Journal of Experimental Botany, Vol. 58, No. 10, pp. 2699–2707, 2007 doi:10.1093/jxb/erm120 Advance Access publication 11 June, 2007

RESEARCH PAPER C-terminal extension of rice (OsGAD2) functions as an autoinhibitory domain and overexpression of a truncated mutant results in the accumulation of extremely high levels of GABA in plant cells Downloaded from https://academic.oup.com/jxb/article/58/10/2699/576570 by guest on 02 October 2021

Kazuhito Akama1,* and Fumio Takaiwa2 1 Department of Biological Science, Shimane University, Nishikawatsu 1060, Matsue, Shimane 690-8504, Japan 2 Transgenic Crop Research and Development Center, National Institute of Agrobiological Sciences, Kannondai 2-1-2, Tsukuba, Ibaraki 305-8602, Japan

Received 17 April 2007; Accepted 1 May 2007

Abstract a strong autoinhibitory domain, and that truncation of this domain causes the to act constitutively, Glutamate decarboxylase (GAD) converts L-glutamate with higher activity both in vitro and in vivo. to g-aminobutyric acid (GABA), which is a non-protein amino acid present in all organisms. Plant GADs carry Key words: Agrobacterium, amino acid, , GABA, 2+ a C-terminal extension that binds to Ca /calmodulin GAD, c-aminobutyric acid, glutamate decarboxylase, (CaM) to modulate enzyme activity. However, rice overexpression, rice. possesses two distinct types of GAD, OsGAD1 and OsGAD2. Although they both have a C-terminal extension, the former peptide contains an authentic CaM-binding domain (CaMBD), which is common to Introduction dicotyledonous plants, while the latter does not. There- Glutamate decarboxylase (GAD) is an enzyme that fore, the role of the C-terminal extension in functional catalyses the conversion of L-glutamate to c-aminobutyric expression of OsGAD2 was investigated. An in vitro acid (GABA), which is a non-protein amino acid that is enzyme assay using recombinant OsGAD2 proteins commonly present in both prokaryotes and eukaryotes. In revealed low activity in the presence or absence of 2+ animals, GABA is known to be a major inhibitory Ca /CaM. However, a truncated version of GAD2 . An alteration in the GABA content of (OsGAD2DC) had over 40-fold higher activity than animal cells can result in severe diseases (Kriegstein, wild-type GAD at physiological pH. These two DNA 2005). By contrast, although the role of GABA in plants constructs were introduced simultaneously into rice remains controversial (Bouche´ and Fromm, 2004; Bown calli via Agrobacterium to establish transgenic cell et al., 2006), various environmental stresses, such as lines. Free amino acids were isolated from several oxygen, water, temperature, and mechanical stresses, lines for each construct to determine GABA content. prompt the accumulation of GABA in plant cells (Shelp Calli overexpressing OsGAD2 and OsGAD2DC had et al., 1999). Because these stresses result in a concurrent about 6-fold and 100-fold the GABA content of wild- increase in intracellular Ca2+ concentration (Bush, 1995), type calli, respectively. Regenerated OsGAD2DC rice it has been hypothesized that the Ca2+/calmodulin path- plants had aberrant phenotypes such as dwarfism, way is involved in cross-talk with the GAD activation etiolated leaves, and sterility. These data suggest that pathway. Baum et al. (1993) isolated a nuclear the C-terminal extension of OsGAD2 plays a role as coding for GAD from petunia, in which the CaM-binding

* To whom correspondence should be addressed. E-mail: [email protected]

ª The Author [2007]. Published by Oxford University Press [on behalf of the Society for Experimental Biology]. All rights reserved. For Permissions, please e-mail: [email protected] 2700 Akama and Takaiwa domain (CaMBD) was present at the C-terminal proximal and an extremely high level of GABA accumulated in region. Interestingly, in dicotyledonous plants, without GAD2DC transgenic plant cells, showing that the any exception, the encoding putative GADs C-terminal extension of GAD2 functions as a negative reported so far carry at their C-termini the Ca2+/CaMBD, regulatory domain for enzymatic activity both in vitro and which is lacking in bacterial and animal GADs (Turano in vivo. Overexpression of GAD2DC in plants caused an and Fang, 1998; Yun and Oh, 1998; Zik et al., 1998; aberrant phenotype: transgenic plants had pale green, Yevtushenko et al., 2003; Liu et al., 2004; Oh et al., curled leaves, and were sterile. 2005). In in vitro studies in these reports it has also been found that increases in plant GAD activity are dependent upon Ca2+/CaM. Therefore, Ca2+ signalling and enhance- Materials and methods ment of GAD enzymatic activity would connect environ- mental stress and GABA production. Furthermore, Chen Plant materials and growth conditions

