Study on Chemical Profile and Neuroprotective Activity of Myrica rubra Leaf Extract

The chemical profile of Myrica rubra (a native species in China) leaf extract was investigated by UPLC-PDA-HRMS, and the neuroprotective activity of two characteristic constituents, myricanol and myricetrin, was evaluated with N2a cells using H2O2-inducedoxidative challenge through a series of methods, e.g., MTT assay, ROS assay and [Ca2+]i assay. Among the 188 constituents detected in the extract of Myrica rubra leaf, 116 were identified definitely or tentatively by the comprehensive utilization of precise molecular weight and abundant multistage fragmentation information obtained by quadrupole orbitrap mass spectrometry. In addition, 14 potential new compounds were reported for the first time. This work established an example for the research of microconstituents in a complex analyte and revealed that suppression of H2O2-induced cytotoxicity in N2a cells was achieved by the pretreatment with myricanol. The evidence suggested myricanol may potentially serve as a remedy for prevention and therapy of neurodegenerative diseases induced by oxidative stress.


Introduction
Myrica rubra is a subtropical perennial plant belonging to the Myricaceae genus [1], and widely distributed in the east of Asia, e.g., China, Japan and Korea. For a long period of time, it had been used as traditional medicine for the treatment of skin diseases and diarrhea [2]. Moreover, it could be taken orally for traumatic injury, bone fracture, diarrhea, stomach and duodenal ulcer in ethno-medicine [3]. Various phytochemicals isolated from the leaves of Myrica rubra have been extensively investigated and generally reported to show numerous bioactivities, such as melanin synthesis-inhibitory [4], antitumor [5], and anti-influenzavirus activity [6]. However, there had been little information concerning the possible neuroprotective activity of Myrica rubra.

Results and Discussion
The widespread use of Myrica rubra in folk medicine has motivated intensive exploration of its pharmacological activities. Prior to that, the identification of constituents is of great importance and the chemical profile should be studied. Then the typical constituents were chosen to evaluate their neuroprotective activity. An approach based on UPLC-PDA-HRMS analysis and neuroprotective activity evaluation was proposed. The schematic diagram is illustrated in Figure 1.

Identification of Constituents in the Whole Extract of Myrica rubra
The precise mass of the precursor ions from the MS 1 spectra and product ions from MS 2 spectra can meet regulatory requests for qualitative analysis, and the combinations of both have been extensively used in the analysis of complex samples [28][29][30][31][32][33]. A total number of 188 compounds were detected in the extract of Myrica rubra leaf (listed in Supplementary Materials Table S1), and the total ions current (TIC) chromatograms by UPLC-HRMS were displayed with negative and positive modes in Figure 2A,B, respectively. Chemical structures of some identified compounds are depicted in Figure 3, including organic acids and their derivatives, flavonoids, cyclic diarylheptanoids, amino acids and peptides etc. Among them, eight compounds (seen in Section 3.3) were explicitly identified by comparison of their retention times and MS spectra with the reference substances. By analyzing the molecular formula and MS 2 fragmentation information, another 108 compounds were characterized tentatively with the aid of the Chemical Abstracts database. A mass error of less than 6 ppm was routinely achieved for the accurate mass measurement.

Identification of Organic Acids and Their Derivatives
Twenty four organic acids and their derivatives were detected in the extract of Myrica rubra leaf. They were only detected in negative ion mode for the presence of carboxyl, so further analysis and discussion were employed in this mode. Among them, compounds 11, 23, 45 and 68 were explicitly verified to be gallic acid, protocatechuic acid, 4-hydroxybenzoic acid, and caffeic acid in accordance with reference standards. They all produced their characteristic fragment ions by the neutral loss of CO2, which suggested the existence of a carboxyl group in their molecular structures [34].
Compound 51demonstrated a [M − H] − ion at m/z 183.02815 for an elemental composition of C8H7O5, which was 15 Da more than that of gallic acid (compound 11, m/z 169), implying the pesence of a methyl group attached to gallic acid. Therefore, compound 51 was deduced to be methylgallic acid. Compounds 20, 42 and 49 were tentatively ascribed to be dihydroxybenzoic acids for all of them shared the identical quasi-molecular ions and fragment routes with protocatechuic acid (compound 23). Compound 38 was xyloside for it generated an aglycone ion at m/z 153 by the loss of xylosyl, followed by the same fragmentation pathway. Compound 110 was observed as a deprotonated molecule at m/z 193 in negative ion mode, which was 14 Da (CH2) more than that of caffeic acid (compound 68), indicative of a methyl group connected to it. By searching the compounds reported in natural products, compound 110 was inferred to be ferulic acid. Meanwhile, compound 94 generated a precursor ion at m/z 163 (C9H7O3), which was 16 Da less than that of caffeic acid, implying there were only one hydroxyl in the benzene ring, so it was assumed to be coumaric acid. Furthermore, compound 61 was identified as its glucoside due to the occurrence of fragmentation ions at m/z 163

