Research Article

Potent peptidic fusion inhibitors of influenza virus

See allHide authors and affiliations

Science  27 Oct 2017:
Vol. 358, Issue 6362, pp. 496-502
DOI: 10.1126/science.aan0516

Broadly reactive drugs for flu

Drugs for influenza are limited. For those available, viral resistance is rife. Part of the problem is that the virus is constantly mutating. Kadam et al. tested the cell entry stage of the virus life cycle as a drug target (see the Perspective by Whitehead). Cell entry is mediated by the major surface glycoprotein hemagglutinin (HA). This stage can be blocked by broadly neutralizing antibodies binding to HA. The authors generated small cyclic peptides that bind to the same sites on HA as the antibodies and mimic their activity. The peptides are cheap and easy to synthesize, are nontoxic to mice, and prevented infection of cells by many types of influenza virus.

Science, this issue p. 496; see also p. 450

Abstract

Influenza therapeutics with new targets and mechanisms of action are urgently needed to combat potential pandemics, emerging viruses, and constantly mutating strains in circulation. We report here on the design and structural characterization of potent peptidic inhibitors of influenza hemagglutinin. The peptide design was based on complementarity-determining region loops of human broadly neutralizing antibodies against the hemagglutinin (FI6v3 and CR9114). The optimized peptides exhibit nanomolar affinity and neutralization against influenza A group 1 viruses, including the 2009 H1N1 pandemic and avian H5N1 strains. The peptide inhibitors bind to the highly conserved stem epitope and block the low pH–induced conformational rearrangements associated with membrane fusion. These peptidic compounds and their advantageous biological properties should accelerate the development of new small molecule– and peptide-based therapeutics against influenza virus.

The influenza drug therapies that are currently available worldwide target only two viral proteins—the M2 channel and neuraminidase (NA)—both of which function at critical stages of the virus life cycle. M2 is involved in proton-conducting activity in the early and late stages of replication, and NA is involved in the release of nascent virions (1, 2). However, escape mutations have resulted in drug-resistant viruses against which the therapeutic effects of these drugs are diminished or ablated (3, 4). Therefore, therapeutics with different mechanisms of action are urgently required to combat the persistent global threat imposed by influenza virus. Therapeutic strategies aimed at targeting the highly conserved functional regions on influenza hemagglutinin (HA) that are involved in viral entry should be highly effective and may reduce the likelihood of generating escape mutants.

Infection by influenza virus is initiated by HA receptor binding at the cell surface and then by HA-mediated fusion of viral and host cellular membranes in the low pH of endosomal compartments (5). HA is the major glycoprotein on the influenza virus surface and is a homotrimer, with each monomer composed of two subunits, HA1 and HA2, linked by a single disulfide bond (6). The HA1 subunit forms the membrane-distal globular head that contains the receptor binding site (RBS), as well as the highly variable immunodominant regions that surround the RBS. HA2 and the N- and C-terminal regions of HA1 form the highly conserved, membrane-proximal stem. HA1 interacts first with sialylated receptors on the host-cell epithelial surface, after which the bound virus is internalized by endocytosis. At the low pH of endosomes, the fusion potential of HA is activated, and HA dramatically rearranges to release its fusion peptide, which anchors to the endosomal membrane, thereby triggering events that lead to fusion of the viral and host membranes (7).

Recent breakthroughs in isolation and characterization of human broadly neutralizing antibodies (bnAbs) [reviewed in (8, 9)] against HA have raised hopes for a more universal vaccine, as well as the possibility of designing therapeutics that target this region and may be less prone to resistance. Such bnAbs target the HA membrane-proximal stem region and block the large conformational rearrangements associated with membrane fusion, thereby neutralizing the virus. Structural characterization of their epitopes identified a highly conserved site of vulnerability at the HA1/HA2 interface in the HA stem that is present on influenza A group 1 and group 2 serotypes—which encompass 18 HA subtypes (H1 to H18)—and on both influenza B lineages (10).

