Discovery and structural optimization of 3-O-β-chacotriosyl oleanane-type triterpenoids as potent entry inhibitors of SARS-CoV-2 virus infections.

Currently, SARS-CoV-2 virus is an emerging pathogen that has posed a serious threat to public health worldwide. However, no agents have been approved to treat SARS-CoV-2 infections to date, underscoring the great need for effective and practical therapies for SARS-CoV-2 outbreaks. We reported that a focused screen of OA saponins identified 3-O-β-chacotriosyl OA benzyl ester 2 as a novel small molecule inhibitor of SARS-CoV-2 virus entry, via binding to SARS-CoV-2 glycoprotein (S). We performed structure-activity relationship profiling of 2 and discovered C-17-COOH of OA was an important modification site that improved both inhibitor potency toward SARS-CoV-2 and selectivity index. Then optimization from hit to lead resulted in a potent fusion inhibitor 12f displaying strong inhibition against infectious SARS-CoV-2 with an IC50 value of 0.97 μM in vitro. Mechanism studies confirmed that inhibition of SARS-CoV-2 viral entry of 12f was mediated by the direct interaction with SARS-CoV-2 S2 subunit to block membrane fusion. These 3-O-β-chacotriosyl OA amide saponins are suitable for further optimization as SARS-CoV-2 entry inhibitors with the potential to be developed as therapeutic agents for the treatment of SARS-CoV-2 virus infections.

Introduction

As with other well-known coronaviruses (CoVs) such as severe acute respiratory syndrome CoV (SARS-CoV) and Middle East respiratory syndrome CoV (MERS-CoV), severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a new enveloped positive-stranded RNA virus [[1], [2], [3], [4]]. SARS-CoV-2 is able to cause high pathogenic and apathogenic human infections because of the possibility of an aerosol mode of transmission, the high fatality rate, as well as the unpredictable nature of the outbreaks, posing a serious threaten to public health and the global economy [[5], [6], [7]]. The outbreak of novel coronavirus disease 2019 (COVID-19) caused by SARS-CoV-2 has resulted in over 61.8 million infections and over 1.4 million deaths worldwide [8], which underscores the need for drug discovery and development efforts to obtain available treatments. There are no clinically effective vaccines for the prevention of COVID-19 infections though several SARS-CoV-2-specific vaccines have entered a Phase III trial [9,10]. In addition, only a few compounds such as lopinavir/ritonavir, chloroquine, hydroxychloroquine and remdesivir have advanced to clinical trials, but they displayed the limited curative effect and toxic side effects for the treatment of COVID-19 patients [11,12]. Thus, there is still an urgent need to develop specific anti-SARS-CoV-2 agents offering efficacy and safety, which can be used alone or in combination with vaccines in future infections.

The surface structural spike glycoprotein (S) of SARS-CoV-2 virus plays a vital role in mediating virus entry into host cell that is the first step in the infection process [13,14]. Structurally, the SARS-CoV-2 S glycoprotein exists as a homotrimer with multiple glycosylation sites and consists of two subunits, an N-terminal S1 subunit and a C-terminal S2 subunit. After SARS-CoV-2 S1 subunit recognizes the cellular ACE2 receptor via receptor-binding domain (RBD), S2 subunit will change its conformation to regulate S-mediated viral/cell membrane fusion [15,16]. Blocking S glycoprotein can prevent SARS-CoV-2 entry into the cell and downstream replication processes, which provides important clues for overcoming SARS-CoV-2.

The SARS-CoV-2 S protein is of particular interest for antiviral development due to its ability in bonding to the human ACE2 receptor, generating membrane fusion events that make possible for the virus to penetrate host human cells [17,18]. Recently, certain antibodies have been found to exhibit potency toward SARS-CoV-2 in vitro, which directly bond to the S1 subunit to inhibit SARS-CoV-2 [[19], [20], [21]]. As is the case for effective SARS-CoV-2-S antibodies, great efforts have been made to identify specific small-molecule entry inhibitors of SARS-CoV-2 on the basis of multiple high-throughput virtual screens of various small-molecule libraries. To our best knowledge, only limited examples of inhibitors that bind directly to the SARS-CoV-2 S protein in enzymatic assays have been reported up to now, including glycyrrhizic acid (GL, Fig. 1 A) that are suggested to block the interaction between RBD domain of the SARS-CoV-2 S1 subunit and ACE2 [22], and Salvianolic acid C (Sal-C, Fig. 1A) that was found to block the formation of six-helix bundle (6-HB) core of S2 protein to inhibit membrane fusion [23]. Despite the fact that more data are needed to prove their antiviral efficacy in vitro or in vivo, the discovery of these known antibodies and small-molecule entry inhibitors highlights SARS-CoV-2 S protein as a promising target for the development of novel anti-SARS-CoV-2 therapeutics.

Fig. 1A

Representative antiviral agents as SARS-CoV-2 entry inhibitors.

Oleanolic acid (OA), is the most well-known oleanane-type pentacyclic triterpene, of which analogs/derivatives have been confirmed in many studies to exert a variety of antiviral activities against human immunodeficiency virus (HIV), influenza A virus (IAV), Ebola virus (EBOV) and hepatitis C virus (HCV) [[24], [25], [26], [27]]. Interestingly, certain OA derivatives are found to display potential inhibition toward IAV entry as does GL [28,29]. Further investigations suggested that these triterpenoids are able to directly bind to multiple viral glycoproteins including GP41 of HIV, hemagglutinin (HA) of influenza, and GP of EBOV, thereby blocking virus-host fusion to disrupt viral entry into the host cells [27]. In particular, CoVs belong to “Type I″ fusion proteins as do other well-known enveloped viruses such as HIV, IAV, and EBOV. Evidence has revealed that these enveloped viruses might have the similar entry mechanism in spite of different cell receptors and diverse glycoproteins [30,31]. Based on these results, it can be assumed that OA derivatives is a potential scaffold for the discovery of potential entry inhibitors of SARS-CoV-2 virus.

From the last decade, our research group has been involved in the discovery and development of novel IAV entry inhibitors structurally relate to OA trisaccharide saponins [27,28,32], which are capable of blocking HA2-mediated membrane fusion. Herein, we report a focused screen of these OA saponins using an efficient S/HIV pseudotyping system with the aim of obtaining potent anti-SARS-CoV-2 inhibitors. This screen identified two potent SARS-CoV-2 entry inhibitors 1 and 2 (Fig. 1 B). Notably, 3β-O-[2, 4-di-O-(α-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oic acid benzyl ester (2) showed good inhibition toward SARS-CoV-2 with an IC50 value of 16.75 μM in vitro though having moderate cytotoxicity. Then optimization of 2 led to a series of novel SARS-CoV-2 entry inhibitors devoid of cytotoxicity. Furthermore, mechanism studies revealed that these OA saponins were identified as potential small-molecular fusion inhibitors that could inhibit SARS-CoV-2 infection through directly binding to the SARS-CoV-2 S2 subunit. This finding has provided validation for the continued development of the novel SARS-CoV-2 entry inhibitors based on pentacyclic triterpene saponins scaffolds.

Fig. 1B

Potent SARS-CoV-2 entry inhibitors 1 and 2 from our screening.

Results and discussion

Screening potential OA saponins against pseudovirus model of SARS-CoV-2

Our previous screen of a pentacyclic triterpene saponin library identified a series of 3-O-β-chacotriosyl oleanane-type triterpenoids as potential H5N1 entry inhibitors [27,28]. In view of their significant inhibitory effect on IAV, we were sought to investigate whether the observed anti-IAV properties of these OA triterpenoids could extend to SARS-CoV-2 and translate into a shared mechanism that was capable of chemically antagonizing the SARS-CoV-2-cell fusion. Notably, one of the challenges of working with the highly pathogenic SARS-CoV-2 virus is that biosafety level 3 (BSL-3) facilities are required to handle this infectious virus due to the safety concerns. For studies of other highly pathogenic enveloped viruses such as HIV, H5N1 and EBOV/MARV, this obstacle has been circumvented by a surrogate pseudovirus (PsV) system, which can be widely utilized to not only identify antiviral agents but also study the entry mechanisms [[33], [34], [35]]. It has been proved that viral PsV systems for CoVs such as SARS-CoV and MERS-CoV are also valid surrogate assays [36,37]. Based on our previous study [13], we utilized a robust HIV pseudotype model with the SARS-CoV-2-S to screen the OA triterpenoids library for discovering entry inhibitors. This S/HIV pseudotype model allowed for direct comparison of SARS-CoV-2-S function with a common lentiviral core and reporter. Here all OA saponins were screened at 20 μM (Table S1), where hits with superior to 50% inhibition rate were retested to obtain accurate IC50 values. Interestingly, several hit compounds have been shown to inhibit the entry of SARS-CoV-2. Among them, 2 as a top hit, exhibited strongest inhibitory activity (IC50 = 16.75 μM) while it had moderate cytotoxicity against 293T-ACE2 cells. Structurally, the inhibitor 2 possesses a α-l-rhamnopyranosyl-(1 → 2)-[α-l-rhamnopyranosyl-(1 → 4)]-β-d-glucopyranosyl residue hydrophilic trisaccharide, named known as a chacotriose, attached to the hydrophobic aglycone OA via β-glycosidic linkage, followed by a benzyl side chain at 17-COOH of OA.

The preliminary structure-activity relationships (SARs)

As depicted in Table S1, these OA saponins were found to show a variety of inhibitory activities against SARS-CoV-2, strongly related to their structures. A preliminary meaningful structure-activity relationships (SARs) could be drawn from these observations to support further chemical optimization, as outlined in Fig. 2 . We first analyzed the effect of branched trisaccharide scaffold of 2 on the inhibition. It was found that removal or positional change of its α-l-rhamnosyl residues as well as replacement of its α-l-rhamnosyl or β-d-glucopyranosyl residue with other monosaccharides, resulted in a significantly diminished inhibition and even a complete loss of potency. These results suggested that the chacotriosyl moiety was essential for inhibitor efficacy. Then evaluation focused on the side chain portion at C-28 position of OA indicated that esterification of the 17-COOH group of OA was helpful for inhibitory activity since deletion of the side chains led to a remarkably reduced inhibition toward SARS-CoV-2. Moreover the enhancement of potency appeared to be correlated to the length and bulk of the substituents at C-28 position of OA, signifying that these bulky side chains may extend to a potential auxiliary hydrophobic pocket to form hydrophobic interaction. Support for this hypothesis can be seen in the reported ester analogs 1 and 2 that dramatically enhance inhibitory activity in comparison with those bearing polar groups at C-28 position of OA. Among the COOH-substituted analogs, introduction of the benzyl group was superior to other alkyl groups, highlighting the superiority of the aryl side chain at C-28 position of OA, which probably allowed either favorable hydrophobic interactions or π-π interactions in the binding site.

Fig. 2

The preliminary SARs and design strategy of OA saponins derivatives against SARS-CoV-2.

