Gynandropsis gynandra (L.) Briq.

D, E. O., e, Sogbohossou, edi, Achigan-Dako, Enoch G., Mumm, Roland, de Vos, Ric C. H. & Schranz, M. Eric, 2020, Natural variation in specialised metabolites production in the leafy vegetable spider plant (Gynandropsis gynandra L. (Briq. )) in Africa and Asia, Phytochemistry (112468) 178, pp. 1-10 : 3-5

2.2. Natural variation in volatile metabolites in Gynandropsis gynandra

A total of 130 volatile metabolites were detected in the leaves of our accessions (Supplementary Table 3). Accessions TOT6440 and ODS-15- 117 were detected as outliers and removed from further analyses. A principal component analysis (PCA) on the remaining 46 accessions and based on the 130 metabolites revealed that there was no clear separation of the accessions along the first two PCs according to their geographic origin ( Fig. 6 ; Supplementary Table 3). A heatmap based on volatile metabolite levels revealed three clusters of accessions (C1, C2 and C3) and two main clusters of metabolites (D1 and D2) ( Fig. 7 ). Cluster C1 consisted of 4 accessions (2 from East Africa and 2 from Asia) with overall low levels of volatiles. Cluster C2 included 30 accessions (19 from East Africa, 7 from Asia and 4 from West Africa) with on average high levels of D1 compounds but generally low levels of D2 compounds. Cluster C3 was made up of 12 accessions (11 from Asia and one from East Africa) that had moderate to high levels of volatile compounds. Of the 130 volatiles detected, 54 were putatively identified. Volatile metabolites present in cluster D1 mainly include aldehydes (e.g. GC987: (E)-2-pentenal; GC1482: (E)-2-hexenal; GC2713: (E,E)-2,4,-heptadienal; GC4069: β- cyclocitral), ketones (e.g. GC3119: 3,5-octadien-2-one; GC2492: 6-methyl-5-hepten-2-one), monoterpenes (e.g. GC2858: eucalyptol) and alcohols (e.g. GC1074: (Z)-2-penten-1-ol). The metabolite cluster D2 mainly included esters (e.g. GC2148: propyl 2-methylbutanoate; GC2667: isobutyl isovalerate; GC3365: 2-methylbutyl 2- methylbutanoate), sesquiterpenes (e.g. GC4862: α- humulene; GC4956: bicyclosesquiphellandrene; GC5093: (E,E)- α- Farnesene; GC5277: cis- Calamenene), and sulphur compounds (e.g. GC798: methylthiocyanate; GC843: methyl isothiocyanate; GC1447: isopropyl isothiocyanate; GC1784: 2-ethyl-thiophene). The profiles of volatile metabolites did not strongly correlate with the geographic origin of the accessions. However, volatile metabolites abundant in cluster C2 were described to have pungent and spicy sensory attributes ( The Good Scents Company, 2019). Some of these compounds included 1-Penten-3-ol (GC600; pungent, horseradish-like), (E) -2-pentenal, (GC987; pungent, green, fruity apple-like) (E,E)-2,4-hexadienal (GC 2033; pungent fatty green), (Z) -2-penten-1-ol, (GC1074; mustard horseradish), (E) -2-hexenal, (GC1576; sharp, penetrating fresh leafy green, spicy). Compounds abundant in cluster C3 including isobutyl isovalerate (GC2667), butanoic acid, 2-methyl-, 3-methylbutyl ester (GC3301), 2-methylbutyl 2-methylbutanoate (GC3365), isobutyl isobutyrate (GC 1858); 3-methylbutyl 2-methylpropanoate (GC2685), propanoic acid, 2-methyl-, 2-methylpropyl 2-mehtyl propanoate (GC 2011) have been described to have a sweet fruity flavour. Other abundant compounds with specific flavour and taste in the C3 cluster included cubenol (GC5447; spicy), linalool (GC3327; floral, citrus scent), (E)-beta-ocimene (GC2893; green, tropical, woody), beta-caryophyllene (GC4609; spicy, woody) ( The Good Scents Company, 2019).

