Antibody and siRNA Nanocarriers to Suppress Wnt Signaling, Tumor Growth, and Lung Metastasis in Triple-Negative Breast Cancer

Abstract

The paucity of targeted therapies for triple-negative breast cancer (TNBC) causes patients with this aggressive disease to suffer a poor clinical prognosis. A promising target for therapeutic intervention is the Wnt signaling pathway, which is activated in TNBC cells when extracellular Wnt ligands bind overexpressed Frizzled7 (FZD7) transmembrane receptors. This stabilizes intracellular β-catenin proteins that in turn promote transcription of oncogenes that drive tumor growth and metastasis. To suppress Wnt signaling in TNBC cells, we developed therapeutic nanoparticles (NPs) functionalized with FZD7 antibodies and β-catenin small interfering RNAs (siRNAs). The antibodies enable TNBC cell-specific binding and inhibit Wnt signaling by locking FZD7 receptors in a ligand unresponsive state, while the siRNAs suppress β-catenin through RNA interference. Compared to NPs coated with antibodies or siRNAs individually, NPs coated with both agents more potently reduce the expression of several Wnt related genes in TNBC cells, leading to greater inhibition of cell proliferation, migration, and spheroid formation. In two murine models of metastatic TNBC, the dual antibody/siRNA nanocarriers outperformed controls in terms of inhibiting tumor growth, metastasis, and recurrence. These findings demonstrate suppressing Wnt signaling at both the receptor and mRNA levels via antibody/siRNA nanocarriers is a promising approach to combat TNBC.

Graphical Abstract

ToC entry

Nanoparticles coated with Frizzled7 antibodies and β-catenin siRNAs can bind triple-negative breast cancer (TNBC) cells to suppress Wnt signaling at both the receptor and effector levels. This reduces TNBC cell proliferation, migration, and spheroid formation in vitro and limits tumor growth, metastasis, and post-resection recurrence in vivo.

1. Introduction

Breast cancer is the most prevalent cancer worldwide, with a remarkable 2.26 million cases and 685,000 deaths in 2020.¹ It is a highly heterogenous disease with multiple subtypes defined by the expression of human epidermal growth factor receptor (HER2) and two hormone receptors (estrogen receptor (ER) and progesterone receptor (PR)). The subtype that expresses none of these receptors is triple-negative breast cancer (TNBC), and it accounts for ~15% of all breast cancer cases.^2–4 TNBC is particularly aggressive and exhibits faster tumor growth, higher rates of recurrence, and poorer prognosis than other forms of breast cancer.³ The standard of care includes chemotherapy, radiation, and surgery, which has remained unchanged for decades and is not sufficient to combat TNBC.⁴ Moreover, TNBC is unsusceptible to conventional targeted or hormonal therapies that leverage HER2 and the hormone receptors. Only recently have targeted therapies for TNBC begun to enter the market, including the antibody-drug conjugate sacituzumab govitecan (which targets Trop-2 for delivery of the topoisomerase I inhibitor SN38) and Talazoparib and Olaparib, two PARP inhibitors approved for patients with BRCA1/2 mutations.^5–9 While these advances are exciting, there is a continued need for new targeted therapies to be developed that exploit the biological pathways known to contribute to TNBC’s aggressive nature.

Given the challenges associated with treating TNBC using conventional methods, researchers have recently begun to develop nanoparticle (NP)-based therapies to combat this lethal disease. These include (i) phototherapeutic NPs that produce heat or singlet oxygen in response to activation with near-infrared light, (ii) gene regulatory agents that deliver small interfering ribonucleic acid (siRNA) or microRNA molecules into TNBC cells to suppress disease-promoting genes, (iii) drug delivery vehicles, and (iv) multifunctional therapies that provide a combination of these or other treatments.^10–19 While these advances are promising, further investigation into potential treatment strategies is required since TNBC is plagued by cellular resistance mechanisms. Herein, we introduce NPs that co-deliver antagnostic antibodies and siRNAs to TNBC cells to attack Wnt signaling, a master regulator of TNBC whose suppression should substantially improve patient outcomes.

Hyperactive Wnt signaling is a key driving force behind TNBC progression that facilitates its rapid growth and metastasis to the lungs.^20–25 Wnt signaling is activated in TNBC cells when external Wnt ligands bind to transmembrane FZD7 receptors, which are overexpressed in ~67% of TNBC tumors.²⁶ This leads to stabilization, cytoplasmic accumulation, and nuclear translocation of intracellular β-catenin molecules, which initiate transcription of multiple genes including Axin2, cyclin D1, and c-Myc.^27–31 Axin2 is a universal indicator of Wnt activity while cyclin D1 and c-Myc promote cell proliferation, migration, and invasion.^{28, 29} Many other oncogenes are also regulated by Wnt signaling, making this pathway a critical target for therapeutic manipulation. We have previously shown that FZD7 antibody nanocarriers can suppress intracellular β-catenin and Axin2 expression in targeted TNBC cells by blocking Wnt ligand/FZD7 receptor interactions.^{32, 33} These nanocarriers are substantially more potent than freely delivered FZD7 antibodies because they exhibit multivalent binding, wherein antibodies on nanocarriers engage multiple receptors simultaneously to increase overall binding strength/avidity and enhance signaling inhibition relatively to freely delivered antibodies.^{32, 33} Here, we expand on our approach to manipulate Wnt signaling by developing NPs that co-deliver both FZD7 antibodies and β-catenin siRNAs. The advantage of this approach is that it targets Wnt signaling at both the receptor (FZD7) and effector (β-catenin) level (Scheme 1). Historically, β-catenin has been considered undruggable due to lack of an effective binding site for small molecule therapeutics.³⁴ Using siRNA to suppress β-catenin through RNA interference (RNAi) will circumvent this problem. While freely delivered siRNAs are not suitable for clinical use owing to their nuclease susceptibility, poor pharmacokinetics, and limited cellular uptake, numerous studies have shown that siRNA-coated NPs can increase RNA stability, circulation, and cellular uptake to enable in vivo gene regulation.^35–37 Hence, we postulated that dual siRNA/antibody nanocarriers would provide both targeted treatment of TNBC and multi-level inhibition of Wnt signaling, resulting in potent anti-tumor and anti-metastasis effects.

Scheme 1. Designing NPs to inhibit Wnt signaling at multiple levels.

NPs coated with FZD7 antibodies and β-catenin siRNAs can inhibit Wnt signaling at both the receptor and effector level. Upon binding FZD7, the NPs block Wnt ligand interaction with FZD7 receptors, leading to intracellular phosphorylation and destruction of β-catenin proteins. siRNA delivered by the NPs is loaded into the RNA induced silencing complex (RISC), resulting in RNA interference mediated cleavage of β-catenin messenger RNA (mRNA). Together, these effects yield robust suppression of triple-negative breast cancer.