New SARS-CoV-2 entry inhibitor design

Current work in our group is committed to the development of novel and potent SARS-CoV-2 entry inhibitors, resulting in the discovery of the high hit 2. However, all these hits displayed moderate cytotoxicity against 293T-ACE2 cells, revealing that further modification might gain better SARS-CoV-2 entry inhibitors with increased potency and drug-like properties. To ablate toxic side effects, we incorporated in our chemical optimization campaign modifications to the preferred hit 2 by the bioisosteric replacement strategy. Accordingly, we attempted to adopt the amide linker to replace the ester linkage at C-17 position of OA, aiming to enhance selectivity index. In parallel to bioisosteric replacement to remove cytotoxicity, modifications were made to enhance affinity for the SARS-CoV-2-S. Our strategy here was to probe and optimize the side chain at 28-position of the hit 2 to identify detailed SARs that could be utilized to improve the potency against SARS-CoV-2, by introducing various phenyl and aralkyl groups at the 28-position Of OA via amide linker. Thus, we designed and synthesized a series of C-28 amide-substituted 3-O-β-chacotriosyl derivatives (Fig. 2, Table 1 ), anticipating that phenyl and aralkyl substituted groups at 28-position of OA would have sufficient length and size, relative flexibility, as well as appropriate hydrophobicity to bind effectively to the potential hydrophobic pocket. Given the importance of the benzyl group as an excellent privileged moiety in our preliminary screening data, we inferred reasonably that these design and optimization strategies would be fruitful.

Table 1

Inhibitory activities of saponins 12a-12p against infection of 293T-ACE2 cells by SARS-CoV-2 PsV.

Comp.	R1	R2	IC50 (μM)	CC50 (μM)	SI
12a	H	phenyl	18.12 ± 1.28	33.46 ± 1.31	1.85
12b	H	2-methoxyphenyl	5.54 ± 0.45	31.24 ± 1.63	5.64
12c	H	3-methoxyphenyl	8.33 ± 0.99	>100.00	>12.00
12d	H	4-methoxyphenyl	9.97 ± 1.09	>100.00	>10.03
12e	H	2-chlorophenyl	6.27 ± 0.84	31.74 ± 1.11	5.06
12f	H	benzyl	4.37 ± 0.82	>100.00	>22.88
12g	H	2-pyridylmethyl	>20.00	NT	NT
12h	H	2-furylmethyl	>20.00	NT	NT
12i	H	2-thienylmethyl	5.89 ± 0.78	34.87 ± 1.32	5.92
12j	Me	benzyl	7.76 ± 0.69	37.56 ± 1.22	4.84
12k	H	2-fluorobenzyl	5.68 ± 0.78	>100.00	>17.61
12l	H	2-chlorobenzyl	6.32 ± 1.11	>100.00	>15.82
12 m	H	2-bromobenzyl	9.57 ± 0.65	>100.00	>8.51
12n	H	2-methoxybenzyl	16.75 ± 1.77	33.42 ± 1.02	2.00
12°	H	2-phenylethyl	7.37 ± 0.81	>100.00	>13.57
12p	H	3-phenylpropyl	13.06 ± 0.66	>100.00	>7.66
2			16.75 ± 1.11	21.63 ± 1.37	1.29
Sal-C			3.85 ± 0.38	>100.00	>25.97

IC50: 50% inhibitory concentration.

CC50: 50% cellular cytotoxicity concentration.

SI: selectivity index as CC50/IC50.

Chemistry

To investigate substitutions at 28-position of the OA scaffold, a set of OA saponins amide analogs 12a-12p were synthesized, containing various aryl moieties, as shown in Scheme 1 . Allylation of commercially available OA with allyl bromide and potassium carbonate in DMF was performed to yield 3. Subsequent glycosylation of 3 with the known 2,3,4,6-tetra-O-benzoyl-d-glucopyranosyl trichloroacetimidate (4) [34] in the presence of TMSOTf afforded the complete β-stereoselectivity product 5 that was attributed to the participation of the neighboring group in a stereoselective reaction at C-2 position in Glu. Removal of all the Bz protecting groups was performed with sodium methoxide in methanol to furnish the intermediate 6. Then regioselective protection of C-6-OH and C-3-OH in Glu with pivaloyl (Piv) group was completed using pivaloyl chloride, leading to the important intermediate 7. Synthesis of the trisaccharide saponin 9 was performed via a coupling reaction with 7 and the known donor 2,3,4-tri-O-acetyl-l-rhamnopyranosyl trichloroacetimidate (8) [38] by means of a similar TMSOTf-catalyzed glycosylation procedure, followed by deprotection of the Ac and Piv groups using NaOH. Then hydrogenolysis of Allyl moiety in 9 was completed using palladium chloride dissolved in THF-MeOH to yield 10. Next, the obtained 10 was subjected to esterification reaction with acetic anhydride in the presence of DMAP to generate the key intermediate 11. The sequence of acyl chloride formation, installation of a variety of amines by utilizing the traditional condensation procedure, and final removal of Ac groups was followed to produce the desired title saponins 12a-12p, respectively.

Scheme 1

Reagents and conditions: (a) AllylBr, K2CO3, DMF; (b) TMSOTf, CH2Cl2; (c) MeONa, CH3OH; (d) PivCl, pyridine; (e) (i) TMSOTf, CH2Cl2; (ii) MeONa, MeOH; (f) PdCl2, MeOH-THF; (g) Ac2O, DMAP, pyridine; (h) (i) (COCl)2, CH2Cl2; (ii) aromatic or aromatic alkyl amines, Et3N, CH2Cl2; (iii) MeONa, MeOH.

Inhibition of the infection of SARS-CoV-2 pseudovirus

All title saponins 12a-12p were tested in 293T-ACE2 cells infected by a SARS-CoV-2-S/HIV pseudotype virus containing a luciferase reporter gene, while VSV-G/HIV pseudoviral transduction was used as a specificity control to exclude inhibitory effect on post-entry for HIV infection. The biological data were summarized in Table 1. Sal-C, a small molecule entry inhibitor previously shown to target SARS-CoV-2-S [23], was used as the positive control in this assay. As shown in Table 1, all the saponins except 12g and 12h displayed inhibition against SARS-CoV-2 pseudovirus with a potency ranging from moderate (IC50 > 10 μM) to potent (IC50 < 10 μM).

Among them, 12f exhibited the strongest inhibitory activity in a dose-dependent manner (Fig. 3 A) with an IC50 value of 4.37 μM on 293T-ACE2 cells stably expressing human-ACE2, while no inhibition was observed on VSV-G pseudoviral transduction (Fig. 3A). Moreover 12f showed negligible cytotoxicity against 293T-ACE2 cells (Fig. 3B). Likewise, other representative saponins 12b, 12i and 12k showed significant inhibitory activity (IC50 < 6 μM, Table 1) but had no effect on VSV-G enveloped pseudovirus (Fig. S1, SI), suggesting that saponins were determined to inhibit the entry of SARS-CoV-2 PsV on 293T-ACE2 cells. Based on these results, we inferred that saponins 12a-12p might block the entry of SARS-CoV-2 virus by targeting S protein, the only glycoprotein enveloped in the pseudovirus system. Support for this hypothesis can be seen in the saponin 12f that showed strong binding affinity to S-trimer of SARS-CoV-2 with a KD value of 0.94 μM on the basis of a surface plasmon resonance (SPR) assay (Fig. 3C).

Fig. 3

(A) Dose-response curve for saponin 12f in the antiviral assay against SARS-CoV-2 PsV and VSV-G PsV. (B) Cytotoxicity of saponin 12f against 293T-ACE2 cells. (C) SPR analysis of the interaction between 12f with SARS-CoV-2 S-trimer.

Taking into account both potency and selectivity index, 12f stood out among all title saponins evaluated, displaying excellent anti-SARS-CoV-2 activity and low cytotoxicity to 293T-ACE2 cells, which could be chosen as the lead compound for SARS-CoV-2 entry inhibitors.

Further SARs development

The goal of this SARs was to improve anti-SARS-CoV-2 activity observed with the hit 2 and explore the significance of amide substitutions at 28-position of OA with respect to SARS-CoV-2 inhibition and selectivity index. As expected, replacement of the ester with its bioisoster amide led to increased inhibition while little cytotoxic activity toward 293T-ACE2 cells (Table 1). It started with the replacement of the benzyl ester in hit 2 with aryl amide to vary the amine substituent. Phenyl amide analogs 12a-12e demonstrated high potency against SARS-CoV-2 with IC50 < 20 μM, suggesting that substitutions at 28-position bearing highly hydrophobic and aromatic nature are preferred that was consistent with earlier finding. To confirm the advantage of substituents introduced to the 28-position of OA, 12a was synthesized to represent the minimal scaffold without substituents attached. It was found that substitution with a phenyl ring (12a) resulted in the similar inhibition but slight decreased cytotoxicity compared to 2. Interestingly, decoration of the phenyl substituents (12b-12e) led to markedly improved inhibition toward SARS-CoV-2 compared to 12a, revealing the importance of structural modifications of phenyl ring for antiviral activity. In comparison to the unsubstituted derivative 12a, antiviral activity was remarkably improved when a methoxy group or chlorine atom was attached to the ortho-position of the phenyl ring while keeping the similar cytotoxicity. In contrast, the attachment of methoxy group to the phenyl ring at the meta- or para-position resulted in enhanced potency and selectivity index though with slightly less intensity than the substituted at the ortho-position.

Next, we focused our attention on novel aralkyl-substituted derivatives (12f-12p) at C-28 position in which the phenyl group of 12a was replaced with relatively long and flexible aralkyl groups containing one to three methylene bridges to probe the effect of length and steric bulk at this position. Encouragingly, a 28-benzyl-substituted analog 12f is significantly more potent than 12a, with a good IC50 value of 4.37 μM and high selectivity index (>22.88). This data revealed that addition of a methylene bridge was beneficial to better flexibility to adapt to the cavity in SARS-CoV-2-S. Inspired by those results, we then employed a bioisosterism strategy to design heteroaromatic aimde derivatives 12g-12i, with the aim of improving the antiviral potency. However, a pyridine-ring analogue of 12g showed a loss of potency against SARS-CoV-2, possibly owing to addition of a polar group to the core of hydrophobic ring, in accordance with the above-established SARs. Likewise, 2-furylmethylsubstituted analogue 12h also lacked potency. Interestingly, the anti-SARS-CoV-2 potency of 2-thienylmethylsubstituted 12i was not significantly right-shifted in relation to 12f, in spite of high cytotoxicity. The impact of N, N-disubstitution was determined with compound 12j. Unfortunately, addition of a methyl moiety at N group on our scaffold led to reduced inhibition of SARS-CoV-2 but stronger cytotoxicity compared to the corresponding monosubstituted 12f.

Given that the N-monosubstituted 12f with the benzyl substituent turned out to be the most potent inhibitor in this subseries 12f-12j, we chose to explore some further modifications targeting the benzyl ring such as the introduction at ortho-position of halogen atoms, as in saponins 12k-12m, introduction of methoxy groups, as in compound 12n, and the introduction of one or two methyl groups in the linker, as in saponins 12o-12p. Considering the inhibitory activity of saponins 12k-12n, it is noted that substitutions in the benzyl ring gave less active compounds relative to the unsubstituted analog 12f. However, some SARs were identified in this series. In general, the halogen substitutions in the benzyl ring did not contribute to the antiviral activity. Despite a fluorine derivative 12k (IC50 = 5.68 μM) showed the comparable potency with 12f, a reduction in the antiviral activity was observed with the increased atomic radius of their halogen atoms attached to ortho-position in the benzyl ring. Also, the introduction of a methoxy group at ortho-position in the benzyl ring (12n) was unfavorable (IC50 = 16.75 μM), conversely coupled with improved cytotoxicity, which supported the electronic characteristics of these groups positively influences inhibitory activity. It seemed that increasing number of methylene has a negative effect on the potency. For example, replacement with a two-carbon chains between the phenyl moiety and the amide, as in 12o (IC50 = 7.37 μM), led to a slight reduction in potency, while a longer linker, as in 12p (IC50 = 13.06 μM), caused a 3-fold drop in potency compared to 12f. These results suggest that either too large (phenylpropyl) or too rigid (phenyl) N-substituted groups are unfavorable for inhibitory activity toward SARS-CoV-2. As observed, not only the electronic characteristics of substituents in aromatic ring but also hydrophobic effects of the substituents at 28-position of OA are responsible for the efficiency of anti-SARS-CoV-2 activity.

