1 INTRODUCTION

↓1

The stability of the tertiary structure (3D structure) of a protein is a function of many weak non-covalent interactions. These interactions involve hydrophobic effect, van der Waals interactions, hydrogen bonds, and ionic interactions. The most intriguing question for a structural biologist is which one of these interactions is likely to stabilise the protein. While there is consensus on the contribution of hydrophobic interactions in protein stability, the role of electrostatic interactions remains enigmatic. Viewed from the perspective of electrostatic interactions, the stability of a protein is the result of a delicate balance of both positive and negative charges of the amino acids. In the recent past, researchers have shown that the contribution of salt bridges (ion pairs) to protein stability is highly variable being sometimes favourable and sometimes unfavourable. Generally, salt bridges are dependent on pH and ion concentrations. Further, the pH of the medium also influences many biochemical and biological processes occurring in the system. In addition, changes in pH result in altered conformation of the protein leading to either protein activation or denaturation. One such example of pH dependency for proteins is the endocytic pathway of viral propogation. Some enveloped viruses enter the host cell by a specialised mechanism called receptor mediated endocytosis.

Endocytosis is a specialized process adopted by eukaryotes for the regulation of the entry or exit of the small and large molecules. While small molecules are transported across the permeable plasma membrane, the large molecules such as micro-organisms are endocytosed. In this process, the material to be ingested is progressively enclosed by a small portion of the plasma membrane, which first invaginates and then pinches off to form an endocytic vesicle containing the ingested substance or particle. Viruses on being internalised, fuse with endosomal membrane to release its genomic content into cytoplasm. Membrane fusion occurs when the lipid bilayers of the apposing membranes are within 1.5 nm distance. In this process, water must be displaced from the hydrophilic surface of the membrane – a process that is energetically highly unfavourable. It seems that specialized fusion proteins catalyse all membrane fusion processes in cells. These fusion proteins provide a way to overcome this energy barrier. Thus membrane fusion is a ubiquitous and critical event in biological systems. It is a required step for the infectious cycle of enveloped viruses, zygote formation, endocytosis, exocytosis, intracellular traffic, and many other biological processes. The mechanism of membrane fusion is best understood in the context of enveloped viruses.

The viral fusion mediating glycoproteins from influenza virus (HA), HIV/SIV (gp160), retroviruses (env), paramyxoviruses (F), and filoviruses (GP) are synthesised as trimers and share similar features. Similarity in sequence suggests similar functional activity (Skehel and Wiley, 1998). Moreover, all the known fusion proteins are class I integral membrane glycoproteins, i.e., they have one transmembrane helix, and the majority of their mass resides on the extracellular side of the host cell (Bechor and Ben-Tal, 2001). Class I fusion proteins are composed of three identical monomers, the functional forms of which are generated from a precursor that is cleaved into two subunits. The cleavage is followed by an irreversible conformational change by which they compensate the energy requirements for fusion. All the above-mentioned proteins are synthesised as precursor molecules before being cleaved to become fusion active. The proteolytic cleavage occurs late in the biosynthetic phase and fusion proteins once cleaved will be in a metastable state (section 1.3.1). To mediate fusion, activated proteins undergo a conformational change triggered either by a pH change or interaction with host cell receptors. During the fusion process, usually the proteins tend to refold into a stable conformation with thermodynamically low energy.

↓2

The fusion of influenza virus with target membranes is one of the most extensively studied viral fusion processes. Colman and Lawrence (review 2003) attributed the understanding of influenza virus HA largely to the pioneering work done by Wiley and Skehel. Till recently, hemagglutinin (HA) of influenza virus is the only characterised protein structurally in both pre and post fusion states. Hence all the fusion models were based essentially on the HA protein (discussed later). Though, it is widely accepted that fusion mechanism is a multi-step process, still the models fall short of explaining the fundamental questions on early steps leading to destabilisation of HA protein. Since HA protein resembles the prototype of a fusion protein, understanding the structure and dynamics of HA is essential for designing novel antiviral agents that can potentially inhibit binding or fusogenic activities not only for HA, but also for other viral fusion proteins.

1.1  The influenza A virus

Influenza is an orthomyxovirus, a term coined by Andrews et al., in 1955 to denote the affinity of the virus for mucus in the form of mucopolysaccharides and glycoproteins. Influenza viruses are classified into A, B and C types, based on the antigenic differences of the internal proteins (nucleoprotein and matrix). The virus has a wide host range of birds and mammals. The different strains of these viruses are nomenclatured based on the host of origin, geographic location, strain number, year of isolation and mostly the antigenic classification of hemagglutinin (HA) and neuraminidase (NA) mentioned in the parenthesis, viz., A/Hong Kong/1/68(H3N2). About 15 antigenic subtypes of HA (HA1-HA15) and 9 subtypes of NA (NA1-NA9) have been reported. Though birds are the natural hosts for all the subtypes, only subtypes H1, H2, H3 and N1, N2 have established stable lineages in humans (Nicholson et al., 2003). The type A virus has been isolated from all the hosts, but so far types B and C were reported only in humans. The most studied strains were mostly from type A, associated with pandemic influenza.

