Synthesis of ~ 20 nm spherical GNPs was performed using sodium citrate (Sigma Aldrich, MO, USA) as a reducing agent, following the methodology established by Enustun and Turkevich^[36]. A 414 μL aliquot of a 50% w/V HAuCl₄ solution was added to 200 mL of purified water (Hylabs, Rehovot, Israel), and the mixture was brought to a boil in a paraffin oil bath. Subsequently, 4.05 mL of a 10% sodium citrate solution was introduced, and the solution was stirred for 5 min. After cooling to room temperature, the GNP suspension (~ 204 mL) was functionalized with a thiol-polyethylene-glycol (PEG) layer using 9.7 μmol total thiolated PEG. The PEG layer comprised 60 mol% mPEG-SH (Mw ≈ 6 kDa; Cat #729159, Sigma Aldrich), 20 mol% heterofunctional thiol-PEG-carboxylic acid (SH-PEG-COOH) of Mw ≈ 5 kDa (Cat #757845, Sigma Aldrich) conjugated to insulin (100 IU), and 20 mol% SH-PEG-COOH Mw ≈ 3.5 kDa (Cat #757837, Sigma Aldrich) for conjugation of the antibodies, namely, trastuzumab and/or pertuzumab (Roche Ltd, Basel, Switzerland), or non-specific human immunoglobulin G (IgG)-1 isotype control antibody (InVivoMAb, cat# BE0297; Bio X Cell, Lebanon, NH, USA). After PEG addition, the suspension was stirred for 2 h; the PEG-coated GNPs were then centrifuged at 20,000 rpm for 20 min, the supernatant was removed, and the pellet was resuspended in purified water. This centrifugation and resuspension procedure was repeated three times. For conjugation, PEG carboxyl groups were first activated for 30 min using 1-ethyl-3-(3-(dimethylamino)propyl) carbodiimide HCl (EDC; Cat #22980, Thermo Fisher Scientific, Waltham, MA, USA) and N-hydroxy-sulfo-succinimide sodium salt (Sulfo-NHS; Cat# 24510, Thermo Fisher Scientific). For this purpose, fresh 30 mg/mL stock solutions of EDC and Sulfo-NHS were prepared immediately before use and added to achieve final concentrations of 1 mM each in the activation reaction. Activation was performed at pH 6.5. Excess EDC/NHS was removed by centrifugation (13,000 rpm, 30 min) and the solvent was washed and replaced with PBS (Hylabs). For nanoparticle formulations used in the in vitro experiments (IgG-GNPs, trastuzumab-GNPs, pertuzumab-GNPs, and BsGNPs), antibodies were added at a total amount of 5 mg per formulation. For formulations intended for the in vivo experiments (trastuzumab-GNPs, pertuzumab-GNPs, and BsGNPs), the antibody input was increased to 10 mg per formulation to achieve a higher antibody loading required for in vivo dosing. For BsGNPs, trastuzumab and pertuzumab were added at a mass ratio of 2:1 in both preparations. GNPs were incubated with the antibodies for 3 h, then centrifuged (13,000 rpm, 30 min), and the antibody-conjugated GNP pellet was separated from the reaction supernatant. The final GNP concentration was 30 mg/mL as determined by inductively coupled plasma optical emission spectrometry (ICP-OES).

The nanoparticles were characterized using transmission electron microscopy (TEM; JEM-1400, JEOL, Tokyo, Japan) to assess their size and shape. Additional characterization was performed through ultraviolet-visible spectroscopy (UV-Vis; UV-1650 PC, Shimadzu Corp., Kyoto, Japan), dynamic light scattering (DLS; NANO-flex, Particle Matrix, Germany), and zeta potential measurements (ZetaSizer 3000HS, Malvern Instruments, Malvern, UK) following each stage of coating. HER2 binding by BsGNPs was assessed by enzyme-linked immunosorbent assay (ELISA) using recombinant human HER2-Fc-coated plates (Cat# HE2-H5253; ACROBiosystems, Newark, DE, USA) at 0.05 µg/well overnight at 4 °C and blocked with 2% BSA. After incubation with BsGNPs (1 h at 37 °C) followed by horseradish peroxidase-conjugated goat anti-human IgG (Jackson ImmunoResearch, West Grove, PA, USA) (1:20,000) for 1 h at 37 °C, binding was detected with 3,3',5,5′-tetramethylbenzidine (TMB; Surmodics, Horsham, PA, USA), and absorbance was measured at 640 nm after blank subtraction. Antibody loading was assessed using NanoDrop spectrophotometry (NanoDrop™ 2000/2000c, Thermo Scientific, Waltham, MA, USA) by measuring unbound antibody concentration in the supernatant after BsGNP purification, and conjugated antibody was determined by mass balance.

