High-performance Countercurrent Chromatography to Access Rhodiola rosea Influenza Virus Inhibiting Constituents

Abstract In a cytopathic effect inhibition assay, a standardized Rhodiola rosea root and rhizome extract, also known as roseroot extract (SHR-5), exerted distinct anti-influenza A virus activity against HK/68 (H3N2) (IC50 of 2.8 µg/mL) without being cytotoxic. For fast and efficient isolation and identification of the extractʼs bioactive constituents, a high-performance countercurrent chromatographic separation method was developed. It resulted in a three-stage gradient elution program using a mobile phase solvent system composed of ethyl acetate/n-butanol/water (1 : 4 : 5 → 2 : 3 : 5 → 3 : 2 : 5) in the reversed-phase mode. The elaborated high-performance countercurrent chromatographic method allowed for fractionation of the complex roseroot extract in a single chromatographic step in a way that only one additional orthogonal isolation/purification step per fraction yielded 12 isolated constituents. They cover a broad polarity range and belong to different structural classes, namely, the phenylethanoid tyrosol and its glucoside salidroside, the cinnamyl alcohol glycosides rosavin, rosarin, and rosin as well as gallic acid, the cyanogenic glucoside lotaustralin, the monoterpene glucosides rosiridin and kenposide A, and the flavonoids tricin, tricin-5-O-β-D-glucopyranoside, and rhodiosin. The most promising anti-influenza activities were determined for rhodiosin, tricin, and tricin-5-O-β-D-glucopyranoside with IC50 values of 7.9, 13, and 15 µM, respectively. The herein established high-performance countercurrent chromatographic protocol enables fast and scalable access to major as well as minor roseroot constituents. This is of particular relevance for extract standardization, quality control, and further in-depth pharmacological investigations of the metabolites of this popular traditional herbal remedy.


ABSTR AC T
In a cytopathic effect inhibition assay, a standardized Rhodiola rosea root and rhizome extract, also known as roseroot extract , exerted distinct anti-influenza A virus activity against HK/68 (H3N2) (IC 50 of 2.8 µg/mL) without being cytotoxic. For fast and efficient isolation and identification of the ex-tractʼs bioactive constituents, a high-performance countercurrent chromatographic separation method was developed. It resulted in a three-stage gradient elution program using a mobile phase solvent system composed of ethyl acetate/nbutanol/water (1 : 4 : 5 → 2 : 3 : 5 → 3 : 2 : 5) in the reversedphase mode. The elaborated high-performance countercurrent chromatographic method allowed for fractionation of the complex roseroot extract in a single chromatographic step in a way that only one additional orthogonal isolation/purification step per fraction yielded 12 isolated constituents. They cover a broad polarity range and belong to different structural classes, namely, the phenylethanoid tyrosol and its glucoside salidroside, the cinnamyl alcohol glycosides rosavin, rosarin, and rosin as well as gallic acid, the cyanogenic glucoside lotaustralin, the monoterpene glucosides rosiridin and kenposide A, and the flavonoids tricin, tricin-5-O-β-D-glucopyranoside, and rhodiosin. The most promising anti-influenza activities were determined for rhodiosin, tricin, and tricin-5-O-β-D-glucopyranoside with IC 50 values of 7.9, 13, and 15 µM, respectively. The herein established high-performance countercurrent chromatographic protocol enables fast and scalable access to major as well as minor roseroot constituents. This is of particular relevance for extract standardization, quality control, and further in-depth pharmacological investigations of the metabolites of this popular traditional herbal remedy.
The HMPC provides standardization protocols to guarantee the content of one or more marker compounds or bioactive compounds within the European regulatory network [10]. In 2011, the ethanolic extract (67-70 %) of roots and rhizomes of R. rosea was monographed by the HMPC as a traditional herbal medicinal product "for temporary relief of symptoms of stress, such as fatigue and sensation of weakness" (EMA/HMPC/232091/2011). During the past decades, the pharmacological activities of this plant extract have been investigated intensively [3,11,12].
Secondary metabolites from different compound classes, namely, phenylethanoids, phenylpropanoids, monoterpenoids, cyanogenic compounds, flavonoids, proanthocyanidines, and tannins, are described as roseroot constituents [3]. In 1970, salidroside (3), belonging to the structure class of phenylpropanoids, was believed to be the main active principle in R. rosea extracts and was therefore closely investigated, not only for adaptogenic effects [3,13] but also for antiviral activities, i.e., inhibition of dengue virus via inhibiting viral protein synthesis and boosting host immunity [8] and anti-coxsackie virus B3 activity to treat viral myocarditis in vitro and in vivo [9]. Concerning influenza A virus, R. rosea, as part of a traditional Chinese medicine composition, was reported to inhibit various influenza virus strains [5]. Commonly, flavonoids are known to exert anti-influenza activity, as recently reviewed [14], which also was confirmed for flavonoids and its glycosides in roseroot [6,15]. However, as of now, the anti-influenza activity of the widely used, standardized hydroalcoholic extract has not been reported.
Influenza viruses account for the majority of acute respiratory tract infections, which remain the most deadly communicable disease, causing 3 million deaths worldwide in 2016 [16]. Moreover, in light of the current SARS-CoV-2 pandemic, the challenge to discover effective antivirals against acute respiratory infections has reached another level of urgency. Here, nature as source for potent drug candidates could play a major role [17]. Despite the implementation of an influenza vaccination and perpetual development of new antivirals, morbidity and mortality rates are still high. Moreover, data on the potential therapeutic effects of R. rosea extract against viruses such as rhinoviruses or bacteria like Staphylococcus aureus and Streptococcus pneumoniae that contribute to the annual burden of acute respiratory infections are lacking [18][19][20].
The diverse chemical character of roseroot constituents requires a universal but scalable method for separation and purification. HPCCC is a separation method taking advantage of the distribution of analytes in two immiscible liquids. The avoidance of irreversible adsorption to solid-phase material renders HPCCC an optimal method for preparative purposes [21], especially with the aim to gain access to the manifoldness of bioactive secondary metabolites of roseroot. In a conventional setup, the isocratic elution mode is applied in HPCCC. However, for the separation of compounds of different polarity, gradient elution can achieve the required flexibility [22]. A few studies describing the separation of R. rosea constituents with countercurrent chromatography have already been conducted. However, these studies focused on the isolation of only selected compound classes [23][24][25][26]. To the best of our knowledge, this is the first report of an HPCCC protocol to target a broad range of substance classes, including major and minor constituents, using an optimized gradient elution HPCCC method.