Identification of Flavonoids
There were three types of flavonoids identified in the extract of Myrica rubra, namely flavan-3ols, flavonols and xanthones. Their UV spectra and fragmentation pathways bear similarities and differences, which would be helpful for their identification and differentiation. The identification of this type of compounds is partly displayed in Figure 4.

Identification of Organic Acids and Their Derivatives
Twenty four organic acids and their derivatives were detected in the extract of Myrica rubra leaf. They were only detected in negative ion mode for the presence of carboxyl, so further analysis and discussion were employed in this mode. Among them, compounds 11, 23, 45 and 68 were explicitly verified to be gallic acid, protocatechuic acid, 4-hydroxybenzoic acid, and caffeic acid in accordance with reference standards. They all produced their characteristic fragment ions by the neutral loss of CO 2 , which suggested the existence of a carboxyl group in their molecular structures [34].
Compound 51demonstrated a [M − H] − ion at m/z 183.02815 for an elemental composition of C 8 H 7 O 5 , which was 15 Da more than that of gallic acid (compound 11, m/z 169), implying the pesence of a methyl group attached to gallic acid. Therefore, compound 51 was deduced to be methylgallic acid. Compounds 20, 42 and 49 were tentatively ascribed to be dihydroxybenzoic acids for all of them shared the identical quasi-molecular ions and fragment routes with protocatechuic acid (compound 23). Compound 38 was xyloside for it generated an aglycone ion at m/z 153 by the loss of xylosyl, followed by the same fragmentation pathway. Compound 110 was observed as a deprotonated molecule at m/z 193 in negative ion mode, which was 14 Da (CH 2 ) more than that of caffeic acid (compound 68), indicative of a methyl group connected to it. By searching the compounds reported in natural products, compound 110 was inferred to be ferulic acid. Meanwhile, compound 94 generated a precursor ion at m/z 163 (C 9 H 7 O 3 ), which was 16 Da less than that of caffeic acid, implying there were only one hydroxyl in the benzene ring, so it was assumed to be coumaric acid. Furthermore, compound 61 was identified as its glucoside due to the occurrence of fragmentation ions at m/z 163 [M -H − glucosyl] − and m/z 119 [M -H -glucosyl − CO 2 ] − .

Identification of Flavonoids
There were three types of flavonoids identified in the extract of Myrica rubra, namely flavan-3-ols, flavonols and xanthones. Their UV spectra and fragmentation pathways bear similarities and differences, which would be helpful for their identification and differentiation. The identification of this type of compounds is partly displayed in Figure 4.
Compound 78 was unequivocally ascertained to be L-epicatechin according to comparison with an authentic standard. In the MS 2 spectrum, the [M − H] − ion at m/z 289.07062 gave the base peak ion at m/z 247 by the loss of C 2 H 2 O. In addition, it produced a product ion of high abundance at m/z 125 through the cleavage of A 1,4 type, along with the B ring fragment ion at m/z 109. In positive ion mode, the [M + H] + ion gave an A 1,3 ion at m/z 139 as the base peak from RDA reaction. There were several subclasses of compounds of this type (shown in Figure 3).
Compounds 24 and 53 had identical quasi-molecular ions at m/z 307 (C 15 H 13 O 7 ) in negative ion mode, which was 16 Da more than that of L-epicatechin, indicating that there was a hydroxyl substituent. Moreover, daughter ions at m/z 125 and 109 could serve as diagnostic ions to presume them to be (epi)gallocatechin. Another series of isomers, compounds 22, 27 and 30, could be deduced as (epi)gallocatechin-(epi)gallocatechin, which displayed a [M − H] − ion at m/z 609, corresponding to dimer of (epi)gallocatechin. In analogy, compound 34 was presumed to be (epi)gallocatechin-(epi)gallocatechin-(epi)gallocatechin, a trimer of (epi)gallocatechin, based on precisemass measurements and MS 2 fragmentation patterns.
For compounds 74 and 90, deprotonated molecules at m/z 457 underwent similar fragment pathways as compounds the mentioned above for their common product ions. The base peak ion at m/z 169 (C 7 H 5 O 5 ) in MS 2 spectra suggested the presence of a galloyl moiety in the molecular structure in negative mode. Consequently, they were tentatively characterized as (epi)gallocatechingallate. Similarly, compound 63 was plausibly inferred to be (epi)gallocatechingallate-(epi)gallocatechingallate, a dimer of (epi)gallocatechingallate. The different fragment pathways between compound 63 and 34, as well as their identification were verified in previous reports [10]. Compounds 31 and 44 were assigned as (epi)gallocatechin-(epi)gallocatechingallate, which combined a unit of (epi)gallocatechin (C 15 H 14 O 7 ) and (epi)gallocatechingallate (C 22 H 18 O 11 ) and underwent similar fragmentation pathways as those described in the literature [9,36].

