Logic-gated antibody pairs that selectively act on cells co-expressing two antigens

Antibodies

Rituximab (MabThera) and obinutuzumab (Gazyvaro) were obtained from the pharmacy (UMC Utrecht). All other antibodies were recombinantly produced at Genmab as described³. Design of mutually dependent antibody mixtures was based on sampling previously described mutations combined with modifications at the shared binding interface for C1q and FcγRs within the IgG1 Fc domain^2,7. Details of the variant selection were described in patent WO 2019/211472, and will be published elsewhere. Mutations to enhance or inhibit Fc–Fc interactions and/or Fc–C1q binding interactions were introduced in expression vectors encoding the antibody heavy chain either using Quikchange technology (Agilent Technologies) or via gene synthesis (Thermo Fisher Scientific), at indicated positions numbered according to Eu nomenclature⁴³. Peptide tags encoding HA (‘YPYDVPDYA’), cMyc (‘EQKLISEEDL’) or FLAG (‘DYKDDDDK’) affinity tags were inserted by gene synthesis after the C-terminal cysteine of the light chain. Quality control of recombinant antibodies was performed by different methods as described previously⁷: capillary electrophoresis–sodium dodecyl sulfate on the Labchip GXII (Caliper Life Sciences/PerkinElmer Hopkinton) under reducing and nonreducing conditions (>90% intact IgG, >95% HC þ LC under reducing conditions), electrospray ionization–time-of-flight–mass spectrometry (Waters) or Orbitrap (Thermo Fisher Scientific), and high-performance–size-exclusion chromatography (aggregate level <5%; Waters Alliance 2975 separation unit, Waters). The following IgG1 antibodies targeting human antigens were used: CD52 (P31358) mAb Campath⁴⁴, CD20 (P11836) mAbs 11B8 and 7D8 (refs. ^45,46,47), CD3 (P07766) mAb huCLB3/4 (ref. ⁴⁸), CD37 (P11049) mAb IgG1-37.3 (ref. ⁴⁹), DR5 (O14763) mAbs DR5-01 and DR5-05 (ref. ³). mAb IgG1-b12 targeting HIV-1 antigen gp120 (Q9IZE4) was used as a nonbinding isotype control⁵⁰.

Cells and reagents

Daudi (human B cell lymphoma), Raji (human B cell lymphoma), Ramos (human B cell lymphoma), COLO-205 (colorectal cancer) and BxPC-3 (pancreatic cancer) cell lines were obtained from the American Type Culture Collection (nos. CCL-23, CCL-86, CRL-1596, CCL-222 and CRL-1687, respectively). The human B lymphoma cell line U-698-M and the human B precursor leukemia cell line REH were obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (cell line numbers ACC 22 and ACC 4, respectively; Braunschweig, Germany). Wien-133 cells (human Burkitt’s lymphoma) were kindly provided by G. Hale (BioAnaLab Limited).

PBMCs derived from patients with CLL were commercially obtained from Discovery Life Sciences. Characteristics of patients with CLL are summarized in Supplementary Table 9. Buffy coats from healthy human donors and complement-competent, pooled NHS (AB positive) were obtained from Sanquin. Whole blood samples from healthy human volunteers were freshly obtained from the University Medical Center Utrecht. Commercially available patient-derived PBMCs and healthy donor blood(-derived) samples were collected at the site of the vendor after patients provided their written and informed consent in accordance with the declaration of Helsinki. All vendors maintained strict ethical compliance, including fully deidentified materials and stringent Institutional Review Board and Ethics Committee compliance. Purified C1q protein and C1 complex were obtained from Quidel and Complement Technology, respectively.