et al. (1994) reported that GABA synthesis is develop- Oryza sativa L. cv. Kitaake was used in this study. For in vitro Downloaded from https://academic.oup.com/jxb/article/58/10/2699/576570 by guest on 02 October 2021 mentally regulated in petunia. Recent studies indicate that tissue culture, rice seeds were immersed in 70% (v/v) ethanol for 30 s, the C-terminal extension of GAD plays an important role rinsed three times with distilled water, surface-sterilized in 50% in normal growth and development of plants (Baum , (v/v) bleach (Kaoh, Japan) for 30 min, then rinsed five times with et al. distilled water. The seeds were transferred onto 2N6 medium to 1996). Furthermore, biochemical analysis has shown that germinate at 28 C under light conditions of 35 lmol m2 s1 2+ Ca /CaM induces dimerization of the C-terminal domain illuminated by white fluorescent tubes and 16/8 h light/dark for of petunia GAD (Yuan and Vogel, 1998; Yap et al., 2 weeks (Hiei et al., 1994). When calli appeared they were 2003). It has also recently been shown that the C-terminal transferred to fresh 2N6 media until infection with Agrobacterium. extension of petunia GAD is required for oligomerization 2+ of the GAD complex and activation via Ca /CaM (Zik Plasmid construction et al., 2006). Therefore, functional expression of plant OsGAD2, a cDNA clone coding for GAD in rice (accession no. GAD is closely regulated via the C-terminal extension, AB056061; Akama et al.,2001)wasusedinthisstudy.First,two 2+ which interacts with Ca /CaM. different recombinant plasmids for overexpressing GAD in E. coli Two nuclear GAD genes have previously been were constructed using pET32a (Novagen) as follows. cDNA identified in the rice genome from cDNA and genomic coding for wild-type OsGAD2 was used as a template for PCR- amplification of the N-terminal coding regions using a specific DNA libraries, each encoding a distinct GAD isoform, primer set (5#-CGTGCGTAGCCATGGTTCTG-3# and 5#-GAT- designated OsGAD1 and OsGAD2 (Akama et al., GTTCACGCACCGGTTCT-3#, NcoI site underlined). The resulting 2001). The coding regions of these two genes have fragment was cleaved using NcoI/XhoI, and the DNA fragment significant homology to other known dicotyledonous obtained (235 bp in length) was subcloned into the NcoI/XhoI site GAD genes, suggesting that both genes are functional in of pET32a to produce pET32a::N-terminal. The C-terminal coding region of GAD2 cDNA was digested using XhoI. The fragment rice. Furthermore, an in vitro CaM-binding assay using was inserted into the XhoI site of the pET32a::N-terminal, yielding recombinant proteins that contain the C-terminal pep- pET32a::GAD2. Finally, clones with sequences in the correct ori- tides of OsGAD1 and OsGAD2, respectively, overex- entation were selected by using restriction enzyme analysis. A GAD pressed in E. coli, and that were purified with an affinity construct with a truncated C-terminal (GAD2DC, lacking 31 amino chromatography, revealed that the former protein binds acid residues at the C terminal) was produced as follows. The C-terminal coding region of GAD2 (amino acid numbers 301–469) CaM, whereas the latter protein does not. Because this is was PCR-amplified using two primer sets specific for GAD2DC the only known example of a plant GAD that does not (5#-AGCTCATCTTCCACATCAAC-3# and 5#-AAAGCATGCC- have a CaM- and it is suggested that the TAGGCGGTCCGGGCGGGG-CC-3#). The PCR fragment was GAD–CaMBD from petunia functions as autoinhibition cleaved with ApaI and cloned into the ApaI/EcoRV site of in the absence of Ca2+/CaM binding (Snedden et al., pBluescript to yield pBluescript containing the C-terminal coding region of GAD2. After the plasmid was confirmed to contain the 1996), it was necessary to determine whether GAD2 is 2+ truncated C-terminal region by sequencing, a KpnI/ApaI fragment actually Ca /CaM-independent with respect to enzyme of pET32a::GAD2, harbouring the N-terminal coding region of activation and whether the C-terminal extension acts as GAD2 and the pET32a vector-coding sequence, was cloned into the an autoinhibitory domain. In order to address these KpnI/ApaI site of the C-terminal coding region of GAD2 in questions, two recombinant , GAD2 and a GAD pBluescript, producing pBluescript containing GAD2DC. This plasmid was then cleaved with NcoI and EcoRI. The open reading with a truncated version of the C-terminal extension frame of GAD2DC was cloned into the NcoI/EcoRI site of pET32a (GAD2DC) were analysed comparatively, with respect to produce pET32a::GAD2DC. For rice transformation, two to their enzymatic activity. Moreover, transgenic lines of different Ti plasmids were constructed. Ti plasmid pCAMBIA1302 rice and tobacco were produced that overexpressed these (CAMBIA, Australia) was first cleaved with NcoI/PlmI. The two E. rice GADs. coli expression vectors produced were cleaved with XhoI, followed by filling-in, and then the vectors were cleaved with NcoI. These It was found that the truncated version of GAD2 had two OsGAD fragments were purified using agarose gel electroph- higher enzyme activity in vitro than its wild-type oresis and were each cloned into the pCAMBIA vector prepared counterpart in the presence or the absence of Ca2+/CaM, as described above. Regulatory domain of rice glutamate decarboxylase 2701 Overexpression of GAD fusion proteins in E. coli was mixed with an equal volume of diethyl ether, and the procedure Two pET plasmids were each introduced into E. coli strain BL21 was repeated twice. The samples were dried by using a centrifuge (DE3) (Novagen) for production of recombinant GAD proteins in concentrator (Tomy, Japan). The resulting samples were suspended bacteria. Induction and purification of fusion proteins were carried in 200 ll of 0.1 N HCl to apply for an amino acid analyser as out in accordance with the method of Akama et al. (2001). Because described above. recombinant proteins GAD2 and GAD2DC have proteolytic traits, these proteins underwent a further purification step using anion- Protein isolation and western blot analysis exchange chromatography (HiTrap Q FF, Amersham) in accordance Total proteins were isolated from calli by using an extraction buffer with the manufacturer’s protocol. The purity of the two recombinant (50 mM NaPO4, pH 7.0, 0.1 mM b-mercaptoethanol, 10 mM proteins was confirmed by SDS-PAGE, and their concentration was EDTA, 0.1% sodium N-laurosyl sarcosinate, 0.1% Triton X-100). determined as described by Bradford (1976). A polyclonal antibody against recombinant OsGAD2 was prepared in Medical and Biological Laboratories (MBL, Japan). SDS–PAGE, Agrobacterium-mediated plant transformation electroblotting and immunodetection were essentially carried out as The constructed Ti plasmids were introduced into Agrobacterium described by Akama et al. (2001). tumefaciens strain GV3101 (Deblaere et al., 1985) via electro- Downloaded from https://academic.oup.com/jxb/article/58/10/2699/576570 by guest on 02 October 2021 poration. Approximately 2-week-old calli cultured in 2N6 medium Microscopic analysis were used for infection with Agrobacterium, essentially using the Longitudinal sections of tobacco stems were prepared by using procedure described by Hiei et al. (1994). The resulting calli were a plant microtome (model MTH-1, MBL, Japan). Samples were selected in medium containing 50 mg l1 hygromycin B for stained with 0.1% toluidine blue O for observation by bright field 4 weeks. Clonal calli were subjected to quantitative analysis of microscopy (Nikon Biophot, Japan). amino acids. Other clonal calli were transferred to N6S3-CH medium (Hiei et al., 1994) to regenerate plants. In parallel, tobacco leaf discs (Nicotiana tabacum cv. Xanthi) were infected with the Results Agrobacterium strains. The transformation procedure was essen- tially as described by De Block et al. (1987). Comparison of the GAD C-terminal extension in rice and petunia In vitro GAD assay Recent sequencing of the rice genome revealed the A rice GAD enzymatic assay was performed by directly measuring GABA production after the GAD reaction as follows: 2 lgof presence of five GAD isoforms, including OsGAD1 and recombinant protein was incubated in 150 ll of reaction mixture OsGAD2 (International Rice Genome Sequencing Project, containing 100 mM of an adequate buffer (bis-tris–HCl for pH 7.0; 2005). All five isoforms are at least transcriptionally pyridine-HCl for pH 5.0, pH 6.0), 1 mM DTT, 5 mM glutamate, active, given the presence of corresponding EST clones. # 0.5 mM pyridoxal 5 -phosphate, 0.5 mM CaCl2, 0.1 lM bovine Figure 1 shows the amino acid sequences of the calmodulin (Sigma), 10% (v/v) glycerol at 30 C. An aliquot of the reaction mixture was mixed with a 1/10 volume of 1 N HCl to stop C-terminal regions of the five rice GADs and the petunia the reaction. The mixture was directly applied to an automated GAD (PhGAD), which is currently the most intensively amino acid analyser (JLC-300, JEOL, Japan), which was based on analysed GAD (Baum et al., 1993). The overall similarity a HPLC separation and quantification with a lithium buffer system among the C-terminal extensions of these plant GADs is after derivatization with ninhydrin of amino acid samples. not high. However, except for GAD2, which lacks CaM- binding ability (Akama et al., 2001), there exists a high RNA extraction and RT-PCR proportion of identity between the rice GADs and the Total RNA was isolated from rice calli by using Sepasol RNAI petunia GAD: several amino acids indicated in the figure Super (Nacalai Tesque, Japan). Single-stranded cDNAs coding for GAD2DC were synthesized from total RNA as the template using are conserved among several GADs. In particular, the Trp reverse transcriptase (SuperScript II, Gibco BRL) and the reverse primer 5#-TCCATCTCGTCCAGGGCCAT-3#. Double-stranded cDNAs were PCR-amplified with the reverse primer described above and the forward primer 5#-AGCTCATCTTCCACATCAAC- 3#. The PCR products (439 bp) were analysed by 2% agarose gel electrophoresis. As a control, rice actin mRNA (accession no. AU312086) underwent the same RT-PCR procedure described above with the primer set 5#-TCCATCTTGGCATCTCTCAG-3# and 5#-GTACCCGCATCAGGCATCTG-3#. Fig. 1. Sequence of the C-terminal portions of GADs from plants. Comparison of the C-terminal amino acid sequences of rice GADs Extraction of free amino acid and quantification of (OsGAD1 to OsGAD5) and petunia GAD (PhGAD). Identical and their content similar amino acids are indicated by black and grey boxes, respectively. Rice calli maintaining in hygromycin-containing 2N6 medium and Trp (W) and the cluster of Lys (K) present in the C-proximal region of transgenic rice and tobacco plants (T generation) cultured in tubes petunia GAD, which are important for in vitro binding to CaM (Arazi o et al., 1995), are indicated by asterisks and a thick line, respectively. containing Murashige and Skoog basal salts (Sigma) for 1 month The positions of two pseudosubstrate residues (E476 and 480) in after regeneration were subjected to determination of GABA PhGAD predicted by Yap et al. (2003) are indicated by hashes. contents. Free amino acids were isolated from these materials by Accession numbers: OsGAD1 (AB56060), OsGAD2 (AB56061), using trichloroacetic acid (TCA) as follows. Tissues were homog- OsGAD3 (AK071556), OsGAD4 (AF377946.3), OsGAD5 enized with 8% (v/v) TCA. After centrifugation, the supernatant (AK070858), PhGAD (L16797). 2702 Akama and Takaiwa (W) residue and the Lys (K) cluster, which contribute to hydrophobic and electrostatic interactions, respectively, are both critical for efficient binding of CaM to the petunia GAD-CaMBD (Arazi et al., 1995). Yap et al. (2003) reported that pseudosubstrate Glu residues (E476 and E480) are present in the N-terminal region of the PhGAD-CaMBD, and that these residues play a role in autoinhibition in the absence of Ca2+/CaM binding, and are frequently observed in other plant GADs. The Glu residues were conserved at corresponding positions in the C-terminal extension of rice GAD1, 3, 4, and 5, implying an autoinhibitory role for the C-terminal regions of these

