Metabolic Alterations in Organic Acids and γ-Aminobutyric Acid in Developing Tomato (Solanum lycopersicum L.) Fruits

Abstract

Salt stress improves the quality of tomato fruits. To clarify the mechanism(s) underlying this phenomenon, we investigated metabolic alterations in tomato fruits exposed to 160 mM salt, focusing on metabolism of organic acids related to the tricarboxylic acid (TCA) cycle and γ-aminobutyric acid (GABA). Quantitative analyses revealed that most amino acids increased in response to salt stress throughout fruit development, and the effect of the stress was greater in the pericarp than in the columella, whereas organic acids did not show a remarkable tendency to salt stress. The transcript levels of 20 genes encoding enzymes of the TCA cycle and peripheral pathways were also analyzed in salt-stressed fruit. Genes responsive to salt stress could be categorized into two types, which were expressed during early development or ripening stages. During fruit development, phosphoenolpyruvate carboxylase 2 and phosphoenolpyruvate carboxykinase displayed contrasting expression patterns between early development and ripening, suggesting a switch of carbohydrate metabolism after the turning stage. Our results revealed a new metabolic pathway for GABA during the development of tomato fruits. At the start of ripening, GABA is first converted to malate via succinate semialdehyde, and it passes into a shunt through pyruvate. Then, it flows back to the TCA cycle and is stored as citrate, which contributes as a substrate for respiration during fruit maturation.

Introduction

The tomato (Solanum lycopersicum L.; Solanaceae) is an excellent model plant for genomic research of solanaceous plants. Recently >80% of the tomato genome sequence was made available to the public, and numerous genomic and genetic resources including numerous mapped traits, developed DNA markers, mutants and expressed sequence tags (ESTs) have become available (reviewed by Matsukura et al. 2008). It is also an excellent berry fruit model that allows the analysis of fruit development, ripening and assimilate metabolism. This horticultural crop also provides good material for studies investigating the effects of salinity on assimilate metabolism in fruit. While excessive salt exposure reduces fruit size, total yield and photosynthesis, and increases blossom end rot (Malundo et al. 1995, Pérez-Alfocea et al. 1996, Gao et al. 1998, Bolarin et al. 2001, Krauss et al. 2006, Saito et al. 2006), moderate salt stress generally improves fruit quality by increasing carotenoids and total soluble solids (sugars, organic acids and amino acids) (Tal et al. 1979, Ho et al. 1987, Adams 1991, Balibrea et al. 1996, Gao et al. 1998, Balibrea et al. 1999, Balibrea et al. 2000, De Pascale et al. 2001, Krauss et al. 2006, Saito et al. 2008a), which are important components of taste in tomatoes. Until a decade ago, these phenomena were explained mainly as a soluble solid ‘concentration effect’ caused by the suppression of fruit growth that results from water stress (Ehret and Ho 1986, Ho et al. 1987, Sakamoto et al. 1999).

However, during the past decade, studies have reported modified carbohydrate allocation and the activation of assimilate metabolic enzymes in tomato fruits exposed to salt (Balibrea et al. 1996, Gao et al. 1998, Saito et al. 2009). Our recent study revealed that activated starch biosynthesis in salt-stressed fruit increases the accumulation of soluble sugars in ripe red fruits and that ADP-glucose pyrophosphorylase is involved in this event at the transcriptional level (Yin et al. 2010). Salt stress also promotes the accumulation of γ-aminobutyric acid (GABA), glutamate and glutamine, and it increases the activity of glutamate synthase and glutamine synthase (GS) (Francisco et al. 1995, Debouba et al. 2006, Saito et al. 2008b). These results suggest that the higher sugar and amino acid levels in salt-stressed tomato fruits are caused by not only a ‘concentration effect’ but also a shift in the metabolism of carbon and nitrogen components. However, because only limited information is available regarding the metabolic changes that occur under salt stress at the molecular level, the mechanisms by which stress enhances the accumulation of assimilates in the fruit are not fully understood. In particular, unlike the metabolism of sugars and amino acids, little information exists on the dynamics of tricarboxylic acid cycle (TCA)-related metabolism. The TCA cycle connects glycolysis to amino acid biosynthesis, and it plays important roles in the regulation of respiration and energy generation by producing ATP and NADH. In fruit, the TCA cycle is involved in organic acid biosynthesis. Although GABA is one of the most abundant nitrogen components in developing tomato fruits, its physiological function remains unclear. GABA reaches its maximum level at the mature-green stage, when it accounts for nearly 50% of the total amino acid content, and then it rapidly declines during tomato fruit ripening (Rolin et al. 2000, Akihiro et al. 2008). However, there is almost no information regarding the fate of GABA that decreases after ripening begins.

In this work, to address the above questions, we focused on the metabolic alterations of the TCA cycle and peripheral pathways in developing tomato fruits exposed to salt stress. We carried out quantitative analyses of the major TCA cycle-related amino acids, organic acids and sugars. We also assessed the transcript levels of genes encoding relevant enzymes. The results suggested the existence of a new metabolic pathway for the supply of carbon/nitrogen (C/N) starting from GABA during fruit development. Moreover, our data suggest a physiological function for GABA in tomato fruits as a substrate for respiration during climacteric ripening.

Results

Growth and fruit development of salt-stressed tomato

Tomato plants, cv. Micro-Tom, grown under control conditions (0 mM NaCl) and with 160 mM NaCl (Fig. 1A) were used to detect changes in assimilate accumulation in salt-stressed fruit. In this study, we divided fruit development into four stages: immature green (IMG), mature green (MG), yellow (Yel) and red (Red) (Fig. 1B). We further separated the fruit into pericarp and columella for quantitative metabolite analyses by gas chromatography–mass spectrometry (GC-MS; Fig. 1C), and we used whole fruit for transcriptional analyses.

Fig. 1

(A) Cv. ‘Micro-Tom’ tomato plants grown under control (0 mM NaCl) and saline conditions (160 mM NaCl). (B) Developing ‘Micro-Tom’ fruits grown under control conditions. IMG, immature green stage; MG, mature green stage; Yel, yellow stage; and Red, red stage. (C) Fruits were divided into pericarp and columella for quantitative metabolite analyses.

