As of January 2023, there have been over 668 million cases of the coronavirus disease COVID-19, caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), across the globe, causing over 1.1 million deaths in the USA and 6.7 million worldwide1. During the pandemic, many individuals have turned to herbal supplements to prevent COVID-192. There are published in silico studies and a few in vitro studies on these extracts3,4,5,6,7,8,9,10,11,12,13,14,15, but the science to support the use of these botanicals to prevent viral infection remains incomplete9,10,13,16. Our group has conducted field studies in global terrestrial biodiversity hotspots and has amassed a large, targeted collection of plant and fungal species used in traditional medicine for general health (as medicinal foods) and for the treatment of infectious and inflammatory disease17,18,19. Viral entry—in which SARS-CoV-2 attaches to the angiotensin-converting enzyme 2 (ACE2) cell surface receptor found on endothelial cells, pneumocytes (type 1 and 2), and ciliated bronchial epithelial cells—presents an attractive option for preventatives20. This is the first extensive investigation of botanical ingredients used in traditional food and medicine systems for their efficacy as viral entry inhibitors for SARS-CoV-2 since the virus emerged in Wuhan, Hubei Province, China, in late 201921.
SARS-CoV-2 is an enveloped RNA virus with a viral spike protein that binds to ACE2 on host cells. Once bound to ACE2, SARS-CoV-2 may enter into the cell through fusion or endocytosis20. Additional host cell membrane proteins may alter the spike protein in order to bind to ACE2 (such as TMPRSS2)20. For endocytosis, chaperone proteins such as clathrin are also recruited to the membrane to encourage membrane formation and entry into the cell22.
There is an urgent need for readily accessible immediate dissemination of orally available therapeutic agents to target viral entry and replication. Since its emergence, various edible and medicinal plants have been proposed as a therapeutic strategy for protecting against infection and treating symptoms9,10,11,12,13. There is emerging data that over-the-counter nutritional and dietary supplements may harbor anti-COVID-19 properties3,4,5,6,7,14,15, but supporting confirmatory data is still largely lacking. Some natural products (NPs) from traditional Chinese medicine (TCM) preparations are reported to inhibit viral entry and modulate host immune responses23. Americans do not typically utilize TCM; however, annually 18–30% of Americans report using botanical dietary supplements24,25, with retail sales over $8 billion in 201726. As Americans are looking to prevent infection, many are hoping dietary herbal interventions can be beneficial13. Therefore, probing botanical extract libraries can provide further guidance to members of the public seeking to prevent COVID-19 infection through supplements and potentially lessen the severity of COVID-19 infections.
Our group has conducted field studies in the USA, Mediterranean, Africa, Asia, and the Balkans for more than a decade and has amassed a large collection of plant specimens and their extracts which were used medicinally by the local people for general health (such as for medicinal foods) and for the treatment of infectious and inflammatory disease18,19,27,28,29,30,31. All 728 species currently in the collection are uniquely linked to ethnobotanical data on their preparation and use in traditional medicine (TM). In total, 2500+ extracts (2497 plant extracts, 83 macrofungal extracts, two algal extracts) representing different tissues from these species, are included in the Quave Natural Products Library (QNPL). Many of the extracts included in the QNPL are created from wild harvested specimens, following WHO guidelines for appropriate plant collection and with necessary permits and permissions32,33. The QNPL also has a collection of botanical materials, representing the 40 top selling botanical dietary supplement ingredients26 which were acquired from commercial sources, vouchered, botanically authenticated and then extracted. This study utilizes the QNPL as a tool to identify potential natural products with viral entry inhibition effects against SARS-CoV-2 (Supplementary Material 1).
Screening the QNPL for promising SARS-CoV-2 inhibitors
A total of 1867 extracts total from the QNPL derived from 660 species (1 Chromista, 27 Fungi, 632 Plantae kingdoms) across 149 families were screened for viral entry inhibition and mammalian cytotoxicity, as well as an additional 18 single compounds that are predominant in botanicals. Hydroxychloroquine (HCQ) was used in the screen as a positive control of viral entry inhibition via the spike-ACE2 complex34. Of these, 310 extracts derived from 188 species across 76 families (3 fungi, 73 plants) exhibited ≥ 50% inhibition activity in the wild-type spike pseudotyped model (Fig. 1; Supplementary Materials 2–4). Of these bioactive extracts, 125 extracts derived from 93 plant species across 53 families exhibited ≥ 85% inhibition activity and ≤ 15% cytotoxicity in the wild-type model (Supplementary Material 2). Once these 125 extracts were identified, an interesting pattern emerged, indicating many hits were from species that are known to be cardiotoxic due to a rich composition of cardiac glycosides, including members of the genera Asclepias and Nerium (Apocynaceae family), Kalanchoe (Crassulaceae), and Drimia (Asparagaceae). For further selection and testing, we reviewed each extract and consulted the literature to eliminate extracts with those cardiac glycosides or similar compounds.
