Patent Application
20170152290
The present invention is related to medicine and the pharmaceutical industry, and in particular to the design and obtainment of synthetic peptides exhibiting antiviral activity against Dengue virus (DENV). The primary structure of these peptides was designed to facilitate the efficient formation of a beta hairpin structure and to mimic functional patches of domain III of the envelope protein (DIIIE) of DENV. The invention is also related to pharmaceutical compositions containing these synthetic peptides, used for the prevention and/or treatment of DENV infections.
DENV is a member of the Flaviviridae family, composed of enveloped viruses whose genome contains a positive-sense, single-stranded ribonucleic acid (RNA) molecule. There are three different genera in the Flaviviridae family: Flavivirus, Hepacivirus and Pestivirus. Flavivirus encompasses over 70 known viruses of which many cause clinically important diseases, such as Yellow Fever Virus (YFV) and Dengue Virus (DENV). The viruses of the Flavivirus genus that cause human disease are usually arthropod-borne (ticks and mosquitoes), and therefore eradicating these diseases is an extremely difficult task (Monath, T. P., F. X. Heinz. 1996. Flaviviruses, p. 961-1034. In: B. N. Fields, D. M. Knipe, P. M. Howley (ed.), Fields virology, 3rd ed., vol. 1. Lippincott-Raven Publishers, Philadelphia, Pa.).
DENV infections have reached pandemic proportions in tropical areas of the world, and their recent re-emergence has become an increasingly difficult challenge for the public health systems of affected countries. About 100 million DENV infections are estimated to take place annually, and 2.5 billion persons live in areas where DENV is endemic (Gubler, D. J. (1998). Clin. Microbiol. Rev. 11, 480-496; Monath, T. P. (1994) Proc. Natl. Acad. Sci USA 91, 2395-2400). During the 1990-1998 period, an average of 514,139 cases and 15,000 deaths due to DENV hemorrhagic fever (DHF) were reported every year to the World Health Organization (WHO), although the actual incidence of DHF is estimated to be several fold higher. No vaccines against DENV are commercially available, and no specific antiviral treatment against this virus exists.
The term DENV actually refers to a complex composed of four antigenically and genetically related viruses or serotypes, denominated DENV1 to DENV4. DENV is transmitted to humans through the bite of a handful of mosquito species, mainly Aedes aegypti. The clinical manifestations of the resulting infection may vary from an asymptomatic disease or a mild febrile state to the more severe DHF and the potentially fatal Dengue shock syndrome (DSS). The most severe clinical manifestations are associated with secondary infections where the virus belongs to a serotype different from that of the primary infection (Kouri G P et al. (1987) Trans Roy Soc Trop Med Hyg; 72: 821-823; Halstead, S. B (2003). Adv. Virus Res. 60:421-67). This observation has been explained through the theory of antibody-dependent enhancement, which states that viral infectivity may be actually enhanced through the formation of non-neutralizing virus-antibody complexes, which would afford the infecting virus an additional port of entry to the target cells via Fc receptors (Halstead S B. (1988). Science; 239: 476-481).
The first step of the viral replication cycle is the binding of virions to the surface of their host cell. It has been shown that DENV virions bind cell surface glycosaminoglycans, and these have been proposed as molecules mediating an initial interaction of infecting viruses with the target cell. Another molecule DENV has also been shown to bind is DC-SIGN (dendritic cell-specific intercellular adhesion molecule-3-grabbing non-integrin), a dendritic cell-specific type C lectin. However, it is thought that both molecules play a passive role, accumulating viruses in the cell surface or propitiating their dissemination in vivo to secondary sites of infection. A productive virus entry requires receptor-mediated endocytosis via subsequent interactions with additional high-affinity receptors or co-receptors, resulting in the internalization of the infecting virions. Once the latter reach the endocytic compartment, the consequent drop in pH triggers the process of fusion of the viral envelope and endosomal membrane, which is mediated by structural changes in the viral fusion protein. The final outcome of this fusion is the release of viral capsids into the cytoplasm, followed by the release of their genomic RNA. Naked cytoplasmic viral RNA then interacts through its 5′ untranslated region (5′UTR) with ribosomes, and the single open reading frame (ORF) it contains gets translated into a precursor viral polyprotein. In the Flavivirus genus, this polyprotein precursor contains three structural proteins (capsid (C) membrane (prM) and envelope (E)) and five non-structural proteins (NS1-5) which are obtained by co- and post-translational modification by viral and host cell proteases. The viral RNA-dependent RNA polymerase produces single-stranded antisense RNA copies, which then function as templates for the synthesis of positive-sense single-stranded genomic RNA. It has been shown that the viral proteins involved in the process of replication of its RNA are physically associated to membrane structures, probably related to the endoplasmic reticulum (ER).
After replication, DENV genomic RNA associates with the capsid protein and, through the membrane of the rough ER or in virus-induced membrane structures, the resulting capsids bud into the ER lumen, already covered with the lipid membrane envelope containing viral proteins. These immature virions then enter the secretory pathway, and are eventually released to the extracellular space as mature virions. As mentioned above, the DENV genome is an approximately 11 kb long single-stranded positive sense RNA molecule (Hencha, E. A. & Putnak, J. R. (1990). Clin. Microbiol. Rev. 3, 376-396) containing an ORF whose exact length varies between different viral serotypes and even between strains of the same serotype (Yabar V, C. (2003). Rev. Perú. Med. Exp. Salud Publica 20, 51-57). The 5′ proximal quarter of this ORF codes for structural proteins, and the remainder codes for non-structural polypeptides (Lindenbach, B. D. & Rice, C. M. (1999). J. Virol. 73, 4611-4621). Protein E of DENV and every other flavivirus plays a fundamental role in the binding of mature virions to cellular receptors, membrane fusion during viral entry and virion assembly. Hence, it influences a number of fundamental virus characteristics, such as host range, virulence and the induction of protective immunity (Modis, Y., et al. (2005). J. Virol. 79, 1223-1231).
Protein E has a molecular weight ranging from 53 to 54 kDa. It is the most conserved of all DENV structural polypeptides, exhibiting an inter-serotype similarity of 60-70%. It has been shown through X-ray crystallography (Modis, Y., (2003). Proc. Natl. Acad. Sci. U.S.A 100, 6986-6991) as well as electron cryo-microscopy studies (Kuhn, R. J., et al. (2002). Cell 108, 717-725) that in DENV, as in other flaviviruses, protein E forms dimers in the surface of mature virions.
Exposing DENV virions to a slightly acid pH (i.e. lower than 6.5) induces an irreversible conformational shift in protein E that dissociates these dimers and leads to the re-association of the resulting monomers into trimers. This conformational change is necessary for the fusion of viral and endosomal membranes that takes place after the internalization of DENV virions by receptor-mediated endocytosis in mammalian cells (Modis, Y., et al. (2004). Nature 427, 313-319).
Out of the approximately 500 amino acid residues of each protein E monomer, 80% form part of its N-terminal ectodomain and the remainder form a transmembrane domain that anchors E to the lipid envelope (Chambers, T. J., et al. (1990). Annu. Rev. Microbiol. 44, 649-688). This transmembrane domain is linked to the ectodomain through a stem of approximately 53 residues (Modis, Y., et al. (2004). Nature 427, 313-319). The E ectodomain in turn folds into three separate structural domains: a central beta-sheet domain (domain I), an elongated dimerization domain (domain II) and a third domain with an immunoglobulin-like fold, domain III (DIIIE) (Modis, Y., et al. (2003). Proc. Natl. Acad. Sci. U.S.A 100, 6986-6991).
DIIIE is the only protein E domain formed by a continuous, uninterrupted segment of its polypeptide chain. DIIIE is found at the C-terminal end of the protein E ectodomain, between residues 294 and 392. Its structure is very similar to that of the globular domains of immunoglobulin constant regions (Modis, Y., et al. (2005). J. Virol. 79, 1223-1231). The tertiary structure of DIIIE includes a disulphide bond between two Cys residues at positions 302 and 333 that are strictly conserved among all flaviviruses.
DIIIE connects to domain I through an extended, flexible 10-residue linker (Modis, Y., et al. (2004). Nature 427, 313-319) that lets DIIIE adopt different orientations relative to the remaining domains (Rey, F. A., et al. (1995). Nature 375, 291-298). During the dimer-to-trimer transition, DIIIE is the domain exhibiting the most drastic structural shifts, as it rotates approximately 70° and its center of mass moves 36 Å towards domain II. In DENV and other flaviviruses, a large portion of the residues determining virulence, cell tropism and viral host range are located in DIIIE (Rey, F. A., et al. (1995). Nature 375, 291-298). Likewise, many neutralization escape mutations map to this domain (Modis, Y., et al. (2005). J. Virol. 79, 1223-1231).
There is plenty of evidence, obtained both from the analysis of the structure of protein E and DENV virions and from experimental work, supporting the notion that DIIIE is part of the region of protein E that interacts with the putative DENV cell receptor. Some of the structural characteristics of this domain consistent with such a role are the fact that DIIIE is the most protruding domain in the virion surface, and therefore the most accessible for interaction with receptor sites (Mukhopadhyay, S., et al. (2005). Nat. Rev. Microbiol. 3, 13-22). Also, the very fact that it adopts an immunoglobulin-like fold suggests that it may engage the putative DENV cell receptor, as similar domains have long been known to be found in a wide variety of cell adhesion proteins. Yet another structural feature of this domain consistent with a possible involvement in receptor binding is the presence of positively charged hydrophilic surface patches, formed by residues 284-310 and 386-411, which might participate in binding of this domain to negatively charged heparin sulfate molecules (Modis, Y., et al. (2005). J. Virol. 79, 1223-1231).
Studies with DIIIE, obtained through recombinant deoxyribonucleic acid (DNA) technology, showed that they bound directly to the surface of cells of the C6/36 and BHK21 lines, where they were able to block viral infection (Hung, J. J., et al. (2004). J. Virol. 78, 378-388).
Previous studies have demonstrated that DIIIE exhibits specific binding to host cells with an apparent dissociation constant (KD) of 30 μM or less, depending on the flaviviral species from which DIIIE is obtained (Halstead, S. B., et al. (2005). Vaccine 23, 849-856). On the other hand, it is known that the monoclonal antibodies (mAb) most efficiently blocking the binding of DENV to its target cells are those whose epitope is located in DIIIE, providing indirect evidence for the involvement of this domain in the interaction of protein E with viral cell receptors (Thullier, P., et al. (2001). J. Gen. Virol. 82, 1885-1892.).
The entry of DENV into its host cells depends on its previous interaction with specific receptor molecules on the cell surface. However, according to the evidence gathered during the study of DENV-host cell interactions, which exact receptor gets used for viral entry may vary, depending on cell type and even virus serotype. The existing data suggest that DENV entry involves an interaction with a multi-molecular complex, where some molecules perform the role of primary receptors, binding and concentrating the virions on the cell surface for later interaction with the putative endocytic receptor.
