The effect of bacterial isochorismate synthase on the Brassica rapa metabolome
Simoh, S.
Citation
Simoh, S. (2008, June 11). The effect of bacterial isochorismate synthase on the Brassica rapa metabolome. Retrieved from https://hdl.handle.net/1887/12944
Version: Corrected Publisher’s Version
License: Licence agreement concerning inclusion of doctoral thesis in the Institutional Repository of the University of Leiden
Downloaded from: https://hdl.handle.net/1887/12944
Note: To cite this publication please use the final published version (if applicable).
Chapter 6
Metabolome analysis of Brassica rapa transformed with a bacterial gene encoding isochorismate synthase
Sanimah Simoh1,2, Huub JM Linthorst3, Alfons WM Lefeber4, Cornelis Erkelens4, Hye Kyong Kim1, Young Hae Choi1, Robert Verpoorte1
1 Section Metabolomics, Institute of Biology, Leiden University, Leiden, The Netherlands
2 Biotechnology Research Centre, Malaysian Agricultural Research & Development Institute (MARDI), Kuala Lumpur, Malaysia
3 Section Plant Cell Physiology, Institute of Biology, Leiden University, Leiden, The Netherlands
4 Division of NMR, Institute of Chemistry, Leiden University, Leiden, The Netherlands
Abstract
Metabolome analysis by 1-dimensional proton nuclear magnetic resonance (1H-NMR) spectroscopy coupled with multivariate data analysis was carried out in Brassica rapa ssp. oleifera plants transformed with a gene encoding bacterial isochorismate synthase (ICS). Partial least square-discriminant analysis (PLS-DA) on selected signals suggested that the resonances which were dominant in the transgenic plants corresponded to a glucosinolate (neoglucobrassicin), phenylpropanoids (sinapoyl malate, feruloyl malate, caffeoyl malate), organic acids (succinic acid and fumaric acid) and sugars (α- and β- glucose). In contrast, amino acids alanine threonine, valine, and leucine were dominant in the untransformed controls. This result suggests that the presence of the bacterial ICS gene in the genome of B. rapa ssp. oleifera has an influence on some metabolites along the phenylpropanoid and indole pathway and also on primary metabolites (sugars and organic acids).
Keywords: isochorismate synthase, Brassica rapa, NMR spectroscopy, multivariate data analysis, salicylic acid
6.1 Introduction
The metabolome is defined as the quantitative complement of the low molecular weight compounds present in an organism at any specific condition (Fiehn et al, 2001; Nicholson et al., 2004). Metabolomics is a new ‘omic’ that complements the previously developed transcriptomics and proteomics. Any change in the physiological state of a cell as a result of gene deletion or overexpression is reflected in its transcriptome, proteome and metabolome and should be measurable through the metabolome as the characterization of the phenotype (Kell et al., 2005). As a key tool to monitor the ultimate gene expression products (Oliver et al., 1998), profiling the metabolome offers a ‘snapshot’ of the physiological state of a cell at a given condition. By integrating all the ‘omic’
technologies it may subsequently provide us a comprehensive and holistic understanding of cellular response to a specific biological condition (Schauer and Fernie, 2006), thus it is crucial in systems biology. In genetic engineering, metabolome analysis allows us to monitor the ‘substantial equivalence’ between the genetically modified products and their wild type counterpart or whether there is significant alteration of unexpected metabolites due to transgene expression that could be detrimental to the safety of consumers.
The progress of metabolome research is limited by the analytical methods. As the ultimate goal of metabolome analysis is the unbiased and non-targeted identification and quantification of all the metabolites present in the organism (Verpoorte et al., 2007), it requires that all the metabolites are analyzed simultaneously in a small number of steps.
