A low CO2-responsive mutant of Setaria viridis reveals that reduced carbonic anhydrase limits C4 photosynthesis

Abstract In C4 species, β-carbonic anhydrase (CA), localized to the cytosol of the mesophyll cells, accelerates the interconversion of CO2 to HCO3–, the substrate used by phosphoenolpyruvate carboxylase (PEPC) in the first step of C4 photosynthesis. Here we describe the identification and characterization of low CO2-responsive mutant 1 (lcr1) isolated from an N-nitroso-N-methylurea- (NMU) treated Setaria viridis mutant population. Forward genetic investigation revealed that the mutated gene Sevir.5G247800 of lcr1 possessed a single nucleotide transition from cytosine to thymine in a β-CA gene causing an amino acid change from leucine to phenylalanine. This resulted in severe reduction in growth and photosynthesis in the mutant. Both the CO2 compensation point and carbon isotope discrimination values of the mutant were significantly increased. Growth of the mutants was stunted when grown under ambient pCO2 but recovered at elevated pCO2. Further bioinformatics analyses revealed that the mutation has led to functional changes in one of the conserved residues of the protein, situated near the catalytic site. CA transcript accumulation in the mutant was 80% lower, CA protein accumulation 30% lower, and CA activity ~98% lower compared with the wild type. Changes in the abundance of other primary C4 pathway enzymes were observed; accumulation of PEPC protein was significantly increased and accumulation of malate dehydrogenase and malic enzyme decreased. The reduction of CA protein activity and abundance in lcr1 restricts the supply of bicarbonate to PEPC, limiting C4 photosynthesis and growth. This study establishes Sevir.5G247800 as the major CA allele in Setaria for C4 photosynthesis and provides important insights into the function of CA in C4 photosynthesis that would be required to generate a rice plant with a functional C4 biochemical pathway.


Introduction
C 4 plants have evolved a combination of anatomical and biochemical specializations that concentrate CO 2 around Rubisco (Hatch, 1987).In C 4 leaves, specialized mesophyll cells (MCs) and bundle sheath cells (BSCs) support a biochemical CO 2 pump (von Caemmerer and Furbank, 2003) where each molecule of CO 2 entering the cytosol of the MCs is converted to bicarbonate (HCO 3 -) by the activity of carbonic anhydrase (CA).This is then incorporated into phosphoenolpyruvate (PEP) by PEP carboxylase (PEPC), yielding the C 4 acid oxaloacetate (OAA).In most agriculturally important C 4 species, the OAA is taken up into the chloroplast of the MCs where it is reduced to malate by NADP-dependent malate dehydrogenase (NADP-MDH).Malate is exported back to the cytosol and then diffuses into the BSCs.It is then transported into the chloroplast and oxidatively decarboxylated by NADP-dependent malic enzyme (NADP-ME), yielding CO 2 , NADPH, and pyruvate.CO 2 is assimilated by Rubisco and pyruvate is transported into the chloroplast of the MCs where it is converted to PEP by pyruvate phosphate dikinase (PPDK).This pump can elevate CO 2 in this compartment by up to 10-fold above air levels (Furbank and Hatch, 1987;von Caemmerer and Furbank, 2003), suppressing photorespiration and CO 2 saturating photosynthesis in air.
The efficiencies of C 4 photosynthesis have long driven interest in engineering a C 4 photosynthetic pathway into C 3 crop species (Matsuoka et al., 2001;Häusler et al., 2002;Miyao, 2003;Kajala et al., 2011;Miyao et al., 2011).In rice, this could potentially lead to increases in radiation use efficiency and yield of up to 50% (Hibberd et al. 2008).The C 4 Rice Consortium (https://c4rice.com) is currently investigating the feasibility of engineering the leaf anatomy and biochemistry required to support a two-celled photosynthetic pathway (Kajala et al., 2011;Lin et al., 2016).To help achieve this, a toolkit of genes is being assembled.Unknown genes are being identified through forward genetics, whereby large populations of C 4 plants are mutagenized and then screened for a loss of C 4 photosynthetic characteristics and function (Furbank et al., 2009).This approach has already led to the identification of some promising mutants with altered leaf anatomy (Rizal et al., 2015(Rizal et al., , 2017)).However, there remains much which is unknown about the efficiencies, genes, and mechanisms underpinning the C 4 photosynthetic process.Among these are the genetic components modulating photosynthetic cellular specialization (suberization of the BSC wall and positioning of the chloroplast), localization of C 3 and C 4 photosynthetic enzymes (C 4 biochemistry, the Calvin cycle, and the photorespiratory cycle), the identity of transporters supporting C 4 metabolic flux, and the regulators of C 4 gene expression.Knowledge of these processes is an essential requirement for engineering a functional C 4 photosynthetic pathway into rice.
In an earlier study, we described the development of highthroughput screening for identifying CO 2 -responsive mutants of the C 4 monocot model species Setaria viridis (L.) P. Beauv.(Coe et al., 2018).The mutant population was generated using N-nitroso-N-methylurea (NMU).There, we could identify 46 candidate mutant lines with reduced F v /F m relative to wildtype (WT) plants under conditions of low CO 2 partial pressure (pCO 2 ) at 15 µbar (0.0015%).This threshold level was determined considering the difference in the CO 2 compensation point in C 3 and C 4 species.Reduction in F v /F m with an increased CO 2 compensation point is used as a signature for decreased photosynthetic efficiency and a loss of C 4 function.
In 1990, Hatch and Burnell (1990) demonstrated that in vivo CA activities in C 4 leaves were only just sufficient to support rates of PEP carboxylation.Since then, it has been debated whether high CA activities are a requirement for evolution of C 4 photosynthesis.The enzyme is a part of a multiple gene family, where one gene may give rise to multiple CA isoforms via alternative splicing of the coding sequence (CDS).Functional evaluation of any of these genes gives different phenotypic expression in different species, which makes it difficult to interpret its role unambiguously.Whether the CA activity in C 4 leaves has been reduced by transgenesis or other genetic means; conflicting reports have suggested either a large inhibition of photosynthesis (von Caemmerer et al., 2004) or that little or no effect was observed (Studer et al., 2014;Osborn et al., 2017).In this study, by adopting a forward genetics approach, we identify the CA allele responsible for the carbon-concentrating mechanism (CCM) in the C 4 model species S. viridis to be the underlying causal gene of the low CO 2 -responsive mutant lcr1 (NM04534 in Coe et al., 2018) using next-generation sequencing and mapping.We study in detail the role of CA in C 4 photosynthesis, plant growth, and isotope discrimination in S. viridis.This work provides insights into the function and genetic regulation of C 4 photosynthesis.