Blocking viral entry into the cells is an attractive strategy for developing an antiviral treatment, as it would target the very first stage of the viral lifecycle and would, if successful, block all other downstream events of the viral infection. Analysis by X-ray crystallography and nuclear magnetic resonance spectroscopy has provided structural information at atomic resolution on the three structural proteins of the Flavivirus genus (protein C, protein prM and protein E). These studies have suggested alternative approaches to inhibit the viral replication cycle like the inhibition of structural transitions of the envelope protein. Similarly, some structural studies have provided data concerning specific regions of the envelope protein involved in interaction with receptor molecules and neutralizing antibodies, which constitute potential binding sites as targets for rational design of antiviral drugs against DENV.
It has been previously proposed that a human membrane protein denominated Low Density Lipoprotein Receptor-Related Protein (LRP1) is the actual endocytic receptor used during DENV entry (Huerta V. et al, WO 2007/124698). LRP1 would, therefore, constitute an attractive target for the design of antiviral drugs, as this molecule efficiently mediates the endocytosis of over 30 natural ligands, and has been previously shown to act as a receptor to an unrelated viral species, the minor group of human Rhinoviruses. LRP1 has been shown to bind DIIIE both directly and through bridging molecules that interact simultaneously with LRP1 and the virus. One example of the latter case is human alfa-2-macroglobulin (α2M), which binds DIIIE and is also an LRP1 ligand.
It has previously been shown that beta hairpin peptides based on the structure of DIIIE can be used to inhibit DENV infection both in vitro and in vivo (Huerta V. et al, WO 2007/124698). These peptides, whose exact sequence depends on the viral serotype from which they are derived, encompass part of the FG beta hairpin of DIIIE, and are stapled through a disulphide bridge between cysteine residues appended at the N- and C-termini of the selected hairpin fragment. Out of this series of peptides, the member exhibiting the highest antiviral activity was peptide HDIII3CL which, although derived from DENV3 DIIIE, exhibits inhibitory activity against all four DENV serotypes.
These stapled beta hairpin peptides do, however, have important disadvantages, such as a relatively low potency. For instance, the 50% inhibitory concentration (IC50) of HDIII3CL, determined through infection inhibition assays in Vero cells, is 15 μM for DENV1, 20 μM for DENV2 and DENV3, and 40 μM for DENV4. This relatively poor potency precludes direct use of these molecules for clinical development, as good drug leads must have potencies at least in the submicromolar range, and preferably in the nanomolar range, or better.
Another disadvantage of this peptide series is that the inter-strand part of the loop contains a neutralizing, immunodominant B-cell epitope (Matsui K, et al. (2009) Virology 384(1):16-20). Document Huerta V. et al, WO 2007/124698, for instance, demonstrates that the neutralizing mAb 3H5 binds peptides containing this inter-strand loop. Since part of the anti-DIIIE antibody response is cross-reactive among different serotypes, there exists the possibility that a pre-existing cross-reactive anti-DIIIE antibody response, induced by a previous infection, might bind this peptide and decrease its effective concentration, thereby requiring even higher therapeutic dosages of the peptide than those already required by their low potency.
The facts presented above demonstrate that no high-potency antiviral drugs against DENV, with IC50 preferably in the nanomolar range, are yet available. The present invention addresses exactly this unmet need.
The present invention solves the problem mentioned above by providing new DENV-inhibiting beta hairpin peptides obtained by optimizing the potency of previously identified peptides. The beta hairpin peptides of the present invention are characterized by having one of the amino acid sequences presented in SEQ ID No. 1 to SEQ ID No. 9, or a sequence analogous to these sequences.
A central objective of the present invention is to design peptides inhibiting DENV infection with a relatively high potency, at least in the submicromolar range, and preferably at the nanomolar range or better. High potencies are necessary to decrease as much as possible the therapeutic dose. Therapeutic synthetic peptides currently in clinical use against other pathologies are typically very potent compounds, whose effectiveness (IC50, 50% effective concentration (EC50), etc.) sits at the nanomolar/subnanomolar range and whose specificity is very high, exhibiting therefore very low toxicity. A low potency peptide would require high therapeutic dosages, which bring along a number of important disadvantages. One of them is the fact that high doses increase the possibility that the drug will exhibit side effects due to non-specific interactions. Another, that the probability of having aggregation problems, either during manufacture and formulation or during its delivery in vivo (due e.g. to non-specific interactions with serum proteins) increases significantly. Yet another is the fact that high doses increase the probability of having immunogenicity/antigenicity problems, where the induction of an anti-peptide antibody response neutralizes its therapeutic activity, especially if the treatment is repeated or prolonged in time, thereby requiring even higher doses to compensate for this loss.
In contrast, the use of high potency peptides with low effective doses exhibits a number of economic advantages, stemming both from lower manufacturing costs and lower prices to be paid by the patients. Peptide synthesis technology is usually much more expensive than the synthesis of drugs based on small molecular weight compounds, and cost considerations may ultimately limit the number of patients or persons with access to peptide-based antiviral drugs against dengue, especially if therapeutic doses are large.
One of the fundamental factors determining the potency of a therapeutic peptide is the affinity of its interaction with its target. However, the task of designing active peptides that mimic the functional surface patches involved in high-affinity protein-protein interactions is far from trivial. The fundamental problem here is that functional surface patches in globular proteins are usually topographic (i.e. involving more than one continuous segment of the polypeptide chain) and conformational (i.e. requiring a well-defined spatial arrangement of the participating residues for the interaction to take place). In contrast, peptides derived from short protein segments, typically 10-20 residues long, are flexible in solution and usually do not adopt one single well-defined conformation. Beta hairpin peptides, for instance, despite being derived from a well defined structural motif in a folded protein, are not structurally stable in solution, existing instead as an ensemble of different conformations in equilibrium.
The conformational status of a peptide in solution intended to mimic a particular conformation is usually analyzed as an equilibrium between two states: the folded state (whose tri-dimensional structure matches the native protein structure it is supposed to mimic) and the unfolded or denatured state (the ensemble of all non-native conformations adopted by the peptide in solution). In the case of a beta hairpin peptide mimicking the interaction between a beta hairpin in a native protein and a separate protein target, the more stable the hairpin is, the more its structure will resemble that of the corresponding segment in the native protein, and the higher the affinity of the resulting interaction due to smaller losses in conformational entropy upon formation of the peptide-receptor complex.
In other words: assuming that a peptide in solution adopts a disordered structure, and assuming that the process of binding to its target proceeds through a conformational selection mechanism, then the variation in free energy of the binding process can be expressed as the sum of the variation in free energy associated to the folding of the peptide into the appropriate conformation and the variation in free energy associated to the binding of the already folded peptide to its docking site on the receptor. In that case, the amount of free energy released by the binding process, or in other words, the stability of the interaction, can be increased by modifying the sequence of the peptide, replacing some residues of the original sequence by other residues that either a) enhance the conformational stability of the peptide in solution (especially convenient when the selected residues do not interact directly with the contact surface at the receptor site) or b) optimize the intermolecular interactions between the peptide and its receptor, increasing the drop in free energy that takes places during the docking step.
An analysis of the sequence of peptide HDIII3CL (SEQ ID No. 10, Table 1) reveals several characteristics that can be taken advantage of to improve the stability of its beta hairpin fold. These are: a) the presence of asparagine residues—such as Asn3 and Asn1 6—on beta strands F and G. Asparagine is a poor beta-sheet former, with a very low intrinsic beta sheet propensity and a tendency to appear in loops and backbone structures characterized by positive torsion angles (Swindells M B, et al. (1995). Nat Struct Biol.; 2(7):596-603); b) the presence of a rather large loop—six residues from strand to strand, much larger (by four residues) than the optimum size of two (Branden C. & Tooze J. Introduction to protein structure. New York: Garland Publishing, 1991). The problem here is that the insertion of new residues into protein loops has an entropic cost associated with loop closure, which grows with loop size; and c) the presence of two glycine residues—Gly7 and Gly9—in the inter-strand loop. Glycine is the residue suffering the largest conformational entropy loss during the folding process, as it is not subject to the backbone torsion angle restrictions associated with the other natural amino acids.
The basis of the sequence changes presented in the current invention, which increase the conformational stability of the disclosed beta hairpin peptides, is to replace existing residues at modifiable positions by residues with a higher structural propensity for the formation of beta hairpin-type structures. The structural propensity of a particular amino acid to occupy a specific position in a beta hairpin can be estimated from the frequency of appearance of said amino acid at that position among experimentally determined beta hairpin structures from protein structure databases. Mathematically, it is calculated as the logarithm of the quotient between its observed and expected frequencies of appearance, deriving the latter from the relative abundance of the relevant residue in the database.
The positions defined as modifiable in the present invention are:
a) Positions in the first beta strand of the beta hairpin peptide that correspond to residues of the internal face of strand F in the structure of DIIIE, that is, residues not exposed to the solvent in the native structure. These residues occupy positions (HB positions) where hydrogen bonds between donor and acceptor atoms in the backbones of the first and second beta strand (corresponding to strand G in the structure of DIIIE) of the peptide are established;
b) The positions corresponding to the inter-strand loop;
c) HB positions in the second beta strand of the peptide (corresponding to strand G of DIIIE), excepting that of residue Trp391 of DIIIE (DENV2 ordinates), which will remain invariable. At the rest of these “c” positions, residues with hydrophobic character are allowed and preferred.
d) The non-HB (NHB) positions of strand F closest to the loop;
e) Possible modifications for the first HB position of the second strand of the peptide (corresponding to strand G in DIIIE) include, in addition to residues with high structural propensity, its replacement by a Lys residue that would mimic the role of a lysine (Lys385, DENV2 ordinates) that is topographically conserved among DIIIE of all four serotypes and may potentially be engaged during binding to the LRP1/α2M* (activated α2M) complex.
The modifiable residues of the present invention defined in point a) are not solvent-accessible in the structure of DIIIE, and therefore are not part of the contact interface between the domain and its receptor or any other relevant receptors during the viral lifecycle. They can therefore be replaced by residues increasing the folding stability of the peptide.
The residues of the inter-strand loop defined in point (b) are considered modifiable because the sequence of this region varies between DENV serotypes, indicating that strict conservation of these residues is not necessary to preserve the interaction of DIIIE with the putative DENV receptor. This line of reasoning is supported by the fact that peptide HDIII3CL inhibits the infection of all four DENV serotypes, and that in general, blocking the α2M*/LPR1 receptor has a serotype-independent antiviral effect. The inter-strand loop variant preferred by this invention consists of the two central residues of a beta turn where positions 1 and 4 of the turn would correspond to the first connecting residue of the adjacent beta strands, as such a variant would increase the conformational stability of the peptide. In general, natural beta hairpins tend to have short connecting loops (Branden C. & Tooze J. New York: Garland Publishing, 1991), and a reduction in loop size concomitantly reduces the conformational entropy of the denatured state, which increases with the size of the polypeptide chain.
One of the two central residues of the two-residue loop (preferred size for the inter-strand loop in the present invention) can take on a functional role by mimicking Lys385 of DIIIE from DENV3. In some of the peptides disclosed in the present invention, this functional role is performed by a lysine residue at position 2 of the loop in the case of type IIP beta turns (this Lys residue would thus occupy the central position of the turn) whose topology were identical to that of the FG hairpin of DIIIE. In another of the peptides disclosed by the present invention, this functional role is fulfilled by a D-Lys residue (a D stereoisomer of L-lysine) placed at position 1 of the loop (position 2 in type IIP beta turns), although in this case, the topology of the peptide is reversed relative to that of the FG beta hairpin in DIIIE.
