New Insights into the Spring-Loaded Conformational Change of Influenza Virus Hemagglutinin

Hemagglutinin (HA) is a glycoprotein of influenza virus which mediates fusion of the viral and host membranes. HA binds the virus to a target cell, allowing the virus to be endocytosed. The low-pH environment of the endosome then triggers conformational changes in HA, which cause fusion. The viral genome enters the cytoplasm, and infection proceeds. HA is the most extensively studied viral fusion protein and therefore serves as a paradigm for enveloped virus fusion. Following synthesis and trimerization, each member of the HA homotrimer is cleaved to two subunits, HA1, containing the sialic acid binding site, and HA2, containing the N-terminal fusion peptide and the C-terminal transmembrane domain. Cleavage primes HA for transition from the native, metastable state to the final, lowest-energy state upon acidification (8). The structures of HA in the native form and a fragment of the low-pH form have been solved by X-ray crystallography (Fig. ​(Fig.1)1) (6, 9, 31). In the native state, HA2 trimerizes through the formation of a parallel coiled coil. At the N terminus of the native coiled coil (Fig. ​(Fig.1A,1A, yellow), a loop region (HA2 55-75) (Fig. ​(Fig.1A,1A, dark blue; designated B region in reference 6) connects to a second α-helix (Fig. ​(Fig.1A,1A, green), which runs antiparallel to the first. At the N terminus of the second α-helix is the fusion peptide (Fig. ​(Fig.1A,1A, red), which is buried in the center of the native trimer. The three HA1 subunits (Fig. ​(Fig.1A,1A, gray) cover the HA2 subunits, making trimeric contacts and acting as a clamp. In the fragment of low-pH-treated HA, in which the fusion peptide, transmembrane domain, and most of HA1 have been removed, the structure of HA2 is quite different (Fig. ​(Fig.1B).1B). The B loop region (dark blue) has refolded into an α-helix, connecting the two original helices of the HA2 ectodomain into one coiled coil, with the fusion peptide (red) at its extreme N terminus. This conformational change is referred to as the spring-loaded conformational change (7). Dramatic changes also occur at the C-terminal end of the original coiled coil. A region of six amino acids has unfolded to a loop (purple), and the helix C-terminal to the new loop has flipped to lie antiparallel to the coiled coil (orange). The crystal structure of a recombinant, slightly longer form of HA2 (Fig. ​(Fig.1B)1B) shows that the region between the end of this helical hairpin structure and the transmembrane domain forms an extended chain that lies in the groove between helices of the N-terminal parallel coiled coil (Fig. ​(Fig.1B,1B, light blue). Hence, in the final low-pH state (Fig. ​(Fig.1B),1B), the transmembrane domain lies near the fusion peptide (9). FIG. 1. Location of mutants and model of disrupted coiled coil. (A) Structure of the native HA ectodomain (31) (Protein Data Base [PDB] accession no. 2HMG). A detail of the relevant region of HA2 is boxed. (B) Structure of E. coli-produced HA2 (EHA2) (9) (PDB ... The differences in structure between the native and low-pH forms of HA suggest a mechanism for fusion in which the head groups (Fig. ​(Fig.1A,1A, gray) separate and then the spring-loaded conformational change occurs, moving the fusion peptides toward the target membrane. The fusion peptides embed in the target membrane, and the helix-to-loop conformational change then pulls the fusion peptide, and therefore the attached target membrane, toward the transmembrane domain (14, 29; see also the White laboratory website). Similar models have been advanced for other viral and cellular fusion proteins that contain coiled coils. Several sets of data have, however, been used to argue against the importance of the spring-loaded conformational change. First, head group separation was not detected when low pH was applied at low temperatures (0°C), even though fusion could occur under these conditions (26). Furthermore, fusion at higher temperatures was shown to occur before head group separation was detected by electron microscopy (EM) (25). Because it seems apparent that some head group separation is required for complete coiled-coil formation, it was argued that fusion occurs prior to the spring-loaded conformational change. These and other lines of evidence led to alternate models for HA fusion which do not require the spring-loaded conformational change (3, 5, 26) (see Discussion). We previously showed that a mutant containing proline at residue 55 of the B loop was impaired for fusion, and mutant V55P/S71P, with two B-loop mutations, displayed no fusion (21). Although these findings supported the importance of the spring-loaded conformational change, our prior study did not directly address coiled-coil formation and did not analyze critical conformational changes in HA2. Hence, the root cause for the fusion defects was not uncovered. Because of these limitations and because the role of the spring-loaded conformational change remains controversial (see Discussion), we further explored its function. We first substituted prolines for all of the residues in the B loop that are found at “a” and “d” positions in the final coiled coil; we also made two new double proline substitutions in this region. We asked the following three questions. Are any of the proline-substituted mutants impaired for fusion? If so, at what stage does this occur (i.e., critical conformational changes or target membrane interactions)? Do fusion-impaired mutants form complete coiled coils? None of the double mutants caused any fusion. Double mutant F63P/F70P was analyzed in detail. Like wild-type (WT) HA, F63P/F70P underwent key conformational changes, including head group separation and fusion peptide exposure, and bound tightly to target membranes. Several lines of evidence indicated, however, that, instead of forming a complete coiled coil, F63P/F70P formed a coiled coil that was only about one-half the length of that formed by the WT and that was splayed in its N-terminal half. Our results demonstrate that complete coiled-coil formation is not necessary for fusion peptide exposure and membrane binding but is crucial for fusion to progress from target membrane binding to membrane merger.