We report the synthesis of silica core/gold shell nanoshells (NS) coated with both FZD7 antibodies and β-catenin siRNAs (Combo-NS) to enable both TNBC cell targeting and multi-level Wnt inhibition (Scheme 1, Figure 1). NS were selected as the core nanocarrier because they have proven safety in human clinical trials^{38, 39} and their gold surface enables simple bioconjugation to siRNAs and antibodies via gold-thold attachment chemistry. Additionally, NS bioconjugates are large enough to enable multivalent receptor binding but small enough to be internalized by TNBC cells.^{32, 33} In this study, NS coated with antibodies or siRNAs individually (FZD7-NS and βcat-NS, respectively) were prepared as controls (Figure 1). We evaluated the impact of each formulation on TNBC cells in vitro, including examination of cellular uptake, target gene expression, proliferation, migration, and sphere formation capacity (Figures 2–6). Multiphoton microscopy studies showed that FZD7 antibodies increase NS binding to MDA-MB-231 TNBC cells compared to non-cancerous MCF10A cells that have lower FZD7 expression (Figure 2). Reverse transcription quantitative polymerase chain reaction (RT-qPCR) analyses revealed that Combo-NS suppress β-catenin and several downstream target genes to a greater extent than FZD7-NS or βcat-NS (Figure 3), which allowed the Combo-NS to inhibit cell proliferation (Figure 4), migration (Figure 5), and spheroid formation (Figure 6) more effectively. These data support that co-delivery of antibodies and siRNAs is advantageous versus delivering either agent individually. The impact of the three Wnt inhibitory NPs was also evaluated in two murine TNBC models (Figures 7–11). When tested in an experimental lung metastasis model wherein MDA-MB-231 TNBC cells are injected in the tail vein and form metastatic nodules in the lung, Combo-NS reduced metastatic burden to a greater extent than the monotherapies (Figure 7). Post-mortem biodistribution analyses showed that both Combo-NS and FZD7-NS coated with FZD7 antibodies had more accumulation in lungs with metastatic sites than the βcat-NS that lacked antibodies (Figure 8). In a spontaneous syngeneic lung metastatic model in which 4T1 TNBC tumors are grown in the mammary fat pad and then spread to the lungs, all three Wnt inhibitory NPs reduced primary tumor growth to a similar extent (Figure 9), but Combo-NS were more effective in preventing lung metastasis (33% of mice treated with Combo-NS did not form lung metastasis, compared to ~15% of mice treated with FZD7-NS and βcat-NS) (Figure 10). Biodistribution assays confirmed the two antibody-loaded NPs (Combo-NS and FZD7-NS) accumulated in primary tumors and lung metastases at higher levels than the βcat-NS (Figure 11), supporting the use of FZD7 antibodies for tumor and metastasis targeting.

Figure 1. Characterization of antibody/siRNA nanocarriers.

(A) Scheme depicting the steps to functionalize NS with FZD7 antibodies and β-catenin siRNAs. (B) Hydrodynamic diameter and zeta potential of bare NS compared to NS conjugates. (C) Antibody and/or siRNA loading on FZD7-NS, βcat-NS, and Combo-NS. Data in B and C represent the mean ± standard error of n=11–19 batches for each NP type.

Figure 2. NP cellular binding and uptake analyzed by microscopy and flow cytometry.

(A) Scheme of NP treatment. (B) Multiphoton microscopy images of NP binding to MDA-MB-231 cells that overexpress FZD7 receptors or MCF10A cells with low FZD7 expression. Scales = 20 μm. (C) Cellular uptake of NPs after 1 h incubation quantified via flow cytometry. Data show the mean ± standard error of n=4 biological replicates per NP in each cell line; **p<0.01 and *p<0.05 by one-way ANOVA with post hoc Tukey (all p-values provided in supplement). (D) Representative flow cytometry histograms showing NP interaction with MDA-MB-231 cells.

Figure 6. MDA-MB-231 spheroid formation in response to NP treatment.

(A) Scheme of NP treatment and cell reseeding to allow for spheroid formation. (B) Metabolic activity of spheroids measured by Alamar Blue assay after one week of formation. Data are normalized to the NT group at each time point and depict the mean ± standard error of n=3 biological replicates per NP. **p<0.01 by one-way ANOVA with post hoc Tukey. (C) Representative brightfield images of spheroids formed after one week with 72 h NP treatment time. Two types of spheres were observed to form after NP treatment, including loose clusters of cells (top row) and less opaque/dense spheres (bottom row). Scales = 200 μm.

Figure 3. NP-mediated gene regulation quantified via RT-qPCR.

(A) Scheme of NP treatment. (B) Relative mRNA expression of several Wnt target genes in MDA-MB-231 cells following NP treatment. Data are normalized to the NT group within each gene and show the mean ± standard error of n = 3 biological replicates for all genes except C-myc, which is n=2 biological replicates. **p<0.01, *p<0.05, and #p<0.10 versus untreated cells (NT) within each target gene by one-way ANOVA with post hoc Tukey.

Figure 4. Assessment of cell proliferation following NP treatment.

(A) Scheme of NP treatment and EdU incorporation assay. (B) Normalized count of proliferating MDA-MB-231 cells following NP treatment for 48, 72, or 96 h. Data are normalized to the NT group within each time point and depict the mean ± standard error of n=3 biological replicates per NP for each time point. #p<0.10 and *p<0.05 by one-way ANOVA with post hoc Tukey.

Figure 5. MDA-MB-231 cell migration in response to NP treatment.

(A) Scheme of NP treatment and cell reseeding in transwell insert for migration assay. (B) Count of migrated MDA-MB-231 cells in each treatment group at various imaging timepoints post reseeding. Data represent mean ± standard error (n=3 biological replicates per NP at each time point). No statistical differences between groups within each timepoint were determined by one-way ANOVA with post hoc Tukey. (C) Representative fluorescence microscopy images of migrated GFP-231 cells at the 24 h timepoint in each treatment group. Scales = 1 mm.

Figure 7. Analysis of NP treatment effects on an experimental TNBC lung metastasis model.

(A) Scheme of experimental design. Created with BioRender.com. (B) Schedule of cell injection and NP treatment. (C) Quantified luminescence of lungs in mice imaged by IVIS at day 58. Data are mean ± standard error (n=3 mice in the FZD7-NS and βcat-NS groups and n=4 mice in the saline and Combo-NS groups). *p<0.05 and #p<0.10 versus saline by one-way ANOVA with post hoc Tukey-Kramer. (D) Representative IVIS images of mice in each treatment group at day 55.

Figure 11. ICP-MS quantification of gold in primary tumors, recurrent tumors, and metastatic lungs.

(A) Amount of gold detected in primary tumors for each treatment group. (B) Amount of gold detected in lungs and recurrent tumors under treatment schedule 1 (left) and schedule 2 (right). Data are the mean ± standard error (n=12–14 mice per NP for the primary tumor; n=6–7 mice per NP for the lungs and secondary tumor in each dosing schedule). Statistical differences denoted as *p<0.05, **p<0.01 by one-way ANOVA with post hoc Tukey-Kramer.

Figure 8. Quantification and visualization of NP accumulation in lungs bearing TNBC lesions.

(A) Quantification of gold in excised lungs determined by ICP-MS. Data are mean ± standard deviation after subtracting background from saline-injected mice. **p<0.01 and *p<0.05 versus βcat-NS by one-way ANOVA with post hoc Tukey-Kramer. (B) Representative darkfield image of FZD7 antibody-coated NS accumulation in metastatic lung nodules versus adjacent normal lung tissue. Scale = 50 μm.

Figure 9. Effect of Wnt inhibitory NPs on growth of primary tumors in a syngeneic murine TNBC model.

(A) Overview of murine model and NP treatment schedules. Created with BioRender.com. (B) Analysis of primary tumor growth in the mammary fat pad between the day of cell injection (day 0) and the day of primary tumor removal (day 13). (Left) Mean ± standard error of primary tumor volume in each treatment group at distinct timepoints. (Right) Change in primary tumor volume between the first day of treatment (day 0) and the day of tumor removal (day 13). Data show mean ± standard error. n=12–14 mice per group. *p<0.05 by one-way ANOVA with post hoc Tukey-Kramer. (C) Analysis of β-catenin protein expression (brown) in primary tumors that were surgically removed on day 13. Sections were counterstained with hematoxylin (blue/purple) (Scale bars = 100 μm). The average percent β-catenin positive cells (shown as mean ± standard deviation) in three tumor sections per group was quantified using QuPath software. **p<0.01, *p<0.05, and #p<0.10 by one-way ANOVA with post hoc Tukey.

Figure 10. Analysis of NP treatment effect on recurrence and spontaneous TNBC metastasis to lungs.