Results
In order to achieve an efficient HPCCC separation with high resolution of constituents of the R. rosea extract, 11 solvent systems were evaluated on a small scale in test tubes. A solvent system is considered suitable when K values of the constituents to be isolated range in the area of 0.5-2 [21]. K values are expressed as the ratio of the concentration of an analyte in the stationary phase to its concentration in the mobile phase (Fig. 1S, Supporting Information). Taking this into consideration, the RP mode of the HEMWat system 4 composed of a solvent mixture of ethyl acetate, methanol, and water (3 : 2 : 5) turned out to result in the best K values for the majority of compounds (▶ Tables 1 and 2).
An isocratic elution in both modes (RP and NP) with HEMWat system 4 revealed the RP mode as more suitable based on predicted and experimentally verified K values. As the compounds in R. rosea extract cover a broad range of polarity, an improvement of the isocratic distribution of compounds via introduction of a stepwise gradient elution was performed. The upper phase of HEMWat 4 remained as a stationary phase whereas for the starting mobile phase, the more polar system HEMWat 3 (lower ethyl acetate content in favor of methanol) was selected. Approaching the polarity of the stationary phase, a stepwise gradient elution with HEMWat 4 was continued. A trial to separate more hydrophobic compounds by consecutive elution with the mobile phase of HEMWat 5 or even 6 did not result in an improved separation of the nonpolar compounds.
With special focus on a high-resolution separation of major compounds, i.e., phenylalkanoids, which are of a rather hydrophilic character, the last optimizing step consisted of proceeding with the HEMWat system 2 as the mobile phase and HEMWat 4 as the stationary phase throughout the whole separation process (S F = 60 %). This stepwise introduced gradient accordingly ranged from the mobile phase of HEMWat 2 → 3 → 4 (exact procedure described in the Material and Methods section) and was revealed as the best suited elution program. The final optimized HPCCC fractionation protocol resulted in a three-stage gradient program, as described in ▶ Table 3, with a total duration of 135 min. After application of the R. rosea extract, the contents of collected test tubes were combined according to their similarity of TLC fingerprints, yielding 13 fractions (FI-FXIII). Seven fractions (FIII, FV, FVI, FVIII, FIX, FX, and FXIII) were selected for further purification due to their promising TLC pattern and sufficient yield (Table 1S, Supporting Information).
▶ Table 1 Partition coefficients (K) of identified compounds in the reversed-phase mode. The solvent system as the starting point for method optimization is highlighted by bold letters. Solvent ratios are given in v/v/v/v/v.

