Photosynth Res DOI 10.1007/s11120-013-9954-7

REGULAR PAPER

A chloroplast pump model for the CO2 concentrating mechanism in the diatom Phaeodactylum tricornutum Brian M. Hopkinson

Received: 13 August 2013 / Accepted: 18 November 2013 Ó Springer Science+Business Media Dordrecht 2013

Abstract Prior analysis of inorganic carbon (Ci) fluxes in the diatom Phaeodactylum tricornutum has indicated that transport of Ci into the chloroplast from the cytoplasm is the major Ci flux in the cell and the primary driving force for the CO2 concentrating mechanism (CCM). This flux drives the accumulation of Ci in the chloroplast stroma and generates a CO2 deficit in the cytoplasm, inducing CO2 influx into the cell. Here, the ‘‘chloroplast pump’’ model of the CCM in P. tricornutum is formalized and its consistency with data on CO2 and HCO3- uptake rates, carbonic anhydrase (CA) activity, intracellular Ci concentration, intracellular pH, and RubisCO characteristics is assessed. The chloroplast pump model can account for the major features of the data. Analysis of photosynthetic and Ci uptake rates as a function of external Ci concentration shows that the model has the most difficulty obtaining sufficiently low cytoplasmic CO2 concentrations to support observed CO2 uptake rates at low external Ci concentrations and achieving high rates of photosynthesis. There are multiple ways in which model parameters can be varied, within a plausible range, to match measured rates of photosynthesis and CO2 uptake. To increase CO2 uptake rates, CA activity can be increased, kinetic characteristics of the putative chloroplast pump can be enhanced to increase HCO3- export, or the cytoplasmic pH can be raised. To increase the photosynthetic rate, the permeability of the pyrenoid to CO2 can be reduced or RubisCO content can be increased.

B. M. Hopkinson (&) Department of Marine Sciences, University of Georgia, Athens, GA 30602, USA e-mail: [email protected]

Keywords Model

CO2 concentrating mechanism  Diatom 

Introduction Diatoms are major primary producers in the ocean, accounting for a significant fraction of total primary production in the ocean and dominating high productivity regions such as upwelling environments and spring blooms (Ducklow and Harris 1993; Jin et al. 2006). In these highly productive environments, CO2 concentrations can be significantly depleted, and diatoms have evolved an effective and relatively efficient CO2 concentrating mechanism (CCM) to maintain elevated CO2 concentrations around RubisCO despite low environmental concentrations (Morel et al. 2002; Reinfelder 2011). Although interest in the diatom CCM is relatively recent, there have been significant, rapid advances in our understanding of its molecular components and physiology, particularly in the model diatom Phaeodactylum tricornutum. Phaeodactylum tricornutum operates a strong CCM, which is necessary for success in its natural habitat (Martino et al. 2007). CO2 uptake generally supplies most of the inorganic carbon required for photosynthesis, but HCO3transport contributes as well (Burkhardt et al. 2001; Cassar et al. 2002). CO2 uptake is a passive process driven by an inward concentration gradient from seawater to the cytoplasm, but HCO3- uptake requires a membrane-embedded transporter. Sequencing of the P. tricornutum genome unexpectedly revealed genes with homology to bicarbonate transporters from the SLC4 and SLC26 families, which have been thoroughly studied in humans (Romero et al. 2004). Recently, one of these SLC4 homologs has been shown to be a Ci transporter, establishing that function has

123

Photosynth Res

been retained in this family despite the evolutionary distance between the diatom proteins and homologs in humans (Nakajima et al. 2013). When acclimated to low CO2, P. tricornutum can maintain CO2 and HCO3- uptake and, consequently, photosynthesis, even at low extracellular Ci concentrations (Burkhardt et al. 2001). High rates of photosynthesis, even at low extracellular Ci concentrations, are accomplished with a modest accumulation of Ci within the cell, of 2- to 6-fold that of extracellular concentrations (Burns and Beardall 1987; Colman and Rotatore 1995). P. tricornutum can sense and respond to changes in inorganic carbon availability, using a cyclic AMP intermediate to signal cellular Ci status and ultimately to regulate expression of CCM genes (Harada et al. 2006; Ohno et al. 2012). Although Ci uptake capabilities are reasonably well characterized, less is known about intracellular Ci fluxes in P. tricornutum. The localization of a highly expressed carbonic anhydrase (CA), PtCA1, to the pyrenoid strongly suggests that accumulated HCO3- is converted to CO2 within the pyrenoid for immediate fixation (Tanaka et al. 2005; Tachibana et al. 2011). Although there is some evidence that a C4 intermediate may be involved in carbon transport from the cytoplasm to the chloroplast, recent knock down of a critical C4 gene showed that a C4 mechanism, at least in its canonical form, is not present (McGinn and Morel 2008; HaimovichDayan et al. 2013). Analysis of inorganic carbon fluxes in P. tricornutum using 18O-exchange indicated that transfer of inorganic carbon from the cytoplasm to the chloroplast is the major active flux in the CCM (Hopkinson et al. 2011). Based on these findings, we proposed a model for the P. tricornutum CCM, here termed the ‘‘chloroplast pump’’ model (Hopkinson et al. 2011). In this model, the major driving force for the CCM is active transport of HCO3into the chloroplast, which generates two important CO2 gradients: (1) the elevation of CO2 around RubisCO over the extracellular concentration and (2) the CO2 deficit in the cytoplasm relative to the extracellular concentration, which drives diffusive CO2 influx (Fig. 1; Hopkinson et al. 2011). HCO3- transported by the chloroplast pump accumulates in the chloroplast stroma and then diffuses into the pyrenoid, where CA converts it to CO2, elevating the CO2 concentration around RubisCO. Some of this CO2 is fixed by RubisCO (approximately one-third under typical oceanic conditions), but the remainder leaks out of the chloroplast to the cytoplasm, where it is converted back to HCO3- by a CA. Net Ci import for photosynthesis comes partially from HCO3- uptake via a membrane-bound transporter, but most is supplied by diffusive CO2 influx driven by the reduced CO2 concentrations in the cytoplasm, which is ultimately maintained by transport of HCO3from the cytoplasm to the chloroplast. The chloroplast pump model was proposed based primarily on measurements of inorganic carbon fluxes and

