Identification of a new epitope for HIV-neutralizing antibodies in the gp41 membrane proximal external region by an Env-tailored phage display library

Identification of EC plasma with broad neutralizing capacity

We recruited a cohort of 12 ECs and 2 viremic controllers (VCs) for this study, the characteristics of which are summarized in Supporting Information Table 1. To characterize the nAb profile in plasma from these ECs/VCs, we first used tier 1 Env-pseudotyped viruses HIV-1SF162.LS, HIV-1Bal.26, and HIV-1ss1196.1 in a standardized TZM-bl assay 15. Supporting Information Table 2 summarizes the neutralizing Ab titers of the plasma samples expressed as the reciprocal plasma dilution resulting in 50% inhibition of the infection (IC50). Strain SF162.LS was neutralized by all EC/VC plasma samples except for EC06 and VC12, with IC50 values ranging from 178 to >12,150. Plasma samples from EC07, EC08, EC09, EC10, EC26, and VC11 neutralized all three subtype B tier 1 strains with titers comparable or better than the broadly neutralizing control sera VI-2992 and VI-3196, which are, however, derived from a subtype A infected individual 30. Plasma from EC26 ranked top against HIV-1SF162.LS and HIV-1ss1196.1 and ranked second against HIV-1Bal.26. The breadth of nAbs of EC26 plasma was therefore further determined against an Env pseudotyped virus panel of 11 tier 2 isolates from subtypes A, B, C, and CRF BC. EC26 plasma neutralized 10/11 viruses out of the tier 2 panel with titers comparable with broadly neutralizing plasma samples from the control patient ITM1 (IC50 values between 21 and 76, data not shown). Due to the broad neutralizing activity of the EC26 plasma sample, we continued using this sample to identify the respective epitopes targeted by neutralizing antibodies.

Selection of epitopes for nAbs in EC26 plasma based on an HIV-1 Env-tailored phage display library

Neutralizing antibodies target various regions of the HIV-1 envelope involved in virus entry. In order to select such Env epitopes, we generated a phage display library displaying Env peptides from shotgun cloned PCR fragments derived from the entire gp160 sequence of the HIV-1ADA strain at the N-terminus of the pIII phage protein. Randomly analyzed library clones showed that about 15% of the clones contained HIV-1ADA env DNA fragments positioned in frame with the pelB leader and the phage gIII gene and encoding Env open reading frames of 30–100 aas.

The biopanning procedure was initiated by a negative selection with immobilized IgGs from pooled HIV negative plasma. Subsequently, positive selections were performed with the depleted library and IgG from the EC26 plasma sample. After two rounds of negative and positive selections, 150 phage clones were analyzed by ELISA to determine their binding specificity for EC26 IgG. From these, the 22 most reactive clones were subjected to sequence analysis. By comparing the peptide insert sequences to HIV-1ADA Env sequences, the 22 sequenced clones could be grouped into five distinct clusters. The ELISA reactivity of one representative phage of each group is shown in Figure 1A. Peptide inserts showed linear homology to Env or to the second exon of Rev. Two sequences could be allocated completely to gp41, one mapped to the C-terminus of gp120, but extended into the N-terminal part of gp41 (Supporting Information Fig. 1). Clone EC26-2H10 covered the well-known KLIC motif, a linear immunodominant region corresponding to gp41 cluster II 21, 31. An interesting clone, EC26-2A4, mapped to the MPER region of gp41 known to be targeted by broadly neutralizing mAbs 2F5, 4E10, and Z13. The peptide insert of the clone EC26-2A4 (IEESQNQQEKNEQELLELDKWASLWNWFD) contains the entire 2F5 epitope (ELDKWA), whereas the 4E10 epitope (NWFDIT) is only partially contained at the C-terminus. To characterize further the binding activity of the epitope EC26-2A4 with mAbs 2F5 and 4E10, we performed Ab phage ELISA. As expected from the sequence data, EC26-2A4 strongly bound mAb 2F5, whereas mAb 4E10 was not bound (Fig. 1B).