four GADs. Taken together, these results show that the Downloaded from https://academic.oup.com/jxb/article/58/10/2699/576570 by guest on 02 October 2021 typical features found in PhGAD are conserved in all rice GAD isoforms except GAD2. Although GAD genes have been reported from various plant species, OsGAD2 is the only one found so far that is unable to bind CaM. Fig. 2. Purification of recombinant GADs. GAD constructs (OsGAD2, comprising 500 amino acids, and the truncated version of OsGAD2, OsGAD2DC, in which 31 amino acids were deleted in the C-terminal Effect of truncating the C-terminus of OsGA2 on region) were cloned into the expression vector pET32a for introduction into E. coli strain BL21 (DE3). Bacterial culture and induction of in vitro enzymatic activity expression were carried out as described in the Materials and methods. In order to explore the role of the 31-amino acid Intact recombinant proteins were purified from crude protein extracts using affinity chromatography and then anion-exchange chromatog- C-terminal extension in rice GAD2, two expression raphy. Protein samples were separated using 10% SDS–PAGE, followed vectors were constructed harbouring rice GAD cDNAs: by Coomassie Brilliant Blue staining of the gel. Samples after anion- one corresponded to wild-type OsGAD2 and the other was exchange chromatography for protein fractions are presented. Lanes 1, 2, and 3 represent protein size marker, OsGAD2 and OsGAD2DC, based on wild-type OsGAD2, but lacked the coding region respectively. for the C-terminal extension, OsGAD2DC. These plasmids were introduced into E. coli BL21 (DE3) for over- expression of the recombinant proteins. Each bacterial Accumulation of GABA in calli overexpressing strain was incubated to induce recombinant protein rice GADs expression with IPTG. Figure 2 shows the recombinant In order to determine whether the rice GAD2 enzyme proteins after two rounds of purification by nickel-affinity behaved the same way in plant cells as in vitro, wild-type chromatography and anion-exchange chromatography. and truncated OsGAD2 cDNAs were inserted into the The band sizes of these two recombinant proteins, T-DNA region of the Ti plasmid, under the control of calculated with a molecular size marker, were 78 kDa and the cauliflower mosaic virus (CaMV) 35S and 74 kDa for OsGAD2 and OsGAD2DC, respectively, that the nopaline synthase gene terminator (Fig. 3). These corresponded well to the protein molecular mass eluci- plasmids were then introduced into a strain of Agro- dated from the sequence information and a tag sequence bacterium. Scutellum-derived rice calli were infected with encoding the expression vector, i.e. 76.6 kDa for the these strains of Agrobacterium (with hygromycin selec- former and 73.4 kDa for the latter. The resulting tion) to establish independent transformed cell lines. Free recombinant proteins were assayed for enzymatic activity amino acids were then extracted from the callus cells to in vitro (Table 1). Recombinant GAD2 had relatively low determine the GABA content. As shown in Fig. 4, the activity at pH 7.0, but activity increased as the pH became concentration of GABA present in non-transformed rice lower. As expected, Ca2+/CaM did not stimulate this calli was reproducibly within the range of 200–300 pmol enzyme. Because recombinant OsGAD1 which possesses mg1 FW. When calli overexpressing GAD2 were an authentic CaMBD resulted in a 5-fold increase in its analysed, the GABA concentration varied from 300 pmol enzyme activity in the presence of Ca2+/CaM (data not to 2000 pmol mg1 FW, an up to 6-fold increase in shown), the possibility could be ruled out of the reaction GABA concentration. Some of the cells overexpressing system used in this study that is essentially Ca2+/CaM OsGAD2DC also had an extremely high concentration of independent. The tendencies observed in OsGAD2 were GABA: clone nos 1 and 12 had about a 100-fold increase also done in OsGAD2DC. Comparing the activity be- in GABA content relative to non-transformed cells tween these two enzymes, C-terminal truncated OsGAD2 (Fig. 4). In order to confirm that the GABA accumulation had higher activity than the wild-type enzyme at any pH was due to overexpression of OsGAD2DC, RT-PCR and tested and it was over 40-fold higher than the wild-type at western blot analysis were performed. RT-PCR revealed physiological pH. that seven clones containing OsGAD2DC had a higher Regulatory domain of rice glutamate decarboxylase 2703 Table 1. In vitro enzyme assay for recombinant rice GAD2 and GAD2DC Effects of pH and Ca2+/CaM on activation of rice GAD2 and GAD2DC. The GAD assay was performed to measurement of GABA production as described in the Materials and methods. –Ca2+/CaM: without Ca2+/CaM; +Ca2+/CaM, with 0.5 mM Ca2+ and 0.1 lM bovine calmodulin (Sigma) (final concentration). Data represent the mean 6SD of two and four independent experiments for GAD2 and GAD2DC, respectively. In parentheses are the activity relative to that of GAD2 under the condition of –Ca2+/CaM at each pH. pH Activity (nmol of GABA min1 mg1 protein)

