The plant material from M. alba (root bark, batch no. 460797) was purchased from Plantasia in 2018. A voucher specimen (JR-20190928-A1) is deposited at the Department of Pharmaceutical Sciences, Division of Pharmacognosy, University of Vienna, Austria.

Double distilled water, HPLC-grade acetonitrile (VWR) and formic acid (VWR) were used for the chromatographic separation. For extract preparations, petroleum ether and n-hexane were distilled according to the Austrian Pharmacopoeia. Isopropanol (VWR) was purchased in analytical grade. Prior to LC-MS analyses, samples were dissolved in HPLC-grade methanol (VWR).

The MA60 extract was prepared as previously reported (Langeder et al., 2023). Briefly, a two-step extraction method was performed on a Dionex ASE 350 accelerated solvent extraction (ASE) instrument from Thermo Fisher Scientific, United States. Approximately 15 g of ground M. alba root bark were filled in a 34 mL extraction cell. In a first defatting step, the root bark was extracted with n-hexane in the flow mode (120°C) to remove highly lipophilic compounds. The obtained n-hexane extract was discarded. The second and essential extraction step was carried out with a mixture of isopropanol and petroleum ether in the ratio 2:1 in the flow mode (80°C). This procedure was repeated multiple times to obtain 2.8 g extract from 300 g plant material. MA60 consists of 10.7% sanggenon C and of 6.9% sanggenon D. Further, it contains other sanggenons, i.e., sanggenon B (1.2%), sanggenon G (1.2%), sanggenon O (0.8%) and sanggenon E (1.0%).

Sanggenons C and D were initially isolated from a methanolic M. alba root bark (2 kg) extract. Eleven fractions were obtained via flash chromatography (Interchim puriFlash^® 4250) separated on a Puriflash 25 Silica HC 200G 25 µm column. Fractions 5 and 6 were further purified on a Gemini-Nx C₁₈ phemomenex^® column yielding 515 mg of sanggenon C (purity: 99%) and 730 mg of sanggenon D (purity: 99%), as previously described in detail (Langeder et al., 2023).

Female mice are known to be more susceptible to influenza virus infection than male mice (Robinson et al., 2011). Therefore, female BALB/c mice (eight-weeks-old; 16–18 g; n = 66) were used in the present in vivo studies. Mice were purchased from Charles River (Bad Sulzfeld, Germany). They were housed in individually ventilated cages, at 22°C ± 2°C with a relative humidity of 55% ± 10% and a 14/10 h light/dark cycle. Mice were given food and water ad libitum.

Madin Darby canine kidney (MDCK) cells (Friedrich Loeffler Institute, Riems, Germany) were applied in Eagle’s minimal essential medium supplemented with 2 μg/mL trypsin, 2 mM L-glutamine, and 1% nonessential amino acids.

Isolation and propagation of the working passage of influenza virus A(H1N1)pdm09 HA-G222-mpJena/5258 were published (Seidel et al., 2014).

All trial procedures and animal care activities were conducted following the German Animal Protection Law. Experiments were approved by the Thüringer Landesamt für Verbraucherschutz (Reg.-Nr.: UKJ-18-015).

For the development of the standard operating procedure (SOP) for the preparation of MA60 suspensions, the solubility of MA60 was evaluated. Based on previous studies from literature, 0.3% carboxymethyl cellulose (CMC) was used as a vehicle for preparing an MA60 extract suspension for oral application (Lim et al., 2013; Yimam et al., 2017). The 300 mg/kg dose was selected as maximum dose for the acute toxicity study. This was the maximum dose still resulting in a nearly homogenous solution (Supplementary Figure S1A) which is the prerequisite for oral gavage.

Aiming to define the dose and to exclude/reduce the risk of unacceptable side effects for experiment II (efficacy study), an acute toxicity study was performed. Twenty mice were randomly divided into four groups of five mice. One group was treated with the solvent (placebo: 0.3% CMC) and the other groups with 30, 100, and 300 mg/kg of solved MA60 (Supplementary Figure S1B). According to our previous results, influenza virus replication decreases until day 7 after infection (Seidel et al., 2014). The time schedule for MA60 administration was selected based on these observations. For the administration of MA60, oral gavage once daily was applied based on promising literature data reported for an ethanol extract from M. alba rootbark in 0.3% CMC applying 200 or 400 mg/bw/daily in a mouse model of LPS-induced airway inflammation (Lim et al., 2013).

