For all experiments, a sample of a production batch (EXCh. 878) of EPs 7630, an extract of Pelargonium sidoides DC. roots (1:8–10), dried, extraction solvent: ethanol 11% (w/w) and fractions thereof were used. Roots of Pelargonium sidoides DC. were collected in South Africa (e.g., Eastern Cape). The dried material was tested in an array of DNA-based and phytochemical methods to confirm the quality and identity of the herbal material. Pharmacognosy was done by the quality control department of Dr. Willmar Schwabe GmbH and Co. KG. Voucher specimens of every lot are deposited in the Department of Pharmacognosy to be retained for 10 years.

For fractionation, 10 g of EPs 7630 extract was dissolved in 200 ml of an aqueous 15% ethanol (v/v) solution and fractionated by ultrafiltration, using a 50 ml Merck Millipore Amicon stirred cell with compatible Ultracel filter discs. The fractionation was carried out in a serial fashion, beginning with ultrafiltration through a filter disc with a cut-off of 30 kDa. The 200 ml extract solution was applied to the stirred cell in portions and concentrated to about 10–20 ml. Subsequently, it was washed with 3 × 30 ml aqueous 15% ethanol (v/v) solution. The retentate was collected, ethanol was evaporated on a rotary evaporator and the resulting aqueous solution was freeze-dried for further use. The filtrate of the ultrafiltration was also collected and applied to the next serial fractionation step with a filter disc of 10 kDa cut-off. The aforementioned procedure was repeated with filter discs with cut-offs of 5 kDa, 3 kDa, and 1 kDa, resulting in dry fractions of >30 kDa, 10–30 kDa, 5–10 kDa, 3–5 kDa, 1–3 kDa, and <1 kDa.

The HPLC-UV chromatograms were recorded on a Thermo Vanquish UHPLC coupled to a diode array detector (DAD). For separation, a Waters Atlantis T3 (3 μM, 2 × 150 mm) column without pre-column was used. Eluent A consisted of 2.5% (v/v) acetonitrile and 0.5% (v/v) formic acid in water. Eluent B consisted of 5% (v/v) water and 0.5% (v/v) formic acid in acetonitrile. The separation parameters were as follows: 0.2 ml/min flow rate, column temperature of 25°C, UV detection wavelength of 280 nm, and injection volume of 4 µl of a 5 mg/ml Pelargonium sidoides DC. extract EPs 7630 and 2 µl of 10 mg/ml ultrafiltration fractions, all dissolved in an aqueous 40% acetonitrile solution. The gradient was as follows: from 0.0–10.0 min linear from 0 to 5% Eluent B, from 10.0–65.0 min linear from 5 to 50% Eluent B, from 65.0 to 66.0 min linear from 50 to 100% Eluent B, from 66.0 to 71.0 min isocratic 100% Eluent B column wash, from 71.0–72.0 min linear from 100 to 0% Eluent B followed by 8 min equilibration period with 0% Eluent B, resulting in a total run time of 80.00 min.

Analysis of the chromatogram was performed using ACDLabs Spectrus Processor Software v2017.2.1. The fractions were complementarily characterized using NMR spectroscopy and gel permeation chromatography. NMR spectra were acquired on a Bruker Avance III HD System equipped with an inverse TCI Prodigy Cryoprobe. The Larmor frequencies for ¹H and ¹³C were 600 and 150 MHz, respectively. The samples >1 kDa were dissolved at concentrations of 20 mg in 600 µl DMSO-d₆. The sample <1 kDa was not soluble in DMSO-d₆ and D₂O was used for this sample instead with the same concentration of 20 mg in 600 µl. For referencing, tetramethylsilane, or trimethylsilylpropionic acid-d₄ were used, respectively, and all spectra were acquired with a sample temperature of 25°C. 1D-¹H spectra were acquired with 32 accumulated scans with a spectral width of 18 ppm, a transmitter frequency offset of 8 ppm, and a digital resolution of 64 k data points. The spectra were processed with an exponential window function with a line broadening of 0.3 Hz. ¹H-¹³C-HSQC spectra were acquired with 16 accumulated scans, 256 increments for the ¹³C dimension, and 1730 data points for the ¹H dimension. The spectral widths were 12.0 ppm at 5.5 ppm transmitter offset and 186.0 ppm at 88.0 ppm transmitter offset for ¹H and ¹³C, respectively. The spectra were processed with 1 k data points and a sine square window function with a sine bell shift of 2 for both dimensions. The spectra were recorded and processed with Bruker Topspin software (v3.6pl7) and analyzed with ACD/Labs Spectrus Processor (v2017.2.1).