The epidemiological behaviour of influenza in human population is related to the two types of antigenic variation of its envelope glycoproteins (HA and NA) namely antigenic drift and antigenic shift (reviewed in Wright and Webster, 2001). Antigenic drift involves point mutations in the surface antigens resulting from an immune selective pressure. The changes in the antigen structure allow the virus to evade the immune system of the host. As a result, new antigenic variants (new subtypes) evolve, but are still related to those circulating during preceding epidemics. On the other hand, the antigenic shift occurs much less frequently but leads to a major antigenic change. This change results from a replacement of the genomic RNA segment encoding surface antigens resulting in emergence of a “new” potentially pandemic, influenza A virus. The new virus would be antigenically distinct from earlier human viruses and could not have arisen from them by mutation.

↓3

The wide host range coupled by a high mutation rate and cross species interactions generally results in the development of new virus strains, which would naturally be the major obstacle in controlling the disease by vaccination.

1.1.1  Structure of influenza virus

Influenza A viruses are pleiomorphic, but those adapted to cell culture occur as spherical particles of 100 nm in diameter. These viruses are enveloped i.e., have a lipid membrane surrounding the nucleocapsid. The viral genome is negative single stranded RNA (reviewed in Lamb and Krug, 2001) and non-infectious. The mRNAs will be transcribed from the virion RNA (vRNA) by the virion associated RNA dependent RNA transcriptase. Thus as a convention mRNA is plus-stranded.

Figure 1. 1.1 : Schematic diagram of structure of the influenza A virus with viral proteins

The viral genome consists of eight separate RNA segments and the segments 1, 2 and 3 code for viral polymerase i.e., PB2, PB1, and PA respectively, segment 4 for hemagglutinin (HA), segment 5 for Nucleoprotein (NP), segment 6 for neuraminidase (NA), segment 7 for matrix protein (M1), and membrane channel proteins (M2), segment 8 for non-structural proteins (NS1 and NS2).

1.1.2  Viral proteins

↓4

The viral genome encodes 10 different viral proteins. These proteins can be separated into three major subviral components:

The viral envelope containing hemagglutinin (HA), neuraminidase (NA), and M2 proteins;

Matrix protein (M1);

↓5

Viral nucleocapsid (viral ribonucleoprotein, vRNP).

Considering A/Puerto Rico/8/34 virus (PR8) as an example, the influenza viral gene segments range from 890 to 2341 nucleotides in length containing approximately 20-45 non-coding nucleotides at the 3’ end and 23-61 at the 5’ end, depending on the segment (Steinhauer and Skehel 2002). Usually the terminal 13 and 12 nucleotides of the 5’ and 3’ ends respectivelyare conserved in all eight vRNA gene segments and form the vRNA promoter. This base pairing for vRNA promoter was found to be critical for efficient virus replication in MDBK cells(Fodor et al., 1998).

Membrane proteins

↓6

The viral membrane constitutes three distinct proteins i.e., spike like hemagglutinin (HA), neuraminidase (NA) and a membrane channel protein (M2).

The 3D structure of HA, and NA have been determined (mentioned below). HA exists as homotrimer and NA as a homotetramer. These proteins are critically involved in viral pathogencity, its virulence, and host and tissue tropism. When examined by negative staining, the virus particles show the presence of “spikes” representing the two glycoproteins namely HA and NA. Overall, HA represents about 25% of viral protein and NA represents about 5%. The M2 ion channel protein however, is present in low quantities, at only a few copies per particle.

Hemagglutinin: HA primarily serves for the attachment of virus to cells and subsequently for the penetration of viral components into the cells. The HA serves as a major antigen and elicits a strong immune response. The antibodies associated with this protein neutralise viral infectivity. In X-31 virus strain, HA is synthesised as a single polypeptide chain of 550 amino acids with a molecular weight of 77 kD. Each monomer of the activated HA harbours two chains – HA1 and HA2 linked by a disulphide bond. Depending on the virus strain, host cell type and growth conditions, the precursor protein HA0 can be proteolytically cleaved into HA1 and HA2. This cleavage was reported to be essential for virus infectivity (Klenk et al., 1994; Zhirnov et al., 2002). The infectivity of poorly infectious virus could be increased by trypsin treatment. The sensitivity of HA to host proteases is determined by the composition of the proteolytic site in the external loop of the HA0 molecule which links HA1 and HA2 (Chen et al., 1998). This loop typically contains either a single Arg/Lys residue (monobasic cleavage site) or several Lys and/or Arg residues, with an R-X-K/R-R motif, forming a multibasic cleavage site. Influenza A viruses with multibasic cleavage sites (only H5, H7 subtypes) could be more virulent and induce systemic infection in hosts than these viruses (all other influenza A viruses) with monobasic cleavage site (Klenk et al., 1994). The crystal structure of the HA ectodomain was determined with a resolution of 3 Å (Wilson et al., 1981). The ectodomain of HA was obtained by bromelain cleavage (BHA). The bromelain cleaves the HA2 chain just beyond the N-terminal end of the trans-membrane sequence. The HA2 portion of HA is essentially hydrophobic and highly conserved in all types of influenza virus strains. It has been implicated that HA2 participates in fusion activity (see below).