GAD2 GAD2DC

–Ca2+/CaM +Ca2+/CaM –Ca2+/CaM +Ca2+/CaM

7.0 15.060.6 (1.0) 17.369.2 (1.0) 641.4660.3 (42.7) 605.5661.5 (40.3) 6.0 199.0629.7 (1.0) 197.5634.6 (1.0) 2775.06238.6 (13.9) 2719.16685.6 (13.6) 5.0 685.0612.7 (1.0) 710.0614.1 (1.0) 4670.561277.7 (6.8) 4849.76972.3 (7.0) Downloaded from https://academic.oup.com/jxb/article/58/10/2699/576570 by guest on 02 October 2021

Fig. 3. T-DNA constructs used for rice transformation. Two different GAD cDNAs (see the legend for Fig. 2) were cloned into the T-DNA of pCAMBIA1302 (CAMBIA). Ti plasmids carrying these T-DNA constructs were each introduced in Agrobacterium strain GV3101. Rice transformation was carried out as described by Hiei et al. (1994). 35S Pro, cauliflower mosaic virus (CaMV) 35S promoter; Nos T, nopaline synthase gene terminator; HPT, hygromycin phosphotransferase gene; 35S T, CaMV 35S terminator; LB, left border; RB, right border; C-terminal region, coding region for the C-terminal extension corresponding to the CaMBD found in plant GADs.

