Ethics statement

This study was carried out in strict accordance with the Dutch Experiments on Animals Act 1997. In accordance with article 18 of the Act the protocol was assessed and approved by the Animal Ethics Committee of the Utrecht University (Permit Number: DEC#2007.III.01.013). All efforts were made to minimize discomfort related to immunizations and blood sampling. The animal welfare officers of the Utrecht University checked the mandatory administration and supervised the procedures and the well-being of the llamas that were used.

Preparation of gp41 constructs

The HIV-1 gp41 constructs ‘gp41CHRTM’, clade B C-terminal Heptad Repeat (CHR), MPER and transmembrane (TM) regions (residues 629–706, HXB2c numbering), ‘gp41-GCN’, gp41 residues 629–683 (HXB2c numbering) fused to a trimeric GCN4 leucine zipper [79], and ‘gp41_INT’, which has a similar design as reported [60] with the exception that the trimeric GCN4 zipper was fused in frame with part of gp41 NHR, were cloned into pETM-11 (EMBL, Heidelberg) or pETM-13 using standard PCR techniques. Proteins were expressed in Escherichia coli strains C41 or BL21. Cells were grown to an OD₆₀₀ of 0.8 and induced with 1 mM isopropyl-ß-D-thiogalactopyranoside (IPTG). Cells were harvested by centrifugation and lysed in either phosphate buffered saline (PBS) supplemented with 1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) or lysed in 20 mM tris(hydroxymethyl)aminomethane hydrogen chloride (Tris-HCl) pH 8.0, 100 mM NaCl, or in 20 mM Bicine pH 9.3, 100 mM NaCl, for ‘gp41CHRTM’, ‘gp41-GCN’ and ‘gp41_INT’ respectively. Purification was performed using a Ni²⁺ affinity or Q-sepharose column (‘gp41_INT’) and gel filtration on a Superdex 200 column (GE Healthcare). Gp41_528–683 was produced as described [29].

Preparation of immunogen

The concentration of gp41CHRTM was adjusted to 1 mg/ml. Lipids POPC, POPE, POPS, Cardiolipin/Sphingomyelin and cholesterol (Avanti Polar Lipids) were dissolved in chloroform and mixed in a 1∶2∶1∶2∶5 ratio (w/w). After solvent evaporation, the lipid film was dried under vacuum. Multilamellar vesicles were obtained by resuspending the lipid film in PBS to a final concentration of 1 mg/ml. After three quick freeze/thaw rounds the liposome suspension was extruded through a 100 nm membrane. Liposomes and gp41CHRTM were mixed in a 1∶1 ratio (w/w). CHAPS was removed by extensive dialysis with PBS containing 0.1 mg/ml BioBeads (BioRad), changing the solution every eight hours for two days. The quality of gp41CHRTM proteoliposomes was confirmed by dynamic light scattering (Malvern Nano ZS).

Immunization and construction of phage libraries

Two llamas (L6 and L7) were immunized intramuscularly with the gp41CHRTM reconstituted in the trimeric form into synthetic liposomes (gp41 proteoliposomes) without adjuvant. The immunization scheme consisted of a priming immunization at day 0 followed by 5 boosts, which were given at days 7, 14, 21, 28 and 35. During the prime and the first boost 0.25 mg of gp41 and 0.5 mg of lipids were used per injection. These amounts were halved for the remaining boosts. Importantly no adjuvants were used. The immune response was measured in the serum taken at day 21 and compared to day 0. At day 43 lymphocytes were collected from the two llamas for construction of two separate phage libraries, as described previously [80]. The genes encoding for the VHH were amplified and cloned in the phage-display plasmid, pAX50. The plasmids were transferred to E. coli strain TG1 [supE hsdΔ5 thi Δ(lac-proAB) F′(traD36 proAB+,lacIq lacZΔM15)].

Selection of gp41-specific VHH

Two rounds of selection with phage libraries were carried out by panning phages onto gp41-proteoliposomes immobilized in microtiter plate wells. In the first round selection, gp41 proteoliposomes at 5, 1 and 0.2 µg were coated in the wells of PolySorp (Nunc) microplate overnight at 4°C followed by PBS containing 4% Marvel dried skimmed milk (Premier International Foods UK Ltd) for one hour at room temperature. approximately 10¹⁰ of the phages, pre-incubated for 30 minutes in 2% skimmed milk in PBS and empty liposomes to reduce nonspecific binding, were then added to the wells for 2 hours followed by extensive washing (15 times) with PBS containing 0.05% Tween 20 (PBS-T)(the 5^th, 10^th and 15^th wash steps were done for 10 min) and three times with PBS. The bound phages were eluted with 100 mM TEA, neutralized with 1 M Tris-HCl pH 7.5, and rescued by infection of the E. coli strain TG1, which were then plated on LB agar plates containing 100 µg/ml ampicillin and 2% glucose. The phages eluted from 1 µg coated gp41 proteoliposomes in the first round yielded the highest enrichment of ∼100 times. These were therefore chosen for further selection in a second round. The selected polyclonal phages were amplified in E. coli strain TG1 after infection with helper phages VCSM13 (Stratagene). After purification from the culture supernatant, phages were used as input for the second round selection. Lower concentrations of proteoliposomes (0.5 and 0.1 µg per well) were coated. ∼10⁹ phages of the amplified phages were incubated in each well. After extensively washing away unbound phages, the bound phages were eluted by 100 µl of 150 µg/ml human monoclonal antibodies of either 2F5 or 4E10. TG1 cells were infected with the eluted phages. The infected TG1 cells were grown on LB agar plates containing 100 µg/ml ampicillin and 2% glucose to obtain individual colonies. The infected TG1 were also spotted on the LB agar plates supplemented with 100 µg/ml ampicillin and 2% glucose for the assessment of the efficiency of the selection. The enrichment of specific phages was estimated in comparison with the phages that were eluted by an irrelevant human monoclonal antibody (Remicade, Schering-Plough) or with the phages eluted from the empty wells. DNA inserts of a number of selected phages binding to gp41 were recloned to plasmid pAX51 using the restriction sites SfiI and BstEII.

