3-O-Galloylated Procyanidins from Rumex acetosa L. Inhibit the Attachment of Influenza A Virus

Infections by influenza A viruses (IAV) are a major health burden to mankind. The current antiviral arsenal against IAV is limited and novel drugs are urgently required. Medicinal plants are known as an abundant source for bioactive compounds, including antiviral agents. The aim of the present study was to characterize the anti-IAV potential of a proanthocyanidin-enriched extract derived from the aerial parts of Rumex acetosa (RA), and to identify active compounds of RA, their mode of action, and structural features conferring anti-IAV activity. In a modified MTT (MTTIAV) assay, RA was shown to inhibit growth of the IAV strain PR8 (H1N1) and a clinical isolate of IAV(H1N1)pdm09 with a half-maximal inhibitory concentration (IC50) of 2.5 µg/mL and 2.2 µg/mL, and a selectivity index (SI) (half-maximal cytotoxic concentration (CC50)/IC50)) of 32 and 36, respectively. At RA concentrations>1 µg/mL plaque formation of IAV(H1N1)pdm09 was abrogated. RA was also active against an oseltamivir-resistant isolate of IAV(H1N1)pdm09. TNF-α and EGF-induced signal transduction in A549 cells was not affected by RA. The dimeric proanthocyanidin epicatechin-3-O-gallate-(4β→8)-epicatechin-3′-O-gallate (procyanidin B2-di-gallate) was identified as the main active principle of RA (IC50 approx. 15 µM, SI≥13). RA and procyanidin B2-di-gallate blocked attachment of IAV and interfered with viral penetration at higher concentrations. Galloylation of the procyanidin core structure was shown to be a prerequisite for anti-IAV activity; o-trihydroxylation in the B-ring increased the anti-IAV activity. In silico docking studies indicated that procyanidin B2-di-gallate is able to interact with the receptor binding site of IAV(H1N1)pdm09 hemagglutinin (HA). In conclusion, the proanthocyanidin-enriched extract RA and its main active constituent procyanidin B2-di-gallate protect cells from IAV infection by inhibiting viral entry into the host cell. RA and procyanidin B2-di-gallate appear to be a promising expansion of the currently available anti-influenza agents.


Introduction
Influenza A and B viruses (IAV, IBV) circulating in the human population are responsible for seasonal epidemics of varying extent. At present, the global annual disease burden of seasonal influenza is estimated to be 1 billion infections, 3 to 5 million of severe infections, and 300 000 to 500 000 fatalities. Without doubt, vaccination remains the most important strategy for prophylaxis and control of seasonal influenza [1]. Although predominantly associated with mild symptoms of upper respiratory tract infection, the first pandemic of the 21st century caused by IAV(H1N1)pdm09 impressively demonstrated the global health risks associated with IAV. Ongoing zoonotic infections with avian IAV(H5N1) and (H7N9) in the human population underscore the permanent threat of pandemic outbreaks, of which the ''Spanish flu'' pandemic of 1918-19 with an estimated number of 50 million deaths world-wide has been the most devastating [2].
Two classes of antiviral drugs have been licensed for the treatment and prophylaxis of influenza [3]. Matrix protein inhibitors, such as amantadine and rimantadine, inhibit viral uncoating. They are ineffective against IBV and are currently not recommended for the treatment of IAV infections due to high levels of resistance [4]. Neuraminidase inhibitors (NAI), such as oseltamivir and zanamivir, inhibit the release of virus progeny from infected cells and viral spread, are effective against IAV and IBV and have been licensed for first-line therapy of influenza. Although the vast majority of currently circulating IAV(H3N2) and (H1N1)pdm09 is sensitive to oseltamivir, the wide-spread use of oseltamivir has led to a high level of IAV(H1N1) resistance in 2008-9 [3,5]. In IAV(H1N1)pdm09 resistance against oseltamivir is almost exclusively caused by a single amino acid exchange (H275Y) in the neuraminidase [6]. Recently, two novel NAIs have been approved for the treatment of influenza, peramivir and laninamivir octanoate, the latter being effective also against oseltamivir-resistant influenza virus strains [3,7]. Since monotherapy with each of the NAIs currently licensed may eventually lead to the selection of resistant virus, drug combinations directed against different molecular targets of influenza virus may be a promising strategy to delay the development of resistance and to achieve synergistic effects. Thus, novel viral targets, antiviral agents and therapeutic strategies such as inhibitors of the viral RNA polymerase complex and broadly neutralizing antibodies should be developed and utilized for the treatment and prophylaxis of influenza [8,9].
