Truncation of the calmodulin binding domain in rice glutamate decarboxylase 4 (OsGAD4) leads to accumulation of γ-aminobutyric acid and confers abiotic stress tolerance in rice seedlings

Abstract

GABA (Gamma-aminobutyric acid) is a non-protein amino acid widely known as major inhibitory neurotransmitter. It is synthesized from glutamate via the enzyme glutamate decarboxylase (GAD). GAD is ubiquitous in all organisms, but only plant GAD has ability to bind Ca²⁺/calmodulin (CaM). This kind of binding suppresses the auto-inhibition of Ca²⁺/calmodulin binding domain (CaMBD) when the active site of GAD is unfolded resulting in stimulated GAD activity. OsGAD4 is one of the five GAD genes in rice genome. It was confirmed that OsGAD4 has ability to bind to Ca²⁺/CaM. Moreover, it exhibits strongest expression against several stress conditions among the five OsGAD genes. In this study, CRISPR/Cas9-mediated genome editing was performed to trim the coding region of CaMBD from the OsGAD4 gene, to remove its autoinhibitory function. DNA sequence analysis of the genome edited rice plants revealed the truncation of CaMBD (216 bp). Genome edited line (#14–1) produced 11.26 mg GABA/100 g grain, which is almost nine-fold in comparison to wild type. Short deletion in the coding region for CaMBD yielded in mutant (#14–6) with lower GABA content than wild type counterpart. Abiotic stresses like salinity, flooding and drought significantly enhanced GABA accumulation in #14–1 at various time points compared to wild-type and #14–6 under the same stress conditions. Moreover, upregulated mRNA expression in vegetative tissues seems correlated with the stress-responsiveness of OsGAD4 when exposed to the above-mentioned stresses. Stress tolerance of OsGAD4 genome edited lines was evidenced by the higher survival rate indicating the gene may induce tolerance against abiotic stresses in rice. This is the first report on abiotic stress tolerance in rice modulated by endogenous GABA.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11032-024-01460-1.

Introduction

Rice is a crop of life all over the planet, with unique economic and cultural values. Global warming and concomitant climate change may create unfavorable environments for the production of food grains including rice (Ramegowda and Senthil-Kumar 2015). Adverse growing conditions have consistently resulted in crucial declines in the yield potential of crops (Datta et al. 2012). Adverse conditions such as salinity, flooding, and drought are compromising food security. To minimize yield loss, it is necessary to understand the underlying processes that take place within cells and plant responses to stresses. In the current study, we focused on the above-mentioned abiotic stresses in-vitro at early vegetative stage of rice seedling.

Gamma-aminobutyric acid (GABA) is a four-carbon, non-proteinaceous amino acid that exists universally in prokaryotic and eukaryotic organisms. It functions as a vital inhibitory neurotransmitter in the central nervous system of mammals (Curtis and Johnston 1974). GABA is metabolized through the “GABA shunt” pathway, which bypasses two steps of the tricarboxylic acid (TCA) cycle. GABA shunt enzymes have unique traits in plants that are missing in other organisms. A Ca²⁺/calmodulin (CaM)-dependent calmodulin binding domain (CaMBD) is often found at the C-terminus of plant glutamate decarboxylases (GADs), allowing plants to control the activation of GAD in response to biotic and abiotic stressors (Bouché and Fromm 2004). GABA biosynthesis commences with the decarboxylation of glutamate (Glu) into GABA by GAD in cells. A secondary pathway of GABA biosynthesis might be via polyamine degradation (Fait et al. 2008; Shelp et al. 2012). GAD genes in plants were isolated for the first time from petunia (Baum et al. 1993), and subsequently GAD homologs have been identified in other plants. Plant GAD normally comprises an extended region (mostly 30 amino acids long) at the C-terminus, referred as the calmodulin binding domain (CaMBD) (Baum et al. 1993; Yap et al. 2003).

To fight against various stresses GABA contributes to primary and secondary metabolic pathways in plants as a nitrogen metabolism intermediary (Ramos-Ruiz et al. 2019). Research has offered extensive evidence regarding the interactions between GABA and polyamines and other plant hormones, including auxins, gibberellins, cytokinin, ethylene, and abscisic acid, which are critical for plants in various stress circumstances (Podlešáková et al. 2019). Drastic accumulation of GABA in plant tissue was reported in multiple stresses such as salinity (Su et al. 2019), drought (Xu et al. 2021), and flooding (Souza et al. 2016). Furthermore, exogenous GABA has positive impacts on plant growth in both stressed and non-stressed conditions. According to Wang et al. (2017), exogenously applied GABA increased endogenous GABA biosynthesis, which conferred resistance to salt tolerance. During salt stress, a significant portion of assimilated carbon is diverted to respiration (Che-othman et al. 2020). Thus, it can be speculated that the GABA shunt couples with TCA cycle providing ATP needed for the cell to survive, because a high amount of ATP is needed to maintain ion homeostasis, osmotic balance, and for reactive oxygen species (ROS) scavenging.

Many studies have shed light on the involvement of GABA in abiotic stress tolerance in a wide range of plant species (Furlan et al. 2017; Liu et al. 2019; Su et al. 2019), but the number of studies performed in rice plants is limited. However, the Transcriptome ENcyclopedia of Rice (TENOR) database (https://tenor.dna.affrc.go.jp/) provides basic information of transcriptional activity in the rice genome at the nucleotide level based on the RNA-Seq data during multiple environmental stresses (Kawahara et al. 2016). Additionally, in the databank of RiceXPro (http://ricexpro.dna.affrc.go.jp/), the gene expression profile throughout the growth cycle of rice plant has been published (Sato et al. 2013). We adopted the concept of stage-specific abiotic stress experiments from these basic datasets.

It is worth noting that conventional methods of genetic modification are often time-consuming and labor-intensive. As a popular alternative, genome editing technology has offered a revolution in crop improvement, which enables a simple, efficient, and time-saving approach to make the desired modifications of target genes. Genome editing has advanced significantly in recent years with the introduction of CRISPR-Cas9 (Clustered Regularly Interspaced Short Palindromic Repeats-CRISPR-associated) systems to induce targeted mutations for studying gene function (Voytas and Gao 2014; Gao 2015). CRISPR/Cas9 is a booming genome editing tool in crop improvement (Mishra et al. 2018). Considering its ability to manipulate the genome of interest, earlier studies were conducted to modify GAD genes in tomato (Nonaka et al. 2017) and rice (Akama et al. 2020). A 7- to 15-fold increase in GABA accumulation was achieved by using CRISPR/Cas9 technology to boost the concentration of GABA in tomato fruits. This was done by removing the autoinhibitory domain of key enzymes involved in GABA biosynthesis (Nonaka et al. 2017). Moreover, Akama et al. (2020) effectively performed CRISPR/Cas9-mediated genome editing to remove the calmodulin-binding domain (CaMBD) from OsGAD3, which resulted in enhanced GABA content in grain that was seven times higher, as well as increased seed weight and protein content. This proven technology was employed in our study to develop OsGAD4 genome-edited mutants.

OsGAD4 has been reported to exhibit the strongest expression levels during several stress conditions (especially salinity, flooding, and drought) among the five OsGAD genes in the TENOR database (Kawahara et al. 2016). The expression level of OsGAD4 was highest in germinated brown rice with a giant embryo among the five OsGAD genes (Zhao et al. 2017), this information led us to consider that GAD4 is a potential focus for genetic manipulation of the GABA shunt. Thus, OsGAD4 was selected as our target gene for this study. The objective of the study was to produce GABA-enriched and stress-tolerant rice plants by genome editing of the rice GAD4 gene i.e., truncation of Ca²⁺/CaMBD in the C-terminal region of GAD4 and a subsequent detailed study to explore the response of the mutants upon exposure to adverse conditions.

Materials and methods

Plant materials and cultivation condition

Oryza sativa L. cv. Nipponbare (Ni) was used for rice transformation in this study. After threshing, rice seeds were surface sterilized in 75% (v/v) ethanol for 1 min with gentle shaking, followed by washing with double distilled water (ddH₂O) and then transfer to 50% (v/v) commercial bleach solution (Haitar; Kao Co., Ltd., Tokyo, Japan) for 30 min with continuous shaking. After rinsing 5 times with ddH₂O, seeds were sown in 0.5 × Murashige and Skoog (MS) solid media (Murashige and Skoog 1962). Then, the seeds were germinated in a 25 °C growth chamber (SANYO, Osaka, Japan), with a photoperiod of 16 h light/8 h dark. Later, the geminated rice seedlings were cultivated in a growth room with LED light (LCL-W5-1200, Churitsu Electric Corporation, Nagoya, Japan) at 25 °C temperature with the same light and dark condition. The plants were transferred to a naturally lighted non-contained greenhouse (40 m² in area) located at Shimane university, Matsue, Japan, as reported in Akama et al. (2009).

