Introduction

Diatoms are responsible for 40% of marine carbon fixation (Field et al. 1998) and are found across all latitudes of the ocean. Increasing anthropogenic inputs of CO2 into the atmosphere is driving both ocean warming and an increase in the ocean’s absorption of anthropogenic CO2 leading to decreasing oceanic pH (ocean acidification) and changes in oceanic carbon chemistry with increasing concentrations of dissolved CO2 and HCO3− (Zeebe 2012; Cheng et al. 2022). While different marine phytoplankton species have differing responses to the elevation of CO2 (Dutkiewicz et al. 2015), diatoms’ ability to fix carbon appears largely insensitive to ocean acidification due to their highly effective carbon concentrating mechanisms (CCMs) that can utilize the much larger pool of HCO3− available in seawater. Most diatoms studied to date possess a biophysical CCM that uses a compartment (pyrenoid) capable of elevating CO2 around the CO2 fixing enzyme, RuBisCO (Roberts et al. 2007; Hopkinson et al. 2016). Some diatoms may also possess a biochemical CCM (e.g., a C4 pathway) but this is still open for debate (Reinfelder et al. 2000; Morel et al. 2002; Roberts et al. 2007; Clement et al. 2016) and is not the focus of this study. While pyrenoids are common in diatoms, the mechanism by which the CCM elevates CO2 around RuBisCO varies among species. Broadly, there appears to be two types of CCM, (1) those where inorganic carbon (Ci) uptake is dependent on an extracellular carbonic anhydrase (eCA) and (2) those lacking an eCA and instead rely on HCO3− transporters on the cell membrane for Ci uptake (Tsuji et al. 2017a).

While algal CCMs are highly effective, they are also energy intensive (Raven 2010; Raven et al. 2014). CCMs elevate pyrenoid CO2 concentrations (needed to nearly saturate RuBisCO) to levels that far exceed CO2 concentrations in seawater, creating large concentration gradients, which encourages diffusive CO2 efflux across highly permeable membranes. Manipulation of internal CO2 pools and active transport of HCO3− across membranes requires energy input. Numerous studies have shown that increased extracellular CO2 or changes in extracellular or intracellular pH can reduce the energetic requirement of the CCM (Hopkinson 2014; Mangan et al. 2016), however, there has been mixed evidence in diatoms on whether this can result in enhanced carbon fixation rates (Gao et al. 2014). Optimal CCM function requires synergistically balancing different CCM components, which facilitate HCO3− and CO2 influx into the cell, recycling of leaked CO2 and the maintenance of high CO2 concentrations inside the pyrenoid.

Temperature plays a large but relatively understudied role on the CCM. As temperatures increase, the solubility of CO2 and O2 in seawater is reduced, along with a decrease in RuBisCO substrate affinity and CO2/O2 specificity (Galmés et al. 2016). However, membrane permeability, rates of diffusion and biochemical reaction rates all increase with rising temperatures, until an optimum is reached (Beardall et al. 2009; Li and Young 2022). It is likely that the temperature sensitivity of these different CCM components vary, leading to imbalanced Ci fluxes with fluctuating temperature. Therefore, diatoms need to regulate their CCM components to maintain balanced Ci fluxes to support high photosynthetic efficiency. There has been extensive research on the response of C4 plants (biochemical CCM) to varying temperature (Sage and Kubien 2007) but little is known about CCM response to temperature in the marine environment. While temperature fluctuations may not be as large in marine compared to terrestrial environments, marine heatwaves can be 5–6 °C above average temperatures and are devastating for marine ecosystems (Frölicher and Laufkötter 2018). It is likely these marine heatwaves are becoming more common, particularly in coastal regions (Lima and Wethey 2012; Hughes et al. 2018).

