Cell Lines
K562-E cells (K562 cells stably expressing HLA-E) and K562-E/UL49.5 cells (with a TAP-inhibitor UL49.5) are kindly provided by Dr. Thorbald van Hall from Leiden University37. All the other cells used in this study are from ATCC. 293T cells (ATCC CRL-3216) were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM; Gibco, Catalog# 10564) supplemented with 10% fetal bovine serum (FBS; Gibco, Catalog# 10099141) and 1% penicillin/streptomycin (Gibco, Catalog# 10378016). K562 cells (ATCC CCL-243), K562-E cells and K562-E/UL49.5 cells were cultured in Iscove’s modified Dulbecco’s Medium (IMDM; Hyclone, Catalog# SH30228.01) supplemented with 10% FBS. Jurkat, DU-4475 and U-937 cells were cultured in RPMI-1640 medium (Gibco, Catalog# 72400) supplemented with 10% FBS. SiHa cells were cultured in Minimum Essential Medium (MEM; Gibco, Catalog# 11095080) supplemented with 10% FBS. The NK-92 human cell line (ATCC CRL-2407) was cultured in Alpha Minimum Essential medium (α-MEM; Gibco, Catalog# 12561072) supplemented with 2 mM l-glutamine, 0.2 mM inositol, 0.1 mM 2-mercaptoethanol, 0.02 mM folic acid, 100 U/ml recombinant IL-2 (Biolegend, Catalog# 589108), 12.5% horse serum (Gibco, Catalog# 16050122) and 12.5% FBS. All the cells were maintained at 37 °C, 5% CO2 in humidified incubators.
Animals
Transgenic mice carrying human β2-microglobulin (β2m) and HLA-B*27:05 genes were obtained from Jackson lab (B6.Cg-Tg(B2M,HLA-B*27:05)56-3Trg/DcrJ; stock# 003428). Hemizygous mice were used in this experiment, as this strain is homozygous lethal. For hemizygous mice genotyping, peripheral blood lymphocytes (PBLs) were isolated and stained using mouse CD45 antibody (Biolegend, Catalog# 103122), human HLA class I antibody (Biolegend, Catalog# 311406) and human β2m antibody (Biolegend, Catalog# 316312). All animal experiments were conducted with approved protocols from the Duke University Institutional Animal Care and Use Committee.
Human subjects
Human leukapheresis frozen vials were collected by the External Quality Assurance Program Oversight Laboratory (EQAPOL)56,57. Samples from four male donors were used in this study. Supplementary Data 2 shows the clinical characteristics of the individuals studied. All experiments that related to human subjects was carried out with the informed consent of trial participants and in compliance with Institutional Review Board protocols approved by Duke University Medical Center.
Peptide synthesis
The VL9 peptide (VMAPRTVLL) was synthesized to >85% purity via Fmoc (9-fluorenylmethoxy carbonyl) chemistry by Genscript USA and reconstituted to 200 mM in dimethyl sulfoxide (DMSO).
HLA-E-peptide protein refolding and purification
β2-microglobulin, previously purified from inclusion bodies in a Urea-MES buffer, was added to a refolding buffer to achieve a final concentration of 2 μM. The refold buffer comprised 100 mM Tris pH 8.0, 400 mM l-arginine monohydrochloride, 2 mM EDTA, 5 mM reduced glutathione and 0.5 mM oxidized Glutathione and was prepared in MiliQ water. A 20 μM concentration of VL9 peptide (VMAPRTVLL), previously reconstituted to 200 mM in DMSO, was added to the refolding buffer followed by HLA-E*0103 heavy chain, which was pulsed into the refold to a final concentration of 1 μM. Once the refold had incubated for 72 h at 4 °C it was filtered through a 1.0 μm cellular nitrate membrane and concentrated in the VivaFlow 50R and VivaSpin Turbo Ultrafiltration centrifugal systems with 10 kDa molecular weight cutoffs. The concentrated samples were injected onto a Superdex S75 16/60 column and refolded protein eluted according to size into phosphate-buffered saline (PBS). Eluted protein complexes were validated by non-reducing sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) electrophoresis on NuPAGE 12% Bis-Tris protein gels and further concentrated via VivaSpin Turbo Ultrafiltration centrifugal device to 1.1 mg/ml.
