calculated from two measured values (DIC and TA or DIC and pHf) and known salinity, temperature, and phosphate concentrations with the software CO2SYS (Lewis & Wallace, 1998) using equilibrium constants determined by Roy et al. (1993). Biological response data are plotted against the means of the initial and final values of the carbonate chemistry. Error bars in plotted carbonate chemistry parameters denote the mean change of the three replicates of the particular carbonate chemistry parameter from the beginning of the experiment to the end.
Sampling, measurements and calculations of growth, organic, and inorganic carbon production rates
Sampling started 2 h after the onset of the light period and lasted not longer than 2.5 h. Duplicate samples for total particulate carbon (TPC) and particulate organic carbon (POC) were filtered (200 mbar) on to precombusted (5 h at 500°C) GF/F filters. To remove HEPES from the filters of the constant-pH experiments, samples were rinsed with 60 ml of artificial seawater medium supersaturated with respect to calcium carbonate and free of HEPES buffer immediately after filtration. Filters were stored at −20°C until measurements were carried out. POC filters were placed for 2 h in a desiccator containing fuming HCl to remove all calcite and then dried for c. 6 h at 60°C. TPC filters were dried under the same conditions but without the acid treatment. TPC and POC analyses were performed using an elemental analyzer (HEKATECH, Wegberg, Germany) combined with an isotope ratio mass spectrometer (FINNIGAN, Schwerte, Germany). Particulate inorganic carbon (PIC) was calculated as the difference between TPC and POC.
Cell numbers were determined with a Coulter Counter (Beckman Coulter, Krefeld, Germany) at the beginning and the end of the experiments c. 4 h after the onset of the light period. Growth rates (μ) were calculated as
where t0 and tfin are the cell numbers at the beginning and the end of the experiments, respectively, and d is the growth period in days. POC and PIC production rates were calculated by multiplying growth rates with the cellular POC or PIC contents.
Treatments were further analyzed by scanning electron microscopy (SEM) and cross-polarized light microscopy to confirm the presence or absence of internal coccoliths (Bach et al., 2012). Cells were considered to be actively calcifying if coccoliths were present.
For gene expression analysis, c. 10 million cells were filtered (200 mbar) onto polycarbonate filters with a pore size of 0.8 μm and subsequently rinsed off the filters with 1 ml RNAlater (Qiagen). This cell suspension was kept on ice until storage at −20°C.
Quantitative reverse-transcriptase polymerase chain reaction was performed for 15 target genes (Table 1). Each sample was measured in triplicate. Experimental procedures were performed as described previously (Mackinder et al., 2011). Primers were designed using expressed sequence tag (EST) clusters from von Dassow et al. (2009), the E. huxleyi Genome Project (http://genome.jgi-psf.org/Emihu1/Emihu1.home.html) or from the current literature (Supporting Information, Table S1). Efficiency curves for each primer pair were generated using serial dilutions on pooled cDNA from all samples. All primers except beta-carbonic anhydrase (βCA) had efficiencies between 90 and 105% and generated curves with R2 values > 0.99. βCA efficiency remained undetermined as a result of the low cycle threshold (CT) values of pooled cDNA even at undiluted levels. For relative expression calculations, its efficiency was assumed to be 100%. This assumption results in a potential decrease in the accuracy of the absolute fold changes, but the trend of expression and the order of magnitude will remain unaffected. For each sample, 2–20 ng of RT RNA were analyzed in technical triplicates. For each primer pair, all samples were analyzed across three plates, and in order to allow for the correction of between-plate variation two standards in triplicate were run on each plate. GeNorm (Vandesompele et al., 2002) was used to test the stability of four potential endogenous reference genes (ERGs).
Table 1. Emiliania huxleyi genetic response to carbonate system manipulations
Correlation to carbonate system parameter
Function and location
CO2 (μmol kg−1)
Possible location – experimentally or by analogy
Predicted location by WoLF PSORT (Horton et al., 2007)
Arrows indicate a significant up-regulation (P <0.05) of the investigated genes at low CO2 (< 10 μmol kg−1), (< 1000 μmol kg−1), and (< 50 μmol kg−1), compared with the lowest measured expression at high dissolved inorganic carbon (DIC) (> c. 1000 μmol kg−1). Numbers behind arrows denote the maximum fold change that has been observed. Shading indicates a consistent up-regulation with decreasing levels of the respective carbonate system parameter (see Fig. 4).