Phaeodactylum tricornutum (Pt) is a model coastal marine diatom with a well-established CCM model at 20 °C (Hopkinson et al. 2011; Hopkinson 2014). While Pt is of low abundance in the open ocean, it is widely distributed globally from sub-polar to tropical environments and able to grow between 5 and 28 °C in the lab (Kudo et al. 2000; De Martino et al. 2011; Rastogi et al. 2020). Here, we acclimated the model diatom Pt to three different temperatures over its growth range and measured CCM activity using membrane inlet mass spectrometry (MIMS). Experimental data were fit within the Pt CCM model, based on Fragilariopsis cylindrus (Fcyl) CCM (Li and Young 2022) adapted from the “chloroplast pump” model (Hopkinson 2014), to explore how the Pt CCM functions across temperatures.

Materials and methods

Cell culture conditions and growth rate

Phaeodactylum tricornutum CCMP632 (Pt) cells were cultured in buffered (25 mM HEPES, pH 8.10) f/2 media with salinity at 35‰. The pH of the culture media was adjusted to 8.10 for each growth temperature, i.e., 10 °C, 18 °C and 25 °C. Light intensity was set at ~ 120 µmol m−2 s−1 with 16 h:8 h, light:dark cycle. Cell cultures were subcultured weekly and acclimated at each temperature for at least two weeks. Initial inoculum was 1 × 104 cells ml−1 (or 1 × 105 cells ml−1 at 10 °C) at day zero and cells were harvested during exponential growth phase, at cell densities between 5 × 105 and 1 × 106 cells ml−1 on the day of each experiment. Growth rates were calculated as described in Li and Young (2022). Beckman Coulter Z2 cell counter was used to determine cell density and size, assuming a spherical cell shape for diameter calculation. The nonspherical shape of Pt cells, which determines the cell surface area, does not impact the estimation of volume-based mass transfer coefficients.

Membrane inlet mass spectrometry

Membrane Inlet Mass Spectrometry (MIMS), including gas concentration calibrations and data analyses are as described in Li and Young (2022) for Fcyl. Photosynthesis and Ci usage measurements with Pt cells were carried out in a similar fashion though at different temperatures (10, 18 and 25 °C) with the presence of the eCA inhibitor acetazolamide (AZ) as detailed in previous studies (Hopkinson et al. 2011; Hopkinson 2014; Li and Young 2022). Like our Fcyl experiments, 18O, 13C-labeled NaHCO3 (final concentration ~ 2 mM) was used to determine HCO3− dehydration rate, CO2/HCO3− mass transfer coefficient (fc, fb) and CA activities at different temperatures (Li and Young 2022). Pt cells were concentrated before loading into the MIMS chamber with final average densities of 3 ~ 4 × 106 cells ml−1.

Photosynthetic rate and Ci uptake versus DIC concentration

Additional experiments to investigate the apparent kinetic properties of Pt Ci uptake were conducted by measuring rates of O2 evolution under varying dissolved inorganic carbon concentrations ([DIC]). Similar to the previous study of Hopkinson (2014), stepwise increases in [DIC] were created by loading 13C-NaHCO3 into the MIMS chamber in the presence of 25 mM HEPES buffered (pH 8.1) artificial seawater. For each different [DIC], Pt cells were treated to a dark–light-dark cycle until quasi-steady state was reached (see Fig. S1, e.g.). The CO2 and O2 signals were used for calculating photosynthetic rate and Ci uptake according to earlier studies (Badger et al. 1994; Hopkinson et al. 2011; Li and Young 2022). The cytosolic CO2 concentration was estimated using following equation:

$$\left[ {CO_{2} } \right]_{cyt} = \left[ {CO_{2} } \right]_{bulk} - \frac{{U_{{CO_{2} }} }}{{f_{c} }}$$
(1)

where [CO2]cyt and [CO2]bulk are the cytosolic and bulk CO2 concentrations respectively; UCO2 is the CO2 uptake rate and fc is the mass transfer coefficient of CO2. The rate of photosynthesis (O2 evolution) as a function of [CO2] was plotted and fit to the Michaelis–Menten equation.

