1

A monomeric uncleaved RSV F antigen retains pre-fusion-specific 1

neutralizing epitopes 2

3

Kurt A. Swanson,a,* Kara Balabanis,a Yuhong Xie,a Yukti Aggarwal,a Concepción Palomo,b 4

Vicente Mas,b Claire Metrick,a,§ Hui Yang,a,& Christine A. Shaw,a José A. Melero,b Philip R. 5

Dormitzer,a Andrea Carfia,# 6

7 a Novartis Vaccines, 45 Sidney Street, Cambridge, MA 02139, USA 8

b Centro Nacional de Microbiología and CIBER de Enfermedades Respiratorias, Instituto de 9

Salud Carlos III, Majadahonda, 28220 Madrid, Spain 10

11 # Address correspondence to: andrea.carfi@novartis.com 12 13

*Present address: Sanofi, 270 Albany Street, Cambridge, MA 02139-4234, USA 14

§Present address: Department of Molecular Biology and Microbiology and Graduate Program in 15

Molecular Microbiology, Sackler School of Graduate Studies, Tufts University School of 16

Medicine, Boston, MA 02111, USA 17

&Present address: Novartis Institutes for Biomedical Research, 250 Massachusetts Avenue, 18

Cambridge, MA 02139, USA 19

20

Running Head: A monomeric RSV F adopts a pre-fusion-like conformation 21

22

Words Text: 3647 23

Words Abstract: 193 24

25

JVI Accepts, published online ahead of print on 30 July 2014J. Virol. doi:10.1128/JVI.01225-14Copyright © 2014, American Society for Microbiology. All Rights Reserved.

2

Abstract 26

Respiratory syncytial virus (RSV) is the leading cause of severe respiratory disease in 27

infants and a major cause of respiratory illness in the elderly. It remains an unmet vaccine need 28

despite decades of research. Insufficient potency, homogeneity and stability of previous RSV 29

fusion protein (F) subunit vaccine candidates have hampered vaccine development. RSV F and 30

related parainfluenza virus (PIV) F proteins are cleaved by furin during intracellular maturation, 31

producing disulfide-linked F1 and F2 fragments. During cell entry, the cleaved Fs rearrange from 32

pre-fusion trimers to post-fusion trimers. Using RSV F constructs with mutated furin cleavage 33

sites, we isolated an uncleaved RSV F ectodomain that is predominantly monomeric and requires 34

specific cleavage between F1 and F2 for self-association and rearrangement into stable post-35

fusion trimers. The uncleaved RSV F monomer is folded, homogenous and displays at least two 36

key RSV neutralizing epitopes shared between the pre-fusion and post-fusion conformations. 37

Unlike the cleaved trimer, the uncleaved monomer binds the pre-fusion-specific monoclonal 38

antibody D25 and human neutralizing immunoglobulins that do not bind to post-fusion F. These 39

observations suggest that the uncleaved RSV F monomer has a prefusion-like conformation and 40

is a potential pre-fusion subunit vaccine candidate. 41

Importance 42

RSV is the leading cause of severe respiratory disease in infants and a major cause of 43

respiratory illness in the elderly. Development of an RSV vaccine was stymied when a clinical 44

trial using a formalin inactivated RSV virus made disease, following RSV infection, more severe. 45

Recent studies have defined the structures that the RSV F envelope glycoprotein adopts before 46

and after virus entry (pre-fusion and post-fusion conformation, respectively). Key neutralization 47

3

epitopes of pre-fusion and post-fusion RSV F have been identified, and a number of current 48

vaccine development efforts are focused on generating easily produced subunit antigens that 49

retain these epitopes. Here we show that a simple modification in the F ectodomain results in a 50

homogeneous protein that retains critical pre-fusion neutralizing epitopes. These results improve 51

our understanding of RSV F protein folding and structure and can guide further vaccine design 52

efforts.53

4

Introduction 54

RSV is a member of the Paramyxoviridae family of RNA viruses, which also includes 55

human metapneumovirus, measles virus, mumps virus, Newcastle disease virus (NDV), human 56

parainfluenzavirus (PIV) 1-4, and PIV5. RSV is the major cause of bronchiolitis and pneumonia 57

in infants. It is the leading cause of infant hospitalization in developed countries and is 58

responsible for an estimated 200,000 infant deaths in developing countries each year (1, 2). RSV 59

also causes substantial morbidity and mortality among the elderly (3, 4). There is no specific 60

antiviral treatment recommended for RSV infection, and the only currently available 61

prophylactic is a monoclonal antibody, Palivizumab (Synagis™), used to prevent disease in the 62

highest risk infants (5). The cost of Synagis prevents general use, and the need for a vaccine is 63

clear. However, despite decades of research there remains no licensed vaccine for RSV. 64

Development of a vaccine was stymied in the 1960’s when a formalin-inactivated RSV vaccine 65

candidate made subsequent RSV disease more severe (6). Increased structural understanding of 66

key RSV neutralization epitopes has supported a resurgence of interest in developing an RSV 67

subunit-based vaccine. 68

RSV-neutralizing antibodies target the two major RSV surface antigens, the attachment 69

protein (G) and the fusion protein (F) (7). G is variable in sequence, whereas F is highly 70

conserved between strains, making F the more attractive vaccine antigen. RSV F is a type I viral 71

fusion protein, responsible for driving fusion of the viral envelope with host cell membranes 72

during viral entry. Crystal structures of RSV F ectodomain trimers have documented two 73

conformational states – pre-fusion and post-fusion (Fig. 1C and D) (8-11). In the pre-fusion 74

conformation (Fig. 1C), the heptad repeat A (HRA) region is associated with the globular head, 75

and the tip of the fusion peptide is mostly buried in the center of the protein. In the post-fusion 76