Confirmation of the lead 12f against infectious SARS-CoV-2

To validate potency of the lead 12f, we tested it against wild-type SARS-CoV-2 (wuhan-HU-1 variant) strain where chloroquine (CQ) was used as a positive control [23]. Specifically, we examined the protective effects of 12f on Vero-E6 cells from highly pathogenic SARS-CoV-2 virus infection via observing virus distribution profile reflected by viral nucleoprotein (NP) expression utilizing immunofluorescence microscopy (IFA) at 48 hpi. As shown in Fig. 4 A, compound 12f was capable of inhibiting SARS-CoV-2 infection in a dose-dependent manner as do the control group, confirming the anti-SARS-CoV-2 efficacy of 12f and the validity of the pseudotype model. Consistently, 12f exhibited the potent antiviral activity with an IC50 value of 0.97 μM as observed in Fig. 4B. Moreover, 12f had negligible cytotoxicity against Vero-E6 cells (Fig. 4C) that was agreement with that in 293T-ACE2 cells. Collectively, the saponin 12f was an improved lead compound in relation to the hit 2, which was a good candidate as a potential antiviral treatment for SARS-CoV-2.

Fig. 4

(A) Antiviral activity of 12f against SARS-CoV-2 infection was detected by indirect immunofluorescence assay with chloroquine (CQ) as a positive control, scale bar = 200 μm. (B) Dose-response curve for 12f in the authentic SARS-CoV-2 virus infection assay. (C) Cytotoxicity of 12f against Vero-E6 cells.

12f does not block binding with S1 subunit of SARS-CoV-2-S

The spike protein of SARS-CoV-2 is consisted of S1 and S2 subunits, which is responsible for recognition of host cells and virus-cell membrane fusion, respectively. To investigate the region (s) in S protein of 12f, we conducted a Co-Immunoprecipitation (Co-IP) assay to detect the effect of 12f on SARS-CoV-2 S1 subunit binding to human ACE2. As shown in Fig. 5 A, compound 12f displayed no effect on inhibition of SARS-CoV-2 S binding to the ACE2 receptor, demonstrating that SARS-CoV-2 S1 subunit was not a potential binding site of 12f. Consistent with the Co-IP results, 12f showed much lower binding affinity with S1 subunit (at millimolar level, Fig. 5B) compared to that with S protein based on SPR. This data further confirmed that 12f did not mainly target S1 subunit of SARS-CoV-2. Additionally, we determined the binding affinities between 12f and S2 subunit. On the contrary, 12f exhibited a more than 1000-fold higher affinity with S2 subunit (K D = 2.16 μM, Fig. 5C) in relation to S1 subunit (K D = 3.69 mM, Fig. 5B). Taken together, these data suggested that 12f has the tendency to bind to the regions in S2 subunit of SARS-CoV-2 S protein.

Fig. 5

(A) The binding of SARS-CoV-2 S protein and ACE2 (Anti-Flag) in the presence or absence of 12f (40 μM) were detected by Co-IP assays. IgG was included as a negative control. (B–C) SPR analysis of the interaction between 12f with SARS-CoV-2 S1 subunit (B) and S2 subunit (C).

12f could block cell-cell fusion mediated by SARS-CoV-2 S protein

With the aim of further confirming mode of action of 12f, we next utilized the cell-cell fusion assay mediated by SARS-CoV-2 S protein to probe whether 12f is a potential fusion inhibitor based on our previous studies on seeking for h-CoVs fusion inhibitors [39]. As shown in Fig. 6 , 12f could remarkably inhibit the membrane fusion of Vero-E6 cells mediated by SARS-CoV-2 S protein in a dose-dependent manner, of which the result was consistent with our SPR data. On the basis of these results, we inferred that 12f, as a potential small-molecular fusion inhibitor, could inhibit SARS-CoV-2 infection by binding to the conserved SARS-CoV-2 S2 subunit to prevent membrane fusion, thus blocking viral entry, although the specific binding characteristics of SARS-CoV-2 S2 subunit with 12f remained to be explored.

Fig. 6

The dose-dependent inhibitory efficacy of 12f in the cell-cell fusion assay.

Conclusion

This study presented here discovered several new SARS-CoV-2 entry inhibitors, via a focused screen of the OA saponin library against pseudotyped SARS-CoV-2. Based on the structure of the top hit 2, rational drug design and subsequent chemical optimization resulted in the development of a potent set of SARS-CoV-2 fusion inhibitors with 3-O-β-chacotriosyl OA amide skeleton. The lead compound 12f had favorable SI values when tested with infectious SARS-CoV-2. Intensive SARs indicated that the chacotriosyl moiety was indispensable for efficacy and introduction of a hydrophobic aromatic ring through a flexible amide linker at 28-position of OA favored antiviral activity and selectivity index. Utilizing the Co-IP assay, SPR analysis and cell-cell fusion studies, we confirmed that the lead 12f was capable of direct binding to SARS-CoV-2 S2 subunit to block membrane fusion. Overall, 3-O-β-chacotriosyl OA amide saponins represent a promising scaffold for the further development as a potential treatment for SARS-CoV-2 infections.

Experimental protocols

General methods

Solvents were purified in a conventional manner. Thin layer chromatography (TLC) was performed on precoated E. Merck silica gel 60 F254 plates. Flash column chromatography was performed on silica gel (200–300 mesh, Qingdao, China). 1H NMR and 13C NMR spectra were taken on a JEOL JNM-ECP 600 spectrometer with tetramethylsilane as an internal standard, and chemical shifts are recorded in ppm values. Mass spectra were recorded on a Q-TOF Global mass spectrometer.

3-Hydroxy-olean-12-en-28-oic acid allyl ester (3)

To a solution of OA (2.00 g, 4.38 mmol) in dry DMF (50 mL) was added potassium carbonate (1.20 g, 8.76 mmol). After 4 h of stirring at r.t., allyl bromide (0.96 mL, 13.14 mmol) was added and the mixture was stirred for an additional 8 h. The solvents were evaporated in vacuo and the residue was dissolved in a mixture of CH2Cl2 (150 mL), washed with brine (2 × 100 mL). The organic layer was dried over anhydrous Na2SO4, filtered, and concentrated under diminished pressure. Recrystallization from EtOH provided 3 (2.00 g, 92%) as a colorless solid; m. p.: 220.8–222.3 °C; 1H NMR (600 MHz, CDCl3): δ 5.96–5.82 (m, 1H, CH CH2), 5.35–5.25 (m, 2H, CH CH2-1, H-12), 5.19 (d, 1H, J = 10.5 Hz, CH CH2-2), 4.51 (t, J = 5.3 Hz, 2H, OCH 2), 3.20 (dd, 1H, J = 11.1, 4.4 Hz, H-3), 2.87 (dd, 1H, J = 13.6, 3.9 Hz, H-18), 1.12, 0.97, 0.91, 0.89, 0.87, 0.76, 0.71 (each s, each 3H, each CH3); 13C NMR (150 MHz, CDCl3): δ 177.4 (C-28), 143.7 (C-13), 132.5 (C-12), 122.4, 117.7, 79.0, 64.8, 55.2, 47.6, 46.7, 45.9, 41.7, 41.3, 39.3, 38.8, 38.4, 37.0, 33.9, 33.1, 32.7, 32.4, 30.7, 28.1, 27.7, 27.2, 25.9, 23.6, 23.4, 23.1, 18.3, 17.0, 15.6, 15.3; HRMS (ESI) m/z calcd for C33H53O3 [M+H]+ 497.3971, found 497.3978.

3β-O-(2,3,4,6-tetra-O-benzoyl-β-d-glucopyranosyl)-olean-12-en-28-oic acid allyl ester (5)

To a solution of 3 (2.00 g, 4.03 mmol), 2,3,4,6-tetra-O-benzoyl-d-glucopyranosyl trichloroacetimidate (4) (3.59 g, 4.84 mmol) and 4 Å molecular sieves in dry CH2Cl2 was added TMSOTf (133 μL g, 0.60 mmol) at 0 °C under argon. The reaction mixture was stirred for 1 h and warmed to r.t. for 1 h. The reaction was quenched by Et3N and concentrated. The residue was purified by silica gel column chromatography (petroleum ether-EtOAc, 6:1) to afford 5 (4.03 g, 93%) as a colorless solid; m. p.: 232.1–233.6 °C; 1H NMR (600 MHz, CDCl3): δ 8.04–8.00 (m, 2H, Ar–H), 7.96–7.89 (m, 4H, Ar–H), 7.86–7.81 (m, 2H, Ar–H), 7.57–7.46 (m, 3H, Ar–H), 7.45–7.32 (m, 7H, Ar–H), 7.29 (d, 2H, J = 7.9 Hz, Ar–H), 5.91 (t, 1H, J = 9.7 Hz, H-3′), 5.92–5.83 (m, 1H, CH CH2), 5.58 (t, 1H, J = 9.7 Hz, H-4′), 5.57 (dd, 1H, J = 9.8, 7.9 Hz, H-2′), 5.35–5.26 (m, 2H, CH CH2-1, H-12), 5.19 (dd, 1H, J = 10.4, 1.4 Hz, CH CH2-2), 4.85 (d, 1H, J = 7.9 Hz, H-1′), 4.62–4.49 (m, 4H, H-6′, OCH 2), 4.18–4.10 (m, 1H, H-5′), 3.10 (dd, 1H, J = 11.7, 4.5 Hz, H-3), 2.88 (dd, 1H, J = 13.7, 3.7 Hz, H-18), 1.09), 0.93, 0.91, 0.82, 0.68, 0.67, 0.62 (each s, each 3H, each CH3); 13C NMR (150 MHz, CDCl3): δ 177.3 (C-28), 166.0, 165.9, 165.3, 165.0, 143.7 (C-13), 133.5, 133.2, 133.1 (two), 132.5 (C-12), 129.9 (two), 129.8 (two), 129.7 (five), 129.4, 128.8, 128.8, 128.4 (four), 128.3 (four), 122.4, 117.7, 103.3 (C-1′), 90.8, 73.0, 72.1, 72.0, 70.3, 64.8, 63.4, 55.4, 47.6, 46.7, 45.9, 41.6, 41.3, 39.3, 38.7, 38.2, 36.6, 33.9, 33.2, 32.6, 32.4, 30.7, 27.7 (two), 25.9, 25.8, 23.7, 23.4, 23.0, 18.1, 16.9, 16.2, 15.2; HRMS (ESI) m/z calcd for C67H79O12 [M+H]+ 1075.5505, found 1075.5546.