level of OsGAD2DC mRNA expression than non- transformed calli, and that OsGAD2DC mRNA was more abundant in calli with an extremely high level of GABA (clone nos 1 and 12; Fig. 5A). Protein analysis using an anti-GAD2 antibody showed that the accumulated GABA (as shown in Fig. 4) is directly linked to overexpression of the OsGAD2DC protein (Fig. 5B). Therefore, taken together with the results of the comparative in vitro enzyme assay of these two proteins shown in Table 1, it is speculated that overexpression of C-terminal truncated GAD2 generates a much higher level of GAD activity than overexpression of the wild-type enzyme, leading to accumulation of very high levels of GABA in cells. Crude proteins were extracted from callus cells to measure GAD activity. In the absence of Ca2+/CaM, calli that accumu- lated GABA had activity that was five times as high as wild-type calli (wild-type, GAD2DC-1 and GAD2DC-12 produced 655, 3483, and 3884 pmol of GABA min1 mg1 protein, respectively). Furthermore, as shown in Table 2, a comparison of the content of the free amino acids of wild-type and truncated GAD2 callus cells revealed that not only the GAD substrate Glu, but also many other amino acids are present at lower levels in the cells in which GABA is accumulated to a very high level.

Fig. 4. GABA levels in calli of a wild-type (wt) rice line and transgenic Characteristics of transgenic plants overexpressing lines overexpressing either OsGAD2 or OsGAD2DC. Total free amino OsGADs acids were extracted from callus cells by using the TCA method. The GABA content of each sample was determined with an automated From calli overexpressing the rice GAD2DC gene, trans- amino acid analyser. genic shoots were regenerated. Amino acid content was 2704 Akama and Takaiwa Discussion Functional analysis of a glutamate decarboxylase (GAD2) from rice, a monocotyledon, was carried out in this study. Many kinds of calmodulin-binding proteins (CaMBPs) have been found in plants (Reddy et al., 2002; Bouche´ et al., 2005), indicating that a major regulatory path- way for plant enzymes involves Ca2+/CaM (Yang and Poovaiah, 2003; McCormack et al., 2005). It is possible that plants have more varied CaMBPs than other eukaryotes, and GAD is one of the most abundant CaMBPs in plants (Ling et al., 1994). Arabidopsis and rice are commonly Fig. 5. Analyses of rice calli overexpressing OsGAD2DC. (A) RT-PCR used as model plants, and the genomes of both have been analysis. Total RNAs were isolated by using Sepasol RNAI Super Downloaded from https://academic.oup.com/jxb/article/58/10/2699/576570 by guest on 02 October 2021 (Nacalai Tesque). Rice actin mRNA was used as a control. wt, wild- sequenced, with the result that both plants have been type calli. Seven independent rice cell lines (nos 1 to 12) overexpressed found to possess five GAD genes each. Arabidopsis GADs OsGAD2DC. GAD2 cDNA, OsGAD2 cDNA cloned into the pBlue- script vector with different concentrations as a PCR standard. (B) all contain the typical CaMBD; however, the GAD Western blot of the crude proteins extracted from callus cells. Rabbit C-terminal extension (e.g. the basic amphiphilic a-helix anti-GAD2 antibody and a goat anti-rabbit IgG antibody alkaline and conserved Trp and Lys-cluster), which is ubiquitously phosphatase conjugate were used as primary and secondary antibodies, respectively. Samples no. 1 to 12 correspond to those in (A). found in dicotyledonous plants, is not found in rice GAD2 (Fig. 1). Extensive database searches against all plant DNA sequences so far reported did not identify any Table 2. Amount of free amino acids in wild-type calli (WT) putative GAD2 orthologue. Therefore, it is tempting to and calli overexpressing OsGAD2DC speculate that OsGAD2 is involved in a unique regula- In parentheses are the relative percentages of each amino acid. tion pathway, which most probably occurs via a novel 2+ a a Ca /CaM-independent route. Amino acid WT OsGAD2DC (nmol mg1 FW) (nmol mg1 FW) In order to determine the role of the C-terminal extension of OsGAD2, a truncated version was con- Asp 3.2460.07 (12.3) 1.1560.12 (5.2) Thr 2.0960.49 (7.9) 0.2460.03 (1.0) structed to allow a comparison of its activity with that of Ser 0.3560.08 (1.3) 0.4060.08 (1.8) the non-truncated version. It was found that the recom- Asn 2.1060.24 (8.0) 0.6560.01 (2.9) binant OsGAD2 enzyme was not activated in the presence Glu 2.7160.37 (10.3) 0.9460.34 (4.2) of Ca2+/CaM (Table 1). This is the first evidence of Gln 1.4760.19 (5.6) 0.5060.02 (2.2) 2+ Gly 1.7960.01 (6.8) 1.8560.16 (8.4) aCa /CaM-independent GAD on its activation in plants. Ala 1.0860.28 (4.1) 0.0960.01 (0.4) This phenomenon might occur because the C-proximal GABA 0.1060.03 (0.3) 10.2761.31 (46.7) Allb 26.0262.15 (100.0) 21.9861.14 (100.0) region of OsGAD2 functions as a negative regulatory domain, and because the C-terminus of this enzyme is not a Values indicate the mean 6SD of two independent extraction. a target for Ca2+/CaM, no conformational change of this b All indicates value of total amino acids. domain would take place in the presence of Ca2+/CaM to stimulate enzyme activity. Under the different pH con- determined for samples from three different tissues ditions tested, as pH became lower, enzyme activity (leaves, roots, stems) of regenerated rice plants. Compared tended to increase, indicating that GAD activity is high with samples from wild-type plants, GABA levels were under acidic conditions, and that GABA production is increased in every tissue for transgenic rice and in stems stimulated by reducing cytosolic pH (Crawford et al., and roots for tobacco (Table 3). Interestingly, plants 1994; Snedden et al., 1995). Truncation of the C-terminal overexpressing OsGAD2DC had a dwarf phenotype, with extension of OsGAD2 conferred a 7–40-fold increase in pale, curled leaves, and were infertile, whereas plants enzyme activity relative to wild-type in the conditions at overexpressing OsGAD2 had a normal phenotype pH 5 to pH 7 and in the presence or absence of Ca2+/CaM (Fig. 6A). The same Agrobacterium strains were used for (Table 1). As far as is known, this is the first instance in tobacco transformation. The same tendencies (higher plants where removal of a limited C-terminal region has GABA content, dwarfism, infertility) were observed in been found to contribute directly to drastic enhancement the OsGAD2DC-overexpressing tobacco plants (Fig. 6B). of enzyme activity. The CaMBD in CaMBP is known to Histological analysis of tobacco plant tissues showed that overlap with an autoinhibitory domain (Hoeflich and cells in the stem cortex were distinct in their size among Ikura, 2002). The extremely high levels of accumulated wild-type and transformed plants: wild-type parenchyma GABA observed in the present study led to the hypothesis cells were large and elongated, but they were smaller and that the C-terminal extension of OsGAD2 functions as shorter in GAD2DC plants (Fig. 6C, D). a strong autoinhibitory domain, thus artificially truncating Regulatory domain of rice glutamate decarboxylase 2705 Table 3. Comparison of GABA content in wild-type and transgenic plants Three different tissues from wild-type and transgenic (overexpressing OsGAD2 or OsGAD2DC) rice and tobacco plants were analysed for GABA content. Free amino acids were extracted by using the TCA method. Data represent mean 6SD of two separate isolations of tissues from the same plant.

