Laguncularia racemosa Phenolics Profiling by Three-Phase Solvent System Step-Gradient Using High-Performance Countercurrent Chromatography with Off-Line Electrospray Mass-Spectrometry Detection

The detailed metabolite profiling of Laguncularia racemosa was accomplished by high-performance countercurrent chromatography (HPCCC) using the three-phase system n-hexane–tert-butyl methyl ether–acetonitrile–water 2:3:3:2 (v/v/v/v) in step-gradient elution mode. The gradient elution was adjusted to the chemical complexity of the L. racemosa ethyl acetate partition and strongly improved the polarity range of chromatography. The three-phase solvent system was chosen for the gradient to avoid equilibrium problems when changing mobile phase compositions encountered between the gradient steps. The tentative recognition of metabolites including the identification of novel ones was possible due to the off-line injection of fractions to electrospray ionization mass spectrometry (ESI-MS/MS) in the sequence of recovery. The off-line hyphenation profiling experiment of HPCCC and ESI-MS projected the preparative elution by selected single ion traces in the negative ionization mode. Co-elution effects were monitored and MS/MS fragmentation data of more than 100 substances were used for structural characterization and identification. The metabolite profile in the L. racemosa extract comprised flavonoids, hydrolysable tannins, condensed tannins and low molecular weight polyphenols.


Introduction
Countercurrent chromatography (CCC) is an all-liquid method, with no solid support, in which the stationary liquid phase is retained in the apparatus using centrifugal force only [1]. The principle behind this technique underlies the partitioning of a sample in a biphasic liquid solvent system [2]. Among many advantages, the technique is highly versatile; has high loading capacity; is easy to scale up; and eliminates sample loss by chemical degradation and irreversible adsorption [1,3,4]. CCC is a powerful tool in the phytochemical working field as it enables plant extract fractionation with existing major compounds and also isolation or fortification of minor compounds [5,6]. This characteristic is even more noticeable when semi-preparative and preparative scales are employed [7,8]. Standard CCC separation experiments do not provide the full flexibility, solely operating with a two-phase solvent system, and the isocratic elution-extrusion mode [9,10].
To improve the polarity range in the CCC operation field, three-phase solvent systems were developed/applied, due to the great differences in polarities between the upper, middle and lower phases. Recently, some applications have been reported but only half of them actually were used as three phases in the separation process [11][12][13][14][15]. Tri-phasic systems are built from a two-phase system normally composed of n-hexane, acetonitrile and water in combination with a fourth solvent such as methyl acetate, ethyl acetate, methyl tert-butyl-methyl-ether or dichloromethane to create the third phase. Very few tri-phasic systems are described due to limited solvent combinations that form these three stable phases in a convenient volume percentage [11].
In analogy to solid phase chromatography, gradient elution in CCC intends to shorten the duration of a separation process and may also improve resolution. A common way to perform gradient elution is to change the mobile phase polarity over time [16], although gradient elution mode in CCC is less frequently used as the biphasic liquid system is in the equilibrium state and the change of composition in one phase corresponds directly to a change in the respective other liquid phase [17]. However, if a three-phase solvent system is used for gradient elution purposes, all phases involved for the experiment were previously in contact. Disturbance of the equilibrium and collapse of phase layers are omitted during the separation process while maintaining the broad polarity range for the recovery process.
In this work, a three-phase solvent system in step-gradient elution mode high-performance countercurrent chromatography (HPCCC) with off-line ESI-MS/MS detection was used for metabolite profiling of the ethyl acetate partition of the leaves from the mangrove plant Laguncularia racemosa. A similar approach was used on Anogeissus leiocarpus for compound identification but using centrifugal partition extraction (CPE), and off-line NMR detection [15]. L. racemosa (Combretaceae), popularly known as white mangrove, is the only occurring specie in the genus [18], and is considered as a strict mangrove [19], characteristic for growing in brackish coastal environments [20], and with excellent function for stabilizing shorelines against erosion [21]. From aspects of ethno-medicinal use, the plant is applied as astringent and tonic for dysentery and fever [22]. To date, there are only a few studies on the production of secondary metabolites [23][24][25][26][27], probably due to its high complexity of polar natural products.