Phage ELISA

MaxiSorp (Nunc) plate wells were coated overnight at 4°C, with gp41-GCN in PBS. After blocking of the wells with 4% skimmed milk in PBS, they were incubated with VHH fused to M13 phages in the presence of 2% skimmed milk. The wells were washed with PBS-T, and bound phages were detected by incubation with a mouse monoclonal antibody against M13 phage coupled to horse radish peroxidase (HRP). The amount of HRP was visualized by the addition of O-phenylenediamine (OPD) supplemented with 0.03% H₂O₂. The reactions were stopped by the addition of 1 M H₂SO₄ and the optical density was measured at 490 nm.

Production and purification of gp41-specific VHH

TG1 strains expressing monoclonal VHH were cultivated in 2×YT medium supplemented with 100 µg/ml of ampicillin until the optical density of culture medium reached 0.5 at 600 nm. Then 1 mM IPTG was added to induce VHH production for 5 hours at 37°C. VHH were extracted from the periplasmic fraction and bound to Talon metal-affinity resin (Clontech 635504). After washing away unbound material, the bound VHH were eluted by 300 mM imidazole in PBS. The eluted VHH were dialyzed against PBS with two buffer changes. The VHH were dialyzed twice for 1.5 hours at room temperature, followed by dialysis overnight at 4°C. The VHH were analyzed on 15% acrylamide sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) stained by Coomassie Brilliant Blue and a Western blot using the antibody against the myc-tag (9E10).

Binding of VHH to HIV-1 envelope proteins in ELISA

The binding of the VHH to HIV-1 envelope was performed on the immobilized gp41 and gp140-92UG037 in the microtiter plate (MaxiSorp from Nunc) in ELISA. 200 ng Gp41-GCN or 280 ng gp140-92UG037 protein were coated in the wells of microtiter plates overnight at 4°C followed by incubation with 4% skimmed milk to block non-specific binding. Dilution series of purified VHH were incubated with the coated wells for one hour at room temperature followed by the removal of unbound VHH by washing three times with PBS-T and one time with PBS. Bound VHH were detected with anti-myc tag antibody (9E10) and donkey anti mouse peroxidase (DAMPO). The antibodies were incubated for one hour and washed with PBS-T and PBS between each incubation. The plates were developed using OPD containing 0.03% H₂O₂. The reaction was stopped with 1 M H₂SO₄. The optical density was measured at 490 nm.

Binding of VHH to membrane anchored Env

Titrating amounts of antibodies were added to JR-FL HIV-1 Env-transfected 293T cells expressing either cleaved or uncleaved Env, incubated for 1 h at 37°C and washed with FACS buffer (PBS, 5% heat-inactivated fetal calf serum [HIFBS], 0.02% azide). MAbs b6, b12, and 4E10 were detected with a secondary anti-IgG Ab conjugated to phycoerythin (Jackson ImmunoResearch). MAb b12 and the VHH 2H10, bi-2H10 and bi-2H10 W100A were biotinylated and detected with streptavidin PE. MAb b6 binding is weak on cleaved Env, confirming that most Env is cleaved [81]. Binding was analyzed using flow cytometry, and binding curves were generated by plotting the mean fluorescence intensity of antigen binding as a function of antibody concentration.

Bivalent VHH construction

PCR was used to amplify the VHH sequences. Different primer sets designed to amplify the VHH that will be located at the N terminus and the VHH that will be located at the C terminus of the bivalent VHH. The primers at the 3′ of the N-terminal VHH and at the 5′ of the C-terminal VHH, may encode a flexible sequence glycine-serine (GS) linker represented by a repeat of the pentapeptide “G-G-G-G-S”, also called 5GS. These same primers contain a unique restriction site (BamHI). We constructed bi-2H10 with 5GS, 15GS, and 25GS linkers and with the linker GGGGSGGGGSGGGGGGS, called 17GS. For the studies we used the bi-2H10 with the 17GS linker, unless specified differently. After PCR amplification, the generated fragments were cleaved at a unique N-terminal restriction site (SfiI restriction site) and BamHI for the VHH that will be located at the N terminus, and with BamHI and at a unique C-terminal restriction site (BstEII) for VHH that will be located at the C terminus. The fragments are ligated into pAX51, which was digested with SfiI and BstEII.

Lipid binding assays

Antibodies were mixed with liposomes having the lipid composition as described above in PBS and their buffer was adjusted to 40% sucrose. The 40% sucrose mixture was overlaid stepwise with 30, 20, 10 and 5% sucrose and centrifuged in a Ti55 swinging bucket rotor for 18 h at 50 K rpm. Six fractions were recovered from the gradient and samples thereof were analyzed by SDS-PAGE.

For ELISA analysis, cardiolipin from bovine heart (SIGMA, C0563) and l-α-phosphatidyl-l-serine from bovine brain (SIGMA, P7769) were dissolved in ethanol to a concentration of 5 mg/ml and further diluted with PBS to 50 µg/ml of which 100 µl was coated per well in PolySorp plates (Nunc) overnight at room temperature without a cover foil to allow complete evaporation of the sample. As a control gp140-92UG037 was coated on a MaxiSorp plate (Nunc). Dilution series were prepared of 2H10, bi-2H10, 2F5 and 4E10. 2F5 and 4E10 were detected with the mouse anti-human IgG monoclonal 3E8 (SANTA CRUZ). The rest of the ELISA was performed as described above.