5

conformation (Fig. 1D), HRA and the fusion peptide (not present in the published crystal 77

structures (14,18)) have extended from the globular head to attach to the target membrane, and 78

the heptad repeat B (HRB) region has rearranged to associate with the HRA region, forming a 79

stable 6-helix bundle. This rearrangement places the host membrane bound by the fusion peptide 80

and the viral membrane bound by the transmembrane region in close proximity to drive 81

membrane fusion. The commercial product Respigam™ (Medimmune), made by purifying 82

antibodies from human sera with high RSV-neutralizing titers, was shown to include antibodies 83

specific for the pre-fusion F conformation (12). Indeed, depleting Respigam of antibodies that 84

bind G and post-fusion F demonstrated that the pre-fusion-specific F antibodies were 85

predominantly responsible for virus neutralization by the product. The crystal structure of a RSV 86

F ectodomain (Fecto) stabilized by C-terminal addition of a trimerization tag (foldon) bound to the 87

FAb of the pre-fusion specific antibody D25 identified a new antigenic site designated site Ø (9). 88

Site Ø is formed in pre-fusion RSV Fecto by the packing of the HRA region against the rest of the 89

F globular head (a structural feature not shared by post-fusion F; Fig. 1C and D). In addition, a 90

pre-fusion RSV Fecto antigen with a trimerization tag and mutations that stabilized the HRA-91

globular head interactions elicited a higher neutralizing titer than did post-fusion Fecto in mice 92

and non-human primates (11). 93

Fusion proteins are expressed as uncleaved F0 precursors (Fig. 1A and B). Crystal structures 94

of uncleaved PIV3, PIV5 and NDV Fecto revealed that furin cleavage is not required for these 95

proteins to trimerize (13-15). PIV3 Fecto was crystalized in the post-fusion conformation whereas 96

PIV5 Fecto was trapped in the pre-fusion conformation by the addition of a C-terminal GCN 97

trimerization tag (14). Unlike PIV Fs, which each contain a single furin cleavage site, RSV F has 98

two sites separated by a 27-amino acid fragment, p27 (Fig. 1B). Activation of RSV F for 99

6

membrane fusion requires cleavage by furin at the two sites (16). Our initial attempts to generate 100

a pre-fusion RSV Fecto trimer relied on mutating the furin cleavage sites and adding a GCN 101

trimerization tag, similar to the strategy used to generate the pre-fusion PIV5 Fecto (14). However, 102

addition of the GCN trimerization domain to the uncleaved RSV Fecto greatly reduced the amount 103

of secreted protein. In the course of these manipulations we observed that uncleaved RSV Fecto 104

was predominantly monomeric in solution. We demonstrate that this monomeric species is a 105

folded, soluble protein which harbors key neutralizing epitopes including the site Ø pre-fusion 106

epitope. This uncleaved species might represent a potent pre-fusion Fecto vaccine candidate. 107

Materials and Methods 108

Fusion protein constructs design, expression and purification 109

RSV Fecto constructs with mutations to the furin cleavage sites and fusion peptide region 110

(residues 109-148) were designed (Fig. 1A and B). ∆p27 furinmut Fecto contains a deletion of 23 111

residues from the p27 fragment and point mutations to the furin cleavage sites. These point 112

mutations prevent cleavage by intercellular furin but permit in vitro cleavage by trypsin. ∆FP 113

furinwt Fecto retains the wild-type furin cleavage sites but has a fusion peptide deletion. Unlike 114

∆p27 furinmut Fecto, which is purified as an uncleaved F0 species, ∆FP furinwt Fecto is cleaved by 115

the cell into the F1/F2 species. DNA constructs encoding Fecto with these mutations and a C-116

terminal histidine tag were synthesized (Geneart) and cloned into the pFastBac baculovirus 117

system (Invitrogen). RSV Fecto constructs were expressed in 1 L cultures for three days and 118

purified by nickel affinity followed by size exclusion chromatography (GE Healthcare, Superdex 119

P200). For limited proteolysis studies, Δp27 furinmut RSV Fecto was digested with trypsin (Sigma) 120

7

at a weight ratio of 1:1000 trypsin to Δp27 furinmut RSV Fecto, and the resulting cleaved protein 121

was purified from a size exclusion column void volume. 122

Size exclusion chromatography and multi-angle light scattering (SEC-MALS) 123

Size exclusion high pressure liquid chromatography (HPLC) was performed using a Waters 124

2685 Separations Module HPLC with an isocratic flow rate of 0.75 mL/min coupled to a multi-125

angle light scattering (MALS) system. The mobile phase consisted of 2x PBS. The size exclusion 126

column (Wyatt Technologies WTC-030S5) was used at ambient temperature with a run time of 127

30 min. UV was detected at 280 nm (Waters PDA 2996), and a light scattering detector (Wyatt 128

Technologies miniDawn TREOS) using a laser wavelength of 690.0 nm and a refractive index 129

detector (Wyatt Technologies Optilab rEX) were added in series. A gel filtration standard 130

solution (Bio-Rad) was used to generate a Stokes radius standard curve. Immediately prior to the 131

measurement of the samples, the system was calibrated using a known BSA (bovine serum 132

albumin) solution. 133

A 100 μL injection for a target column load of 100 μg protein (RSV Fecto alone or 1:1 molar 134

ratio RSV Fecto:FAb) was used. Peak retention times were used to estimate the molecular mass of 135