Gel permeation chromatography was carried out on a Hitachi LaChrom system, coupled to a DAD. For separation, a Polymer Standards Service MCX analytical 100 Å column (5 μm, 8 × 300 mm) with a total run time of 30 min and an isocratic elution of an aqueous 40% vol. acetonitrile solution with a flow rate of 0.8 ml/min was used. The column temperature was 50°C and signals were detected by UV at 230 nm. The injection volume was 10 μl at a sample concentration of 4.5 mg/ml in elution solvent. For calibration of the retention times, epicatechin 0.8 mg/ml (monomer), procyanidin B2 1.0 mg/ml (dimer), and procyanidin C1 0.8 mg/ml (trimer) were used, respectively. Standards for higher oligomeric proanthocyanidins were not commercially available. The molecular weight of the sample was estimated by extrapolating the elution volumes of the mono, di and trimer.

VeroFM (ATCC CCL-81), VeroE6 (ATCC CRL-1586), and Calu-3 (ATCC HTB-55) cells were grown in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 1% penicillin/streptomycin, 1% non-essential amino acids, and 1% sodium pyruvate at 37°C and 5% CO₂. All cell lines were cultivated under sterile laboratory conditions and tested for simian virus 5 and mycoplasma contamination.

The SARS-CoV-2 strain Munich/2020/984 (BetaCoV/Munich/BavPat1/2020) was isolated from a respiratory swab obtained from the early 2020 Munich patient cohort (GenBank: MT270101; GISAID: EPI_ISL_406,862). For virus infection with SARS-CoV-2, 2x10e5 to 3x10e5 cells/ml were seeded in 6-well plates or 24-well plates. After 24 h, cells were infected with SARS-CoV-2 in a serum-free medium. After 1 h, virus dilutions were removed, and the wells were washed twice with PBS and refilled with DMEM (supplemented as described previously). Samples were taken at the indicated time points. Infection with SARS-CoV (isolate Hong Kong/6109) (Hattermann et al., 2005) and MERS-CoV (Human betacoronavirus 2c EMC/2012; Genbank: JX869059.2), as well as SARS-CoV-2 Alpha (BetaCoV/Baden-Wuerttemberg/ChVir21528/2020; EPI_ISL_754,174) and Beta (Baden-Wuerttemberg/ChVir22131/2021; EPI_ISL_862,149) variants were performed in an analogous procedure using the same multiplicity of infection (MOI). All virus infection experiments were conducted under biosafety level 3 conditions with enhanced respiratory personal protection equipment.

Calu-3 and VeroE6 cells were seeded in 24-well plates at a seeding density of 3x10e5 and 2x10e5 cells/well, respectively. 2 h before virus infection, cells were washed once with PBS and pretreated with DMEM containing the indicated substances or DMSO as vehicle control for niclosamide. Virus infections were performed on ice and cells were transferred to 4°C for 30 min after the addition of virus inoculum to allow virus attachment without receptor-mediated virus entry. Cells were subsequently washed five times with cold PBS to remove unbound virus particles. For 0 h p. i. samples, the infected cells were immediately lysed in external lysis buffer (Roche). For 4 h p. i. samples, infected cells were further incubated at 37°C in DMEM containing the same substances and concentrations used for pretreatment. RNA isolation and RT-qPCR were performed as described earlier (Dagotto et al., 2021; Schroeder et al., 2021). For viral subgenomic mRNA (sgmRNA) quantification of the N gene, an MOI of 1 was used.

