This study aimed at further investigating the molecular mechanism of action by which arbidol (Arb) inhibits virus cell entry and membrane fusion, using HCVpp as a model of an enveloped virus.
We showed that Arb displayed a dual binding capacity, on lipid membranes interface on one hand and on the aromatic residue tryptophan of proteins on the other hand. It therefore appears plausible that the observed inhibitory effect of Arb on viral entry and membrane fusion might result from a combined effect of binding of Arb on membranes and on (fusion) proteins.
From a physico-chemical point of view, Arb displayed tropism for membranes or membrane-like environments such as detergent micelles, particularly prominent at low pH 
. By combining several biochemical approaches, we show here that Arb has the propensity to bind to and incorporate into lipid bilayers, with calculated apparent affinities in a similar range as the IC50 value for fusion, i.e. ca. 10 µM. Our NMR studies of Arb interaction with DMPC leads to a model where Arb binds at the membrane interface and establishes contacts mainly with the polar heads of phospholipids ().
Altogether these data suggest that at least part of Arb inhibitory activity could be explained by its membranotropism. This physico-chemical property has been further emphasized in a recent work by Villalain 
, using Fourier-transform infrared spectroscopy. Arb interaction with phospholipids would disturb membrane fluidity crucial to the fusion process, thereby rendering the lipid bilayer less prone to fusion. Such a model is consistent with the behavior of other indole derivatives, that were shown to exhibit a preference for membrane interfaces 
, due to the flat rigid structure of these molecules and to their aromaticity, which allows them to establish cation-π interactions with the positively charged quaternary ammonium lipid headgroups 
. At low pH, the optimal pH for HCV fusion, these interactions would be favored due to the protonation of the amino groups. As was described for other substituted indoles 
, it is possible that protonation of the carbon bearing the ester group of Arb could displace this group out of the indole plane, and place it in a better position to bond with neighboring molecules. This could in turn lead to a better membrane association. Arb might therefore have the propensity to intercalate into lipids of the viral and target membranes while adopting a consistent orientation by filling the gaps between lipid molecules. The interfacial region of the lipid bilayer provides a suitable environment for a wide range of chemical groups, as long as they possess a large enough hydrophobic moiety and a group capable of forming hydrogen bonds with the lipid carbonyl groups. Several compounds with antiviral pharmacological properties belong to this category, in particular adamantanes active against influenza A viruses 
and against some HCV clones but not all 
, the natural triterpene glycyrrhizin efficient in the treatment of chronic viral hepatites 
and the flavonolignan molecules composing silymarin, an herbal extract with potent anti-HCV activities 
In a previous study, we noticed that Arb inhibition of cell entry concerned HCVpp and pseudoparticles bearing the influenza hemagglutinin (HApp), but not pseudoparticles bearing the envelope glycoprotein of a feline oncogenic retrovirus (Rd114pp) 
. These data suggest that Arb might display selectivity for the recognition of key motifs inside envelope proteins. This hypothesis was tested by assessing the influence of Arb on the fluorescence properties of aromatic compounds derived from tryptophan (Trp) and of peptides containing Trp. Trp is a component of proteins with interfacial properties 
, often located at the lipid/water interface and grouped into so-called tryptophan-rich motifs crucial to protein/membrane association 
, and found in the envelope (fusion) proteins of the SARS coronavirus or HIV-1 
. Trp is also enriched at protein/protein binding interfaces of the small envelope protein of the hepatitis B virus 
and of membrane proteins in general 
. We demonstrated here that Arb was able to alter/quench the fluorescence properties of small Trp derivatives in solution (NATA), in detergent micelles and in liposomes (TOE, 
), in a dose-dependent manner. This occurred most likely through stacking of the aromatic rings of both molecules which is often involved in stabilization of inter-cations. Interestingly the apparent affinity of the Arb/Trp derivative interaction was in the order: lipid bilayers>micelles>solution, indicating that Arb binding strength for Trp could increase in membrane environments where both molecules accomodate and get packed. Indeed Arb apparent affinity for TOE in liposomal membranes was in the 10 µM range, a value comparable to the IC50 of fusion. Arb affinity was even greater for membrane peptides containing Trp and tyrosine (Tyr) residues (ca. 4 µM). Due to its indole group, it is conceivable that Arb might display selectivity not only for indole rings (Trp) but more generally for aromatic groups, as the phenol ring of Tyr. A greater number of Arb molecules could therefore interact with aromatic residues in peptide sequences, leading to some cooperativity in the quenching effect and to an overall larger apparent affinity.