Whole blood cytotoxicity

Cytotoxicity assays were performed with healthy human donor blood samples that were hirudin anticoagulated, or EDTA anticoagulated and recalcified using 5 mM CaCl₂ (Sigma Aldrich) for 30 min in the presence of 10 µg ml⁻¹ hirudin (Genscript) to preserve or restore complement activity, respectively. While hirudin anticoagulated blood samples were preferred for this study due to optimal preservation of complement activity and enabling ADCC activity, availability of such samples was limited due to discontinuation of the product. Recalcified EDTA anticoagulated blood samples were used as an alternative whole blood source, albeit with limited incubation times (<4 h) to prevent blood clotting. For the hirudin anticoagulated blood samples analyzed in Supplementary Fig. 2 and in Fig. 3, whole blood was incubated with antibodies for 18 h (to enable detection of both CDC and ADCC/ADCP-mediated depletion of cell subsets) at 37 °C/5% CO₂. After 18 h incubation, red blood cells were lysed first and next samples were stained as described above, before measuring on an LSR Fortessa (BD Biosciences) flow cytometer. For the recalcified, EDTA anticoagulated blood samples used in Fig. 6 and Supplementary Fig. 6, whole blood was incubated with antibodies for 45 min (to enable detection of CDC) at 37 °C/5% CO₂. After 45 min of incubation, samples were stained for 30 min at 4 °C with fluorochrome-labeled lineage-specific antibodies and fixable viability stain (FVS-BV510; BD Biosciences) to characterize cell subsets and dead or dying cells, respectively. Next, red blood cells were lysed (lysis buffer: 10 mM KHCO₃, 0.01 mM EDTA and 155 mM NH₄Cl) and cells were optionally fixed in 2% PFA before measuring on a flow cytometer. Cell markers used to define cell populations were: CD45-PerCP (Biolegend), CD66b-PE-Cy7 (Biolegend), CD3-eFluor450, CD4-APC-eFluor780 (eBioscience) and CD19-BV711 (Biolegend). For Fig. 6 and Supplementary Fig. 6, additional cell markers included were: CD16-BV605 (Biolegend), CD56-PE-CF594 (BD Biosciences) and CD8-AF700 (Biolegend). The gating strategy used to define cell populations is described in Supplementary Fig. 7a. Cytotoxicity was calculated as the fraction (%) of cells remaining after treatment relative to a nontreated control sample (100%).

C1q binding

Wien-133 cells were opsonized with antibody serial dilutions for 15 min at 37 °C. Subsequently, cells and antibodies were put on ice, purified human complement component C1q (2.5 µg ml⁻¹) was added and incubated for 45 min. After washing, C1q binding was detected using a fluorescein isothiocyanate (FITC) conjugated rabbit antihuman C1q antibody (Dako) and quantified as the FITC geometric mean fluorescent intensity (gMFI) determined using an iQue Screener flow cytometer (Sartorius).

CDC

CDC assays were performed using tumor cells incubated with antibody concentration series or a fixed antibody concentration as indicated, for 45 min at 37 °C in the presence of NHS (20% final concentration) as source of complement. Killing was calculated as the fraction of propidium iodide (PI)-positive cells (%) determined by an iQue Screener flow cytometer for tumor B cell lines, and as the fraction of TO-PRO-3-positive cells (%) determined by an LSR Fortessa flow cytometer for CD19⁺ CLL B cells.

FcγR binding

Binding of antibody variants to the monomeric extracellular domain (ECD) of FcγRIA (FCGR1AECDHis) and to dimeric ECDs of FcγRIIA allotype 131H (diFCGR2AH-HisBAP), FcγRIIA allotype 131R (diFCGR2AR-HisBAP), FcγRIIIA allotype 158F (diFCGR3AF-HisBAP) and FcγRIIIA allotype 158V (diFCGR3AV-HisBAP) was tested in enzyme-linked immunosorbent assays (ELISAs)⁵¹. For binding to dimeric FcγR variants, 100 µl of goat antihuman F(ab’)₂ (1 μg ml⁻¹) was added per well for coating overnight at 4 °C. After washing the plates, nonspecific binding was blocked for 1 h at room temperature by adding 200 µl per well PBS/0.2% BSA. With washings in between incubations, plates were sequentially incubated with 100 µl of 20 µg ml⁻¹ antibody variants in PBS with Tween with 0.2% BSA buffer for 1 h at room temperature, 100 µl of the recombinant dimeric FcγR constructs (1 µg ml⁻¹) for 1 h at room temperature and 100 µl of streptavidin-labeled Poly-HRP (1:10,000) for 30 min at room temperature. Development was performed for 10–30 min with 1 mg ml⁻¹ ABTS (Roche). To stop the reactions, 100 µl per well of 2% oxalic acid was added. Absorbance was measured at 405 nm in a microplate reader (BioTek) using BioTek Gen5 V1.04.5 software. To detect binding to monomeric FcγRIa, plates were coated with monomeric His-tagged FCGR1A ECD and after antibody incubation, goat antihuman-kappaLC-HRP (1:5,000) was used as detection antibody.