3β-O-(β-d-glucopyranosyl)-olean-12-en-28-oic acid allyl ester (6)

Compound 5 (4.03 g, 3.75 mmol) was dissolved in CH2Cl2 and MeOH (V:V = 1:1, 80 mL) and then NaOMe was added until pH = 10, respectively. After stirred at r.t. overnight, the solution was neutralized with Dowex 50 × 8 (H+) resin until pH = 7, filtered and concentrated in vacuo. Then the residue was purified by silica gel column chromatography (CH2Cl2–MeOH, 10:1) to give 6 (2.35 g, 95%) as a colorless solid; m. p.: 153.8–155.3 °C; 1H NMR (600 MHz, CDCl3): δ 5.98–5.86 (m, 1H, CH CH2), 5.38–5.28 (m, 2H, CH CH2-1, H-12), 5.23 (dd, 1H, J = 10.5, 1.2 Hz, CH CH2-2), 4.61–4.49 (m, 2H, OCH 2), 4.35 (d, 1H, J = 7.1 Hz, H-1′), 3.56–3.50 (m, 2H, H-6′), 3.65–3.56 (m, 1H, H-4′), 3.56–3.50 (m, 1H, H-3′), 3.47–3.38 (m, 1H, H-2′), 3.27 (d, 1H, J = 11.4 Hz, H-3), 3.20–3.10 (m, 1H, H-5′), 2.91 (dd, 1H, J = 13.3, 2.7 Hz, H-18), 1.14, 1.01, 0.95, 0.93, 0.91, 0.82, 0.74 (each s, each 3H, each CH3); 13C NMR (150 MHz, CDCl3): δ 178.3 (C-28), 144.3 (C-13), 133.1 (C-12), 123.3, 117.6 (two), 106.1 (C-1′), 90.1, 77.6, 77.0, 75.0, 71.0, 65.4, 62.1, 56.4, 47.5, 46.5, 42.2, 42.1, 40.1, 39.5, 39.2, 37.2, 34.2, 33.4, 33.1, 32.9, 31.0, 28.1, 27.9, 26.4, 25.8, 23.9, 23.5, 23.3, 18.7, 17.2, 16.4, 15.4; HRMS (ESI) m/z calcd for C39H63O8 [M+H]+ 659.4478, found 659.4434.

3β-O-(3,6-di-O-pivaloyl-β-d-glucopyranosyl)-olean-12-en-28-oic acid allyl ester (7)

To a cold solution of compound 6 (2.35 g, 3.56 mmol) in dry CH2Cl2 and pyridine (V:V = 1:1, 100 mL), pivaloyl chloride (2.19 mL, 17.8 mmol) dissolved in dry CH2Cl2 (5 mL) was added dropwise over a period of 10 min at −10 °C and the reaction mixture was allowed to stir for another 12 h. After the reaction was complete as judged by TLC, the mixture was concentrated in vacuo. Then the residue was diluted with EtOAc (100 mL) and washed with 1 M HCl (50 mL × 3), saturated aq NaHCO3 (50 mL × 3), and brine (50 mL). The organic layer was dried over Na2SO4, filtered, and concentrated under diminished pressure. Purification of the crude reaction product by column chromatography on silica gel (petroleum ether-EtOAc, 3:1) furnished 7 (2.15 g, 73%) as a colorless solid; m. p.: 114.7–116.4 °C; 1H NMR (600 MHz, CDCl3): δ 5.93–5.84 (m, 1H, CH CH2), 5.30 (d, 2H, J = 17.7 Hz, CH CH2-1, H-12), 5.19 (d, 1H, J = 10.5 Hz, CH CH2-2), 4.88 (t, 1H, J = 9.2 Hz, H-3′), 4.52 (d, 1H, J = 5.1 Hz, OCH 2-1), 4.50 (d, 1H, J = 5.0 Hz, OCH 2-2), 4.45 (dd, 1H, J = 11.8, 1.9 Hz, H-6′-1), 4.40 (d, 1H, J = 7.8 Hz, H-1′), 4.17 (dd, 1H, J = 11.0, 8.0 Hz, H-6′-2), 3.63–3.49 (m, 2H, H-5′, H-4′), 3.44 (t, 1H, J = 9.2 Hz, H-2′), 3.14 (dd, 1H, J = 11.7, 4.2 Hz, H-3), 2.87 (dd, 1H, J = 13.7, 4.0 Hz, H-18), 1.23 (s, 9H, C(CH3)3), 1.19 (s, 9H, C(CH3)3), 1.11, 0.98, 0.91, 0.89, 0.88, 0.80, 0.71 (each s, each 3H, each CH3); 13C NMR (150 MHz, CDCl3): δ 180.2 (C-28), 178.6, 177.4, 143.7 (C-13), 132.5 (C-12), 122.4, 117.7, 104.8 (C-1′), 90.3, 77.8, 74.1, 72.7, 70.1, 64.8, 63.8, 55.5, 47.6, 46.7, 45.9, 41.7, 41.3, 39.4, 39.0, 39.0, 38.8, 38.4, 36.7, 33.9, 33.1, 32.7, 32.4, 30.7, 28.3, 27.6, 27.2 (two), 27.1 (two), 26.9, 26.5, 25.9, 25.8, 23.6, 23.4, 23.0, 18.2, 16.9, 16.6, 15.3; HRMS (ESI) m/z calcd for C49H79O10 [M+H]+ 827.5540, found 827.5540.

3β-O-[2,4-di-O-(α-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oic acid allyl ester (9)

To a mixture of 7 (2.15 g, 2.60 mmol) and 4 Å molecular sieves in dried CH2Cl2 (50 mL) at 0 °C under argon was added TMSOTf (141 μL, 0.78 mmol), followed by a solution of 2,3,4-tri-O-acetyl-l-rhamnopyranosyl trichloroacetimidate (8) (5.65 g, 13 mmol) in CH2Cl2 (5 mL). After stirring at 0 °C for 2 h and then at r.t. for 1 h, the reaction was quenched with triethylamine. The solid was filtered, and the filtrate was concentrated under vacuum to give a yellow oil. Next the residue was subjected to column chromatography on silica gel (EtOAc-petroleum ether, 1:4) to provide the crude trisaccharide, which was dissolved in THF-CH3OH-H2O (V:V:V = 1:1:1, 90 mL) and then NaOH (1.46 g, 36.40 mmol) was added. After stirred at 50 °C overnight, the solution was neutralized with Dowex 50 × 8 (H+) resin until pH = 7, filtered and concentrated in vacuo. The crude product was purified by silica gel column chromatography (CH2Cl2–MeOH, 6:1) to afford 9 (1.82 g, 70%, for two steps) as a colorless solid; m. p.: 252.3–254.1 °C; 1H NMR (600 MHz, CD3OD): δ 5.99–5.86 (m, 1H, CH CH2), 5.36 (dd, J = 3.8, 1.5 Hz, 2H), 5.33–5.24 (m, 2H, CH CH2-1, H-12), 5.22 (dd, 1H, J = 10.5, 1.3 Hz, CH CH2-2), 5.00 (d, 1H, J = 1.4 Hz, Rha-H-1), 4.85 (d, 1H, J = 1.2 Hz, Rha-H-1), 4.53 (d, 1H, J = 5.6 Hz, OCH 2-2), 4.51 (d, 1H, J = 5.5 Hz, OCH 2-2), 4.43 (d, 1H, J = 7.7 Hz, H-1), 4.02–3.95 (m, 2H), 3.95–3.87 (m, 1H), 3.84 (dd, 1H, J = 2.9, 1.7 Hz, Rha-H-2), 3.80 (dd, 1H, J = 12.7, 1.8 Hz, H-6′-1), 3.75 (dd, 1H, J = 9.6, 3.4 Hz, Rha-H-3), 3.69 (dd, 1H, J = 8.3, 3.6 Hz, Rha-H-3), 3.66–3.60 (m, 2H), 3.56 (t, J = 8.9 Hz, 2H), 3.47–3.35 (m, 4H), 3.16 (dd, 1H, J = 11.6, 4.0 Hz, H-3), 2.89 (dd, 1H, J = 13.6, 3.9 Hz, H-18), 1.27 (d, 3H, J = 6.3 Hz, Rha-H-6), 1.21 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.16, 1.05, 0.94, 0.34, 0.92, 0.86, 0.75 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 176.3 (C-28), 142.2 (C-13), 131.0 (C-12), 121.2, 115.5, 102.8 (C-1′), 100.3 (Rha-C-1), 99.3 (Rha-C-1), 87.6, 77.6, 76.5, 75.4, 73.7, 71.5, 71.2, 70.9, 70.3, 69.7, 69.4, 69.3, 68.0, 67.3, 66.5, 63.3, 59.2, 54.6, 40.1, 40.1, 38.0, 37.5, 37.3, 35.1, 32.1, 31.3, 31.0, 30.8, 28.9, 26.0, 25.8, 24.5, 23.8, 21.8, 21.4, 21.3, 16.6, 15.4, 15.3, 15.2, 15.0, 14.5, 13.4; HRMS (ESI) m/z calcd for C51H83O16 [M+H]+ 951.5527, found 951.5585.

3β-O-[2,4-di-O-(α-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oic acid (10)

Compound 9 (1.82 g, 1.82 mmol) was dissolved in MeOH and CH2Cl2 (V:V = 1:1, 40 mL), PdCl2 (161 mg, 0.91 mmol) was then added. The mixture was vigorously stirred at r.t. for 3 d and then filtered over a celite pad and concentrated under vacuum. The crude product was subjected to column chromatography on silica gel (CH2Cl2–MeOH, 5:1) to provide 10 (1.49 g, 90%) as a colorless solid; m. p.: 201.6–203.4 °C; 1H NMR (600 MHz, CD3OD): δ 5.38 (s, 1H, Rha-H-1), 5.26 (t, 1H, J = 3.0 Hz, H-12), 4.87 (s, 1H, Rha-H-1), 4.44 (d, 1H, J = 7.8 Hz, H-1′), 4.0–43.97 (m, 1H), 3.98 (dd, 1H, J = 3.2, 1.8 Hz, Rha-H-2), 3.94–3.91 (m, 1H), 3.85 (dd, 1H, J = 3.2, 2.0 Hz, Rha-H-2), 3.82 (d, 1H, J = 10.8 Hz, H-6′-1), 3.76 (dd, 1H, J = 9.5, 3.3 Hz, Rha-H-3), 3.68 (dd, 1H, J = 12.0, 4.0 Hz, H-6′-2), 3.64 (dd, 1H, J = 9.3, 3.2 Hz, Rha-H-3), 3.60 (t, 1H, J = 8.7 Hz), 3.55 (t, 1H, J = 8.8 Hz), 3.46 (t, 1H, J = 8.2 Hz), 3.42 (t, 1H, J = 9.3 Hz), 3.38 (t, 1H, J = 9.5 Hz), 3.18 (dd, 1H, J = 11.6, 4.0 Hz, H-3), 2.86 (dd, 1H, J = 13.9, 3.7 Hz), 1.28 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.22 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.17, 1.06, 0.96, 0.95, 0.92, 0.87, 0.83 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 180.8 (C-28), 143.8 (C-13), 122.2 (C-12), 104.0 (C-1′), 101.6 (Rha-C-1), 100.6 (Rha-C-1), 89.0, 79.0, 77.8, 76.7, 75.0, 72.5, 72.3, 71.0, 70.7, 70.6, 69.4, 68.6, 60.6, 55.9, 46.3, 45.9, 41.5, 41.3, 39.2, 38.8, 38.7, 36.5, 33.5, 32.6, 32.4, 32.2, 30.2, 27.4, 27.1, 25.8, 25.0, 23.1, 22.7, 22.6, 17.9, 16.6, 16.5, 16.3, 15.8, 14.6; HRMS (ESI) m/z calcd for C48H79O16 [M+H]+ 911.5228, found 911.5283.