↓7

Neuraminidase: The NA is encoded by the sixth gene segment and is the second major surface antigen. The neuraminidase (NA) protein plays a crucialrole late in the infection by removing sialic acid from sialyloligosaccharides, releasing newly assembled virions from the cell surface and preventing the self-aggregation of virus particles. The neuraminidase protein was found to be an identical tetramer, linked by disulphide bonds. The 3-D structure of pronase-isolated NA heads was determined by X-ray crystallography at 2.2Å resolution (Varghese et al., 1991). The NA protein has long been considered a valid target for antiviral therapy (Oxford et al., 1998). The amino acid residues in the active site of the enzyme are highly conserved between different NA subtypes. The inhibitors against NA were shown to have antiviral activities against a broad range of influenza viruses (Varghese et al., 1995). Several potent and selective medically used inhibitors, for instance, oseltamivir carboxylate (Tamiflu, Ro64-0802, and GS4071) and zanamivir (Relenza, GG167) have been discovered through structure-based rational drug design (Jacksonet al., 2000).

M2 protein: It forms a proton-specific transmembrane (TM) ion channel, and is activated at acidic pH (reviewed by Kelly et al., 2003). The M2 protein plays significant role in both early and late stages of virus infection. During early stages, the M2 ion channel functions between the steps of virus penetration and uncoating (reviewed by Lamb et al., 1994).

Matrix protein (M1): The M1 matrix protein is the major structural component inside the virion and forms a shell around the viral ribonucleoproteins (vRNPs). In addition, M1 protein is considered to be the driving force behind the budding of the virus (Gomez-Puertas et al., 2000).

↓8

Viral ribonucleoproteins (vRNP): The vRNPs are comprised of 4 proteins - the nucleoprotein (NP), and the three subunits of the polymerase (proteins PB1, PB2, and PA), and are associated with each of the viral genomic RNA, forming ribonucleoprotein (RNP) complexes.

The last two proteins are encoded by the eighth gene segment and designated as non-structural proteins (NS1 and NS2). The NS1 was found to be responsible for optimal replication of the virus in the host cell. Importantly, the NS1 protein represses the host cell antiviral response by multiple mechanisms (Geiss et al., 2002). The NS2 is also referred to as nuclear export protein (NEP) and was found to be essential for export of progeny vRNPs (Neumannet al., 2000) by acting as an adapter protein between the nuclear export machinery and the M1-vRNP complex. In addition, the C terminus of NS2 has been shown to interact with M1 both in vivo and in vitro (Wardet al., 1995).

1.1.3  Viral propagation

Influenza virus has affinity primarily for cells in the epithelial lining of the respiratory mucosa. The virus needs to infect these cells in order to propagate itself. Virus spreads by aerosol route. As it is typical for enveloped viruses, influenza virus also initiates infection by binding to host cell surface followed by the fusion of viral and cell membranes.

↓9

Infection by influenza virus can be categorised into

A) Early stages

Attachment; binding; and fusion.

↓10

B) Late stages

Primary transcription of vRNA;

Replication of vRNA and secondary transcription;

↓11

Translation of viral mRNAs to produce viral proteins;

Post-translational modification of viral proteins;

Assembly of viral structural components and release of progeny virus.

↓12

The early steps are mediated by the surface glycoprotein hemagglutinin (HA). The proteolytic cleavage of HA0 generates HA1 and HA2 and is mostly mediated extracellularly, possibly by proteases such as those secreted by Clara cells in the lung or by co-infecting bacteria such as Staphylococcus aureus, Haemophilus influenzae , Streptococcus pneumoniae etc. (Whittaker, 2001). The HA mediates binding of virus to cell surface receptors containing sialic acid and subsequently mediates the fusion of viral and host cell membranes. The cell receptor contains sialic acid and the viral acceptor (anti-receptor) is at the distal end of the HA molecule. Binding to sialic acid occurs via a shallow cavity near the membrane-distal tip of the HA glycoprotein. Subsequently, influenza virus enters the cell by receptor-mediated endocytosis.

The low pH of the endosomes triggers a dramatic conformational change in HA protein, exposing the otherwise deeply embedded fusion peptide (mentioned below). Alterations of HA pull the membranes of the virus and of the endosome together. Eventually both the membranes merge, creating a fusion pore through which the viral content is delivered into the cytoplasm. Simultaneously, acidification also activates the M2 proton channel mediating the transfer of protons into virus particles. This is assumed to be a pre-requisite for uncoating of nucleocapsid and further for releasing the RNPs and M1 protein complex into the cytoplasm of host cell. During early viral infection, dissociation of M1 from RNP is required for entry of viral RNP into the cytoplasm of the host cell (Bui et al., 1996). The M1 protein was shown to inhibit viral transcription and might contribute to the shift of viral RNA synthesis towards replication in the late phase of infection (Watanabe et al., 1996). In addition, antiviral compounds like amantadine and rimantadine (Monto, 2003) were shown to inhibit the M2 proton channel activity and thus the replication of the virus.

As influenza virus life cycle does not involve any DNA coding stage, the transcription occurs in the nucleus. Accordingly, after release of the RNPs into the cytoplasm, they migrate into the nucleus, by an active process and are mediated by the cellular importin a/b pathway (O’Neill et al., 1995). Upon RNPs migrating into nucleus, the viral polymerase directs transcription of the negative sense RNA into a positive sense RNA that forms the template for either viral messenger RNAs (vmRNA)or for self-replication. The vmRNAs move into cytoplasm through the nuclear pores. The vmRNA directs synthesis of nucleoproteins, matrix proteins, and transmembrane proteins (HA and NA). The matrix protein and NPs synthesised are redirected to nucleus.

