1

Bicarbonate-induced redox tuning in 1

Photosystem II for regulation and protection 2

Katharina Brinkerta, Sven De Causmaecker

a, Anja Kieger-Liszkay

b, Andrea Fantuzzi

a* and 3

A.William Rutherforda* 4

5

aDepartment of Life Sciences, Imperial College London, London SW7 2AZ, UK 6

bInstitute for Integrative Biology of the Cell (I2BC), CEA, CNRS, Univ Paris-Sud, Université 7

Paris-Saclay, F-91198, Gif-sur-Yvette, cedex, France 8

9

Running Title: Bicarbonate redox tuning of QA in Photosystem II 10

11

12

Supporting Information 13

14

1 Materials and methods 15

16

Chemicals: The redox mediators anthraquinone-2-sulfonate (Em = -195 mV, pH 6.5), 2-17

hydroxy-1,4-naphthoquinone (Em = -100 mV, pH 6.5) and N,N,N',N'-tetramethyl-p-18

phenylenediamine (TMPD, Em = +300 mV, pH 6.5) and additional chemicals were purchased 19

from Sigma Aldrich. 20

21

Spectroelectrochemical redox titrations 22

Spectroelectrochemical redox titrations of plant PSII-enriched membranes were carried out at 23

15 °C in an optically transparent thin-layer (spectro)electrochemical cell (optical path-length 24

0.1 mm, Als Co., Ltd.). The working electrode (100 mesh, gold gauze, Als Co., Ltd.) was 25

cleaned prior to each measurement by exposing it to an oxygen plasma (Emitech K1050X 26

Plasma Asher). The clean electrode was modified by incubation, under a stream of argon, in a 27

1 mM aqueous solution of 4,4’-dithiodipyridine (Sigma Aldrich) (1). The electrode was 28

carefully rinsed with water before use. The reference electrode was a Ag/AgCl (3 M KCl, 29

diameter 6 mm) electrode and the counter electrode was a platinum electrode (80 mesh, 30

platinum gauze, Als Co., Ltd.). The electrode potential was controlled by a potentiostat 31

2

(Metrohm/Eco Chemie Autolab potentiostat/galvanostat, electrochemical analyser, 32

PGSTAT12). Redox titrations were carried out in an anaerobic glove box using conditions 33

reported earlier by Shibamoto et al. (2, 3). The redox state of QA- was monitored by 34

measuring chlorophyll a fluorescence. Photosystem II was diluted to a [Chl] = 150 μg/mL in 35

a buffer containing 50 mM MES-NaOH (pH 6.5), 0.2 M KCl, 0.1 % dodecyl-β-D-maltoside, 36

1 M glycine-betaine and 1% taurine. A combination of redox mediators was added: 100 μM 37

anthraquinone-2-sulfonate, 100 μM 2-hydroxy-1,4-naphthoquinone and 200 µM TMPD. The 38

sample was kept in the electrochemical cell in complete darkness except when chlorophyll a 39

fluorescence was excited for ~5 s every ~10 min to monitor the change in fluorescence and to 40

evaluate the electrochemical equilibrium. The excitation was carried out with a DUAL-PAM-41

100 P700 & Chlorophyll Fluorescence Measuring System (Heinz WALZ GmbH), using the 42

pulse amplitude modulation (PAM) method and a very weak monochromatic beam (460 nm, 43

3 μE m-2

s-1

). The emission (> 625 nm, RG665 filter) was detected in line with the measuring 44

beam using a Photomultiplier-Detector Unit (DUAL-DPM). The fluorescence intensity 45

changes were plotted against the applied potential and the data fitted to the Nernst equation. 46

All of the potential are expressed versus SHE. 47

48

Influence of freeze-thawing on bicarbonate binding 49

With samples that were frozen and thawed titrations often became less reversible, in that the 50

return from reducing potentials showed indications of higher potential transitions. Good 51

reversible titrations were found with samples that had not been frozen and thawed. One 52

freeze-thaw cycle was apparently enough to decrease bicarbonate binding, while a second 53

freeze-thaw cycle led to relatively easy loss of the bicarbonate. In kinetic experiments the 54

addition of bicarbonate showed varying degrees of acceleration of the rate of forward 55

electron transfer in frozen and thawed samples. These effects where not characterised in a 56

systematic way, but the data shown in this work were done with sample that were either 57

unfrozen or had undergone one freeze-thaw cycle (where indicated). It is of note Shibamoto 58

et al (3) specified that their titrations required unfrozen samples. We suggest that this effect 59

could be due to changes in access to HCO3- binding site. 60

61

Connectivity determination 62

Due to antenna connectivity the exciton visiting a closed centre has a significant probability 63

of continuing its random walk and eventually hitting an open trap (4, 5). A consequence is 64