This relatively recent structural information from bnAb-HA complexes has already inspired and guided the design of nonimmunoglobulin protein scaffolds against the HA stem (11, 12). Small proteins, such as HB80 and HB36, were engineered de novo on the basis of the paratope of bnAb CR6261 and its interactions with the HA stem. Particular amino acid side chains were designed to occupy conserved hydrophobic pockets in the HA stem and then displayed in the appropriate configuration and conformation on selected scaffold proteins. These small proteins show comparable affinity, mode of binding, and neutralization breadth to CR6261, and they inhibit the low-pH conformational change in HA. Improved variants of HB36 also protect mice against lethal challenge from the 2009 H1N1 pandemic virus (13).

The success of these effective protein scaffolds encouraged us to explore the feasibility of engineering even smaller ligands against influenza virus. Smaller peptidic ligands present a promising alternative to biologics by retaining high affinity and selectivity and overcoming challenges such as cell permeability, low bioavailability, and high production cost (14). Furthermore, passive immunotherapy against influenza currently requires hospitalization for intravenous treatment (15). Strategies to develop peptides based on complementarity-determining region (CDR) loops have been reported for other viral targets (1618). However, the absence of any structural insights into their mechanism of action has hindered further development. Our approach was guided by the collective structural information derived from the interactions of several hydrophobic pockets in the influenza HA stem with heavy-chain CDR (HCDR) loops and framework region 3 (FR3) of bnAbs FI6v3 and CR9114. Here we report on the successful development and structural characterization of potent cyclic peptides as influenza HA fusion inhibitors.

bnAb-guided peptide design

A rich compendium of structural and functional information is available on HA stem–targeting bnAbs encoded by germline genes VH1-69 [CR9114 (10), CR6261 (19), F10 (20), and A06 (21)] and VH3-30 [FI6v3 (22), 3.1 (23), and 39.29 (24)]. BnAbs CR6261, F10, A06, and 3.1 are specific against influenza A group 1 viruses, whereas FI6v3, 39.29, and CR9114 show pan-influenza reactivity against influenza A group 1 and 2, and CR9114 also shows reactivity against influenza B viruses. All of these bnAbs interact with a highly conserved hydrophobic groove at the HA1/HA2 interface in the HA stem. For CR9114, the key interacting residues are derived from HCDR2, HCDR3, and FR3, whereas in FI6v3, HCDR3 serves as the dominant interacting motif (Fig. 1A). Therefore, to recapitulate most of these interactions within a single peptide with greater feasibility of cyclization, the HCDR3 loop of FI6v3 was chosen as the starting point for construction of peptidic ligands (Fig. 1, B and C).

Fig. 1 bnAb-based cyclic peptide design strategy.

(A) Overlay of structures of influenza hemagglutinin (HA) from group 1 strain H1/PR8 [Protein Data Bank (PDB) ID, 1RU7; gray), with HA-interacting loop residues from the Fabs of HA-Fab complexes of CR9114 (PDB ID, 4FQI; green) and FI6v3 (PDB ID, 3ZTN; yellow). One HA protomer of H1/PR8 is represented as a cartoon, and the other two protomers are shown in surface representation (HA1, dark gray; HA2, light gray). Loop residues from the complementarity-determining regions of the heavy (H) and light (L) chain (CDR1, -2, and -3) and framework region 3 (FR3) of each antibody that interact with the hydrophobic groove in the HA1/HA2 interface are in stick and ribbon representation. (B) Overlay of the hydrophobic groove–interacting side chains from CR9114 HCDR2, HCDR3, and FR3 (green) with the HCDR3 loop from FI6v3 (yellow). CR9114 FR3 residues occupying one of the hydrophobic cavities on the HA stem not occupied by FI6v3 are highlighted with a dashed red oval.(C) Blueprint of the peptides constructed by merging features from the HCDR3 loop of FI6v3 (yellow) with FR3 of CR9114 (green). Nonproteinogenic amino acids (NPAAs) used to link features derived from the two antibodies are shown in magenta. (D) Representative peptides from this study. Amino acid sequences are shown for HCDR3 of FI6v3, linear peptide P1, and cyclized peptides P2 to P7, with the chemical structures of corresponding NPAAs illustrated below. Suc, succinyl; Ac, acetyl; ter., terminal. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

