ImmunoPET Imaging of Pancreatic Tumors with ⁸⁹Zr-labeled Gold Nanoparticle-Antibody Conjugates

Abstract

Purpose:

Targeted delivery in vivo remains an immense roadblock for the translation of nanomaterials into the clinic. The greatest obstacle is the mononuclear phagocyte system (MPS), which sequesters foreign substances from general circulation and causes accumulation in organs such as the liver and spleen. The purpose of this study was to determine whether attaching an active targeting antibody, 5B1, to the surface of gold nanoparticles and using clodronate liposomes to deplete liver and splenic macrophages could help to minimize uptake by MPS organs, increase targeted delivery to CA19.9-positive pancreatic tumors, and enhance pancreatic tumor delineation.

Procedures:

To produce the antibody-gold nanoparticle conjugate (Ab-AuNP), the Ab was conjugated to p-isothiocyanatobenzyl-desferrioxamine (p-SCN-DFO) and subsequently conjugated to NHS-activated gold nanoparticles. The Ab-AuNP was characterized by transmission electron microscopy (TEM) and atomic force microscopy (AFM). Modified Lindmo assay was performed to assess binding affinity and internalization potential in vitro. The Ab-AuNP was radiolabeled with ⁸⁹Zr and injected into CA19.9-positive BxPc-3 pancreatic orthotopic tumor-bearing mice pretreated with or without clodronate liposomes for PET imaging and biodistribution studies. Inductively coupled plasma-optical emission spectrometry (ICP-OES) analysis was used to confirm delivery of gold nanoparticles to BxPc-3 pancreatic subcutaneous xenografts.

Results:

Mice pretreated with clodronate liposomes in an orthotopic setting demonstrated decreased liver uptake at early time points (12.2 ± 2.3% ID/g vs. 22.8 ± 3.8% ID/g at 24 hours) and increased tumor uptake at 120 h (13.8 ± 8.0% ID/g vs 6.0 ± 1.2% ID/g). This allowed for delineation of orthotopic pancreatic xenografts in significantly more mice treated with clodronate (6/6) than in mice not treated with clodronate (2/6) or mice injected with gold nanoparticles labeled with a nonspecific antibody (0/5).

Conclusions:

The combination of clodronate liposomes and an active targeting antibody on the surface of gold nanoparticles allowed for PET/CT imaging of subcutaneous and orthotopic pancreatic xenografts in mice.

Introduction

The potential of gold nanoparticles in the areas of cancer imaging and therapy is enormous [1, 2]. Applications for gold nanoparticles range from CT contrast agents [3–5], to drug delivery agents [6–8], to radiosensitizers [9–11]. However, their use is limited by their poor delivery efficiency in vivo. The main roadblock associated with nanoparticle delivery is the mononuclear phagocyte system (MPS) [12, 13]. The MPS is part of the immune system and contains phagocytic cells such as monocytes in the spleen and Kupffer cells in the liver, which sequester nanoparticles from the general circulation, leading to off-target accumulation. Due to these effects, the vast majority (> 90%) of systemically injected nanoparticles accumulate in the liver and spleen [14, 15]. This nonspecific uptake makes imaging solid tumors with nanoparticles difficult, as high background in organs of the MPS can overshadow the signal from the tumor. One strategy to increase tumor-specific signal is to “actively target” the nanoparticles by covalently attaching a targeting moiety such as an internalizing peptide or antibody to the surface of nanoparticles that targets a specific antigen overexpressed on the cells of interest [16, 17]. The targeting moiety will allow for cell-specific interaction and internalization and thereby increase nanoparticle accumulation in cells that express the antigen over that which can be achieved by EPR alone. In theory, this should maximize delivery efficiency to cells of interest and thereby improve the target-to-non-target ratio. In this work, we use gold nanoparticles for two reasons: (1) the gold content of the nanoparticle indirectly facilitates the chemical conjugation of ⁸⁹Zr-labeled antibodies and allows for the detection of gold itself through elemental analysis via ICP-OES; (2) the high uptake seen with ⁸⁹Zr-5B1-AuNP warranted further study, as it may be helpful for future applications of AuNPs in cancer and other disease contexts.