At present there is no single analytical method capable to do so. The analytical methods used for metabolomics are normally based on chromatography hyphenated diverse spectroscopy such as ultraviolet or mass spectrometry, or spectroscopy as itself, e.g.
nuclear magnetic resonance (NMR) and Fourier transform–infrared spectroscopy (FT- IR). The application of NMR as a promising analytical tool in metabolomics has been highlighted (Verpoorte et al., 2007) due in part to its potential to quantify and identify diverse groups of metabolites including amino acids, carbohydrates, lipids, phenolics and terpenoids (Choi et al., 2006; Kim et al., 2006; Moing et al., 2006). Although NMR is considered as less sensitive than other methods such as GC/LC-MS, it provides a direct profile of the system in a single spectrum, in which the metabolite information obtained can be further extracted by multivariate data analyses (Manetti et al., 2006). In the field
of plant biotechnology, NMR-based metabolomics has been utilized to identify unintended metabolic effects and to monitor the substantial equivalence in transgenic tomatoes (Gall et al., 2003; Noteborn et al., 2000). Also metabolic profiles of transgenic potatoes (Defernez et al., 2004), and tobacco (Choi et al., 2004) have been reported. The advancement of such analytical technologies for measurement of the metabolome generates massive data sets in which the identification of compounds remains a major challenge. The application of multivariate data analysis methods such as principal component analysis (PCA), partial least square regression (PLS) and partial least square- discriminant analysis (PLS-DA) is of great importance for data handling in such a way that complex data sets can be reduced to allow analysis and interpretation.
We have previously reported that metabolic engineering of salicylic acid (SA) biosynthesis with the two bacterial genes entC (encoding for isochorismate synthase (ICS)) and pmsB (encoding for isochorismate pyruvate lyase (IPL)) led to constitutive SA production in tobacco plants (CSA plants) when both the ICS and IPL were targeted to the chloroplast (Verberne et al., 2000). The profiling of these CSA plants by High Performance Liquid Chromatography (HPLC) showed suppression of several flavonoids, such as quercetin, kaempferol, rutin, and also chlorogenic acid, all products of the phenylpropanoid pathways (Nugroho et al., 2002). The metabolite profiles of Tobacco Mosaic Virus (TMV) infected and non-infected wild type and CSA plants by 1H-NMR spectroscopy showed a clear discrimination between the transgenic and wild type plants.
The major compounds contributing to the discrimination were chlorogenic acid, malic acid, glucose and sucrose (Choi et al., 2004).
Prior to the present study, we transformed a single ICS gene into B. rapa ssp. oleifera in which this gene product was also targeted at the chloroplast. The introduction of a gene utilizing chorismate as a substrate may alter directly or indirectly the metabolites present in B. rapa (Figure 6.1). This study aimed at the identification of metabolic pathways that are affected by the introduction of the ICS gene in this plant. 1H-NMR based metabolomics coupled with multivariate data analysis was applied to distinguish between greenhouse-grown control and ICS transgenic (T0) plants of B. rapa ssp. oleifera.
6.2 Materials and methods 6.2.1 Growth of plant materials
Nine and 18 months old primary transformants (T0 transgenic plant) (L2, L4, L5, L6, L9, L10 and L12) of B. rapa ssp. oleifera transformed with a gene encoding bacterial ICS and control plants, grown in a greenhouse at 22 °C with 16:8 h light/dark photoperiod, were used for this experiment. Old (the lowest green leaf) and young (upper three/four) leaves of transgenic and control plants were harvested at the same time and ground in liquid nitrogen prior to extraction. Control plants used for this experiments were tissue culture generated plants (CTRL) developed by using the same media and growth conditions as transformed plants. Also wild type plants (WT) were generated from seeds.
6.2.2 Extraction of plant materials
For each individual plant, the old and young leaves were ground in liquid nitrogen and pooled before subjected to freeze drying. Fifty milligrams of freeze-dried material were transferred to a microcentrifuge tube before adding 750 µl of methanol-d4 in D2O and 750
CHORISMATE shikimate
isochorismate
Vit K SA
p-hydroxybenzoic acid p-aminobenzoic acid
anthranilate
phenylalanine tyrosine
glucosinolates Krebs cycle glycolysis cinnamic acid derivatives
SA
lignin
flavonoids
phenylpropanoids
Figure 6.1 Metabolites originated from chorismate
µl of KH2PO4 buffer, pH 6.0, containing 0.1% trimethyl silyl propionic acid sodium salt (w/v). The mixture was vortexed for 2 minutes and sonicated for 15 minutes, followed by centrifugation at 13,000 rpm for 20 minutes at room temperature. Eight hundred microliters of the supernatant were then transferred into 5 mm NMR tubes for analysis.