123

Fig. 1 Diagram of the chloroplast pump model with representative Ci concentrations. CO2 and HCO3- are treated in four compartments (extracellular solution, cytoplasm, chloroplast stroma, and pyrenoid) subject to hydration and dehydration within each compartment (indicated with dark blue arrows), photosynthetic removal in the pyrenoid (green arrow), and fluxes between compartments. Diffusive fluxes are indicated with wavy black arrows and are parameterized using mass transfer coefficients (fx), whereas active fluxes are indicated with straight light-blue arrows and are parameterized using Michaelis–Menten kinetics

was previously shown to be consistent with 13C isotope fractionation data (Hopkinson et al. 2011). In this study, the consistency of the chloroplast pump model with a wide range of physiological data and current molecular data is tested. Refinements in the model are made to improve agreement between data and model predictions.

Methods Model structure A mechanistic, numerical model of the chloroplast pumpbased CCM was developed to assess the agreement of this model with data and constraints on its function. CO2 and HCO3- concentrations are modeled in four compartments (extracellular solution, cytoplasm, chloroplast stroma, and pyrenoid), subject to hydration–dehydration reactions, diffusion, active transport, and photosynthesis (Fig. 1). Because equilibrium between HCO3- and CO32- is established very rapidly (Zeebe and Wolf-Gladrow 2001), CO32- is treated implicitly as part of the HCO3- pool. CO2 is allowed to diffuse between all compartments, as described using mass transfer coefficients, but HCO3- can only diffuse into the pyrenoid because cellular membranes are generally impermeable to HCO3-. The only active transport steps postulated in the chloroplast pump model are the transport of HCO3- into the cytoplasm from the extracellular solution (Uc) and the transport of HCO3- into the chloroplast from the cytoplasm (Up). CO2 hydration and HCO3- dehydration are treated with first order rate

Photosynth Res Table 1 Model Notation Symbol

Definition

Value

Units

Source

?P?COa2

Base cx

CO2 concentrations

Varies

lM

bx

HCO3- concentrations

Varies

lM

kuf

Uncat

3.7 9 10-2

s-1

Johnson (1982)

kcf

Cytoplasm

450

s-1

Hopkinson et al. (2011)

kpf

Plastid–stroma

3.7 9 10-2

s-1

Johnson (1982)

kyf

Pyrenoid

1.8 9 104

s-1

Hopkinson et al. (2011)

pHe

Extracellular pH

8.0

–

Measured, Zhang and Byrne (1996)

pHc

Cytoplasmic pH

7.3

–

Herve et al. (2012)

pHp

Chloroplast pH

8.15

–

Anning et al. (1996)

Uc

HCO3- uptake rate into cytoplasm

Varies

mol cell-1 s-1

Vm-Bc

Maximal HCO3- uptake rate into cytoplasm

4 9 10-18

mol cell-1 s-1

Badger et al. (1994) Knauf et al. (2002)

-

600

Km-Bc

Half-saturation constant for HCO3 uptake into cytoplasm

140

lM

Up

HCO3- uptake rate into chloroplast

Varies

mol cell-1 s-1

Vm-Bp

Maximal HCO3- uptake rate into chloroplast

7 9 10-17

mol cell-1 s-1

Hopkinson et al. (2011)

lM

Knauf et al. (2002)

-

Km-Bp

Half-saturation constant for HCO3 uptake into chloroplast

140

P

Photosynthetic rate (CO2 fixation rate)

Varies

35

mol cell-1 s-1 -18

mol cell-1

Losh et al. (2013)

3.4

s-1

Whitney et al. (2001)

RubisCO half-saturation constant for CO2

41

lM

Whitney et al. (2001)

fc-c

CO2 mass transfer coefficient from solution to cytoplasm

2.3 9 10-8

cm3 s-1

Hopkinson et al. (2011)

fc-p

CO2 mass transfer coefficient from cytoplasm to chloroplast

6 9 10-9

cm3 s-1

Hopkinson et al. (2011)

fc-y

CO2 mass transfer coefficient from chloroplast to pyrenoid

6.7 9 10-10

cm3 s-1

Hopkinson et al. (2011)

fb-y

HCO3- mass transfer coefficient from chloroplast to pyrenoid

7.5 9 10-9

cm3 s-1

Hopkinson et al. (2011)

Ve

Volume of extracellular solution

1

mR

RubisCO content

6.3 9 10

kcat

RubisCO turnover rate

Km-R

7.0 9 10

-18b

4.0 9 10-10b

cm3

Measured

-11

Vc

Cytoplasmic volume

6.6 9 10

cm3 cell-1

Measured, Coulter counter

Vp

Chloroplast stroma volume

6.0 9 10-12

cm3 cell-1

Hopkinson et al. (2011)

Vy N

Pyrenoid volume Cell number

2.1 9 10-13 Varies

cm3 cell-1 cell

Hopkinson et al. (2011) Measured, Coulter counter

a

parameters are the same in the naı¨ve and refined models unless listed

b

fc-y was decreased and mR was increased in the base?P model to increase the Ci-saturated rate of photosynthesis

constants. The first order rate constants for CO2 hydration (kxf) are specified (Table 1), and HCO3- dehydration rate constants (kxr) are determined from the forward rate constant, compartmental pH, and the CO2/HCO3- equilibrium constant, assuming microscopic reversibility. The CO2/ HCO3- equilibrium constant was calculated using the equations of (Zeebe and Wolf-Gladrow 2001). The model is described by the following system of equations:

dce fcc N ¼ kuf ce þ kur be þ ðc c  ce Þ Ve dt

ð1Þ

dbe N ¼ kuf ce  kur be  Uc ð2Þ Ve dt
 dcc 1  ¼ kcf cc þ kcr bc þ fcc ðce  cc Þ þ fcp cp  cc Vc dt ð3Þ