The current manuscript demonstrates a method for using a zirconium-89 (⁸⁹Zr) radiolabeled antibody, 5B1, to track nanoparticle biodistribution over time and quantitate uptake in various organs in comparison to an isotype IgG. ⁸⁹Zr is a positron-emitting radionuclide with a half-life of 78.4 h, which nicely matches the biological half-life of large biomolecules such as antibodies. This method also allows for quantitating differences between models (subcutaneous vs. orthotopic) as well as added variables in the model (clodronate liposomes). This work will not change the state of radiolabeled antibodies for imaging, but it represents a robust method for the evaluation of nanoparticle behavior in vivo in the context of actively targeted nanoparticle formulations.

5B1 is a fully human monoclonal antibody that targets the CA 19.9 antigen, which is overexpressed in pancreatic ductal adenocarcinoma (PDAC) [18]. The radiolabeled version of this antibody, [⁸⁹Zr]Zr-5B1, has been evaluated in pancreatic and bladder cancer preclinical models, and is now in clinical trials for imaging pancreatic cancer at Memorial Sloan Kettering Cancer Center (NCT02687230) [19–21]. By using a radiolabeled antibody as the active targeting component of a nanoparticle system, it is possible to use a PET scanner to image and track nanoparticle accumulation in vivo and to quantify the biodistribution in all major organs [22, 23].

Another strategy to improve the delivery efficiency of nanoparticles is to evade or depress the MPS so that it cannot sequester the nanoparticles from general circulation as rapidly [24–26]. To evade the MPS, particles can be modified to include surface coatings that shield the particles from phagocytic cells. The MPS can also be chemically depressed. For example, phagocytic cells such as macrophages can be depleted using clodronate liposomes [27, 28]. Clodronate is a bisphosphonate that is toxic to macrophages. When encapsulated in liposomes and injected in vivo, the liposomes are ingested and digested by macrophages, releasing the toxic drug and causing apoptosis. With macrophages depleted, the nanoparticles can circulate longer, with a greater fraction of them making it to their intended destination.

While these existing strategies should theoretically improve nanoparticle delivery to tumors, recent publications have called their utility into question [29]. The observation of “active targeting” has been shown to derive mostly from nonspecific uptake and accumulation in tumor-associated macrophages (TAMs) rather than specific ligand-receptor interactions. And although macrophage depletion has enhanced nanoparticle delivery, it was not to the extent that was expected: depletion increased tumor delivery efficiency to only 2% of the total injected dose [30].

In this work we combine an active targeting moiety, the 5B1 antibody and clodronate liposomes to deliver gold nanoparticles to both subcutaneous and orthotopic pancreatic tumors. Here we present data confirming the utility of an antibody to improve the specific delivery of nanomaterials and the potential of MPS depression strategies to aid in delineation of tumors in close proximity to organs such as the liver and spleen.

Results

The gold immunoconjugate was prepared in two steps (Figure 1A). The antibody 5B1 was functionalized with an isothiocyanate derivative of the chelator desferrioxamine (DFO) and subsequently radiolabeled with ⁸⁹Zr with a radiochemical yield > 90%. The radiolabeled antibody was then conjugated to the 10 nm NHS-activated gold nanoparticles with an efficiency of 30–35%. Control immunoconjugates were prepared identically with an IgG antibody. Purification was accomplished by serial centrifugation to yield the antibody-labeled gold nanoparticles. NHS-activated gold nanoparticles were characterized by transmission electron microscopy (TEM - JEOL 1400 Plus TEM) and atomic force microscopy (AFM). The diameter of the gold nanoparticle core was 10 nm as determined by TEM (Figure 1B). Negative staining with uranyl acetate allowed for visualization of the polyethylene glycol (PEG) corona surrounding the gold nanoparticle cores. AFM was used to confirm the structure and purity of antibody-labeled nanoparticles. AFM images showed 3–4 antibodies bound to the surface of the gold nanoparticles and confirmed the diameter of the gold nanoparticles as seen by TEM (Figure 1C–D and Supporting Figure 1). Dynamic light scattering showed a hydrodynamic diameter of 34.86 nm (PDI: 0.27) of the 5B1 labeled particles and a zeta potential of −21.3 mV. Serum stability was analyzed by radio-iTLC to determine how much ⁸⁹Zr remained bound to the antibody-labeled gold nanoparticles. Retention of ⁸⁹Zr was > 90% after 5 days.

