IgE--Dependent Immunotherapy of Solid Tumors: Cytotoxic and Phagocytic Mechanisms of Eradication of Ovarian Cancer Cells This information is current as of September 29, 2021. Sophia N. Karagiannis, Marguerite G. Bracher, James Hunt, Natalie McCloskey, Rebecca L. Beavil, Andrew J. Beavil, David J. Fear, Richard G. Thompson, Nicholas East, Frances Burke, Robert J. Moore, David D. Dombrowicz, Frances R. Balkwill and Hannah J. Gould Downloaded from J Immunol 2007; 179:2832-2843; ; doi: 10.4049/jimmunol.179.5.2832 http://www.jimmunol.org/content/179/5/2832 http://www.jimmunol.org/

References This article cites 68 articles, 24 of which you can access for free at: http://www.jimmunol.org/content/179/5/2832.full#ref-list-1

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision by guest on September 29, 2021

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2007 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

IgE-Antibody-Dependent Immunotherapy of Solid Tumors: Cytotoxic and Phagocytic Mechanisms of Eradication of Ovarian Cancer Cells1,2

Sophia N. Karagiannis,3* Marguerite G. Bracher,* James Hunt,* Natalie McCloskey,* Rebecca L. Beavil,* Andrew J. Beavil,* David J. Fear,* Richard G. Thompson,† Nicholas East,† Frances Burke,† Robert J. Moore,† David D. Dombrowicz,‡ Frances R. Balkwill,† and Hannah J. Gould*

Abs have a paramount place in the treatment of certain, mainly lymphoid, malignancies, although tumors of nonhemopoietic origin have proved more refractory ones. We have previously shown that the efficacy of immunotherapy of solid tumors, in Downloaded from particular ovarian carcinoma, may be improved by the use of IgE Abs in place of the conventional IgG. An IgE Ab (MOv18 IgE) against an ovarian-tumor-specific Ag (folate binding protein), in combination with human PBMC, introduced into ovarian cancer xenograft-bearing mice, greatly exceeded the analogous IgG1 in promoting survival. In this study, we analyzed the mechanisms by which MOv18 IgE may exert its antitumor activities. were essential IgE receptor-expressing effector cells that mediated the enhanced survival of tumor-bearing mice by MOv18 IgE and human PBMC. Monocytes mediated MOv18 IgE- dependent ovarian tumor cell killing in vitro by two distinct pathways, cytotoxicity and phagocytosis, acting respectively through http://www.jimmunol.org/ the IgE receptors Fc␧RI and CD23. We also show that human were potent effector cells in MOv18 IgE Ab-dependent ovarian tumor cell cytotoxicity in vitro. These results demonstrate that IgE Abs can engage cell surface IgE receptors and activate effector cells against ovarian tumor cells. Our findings offer a framework for an improved immunotherapeutic strategy for combating solid tumors. The Journal of Immunology, 2007, 179: 2832–2843.

varian cancer ranks second among gynecological can- Campath-1H (alemtuzumab) IgG Ab subclasses in complement- cers in the number of new cases and first in the number dependent immunotherapy of non-Hodgkin’s lymphoma. Several of deaths each year (1, 2). In the early stages, it grows in other mechanisms are now known to operate in Ab immunother-

O by guest on September 29, 2021 and around the ovaries and it metastasizes in the peritoneal cavity apies (3, 9). However, non-IgG isotypes have never been tried in and later to the internal organs. It causes no symptoms in the early cancer patients. IgG Abs are now increasingly being used for the stages, and is essentially incurable in the late stages of the disease; treatment of blood malignancies, but “solid” tumors are often re- around 50–60% of patients die within 5 years (1, 2). Ovarian fractory (3–5). Probable reasons are diffusional barrier of tissues, cancer is intractable to current chemotherapies, but immunother- slow recruitment of effector cells, and low affinity of IgG for its apies using IgG Abs currently in clinical trials are showing some receptors (8–10). promising results (3–6). Abs of the IgE class may offer an alternative to the conventional The current use of the IgG1 Ab in Ab immunotherapy treatments with IgG Abs, particularly for solid tumors such as stems from Waldmann’s classic work (7, 8) on the efficacy of those of the ovary. Unlike Abs of the IgG class, IgE binds to its receptors with very high affinity. The affinity of IgE for its high ␧ ϭ 11 Ϫ1 affinity receptor, Fc RI (Ka 10 M ), is two to five orders of *Randall Division of Cell and Molecular Biophysics, New Hunt’s House, King’s magnitude higher than that of IgG for the Fc␥Rs (Fc␥RI-III) (9– † College London, Guy’s Campus, London, United Kingdom; Centre for Translational 11). Its affinity for the low affinity receptor, CD23 (K ϭ 108 Oncology, Institute of Cancer and the CR-UK Clinical Centre, Barts and The London, a Ϫ1 Queen Mary’s School of Medicine and Dentistry, John Vane Science Centre, Char- M ), is as high as that of IgG for Fc␥RI (11). Furthermore, the terhouse Square, London, United Kingdom; and ‡Institut National de la Sante´etdela peritoneal cavity, where ovarian cancer spreads, is home to IgE Recherche Me´dicale, Institut Fe´de´ratif de Recherche 17, Institut Pasteur de Lille, Unite´547, Lille, France receptor-expressing cells, such as (12, 13), mast cells (14), and dendritic cells (15). These properties may translate to Received for publication February 15, 2007. Accepted for publication June 28, 2007. strong retention of Abs in tissues and longer antitumor immune 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 surveillance. IgE may therefore have greater efficacy than IgG in with 18 U.S.C. Section 1734 solely to indicate this fact. targeting tumors in nonhemopoietic tissues. 1 This study was supported by the Association for International Cancer Re- A number of epidemiological studies on the relation between search, U.K. and the risk of cancer support the idea that IgE might 2 The costs of publication of this article were defrayed in part by the payment of page have advantages over IgG for the treatment of certain types of charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. section 1734 solely to indicate this fact. cancer. Mills et al. (16) conducted a prospective study in 34,198 3 Address correspondence and reprint requests to Dr. Sophia N. Karagiannis, Randall Seventh-Day Adventists. They found that the risk of prostate, Division of Cell and Molecular Biophysics, King’s College London, Room 3.8, New breast cancer, and lymphatic or hematopoetic cancers actually in- Hunt’s House, Guy’s Campus, St Thomas’s Street, London, U.K. E-mail address: creased with , and further increased with an increasing num- [email protected] ber of allergies. However, the risk of ovarian cancer was de- Copyright © 2007 by The American Association of Immunologists, Inc. 0022-1767/07/$2.00 creased with allergy and decreased with an increasing number of www.jimmunol.org The Journal of Immunology 2833

allergies. A recent meta-analysis has demonstrated a significant us to distinguish between two modes of IgE-dependent - inverse association between a history of both asthma and hay fever mediated tumor cell killing, cytotoxic cell killing (ADCC), and and overall cancer mortality and colorectal cancer (17). Recent phagocytosis (ADCP), and establish that Fc␧RI is responsible for studies have also confirmed the increased risk of prostate and ADCC and CD23 for ADCP. breast cancer (18), and shown an increased risk for lung cancer Eosinophils express low levels of Fc␧RI, which mediates IgE- with asthma and a decreased risk of pancreatic cancer and glioma dependent stimulation of IL-10 secretion and defense against par- (19–24). asites (47). Using purified eosinophils from blood, we now show Experimental studies have also supported the concept of har- that they, too, are potent effector cells in MOv18 IgE Ab-depen- nessing IgE for cancer therapy. Nagy et al. (25) demonstrated that dent ovarian tumor cell killing in vitro. a mouse monoclonal IgE, directed against the murine mammary tumor virus, prevented the growth of the tumor in mice. Kershaw Materials and Methods et al. (26) showed that a mouse monoclonal IgE, directed against Abs and reagents a human colon carcinoma Ag, conferred a brief survival advantage to mice implanted with colon tumor cells. We have shown that a Chimeric Abs MOv18 IgE against folate binding protein and 4-hydroxy- 3-nitro-phenacetyl (NIP) IgE specific for the NIP were prepared as chimeric Ab, MOv18 IgE, directed against an ovarian tumor Ag, before (27, 48). We used goat anti-human IgE-FITC Ab (Vector Labora- folate binding protein, in combination with human PBMC, was tories), anti-CD89-PE (BD Biosciences), anti-CD14-PE, isotype control

more active than MOv18 IgG1 in protection of mice from ovarian mAbs, and anti-mouse IgG (Fab)2-FITC (DakoCytomation). Anti-CD23 tumor growth in two xenograft models of ovarian carcinoma in mAbs IDEC-152 and IDEC-152 Fab (Dr. J. Hopp, Biogen Idec, San Diego, CA) recognize the IgE binding site (49). Anti-Fc␧RI mAb 22E7 recognizes scid and nude mice (27, 28). Downloaded from ␧ an unaffected by IgE occupancy (Hoffmann-La Roche) (50). The T cells do not express Fc RI, and hence treatment with IgE Abs soluble Fc␧RI ␣-chain (sFc␧RI␣) was prepared as before (51). Human would not exploit their known cytotoxic functions. Kershaw et al. rIL-4 (1U ϭ 34.5 pg) was obtained from R&D Systems. Propidium iodide (29) expressed the extracellular portion of the high-affinity IgE (PI), CFSE dye, tissue culture medium, and reagents were obtained from receptor, Fc␧RI ␣-chain, joined to the membrane sequence of Invitrogen Life Technologies. mouse Fc␥RII and cytoplasmic sequences of CD28 and the TCR Flow cytometric evaluation of receptor expression and IgE