3β-O-[2,4-di-O-(2,3,4-tri-O-acetyl-α-l-rhamnopyranosyl)-β-(3,6-two-O-acetyl)-d-glucopyranosyl]-olean-12-en-28-oic acid (11)

A mixture of 10 (1.49 g, 1.64 mmol), Ac2O (2.48 mL, 26.24 mmol) and DMAP (300 mg, 2.46 mmol) in pyridine (20 mL) was stirred at r.t. overnight. The mixture was concentrated and dissolved in CH2Cl2 (100 mL), then washed with 1 M HCl (3 × 50 mL), saturated NaHCO3 (3 × 50 mL) and brine (2 × 50 mL). The organic layer was dried over anhydrous Na2SO4, filtered, and concentrated in vacuo. Then the residue was purified by silica gel column chromatography (petroleumether-EtOAc, 2:1) to produce 11 (1.89 g, 93%) as a colorless solid; m. p.: 151.7–153.4 °C; 1H NMR (600 MHz, CDCl3): δ 5.29 (t, 1H, J = 3.2 Hz, H-12), 5.25–5.23 (m, 2H), 5.18 (dd, 1H, J = 10.2, 3.2 Hz, Rha-H-3), 5.12 (dd, 1H, J = 3.2, 1.6 Hz, Rha-H-2), 5.06–5.00 (m, 4H), 4.80 (d, 1H, J = 1.8 Hz, Rha-H-1), 4.55 (d, 1H, J = 7.7 Hz, H-1′), 4.48 (dd, 1H, J = 12.2, 1.8 Hz, H-6-1′), 4.28 (dd, 1H, J = 12.3, 4.2 Hz, H-6-2′), 4.24–4.21 (m, 1H), 3.88–3.84 (m, 1H), 3.78 (t, 1H, J = 9.4 Hz), 3.68 (t, 1H, J = 8.4 Hz), 3.63–3.61 (m, 1H), 3.16 (dd, 1H, J = 11.6, 4.0 Hz, H-3), 2.84 (dd, 1H, J = 13.4, 4.2 Hz, H-18), 2.14, 2.14, 2.13, 2.10, 2.05, 2.02, 1.99, 1.98 (each s, each 3H, each CH3CO), 1.17 (d, 6H, J = 6.2 Hz, 2 × Rha-H-6), 1.14, 1.04, 0.94, 0.92, 0.91, 0.81, 0.76 (each s, each 3H, CH3); 13C NMR (150 MHz, CDCl3): δ 182.7 (C-28), 170.6, 170.2, 170.1, 170.0 (three), 169.9, 169.6, 143.4 (C-13), 122.5 (C-12), 103.6 (C-1′), 99.5 (Rha-C-1), 97.1 (Rha-C-1), 89.9, 78.0, 75.6, 75.4, 72.1, 71.1, 70.5, 69.9, 69.7, 68.6, 68.5, 67.9, 66.7, 62.2, 55.9, 47.6, 46.5, 45.8, 41.6, 41.1, 39.3, 39.0, 38.9, 36.7, 33.8, 33.1, 32.6, 32.5, 30.7, 29.7, 27.8, 27.6, 26.0, 25.9, 24.5, 23.6, 23.4, 22.9, 21.4, 20.9 (two), 20.8 (two), 20.7 (two), 18.2, 17.2, 17.1, 16.9, 16.3, 16.2, 15.4; HRMS (ESI) m/z calcd for C64H95O24 [M+H]+ 1247.6149, found 1248.6047.

General procedure for the preparation of 12a-12p

To a solution of 11 (1.0 mmol) in dry CH2Cl2 (20 mL) was added oxalyl chloride (1 mL), and the mixture was allowed to stir at r.t. for 24 h under argon. Next, the dichloromethane was removed via vacuum. The prepared acid chloride was dissolved in dry CH2Cl2 (30 mL) and put under a nitrogen gas atmosphere. The appropriate amine (2.0 mmol) was then added followed by triethylamine (4.0 mmol). The resulting mixture was stirred at r.t. for 5 h and then concentrated in vacuo. The obtained residue was dissolved in MeOH and CH2Cl2 (V:V = 2:1, 15 mL) and then NaOMe was added until pH = 10. After stirred at r.t. for 3 h, the solution was neutralized with Dowex 50 × 8 (H+) resin until pH = 7, filtered and concentrated in vacuo. Then the crude product was purified by silica gel column chromatography (CH2Cl2–MeOH, 6:1) to provide 12a-12p, respectively.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-aniline (12a)

Compound 12a was obtained as a white solid in 92% yield; m. p.: 197.1–198.3 °C; 1H NMR (600 MHz, CD3OD): δ 7.46 (d, 2H, J = 7.7 Hz, Ar–H), 7.29 (t, 2H, J = 7.9 Hz, Ar–H), 7.09 (t, 1H, J = 7.4 Hz, Ar–H), 5.45 (t-like, 1H, H-12), 5.36 (d, 1H, J = 1.0 Hz, Rha-H-1), 4.42 (d, 1H, J = 7.7 Hz, H-1′), 3.91 (dd, 1H, J = 9.5, 6.2 Hz, Rha-H-4), 3.84 (dd, 1H, J = 2.9, 1.8 Hz, Rha-H-2), 3.80 (d, 1H, J = 11.8 Hz, 1.4 Hz, H-6′-1), 3.74 (dd, 1H, J = 9.6, 3.4 Hz, Rha-H-3), 3.66 (dd, 1H, J = 12.1, 4.0 Hz, H-6′-2), 3.63 (dd, 1H, J = 9.5, 3.3 Hz, Rha-H-3), 3.60–3.56 (m, 1H), 3.54 (t, 1H, J = 9.1 Hz), 3.46–3.36 (m, 3H), 3.16 (dd, 1H, J = 11.7, 4.2 Hz, H-3), 2.95 (dd, 1H, J = 13.0, 3.3 Hz, H-18), 1.26 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.20 (d, 6H, J = 6.0 Hz, Rha-H-6, CH3), 1.04, 0.98, 0.94, 0.92, 0.84, 0.74 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 177.3 (C-28), 144.0 (C-13), 138.1 (C-12), 128.2 (two), 124.1, 122.8, 121.1 (two), 104.0 (C-1′), 101.6 (Rha-1-C), 100.6 (Rha-1-C), 88.9, 79.0, 77.8, 76.7, 75.0, 72.5, 72.3, 71.0, 70.7, 70.6, 69.3, 68.6, 60.6, 55.9, 48.4, 46.9, 46.3, 41.6, 41.2, 39.3, 38.8, 38.6, 36.4, 33.7, 32.4, 32.2, 30.2, 27.2, 27.1, 25.8, 25.2, 23.2, 22.7 (two), 17.9, 16.6, 16.5, 16.4, 15.8, 14.6; HRMS (ESI) m/z calcd for C54H84NO15 [M+H]+ 986.5774, found 986.5776.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-2-methoxyaniline (12b)

Compound 12b was obtained as a white solid in 94% yield; m. p.: 201.3–202.7 °C; 1H NMR (600 MHz, CD3OD): δ 8.26 (d, 1H, J = 8.0 Hz, Ar–H), 7.05 (t, 1H, J = 7.2 Hz, Ar–H), 6.99 (d, 1H, J = 7.9 Hz, Ar–H), 6.90 (t, 1H, J = 7.7 Hz, Ar–H), 5.54 (t-like, 1H, H-12), 5.36 (s, 1H, Rha-H-1), 4.42 (d, 1H, J = 7.7 Hz, H-1′), 3.90 (s, 3H, OCH3), 3.80 (d, 1H, J = 11.1 Hz), 3.74 (dd, 1H, J = 9.5, 3.2 Hz, Rha-H-3), 3.66 (dd, 1H, J = 12.0, 3.8 Hz, H-6′-1), 3.63 (dd, 1H, J = 9.4, 3.1 Hz, Rha-H-3), 3.59 (t, 1H, J = 8.7 Hz), 3.54 (t, 1H, J = 9.1 Hz), 3.47–3.36 (m, 3H), 2.74 (d, 1H, J = 12.1 Hz, H-18), 1.27 (d, 3H, J = 6.1 Hz, Rha-H-6), 1.20 (d, 6H, J = 8.4 Hz, Rha-H-6, CH3), 1.04, 0.96, 0.94, 0.89, 0.83, 0.62 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 177.1 (C-28), 148.7, 143.4 (C-13), 127.2 (C-12), 123.9, 120.4, 119.5, 109.9, 104.0 (C-1′), 101.7 (Rha-C-1), 100.6 (Rha-C-1), 88.9, 79.0, 77.8, 76.7, 75.0, 72.5, 72.3, 71.0, 70.7, 70.6, 69.3, 68.6, 60.6, 55.8, 55.0, 47.6, 46.6, 46.5 (three), 42.3, 41.7, 39.3, 38.8, 38.7, 36.4, 33.7, 32.4, 32.2, 32.0, 30.2, 27.1, 27.0, 25.8, 25.0, 23.5, 23.3, 22.6, 17.8, 16.6, 16.5, 15.7, 15.6, 14.7; HRMS (ESI) m/z calcd for C55H86NO16 [M+H]+ 1016.5891, found 1016.5892.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-3-methoxyaniline (12c)

Compound 12c was obtained as a white solid in 91% yield; m. p.: 176.2–178.2 °C; 1H NMR (600 MHz, CD3OD): δ 7.20 (t, 1H, J = 2.0 Hz, Ar–H), 7.18 (t, 1H, J = 8.2 Hz, Ar–H), 7.01–6.98 (m, 1H, Ar–H), 6.66 (dd, 1H, J = 8.3, 2.3 Hz, Ar–H), 5.44 (t-like, 1H, H-12), 5.36 (s, 1H, Rha-H-1), 4.42 (d, 1H, J = 7.7 Hz, H-1′), 3.91 (dd, 1H, J = 9.5, 6.3 Hz, Rha-H-4), 3.77 (s, 3H, OCH3), 3.74 (dd, 1H, J = 9.6, 3.4 Hz, Rha-H-3), 3.66 (dd, 1H, J = 12.0, 3.9 Hz, H-6′-1), 3.63 (dd, 1H, J = 9.5, 3.3 Hz, Rha-H-3), 3.59 (t, 1H, J = 8.7 Hz), 3.54 (t, 1H, J = 9.0 Hz), 3.46–3.36 (m, 3H), 2.94 (dd, 1H, J = 13.2, 3.2 Hz, H-18), 1.26 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.21 (d, 3H, J = 6.3 Hz, Rha-H-6), 1.20, 1.04, 0.98, 0.94, 0.92, 0.84, 0.74 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 177.3 (C-28), 159.9, 144.0 (C-13), 139.3 (C-12), 128.99, 122.8, 113.1, 109.6, 106.8, 104.1 (C-1′), 101.6 (Rha-C-1), 100.6 (Rha-C-1), 88.9, 79.0, 77.8, 76.7, 75.0, 72.5, 72.3, 71.0, 70.7, 70.6, 69.3, 68.6, 60.6, 55.9, 54.3, 48.5, 46.9, 46.4 (two), 46.3, 41.6, 41.2, 39.3, 38.8, 38.6, 36.4, 33.8, 32.4, 32.3, 32.2, 30.2, 27.2, 27.1, 25.8, 25.2, 23.2, 22.7, 17.9, 16.6, 16.5, 16.4, 15.8, 14.7; HRMS (ESI) m/z calcd for C55H86NO16 [M+H]+ 1016.5859, found 1016.5854.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-4-methoxyaniline (12d)

