PURPOSE

There is a gradual shift in focus in clinical medicine from reductionism to systems approach and monodrug to combination drugs for treatment¹. This change in viewpoint needs to be applied to plants as well since they play a vital role not only in diet and nutrition but also as therapeutic remedies. Many of the drugs are either plant-based or have taken lead from traditional systems of medicine, which uses plants extensively². However, objectively measurable system properties are required to adopt a systems approach to study of plants. In this context, organoleptic property of taste is a suitable parameter for evaluating plants as a whole. Although there are a number of contributing factors affecting sensorial parameters such as taste, the latter will be a suitable system parameter since it reflects both the chemical and functional aspects of a plant. The purpose of this study is to use NMR coupled with Etongue and multivariate analysis for chemosensory signature of medicinal plants.

METHOD

Plant materials: The classification and sub-classification of medicinal plants under different single taste categories have been taken from ethnobotanical information³. Seventy four authenticated plant samples classified under five different taste groups (sweet, sour, bitter, pungent and astringent) and subgroups within major categories were studied. Some of the plants studied were: Cucumis sativus, Asparagus racemosus, Agaricus campestris, Phaseolus aerous, Vitis vinifera, Phoenix sylvestris (sweet); Mentha piperata, Cinnamomum tamala, Trigonella foeneum, Cuminum cyminum, Piper longum, Zingiber officinale (pungent); Terminalia arjuna, Ficus bengalensis, Bauhinia purpurea, Woodfordia floribunda (astringent); Swertia chirata, Momordica charantia, Nyctanthis arbortristis, Picrorhiza kurrao (bitter); Tamarindus indica, Garcinia indica, Thespesia populnea (sour). The samples were prepared as aqueous solution using the method of cold infusion.

NMR: Water suppressed 1D proton NMR spectra were recorded at 700 MHz (Agilent, USA) using the following parameters: relaxation delay - 14 sec, number of scans - 32, spectral width - 15ppm and data point - 32 K. Deuterated TSP in a coaxial insert was used as an external reference. Peak assignments were carried out using 2D data and NMRshiftDB data library.

E-tongue: Taste was assessed objectively with a 16 autosampler E-Tongue (Alpha MOS, France) using sensors which work on the principle of chemical modified field effect transistor⁴.

Data analysis: Principle Component Analysis (PCA) was performed on both NMR and Etongue data. The NMR data were binned and bucketed at intervals of 0.04 ppm and then subjected to different scaling methods for the cluster analysis. Spectral data post-processing, binning and bucketing were done using MestReC. Multivariate analysis was carried out with Unscrambler X10 and metaboanalyst 3.0 (NMR data) and alphasoft 4.0 (Etongue data). The inhouse built data library of taste standards was used as reference for analysis of the Etongue data.

RESULTS

Figure 1 shows the proton NMR spectra of plants from four different taste categories. Primary metabolites such as carbohydrates (α- and β-glucose), pyruvate, amino acids (tyrosine, tryptophan, alanine, leucine, isoleucine, valine, methionine, proline, pyruvate, threonine and phenylalanine) and β-OH butyrate were seen in the aliphatic region of all the spectra, although with varying intensities. Aromatic region (5.2-10 ppm), which generally predominates in secondary metabolites, showed differences between the groups. For example, polyphenols (sweet category), high presence of flavonoids, flavonol glycosides and also polyphenols (pungent), phenylalanine (bitter) were observed.

The distinct differences observed in the aromatic region of the spectra suggest possibility of NMR based fingerprinting for taste phenotypes of medicinal plants. Some representative data are shown in Figures 2 and 3. PCA plots of the spectral data from four different major taste categories are shown in Figure 2. Statistically significant distinct clusters were observed between sweet and astringent (Fig. 2a), pungent and bitter (Fig. 2b), bitter and astringent (Fig. 2c). However, there were also some outliers and overlaps. Spectral data from sub-groups (within sweet category) when used for the PCA analysis, resulted in markedly better discrimination (Figs. 3a-3c). Clear differentiation were observed between the sweet sub-groups and the major groups of bitter (Fig. 3a), sour (Fig. 3b) and pungent (Fig. 3c) categories of plants. PCA analysis of the Etongue data also showed similar clustering pattern indicating that NMR has the potential to provide chemosensory signature.

CONCLUSION

This study has introduced a conceptually new approach of using NMR to study sensorial properties of plants. While the spectral data has shown statistically significant chemosensory signatures, individual phytomarkers were also identified for the different taste groups. Further indepth studies are underway to explore this novel application of NMR in chemosensory science and system evaluation of medicinal plants.