Immunoconjugate Induces Cytotoxicity in Human Uveal Melanoma Cells

In Vitro Targeting of NG2 Antigen by 213Bi-9.2.27 ␣-Immunoconjugate Induces Cytotoxicity in Human Uveal Melanoma Cells

Yong Li,1,2 Jian Wang,1,3 Syed M. Abbas Rizvi,1 Martine J. Jager,4 Robert M. Conway,2 Francis A. Billson,2 Barry J. Allen,1,2,3 and Michele C. Madigan2

PURPOSE. To examine uveal melanoma cell lines for the expres- mately 95% of all uveal melanomas. Primary uveal melanomas sion of human melanoma proteoglycan (NG2) using monoclo- may be treated successfully with enucleation, local tumor re- nal antibody (mAb) 9.2.27 and subsequently to assess the in section, laser photocoagulation, plaque radiotherapy, proton vitro specificity and cytotoxicity of mAb 9.2.27 conjugated to beam therapy, or combinations of therapies.1–6 Earlier studies 213 213 the ␣-particle–emitting radioisotope bismuth ( Bi-9.2.27) found minimal benefits using intravenous chemotherapy or for uveal melanoma cells. immunotherapy for primary and metastatic uveal melanoma, METHODS. Immunocytochemistry and flow cytometry were though some success with these modalities has been reported used to examine OCM-1, OCM-3, OCM-8, OMM-1, Mel202 and recently.6–9 Proton beam therapy or local plaque therapy gen- 92–1 melanoma cell lines for NG2 expression. Melanoma cells erally produces good clinical responses.10 However, radiation- were treated with test (213Bi-9.2.27) or control (213Bi-A2) ␣-im- associated morbidity remains a problem and usually involves munoconjugates (AICs). The specific cytotoxicity of 213Bi- long-term visual complications. Most patients with uveal mel- 9.2.27 AIC was evaluated using an MTS (3-(4,5-dimethylthiazol- anoma display no detectable evidence of metastases at the time 2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfenyl)-2H-tetrazolium, of diagnosis; however, within 5 years of enucleation, meta- inner salt) assay. Cell death was also assessed using TUNEL. static disease, predominantly to the liver, develops in approx- RESULTS. OCM-1, OCM-8, OMM-1, and Mel202 cells strongly imately 40% of patients. Once detected clinically, metastatic expressed NG2. OCM-3 cells showed moderate expression and disease is resistant to most therapies and is usually fatal within 92–1 cells were NG2-negative. 213Bi-9.2.27 specifically killed 12 months. Local therapies are ineffective for micrometastatic NG2-positive OCM-1, OCM-8, and OMM-1 cells in a concentra- disease, and adjunctive therapies to prevent or treat melanoma tion-dependent manner. D0 values for 37% cell survival of metastases are lacking. New approaches to treat uveal mela- NG2-positive OCM-1, OCM-8, and OMM-1 cells were 5.8, 5.0, noma metastases are, therefore, necessary. ␮ ␮ and 5.6 Ci, respectively, and the value was 43.4 Ci for Targeted ␣-particle therapy (TAT) is an emerging therapeu- NG2-negative 92–1 cells. tic modality that uses a labeled antibody or protein to selec- 213 CONCLUSIONS. The specific cytotoxicity of Bi-9.2.27 AIC for tively target cancer cells and to deliver a lethal dose of short- NG2-positive, but not NG2-negative, cells suggests NG2 is a range, highly cytotoxic ␣-radiation. This approach has the 213 suitable target for ␣-immunotherapy in uveal melanoma. Bi- capacity to greatly increase the efficacy of tumor cell killing 9.2.27 AIC used directly or as adjunct therapy may be a prom- and to reduce damage to the surrounding normal tissue. Stable ising new agent for treating NG2-positive uveal melanomas or ␣-conjugates have been prepared with labeling yields of up to metastases. (Invest Ophthalmol Vis Sci. 2005;46:4365–4371) 95% for antibodies and proteins labeled with bismuth-213 DOI:10.1167/iovs.05-0559 (213Bi) using cyclic diethylenetriaminepentaacetic acid anhydride (cDTPA) or DTPA-CHX-A’’ as chelators.11 213Bi has a ϭ ␣ veal melanomas affect the iris, ciliary body, and choroid short half-life (t1/2 46 minutes) and emits -particles with Uand are the most common primary eye tumors in adults; high linear energy transfer radiation and a short range (80 ␮m). ciliary body and choroidal melanomas constitute approxi- As such, ␣-particles have several advantages over ␤-particles. They cause double-strand DNA damage that is not easily re- paired by the cell because of the very high linear energy transfer (ϳ100 times greater than for ␤-particles); their cyto- 1 From the Centre for Experimental Radiation Oncology, Cancer toxicity is not affected by oxygen; and they are much more Care Centre, St. George Hospital, Kogarah, NSW, Australia; 2Discipline cytotoxic, requiring as few as 6 or 7 disintegrations for inter- of Clinical Ophthalmology, Save Sight Institute, University of Sydney, ␣ Sydney, NSW, Australia; 3Department of Medicine, University of New nalized -particles and approximately 25 disintegrations for South Wales, Kensington, NSW, Australia; and 4Department of Oph- surface-bound ␣-emitters to kill a cell.12 ␣-Particle therapy has thalmology, Leiden University Medical Center, Leiden, The Nether- been used in single-cell disorders, such as leukemia, lym- lands. phoma, and micrometastatic carcinoma,13–15 in which rapid Supported in part by grants from the Sydney Foundation for targeting to cancer cells is possible. Medical Research (MCM), the Ophthalmic Research Institute of Aus- mAb 9.2.27 is highly specific for the melanoma-associated tralia, and the US Department of Defense (YL). 16,17 Submitted for publication May 9, 2005; revised August 12, 2005; chondroitin sulfate proteoglycan NG2, which is expressed 18,19 20 accepted October 18, 2005. on most cutaneous and uveal melanomas. Stably che- Disclosure: Y. Li, None; J. Wang, None; S.M. Abbas Rizvi, None; lated 213Bi-9.2.27 has recently been found to be highly specific M.J. Jager, None; R.M. Conway, None; F.A. Billson, None; B.J. and cytotoxic to skin melanoma cells in vitro11 and to com- Allen, None; M.C. Madigan, None pletely regress tumor growth in a xenograft mouse model of The publication costs of this article were defrayed in part by page skin melanoma after local injection.14,21 In this study, we charge payment. This article must therefore be marked “advertise- ment ” in accordance with 18 U.S.C. §1734 solely to indicate this fact. examined human uveal melanoma cell lines for NG2 expres- 213 Corresponding author: Michele C. Madigan, Save Sight Institute, GPO sion and evaluated the in vitro efficacy and specificity of Bi- Box 4337, Sydney NSW 2001, Australia; [email protected]. 9.2.27 as a cytotoxic agent for these cell lines.