Flavonols
Flavonols were also detected in the extract of Myrica rubra leaf. These compounds produced abundant fragment ions in both positive and negative ion mode, while the ion response of the latter one was better. As a result, we only discuss the fragmentation pathways observed in negative ion mode.
One main structure unit of this type was myricitrin (compound 120) [37], which was explicitly characterized by comparison with a reference sample. The fragment ion at m/z 316 in the MS 2 spectrum was generated by the loss of a rhamnosyl residue (146 Da), and further yielded characteristic ions at m/z 151 and 179 owing to cleavages of the A 1,3 and B 0,3 type, which are often shown in flavonols [33]. Compound 72 exhibited a pseudomolecular ion [M − H] − at m/z 915.16467 (C 42 H 37 O 24 ), whose MS 2 fragmentation information resembled that of compound 120. The molecular formula was a double confirmation to ascribe it to myricitrinyl-myricitrin.
Compounds 134 and 140 gave rise to identical deprotonated molecules at m/z 615, which fragmented to [M − H − galloyl-rhamnoyl] − ion at m/z 371 with high abundance, implying the existence of myricitrin. The product ion at m/z 169 in MS 2 spectrum also indicated the presence of a gallic acid fragment in the structure. According to the chemical constituents reported in [10,38], these two compounds were tentatively deduced to be myricitringallate.
From all the characterization discussed above, it's not difficult to find the composition of flavonoids in the extract of Myrica rubra leaf. The units were gallic acid (C 7 O 19 given by high resolution mass spectra. Their UV spectra shapes were alike, with λ max at 270 nm and they appeared to belong to proanthocyanidins [10]. Typical fragment ions at m/z 125, 151, 169, and 179 were observed in the MS 2 spectra in accordance with what we summarized previously. Thus they were deduced to contain one (epi)gallocatechin and one myricitrin moiety in their molecular structures, and were named as myricitrinyl-(epi)gallocatechin. Another pair of isomers belonging to this type were compounds 93 and 108 according to their consistent fragmentation pathways. By analyzing the molecular formula (C 36 H 28 O 20 ) obtained from HRMS, we inferred the extract comprised a molecule of myricetin and a molecule of myricitrin, although their linkage position remained uncertain. Additionally, some gallates esterified with myricitrinyl-myritrin were also identified in the extract, i.e., compounds 98, 111 and 123. As all these compounds couldn't be found in the Chemical Abstracts database, they may be novel compounds.
The other primary structure unit of this type was quercitrin (compound 139), which displayed a [M − H] − ion at m/z 447 in negative ion mode, resulting in a base peak ion at m/z 300 and another dominant ion at m/z 301 by eliminating a rhamnosyl unit, and the fragment behavior of aglycone was similar to that of myricitrin. Both compounds 146 and 150 exhibited a deprotonated molecular ion at m/z 489, which was 42 Da (C 2 H 2 O) higher than that of compound 139 and their MS 2 spectra were homologous, implying that an acetyl group was linked to the quercitrin, however the position of the linkage was still not determined, so they were provisionally presumed to be acetylquercitrin. Compound 136 was identified as quercitrinyl-(epi)gallocatechin methyl gallate in a similar way.
Moreover, a range of new compounds that haven't been reported previously were discovered by means of permutations and combinations of the aforementioned units. Compound 114 showed a molecular formula of C 41 H 36 O 23 , producing a base peak ion at m/z 301 (C 15 H 10 O 7 ), together with other diagnostic ions at m/z 125, 151 and 179, in its MS 2 spectrum, thus it's speculated that a quercitrin (C 21 H 20 O 11 ) group connected to another (C 20 H 18 O 12 ) unit existed in its molecular structure. By searching the Chemical Abstracts database, C 20 H 18 O 12 may be quercetin-galactoside, another flavonol of the same aglycone. As a result, compound 114 was plausibly characterized as quercitrinyl-quercetin-galactoside. Homoplastically, compounds 104 and 109 were assigned to be consisting of myricitrin and quercetin-arabinoside.