Body weight and clinical score (weight change, physiognomy, and behavior) of mice were determined daily and considered for the determination of the MTD. Mice were sacrificed and organ tissue samples (lung, liver, heart, spleen) were aseptically removed and weighed 3 hours after the last treatment. In addition, serum was obtained from collected blood samples by centrifugation at 855 g for 15 min. The lung, liver, spleen, and serum samples of mock-infected mice were also used to establish and validate a UPLC-MS method for the quantitation of the marker compounds sanggenon C and D.

To determine the antiviral effect of MA60 in vivo, 46 mice were randomly divided into experimental groups as summarized in Supplementary Table S1. Mice were treated orally (per oral gavage) once daily for a maximum of 7 days with either the solvent (placebo: 0.3% CMC; n = 23) or 100 mg/kg of solved MA60 (n = 23). One hour after the first application of MA60, groups of mice were mock-infected (NaCl) or infected with the influenza virus in NaCl. For this procedure, mice under isoflurane anesthesia were inoculated intranasally with 20 µL NaCl or 10⁵ TCID₅₀/20 µL of HA-G222-mpJena/5258 diluted in NaCl. The selected number of mice as well as the statistical evaluation are based on literature data (Richardson and Overbaugh, 2005).

Body weight and clinical score (weight change, physiognomy, and behavior) of mice were determined at least once daily. Mice that lost more than 20% of their initial body weight over 48 h or 25% of their initial body weight and/or became severely ill were sacrificed for ethical reasons (n = 3; two placebo-treated, influenza virus-infected mice and one MA60-treated, influenza virus-infected mouse).

One hour after the second treatment (day 1 p.i.), as well as 24 h after the last treatment (day 7 p.i.), groups of mice were sacrificed and organ tissue samples (lung, liver, spleen) were aseptically removed. Serum was obtained from collected blood samples by centrifugation at 855 g for 15 min. The superior lobe of influenza virus-infected right lung was homogenized in test medium and used for virus titer determination (Reed and Muench, 1938). Another right lung lobe was frozen for analysis of cytokine mRNA levels as described in 2.6. The left lobe was fixed in 10% formalin and embedded in paraffin for histopathological evaluation. For that, sections of 4 µm were prepared and stained with hematoxylin and eosin. Further two right lung lobes, liver, spleen, and serum of mock-infected mice were used for the determination of the content of sanggenons C and D.

Lung samples were homogenized in lysis buffer and TissueLyzer beads (Qiagen, Hilden) in the TissueLyzer II (Qiagen, Hilden). Then RNA was isolated with the RNeasy^® Mini Kit (Qiagen, Hilden) according to the producer’s manual. NP-vRNA copy number was determined after reverse transcription of 1,000 ng RNA with uni12 primer (Hoffmann et al., 2001) by a quantitative PCR with two NP-gene targeting primers and a NP-plasmid standard. The expression profile of cytokines (IP10, IL-6, IFN-ß, CXCL9 and Eif2ak2) was determined after reverse transcription of 1,000 ng RNA with oligo dT primer by a semi quantitative PCR with cytokine mRNA-targeting primers. The expression of GAPDH mRNA was used as a baseline control. Results are calculated and expressed as Log2(2^-ΔΔCt)-fold change of mRNA expression related to control (Pfaffl et al., 2002). Primer sequences and PCR temperature profiles are summarized in Tables (Supplementary Tables S2–S4).

Semi-quantitative PCR results were analyzed using EXCEL 2016 software according to the method described by Pfaffl (Pfaffl, 2004). Data are presented as mean with standard deviation (SD) calculated with EXCEL 2016.