Plant material
A low CO 2 -responsive mutant (NM04534, lcr1) was identified from a population of S. viridis A10.1 NMU mutants as described in Coe et al. (2018).The mutant was identified based on its reduced F v /F m values in ambient (390 µbar pCO 2 ) and in low CO 2 conditions (15 µbar pCO 2 ).The F v /F m values of the successive generations of lcr1 were checked in ambient and low CO 2 conditions using a PlantScreen Compact System (Photon Systems Instruments) machine with replicated progenies to finally achieve a stable phenotype in the M 5 generation.A single M 5 mutant line (progeny number 23-50) was selected and advanced for further characterization and gene identification.The mutant is characterized for its photosynthetic, leaf anatomy, and biochemical characters mostly on the plants of the M 5 and M 6 generation, except the F v /F m value, which was measured throughout as a marker to select the true candidate mutants.The seeds of S. viridis A10.1 were obtained from Dr Thomas Brutnell (Danforth Plant Science Center), and the rice seeds were obtained from the genebank of the International Rice Research Institute (IRRI), Philippines.
Seed dormancy was broken by soaking in 5% liquid smoke (Wright's Liquid Smoke, B&G Foods Inc., Roseland, NJ, USA) for 24 h; they were then thoroughly rinsed with water and sown on moist filter paper.When the hypocotyl was ~5 mm long, seedlings were transferred to 0.5 litre pots containing sterilized soil from the IRRI upland farm mixed with 0.4 g l -1 Osmocote Plus 15-9-12 (The Scotts Company Ltd, Thorne UK).Plants were cultivated in a greenhouse at ambient CO 2 or inside custom-made growth chambers at different controlled CO 2 concentrations, as mentioned above.Irradiance in the greenhouse ranged between 500 µmol m -2 s -1 and 2000 µmol m -2 s -1 with a day length of between 11 h and 13 h.Day and night temperature ranged between 21 °C and 34 °C, with a relative humidity of between 60% and 70%.Inside the growth chambers, ~63% of the ambient solar irradiance is transmitted at the canopy level; the air temperature was maintained at ~30 °C and a relative humidity of 60-70%.
At the ANU, seeds were germinated in garden soil mix fertilized with Osmocote (Scotts, Australia) in small containers before being transferred to individual 2 litre pots.Plants were always grown in controlledenvironment chambers at an elevated CO 2 level 2 (i.e. 2% CO 2 ), irradiance 500 µmol photons m -2 s -1 , 16 h photoperiod, and day and night temperature of 28 °C and 24 °C, respectively.Pots were watered daily.

Chlorophyll fluorescence imaging
Chlorophyll fluorescence images were acquired with the PlantScreen™ Compact System (Photon System Instruments) as described in detail in the previous report by Coe et al. (2018).Images were analysed using Fluorcam 7 v1.024.2 (Photon Systems Instruments), and fluorescence parameters were calculated in the software according to Genty et al. (1989).Plants are imaged for F v /F m after 30 min of dark adaptation, and data were used for candidate selection, checking heritability of the low F v /F m character in the mutant, confirmation of BC 1 F 1 , and BC 1 F 2 population screening.The maximum quantum efficiency of PSII photochemistry (F v /F m ), which gives a good understanding of a plant's photosynthetic efficiency, was calculated as: (F m -F o )/F m .Detailed kinetic studies were performed on the M 5 generation mutant plants and the WT.Measurements of rapid chlorophyll fluorescence kinetics were made following acclimation for 20 s at six different actinic light intensities between 25 µmol m -2 s -1 and 1000 µmol m -2 s -1 .Plants were characterized for the maximum efficiency of PSII in the light if all centres were open (F v ′/F m ′), (F m ′-F o ′)/F m ′; PSII operating efficiency (ΦPSII), (F m ′-F′/F m ′); and photochemical (qP) and non-photochemical quenching (NPQ).qP (F m ′-F′)/(F m ′-F o ′), relates PSII maximum efficiency to operating efficiency; and NPQ [(F m /F m ′)-1] estimates the rate constant for heat loss from PSII.F and F′ are the fluorescence emission from a dark-or lightadapted leaf; F o and F o ′ are the minimal fluorescence from a dark-and light-adapted leaf; F m and F m ′ are maximal fluorescence from a dark-and light-adapted leaf, respectively; and F v and F v ′ are the variable fluorescence from a dark-and light-adapted leaf.Details of the method are given in Coe et al. (2018).Fluorescence values represent the median of fluorescence values obtained for each plant and nine plants per line.

Plant biomass
The number of tillers was counted and plant height was measured at the maximum tillering stage taking measurements from the soil surface to the tip of the youngest fully expanded leaf prior to destructive biomass measurements.All above-ground biomass (leaves, stems, and sheaths) was harvested, weighed, placed in papers bags, and oven-dried at 70 °C until a constant dry biomass weight was achieved.Leaf chlorophyll content was estimated at the mid-tillering stage using the upper fully expanded leaves employing a SPAD Chlorophyll Meter (SPAD, Konica Minolta, Japan).Values given are the average ±SD of 10 plants per line.

Leaf microscopy
Leaf anatomy measurements were scored from the three cleared transverse leaf section images on the middle portion of one leaf each taken from five plants per line.Cleared leaf sections and fresh leaf sections are prepared and imaged as described in Chatterjee et al. (2016).The sections were imaged by an Olympus BX51 compound microscope using bright field for cleared sections and an Olympus BX52 disc spinning fluorescent microscope for fresh sections.The images were acquired with an Olympus DP71 camera.These were analysed with Olympus CellSens software (http://www.olympus-lifescience.com) and ImageJ software v.1.43(Schindelin et al. 2015) to calculate leaf thickness (µm), BSC (µm 2 ), MC (µm 2 ), and epidermal cell (µm 2 ) area, and area of the chloroplast in BSC area (µm 2 ).

Stomatal measurement
Stomatal measurements were performed on a total of five leaves taken from five plants per line.Leaf nail varnished images of abaxial imprints were used to score stomatal density and stomatal size under ×20 and ×40 magnification with the BX51 Olympus bright field microscope.Stomatal density was counted from 25 random observations and the length and width were determined from observation of 25 random stomata per line.