Tissue Rice Tobacco

Wild-type GAD2DC-12 Wild-type GAD2 GAD2DC (pmol mg1 FW) (pmol mg1 FW) (pmol mg1 FW) (pmol mg1 FW) (pmol mg1 FW)

Leaf 675657 29 20668389 37560 410690 381616 Stem 150611 28 09065724 431647 8346108 14 07263790 Root 5496165 467261885 18563 341660 15 7766316 Downloaded from https://academic.oup.com/jxb/article/58/10/2699/576570 by guest on 02 October 2021

Fig. 6. Morphology of transgenic plants (A, rice; B, tobacco). Note that both plants over-expressing GAD2DC have not undergone normal development. (C, D) Longitudinal radial sections of the stem cortex of wild-type (wt) and GAD2DC-overexpressing tobacco plants, respectively. Bar¼100 lm. it would result in abolition of this autoinhibition. hygromycin to establish clonal cell lines. GABA content Alternative approaches to activating enzyme(s) via a was determined for eight or four clones per construct CaMBD, such as monoclonal antibodies directed against (Fig. 4). Calli overexpressing GAD2 had a 6-fold greater the CaMBD and controlled proteolysis have been de- GABA concentration compared with wild-type calli. scribed (Rasi-Caldogno et al., 1993; Snedden et al., Some of the GAD2DC-overexpressing clones produced 1996). Zik et al. (2006) used petunia GADDC9, in which had an extremely high concentration of GABA (a ;100- nine C-terminal amino acid residues were removed fold increase relative to wild-type). To date, other than (a similar construct to the one used in the present study) ourselves, three different research groups have produced and found that in vitro enzymatic activity was lower than transgenic plants (tobacco and rice) and bacteria into that of wild-type GAD, and that Ca2+/CaM-induced which a plant GAD gene lacking a portion of the CaMBD activation did not occur. GADDC9 bound to Ca2+/CaM, has been introduced (Baum et al., 1996; McLean et al., but could not form a higher structure (;680 kDa), unlike 2003; Park et al., 2005). In all cases, transgenes were wild-type GAD, thus causing lower activity and no expressed at high levels, and transgenic organisms induction of activity. contained about 10–50 times higher concentrations of In the present study, the two constructs used for the in GABA than controls. These values are higher than that vitro assay were both introduced in the Ti plasmid vector found for OsGAD2 in the present study. When via Agrobacterium (Fig. 3). After infection with Agro- OsGAD2DC was overexpressed in rice cells in the present bacterium, calli were incubated on media containing study, a GABA concentration of 17–22 nmol mg1 FW 2706 Akama and Takaiwa calli was measured (Fig. 4), corresponding to 34–46% of content in the plant tissues was higher than that of total free amino acids, almost equivalent to the values controls, indicating an imbalance of amino acids in cells, reported by Baum et al. (1996). In transgenic plants the i.e. an increase in GABA and a decrease in many other content of the free amino acids was reduced, meaning that amino acids would be involved in these phenotypes in rice the increase in GABA concentration influenced the as well as tobacco. However, in leaves of tobacco there proportion of the other amino acids examined. It is noted was reproducibly little difference of GABA content that the substrate of GABA, Glu, was present at lower among wild-type and transgenic lines, although there is levels than in wild-type plants (Table 2). Due to the close no explanation for this. Interestingly, a very similar connections between the metabolic pathways of Glu and phenotype was observed by Baum et al. (1996), irrespec- most of the other amino acids, a decrease in Glu directly tive of the kind of GAD tested, suggesting that an might influence the concentrations of many of the other extremely high level of GABA and a decrease in Glu at amino acids. least has similar effects on development and morphogen-