Medicinal plant extracts with anti-IAV activity have been described in many publications [10][11][12]. Although in most plantderived preparations active compounds and structure-activity relationships remain to be elucidated, polyphenols have been frequently identified to be the antiviral principle in plant extracts [13]. In particular, the broad antiviral and antimicrobial activity of green tea and its components has received much attention [14,15]. In green tea and a number of other polyphenol-rich plant extracts, catechins and proanthocyanidins, a subgroup of polyphenols derived from oligomerized flavan-3-ols, were found to exert antiviral effects against influenza viruses and other enveloped and non-enveloped viruses [16][17][18][19][20]. Recently, we have shown inhibition of viral attachment of herpes simplex virus type-1 by proanthocyanidin-enriched extracts from Rumex acetosa L. (Polygonaceae) and Myrothamnus flabellifolia Welw. [21,22]. Extracts from R. acetosa are a component of modern phytotherapeutical preparations with nationally registered drug status in Europe, and are used in the treatment of acute and chronic respiratory viral infections [23].
Aim of the present study was to investigate the anti-IAV activity of the R. acetosa extract (RA) in vitro, to identify relevant compounds and structural requirements for anti-IAV activity and to characterize their mode of action. Our results show that RA strongly inhibits growth of IAV by blocking viral entry. The dimeric, digalloylated procyanidin epicatechin-3-O-gallate-(4bR8)-epicatechin-39-O-gallate (syn. procyanidin B2-di-gallate) was identified as main active principle in RA. Galloylation of the procyanidin backbone was found to be a prerequisite for anti-IAV activity.

Plant material, extract and isolated compounds of Rumex acetosa
Starting materials and preparation of the Rumex acetosa L. extract RA have been described recently [21]. Isolation and analytical characterization of proanthocyanidins from RA have been reported by Bicker et al. (2009) [24]. Structural features, sources and purity of flavan-3-ols, oligomeric proanthocyanidins, hydrolyzable tannins, depsides and building blocks of tannins used for antiviral bioassays used in this study are given in Figure 1 and Table 1. Sodium heparin (100,000 IU/g) was purchased from Roth (Karlsruhe, Germany).
The number of infectious particles in viral stocks was assessed by plaque titration. MDCK II cells were infected with serial dilutions of IAV in PBS containing 0.21% bovine albumin (MP Biochemicals, Eschwege, Germany), 100 U/mL penicillin (Biochrom), 100 mg/mL streptomycin (Biochrom), 230 mmol/L MgCl 2 (Roth) and 514 mmol/L CaCl 2 (Roth) for 30 min. at 37uC (500 mL/well). After discarding the inoculum, cells were washed with PBS and covered with 2 mL of overlay medium (MEM [Gibco, Life Technologies, Darmstadt, Germany] containing 100 U/mL penicillin, 100 mg/mL streptomycin, 0.21% NaHCO 3  Cytotoxicity assay, antiviral assays Cytotoxicity assay. The effect of RA and its components on the proliferation of MDCK II cells was determined in 96-well plates (TPP, Trasadingen, Switzerland) using the MTT assay [27] essentially as described by   [21] with the exception that samples were incubated at 37uC for 1 h prior to addition to cells and remained on the cells for 48 h. The cytotoxic concentration of RA or its components which reduced the cells' viability by 50% (IC 50 ) was determined from dose-response curves. The untreated control was arbitrarily set as 100%.
MTT IAV assay. The inhibitory effects of RA and other test compounds on the cytopathic effect induced by IAV replication was determined in a MDCK II cell-based assay measuring cell viability by MTT stain (MTT IAV assay) [28]. An inoculum of 1610 4 pfu IAV/well (corresponding to a multi plicity of infection of 0.1) was used to infect 96-well plates. All incubation steps were performed with serum-free MEM. In the elementary assay, IAV was pre-incubated with test compounds for 1 h at 37uC and subsequently MDCK II cells were incubated with this RA/IAV mixture for 48 h. In modified assays, either the test compound/ IAV mixture was removed from the cells after 60 min., or cells were pre-incubated with test compounds alone for 1 h prior to infection with IAV, or test compounds were added to the cells following a 1 h infection period with IAV.
The antiviral activity was calculated according to the following formula [29]: (OD T ) IAV represents the optical density of cells, which were infected by IAV (index: IAV) and treated with RA. (OD C ) IAV corresponds to the optical density measured for the untreated IAV-infected cells and (OD C ) mock is the optical density of untreated, mock-infected cells. The antiviral dose of RA which protected the cells by 50% was defined as the 50% inhibitory concentration (IC 50 ).
