Design and Synthesis of Oleanolic Acid Trimers to Enhance Inhibition of Influenza Virus Entry

Abstract

Influenza is a major threat to millions of people worldwide. Entry inhibitors are of particular interest for the development of novel therapeutic strategies for influenza. We have previously discovered oleanolic acid (OA) to be a mild influenza hemagglutinin (HA) inhibitor. In this work, inspired by the 3D structure of HA as a homotrimeric receptor, we designed and synthesized 15 OA trimers with different linkers and central region via the copper-catalyzed azide–alkyne cycloaddition reaction. All of the OA trimers were evaluated for their antiviral activities in vitro, and 12c, 12e, 13c, and 13d were observed to exhibit robust potency (IC₅₀ in the submicromolar range) against influenza A/WSN/33 (H1N1) virus that was stronger than that observed with oseltamivir. In addition, these compounds also displayed strong biological activity against A/Hong Kong/4801/2014 and B/Sichuan/531/2018 (BV). The results of hemagglutination inhibition assays and surface plasmon resonance binding assays suggest that these OA trimers may interrupt the interaction between the HA protein of influenza virus and the host cell sialic acid receptor, thus blocking viral entry. These findings highlight the utility of multivalent OA conjugates to enhance the ligand–target interactions in anti-influenza virus drug design and are also helpful for studying antiviral drugs derived from natural products.

Influenza A virus is the major cause of seasonal influenza epidemics and respiratory deaths.¹ It was estimated by the World Health Organization (WHO) that annual epidemics result in approximately 3–5 million cases of severe illness and around 290 000 to 650 000 respiratory deaths worldwide.² Currently, seasonal influenza vaccination remains the main strategy to prevent infection. However, the increase in resistance and the difficulties of production of cross-protective vaccines make the development of novel anti-influenza therapies more pressing.³ In contrast to vaccines, chemical drugs offer the advantages of oral bioavailability, high stability, and relatively low production costs. Oseltamivir, zanamivir, peramivir, and baloxavir are the major current and approved therapeutic options for new influenza outbreaks, while agents such as amantadine and rimantadine are no longer recommended for clinical therapy.^4,5 Unfortunately, the high mutation rate of the influenza virus RNA genome coupled with long-term clinical single drug use has promoted antigen diversity and new subtypes, making the virus resistant to existing antiviral drugs. It is particularly important to develop new antiviral drugs as alternatives to influenza virus treatment to fight drug-resistant strains.⁶

Influenza hemagglutinin (HA) is an attractive target for antiviral therapy because of its essential role in mediating viral entry into the host cell.⁷ Our previous studies revealed that oleanolic acid (OA), a pentacyclic triterpene compound, was able to block the pathogen’s HA from binding to a host cell’s sialic acid (SA)-containing sugars.⁸⁻¹¹ However, interactions between HA and SA involve at least two large interacting interfaces. Consequently, it is a significant challenge for small molecules such as OA to efficiently and completely disrupt the full range of physical HA–SA interactions.

Recent studies have widely explored multivalency as a strategy in protein–protein interactions to bridge binding sites or inhibit pathogen binding.¹² HA is a homotrimeric receptor, and thus, tri- or oligovalent ligands may also have a high binding affinity. Waldmann et al. described a trivalent glycopeptide mimetic that is able to bind to hemagglutinin H5 of avian influenza with a dissociation constant (K_D) of 446 nM.¹³ Lu et al. reported that linking sialylated LAcNAc units to trivalent scaffolds furnished inhibitors with >400-fold enhanced inhibition in various assays.¹⁴ Xiao et al. showed that star-shaped compounds containing several pentacyclic triterpene pharmacophores on cyclodextrin scaffolds exhibited very potent antiviral activity against H1N1 virus (A/WSN/33).⁸

On the basis of these results, we designed and synthesized a series of trivalent OA derivatives, and the anti-influenza virus activity of each of the compounds was evaluated. Upon optimization of the length and hydrophobicity of the linker as well as the central region, we identified several compounds that were able to exhibit dramatically potent activity with IC₅₀ values in the submicromolar range. Mechanistic studies suggested that these conjugates potentially bind to HA, thus blocking the interaction between HA and SA. These inhibitors may pave the way for the development of new combination therapies by synergizing with existing agents.

The copper-catalyzed azide–alkyne cycloaddition (CuAAC) reaction is one of the most prominent click reactions because of its atom economy and experimental simplicity. We previously utilized this reaction for the preparation of a structurally diverse library of triterpene derivatives. Biological evaluation of these compounds, including investigation of anti-hepatitis C virus (anti-HCV) activity and anti-influenza A virus (anti-IAV) activity, revealed that the triazole group within the linker significantly promoted the observed antiviral effects.^8−11,15 Hence, we decided to synthesize trivalent OA analogues via the CuAAC reaction and investigate the structure–activity relationship (SAR) for these analogues against IAV infection.