Xanthones
Only compound 119 was identified as a xanthone. It yielded a dominant ion as the base peak at m/z 125 after the breakdown of C1-4 in the MS 2 spectrum, indicating the existence of two substituent groups in the B ring. Other fragment ions at m/z 259, 243 and 223 corresponded to the characteristic losses of CO, CO 2 , and 2 × CH 4 O. Consequently, compound 119 was ascribed to be dimethoxy-dihydroxyxanthone with an as yet unknown linkage form.
The quasi-molecular ion [M + H] + of compound 153 experienced a series of cleavages and rearrangements, and formed fragment ions of moderate abundance at m/z 107 and 123, which were attributed to a methyl substituted phenol residue and a methyl substituted resorcinol residue, respectively. On the other hand, the elimination of a diaryl moiety led to the base peak ion at m/z 131 in the MS 2 spectrum.
In the positive ion mode MS spectrum compound 165 showed a remarkable protonated [M + H] + molecule at m/z 359, and produced a base peak ion at m/z 271 by the loss of a pentanol group. Other fragment ions at m/z 259 and 341 originated from the elimination of an octanol moiety and H 2 O. Moreover, a weaker ion at m/z 211 occurred after the cleavage of the C1-2 bond followed by the loss of a molecule of C 9 H 8 O 2 .

Identification of Amino Acids and Peptides
Amino acids and peptides are widespread in natural products, showing various biological activities such as antimicrobial, anticancer, anti-cardiovascular disease effects and so on. There were four amino acids containing primary amines, five amino acids containing secondary amines, and three peptides detected. The primary amino acids yield protonated [M + H] + molecules and subsequently underwent successive losses of several NH 3 , H 2 O and CO 2 groups. However, compounds 85 and 91, secondary amino acid isomers, preferentially lost a molecule of methacrylic acid (C 4 H 6 O 2 ) to form the base peak ions at m/z 130. Based on the MS n fragment information and high resolution mass spectrometry, they could be tentatively identified as monascumic acid or its isomers. Compounds 76, 83 and 102 showed a pseudomolecular ion [M + H] + at m/z 409 of which the molecular formula was C 16 H 28 O 10 N 2 . They tended to eliminate the C 6 H 13 NO 5 group to afford a fragment ion at m/z 230, ascribed to be N-(1-deoxyfructopyranos-1-yl)isoleucyl aspartic acid or its isomers. For the peptides, their MS 2 fragmentation spectra gave fragment ions by the cleavage of peptide bonds except the neutral loss of NH 3 , H 2 O and CO 2 , which were helpful for the prediction.

Identification of Other Types of Constituents
There were also other types of micro-constituents detected in the extract of Myrica rubra, including naphthoquinones, terpenoids, polysaccharides, steroids, etc. For compounds 37 and 69, preliminary mass spectra performed in TIC mode showed [M − H] − ions at m/z 451 in positive ion mode. Then they generated a series of product ions of glycosyl segments depending on the voltage applied to the source, such as ions at m/z 59, 89, 101, and 119. Another pentyl moiety daughter ion was observed at m/z 71. Finally, their structures were established as glucosyl-octenyl-glucose. However, it was arduous to distinguish the linkage position between the octenyl units and the glucosyl group. In a similar way, compounds 75, 84 and 97 were tentatively assigned as glucosyl-hexanoyl-glucose.
Compounds 64 and 70 showed a pseudomolecular ion [M − H] − at m/z 447 with fragment ions at m/z 59, 89, 101, and 161, which may be due to the existence of a glycosyl group. An additional product ion at m/z 71 corresponded to the successive loss of an oxygen heterocycle and acetic acid. By searching the Chemical Abstracts database, they were deduced to be O-acetylshanzhiside methyl ester or its isomers, which are identified herein in the Myrica genus for the first time. The predominant fragment ions of compounds 73, 81 and 101 were rather similar, except for the molecular weight of 16 Da less, so probably the O-acetyl group was replaced by a propyl. Consequently, the three isomers were tentatively identified as propyl shanzhiside methyl ester or its isomers, although the linkage position still could not be determined, which might be novel compounds.

