Gene Therapy (2001) 8, 1721–1728  2001 Nature Publishing Group All rights reserved 0969-7128/01 $15.00 www.nature.com/gt RESEARCH ARTICLE Bypassing tumor-specific and bispecific : triggering of antitumor immunity by expression of anti- FcγR scFv on cancer cell surface

N Gruel1, WH Fridman2 and JL Teillaud1,2 1Laboratoire de Biotechnologie des Anticorps, Institut Curie, Paris, France; and 2Unite´ INSERM U255, Institut Curie, Paris, France

We have developed a novel immunostimulatory molecule (KIR)-mediated inhibition of NK cell cytotoxicity. This anti- against tumor cells, composed of an anti-Fc␥RIII (CD16) CD16 scFv tumor expression also enhanced tumor phago- scFv fused to the platelet-derived growth factor cytosis by IFN␥-activated , a mechanism (PDGFR) transmembrane region. This fusion molecule was known to induce a protective long-term adaptative immunity stably expressed on the tumor cell surface and retained the to tumors. In vivo Winn tests performed in SCID mice ability of the parental to bind soluble CD16. Tumor showed that the expression of anti-CD16 scFv on tumor cells expressing anti-CD16 scFv triggered the release of IL- cells, but not of the negative control anti-phOx scFv, pre- 2 by Jurkat-CD16/␥ cells and of TNF␣ by when vented tumor cell growth. Thus, expression of FcR anti- co-cultured with these cells. Furthermore, NK cells could kill bodies or other FcR-specific ligands on tumor cells rep- scFv-transfected HLA+ class I H1299 lung carcinoma tumor resents a novel and potent antibody-based gene therapy cells, but not the parental cells, indicating that anti-CD16 approach, which may have clinical applications in cancer scFv tumor expression prevents the killer inhibitory receptor therapy. Gene Therapy (2001) 8, 1721–1728.

Keywords: single chain antibody; Fc␥R; immunotherapy; monocytes/macrophages; NK cells

