Generating and characterizing single- and multigene mutants of the Rubisco small subunit family in Arabidopsis

Abstract

The primary CO₂-fixing enzyme Rubisco limits the productivity of plants. The small subunit of Rubisco (SSU) can influence overall Rubisco levels and catalytic efficiency, and is now receiving increasing attention as a potential engineering target to improve the performance of Rubisco. However, SSUs are encoded by a family of nuclear rbcS genes in plants, which makes them challenging to engineer and study. Here we have used CRISPR/Cas9 [clustered regularly interspaced palindromic repeats (CRISPR)/CRISPR-associated protein 9] and T-DNA insertion lines to generate a suite of single and multiple gene knockout mutants for the four members of the rbcS family in Arabidopsis, including two novel mutants 2b3b and 1a2b3b. 1a2b3b contained very low levels of Rubisco (~3% relative to the wild-type) and is the first example of a mutant with a homogenous Rubisco pool consisting of a single SSU isoform (1B). Growth under near-outdoor levels of light demonstrated Rubisco-limited growth phenotypes for several SSU mutants and the importance of the 1A and 3B isoforms. We also identified 1a1b as a likely lethal mutation, suggesting a key contributory role for the least expressed 1B isoform during early development. The successful use of CRISPR/Cas here suggests that this is a viable approach for exploring the functional roles of SSU isoforms in plants.

Arabidopsis thaliana, chloroplast, CRISPR/Cas9, photosynthesis, protoplasts, Rubisco, SpCas9, T-DNA

Introduction

Rubisco (EC 4.1.1.39) catalyses the reaction between CO₂ and ribulose-1,5-bisphosphate (RuBP), and is responsible for net carbon gain in all oxygenic photosynthetic organisms including plants. Despite its importance, Rubisco is a relatively slow and error-prone enzyme that limits the efficiency of photosynthesis. This phenomenon is particularly evident in C₃ crop plants, where increasing the operating efficiency of Rubisco and reducing photorespiration are considered promising strategies for improving growth and productivity (Rae et al., 2017; Kubis and Bar-Even, 2019; Simkin et al., 2019; South et al., 2019). In plants, Rubisco (Form IB) is composed of eight large subunits (LSUs) and eight small subunits (SSUs) that form an L₈S₈ complex (Bracher et al., 2017). The LSU is encoded by a single, highly conserved gene on the chloroplast genome (rbcL), while SSUs are encoded by a family of nuclear rbcS genes that show significantly more diversity between species compared with rbcL. The size of the SSU family also differs between species, with up to 20 isoforms reported in cultivated polyploid wheat varieties (Sasanuma, 2001).

In Form I Rubisco, the eight LSUs form dimers that carry two active sites per dimer. The SSUs are located distal to the active sites and are not required for carboxylation, but are essential to maximize Rubisco activity and are thought to provide structural stability to the L₈S₈ complex (Andersson and Backlund, 2008; van Lun et al., 2011). A substantial body of work in the green alga Chlamydomonas reinhardtii, cyanobacteria, and higher plants has demonstrated that directed mutagenesis of the SSU, or expression of heterologous SSUs, can significantly modify the catalytic properties of Rubisco, including the turnover rate (k_cat), CO₂/O₂ specificity (S_c/o), and the ability to assemble the L₈S₈ complex (Schneider et al., 1990; Paul et al., 1991; Read and Tabita, 1992; Kostov et al., 1997; Getzoff et al., 1998; Spreitzer et al., 2005; Genkov and Spreitzer, 2009; Ishikawa et al., 2011; Esquivel et al., 2013; Whitney et al., 2015; Atkinson et al., 2017; Fukayama et al., 2019; Orr et al., 2020). More recently, several native, specialized SSU isoforms have been shown to enhance the catalytic properties of Rubisco in plants (Morita et al., 2014; Laterre et al., 2017; Pottier et al., 2018; Lin et al., 2019, Preprint). This suggests that SSUs naturally play a role in modifying Rubisco performance and thus could be an important target for crop improvement.

The expression levels of SSUs also play a key role in regulating Rubisco abundance, and thus affect whole-plant nitrogen and carbon partitioning, and the overall capacity for carbon uptake (Rodermel, 1999). Previous work has shown that reducing SSU content by antisense RNA or T-DNA insertion results in a decrease in rbcL mRNA translation and, subsequently, a decrease in Rubisco content (Rodermel et al., 1996; Wostrikoff and Stern, 2007; Izumi et al., 2012; Ogawa et al., 2012; Wostrikoff et al., 2012; Atkinson et al., 2017). The transcript abundances of SSU families have been explored in detail only for a small number of plant species, such as tomato, wheat, rice, and Arabidopsis (Wanner and Gruissem, 1991; Galili et al., 1992; Suzuki et al., 2009; Izumi et al., 2012). Within a species, the strength of promoters for each SSU can vary significantly, resulting in a range of expression levels between different SSU isoforms. Furthermore, expression levels of individual SSUs can vary depending on tissue type and developmental stage, and in response to the growth environment (Wehmeyer et al., 1990; Meier et al., 1995; Ewing et al., 1998; Day et al., 2000; Morita et al., 2014; Laterre et al., 2017). Functional studies for individual SSUs in a species are challenging due to the high sequence homology between SSU isoforms within a family (Yamada et al., 2019). Nevertheless, a better understanding of how the expression of different SSU isoforms is co-ordinated in response to the environment could lead to novel strategies to improve plant growth performances (Cavanagh and Kubien, 2014).

The model plant Arabidopsis thaliana (hereafter Arabidopsis) has four SSU genes, rbcS1A, rbcS1B, rbcS2B, and rbcS3B (hereafter 1A, 1B, 2B, and 3B, respectively), which are divided into A and B subfamilies based on linkage and sequence similarities (Krebbers et al., 1988; Schwarte et al., 2011). 1A and 3B are typically reported as the dominant SSU isoforms, while 1B and 2B are expressed at lower levels (Izumi et al., 2012; Klepikova et al., 2016). The Arabidopsis SSU family show signs of spatially overlapping and distinct expression during early leaf development (Sawchuk et al., 2008). Although the response of Arabidopsis SSUs to environmental stimuli is relatively well studied (Dedonder et al., 1993; Cheng et al., 1998; Yoon et al., 2001; Sawchuk et al., 2008), a clear understanding of their impact on growth and performance is still lacking.