Results and Discussion
The composition of L. racemosa ethyl acetate solvent partition (EtOAcPart) was initially investigated by TLC and LC-ESI/TOFMS analysis (Supplementary Figure S1). The liquid mass-spectrometry profile showed a high chemical complexity containing metabolites in a larger polarity range, making this mixture an ideal case study for the application of three-phase gradient elution. Some multiple-solvent biphasic systems have been proposed to great extend the polarity in CCC [28]. However, as the phases are in steady mixing contact during the complete separation process, a change of the mobile phase composition during gradient elution directly influences the stationary phase composition as well, and as consequence, disturbs the equilibrium and could lower or even lead to low chromatographic resolution [29,30]. To circumvent the equilibrium obstacle during the gradient elution procedure, a three-phase solvent system was used instead of two (or more) different biphasic systems. In this approach all phases involved in the separation were previously saturated with each other. The tri-phasic system n-hexane-tert-butyl-methyl ether-acetonitrile-water 2:3:3:2 (v/v/v/v) was used in the semi-preparative purification of L. racemosa EtOAcPart. The L. racemosa EtOAcPart was separated by semi-preparative HPCCC chromatography, and the off-line injection profiling by injections of recovered fractions to ESI-MS/MS distinguished 17 principal phenolic constituents ( Figure 1). However, as a result of the highly concentrated injections of respective HPCCC fractions, the selected single ion-based projection of the HPCCC experiment revealed more than 100 different metabolites (1-109) (Partially shown in Figure 1 and Supplementary Figure S2). Preliminary LC-ESI/TOF MS analysis was not capable of detecting all minor compounds due to concentration levels below the detection limits. Figure 1. Selected electrospray ionization mass spectrometry electrospray ionization mass spectrometry ions traces (negative mode) of phenolics of L. racemosa EtOAcPart detected in the off-line injected high-performance countercurrent chromatography (HPCCC) fractions. HPCCC separation using n-hexane-tert-butyl-methyl ether-acetonitrile-water 2:3:3:2 (v/v/v/v) as triphasic solvent system.
Large advantage of the off-line injection profiling methodology of preparative HPCCCfractions by an ESI-MS detector in the sequence of recovery is the 'on-the-fly' delivery of the respective molecular weight-and MS/MS-fragmentation data of all ionizeable compounds in one step. This is a very fast process to get the required data for immediate compound identifications in the respective HPCCC-fractions. A full mass-spectrometry guided metabolite profile with more than ten automatically selected precuresur ions for MS/MS on a larger lab-scale preparative HPCCC fractionation can be achieved in a 60 to 100 min experimental mass-spectrometry time frame. This mass-spectromtry approach is roughly by a factor of hundred faster than the single investigation of resepctive HPCCCC fractions by LC-ESI-MS/MS analysis. The results displayed in a single data file are ready to use and not mutiple analysis sets need to be compared for guiding the decisions in fractionation process. This powerful approach was previously applied by Costa et al. [8] on the complex metabolite mixture extracted from the Brazilian plant Salicornia gaudichiana.
In case of the investigated L. racemosa ethylacetate solvent partition, the elution ranges of a large selection of higher and lower concentrated target molecules (Table 1) were visualized in the recovered HPCCC-fractions by selected single ion traces for performing the accurate fractionation, recovery and preventing unintentional mixing of already separated compounds. Additionally, the existing compound co-elution effects, and the sequential elution orders of separated isobars/ isomers were clearly detected and visualized ( Figure 1).    Table 1) co-eluted in this HPCCC run as seen in Figure 1, as well as the penta-galloyl-hexoside (85) ( Table 1) Table 1). Although the constitution of certain compounds had been different, the polarity differences were not sufficient for a successful HPCCC separation as seen for the selected ion traces [M − H] − m/z 585 (quercetin pentoside gallate, 32), and m/z 599 (quercetin desoxyhexoside gallate, 33) (Figure 1).