Ca²⁺/CaM-Binding assay

In-vitro Ca²⁺/CaM-Binding assay was performed using recombinant protein with OsGAD4-CaMBD and OsGAD3-CaMBD fusion protein essentially following the methods stated in Akama et al. (2001).

Construction of vector

Single guide RNAs (gRNAs) for genome-editing were designed using the CRISPR-P program (Lei et al. 2014) for truncation of the C-terminal region of OsGAD4: guide RNA (gRNA) in the coding region for the OsGAD4-CaMBD targeted upstream (gRNA-F1) and downstream (gRNA-R1). A couple of the synthesized 20-nucleotide target sequences for gRNA-F1 and gRNA-R1 (Fig. 2b) were annealed to make double-stranded DNA, respectively. Then inserted into the BbsI site of the gRNA cloning vector pU6gRNA (Mikami et al. 2015). A PvuII and AscI fragment from pU6gRNA carrying gRNA-R1was inserted into pU6gRNA carrying gRNA-F1 via the EcoRV and AscI sites respectively, resulting in pU6gRNA_F1 and R1. A PvuII and AscI fragment from this pU6gRNA derivative was inserted at the AscI and PlmI sites of a Ti plasmid pZH_gYSA_MMCas9 (Mikami et al. 2015) for rice transformation.

Fig. 2.

Gene structure and target region of genome editing in OsGAD4. a Schematic diagram showing the gene structure of OsGAD4 (Os03g0720300). Exons and introns are indicated by blue boxes and lines, respectively; CaMBD is shown as a pink box. UTR indicates untranslated region shown as a green box. Double slash between the first intron indicates the sequence gap in the longest intron of this gene. Upstream and downstream arrows direct the translation initiation and termination sites, respectively. The number in brackets denotes the distance of the exon sequence to the initiation codon (ATG). Scale bar = 100 bp. b Downstream region of OsGAD4 including the target site of CaMBD. Nucleotide position is indicated for 1639 to 1918. Amino acids are shown below nucleotide sequence. Upstream and downstream guide RNA consisting of a 20-nucleotide long sequence is denoted by F1 and R1 respectively. Black arrows show the targeted cleavage site in gRNA. An asterisk indicates a termination codon. The protospacer adjacent motif (PAM) sequence is highlighted in yellow, and amino acid sequences underlined in blue specify the CaMBD. Two termination codons are shown in red letters. The first one is an authentic termination codon and the later one is used after cleavage at the two cleavage sites and joining

Rice transformation

The binary vector was introduced into Agrobacterium strain EHA105 (Hood et al. 1993). Transformation of rice calli via Agrobacterium, selection of calli with N6D media supplemented with 50 mg/L of hygromycin B, and subsequent regeneration was done essentially as described by Ozawa (2009).

Analysis of transgene and PCR screening

For screening of transgenic lines, DNA lysates were prepared from leaves of young seedlings. A small section (2–3 mm²) of leaf was taken and placed in a 96-well PCR plate containing 50 μl of elution buffer (1 M KCl, 100 mM Tris–HCl [pH: 8.0], 10 mM EDTA) in each well. Then, it was boiled for 5 min at 95 °C.

KAPA (KAPA BIOSYSTEMS, United States) was used for amplification from the DNA lysate for screening purposes along with forward and reverse primers of the target region. PCR conditions were hot start at 95 °C for 3 min and then 38 cycles of denaturation at 95 °C for 15 s, annealing at 60 °C for 15 s, and extension at 72 °C for 20 s. The PCR products were analyzed using an automated microchip electrophoresis system MultiNA (Shimadzu Co. Ltd., Kyoto, Japan).

Leaf materials were used to isolate positive transgenic lines from the next generation by first and second screening. To analyze the transgenic plants, rice leaves were collected from the plants grown in the greenhouse to extract genomic DNA using the cetyltrimethylammonium bromide (CTAB) method (Murray and Thompson 1980). PCR analysis of the genomic DNA samples was carried out using 10 × Taq buffer (100 mM Tris–HCl: pH 9.0 at 25 °C, 500 mM KCl, and 15 mM MgCl₂ and 1% Triton X-100), Taq DNA polymerase, dNTP, and forward and reverse primers (F1 and R-356 respectively) (Supplementary Table S2).

GAD activity measurement

Upon obtaining seeds from genome edited line, GAD enzymatic assay was executed as described by Akama and Takaiwa (2007) with some modifications. Crude protein was extracted from rice powder (embryo and pericarp layer) using GUS isolation buffer comprised of 50 mM NaPO₄, pH 7.0, 0.1 mM β-mercaptoethanol, 10 mM EDTA, 0.1% sodium N-laurosyl sarcosinate, 0.1% Triton X-100. After mixing well with the buffer, crude protein was obtained in the supernatant after centrifugation. The protein concentration was measured using Qubit (Flurometer) (Thermo Fisher Scientific, USA) and stored for further experimental processes.

Amino acid analysis by gas chromatography/mass spectrometry

Amino acids were extracted from seeds and vegetative tissues by using 8% (v/v) trichloroacetic acid (TCA). After mixing and centrifugation of the solution, the supernatant was collected. Then it was mixed with diethyl ether with at least an equal volume of supernatant. After repeating this process twice, the remaining diethyl ether was removed by air drying to obtain the amino acids. GABA content from the substrate glutamate was measured using GABase enzyme (Sigma) in accordance with Akama et al. (2009).

Measurement of free amino acids along with GABA in T₂ plants were performed as explained in Kowaka et al. (2015). The concentration of each free amino acid was determined by GC/MS as follows: samples were derivatized using a dedicated EZ:Faast™ kit (Phenomenex, Torrance, CA), followed by analysis using a GC/MS-QP2010 system (Shimadzu Co. Ltd., Kyoto, Japan) with electronic pressure control and equipped with a split capillary inlet.

RNA extraction and RT-qPCR

Total RNA was extracted from vegetative tissues (shoot and root) of 2-week-old rice seedlings (wild-type Ni and genome-edited line #14–1 and #14–6) using Sepasol RNAI Super G (Nacalai Tesque, Japan). Single-stranded cDNAs coding for GAD4 and its truncations were synthesized from total RNA as the template using reverse transcriptase (ReverTra Ace, TOYOBO, Osaka, Japan). cDNAs were used for RT-qPCR analysis in an ECO Real Time PCR System (PCR max, United Kingdom). As an internal control, TATA-binding protein 2 (TBP2) was amplified with primers TBP2-F and TBP2-R. The results of RT-qPCR were calculated using the 2^−∆∆CT method (Livak and Schmittgen 2001).

Abiotic stress experiments

Before planning abiotic stress experiments, we separated the expression data of five GAD genes (retrieved from TENOR database). From the comparison it is obvious that OsGAD4 has prominent expression among five GAD genes (Supplementary Fig. S7).

For in-vitro stress induction, rice seeds from line #14–1 (C-terminal truncated version of CaMBD of OsGAD4) and #14–6 (small deletion at the CaMBD-coding region) were germinated and allowed to grow for 2 weeks in petri dishes containing 0.5 X MS media inside a growth chamber at 25 °C with a 16 h light/8 h dark period. Then the seedlings (Ni, #14–1 and #14–6 in each stress) were subjected to salinity, flooding, and drought stresses. For salinity stress, plants were held in 150 mM NaCl solution for a time course of 0, 1, 3, and 6 h. Samples of the vegetative tissues (shoot and root) were frozen in liquid nitrogen and preserved in -80 °C for further experiments. Flooding stress was applied by complete submergence of the seedlings in liquid MS media, and similar time lapse treatments (0, 1, 3, and 6 h) were carried out to prepare samples. On the other hand, drought stress was implemented by removing the plants from growing media and a longer duration of stress (0, 1, 3, 6, 12, and 24 h) was applied to measure the induction of responses against drought stress.