Plaque reduction assay. IAV was incubated with antiviral compounds for 1 h at 37uC, both diluted in PBS containing 100 U/mL penicillin, 100 mg/mL streptomycin, 230 mmol/L MgCl 2 and 514 mmol/L CaCl 2 . MDCK II cells, cultivated in 12well culture plates (Greiner Bio-One, Frickenhausen, Germany), were washed with PBS and infected with 300 mL/well IAV/RAsuspension (100 pfu/well). After 30 min. of incubation, the inoculum was removed, 1 mL of overlay-medium without bovine albumin was added and the plates were cultivated for 72 h at 37uC. Subsequently, cells were stained as described above, virus plaques were counted and antiviral activity was calculated by the following formula [21]: Penetration assay. The effect of extract RA and antiviral compounds on viral penetration was determined by a modified plaque reduction assay. In contrast to the basic assay, cells were treated with RA after virus attachment to the cell surface. Penetration of IAV during the attachment and treatment phase was prevented by strictly performing all steps at 4uC.
MDCK II cells, cultivated to 95% confluence in 12-well culture plates, were pre-cooled to 4uC for 15 min. and washed with PBS. 600 pfu IAV, diluted in PBS (400 mL/well) containing 100 U/mL penicillin, 100 mg/mL streptomycin, 230 mmol/L MgCl 2 and 514 mmol/L CaCl 2 , were allowed to attach to the cells. After 20 min. the inoculum was removed, cells were washed with PBS, PBS containing a 2-fold serial dilution of RA was added and cells were incubated for another 30 min. at 4uC. Before shifting culture plates to 37uC for initiation of viral penetration, cells were washed with PBS and covered with serum-free cultivation medium (see above). Following 30 min. incubation at 37uC, medium was removed and cells were treated with low pH citrate buffer (135 mM NaCl, 10 mM KCl, 40 mM citric acid, pH 3.0) for 15 s to stop penetration and inactivate attached, non-penetrated virions. Low pH buffer was removed by washing twice with PBS, and overlay medium was added. Further cultivation and quantitation of plaques was performed as described above. Mocktreatment of attached virus and inactivation of attached mock-treated virus by low pH citrate buffer immediately prior to the 37uC shift served as controls.

Hemagglutination inhibition test (HIT)
Twofold serial dilutions (25 mL) of test compounds in PBS and 4 hemagglutinating units (HU) of IAV (25 ml) were mixed carefully in 96-well plates with U-shaped bottom (Thermo Fisher Scientific Nunc, Schwerte, Germany). Plates were shaken for 5 min. and incubated for 25 min. at room temperature (RT). 50 mL of a 1.5% suspension of newborn chicken erythrocytes (RBC) in PBS (Labor Dr. Merk & Kollegen, Ochsenhausen, Germany) were added, and plates shaken again. Assays were read following a 2 h incubation period at RT, and the minimum inhibitory concentration (MIC), defined as the highest test compound dilution showing complete inhibition of the agglutination of erythrocytes, was determined. In every assay, a test compound control (compound plus RBC without addition of IAV), and erythrocyte controls (A: IAV plus RBC, without addition of test compound; B: RBC, without addition of test compound or IAV) were included. Test results were accepted if the back titration of IAV revealed 4 HU and the controls yielded correct results.

Statistical analysis
Data represent the means 6SD of at least three independent experiments. Statistical significance was evaluated by a two-tailed one sample t-test. A P value of ,0.05 indicated a statistically significant difference.

In silico protein-ligand docking
For in silico analyses the HA of influenza virus A/California/ 04/2009 (H1N1) [30] (protein data base ID 3LZG) was used. HA of A/California/04/2009 (H1N1) is closely related to HA of the vaccine strain A/California/07/2009 (H1N1) and HAs of IAV(H1N1)pdm09 strains circulating in the post-pandemic era in Europe and Asia [31,32]. Epicatechin (2), epigallocatechin-3-Ogallate (EGCG) (6), procyanidin B2 (7) and procyanidin B2-digallate (8) were docked to the HA of influenza virus in silico by the software Molecular Operating Environment (MOE) version 2011.10 (Chemical Computing Group, Montreal, Canada). After identifying potential binding sites at HA with the MOE module ''Site Finder'', the test compounds were docked into the 30 cavities with the best PLB (propensity for ligand binding) score using the MMFF946 force field as implemented in MOE. The flexible docking method (induced fit, i.e. both the ligand and the protein binding site were treated as flexible) was applied. The best score of each cavity-compound pair was compared to the best score of the remaining 29 cavities for each of the four compounds. The docking pose represents the best geometry (lowest score) of all investigated orientations of all compounds with respect to all cavities taken into account.