Primary tumors were surgically removed on day 13, and treatment continued under two dosing schedules (as depicted in Figure 9A) with growth of recurrent tumors and lung metastasis monitored by bioluminescence imaging. n=6 – 7 mice per group for each dosing schedule. (A) Analysis of treatment effect on recurrent tumor growth. Mean ± standard error of tumor luminescence in each group at distinct timepoints when treated under schedule 1 (left) or schedule 2 (right). #p<0.10 compared to saline control by one-way ANOVA with post hoc Tukey-Kramer. (B) Quantified luminescence of lung metastases in mice treated with NPs or saline under schedule 1 (left) or schedule 2 (right). Data are mean ± standard error. (C) Percentage of mice in each treatment group that did not form metastasis by the end of the study; (left) schedule 1 and (right) schedule 2.

Taken together, these in vitro and in vivo data indicate Wnt inhibitory NPs hold great potential as TNBC therapeutics, as they can inhibit primary tumor growth, reduce the burden of pre-existing metastasis, and limit the formation of new metastasis. While FZD7 antibody nanocarriers (FZD7-NS) and β-catenin siRNA nanocarriers (βcat-NS) show promise individually, they are less effective than Combo-NS that co-deliver these molecules to attack Wnt signaling at both the receptor and mRNA level simultaneously. These results support the continued development of dual antibody/RNA nanocarriers to target Wnt signaling in TNBC, as well as this and other signaling pathways in distinct cancer types.

2. Results and Discussion

2.1. Characterization of Antibody and siRNA Nanoshell Conjugates

Nanoshells (NS) comprised of ~120 nm diameter silica cores and ~15 nm thick gold shells were synthesized per established methods.⁴⁰ These were then coated with biomolecules to produce three types of Wnt inhibitory NPs: (1) NS coated with FZD7 antibodies and methoxy-poly(ethylene glycol)-thiol (mPEG-SH) (FZD7-NS), (2) NS coated with β-catenin siRNAs and mPEG-SH (βcat-NS), and (3) NS coated with both FZD7 antibodies and β-catenin siRNAs, along with mPEG-SH (Combo-NS) (Figure 1A). Detailed synthesis methods are provided in Section 4.1 and the Supporting Information, and a discussion of how antibody and RNA addition was tuned to yield equivalent loading between mono- and dual-loaded NS is provided in the Supporting Information (Scheme S1, Figure S1-S4). siRNA sequence information is provided in Table S1. Dynamic light scattering (DLS) showed that the conjugated NS were ~20 nm larger than bare NS and were also more neutral in charge, indicating biomolecule attachment (Figure 1B). To quantify the antibody and siRNA loading on NS, we performed solution-based enzyme linked immunosorbent assays (ELISAs) and Oligreen assays, respectively, as previously reported.^{32, 41, 42} Combo-NS and FZD7-NS had ~130 antibodies per NS and Combo-NS and βcat-NS had ~2,400 siRNA duplexes per NS (Figure 1C). This corresponds to the following NS, antibody, and siRNA concentration for reference: NS diluted to an optical density of 1 at the peak resonance wavelength near 810 nm (i.e., OD^{810 nm} =1) contain ~3×10⁹ NPs mL⁻¹, which is ~0.8 nM FZD7 antibodies and ~12 nM siRNA. Similarly, NS at OD^{810 nm}=5 have ~1.5×10¹⁰ NPs mL⁻¹, which is ~4 nM FZD7 antibodies and ~64 nM siRNA.

2.2. Wnt Inhibitory NPs Incorporating FZD7 Antibodies Exhibit Preferential Binding/Uptake by TNBC Cells

To demonstrate the advantage of incorporating FZD7 antibodies in the NP coating, we evaluated the binding of each NP type to MDA-MB-231 TNBC cells versus MCF10A non-cancerous breast epithelial cells (Figure 2A). Literature has established higher expression of FZD7 receptors in MDA-MB-231 versus MCF10A cells through immunocytochemistry and western blotting.^{26, 32} To visualize NP binding to each cell type, we used two-photon microscopy to detect the photoluminescence emitted by NS in response to their excitation with a femtosecond-pulsed laser tuned to the peak plasmon resonance wavelength.⁴³ As expected, the FZD7-coated NPs (FZD7-NS and Combo-NS) exhibited greater binding to the MDA-MB-231 cells than the PEG-NS and βcat-NS controls (Figure 2B). We also observed minimal attachment of all four NP types to control MCF10A cells.

To corroborate the imaging data, we used flow cytometry to examine cellular uptake of NPs passivated with Cy5-tagged mPEG-SH, where we ensured the NPs had equivalent Cy5 loading (Figure S5). As with the multiphoton microscopy data, flow cytometry showed the antibody-loaded NPs were bound or internalized by MDA-MB-231 cells more than the PEG-NS and βcat-NS (Figure 2C,D), as evidenced by greater Cy5 signal (which was taken to correlate with NP binding/uptake). Additionally, the Cy5 signal was higher in MDA-MB-231 (Figure 2C,D) cells than in MCF10A cells, indicating selective binding (Figure S6; Figure S7). Uptake analysis at 4 and 8 h of NP incubation exhibited similar trends where FZD7-NS and Combo-NS were taken up to a greater extent than PEG-NS or βcat-NS (Figure S8). Generally, we observed more binding/uptake in MCF10A cells by flow cytometry than multiphoton microscopy. We posit that the Cy5 tag on the PEG contributed to an increase in non-specific cell binding. Nonetheless, at the 1 h timepoint there is ~1.5 fold more binding of antibody-loaded NPs to target MDA-MB-231 cells than to control MCF10A cells.

2.3. Antibody/siRNA Nanoconjugates Suppress Several Wnt Target Genes in TNBC Cells

Following assessment of NP uptake, we next evaluated gene regulation at the mRNA level via RT-qPCR. In addition to analyzing expression of β-catenin, the key mediator of Wnt signaling, we also evaluated several downstream target genes (Axin2, CCND1, C-myc, Nanog, Snail, and Survivin) (primer sequences provided in Table S2). In initial experiments, we validated the efficacy of the β-catenin siRNA by transfecting it or scrambled control siRNA into MDA-MB-231 cells with Dharmafect, and RT-qPCR analysis confirmed the siβ-catenin could suppress Wnt target genes (Figure S9). We then evaluated the NPs’ ability to suppress Wnt target genes in MDA-MB-231 cells (Figure 3A). At 72 h post-addition to cells, all three NP types decreased mRNA expression of the evaluated genes relative to untreated cells (NT=non-treated), and Combo-NS provided the most robust inhibition of β-catenin, C-myc, and Snail (Figure 3B). Impressively, the NPs yielded similar magnitude of gene knockdown as the transfection experiments while delivering ~4X less RNA (the siRNA dose in the NP experiments was 12 nM, compared to 50 nM in the transfection studies), indicating the high efficiency of the system. Analysis at 24 and 48 h also revealed the greatest level of gene silencing for Combo-NS, with more robust effects observed at 48 h compared to 24 h (Figure S10). PEG-NS did not alter gene expression in cells at any timepoint, confirming the gene regulation is due to the delivered FZD7 antibodies and β-catenin siRNAs (Figure S11).

2.4. Wnt Inhibitory NPs Repress Proliferation of TNBC Cells

Cellular proliferation is a hallmark of cancer and is associated with genes including C-myc and CCND1. Since the Wnt inhibitory NPs reduced expression of these genes, we evaluated their impact on cell proliferation using an EdU assay (Figure 4A). After 48, 72, and 96 h, βcat-NS treated cells had comparable proliferation rates as the NT group. In contrast, both FZD7-NS and Combo-NS decreased cell proliferation at 48 h, but only Combo-NS maintained suppression of proliferation through 96 h (Figure 4B; complete statistics provided in Figure S12). This affirms that co-delivery of β-catenin siRNAs and FZD7 antibodies has potential for improved therapeutic effects. In future work it would be interesting to assess the NPs’ effect on cell cycle arrest or evaluate the impact of multiple NP doses, which could correlate to in vivo and clinical studies that typically use multiple rounds of therapy administration.