↓13

The production of HA and NA starts in the rough endoplasmic reticulum and progresses through the Golgi apparatus before being released onto the cell surface. As a result, these membrane glycoproteins undergo posttranslational modifications such as disulphide bond formations, glycosylation and protein folding. The oligomer formation occurs in pre-Golgi (Tatu et al., 1997) and the protein undergoes trimming of carbohydrate chains in Golgi.

The vmRNAs coding for internal proteins (other than membrane associated proteins) are synthesised earlier than vmRNA coding for glycoproteins, implying some transcriptional regulation. In the nucleus, the positive sense RNA replicates to create further copies of the viral genome. These new negative sense viral genomic RNAs become associated with nucleoproteins and some matrix proteins that have migrated into the nucleus. The whole complex leaves the nucleus via nuclear pore.

The last step in the viral morphogenesis involves budding and release of the virus into extra-cellular medium. Budding takes place at the apical plasma membrane and is heavily dependent on the presence of lipid microdomains, or “rafts,” which are enriched in cholesterol and sphingomyelin. Both the HA and NA, are transported from the trans-Golgi network (TGN) and associate with lipid microdomains (Suomalainen, 2002). As a result, a complete virion is formed only at the point of budding. The viral components (HA, NA, M2 forming viral envelope; matrix protein M1; and vRNP) upon transportation to assembly site on plasma membrane interact with one another in an orderly manner (Avalos et al., 1997). The viral membrane anchored proteins (HA, NA and M2) co-localize in the membrane by a selective process excluding cell proteins and come in contact with the viral nucleocapsid (the RNPs coated by M1 proteins) at the cytoplasmic leaflet of the plasma membrane. The M1 binds to the cytoplasmic tail and transmembrane protein domain of HA and NA (Ali et al., 2000) on the outer side and vRNP on the inner side at the budding site. Finally, the plasma membrane at the assembly site bends, causing an outward membrane curvature, and pinches off, releasing the enveloped progeny virus particle into the extracellular medium.

↓14

Fig 1.1.3: Replication cycle of an influenza virus (picture along with the legend from Whittaker, 2001)

(a) The virus binds to receptors on the surface of the host cell and (b) is internalised into endosomes. (c) Fusion and uncoating events, which are pH dependent, result in (d) the release of the viral genome (in the form of viral ribonucleoproteins; vRNPs) into the cytoplasm. The vRNPs are then imported into the nucleus for (e) replication. (f) Positive-sense viral messenger RNAs (mRNAs) are exported out of the nucleus into the cytoplasm for (g) protein synthesis. (h) Some of the proteins are imported into the nucleus to assist in viral RNA replication and (i) vRNP assembly, which also occur in the nucleus. (j) Late in infection, the vRNPs form and leave the nucleus, and (k) progeny viruses assemble and (l) bud from the plasma membrane. The sites of action of anti-viral drugs are shown in red, italic text.

1.2 The influenza virus hemagglutinin – its role in fusion mechanism

Fig.1.2: 3D structure (PDB:1QU1) of the ectodomain of the HA monomer at neutral pH

Green coloured portion is the HA1, HA2 is cyan coloured, and the cleavage site is red coloured.

The spike protein HA of influenza virus is the most and best-studied viral membrane glycoprotein. The HA0 is synthesised as a precursor that trimerises in ER and is transported to cell surface through Golgi apparatus. In X-31 strain, the cleavage of HA0 occurs at residue 328, and is absolutely required for infectivity and further for acid-induced fusion mechanism of the virus. Cleavage occurs either at the cell surface or on the released viruses. In H7 subtype, the cleavage occurs in the trans-Golgi network by furin proteases. The structure (Fig 1.2) is divided into two distinct regions: 1) a long fibrous stem region, containing residues from both HA2 and HA1; 2) a globular head region containing residues entirely from HA1. A hinge region connects these two regions. The head is composed of three independently folded globular domains made up of entirely HA1. The sialic acid binding site as well as the major antibody sites – loop, hinge, and tip/interface are located in this region. The interesting feature of the stem domain is the triple stranded coiled coil, formed by a complex of three long α-helices (from each HA2 subunit) stabilised by electrostatic, hydrophobic and other non-covalent interactions (Wilson et al., 1981). The presence of cleaved HA is sufficient for fusion. This significance is confirmed by the expression of HA protein alone in tissue culture. HA expressed in mature cleaved form is capable of promoting fusion when exposed to low pH (White et al., 1982). Fusion studies with reconstituted vesicles containing isolated HA has further confirmed the role of HA in the fusion process (Stegmann et al., 1987). Thus, HA in the absence of other viral proteins can catalyse the fusion reaction, provided it is anchored in one of the fusing membranes (White et al., 1982).

1.2.1  Conformational changes due to cleavage

↓15

The crystal structure of uncleaved HA0 of A/Hong Kong/68 virus was determined using a mutant R329Q, to prevent its cleavage into HA1 and HA2 (Chen et al., 1998). Only 19 residues are positioned differently in the uncleaved as compared to the cleaved HA form. These residues are HA1 323- 328, R329Q and 1-12 of HA2. All of these 19 residues formed a loop at the cleavage site projecting 8 residues (HA1 327 to HA2 5) away from the molecular surface (Skehel and Wiley 2000).