RSV Fecto or RSV Fecto:FAb complexes based on the Stokes radius standard curve. Protein 136

concentrations for identified peaks were calculated using calculated UV extinction coefficients, 137

and the molecular masses for RSV Fecto or RSV Fecto:FAb complexes were determined from 138

multi-angles light scattering data using Astra 5.3.4 software (Wyatt Technologies). 139

Binding studies by surface plasmon resonance (SPR) 140

The affinity of RSV Fecto constructs for Motavizumab or 101F FAbs was measured by SPR 141

with a Biacore T100 instrument. FAbs were directly immobilized on CM5 sensor chips using 142

8

amine coupling at very low levels (75 response units) and RSV Fecto construct preparations at 143

various concentrations were injected at a high flow rate (50 μL/min) to avoid avidity effects and 144

higher than 1:1 binding interaction. The data were processed using Biacore T100 evaluation 145

software and double referenced by subtraction of the blank surface and buffer-only injection 146

before global fitting of the data to a 1:1 binding model. 147

Depletion of Respigam antibodies 148

Antibody depletion was carried out by a method similar to that previously described (17). 149

Briefly, the Δp27 furinmut Fecto and ΔFP furinwt Fecto preparations were bound covalently to 150

CNBr-activated Sepharose 4g (GE Healthcare) following the manufacturer’s instructions (500 151

µg of protein/160 mg Sepharose). The resins were packed into columns that were equilibrated in 152

PBS. Three mg of Respigam were loaded on each of these columns. The unbound material was 153

collected (referred to as Fecto-depleted Respigam). After washing the column with PBS, bound 154

antibodies were eluted with glycine-HCl pH 2.5 and immediately neutralized with saturated Tris, 155

concentrated and buffer exchanged to phosphate-buffered saline (PBS). The eluted antibodies 156

are referred to as Fecto-captured Respigam. 157

Enzyme-linked immunosorbent assay (ELISA) with depleted Respigam antibodies 158

ELISA of depleted antibodies was performed as previously described (12). Briefly, purified 159

101F antibody was used to coat 96 well microtiter plates overnight at 4ºC. Nonspecific antibody 160

binding was blocked with 2% porcine serum in PBS with 0.05% Tween 20. An excess of 161

purified proteins (either Δp27 furinmut Fecto or ΔFP furinwt Fecto, 2 µg/well) was added to each 162

well, and incubation continued for 1 hour at 37ºC followed by addition of serially diluted 163

antibodies and incubation for 1 hour at 37ºC. The plates were extensively washed with water 164

9

after each step. Finally, bound antibodies were detected with peroxidase-labelled rabbit anti-165

human immunoglobulins and o-phenylenediamine (OPD) as substrate (GE Healthcare). The 166

reaction was stopped with 2N sulfuric acid, and absorbance was read at 490 nm. 167

Neutralization assay using depleted Respigam antibodies 168

RSV neutralization by depleted Respigam antibodies was performed using a 169

microneutralization assay described originally by Anderson et al.,1988 and modified by Martínez 170

and Melero, 1998 (18, 19). Briefly, immunoglobulin dilutions were incubated with 1x105 plaque 171

forming units (pfu) of the RSV Long strain for 30 minutes at 37ºC in a total volume of 50 µl. 172

These mixtures were used to infect 1x105 HEp-2 cells growing in 96 well plates with Dulbecco’s 173

modified Eagle’s medium (DMEM) supplemented with 2.5% fetal calf serum (FCS) that had 174

been inactivated 30 minutes at 60ºC. After one hour adsorption, DMEM with 2.5% FCS was 175

added, and the cultures were incubated 72 hours at 37ºC with 5% CO2. The plates were washed 176

three times with 0.05% Tween 20 in PBS and fixed with 80% cold acetone in PBS. After air 177

drying, viral antigen production in the fixed monolayers was measured by ELISA with a pool of 178

anti-G and anti-F murine antibodies and a horseradish peroxidase (HRP)-labelled anti-mouse 179

immunoglobulin preparation (Sigma). 180

Electron microscopy 181

RSV Fecto preparations (30 μg/mL in 25 mM Tris, 300 mM NaCl) were absorbed onto a 400-182

mesh carbon-coated grid (Electron Microscopy Sciences) and stained with 0.75% uranyl formate. 183

A JEOL 1200EX microscope, operated at 80 kV, was used to analyze the samples. Micrographs 184

were taken at 65,000× magnification. 185

10

Results 186

∆P27 furinmut and ∆FP furinwt Fecto oligomerization states by SEC-MALS 187

PIV F ectodomains with furin cleavage site mutations form trimers in the post-fusion F 188

trimer conformation (13-15, 20). To determine if ∆P27 furinmut Fecto similarly formed trimers, we 189

analyzed its oligomerization state with analytical SEC coupled with MALS (Fig. 2). ∆P27 190

furinmut Fecto principally eluted from the analytical column at a retention volume indicating a 191

monomer, as estimated by Stokes radius. In addition, MALS analysis of the principal ∆P27 192

furinmut Fecto peak gave a mass consistent with a monomer. A minor peak in the ∆P27 furinmut 193

Fecto analysis ran consistent with the retention time of a trimer, suggesting some protein in this 194

preparation was competent to for trimers in solution (red arrow in Fig. 2). For comparison, we 195

performed SEC-MALS analysis of ∆FP furinwt Fecto, which is known by its crystal structure and 196

EM analysis to be a post-fusion trimer (8). ∆FP furinwt Fecto eluted at a retention time consistent 197

with a trimer, and analysis of the protein peak by MALS gave a mass consistent with a trimer 198

(Fig. 2). 199

Motavizumab and 101F FAb binding to ΔP27 furinmut Fecto by SPR and SEC 200

We tested whether ∆P27 furinmut Fecto binds FAbs from two structurally-characterized 201

neutralizing antibodies (site II (defined as Site A in (5)) binding Motavizumab and site IV 202