No statistical evaluation was possible because some replicates were below the detection limit.
Indicates partially characterized in coccolithophores.
Anion exchanger like 1
transport, probably coupled to Na+, Cl− and/or H+ transport
Plasma membrane or chloroplast. A diatom homolog has been shown to be plasma membrane-located (Nakajima et al., 2013)
α carbonic anhydrase 1
+ H+ ↔ CO2 + H2O
αCAs are distributed throughout all kingdoms of life. In Chlamydomonas they are found in the periplasmic space and the thylakoid lumen (Spalding, 2008). In the diatoms Phaeodactylum tricornutum and Thalassiosira pseudonana, putative αCAs were located in the four layered plastid membrane system and inside the chloroplast, respectively (Tachibana et al., 2011)
Analysis of qRT-PCR data was done using an efficiency-corrected ∆∆Ct method, normalizing to the geometric mean of three ERGs (Vandesompele et al., 2002). For each gene, all samples were plotted relative to the sample with lowest expression from all three experiments. The sample with the lowest expression level was normalized to 1, allowing the expression ratios between samples to be easily identified.
We tested if the carbonate chemistry had a statistically significant effect (P <0.05) on individual physiological and molecular response parameters with either a one-factorial analysis of variance (ANOVA) using Statistica (Statsoft, Hamburg, Germany) in case the data subsets were normally distributed, or with a permutational multivariate analysis of variance (PERMANOVA) using Primer 6 in case they were not. Normality was tested with Shapiro–Wilk's test (P = 0.05). Nonnormally distributed subsets were Box–Cox-transformed. Subsets that remained nonnormally distributed were analyzed with the PERMANOVA.
ANOVA: The difference of individual treatments within an experiment was tested with Tukey's HSD post-hoc test (P values from post-hoc tests are denoted by Ppost hoc). Homogeneity of variance was tested using Levene's test and was accepted if the P-value was > 0.05. Where P was smaller, the significance level (P-value of the ANOVA and the post-hoc test) was decreased to 0.01. Subsets treated this way are marked in Table S2.
PERMANOVA: A resemblance matrix was created using the Euclidian distance function and further processed with a one-factorial PERMANOVA design choosing type III partitioning of the sum of squares. In cases where statistically significant differences were detected, a pairwise comparison of treatments (analog to a post-hoc test) was conducted in a second PERMANOVA run. The numbers of permutations for each run are given in Table S2. In pairwise PERMANOVA runs, these numbers were not sufficiently high (< 100) to get reasonable results for P, so that an additional Monte Carlo test was conducted. Significance levels of the PERMANOVA analysis are the same as for the ANOVA, but by convention are termed P(perm) for the permutation P-value and Ppost hoc(MC) for the Monte Carlo P-value to distinguish them.
Growth and POC production rates are sensitive to low CO2 (and ) and to low pH, but not to elevated CO2
To determine the importance of individual components of the carbonate system for E. huxleyi physiology, cells were grown in three separate experiments at constant pHf (7.74 and 8.34) and constant CO2 (16 μmol kg−1). Fig. 1 shows how the carbonate system changed within the three experiments. By maintaining relatively low cell concentrations, changes in carbonate chemistry as a result of biological processes were kept to a minimum over the time of the experiments. This is indicated by the error bars in Fig. 1 with the corresponding values in Table S3.