Downloaded from jov.arvojournals.org on 09/28/2021 4366 Li et al. IOVS, December 2005, Vol. 46, No. 12

MATERIALS AND METHODS Ridge, National Laboratory, Oak Ridge, TN). Using published methods,11 mAbs 9.2.27 and A2 were conjugated with the chelator, cyclic Antibodies diethylenetriaminepentaacetic acid anhydride (cDTPA; Sigma-Aldrich Pty., Ltd., Castle Hill, NSW, Australia). Conjugated mAbs 9.2.27 and A2 mAb 9.2.27, with a high specificity for human melanoma cells related were measured by plate reader at 280 nm using commercial software to a 250-kDa N-linked glycoprotein and a Ͼ400-kDa proteoglycan (ProMax; Bio-TEC Instruments Inc., Winooski, VT) and were purified component, was kindly provided by Peter Hersey (Department of on a PD-10 column (Amersham Biosciences Ltd., Bucks, UK). 213Bi was Oncology and Immunology, Newcastle Mater Misercordiae Hospital, eluted from the 225Ac column with 250 ␮L freshly prepared 0.15 M Newcastle, NSW, Australia). Rabbit anti–NG2 chondroitin sulfate pro- hydriodic acid as the (BiI )2- anion species, neutralized to pH 4 to 4.5 teoglycan polyclonal antibody (pAb) was obtained from Chemicon 5 with the addition of 3 M ammonium acetate and was immediately used International (Temecula, CA). Nonspecific mouse anti–human IgG1 to radiolabel the mAb construct. Two to 3 hours was allowed for 213Bi monoclonal antibody (A2) was kindly provided by Andrew Collins to grow back in the generator for the next elution. Radiolabeling (Department of Microbiology, University of New South Wales, Sydney, efficiency was determined by instant thin-layer chromatography, with NSW, Australia). Mouse myeloma IgG2a isotype control and rabbit a 10-␮L aliquot of the final reaction mixture applied to silica gel– immunoglobulin isotype control were supplied by Zymed Laboratories coated fiber sheets (Gelman Science Inc., Ann Arbor, MI). The paper Inc. (San Francisco, CA). Goat anti–mouse IgG (Alexa-488) and strepta- strips were developed using 0.5 M sodium acetate (pH 5.5) as the vidin (Alexa-594) conjugates were purchased from Molecular Probes solvent. Then the paper strips were cut into four sections, and the Inc. (Eugene, OR). Goat anti–mouse or sheep anti–rabbit fluorescein ␥-emissions from the radioisotope were counted in each section using isothiocyanate (FITC)–conjugated antibodies were purchased from a 340- to 540-keV window. Radiolabeled protein stays in the section of Chemicon International. origin, whereas free radioisotope moves with the solvent front section. 213 Cell Culture Radiolabeling efficiency was 80% to 95% for Bi-9.2.27 (test) and 213Bi-A2 (control) AICs. Five human uveal melanoma cell lines (OCM-1, OCM-3, OCM-8, Mel202, 92–1) and one cell line derived from a uveal melanoma skin In Vitro Cytotoxicity Assay metastasis (OMM-1) were grown in either RPMI 1640 medium (OCM-3, OCM-8, 92–1, Mel202) or DMEM medium (OCM-1 and OMM-1) sup- Three NG2-positive cell lines (OCM-1, OCM-8, OMM-1) and a NG2- negative cell line (92–1) were selected for in vitro AIC treatment. To plemented with 10% fetal bovine serum (FBS), 2 mM glutamine, 50 213 213 IU/mL penicillin, and 50 ␮g/mL streptomycin. Tissue culture reagents ensure equal activities, Bi-9.2.27 and Bi-A2 preparations were were supplied by ThermoElectron Pty. Ltd. (Noble Park, Victoria, measured using a radioisotope calibrator (Atomlab 200; Biodex Medi- Australia). Cells were maintained in a humidified incubator at 37°C and cal System, Shirley, NY). Conjugate solutions were neutralized to pH 5% CO . 