6.2.3 Solvents and chemicals
D2O (99%) and methanol-d4 (99.8%) were obtained from Cambridge Isotope Laboratories Inc (Miami, FL, USA). NaOD was purchased from Cortec (Paris, France).
Potassium dihydrogen phosphate and trimethylsilane propionic acid sodium salt (TSP) were purchased from Merck (Darmstadt, Germany). KH2PO4 was added to D2O as a buffering agent. The pH of the D2O was adjusted to 6.0 using a 1 M-NaOD solution.
6.2.4 NMR measurement
1H-NMR, 2D-J-resolved spectra were recorded at 25 oC on a 500 MHz Bruker DMX-500 spectrometer (Bruker, Karlsruhe, Germany). 1H-1H-correlated spectroscopy (COSY), heteronuclear single quantum coherence (HSQC), and heteronuclear multiple bonds coherence (HMBC) spectra were recorded on a 600 MHz Bruker DMX-600 spectrometer (Bruker, Karlsruhe, Germany). All the NMR parameters were the same as those of our previous reports (Jahangir et al., 2008; Abdel-Farid et al., 2007).
6.2.5 Data analysis
Spectral intensities of 1H-NMR spectra were scaled to total intensity and reduced to integrated regions of equal width (0.04) corresponding to the region of δ 0.4- δ 10.0. The regions of δ 4.8-δ 4.9 and δ 3.28-δ 3.40 were excluded from the analysis because of the residual signal of water and MeOH. PCA and PLS-DA were performed with the SIMCA- P software (v. 11.0, Umetrics, Umeå, Sweden).
6.3 Results and discussion
6.3.1 1H-NMR spectra and identification of metabolites
Leaves of nine and eighteen months old primary transgenic and tissue-culture generated plants of B. rapa ssp. oleifera were subjected to metabolomic analysis by 1H-NMR and
diverse two dimensional spectroscopy. The assignments of 1H-NMR signals of amino/organic acids, carbohydrates and phenolic compounds were made based on our previous reports (Jahangir et al., 2008; Abdel-Farid et al., 2007; Hendrawati et al., 2006;
Liang et al., 2006a; 2006b; Widarto et al., 2006), our home-made database of ca. 500 metabolites and confirmed with the aid of two dimensional spectroscopy including J- resolved, 1H-1H COSY, HSQC and HMBC. Some of the characteristic chemical shifts and coupling constants of the compounds identified from 1H-NMR and J-resolved spectra are summarized in Table 6.1.
The aliphatic and sugar region (δ 0.8-5.0) shows signals of alanine (δ 1.48), threonine (δ 1.32), valine (δ 1.04), leucine (δ 0.96), isoleucine (δ 0.95), α-glucose (δ 5.20), β- glucose (δ 4.64), sucrose (δ 5.40) and the fructose moiety of sucrose (δ 4.17). In the organic acid region, malic acid (δ 4.32), glutamic acid (δ 2.08), fumaric acid (δ 6.56), formic acid (δ 8.46) and succinic acid (δ 2.52) are observed. Glucosinolates, compounds that are found almost exclusively in the Brassicaceae family were also detected. The presence of progoitrin (Figure 6.2a) was firstly confirmed based on the characteristic of the anomeric proton of glucose attached to sulfur (S-Glc) in the J-resolved spectrum, which has an upfield chemical shift (δ 5.04 ppm) and higher coupling constant (d, J=9.8 Hz) than the normal anomeric proton for α-and β-glucose (Jaki et al., 2002). This signal together with the signals of H-2α at δ 2.79 (dd, J=15.5, 8.0 Hz) and H-2β at δ 2.73 (dd, J=15.5, 5.0 Hz) of the glucosinolate correlates with the signal at δ 162.4 (C-1) in the HMBC spectrum. Also it was confirmed in the COSY spectrum, H-2, H-3, H-4 and H-5 of the glucosinolate correlate with each other (Figure 6.3). The indole glucosinolate neoglucobrassicin (Figure 6.2b) was also identified in the extracts. In the HMBC spectrum the signal of C-1 at 164.5 correlates with the anomeric proton of glucose (S- Glc) at δ 4.73 (d, J=10.0 Hz) and with the signal of H-2α at δ 4.27 (d, J=18.0 Hz) and H- 2β at δ 4.10 (d, J=16.0 Hz) of neoglucobrassicin. The signals of H2 (α and β) were confirmed by their correlation with signal at δ 32.4 in the HSQC spectrum. In the COSY
Table 6.1 The chemical shifts (δ in ppm) and coupling constants (J in Hz) of some metabolites detected on 1H-NMR and J-resolved spectra of the transgenic and control plants of Brassica rapa ssp. oleifera. s=singlet, d=doublet, dd=double of doublet, t=triplet, td=triple of doublet, m=multiplet.