The modifiable residues defined in point (c) are oriented to the same face of the hairpin as those of point a), to which they are therefore adjacent, and with which they form hydrogen bonds involving backbone atoms. The corresponding residues in the structure of DIIIE are partially solvent-exposed, as strand G forms the edge of the beta sheet. The position in the beta hairpin peptide corresponding to that of Trp391 (DENV2 ordinates) in DIIIE is defined as non-modifiable in the present invention, as this residue is strictly conserved across all DENV serotypes and is, very likely, functional. Indeed, Example 2 (Table 2) demonstrates that this residue is essential for the antiviral activity of peptide HDIII3CL (and that of peptide HDIII3CL2 as well, see Table 1), and is lost upon its replacement by an alanine residue. A lysine substitution is also allowed at the first position of the second strand (modifiable residue referred to in point e)), as such a substitution would constitute a structural/functional mimic of residue Lys385 of DIIIE of DENV3. Lys385 is a cationic residue involved in the interaction of DIIIE with α2M*, one of the constituent proteins of the putative α2M*/LRP1 endocytic receptor complex.
Although the modifiable positions referred to in point d) correspond to residues in the solvent-exposed surface of DIIIE, these are still defined as modifiable because they are not strictly conserved across DENV serotypes (
Once the set of modifiable positions in peptide HDIII3CL was defined, its sequence was optimized by selecting, for each modifiable position, the residue with the highest structural propensity index. This parameter, as mentioned above, is derived from the observed frequency of appearance of each residue in 8-residue fragments adopting a beta hairpin structure with a central 2-residue loop (that is, residues 4 and 5 of the fragment would correspond to the central residues of a beta turn), taken from databases of experimentally determined protein structures. The database used in this instance was a non-redundant set of tridimensional protein structures with a resolution higher than 2.5 Å (WHAT IF database; Vriend, G. (1990), J. Mol. Graph. 8, 52-6). The structural propensity index was defined as the logarithm of the quotient between the number of occurrences of a particular amino acid at a particular position in the hairpin and the expected number of occurrences based on the relative abundance of the residue in the database, and was calculated separately for beta hairpins containing type IP and IIP beta turns. The final selection required building models of the tridimensional structures of the proposed beta hairpin peptides, taking into account the most common side chain rotamers and inter-strand contacts.
The present invention also discloses the introduction of non-natural amino acids into positions of the beta hairpin that are structurally compatible with their chemical structure and stereochemistry. Such is the case of D-stereoisomers of natural amino acids, introduced into positions adopting positive backbone torsion angles. D-Pro, for instance, is favorable as the second residue of type IIP beta turns (equivalent to the first residue of a 2-residue inter-strand loop in the present invention). Another example is D-Lys, also introduced into the position corresponding to the second residue of a type IIP beta turn, but in peptides with a reversed topology. In these cases, L-Lys is not structurally favored for this position (its structural propensity index is not high, reflecting the fact that Lys is not good at adopting torsion angles located at the right bottom quadrant of the Ramachandran plot), which is unfortunate in light of the fact that a Lys residue in this position would potentially mimic Lys285 of DIIIE from DENV3. Introducing a D-Lys residue instead solves this conundrum, as it can easily adopt the torsion angles required for position 2 of type IIP beta turns.
The definition of modifiable residues, as described in the paragraphs above, corresponds to beta hairpin turns whose topology is identical (native topology) to that of beta hairpin FG in the structure of DIIIE. However, the present invention also discloses peptides with the reverse topology. In these peptides, the first and second beta strand of the sequence correspond, functional and structurally, to strands G and F of DIIIE, respectively, and the residues occupying NHB positions correspond to HB positions in native topology peptides (and vice versa).
The peptides designed here that maintain the native topology of the FG beta hairpin of DIIIE include a disulphide bridge at an NHB position, which increases the conformational stability of beta hairpin structures. The positions occupied by these Cys residues are: a) the first residue of strand F and b) the last residue of strand G (
Several of the peptides disclosed in the present invention exhibit longer beta strands than those of peptide HDIII3CL. These longer strands increase the conformational stability of the peptide, as they introduce two additional hydrogen bonds (one HB residue per strand).
In addition to stabilizing their conformation, the affinity of the interaction of the disclosed peptides with their target (and hence, their antiviral activity) can also be increased by directly optimizing said interaction. This can be done by modifying interface residues whose contribution to the energetics of the interaction is either negligible or actually negative.
In general, it has been found that the energetics of protein-protein binding is dominated by the interactions of a rather small set of residues. Hence, most interface residues play a relatively limited role in the binding process, if at all. Although no structural data are available about the interaction between the peptides disclosed in this invention and their target, or about the interaction between DIIIE and its putative receptor, it can be assumed, as a first approximation, that every solvent-exposed residue of the FG hairpin of DIIIE is engaged and/or plays a functional role in the peptide-receptor interface.
Examples 2 (Table 2) and 6 of the present invention demonstrate that replacing Lys14 (corresponding to Lys388 in DENV3 DIIIE) or Trp17 (corresponding to Trp391 in DIIIE) of peptide HDIII3CL2 by an alanine residue abolishes in either case the antiviral activity of this peptide in Vero cells and affects its binding to the LRP1 receptor. Both residues are hence essential for the biological activity of HDIII3CL2 and are very likely located at the interface, making essential contributions to the interaction. Therefore, none of these residues are substituted in the present invention. A double substitution of residues Cys1 and Cys18 by alanine also abolishes the antiviral activity and binding to LRP1 of HDIIICL2, but the effect in this case is thought to be caused by the loss of conformational stability in solution of the beta hairpin rather than the loss of favorable intermolecular interactions.
The present invention also discloses that extending the potentially functional surface of beta hairpin peptides constitutes an additional means for increasing their potency. In this particular case, four residues are added to the structure of the hairpin, of which two correspond to the exposed face of the FG hairpin in DIIIE and had not been previously included in the HDIII3CL series of analogue peptides. Specifically, two residues are added to each beta strand, where NHB positions are occupied by the amino acids corresponding to residues 375 (F strand) and 392 (G strand) of DENV3 DIIIE. Although in DIIIE the identity of position 375 is not strictly conserved across all four serotypes, since either Asp or Glu are observed in different isolates, this is a conservative substitution. Therefore, both amino acids are eligible for the equivalent position of the beta hairpin peptides disclosed in the present invention, although Glu is the preferred solution due to the lower propensity of Asp to be part of extended conformations such as beta sheets. In the case of the peptide position equivalent to residue 392 of DIIIE, the preferred amino acids are Phe or Tyr, which are the residues most frequently observed at this position among the four DENV serotypes. The modifiable residues of these extended portions of the hairpin then correspond to the HB positions, which would be used to optimize conformational stability by including preferentially hydrophobic amino acids with high beta sheet forming propensities.
Excepting the essential residues disclosed in Example 2 (those corresponding to Lys388, Trp391 and Glu/Asp392 in DIIIE), all remaining residues of the external face (NHB positions) are considered modifiable in the present invention, as their corresponding amino acids in DIIIE exhibit more variability across DENV serotypes.
Concerning the residue type selectable for the modifiable positions of the beta hairpin peptides of the present invention, the preferred solutions are those residue types which appear at the corresponding positions of homologous sequences of the four serotypes of DENV. Such is the case of the residue corresponding to position 377 in DIIIE, where the preferred solution is Tyr (appearing in DIIIE from DENV1 and DENV4) rather than Asn (appearing in DIIIE from DENV3). Although both Tyr and Asn are polar and hydrogen bond donor/acceptors, Tyr exposes a larger nonpolar surface and appears more frequently at protein-protein interaction interfaces, has a higher propensity for adopting extended structures and forming part of beta strands, and may contribute a more favorable (pi-cation) interaction with the residue corresponding to Lys388 from DIIIE, thus contributing to the conformational stability of the beta hairpin.
Non-essential but potentially functional positions can usually be identified through combinatorial methods, such as the use of phage peptide libraries where essential/structural positions have been fixed and the remaining positions have been randomized. In this case, the library is screened for optimal sequences through the use of binding assays that select those phage expressing high-binding peptides to a specific ligand, such as LRP1 and/or α2M*. Optimal sequences can also be obtained through the use of rational design methodologies.
For the purposes of the present invention, the disclosed beta hairpin peptides may be synthesized chemically or, if their sequence contains only natural amino acids, through recombinant DNA technology, either alone or as fusion proteins. The use of fusion proteins, should recombinant DNA technology be applicable, may increase expression levels and the stability of the recombinant peptide against host proteases. The peptide may be bound to its fusion partner through a protease recognition site, which can then be used to release the peptide by treatment with the cognate protease.
Example 1 of this invention discloses the design of nine beta hairpin peptides (SEQ ID No. 1-SEQ ID No. 9). Basically, their structure consists of four segments: two beta-strand segments (structurally analogous to the FG hairping of DIIIE), separated by a beta turn and followed by a C-terminal cationic extension containing three lysine residues. These peptides were designed following the structural/functional criteria discussed above for optimizing their antiviral potency and their binding to cellular receptors.
For the purposes of the present invention, a peptide sequence is considered to be analogous to that of the herein disclosed beta hairpin peptides (peptides PHB1-9 of table1, SEQ ID1-9) if the sequence identity of said peptide to at least one of the sequences of the PHB1-9 beta hairpin peptides is equal to or higher than 70%, and preferably 80%. The sequences of said analogous peptides differ from one another in one or several positions, selected from: 1) modifiable positions a)-e) described above, where a particular residue or residues in PHB1-9 is/are substituted by residues which also exhibit a high structural propensity for occupying that position in beta hairpins; 2) the potentially functional positions described above, whereby the relevant position(s) in PHB1-9 is/are occupied instead by a residue or residues of the FG hairpin of DIIIE from a particular DENV serotype, wherein said residues(s) occupy an analogous position in the beta hairpin of PHB1-9 peptides; 3) positions corresponding to residues of the C-terminal lysine tag; in this case, the tag may comprise two or three lysine residues, preferably three; and 4) positions corresponding to the cysteines forming the disulfide bonds of peptides PHB1-4 and PHB7-9, wherein one Cys residue may be replaced by Asp/Glu as long as the opposing Cys residue is replaced by Lys, such that the peptide is stapled through the formation of an amide bond between the side chains of said residues, that is Asp/Glu on one side and Lys on the other.
The activity of the PHB1-9 peptides was evaluated in a plaque inhibition assay on Vero cells, whose results are shown in Example 2. All of them exhibited antiviral activity in this experimental model, six of them at higher levels than peptide HDIII3CL. Peptide PHB4 in particular exhibited a very potent antiviral effect at the nanomolar range against all four DENV serotypes, with very good selectivity indexes (1290 to 6450, depending on the specific serotype). Hence, this peptide constitutes an excellent lead molecule for the development of antiviral drugs against DENV, with a better potency than previously reported antiviral drug candidates.