Introduction rapid catabolism. Circulating antibodies also provoke side-toxicity such as fevers, nausea, vomiting and leuko- Antibodies are attractive tools for cancer immunotherapy penia, likely due to the binding and activation of FcγR+ due to their ability to bind tumor cells and to activate cells.7,9 In addition to the toxicity, mouse antibodies complement and/or to recruit effector cells through the induce human anti-mouse antibody (HAMA)3 responses interaction of their Fc regions with receptors for IgG in patients, limiting their use for clinical purposes. γ α (Fc R) or for IgA (Fc R). NK cells, monocytes and macro- An alternative to the use of bispecific or tumor-specific phages are able to kill tumor cells by antibody-dependent antibodies is the recruitment of immune effector cells by cell cytotoxicity following the engagement of CD16 or the tumor itself. Tumor cells have been genetically engi- γ γ CD32, the low-affinity Fc RIII and Fc RII, or CD64, the neered to enhance their immunogenicity. Tumor cells γ 1 high-affinity Fc RI. In addition, the internalization of that express recombinant cytokines,10,11 MHC,12 costimu- IgG containing immune-complexes bound to these recep- latory molecules,13,14 or combinations of these mol- 2 tors enhances antigen presentation to T cells. NK cells ecules,15 have been used to induce tumor regression in and monocytes also produce a variety of cytokines upon mouse models. However, the cells recruited in these cases γ 1,3 Fc R engagement. Bispecific antibodies directed against are mostly limited to tumor-specific T cells or to a restric- mostly CD16 or CD64 and tumor-associated antigens ted subset of cells responsive to a given cytokine. An 4–6 have been therefore generated and clinical trials in ideal strategy would be to recruit a large variety of cells 7–9 patients have been initiated. from both the innate and the adaptative immunity that However, the use of antibodies or bispecific antibodies could exert a strong antitumor response through various in cancer therapy requires the generation of antitumor mechanisms and which could be memorized. antibodies specific for a given tumor type. This has An attempt to express single chain Fv (scFv) directed strongly limited the use of therapeutic antibodies in can- against FcγR by tumor cells is therefore attractive because cer patients over the last 20 years. Moreover, the injection of the broad expression of these molecules at the surface of antibodies into the peripheral blood of patients ham- of cells playing a major role in the antitumor immunity. pers the efficiency of their targeting to tumor sites due Here, we describe a new fusion molecule composed of γ + to the nonspecific binding to Fc R circulating cells and one scFv directed against CD16 fused to the transmem- brane region of platelet-derived growth factor receptor (PDGFR). When expressed at the surface of tumor cells, Correspondence: JL Teillaud, Unite´ INSERM 255, Centre de Recherches Biome´dicales des Cordeliers, 15 rue de l’Ecole de Me´decine, 75270 Paris this molecule allows the recruitment and activation of cedex 06, France NK cells, monocytes and IFN-γ-activated macrophages. Received 17 May 2001; accepted 16 August 2001 It induces cytokine release, tumor cell cytotoxicity and Antitumor immunity with surface anti-FcR scFv N Gruel et al 1722 overcomes the killer inhibitory eceptor (KIR)-mediated CD16 scFv or anti-phOx scFv cDNAs were then isolated. inhibition of NK cells. It increases also the tumor cell Indirect immunofluorescence analyses performed with uptake by phagocytic cells. Finally, this fusion molecule polyclonal rabbit IgG anti-scFv17 (Figure 1b) showed that prevents tumor cell growth in Winn tests performed with cells from selected clones express scFv at their surface. SCID mice. Clone 25 (anti-CD16 scFv) and clone 57 (anti-phOx scFv) Thus, this approach may have a broad use in triggering were used in all further studies. protective immunity against tumors or various patho- Anti-CD16 scFv expressed at the cell surface of H1299 gens as it could be also adapted using scFv directed cells, or at least part of them, retain the ability of their against FcγrRI, or any component which binds to a Fc soluble scFv counterparts to bind CD16. Biotinylated receptor. recombinant sCD16 corresponding to the two extracellu- lar domains of the NA2 form of the CD16-B receptor18 Results was incubated with H1299-anti-CD16 scFv cells. Its pres- ence at the cell surface could be demonstrated by Cell surface expression of anti-CD16 scFv and binding addition of FITC-streptavidin (Figure 1c, middle panel). of biotinylated sCD16 Furthermore, this binding could be inhibited by mixing the 3G8 anti-CD16 mAb with biotinylated sCD16 before An anti-CD16 scFv was derived from the 3G8 mAb16 and incubation with the cells (Figure 1c, right panel). No fused to the transmembrane domain of the PDGFR as direct binding of FITC-streptavidin to H1299-anti-CD16 described in Materials and methods. The expression of scFv was observed (Figure 1c, left panel). the anti-CD16 scFv fusion at the cell surface was first assessed by immunofluorescence with transiently trans- Cytokine production by CD16+ cells stimulated with fected H1299 lung carcinoma tumor cells. Anti-CD16 scFv H1299-anti-CD16 scFv cells was detected at the cell surface of H1299 cells (Figure In a first set of experiments, the ability of Jurkat-CD16/γ 1a), as observed for the anti-phOx scFv fusion (data not cells19 to produce IL-2 during a co-culture with H1299- shown). H1299 clones stably transfected with either anti- anti-CD16 scFv cells for 24 h was examined. Co-cultures in presence of PMA (10 ng/ml) triggered IL-2 release as demonstrated by a specific quantitative ELISA (data not shown). The biological activity of the IL-2 produced dur- ing these co-cultures was demonstrated by a proliferation assay using the IL-2 dependent CTLL-2 cells (Figure 2a). There was no significant IL-2 production with H1299- anti-phOx scFv cells co-cultured in similar conditions. As expected, no activation of Jurkat-CD16/γ cells was observed with PMA alone or with H1299-anti-CD16 scFv cells in absence of PMA (data not shown), since the trig- gering of IL-2 production by Jurkat cells requires two signals.20 Similarly, the production of TNFα by freshly isolated human monocytes was induced when monocytes were co-cultured with H1299-anti-CD16 scFv cells, but not with H1299-anti-phOx scFv cells, as shown by the cyto- toxicity induced by the corresponding co-culture super- natant on TNFα-sensitive L929 cells21 (Figure 2b). No spontaneous TNFα production by freshly isolated mono- cytes was detected. Similarly, scFv-expressing H1299 cells did not produce TNFα when they were cultured alone or in the presence of sCD16.