Figure 1.

(A) Schematic representation of [⁸⁹Zr]Zr-5B1-AuNP preparation showing DFO conjugation to 5B1 and then to the surface of the gold nanoparticle. B) TEM image (uranyl acetate stained) of 10 nm NHS-activated gold nanoparticles. (C) AFM phase contrast image of [⁸⁹Zr]Zr-5B1-AuNP demonstrating 3–4 antibodies per gold nanoparticle. (D) AFM image of [⁸⁹Zr]Zr-5B1-AuNP confirming size of [⁸⁹Zr]Zr-5B1-AuNP. (E) Immunoreactivity of [⁸⁹Zr]Zr-5B1-AuNP and [⁸⁹Zr]Zr-IgG-AuNP in both BxPC-3 (CA 19.9 positive) and MiaPaCa-2 (CA 19.9 negative) cell lines. (F) Internalization assay results for [⁸⁹Zr]Zr-5B1-AuNP in both BxPC-3 (CA 19.9 positive) and MiaPaCa-2 (CA 19.9 negative) cell lines. (G) Internalization assay results for [⁸⁹Zr]Zr-IgG-AuNP in BxPC-3 and MiaPaCa-2 cell lines. **p<0.001, ***p<0.0001 determined by t-test.

In vitro assays were conducted to assess the binding affinity and internalization potential of the gold immunoconjugates. Retention of binding affinity was demonstrated by modified Lindmo assay [31]. Immunoreactivity of [⁸⁹Zr]Zr-5B1-AuNP was 49.5% in BxPC-3 (CA 19.9-positive) cells with no significant binding in MiaPaCa-2 (CA 19.9-negative) cells. [⁸⁹Zr]Zr-IgG-AuNP showed no significant binding in either cell line, as expected (Figure 1E). Internalization of [⁸⁹Zr]Zr-5B1-AuNP was evaluated in BxPC-3 and MiaPaCa-2 cells. [⁸⁹Zr]Zr-5B1-AuNP exhibited increasing internalization over time, reaching a maximum of 20.2 ± 0.50% at 4 h in BxPC-3 cells with negligible uptake in MiaPaCa-2 cells. A blocking study was also performed to confirm that the internalization was due to specific binding of the antibody-nanoparticle conjugate to the target antigen (Figure 1F). Adding an excess of unlabeled 5B1 to the wells one hour before the nanoparticle conjugates were introduced could block uptake of [⁸⁹Zr]Zr-5B1-AuNP. IgG labeled particles showed negligible uptake in both cell lines (Figure 1G).

For the primary in vivo assessment of [⁸⁹Zr]Zr-5B1-AuNP, mice were xenografted in the hind flank with BxPC-3 tumors. After three weeks, 80 μg of [⁸⁹Zr]Zr-5B1-AuNP or the control particle [⁸⁹Zr]Zr-IgG-AuNP were injected intravenously. This quantity was determined by the specific activity of [⁸⁹Zr]Zr-5B1-AuNP. Ultimately, to achieve an injectable activity that would produce acceptable image quality, 80 μg (~80-100 μCi) was required.

Imaging and biodistribution was determined at 24, 48, 72, and 120 h post-injection. Tumor uptake of [⁸⁹Zr]Zr-5B1-AuNP in antigen-positive BxPC-3 tumors was rapid, enabling clear visualization by PET at 24 h post-injection (24.0 ± 11.6% ID/g). This uptake remained constant over the course of 120 h. No tumor visualization was seen with the control particle. Liver and spleen uptake (< 20% ID/g in both cases) was also evident at all time points. At later time points, high uptake in the axillary lymph node obscured the tumor in the maximum intensity projection (MIP) but clear delineation could still be seen in the coronal slices. Biodistribution in all organs except the blood was unchanged from 24 h to 120 h post-injection. This suggests the antibody-nanoparticle conjugates had a relatively short blood circulation time, as the time needed for a radiolabeled antibody to accumulate in the tumor is approximately 3 to 5 days. The much larger size of these conjugates prevented any increased accumulation in the tumor (Figure 2). Another cohort of mice was xenografted with MiaPaCa-2 cells, which do not express CA 19.9. These tumors showed little accumulation of the 5B1-labeled particles (4.0 ± 1.2% ID/g), which can be attributed to EPR effect.