Viruses

HIV-1 IIIB was obtained from the Centralised Facility for AIDS Reagents (CFAR), NIBSC and propagated in H9 cells. Replication-competent NL43 stocks were prepared from a replication-competent HIV-1 molecular clone by transfection of 293T cells. All other used HIV-1 viruses were HIV-1 envelope pseudotyped viruses and were produced in 293T cells by co-transfection with the pSG3Δenv plasmid and the relevant envelope plasmid. The subtype B HIV-1 reference panel of envelope clones and the subtype C clones were obtained through the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH, USA. AC10.0 clone 29 and SC422661 clone 8 were contributed by David Montefiori, Feng Gao and Ming Li; ZM109F.PB4 was contributed by David Montefiori, Feng Gao, S Abdool Karim and G Ramjee; RHPA4259 clone 7 was contributed by B.H. Hahn and J.F. Salazar-Gonzalez. The SHIV clones were constructed at the BPRC by Zahra Fagrouch and Ernst Verschoor. The panel of transmitted subtype B (T/F) clones (WEAU-d15.410.787, 1006-11.C3.1601, 1054-07.TC4.1499, 1056-10.TA11.1826, 1012-11.TC21.3257) were provided by George Shaw and Beatrice Hahn.

Cells

TZM-bl cells were obtained through the NIH AIDS Research and Reference Reagent Program from J. C. Kappes, X. Wu, and Tranzyme, Inc., and cultured in Dulbecco's modified Eagle medium (Invitrogen) containing 10% (v/v) fetal calf serum (FCS).

Neutralization assay

Neutralization was measured using pseudovirus or 200 50% tissue culture infective doses (TCID50) of virus in the TZM-bl cells as targets following the methods of Derdeyn et al. [82] Wei et al. [83] and Li et al. [84], with Bright-Glo luciferase reagent (Promega, Southampton, UK). The neutralization activity of each VHH was assayed in duplicate. No virus inactivation was observed with a negative control VHH. VHH IC50 titers were calculated using the XLFit4 software (ID Business Solutions, Guildford, UK). The HIV-1 neutralization potency of llama 6 & 7 serum/plasma samples were also evaluated in TZM-bl. Neutralization was measured using 200 TCID50 of IIIB virus in the TZM-bl cell-based assay, as described above, using a Glomax plate reader (Promega). Serum samples were heat-inactivated to inactivate complement by incubation at 56°C for 1 h before use in neutralization assays. Threefold serial dilutions of llama sera were then tested, starting at a 1∶5 dilution.

Epitope mapping

The binding of VHH and mAb to peptides was assessed in a Pepscan-based ELISA [85]. Structural aspects of antibody antigen interaction were revealed through small random peptide libraries [86]. Each VHH and mAb was titrated to ensure that optimal binding was achieved and that nonspecific binding was avoided. Each well in the card contained covalently linked peptides that were incubated overnight at 4°C with VHH and mAb, between 1 and 10 ng/ml in PBS containing 5% horse serum (v/v), 5% OVA (w/v), and 1% (v/v) Tween 80, or in an alternative blocking buffer of PBS containing 4% horse serum (v/v), and 1% (v/v) Tween 80. After washing, the plates were incubated with a HRP-linked rabbit anti-Ab (DakoCytomation) for 1 h at 25°C. After further washing, peroxidase activity was assessed using substrate (0.5 g/l 2,2′-azino-di[3-ethyl-benzthiazolinesulfonate(6)]diammonium salt (ABTS)) with 0.006% H₂O₂ in 0.18 M Na₂HPO₄, 0.22 M citric acid was added until the solution had a pH of 4. The color development was quantified after 60 minutes using a charge-coupled device camera and an image-processing system.

Protein expression and purification of 2H10 for crystallization

The cDNA corresponding to 2H10 was synthesized (GeneArt) and cloned into vector pMEK220 (without 6his-tag or myc tag) and 2H10 was expressed in E. coli strain BL21gold (DE3) (Invitrogen). Cells were grown to an OD600 of 0.7 and induced with 1 mM IPTG at 37°C. After 5 hours cells were harvested by centrifugation and lysed by sonication in 0.02 M 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 0.1 M NaCl, pH 7.4 (HEPES buffer). 2H10 was purified by affinity chromatography using a Protein A column (Roche). 2H10 was eluted with 0.01 M glycine, pH 2.5, and fractions containing 2H10 were neutralized with 1 M Tris-HCl, pH 8. These fractions were further purified by gel filtration on a Superdex 200 column (GE Healthcare) in HEPES buffer. Purity of the sample was confirmed on a 12% SDS-Tris-Tricine gel.

Crystallization, data collection and structure determination

2H10 in complex with peptide 903 was subjected to crystallization at a concentration of 12 mg/ml. Crystals were obtained by the vapor diffusion method in hanging drops, with equal volumes of antibody and reservoir solution (0.1 M Bicine, 3.2 M NH₄SO₄, pH 9). The crystal was cryo-cooled at 100 K in reservoir buffer containing 25% (v/v) glycerol. A complete dataset was collected at the ESRF (Grenoble, France) beamline ID14-4. Data were processed and scaled with MOSFLM [87], and SCALA [88]. The crystals belong to space group I23 with unit cell dimensions of a = 89.95 Å, b = 89.95 Å, c = 89.95 Å and α = 90.00°. The structure was solved by molecular replacement using PHASER [89] and the VHH structure from Protein Data Bank (PDB) ID 3EZJ as a search model. An initial model was built with ARP/wARP [90] and completed by several cycles of manual model building with Coot [91] and refinement with refmac [92] using data to 1.3 Å resolution. The final model contains 2H10 residues 1–122. Molecular graphics figures were generated with PyMOL (W. Delano; http://www.pymol.org). Coordinates and structure factures have been deposited in the Protein Data Bank with accession ID 4B50.

