2.3. Identification of α -glucosidase inhibitory compounds from Alnus firma A previous study reported that H 2 O or MeOH extracts of A. firma leaves exhibited inhibitory activity against α- glucosidase (Yu et al., 2007). However, none of the compounds isolated in this study showed α- glucosidase inhibitory activity, and other metabolites from Alnus have not been reported to exhibit α- glucosidase inhibition, except three cyclic diarylheptanoids isolated from A. sieboldiana (Chiba et al., 2013). We hypothesized that the chemical information derived from MS/MSbased dereplication in this study could be used to reveal which compounds contribute to the α- glucosidase inhibitory activity of A. firma, and tried to associate the annotations to bioactive extracts. 15 Alnus extracts were evaluated for their α- glucosidase inhibitory activity as shown in Table 1. As previously reported, the leaf extract of A. firma exhibited potent inhibitory activity showing IC 50 values of 12.29 μg/ mL. Extracts of A. firma bark, twigs, and fruits, and A. hirsuta var. sibirica fruits also showed potent activity with IC 50 ranging from 6.80 to 8.48 μg/mL. Extract of A. japonica leaves and A. hirsuta var. sibirica leaves exhibited moderate inhibitory activity with IC 50 values of 23.36 and 29.84 μg/mL, while other extracts showed very weak or no inhibitory activity against α- glucosidase. To bridge the bioactivity and chemical information, we applied the workflow named “bioactive molecular networking”, which was developed in our previous study (Nothias et al., 2018). In this workflow, the Pearson correlation coefficients (r) and their significances (p values) between the semi-quantitative intensities of MS/MS features and the bioactivity value are calculated. By plotting r and p values on the molecular network, it was easily visualized that MS/MS spectra showing negative r values (negative correlation with IC 50; meaning possibly inhibitory against α- glucosidase) with significance (p <0.05) are mainly clustered within the molecular families B (ellagitannins) and G (gallotannins) (Fig. S3, Supplementary Data). In order to narrow down the candidates from chemical classes to single compounds, we filtered the candidate list using the Bonferroni correction for multiple hypothesis testing (Nothias et al., 2018). As a result, only one MS/MS feature (10) was left as a possible contributor to the bioactivity showing an r value of −0.8625 (IC 50 values were applied as bioactivity index, so negative r values mean higher contribution to bioactivity) and a p value of 3.52 × 10 −5. The MS/MS feature 10 was annotated as gallic acid by spectral matching, and the identification was confirmed by standard compound injection. We inspected the MS/MS feature table and confirmed that active extracts (Af-B, Af-T, Af-L, Af-F, Aj-L, Ahv-L, and AhvF; ion intensity for 10 ranging from 49,854 to 96,211) show a higher intensity of the gallic acid ion than inactive extracts (Aj-B, Aj-T, Aj-F, Ah-L, Ah-F, Ah-B, and Ahv-T; ion intensity for 10 ranging from 520 to 11,358) as shown in Fig. 5. A previous study reported gallic acid as a potent α- glucosidase inhibitor (Wansi et al., 2007), so it could be hypothesized that gallic acid is the main contributor of α- glucosidase inhibition of these Alnus extracts. Although this result did not provoke a discovery of a previously unknown bioactive compound, it demonstrated that the digitized MS/MS-based dereplication strategy can reveal bioactive components from complex phytochemical extracts and reduce the unnecessary effort spent in re-isolation of previously known bioactive compounds.

Published as part of Kang, Kyo Bin, Woo, Sunmin, Ernst, Madeleine, van der Hooft, Justin J. J., Nothias, Louis-Félix, da Silva, Ricardo R., Dorrestein, Pieter C., Sung, Sang Hyun & Lee, Mina, 2020, Assessing specialized metabolite diversity of Alnus species by a digitized LC-MS / MS data analysis workflow, pp. 1-10 in Phytochemistry (112292) (112292) 173 on pages 3-4, DOI: 10.1016/j.phytochem.2020.112292, http://zenodo.org/record/8294564