Epitope EC26-2A4 is targeted by HIV-1 neutralizing antibodies present in plasma EC26

To clarify whether the selected EC26-2A4 peptide is targeted by neutralizing antibodies contributing to the neutralizing activity of plasma from EC26, we first preabsorbed EC26 plasma with the purified phage EC26-2A4 or Hyperphage (control without peptide insert), followed by neutralization studies of the depleted plasma. Preincubation with phage EC26-2A4 resulted in a significant reduction in the neutralizing activity of EC26 plasma against HIV-1SF162.LS compared with that of plasma preincubated with the control Hyperphage (Fig. 2B). The IC50 values of EC26 plasma samples preincubated with EC26-2A4 phage and Hyperphage, respectively, differed significantly (11,547 versus 20,535). Likewise, preincubation of mAb 2F5 with the EC26-2A4 phage resulted in reduced neutralization compared with that of preincubation with Hyperphage (Fig. 2A). The fact, that preincubation with Hyperphage also reduced the mAb 2F5 neutralizing activity, is probably due to unspecific blocking due to the large size of the phage coat (>2500 copies for the pVIII protein) as opposed to the three to five copies of the specific epitope displayed on the phage linked to the pIII protein.

We then affinity purified IgG specific for the EC26-2A4 peptide from EC26 plasma with a fusion protein (EC26-2A4-MBP), where EC26-2A4 has been fused N-terminally to the maltose-binding protein (MBP). This fusion protein showed similar binding for EC26 IgG as the original EC26-2A4 phage (data not shown). The purified IgG fraction (referred to as EC26-2A4-IgG) was then analyzed for neutralization of HIV-1 on TZM-bl cells. A total of 50 μg/mL EC26-2A4-IgG neutralized 91.8 % of SF162.LS, 53.1% of ss1196.1, and 48.3% of Bal.26, respectively (Fig. 3). Thus, EC26 plasma indeed contains neutralizing antibodies targeting the MPER peptide EC26-2A4, which contribute significantly to the neutralizing activity in the plasma.

As neutralizing mAbs against MPER, in particular mAb 4E10 and to a lesser extent 2F5, are associated with autoreactivity against phospholipids, we analyzed the reactivity of plasma EC26 and the antibodies affinity purified with EC26-2A4-MBP for reactivity with phospholipids. The plasma as well as the purified EC26-2A4-IgG did not show significant reactivity with cardiolipin by ELISA at 100 μg/mL, which is a clinically relevant concentration 32, 33 (Fig. 4A); however, the purified antibodies showed low reactivity with phosphatidylserine, which was comparable with that of mAb 2F5 (Fig. 4B). The positive control, mAb 4E10, showed strong reactivity with cardiolipin as well as with phosphatidylserine at the same concentration, whereas the negative control mAb 447-52D did not react with either of the Ags.

The core epitope of EC26-2A4-IgG is distinct but overlaps with the epitope of mAb 2F5

The core epitope for purified EC26-2A4-IgG was further fine mapped on peptide arrays (Fig. 5). By comparing the reactivity of sets of overlapping peptides shifted by one aa with mAb 2F5 and purified EC26-2A4-IgG we could define more precisely the core epitope recognized by EC26-2A4-IgG. As expected from published studies, mAb 2F5 recognized the core epitope LDKW on these arrays, whereas purified EC26-2A4 IgG targeted an epitope partially overlapping the 2F5 epitope, but slightly shifted toward the N-terminus. This new core epitope with the sequence (⁶⁵⁹ELLELDKW⁶⁶⁶) is termed EC26-2A4MZ. W666 was absolutely essential for the reactivity with mAb2F5, whereas the absence of this aa weakened the reactivity with EC26-2A4-IgG, but did not abolish it completely. Furthermore, the EC26-2A4MZ core epitope includes E659, which is neither present in the 2F5 epitope nor in the recently described m66.6 epitope (see below).

Subjects studied

Plasma samples from ECs and viremic controllers were obtained from the Hospital Carlos III in Madrid (Spain) and the HIV Center from the Goethe Medical School in Frankfurt (Germany) after receiving informed consent. Twelve ECs enrolled had (i) documented HIV-1 infection for more than 10 years and (ii) persistently undetectable VL < 50 copies per mL plasma in the absence of any antiretroviral therapy. Two viremic controllers had VL from 100 to 150 copies/mL plasma. Control sera VI2992 and VI3196, kindly provided by Guido Vanham (Antwerpen), were from a therapy naïve HIV-1 subtype A infected patient ITM-1, and previously reported to broadly neutralize primary isolates on PBMC and TZM-bl cells 30. Further, control sera for the neutralization studies were previously published 14 and include LTNPs MH01 and MH04 with high titers of neutralizing antibodies against a panel of recombinant isolates from clade B and C and sera from four progressors with comparable CD4 counts (>400 per mL) and low VL (<11,000 copies per mL) (Supporting Information Table 1). The study was approved by the Ethics Commission of the Goethe University of Frankfurt am Main, Germany.