Rumex acetosa extract RA specifically inhibits IAV-infection in cell culture
Extract RA and its constituents were screened for anti-IAVactivity by single cycle, MDCK II cell-based MTT IAV assay. Depending on the IAV isolate, the screening window coefficient Z9 of the MTT IAV assay ranged from approx. 0.6 to 0.63, indicating that this assay is well suited to detect inhibitors of IAV entry and replication [28,33]. Extract RA exhibited 100% antiviral activity against IAV PR8 at concentrations.5 mg/mL with an IC 50 of 2.5 mg/mL. At extract concentrations$25 mg/mL a dose-dependent, increasing reduction of cell vitality was observed. The CC 50 of extract RA was determined to be approximately 80 mg/mL which corresponds to a selectivity index (SI = CC 50 /IC 50 ) of 32 ( Figure 2A). Almost identical data were found for the clinical isolate I1 of IAV(H1N1)pdm09 with an IC 50 of 2.2 mg/mL, and a SI of 36 ( Figure 2B). The results obtained by MTT IAV assay were corroborated by plaque reduction assay. At a concentration of 100 ng/mL extract RA reduced plaque formation of IAV(H1N1)pdm09 I1 in a highly significant manner by 67%, at 1 mg/mL by 100% ( Figure 3).
The antiviral effect of extract RA was tested in two additional clinical isolates of IAV(H1N1)pdm09 obtained in consecutive samples of a patient with acute respiratory distress syndrome. The oseltamivir-sensitive isolate NRW172 was obtained early after hospitalization, the oseltamivir-resistant isolate NRW173 was isolated after completion of oseltamivir therapy. Extract RA inhibited growth of NRW172 and NRW173 with similar efficiency. The IC 50 values determined for NRW172 (19 mg/ mL) and NRW173 (37 mg/mL) in MTT IAV assay were approximately 10-fold higher as observed in IAV PR8 and IAV I1 ( Figure 2C, D). Previous work indicated that a high protein load of samples may reduce the antiviral activity of extract RA [22]. Since stocks of IAV NRW172 (6.6610 6 pfu/mL) and NRW173 (8.3610 6 pfu/mL) contained significantly lower virus titers than stocks of IAV PR8 (3.2610 8 pfu/mL) and I1 (3.4610 7 pfu/mL), inhibitory effects of residual allantoic fluid on the anti-IAV activity of extract RA were studied. Retesting IAV I1 diluted to 6.6610 6 pfu/mL in allantoic fluid of a noninfected egg led to an approx. four-fold increase in the IC 50 of extract RA (8.2 mg/mL) ( Figure S1). Thus, inhibitory effects of residual allantoic fluid on the anti-IAV activity of extract RA appear to account for the differences in IC 50 values observed in MTT IAV assay. Accordingly, the consistently lower IC 50 values observed in plaque reduction assay are most likely due to higher dilution of virus stocks during incubation with extract RA. Whether strain specific factors also determine the susceptibility of IAV to extract RA as observed for a polyphenolic extract of Pelargonium sidoides DC [34] remains to be clarified.
Structure-activity relationship: epicatechin-3-O-gallate-(4bR8)-epicatechin-39-O-gallate (procyanidin B2-di-gallate) (8) is responsible for the antiviral activity of RA The lead compounds in extract RA have been recently described to be flavan-3-ols and oligomeric proanthocyanidins [24]. To pinpoint the plant secondary products responsible for the antiviral effect of the extract, the dominant proanthocyanidins isolated from extract RA were tested for antiviral effects against IAV I1 and cytotoxicity (Table 2) (for numbering of compounds compare Table 1) at concentrations of 2, 20 and 200 mM, respectively, by MTT IAV and cytotoxicity assay. Additionally EGCG (6), a known inhibitor of IAV replication from extracts of green tea which is not present in extract RA [17,24] was included ( Table 2).