Statistical analysis of data is based on “basic statistical considerations in virological experiments” published by (Richardson and Overbaugh, 2005). The number of animals (n) per experimental group is given in Supplementary Figure S1 (acute toxicity study) and Supplementary Table S1 (efficacy study). Mice succumbing to infection before day 7 are mentioned in paragraph 2.5 and 3.2. After performing normality test (Shapiro-Wilk) and equal variance test (Brown-Forsythe), the significance of differences within multiple cardinal data sets (body weight, cytokine mRNAs) were analyzed with One-way ANOVA (normal distribution) or Kruskal–Wallis One Way Analysis of Variance on Ranks (non-normal distribution) with Sigmaplot 14.0. Virus titers and nucleoprotein gene copies of the mock-treated, influenza virus-infected and the MA60-treated influenza virus-infected group of mice were analyzed with the unpaired, two-sided Student’s t-test by using EXCEL 2016. p ≤ 0.05 was set as cut off for statistical significance.

The bio sample extraction was adapted according to the published workflow from Yuan and co-workers (Yuan et al., 2012). 50 μL of thawed serum sample was transferred to a 1.5 mL Eppendorf tube. 1,000 μL of MTBE was added and shaken for 1 h. After a centrifugation step at 13,400 rpm for 15 min, 900 µL of supernatant was removed and the solvent was evaporated until complete dryness.

To analyze tissue samples, an aliquot of approximately 200 mg of each mouse liver, the whole spleen and lung were homogenized in a Zentrimix 380R (Hettich) at 4°C for 15 min at 1,500 rpm in a separate tube by three zirconium oxide beads (Ø 3 mm) in 500 µL water. 900 μL cooled methanol (−70°C) was added and ultrasonicated for 5 min in an ice bath. After vortexing, the samples were incubated for 1.5 h at −70°C. Then, tissue homogenates were centrifuged at 15,000 rpm in a refrigerated centrifuge (4°C) for 10 min. Supernatants were collected and the incubation cycle was repeated for exhaustive extraction with 400 µL fresh methanol (−70°C) followed by ultrasonication (10 min on ice) and vortexing under cooled conditions (4°C). Samples were transferred to a refrigerator for a 30 min incubation at −70°C. After centrifugation, supernatants were combined and dried in a GeneVac EZ-2 plus using the program “Aqueous” set to 30°C. For analysis, serum, as well as tissue samples, were dissolved in 50 µL MeOH, centrifuged and 5 µL were injected in duplicates for analysis.

UHPLC-ESI-MS (Thermo Fisher Scientific, CA) analysis was performed on a Dionex UltiMate 3,000 system coupled to an LTQ XL ion trap mass spectrometer. MS detection was carried out using HESI source (300°C heater temperature, 40/10/1 arb. units for the sheath, aux and sweep gases, respectively and 3.5 kV spray voltage at 275°C capillary temperature) to achieve negative ion mode ionization. For quantitation, SIM (single ion monitoring) was applied in the negative mode. The [M-H]^- adduct 707.21 Da was used as a center mass with a 1.00 Da window. For confirmation and method validity, selective MS/MS scans of the 3 most abundant ions were achieved through collisional induced dissociation (CID) fragmentation at 30% normalized collision energy.

Separation was carried out on a Waters ACQUITY BEH phenyl column (2.1 × 100 mm, 1.7 µm) within 13 min. Solvent A: Water/formic acid (99.9:0.1), Solvent B: Acetonitril/formic acid (99.9:0.1). The following gradient was applied: 0–2 min: 30% B, 2–13 min: 30%–98% B, followed by a washing step: 13–17 min 98% B and an equilibration step for 3 min 5 μL of dissolved sample in methanol were injected. Data acquisition and evaluation was carried out using Xcalibur (version 4.27.0.19). The UHPLC-ESI-MS method was validated for specificity, linearity, precision, and extraction recovery according to ICH guidelines ³ . In terms of specificity, chromatograms of the spiked samples were compared to respective blank serum or liver samples. Calibration curves were established by serial dilution of standards of sanggenons C and D, respectively in a ratio of 1:3 in methanol (level 0: 1 mg/mL) followed by injection of each level in triplicate. The peaks were integrated using Thermo Xcalibur software using the integration algorithm ICIS. Limit of detection (LOD) and limit of quantitation (LOQ) were evaluated visually as a signal-to-noise ratio of more than 3 times (LOD) or 10 times (LOQ). Linearity was determined as the range of all included calibration levels in the regression equation. For precision, intraday (within 1 day, sample #64) and interday (over 3 days, sample #65) measurements were evaluated accordingly. To assess extraction recovery, spiking experiments were conducted as follows: Stock solutions of the extract MA60 were prepared at a concentration of 1 mg/mL. 10 μL of the stock solution was added to 50 µL of blank serum or blank liver homogenate and treated as described above. By integration of the AUC of the peak of sanggenon C (serum samples) and sanggenon D (tissue samples) and comparing it to the peak area of sanggenons C and D, respectively in a 1:5 dilution of the stock solution, the recovery rate was calculated accordingly.