RuBisCO kinetics and quantification

In vitro RuBisCO kinetics at different temperatures (6, 12, 18, 25 °C) were measured according to previous studies (Sharwood et al. 2008; Young et al. 2016). Briefly, RuBisCO active sites were quantified using [14C] 2-CABP binding assay as described by Sharwood et al. (2008). For the ribulose-P2-dependent 14CO2-fixation assays, crude extracts of soluble diatom protein were incubated with 15 mM NaH14CO3 and 15 mM MgCl2 at the relevant temperature for 10–15 min to activate RuBisCO. This extract was added to 7 ml septum-capped scintillation vials containing reaction buffer (0.5 ml of 100 mM EPPS-NaOH, pH 8, 15 mM MgCl2, 0.6 mM ribulose-P2, 0.1 mg ml−1 CA) equilibrated with 21% (v/v) O2 in N2 and five differing concentrations of 14CO2 (between 10 and 115 μM). Values for the half saturation constant for CO2 in air (KCair) and maximal carboxylase activity in air (VCmax) were extrapolated from the data using the Michaelis–Menten equation as described previously (Sharwood et al. 2008; Whitney et al. 2011). The carboxylation turnover rate of RuBisCO (kcatC) was calculated by dividing VCmax by the number of RuBisCO active sites as determined in [14C] 2-CABP-binding assay.

Temperature dependence of kcatC and KCair was described by fitting following equation as described by Arcus et al. (2016):

$$lnk = a \times lnT + b \times \left( \frac{1}{T} \right) + c$$
(2)

The best-fit constants (a, b, c) were used to interpolate kcatC and KCair values at different temperature (T).

Western Blots were used to quantify the relative abundance of RuBisCO in Pt cells cultured at different temperatures, with experimental details described previously (Li and Young 2022). However, Western Blots using commercial anti-RbcL antibody and standards consistently underestimated absolute RuBisCO concentrations in diatoms. Thus, for modeling purposes, average RuBisCO quantities were estimated by assuming Pt RuBisCO reaches 80% saturation rate during photosynthesis at 18 °C, based on previous modeling and field research work (Hopkinson 2014, Kranz et al. 2015).

Modeling and statistical analyses

We previously adapted the “chloroplast pump” models (Hopkinson 2014; Li and Young 2022) to simulate the Ci fluxes of Fcyl. By adjusting the HCO3− uptake behavior observed in our Pt data, as well as other physiochemical parameters acquired in this study, we modified the Fcyl CCM model to simulate Pt Ci fluxes during steady state photosynthesis at different temperatures. The model was constrained to describe observed CO2 concentrations, net O2 evolution rates and Ci uptake rates. The Python script with modeling parameter files can be found on GitHub (https://github.com/limengwsu/Pt_CCM_T). All descriptive statistical analyses including mean, stand deviation, tukey HSD p values, etc. were calculated using Python packages including pandas, numpy, scipy (Virtanen et al. 2020) and statsmodels (Seabold and Perktold 2010).

Results

Growth rates and photosynthesis with temperature

Pt net photosynthesis increased steadily across all temperatures tested, with an ~ fourfold increase between 10 and 25 °C, as measured by the rate of O2 evolution in the light (Fig. 1a). Respiration (O2 consumption in the dark) also increased significantly at 25 °C compared with 10 and 18 °C (p ≤ 0.001, Fig. 1b). The estimated daily net O2 evolution (calculated as 1/3 dark respiration and 2/3 light O2 evolution over a 16:8 daylight cycle), increased with temperature (p ≤ 0.001, Fig. 1c). However, the specific growth rate (d−1) of Pt only increased between 10 and 18 °C, not between 18 and 25 °C (Fig. 1c). This disparity could be attributed to changing rates of respiration and photosynthesis over a diel cycle that were not captured in our short-term MIMS experiments. In addition, cell size also increased significantly between 18 and 25 °C (p ≤ 0.001), though only by ~ 10% in volume, and may partly explain the difference observed between daily net photosynthesis and growth rates (Figs. 1c, S2).