FcγR activation

Activation of FcγRIIa- (allotype H-131) and FcγRIIIa-mediated (allotype V-158) intracellular signaling was quantified using Luminescent Reporter Bioassays (Promega), according to the manufacturer’s recommendations.

ADCC

ADCC was assessed using a DELFIA EuTDA Cytotoxicity assay (PerkinElmer) according to the manufacturer’s recommendations. Briefly, Wien-133 target cells were loaded with BATDA reagent, and 1 × 10⁴ cells were incubated with antibody serial dilutions and human healthy donor PBMCs (isolated from buffy coats through centrifugation using Leucosep tubes according to the manufacturer’s instructions; Greiner Bio-One, catalog no. 227288) as effector cells, at a 100:1 effector to target ratio, for 2 h at 37 °C in a total volume of 160 µl. After incubation and centrifugation, 20 µl of supernatant was transferred to a 96-well plate, 200 µl of Europium Solution was added and the mixture was incubated for 15 min at room temperature while shaking. EuTDA release and time-resolved fluorescence was measured on an EnVision Multilabel Reader (PerkinElmer). Maximal and spontaneous release were determined using target cells incubated with 0.1% Triton X-100 or target cells in medium without effector cells, respectively. Specific release was calculated as:

ADCP

ADCP assays were performed as described in ref. ⁵². In short, human CD14⁺ monocytes were obtained from healthy donor PBMCs (isolated from buffy coats through centrifugation using Leucosep tubes according to the manufacturer’s instructions) through positive isolation using CD14 MicroBeads (Miltenyi Biotec) according to the manufacturer’s instructions. Monocytes were cultured in culture medium (CellGenix GMP DC serum-free medium with 50 ng ml⁻¹ M-CSF) in Nunc dishes with UpCell surface (Thermo Fisher Scientific) at 37 °C/5% CO₂ for 7–8 days to obtain human monocyte-derived macrophages (h-MDM). h-MDMs were characterized by flow cytometry for expression of myeloid- and macrophage-specific maturation markers (Supplementary Table 1). ADCP was determined using Raji cells labeled with calcein AM (Life Technologies) according to the manufacturer’s instructions and opsonized with antibodies for 15 min at 37 °C. h-MDM were added at an effector to target (E:T) ratios of 2:1 and incubated for 4 h at 37 °C/5% CO₂. After incubation, tumor cells and h-MDM were stained for surface markers using fluorochrome-conjugated antibodies (Supplementary Table 2) for 30 min at 4 °C and analyzed on an LSR Fortessa flow cytometer. The gating strategy used to define cell populations is described in Supplementary Fig. 7b. ADCP was calculated as the fraction of CD11b⁺/calcein AM⁺/CD19⁻ cells within the total h-MDM (CD11b⁺) cell population.

Target expression

Expression levels of cellular markers were determined using an indirect immunofluorescence assay (QIFIKIT, Agilent Technologies and Human IgG Calibrator kit, BioCytex) according to the manufacturer’s instructions.

PBMC activation

PBMC CDC assays were performed as follows. Human PBMCs from eight healthy donors (isolated from buffy coats through centrifugation using Leucosep tubes according to the manufacturer’s instructions) were incubated with antibodies in culture medium (Iscove’s Modified Dulbecco’s Medium (IMDM) supplemented with 1 U ml⁻¹ penicillin and 1 µg ml⁻¹ streptomycin, Lonza) and 20% (final concentration) pooled NHS (Sanquin) as a source of complement, in three individual samples per antibody. After 20–24 h incubation at 37 °C/5% CO₂, culture supernatant was collected and stored at −20 °C until further analysis. Cytokine quantitation was performed on the culture supernatants using the U-PLEX Proinflam Combo 1 Human kit and Discovery Workbench software v.4.0 (Meso Scale Diagnostics), according to the manufacturer’s instructions. Two donors were excluded based on a high induction of cytokine production by isotype controls.