SARS-CoV-2, mumps virus, and Rift Valley fever virus clone 13 (Muller et al., 1995) plaque-forming units (PFU) were quantified by plaque titration on VeroE6 cells as described before (Dulbecco, 1952; Herzog et al., 2008; Forcic et al., 2010). Briefly, monolayers of VeroE6 cells were seeded in 24-well plates with ∼90% confluency and washed with PBS, incubated with serial dilutions of SARS-CoV-2-containing cell culture supernatants, and overlaid with 1.2% Avicel in DMEM 24 h after seeding. 72 h post-infection, cells were fixed with 6% formalin and visualized by crystal violet staining. The assay-specific cut-off is 10 PFU/ml. All titration experiments were performed in duplicate wells.

Cellular RNA of diverse immune genes was quantified by RT-qPCR using the Superscript III OneStep RT-PCR (Thermo Fisher Scientific) kit and the oligonucleotides listed in Supplementary Table S1. Total RNA was isolated from compound-treated or SARS-CoV-2-infected Calu-3 cells by automated pipetting using the MagNA Pure 96 extraction system (Roche). Gene expression was calculated relative to TBP reference gene expression using the ΔΔCT method (Livak and Schmittgen, 2001). SARS-CoV-2 N gene sgmRNA was quantified in RNA extracts from infected VeroE6 and Calu-3 cells using RT-qPCR and TBP as reference gene. Gene expression was calculated relative to TBP reference gene expression using the ΔΔCT method (Livak and Schmittgen, 2001). To monitor virus growth, SARS-CoV-2 RNA was quantified in cell culture supernatants by RT-qPCR targeting the SARS-CoV-2 E gene, as described before (Corman et al., 2020).

SARS-CoV-2 spike (S) protein-dependent viral entry was assessed using an established vesicular stomatitis virus (VSV) pseudo-particle (VSVpp) assay as described elsewhere (Kleine-Weber et al., 2019; Hoffmann et al., 2020a; Zettl et al., 2020). Briefly, VeroFM and Calu-3 cells were seeded in DMEM in 96-well plates with a density of 50–70% 24 h before infection with VSVpp. For pretreatment, medium was removed 2 h before infection and replaced by fresh DMEM containing EPs 7630, fractions of EPs 7630, camostat mesylate, niclosamide, or DMSO as vehicle control for niclosamide in indicated concentrations. After pre-incubation, medium was removed and fresh DMEM containing VSVpp carrying the SARS-CoV-2 spike (BetaCoV/Munich/BavPat1/2020) or pCG1 vector control was added to the cells. Plates were then centrifuged for 30 min at 4°C and 300 x g to achieve synchronized infection. After additional incubation for 90 min at 37°C, 5% CO₂, compound-containing medium was added to the cells. To measure luciferase production, which correlates with successful viral entry, cell lysates were prepared after 16 h (VeroFM) or 24 h (Calu-3) using passive lysis buffer (Promega). Lysates were then transferred to opaque 96-well plates and luminescence was measured in a multi-well plate reader (Berthold) using Luciferase Assay Substrate (Promega) according to the manufacturer’s recommendations.

To assess cytokine levels in Calu-3 cell supernatant, 25 µl of supernatant were sampled before infection and at 8 and 48 h post-infection with SARS-CoV-2. Cytokines were quantified using a Human Cytokine/Chemokine/Growth Factor Panel A 48-Plex Premixed Magnetic Bead Multiplex Assay (Merck Millipore), using the Luminex MAGPIX System in 96-well plate format, according to the manufacturer’s instructions. Plate washing steps were performed using the HydroFlex Microplate Washer (Tecan). Calibration and verification checks were met for all of the analytes. Analytes for which quality controls were out of the expected range or for which standard curves exhibited an R ² value of <0.80 were excluded from the dataset. All remaining analytes had standard curves with R ² values > 0.90, except for IL-9 (R ² = 0.84).

The viability of EPs 7630- or fraction-treated VeroFM and Calu-3 cells was assessed using the CellTiter-Glo 2.0 Cell Viability Assay (Promega) according to the manufacturer’s instructions. Briefly, cells were seeded in 96-well plates and treated with the indicated concentrations of EPs 7630 or ultracentrifugation fractions. After 48 h cells were lysed and the luminescence signal was measured using a multi-well plate reader (Berthold). Viability was calculated in relation to untreated cells and reported as percent of control.