123

Photosynth Res

 dbc 1  ¼ kcf cc  kcr bc þ Uc  Up ð4Þ Vc dt

 dcp 1  ¼ kpf cp þ kpr bp þ fcp cc  cp þ fcy cy  cp Vp dt ð5Þ
 dbp 1  ¼ kpf cc  kpr bc þ fby by  bp þ Up Vp dt

ð6Þ


 dcy 1  ¼ kyf cy þ kyr by þ fcy cp  cy  P Vy dt

ð7Þ


dby fby ¼ kyf cy  kyr by þ bp  by dt Vy

ð8Þ

where rates of photosynthesis and HCO3- uptake are described by Michaelis–Menten kinetics: P¼

mR kcat cy KmR þ cy

ð9Þ

Uc ¼

VmBc be KmBc þ be

ð10Þ

Up ¼

VmBp bc KmBp þ bc

ð11Þ

The notation is detailed in Table 1. The model was solved in Matlab using built-in differential equation solvers for stiff systems. Model parameterization The model was initially parameterized using values obtained from the literature (Table 1). Wherever possible, parameter values for P. tricornutum or other diatoms were used, but in some cases this was not possible. The model implemented with these literature value parameters was termed the ‘‘base’’ model. Upon comparing the ‘‘base’’ model to data, some discrepancies were noted and parameters were varied to obtain two refined model implementations: (1) ‘‘base?P,’’ in which Ci-saturated photosynthetic rates were increased to values close to observed rates by modifying pyrenoid and RubisCO parameters; and (2) ‘‘base?P?CO2,’’ in which several additional parameters were modified to increase CO2 uptake rates (Table 1). Many important parameters are discussed further in detail in ‘‘Results and Discussion’’, but a few parameters will be discussed here. The kinetics of HCO3- transport into the cytoplasm are specified using the maximal observed rate of HCO3- uptake (based on the Badger et al. 1994 method) as the maximal rate for the transporter and using the half-saturation constant of a human HCO3- transporter (Knauf et al. 2002). RubisCO content per cell was determined from the absolute quantification of RubisCO by Losh et al. (2013), who reported

123

RubisCO content as a percentage of total protein, and from the protein per cell data provided by the authors (J. Losh, personal communication). Photosynthesis and Ci uptake versus Ci concentration P. tricornutum (CCMP 632) was grown in Aquil media (Price et al. 1988), made from a natural seawater base, and supplemented with 5 mM EPPS buffer at pH 8.0. The culture was grown in a lighted incubator (125–150 lmol photon m-2 s-1) on a light–dark cycle (16 h on, 8 h off) at 18 °C. Cell numbers were counted daily with a Coulter Counter, and cells were harvested by filtration during exponential growth. After resuspension in Ci-free artificial seawater buffered to pH 8.0 with 20 mM TRIS, the cells were placed in the cuvette of a membrane inlet mass spectrometer (MIMS) thermostated at 20 °C. Light (200 lmol photons m-2 s-1 from a halogen bulb) was applied to induce photosynthesis and thus consume any residual Ci. When net photosynthesis stopped (*10 min) the light was turned off and the photosynthesis vs. Ci experiment was begun. Ci was gradually added to the sample, and photosynthesis and CO2 and HCO3- uptake were measured in a series of light–dark cycles following the method of Badger et al. (1994). Net photosynthetic rates were determined from the rate of O2 production, and net CO2 uptake rates were calculated from the extent to which CO2 was drawn down below equilibrium. Net HCO3- uptake was then calculated as the difference between photosynthesis and net CO2 uptake. The MIMS system consists of a water-jacketed chamber with optical ports and a thin Teflon membrane on the base, interfaced to a Pfeiffer QMS 220 quadrupole mass spectrometer.

Results and discussion Overview To test the consistency of the chloroplast pump model of the P. tricornutum CCM with data, a numerical box model representing important intracellular compartments (cytoplasm, chloroplast stroma, and pyrenoid) and Ci fluxes was developed (Fig. 1). The ability or inability of the model to explain Ci uptake and photosynthesis as a function of extracellular Ci, subject to literature constraints on CA activity, intracellular pH, and RubisCO content was found to be especially informative. Therefore, the model tests focus on analysis of this data (Fig. 2). The ‘‘base’’ model, parameterized from literature data, did a reasonable job explaining the shape of the photosynthesis and Ci uptake curves as a function of extracellular Ci, but underestimated

Photosynth Res

Fig. 2 Results from three different implementations of the chloroplast pump model in an attempt to match photosynthesis and Ci uptake vs. extracellular Ci data from a representative experiment. a, b, c Agreement between modeled (solid lines) and measured (points) rates of photosynthesis (P), CO2 uptake, and HCO3- uptake as a function of extracellular Ci. Also shown is the modeled rate of HCO3- transport by the chloroplast pump (Up). d, e, f Modeled CO2 concentrations in the extracellular solution (ce), cytoplasm (cc), chloroplast (cp), and pyrenoid (cy) as a function of extracellular Ci. g, h, i Modeled HCO3-

concentrations in the extracellular solution (be), cytoplasm (bc), chloroplast (bp), and pyrenoid (by) as a function of extracellular Ci. j, k, l Agreement between observed and modeled CO2 data during a photosynthesis vs. Ci experiment. CO2 drawdown occurred during the light periods (white background), when photosynthesis was occurring and then increased rapidly immediately after the light was turned off (dark periods indicated by gray background). After the CO2 concentration returned to equilibrium in each dark period, a Ci addition was made leading to another sudden increase in the CO2 concentration

123

Photosynth Res Table 2 Sum of square error between data sets and model fits (910-35) Model