Figure 2.

(A) PET images (MIPs and coronal slices) of [⁸⁹Zr]Zr-5B1-AuNP and control particle in nude mice bearing BxPC-3 (CA 19.9 positive) xenografts on the hind flank. White arrows denote location of tumors in all mice. (B) Select organ biodistribution data for [⁸⁹Zr]Zr-5B1-AuNP and control particle in subcutaneous xenograft model.

The short blood half-life of the gold immunoconjugates is partly due to rapid sequestration of the nanoparticles from the general circulation by macrophages, primarily in the liver (Kupffer cells) and spleen. We used clodronate liposomes to determine whether circulation time and tumor accumulation would increase in an environment where macrophages have been depleted.

First, to validate macrophage depletion by clodronate liposomes, a group of mice bearing subcutaneous BxPC-3 tumors were injected intraperitoneally with 200 μL (7 mg/mL) of clodronate liposomes. All mice were injected with same dose, as their weights were similar. The choice of IP injection was motivated by previous work demonstrating that the effects of macrophage depletion are delayed, whereas this route extends depletion, as the clodronate liposomes arrive more gradually and over a longer period of time than with intravenous injection [32]. To mimic experimental conditions, mice were sacrificed at 72 and 120 h post-injection of the liposomes. These time points would correspond to 24 and 72 h post-injection of the antibody-nanoparticle conjugates, allowing 48 h for macrophage depletion [33]. At these time points the liver, spleen, and tumors were harvested from each mouse. A group of control mice were not treated and were also sacrificed at the same time points. Sections of 5 μm were stained with an immunofluorescent marker, Ibal, which is a general marker for macrophages. Immunofluorescent staining of liver and spleen tissues from mice treated with clodronate liposomes showed a decrease in staining intensity for macrophages compared to untreated controls at 72 h. However, this depletion was transient, and by 120 h post-injection the macrophages appeared to rebound to pre-injection levels. Tumors did not show any evidence of macrophage depletion (Supporting Figure 2).

To determine the effects of this transient macrophage depletion on nanoparticle accumulation, a cohort of mice bearing subcutaneous BxPC-3 tumors were injected intraperitoneally with clodronate liposomes 48 h before the nanoparticle conjugates were injected intravenously, allowing time for macrophages to be depleted. The biodistribution of [⁸⁹Zr]Zr-5B1-AuNP was assessed at 24, 48, and 120 h post-injection. At 24 h, uptake in the tumor was comparable to untreated mice bearing subcutaneous BxPC-3 xenografts. In mice injected with clodronate, the tumor uptake continued to increase over time to 71.5 ± 21.3% ID/g at 120 hours (Figure 3). This may be due to macrophage depletion, as sequestration in the liver is decreased after clodronate administration (12.8 ± 2.0% ID/g vs. 16.5 ± 2.6% at 48 hours).

Figure 3.

Biodistribution of [⁸⁹Zr]Zr-5B1-AuNP in subcutaneous BxPC-3 xenografts in mice treated with clodronate liposomes 48 hours before nanoparticle injection.

To confirm the delivery of gold nanoparticles to subcutaneous xenografts, a cohort of BxPC-3-tumor-bearing mice were injected with either [⁸⁹Zr]Zr-5B1-AuNP or [⁸⁹Zr]Zr-IgG-AuNP and their organs were harvested for inductively coupled plasma-optical emission spectrometry (ICP-OES) analysis. Gold content analysis demonstrated comparable biodistribution to gamma counting analysis of ⁸⁹Zr distribution. The analysis showed a 15x greater gold content in tumors that were actively targeted by the 5B1-labeled particles as compared to the isotype control. This demonstrates that the gold immunoconjugates are stable in vivo and that the gold nanoparticle and radiolabeled antibody remain bound to each other after intravenous injection (Figure 4).