2.5. Cellular Migration is Reduced by Treatment with Wnt Inhibitory NPs

Cellular migration and invasion contribute to cancer metastasis and are also regulated by Wnt signaling. To determine whether the Wnt inhibitory NPs could suppress cellular migration, we pre-treated GFP-expressing MDA-MB-231 cells with NPs for 96 h and then reseeded 10,000 cells in transwell porous membranes to evaluate their migratory behavior over the following 6–48 h (Figure 5A). At 6 h post reseeding, all three Wnt inhibitory NPs reduced cell migration across the membrane compared to the NT group, with the most dramatic reduction observed in Combo-NS treated cells (Figure 5B). Over time, the cells treated with Combo-NS continued to migrate more slowly than cells in the other treatment groups (Figure 5B,C). In fact, while the effects of FZD7-NS and βcat-NS were diminished by 16 h, the Combo-NS suppressed migration through 48 h. It should be noted that at this later timepoint the cells may have divided, such that the results indicate effects on both migration and proliferation, but timepoints before 24 h demonstrate the effect on migration alone.

Interestingly, when optimizing conditions for the migration assay, we had pre-treated cells for 72 h and reseeded them into transwell inserts at densities of 5k, 10k, or 20k cells per well prior to analyzing migration. At seeding densities of 5k or 10k cells per insert, all three NPs reduced migration to a similar extent (Figure S13; Supporting Information). However, at the higher seeding density of 20k cells/insert, the FZD7-NS and Combo-NS had more dramatic impacts on migration than the βcat-NS. Collectively, these data confirm that NP-mediated Wnt inhibition can reduce cell migration, with co-delivery of β-catenin siRNAs and FZD7 antibodies being advantageous therapeutically.

2.6. Wnt Inhibitory NPs Hinder Spheroid Formation

Wnt signaling is also associated with cancer stemness, which contributes to tumor initiation, treatment resistance, relapse, and metastasis.⁴⁴ Two stemness genes linked to aggressive cancer phenotypes are Nanog and Survivin,^45–47 which were down-regulated by the Wnt inhibitory NPs (Figure 3). Therefore, we evaluated the impact of NP treatment on stemness/self-renewal through spheroid formation assays (Figure 6A). Briefly, MDA-MB-231 cells were pre-treated with NPs for 24 – 72 h and then reseeded in sphere forming media in low-adhesion u-bottom well plates to facilitate spheroid formation. After one week, the spheroids were imaged and their metabolic activity assessed via an alamarBlue^™ assay. This revealed that an NP treatment time of 72 h is needed to impair sphere viability (Figure 6B; statistics in Figure S14). Under these conditions, spheroid metabolic activity was reduced by ~34% for Combo-NS, ~18% for βcat-NS, and ~13% for FZD7-NS compared to the NT group (Figure 6B). Upon imaging the spheres, we found that all three NPs physically hindered spheroid formation, with some NP-treated samples forming loose cell clusters rather than spheres (Figure 6C, top row) and others forming less dense spheres as indicated by reduced opacity (Figure 6C, bottom row). The appearance of loose cell clusters indicates the NPs have reduced the cells’ ability to condense into compact spheroids, while the appearance of less dense spheres indicates the NPs’ have reduced the proliferation of cells within the spheres (which matches the impact on spheroid metabolic activity observed in Figure 6B). As with the other in vitro assays, Combo-NS had the most dramatic effects on spheroid formation and viability, indicating a potential benefit of suppressing Wnt signaling in TNBC cells at both the receptor and effector level.

2.7. Wnt Inhibitory NPs Reduce the Growth of Pre-Existing Metastases in an Experimental Lung Metastasis Model

To test the NPs’ ability to reduce the growth of established metastases, we used an experimental metastasis model in which firefly luciferase-expressing MDA-MB-231 cells are inoculated into the tail vein of female nude mice leading to growth of metastatic tumor nodules in the lungs. Once lung metastases formed, mice received saline or NPs intravenously, for 6 total injections, and bioluminescence imaging was used to monitor tumor growth (Figure 7A). For these studies, the NPs were coated with FZD7 antibodies using 2 kDa orthopyridyl disulfide-PEG-succinimidyl valerate (OPSS-PEG-SVA) linkers rather than 5 kDa OPSS-PEG-SVA linkers used in other experiments (note that the 5 kDa linkers increase antibody accessibility, which is why they were used in all other experiments). Using the dose schedule depicted in Figure 7B, we found that all three Wnt inhibitory NPs decreased metastasis burden, with the lowest luminescent signal observed following treatment with Combo-NS (Figure 7C,D). While saline-treated mice exhibited large increases in mean luminescence versus time indicative of metastatic growth, Combo-NS-treated mice had stable or significantly reduced tumor signal (Figure 7D). Consequently, there was a large difference in metastatic burden between these groups at the study end. After animal sacrifice, we quantified gold content in excised lungs and other tissues via ICP-MS and found that the Combo-NS and FZD7-NS exhibited 4–5X higher accumulation in the lungs than the βcat-NS (Figure 8A, Figure S15). Darkfield microscopy images of lung sections showed the antibody-modified NPs were associated primarily with tumor tissue rather than with adjacent normal lung (Figure 8B). These data indicate antibody incorporation into the NPs increased delivery to the desired site and support the potential of Wnt inhibitory NPs to treat pre-existing TNBC metastases. Note that while Combo-NS and FZD7-NS exhibited greater tumor delivery than βcat-NS, all three formulations reduced metastatic growth to a similar extent. This suggests that the siRNA might be a more potent cargo, such that βcat-NS could inhibit disease burden to a similar degree despite lower accumulation in the tumor-bearing lungs. Importantly, brightfield microscopy of hematoxylin and eosin (H&E)-stained tissues (heart, spleen, kidney, and liver) showed no noticeable differences in structure between saline and Combo-NS treated mice (Figure S16), suggesting the NPs are well tolerated under the dosing schedule and concentration tested here.

2.8. Wnt Inhibitory NPs Limit Primary Tumor Growth, Recurrence, and Spontaneous Metastasis in a Syngeneic Murine TNBC Model

Given the promising observations in the experimental lung metastatic model, we wondered if the Wnt inhibitory NPs would also be effective against TNBC in a syngeneic murine model that uses immune competent rather than nude mice. In shifting to this model, we needed to confirm FZD7 is expressed in 4T1 cells, a murine cancer cell line that accurately mimics the aggressive nature of human TNBC and which can spontaneously metastasize to distant sites after implantation in the mammary fat pad of Balb/c mice (Charles River). Using immunocytochemistry (ICC), we confirmed FZD7 is more abundantly expressed on 4T1 cells than noncancerous MCF10A cells (which also confirmed our antibody could bind murine FZD7) (Figure S17A). We also had to alter the siRNA sequence to target murine β-catenin (Table S1), and hence used RT-qPCR to validate that the new sequence could suppress β-catenin in 4T1 cells following transfection with Lipofectamine (Figure S17B; primers in Figure S17C). With this confirmation, we re-formulated the Wnt inhibitory NPs using both antibodies and siRNAs designed to target murine FZD7 and β-catenin. The characteristics of these NPs are provided in the Supporting Information (Table S3). We also validated that the reformulated NPs could suppress proliferation of murine 4T1 cells using an EdU assay (Figure S18) before moving into in vivo experiments.

