Acute Statin Treatment Improves Antibody Accumulation in EGFR- and PSMA-Expressing Tumors

This article is featured in Highlights of This Issue, p. 6075

Statins, at nanomolar concentrations, bind 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA), a rate-limiting enzyme upstream in the cholesterol synthesis pathway. The binding of statins to HMG-CoA results in inhibition of the conversion of HMG-CoA to l-mevalonate and further reduction of downstream cholesterol biosynthesis (1–3). Statins' low cost and oral administration route make them favorable therapeutic agents of hypercholesterolemia. More than 200 million people around the world are being treated with statins. Statins can be classified into natural drugs (lovastatin, mevastatin, pravastatin, and simvastatin) or synthetic drugs (fluvastatin, atorvastatin, cerivastatin, and rosuvastatin); their different chemical structures result in compounds with distinct lipophilicities, half-lives, and potencies (2, 3). In addition to their hypolipidemic effects, statins exert many other pleiotropic biologic effects. In this context, statins have demonstrated anti-cancer properties (1, 4–12).

Cholesterol is a major structural element of the cell membrane: it forms a semipermeable barrier separating cellular compartments, controls membrane dynamics, and interacts with other lipids as well as with specific membrane proteins (13–15). Caveolin-1 (CAV1) is a cholesterol-binding protein (16) and the main structural protein of caveolae (small, flask-shaped invaginations at the cell membrane; ref. 17). CAV1 plays key roles in lipid trafficking, cell signaling, and endocytosis (16). Although caveolae downregulate receptors at the cell surface of cancer cells (18–20), caveolae in fibroblastic cell lines and endothelial cells have lower dynamics when compared with noncaveolae endocytosis (reviewed in ref. 16). Cholesterol is essential for the formation of caveolae: it regulates CAV1 at the transcriptional level through a steroid regulatory binding element of the CAV1 promoter and it enhances CAV1 expression in stromal cells (reviewed in ref. 21). Cholesterol depletion is performed in laboratory protocols to pharmacologically modulate caveolae-mediated endocytosis (reviewed in ref. 22) by using CAV1 small interfering RNA (siRNA), CAV1-knockout mice, inhibition of cholesterol synthesis (statins), removal of cholesterol (methyl-β-cyclodextrin), and sequestration of cholesterol (filipin and nystatin). Cholesterol depletion perturbs not only caveolae but also clathrin-dependent endocytosis (23, 24), fast endophilin-mediated endocytosis (FEME) and the CLIC/GEEC pathway (25, 26).

Cetuximab [a human-mouse chimeric monoclonal antibody (mAb), IgG1] and panitumumab (a fully human mAb, IgG2k) are therapeutic mAbs that target the epidermal growth factor receptor (EGFR). Cetuximab and panitumumab have been used in the treatment of EGFR-mutant non–small cell lung cancer, KRAS wild-type tumors of patients with colorectal cancer, and squamous cell carcinoma of the head and neck (27). J591 targets prostate-specific membrane antigen (PSMA; ref. 28); ¹⁷⁷Lutetium- and ⁹⁰Yttrium-labeled J591 have demonstrated antitumor activity in prostate cancer metastases (29). The SC16 antibody targets the delta-like ligand 3 (DLL3), which is a transmembrane Notch ligand selectively expressed in small cell lung cancer (30). Although the SC16 antibody–drug conjugate Rova-T (31) demonstrated encouraging results in the first-in-human clinical trial for patients with small cell lung cancer, Rova-T showed disappointing results in the phase II TRINITY study (NCT02674568). In therapies using mAbs targeting membrane receptors, a prerequisite for antibody binding to tumors is the availability of the target antigen at the surface of cancer cells. It is clear that this process is governed in part by receptor trafficking from the cell membrane to the intracellular compartment; recent clinical studies have shown the potential of modulating endocytosis to improve antibody-directed therapies (32). In caveolae-mediated endocytosis, the endocytic pathway requires the cholesterol-binding protein CAV1, and therefore the process might be subject to modulation by statins. In agreement with this suggestion, the acute administration of statins enhances the availability of the HER2 for binding the anti-HER2 antibodies trastuzumab and pertuzumab (19, 33). How this pharmacologic approach works mechanistically and whether it can be applied to other membrane receptors is still unclear.

EGFR colocalizes with CAV1 in cancer cells (34). However, the contribution of caveolae in mediating EGFR endocytosis remains unclear. While some studies have suggested a caveolae-mediated endocytic pathway for EGFR (35), others have shown EGFR internalization through a clathrin pathway (36). In A431 cancer cells, 95% of EGFR is endocytosed by caveolae-independent pathways (32, 37). The differences in EGFR internalization can be explained by a “switch” process mediated by the EGFR ligand (EGF; ref. 38). At low EGF doses, the receptor internalizes through a clathrin-mediated mechanism. On the other hand, at higher EGF doses, EGFR is endocytosed through a caveolae-mediated pathway that leads into EGFR downregulation (38). Indeed, depletion of cholesterol inhibits the internalization of EGFR (38) and stabilizes EGFR at the cell surface (14). Others have shown that nystatin-mediated cholesterol sequestration switches EGFR trafficking from lipid rafts to clathrin-mediated internalization, enhancing anti-EGFR antibody accumulation in EGFR-expressing tumors (39).

Besides binding receptors of the HER family, CAV1 interacts with the PSMA, and PSMA internalizes via a caveolae-dependent mechanism in endothelial cells (18). In prostate cancer cells, approximately 60% of membrane PSMA internalizes through a clathrin-mediated mechanism, and the binding of anti-PSMA antibodies to PSMA further induces PSMA internalization (40, 41).

In this study, three different mouse models and the antibodies cetuximab, panitumumab, huJ591, and SC16 were used to monitor antibody binding to EGFR-, PSMA-, and DLL3-expressing tumors after administering mice with lovastatin, simvastatin, or rosuvastatin. Endocytic proteins and antibody–tumor accumulation were monitored in tumor xenografts by ex vivo analyses and noninvasive molecular imaging in mice after treatment with statins.

EGFR-expressing A431 epidermoid carcinoma cells, PSMA-overexpressing PC3pip prostate cancer cells, and DLL3-expressing H82 lung cancer cells were obtained from the ATCC.

A431 cells (EGFR-expressing) were cultured in DMEM supplemented with 10% FBS, 4.5 g/L glucose, 100 U/mL penicillin, and 100 μg/mL streptomycin; the human lung cancer cell line H82 (DLL3-expressing) was cultured in RPMI1640 supplemented with 10% FBS, 2 mmol/L l-glutamine, 10 mmol/L HEPES, 1 mmol/L sodium pyruvate, 4.5 g/L glucose, 1.5 g/L sodium bicarbonate, 100 U/mL penicillin, and 100 μg/mL streptomycin; and the prostate cancer cell line PC3pip (PSMA-overexpressing) was cultured in F12K media supplemented with 10% FCS, 1.5 g/L sodium bicarbonate, 2 mmol/L l-glutamine, 6 μg/mL puromycin, 100 U/mL penicillin, and 100 μg/mL streptomycin.