Additionally, abiotic stress tolerance of the wild type and mutant rice plants was checked by increasing the duration and strength of adversity. For high-salt treatment, 150 mM NaCl solution was used to dip the seedlings for 2 d and then removed from salt stress to grow for 17 d in small pots containing soil in normal conditions. In the case of flooding stress, plants were allowed to stay immersed in 1 × MS (liquid) media for 3 d in almost hypoxic conditions and thereafter removed from this condition and grown normally for another 17 d. Drought stress was applied by removing the plants from growth media and being left on a petri dish in the open air and at room temperature until the fresh weight of the wild-type plant (Ni) was reduced to approximately 25% (6 h). Afterwards, plants were rehydrated for 3 h for recovery and grown for 17 d similar to the other stress treated plants to check the survival rate. Moreover, biomass comparisons were performed by recording the fresh weight of stress surviving plants and kept overnight in an oven to dry and next day measured the dry weight.

Statistical analyses

All the results of different experiments were based on the value of three biological repeats. Statistical analyses were done with Student’s t-test in Microsoft Excel to determine the significance of differences. Analysis of variance (ANOVA) was performed to distinguish the differences among various treatments. Single and double asterisks denote the values are significant at P < 0.05 or P < 0.01, respectively.

Results

C-terminal region of OsGAD4 has the ability to bind to Ca²⁺/CaM

Eight amino acid sequences in the C-terminal regions of GADs from 3 plant species were compared (Fig. 1a). A Ca²⁺/CaMBD is found in almost all plant GADs in the C-terminal region, and it was first reported in Petunia hybrida GAD (PhGAD) to modulate the GABA biosynthesis in plants (Baum et al. 1993). On the other hand, the existence of the rice GAD2 isoform lacking a CaMBD also has been found (Fig. 1a) (Akama et al. 2001). As shown in Fig. 1a, very little similarity of the C-terminal extension in plant GADs is observed, but they have some common features, including several conserved clusters in each GAD. In particular, a highly conserved tryptophan (W) residue is found at the central position of the domain. Moreover, 2 to 4 lysine (K) and arginine (R) clusters can be seen at the N-terminus and C-terminus of GAD. The tryptophan (W) residue and lysine (K) clusters have crucial roles in binding CaM to the CaMBD (Arazi et al. 1995). According to a report by Yap et al. 2003, E476 and E480 work as pseudosubstrates of glutamic acid (Glu) in Petunia GAD and have autoinhibitory functions in the absence of Ca²⁺/CaM binding. Results such as these leave us in no doubt that the Glu residues are conserved in all the C-termini of rice isoforms, excluding OsGAD2, indicating a probable conserved function. Between two tea GADs (CsGAD), CsGAD2 does not have the ability to bind to CaM but it is upregulated by mechanical stress, whereas CsGAD1 can bind to CaM (Mei et al. 2016). From the sequence alignment, it is observed that CsGAD2 lacks one Glu (E471) and one Lys (K487) residue at the C-terminus, whereas CsGAD1 has the conserved motifs. The in-vitro Ca²⁺/CaM binding abilities of PhGAD, OsGAD1 and OsGAD3 have already been shown (Baum et al. 1993; Akama et al. 2001, 2020), thus we anticipated that OsGAD4 potentially has the same ability, because it contains all the characters that are required for binding to Ca²⁺/CaM. It was speculated from the structural features of OsGAD4-CaMBD that OsGAD4 is a common plant GAD that demonstrates Ca²⁺/CaM-dependent activation (Fig. 1a). In α-helical wheel analysis, all of the hydrophobic residues are grouped on one side of PhGAD, whereas residues with positive charges are on the opposite side (Fig. 1b). For effective CaM binding to the CaMBD of PhGAD, the Trp (W) residue and Lys (K) cluster are essential, which contribute to hydrophobic and electrostatic interactions, respectively (Arazi et al. 1995). At these crucial positions, OsGAD4 was identical to PhGAD, whereas OsGAD2 had a unique structure. Besides, CsGAD2 showed disparity in two conserved motifs in the putative essential domain. This α-helix analysis in Fig. 1b supports the assumption of Ca²⁺/CaM binding ability of OsGAD4.

Fig. 1.

Comparison of the C-terminal regions of plant GADs. a Multiple sequence alignment of the C-terminal regions of GADs in rice, petunia, and tea. Trp (W) and Lys (K) essential for in vitro binding to CaM are indicated by stars and a thick line, respectively. The positions of two pseudosubstrate residues E476 and 480 in PhGAD are indicated by rectangles, reported by Arazi et al. (1995). Os: Oryza sativa, Ph: Petunia hybrida, Cs: Camellia sinensis. OsGAD1 (AB056060), OsGAD2 (AB056061), OsGAD3 (AK071556), OsGAD4 (AK101171), OsGAD5 (AK070858), PhGAD (L16977), CsGAD1 (KT728367), CsGAD2 (KT728368). b α-Helical wheel projection of amino acid residues displayed using EMBOSS-pepwheel (https://www.bioinformatics.nl/cgi-bin/emboss/pepwheel). Hydrophobic residues are marked with squares, hydrophilic residues with diamonds, and positively charged residues with octagons. The hydrophilic area is clustered on the opposite side of the Trp (W)-centered hydrophobic residues (purple arrows). W and the Lys (K) cluster (black arrows) at the C-terminus are essential for the effective binding of CaM to the GAD-CaMBD. They contribute to hydrophobic and electrostatic interactions. The red arrowed Glu (E) residues serve as pseudosubstrates. c In-vitro CaM-binding assay with recombinant OsGAD4-CaMBD fusion protein and vector control. An expression vector carrying a DNA fragment for the C- terminal domain of OsGAD4 was constructed. Coding sequence for the C-terminal peptide of OsGAD4 cDNA was subcloned in-frame in pET32a (Novagen) and transformed into E. coli strain BL21 (DE3) pLysS. Recombinant protein carrying a polyhistidine-tag was induced for purification using a nickel-affinity resin. Expression of the fusion protein (OsGAD4-CaMBD), its purification and an in vitro CaM binding assay were performed in accordance with Akama et al. (2001). As a negative control, an empty expression vector (pET32a) was used. The protein was purified to incubate with bovine CaM agarose (Sigma) in the presence of Ca²⁺, then the agarose beads were washed with an excess amount of Ca²⁺-containing buffer four times. CaM-binding proteins were eluted with elution buffer containing EGTA three times. Protein samples were electrophoresed on a 15% SDS–polyacrylamide gel, followed by Coomassie Brilliant Blue staining. Lane 1: fusion protein in case of OsGAD4-CaMBD panel and vector encoding protein in case of vector control panel, Lanes 2–5: effluent fractions, Lanes 6–8: eluted fractions with EGTA-containing buffer

Ca²⁺/CaM -Binding Domain of OsGAD4 is capable of CaM binding

To confirm whether C-terminal domain of OsGAD4 possess the Ca²⁺/CaM -binding ability, an expression vector carrying a DNA fragment for the C- terminal domain of OsGAD4 was constructed to over-express in E. coli strain BL21 (DE3) pLysS and to purify recombinant protein. The protein was incubated with bovine CaM agarose beads in the presence of Ca²⁺. As a result of binding with Ca²⁺/CaM, it will be expected that the C-terminal region of GAD4 is eluted with EGTA-containing buffer, whereas the control protein expressed from only vector is detected on the reverse side. As shown in Fig. 1c, the recombinant protein with the C-terminal region of OsGAD4 was not detected in the effluent fractions (Lanes: 2 to 5) but was detected in the elution fractions (Lanes: 6 to 8; corresponds to recovery of first, second, and third elution with elution buffer with EGTA, respectively). On the contrary, no protein was detected in the elution fractions of the vector control. Moreover, for further confirmation of the result of this experiment, another expression vector carrying a DNA fragment for the C-terminal domain of OsGAD3 was used as a positive control (Akama et al. 2020). Similar experiments with the recombinant protein showed the evidence of having Ca²⁺/CaM-binding ability of the C-terminal domain of OsGAD3 (Supplementary Fig. S1). Whereas it was observed that without Ca²⁺, CaM/GAD interactions were drastically reduced both in case of OsGAD3 and OsGAD4 (Supplementary Fig. S1). Therefore, in-vitro experiment depicted that alike typical GADs in dicotyledonous plants, GAD4 has the ability to bind to Ca²⁺/CaM, suggesting that OsGAD4 is in fact a Ca²⁺/CaM-dependent enzyme.