Gene knockout (KO) mutants generated by T-DNA insertion have been useful tools for functional studies in Arabidopsis (Izumi et al., 2012; Atkinson et al., 2017), but this approach does have limitations. First, T-DNA insertion sites are prone to small deletions, duplications, and filler sequence of unknown origin, while T-DNA lines have been shown to contain chromosomal translocations, and in some cases significant chromosomal rearrangements (Nacry et al., 1998; Clark and Krysan, 2010). Thus, multiple T-DNA insertion lines for a given gene are typically required to verify experimental findings. However, in some cases, only a limited number of T-DNA lines are available for a given locus. Secondly, generating multiple gene KO lines is time consuming and not feasible for genes with loci in close proximity, as is the case for the three SSU B subfamily genes which, in Arabidopsis, are in a tandem array on chromosome 5 (Krebbers et al., 1988; Niwa et al., 1997).

To overcome this challenge, we have utilized a pooled clustered regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated protein 9 (Cas9) approach and available T-DNA insertions lines to generate a novel suite of rbcs mutants. We produced new single rbcs mutants for all four rbcS genes, a double rbcs mutant (2b3b), and a triple rbcs mutant (1a2b3b). Molecular characterizations were performed to examine the impact of specific SSU mutations on protein and Rubisco contents, while physiological analysis under near-outdoor light levels (1000 µmol photons m^–2 s^–1) provided novel insights into the contributions of different SSU isoforms to growth performance. This study serves as a proof of principle for future studies to examine the roles of different SSU isoforms in other species.

Materials and methods

Plant material and growth conditions

Arabidopsis [Arabidopsis thaliana (L.) Heyn. Col-0] seeds were sown on soil and stratified for 3 d at 4 °C, and grown at 22 °C, ambient CO₂, 70% relative humidity, and a photosynthetic photon flux (PPFD) of 200 µmol photons m^–2 s^–1 (standard lab conditions) or 1000 µmol photons m^–2 s^–1 (high light) supplied by cool white fluorescent lamps (Percival SE-41AR2, Clf Plantclimatics GmbH) in 12:12 h light:dark cycles. For comparison of different genotypes, plants were grown from seeds of the same age and storage history, harvested from plants grown in the same environmental conditions. Arabidopsis T-DNA insertion lines 1a [GABI_608F01 (At1g67090)], 1b [SAIL_755_D09 (At5g38430)], 2b [GABI_324A03 (At5g38420)], and 3b [SALK_117835 (At5g38410)] were sourced from the Nottingham Arabidopsis Stock Centre (NASC). The 1a2b mutant (GABI_608F01; GABI_324A03) generated previously (Atkinson et al., 2017) was backcrossed with a wild-type (WT) plant to remove potential background mutations. The 1a3b mutant (GABI_608F01; SALK_117835) was provided by Hiroyuki Ishida, Department of Applied Plant Science, Tohoku University, Japan. Homozygous T₃ seed stocks for mutants generated via CRISPR/Cas9 in this study can be obtained through the NASC (http://arabidopsis.info) (NASC IDs N2109789 – N2109802).

Construction of CRISPR/Cas9 vectors

Plasmid vectors were assembled using the Plant MoClo Golden Gate modular cloning kit (Engler et al., 2014). New Level 0 parts were made according to Patron et al. (2015). Level 0 vectors (100 ng each) carrying the UBI10 promoter, the SpCas9 coding sequence (Parry et al., 2016), or the heat shock protein (HSP) terminator (Nagaya et al., 2010) were assembled into the Level 1 Position 2 (L1P2) acceptor vector in a 20 µl assembly reaction [BsaI (ThermoFisher Scientific, UK) (10 U), 1× Buffer G, T4 DNA ligase (ThermoFisher Scientific) (400 U), and 20 nmol ATP] as in Vasudevan et al. (2019). PCR amplicons of each complete guide RNA (gRNA; the spacer fused to the RNA scaffold) were combined with a Level 0 vector carrying the U6 promoter for assembly (see Supplementary Table S1 at JXB online). Each pair of gRNA expression cassettes were constructed in L1P3 and L1P4, respectively, as described in Raitskin et al. (2019). Four Level 1 transcriptional units [the pFAST selection marker (Shimada et al., 2010) in L1P1, L1P2, L1P3, and L1P4] were then assembled into a Level 2 acceptor vector in a 20 µl assembly reaction [BpiI (ThermoFisher Scientific) (10 U), 1× Buffer G (ThermoFisher Scientific), T4 DNA ligase (ThermoFisher Scientific) (400 U), and 20 nmol of ATP] (see Supplementary Data S1) as in Vasudevan et al. (2019).

DNA and RNA extraction, PCR, and RT–qPCR

DNA was extracted from a mature leaf as described in Li and Chory (1998). PCRs were performed as in McCormick and Kruger (2015) using gene-specific primers (Supplementary Table S2). Total RNA was isolated from leaves using the RNeasy plant mini kit (Qiagen, USA). Isolated RNA was treated with DNase (Qiagen) and reverse transcribed with random primers (Promega, USA). Gene-specific primers amplifying the unique 3' region of the transcript were used for quantitative reverse transcription–PCR (RT–qPCR) (Izumi et al., 2012). A DNA fragment containing regions matching the target loci of the rbcS for RT–qPCR primers was synthesized (Gblock, IDT) (Supplementary Fig. S1). RT–qPCR calibration curves were constructed using known concentrations of the standard to quantify mRNA levels for each rbcS transcript pool. For quantitative analysis, an aliquot of cDNA derived from 4 ng of RNA was used (total volume 20 µl) with SYBR Green Master Mix (Eurogentec, Belgium).