7.0 by the addition of 10% (vol/vol) 1 M NaHCO3 (pH 9.0). After this, 2 five serial activities of 213Bi-9.2.27 (1 ␮Ci, 2 ␮Ci, 4 ␮Ci, 8 ␮Ci, 10 ␮Ci) 213 Immunocytochemistry and a dose of Bi-A2 at the highest activity (10 ␮Ci) were prepared in 100 ␮L RPMI 1640/5% FBS and were added to 96-well plates in Cells were seeded onto glass coverslips in a 24-well plate at a density triplicate containing 2 ϫ 104 OCM-1, OCM-8, OMM-1, or 92–1 cells, 4 of 10 cells/well and were cultured overnight in growth medium. After respectively. Controls were performed in triplicate in the same 96-well rinsing in PBS (pH 7.4), cells were fixed in cold acetone for 5 minutes, plate for each experiment and consisted of cDTPA-9.2.27, mAb 9.2.27, rinsed again in PBS, and blocked in 10% normal goat serum (NGS)/PBS and RPMI 1640/5% FBS medium alone. Plates were incubated over- for 20 minutes at room temperature (RT). Cells were incubated overnight at 37°C, and cell morphology was subsequently assessed before night at 4°C in mouse anti–human NG2 (IgG2a; 1:100 dilution), rabbit the MTS assay. anti–human NG2 pAb (1:200 dilution), or isotype control (mouse Cells were then washed in DPBS, incubated in 100 ␮L serum-free, IgG2a; 1:100 dilution or rabbit immunoglobulins, 1 ␮g/mL). After phenol-red free RPMI 1640 containing 20 ␮L reagent (Cell Titer 96 rinsing in PBS, coverslips were incubated for 1 hour at RT in Alexa- Aqueous One Solution; Promega, Madison, WI) for 3 hours at 37°C. 488–conjugated goat anti–mouse or goat anti–rabbit antibody (1:1000 The reaction was stopped by the addition of 10% sodium dodecyl dilution). After a final rinse in PBS, coverslips were mounted on slides sulfate, and the absorbance of each well was recorded at 490 nm using in glycerol, and viewed using fluorescence microscopy. a plate reader (Spectro Max; Bio-Rad, Hercules, CA). The absorbance reflects the number of surviving cells. Blanks were subtracted from all Flow Cytometry data, and results were analyzed (Prism software; GraphPad Software Cell-surface NG2 expression was detected using flow cytometry. Inc., San Diego, CA). Briefly, confluent adherent cells were rinsed twice with PBS and detached with a scraper. Cells (0.5 ϳ 1.0 ϫ 106) were washed twice Assessment of Cell Death in cold Dulbecco’s phosphate-buffered saline (DPBS) with 5% FBS TUNEL Technique. To assess whether AIC treatment induced (DPBS/FBS; 200g, 8 minutes) and resuspended in 80 ␮L DPBS/FBS. apoptotic cell death, cultured OCM-1, OCM-8, OMM-1, and 92–1 cells Cells were then incubated for 60 minutes at 4°C in either an isotype were treated with a low or a high concentration of 213Bi-9.2.27 (2 ␮Ci control (mouse IgG2a or rabbit immunoglobulin), mouse anti–human or 10 ␮Ci, respectively) and with 213Bi-A2, cDTPA-9.2.27, or medium NG2 (IgG2a), or rabbit anti–human NG2 pAb. After washing with alone at 37°C overnight. After treatment, cells were washed with DPBS DPBS/FBS, the cells were resuspended and incubated in goat anti– and were harvested by scraping. Cell cytospins using 3 ϫ 104 cells/100 mouse FITC-conjugated antibody (1:80 dilution) or sheep anti–rabbit ␮L were prepared using a centrifuge (Heraeus Megafuge 1.0R; DJB FITC-conjugated antibody (1:40 dilution) for 45 minutes in the dark at Labcare, Buckinghamshire, UK) and were fixed in 2% paraformalde- 4°C. The cells were washed again and resuspended in 0.5 mL DPBS/ hyde at RT for 30 minutes. Apoptosis was detected using a modified FBS. Autofluorescence was subtracted in all experiments. Ten thou- TUNEL method.23,24 Briefly, cytospins were rinsed in terminal deoxy- sand cells in each sample were counted, and the data were presented nucleotidyl transferase (TdT; pH 7.2) buffer for 10 minutes at RT and as histograms. All data were analyzed (CELLQuest software; Becton- then incubated at 37°C for 1 hour in reaction mixture containing 10.5% Dickinson, San Jose, CA). TdT, 0.42% biotin-16 to 2Ј-deoxyuridine-5Ј-triphosphate (dUTP; Roche Molecular Sciences Pty. Ltd., Sydney, NSW, Australia), and 0.13% ter- Radiolabeling of mAbs with Radioisotope minal transferase enzyme (Roche Molecular Sciences) in Milli-Q water. 213Bi was produced from the 225Ac/213Bi generator22; the 225Ac column The reaction was stopped by immersion in 2ϫ SSC (3 M sodium was purchased from the United States Department of Energy (Oak chloride and 0.3 M sodium citrate, pH 7.0) for 15 minutes at RT,