Photosynthetic measurements
Gas exchange measurements were performed on M 6 generation plants.Leaf gas exchange measurements were made using a Li-6400XT infrared gas analyzer (LI-COR Biosciences) on the plants grown at ambient or in 1% CO 2 (elevated CO 2 level 1).The cuvette was fitted with a standard 2×3 cm leaf chamber and a 6400-02B light source.Measurements were made after the leaves were acclimated in the cuvette for ~30 min, with a constant air flow rate of 400 μmol s -1 , leaf temperature of 30 °C, leaf to air vapour pressure deficit between 1.0 kPa and 1.5 kPa, and relative humidity of 60-65%.Data were acquired between 08.00 h and 13.00 h.Measurements were made on at least three plants during the tillering stage on the mid-portion between the proximal and distal end of a fully expanded leaf.The response curves of the net CO 2 assimilation rate (A, µmol m -2 s -1 ) to changing intercellular pCO 2 (C i , µbar) were acquired by first stabilizing the plants in 400 µbar pCO 2 and then increasing the C a (pCO 2 in the cuvette) stepwise from 20 µbar to 2000 µbar at a photosynthetic photon flux density (PPFD) of 1800 µmol photon m -2 s -1 at an appropriate O 2 level (2, 21, 40, 60, and 80% O 2 ).The CO 2 compensation point (Γ) and maximum carboxylation efficiency (CE) were calculated from the intercept (Vogan et al., 2007) and slope (Wang et al., 2006) of the CO 2 -response curves.Light-response curves were acquired by increasing the PPFD from 0 to 2000 µmol photon m -2 s -1 at a C a of 400 µbar.The quantum efficiency for CO 2 assimilation (Φ) was calculated from the slope of the light-response curves (PPFD <100 μmol photons m -2 s -1 ).Dark respiration rate (R d ) measurements were calculated at a C a of 400 μbar.

Gas exchange and isotopic discrimination
Carbon isotope discrimination was determined by combusting dry leaf tissue utilizing a Eurovector elemental analyser using plants grown in ambient pCO 2 at IRRI.The three youngest fully expanded leaves on the main stem of three different 25 days-old-plants were sampled and dried in an oven at 70 °C, ground into a fine powder (Precellys 24, Bertin Instruments, France), and 1.1-1.3mg of leaf sample was sent to the Analytical Service Laboratory at the IRRI for quantification of dry carbon isotope.The ratio of 13 C to 12 C (δ) is reported relative to the Pee Dee Belemnite standard (Craig, 1957).Based on the equation from Farquhar et al. (1989), the carbon isotope composition (δ p ) was calculated as: where, R S is the molar abundance ratio ( 13 C/ 12 C) of the PDB standard, and R P is the molar abundance ratio ( 13 C/ 12 C) of a plant sample.Carbon isotope discrimination (∆, ‰) was calculated as: where R a is the molar abundance ratio ( 13 C/ 12 C) in air, and δ a is the ratio of 13 C/ 12 C of free atmospheric CO 2 on the PDB scale and is approximately -8‰ according to the Earth Systems Research Laboratory.Isotopic discrimination was also measured in combination with gas exchange measurement using tunable diode laser (TDL) spectroscopy as described by Evans and von Caemmerer (2013) for 13 C isotope discrimination and as described by Osborn et al. (2017) for C 18 O 16 O discrimination.Gas exchange and carbon isotope discrimination measurements were made using a 6 cm 2 chamber of the LI-6400 with a red/blue lightemitting diode (LED) light source (Li-Cor, Lincoln, NE, USA).Two LI-6400 chambers and the plants were placed in a temperature-controlled cabinet with fluorescent lights (TRIL1175, Thermoline Scientific Equipment, Smithfield, Australia).The CO 2 in the leaf chamber was set at 380 µmol mol -1 , flow rate at 200 µmol s -1 , and irradiance at 1500 µmol quanta m -2 s -1 .Leaf temperature was controlled at 25 °C.O 2 [2% in N 2 , mixed by mass flow controllers (Omega Engineering Inc., Stamford, CT, USA)], was supplied to the LI-6400s after humidification of the air by adjusting the temperature of water circulating around a Nafion tube (Perma Pure LLC, Toms River, NJ, USA, MH-110-12P-4).For 13 C discrimination measurements, the gas exchange was coupled to a TDL (TGA100a, Campbell Scientific, Inc., Logan, UT, USA).Measurements were made at 4 min intervals for 20 s, with 10-12 measurements per leaf, and the last five measurements were averaged.The measurements of C 18 O 16 O discrimination were made as described by (Osborn et al., 2017).Simultaneous measurements of CO 2 , H 2 O, C 18 O 16 O, and H 2 18 O were made by coupling two LI-6400XT gas exchange systems to a TDL (model TGA200A, Campbell Scientific Inc.) and a cavity ring-down spectrometer (L2130-i, Picarro Inc., Sunnyvale, CA, USA) to measure the oxygen isotope composition of water vapour.The system is essentially that described above except that the TGA100a was replaced by a TGA200A and the additional laser for water vapour measurements was added together with a 16 port distribution manifold.The sample and reference gas streams were sampled with a T junction in the match valve tubing and in the reference line of the LI-6400XT, respectively.This allowed leaves of two plants to be measured in sequence, with each LI-6400XT sampled by the TDL at 4 min intervals for 20 s at the sample and reference line.The Picarro cavity ring-down spectrometer sampled for 3 min, so that leaves were sampled at 6 min intervals.Gas exchange was calculated using the equations presented by von Caemmerer and Farquhar (1981), and ∆ was calculated from the equation presented by Evans et al. (1986).The measurements were performed on 5-week-old plants grown at 2% CO 2 at the ANU.Measurements were made at 2% O 2 , 380 µmol mol -1 CO 2 , leaf temperature of 25 °C, irradiance of 1500 µmol quanta m -2 s -1 , and relative humidity of 55%.Each leaf was measured at 4 min intervals, and 10 readings were taken.

Generation of the segregating population and whole-genome sequencing of bulked segregants
The M 5 progeny of lcr1 was crossed with WT plants, the latter as pollen donor, to obtain the BC 1 F 1 generation (following the protocol of Jiang et al., 2013).BC 1 F 1 lines were checked for a recovery of the F v /F m value in true F 1 , and then selfed to produce the BC 1 F 2 segregating population, which was used for whole-genome sequencing.From a BC 1 F 2 population of 300 plants, equal quantities of genomic DNA were pooled from 45 individuals exhibiting the lcr phenotype (homozygous mutant pool) and 54 individuals without the lcr phenotype (azygous pool).Total genomic DNA was extracted from young leaves of a mid-tillering plant using the CTAB (cetyltrimethylammonium bromide) protocol (Murray and Thompson, 1980).Genomic DNA was also pooled from 50 WT plants (WT pool).These pooled samples were sequenced using Illumina HiSeq2000 platforms at Beijing Genome Institute BGI Tech Solutions (Hongkong) Co. Ltd, Shenzhen, China.The number of paired-end reads generated of size 125 bp was ~108.1 million for the WT pool, ~91 million for the azygous pool, and ~107.3 millions for the mutant pool, yielding 23-33 Gb of sequence data with a genome coverage of 45-65x for each.