CRISPR/Cas9 cassettes in protoplasts

Leaves from 4-week-old plants were cut vertically into 1 mm strips and digested in 10 ml of maceration glycine glucose (MGG) digestion solution as in Chupeau et al. (2013) containing cellulase ‘Onozuka’ R-10 [1.5% (w/v)] and Macerozyme R-10 [0.4% (w/v); Yakult Pharmaceutical, Japan] for 3 h. Released protoplasts were filtered from the digestate using a 70 μm cell strainer and washed three times with MGG not containing enzymes to remove traces of the enzyme solution and cell debris. Protoplasts were resuspended in MMM solution [0.4 M mannitol, 15 mM MgCl₂, 0.1% (w/v) MES (pH 8)] to a concentration of 5×10⁵ cells ml^–1 in a 5 ml glass test tube. For protoplast transformations, 8 μl of DNA (4 μg total) was added to 75 μl of the protoplast suspension, followed by addition of 83 μl of polyethylene glycol (PEG) solution [0.4 M mannitol, 0.1 M Ca(NO₃)₂·4H₂O, 40% (w/v) PEG 4000 (pH 8)]. Following a 1 min incubation, 2 ml of MGG solution was added. Following a further 1 h incubation at room temperature, the protoplasts were centrifuged at 70 g for 5 min and the supernatant removed. Fresh MGG solution was added (100 μl), and transfected protoplasts were incubated in the dark for 18 h at room temperature. The target loci of each CRISPR/Cas9 vector were analysed by PCR of protoplast DNA extracts (Supplementary Table S2).

Expression of Cas9 and gRNA in Arabidopsis

Binary vectors (Level 2) were transformed into Agrobacterium tumefaciens (AGL1) for stable insertion in Arabidopsis by floral dipping (Clough and Bent, 1999). T₁ plants were screened for the presence of the transgene by the pFAST selectable marker (Shimada et al., 2010), and for the presence of CRISPR/Cas9-mediated mutations by PCR and Sanger sequencing (Supplementary Table S2). Stable mutations in transgene-free T₂ plants were confirmed by Sanger sequencing.

Protein quantification and Rubisco content

Leaf samples (20–40 mg) were collected from 35-day-old plants, snap-frozen, and stored at –80 °C prior to extraction. Samples were ground rapidly in an ice-cold mortar and pestle in 200 µl of protein extraction buffer [50 mM Bicine-NaOH pH 8.2, 20 mM MgCl₂, 1 mM EDTA, 2 mM benzamidine, 5 mM ε-aminocaproic acid, 50 mM 2-mercaptoethanol, 10 mM DTT, 1% (v/v) protease inhibitor cocktail (Sigma-Aldrich, USA), and 1 mM phenylmethylsulfonyl fluoride] for ~1 min followed by centrifugation at 14 700 g at 4 °C for 1 min. Supernatant (90 µl) was then mixed with 100 µl of carboxyarabintol-1,5-bisphosphate (CABP) binding buffer {100 mM Bicine-NaOH (pH 8.2), 20 mM MgCl₂, 20 mM NaHCO₃, 1.2 mM (37 kBq µmol^–1) [¹⁴C]CABP}, incubated at room temperature for 25 min, and Rubisco content was determined via [¹⁴C]CABP binding (Sharwood et al., 2016). Bradford assay was used to determine total soluble protein in the same supernatant as prepared for Rubisco content analysis (Bradford, 1976).

Extracts were subjected to SDS–PAGE on a 4–12% (w/v) polyacrylamide gel (Bolt^® Bis-Tris Plus Gel) (ThermoFisher Scientific, UK), transferred to a PVDF membrane, then probed with rabbit serum raised against wheat Rubisco at 1:10 000 dilution (Howe et al., 1982) followed by LI-COR IRDye® 800CW goat anti-rabbit IgG (LI-COR Biosciences, USA) at 1:10 000 dilution, then viewed on an LI-COR Odyssey CLx Imager. The relative abundances of LSUs and SSUs were estimated densitrometically using Image Studio Lite (LI-COR Biosciences) and the values were means ±SE based on three immunoblots as in Atkinson et al. (2017)

Chlorophyll quantification

Leaf discs (20 mm² in total) were frozen in liquid N₂, powdered, and then mixed with 1 ml of ice-cold 80% (v/v) acetone, 10 mM Tris–HCl. Following centrifugation at 17 200 g for 10 min, chlorophyll was quantified according to Porra et al. (1989).

Measurement of photosynthetic parameters

Gas exchange and chlorophyll fluorescence were determined using a LI-COR LI-6400 portable infra-red gas analyser (LI-COR Biosciences) with a LI6400-40 leaf chamber (2 cm² area) on either the sixth or seventh leaf of 35- to 45-day-old rosettes grown under 200 μmol photons m^–2 s^–1 in large pots to generate leaf area sufficient for gas exchange measurements (Atkinson et al., 2017). For all gas exchange experiments, leaf temperature and chamber relative humidity were 25 °C and ~65%, respectively. The response of the net CO₂ assimilation (A) to the intercellular CO₂ concentration (C_i) was measured at various CO₂ concentrations (50, 100, 150, 200, 250, 300, 350, 400, 500, 700, 900, and 1200 µmol mol⁻¹) under saturating light (1800 μmol photons m^–2 s^–1) (Supplementary Fig. S2). Gas exchange data were corrected for CO₂ diffusion from the measuring chamber as in Bellasio et al. (2016). To calculate the maximum rate of Rubisco carboxylation (V_cmax), the A/C_i data were fitted to the C₃ photosynthesis model as in Ethier and Livingston (2004) using the catalytic parameters K_c^air and affinity for O₂ (K_o) values for WT Arabidopsis Rubisco at 25 °C from Atkinson et al. (2017). Estimates of the light- and CO₂-saturated photosynthetic electron transport rate (J_max) were not included as several of the SSU mutants were probably Rubisco limited even at high CO₂ concentrations. Maximum quantum yield of PSII (F_v/F_m) was measured using a Hansatech Handy PEA continuous excitation chlorophyll fluorimeter (Hansatech Instruments, UK) (Maxwell and Johnson, 2000).

Rosette area and biomass

Rosettes were imaged daily during growth experiments. Rosette area was calculated using iDiel Plant software (Dobrescu et al., 2017). For biomass measurements, aerial rosette tissue was removed and weighed to determine fresh weight. Samples were then dried in an oven (80 °C for 3 d) and weighed to determine dry weight.

