DPPE was purchased from Wako Pure Chemical Industries, Ltd. (Japan). Dioleoylphosphatidylcholine (DOPC) and 1-palmitoyl-2-oleoyl-sn-glycelo-3-phosphocholine (POPC) were purchased from Sigma-Aldrich Co., LLC (St. Louis, MO, USA). Ganglioside Neu5Acα2–3Galβ1–4Glcβ1–1′Cer (GM3) from the bovine brain was purchased from Hytest Ltd. (Finland). Influenza hemagglutinins (HAs) derived from A/New Caledonia/20/99 (H1N1) and A/New York/55/2004 (H3N2) viruses were provided by Yujiro Suzuki (The Kitasato Institute, Japan; Matsubara et al., 2010). The human IFV strain A/Puerto Rico/8/34 (H1N1) was provided by Dr. Kyosuke Nagata (University of Tsukuba, Japan; Matsubara et al., 2009).

A sialic acid-mimic pentapeptide with the ability to bind to the receptor-binding site of HA was previously identified (Matsubara et al., 2010). An azide group-conjugated peptide amide, Ala-Arg-Leu-Pro-Arg-Lys(N₃)-NH₂, was prepared by solid-phase peptide synthesis using standard 9-fluorenylmethoxycarbonyl (Fmoc) chemistry (Matsubara et al., 2009). Briefly, Fmoc-Lys(N₃)-OH was loaded onto Fmoc-NH-SAL Resin (Watanabe Chemical Industries, Ltd., Japan), and the peptide was elongated manually in multiple batches on a 0.1-mmol scale. In order to cleave the peptide from the resin, the resin (0.1 mmol) was treated with 1 mL of a cleavage cocktail (trifluoroacetic acid/water/triisopropylsilane, 95:2.5:2.5 by volume) on ice for 2 h (Schneggenburger et al., 2010). The crude peptide was purified by high-performance liquid chromatography (HPLC), and the fraction of the product was lyophilized. Purity (>98%) and the expected structure were verified by HPLC and matrix-assisted laser desorption ionization/time-of-flight mass spectrometry (MALDI-TOF MS). Ala-Arg-Leu-Pro-Arg-Lys(N₃)-NH₂ 1 (16 mg, 21%): MALDI-TOF MS (m/z: calcd exact mass for C₃₂H₆₀N₁₆O₆ [M+H]⁺ 765.5, found 765.8).

In order to obtain an alkyne-modified lipid, DPPE was conjugated with 4-pentynoic acid using 4-(4,6-dimethoxy-1,3,5-triazin-2-yl)-4-methylmorpholinium chloride (DMT-MM) as a coupling reagent (Watanabe et al., 2004). DPPE (0.01 mmol), 4-pentynoic acid (0.03 mmol), and DMT-MM (0.01 mmol) were stirred at 25°C for 12 h in a mixture of chloroform (3 mL)/methanol (0.2 mL)/triethylamine (0.05 mL). The product was extracted with chloroform/sodium hydrogen carbonate solution, and solvents of the organic layer were evaporated. N-(4-pentynoyl)-DPPE 2 (7.2 mg, 90%) was obtained as oil: R_f 0.5 [8:2 (v/v) CHCl₃–MeOH]; MALDI-TOF MS (m/z: calcd exact mass for C₄₂H₇₈NO₉P [M+Na]⁺ 794.5, found 794.4).

Azide-containing peptide 1 (0.2 μmol), pentynoyl-DPPE 2 (0.4 μmol), CuSO₄ (1.6 μmol), sodium ascorbate (1.0 μmol), and tris(benzyltriazolylmethyl)amine (TBTA)(0.2 μmol) were stirred at 25°C for 3 h in 1 mL of 50% methanol (Tornøe et al., 2002). The reaction was stopped on ice and the crude product was purified by HPLC on an ODS column with a liner gradient of 20–60% acetonitrile in 0.1% TFA. After lyophilization, purity (>98%) and the expected structure were verified by HPLC and MALDI-TOF MS. Peptide-conjugated DPPE (pep-PE) 3 (0.19 mg, 63%): MALDI-TOF MS (m/z: calcd exact mass for C₇₄H₁₃₈N₁₇O₁₅P [M+K]⁺ 1575.0, found 1576.2).