A series of linear peptides varying in length, terminal-capping moieties, and mutations (table S1 and fig. S1) was synthesized, characterized by liquid chromatography–mass spectrometry (LC-MS), and tested in an AlphaLISA competition binding assay against a panel of HAs from viruses representing influenza A group 1 [H1N1 A/California/07/2009 (H1/Cal), H1N1 A/New Caledonia/20/1999 (H1/NCa), and H5N1 A/Vietnam/1203/2004 (H5/Viet)], influenza A group 2 [H3N2 A/Brisbane/10/2007 (H3/Bris) and H7N7 A/Netherlands/219/2003 (H7/Neth)], and influenza B [B/Florida/4/2006) (B/Flo)] (supplementary text S1). The linear sequence variant P1 displayed weak binding competition on group 1 H1 and H5 HAs (Figs. 1D and 2A and fig. S2). Peptide P1 is identical to FI6v3 HCDR3, except at position 4, where Leu100B in FI6v3 was replaced by Glu4 to improve peptide solubility (Fig. 1D). (Peptide residues are indicated in bold throughout.)

Fig. 2 In vitro binding, neutralization, and crystal structure of the peptides.

(A) Peptide-mediated binding potencies, determined using an AlphaLISA competition (ALC) assay, reported as IC50. Peptides were tested against group 1 H1 HAs (H1/Cal and H1/NCa), an H5 HA (H5/Viet), a group 2 H3 HA (H3/Bris), an H7 HA (H7/Neth), and an influenza B HA (B/Flo). Either CR9114 Fab or small protein HB80.4 was used as a positive control. (B) Virus neutralization potential of peptides, determined by virus neutralization assay (VNA), reported as EC50. N.D., not determined. (C) SPR kinetic data for the peptides, as determined for group 1 H1 and H5 HAs. (D to F) Crystal structure of peptide P2 in complex with H1/PR8 HA. (D) P2 binding residues on the molecular surface of H1/PR8 HA (HA1, dark gray; HA2, light gray). (E) P2 (yellow sticks) in complex with H1/PR8 HA. A 2Fo-Fc electron density map (black mesh) contoured at 1σ is displayed around P2. Residues of the peptide are numbered as in Fig. 1D. (F) Noncovalent interactions of P2 with H1/PR8 HA. The peptide is shown as yellow sticks, HA in gray, and waters as red spheres. The distances are in angstroms.

In vitro affinity maturation of stem-targeting peptides

Starting from the sequence of P1, the peptide conformation was constrained by cyclization. To determine optimal ring size, a library of cyclic peptides of varying lengths was constructed, incorporating nonproteinogenic amino acids (NPAAs) for lactam formation and using various cyclization strategies (head to tail, side chain to side chain, and side chain to tail). In the side chain to tail constructs, an ornithine (Orn2) side chain was fused with the carboxyl terminus of β-alanine (XD11) (Fig. 1, C and D). Synthesis and screening of this peptide library (table S2 and supplementary text S2) led to P2, with micromolar binding and potency against H1/Cal, H1/NCa, and H5/Viet HAs [equilibrium dissociation constants (KD) of 1 to 3 μM, determined by surface plasmon resonance (SPR); median inhibitory concentrations (IC50) of 2 to 8 μM, determined by AlphaLISA assays]; however, no binding was observed to group 2 or influenza B HAs (Figs. 1D and 2, A to C, and figs. S1 to S3). Thus, constraining the peptide backbone through cyclization improved the performance of the peptidic ligands against group 1 HAs.