The cells were utilized within 15 passages upon delivery from ATCC, confirmed to be negative for Mycoplasma, and their identity verified by the Memorial Sloan Kettering Cancer Center (MSKCC) Integrated Genomics Operation Core using short tandem repeat profiling. Cells were cultured under aseptic conditions at 37°C in a humidified incubator at 5% CO₂.

Cells were incubated with 25 μmol/L of the active form of lovastatin (Millipore, 438187), active form of simvastatin, or with rosuvastatin (Sigma, SML1264). Simvastatin (Millipore, 567020) was activated before use by dissolving 5 mg of the statin powder in a solution containing 190 μL of DMSO and 810 μL milli-Q water. The pH of the solution was then adjusted to 7.2 with 0.1 mol/L NaOH and the solution was heated at 70°C for 2 hours. Cells were incubated in media containing a solution of simvastatin in 0.2% DMSO; control experiments were performed in media containing 0.2% DMSO instead of simvastatin.

A pool of three target-specific 20–25 nt siRNA CAV1 or siGENOME Human CAV1 siRNA was used to deplete CAV1 in A431 and PC3pip cancer cells (19). The cells were treated with 2.4 μmol/L of siRNA CAV1 or control scrambled (scr) siRNA for 5 hours. After a 48-hour incubation, membrane extracts of the transfected cells were collected by biotin pull down of cell-surface proteins using EZ-LINK Sulfo-Biotin (Thermo Fisher Scientific) and NeutrAvidin Agarose Resins (Thermo Fisher Scientific). At this time point, whole protein extracts were also collected using radioimmunoprecipitation assay buffer [RIPA buffer; 150 mmol/L sodium chloride (NaCl), 50 mmol/L Tris hydrochloride (Tris-HCl), pH 7.5, 5 mmol/L ethylene glycol tetraacetic acid (EGTA), 1% Triton X-100, 0.5% sodium deoxycholate (DOC), 0.1% SDS, 2 mmol/L phenylmethanesulfonyl (PMSF), 2 mmol/L iodoacetamide (IAD), and 1  ×  protease inhibitor cocktail (Roche)]. Following protein denaturation, equal amounts of proteins were loaded and separated on SDS-PAGE gels. The proteins were then transferred to polyvinylidene difluoride (PVDF) membranes (Bio-Rad). The membranes were blotted overnight with mouse anti-β-actin 1:20,000 (Sigma, A1978), rabbit anti-EGFR 1:1,000 (Abcam, ab52894), rabbit anti-CAV1 1:500 (Abcam, ab2910), or rabbit anti-PSMA 1:50,000 (Abcam, ab133579) after blocking them with 5% (m/v) BSA in Tris-buffered saline buffer-Tween (TBS-T, Cell Signaling Technology). The membranes were washed with TBS-T and then incubated with secondary antibodies IRDye 800CW anti-rabbit (925–32211) or anti-mouse (925–32210) IgG 1:15,000 (LI-COR Biosciences) and visualized using the Odyssey Infrared Imaging System (LI-COR Biosciences) followed by densitometric analysis using Fiji software (https://imagej.net/Fiji).

Biotin pull down and Western blot analyses were performed in A431 and PC3pip cells at 4 hours after treatments with 25 μmol/L of the active forms of lovastatin and simvastatin, or with rosuvastatin.

Western blots were also performed in total extracts of A431 and PC3pip xenografts using antibodies against clathrin 1:100 (Thermo Fisher Scientific, MA1065), dynamin 1:1,000 (BD Biosciences, 610245), cortactin 1:1,000 (MD Millipore, 05–180), endophilin 1:1,000 (Santa Cruz Biotechnology, sc-365704), and cavin-1 1:1,000 (Abcam, ab48824).

Panitumumab and cetuximab were obtained from the MSK Hospital Pharmacy. Rovalpituzumab (SC16.56) was synthesized and produced as recombinant protein based on the sequence disclosed in U.S. patent US9089616B2 (GenScript). J591 was a generous gift of Professor Neil Bander at Weill Cornell (New York, NY; ref. 28). The antibodies were conjugated with the bifunctional chelate p-isothiocyanatobenzyl-desferrioxamine (DFO-Bz-NCS; Macrocyclics) and then radiolabeled with zirconium-89 (⁸⁹Zr) following previously reported methods (19, 33, 42–46). [⁸⁹Zr]Zr-DFO-antibody radiochemical purity (RCP) was determined by instant thin-layer chromatography and used for in vitro and in vivo studies.

Internalization of [⁸⁹Zr]Zr-DFO-cetuximab or [⁸⁹Zr]Zr-DFO-panitumumab was evaluated in EGFR-expressing A431 cells. PSMA-overexpressing PC3pip cells were used to evaluate internalization of [⁸⁹Zr]Zr-DFO-huJ591. The radiolabeled antibody was dissolved in cell culture media (1 μmol/L) and then added to control or statin-treated cells for 90 minutes at 37°C. Following the incubation period, the media containing non–cell-bound radiotracer was collected and the cells were rinsed twice with PBS. Membrane-bound activity was collected by incubating the cells with 0.2 mol/L glycine buffer containing 0.15 mol/L sodium chloride (NaCl), and 4 mol/L urea (pH 2.5) at 4°C for 5 minutes. The cells were then lysed with 1 mol/L sodium hydroxide (NaOH) to collect the internalized fraction. All three fractions were collected and measured for ⁸⁹Zr radioactivity counts using a gamma counter.

For the binding assays, solutions of ⁸⁹Zr-labeled antibody (0–256 nmol/L) or ⁸⁹Zr-labeled control IgG (100 nmol/L) were prepared in PBS (pH 7.5) containing 1% w/v human serum albumin (Sigma) and 0.1% w/v sodium azide (Acros Organics). Control or statin-treated cells were incubated with the radiolabeled antibody for 2 hours at 4°C. Unbound radioactivity was removed and cells were washed twice with PBS. The cells were solubilized in 100 mmol/L NaOH and total cell-bound radioactivity was measured on a gamma counter calibrated for ⁸⁹Zr. Specific binding was plotted versus the concentration of ⁸⁹Zr-antibody; the data were fit via nonlinear regression with a one-site binding model in RStudio (http://www.rstudio.com/) to determine B_max and K_D.

All studies involving animal handling and experimentation were conducted according to the guidelines approved by the Research Animal Resource Center and Institutional Animal Care and Use Committee at MSKCC (New York, NY). The first author (Pereira) has a Category C accreditation for animal research from the Federation of European Laboratory Animal Science (FELASA). We adhere to the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines and to the guidelines for the welfare and use of animals in cancer research.