Downloaded from jov.arvojournals.org on 09/28/2021 IOVS, December 2005, Vol. 46, No. 12 ␣-Immunotherapy for NG2-Positive Uveal Melanoma 4367

followed by rinsing in 1% bovine serum albumin (BSA) in PBS. To label DNA fragments, cytospins were then incubated in streptavidin-Alexa 594 conjugate (1:1000 dilution) for 1 hour at RT, rinsed in PBS, coverslipped (Universal Mount; Invitrogen Life Technologies, Carls- bad, CA), and examined using a fluorescence microscope (Diaplan; Leitz, Wetzlar, Germany) at 50ϫ magnification. The specificity of TUNEL reactivity was confirmed in parallel negative (omitting terminal transferase enzyme from the reaction mixture) and positive (human retina) controls. Apoptotic cells were identified by TUNEL labeling and the presence of nuclear chromatin fragments. Acridine Orange/Ethidium Bromide Staining. Acridine orange/ethidium bromide (AO/EB) staining was used to identify apoptotic cells, as described previously.25 Briefly, cultured OCM-1, OCM-8, OMM-1, and 92–1 cells were treated with a low or a high concentration of 213Bi-9.2.27 (2 ␮Ci and 10 ␮Ci, respectively) or with 213Bi-A2, cDTPA-9.2.27 or medium alone at 37°C overnight. At 24 and 48 hours, 25 ␮L floating control or treated cells were stained with 1 ␮L 100 ␮g/mL AO/EB in PBS; 10 ␮L stained cells was placed onto a glass slide, coverslipped, and examined immediately using a fluorescence microscope (Diaplan; Leitz).

RESULTS Expression of NG2 on Uveal Melanoma Cell Lines Immunocytochemistry. OCM-1, OMM-1, OCM-3, OCM-8, and Mel202 uveal melanoma cell lines displayed strong immunoreactivity for NG2 antigen compared with the isotype control (Fig. 1A–E). The 92–1 cell line displayed minimal NG2 immunoreactivity (Fig. 1F), similar to the isotype control (Fig. 1G). NG2-immunolabeled cells displayed obvious punctate staining clearly localized to cell membranes (Fig. 1A-E, H). Similar patterns of immunolabeling were seen with mAb 9.2.27 and pAb human NG2 (Fig. 1H). Flow cytometry conﬁrmed these observations. Flow Cytometry. OCM-1, OMM-1, OCM-8, and Mel202 cell lines strongly expressed NG2 immunolabeling (98%, 98%, 99%, FIGURE 1. NG2 immunoreactivity in melanoma cell lines. (A) OCM-1, (B) OMM-1, (C) OCM-3, (D) OCM-8, and (E) Mel202 cell lines display and 94% positive cells, respectively; Figs. 2A, 2C, 2D, 2F) strong immunostaining for NG2, with obvious punctate staining clearly compared with the isotype control, and OCM-3 cells showed localized to cell membranes. (F) 92–1 cells are negative for NG2 moderate immunolabeling (85% positive cells; Fig. 2E). NG2 antigen, similar to the isotype control (G). (H) Distinct cell membrane immunolabeling of 92–1 cells was similar to that of the isotype immunolabeling is also obvious using polyclonal human anti–NG2 control (2% positive cells; Fig. 2B). Similar results were found antibody, seen in this example of OMM-1 melanoma cells. using rabbit anti–human NG2 pAb (not shown).

Cytotoxicity death, and displayed the spindle or epithelioid morphology characteristic of the respective cell lines (Figs. 4A, 4B, 4E, 4F). 213Bi-9.2.27 AIC was specifically cytotoxic to NG2-positive melanoma cells (OCM-1, OMM-1, and OCM-8) in a concentra- Cell Death tion-dependent fashion (Fig. 3A–C) but not to 92–1 NG2-negative melanoma cells (Fig. 3D). For 37% cell survival, the Cells treated with nonspecific AIC or medium alone displayed no obvious TUNEL labeling (Figs. 5A, 5B). Some TUNEL-posi- corresponding D0 values for the NG2-positive cell lines— ␮ ␮ tive OCM-1, OMM-1, and OCM-8 cells were observed after OCM-1, OMM-1, and OCM-8—were 5.8 Ci, 5.0 Ci, and 5.6 213 ␮Ci, respectively, and 43.4 ␮Ci for NG2-negative 92–1 mela- treatment with Bi-9.2.27 AIC (Figs. 5C, 5D). However, when noma cells. A nonspecific control ␣-immunoconjugate (213Bi- the terminal transferase enzyme was omitted, no TUNEL-positive cells were seen (Fig. 5E). AO/EB staining of floating cells A2) that did not specifically target the NG2 protein induced 213 minimal cytotoxicity in all cell lines (Fig. 3A–D). 24 hours after treatment with Bi-9.2.27 AIC showed some cells with nuclear chromatin condensation and fragmentation—morphologic features consistent with apoptotic cell Morphology death (Figs. 5F, 5H). Viable cells were also seen in control After 24-hour treatment with 213Bi-9.2.27 (2-␮Ci and 10-␮Ci cultures (Fig. 5G). We also observed degenerate cells25 that doses), NG2-positive cells (OCM-1, OCM-8, OMM-1) detached stained with EB in 213Bi-9.2.27 AIC–treated cultures (not from the plate and rounded up in a concentration-dependent shown). fashion. Pyknotic cells and some necrotic cells were seen in the floating cell population (Figs. 4C, 4D, 4G, 4H); adherent DISCUSSION cells were also visible, though much reduced in numbers compared with controls (Figs. 4D, 4H). Cells cultured with Monoclonal antibodies are being increasingly recognized for isotype control AIC (213Bi-A2), cDTPA-9.2.27, or medium alone their potential in anticancer therapeutics, with strategies that remained adherent to the plate, showed little evidence of cell include mAbs combined with cytotoxic drugs or conjugated