Changes in the levels of organic acids and amino acids under salt stress

The levels of organic acids involved in the TCA cycle were determined in both pericarp and columella tissue using GC-MS (Table 1). Under both control and salt stress conditions, pyruvate levels gradually decreased during fruit maturation. Salt stress suppressed pyruvate accumulation to 73 and 44% of the control in columella tissue at the IMG and MG stages, respectively. In contrast, salt increased pyruvate levels 1.79-fold in pericarp tissue at the Yel stage (Table 3). Under the control conditions, while citrate levels remained relatively steady in pericarp, they prominently increased in the Yel stage and then declined in the Red stage in columella tissue. The responsiveness of citrate accumulation to salt stress varied in different tissue types (Table 1). Salt stress enhanced citrate accumulation in columella tissue (1.66-fold) only at the MG stage and then suppressed it during ripening, whereas it promoted citrate accumulation from the MG and Red stages by 1.21- to 1.61-fold in pericarp tissue (Table 3). Succinate was maximal at the IMG stage and then markedly decreased during the later developmental stages. Salt stress had contrasting effects on succinate accumulation in columella and pericarp tissues. It was enhanced 1.09- to 1.85-fold in pericarp in most of the developmental stages, whereas it was suppressed in columella except at the Yel stage, with a maximal decrease of 48% at the MG stage (Table 3). Fumarate was also maximal at the IMG stage, and it then gradually decreased as the fruit developed in both tissue types. Fumarate accumulation was enhanced in most fruit stages in both tissues, except in the early stages in columella (Table 1). Malate levels peaked at the MG stage in columella and at the Yel stage in pericarp under control conditions. It dramatically decreased during fruit ripening and was lowest at the Red stage. In spite of relatively high variability, salt stress tended to enhance malate accumulation only in the IMG stage. However, malate levels were considerably suppressed after the MG stage in both tissue types (Tables 1, 3). The 2-oxoglutarate levels remained almost unchanged throughout all developmental stages and under both conditions in pericarp tissue, but in columella tissue, the 2-oxoglutarate level decreased during fruit development under control conditions. Under salt stress, 2-oxoglutarate levels were suppressed at the early stages but enhanced at the Yel and Red stages (Table 1). Taken together, these results indicate that salt tended to suppress the accumulation of organic acids (except citrate) in columella tissue during early stages, whereas organic acid accumulation (except that of malate) was enhanced in pericarp tissue during the ripening stages (Table 1). However, the responsiveness of the organic acids to salinity was lower than that of the soluble sugars. Although pyruvate, succinate, fumarate and 2-oxoglutarate accumulation were all affected by salt stress, the actual levels of these components were very low in fruit compared with citrate and malate (Tables 1, 3).

Table 1

Organic acid contents (nmol g FW⁻¹ or μmol g FW⁻¹) in developing tomato fruits grown under control (0 mM NaCl) and salt stress (160 mM NaCl) conditions

Organic acid	Tissue	Treatment	IMG	MG	Yel	Red
Pyruvate (nmol g FW−1)	Pericarp	Control	13.3 ± 1.1	14.4 ± 0.6	10.0 ± 1.7	8.3 ± 1.0
		+NaCl	15.8 ± 4.5	13.0 ± 1.0	17.9 ± 0.9**	7.2 ± 0.7
	Columella	Control	12.0 ± 2.4	20.6 ± 1.8	10.6 ± 1.9	6.0 ± 1.6
		+NaCl	8.8 ± 1*	9.1 ± 0.7**	10.7 ± 2.1	6.7 ± 0.7
Citrate (μmol g FW−1)	Pericarp	Control	43.5 ± 11.7	30.9 ± 6.1	48.6 ± 3.3	47.8 ± 4.4
		+NaCl	32.2 ± 2.2	41.3 ± 3.4*	78.3 ± 8.9**	58.0 ± 5.4*
	Columella	Control	28.4 ± 2.9	38.5 ± 3.8	121.3 ± 16.4	70.9 ± 5.9
		+NaCl	27.3 ± 1.2	63.8 ± 7.1**	62.1 ± 4.1**	64.2 ± 5.0
Succinate (nmol g FW−1)	Pericarp	Control	254.1 ± 33.3	31.8 ± 3.1	26.5 ± 3.1	21.5 ± 3.6
		+NaCl	386.6 ± 253.1	34.5 ± 5.2	40.2 ± 2.3**	39.9 ± 27.8
	Columella	Control	155.7 ± 30.0	48.4 ± 3.6	16.4 ± 2.8	13.9 ± 1.3
		+NaCl	148.8 ± 24.3	25.3 ± 2.2**	24.6 ± 13.5	9.2 ± 1.2**
Fumarate (nmol g FW−1)	Pericarp	Control	208.1 ± 9.2	183.1 ± 9.8	171.3 ± 4.2	164.2 ± 4.3
		+NaCl	293.2 ± 83.1	204.2 ± 10*	219.0 ± 8.9**	226.9 ± 50.2*
	Columella	Control	188.9 ± 11.1	183.2 ± 6.7	123.2 ± 3.4	112.9 ± 20.4
		+NaCl	162.9 ± 12.8*	147.2 ± 10**	172.6 ± 42.6	139.8 ± 19.0
Malate (μmol g FW−1)	Pericarp	Control	109.6 ± 87.0	234.7 ± 146.8	270.4 ± 151.6	7.7 ± 8.9
		+NaCl	388.5 ± 377.9	152.8 ± 97.2	78.3 ± 3.7*	2.8 ± 0.1
	Columella	Control	23.6 ± 10.5	475.4 ± 292.4	212.5 ± 173.6	3.9 ± 1.3
		+NaCl	32.2 ± 5.1	341.0 ± 336.3	50.3 ± 45.4	2.2 ± 0.1*
2-Oxoglutarate (nmol g FW−1)	Pericarp	Control	1025.7 ± 7.3	987.4 ± 4.5	1002.9 ± 6.5	982.1 ± 8.8
		+NaCl	1011.4 ± 27.0	977.5 ± 2.2**	1032.9 ± 23*	979.7 ± 6.5
	Columella	Control	1112.1 ± 30.4	1003.7 ± 3.8	742.9 ± 20.5	686.5 ± 154.9
		+NaCl	963.7 ± 50.6**	782.5 ± 92**	1034.2 ± 196.9*	911.4 ± 111.8

The asterisks indicate statistical significance of means in the same tissue and same developmental stage estimated using Fisher’s PLSD test (*P < 0.05; **P < 0.01).

IMG, immature green; MG, mature, green; Yel, yellow; and Red, red.

The amino acid levels in pericarp and columella tissues in developing fruit grown under control conditions and salt stress were determined (Table 2). During the early stages, the dominant amino acids were glutamine and GABA. While glutamine levels declined gradually during ripening, GABA declined dramatically. In contrast, aspartate clearly increased during fruit ripening and became the dominant amino acid at the Red stage. Although glutamate levels tended to increase during fruit development, this increase was not pronounced compared with that of aspartate. Asparagine levels remained relatively unchanged in pericarp tissue throughout all developmental stages, whereas they declined during ripening in columella tissue. Aspartate substantially increased 40.6- to 42.2-fold from the IMG to the Red stage, which was the greatest accumulation of any of the quantified amino acids. In both pericarp and columella tissues and during most stages, salt stress enhanced the accumulation of glutamine by 1.1- to 1.8-fold, glutamate by 1.36- to 2.44-fold and GABA by 1.02- to 4.26-fold (Tables 2, 3). Asparagine accumulation was enhanced 1.55- to 2.4-fold by salt stress, but only in pericarp. Salt stress had little effect on the accumulation of aspartate in either tissue or on that of asparagine in columella tissue. Proline accumulation, which is a reliable indicator of the response to salt stress in higher plants, was strongly enhanced by salt stress by 5.36- to 17.50-fold at all stages in both pericarp and columella tissue (Table 2). The effect of salt stress on amino acid accumulation was higher in pericarp tissue than in columella tissue, especially GABA and proline accumulation (Tables 2, 3).