The present invention, therefore, also discloses a pharmaceutical composition comprised of one or more beta hairpin peptides with an amino acid sequence selected from those presented in SEQ ID No. 1 to SEQ ID No. 9, or an analogous sequence thereof, and at least one pharmaceutically acceptable excipient.
In one embodiment of the present invention, the peptides comprising said pharmaceutical composition are forming supramolecular aggregates. Since peptide PHB4 (and, in general, the beta hairpin peptides disclosed in the presence invention) is a molecule that is designed to adopt the structure of an amphiphilic beta strand, it can form supramolecular aggregates depending on peptide concentration and exact solvent composition.
Example 5 shows how the peptide self-assembles into nanostructures, whose size and form depend on experimental conditions such as peptide concentration, temperature, solvent, solution age, presence of different additives, etc. The antiviral activity of this peptide may change, depending on these conditions.
In the context of the present invention, the term ‘supramolecular aggregates’ is intended to denominate aggregates formed by several peptide molecules, preferably more than 10.
Example 8 shows how the addition of human serum albumin (HSA) at specific time points after peptide PHB4 is dissolved into saline produces formulations with potencies at the low nanomolar/subnanomolar range. This result indicates that it is possible to formulate the peptide in the presence of HSA, and that the presence of this protein leads to the formation of peptide:HSA complexes exhibiting very high antiviral activity. This is an important finding from a pharmacokinetic point of view if the peptide is to be developed into an antiviral compound, as not only is the serum half-life of HSA very long, but HSA may protect the peptide against serum proteases and unwanted interactions with other serum constituents. Therefore, in one embodiment of the present invention, the pharmaceutical composition comprised of the disclosed beta hairpin peptides also contains HSA.
Regardless of the presence of HSA, the fact that peptide PHB4 self-assembles into nanoparticles is a desirable characteristic from pharmacokinetic and pharmacodynamic points of view. Owing to their size, nanoparticles tend to exhibit longer serum half-lives than monomeric peptides, which are usually cleared and metabolized in a very short time span. In addition, monomeric peptides are very susceptible to proteolytic attack by host proteases, against which their assemblage into nanoparticles tends to provide some protection. Also, nanoparticles offer a structural context on which there would be multiple binding sites in a single entity, which may lead to avidity effects due to multivalent binding and thus, a higher apparent affinity for the peptide-receptor interaction or, in other words, better antiviral potency.
Examples 3 and 6 demonstrate that the beta hairpin peptides of this invention can bind the α2M* and LRP1 proteins, which are components of the putative endocytic receptor for DENV and would, therefore, constitute a target for the development of antiviral compounds, as reported in Huerta H. et al, WO 2007/124698.
An essential part of the present invention is the analysis of the biological activity of the disclosed beta hairpin peptides at the nanomolar range. As shown in Example 3, peptides PHB2, PHB4, PHB5, PHB8 and PHB9 inhibit the binding of a biotinylated variant of recombinant DENV1 DIIIE (Jamaica strain; DIIIE1Jbiot) to protein α2M* at the nanomolar/submicromolar range, and said inhibitory capacity is clearly higher than that exhibited by non-biotinylated DIIIE (DIIIE1J), since their inhibition percentages are 1.5- to 3-fold higher than that of DIIIE1J. In contrast, the inhibitory capacity exhibited by HDIII3CL at this concentration range is very low, compared to DIIIE1J. Example 6 demonstrates that the beta-hairpin peptides of the present invention bind receptor LRP1 with high avidity, and that peptide PHB4 exhibits the highest avidity in this regard. This result is coherent with the fact that PHB4 was also the most potent peptide on antiviral activity assays.
Example 8 demonstrates that the antiviral effect of beta hairpin peptides correlates with their ability to bind receptor LRP1 and protein α2M*, solidly supporting the notion that these peptides exert their antiviral effect by inhibiting the entry of DENV to its target cells. This result also underscores the rationality of the strategy disclosed in this invention for optimizing the antiviral potency of the beta hairpin peptides.
The capacity of the beta hairpin peptides disclosed in the present invention for binding protein α2M* and receptor LRP1 can be harnessed to develop therapeutic agents for the control of diseases or clinical conditions mediated by these proteins. Their potent inhibition of DENV infection, as disclosed in the present invention, constitutes one example of that possibility.
Blocking the binding of DENV (in other words, inhibiting the binding of protein E) to the putative endocytic receptor α2M*/LRP1 is a better choice for developing antiviral drugs against this virus than the other receptors known in the state of the art. The receptors described by other authors are adhesion receptors, that is, they mediate the binding of the virus to the cytoplasmic membrane, but do not internalize the virus by endocytosis. Since the stage inhibited by the peptides disclosed in the present invention is located downstream the initial adhesion stage, the peptides would be effective independently of what specific receptor or receptors the virus uses for adhering to the target cell.
Taking into account these findings, the present invention also comprises the use of the beta hairpin peptides described thereof for the manufacture of a drug. In one embodiment, the drug is used to inhibit or attenuate an infection by DENV. In another embodiment, the drug manufactured with the beta hairpin peptides disclosed in the present invention is employed for the treatment of clinical conditions mediated by proteins α2M* or LRP1.
The present invention also provides a method for inhibiting or attenuating an infection by DENV in an affected patient, characterized by the administration to said patient of one or more of the beta hairpin peptides disclosed in the present invention, or a pharmaceutical composition comprising at least one of these peptides.
A total of nine beta hairpin peptides, denominated PHB1-9 in Table 1, whose sequences are shown in SEQ ID No. 1-9, were designed starting from peptide HDIII3CL (SEQ ID No. 10), following structural/functional criteria aimed at increasing the potency of their antiviral activity and the strength of their binding to cell receptors when compared with HDIII3CL. The structure of these peptides basically consists of four segments: two beta strand segments (structurally analogous to those of the FG beta hairpin of DIIIE), separated by a beta turn and followed by a cationic C-terminal extension composed of three lysine residues. Two different topologies were used: the native topology, used in peptides PHB1-4 and PHB7-9, where the polypeptide backbone runs in the same direction as in the FG beta hairpin of DIIIE, and the reverse topology, used in peptides PHB5 and PHB6. In native topology peptides, the first beta strand segment (Beta1 in Table 1) corresponds structurally to the F beta strand, and the second beta strand segment (Beta2) corresponds to the G beta strand (
Residues Cys1, Asn/Tyr3 and Thr5 from peptides PHB1, PHB2 and PHB7 correspond to exposed NHB positions in the FG beta hairpin of DIIIE. An Asn residue, equivalent to residue Asn377 in DENV3 DIIIE, was chosen for position 3 (Asn3) of peptides PHB1 and PHB7, and a Tyr residue, equivalent to Tyr377 of DIIIE from DENV1, DENV2 and DENV4, was chosen for position 3 (Tyr3) of peptide PHB2. Tyr3 is preferred over Asn3, as it is more conserved across DENV serotypes, has a higher propensity for beta sheet formation, and is arranged diagonally to Lys10, facilitating a favorable pi-cation interaction between their corresponding side chains (
Residues Val2, Trp/Val4, Glu/Arg/Ile6, Lys/Trp9, Val/Tyr11 and Trp13 form part of the opposite face of the hairpin, occupying HB positions. Positions 2, 4 and 6 are adjacent to positions 13, 11 and 9 respectively, and form backbone hydrogen bonds. Position 13 is occupied only by Trp, as explained earlier, and the remaining positions are chosen according to the value of the preference parameter for each residue (
Peptides PHB1 and PHB2 have been designed to that a type IIP beta turn forms between residues 6 and 9. Position 7 in these peptides is occupied by a d-Pro residue (a D-stereoisomer of proline), which is a favorable amino acid for the second position of type IIP beta turns (Pantoja-Uceda D, et al. (2006) Methods Mol Biol.; 340:27-51). Asp was chosen for position 8 in peptide PHB1 due to the high scores of the preference parameter for this amino acid at this position. In contrast, a Lys residue was selected instead for position 8 in peptide PHB2, with the intention of mimicking Lys386 from DIIIE. For peptide PHB7, a type IP beta turn was introduced between positions 6 and 9. The choice of residue for positions 7 and 8 (Asn7 and Gly8) was driven by the value of the preference parameter (
The sequence of the beta hairpin of peptides PHB3 and PHB4 comprises the sequence of the corresponding hairpin in peptides PHB1 and PHB2 respectively (Table 1). In this case, positions 4-15 in the former are equivalent to positions 2-13 in the latter, and so the residues chosen for positions 4-15 of PHB3 and PHB4 used the same criteria described above for peptides PHB1 and PHB2.
Similarly, peptides PHB8 and PHB9 comprise the sequence of peptide PHB7, and the residues for positions 4-15 were chosen following the same criteria as for positions 2-13 in PHB7. Residues Glu3 and Phe16 in PHB3, PHB4, PHB8 and PHB9 were chosen with the purpose of mimicking Glu375 (conserved in DENV1 and DENV3, Asp in DENV2 and DENV4) and Phe392 (conserved in DENV1, DENV2 y DENV4, Tyr in DENV3) from DIIIE (
The topology of peptides PHB5 and PHB6 is reversed to that of the native FG hairpin loop in DIIIE. In their case, residues Phe1, Asn3, Lys5, Thr12, Asn14 and Glu16 are structurally equivalent to Phe16, Asn14, Lys12, Thr7, Asn5 and Glu3 in peptide PHB3, but they occupy HB positions interacting together through hydrogen bonds. Residues Trp4 and Trp6, which are adjacent to residues Trp11 and Trp13 in the folded hairpin, occupy NHB positions, and were chosen so as to form Trp zippers that increase the conformational stability of the beta hairpin. Residue Trp2 corresponds structurally to residue Trp391 of DIIIE, and was chosen due to the reasons explained earlier. Residues Glu7 and Lys10 were chosen based on high scores for the preference parameter and on the fact that they can form a salt bridge that would provide further stability to the hairpin. The choice of Asn9 was also based on a high score for the preference parameter. Residue d-Pro8 in PHB6 was selected because this is a favorable amino acid for position 2 in type IIP beta turns. The d-Lys residue introduced in PHB5 (at position 8) is aimed at mimicking the Lys385 residue of DIIIE; in addition, d-Lys is more favorable for position 2 of type IIP beta turns than its L stereoisomer. Alternatively, Lys10 in PHB5 and PHB6 may also mimic Lys385 from DIIIE.
The C-terminal extensions introduced in peptides PHB1-9 consist of a Lys tripeptide. Their purpose is to increase the solubility of the disclosed beta hairpin peptides and to confer them a cationic character, thus favoring their interaction through electrostatic forces with receptor LRP1, which is an anionic protein. In the case of peptides PHB5 and PHB6, the cationic tripeptide is joined to the beta hairpin through a di-glycine dipeptide spacer, intended to provide some flexibility between both segments and to block the prolongation of the extended beta structure of strand Beta2.