Neutralization assays on TZM-bl cells

Neutralization studies were performed using a panel of standardized HIV-1 pseudoviruses on TZM-bl cells 15. All plasma samples were heat-inactivated at 56°C for 1 h prior to use. Neutralizing Ab titers of all samples were first analyzed against three standard HIV-1 reference subtype B tier 1 strains (SF162.LS, SS1196.1, and Bal.26). The breadth of nAbs in the best neutralizing samples was further determined against HIV-1 tier 2 pseudoviruses derived from standard primary isolates of subtype B (THRO4156.18, WITO4160.33, RHPA4359.7, TRJO4551.58, and REJO4541.67), subtype A (Q769.d22 and Q23.17 ), subtype C (DU156.12 and DU422.1), and CRF BC (CH110.2 and CH117.4) strains as previously described 15. Briefly, pseudoviruses were produced by cotransfection of 293T/17 cells with an HIV-1 env-expressing vector and an env-deficient backbone vector (pSG3Δenv) and titrated on TZM-bl cells. After preincubation of the pseudovirus stock with serial dilutions of plasma or Abs for 1 h at 37°C, TZM-bl cells were infected in a 96-well plate for 48 h. Relative luminescence units were determined and the 50% inhibitory dose (IC50) was determined compared with virus only controls.

Construction of an HIV-1 Env-tailored phage display library

The env gene of HIV-1ADA was amplified from the pSV3-ADA env plasmid (kindly provided by Dr. Rolf Kaiser, Cologne) and the PCR product was sonicated to generate random DNA fragments with a size of 100–1200 bp. After purification and polishing, the fragments were cloned into the pHORF3 vector 16 and electroporated into electro-competent TOP10F′ (Invitrogen). The grown colonies were harvested by suspending in 40 mL 2× tryptone yeast medium. The library was either directly used for phage packaging using Hyperphage 81, 82, or after addition of glycerin (20%), the library was stored at −80°C in 1 mL aliquots.

Colony PCR and sequence analysis of inserts

The size distribution of the inserts from single clones bearing pHORF3 was analyzed by colony PCR using the MHLacZ-Pro-f primer (5′-GGCTCGTATGTTGTGTGG-3′) and MHgIII-r primer (5′-GGAAAGACGACAAAACTTTAG-3′) 83 following the protocol: 94°C, 1 min; 56°C, 1 min; 72°C, 2 min; 30 cycles. Plasmids from 136 clones bearing pHORF3 were purified and sequenced using the same primer pair on an ABI PRISM 310 Genetic Analyzer to determine the representation of different env regions in the library.

Selection of phage reactive with HIV antibodies from EC

Plasma samples from ECs and pooled HIV negative plasma were diluted 1:100 in 2% MPBST and added to microtiter plates coated with a goat antihuman Fc Ab for 2 h at RT. After washing, 100 μL (∼10⁷ phage particles) of the HIV-1 Env-tailored phage display library were first incubated with the wells containing pooled HIV negative IgGs for 2 h at RT (negative selection). Then supernatants were transferred into wells coated with the IgGs from EC for 2 h at RT. The nonbinding polypeptide phage particles were removed by ten stringent washing steps with PBST. After washing, the bound phage were eluted with 200 μL/well Trypsin (10 μg/mL) for 30 min at 37°C. Ten microliters of the eluted phage were used for titration as reported 84. A total of two rounds consisting of negative and positive selections were performed. Single phage clones were produced as previously described 16.

Phage capture ELISA

Twenty-five nanograms of mouse anti-M13 (B62-FE2, Progen) in PBS were coated at 4°C overnight followed by blocking with 2% MPBST. After washing three times with PBST, 100 μL of the monoclonal phage were added for 2 h. Plasma from ECs or pooled HIV negative plasma samples were diluted 1:100 in 2% MPBST supplemented with 10% v/v TOP10F' cell lysate and 1 × 10⁸ CFU/mL Hyperphage, added to the captured phage particles and incubated for 2 h at room temperature. Bound IgGs were detected with HRP-conjugated goat antihuman IgG for 1 h and visualized with TMB microwell peroxidase substrate (KPL, USA). The staining reaction was stopped by adding 100 μL 1 N HCl. The absorbance at 450 nm and scattered light at 620 nm were measured as reference using an ELISA reader (MWG-BIOTECH).

