Dereplication-Guided Isolation of New Phenylpropanoid-Substituted Diglycosides from Cistanche salsa and Their Inhibitory Activity on NO Production in Macrophage

Dereplication allows for a rapid identification of known and unknown compounds in plant extracts. In this study, we performed liquid chromatography-mass spectroscopy (LC-MS)- based dereplication using data from ESI+ QTOF-MS for the analysis of phenylpropanoid-substituted diglycosides, the major active constituents of Cistanche salsa (C. A. Mey.) Beck. Using TOF-MS alone, the substructures of these compounds could be unambiguously confirmed based on the characteristic fragmentation patterns of various product ions. HPLC-MS based profiling of C. salsa also allowed for the detection of new phenylpropanoid-substituted diglycosides from this plant. Of them, five new phenylpropanoid-substituted diglycosides, named cistansalsides A–E (5, 6, 12, 17 and 18), were isolated. Their structures were elucidated through spectroscopic methods including NMR and MS analysis. All the isolates were tested for their inhibitory activity against NO production in RAW 264.7 cells stimulated by LPS. Of the tested compounds, compounds 5, 11, 13 and 18 showed moderate inhibitory activity on inducible NO synthase. Compounds 11, 13 and 18 also inhibited the phosphorylation of NF-κB in macrophages. None of the compounds displayed significant cytotoxicity.


Introduction
Cistanche salsa (C. A. Mey.) Beck, belonging to the family Orobanchaceae, is a parasitic plant that obtains its nutrition from the root of Haloxylon ammodendron (Chenopodiaceae) and other desert plants [1]. This plant has been used in traditional medicine for the treatment of neurasthenia, sexual dysfunction and kidney deficiency [2,3]. In previous phytochemical studies, it has been reported that the whole plant of C. salsa contained various types of compounds including phenylethanoid glycosides and iridoid glycosides [4][5][6][7]. Phenylethanoid glycosides, such as acteoside and echinacoside, are the major active constituents of the plant [8]. The extracts of C. salsa showed beneficial properties, including immunomodulatory, anticancer and antiinflammatory activities [9,10].
Dereplication is a process by which sample mixtures would be tested to differentiate unknown constituents from known compounds. The dereplication strategies are based on the analytical techniques and database searching to identify secondary metabolites early in the phytochemical research process [11]. Of the analytical techniques, ESI-QTOF-MS (electrospray ionization-quadrupole-time of flight-mass spectroscopy) could provide valuable information about chemical structures of secondary metabolites. The LC-MS-based dereplication-guided fractionation research process [11]. Of the analytical techniques, ESI-QTOF-MS (electrospray ionizationquadrupole-time of flight-mass spectroscopy) could provide valuable information about chemical structures of secondary metabolites. The LC-MS-based dereplication-guided fractionation has been demonstrated to enable extraction and purification of target metabolites from crude extracts of plants with high efficiency [12][13][14][15].
This study performed the LC-MS-based dereplication using data from ESI + TOF-MS for analysis of phenylpropanoid-substituted diglycosides, the major active constituents of C. salsa. The TOF-MS data could suggest the substructures of these compounds based on characteristic fragmentation patterns of various product ions. Based on this dereplication, LC-MS profiles of ethyl acetate (EtOAc) fraction and water-soluble fraction were analyzed. The EtOAc fraction was subjected to the dereplication strategy for further separation, resulted in the isolation of five new phenylpropanoidsubstituted diglycosides and 13 known compounds (Figure 1). It was confirmed that tentatively predicted structures of phenylpropanoid-substituted diglycosides were correctly matched to their real structures. In addition, the anti-inflammatory activities of the isolates were explored.

Results and Discussion
The phenylpropanoid-substituted diglycosides isolated from C. salsa usually have structures based on disaccharide glycosides, which consist of a glucose and a rhamnose with a Rha (1→3) Glc linkage and one cinnamoyl substituent, such as coumaric acid (Cou), caffeic acid (Caf) and feruloyl acid (Fer), at the C-4 or C-6 position of glucose. The aglycone is commonly attached at the C-1 position of glucose. The structures of phenylpropanoid-substituted diglycosides with an acetyl group at the C-2 of glucose have frequently been reported [6,12,16].