In addition to peptides PHB1-9, Table 1 shows the sequence of peptide HDIII3CL (SEQ ID No.10) and several variants thereof, bearing substitutions of selected residues to an alanine residue, which were used to investigate the importance of the substituted residues for the antiviral activity of this peptide. The peptides thus designed were HDIII3CL2, where peptide HDIII3CL was extended C-terminally with 3 lysine residues (SEQ ID No. 11); HDIII3CLW-, a Trp17→Ala mutant of HDIII3CL2 (SEQ ID No. 12); HDIII3CLK-, a Lys14→Ala mutant of HDIII3CL2 (SEQ ID No. 13); and HDIII3CLC-, a double Cys1→Ala and Cys18→Ala mutant of peptide HDIII3CL2 (SEQ ID No. 14).
The peptides PHB1-9, disclosed by the present invention, represent specific examples of beta hairpin peptides with highly potent antiviral activity against DENV that can be designed by following the general principles exposed in the Description section and in this Example of the present invention. The present application, therefore, covers any beta hairpin peptide whose sequence is analogous to at least one of the peptides of the PHB1-9 set, such that its sequence identity is equal to or higher than 70%, and preferably 80%. Such analogous peptides would bear differences in one or several positions selected among: 1) modifiable positions a-e), described in the Description of the Invention section; in this case, the residue(s) in PHB1-9 would be replaced by residues also exhibiting a high structural propensity for occupying that position in beta hairpins; 2) potentially functional positions, described in the Description of the Invention section; in this case, the residue(s) in PHB1-9 would be replaced by a residue from the equivalent position of the FG beta hairpin of DIIIE from a specific DENV serotype; 3) positions corresponding to the C-terminal Lys extension; in this case, the extension comprises two or three lysine residues, preferably three, and 4) positions corresponding to the cysteines forming the disulfide bonds of peptides PHB1-4 and PHB7-9; in these cases one member of the Cys pair may be replaced by Asp/Glu or Lys and the other by Lys or Asp/Glu, such that the peptide may be stapled by forming an amide bond between the side chains of these residues, that is, Asp/Glu on one side and Lys on the other.
The sequences of beta hairpin peptides analogous to peptides of the PHB1-9 set may be described, in general terms, as follows:
i. Peptides analogous to PHB1 and PHB2: these are peptides exhibiting a sequence identity of 70% or higher (preferably 80% or higher) with PHB1 or PHB2, wherein their sequence consists of,
ii. Peptides analogous to PHB3 and PHB4: These are peptides with a sequence identity of 70% or higher (preferably 80% or higher) relative to peptides PHB3 and PHB4, wherein said analogous peptides have a sequence consisting of:
iii. Peptides analogous to PHB7: These are peptides with a sequence identity of 70% or higher (preferably 80% or higher) with peptide PHB7, wherein said analogous peptides have a sequence consisting of,
iv. Peptides analogous to PHB8 and PHB9: These are peptides with a sequence identity of 70% or higher (preferably 80% or higher) relative to peptides PHB8 and PHB9, wherein said analogous peptides have a sequence consisting of:
v. Peptides analogous to PHB5 and PHB6: These are peptides with a sequence identity of 70% or higher (preferably 80% or higher) relative to peptides PHB5 and PHB6, wherein said analogous peptides have a sequence consisting of:
The peptides listed in Table 1 were obtained by solid phase synthesis on Fmoc-AM-MBHA resin, following the Fmoc/tBu strategy (Barany, G. & Merrifield, R. B. (1977) J Am Chem Soc. 99 7363-7365). The process was performed manually, on 10 mL syringes fitted with a porous fritter, so that all reagents and solvents could be conveniently removed by filtration under vacuum. The amino acids were coupled using the DIC/HOBt activation method, and completion of the coupling reaction was verified with the ninhydrin test (Kaiser, E., et al. (1970) Anal Biochem. 34 595-598). The synthesized peptides were released by treating the resin with a trifluoroacetic acid/EDT/H2O/TIS (94%/2.5%/2.5%/1%) solution, precipitated with ether and lyophilized for 72 h, after which they were stapled by forming a disulfide bond via oxidation with dimethyl sulf oxide (DMSO) (Andreu, D., et al. (Eds), Peptide Synthesis Protocols, Methods in Molecular Biology, Totowa, N.J., 1994, pp. 91-169). The resulting peptides were purified by preparative RP-HPLC in an RP-C18 column, the collected fractions were analyzed independently by analytical RP-HPLC, and the final peptide preparation was obtained by pooling the fractions of purity higher than 99%. Mass spectrometry (ESI-MS) was used to verify the molecular weight of the final preparation.
Mass spectra were acquired with a hybrid octagonal geometry QTOF-2TM mass spectrometer (Micromass, UK) fitted with a Z-spray electrospray ionization source. The software package MassLynx, ver. 3.5 (Waters, USA) was used for spectra acquisition and processing.
In order to demonstrate that the beta hairpin peptides disclosed by the present infection can inhibit DENV infection in vitro, said peptides were evaluated in a plaque reduction assay using the Vero cell.
The cells were grown in 24 well plates until the monolayer reached approximately 90% confluence, after which they were washed twice with MEM medium without Fetal Bovine Serum (FBS). Then, peptide dilutions were added and the cell-peptide mixtures were incubated typically for 1 hour at 37° C., followed by the addition of a DENV serotype 2 preparation (NIBSC code S16803) at a multiplicity of infection (m.o.i.) of 0.001. The cells were then incubated for 1 hour at 37° C., and once the period of incubation with the viral inoculum concluded, were washed again to remove unbound virions and incubated for 5 days at 37° C. in high density medium (MEM supplemented with non-essential amino acids, 1% FBS, 1% carboxymethylcellulose) in order to let viral plaques form. Afterwards, the cells were stained with 0.1% Naphtol Blue Black in 0.15 M sodium acetate. Two replicates per experimental point were used in each assay, and three independent determinations were performed for each sample.
The toxicity of the beta hairpin peptides was assessed with the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide, Invitrogen, USA) assay (Mosmann, T. (1983). J. Immunol. Methods 65, 55-63.), and by manual cell counting using a Neubauer chamber. Ninety-six well plates (Costar, USA) were seeded with 200 μL/well of a suspension containing Vero cells at 1×105 cells/mL and incubated until the monolayers reached approximately 90% confluence, typically for 18-24 h. After washing the cells twice with DMEM (Dulbecco's Modified Eagle Medium) medium, 50 μL/well were added of the peptide or additive dilutions to be evaluated (Triton X-100™ dissolved in PBS was used as positive toxicity control) and the plates were incubated at 37° C. under a 5% CO2 atmosphere for 2 or 24 h. Two different procedures were then employed to measure toxicity:
a) A 2 mg/mL solution of MTT in PBS was added (50 μL/well) and the plates were incubated for further 4 h at 37° C., 5% CO2. The medium was then removed, and 100 μL/well of DMSO were then added to solubilize formazan precipitates. Finally, the optical density (OD) of the supernatants at a wavelength of 540 nm was measured using a plate reader (Sensident Scan™, Merck, Germany).
b) The plates were washed with 200 μL/well of DMEM medium and then a trypsin solution (Sigma, USA) was added at 50 μL/well. After inactivating the trypsin by adding 150 μL/well of DMEM supplemented with 10% FBS, the cells were counted using the Trypan Blue (Gibco, USA) vital stain. The data were processed using the statistical software package Prism v5.3 (GraphPad, USA), using non-linear regression to fit them to a log10 (concentration) vs response curve. Table 2 shows the calculated IC50 values corresponding to the antiviral activity of the beta hairpin peptides against DENV2 in Vero cells, as well as their toxicity and selectivity index. All the analyzed peptides exhibited detectable antiviral activity in this system. Peptides PHB2, PHB3, PHB4, PHB5, PHB8 and PHB9 exhibited higher antiviral potency than peptide HDIII3CL, although in the case of peptides PHB3, PHB5 and PHB8 the difference was small. In this in vitro system, peptides PHB1, PHB6 and PHB7 were actually less potent than peptide HDIII3CL. Peptide PHB4 exhibited antiviral activity in the nanomolar range and an excellent selectivity index of 4333. This peptide therefore constitutes a lead molecule with excellent properties as a candidate antiviral drug against DENV.
The results also demonstrated that the six-residue inter-strand loop of peptide HDIII3CL is completely dispensable, as it has been deleted from the sequence of the beta hairpin peptides disclosed by the present invention and replaced with a beta turn with two central residues, with no similarity, complete or partial, to the original sequence of the DIIIE loop, without deleterious effects on antiviral activity in vitro. This is a desirable characteristic if the peptides are to be used during secondary DENV infections, because this loop is immunodominant during the human antibody response against DENV (Sukupolvi-Petty S, et al. (2007). J Virol.; 81(23):12816-26) and pre-existing antibodies might, therefore, neutralize the antiviral activity of FG beta hairpin-based peptides should the loop not be eliminated. Hence, deleting the loop in the peptides disclosed in the present invention eliminates this possibility.
The data also points at several common features among the most potent beta hairpin peptides, which probably are playing an important role on increasing their antiviral activity. One of them, for instance, is the presence of a lysine (or d-lysine) in a central position on the inter-strand beta turns. Peptide PHB5 is more potent than PHB6, and their only difference is the absence of a lysine residue on the beta turn of the latter (Table 1). Also, PHB2 and PHB4 are more potent than their counterparts PHB1 and PHB3, and neither PHB1 nor PHB3 have a lysine residue at the central positions of the loop (this lysine is thought to mimic functionally the Lys386 residue of DIIIE). Another distinctive feature of the most potent peptides is the presence of a tyrosine residue at the position equivalent to that of residue 377 in DIIIE (ordinates from the E protein of DENV3). For instance, peptide PHB9 is more potent than PHB8, and their only difference is that Tyr5 has been replaced by Asn5 in the latter. Also, PHB2 (containing Tyr3) and PHB4 (containing Tyr5) are more potent than their peptide counterparts PHB1 and PHB3, where these tyrosine residues have been replaced by asparagine residues.
Finally, the data confirmed that extending the beta strands by adding four (two per strand) additional amino acids had a favorable effect on the antiviral activity of the resulting peptides. The potency of peptide PHB4, for instance, was higher than that of PHB2, and that of peptides PHB8 and PHB9 was higher than that of PHB7. Peptide PHB4 is the most potent of the entire series, and the only one whose sequence contains all the three distinctive features of potent peptides identified above. Hence, the experimental data obtained in this example demonstrated, taken together, that the criteria followed for designing the beta hairpin peptides disclosed in the present invention were judiciously chosen.
‡non toxic at 1000 μM.
Next, it was decided to compare the antiviral activity of peptides PHB4 and HDIII3CL against all four DENV serotypes, with the objective of comparing their respective IC50 values. This comparison was performed with plaque reduction assays on Vero cells, adding the viral inoculum for each serotype (DENV1, NIBSC code West Pac 74, DENV2, NIBSC, code S16803; DENV3, NIBSC, code CH53489; DENV 4, NIBSC code TVP360) at a m.o.i. of 0.001. As can be observed in Table 3, peptide PHB4 was active against the four serotypes, and much more so than peptide HDIII3CL (4000- to 7500-fold more potent, depending on serotype). The fact that PHB4 exhibited inhibitory activity against the four serotypes is consistent with the proposed role of the α2M*/LRP1 complex as endocytic receptor for all DENV serotypes and with the experimental evidence demonstrating the binding of peptide PHB4 to these proteins, as shown later in this application.