Before attempting NP treatment of the syngeneic TNBC model, we first characterized the rate of tumor growth and metastatic spread as a function of the number of luciferase-expressing 4T1 cells initially implanted in the 4^th inguinal mammary fat pad (Scheme S2, Figure S19, Figure S20). Based on the findings, we injected 100k cells in the mammary fat pad for therapeutic studies. Two different NP treatment schedules were evaluated in two separate studies (Figure 9A). In both schedules, NP treatment began when tumor volume reached ~100 mm³ (defined as Day 0), and the second NP dose was administered on Day 7. Primary tumors were removed from mice on Day 13, after which the treatment schedules differed in terms of frequency and total number of injections (Figure 9A). Since the treatment from Day 0 through 13 was the same in both schedules, the datasets across the two studies were combined (allowing for n=12–14 mice/group) to understand the effect of NP treatment on primary tumor growth. Excitingly, all three Wnt inhibitory NPs significantly reduced primary tumor growth compared to saline-injected controls (Figure 9B, Left). While tumors in saline-treated mice grew by an average of 315 mm³ between Day 0 and 13, those in the NP-treated mice grew by an average of 184–201 mm³, which corresponds to an inhibited growth rate of 36–42% (Figure 9B, Right). After surgical excision, primary tumors were sectioned and stained for β-catenin with hematoxylin as a counterstain to quantify the average percent positive cells (Figure 9C). Combo-NS significantly reduced the fraction of β-catenin positive cells to ~47%, compared to ~71% for saline treated mice, 60% for FZD7-NS treated mice, and 56% for βcat-NS treated mice. This demonstrates the NPs suppress intra-tumoral Wnt signaling as expected.

Following primary tumor removal, tumor regrowth and lung metastatic spread was monitored via IVIS imaging, and the datasets from the mice receiving the different NP dose schedules were kept separate (n=6–7 mice/group). All three NPs slowed recurrence compared to saline-treated mice under both dosing schedules based on average luminescent signal, although the differences were not significant at the 95% confidence level, likely due to the high variability in recurrent tumor burden, particularly in the saline control (Figure 10A). We expected that schedule 2, which had more NP doses at a higher frequency, would slow the growth of recurrent tumors compared to schedule 1, but we found no major differences in rate of recurrence between the two schedules (Figure 10A). This may have been due to the inherent variability of tumor growth/aggressiveness between the two studies. To better compare different treatment schedules in the future, we would perform concurrent analysis in a single cohort of mice inoculated with the same batch of cells on the same date. Performing less frequent dosing or adjusting the NP concentration, along with increasing the sample size (number of animals) in the study, may also help tease out differences between the NP treatments in terms of their anti-recurrence effects.

Evaluation of lung metastasis revealed more differences between the NP types and treatment schedules. For both schedules, lung metastasis occurred more slowly in mice treated with Wnt inhibitory NPs versus saline (Figure 10B). Mice treated with NPs didn’t have metastatic signal appear until around day 25 compared to ~day 20 for the saline group (Figure S21). While the metastatic burden (based on luminescent signal) was equivalent between NP groups for mice that formed distant disease, it is noteworthy that the fraction of mice that formed metastasis differed based on treatment type and dosing schedule. Under schedule 1, approximately 15% of mice treated with βcat-NS or FZD7-NS did not form metastasis, compared to 33% of mice treated with Combo-NS (Figure 10C). This demonstrates a distinct benefit of antibody/siRNA co-delivery under this treatment schedule. With schedule 2, both FZD7-NS and Combo-NS prevented metastasis in 15% of mice, while βcat-NS did not prevent metastasis in any mice. As noted before, it is difficult to directly compare the two schedules given that they were evaluated in two distinct cohorts of mice, but it is promising that Combo-NS and FZD7-NS prevented metastasis in a subset of mice under both schedules, while βcat-NS prevented metastasis in a subset of mice under schedule 1. We postulate that prevention of lung metastasis could be further enhanced by removing primary tumors at an earlier stage and beginning NP treatment sooner post initial cell inoculation in the mammary fat pad.

To shed light on the therapeutic results, the accumulation of the NPs in primary tumors excised at Day 13, as well as in recurrent tumors and metastatic lungs excised at animal sacrifice, was assessed by measuring gold content in the tissues via ICP-MS. As expected, the NPs that incorporated FZD7 antibodies (i.e., FZD7-NS and Combo-NS) exhibited increased delivery to primary tumors (Figure 11A). In the lungs, Combo-NS exhibited the most accumulation under schedule 1, and the least accumulation under schedule 2, although the level of accumulation was not significantly different amongst the three NP types under either schedule (Figure 11B). In recurrent tumors, both FZD7-NS and Combo-NS exhibited at least 2X more accumulation than βcat-NS, with the overall accumulation being greater under schedule 2 than schedule 1 (Figure 11B). We also analyzed gold content in major clearance and non-clearance organs (Figure S22, S23; Supporting Information), which indicated distribution to the liver and spleen, with lesser delivery to intestines and other tissues. Importantly, microscopic evaluation of H&E-stained liver, spleen, kidney, and heart (Figure S24) showed that all three NPs were well tolerated in this model, matching the results obtained in the experimental metastasis model (Figure S16). Animal weight versus time (Figure S25) was also consistent between treatment groups, suggesting the NPs exhibit satistfactory biosafety under the dosing conditions tested in this work.

3. Conclusion

These studies demonstrate that the dual loading of antibodies and siRNAs on NS can be achieved and carefully controlled by tailoring synthesis parameters, and that co-delivery of antibodies and siRNAs to suppress Wnt signaling at both the receptor and effector level is advantageous therapeutically. Prior work has shown FZD7-NS can suppress Wnt signaling in TNBC cells, resulting in impaired cell function when applied alone or in combination with freely delivered drugs.^{32, 33} Here, we developed NS for co-delivery of FZD7 antibodies and β-catenin siRNAs and compared their effect to NS that deliver antibodies or siRNAs individually.

Both FZD7-NS and Combo-NS exhibited preferential TNBC cell binding/uptake, as evidenced by multiphoton microscopy and flow cytometry studies (Figure 2). This improved binding translated to increased accumulation of FZD7-NS and Combo-NS in established TNBC lung metastasis in nude mice (Figure 8), as well as in primary tumors and recurrent tumors in the mammary fat pad of Balb/c mice (Figure 11). In vitro, all three Wnt inhibitory NPs reduced the expression of several genes that are implicated in cell proliferation, survival, stemness, and motility (Figure 3), which allowed them to impair cell proliferation, migration, and spheroid formation (Figure 4–6). From these functional assays we found a trend in which the effects of Combo-NS were sustained for longer periods of time than the effects of βcat-NS or FZD7-NS. In vivo, the Wnt inhibitory NPs were able to effectively treat pre-existing lung metastases in nude mice (Figure 7), with the most robust inhibition observed for Combo-NS. Additionally, all three Wnt inhibitory NPs reduced the growth of primary tumors, decreased the rate of recurrence, and limited the formation of spontaneous lung metastases in a syngeneic murine model (Figure 9–10). Immunohistochemical examination of the primary tumors confirmed all three NPs decreased β-catenin expression, with Combo-NS having the most significant impact (Figure 9C).

Collectively, these studies demonstrate Wnt inhibitory NPs are promising tools for targeted gene regulation of TNBC. Future studies that build on this work could study the influence of antibody and siRNA loading density on the cellular uptake, biodistribution, gene regulation potency, and tumor/metastasis inhibition capabilities of the system. We postulate that lower antibody loading and higher siRNA loading would provide increased efficacy. In performing this work, detailed studies should be completed to reveal the intracellular trafficking of Combo-NS. We presume the NPs escape endosomes to reach the cytosol based on their observed gene silencing capacity (Figure 3, Figure 9C), but a mechanistic examination of their colocalization with different intracellular compartments would validate this hypothesis. In prior work, we have shown that FZD7-NS do not heavily co-localize with lysosomes upon cellular uptake,³³ providing scientific basis for the ability of Combo-NS to exhibit similar endosomal escape.