Figure 4.

ICP-OES gold content analysis of [⁸⁹Zr]Zr-5B1-AuNP and [⁸⁹Zr]Zr-IgG-AuNP in select organs from mice bearing BxPc-3 subcutaneous xenografts.

To evaluate the nanoparticle conjugates in an orthotopic context, nude mice bearing orthotopic pancreatic xenografts were injected with [⁸⁹Zr]Zr-5B1-AuNP or [⁸⁹Zr]Zr-IgG-AuNP. Mice injected with [⁸⁹Zr]Zr-5B1-AuNP were examined in both an MPS-depressed (clodronate injected) and normal setting (saline injected). Another cohort of mice was injected with the 5B1-radiolabeled antibody only (Supporting Figure 3). In mice injected with [⁸⁹Zr]Zr-5B1-AuNP, biodistribution showed a significant decrease in liver uptake in the clodronate-treated mice compared to untreated controls at 24 h (12.2 ± 2.3% ID/g vs. 22.8 ± 3.8% ID/g) and 72 h (11.28 ± 2.0% ID/g vs 18.6 ± 1.9% ID/g). This also corresponded with significantly higher spleen uptake in the clodronate-treated group as compared to the untreated controls at the same time points. Tumor uptake was significantly higher in the clodronate group than in the untreated control group at 120 h post injection (13.8 ± 8.0% ID/g vs. 6.0 ± 1.2% ID/g). By PET/CT imaging, orthotopic pancreatic tumors could be visualized in all mice treated with clodronate (6/6), in 33% of mice not treated with clodronate (2/6) and none of the mice injected with the isotype-control-labeled nanoparticle (0/5) (Supporting Figure 4). This demonstrates the ability of clodronate liposomes to improve the detection of orthotopic pancreatic tumors with gold nanoparticle immunoconjugates (Figure 5, 6).

Figure 5.

PET/CT images of pancreatic orthotopic xenograft bearing mice injected with either [⁸⁹Zr]Zr-5B1-AuNP in mice treated with clodronate, without clodronate, or [⁸⁹Zr]Zr-IgG-AuNP.

Figure 6.

Selected organ biodistribution for mice bearing orthotopic pancreatic xenografts comparing the effect of an active targeting antibody and clodronate liposomes for macrophage depletion. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 by 2-way ANOVA.

Discussion

Here, we wanted to assess the ability of active targeting ligands on the surface of gold nanoparticles to enhance delivery to pancreatic tumors in both the subcutaneous and orthotopic setting. We also wanted to observe the ability of clodronate liposomes to deplete macrophages in MPS organs such as the liver and spleen. To accomplish this, we aimed to directly quantify nanoparticle delivery to several organs of interest by PET imaging, organ analysis of radioactivity, and organ gold content by ICP-OES as proof-of-concept in mice that do not express CA19.9 endogenously.

There has been some disagreement as to whether “active targeting” of nanoparticles is a sound strategy for delivery, as recent publications have shown that the majority of “actively targeted” nanoparticle systems actually end up in nonspecific cell types such as tumor-associated macrophages [29]. However, our results here imply that, at least in this particular tumor model and nanoparticle system, active targeting does indeed increase the delivery efficiency of nanoparticles to a given cell type, unlike control nanoparticles labeled with a non-specific targeting component.

The [⁸⁹Zr]Zr-5B1-AuNP antibody-gold nanoparticle conjugate was successfully synthesized and characterized by TEM and AFM. In vitro assays showed high specific uptake in BxPc-3 (CA19.9-positive) cells with internalization reaching a maximum of 20.2 ± 0.50% at 4 h. The internalization is not normalized to account for the 50% immunoreactivity observed. Likewise, adding unlabeled 5B1 prior to addition of the conjugate showed that the target could be blocked. Since the uptake could be blocked, one can infer that the internalization mechanism of [⁸⁹Zr]Zr-5B1-AuNP was due to specific receptor-ligand interactions.