Compounds Chemical shift (ppm) and coupling constant (Hz)
Alanine δ 1.48 (d, J=7.0, H-3), δ 3.76 (m, H-2)
Valine δ 1.04 (d, J=8.0, H-4), δ 1.06 (d, J=8.0, H-5), δ 2.39 ( m, H-3)
Threonine δ 1.32 (d, J=6.6, H-4), δ 4.33 (m, H-3)
Leucine δ 0.96 (d, J=6.8), δ 0.98 (d, J=6.8)
Isoleucine δ 0.95 (t, J=7.5 ), δ 1.02 (d, J=7.0), δ 3.65 (d, J=7.0)
β-glucose δ 4.64 (d, J=7.8, H-1)
α-glucose δ 5.20 (d, J=3.6, H-1))
Sucrose δ 5.40 (d, J=4.0, H-1)
Fructose moiety of sucrose δ 4.17 (d, J=8.0)
Succinic acid δ 2.52 (s)
Fumaric acid δ 6.56 (s)
Formic acid δ 8.46 (s)
Adenine δ 8.20 (s), 8.21(s)
Indole-3-acetic acid δ 3.24 (d, J=16.0), δ 3.39 (d, J=16.0), δ 7.10 (s), δ 7.13 (t, J=7.8), δ 7.21 (t, J= 7.8), δ 7.47 (d, J=7.8), δ 7.72 (d, J=7.8)
Choline δ 3.21 (s)
Glutamic acid δ 2.08 (m, H-3β), δ 2.14 (m, H-3α), δ 2.38 (td, J=8.0, 2.0), δ 3.79 (dd, J=7.0, 3.5) Malic acid δ 2.58 (dd, J=14.0, 6.5), δ 2.82 (dd, J=17.0, 3.5)
δ 4.32 (dd, J= 11.5, 4.0) Malic acid conjugated with trans
phenylpropanoid
δ 2.68 (dd, J= 17.0, 9.0), δ 2.85 (dd, J=17.0, 7.0), δ 5.22 (dd, J= 9.5, 3.0)
Progoitrin
δ 5.96 (m, H-4), δ 5.34 (td, J=17.0, 3.0, H-5α ), δ 5.21 (td, J=10.0, 3.0, H-5β ), δ 4.64 (m, H-3),
δ 2.79 (dd, J=15.5, 8.0, H-2α), δ 2.73 (dd, J=15.5, 5.0, H-2β)
Neoglucobrassicin
δ 4.09 (s, -OCH3), δ 4.10 (d, J=16.0, H-2β), δ 4.27 (d, J=18.0, H-2α ), δ 4.73 (d, J=10.0, H-1’),
δ 7.18 (t, J=8.0, H-5”), δ 7.31 (t, J=8.0, H-6”), δ 7.48 (s, H-2’), δ 7.49 (d, J= 8.0, H-7”), δ 7.72 (d, J= 8.0, H-4”)
trans-sinapoyl malate δ 6.48 (d, J=16.0, H-8’), δ 7.65 (d, J=16.0, H-7’) δ 7.0 (s, H-2, H-6’), δ 3.88 (s, OCH3)
trans-feruloyl malate δ 6.48 (d, J=16.0, H-8’), δ 7.66 (d, J=16.0, H-7’) δ 7.15 (d, J=2.0, H-2’), δ 7.07 (dd, J=8.5, 2.0, H-6’), δ 6.88 (d, J=8.5, H-5’)
trans- caffeoyl malate δ 6.41 (d, J=16.0, H-8’), δ 7.60 (d, J=16.0, H-7’) δ 6.98 (d, J=2.0, H-2’), δ 6.84 (d, J=8.0, H-5’), δ 7.07 (dd, J=8.0, 2.0, H-6’)
trans-coumaroyl malate δ 6.45 (d, J=16.0, H-8’), δ 7.66 (d, J=16.0, H-7’), δ 7.06 (d, J=8.0, H-3’), δ 7.02 (d, J=8.0, H-5’), trans-5-hydroxyferuloyl malate δ 6.41 (d, J=16.0, H-8’), δ 7.61 (d, J=16.0, H-7’),
δ 6.77 (s, H-2’), δ 6.73 (s, H-6’)
Kaempferol analogue δ 6.47 (d, J=2.2, H-6), δ 6.82 (d, J=2.2, H-8)
δ 7.91 (d, J=8.0, H-2’, H-6’), δ 7.02 (d, J=8.0, H-3’, H-5’)
O HOH2C
OH OH HO
S
N OSO3 OH
N
S O HOH2C
OH OH HO
N OSO3
OCH3
O
O OH OH
O O
R2 R1
HO
a
c b
2
3 4
5
1
H-5 H-3
H-4
H-5 H-3 H-4