Trastuzumab was biotinylated using EZ-Link Sulfo-NHS-LC-Biotin (Cat# 21335, Thermo Fisher Scientific) according to the manufacturer’s protocol, followed by dialysis against PBS (Slide-A-Lyzer MINI, 50-100 kDa MWCO). The biotinylated antibody was subsequently conjugated to streptavidin-phycoerythrin (PE) at a 1:1 molar ratio for 30 min at room temperature in the dark. Pertuzumab was fluorescently labeled with NHS-Fluorescein (Cat# 46410, Thermo Fisher Scientific) by incubation at a 100:1 molar ratio (dye:antibody) for 2 h at room temperature, protected from light. Excess unbound dye or PE was removed by several centrifugal filtration repeats (Amicon Ultra, 50 kDa MWCO). The resulting fluorescent antibody conjugates were stored at 4 °C in the dark until use. Fluorescently labeled antibodies (trastuzumab-phycoerythrin and pertuzumab-fluorescein) were incubated with activated PEGylated GNPs for 3 h at room temperature under gentle agitation, resulting in three types of conjugates: GNPs bearing only trastuzumab-PE, only pertuzumab-fluorescein, or both antibodies simultaneously. Conjugates were purified by centrifugation (14,000 rpm, 20 min) and resuspended in PBS. The dual-labeled antibody-GNP conjugates were stored at 4 °C in the dark until further analysis. Fluorescence intensity was measured using a Synergy H1 multimode microplate reader (Agilent, Santa Clara, CA, USA).

The fluorescence lifetime imaging microscopy (FLIM) setup consisted of a Lambert Instruments system (LIFA, Groningen, The Netherlands), frequency-domain (FD) FLIM, with the excitation source of a multi-light-emitting diode (LED) working in 468 nm wavelength mode, modulated in a sinusoidal mode by a signal generator (Prior OptiScan I, Rockland, Massachusetts, United States) and a time-resolved intensified low-light imaging CMOS camera with ultra-short gating. In this experiment, three scanning-mode frequencies (34 MHz, 38 MHz, and 42 MHz) were used to excite the samples. An Olympus IX-81 model inverted microscope with a 10× NA¼ 0.4 objective (Olympus, Tokyo, Japan) was used to achieve precise focusing on the samples. Samples were prepared on glass slides by taking 5 ml of each solution onto separate slides using a pipette. Each slide was then placed under the microscope and analyzed using LIFA FD-FLIM software. Average and standard deviation values were estimated using LIFA FLIM v1.2.26 software (Lambert Instruments).

Human breast cancer SKBR3 cells (ATCC Cat# HTB-30, RRID: CVCL_0033), human breast cancer BT-474 cells (ATCC, Cat# HTB-20, RRID: CVCL_0179), or epidermoid carcinoma A431 cells (ATCC, Cat# CRL-1555, RRID: CVCL_0037) were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Sigma-Aldrich) supplemented with 10% fetal bovine serum (FBS, Cat# 04-001-1A, Biological Industries, Beit HaEmek, Israel), 2 mM L-Glutamine, and 1% Penicillin/Streptomycin (Biological Industries, Beit HaEmek, Israel). HepG2 human hepatocellular carcinoma cells (ATCC, cat# HB-8065, RRID CVCL_0027) were maintained in RPMI 1640 (#SH30096.01, HyClone, Wilmington, USA) supplemented with 10% FBS, 2 mM L-glutamine, 100 U/mL penicillin, and 100 µg/mL streptomycin; viability assays were performed in the corresponding antibiotic-free medium. KUP5 cells (mouse Kupffer cell line clone 5) (RIKEN BRC, cat# RCB4627, RRID CVCL_6C88) were cultured in high-glucose DMEM (#11995-065, Gibco, USA) supplemented with 10% FBS, 10 µg/mL human insulin, and 250 µM monothioglycerol (Sigma-Aldrich). Cells were maintained at 37 °C under humidified conditions with 5% CO₂. All cell lines were used between passages 3 and 6 post-thaw. The absence of mycoplasma contamination in all cell lines was confirmed prior to experimental use by Myco-One Step Mycoplasma Detector (Cat# D201-02, Vazyme Biotech, Nanjing, China).