Fig 1.2.1: Three structures of the influenza virus HA.  (from Colman and Lawrence 2003).

a) The extracellular domain of the neutral pH, cleaved HA trimer. The HA1 polypeptide is shown as a pink-shaded surface, and two of the three receptor-binding sites are labelled by 'R'. HA2 is represented by its backbone, which is coloured blue and yellow according to the positions of the helical regions that are shown in the structure in panel c.
b) An expanded view of the cleavage site (white sphere), which was determined from the structure of a mutant, uncleavable form of HA0. The uncleaved loop in HA0 is shown in red, and its post-cleavage conformation is shown in green. The blue and red spheres mark the amino terminus of HA2 (which is the end of the fusion peptide) and the carboxyl terminus of HA1, respectively, and the pink, blue and yellow shading are as in panel a. Two aspartic acid residues in HA2 — Asp109 and Asp112 — contribute to the hydrophilic cavity into which the fusion peptide folds on cleavage.
c) A comparison of the structure of the HA2 polypeptide in neutral pH, cleaved HA (left; taken from part a) and in recombinant, bacterially-expressed HA2 (right). The latter structure corresponds to the low pH form of HA2. In both cases, only one monomer of the trimer is shown for clarity.

The fusion peptide formed by 1-10 residues after cleavage inserts itself into a negatively charged cavity adjacent to the loop. This insertion is probably guided by an electrostatic interaction between the positively charged amino terminus of HA2 and the negatively charged cavity (Skehel and Wiley, 2000). The cavity contains Asp of HA2 109 and112 and His 17 of HA1, which upon cleavage are buried without pairing to other ionisable residues. Two HA1 C-terminal residues, Glu 325 and Arg 321, move out of the cavity on cleavage.

1.2.2  Conformational changes at low pH

↓16

The hydrophobic fusion peptide of HA2, that is originally buried in the interior of HA trimer becomes exposed at a pH of 5.0 – 6.0 depending on the strain of the virus. This conformational change presented a difficulty for crystal structure determination of low pH HA2. The exposure of hydrophobic fusion peptide leads to protein aggregation and thus prevents crystal formation. Nevertheless, the X-ray crystal structure of the HA2 at low pH (TBHA2) was determined (Bullough et al., 1994; Chen et al., 1999) after solubilising the low pH treated BHA with either protease Lys C or trypsin into HA1 monomers. The left over aggregates from such a solubilised BHA were further proteolysed by thermolysin, which removed the fusion peptide region from HA2 domain. The crystal structure of the remaining fragment (TBHA2) containing HA2 residues 38-175 disulphide linked to HA1 residues 1-27 in the low pH conformation was identified.

The X-ray crystal structure of TBHA2 revealed that the residues from 76-105 remain unchanged with respect to BHA at the neutral pH. However, this structure reveals three major changes in BHA at low pH. First, the residues 54 to 76 of HA2 (HA2 54–76) which were unstructured in neutral-pH BHA are helical in TBHA2. Second, residues 106 to 112 of HA2 (HA2 106–112 region) have undergone a helix-to-loop transition at low pH. Third, the helix C-terminal of the new loop has flipped to lie antiparallel to the coiled coil. Chen et al, (1999) from their E. coli expressed TBHA2, reported that the monomer structure is 110 Å long. The overall conformational change brings the fusion peptide in proximity to the target membrane and also bends the molecule in such a way that the fusion peptide and viral membrane anchor are towards the same end. Upon acidification, the conformational change described above is rapid compared to the fusion process (Godley et al., 1992).

1.2.3  Structure of the fusion peptide

The fusion peptide constitutes of the first 20-25 amino acids of the N-terminus of HA2 chain formed by cleavage (Epand, 2003). These residues are essentially hydrophobic. They are highly conserved among different strains of the influenza virus. Even very conservative single point mutations have been shown to effect fusion. The fusion peptide is rich in Gly providing it greater flexibility of the peptide. The fusion peptide is followed by nine polar residues, which though form a part of the low pH crystallised structure, do not form any ordered structure in these crystals. For this reason, it is likely that the fusion peptide constitutes an independently folded domain when it is inserted into target membrane. Although there is convincing evidence that the fusion peptide inserts into the hydrophobic phase of the membrane, it is still unclear whether it solely inserts into the target membrane (Shangguan et al., 1998) or whether it inserts at an early stage into the viral membrane before relocation to the target membrane (Kozlov and Chernomordik 1998). Kozlov and Chernomordik (1998), and Bentz (2000) reason that HAs with the fusion peptide embedded initially into the viral membrane are along the fusion pathway. Though the structure of HA fusion peptide has been observed in random coil, α-helical and β-sheet based on environment, there is strong evidence that it is the helical form which promotes fusion. The mutants inhibiting fusion are less helical and have a tendency to self-associate into β-sheets (Li et al. 2003).