Previous studies showed that respiratory viruses such as common cold CoV and Influenza A virus can be inhibited by EPs 7630 treatment in vitro (Michaelis et al., 2011; Theisen and Muller, 2012), while in vivo treatment of influenza-infected mice with EPs 7630 led to a reduction of viral burden, attenuation of body weight loss, and improved survival (Theisen and Muller, 2012). In patients with common cold, the EPs 7630 treatment outcomes assessed as reduction of symptoms were as favorable in patients with confirmed HCoV infection as in patients with other viral infections (Keck et al., 2021). To explore putative inhibitory in vitro effects of EPs 7630 on highly pathogenic CoV (SARS-CoV, MERS-CoV, and SARS-CoV-2), we infected IFN-deficient (VeroFM) monkey kidney cells and IFN competent human lung cells (Calu-3) with a low MOI (0.0005) in the absence and presence of EPs 7630 (10, 100 μg/ml). Infectious virus particles in cell culture supernatants were determined by virus plaque assay at 48 h post-infection. EPs 7630 significantly inhibited propagation of all highly pathogenic CoV at 100 μg/ml at 48 h post-infection by up to >99% (Figure 1A). In the case of SARS-CoV-2, significant growth inhibition was already observed using 10 μg/ml in both cell lines. Since clinical applications and repurposing approaches rely on low inhibitory concentrations, we determined IC50 values for SARS-CoV-2 in both cell lines. The EPs 7630-specific IC50 for SARS-CoV-2 inhibition was 0.48 μg/ml in VeroFM cells and 1.61 μg/ml in Calu-3 cells (Figure 1B, green) being well within non-cytotoxic ranges (Figure 1B, yellow) and highly comparable to previous studies (Michaelis et al., 2011; Walther et al., 2020). To verify the antiviral activity of EPs 7630 against newly emerging SARS-CoV-2 variants, we analyzed its impact on SARS-CoV-2 Alpha and SARS-CoV-2 Beta propagation, both of which were classified as variants of concern (VOC) by the World Health Organization (Abdool Karim and de Oliveira, 2021; Konings et al., 2021). We found substantial inhibition of SARS-CoV-2 VOC propagation, comparable to the original Munich patient isolate in Calu-3 cells using 100 μg/ml EPs 7630 (Supplementary Figure S1A). At low concentrations of 10 μg/ml EPs 7630, we did not detect a significant reduction of viral RNA levels, possibly as a consequence of increased replicative fitness reported for the analyzed VOCs (Pyke et al., 2021; Rosenke et al., 2021; Touret et al., 2021). To exclude unspecific EPs 7630-induced inhibitory effects in Calu-3 cells, virus growth of other enveloped viruses, specifically mumps virus (MuV) and a Rift Valley fever virus (RVFV) clone 13-based reporter virus (Kuri et al., 2010) was tested in the absence and presence of EPs 7630 (Supplementary Figure S1B). Whereas RVFV growth was significantly reduced at treatment concentration 100 μg/ml, mumps virus, in analogy to measles virus (Michaelis et al., 2011), was not inhibited by EPs 7630.

As EPs 7630 was shown to inhibit virus entry and release (Theisen and Muller, 2012), we next analyzed the SARS-CoV-2 cellular entry process upon EPs 7630 treatment. SARS-CoV-2 enters cells via ACE2 receptor-mediated endocytosis and direct TMPRSS2-mediated fusion of ACE2-bound SARS-CoV-2 particles with the plasma membrane (Hoffmann et al., 2020a). To model the two entry pathways, we infected TMPRSS2-negative Vero cells (endosomal entry) and TMPRSS2-positive Calu-3 cells (endosomal entry and direct fusion) with SARS-CoV-2 spike (SARS-CoV-2-S) protein-carrying VSV-based pseudo particles (VSVpp). As controls, we applied niclosamide (10 µM), which blocks SARS-CoV-2 replication and possibly restricts endosomal entry (Jurgeit et al., 2012; Prabhakara et al., 2021; Gassen et al., 2021), and camostat mesylate (100 µM), a proven TMPRSS2 inhibitor (Hoffmann et al., 2020a). SARS-CoV-2-S VSVpp entry was efficiently blocked (>98%) by niclosamide in both cell lines (Figure 2A, unprocessed data and controls in Supplementary Figure S2) whereas, expectedly, camostat inhibited VSVpp entry only in TMPRSS2-positive Calu-3 cells. EPs 7630 inhibited SARS-CoV-2-S VSVpp entry more efficiently in Calu-3 cells (59%) as compared to Vero cells (17%) suggesting that the endosomal, as well as the plasma membrane fusion-mediated entry processes, are affected. To further confirm this finding, we analyzed EPs 7630-mediated entry inhibition using infectious SARS-CoV-2 as previously described (Dagotto et al., 2021; Schroeder et al., 2021). Synchronized virus infection of cells with a high MOI (MOI = 1) was followed by PCR-based detection of subgenomic (sg) SARS-CoV-2 nucleocapsid (N) mRNA at 4 h post-infection. Differences in N gene sgmRNA levels at this early time point allowed us to monitor subtle differences in virus entry. In agreement with the VSVpp data, we observed a significant entry inhibition upon niclosamide-treatment in both cell lines, whereas camostat exclusively inhibited SARS-CoV-2 entry in Calu-3 cells (Figure 2B). EPs 7630 clearly showed more efficient entry inhibition in Calu-3 as compared to Vero cells confirming the VSVpp-based data.