In-vivo truncation of the C-terminal region of OsGAD4-CaMBD by genome editing

The structure of the OsGAD4 gene including exon/intron positions is presented in Fig. 2a, where the position of the CaMBD is presumed to be in the proximal region of the last exon. To remove the C-terminal extension of OsGAD4-CaMBD, guide RNAs (gRNAs) were designed, as shown in Fig. 2b. We predicted that the upstream and downstream cleavage of F1 and R1 will result in a 216 bp deletion (Fig. 2b). Subsequent DNA repair caused the deletion of almost the entire CaMBD and permitted the production of GAD4 protein without the C-terminal domain.

Agrobacterium-mediated plant transformation and screening after regeneration

Transformation of rice scutellum-derived calli was done using Agrobacterium. A binary vector harboring gRNAs and the Cas9 gene cassette was introduced into the Agrobacterium strain. In total, 27 independent transgenic lines (T₀) were obtained after the regeneration process.

Ten lines having shorter bands at around 180 bp along with the wild-type band 400 bp were selected as candidate lines after genome editing (Supplementary Fig. S2). DNA sequencing of the two alleles (Allele 1: 400 bp and Allele 2: 180 bp) in three independent T₀ line (#11, #14 and #26) was performed to know the exact mutation of the target site. In allele 1, two lines (#14 and #26) had the identical sequence as wild type, while the other had multiple substitutions (#11) leading to minor change in amino acid chain. On the other hand, allele 2 had the same deletion of 216 bp in the three lines indicating the efficient truncation of intact CaMBD (Supplementary Fig. S2). Seeds from all the T₀ lines were grown for further confirmation of desired mutants in segregating generation. Afterwards, seeds were harvested and grown to analyze targeted genome editing and GABA content determination.

For the purpose of PCR screening, seedling derived from each T₀ lines were considered. For instance, twelve seedlings (T₁) germinated from single panicle (#14 is shown in Fig. 3a) were analyzed by PCR. PCR analysis of DNA extracted from the candidate transgenic lines (T₁ generation) showed three different patterns of amplified bands (Fig. 3a). The first (A type) had a band at approximately 400 bp, which is close to wild-type Nipponbare; second (B type) with two bands at 400 bp and at approximately 180 bp; and third pattern (C type) with only one band at approximately 180 bp. The latter was the expected amplicon size for the genome-edited lines, resulting from the deletion of 216 bp. Among the 3 different band patterns, A represents close to wild type GAD4 mutants, B represents the heterozygous condition (bi-allelic) and C indicates the expected homozygous mutant i.e., genome-edited bands (Fig. 3a). Seedling numbers 1, 2, 3, and 7 were found to be A type; number 5, 6, and 9 were heterozygous B type, and numbers 4, 8, 10, 11, and 12 had the expected band size for homozygous genome-edited lines, C type.

Fig. 3.

Production of genome edited plant and analysis. a PCR screening of transgenic plantlets (T₁ generation) with MultiNA (Shimadzu, Kyoto, Japan). DNA lysates were prepared by a boiling method, using young leaves of seedling; then this lysate was used as a template for PCR. A DNA size marker (20 bp to 500 bp) was used to identify the PCR product size. Here, A represents the mutant type close to wild-type (band size is 400 bp), B represents the heterozygous mutant (bi-allelic band), and C indicates the homozygous mutant (a band with an expected truncated size of around 180 bp) i.e., genome-edited bands. Twelve seedlings derived from the seeds of the same panicle were analyzed by PCR. Seedling numbers 1, 2, 3, and 7 were found to be close to wild-type (A type); numbers 5, 6, and 9 were bi-allelic (B type), and the rest had the expected band size for genome-edited lines (C type). b Nucleotide and amino acid sequences of OsGAD4 transgenic plants; Wild-type Nipponbare (Ni) denotes the reference nucleotide sequence. The intended CRISPR/Cas9 cleavage sites are indicated by two black arrows (F1 and R1). Underlined text indicates the amino acid sequence from the CaMBD. Sequence gaps and amino acid gaps are indicated by double slash marks. c Nucleotide sequence of genome-edited lines numbers 1 to 6. In the genome-edited sequence, hyphens indicate the deletion; base pairs in parentheses indicate the length of the deletion. Small letters inside box indicate insertions. d Amino acid sequences of wild-type Ni and genome-edited lines (GE 1 to GE 6) resulted from the insertion and/or deletion of nucleotide sequences in the target region of CRISPR/Cas9, shown in Fig. 3c. The dots denote the amino acid sequence not shown in the polypeptide chain. The nucleotide sequence numbers 1 to 6 shown in Fig. 3c correspond to translated polypeptide chains GE 1 to GE 6 in Fig. 3d, respectively. In parentheses, the number along with AA indicates the length of the amino acid chain. DNA sequences were analyzed with GENETYX-MAC Software v.22.0.1 (GENETYX Corporation). e GABA content in grains of C (sequence number 1), B (sequence number 2) and A type (sequence number 3 to 6) mutant. The concentration of GABA was measured by GABase assay. Error bars indicate SD (n = 3). f In-vitro GAD enzymatic assay of total proteins extracted from Ni, #14–1 and #14–6 grains. This assay was performed to measure GABA production, in accordance with Akama and Takaiwa (2007) at physiological pH (pH 7). –Ca²⁺/CaM; without Ca²⁺/CaM, + Ca²⁺/CaM; with 0.5 mM Ca²⁺ and 0.1 μM bovine calmodulin (Sigma). Error bars indicate SD (n = 3). Two-way Analysis of variance (ANOVA) was used to determine the differences among various treatments. Asterisks indicate significant differences versus Ni (*P < 0.05, **P < 0.01). ns = not significant (no difference within groups)

After a series of transformation events, we obtained 259 T₁ transgenic plants in total. Out of these, 75 plants were found to be A type (29%), 114 plants were B type (44%), and 70 plants were C type (27%) (Supplementary Table S1a). Separated genotyping data showed the diverse pattern of segregation in T₁ generation (Supplementary Table S1b).

To confirm the genome editing, we have performed DNA sequencing in the T₁ generation and analyzed the sequence in comparison with a Ni reference. Wild-type Ni indicates the reference nucleotide (Fig. 3b); sequence analysis resulted in different types of genome-edited patterns is numbered 1 to 6. Progeny of #11, #14 and #26 lines were used for DNA sequencing. Here, sequence number 1 corresponds to type C (line #14), and sequence number 2 corresponds to the slower migrating band in type B (line #14) (Fig. 3a). The remaining sequences are all A type (line of #11, #14 and #26) with various sequence patterns. We observed alteration in amino acid sequence of wild-type Ni and genome-edited lines in the target region of CRISPR/Cas9 (Fig. 3d). Nucleotide sequences 1 to 6 shown in Fig. 3c correspond to translated polypeptide chains GE 1 to GE 6 in Fig. 3d, respectively. The deletion of the CaMBD resulted in a peptide with 9 artificial amino acids after the authentic N-terminal VVAN. The other mutations included longer polypeptide chains.

Subsequently, after getting different sequence patterns in the B, C and A type plants (T₁ generation), GABA content was determined from the grain. Diverse value of GABA concentration was observed in comparison to wild-type Ni (Fig. 3e). Variable GABA content is most probably an outcome of frameshift mutation leading to an extension of unexpected peptides in place of CaMBD. Rice grain of C type plant (GE 1: truncation of CaMBD) yielded highest GABA content compared to B and A type plants; thus, these plants were subjected to further analysis in T₂ generation. Moreover, among C type plants, we analyzed T₁ seeds derived from #11, #14 and #26 lines. Out of these the highest GABA content (0.57 nmol/mg) was found from line number #14 (named as #14–1), in comparison with #11 (0.40 nmol/mg) and #26 (0.43 nmol/mg).

Therefore, mutant line #14–1, having the nucleotide sequence as no. 1 in Fig. 3c and corresponds to the amino acid sequence of GE1 in Fig. 3d was selected to be used in following experiments. Moreover, another mutant line, namely #14–6 derived from short deletion of 14 bp and 8 bp separately for CaMBD-coding region as shown in no.6 in Fig. 3c resulting in polypeptide chain of 53 amino acids (GE 6 in Fig. 3d) was used to evaluate the truncation effect on phenotypes and stress response.