In order to investigate the surface topography of the lipid bilayer, lipid bilayers were prepared on mica as described previously (Iijima et al., 2009; Matsubara et al., 2013). Briefly, a POPC lipid monolayer at an air–water interface was prepared on a Langmuir–Blodgett trough at 25°C with a subphase of water, and transferred to freshly cleaved mica (1 × 1 cm) by horizontal deposition at a surface pressure of 35 mN m⁻¹ (POPC-coated mica). A second lipid monolayer of pep-PE (or DPPE, DOPC, GM3, or pep-PE/DOPC) was loaded onto POPC-coated mica by horizontal deposition at a surface pressure of 30 mN m⁻¹ to give a lipid bilayer.

The lipid bilayer was incubated with HA solution for 1 h at 25°C to observe the binding of H1 HA. After washing three times with phosphate-buffered saline (PBS), the lipid monolayer was subjected to AFM measurements.

AFM measurements of lipid bilayers on mica were carried out using SPM-9600 (Shimadzu Co., Japan) in water at 25°C. A 38-μm-long soft cantilever (BL-AC40TS-C2, Olympus) with integrated pyramidal silicon nitride tips having a spring constant of 0.1 Nm⁻¹ was used for all measurements.

A number of topographic images were taken in the dynamic mode at a scanning rate in the range of 1–2 Hz, and the occupied area of target domains was estimated from typical multiple images (n = 3, 1 × 1 μm). In the estimation of domains, AFM images were binarized on the basis of the heights of the membrane, and pixels were counted using image processing software (Adobe Photoshop Elements). For example, the white area of a binarized image by thresholding at 8 nm from the bottom was identified as the area of the HA-bound domain of the pep-PE membrane (Figure 3).

A lipid monolayer of pep-PE/DOPC (50:50) or DOPC was directly loaded onto a few dozen plastic plates (13.5 mm in diameter; code 174950, Nalge Nunc international) by horizontal deposition at a surface pressure of 30 mN m⁻¹. IFV [800–5600 plaque-forming units (pfu)] in 0.2 mL of PBS was incubated at 25°C for 1 h in pep-PE/DOPC- or DOPC-transferred plastic plates. The plates were washed three times with PBS, and their contents were then transferred into a 24-well plates that were blocked with 5% bovine serum albumin (BSA)/PBS at 4°C overnight.

The IFV-bound plates were incubated with a 1:1000 (v/v) dilution of an anti-hemagglutinin (A/H1N1) antibody (RayBiotech Inc.) for 1 h, and the primary antibody was then labeled with a 1:1000 (v/v) dilution of a peroxidase-conjugated anti-mouse IgG antibody (Merck Millipore) for 1 h. Color was developed using o-phenylenediamine, and changes in absorbance (ΔA) at 492 nm were determined by a microplate reader. Each experiment was performed in triplicate.

The pep-PE/DOPC (50:50) lipid monolayer was attached horizontally to plastic plates as described above. IFV (H1N1, A/Puerto Rico/8/34; 1600 pfu in 0.14 mL of PBS) was incubated at 25°C for 1 h on the pep-PE/DOPC-transferred plastic plates. After washing with PBS, plates were incubated with AVL viral lysis buffer (QIAamp Viral RNA Mini Kit, QIAGEN) for 10 min. Viral RNA was extracted according to the manufacturer's instructions.

One-step RT-PCR was performed using QuantiTect SYBR Green PCR Kit (QIAGEN) with forward and reverse primers for the sequences of the matrix protein (M) gene for 244 bp (forward M30F2/08: 5′-ATGAGYCTTYTAACCGAGGTCGAAACG-3′; reverse M264R3/08: 5′-TGGACAAANCGTCTACGCTGCAG-3′; Eisfeld et al., 2014). The reaction was performed using a PikoReal Real-Time PCR system (TCR0096, Thermo Scientific). PCR was set up in a 10-μL reaction volume containing 5 μL of 2 × QuantiTect SYBR Green RT-PCR master mix, 0.1 μL of QuantiTect RT Mix, 0.5 μL of RNA template, 1 μL of forward and reverse primers (10 μM each), and 3.4 μL of RNase free water. The optimized cycling conditions were as follows: RT reaction at 50°C for 2 min, 95°C for 15 min, initial denaturation at 95°C for 30 s, followed by 50 cycles of denaturation at 95°C for 5 s, primer annealing at 57°C for 20 s, and extension at 72°C for 10 s. Fluorescence was measured at the end of each cycle. A melt curve analysis was performed following amplification in order to verify the specificities of the amplified products. A melting curve analysis consisted of 60°C for 30 s, and followed by a temperature increase to 95°C for 10 s with the continuous reading of fluorescence. The amplified products were analyzed by agarose gel electrophoresis (2.5% agarose) and detected by staining with ethidium bromide (Supplementary Figure 3A).