Within the ranges examined, growth and POC production rates were primarily influenced by changes in carbonate chemistry from low to intermediate (160–2000 μmol kg−1) and CO2 (0.8–20 μmol kg−1) (Fig. 2) with neither pH nor having a pronounced influence (Fig. S1). Growth rates increased in all experiments with increasing concentrations of and CO2 until reaching maximum rates of c. 1.1 d−1 where further CO2 or increases had no effect on growth rates. The constant-pH experiments allow us to differentiate between the effects of CO2 and on growth rate. CO2 demonstrates a good correlation with growth rate in both constant-pH experiments, whereas the influence of on growth rate is more variable (Fig. 2a,b), suggesting that CO2 is the principal factor responsible for growth inhibition below a [CO2] of c. 7.5 μmol kg−1 (Fig. 2b). No effect of pH on growth rate was observed in the constant-pH treatments (7.74 and 8.34). However, at constant CO2, growth rates are significantly lower at pH 7.58 than at pH 7.83 (Ppost-hoc = 0.009), which cannot be explained by a decrease in [CO2] or  (Fig. 2a,b). Thus, below pHf 7.74, [H+] appears to have a direct negative influence on growth rate.
Particulate organic carbon production rates in both constant-pH experiments were highly sensitive to and CO2 when the concentrations dropped below c. 2000 and 10 μmol kg−1, respectively (Fig. 2c,d). The rates appear to correlate best to CO2 at concentrations < c. 5 μmol kg−1, although there are limited data points in this range. At a constant CO2, the lowest treatment also showed significantly lower POC production rates than at intermediate (Ppost hoc = 0.002, Fig. 2c). At concentrations > c. 2000 μmol kg−1 POC production rates display a slight but significant decrease of c. 20% at a constant pHf of 8.34 and 10% at a constant pHf of 7.74 up to the highest concentrations (Fig. 2c; pH 8.34, Ppost hoc < 0.001; pH 7.74, Ppost hoc (MC) = 0.004). In summary, POC production showed no clear overall correlation with any of the carbonate chemistry parameters, but appears to be driven by CO2 in the very low CO2 range (< c. 5 μmol kg−1) and decreased by at concentrations > 2000 μmol kg−1.
At low DIC, C : N ratios decreased significantly in the constant-pH experiments, which appear to be driven primarily by a reduction in CO2 (Table S2). This is supported by no significant changes at constant CO2 (Table S2). Differences in C : N between treatments probably reflect variable cellular amounts of nitrogen-free relative to nitrogen-rich organic compounds. As 40–60% of the total cellular carbon in E. huxleyi is in the form of lipids (Fernandez et al., 1994), the decrease in C : N is likely to reflect reduced assimilation of lipids and polysaccharides at low DIC.
Calcification is primarily driven by and does not act as a CCM
Calcification rates (PIC production) increased similarly in all experiments with increasing  (Fig. 2g). Maximum calcification rates at constant pHf values of 8.34 and 7.74 were identical, but were reached at lower CO2 and higher at a constant pHf of 8.34, indicating that calcification is not primarily dependent on [CO2] or  (Figs 2h, S1h). A limited control of calcification by [CO2] is further supported by the decrease in calcification rates found in the constant-CO2 experiment. Here, calcification rates would have to remain constant if [CO2] were of primary importance. No signs of calcification could be found in the two lowest treatments at a constant pHf of 7.74 and in two replicates of the lowest treatment at constant CO2 (Table 2). In these treatments, calcite saturation (Ωcalcite) is < 0.31, so post-production dissolution could potentially have taken place. However, cross-polarized light microscopy and scanning electron microscopy show the absence of internal coccoliths under these conditions, indicating that the production of coccoliths is inhibited (Bach et al., 2012).
Table 2. Presence or absence of Emiliania huxleyi coccoliths from SEM investigations
H+ (nmol kg−1)
Table showing SEM analysis of individual replicates of treatments where particulate inorganic carbon (PIC) production was < 0.5 pg per cell d−1. Note that coccoliths were found in all treatments and replicates not listed in this table.
Cell concentrations were higher in this replicate at the end of the experiment (76 000 compared with 36 000 cell ml−1 in first replicate). This caused a stronger decrease in  and [H+].