3

that the fluorescence yield F depends on the fraction of closed centres, c, in a hyperbolic 65

rather than linear way, according to the equation: 66

67

68

𝐹(𝑐) = 𝑐

1+𝐽−𝐽𝑐 (SE1) 69

70

71

The parameter J expresses the antenna connectivity. The non-linear relationship between the 72

fluorescent yield F and the concentration of reduced QA has consequences on the correct 73

determination of the fraction of reduced QA as a function of the redox potential. This effect 74

can be corrected using a calibration of the F(c) relationship, where c(t) can be determined 75

directly from the fluorescence (6). A weak sub-saturating flash that excites approximately 76

one sixth of the centres (fluorescence amplitude 15% of the Fmax with saturating flash) elicits 77

a fluorescence decay that can be taken as the true c(t) decay (6). 78

79

80

81

Figure S1 The fluorescence intensity elicited from a weak flash (wf) (5% of the Fmax) plotted as a 82

function of that using a saturated flash (sat). The data is fitted the equation SE1. (see text for details). 83

84

4

Thus the time course of the fluorescence decay was followed with a Joliot-type 85

spectrophotometer, JTS-10 from Biologique using i) a saturating and ii) a flash attenuated to 86

the point where it gave approximately 15% of the maximum fluorescence change measured 87

with the saturating flash. Figure S1 shows the plot of fluorescence with the weak flash (“real” 88

QA- concentration) as a function of the fluorescence seen on the saturating flash. Using 89

equation SE1 and substituting the measured values of fluorescence to obtain c, the mole 90

fraction of QA-, the resulting curve was fitted. A connectivity parameter J of 1.6 was obtained 91

and this was used to determine the concentration of QA- as a function of the electrode 92

potential in the redox titrations. In these experiments the saturating actinic flash (5 ns, 93

690 nm, 1.5 mJ) was provided by a dye laser (6 mM LDS 698 (CAS 89846-19-5) in DMSO), 94

using a Minidye accessory (GSI Group France) pumped by a frequency-doubled Nd:YAG 95

laser (532 nm; Minilite II, Continuum). The fluorescence emission was probed by weak 96

monochromatic flashes (450 nm) from the JTS-10. 97

98

Chlorophyll a fluorescence decay measurements. 99

The flash-induced increase and subsequent decay of chlorophyll fluorescence yield 100

were measured by a double fluorometer FL3000 (PSI Instruments, Brno, Czech Republic). A 101

sequence of 5 actinic flashes spaced by 1 second was used. The sample concentration was 5 102

μg Chl/ml. Ferricyanide at a final concentration of 250 μM was added to all samples prior to 103

the measurement in order maintain the plastoquinone pool in its oxidised form. 104

To verify HCO3--depletion according to the method (7), the sample (on ice under Ar), 105

was transferred in the dark with a gas-tight syringe into a sealed cuvette previously flushed 106

with Ar. Ferricyanide was added to the sample from a degassed and sealed stock to a final 107

concentration of 250 μM using a gas-tight syringe. The solution was allowed to equilibrate 108

for 1 minute before starting the flash sequence. Measurements were carried out at room 109

temperature (+20°C) in 40 mM MES pH 6.5, 15 mM CaCl2, 15 mM MgCl2. 110

To test the light-dependent dissociation of HCO3- the PSII sample was placed into a 111

sealed cuvette and degassed by bubbling argon for 15 minutes at +15 °C. The sample was 112

maintained at +15 °C and then illuminated for 4 minutes through a 590 nm cut-off filter and a 113

light intensity of 250 μmol photons m-2

s-1

(measured with a QRT-1 light meter from 114

Hansatech). The sample in the sealed cuvette was then transferred into the spectrofluorometer 115

where ferricyanide was added to the sample and the flash fluorescence measured were 116

performed as described in the previous paragraph. 117

118

5

Singlet oxygen measurements. 119

Singlet oxygen was trapped using the water soluble spin-probe 2,2,6,6-tetramethyl-4-120

piperidone (TEMPD) hydrochloride (11) and measured with a Bruker e-scanTM

(Bruker 121

Biospin, Rheinstetten, Germany). HCO3--depleted samples were diluted in non-degassed 122

buffer (0.3 M sucrose, 10 mM NaCl, 20 mM MES pH 6.5) containing 100 mM TEMPD and 123

shaken vigorously in room light to allow air to enter the solution and measured immediately. 124