A series of two-fold dilutions of the virus solution starting from 400 to 3200 pfu were prepared in order to construct a standard curve. A liner regression relationship was observed between the amount of the virus and threshold cycle (C_t) values with a coefficient of determination (R²) of 0.954 (Supplementary Figure 3B). The amount of virus that remained on the pep-PE/DOPC (50:50) membrane was estimated from the standard curve.

The sialic acid-mimic pentapeptide (Ala-Arg-Leu-Pro-Arg; Matsubara et al., 2010) was conjugated with DPPE by click chemistry to immobilize the HA-binding peptide on the lipid membrane (Figure 1A). Prior to the click reaction, the pentapeptide was modified with an azide group through the side chain of Lys (Ala-Arg-Leu-Pro-Arg-Lys(N₃)-NH₂, 1; Supplementary Figure 1A). On the other hand, DPPE was modified with 4-pentynoic acid to give N-(4-pentynoyl)-DPPE (2, Supplementary Figure 1B). The click reaction of 1 and 2 quantitatively gave peptide-conjugated DPPE (pep-PE, 3), and pep-PE was purified by HPLC and determined by MALDI-TOF MS (Supplementary Figure 1C).

A lipid monolayer of pep-PE was loaded onto POPC-coated mica to give a lipid bilayer that exposed the pep-PE layer (pep-PE membrane, Figure 1B; Matsubara et al., 2013). The surface topography of the pep-PE membrane was observed by AFM, and revealed that the height of the lipid domain was ~5.5 nm (Figure 2A). On the other hand, the height of the DPPE domain was ~3.5 nm (Figure 2B); therefore, the height of pep-PE was higher than that of DPPE. The difference in height (2.0 nm) was attributed to the size of the pentapeptide.

HA of the H1N1 strain (A/New Caledonia/20/99) was interacted with the pep-PE membrane at 25°C for 1 h to investigate the binding of influenza HA, and the surface topography was observed by AFM. In order to illuminate the HA-bound area, the original AFM image was binarized on the basis of the height of the membrane (8 nm threshold) because the height of the bare pep-PE membrane was not greater than 8 nm (Figures 2A, 3A). The percentage of white pixels in binarized images was regarded as the HA-bound area. The HA-bound area increased in proportion to HA concentrations, and 65% of the membrane surface was covered at 24 nM of HA (Figure 3B). On the other hand, HA showed no significant binding to the DPPE membrane (10% or lower). These results indicated that the binding of HA was responsible for the sialic acid-mimic peptide of pep-PE.

The ganglioside GM3 is an IFV receptor because HA binds to GM3 through the sialyllactose structure (Suzuki et al., 1986; Sato et al., 1996). The binding of HA to the GM3 membrane was investigated using AFM as well as pep-PE. The lipid bilayer that exposed the GM3 layer (the GM3 membrane) was prepared, and its surface topography was then observed by AFM. GM3 was similar in height to DPPE, at ~4 nm (Figure 4A). After the H1 HA incubation, the HA-bound area was estimated by the binarization of AFM images (6 nm threshold; Figure 4B, left). Similar to pep-PE, the HA-bound area increased in proportion to HA concentrations (Figure 4B, right). These results indicated that pep-PE has potential as an alternative to capture HA instead of GM3.

Glycosphingolipids (GSLs) such as gangliosides and neutral GSLs are enriched in the membrane microdomain, e.g., membrane (lipid) raft, of animal cell membranes (Pike, 2006). The membrane microdomain is one of the functional units in the membrane, and is considered to contribute to the many biological activities of lipids and proteins (Simons and Ikonen, 1997). The phase separation of the (glyco) sphingolipid/cholesterol/PC ternary system occurs in the presence of phosphatidylcholine (PC) with unsaturated acyl chain(s) such as POPC (Simons and Ikonen, 1997; de Almeida et al., 2003). Phase separation is detectable by height differences between the three phases; solid-ordered (s_o), liquid-ordered (L_o), and liquid-disordered (L_d) domains, by AFM imaging (Johnston, 2007). We previously investigated the interaction between wheat germ agglutinin (WGA) and a GM3-containing membrane composed of lipids extracted from mouse B16 melanoma cells (Iijima et al., 2009). Since the results of AFM indicated that WGA binds to the highest domain [L_o and/or s₀ phase(s)], this domain was identified as the GM3-enriched area.