Fig. 1
figure 1

Pt net photosynthetic, respiratory and growth rate at different temperature. a. Net oxygen evolutionary rate in the light per cell across different temperatures. b. Oxygen consumption (dark respiratory) rate per cell across different temperatures. c. The correlation between growth rate and averaged daily primary production rate, presented as daily net O2 evolution rate per cell. All box plots have whiskers extending maximum 1.5 times of IQR from box

Full size image

RuBisCO kinetics and quantity

Faster growth and photosynthetic rates at higher temperatures could be attributed to the temperature-driven increase in RuBisCO catalytic activity (kcatC). From 10 to 25 °C, RuBisCO kcatC increased from 0.9 to 3.3 s−1, an almost four-fold increase between 10 and 25 °C, similar in magnitude to the increase in photosynthetic oxygen evolution (Fig. 2a). Between 10 and 25 °C, there is also a 2.5-fold increase in KCair from 22 to 55 µM (Fig. 2a), thus at higher temperatures RuBisCO needs a higher CO2 concentration for similar saturation levels. As KCair was less temperature sensitive than kcatC, carboxylation efficiency (kcatC/KCair) increases with temperature.

Fig. 2
figure 2

The impact of temperature on RuBisCO kinetic properties and quantity in Pt. a. RuBisCO carboxylation kinetic parameters kcatC and KCair in the presence of air. Curve fitting and interpolation were used to derived parameters at different temperatures. b. Western Blot (WB) quantification of RbcL in Pt cultured at different temperatures. Error bars represent SD with n = 3

Full size image

There was little change in RuBisCO abundance per cell across temperatures in Pt (Figs. 2b, S3a). The Western Blot method tracks changes in relative abundance but appears to underestimate diatom RbcL quantity in Pt, which was also observed in our previous work on Fcyl (Li and Young 2022). Our estimation of average gross CO2 fixation rates at RuBisCO (based on estimated gross O2 evolution rates, Fig. S3b), increased from 0.90 to 3.0 × 10–17 mol cell−1 s−1 with a temperature increase from 10 to 25 °C. Dividing the gross CO2 fixation rates by kcatC, the results lead to an average minimum RuBisCO quantity at 1 × 10–17 mol per Pt cell. For modeling purposes and to estimate the relative substrate (CO2) concentration at RuBisCO, 80% saturation was assumed at 18 °C, which resulted in an estimated abundance of 1.24 × 10–17 mol RuBisCO per Pt cell across different temperatures. Combined with KCair data, we estimated the average CO2 concentration around RuBisCO during steady state photosynthesis to be 95, 147, and 151 µM at 10, 18, and 25 °C, respectively.

Inorganic carbon usage

An increase in cellular inorganic carbon (Ci) uptake was required to supply the faster carboxylation rates and maintain higher CO2 concentrations within the pyrenoid with increasing temperature. Total Ci uptake increases at higher temperatures, which was largely driven by an increase in CO2 uptake from 10 to 18 °C (Fig. 3a) and by an increase in HCO3− uptake from 18 to 25 °C (Fig. 3b). This results in the proportional supply of Ci in Pt to be ~ 80% CO2 at the lower temperatures of 10 and 18 °C, but only ~ 50% CO2 supply at 25 °C (p ≤ 0.006, Fig. 3c).

Fig. 3
figure 3

Inorganic carbon uptake by Pt at different temperatures. a and b Uptake rate of CO2 (UCO2) and HCO3− (UHCO3-) per cell basis. c. Fraction of CO2 supply normalized to O2 evolution

Full size image

Using 18O-exchange method (Hopkinson et al. 2011; Li and Young 2022), we estimated the mass transfer coefficients of CO2 and HCO3− (fc, fb respectively) in Pt at different temperatures. The permeability of CO2 across the Pt cell membrane (and frustule), measured as fc, steadily increased with temperature (p ≤ 0.006 for each temperature pair, Fig. 4a). The permeability for HCO3−, indicated by fb, was three orders of magnitude lower than CO2 and insensitive to temperature (Fig. 4a). Thus, as noted in previous studies (Hopkinson et al. 2011), the cell membrane can be treated as essentially impermeable for HCO3−. The increased fc at higher temperatures may facilitate the influx of CO2, where CO2 uptake rates are faster at higher temperatures than at 10 °C (p ≤ 0.007, Fig. 3b). As discussed by Hopkinson (2014), the cytosolic CA can also facilitate CO2 uptake. Like fc, the estimated catalytic rate of intracellular CA in Pt was higher at warmer temperatures (p < 0.02 for each temperature pair, Fig. 4b).

