16. A strong link between an unknown metabolite trait (m/z
256.0810, retention time ¼ 1.05) and the candidate gene
GC1 (guanylyl cyclase 1 [35], AT5G05930) was supported
by both GWAS (Fig. 2a) and network analysis in 21-D and
32-D conditions (Fig. 2b). In GC1, two SNPs (m164251,
G/T, lead SNP; m164253, C/A) result in an altered protein
amino acid sequence (Fig. 2c). Together with the results from
linkage disequilibrium (LD) (Fig. 2d) and haplotype (Fig. 2e)
analyses, this finding suggests that these polymorphic variants
are likely to constitute the functional variation underlying this
association. Isotope-labeling results suggested the possible
chemical formula C 11 H 13 NO 6 (Fig. 2f). Assaying the standard
compound nicotinate D-ribonucleoside revealed that it shares
the same MS/MS fragmentation pattern as our metabolite
(Fig. 2g), albeit with a slight retention-time shift, suggesting
that our metabolite is structurally highly similar to nicotinate
D-ribonucleoside.
17. A significant association (LOD ¼ 5.46), only appearing in the
control-condition GWAS, was detected between gamma-Lglutamyl-L-cysteine and a locus on chromosome 5 (Fig. 3a)
harboring two candidate genes: CYDS2 (cysteine synthase D2,
AT5G28020)
and
DES1
(L-cysteine
desulfhydrase
1, AT5G28030). Both correlations between the metabolite
feature and the two genes were supported by network analysis
in darkness-related conditions (Fig. 3b). We confirmed the
metabolites as gamma-L-glutamyl-L-cysteine using isotope
labeling (Fig. 3c) and MS/MS fragmentation analysis with a
standard (Fig. 3d). DES1 was reported to catalyze the desulfuration of cysteine to sulfide [36], whereas the function of
CYSD2 remains elusive to date. Our results provide genetic
evidence that the associations between gamma-L-glutamyl-Lcysteine and CYDS2 and DES1 are involved in cysteine metabolism (Fig. 3e).
18. The integration of genetic mapping and network analysis at this
stage still mainly focused on limited classes of metabolites in
Arabidopsis [34, 46, 47] and crop species [19, 48, 49]. It
indicates that this integrative strategy coupled with large-scale
untargeted metabolomics remained yet to be widely and deeply
exploited, which can provide more global insights into the
metabolic landscape of plants. In addition, plants interact
with their surrounding environments from time to time by
their numerous metabolites. Given the fact that many secondary metabolites are “silent” under control conditions, imposition of stresses may assist in the elucidation of novel
biochemical pathways in plants. Therefore, the investigations
of GWAS and network analysis with environmental stressdriven perturbations of metabolic homeostasis would largely
408
Si Wu et al.
256.0810, retention time ¼ 1.05) and the candidate gene
GC1 (guanylyl cyclase 1 [35], AT5G05930) was supported
by both GWAS (Fig. 2a) and network analysis in 21-D and
32-D conditions (Fig. 2b). In GC1, two SNPs (m164251,
G/T, lead SNP; m164253, C/A) result in an altered protein
amino acid sequence (Fig. 2c). Together with the results from
linkage disequilibrium (LD) (Fig. 2d) and haplotype (Fig. 2e)
analyses, this finding suggests that these polymorphic variants
are likely to constitute the functional variation underlying this
association. Isotope-labeling results suggested the possible
chemical formula C 11 H 13 NO 6 (Fig. 2f). Assaying the standard
compound nicotinate D-ribonucleoside revealed that it shares
the same MS/MS fragmentation pattern as our metabolite
(Fig. 2g), albeit with a slight retention-time shift, suggesting
that our metabolite is structurally highly similar to nicotinate
D-ribonucleoside.
17. A significant association (LOD ¼ 5.46), only appearing in the
control-condition GWAS, was detected between gamma-Lglutamyl-L-cysteine and a locus on chromosome 5 (Fig. 3a)
harboring two candidate genes: CYDS2 (cysteine synthase D2,
AT5G28020)
and
DES1
(L-cysteine
desulfhydrase
1, AT5G28030). Both correlations between the metabolite
feature and the two genes were supported by network analysis
in darkness-related conditions (Fig. 3b). We confirmed the
metabolites as gamma-L-glutamyl-L-cysteine using isotope
labeling (Fig. 3c) and MS/MS fragmentation analysis with a
standard (Fig. 3d). DES1 was reported to catalyze the desulfuration of cysteine to sulfide [36], whereas the function of
CYSD2 remains elusive to date. Our results provide genetic
evidence that the associations between gamma-L-glutamyl-Lcysteine and CYDS2 and DES1 are involved in cysteine metabolism (Fig. 3e).
18. The integration of genetic mapping and network analysis at this
stage still mainly focused on limited classes of metabolites in
Arabidopsis [34, 46, 47] and crop species [19, 48, 49]. It
indicates that this integrative strategy coupled with large-scale
untargeted metabolomics remained yet to be widely and deeply
exploited, which can provide more global insights into the
metabolic landscape of plants. In addition, plants interact
with their surrounding environments from time to time by
their numerous metabolites. Given the fact that many secondary metabolites are “silent” under control conditions, imposition of stresses may assist in the elucidation of novel
biochemical pathways in plants. Therefore, the investigations
of GWAS and network analysis with environmental stressdriven perturbations of metabolic homeostasis would largely
408
Si Wu et al.