Candidate gene discovery
The raw sequence data were processed so that reads with poor quality bases were either trimmed or filtered using Trimmomatic v0.32 (Bolger et al., 2014).The minimum Phred quality score was set to 20, and reads containing partial/complete Illumina adaptor sequences were removed.The processed reads were aligned against the S. viridis reference genome (version 1.1 from the Phytozome database) using the BWA aligner (version 0.7.12;Li, 2013, Preprint).The alignment was improved by discarding the duplicate read pairs using the PICARD tool (https://broadinstitute. github.io/picard/),and the base quality was re-calibrated with the help of the highest quality variants, present within the sequence data itself, as a gold standard variant set, using the Genome Analysis Tool Kit (GATK-3.7-0;McKenna et al., 2010).Variant calling was also done by using the same tool such that alleles were first called at all genomic positions, followed by variant calling (DePristo et al., 2011).
The variants with an allele frequency of ≤0.3 were assumed to result from incorrect alignment of minority reads, and were thus removed from the analysis.Only variants occurring with support of at least 10 reads, and a genotype quality (GQ) of ≥30, at least in the mutant pool, were considered for downstream analysis.Variant(s) associated with the mutant phenotype are likely to be common to all individuals of the mutant pool (yielding a mutant allele frequency of 1); however, such variants will either be absent or in a heterozygous state in the individuals of the azygous pool (yielding a mutant allele frequency of ~0.33).Such variants (mutant pool) are also likely to show a typical linkage pattern with the neighbouring variants in the mutant pooled sample.Therefore, genomic loci harbouring such variants were extracted from the mutant pool and were considered as candidates.
The candidate mutations were annotated using the S. viridis genome annotation available at the Phytozome database (release 12; https:// phytozome.jgi.doe.gov/pz/portal.html).Mutations located within a gene leading to an alteration to protein sequence became the basis to identify candidate genes.

Functional characterization of amino acid substitution by sequence conservation and protein structural analysis
In order to examine the effect of Leu156Phe substitution, the amino acid conservation profile was first obtained.Members of the 'Carbonic anhydrase 2' gene family from Angiosperms were obtained from the Phytozome database (https://phytozome.jgi.doe.gov),available under the 'Gene ancestry' tab of the Sevir.5G247800information page.Using the Biomart tool of Phytozome, 226 peptide sequences were extracted and multiple-aligned using the MAAFT tool (version 7; https://mafft.cbrc.jp/alignment/server/;Mode: L-INS-i).Fifteen sequences, which appeared partial/incomplete, were dropped and the fully conserved residues were finally identified from the multiple alignment itself.
The mechanism of β-CA function in plants had been thoroughly examined in the β-CA protein of Pisum sativum by Kimber and Pai (2000), wherein the residues were identified for Zn binding, active site, catalytic cleft formation, non-catalytic bicarbonate-binding pocket, and solvent access.To further explore the role of the residues in β-CA function, and loss of function due to Leu156Phe substitution, structural models were generated and analysed.The native and mutant peptide sequence of Sevir.5G247800 was given as input to the Rosetta tool to build 3-D protein structures using the comparative modelling approach (Kim et al., 2004).Rosetta provided five models with a confidence score ranging from 0.0 to 1.0, where a confidence score of 0.0 is indicative of a bad model and 1.0 of a perfect model (Song et al., 2013).
Among the five models, the structure present in the same conformation as in Type I CA (based on the classification of Rowlett, 2010) was selected for further analysis of (i) contact order (CO) calculation; (ii) estimation of long-range interacting residues by identifying the primary, secondary, and tertiary shell of residues; and (iii) analysing the dimer interface of native and mutant proteins.The CO quantifies the global protein topology and correlates with protein folding rates (Plaxco et al., 1998); thus, by comparing intraprotein interactions of the native and mutant proteins, the stabilizing forces can be quantified.The higher the CO value, the higher the stabilizing factor of the interaction type (Rader et al., 2012).
As a complementary method of identifying short-and long-range interactions, inspiration came from electron shells surrounding the nucleus.In the case of proteins, residue 156 was considered as the nucleus, and the residues interacting with 156 (Leu or Phe) were considered as the first shell of interacting residues; the residues that interact with these first shell of interacting residues were considered as the second shell of interacting residues, and similarly the residues that interact with the second shell of interacting residues were identified as the third shell of interacting residues (Drawz et al., 2009;Brodkin et al., 2011Brodkin et al., , 2015) ) In each shell, redundancy was removed and directionality of moving away from the nucleus was maintained.
The third analysis involved determining the residues interacting between the monomers of Sevir.5G247800.This directly established the various unique interactions in native and mutant proteins that stabilize the dimer stoichiometry of Type I CA.This analysis was performed on three protein structures; the homodimer of the native, the homodimer of the mutant, and a heterodimer made of β-CA isoforms (Sevir.5G247800and Sevir.5G247900.1).This was done so that an isoform was hypothesized to interact with other isoforms, leading to re-establishment of the lost function due to mutation at position 156.

Leaf protein and western blot
Progenies of the M 6 generation were used for protein analysis.Leaf samples were harvested between 09.00 h and 11.00 h from the fourth fully expanded leaf of plants grown at ambient CO 2 conditions at IRRI.Proteins were extracted by homogenizing leaf material in 250 µl of buffer containing 100 mM Bicine-KOH, pH 9.8, and 25 mM DTT (Lin et al., 2016).Proteins were fractionated by 12% (w/v) SDS-PAGE with Precision Plus Protein™ Unstained Standards (BIORAD, USA) used as the loading control.Samples were loaded based on equal leaf weight [34 µg for PEPC and Rubisco large subunit (LSU), 136 µg for PPDK, and 340 µg for MDH, ME, and CA].After electrophoresis, proteins were electroblotted onto a polyvinylidene difluoride membrane and probed with rabbit antisera containing antibodies made against PEPC protein, MDH protein, ME protein, Rubisco LSU protein (all provided by Richard Leegood, University of Sheffield, UK), PPDK protein (provide by Chris Chastain, Minnesota State University, USA), and CA protein (provided by Jim Burnell, James Cook University, Australia).The dilutions of PEPC, PPDK, MDH, ME, Rubisco LSU, and CA antibodies were 1:20 000, 1:20 000, 1:5000, 1:5000, 1:30 000, and 1:5000, respectively.A peroxidase-conjugated secondary antibody (Sigma Aldrich, USA) was used at a dilution of 1:5000, and immunoreactive bands were visualized with ECL Western Blotting Detection Reagents (GE Healthcare, UK).Total soluble leaf protein content was determined using the Bradford method (Bradford, 1976).

Enzyme activity
Enzyme activities were measured on 5-week-old plants that were grown in 2% CO 2 conditions (elevated CO 2 level 2).A high CO 2 concentration was used to obtain sufficient healthy leaves of mutants to perform the assays.The activities of CA, Rubisco, and PEPC were determined as described by Osborn et al. (2017).

Statistical analysis
The significant differences in the results of the WT and the mutant were obtained by statistical analysis performed in R version 3.0.0(The R Foundation for Statistical Computing, Vienna, Austria) or OriginPro 9.1 (OriginLab Corporation) using a one-way ANOVA and a least significant difference test or a two-sample t-test.