Photosynthesis

CO2 uptake

HCO3- uptake

base

8.1

7.2

0.2

base?P

2.9

2.6

0.2

base?P?CO2

0.8

1.2

0.2

photosynthetic and CO2 uptake rates (Fig. 2a, d, g, j). These deficiencies were sequentially corrected by altering model parameters, resulting in two additional model implementations. In the first refinement (‘‘base?P,’’ i.e., increased photosynthesis), the Ci-saturated photosynthetic rate was increased by altering pyrenoid and RubisCO characteristics (Fig. 2b, e, h, k; Table 2). A second set of refinements was then made to increase CO2 uptake, resulting in the ‘‘base?P?CO2’’ implementation (i.e., increased photosynthesis and CO2 uptake) (Fig. 2c, f, i, l; Table 2). The effects of these parameter modifications are not entirely independent. For example, the modifications to increase CO2 uptake also increase the maximal rate of photosynthesis (compare Fig. 2b, c), but for simplicity, they will be treated as such. In the sections that follow, the compatibility of the chloroplast pump model with a wide variety of physiological and molecular data is assessed. Unless otherwise specified, the model implementation discussed is the ‘‘base?P?CO2’’ case, which best matches the data. Comparisons of the various model implementations are used to explain features of the model necessary to match Ci uptake and photosynthetic rates. Because these rates are emergent properties of the model, dependent on multiple factors, they will be discussed after more basic features of the model have been introduced. CA activity and localization The level of CA activity in each cellular compartment is critical to proper operation of the chloroplast pump CCM. CO2 that enters the cytoplasm, either through diffusive influx from the extracellular solution or by leakage out of the chloroplast, is converted to HCO3- by CA for transport by the chloroplast pump (Fig. 1). In the chloroplast stroma, CA must be absent or at a low concentration to allow the accumulation of HCO3-, whereas CA must be present in the pyrenoid to rapidly convert HCO3- to CO2 for fixation by RubisCO. The distribution of CAs in P. tricornutum is largely consistent with that required by the chloroplast pump model. GFP tagging of P. tricornutum CAs has shown CA to be localized to the pyrenoid, mitochondria, and chloroplast membranes or to compartments between chloroplast membranes, but none are localized to the chloroplast stroma (Tanaka et al. 2005; Tachibana et al.

123

Fig. 3 The effect of cytoplasmic CA activity on the modeled CO2 uptake rate at 300 lM extracellular Ci

2011). As yet, no CA has been localized to the cytoplasm, but as discussed below, one of the chloroplast membrane CAs could serve the same role. 18O-exchange measurements allow total intracellular CA activity to be measured on whole, live cells, which avoids CA inactivation that can occur during processing, but provides no direct information about CA localization. Hopkinson et al. (2011) measured CA activities using this approach and then applied a procedure to partition CA activity between the pyrenoid and the cytoplasm in a manner consistent with the 18O exchange data. When P. tricornutum was grown at typical oceanic conditions (pH 8.2, 2 mM DIC), CA activity in the cytoplasm and pyrenoid were determined to be 450/s and 18,000/s, respectively. Using the CA activities determined by Hopkinson et al. (2011) in the model provides a reasonable fit to the photosynthesis vs. Ci data (Fig. 2), but the model performance is sensitive to the cytoplasmic CA activity around the measured value, as illustrated by examining the effect of cytoplasmic CA activity on CO2 uptake at 300 lM Ci, the approximate half-saturation point of photosynthesis (Fig. 3). Increasing cytoplasmic CA activity increases the CO2 uptake rate modestly, improving model agreement with the data (compare Fig. 2a, b, c), whereas decreasing CA activity below 450/s quickly leads to a dramatic reduction in CO2 uptake (Fig. 3). Mechanistically, increased CA activity allows CO2 to be drawn down further in the cytoplasm and thus increases CO2 influx, while maintaining a similar rate of HCO3- generation from CO2. The HCO3- generated is continuously exported to the chloroplast by the chloroplast pump. Conversely, lowered CA activity limits the rate of HCO3- production, causing a build up of CO2 in the cytoplasm and thus a reduced CO2 uptake.

Photosynth Res

Fig. 4 The effect of compartmentalization of cytoplasmic CA on the modeled CO2 uptake rate at 300 lM extracellular Ci. a A diagram of the modified model structure, incorporating a chloroplast envelope compartment where CA is concentrated. b Effect of compartment pH

on rates of CO2 uptake at 300 lM extracellular Ci in the modified model (open symbols). For reference, the rate of CO2 uptake in the ‘‘base?P?CO2’’ model realization is also shown (filled symbols)

The measured cytoplasmic CA activity (450/s) is nearly optimal, but as shown in Fig. 3, CO2 uptake could be increased to some extent if cytoplasmic CA was increased. Total CA activity does increase in the light, but this increase was previously attributed to increased CA activity in the pyrenoid as a result of a rise in stromal pH during photosynthesis (Hopkinson et al. 2011). However, it has recently been reported that PtCA1 and PtCA2 are redox regulated, suggesting that diatoms can regulate CA activity (Kikutani et al. 2012). Some of the increased CA activity during photosynthesis may therefore be caused by activation of cytoplasmic CA, which enhances CO2 uptake. Another consideration is that though the model treats CA as if it is distributed homogeneously throughout the cytoplasm, diatom CAs are frequently compartmentalized or associated with membranes (Tanaka et al. 2005; Tachibana et al. 2011). The potential effect of compartmentalization was assessed by adding an additional compartment between the cytoplasm and the chloroplast stroma to the model, representing an internal compartment of the chloroplast envelope, where several CAs have been localized. CA activity was transferred from the cytoplasm to this compartment, and the chloroplast pump was located between the new compartment and the chloroplast stroma (Fig. 4a). On its own, the relocation of cytoplasmic CA activity to this compartment does not alter CCM performance significantly because the same rates of net CO2 hydration are obtained, leading to similar, but slightly lower rates of CO2 uptake and chloroplast pump rates in the compartmentalized model. However, if other characteristics of the compartment are modified, such as pH, the performance of the compartmentalized model can be enhanced slightly. Increasing pH shifts the CO2/HCO3equilibrium toward HCO3-, creating a greater driving force