Female athymic nude mice nu/nu (8 to 10 weeks old, Charles River Laboratories) were subcutaneously implanted with A431 or H82 cells. Male athymic nude mice nu/nu were used for studies with PC3pip model. A total of 5 × 10⁶ cells were suspended in 150 μL of a 1:1 v/v mixture of medium with reconstituted basement membrane (BD Matrigel, BD Biosciences) and injected subcutaneously in each mouse. Mice studies were initiated once tumors reached 100–150 mm³ in volume as determined by caliper measurements.

Mice bearing A431 or PC3pip subcutaneous tumors were administered lovastatin (8.3 mg/kg of mice) or saline solutions. The solutions were administered twice with an interval of 12 hours between each administration (19). At 48 hours after the first dose of lovastatin, tumor sections were collected, formalin-fixed, paraffin-embedded, and processed into 10-μm thick sections using a microtome. Sections were then incubated twice in NeoClear solution (Sigma, 4741–6-5–7) for 5 minutes, followed by incubation in solutions containing decreasing percentages of ethanol (ethanol 100% 30 seconds, ethanol 100% 30 seconds, ethanol 96% 30 seconds, ethanol 70% 30 seconds, ethanol 70% 30 seconds, and water 1 minute). The sections were washed with water, permeabilized for 30 minutes with 0.25% v/v Triton X-100 (Tx-100; Sigma, T8787) in PBS with 0.022% w/v BSA (PBS/BSA) and blocked for 30 minutes with 10% v/v normal goat serum (Thermo Fisher Scientific, 50062Z) before incubation overnight at 4°C with primary antibodies (1:100 v/v) rabbit anti-HER2, rabbit anti-EGFR, rabbit anti-PSMA, rabbit anti-cavin-1, or rabbit anti-CAV1. Sections were rinsed with PBS and incubated with DAPI (1:1,000 v/v, Sigma, D9542) for nuclear staining and the secondary anti-rabbit fluorescent antibodies (1:300 v/v, goat anti-rabbit IgG H+L cross-adsorbed secondary antibody, Alexa Fluor 488, A-11008) for 1 hour at room temperature. The coverslips were mounted using the glycergel mounting medium (Agilent Technologies, C056330) and sealed with nail polish. The samples were stored at 4°C until acquisition of images by fluorescence microscopy.

For quantification of CAV1, EGFR, or PSMA staining, whole-slide digital images were generated on Pannoramic MIDI scanner (3DHistech) at a resolution of 0.3250 μm per pixel. Staining area quantification was performed with the Halo 2.3.2089.23 software with Area Quantification FL 1.2 and Tissue Classifier 2.3 modules.

Mice bearing A431, PC3pip, or H82 subcutaneous tumors (100–150 mm³ in volume) were randomized before treatment with statin or saline solutions (n = 5 mice per group for biodistribution and n = 3 mice per group for PET imaging). We used statin doses and treatment schedules as reported previously (19). Statin solutions were performed by dissolving the statin powder in a saline solution. Lovastatin, simvastatin, or rosuvastatin were orally administered twice (8.3 mg/kg of mice) with an interval of 12 hours between each administration. Control mice were administered saline instead of statin. [⁸⁹Zr]Zr-DFO-cetuximab (3.70–5.55 Mbq, 23.8–35.7 μg protein), [⁸⁹Zr]Zr-DFO-panitumumab (3.70–5.55 Mbq, 22.2–33.3 μg protein), [⁸⁹Zr]Zr-DFO-huJ591 (3.70–5.55 Mbq, 18.2–27.3 μg protein), or [⁸⁹Zr]Zr-DFO-SC16 (3.70–5.55 Mbq, 25.3–37.9 μg protein) were administered by tail vein injection at the time of the second dose of statin. PET imaging and ex vivo biodistribution were performed at 4, 8, 24, and 48 hours after intravenous injection of [⁸⁹Zr]Zr-DFO-antibody according to previously reported methods (47). The imaging and biodistribution studies were performed within 48 hours post antibody administration to allow the use of nonnecrotic tumors with a volume of 10% within the initial 100–150 mm³. The percentage of the injected dose of [⁸⁹Zr]Zr-DFO-antibody per gram of tumor tissue (%ID/g) was determined ex vivo using biodistribution data.

The CAV1, PSMA, and DLL3 messenger RNA (mRNA) expression data were obtained from the Cancer Cell Line Encyclopedia (https://portals.broadinstitute.org/ccle) database. In the CAV1/PSMA and CAV1/DLL3 analyses, we selected prostate and lung cancer cell lines. CAV1 and DLL3 protein levels in human lung tumors were obtained from data available in The Cancer Genome Atlas (TCGA; https://www.cancer.gov/tcga). CCLE and TCGA mRNA data were analyzed using RStudio (http://www.rstudio.com/). CAV1 mRNA, PSMA mRNA, and DLL3 mRNA were assembled into a single table file consisting of columns of proteins and rows of mRNA values. Pearson correlation coefficient (r) values between expressions of CAV1 and PSMA or DLL3 were determined. Significance of r values were assessed with P values calculated with cor.test function in R. Pair.

Data were analyzed using RStudio (http://www.rstudio.com/) or GraphPad Prism 7.00 (www.graphpad.com). Data are expressed as mean ± SEM. Groups were compared using the Student t test.

Given that the cholesterol-depleting drug lovastatin improves the availability of HER2 at the cell membrane for binding anti-HER2 antibodies in vitro and in vivo (19, 33), we sought to determine whether the use of lovastatin would result in improved binding of antibodies to membrane receptors other than HER2. In this preclinical work, we determined the ability of statins to modulate the membrane availability of EGFR, PSMA, and DLL3 (Fig. 1). Besides exploring this approach in three different receptors, we used three chemically different statins–lovastatin, simvastatin, and rosuvastatin. The statins used in our studies have demonstrated anticancer activity and ability to modulate CAV1 protein (4–7, 12). Lovastatin, simvastatin, and rosuvastatin have distinct pharmacodynamic profiles (1, 2); lovastatin and simvastatin are lipophilic lactone prodrugs (2). Simvastatin differs from lovastatin in that it has an additional methyl group in the lateral chain (Fig. 1). Upon oral administration, the statin prodrugs are enzymatically hydrolyzed to their active β-hydroxy acid form. Unlike lovastatin and simvastatin, rosuvastatin is a hydrophilic statin administered in the active hydroxy acid form (2).