1.2.4  Role of the ectodomains in membrane fusion

↓17

The role of structural changes of the HA ectodomain in membrane fusion was highlighted from polarised Fourier transform infrared (FTIR) studies of full-length HA and some of its fragments (Tatulian and Tamm, 2000; Gray and Tamm, 1998). These experiments provided some information about the possibility of the low pH structure of HA or of its intermediates to fit in between the two membranes that are to be fused. Gray and Tamm (1998) reported that helical coiled coils were tilted by a large angle at pH 5. The tilting might occur before or after insertion of fusion peptide providing a mechanism to pull the target and viral membranes into close proximity. The distance between the two membranes is reduced from 13 nm to approximately 4 nm if the ectodomains tilt by an angle of 70°. Further, Tatulian et al. (1996) reported that tilting was a reversible process in the absence of the target membrane. Gray and Tamm (1997) confirmed that tilting occurred in X-31 HA and that the hinge region was at the base of the HA near the transmembrane (TM) domain. Tamm (2003) hypothesised that the process was more a membrane-driven rather than a protein-driven process. It is attributed to a protein clamp, which could hold the structure at neutral pH and release at low pH. The clamp may be reattached in the absence of target membranes, but removed from interaction with the core protein in the presence of target membranes, if the clamp interacts with these membranes. The possible candidates for this protein clamp could be parts of the HA1 domain shown to be interacting with bilayer model membranes (Bui et al., 1996). A second hinge domain was also identified by the same research group between the fusion peptide and the ectodomain. It is hypothesised that several trimers tilt simultaneously towards each other, pull the membranes that are to be fused into close proximity, and thereby establish a precursor of the fusion pore.

1.3 Models for membrane fusion

The rate of fusion is influenced both by the HA surface density and the target lipid composition (Clague et al., 1991). The number of trimers required for initiating fusion pore formation is not clear and the number varied from three to eight trimers. Daniele et al. (1996) reported the number as either three or four; Blumenthal et al. (1996) suggested six trimers and Bentz et al. (2000) proposed that at least eight trimers have to be at the fusion site, of which only two or three trimers might be needed to enable the fusion competent conformational changes.

All the models proposed for the mechanism of HA mediated fusion were based on the low pH conformational changes of the HA protein, but differ in that the formation of the initial fusion pore to be either lipidic or proteinaceous. Bentz and Mittal (2000) took into account various fusion models and suggested four distinct intermediates, subsequent to close apposition of the membranes and low pH induced conformational changes of HA. These are

↓18

  1. Fusion proceeds from an aggregation of HA, formed either before or after acidification.
  2. The sign of a fusion pore is defined by the first measurable conductivity (2-5 nS) across the membranes. Additional flickering of the pore may follow which eventually may lead to the formation of an irreversible opening of the pore.
  3. Formation of a lipidic channel, monitored by lipid dye transfer between membranes.
  4. The fusion site monitored by the mixing of aqueous contents (e.g., fluorophors) and the stable merging of the two membranes and complete-mixing of aqueous contents.

The mechanisms of fusion mediated by HAwere investigated by monitoring fusion of erythrocytes(RBCs) stained with fluorescent lipids and solutes with cellsexpressing HA (Blumenthal et al., 1996). A delay in lipid redistribution following the low pH treatment implies that fusion could be a multistep process (Clague et al., 1991). The kinetic studies of fusion cascade have revealed a number of intermediates, which include formation of ahemi-fusion diaphragm which allows lipid redistribution, transient fusion pore, anda large pore which allows transfer of solutes (Blumenthal et al., 1996). Hemifusion (Chernomordik et al., 1999) was referred to as a stage of the stalk wherein the contacting monolayers are in contact while the distal (trans) monolayers are apart giving the appearance of an “hour-glass”.

In the proteinaceous model (Lindau and Almers, 1995; Tse et al., 1993) the fusion pore is an oligomeric ring structure formed by aggregates of HA trimers. Thus, the aqueous fusion pore would be initially lined by proteins and subsequently lipids would be directed to the fusion site. On the other hand, the lipid theory (Hernandez et al., 1996; Jahn and Sudhof, 1999) assigns a greater role for lipids in the formation of the fusion intermediate. This method advocates a hemifusion intermediate, which further continues to expand resulting in the formation of a lipid lined fusion pore.

↓19

However the model of stalk-lipidic pore was strongly supported by Chernomordik et al, (1998) showing lipids which do not favour the formation of stalk inhibits fusion while lipids which support stalk formation promote fusion. The concept of hemifusion was supported by glycosylphosphatidylinositol-linked ectodomain of HA (GPI-HA) lacking the transmembrane domain and cytoplasmic tail (Kemble et al., 1994) which showed lipid mixing with similar time course and efficiency as wt-HA, but did not mediate transfer of soluble contents. The hemifusion under these sub optimal conditions could not make progress to complete fusion even after reversal to optimal conditions suggested it to be a branch of the regular HA mediated fusion pathway. Chernomordik et al. (1998) described another fusion intermediate – FIF (frozen intermediate of fusion). The low temperature (4° C) arrested-intermediate, (after low pH conformational change) has not shown either lipid mixing or aqueous pore formation. Furthermore, it was revealed to be a part of the pathway leading to hemifusion.

1.3.1  HA – a metastable conformation

The fusogenic state of HA molecule is more stable than the cleaved, non-fusogenic HA molecule at neutral pH (Carr and Kim 1993). In addition, the ectodomain is kinetically trapped behind an energy barrier because of extensive non-covalent interactions. It has been experimentally deduced (Carr et al., 1997) that the HA molecule at non-fusogenic state was metastable and that the conformational change could be triggered even at neutral pH either by heat or by urea-mediated denaturation leading to membrane fusion activity. Chen et al. (1995) reported that the bacterial expressed ectodomain of HA2 (comprising amino acids 23-185) at neutral pH, folds spontaneously into a fusion-pH-induced conformation. This observation supports the metastable nature of HA.