NMR analyses of 2H10 and 2H10 in complex with its peptide epitope

A ¹³C,¹⁵N-labeled protein sample was prepared at a concentration of 0.95 mM in 20 mM HEPES buffer and 75 mM NaCl at pH 6.7 with 10% D₂O. For the study of the peptide∶2H10 complex, a solution of gp41 peptide (₆₅₅KNEQELLELDKWASL₆₆₉.) was prepared at 7.9 mM in the same HEPES buffer and 5 µl were added step by step to the ¹³C,¹⁵N-labeled protein sample. A final protein∶peptide ratio of 1∶1.5 was obtained and ¹⁵N HSQC experiments were recorded for each peptide addition in order to control the complete disappearance of the free form. Sequential backbone resonance assignments of free and peptide bound 2H10 were performed by recording series of 3D experiments using the BEST scheme (HNCO, HNCA, HNcoCACB, HNCACB) and the BEST-TROSY scheme (hNcocaNH, hNcocaNH) [93].

All NMR experiments were performed at 37°C on Agilent VNMRS 600 or 800 MHz spectrometers equipped with triple-resonance (¹H, ¹³C, and ¹⁵N) coldprobes and shielded z-gradients. NMR data were processed with NMRPipe [94] and analyzed using the CcpNmr Analysis routines of CCPN [95]. Chemical shifts were referenced with respect to the H₂O signal at 4.66 ppm relative to DSS, using the ¹H∶X frequency ratios of the zero point according to Markley et al. [96].

Surface plasmon resonance

Surface plasmon resonance (SPR) analysis was performed with a Biacore T100 (GE Healthcare). As a flow buffer 10 mM HEPES, 150 mM NaCl, pH 7.4 with 0.005% Tween-20 was used. Gp140-92UG037 was immobilized to 1500 response units using 50 µg/ml protein in flow buffer on an activated CM-5 sensor chip (GE Healthcare: BR-1000-50) according to the manufacturer's instructions. Specific binding to the target protein was corrected for nonspecific binding to the deactivated control channel. The flow rate was 30 µL/min. Regeneration of the sensor chip was achieved with 4 M MgCl₂ for 60 seconds at 50 µL/min. Single concentration injections were all performed with 100 nM of VHH for 60 seconds each. All data were measured in duplo and the curves superpose quite well. Data were analyzed with the Biacore T100 software version 2. Curves obtained for 2H10 could not be fitted well with the 1∶1 Langmuir or heterogeneous ligand algorithms, but only with the two-state reaction algorithm. To establish whether 2H10 indeed binds with a two-state reaction, we injected 2H10 with different injection durations. This clearly showed that the dissociation rate is dependent on the injection time. The longer the duration of the injection, the stronger the interaction becomes and the dissociation rate decreases. We, therefore, fitted the curves belonging to the monovalent 2H10 VHH with the two-state reaction model. Curves of the bivalent 2H10 should be fitted with a combination of two-state reaction and bivalent analyte model, but due to a lack of availability of this model, the curves were fitted with the heterogeneous ligand fit and the two-state reaction fit.

Additional SPR analysis was performed using a Biacore 3000 (GE Healtcare). As a flow buffer 10 mM HEPES, 150 mM NaCl, pH 7.4, 50 µM EDTA with 0.005% P-20 (GE Healthcare) was used. Gp41_INT or gp41_528–683 was immobilized to 1000 response units in a sodium acetate buffer, pH 4.5, on an activated CM-5 sensor chip according to the manufacturer's instructions. Specific binding to the target protein was corrected for nonspecific binding to the deactivated control channel. The flow rate was 20 µl/min. Regeneration of the sensor chip was achieved with 25 mM NaOH for 30 seconds at 50 µl/min. Data were analyzed with the BIAevalution software version 4.1. The curves for 2H10 and 2H10-W100A on gp41_INT were fitted with separate association and dissociation, because global fittings, either 1∶1 langmuir or two-state reaction, did result in reasonable Chi² scores and residuals.

Isothermal titration calorimetry

Isothermal Titration Calorimetry (ITC) measurements were carried out at 25°C on a VP-ITC microcalorimeter (MicroCal). Both 2H10 and peptides were purified or dissolved, respectively, in a solution containing 20 mM HEPES, pH 7.4 and 100 mM NaCl. The samples were degassed for 20 min and centrifuged to remove any residuals prior to the measurements. The concentration of 2H10 in the sample cell was 10 µM and that of peptide 903 or 904 in the syringe was approximately 100 µM. The peptides injected into buffer alone was used as a negative control. Data were fit to a single binding site model and analyzed using Origin version 5 (MicroCal) software.

Site directed mutagenesis

Site directed mutagenesis was performed by PCR with complementary oligos which carried the desired mutation, to amplify the whole plasmid. The oligo's were designed to be between 42–50 nucleotides long and have a melting temperature (salt adjusted, calculated on http://www.basic.northwestern.edu/biotools/oligocalc.html) of approximately 80°C. The PCR reaction was performed with 20 ng of template DNA (2H10-wild type plasmid), 125 ng each primers, 2 mM DNTPs, Pfu buffer with 2 mM, MgSO₄ and 0.5 units of Pfu polymerase (Fermentas). The protocol consisted of an initial denaturation step at 95°C for 5 min followed by 16 cycles of 95°C for 30 seconds, 60°C for 1 minute and 68°C for 7 minutes, and a final step extension at 68°C for 10 min. DNA was cleaned with PCR Clean-up kit (Macherey-Nagel). Parental methylated DNA was digested with DpnI (Fermentas), followed by transformation into TG1 E. coli cells.