Table 2

Amino acid contents (μmol g FW⁻¹) in developing tomato fruits grown under control (0 mM NaCl) and salt stress (160 mM NaCl) conditions

Amino acid	Tissue	Treatment	IMG	MG	Yel	Red
Glutamine (μmol g FW-1)	Pericarp	Control	19.8 ± 3.1	19.2 ± 3.3	16.0 ± 2.5	14.4 ± 1.2
		+NaCl	27.5 ± 7.2	32.2 ± 6**	28.8 ± 6**	18.2 ± 1.1**
	Columella	Control	16.6 ± 4.7	22.2 ± 4.3	16.6 ± 3.8	11.0 ± 1.2
		+NaCl	27.6 ± 3.2**	32.2 ± 3.5*	23.2 ± 8.4	12.1 ± 0.8
Glutamate (μmol g FW-1)	Pericarp	Control	3.0 ± 0.6	3.6 ± 0.2	3.7 ± 0.5	5.4 ± 0.5
		+NaCl	5.8 ± 1.8 *	6.3 ± 1.5*	6.7 ± 1.5*	10.3 ± 1.4**
	Columella	Control	5.4 ± 1.1	4.3 ± 0.4	4.2 ± 0.9	5.7 ± 0.2
		+NaCl	7.4 ± 1.2*	10.4 ± 1.5**	9.5 ± 6.1	7.8 ± 1**
GABA (μmol g FW-1)	Pericarp	Control	11.5 ± 1.9	14.1 ± 1.8	6.4 ± 0.7	3.0 ± 0.2
		+NaCl	28.3 ± 10.1*	19.4 ± 5.1	17.3 ± 1.7**	13.0 ± 1.3**
	Columella	Control	20.0 ± 5.2	38.6 ± 4.1	21.3 ± 2.5	7.5 ± 0.7
		+NaCl	26.9 ± 1.8*	39.3 ± 2.1	28.6 ± 6.7	16.0 ± 0.7**
Asparagine (μmol g FW-1)	Pericarp	Control	0.7 ± 0.1	1.3 ± 0.1	1.3 ± 0.2	1.2 ± 0.0
		+NaCl	1.4 ± 0.3**	2.0 ± 0.3**	2.3 ± 0.5**	3.0 ± 0.6**
	Columella	Control	1.2 ± 0.3	2.0 ± 0.2	1.5 ± 0.5	0.9 ± 0.2
		+NaCl	1.3 ± 0.3	2.1 ± 0.1	2.0 ± 1.1	1.1 ± 0.0
Aspartate (μmol g FW-1)	Pericarp	Control	0.9 ± 0.1	2.6 ± 0.6	14.6 ± 0.2	37.3 ± 3.4
		+NaCl	2.2 ± 0.4**	4.2 ± 1*	17.7 ± 1.8*	47.5 ± 30.2
	Columella	Control	0.8 ± 0.2	3.9 ± 0.6	16.8 ± 1.9	30.5 ± 4.3
		+NaCl	1.1 ± 0.1*	3.9 ± 0.5	14.2 ± 6.0	27.0 ± 3.3
Proline (μmol g FW-1)	Pericarp	Control	2.9 ± 0.9	1.3 ± 0.3	1.1 ± 0.1	1.5 ± 0.2
		+NaCl	50.5 ± 9.8**	22.3 ± 6.2**	14.9 ± 7.2**	20.7 ± 5.8**
	Columella	Control	1.3 ± 0.4	1.2 ± 0.2	0.8 ± 0.1	0.8 ± 0.1
		+NaCl	11.2 ± 2.2**	7.8 ± 2.3**	4.5 ± 2.3*	7.3 ± 1.8**

The asterisks indicate statistical significance of means in the same tissue and the same developmental stage estimated using Fisher’s PLSD test (*P < 0.05; **P < 0.01).

IMG, immature green; MG, mature, green; Yel, yellow; and Red, red.

Table 3

The relative ratio of the major C/N metabolites in tomato fruits grown under the salt stress condition (160 mM NaCl) vs. that of those grown under the control (0 mM NaCl) condition in each tissue

	Tissue	Soluble solids	IMG	MG	Yellow	Red
Organic acid	Pericarp	Pyruvate	1.18	0.90	1.79**	0.86
		Citrate	0.74	1.34*	1.61**	1.21*
		Succinate	1.52	1.09	1.51**	1.85
		Fumarate	1.41	1.12*	1.28**	1.38*
		Malate	3.55	0.65	0.29*	0.36
		2-Oxoglutarate	0.99	0.99**	1.03*	1.00
	Columella	Pyruvate	0.73*	0.44**	1.01	1.11
		Citrate	0.96	1.66**	0.51**	0.90
		Succinate	0.96	0.52**	1.50	0.66**
		Fumarate	0.86*	0.80**	1.40	1.24
		Malate	1.36	0.72	0.24	0.57*
		2-Oxoglutarate	0.87**	0.78**	1.39*	1.33
Amino acid	Pericarp	Glutamine	1.39	1.67**	1.80*	1.27**
		Glutamate	1.92*	1.75*	1.79*	1.92**
		GABA	2.45*	1.38	2.73**	4.26**
		Asparagine	1.94**	1.55**	1.76**	2.40**
		Aspartate	2.43**	1.61*	1.21*	1.27
		Proline	17.50**	16.86**	13.30**	13.93**
	Columella	Glutamine	1.66**	1.45*	1.40	1.10
		Glutamate	1.38*	2.44**	2.24	1.36*
		GABA	1.34*	1.02	1.70	2.14**
		Asparagine	1.09	1.03	1.37	1.19
		Aspartate	1.44*	0.99	0.85	0.88
		Proline	8.79**	6.60**	5.36*	9.25**

Asterisks indicate the statistical significance of means in the same tissue and developmental stage estimated using Fisher’s PLSD test (*P < 0.05; **P < 0.01).

IMG, immature green; MG, mature green; Yel, yellow; and Red, red.

In the GC-MS measurements, the soluble sugar contents (sucrose, glucose and fructose) of the developing fruit were also determined in both pericarp and columella tissue under control and saline conditions (Supplementary Table S2). Salt stress tended to enhance the accumulation of soluble sugars throughout fruit development, except the IMG stage. The effects of salt stress on soluble sugar accumulation were clearer in the pericarp than in the columella (Supplementary Table S2).

Expression of key enzymes of the TCA cycle and peripheral metabolic pathways

GAD activity and gene expression. GABA levels were high in early developing fruit and rapidly decreased during ripening, and they were enhanced by salt stress (Table 2). To dissect this pattern of GABA accumulation, the gene expression and activity of the GABA biosynthetic enzyme glutamate decarboxylase (GAD) were analyzed (Fig. 2). Because Akihiro et al. (2008) reported that the important isoforms for GABA biosynthesis in fruit are SlGAD2 and SlGAD3, we also focused on these two genes in this work. Consistent with the GABA accumulation data, both of the genes were highly expressed in the IMG and MG stages, and their expression declined after the Yel stage under both the control condition and salt stress. However, unlike the GABA accumulation, salt stress suppressed the expression of SlGAD2 at the IMG stage and that of SlGAD3 at the IMG and MG stages (Fig. 2A, B). Although the expression of SlGAD2 was enhanced by salinity at the MG stage and that of SlGAD3 was enhanced at the Yel and Red stages, these effects were very small. Salt stress did not affect GAD enzymatic activity per protein in the IMG or MG stages (Fig. 2C), but it was significantly enhanced in both stages on a fresh weight basis (Fig. 2D).