Samples (10 µg Chl ml-1

) were illuminated for 2 min with 500 µmol quanta m-2

s-1

red light 125

(RG 630). 126

127

128

6

129

2 Additional Data 130

131

A) Redox titration data showing reconstitution of bicarbonate and recovery of the 132

lower potential Em values. 133

134

Fig. S2 Redox titration curves of QA in Mn-containing (A) and Mn-depleted (B) PSII 135

membranes in HCO3--depleted PSII (triangles) and (10mM) HCO3

- reconstituted PSII 136

(cirlces), measured as fluorescence yield. Filled symbols, oxidative titration, open symbols, 137

reductive titration. The concentration of reduced QA was determined by correcting the 138

fluorescence values for connectivity according to equation SE1 considering excitation energy 139

transfer between two PSII unit with J = 1.6. The data points were fitted to the Nernst equation 140

with n=1. The error bars indicate the standard deviation of three independent redox titrations. 141

The calculated mean Em values were -124 mV ± 3 mV and -60 mV ± 2 mV (A) and -22 mV ± 142

2 mV and +64 mV ± 3 mV (B), respectively 143

7

B) Redox titration data prior to correction for connectivity 144

145

i)Effect of depletion of depletion of bicarbonate on intact and Mn dpeleted PSII 146

147

148

149

Fig. S3 Redox titration curves of fluorescence yield levels uncorrected for connectivity in 150

intact (A) and Mn-depleted (B) PSII core complexes from PSII-membranes in the presence 151

(circles) and absence (triangles) of the HCO3- ligand measured as fluorescence yield. Filled 152

symbols, oxidative titration, open symbols, reductive titration. The data points were fitted to 153

the Nernst equation with n=1. The error bars indicate the standard deviation of three 154

independent redox titrations. The calculated mean Em values were -138 mV ± 2 mV and -61 155

mV ± 3 mV (A) and -23 mV ± 3 mV and +65 mV ± 3 mV (B), respectively. These are the 156

same data as shown in Figure 2A and B in the main text but prior to correction for 157

connectivity. 158

8

ii) Effect of bicabonate depeletion using formate 159

160

161

162

163

164

Fig. S4 Redox titration curves of fluorecence yield in intact PSII membranes in the presence 165

of the HCO3- ligand (circles) and upon replacement of the HCO3

- ligand by CHO2

- (triangles) 166

measured as fluorescence yield uncorrected for connectivity. Filled symbols, oxidative 167

titration, open symbols, reductive titration. The data points were fitted to the Nernst equation 168

with n=1. The error bars indicate the standard deviation of three independent redox titrations. 169

The calculated mean Em values were -138 mV ± 2 mV and -95 mV ± 2 mV, respectively. 170

171

9

iii) Effect of illumination on the redox titration 172

173

174

175

Fig. S5 Redox titration curves of fluorescence yield (uncorrected for connectivity) in intact 176

PSII membranes measured in the presence of the HCO3- ligand (circles and red triangle) in 177

the dark and and after illumination. The following protocol (triangles): a potential of -58 mV 178

was applied in the dark for 12 min and the fluorescence value recorded (red triangle), the 179

potential was then switched off for 10min in the dark, followed by QA reduction induced by 180

illuminating the sample for 4 min with a 620 nm actinic light (273 μEm-2

s-1

). The light was 181

switched off for 10 min to allow QA- to re-oxidize, after which a potential of -58 mV was 182

applied for 8 min in the dark and the fluorescence yield was measured. A reductive titration 183

was performed up to -150 mV followed by an oxidative titration. Filled symbols, oxidative 184

titration, open symbols, reductive titration. The data points were fitted to the Nernst equation 185

with n=1. The error bars indicate the standard deviation of three independent redox titrations. 186

The calculated mean Em values were -138 mV ± 2 mV and -61 mV ± 4 mV respectively. 187