In vitro enzyme activity from total protein extracted from WT, #14–1 and #14–6 seeds

To examine GAD enzymatic activity at physiological pH in the presence or absence of Ca²⁺/CaM, we isolated crude protein extract from rice grains, which was utilized in the GAD enzymatic reaction. Crude protein of #14–1 exhibited higher GAD activity compared with its intact authentic extract (Ni) and #14–6, possibly because of higher activity of GAD4ΔC in #14–1 (Fig. 3f). At physiological pH (pH 7), Ca²⁺/CaM induced 1.3-fold higher activity in wild-type Ni, whereas GAD activity was higher in #14–1 with or without Ca²⁺/CaM. GAD activity increased 2.3- and 2.2-fold in #14–1 at pH 7 in the absence or presence of Ca²⁺/CaM in comparison to wild-type control without Ca²⁺/CaM. This implies that the C-terminal domain functions as a potential autoinhibitory domain in OsGAD4; as a result, when this domain is truncated, the enzyme acts constitutively and exhibits increased activity. On the other hand, GAD activity was almost similar in #14–6 as wild type Ni. Notably, there was an increasing trend of GAD enzymatic activity in Ni wild-type seeds in the presence of Ca²⁺/CaM in contrast to the absence of that. RT-qPCR analysis was performed to confirm that the elevated GAD enzyme activity in genome edited seeds is resulted by truncated OsGAD4 (Supplementary Fig. S4). Expression level of four GAD genes tested revealed almost alike pattern in all the tissues of wild-type and truncated GAD4 mutant, meaning no other GAD gene expression is upregulated in the plant tissues of truncated GAD4 mutant.

Measurement of free amino acids

The concentration of free amino acids (T₂ generation) was measured using gas chromatography-mass spectrometry (GC/MS) (Table 1). The quantities of Asn and Trp were significantly lower in the #14–1 line compared with the wild-type, whereas most other free amino acids levels were increased. Among the proteinaceous amino acids, the accumulation of Val, Ile, Leu, Glu, and Phe was significantly higher compared with wild-type Ni. The highest GABA content was found in the line #14–1, which was almost 9-times higher compared with wild-type Ni. However, the other mutant #14–6 yielded much lower GABA than wild type but found to produce significantly higher Leu, Pro and Glu.

Table 1.

The concentration of free amino acids in rice grains (T₂) (nmol/g grain) determined by GC/MS

Amino acid	Ni	#14–1	#14–6
Ala	56.1 ± 4.1	113.2 ± 7.1** (2.0)	72 ± 11.7 (1.3)
Gly	15.9 ± 8.3	24.8 ± 1.6 (1.6)	12.5 ± 3.7 (0.8)
Val	7.3 ± 0.8	39.1 ± 2.1** (5.4)	11.9 ± 5.2 (1.6)
Leu	2.4 ± 0.6	12.9 ± 0.2** (5.3)	12.6 ± 1.6** (5.2)
Ile	2.3 ± 0.1	12.2 ± 0.4** (5.4)	6.6 ± 2.4* (2.9)
Ser	15.9 ± 1.5	31.1 ± 1.6** (2.0)	15.1 ± 1.8 (0.9)
Pro	13.7 ± 2.8	8.7 ± 3.8 (0.6)	34.9 ± 6.1** (4.0)
Asn	134.2 ± 15.0	44.8 ± 12.1** (0.3)	128.6 ± 26.4 (1.0)
Asp	92.1 ± 26.0	271.4 ± 21.6** (2.9)	141.7 ± 51.5 (1.5)
Met	42.4 ± 7.1	110.3 ± 9.7** (2.6)	59.7 ± 10.7* (1.4)
Glu	149.0 ± 7.8	804.6 ± 32.2** (5.4)	313.3 ± 20.7** (2.1)
Phe	2.1 ± 0.1	9.6 ± 1.0** (4.5)	16.7 ± 5.7* (7.8)
Gln	22.2 ± 1.6	48.0 ± 4.4* (2.2)	27.9 ± 4.7 (1.3)
His	17.7 ± 2.4	76.8 ± 9.1** (4.3)	18.2 ± 7.6 (1.0)
Tyr	7.6 ± 5.0	2.3 ± 1.2 (0.3)	6.0 ± 1.9 (0.8)
Trp	8.6 ± 1.0	1.7 ± 0.4** (0.2)	1.6 ± 0.6** (0.2)
GABA	14.7 ± 1.2	129.8 ± 20.9** (8.8)	11.3 ± 3.0 (0.8)

Value: average ± standard deviation (SD); data comprise of three biological replicates

^*P < 0.05, **P < 0.01 versus Ni control; Data in parentheses are fold change in comparison to wild-type Ni

GABA content in vegetative tissues was measured using GABase assay. GABA concentration in line #14–1 was compared with wild-type Ni (Supplementary Fig. S6). Stem and root tissues had higher GABA contents than wild-type Ni. Among the vegetative tissues of line #14–1, root tissues accumulated the highest GABA levels. Whereas all the vegetative tissues remained low in GABA content in case of #14–6 compared with Ni.

Abiotic stress induced GABA accumulation in vegetative tissues

Earlier studies demonstrated that abiotic stresses increase endogenous GABA concentrations in plant tissues, although the rate of its accumulation varied widely (Li et al. 2021). We observed the responses to stresses in rice seedling at an early vegetative stage. Firstly, salt stress treatment of 2-week-old rice seedlings was performed in vitro using 150 mM NaCl solution for various time periods. Secondly, flooding stress treatment of rice seedling was performed by completely submerging the seedlings in liquid MS media, and finally seedlings at the same age were allowed to grow in dehydrated condition without any media for drought stress treatment.

Salt stress induced accumulation of GABA in shoot tissues of line #14–1 after 1 h of treatment, which was higher than wild-type Ni (Fig. 4a). GABA levels were reduced after 3 h of treatment and climbed again at the highest point after 6 h of stress treatment, reaching 2.6-fold higher compared with wild-type Ni. In root tissues, GABA induction demonstrated a gradual increase with the duration of stress and rose to maximum at 6 h, producing an extreme yield of almost 4.3-fold higher GABA levels compared with wild-type. On the contrary, #14–6 had shown a considerable increase in GABA content of root tissue after 3 h of salt treatment that was almost 1.4-fold in comparison to wild type.

Fig. 4.

Response to abiotic stresses as GABA accretion. a GABA accumulation in vegetative tissues (shoot and root) of 2-week-old Ni, #14–1 and #14–6 seedlings when challenged with salinity. For salinity stress, plants were placed in 150 mM NaCl solution for a time course of 0, 1, 3, and 6 h. b Flooding stress was implemented by complete submergence of the seedlings in liquid MS media and similar time lapse treatments of 0, 1, 3, and 6 h and c Drought stress was carried out by removing the plants from the growing media and involved a longer duration of stress (0, 1, 3, 6, 12, and 24 h) to record the induction of response. Ni = wild-type control; #14–1 = OsGAD4 genome-edited plant with intact deletion of CaMBD; #14–6 = OsGAD4 genome-edited plant with short deletion in the coding region for CaMBD. Error bars indicate SD (n = 3). Levels of significance were determined by comparing with the wild-type values in the same stress conditions. Asterisks indicate significant differences (*P < 0.05, **P < 0.01)

With flooding treatment, GABA accumulation started to rise drastically in root tissues of line #14–1 and it reached at peak of approximately 3.3-fold higher GABA accumulation in comparison to wild-type Ni at 1 h (Fig. 4b). After this, GABA levels declined at 3 h and 6 h of flooding stress. In shoot tissues, there was similar trend of GABA production, producing significantly higher GABA levels after 1 h. Flood stress response in short deletion mutant (#14–6) was observed almost at similar level as wild type in both vegetative tissues.

Furthermore, drought stress was imposed in young rice seedlings to evaluate the response in GABA accumulation. Shoot tissues exhibited elevated GABA concentration in response to drought stress starting from 6 h, with a slight reduction at 12 h, and again showing greater accumulation at 24 h (Fig. 4c). At this time point, line #14–1 produced approximately 2.2-fold higher GABA compared with wild-type and line #14–6 was found with enhanced GABA accumulation as well. Additionally, in root tissues of #14–1, drought stress induced a gradual rise in GABA content with time, reaching its maximum level at 24 h.