Effects on H 2 O 2 Induced Cell Death by MTT Assays
To establish the conditions for the research, the effect of H 2 O 2 and serum on N2a cells were investigated. Cells incubated with or without serum medium both displayed a H 2 O 2 -induced anti-cell proliferation at the concentration of 50-100 µM. For the medium supplemented with serum, the concentration of H 2 O 2 did not demonstrate significant difference, and 100 µM was chosen for the latter trail. Cell viability was measured by MTT reduction assay. Results were means ± S.D. from three independent experiments. As shown in Figure 5, exposure of N2a cells to H 2 O 2 100 µM alone sharply reduced the cell viability (approximately 50% of the control group value). The anti-cell proliferation was dramatically attenuated by pretreatment of N2a cells with myricanol at various concentrations without cytotoxicity, and the relative cell viability came to the highest at the concentration of 0.84 mM. Besides, pretreatment with the whole extract of Myrica rubra leaf at the concentration of 20 µg/mL, followed by 100 µM H 2 O 2 for 8 h, it suppressed the reduction in relative cell viability (82.16 ± 4.15%). Pretreatment of N2a cells with myricetrin altered the relative viability (74.60 ± 3.45%) at the highest non-toxic concentration (0.65 mM). The result demonstrated a protective effect of myricanol against H 2 O 2 -induced cytotoxicity.

Effects on H2O2 Altered Cell Morphology
Control cells with good growth were plump and intact. Upon exposure to 100 μM H2O2, N2a cells floated in the culture media and appeared round, with completely altered neuronal outgrowths, indicative of the cytotoxicity of H2O2. In H2O2+myricanol group, there were noticeably fewer damaged and floating cells compared with the H2O2 group, cells showed healthy morphology similar to that of untreated control group ( Figure 6). This confirmed the protective effect of myricanol on H2O2-induced cell damage and was consistent with the MTT assay result.

Effects on H2O2-Induced Intracellular ROS
Oxidative stress is thought to be induced by excess of ROS, which may be byproducts of cellular metabolism and could be quantified by the fluorescent probe 2′,7′-dichlorofluorescin diacetate (DCFH-DA). DCFH can be oxidized to a highly fluorescent DCF, indicating the resultant oxidative stress due to overproduction of ROS [42]. N2a cells exposed to H2O2 alone with myricetrin displayed a significant increase in DCF fluorescence, while when pretreated with myricanol at a concentration of 0.84 mM, the DCF signal was attenuated compared with the H2O2 group (shown in Figure 7). In addition, myricetrin did not exert significant protective effect.

Effects on H 2 O 2 Altered Cell Morphology
Control cells with good growth were plump and intact. Upon exposure to 100 µM H 2 O 2 , N2a cells floated in the culture media and appeared round, with completely altered neuronal outgrowths, indicative of the cytotoxicity of H 2 O 2 . In H 2 O 2 +myricanol group, there were noticeably fewer damaged and floating cells compared with the H 2 O 2 group, cells showed healthy morphology similar to that of untreated control group ( Figure 6). This confirmed the protective effect of myricanol on H 2 O 2 -induced cell damage and was consistent with the MTT assay result.

Effects on H2O2 Altered Cell Morphology
Control cells with good growth were plump and intact. Upon exposure to 100 μM H2O2, N2a cells floated in the culture media and appeared round, with completely altered neuronal outgrowths, indicative of the cytotoxicity of H2O2. In H2O2+myricanol group, there were noticeably fewer damaged and floating cells compared with the H2O2 group, cells showed healthy morphology similar to that of untreated control group ( Figure 6). This confirmed the protective effect of myricanol on H2O2-induced cell damage and was consistent with the MTT assay result.