For proliferation assays, cells were counted using a hemocytometer and seeded at 1 × 10⁴ cells/well in 96-well plates under normal growth conditions. After 24 h, cells were treated with BsGNPs or IgG-GNPs (0.3, 0.6, 1.5, or 3 µg/mL of bound antibody corresponding to gold concentrations of 10, 20, 50, or 100 µg/mL), 3 µg/ml total free antibodies, or the single-antibody-GNP cocktail (trastuzumab-GNPs and pertuzumab-GNPs; 3 µg/ml of bound antibody). Antibody-equivalent concentrations were determined based on the measured antibody loading per GNP. Treatments were diluted in culture medium to the indicated concentrations, with a final volume of 200 µL per well (three technical replicates per condition). Cells were maintained under standard culture conditions for 6 days; on day 4, medium was replaced, and treatments were replenished at the same concentrations. Experimental data comprised 3-4 biological replicates performed on different days with three technical replicates per biological replicate. Technical replicates were averaged within each independent experiment. Cell proliferation was assessed using the CyQUANT^TM NF Cell Proliferation Assay kit (#C35006, Thermo Fisher Scientific, Waltham, MA, USA), following the manufacturer's protocol. For viability testing, HepG2 or KUP5 cells were seeded (5 x 10⁵ cells/mL); 24 h later, cells were incubated with the BsGNPs in triplicate for 48 h. Cell viability was quantified using the MTT reduction assay [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; Sigma-Aldrich, M5655].

A BT-474 breast cancer tumor model was established by subcutaneous injection of 5 × 10⁶ BT-474 cells into the back flank of 6-week-old Hsd:Athymic nude-Foxn1nu female mice (Harlan Laboratories, Jerusalem, Israel). Once the tumors reached 40 mm² (day 14), mice were randomly divided into groups and received weekly IP injections of either free antibodies, a cocktail of single-antibody GNPs (trastuzumab-GNPs and pertuzumab-GNPs), or BsGNPs. All treatments were administered at an equivalent antibody dose (free or bound) of 0.945 mg per mouse per injection (with 12.6 mg Au/injection administered in the GNP-treated groups). Treatments were given on days 14, 21, 28, and 35. Tumor growth was monitored in a blinded manner at several time points over the study. Tumor length and width were measured using a caliper, and tumor area was calculated as length × width (mm²). At the end of the experiment, the mice underwent CT imaging to evaluate gold accumulation within the tumor. Following euthanasia, tissue samples were collected to assess the biodistribution of gold in various organs using ICP-OES.

At 24 h post-injection, CT scans were performed to assess nanoparticle distribution and tumor morphology. The abdominal (tumor) region was imaged using a Bruker Skyscan 1176 micro-CT scanner with a 35 μm nominal resolution, a 0.2 mm aluminum filter, and a tube voltage of 45 kV. Reconstruction was carried out using a modified Feldkamp algorithm with GPU acceleration via SkyScan NRecon software. Image processing included ring artifact reduction, Gaussian smoothing (3%), and beam hardening correction (20%). Volume-rendered 3D images were generated using RGBA transfer functions in SkyScan CT-Volume (“CT-Vol”) and CT-Voxel (“CT-Vox”) software.