Downloaded from jov.arvojournals.org on 09/28/2021 4368 Li et al. IOVS, December 2005, Vol. 46, No. 12

FIGURE 2. Flow cytometric analysis of NG2 immunoreactivity on melanoma cell lines using mAb 9.2.27. Data are shown as histograms using a mouse IgG2a-negative control to de- termine background ﬂuorescence and to set the marker (M1). All cell lines except 92–1 show high levels of NG2 expression. (A) OCM-1, 98%; (B) 92–1, 2%; (C) OMM-1, 98%; (D) Mel202, 99%; (E) OCM-3, 85%; and (F) OCM-8, 94%; FL1-H log ﬂuores- cence intensity.

with radionuclides or immunologic effector cells. The selec- tions on primary uveal melanoma and the apparent absence of tion of target antigens and targeting vectors requires high mAb 9.2.27 immunoreactivity on normal ocular tissues,20 these specificity and affinity to cancer tissue. To date, one of the results support the use of NG2 as a potential effective tumor most widely explored strategies for enhancing the efficacy of cell surface marker for targeting primary uveal melanoma, antitumor antibodies is direct arming by chelator linkage to similar to observations in cutaneous melanoma.11,14,21 toxins or radionuclides.26 Targeted radiotherapy that uses ra- In the present study, we also observed similar levels of NG2 dioactive isotopes linked to mAbs or proteins specific for immunoreactivity on OCM-1 and OMM-1 cell lines—uveal and cancer cells has been proposed for various tumors.12,27–29 metastatic melanoma cell lines, respectively. A recent study30 We recently described NG2 expression in a series of pri- comparing mRNA expression profiling of cancer-related genes mary uveal melanomas and normal control eyes.20 In that in uveal melanoma and liver metastases from the same patient study, 95% (18/19) of uveal melanoma specimens expressed found some similarities in patterns of gene expression. These moderate to high levels of tumor cell surface mAb 9.2.27 observations suggest that circulating cancer clones from pri- immunoreactivity (NG2).20 However, in most melanoma-af- mary uveal melanomas may not necessarily lose NG2 expres- fected eyes, the retina, retinal pigmented epithelium, and cho- sion, but this remains to be confirmed in primary metastatic roid displayed low-level immunostaining. Furthermore, in con- uveal melanomas. Application of 213Bi-9.2.27 AIC to target trol eye specimens (without known disease; n ϭ 5), the retina metastatic uveal melanoma may be possible either systemically and choroid appeared negative for mAb 9.2.27.20 We did ob- or through isolated hepatic perfusion (IHP). A recent study serve evidence of moderate immunoreactivity in optic nerve using IHP of high-dose chemotherapy in a small group of axon bundles of control and melanoma-affected eyes; however, patients with uveal melanoma with metastases confined to the this immunostaining is not cellular but is associated with the liver indicates that this approach may produce tumor response myelin (unpublished observation, 2004). The present study in some patients.31 also demonstrates high levels of in vitro cell surface NG2 The in vitro cytotoxicity of 213Bi-9.2.27 to uveal melanoma expression on uveal melanoma cell lines OCM-1, OCM-3, cell lines in the present study was specific, concentration OCM-8, Mel202, and OMM-1 (but not 92–1 cells) using flow dependent, and directly related to the level of NG2 expressed

cytometry and immunocytochemistry with mAb 9.2.27 and a on these cells. The D0 values for the NG2-positive cell lines polyclonal NG2 antibody. Taken together with earlier observa- (OCM-1, OMM-1, and OCM-8) were 5.8 ␮Ci, 5.0 ␮Ci, and 5.6

Downloaded from jov.arvojournals.org on 09/28/2021 IOVS, December 2005, Vol. 46, No. 12 ␣-Immunotherapy for NG2-Positive Uveal Melanoma 4369

FIGURE 3. Representative cytotoxicity of (A) OCM-1, (B) OMM-1, (C) OCM-8, and (D) 92–1 melanoma cell lines. Twenty thousand cells are seeded in triplicate and treated with a range of concentrations of 213Bi- 9.2.27 or a maximum concentration (10 ␮Ci) of 213Bi-A2 nonspeciﬁc isotype control, respectively. After overnight incubation, cell survival is measured by MTS assay at 24 hours and is expressed as a percentage of controls. Each point represents a mean of three experiments.