Effects on H2O2-Induced Intracellular ROS
Oxidative stress is thought to be induced by excess of ROS, which may be byproducts of cellular metabolism and could be quantified by the fluorescent probe 2′,7′-dichlorofluorescin diacetate (DCFH-DA). DCFH can be oxidized to a highly fluorescent DCF, indicating the resultant oxidative stress due to overproduction of ROS [42]. N2a cells exposed to H2O2 alone with myricetrin displayed a significant increase in DCF fluorescence, while when pretreated with myricanol at a concentration of 0.84 mM, the DCF signal was attenuated compared with the H2O2 group (shown in Figure 7). In addition, myricetrin did not exert significant protective effect.

Effects on H 2 O 2 -Induced Intracellular ROS
Oxidative stress is thought to be induced by excess of ROS, which may be byproducts of cellular metabolism and could be quantified by the fluorescent probe 2 ,7 -dichlorofluorescin diacetate (DCFH-DA). DCFH can be oxidized to a highly fluorescent DCF, indicating the resultant oxidative stress due to overproduction of ROS [42]. N2a cells exposed to H 2 O 2 alone with myricetrin displayed a significant increase in DCF fluorescence, while when pretreated with myricanol at a concentration of 0.84 mM, the DCF signal was attenuated compared with the H 2 O 2 group (shown in Figure 7). In addition, myricetrin did not exert significant protective effect.

Effects on H2O2-Induced Intracellular Calcium Concentration
It had been widely reported that calcium invokes many Ca 2+ -dependent enzymes which play a remarkable role in regulating various cellular components, while the excessive entry of Ca 2+ may cause neurotoxicity through cytoplasmic and nuclear processes [43,44]. In the study, we estimated the effect of two compounds on cytosolic Ca 2+ shifts, originating from both extra and intracellular Ca 2+ sources. H2O2 was reported to induce biphasic elevation of [Ca 2+ ]i, resulting in disruption of cytosolic calcium homeostasis [45]. In this study, [Ca 2+ ]i increased to 3045.51 ± 572.69 (p < 0.05) upon cells' exposure to 100 μM H2O2 (Table 1). However, when pretreated with myricanol, no obvious increase in cytosolic Ca 2+ was observed (777.81 ± 23.49). The Ca 2+ shifts were only moderate when pretreated with myricetrin by contrast (1178.92 ± 106.93). This observation further verified the protective effects of myricanol against H2O2-induced cytotoxicity via relieving [Ca 2+ ]i overload in N2a cells.

Discussion
UPLC-PDA-HRMS was used for the identification of chemicals in the extract of Myrica rubra leaf. One hundred and sixteen compounds were firmly or tentatively identified, among which there were 24 pairs of isomers, whose detailed structures could not be determined via mass spectrometry and need the aid of nuclear magnetic resonance. What's more, we only studied the components of the high polarity fraction, leaving the low polar one which may contain abundant constituents and be a complement for the overall chemical profile displayed by Myrica rubra.
Neurodegenerative disorders have been linked to oxidative stress, which may contribute to the generation of ROS [46,47]. H2O2 is often used as neurotoxin to induce oxidative stress and damage in neuronal cells via production of ROS [24] and overload of calcium ions [48]. The cytotoxicity of H2O2 to N2a cells was proved in this study by MTT assays of cell viability and morphological observation (Figures 4 and 5). The cells pretreated myricanol showed significantly increased cell viability, and the cell morphology resembled that of the control groups even after exposure to H2O2, suggesting its protective role in H2O2-induced oxidative stress. The generation of ROS and the shift of [Ca 2+ ]i were

Effects on H 2 O 2 -Induced Intracellular Calcium Concentration
It had been widely reported that calcium invokes many Ca 2+ -dependent enzymes which play a remarkable role in regulating various cellular components, while the excessive entry of Ca 2+ may cause neurotoxicity through cytoplasmic and nuclear processes [43,44]. In the study, we estimated the effect of two compounds on cytosolic Ca 2+ shifts, originating from both extra and intracellular Ca 2+ sources. H 2 O 2 was reported to induce biphasic elevation of [Ca 2+ ]i, resulting in disruption of cytosolic calcium homeostasis [45]. In this study, [Ca 2+ ]i increased to 3045.51 ± 572.69 (p < 0.05) upon cells' exposure to 100 µM H 2 O 2 (Table 1). However, when pretreated with myricanol, no obvious increase in cytosolic Ca 2+ was observed (777.81 ± 23.49). The Ca 2+ shifts were only moderate when pretreated with myricetrin by contrast (1178.92 ± 106.93). This observation further verified the protective effects of myricanol against H 2 O 2 -induced cytotoxicity via relieving [Ca 2+ ]i overload in N2a cells.