Cpd CCC-Fraction MS [M − H] − (m/z) MS/MS [M − H] − (m/z) LC-RT(min) ESI/TOF MS Formula (Error in ppm)
Using literature to guide the identification process of the minor, and very minor concentrated derivatives, literature was verified and the few previously isolated compounds in L. racemosa were listed with molecular weights as a comparative database. Most of the unknown compounds were characterized by ESI-MS/MS fragmentation and indicative neutral loss pattern. High accuracy molecular weights acquired by LC-ESI/TOF MS were used to ratify and/or verify the proposed molecular formulas. Phytochemical investigations describing chemical compounds on other genus of Combretaceae helped to support the results based on chemotaxonomic knowledge. From the aspect of natural product classes, the chemical composition of the EtOAcPart was distinguished in four main groups as flavonoids, hydrolysable tannins, condensed tannins and other low molecular weight polyphenols (Supplementary Figure S2). The chemical structures and substitution patterns of fractionated and identified compounds are shown in Figure 2. Fractions had been combined on the basis of TLC analysis and the electrospray massspectrometry profiling experiment. Supplementary Figure S3 displays the TLC-analysis on the combined fractions of the HPCCC experiment. Table 1 lists HPCCC chromatographic elution, and ESI-MS/MS informations; LC ESI-TOF-MS data (when present) and tentative identification. Although L. racemosa EtOAcPart showed quite complex constituents, most phenolic compounds were well separated.

Flavonoids and Derivatives
Flavonoid derivatives were detected and identified in L. racemosa EtOAcPart by ESI-MS/MS as principal compounds in the recovered HPCCC fractions ( Table 1, compounds  1-56). Flavonoids including flavonols, flavones, flavanols and flavanones were found in free form, linked to one sugar unit, as well as in the presence of galloyl substituents. The tentative identification of the flavonoid-aglyca (compounds 1-13) was done by comparison to specific fragmentation patterns, as the spectra of this flavonoid often displayed loss of small neutral fragments contributing to structure information [31][32][33]. The free flavonoid aglyca eluted during the first step of the gradient before the glycoside linked flavonoids, in accordance to mobile phase/compound polarity in the tail-to-head mode.

Low Molecular Weight Polyphenols
Other compounds were recognized and characterized based on precursors/derivatives of existing identified compounds in the off-line ESI-MS/MS profile or on the L. racemosa chemical database. Simple phenolic compounds included catechol (93) and pyrogallol (95), common occurring products in the hydrolysable tannins pathway [35]. Benzoic acid derivatives with frequent [M − H − 44] − corresponding to the neutral loss of CO 2 , comprised protocatechuic (96) and vanillic (97) acids [37]. The amino derivatives aminocatechol (94) and amino protocatechuic acid (98) were also detected [52]. The chromone detected at m/z 193, was identified as trihydroxy-chromone (99) and had its molecular formula confirmed by HRMS.
The jasmonic acid (100) and its sulphated derivative 5 -hydroxy-sulphonyloxy jasmonic acid (105), earlier isolated from the L. racemosa twigs and leaves [23], could be found at m/z 209 and 305, respectively. ESI/TOF MS data confirmed the proposed compounds. Ordinary oleic and linoleic fatty acids (102-104), jasmonic acid biosynthetic precursor, were further encountered. Another sulphated derivative isolated from L. racemosa leaves [26] was found at [M − H] − at m/z 707 and was identified as integracin D (107) [26]. Due to concentration limits, the compound could not be detected in the ESI/TOF MS analysis and structure was not fully confirmed.

Chemical Reagents and Solvents
Preparation of extracts was carried out with analytical grade solvents from Tedia Brazil (Rio de Janeiro, Brazil). LC-ESI/TOF-MS/MS analyses used HPLC grade solvents from Tedia Brazil (Rio de Janeiro, Brazil). HPCCC separations were performed with analytical grade solvents from Fisher Chemicals (Loughborough, UK). ESI-MS/MS analyses were done with HPLC grade solvents from VWR Chemicals (Radnor, PA, USA). NMR analyses used deuterated solvents from Cambridge Isotope Laboratories (Tweksbury, MA, USA) and TMS as internal standard. All aqueous solutions were prepared with pure water produced by Milli-Q water (18.2 MΩ) system (Thame, UK).