Up-regulation of carbonic anhydrases (CAs) at low DIC
Three out of five investigated CAs showed an up-regulation in expression at low DIC
Location of δCA at plasma membrane
Presence of a putative membrane anchor; localization of a dinoflagellate δCA to the plasma membrane; strong up-regulation at low CO2 has also been demonstrated in TWCA1, a δCA from Thalassiosira weissflogii; up-regulation at low DIC
Soto et al. (2006); Lapointe et al. (2008); McGinn & Morel (2008); this study
Up-regulation of extracellular CA at low DIC
Increased extracellular CA activity at low DIC/high pH; up-regulation of δCA at low DIC
Two βCAs from diatoms have been shown to localize to the chloroplast – specifically the pyrenoid; CA activity in the stroma chloroplast fraction of the coccolithophore Pleurochrysis sp.
Kitao et al. (2008) and Tachibana et al. (2011); Quiroga & González (1993)
Probable absence of cytosolic CA
Cytosolic acidification at high – presence of CA would result in rapid buffering; expression of a human CA in the cytoplasm of cyanobacteria resulted in a high CO2-requiring phenotype; cytosolic CA would increase cytosolic CO2, leading to increased leakage at low external CO2
Suffrian et al. (2011); Price & Badger (1989); this study
Switching off of calcification at low DIC to increase DIC availability for photosynthesis
The decrease in calcification before a reduction in particulate organic carbon (POC) and growth rates. The complete termination of calcification at low DIC and pHf
is the principal substrate for calcification
Previous 14C labeling studies; strong correlation of calcification with concentration
Sikes et al. (1980) and Nimer et al. (1997); Buitenhuis et al. (1999) and this study
The use of pH gradients within the CCM
Up-regulation of putative H+ transporters at low DIC
Up-regulation of RubisCO to compensate for its decrease in efficiency as a result of an increased oxygenase : carboxylase ratio at low CO2
Low DIC therefore results in a decrease in growth rate and POC production as well as in calcification (Figs 2, S2). However, PIC production appears to be the most sensitive to low DIC, with low calcification rates observed in several low-DIC conditions where there was no appreciable effect on POC production and growth rate (Fig. S2). This indicates that POC production is prioritized over PIC production under Ci limitation (Fig. S2), and suggests that reducing calcification rate may enable cellular resources (such as those relating to uptake) to be used for photosynthesis. Calcification is clearly not operating as a CCM at low DIC, as in this case we would expect a stimulation of calcification at low DIC.
At a genetic level, the CCM is up-regulated only at low CO2 and is not induced at current ocean CO2 concentrations
In order to identify the molecular basis of the physiological response of E. huxleyi to the individual carbonate system parameters, 15 genes with putative roles in carbon transport, pH homeostasis and biomineralization were chosen for investigation (Table S1). The measurement of relative transcript abundance was chosen as the most suitable approach to allow the expression profiles of multiple genes to be accurately determined. Although transcript abundance is not a direct measurement of protein abundance or activity, it gives a good insight into the cellular demand for specific proteins and provides a strong foundation for the further characterization of genes related to a particular cellular process. All genes are normalized to three endogenous reference genes (ERGs; ACTIN, α-TUBULIN and EFG1-α) with expression plotted relative to the lowest expression level, which is set to one.
Plotting gene expression against DIC indicates the transcriptional response to changes in total DIC (Fig. 3, Table S3). Out of the 15 genes investigated, 11 showed a marked increase in expression when the cells became DIC-limited (DIC < 1000 μmol kg−1) but showed no repression above this concentration. This corresponds to [CO2] and  thresholds of c. 7.5 and 800 μmol kg−1, respectively, below which CCM gene up-regulation occurs (Fig. 4a,b). Both of these values are approximately half that of average current oceanic values (i.e. similar to pre-industrial values), suggesting that the E. huxleyi CCM, at least in this strain, is actually only induced at DIC concentrations lower than ambient.