Fig. 4
figure 4

Temperature impact on Pt inorganic carbon mass transfer coefficient (a) and intracellular CA catalytic rate (b)

Full size image

Ci uptake rate versus Ci concentration

To understand the kinetics of Ci uptake by Pt and to model Ci fluxes during photosynthesis at different temperatures, we investigated the effect of [DIC] on net photosynthetic rates (Fig. 5a). As expected, lower [DIC] limited net photosynthetic rates across different temperatures. However, when Ci uptake was analyzed, the limitation of photosynthesis was not from the limited HCO3− uptake under our experimental conditions. HCO3− uptake was not responsive to bulk [HCO3−] in the range of 0.08–2 mM, regardless of temperature (Fig. 5b). This indicates that Pt has highly specific HCO3− transporters that operate with half saturation constants (K0.5) <  < 100 µM (agreeing with earlier research on Pt (Nakajima et al. 2013; Tsuji et al. 2017a; Nawaly et al. 2022). The observed insensitivity of HCO3− uptake to bulk [HCO3−] (and [DIC]) suggests an alternative entry of Ci into the cytosol (Fig. S4), i.e., through CO2 diffusion, which causes the limit on photosynthesis at low [DIC].

Fig. 5
figure 5

The impact of DIC (a) and HCO3− (b) concentration on net O2 evolution rate and uptake rate of HCO3− respectively. Data points were pooled from 4, 3, 5 independent cultures for temperature at 10, 18, 25 °C respectively

Full size image

The diffusion of CO2 into the cell is driven by the CO2 concentration gradient between the extracellular environment and the cytosol (Hopkinson 2014). Ci from the cytosol is then concentrated within the chloroplast by the “chloroplast pump”. The “chloroplast pump” as defined by Hopkinson (2014) intensively “pumps” HCO3− from the cytosol to the chloroplast, maintaining low HCO3− concentrations within the cytosol. Rapid cytosolic-CA-mediated net flux of CO2 to HCO3− then drives the diffusive influx of CO2 into the cytosol from outside the cell. With our measured fc, we estimated cytosolic CO2 concentrations using Eq. 1, which allowed us to investigate the relationship between cytosolic CO2 concentration and net photosynthetic rate (Fig. S5b). Fitting the net O2 evolution rate versus extracellular and cytosolic CO2 concentrations to the Michaelis–Menten curve separately revealed a strong “chloroplast pump”, with average K0.5 ≤ 1 µM for both extracellular and cytosolic CO2 (Fig. S5). With the equilibrium constant of CO2/HCO3− interconversion being ~ 100, the K0.5 for HCO3− transport from the cytosol to the chloroplast was estimated to be < 100 µM. With our experimental setup, the temperature impact on the affinity (K0.5) of HCO3−/CO2 at the chloroplast level was not distinguishable.