␮Ci, respectively, and 43.4 ␮Ci for the NG2-negative 92–1 and may trigger cytoskeleton-dependent changes in cell mor- melanoma cell line. A similar pattern of cell killing was ob- phology and motility in response to the extracellular environ- served for OCM-1, OMM-1, and OCM-8 cells, consistent with ment.32–34 Both the growth control and the cell motility as- their similar levels of NG2 expression. Significantly, there was pects of NG2 function have been observed in studies of no cell killing for cDTPA-9.2.27 cold conjugate or mAb 9.2.27 cutaneous melanoma progression, with NG2-expressing mouse alone groups. From these observations, Ͻ10% of melanoma melanoma cells reported to grow faster and to be more meta- cells would be expected to survive after 10 ␮Ci 213Bi-9.2.27 static than their NG2-negative counterparts.18,35 Overexpres- AIC compared with survival of Ͼ90% melanoma cells for the sion of NG2 has also been found to increase tumor initiation, same activity of nonspecific 213Bi-A2. Clearly, 213Bi-9.2.27 can growth rates, neovascularization, and cellular proliferation, fac- specifically target and kill NG2-positive cells while sparing tors that predispose to poorer survival outcome.36 NG2-negative (92–1) uveal melanoma cells. Perivascular cells (mature and immature smooth muscle NG2 proteoglycan has been implicated in the growth and cells and pericytes) on arterioles and capillaries have been invasion of cutaneous melanoma, though its function in uveal observed to express NG2 immunoreactivity during normal de- melanoma is unclear. As a membrane-spanning molecule, NG2 velopment and in pathologic vascular remodeling, as seen, for can interact with extracellular and intracellular components example, during tumor growth.36–38 Smooth muscle cells and

FIGURE 4. Morphologic features of OCM-1 (A–D) and OCM-8 (E–H) cells 24 hours after AIC treatment. Cells cultured with isotype control AIC (213Bi-A2), cDTPA-9.2.27, or medium alone remain adherent to the plate, show little evidence of cell death, and display the characteristic spindle/epithelioid morphology of the respective cell lines (A, B, E, F). After treatment with 213Bi-9.2.27 AIC [(C, G)2␮Ci; (D, H)10␮Ci], many NG2-positive cells detach from the plate and round up, more obviously with 10 ␮Ci 213Bi-9.2.27 AIC (C, D, G, H). Some apoptotic and necrotic cells are seen in these detached cell populations (C, D, G, H). Adherent cells are also visible 24 hours after treatment (D, H).

Downloaded from jov.arvojournals.org on 09/28/2021 4370 Li et al. IOVS, December 2005, Vol. 46, No. 12

dem U. IOVS 2005;46:ARVO E-Abstract 4620). Further studies of NG2 antigen expression on vascular-associated cells in normal and tumor-affected human retina and choroid and in uveal melanoma are being pursued. Our recent studies of prostate cancer cells found a high percentage of TUNEL-positive cells after treatment with AICs, indicating the induction of cell death predominantly by apoptosis.42–44 An earlier electron microscopy study of AIC- treated murine lymphoma cells reported bizarre blebbing patterns, condensation of chromosomal material, and internu- cleosomal DNA fragmentation, also consistent with the induction of apoptotic cell death.45 In the present study, TUNEL- positive cells were seen 24 hours after 213Bi-9.2.27 AIC treatment of NG-2 positive cell lines; however, apoptosis was not the only mode of cell death observed. Many factors, including antigen affinity and antigen density, play important roles in the killing of targeted antigen-positive cells. For example, after binding the cell NG2 antigen, 213Bi-9.2.27 may form 213Bi- 9.2.27-NG2 complexes at the cell membrane, emitting ␣-particles that can kill uveal melanoma cells by causing double-DNA strand breaks.12 Alternatively, surface-bound 213Bi-9.2.27-NG2 complexes may be internalized by the cell with increased cell killing efficiency,14,15 as suggested in 213Bi-Herceptin–mediated killing of prostate cancer cells.44 The relative importance of these factors (antigen density, antigen affinity, and internal- ization) in AIC-mediated killing of uveal melanoma cells remains to be determined. We are conducting a phase 1 clinical trial for secondary or recurrent skin melanoma using intralesional injection of 213Bi- 9.2.27 AIC. To date, this study shows almost complete tumor cell kill in a diffuse area around the injection site compared with lesions injected with antibody alone, consistent with specific targeting of this AIC (unpublished data, 2004). The present study shows that 213Bi-9.2.27 AIC can also selectively kill NG2-positive uveal melanoma cells and that it has the potential to target human primary uveal melanoma cells that express high levels of NG2. These results support further investigations into the efficacy of this AIC for local and sys- FIGURE 5. TUNEL assay of OCM-8 (A–C) and OCM-1 (D, E) melanoma temic therapy in animal models of uveal melanoma, particu- cells. No TUNEL-positive cells are seen after treatment with (A) non- larly with liver metastases. specific AIC or (B) medium alone. Some TUNEL-positive cells (arrows) are seen in cultures 24 hours after treatment with (C)2␮Ci and (D)10 ␮Ci 213Bi-9.2.27 AIC. (E) No labeling is seen when terminal transferase Acknowledgments enzyme is omitted from the reaction mixture. AO/EB staining of OCM-1 cells is also shown (F–H). After treatment with 213Bi-9.2.27 AIC, The authors thank John Kearsley, Director, St. George Cancer Services, some cells display nuclear chromatin fragmentation and condensation, for ongoing support and David Zhang for assistance with flow cytom- characteristic of apoptosis (arrows, F, H). Viable control cells are etry experiments. OCM-1, OCM-3, and OCM-8 cell lines were kindly shown (G). provided by June Kan-Mitchell (University of California at San Diego, CA); OMM-1 cells were kindly provided by Gregorius P. M. Luyten pericytes have also been observed to be NG2-immunoreactive (Erasmus University, Rotterdam, The Netherlands); and Mel202 cells in developing and adult rat retina39,40 and in diabetic human were kindly provided by Bruce R. Ksander (Schepens Eye Research retina.41 However, in our earlier study, obvious NG2 immuno- Institute, Harvard Medical School, Boston, MA). labeling of perivascular cells was not apparent in human retina, 20 choroid, or primary uveal melanoma specimens. These dif- References ferences may be related to the use of different antibodies, detection systems, or immunolabeling techniques. However, it 1. Karolis C, Frost RB, Billson FA. A thin I-125 seed eye plaque to treat is important to establish whether normal adult human choroi- intraocular tumors using an acrylic insert to precisely position the dal and retinal perivascular cells do express NG2 antigen, sources. Int J Radiat Oncol Biol Phys. 1990;18:1209–1213. particularly if local therapy is to be considered. Furthermore, 2. Shields CL, Shields JA, Gunduz K, Freire JE, Mercado G. Radiation NG2-expressing pericytes and smooth muscle cells may repre- therapy for uveal malignant melanoma. Ophthalmol Surg Lasers. 1998;29:397–409. sent targets for therapy (including 213Bi-9.2.27) in uveal mela- 3. Robertson DM. Changing concepts in the management of choroi- noma, as proposed in a recent study of orthotopic human uveal dal melanoma. Am J Ophthalmol. 2003;136:161–170. melanomas induced in immunosuppressed wild-type and NG2- 4. Damato B, Lecuona K. Conservation of eyes with choroidal mela- knockout mice (Ozerdem U. IOVS 2005;46:ARVO E-Abstract noma by a multimodality approach to treatment: an audit of 1632 4620). Pericytes were observed throughout these uveal mela- patients. Ophthalmology. 2004;111:977–983. nomas, with a significant decrease in tumor microvascular 5. Shields CL, Shields JA. Recent developments in the management of density in NG2-knockout mice; the importance of pericytes choroidal melanoma. Curr Opin Ophthalmol. 2004;15:244–251. and NG2 proteoglycan in tumor vascularization suggests po- 6. Damato B. Developments in the management of uveal melanoma. tential therapeutic targets in controlling tumor growth (Ozer- Clin Exp Ophthalmol. 2004;32:639–647.