To perform the dereplication, MS fragmentation patterns of these compounds were analyzed by positive mode ESI-QTOF-MS. In MS spectra, all the phenylpropanoid-substituted diglycosides

Results and Discussion
The phenylpropanoid-substituted diglycosides isolated from C. salsa usually have structures based on disaccharide glycosides, which consist of a glucose and a rhamnose with a Rha (1→3) Glc linkage and one cinnamoyl substituent, such as coumaric acid (Cou), caffeic acid (Caf) and feruloyl acid (Fer), at the C-4 or C-6 position of glucose. The aglycone is commonly attached at the C-1 position of glucose. The structures of phenylpropanoid-substituted diglycosides with an acetyl group at the C-2 of glucose have frequently been reported [6,12,16].
To perform the dereplication, MS fragmentation patterns of these compounds were analyzed by positive mode ESI-QTOF-MS. In MS spectra, all the phenylpropanoid-substituted diglycosides produced adduct ion peaks at [M + NH 4 ] + , [M + K] + and [M + Na] + , which provided the molecular weight and formula. The pattern of fragment ions could be found by successive losses of aglycone and glycoside residues ([M + H − Aglycone] + , [M + H − Aglycone − Rha] + and [M + H − Aglycone -Rha − Glc (or Acetyl-Glc)] + ), which were useful for predicting the type of cinnamoyl substituent and sugars. The fragment ions at m/z 163 of the caffeoyl group, m/z 147 of the coumaroyl group or m/z 177 of the feruloyl group give the characteristic signal of a cinnamoyl substituent in the phenylpropanoid-substituted diglycosides [4,15] (Figure 2). The analysis of the fragment ions would provide useful information for the identification of the structures of phenylpropanoid-substituted diglycosides. However, their isomers could not be differentiated by MS spectrometry alone. For accurate identification of their complete structures, NMR spectra are required. Rha − Glc (or Acetyl-Glc)] + ), which were useful for predicting the type of cinnamoyl substituent and sugars. The fragment ions at m/z 163 of the caffeoyl group, m/z 147 of the coumaroyl group or m/z 177 of the feruloyl group give the characteristic signal of a cinnamoyl substituent in the phenylpropanoidsubstituted diglycosides [4,15] (Figure 2). The analysis of the fragment ions would provide useful information for the identification of the structures of phenylpropanoid-substituted diglycosides. However, their isomers could not be differentiated by MS spectrometry alone. For accurate identification of their complete structures, NMR spectra are required. C. salsa was analyzed and the fingerprint of the EtOAc fraction was generated using the HPLC-DAD (diode array detector)-ESI-QTOF-MS method ( Figure 3). Each peak in the fingerprint of C. salsa was predicted according to MS fragmentation features (Table 1). Many phenylpropanoid-substituted diglycosides were screened out from this fraction, which was subjected to HPLC-QTOF-MS-guided isolation for the discovery of new phenylpropanoid-substituted diglycosides. Eighteen peaks including five new compounds were further identified and their structures were elucidated through extensive spectroscopic analysis. C. salsa was analyzed and the fingerprint of the EtOAc fraction was generated using the HPLC-DAD (diode array detector)-ESI-QTOF-MS method ( Figure 3). Each peak in the fingerprint of C. salsa was predicted according to MS fragmentation features (Table 1). Many phenylpropanoid-substituted diglycosides were screened out from this fraction, which was subjected to HPLC-QTOF-MS-guided isolation for the discovery of new phenylpropanoid-substituted diglycosides. Eighteen peaks including five new compounds were further identified and their structures were elucidated through extensive spectroscopic analysis.    Table 2).   A 3,4-dihydroxyphenyl group was suggested by the HMBC correlations between H-2 and a quaternary aromatic carbon at δ C 149.4 (C-4 ) and between H-5 and C-1 (δ C 125.6) and C-3 (δ C 147.9). From the HMBC NMR spectrum, the correlations between a carbonyl carbon at δ C 165.8 and H-8 and between H-6 and C-7 (δ C 145.5) suggested a 3,4-dihydroxylated cinnamoyl group. The HMBC correlation between the methoxy proton and C-3 and the NOESY correlation between the methoxy proton and H-2 confirmed the cinnamoyl substituent to be an (E)-feruloyl group.