Modeling of the gp41 peptide interaction with 2H10

Models of gp41 peptide docked onto 2H10 were built with the webserver version “The HADDOCK web server for data-driven biomolecular docking” of HADDOCK2.1 [97] using CNS1.2 [98] for the structure calculations. The coordinates of 2H10 were taken from the crystal structure, and the gp41 peptide coordinates were based on those of the helical peptide (653-QEKNEQELLELDKWASL-669) from the crystal structure of a late fusion intermediate of HIV-1 gp41 (PDB code: 2X7R [29]). Ambiguous interaction restraints were defined, based on the NMR chemical shift perturbation (CSP) mapping using the program SAMPLEX [99] and the epitope mapping. The SAMPLEX program takes as input a set of data and the corresponding three-dimensional structure and returns the confidence for each residue to be in a perturbed or unperturbed state. Residues 47–51, 57–62 and 95–99 identified as perturbed in 2H10 upon binding to gp41 and residues 657, 658, 661, 662 and 665 identified from the epitope mapping, were used as interaction restraints for docking. Phi-psi angles were deduced from the ¹³C chemical shifts using TALOS+ [100] and the resulting dihedral and H-bond restraints were used in the modeling as well. The docking was performed with default HADDOCK parameters, except that random removal of restraints was turned off and the clustering cutoff was decreased to 2.5 Å because of the small size of the peptide. Non-bonded interactions were calculated with the OPLS force field using a cutoff at 8.5 Å. The electrostatic potential (E_elec) was calculated by means of a shift function, while a switching function – between 6.5 and 8.5 Å – was used for the Van der Waals potential (E_vdw). The HADDOCK score is used to rank the generated models. It consists of a weighted sum of intermolecular electrostatics, Van der Waals, desolvation (ΔG_solv) [101] and ambiguous interaction restraint (AIR) energies, defined as: Haddock Score = 0.2E_elec+1.0E_VdW+1.0E_ΔGsolv+0.1E_AIR

Immunization of llamas, library construction and selection of VHH against gp41

Two llamas (L6 and L7) were immunized with the gp41 proteoliposomes composed of an HIV-1-like lipid envelope [102] and membrane-anchored gp41CHRTM (Fig. 1). The presence of anti-gp41CHRTM proteoliposome antibodies in the plasma was confirmed by ELISA at day 21 post immunization. However, no significant neutralization was detected in the sera/plasma at day 21 and 43 compared to those of day 0. Two phage display libraries were constructed and the selection of VHH targeting gp41 was performed in two rounds by direct panning of the phage display library on immobilized gp41 proteoliposomes. In the first round triethyl-amine (TEA) was used to elute the phages. In the second round a specific competitive elution with bnAbs 2F5 or 4E10 was performed. Monoclonal VHH expressing TG1 clones from first and second round selections were screened for binding to detergent-solubilized gp41CHRTM and gp41-GCN (with pIIGCN in place of the transmembrane region) by phage ELISA. Approximately 80% of all monoclonal VHH displaying phages were positive for binding. In order to find VHH targeting specific binding sites, a competition ELISA between phages and bnAb 2F5 or 4E10 was performed. The phages with either strong binding signals, and/or binding signals that were inhibited by bnAb 2F5 or 4E10 were chosen for further investigation with purified VHH. Using ELISA, three of the VHH were found to bind to gp41-GCN, gp41 proteoliposomes and gp140-92UG037. 2H10, which originated from llama 7 (L7), exhibited the highest maximum signal and was therefore chosen for the subsequent studies (see Fig. 2 for the sequence of 2H10). 2H10 competed for binding to gp41 with 2F5, but not with 4E10.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. The gp41CHRTM antigen used for immunization.

A) Schematic representation of gp41 and of the regions present in gp41CHRTM. FP, fusion peptide; HR1, N-terminal heptad repeat; C-C loop, cysteine loop, HR2, C-terminal heptad repeat; MPER, membrane proximal external region; TM, transmembrane region. The residue numbers at the domain/region boundaries are given. B) Gel filtration chromatogram of recombinant gp41CHRTM, which elutes at 13.3 ml from the column, similar to the elution profile of a marker protein of 158 kDa. This indicates that gp41CHRTM is most likely trimeric and may have an elongated structure. The inset shows a Coomassie stained SDS-PAGE gel with the gp41CHRTM protein band at the left and a protein marker at the right with the marker protein sizes in kDa indicated at the right.

https://doi.org/10.1371/journal.ppat.1003202.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Amino acid and DNA sequence of 2H10.

A) Amino acid sequence. Secondary structure assignment is based on the crystal structure. β-strands are depicted as blue arrows. CDR regions are indicated with CDR1 to CDR3 and are underlined. Residues depicted in bold were mutated. Mutants that still showed binding are shown in green and mutants that do not bind to the antigen anymore are shown in red. B) DNA sequence alignment of 2H10 with the germline V-gene from which it originated. The asterisks indicate identical nucleotides. The CDR1 and CDR2 coding regions are underlined. Codons with at least two point mutations are in boldface.

https://doi.org/10.1371/journal.ppat.1003202.g002

Epitope mapping of 2H10 with Pepscan analysis

To determine the epitope of 2H10, its binding to a set of overlapping linear and cyclic peptides covering the gp41 MPER was measured. The best binding cyclic 15-mer peptide covered the gp41 region from amino acid 655 to 669 (₆₅₅KNEQELLELDKWASL₆₆₉). This peptide was used as a seed for a library in which the amino acids on each position of the peptide was substituted by all other 19 natural amino acids. Probing this library with 2H10 and 2F5 Fab allowed the mapping of the 2H10 epitope and identified five residues (E657, Q658, L661, E662 and K665) important for 2H10 binding. Notably, three out of the five are part of the 2F5 epitope (Fig. 3A and 3B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Epitope mapping of 2H10 and binding to helical peptides.