Statistical analysis

Significant differences between sample groups were assessed by one-way ANOVA followed by Tukey’s honest significant difference (HSD) post-hoc test (IBM SPSS Statistics Ver. 26.0, USA) for individual parameters. Difference in growth performance (as measured by rosette area) were assessed by repeated measures ANOVA followed by Tukey’s HSD post-hoc test.

Results

Identification of T-DNA insertion lines and double knockout mutants

We initially performed a search for available Arabidopsis mutant lines on T-DNA Express (http://signal.salk.edu/cgi-bin/tdnaexpress) that carry a single homozygous T-DNA insertion, ideally located in exonic regions for members of the SSU gene family. We identified only one such line for 1A (GABI_608F01) and 2B (GABI_324A03) (Fig. 1A). The only available homozygous mutants for 1B (SAIL_755_D09) and 3B (SALK 117835) had insertion sites located in the 5'-untranslated region (UTR). For the latter, 3B expression is reportedly reduced but not absent (Izumi et al., 2012).

Fig. 1.

The Rubisco small subunit gene family in Arabidopsis thaliana. (A) The sites of the T-DNA insertions for mutant lines used in this study are shown in blue, with the orientation of the left border (LB) indicated. The locations targeted by CRISPR/Cas9 are shown, with the names of the targeting gRNAs in green boxes (see Supplementary Table S1 and vector maps in Supplementary Data S1 for gRNA sequences). The 5'- and 3'-untranslated regions (UTRs) are shown for each gene. (B) The B subfamily is located in tandem in the 1B–3B locus. A promiscuous pair of gRNAs targeting the three small subunits in the 1B–3B locus is indicated.

Double mutants 1a2b and 1a3b were generated previously by crossing available T-DNA lines (Izumi et al., 2012; Atkinson et al., 2017). Upon further characterization of 1a2b, we observed that the mutant had a defective silique phenotype, with a reduced silique size and seed count per silique (Supplementary Fig. S3A, B). Backcrossing with a WT plant and then re-segregating 1a2b in the F₂ generation removed the observed silique phenotype, indicating that this trait was not attributable to the absence of 1A and 2B. Furthermore, the new 1a2b showed a more robust growth phenotype compared with the original double mutant (Supplementary Fig. S3C). Single rbcs mutant lines for 1a or 2b did not show a silique phenotype. We identified plants in the segregating F₂ population with the silique phenotype or reduced growth, but with no T-DNA insertions in either 1A or 2B, indicating that these traits were possibly linked to a recessive heterozygous mutation in either the 1a or 2b parental line. The new 1a2b double mutant was used for the remainder of this study. We also attempted to generate a 1a1b double mutant by crossing the T-DNA insertion mutants for 1a and 1b, and were successful in generating a heterozygote F₁ (1A1a1B1b) line (Supplementary Fig. S4). We were not able to recover a 1a1b mutant after screening 125 F₂ plants, but did identify a 1A1a1b1b line. Analysis of 15 F₃ progeny of the latter line failed to show a Mendelian distribution for the 1A locus (8:7:0 for WT:heterozygous:homozygous KO), suggesting that 1a1b may be lethal. The germination rates of seeds recovered from the mutant lines were comparable with WT plants (WT, 93%; 1A1a1B1b, 93%; and 1A1a1b1b, 96% of ~100 seeds), indicating that selection against a 1a1b genotype occurred prior to seed development.

gRNA targeting strategy and transient expression of Cas9/gRNAs in protoplasts

To generate new rbcs mutants for individual members of the Arabidopsis SSU family using the CRISPR/Cas9 method, we initially designed two unique pairs of gRNAs to target specific regions of each of the four rbcS genes, for a total of eight gRNA pairs (Fig. 1A; Supplementary Table S1). Furthermore, to knock out 1B, 2B, and 3B simultaneously, we designed a promiscuous pair of gRNAs to target homologous regions in the 1B–3B locus (Fig. 1B; Supplementary Table S1). A paired gRNA approach was chosen for two reasons: (i) to increase the probability of generating mutations and (ii) the generation of larger deletions between the two gRNAs could potentially be screened more easily and cheaply (e.g. by PCR). Each gRNA pair was assembled using the Plant MoClo system as an individual expression cassette in a Level 2 binary vector containing a Cas9 expression cassette (Supplementary Data S1) (Engler et al., 2014).

Vectors with each rbcS-specific gRNA pair were initially tested using a novel Arabidopsis protoplast transient expression system to estimate the efficiency of generating deletion events (Fig. 2A). Following transfection of the gRNA pairs targeting 1A or 1B, amplification of the respective gene loci produced the expected WT band and a second lower band indicating a deletion event based on the target sites of the gRNA pair. Pairs 1AP2 and 1BP2 most consistently produced prominent ‘deletion bands’ for 1A and 1B, respectively, and thus were selected for in planta expression. Only one gRNA pair each for 2B (2BP2) and 3B (3BP1) produced a deletion band. Sequence analysis of the deletion bands indicated cleavage 3–4 bp upstream of the gRNA PAM (photospacer adjacent motif) sites, consistent with the activity of SpCas9, and deletions ranging from 96 bp to 180 bp (Fig. 2B).

Fig. 2.

Targeted mutagenesis of Rubisco small subunits in protoplasts of Arabidopsis thaliana using vectors encoding Cas9 and gRNA pairs. (A) Detection of mutations using PCR. Lane 1, DNA marker; lanes 2–9, PCR products of genomic DNA from protoplasts transfected with vectors carrying gRNA pairs targeting Rubisco small subunit (SSU) genes 1A (1AP1, 1AP2), 1B (1BP1, 1BP2), 2B (2BP1, 2BP2), and 3B (3BP1, 3BP2) (see Supplementary Table S2 for primer details). (B) Sequencing results of the lower ‘deletion band’ showed deletion events 3–4 bp upstream of the PAM site (underlined). The gRNA target sequences are shown in bold.