Preparation of the Extract
Laguncularia racemosa (3 kg) was collected at Guaratiba Biological and Anthropological Reserve (Rio de Janeiro, Brazil) in November 2010. Specialist researchers from the Nucleus of Mangrove Studies (University of the State of Rio de Janeiro) helped in the localization, identification and collection of the plant. The leaves were dried and grounded in a laboratory mill (Laboratory Retsch mill, Haan, Germany) and 1800 g were submitted to maceration with ethanol-water 8:2 (v/v) in 10 cycles of 24 h. The solvent was evaporated under reduced pressure at 50 • C and the crude extract (255 g) was partitioned between water and organic solvents, affording different extracts: n-hexane (4 g), dichloromethane (8 g), ethyl acetate (15 g) and aqueous (215 g).

LC-ESI/TOF MS Preliminary Analysis
The EtOAcPart was also analysed by LC-ESI/TOF-MS with a 1200 Series LC-chromatograph (Agilent, Palo Alto, CA, USA) coupled with a MicrOTOF II time-of-flight mass spectrometer (Bruker Daltonics, Inc., Billerica, MA, USA). 5 µL injection was performed with an autosampler on a Poroshell EC-C18 column (100 × 2.1 mm; 2.7 µm, Agilent, Palo Alto, CA, USA). The source temperature was set at 200 • C, the drying gas (nitrogen) flow rate was 10.0 L/min and the nebulizer gas (nitrogen) pressure was 4 bar. Data were acquired in negative mode in the range of m/z 100-1500. The capillary voltage was 3.8 kV, the capillary exit voltage was −150 V, the skimmer 1 and 2 voltages were 50 V and 23 V, respectively, the hexapole 1 voltage was set to −23 V, the hexapole RF voltage was 120 Vpp, lens 1 transfer was 68 µs and lens 1 pre plus stage was 7 µs. Mass calibration was achieved by infusing ammonium formate in an isopropanol-water mixture (1:1, v/v) as an external standard. All data were analysed using Bruker Daltonics ESI Compass Data Analysis Version 4.0 SP 1 (Bruker Daltonics Inc., Billerica, MA, USA). The mobile phase consisted of spectroscopic grade methanol (B) and ultrapure water (A) containing 0.05% (v/v) formic acid. The linear gradient elution was set from 10% to 100% of B in 90 min at a flow rate of 0.3 mL/min.

Equipment
CCC separations were performed on a semi-preparative HPCCC system (model Spectrum, Dynamic Extractions Ltd., Gwent, UK) equipped with two counter-balanced bobbins with perfluoroalkoxypolymer (PFA) tubing (1.6 mm i.d.) wound in multi-layer coiledcolumns, resulting in 143.5 mL total volume (V C ). The rotation speed was adjusted to the maximum velocity of 1600 rpm (240 g). Solvent phase systems were delivered by a constant flow pump (Agilent HP1200, Palo Alto, CA, USA) to the HPCCC system. A semi-preparative sample loop (7.15 mL) was used to inject the dissolved sample over a low-pressure valve (Upchurch Model V-450, with 1.6 mm i.d. fittings) to the chromatographic system. Fractions were collected by a fraction collector (Agilent HP1200, Palo Alto, CA, USA).

Three-Phase Solvent System Test Evaluation
The three-phase solvent systems were composed of n-hexane-methyl acetate -acetonitrilewater and n-hexane-tert-butyl methyl ether-acetonitrile-water [11,53,54]. For the experiments for solvent system evaluation, 2 mg of the EtOAcPart were dissolved in a test tube containing 2 mL of each phase of the thoroughly equilibrated solvent systems. The test tubes were shaken vigorously for compound partition. After the phase layers had completely separated and distribution equilibrium was established, the resulting phase layers were analyzed by TLC (Supplementary Figure S4).