for CO2 conversion to HCO3- and ultimately increasing the CO2 uptake rate (Fig. 4b). By increasing the HCO3concentration, the pH shift also allows higher chloroplast pump rates (Fig. 4b). Although CA compartmentalization offers no major benefit in the chloroplast pump model, it is not detrimental to CCM performance, which is notable because to date no CA protein has been localized to the cytoplasm of P. tricornutum. CAs have been localized to various layers of the chloroplast envelope (Tachibana et al. 2011), and one of these layers could be the region of CO2 convergence, where CO2 leaking from the chloroplast and entering the cell from the extracellular environment is converted to HCO3- for uptake by the chloroplast pump. CA activity in the pyrenoid is thought to be very high based on analysis of 18O-exchange data and on the localization of one of the most abundant CAs, PtCA1, to this compartment (Satoh et al. 2001; Hopkinson et al. 2011; Tachibana et al. 2011). The model performance is not sensitive to pyrenoid CA activity over a wide range. The activity in the model is sufficient to equilibrate HCO3- and CO2 in the pyrenoid, but even if activity is reduced such that equilibrium is no longer established, HCO3- builds up in the stroma until the dehydration rate in the pyrenoid is fast enough to generate a particular CO2 concentration set by the chloroplast pump rate (Fig. 5). The CO2 concentration, and thus photosynthetic rate, achieved is not altered by pyrenoid CA activity even at quite low activity. This counterintuitive result is obtained because the Ci loss terms from the chloroplast (photosynthesis and CO2 leakage) are dependent on the CO2 concentration, but independent of the HCO3- concentration (Eqs. 7, 9). At steady state, Ci losses to photosynthesis and CO2 leakage must equal the Ci import rate by the chloroplast pump. Therefore, the chloroplast pump rate directly determines the pyrenoid CO2

123

Photosynth Res

Fig. 5 Effect of pyrenoid CA activity on pyrenoid CO2 and HCO3concentrations

concentration, whereas the pyrenoid CA activity and pH effectively control the HCO3- concentration needed to achieve that CO2 concentration. However, extremely low rates of CA activity in the pyrenoid require a build up of HCO3- concentrations to levels inconsistent with cellular Ci measurements (see ‘‘Ci accumulation’’ below). In summary, the chloroplast pump model performs reasonably well with measured CA activities, although it is sensitive to the cytoplasmic CA activity near the measured value. Ci accumulation In the chloroplast pump model, Ci is primarily accumulated as HCO3- within the chloroplast stroma and, in fact, is depleted relative to seawater in the cytoplasm (Fig. 2g, h, i). The average intracellular concentration is strongly dependent on the stromal Ci concentration, which in turn is very sensitive to the stromal pH (Fig. 6). This dependence occurs because the chloroplast membranes are impermeable to HCO3-, and therefore HCO3- accumulates in the chloroplast stroma until it is in equilibrium with the CO2 concentration around RubisCO. There have been two direct measurements of the Ci concentration in P. tricornutum, which showed that Ci was accumulated intracellularly to 2 mM at 1 mM extracellular Ci (Colman and Rotatore 1995) and 1 mM at 180 lM extracellular Ci (Burns and Beardall 1987). In the model, 2 mM average intracellular Ci at 1 mM extracellular Ci can be achieved with a pH of 8.15 in the stroma (Fig. 6), which was consequently chosen as the stromal pH for the model. This pH is within the range of measured stromal pH values (*7.5–8.5) in plants and algae (Heldt et al. 1973; Anning et al. 1996). The model has difficulty achieving 1 mM average intracellular Ci at 180 lM extracellular Ci. However, the P. tricornutum culture used in Burns and Beardall’s work (1987) had a

123

Fig. 6 Effect of changing the chloroplast stroma pH on average intracellular Ci content

much lower half-saturation constant of photosynthesis for Ci than the culture used here (17 vs. 300 lM). Whether this is due to culture conditions or strain variability is unclear, but it means that the Burns and Beardall cultures were saturated for photosynthesis at 180 lM extracllular Ci and 1 mM intracellular Ci, and the model is able to achieve intracellular Ci in excess of 1 mM over a range of stromal pHs, when the extracellular Ci concentration is saturating for photosynthesis (Fig. 6). Intracellular pH The only direct work on intracellular pH on P. tricornutum reported an average intracellular pH of 7.6 at an extracellular pH of 7.5, but the method used did not permit determination of the pH of specific compartments, which is critical to evaluate model performance (Burns and Beardall 1987). Cytoplasmic pH in another diatom, Thalassiosira weissflogii, varied between 7.2 and 7.5 over the range of ocean pH (7.8–8.4) (Herve et al. 2012) and is generally similar to the cytoplasmic pH of other algae (Raven and Smith 1978; Anning et al. 1996; Braun and Hegemann 1999). Based on these data from the literature, the cytoplasmic pH was set to 7.3. Model agreement with Ci uptake data could be improved by raising the cytoplasmic pH, which increases the HCO3- concentration in the cytoplasm at low extracellular Ci, allowing the chloroplast pump to operate at higher rates. For example, the CO2 uptake rate in the ‘‘base’’ model at 300 lM extracellular Ci is 2.3 9 10-18 mol/cell/s with a cytoplasmic pH of 7.3, but can be increased to 2.8 9 10-18 mol/cell/s with a cytoplasmic pH of 7.5. However, enhanced cytoplasmic CA activity fulfilled the same objective and fits the data