(defined as Site C in (5)) binding 101F) using SPR (Fig. 3). Both ∆P27 furinmut and ∆FP furinwt 203

Fecto bound tightly to Motavizumab and 101F FAbs. The KDs for the ∆P27 furinmut Fecto-204

Motavizumab FAb interaction (3.23 x 10-11 M) and the ∆FP furinwt Fecto-Motavizumab FAb 205

interaction (2.05 x 10-11 M) were similar, as were the KDs for the ∆P27 furinmut Fecto-101F FAb 206

(4.19 x 10-10 M) and ∆FP furinwt Fecto-101F FAb (1.54 x 10-10 M) interactions. 207

11

To determine if FAb binding could stabilize trimerization of ∆P27 furinmut Fecto, we 208

incubated the ∆P27 furinmut Fecto with a molar equivalent of either Motavizumab or 101F FAb 209

and analyzed the resulting complex by SEC-MALS (Fig. 4). The retention times of the principal 210

peaks for the ∆P27 furinmut Fecto-Motavizumab FAb complex and of the ∆P27 furinmut Fecto-101F 211

FAb complex were consistent with 1:1 F monomer:FAb species as determined by Stokes radius 212

(Fig. 4A). MALS analysis also gave masses consistent with a 1:1 F monomer:FAb complexes. 213

For comparison, Stokes radius and MALS analysis for the peaks from ∆FP furinwt Fecto-FAb 214

complexes were consistent with 3:3 F-FAb complexes (or an F trimer bound to three FAbs) (Fig. 215

4B). As previously discussed, a shoulder peak with a retention time consistent with an F trimer 216

was observed in ∆P27 furinmut Fecto preparations and corresponding shoulder peaks (indicated by 217

a red arrow) were observed in the ∆P27 furinmut Fecto-FAb complex preparations with a retention 218

time consistent with F trimers bound by three FAbs. 219

∆P27 furinmut Fecto binding to D25 FAb by SPR and SEC 220

D25 FAb binds the pre-fusion Fecto but does not bind post-fusion Fecto (9). To determine if 221

∆P27 furinmut Fecto is competent to bind the pre-fusion site Ø-specific D25 FAb, we tested 222

binding by SPR and SEC (Fig. 5). The KD for the ∆P27 furinmut Fecto-D25 FAb interaction was 223

2.38 x 10-11 M. In the RSV Fecto -D25 crystal structure, the D25 FAb has principal contacts with 224

a single protomer and modest contacts with an adjacent protomer, suggesting that binding of D25 225

might occur with F trimers rather than monomers (9). To determine if binding to D25 FAb 226

induced trimerization of ∆P27 furinmut Fecto, we incubated the F protein with a molar equivalent 227

of D25 FAb and analyzed the complex by SEC-MALS (Fig. 5B). The retention time and Stokes 228

radius of the D25-bound ∆P27 furinmut Fecto was consistent with a 1:1 F monomer:FAb species. 229

Similarly, MALS analysis of the D25-bound ∆P27 furinmut Fecto gave a mass consistent with a 1:1 230

12

F monomer:FAb species. We conclude that binding to D25 FAb does not induce trimerization of 231

∆P27 furinmut Fecto. 232

Respigam antibodies binding to ∆P27 Furinmut Fecto 233

To determine if ∆P27 furinmut Fecto is competent to bind pre-fusion-specific human 234

antibodies we performed binding experiments using Respigam-depletion assays similar to those 235

published in Magro et al. 2012 (12). ∆P27 furinmut Fecto or ∆FP furinwt Fecto (post-fusion F trimer) 236

was immobilized on resin, and the protein-bound resin was incubated with Respigam. Antibodies 237

that did not bind the F protein were collected in the column flow-though (referred to as ∆P27 238

furinmut Fecto-depleted Respigam or ∆FP furinwt Fecto -depleted Respigam, respectively). 239

Antibodies that bound ∆P27 furinmut Fecto and were eluted are referred to as ∆P27 furinmut Fecto-240

captured Respigam. Untreated Respigam is able to bind both ∆P27 furinmut Fecto and ∆FP furinwt 241

Fecto (Fig. 6A). To demonstrate that antibodies that specifically bind ∆P27 furinmut exist in 242

Respigam, ∆FP furinwt Fecto–depleted Respigam was tested in ELISA for its ability to bind ∆P27 243

furinmut Fecto (Fig. 6B). Although ∆FP furinwt Fecto–depleted Respigam had no significant binding 244

to ∆FP furinwt Fecto (demonstrating sufficient depletion of the sera) ∆FP furinwt Fecto–depleted 245

Respigam retained binding for ∆P27 furinmut Fecto (i.e. monomeric F). For comparison, ∆P27 246

furinmut Fecto–depleted Respigam showed no significant binding to either ∆FP furinwt Fecto or 247

∆P27 furinmut Fecto (data not shown). 248

It has been previously shown that post-fusion F-depleted Respigam retained the majority of 249

its neutralizing titer, and pre-fusion F-depleted Respigam had greatly diminished neutralizing 250

titer (12). Furthermore, the antibodies that bind pre-fusion F represent a greater fraction of 251

Respigam neutralizing titer (12). To test whether the antibodies that bind ∆P27 furinmut Fecto 252

represent the majority of the neutralizing titer of Respigam, we tested the neutralizing titer of 253

13

∆P27 furinmut Fecto-depleted versus ∆P27 furinmut Fecto-captured Respigam (Fig 6C). Untreated 254

Respigam neutralized RSV infectivity well relative to the poorly neutralizing polyclonal 255

antibodies purified from a non-immunized rabbit (negative control). ∆P27 furinmut Fecto-depleted 256