␨-chain (CD28-␨) in primary human T cells (30). Fc␧RI-CD28-␨ T http://www.jimmunol.org/ binding to monocytes cells, together with an IgE Ab directed against CD8, protected scid mice from the growth of a human thymoma tumor. This strategy Monocytes were incubated with 10 ␮g/ml mAb 22E7 or MHM6, anti- mouse IgG (Fab) –FITC, followed by anti-CD14-PE. To assess IgE bind- should allow any tumor Ag-specific IgE or combination of IgEs to 2 ing, monocytes were given 5 ␮g/ml MOv18 IgE or no Ab for 30 min at be used in adoptive cell immunotherapy of any type of cancer. 4°C, followed by 10 ␮g/ml goat anti-IgE-FITC for 30 min at 4°C. To block Another way of exploiting the inherent advantages of IgE over IgE binding to cell surface receptors, 5 ␮g/ml MOv18 IgE were incubated IgG is based on oral vaccination of mice with tumor Ags under alone or with 62 ␮g/ml sFc␧RI␣ for 30 min at 37°C, followed by addition alkaline conditions, which favors the production of IgE Abs (31, of monocytes and anti-IgE-FITC. Incubations and washing steps were per- formed in FACS buffer (PBS, 5% normal goat serum). 32). Jensen-Jarolim et al. (33) showed that oral vaccination with a by guest on September 29, 2021 breast tumor “Ag” stimulates the production of IgE Abs that ac- Cell purification, stimulation, and culture tivate IgE effector cells and mediate tumor cell lysis in vitro. Reali et al. (34) have shown that passive immunization with IgE Abs can The human ovarian carcinoma IGROV1 cells were grown in RPMI 1640, 10% FCS complete medium at 37°C in 5% CO2 (52). Monocytes from also result in stimulating an active immune response. This reflects human venous blood were isolated to 70–80% purity as described before the sensitization of APCs bearing Fc␧RI, which results in uniquely (28). Cells were incubated overnight at 37°C in AIM-V medium, 5% FCS, strong stimulation. It is notable that no adverse effects of IgE Ab in VueLife (FEP) culture bags (American Fluoroseal Corporation) (53). Ϫ1 treatment in the mouse models of cancer have been observed in Monocytes were incubated for 20 h with 320 U/ml (10 ng ml ) human ␮ ␧ any of the above-mentioned studies. rIL-4 to stimulate CD23 or 2 g/ml MOv18 IgE to stimulate Fc RI. For in vivo experiments, PBMCs were isolated from human venous blood as be- The present study focuses for the first time on the mechanism of fore (27, 28). PBMCs were depleted of monocytes by incubation with IgE-dependent tumor cell killing. For this, we made use of MOv18 anti-CD14-coated immunomagnetic beads to label monocytes, followed by IgE developed against the tumor-associated Ag folate binding pro- the removal of labeled monocytes using a VarioMACS immunomagnetic tein (35, 36), which is overexpressed in 80% of ovarian cancers device (Miltenyi Biotec) according to the manufacturer’s instructions. Eo- sinophils were isolated to Ͼ95% purity by Percoll gradient centrifugation (37, 38). In a previous study, we observed that monocytes infil- (density 1.082 g/ml) (GE Healthcare) followed by immunomagnetic sep- trated human ovarian tumors growing in the nude mice treated aration with anti-CD16-coated immunomagnetic beads as previously de- with MOv18 IgE and human PBMC (28). In the present study, we scribed (47), and used for assays immediately. All work was performed show that monocytes are necessary for the protection of the mice with the approval of the Guy’s Research Ethics Committee and with the by human PBMC. volunteers’ written informed consent. Human monocytes express the two IgE receptors, the high-af- Flow cytometric cytotoxicity/phagocytosis (ADCC/ADCP) assay finity receptor, Fc␧RI, and the low-affinity receptor, CD23 (39– Cell treatment. A three-color flow cytometric assay was used to si- 41). CD23b expression is induced by IL-4 on a wide range of multaneously study tumor cell cytotoxicity (ADCC) and phagocytosis hemopoietic cells, including monocytes (42, 43) and has been (ADCP) of IGROV1 cells by human effector cells as previously de- shown to act in IgE Ab-dependent phagocytosis (ADCP)4 of hap- scribed (46). IGROV1 cells were labeled with 10Ϫ5 mM CFSE for 10 ten-coated red cells (44). The nearest IgG receptor homologue to min at 37°C 1 day before assays. A total of 1.3 ϫ 105 CFSE-labeled ϫ 5 Fc␧RI is Fc␥RIII, which acts in IgG Ab-dependent -, NK IGROV1 cells were mixed with 1.3 10 unstained effectors (E:T ratio ϭ 1:1) and 5 ␮g/ml MOv18, NIP IgE, or no Ab, followed by cell-, and -mediated tumor cell cytotoxicity (Ab-de- incubation for 2.5 h at 37°C. In blocking experiments, 25 ␮g/ml IDEC- pendent cell-mediated cytotoxicity, ADCC) (9, 10, 45). In this 152 Fab were added to monocytes for 30 min at 37°C before assays. In study, we use a novel three-color cytometric assay (46) to enable others, 62 ␮g/ml sFc␧RI␣ were combined with 5 ␮g/ml IgEs or with complete medium alone for 30 min at 37°C, followed by addition of cells. All conditions were tested in triplicate. Cells were then incubated ␮ 4 Abbreviations used in this paper: ADCP, Ab-dependent cell-mediated phagocytosis; with 10 g/ml anti-CD89-PE mAb to label monocytes or anti- ADCC, Ab-dependent cell-mediated cytotoxicity; NIP, 4-hydroxy-3-nitro-phenac- CD49d-PE to label eosinophils for 25 min at 4°C, washed, and treated etyl; sFc␧RI␣, soluble Fc␧RI ␣-chain; PI, propidium iodide; SL, spontaneous loss. with 0.25 ␮g/ml PI for 15 min at 4oC to identify dead cells. Following 2834 MECHANISMS OF IgE ANTIBODY-DEPENDENT TUMOR CELL KILLING

FIGURE 1. Setup of the three-color flow cytometric assay. A three-color flow cytometric assay can simulta- neously measure cytotoxicity and phagocytosis of tu- mor cells by effector cells. Dot plots of mixed monocyte effector and IGROV1 tumor target cells from which cal- culations were made. Region 1 (R1, green) represents total CFSEϩ tumor cell targets. Region 2 (R2, orange) depicts the CFSEϩ tumor cells present within PE- stained monocytes (CFSEϩ/PEϩ), depicting phagocyto- sis. Region 3 (R3, red) contains tumor cells killed ex- ternally by effector cells (cytotoxicity) and are therefore CFSEϩ/PIϩ.