Reversed phase
HEMWat system no.
HEMWat solvent ratio a Compound no. The isolated compounds are annotated in the chromatogram of the extract in ▶ Fig. 2 and Fig. 2S, Supporting Information. To obtain a better separation, the BEH C 18 column using an extended run time was developed for purity control, as described in Material and Methods. The hydroalcoholic R. rosea extract showed a distinct dose-dependent inhibition of the CPE caused by the H3N2 influenza virus A/Hong Kong/68 with an IC 50 value of 2.8 µg/mL in MDCK cells (CC 50 > 100 µg/mL). In contrast, the extract showed no activities against the following respiratory pathogens: rhinovirus A2 as well as the bacterial strains S. aureus ATCC 25923 and ATCC 43300, and S. pneumoniae DSM 20566. Pleconaril, ampicillin, and rifampicin, used as reference antirhinoviral/antibacterial agents, performed as expected [38,39].

Discussion
The continuing impact of natural products in modern medicine [40] emphasizes the importance of scrutinizing the phytochemical universe of plant secondary metabolites. Getting access to iso-▶ Table 3 Stepwise gradient elution program of the developed HPCCC method.

Gradient elution steps
HEMWat system no.  lates forms the basis for quality control and standardization procedures. Further, their physical availability is a requirement for assaying and thus augmenting the pharmacological spectrum of herbal constituents. This work established a gradient elution method by HPCCC that allowed for a preparative enrichment of R. rosea constituents. A fast orthogonal chromatographic step enabled the purification of 12 secondary metabolites from the R. rosea extract. The applicability of this method for a broad and structurally diverse spectrum of compounds of a wide polarity range is further represented by obtaining a cyanogenic compound (3) and two mono-terpenoids (8, 11) as well as minor flavonoid glycosides (7, 10) and the methoxylated flavonoid 12.
Since HPCCC separation entirely relies on liquid/liquid partition, any adsorption effect on the stationary solid phase is avoided, which is a clear advantage in contrast to classic column chromatographic approaches.
Moderate anti-influenza A virus activity in the CPE assay has already been reported for an ethyl acetate and a water extract of R. rosea [6]. In the present study, the standardized hydroethanolic extract exhibited distinct and specific anti-influenza virus activity. This was mainly observed for one particular compound class, i.e., the flavonoids rhodiosin (10), tricin (12), and tricin-5-O-β-D-glu-copyranoside (7). The anti-influenza A virus activity of rhodiosin (10) has already been reported for the subtypes H1N1 and H9N2 in a CPE assay [6] but not yet for the here used H3N2 strain. Moreover, our results further underline the anti-influenza virus activity of tricin (12) [15]. For the first time, the 5-O-β-D-glycosylated congener (7) is reported to have similar anti-influenza A virus activity as its aglycone. However, the activities of the isolated compounds do not fully explain the distinct inhibitory activity of the investigated R. rosea extract.

Plant material and reagents
The R. rosea extract SHR-5 (batch no. 1521229) was obtained from the Swedish Herbal Institute in 2018 and prepared with 70 % ethanol according to HMPC guidelines. A voucher specimen (JR-20180904-A1) is deposited at the Department of Pharmacognosy, University of Vienna, Austria.
All solvents were purchased from VWR. Acetonitrile for UPLC analysis was of chromatographic grade. For NMR measurements, (ultrahigh-)gradient grade deuterated methanol was used (Merck).

General experimental procedures
TLC was used for monitoring of the fractionation process. The mobile phase was composed of ethyl acetate-methanol-water-formic acid (77 : 13 : 10 : 2). Merck silica gel 60 F 254 plates served as the stationary phase. Detection was carried out under UV 254 , UV 366 , and after derivatization with H 2 SO 4 (5 % in methanol) under visible light.
For calculation of the K values and monitoring of the fraction purity and purity determinations, a Waters ACQUITY H-Class UPLC comprising a sample-, quaternary solvent-, and column manager with an ELSD and PDA (205 nm, scan from 190-400 nm) detection was used. The instrument was controlled via Empower 3 software.
Due to resolution reasons, fractions of interest and purity determinations of isolates were analyzed with an Acquity BEH C 18 column (dimensions: 3 mm × 100 mm, 130 Å, 1.7 µm), 40°C, and the same mobile phase solvents as before, but with a slightly extended gradient as follows: start with 15 % B for 0.3 min, from 0.3-9 min increase from 15-40 % B, 9-13.5 min increase from 40-98 %, and a subsequent washing step at 98 % solvent B (Fig. 2S, Supporting Information).
MS analysis was performed on a DIOXNEX UltiMate 3000 RSLC instrument consisting of an RS autosampler, autopump, auto column compartment, DAD, and a Corona Ultra RS hyphenated to Bruker Daltonics HCT equipped with a 3D quadrupole ion trap and orthogonal ESI. The Acquity BEH C 18 column (40°C) used water as solvent A and ACN-MeOH (80 : 20) as solvent B with a gradient starting from 5 % B for 2 min, 2-15 min increase from 5-40 %, 15-15.5 increase from 40-98 % solvent B, and a subsequent washing step. The software Xcalibur was used.
NMR experiments were performed on a Bruker Advance 500 NMR spectrometer (Ultra-Shield) enhanced by a TCI Prodigy CryoProbe (5 mm, triple resonance inverse detection probe head). Spectra were analyzed using MestreNova software.