HLA-E-peptide biotinylation and tetramer generation
HLA-E-peptide samples requiring biotinylation were subsequently buffered exchanged on Sephadex G-25 PD10 columns (GE Healthcare, UK) into 10mM Tris buffer using commercially available BirA enzyme (Avidity, USA) following the manufacturer’s instructions. Following overnight biotinylation, protein samples were subsequently purified into 20 mM Tris pH 8, 100 mM NaCl buffer or PBS on a HiLoad 16/600 Superdex 75 pg column using an AKTA size-exclusion fast protein liquid chromatography (FPLC) system. Correctly folded β2m-HLA-E*01:03-peptide complexes were subsequently concentrated to 2 mg/ml and snap frozen.
HLA-E*01:03 tetramers were generated via conjugation to various fluorescent labels including Extravidin-PE (Sigma), Streptavidin-bound APC (Biolegend, San Diego) or BV421 (Biolegend, San Diego) at a Molar ratio of 4:1 as previously described12.
Immunization in HLA-B27/β2m transgenic mice
HLA-B27/β2m transgenic mice (n = 23) were intramuscularly (i.m.) immunized with pooled HLA-E-RL9HIV complex (12.5 μg/animal) and HLA-E-RL9SIV complex (12.5 μg/animal) adjuvanted with STR8S-C58 at weeks 0, 2, 4, 6, 12 and 16. MAb 3H4 was isolated from this study. In another experiment, HLA-B27/β2m transgenic mice (n = 10) were i.p. immunized with either HLA-E-RL9HIV single-chain trimer (SCT) transfected 293T cells (2 x 106 cells/animal) or HLA-E-RL9SIV SCT transfected 293T cells (2 x 106 cells/animal) at weeks 0, 2, 4, 6, 17 and 19. MAb 13F11 was isolated from this study. In the third experiment, HLA-B27/β2m transgenic mice (n = 10) were i.m. immunized with HLA-E-VL9 complex (25 μg/animal) adjuvanted with STR8S-C at Week 0, 2 and 4, following by intraperitoneally (i.p.) immunization with HLA-E-VL9 SCT transfected 293T cells (2 x 106 cells/animal) at week 14, 16 and 18. MAb 10C10 and 2D6 were isolated from this study. Serum titers were monitored by ELISA Mice with high-binding antibody titers were selected for the subsequent spleen cell fusion and B-cell sorting experiments.
Hybridoma cell line generation and monoclonal antibody production
Mice were boosted with the indicated priming antigen 3 days prior to fusion. Spleen cells were harvested and fused with NS0 murine myeloma cells using PEG1500 to generate hybridomas. After 2 weeks, supernatant of hybridoma clones were collected and screened by flow cytometry-based high-throughput screening (HTS). Specifically, we tested for antibodies differentially binding 293T cells transiently transfected with plasmid DNA expressing single-chain peptide-HLA-E-ß2m trimers so that they expressed HLA-E-RL9HIV, HLA-E-RL9SIV or HLA-E-VL9 at the cell surface. Hybridomas cells that secreted HLA-E-VL9 antibodies were cloned by limiting dilution for at least five rounds until the phenotypes of all limiting dilution wells are identical. IgG mAbs were purified by protein G affinity chromatography, while IgM mAbs were purified by ammonium sulfate precipitation and by Superose 6 column size-exclusion chromatography in AKTA Fast Protein Liquid Chromatography (FPLC) system. The VH and VL sequences of mAbs were amplified from hybridoma cell RNA using primers reported previously59,60.