In the next round, further optimization was aided by our cocrystal structure of P2 with H1 HA from A/Puerto Rico/8/1934 (H1N1) (H1/PR8) (Fig. 2, D to F). The Arg1 guanidinium moiety was found to have unfavorable interactions with a hydrophobic cavity in the stem. However, the corresponding stem cavity is occupied by Ile73 and Phe74 of FR3 in CR9114 (Fig. 1, A and B). Thus, Arg1 of P2 was substituted by the hydrophobic 5-phenyl-norvaline (XA1) in P3 (Fig. 1D), which improved binding by 11- to 26-fold (KD = 104 to 125 nM) and potency by 9- to 15-fold (IC50 = 177 to 766 nM) compared with P2 (Fig. 2, A and C). The P3-HA complex was also slightly more stable, with a slower dissociation rate (koff ≈ 0.2 s−1) than P2-HA (koff > 1 s−1) (Fig. 2C and fig. S3).

Peptides P4, P5, and P6 (Fig. 1D) were then constructed on the basis of P2 and P3, which suggested that rigidification of the peptide macrocycle and optimization of the HA-peptide interactions through incorporation of NPAAs could improve stability of the complex and enhance binding and neutralization. All three peptides demonstrated improved binding (KD), complex stability (koff), and potency (IC50) compared with P3 (Fig. 2, A and C). Whereas P4 showed modest improvement of two- to threefold (KD = 47 to 75 nM; IC50 = 76 to 228 nM), P5 and P6 exhibited three to 13 times the affinity and potency (KD = 17 to 37 nM; IC50 = 30 to 70 nM) against all three HAs tested (Fig. 2, A and C, and figs. S2 and S3). Furthermore, these peptides neutralized H1N1 and H5N1 viruses (Fig. 2B).

To investigate whether binding and neutralization could be further improved, the enhancing features of P4, P5, and P6 were incorporated into a single peptide, P7 (Fig. 1D). P7 showed similar affinity, potency, and virus neutralization to P5 and P6 (Fig. 2, A to C, and figs. S2 and S3), but the HA-P7 complex was more stable, with koff lower by a factor of 2 to 25 (0.01 to 0.03 s−1) and a half-life (t1/2) 2 to 12 times as long (27 to 48 s) (Fig. 2C). Further, P7 demonstrated broad group 1 specificity when tested against a panel of HAs from the influenza A (groups 1 and 2) and B viruses (figs. S4 and S5 and table S3).

Structural analysis of cyclic peptide binding to HA

Crystal structures of cyclic peptides (P2 to P7) in complex with H1/PR8 HA were determined at resolutions of 2.28 to 3.10 Å (Figs. 2 to 4, figs. S6 to S9, and tables S4 and S5). All peptides recognize the highly conserved hydrophobic stem groove (Fig. 2, D and E, and fig. S6). The peptides contact HA1 residues 18, 38 to 42, and 318, the HA2 A-helix (residues 38 to 56), and a turn encompassing HA2 residues 19 to 21 (Fig. 2D). These residues are similar to the epitopes of the stem-targeting bnAbs (CR9114, FI6v3, CR6261, and F10) (Fig. 3, A and B).

Fig. 4 Structural elucidation of the group 1 HA specificity of the peptides.

(A) Superimposition of H1 HA bound to FI6v3 HCDR3 (PDB ID, 3ZTN; magenta) with H1/PR8 HA bound to peptide P6 (yellow). Only P6 is shown from the P6-H1/PR8 complex. (B) Superimposition of H3 HA bound to FI6v3 HCDR3 (PDB ID, 3ZTJ; magenta) with H1/PR8 HA bound to P6 (yellow). (C to E) Depiction of potential steric clashes between Phe6 from P6 and Trp21 from group 2 HA. (C) Zoomed in view of Phe100D-Phe6 interactions with Trp21 from group 1 H1 HA. (D) Difference in orientation of Trp21 between group 1 and group 2 HAs. (E) Displacement of Phe100D to avoid a steric clash with Trp21 from group 2 HA and the predicted clash of Phe6 with Trp21 from group 2 HA. (F) Potential steric clash between XA1, Leu3, and Tyr5 from P6 and the glycosylated Asn38 (green) in group 2 HAs. (G) Potential steric clash between ornithine (Orn2) from P6 and Asn49 (light blue) in group 2 HAs, from an overlay of H1/PR8-bound P6 with H3 HA.