Human α2M was purified from 380 mL of human plasma, obtained by pooling plasma samples from healthy 30- to 40-years old volunteers. The plasma was dialyzed against deionized water, with frequent changes, for 72 at 4° C. Insoluble material was removed from the resulting dialysate by centrifugation at 10,000×g for 30 min, and the supernatant was equilibrated to PBS pH 6.0 by dialysis and then loaded onto an XK 50/30 column (Amersham, UK) packed with 65 mL of Chelating Sepharose Fast Flow (Amersham, UK) previously loaded with Zn2+ and equilibrated with PBS pH 6.0. After loading the sample, the column was washed with PBS pH 6.0 until the absorbance of the eluent dropped to baseline, and bound proteins were eluted with 10 mM sodium acetate/150 mM sodium chloride pH 5.0 buffer. The collected eluate was concentrated by ultrafiltration and then loaded onto a Superdex 200 (Amersham, UK) gel filtration column equilibrated with PBS pH 7.8. The presence of α2M in the highest molecular weight fraction was verified by Western blotting with an anti-human α2M polyclonal antibody preparation (Sigma, USA).
Purified α2M was activated by incubating it with 200 mM methylamine in 50 mM sodium phosphate, 150 mM sodium chloride, pH 7.4. The resulting α2M_MeNH2 (α2M*) preparation was extensively dialyzed against 50 mM sodium phosphate/0.5 M sodium chloride pH 7.8.
The ability of peptides PHB1-9 and HDIII3CL to inhibit the interaction of recombinant DIIIE1J with α2M* was analyzed using a competition ELISA format. The plates were coated with purified α2M* and incubated with previously purified and biotinylated DIIIE1J (DIIIE1Jbiot) in the presence of different concentrations of the test peptides and/or recombinant DIIIE1J, detecting α2M*-bound DIIIE1Jbiot with a streptavidin-peroxidase conjugate. Details of the experimental procedure followed to obtain recombinant DIIIE1J (SEQ ID No. 19) are presented later, but essentially, the molecule consists of residues 289-400 (numbering according to sequence PIR:A32401) of protein E of DENV1, strain 1636 (Chu, M. C., et al. (1989). Journal of General Virology 70 (Pt 7), 1701-1712.), corresponding to domain III followed by a C-terminal six-histidine tag.
At peptide concentrations in the low micromolar to submicromolar/nanomolar range, the beta hairpin peptides PHB5, PHB7, PHB8 and PHB9 partially inhibit the binding of DIIIE1Jbiot to α2M* (
The maximum inhibition percentage exhibited by peptide HDIII3CL in these assays was only 18%; lower than the inhibition percentage of even PHB7, the smallest beta hairpin peptide. This indicates that in practice, the residues of the inter-strand loop of the FG hairpin of DIIIE are dispensable for the purpose of interacting with protein α2M*, as the original 6-residue long loop has been replaced in peptide PHB7 by a type IP beta turn with only two central residues. The sequence of this beta turn bears no similarity—total or partial—to that of the original loop, its residues having been chosen based on structural criteria (high type IP beta turn forming propensity, beta hairpin connectors). The only residue from the original HDIII3CL loop that still remains in peptide PHB7 (and PHB8 and PHB9 as well) is Lys11 (HDIII3CL numbering), whose functional role is mimicked by Lys9 of PHB7, which occupies a similar (structurally quasi-equivalent) spatial position.
In addition to Lys9 (which mimics Lys11 from HDIII3CL) there are other PHB7 residues mimicking the role of HDIII3CL residues: a) Asn3, Lys10 and Asn12 from PHB7 mimic residues Asn3, Lys14 and Asn16 from HDIII3CL (these are the beta hairpin residues that do not form hydrogen bonds), whose side chains project into the exposed face of the FG hairpin of DIIIE and are, therefore, potentially functional; b) Trp13 from PHB7 is equivalent to Trp17 from HDIII3CL, and corresponds to a Trp residue that is strictly conserved across all flaviviruses; c) Cys1 and Cys14 from PHB7 form a disulfide bridge equivalent to that formed by Cys1-Cy18 from HDIII3CL. The fact that PHB7 indeed retains, and as a matter of fact improves, upon the DENV inhibitory activity of HDIII3CL suggests that the features that were retained during the design of the former do play an important role for the functional activity of the peptides disclosed by the present invention.
Peptides PHB8 and PHB9 are better inhibitors than peptide PHB7. This indicates that the addition of four residues (two per strand) to PHB8 and PHB9 contributes favorably to the DENV inhibitory activity of the peptides disclosed by the present invention.
As can be observed in
An essential element of the present invention is the analysis of the biological activity of the disclosed peptides at the nanomolar range. High potency peptides allow the use of lower therapeutic dosages and the advantages they entail in terms of lower costs and lower likelihood of appearance of a number of issues associated with high dosages, such as aggregation, non-specific interactions, antigenicity and immunogenicity. As shown in
In order to determine which residues in DIIIE from DENV1 play an important role in the interaction of said molecule with a number of different ligands, a library of single-residue mutants was prepared where each solvent-exposed residue was systematically replaced by an alanine residue (a technique more commonly known as ‘alanine scanning’) in order to later study the binding of said mutant variants to the ligands to be analyzed. In this type of experiments, any variation regarding the binding of a particular variant to the ligand is interpreted as evidence of the involvement of the mutated residue in the interaction of the wild-type protein with said ligand.
In order to prepare this library, the initial analysis was circumscribed to residues 289 to 395 (numbering according to GenBank: AAN32775.1) of the envelope protein of DENV1, strain PRS 288690 (Goncalvez, A. P., et al. (2002), Virology 303 (1), 110-119.). The sequence corresponding to this fragment of the viral polyprotein, herein defined as DIIIE from DENV1 strain PRS 288690 (DIIIE1PRS), is shown in SEQ ID No. 15.
Instead of a brute-force approach where every residue was replaced by alanine one at a time, it was decided to make a previous selection based on relative solvent accessibility, as estimated using the WHAT IF version 2005091 9-1718 software package (Vriend, G., (1990), J. Mol.Graph. 8, 52-6), and based on the estimation of the difference in stability of every possible variant with respect to the wild-type protein, expressed as the ΔΔG calculated by FoldX version 6.0 (Schymkowitz, J., et al. (2005). Nucleic Acids Res. 33, W382-W388). Both calculations were based on homology models of DIIIE1PRS (residues 1-105 in SEQ ID No. 15), obtained from the crystallographic coordinates of protein E from DENV3 (Modis, Y., et al. (2003). Proc. Natl. Acad. Sci. U.S.A 100, 6986-6991) and submitted to an energy minimization process.
The criterion used to select residues to be mutated to Ala was a relative accessibility higher than 15% and a ΔΔG lower than 4 kcal/mol. The 79 positions meeting this criterion and the results of the calculations for the parameters mentioned above are shown in Table 4.
After selecting the DIIIE1 PRS residues to be used for the alanine scan, recombinant plasmids were prepared for the heterologous expression of each Ala-mutant variant of the library in E. coli. This was accomplished by synthesizing, using the method of Agarwal et al. (Agarwal K L, et al. (1970), Nature 227, 27-34), and starting from oligonucleotides synthesized on solid phase via phosphoramidite chemistry (Beaucage S L & Caruthers M H (1981). Tetrahedron Letters, 22, 1859), a double-stranded DNA molecule coding for residues 289-400 (numbering according to GenBank: AAN32775.1) of protein E of DENV1, strain PRS 288690 (Goncalvez, A. P., et al. (2002). Virology 303 (1), 110-119), followed by a C-terminal 6-histidine tag; a recombinant protein defined here as recombinant DIIIE1PRS (rDIIIE1PRS, SEQ ID No. 17). This double-stranded DNA molecule (SEQ ID No. 16) contains recognition sites for the Nde I and Xho I restriction enzymes, designed so that the fragment can be inserted into plasmid pET22b (Novagen Inc., USA) in the same reading frame as the start codon provided by said plasmid. After digesting this double-stranded DNA molecule with the Nde I and Xho I restriction enzymes under the conditions specified by their manufacturer, the digested fragment was ligated, using T4 DNA ligase under the conditions specified by its manufacturer, to plasmid pET22b (Novagen Inc., USA) previously digested in the same manner. The ligation mixture thus obtained was transformed into the XL-1 Blue strain of E. coli (Bullock W O, et al. (1987). Biotechniques; 5:376-8) as described by Sambrook et al. (Sambrook J, et al. Molecular cloning: A laboratory manual. New York, USA: Cold Spring Harbor Laboratory Press; 1989), and the plasmids from colonies grown in selective medium were screened, purified and sequenced to obtain a plasmid whose sequence corresponded to the expected sequence. Said plasmid was denominated pET-DIII DENV1 (SEQ ID No. 18), and is represented diagrammatically in
The construction of recombinant plasmids to express the 79 previously selected Ala mutants of DIII1PRS was performed as described above for pET-DIII DENV1, but synthesizing in each case a different double-stranded DNA molecule in which the codon corresponding to the residue to be mutated was replaced by a GCG triplet, corresponding to an alanine codon. The codon that was replaced in each variant, together with the amino acid for which it coded, is shown in Table 5.
Transformation of the Plasmids of the Library of DIIIE1PRS Mutants into E. Coli and Cryopreservation of the Obtained Clones
In order to obtain clones of E. coli cells containing the plasmids for expressing in this host the selected DIIIE1 PRS variants, transformation-competent cells of the E. coli strain BL21(DE3) (Studier, F. W. & Moffatt, B. A. (1986) J. Mol. Biol. 189(1), 113-130) were prepared and divided into aliquots, each of which was separately transformed with 20 ng of one of the plasmids of the mutant library, using methods known to those skilled in the art (Sambrook, J., et al. Molecular cloning: A laboratory manual. 1989. New York, USA, Cold Spring Harbor Laboratory Press). The transformed aliquots were plated separately onto LB-agar plates containing ampicillin at 100 μg/mL. After incubation for 12 hours at 37° C. to enable bacterial growth, one well-isolated colony from each plate was inoculated into separate test tubes containing 5 mL of LB broth each, supplemented with ampicillin at 100 μg/mL, and incubated at 37° C. under agitation (200 rpm) until the appearance of visible turbidity. The cultures were then centrifuged aseptically at 3000×g for 20 min. at 25° C., each resuspended into 250 μL of fresh LB broth+250 μL 40% (v/v) glycerol, and in turn split into 100 μL aliquots that were stored at −70° C.