Selection of the solvent system
Ten mixtures composed of different volume ratios of the following five solvents were tested: n-hexane, ethyl acetate, methanol, n-butanol, and water (HEMWat). Additionally, a previously published solvent system composed of chloroform-methanol-isopropanol-water was tested [23]. The K value of compounds in the R. rosea extract in these different solvent mixtures was calculated using UPLC as follows: The respective solvents (▶ Tables 1 and 2) were mixed and after shaking vigorously and phase separation, 1 mg of sample was weighted and dissolved in 1 mL of the upper and 1 mL of the lower phase of the before prepared solvent mixture. Next, 500 µL of the upper and lower phase, respectively, were removed and 5 µL of each phase was injected into the UPLC. To obtain partition coefficients, six selected peaks were integrated ( ▶ Tables 1 and 2). The K value is conveyed by the AUC of the stationary phase divided by the AUC of the mobile phase [41]. Hence, AUCs (PDA detection at 205 nm) were divided K = AUC upper /AUC lower for the RP mode and K = AUC lower /AUC upper for the NP mode.

High-performance countercurrent chromatography apparatus and general separation settings
For HPCCC experiments, a Spectrum instrument from Dynamic Extractions combined with an Accel 500 LC chiller by Thermo Scientific connected to an Alpha 10+ Isocratic Pump by Ecom was used. For analytical and preparative experiments, the multilayer column was completely filled with the stationary phase, rotated at maximum speed of approximately 1600 rpm and then the mobile phase was pumped into the system. After reaching hydrodynamic equilibrium, the sample was injected. Samples were filtered through a paper filter before injection. Due to the aqueous character of the fractions, they were subsequently dried in a GeneVac Standard (EZ-2 series) in the "aqueous" mode to evaporate the solvent. The fractionation was monitored by TLC (dried fractions dissolved in 1 mL MeOH, 3 µL applied to the TLC plate).

Analytical high-performance countercurrent chromatography
An analytical column with a volume of 22 mL and a sample loop of 1 mL was used. The instrument was operated in NP and RP modes for method development with a flow rate of 1 mL × min −1 . Approximately 25 mg of roseroot extract were dissolved in 1 mL of a 1 : 1 mixture of mobile and stationary phases. Stepwise gradient elution was performed and repeated in a multistep elution program 2-4 times as follows: 50 mL of each mobile phase was pumped into the system. When 25 mL of the mobile phase solvent was consumed, the same amount of the next step mobile phase was added to yield 50 mL. After dropping down to 25 mL, it was then filled up with the remaining 25 mL to 50 mL. A volume of 1.5 mL per fraction was collected in gradient elution and extrusion.

Preparative high-performance countercurrent chromatography
A preparative column covering a volume of 136 mL with a sample loop of 12 mL was used. The instrument operated only in the RP mode with a flow rate of 6 mL × min −1 . Then, 2 × 370 mg were dissolved in 10 mL of a 1 : 1 mixture of mobile and stationary phases. A stepwise gradient elution was performed and repeated in a multistep elution program 2-4 times as follows: 285 mL of each mobile phase was pumped into the system. When 210 mL of the mobile phase solvent was consumed to 75 mL, the same amount of the next step mobile phase was added to yield 150 mL. After dropping down to 75 mL, it was then filled up with the remaining 210 mL to 285 mL (▶ Table 3). A volume of 6 mL per fraction was collected in gradient elution and 5 mL per fraction in extrusion. The preparative procedure was repeated once again to increase the yield. According to the TLC fingerprint, fractions of the two preparative runs were combined to yield 13 fractions (FI-FXIII) with an overall sum of approximately 700 mg in total.