A) The peptide Sym-903 was used for the replacement analysis with 2F5 and 2H10. Underlined cyan colored residues are recognized by the antibodies. B) Full substitution analysis of 2H10 epitope on gp41 MPER. The binding activity of VHH 2H10 at 2 ng/ml with a peptide is shown as a vertical line proportional to the Pepscan ELISA signal. Each group of 19 lines corresponds to the replacement set of all residues except cysteine for each amino acid position in the original 15-mer cyclic peptide: KNEQELLELDKWASL. Within each group of 19 lines the substitutions are in alphabetical order: based on the one-letter amino acid code (ADEFGHIKLMNPQRSTVWY) and the reactivity of the original nonapeptide is shown as a red line. The peptides contained an N- and C-terminal cysteine and were cyclised by addition of a T2-CLIPS [115]. These results show that EQ at position 3 and 4 and LE at position 7 and 8 and K at position 11, cannot be replaced by another amino acid without loss of binding and are thus essential for binding 2H10. C) Binding of VHH 2H10 (100 ng/ml) and D) Mab 2F5 (1 ng/ml) to gp41-MPER peptides grafted in GCN4-like helical templates that contain a stabilizing Ile/Leu heptad repeat (top panel) and binding to peptides in which the zipper motif is mutated to destabilizing glycine residues (lower panel). Peptide sequences are all based on hybrids of HIV-1 gp41 and GCN4. Alignment is shown of peptide sequence with heptad repeat positions (top line). Residues essential for 2H10 binding according to the substitution analysis are underlined and residues essential for 2F5 binding are in green. Mutations are shown with one-letter codes and dots indicated that the residue is not changed.

https://doi.org/10.1371/journal.ppat.1003202.g003

Next, we tested whether 2H10 recognizes its epitope in a helical conformation. To achieve this, the peptide sequence (residues 655–669) was fused to a short coiled-coil that was shown to favor a helical conformation of the fused sequences [103]. The fusion construct was designed in such a way that the 2H10 residues required for interaction are exposed. Binding assays demonstrated that 2H10 binding was unaffected when the MPER peptide was fused to the coiled-coil template (Fig. 3C). In contrast, 2F5, which binds its epitope in a β-hairpin conformation [24] shows a dramatically reduced binding to the coiled-coil MPER construct (Fig. 3D). Replacement of heptad positions by Gly, which are helix breakers and thus do not support a helical MPER conformation, further leads to the loss of 2H10 interaction (Fig. 3C).

2H10 binding to Env and construction of bivalent 2H10

Because MPER antibodies primarily target the fusion intermediate state of gp41 [60], [104], [105], binding of 2H10 was tested to a soluble form of the fusion intermediate conformation of gp41, gp41_INT [60]. SPR measurements revealed a K_D of 29 nM for the 2H10 monohead (Fig. S1A). The avidity increased substantially to 2.5 nM, when two 2H10 VHH were linked with a 15-GS linker (bi-2H10) (Fig. S1B). Surprisingly, 2H10 and bi-2H10 interacted with similar affinities of 18 nM and 1.1 nM, respectively with gp41_523–883, a potential late fusion intermediate conformation [29] (Figs. S1C and S1D). In addition binding to gp140 from clade 92UG037 was recorded. 2H10 interaction with gp140 was fit with a two state reaction model, because an experiment with different injection durations with 2H10 indicated that it binds in two states, a low affinity followed by a high affinity state (Fig. S1E), which may be due to heterogeneous conformational states of gp41 present in recombinant gp140 consistent with the presence of non-native Env on virions [106]. Again the affinity of 2H10 (K_D = 4 nM) is inferior to bi-2H10 binding (K_D = 0.15 nM), based on both the heterogeneous ligand fit and the two state reaction fit (Figs. S1F and S1G). In order to confirm the pepscan 2H10 binding to the linear gp41 peptide epitope, isothermal titration calorimetry was used to determine the affinity of 2H10 to linear and cyclic forms of peptide ₆₅₅KNEQELLELDKWASL₆₆₉. This revealed affinities of 35 nM and 84 nM, respectively, indicating that cyclisation of the peptide epitope is unfavorable and reduces the binding affinity. However, the K_D of 35 nM is in good agreement with the SPR data and indicates that no other determinants of gp41 (present in gp41_int) contribute to binding. We next tested whether 2H10 interacts with wild type native Env expressed on the plasma membrane, which revealed no interaction with either cleaved or uncleaved forms of JR-FL Env comparable to the lack of significant 4E10 interaction (Fig. 4). We conclude that 2H10 interacts with a linear sequence with a helical conformation present in two gp41 conformations tested.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. 2H10, bi-2H10 and bi-2H10 W100A do not interact with membrane anchored Env.