Fig. 2

Relative expression level and enzymatic activity of glutamate decarboxylase in developing tomato fruits grown under control (0 mM NaCl) and saline conditions (160 mM NaCl). (A) Transcriptional level of SlGAD2. (B) Transcriptional level of SlGAD3. Open and filled bars indicate control (0 mM NaCl) and saline treatments (160 mM NaCl), respectively. The values are expressed relative to the level of SlGAD2 expression in control fruit at the IMG stage. The values are means ± SD (n = 4). (C) GAD activity per total protein in early developmental stages. (D) GAD activity on a fresh fruit weight basis in early developmental stages. The open and shaded bars indicate control (0 mM NaCl) and saline treatments (160 mM NaCl), respectively. The horizontal axis indicates fruit developmental stages. IMG, immature green; MG, mature green; Yel, yellow; and Red, red. The values are means ± SD (n = 5). Asterisks indicate the statistical significance of means in the same developmental stage estimated using Fisher’s PLSD test (*P < 0.05, **P < 0.01).

Gene expression of key enzymes connecting glycolysis and the TCA cycle. To clarify how the balance between glycolysis and the TCA cycle is regulated at the molecular level during fruit development and also under salt stress, transcriptional analyses of phosphoenolpyruvate carboxylase (PEPC1, 2), phosphoenolpyruvate carboxykinase (PEPCK), malate dehydrogenase (cMDH, mMDH, gMDH1, 2), malic enzyme (ME1, ME2) and pyruvate kinase (PK1, PK2) were performed at each stage of fruit development using quantitative reverse transcription–PCR (qRT–PCR) (Fig. 3). PEPC1 and PEPC2 showed the highest expression in the MG stage. While PEPC1 expression declined gradually during ripening, that of PEPC2 declined rapidly after the Yel stage. PEPC2 was up-regulated by salt stress at all stages, whereas PEPC1 did not show a clear response to salt stress (Fig. 3A). In contrast to PEPC2, the expression of PEPCK was lower at the early stages and then increased during fruit ripening; its expression was down-regulated during the early stages and tended to increase during ripening in response to salt stress (Fig. 3A).

Fig. 3

Relative expression levels of genes encoding key enzymes of the TCA cycle and peripheral pathways in developing tomato fruits grown under control (0 mM NaCl, white bar) and saline conditions (160 mM NaCl, black bars). (A) PEPC1, PEPC2, PEPCK. (B) cMDH, mMDH, gMDH1, gMDH2. (C) ME1, ME2. (D) PK1, PK2. (E) CS. (F) 2OGDE1, 2OGDE2. The values are relative to the expression levels of (A) PEPC1, (B) cMDH, (C) ME1, (D) PK1, (E) CS and (F) 2OGDE1 in control fruits at the IMG stage. The horizontal axis indicates fruit developmental stages: IMG, immature green; MG, mature green; Yel, yellow; and Red, red. The values are means ± SD (n = 4). Asterisks indicate the statistical significance of means in the same developmental stage estimated using Fisher’s PLSD test (*P < 0.05, **P < 0.01).

In tomato, four MDH genes have been reported in previous studies (Nunes-Nesi et al. 2005). Among them, cytosolic MDH (cMDH, AY725475), mitochondrial MDH (mMDH, AY725474) and glyoxysomal MDH (gMDH1, AY725476) were chosen for qRT–PCR analyses. We also newly obtained a gMDH homolog cDNA (SGN-U574918) from an EST and included it as gMDH2 in the analyses. In the control fruit, cMDH was highly expressed during the early stages and then declined after the Yel stage. Under salt stress, it was significantly enhanced at the MG and Yel stages, whereas it was suppressed at the IMG and Red stages (Fig. 3B). The mMDH gene peaked at the Yel stage in control fruit. While salt stress enhanced its transcription by 1.5- and 1.9-fold at the MG and Red stages, transcription was suppressed at the Yel stage (Fig. 3B). The two gMDH genes showed different expression patterns during fruit development. The expression of gMDH1 peaked at the MG stage; however, there was almost no change at the other stages. Salt stress suppressed gMDH1 expression at the MG stage (Fig. 3B). However, in control fruit, gMDH2 expression was low at the IMG and MG stages and increased after the Yel stage. Relative to the IMG stage, the transcript level of gMDH2 was about 16-fold higher at the Yel and Red stages. Although gMDH2 expression was moderately suppressed at the Yel stage, it was significantly enhanced by 1.56-fold at the Red stage by salt stress compared with control fruit of the same stage (Fig. 3B). Among the MDH family, the isoform that exhibited the highest expression level in developing tomato fruit was gMDH2 and the lowest was cMDH (Fig. 3B). The expression level of gMDH2 was 59.7- and 36.6-fold higher than that of gMDH1, and 9.1- and 12.6-fold higher than that of mMDH at the Yel and Red stages, respectively. In this work, the transcript levels of ME, which synthesizes pyruvate by decarboxylating malate, were also analyzed. Two tomato isoforms were obtained as cDNAs from the EST database based on sequence homology, and designated as ME1 and ME2. The expression of ME1 remained almost unchanged throughout development in control fruit. Salt stress enhanced its expression only at the Yel stage (Fig. 3C). However, ME2 expression increased during fruit development in both conditions. Salt stress did not have an obvious effect on ME2 expression except at the Yel stage (Fig. 3C).

Two tomato EST clones encoding different types of pyruvate kinase, PK1 and PK2, were also newly obtained through a homology search based on sequence information from other plant species. The expression of both the pyruvate kinase isozyme A (PK1) and the pyruvate kinase cytosolic isozyme (PK2) genes was low at the early stages and increased during ripening; PK1 and PK2 peaked at the Red and Yel stages, respectively, and both genes were up-regulated by salt stress (Fig. 3D).

Gene expression of citrate synthase (CS) and 2-oxoglutarate dehydrogenase (2OGD). The transcript levels of CS were investigated by qRT–PCR in fruit grown under control conditions and salt stress (Fig. 3E). CS expression gradually increased during fruit development and peaked at the Red stage; however, salt stress led to a down-regulation at the IMG and MG stages. Regarding 2OGD, two isoforms exist in the tomato database. The transcript levels of both 2OGDE genes were lower in the early stages and increased during ripening. 2OGDE1 was up-regulated at all stages of fruit development by salt stress and peaked at the Red stage. 2OGDE2 expression also peaked at the Red stage, and it was down-regulated by salinity at the IMG and Red stages (Fig. 3F).