188

189

190

191

10

iv) Effect of bicarbonate reconstitution on bicarbonate-depleted PSII 192

193

194

195

196

Fig. S6 Redox titration curves of fluorescence yield (uncorrected for connectivity) in Mn-197

containing (A) and Mn-depleted (B) PSII membranes in HCO3--depleted PSII (triangles) and 198

(10 mM) HCO3- reconstituted PSII (cirlces). Filled symbols, oxidative titration, open 199

symbols, reductive titration. The data points were fitted to the Nernst equation with n=1. The 200

error bars indicate the standard deviation of three independent redox titrations. The 201

calculated mean Em values were -126 mV ± 4 mV and -61 mV ± 3 mV (A) and -20 mV ± 3 202

mV and +65 mV ± 3 mV (B), respectively. 203

204

11

D) The thermodynamic cycle showing the relationship between Em and HCO3- binding. 205

206

207

208

Fig. S7 Thermodynamic cycle for the intact (A) and Mn-depleted (B) PSII linking QA/QA-· 209

reduction potential and HCO3- dissociation. The HCO3

- dissociation constant with QA 210

reduced (in red) was calculated based on the ΔΔG beween the reduction reaction with HCO3- 211

bound and that with HCO3- dissociated. 212

213

214

215

12

E) Results of EPR spin trapping experiments 216

217

218

219

sample EPR signal size

control 91%

91%

88%

Control + CO2 100%

-CO2 54%

59%

58%

-CO2 + 1 mM

NaHCO3

78%

99%

98%

220

221

Table S1 Results of the EPR spin-trapping measurements. Two different PSII enriched 222

membrane preps were used which were from a frozen sample. The control sample + 1 mM 223

NaHCO3 was set to 100%. 224

225

226

227

228

229

230

231

232

13

F) The Em values measured here compared to those discussed from the literature 233

234

235

Table S2 QA/QA- reduction potential values from spinach obtained in this work compared to 236

relevant literature values (Krieger et al 1995 (8); Shibamoto et al 2010 (2); Allakhverdiev et 237

al 2011 (9)). The four states of PSII are indicated by the colored squares for the data 238

presented here: intact with bicarbonate (blue), intact without bicarbonate (red), Mn-depleted 239

with bicarbonate (green) and Mn-depleted without bicarbonate (orange). The shading 240

represents the interpretation made in this work of the literature data. It is possible to do this 241

for the dozens of values in the literature but we chose to focus on the limited set of values 242

discussed in the main text. We consider the new values to have only a small error. Some 243

literature values have much greater error however the breadth of the shading does not 244

represent that error it is merely a means of illustrating the new interpretation of the data. 245

246

247

248

249

250

14

3 Additional Discussion 251

252

a) Relevance to donor-side effects of bicarbonate: further discussion 253

Of additional relevance to the donor-side debate is the observation made here that the 254

size of the QA/QA-• Em shift is greater in Mn-depleted PSII than in intact PSII. This seems to 255

indicate an influence of the donor-side on the acceptor-side. Donor-side effects on the 256

electron acceptor-side have been reported and discussed before (8, 10-14). In the present 257

studies we had the impression that when the HCO3- was lost, it was easier to lose the Mn 258

cluster. This needs to be verified by systematic study but it could fit with the reported 259

susceptibility to Mn loss that occurred in material that we now consider to have been depleted 260

of HCO3- (8,10,11). The bigger effect of HCO3

- binding on the Em of QA in Mn depleted PSII 261

and the consequent bigger shift in the dissociation constant of HCO3- when QA

- is present 262

could have specific importance during the process of photoassembly, where two QA redox 263

tuning events may occur, one involving Ca binding and the other bicarbonate binding. The 264

sequence and mechanistic significance of these events is worth investigating. 265

In a recent study Kahn et al (15) questioned whether Ca2+

/Sr2+

ion binding in the Mn 266

cluster at the donor-side of PSII is in fact responsible for the decrease in the Em of QA 267

described in the introduction. Instead they suggest that Ca2+

/Sr2+

ions bind to the glutamate 268

patch close to QA and that this is responsible for the changes in the Em of QA reported earlier 269