H1299-anti-CD16 scFv tumor cells induce PBMC cytotoxicity Whether the expression of anti-CD16 scFv at the surface of tumor cells could trigger antitumor cytotoxicity by freshly isolated huPBMC was then examined. H1299 lung Figure 1 Expression of anti-CD16 scFv at the surface of the human lung carcinoma cells transfected with anti-CD16 scFv or anti- carcinoma cell line H1299. (a) Transiently transfected cells cultured in phOx scFv encoding plasmids were mixed with huPBMC Labtek were analyzed by epifluorescence. Cell surface scFv were revealed at different ratios and incubated at 37°C for 4 h before with anti-scFv rabbit IgG and FITC-labeled goat anti-rabbit IgG anti- bodies. Cell nuclei were stained with DAPI. (b) H1299-anti-CD16 scFv LDH measurement. HuPBMC killed H1299 cells only (clone 25) or H1299-anti-phOx scFv (clone 57) clones were analyzed for when these cells express anti-CD16 scFv. Figure 3a shows cell surface scFv expression using fluorescence flow cytometry. In both the cytotoxicity obtained with the huPBMC from one cases, the presence of scFv was revealed with anti-scFv rabbit IgG and donor. 17.5% PBMC from this donor expressed CD16+ FITC-labeled goat anti-rabbit IgG antibodies. (c) H1299-anti-CD16 scFv (data not shown). The cytotoxicity was inhibited by clone 25 was analyzed for its ability to bind biotinylated sCD16. There sCD16 in a dose-dependent manner (Figure 3b). was no direct binding of FITC-streptavidin on H1299-anti-CD16 scFv cells (left panel). The binding of biotinylated sCD16 was revealed with A cytotoxicity ranging from 32 to 82% (mean value: FITC-streptavidin (middle panel). Inhibition of the binding was performed 52.8% ± 14.1%; E:T = 100:1) could be achieved depending by adding the parental 3G8 mAb (right panel). on the donor in five independent experiments. The killing

Gene Therapy Antitumor immunity with surface anti-FcR scFv N Gruel et al 1723

Figure 3 Cytotoxic activity of peripheral blood mononuclear cells (huPBMC) induced by H1299-anti-CD16 scFv cells. (a) Freshly isolated + huPBMC were incubated for 4 h with H1299-anti-CD16 scFv cells (in the Figure 2 IL-2 and TNFα production by CD16 cells co-cultured with presence of sCD16 or not), or with H1299-anti-phOx scFv cells. Specific H1299-anti-CD16 scFv cells. (a) Jurkat-CD16/␥ or Jurkat cells were co- cytotoxicity was determined as described in Materials and methods. (b) cultured for 24 h with H1299-anti-CD16 scFv or H1299-anti-phOx scFv Dose-dependent inhibition of cytotoxicity by sCD16 (E:T ratio: 50:1). cells in the presence of PMA. IL-2 was quantified with a bio-assay using the IL-2-dependent CTLL-2 cells. Results are from duplicates. (b) Mono- cytes were cultured alone, or co-cultured with H1299-anti-CD16 scFv or H1299-anti-phOx scFv cells for 15 h. TNF␣ was detected by a cytotoxicity mAb or not (Figure 4b). H1299-anti-phOx scFv cell assay using the TNF␣-sensitive L929 cells. Results are from triplicates. extracts were used as control. A strong red fluorescence corresponding to phagocytosed PKH26-labeled H1299- anti-CD16 scFv cell extracts was observed within FITC- was due to the binding of anti-CD16 scFv to CD16 CD14+/CD64+ IFNγ-activated macrophages (Figure 4b, expressed at the surface of huPBMC, as it was inhibited left panel). In contrast, no or only a faint red fluorescence by the addition of 25 ␮g/ml sCD16 (% inhibition mean was observed when IFNγ-activated macrophages were value: 73% ± 20.6; E:T = 100:1; n = 4). In contrast, only incubated with sonicated H1299-anti-phOx scFv cells 5.7% ± 6.4 cytotoxicity (E:T = 100:1; n = 5) were observed (Figure 4b, middle panel) or with 3G8 mAb before the when huPBMC were mixed with H1299-anti-phOx scFv addition of sonicated H1299-anti-CD16 scFv cells (Figure cells. 4b, right panel).