The gating strategy used to define cell populations was essentially the same as described for whole blood cytotoxicity analysis, and is detailed further in Supplementary Fig. 7. In addition, PBMCs were stained for cell surface markers CD25 (CD25-PE, eBioscience) and CD69 (CD69-FITC, BD Biosciences). Staining was performed for 30 min at 4 °C and analyzed on an LSR Fortessa X20 flow cytometer (BD Biosciences).

FRET

Proximity-induced FRET was determined by measuring energy transfer between AF555-conjugated donor and AF647-conjugated acceptor antibodies incubated with cells^9,26. Briefly, purified B cells (isolated from buffy coats using Dynal Dynabeads Untouched Human B cell isolation kit (Life Technologies) according to the manufacturer’s instructions) were incubated with AF555-conjugated donor mAbs and/or AF647-conjugated acceptor mAbs in the presence or absence of purified human C1q (Quidel, 2.5 µg ml⁻¹) or C1 (Complement Technology, 2.42 µg ml⁻¹). gMFI values were measured using an LSR Fortessa flow cytometer by recording events at 585/42 nm (FL2, donor AF488) and ≥670 nm (FL3), both excited at 488 and at 660/20 nm (FL4, acceptor AF647), excited at 635 nm. Unquenched donor fluorescence intensity was determined with cells incubated with AF555-conjugated donor mAbs, and nonenhanced acceptor intensity was determined with cells incubated with AF647-conjugated acceptor mAbs. Proximity-induced FRET was determined by measuring energy transfer between cells incubated with AF555-conjugated donor and AF647-conjugated acceptor mAbs. gMFI values allowed calculation of FRET according to the following equation:

where a is FL2(D)/FL3(D), b is FL4(A)/FL3(A), D is donor, A is acceptor and FLn (D, A) are donor + acceptor. ET values obtained were normalized:

Viability assay

Cell viability was determined using a CellTiter-Glo luminescent cell viability assay, according to the supplier’s protocol (Promega). Cells were seeded in white OptiPlates (PerkinElmer) and allowed to adhere overnight at 37 °C. The following day, antibody serial dilutions and purified human C1q (Complement Technology, 2.5 µg ml⁻¹) were added and incubated for 3 days at 37 °C. Then 5 µM staurosporine (Sigma no. S6942) treated cells and untreated cells were included as positive and negative controls of cell death induction, respectively. After incubation, Luciferin Solution Reagent was added and plates were incubated for 1.5 h at 37 °C. Luminescence was measured on an EnVision Multilabel Reader. The percentage of viable cells was calculated using the following formula:

where T is luminescence of the test sample, P is luminescence of staurosporine control sample and V is the luminescence of the medium control sample.

Human FcRn binding

Kinetic rate and affinity constants were determined using a Biacore 8K SPR system (Cytiva) equipped with a Biacore Series S Sensorchip CM5 (Cytiva catalog no. 29104988) at 25 °C in running buffer of PBS-P+ (10×, Cytiva catalog no. 28995084; diluted and pH adjusted to a final composition of 20 mM phosphate pH 6.0, 137 mM NaCl, 2.7 mM KCl, 0.05% Tween-20). Briefly, anti-His-tag antibody (Cytiva catalog no. 28995056) was covalently linked in all flow cells by NHS/EDC coupling (Cytiva, catalog no. BR100050) according to supplier instructions. After activation, anti-His-tag antibody (50 μg ml⁻¹) in 10 mM sodium acetate pH 4.5 (His Capture Kit) was coupled for 420 s at a flow rate of 10 µl min⁻¹. After a minimum of 1 h of stabilization in running buffer, 5 nM His-tagged FcRn (Sino Biological, catalog no. CT071-H27H-B) in running buffer was captured for 60 s at 10 µl min⁻¹ to capture levels of 45–60 resonance units only in active flow cells, while running buffer was injected in reference flow cells. A twofold, eight-step dilution series of indicated IgG1 molecules (75 nM to 0.29 nM) was prepared in running buffer. IgG dilution series were injected in reverse order in both flow cells for 120 s at 30 μl min⁻¹, with a 300 s dissociation phase at 30 μl min⁻¹. Samples were injected in a multi-cycle manner over freshly captured FcRn per cycle, by a single regeneration of capture surfaces with 10 mM glycine, pH 1.5 (Cytiva catalog no. BR100354), for 30 s at 30 μl min⁻¹. Three independent blank cycles were injected per interaction at the start of the assay, the third of which was used during double referencing. Data were processed and analyzed using Biacore 8K Evaluation Software (v.3.0.12) (GE Healthcare Bio-Sciences Corporation). Responses in the reference cell were subtracted from the active cell to yield reference-subtracted data. A blank sensorgram (recorded in the active cell before injection of IgG) was subtracted from reference-subtracted data to yield double-referenced data. Double-referenced data were fit to a 1:1 binding model to determine the apparent association (k_a) and dissociation (k_d) rate constants. The apparent equilibrium dissociation constant, or affinity constant, K_D, was calculated as K_D = k_d/k_a.