H-4, H-3 (progoitrin) H-4, H-5
(progoitrin)
H-7’ H-8’
H-7’
H-8’
H-7’, H-8’(phenylpropanoids)
Figure 6.3 COSY spectrum (in the region of δ 4.0-8.0) of the transgenic plants of Brassica rapa ssp. oleifera, indicating the correlation of H-3, H-4 and H-5 of aliphatic glucosinolate, progoitrin and the correlation between H-7’ and H-8’ of phenylpropanoids
Figure 6.2 The chemical structure of the metabolites identified in Brassica rapa ssp. oleifera (a) progoitrin (b) neoglucobrassicin and (c) phenylpropanoyl malate R1=OCH3, R2=OCH3, sinapoyl malate, R1=H, R2=H, coumaroyl malate R1=OH, R2=H, caffeoyl malate, R1=OCH3, R2=H, feruloyl malate R1= OCH3, R2=OH, 5-hydroxyferuloyl malate
4’
2’
5’
6’
7’
8’
1 2
3 4
3’ 9’
2’
5” 1
7”
4’’ 3’
6’’
The presence of cis and trans-cinnamic acid derivatives conjugated with malic acid in a B. rapa variety has already been document by previous researchers (Liang et al., 2006b).
In the variety used in this study, these compounds (Figure 6.2c) were clearly detected.
The signal of H-2 of malic acid at 5.18 (dd, J=9.5, 3.0) correlates with its H-3α at δ 2.83 (dd, J=17.0, 7.0) and H-3β at 2.71 (dd, J=17.0, 7.0) (cis configuration) in the COSY spectrum whereas the signal of trans configuration at 5.22 (dd, J=9.5, 3.0) correlates with signals at δ 2.68 (dd, J= 17.0, 9.0) and δ 2.85 (dd, J=17.0, 7.0). The most characteristic region for phenylpropanoids in the 1H-NMR spectrum is in the range of δ 6.0–δ 6.5 for H-8’ (Sanchez-Sampedro et al., 2007). In this region five major doublets of H-8’ of the trans olefinic proton of phenylpropanoids was confirmed with a coupling constant of 16.0 Hz, correlated with the H-7’ proton in the region of δ 7.5- δ 7.7 in the COSY spectrum (Figure 6.3). The signals of H-8’ and H-7’ were correlated with a signal of C-9 at δ 171.8 in the HMBC spectrum. A thorough investigation of the 1H-NMR and 2D- NMR showed that the trans-configuration of sinapoyl malate, feruloyl malate, coumaroyl malate, caffeoyl malate and 5-hydroxyferuloyl malate are present in the samples. Five doublets with a coupling constant of 12.0 Hz in the region of δ 5.91-5.98 in the J- resolved spectrum were believed to be the signals of H-8’ of cis olefinic proton, however the cis forms might be the artifacts of trans forms (Pauli et al., 1998). Figure 6.4 shows the J-resolved spectra of some of the compounds identified in the transgenic plants.