Two sugar moieties, suggested by the MS fragment pattern, were double-checked by the NMR spectra and HPLC analysis of the acid hydrolysate. The absolute configurations of them were determined to be D-glucose and L-rhamnose using HPLC analysis of the acid hydrolysate [17]. A β-glucose moiety and an α-rhamnose moiety were established by coupling constants of the anomeric protons. The 1 H-1 H COSY spectrum showed the sequential correlations from H-1 to H-5 and from H-1 to H-6 ( Figure 4).
Two sugar moieties, suggested by the MS fragment pattern, were double-checked by the NMR spectra and HPLC analysis of the acid hydrolysate. The absolute configurations of them were determined to be D-glucose and L-rhamnose using HPLC analysis of the acid hydrolysate [17].    Table 2).
From the HMBC spectrum, the 3,4-dihydroxyphenyl group was suggested by the correlations between H-2 and a quaternary aromatic carbon at δ C 149.0 (C-4 ) and between H-5 and another quaternary carbons at δ C 125.5 (C-1 ) and 148.5 (C-3 ). From the HMBC NMR spectrum, the correlations between H-6 and C-7 (δ C 145.6) and between the carbonyl carbon at δ c 165.7 (C-9 ) and H-8 suggested an (E)-caffeoyl group.
A 3-methylbutenyl group, an aglycone substructure, was suggested by the COSY correlations of H-2 with H-1a and H-1b and the HMBC correlations between H-4 and H-5 and the olefinic carbons, at δ C 120.7 (C-2) and 136.4 (C-3).
Two sugar moieties were established by the NMR spectra analysis and HPLC spectra analysis of the acid hydrolysate, with MS fragment pattern. The absolute configuration of them were determined to be D-glucose and L-rhamnose using HPLC analysis of the acid hydrolysate [17]. These sugar moieties were defined as a β-glucose and an α-rhamnose by coupling constants of the anomeric protons. The 1 H-1 H COSY spectrum showed the sequential correlations from H-1 to H-6 and from H-1 to H-6 ( Figure 4).
A downshifted glucose proton at δ H 4.70 (H-4 ) suggested an acyl-substituent on glucose. From the HMBC spectrum, the correlation between H-4 and C-9 confirmed the position of caffeoyl substituent. The HMBC correlations between H-1 and C-1 (δ C 64.5) and between H-3 (δ H 3.68) and C-1 (δ C 101. Comparison of the NMR spectra of 12 with those of 6 showed that they were similar except for the aglycone structure. In the NMR spectra of 12, two paraffinic carbons at δ C 38.0 (C-2) and 24.4 (C-3) were observed instead of two olefinic carbons at δ C 120.7 (C-2) and 136.4 (C-3) in the aglycone of 6. The germinal methyl groups (δ H 0.88) in the aglycone of 12 were shifted upfield relative to H-4 and H-5 (δ H 1.71 and 1.63) in the aglycone of 6 ( Table 2). The aglycone of 12 was suggested to be a 3-methylbutyl group, which was confirmed by the 1 H and COSY NMR spectra. Peaks of 3-methylbutyl group were observed at δ Two sugar moieties were reaffirmed by the HPLC analysis of the acid hydrolysate and the NMR spectra analysis as well as MS fragment pattern. The absolute configurations of the sugars were identified as D-glucose and L-rhamnose using HPLC analysis of the acid hydrolysate [17]. A β-glucose moiety and an α-rhamnose moiety were confirmed by coupling constants of the anomeric protons. The 1 H-1 H COSY spectrum showed the sequential correlations from H-1 to H-5 , from H-1 to H-3 and from H-6 to H-4 ( Figure 4).