Photosynth Res

slightly better and was thus modified in the refined models rather than modifying cytoplasmic pH. Measurements of pH in the chloroplast stroma are even rarer than cytoplasmic pH measurements, but typically during photosynthesis, the stromal pH rises above the cytoplasmic pH to 7.5–8.5 (Anning et al. 1996; Heldt et al. 1973; Kramer et al. 1999). Ci uptake and photosynthesis are not altered by chloroplast pH over a reasonable range (7.5–8.5), but as discussed above Ci, accumulation is very sensitive to stromal pH, and the available data are more consistent with a higher stromal pH in the chloroplast pump model. RubisCO content and kinetics P. tricornutum has a Form ID RubisCO with an intermediate affinity for CO2 (Kair c = 41 lM) and a moderatelylow maximal turnover rate (3.4 s-1) typical of non-green algae (Badger et al. 1998; Whitney et al. 2001). This enzyme composes 1–3 % of total protein in the cell, which is less than that in land plants, but similar to that in other algae (Losh et al. 2013). Based on RubisCO kinetics and cellular RubisCO content, the enzyme appears to be working near its maximal rate as the estimated maximal rates of carbon fixation are close to observed photosynthetic rates (Losh et al. 2013). In the ‘‘base’’ chloroplast pump model, RubisCO kinetic characteristics were specified using literature values and using the RubisCO content based on measurements from Losh et al. (2013). These initial parameters underestimated maximal rates of carbon fixation because pyrenoid CO2 concentrations were not sufficient to saturate RubisCO, which is necessary to match measured CO2 fixation rates (Fig. 2a, d). The kinetic parameters were considered to be relatively accurate because they agree with the values for other Form 1D RubisCOs and are consistent with a theoretical analysis of the enzyme (Badger et al. 1998; Tcherkez et al. 2006). Matching observed Ci-saturated photosynthetic rates can be achieved either by increasing RubisCO content by *30 % or decreasing the pyrenoid’s CO2 mass transfer coefficient by *50 %. It would also be possible to match Ci-saturated photosynthetic rates by increasing the maximal uptake rate of the chloroplast pump, but that would result in pump rates exceeding observed values (*39 the rate of photosynthesis; Hopkinson et al. 2011). In the improved models (‘‘base?P’’ and ‘‘base?P?CO2’’), the RubisCO content was increased moderately (10 %), and the pyrenoid’s CO2 mass transfer coefficient was decreased by 40 %. The RubisCO content measurements showed little variability, so the RubisCO content value in the model was changed only slightly. These modifications to achieve high rates of photosynthesis lead to higher CO2 concentrations in the pyrenoid (upto 110 lM) than might be expected based on the

regulatory response of the CCM (Fig. 2e, f). The chloroplast pump model predicts that the CCM should be downregulated when the extracellular CO2 concentration approaches the pyrenoid CO2 concentration because CO2 leakage out of the chloroplast is then decreased, reducing the need to import HCO3- into the chloroplast. P. tricornutum’s CCM is down-regulated over the range of *1–50 lM CO2 (Burkhardt et al. 2001; Nakajima et al. 2013), suggesting that the pyrenoid CO2 is somewhere in the range of 50–75 lM. A pyrenoid CO2 concentration in this range could be obtained if the increased photosynthetic rate was explained solely by increasing the RubisCO content. Ci uptake rates and transporters Physiological data show that P. tricornutum is capable of both net CO2 and net HCO3- uptake from the environment (Burkhardt et al. 2001; Rost et al. 2007) and of transfer of carbon from the cytoplasm to the chloroplast (Hopkinson et al. 2011). There are a number of putative SLC4-type HCO3- transporters in the genome that may be responsible for HCO3- fluxes, one of which has recently been confirmed to function as a Na?-dependent HCO3- transporter on the outer membrane (Nakajima et al. 2013). CO2 uptake across the cytoplasmic membrane is diffusive, driven by a concentration gradient. The ability of the chloroplast pump model to explain Ci uptake and photosynthetic rates as a function of extracellular Ci was assessed. The ‘‘base’’ chloroplast pump model matches the shape of the photosynthesis vs. Ci data, but underestimates rates of photosynthesis and CO2 uptake (Fig. 2a). In absolute terms, the cell is able to achieve remarkably low cytoplasmic CO2 concentrations (a minimum of 0.4 lM), but at low extracellular CO2 concentrations, the gradient is not sufficient to support observed CO2 uptake rates. This inability to match the observed CO2 uptake rates is shown clearly by the overestimate of the extracellular CO2 concentration in the ‘‘base’’ model at low Ci concentration in a photosynthesis vs. Ci experiment (Fig. 2j). Modeled HCO3- uptake matches the data reasonably well, though because the maximal uptake rate was set from the data and the half-saturation constant for SLC4 transporters is low (140 lM; Knauf et al. 2002), there is not much informative variability in the data to fit. Diffusive CO2 uptake is dependent on the ability of the cell to maintain a low CO2 concentration in the cytoplasm and is ultimately dependent on the net export of Ci from the cytoplasm and subsequent CO2 fixation. No single parameter can be changed to obtain agreement between the model and the CO2 uptake data. First, the maximal photosynthetic rate must be increased by decreasing the pyrenoid’s CO2 mass transfer coefficient and by increasing

123

Photosynth Res

RubisCO content (see previous section). Lower CO2 concentrations in the cytoplasm must then be achieved through increased HCO3- efflux via the chloroplast pump. In the ‘‘base?P?CO2’’ model, the cytoplasmic CA activity is raised moderately (from 450 to 600 s-1), increasing the rate of CO2 hydration, and the half-saturation constant of the chloroplast pump for HCO3- is decreased from 140 to 35 lM. Although this Km is lower than literature values for human SLC4 transporters, it is consistent with the Km for the cyanobacterial BicA HCO3- transporter from the related SLC26 family (Price et al. 2004). These modifications to the model permit higher rates of HCO3- efflux from the cytoplasm via the chloroplast pump, driving faster rates of CO2 influx from the environment and nearly matching observed rates of CO2 uptake and extracellular CO2 concentrations, though there is still some discrepancy between the modeled rates and observed rates of CO2 uptake (Fig. 2c, l; Table 2).

