Twenty-five-millimeter #1.5 coverslips (Warner Instruments) were cleaned and coated with fibronectin-like engineered protein [25 μg/ml in phosphate buffer saline (PBS), pH 7.4, Sigma] as described before (32).

MDA-MB-468 and SK-BR-3 cells (originally obtained from the American Type Culture Collection, ATCC) were cultured in Phenol Red-free Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum, 1 mM sodium pyruvate, 100 units/ml penicillin, 100 units/ml streptomycin, and 2 mM L-alanyl-L-glutamine. MDA-MB-453 cells (also from ATCC) were cultured in DMEM-F12 media supplemented with 10% fetal bovine serum, 0.5 mM sodium pyruvate 100 units/ml penicillin and 100 units/ml streptomycin. BT-474 cells were cultured as described previously (32).

A perbB2-EGFP pcDNA3.1(+) plasmid was purchased from Adgene (plasmid # 39321). This plasmid was used for iterative site directed mutagenesis to mutate the EGFP coding region into paGFP. The following primer pairs were used for iterative site directed mutagenesis to alter five amino acids in the EGFP coding region of the plasmid:

Primer pair 1

5’ CCCACCCTCGTGACCACCTTTAGTTACGGCGTGCAG TGCTTC 3’

5’ GAAGCACTGCACGCCGTAACTAAAGGTGGTCACGAG GGTGGG 3’

Primer pair 2

5’ GAACGGCATCAAGGCGAACTTCAAGATCC 3’

5’ GGATCTTGAAGTTCGCCTTGATGCCGTTC 3’

Primer pair 3

5’ GACAACCACTACCTGAGCCATCAGTCCAAACTGAG CAAAG 3’

5’ CTTTGCTCAGTTTGGACTGATGGCTCAGGTAGTGG TTGTC 3’

After each round of site directed mutagenesis, the DNA was transformed into BP5alpha competent cells (Biopioneer) and plated on LB/ampicillin agar plates at 37°C. Single colonies were selected from the plate and amplified in 10 ml LB culture to purify plasmid DNA to be used for the next round of site directed mutagenesis. After the final round of SDM was complete, the plasmid was sent for Sanger sequencing at the City of Hope Integrative Genomics Core.

Pertuzumab and trastuzumab (Genetech) were clinical grade. Trastuzumab memAb I83E (referred to here as trastuzumab memAb, or TmemAb) and wild type trastuzumab (for ADCC experiments) were prepared similarly as described before (34).

The Fc used for control studies was obtained from a papain digest of clinical trastuzumab. The Fc was purified by application of the papain digest material to first a protein L column (GE, 5 ml) with collection of the flow through. The flow through material was concentrated and further purified by gel filtration chromatography (GE, superdex 10/300 GL). The fractions were concentrated and stored at −80°C.

Fab trastuzumab was generated through the digestion of the clinical trastuzumab with immobilized papain (Pierce) and purified by reverse purification with protein A (GE Healthcare) and SEC on a HiLoad 16/600 pg Superdex 75 column (GE Healthcare).

MemAbs were labeled with Alexa Fluor 647 (AF647) dyes presenting an N-hydroxysuccinidimidyl ester (NHS) group for protein conjugation. A solution with 4–6 x excess of dye dissolved in dimethyl sulfoxide was mixed with a solution of 1 mg/ml of TmemAb in PBS, pH 7.4, and 0.02 M NaHCO₃. The conjugation reaction solution was placed on a rotator for 30 min at room temperature and quenched with 1.5 M hydroxylamine (pH 8.5) for 10 min. Unconjugated dye was removed by passing the solution through a size exclusion chromatography column (Bio-Rad, Hercules, CA, USA) while any potential aggregates were removed by passing labeled Ab through a 300-kDa concentrator. Measurements from a NanoDrop 1000 (Thermo) were used to calculate (with respect to the dye correction factor) the final concentration and degree of labeling for each fluorescent antibody. Approximately one dye per antibody (degree of labeling ∼1) was obtained in all cases for antibodies labeled with AF647. These conditions are desired as an increased degree of labeling has been reported to decrease affinity for trastuzumab (39).