Respigam has a dramatically lower neutralizing titer than Respigam, similar to the observations 257

for pre-fusion F-depleted Respigam (12). ∆P27 furinmut Fecto -captured Respigam had a higher 258

neutralizing titer per mass IgG than untreated Respigam. Taken together, the data indicate that 259

most of the neutralizing activity of Respigam is from antibodies that bind to ∆P27 furinmut Fecto 260

(i.e. monomeric F) but not to the post-fusion ∆FP furinwt Fectoand that this monomeric RSV F 261

ecto-domain retains pre-fusion F specific epitopes. 262

Limited-proteolysis of ∆P27 furinmut Fecto 263

Uncleaved post-fusion PIV trimers (F0) form rosettes, self-associated by aggregation of 264

their fusion peptides, after in vitro cleavage into the F1/F2 species (20). To determine if 265

uncleaved ∆P27 furinmut Fecto was competent to form rosettes of post-fusion trimers upon in vitro 266

cleavage, ∆P27 furinmut Fecto was digested with trypsin (Fig. 7A). Upon treatment with trypsin 267

uncleaved ∆P27 furinmut Fecto was cleaved into the native F1/F2 species of wild-type Fecto, further 268

suggesting that the protein retains a mostly globular fold. Cleaved ∆P27 furinmut Fecto was further 269

purified from the SEC void volume and analyzed by electron microscopy (Fig. 7B). ∆FP furinwt 270

Fecto was previously described as mono-dispersed elongated crutches (8); however, cleaved ∆P27 271

furinmut Fecto forms rosettes of elongated crutches similar to the post-fusion rosettes described for 272

RSV and paramyxoviruses (20, 21). The observations that ∆P27 furinmut Fecto is resistant to 273

degradation by limited proteolysis and competent to forms rosettes of post-fusion F trimers 274

suggest that the uncleaved ∆P27 furinmut Fecto (i.e.. monomeric F) is a well-folded pre-fusion-like 275

molecule structurally competent to rearrange into the post-fusion conformation (Fig. 7C). We 276

14

also attempted to observe the monomeric RSV F by electron microscopy, but we were 277

unsuccessful, which is likely due to the protein’s relatively small size (i.e. 65 KDa). 278

Discussion 279

RSV remains one of the most important, unmet vaccine needs. Renewed interest in 280

developing an RSV F subunit-based vaccine was triggered first by information from the 281

structurally-characterized PIV F proteins and now from structural studies of RSV F itself. Recent 282

data have demonstrated that RSV Fecto stabilized in a pre-fusion conformation elicits higher 283

neutralizing titers than post-fusion RSV Fecto (11, 12), and there is high interest in developing a 284

pre-fusion RSV F subunit vaccine. The novel, pre-fusion-specific site Ø epitope, recognized by 285

the D25 FAb and formed by the HRA region folding against the globular head, appears to be 286

largely responsible for the improved immunogenicity (9). For this reason, antigens presenting the 287

site Ø epitope will likely become components of next generation RSV vaccine candidates. 288

Our initial attempts to generate a stable pre-fusion RSV Fecto antigen focused on the strategy 289

used to stabilize PIV5 F in its metastable, pre-fusion conformation (data not shown) (14). We 290

attempted to trimerize an uncleaved (furin sites mutated) or cleaved (wild type furin sites) RSV 291

Fecto using a C-terminal GCN trimerization domain, but found the protein secretion from the cell 292

was greatly reduced. We similarly tried expressing a cleaved RSV Fecto using a C-terminal foldon 293

tag but this protein formed rosettes similar to a post-fusion Fecto. For this reason, we examined 294

the oligomerization states of RSV F proteins without trimerization tags. SEC-MALS 295

demonstrated that uncleaved ∆P27 furinmut Fecto is predominantly a monomer in solution (Fig. 2). 296

This observed difference in the cleavage-dependence of PIV F and RSV F trimerization may be 297

linked to the presence of a single furin cleavage site in PIV F but two cleavage sites in RSV F, 298

15

flanking a p27 insertion (Fig. 1). The uncleaved, monomeric Fecto (∆P27 furinmut Fecto) has some 299

features consistent with the post-fusion trimer (∆FP furinwt Fecto). Specifically, two neutralizing 300

epitopes present on both pre-fusion and post-fusion Fecto, sites II and IV, are available for binding 301

by their respective FAbs, from Motavizumab and 101F, as demonstrated by SPR (Fig. 3) and 302

SEC-MALS (Fig. 4). A linear peptide harboring the Motavizumab epitope is capable of 303

transiently adopting its native conformation and being stabilized by FAb binding; however, the 304

affinity of the FAb-peptide interaction is ~6,000-fold less than the folded post-fusion protein-305

FAb affinity (10). Thus, the tight binding of uncleaved, monomeric Fecto to the Motavizumab 306

FAb suggests the epitope is pre-formed on the protein surface, as it is on the trimer (Fig. 3). 307

SEC-MALS indicates that, upon binding either the Motavizumab FAb or the 101F FAb, the 308

monomer does not self-assemble into a trimer (Fig. 4). This finding suggests that the constraints 309

preventing trimerization of the uncleaved monomer are not relieved by FAb binding. 310

