*2.3. Qualitative Analysis*

As an important edible medicinal plant for Mongolian people, *A. mongolicum* has made a great contribution to the development of the local economy, yet there is a lack of analysis of its quality until now.

Our systematic phytochemistry isolation results indicated the main constituents of AM were flavonoids and phenolic acids. The aglycones in the plant mainly included quercetin, kaempferol, as well as isorhamnetin for flavonoid glycosides; while coumaric acid, caffeic acid, and ferulic acid for phenolic acid glycosides. The sugars consisted of β-d-glucopyranoside (Glc), α-d-glucopyranoside (α-Glc), β-d-glucuronic acid (Glu), and α-l-rhamnopyranoside (Rha). While α-Glc was only found in phenolic acid glycosides, Glu and Rha substituted only for flavonoid glycosides.

As for flavonoids, 3-, 7-, and 4- -OH of quercetin, kaempferol, and isorhamnetin were easily substituted by various glycosyls to format *O*-glycosides. Among them, 7- and 4- -OH was substituted by monosaccharose such as Glc and Glu, while Glu was found to only link with their 7-position. Meanwhile, 3-OH was with a high degree of glycosylation, having one to three sugar moieties, and all of the glycosyl groups directly linked to flavonoid was Glc group, then its 2-, 4-, or 6-position was substituted by another Glc continuously; moreover, its 6-position could also be replaced by Rha [to form rutinosyl (Rut)] or acetyl group (Figure 5).

On the other hand, the carboxyl of obtained phenolic acids from AM was easily substituted by sugar moiety such as Glc(1→2)Glc–, α-Glc(1→2)Glc–, or Glc(1→6)Glc(1→2)Glc– on their 9-position, while 4-OH of them was only substituted by monosaccharose, Glc (Figure 6).

Herein, on the basis of above-mentioned phytochemistry study, a fast analysis method for flavonoids and phenolic acids in AM was established by LC-MS on an ESI-Q-Orbitrap MS in negative ion mode (Figure 7). According to the chromatographic retention time (*t*R) and the exact mass-to-charge ratio (*m*/*z*), 31 compounds (**1**–**31**) were unambiguously identified by comparing to the standard references. Meanwhile, the rules of the MS/MS fragmentation pattern and chromatographic elution order have been generalized. Then, five flavonoid glycosides (**32**–**36**) and one phenolic acid glycoside (**37**) were tentatively speculated. Among them, **36** was a potential new compound (Table S1, Figure 8).

**Figure 5.** The structure of aglycones and glycosyls of flavonoids from the aerial parts of *A. mongolicum*.

**Figure 6.** The structure of aglycones and glycosyls of phenolic acids from the aerial parts of *A. mongolicum*.

**Figure 7.** *Cont*.

**Figure 7.** Base peak chromatograms of AM (the extract obtained from fresh the aerial parts of *A. mongolicum* heated reflux with 95% EtOH and 50% EtOH one time each, successively), AMH (H2O layer extract), AME (EtOAc layer extract), and mixed standard references.

**Figure 8.** The structures of tentatively presumed compounds from the aerial parts of *A. mongolicum*.

#### 2.3.1. Structural Elucidation of Flavonoids

Peaks 3- –6- , 9- , 10- , 13- , 16- , 21- , 23- –27 and 29- –37 were identified by comparison with reference standards (Table S1, Figure 7).