Compound 12d was obtained as a white solid in 91% yield; m. p.: 225.3–227.1 °C; 1H NMR (600 MHz, CD3OD): δ 7.33 (d, 2H, J = 9.0 Hz, Ar–H), 6.86 (d, 2H, J = 9.0 Hz, Ar–H), 5.43 (t-like, 1H, H-12), 5.36 (s, 1H, Rha-H), 4.42 (d, 1H, J = 7.7 Hz, H-1′), 3.91 (dd, 1H, J = 9.4, 6.2 Hz, Rha-H-4), 3.80 (d, 1H, J = 10.6 Hz), 3.76 (s, 3H, OCH3), 3.66 (dd, 1H, J = 12.0, 4.0 Hz), 3.63 (dd, 1H, J = 9.4, 3.0 Hz, Rha-H-3), 3.59 (t, 1H, J = 8.7 Hz), 3.54 (t, 1H, J = 9.1 Hz), 3.46–3.36 (m, 3H), 3.15 (dd, 1H, J = 11.7, 4.2 Hz, H-3), 2.93 (dd, 1H, J = 13.0, 3.3 Hz, H-18), 1.26 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.21 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.20, 1.04, 0.98, 0.93, 0.92, 0.84, 0.76 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 177.2 (C-28), 156.7, 144.0 (C-13), 131.0 (C-12), 123.0, 122.8 (two), 113.4 (two), 104.0 (C-1′), 101.6 (Rha-1-C), 100.6 (Rha-1-C), 88.9, 79.1, 77.8, 76.7, 75.0, 72.5, 72.3, 71.0, 70.7, 70.6, 69.4, 68.6, 60.6, 55.9, 54.5 (two), 53.4, 46.7, 46.3, 41.6, 41.2, 39.3, 38.8, 38.7, 36.5, 33.8, 32.5, 32.4, 32.2, 30.2, 27.2 (two), 25.8, 25.2, 23.2, 22.7, 17.9, 16.6, 16.5 (two), 15.8, 14.7, 7.8; HRMS (ESI) m/z calcd for C55H86NO16 [M+H]+ 1016.5870, found 1016.5870.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-2-chloroaniline (12e)

Compound 12e was obtained as a white solid in 92% yield; m. p.: 239.1–240.7 °C; 1H NMR (600 MHz, CD3OD): δ 8.09 (d, 1H, J = 8.2 Hz, Ar–H), 7.43 (dd, 1H, J = 8.0, 1.2 Hz, Ar–H), 7.36–7.20 (m, 1H, Ar–H), 7.20–7.04 (m, 1H, Ar–H), 5.47 (t-like, 1H, H-12), 5.36 (s, 1H, Rha-H-1), 4.42 (d, 1H, J = 7.7 Hz, H-1′), 3.91 (dd, 1H, J = 9.5, 6.2 Hz, Rha-H-4), 3.84 (dd, 1H, J = 2.9, 1.7 Hz, Rha-H-2), 3.74 (dd, 1H, J = 9.5, 3.3 Hz, Rha-H-3), 3.66 (dd, 1H, J = 12.1, 3.9 Hz, H-6′-1), 3.63 (dd, 1H, J = 9.4, 3.3 Hz, Rha-H-3), 3.58 (t, 1H, J = 8.7 Hz), 3.54 (t, 1H, J = 9.0 Hz), 3.46–3.36 (m, 3H), 3.15 (dd, 1H, J = 11.7, 4.2 Hz, H-3), 2.85 (dd, 1H, J = 12.4, 2.8 Hz, H-18), 1.26 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.20 (d, 3H, J = 6.3 Hz, Rha-H-6), 1.21, 1.04, 0.98, 0.95, 0.90, 0.84, 0.71 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 177.2 (C-28), 143.3 (C-13), 134.6 (C-12), 129.0, 127.2, 125.3, 125.0, 124.0, 123.3, 104.0 (C-1′), 101.6 (Rha-1-C), 100.6 (Rha-1-C), 88.9, 79.0, 77.8, 76.7, 75.0, 72.5, 72.3, 71.0, 70.7, 70.6, 69.4, 68.6, 60.6, 55.8, 53.4, 47.6, 46.5, 46.4, 42.0, 41.6, 39.3, 38.8, 38.6, 36.4, 33.7, 32.6, 32.3, 32.0, 30.2, 27.1, 25.8, 25.2, 23.3, 23.2, 22.6, 17.8, 16.6, 16.5, 16.2, 15.7, 14.7, 7.9; HRMS (ESI) m/z calcd for C54H83ClNO15 [M+H]+ 1020.5378, found 1020.5389.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-benzylamine (12f)

Compound 12f was obtained as a white solid in 90% yield; m. p.: 198.6–200.4 °C; 1H NMR (600 MHz, CD3OD): δ 7.70 (s, 1H, J = 5.8 Hz, NHCH2), 7.30–7.27 (m, 4H, Ar–H), 7.25–7.20 (m, 1H, Ar–H), 5.37 (s, 1H, Rha-H-1), 5.32 (t-like, 1H, H-12), 4.85 (s, 1H, Rha-H-1), 4.42 (d, 1H, J = 7.7 Hz, H-1′), 4.38 (dd, 1H, J = 14.7, 5.9 Hz, NHCH 2-1), 4.27 (dd, 1H, J = 14.7, 5.7 Hz, NHCH 2-2), 4.01–3.95 (m, 2H), 3.94–3.88 (m, 1H), 3.85–3.83 (m, 1H, Rha-H-2), 3.80 (d, 1H, J = 10.6 Hz), 3.75 (dd, 1H, J = 9.6, 3.3 Hz, Rha-H-3), 3.66 (dd, 1H, J = 12.0, 3.9 Hz, H-6′-1), 3.63 (dd, 1H, J = 9.4, 3.2 Hz, Rha-H-3), 3.58 (t, 1H, J = 8.7 Hz), 3.54 (t, 1H, J = 9.1 Hz), 3.47–3.36 (m, 3H), 3.15 (dd, 1H, J = 11.7, 4.1 Hz, H-3), 2.82 (dd, 1H, J = 13.2, 3.5 Hz, H-18), 1.27 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.22 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.16, 1.05, 0.95, 0.91, 0.90, 0.86, 0.59 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 179.9 (C-28), 179.8, 145.0 (C-13), 140.1, 129.2 (two), 128.6, 127.8, 123.8 (C-12), 105.2 (C-1′), 102.8 (Rha-C-1), 101.7 (Rha-C-1), 90.1, 80.2, 78.9, 77.9, 76.2, 73.7, 73.4, 72.2, 71.9, 71.7, 70.5, 69.7, 61.7, 57.0, 47.4, 47.3 (two), 44.2, 44.0, 42.7, 42.4 (two), 40.4, 40.0, 39.8, 37.6, 34.9, 34.1, 33.6, 33.3, 31.4, 28.3, 28.2, 26.9, 26.2, 24.3, 23.8, 19.0, 17.8, 17.7, 17.6, 16.9, 15.8; HRMS (ESI) m/z calcd for C55H86NO15 [M+H]+ 1000.5997, found 1000.5923.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-2-pyridinemethanamine (12g)

Compound 12g was obtained as a white solid in 91% yield; m. p.: 198.0–199.2 °C; 1H NMR (600 MHz, CD3OD): δ 8.49 (d, 1H, J = 4.5 Hz, Ar–H), 7.80–7.76 (m, 1H, Ar–H), 7.36 (d, 1H, J = 7.9 Hz, Ar–H), 7.31–7.28 (m, 1H, Ar–H), 5.39 (t-like, 1H, H-12), 5.36 (s, 1H, Rha-H-1), 4.85 (s, 1H, Rha-H-1), 4.46 (d, 1H, J = 15.8 Hz, NHCH 2-1), 4.42 (d, 1H, J = 7.6 Hz, H-1′), 4.41 (d, 1H, J = 15.9 Hz, NHCH 2-2), 3.91 (dd, 1H, J = 9.5, 6.2 Hz, Rha-H-4), 3.84 (dd, 1H, J = 2.8, 1.8 Hz, Rha-H-2), 3.80 (d, 1H, J = 10.6 Hz), 3.75 (dd, 1H, J = 9.6, 3.4 Hz, Rha-H-3), 3.66 (dd, 1H, J = 12.1, 4.0 Hz, H-6′-1), 3.63 (dd, 1H, J = 9.5, 3.3 Hz, Rha-H-3), 3.60–3.56 (m, 1H), 3.54 (t, 1H, J = 9.1 Hz), 3.46–3.42 (m, 1H), 3.42–3.37 (m, 2H), 3.15 (dd, 1H, J = 11.7, 4.2 Hz, H-3), 2.82 (dd, 1H, J = 13.1, 3.5 Hz, H-18), 1.26 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.21 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.17, 1.04, 0.96, 0.92, 0.86, 0.84, 0.54 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 178.9 (C-28), 157.4, 148.3, 143.6 (C-13), 137.3 (C-12), 123.0, 122.4, 122.0, 104.1 (C-1′), 101.6 (Rha-C-1), 100.6 (Rha-C-1), 88.9, 79.0, 77.8, 76.7, 75.0, 72.5, 72.3, 71.0, 70.7, 70.6, 69.4, 68.6, 60.6, 55.8, 46.3, 46.2, 44.1, 41.5, 39.2, 38.8, 38.6, 36.4, 33.7, 32.7, 32.4, 32.2, 30.2, 29.3 (two), 27.1 (two), 25.8, 25.1, 23.1, 22.8, 22.6, 17.8, 16.6, 16.5, 16.1, 15.8, 14.6, 7.8; HRMS (ESI) m/z calcd for C54H85N2O15 [M+H]+ 1001.5950, found 1001.5924.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-2-furyl-methylamin (12h)

Compound 12h was obtained as a white solid in 91% yield; m. p.: 214.8–215.9 °C; 1H NMR (600 MHz, CD3OD): δ 7.38 (d, 1H, J = 0.9 Hz, Furan-H), 6.32 (dd, 1H, J = 2.9, 1.9 Hz, Furan-H), 6.20 (d, 1H, J = 3.1 Hz, Furan-H), 5.36 (s, 1H, Rha-H-1), 5.32 (t-like, 1H, H-12), 4.42 (d, 1H, J = 7.7 Hz, H-1′), 4.34 (d, 1H, J = 15.4 Hz, NHCH 2-1), 4.28 (d, 1H, J = 15.4 Hz, NHCH 2-2), 3.91 (dd, 1H, J = 9.4, 6.2 Hz, Rha-H-4), 3.80 (d, 1H, J = 10.7 Hz), 3.75 (dd, 1H, J = 9.6, 3.3 Hz, Rha-H-3), 3.66 (dd, 1H, J = 12.1, 4.0 Hz, H-6′-1), 3.63 (dd, 1H, J = 9.5, 3.2 Hz, Rha-H-3), 3.58 (t, 1H, J = 8.7 Hz), 3.54 (t, 1H, J = 9.0 Hz), 3.47–3.36 (m, 3H), 3.15 (dd, 1H, J = 11.7, 4.1 Hz, H-3), 2.78 (dd, 1H, J = 13.1, 3.4 Hz, H-18), 1.26 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.21 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.15, 1.04, 0.94, 0.91, 0.91, 0.85, 0.61 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 178.6 (C-28), 151.8, 143.8, 141.6 (C-13), 122.7 (C-12), 109.9, 106.8, 104.1 (C-1′), 101.5 (Rha-1-C), 100.6 (Rha-1-C), 89.0, 79.0, 77.8, 76.7, 75.0, 72.5, 72.3, 71.0, 70.7, 70.6, 69.3, 68.6, 60.6, 55.9, 46.3, 46.2, 41.5, 41.2, 39.2, 38.8, 38.7, 36.4, 35.9, 33.7, 32.7, 32.5, 32.2, 30.2, 27.1, 27.0, 25.8, 25.1, 23.2, 22.7, 22.6, 17.9, 16.6, 16.5, 16.1, 15.8, 14.7; HRMS (ESI) m/z calcd for C53H84NO16 [M+H]+ 990.5727, found 990.5743.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-2-thiophenemethylamine (12i)