Human epidermoid carcinoma A431 cells contain high levels of EGFR (approximately 2 × 10⁶ EGFRs/cell; ref. 48) and were used as a biological model for experiments with the anti-EGFR antibodies cetuximab and panitumumab. The EGFR⁺ A431 cells express CAV1 and have an abundance of caveolae (49). The prostatic cancer cell line PC3 contains low levels of PSMA and it has been transfected to overexpress PSMA (50), generating the PSMA-overexpressing PC3pip cell line herein used for studies with the anti-PSMA antibody huJ591. Although PC3 cells express CAV1, these cells do not form caveolae as they do not express proteins of the cavin family (51). The H82 lung cancer cell line contains median DLL3 protein levels, within the range of that clinically reported in human small cell lung cancer (42); they were the model of choice for studies with the anti-DLL3 antibody SC16. The DLL3-expressing H82 cells contain lower levels of CAV1 when compared with A431 and PC3 cells (Supplementary Fig. S1).

We used immunoPET to image dynamic changes in antibody–xenograft binding after administering mice with a statin. The anti-EGFR antibodies (cetuximab and panitumumab), anti-PSMA antibody huJ591, and anti-DLL3 antibody SC16 were conjugated with the acyclic bifunctional chelator DFO-Bz-SCN and then radiolabeled with ⁸⁹Zr (19, 33, 42–46). Figure 1 summarizes radiolabeling yield, purity, and specific activities of ⁸⁹Zr-labeled antibodies, which compared favorably with the previously reported values for ⁸⁹Zr-labeled cetuximab (43), ⁸⁹Zr-labeled panitumumab (44), ⁸⁹Zr-labeled huJ591 (45, 46), and ⁸⁹Zr-labeled SC16 (42). The ⁸⁹Zr-labeled antibody conjugates were prepared with radiochemical yields ranging from 88.9% to 94.6% and specific activities in the range of 21.98–30.53 MBq/nmol. In vitro, the radiolabeled antibodies retained the antibody's affinity to their respective target with immunoreactivities above 82% (Fig. 1).

Altogether, we have selected three chemically different statins (two prodrugs and the hydroxy acid form rosuvastatin), three distinct membrane receptors (EGFR, PSMA, and DLL3), and four ⁸⁹Zr-labeled antibodies (cetuximab, panitumumab, huJ591, and SC16) to noninvasively image in vitro and in vivo the ability of statins to enhance tumors' accumulation of mAbs.

Previous studies have demonstrated that cholesterol contributes to the membrane dynamics of EGFR (14, 39). Treatment of tumor cells with statins inhibits EGFR internalization (12). Because A431 cells have an abundance of caveolae and caveolae/lipid rafts have been found to participate in ligand-induced EGFR endocytosis (38), we first determined the effect of CAV1 depletion on EGFR levels at the cell surface and anti-EGFR antibody internalization (Fig. 2). In our studies, the cell-surface biotinylation experiments (Fig. 2A) were performed using two different siRNA reagents to visualize changes in EGFR protein levels at the cell membrane in siRNA Scr (control) or siRNA CAV1-depleted A431 cells (Fig. 2B; Supplementary Fig. S2). A431 cells showed a 9.7-fold (Fig. 2B) and 7.3-fold (Supplementary Fig. S2) increase in EGFR expression at the cell membrane after siRNA-mediated CAV1 depletion. The knockdown of CAV1 did not impact EGFR levels in total protein extracts (Fig. 2B; Supplementary Fig. S2). We next performed acute pharmacologic depletion of CAV1 using cholesterol-depleting drugs (lovastatin, simvastatin, and rosuvastatin; refs. 19, 33). Previous in vitro studies have shown CAV1 depletion in cells treated with statin solutions in the μmol/L range (5, 6, 10, 19, 33). In our experiments, cells were treated with 25 μmol/L of active lovastatin, active simvastatin, or rosuvastatin. We observed an increase in EGFR at the cell membrane after incubating A431 cells with statins (Supplementary Fig. S3). We next expected that changes in EGFR presence at the cell membrane upon statin treatment would affect the ability of the anti-EGFR antibody to bind A431 cells. Cellular fractionation of lovastatin-treated or simvastatin-treated A431 cells incubated with ⁸⁹Zr-labeled panitumumab revealed a significant increase (P < 0.05, Student t test) in membrane-bound radioactivity (Fig. 2C). We observed a significant decrease (P < 0.05, Student t test) in internalized ⁸⁹Zr-labeled panitumumab after treatment with lovastatin. Treatment of A431 cells with rosuvastatin did not interfere with membrane-bound or internalized ⁸⁹Zr-labeled panitumumab. The ratio of antibody binding to the cell membrane versus internalized antibody (Fig. 2D) was significantly higher in rosuvastatin-treated A431 cells when compared with untreated cells. Additional ⁸⁹Zr-labeled panitumumab saturation-binding assays in statin-treated cells (Fig. 2E) demonstrated an increase in EGFR density at the cell membrane (B_max) after treatment with lovastatin or simvastatin (Fig. 2F). Control experiments demonstrated similar low binding of a control IgG in untreated and statin-treated cells (Supplementary Table S1).

Altogether, the internalization and binding studies indicate that statins improve panitumumab binding on EGFR-expressing A431 cancer cells.

Premised on our in vitro findings (Fig. 2), we expected that mice treatment with statins might result in increased binding of anti-EGFR antibodies in EGFR⁺ A431 tumor xenografts. Mice were treated with statins as described previously (19, 33); the mice received two oral administrations of statin (8.3 mg/kg) given with an interval of 12 hours between each dose (Supplementary Fig. S4). The 12-hour administration window between the two doses of statin was selected on the basis of our previously reported studies showing statin-mediated pharmacologic CAV1 modulation (19, 33). According to the Food and Drug Administration's formula of mouse-to-human dose conversion, an 8.3 mg/kg statin dose in mice equates to 0.68 mg/kg for an adult weighing 60 kg. These doses are within the recommended dose range of statins in the clinic: 1.33 mg/kg/day for lovastatin or simvastatin and 0.66 mg/kg/day for rosuvastatin (2).

We used immunofluorescence to visualize changes in EGFR staining at the cell membrane in A431 tumors of mice treated with saline versus lovastatin (Fig. 3A). EGFR localization at the cell membrane increased in A431 tumors collected from mice treated with lovastatin. EGFR⁺ tumors showed a 2.9-fold decrease in CAV1 protein levels after mice treatment with lovastatin (Fig. 3B).

Cholesterol depletion also interferes with noncaveolae pathways (23–26). Therefore, we determined changes in endocytic proteins other than CAV1 after administering mice with lovastatin (Fig. 3C). Similar to our previous studies in NCIN87 tumor xenografts (33), lovastatin induced temporal changes in CAV1 protein levels in A431 tumors. We observed a decrease in CAV1 protein levels between 12 and 16 hours after lovastatin administration. The 12-hour time point corresponds to the time of antibody administration (Supplementary Fig. S4). Similar results were observed for the cavin-1 protein (another abundant and essential structural component of caveolae) and for endophilin. Additional studies demonstrated an increase in clathrin and dynamin protein levels at 16 hours after lovastatin administration.