When examined in a subcutaneous model, [⁸⁹Zr]Zr-5B1-AuNP accumulated in tumors more than the IgG-labeled control particle did, implying receptor-ligand interaction rather than nonspecific uptake. Accumulation of [⁸⁹Zr]Zr-5B1-AuNP at the site of the tumor via EPR is further enhanced and retained by active tumor targeting via the internalization of the antibody-nanoparticle conjugate by the BxPc-3 cells [34]. With no active targeting moiety on the surface of the IgG particles, a potential mechanism for their significantly decreased uptake may derive from quick MPS clearance from the blood. To confirm that the antibody-labeled nanoparticle remained intact in vivo, the gold content of specific organs was determined by ICP-OES in mice that were injected with the 5B1 active targeting nanoparticle or an IgG-labeled nonspecific control. The analysis confirmed a significant increase (15×) in gold accumulation in the tumors of mice injected with the 5B1-labeled nanoparticle compared to the IgG-control-labeled nanoparticle. Both quantitative methods demonstrate the benefit of an active targeting component to more efficiently deliver nanoparticles to a solid tumor.

In an orthotopic model of PDAC, an actively targeted nanoparticle, [⁸⁹Zr]Zr-5B1-AuNP, and a nonspecifically targeted nanoparticle, [⁸⁹Zr]Zr-IgG-AuNP, were examined. For the actively targeted particle, the imaging and biodistribution was also examined in the context of macrophage depletion in mice that were treated with clodronate liposomes. In macrophage-depleted mice, liver uptake was significantly reduced in the first 72 h post-injection, while during that same time period spleen uptake was significantly higher. One possible explanation for this high spleen uptake may be the quicker recovery of splenic macrophages in comparison to the recovery of Kupffer cells in the liver; however, further research would have to be done to answer this question. By 120 h post-injection, both liver and spleen uptake had normalized to the levels of controls. Tumor uptake was not significantly different until 120 h post-injection. This significance was due to a constant level of uptake in the clodronate-treated mice and an apparent decrease in the tumor uptake in mice not treated with clodronate at 120 h. Nevertheless, mice treated with clodronate liposomes retained their tumor uptake throughout the experiment, which allowed for the visualization of orthotopic pancreatic tumors in more cases than with mice not treated with clodronate. Mice injected with the nonspecifically targeted nanoparticle showed very low levels of tumor uptake (4–7 times less) as compared to the actively targeted particle, again demonstrating the ability of a targeting ligand on the surface of nanoparticles to increase accumulation in target tissues.

Conclusion

In this work we have demonstrated the ability to deliver specifically targeted antibody-labeled gold nanoparticles to both subcutaneous and orthotopic pancreatic xenografts. When utilizing the humanized antibody 5B1, gold nanoparticles accumulated in subcutaneous pancreatic xenografts bearing the target antigen at 24.0 ± 11.6% ID/g compared to 4.0 ± 1.2% ID/g for the IgG-labeled control. This accounts for a 6–8x increase in tumors that expressed the target antigen. In an orthotopic model, 5B1-labeled gold nanoparticles accumulated 4–7 times more in tumors than did the IgG-labeled controls. Further, the ability of clodronate liposomes to enhance imaging of orthotopic pancreatic xenografts has also been demonstrated. This work carries implications for the development of methods of tracking nanoparticle biodistribution in real time in vivo. By incorporating a PET-active radionuclide onto our nanoparticles, we could visualize and analyze the effects of an active targeting moiety and macrophage depletion strategies. This could in theory then be used for other modifications made to any nanoparticle system that needs to be evaluated in vivo for clinical applications. The incorporation of a PET nuclide can enable the noninvasive tracking of drug effects or other factors in the biodistribution of nanoformulations. Further, gold nanoparticles themselves provide advantages such as (a) in vivo safety, (b) easy modification, and (c) the ability to target both passively through localization to the tumor microenvironment and actively by tumor cells. Lastly, this study supports a pharmacologic strategy that uses clodronate liposomes to enhance the biodistribution of a gold nanoparticle-antibody conjugate in order to enable imaging by PET.