Compound 12i was obtained as a white solid in 94% yield; m. p.: 227.1–228.3 °C; 1H NMR (600 MHz, CD3OD): δ 7.25 (d, 1H, J = 5.1 Hz, Thiophene-H-5), 6.96 (d, 1H, J = 3.4 Hz, Thiophene-H-3), 6.91 (dd, 1H, J = 5.0, 3.5 Hz, Thiophene-H-4), 5.36 (s, 1H, Rha-H-1), 5.32 (t-like, 1H, H-12), 4.85 (d, 1H, J = 1.1 Hz, Rha-H-1), 4.54 (d, 1H, J = 15.1 Hz, NHCH 2-1), 4.44 (d, 1H, J = 14.9 Hz, NHCH 2-2), 4.42 (d, 1H, J = 7.7 Hz, H-1′), 3.91 (dd, 1H, J = 9.5, 6.2 Hz, Rha-H-4), 3.84 (dd, 1H, J = 2.7, 1.9 Hz, Rha-H-2), 3.79 (dd, J = 11.9, 1.4 Hz, 1H, H-6′-1), 3.74 (dd, 1H, J = 9.6, 3.4 Hz, Rha-H-3), 3.66 (dd, 1H, J = 12.1, 4.1 Hz, H-6′-2), 3.63 (dd, 1H, J = 9.5, 3.3 Hz, Rha-H-3), 3.58 (t, 1H, J = 8.7 Hz), 3.54 (t, 1H, J = 9.1 Hz), 3.46–3.37 (m, 3H), 3.15 (dd, 1H, J = 11.8, 4.2 Hz, H-3), 2.78 (dd, 1H, J = 13.1, 3.8 Hz, H-18), 1.26 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.21 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.15, 1.04, 0.94, 0.91, 0.90, 0.85, 0.61 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 177.0 (C-28), 142.2 (C-13), 140.0 (C-12), 124.6, 124.0, 122.9, 121.2, 102.5 (C-1′), 100.1 (Rha-1-C), 99.1, 87.5, 77.6, 76.3, 75.2, 73.5, 71.0, 70.8, 69.5, 69.2, 69.1, 67.9, 67.1, 59.1, 54.4, 46.1, 44.8, 44.6, 40.0, 39.7, 37.7, 37.3, 37.1, 36.1, 34.9, 32.2, 31.2, 30.9, 30.6, 28.7, 27.8, 25.6 (two), 24.3, 23.5, 21.6, 21.2, 21.1, 16.4, 15.1, 15.0, 14.7, 14.3, 13.1; HRMS (ESI) m/z calcd for C53H84NO15S [M+H]+ 1006.5562, found 1006.5507.

N-methyl-N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-benzylamine (12j)

Compound 12j was obtained as a white solid in 91% yield; m. p.: 211.3–212.4 °C; 1H NMR (600 MHz, CD3OD): δ 7.35–7.28 (m, 2H, Ar–H), 7.28–7.19 (m, 3H, Ar–H), 5.37 (s, 1H, Rha-H-1), 5.21 (t-like, 1H, H-12), 4.43 (d, 1H, J = 7.7 Hz, H-1′), 4.01–3.96 (m, 2H), 3.94–3.88 (m, 1H), 3.84 (dd, 1H, J = 3.0, 1.7 Hz, Rha-H-2), 3.80 (d, 1H, J = 10.7 Hz), 3.76 (dd, 1H, J = 9.6, 3.3 Hz, Rha-H-3), 3.66 (dd, 1H, J = 13.9, 3.4 Hz), 3.63 (dd, 2H, J = 9.5, 3.1 Hz), 3.60–3.53 (m, 2H), 3.48–3.36 (m, 3H), 3.16 (dd, 1H, J = 11.7, 3.9 Hz, H-3), 3.06 (s, 3H, NCH 3), 1.27 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.22 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.17, 1.06, 0.97, 0.96, 0.92, 0.87, 0.78 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 178.5 (C-28), 145.7 (C-13), 138.7, 130.3, 129.4 (two), 129.3, 128.3, 128.0, 127.1, 127.0, 122.9 (C-12), 105.2 (C-1′), 102.8 (Rha-C-1), 101.7 (Rha-C-1), 90.1, 80.2, 79.0, 77.9, 76.2, 73.7, 73.4, 72.2, 71.9 (two), 71.7, 70.5, 69.8, 61.7, 57.2, 54.4, 47.5, 42.9, 40.3, 40.0 (two), 39.8, 37.7 (two), 36.9, 34.7, 33.9, 33.3, 31.1 (two), 28.7, 28.3, 27.0, 26.4, 24.3, 19.1, 17.8, 17.7, 17.6, 16.9, 15.9; HRMS (ESI) m/z calcd for C56H88NO15 [M+H]+ 1014.6088, found 1014.6086.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-2-fluorobenzylamine (12k)

Compound 12k was obtained as a white solid in 92% yield; m. p.: 245.1–246.3 °C; 1H NMR (600 MHz, CD3OD): δ 7.36–7.31 (m, 1H, Ar–H), 7.30–7.23 (m, 1H, Ar–H), 7.13–7.08 (m, 1H, Ar–H), 7.08–7.01 (m, 1H, Ar–H), 5.37 (s, 1H, Rha-H-1), 5.32 (t-like, 1H, H-12), 4.85 (s, 1H, Rha-H-1), 4.46–4.32 (m, 3H, H-1′, NHCH 2Ph), 3.91 (dd, 1H, J = 9.3, 6.2 Hz, Rha-H-4), 3.86–3.82 (m, 1H), 3.79 (d, 1H, J = 11.0 Hz), 3.75 (dd, 1H, J = 9.5, 3.2 Hz, Rha-H-3), 3.69–3.60 (m, 2H), 3.60–3.50 (m, 2H), 3.47–3.35 (m, 3H), 3.15 (dd, 1H, J = 11.5, 3.6 Hz, H-3), 2.80 (dd, 1H, J = 13.3, 2.7 Hz, H-18), 1.27 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.21 (d, 3H, J = 6.1 Hz, Rha-H-6), 1.15, 1.04, 0.95, 0.91, 0.90, 0.85, 0.55 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 180.0 (C-28), 163.2, 160.8, 144.9 (C-13), 130.9, 130.3, 129.9, 129.3, 127.1, 126.9, 124.9 (C-12), 116.0, 115.7, 105.2 (C-1′), 102.8 (Rha-C-1), 101.7 (Rha-C-1), 90.1, 80.1, 78.9, 77.9, 76.2, 73.7, 73.4, 72.2, 71.9, 71.7, 70.5, 69.7, 61.7, 57.0, 47.4, 42.6, 42.5, 40.3, 40.0, 39.8, 38.0, 37.6, 34.9, 34.0, 33.6, 33.3, 31.4, 30.5, 28.3, 26.9, 26.2, 24.3, 23.8, 19.0, 17.8, 17.7, 17.4, 16.9, 15.8; HRMS (ESI) m/z calcd for C55H85FNO15 [M+H]+ 1018.5842, found 1018.5842.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-2-chlorobenzylamine (12l)

Compound 12l was obtained as a white solid in 89% yield; m. p.: 205.5–206.3 °C; 1H NMR (600 MHz, CD3OD): δ 7.61 (t-like, 1H, NHCH2), 7.40–7.32 (m, 2H, Ar–H), 7.29–7.21 (m, 2H, Ar–H), 5.36 (s, 1H, Rha-H-1), 5.34 (t-like, 1H, H-12), 4.50–4.36 (m, 3H, H-1′, NHCH 2), 4.03–3.95 (m, 2H), 3.94–3.86 (m, 1H), 3.84 (s, 1H, Rha-H-2), 3.80 (d, 1H, J = 11.4 Hz, H-6′-1), 3.75 (dd, 1H, J = 9.6, 3.2 Hz, Rha-H-3), 3.70–3.60 (m, 2H), 3.60–3.53 (m, 2H), 3.44–3.40 (m, 3H), 3.15 (dd, 1H, J = 11.5, 3.8 Hz, H-3), 2.79 (dd, 1H, J = 13.3, 3.5 Hz, H-18), 1.27 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.22 (d, 3H, J = 6.1 Hz, Rha-H-6), 1.16, 1.04, 0.96, 0.92, 0.90, 0.85, 0.54 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 180.0 (C-28), 144.9 (C-13), 137.0, 134.3, 130.7, 130.2, 129.5, 127.8, 124.0 (C-12), 105.24 (C-1′), 102.9 (Rha-C-1), 101.7 (Rha-C-1), 90.1, 80.2, 79.0, 77.9, 76.2, 73.7, 73.5, 72.2, 71.9, 71.7, 70.5, 69.7, 64.9, 61.8, 57.0, 47.5 (two), 42.7 (two), 42.2, 40.4 (two), 40.0 (two), 37.6 (two), 34.9, 34.0, 33.6, 33.3, 31.4, 28.3, 26.9, 26.2, 24.3, 23.9, 23.8, 19.0, 17.8, 17.7, 17.4, 16.9, 15.8; HRMS (ESI) m/z calcd for C55H85ClNO15 [M+H]+ 1034.5509, found 1034.5516.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-2-bromobenzylamine (12m)

Compound 12m was obtained as a white solid in 90% yield; m. p.: 175.4–176.3 °C; 1H NMR (600 MHz, CD3OD): δ 7.62–7.52 (m, 2H, Ar–H), 7.39–7.27 (m, 3H, Ar–H), 7.17 (t, 1H, J = 7.4 Hz, NHCH2), 5.39–5.30 (m, 2H, Rha-H-1, H-12), 4.48–4.35 (m, 3H, H-1′, NHCH 2), 4.01–3.94 (m, 2H), 3.94–3.86 (m, 1H), 3.86–3.83 (m, 1H, Rha-H-2), 3.80 (d, 1H, J = 11.4 Hz, H-6′-1), 3.75 (dd, 1H, J = 9.5, 3.0 Hz, Rha-H-3), 3.70–3.60 (m, 2H), 3.60–3.50 (m, 2H), 3.48–3.36 (m, 3H), 3.15 (dd, 1H, J = 11.5, 3.6 Hz, H-3), 2.78 (dd, 1H, J = 12.6, 3.0 Hz, H-18), 1.27 (d, 3H, J = 6.1 Hz, Rha-H-6), 1.22 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.16, 1.05, 0.96, 0.92, 0.90, 0.85, 0.55 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 180.0 (C-28), 144.9 (C-13), 138.6, 133.5, 130.8, 129.8, 128.4, 124.2 (C-12), 124.1, 107.4, 105.2 (C-1′), 102.8 (Rha-C-1), 101.7 (Rha-C-1), 90.1, 80.2, 79.0, 77.9, 76.2, 73.7, 73.4, 72.2, 71.9 (two), 71.7, 70.5, 69.7, 61.7, 57.0, 47.5, 47.4, 44.6, 42.7 (two), 40.4, 40.0, 39.8, 37.6, 34.9, 34.0, 33.6, 33.3, 31.4 (two), 28.3, 26.9, 26.2, 24.3, 23.9, 23.8, 19.0, 17.8, 17.7, 17.4, 16.9, 15.8; HRMS (ESI) m/z calcd for C55H85BrNO15 [M+H]+ 1078.5037, found 1078.5027.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-2-methoxybenzylamine (12n)