M2Fc contains the meditope sequence fused to the N-terminus of the Fc using a flexible 37 amino acid sequence (Supplementary methods). For the M4Fc moiety, the meditope sequence is placed at both N and C-termini using flexible 39 and 30 amino acid linkers respectively (Supplementary methods). Each were produced in insect cells. Specifically, baculovirus encoding each were produced in TNI insect cells according to the manufacturer’s protocol by transfecting M2Fc or M4Fc expression vectors and propagating the virus three times (Expression Systems). TNI cells were seeded in ESF921 media at a density of 1e6 cells/ml and allowed to grow overnight. The following day, high titer baculovirus containing the M2Fc or M4Fc expression vectors was added to the TNI cells at a MOI of 50 and the cells were allowed to produce protein. After three days, cells were separated from the media by centrifugation and media containing M2Fc or M4Fc was placed over a protein A column for purification. M2Fc or M4Fc in PBS was further purified using fast protein liquid chromatography.

Antibody dependent cell-mediated cytotoxicity assays were conducted using a luciferase reporter-based core kit according to manufacturer instructions (G7010, Promega). SK-BR-3 cells were used as the target cell line and seeded at 5,000 cells per well in a 96-well white, tissue-culture treated plate in DMEM with 10% FBS one day prior to conducting the ADCC assay. The next day, media was replaced with 4% low IgG serum in RPMI media, according to kit protocol. Dilutions of listed antibodies were made at 1:2 starting at 1 nM antibody in the presence or absence of 10 nM M4Fc and added to cells along with the effector cells provided. Antibodies at 1 nM, with or without M4Fc, were also added to wells containing no target or effector cells as a negative control. Cells were incubated at 37^°C for 6 h prior to adding luciferase substrate. Luminescence was measured on a Synergy 4 multi-mode microplate reader (BioTek) with a 0.5 s integration time. Each experiment was repeated at least three times with three technical replicates per experiment.

MDA-MB-468 cells were transiently transfected 48 h after plating on coverslips using Jet Prime (PolyPlus) according to the manufacturer’s instructions. For steady-state measurements, cells were washed with PBS at 37°C and fixed with 4% (w/v) paraformaldehyde and 0.2% (w/v) glutaraldehyde (EMS, Cat# 157-8 and 16019, respectively, Hatfield, MA, USA) in PBS for 30 min at room temperature; fixation was quenched with 25 mM glycine for 10 min as described before (32).

For Ab/Ab fragment treatments, 24 h after HER2-paGFP transfection, MDA-MB-468 cells were washed with media and treated with: 20 nM Fab trastuzumab, 10 nM trastuzumab, 10 nM pertuzumab, or 10 nM trastuzumab combined with 10 nM pertuzumab in media for 10 min. Cells were washed again with warm media and fixed/quenched as described above. For multivalent meditope treatment, MDA-MB-468 cells transfected with HER2-paGFP were first incubated with meditope-enabled trastuzumab memAb for 10 min. After a quick warm media wash, cells were incubated with 10 nM Fc, 10 nM M2Fc, 20 nM M2Fc, or 10 nM M4Fc in media for indicated times and fixed. To assess the effect of multivalent meditopes on endogenous HER2, BT-474 cells were incubated with 10 nM TmemAb-AF647, alone or premixed with 10 nM multivalent meditope (i.e., M2Fc and its variants or M4Fc), for 10 min at 37°C and subsequently fixed as described above. Alternatively, cells were stained with 10 nM TmemAb-AF647 postfixation as indicated. All incubations were performed at 37°C in the cell culture incubator.

About 0.1 μm TetraSpeck beads (Life Technologies) served as fiducial markers in all experiments to correct any lateral drift as described before (40). Coverslips in Attofluor cell chambers (Life Technologies) were imaged immediately after preparation in PBS (to detect paGFP using photoactivated localization microscopy, PALM) or direct stochastic optical reconstruction microscopy (dSTORM) imaging buffer (41) (to detect AF647, using dSTORM).