After preparing a library of cryopreserved E. coli clones expressing each of the variants of the DIIIE1PRS mutant library, a small-scale process was used to purify said DIIIE1 PRS variants. Briefly, for each variant, a single cryopreserved aliquot of the E. coli clone containing the corresponding plasmid was used to inoculate 50 mL of ZYM50502 medium (Studier, F. W. (2005). Protein Expr. Purif. 41(1), 207-234) supplemented with ampicillin at 100 μg/mL in a 1 L Erlenmeyer flask, which was then incubated for 12 hours at 37° C., 300 r.p.m. Afterwards, the culture was centrifuged at 3000×g, 25° C. for 20 min, the supernatant was discarded, and the resulting biomass was lysed by resuspension into 19 mL of AG buffer (PBX 1X, NaCl 0.3 mol/L, imidazole 20 mM) containing 6 M guanidinium hydrochloride (GuHCl), eliminating the viscosity of the homogenate by brief sonication for 30 with an appropriate probe. After clarifying the homogenate by centrifugation at 3000×g, 25° C. for 45 min, the supernatant was incubated for 1 h at 25° C. in a slow rotary shaker with 0.3 mL of Ni2+-nitrilotriacetic acid-agarose resin (Ni-NTA agarose, Qiagen, Germany) and the resulting slurry was gravity-packed into an empty NAP-10 column (GE Healthcare, USA) and washed consecutively with AG buffer containing decreasing concentrations of GuHCl (6 M to 1.2) and, finally, with buffer A (PBS 1X, NaCl 0.3 M, imidazole 20 mM). Then, the protein was eluted with 0.9 mL of buffer E (PBS 1X, NaCl 0.3 M, imidazole 300 mM), and the eluate was subjected immediately to buffer exchange into PBS 1X by gel filtration on Sephadex G25 using pre-packed PD-10 columns (Amersham, UK). Total protein concentration of the resulting preparation was determined by the bicinchoninic acid (BCA) method (Smith, P. K. (1985). Anal. Biochem. 150(1), 76-85), and purity was assessed by SDS-PAGE under reducing conditions (Laemmli, U. K. (1970). Nature 227(259), 680-685), to ensure that no degradation products were present. The library of purified DIIIE1 PRS variants was stored at −20° C. until used.
The ability of each DIIIE1 PRS alanine mutant to inhibit the interaction of recombinant DIIIE with α2M* was analyzed using a competition ELISA format. The plates were coated with purified α2M* and incubated with biotinylated DIIIE1 PRS (DIIIE1PRSbiot) in the presence of varying concentrations of each DIIIE1 PRS alanine mutant, then detecting α2M*-bound DIIIE1PRSbiot with a streptavidin-peroxidase conjugate. Optical density vs. concentration readings were fitted to a dose-response curve, used to calculate the IC50 of each mutant as well as that of wild-type recombinant DIIIE1 PRS.
As observed in
These results imply that binding of DIIIE to α2M* involves two independent surface patches in DIIIE (
The second patch mentioned above, located to the lower surface of DIIIE (
The results obtained in this experiment addressed at mapping the sites on the surface of DIIIE involved on its interaction with α2M* demonstrate the soundness of the principles followed for designing the beta hairpin peptides of the present invention. For instance, care was taken to include lysine residues that constituted structural/functional mimics of Lys385 of DIIIE, and it turns out that this is the residue individually contributing the most to the interaction (its mutation to alanine reduces the affinity of the interaction by over 22-fold). The peptide residues mimicking this residue are: Lys8 in PHB2 and d-Lys8 in PHB5, Lys9 in PHB1, PHB6 and PHB7, Lys10 in PHB4 and Lys11 in PHB3, PHB8 and PHB9. In another example, four of the beta hairpin peptides that most potently inhibit the interaction of DIIIE with α2M* (PHB4, PHB5, PHB8 and PHB9,
Light scattering was used to monitor the aggregation of peptide PHB4. The measurements were performed on a Shimadzu™ RF-5301PC (Shimadzu, Japan) spectrofluorimeter, acquiring data with the software provided by the instrument manufacturer. The wavelengths of the incident light beam (excitation) and the detector (emission) were set to 320 nm, using excitation and emission apertures of 5 nm. The scattering of light due to the peptide was calculated by subtracting, from the intensity of the light dispersed by the peptide, the intensity of the light dispersed by a solution without peptide (blank). Variations on light scattering due to the dissolution of the peptide in PBS were studied by acquiring data in time course mode, with a measurement frequency of 1 or 2 per second. The solutions of PHB4 in PBS were prepared by mixing one part of a solution of the peptide in water with one part of a PBS 2X solution.
The data acquired in this experiment revealed that once dissolved into PBS, and after a latency period shorter than 2 min, the intensity of the light dispersed by the peptide increases rapidly during the first 30-40 min and then continues to slowly increase at a linear rate up to 300 min, when the measurement finished.
for peptide concentrations of 5 and 10 μM. The profiles for both concentrations are similar, although the
maximum at 5 μM is larger (11 vs. 7) than that at 10 μM, suggesting that the aggregation that takes place at smaller peptide concentrations leads to the formation of aggregates of a larger average size compared to the original average size. The charts were prepared with SigmaPlot 10.0 (Systat Software Inc., USA).
Analysis by Transmission Electron Microscopy of the Morphology of the Supramolecular Structures formed by Peptide PHB4
In order to analyze the morphology of the aggregates formed by peptide PHB4, peptide solutions were subjected to different experimental treatments where conditions such as temperature, incubation time and the presence of electrolytes were manipulated to facilitate aggregation (
As observed in
Also, the study clearly indicates that peptide PHB4 forms supramolecular aggregates whose morphology depends on exact medium composition and temperature. The dimensions of the observed fibrillary structures are consistent with those of amyloid-type structures, which typically grow as elongated fibers through the formation of extended beta sheets.
The outcome of protein aggregation processes depends on the exact balance between repulsion and attraction among protein molecules (Juarez, J., et al. (2009) Biophysical Journal 96[6], 2353-2370). It is known that adding electrolytes to protein solutions often increases the rate of formation of fibrils, due to the shielding of electrostatic repulsions between protein molecules by said electrolytes, which thus tips the balance towards attractive intermolecular interactions and, hence, the formation of aggregates (Juárez, J., et al. (2009) Biophysical Journal 96[6], 2353-2370; Sagis, L. M., et al. (2004). Langmuir 20, 924-927). The kinetics of the process of intermolecular interaction is also favored by temperature, which has long been known to play a fundamental role on the induction of protein aggregation.
It is known that DENV infection can be blocked by molecules that interfere with the interaction between DENV virions and the cell receptor known as LRP1 (Huerta H. et al, WO 2007/124698). LRP1 is an integral membrane protein (
It is the a chain of LRP1 that interacts with extracellular LRP1 ligands. This chain exhibits the typical domain architecture of members of the family of low density lipoprotein receptors, to which LRP1 belongs. Specifically, the a chain of LRP1 contains four clusters containing 2 (cluster I), 8 (cluster II), 10 (cluster III) and 11 (cluster IV) ligand binding sites structurally related to the Cys-rich regions of proteins of the complement cascade. After each ligand binding site cluster there is an Epidermal Growth Factor (EGF)-like domain, formed by Cys-rich regions and YWTD domains.
In order to determine whether the ligand binding domains of LRP1 interact with DIIIE from DENV, recombinant DIIIE proteins were prepared from residues 289-399 of protein E from DENV1 strain Jamaica/CV1636/1977 (DIIIE1J); residues 289-399 of protein E from DENV2 strain Jamaica 1409 (DIIIE2); residues 287-397 of protein E from DENV3 strain H-87 (DIIIE3) and residues 289-399 of protein E from DENV4, strain Dominica 814669 (DIIIE4), fused to a C-terminal hexahistidine tag (SEQ ID No. 19-22), as shown in
Also, the experiment used three recombinant proteins containing clusters II, III or IV from LRP1, denominated sLRP1-CII (SEQ ID No. 23), sLRP1-CIII (SEQ ID No. 24) and sLRP1-CIV (SEQ ID No. 25), respectively, fused to the constant region of a human IgG1 molecule (R&D Systems, USA) (
The evaluation of DIII/LRP1 interactions was performed with an ELISA assay, in which 96-well plates were coated with proteins DIIIE1J or DIIIE2-DIIIE4 at 10 μg/ml, then blocked with Bovine Serum Albumin (BSA) and incubated for 1 hour at 37° C. with either sLRP1 CII, sLRP1 CIII or sLRP1 CIV at 10 μg/m1 in 20 mM HEPES pH 7.5/150 mM NaCl/1 mM CaCl2/0.05% Tween 20. After washing, the plates were then incubated for 1 hour at 37° C. with a 1:1000 dilution of an anti-human IgG-peroxidase conjugate and developed with o-phenylenediamine/H2O2. The obtained data revealed a similar pattern of interactions for DIIIE from all four serotypes, where sLRP1-CIV exhibited the strongest binding followed by sLRP1-CII while sLRP1-CIII showed no detectable binding to any DIIIE molecule (
The interaction of the beta hairpin peptides with soluble fragments of the LRP1 receptor was studied by Surface Plasmon Resonance (SPR), using a Biacore X unit. In this technique, one of the interaction partners to be studied is immobilized on the surface of a chip, and the other interaction partner is dissolved in a solution that runs through the measurement cell on top of the interaction surface, under conditions of controlled flow. The association and dissociation of interacting partners is measured in resonance units (RU) and plotted, as a function of time, on charts known as ‘sensorgrams’. SPR, therefore, enables the experimenter to obtain detailed information about the interaction without having to label either interaction partner.
In this example, one channel of a CM5 chip (GE Healthcare, USA) was used to immobilize protein sLRP1-CIV on its surface at high density (see Table 6) and recombinant Human Serum Albumin (rHSA, Sigma-Aldrich, USA) was immobilized on the other channel. Based on the RU density achieved for each molecule, and assuming that there are 11 potential interaction sites for each sLRP1-CIV molecule and one potential interaction site for each HSA molecule, the surfaces at both channels would thus exhibit a similar density of binding sites per surface area unit (pmol/mm2).
The reactivity of the immobilized protein was evaluated using two known LRP1 ligands; namely, α2M and the protein known as Receptor-Associated Protein (RAP). It has been previously demonstrated that both molecules interact with cluster IV of the LRP1 receptor (Jaap G. Neels, et al. (1999). Journal of Biological Chemistry, 274, 44: 31305-31311).