These results suggest that statins temporarily modulate the protein levels of CAV1, cavin-1, clathrin, dynamin, and endophilin in EGFR-expressing A431 xenografts.

We next used EGFR-targeted PET imaging to monitor the binding of panitumumab to EGFR-expressing A431 tumors after treatment with lovastatin, simvastatin, or rosuvastatin (Fig. 3D). Panitumumab is rapidly internalized in an EGFR-dependent manner, resulting in EGFR downregulation at the cell membrane (52). Mice with A431 tumors were administered statin before tail vein injection of ⁸⁹Zr-labeled panitumumab (Supplementary Fig. S4). Control mice were orally administered saline instead of statin. Longitudinal PET imaging (Fig. 3D) and biodistribution (Fig. 3E; Supplementary Figs. S5 and S6) of saline-treated mice demonstrated a gradual increase of ⁸⁹Zr-labeled panitumumab at the tumor site throughout 48 hours. In lovastatin- and simvastatin-treated mice, we observed an increased accumulation of ⁸⁹Zr-labeled panitumumab in A431 tumors. The increase in antibody accumulation was observed for time points as early as 4 hours after antibody administration. In addition, administering mice with rosuvastatin increased anti-EGFR antibody accumulation in the tumor. Biodistribution of these mice at 48 hours postadministration of ⁸⁹Zr-labeled panitumumab (Fig. 3E; Supplementary Figs. S5 and S6) demonstrated a significant difference (**, P < 0.01, Student t test) in the tumor uptake of the radioimmunoconjugate between the saline- versus statin-treated A431 tumors. The percentage of the injected dose of [⁸⁹Zr]Zr-DFO-panitumumab per gram of tumor tissue (%ID/g) was higher in statin-treated mice when compared with saline-treated mice (16.05 ± 3.65 saline-treated, 27.89 ± 5.98 lovastatin-treated, 25.37 ± 2.19 simvastatin-treated, and 23.59 ± 3.26 rosuvastatin-treated mice).

Additional biodistribution studies were performed with the anti-EGFR antibody [⁸⁹Zr]Zr-DFO-cetuximab (Supplementary Figs. S7 and S8). Although cetuximab and panitumumab share the same target (EGFR), these antibodies have distinct degradative pathways (53). Biodistribution studies in saline-treated mice at 48 hours post antibody administration demonstrated similar antibody accumulation for [⁸⁹Zr]Zr-DFO-cetuximab (15.24 ± 8.67 %ID/g) and [⁸⁹Zr]Zr-DFO-panitumumab (16.05 ± 3.65 %ID/g; Fig. 3E; Supplementary Figs. S5–S8). [⁸⁹Zr]Zr-DFO-cetuximab accumulation was similar in saline-treated and lovastatin-treated tumors at 4 and 24 hours after antibody administration. Enhanced antibody accumulation in A431 tumors was observed in mice treated with lovastatin at 8 hours (2.63 ± 1.07 saline-treated vs. 5.22 ± 0.27 statin-treated mice) and 48 hours postadministration of [⁸⁹Zr]Zr-DFO-cetuximab (15.24 ± 8.67 saline-treated vs. 25.47 ± 2.54 statin-treated mice).

Taken together, these studies support the potential of statins to enhance the avidity of EGFR-expressing A431 tumors for cetuximab and panitumumab.

Having found that statins enhance tumor avidity for antibodies targeting receptors of the HER family [i.e., HER2 (19, 33) and EGFR; Figs. 2 and 3], we next investigated how statin treatment would affect antibody binding to receptors other than HERs. We examined the binding of the anti-PSMA antibody huJ591 to PSMA-overexpressing PC3pip cells and xenografts (Figs. 4 and 5). Previous studies have reported the potential for statins to treat castration-resistant prostate cancer by reducing androgen receptor–mediated signaling through the regulation of CAV1 expression (5). We analyzed the PSMA and CAV1 protein levels of human prostate tumors and prostate cancer lines using data obtained from The Cancer Genome Atlas (TCGA; https://www.cancer.gov/tcga) and the Cancer Cell Line Encyclopedia (CCLE; https://portals.broadinstitute.org/ccle) databases (ref. 54; Supplementary Figs. S9 and S10). We observed a negative correlation between PSMA and CAV1 at the protein level with Pearson rank correlation coefficients (r) of −0.685 (CCLE, P  = 0.0608) and −0.233 (TCGA, P = 1.568 × 10⁻⁷). In a panel of seven different prostate cancer cells, the PC3 cell line exhibited the highest expression of CAV1 (Supplementary Fig. S9). However, PC3 cells do not express high levels of PSMA and, therefore, PC3pip cells (PC3 cells transfected to overexpress PSMA) were used in our studies (50).

Although PC3pip cells contain high levels of CAV1, they do not express cavin-1 (Supplementary Fig. S11) and previous studies have reported that PC3 cells do not form caveolae (51). We depleted CAV1 in PC3pip cells using siRNA (Supplementary Figs. S12 and S13) and statins (Supplementary Fig. S14). We did not detect changes in PSMA expression at the cell membrane in CAV1-depleted versus control Scr-treated PC3pip cells (Supplementary Figs. S12 and S13). Additional studies demonstrated that PC3pip cells' treatment with statins did not change PSMA protein levels at the cell surface (Supplementary Fig. S14).

These studies demonstrate that PSMA negatively correlates with CAV1 at the protein level, but CAV1 depletion does not enhance PSMA membrane availability in a prostate cancer cell line that does not form caveolae.

Next, the binding of [⁸⁹Zr]Zr-DFO-huJ591 was determined in saline- and statin-treated PC3pip tumor cells. Previous studies have reported rapid internalization for [⁸⁹Zr]Zr-DFO-huJ591 in PSMA⁺ prostate cancer cells (45). In our studies, we detected 16.6% of [⁸⁹Zr]Zr-DFO-huJ591 being internalized by PC3pip cells at 1.5 hours. Lovastatin-treated PC3pip cells showed a significant increase (*, P < 0.05, Student t test) in [⁸⁹Zr]Zr-DFO-huJ591 binding to the cell membrane (Fig. 4A and B). On the other hand, simvastatin and rosuvastatin increased [⁸⁹Zr]Zr-DFO-huJ591 internalization (**, P < 0.01, Student t test). Additional binding assays (Fig. 4C and D) demonstrated that lovastatin and simvastatin increased huJ591 binding to PC3pip cells (B_max).

Immunofluorescence staining of PSMA was similar in tumors of saline-treated versus statin-treated mice (Fig. 5A). Additional immunofluorescence analyses of PC3pip tumors confirmed a 1.9-fold decrease in CAV1 protein levels in statin-treated mice (Fig. 5B). Similar to what was observed for A431 xenografts (Fig. 3C), lovastatin induced temporal changes in CAV1, clathrin, dynamin, and endophilin (Fig. 5C). A decrease in CAV1 and endophilin was associated with an increase in clathrin and dynamin at 16 hours after administration of lovastatin to mice bearing PC3pip xenografts.