At 24 h post-injection and following CT imaging, mice (n = 3) were sacrificed, tumors and major organs (liver, spleen, kidneys, lungs, and heart) were excised, and blood (~ 1.5 mL) was collected. The mean ± standard deviation (SD) weights of the major organs were 0.918 ± 0.036 g for the liver, 0.076 ± 0.011 g for the spleen, 0.218 ± 0.018 g for the kidneys, 0.222 ± 0.074 g for the lungs, and 0.112 ± 0.010 g for the heart. Tissues were digested in aqua regia (a mixture of nitric acid and hydrochloric acid at a 1:3 volume ratio) and diluted with purified water to a final volume of 5 mL. After filtration, gold concentrations in tumors and organs were quantified using inductively coupled plasma optical emission spectroscopy (ICP-OES; 710, Agilent Technologies) with calibration curves prepared from standard gold solutions (0, 0.1, 0.5, 1, 5, and 10 mg/L). Percentage of injected dose (%ID) was determined by calculation of ICP-OES measured gold concentration (ppm) × final sample volume (mL) / total amount of injected gold (mg) × 100.

Blood samples collected for serum chemistry analysis were allowed to clot for 30 min at room temperature and centrifuged at 3,000 × g for 10 min at 22 °C. Serum creatinine, urea, alanine aminotransferase (ALT), aspartate aminotransferase (AST), and alkaline phosphatase (ALP) levels were measured using an automatic biochemistry analyzer (Roche Cobas c311) with commercial assay kits (Cat# 03263991190, 04460715190, 20764957322, 20764949322, and 03333701190; Roche Diagnostics, Basel, Switzerland) according to the manufacturer's instructions.

All animals were housed under controlled conditions, maintaining a 12:12 h light/dark cycle, a stable temperature of 23 °C, and 50% humidity, with unrestricted access to food and water. All animal procedures were conducted at the Animal Research Center of Bar-Ilan University in accordance with the guidelines and regulations of the National Institutes of Health (DHEW Pub. No. NIH 78-23). The Bar-Ilan University Institutional Animal Care and Use Committee approved the study protocol, and the study was conducted under its supervision (protocol no. 2401-104-5).

Statistical analysis was conducted using GraphPad Prism 10 (GraphPad Software, Inc.). Two-tailed paired Student’s t-tests were used for paired two-group comparisons. For in vitro experiments, one-way analysis of variance (ANOVA) with Tukey’s multiple-comparison test or two-way ANOVA with Holm-Šídák’s multiple-comparisons test was used for independent-group analyses, with biological replicates as the experimental units. Longitudinal tumor-size data were analyzed using two-way ANOVA with repeated measurements followed by Tukey’s multiple-comparisons test. Parametric analyses assumed approximately Gaussian residuals (formal normality testing was not performed because of the small sample sizes). Homogeneity of variance was assessed, where applicable, using the Brown-Forsythe test. A P-value of < 0.05 was considered statistically significant.

Gold nanoparticles were synthesized and subsequently functionalized with PEG linkers to which trastuzumab and pertuzumab were covalently attached. Transmission electron microscopy confirmed a uniform, spherical morphology with an average core diameter of 20.0 ± 3.4 nm and 20.0 ± 3.1 nm for bare nanoparticles and BsGNPs, respectively [Figure 1A]. We previously showed that this size optimally balances tumor delivery and CT imaging capability^[22]. UV-visible spectroscopy revealed stepwise red-shifts in the surface plasmon resonance following PEGylation and antibody conjugation [Figure 1B], consistent with progressive surface modification. Dynamic light scattering measurements showed an increase in hydrodynamic diameter from 25 ± 3 nm for bare GNPs to 65 ± 8 nm for BsGNPs [polydispersity Index (PDI) = 0.1878; Figure 1C], and zeta potential values shifted from -39.8 mV to -13.7 ± 9 mV after antibody conjugation [Figure 1D]. Together, these size and surface charge changes confirmed successful antibody conjugation. These physicochemical parameters remained comparable after three months of storage at 4 °C (hydrodynamic diameter, 60 ± 7 nm; PDI 0.1944; zeta potential, -14.0 ± 6.4 mV). In addition, ELISA showed that HER2-binding activity was maintained over one week at 4 °C storage (EC50 of 0.134 ± 0.084 µg/mL after synthesis and 0.118 ± 0.058 µg/mL after a week; mean ± SD, n = 3 independent samples; two-tailed paired Student’s t-test, P = 0.62). These data support the physicochemical stability of BsGNPs and retention of their binding functionality. NanoDrop-based analysis indicated an antibody loading of 9.9 ± 2.0 antibodies per BsGNP for the preparations used in the in vitro experiments. To achieve higher antibody loading required for in vivo dosing, the in vivo preparations were synthesized with increased antibody input, yielding an average of 24.7 ± 1.8 antibodies per BsGNP (mean ± SD, n = 3 independent samples).