Fig. 3 bnAb-guided affinity maturation of the peptides.

(A and B) The stem epitope of bnAbs on H1 HA from H1/PR8 is shown in surface representation (HA1, dark gray; HA2, light gray) (see Fig. 1 for overall location). Antibodies in complex with HA [CR9114 Fab (PDB ID, 4FQI), green; CR6261 Fab (PDB ID, 3GBN), magenta; FI6v3 Fab (PDB ID, 3ZTN), yellow; and F10 Fab (PDB ID, 3FKU), cyan] were superimposed on H1/PR8 HA. The side chains of hydrophobic residues from HCDR1, -2, and -3 and FR3 of each antibody that occupy pockets in the stem hydrophobic groove are shown in ball-and-stick representation (CR9114, green; CR6261, magenta; FI6v3, yellow; F10, cyan). (C) Sequential changes introduced into the peptides during the process of affinity maturation are highlighted with blue and red surfaces on the crystal structures of peptides P2 to P7 in complex with H1/PR8 HA. (D to G) Interactions of peptides with H1/PR8 HA introduced by the sequential changes shown in (C). (D) In P3, nonproteinogenic amino acid XA1 occupies a hydrophobic cavity in the stem epitope. The distance between the centroid of the phenyl ring and the surrounding hydrophobic residues is given in angstroms. (E) Water-mediated H-bond interaction (distance in angstroms) of the backbone amide of Leu3 from P5 with Thr49 in the HA2 A-helix. (F) σ hole–π interactions between the nonproteinogenic amino acid XC6 from P5 with His18 and Trp21 from HA1 and HA2, respectively. The distance between the centroid of the aromatic ring and the chlorine atoms of XC6 is reported in angstroms. (G) Intramolecular H bonds (distance in angstroms) between Glu4 and XE11 in P6.

The nonpolar contacts made by the hydrophobic residues in the peptide macrocycles can be grouped into two distinct regions on the HA stem (Fig. 2D). The first region, consisting of Val40, Leu42, and Thr318 in HA1 and Ile48, Val52, Val55, and Ile56 on the HA2 A-helix (HA2 residues are distinguished by italics throughout), is contacted by peptide residues Arg1 or XA1, Leu3 or XB3, and Tyr5 (Fig. 2, D and E; peptide residue abbreviations are in Fig. 1D). Phe6 or XC6, Trp8, and Leu9 contact the second region consisting of His18, His38, Trp21, Gln38, Thr41, Gln42, and Ile45 (Fig. 2, D and E). Polar interactions include 11 hydrogen bonds; six are direct and five are mediated through water molecules (Fig. 2F). Four hydrogen bonds are involved in recognition of the A-helix at Asn53 and Gln42 by the peptide macrocycle backbone carbonyls of Arg1 (or XA1) and Ser10 (Fig. 2F). The Tyr5 and Trp8 side chains hydrogen-bond with Thr318 and the Asp19 main-chain carbonyl (Fig. 2F). The cyclic peptides bury ~544 to 593 Å2 on the HA surface, which is fairly comparable to the surface area buried by the anti-stem Fabs in their HA complexes (~630 to 680 Å2) (10, 19, 22).