We next conducted in vivo PET imaging and biodistribution assays to investigate possible differences in [⁸⁹Zr]Zr-DFO-huJ591 binding to PC3pip tumor xenografts of saline-treated versus statin-treated mice. To further compare huJ591 accumulation in PC3pip tumors between saline- and statin-treated mice, PET imaging and biodistribution studies were performed at 4, 8, 24, and 48 hours after tail vein injection of [⁸⁹Zr]Zr-DFO-huJ591. We did not detect significant differences in antibody tumor binding for the time points of 4 hours or 24 hours (Fig. 5D; Supplementary Fig. S15). At 48 hours after [⁸⁹Zr]Zr-DFO-huJ591 injection, the %ID/g was higher in statin-treated mice when compared with control saline-treated mice (25.55 ± 5.41 saline-treated, 35.67 ± 3.28 lovastatin-treated, 32.21 ± 2.48 simvastatin-treated, and 34.97 ± 5.32 rosuvastatin-treated mice; Fig. 5E).

These results indicate that statins can increase the accumulation of [⁸⁹Zr]Zr-DFO-huJ591 in PSMA-overexpressing PC3pip tumors, either by increasing antibody binding to the cell membrane or by increasing antibody internalization.

We showed that statins increase the accumulation of anti-HER2 antibodies (19, 33), anti-EGFR antibodies (Figs. 2 and 3), and anti-PSMA antibodies in tumor xenografts (Figs. 4 and 5). We next used DLL3-targeted immunoPET to determine lovastatin-mediated pharmacologic modulation of the Notch ligand DLL3. DLL3 is characterized by rapid internalization; antibody–drug conjugates, targeting the DLL3 receptor, internalize in an antigen-dependent manner (31). However, the endocytic mechanisms by which DLL3 and anti-DLL3 antibodies internalize in cancer cells are not completely understood. Similar to what we previously observed for HER2 (19) and PSMA (Supplementary Fig. S10), DLL3 protein expression negatively correlates with CAV1 protein levels in lung cancer cell lines (CCLE, P  = 2.2 × 10⁻¹⁶, Pearson correlation r = −0.5624) and human lung tumors (TCGA, P  = 1.6 × 10⁻⁷, Pearson correlation r = −0.2333, Supplementary Fig. S16). DLL3 expression in H82 tumor cells is lower when compared with HER2, EGFR, and PSMA protein expression in their respective in vitro biological models (NCIN87, A431, and PC3pip; Supplementary Fig. S1). Although expression of DLL3 at the cell surface is low when compared with other targets in immunoPET, we have previously reported the ability of the radiolabeled antibody [⁸⁹Zr]Zr-DFO-SC16 to image DLL3⁺ H82 tumors in vivo (42). In our next experiments, we determined [⁸⁹Zr]Zr-DFO-SC16 antibody accumulation in DLL3-expressing H82 tumors from mice treated with saline or lovastatin (Supplementary Fig. S17). The in vivo biodistribution of the radioimmunoconjugates in H82 tumors demonstrated a progressively increasing concentration of radioactivity in the tumors over time. The tumoral uptake values of the [⁸⁹Zr]Zr-DFO-SC16 radioimmunoconjugate were similar in tumors of saline-treated and statin-treated mice (Fig. 6). Additional immunofluorescence demonstrated that H82 tumors contain an undetectable expression of CAV1 protein, which is in accordance with the low CAV1 protein levels in H82 cells (Supplementary Fig. S1).

These results suggest that statins do not increase anti-DLL3 antibody accumulation in H82 tumors.

In HER2-expressing NCIN87 and EGFR-expressing A431 xenografts, administering mice with a statin enhanced ⁸⁹Zr-labeled antibody accumulation at the tumor site (Fig. 6). ⁸⁹Zr-labeled huJ591 tumor accumulation was higher in PSMA-overexpressing tumors of statin-treated mice when compared with tumors of saline-treated mice at later time points of antibody accumulation (24 hours and 48 hours), but the values were similar at 4 hours and 8 hours. These results suggest that statins increase, but do not accelerate, anti-PSMA antibody accumulation in the PSMA-overexpressing PC3pip tumors. In DLL3⁺ H82 xenografts, we did not observe differences in antibody accumulation at the tumor site between saline- or statin-treated mice. H82 exhibited lower levels of CAV1 when compared with A431 and PC3pip tumor cells (Supplementary Fig. S1). Taken together, these results seem to indicate that for receptors that are overexpressed in tumor cells and rapidly internalized, a pharmacologic approach modulating receptors with a statin has the potential to enhance antibody accumulation in EGFR-expressing and PSMA-overexpressing tumors.

The mevalonate pathway is essential in the synthesis of cellular cholesterol. Statins are low-cost drugs clinically prescribed for patients with cardiovascular diseases and hypercholesterolemia (2). The ability of statins to deplete cholesterol is explained by inhibition of HMG-CoA. Because cellular membranes contain cholesterol-rich domains, cholesterol biosynthesis through the mevalonate pathway is also important in the membrane dynamics and intracellular traffic (55). The exposure of cell membranes to cholesterol-depleting drugs causes the invaginated caveolae to flatten (17) while temporarily reducing caveolin-1 protein levels (5, 6, 10, 19, 33) and dissociating cavins from the cell membrane (reviewed in ref. 16). In addition to modulating caveolae-mediated endocytosis, cholesterol depletion interferes with clathrin-dependent and clathrin-independent endocytic pathways (25). At the cell membrane, distinct receptor–ligand complexes concentrate in cholesterol-rich structures (16, 19, 34). This leads to aberrant membrane trafficking and provides proximity for receptor interaction, which can induce cross-talk of oncogenic signaling pathways (16, 56). The endocytic processes interfere with surface levels of specific membrane receptors–important, for example, for antibody binding to tumors in approaches to imaging or therapy that use antibodies directed to cell-surface receptors. Here we show that three chemically different statins, clinically prescribed as cholesterol-depleting drugs, can temporarily modulate endocytic processes to improve antibody binding to tumor xenografts.