a further wash, cells were mixed thoroughly to interrupt cell-cell con- Student’s t test and significance was accepted at p Ͻ 0.05. Changes in IgE tact, and 20,000 cellular events were acquired by flow cytometry using receptor expression and in tumor cell killing by unstimulated and IgE- a dual laser FACSCalibur flow cytometer (BD Biosciences). stimulated monocytes in vitro were studied by means of the paired two- Assay setup and calculations. Acquisition and measurement of single tailed Student’s t test. cell events were monitored by forward scatter vs side scatter dot plots and Downloaded from compared with control single- and mixed- population samples. CFSE-la- beled IGROV1 were detected in FL1 (530/30 nm band pass filter), PE- Results labeled effectors in FL2 (582/42 nm band pass filter), and PIϩ dead cells in CD23 expression on monocytes and role in IgE-mediated ADCP FL3 (670 nm LP band pass filter) channels. Appropriate controls were set of tumor cells for compensation adjustments between fluorochromes and ADCC and ADCP (46, 54). To calculate ADCC and ADCP (Fig. 1), two dot plots were In previous studies, we used standard cytotoxicity assays to ex- ϩ generated and three regions were identified: 1) R1 (green), total CFSE amine the efficacy of tumor Ag-specific IgE for the immunother- http://www.jimmunol.org/ ϩ ϩ tumor cells ϭ total tumor cells/sample; 2) R2 (orange), CFSE /PE apy of ovarian cancer (27, 28). Standard cytotoxicity assays do not cells ϭ tumor cells phagocytosed by PEϩ effector cells; and 3) R3 (red), ϩ ϩ measure phagocytosis. Thus, they may underestimate tumor cell CFSE /PI cells ϭ intact dead tumor cells. Deviations between samples were accounted for by “R1 Spontaneous Loss (SL) Control” ϭ the average killing and hence the potential of an Ab for immunotherapy of R1 of three control samples (i.e., effector and target cells without mAb). cancer. To analyze the mechanisms by which IgE effector cells Below are calculations to determine the proportion of IGROV1 tumor cells mediate tumor cell killing, we have developed a three-color cyto- killed by ADCC and ADCP: metric assay (Fig. 1) to simultaneously measure cytotoxicity and R1 SL control – R1 ϭ X ADCC ϭ [(X ϩ R3)/R1 SL Control] ϫ 100 phagocytosis of tumor cells by effector cells and tumor Ag-specific ADCP ϭ [R2/R1 SL Control] ϫ 100 IgE (46). Two-color flow cytometric dot plots of CD14ϩ human mono- by guest on September 29, 2021 Immunofluorescence imaging of cells cytes show that Ͻ3% of monocytes cultured without IL-4 stimu- Monocytes were incubated on glass chamber slides (SLS) with IGROV1 lation express CD23 (Fig. 2A, top). After overnight stimulation and mAbs to assess contact between cells and ADCP as previously de- with IL-4, 58% of the cells express CD23. The proportion of scribed (28, 46). Following incubations, monocytes were given anti- monocytes expressing Fc␧RI (31%) remained unchanged after CD89-PE mAb. In similar assays, anti-CD49d-PE mAb was used to label eosinophils. Slides were washed, fixed in 1% paraformaldehyde-FACS IL-4 stimulation (Fig. 2A, middle). Thus, incubation of monocytes buffer, and mounted with fluorescence preserver (DakoCytomation). with IL-4 stimulated expression of CD23, but did not affect ex- Slides were observed using an Axioskop 20 upright microscope (Carl pression of Fc␧RI on the cell surface. With IL-4 stimulation, the Zeiss) equipped with a Zeiss A-Plan 40ϫ/0.65 Ph2 lens, an AxioCam proportion of monocytic cells capable of binding IgE increased 14-bit camera, and AxioVision Version 3.0.2 imaging system (Imaging from 32% in untreated monocytes to 56% in IL-4-treated cells Associates). Light and fluorescent images were superimposed as de- scribed before (28). (Fig. 2A, bottom), suggesting that newly expressed CD23 on the surface of monocytes was capable of binding MOv18 IgE. Experiments in the human ovarian carcinoma xenograft model We measured ADCC and ADCP of IGROV1 cells cultured for The human ovarian carcinoma xenograft HUA was established in 8–12- 2.5 h with human peripheral blood monocytes and either MOv18 wk-old specific pathogen-free female nude mice and implanted i.p. as de- IgE, the hapten-specific anti-NIP IgE, or no Ab as controls (Table scribed before (28, 55). Each mouse received an i.p. injection of primary I; n ϭ 6). Unstimulated monocytes mediated 34.4% ADCC, com- cells with or without Abs 8 h following tumor challenge. The treatments pared with Ͻ15% for the NIP IgE and the no Ab controls (Fig. 2B, were with PBS or 4 ϫ 106 unfractionated PBMC per mouse, 3.2 ϫ 106 PBMC depleted of monocytes using anti-CD14-coated immunomagnetic Table I). No phagocytosis of tumor cells was detected by compar- beads and the MACS system, or 3.2 ϫ 106 PBMC depleted of monocytes ison to controls. Following IL-4 stimulation of the monocytes, and reconstituted with 0.8 ϫ 106 purified monocytes. All cell treatments MOv18 IgE ADCC was 32.5%, compared with Ͻ10% for the were administered with or without 100 ␮g MOv18 IgE in a total volume of controls, whereas MOv18 IgE-mediated ADCP measured at 0.2 ml per mouse. Mice received one further dose of treatments after 14 Ͻ days and were assessed for the duration of survival. All animal work was 22.2%, compared with 15% for the control samples. conducted according to the guidelines specified by the U.K. Home Office Thus, IL-4 stimulation did not affect MOv18 IgE-mediated Animals Scientific Procedures Act 1986. ADCC ( p ϭ 0.77, n ϭ 6), but it significantly enhanced MOv18 IgE ADCP ( p ϭ 0.0007, n ϭ 6). Levels of ADCP in control Data handling and statistical analyses samples without Ab or with NIP IgE were also slightly elevated Flow cytometry experiments of receptor expression and blocking of IgE compared with those for unstimulated monocytes (Table I). This binding were repeated at least three times, and representative experiments suggests that IL-4 may enhance the innate phagocytic capacity of are illustrated. In vitro ADCC/ADCP assays were performed in triplicate, and data are shown as mean ADCC Ϯ SD and ADCP Ϯ SD of a number monocytes. of independent experiments. Statistical analyses of ADCC/ADCP assays Because IL-4 stimulates the expression of CD23, but not Fc␧RI and of mouse survival were performed by means of the unpaired two-tailed (Fig. 2A), the results of the three-color assay imply that ADCP The Journal of Immunology 2835 Downloaded from

FIGURE 2. CD23 up-regulation by IL-4 and role in MOv18 IgE-mediated ADCP of ovarian tumor cells. A, Two-color flow cytometric dot plots show IgE receptor expression and binding of MOv18 IgE in primary monocytes. Primary http://www.jimmunol.org/ monocytes, untreated (left) or incubated with IL-4 (right), were labeled with anti-CD23 mAb IDEC-152 (top) or anti-Fc␧RI mAb 22E7 (mid- dle), followed by anti-mouse IgG (Fab)2–FITC (x-axis) and CD14-PE (y-axis). For binding of MOv18 IgE to untreated (bottom left) and IL-4- stimulated monocytes (bottom right) cells were labeled with MOv18 IgE, anti-IgE FITC (x- axis), followed by CD14-PE (y-axis). B, MOv18

IgE-mediated killing of IGROV1 tumor cells by by guest on September 29, 2021 untreated (left) and IL-4-stimulated (right) monocytes. C, ADCP killing of tumor cells by IL-4-stimulated primary monocytes (left) was blocked by anti-CD23 IDEC-152 mAb Fab (right). Cytotoxicity (ADCC), black bars; phagocytosis (ADCP), gray bars. Results shown are means Ϯ SD of six independent experi- ments. Significance of values compared with samples given MOv18 IgE by the Student’s t ;p Ͻ 0.005 ,ءء ;p Ͻ 0.05 ,ء ;test. n/s, p Ͼ 0.05 .p Ͻ 0.0005 ,ءءء 2836 MECHANISMS OF IgE ANTIBODY-DEPENDENT TUMOR CELL KILLING

Table I. Blocking of IgE-mediated killing of IGROV1 tumor cells by monocytesa

% ADCC Ϯ SD % ADCP Ϯ SD % ADCC Ϯ SD % ADCP Ϯ SD

Untreated Monocytes (n ϭ 6) IL-4-Treated Monocytes (n ϭ 6) No Ab 10.5 Ϯ 2.1*** 8.5 Ϯ 2.3n/s 9.9 Ϯ 1.2** 11.2 Ϯ 6.5* NIP IgE 10.3 Ϯ 0.8*** 8.9 Ϯ 1.5* 8.6 Ϯ 1.0*** 11.6 Ϯ 6.7* MOv18 IgE 34.4 Ϯ 11.2 6.4 Ϯ 1.4 32.5 Ϯ 11.5 22.2 Ϯ 8.0 IL-4-Treated Monocytes (n ϭ 6) IL-4-Treated Monocytes ϩ IDEC-152 Fab (n ϭ 6) No Ab 10.9 Ϯ 3.1** 13.4 Ϯ 3.0** 11.8 Ϯ 3.0** 14.0 Ϯ 4.1n/s NIP IgE 9.4 Ϯ 2.1** 14.2 Ϯ 4.3* 10.5 Ϯ 2.8** 14.7 Ϯ 4.9n/s MOv18 IgE 26.2 Ϯ 8.7 25.8 Ϯ 6.9 26.1 Ϯ 9.3 13.9 Ϯ 4.7 Untreated Monocytes (n ϭ 6) Untreated Monocytes ϩ sFc⑀RI␣ (n ϭ 6) No Ab 8.1 Ϯ 3.0*** 4.6 Ϯ 0.7n/s 9.7 Ϯ 3.1n/s 6.5 Ϯ 0.7n/s NIP IgE 7.7 Ϯ 3.5*** 6.4 Ϯ 1.3n/s 9.8 Ϯ 5.0n/s 7.3 Ϯ 0.5* MOv18 IgE 27.6 Ϯ 6.2 5.4 Ϯ 1.1 13.2 Ϯ 4.7 6.5 Ϯ 0.5 IL-4-Treated Monocytes (n ϭ 6) IL-4-Treated Monocytes ϩ sFc⑀RI␣ (n ϭ 6)