MA60 contains a well-characterized amount of bioactive MDAAs of 29% (Langeder et al., 2023). Sanggenons C and D calibration curves were obtained by serial dilution down to 1.88 ng/mL (Figure 2). A quadratic fit of the calibration curves revealed correlation coefficients of R ² ≥ 0.9974 (Table 1). Validation parameters (specificity, linearity, precision, and extraction recovery) were evaluated in accordance with ICH guidelines. ⁴ The specificity was assessed by comparing the selected MS/MS spectra of serum and tissue samples to the fragment spectra of sanggenons C and D. As there was a precise matching of the standard fragment ions to the fragment ions found in serum and tissue samples, it can be concluded that the established method is specific for the marker compounds (Supplementary Figure S2).

The extraction efficiency for serum and tissue samples was probed by performing spiking experiments. 10 μL of a stock solution of MA60 (1 mg/mL) was spiked to samples of blank serum and liver. Sanggenons C and D were extracted according to the given protocol. Their peak areas were compared to the extract solution (diluted 1:5), respectively. From the spiked samples, 97% of sanggenon C in serum samples and 85% of sanggenon D in liver samples were retrieved. These results confirmed the suitability of the extraction protocol for sanggenon C in serum samples and sanggenon D in liver samples (Supplementary Figure S2). Interestingly, the extraction recoveries for sanggenon D in serum samples (84%) and sanggenon C in liver samples were significantly lower (26%), and therefore not applied as marker compounds in the respective matrices.

BALB/c mice were treated per gavage feeding with 30, 100, and 300 mg/kg of MA60 in 0.3% CMC or 0.3% CMC (placebo) once daily for 7 days to determine the MTD for the subsequent in vivo efficacy study. Daily monitoring of body weight revealed that neither treatment with 30 nor with 100 mg/kg of MA60 caused a significant body weight change compared to placebo (Figure 3A; Supplementary Table S5). In contrast, mice treated with 300 mg/kg of MA60 continuously lost body weight until day 4 of treatment (p = 0.03 at day 4 of treatment compared to the placebo-treated group). The 300 mg/kg dose of MA60 did not affect the weight of liver, lung, heart, and spleen (Supplementary Table S6). The applied tests and the results of statistical analysis of the percentage of daily body weight (% of day 0) are summarized in Supplementary Table S5. After day 4, a slight amelioration was observed.

In summary, body weight loss was the only adverse effect observed after treatment (clinical score not shown). The 300 mg/kg dose of MA60 causes more than a 10% depression in body weight gain. Thus, the 100 mg/kg dose of MA60 was identified as MTD in the acute toxicity study.

Based on the dose-finding experiment, the 100 mg/kg dose of MA60 was well tolerated when administered orally once a day for 7 days to 8-weeks-old female BALB/c mice (MA60-treated, mock-infected). Neither body weight (Figure 3B) nor lung histology (Figures 4A,C and Supplementary Figure S3A, S3C) and cytokine mRNA expression (Figure 6) were significantly affected when applying this MA60 dose (Supplementary Tables S7, S8).