Of the selected genes with putative roles in DIC transport, AEL1 (anion exchanger like 1, belonging to the solute carrier 4 (SLC4) family), αCA1 (alpha-carbonic anhydrase 1), δCA (delta-carbonic anhydrase 1) and rubisco (RubisCO large subunit) showed a significant DIC limited up-regulation between four and 11-fold (Table 1, Fig. 3a). Two genes, βCA and LCIX (low CO2 induced gene X), had a large response at low DIC, with a respective 450- and 180-fold up-regulation at the lowest DIC value in the constant-pHf (= 8.34) experiment relative to the treatment with the lowest expression (Table 1, Fig. 3e,f). βCA encodes a putative carbonic anhydrase responsible for catalyzing the interconversion of CO2 and , whereas LCIX exhibits similarity to the Chlamydomonas LCIB protein, which is located in the chloroplast and plays a crucial role in uptake (Miura et al., 2004; Wang & Spalding, 2006). Furthermore, βCA showed a highly correlated expression with LCIX (R2>0.99, data not shown), indicating that these genes could be under the same transcriptional control.
Of the putative H+ transport-related genes, CAX3 (Ca2+/H+ exchanger 3), NhaA2 (Na+/H+ exchanger 2), ATPVc′/c (vaculoar-type H+ pump) and PATP (plasma membrane-type H+ pump) showed a 4–7.5 fold up-regulation (Table 1, Fig. 3b). Four genes with potential roles in H+ and DIC transport, HVCN1 (H+ channel), AQP2 (aquaporin 2), αCA2 (alpha-carbonic anhydrase 2), and γCA (gamma-carbonic anhydrase), showed no significant transcriptional response over the carbonate system range tested (Fig. 3c; Table S2). Above 1000 μmol kg−1, DIC changes in gene expression of most investigated genes was minimal with no repression of DIC-responsive genes, but a small but significant decrease (Ppost hoc = 0.02) seen in GPA expression > c. 2000 μmol kg−1 (Fig. 3d).
The CCM is responsive to CO2 and but not to pH
An understanding of the regulation of the E. huxleyi CCM may provide important information about its mode of operation and cellular function. An examination of the individual carbonate system parameters indicated that the expression of these genes correlates closely with [CO2] and  at low DIC (Fig. 4a,b). This indicates that although pH and may have a synergistic effect with other factors on the expression of some genes, they do not appear to be the main parameters of the carbonate system driving the genetic responses (Fig. 4c,d). Table 1 summarizes the responses of the investigated genes along with their putative or confirmed function and potential cellular locations.
Transcriptional response to reduced calcification
Previously we demonstrated that the expression of several genes with putative roles in DIC, Ca2+ and H+ transport (AEL1, CAX3 and ATPVc′/c) show a close correlation with calcification rate, suggesting that these genes play a direct role in the calcification process (Mackinder et al., 2011). When calcification was inhibited by the removal of Ca2+, the expression of these calcification-associated genes was strongly repressed (Mackinder et al., 2011). However, in the present study, these genes were all induced at low DIC (Fig. 3), whereas calcification was inhibited. This indicates that these genes may play a dual role within the cell, supporting calcification under ambient conditons but switching to support photosynthesis when DIC becomes limiting.
Growth and calcification responses to the carbonate system
The predicted changes in the ocean's carbonate system caused by increasing atmospheric CO2 may have multiple impacts on coccolithophore physiology (Riebesell & Tortell, 2011). Using experimental manipulation of the carbonate system, we show that individual aspects of E. huxleyi physiology can be attributed to separate components of the carbonate system.
Growth rates presented in this study correlate closely to [CO2] (Fig. 2a), with pHf having a significant negative impact below values of c. 7.7 (Fig. S1a). Although POC production does not show such a clear coupling to [CO2] as growth rates (Fig. 2d), it also responds negatively to pHf when it drops below c. 7.7. A similar regulation of pH and CO2 on growth and POC production was also seen in Bach et al. (2011) with a linear decrease from a pHf of c. 7.7–7.0 and CO2 dependence above a pHf of 7.7. However, a study by Buitenhuis et al. (1999) saw no clear tightly coupled correlation between E. huxleyi growth rate and [CO2]. Instead, the authors suggested that both CO2 and are important for growth rates. The reason behind this discrepancy is unclear, although it should be kept in mind that threshold values for individual carbonate system components may differ between strains and may be modulated by light conditions (Langer et al., 2009; Rokitta & Rost, 2012).
