Planodes, Newman, 1842

Agerbirk, Niels, Hansen, Cecilie Cetti, Olsen, Carl Erik, Kiefer, Christiane, Hauser, Thure P., Christensen, Stina, Jensen, Karen R., Ørgaard, Marian, Pattison, David I., Lange, Conny Bruun Asmussen, Cipollini, Don & Koch, Marcus A., 2021, Glucosinolate profiles and phylogeny in Barbarea compared to other tribe Cardamineae (Brassicaceae) and Reseda (Resedaceae), based on a library of ion trap HPLC-MS / MS data of reference desulfoglucosinolates, Phytochemistry (112658) 185, pp. 1-19 : 7-8

2.2.2. The genus Planodes

The genus Planodes comprises two species only. Seeds of the small weedy plant P. virginica had been investigated previously using now historical methods ( Gmelin et al., 1970). Those authors reported the presence of EBAR (40 R). 7mSOh (66) and 8mSOo (69), two of which were deduced for the first time in that paper, and one or more lipophilic GSLs that were not identified. We wished to repeat and expand this experiment, testing all major plant parts, with focus on searching for similarities with and contrast to Barbarea and Cardamine species. In particular, we wanted to search for oxidized derivatives of homoPhe derived GSLs, for distubstituted Trp-derived in roots, for acyl derivatives in seeds, and for GSLs derived from BCAAs, as many of these kinds of structures were potentially due to de novo evolution in one or more species in the tribe. Likewise, we wished to know the full complement of n-homoMet-derived GSLs in the species, as this profile was a candidate for an ancestral profile of the Barbarea species.

The historical report on the dominant glucosinolates in P. virginica seeds was fully confirmed ( Fig. 3A ). Of particular interest was the dominance of EBAR that also occurs in most Barbarea species ( Table 4 ). In addition, minor levels of BAR and NAS were discovered. Of Trp-derived GSLs, only the parent structure IM was detected, in contrast to the prominence of monosubstituted indoles in roots of most other known crucifers. Clearly, this species did not accumulate the peculiar disubstituted indole, 1,4moIM, known from roots of one Barbarea species.

In addition to the previously known dominating methylsulfinyl alkyls with chain length 7–8, a wide range of chain lengths were found ( Table 5 ), supported by coelution with authentic standards except in the case of 9mSOn (68). For 5mSOp (72), the only authentic standard available was A. thaliana seeds, which was considered reliable. The retention times of the homologs increased in an expectable way for a homologous series (Supplementary Fig. S3C ). The characteristic MS2 fragmentation (including loss of CH 3 SOH) was observed for all major peaks including the tentatively identified 9mSOn (Supplementary Fig. S3D ). From the latter fragmentation, isomeric hydroxylated methylthioalkyls could be ruled out. The ‘bell-shaped’ level distribution of chain lengths ( Table 5 ) was similar to distributions in some other species, as expected from the properties of MAM-enzymes that are responsible for the chain elongation ( Textor et al., 2007; Kumar et al., 2019; Petersen et al., 2019).

The more lipophilic methylthioalkyl glucosinolates were mainly detected in seeds ( Table 5 ). Only two were identified with certainty, 8mSo from comparison with an authentic standard, and 7mSh from comparison with A. thaliana seeds, considered a reliable reference material. However, additional members of the entire homologous series from n = 5 to n = 9 were tentatively detected, supported by reasonable t R compared to that of 8mSo, 7mSh and of 4mSb available as standards.

Due to the wide range of GSLs confirmed in P. virginica seeds and their public availability at Nordgen, we imagined that they could be useful as reference material and compiled various illustrative chromatograms (Supplementary Fig. S3 ). Pioneers in the study of P. virginica also confirmed the R -configuration of the chiral sulfur in the side-chain methylsulfinyl groups ( Gmelin et al., 1970), which adds further reliability to this reference material.