The trivalent OA analogues were composed of center, linker, and OA moieties. First, three different centers were synthesized, as shown in Scheme 1. As previously reported, starting from 3-bromo-2,2-bis(bromomethyl)propan-1-ol, 1,3,5-tris(bromomethyl)benzene, and phloroglucinol, we introduced azide or alkynyl modules to obtain compounds 1, 2, and 3, respectively, which were used as the centers of trivalent OA analogues.¹⁶⁻¹⁸

Scheme 1.

Reagents and conditions: (a) NaN₃, DMSO, 90 °C, 17 h; (b) NaN₃, DMSO, 90 °C, overnight; (c) 3-bromo-1-propyne, DMF, Na₂CO₃, 80 °C, overnight; (d) TsCl, TEA, DCM, 0 °C to rt, overnight; (e) NaN₃, DMF, 50 °C, overnight; (f) PPh₃, THF/H₂O, 50 °C, overnight; (g) TsCl, TEA, DCM, 0 °C to rt, overnight; (h) NaN₃, DMF, 50 °C, overnight; (i) PPh₃, EtOAc/HCl, rt, overnight.

In order to explore the SAR between the linker and the antiviral activity of the OA trimer, alkyl and poly(ethylene glycol) (PEG) chains of different lengths were chosen as the linkers. As shown in Scheme 1, different alkynols were used as raw materials, and the hydroxyl group was converted into an amine (compounds 6a–e). Similarly, the two hydroxyl groups of PEG were converted into azides, and then via selective reduction, a key intermediate with an azide group at one terminus and an amine group at the other terminus were obtained (compounds 9a–e).^19,20

As outlined in Scheme 2, after the 17-COOH of OA was activated using TBTU, the intermediate reacted with compounds 6a–e and 9a–e to give the OA–linker conjugates with the terminal group as an alkyne or azide module, respectively. To perform the CuAAC reaction, the OA–linker conjugates and the corresponding centers were dissolved in CH₂Cl₂/H₂O, followed by the addition of sodium l-ascorbate and CuSO₄. The reaction proceeded smoothly at room temperature to afford three series of OA trimers (12a-12e, 13a-13e and 14a-14e).¹² The ¹H and ¹³C NMR spectra of the synthesized compounds are shown in the Supporting Information.

Scheme 2.

Reagents and conditions: (a) TBTU, DIEA, THF, rt, 12 h; (b) 6a–e, TEA, DMF, rt, overnight; (c) 1, CuSO₄·5H₂O, sodium l-ascorbate, DCM/H₂O, rt, overnight; (d) 2, CuSO₄·5H₂O, sodium l-ascorbate, DCM/H₂O, rt, overnight; (e) 9a–e, TEA, DMF, rt, overnight; (f) 3, CuSO₄·5H₂O, sodium l-ascorbate, DCM/H₂O, rt, overnight.

Cytopathic effect (CPE) screening was utilized to identify compounds causing a reduction in CPE by the influenza A/WSN/33 H1N1 virus. As shown in Figure 1, we found that compound 12c was able to significantly reduce the viral CPE in MDCK cells at 100 μM. The reduction in CPE was confirmed by direct microscopic observation, which detected far less CPE in the drug-treated sample than in the DMSO control. In addition, we utilized the CellTiter-Glo screening assay, which monitors cell viability, in order to evaluate the anti-IAV activity of the synthesized conjugates and to exclude compounds with significant toxicity toward MDCK cells. Culture medium containing 1% DMSO served as a negative control. We observed no significant cytotoxicity against MDCK cells for any of the synthesized OA trimers at concentrations of 100 μM (Figure 2A). All of the OA trimers tested, with the exception of compounds 14b, 14d, and 14e, exhibited potent anti-H1N1 activity with inhibition rates of >80% at 100 μM, which is superior to that observed with OA (Figure 2B). The results of the dose–response assays, expressed as IC₅₀, CC₅₀, and selectivity index (SI), are summarized in Table 1. These data suggest that the antiviral activities corresponded with the type and length of the linker structure. When the linker was composed of PEG, in general the OA trimers (14a–e) displayed less potent antiviral activities than the compounds with linkers composed of methylene groups (12a–13e).

Figure 1.

Compound 12c significantly reduced the viral CPE in MDCK cells at 100 μM.

Figure 2.