Calcification rates are tightly coupled to  (Fig. 2g), suggesting that is the primary carbon source used for CaCO3 precipitation in E. huxleyi. This is in agreement with previous studies (reviewed in Paasche, 2001). Simulated ocean acidification has been shown to affect coccolithophore calcification mostly negatively (Riebesell & Tortell, 2011). By comparing ocean acidification with constant-pH experiments, Bach et al. (2011) showed that it is the increase in H+ at elevated CO2 that negatively affects calcification rates of E. huxleyi. It is also known that intracellular pH in coccolithophores is particularly sensitive to changes in external pH (Suffrian et al., 2011; Taylor et al., 2011). Under these considerations, it could be expected that calcification rates would remain consistently lower throughout the constant pHf = 7.74 experiment compared with the constant pHf = 8.34 experiment. Surprisingly, however, this is not the case. Instead, maximum calcification rates are similar in both constant-pH experiments (Fig. 2g,h). This indicates that the direct negative effect of high [H+] on calcification rates may at some point be overcome by increasing availability of substrate. This is further supported by our finding that higher  was necessary to initiate calcification when [H+] in the seawater medium was higher (Table 2). Considering carbonate chemistry conditions of the past, this might provide a further explanation as to why coccolithophores were able to thrive in the early Mesozoic era, a time that was characterized by relatively low sea water pH (as low as pH 7.7) and high DIC substrate (up to 5000 μmol kg−1; Ridgwell, 2005).
The nature and regulation of the CCM
Previous mass spectrometrically based work by Rost et al. (2003) showed that E. huxleyi operates a regulated CCM but gave no indication of the mechanism. Our results support the presence of a regulated CCM and furthermore have identified several of its molecular components, the carbonate species to which it responds, the threshold at which it is induced, and its possible interactions with calcification.
The transcriptional data identify the genetic basis of a CCM in E. huxleyi with a clear up-regulation in multiple putative CCM-related genes as DIC becomes limiting for growth, POC and PIC production (Fig. 3, Table 1). The majority of genes were up-regulated when or CO2 dropped below c. 800 and 7.5 μmol kg−1, respectively. Interestingly, most of the DIC-responsive genes were not further repressed at CO2 > c. 7.5 μmol kg−1 ( c. 800 μmol kg−1); this indicates a potential basal level of the CCM, with a low amount of active DIC transport taking place even when growth rates and POC production are saturated. The presence of active transport at ambient CO2 and is supported by Schulz et al. (2007), who showed active DIC uptake even at ambient conditions.
Photosynthetic O2 evolution curves and 14C incorporation studies have indicated that photosynthesis is not saturated at ambient CO2 (Paasche, 1964; Herfort et al., 2002; Rost et al., 2003). This is not supported by our data with growth rates and organic carbon fixation both saturated at or below ambient [CO2]. However, these differences could theoretically be attributed to the different light intensities used between the studies and to the fact that O2 evolution is a measurement of photosystem II activity, not a direct measurement of CO2 fixation. Furthermore, these thresholds may vary between strains, as seen with strain-specific responses in calcification and growth to changing carbonate chemistry (Langer et al., 2009). These responses do not necessarily indicate that the underlying cellular mechanisms differ between strains, but most likely highlight differences in the regulation of cellular processes, such as calcification. This is further supported by an optimum curve response, with different strains and species having varying optimum calcification rates in relation to pCO2, but the overall response (i.e. the shape of the curve) being very similar (Langer et al., 2006, 2009; Ridgwell et al., 2009; Bach et al., 2011; Krug et al., 2011). However, a greater understanding at the molecular level of the response of different E. huxleyi strains and coccolithophore species to changes in carbonate chemistry is critical to extrapolate our data to other coccolithophores.
The CCM of E. huxleyi shows a number of differences from those of other partially characterized eukaryotic algae. One outstanding feature is its low affinity for CO2 (Rost et al., 2003) with a K1/2 for CO2 that is several-fold higher than the K1/2 for the prymnesiophyte Phaeocystis globosa and several diatom species (Johnston & Raven, 1996; Rost et al., 2003; Trimborn et al., 2009). Another feature of the E. huxleyi CCM is that up-regulation of molecular components seems to occur only when very low CO2 concentrations are reached. This is strikingly different from diatoms and Chlamydomonas, where molecular CCM components are already strongly induced at ambient CO2 and even above (Harada et al., 2005; Brueggeman et al., 2012).