Solvent System and Sample Preparation
The selected solvent system n-hexane-tert-butyl methyl ether-acetonitrile-water (2:3:3:2, v/v/v/v) was thoroughly equilibrated in a separatory funnel at room temperature. The three phases were separated shortly before use and degassed by ultra-sonication for 5 min. The sample solution was prepared by dissolving the sample at fixed concentration (100 mg/mL) and coil-volume (5% V C ) in the lower aqueous phase only.

HPCCC Separation Procedure
Separation was performed in a normal step-gradient elution mode. The more aqueous lower phase was used as the stationary phase while organic upper and middle phases were used as mobile phases as shown in Figure 3. The system was completely filled with the lower aqueous stationary phase. Rotation was set to 1600 rpm. For the separation, the upper organic mobile phase was pumped at 4.0 mL/min. After reaching hydrodynamic equilibrium, the sample was injected to the HPCCC column. For the first elution step, 214.5 mL mobile phase (1.5 V C ) of upper phase was pumped through. For the second elution step, 2 V C (286 mL) of the middle phase was pumped through the HPCCC system. Fractions were collected at 1 min intervals. For the extrusion step, rotation was reduced to 200 rpm and the column contents were pushed out of the system by lower phase at 8.0 mL/min and fractions were collected at 30 s intervals. The temperature control was maintained at 30 • C.

Metabolite Profiling by Offline Injections to ESI-MS/MS
The molecular weight profiles of the recovered HPCCC fractions were monitored by off-line ESI-MS/MS and were recorded in a single data file using the ion-trap massspectrometer HCT-Ultra ETD II (Bruker Daltonics, Bremen, Germany). Aliquots of 0.75 µL of odd numbered CCC fractions were directly filled to vials, dried and redissolved in 1.0 mL of methanol for conducting the ESI-MS analysis. Fractions were delivered to the ESI-MS/MS by a HPLC-pump (binary pump, G1312 A, 1100 Series, Agilent, Waldbronn, Germany) using the make-up solvent system with a flow rate of 0.5 mL/min composed of acetonitrile and water (1:1, v/v). ESI-MS/MS parameter settings were in the negative ionization mode, with scan-range between m/z 100-2000, where mostly deprotonated [M − H] − ion signals were generated. An auto-MS/MS method fragmented the nine most intense peaks and to monitor and characterize co-eluting compounds. Drying gas was nitrogen (flow rate 10.0 L/min, 310 • C), and nebulizer pressure was set to 60 psi. Ionization voltage at HV capillary was 3500 V, HV end plate off set −500 V, trap drive 61.8, octupole RF amplitude 187.1 Vpp, lens 2 60.0 V, Cap Ex −115.0 V, max. accumulation time 200 ms, averages 5 spectra, trap drive level 120%, target mass range: m/z 500, compound stability 80%, Smart ICC target 70.000, ICC charge control 'on' and smart parameter setting 'active'.

Conclusions
The combination of analysis of preparative HPCCC fractions with off-line injections to an ESI-MS/MS device was proven to be highly effective for a full metabolite chemical profile for polyphenols using the negative ionization mode. The use of a three-phase solvent system for HPCCC in a step-gradient elution mode was adequate to maintain equilibrium and chromatographic resolution while improving mobile phase strength. The ESI-MS/MS projection of the semi-preparative HPCCC experiment visualized over 100 compounds by selected single ion traces and was an adequate confirmation of the LC-ESI/TOF MS analysis. This study detected a variety of metabolites from different classes occurring in L. racemosa EtOAcPart and used chemotaxonomic data to guide the MS/MS putative structure elucidation.
Supplementary Materials: The following are available online, Figure S1: EtOAcPart preliminary analyses by LC-ESI-TOF-MS. The mobile phase was methanol (B) and water (A) containing 0.05% (v/v) formic acid. The linear gradient elution was set from 10% to 100% of B in 90 min. Figure  Author Contributions: F.d.N.C. is responsible for the idea, execution and draft preparation; S.I. is responsible for supervision, funding administration and paper reviewing; G.J. is responsible for MS execution, MS data analysis and paper reviewing; P.H. is responsible for CCC execution and paper reviewing including English improvement; F.d.S.F. is responsible for MS data analysis, bibliography review and paper editing. All authors have read and agreed to the published version of the manuscript.