No Ab 7.9 Ϯ 3.1*** 7.4 Ϯ 2.4** 8.7 Ϯ 3.5n/s 6.2 Ϯ 1.3n/s Downloaded from NIP IgE 9.8 Ϯ 4.1** 7.5 Ϯ 2.0** 9.9 Ϯ 2.3n/s 5.5 Ϯ 0.6n/s MOv18 IgE 27.8 Ϯ 8.9 16.5 Ϯ 4.2 12.0 Ϯ 1.6 5.8 Ϯ 1.0

a Monocytes were incubated with or without human rIL-4 to stimulate CD23 expression. IGROV1 tumor cells were incubated with monocyte effector cells and MOv18 IgE, NIP IgE, or no Ab, followed by incubation for 2.5 h at 37°C. To block CD23, IDEC-152 Fab was added to monocytes prior to assays. To block binding of IgE to receptors, ;p Ͻ 0.005 ,ءء ;p Ͻ 0.05 ,ء ;sFc⑀RI␣ was combined with IgE before addition of cells. Significance compared to samples given MOv18 IgE by the Student’s t test: n/s, p Ͼ 0.05 .p Ͻ 0.0005 ,ءءء http://www.jimmunol.org/

may be attributed to CD23. To test this possibility, we preincu- compared with Ͻ15% with the controls. Blocking of IgE binding bated monocytes with 20-fold molar excess of the IDEC-152 to CD23 with IDEC-152 Fab resulted in MOv18 IgE ADCC mea- CD23-blocking Ab Fab (Fig. 2C; n ϭ 6). Following IL-4 stimu- sured at 26.1%, compared with Ͻ15% for the controls. Monocytes lation of monocytes, MOv18 IgE-induced ADCC was 26.2%, treated with the CD23 blocking Ab Fab, however, showed low compared with Ͻ15% for the controls. As before (Fig. 2B), with MOv18 IgE ADCP (13.9%), a value similar to that measured with IL-4 stimulation, MOv18 IgE-induced ADCP measured at 25.8%, the controls. by guest on September 29, 2021

FIGURE 3. Fc␧RI is responsible for monocyte ADCC by tumor Ag-specific IgE. A, Two-color flow cytometric dot plots show Fc␧RI expression in primary monocytes. Monocytes were incubated in medium alone (left) or stimulated with MOv18 IgE for 20 h (middle). Cells were labeled with anti-Fc␧RI ␧ mAb 22E7, anti-mouse IgG (Fab)2–FITC (x-axis), followed by anti-CD14-PE (y-axis). The proportion of monocytes showing cell surface Fc RI expression increased from 31.0 Ϯ 9.6% to 37.8 Ϯ 10.1% following overnight stimulation with IgE (n ϭ 8). B, Increased Fc␧RI cell surface expression correlated with enhanced ADCC of IGROV1 tumor cells by primary monocytes and MOv18 IgE. Primary monocytes were incubated overnight in medium alone or stimulated with medium containing MOv18 IgE, followed by ADCC/ADCP tumor killing assays. Monocytes from eight donors were assessed, and each set of ADCC measurements represents mean values of triplicate calculations. Tumor cell killing by IgE-stimulated monocytes and MOv18 IgE increased to 30.7 Ϯ 8.1% compared with 24.4 Ϯ 7.7% with unstimulated monocytes and MOv18 IgE (n ϭ 8). Phagocytosis was not significantly influenced by IgE stimulation (ADCP: 8.1 Ϯ 3.2% with and 6.4 Ϯ 1.4% without IgE, n ϭ 8, p Ͼ 0.05). Significance of changes in IgE receptor expression and in tumor cell .p Ͻ 0.0005 ,ءءء ;p Ͻ 0.005 ,ءء :killing by the paired two-tailed Student’s t test The Journal of Immunology 2837

FIGURE 4. sFc␧RI␣ blocked bind- ing of IgE to monocytes and IgE-medi- ated tumor killing. A, Flow cytometric histograms show binding of MOv18 IgE to monocytes. IL-4-stimulated monocytes bound fresh MOv18 IgE via both Fc␧RI and CD23 (left). Preincuba- tion of MOv18 IgE with sFc␧RI␣ pre- vented binding of fresh IgE on the monocyte surface (right). (MOv18 IgE, anti-IgE-FITC, black lines; anti- IgE-FITC, gray lines). B, MOv18 IgE mediated killing of IGROV1 tumor cells by unstimulated monocytes (left). sFc␧RI␣ blocked MOv18 IgE-medi- ated IGROV1 tumor killing by un- stimulated monocytes (right). C, Downloaded from MOv18 IgE-mediated ADCC and ADCP by IL-4-stimulated monocytes (left). Both ADCC and ADCP were blocked by incubation of MOv18 IgE with soluble Fc␧RI␣ (right). Cytotox- icity (ADCC), black bars; phagocyto- http://www.jimmunol.org/ sis (ADCP), gray bars. Results are means Ϯ SD of six independent mea- surements. Significance of values compared with samples given MOv18 IgE by the Student’s t test: p Ͻ ,ءء ;p Ͻ 0.05 ,ء ;n/s, p Ͼ 0.05 .p Ͻ 0.0005 ,ءءء ;0.005 by guest on September 29, 2021

Whereas MOv18 IgE ADCC remained unchanged at 26% (Fig. 3A, representative dot plots). Without stimulation, 31.0% with or without CD23 blocking ( p ϭ 0.97), MOv18 IgE ADCP of CD14ϩ cells expressed Fc␧RI and this proportion increased was reduced from 25.8% to 13.9% with blocking of CD23 ( p ϭ to 37.8% of the cells in IgE-stimulated monocytes (Fig. 3A, 0.006). Thus, CD23 mediated the IgE ADCP of ovarian tumor right; n ϭ 8). Increased expression of Fc␧RI on monocytes cells effected by monocytes. The observed levels of expression stimulated with MOv18 IgE in culture corresponded to an of Fc␧RI and CD23 in unstimulated and IL-4-stimulated mono- equivalently modest increase in MOv18 IgE-induced ADCC cytes (Fig. 2A), together with the activity of anti-CD23, suggest from 24.4% in unstimulated monocytes to 30.7% in IgE-stim- that Fc␧RI mediates the constitutive ADCC and CD23 mediates ulated monocytes (Fig. 3B; n ϭ 8). ADCP was measured at the superimposed ADCP in IL-4-stimulated monocytes (Fig. 2, background levels in unstimulated and IgE-stimulated mono- B and C). cytes. These results support a role for Fc␧RI in IgE- mediated ADCC. ␧ Up-regulation of Fc RI on monocytes and function in IgE IgE bound to IL-4-stimulated monocytes was assayed using goat ADCC of tumor cells anti-IgE-FITC. We detected endogenous IgE bound to 25% of the We established that monocyte-mediated ADCP is due to the action monocytes and, after addition of MOv18 IgE, 58% of the mono- of CD23 by correlating the increase in ADCP with the up-regula- cytes bound IgE (Fig. 4A). When 100-fold molar excess of tion of CD23 by IL-4, and by the inhibitory effect of the anti- sFc␧RI␣ was added to MOv18 IgE before incubation with the CD23-blocking Ab. Similarly, we have been able to attribute monocytes, the proportion of IgEϩ monocytes did not increase ADCC to the action of Fc␧RI. We up-regulated Fc␧RI on mono- with the addition of MOv18 IgE (Fig. 4A). This demonstrates that cytes by incubation with MOv18 IgE, and examined the effects of sFc␧RI␣ prevented binding of MOv18 IgE on the surface of mono- increased expression on tumor cell killing by IgE. We also tested cytes. The binding sites of sFc␧RI and CD23 on IgE are compet- the effects of a soluble fragment of the IgE receptor ␣-chain itive (11, 56, 57), so that sFc␧RI␣ in 100-fold molar excess must (sFc␧RI␣) on ADCC and ADCP. prevent the association of IgE to both receptors. Primary monocytes were incubated for 20 h with or without When a 100-fold molar excess of sFc␧RI␣ was mixed with MOv18 IgE, and then labeled with CD14-PE and the 22E7 anti- MOv18 IgE before incubation with monocytes, MOv18 IgE Fc␧RI Ab to determine the level of receptor expression by flow ADCC dropped from 27.6 to 13.2% ( p ϭ 0.001, n ϭ 6) (Fig. 4B, cytometry. Preincubation with MOv18 IgE led to a modest in- Table I). MOv18 IgE-mediated ADCP was low with (6.5%) and crease, in the proportion of Fc␧RIϩ-expressing CD14ϩ cells without (6.2%) treatment with sFc␧RI␣ ( p ϭ 0.052, n ϭ 6). This 2838 MECHANISMS OF IgE ANTIBODY-DEPENDENT TUMOR CELL KILLING Downloaded from http://www.jimmunol.org/ by guest on September 29, 2021

FIGURE 5. Superimposed bright-field and fluorescent images of IGROV1-monocyte interactions. CFSE-stained IGROV1 tumor cells (green) and CD89-PE-labeled monocytes (red) were incubated with IgE Abs for 2.5 h. A, After incubation with MOv18 IgE, unstimulated monocytes were in frequent contact with IGROV1 cells (left). Rare monocyte-IGROV1 cell contact was seen in samples given control NIP IgE (middle) and when MOv18 IgE was treated with sFc␧RI␣ before assays (right). B, IL-4-stimulated monocytes show enhanced contact with IGROV1 tumor cells (green) and phagocytosis of tumor cells (green fluorescent CFSE inside monocytes; orange; arrow) after incubation with MOv18 IgE. Little E:T cell contact and phagocytosis are seen with control Ab NIP IgE, or when MOv18 IgE was preincubated with sFc␧RI␣. IL-4-treated monocytes given anti-CD23 IDEC-152 blocking mAb Fab before MOv18 IgE exhibit reduced effector-tumor cell contact and lack of phagocytosis (absence of CFSE dye inside monocytes; red). Original magni- fication, ϫ400. Scale bars, 20 ␮m.