Although it was decided to immobilize sLRP1-CIV at high density because these conditions make easier the detection of any potential interactions with compounds whose molecular weight is lower than that of average proteins (such as the peptides this experiment is intended to study), it must be stressed that in high-density surfaces, mass transfer limitations and the possibility of multivalent binding make difficult the accurate determination of kinetic constants. Therefore, and in order to characterize in greater detail the sLRP1-CIV binding surface, different RAP dilutions (0.6-20 μM) were loaded, and RU were registered under conditions of interaction equilibria (250 sec, see
The binding of beta hairpin peptides to sLRP1-CIV was studied by loading 20 μM dilutions in running buffer of the different peptide variants onto the sLRP1-CIV chip characterized above. The high strength of the resulting signals is consistent with the binding of aggregated, rather than monomeric forms of the peptides. Therefore, the observed variations between different peptides may reflect either differences in the affinity of unitary peptide interactions or differences in aggregation numbers and/or geometries that lead to variations in the number of peptide units able to simultaneously engage immobilized sLRP1-CIV molecules in a multivalent manner. In the case of peptides of the HDIII3CL family (peptides 10-14 in Table 1), variant HDIII3CL2 exhibits the strongest binding. There is also detectable specific binding in the case of variants HDIII3CLW and HDIII3CLK, although the strength of the interaction is much lower, suggesting that the replacement of residues Trp17 and Lys14 has a deleterious effect on binding to sLRP1-CIV. There is an at most marginal interaction in the case of variant HDIII3CLC, indicating that the replacement of residues C1 and C18 (in other words, the elimination of the disulfide bridge) affects dramatically the strength of the interaction of the peptide with sLRP1-CIV. This result is consistent with previous data on the potency of the antiviral activity exhibited by the peptides, since HDIII3CLC, HDIII3CLW and HDIII3CLK, whose binding to sLRP1-CIV is decreased relative to HDIII3CL, did not exhibit in vitro antiviral activity against DENV2 in Vero cells (see Table 2). On the other hand, the stronger binding to sLRP1-CIV exhibited by HDIII3CL2 does not translate into a more potent antiviral activity compared to HDIII3CL (see Table 2), indicating that receptor binding per se is necessary but not sufficient for antiviral activity. One possible explanation for this finding is that there is a fraction of HDIII3CL2 that binds to sites in sLRP1-CIV that do not play a relevant role in the virus-receptor interaction, and/or that it is binding with very low affinity; a reasonable supposition taking into account the multi-domain architecture of receptor LRP1 and its clusters. The difference between peptides HDIII3CL2 and HDIII3CL is that the former has an additional C-terminal lysine and has, therefore, a stronger cationic character, which may facilitate additional electrostatic interactions with the negatively charged LRP1 receptor. In addition, it must be pointed out that Lys21 does not form part of the hairpin, which is the topological region that mimics the functional patch in DIIIE described previously (Huerta V. et al, WO 2007/124698) and in this invention.
The same procedure was followed to evaluate the interaction of the beta hairpin peptides PHB1, PHB2, PHB3, PHB4, PHB5, PHB8 and PHB9 with the sLRP1-CIV surface. As shown in
Receptor LRP1 has been previously proposed as the putative endocytic receptor for DENV, and a ‘bridging’ or ‘carrier’ role has been ascribed to protein α2M* (an LRP1 ligand) in the process of virus entry into the cells (Huerta V. et al, WO 2007/124698). The previous example (Example 6) demonstrated that the beta hairpin peptides disclosed in the present invention can bind a fragment of receptor LRP1. The present example, in turn, examines whether the antiviral activity observed in Example 2 correlates quantitatively with the capacity of these peptides to bind receptor LRP1. As can be observed in
The plC50 data for the antiviral activity of peptides with a common topology exhibits a statistically significant correlation with RUrem values (r(Pearson)=−0.9957 and P<0.0001). If the peptide PHB4 data are excluded, the values of Pearson's correlation coefficient remain similar (r(Pearson)=−0.9868, P=0.0018). The analyses of correlation, linear regression and the charts were prepared with Prism v5.03 (GraphPad Software Inc., USA).
These results are consistent with an antiviral action mechanism whereby PHB peptides disturb the interaction between virions and the LRP1 receptor, leading to the inhibition of productive virus entry into the cells.
A similar analysis was carried out regarding the ability of PHB peptides to inhibit the binding of DIIIE to protein α2M* in the submicromolar/nanomolar range. As shown in Example 3, at that concentration range the peptides inhibited partially the binding of biotinylated DIIIE1J (DIIIE1Jbiot) to α2M*, with peptides PHB2, PHB4, PHB5, PHB8 and PHB9 exhibiting better (1.5- to 3-fold) inhibition percentages than DIIIE1J alone. In this case, the inhibitory capacity of the peptides (relative to the capacity of recombinant DIIIE1J to inhibit DIIIE1Jbiot binding to α2M*) also correlates with their antiviral activity against DENV2 in Vero cells, as shown in
The antiviral activity of peptide PHB4 was evaluated after treatments facilitating its aggregation or in the presence of different agents that modulate aggregation, with the purpose of examining their effect on the inhibitory activity of this peptide for the infection of Huh7.5 cells with DENV2 strain S16803. Huh7.5, a human hepatoma-derived cell line, was selected not only because it is permissive for DENV2, but because of its relevance as an experimental system from the viewpoint of DENV pathogenesis (Lin Y L, et al. (2000) J Med Virol.; 60(4):425-31).
Based on the results shown in Example 5, which evidence that changes in ionic strength result in changes to the size and morphology of peptide aggregates, an experiment was designed in which, starting from a concentrated solution of PHB4 in water, the peptide was dissolved either in (i) MEM medium or (ii) PBS (containing respectively 6.8 g/L and 8.0 g/L of NaCl, and with pH adjusted to 7.4) and subjected to the different conditions to be evaluated.
Antiviral activity was evaluated in virus yield assays, in which 80-90% confluent Huh7.5 monolayers were washed twice with non-supplemented DMEM and then infected at a m.o.i of 0.01 for 2 h at 37° C. and 5% CO2 in the presence of the different peptide preparations, after which the viral inoculum was removed by washing with non-supplemented DMEM, and the cells were grown in DMEM supplemented with 2% FBS. Cell supernatants were collected at 24 h p.i., and virus yields were determined by titrating the supernatant in a plaque formation assay (Morens D. M., et al. (1985). N. J. Clin. Microbiology; 22: 250-254).
It is well known that temperature influences significantly the aggregation, or self-assemblage of peptides into supramolecular structures (Sabaté R, et al. (2012). Biomacromolecules; 13(2):474-83). Therefore, a first series of assays was set up in which, starting from a 20 μM solution of the peptide in water, dilutions were prepared in MEM spanning the 10-0.01 μM concentration range, which were then incubated in parallel at 10° C. and 37° C. for 2 hours and then used in a viral inhibition assay. Non-incubated peptide dilutions (that is, prepared right before the viral inhibition assay) were used as controls.
The results show that the non-incubated peptide, dissolved in MEM medium, inhibited viral infection in a dose-dependent manner, with an IC50 of 0.16 μM and an IC100 of 10 μM (
A second experiment was designed to determine the influence of aggregation time on the antiviral potency of peptide PHB4 (
As shown in
The results evidence that the inhibitory activity of the peptide is sensitive to the length of the incubation period in PBS for the formation of aggregates of higher potency, where potency is understood, in this case, as the lowest peptide concentration that decreases by 50% the viral yield of the infection control for the assay. On the other hand, the incubation with HSA at 10° C. leads to the formation of lower potency variants, compared to the results obtained by incubation at 37° C. This confirms that there probably is an interaction between the peptide, or still-growing aggregates thereof, with HSA, wherein the course of the aggregation process is changed, leading to the formation of high-potency variants such as those obtained by incubation for 45-60 min. in PBS and then incubation with HSA at 37° C., that can produce total inhibition of the infection by DENV in this assay format at peptide concentrations as low as 1 nM-10 pM.
The starting concentration of the peptide is another parameter that may influence the kinetics and final results of the aggregation process in a significant manner. In order to dissect how this parameter influences the antiviral activity of peptide PHB4, an assay was set up (represented diagrammatically in
The results demonstrate that there is a relationship between antiviral activity and the initial concentration of peptide PHB4 in water before its addition to a solution of higher ionic strength to trigger the process of aggregate growth (
In all, the results evidence that peptide PHB4 can assemble into aggregated forms with an antiviral potency in the nanomolar/subnanomolar range, depending on aggregation conditions. It should be stressed that the antiviral activity assay used in these experiments is one of the most stringent tests for this purpose, as the molecule under evaluation is present only during the initial stage of the infection and is later removed during subsequent washes of the cell monolayer. Therefore, that fact that antiviral activity was detectable with this experimental set up is consistent with a mechanism of action for peptide PHB4 whereby it inhibits one of the early stages of viral entry into the cell; namely binding to the cell surface, internalization or membrane fusion.
An experiment was set up to evaluate the in vivo antiviral activity of peptide PHB4 and compare it to that of peptide HDIII3CL. For this purpose, groups of 12 adult Balb/c mice (20 g average body weight) were anesthetized by ether inhalation and inoculated intracranially with 20 μL of viral preparations containing 10 times the median lethal dose of this model and the peptide to be evaluated. The animals recovered quickly and already appeared alert five minutes after the treatment, and were monitored afterwards for 21 days. The data were evaluated using the Kaplan-Meier logarithmic rank test.
Table 8 contains the results of the experiments performed with the four DENV serotypes. At the assayed dose, peptide PHB4 affords total protection to the mice, and is therefore more potent than peptide HDIII3CL. In the case of the DENV2 strain, the experiment was repeated thrice.
aThe lethal dose for the DENV3 strain used in this assay could not be determined. Therefore, morbidity was evaluated instead, monitoring the mice for the appearance of the following symptoms: ruffled fur, limb paralysis, and hunched posture, and mice showing these signs were euthanized. ruffled fur, hunched posture, prostration, and hind limbs paralysis; and for their duration in days.
We have previously pointed out that the beta hairpin peptides disclosed by the present invention may be produced, using chemical synthesis or recombinant DNA technology, alone or fused to other protein/peptides. Also, it was demonstrated in Example 5 that the beta hairpin peptides disclosed by the present invention may aggregate and adopt different quaternary structures, depending on starting peptide concentration, solvent type and experimental conditions such as temperature, pH, ionic strength, sonication, presence of additives, non-covalent binding to proteins such as HSA, etc. The formation of these supramolecular structures greatly increases the antiviral activity of the beta hairpin peptide (See Example 8), probably due to the fact that the avidity of the interactions of peptide-based nanoparticles with their putative receptor is much higher due to multivalent binding (simultaneous interaction of several monomers in the same nanoparticle with several ligand binding domains in the receptor, or with more than one receptor on the cell surface). In addition, the formation of nanoparticles may contribute positively to the pharmacokinetic properties of the peptide by increasing its serum half life, conferring resistance to serum proteases, etc.
The nanoparticles that the beta hairpin peptides disclosed by the present invention have been shown to form may be further stabilized by fusing said peptides to ‘acceptor’ (A) groups that can bind non-covalently to ‘spacer’ (E) molecules or atoms, with a stoichiometry of two or more ‘A’ groups for each ‘E’ group. Such a design would enable the association of two or more PHB peptides with a single ‘E’ spacer. In one embodiment of the present invention, we have thus appended, via chemical synthesis, a tag consisting of five histidine residues to peptide PHB4 (
These peptide preparations (PHB4H5 and H5PHB4, either alone or in complexes with either Zn or EGTA(Zn2)) were then assayed for antiviral activity against DENV2 (strain M2C) in Vero cells. Briefly, the cells were incubated for 1 h at 37° C. in the presence of the peptide preparations, and then the viral inoculum was added at a m.o.i. of 0.001. Two hours later, a layer of high-density medium was added and the plates were incubated for 3 days at 37° C., 5% CO2. Cells incubated only with solutions of ZnCl2 and EGTA(Zn2) at the corresponding concentrations before viral challenge were used as negative controls. The viral plaques were detected using an immunofoci technique with a monoclonal antibody specific for the viral envelope, and inhibition percentages were calculated with the following expression:
I=100−100×NPpeptide/NPNegative control
Where I stands for the inhibition percentage and NP for the number of viral plaques, averaged across replicates.
As can be observed in
| Number | Date | Country | Kind |
|---|---|---|---|
| CU 2014-0026 | Mar 2014 | CU | national |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/CU2015/000002 | 2/26/2015 | WO | 00 |