Gene expression of GS, nitrate reductase(NR) and 2-oxoglutarate/malate translocator (2OGMT). To analyze the effects of salinity on nitrogen assimilation, the transcript levels of GS and NR, both of which are key nitrogen assimilation enzymes in higher plants, were investigated in developing fruit grown under both conditions (Fig. 4). To date, two GS-encoding genes, GS1 (cytoplasm type) and GS2 (plastid type), have been reported in tomato and are proposed to regulate nitrogen assimilation and glutamine biosynthesis in developing tomato fruits (Gallardo et al. 1988, Scarpeci et al. 2007). In the qRT–PCR analyses, GS1 and GS2 showed contrasting transcriptional patterns: GS2 was mainly expressed at the early stages and declined after the Yel stage (Fig. 4B), whereas GS1 expression was low at the early stages but became the dominant isoform at the ripening stages (Fig. 4A). The expression level of GS1 was 20- to 30-fold higher than that of GS2. Although salt stress suppressed GS1 expression from the IMG to the Yel stages, it enhanced GS1 expression by 1.5-fold at the Red stage (Fig. 4A). GS2 was up-regulated 1.5-fold at the MG stage, but it was suppressed to 15% of the control at the Yel stage under salt stress (Fig. 4B). NR exhibited an expression pattern similar to that of GS1, which was low at the IMG and MG stages and higher after the Yel stage (Fig. 4C). Salt stress suppressed its transcription only at the IMG stage and tended to enhance it in the later developmental stages. The transcript levels of 2OGMT were higher in ripening stages than in the earlier stages and peaked at the Yel stage under control conditions. Although 2OGMT was down-regulated by salt stress at the IMG and Yel stages, a consistent effect of salinity was not observed between fruit stages (Fig. 4D).

Fig. 4

Relative expression levels of GS, NR and 2OGMT in developing tomato fruits grown under control (0 mM NaCl) and saline conditions (160 mM NaCl). (A) GS1, (B) GS2, (C) NR, (D) 2OGMT. The open and shaded bars indicate control (0 mM NaCl) and saline treatments (160 mM NaCl), respectively. In A and B, the values presented are relative to the expression level of GS1 in control fruits at the IMG stage. In C and D, the values presented are relative to the expression level of each gene in control fruits at the IMG stage. The horizontal axis indicates fruit developmental stages: IMG, immature green; MG, mature green; Yel, yellow; and Red, red. The values are means ± SD (n = 4). Asterisks indicate the statistical significance of means in the same developmental stage estimated using Fisher’s PLSD test (*P < 0.05, **P < 0.01).

[¹⁴C]GABA tracer analysis during fruit ripening

To clarify the fate of the GABA that accumulates in the fruit during ripening, tracer analysis was performed with [¹⁴C]GABA (Fig. 5). Fruits at the MG, orange (Or) and Red stages were incubated with [¹⁴C]GABA (Fig. 5A) for 6 h and analyzed to determine the ratio of GABA incorporated into CO₂ and into insoluble and soluble fractions. As shown in Fig. 5B, 43, 73 and 53% of the [¹⁴C]GABA incorporated into MG, Or and R fruits was discharged as [¹⁴C]CO₂. However, 47, 17 and 37% of the [¹⁴C]GABA was retained in fruit tissues in soluble form (Fig. 5B). The ¹⁴C retained in the insoluble form constituted about 10% of the total throughout all of the tested fruit stages. [¹⁴C]GABA incorporation was highest at the MG stage and then decreased during ripening (Fig. 5C). The intake of [¹⁴C]GABA by the Or and Red stage fruits was a quarter and one-eighth that of MG stage fruits, respectively.

Fig. 5

[¹⁴C]GABA tracer analysis during the ripening of control fruit. (A) Feeding of [¹⁴C]GABA into the fruit. A total of 300 mg of fruit cubes was incubated with 2 ml of reaction buffer containing 0.1 μCi of 1-[¹⁴C]GABA for 6 h. (B) Distribution ratio and (C) intake of ¹⁴C (%) in ripening fruit. The filled, open and shaded columns indicate the soluble fraction, the insoluble fraction and the CO₂ fraction, respectively. The horizontal axis indicates fruit developmental stages: MG, mature green; Or, orange; and Red, red. The values are means ± SD (n = 3).

Discussion

In this work, we investigated the metabolic shifts of organic acids and amino acids during fruit development of tomato plants exposed to salt stress. ‘Micro-Tom’ exhibited considerable tolerance to elevated salt stress (Fig. 1A) with an increasing brix (%) (1.6-fold that of control fruit at the Red stage) and decreasing fruit size (about 80–60% of the control condition). Although plant size decreased under salt stress, the ratio of fruit to foliage remained almost unchanged in terms of both fresh and dry weight, which indicated that the source–sink balance was the same in both growth conditions, as reported in our previous work (see Supplementary Data in Yin et al. 2010). In contrast, the accumulation of proline, which is a reliable indicator of responses to environmental stresses such as drought and salinity in higher plants (Ashton and Verma 1993, Claussen 2005, Zushi and Matsuzoe 2006), was strongly promoted by 160 mM NaCl at all fruit developmental stages in both pericarp and columella tissue (Table 2). These results indicate that the level of salt stress adopted in these experiments was effective, and this experimental system using ‘Micro-Tom’ is available for analyses of metabolic alterations under salt stress in fruit.

In tomato fruits, the dry matter is about 50% soluble sugars (glucose 22%, fructose 25% and sucrose 1%) and 13% organic acids (citrate 9% and malate 4%) (Davies and Hobson 1981). We found that soluble sugar accumulation was enhanced by salt stress in most fruit developmental stages, which reached a maximum of 2.60-fold higher than the control condition at the Yel stage (Supplementary Table S2). Actually, the effect of the stress on the sugar accumulation was apparently higher than that on suppression of fruit weight. Additionally, the responsiveness of the metabolite level to the salt stress was different even in the same fruit (Table 3). These results indicate that active metabolic modification should occur during this event along with a concentration effect due to the reduction of fruit size. It has been reported that salt stress prominently enhances and extends the accumulation of starch during early fruit development (Balibrea et al. 1996, Gao et al. 1998, Balibrea et al. 1999). Because carbohydrate influx into fruit only occurs in the early stages (Yin et al. 2010), the enhanced starch accumulation probably results in an increase in soluble solids in ripe fruit under salt stress. Interestingly, the responsiveness to salinity was higher in the pericarp than in the columella (Table 3). Differential carbohydrate accumulation and metabolism among tissue fruit types such as pericarp, columella and placenta was reported in tomatoes grown under normal conditions by Hazem et al. (2004) and Baxter et al. (2005). Our results indicate that salinity responsiveness varies among tissue types in tomato. At the IMG stage, although sugars function as osmoprotectants in plant cells, glucose, fructose and total sugar levels were obviously suppressed by salt stress (Supplementary Table S2). Because carbohydrates imported into fruit at the IMG stage are mainly stored as starch, the active conversion of glucose and fructose to starch would cause such a decrease in glucose and fructose levels in salt-stressed fruit. Salt stress enhances starch accumulation in the early developmental stages (Yin et al. 2010). This responsiveness suggests that salt stress affects the intensity of the metabolic process but does not change the process itself in tomato fruit development.