Expression of CRISPR/Cas9 in planta to generate stable mutants for each rbcS isoform

Arabidopsis plants were stably transformed with CRISPR/Cas9 binary vectors containing gRNA pairs 1AP2, 1BP2, 2BP2, 3BP1, and the promiscuous pair 1B3B. Transformed T₁ seeds were visually selected using the pFAST red fluorescent seed coat marker for the presence of Cas9 and gRNAs (Shimada et al., 2010). The loci of the five gRNA pairs were initially screened by PCR for large deletions, which were detected in transformants for 1AP1, 2BP2, and 3BP1, but not for 1BP2 (Supplementary Fig. S5A). The deletion bands were consistent with those observed in protoplasts (Fig. 2A). Several deletion bands were detected for 1B3B that were consistent with predicted amplicon sizes following multiple cleavage events within the 1B–3B locus (Supplementary Fig. S5B). Sequencing of each amplicon showed the expected cleavage position upstream of the PAM sites.

The efficiency with which deletions were induced by gRNA pairs ranged from 1% to 14% (Table 1). However, all deletions that were detected were accompanied by a significantly brighter WT band, indicating chimeric rather than heritable genomic mutations (i.e. homozygous or heterozygous) (Feng et al., 2014; Pauwels et al., 2018). T₁ transformants were also screened by Sanger sequencing to detect potential small indel mutations at each gRNA target site (Supplementary Fig. S5C). Indels were generally detected at a higher frequency than large deletions (3–31%).

Table 1.

Editing efficiency of paired gRNAs targeting each rbcS gene

			Number of T1 plants (with transgene)		Number of T2 plants (transgene free)
gRNA pair	Target gene	Background	Large deletion	Indels/PM	Large deletion	Indels/PM
1AP2	1A	WT	8/92 (9%)	26/106 (25%)	0/8 (0%)	24/204 (12%)
1BP2	1B		0/112 (0%)	11/35 (31%)	N/A	42/66 (64%)
2BP2	2B		1/69 (1%)	12/63 (19%)	0/1 (0%)	30/78 (38%)
3BP1	3B		10/70 (14%)	1/32 (3%)	0/10 (0%)	24/66 (36%)
1B3B	1B	WT	8/76 (11%)	N/A	0/8 (0%)	0/64 (0%)
	2B					1/64 (2%)
	3B					8/64 (13%)
1BP2	1B	1a2b	0/33 (0%)	8/33 (24%)	N/A	N/A
3BP2	3B	1a2b	0/30 (0%)	3/30 (10%)	N/A	N/A

T₁ plants containing Cas9 were screened by PCR for large deletions and Sanger sequencing for indels and point mutations (PMs). T₁ lines containing large deletions or indels/PMs were screened for heritable mutations in Cas9-free plants in the T₂ generation. Six of the Cas9-free T₂ plants were sequenced for each T₁ line (34 lines for 1AP2, 11 lines for 1BP2, 13 for 2BP2, 11 lines for 3BP1, and 8 lines for 1B3B).

Screening for heritable mutations in the T₂ generation was performed on segregated non-red seed progeny (i.e. containing no CRISPR/Cas9 insertion). Eight progeny from each T₁ line that showed large deletions were screened by PCR (Table 1). No heritable large deletions were detected for any gRNA pair, confirming that the observed deletions in T₁ were chimeric mutations. Sanger sequencing was then performed on T₂ lines that contained large deletions or indels in T₁. Stable homozygous indels (i.e. frameshift mutations in exonic regions that produced early stop codons) were identified for all four rbcS genes targeted by specific gRNA pairs (Supplementary Table S3; Supplementary Fig. S6). For the promiscuous gRNA pair 1B3B, each of the B subfamily genes was sequenced in 64 non-red T₂ plants from eight T₁ lines showing large deletions. No mutations were observed in 1B in any of the lines tested. However, we did identify a 147 bp deletion in 3B common to the progeny of a single T₁ line. Sequencing of the eight progeny revealed an additional 4 bp indel deletion in 2B in a single plant, thus producing a 2b3b double mutant. For 2b3b, both indels led to frameshift mutations that produced early stop codons in 2B and 3B. For all rbcs lines, T₂ plants containing homozygous mutations were selected and seeds were collected for subsequent molecular and physiological characterizations in the T₃ generation.

Generating triple rbcs mutants using the 1a2b T-DNA insertion mutant

We next introduced CRISPR/Cas9 binary vectors containing gRNA pairs 3BP1 or 1BP2 into the new 1a2b double T-DNA insertion mutant to attempt to produce the novel triple mutants 1a2b3b or 1a1b2b. Out of 30 T₁1a2b transformants for 3BP1, three showed a slow-growth, pale leaf phenotype compared with the 1a2b phenotype (Table 1; Fig. 3A). Sanger sequencing of the 3B locus confirmed the presence of a homozygous and biallelic frameshift mutation in all three T₁ plants (Supplementary Fig. S6). The heritability of those mutations was confirmed in non-red seed progeny of the T₂ generation for each line. In contrast, 1B is a minor SSU isoform, so no growth phenotype was expected in T₁1a2b transformants for 1BP2. Sanger sequencing of the 1B locus in 33 T₁ plants showed mutations in 1B for eight plants (~ 25%), indicating that the CRISPR/Cas9 was functional and efficient. However, in all cases, only single base pair changes (i.e. a single codon substitution) or silent substitutions were observed (Supplementary Fig. S7).

Fig. 3.

Total soluble protein and Rubisco contents in rbcs mutants of Arabidopsis thaliana. (A) The 1a2b3b mutant compared with 1a3b and WT plants grown under standard conditions. All images represent 28-day-old rosettes unless otherwise stated. (B) Total soluble protein content shown for 35-day-old plants. (C) Rubisco content determined by [¹⁴C]CABP binding and subunit ratios estimated by immunoblotting. Values are means ±SE of five measurements. (D) Representative immunoblots of rbcs mutants probed with a serum containing polyclonal antibodies against Rubisco to illustrate the reduction or absence of each SSU subfamily. The LSU (55 kDa) and SSU (14.7 kDa for 1A and 14.8 kDa for the B subfamily) are shown. See Supplementary Table S5 for further results and statistical analysis.