Since P. virginica shared high levels of EBAR with Barbarea , some other characteristic Barbarea GSLs were searched for ( Table 4 ). ThioGlcacylated GSLs were searched for but not found. Similarly, hydroxyl and methoxyl derivatives of the major EBAR were searched for. A possible trace of the 4-hydroxy derivative 4hEBAR (139 R) was detected in many samples at very low levels but not confirmed by MS2 using ion trap HPLC-MS. As 4-hydroxylation is seasonally regulated in Barbarea (Agerbirk and Olsen, 2015) , a new set of samples were collected earlier in the next year (accession 2), in an attempt to confirm or reject the presence of 4hEBAR. The putative trace of 4hEBAR was also observed in this second accession, but of the same very low level as in the first accession and still MS2 confirmation was not obtained by ion trap HPLC-MS. If 4hBAR or 4hEBAR were present, it would be the first occurrence outside the genera Barbarea and Arabis . In order to settle the uncertainty, we carried out more sensitive UHPLC-QToF MS/MS with higher mass- and chromatographic resolution. At these conditions, authentic desulfo 4hBAR and 4hEBAR mainly resulted in proton adducts after loss of water; [M-H 2 O+H] +. Re-analysis of three P. virginica leaf samples of accession 2 and seeds of accession 1 demonstrated that 4hEBAR was present at very low levels, supported by matching t R and MS2. The high-resolution mass matched the expected (found: 358.0953 (mean, N = 4), calculated for C 15 H 20 O 7 NS + ([M-H 2 O+H] +): 358.0955). Critical MS2 features supporting the identification was loss of anhydroGlc and a characteristic major loss of m / z 220, plus a combined loss of anhydroGlc and hydroxylamine (Supplementary Fig. S4 A-C). In the seed sample, which had the highest level, loss of hydroxylamine from the [M–H 2 O–ahGlc+H] + ion was also confirmed (Supplementary Fig. S4E ). We concluded that 4hEBAR had been detected conclusively in two accessions of P. virginica , but at rather low levels ( Table 4 ). We did not detect any other known derivatives of EBAR (the 3-hydroxy derivative 3hEBAR (142 R), the 4-methoxy derivative 4moEBAR (50 R) or the 3/4 methoxy-hydroxy derivative “ x1 ”) in any sample. As P. virginica shared chain elongated Met derived GSLs with A. thaliana , some rare and characteristic GSLs from this species were searched for but not found (hydroxyalkyls and their benzoyl esters). This negative result was likewise obtained for two more species discussed below ( Table 5 ).

We also searched for GSLs derived from BCAAs, because their distribution in the tribe was poorly known. No GSL derived from non-chain elongated BCAAs was detected ( Table 5 ). However, a peak corresponding to an isomeric hydroxyhexylGSL was detectable at minute levels, with a further isomer at close to trace levels (Supplementary Figure 4H , panels H1–H3). The sum formula was supported by QToF-MS (found 340.1424, calculated for [C 13 H 26 O 7 NS] +: 340.1425) (panel H1). Two such isomers are known, 141 and 149. The t R of the unknown dGSLs were far from that of d 141 but the major was close to d 149. However, a reproducible slight (0.1 min) mismatch with d 149 was observed using UHPLC, confirmed by spiking that resulted in a splitpeak (results not shown). A quite characteristic MS2 ion trap spectrum, with six major fragment ions, was confirmed in several samples (Supplementary Fig. S3G ). The ion trap MS2 included an e-fragment ( Table 1), which is indicative of a β- hydroxylated dGSL ( Olsen et al., 2016). We conclude the presence of a hitherto unknown isomeric β- hydroxyhexyl GSL at levels around 0.01 μmol/g dry wt. in all tested plant parts of both P. virginica accessions. Possible identities could be a straight chain isomer or “2h4mPe” derived from 2-homoLeu, not yet known to science. Several structurally related known GSLs were searched for but not found, including the parent GSL 3mPe, the δ- hydroxylated isomer 3hmPe and lower homologs ( Table 5 ).

Figures

Fig. 3. HPLC-MS chromatograms of desulfoglucosinolates (dGSLs) prepared from glucosinolates (GSLs) in Planodes virginica (A) and Nasturtium officinale (B–C) seeds, showing qualitative similarities and quantitative contrasts. Major peaks (B) from N. officinale revealed many of the same GSLs as in A, but levels of EBAR (40R) were much lower while levels were much higher for the biosynthetic precursor PE (105). A focus on trace peaks from N. officinale (C) revealed sharp peaks representing a range of minor constituents. Due to the closely eluting peaks, the latter chromatograms (B–C) were made by combining extracted ion chromatograms corresponding to [M+Na]+ of the indicated dGSLs. In C, the m/z 366 signal of d105 was omitted to allow visualization of minor coeluting peaks. An asterisk after a peak number indicates tentative identification. HPLC-MS conditions as in Olsen et al. (2016). TIC, total ion chromatogram, EIC, extracted ion chromatogram.

Fig. 4. HPLC-MS chromatogram of desulfoglucosinolates (dGSLs) prepared from glucosinolates (GSLs) from leaves (A) and petioles (B) of horseradish (Armoracia rusticana), focusing on trace level GSLs. The chromatograms were made by combining extracted ion chromatograms corresponding to [M+Na]+ of the indicated dGSLs, from analyses that were much overloaded with respect to the dominating dGSL d107 from Pren. In panel A, an insert shows magnification of the chromatogram from 5.2 to 5.8 min. Neither suggested BAR nor EBAR were detectable. In panel B, only extracted ion chromatograms of m/z 382 (BAR/EBAR), 352 (BZ), 366 (PE), 380 (3PP), 394 (4PB), 408 (5PP at high tR and 6mSOh at 5.4 min), 422 (7mSOh) and 436 (8mSOo) are included. Unlabeled trace peaks did not exhibit a combination of tR and m/z suitable for any of the mentioned candidates. HPLC conditions as in Olsen et al. (2016). Panel A depicts analysis of the Copenhagen garden accession; panel B from the naturalized population at Lake Fures¨o. An asterisk after a peak number indicates tentative identification.