Further evidence suggests that the RSV Fecto monomer is a folded species. When the RSV Fecto 311

monomer is treated with trypsin, it is cleaved into its biologically relevant F1/F2 species. This 312

limited and specific proteolysis suggests that the RSV Fecto monomer has a globular fold, 313

protecting it from non-specific trypsin digestion. Upon cleavage into F1/F2, RSV Fecto 314

spontaneously self-associates into rosettes of post-fusion trimers (Fig. 7B). This behavior is 315

analogous to the rosette formation observed for the PIV Fs (20). 316

If the RSV Fecto monomer is in a pre-fusion-like conformation, the HRA region would be 317

packed against the globular head, presenting novel HRA epitopes lost in the post-fusion 318

conformation (Fig. 7C). Unlike the post-fusion Fecto trimer, the Fecto monomer is capable of 319

binding the site Ø, pre-fusion-specific, FAb D25 (Fig. 5). Furthermore, the Fecto monomer binds 320

antibodies in the human sera-derived Respigam commercial product that do not bind the post-321

16

fusion Fecto antigen (Fig. 6B). Finally, previous studies have shown that post-fusion Fecto antigens 322

are unable to bind and deplete the majority of neutralizing activity from Respigam (12). 323

However, the Fecto monomer is able to bind and deplete the majority of neutralizing antibodies 324

from Respigam, suggesting that the Fecto monomer retains the important site Ø epitopes present 325

in other pre-fusion antigens and presumably on the RSV F surface on the virion. 326

Retention of site Ø is a desired characteristic of next generation RSV F antigens, and 327

therefore the uncleaved RSV F monomer here described represents a potential pre-fusion subunit 328

vaccine candidate. However, the RSV F monomer described here readily refolds in the post-329

fusion conformation in presence of trace amounts of trypsin or contaminant trypsin-like 330

proteases and, therefore, is not stable enough for evaluation in immunization studies. Future 331

work will be aimed at stabilizing the RSV F monomer by introducing additional mutations that 332

lock jt more stably in a pre-fusion conformation amenable to in vivo studies. 333

334

Acknowledgements 335

Work in Madrid was supported in part by grant SAF2012-31217 from Plan Nacional I+D+I 336

(Ministerio de Economía y Competitividad). 337

17

References 338 339 1. Nair H, Nokes DJ, Gessner BD, Dherani M, Madhi SA, Singleton RJ, O'Brien KL, Roca A, Wright 340

PF, Bruce N, Chandran A, Theodoratou E, Sutanto A, Sedyaningsih ER, Ngama M, Munywoki PK, 341 Kartasasmita C, Simoes EA, Rudan I, Weber MW, Campbell H. 2010. Global burden of acute 342 lower respiratory infections due to respiratory syncytial virus in young children: a systematic 343 review and meta-analysis. Lancet 375:1545-1555. 344

2. Hall CB. 2010. Respiratory syncytial virus, p. 2207-2221. In Mandell GL BJ, Dolin R (ed.), Mandell, 345 Douglas, and Bennett's Principles and Practice of Infectious Disease, vol. 7th edition, Churchill 346 Livingstone, New York. 347

3. Thompson WW, Shay DK, Weintraub E, Brammer L, Cox N, Anderson LJ, Fukuda K. 2003. 348 Mortality associated with influenza and respiratory syncytial virus in the United States. JAMA : 349 the journal of the American Medical Association 289:179-186. 350

4. Han LL, Alexander JP, Anderson LJ. 1999. Respiratory syncytial virus pneumonia among the 351 elderly: an assessment of disease burden. The Journal of infectious diseases 179:25-30. 352

5. Johnson S, Oliver C, Prince GA, Hemming VG, Pfarr DS, Wang SC, Dormitzer M, O'Grady J, 353 Koenig S, Tamura JK, Woods R, Bansal G, Couchenour D, Tsao E, Hall WC, Young JF. 1997. 354 Development of a humanized monoclonal antibody (MEDI-493) with potent in vitro and in vivo 355 activity against respiratory syncytial virus. The Journal of infectious diseases 176:1215-1224. 356

6. Kim HW, Canchola JG, Brandt CD, Pyles G, Chanock RM, Jensen K, Parrott RH. 1969. 357 Respiratory syncytial virus disease in infants despite prior administration of antigenic inactivated 358 vaccine. American journal of epidemiology 89:422-434. 359

7. Anderson R, Huang Y, Langley JM. 2010. Prospects for defined epitope vaccines for respiratory 360 syncytial virus. Future microbiology 5:585-602. 361

8. Swanson KA, Settembre EC, Shaw CA, Dey AK, Rappuoli R, Mandl CW, Dormitzer PR, Carfi A. 362 2011. Structural basis for immunization with postfusion respiratory syncytial virus fusion F 363 glycoprotein (RSV F) to elicit high neutralizing antibody titers. Proceedings of the National 364 Academy of Sciences of the United States of America 108:9619-9624. 365

9. McLellan JS, Chen M, Leung S, Graepel KW, Du X, Yang Y, Zhou T, Baxa U, Yasuda E, Beaumont 366 T, Kumar A, Modjarrad K, Zheng Z, Zhao M, Xia N, Kwong PD, Graham BS. 2013. Structure of 367 RSV Fusion Glycoprotein Trimer Bound to a Prefusion-Specific Neutralizing Antibody. Science. 368

10. McLellan JS, Yang Y, Graham BS, Kwong PD. 2011. Structure of respiratory syncytial virus fusion 369 glycoprotein in the postfusion conformation reveals preservation of neutralizing epitopes. 370 Journal of virology 85:7788-7796. 371

11. McLellan JS, Chen M, Joyce MG, Sastry M, Stewart-Jones GB, Yang Y, Zhang B, Chen L, 372 Srivatsan S, Zheng A, Zhou T, Graepel KW, Kumar A, Moin S, Boyington JC, Chuang GY, Soto C, 373 Baxa U, Bakker AQ, Spits H, Beaumont T, Zheng Z, Xia N, Ko SY, Todd JP, Rao S, Graham BS, 374 Kwong PD. 2013. Structure-based design of a fusion glycoprotein vaccine for respiratory 375 syncytial virus. Science 342:592-598. 376