Minor signals of a flavonoid moiety are also present in the extract. Signals at δ 6.47 (d, J=2.2, H-6), δ 6.82 (d, J=2.2, H-8), δ 7.91 (d, J=8.0, H-2’, H-6’) and δ 7.02 (d, J=8.0, H-3’, H-5’) were elucidated as kaempferol analogues (Hendrawati et al., 2006). In the COSY spectrum, the signals of H-3’, H-5’ correlate with the signals of H-2’, H-6’.
6.3.2 PCA and PLS-DA of the 1H NMR spectra pinpoint metabolite differences between the transgenic and the control plants
Multivariate data analysis was performed on the raw dataset obtained from 1H-NMR spectra. In general 1H-NMR spectra allow producing great amounts of variables for given samples (>200 signals). The first requirement is the data reduction by mathematical and statistical procedures in order to draw a conclusion e.g. grouping of samples or pattern
Figure 6.4 Two dimensional 1H-1H J-resolved spectra of transgenic plants of Brassica rapa ssp. oleifera. (a) in the range of δ 6.3-8.0, 1. H-2’, H-6’ of kaempferol 2. H-4” of neoglucobrassicin 3. H-4 of indole acetic acid (IAA) 4. H-7’ of trans-phenylpropanoids, 5. H-7” of neoglucobrassicin 6. H-7 of IAA 7. H-6” of neoglucobrassicin 8. H-6 of IAA 9. H-5” of neoglucobrassicin 10. H-5 of IAA 11. H-5’ of feruloyl malate, 12. H-5’ of caffeoyl malate 13. H-8’ of trans-phenylpropanoids. (b) in the range of δ 5.0-6.2 14. H-4 of progoitrin, 15. sucrose 16. H5α- of progoitrin 17. H5β-of progoitrin 18. α-glucose 19.
Anomeric proton of progoitrin.
4
a
b
16
18
14 15 17 19
1 2,3 4 5,6 7 8 9 10 11, 12 13
recognition. The most common unsupervised multivariate data analysis is principal component analysis (PCA). As the first step of multivariate data analysis, PCA of 1H- NMR spectra was initially performed to discriminate the samples. The PCA score plot showed that PC1 and PC2 explained 34% and 20% of the variation, respectively. The score plot of PCA (Figure 6.5a) could only clearly reveal the separation between the age of the plants (9 and 18 months old) but the transgenic and control plants showed only partial separation on PC1 and PC2. PCA is basically obtained from the maximum variation between samples. When the variation between interesting groups is smaller than that between individual samples, the difference cannot be deduced. In the dataset in this study, the difference between the control and transgenic plants are smaller than between the developmental stages (9 and 18 months old). Based on all the 1H-NMR signals, the plants of the same ages are closely clustered but the metabolic resemblance within the controls or transgenic plants are quite far from each other. For solving the problems a supervised multivariate data analysis, PLS-DA was applied.
A sort of PLS, PLS-DA uses discrete class matrix (0 and 1). In contrast to PCA which only uses the information of one matrix, PLS-DA also takes into account the information in another matrix (Berrueta et al., 2007). When PLS-DA was applied, the separation between the control and transgenic plants considerably improved. All the transgenic and control plants could be clearly seen as separate clusters in the PLS-DA score plot (Figure 6.5b). Most of the transgenic plants are located on the negative side of the PLS component 1 axis while all the controls are on the positive side. The 18 months old plants are fully separated from the 9 months old plant by PLS component 2 (t[2]). The model diagnostics for the first component showed an explained variation (R2Y) of 0.67 and the goodness of fits (Q2Y) of 0.28.
To find out precisely which metabolites contributed to the discrimination between the transgenic and control plants, a PLS-DA loading plot c(PLS weight w*c) was made as shown in Figure 6.5c. Variables located closely to the “dummy” variables Y (transgenic and control) contribute strongly to the class separation. Positive values of wc*[1] are seen for the variables associated with the control plants whereas negative values are associated with transgenic plants. Investigation of this loading plot suggested that the contributing metabolites that strongly influence the separation in the control plants were mostly
-20 -10 0 10 20
-20 -10 0 10 20
-10 0 10
-20 -10 0 10 20
Figure 6.5 Score plot of PCA (a) and PLS-DA (b) of the transgenic and control plants of Brassica rapa ssp. oleifera (∆: Transgenic plants (9 months old), ▲: Control plants (9 months old), □ Transgenic plants (18 months old), ■ Control plants (18 months old).