Future in vivo work could explore how the frequency, number, timing, and concentration of NP doses correlate with response, as well as evaluate the biocompatibility of the system through analysis of blood chemistry, serum cytokines, liver enzymes, and more. This information would be insightful towards future clinical translation. Future studies could also include comparison to a control of PEG-NS to ensure the observed effects in vivo are due to the delivered siRNAs and antibodies and not due to the carrier material. While PEG-NS did not inhibit Wnt target gene expression in this study or in our prior research³², literature has shown that gold NPs can be self-therapeutic in cancer and psoriasis models, for example by disrupting cellular crosstalk in the disease microenvironment.^48–51 Hence, it will be important to understand the potential impact of the core NS on TNBC progression. Lastly, future studies should quantify the expression of Wnt target genes in both recurrent tumors and metastatic lesions to confirm the NPs are acting through the expected molecular mechanism of action and verify the results observed in primary tumors in this study.

In conclusion, the results presented here indicate Wnt inhibitory NPs are promising gene regulatory agents that may offer a new targeted arsenal in the fight against TNBC. With continued development, this platform could advance the treatment of TNBC and other cancers whose growth and dissemination are diven by hyperactive Wnt signaling mediated through overexpressed FZD receptors.

4. Experimental Section

4.1. Nanoshell Synthesis and Biomolecule Functionalization

NS were synthesized by the Oldenburg method,⁴⁰ as expanded upon in the Supporting Information. Synthesized NS were stored in milliQ water at ~6×10⁹ NS mL⁻¹ (OD^{810 nm} = 2) at 4 °C. Prior to bioconjugation, the NS were treated with 0.1% diethyl pyrocarbonate (DEPC) (Sigma) for 3 days rocking at 37 °C to render them RNase-free.

Human and mouse cross reactive FZD7 antibodies (LSBio; provided at 1 mg mL⁻¹ in 100–200 μL volumes) were diluted, purified, and resuspended in 1XPBS as detailed in the Supporting Information. To facilitate NS conjugation, the FZD7 antibodies were incubated with 5 kDa orthopyridyl disulfide-PEG-succinimidyl valerate (OPSS-PEG-SVA, Laysan Bio) in 100 mM sodium bicarbonate at a 2:1 PEG:antibody molar ratio, followed by rocking overnight at 4 °C. The PEGylated antibodies were aliquoted into microcentrifuge tubes and stored at −20 °C until used for NS conjugation.

β-catenin siRNA oligonucleotides were purchased as single strands from Integrated DNA Technologies (Table S1). The sense strands had 3′ thiol modification to allow functionalization to the gold surface of NS. Complementary sense and antisense strands were mixed in duplex buffer (IDT) in equimolar amounts, heated at 95 °C for 5 min in a thermomixer, and slowly cooled to 37 °C for over 1 h to facilitate duplexing. Duplexed siRNA was aliquoted into small volumes and stored at −80 °C until use.

To coat NS with biomolecules (Figure 1A), the NS were diluted to OD^{810 nm} = 1.5 in RNase-free milliQ water, then PEGylated FZD7 antibodies were added at a ratio of 1,200 or 1,000 antibodies per NS to prepare FZD7-NS or Combo-NS, respectively. After rocking at 4 °C for 1 h, the solution was briefly vortexed and bath sonicated followed by addition of 0.2% Tween-20 and NaCl (at concentrations of 40 mM and 12 mM for Combo-NS and βcat-NS, respectively). After 5 min incubation at RT, siRNA duplexes were added at 0.25 or 0.1 nmol siRNA per mL of NS for Combo-NS and βcat-NS, respectively. The solution was vortexed and rocked at 4 °C for 6 h, with brief vortexing and bath sonication performed every 1.5 h. At every 3 h NaCl was added to an end concentration of 100 or 200mM (for Combo-NS and βcat-NS, respectively), which is detailed further in the Supporting Information. The solution rocked at 4 °C for at least 12–14 h, then the functionalized NS were vortexed and bath sonicated. Lastly, 5 kDa mPEG-SH (Laysan) was added to the NS at a concentration of 10 or 20 μM (for βcat-NS and Combo-/FZD7-NS, respectively). After rocking 4 h at 4 °C, unbound biomolecules were removed via centrifugation (performed thrice at 500 g). After removal of the supernatant, the NPs were diluted in RNase-Free 1XPBS with 100X less volume than the starting NS volume. The NS conjugates were stored in LoBIND centrifuge tubes (FisherSci) at 4 °C until use. All NS concentrations were calculated based on Beer’s law using the peak extinction (~810 nm) measured on a Cary 60 UV-visible spectrophotometer. The concentrations of antibodies, siRNA duplexes, mPEG-SH, Tween-20, and NaCl differed in the preparation of each NP type to yield equivalent antibody loading between Combo-NS and FZD7-NS and equivalent siRNA loading between Combo-NS and βcat-NS. We performed extensive studies investigating parameters that influence biomolecule loading on NS, which are detailed in the Supporting Information.

4.2. Nanoparticle Characterization

The hydrodynamic diameter and zeta potential of the bare and functionalized NS were measured using an Anton Paar Litesizer with at least 3 different batches of NPs diluted in milliQ water. At least 300 μL of sample at 3×10⁹ NS mL⁻¹ (OD^{810 nm} = 1) was used in the Litesizer Uvettes and Omega Cuvettes. Dynamic Light Scatting (DLS) replicates had 60 runs each while zeta potential replicates included 1,000 runs. RNA loading on NS was measured using a Quant-iT OliGreen^™ ssDNA quantification kit as detailed in the Supporting Information and previously reported.⁴² Antibody loading on NS was quantified using an ELISA assay,⁴¹ which is described in the Supporting Information.

4.3. Cell Culture and NP Treatment Experimental Setup

MDA-MB-231 TNBC cells purchased from American Type Culture Collection (ATCC) were cultured in Dulbecco’s Modified Eagle Medium (DMEM) (VWR) supplemented with 10% fetal bovine serum (FBS) (Gemini Bio) and 1% penicillin-streptomycin (VWR). Cells were cultured in T75 cell culture flasks (VWR), passaged with 0.25% trypsin-EDTA (VWR), and incubated at 37 °C in a 5% CO₂ humidified environment. Cells were cultured to 80–90% confluency prior to plating for experiments, used within passage numbers of 40 – 55, and counted using a glass slide hemocytometer for all experiments. GFP-expressing MDA-MB-231 cells (Angio-Proteomie) were cultured with the same conditions. Noncancerous MCF10A breast epithelial cells were provided by Dr. Kenneth Van Golen and cultured in 1:1 DMEM and F12 base medium supplemented with 5% FBS (Gemini Bio), 1% penicillin-streptomycin (VWR), 10 μg mL⁻¹ insulin (ThermoFisher), 0.5 μg mL⁻¹ hydrocortisone (Sigma), 50 μg mL⁻¹ bovine pituitary extract (ThermoFisher), 20 ng mL⁻¹ epidermal growth factor (ThermoFisher), and 100 ng mL⁻¹ cholera toxin (Sigma). MCF10As were used within passage numbers 20 – 30.

The general experimental setup described below was used for studies to assess the impact of co-delivered NPs on TNBC cells. Cells were plated in well plates with complete cell culture media and placed in a humidified incubator at 37 °C, 5% CO₂ for at least 20 h. The cells were then incubated with FZD7-NS, βcat-NS, Combo-NS, or full media at 3×10⁹ NPs mL⁻¹ (OD^{810 nm} = 1) or 1.5×10¹⁰ NPs mL⁻¹ (OD^{810 nm} = 5) for distinct periods of time depending on the experiment. OD^{810 nm} = 1 corresponds to ~ 0.8 nM FZD7 antibodies and ~12 nM siRNA. OD^{810 nm} = 5 corresponds to ~ 4 nM FZD7 antibodies and ~64 nM siRNA.