Enhanced of H1299-anti-CD16 scFv cells Winn test by macrophages The ability of H1299-anti-CD16 scFv tumor cells to recruit The phagocytosis of H1299-anti-CD16 scFv cells by IFNγ- and activate CD16+ huPBMC in vivo was then tested in activated macrophages22 was then examined. First, the Winn tests. Anti-Asialo GM1-treated SCID mice were co- binding of IFNγ-activated macrophages to H1299-anti- engrafted with H1299 tumor cells expressing either anti- CD16 scFv cells after 4 h of incubation at 4°C was ana- CD16 scFv or anti-phOx scFv and freshly isolated lyzed. It was higher than the binding to H1299-anti-phOx huPBMC (ratio 1:40). All the animals from the control scFv (11.5% versus 1.8%, respectively) (data not shown). group (H1299-anti-phOx scFv; n = 5) developed solid At 37°C, IFNγ-activated macrophages bound and phago- tumors (2/5) or tumor ascites (3/5) (Figure 5a). These cytosed more H1299-anti-CD16 scFv cells (Figure 4a, left tumor cells still expressed anti-phOx scFv after growing panel) than H1299-anti-phOx scFv cells (Figure 4a, right in vivo (data not shown). In contrast, only one mouse panel). The percentage of binding and phagocytosis was (1/6) engrafted with H1299-anti-CD16 scFv developed a 15.7% and 4%, respectively. In this experiment, 79.5% of solid tumor (Figure 5a). Cells isolated from this tumor IFNγ-activated macrophages expressed CD16 (MFI = still expressed anti-CD16 scFv. Thus, co-engraftment of 119.8). The phagocytosis was confirmed by immunofluo- tumor cells expressing anti-CD16 scFv and huPBMC rescence analysis of adherent IFNγ-activated macro- results in a strong inhibition of tumor development in phages incubated with H1299-anti-CD16 scFv sonicated SCID mice. H1299-anti-CD16 scFv and H1299-anti-phOx cell extracts, in the presence of 25 ␮g/ml anti-CD16 3G8 scFv cells exhibited the same growth kinetics in absence

Gene Therapy Antitumor immunity with surface anti-FcR scFv N Gruel et al 1724

Figure 4 Phagocytosis of H1299-anti-CD16 scFv cells by IFNγ-activated human macrophages. (a) PKH26-labeled tumor cells (H1299-anti-CD16 scFv or H1299-anti-phOx scFv cells) were incubated with IFNγ-activated macrophages for 4 h at 37°C (E:T ratio: 1:1). Macrophages were then stained with FITC-CD14/CD64 mAbs and 10 000 cells were analyzed using a FACScalibur. Double-labeled cells (upper right quadrants) correspond to IFN␥- activated macrophages stained with FITC-CD14+/CD64+ mAbs that have bound and/or phagocytosed PKH26+-labeled tumor cells. (b) IFN␥-activated ° human macrophages were cultured on Labteck slides for 16 h at 37 Cina5%CO2 atmosphere and incubated for 30 min on ice in the presence of 3G8 mAb (right panel) or not (left and middle panels). Sonicated PKH26+-H1299-scFv cell extracts were then added and macrophages were incubated for a further 4 h period at 37°C. Immunofluorescence assays were performed as described in Materials and methods.

of huPBMC (data not shown). In addition, all mice med in SCID mice indicated that the expression of anti- injected with H1299-anti-CD16 scFv cells were alive after CD16 scFv at the tumor cell surface prevents the growth 92 days, while only two of five control mice were alive of tumor cells. It should be noted that a murine tumor at that time (Figure 5b). model using immunocompetent mice is difficult to develop as the anti-mouse FcγR mAb which is available, γ γ Discussion 2.4G2, is directed both against Fc RII and Fc RIII, and does not discriminate between ITIM and ITAM bearing In this report, a novel antibody-based strategy allowing receptors.1 the recruitment and activation of immune effector cells The targeting of FcγRorFcαR by a specific is described (Figure 6). It is based on the expression of expressed on tumor cells has multiple advantages com- anti-FcγR scFv at the tumor cell surface. It allows FcγR- pared with the surface expression of other such dependent cell functions to be triggered and tumor cell as CD323 or co-stimulatory molecules.13–15 In contrast to cytotoxicity to be induced without requiring the isolation these molecules, FcR are widely expressed on a large var- and use of specific antitumor antibodies. This represents iety of cells of the (Figure 6) and their an important benefit since many tumors do not express engagement provokes the activation of multiple cell func- specific tumor antigens. We showed that a functional tions.1,24 Interestingly, cells from both innate and adapt- scFv directed against FcγRIII (CD16) can be expressed at ative immunity can be recruited and activated by the the surface of human tumor cells and can activate binding to and triggering of FcR.1 On the one hand, FcγRIII+ cells. The PDGFR-TM domain fused to the two potent FcγR+ killer cells such as G-CSF-activated neutro- c-myc tags allows the binding of the anti-CD16 scFv to phils or NK cells can be recruited to directly kill tumor surface CD16 and the triggering of various FcγRIII- cells through the activation of FcγRI or FcγRIII, respect- mediated functions, as observed with anti-CD16 anti- ively.4,25 On the other hand, the enhancement of phago- bodies. The fusion expressed at the tumor cell cytosis of tumor cells or cell extracts by antigen- surface induces the secretion of IL-2 and TNFα when presenting cells through FcγR could induce a protective tumor cells are co-cultured with CD16+ T cells (Jurkat- long-term immunity to tumors by allowing recruitment CD16/γ) or monocytes, respectively. Furthermore, NK of T cells.26–28 The expression of anti-CD16 scFv at the cells killed anti-CD16 scFv-transfected HLA+ class I tumor cell surface allowed demonstration of the fact that H1299 lung carcinoma tumor cells, but not the parental CD16 expressed on activated human macrophages is cells, indicating that the expression of anti-CD16 scFv at involved in phagocytosis. Of note is that the present the tumor cell surface can overcome the KIR-mediated approach provides a distinct advantage in that no parti- inhibition of NK cell cytotoxicity. ScFv CD16 tumor cular tumor antigen needs to be known for antigen- expression also enhanced tumor phagocytosis by FcγR+ presenting cell targeting. Another advantage in using the antigen-presenting cells. Last, in vivo Winn tests perfor- present approach is the absence of co-recruitment of