PALM and dSTORM imaging were performed using a 3D N-STORM super-resolution microscope system (32). Data was acquired using NIS Elements 4.3 software and ANDOR Solis 4.23. Images of 256 × 256 pixels (27 μm × 27 μm for PALM and 41 μm × 41 μm for dSTORM) were collected with a frame rate of 100 ms (PALM) and 10 ms (dSTORM). paGFP, which is a monomeric optical highlighter protein with good signal to noise ratio, was activated and excited with a 488 nm laser with power values ranging from 1.5 to 2 mW (measured out of the optical fiber) and a 505/15 emission filter. paGFP molecules were imaged until the signal was completely exhausted—typically 15,000–25,000 frames were acquired. For fluorescently labeled trastuzumab memAb (TmemAb-AF647), dSTORM imaging was performed similarly as before (32), with a laser power of 120 mW (measured out of the optical fiber) and acquiring 20,000–40,000 frames until the AF647 signal was exhausted. Additionally, NIS-Elements software was used with the following identification settings to capture positive signal and produce localization data for analysis: 700 as the minimum number of photons/localization, 200 nm minimum localization width, 400 nm maximum localization width, 300 nm initial fit width, 1.3 maximum axial ratio, and 1 pixel maximum displacement.

We characterized the blinking behavior of paGFP and TmemAb-AF647. As described previously (31, 32), the average number of localizations was approximately 5 for paGFP, whereas the average number of localizations for TmemAb-AF647 was approximately 3. These values represent the average number of localizations (discrete appearances α) of the fluorescent probe for a given set of imaging conditions and particular optical setup (31, 40, 42). We have demonstrated that robust molecular counting can be obtained using this approach (31, 32).

Pair correlation (PC) analysis and k-means-like clustering analysis were performed using MATLAB (The Mathworks, Inc., Natick, MA, USA) on 10–18 μm² regions of interest (ROIs) as described previously (31, 32, 40, 42, 43). To briefly summarize key steps in the analysis, images of cells were first binarized using localization xy-coordinate centers obtained from NIS-elements. Localizations corresponding to noise (precision values outside the 98th percentile) were removed from these images. Individual ROIs were placed across the cells and the total number of localizations from within these regions was divided by a constant value (α, respective to the specific fluorophore) to obtain detected densities in terms of the number of molecules. Auto-correlation functions were computed using fast Fourier transforms to obtain the number of molecules per cluster and cluster radius (40, 42). Subsequently, localization precision and the cluster radius from PC analysis were directed into a k-means-like clustering algorithm (32, 43) to determine the fraction of clustered receptors. This calculation groups detected localizations via thresholds for spatial parameters (PC cluster radius and average localization precision) and a temporal parameter (maximum fluorophore dark time). Molecules were counted as part of a cluster (more than two receptors) if these spatiotemporal requirements are met; otherwise, molecules were counted as unclustered. All codes for this analysis workflow have been provided previously (32, 42) and the described approach was validated using Monte Carlo simulations (32).

The calculation for all p-values was performed in Excel using the Student’s t-test with a one-tailed distribution and heteroscedastic two-sample unequal variance type.

Cells were treated with PBS as vehicle control, 10 μM afatinib (LC LABS) for the indicated times. For multivalent meditope treatment, cells were treated with 10 nM trastuzumab memAb for 10 min, washed with warm media briefly and subsequently treated with either 10 nM Fc, 10 nM M2Fc, or 10 nM M4Fc for the indicated times. For parenteral antibody treatment, cells were treated with 10 nM trastuzumab and 10 nM pertuzumab in combination for the indicated times. The meditope treated and parental antibody treated cells were then washed briefly with warm media and treated with or without 10 ng/ml EGF (Genscript). All drug incubations were performed at 37°C in a cell culture incubator. To test for HER2-paGFP transfection activity, cells were treated with 10 ng/ml EGF for 30 min.

Cells were treated with multivalent meditopes as described above for the indicated times. Cells were lysed for immunoblotting with phospho-Akt, total Akt, phospho-HER2, total HER2, phospho-Erk1/2, total Erk1/2, phospho-EGFR, and total EGFR. Blots were imaged and quantified using the Image Lab software (Biorad). Akt, Erk1/2, and EGFR phosphorylation at each time point was quantified and normalized by calculating the ratio of pAkt over total Akt, pErk1/2 over total Erk1/2, and pEGFR over total EGFR, respectively, in each lane. The relative phosphorylation was normalized to the maximum response by control Fc at each time point. The experiment was repeated five times independently to calculate an average normalized relative phosphorylation.