PEPC2 and PEPCK regulate carbohydrate flow in the oxaloacetate (OAA) and PEP metabolic pathway

PEP is one of the most important metabolites for both glycolysis and the TCA cycle. At the last step of glycolysis, PEP is converted to pyruvate by PK, and then it is converted to citrate by CS or to OAA by PEPC. Meanwhile, PEP is synthesized from OAA by PEPCK in the cytoplasm as the first step of gluconeogenesis (Plaxton 1996). PK, PEPC and PEPCK are regulatory steps in each pathway. In tomato, two PEPC isoforms have been reported: PEPC1 is systemically expressed in all organs and tissues, and PEPC2 is specifically expressed in pericarp, placenta and gel tissue of early developing fruit (Guillet et al. 2002). In contrast, PEPCK is a single-copy gene that is highly expressed in pericarp tissue of ripening tomato fruits (Bahrami et al. 2001). However, the physiological function of PEPCK in ripening fruit is unclear, although early work suggested that gluconeogenesis occurs in the ripening stage (Halinska and Frenkel 1991). In this study, we investigated the expression patterns of PEPC and PEPCK in developing fruit. As shown in Fig. 3A, PEPC2 was more highly expressed at the MG stage, whereas PEPCK was more highly expressed at the Yel and Red stages. Salt stress tended to promote the expression of both genes at their dominant stages. Because such up-regulation of PEPCK by salt stress was also reported in another cultivar (Saito et al. 2008a), this is a common reaction in tomato fruits. Although the PEPC1 transcript was detected in fruit, its expression was much lower than that of PEPC2 (Fig. 3A), and it did not show a clear response to salt stress, indicating that it is not involved in this event. The opposite expression profiles of PEPC2 and PEPCK suggest that the carbohydrate metabolic flow runs from glycolysis to the TCA cycle in the early developmental stages and then switches to the opposite direction during the ripening stages (Fig. 3A).

An active shunt composed of MDH, PEPCK and PK is involved in organic acid metabolism

The TCA cycle connects glycolysis to the organic acid and amino acid metabolic pathways. In tomato fruits, the major organic acids are malate in the early stages and citrate in the ripening stages (Hobson and Davies 1971, Mattoo et al. 1975, Tucker 1993). Our quantification of organic acids yielded results similar to those of previous reports; most TCA cycle-related organic acids were very low, except for malate and citrate. In particular, the malate content peaked at the IMG to MG stages and then dramatically decreased during ripening. Its responsiveness to salt stress was different from that of the other organic acids, which were clearly suppressed in columella and pericarp tissue after the MG stage (Tables 1, 3). Malate is alternatively converted to OAA by MDH or to pyruvate by ME. However, to date, there has been very little information regarding which is the major metabolic pathway in tomato fruits and where the malate is metabolized in fruit cells. To determine this, we performed qRT–PCR analyses with primers specific for four MDH genes and two ME genes. Among the MDH genes, gMDH2 showed the highest expression level and an expression pattern similar to that of PEPCK (Fig. 3B). The expression levels of cMDH, mMDH and gMDH1 were lower than that of gMDH2, especially in the Yel and Red stages when a dramatic decrease in malate was observed (Fig. 3B, Table 1). The cellular localization of MDH enzymes has not been fully determined in tomato except for that of mitochondrial MDH (Nunes-Nesi et al. 2005), and the remaining genes were annotated by in silico predictions based on their sequences. Because MDH activity in ripening fruit is mainly detected in the cytoplasm and not in glyoxysomes (Jeffery et al. 1986), and because the level of cMDH transcripts was extremely low (Fig. 3B), gMDH2 probably encodes a cytoplasmic isoform in tomato. The expression level of ME1 was higher than that of ME2. However, both levels were much lower than that of gMDH2 (Fig. 3B, C). These results suggest that the dehydration of malate mainly occurs in the cytoplasm. Indeed, the amount of ME decreases during ripening (Knee and Finger 1992, Bahrami et al. 2001). On the other hand, the expression of PEPCK increased markedly after the Yel stage (Fig. 3A). These results lead to the hypothesis that the dramatic decrease in malate at the Red stage (Table 1) could be the result of metabolism by the cytoplasmic MDH and the PEPCK-related pathway.

The levels of PK1 and PK2 transcripts peaked in fruit at the Red and Yel stages, respectively (Fig. 3D). Although both isoforms were enhanced by salt stress in the ripening stages, PK2 is probably the major isoform involved in pyruvate production in tomato fruits because its expression pattern corresponds well to the enzymatic activity reported by Hazem et al. (2004), and PK1 expression levels are much lower than those of PK2 (Fig. 3D). Furthermore, the up-regulation of gMDH2, PEPCK and PK2 (Fig. 3) by salt stress strongly suggests that active OAA consumption occurs for the biosynthesis of PEP and subsequently pyruvate in ripening fruit. In contrast to malate, citrate and CS expression increased as fruit ripened (Table 1, Fig. 3E). These results also suggest that malate was converted to citrate via PEP during ripening stages. Considering the results of the metabolite profiling and the transcriptional analyses, we conclude that organic acid synthesis is globally activated by salt stress and should involve the pathway consisting of malate–OAA–PEP–pyruvate–citrate (Fig. 6). However, we do not rule out the possibility that the accumulated malate is consumed for gluconeogenesis. To clarify the ratio of malate available for gluconeogenesis and/or citrate biosynthesis during ripening, transgenic analysis using a knockdown strategy for the PEPCK or PK2 genes will be indispensable.

Fig. 6

Schematic illustration of the C/N metabolic pathways of and around the TCA cycle. Compounds that prominently accumulate in fruit are enclosed in colored boxes (green, IMG and MG stages; red, Yel and Red stages). The boxed enzymes indicate the regulatory steps in the pathway. The metabolic pathway of GABA during fruit ripening is indicated by red arrows.

In the many tested genes in Figs. 3 and 4, the changes in the expression levels due to the salt stress were not significantly different from or were smaller than those in the metabolites (Tables 1, 2). Actually, total protein yield was high in salt-stressed fruit compared with the control (Supplementary Fig. S1). Therefore, the salt stress would affect the stability of an enzyme protein in addition to the substrate availability (Tables 1, 2). Because a level of primary metabolite is generally a balance of continuous biosynthesis and degradation, the smaller fluctuation in the expression level of metabolic enzymes is not an uncommon phenomenon.

GABA is utilized as a substrate for respiration and functions as an energy source for climacteric ripening of tomato fruits

In this work, the levels of amino acids such as glutamine, glutamate, GABA, asparagine and aspartate were quantified in salt-stressed fruit by GC-MS (Table 2). All of the quantified amino acids increased under salt stress except for aspartate. While the glutamine, glutamate and asparagine levels did not fluctuate significantly, the GABA and aspartate levels changed dramatically between the early and ripening stages. As reported previously (Rolin et al. 2000, Akihiro et al. 2008, Saito et al. 2008b), GABA accumulation peaked at the MG stage and then dramatically decreased during the ripening stages (Table 2). Akihiro et al. (2008) showed the essential roles of GAD (encoded by SlGAD2 and SlGAD3) and 2-oxoglutamate-dependent GABA transaminase (GABA-TK) in GABA accumulation and metabolism in tomato fruits. In this work, the expression of both of the GAD genes and their enzymatic activity per protein were not enhanced in the early developmental stages of tomato fruits (Fig. 2A–C). However, the enzymatic activity per fresh weight was higher at the IMG and MG stages (Fig. 2D). Indeed, the total protein yield on a fresh weight basis was higher by 1.5- to 1.9-fold in salt-stressed fruit compared with control fruit (Supplementary Fig. S1). In conjunction with substrate (glutamate) availability, these enhanced GAD protein levels would result in higher GABA levels in the early developmental stages of tomato fruits.