Thus, within the complex mixture of extract RA dominated by flavan-3-ols and proanthocyanidins with different degrees of polymerization and galloylation, the antiviral activity is mostly mediated by galloylated oligomers. The dimeric compound procyanidin B2-di-gallate (8) was assessed as the main principle of antiviral activity in extract RA. The content of procyanidin B2di-gallate (8) in extract RA was determined by UHPLC to be 0.96%. The strong antiviral effect of procyanidin B2-di-gallate (8) was confirmed by plaque reduction assay (Figure 3). Purified galloylated higher oligomers present in extract RA were not available for antiviral testing, however, most likely are also active against influenza virus. Generally, a higher number of pyrogalloyl moieties, an increased degree of polymerization and a 4bR8 interflavan linkage amplify the anti-IAV activity of polyphenols from extract RA. These findings are in accordance with the results published by De Bruyne et al. (1999) [35] describing similar structural requirements of polyphenols active against HSV and HIV. In addition, trihydroxylation of the B-ring of non-galloylated oligomeric proanthocyanidins has been reported to mediate antiinfluenza virus activity [34].
An insignificant anti-influenza activity of the monomeric flavan-3-ols catechin (1) and epicatechin (2) has been reported earlier  Table 2. Anti-IAV activity and effect on cell vitality of flavan-3-ols and oligomeric proanthocyanidins from Rumex acetosa extract RA and structurally related polyphenolic compounds.  [16,19]. Interestingly, Song et al. (2005) [19] showed that ECG (5) a main constituent from green tea strongly inhibited anti-IAV and IBV in cell culture whereas EGC (4) exhibited little antiviral activity. Yang et al. (2014) [16] found that procyanidin B2 (7) significantly inhibited growth of IAV. This is in contrast to our findings where ECG (5) and procyanidin B2 (7) were screened negative for anti-IAV activity at noncytotoxic concentrations. Most likely, this reflects differences in the test format used, e.g. MTT IAV assay vs. plaque reduction assay and cytopathic effect inhibition assay, respectively. In particular, the assays used by Song et al. (2005) [19] and Yang et al. (2014) [16] imply multicycle replication of IAV and thus should also detect inhibitory effects of compounds on late steps of the viral replication cycle, such as assembly, maturation and release as reviewed by Beyleveld et al. (2013) [28]. Accordingly, Song et al. (2005) [19] detected a direct inhibition of the viral neuraminidase activity by ECG (5), however, not by EGC (4). A prominent virucidal activity of EGCG (6) from green tea has been first reported by Nakayama et al. (1993) [17]. As reported for ECG (5), EGCG (6) also directly inhibits the viral neuraminidase. In addition to anti-influenza activity, EGCG offers broad antiinfective properties against various viral, bacterial and fungal pathogens as reviewed by Steinmann et al. (2012) [15].
After oral application, proanthocyanidins exhibit a very limited bioavailability as reviewed by Zumdick et al. (2012) [36]. Thus, the oral application of active compounds such as procyanidin B2-digallate (8) for the systemic treatment of influenza virus infection appears to be inappropriate. As an alternative, the local application of procyanidins in the upper respiratory tract, either by lozenges, chewing gums etc. or by inhaling devices allows the active compounds to directly contact the virus and should be preferred.
Because proanthocyanidins are known to have tannin-like effects it might be assumed that these polyphenols from extract RA nonspecifically inactivate essential viral structural proteins. Therefore we included other polyphenols not being part of extracts from R. acetosa, but with known strong astringent activity (Table 1). Pentagalloyl-glucose (PGG) (12), a well characterized hydrolyzable tannin [37], showed moderate antiviral activity, however, significant cell toxicity in the MTT IAV assay ( Table 2). Also the ellagitannins geraniin (13), corilagin (14) [38] and ellagic acid (15) were inactive at the highest concentration tested (200 mM) ( Table 2). When added at concentrations in the millimolar range, ellagic acid (15) has been reported to exhibit broad anti-influenza activity in vitro and in vivo [39]. The depside rosmarinic acid (16), known as tannin-like compound, was also inactive. Keeping in mind that also oligomeric procyanidins such as procyanidin B2, C1 or D1 (7, 10, 11) are known to interact strongly with proteins in a tannin-like manner, nonspecific denaturing effects do not appear to account for most of the antiviral activity observed for procyanidin B2-di-gallate (8). Otherwise, a more potent activity of the hydrolyzable tannins geraniin (13) and corilagin (14) should have been observed. An exception appears to be PGG (12), which exhibited moderate anti-IAV activity in MTT IAV assay with an IC 50 of 22 mM. This might be due to its flexible structure. In contrast to geraniin (13), PGG (12) owns the capacity to rotate its galloyl moieties relatively to the glucose. As a result PGG (12) may be able to bind more strongly to proteins. In accordance with our results, PGG (12) has been recently reported to possess anti-IAV activity at micromolar concentrations and to inhibit viral entry, budding and release [40].