EPs 7630 comprises a multitude of molecules including carbohydrates, minerals, peptides, purine derivatives, highly substituted benzopyranones, and oligo- and polymeric prodelphinidins (Schoetz et al., 2008). To identify the chemical substance classes mainly responsible for the observed SARS-CoV-2 entry inhibition, we generated extract fractions by sequential ultrafiltration that contain substances with different molecular weight ranges. The fractions generated by ultrafiltration were analyzed using HPLC-UV, gel permeation chromatography (GPC), and NMR spectroscopy. HPLC analysis was used to monitor the purine derivatives and benzopyranones in the fractions according to the LC-MS assignment reported by Roth et al. (Roth et al., 2021). Polymeric and oligomeric prodelphinidins appeared as broad signals in HPLC analysis, whereas carbohydrates were not detectable due to the absence of chromophores. HPLC analysis revealed that the fractionation by ultrafiltration was successful, yielding major amounts of the small molecule fraction (gallocatechin/epigallocatechin, purine derivatives, benzopyranones) in the <1 kDa fraction (Table 2, Supplementary Figure S3). To estimate the oligomerization degree of the prodelphinidins, a GPC analysis was applied, calibrated by epicatechin (monomer), procyanidin B2 (dimer), and procyanidin C1 (trimer). The prodelphinidins appeared as broad signals in GPC as well, representing the distribution of different oligomerization degrees in the fractions. Comparison and extrapolation of the elution times of the calibration substances indicated the following distribution: fraction 1–3 kDa: di-to tetramers, fraction 3–5 kDa: di-to pentamers, fraction 5–10 kDa: tri-to hexamers, and fraction 10–30 kDa: tetra-to about decamers. The fraction >30 kDa consisted mainly of polymers eluting in the void volume (Supplementary Figure S4). 1D and 2D NMR spectroscopy was used for the detection of carbohydrates and solvent residue. 1D NMR analysis demonstrated a major amount of mono- and dimeric carbohydrates in the <1 kDa fraction (Supplementary Figure S5). According to the intensities of the NMR signal patterns, carbohydrates exceeded the amounts of benzopyranones by approximately five-fold, which is consistent with the constituent ratios described previously (Schoetz et al., 2008). An exact quantitative analysis was not pursued, since this is outside the scope of the present study.