To prepare protein extracts for immunoblotting, cells were pelleted by centrifugation and washed two times using ice-cold PBS. The pellets were subsequently resuspended in lysis buffer (150 mM sodium chloride, 50 mM Tris, pH 8.0, 1% NP-40, Protease and Phosphatase Inhibitor Mini Tablets, Pierce) and rotated at 4°C for 30 min to lyse the cells. The cells were then centrifuged for 20 min at 10,000 rpm at 4°C. The supernatant consisting of the protein lysate was stored at −80°C. SDS-polyacrylamide gel electrophoresis and western blotting procedures were carried out using the treated cell lysates as per standard protocols. Primary antibodies used included anti-phospho-EGFR (Tyr 1068) (rabbit monoclonal, Cell Signaling), anti-EGFR (rabbit monoclonal, Abcam), anti- phospho-HER2 (tyr877) (rabbit monoclonal, Abcam) anti-HER2 (rabbit monoclonal, Abcam), anti-phospho-Akt (S473) (rabbit polyclonal, Abcam), anti-Akt (rabbit polyclonal, Abcam), anti-phospho-Erk1/2 (Thr202/Tyr204) (rabbit monoclonal, Cell Signaling); anti-Erk1/2 (rabbit polyclonal, Abcam); and anti-β-actin (mouse monoclonal, Cell signaling). Proteins were detected with Pierce ECL detection reagents (Pierce). The blots were imaged on a Biorad Chemidoc imager.

In SMLM, target molecules of interest are detected via fluorescent reporters. Since one of our objectives was to use qSMLM to map the membrane organization of HER2 in breast cancer cells, we genetically tagged HER2 with photoactivatable green fluorescent protein (HER2-paGFP) and expressed this construct in MDA-MB-468 cells. This breast cancer cell line has a very low expression level of HER2 (32) and high expression of EGFR (44) and is considered HER2 negative. We then compared levels of HER2-paGFP expressed in MDA-MB-468 cells to native HER2 found in HER2 overexpressing SK-BR-3 breast cancer cells. According to our results from Western blot analysis (Supplementary Figure 1), the amount of expressed HER2-paGFP in MDA-MB-468 cells was comparable to endogenous HER2 in SK-BR-3 cells. In addition, both cell lines were exposed (for 0, 5, 30 min) to one of two treatments: either epidermal growth factor (EGF) or a combination of EGF + afatinib. EGF is a ligand for EGFR and has been shown to stimulate growth and differentiation, activate HER2, and promote phosphorylation of receptor tyrosine kinases. Conversely, afatinib is an irreversible inhibitor of both EGFR and HER2. During the treatments, we followed the phosphorylation status of five protein kinases involved in HER2 signaling: HER2, EGFR, protein kinase B (Akt), and extracellular signal-regulated kinases 1 and 2 (Erk1/2). Treating both cell lines with 10 ng/ml EGF led to the phosphorylation of all four kinases. Phosphorylation was reduced when both cell lines were additionally treated with 10 μM afatinib.

We next assessed how the phosphorylation of EGFR (Y1068) was affected by HER2-paGFP expression. Using Western blot analysis (Supplementary Figure 2), we probed the phosphorylation of EGFR when 10 ng/ml EGF was applied to the following cells: SK-BR-3, MDA-MB-468, MDA-MB-468 expressing HER2-paGFP, and MDA-MB-453 cells. MDA-MB-453 breast cancer cells do not express significant amounts of EGFR and HER2. As expected, MDA-MB-453 cells did not show EGFR phosphorylation following EGF treatment. Treatment of the other three cell lines with EGF showed increased phosphorylation of EGFR. This increase was more pronounced in MDA-MB-468 expressing HER2-paGFP than in MDA-MD-468 cells, which have minimal endogenous HER2. Altogether, results in Supplementary Figures 1, 2 suggest that the HER2-paGFP construct is functional.