In this study, we show that cholesterol-depleting drugs induced a significant increase in tumors' avidity for anti-EGFR and anti-PSMA antibodies in A431 and PC3pip xenografts, but not in DLL3-expressing H82 tumors (Fig. 6). Future studies are necessary to determine whether the low efficacy of statins in enhancing anti-DLL3 antibody tumor accumulation is due to the low levels of CAV1 in H82 tumors. The cholesterol-depleting drugs lovastatin, simvastatin, and rosuvastatin were used, within the recommended dose range of statins in the clinic. Contrary to our preclinical studies using only two doses of a statin, statins are taken by patients every day. In addition to interfering with caveolae, cholesterol-depleting drugs affect other clathrin-dependent (23, 24) and clathrin-independent endocytosis pathways, such as FEME and the CLIC/GEEC pathway (25, 26). Similar to our previous studies in the context of HER2 (19, 33), lovastatin depleted CAV1 in A431 and PC3pip xenografts (Figs. 3 and 5). In addition to CAV1, lovastatin induced temporal alterations in other endocytic proteins. These results suggest that statin-mediated generalized disruption of endocytic pathways leads to enhanced antibody tumor accumulation. While lovastatin depleted cavin-1 and endophilin, the statin enhanced clathrin and dynamin between 12 and 16 hours upon mice gavage. Previous studies have demonstrated that EGFR internalizes through clathrin and FEME pathways (32, 37), and PSMA or anti-PSMA antibodies internalize through a clathrin pathway (40). An increase in clathrin levels upon lovastatin administration could explain the enhancement in antibody accumulation in A431 and PC3pip xenografts. In addition, statin-mediated alteration of several endocytic mechanisms could also explain the previously reported enhancement in anti-HER2 antibody tumor accumulation after administering mice with lovastatin (19, 33). Because CAV1 knockdown or treatment with statins enhances EGFR at the cell membrane, future studies are necessary to determine whether an increase in clathrin upon CAV1 depletion occurs as a transient compensatory mechanism for antibody uptake. Additional studies in a genetically engineered mouse model are necessary to define the specific contribution of CAV1 in anti-HER2, anti-EGFR, and anti-PSMA tumor accumulation.

Tumors overexpressing HMG-CoA exhibit dysregulation of the mevalonate pathway, which ultimately leads to cell transformation and induction of oncogenic signaling (e.g., Ras and RhoA; ref. 57). There are several mechanisms by which statins induce cell death (1, 58, 59): inhibition of proliferation (9), inhibition of GTPase oncogenes, and by decreasing the production of inflammatory cytokines (60). In castration-resistant prostate cancer, CAV1 has been suggested as a predictive biomarker and statin-mediated CAV1 depletion has demonstrated therapeutic efficacy (5). Others have reported that in prostate cancer cells, statins decrease androgen receptor signaling and cell proliferation (11). Statins have also demonstrated synergy with therapeutic mAbs and tyrosine kinase inhibitors against HER2 and EGFR (7, 15, 19). Further studies are necessary to evaluate whether an increase in anti-PSMA or anti-EGFR antibody accumulation in xenografts (Fig. 6) results in improved targeted therapies.

Previous therapeutic studies have reported higher anti-cancer efficacy for lipophilic than hydrophilic statins due to their ability to cross cell membranes (4, 8). Our in vivo studies suggest that lipophilic statins enhance antibody accumulation to a higher extent than hydrophilic rosuvastatin (Figs. 3 and 5). Lovastatin and simvastatin are lipophilic prodrugs and less than 5% of an oral dose of these drugs reaches systemic circulation in the open hydroxy acid form. This suggests that, after entering systemic circulation, only low amounts of lovastatin or simvastatin could accumulate in tumors. In EGFR-expressing A431 tumors, lovastatin and simvastatin increased membrane-bound panitumumab (Fig. 2C). However, a different pattern was observed in PSMA-overexpressing tumors. While lovastatin increases membrane-bound huJ591, simvastatin enhances antibody internalization (Fig. 4A). Differences in the pharmacodynamic profile of statin prodrugs might explain the inefficacy of simvastatin to increase huJ591 accumulation in PC3pip xenografts (Fig. 5D). Indeed, oral absorption of simvastatin lactone is more complete (80%–85%) when compared with the lactone prodrug lovastatin. The hydrophilic rosuvastatin is characterized by a bioavailability of 20%. While lipophilic statins accumulate in hepatocyte cell membranes or other cell membranes (such as cancer cells) by passive diffusion, rosuvastatin accumulation in cells is governed by an active mechanism (3). Contrary to hepatocytes, cancer cells do not express the OATP1B1 and 1B3 transporters responsible for rosuvastatin cellular uptake (61). The low ability of rosuvastatin to accumulate in cancer cells when compared with lipophilic lovastatin or simvastatin could result in low cholesterol depletion at the tumor site. Indeed, tumors of mice treated with rosuvastatin demonstrated lower antibody accumulation when compared with tumors from lovastatin-treated mice (Figs. 3D and 5D). Future pharmacodynamic studies are necessary to determine the ability of statins to cross cancer cell membranes and accumulate at the tumor site after oral administration.

HER2 and EGFR are tumor biomarkers characterized by a heterogeneous expression at the cell membrane of cancer cells (32, 62). Because the efficacy of monoclonal antibodies is dependent on target availability at the cell surface, previous studies demonstrated that tumors with low EGFR endocytosis respond better to cetuximab treatment (63). For tumors with heterogeneous expression of HER2 and EGFR, statin-mediated temporal modulation of endocytosis enhances HER2 (19) and EGFR presence at the cell membrane (Fig. 3A). Temporal modulation of the endocytosis of membrane receptors has the potential to enhance antibody-directed imaging and therapeutic approaches (19, 32, 33). After trastuzumab, cetuximab, or panitumumab bind HER2 or EGFR at the cell surface, the antibody-bound receptor is rapidly internalized, resulting in receptor downregulation. The internalized HER2 or EGFR can either be degraded or recycled back to the cell surface. In our experiments, the chimeric monoclonal IgG1 antibody cetuximab and the fully human monoclonal IgG2 antibody panitumumab were used in experiments with EGFR⁺ A431 cells. The treatment of mice with statins enhanced panitumumab accumulation in A431 tumors. On the other hand, cetuximab accumulation was similar at the time points 4 and 24 hours of antibody administration in tumors of control or statin-treated mice. Changes in antibody accumulation can be explained by differences in antibody internalization. Previous studies have demonstrated that cetuximab internalizes more quickly than panitumumab (53).

In addition to HER2 and EGFR, PSMA has also been described as a heterogeneous tumor biomarker (64). Our immunofluorescence experiments (Fig. 5A) support previous reports describing how PC3pip xenografts homogenously and stably express PSMA at the cell membrane (65). Although we observed a negative correlation between PSMA and CAV1 at the protein levels, we did not detect changes in PSMA staining at the cell membrane after CAV1 knockdown or statin treatment. The weak ability to enhance PSMA membrane availability after CAV1 knockdown can be also explained by the low levels of caveolae in this cell line (51). Similarly, depletion of CAV1 did not enhance HER2 membrane availability in cell lines that do not possess caveolae (e.g., HER2-expressing SKBR3 breast cancer cells; ref. 19). PSMA internalization occurs through multiple mechanisms that are dependent on the cell line (18, 40). PSMA–huJ591 complexes have been suggested to internalize via a clathrin-mediated mechanism (40, 41). Our in vivo data show that, although huJ591 has similar tumor uptake at initial time points, PC3pip tumors of statin-treated mice showed higher huJ591 uptake at 48 hours after antibody administration (Fig. 6). In our in vitro experiments, simvastatin and rosuvastatin increased huJ591 antibody internalization (Fig. 4C), suggesting that statin-mediated clathrin increase results in enhanced anti-PSMA antibody uptake in PC3pip tumors.