Inhibitory effects of OA trimers against influenza A/WSN/33 (H1N1) virus. (A) Cytotoxic-effect-based screen of OA trimers (100 μM). (B) Cytopathic-effect-based screen of OA trimers (100 μM). MDCK cells were utilized as the host cells to test influenza A/WSN/33 (H1N1) virus infection; DMSO was used as the negative control, and oseltamivir phosphate (OSV-p) was used as the positive control. Error bars indicate standard deviations of triplicate experiments.

Table 1. Inhibitory Activities of Compounds 12a–14a and 14c against A/WSN/33 (H1N1).

compound	IC50 (μM)a	CC50 (μM)b	SIc
12a	4.69 ± 6.03	>100	>21.32
12b	2.87 ± 2.37	>100	>34.84
12c	0.31 ± 0.17	>100	>322.58
12d	25.07 ± 1.05	>100	>3.99
12e	0.57 ± 2.64	>100	>175.44
13a	6.86 ± 0.58	>100	>14.58
13b	31.58 ± 1.94	>100	>3.17
13c	0.38 ± 0.28	>100	>263.16
13d	0.23 ± 0.13	>100	>434.78
13e	NDd	>100	–
14a	16.72 ± 0.75	>100	>5.98
14c	9.84 ± 1.36	>100	>10.16
OA	72.27 ± 3.18	>100	>1.38
OSV-pe	1.82 ± 0.25	>100	>54.95

^aIC₅₀ is the concentration required to inhibit viral-infection-induced CPE by 50%. Values are means of duplicate samples from three independent experiments.

^bCC₅₀ is the concentration required to reduce the viability of normal, noninfected MDCK cells by 50%.

^cSI is the selectivity index, defined as CC₅₀/IC₅₀.

^dNot detectable.

^eOseltamivir phosphate, the positive control.

There are two factors that may explain why 14a–e exhibited lower antiviral activities. The first factor is the optimal distance among OA moieties that can bridge three sialic acid binding sites on HA. The second factor is the molecular mobility, which may have affected the binding affinities of OA trimers with HA. As shown in Scheme 2, compounds 14a–e have additional hydrophilic components [−(CH₂CH₂O)−] in the linker, which could potentially lead to a higher molecular mobility compared with the 12a–13e series. This may cause OA to detach from the binding pockets of HA, thereby decreasing the magnitude of the antiviral activity.

The existence of a larger cave in the middle of the HA trimer translates into better tolerance for the central part of the OA trimer. Therefore, this feature likely will have no significant influence on the antiviral activity, regardless of whether the center of these compounds is a hydroxyethyl group (compounds 12a–e) or a benzene ring (compounds 13a–14e). Among these OA trimers, compounds 12c, 12e, 13c and 13d exhibited robust potent antiviral activity, with IC₅₀ values at the submicromolar level. We conclude that in addition to its length, the hydrophobicity of the linker exerts a significant influence on the potency of the OA trimers.

Having identified that OA trimers display strong anti-influenza virus activity, we then explored the potential underlying molecular mechanism for the activity of these compounds against IAV. We chose compound 13d to perform a time-of-addition experiment. The life cycle of IAV can be divided into three steps: (1) viral entry (0–2 h), (2) viral genome replication and translation (2–8 h), and (3) progeny viral particle release (8–10 h). The expression levels of influenza M2 and NP protein in infected MDCK cells were measured at five time intervals: 0–10, 0–2, 2–5, 5–8, and 8–10 h (Figure 3A). As shown in Figure 3B, compared with the DMSO control, the M2 and NP levels at the interval 0–10 h (covering the whole life cycle) and 0–2 h (covering the entry step) were each reduced over 90%. These data suggest that 13d is mainly effective at the early stage (0–2 h) of the viral lifecycle, presumably during attachment or fusion of the virus with the host cell.

Figure 3.

(A) Design of time-of-addition experiments to identify which step of the influenza virus life cycle the oleanolic acid trimers target, i.e.. virus entry (0–2 h), viral genome replication and translation (2–8 h), or progeny virion release (8–10 h). (B) Detection of the expression levels of influenza M2 and NP protein in infected MDCK cells at five time intervals: 0–10, 0–2, 2–5, 5–8, and 8–10 h. At only two intervals, 0–10 h (covering the whole life cycle) and 0–2 h (covering the entry step), was M2 and NP expression significantly reduced compared with the DMSO control. (C) Anti-HA antibody can inhibit influenza-virus-induced aggregation of chicken erythrocytes. (D) Compound 12c exerted a similar capability to inhibit hemagglutination as anti-HA antibody.