Phase separation is responsible for differences in the phase transition temperatures (T_m) of lipids; therefore, we have the ability to design model membranes composed of a low T_m lipid (e.g., unsaturated PC), high T_m lipid (a saturated PC or a sphingolipid), and cholesterol (de Almeida et al., 2003). In the present study, DOPC was mixed with pep-PE to prepare a membrane raft-like domain, and the binding of HA was investigated. A lipid monolayer of a mixture of pep-PE and DOPC (50:50, molar ratio) was loaded onto POPC-coated mica to give a lipid bilayer that exposed the pep-PE/DOPC layer (pep-PE/DOPC membrane). AFM images of the pep-PE/DOPC (50:50) membrane clearly showed phase separation, and the highest domain was estimated to be an area of 50% from a binarized image (5.5 nm threshold; Figure 5A). Since the height of DOPC was ~3.5 nm (Figure 5B), the highest domain in the pep-PE/DOPC (50:50) membrane was considered to be the pep-PE-containing domain.

The HAs of H1 and H3 were interacted with the pep-PE/DOPC (50:50) membrane at 25°C for 1 h, and the HA-bound area was estimated from binarized images as described above (8 nm threshold). In the case of H1 HA, the HA-bound area showed a saturation curve against HA concentrations (Figure 6A). Although the content of pep-PE was reduced to 50 mol%, the HA-bound area of pep-PE/DOPC (50:50) membrane at around 5–20 nM was comparable to that of pep-PE membrane (Figure 3). The HA-bound area (49%) at 24 nM was similar to that of pep-PE domain (50%; Figure 5A), indicating that the pep-PE domain was covered with HA. Furthermore, in the case of the GM3/DOPC (50:50) membrane, the HA-bound area was significantly smaller (22%) than that of the GM3 membrane (48%; Figures 6B, 4B), suggesting the superiority of pep-PE over GM3 for HA binding.

On the other hand, the binding of HA to pep-PE/DPPC (50:50) was significantly less than that to pep-PE/DOPC (50:50; Figure 6A). A comparison with the AFM image of pep-PE/DOPC (50:50), as shown in Figure 5A, revealed the absence of the distinct phase separation of pep-PE/DPPC (50:50; Supplementary Figure 2). This was attributed to the T_m value of pep-PE being markedly different from that of DOPC (−3°C; Koster et al., 1994) and closer to that of DPPE (64°C; Ramezani et al., 2009) and DPPC (41.5°C; Parasassi et al., 1991). Hashizume et al. previously reported that an extensive affinity for lectin was induced by the phase separation of lactosylceramide (LacCer) in a LacCer/DOPC membrane (Hashizume et al., 1998). These findings suggest that the affinity of HA for a pep-PE-containing membrane increases with the formation of a raft-like domain in the presence of DOPC.

In order to investigate the interactions of pep-PE with other proteins, H3 HA, the anti-GM3 antibody, WGA, and two types of serum albumin were incubated with pep-PE/DOPC (50:50; Figure 6C). The amount of H3 HA that bound to the pep-PE/DOPC (50:50) membrane was lower than that of H1 HA (Figure 6A). An anti-GM3 antibody and WGA showed a lower amount of binding to the pep-PE/DOPC (50:50) membrane than HAs, and no binding was observed by human serum albumin (HSA) or bovine serum albumin (BSA) was observed (Figure 6C).

These results indicate that HA specifically interacts with the pep-PE/DOPC (50:50) membrane, and the composition of pep-PE/DOPC (50:50) has the potential to detect HA and IFV effectively.

The binding of IFV of the H1N1 strain (A/Puerto Rico/8/34) to the pep-PE/DOPC membrane and DOPC membrane (as control) was evaluated by ELISA. IFV solutions containing 800, 1600, and 5600 pfu were incubated with membranes at 25°C for 1 h, and IFV that bound to the membranes was then detected. The absorbance of the pep-PE/DOPC (50:50) membrane was significantly increased at more than 1600 pfu (Figure 7A).

The amount of IFV that bound to the pep-PE/DOPC (50:50) membrane was quantified by rRT-PCR as described elsewhere (Tsukamoto et al., 2009). The specific amplification of the products (244 bp) from matrix protein (M) gene (Eisfeld et al., 2014) was confirmed by a melting curve analysis and agarose gel electrophoresis (Supplementary Figure 3A). A standard curve was constructed, and a linear regression relationship was observed between the amount of the virus (pfu) and the threshold cycle (C_t) values was observed with a coefficient of determination (R²) of 0.954 and slope of −0.00199 (Supplementary Figure 3B). After the incubation of virus (1600 pfu) with the membrane for 1 h, 570 pfu of the virus was captured on the pep-PE/DOPC (50:50) membrane, and 100 pfu on the DOPC membrane (Figure 7B). These results indicate that IFV selectively binds to pep-PE-containing membranes.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.