Gene Therapy Antitumor immunity with surface anti-FcR scFv N Gruel et al 1725

Figure 6 Diagram representing the targeted immunotherapy (the ‘Hook- ing tumor strategy’) based on the expression of recombinant antibody frag- ments directed against FcγR on tumor cells. Cells from both innate and adaptative immunity can be recruited and activated by the triggering of Fc␥R. Fc␥R+ killer cells, such as G-CSF-activated polymorphonuclear cells (PMN) or NK cells, can be recruited to kill tumor cells through the activation of FcγR. The enhancement of phagocytosis of tumor cells or cell extracts by antigen-presenting cells such as macrophages or immature dendritic cells (DC) through Fc␥R could also induce a protective long- term immunity to tumors by allowing recruitment of specific T cells. The engagement of Fc␥R also triggers the production of various cytokines and pro-inflammatory molecules involved in antitumor activities.

This antibody-based therapy strategy for cancer treat- ment also has several potential advantages over the use of engineered bispecific antibodies directed against FcγR and tumor antigens.32–34 First, as already discussed, no tumor-specific antibody has to be isolated and the mol- ecule can be used for the treatment of different tumor types that do not express known tumor-associated anti- gens. Second, the recruitment of effector cells should be improved by using an intra-tumoral gene delivery sys- Figure 5 Inhibition of H1299-anti-CD16 scFv tumor growth by tem, which allows the expression of the anti-FcγR scFv in huPBMC. Groups of six mice were engrafted with 1.5 × 106 transfected situ. No systemic delivery is needed, avoiding the rapid × 6 H1299-scFv cells and 50 10 huPBMC. (a) Percentage of animals with- trapping and clearance of the injected antibody by FcγR+ out tumor in the control group (H1299-anti-phOx scFv cells) (open circles), or in the group of mice engrafted with H1299-anti-CD16 scFv cells present in the circulating blood, liver and other cells (black squares). (b) Percentage of survival in the control group organs. In addition, it should decrease the appearance of (H1299-anti-phOx scFv cells) (open circles), or in the group of mice the side-effects provoked by the activation of circulating engrafted with H1299-anti-CD16 scFv cells (black squares). FcγR+ cells upon binding of IgG antitumor antibodies or bispecific antibodies directed against FcγR and tumor antigens. Third, it is likely to be weakly immunogenic, FcγRII and FcγRIII expressed on , which since no constant domain of murine immunoglobulin is induces a severe toxicity.29 Our data also show that it is present in the recombinant antibody fragment. A weak possible to block the negative immuno-modulatory func- immunogenicity may even have some benefits, by trig- tion exerted by KIR on NK cells by expressing anti-CD16 gering an adjuvant immune effect against tumor cells antibodies on HLA+ tumor cells otherwise resistant or expressing the transgene. In the case of an HAMA poorly sensitive to NK cell activity. The balance between immune response that could block the interaction of sur- the positive signal triggered by the engagement of face scFv and immune cells, an alternative would be to FcγRIII-A on NK cells and the negative signal due to the use human anti-FcγR scFv. However, it should be pointed KIR/HLA interaction can therefore be modified in favor out that our approach, as other cancer gene therapy of the former one by expressing anti-FcγR scFv at the approaches, relies on the use of an efficient gene delivery tumor surface. It clearly demonstrates the existence of system, allowing repeated injections into the tumor. fine tuning between the activation and inhibition path- Although one can expect that the recruitment of even a ways in these cells.30 The ability to trigger the killing of limited number of FcγR+ cells from various hematopoietic HLA class I+ tumor cells may be useful, since it has been lineages can be sufficient to trigger a potent antitumor postulated that KIR expression may result in the inability immune response, it has still to be documented. to control tumor growth by killer cells.31 In conclusion, our data demonstrate that anti-FcγR scFv