Using qSMLM, we assessed the distribution of HER2-paGFP in cultured MDA-MD-468 cells. As shown in Figure 1, HER2 was imaged both in the steady state and upon the following four treatments: Fab trastuzumab, trastuzumab, pertuzumab, and the combination of trastuzumab + pertuzumab. While some degree of HER2 clustering was observed in both the steady state and upon treatment with Fab trastuzumab, an increase in clustering was evident for all mAb treatments (Figure 1A). We assessed the imaging data using analysis algorithms to define the distribution of detected HER2 molecules per cluster (Figure 1B), the distribution of HER2 cluster radii (Figure 1C), and the fraction of clustered HER2 molecules (Figure 1D). The detected HER2 densities are shown in Supplementary Figure 3A and corresponding localization precision distributions are shown in Supplementary Figure 4A. Individual mAb treatments, trastuzumab or pertuzumab, resulted in an increased frequency of larger clusters with radii > 80 nm (Figure 1C, in gray). Trastuzumab + pertuzumab treatment produced the highest frequency of larger clusters (>80 nm radii) that contained > 8 HER2 molecules (Figures 1B, C, in gray). Additionally, the fraction of clustered HER2 molecules increased when breast cancer cells were treated with trastuzumab, pertuzumab, or a combination of trastuzumab + pertuzumab. Overall, the combination of trastuzumab + pertuzumab led to the most pronounced increase in HER2 clustering.

Meditopes are cyclic 12-mer peptides that bind tightly (∼400 nM) to a unique site within the Fab arm of cetuximab (34). Previously, we demonstrated that this meditope-binding site is absent in human mAbs but can be readily grafted onto them; we termed the constructs meditope-enabled Abs (memAbs). We have demonstrated that the presence of the meditope does not affect antigen binding of memAbs (35–38). Here, we used the technology, in concert with qSMLM, to determine how the clustering of HER2 was affected by the valency. To this end, we fused the meditope sequence to either the N-termini or both the N- and C-termini of an IgG1 Fc (CH₂-CH₃) domain. We used this approach to generate two multivalent meditopes: Fc-divalent meditope (M2Fc) and Fc-tetravalent meditope (M4Fc).

Using qSMLM, we identified how the clustering of HER2-paGFP in MDA-MB-468 cells was affected by trastuzumab memAb (TmemAb) in combination with one of the following ligands: Fc domain (monovalent control), M2Fc (divalent meditope ligand), and M4Fc (tetravalent meditope ligand). Treatments were assessed at 3 and 10 min by qSMLM (Figure 2A and Supplementary Figure 5). The detected HER2 densities are shown in Supplementary Figure 3B and the associated localization precision distributions are shown in Supplementary Figure 4B. We assessed the imaging data upon each treatment using the same analysis as the therapeutic antibodies to obtain quantitative information on the distribution of detected HER2 per cluster (Figure 2B), the distribution of HER2 cluster radii (Figure 2C), and the fraction of clustered HER2 (Figure 2D). While some HER2 clustering was observed for all treatments at 3 min, it was most pronounced for tetravalent meditope M4Fc combined with TmemAb. This combination produced the highest frequency of larger-sized clusters (>80 nm radii) occupied with >8 HER2 molecules (Figures 2B, C in gray), and the highest fraction of clustered HER2 (Figure 2D). While still present at 10 min, the magnitude of HER2 clustering was reduced across all treatments.

In addition to mapping the nanoscale organization of HER2-paGFP, where HER2 is covalently attached to the fluorescent reporter, we assessed the organization of endogenous HER2 in breast cancer BT-474 cells. To this end, TmemAb was covalently labeled with a suitable fluorescent reporter, Alexa Fluor 647 (AF647), for qSMLM. For steady state, cells were first fixed and subsequently stained with TmemAb-AF647 postfixation (PF). To assess the effects of multivalent ligands, live cells were first incubated with reagents and subsequently fixed. In the latter scenario, we tested the following six treatments: (1) TmemAb-AF647 alone; (2–5) TmemAb-AF647 + one of four M2Fc ligands; and (6) TmemAb-AF647 + M4Fc. In all cases, TmemAb-AF647 was administered at a 10 nM concentration. In the case of M2Fc, the length of the meditope linker was varied (c10, c20, c30, and c37) to probe the impact of meditope geometry on HER2 clustering.