Modeling the CCM with temperature

Combining the quantitative data of RuBisCO kinetics, CA activities, net O2 evolution, Ci uptake rates and its kinetic properties, etc., we modeled the Ci fluxes in Pt during photosynthesis at different temperatures (Fig. 6). At steady state in general, CO2 diffuses from the bulk environment through the surface layer and cytoplasmic membrane into the cytosol, with additional CO2 supplied to the cytosol from mitochondrial respiration and leakage from the chloroplast (gray arrows pointing at cytosolic CO2, Fig. 6). The concentrations of CO2 and HCO3− in the cytosol favor the net conversion of CO2 to HCO3− by cytosolic CA. The CA supplied HCO3−, combined with extracellular HCO3− uptake through cytoplasmic membrane, is pumped into the chloroplast by potential transporter(s). This allows the accumulation of high concentrations of HCO3− within chloroplast and subsequently in the pyrenoid, where HCO3− is converted to CO2 by pyrenoid CA and fixed by RuBisCO (green arrows, Fig. 6). Due to the concentration gradient between pyrenoid and chloroplast stroma, some proportion of the CO2 leaks out from the pyrenoid into the chloroplast stroma and subsequently into the cytosol. As discussed by Hopkinson (Hopkinson 2014), the lack of CAs in the chloroplast stroma allows the buildup of HCO3−, with a very small fraction converted to CO2 before reaching the pyrenoid.

Fig. 6
figure 6

CCM models of Pt carbon (Ci) fluxes during photosynthesis at different temperatures. Gray, black, blue, and green arrows along with net flux rates (unit in 10–18 mol cell−1 s−1) represent diffusive, interconversion, active uptake, and fixation of inorganic carbon species, respectively. Arrow widths are drawn proportional to flux rates. The concentrations of CO2 and HCO3− are labeled in each compartment with units. The “Surface” compartment (Hopkinson et al. 2013) denotes the cellular layer outside the cytoplasmic membrane, including periplasmic space and frustule

Full size image

As temperature increases, the model was able to capture the experimentally observed fluxes of Ci uptake and photosynthetic rates, including faster diffusive CO2 uptake from the bulk environment to the cytosol at 18 °C compared with 10 °C, and enhanced HCO3− uptake into the cytosol at 25 °C compared with 18 °C (Figs. 3, 6). The model also enabled visualization of the intracellular Ci fluxes that are less accessible through experiments. All intracellular Ci fluxes operate at faster rates as temperatures rise (Fig. 6). Notably the net fluxes of cytosolic CA-catalyzed conversion of CO2 to HCO3−, HCO3− transport from the cytosol to the chloroplast, and the leakage of CO2 from the chloroplast to the cytosol are faster at higher temperatures. For every mole of HCO3− transported into the chloroplast, there are 0.37, 0.45, and 0.43 mol of CO2 leaking out of the chloroplast at 10, 18, and 25 °C, respectively. In other words, for every carbon fixed by RuBisCO, there are 1.6, 1.8 and 1.8 HCO3− molecules transported into the chloroplast at 10, 18, and 25 °C, respectively. When HCO3− uptake from the surface/bulk environment into the cytosol is included (i.e., total active HCO3− transport), the ratio of active HCO3− transport to carboxylation rate at RuBisCO is approximately 2:1 for all three temperatures. The HCO3− concentrations within the chloroplast are ~ 20 times of that within cytosol at 10 °C, and ~ 40 times at 18 and 25 °C. This is the result of cytosolic HCO3− concentrations remaining ~ 0.7 mM across all temperatures, whereas the chloroplast HCO3− concentration increased from 15 mM at 10 °C to ~ 30 mM at 18 and 25 °C.

Discussion

The CCM of Pt is essential to supply elevated CO2 concentrations around RuBisCO (Hopkinson et al. 2011; Hopkinson 2014). While there has been much focus on the response of diatom carbon fixation to ocean acidification (Wu et al. 2014; Shi et al. 2019; Wada et al. 2021), less is known about how marine heatwaves and rising sea surface temperatures (SST) may impact the diatom CCM and ultimately the ability of diatoms to sequester carbon.