Next, we elucidated the structural basis for the improvement of the optimized peptides relative to P2. The rationale behind substituting Arg1 of P2 with XA1 in P3 was to mimic interactions made by FR3 Ile73 and Phe74 in bnAb CR9114 (Figs. 1 and 3, A and B). Indeed, the major improvement in the affinity and potency of P3 can be explained by a gain of hydrophobic contacts of XA1 with a small hydrophobic cavity formed by Val40, Leu42, Val52, and Ile56 (Fig. 3, C and D). N-methylation (Leu3XB3) rigidifies the peptide backbone in P4 and also mediates entropy gain by displacing a conserved water molecule bound to the A-helix (Fig. 3, C and E), whereas incorporation of dichloro-phenylalanine (Phe6XC6) in P5 results in σ hole–π interactions with His18 and Trp21, respectively (Fig. 3, C and F). Such interactions in halogen-aromatic systems can contribute binding energies of up to ~2.0 kcal/mol (25). In P6, XE11 rigidifies the peptide macrocycle by forming an intramolecular hydrogen bond with Glu4 (Fig. 3, C and G).

Group 1 HA binding specificity

Despite the presence of interacting residues from pan-influenza antibodies FI6v3 and CR9114, and despite targeting the same epitope on the surface of HA, the peptides neutralized group 1 and not group 2 or influenza B viruses (Fig. 2B). This activity can be rationalized by comparing the structures of apo and bnAb-complexed HAs from groups 1 and 2 with the peptide-HA complexes determined in this study. The key differences between the peptide epitope on H1 HAs and the corresponding region on H3 HAs are the orientation of HA2 Trp21, a glycosylation site at HA1 Asn38 in group 2 HAs, and Asn49 in the A-helix of group 2 H3 HAs (which is Thr49 in group 1 H1 HAs) (26). A conserved Phe (from different locations) in group-specific and more broadly neutralizing antibodies makes edge-to-face aromatic interactions with Trp21. However, only FI6v3 and CR9114 appear to demonstrate sufficient flexibility in their key interacting CDR loops—containing Phe100D (HCDR3) and Phe54 (HCDR2), respectively—to accommodate the different orientations of Trp21 in the two HA groups (10, 22).

Comparison of peptide P6 and FI6v3 bound to group 1 H1 HA shows a perfect overlay of Phe6 and Phe100D (Fig. 4, A and C), whereas when FI6v3 is bound to group 2 H3 HA, Phe100D is displaced outwardly by ~1 Å relative to Phe6 (Fig. 4, B and E). Consequently, although FI6v3 can interact favorably with Trp21 of H3 HA, P6 and the other peptides evidently are more constrained and, hence, cannot avoid steric clashes with H3 HA Trp21 (Fig. 4D). Similarly, steric clashes between the corresponding Phe of group 1–specific antibodies (CR6261 and F10) and Trp21 of group 2 HA probably contribute to their group 1 specificity.

The second difference pertains to a glycosylation site at HA1 Asn38 in group 2 HAs. In unliganded (apo) group 2 HAs, the Asn38 glycan projects toward the HA2 A-helix of the same HA subunit so that it would overlap with the expected footprint of the stem-targeting antibodies and the peptides. Notwithstanding this, when bnAbs FI6v3 and CR9114 bind group 2 HAs, they are able to reorient the glycan to insert their heavy-chain CDRs into the corresponding hydrophobic groove and acquire cross-group reactivity. The peptides do not appear to be capable of reorienting the glycan and, therefore, would experience steric clashes of XA1, Leu3, and Tyr5 with the Asn38 glycan on group 2 HAs (Fig. 4F).

The third key difference arises at position 49 in the A-helix, where the bulkier Asn49 in H3 HAs would clash with the ornithine (Orn2) introduced into the peptides for cyclization (Fig. 4G). Overall, the crystal structures indicate that the cumulative effects of these differences account for why the peptides do not interact with group 2 HAs.