Identification of the Setaria mutant responsive to low pCO 2
The lcr1 mutant was identified from a screening of the F v /F m response to low CO 2 of 7900 NMU mutant lines.A cut-off value of 0.6 QY of PSII (F v /F m ) was set to identify the low CO 2 mutants, as we previously established that the F v /F m value decreased to 0.6 in WT Setaria after 48 h of low CO 2 treatment (Coe et al., 2018).Initially two individuals of M 3 progeny of the mutant line were identified with a reduced F v /F m relative to WT plants at low pCO 2 (Fig. 1A).Its progeny in M 4 inherited mutant-like F v /F m values (<0.6) after the low CO 2 treatment (Fig. 1B).By the M 5 generation, a stable F v /F m response to low pCO 2 was observed (Fig. 1C, D), which was very distinct from the WT response and always in the range of the mutant category.Two of the M 5 lines were studied further for detailed characterization.The lcr1 mutant exhibited heritable reductions of ~50% in F v /F m after low pCO 2 treatment for 48 h (Fig. 2).This response was significantly different from that of the WT, where the values were only 25% less after the treatment.The F v /F m values of lcr1 were only slightly reduced under normal pCO 2 conditions.lcr1 showed reduction in overall growth at ambient CO 2 , but recovered at high CO 2When grown at ambient pCO 2 , the mutant plants were significantly shorter, had fewer tillers and panicles, and accumulated less fresh and dry biomass (Fig. 3A; Table 1).Plants were visually pale, and leaves were significantly thinner and had lower MC, BSC, and adaxial cell areas (Fig. 3B; Table 1).The chloroplast area in the BSCs was approximately half that of WT plants.However, when grown at elevated pCO 2 (level 1), the mutant was almost indistinguishable from WT plants (Fig. 3A) and defects in leaf anatomy recovered significantly (Table 1; Supplementary Fig. S1).Plants grown in elevated CO 2 level 2 also did not show any growth abnormalities.Stomatal number was significantly increased in mutant plants under ambient growth conditions, but were not different when grown at elevated CO 2 levels 1 and 2 (Table 1; Supplementary Table S1) Physiological properties of lcr1 lcr1 shows reduced quantum efficiency and photosynthesis in response to light The CO 2 assimilation rate (A) was severely affected in lcr1.In response to light, CO 2 assimilation in the mutant was saturated  at very irradiance (~100 µmol m -2 s -1 ), whereas the WT plants exhibited a typical C 4 response, with a curvilinear increase in CO 2 assimilation rate in response to increasing irradiance (Fig. 4A).Values for Φ were also significantly reduced, suggesting that light utilization efficiency was affected (Table 2).Analysis of the rapid response of chlorophyll fluorescence kinetics of ambient-grown plants revealed that NPQ was increased in the mutant relative to WT plants between 200 µmol m -2 s -1 and 500 µmol m -2 s -1 (Supplementary Fig. S2a).Corresponding decreases were observed in the maximum efficiency of PSII in the light (Supplementary Fig. S2b), ΦPSII (Supplementary Fig. S2c), and qP (Supplementary Fig. S2d).We have seen recovery in light response of the mutant to some extent when grown in elevated CO 2 conditions.When mutants were grown at elevated pCO 2 , CO 2 assimilation rates were saturated at a higher irradiance (~400 µmol m -2 s -1 ), although the response was still uncharacteristic of a typical C 4 photosynthetic response, compared with WT plants (Fig. 4A).No significant differences were observed in the values for Φ of these plants (Table 2).
lcr1 showed reduced CO 2 assimilation in response to intercellular CO 2 In response to changes in intercellular pCO 2 (C i ), CO 2 assimilation rates in mutant plants grown at ambient pCO 2 increased linearly, reaching a semi-plateau at ~1600 µmol CO 2 m -2 s -1 (Fig. 4B).Partial recovery in CO 2 response was observed when the plants were grown at elevated CO 2 .At elevated pCO 2 level 1, CO 2 assimilation rates of lcr1 plants were higher than in ambient levels at all intercellular pCO 2 , with a linear increase up to 1000 µbar.The carboxylation efficiency (CE) of mutants was significantly lower compared with WT plants at all growth pCO 2 .CO 2 compensation points (Γ) were 66.16 µbar and 18.00 µbar in lcr1 plants grown at ambient and elevated pCO 2 compared with 2.57 µbar in WT plants.Rates of respiration were ~1.5 times higher in plants grown at ambient pCO 2 , but no significant differences were found for plants grown at elevated pCO 2 (Table 2).

Isotope discrimination is severely affected in lcr1
Gas exchange measurements were repeated at ANU on the plants grown in high CO 2 conditions (2% pCO 2 ).The gas exchange measurement, coupled with a TDL trace gas analyser to measure 13 C isotope and 18 O (C 18 O 16 O) discrimination, showed similar results to the dry matter isotope measurements at IRRI, confirming a severe block of CO 2 assimilation in the mutant (Table 3).Stomatal conductance remained either unaffected or slightly better in the mutant.
The mutant showed a very high C i /C a value because of low CO 2 assimilation rates (Table 3).The exchange of 18 O between CO 2 and water is facilitated by the presence of CA in leaves, which catalyses the interconversion of CO 2 and bicarbonate (HCO 3 -).The C 18 O 16 O discrimination is   dependent on C i /C a ratio and CA activity.Although the C 18 O 16 O discrimination values of the WT and mutants were not significantly different, the values for the mutant were lower than expected at the given C i /C a (Table 3).There is a significant increase in carbon isotope discrimination (∆ 13 C) in lcr1 at both ambient and high CO 2 levels.The ∆ 13 C value fell in between the range of ∆ 13 C values of any of the C 3 or C 4 species when measured on the ambient CO 2 -grown plants (Fig. 5).
Mutation in Sevir.5g247800,coding for β-CA2, caused the lcr phenotype For identifying the causal mutation and the gene thereby affected, a BC 1 F 2 population segregating for F v /F m values was generated for gene mapping by sequencing (Fig. 6A).The F v /F m values of this population ranged from 0.65 to 0.78 (median 0.78) at ambient pCO 2 , and from 0.39 to 0.79 (median 0.69) at low pCO 2 (Fig. 6A).The percentage of BC 1 F 2 progeny with a low F v /F m (representing the lcr phenotype) was ~25% (Fig. 6A), indicative of a single recessive allele for this phenotype.Mapping-by-sequencing analysis led to a 3 Mb long region in chromosome 5, with mutant allele frequency peaking at position ~29 Mb (Fig. 6B).A total of 14 variants with alternative allele frequency of 1.0 in the mutant pool, but <0.5 in the azygous pool, were identified within this region (Supplementary Table S2).All 14 variants were single nucleotide polymorphisms (SNPs), and on functional annotation only two were found located within CDSs of genes, the rest were of low/no impact (Supplementary Table S2).Both the mutations within the gene CDSs were non-synonymous, leading to a predicted change in the encoded protein sequence.One of these mutations led to a substitution of serine to proline in a gene encoding a DNAbinding protein (Sevir.5G251900);as this gene is not expressed in any of the plant tissues (from public database, PhytoMine), this was not attributed as the causal mutation.The second mutation led to an amino acid change at position 156 from leucine to phenylalanine (Leu156Phe) in Sevir.5G247800 which encodes a β-CA2 gene.Given its high expression in leaf tissues of C 4 relatives such as maize and sorghum, it was identified as the causal gene for the lcr phenotype.There are two orthologues of this gene in maize (GRMZM2G121878 and GRMZM2G348512), and one in sorghum (Sobic.003G234500).Its sorghum orthologue shows high expression in leaf tissue (Supplementary Table S3) and both the maize genes have strong differential expression between BSCs and MCs (Supplementary Fig. S5).