To elucidate whether the HA conformational transition is associated with exothermic reaction; Differential scanning calorimetry (DSC) is the method of choice (Huang et al., 2003). There is difference in opinion about the energetics of conformational change of HA ectodomain.

↓20

Investigating the energetics, Remeta et al. (2002), and Epand and Epand (2002) found endotherm DSC peaks during the unfolding process of HA ectodomain. In addition, they report that neutral pH; non-fusogenic HA is very stable and is not metastable.

Contradicting the above statement, Huang et al. (2003) opined that the formation of extended coiled coil involves exothermic reaction and the experiments from Remeta et al. (2002) and Epand and Epand, (2002) did not cover the temperature region at which the exothermic peak occurs. Further, it may be envisaged that both the DSC studies may have failed to detect an exothermic peak. Probably, theextended coiled-coil could have formed already priorto DSC measurement as already acknowledged by Epandand Epand (2002).

1.3.2  Opening of the HA1 distal domain – an essential step for the conformational change

Extensive studies have been performed on the acid-triggered conformational changes that are related to the fusogenic activity of influenza HA (Skehel and Wiley, 2000). White and Wilson (1987) using a panel of anti-HA antibodies concluded that acid triggered conformational change of isolated HA occurs in two steps:

↓21

The characterisation of HA of X-31 at 0°C and the HA from the Japan strain of influenza virus at 37°C supported this two stage conformational change (Puri et al., 1990; Stegmann et al., 1987 and 1990). HA after conformational change is found to be extremely susceptible to proteolysis at specific residues. Studies using circular dichroism and monoclonal antibodies indicated that the molecule has not just denatured, but low pH-induced changes involved movement of molecular domains relative to each other (Skehel et al., 1982; Daniels et al., 1985). Furthermore, Wiley and Skehel (1987) suggested that interactions between the membrane distal globular domains containing the receptor binding sites and the epitopes for infectivity-neutralising monoclonal antibodies were perturbed specifically at fusion pH. Studies from electron microscopy, chemical cross-linking (Ruigrok et al., 1986) and size estimates of HA fragment in solution (Bizebard et al., 1995) clearly indicated that the dissociation of HA membrane-distal domains occur at low pH. The locking of the HA1 subunits by intermolecular di-sulphide bonds in the distal part (Thr 212 and Asp 216 residues of HA1) prevented the conformational change of HA and abolished the fusion activity (Godley et al., 1992; Kemble et al., 1992). In a comparable approach, Barbey-Martin et al., 2002 cross-linked HA monomers using an antibody whose epitope comprised residues from two HA monomers (complex consisted of HA monomer and Fab in 2:1 ratio). This complex was resistant to any conformational change even at low pH demonstrating the need for a partial dissociation and reorientation of HA membrane distal domains for fusion activity. Though it is clear from the above experiments that dissociation of globular domains forms the initial step towards successful fusion activity, the mechanism of dissociation and the forces involved remain still unsolved.

1.4 Protonation effects – possible role in dissociation of HA1 domains

An attractive model for the stability of the non-fusogenic state of HA protein is provided by Huang et al. (2002). The model posits that the stability of HA is contributed by the net charges of the HA1 and HA2. At neutral pH, the charges are balanced viz., zero. This stability is disturbed only due to enhanced protonation at low pH (discussed below).

↓22

This model takes into consideration, the crystal structure of HA (X31 strain) at neutral pH (PDB: 1HGF). The theory is based on molecular modelling techniques (Beroza et al., 1991; Ullmann and Knapp 1999). The protonation calculations were performed on HA at neutral pH with respect to dependence on pH and temperature. Based on the calculations, the model proposes that the net charge of HA1 domain is positive and the HA2 domain, in particular the distal part, is locally enriched with negative charges. The HA1 domains by themselves would not be able to form a stable trimer. This indicates that, in the HA ectodomain at neutral pH, the electrostatic force between either three subunits of HA1 or of HA2 is repulsive, however that between HA1 and HA2 domains is attractive. The work also involves comparing the protonation patterns with that of the HA ectodomain stability assessed by proteolysis (proteinase K assay).

Fig 1.4: Surface electrostatic potential of HA1 domain and HA2 domain with GRASP. ( Figure courtesy (Huang et al., 2002 Biophysical Journal 82; 1050–1058))

Electrostatic potential is color-coded using a sliding scale indicated in the lower part of the figure (unit in k B T/e). Red represents negative electrostatic potential; blue represents positive electrostatic potential, and white is neutral.

According to this model, the HA1 domains exposed to solvent show enhanced protonation at low pH and further, the enhanced protonation gives rise to electrostatic repulsion resulting in destabilisation of the head domains of HA1. Indeed the largest contribution to the electrostatic potential within a protein may arise from protonable amino acids that can carry a net charge. The model argues that the solvent components (water, proton, other ions etc.) interact directly with the surface of HA1 domain. The HA2 domain being buried under the HA1 domain is shielded from the interaction with the solvent. Thus the authors surmise that initially, the protonation state of HA1 domain could be influenced by a pH decrease, whereas the protonation of the HA2 subunits only after an initial destabilisation and re-arrangement of the HA1 domains of a trimer. The hypothesis points out that at low pH, the enhanced protonation of HA1 affects the non-covalent interactions (van der Waals and electrostatic forces) between the HA1 subunits, in particular the electrostatic forces. As the three subunits of a trimer are identical and symmetrical (Wilson et al., 1981), the protonation state and respective changes at low pH should be similar for each subunit.