Taken together, it can be seen that GABA induction in vegetative tissues of young rice seedling showed diverse responses to different abiotic stresses, and GABA accumulation was augmented in vegetative tissues to respond to abiotic challenges.

mRNA expression levels are upregulated in vegetative tissues upon exposure to abiotic stresses

To explore changes in the expression levels of OsGAD4, reverse transcription quantitative polymerase chain reaction (RT-qPCR) of vegetative tissues was done at different time points while exposed to the same abiotic stresses as for GABA content measurement. Here wild-type Ni, #14–1 and #14–6 were analyzed but having the same promoter region they exhibited almost identical expression levels. In the case of salt stress, mRNA expression levels were induced significantly in root tissues at 1 h, then a slight drop at 3 h (Fig. 5a). Strong expression at 6 h reaching its peak at 3.1-fold compared with the untreated control. Similarly, shoot tissues showed consistent expression levels as root tissues, with comparatively lower expression levels at the same time point in response to salt stress.

Fig. 5.

OsGAD4 gene expression in response to abiotic stresses a salinity, b flooding, and c drought in genome-edited rice plants. The stress treatments and durations are described in the text. The expression level was analyzed by RT-qPCR. TATA-binding protein 2 (TBP2) was used as an internal control. Relative expression level was analyzed using the 2^−ΔΔCT method. Error bars indicate SD (n = 3). Levels of significance were determined by comparing values of stress treated samples to the respective non-stressed condition. Asterisks indicate significant differences (*P < 0.05, **P < 0.01). The asterisk above the bar indicates level of significance in three lines simultaneously compared to the non-stressed control

During flooding stress, mRNA expression levels were upregulated rapidly at 1 h, with an approximately 3.1-fold increase in root tissue compared with the untreated control (Fig. 5b). Surprisingly, in the same stress condition, shoot tissues did not show much upregulation, though levels were higher compared with the control. After this time point, the expression in root tissues declined slightly up to 3 h but then increase at 6 h. Again, a similar tendency was observed in shoot tissues, with a trend of declining expression at 3 h but then significantly increased expression at 6 h.

With respect to drought stress, both shoot and root tissues exhibited steady upregulation of expression at the onset of stress treatment at 1 h (Fig. 5c). At the 3 h time point, expression levels had a trend towards a decrease in shoot tissues, almost similar to untreated samples. However, root tissue showed increased expression at this point. Eventually, with an increase in duration of stress the expression level increased moderately and reached a maximum at 24 h. In shoot and root tissues, the increase was 7.5-fold and 6.3-fold, respectively, compared with the untreated control. Our mRNA expression data is mostly in compliance with the gene expression profile or transcriptional activity of the TENOR database (Kawahara et al. 2016).

Abiotic stress tolerance was significantly improved in #14–1 rice plants

The tolerance level of OsGAD4 genome-edited plants to salt, flooding, and drought stresses were examined compared with that of wild-type plants. Two-weeks old seedlings were treated with multiple stresses as described in the material and method section. After stress treatment, the seedlings were washed and rehydrated for 3 h before transferring to small pots containing soil (Fig. 6). Salt stress tolerance in line #14–1 was demonstrated in our experiment using 150 mM NaCl solution. The survival rate of line #14–1 was 44.2% against salt stress, whereas 17.3% and 15.2% of wild-type and #14–6 plants respectively survived the stress (Table 2). For flooding stress, 25.6% of wild-type plants survived, and line #14–1 showed a survival rate of 59.6%. With drought stress, 34.5% of line #14–1 seedlings endured the stress, whereas the survival rate of wild-type was only 12.2%, besides line #14–6 showed even less survival rate of 11.7% (Table 2). Our results showed that genome-edited mutant of the truncated version of CaMBD OsGAD4 significantly improved abiotic stress tolerance at the early vegetative stage.

Fig. 6.

Abiotic stress (salinity, flooding, and drought) tolerance of wild-type, #14–1 and #14–6 seedlings at the early vegetative stage. The stress treatments were carried out as described in the text. High-salt treatment: 150 mM NaCl solution was used to dip the seedlings for 2 d and then removed from the high-salt conditions to grow for 17 d in small pots containing soil in normal conditions. Flooding stress: plants were allowed to stay immersed in liquid MS media for 3 d and thereafter removed and grown normally for another 17 d. Drought stress: the plants were removed from growth media and left on a petri dish in the open air and at room temperature until the fresh weight of the wild-type plant (Ni) was reduced to approximately 25% (6 h). Afterwards, plants were rehydrated and grown for 17 d. For each line, 12 seedlings were used in one replication of an individual experiment. Independent stress experiment was repeated at least 3 times. Scale bar = 10 cm

Table 2.

Quantitative data on survival rate and biomass reduction after the seedlings survived abiotic stresses

Type of stress	Survival rate (%)			Biomass loss % (FW)			Biomass loss % (DW)
	Ni	#14–6	#14–1	Ni	#14–6	#14–1	Ni	#14–6	#14–1
Salinity	17.3 ± 8.1	15.2 ± 10.0	44.2 ± 12.3**	33.1 ± 4.2	33.5 ± 7.4	10.8 ± 3.7**	27.9 ± 14.5	31.4 ± 9.5	10.2 ± 6.2**
Flooding	25.6 ± 6.3	21.3 ± 5.2	59.6 ± 20.1*	25.9 ± 7.8	28.0 ± 8.6	10.2 ± 1.7**	22.3 ± 8.4	23.8 ± 6.1	12.1 ± 7.9*
Drought	12.2 ± 4.8	11.7 ± 4.6	34.5 ± 9.7**	35.6 ± 4.2	36.7 ± 3.7	15.0 ± 3.3**	29.1 ± 4.0	29.4 ± 5.7	9.6 ± 1.8**

DW dry weight, FW fresh weight. The values are calculated from averages of the surviving plants in each stress with wild-type, line #14–1 and line #14–6. The average number of surviving plants was converted to percentage to show the survival rate. These data represent the combined result of two independent experiments each with three replications in each stress condition. In the case of biomass, the control or untreated fresh and dry weight was assumed as 100%. Biomass loss was calculated by deducting the biomass of stress surviving plants from that of the untreated one. Asterisks indicate significant differences compared to wild-type (*P < 0.05, **P < 0.01)

Biomass comparison was done by measuring the fresh weight and dry weight of the surviving plants. Fresh weight was measured after removal from soil and washing to remove dirt. For measurement of dry weight, the cleaned plant sample was dried overnight in an oven at 45 °C. Results revealed that biomass reduction as fresh weight in #14–1 plants in the three stress conditions was significantly lower than that in the wild-type plants (Table 2). Similarly, dry weight loss was more in wild-type plants compared with #14–1 plants in response to the abiotic stresses, suggesting a protective mechanism enabled #14–1 plants to tolerate the adverse conditions. While biomass loss in #14–6 was found almost similar to wild type predicting its inability to defend the plant against stress conditions.

Our current results indicated that elevated GABA levels are possibly associated with abiotic stress tolerance in plants by playing a protective role. The underlying mechanism needs to be studied further.

Discussion

In the current study GABA-enriched rice plants were generated. CRISPR/Cas9-mediated genome editing was performed to truncate the coding region of CaMBD from the OsGAD4 gene. Comparison of the C-terminal region of plant GADs provided the foundation to predict the putative features of OsGAD4 including having an authentic CaMBD and Ca²⁺/CaM-induced activation (Fig. 1a). Earlier studies with OsGAD1 proved that it can bind to CaM; however, mutants of OsGAD2 lacking the C-terminal extension had higher GABA levels but cannot bind CaM (Akama et al. 2001; Akama and Takaiwa 2007). In addition to rice, unique structures of GADs without an authentic CaMBD have been confirmed in apple and tea (Trobacher et al. 2013; Mei et al. 2016). Because of a lack of two or more essential residues compared with authentic GAD in PhGAD (Fig. 1b); OsGAD2, and CsGAD2 are unable to bind Ca²⁺/CaM and are not activated by CaM (Akama et al. 2001; Mei et al. 2016). Truncation of the C-terminal coding region of OsGAD3 resulted in seven-fold higher GABA accumulation in rice grains compared with wild-type (Akama et al. 2020). In vitro experiment proved that OsGAD4 is in fact a Ca²⁺/CaM-dependent enzyme having ability to bind to Ca²⁺/CaM (Fig. 1c).