Data presented here show that the mutant GAD2 gene esis in plants. Recently, functional genomic studies in Downloaded from https://academic.oup.com/jxb/article/58/10/2699/576570 by guest on 02 October 2021 contributes to the extremely high level of GABA Arabidopsis have revealed the importance of close accumulated in cells. First, the mutant GAD2 enzyme had regulation of the GABA shunt in plant development and a much higher level of Ca2+/CaM-independent activity stress responses (Palanivelu et al., 2003; Bouche´ et al., than the wild-type enzyme in vitro (Table. 1). Second, 2003, 2004). RT-PCR analysis indicated overexpression of the trans- Bown et al. (2006) proposed that plants produce GABA gene at the RNA level (Fig. 5A). Third, western blot for defence against invertebrate pests. The transgenic rice analysis using an anti-GAD2 antibody indicated a positive plants generated in this study that show constitutive correlation between GAD2DC protein concentration and expression of GABA may thus acquire enhanced re- GABA accumulation (Fig. 5B). Fourth, crude protein sistance against various natural pests. Because it was extracts isolated from wild-type and GAD2DC calli possible to establish transgenic rice plants that accumu- showed that, in the absence of Ca2+/CaM, GAD in lated extremely high levels of intracellular GABA, by GAD2DC cells had a higher level of activity than in wild- controlling this expression using tissue- or stage-specific type cells. Taken together, these data show without promoters or stress-inducible promoters, many different question that overexpression of GAD2DC increases GAD kinds of dominant-negative rice mutants could potentially enzyme activity and leads to extremely high levels of be produced, which would contribute to exploring GABA GABA in plant cells. By contrast, transgenic tobacco function. Also, because GABA plays a role in reduction plants overexpressing truncated petunia GAD of blood pressure (Takahashi et al., 1955), a molecular (PhGADDC), which accumulated extremely high levels of breeding approach involving the accumulation of GABA GABA, had a GAD enzyme activity in crude protein in the rice grain might lead to development of a ‘functional extract that was Ca2+/CaM-independent, but at a lower food’ strain of rice in the near future. level than that in wild-type plants (Baum et al., 1996), unlike the GAD2DC discussed in the present study. Baum et al. concluded that abnormal regulation of GAD activity Acknowledgements in the PhGADDC-overexpressing plant was the main This research was supported by grants for functional analysis of reason for the extremely high GABA level. genes relevant to agriculturally important traits in the rice genome In a recent structural study it was found that E. coli from the Ministry of Agriculture, Forestry and Fisheries of Japan GAD that does not contain the CaMBD forms a 330 kDa (Green Technology Project IP-2004). homohexamer in acidic conditions, which is important for enzyme activation (Capitani et al., 2003). Inatomi and Slaughter (1975) reported tissue-specific GAD complexes References occurring in barley. Because the structure that the trun- Akama K, Akihiro T, Kitagawa M, Takaiwa F. 2001. Rice cated GAD2 forms in the present study was not inves- (Oryza sativa) contains a novel isoform of glutamate decarboxy- tigated, the possibility cannot be excluded that GAD2DC lase that lacks an authentic calmodulin-binding domain at the forms a complex with endogenous GAD subunits, which C-terminus. Biochimica et Biophysica Acta 1522, 143–150. would contribute to the activity of the complex in cells. Arazi T, Baum G, Snedden WA, Shelp BJ, Fromm H. 1995. Overexpression of OsGAD2DC conferred aberrant pheno- Molecular and biochemical analysis of calmodulin interactions types in both rice (line 12) and tobacco plants (Fig. 6). with the calmodulin-binding domain of plant glutamate decarboxy- Rice plants had small, curled, etiolated leaves, indicating lase. Plant Physiology 108, 551–561. Baum G, Chen Y, Arazi T, Takatsuji H, Fromm H. 1993. A some interference with the normal development of plant glutamate decarboxylase containing a calmodulin binding plastids. Transgenic tobacco plants with OsGAD2DC had domain. Journal of Biological Chemistry 268, 19610–19617. a typical dwarf phenotype, with a reduction in the size of Baum G, Lev-Yadun S, Fridmann Y, Arazi T, Katsnelson H, parenchyma cells in the stem cortex. In both cases, GABA Zik M, Fromm H. 1996. Calmodulin binding to glutamate Regulatory domain of rice glutamate decarboxylase 2707 decarboxylase is required for regulation of glutamate and GABA MacLean MD, Yevtushenko DP, Deschene A, Van and normal development in plants. EMBO Journal Cauwenberghe OR, Makhmoudova A, Potter JW, Bown AW, 15, 2988–2996. Shelp BJ. 2003. Overexpression of glutamate decarboxylase in Bouche´ N, Fait A, Bouchez D, Møller SG, Fromm H. 2003. transgenic tobacco plants confers resistance to the northern root- Mitochondrial succinic-semialdehyde dehydrogenase of the c- knot nematode. Molecular Breeding 11, 277–285. aminobutyrate shunt is required to restrict levels of reactive McCormack E, Tsai Y-C, Braam J. 2005. Handling calcium oxygen intermediates in plants. Proceedings of the National signaling: Arabidopsis CaM and CMLs. Trends in Plant Science Academy of Sciences, USA 100, 6843–6848. 