Cell-surface staining and high-throughput screening (HTS)
HLA-E SCT constructs encoding HLA-E-VL9, HLA-E-RL9HIV, or HLA-E-RL9SIV were transfected into 293T cells using GeneJuice transfection reagent (Novagen, Catalog# 70967). For epitope mapping experiment, a panel of HLA-E-VL9 SCT constructs with single amino acid mutations were transfected into 293T cells using the same method. Cells were dissociated with 0.1% EDTA at 48 h post-transfection and stained with a Fixable Near-IR Dead Cell Stain Kit (Thermo Fisher, Catalog# L34976). After washing, primary antibodies (supernatant from hybridoma cells, supernatant from transfected cells, or purified antibodies) were added and incubated with cells for 1 h at 4 °C, following by staining with 1:1000 diluted secondary antibodies for 30 min at 4 °C. For mouse primary antibodies, we used Alexa Fluor 555 (AF555) conjugated goat anti-mouse IgG (H + L) (Thermo Fisher, Catalog# A32727) or Alexa Fluor 647 (AF647) conjugated goat anti-mouse IgG (H + L) (Thermo Fisher, Catalog# A32728) as secondary antibodies; for human primary antibodies, we used AF555 conjugated goat anti-human IgG (H + L) (Thermo Fisher, Catalog# A-21433) or AF647 conjugated goat anti-human IgG (H + L) (Thermo Fisher, Catalog# A-21445) as secondary antibodies. Cells were then washed three times and resuspended in fixation buffer (1% formaldehyde in PBS, pH 7.4). Data were acquired on a BD LSR II flow cytometer and analyzed using FlowJo version 10.
3H4 Fab production
A humanized version of the 3H4 antibody (3H4-huIgG1) was digested to produce Fab fragments using the Pierce Fab Preparation kit (ThermoFisher SCIENTIFIC). 3H4 Fab-retrieved sample was further purified by size-exclusion on a Superdex S75 16/60 column and eluted into PBS buffer. Following concentration to 1.1 mg/ml and SDS-PAGE gel-based validation, 3H4 Fab purified material was incubated for 1 h on ice with freshly purified HLA-E-VL9. The combined 3H4:Fab-HLA-E-VL9 sample was concentrated to 7.5 mg/ml prior to crystallographic set-up.
Crystallization screening
Crystals were grown via sitting drop vapor-diffusion at 20 °C in a 200 nl drop with a 1:1 protein to reservoir ratio61. The 3H4 Fab-HLA-E(VL9) co-complex crystallized in 20% PEG 8000, 0.1 M Na HEPES at pH 7, in the ProPlex sparse matrix screen. Crystals were cryo-preserved in 25% glycerol and diffraction data were collected at the I03 beamline of Diamond Light Source.
Crystallographic analysis
Two copies of the co-complex structure of 3H4 Fab bound to HLA-E-VL9 were present in the asymmetric unit, a single copy constituted the focus of further discussion since root-mean-square deviation (RMSD) calculations from Cα-atom pairwise alignment of the two copies indicated minimal repositioning of interfacing residues at the HLA-E-3H4-binding site (Table 1b–f). Additionally, pairwise alignment with the previously published non-receptor-bound HLA-E coordinates (PDB ID: 1MHE)62 revealed minimal structural changes in HLA-E upon 3H4 engagement (Table 1c).
Diffraction data were merged and indexed in xia2 dials63. Outer shell reflections were excluded from further analysis to ensure the CC1/2 value exceeded the minimum threshold (>0.5) in each shell64. Sequential molecular replacement was carried out in MolRep of the CCP4i suite using molecule one of the previously published Mtb44-bound HLA-E structure with the peptide coordinates deleted (PDB ID: 6GH4) and one molecule of the previously published anti-APP-tag Fab structure (PDB ID: 6HGU) as phasing models65,66. Rigid body and retrained refinement were subsequently carried out by Phenix.refine67 in between manual model building in Coot68. Model geometry was validated by MolProbity69 and structural interpretation was conducted using the PyMOL Molecular Graphics System, version 2.0 (Schrödinger, LLC) in addition to the PDBePISA70 and PDBeFOLD71 servers.