proves that ADCC depends on MOv18 IgE binding to Fc␧RI on monocytes and IGROV1 cells (Fig. 5A), not seen with control Ab monocytes. In the case of IL-4-treated monocytes, sFc␧RI␣ led to NIP IgE, or when the MOv18 IgE is preincubated with 100-fold decreases in both MOv18 IgE ADCC (from 27.8% to 12.0%; p ϭ molar excess of sFc␧RI␣ (Fig. 5A). When the monocytes were 0.002, n ϭ 6) and MOv18 IgE ADCP (from 16.5% to 5.8%; p ϭ stimulated by IL-4, contact between the tumor and effector cells 0.0001, n ϭ 6), approaching the values observed in controls (Fig. was enhanced, and phagocytosis was clearly visible in the 4C). This was expected because, as mentioned above, sFc␧RI␣ merged image of the green tumor cells inside the red mono- blocks the binding of IgE to both receptors (11, 56, 57). We con- cytes, resulting in the yellow color (Fig. 5B; arrow); this is not clude that the binding of MOv18 IgE to both IgE receptors on seen with NIP IgE, or when IgE binding to its receptors is monocytes contributed to IgE-mediated tumor cell targeting and blocked by sFc␧RI␣. Addition of the IDEC-152 anti-CD23 killing. blocking Ab Fab to the monocytes cultured with IL-4 substan- tially reduced, but did not eliminate, the contact between mono- Visualization of contact between tumor cells and monocytes cytes and IGROV1 cells. IDEC-152 Fab inhibited phagocytosis, To visualize the results, shown quantitatively in the flow cytomet- seen by the absence of yellow color inside the monocytes (Fig. ric assays described above, we labeled IGROV1 with CFSE before 5B). In contrast, incubation with sFc␧RI␣ completely inhibited incubation with IgE Abs for 2.5 h and then stained monocytes with contact between monocytes and IGROV1 cells, consistent with anti-CD89-PE. Cell interactions were the observed by fluorescence IgE binding to both Fc␧RI and CD23 and the inhibition of cy- microscopy. Incubation with MOv18 IgE led to contact between totoxicity as well as phagocytosis (Fig. 4, B and C). The Journal of Immunology 2839

FIGURE 6. Effect of monocytes and MOv18 IgE on survival of human ovarian carcinoma xenograft-bearing mice. Nude mice were transplanted with the HUA hu- man ovarian cancinoma xenografts and given various treatments with primary human cells, with or without MOv18 IgE. A repeat treatment dose was given 2 wk following tumor challenge. Dot, Survival following tu- mor challenge (days). Horizontal bars, Mean survival for each population (days). Significance of survival Downloaded from p Ͻ ,ء ;compared by the Student’s t test. n/s, p Ͼ 0.05 .p Ͻ 0.0005 ,ءءء ;p Ͻ 0.005 ,ءء ;0.05 http://www.jimmunol.org/

Monocytes are necessary for IgE-mediated survival of mice cytes from the PBMC population thus abolished the survival advan- by guest on September 29, 2021 bearing ovarian tumor xenografts tage of mice given PBMC and MOv18 IgE, implying that monocytes We tested the ability of monocytes to promote survival in nude play a crucial role in targeting IgE against tumor cells and promoting mice bearing the human ovarian carcinoma xenograft HUA, which mouse survival. expresses the folic acid receptor at moderate levels (28). HUA The role of monocytes in delaying tumor growth and prolonging cells were derived from a patient with ovarian cancer; they have mouse survival was further confirmed with reconstitution of mono- not yet been adapted to tissue culture and their characteristics have cyte-depleted PBMCs with purified monocytes. Adding back pu- not changed in years of passage in mice. The tumors grow and rified monocytes to depleted PBMCs at proportions equivalent to spread in the peritoneal cavity of the mice, as in ovarian cancer those in unfractionated PBMCs restored the ability of PBMCs and patients, and their histological and profiles are similar to MOv18 IgE to increase mouse survival (44 days) to levels statis- the original ascites. This is, therefore, considered a more clinically tically equivalent to those seen in mice given whole PBMCs and relevant model than the ovarian carcinoma cell line IGROV1 used MOv18 IgE (Fig. 6, Table II). This survival was significantly in this study in our in vitro assays and in our previous in vitro and longer than monocyte-reconstituted PBMCs alone (25 days) or de- in vivo studies (27, 28, 55). HUA ascites were introduced into pleted PBMCs with and without MOv18 IgE. Purified monocytes nude mice by i.p. injection. Unlike human IgG1, the Fc fragment of the human IgE (Fc␧) is not recognized by the murine Fc␧Rs Table II. Monocytes are crucial effector cells in IgE-mediated survival a (10). Thus, to study the effect of human IgE and the role of IgE of ovarian carcinoma xenograft-bearing mice receptor-bearing effector cells in targeting a human ovarian carci- Ϯ noma xenograft, we introduced human mononuclear cells from Treatment Survival SD peripheral blood. Mice were treated with human effector cells, with PBS (n ϭ 21) 15.8 Ϯ 2.7*** or without MOv18 IgE, and the length of survival was assessed PBMC (n ϭ 21) 20.6 Ϯ 5.1** (Fig. 6). PBMC ϩ MOv18 IgE (n ϭ 21) 38.7 Ϯ 23.9 ϭ Ϯ Mice treated with PBMCs plus MOv18 IgE survived for 39 days Monocyte-depleted PBMC (n 13) 18.4 3.0** Monocyte-depleted PBMC ϩ MOv18 IgE 20.5 Ϯ 5.8* and they had a significant survival advantage compared with mice that (n ϭ 13) received PBMC alone (21 days) or buffer controls (16 days) (Table Monocyte-depleted PBMC ϩ Monocytes 24.9 Ϯ 9.4* II). As observed in our previous studies (27, 28), mice given PBMC (n ϭ 13) ϩ ϩ Ϯ n/s alone survived longer than buffer controls. Mice given PBMCs de- Monocyte-depleted PBMC Monocytes 44.4 16.4 MOv18 IgE (n ϭ 13) pleted of monocytes plus MOv18 IgE survived for 21 days, while those treated with PBMCs depleted of monocytes without IgE sur- a Results are mean Ϯ SD. Numbers in brackets denote the number of nude mice examined in each group. Significance levels for comparison to mice treated with p Ͻ ,ءء ;p Ͻ 0.05 ,ء ;vived for 18 days and their survival was shorter than that of mice PBMC and MOv18 IgE by the Student’s t-test: n/s, p Ͼ 0.05 .p Ͻ 0.0005 ,ءءء ;given unfractionated PBMC with MOv18 IgE. Removal of mono- 0.005 2840 MECHANISMS OF IgE ANTIBODY-DEPENDENT TUMOR CELL KILLING Downloaded from http://www.jimmunol.org/ FIGURE 7. Assessments of MOv18 IgE-mediated tumor cell interaction and killing by human eosinophils. A, MOv18 IgE mediated IGROV1 killing of tumor cells by primary eosinophils. Cytotoxicity (ADCC), black bars; phagocytosis (ADCP), gray bars. Results are means Ϯ SD of four independent .p Ͻ 0.0005 ,ءءء ;p Ͻ 0.005 ,ءء ;p Ͻ 0.05 ,ء ;experiments. Significance was compared with samples given MOv18 IgE by the Student’s t test. n/s, p Ͼ 0.05 B, Superimposed bright-field and fluorescent images of IGROV1- interactions. CFSE-stained IGROV1 tumor cells (green) and CD49d-PE- labeled eosinophils (red) were incubated with IgE Abs for 2.5 h. After incubation with MOv18 IgE, eosinophils were in frequent contact with IGROV1 cells and IGROV1 tumor cell destruction was evident (left). Eosinophil-IGROV1 cell contact was observed less frequently in samples given control NIP IgE (middle), or no Ab (right). Original magnification, ϫ400. Scale bars, 20 ␮m.