Figure 9 and Figure S74 showed the MS/MS fragmentation pattern of flavonoid glycosides with kaempferol and quercetin aglycones, which suggested both of two kinds of flavonoid glycosides could be ionized to generate heterolytic cleavage with fragments ion peak at *m*/*z* 285.03936 (YK0−) for kaempferol and *m*/*z* 301.03428 (YQ0<sup>−</sup>) for quercetin, as well as hemolytic cleavage with fragments ion peak at *m*/*z* 284.03154 [YK0<sup>−</sup> − H]<sup>−</sup> for kaempferol and *m*/*z* 300.02645 [YQ0<sup>−</sup> − H]<sup>−</sup> for quercetin, respectively. Then, kaempferol aglycone could be further cleavage to generate fragment ion peaks at *m*/*z* 255.02880, 179.02371, and 151.00259. The fragment ion peaks at *m*/*z* 271.02371, 255.02880, 243.02880, 179.02371, as well as 151.00259, were yielded from quercetin by a series of reactions

including decarbonylation, dehydrogenation, retro Diels–Alder reaction, and the reaction to remove the B ring. The above-mentioned characteristic fragment ions could be used to distinguish the type of aglycone.

**Figure 9.** The proposed fragmentation pathways of kaempferol and quercetin glycosides.

Meanwhile, when the 4- -position of flavanol aglycone was glycosylated to format *O*-glycoside, the debris ions peaks ([YK0<sup>−</sup>2H]−) at *m*/*z* 283.02371 for kaempferol and *m*/*z* 299.01863 ([YQ0<sup>−</sup>2H]−) for quercetin glycosides were stronger than those of *m*/*z* 284.03154 and 300.02645, respectively (Table S1). Therefore, their ionic strength could be used to quickly determine whether the C-4 position of the aglycone was replaced by sugar.

Peaks 9- , 22- , 23- , and 27 were obtained by extracting ion of *m*/*z* 771.19783 from the total ion chromatogram of AM (Figure 10), among them, 9- , 23- , and 27 were clarified to be kaempferol-3,7,4- -tri-*O*-β-glucoside (**18**), kaempferol-3-*O*-gentiobioside-4- -*O*-glucopyranoside (**17**), and mongoflavonoside A2 (**2**) by comparing with reference standards. Then, according to the above-mentioned biosynthetic pathway of substituted sugar, peak 22 was tentatively presumed to be kaempferol-3-*O*-β-d-glucopyranosyl(1→2)-*O*-β-d-glucopyranosyl-4- -*O*-β-d-glucopyranoside (**36**), which was one new compound.

**Figure 10.** The EIC of the *m*/*z* 771.19783.

Moreover, during the comparison of the chromatographic retention behavior of peaks 9- , 22- , 23- , and 27- , we discovered the effect of sugar substitution position on *t*<sup>R</sup> was 3,7,4- -tri-*O*-Glc < 3-*O*-Glc(1→2)-Glc-4- -*O*-Glc < 3-*O*-Glc(1→6)-Glc-4- -*O*-Glc < 3-*O*-Glc(1→4)-Glc-4- -*O*-Glc.

The molecular formula of peaks 3- (*m*/*z* 801.17407), 8- (*m*/*z* 801.17462), 11- (*m*/*z* 801.17389), and 14- (*m*/*z* 801.17200) were all C33H38O23 (Figure 11). Peak 3 was unambiguously identified as mongoflavonoside B2 (**6**) by comparison with reference standard. According to the MS/MS fragment ion peaks at *m*/*z* 301.03428, 300.02645, 299.01863, 271.02371, 255.02880, and 151.00259, peaks 8- , 11- , and 14 were deduced to be with quercetin aglycone. On the other hand, the fragment ion peaks at *m*/*z* 625.13993 [M − H − 176]−, 301.03428 [M <sup>−</sup> <sup>H</sup> <sup>−</sup> <sup>176</sup> <sup>−</sup> <sup>162</sup> <sup>−</sup> 162]<sup>−</sup> suggested the presences of one <sup>β</sup>-d-Glu and two <sup>β</sup>-d-Glc in them. Since the strength of fragment ion peak at *m*/*z* 299.01863 was weaker than that of *m*/*z* 300.02645, we could propose that 4- -OH of quercetin was not be glycosidated. According to the above-mentioned biosynthetic pathway of substituted sugar and effect of sugar substitution position on *t*R, peaks 8- , 11- , and 14 were tentatively presumed to be quercetin-3-*O*-β-d-glucopyranosyl(1→2)-β-d-glucopyranosyl-7-*O*-β-d-glucuronide (**32**), quercetin-3-*O*-β-d-glucopyranosyl(1→6)-β-d-glucopyranosyl-7-*O*-β-d-glucuronide (**33**), and quercetin-3-*O*-β-d-glucopyranosyl(1→4)-β-d-glucopyranosyl-7-*O*-β-d-glucuronide (**34**), respectively.