12. Magro M, Mas V, Chappell K, Vazquez M, Cano O, Luque D, Terron MC, Melero JA, Palomo C. 377 2012. Neutralizing antibodies against the preactive form of respiratory syncytial virus fusion 378 protein offer unique possibilities for clinical intervention. Proceedings of the National Academy 379 of Sciences of the United States of America 109:3089-3094. 380

13. Yin HS, Paterson RG, Wen X, Lamb RA, Jardetzky TS. 2005. Structure of the uncleaved 381 ectodomain of the paramyxovirus (hPIV3) fusion protein. Proceedings of the National Academy 382 of Sciences of the United States of America 102:9288-9293. 383

18

14. Yin HS, Wen X, Paterson RG, Lamb RA, Jardetzky TS. 2006. Structure of the parainfluenza virus 384 5 F protein in its metastable, prefusion conformation. Nature 439:38-44. 385

15. Swanson K, Wen X, Leser GP, Paterson RG, Lamb RA, Jardetzky TS. 2010. Structure of the 386 Newcastle disease virus F protein in the post-fusion conformation. Virology 402:372-379. 387

16. Gonzalez-Reyes L, Ruiz-Arguello MB, Garcia-Barreno B, Calder L, Lopez JA, Albar JP, Skehel JJ, 388 Wiley DC, Melero JA. 2001. Cleavage of the human respiratory syncytial virus fusion protein at 389 two distinct sites is required for activation of membrane fusion. Proceedings of the National 390 Academy of Sciences of the United States of America 98:9859-9864. 391

17. Sastre P, Melero JA, Garcia-Barreno B, Palomo C. 2005. Comparison of affinity chromatography 392 and adsorption to vaccinia virus recombinant infected cells for depletion of antibodies directed 393 against respiratory syncytial virus glycoproteins present in a human immunoglobulin 394 preparation. Journal of medical virology 76:248-255. 395

18. Anderson LJ, Bingham P, Hierholzer JC. 1988. Neutralization of respiratory syncytial virus by 396 individual and mixtures of F and G protein monoclonal antibodies. Journal of virology 62:4232-397 4238. 398

19. Martinez I, Melero JA. 1998. Enhanced neutralization of human respiratory syncytial virus by 399 mixtures of monoclonal antibodies to the attachment (G) glycoprotein. The Journal of general 400 virology 79 ( Pt 9):2215-2220. 401

20. Connolly SA, Leser GP, Yin HS, Jardetzky TS, Lamb RA. 2006. Refolding of a paramyxovirus F 402 protein from prefusion to postfusion conformations observed by liposome binding and electron 403 microscopy. Proceedings of the National Academy of Sciences of the United States of America 404 103:17903-17908. 405

21. Calder LJ, Gonzalez-Reyes L, Garcia-Barreno B, Wharton SA, Skehel JJ, Wiley DC, Melero JA. 406 2000. Electron microscopy of the human respiratory syncytial virus fusion protein and 407 complexes that it forms with monoclonal antibodies. Virology 271:122-131. 408

409

410

19

Figure legends 411

Figure 1. RSV F ectodomain structure. (A) Linear diagram listing residue numbers 412

corresponding to the N terminus of each segment. The furin cleavage sites (arrowheads) divide 413

the protein into F1 and F2. DI-III, domains I-III; p27, excised peptide; FP, fusion peptide; HRA, 414

-B, and –C heptad repeats A, B, and C are indicated. (B) Sequences showing the mutations 415

introduced into p27 and FP region of Fecto constructs. Top, wild-type RSV F sequence with p27 416

and fusion peptide unaltered. Middle, ∆P27 furinmut Fecto sequence with mutations to the furin 417

sites and deletion of p27 residues. Bottom, ∆FP furinwt Fecto sequence with fusion peptide 418

residues deleted. Deleted residues shown as hyphens. (C) Surface representation of the pre-419

fusion RSV Fecto (9). One subunit colored by domains and heptad repeat regions as in A; the 420

other two are white and gray. (D) Surface representation of the post-fusion RSV Fecto (8), colored 421

as in C to highlight the rearrangement of domains and heptad repeat regions after conformational 422

change. Notice the HRA (red) largely buried by the HRB (blue) upon forming the six helix 423

bundle (6HB). 424

Figure 2. Analytical size exclusion chromatography analysis of RSV Fecto constructs. Top, 425

open circles represent the chromatogram of ∆P27 furinmut Fecto in which closed circles represent 426

the chromatogram of ∆FP furinwt Fecto. The principal peak of the cleaved ∆FP furinwt Fecto is 427

approximately 10.5 minutes. The principal peak of the uncleaved ∆P27 furinmut Fecto is 428

approximately 11.2 minutes, with a minor shoulder peak at approximately 10.5 minutes (arrow). 429

Bottom, masses of Fecto constructs measured by SEC-MALS. The molecular masses estimated by 430

the protein sequence for a monomer ∆P27 furinmut Fecto and trimer ∆FP furinwt Fecto are shown in 431

the first column. The masses for the RSV Fecto constructs as measured by Stokes radius and 432