4.4. Evaluation of NP Binding to TNBC Cells versus Non-cancerous Cells

To evaluate the targeting afforded by coating NS with FZD7 antibodies, target MDA-MB-231 and control MCF10A cells were seeded in Nunc Lab-Tek #1 8-well chamber slides (Thermo) at 250,000 and 100,000 cells per well, respectively, and left to adhere for at least 20 h while incubating at 37 °C, 5% CO₂. The next day media was removed and 300 μL NPs (PEG-NS, FZD7-NS, βcat-NS, and Combo-NS) was added at OD^{810 nm} = 1 in complete media for both cell lines. The cells were incubated at 37 °C for 1 h, the media was gently removed, and the cells were rinsed thrice with warm Dulbecco’s PBS (DPBS, ThermoFisher). Cells were then fixed with 4% paraformaldehyde in 1X PBS for 15 min at RT and neutralized with 1X PBS afterwards. Next, the cell membranes were dyed using CellVue Claret Red Membrane Dye (Sigma) for 8 – 10 min at RT covered, with the dye prepared per manufacturer’s instructions. The dye was neutralized with 1% w/v BSA in 1X PBS. The dye solution and neutralizer were than aspirated and cells covered with 1X PBS. Slides were imaged using a Zeiss LSM 880 Multiphoton Microscope with a 20x/0.8 NA water objective, which was immersed in a droplet of water to provide the correct contrast and visualization upon connection to the slides. The NPs in the slide samples were excited by the multiphoton laser tuned to λ_excitation = 800 nm with a pinhole of 1.57 AU and a detection range λ_emission of 400 – 550 nm. The CellVue Claret Red-labeled cell membranes were visualized at λ_excitation = 655 nm and λ_emission of 675 nm.

To analyze cellular uptake of NPs via flow cytometry, the NPs were passivated with Cy5-PEG-SH instead of mPEG-SH. The amount of Cy5-PEG-SH added to each NP formulation was adjusted to ensure equal Cy5 signal across groups. Following synthesis and purification of the Cy5-tagged NPs, MDA-MD-231 and MCF10A cells were seeded in 96 well plates at 75,000 and 50,000 cells per well respectively. The following day NPs (no NP, PEG-NS, FZD7-NS, βcat-NS, and Combo-NS) were added to the wells at OD^{810 nm} = 1 in complete media for both cell lines. The cells incubated for 1, 4, and 8 h at 37 °C. Following each incubation period, the media was gently removed, and the cells rinsed thrice with warm DPBS with thorough pipetting to dislodge any unbound/non-internalized NPs. Cells were then trypsinzed, neutralized, and transferred to Eppendorf tubes for flow cytometry performed with an Acea Novocyte 2060 Flow Cytometer. The parameters were set to 100 μL min⁻¹ to a cell cut off at 10,000 cells gated for singlet cells. Cells were then further gated using the APC-Cy7 filter where cells were excited at 640 nm and detected using the 780/60 nm filter to assess the shifts in Cy5 signal. In processing the data, the raw median value of Cy5 signal was averaged in triplicate and the NT (non-treated) group values were subtracted from each NP group to evaluate the shift from the baseline cells, which should have minimal Cy5 signal.

4.5. RT-qPCR Analysis of the mRNA Expression of Wnt Related Genes

RT-qPCR was utilized to assess the impact of NP treatment on mRNA expression of relevant Wnt-associated genes. MDA-MB-231 cells were plated at 40,000 cells/well in 12-well plates prior to NP treatment. Cells were treated with no NPs, PEG-NS, FZD7-NS, βcat-NS, or Combo-NS at OD^{810 nm} = 1. Following 24, 48, or 72 h, RNA was extracted using the Isolate II RNA Mini Kit (Bioline). Modifications to the protocol include omitting the β-mercaptoethanol addition. RNA concentration was determined by reading the sample absorbance at 260 nm using a Take3 Plate on a Synergy H1 plate reader. RT-qPCR was then performed with a concentration of 75 ng RNA per 2 μL using the SensiFAST SYBR One-Step Master Mix (Bioline) on a LightCycler96 (Roche). The mRNA expression of each target gene was normalized to that of GUSB and further normalized to NT (no treatment) group. Primer sequences are listed in Table S2. These experiments were performed in triplicate and analyzed using a one-way ANOVA with posthoc Tukey.

4.6. Analysis of Cellular Proliferation Following NP Treatment

An EdU assay was employed to evaluate MDA-MB-231 cell proliferation after NP treatment. Cells were plated at 10,000 cells per well in 24-well plates prior to NP treatment. Each treatment group had 2 wells to serve as technical replicates for individual experiments. Cells were treated with no NPs, FZD7-NS, βcat-NS, or Combo-NS at OD⁸¹⁰ nm = 5. Following 48 – 96 h, the cells were assessed for proliferation using the Click-iT^™ EdU Cell Proliferation Kit (ThermoFisher) per manufacterer’s instructions. These experiments were performed in triplicate (i.e., with three biological replicates) and the proliferation in NP-treatment groups was normalized to that in the NT control prior to analysis by one-way ANOVA with posthoc Tukey.

4.6. Analysis of Cellular Migration Following NP Treatment

A transwell migration assay was employed to evaluate the invasive and migratory capacity of MDA-MB-231 cells after NP treatment. GFP-expressing MDA-MB-231 cells were plated at 10,000 cells/well in 48-well plates for migration studies. Each treatment group had technical replicates of 2 wells each. Cells were treated with no NPs (NT), FZD7-NS, βcat-NS, or Combo-NS at OD^{810 nm} = 5. At 96 h post incubation, the cells were trypsinized, pelleted, resuspended in non-supplemented DMEM, and counted. The cells were then reseeded at 10,000 cells onto the tops of black transwell inserts (Corning) in 24-Well Glass Bottom Black Walled Plates (CellVis); non-supplemented DMEM was in the top of the insert and complete media was in the bottom of the wells. The cells were left to migrate across the 8.0 μm polyester membranes over 48 h. The cells were imaged at 6, 16, 24, and 48 h using the GFP filters and tile imaging on an Axioobserver Z1 Inverted Fluorescent Microscope (Zeiss). These experiments were performed in triplicate and analyzed by one-way ANOVA with posthoc Tukey.

4.8. Analysis of Spheroid Formation Following NP Treatment

A spheroid formation assay was employed to evaluate the self-renewal capacity of MDA-MB-231 cells after NP treatment. MDA-MB-231 cells were plated at 20,000 cells/well in 24-well plates and treated with no NPs, FZD7-NS, βcat-NS, or Combo-NS at OD^{810 nm} = 5. At 24, 48, and 72 h post incubation, the cells were lifted with trypsin, pelleted, resuspended MammoCult Basal Medium (StemCell) supplemented with the provided MammoCult Proliferation Supplement, 4 μg mL⁻¹ heparin (StemCell), and 0.48 μg mL⁻¹ hydrocortisone (StemCell). The cells were then reseeded at 10,000 cells in triplicate in low adhesion U-bottom 96 well plates (BrandTech) treated for suspension tissue cultures. The cells were placed in a 37 °C, 5% CO₂ humidified incubator and allowed to form spheroids over 7 days, with brightfield imaging performed every other day using an Axioobserver Z1 Inverted Fluorescent Microscope (Zeiss) and a 10X objective. Following 7 days of spheroid formation, the metabolic activity of the spheroids was assessed via an alamarBlue^™ assay in which 10% of alamarBlue^™ was added to the spheroid containing media and left to incubate. After 24 h, the absorbance was read at 570 nm using the Synergy H1 plate reader. These experiments were performed in triplicate and analyzed by one-way ANOVA with posthoc Tukey.