Antigen-specific single B-cell sorting
HLA-E-VL9-specific human B cells were sorted in flow cytometry using a three-color sorting technique. Briefly, the stabilized HLA-E-β2M-peptide complexes were made as tetramers and conjugated with different fluorophores. Human pan-B cells, including naive and memory B cells, were isolated from PBMCs of healthy donors using human pan-B-cell enrichment kit (STEMCELL, Catalog# 19554). The isolated pan-B cells were then stained with IgM PerCp-Cy5.5 (Clone# G20-127, BD Biosciences, Catalog# 561285), IgD FITC (Clone# IA6-2, BD Biosciences, Catalog# 555778), CD3 PE-Cy5 (Clone# HIT3a, BD Biosciences, Catalog# 555341), CD235a PE-Cy5 (Clone# GA-R2, BD Biosciences, Catalog# 559944), CD10 PE-CF594 (Clone# HI10A, BD Biosciences, Catalog# 562396), CD27 PE-Cy7 (Clone# O323, eBioscience, Catalog# 25-0279), CD16 BV570 (Clone# 3G8, Biolegend, Catalog# 302035), CD14 BV605 (Clone# M5E2, Biolegend, Catalog# 301834), CD38 APC-AF700 (Clone# LS198-4-2, Beckman Coulter, Catalog# B23489), CD19 APC-Cy7 (Clone# LJ25C1, BD Biosciences, Catalog# 561743) and tetramers at 2 μg/million cells (including BV421-conjugated HLA-E-VL9 tetramer, PE-conjugated HLA-E-VL9 tetramer, APC-conjugated HLA-E-RL9SIV tetramer and APC-conjugated HLA-E-RL9HIV tetramer). The cells were then stained with a Fixable Aqua Dead Cell Stain Kit (Invitrogen, Catalog# L34957). HLA-E-VL9-specific B cells were sorted in BD FACSAria II flow cytometer (BD Biosciences) for viable CD3neg/CD14neg/CD16neg /CD235aneg/CD19pos/HLA-E-VL9double-pos/HLA-E-RL9HIVneg/HLA-E-RL9SIVneg subset as single cells in 96-well plates.
PCR amplification of human antibody genes
The VHDHJH and VLJL genes were amplified by reverse transcription polymerase chain reaction (RT-PCR) from the flow cytometry-sorted single B cells using the methods as described previously72,73 with modification. The PCR-amplified genes were then purified and sequenced with 10 μM forward and reverse primers. Sequences were analyzed by using the human library in Clonalyst for the VDJ arrangements of the immunoglobulin IGHV, IGKV, and IGLV sequences and mutation frequencies41. Clonal relatedness of VHDHJH and VLJL sequences was determined as previously described74.
Expression of VHDHJH and VLJL as full-length IgG recombinant mAbs
Transient transfection of recombinant mAbs was performed as previously described73. Briefly, purified PCR products were used for overlapping PCR to generate linear human antibody expression cassettes. The expression cassettes were transfected into 293i cells using ExpiFectamine (Thermo Fisher Scientific, Catalog# A14525). The supernatant samples containing recombinant antibodies were used for cell surface staining and HTS assay to measure the binding reactivities.
The selected human antibody genes were then synthesized and cloned (GenScript) in a human IgG1 backbone with 4A mutations75. Recombinant IgG mAbs were then produced in HEK293i suspension cells by transfection with ExpiFectamine and purified using Protein A resin. The purified mAbs were run in SDS-PAGE for Coomassie blue staining and western blot. Antibodies with aggregation were further purified in AKTA FPLC system using a Superdex 200 size-exclusion column.
Surface plasmon resonance (SPR)
Surface plasmon resonance assays were performed on a BIAcore 3000 instrument, and data analysis was performed with BIAevaluation 3.0 software as previously described76. Purified mAbs flowed over CM5 sensor chips at concentrations of 100 μg/ml, and antibody binding was monitored in real-time at 25 °C with a continuous flow of PBS at 30 μl/min. For SPR affinity measurements, antibody binding to HLA-E-VL9 complex protein was performed using a BIAcore S200 instrument (Cytiva, formerly GE Healthcare, DHVI BIA Core Facility, Durham, NC) in HBS-EP+ 1x running buffer. The antibodies were first captured onto CM5 sensor chip to a level of ~9000 RU. The HLA-E-VL9 soluble proteins were injected over the captured antibodies at a flow rate of 30 μl/min. After dissociation, the antibodies were regenerated using a 30 second pulse of Glycine pH 2.0. Results were analyzed using the Biacore S200 Evaluation software (Cytiva). Subsequent curve fitting analyses were performed using a 1:1 Langmuir model with a local Rmax. The reported binding curves are representative of two data sets.