Loading plot of PLS-DA (c). Variables (▲) located closely to the “dummy” variable Y (■
control, ● transgenic) contribute strongly to the separation between the control and transgenic plants.
PLS component 2 (18%)
PLS component 1 (26%) PC1 (34%)
PC2 (20%) (20%) 9 months
18 months
a
b
-0.1 0.0 0.1
-0.12 -0.10 -0.08 -0.06 -0.04 -0.02 -0.00 0.02 0.06 0.08 0.10 0.12 0.14
w*c[2]
malic acid
alanine threonine, leucine
glutamic acid
neoglucobrassicin
feruloyl malate
caffeoyl malate
succinic acid α/β glucose
adenine
●
■
wc*[2]
c
IAA valine
fumaric acid
sinapoyl malate
wc*[1]
transgenic
control
amino acids and organic acids alanine (δ 1.48), valine (δ 1.04), leucine (δ 0.96), threonine (δ 1.32), glutamic acid (δ 2.08), malic acid (δ 4.32) and adenine (δ 8.21). In the transgenic plants, the responsible loadings corresponded to resonances of sugars (α- glucose at δ 5.20 and ß-glucose at δ 4.64), succinic acid (δ 2.52), indole acetic acid (IAA, δ 3.24) and fumaric acid (δ 6.56). Resonances corresponding to secondary metabolites including the glucosinolate neoglucobrassicin at δ 7.72 and δ 7.48, and the phenylpropanoids i.e. sinapoyl malate at δ 3.88 and δ 7.0, feruloyl malate at δ 6.88 and δ 6.48, caffeoyl malate at δ 6.84 and δ 7.60 were also detected in the left quadrant of the PLS-DA loading plot where all the samples of transgenic plants were located. These results indicate that the transgenic plants have higher levels of these compounds in comparison to control plants which have higher levels of amino acids.
The increase of neoglucobrassicin in this analysis was in accordance with our HPLC results which showed that not only neoglucobrassicin was increased in the transgenic plants but also other indole glucosinolates such as glucobrassicin and 4- methoxyglucobrassicin (Chapter 5). Increased accumulation of indole glucosinolates was observed previously upon herbivory and fungal infection of Chinese cabbage (Rostas et al., 2002) and in methyl jasmonate-treated B. rapa (Liang et al., 2006). The increase of glucose in the transgenic plants was expected as this is in parallel with the increase of glucosinolate neoglucobrassicin. Glucose is not only a critical nutrient source for plants but it is also believed to play a role in environmental responses, since a large number of stress responsive genes were induced by this compound (Price et al., 2004).
The increase of certain phenylpropanoid compounds identified in the transgenic plants suggests that the presence of the bacterial ICS, utilizing chorismate as a substrate, has an influence on other competitive pathways at the chorismate branching point. The involvement of the phenylpropanoid pathway in the biosynthesis of secondary metabolites important for plant development and the plant’s interaction with the environment are well known (Rasmussen and Dixon, 1999). For example, many phenylpropanoids exhibit a broad spectrum of antimicrobial activity, which is believed to be involved in the plant defense response (Dixon, 2001). Up and down regulation of these compounds is often observed in plants after infection or stress. Tobacco plants overexpressing SA as a result of insertion of both ICS and IPL showed suppression of
quercetin, kaempferol, rutin and chlorogenic acid, all products of the phenylpropanoid pathway (Nugroho et al., 2002). This suggests that the suppression of these compounds might be due to the channeling of chorismate into isochorismate and away from the phenylpropanoid pathway.
Prior to the present study, HPLC analysis on the same transgenic plants showed that SA and its glucoside (SAG) were increased significantly in comparison to non-transgenic control plants (Chapter 4). The increase of phenylpropanoids together with the increase of SA as observed in these transgenic plants suggested that the accumulation of SA in B.
rapa does not compete with the phenylpropanoid pathway. A similar phenomenon occurs in the indole pathway. Phylloquinone (Chapter 5) was expected to be influenced by the insertion of the single ICS gene as shown by the results with tobacco CSA plants (Verberne et al., 2000), but in B. rapa the level of this compound was not altered.