Methods

Preparation of DFO-Ab:

IgG from human serum was obtained by Sigma-Aldrich. 5B1 antibody was provided by MabVax pharmaceuticals. DFO-SCN was obtained from Macrocyclics. DFO was conjugated to the antibody (5B1 or IgG) by adding 5 μL of a 10 mM DFO-NCS solution in DMSO to a 2 mg/mL solution of antibody in 1 mL of PBS. The pH of the solution was adjusted to 9 with 30 μL of 1 M sodium carbonate. The reaction vial was then incubated for 1 hour at 37°C on a thermomixer followed by purification on a pre-packed disposable PD-10 desalting column (GE life sciences). Ab-DFO was concentrated with a 50,000 MWCO Amicon centrifugal filter. Final Ab-DFO concentration was assessed on a NanoDrop 2000 spectrometer (Thermo Scientific).

Radiolabeling of DFO-Ab:

⁸⁹Zr was supplied by the RMIP core at MSK in 0.1 M oxalic acid. The pH of the solution was adjusted to 6.8 with 1 M sodium carbonate before adding 1.2 mg of DFO-Ab to 4-6 mCi (148-222 MBq) of ⁸⁹Zr. The solution was placed on a thermomixer at 37°C for 1 hour before purification on a pre-packed disposable PD-10 desalting column. Radiochemical purity was assessed by radio-iTLC.

[⁸⁹Zr]Zr-DFO-Ab-AuNP Conjugation:

NHS-activated gold nanoparticles have a 5 kDa PEG chain between the surface and the NHS ester. For conjugation to the surface of 10 nm NHS-activated gold nanoparticles (Cytodiagnostics), 1.2 mg of [⁸⁹Zr]Zr-DFO-Ab was added to a mixture of 1.2 mg nanoparticles, protein resuspension, and reaction buffers. The mixture incubated at room temperature while mixing overnight. After 12 h, unbound [⁸⁹Zr]Zr-DFO-Ab was removed from the mixture by serial centrifugation (3x) at 20,000 g for 1 hour. After each successive centrifugation the supernatant, which contained unbound radiolabeled antibody, was removed. Approximately 30–35% of radiolabeled antibody was successfully conjugated to the surface of the gold nanoparticles. The addition of the antibody to the surface of the gold nanoparticles is non-site specific and there will be a range of Abs/NP; but on average by AFM it appears to be about 4 Abs/NP.

Characterization of Ab-AuNP:

10 nm NHS-activated gold nanoparticles were characterized by TEM and AFM. For AFM analysis, the nanoparticle and antibody-nanoparticle samples were diluted and plated on a mica surface. Once plated, the samples were washed with water and dried using nitrogen gas. Analysis was conducted using semi-contact (tapping) mode. A Malvern Zetasizer was used for hydrodynamic radius and zeta potential measurement samples were measured in water.

Cell culture:

The pancreatic cancer cell lines, BxPC-3 and MiaPaCa-2, were maintained at 37°C and 5% CO₂ atmosphere. BxPC-3 cells were cultured in Roswell Park Memorial Institute (RPMI)-1640 containing 2 mM L-glutamine, 10 mM HEPES, 1 mM sodium pyruvate, 4.5 g/L glucose, 1.5 g/L sodium carbonate, and 10% FBS. MiaPaCa-2 cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% FBS and 2.5% horse serum.

In vitro assessment of [⁸⁹Zr]Zr-Ab-AuNP:

Six-well plates were seeded with 1x10⁶ cells in 1 mL of media and allowed to adhere overnight in an incubator at 37°C and 5% CO₂. The following day 1 uCi of [⁸⁹Zr]Zr-5B1-AuNP was added to each well and returned to the incubator for 1, 4, 12, and 24 h. At each time point the media was collected and wells were washed with phosphate-buffered saline twice. The surface-bound fraction was collected by washing wells twice with cold 05 M glycine buffer (pH = 2.8) followed by a PBS rinse at 4°C. The internalized fraction was collected by lysing cells with 1 M NaOH followed by two PBS washes. All fractions were then counted on a Wizard² automatic gamma counter (Perkin Elmer).

Animals and Tumor models:

All animal experiments were performed according to protocols approved by the Institutional Animal Care and Use Committee at MSKCC and followed National Institute of Health directives for animal welfare.