Molecular characterisation of rbcs mutants

The expression profiles of the rbcS family and rbcL were quantified in T₃ plants for each rbcS mutant line. Transcript abundances were generally reduced for rbcS genes in lines targeted by specific T-DNA insertions or CRISPR/Cas9 editing (Supplementary Table S4). Consistent with previous observations, the relative expression of rbcL was more repressed in mutants with greater reductions in overall rbcS expression (i.e. 1a3b and 1a2b3b) (Wostrikoff and Stern, 2007; Izumi et al., 2012; Wostrikoff et al., 2012; Atkinson et al., 2017).

The total soluble protein content in leaves of all single mutants, and double mutants 1a2b and 2b3b, was similar to that of WT plants, but was significantly reduced by 20% and 82% for 1a3b and 1a2b3b, respectively (Fig. 3B;Supplementary Table S5). Leaf Rubisco content was generally decreased in single rbcS mutants compared with WT plants, and was significantly reduced in all double mutants and 1a2b3b (Fig. 3C). Specifically, 1a2b and 2b3b had Rubisco levels reduced by 42% and 38%, respectively, 1a3b by 61%, and 1a2b3b by 97% relative to WT plants. Disruption of 1A led to an absence of the lower 1A SSU (14.7 kDa) band in 1a, 1a2b, 1a3b, and 1a2b3b mutants as detected by western blot (Fig. 3D). Similarly, disruption of 3B resulted in a reduction in the intensity of the upper SSU band of the B subfamily genes (14.8 kDa) in 3b, 2b3b, 1a3b, and 1a2b3b mutants. In contrast, no observable reductions in upper band intensity were observed with mutants of the less expressed isoforms 1B and 2B. Nevertheless, the SSU band intensity of 1a2b3b was 30-fold lower than that of the overall intensity of the WT SSU bands.

Characterization of rbcs mutants under normal light and high light

Growth assays were performed under standard lab conditions for Arabidopsis plants (i.e. PPFD of 200 µmol photon m^–2 s^–1) and under near-outdoor levels of light (i.e. PPFD of 1000 µmol photon m^–2 s^–1) to characterize the growth performance of the rbcs mutants. Under standard lab conditions, the gene KO mutants for individual rbcS genes generated by CRISPR/Cas9 or T-DNA insertion (i.e. 1a, 1b, 2b, and 3b), as well as double mutants 1a2b and 2b3b, did not differ from WT plants in terms of rosette expansion rate, FW, DW, specific leaf area (SLA), maximum potential quantum efficiency of PSII (F_v/F_m), or chlorophyll content (Fig. 4; Supplementary Table S6). In contrast, 1a3b and 1a2b3b showed significant reductions in area, FW, DW, SLA, chlorophyll content, and F_v/F_m. 1a2b3b was more slow growing than 1a3b, but eventually developed to a similar level under standard lab conditions and produced viable seeds (Fig. 3A).

Fig. 4.

Growth phenotypes of rbcs mutants of Arabidopsis thaliana grown under standard and high light conditions. (A) Rosette area expansion of rbcs single T-DNA and CRISPR/Cas9 (CC) mutants, (B) rosette area expansion of double and triple mutants (the insert shows the expansion of 1a2b3b at appropriate scale), and (C) fresh and dry weights of 28-day-old rosettes of all mutant lines grown under standard conditions. (D) Rosette area expansion of 1b, 2b, and 3b T-DNA and CRISPR/Cas9 mutants, (E) rosette area expansion of 1a, 2b3b, 1a2b, and 1a3b mutants, and (F) fresh and dry weights of 28-day-old rosettes for all mutant lines grown under high light conditions. (G) Representative examples of 20-day-old rosettes of plants grown in standard conditions and (H) under high light conditions. Letters in the keys for (B) and (E) indicate a significant difference between plant lines (P < 0.05) as determined by repeated measures ANOVA followed by Tukey’s HSD tests. See Supplementary Table S6 for further results and statistical analysis.

Under high light, 1a and all three double mutants had significantly lower area, FW, DW, and chlorophyll content compared with the WT. A significant reduction in FW and DW was also observed for 3b mutants, while rosette area remained similar to that of the WT. As a result, 3b mutants also had a significantly higher SLA than the WT. High light was lethal to the 1a2b3b triple mutant, which did not survive past 15 d after germination.

The response of A to C_i under saturating light (A/C_i curves) was measured for all rbcs mutant lines grown under standard lab conditions (Fig. 5) as well as key photosynthetic variables (Table 2). Stomatal conductance to CO₂ at ambient CO₂ (g_s) and respiration rates in the dark (R_d) were the same in all lines, consistent with previous work showing that reductions in Rubisco do not affect stomatal behaviour or mitochondrial respiration in Arabidopsis (Atkinson et al., 2017). Single rbcs mutants for 1A and 3B showed significant decreases in the maximum rate of Rubisco carboxylation (V_cmax) compared with WT plants. In contrast, 1b and 2b mutants were similar to the WT. All three double mutants (1a2b, 2b3b, and 1a3b) had decreased V_cmax, and 1a3b also showed an increased substomatal CO₂ compensation point (Γ). However, 1a2b3b was significantly different from all other plant lines, with a 4-fold higher Γ value, and V_cmax at 10% of WT values.

Table 2.