Conclusions The congruence among molecular data, physiological data, and the chloroplast pump model for the P. tricornutum CCM is encouraging, but there are still significant gaps in our knowledge. Achieving good agreement between the model and data required the modification of some parameter values from their best estimates in the literature, and there are multiple parameter values that result in similar model outputs. The parameters most critical for model performance, but with few empirical constraints are the permeability of the pyrenoid to CO2 and the kinetic characteristics of the chloroplast pump transporter. Because there is little excess RubisCO capacity, the enzyme must be nearly saturated to support observed photosynthetic rates. At the inferred rates of HCO3- import into the chloroplast, this requires that the pyrenoid be a significant barrier to CO2 efflux, *10–20-fold lower than the diffusion limited rate. The lack of excess RubisCO capacity also requires CO2 concentrations to be quite high in the pyrenoid (on the order of 100 lM), which is somewhat at odds with expectations of a lower value (*50 lM) based on the regulatory response of P. tricornutum’s CCM to external CO2 concentrations (Burkhardt et al. 2001; Hopkinson et al. 2011). Maintenance of CO2 influx at low extracellular CO2 concentrations require that CO2 and HCO3- concentrations in the cytoplasm be depleted to very low levels. To achieve these low levels of CO2 and HCO3-, the chloroplast pump transporter must have a half-saturation constant more similar to a cyanobacterial transporter than to the human SLC4 transporter. However, given the diversity of this family and the environmental constraints on diatom

123

HCO3- transporters, such a half-saturation constant does not seem implausible. Obtaining direct data on HCO3transporter kinetics may be achieved through overexpression in model systems (Romero and Boron 1999). Alternative tests of the chloroplast pump model include overexpression of a CA in the chloroplast stroma, which would be expected to dissipate the HCO3- pool, thwarting the CCM (e.g., Price and Badger 1989), localization of additional putative HCO3- transporters to determine if any are collocated with CAs in a chloroplast envelope compartment (Fig. 4a), and further physiological characterization of Ci uptake and processing, particularly to determine if flux rates are regulated according to the CO2 gradient between the pyrenoid and environment, as previously predicted (Hopkinson et al. 2011). Because the CCM has not been as well characterized in other diatoms, it is unclear whether this model is applicable to other species. Many diatoms have physiological characteristics (photosynthesis and Ci uptake vs. extracellular Ci, CA activity, RubisCO content, etc.) similar to P. tricornutum. Therefore, the chloroplast pump model should be broadly compatible with such data in other diatom species (Burkhardt et al. 2001; Rost et al. 2003; Losh et al. 2013). However, the presence of a CA in the chloroplast stroma of Thalassiosira pseudonana and C4 features in T. weissflogii suggest that the CCMs of other diatoms may function based on significantly different principles, and further research will be required to understand their mechanics (Reinfelder et al. 2000; Roberts et al.2007; Tachibana et al. 2011). Acknowledgments This work was supported by grants from the National Science Foundation (EF 1041023 and MCB 1129326 to B.H.). J. Losh and F.M.M. Morel (Princeton University) are thanked for providing data on total protein and RubisCO content in P. tricornutum. Comments from two anonymous reviewers improved the work.

References Anning T, Nimer N, Merrett MJ, Brownlee C (1996) Costs and benefits of calcification in coccolithophorids. J Marine Syst 9:45–56 Badger MR, Palmqvist K, Yu JW (1994) Measurement of CO2 and HCO3- fluxes in cyanobacteria and microalgae during steadystate photosynthesis. Physiol Plantarum 90:529–536 Badger MR, Andrews TJ, Whitney SM, Ludwig M, Yellowless DC, Leggat W, Price GD (1998) Diversity and coevolution of RubisCo, plastids, pyrenoids, and chloroplast-based CO2 concentrating mechanisms in algae. Can J Bot 76:1052–1071 Braun FJ, Hegemann P (1999) Direct measurement of cytosolic calcium and pH in living Chlamydomonas reinhardtii cells. Eur J Cell Biol 78:199–208 Burkhardt S, Amoroso G, Riebesell U, Su¨ltemeyer D (2001) CO2 and HCO3- uptake in marine diatoms acclimated to different CO2 concentrations. Limnol Oceanogr 46:1378–1391