Previously, transgenic strategy for C-terminal deletion resulted in enhanced GABA content in several species, such as tobacco, tea, and tomato (Baum et al. 1996; Mei et al. 2016; Takayama et al. 2017). Therefore, we planned a complete truncation of the coding region for CaMBD using the CRISPR/Cas9 approach, by means of in vivo mutagenesis. Upstream and downstream gRNA were designed to obtain an intact deletion of the domain (Fig. 2b). We managed to establish 27 lines (T₀), among which we selected 10 candidate lines (Fig. 3a). Out of these candidate lines (259 seedlings) which were analyzed at the DNA level, 70 seedlings (Supplementary Table S1b) contained the expected 216 bp deletion (27% of total). The remaining lines were found with variable number of base insertions or simultaneous insertions and deletions resulting in different polypeptide chain (Fig. 3c-d).

Phenotypic difference (variable GABA content) was observed in the lines with diverse nucleotide sequences (Fig. 3e) which may be caused by frameshift mutation for multiple reasons such as the production of an altered protein completely different from CaMBD. Several patterns of InDels in the sequences leads to different length and function of protein. Altered protein might have less ability to recover the autoinhibition of the GAD enzyme. That may regulate the GABA biosynthesis in these mutants independently and produce almost same level or lower GABA compared to wild-type with the exception of intact truncated version of CaMBD. Similarly, Nonaka et al. (2017) reported that frameshift mutations just upstream of CaMBD-coding sequence in genome edited tomato GAD3 resulted frequently in an extension of altered peptides, mostly showing the same level of GABA accumulation as wild-type.

In earlier study of OsGAD3, 41% of lines had the desired truncation of 122 bp (Akama et al. 2020), but the rest had shorter mutations. Whereas targeted mutagenesis in tomato produced deletions in 59.9% of line, varying from a single nucleotide to more than 200 nucleotides in the T₀ generation, with the remaining mutations comprises single insertions and simultaneous insertions and deletions (Nonaka et al. 2017).

GAD enzymatic activity data demonstrated that wild-type Ni increased 1.3-fold with GAD activity modulated by the presence of Ca²⁺/CaM, whereas truncation of the CaMBD from OsGAD4 resulted in 2.3- and 2.2-fold increases in GAD activity at pH 7.0 with or without Ca²⁺/CaM, respectively (Fig. 5). Decreased GAD activity in #14–6 could be observed because of the small deletion in the coding region of CaMBD, which might interfere with essential residues required for the GAD and calmodulin interaction. However, ANOVA results indicate that #14–1 had significantly high enzymatic activity in contrast to Ni and #14–6. Whereas GAD activity between each genotype was not statistically significant in the presence or absence of Ca²⁺/CaM. Extreme enzymatic activity was revealed in crude protein extracts of an OsGAD2 truncated mutant in contrast to wild-type (Akama and Takaiwa 2007). Moreover, another study reported a rise in activities in truncated tomato GADs at C-terminal positions, GAD3ΔC; 16-fold at pH 7.0 compared with the full-length tomato GAD (Nonaka et al. 2017). These results suggested that the C-terminal extension region of SlGAD3 also functions as an autoinhibitory domain, like that of other plant GADs (Baum et al. 1996; Akama and Takaiwa 2007). Additionally, removal of the C-terminal extension in GADs has enhanced the activity of these enzymes in other plant species such as apple (Trobacher et al. 2013). Hence it can be assumed that in neutral pH conditions GAD activity is enhanced in a Ca²⁺/CaM-dependent manner and truncated version of GAD4 showed much higher activity.

Moreover, varying gene expression was observed in leaf, stem, root and seed tissue of GAD homologues, which is mainly tissue specific. (Supplementary Fig. S4) Most importantly, individual GAD gene expression wild type Ni and #14–1 was found similar, suggesting there is no upregulation of expression level to compensate truncated version of GAD4. Hence, it can be evidently said that increased GAD activity in crude protein of whole truncated version of OsGAD4 genome edited line (#14–1) was the result of truncation of CaMBD.

As anticipated, the GABA content in line #14–1 was almost ninefold higher than wild-type Ni (Table 1). In our study, an increase in most of the amino acids was seen in genome-edited line, with exception in Pro, Asn, Tyr and Trp. These enhancements in analyzed amino acids seemed to be closely associated with that of GABA, roughly proportional to GABA. On the contrary line #14–6 was observed to produce almost similar level of amino acids and less amount of GABA however unexpectedly yielded higher concentration of Leu, Pro and Glu compared to wild type. Akama et al. (2020) also reported similar results in a transgenic line where the C-terminal region of CaMBD was truncated (GAD3ΔC) and mutant with single base deletion. The accumulation of GABA in rice grains, which serve as sink organs, may differentially affect the composition of the amino acid pool (Akama et al. 2009). The accumulation of other free amino acids may be triggered by an increase in GABA levels in rice grains, indicating a possible connection between GABA and amino acid accumulation.

Kisaka et al. (2006) reported the development of transgenic tomato fruits with antisense-suppressed GAD expression. In three out of four transgenic lines, an amino acid study of the fruit showed an increase in both Glu and other free amino acids. Apparently, genetic modification of the GABA shunt in rice can result in an increase in other amino acids as well as stable and high levels of GABA accumulation in the rice kernels. The stable accumulation of GABA and other amino acids in the edible portions of the rice was caused by the modification of GABA biosynthesis and catabolism in rice, especially the suppression of GABA-transaminase (GABA-T) (Shimajiri et al. 2013a).

In 2017, Nonaka et al. demonstrated the modification of two SlGAD genes by removing the coding region of CaMBD by means of a CRISPR/Cas9 strategy. Seven- to 15-fold more GABA accumulation was seen by their targeted mutagenesis in tomato fruits in comparison to wild-type. Genome-edited rice (#14–1) developed in the current study yielded almost ninefold higher levels of GABA than wild-type. Whereas the C-terminal truncation of OsGAD3 resulted in sevenfold higher GABA accumulation in rice grains (Akama et al. 2020). Both rice GAD3 and GAD4 undergoing similar genome editing approaches yielded variation in GABA production. The reason behind this can be different transcriptional regulators or signaling molecules, for example, can modify the expression or activity of GAD gene. Evidence showed the stronger transcription of GAD3 in rice seeds compared with GAD1 and GAD2 (Liu et al. 2005).

Seemingly, high GABA accumulation in genome edited line correlates with GABA metabolism which involves balance between higher synthesis by GAD4 and constant degradation by GABA-T. Rice genome has four GABA-T and five GAD genes, where three and four genes predominantly expressed, respectively (Shimajiri et al. 2013b; Akama et al. 2020). To rule out the potential possibility such as somaclonal variation which may result in increase in GABA content, RT-qPCR analysis was performed to reveal that all the GABA-T and GAD expressions tested remain consistent in wild type, #14–1 and #14–6 line (Supplementary Fig. S4 and Fig. S5). Therefore, elevated GABA levels can be attributed to higher GAD enzymatic activity in #14–1 compared with wild type (Fig. 3f).

We observed significant GABA accumulation specially in the roots of #14–1 seedlings (4.3-fold) in response to a 6 h duration of salt stress compared with wild-type subjected to the same stress conditions (Fig. 4a). Banerjee et al. (2019) reported a drastic increase in endogenous GABA content (3.7-fold compared with non-treated lines) in the Kalonunia (KN) cultivar due to salt stress subsequently it plays a vital role in regulating stress tolerance. The rise in GABA shunt activity offers an additional carbon source for the TCA cycle to function in mitochondria and by-pass salt-sensitive enzymes to help wheat plants increase leaf respiration (Che-Othman et al. 2020), consequently alleviating salt damage to plants. Again, flooding stress induced higher GABA deposition in GAD4 genome edited root tissues (3.3-fold) after 1 h of treatment compared with wild-type (Fig. 4b). These data are in agreement with the previous study where GABA noticeably accumulated along with other hypoxia-induced metabolites in the root systems of grapevine (Ruperti et al. 2019) when submerged in water. Activation of the GABA shunt triggered an increase in GABA in response to hypoxia (António et al. 2016). In the present study, we recorded notably higher GABA accumulation in shoot tissues of line #14–1 after 24 h of drought stress, which was 2.2-fold in contrast to wild-type in the same stress conditions (Fig. 4c). Elevated GABA deposition in gad1/2 × pop2 triple mutants has been demonstrated in Arabidopsis (Mekonnen et al. 2016) as a drought stress response that is linked with regulation of the stomatal aperture, preventing water loss. Increased accumulation of GABA in different forms of stress suggested that it is not used by the Krebs cycle and instead plays a signaling role (Jantaro and Kanwal 2017; Carillo 2018). Conversely, low GABA content in stress conditions for mutant with a short deletion in coding region for CaMBD (#14–6) most likely occurs from a combined effect of reduced enzymatic activity and calmodulin interaction impairment.