were also tested in this model, but they did not confer any survival with IgE Abs for 2.5 h, and then stained eosinophils with anti- by guest on September 29, 2021 advantage to tumor-bearing mice (results not shown). This may be CD49d-PE. Cells were then visualized to examine E:T cell inter- due to the requirement of survival factors or trafficking signals actions by fluorescence microscopy (Fig. 7B). Microscopic obser- conferred by other cells in the PBMC preparation in these models. vations showed that incubation with MOv18 IgE led to contact between eosinophils (red) and IGROV1 tumor cells (green), fre- Eosinophils mediate IgE-dependent ADCC quently accompanied by eosinophil degranulation, loss of tumor Eosinophils were lost in the purification of PBMCs from blood. To cell architecture, and apparent tumor cell death. Less frequent con- test their ability to direct IgE Ab-dependent tumor cell killing, we tact of tumor cells with eosinophils was seen with NIP IgE or in purified eosinophils and used them as effector cells in the ADCC/ the absence of Ab. In the latter specimens, eosinophil degranula- ADCP assay. We measured the ADCC and ADCP of IGROV1 tion was less common and tumor cells appeared more brightly cells incubated with human eosinophils and MOv18 IgE, NIP IgE fluorescent and viable. As previously reported, we found eosino- or no Ab (Fig. 7A; n ϭ 4). Eosinophils mediated elevated ADCC phils to express low levels of the Fc␧RI receptor but not CD23 (32.4%) with MOv18 IgE, compared with controls. No phago- (results not shown) (47). These results support a role for Fc␧RI in cytosis of tumor cells was detected by comparison to controls IgE ADCC of tumor cells by eosinophils as well as monocytes. (Table III). We examined the interaction between eosinophils and tumor Discussion cells in the presence or absence of MOv18 IgE by fluorescence Following our previous demonstration that MOv18 IgE is superior microscopy. We labeled IGROV1 with CFSE before incubation to MOv18 IgG1, together with human PBMC, in prolonging the survival of mice in our xenograft models of ovarian cancer (27, 28), we sought to identify the IgE effector cells and the IgE re- Table III. IgE-mediated killing of IGROV1 tumor cells by eosinophilsa ceptors and mechanisms involved in tumor cell killing. The ob- servation of monocytes infiltrating the xenografts in the mice (28) Fresh Eosinophils (n ϭ 4) directed our attention to monocytes. We also examined the activity of eosinophils, which are potential IgE effector cells not present % ADCC Ϯ SD % ADCP Ϯ SD in PBMC. No Ab 14.6 Ϯ 5.4* 6.6 Ϯ 3.7n/s In earlier work, we observed that Fc␧RI, but not CD23, was Ϯ Ϯ n/s NIP IgE 16.7 5.2* 6.8 3.9 constitutively expressed on primary monocytes, implicating the Mov18 IgE 32.4 Ϯ 8.6 5.9 Ϯ 3.1 former receptor in IGROV1 cell killing (28). In this study, we a Human eosinophils freshly isolated from peripheral blood were incubated confirmed this finding and found that Fc␧RI-expressing primary with IGROV1 tumor cells and MOv18 IgE, NIP IgE or no Ab for 2.5 h at 37°C. ␧ Significance compared to samples given MOv18 IgE by the Student’s t test: n/s, monocytes exerted primarily ADCC. Upon up-regulation of Fc RI p Ͻ 0.05. by preincubation of the monocytes with MOv18 IgE, we observed ,ء ;p Ͼ 0.05 The Journal of Immunology 2841 an increase in ADCC, further suggesting that Fc␧RI mediates cy- IgE-Fc␧RI complexes on eosinophils are known to exert ADCC totoxic cell killing. ADCC and the contact between the monocytes against parasites (61, 62). In this study, we show for the first time and tumor cells, observed by fluorescence microscopy, were in- that a tumor Ag-specific Ab can direct eosinophils to kill tumor hibited by sFc␧RI␣, confirming this mechanism of Fc␧RI action. cells by ADCC. As observed with monocytes, and consistent with In contrast, up-regulation of CD23 by IL-4, led to MOv18 IgE the expression of low levels of Fc␧RI on the surface of eosinophils ADCP without affecting the level of ADCC. Moreover, incubation (47), this IgE receptor on eosinophils appears to mediate MOv18- with the IDEC-152 anti-CD23 Ab Fab inhibited ADCP, but not dependent cytotoxicity of tumor cells. Engagement of IgE Abs on ADCC. Therefore, we were able to attribute IgE-mediated cyto- the surface of eosinophils triggers the release of inflammatory and toxicity of tumor cells to Fc␧RI and IgE-mediated phagocytosis cytotoxic mediators, such as eosinophil peroxidase and major basic to CD23. protein (63), which may contribute to tumor cell death. Our data, Our tumor cell killing assays and microscopic observations therefore, raise the possibility that eosinophils, either resident in suggest that contact of tumor target and monocytes is necessary intratumoral areas or recruited from the circulation, could contrib- for IgE-mediated ADCC and ADCP. Addition of tumor Ag- ute to the elimination of tumor cells in vivo. Indeed, eosinophil specific IgE was needed for tumor cell death, as the hapten- infiltrates are associated with host inflammatory responses against specific NIP IgE did not mediate tumor killing. This implies the a number of cancers (62, 64). requirement for tumor cell recognition by a tumor Ag-specific Ab. The human operates with nine Ab classes, but Blocking of MOv18 IgE binding to receptors on monocytes by cancer immunotherapy has been attempted only with Abs of the preincubation of IgE with sFc␧RI␣ drastically reduced MOv18- IgG class. It seems unlikely that this would afford optimal efficacy IgE-mediated ADCC and ADCP. Therefore, engagement of IgE for all types of tumors, in every anatomical location and in every Downloaded from receptors by IgE is also required. Finally, microscopic observa- cancer patient. Other workers have shown IgA ADCC and ADCP tions show a correlation between E:T cell contact and the presence of breast tumor cells (54, 65), melanoma cells (66), and non- of MOv18 IgE. Therefore, bridging tumor cells and IgE receptors Hodgkins lymphoma (67) by monocyte-derived macrophages (65) on monocytes by IgE is necessary for both mechanisms of IgE- or (54, 66, 67). IgA Abs were directed to tumor cells ␣ mediated tumor killing, ADCC and ADCP, to occur. in the form of bispecific Abs against the IgA receptor, Fc RI, or Our observations in human ovarian carcinoma xenograft-bear- CD89, on the effector cells and the specific tumor Ag on the http://www.jimmunol.org/ tumor cells. Bispecific Abs were used to compensate for the low ing mice clearly suggest a pivotal role for monocytes in enhancing affinity of IgA for Fc␣RI (ϳ106 MϪ1), which precludes the mouse survival effected by MOv18 IgE. In this in vivo system, stable interaction of Ab with the effector cells. Unlike either monocytes in PBMC, together with MOv18 IgE, are required to IgG or IgA, IgE binds to Fc␧RI with sufficiently high affinity enhance mouse survival, compared with controls. This implies that (ϳ1011 MϪ1) to be effective as a monospecific Ab. Consistent IgE-receptor-bearing cells like monocytes can act, not only in with the stability of the IgE-Fc␧RI interaction, circulating vitro, but also in vivo as effector cells. Monocytes-macrophages monocytes and eosinophils are able to transport IgE into tissues are recruited in the intratumoral environment by tumor-derived where this IgE may engage effector cells in tumor surveillance chemoattractant signals. Tumor-associated macrophages can be by guest on September 29, 2021 and killing, as we have shown (11, 13, 59, 61, 62). This prop- activated to promote tumor growth in situ and fail to mount an erty is also likely to enhance Ag presentation by Fc␧RI APCs antitumor response (13, 53, 58). Resident macrophages express (e.g., dendritic cells) leading to active . IgE receptors (59) and are capable of mediating NO release (60). There is an increasing call for rational therapies for the treat- Based on our observations on the IgE-mediated antitumor effects ment of cancer, requiring knowledge of mechanisms of drug action of monocytes in vitro as well as in vivo, we propose that these cells (68). There are no previous reports that analyze the mechanisms of may be activated and reprogrammed by a tumor Ag-specific IgE to IgE-dependent tumor cell killing and the role of IgE receptor-ex- kill and phagocytose tumor cells, rather than promote their growth. pressing effector cells. Our findings offer a framework for an im- The PBMC used in our xenograft model contain around 2% proved immunotherapeutic strategy for combating solid tumors. human (28), which are known to express Fc␧RI. We have shown that these cells can be activated to secrete histamine in Acknowledgments the presence of PBMC, tumor cells, and MOv18 IgE (27). It is This study was supported by the Association for International Cancer Re- likely that IgE-dependent mast cell and activation by tu- search. We thank Kate Kirwan for expert assistance with the figures. We mor cell Ags would enhance the recruitment of inflammatory cells are grateful to Dr. T. Hagemann for technical help and practical advice for to the site of a tumor, but this remains to be tested. the growth of primary monocytes and to Dr. Silvana Canevari, Dr. Marie- The PBMC used as effector cells lack other IgE receptor-ex- He´le`ne Jouvin, and Dr. J. Hopp, for the generous provision of materials and pressing cells, such as eosinophils, tissue mast cells, and macro- advice. phages that might mediate strong antitumor responses. The ab- sence of these potent human effector cells in this system may have Disclosures contributed to the observed limited efficacy that MOv18 IgE con- The authors have no financial conflict of interest. fers in this system. We have previously observed a lack of en- hanced mouse survival using tumor-specific IgE to treat estab- References lished tumors; these observations could also be attributed to lack of 1. Bray, F., A. H. Loos, S. Tognazzo, and C. La Vecchia. 2005. Ovarian cancer in potent human effector cells. Human PBMC have a limited lifespan Europe: cross-sectional trends in incidence and mortality in 28 countries, 1953– 2000. Int. J. Cancer 113: 977–990. in the mouse and their antitumor effects may therefore be limited 2. Quirk, J. T., and N. Natarajan. 2005. Ovarian cancer incidence in the United to a short time following treatments. released by human States, 1992–1999. Gynecol. Oncol. 97: 519–523. effector cells may not act on murine effector cells and, thus, the 3. Glennie, M. J., and J. G. J. van de Winkel. 2003. Renaissance of cancer thera- peutic . Drug Discov. Today 8: 503–510. xenograft model may not reveal indirect mechanisms of tumor cell 4. Luborsky, J. L., A. Barua, S. V. Shatavi, T. Kebede, J. Abramowicz, and killing. Our results may point toward a beneficial role for IgE J. Rotmensch. 2005. Anti-tumor antibodies in ovarian cancer. Am. J. Reprod. Immunol. 54: 55–62. therapy following surgical intervention on ovarian cancer patients 5. Nicodemus, C. F., and J. S. Berek. 2005. therapy of ovarian with smaller tumor burden or with minimal residual disease. cancer. Expert Rev. Anticancer Ther. 5: 87–96. 2842 MECHANISMS OF IgE ANTIBODY-DEPENDENT TUMOR CELL KILLING