Leu156Phe substitution in β-CA affected amino acid interactions within the protein, including new interactions with functionally important residues
The conservation profile of the amino acids in β-CA protein sequences across plant genomes showed that Leu156 is one of the 15 fully conserved amino acids (Supplementary Table S4).Eight of them were reported in the literature to play important functional roles in β-CA enzymatic activity (Kimber and Pai, 2000); Leu156, however, was not among the previously characterized residues (Supplementary Table S4).Three-dimensional structural models of WT and mutant Sevir.5G247800(with The plants were grown in 2% CO 2 .The measurements were made at 2% O 2 , 25 °C leaf temperature, and an irradiance of 1500 µmol quanta m -2 s -1 in a gas exchange system coupled to tuneable diode lasers that could measure either 13 C or 18 O discrimination.Asterisks denote statistically significant differences (*P<0.05,**P<0.01,***P<0.001)relative to the wild type according to two-sample t-test.n=4. a confidence score of 0.7) showed a characteristic dimer of β-CA, with both Leu156 and Phe156 buried inside the hydrophobic core of their respective structures, and located ~8 Å away from the catalytic site (Supplementary Fig. S6).
Comparison of the CO of the mutant structure with that of the WT showed an increase in aromatic-aromatic interactions (∆=14), but a substantial decline in aromatic-sulfur interactions (∆= -25) (Supplementary Table S5).Leu156 interacts with some of the functionally important residues (directly or indirectly).Three such residues, present in the secondary shell of Leu156, are Gly151, Ile111, and Val170, which are involved in the active site, the catalytic cleft, and solvent access, respectively; whereas two other residues present in the tertiary shell are Cys87 and His147, both involved in ligand binding (Supplementary Table S8).The mutant protein showed new covalent interactions at primary, secondary, and tertiary interacting shells of residues to accommodate the side chain of Phe156.Some of them were unique to Phe156 interaction shells, which also included residues involved in β-CA function in plants (Supplementary Fig. S7; Supplementary Table S8).Details of the functionally important amino acids of β-CA are shown in Supplementary Table S9.Even the inter-monomer interactions in the homodimer were also affected in mutant dimers, with loss of seven out of 11 salt bridges (Supplementary Table S6).The possibility of a heterodimer seemed unlikely as none of the interactions of native homodimers was observed in heterodimers (Supplementary Tables S6, S7).

Reduced abundance of CA transcript in lcr1
To confirm that the mutation had led to suppression of β-CA transcript accumulation, quantitative real-time PCR (RT-qPCR) was performed on the mutated β-CA transcript of lcr1.Gene Expression Atlas data for S. viridis shows that Sevir.5G247800 is the most expressed β-CA isoform in the S. viridis leaf (Supplementary Fig. S8A).Expression of both the splice forms of the candidate gene (Sevir.5g247800.1 and Sevir.5g247800.2) was reduced in lcr1, and is 0.8-and 0.5-fold lower than in WT plants (Fig. 7).Interestingly, the transcript abundance of two other tandem β-CA genes on chromosome 5 (Sevir.5G247900and Sevir.5G248000) was also reduced in lcr1 (Supplementary Fig. S8B).

Reduced CA activity and abundance in lcr1
Enzyme assays revealed that the activity of CA in lcr1 leaves was only 2% of that of the WT (Table 4).Immunoblotting with an anti-CA antibody showed an ~30% reduction in protein accumulation of CA in the mutant (Fig. 8).The abundance of primary C 4 cycle enzymes was investigated to examine the effect of the mutation on C 4 metabolism.The accumulation of MDH (14% lower) and ME (15% lower) was also found to be reduced in the mutant, while the accumulation of PEPC increased (10% higher) (Fig. 8).No significant differences in the accumulation of PPDK or the Rubisco LSU were observed (Fig. 8).The activities of Rubisco or PEPC (Table 4) were both significantly higher in lcr1 (Table 4).No differences were observed in the localization of the proteins within the cellular compartments (Supplementary Figs S9 -S11).

Discussion
Carbonic anhydrase (EC 4.2.1.1)catalyses the reversible conversion of CO 2 to HCO 3 , which is the first step of C 4 photosynthesis (Hatch and Burnell, There are multiple of CAs in plants, of which the β-CA isoforms are the most prevalent in higher plants (DiMario et al., 2017).There are also considerable variations in CA activity between species, regardless of their photosynthetic capacities (Gillon and Yakir, 2001;Cousins et al., 2008).In C 4 species, localized within the cytosol of MCs, CA provides the substrate for PEPC (Hatch and Burnell, 1990) by accelerating the interconversion of CO 2 to HCO 3 -by up to 10 4 -fold (Badger and Price, 1994), but the importance of CA in C 4 photosynthesis is still a matter of debate.The discrepancies among the observed phenotypes of different CA mutants have remained confusing for years.Early reports said that in vivo CA activity appears to be just enough to support the observed rates of photosynthesis (Jenkins et al., 1989a;Hatch and Burnell;1990), but not necessarily limiting for C 4 photosynthesis.The growth inhibition and photosynthetic phenotype of lcr1 is similar to that exhibited by antisense suppression of a putative cytosolic CA (CA3) in the C 4 dicot Flaveria bidentis, where reduction of CA activity to 10% of that of WT plants showed marked inhibition of CO 2 assimilation and a requirement of high CO 2 for growth (von Caemmerer et al., 2004;Cousins et al., 2006).However, our results contrast with those reported for the monocot Zea mays (Studer et al., 2014) and gene suppression work in S. viridis (Osborn et al., 2017) in which CA activity was reduced to 50% and <13% of WT levels, respectively, but no defect in photosynthesis at ambient conditions was observed.Interestingly, in lcr1, mutation in a single major CA gene (Sevir.5G247800)was strong enough to cause a subsequent reduction in expression of the transcript of other CA isoforms too (Supplementary Fig. S8B), which possibly could have happened due to post-transcriptional silencing of a mutated copy leading to concomitant downregulation of other homologous sequences.However, this was not observed when a major leaf CA isoform was silenced in S. viridis by Osborn et al. (2017).This could be a possible explanation for why the mutant phenotypes differed between these experiments despite the same gene being targeted.