Information about endogenous GABA-induced gene expression in rice under abiotic stress conditions are scanty so far. Our RT-qPCR results showed that upregulated expression of the OsGAD4 gene was induced in root tissue after 6 h of salinity stress reaching 3.2-fold higher than the untreated control (Fig. 5a). Unexpectedly, the expression pattern of OsGAD4 in root was not consistent with the GABA content, mRNA level at 3 h in root tissue was decreasing, while GABA accumulation increased. The potential reason for this discrepancy could be the compensatory mechanisms involve the regulation of other GAD isoforms like OsGAD1 and/or OsGAD3 expression under that specific salinity condition.

A previous study concluded that in comparison to the control, salt stress significantly increased the expression of SlGAD1-3 in leaves of tomato seedling (Wu et al. 2020) when applied in combination with exogenous GABA. Compared with sole treatment with salt or GABA, integrated treatment of these two boosted gene expression levels. This result implied that both salt and exogenous GABA had an additive impact on the transcriptional and metabolic levels of endogenous GABA content. In response to flooding stress, mRNA expression was upregulated swiftly in the first hour, by almost 3.1-fold in root tissue compared with the untreated control (Fig. 5b). The strongest upregulation of GAD2 was reported among five GAD isoforms in soybean after flooding stress which is supported by concomitant enhancement of GABA accumulation (Souza et al. 2016). However, drought stress showed extreme responses in mRNA expression (7.5-fold) after 24 h of stress treatment in shoot tissues in rice plants in contrast to untreated samples (Fig. 5c). Reactive oxygen species (ROS) are known to be produced by drought stress in plant cells. Reports suggested that stress-induced ROS may have a signaling role (Hasan et al. 2020). On the contrary, a substantial rise in ROS is seen in cam5-4 and cam6-1 mutants (T-DNA insertion of CAM genes) that do not express GABA shunt genes, indicating that the shunt plays a significant role in pathways involved in the removal of ROS (AL-Quraan 2015). Through the CaM-regulated GABA biosynthesis and the GABA shunt metabolic pathway, CaM might have an indirect role in controlling ROS concentration (Bouche et al. 2003). In this way, plants may be able to maintain ROS (for example H₂O₂) homeostasis and hence provide protection against stress-related damage by the process of feedback regulation of several enzymes through CaM modulation (Vridi et al. 2015).

In the present study, wild-type, genome-edited line #14–1 and #14–6 contains same promoter region, hence they exhibited similar expression levels, whereas the #14–1 had significantly increased GABA content when subjected to abiotic stresses. As mentioned above, OsGAD4 with a truncated CaMBD has greater enzymatic activity than the wild type, this may result in a higher rate of glutamate to GABA conversion and a consequent increase in GABA content in tissues subjected to abiotic stresses.

We endeavoured to investigate the salinity, flooding, and drought stress tolerance in OsGAD4 genome-edited plants. #14–1 line showed significantly higher survival rates (44.2%, 59.6%, and 34.5%, respectively) compared with wild-type controls (17.3%, 25.6% and 12.2%, respectively) but #14–6 was spotted to be like wild type (Fig. 6, Table 2). A trend towards lower biomass reductions in #14–1 plants (Table 2) coincided with the defensive role of GABA-enriched plants to minimize stress damage. Transgenic rice plants overexpressing the OsDREB1A and OsDREB1B genes showed improved tolerance to drought, high-salt, and cold stresses, which likely contributed to their increased survival rate in these conditions (Ito et al. 2006). However, being a key molecule of the defence system, the interaction of GABA with plant growth regulators and other signalling molecules may contribute to potential responses in plant abiotic stress tolerance though the candidates remain inconclusive.

Our study reported the genetic modification of OsGAD4 gene inducing abiotic stress tolerance in rice seedlings at an early vegetative stage. Likewise, transgenic rice plants over-expressing AP37 and AP59 (members of ethylene response factor [ERF] family) confer tolerance to drought and high salinity (Oh et al. 2009). Additionally, plant tolerance to diverse stress conditions through OsCPK4 expression was reported by Campo et al. (2014), suggesting the possibility that this gene encoding a key protein in the stress response pathway is probably capable of providing sufficient tolerance to multiple stresses.

Notably, C-terminal truncated mutant of OsGAD1, OsGAD3 and their hybrid line also showed significant tolerance to abiotic stresses in various extent indicating high endogenous GABA actively regulates the stress resistance mechanism in rice seedling (unpublished data). Taken together, improved stress tolerance in OsGAD4 genome-edited plants may be credited to enhanced expression of GABA-shunt genes and activation of antioxidant enzymes reported as elevated GABA-induced counteracting mechanisms in earlier studies (Kumar et al. 2017; Wang et al. 2017) and it is tempting to establish a relationship with high GABA production with stress tolerance in vegetative tissue of rice plants. To our best knowledge this is the first report to evaluate abiotic stress responses of rice GAD gene modulated by endogenous GABA. This approach of modification of putative GAD by removing CaMBD through genome editing could be applied in other crop plants to improve GABA accumulation and thereby abiotic stress tolerance.

In our study, C-terminal truncation of the coding region of OsGAD4 was performed successfully by genome editing using CRISPR/Cas9. The genome-edited rice line (#14–1) produced almost nine-fold higher GABA content than wild-type Ni. Furthermore, abiotic stress conditions promoted the accumulation of GABA in vegetative tissues. Expression levels of OsGAD4 were upregulated when subjected to abiotic stresses and expression levels were consistent with the GABA accumulation trend in stress conditions. Additionally, it is obvious that the survival rates of truncated version of OsGAD4 genome-edited rice plants were significantly higher than that of control plants, suggesting the stress tolerance induction of OsGAD4. Besides, mutants with small deletion at the C-terminal region of CaMBD (#14–6) resulting in longer polypeptide chain neither produce significant GABA nor considerable stress tolerance response. The mutant's lower GABA concentration may interfere with stress signalling pathways and metabolites, making it more difficult for the plant to withstand stress conditions.

As a powerful tool, genome editing technique allows targeted modifications at specific genomic locations for crop improvement; though transgenic method offers much more choices to consider by integrating genes of interest under strong constitutive or inducible promoters. Precise control over the expression of target genes in a variety of tissues is possible with transgenic overexpression which would be quite difficult in genome editing approach. For instance, GABA-fortified rice was developed through transgenic method yielding almost 60-fold GABA in comparison to wild type (Kowaka et al. 2015) whereas, in genome editing GABA increment remained under tenfold.

However, the potential application of the current study could outweigh the constraints by the production and consumption of GABA-enriched rice for the conscious consumers because of its potential health benefits. Moreover, genome editing can shorten the time needed to create new rice genotypes that are more resilient to abiotic stresses and it has immense prospect to ensure sustainable food security in areas vulnerable to climate-related adversities.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

bp

Base pairs

CaM

Calmodulin

CaMBD

Calmodulin binding domain

CRISPR/Cas

Clustered regularly interspaced palindromic repeats/CRISPR-associated protein

GABA

γ-Aminobutyric acid

GAD

Glutamate decarboxylase

gRNA

Guide RNA

Ni

Nipponbare

PAM

Protospacer adjacent motif

RT-qPCR

Reverse transcription quantitative polymerase chain reaction

TBP2

TATA-binding protein 2

UTR

Untranslated region

Author contribution

KA conceptualized the study and designed the experiments. Experimentation, data collection, and analysis were performed by NA. UK, MM and NY performed additional experiments. The draft of the manuscript was written by NA. All the authors read and approved the manuscript.

Funding

This work was supported by JSPS KAKENHI (Grant No. 23K05166), Cabinet Office, Government of Japan, Moonshot Research and Development Program for Agriculture, Forestry and Fisheries (funding agency: Bio-oriented Technology Research Advancement Institution) grant no. JPJ009237, JST A-STEP TRYOUT (Grant No. JPMJTM20FU), and the Shimane Industrial Promotion Foundation.

Data availability

The vectors used in this study will be available upon request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors have no conflicts of interest to declare.