To confirm that trastuzumab and pertuzumab are co-conjugated on the same nanoparticle, rather than distributed across separate, singly functionalized populations, two complementary BsGNP formulations were synthesized. In the first formulation, trastuzumab was labeled with phycoerythrin (PE) while pertuzumab remained unlabeled; in the second, pertuzumab was labeled with fluorescein (FLU) while trastuzumab remained unlabeled [Figure 2A]. This approach enabled independent verification of each antibody’s incorporation into the BsGNPs while controlling potential cross-labeling artifacts. Fluorescence intensity analysis revealed clear and mutually exclusive signal detection in the expected channels [Figure 2B and C]: BsGNPs bearing PE-labeled trastuzumab exhibited strong fluorescence in the PE channel (MFI > 3 × 10⁴) with negligible signal in the FLU channel, whereas BsGNPs bearing FLU-labeled pertuzumab showed the opposite pattern, with robust FLU fluorescence (MFI ~ 950) and minimal PE background. BsGNPs lacking fluorescent labels displayed negligible signal in both channels, confirming assay specificity. To confirm co-localization of both antibodies on individual nanoparticles, FLIM was performed on BsGNPs functionalized with either single-labeled or dual-labeled antibodies [Figure 2D]. All labeled samples exhibited discrete, punctate fluorescent signals corresponding to antibody-functionalized nanoparticles, whereas unlabeled BsGNPs showed no detectable fluorescence. Nanoparticles carrying only fluorescein- or PE-labeled antibodies displayed fluorescence lifetimes corresponding to the native lifetimes of the free dyes in solution (~ 4 ns for FLU and ~ 2 ns for PE; Figure 2E). In contrast, dual-labeled BsGNPs exhibited multicomponent fluorescence lifetimes spanning both characteristic lifetime ranges, which were detected within the same regions of interest, confirming the co-localization of both fluorophores and thus both antibodies, on individual nanoparticles [Figure 2E]. Collectively, these results demonstrate successful dual functionalization of trastuzumab and pertuzumab on the gold nanoparticle surface.

The antiproliferative effects of BsGNPs were first evaluated in vitro. The HER2-overexpressing breast cancer cell line SKBR3 was treated with either BsGNPs or non-specific IgG-conjugated GNPs (IgG-GNPs) at increasing concentrations (0.3-3 µg/mL of bound antibody). IgG-GNPs showed no concentration-dependent reduction in cell proliferation, whereas BsGNPs progressively reduced proliferation with increasing concentration. At matched antibody concentrations, proliferation was significantly lower following BsGNP treatment than following IgG-GNP treatment at 0.6, 1.5, and 3 µg/mL (P < 0.01, P < 0.001, and P < 0.0001, respectively). Within the BsGNP group, 1.5 µg/mL significantly reduced proliferation compared with 0.3 µg/mL (P < 0.05), while 3 µg/mL significantly reduced proliferation compared with 0.3 and 0.6 µg/mL (P < 0.0001 and P < 0.01, respectively; two-way ANOVA with Holm-Šídák’s multiple-comparisons test; Figure 3A). Next, using the most effective concentration, we compared BsGNP activity to an equivalent dose of free trastuzumab and pertuzumab or a cocktail of single-antibody GNPs consisting of trastuzumab-GNPs and pertuzumab-GNPs (total antibody concentration of 3 µg/mL per group). BsGNPs exhibited significantly greater antiproliferative activity than both the free antibody combination and the single-antibody GNP cocktail (trastuzumab-GNPs and pertuzumab-GNPs) (P < 0.01 and P < 0.05, respectively; Figure 3B).