The position of substituents were confirmed by means of the HMBC analysis. In the HMBC spectrum, the correlations between H-1 and C-1 (δ C 67.2), between H-3 (δ H 3.68) and C-1 (δ C 101.2) and between H-4 (δ H 4.70) and C-9 (δ C 165.7) were detected. Consequently  Table 2). From the HMBC NMR spectrum, the correlations between a carbonyl carbon at δ c 165.5 (C-9 ) and H-8 and between H-2 , 6 and C-7 (δ c 145.4) suggested an (E)-coumaroyl group. The HMBC correlation between a methyl proton peak and a carbonyl carbon at δ c 169.0 confirmed the presence of an acetyl group.
Two sugar moieties, suggested from MS fragment pattern, were reconfirmed by the HPLC analysis of the acid hydrolysate and NMR spectra. D-glucose and L-rhamnose were elucidated using HPLC analysis of the acid hydrolysate [17]. A β-glucose moiety and an α-rhamnose moiety were established by coupling constants of the anomeric protons. The 1 H-1 H COSY spectrum showed the sequential correlations from H-1 to H-5 and from H-1 to H-6 ( Figure 4).  Table 2). A phenylethyl group, an aglycone substructure, was suggested by the HMBC correlations between H-7 (2H, δ H 2.80, m) and C-1 (δ C 138.8) and between H-7 and C-2, 6 (δ C 128.9) and the COSY NMR signals of H-7 with H-8a (1H, δ H 3.99, m) and H-8b (1H, δ H 3.63, m).
Two sugar moieties were reaffirmed by the HPLC analysis of the acid hydrolysate and NMR spectra analysis. The absolute configurations of the sugars were elucidated using HPLC analysis of the acid hydrolysate, which were confirmed to be D-glucose and L-rhamnose [17]. A β-glucose moiety and an α-rhamnose moiety were established by coupling constants of the anomeric protons. The 1 H-1 H COSY spectrum showed the sequential correlations from H-1 to H-5 , from H-1 to H-2 and from H-6 to H-3 ( Figure 4).
Most of the trans-cinnamoyl substituents were isomerized to the cis-isoform in vitro. Light has been reported to convert trans-cinnamic acid derivatives into cis-isoforms [18,19]. The equilibrium of the trans-cis conversion of the cinnamoyl substituents was observed to maintain approximately 70% of the isolates in the trans-isoform. were observed for trans form [20]. In the 1 H-NMR spectrum of trans-cis mixtures, peaks for two olefinic protons (H-7 and H-8 ) were observed in the ratio of 7:3 (trans:cis). 13 C-NMR peaks of the cis form were similar to those of the trans form.
All the isolates were tested for their inhibitory effects on LPS-induced NO production in RAW 264.7 cells. Dexamethasone was used as a positive control and its IC 50 was 7.0 µM. Of the tested compounds, compounds 5 (IC 50 42.7 ± 6.6 µM), 11 (IC 50 37.3 ± 2.2 µM), 13 (IC 50 40.0 ± 4.0 µM) and 18 (IC 50 27.9 ± 0.8 µM) showed moderate inhibitory activities on inducible NO synthase, while the other compounds were inactive in this assay (IC 50 values > 100 µM). To verify whether these compounds had cytotoxicity, cell viability was measured employing MTT assay. As a result, none of them displayed significant cytotoxicity (Supplemental Figure S6-1). These four compounds were selected to evaluate for their inhibitory activity against NF-κB pathway in LPS-stimulated RAW 264.7 cells. Stimulation of RAW 264.7 cells with LPS induced the phosphorylation of IκBα and NF-κB (p65) after 0.5 h of incubation. The phosphorylation of NF-κB (p65) was significantly reduced by pretreatment with compounds 11, 13 and 18 as shown by western blot analysis ( Figure 5). Therefore, compounds 11, 13 and 18 might exert anti-inflammatory effects via the inhibition of NF-κB in macrophages. Most of the trans-cinnamoyl substituents were isomerized to the cis-isoform in vitro. Light has been reported to convert trans-cinnamic acid derivatives into cis-isoforms [18,19]. The equilibrium of the trans-cis conversion of the cinnamoyl substituents was observed to maintain approximately 70% of the isolates in the trans-isoform. For the olefin protons of the cis form, peaks at approximately 6.90 ppm (d, J = 12~13 Hz, H-7′′′) and 5.80 ppm (d, J = 12~13 Hz, H-8′′′) were assignable in the 1 H-NMR spectra, whereas peaks at approximately 7.55 ppm (d, J = 15.8 Hz, H-7′″) and 6.40 ppm (d, J = 15.8 Hz, H-8′′′) were observed for trans form [20]. In the 1 H-NMR spectrum of trans-cis mixtures, peaks for two olefinic protons (H-7′′′ and H-8′′′) were observed in the ratio of 7:3 (trans:cis). 13 C-NMR peaks of the cis form were similar to those of the trans form.