GABA is generally metabolized to succinate and then flows back to the TCA cycle in plant cells (Bouché and Fromm 2004). However, even though tomato fruits are one of the highest GABA accumulators among vegetable crops, GABA metabolism and its physiological function during tomato fruit ripening has remained unclear. In this work, we demonstrated that ¹⁴C-labeled carbon dioxide was discharged from fruit fed ¹⁴C-labeled GABA during ripening (Fig. 5). Although it is well known that a burst of respiration occurs in isolated tissues as a result of wounding, this [¹⁴C]tracer analysis result directly proves that GABA is utilized as a substrate for respiration by the TCA cycle during fruit ripening (Fig. 5). Furthermore, thin-layer chromatography (TLC) with the soluble fraction from fruit at the Or stage fed ¹⁴C-labeled GABA proved that GABA is converted to malate, citrate, cis-aconitate, isocitrate and 2-oxoglutarate (Supplementary Fig. S2). These results suggest that the dissimilated GABA is converted to malate and then is metabolized to 2-oxoglutarate, passing through a shunt consisting of OAA–PEP–pyruvate and the TCA cycle (presented as red arrows in Fig. 6). Considering that tomato is a climacteric-type fruit, and ethylene production and its respiration dramatically increase during ripening, the shunt passing through OAA–PEP–pyruvate is beneficial for the plant because MDH and PK generate NADH₂⁺ and ATP via conversion of malate to OAA and of PEP to pyruvate (Fig. 6).

GS1 and NR play important roles in nitrogen assimilation during fruit ripening

GS is a regulatory enzyme for nitrogen assimilation as well as glutamine synthesis in higher plants. Most plant species have two GS isoforms localizing as cytosolic (GS1) and plastidic (GS2) forms (Gallardo et al. 1988, Scarpeci et al. 2007). According to early studies, the GS2 protein mainly functions during the earlier stages of fruit ripening, it keeps the fruit green and its expression is promoted by low temperatures and excess nitrate (Boggio et al. 2000, Lu et al. 2005, Zozaya-Hinchliffe et al. 2005). In contrast, GS1 expression is induced at the Red stage under high N conditions (Scarpeci et al. 2007). In this work, GS1 expression was much higher than that of GS2, reaching 3- to 5-fold that of control fruit in the IMG and MG stages (Fig. 4A, B), suggesting that GS1 mainly contributes to glutamine biosynthesis in ripening fruit. NR showed a coordinated expression pattern similar to that of GS1, and it tended to be promoted by salt stress at the Yel and Red stages (Fig. 4C). Nitrogen availability should be higher in ripening fruit cells because the photosynthetic enzymes are degraded in the transition process from chloroplast to chromoplast. The higher expression of GS1 and NR at the Yel and Red stages indicates that active nitrogen assimilation is occurring in fruit during ripening. Furthermore, the higher expression of 2OGDE and 2OGMT during the ripening stages (Figs. 3F, 4D) and the TLC results (Supplementary Fig. S2) indicating that GABA is converted to 2-oxoglutarate suggest that the accumulated citrate functions as a carbon skeleton donor for glutamate/glutamine biosynthesis (Fig. 6).

Taken together, these results indicate that the pool of GABA accumulated in mature green fruit is utilized as a substrate for respiration by passing through the TCA cycle, and it functions as an energy source for the climacteric ripening of tomato fruits. This previously unreported shunt composed of MDH, PEPCK and PK is provably involved in metabolism.

Materials and Methods

Plant materials and sampling

Seeds of ‘Micro-Tom’ (Scott and Harbaugh 1989) were sown on moist paper in a culture room at 25°C under 16 h light/8 h dark conditions. After germination, seedlings were transplanted into rockwool (5 × 5 × 5 cm) on a slide and grown for 6 weeks with a commercial nutrient solution (Otsuka A; Otsuka Chemical Co., Ltd., Osaka, Japan) adjusted to EC (electrical conductivity) 2.0 dS m⁻¹. After flowering of the first truss, half of the plants were transferred to saline nutrient solution that was adjusted to EC 15.0 dS m⁻¹ by adding NaCl (equivalent to 160 mM NaCl), while the remaining plants were kept in nutrient solution at EC 2.0 dS m⁻¹ (0 mM NaCl) as a control. Fruits were sampled from plants grown under both conditions at four stages: IMG, MG, Yel and Red, which correspond to 14, 22, 34 and 42 days after flowering, and they were kept at −80°C until use. For the metabolite quantification, these fruits were divided into pericarp and columella tissue before freezing, as shown in Fig. 1C.

Metabolite analysis

The frozen fruits were divided into pericarp and columella tissue and ground in liquid nitrogen using a pestle and mortar. For each sample, 50 mg of frozen tissue powder was put into a 2 ml plastic tube and rapidly homogenized with 250 μl each of methanol and chloroform. After the addition of 50 μl of 0.2 mg ml⁻¹ ribitol solution as an internal standard and 175 μl of MilliQ water, the samples were vigorously mixed. These extracts were centrifuged at 18,000 × g for 10 min at room temperature. Then 200 μl of the supernatant was filtered through an Amicon (Millipore, Billerica, MA, USA) Ultrafree-MC (Millipore/719-23-32-02), and part of the flowthrough fraction (160 μl) was evaporated to dryness in a centrifuge evaporator (model 1910; Kubota, Tokyo, Japan). For methylation, 40 μl of methoxylamine (20 mg ml⁻¹ pyridine) was added to the samples and incubated for 90 min at 37°C. Trimethylsilylation was performed by adding 50 μl of N-methyl-N-(trimethylsilyl)-trifluoroacetamide (MSTFA) solution for 30 min at 37°C.

A GC 6890 (Agilent Technologies, Santa Clara, CA, USA) was operated under electronic pressure control and equipped with a split/splitless capillary inlet. A 30 m DB-17-ms column (0.25 mm ID, 0.25 μm film thickness; J&W Scientific, Sacramento, CA, USA) was used. A 1 μl aliquot of sample was injected in the splitless mode using the methods described by Akihiro et al. (2008).

Enzyme extraction and measurement of GAD enzymatic activity

Frozen fruits were ground in liquid nitrogen and 200 mg of frozen powder was homogenized in a 5-fold volume of ice-cold extraction buffer [0.1 M Tris–HCl, pH 7.0, 10 mM dithiothreitol, 5 mM EDTA, 1 mM pyridoxal-5-phosphate, 1% (w/v) insoluble polyvinylpyrrolidone] (Rolin et al. 2000). The homogenate was centrifuged at 10,000 × g for 15 min at 4°C. A total of 800 μl of the supernatant was collected and desalted with Sephadex G-50 (GE Healthcare, UK). GAD enzymatic activity was measured as glutamate-dependent GABA production according to Akihiro et al. (2008).