Since only the galloylated compounds (6) and (8) [41] recently reported that gallic acid (17) inhibits reporter gene expression of the recombinant IAV laboratory strain A/Puerto Rico/8/34-NS116-GFP in a multi-cycle assay with an EC 50 of approx. 50 mM and a SI of approx. 15. Similar to our findings, however, preincubation of IAV(H1N1)pdm09 particles for 2 h with 50 mg/mL (corresponding to 265 mM) gallic acid (17) had only little effect on virus replication in A549 cells. Furthermore, gallic acid (17) poorly inhibited IAV neuraminidase with an IC 50 of.500 mM. Thus, the inhibitory mechanism of gallic acid (17) on IAV replication remains to be clarified.

Extract RA affects viral attachment
To identify steps in the viral life cycle that were affected by extract RA, virus and cells were treated with extract RA at different times pre and post infection. If pre-treated IAV was added to cells for 1 h, viral replication was inhibited completely at concentrations of extract RA.10 mg/mL. In contrast, if cells were infected with IAV and extract RA was added after 1 h, no antiviral effect was observed at #10 mg/mL, indicating that extract RA does not operate in the post-entry phase (data not shown).
To determine whether extract RA interacts with target molecules of the host cells or of the virus, MDCK II cells were incubated with extract RA for 1 h and subsequently infected with IAV. At concentrations of #10 mg/mL this preincubation of the host cells did not result in any antiviral effects (data not shown). This suggests that the anti-IAV activity of extract RA is caused by direct interaction with IAV particles and inhibition of viral entry as shown for a number of polyphenol and tannin-rich plant extracts in earlier reports [17][18][19]39,[41][42][43][44].
To reconnoiter the effect of extract RA to inhibit penetration of IAV particles already attached to the cell surface we used a penetration assay. Cells were infected at 4uC, unbound viral particles were removed by washing, extract RA was added at 4uC for 30 min., and penetration was allowed to occur by a temperature shift to 37uC (30 min.) followed by washing with pH 3.0 citrate buffer to inactivate non-penetrated virus. As shown in Figure 4, extract RA also blocks viral penetration. However, in comparison to incubation of IAV with extract RA prior to entry, significantly higher concentrations of extract RA were needed to achieve comparable antiviral effects. Washing of cells with pH 3.0 citrate buffer at 4uC immediately after the adsorption period and prior to shifting the temperature to 37uC completely abrogated plaque formation. These observations suggested that RA affects virus entry primarily by inhibiting viral attachment. Similar results were also obtained with EGCG (6) and procyanidin B2-di-gallate (8) (Figure 4). As discussed above, the relatively high protein load due to the presence of cells and culture media components may increase the concentration of RA and its active constituents needed to inhibit penetration of IAV already attached to the cell surface. When added after the infection of MDCK cells, high concentrations of green tea extract and EGC (4) have been reported to affect the early phase of influenza virus infection, possibly by interference of the polyphenolic compounds with the acidification of endosomes [18].

RA and galloylated oligomeric procyanidins interact with IAV hemagglutinin
Data presented above suggested that extract RA, EGCG (6) and procyanidin B2-di gallate (8) may interfere with the sialic acid receptor binding function of the viral HA. Therefore, effects on HA-mediated attachment of IAV to the cell surface were further investigated in a hemagglutination inhibition assay. Using four hemagglutinating units of IAV(H1N1)pdm09 I1 in allantoic fluid (5.5610 7 pfu/mL) to agglutinate chicken erythrocytes, pretreatment of the IAV suspension with extract RA inhibited erythrocyte agglutination at a minimum inhibitory concentration of 156 mg/ mL (Table 3). At higher concentrations, hemagglutination reappeared due to direct agglutination of erythrocytes by extract RA. By serial dilution of extract RA in PBS the minimal concentration needed to agglutinate erythrocytes in the absence of IAV was determined to be 156 mg/mL. Thus, treatment of IAV with extract RA appears to directly interfere with the cell surface receptor-binding function of IAV HA. Procyanidin B2-di-gallate (8) did not inhibit IAV-mediated hemagglutination, however, was able to directly agglutinate erythrocytes at a concentration$ 39 mM. In accordance to Theissen et al. (2014) [41] EGCG (6) showed no inhibitory effect on IAV-mediated hemagglutination, however, directly agglutinated erythrocytes ( Table 3). None of the test compounds induced hemolysis (data not shown). Strong, IAV-strain specific differences in the concentrations of EGCG (6) needed to inhibit hemagglutination have been reported earlier [19], and may account for the failure to detect inhibitory effects of procyanidin B2-di-gallate (8) and EGCG (6) on IAV(H1N1)pdm09 induced hemagglutination.