Mechanism of viral fusion inhibition

Viral fusion with the host cell membrane is mediated by a large conformational change in HA that is triggered by the low pH of the endosome. Stem-targeting bnAbs, such as CR9114, FI6v3, and CR6261, inhibit viral fusion by stabilizing the trimer even at low pH (10, 19, 22). Because the peptides mimic the binding modes of these antibodies, we expected that they would have a similar mechanism of action. The first piece of evidence came from crystallization of the P3, P4, P5, and P7 peptide–HA complexes under low-pH conditions (pH 4.0 to 5.0), well below the pH of membrane fusion (table S4) (7). The H1/PR8 HA structure in complex with various peptides is essentially identical to that of the prefusion apo HA (fig. S9). Therefore, the peptides appear to stabilize the prefusion state of HA and prevent the pH-dependent conformational rearrangements that lead to membrane fusion (Fig. 5A).

Fig. 5 Mechanism of HA conformational change inhibition by the peptides.

(A) Cartoon representation of the mechanism of conformational change inhibition by peptides. HA trimeric is depicted as HA1, green; HA2, gray; fusion peptide, yellow; and cyclic peptide, red and blue. (B) The trypsin susceptibility assay establishes that peptide P7 inhibits the low pH–induced conformational changes in H1/PR8 HA. Exposure to low pH renders the H1/PR8 HA sensitive to trypsin digestion (lanes 7 versus 8), but P7 prevents its conversion to a trypsin-susceptible conformation (lanes 11 versus 12). The mechanism is similar to that of fusion inhibiting CR9114 Fab (lanes 9 versus 10). (C to E) Correlation of IC50 with SPR-derived constants KD and t1/2, and of EC50 with t1/2, were derived from conformational change inhibition (CCI) and virus neutralization assays. R2, coefficient of determination.

To directly assess inhibition of the HA rearrangements that are associated with fusion activation by the peptides, we performed conformational change inhibition (CCI) and trypsin susceptibility (TS) assays (19, 27) (Fig. 5, B to D; figs. S10 and S11; and table S6). All of the cyclic peptides prevented the conformational change at low pH in the CCI assay (table S6), and the CCI-derived IC50 and virus neutralization assay–derived median effective concentration (EC50) values correlated well with kinetic parameters derived from SPR (Fig. 5, C to E, and supplementary text S3). The TS assay with representative peptide P7 corroborated the CCI assay results. Therefore, the reported peptides are not only strong HA binders, but also effective inhibitors of the low-pH conformational change. These data imply that, like the bnAbs, the cyclic peptides function by stabilizing the HA trimer throughout endosomal entry and subsequent trafficking to late endosomes (28).

In vitro stability and in vivo pharmacokinetic profiling of a peptide

To investigate the translational potential of the cyclic peptides identified in this study, we performed in vitro stability and in vivo pharmacokinetic analyses (Fig. 6). In vitro stability of peptide P7 was assessed in human and mouse plasma. P7 (~2 μg/ml) was incubated with the plasma at 37°C and measured by LC-MS/MS. P7 did not show degradation over the entire time course (~4 hours; Fig. 6A), implying that cyclization and incorporation of NPAAs endow the peptide with resistance to proteolytic cleavage. In vivo pharmacokinetic profiling was then used to assess the stability and clearance of P7 after intravenous administration in BALB/c mice. P7 has a t1/2 of ~2.7 hours and was cleared from plasma in ~24 hours (Fig. 6B). Furthermore, the peptides did not show any cytotoxic effects in the human lung–derived Calu-3 cell line (fig. S12). These translational data demonstrate that the peptidic fusion inhibitors discovered in this work are highly promising for the development of influenza therapeutics.

Fig. 6 Plasma stability and in vivo pharmacokinetic profile of peptide P7.

(A) Stability of P7 in BALB/c mouse (blue circles) and human (red triangles) plasma at 37°C. (B) Plasma concentration of P7 after intravenous injection at 2.4 mg per kilogram of body weight in male BALB/c mice (n = 3 per group). Peptide concentrations from the plasma samples were analyzed by LC-MS/MS, and individual plasma concentration–time profiles were subjected to noncompartmental pharmacokinetic analysis.