(15). This suggestion is contradicted by the data. Just as in earlier work, all the redox data 270

presented here, including those with the large positive shift in redox potential in PSII when 271

the Mn4O5Ca was removed, were obtained with identical cation concentrations and without 272

divalent cations. In some earlier work low potential values were obtained when reducing the 273

sample but high potential values were obtained in the same sample when reoxidising after 274

reduction because the Mn (and hence Ca) ions had been released due to reduction and loss of 275

the Mn cluster : thus the cation concentration barely changed under these conditions only the 276

integrity of the cluster changed (8). The redox titration work using Sr2+

was due to a single 277

Sr2+

ion per PSII shown spectroscopically to be located in the Ca2+

site in the Mn cluster and 278

the redox titrations comparing the influence of Sr2+

exchange were done in the presence of 279

identical concentrations of Ca2+

(13). It seems clear that the changes in Em described cannot 280

be attributed to changes in divalent cation binding to the electron acceptor side. Nevertheless, 281

given the proximity of the glutamate patch to QA (16), it does seem likely that binding of 282

cations to this site could influence the potential of QA. However, if this were the case the Em 283

15

of QA/QA- would be expected to increase (due to an electrostatic effect) rather than decrease, 284

and so far titration studies have not shown such effects. 285

286

b) On the relevance of the new model to published literature 287

It is worth considering if the new H3CO--mediated regulation mechanism presented here can 288

explain observations in the literature. Given the extensive literature, we provide only one 289

recent example. In recent work the effects of illumination regimes on CO2-limited 290

cyanobacterial cells were investigated in cyanobacteria under CO2 restricted conditions (17). 291

A series of observations related to photosynthetic electron transfer (Chl fluorescence, O2 292

evolution, cytochrome f, P700+, NADP

+) were reported leading the authors to postulate that 293

the over-reduced PQ pool prevents the cytochrome b6f complex from functioning after a 294

pulse of strong light. This reduced-pool mechanism of electron transfer inhibition could 295

indeed explain the phenomena reported, however the new HCO3--mediated regulatory 296

mechanism reported here could account for the observations made. Given the current model it 297

should be possible to distinguish the two mechanisms experimentally. 298

299

c) On the occurrence of the HCO3--mediated regulation of PSII in vivo 300

A wide range of strategies have evolved to deal with pressures associated with CO2 limitation 301

when encountered by different species in different environments. Several types of alternative 302

electron transfer mechanisms have evolved to alleviate problems encountered with 303

photosynthetic electron transfer under CO2-limiting conditions, including O2 reduction 304

(involving flavo-di-iron proteins, plastid alternative oxidase, photorespiration etc.) and cyclic 305

electron flow around PS I (e.g. 18-21). Similarly different mechanisms have evolved to 306

mitigate CO2 limitation, (CO2 concentrating mechanisms, CAM and C4 carbon fixation) and 307

these will have different influences on the stromal HCO3- concentrations (22-23). It thus 308

seems likely that the importance of HCO3--mediated regulation of PSII in vivo could vary 309

greatly depending on the species. These considerations will have to be taken into account in 310

future efforts to demonstrate its role in vivo. 311

312

References 313

1) Tanaguchi I, Toyosawa K., Yamaguchi H, Yasukouchi K (1982) Voltammetric response to 314

horse heart cytochrome c at a gold electrode in the presence of a sulfur-bridged bipyridine. J 315

Electroanal Chem 140:187-193. 316

16

2) Shibamoto T, Kato Y, Nagao R, Yamazaki T, Tomo T, Watanabe T (2010) Species-317

dependence of the redox potential of the primary quinone electron acceptor QA in 318

photosystem II verified by spectroelectrochemistry. FEBS Lett, 584:1526-1530. 319

3) Shibamoto T, Kato Y, Sugiura M, Watanabe T (2009) Redox potential of the primary 320

plastoquinone electron acceptor QA in photosystem II from Thermosynechococcus elongatus 321

determined by spectroelectrochemistry. Biochemistry, 48:10682-10684. 322

4) Joliot P, Joliot A (1964) Etude cinétique de la reaction photochimique libérant l’oxygène 323

au cours de la photosynthèse. C.R. Hebd Seances Acad Sco Paris 258:4622-4625. 324

5) Lavergne J, Trissl H-W (1995) Theory of fluorescence induction in photosystem II: 325

derivation of analyical expressions in a model including exciton-radical-pair equilibrium and 326

restricted energy transfer between photosynthetic units. Biophys. J. 68:2474-2492. 327