In addition, the physical interaction of extract RA and its active compounds with recombinant, soluble HA was studied by SDS-PAGE and immunoblotting. Incubation of HA with high concentrations of extract RA, i.e. 2.5 to 10 mg/mL, for 1 h led to the almost complete disappearance of the 75 -85 kDa HAspecific band in SDS-PAGE ( Figure 5) and abrogated reactivity of HA with an HA-specific monoclonal antibody in immunoblotting (data not shown). Extract RA-treated HA appeared to be retained in the gel pockets, most likely due to the formation of large, electrophoretically immobile complexes. At lower concentrations, i.e. 1 to 0.1 mg/mL, extract RA had no effect on the electrophoretic mobility and immunoreactivity of HA, respectively. Taking into consideration that the IAV-specific IC 50 value of extract RA in MTT and plaque reduction assay is approximately 100 to 1,000-fold lower, this finding supports the conclusion that most of the anti-IAV activity of extract RA is not due to nonspecific tannin-like effects on viral proteins.
Incubation of HA with high concentrations of procyanidin B2di-gallate (8) (1.13 mM) and EGCG (6) (2.18 mM) led to a time dependent slight reduction of the monomeric HA band and the appearance of HA aggregates being visible in Coomassie-stained gels as a broad 75 to.200 kDa ''smear'' ( Figure 6A, C). After incubation of HA with EGCG (6) for 4 h to 24 h a faint band corresponding to HA dimers became visible ( Figure 6A). Higher oligomers of HA could not be detected. As compared to mock treated HA, incubation with the galloylated oligomeric proanthocyanidins (6) and (8) only led to a moderate decrease in the intensity of the band corresponding to monomeric HA in Coomassie-stained gels ( Figure 6A, C). Both compounds, however, reduced the strength of the HA monomer-specific signal in immunoblot ( Figure 6B, D). The decrease in immunoreactivity of HA appeared to be more pronounced for (6).
Thus, (6) and (8) exhibit tannin-like astringent effects on HA when applied for prolonged times at high concentrations, i.e., at concentrations approx. 100 to 10,000-fold higher than the respective IC 50 values in MTT IAV assay and plaque reduction  assay, respectively. The observed ''smear'' in SDS-PAGE and immunoblots may stem from HA literally coated with various amounts of (6) and (8). This may also account for the reduced reactivity of the His-tag-specific monoclonal antibody used to detect recombinant soluble HA. Similar effects were observed with an HA-specific monoclonal antibody (data not shown). The effects of high concentrations of extract RA and its active compounds on HA are in good accordance with the model suggested by Haslam (1996) [45] by describing the aggregation of proteins by polyphenols, and confirms earlier findings in HSV-1 [21]. On the other hand, antiviral effects of (6) and (8) are detectable at much lower concentrations. Therefore, similar to what was observed for RA, tannin-like astringent effects are unlikely to mediate most of the antiviral activity of these compounds.