The members of the HER family (including EGFR and HER2) dimerize when present at the cell membrane. HER activation and dimerization induce downstream oncogenic signaling (27). Therefore, it would seem counter-productive to delay HER internalization by using statins, as this could amplify the downstream oncogenic cascade induced by the dimerized receptor. However, our data demonstrate that the increase in receptor availability at the cell membrane leads to enhanced antibody-tumor binding at early time points. Being a transient and controlled pharmacologic approach that occurs upon administration of a statin, normalization of the cellular distribution of the receptor occurs for later time points. In our previous studies, we have demonstrated that lovastatin increases HER2–HER2 and HER2–EGFR inactive dimers (33), plausibly because our lovastatin pharmacologic approach induced a transient effect on increasing receptor availability at the cell membrane. In a context different from HER signaling, other studies have demonstrated that a release of cavin-1 from caveolae promotes a proapototic signaling pathway (49). Further studies are necessary to determine whether the alterations observed in endocytic proteins and membrane receptors result in changes in oncogenic signaling after treatment with single, fractionated, and prolonged doses of lovastatin in the presence and absence of antibody therapies.

Noninvasive molecular imaging is a powerful technology to image the membrane dynamics of antibody binding to tumors during modulation of target accessibility (19, 33, 47, 66). Similar to our previous studies (33, 47), preclinical immunoPET shows the potential of cholesterol-depleting drugs as a pharmacologic approach to enhance antibody accumulation. Our approach using statins enhanced ⁸⁹Zr-labeled panitumumab, cetuximab, and huJ591 tumor accumulation at 48 hours after tracer injection, suggesting the use of this strategy to improve antibody accumulation in tumors characterized by a heterogeneous expression of the tumor target. In the field of molecular imaging using radiolabeled antibodies for tumor detection, our studies present an important approach. The pharmacokinetics of full-length antibodies requires their labeling with long-lived positron-emitting radionuclides; images with high contrast are obtained between 3 and 5 days after administration of the radiolabeled antibody. In our studies, we observed high-contrast images between 1 and 2 days in mice treated with statins. These results suggest that a PET scan could be acquired in patients at early time points; this could result in a faster diagnosis of the disease and would not require the patient to return to the clinic for a PET scan 3–5 days after injection of the radiolabeled antibody. Further PET studies are necessary to determine tumor populations that could benefit from statin administration to enhance antibody accumulation and efficacy.

Limitations of our study include the in vitro use of high statin concentrations. Although previous studies have shown that statin solutions in the micromolar range deplete CAV1 (19) in vitro, the concentration of statins in the serum of patients ranges from 0.002 to 0.1 μmol/L (67), and their accumulation within tumor tissues remains unclear. In addition, statin-mediated modulation of caveolae in vivo requires further investigation; the endocytic potential of caveolae in vivo is still unclear (reviewed in ref. 16). Furthermore, it is not clear whether the use of a prostatic cancer cell line containing lower and heterogeneous PSMA levels when compared with PC3pip (e.g., LNCaP) could be a better biological model for our studies. Because PC3pip cells correspond to prostate cancer cells that were transduced to stably express PSMA at high levels, we expect that modulation of membrane PSMA with statins will be more pronounced in a model that expresses PSMA physiologically and heterogeneously and that possess caveolae. Furthermore, the DLL3⁺ lung tumor model used in our studies contain low levels of CAV1 that might explain the similar SC16 antibody accumulation in saline-treated versus statin-treated tumors. Also unknown are the mechanisms by which DLL3 internalizes in tumor cells. Future experiments are necessary to determine SC16 antibody accumulation in DLL3⁺ lung tumors expressing CAV1 as well as in biological models where CAV1 is modulated using nonpharmacologic approaches.

Our data suggest that acute statin treatment with appropriate pharmacokinetics and pharmacodynamics are potential adjuvants for specific antibody-targeted therapies. Lovastatin, simvastatin, and rosuvastatin enhanced anti-EGFR and anti-PSMA antibody accumulation while also temporarily modulating proteins of the endocytic trafficking systems and enhancing target density at the cell membrane. Further investigations are necessary to define the statin doses necessary for tumoral cholesterol modulation and membrane receptor modulation, and the therapeutic value of combining cytotoxic antibodies with statins.

P.M.R. Pereira reports grants from NIH, Tow Foundation from the MSKCC Center for Molecular Imaging and Nanotechnology, and Alan and Sandra Gerry Metastasis and Tumor Ecosystems Center of MSKCC during the conduct of the study. J.S. Lewis reports grants from NIH during the conduct of the study. No potential conflicts of interest were disclosed by the other authors.

P.M.R. Pereira: Conceptualization, data curation, formal analysis, supervision, validation, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing. K. Mandleywala: Validation, investigation, visualization, methodology, writing-review and editing. A. Ragupathi: Visualization, methodology. J.S. Lewis: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, writing-original draft, project administration, writing-review and editing.

The authors acknowledge the Radiochemistry and Molecular Imaging Probe Core, and Molecular Cytology Core Facility, which were supported by NIH grant P30 CA08748. This study was supported in part by the Geoffrey Beene Cancer Research Center of MSKCC (to J.S. Lewis) and NIH NCI R35 CA232130 (to J.S. Lewis). We gratefully acknowledge Mr. William H. and Mrs. Alice Goodwin and the Commonwealth Foundation for Cancer Research and The Center for Experimental Therapeutics of MSKCC. P.M.R. Pereira acknowledges the Tow Foundation Postdoctoral Fellowship from the MSKCC Center for Molecular Imaging and Nanotechnology and the Alan and Sandra Gerry Metastasis and Tumor Ecosystems Center of MSKCC. We thank Dr. Sai Kiran Sharma and Dr. Nagavarakishore Pillarsetty for critical discussions in our experiments. J591 was a generous gift of Professor Neil Bander at Weill Cornell. We would also like to acknowledge Ricardo D'Oliveira Albanus from Department of Computational Medicine & Bioinformatics, University of Michigan for assistance in RStudio analyses. Part of the results shown in Supplementary Information of this study are based upon data generated by the TCGA Research Network (https://www.cancer.gov/tcga) and CCLE (https://portals.broadinstitute.org/ccle). Figure 2A was made using Biorender.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.