4.9. Monitoring Tumor Growth and Metastasis via Caliper Measurements and IVIS Imaging

All in vivo experiments described in Section 4.9 through Section 4.12 were performed under animal use protocol (AUP) 1318 approved by the University of Delaware Institutional Animal Care and Use Committee. Tumor growth in the mammary fat pad (m.f.p.) was monitored with physical caliper measurements and/or IVIS imaging of bioluminescent cells. IVIS imaging was also used to monitor lung metastasis. For the m.f.p. measurements, a digital caliper was used to measure the length (longer) and width (shorter) of the tumor mass. The tumor volume was calculated with the following equation: (length × width²)/2. For IVIS imaging of the tumors and any metastatic lesions, the mice were injected intraperitoneally with 150 mg kg⁻¹ of D-luciferin (Biotium) diluted in saline. At 10 min post injection, the mice were anesthetized with isoflurane and placed belly up in an IVIS Lumina III (fitted with an additional field of view lens) to read the exposure with the following parameters: imaging mode(s) = luminescence and photograph, exposure time = auto, binning = medium, F/stop = 1, excitation filter = closed, emission filter = open, overlay = checked, field of view = E, subject height = 1.5 cm, focus = use subject height, and bin = medium. Acquisition of the primary tumors was typically taken first. The chest area was then imaged while hiding the primary tumors with a black cover to allow for detection of metastases.

4.10. Evaluation of Therapeutic Impact on an Experimental TNBC Lung Metastasis Model

We evaluated the NPs’ impact on an experimental lung metastasis model, in which 10⁵ Firefly luciferase expressing MDA-MB-231 cells (ATCC) were administered into the vail vein of 6 – 8-week-old female nude mice. Therapy began within 2 – 3 weeks after TNBC cell delivery after confirming the formation of lung metastases via IVIS imaging. Mice were randomized to groups to receive 100 μL Saline, Combo-NS, FZD7-NS, or βcat-NS intravenously following the dosing schedule in Figure 7B. The concentration corresponded to 250 pM NS, equal to 300 nM siRNAs and 15 nM antibodies. During the study, metastasis burden in the lungs was monitored by bioluminescence imaging. At the end of the study (day 59), mice were euthanized by carbon dioxide asphyxiation followed by cervical dislocation and the lungs excised for ICP-MS measurement of gold content as well as darkfield microscopy visualization of NP location in histological sections. Darkfield microscopy was performed using a darkfield condenser attached to a Zeiss Axioobserver Z1 microscope. To understand biosafety, the heart, kidney, liver, and spleen from mice treated with saline or Combo-NS were sectioned, stained with hematoxylin and eosin, and imaged on a Zeiss Axioobserver Z1 microscope (Figure S16). For this histology, excised organs were rinsed in 1× PBS and placed into embedding cassettes. Cassettes were then placed in 4% paraformaldehyde (in 1× PBS) for 72 h while rocking at 4 °C. Next, cassettes were rinsed three times in 70% ethanol for 15 min at 4 °C while rocking. Tissues were then stored in fresh 70% EtOH solution at 4 °C prior to histological preparation and processing. The fixed tissues were processed, embedded with paraffin, and cut into 5 μm slices using a rotary microtome. Sample sections were deparaffinized with xylene and rehydrated prior to hematoxylin staining and subsequent counterstaining with eosin.

4.11. Evaluation of Therapeutic Impact on Primary Tumors, Recurrence, and Spontaneous Metastasis in a Syngeneic Murine TNBC Model

A preliminary study was performed to determine the rates of tumor growth and spontaneous metastasis in 6–8 week old female Balb/c mice inoculated with 5k-100k 4T1-Luc2 cells (ATCC) in the 4^th mammary fat pad (Scheme S2, Figure S19, Figure S20). Based on the results, a therapeutic study was performed in 6–8 week old female Balb/c mice inoculated with 100K 4T1-Luc2 cells in the 4^th mammary fat pad. Once primary tumors reached ~100 mm³ in volume based on caliper measurements (defined as Day 0), NP treatment began where mice were randomized to receive 100 μL saline, Combo-NS, FZD7-NS, or βcat-NS intravenously at a concentration of 1.9×10¹¹ NS mL⁻¹ (OD^{810 nm} = 60), which corresponds to ~73 nM FZD7 antibodies and ~640 nM siRNA. After two injections on Day 0 and Day 7, the primary tumors were removed on Day 13 for assessment of gold content by ICP-MS (described in Section 4.12). Additionally, sections of primary tumors were stained for β-catenin to evaluate Wnt signaling inhibition. In brief, tumors were paraffin embedded and cut into 5 μm sections as described in Section 4.10. The tumor sections were stained for β-catenin using rabbit anti-mouse β-catenin antibodies (Cell Signaling, 1:400 dilution), followed by addition of peroxidase-conjugated secondary antibodies (1:800 dilution) and 3,3’-diaminobenzidine (DAB) to produce a brown signal indicative of β-catenin. Sections were counterstained with hematoxylin, coverslipped, and imaged using an EVOS M5000 microscope. Images of β-catenin-stained tumor sections (3 per group) were analyzed with QuPath to calculate the average percent positive cells in each treatment group.

After primary tumors were removed from mice, NP treatment continued under two dose schedules (Figure 9A). During this time, tumor recurrence in the m.f.p. and metastasis in the lungs were monitored using bioluminescence imaging. At the end of the study, mice were euthanized by carbon dioxide asphyxiation followed by cervical dislocation, and tissues collected for analysis of gold content by ICP-MS. Sections of liver, spleen, kidney, and heart were stained with H&E as described above and imaged with brightfield microscopy on a EVOS M5000 microscope to visualize tissue structure (Figure S24). Animal weight was also measured throughout the study to provide a measure of health and NP biosafety (Figure S25).

4.12. Quantifying Gold Content in Tissue Samples via ICP-MS

Inductively Coupled Plasma-Mass Spectrometry (ICP-MS) was used to measure gold content in primary tumors, recurrent tumors and other organs dissected at the end of each study (blood, heart, lungs, small intestine, spleen, kidney, liver, brain). Following removal from the body, one-quarter of each excised organ was placed into 15 mL conical tubes that were weighed beforehand to record a tare weight. A wet weight was recorded with the organ in the tubes. The tubes were then frozen overnight at −80 °C, lyophilized, and the dry weight of the tissues in tubes was recorded. The dried samples were digested in a 97% trace metal grade nitric acid (Fisher) and 3% trace metal grade hydrochloric acid (Fisher) solution in a heat block at 60 °C in a fume hood for 1 – 2 h (acid volume and digestion time varied depending on the tissue). Next the samples were diluted in an acid matrix containing 2% nitric acid and 2% hydrochloric acid in MilliQ water (dilution volume varied depending on the tissue), and a standard curve containing 0 – 750 parts per billion (ppb) gold using TraceCert Gold Standard for ICP (Sigma) was prepared. An internal baseline of 5 parts per million (ppm) was added to each standard and sample. Samples were analyzed on an Agilent 7500c ICP-MS instrument and the gold content in ppb was recorded for each sample. The gold content per gram of dried tissue was calculated based on the raw gold ppb values, accounting for the dilution factors throughout the sample preparation, and the dry weight of the organ (subtracting out the tare weight of the tubes).

4.13. Statistical Analysis

All data are expressed as mean ± standard deviation or standard error of the mean as noted in the figure captions. Normalization was performed as explained in each figure caption. The percentage of β-catenin positive cells in stained images of tumor sections was determined using QuPath software. All statistical tests were performed using JMP Pro 17.2 software. One-way analysis of variance (ANOVA) with post hoc Tukey (for equal sample sizes) or Tukey-Kramer (for unequal sample sizes) was used to determine statistical significance between groups at the 95% confidence level in all experiments. Significance is depicted as **p<0.01, *p<0.05, and #p<0.10. For NP characterization, 11 – 19 batches of each NS formulation were analyzed. For in vitro assays, at least 3 biological replicates were conducted for each treatment group. For in vivo experiments, the number of mice per treatment group was calculated by power analysis in G*Power software using α=.05 and β = 0.80. Sample sizes for each study are explained in the text and figure captions.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.