Gene Therapy Antitumor immunity with surface anti-FcR scFv N Gruel et al 1726 expressed at the tumor cell surface can be used to trigger sodium pyruvate, 100 IU/ml penicillin and 100 ␮g/ml efficient FcγR-dependent immune responses against streptomycin. tumors. We developed a new versatile therapeutic tool IFNγ-activated macrophages were obtained as that may have clinical applications in cancer therapy and previously described.22 can easily be adapted to target other FcγRorFcαRI (CD89), using adequate anti-FcR antibodies or other FcR- Immunofluorescence assays specific ligands. Further preclinical studies in mouse H1299 cells were grown in two-well-Labtek chamber tumor models should allow optimization of its use for slides (Nunc, Roskilde, Denmark) (2 × 105 cell/well) for cancer treatment. 24 h and transfected with 250 ng of plasmid encoding anti-CD16 scFv or anti-phOx scFv, or mock transfected. Indirect immunofluorescence assays were performed 24 Materials and methods h after transfection. After two washes in PBS with cal- cium and magnesium (PBS-Ca++Mg++) (Gibco-BRL), cells Construction of anti-CD16 surface display vector were fixed with PBS-Ca++Mg++-4% formaldehyde for 15 The anti-CD16 scFv cDNA was derived from the 3G8 min at RT. After two washes in PBS-Ca++Mg++ and satu- mouse mAb16 by a three-step PCR method as previously ration with PBS-Ca++Mg++-1% BSA (30 min, RT), slides described.35 It was then cloned into the eukaryotic were incubated with rabbit anti-scFv IgG17 diluted in expression NeoR pHook-2 vector (kind gift of Dr M Rus- PBS-Ca++Mg++-1% BSA (2 ␮g/ml). Cells were then sel, Invitrogen, San Diego, CA, USA), using the SfiI and washed with PBS-Ca++Mg++ and incubated with goat anti- NotI restriction sites. The pHook-2 vector initially con- rabbit IgG antibodies coupled to FITC (Jackson Immuno- tains a cDNA insert encoding for an anti-4-ethoxy-meth- research, West Grove, PA, USA). The slides were stained ylene-2-phenyl-2-oxazolin-5-one scFv (anti-phOx scFv) with Vectashield containing DAPI (Vector, Burlingame, fused to a mouse PDGFR transmembrane domain CA, USA) and analyzed with an epifluorescence micro- through a repeated c-myc motif, under the control of a scope. RSV promoter.36 The cDNA encoding anti-phOx scFv Stable transfected cells were examined for scFv surface was removed and replaced by the anti-CD16 scFv cDNA expression using indirect immunofluorescence assays. 5 after deletion of an extra-3Ј NotI site by site-directed PCR × 105 transformed cells were washed twice with PBS and mutagenesis. The anti-phOx scFv construct was used as incubated for 30 min at 4°C with rabbit IgG anti-scFv negative control in all further experiments. diluted in PBS (2 ␮g/ml). After two washes in PBS, scFv were revealed with FITC-labeled goat Ig anti-rabbit IgG. Cell lines and transfections Cells were then washed in PBS and fixed with 500 ␮lof The non-small cell lung carcinoma H1299 adherent cells PBS-1% formaldehyde. Ten thousand cells per experi- (kind gift of Dr L Bracco, Aventis, Vitry sur Seine, ment were analyzed with a FACScalibur cytometer France), were cultured with DMEM (Seromed, Berlin, (Becton Dickinson, San Jose, CA, USA) using the Cell- Germany), 10% heat-inactivated fetal calf serum (FCS) quest program. (Hyclone, Logan, UT, USA), 2 mml-glutamine (Seromed), 1 mm sodium pyruvate (Seromed), 100 IU/ml Biotinylated recombinant soluble CD16 binding penicillin (Seromed) and 100 ␮g/ml streptomycin Purified recombinant soluble CD16 (sCD16),37 kindly (Seromed) (culture medium). 