Variables derived from photosynthetic CO₂ response curves, based on leaf gas exchange analysis

	V  cmax (μmol CO2 m–2 s–1)	R  d (μmol CO2 m–2 s–1)	g  s (mmol H2O m–2 s–1)	Γ (μmol CO2 mol–1)
WT	55.5 ± 3.5 a	1.01 ± 0.27 a	0.25 ± 0.03 a	52.4 ± 4.7 a
1a T-DNA	38.0 ± 3.2 bc	1.18 ± 0.11 a	0.20 ± 0.03 a	65.7 ± 10.1 ab
1a CC1	38.1 ± 2.0 bc	0.93 ± 0.08 a	0.25 ± 0.01 a	61.3 ± 1.0 ab
1b T-DNA	50.3 ± 5.2 ab	1.20 ± 0.11 a	0.24 ± 0.01 a	63.2 ± 3.2 ab
1b CC1-3	49.4 ± 1.4 ab	0.83 ± 0.08 a	0.28 ± 0.01 a	53.8 ± 1.3 a
2b T-DNA	48.2 ± 4.7 ab	1.17 ± 0.08 a	0.24 ± 0.07 a	60.1 ± 2.3 ab
2b CC1-3	48.9 ± 1.5 ab	0.81 ± 0.06 a	0.28 ± 0.02 a	52.7 ± 1.3 a
3b T-DNA	39.8 ± 2.0 b	0.73 ± 0.05 a	0.26 ± 0.01 a	52.5 ± 0.8 a
3b CC1-3	38.5 ± 1.5 bc	0.77 ± 0.03 a	0.27 ± 0.01 a	54.7 ± 2.4 a
2b3b	36.9 ± 3.2 bc	0.85 ± 0.07 a	0.25 ± 0.02 a	65.3 ± 2.9 ab
1a2b	36.3 ± 2.9 bc	0.78 ± 0.08 a	0.27 ± 0.01 a	64.7 ± 2.3 ab
1a3b	24.6 ± 1.0 c	0.71 ± 0.04 a	0.26 ± 0.01 a	70.0 ± 1.5 b
1a2b3b	5.4 ± 0.3 d	0.84 ± 0.13 a	0.23 ± 0.04 a	200.1 ± 3.8 c

Values are means ±SE of measurements made on three or four leaves from different plants (35- to 45-day-old non-flowering rosettes for the WT and mutants and 80-day-old non-flowering rosettes for 1a2b3b). Values followed by the same letters in the same column are not significantly different (P < 0.05) as determined by ANOVA followed by Tukey’s HSD tests. Abbreviations: Γ, substomatal CO₂ compensation point; g_s, stomatal conductance at 400 ppm CO₂; R_d, mitochondrial respiration in the light; V_cmax, maximum rate of Rubisco carboxylation.

Fig. 5.

Photosynthetic CO₂ response curves of rbcs mutants of Arabidopsis thaliana. Measurements were made on the fully expanded sixth or seventh leaf of 35- to 45-day-old non-flowering rosettes for the WT and mutants and 80-day-old non-flowering rosettes for 1a2b3b. The A/C_i curves show the response of net CO₂ assimilation (A) to different substomatal concentration of CO₂ (C_i) under saturating light (1800 µmol photon m^–2 s^–1) for (A) 1b and 2b mutants, (B) 1a and 3b mutants, and (C) 2b3b, 1a2b, 1a3b, and 1a2b3b mutants. Each value represents the means ±SE of measurements made on individual leaves from 3–4 different rosettes.

Discussion

Although the uptake of CRISPR/Cas9 in plant biology has increased dramatically in recent years, many aspects concerning efficiency are still unclear, especially as established guidelines for CRISPR/Cas9 in other biological systems are not necessarily applicable to plants (Liang et al., 2016; Hahn and Nekrasov, 2019). As the generation of stable plant transformants is time-consuming, we adopted a more rapid, transient expression approach using Arabidopsis protoplasts to test the efficiency of different gRNA pairs and select robust candidates (Fig. 2). All selected gRNA pairs were active in Arabidopsis and resulted in stable rbcs mutants. Thus, in agreement with previous studies (e.g. Li et al., 2013; Durr et al., 2018), screening gRNAs in protoplasts appears to be a reliable method for selecting functional gRNAs in planta. Overall, CRISPR/Cas appeared a viable approach to examine the impact of mutating rbcS isoforms and exploring their functional roles in plants.

CRISPR/Cas9 editing can be utilized to edit closely linked genes, which is not technically feasible through crossing T-DNA insertion lines, such as genes in the rbcS B subfamily. Here, the 2b3b double rbcs mutant was successfully generated by targeting homologous regions by the gRNA pair 1B3B. In addition, the triple mutant 1a2b3b was generated by targeting 3B in the 1a2b T-DNA line. Interestingly, disruption of 1B in conjunction with other rbcS genes was not accomplished using T-DNA insertion or CRISPR/Cas9 approaches despite successful generation of 1b mutants with both methods (Supplementary Fig. S4). 1B is the least expressed rbcS isoform and the mature peptide differs from 2B and 3B by only two amino acid residues (Izumi et al., 2012). Thus, it is unlikely that the structure of 1B has a unique impact on Rubisco activity (Valegård et al., 2018). Nevertheless, the temporal and/or spatial expression of 1B may contribute a specialized functional role that is critical for Arabidopsis development or fitness (i.e. during procreation). For example, previous work has shown that 1B localizes exclusively to the abaxial side of primordia and young leaves (Sawchuk et al., 2008). It would be interesting to further examine 1B localization in reproductive tissues (e.g. siliques or flowers). As a product of gene duplication, 1B is subject to selection pressure and is not retained in all accessions of Arabidopsis, although loss was correlated with disruption of the promoter (Schwarte and Tiedemann, 2011). Retention of 1B may be linked to sublocalization (Qiu et al., 2019). Differential expression patterns of rbcS genes in different organs have also been observed in other plant species including tomato, maize, tobacco, and rice (Wanner and Gruissem, 1991; Ewing et al., 1998; Morita, et al., 2014; Laterre et al., 2017). However, the extent of how important these organ-specific SSUs are has not yet been explored.