FACS analysis of HEK293T cells transfected with JR-FL Env in its cleaved (A) and un-cleaved form (B). The increase in mean fluorescence intensity (MFI) has been blotted as a function of antibody concentration. Control antibodies include b12, b6, b12-Bio (biotinylated version of B12) and 4E10. The lack of significant b6 interaction with the cells expressing cleaved Env indicates that most Env has been processed.

https://doi.org/10.1371/journal.ppat.1003202.g004

Bi-2H10 neutralizes HIV-1

The 2H10 VHH monohead did not neutralize any of the HIV-1 strains tested (Table 1). However, bivalent 2H10, which showed an increased affinity for gp41 as compared to the monohead, was able to neutralize HIV-1 isolates and pseudoviruses in a TZM-bl assay. No substantial difference was observed for bi-2H10 with a 15 or 17 amino acid linker. The activity of bi-2H10 with the 17 amino acid linker (bi-17GS) was mainly directed towards clade B, neutralizing 14 of 26 B isolates, one clade A (i.e. 92UG037) and none of the clade C isolates due to the sequence conservation of the epitope. Eight strains neutralized are Tier 1 and seven are Tier 2 viruses. Bi-2H10 also neutralized both the neutralization sensitive and resistant variants of SHIV SF162 consistent with SF162.LS neutralization (Table 1). The IC50 values were between equal to and 100-fold lower than those of mAb 2F5, which neutralized 18 out of 23 clade B viruses tested (Table 1). Bi-2H10, like 2H10, did not interact with cell surface expressed JR-FL Env (Fig. 4), although the virus was neutralized. We conclude that that bi-2H10 acts downstream of receptor-induced conformational changes in Env and the increased avidity of bi-2H10 renders the VHH neutralization active.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. 2H10, bivalent 2H10 and mutant bi-2H10 IC50 titers against HIV-1 in TZM-bl cells.

https://doi.org/10.1371/journal.ppat.1003202.t001

2H10 exposes W100 at the tip of CDR3 which is required for neutralization

To study the interaction of 2H10 with its epitope, several attempts were made to obtain crystals of 2H10 in complex with various MPER peptides based on the Pepscan analyses. However, in each attempt, diffraction quality crystals were obtained that contained the VHH, but not the peptide. The best 2H10-containing crystals diffracted to 1.3 Å resolution and the structure was solved by molecular replacement (Table 2). The structure resembles the framework of known VHH structures [76], with the major exception that most other VHH structures have CDR3 loops that fold back on the framework of the VHH, whereas the CDR3 of 2H10 protrudes from the framework of 2H10. However, it should be noted that its orientation may have been influenced by crystal contacts. Its most notable feature is the exposure of W100 at the tip of CDR3 (Fig. 5A). CDR3 W100 resembles the presence of hydrophobic residues within the CDR3 heavy chains of 2F5 and 4E10 [24], [25], which are not essential for epitope interaction although changes in affinity have been reported depending on the nature of the antigen for some mutants. In contrast, mutations of hydrophobic CDR3 residues affect the neutralization potency of these antibodies dramatically [48]–[50]. Similarly, several 2H10 W100 mutants (i.e. W100A, W100F and W100L) interacted with gp41_INT and gp140 with affinities comparable to wild type 2H10 (Fig. 6A and Fig. S1H), confirming that W100 is non-essential for epitope interaction. To study whether W100 is required for neutralization, the W100A mutation was introduced into bi-2H10 (bi-2H10-W100A). SPR confirmed that bi-2H10-W100A binds to gp140 comparable to wild type bi-2H10 revealing the same dissociation rate together with a slightly slower on-rate (Fig. 6B). Although not essential for gp41 interaction, the exchange of W100 to alanine completely abrogated neutralization of most strains. Four other strains tested showed a reduced potency as evident from the 4-and 6-fold increased IC50 values for these viruses (Table 1). In addition, bi-2H10 W100A did not interact with surface-expressed Env as expected (Fig. 4). We thus conclude that CDR3 W100 is dispensable for epitope interaction but is essential for neutralization similar to the role of hydrophobic CDR3 heavy chain residues present in mAbs 2F5 and 4E10 [48]–[50]. In order to map the gp41 binding site on 2H10 the following mutants (S29F, D53T/R54S, R54S, R56A, R71S, R93A, E95A, R97A) were designed based on the structure (Fig. 5A) and the VHH germline sequence of CDR1 and CDR2 (Fig. 2). Assessment of binding by SPR revealed that the single mutants R56A, R93A, E95A, and R97A lost binding to gp140-92UG037 (Fig. 6A). All other mutants (S29F, D53T/R54S, R54S, R71S) still bound gp140, although the affinities for some seemed to be marginally lower than for wild-type 2H10. This thus implicates the charged residues R56 in CDR2 and R93, E95 and R97 of CDR3 in gp41 recognition.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Crystal structure of 2H10 and NMR analysis of 2H10 interaction with the gp41 peptide.

A) Ribbon representation of 2H10. CDR 1, 2 and 3 are colored blue, yellow and green, respectively. Residues implicated in gp41 interaction by mutagenesis are shown as sticks. W100 at the tip of CDR3 is required for neutralization. B) Selected region of the ¹⁵N HSQC spectra (see Fig. S3) recorded on a ¹³C,¹⁵N labeled 2H10 sample. Increasing concentrations of the gp41 peptide were titrated into the 2H10 solution and induced specific chemical shifts; spectra were recorded with a protein∶peptide ratio of 1∶0, in grey, 1∶0.5 in blue and 1∶1 in red.C) Chemical shift perturbations (CSPs) were mapped onto the 2H10 structure. Residues showing ¹⁵N,¹H chemical shift perturbations greater than 0.15 ppm are shown in red and residues with an amide resonance disappearing in the free or the bound form are colored in green.

https://doi.org/10.1371/journal.ppat.1003202.g005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Surface plasmon resonance affinity measurements reveal essential and non-essential 2H10 residues for gp41 interaction.