MALS analysis are shown in second and third columns, respectively. 433

20

Figure 3. Binding studies by SPR of Fecto and FAbs. 101F and Motavizumab FAbs were 434

immobilized on a sensor chip, and varying concentrations of Fecto were injected with mass 435

absorption response units (RU) monitored as a function of time (sensorgrams are labeled 436

according to Fecto construct, FAb chip, and concentration as indicated). Fitting the resulting 437

curves to a 1:1 binding stoichiometry resulted in ka and kd and fit values as indicated in the 438

table. 439

Figure 4. Analytical size exclusion chromatography analysis of RSV Fecto constructs bound 440

to FAbs. (A) Black circles represent the chromatogram of ∆P27 furinmut Fecto where blue and red 441

circles represent the chromatogram of 101F FAb-bound and Motavizumab FAb-bound ∆P27 442

furinmut Fecto, respectively. Upon FAb binding the principal peak of ∆P27 furinmut Fecto shifts from 443

approximately 11.2 minutes to 10.7 minutes. An arrow shows a shoulder peak shifting to a 444

retention time of approximately 9.5 minutes. (B) Chromatogram of ∆FP furinwt Fecto with or 445

without FAb binding, colored as in A. Upon FAb binding the principal peak of ∆FP furinwt Fecto 446

shifts from approximately 10.5 minutes to approximately 9.5 minutes. C) Masses of Fecto 447

constructs predicted by the protein sequence for a monomer ∆FP furinwt Fecto and trimer ∆P27 448

furinmut Fecto bound to a single or three FAbs, respectively, are shown in the first column. The 449

masses for the RSV Fecto constructs bound to FAbs as measured by Stokes radius and MALS 450

analysis are shown in columns two and three, respectively. 451

Figure 5. Pre-fusion-specific D25 FAb binding to ∆P27 furinmut Fecto by SPR and SEC-452

MALS. (A) D25 FAb was immobilized on a sensor chip, and varying concentrations of ∆P27 453

furinmut Fecto were injected with mass absorption response units (RU) monitored as a function of 454

time (Fecto concentrations are indicated). Fitting the resulting curves to a 1:1 binding 455

21

stoichiometry resulted in ka and kd and fit values as indicated in the table below. The resulting 456

KD for D25 FAb binding to ∆P27 furinmut Fecto is 2.38 x 10-11 M. (B) Top, SEC chromatogram of 457

∆P27 furinmut Fecto. Bottom, chromatogram of D25 FAb-bound ∆P27 furinmut Fecto. Arrows 458

indicate the expected retention time of RSV Fecto monomer (FM), unbound FAb (FAb), and RSV 459

Fecto monomer bound to FAb (FM+FAb). Below, masses of unbound ∆P27 furinmut Fecto or FAb-460

bound ∆P27 furinmut Fecto predicted by sequence are shown in the first column. The measured 461

masses for ∆P27 furinmut Fecto and D25 FAb-bound ∆P27 furinmut Fecto by Stokes radius and 462

MALS analysis are shown in the second and third columns, respectively. 463

Figure 6. ∆P27 furinmut Fecto is recognized by unique neutralizing antibodies of Respigam. 464

(A) ∆FP furinwt Fecto or ∆P27 furinmut Fecto was adsorbed on an ELISA plate, and different 465

concentrations of Respigam were analyzed for binding. Respigam bound both ∆FP furinwt Fecto 466

(circles) and ∆P27 furinmut Fecto (squares) similarly. (B) ∆FP furinwt Fecto or ∆P27 furinmut Fecto 467

was adsorbed on an ELISA plate and ∆FP furinwt Fecto-depleted Respigam was analyzed at 468

different concentrations for binding. ∆FP furinwt Fecto-depleted-Respigam did not significantly 469

bind ∆FP furinwt Fecto (circles) but retained significant binding to ∆P27 furinmut Fecto (squares). (C) 470

∆P27 furinmut Fecto-depleted and –daptured Respigam were tested for neutralizing titer. Relative 471

to the strongly neutralizing, untreated Respigam (circles), antibodies from ∆P27 furinmut Fecto–472

depleted Respigam (squares) retain a poorer neutralizing titer whereas antibodies from ∆P27 473

furinmut Fecto–captured Respigam (diamonds) retain a higher neutralizing titer per mass IgG 474

(plotted on the x-axis). 475

Figure 7. Limited proteolysis of ∆P27 furinmut Fecto induces protein trimerization and 476

association by the fusion peptide. (A) SDS-PAGE gel showing limited proteolysis of RSV F 477

22

constructs. RSV ∆P27 furinmut Fecto prior to digestion with trypsin is labeled as “Monomer” and 478

produces the F0 band. Two lots of RSV ∆FP furinwt Fecto are labeled as “Trimer” and produce F1 479

and F2 bands in the absence of trypsin. RSV ∆P27 furinmut Fecto produces F1 and F2 bands after 480

digestion with trypsin. (B) Electron microscopy of the RSV Fecto rosettes formed spontaneously 481

by trypsin treatment of ∆P27 furinmut Fecto. The rosettes have the elongated crutch shape of post-482

fusion RSV Fecto and proteins are associated at the end of the stalk where the hydrophobic fusion 483

peptide is located. (C) Hypothetical model of ∆P27 furinmut Fecto as it undergoes conformational 484

rearrangement after cleavage into the F1/F2 species. Top, model of uncleaved ∆P27 furinmut Fecto 485

based on pre-fusion structure with Site A (sites II), Site C (site IV) and site Ø formed on the 486

protein surface. HRB in blue is likely unfolded and represented as a dashed line. A black arrow 487

represents trypsin digestion of the monomer into F1/F2 leading to a conformational change of the 488

pre-fusion-like monomer into the post-fusion trimer. Bottom, a model of RSV Fecto rosettes 489

formed by post-fusion trimers associated by interactions between their fusion peptides (green). 490

Two neutralizing epitopes, Sites A and C, remain on the protein surface while the pre-fusion site 491

Ø epitope is lost as the HRA (red) is largely buried by the HRB (blue). 492

493