6) Cuni A, Xiong L, Sayre R, Rappaport F, Lavergne J (2004) Modification of the 328

pheophytin midpoint potential in photosystem II: modulation of the quantum yield of charge 329

separation and of charge recombination pathways. Phys. Chem. Chem. Phys 6:4825-4831. 330

7) Shevela D, Klimov V, Messinger J (2007) Interaction of photosystem II with bicarbonate, 331

formate and acetate. Photosynth Res 94:247-264 332

8) Krieger A, Rutherford AW, Johnson GN (1995) On the determination of the redox 333

midpoint potential of the primary quinone electron acceptor, QA, in Photosystem II Biochim. 334

Biophys Acta 1229:193-201. 335

9) Allakhverdiev S I, Tsuchiya T, Watabe K, Kojima A, Los DA, Tomo T, Klimov V, 336

Mimuro M (2011) Redox potentials of primary electron acceptor quinone molecule QA- and 337

conserved energetics of photosystem II in cyanobacteria with chlorophyll a and chlorophyll 338

d. Proc Natl Acad Sci USA 108:8054-8058. 339

10) Cardona T, Sedoud A, Cox N, and Rutherford AW (2012) Charge separation in 340

Photosystem II: a comparative and evolutionary overview. Biochim Biophys Acta 1817: 26-341

43. 342

11) Johnson GN, Rutherford AW, Krieger A (1995) A change in the midpoint potential of QA 343

in Photosystem II associated with photoactivation of oxygen evolution. Biochim. Biophys. 344

Acta 1229: 202-207. 345

12) Krieger A, Weis E, Demeter S (1993) Low-pH-induced Ca2+

ion release in the water-346

splitting system is accompanied by a shift in the midpoint redox potential of the primary 347

quinone acceptor-QA. Biochim Biophys Acta 1144: 411-418. 348

13) Kato Y, Shibamoto T, Yamamoto S, Watanabe T, Ishida N, Sugiura M, Rappaport F, 349

Boussac, A (2012) Influence of the PsbA1/PsbA3, Ca2+

/Sr2+

and Cl-/Br

- exchanges on the redox 350

17

potential of the primary quinone QA in Photosystem II from Thermosynechococcus elongatus 351

as revealed by spectroelectrochemistry Biochim. Biophys. Acta 1817:1998-2004. 352

14) Kato Y, Noguchi T (2014) Long-Range interaction between the Mn4CaO5 cluster and the 353

non-heme iron center in Photosystem II as revealed by FTIR spectroelectrochemistry 354

Biochemistry 53:4914-4923 355

15) Khan S, Sun JS, Brudvig GW 2015 Cation effects on the electron acceptor side of 356

Photosystem II. J Phys Chem B 119, 7722-7728 357

16) Umena Y, Kawakami K, Shen JR, Kamiya N (2011) Crystal structure of oxygen-evolving 358

Photosystem II at a resolution of 1.9 Å. Nature 473:55-60. 359

17) Shaku K, Shimakawa G, Hashiguchi M, Miyake C (2015) Reduction-induced suppression 360

of electron flow (RISE) in the photosynthetic electron transport system of Synechococcus 361

elongatus PCC 7942. Plant Cell Physiol 57, 1443-1453 362

18) Mullineaux CW (2014) Co-existence of photosynthetic and respiratory activities in 363

cyanobacterial thylakoid membranes. Biochim. Biophys. Acta 1837: 503–511. 364

19) Allahverdiyeva Y et al (2013) Flavodiiron proteins Flv1 and Flv3 enable cyanobacterial 365

growth and photosynthesis under fluctuating light. Proc. Natl. Acad. Sci. USA 110: 4111–366

4116. 367

20) Helman Y et al (2003) Genes encoding A-type flavoproteins are essential for 368

photoreduction of O2 in cyanobacteria. Curr. Biol. 13: 230–235. 369

21) Trouillard M et al (2012) Kinetic properties and physiological role of the plastoquinone 370

terminal oxidase (PTOX) in a vascular plant. Biochim Biophys Acta 1817:2140–2148 371

22) Wang Y, Stressman D.J. and Spalding MH (2015) The CO2 concentrating mechanism 372

and photosynthetic carbon assimilation in limiting CO2: how Chlamydomonas works against 373

the gradient The Plant Journal 82, 429–448 374

23) Sage, R. F. (2013) Photorespiratory compensation: a driver for biological diversity Plant 375

Biology 15, 624-638 376