Although HCV entry inhibition by Arb was found genotype-dependent, HCV membrane fusion was inhibited by Arb in a genotype-independent manner. HCV entry and fusion are early steps in the life cycle of the virus 
. HCV first interacts through its envelope glycoproteins with a set of coreceptors at the plasma membrane level (recently reviewed in 
) and eventually becomes endocytosed 
. Due to a combined action of acidification in the endosome and particular lipids like cholesterol and sphingomyelin 
, viral fusion occurs over a broad spectrum of pH's ranging from 6.3 to 5.0 
. HCV binding to the hepatocyte membrane followed by endocytosis therefore requires several cellular proteins, and most likely involves several levels of interactions (interactions between viral proteins, between cellular and viral proteins, between viral/cellular proteins and lipids). These features might explain the differential effect exerted by Arb on entry of various HCV genotypes: indeed subtle differences in protein sequences could translate into modified interactions with several partners and/or at several levels. Conversely some common principles of action apply to all fusion reactions, viral fusion and cellular fusion processes alike 
. Indeed all fusion processes involve two partners: lipids and the fusion protein(s). This might account for the similar inhibitory effect of Arb on HCV fusion observed for all genotypes. This is in line with the observations that Arb displayed potent antiviral activity against some antigenic serotypes of influenza viruses, but not against all 
Previously we noticed that Arb inhibition of primary infection of Huh-7.5.1 cells with HCV (clone JFH-1) was efficient only when cells were preincubated with Arb 24 or 48h before infection 
; in addition, inhibition of HCVpp and HApp cell entry was most efficient when Arb was pre-incubated with both viral and cell membranes 
. Here, using our in vitro
fusion assay, we observed that Arb inhibition of HCVpp fusion was maximal when both viral and target membranes were incubated with Arb, before fusion was initiated. This suggests that a certain level of membrane impregnation and/or saturation with Arb must be achieved to efficiently inhibit viral infection. Membranes might therefore act as “concentrators” of arbidol, and high concentrations of the molecule might be locally achieved. This could explain why Arb, exhibiting an apparent (medium to low) affinity for membranes in the µM range, exerts a relevant antiviral activity without noticeable membrane damages. Along these lines, in spite of its marked membranotropism, Arb displays only low toxicity 
. Arb exhibited a comparable micromolar apparent affinity for aromatic residues present in membrane peptides in a membrane environment. Altogether, these observations lead us to propose a mechanistic model of the way Arb would inhibit HCV entry and fusion. Through its membranotropism, Arb is able to freely interact with viral and target membranes, and could locally get highly concentrated. Arb is also able to interact with aromatic residues within viral proteins involved in membrane interactions and membrane destabilization necessary for fusion. Through this dual binding capacity, Arb could then locally impede
the structural rearrangements required for the fusion protein to adopt its fusion conformation. The fact that Arb is active in the µM range suggests that Arb would act by reducing the overall speed of the fusion reaction rather than by blocking a specific protein conformation. This could therefore explain the broad antiviral spectrum of Arb, and the genotype independence of its inhibitory effect on HCV fusion, since HCV envelope proteins contain well-conserved aromatic residues in all genotypes. Mechanistically, the key point is the relative accessibility of these residues to Arb at the membrane interface. A cooperative effect between Arb and several aromatic residues might therefore occur. Also the local environment of these aromatic aa is important, since the presence of residues such as histidines (His) in their vicinity could modify their accessibility with respect to pH. Interestingly enough, in the sequence of both HCV E2 peptides studied here () and shown to be involved in HCV fusion 
, His is contiguous to Trp, and in the 606–625 peptide, His is surrounded by three tyrosines. The concept of His as a critical pH sensor at a key intramolecular domain interface in a viral fusion protein has recently emerged 
. Indeed, the protonation of a sole His in the E protein of the tick-borne encephalitis flavivirus (TBEV) triggers large-scale conformational changes leading to viral fusion. Concerning HCV, Rey and coworkers recently proposed a model of the 3D arrangement of the E2 ectodomain 
. In this model, the fusion loop/peptide would lie within the poorly structured domain II, and the E2 606–625 peptide would be found in the globally unstructured domain III, where a critical His residue is disposed at the interface with domain I. The putative fusion loop contains a phenylalanine and a tyrosine 
. At low pH, the optimal pH for HCV membrane fusion, key histidine(s) could become protonated. This could result in conformational rearrangements and, in the context of Arb fusion inhibition, aromatic residues might consequently become more or less accessible to Arb molecules present in their vicinity. We noted that the apparent affinity of Arb for HCV peptides was weaker at pH 5.0 than at pH 7.4. At low pH, Arb is also protonated, and this protonated form could exhibit a greater preference for the interfacial region of the lipid bilayer than the deprotonated form, as demonstrated for adamantanes 
. Combined with the notion that key aromatic and His residues would also display interfacial (re)localization at low pH, this would in turn explain the higher efficiency of Arb at inhibiting fusion at acidic pH 
In conclusion our data reveal that Arb directly interacts with the lipid membrane-water interface, and is able to bind to aromatic residues present in HCV glycoproteins, in their membrane-associated form. Through a subtle binding interplay between Arb, lipids, viral and cellular proteins, Arb might efficiently block HCV entry and membrane fusion interacting with the main actors of the early steps of viral entry. Most interestingly, Arb inhibition of these processes demonstrated an affinity in the µM range, although the membranotropic properties of Arb suggest that it could become locally more concentrated in membranes. Together, these findings suggest that Arb could increase the strength of viral glycoprotein's interactions with the membrane due to a dual binding mode, involving aromatic residues and phospholipids. The resulting complexation would inhibit the expected viral glycoprotein conformational changes required during the membrane fusion process.
The antiviral mechanism of Arb therefore opens promising perspectives for the development of small membranotropic low affinity molecules, that would become locally concentrated in membranes and would mainly act on the kinetics of the conformational rearrangements of the viral fusion protein.