SMLM images in Figure 3A illustrate that M2Fc and M4Fc induce an increase in HER2 clustering in combination with the TmemAb. Figure 3B summarizes the average detected molecular density of HER2 (as detected with TmemAb-AF647) and the corresponding localization precision distributions are provided in Supplementary Figure 4C. The p values for the HER2 densities are given in Supplementary Table 1. The highest HER2 densities were detected when TmemAb-AF647 was paired with either M2Fc (c37) or M4Fc (Figure 3B). According to studies in different cell lines, the normal function of HER2 may be influenced by local differences in HER2 spatial arrangement on the cell membrane: monomers, dimers, and clusters (29–33). We thus analyzed the imaging data to determine nanoscale organization of endogenous HER2 upon multivalent ligand treatment, Figures 3C, D. We identified the fraction of TmemAb-bound HER2 as either an isolated receptor (monomer), a cluster of two receptors, or a cluster with more than two receptors; the p values for the fraction of clustered HER2 are given in Supplementary Table 2. Based on the analysis (Figure 3C), HER2 monomers were most abundant when TmemAb-AF647 was used alone (PF or in live cells) and lower frequencies of HER2 monomers were observed when TmemAb-AF647 was combined with multivalent meditopes. By far the highest percentage of clustered HER2 was observed for the tetravalent meditope (Figure 3D).

Given these findings, we were interested in identifying how multivalent treatments affected downstream HER2 signaling pathways. For example, data suggests HER receptors may be highly expressed in a trastuzumab resistant setting (45–52) and may associate into signaling platforms to activate pathways and compensate for trastuzumab-induced inhibition (45, 52). We thus tested how exposing MDA-MB-468 cells that express HER2-paGFP to different treatments affected the phosphorylation of HER2, EGFR, Akt, and Erk1/2. Western blot analysis was used to identify the impact of four different treatments: (1) TmemAb + Fc; (2) TmemAb + M2Fc; (3) TmemAb + M4Fc; and (4) trastuzumab + pertuzumab. After the treatments were administered for the indicated times (Figure 4A), cells were incubated for an additional 30 min at 37°C either in the presence or absence of EGF. Images of the Western blots are shown in Figure 4A. The relative phosphorylation of Akt, HER2, and EGFR was calculated as a ratio (pReceptor divided by total Receptor) and normalized to the maximum response by Fc (Figure 4B). Remarkably, both TmemAb + M2Fc and TmemAb + M4Fc effectively blocked EGFR, HER2, and Akt phosphorylation at 10 min. Compared to the control (TmemAb + Fc), phosphorylation of EGFR, HER2, and Akt was attenuated with TmemAb and multivalent meditope ligands at 30 min; the effect was more pronounced with M4Fc. Compared to the control, phosphorylation of EGFR, HER2, and Akt was attenuated with trastuzumab + pertuzumab at 10 and 30 min, but to a lower extent.

One mechanism promoted by trastuzumab, which may lead to improved clinical outcomes, is ADCC (15, 16). Similarly, pertuzumab has been shown to activate ADCC and the enhanced efficacy of the antibody combination treatment has been explained, in part, by this mode of action (9, 17). Since clustering and phosphorylation were significantly increased by the tetravalent meditope, we were interested in measuring how it affected ADCC. To this end, SK-BR-3 cells were exposed for 6 h at 37°C to one of four treatments: (1) trastuzumab; (2) trastuzumab + M4Fc; (3) TmemAb; (4) TmemAb + M4Fc. The concentration of M4Fc was held at 10 nM while we assessed different concentrations of the Abs. Consistently across the various concentrations (Figure 5), the highest levels of ADCC activity were observed when cells were treated with TmemAb + M4Fc. A control experiment (Supplementary Figure 6) was performed, which demonstrated that ADCC activity was negligible without target or effector cells. Overall, these results suggest that the multivalent meditope, in combination with meditope-enabled trastuzumab, enhances ADCC activity.