All the isolates were tested for their inhibitory effects on LPS-induced NO production in RAW 264.7 cells. Dexamethasone was used as a positive control and its IC50 was 7.0 μM. Of the tested compounds, compounds 5 (IC50 42.7 ± 6.6 μM), 11 (IC50 37.3 ± 2.2 μM), 13 (IC50 40.0 ± 4.0 μM) and 18 (IC50 27.9 ± 0.8 μM) showed moderate inhibitory activities on inducible NO synthase, while the other compounds were inactive in this assay (IC50 values > 100 μM). To verify whether these compounds had cytotoxicity, cell viability was measured employing MTT assay. As a result, none of them displayed significant cytotoxicity (Supplemental Figure S6-1). These four compounds were selected to evaluate for their inhibitory activity against NF-κB pathway in LPS-stimulated RAW 264.7 cells. Stimulation of RAW 264.7 cells with LPS induced the phosphorylation of IκBα and NF-κB (p65) after 0.5 h of incubation. The phosphorylation of NF-κB (p65) was significantly reduced by pretreatment with compounds 11, 13 and 18 as shown by western blot analysis ( Figure 5). Therefore, compounds 11, 13 and 18 might exert anti-inflammatory effects via the inhibition of NF-κB in macrophages.

General Experiment Procedure
Optical rotations were measured with a Jasco P-2000 digital polarimeter (Jasco, Tokyo, Japan). UV spectra were recorded on a Chirascan plus Circular Dichroism spectrometer (Chirascan, APL, UK). IR spectra were recorded using Jasco FT/IR-4200 spectrophotometer. High-resolution electrospray

Immunoblot Analysis
Protein expression was assessed by western blotting according to standard procedures. Briefly, RAW264.7 cells were cultured in 60 mm culture dishes (2 × 10 6 /mL), following by pretreatment 50 µM of compounds. Cells were washed twice in ice cold PBS (pH 7.4), the cell pellets were resuspended in lysis buffer on ice for 15 min, and the cell debris was then removed by centrifugation. Protein concentration was determined using Bio-Rad protein assay reagent according to the manufacturer's instructions. Protein (20-30 µg) was mixed 1:1 with 2× sample buffer (20% glycerol, 4% SDS, 10% 2-ME, 0.05% bromophenol blue, and 1.25 M Tris [pH 6.8]), loaded onto 8 or 15% SDS-PAGE gels, and run at 150 V for 90 min. Cellular proteins were transferred onto ImmunoBlot polyvinylidene difluoride membranes (Bio-Rad) using a Bio-Rad semi-dry transfer system according to the manufacturer's instructions. The membranes were then incubated overnight with the resprective p-NF-κB, NF-κB, p-IκBα and β-actin primary antibodies (Abcam, Cambridge, UK) in Tris-buffered saline containing 5% skimmed milk and 0.1% Tween 20. The following day, the blots were washed three times with Tris-buffered saline (0.1% Tween 20) and incubated for 1 h with an HRP-conjugated secondary anti-IgG antibody (diluted 1:2000-1:20,000). The blots were washed again three times with Tris-buffered saline (0.1% Tween 20), and immunoreactive bands were developed using the chemiluminescent substrate ECL Plus (Amersham Biosciences, Piscataway, NJ, USA).

Statistical Analysis
Experimental data are presented as the mean ± SEM. The level of statistical significance was determined by analysis of variance (ANOVA) followed by Dunnett's t-test for multiple comparisons. p Values less than 0.05 were considered significant.
Supplementary Materials: Supplementary data associated with this article can be found in the PDF file (supplementary material).