Figure 1 Phylogenetic distribution of plant species tested for SARS-CoV-2 viral entry inhibition. A total of 1867 extracts derived from 660 species (1 Chromista, 27 Fungi, 632 Plantae) across 149 families in the Quave Natural Products Library of plant extracts were screened at a concentration of 20 µg/mL for viral entry inhibition using a wild-type spike (SARS-CoV-2, GeneBank #QHD43416.1) pseudotyped lentivirus model with human embryonic kidney cells expressing human angiotensin converting enzyme 2 (HEK-293T-hACE2). The phylogenetic distribution of plant species demonstrating viral entry inhibition (at ≥ 50% at the screening concentration) were mapped on the family-level maximally resolved complete euphyllophyte phylogenetic tree according to Angiosperm Phylogeny Group IV68. Major clades are indicated in different colors and family names are only listed for those genera reported in the present study. Gray bars indicate the percentage of species tested that demonstrated ≥ 50% bioactivity in each family. A high-resolution version of this figure is provided as Supplementary File 4. Full size image
Phylogenetic assessment
A phylogenetic assessment of the 185 plant species demonstrating bioactivity (≥ 50% inhibition) identified predominant activity in the asterid families and many species in Fagales and Rosales within the fabid clade of plants (Fig. 1, Supplementary Material 5). A considerable proportion of eudicot families included in this study (57 out of 92 families) exhibited bioactivity. Among those, the fabid clade predominates in terms of number of species that showed viral entry inhibition activity. Fabales, Fagales, Rosales and Malpighiales are the major orders in the fabid clade. Fabaceae is the largest family in this clade, with nine individual species from the QNPL demonstrating bioactivity; however, this represents a relatively small percentage compared to the total number of species included in this study (n = 42) and total species diversity of the family. Fagales is a small clade, with about 1175 species and all four families included in the study showed activity. Five out of six Rosales families demonstrated bioactivity. In total, 58 species from Rosaceae and Fagaceae are included in this study, and 50% of these demonstrated bioactivity (Fig. 1, Supplementary Material 4). Only the Lamiales and some Asteraceae families demonstrated activity in the asterid clade. Lamiid is one of the largest clades in the asterids and 10 out of 15 families studied showed significant antiviral activity. Among the lamiid families, Apocynaceae (n = 6) and Lamiaceae (n = 5) species demonstrated bioactivity. Asteraceae is the largest angiosperm plant family; however, only 12 of 64 species tested from Asteraceae showed activity.
Outside of the asterid and fabid clades, there were a few distinct small families that also demonstrated bioactivity. Two out of nine fern families and two gymnosperms that are included in the study demonstrated activity. Only three out of seven magnoliid families exhibited activity. Bioactivity was low across the monocot species assessed (8 out of 21). Poaceae—one of the largest plant groups in the monocots with high species diversity—had only three species out of nine that exhibited inhibition activity. Only 24 species out of 61 tested in the malvid clade showed bioactivity, with 18 species belonging to Malvales and Sapindales orders. Cistaceae is a small family of the Malvales with 270 species, and all six species included in this study demonstrated bioactivity. The majority of Anacardiaceae (four out of six) and Sapindaceae (four out of five) species from the Sapindales included in the study also demonstrated bioactivity.
Further testing of hit extracts against emerging variants
Concentration-dependent response assays against the wild-type spike pseudotyped model were performed at 2 to 64 µg/mL on nine extracts representing seven species of interest. Initial concentration response assays demonstrated that three extracts exhibited potent antiviral activity, low cytotoxicity, a lack of cardiotoxicity concerns based on the literature, and were not identified as common allergens (Table 1, Supplementary Material 6): Solidago altissima L. (extract 1428); Salix nigra Marshall (extract 1749); and Pteridium aquilinum (L.) Kuhn (extract 1804). These extracts were then tested for cell viability, cytotoxicity by lactate dehydrogenase (LDH) assay, and viral entry inhibition at 2 to 128 µg/mL on HEK-293T-hACE2 and HaCaT cells (Table 2). Extracts 1428, 1749, and 1804 demonstrated robust activity at lower concentrations, as well as minimal cytotoxicity at higher concentrations. Concentration-dependent testing of these three extracts in pseudotyped variants (Delta/B.1.617.2, Alpha/B.1.1.7, Gamma/P.1, and Beta/B.1.351) demonstrated bioactivity as well (Table 3, Fig. 2) with EC 50 values all below 10 µg/mL. In summary, all three lead extracts exhibited activity in pseudotyped models of the wild-type virus and the four variants tested.