Enzyme-linked immunosorbent assay (ELISA)
Direct binding ELISAs were conducted in 384-well ELISA plates coated with 2 μg/ml of C-trap-stabilized HLA-E-VL9, C-trap-stabilized HLA-E-RL9HIV or C-trap-stabilized HLA-E-RL9SIV in 0.1 M sodium bicarbonate overnight at 4 °C. Plates were washed with PBS + 0.05% Tween 20 and blocked with 3% BSA in PBS at room temperature for 1 h. MAb samples were incubated for 1 h in threefold serial dilutions starting at 100 μg/ml, followed by washing with PBS-0.05% Tween 20. HRP-conjugated goat anti-human IgG secondary Ab (SouthernBiotech, catalog# 2040-05) was diluted to 1:10,000 in 1% BSA in PBS-0.05% Tween 20 and incubated at room temperature for 1 h. For sandwich ELISA, 384-well ELISA plates were coated with HLA-E-VL9 antibodies in a threefold dilution starting from 100 μg/ml in 0.1 M sodium bicarbonate overnight at 4 °C. Plates were washed with PBS + 0.05% Tween 20 and blocked with 3% BSA in PBS at room temperature for 1 h. C-trap-stabilized HLA-E-VL9, C-trap-stabilized HLA-E-RL9HIV, C-trap-stabilized HLA-E-RL9SIV, or diluent control were then added at 2 μg/ml and incubated at room temperature for 1 h. After washing, HRP-conjugated anti-human β2M antibody (Biolegend, catalog# 280303) were added at 0.2 μg/ml and incubated at room temperature for 1 h. These plates were washed for four times and developed with tetramethylbenzidine substrate (SureBlue Reserve). The reaction was stopped with 1 M HCl, and optical density at 450 nm (OD450) was determined.
Antibody poly-reactivity assays
All mAbs isolated from mice and human were tested for ELISA binding to nine autoantigens—Sjogren’s syndrome antigen A (SSA), Sjogren’s syndrome antigen (SSB), Smith antigen (Sm), ribonucleoprotein (RNP), scleroderma 70 (Scl-70), Jo-1 antigen, double-stranded DNA (dsDNA), centromere B (Cent B), and histone as previously described77,78. Indirect immunofluorescence assay of mAbs binding to HEp-2 cells (Inverness Medical Professional Diagnostics, Princeton, NJ) was performed as previously described78,79. MAbs 2F580 and 17B81 were used as positive and negative controls, respectively. All antibodies were screened in two independent experiments.
Negative-stain electron microscopy of IgM antibodies
FPLC purified IgM antibodies were diluted to 0.08 mg/ml in HEPES-buffered saline (pH 7.4) + 5% glycerol, and stained with 2% uranyl formate. Images were obtained with a Philips EM420 electron microscope at 82,000 magnification and processed in Relion 3.0.
Peptide-pulsing in K562-E cells
K562-E cells and K562-E/UL49.5 cells were resuspended with fresh IMDM media with 10% FBS at 2 x 106 cells/ml. Peptides were added into cell suspension at a final concentration of 100 μM. The cell/peptide mixtures were incubated at 26 °C with 5% CO2 for 20–22 h and were transferred to 37 °C for 2 h with 5% CO2 before use. In the following mAb staining experiment, medium with 100 μM peptides was used to maintain peptide concentration.
NK cell cytotoxicity assay
NK cell cytotoxicity was measured by 51Cr release assay. A NKG2A-positive, CD16/CD32/CD64-negative NK-92 cells were used as effector cells in our study. Transfected or untransfected 293T cells were used as target cells. Target cells were counted, washed, resuspended in R10 at 1 × 107 cell/ml, and labeled with Na251CrO4 at 250 μCi/ml for 2 h at 37 °C. After washing three times using R10, cells were mixed with the testing antibody and effector cells in a final effector to target (E:T) ratio of 20:1 and 6:1 in triplicate wells in a flexible 96-well round bottom plates (PerkinElmer, Catalog# 1450-401). The plates were inserted in flexible 96-well plate cassettes (PerkinElmer, Catalog# 1450-101), sealed and incubated at 37 °C for 4 h. After the incubation, cells were pelleted by centrifugation, and from the top of the well, add 25 μl of supernatant to a rigid 96-well isoplates (PerkinElmer, Catalog#1450-514) containing 150 μl of Ultima Gold LSC Cocktail (Sigma, Catalog# L8286). The plates were inserted in rigid 96-well plate cassettes (PerkinElmer, Catalog# 1450-105), sealed and counted on Perkin Elmer Microbeta Triux 1450 counter. 51Cr labeled target cells without effector cells were set as a spontaneous release control, and 51Cr labeled target cells mixed with detergent (2% Triton X-100) were used as a maximum release control. The percentages of specific lysis were calculated with the formulation: The Percentages of Specific Lysis (51Cr Release%) = [(Experimental Release – Spontaneous Release)/(Maximum Release – Spontaneous Release)] × 100.