Compound 12n was obtained as a white solid in 89% yield; m. p.: 174.9–175.8 °C; 1H NMR (600 MHz, CD3OD): δ 7.25 (t, 2H, J = 6.1 Hz, Ar–H), 7.19 (d, 1H, J = 7.0 Hz, Ar–H), 6.95 (d, 1H, J = 8.3 Hz, Ar–H), 6.88 (t-like, 1H, NHCH2), 5.36 (s, 1H, Rha-H-1), 5.31 (t-like, 1H, H-12), 4.42 (d, 1H, J = 7.6 Hz, H-1′), 4.35 (d, 2H, J = 5.3 Hz, NHCH 2), 4.00–3.95 (m, 2H), 3.94–3.90 (m, 1H), 3.88 (s, 3H, OCH3), 3.84 (s, 1H, Rha-H-2), 3.80 (d, 1H, J = 11.6 Hz), 3.75 (dd, 1H, J = 9.6, 3.1 Hz, Rha-H-3), 3.69–3.60 (m, 2H), 3.59–3.53 (m, 2H), 3.47–3.36 (m, 3H), 3.14 (dd, 1H, J = 11.3, 3.7 Hz, H-3), 2.67 (dd, 1H, J = 12.9, 2.5 Hz, H-18), 1.27 (d, 3H, J = 6.4 Hz, Rha-H-6), 1.22 (d, 3H, J = 5.8 Hz, Rha-H-6), 1.04, 0.94, 0.91, 0.87, 0.85, 0.46 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 179.5 (C-28), 158.7, 145.1 (C-13), 130.1, 129.6, 127.1, 123.8 (C-12), 121.3, 111.1, 105.2 (C-1′), 102.8 (Rha-C-1), 101.7 (Rha-C-1), 90.1, 80.1, 78.9, 77.8, 76.2, 73.6, 73.5, 72.1, 71.8, 71.7, 70.5, 69.7, 61.7, 61.3, 57.0, 55.6, 47.6, 47.5, 47.4, 42.9, 42.7, 40.3, 40.1, 40.0, 39.8, 37.5, 34.9, 33.8, 33.3, 31.3, 28.3, 28.1, 26.9, 26.2, 24.3, 24.0, 23.7, 19.0, 17.8, 17.7, 17.2, 16.9, 15.9, 9.2; HRMS (ESI) m/z calcd for C56H88NO16 [M+H]+ 1030.6031, found 1030.6031.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-2-phenylethylamine (12o)

Compound 12o was obtained as a white solid in 93% yield; m. p.: 199.6–200.7 °C; 1H NMR (600 MHz, CD3OD): δ 7.34–7.26 (m, 2H, Ar–H), 7.26–7.18 (m, 3H, Ar–H), 6.94–6.87 (t-like, 1H, NHCH2), 5.36 (s, 1H, Rha-H-1), 5.13 (t-like, 1H, H-12), 4.42 (d, 1H, J = 7.7 Hz, H-1′), 4.02–3.95 (m, 2H), 3.95–3.86 (m, 1H), 3.86–3.82 (m, 1H, Rha-H-2), 3.82–3.77 (m, 1H, H-6-1′), 3.74 (dd, 1H, J = 9.5, 3.1 Hz, Rha-H-3), 3.70–3.60 (m, 2H), 3.60–3.53 (m, 2H), 3.48–3.35 (m, 3H), 3.26–3.18 (m, 1H, NHCH 2CH 2), 3.15 (dd, 1H, J = 11.7, 3.8 Hz, H-3), 2.88–2.69 (m, 3H, NHCH 2CH 2), 2.59 (dd, 1H, J = 12.6, 3.6 Hz, H-18), 1.27 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.21 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.13, 1.04, 0.92, 0.91, 0.90, 0.86, 0.65 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 180.1 (C-28), 144.9 (C-13), 140.3, 129.7 (two), 129.4 (two), 127.3 (C-12), 105.2 (C-1′), 102.8 (Rha-C-1), 101.8 (Rha-C-1), 90.1, 80.2, 79.0, 77.9, 76.2, 73.7, 73.4, 72.2, 71.9, 71.7, 70.5, 69.8, 61.7, 57.0, 47.5, 47.3, 42.6, 42.5, 41.8, 40.4 (two), 40.0, 39.8, 37.6 (two), 36.1, 34.9, 33.7, 33.4, 33.3, 31.3 (two), 28.3, 28.2, 26.9, 26.2, 24.3, 24.0, 23.7, 19.0, 17.8, 17.7, 17.5, 16.9, 15.8; HRMS (ESI) m/z calcd for C56H88NO15 [M+H]+ 1014.6089, found 1014.6089.

N-{3β-O-[2, 4-di-O-(a-l-rhamnopyranosyl)-β-d-glucopyranosyl]-olean-12-en-28-oyl}-3-phenylpropanamine (12p)

Compound 12p was obtained as a white solid in 92% yield; m. p.: 203.4–204.2 °C; 1H NMR (600 MHz, CD3OD): δ 7.28–7.22 (m, 2H, Ar–H), 7.20–7.12 (m, 3H, Ar–H), 5.37 (s, 1H, Rha-H-1), 5.34 (t-like, 1H, H-12), 4.42 (d, 1H, J = 7.6 Hz, H-1′), 4.00–3.95 (m, 2H), 3.95–3.86 (m, 1H), 3.86–3.83 (m, 1H, Rha-H-2), 3.80 (d, 1H, J = 11.6 Hz, H-6′-1), 3.75 (dd, 1H, J = 9.6, 3.2 Hz, Rha-H-3), 3.69–3.60 (m, 2H), 3.56 (dd, 2H, J = 11.2, 9.0 Hz), 3.48–3.34 (m, 3H), 3.27–3.20 (m, 1H), 3.15 (dd, 1H, J = 11.5, 3.8 Hz, H-3), 3.12–3.05 (m, 1H), 2.76 (dd, 1H, J = 12.7, 2.5 Hz, H-18), 2.66–2.59 (m, 2H), 1.27 (d, 3H, J = 6.2 Hz, Rha-H-6), 1.22 (d, 3H, J = 6.1 Hz, Rha-H-6), 1.16, 1.05, 0.95, 0.92, 0.91, 0.86, 0.75 (each s, each 3H, each CH3); 13C NMR (150 MHz, CD3OD): δ 180.0 (C-28), 179.9, 145.1 (C-13), 142.8, 129.2 (two), 129.1 (two), 126.7, 123.7 (C-12), 105.2 (C-1′), 102.8 (Rha-C-1), 101.7 (Rha-C-1), 90.1, 80.2, 78.9, 77.9, 76.2, 73.7, 73.4, 72.2 (two), 71.9, 71.7, 70.5, 69.7, 61.7, 57.0, 47.4, 47.2, 42.7, 42.4, 40.4, 40.3, 40.2, 40.0, 39.8, 37.6, 34.9, 34.2, 34.1, 33.6, 33.3, 31.9, 31.4, 28.3, 26.9, 26.3, 24.3, 23.8, 19.0, 17.8, 17.7 (two), 17.7, 16.9, 15.8; HRMS (ESI) m/z calcd for C57H90NO15 [M+H]+ 1028.6249, found 1028.6249.

Pseudotyped SARS-CoV-2 infection assay

HEK293T cells were seeded in 6 plate wells at 4 × 105/mL and cultured overnight. Pseudovirus (PsV) were produced by co-transfection HEK293T cells with pNL4-3.Luc.R-E− and plasmids encoding either SARS-CoV-2 S, or VSV-G by using Polyjet (SignaGen, USA). After 48 h of transfection, supernatants containing PsV were collected and cleared by 0.45 μM filtrations [23]. 293T-ACE2 cells were seeded (20,000 cells per well) in 96 wells plates. After 24 h, a mixture that was consisted of 50 μL gradient concentrations of compounds and 50 μL PsV supernatant was added to each well and incubated with cells for 48 h. Cells were lysed and luciferase activities were quantified by Luciferase assay system (Promega, USA).

Cytotoxicity assay

The cytotoxicity effect of the compounds on 293T-ACE2 or Vero-E6 cells were measured by Methyl Thiazolyl Tetrazolium (MTT) assay [13,28]. In brief, monolayers cells in 96-well plates were rinsed by PBS and incubated with indicated compounds for 48 h subsequently. Then, cells were treated with 0.5 mg/mL MTT for 4 h at 37 °C and formazan crystals in viable cells were dissolved in 150 μL DMSO. The absorbance of solubilized formazan was measured by Synergy multimode reader (BioTek, American) at 570 nm.

Indirect immunofluorescence assay (IFA)

After the authentic SARS-CoV-2 inhibition assay, we fixed the cells with 4% paraformaldehyde (Bio-Rad) and permeabilized by using Triton X-100 (Sigma, USA). Cells were stained with polyclonal rabbit anti-NP antibody and a secondary peroxidase-labeled goat anti-rabbit IgG H&L (Alexa Fluor® 488) (Abcam). DAPI was used to color nucleus and images were acquired by Axio Observer microscope (Zeiss, Germany) [40] and chloroquine (CQ) was used as a positive control [23].

Authentic SARS-CoV-2 inhibition assay

Authentic SARS-CoV-2 inhibition assay was performed by Wuhan Institute of Virology, Chinese Academy of Sciences, as described previously [13,23]. Briefly, SARS-CoV-2 (Wuhan-HU-1) was incubated with different concentrations of compound 12f for 1 h before placing the mixture on vero-E6 cells for another 1 h incubation (MOI = 0.01). After that, fresh medium with variant concentrations of 12f was added and viral total RNA were extracted by Viral RNA/DNA Extraction Kit (Takara, Japan) after 24 h, and the S gene copies were quantified on ABI 7500 (Takara TB Green® Premix Ex Taq™ II, Japan) by using the primers reported before [23], Foward: GCTCCCTCTCATCAGTTCCA; Reverse: CTCAAGTGTCTGTGGATCACG.

Co-immunoprecipitation (Co-IP) assay and western blotting

Plasmids pcDNA3.1-ACE2-Flag and pcDNA3.1-SARS-S were co-transfected into 293T cells. Total protein was extracted after 48 h and then incubated with protein A Sepharose in conjunction with anti-flag antibody or mouse IgG. The samples were incubated overnight and separated by 10% PAGE subsequently, followed by transfer to nitrocellulose membranes (Roche, Germany). ACE2 and SARS-S were detected by anti-Flag (Sigma, USA) and anti-SARS-CoV-S (Sinol biological Inc., China) with mouse anti-goat-horseradish peroxidase (HRP) (Fude biological Technology Co., LTD., China) as the secondary antibody, respectively [41].

Surface plasmon resonance (SPR) measurement

Compound 12f was fixed on the chip by photo-crosslinking, then recombinant SARS-CoV-2 S-trimer, S1 subunit or S2 subunit protein (DRA 47, DRA 48, DRA 49, Novoprotein Inc. Shanghai) at indicated concentrations were injected sequentially into the chamber in buffer PBST (0.1% Tween 20, pH 7.4), the interaction of S-trimer, S1 subunit or S2 subunit with 12f fixed was detected by PlexArrayTM HT SPRi (Seattle, US). The reaction temperature was controlled at 4 °C, binding time was 600 s, disassociation time was 360 s, flow rate was 0.5 μL/s, the chip was regenerated with Glycine Hydrochloride (pH 2.0). The data of interaction signals was retrieved and analyzed with PlexeraDE software [42].

Cell-cell fusion assays

HEK-293T cells were transfected with pAAV-IRES-GFP-SARS-CoV-2 S or vehicle plasmid pAAV-IRES-GFP to construct effector cells by using Polyjet (SignaGen, USA) [13,23]. Targeted cells (Vero-E6) were seeded in 96 well plates (20,000 cells per well) 6 h prior to cell-cell fusion. Effector cells were firstly treated with different concentrations of compounds for 30 min, then were overlaid on targeted cells. After 24 h, three random fields were imaged by inverted fluorescence microscope (Zeiss, Germany) [13].

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.