Although the E. huxleyi CCM may be of a lower affinity, the basic components appear to be similar to other eukaryotic algae. CAs play fundamental roles within algal CCMs, and CAs associated with the CCM are generally up-regulated under carbon limitation (Badger, 2003; Raven & Giordano, 2009). Genome analysis shows that E. huxleyi has nine putative CAs belonging to the α, β, γ and δ families. This CA composition demonstrates strong similarities with Chlamydomonas, which has 10 putative CAs in its genome belonging to the α, β and γ families (Spalding, 2008). It is also very similar to the diatom CA repertoire, with Phaeodactylum tricornutum also having nine CAs distributed across the same four families (Tachibana et al., 2011). Diatoms also possess multiple homologs to AEL1. The characterization of P. tricornutum SLC4-2 shows that it is induced at low CO2, localizes to the plasma membrane and stimulates uptake and photosynthesis (Nakajima et al., 2013). Wolf PSORT predicts a plasma membrane location for AEL1 (Table 1) and its low /CO2-dependent expression suggests a related function in E. huxleyi.
Localized intracellular pH gradients and regulation are thought to be a fundamental part of CCMs (Raven, 1997). The increased expression of putative proton pumps (ATPVc′/c and PATP) and cation/H+ exchangers (NhaA2 and CAX3) suggests an increased demand of these transporters to maintain pH homeostasis, membrane potential or alter compartmental pH in order to promote changes in CO2 : ratios. More alkaline regions would maintain DIC as , which is one million times less permeable to membranes than CO2 (Moroney et al., 2011). This could prevent CO2 loss via diffusion across membranes, while more acidic regions in the proximity of RubisCO would result in a shift to CO2 (Raven, 1997).
Although use appears to become increasingly important at low DIC (Rost et al., 2003; Schulz et al., 2007; AEL1 up-regulation at low DIC shown here), growth rates are ultimately determined by CO2 (Fig. 2b). By operating a CCM, the cell actively accumulates and CO2 at a higher concentration in the proximity of RubisCO than externally. DIC has to be presented to RubisCO as CO2, so ultimately accumulated for carbon fixation will have to be converted to CO2. If the external CO2 concentration is very low, the diffusion gradient from the chloroplast to the outside will be large and leakage increases (Rost et al., 2006). Leakage in E. huxleyi has been measured to be c. 79% at ambient CO2 (Schulz et al., 2007) and shown to increase as CO2 decreased (Rost et al., 2006). Thus, external [CO2] largely determines how much accumulated DIC stays within the cell as a result of the strong inside-to-out CO2 gradient and high permeability of membranes to CO2.
Calcification as a CCM
Coccolithophores have maintained calcification since coccoliths appeared in the fossil record c. 220 million yr ago (Bown et al., 2004). A proposed role for the maintenance of calcification in coccolithophores is to support photosynthesis by using H+ generated by the production of calcium carbonate from bicarbonate (Paasche, 2001). Whilst carbon fixation by photosynthesis and calcification can occur at a similar rate within a cell, there is increasing evidence suggesting that the two processes are not tightly linked. It is possible to inhibit calcification by limiting calcium (Herfort et al., 2004; Trimborn et al., 2007; Leonardos et al., 2009) or DIC (Buitenhuis et al., 1999; this study), whilst photosynthesis, growth and POC production rates remain unaffected (Trimborn et al., 2007; constant-CO2 experiment of this study). Photosynthesis therefore appears to have no mechanistic dependence on calcification (Leonardos et al., 2009). Our data support this and strongly suggest that calcification does not function as a CCM at low DIC.