Apparently in B. rapa, the channeling of chorismate into SA seems not to directly effect the fluxes through the chorismate branches, but SA as a signal compound may indirectly induce some of these pathways leading to increased levels of neoglucobrassicin and phenylpropanoids (Figure 6.6).
PCA of only the data from transgenic plants was applied to identify the contributing metabolites responsible for the separation between 9 and 18 months old plants. A score scatter plot of PC1 over PC3 (Figure 6.7a) showed that all 9 months old plants are clearly separated from the 18 months old plants. A loading plot of PCA along the PC1 was investigated to find the contributing metabolites responsible for the separation (Figure 6.7b). Nine months old transgenic plants appeared to have more indole acetic acid (IAA), neoglucobrassicin, caffeoyl malate, sucrose, progoitrin, succinic acid and valine as indicated by their location on the positive side of the loading plot. In contrast, 18 months old plants have higher amounts of adenine, flavonoid moiety, feruloyl malate, sinapoyl malate, malate/conjugated malate, glutamic acid, alanine and leucine (negative side of the PC1 loading). The results suggest that induction of certain metabolites varies with the age of the plants. Some higher oxidized phenylpropanoids accumulate in the later stage.
Figure 6.6 The effect of the introduction of the bacterial isochorismate synthase gene in Brassica rapa ssp. oleifera on the metabolites originated from chorismate. (+); induced, (+/-; no change).
AS; anthranilate synthase, CM; chorismate mutase, ICS; isochorismate synthase
-10 0 10
-20 -10 0 10 20
-0.10 -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08 0.10
8.8 8.64 8.44 8.24 8.04 7.84 7.64 7.44 7.24 7.04 6.84 6.64 6.44 6.24 6.04 5.84 5.64 5.44 5.24 5.04 4.6 4.4 4.2 4 3.8 3.6 3.4 3.16 2.96 2.76 2.56 2.36 2.16 1.96 1.76 1.56 1.36 1.16 0.96 0.76
Figure 6.7 (a) Score scatter plot (PC1/PC3) of the transgenic plants of Brassica rapa ssp. oleifera (b) Loading column plot showing the metabolites that most strongly influence the separation in the positive and negative PC1 which corresponded to the nine months and eighteen months old transgenic plants respectively. 1. adenine 2. kaempferol 3. indole acetic acid 4. neoglucobrassicin 5. caffeoyl malate 6. sinapoyl malate 7. progoitrin 8. sucrose 9. α-glucose 10. β-glucose 11.
malate/malate conjugated 12. succinic acid 13. glutamic acid 14. alanine 15. valine 16. leucine CM
PC1 (37%) PC3 (17%)PC 1
Tryptophan GLUCOSINOLATE (+)
Phenylalanine PHENYLPROPANOIDS (+)
SALICYLIC ACID (+) AS
ICS
PHYLLOQUINONE (+/-) CHORISMATE
9 months 18 months
a
1 2
3 4 5
6 7
8
9
10
11 12
13 14 15
16
b
This study shows that NMR-based metabolomics coupled with multivariate data analysis is a useful tool for a macroscopic approach, allowing us to identify a broad range of primary and secondary metabolites in crude samples without the need for further purification. This non targeted approach is especially useful as an easily applicable method to quickly investigate effects of a transgene on a large number of metabolites or to detect potential unintended effects in genetically modified crops.
6.4 Conclusion
The results obtained in this study show that the expression of the bacterial isochorismate synthase (ICS) gene in B. rapa ssp. oleifera seems not to affect fluxes into pathways to other groups of secondary metabolites through competition for the same precursor. On the contrary, the biosynthesis of isochorismate derived products (SA) seems to induce the competitive pathways via phenylalanine (phenylpropanoids) and tryptophan (IAA and indole glucosinolates).
6.5 Acknowledgements
We thank Malaysian Agricultural Research and Development Institute (MARDI), Malaysia for the Ph.D grant to Sanimah Simoh.