Table 1 EC 50 values were calculated by modeled dose–response curves against the wild type SARS-CoV-2 pseudovirion model using non-linear regression. Full size table
Table 2 EC 50 values were calculated by modeled dose–response curves against the wild-type SARS-CoV-2 pseudovirion model using non-linear regression. Full size table
Table 3 EC 50 values were calculated by modeled dose–response curves against the wild type SARS-CoV-2 pseudovirion model using non-linear regression. Full size table
Figure 2 All variants were tested at a titer of 1.2 × 105: Gamma/P.1 variant (BPS Biosciences #78144-1), Beta/B.1.351 variant (BPS Biosciences #78142-1), Delta/B.1.617.2 variant (BPS Biosciences #78215-1), and Alpha/B.1.1.7 variant (BPS Biosciences #78112-1). Concentration-dependent response data (2–128 μg/mL) against SARS-CoV-2 variants for three select extracts (1428, 1749, and 1804) identified as leads for antiviral activity in the SARS-CoV-2 pseudovirion model. Hydroxychloroquine was not included due to study limitations, refer to Tables 1 and 2 for the EC 50 tested against the wild-type pseudotype virus. Full size image
Antiviral activity shown in infectious virus
To determine the potential antiviral effects of the top three extracts against in vitro replication of SARS-CoV-2 in cell culture, a confluent monolayer of African Green Monkey kidney (Vero) cells in a 96-well cell culture microplate was treated with 20 μg/mL of compound followed by inoculation with 0.1 multiplicity of infection (MOI) of the virus. This concentration was selected to be consistent with the pseudotyped virus screening against HEK-293T-hACE2 cells. Cells were treated with 12 μg/mL remdesivir as a positive control. Extract 1428 and extract 1804 demonstrated > 99% inhibition at 20 μg/mL in Vero cells, while extract 1749 only demonstrated 45.9% inhibition. The antiviral activity of extract 1428 and extract 1804 was further confirmed by virus yield reduction assay using specific qRT-PCR for SARS-CoV-2 by measuring the RNA copy number of the virus after 2-days post-treatment (for Vero cells) and after 3-days post-treatment (for Calu-3 and Caco-2 cells) in supernatant of treated-infected cells in a concentration–response manner (Table 4, Fig. 3). Extracts 1428 and 1804 showed CC 50 values of > 100 and > 80 μg/mL respectively, in MTS assays against human peripheral blood mononuclear (PBM) and Vero cells.
Table 4 The antiviral activity of 1804 and 1428 has been further confirmed by virus yield reduction assay using specific qRT-PCR for SARS-CoV-2 to measure the RNA copy number of the virus either 2 days post-treatment (for Vero cells) or 3 days post-treatment (for Calu-3 and Caco-2 cells) in treated-infected cells in a dose response manner. Full size table
Figure 3 Infectious virus results in different cell lines. (A) The antiviral activity of 1804 and 1428 has been further confirmed by virus yield reduction assay using specific qRT-PCR for SARS-CoV-2 to measure the RNA copy number of the virus either 2 days post-treatment (for Vero cells) or 3 days post-treatment (for Calu-3 and Caco-2 cells) in treated-infected cells in a dose response manner. Remdesivir was used as a positive control for both assays. (B) Cytotoxicity assays measuring cell proliferation in PBM and Vero cells were performed in parallel to the antiviral assays. Cycloheximide, a known protein synthesis inhibitor, was used as the positive control. Full size image
Chemical analysis of top hits
Chemical analysis of these top two candidates was performed using high-resolution mass spectrometry. The major metabolites from both extracts were tentatively identified based on the MS/MS fragmentation data compared with literature, in silico prediction, and web-based databases (Fig. 4, Supplementary Materials 7–9). In extract 1804 (from Pteridium aquilinum), a variety of metabolites were detected, including phenylpropanoids, proanthocyanidins, flavonoids, and triterpenes. In extract 1428 (from Solidago altissima), phenylpropanoids, glycosidic triterpenoids, and fatty acids were detected as major chemical classes.