The influenza viral envelope protein HA plays a critical role in the early stage of the IAV life cycle. It can bind to sialic acid receptors on the surface of red blood cells (RBCs) causing hemagglutination. Thus, reagents that are able to inhibit the functions of HA may potentially interfere with the attachment of the virus to the RBC, thereby inhibiting the hemagglutination. Thus, we evaluated the interactions of the synthesized OA trimers with IAV using the hemagglutination inhibition (HI) assay. In this assay, IAV was pretreated with compound 12c, 12e, 13c, or 13d prior to the addition of chicken erythrocytes. Following this, a further incubation was conducted at room temperature for 25 min. As shown in Figures 3D and S1, all of the compounds significantly inhibited hemagglutination at submicromolar concentrations, with a capacity similar to that of anti-HA antibody (shown in Figure 3C as a positive control). These results suggest that OA trimers may have the same target as anti-HA antibody and thus may block the interactions of viruses with target cells.

Surface plasmon resonance (SPR) was also used to determine the interactions between the OA trimer and the HA protein of influenza A virus. The H1N1 HA proteins were fixed on CM5 chips, and then OA and compound 13d were flowed over their surfaces, respectively. We found that both OA and 13d showed strong dose-dependent interactions with HA protein of H1N1. As shown in Figure 4, the binding affinity of 13d and HA protein (K_D = 2.63 μM) was higher than that of OA and HA protein (K_D = 16.7 μM). These results suggest that the OA trimers represented by compound 13d can block virus entry by binding to HA protein.

Figure 4.

SPR binding assay of OA and 13d. (A, B) Representative sensorgrams for binding of OA and 13d to surface-bound HA. (C, D) OA and 13d showed a dose-dependent reaction with HA protein of H1N1 with K_D = 16.7 and 2.63 μM, respectively.

Given that OA trimers displayed potent anti-H1N1 activity, we inquired whether these compounds exert broad-spectrum antiviral activity. Four potent compounds, 12c, 12e, 13c, and 13d, were evaluated against two clinical isolates infected with influenza A and B viruses (Table 2). A/Hong Kong/4801/2014 belongs to the H3N2 subtype, and B/Sichuan/531/2018 (BV) belongs to the influenza B subtype; both showed strong OSV-p resistance. The IC₅₀ values of these compounds and the positive control oseltamivir were determined. As shown in Table 2, the tested OA trimers exhibited remarkable activity against A/Hong Kong/4801/2014 with IC₅₀ values in the range of 1.20–4.58 μM. In addition, compound 13d displayed robust activity against B/Sichuan/531/2018 (BV). However, compared with IAV, the inhibitory effects of these compounds against influenza B virus were considerably less potent.

Table 2. Broad-Spectrum Anti-influenza Activity of Compounds 12c, 12e, 13c, and 13d.

		A/Hong Kong/4801/2014 (H3N2)		B/Sichuan/531/2018 (Victoria)
compound	CC50 (μM)	IC50 (μM)	SI	IC50 (μM)	SI
12c	>100	1.71 ± 0.44	>58.48	37.63 ± 2.03	>2.66
12e	>100	1.20 ± 0.27	>83.33	>100	–
13c	>100	2.67 ± 0.26	>37.45	28.35 ± 1.61	>3.53
13d	>100	4.58 ± 0.13	>21.83	1.87 ± 0.23	>53.48
OSV-p	>100	>100	–	>100	–

In this study, we designed and synthesized a series of OA trimers with different linkers and central regions by applying an efficient CuAAC reaction. All of the target compounds were evaluated for their anti-influenza activities in vitro. Among them, compounds 12c, 12e, 13c, and 13d showed the most potent activities against IAV, with IC₅₀ values of 0.31, 0.57, 0.38, and 0.23 μM, respectively. In addition, these four OA trimers exhibited broad-spectrum activity, even against OSV-p-resistant type A and B viruses. The results of the hemagglutination inhibition assay and SPR binding assay suggest that these OA trimers can target the viral HA protein, thus blocking the attachment of influenza viruses to host cells, confirming the initial design notion. Compared with the sialic acid polyvalent HA inhibitors, OA trimers were observed to be more stable and easily synthesized. In future studies, we may apply nanotechnology and nanomaterials in an attempt to further improve the anti-IAV efficacy that we have observed with the OA derivatives described here.

Glossary

Abbreviations

OA

oleanolic acid

HA

hemagglutinin

SA

sialic acid

HCV

hepatitis C virus

IAV

influenza A virus

CPE

cytopathic effect

M2

matrix protein 2

NP

nucleoprotein

NMR

nuclear magnetic resonance

HRMS

high-resolution mass spectrometry

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.1c00374.

• Experimental details, selected ¹H and ¹³C NMR spectra, the hemagglutination inhibition test of compounds 12e, 12c, and 13d (PDF)

Author Contributions

^⊥ L. Shao and F. Yang contributed equally to this work.

The authors declare no competing financial interest.