Photosynth Res Burns BD, Beardall J (1987) Utilization of inorganic carbon by marine microalgae. J Exp Mar Biol Ecol 107:75–86 Cassar N, Laws EA, Popp BN, Bidigare RR (2002) Sources of inorganic carbon for photosynthesis in a strain of Phaeodactylum tricornutum. Limnol Oceanogr 47:1192–1197 Colman B, Rotatore C (1995) Photosynthetic inorganic carbon uptake and accumulation in two marine diatoms. Plant Cell Environ 18:919–924 Ducklow HW, Harris RP (1993) Introduction to the JGOFS NorthAtlantic bloom experiment. Deep Sea Res II 40:1–8 Haimovich-Dayan M, Garfinkel N, Ewe D, Marcus Y, Gruber A, Wagner H, Kroth PG, Kaplan A (2013) The role of C4 metabolism in the marine diatom Phaeodactylum tricornutum. New Phytol 197:177–185 Harada H, Nakajima K, Sakaue K, Matsuda Y (2006) CO2 sensing at ocean surface mediated by cAMP in a marine diatom. Plant Physiol 142:1318–1328 Heldt HW, Werdan K, Milovancev M, Geller G (1973) Alkalization of the chloroplast stroma caused by light-dependent proton flux into the thylakoid space. Biochim Biophys Acta 314:224–241 Herve V, Derr J, Douady S, Quinet M, Moisan L, Lopez PJ (2012) Multiparametric analyses reveal the pH-dependence of silicon biomineralization in diatoms. PLoS One 7:e46722 Hopkinson BM, Dupont CL, Allen AE, Morel FMM (2011) Efficiency of the CO2-concentrating mechanism of diatoms. Proc Natl Acad Sci USA 108:3830–3837 Jin X, Gruber N, Dunne JP, Sarmiento JL, Armstrong RA (2006) Diagnosing the contribution of phytoplankton functional groups to the production and export of particulate organic carbon, CaCO3, and opal from global nutrient and alkalinity distributions. Global Biogeochem Cycles. doi:10.1029/2005GB002532 Johnson KS (1982) Carbon dioxide hydration and dehydration kinetics in seawater. Limnol Oceanogr 27:849–855 Kikutani S, Tanaka R, Yamazaki Y, Hara S, Hisabori T, Kroth PG, Matsuda Y (2012) Redox regulation of carbonic anhydrases via thioredoxin in chloroplast of the marine diatom Phaeodactylum tricornutum. J Biol Chem 287:20689–20700 Knauf PA, Law FY, Leung TWV, Gehret AU, Perez ML (2002) Substrate-dependent reversal of anion transport site orientation in the human red blood cell anion-exchange protein, AE1. Proc Natl Acad Sci USA 99:10861–10864 Kramer DM, Sacksteder CA, Cruz JA (1999) How acidic is the lumen? Photosyn Res 60:151–163 Losh JL, Young JN, Morel FMM (2013) Rubisco is a small fraction of total protein in marine phytoplankton. New Phytol 198:52–58 Martino AD, Meichenin A, Shi J, Pan K, Bowler C (2007) Genetic and phenotypic characterization of Phaeodactylum tricornutum (Bacillariophyceae) accessions. J Phycol 43:992–1009 McGinn PJ, Morel FMM (2008) Expression and inhibition of the carboxylating and decarboxylating enzymes in the photosynthetic C4 pathway of marine diatoms. Plant Physiol 146:300–309 Morel FMM, Cox EH, Kraepiel AML, Lane TW, Milligan AJ, Schaperdoth I, Reinfelder JR, Tortell PD (2002) Acquisition of inorganic carbon by the marine diatom Thalassiosira weissflogii. Funct Plant Biol 29:301–308 Nakajima K, Tanaka A, Matsuda Y (2013) SLC4 family transporters in a marine diatom directly pump bicarbonate from seawater. Proc Natl Acad Sci USA 110:1767–1772 Ohno N, Inoue T, Yamashiki R, Nakajima K, Kitahara Y, Ishibashi M, Matsuda Y (2012) CO2-cAMP-responsive cis-elements

targeted by a transcription factor with CREB/ATF-like basic zipper domain in the marine diatom Phaeodactylum tricornutum. Plant Physiol 158:499–513 Price GD, Badger MR (1989) Expression of human carbonic anhydrase in the cyanobacterium Synechococcus PCC7942 creates a high CO2-requiring phenotype: evidence for a central role for carboxysomes in the CO2 concentrating mechanism. Plant Physiol 91:505–513 Price NM, Harrison GI, Hering JG, Hudson RJ, Nirel PMV, Palenik B, Morel FMM (1988) Preparation and chemistry of the artificial algal culture medium Aquil. Biol Oceanogr 6:443–462 Price GD, Woodger FJ, Badger MR, Howitt SM, Tucker L (2004) Identification of a SulP-type bicarbonate transporter in marine cyanobacteria. Proc Natl Acad Sci USA 101:18228–18233 Raven JA, Smith FA (1978) Effect of temperature and external pH on the cytoplasmic pH of Chara corallina. J Exp Bot 29:853–866 Reinfelder JR (2011) Carbon concentrating mechanisms in eukaryotic marine phytoplankton. Ann Rev Mar Sci 3:291–315 Reinfelder JR, Kraepiel AML, Morel FMM (2000) Unicellular C4 photosynthesis in a marine diatom. Nature 407:996–999 Roberts K, Granum E, Leegood RC, Raven JA (2007) C3 and C4 pathways of photosynthetic carbon assimilation in marine diatoms are under genetic, not environmental, control. Plant Physiol 145:230–235 Romero MF, Boron WF (1999) Electrogenic Na?/HCO3- cotransporters: cloning and physiology. Ann Rev Physiol 61:699–723 Romero MF, Fulton CM, Boron WF (2004) The SLC4 family of HCO3- transporters. Pflug Arch Eur J Physiol 447:495–509 Rost B, Riebesell U, Burkhardt S, Su¨ltemeyer D (2003) Carbon acquisition of bloom-forming marine phytoplankton. Limnol Oceanogr 48:55–67 Rost B, Kranz SA, Richter KU, Tortell PD (2007) Isotope disequilibrium and mass spectrometric studies of inorganic carbon acquisition by phytoplankton. Limnol Oceanogr-Meth 5:328–337 Satoh D, Hiraoka Y, Colman B, Matsuda Y (2001) Physiological and molecular biological characterization of intracellular carbonic anhydrase from the marine diatom Phaeodactylum tricornutum. Plant Physiol 126:1459 Tachibana M, Allen AE, Kikutani S, Endo Y, Bowler C, Matsuda Y (2011) Localization of putative carbonic anhydrases in two marine diatoms, Phaeodactylum tricornutum and Thalassiosira pseudonana. Photosyn Res 109:205–221 Tanaka Y, Nakatsuma D, Harada H, Ishida M, Matsuda Y (2005) Localization of soluble b-carbonic anhydrase in the marine diatom Phaeodactylum tricornutum. Sorting to the chloroplast and cluster formation on the girdle lamellae. Plant Physiol 138:207–217 Tcherkez GBB, Farquhar GD, Andrews TJ (2006) Despite slow catalysis and confused substrate specificity, all ribulose bisphosphate carboxylases may be nearly perfectly optimized. Proc Natl Acad Sci USA 103:7246–7251 Whitney SM, Baldet P, Hudson GS, Andrews TJ (2001) Rubiscos from non-green algae are expressed abundantly but not assembled in tobacco chloroplasts. Plant J 26:535–547 Zeebe RE, Wolf-Gladrow D (2001) CO2 in seawater: equilibrium, kinetics, isotopes. Elsevier, Amsterdam Zhang HN, Byrne RH (1996) Spectrophotometric pH measurements of surface seawater at in situ conditions: absorbance and protonation behavior of thymol blue. Mar Chem 52:17–25

123