Animals

Animal experiments were performed in compliance with the Dutch animal protection law (WoD) translated from the directives (2010/63/EU) and if applicable, the Code of Practice ‘animal experiments for cancer research’ (Inspection V&W, Zutphen, Netherlands, 1999) and were approved by the Dutch Central Committee for animal experiments and by the local ethical committee. Animals were housed and handled in accordance with good animal practice as defined by the Federation of European Laboratory Animal Science Associations, in an association for assessment and accreditation of laboratory animal care and an ISO 9001:2000 accredited animal facility (GDL, Utrecht, Netherlands).

Pharmacokinetic analysis

Pharmacokinetic studies were performed using 11–12-week-old female tumor-free C.B-17/IcrHan Hsd-Prkdcscid mice (SCID, Envigo). Mice were injected intravenously (IV) with a single dose of 500 µg of each test reagent per mouse (n = 3). Blood samples were drawn from the saphenous vein at 10 min, 4 h and 1, 2, 7, 14 and 21 days after antibody administration and collected into heparin-containing vials. Vials were centrifuged (10 min at 14,000g) to separate plasma from cells and plasma was stored at −20 °C until further use. Total human IgG concentration in plasma samples was analyzed by ELISA. Plates were coated overnight at 4 °C with 2 μg ml⁻¹ of an in-house generated antihuman IgG directed mouse-IgG2a recombinant Fab fragment in PBS, and plasma human IgG was detected by a peroxidase-conjugated AffiniPure goat antihuman IgG Fcγ-specific antibody (Jackson, catalog no. 109-035-098). Absorbance was measured at 405 nm in a microplate reader (BioTek) and processed using BioTek Gen5 V1.04.5 software. Area under the curve (AUC) up to day 21 was determined using GraphPad Prism and clearance was calculated as (Dose (mg kg⁻¹) × 1,000/AUC).

In vivo proof of concept studies

Human PBMCs were isolated from buffy coats obtained from healthy donors, as described above, and were frozen overnight at −80 °C and stored in liquid nitrogen after 24 h. On thawing, monocytes and Natural Killer cells were depleted using CD14 and CD56 MicroBeads (Miltenyi, catalog no. 130-050-201 9CD14) and 130-050-401 (CD56)) according to the manufacturer’s protocol. One day before the study, NSG mice (NOD.C.B-17-Prkdc scid/J mice; Charles River Laboratories; female; age 7–8 weeks) were injected intraperitoneally with 8 mg IVIg (Octagam, Pharmacy of Veterinary Medicine, Utrecht University, catalog no. 5430000803496). NSG mice were injected intravenously with antibody mixtures (combined dose of both antibody components was 1.0 mg kg⁻¹) and immediately thereafter 8 × 10⁶ human T and B cells from three separate donors were injected intraperitoneally. The next day, mice were euthanized and the human cells were recovered by peritoneal lavage. The number of human cells was analyzed by hCD45(⁺) staining, T cells were identified as hCD3(⁺)/hCD2(⁺), B cells as hCD19(+)/hCD22(+), and the number of cells recovered was expressed relative to a fixed number of beads added at the start of the staining. Deposited C3 was detected by a directly labeled mAb against mouse C3 (details in Supplementary Table 5). Significance of comparisons between log-transformed cell counts or mean fluorescence intensities of C3 deposition were assessed by one-way analysis of variance after Welch correction for nonequal s.d., followed by Dunnett’s T3 multiple comparisons test.