Discussion
UPLC-PDA-HRMS was used for the identification of chemicals in the extract of Myrica rubra leaf. One hundred and sixteen compounds were firmly or tentatively identified, among which there were 24 pairs of isomers, whose detailed structures could not be determined via mass spectrometry and need the aid of nuclear magnetic resonance. What's more, we only studied the components of the high polarity fraction, leaving the low polar one which may contain abundant constituents and be a complement for the overall chemical profile displayed by Myrica rubra.
Neurodegenerative disorders have been linked to oxidative stress, which may contribute to the generation of ROS [46,47]. H 2 O 2 is often used as neurotoxin to induce oxidative stress and damage in neuronal cells via production of ROS [24] and overload of calcium ions [48]. The cytotoxicity of H 2 O 2 to N2a cells was proved in this study by MTT assays of cell viability and morphological observation (Figures 4 and 5). The cells pretreated myricanol showed significantly increased cell viability, and the cell morphology resembled that of the control groups even after exposure to H 2 O 2 , suggesting its protective role in H 2 O 2 -induced oxidative stress. The generation of ROS and the shift of [Ca 2+ ]i were determined by the fluorescent DCFHDA probe and Fura 2-AM probe, respectively. The increase of ROS induced by H 2 O 2 was markedly attenuated by the myricanol treatment. Little elevation of [Ca 2+ ]i was observed when pretreated with myricanol compared with H 2 O 2 group.
The investigation clearly revealed that myricanol inhibited H 2 O 2 -induced neuronal death, ROS generation and [Ca 2+ ]i overload. Myricanol is a typical cyclic diarylheptanoid containing two phenolic hydroxyl groups and one alcoholic hydroxyl group, which may lead to powerful ROS scavenging properties and neuroprotective activities. Myricitrin is a typical flavonoid, also containing numerous phenolic hydroxyl groups, however, it demonstrated insignificant pharmacological activity. This was only a preliminary investigation of the neuroprotective effects of myricanol. Relevant aspects (cell viability, ROS production and intracellular Ca 2+ ) leading to the stress condition merit further research, as well as the difference of the structure and neuroprotective mechanism between these compounds. This work may also provide experimental platform for further research in human cell lines to test the neuroprotective activity.
High resolution MS analysis was recorded on a Q Exactive quadrupole Orbitrap mass spectrometer (Thermo, Bremen, Germany) coupled with an ESI source in both negative and positive mode. Data were acquired by Higher Energy Collision Induced Dissociation. The full scan-ddMS2 mode was applied with the optimized MS parameters set as: spray voltage 2.5 kV for ESI − and 3.5 kV for ESI + ; sheath gas 49 arb; Aux gas 12 arb; probe heater temperature 420 • C; capillary temperature 260 • C; S-lens RF level 50; scan rang 100-1500 Da; resolution 70 k FWHM (at m/z 200); HCD fragmentation energy 30%. MS data collected were processed utilizing Thermo Scientific TM Xcalibur TM platform (San Jose, CA, USA).
For evaluation of neuroprotective activity, 1 mg powder of myricanol and myricitrin were dissolved in DMSO to a concentration of 50 mg/mL. Fifty mg of freeze-dried powder of the whole extract was dissolved in 1 mL DMSO.

Cell Culture
N2a cells were maintained in high-glucose DMEM containing 10% FBS and 2 mM L-glutamine in an atmosphere of 95% relative humidity (5% CO 2 ) at 37 • C. The media was changed every other day.

Analysis of Cell Viability by MTT Assay
Cells seeded into 96-well microplate (2×10 4 cells/well) were challenged with 100 µM H 2 O 2 to induce oxidative stress. For administration groups, cells were pretreated with various non-cytotoxic concentrations of myricanol and myricetrin for 12 h (not withdraw before adding H 2 O 2 ). After 8 h's incubation with H 2 O 2 , 10 µL of 5% MTT was added into the medium. Following 2-4 h incubation, the medium was discarded and replaced by 10 mL acidified isopropyl alcohol. Then cell viability was evaluated in an ELISA reader (INFINITE M1000PRO, Tecan US, Morrisville, NC, USA) by measuring the optical densities at 570 nm. For the control group, cell cultures were incubated with the same culture medium of equal volume. The culture medium containing 10 µL of 5% MTT was adopted as the blank group. The viability (%) was calculated with the following formula: According to aforementioned procedure of Section 3.4.2, N2a cells were cultured in 96-well microplate (2 × 10 4 cells/well) and exposed to 100 µM H 2 O 2 for 8 h. For the trial, myricanol and myricetrin isolated from Myrica rubra were added in advance to assess their neuroprotective effect. The observation and visualization of cellular morphology were conducted by an inverted microscope (Leica, Malvern, PA, USA) equipped with a DF450C system.