Moreover, our data reveal that calcification is actually inhibited at low DIC, rather than induced. Current evidence indicates that coccolithophores largely use CO2 for photosynthesis and for calcification (reviewed in Paasche, 2001), which is supported by our own observations. Thus, inhibition of calcification would enable the cell to utilize the normally acquired for calcification as a substrate for photosynthesis. Here we provide the first transcriptional dataset in support of this hypothesis. We found that the expression of three putative calcification-related ion transporters was elevated under limiting DIC, whilst calcification was inhibited. For example, assuming AEL1 functions as a plasma membrane transporter in E. huxleyi, as with SLC4-2 in diatoms, under normal conditions it most probably acts to transport into an intracellular pool for calcification (Fig. 5a). This is supported by AEL1 expression being repressed when calcification is inhibited by calcium limitation or in noncalcifying strains (Mackinder et al., 2011). However, under low CO2 and availability, AEL1 is induced, whereas calcification is inhibited. This suggests that there is an increased need for transport at low DIC, but that this is diverted away from the coccolith vesicle into the chloroplast for photosynthetic carbon fixation (Fig. 5b). Further functional characterization and localization of AEL1 and other CCM/calcification components is critical to validate this model and to fully understand this process at the molecular level.
Extrapolation to the real ocean
The expression data indicate an up-regulation of the CCM occurring at low DIC ([CO2] c. 7.5 μmol kg−1), suggesting that an inducible CCM is redundant in this E. huxleyi strain under current oceanic [CO2] (c. 16 μmol kg−1). However, in their natural habitat, it is possible that cells sporadically experience [CO2] < 7.5 μmol kg−1, in particular at the end of a bloom where [CO2] is reduced as a result of photosynthetic carbon draw-down. Values as low as c. 5 μmol kg−1 were seen in a mesocosm experiment where an E. huxleyi bloom occurred after a Phaeocystis sp. and diatom bloom (Purdie & Finch, 1994). Furthermore, [CO2] was significantly lower before the onset of anthropogenic CO2 release c. 200 yr ago, so that limiting DIC concentrations might have occurred more frequently in the past. A third aspect, which has to be considered, is a possible variability in the threshold DIC concentration below which the CCM is up-regulated. Variable thresholds either could result from strain-specific differences between E. huxleyi clones (Langer et al., 2009) and/or could be altered by culture conditions (Rokitta & Rost, 2012). At very high light conditions, for example, it is possible that the CCM becomes up-regulated at a higher CO2 threshold, owing to the cell having a larger DIC demand. Finally, the necessity of an inducible CCM in E. huxleyi can only be reliably determined by in field experiments where regulation patterns are investigated in in situ conditions.
Increased pCO2 has been shown to affect intracellular processes like calcification and photosynthesis in coccolithophores (Riebesell et al., 2000; Langer et al., 2006, 2009). In contrast to these physiological responses, our data suggest that the regulatory response to these changes at a genetic level is very limited. CO2 and only enhanced transcription of genes at concentrations significantly below those currently experienced and well below concentrations predicted in the near future. Furthermore, none of the investigated genes – even putative H+ pumps – were responsive to increasing sea water [H+]. There are two possible explanations for this lack of regulatory response: we have simply missed the critical pH and high CO2 responsive genes; or E. huxleyi does indeed entirely lack a regulatory machinery to cope with ocean acidification. The former can only be addressed in similar future studies that investigate the whole transcriptome. However, if future studies support the latter then the inability to regulate to changing pH could offer an explanation as to why calcification and photosynthesis are negatively affected below certain pH thresholds.
The novel approach applied in this study has allowed us to tease out the complexities of, and interactions between, photosynthesis and calcification in the ecologically important phytoplankton, E. huxleyi, and their responses to changing pCO2. The data presented provide a significant step forward in understanding the underlying cellular and molecular mechanisms of these processes, providing strong evidence that calcification does not function as a CCM and indicating that E. huxleyi may have evolved mechanisms to deal with limiting rather than elevated pCO2.
Silke Lischka is acknowledged for support on statistics and Janett Voigt for support during sampling. Furthermore, we thank three anonymous reviewers for their valuable comments, which helped to improve the manuscript. The work was funded by CalMarO a FP7 Marie Curie Initial training network and the Federal Ministry of Education and Research (Bundesministerium für Bildung und Forschung; 03F0608A) in the framework of the Biological Impacts of Ocean Acidification (BIOACID) project (subproject 3.1.1).