10, 383–389. Bouche´ N, Fromm H. 2004. GABA in plants: just a metabolite? Oh SH, Choi WG, Lee IT, Yun SJ. 2005. Cloning and charac- Trends in Plant Science 9, 110–115. terization of a rice cDNA encoding glutamate decarboxylase. Bouche´ N, Fait A, Moriyah Zik, Fromm H. 2004. The root- Journal of Biochemistry and Molecular Biology 38, 595–601. specific glutamate decarboxylase (GAD1) is essential for sustain- Palanivelu R, Brass L, Edlund AF, Preuss D. 2003. Pollen tube ing GABA levels in Arabidopsis. Plant Molecular Biology 55, growth and guidance is regulated by POP2,anArabidopsis gene 315–325. that controls GABA levels. Cell 114, 47–59. Bouche´ N, Yellin A, Snedden WA, Fromm H. 2005. Plant- Park KB, Ji GE, Park MS, Oh SH. 2005. Expression of rice specific calmodulin-binding proteins. Annual Review of Plant glutamate decarboxylase in Bifidobacterium longum enhances Downloaded from https://academic.oup.com/jxb/article/58/10/2699/576570 by guest on 02 October 2021 Biology 56, 435–466. c-aminobutyric acid production. Biotechnology Letters 27, Bown AW, MacGregor KB, Shelp BJ. 2006. Gamma-amino- 1681–1684. butyrate: defense against invertebrate pests? Trends in Plant Rasi-Caldogno F, Carnelli A, De Michelis MI. 1993. Controlled Science 11, 424–427. proteolysis activates the plasma membrane Ca2+ pump of higher Bradford MM. 1976. A rapid and sensitive method for the quan- plants. Plant Physiology 103, 385–390. titation of microgram quantities of protein utilizing the principle of Reddy VS, Ali GS, Reddy ASN. 2002. Genes encoding calmod- protein–dye binding. Analytical Biochemistry 72, 248–254. ulin-binding proteins in the Arabodopsis genome. Journal of Bush DS. 1995. Calcium regulation in plant cells and its role in Biological Chemistry 277, 9840–9852. signaling. Annual Review of Plant Physiology and Plant Shelp BJ, Bown AW, McLean MD. 1999. Metabolism and Molecular Biology 46, 95–122. functions of gamma-aminobutyric acid. Trends in Plant Science Capitani G, De Biase D, Aurizi C, Gut H, Bossa F, Gru¨tter MG. 4, 446–452. 2003. Crystal structure and functional analysis of Snedden WA, Arazi T, Fromm H, Shelp BJ. 1995. Calcium/ glutamate decarboxylase. EMBO Journal 22, 4027–4037. calmodulin activation of soybean glutamate decarboxylase. Plant Chen Y, Baum G, Fromm H. 1994. The 58-kilodalton calmodulin- Physiology 108, 543–549. binding glutamate decarboxylase is a ubiquitous protein in Snedden WA, Koutsia N, Baum G, Fromm H. 1996. Activation petunia organs and its expression is developmentally regulated. of a recombinant petunia glutamate decarboxylase by calcium/ Plant Physiology 106, 1381–1387. calmodulin or by a monoclonal antibody which recognizes the Crawford LA, Bown AW, Breitkreuz KE, Guinel FC. 1994. The calmodulin binding domain. Journal of Biological Chemistry 271, synthesis of c-aminobutyric acid in response to treatments 4148–4153. reducing cytosolic pH. Plant Physiology 104, 865–871. Takahashi H, Tiba M, Iino M, Takayasu T. 1955. The effect of Deblaere R, Bytebier B, De Greve H, Deboeck F, Schell J, Van gamma-aminobutyric acid on blood pressure. Japanese Journal of Montagu M, Leemans J. 1985. Efficient octopine Ti plasmid- Physiology 5, 334–341. derived vectors for Agrobacterium-mediated gene transfer to Turano FJ, Fang TK. 1998. Characterization of two glutamate plants. Nucleic Acids Research 13, 4777–4788. decarboxylase cDNA clones from Arabidopsis. Plant Physiology De Block M, Botterman J, Vandewiele M, Dockx J, Thoen C, 117, 1411–1421. Gossele´ V, Rao Movva N, Thompson C, Van Montagu M, Yang T, Poovaiah BW. 2003. Calcium/calmodulin-mediated signal Leemans J. 1987. Engineering herbicide resistance in plants by network in plants. Trends in Plant Science 8, 505–512. expression of a detoxifying enzyme. EMBO Journal 6, 2513–2518. Yap KL, Yuan T, Mal TK, Vogel HJ, Ikura M. 2003. Structural Kriegstein AR. 2005. GABA puts the brake on stem cells. Nature basis for simultaneous binding of two carboxy-terminal peptides Neuroscience 8, 1132–1133. of plant glutamate decarboxylase to calmodulin. Journal of Hiei Y, Ohta S, Komari T, Kumashiro T. 1994. Efficient Molecular Biology 328, 193–204. transformation of rice (Oryza sativa L.) mediated by Agro- Yevtushenko DP, McLean MD, Peiris S, Van Cauwenberghe OR, bacterium and sequence analysis of the boundaries of the Shelp BJ. 2003. Calcium/calmodulin activation of two divergent T-DNA. The Plant Journal 6, 271–282. glutamate decarboxylases from tobacco. Journal of Experimental Hoeflich KP, Ikura M. 2002. Calmodulin in action: diversity Botany 54, 2001–2002. in target recognition and activation mechanisms. Cell 108, Yuan T, Vogel HJ. 1998. Calcium-calmodulin-induced dimerization 739–742. of the carboxyl-terminal domain from petunia glutamate decarboxy- Inatomi K, Slaughter JC. 1975. Glutamate decarboxylase from lase. Journal of Biological Chemistry 273, 30328–30335. barley embryos and roots. Biochemical Journal 147, 479–484. Yun SJ, Oh SH. 1998. Cloning and characterization of a tobacco International Rice Genome Sequencing Project. 2005. The map- cDNA encoding calcium/calmodulin-dependent glutamate decar- based sequence of the rice genome. Nature 436, 793–800. boxylase. Molecules and Cells 8, 125–129. Ling V, Snedden WA, Shelp BJ, Assmann SM. 1994. Analysis of Zik M, Arazi T, Snedden WA, Fromm H. 1998. Two isoforms of a soluble calmodulin binding protein from fava bean roots: identi- glutamate decarboxylase in Arabidopsis are regulated by calcium/ fication of glutamate decarboxylase as a calmodulin-activated calmodulin and differ in organ distribution. Plant Molecular enzyme. The Plant Cell 6, 1135–1143. Biology 37, 967–975. Liu LL, Zhao L, Li Q, Jiang L, Wang CM, Zhang WW, Liu SJ, Zik M, Fridmann-Sirkis Y, Fromm H. 2006. C-terminal residues Chen LM, Zhai HQ, Wan JM. 2004. Molecular cloning and of plant glutamate decarboxylase are required for oligomerization expression of a novel glutamate decarboxylase gene in rice. Rice of a high-molecular weight complex and for activation by calcium/ Genetics Newsletter 21, 39–41. calmodulin. Biochimica et Biophysica Acta 1764, 872–876.