Desmodium incanum, DC.

Hao, B., Caulfield, J. C., Hamilton, M. L., Pickett, J. A., Midega, C. A. O., Khan, Z. R., Wang, J. & Hooper, A. M., 2016, Biosynthesis of natural and novel C-glycosylflavones utilising recombinant Oryza sativa C-glycosyltransferase (OsCGT) and Desmodium incanum root proteins, Phytochemistry 125, pp. 73-87 : 77

2.6. Synthesis of 6,8-di-C-glycosylflavones using 6-C-glucosyl-2- hydroxyflavanone substrates created by OsCGT and CGT enzymes from D. incanum

The results of incubating synthetic 2-hydroxyflavanones with UDP-glucose and OsCGT ( Tables 1 and 2 ) revealed substrate that can be used to generate C -glucosyl-2-hydroxyflavanones (4) that may be themselves substrates for a second glycosylations. In our work to understand the biosynthesis of di- C -glycosylflavones in the allelopathic root exudates of D. incanum we have demonstrated that the soluble plant enzyme extracts can C -glucosylate, C -galactosylate and C -arabinoslyate the mono- C -glucosylflavanones 4a to produce di- C -glycosylflavonones that dehydrate to the respective flavones ( Hamilton et al., 2012; Hao et al., 2015). We therefore used synthetic 2-hydroxyflavanones that are well accepted by the OsCGT to generate natural and novel C -glucosyl-2-hydroxyflavanones ( Table 3 ). By using recombinant OsCGT, these unstable intermediates can be generated in good yields and purity without the need for separation from other flavonoids. The incubation of the substrates with UDP-glucose is expected to give three products (7–9), rather than one containing two glucose groups, as the D. incanum soluble enzyme extract is able to isomerise UDP-glucose to UDP-galactose which is subsequently incorporated and after dehydration can be fixed in either the C- 6 or C- 8 position to give two regiosiomers ( Scheme 1 ) ( Hao et al., 2015). The incorporation of UDP-arabinose likewise produces two products (10, 11), regioisomers with the C -arabinose in either the 6 or 8 position. In all cases, where the correct number of novel products was detected with the correct molecular weight, the fragmentation pattern was subsequently carried out under high cone voltage ESIMS conditions to demonstrate that the products are indeed di- C -glycosylflavones. In the case of C -hexosylflavones, loss of 90 and 120 amus indicate the correct C -link and with C -arabinosylflavones, loss of 60 and 90 amus is indicative of the correct structure ( Fig. 1 ). The results of these experiment showed that there was a greater relaxed specificity for C -glucosylation and C -galactosylation than for C -arabinosylation. Substrates 4a, b, d, f, g, h, k,o,p,s all produced 3 new metabolites when incubated with UDP-glucose and D. incanum root protein, deduced to be structures 7, 8 and 9 by ESIMS mass spectrometry (Supplementary data). In the case of 4b, as the products are already present in D. incanum and are co-extracted with the proteins, the labelled sugar moiety UDP-α- D-[UL- 13 C 6]glucose was used and incorporation judged by presence of the labelled [ M — H +6] — 615 ion with six 13 C atoms. The ion trace, low cone voltage ESIMS spectrum and fragmented high cone voltage ESIMS spectra from compounds 7h, 8h, and 9h produced from 4h are shown ( Fig. 3 ). The selective ion trace at 621 amus ( Fig. 3A ) shows three new metabolites with the correct ion ([ M — H] — 621) which was shown to be the anticipated molecular ion ( Fig. 3C ). Fragmentation shows sequential losses of 90 ([ M — H — 90] — 531 and [ M — H — 180] — 441} and 120 amus ([ M — H — 120] — 501 and [ M — H — 240] — 381) and combinations of losses ([ M — H — 90-120] — 411 ( Fig. 3D ) indicating both sugars must be C -linked to the apigenin core ( Fig. 1 ). This has previously been shown for 4a where the three new metabolites produced were characterised through the NMR characterisation of natural standards as the di- C -glucosylated compound and two C -galactosylated regioisomers 7a, 8a and 9a ( Hooper et al., 2015). The substrates that were not incorporated were 4e and 4i which both possess two oxygen atoms in meta positions. The more sterically bulky substrate 4l was also not C -hexosylated. In the case of C -arabinosylation, however, there was a distinctly lower tolerance for different B-ring functionality with products only generated from substrates 4f, 4h and traces from 4g, all very close in structure to the natural substrate 4a. These three substrates were incorporated to yield 10f, g, h and 11f, g, h (Supplementary data). Representative data ( Fig. 4 ) reveal two compounds in the ion trace ( Fig. 4B ) with ions of 591 ( Fig. 4A ) which are revealed as molecular ions [ M — H] — 591 ( Fig. 4C ) and which fragment to give losses of 60 ([ M — H — 60] — 531, losses of 90 [ M — H — 90] — 501}, losses of 120 ([ M — H — 120] — 471 and combinations of losses ([ M — H — 60-90] — 441, [ M — H — 60-120] — 411 and [ M — H — 90-120] — 381 ( Fig. 4D ) indicating both sugars must be C -linked to the apigenin core ( Fig. 1 ).

Figures

Scheme 1. Production of di-C-glycosides by the soluble root protein of D. incanum and suitable sugar donor cofactors from the subtrate generated by recombinant OsCGT.

Fig. 1. Fragmentation of C-hexosides and C-pentosides under high cone voltage ESIMS conditions to generate fragments losses of 60, 90 and 120 amus.

Fig. 3. (A) Ion scan for [M—H] — 621 showing three new metabolites found in (B) the LCMS ion trace. (C) Molecular ions of three new metabolites and (D) fragmentation under high cone voltage showing C-link fragments.

Fig. 4. (A) Ion scan for [M—H] — 591 showing two new metabolites found in (B) the LCMS ion trace. (C) Molecular ions of two new metabolites and (D) fragmentation under high cone voltage showing C-link fragments.