Identification of selected epitopes

Plasmids from clones with strong ELISA reactivity were purified and the inserts sequenced with the above-mentioned MHLacZ-Pro-f and MHgIII-r primers. Deduced aa sequences were compared with published HIV-1 protein sequences (http://www.ncbi.nlm.nih.gov/blast/blast.cgi) and similarities with known HIV-1 neutralizing Ab epitopes were studied using the online HIV database (http://www.hiv.lanl.gov/content/index).

Depletion of epitope-specific neutralizing antibodies from plasma

Plasma or antibodies were preincubated with 10⁶ CFU of twice PEG-purified and filtered EC26-2A4 phage/Hyperphage or MBP-EC26-2A4 fusion protein (see below, 10 μg/mL) for 1 h at 37°C followed by neutralization assays on TZM-bl cells as described above.

Cloning of selected phage epitopes in fusion with MBP

The peptide insert EC26-2A4 was amplified from the phage by PCR and cloned N-terminally to the MBP into the expression vector pMAL-pIII (NEB). Briefly, EC26-2A4 was amplified using the primers EC26-2A4-fw (5′- CGG GGT ACC TAT TGA AGA ATC GCA GAA CCA GCA A-3′) and EC26-2A4-rev (5′-TCC CCG GCC GAG TCA AAC CAA TTC CAC AAA CTT GC-3′) to introduce restriction sites KpnI and EagI, respectively, for cloning. After sequence analysis, the EC26-2A4 peptide fused to MBP was expressed following the NEB manual, purified by amylose column (NEB) and quantified using the BCA assay kit (Thermo).

ELISA

Plates were coated with 200 ng/well proteins in PBS at 4°C overnight. After washing three times, plates were blocked with 5% MPBST for 2 h at room temperature and incubated overnight with plasma diluted 1:100. Plates were washed three times and incubated with HRP-conjugated antihuman IgG antibodies for 1 h at room temperature. After washing five times, plates were developed using 100 μL/well Sureblue TMB substrate (KPL), stopped with 100 μL 1 N HCl and read at 450 nm and 620 nm as a reference. Reactivity of selected phage clones with mAbs 2F5 and 4E10 was tested by coating the plates with 200 ng/well antibodies, washing and blocking as above, followed by incubation with selected monoclonal phage, which were then detected by anti-M13-HRP Ab. ELISAs to determine the reactivity of plasma and purified IgG against cardiolipin and antiphosphatidylserine IgG were performed using the Cardiolipin IgG ELISA kit (Abnova) and the antiphosphatidylserine ELISA kit (Biorad) using the included controls and according to the manufacturer's instructions.

Affinity purification of plasma antibodies with the selected epitope

Purification of EC26-2A4-binding antibodies from EC26 plasma was performed using the AminoLink Plus kit (Thermo) according to the manufacturer's instructions. Briefly, 1.5 mg EC26-2A4-MBP proteins were coupled to the resin. Plasma samples were injected into the fusion protein conjugated column and incubated overnight at 4°C. After washing, the binding fraction was eluted by IgG elution buffer (Thermo). The eluted fractions were combined and concentrated to 100 μg/mL using Amicon 30 centrifugal filter units.

Mapping the core epitope of EC26-2A4 purified IgG by spot synthesis peptide arrays

Peptide arrays were synthesized by Fmoc chemistry at activated PEG spacers on cellulose membranes by automated parallel peptide synthesis on a MultiPep RS instrument (Intavis) 85. Overlapping peptides of the EC26-2A4-epitope were synthesized consisting of 18mers, 15mers, 12mers, 10mers, 8mers, and 6mers, each shifted by one aa. After rehydration, washing, and blocking with 5% (w/v) skimmed milk powder in PBS supplemented with 0.5% Tween-20 (5% MPBST) for 3 h at RT, membranes were incubated with 500 ng/mL 2% MPBST diluted affinity purified plasma antibodies or 2F5 mAb as positive control at 4°C overnight. After washing three times with PBS, bound antibodies were detected using the ECL kit (Thermo).

Statistical analysis

The graphs including mean and SD were performed using Graph Pad Prism5 (Graph-Pad Inc., San Diego, CA, USA). To strengthen our results, we used two-way or one-way ANOVA followed by t-test for all statistical analyses with the help of Graph Pad Prism5, if not indicated otherwise. Statistically significant p values are marked by asterisks in the graphs (***p < 0.001; **p < 0.01; *p < 0.05).