6. Weiner, L. M., and H. Borghaei. 2006. Targeted therapies in solid tumors: mono- tumor-specific IgE antibodies by oral mimotope vaccination. Cancer Res. 67: clonal antibodies and small molecules. Hum. Antibodies 15: 103–111. 3406–3411. 7. Bru¨ggemann, M., G. T. Williams, C. I. Bindon, M. R. Clark, M. R. Walker, 34. Reali, E., J. W. Greiner, A. Corti, H. J. Gould, F. Bottazzoli, G. Paganelli, R. Jefferis, H. Waldmann, and M. S. Neuberger. 1987. Comparison of the effector J. Schlom, and A. Siccardi. 2001. IgEs targeted on tumor cells: therapeutic ac- functions of human immunoglobulins using a matched set of chimeric antibodies. tivity and potential in the design of tumour . Cancer Res. 61: 5517–5522. J. Exp. Med. 166: 1351–1361. 35. Campbell, I. G., T. A. Jones, W. D. Foulkes, and J. Trowsdale. 1991. Folate 8. Dyer, M. J. S., G. Hale, F. G. J. Hayhoe, and H. Waldmann. 1989. Effects in vivo binding protein is a marker for ovarian cancer. Cancer Res. 51: 5329–5338. in patients with lymphoid malignancies; influence of antibody isotype. Blood 73: 36. Coney, L. R., A. Tomassetti, L. Carayannopoulos, V. Frasca, B. A. Kamen, 1431–1439. M. I. Colnaghi, and V. R. Zurawski, Jr. 1991. Cloning of a tumor-associated 9. Clynes, R. A., T. L. Towers, L. G. Presta, and J. V. Ravetch. 2000. Inhibitory Fc : MOv18 and MOv19 antibodies recognize a folate-binding protein. Can- receptors modulate in vivo cytoxicity against tumor targets. Nat. Med. 6: cer Res. 51: 6125–6132. 443–446. 37. Coney, L. R., D. Mezzanzanica, D. Sanborn, P. Casalini, M. I. Colnaghi, and 10. Ravetch, J. V., and J. P. Kinet. 1991. Fc receptors. Annu. Rev. Immunol. 9: V. R. Zurawski, Jr. 1994. Chimeric murine-human antibodies directed against 457–492. folate binding receptor are efficient mediators of ovarian carcinoma cell killing. 11. Gould, H. J., B. J. Sutton, A. J. Beavil, R. L. Beavil, N. McCloskey, H. A. Coker, Cancer Res. 54: 2448–2455. D. Fear, and L. Smurthwaite. 2003. The biology of IgE and the basis of allergic 38. Toffoli, G., C. Cernigoi, A. Russo, A. Gallo, M. Bagnoli, and M. Boiocchi. 1997. disease. Annu. Rev. Immunol. 21: 579–628. Overexpression of folate binding protein in ovarian cancers. Int. J. Cancer 74: 12. Schoppmann, S. F., P. Birner, J. Stockl, R. Kalt, R. Ullrich, C. Caucig, 193–198. E. Kriehuber, K. Nagy, K. Alitalo, and D. Kerjaschki. 2002. Tumor-associated 39. Williams, J., S. Johnson, J. J. Mascali, H. Smith, L. J. Rosenwasser, and macrophages express lymphatic endothelial growth factors and are related to L. Borish. 1992. Regulation of low affinity IgE receptor (CD23) expression on peritumoral lymphangiogenesis. Am. J. Pathol. 161: 947–956. mononuclear in normal and asthmatic subjects. J. Immunol. 149: 13. Gordon, I. O., and R. S. Freedman. 2006. Defective antitumor function of mono- 2823–2829. cyte-derived macrophages from epithelial ovarian cancer patients. Clin. Cancer 40. Wong, H. L., M. T. Lotze, L. M. Wahl, and S. M. Wahl. 1992. Administration of Res. 12: 1515–1524. recombinant IL-4 to humans regulates gene expression, phenotype, and function in circulating monocytes. J. Immunol. 148: 2118–2125. 14. Chan, J. K., A. Magistris, V. Loizzi, F. Lin, J. Rutgers, K. Osann, P. J. DiSaia, Downloaded from and M. Samoszuk. 2005. Mast cell density, angiogenesis, blood clotting, and 41. Boltz-Nitulescu, G., M. Willheim, A. Spittler, F. Leutmezer, C. Tempfer, and prognosis in women with advanced ovarian cancer. Gynecol. Oncol. 99: 20–25. S. Winkler. 1995. Modulation of IgA, IgE, and IgG Fc receptor expression on ␣ 15. Wei, S., I. Kryczek, L. Zou, B. Daniel, P. Cheng, P. Mottram, T. Curiel, human mononuclear phagocytes by 1 ,25-dihydroxyvitamin D3 and cytokines. A. Lange, and W. Zou. 2005. Plasmacytoid dendritic cells induce CD8ϩ regu- J. Leukocyte Biol. 58: 256–262. latory T cells in human ovarian carcinoma. Cancer Res. 65: 5020–5026. 42. Mossalayi, M. D., N. Paul-Eugene, F. Ouaaz, M. Arock, J. P. Kolb, E. Kilchherr, ␧ 16. Mills, P. K., W. L. Beeson, G. E. Fraser, and R. L. Phillips. 1992. Allergy and P. Debre, and B. Dugas. 1994. Involvement of Fc RII/CD23 and L-arginine- cancer: organ site-specific results from the Adventist Health Study. dependent pathway in IgE-mediated stimulation of human monocyte functions. Am. J. Epidemiol. 136: 287–295. Int. Immunol. 6: 931–938. 43. Paul-Eugene, N., D. Mossalayi, M. Sarfati, K. Yamaoka, J. P. Aubry, http://www.jimmunol.org/ 17. Turner, M. C., Y. Chen, D. Krewski, P. Ghadirian, M. J. Thun, and E. E. Calle. ␧ 2005. Cancer mortality among U.S. men and women with asthma and hay fever. J. Y. Bonnefoy, B. Dugas, and J. P. Kolb. 1995. Evidence for a role of Fc RII/ Am. J. Epidemiol. 162: 212–221. CD23 in the IL-4-induced nitric oxide production by normal human mononuclear phagocytes. Cell. Immunol. 163: 314–318. 18. Wiemels, J. L., J. K. Wiencke, J. Patoka, M. Moghadassi, T. Chew, A. McMillan, 44. Yokota, A., K. Yukawa, A. Yamamoto, K. Sugiyama, M. Suemura, Y. Tashiro, R. Miike, G. Barger, and M. Wrensch. 2004. Reduced and T. Kishimoto, and H. Kikutani. 1992. Two forms of the low-affinity Fc receptor allergy among adults with glioma compared with controls. Cancer Res. 64: for IgE differentially mediate endocytosis and phagocytosis: identification of the 8468–8473. critical cytoplasmic domains. Proc. Natl. Acad. Sci. USA 89: 5030–5034. 19. Brenner, A. V., M. S. Linet, H. A. Fine, W. R. Shapiro, R. G. Selker, P. M. Black, 45. Dyall, R., L. V. Vasovic, R. A. Clynes, and J. Nikolic-Zugic. 1999. Cellular and P. D. Inskip. 2002. History of allergies and autoimmune diseases and risk of requirements for the monoclonal antibody-mediated eradication of an established brain tumors in adults. Int. J. Cancer 99: 252–259. solid tumor. Eur. J. Immunol. 29: 30–37. 20. Wiemels, J. L., J. K. Wiencke, J. D. Sison, R. Miike, A. McMillan, and 46. Bracher, M. G., H. J. Gould, B. J. Sutton, D. Dombrowitz, and S. N. Karagiannis.