CCM activity increases with temperature

Our experimental conditions of nutrient replete, saturating light resulted in increasing carbon fixation rates in Pt as temperatures increased. Concomitantly, there was also an increase in CCM activity to support the increased demand of CO2 for fixation, overcoming the challenges at higher temperatures of lower CO2 solubility in the bulk medium, higher half saturation constants for CO2 by RuBisCO and lower CO2/O2 specificity of RuBisCO (Galmés et al. 2016). Increased activity of the CCM occurred due to an overall increase in catalytic rates of CAs and HCO3− transporters, as well as enhanced membrane permeability to CO2 and increased rates of CO2 diffusion. As temperatures increased, we observed a shift towards a larger proportion of HCO3− to CO2 uptake from the bulk media, and a slight lowering of RuBisCO saturation state (from 81 to 73%). Modeling of the CCM revealed that Pt initially increased CO2 supply to RuBisCO by enhancing its “chloroplast pump” (from 10 to 18 °C). This involved increasing HCO3− transport across the chloroplast membranes. The rapid removal of inorganic carbon from the cytosol was quickly replaced by predominantly diffusive entry of CO2 but also active HCO3− uptake into the cell. The ratio of cellular CO2: HCO3− uptake did not change between 10 and 18 °C (Fig. 3c). However, between 18 and 25 °C it appeared that the “chloroplast pump” cannot be further enhanced, either unable to elevate chloroplast HCO3− concentrations above 30 mM or unable to maintain a chloroplast-to-cytosol HCO3− gradient greater than ~ 40 fold. Once this limit had been reached, the increased demand for inorganic carbon was supplied by enhancing the uptake of HCO3− from the bulk media. A comparable intracellular HCO3− concentration ~ 30 mM has been estimated in cyanobacteria previously (Woodger et al. 2005; Mangan and Brenner 2014).

Temperature effects on kinetics or molecular regulation on CCM components?

As temperature increases, catalytic activity, diffusion rates, and membrane fluidity also increase until an optimum is reached. Observed increased rates of CA activity, HCO3− transport, and CO2 carboxylation at RuBisCO are of magnitudes that could be explained by the direct impact of temperature (for enzymes assuming a Q10 around 2) without changes in enzyme abundance. Likewise, the differences in the magnitudes of various Ci fluxes (Fig. 6b, c) between 10 and 18 °C could also be entirely due to different temperature sensitivities of CCM components. However, at 25 °C the shifted ratio of CO2 to HCO3− uptake fluxes (Fig. 6a) and the upheld saturation state of RuBisCO suggest that there is a temperature-driven physiological and molecular regulation of CCM components. Further investigations (e.g., transcriptional studies) beyond our current model are needed to test whether Pt regulates its CCM through gene expression in response to changing temperatures. Previous studies have demonstrated that CCM components (e.g., CA, HCO3− transporters) are transcriptionally regulated via CO2 availability (Harada et al. 2006; Ohno et al. 2012; Nakajima et al. 2013; Hennon et al. 2015) indicating that the CCM is dynamically able to respond to changing environmental conditions.

Temperature impacts on the energy cost of the Pt CCM

The energetic cost of the CCM is thought to be primarily driven by the active uptake of HCO3− across membranes and to be strongly correlated with the amount of CO2 leaked from the pyrenoid. While CO2 leaks from the pyrenoid and chloroplast in the Pt CCM model, it is recycled in the cytosol by CA into HCO3−. This “chloroplast pump” model in Pt (Hopkinson et al. 2011; Hopkinson 2014) revealed the dependence on active pumping of HCO3− from the cytosol into the chloroplast to maintain a low cytosolic CO2 concentration. This low cytosolic CO2 concentration drives the net influx of CO2 from the bulk environment into the cytosol, along with rapid recycling of CO2 respired from the mitochondria or leaked from the chloroplast. Below we discuss the energy costs of this model in the context of temperature but acknowledge that other diatoms have different CCM strategies such as relying on external CA for extracellular HCO3 uptake while CO2 could leak from the cell (Tsuji et al. 2017a; Li and Young 2022).