RNA extraction and first-strand cDNA synthesis

Frozen fruits (5 g) were ground to a fine powder in liquid nitrogen and mixed with RNA extraction buffer [100 mM Tris–HCl (pH 9.5), 100 mM LiCl, 10 mM EDTA, 1% SDS] and TE-saturated phenol (1/1, v/v). The mixture was incubated first at 80°C for 15 min and then at room temperature for 20 min with shaking, and finally extracted three times with phenol/chloroform (1/1, v/v). After phase separation, the aqueous phase was rewashed with chloroform. Total nucleic acids were precipitated with one-third volume of 8 M LiCl. The pellet was dissolved in 500 μl of diethyl pyrocarbonate (DEPC)–water, extracted with phenol/chloroform (1/1, v/v) and rewashed with chloroform. Total RNA was precipitated using 3 M sodium acetate (pH 5.3) and 99.9% ethanol. The pellet was washed with 70% ethanol, dried, re-suspended in DEPC–water and stored at −80°C until use.

A 10 μg aliquot of total RNA was treated with DNase and then cleaned up with phenol/chloroform (1/1, v/v), 3 M sodium acetate, 99% ethanol and then 70% ethanol. The RNA was then dried and resuspended in DEPC–water. The RNA was subjected to reverse transcription with an Oligo-(dT)¹⁵ Primer (Promega, Madison, WI, USA) using Super Script II (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions for gene expression analysis.

Database search of EST sequences encoding sugars, amino acids and organic acid metabolic enzymes

For transcriptional analyses, sequence information of the target genes encoded in the tomato genome was obtained through searches of multiple databases. The International Tomato Genome Sequencing Project is accumulating genome and EST information, which is available through the Sol Genomic Network (http://www.sgn.cornell.edu/index.pl). In this experiment, we used the DDBJ database (http://www.ddbj .nig.ac.jp/index-j.html), the TIGER database (http://compbio .dfci.harvard.edu/tgi/cgi-bin/tgi/Blast/index.cgi) and MiBase (http://www.pgb.kazusa.or.jp/kaftom/index.html) to search for and design gene-specific primers for qRT–PCR analysis. The cDNA fragment sequences for all of the enzymes were identified from these databases (Supplementary Table S1).

The gene-specific primers were designed with Amplify ver.3.1.4 (Bill Engels; University of Wisconsin-Madison, WI, USA) (Supplementary Table S1). We investigated transcript levels of the target genes with qRT–PCR. The qRT–PCR analyses were carried out with an Mx 3000P qRT–PCR system (Stratagene, La Jolla, CA, USA) according to the manufacturer’s instructions. For normalizing qRT–PCRs, the endogenous actin gene was used as an internal standard (Moniz and Drouin 1996). Specific amplifications were confirmed by a single dissociation peak and calibration curves. Gene expression levels were calculated relative to the expression of the actin gene by the ΔΔCt method (Pfaffl 2001).

[¹⁴C]GABA tracer analysis

Fruits at the Or stage (38 d after flowering) were used for the [¹⁴C]GABA tracer experiment in addition to MG and Red fruits to dissect the metabolism of GABA during fruit ripening. The fruits were sterilized with 70% ethanol and washed with sterilized water several times. The fruits were cut into 3 mm cubes. A total of 300 mg of the fruit was transferred into a custom-made flask in which a short glass tube was welded to the center of the bottom (Fig. 5A). Before the transfer of the sample, 2 ml of reaction buffer {3.7 × 10³ Bq 1-[¹⁴C]GABA (American Radiolabeled Chemicals Inc., St. Louis, MO, USA), 30 mM sodium phosphate, pH 7.0, 5 mM sucrose} was put into the flask. To adsorb CO₂, a strip of filter paper infiltrated with 100 μl of 20% KOH was put into the test tube. After being stoppered with a silicon plug, the flask was incubated with shaking for 6 h at 27°C.

After incubation, the strip of filter paper and the fruit cubes were recovered separately. The strip was soaked in 10 ml of sterilized water at room temperature overnight and then cpm values were counted by a scintillation counter. The fruit cubes were collected into 2 ml tubes and centrifuged at 10,000 × g for 5 min, and then the supernatant was discarded. The precipitate was washed with 1.5 ml of sterile water three times to remove reaction buffer. After being washed, the collected fruits were frozen in liquid nitrogen and ground to a fine powder by a Tissuelyser (QIAGEN, Valencia, CA, USA). A 1 ml aliquot of 80% ethanol was added to the tube and centrifuged at 15,000 × g for 10 min at 4°C. The precipitate and the supernatant were recovered separately as insoluble and soluble fractions, respectively. For the insoluble fraction, 1 ml of sterilized water was added to the precipitate and it was resuspended. The cpm value of each fraction was counted by scintillation counter. The ratio of incorporated [¹⁴C]GABA was calculated from the total cpm values of the CO₂ and the insoluble and soluble fractions in each ripening stage, and was presented as a percentage.

Statistical analyses

Fisher’s PLSD (protected least significant difference) analyses were performed with StatView^® version 4.5 (SAS Institute Inc., Cary, NC, USA).

Funding

This work was supported in part by the "Research and Development Program for New Bio-industry Initiatives" from Bio-oriented Technology Research Advancement Institution (BRAIN), and a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (No. 16780044 and No. 21580075).

Acknowledgments

The authors thank Mr. Taneaki Tsugane, Chiba Prefectural Agriculture Research Center, for technical advice in the gene expression analyses, and Mr. Masaki Sugiyama, University of Tsukuba, for technical assistance in the GC-MS analyses. ‘Micro-Tom’ seeds were provided by the Gene Research Center, University of Tsukuba, through the National Bio-Resource Project, MEXT, Japan.

Abbreviations

Abbreviations
 
• CS

  citrate synthase

 
• DEPC

  diethyl pyrocarbonate

 
• EC

  electrical conductivity

 
• EST

  expressed sequence tag

 
• GABA

  γ-aminobutyric acid

 
• GAD

  glutamate decarboxylase

 
• GC-MS

  gas chromatography–mass spectrometry

 
• GS

  glutamine synthetase

 
• MDH

  malate dehydrogenase

 
• ME

  malic enzyme

 
• NR

  nitrate reductase

 
• OAA

  oxaloacetate

 
• 2OGD

  2-oxoglutarate dehydrogenase

 
• 2OGMT

  2-oxoglutarate/ malate translocator

 
• PEP

  phosphoenolpyruvate

 
• PEPC

  phosphoenolpyruvate carboxylase

 
• PEPCK

  phosphoenolpyruvate carboxykinase

 
• PK

  pyruvate kinase

 
• qRT–PCR

  quantitative reverse transcription–PCR

 
• TCA

  tricarboxylic acid

 
• TLC

  thin-layer chromatography.