M. Wrensch. 2002. History of allergies among adults with glioma and controls. by guest on September 29, 2021 2007. Three-colour flow cytometric method to measure antibody-dependent tu- Int. J. Cancer 98: 609–615. mour cell killing by cytotoxicity and phagocytosis. J. Immunol. Methods 323: 21. Wang, H., and T. L. Diepgen. 2005. Is atopy a protective or a risk factor for 160–171. cancer? A review of epidemiological studies. Allergy 60: 1098–1111. 47. Kayaba, H., D. Dombrowicz, G. Woerly, J. P. Papin, S. Loiseau, and M. Capron. 22. Turner, M. C., Y. Chen, D. Krewski, and P. Ghadirian. 2006. An overview of the 2001. Human eosinophils and human high affinity IgE receptor transgenic mouse association between allergy and cancer. Int. J. Cancer 118: 3124–3132. eosinophils express low levels of high affinity IgE receptor, but release IL-10 23. Wang, H., and T. L. Diepgen. 2006. Atopic dermatitis and cancer risk. upon receptor activation. J. Immunol. 167: 995–1003. Br. J. Dermatol. 154: 205–210. 48. Neuberger, M. S., G. T. Williams, E. B. Mitchell, S. S. Jouhal, J. G. Flanagan, 24. Wang, H., D. Rothenbacher, M. Low, C. Stegmaier, H. Brenner, and and T. H. Rabbitts. 1985. A hapten-specific chimaeric IgE antibody with human T. L. Diepgen. 2006. Atopic diseases, immunoglobulin E and risk of cancer of the physiological effector function. Nature 314: 268–270. prostate, breast, lung and colorectum. Int. J. Cancer 119: 695–701. 49. Wakai, M., P. Pasley, Z. M. Sthoeger, D. N. Posnett, R. Brooks, S. Hashimoto, 25. Nagy, E., I. Berczi, and A. H. Sehon. 1991. Growth inhibition of murine mam- and N. Chiorazzi. 1993. Anti-CD23 monoclonal antibodies: comparisons of mary carcinoma by monoclonal IgE antibodies specific for the mammary tumor epitope specificities and modulating capacities for IgE binding and production. virus. Cancer Immunol. Immunother. 34: 63–69. Hybridoma 12: 25–43. 26. Kershaw, M. H., P. K. Darcy, J. A. Trapani, D. MacGregor, and M. J. Smyth. 50. Riske, F., J. Hakimi, M. Mallamaci, M. Griffin, B. Pilson, N. Tobkes, P. Lin, 1998. Tumor-specific IgE-mediated inhibition of human colorectal carcinoma W. Danho, J. Kochan, and R. Chizzonite. 1991. High affinity human IgE receptor xenograft growth. Oncol. Res. 10: 133–142. (Fc␧RI): analysis of functional domains of the ␣-subunit with monoclonal anti- 27. Gould, H. J., G. A. Mackay, S. N. Karagiannis, C. M. O’Toole, P. J. Marsh, bodies. J. Biol. Chem. 266: 11245–11251. B. E. Daniel, L. R. Coney, V. R. Zurawski, Jr., M. Joseph, M. Capron, et al. 1999. 51. Cook, J. P., A. J. Henry, J. M. McDonnell, R. J. Owens, B. J. Sutton, and Comparison of IgE and IgG antibody-dependent cytotoxicity in vitro and in a H. J. Gould. 1997. Identification of contact residues in the IgE binding site of SCID mouse xenograft model of ovarian carcinoma. Eur. J. Immunol. 29: human Fc␧RI␣. Biochemistry 36: 15579–15588. 3527–3537. 52. Benard, J., J. Da Silva, M. C. De Blois, P. Boyer, P. Duvillard, E. Chiric, and 28. Karagiannis, S. N., Q. Wang, N. East, F. Burke, S. Riffard, M. G. Bracher, G. Riou. 1985. Characterization of a human ovarian adenocarcinoma line, R. G. Thompson, S. R. Durham, L. B. Schwartz, F. R. Balkwill, and H. J. Gould. IGROV1, in tissue culture and in nude mice. Cancer Res. 45: 4970–4979. 2003. Activity of human monocytes in IgE antibody-dependent surveillance and 53. Hagemann, T., J. Wilson, F. Burke, H. Kulbe, N. F. Li, A. Pluddemann, killing of ovarian tumor cells. Eur. J. Immunol. 33: 1030–1040. K. Charles, S. Gordon, and F. R. Balkwill. 2006. Ovarian cancer cells polarize 29. Kershaw, M. H., P. K. Darcy, J. A. Trapani, and M. J. Smyth. 1996. The use of macrophages toward a tumor-associated phenotype. J. Immunol. 176: 5023–5032. chimeric human Fc␧ receptor I to redirect cytotoxic T to tumors. 54. Otten, M. A., E. Rudolph, M. Dechant, C. W. Tuk, R. M. Reijmers, R. H. Beelen, J. Leukocyte Biol. 60: 721–728. J. G. van de Winkel, and M. van Egmond. 2005. Immature neutrophils mediate 30. Teng, M. W., M. H. Kershaw, J. T. Jackson, M. J. Smyth, and P. K. Darcy. 2006. tumor cell killing via IgA but not IgG Fc receptors. J. Immunol. 174: 5472–5480. Adoptive transfer of chimeric Fc␧RI gene-modified human T cells for cancer 55. Balkwill, F. R., A. Lee, G. Aldam, E. Moodie, J. A. Thomas, J. Tavernier, and immunotherapy. Hum. Gene Ther. 17: 1134–1143. W. Fiers. 1986. Human tumor xenografts treated with recombinant human tumor 31. Untersmayr, E., I. Scholl, I. Swoboda, W. J. Beil, E. Forster-Waldl, F. Walter, necrosis factor alone or in combination with interferons. Cancer Res. 46: A. Riemer, G. Kraml, T. Kinaciyan, S. Spitzauer, et al. 2003. Antacid medication 3990–3993. inhibits digestion of dietary proteins and causes food allergy: a fish allergy model 56. Vercelli, D., B. Helm, P. Marsh, E. Padlan, R. S. Geha, and H. Gould. 1989. The in BALB/c mice. J. Allergy Clin. Immunol. 112: 616–623. B-cell binding site on human immunoglobulin E. Nature 338: 649–651. 32. Scholl, I., E. Untersmayr, N. Bakos, F. Roth-Walter, A. Gleiss, G. Boltz-Nitulescu, 57. Keegan, A. D., C. Fratazzi, B. Shopes, B. Baird, and D. H. Conrad. 1991. Char- O. Scheiner, and E. Jensen-Jarolim. 2005. Antiulcer drugs promote oral sensitization acterization of new rat anti-mouse IgE monoclonals and their use along with and to hazelnut in BALB/c mice and humans. Am. J. Clin. chimeric IgE to further define the site that interacts with Fc␧RII and Fc␧RI. Mol. Nutr. 81: 154–160. Immunol. 28: 1149–1154. 33. Riemer, A. B., E. Untersmayr, R. Knittelfelder, A. Duschl, H. Pehamberger, 58. Schoppmann, S. F., P. Birner, J. Stockl, R. Kalt, R. Ullrich, C. Caucig, C. C. Zielinski, O. Scheiner, and E. Jensen-Jarolim. 2007. Active induction of E. Kriehuber, K. Nagy, K. Alitalo, and D. Kerjaschki. 2002. Tumor-associated The Journal of Immunology 2843

macrophages express lymphatic endothelial growth factors and are related to 64. Lotfi, R., J. J. Lee, and M. T. Lotze. 2007. Eosinophilic and damage- peritumoral lymphangiogenesis. Am. J. Pathol. 161: 947–956. associated molecular pattern molecules (DAMPs): role in the inflammatory re- 59. Leung, D. Y., E. E. Schneeberger, R. P. Siraganian, R. S. Geha, and A. K. Bhan. sponse within tumors. J. Immunother. 30: 16–28. 1987. The presence of IgE on macrophages and dendritic cells infiltrating into the 65. Keler, T., P. K. Wallace, L. A. Vitale, C. Russoniello, K. Sundarapandiyan, R. F. Graziano, and Y. M. Deo. 2000. Differential effect of cytokine treatment on skin lesion of atopic dermatitis. Clin. Immunol. Immunopathol. 42: 328–337. Fc␣ receptor I- and Fc␥ receptor I-mediated tumor cytotoxicity by monocyte- 60. Thomsen, L. L., and D. W. Miles. 1988. Role of nitric oxide in tumor progres- derived macrophages. J. Immunol. 164: 5746–5752. sion: lessons from human tumors. Cancer Metastasis Rev. 17: 107–118. 66. van Spriel, A. B., H. H. van Ojik, A. Bakker, M. J. Jansen, and 61. Gounni, A. S., B. Lamkhioued, K. Ochiai, Y. Tanaka, E. Delaporte, A. Capron, J. G. J. van de Winkel. 2003. Mac-1 (CD11b/CD18) is crucial for effective Fc J. P. Kinet, and M. Capron. 1994. High-affinity IgE receptor on eosinophils is receptor-mediated immunity to melanoma. Blood 101: 253–258. involved in defence against parasites. Nature 367: 183–186. 67. Stockmeyer, B., M. Dechant, M. van Egmond, A. L. Tutt, K. Sundarapandiyan, R. F. Graziano, R. Repp, J. R. Kalden, M. Gramatzki, M. J. Glennie, et al. 2000. 62. Cormier, S. A., A. G. Taranova, C. Bedient, T. Nguyen, C. Protheroe, R. Pero, Triggering Fc␣receptor I (CD89) recruits neutrophils as effector cells for CD20- D. Dimina, S. I. Ochkur, K. O’Neill, D. Colbert, et al. 2006. Pivotal advance: directed antibody therapy. J. Immunol. 165: 5954–5961. eosinophil infiltration of solid tumors is an early and persistent inflammatory host 68. Park, J. W., R. S. Kerbel, G. J. Kelloff, J. C. Barrett, B. A. Chabner, response. J. Leukocyte Biol. 79: 1131–1139. D. R. Parkinson, J. Peck, R. W. Ruddon, C. C. Sigman, and D. J. Slamon. 2004. 63. Munitz, A., and F. Levi-Schaffer. 2004. Eosinophils: ‘new’ roles for ‘old’ cells. Rationale for biomarkers and surrogate end points in mechanism-driven oncology Allergy 59: 268–275. drug development. Clin. Cancer Res. 10: 3885–3896. Downloaded from http://www.jimmunol.org/ by guest on September 29, 2021