A) 2H10 single domain mutants. The curves are labeled with the mutant codes in colors corresponding with the colors of the curves. The mutants R56A, R96A, E98A and R100A do not bind at all to gp140-92UG037. Fitting the curves with the two-state reaction algorithm yielded K_Ds of 8.0 nM for 2H10 WT and 10.1 nM for 2H10-W100A. The other fits, except for those of the non-binding mutants, yielded K_Ds between 5.5 nM for 2H10-S29F and 18.7 nM for 2H10-R71S B) bi-2H10-W100A mutant and wild type bi-2H10 on immobilized gp140-92UG037. Because the maximum responses units are slightly different for the two bi-2H10 molecules, the curves were normalized to the maximum response, to be able to compare the dissociation rates well.

https://doi.org/10.1371/journal.ppat.1003202.g006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Data collection and refinement statistics.

https://doi.org/10.1371/journal.ppat.1003202.t002

Because it was reported that mAbs 2F5 and 4E10 have lipid binding activity, we tested whether 2H10 also binds liposomes or lipids. Sucrose gradient flotation analyses showed that 2H10 does not float with liposomes having the lipid composition of native HIV-1 virions. 2H10 is only found in the bottom fractions of the gradient while a substantial amount of 4E10 floats to the upper fraction of the gradient in this assay. However, 2F5 did not reveal any liposome binding either (Fig. S2A). This is consistent with ELISA data which demonstrated binding of mAb 4E10 to cardiolipin and phosphatidylserine, while 2H10, bi-2H10 and 2F5 revealed no binding activity (Fig. S2B). We conclude that 2H10 exerts no measurable lipid binding activity.

2H10-gp41 complex formation: NMR characterization of 2H10 in the free and peptide-bound form

¹⁵N HSQC spectra of 2H10 were recorded in the absence or presence of an excess of the gp41 peptide and revealed a sufficient spectral resolution to allow the 1H,15N, and 13C backbone assignment of the two forms using heteronuclear experiments (Fig. S3). Chemical shift index (CSI) calculated from the CB, CA, N and CO chemical shifts reveal similar secondary structures for the free and the bound forms (Fig. S4). The specific shift of individual residues was observed using ¹⁵N HSQC spectra recorded at different ratios of gp41 peptide to 2H10 and depend on the peptide concentration (Fig. 5B). These data reveal a slow exchange process between the bound and the free form in agreement with the affinity constant measured for this complex. The complete list of chemical shift differences (CSD) induced in 2H10 upon peptide binding is listed in Figure S5 (Fig. S5), which confirmed further that peptide binding, did not induce conformational changes in the tertiary structure of 2H10. Notably most significant CSDs map to one surface of the 2H10 structure providing a first clue on the gp41 ligand position (Fig. 5C).

Modeling of the 2H10 gp41 peptide complex

Modeling of the 2H10 interaction with a helical gp41 peptide conformation with HADDOCK [97] readily produced a top cluster with an average cluster score significantly lower than any other solution (−96.3±0.5 a.u.). This cluster corresponded also to the most populated one with 173 members out of 200 models generated. Its electrostatic energy is −437.0±31.1 kcal/mol and its Van der Waals energy −36.9±6.9 kcal/mol. The model reveals that the short gp41 helix binds in a parallel fashion to the 2H10 beta-sheet involving a network of salt bridges (E654-K60; K665-E95; D664-R56) and hydrogen bonds (E654-Q44; E657-S62 amide; Q658-amide-L47; E662-Y37; R96-L669 carbonyl). In addition, R93 stabilizes the orientation of Y36 and E95, which contact E662 and K665, respectively (Fig. 7A). The network of interactions is consistent with the chemical shift perturbation (Fig. 5B) and mutagenesis data, which implicate R56, R93, E95 and R97 in the interaction. CDR3 S99 and S100D show additional chemical shift differences upon peptide binding. This may indicate that the CDR3 may be flexible enough to contact the peptide or alternatively, that peptide binding induces major changes within the CDR3 conformation or orientation. Among the interacting residues are five of the six framework residues that deviate from the germline (S35, F50, K60, V61 and R93) by at least two nucleotide changes in their codons (Fig. 2B). The sixth residue (L91) is very close to the interaction site. Their occurrence in VHH is rather rare (F50,1.5% of VHH sequences, K60, 0.8%, V61, 0.7% and R93, 0.52% of VHH sequences (https://fungen.wur.nl/). The model also corroborated that CDR3 W100 does not interact with gp41. Subsequent superpositioning of the Cα atoms of the VHH-peptide model (Fig. 7A) and the trimeric structure of the late fusion intermediate [29] yielded a trimer model without any clashes (Fig. 7B). Importantly, it revealed that the side chain of W100 is positioned in such a way that it could dip into the membrane bilayer in the same plane as gp41 MPER residues W678, W680 and Y681 (Fig. 7B). The latter have been proposed to insert into the membrane during the conformational changes of gp41 leading to membrane fusion [29]. We conclude that 2H10 binds a helical epitope of gp41, which allows positioning of the CDR3 W100 towards the membrane. Due to the small size of the VHH, 2H10 may stay associated with gp41 until late stages of membrane fusion.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Molecular modeling of the 2H10-gp41 peptide interaction suggests that 2H10 W100 is oriented towards the membrane.

A) Molecular model of the gp41 peptide interaction with 2H10 produced by HADDOCK. Salt bridges and hydrogen bonds present in the top model and found in most of the top10 models are shown as dashed lines. B) Superpositioning of the Cα atoms of the gp41 peptide derived from the HADDOCK 2H10-peptide model onto the structure of a late fusion intermediate of gp41 [29] shows that the CDR3 W100 is oriented towards the membrane and found in the same plane as gp41 residues W678, W680 and Y681. Note that there are no clashes between the 2H10 VHH and gp41.

https://doi.org/10.1371/journal.ppat.1003202.g007