We further evaluated BsGNP activity across cell lines with differing HER2 expression, including low-HER2-expressing human epidermoid carcinoma A431 cells^[37,38] and the HER2-overexpressing human breast cancer BT-474 and SKBR3 cells^[39,40]. BsGNPs significantly reduced cell proliferation in SKBR3 cells compared with A431 cells (P < 0.05; Figure 3C). While the reduction observed in BT-474 cells did not reach statistical significance compared with A431 cells (P = 0.0503), BsGNP antiproliferative activity in BT-474 cells was not significantly different from that observed in SKBR3 cells [Figure 3C].

To investigate the cytotoxicity of the BsGNPs, we treated both human hepatocarcinoma cells (HepG2) and mouse Kupffer cells (KUP5) with increasing gold concentrations (10-100 µg/mL) for 48 h. MTT assay showed that BsGNPs did not affect cell viability in either cell type at any dose [Figure 3D].

Taken together, these findings demonstrate that BsGNPs induce an enhanced, dose-dependent antiproliferative effect in HER2-expressing cells, outperforming both the free antibody combination and the single-antibody nanoparticle cocktail. The findings also suggest an antiproliferative effect that increases with HER2 expression level. The absence of cytotoxicity in HepG2 and KUP5 cells further supports the favorable in vitro biocompatibility of the BsGNP platform.

We next evaluated the tumor targeting capacity and therapeutic efficacy of BsGNPs in a HER2-positive xenograft mouse model.

BT-474 cells were used due to their established ability to form robust HER2-overexpressing xenografts. Tumor-bearing nude mice were injected IP with either BsGNPs, a single antibody-GNP cocktail (trastuzumab-GNPs and pertuzumab-GNPs), or the combination of free trastuzumab and pertuzumab (all groups at equivalent antibody doses, weekly, over four weeks).

Intratumoral delivery of BsGNPs was evaluated by CT imaging at 24 h post-injection, demonstrating clear accumulation of the nanoparticles within the tumors [Figure 4A]. The single-antibody-GNP cocktail was also detectable within the tumors [Figure 4A]. Quantitative elemental analysis of tumors at this time point showed similar intratumoral gold accumulation for both BsGNPs and antibody-GNP cocktail treatments (4.19% ± 1.1 %ID/g tissue and 4.57 ± 1.85 %ID/g tissue) respectively; mean ± SD, n = 3), indicating similar tumor delivery efficiency.

Next, biodistribution analysis across major organs showed high accumulation of BsGNPs in the liver, consistent with typical clearance patterns for gold nanoparticles, and negligible amounts in the heart [Figure 4B].

Tumor growth was monitored over time following treatment initiation [Figure 4C]. Notably, despite comparable intratumoral gold accumulation between BsGNPs and the antibody-GNP cocktail, BsGNP-treated tumors showed a marked suppression of tumor growth, with near-complete elimination by the endpoint, and tumor size significantly reduced compared to all other groups (P < 0.05-0.001). In contrast, treatment with free antibodies and the antibody-GNP cocktail resulted in partial inhibition of tumor growth and remained significantly less effective than BsGNPs by the final time point.

Mean body weight of the BsGNP-treated mice remained stable during the treatment period and was comparable to the untreated, free antibody- and single-antibody cocktail-treated groups, with no evidence of treatment-associated weight loss [Figure 4D]. Moreover, blood chemistry analysis revealed no abnormalities in key markers of renal and hepatic function following BsGNP treatment. Creatinine, urea, ALT, AST, and ALP levels remained within established reference ranges for mice^[41,42]. Measured levels were 0.185 ± 0.007 mg/dL (reference range: 0.1-0.4 mg/dL), 55.7 ± 8.4 mg/dL (28-76 mg/dL), 30.0 ± 3.7 U/L (22-296 U/L), 129.4 ± 12.9 U/L (38-361 U/L), and 228.4 ± 11.6 U/L (56-356 U/L), respectively (mean ± SD; n = 5). These findings suggest favorable systemic tolerability of BsGNPs.

Together, our results demonstrate that simultaneous presentation of trastuzumab and pertuzumab on a single nanoparticle significantly augments therapeutic efficacy while maintaining biocompatibility.