The characterization of the oligomerization degree of the prodelphinidin was additionally assessed by 2D NMR spectroscopy. In this regard, the integrals of H/C-4 and H/C-4_term in ¹H-¹³C-HSQC spectra were determined, which represent the number of monomer building blocks that are within the oligomer chain and terminal building blocks, respectively. Since H/C-4_term is a methylene group consisting of two protons in contrast to one proton involved in the H/C-4 group, the value of H/C-4_term was divided by two and normalized to a value of 1.0. The integrals of H/C-4 were normalized with the same factor as H/C-4_term. To estimate the mean oligomerization degree, the two normalized integral values were added. In the spectrum of the fraction >30 kDa, the H/C-4 and H/C-4_term signals of prodelphinidins had a suboptimal signal-to-noise ratio, and the signals of this fraction were not integrated. The results of the assessment of the oligomerization degree of the fractions 1–30 kDa by NMR analysis are summarized in Table 1. Apart from some minor amounts of small molecules in the 1–3 kDa fraction, the four fractions between 1–30 kDa were largely free of small molecules, consisting mainly of prodelphinidins of different oligomerization degrees and minor amounts of polymeric carbohydrates of undefined oligomerization degree (Figure 3). The >30 kDa fraction contained mainly polymeric prodelphinidins. Interestingly, also umckalin and umckalin sulfate, as well as minor amounts of purine derivatives, were found in this fraction, as shown in the HPLC chromatograms (Supplementary Figure S3). Although not expected by the fractionation strategy, this finding may be due to non-covalent interactions of umckalin (sulfate) and purine derivatives with putative secondary structures of polymeric prodelphinidins or carbohydrates. The results of the phytochemical analyses including a mass balance are summarized in Table 2. All fractions contained small amounts of solvent residue, namely water and ethanol, as detected by NMR analysis, which may contribute to the excess mass (0.2 g) of totaled fractions.

Next, we aimed to analyze the putative differential inhibitory activity of the 6 different EPs 7630 ultrafiltration fractions on SARS-CoV-2 entry. Pretreated Calu-3 cells were synchronously infected with SARS-CoV-2-S VSVpp in the presence of 10 and 100 μg/ml of the respective fraction during inoculation and post-infection. All fractions showed significant entry inhibition at high concentrations of 100 μg/ml (Figure 4A). Fractions 10–30, 5–10, 1-3, and <1 kDa also showed significant entry inhibition at concentrations of 10 μg/ml, with fraction 1–3 kDa showing the highest efficiency. Confirmatory experiments using infectious SARS-CoV-2 showed significant growth inhibition at 10 μg/ml for all 6 fractions (Figure 4B). Interestingly, medium to high molecular weight fractions (3–5, 5–10, 10–30, and >30 kDa) inhibited virus growth completely at 100 μg/ml whereas low molecular weight fractions 1-3 and <1 kDa reached a maximum inhibition of approximately 90% at 10 and 66 μg/ml, respectively, indicating differential activity of the fractions. Cell viability assays ruled out that cytotoxic effects were responsible for differential inhibition using the indicated concentrations (Supplementary Figure S6).

Apart from inhibiting virus entry, EPs 7630 has immunomodulatory effects (Noldner and Schotz, 2007; Peric et al., 2021) that might be beneficial for counteracting virus infections and preventing inflammation and immune dysregulation (cytokine storm), which is associated with high COVID-19 morbidity and mortality. The variable SARS-CoV-2 inhibition of the different EPs 7630 fractions (Figure 4B) encouraged us to compare immunomodulatory effects of EPs 7630 and its fractions in SARS-CoV-2-infected Calu-3 cells. Whereas EPs 7630 or its fractions (each 100 μg/ml) alone (Figure 5, green bar; Supplementary Figure S7) had no major effects on pro-inflammatory (CCL5, IL6, IL1B), IFN-dependent (IFNB1, IFIT1, MX1), or anti-inflammatory (TNFAIP3) gene expression, SARS-CoV-2 infection resulted in 10 to 100-fold increased expression of all genes except TNFAIP3 at 48 h post-infection (Figure 5, red bars). EPs 7630 treatment post-SARS-CoV-2 infection resulted in a significant reduction of IL1B gene expression and strong upregulation of anti-inflammatory TNFAIP3. All other genes showed a limited but non-significant decrease in gene activation suggesting either limited transcriptional regulation or restoration of mRNA levels late in infection. Notably, reduced virus growth as a consequence of entry inhibition (see Figures 2,4) might also generally limit the SARS-CoV-2-induced upregulation of immune genes. Still, the low molecular weight fractions <1 and 1–3 kDa, which had limited effects on virus propagation using 100 μg/ml (see Figure 4B), resembled EPs 7630-dependent gene regulation patterns with the strongest effects on IL1B and TNFAIP3 during SARS-CoV-2 infection. The 4 fractions between 3 and >30 kDa had minor effects on anti-inflammatory TNFAIP3, but strong inhibitory effects on most of the pro-inflammatory and IFN-dependent genes. The differential gene activation patterns might be explained by the composition of the fractions containing gallocatechins, benzopyranones such as umckalin and umckalin sulfate, and purine derivatives in the two low molecular weight fractions, and increasing amounts of di-, tri-, hexa-, oligo-, and polymeric prodelphinidins in the high molecular weight fractions (Table 1).