5 × 106 cells were transfor- provided by Pr C Saute`s-Fridman, was biotinylated using med with 1 ␮g of plasmid encoding anti-CD16 scFv or EZ-Link Sulfo-NHS-LC-Biotin (Pierce, Rockford, IL, anti-phOx scFv. Transfections were performed using USA), according to the manufacturer’s instructions, with lipofectAMINE (Gibco-BRL, Paisley, UK) according to a 25 biotin/sCD16 molar ratio. The presence of the 3G8 the manufacturer’s instructions. Briefly, 1 ␮g of plasmid epitope on biotinylated sCD16 was assessed with a sand- DNA was mixed with 5 ␮g of lipofectAMINE in 500 ␮l wich ELISA using the 3G8 mAb as capture antibody. The Opti-MEM medium (Gibco-BRL) and incubated for 30 binding of biotinylated CD16 at the cell surface was min at room temperature (RT) before being added to detected by a direct fluorescence assay using FITC-strep- H1299 cells. Cells were then incubated for 4 h at 37°C. tavidin (Jackson Immunoresearch) as the revealing After removing Opti-MEM, 3 ml of fresh culture medium reagent. Inhibition experiments were performed by mix- was added. Stable transformants were selected by adding ing biotinylated sCD16 with purified 3G8 mAb (4:1 molar G418 (Geneticin; Gibco-BRL) (2 mg/ml) 48 h later and ratio) for 30 min at 4°C before addition to the cells. examined for scFv cell surface expression. They were further cloned and grown in the presence of 1 mg/ml IL-2 production assay G418. 105 Jurkat-CD16/γ transfected cells19 were stimulated for Jurkat-CD16/γ transfected cells19 were grown in cul- 24 h with 2 × 105 H1299-anti-CD16 scFv or H1299-anti- ture medium containing 0.4 mg/ml G418. CTLL-2 cells phOx scFv cells, in the presence of 10 ng/ml PMA (kind gift of Dr RS Lee, Institut Curie, Paris, France), were (Sigma, St Louis, MO, USA). Culture plates were then grown in 1640 RPMI medium (Seromed) supplemented centrifuged (800 g, RT, 5 min) and supernatants har- with 10% FCS, 2 mml-glutamine, 100 IU/ml penicillin, vested. The presence of huIL-2 was measured by ELISA 100 ␮g/ml streptomycin, 10 ␮m β-2 mercapto-ethanol with the Quantikine human IL-2 (R&D, Oxford, UK). (Gibco-BRL) and 100 IU/ml recombinant human IL-2 The IL-2 bioactivity was evaluated with the IL-2-depen- (huIL-2) (Chiron, Emeryville, CA, USA). dent mouse line CTLL-2. Cells were cultured in the Mouse fibroblast L929 cells21 were kindly provided by presence of 10 IU/ml huIL-2 for 48 h. After three washes Dr E Tartour (Hoˆpital Europe´en Georges Pompidou, in RPMI 1640 medium, 2 × 104 CTLL-2 cells were cultured Paris, France). They were grown in 1640 RPMI sup- for 48 h in the presence of serial dilutions of co-culture plemented with 10% FCS, 2 mml-glutamine, 1 mm supernatants or of recombinant huIL-2 (from 5 × 10-5 to