Implications for therapeutics against influenza virus

Here we report on the development of effective peptidic inhibitors of influenza virus that neutralize by inhibiting the HA conformational rearrangements at low pH. Extensive data from virus neutralization assays, AlphaLISA assays, SPR, TS assays, CCI assays, x-ray crystallography, and in vitro and in vivo stability studies, as well as their lack of cytotoxicity, provide validation that these peptide fusion inhibitors have potential to translate into the clinic. The peptides were designed on the basis of how the antigen-binding loops of pan–influenza A (and B) antibodies interact with diverse HAs. Constraining the peptide macrocycle by cyclization and addition of nonproteinogenic amino acids led to improvement in the affinity and potency of peptides P4, P5, P6, and P7. Structural characterization illustrated that the peptides recognize the highly conserved HA stem epitope and explained their specificity for influenza A group 1 viruses. Thus, the peptides developed in this work should accelerate the field of small-molecule therapeutics against influenza HA (27). This approach could be further applied to other sites of vulnerability on the HA and to other pathogenic viruses such as HIV-1, Ebola, and Zika.

Supplementary Materials

www.sciencemag.org/content/358/6362/496/suppl/DC1

Materials and Methods

Supplementary Text

Figs. S1 to S12

Tables S1 to S6

References (2939)

References and Notes

Acknowledgments: We thank P. Raboisson for useful discussions on peptide design, M. Jongeneelen and A. van Eijgen for execution of AlphaLISA and virus neutralization assays, S. Blokland for production and site-specific modification of hemagglutinin and setup and execution of conformational change inhibition assays, M. Seijsener-Peeters for scientific guidance of assay execution and data analysis, F. Jahouh for designing the analytical methodology for peptide stability studies, H. Tien for help in setting up automated crystallization screens, W. Koudstaal and J. P. Verenini for help with manuscript formatting, and R. Stanfield and X. Dai for discussions on refinement. R.V. and M.J.P.v.D. own stock in Johnson & Johnson. This work is supported in part by NIH NIAID (National Institute of Allergy and Infectious Diseases) grants R56 AI117675 and R56 AI127371 (to I.A.W.). R.U.K. is grateful to the Swiss National Science Foundation for an Early Postdoc.Mobility fellowship. X-ray data sets were collected at the Advanced Photon Source, Argonne National Laboratory (beamline 23 ID-D); the Advanced Light Source (beamline 5.0.3); and the Stanford Synchrotron Radiation Lightsource (beamline 12-2). GM/CA CAT is funded in whole or in part with federal funds from the National Cancer Institute (Y1-CO-1020) and the National Institute of General Medical Sciences (NIGMS; Y1-GM-1104). Use of the Advanced Photon Source was supported by the U.S. Department of Energy (DOE), Office of Science, Office of Basic Energy Sciences (BES), under contract no. DE-AC02-06CH11357. The Advanced Light Source is supported by the Director, BES, Office of Science of the DOE under contract no. DE-AC02-05CH11231. Use of the Stanford Synchrotron Radiation Lightsource, SLAC National Accelerator Laboratory, is supported by the DOE, Office of Science, BES, under contract no. DE-AC02-76SF00515. The SSRL Structural Molecular Biology Program is supported by the DOE Office of Biological and Environmental Research and the National Institutes of Health (NIH), NIGMS (including P41 GM103393). The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of NIGMS, NIAID, or NIH. All data and code to understand and assess the conclusions of this research are available in the main text, the supplementary materials, and the Protein Data Bank (accession codes 5W6U, 5W6I, 5W5U, 5W5S, 5W6R, and 5W6T). This is manuscript 29451 from The Scripps Research Institute. A patent application related to this work has been filed (application number WO2016EP60438, publication number WO2016180826). Sharing of materials described in this patent application will be subject to standard material transfer agreements.
View Abstract

Navigate This Article