Active transport of HCO3− across the cytoplasmic membrane in Pt is done by the Na+-dependent SLC4-type transporters while the “chloroplast pump” HCO3− transporters are yet to be identified (Nakajima et al. 2013; Tsuji et al. 2017b; Nawaly et al. 2022). Without direct measurements, HCO3− transport across the plasmalemma or chloroplast envelope membranes was estimated to require 0.5 or 1 ATP for each HCO3− transported (Hopkinson et al. 2011; Raven et al. 2014). Considering the HCO3− concentration gradient between the chloroplast and cytosol was ~ 40 fold at maximum (Fig. 6), and factoring in the ~ 20 mV chloroplast envelope membrane potential under light (Demmig and Gimmler 1983; Pottosin and Dobrovinskaya 2015), the energy input required would be ~ 11 kJ/mol HCO3− transported from the cytosol to the chloroplast stroma, which is considerably lower than the −51 kJ/mol for the hydrolysis of ATP (estimated from chloroplast ATP/ADP concentrations (Heineke et al. 1991)). This indicates that a relatively energy efficient “chloroplast pump” is possible, so the estimate of 0.5 ATP for each HCO3− transported can hold. In contrast to HCO3− transport into the chloroplast, the cytoplasmic uptake of HCO3− has a favorable concentration gradient, i.e., the cytosolic HCO3− concentration is lower than the bulk environment. This favorable concentration gradient, together with the unfavorable cytoplasmic membrane potential −84 mV (resting potential, oscillating up to −40 mV) measured from the diatom Odontella sinensis (Taylor 2009), points to an estimated energy requirement of ~ 5.5 kJ/mol for HCO3− transport across the cytoplasmic membrane into the cytosol. Thus, HCO3− transporters (such as the SLC4-type) on the cytoplasmic membrane can operate efficiently using 0.5 or less ATP per HCO3− uptake into the cytosol. The ratio of total active HCO3− transport to CO2 fixed at RuBisCO remained at 2:1 across all temperatures (Fig. 6) indicating that the energy cost of the Pt CCM per CO2 fixed does not change with temperature unless the energetic costs for chloroplast and cytoplasmic HCO3− uptake are different. Nevertheless, higher photosynthetic rates at higher temperatures (under nutrient replete conditions) will need a higher total energy input, requiring higher saturating light intensities.

In addition to the energetic cost of HCO3− transport, energy costs and resource allocation of the CCM must also take into consideration the synthesis and maintenance of the CCM biophysical components (Raven et al. 2014). To compensate for slow catalytic rates at cold temperatures, many CCM components and RuBisCO are likely increased in abundance (Young et al. 2015; Li and Young 2022), this would require a larger investment in resources, such as nitrogen. While the abundance of other CCM components were not directly measured in this study, RuBisCO abundance did not change with temperature.

Combined effects of increasing SST and ocean acidification

Climate change will lead to both increasing SST and lowering of ocean pH. While warming SST leads to lower solubility of CO2 in seawater with a given pH, increasing atmospheric CO2 leads to elevating dissolved CO2 and HCO3− in the ocean, decreasing the oceanic pH (Zeebe 2012; Cheng et al. 2022). We hypothesized that the effect of rising temperatures would put additional pressure on the CCM, negating any relaxation of the CCM due to increasing CO2 concentrations. Our study showed that Pt increases CCM activity with increasing temperature, maintaining RuBisCO saturation above 70%. However, the increase in CCM activity is in pace with faster carboxylation rates, so that the estimated cost of the CCM does not increase per carbon fixed. Our study was conducted under nutrient replete conditions over a temperature range that led to increased carbon fixation rates with temperature. Nutrient limitation is predicted to become more prevalent in conjunction with warming SST due to enhanced stratification and could counter any temperature-driven increases in growth rate (Marinov et al. 2010). How the CCM functions under higher temperatures when nutrient limitation restricts carbon fixation rates has yet to be elucidated though there is evidence of either the Calvin cycle or CCM being downregulated in phytoplankton under phosphate limitation (Beardall et al. 2005; Brembu et al. 2017). Our study also highlighted the dependence of HCO3− transport in Pt’s “chloroplast pump” and an upper limit of either chloroplast HCO3− concentrations or chloroplast-to-cytosol gradients. At this limit (above 18 °C in this study), the CCM is no longer sensitive to CO2 availability and ocean acidification would not affect carbon fixation rates under these experimental conditions.