Procyanidin B2-di-gallate (8) is predicted in silico to interact with the sialic acid binding site of viral hemagglutinin To visualize the binding of components from RA to the viral surface proteins, four selected compounds were docked to HA of influenza virus A/California/04/2009 (H1N1) [30] in silico by means of the software package MOE. Exemplary for the docking results of all investigated cavities of HA, the score of the docking at the sialic acid binding site [46] was -6.29 for procyanidin B2-digallate (8), 25.55 for procyanidin B2 (7), 25.89 for EGCG (6), and 25.28 for epicatechin (2), with (8) showing the best score. The data demonstrated a better score of galloylated compounds in comparison to the respective ungalloylated molecules. Additionally, inspection of the best docking pose revealed the binding of (8) (Figure 7) with both galloyl moieties and the B-ring of the second epicatechin gallate unit deep inside the sialic acid binding pocket, suggesting a notably stronger anchorage of galloylated molecules in contrast to ungalloylated compounds and offering a straightforward explanation for the strong activity of this digalloylated dimer. Aside from this, the investigated dimers (7) and (8) yielded a better docking score than the monomeric (6) and (2). These results further corroborate the observation depicted in the functional bioassays: An increase in the degree of polymerization and galloylation enhances the binding of proanthocyanidins to HA. As discussed already above, these results are in contrast to a model favoring the unspecific ''coating'' of HA by polyphenols. The strong anchoring of the galloylated compounds (6) and (8) in the sialic acid binding pocket of HA disclosed by in silico visualization may block the receptor binding site of HA and consequently specifically inhibit the viral adsorption process. EGCG (6) blocks binding of HIV gp120 to its cellular receptor CD4, and it has been suggested that there is an appropriate binding site of EGCG (6) in the region of CD4 interacting with gp120. The galloyl ring D of EGCG (6) appears to stack against aromatic and basic amino acid side chains within the gp120 binding site of CD4, e.g., Phe 43, Arg 59, Trp62 of CD4, thereby abrogating interaction of gp120 with CD4 [47]. Notably, crystal structure analyses revealed that a subgroup of neutralizing antibodies interferes with receptor binding of HA by targeting the highly conserved Tyr98 and Trp153 at the hydrophobic cavity base of the sialic acid binding site with an aromatic side chain [48,49]. It is therefore worth mentioning that in our docking model, the galloyl moiety of the second epicatechin gallate unit of procyanidin B2-di-gallate (8) is close to the aromatic side chain of Trp153 in the sialic acid binding pocket of HA, where it might interact in terms of a T-shaped p-p interaction. Furthermore, the B-ring of the second subunit is in a position where its phenolic oxygens might form hydrogen bonds with the hydroxyl proton of Tyr98 (both distances O…H,3 Å ; see Figure 7B).

RA does not interfere with cellular responses to TNF-a and EGF
While the extract RA showed little cytotoxic effect over a wide range of concentrations it might still elicit or interfere with intracellular responses in treated cells. Thus, the effect of the addition of high concentrations of RA (100 mg/mL) close to the calculated CC 50 for 1 h at 37uC on TNF-a and EGF induced signal transduction was studied. As shown in Figure 8A stimula-  tion of A549 cells by TNF-a led to similar increases in phosphorylated NF-kB (pNF-kB) in RA-treated or mock-treated cells, respectively. In the absence of TNF-a, neither RA nor mocktreatment led to a significant induction of pNF-kB. Potential effects of RA on Raf/MEK/ERK-signaling were investigated by stimulation of A549 cells by EGF ( Figure 8B). While non-EGFstimulated cells did not express pERK1/2, regardless if pretreated with RA or not (lanes 1 and 3), EGF treatment activated its expression (lane 2). Pretreatment of the cells with RA, followed by stimulation with EGF did not result in a significant decrease in pERK1/2 expression. It was thus concluded that even high concentrations of RA close to the CC 50 are unlikely to significantly elicit or interfere with TNF-a and EGF-induced signal transduction. This is in accordance to recent results showing that cells are inert to LADANIA067, a polyphenol-rich extract of Ribes nigrum folium inhibiting entry of IAV [50].

Conclusions
The proanthocyanidin-enriched extract RA and its main active constituent epicatechin-3-O-gallate-(4bR8)-epicatechin-39-O-gallate (procyanidin B2-di-gallate) protect cells from IAV infection by blocking IAV adsorption and interfering with penetration at higher concentrations. Anti-IAV-activity is dependent on galloylation of the procyanidin backbone. At effective concentrations, cells are unaffected by RA and procyanidin B2-di-gallat. Regarding the need for new and abundantly available anti-influenza therapeutics, RA and procyanidin B2-di-gallate appear to be a promising expansion of the currently available anti-influenza agents. Figure S1 Inhibitory effect of residual allantoic fluid on the antiviral activity of RA. To demonstrate that titres of viral stocks prepared from allantoic fluid of infected eggs have an impact on the outcome of the MTT IAV assay, stocks of isolate I1 (H1N1)pdm09 were approx. 50-fold prediluted in allantoic fluid (from 3.2610 8 pfu/mL to 6.6610 6 pfu/mL). Subsequently, virus was diluted to 1610 4 pfu IAV/well in serum-free medium and the antiviral activity and cell vitality were determined by MTT IAV assay and cytotoxicity assay, respectively (compare Figure 2). Values represent mean 6SD of $3 independent experiments, * p,0.05, ** p,0.01 (two-tailed, unpaired Student's t-test). Statistical significance of antiviral activity was calculated for nontoxic concentrations only (1 to 5 mg/mL). (TIF)