Development and screening of scFv libraries on the surface of yeast
A library was built that contained ~1.1 million 3H4 scFv variants with amino acid diversity at sites that were determined by structural analysis to interact with HLA-E-VL9. Seventeen residues (Supplementary Fig. 3) located in the CDR loops of 3H4 were randomized in groups of four based on their proximity and all the possible combinations of amino acids were sampled at these sites. Library DNA was synthesized on a BioXP 3250 (Codex) system and amplified with High Fidelity Phusion polymerase (New England Biolabs). PCR products were gel extracted (Qiagen Gel Extraction kit) to select full-length genes as per the manufacturer’s protocol. 3H4 scFv variants were displayed in library format on the surface of yeast as previously described82,83. Briefly, S. cerevisiae EBY100 cells were transformed by electroporation with a 3:1 ratio of 12 µg scFv library DNA and 4 µg pCTcon2 plasmid digested with BamHI, SalI, NheI (New England Biolabs). The size of the transformed library, determined by serial dilution on selective plates, was 5 x 107 individual colonies. Yeast Libraries were grown in SDCAA media (Teknova) supplemented with pen-strep at 30 °C and 225 rpm. Eighty percent of the sequences recovered from the transformed libraries were confirmed to contain full length, in-frame genes by Sanger sequencing (Genewiz). scFv expression on the surface of yeast was induced by culturing the libraries in SGCAA (Teknova) media at a density of 1 x 107 cells/ml for 24–36 h. Cells were washed twice in ice-cold PBSA (0.01 M sodium phosphate, pH 7.4, 0.137 M sodium chloride, 1 g/L bovine serum albumin) and labeled with APC-conjugated HLA-E-VL9 tetramer and 1:100 anti-c-myc:FITC (ICL) and incubated for 1 h at 4 °C. Initial selection was conducted with 50 mg/ml labeling concentration of HLA-E-VL9 tetramer; the second round of selection was done at 0.6 mg/ml tetramer. Cells were washed twice with PBSA after incubation with the fluorescently labeled probes and sorted on a BD FACS-DiVa. Double-positive cells for APC and FITC were collected and expanded in SDCAA media supplemented with pen-strep before successive rounds of enrichment. FACS data was analyzed with Flowjo_v10.7 software (Becton, Dickinson & Company). All clones selected by FACS were expanded, and their DNA was extracted (Zymo Research) for analysis by Sanger sequencing (Genewiz). scFv encoding plasmids were recovered from yeast cultures by yeast miniprep with the Zymoprep yeast plasmid miniprep II kit (Zymo Research). Isolated DNA was transformed into NEB5α strain of E. coli (New England Biolabs) and the DNA of individual bacterial colonies was isolated (Wizard Plus SV Minipreps, Promega) and analyzed by Sanger sequencing (Genewiz).
Statistics analysis and reproducibility
Data were plotted using Prism GraphPad 8.0 or visualized using the ComplexHeatmap R package. SAS 9.4 (SAS Institute, Cary, NC) was used to perform the statistical analysis with a P-value < 0.05 considered significant. For 51Cr release assays, mixed effects models were used to make comparisons of antibody to control using a random intercept for the triplicates run within each experiment and fixed effects of E:T ratio, type (antibody or control), and the interaction of E:T ratio by type. For human antibody gene usage analysis, chi-square test of independence was used to compare differences between groups. All attempts at replication were successful in 3–6 independent experiments, as indicated in the figure legends.
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