C 4 phenotypic differences in lcr1
A reduction in CA activity would potentially limit CO 2 hydration and the supply of bicarbonate to PEPC, leading to a distinctive photosynthetic phenotype (von Caemmerer, 2000).Consistent with this, lcr1 exhibited severe disruptions to normal C 4 photosynthetic function.We observed that the CO 2 assimilation was saturated at low irradiance, and increased linearly in response to increasing intercellular pCO 2 .CE and quantum efficiency were significantly reduced, and the respiration rate was increased.All of these observations are indicative of a CO 2 supply limitation, as lcr1 recovered to some extent by application of high CO 2 .The supply of CO 2 is definitely not hindered by the stomatal properties, as stomatal conductance (g s ) was higher at all pCO 2 in plants grown at ambient and elevated pCO 2 (Supplementary Fig. S3).This increased stomatal conductance in the mutant is at odds with a typical CCM mechanism in C 4 species which allows the leaf to function at a relatively low conductance (Jenkins et al., 1989b;Brown and Byrd, 1993;He and Edwards, 1996;von Caemmerer, 2000;Kiirats et al., 2002).Increased Γ is likely  Enzyme activities were measured on 5-week-old plants grown in 2% CO 2 .The activities of carbonic anhydrase, Rubisco, and PEPC were determined as described by Osborn et al. (2017).
to be the result of a reduced CO 2 rate relative to mitochondrial respiration as there is no increase in O 2 sensitivity of Γ.Therefore, we concluded that the photosynthesis response in the mutant still follows a C 4 pathway but with a much reduced assimilation rate, where the BSC permeability remains intact.The increased ∆ 13 C value in the mutants is explained rather by a change in fractionation by C 4 fractionation factor b4, which is dependent on CO 2 hydration (Cousins et al., 2006).The ∆ 18 O decreases with reductions in CA activity (Williams et al., 1996;Cousins et al., 2006;Osborn et al., 2017).Consistent with this, ∆ 18 O values were lower at the high C i /C a in lcr1 compared with the expected values in the WT (Osborn et al., 2017).

Reduction in CA affects the abundance/activity of other C 4 enzymes in leaf
Earlier reports said that CA and PEPC may be controlled by similar regulatory mechanisms (Burnell et al., 1990), and the relationship between CO 2 hydration and PEPC carboxylation has been modelled in Z. mays (Farquhar, 1983).K PEP and K HCO3 of PEPC are negatively related and linked by a conserved serine residue (DiMario and Cousins, 2019).Likewise, we have seen differences in PEPC and other C 4 enzymes due to a change in behaviour of the lcr1 gene.There was an apparent increase in PEPC protein abundance (Fig. 8), possibly due to a feedback response with a significant change in PEPC activity (Table 4).Interestingly, the Rubisco activity is also significantly increased in the mutant.Generally, in C 4 plants, changes in CE are considered to be mostly related to PEPC activity assuming an excess of CA and saturated rates of Rubisco activity or rates of ribulose bisphosphate regeneration (Berry and Farquhar, 1978;Peisker, 1979;Collatz et al., 1992;von Caemmerer and Furbank, 1999).However, our result suggests that the reduction in CA activity in lcr1 is strong enough to alter the normal photosynthetic machinery in lcr1.
A single amino acid change (Leu156Phe) in Sevir.5G247800creates structural and functional alterations in the protein In S. viridis there are four genes encoding β-CA genes, CA5 (Sevir.2G245200)located on chromosome 2, two CA2 (Sevir.5G247800and Sevir.5G247900) and CA4 (Sevir.5G248000)located side by side on chromosome 5.There are two isoforms of the mutated gene (Sevir.5g24780;CA2) (Christin and Osborne, 2013;John et al., 2014) as also confirmed from our transcript analysis (Supplementary Fig. S8).All the genes are expressed in the cytosol of the M cells.Sequencing of lcr1 led to non-synonymous substitution from a leucine residue (L156) to a phenylalanine residue in the major β-CA gene Sevir.5g248000.Although the functional role of Leu156 was not reported by Kimber and Pai (2000), its complete conservation in the angiosperm β-CA family (Supplementary Table S4) indicates a plant-specific role.Due to the proximity of Leu156 to the active site (~8 Å), and even the (hydrophobic) residues lining the narrow channel through which solvent reaches the active site, it interacts with functionally important residues in the protein.Its substitution by phenylalanine altered the existing interactions (Supplementary Fig. S7; Supplementary Table S8), resulting in possible functional perturbation.Additionally, the positional and/or physicochemical nature of Leu156 indirectly played an important role in maintenance of aromatic-aromatic interactions within the β-CA protein structure.The substitution by phenylalanine, an aromatic amino acid, substantially altered the existing aromatic-aromatic interactions, such that there was a large reduction in CO within the mutant protein with respect to the native protein (Supplementary Table S5), which seemed to negatively impact the structural of protein, and eventually the normal functioning of the β-CA enzyme.The mutation even alters transcript accumulation for CA genes located on chromosome 5 and a significant reduction in total leaf CA protein abundance and activity (Supplementary Fig. S8; Fig. 8; Table 4).
Sevir.5g247800 as a target to achieve CA response in C 3 plans CA in C 3 plants has similar kinetic properties to those of C 4 plants (Hatch and Burnell, 1990), but it is localized in a different subcellular compartment (Tanz et al., 2009).
Evolutionary studies predicted that the C 4 CA has evolved from C 3 CA by loss of transit peptide and relocation of the enzyme in the cytosol; eventually, it was recruited in the C 4specific carboxylation function (Ludwig, 2012).It is predicted that the most abundant CA in the C 3 chloroplast stroma was actually modified in this way and present in multiple copies in C 4 species.Evidence from the phylogeny study shows that rice homologues LOC_Os01g45274 and LOC_Os09g28910 are the closest homologues of Sevir.5G247800, with 83.3.% and 48.0% protein similarity, respectively.Both of these rice genes code for chloroplastic precursors of the β-CA gene in rice and might be potential targets to create a CA response in rice.On this note, the transgenic approach is adopted by the C 4 consortium to overexpress the C 4 CA from Z. mays in the MCs of rice using a cell-specific promoter (Kajala et al., 2011).However, it still remains unclear if the multiple roles for CA in C 3 plants are an important consideration for manipulating the accumulation and activity of these proteins in C 3 plants.Therefore, given the involvement of a single CA isoform in the C 4 pathway in S. viridis, it may be a better choice than the CA of Z.mays, and can be incorporated in a C 3 species along with other C 4 enzymes to create a C 4 environment.Even the cis-element of Sevir.5G247800 could be tested for C 4 -specific gene expression in the mesophyll cytosol.Overall, our results demonstrate unequivocally that CA is an absolute requirement for C 4 photosynthesis, and that uncatalysed rates of CO 2 to HCO 3 -appear to be insufficient to support C 4 photosynthetic flux (Hatch and Burnell, 1990;Badger and Price, 1994).Our results highlight the potential importance of expressing at least some CA in the mesophyll cytosol of rice MCs in order to generate a functional C 4 biochemical pathway capable of high rates of photosynthesis.