Control compounds
Pleconaril, an inhibitor with well-known anti-rhinovirus activity [39], was used to validate the assay with rhinovirus A2. The synthesis of pleconaril and its purity (99.9 %) were described previously [42]. The drug zanamivir (GlaxoSmithKline) was used as the reference compound in the anti-influenza virus assay [43]. Stock solutions of pleconaril and zanamivir (10 000 µM) were prepared in dimethyl sulfoxide and bidistilled water, respectively. Ampicillin (purity 99 %; Carl-Roth GmbH) and rifampicin (purity 97 %; Sigma-Aldrich) were used as reference compounds in the antibacterial assays. A stock solution of ampicillin (10 000 µg/mL) in bidistilled water and rifampicin in DMSO (10 000 µg/mL) were prepared. The highest DMSO concentration in the test was 0.05 %.

Cells, viruses, and bacteria
HeLa cells (human cervix carcinoma; Flow Labs) and MDCK (Friedrich Löffler Institute) cells were used in the cytotoxicity and antiviral assays. Viruses included in this study were rhinovirus A2 (Institute of Biochemistry, University, Vienna) and the H3N2 influenza virus A/Hong Kong/68 (Schaper and Brümmer).
Antibacterial assays were performed with the neuraminidase A and B expressing S. pneumonia DSM 20566 (Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures) [44] and S. aureus ATCC 25923 and ATCC 43300.

Cytotoxicity and antiviral activity
The cytotoxicity of the test compounds was determined on twoday-old confluent cell monolayers of HeLa and MDCK cells as described previously [45]. The maximum tested concentration was 100 µg/mL for the extract and 100 µM for the isolated compounds. Cell viability was evaluated as the percentage of the mean value of optical density resulting from the six cell controls, which was arbitrarily set as 100 %. The compound concentration that reduces the viability of the treated cells in comparison to mock-treated control cells (no inhibitor) by half is called 50 % cytotoxic concentration (CC 50 ) and was calculated by using a Four Parameter Logistic (4PL) Curve Calculator [46]. Two-day-old confluent cell monolayers of MDCK cells were used for CPE inhibitory assays with influenza virus A/Hong Kong/ 68. However, HeLa cells were preincubated for only 1 day for the anti-rhinovirus assays. After the aspiration of the cell growth medium, 50 µL of test medium (mock treatment of cell and virus controls) or half-log dilutions of reference compounds or test items in test medium, and a certain multiplicity of infection of the test virus in a volume of 50 µL of the test medium were added to the cell monolayers. The multiplicity of infection of rhinovirus A2 and influenza virus A/Hong Kong/68 was adjusted to 0.03 and 0.008 TCID 50 /cell, respectively. Plates were incubated at 37°C in a humidified atmosphere with 5 % CO 2 for 48 h influenza virus A/ Hong Kong/68) or 72 h for rhinovirus A2. Thereafter, the cell monolayers were fixed and stained with the Dynex Immuno Assay System as described elsewhere [45]. The percentage of antiviral activities of the test compounds was calculated according to Pauwels et al. [47] and used to calculate the IC 50 with the Four Parameter Logistic (4PL) Curve Calculator [46].
A minimum of three independent cytotoxic as well as antiviral experiments were performed, with the exception that compounds without antiviral activity were tested only twice in the antiviral assay.

Evaluation of antibacterial activity by broth microdilution assay
Double-concentrated half-log dilutions of reference antibiotics (ampicillin and rifampicin) and test items (pure extract and compounds) were prepared in brain-heart infusion broth or Todd Hewith broth with yeast extract (all Carl-Roth GmbH) for staphylococci and pneumococci, respectively. A volume of 50 µL of each dilution was dispensed into U-bottomed 96-well sterile plates (Greiner Bio-One GmbH). Each dilution was tested in duplicate per assay.
Fifty µL of bacterial suspension consisting of 2 × 10 5 cfu/mL [38] in brain-heart infusion broth or Todd Hewith broth with yeast extract for staphylococci and pneumococci, respectively, were inoculated to the antibiotic dilutions or to six wells with antibioticfree broth for bacterial growth control. Furthermore, uninoculated wells of the antibiotic-free broth were included in each assay as a check of sterility. After 24 h of incubation at 37°C and 5 % CO 2 , the minimal inhibitory concentration endpoint was read as the lowest concentration of antibiotic or test item at which there was no visible growth.
Three independent assays were performed with active compounds and two were performed when the bacterial growth was not impacted at the maximum tested concentration of 100 µg/mL (extract) or 100 µM (pure compounds).

Supporting information
UPLC analyses of HEMWat system 4 (upper and lower phase), the RR extract, and thereof isolated pure compounds as well as yields of HPCCC fractions, cytotoxicity and anti-influenza A virus activity data are available as Supporting Information.