**Figure 11.** The EIC of the *m*/*z* 801.17201.

The molecular formula of peak 28- (*m*/*z* 799.19391) was C33H38O23. Its MS/MS fragment ion peaks displayed at *m*/*z* 623.15869 [M − H − 176], 315.05048 [M − H − 176 − 162 − 146]−, 300.02713, 271.02469, and 243.02880 suggested the aglycone of it was isorhamnetin and the substituted sugar moieties included one Glu, one Glc, and one Rha. According to the biosynthesis laws summarized above, peak 28was deduced to be isorhamnetin-3-*O*-rutinosyl-7-*O*-β-d-glucuronide (**37**) (Table S1, Figure S75).

#### 2.3.2. Structural Elucidation of Phenolic Acids

Peaks 1- , 2- , 7- , 12- , and 17- –20 were identified unequivocally by comparing with reference standards (Table S1, Figure 7). As what have been mentioned above, the aglycones of phenolic acid glycosides included coumaric acid, caffeic acid, and ferulic acid. It was well known that the characteristic ions of coumaroyl, caffeoyl, and feruloyl were at *m*/*z* 163.03897 ([coumaroyl − H]−), 179.03389 ([caffeoyl − H]−), and 193.04954 ([feruloyl − H]−), respectively [31]. Then, all of the ions would further generate fragment ion peaks (as shown in Figure 12) by removing 44 Da (–CO2), 28 Da (–CO), and 18 Da (–H2O), respectively.

**Figure 12.** The proposed fragmentation pathways of coumaric acid, caffeic acid, and ferulic acid glycosides.

Meanwhile, the phenomenon of the neutral loss 120 Da on the basis of [M − H]<sup>−</sup> were only found in <sup>β</sup>-d-glucopyranosyl(1→2)-β-d-glucopyranosyl substituted phenolic acid glycosides **<sup>7</sup>** (peak 7- ), **28** (peak 12- ), and **29** (peak 17- ) (Figure 13 and Figure S76), which could be used to distinguish the type of substituted sugar moieties.

**Figure 13.** The proposed fragmentation pathways of <sup>β</sup>-d-glucopyranosyl(1→2)-β-d-glucopyranosyl substituted phenolic acid glycosides.

Moreover, comparing the *t*<sup>R</sup> of compounds **10** (peak 18- ) and **29** (peak 17- ), α-d-glucopyranosyl-substituted phenolic acid glycoside was found to have the shorter *t*<sup>R</sup> than that of β-d-glucopyranosyl-substituted ones.

The molecular formula of peak 15- (*m*/*z* 487.14313) was C21H28O13. Its MS/MS fragment ion peaks displayed at *m*/*z* 367.10297, 163.03888, and 145.02829, which was similar to those of peak 12- (Table S1, Figure 14). According to the above-mentioned chromatographic retention behavior, we could deduce that peak 15 was not *<sup>p</sup>*-hydroxycinnamic acid-9-*O*-α-d-glucopyranosyl(1→2)-β-d-glucopyranoside. As Han et al. reported, the *t*<sup>R</sup> of *cis*-phenylpropane glycoside was longer than that of *trans* one when they were analysed by HPLC with the acetonitrile-water system [24]. Consequently, peak 15 was tentatively presumed to be *cis*-*p*-hydroxycinnamate sophorose (**35**).

**Figure 14.** The tandem MS of the [M − H]<sup>−</sup> ions for peaks 12 and 15- .