Conclusion
Our results highlight the potential utility of chemical mutant screens to identify key genes involved in C 4 photosynthesis.The mutant shows a large reduction in carbon assimilation, a higher CO 2 compensation point, and higher carbon isotope discrimination.Analysis reveals mutation in the β-CA gene in the mutant, which not only causes a large reduction in CA activity in the, but also changes C 4 enzyme abundance.This study concludes that despite conflicting evidence, CA is a crucial first step in the C 4 pathway, at least in S. viridis, and potentially in all C 4 plants.In other species, different CA isoforms may take on this role and be responsible for a similar phenotype.

Supplementary data
The following supplementary data are available at JXB online.
Table S4.List of amino acid residues found to be fully conserved in β-CA homologues across entire plant genomes.
Table S5.Contact order values for each interaction type in native and mutant structures.
Table S6.Change in inter-monomer interactions between native and mutant structures of β-CA.
Table S7.Change in inter-monomer interactions between native (homodimer) and isoform heterodimer structures of β-CA.
Table S8.List of amino acid residues found in primary, secondary, and tertiary shells of Leu156 and Phe156.
Table S9.List of functionally important amino acid residues identified in Pisum sativum β-CA by Kimber and Pai (2000) and Rowlett (2010).Fig. S7.Venn diagram comparing combined interaction shells (primary, secondary, and tertiary) of Leu156 and Phe156, and further with the functionally important residues of β-CA (as per Kimber and Pai, 2000).

Fig. 1 .
Fig. 1.Identification of the mutant lcr1.Histogram of F v /F m at ambient pCO 2 (390 μbar) and after 48 h at low pCO 2 (15 μbar).M 3 (A), M 4 (B), and two M 5 generation progenies (C, D); n=80, 100, and 50 for lcr1 at the M 3 , M 4 , and M 5 generations, respectively.Dashed vertical lines are the average F v /F m of wild-type Setaria viridis at low CO 2 for that batch (n= 260, 100, and 50 for M 3 , M 4 , and M 5 respectively).Asterisks show the bin from which the progeny advanced to each generation were selected.

Fig. 2 .
Fig. 2. Chlorophyll fluorescence properties of lcr1.The graph shows the F v /F m of wild-type (WT) and M 5 generation lcr1 plants grown at ambient and low pCO 2 .Values are the average ±SE of 10 individual plants per line.Significant variation in values of the WT and lcr1 are denoted with *P<0.05 and ***P<0.001.

Fig. 3 .
Fig. 3. Plant morphology and leaf anatomy of lcr1.(A) The lcr1 plant shows a dwarf phenotype in ambient pCO 2 , but recovered in elevated pCO 2 .Scale bar=5cm.(B) Leaves are thin in lcr1 in ambient pCO 2 .Quantification of leaf anatomical traits is given inTable 1. Red shows autofluorescence of chloroplast; yellow shows the autofluorescence from the cell walls.Scale bar=50 µm.Plant images are redrawn fromCoe et al. (2018).

Fig. 4 .
Fig. 4. Photosynthetic response of lcr1 in ambient pCO 2 and elevated pCO 2 .(A) Net CO 2 assimilation rate (A) in response to changes in photosynthetic photon flux density (PPFD) and (B) intercellular pCO 2 (C i ).Measurements were made at a pCO 2 (C a ) of 400 µbar or 1800 µmol photons m -2 s -1 and a leaf temperature of 30 °C.Values are the average ±SE of one fully expanded leaf from three different plants of the wild-type (WT, filled symbols) and M 6 generation lcr1 (open symbols) plants grown at ambient (circles) or elevated (10 000 ppm) pCO 2 (triangles).

Fig. 5 .
Fig. 5. Leaf dry matter carbon isotope discrimination in lcr1.∆ 13 C value of rice, wild-type Setaria viridis (WT), and M 6 generation lcr1 plants.Values are the average ±SE of 10 plants per line.Different letters denote statistically significant differences (P<0.05).

Fig. 6 .
Fig. 6.Sequencing of the BC 1 F 2 population.(A) Histogram of F v /F m at ambient pCO 2 and after 48 h at low pCO 2 in a BC 1 F 2 population of lcr1.Solid and dashed vertical lines are the average F v /F m of the wild type (WT) at ambient and low pCO 2 , respectively (n=300 for BC 1 F 2 of lcr1; n= 50 for the WT); ~25% of the population shows mutant response.(B) Mutant allele frequencies in chromosome 5.The graph represents the plot of mutant allele frequency along the length of chromosome 5.The lines at 1 and 0.5 represent the threshold of allele frequency in the mutant and WT pool.

Fig. 7 .
Fig. 7. Transcript accumulation of β-CA spliceforms of the gene Sevir.5g247800as measured by quantitative real-time PCR.Values are expressed as fold changes in transcript accumulation relative to the wild type (average ±SE).Asterisks denote statistically significant differences relative to average wild type values (P<0.05).

Fig. 8 .
Fig. 8. Soluble leaf proteins.(A) Western blots of wild types (WTs) and five individuals of lcr1.Protein was extracted from the fully expanded fourth leaf; samples were loaded on an equal leaf weight (34 µg for PEPC and LSU, 136 µg for PPDK, 340 µg for MDH, ME, and CA).(B) Integrated density values are expressed as fold changes relative to WT plants (average ±SE, n=5).Asterisks denote statistically significance differences relative to average WT values (P<0.05).

Fig. S8 .
Expression pattern of β-CA isoforms in the Gene Expression Atlas of S. viridis (A) and on quantification by RT-qPCR in lcr1 (B).

Fig. S9 .
Representative images of immunolocalization of carbonic anhydrase protein in wild-type and M 6 generation lcr1 plants.Fig. S10.Representative images of immunolocalization of Rubisco protein in wild-type and M 5 generation lcr1 plants.Fig. S11.Representative images of immunolocalization of PEPC protein in wild-type and M 6 generation lcr1 plants.

Table 1 .
Morphological characteristics of lcr1 in ambient and elevated CO 2

Table 2 .
Photosynthetic response of lcr1 in ambient and elevated CO 2 conditions

Table 3 .
Gas exchange measurements of lcr1 on elevated CO 2 level 2-grown plants

Table 4 .
Leaf biochemical properties of lcr1