Downloaded from jov.arvojournals.org on 09/28/2021 IOVS, December 2005, Vol. 46, No. 12 ␣-Immunotherapy for NG2-Positive Uveal Melanoma 4371

7. Knight LA, Di Nicolantonio F, Whitehouse P, et al. The in vitro 27. Mulford DA, Scheinberg DA, Jurcic JG. The promise of targeted effect of gefitinib (‘Iressa’) alone and in combination with cyto- [alpha]-particle therapy. J Nucl Med. 2005;46(suppl 1):199S–204S. toxic chemotherapy on human solid tumours. BMC Cancer. [serial 28. Couturier O, Supiot S, Degraef-Mougin M, et al. Cancer radioim- online]. 2004;4:83. Available at http://www.biomedcentral.com/ munotherapy with alpha-emitting nuclides. Eur J Nucl Med Mol 1471–2407/4/83. Accessed November 23, 2004. Imaging. 2005;32:601–614. 8. Bedikian AY, Plager C, Papadopoulos N, Eton O, Ellerhorst J, Smith 29. Payne G. Progress in immunoconjugate cancer therapeutics. Can- T. Phase II evaluation of paclitaxel by short intravenous infusion in cer Cell. 2003;3:207–211. metastatic melanoma. Melanoma Res. 2004;14:63–66. 30. van der Velden P, Zuidervaart W, Hurks MHM, et al. Expression 9. Pfohler C, Cree IA, Ugurel S, et al. Treosulfan and gemciabine in profiling reveals that methylation of TIMP3 is involved in uveal metastatic uveal melanoma patients: results of a multicenter feasi- melanoma development. Int J Cancer. 2003;106:472–479. bility study. Anticancer Drugs. 2003;14:337–340. 31. Noter SL, Rothbarth J, Pijl ME, et al. Isolated hepatic perfusion with 10. Kodjikian L, Roy P, Rouberol F, et al. Survival after proton-beam high-dose melphalan for the treatment of uveal melanoma metas- irradiation of uveal melanomas. Am J Ophthalmol. 2004;137: tases confined to the liver. Melanoma Res. 2004;14:67–72. 1002–1010. 32. Lin XH, Dahlin-Huppe K, Stallcup WB. Interaction of the NG2 11. Abbas Rizvi SMA, Sarkar S, Goozee G, Allen BJ. Radioimmunocon- proteoglycan with the actin cytoskeleton. J Cell Biochem. 1996; jugates for targeted alpha therapy of malignant melanoma. Mela- 63:463–477. noma Res. 2000;10:281–289. 33. Goretzki L, Lombardo CR, Stallcup WB. Binding of the NG2 pro- 12. Chang CH, Sharkey RM, Rossi EA, et al. Molecular advances in teoglycan to kringle domains modulates the functional properties pretargeting radioimmunotherapy with bispecific antibodies. Mol of angiostatin and plasmin(ogen). J Biol Chem. 2000;275:28625– Cancer Therapy. 2002;1:553–563. 28633. 13. Allen BJ. Targeted alpha therapy: evidence for potential efficacy of 34. Fang X, Burg MA, Barritt D, Dahlin-Huppe K, Nishiyama A, Stallcup alpha-immunoconjugates in the management of micrometastatic WB. Cytoskeletal reorganization induced by engagement of the cancer. Australas Radiol. 1999;143:480–486. NG2 proteoglycan leads to cell spreading and migration. Mol Biol 14. Allen BJ, Rizvi S, Li Y, Tian Z, Ranson M. In vitro and preclinical Cell. 1999;10:3373–3387. targeted alpha therapy for melanoma, breast, prostate and colo- 35. Eisenmann KM, McCarthy JB, Simpson MA. Melanoma chondroitin rectal cancers. Crit Rev Oncol Hematol. 2001;39:139–146. sulphate proteoglycan regulates cell spreading through Cdc42, 15. McDevitt MR, Barendswaard E, Ma D, et al. An ␣-particle emitting Ack-1 and p130cas. Nat Cell Biol. 1999;1:507–513. antibody ([213Bi]J591) for radioimmunotherapy of prostate cancer. 36. Chekenya M, Pilkington GJ. NG2 precursor cells in neoplasia: Cancer Res. 2000;60:6095–6100. functional, histogenesis and therapeutic implications for malignant 16. Morgan AC Jr, Galloway DR, Reisfeld RA. Production and charac- brain tumors. J Neurocytol. 