In contrast, placebo-treated, influenza virus-infected mice lost body weight starting on day 2 after infection (Figure 3B). Due to critical body weight loss, one mouse of this experimental group had to be sacrificed for ethical reasons on day 5 p.i. and another one on day 6 p.i. As of day 2 p.i. or day 3 p.i., the difference in body weight was statistically significant compared to placebo-treated, mock-infected mice (p values between <0.001 and 0.020) or MA60-treated, mock-infected mice (p values between <0.001 and 0.027), respectively. The applied statistical tests and the calculated p values are summarized in Supplementary Table S7. High virus titers and nucleoprotein gene copy numbers were detected on day 1 p.i. (Figures 5A, B). Infectious virus was nearly eliminated on day 7 after virus infection (Figure 5A) whereas nucleoprotein gene copies declined more slowly (Figure 5B). The expression of cytokines was enhanced, however not significantly on the day after influenza virus infection when comparing placebo-treated, mock-infected mice with placebo-treated, influenza virus-infected mice (Figure 6; Supplementary Table S8). In contrast, a significant difference was found between both experimental groups for the expression of IL-6, IP10, and CXCL9 7 days after virus infection (Figure 6; Supplementary Table S8).

The loss of body weight occurred more slowly and was less pronounced in MA60-treated, influenza virus-infected mice compared to placebo-treated, influenza virus-infected mice (Figure 3B). The body weight differences were statistically different between day 3 and 5 p.i. (p values between 0.022 and 0.004) or day 4 p.i. (0.004) when comparing MA60-treated, influenza virus-infected mice with placebo- or MA60-treated, mock-infected mice, respectively. However, the observed difference between MA60-treated, influenza virus-infected mice and placebo-treated, influenza virus-infected mice was not significant (Supplementary Table S7). Only one mouse of this experimental group had to be sacrificed for ethical reasons on day 6 p.i. The treatment effect of 100 mg/kg MA60 administered orally was not strong enough to significantly reduce virus titers (Figure 5A) and the number of viral nucleoprotein genome copies (Figure 5B). At day 1 after infection, the cytokine expression (IL-6, CXCL-9, IP10, and IFN-ß) in lung samples of MA60-treated, influenza virus-infected mice was stronger than in all other experimental groups. The difference was not significant in comparison to the placebo-treated, influenza virus-infected group but it was significant in comparison to the placebo-treated, mock-infected and/or MA60-treated, mock-infected groups (Figure 6; Supplementary Table S8). The comparison of lung histopathology did not reveal differences between placebo-treated and MA60-treated influenza virus-infected mice (Figures 4B, D; Supplementary Figures S3B, S3D).

In the samples from the acute toxicity trial, a dose-dependency could be observed for the content of sanggenon C in serum samples (Figure 7A). In the lowest dose group (30 mg/kg) an amount of sanggenon C below 9.10 ng/mL was found; in the 100 mg/kg dose group between 7.94 ng/mL and 31.53 ng/mL, and in the group of mice treated with 300 mg/kg of MA60 between 21.21 ng/mL and 227.26 ng/mL of sanggenon C were determined (Supplementary Table S9).

In the efficacy trial (100 mg/kg dose), serum concentrations of sanggenon C were found to be between 8.00 ng/mL and 124.22 ng/mL (#64-#67) after 1 h. After 24 h, the concentrations were below 2.77 ng/mL or not detectable (#68-#71).

The amount of sanggenon D was determined in both the acute toxicity as well as the efficacy experiment in lung, liver, and spleen samples (see Figure 7B; Supplementary Table S9). In lung samples, sanggenon D was detected in relatively high amounts between 343 and 10,306 ng/mL in four samples (#29, #32, #33 and #35). In liver samples of the low dose group sanggenon D was not detected. Only in two (#32 and #35) out of five samples, a sanggenon D concentration of 34.18 ng/mL and 26.97 ng/mL was detected, respectively. However, in the high dose group (300 mg/kg) especially in liver tissues derived from #38 and #39, a concentration of almost 1 µg/mL was determined. In general, samples from mice #38-#40 resulted in high concentrations of sanggenon C (serum) and sanggenon D (liver). Comparable to the serum concentrations of sanggenon C in the infection experiment, a high amount of sanggenon D was detected when mice were sacrificed after 1 h. In contrast, sanggenon D could not be determined after 24 h in the liver. None of the marker compounds was found in spleen samples (data not shown).