Human pluripotent stem cell (HPSC) -humanized NSG (NSG-HIS) mice were purchased from Jackson (female, age 16–18 weeks after quality control of immune reconstitution) from three HPSC donors. Humanization levels were confirmed on arrival by flow cytometry (Supplementary Table 8), and randomization was performed based on human CD45(⁺) counts and the donors. One day before the study all animals were injected intraperitoneally with 8 mg IVIg (Octagam). NSG-HIS mice were treated intravenously with antibody mixtures at a combined dose of 0.5 mg kg⁻¹. Mice were euthanized the next day. Blood samples were incubated with fluorescently labeled antibodies (Supplementary Table 8) and red blood cell Lysis buffer (BioLegend, catalog no. 420302). Peritoneal cells were also incubated with fluorescently labeled antibodies after washing in PBS. CountBright Beads (Invitrogen, catalog no. C36950) were added to blood and peritoneal cell samples to determine the number of human cells and C3 fixation by flow cytometry on a FACS Fortessa (BD). Significance of comparisons between log-transformed cell counts or mean fluorescence intensities of C3 deposition were assessed by one-way analysis of variance after Welch correction for nonequal s.d.s, followed by Dunnett’s T3 multiple comparisons test.

All fluorescently labeled antibodies specific for the cell markers used to define cell populations are listed in Supplementary Table 8. The gating strategy used to define cell populations is described in Supplementary Fig. 7.

Super-resolution microscopy sample preparation

Wien-133 suspension cells were cultured in IMDM supplemented with 10% heat-inactivated fetal bovine serum, 5 U ml⁻¹ penicillin, 0.05 mg ml⁻¹ streptomycin and 2 mM l-glutamine at 37 °C and 5% carbon dioxide. Washes described below were performed by pelleting cells at 0.1 RCF using an Eppendorf centrifuge (model no. 5415R) followed by buffer exchange. Cells were washed once with warm IMDM and aliquoted into 1.5 ml Eppendorf tubes at 4 × 10⁶ cells per treatment group. Cells were incubated with 5 µg ml⁻¹ of total IgG for each antibody combination plus 5 µg ml⁻¹ C1q in IMDM for 15 min at 37 °C. After antibody opsonization, cells were washed once with PBS and fixed with 4% paraformaldehyde (PFA) plus 0.2% glutaraldehyde in PBS at room temperature for 2 h. Fixation was halted by removing the PFA/GA fixative via centrifugation, followed by two washes with 10 mM Tris and then blocked for 15 min with 5% BSA/PBS. CD52- and CD20-directed antibody variants used for super-resolution imaging contained an HA-tag at the C terminus of the light chain, allowing for detection using AF647-labeled anti-HA antibody (Novus Biological NB600-363AF647). Cells were incubated with anti-HA-AF647 at 8 µg ml⁻¹ in 2% BSA/PBS for 1 h at room temperature. Labeled cells were washed with PBS and postfixed with 4% PFA for 10 min at room temperature, washed twice with 10 mM Tris and suspended in PBS until imaging.

Super-resolution microscopy image acquisition

For dSTORM imaging, 25 mm glass coverslips (no. 1.5, Warner Instruments, no. CS-25R15) were piranha etched (96% H₂SO₄ + 30% H₂O₂) and treated with 0.1% NABH₄ for 5 min to quench background fluorescence. Coverslips were dried and glow discharged before incubation with 0.01% poly-l-lysine solution plus 0.01% poly-d-lysine hydrobromide (Sigma nos. P4707, P1149) for 3 h at room temperature. Coverslips were rinsed with water and dried before storage at 4 °C. Poly-l/d-lysine coated coverslips were secured into custom-made imaging chambers and rehydrated with PBS for 10 min at room temperature. After removal of PBS used for coverslip hydration, 4 × 10⁶ Wien-133 cells in 1 ml of PBS were allowed to settle onto the coverslip for 8 h at room temperature. Cells were gently washed twice with PBS before addition of 1.5 ml of fresh dSTORM imaging buffer (50 mM Tris, 10 mM NaCl, 10% w/v glucose, 168.8 U ml⁻¹ glucose oxidase (Sigma no. G2133), 1,404 U ml⁻¹ catalase (Sigma no. C9322) and 60 mM 2-aminoethanethiol (MEA), pH 8.0). The chamber was sealed by securing a clean 25 mm coverslip over the top.