2002;31:507–521. terization of monoclonal antibody to a melanoma specific glyco- 37. Ozerdem U, Monosov E, Stallcup WB. NG2 proteoglycan expres- protein. Hybridoma. 1981;1:27–36. sion by pericytes in pathological microvasculature. Microvasc Res. 2002;63:129–134. 17. Bumol TF, Reisfeld RA. Free in PMC unique glycoprotein-proteoglycan complex defined by monoclonal antibody on human mel- 38. Murfee WL, Skalak TC, Peirce SM. Differential arterial/venous ex- anoma cells. Proc Natl Acad Sci USA. 1982;79:1245–1249. pression of NG2 proteoglycan in perivascular cells along microvessels: identifying a venule-specific phenotype. Microcircu- 18. Schroff RW, Woodhouse CS, Foon KA, et al. Intratumor localiza- lation. 2005;12:151–160. tion of monoclonal antibody in patients with melanoma treated 39. Hughes S, Chan-Ling T. Characterization of smooth muscle cell and with antibody to a 250,000-dalton melanoma-associated antigen. pericyte differentiation in the rat retina in vivo. Invest Ophthalmol J Natl Cancer Inst. 1985;174:299–306. Vis Sci. 2004;45:2795–2806. 19. Burg MA, Grako KA, Stallcup WB. Expression of the NG2 proteo- 40. Wiley LA, Rupp GR, Steinle JJ. Sympathetic innervation regulates glycan enhances the growth and metastatic properties of mela- basement membrane thickening and pericyte number in rat retina. noma cells. J Cell Physiol. 1998;177:299–312. Invest Ophthalmol Vis Sci. 2005;46:744–748. 20. Li Y, Madigan C, Lai K, et al. Human uveal melanoma expresses 41. Kuiper EJ, Witmer AN, Klaassen I, Oliver N, Goldschmeding R, NG2 immunoreactivity. Br J Ophthalmol. 2003;87:629–632. Schlingemann RO. Differential expression of connective tissue 21. Allen BJ, Rizvi SMA, Tian Z. Preclinical targeted alpha therapy for growth factor in microglia and pericytes in the human diabetic subcutaneous melanoma. Melanoma Res. 2001;11:175–182. retina. Br J Ophthalmol. 2004;88:1082–1087. 22. Boll RA, Mirzadeh S, Kennel SJ. Optimizations of radiolabeling of 42. Li Y, Tian Z, Rizvi SMA, Bander NH, Allen BJ. In vitro and preclin- immunoproteins with 213Bi. Radiochim Acta. 1997;79:145–149. ical targeted alpha therapy of human prostate cancer with Bi-213 23. Gavrieli Y, Sherman Y, Ben-Sasson SA. Identification of pro- labeled J591 antibody against the prostate specific membrane grammed cell death in situ via specific labeling of nuclear DNA antigen. Prostate Cancer Prostatic Dis. 2002;5:36–46. fragmentation. J Cell Biol. 1992;119:493–501. 43. Li Y, Rizvi SMA, Ranson M, Allen BJ. 213Bi-PAI2 conjugate selec- 24. Georges P, Madigan MC, Provis JM. Apoptosis during development tively induces apoptosis in pC3 metastatic cancer cell lines and of the human retina: relationship to foveal development and reti- shows anticancer activity. Br J Cancer. 2002;86:1197–1230. nal synaptogenesis. J Comp Neurol. 1999;413:198–208. 44. Li Y, Cozzi PJ, Qu CF, et al. Cytotoxicity of human prostate cancer 25. Conway RM, Madigan MC, Penfold PL, Billson FA. Induction of cell lines in vitro and induction of apoptosis using 213Bi-Herceptin apoptosis by sodium butyrate in the human Y-79 retinoblastoma ␣-conjugate. Cancer Lett. 2004;205:161–171. cell line. Oncol Res. 1985;7:289–297. 45. Macklis RM, Lin JY, Beresford B, Atcher RW, Hines JJ, Humm JL. 26. Farah RA, Clinchy B, Herrera L, Vitetta ES. The development of Cellular kinetics, dosimetry, and radiobiology of alpha-particle monoclonal antibodies for the therapy of cancer. Crit Rev Euk radioimmunotherapy: induction of apoptosis. Radiation Res. Gene Exp. 1998;8:321–356. 1992;130:220–226.

Downloaded from jov.arvojournals.org on 09/28/2021