Supervised sPLS-DA with the 130 volatiles revealed that the error rate obtained by the prediction models were stabilized after the first five dimensions (Supplementary Table 3). The maximum distance method showed the lower error rate for the five dimensions compared with the centroids and Mahalanobis distances (Supplementary Table 3). The first dimension explained 15% of variation and discriminated Asian accessions from African ones. The second dimension explained 13% of variation and partially discriminated East/Southern African accessions from Asian and West African ones ( Fig. 8a ). The model selected ten discriminative volatiles, all positively correlated with dimension 2 and thus in relatively low levels in West African accessions. GC1074 (( Z)-2-penten- 1-ol), GC 1944 (heptanal), GC2164 (unknown), GC2586 (unknown), GC2651 (unknown) and GC3426 (nonanal) were positively correlated with both dimensions and characterised East African accessions. GC515 (unknown), GC 1875 (isopropyl isovalerate), GC2858 (eucalyptol) and GC3900 (unknown) were negatively correlated with dimension 1 and characterised Asian accessions.

From 31 volatiles previously reported in G. gynandra ( Nyalala et al.,

2013), 16 were identified in our study and included mainly isothiocyanates, terpenes and aldehydes. Nyalala et al. (2013) highlighted the inactivity of spider mites exposed to 2,4-heptadienal or β -cyclocitral, (Z)-2-pentenol, or methyl isothiocyanate, all compounds that were detected in our study.

2.3. Natural variation in glucosinolates and other plant defence related compounds in G. gynandra

Glucosinolates are known to be involved in plant defence against herbivores. Upon disruption of the leaf tissues, the hydrolysis of glucosinolates by myrosinases releases volatile sulphur compounds, mainly isothiocyanates nitriles that have repellent properties against pests ( Beekwilder et al., 2008). In our collection, one glucosinolate putatively identified was glucocapparin, also known as methylglucosinolate (LC880). A closer look at the metabolite profiles of the accessions for both non-volatile and volatile glucosinolate-related compounds ( Fig. 9 ) revealed three clusters of accessions. Cluster E1 was comprised of nine East African and one West African accessions which had relatively low levels of glucocapparin and isothiocyanates. Accessions in cluster E2 had overall high levels of glucocapparin and isothiocyanates. Slight correlations were found between glucocapparin and both isopropylisothiocyanate (r 2 =0.27, ns) and methylisothiocyanate (r 2 = 0.23, ns). A weak correlation between glucocapparin and methylthiocyanate was detected (r 2 = 0.11, ns). In this study we determined these glucosinolates-related compounds as they are in the intact plant cells, by specifically preventing myrosinases activity. Thus, leaves were flash-frozen and ground in liquid nitrogen, and glucosinolates were subsequently extracted in methanol, which denatures proteins, and the thiocyanates were trapped while inhibiting enzyme activities with saturated CaCl 2. We therefore expected a low correlation between glucosinolates and isothiocyanates levels. The natural variation in glucosinolates in G. gynandra provides a basis for further investigation of the potential of these compounds with herbivore interactions. In our study we identified an aliphatic glucosinolate (glucocapparin). However, Omondi et al. (2017) identified 3-hydroxypropyl glucosinolate as the main glucosinolate present in different plant parts of 30 accessions of G. gynandra collected in various East African countries while Neugart et al. (2017) observed mainly glucocapparin and only traces of indole glucosinolates (glucobrassicin and 4-methoxyglucobrassicin) in G. gynandra leaves. Glucosinolate levels were strongly influenced by plant developmental stages in Aethionema arabicum and tended to decrease after flowering ( Mohammadin et al., 2017). The discrepancies between our results and previous investigations of glucosinolates in G. gynandra may therefore be explained not only by differences in the specific accessions used but also by differences in growth conditions and plant developmental stages.

Figures

Fig. 6. Principal component analysis score plot of relative levels of 130 volatile metabolites detected in the leaves of 46 accessions of Gynandropsis gynandra from Asia (red), East/Southern Africa (black) and West Africa (blue). The first two dimensions explaining 52.9% of the total variation are shown. 95% confidence ellipses are presented for the three regions. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 7. Heatmap of the 130 volatile metabolites detected in the leaves of 46 accessions of Gynandropsis gynandra from Asia (red), East/Southern Africa (black) and West Africa (blue). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 8. Sparse partial least square discriminant analysis on the 48 accessions of Gynandropsis gynandra based on 130 volatile metabolites: (a) Score plot showing the projection of the 48 accessions Asia (red), East/Southern Africa (black) and West Africa (blue) on the two dimensions; (b) Selected variables representation on two dimensions on the correlation circles (0.5 and 1 correlation values). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 9. Relative levels of glucosinolates and isothiocyanates in 43 accessions of Gynandropsis gynandra from Asia (red), East/Southern Africa (black) and West Africa (blue). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)