The strong EPs 7630-induced inhibition of pro-inflammatory IL1B induction and upregulation of anti-inflammatory TNFAIP3 late post-infection encouraged us to have a closer look at the cytokine secretion profile of epithelial Calu-3 cells. Cytokine production at the primary site of SARS-CoV-2 infection, i.e. the lung epithelium, is crucial for activation of immune cells and priming of inflammatory responses, thereby affecting the outcome of disease (Chua et al., 2020; Mathew et al., 2020; Islam et al., 2021).

Luminex-based cytokine detection was done early (8 h) and late (48 h) post-SARS-CoV-2 infection with and without EPs 7630 treatment (100 μg/ml). We detected pronounced production of pro-inflammatory cytokines, chemokines, and immunomodulating growth factors in SARS-CoV-2 infected Calu-3 cells, most of which are associated with COVID-19-induced cytokine storm (Figure 6, right panel: I, II, III, red bars) (Cauchois et al., 2020; Donlan et al., 2021; Pang et al., 2021; Petrey et al., 2021; Sugiyama et al., 2021). During the course of infection, the concentration of all analyzed proteins increased at 48 h p. i., analogously to SARS-CoV-2 viral loads in Calu-3 cells (Figure 1). Interestingly, concomitant treatment of infected cells with EPs 7630 strongly reduced the production of distinct cytokine subsets. Protein levels of key inflammatory cytokines like IL-8, IL-13, IL-18, or TNF-α were greatly reduced in supernatants of SARS-CoV-2 infected Calu-3 cells that were simultaneously treated with EPs 7630. In contrast, we noticed an EPs 7630-mediated increase in IL-1β and IL-6 levels early post-infection (8 h), with enhanced IL-6 levels also detectable late after infection (48 h). These effects were also found in EPs 7630-treated Calu-3 cells in the absence of virus infection, highlighting the versatile immunomodulating activities of EPs 7630 beyond cytokine suppression (Supplementary Figure S8).

In the case of chemokines (Figure 6, right panel: II) and growth factors (Figure 6, right panel: III), which coordinate immune cell attraction and infiltration to the site of infection, we consistently found reduced levels in cell supernatants from SARS-CoV-2-infected and EPs 7630-treated cells. Particularly late upon infection (48 h), reduction of secreted proteins ranged from 75.6% reduction for PDGF-AB/BB to 95.8% reduction for VEGF-A. Importantly, chemokines and growth factors that are associated with COVID-19-induced cytokine storm and severe disease progression like CXCL9, CXCL10 (IP10), or VEGF-A exhibited particularly effective EPs 7630-mediated inhibition (CXCL9: 82% reduction; CXCL10: 80.9% reduction) (Costela-Ruiz et al., 2020; Petrey et al., 2021; Sugiyama et al., 2021). Whereas the strongest EPs 7630-mediated cytokine reduction was detected at 48 h p. i., multiple proteins showed significant inhibition in the early phase of SARS-CoV-2 replication at 8 h p. i. (IL-8, TNF-a, CCL22, CXCL1, CXCL9, FGF-2, PDFG-AA/BB, VEGF-A), hinting towards an intrinsic capacity of EPs 7630 to dampen virus-induced inflammation. Indeed, with few exceptions, EPs 7630 was capable of significantly reducing baseline cytokine production in the absence of virus (Supplementary Figure S8, green bars). In contrast, EPs 7630 induced the production of immune-regulating cytokines with a role in inflammation resolution and immune homeostasis in SARS-CoV-2 infected cells (Figure 6, right panel: IV), except for IL-1RA, which correlates with COVID-19 severity and was reduced in our analysis (Zhao et al., 2020). Interestingly, this effect of EPs 7630 treatment was also detected in non-infected cells, boosting the production of immune-regulating cytokines IL-9, IL-12, and IL-15, and potentially setting the course for a more balanced immune response towards virus infection in epithelial cells (Supplementary Figure S8, IV., bottom panel).