Subcutaneous tumor model:

Female athymic nude mice were injected with 5x10⁶ BxPC-3 or MiaPaCa-2 cells in 150 μL (1:1 cell media:matrigel) in the hind flank. Tumors were allowed to grow for 3–4 weeks before imaging and biodistribution studies.

Orthotopic tumor model:

For orthotopic pancreatic xenografts, female athymic nude mice were used. Mice were anesthetized with 1–2% isoflurane and surgeries were conducted on a heated pad to regulate body temperature during the procedure. Bupivicaine was administered as a local anesthetic intradermally in the area near the incision. Skin around the incision was washed with 3 alternating wipes of povidone-iodine and 70% ethanol. An incision through the skin and peritoneum was made, followed by removal of the spleen and pancreas from the peritoneal cavity. At this time, 6 × 10⁵ BxPC-3 (luciferase-transfected) cells in media and Matrigel (1:1 ratio) were injected into the head of the pancreas. The spleen and pancreas were inserted back into the peritoneal cavity, which was then sutured with 4-0 Vicryl sutures. To close the skin, three sterile wound clips were inserted. Buprenorphine and meloxicam were administered immediately following the surgery. Meloxicam was administered 24 and 48 h post-surgery and wound clips were removed 7 days later. Tumor growth was monitored by bioluminescent imaging (IVIS Spectrum) with tumors reaching optimal size in approximately three weeks. Clodronate liposomes were obtained from Formumax and each mouse received a 200 μL intraperitoneal injection.

ICP-OES analysis:

Analysis was carried out using an Optima 7000 DV spectrometer (Perkin Elmer). The harvested organs containing [⁸⁹Zr]Zr-5B1-AuNP and [⁸⁹Zr]Zr-IgG-AuNP were first digested with an aqua regia solution of HNO₃(65%): HCl (35%) at 75°C overnight and then dissolved in adequate volumes of 5% HNO₃ solutions to be within the calibration curve range (from ppm to ppb). Hydrogen peroxide was added to speed the digestion of the organic materials. Calibration solutions were prepared from certified stock of a gold single element solution. The instrument was calibrated using a six-point calibration curve between 0.01 and 5 ppm and checked by three QC samples at the low, middle and high points on the curve. The operating conditions employed for ICP-OES determination were 1,300 W RF power, 15 L.min⁻¹ plasma flow, 0.5 L.min⁻¹ auxiliary flow, 0.8 L.min⁻¹ nebulizer flow, and 1 mL.min⁻¹ sample uptake rate. Signals at a wavelength of 267.595 nm were monitored. The low limit of quantification was determined to be 0.06 ppm. The comparison between the radioactive biodistribution data and ICP data is meant to be qualitative, in that it shows the same pattern of uptake. The discrepancy noted here we believe is due to the process by which the ICP measurements happen. Depending on the volume of acid required to dissolve organs, some concentrations may be below the limit of detection of the instrument (please note that the error bars are quite large for the ICP data due to this fact). This is a limitation of this method when using whole organs, as the volume of acid required for complete dissolution can be quite high; as well, a low injected mass of NPs, because it is difficult to dissolve, can make detection difficult for some samples, especially from the liver.

Biodistribution studies:

Biodistribution studies were conducted by sacrificing mice at discrete time points after injection of antibody-nanoparticle conjugates. Relevant organs and tumors were harvested, weighed, and counted on a gamma counter. Biodistribution values are presented as the percent of the injected dose per gram of tissue and were calculated by including appropriate standards.

PET/CT Imaging:

Mice were anesthetized with 1–2% isoflurane and images were acquired on an Inveon microPET/CT instrument.

Statistical Analyses:

All data presented are expressed as mean ± SD. Data for the in vitro internalization assay were analyzed by unpaired, two-tailed, t-test using GraphPad Prism 7 software. A correction for multiple comparisons was done using the Holm-Sidak method to determine statistical significance (α = 0.05). To evaluate the nanoparticle conjugates in an orthotopic context, biodistribution data were analyzed using a 2-way ANOVA test (in GraphPad Prism 7 software) with a threshold for statistical significance set at P < 0.05.