dSTORM imaging was performed using a custom-built microscope equipped with a 1.35 NA silicon oil immersion lens (UPLSAPO100XS, Olympus) and an sCMOS camera (C11440-22CU, Hamamatsu). Excitation was with a 647 nm fiber laser (2RU-VFL-P-500-647-B1R, MPB Communications). Emission light was collected with a 708/75 nm band pass filter (FF01-708/75-25, Semrock). Brightfield registration⁵³ was performed before each sequence using a 660 nm LED (M660L3, Thorlabs) illumination lamp and a 3D piezo sample stage (MAX341/M, Thorlabs). A total of 100,000 frames were collected for each cell (20 sequences of 5,000 frames each) at 20 frames per second and approximately 50 mW of excitation laser power.

Super-resolution microscopy image reconstruction and data analysis

dSTORM images were analyzed and reconstructed with custom-built functions in MATLAB R2019b software (MathLinks). For each image frame, subregions were selected based on local maximum intensity. Each subregion was then fitted to a finite pixel Gaussian intensity distribution using a maximum likelihood estimator⁵⁴. Fitted results were accepted or rejected based on log-likelihood ratio, the fit precision that was estimated using the Cramér–Rao lower bound values for each parameter, as well as intensity and background cutoffs⁵⁵ between localizations in close spatial-temporal proximity that were possibly from the same blinking event were connected if the null hypothesis that they originated from a fluorophore at the same location was not rejected at a level of significance of 0.01.

Localizations were used to make estimates of the underlying number and locations of the fluorescence emitters using the Bayesian grouping of localizations (BaGoL) algorithm⁵⁶. We carried out BaGoL analysis using the MATLAB implementation of BaGoL included with smite (https://github.com/LidkeLab/smite/) v.0.1.0 at the UNM Center for High-Performance Computing using MATLAB R2019b. The locations and uncertainty given by the maximum a posteriori number of emitters output of the BaGoL algorithm were used for further analysis (Supplementary Fig. 4). The well-separated fluorophores in the sparsely labeled IgG1-b12-RGE-HA samples were used to estimate the blinking statistics for the anti-HA-AF647 probe. From the blinking statistics of individual antibodies, we found an optimal precision inflation (SE_adjust) parameter of 9 nm. Blinking statistics were modeled with an exponential distribution with mean value of 5.131 and 4.333 blinking events per antibody probe for the two lots of anti-HA-AF647.

ROI of size 2 × 2 μm were selected from the set of images, from which statistics for the number of localizations and nearest neighbor distances were collected per ROI. The nearest neighbor distance in each ROI was computed with locally written MATLAB R2019Bb software using ‘knnsearch’, producing the distance from each localization to its nearest neighbor, and so inferring an average localization density.

Nonmicroscopy data processing

LSR Fortesssa flow cytometry data were processed using FACSDiva v.8 and v.9.0 software (BD). Sartorius iQue flow cytometry data were processed using iQue ForeCyt v.6.2 and v.8.1 software (Sartorius). Flow cytometry data were analyzed using FlowJo V10 software (BD). Graphs were plotted and analyzed using GraphPad Prism v.8.0 (DotMatics). EnVision data were processed using EnVision Workstation 1.13.3009.1401 software (Perkin Elmer). Graphs were plotted and analyzed using GraphPad Prism v.8.0. Dose-response curves were generated using best-fit values of nonlinear dose-response fits using log-transformed concentrations. All data shown are representative of at least three independent replicate experiments or three individual human donors tested. The mean AUC ± standard deviation (s.d.) was determined for all available dose-response analyses and is summarized in Supplementary Table 3. Dose-response data from multiple experimental repeats were pooled, concentrations were log-transformed and the resulting AUC values were normalized relative to the positive control indicated (100%) and negative control nonbinding antibody IgG1-b12 (0%).

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.