↓23

To describe the destabilisation process, the essential features are summarised in the figure.

At neutral pH, ignoring the local thermal vibrations of the atoms, the HA ectodomain can be viewed as a stationary, force-balanced-structure implying the overall forces between the monomers as zero (all attractive and repulsive contributions cancel each other taking also into account the solvent interactions).

When decreasing the pH, the electrostatic repulsive force between the HA1 monomers becomes enhanced by protonation. As the original attractive forces cannot cancel the increased repulsion, the neutral pH, non-fusion active structure becomes force unbalanced.

↓24

Fig 1.4.1: Protonation model for HA stability (Huang et al., 2002 Biophysical Journal 82; 1050–1058)

a) Top view of the HA1 domain of the stable trimeric HA ectodomain at neutral pH;
b) Enhanced protonation of the solvent-exposed HA1 domain weakens the attraction between three HA1 subunits at low pH;
c) The new stable state of HA1 domain after its partial dissociation at low pH is determined by a balance between attractive and repulsive forces.

Thus the three HA1 monomers will move apart and a new balance between the attractive and repulsive forces depending on the atomic coordinates will determine their new positions i.e., the extent of the dissociation depends on the changes of attractive and repulsive forces with respect to the dissociation state itself. The calculations and related experiments by the authors supported that the protonation of HA1 was enhanced at fusion pH conditions.

This modelling approach provides a basis and explanation to the puzzling observations on conformational changes of HA occurring at higher temperatures and neutral pH as reported by Carr et al. (1997). It may be deduced that protonation becomes enhanced by increasing the temperature and results in a conformational change. Furthermore, a weakened attraction between the HA subunits may facilitate not only a relocation of the HA1 subunits, but also enable the subsequent steps of the conformational change of the HA into a fusogenic state by the spring-loaded mechanism. Thus it is quiet evident from the above observation and by disulfide linking experiments from Godley et al. (1992) and Kemble et al. (1992) that the HA1 domains act as “clamp” enclosing the HA2 domains and shielding them from solvent exposure.

↓25

This attractive model emphasises the role of electrostatic interactions in maintaining the stability of HA protein at neutral pH. Also taking into account that the HA protein crystal structure indicates a high number of salt bridges (discussed below), this model needs an experimental approach by selectively targeting ion-pair interactions specifically in the interface region of HA1 and HA2.

1.5 Role of salt-bridges (ion pairs) in protein conformation

When amino acid side chains of opposite charge are in close proximity, they can form an ion pair (also called a salt bridge). They are also capable of hydrogen bonding and hence, are usually found on the surface of the protein. If they can form a salt bridge, they will usually be buried. The maintenance of the tertiary (3D) structure of the protein depends essentially on the non-covalent forces and more so on the salt bridges, as these are the most important interactions involved in the oligomerization of proteins. An ion pair is defined as a salt bridge if the centroids of the side-chain charged group atoms in the residues lie within 4.0 Å each other and at least one pair of Asp or Glu side chain carbonyl oxygen and side chain nitrogen atoms of Arg, Lys, or His are also within this distance (Kumar et al., 1999). As a salt bridge is basically made of charged residues and that charge is affected by pH, salt bridges can be broken by either low or high pH conditions or by high concentrations of salts like Na+, K+, etc. Since the present work is aimed at the effects of protonation on early conformational changes in HA protein, salt bridges obviously would be the focus of the present study. Salt bridges could be broadly categorised as “simple” or “complex” depending upon the number of residues involved (Musafia et al., 1995). A simple salt bridge involves a non-bonded or hydrogen-bonded ion paired interaction that joins a single pair of charged amino acid residues, where as a complex salt bridge joins more than two residues (such as the triad Asp-Arg-Asp) in single or adjacent protein chains. It was reported that complex salt bridges probably serve a more significant role because of the co-operativity involved (Marqusee and Sauer,1994). In recent past, salt bridges were among the more thoroughly investigated functional groups with respect to homo and hetero protein complexes (mostly dimers), but the research related to higher multimeric states like trimers and tetramers and especially those of viral coat proteins has still not been addressed (Jones and Thorton, 1996). Musafia et al. (1995) worked on the statistical analysis of complex salt bridges from 94 proteins found in PDB database and reported that 60 % of the proteins contained complex salt bridges, more so triad salt bridges (involving 3 charged residues) were abundant. Further, the report quoted that influenza virus hemagglutinin-3 (PDB ID: 3hmg) contained the largest complex salt bridge involving 15 charged amino acids interacting among themselves to connect the three subunits. Indeed, a look into the 3 D structure of HA protein (1hge or 1hgf) shows that the protein has an extensive network of salt bridges involving 97 cationic residues forming salt bridges with an equal number of anionic residues. The focus of the present investigation is on the interface region between HA1 and HA2 chains of monomer, which is strongly engaged in the conformational change (described above). Xu et al. (1997) reports that interface regions are similar to the protein surfaces and more hydrophilic than the protein interiors. In addition, the interface regions tend to form more hydrogen bonds and salt bridges than protein interiors. This argument gains credence as the interface region between HA1 and HA2 containing the sequences Lys 58 till Arg 76 of HA2 region forms nearly seven salt bridges (protein explorer view) within the same monomer and also with the adjacent monomers.


© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 4.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML-Version erstellt am:
14.09.2006