Measurement of ROS Production
According to the aforementioned procedure described in Section 3.4.2, N2a cells were cultured in 96-well microplate (2 × 10 4 cells/well) and exposed to 100 µM H 2 O 2 for 8 h. For the trial, N2a cells were pretreated with myricanol and myricetrin for 12 h, followed by treatment with 10 µM DCFH-DA for 30 min. The solution was discarded and washed with PBS for another three times. 100 µL media was added into the microplate. The generation of ROS was determined by fluorimetric detection with an ELISA reader (INFINITE M1000PRO, Tecan US, Morrisville, NC, USA) at an excitation-emission wavelength of 485-538 nm respectively. Alteration of ROS was calculated by the DCF fluorescence intensity in contrast with the control group.

Measurement of Intracellular Calcium Concentration
According to the procedure described in Section 3.4.2, N2a cells were cultured in 96-well microplates (2 × 10 4 cells/well) and exposed to 100 µM H 2 O 2 for 8 h. For the trial, N2a cells were pretreated with myricanol and myricetrin for 12 h. The media were discarded, and washed with PBS for twice, followed by treatment with 5 µM Fura-2-AM for 30 min at 37 • C. The solution was discarded and washed with PBS for another three times. 100 µL media was added into the microplate and the concentration of [Ca 2+ ]i was detected with a ELISA reader (INFINITE M1000PRO, Tecan US, Morrisville, NC, USA). Triton-X 100 was used for disruption of the cell membrane, and EGTA was used for the complexation of calcium. On binding Ca 2+ , the excitation wavelength shifted to 340 and 380 nm, while the emission wavelength remained at 510 nm. The concentration of calcium was calculated with the following formula: [Ca 2+ ]i = K d (F 0 −F min )/(F max −F 0 ), where K d (assuming it as 224 nM) was the dissociation constant of the chemical reaction for Ca 2+ buffering by the fluorescent dye. F 0 , F max and F min stand for the fluorescence measured without adding Triton-X 100 or EGTA, with adding 0.1% Triton-X 100 and by quenching Fluo-3 fluorescence with 5 mM EGTA, respectively.

Statistical Analysis
Except for where stated otherwise, all results were expressed as the mean ± standard deviation (SD) of the indicated measurements of quintuplicate experiment. The significance of differences was determined by Student's t-test, and a p value less than 0.05 was considered statistically significant. All data were analyzed using InStat3 or Prism software 5.0b (GraphPad Software, San Diego, CA, USA).

Conclusions
In this research, UPLC coupled with Q Exactive quadrupole Orbitrap mass spectrometry was applied to identify the chemical constituents in the extract of Myrica rubra leaf. The chemical profile was comprehensively and systematically studied for the first time. By comparison with authentic substances, integration of exact mass, UV spectra, and fragment information, 116 compounds were confirmed or tentatively identified, including 26 organic acids and their derivatives, 36 flavonoids, two cyclic diarylheptanoids, 12 amino acids and peptides, together with 40 other types of compounds. Each category was analyzed in detail to summarize and conclude available fragmentation rules in HRMS. In particular, a convenient and effective strategy was proposed for the rapid characterization of flavonoids through permutation and combination of molecular formula, which were comprised of specific units, i.e., galloyl, (epi)catechin, (epi)gallocatechin, myricetin, myricitrin, and quercitrin.
The neuroprotective activities of two representative components, myricetrin and myricanol (one flavonoid and one cyclic diarylheptanoid, respectively) were evaluated onH 2 O 2 -inducedN2a cells by MTT assays. The results revealed that a significant cell death trigged by H 2 O 2 was neutralized by myricanol, while myricetrin only had moderate effect. Further confirmation was conducted by intracellular ROS and calcium ion assays, for the decrease in ROS production and Ca 2+ shifts were both observed. In summary, myricanol might offer a promising the rapeutic strategy to reduce the neurotoxicity of reactive dicarbonyl compounds, providing a potential benefit agent with age-related neurodegenerative diseases.