Gene editing via CRISPR/Cas9 resulted in a general decrease in mRNA abundance of target genes and significant reductions in Rubisco content for all 1a, 1a2b, 1a3b, and 1a2b3b mutants (Fig. 3; Supplementary Table S4). The observed reduction in mRNA levels was probably due to the presence of early stop codons that emerged from frameshift mutations, which consequently led to the degradation of mRNA through the nonsense-mediated decay process (Hug et al., 2015). Reductions in rbcS transcripts led to a reduction in the rbcL transcript but not to the same extent, as Arabidopsis rbcL transcript is controlled post-transcriptionally at the translation initiation process (Rodermel et al., 1996). Similar to Arabidopsis, rbcL synthesis in tobacco was shown to be partially independent of the rbcS transcript level, and LSU production was subject to state-dependent regulation of assembly that operated at the translational level (Wostrikoff and Stern, 2007).

rbcs mutants with a relatively small reduction in Rubisco content (i.e. <40% of WT levels) showed no change in growth rate and biomass accumulation relative to WT plants under standard lab growth conditions (i.e. PPFD of 200 µmol photons m^–2 s^–1) (Fig. 4; Supplementary Table S6). This was not unexpected as reductions in Rubisco content have previously been shown to be compensated by an increase in activation state of the remaining Rubisco pool under conditions that are non-limiting for Rubisco (Quick et al., 1991). However, under high light conditions (i.e. PPFD of 1000 µmol photons m^–2 s^–1), biomass accumulation (i.e. FW and DW) was significantly decreased in 1a and 3b mutants and all three double mutants (i.e. 2b3b, 1a2b, and 1a3b). This demonstrates that under near-outdoor levels of light, both 1A and 3B are critical for normal growth.

Differential expression of the Arabidopsis rbcS genes has been observed previously [e.g. under changing temperatures, light qualities, and CO₂ concentrations (Dedonder et al., 1993; Cheng et al., 1998; Yoon et al., 2001; Sawchuk et al., 2008)] but the impact of these differences on growth remains unclear. One potential use of our rbcS mutant suite could be to further examine the importance and interaction of specific SSU isoforms during development. All SSUs with the exception of 1B are increased in response to specific wavelengths of light (i.e. blue, red, and far-red light), but at differing relative gene expression levels (Dedonder et al., 1993; Sawchuk et al., 2008). Furthermore, 1A is the major isoform below 20 °C, whereas 3B expression is dominant at 30 °C (Yoon et al., 2001; Izumi et al., 2012).

The rosette area of 3b mutants was not decreased under high light compared with WT plants (Fig. 4; Supplementary Table S6). However, the observed reduction in biomass resulted in a significant increase in SLA and suggested that leaves were thinner in those lines. Furthermore, gas exchange measurements demonstrated a decrease in V_cmax for 3b mutants (Fig. 5; Table 2). Similar increases in SLA and reductions in photosynthetic capacity were observed for 1a and all three double mutants. These observations are in line with previous growth analyses of Arabidopsis Rubisco activase antisense lines under high light (Eckardt et al., 1997), where a reallocation of resources to expand leaf area and reduce thickness was observed when photosynthetic capacity was limiting (Hoshino et al., 2019).

For the most severely Rubisco-limited mutants, 1a3b and 1a2b3b, F_v/F_m was reduced under standard light (Supplementary Table S6). However, F_v/F_m for 1a3b was similar to that of the WT under high light, indicating that a reduction in Rubisco has less impact on the operating efficiency of the light reactions under high light. Thus, failure of 1a2b3b to grow under high light could indicate an inability of the light reactions to coordinate product utilization (i.e. ATP and NADPH) with the extremely low Rubisco content of the triple mutant.

CRISPR/Cas9 is a versatile tool that has been successfully used for genetic editing and the enhancement of breeding strategies in a wide variety of plant and crop species (Khumsupan et al., 2019; Wolter et al., 2019). This study has shown that CRISPR/Cas is a viable approach for characterizing the roles of SSUs in plant species and that Arabidopsis mutants lacking SSU isoforms are useful platforms for the study of functional roles of SSUs. In particular, the triple mutant 1a2b3b is potentially a powerful resource for studying the impact of heterologous SSU expression that has previously been studied in the 1a3b background (Atkinson et al., 2017). Unlike 1a3b, which has a Rubisco content of ~35% relative to the WT (Izumi et al., 2012; Atkinson et al., 2017), 1a2b3b retained only 3% under the conditions tested (Fig. 3; Supplementary Table S5), and is the first example of a plant line with a homogenous Rubisco pool consisting of single SSU and a single LSU isoform. Following complementation, further disruption of the remaining 1B isoform (e.g. by CRISPR/Cas9 editing using the gRNA pair 1BP2) could be used to generate a true hybrid Rubisco pool comprised of only heterologous SSU(s) and the native LSU. In addition, as the contribution of individual SSUs to the Rubisco enzyme is still unclear, the triple mutant could be exploited as a model to knock in native SSUs tagged with different fluorescent probes (Ishida et al., 2008). This method would allow for the visualization of composition of each hexadecamer in the Rubisco enzyme.

Supplementary data

Supplementary data are available at JXB online.

Table S1. List of CRISPR/Cas9 vectors constructed in this study.

Table S2. Sequences of synthetic oligonucleotides used in this study.

Table S3. Editing efficiency of individual gRNAs for each pair of gRNAs.

Table S4. Transcript abundances of the Rubisco gene family in rbcs mutants of Arabidopsis thaliana.

Table S5. Rubisco and total soluble protein contents for rbcs mutants of Arabidopsis thaliana.

Table S6. Rosette area and biomass of rbcs mutants of Arabidopsis thaliana grown at standard versus high light conditions.

Fig. S1. Synthesized nucleic acid sequence containing 3'-UTRs of rbcS genes used for RT–qPCR quantification.

Fig. S2. Light response curve of Arabidopsis WT plants.

Fig. S3. Segregation of a new 1a2b double T-DNA insertion mutant.

Fig. S4. Segregation ratios for progeny of 1a and 1b T-DNA insertion mutants of Arabidopsis thaliana.

Fig. S5. Screening for large deletion and indel events in rbcs mutants of Arabidopsis thaliana.

Fig. S6. A library of Arabidopsis thaliana rbcs mutants produced using CRISPR/Cas9.

Fig. S7. Sequence alignments of the target site for the 1BP2 CRISPR/Cas9 construct in 1a2b T-DNA mutants.

Data S1. Zip file containing sequence maps (.gb files) of the vectors used in this study.

Acknowledgements

This work was supported by the UK Biotechnology and Biological Sciences Research Council (grants BB/I024488/1 to EC-S, and BB/M006468/1 and BB/S015531/1 to AJM) and the Leverhulme Trust (grant RPG-2017–402 to AJM). PK was funded by a postgraduate research scholarship from the Darwin Trust of Edinburgh. We thank Louis Caruana (Lancaster University) for technical support.