Gene Therapy Antitumor immunity with surface anti-FcR scFv N Gruel et al 1727 50 IU). Cell proliferation was measured by the CellTiter Epifluorescence microscopy was performed by cultur- 96 Aqueous Non-Radioactive Proliferation Assay ing 5 × 105 macrophages in two-chamber culture Labtek (Promega, Madison, WI, USA). slides (Nunc) for 16 h. Non-adherent cells were removed and macrophages were washed twice with PBS- activation Ca++Mg++. Macrophages were then incubated on ice for HuPBMC from healthy donors were isolated from 30 min with PBS containing anti-CD16 3G8 mAb (25 cytapheresis rings (obtained from the Etablissement de ␮g/ml) or not. PKH26-labeled H1299-anti-CD16 scFv or Transfusion Sanguine, Hoˆpital de l’Hoˆtel-Dieu, Paris, H1299-anti-phOx scFv cells were washed with PBS- France) by density gradient centrifugation ( Ca++Mg++, sonicated (20 000 Hz, 130 W, 10 s, three times), Separation Medium, d: 1.077, Roche Diagnostics, Meylan, added to macrophages at an E:T ratio of 1:1 and incu- France). Monocytes were isolated by plastic adherence by ° bated for 4 h at 37 C in a 5% CO2 humidified atmosphere. incubating 105 huPBMC in 100 ␮l RPMI 1640 culture ++ ++ ° Cells were then washed twice with PBS-Ca Mg , fixed medium/well of a 96-well microplate at 37 C for 3 h. with PBS-Ca++Mg++-4% formaldehyde, and stained with × 4 Monocytes were then co-cultured with 5 10 H1299- FITC-CD14/CD64 mAbs (Beckman Coulter). Nuclei were anti-CD16 scFv or H1299-anti-phOx scFv cells at 37°C for α stained with Vectashield containing DAPI (Vector) and 15 h. The presence of TNF in culture supernatants was analyzed under an epifluorescence microscope. measured by a cytotoxicity assay with the L929 cell line.21 2 × 104 L929 cells/well were grown overnight in 96 well- microplates in triplicates. Co-culture supernatants or Animals α ␮ recombinant TNF (from 0.1 ng to 37.5 fg) and 2 g/ml Eight-week-old female SCID mice (C.B-17 1cr scid/scid), actinomycin D (Sigma) were then added for a further 20 obtained from IFFA Credo (L’Arbresle, France) were h culture period. The proliferation of L929 cells was mea- ␮ ° maintained under pathogen-free conditions in micro- sured by adding 100 g/well of MTT (Sigma) (4 h, 37 C, isolation cages. Serum Ig levels were quantitated by in the dark) and then 100 ␮l/well of 20% SDS-50% N,N- 40 ° enzyme-linked immunosorbent assay (ELISA). Animals dimethyl formamide (DMF), pH 4.7 (Sigma) (20 h, 37 C, with antibody titers exceeding 1 ␮g/ml were excluded in the dark). The percentage of cytotoxicity was calcu- from the experiments. To eradicate residual NK cells, 25 lated according to the following formula: cytotoxicity (%) ␮ = Ϫ × l of an anti-Asialo GM1 antibody solution (WakoOsaka, 1 (OD550 nm experiment/OD550 nm control) 100. Japan) were administered intraperitoneally 1 day before Cytotoxicity assay tumor-cell grafting and every 5 days up to 25 days after HuPBMC cytotoxicity was determined in a 4 h lactate tumor cell injection. dehydrogenase (LDH) release cytotoxicity assay in tripli- cate, using H1299-anti-CD16 scFv or H1299-anti-phOx scFv cells as target cells. The specific cytotoxicity was cal- Winn assay culated according to the following formula: specific lysis H1299-anti-CD16 scFv and H1299-anti-phOx scFv cells (%) = (experimental – effector spontaneous – target were first adapted for growth in vivo in SCID mice. Cells spontaneous) x 100/target maximun – target spon- were then isolated from subcutaneous tumors and frozen. taneous. They were thawed before each experiment, cultured in H1299-anti-CD16 scFv cells were pre-incubated with vitro for 72 h and analyzed for scFv surface expression different doses of sCD16 for 20 min at RT before adding before engraftment. Groups of six mice were subcutane- × 6 huPBMC in inhibition experiments. ously injected in the left flank with a mixture of 1.5 10 transfected H1299-scFv cells and 50 × 106 huPBMC resus- Tumor binding to and phagocytosis by IFNγ-activated pended in 200 ␮l PBS. Tumors were measured using a macrophages caliper. The longest dimension (a) and perpendicular A two-color immunofluorescence assay was used to width (b) were determined, and tumor volume (v) was determine the binding and phagocytosis of H1299-anti- calculated according to the formula: v = ab2/2. Mouse CD16 scFv or H1299-anti-phOx scFv cells by IFNγ-acti- survival was monitored for 92 days. vated macrophages. Tumor cells were labeled with PKH26 dye (Sigma) according to the manufacturer’s instructions and cultured overnight to eliminate excess dye. 3 × 105 macrophages (100 ␮l of RPMI 1640 culture Acknowledgements medium) were distributed in polypropylene round-bot- tom capped tubes (Becton Dickinson). 100 ␮lof3× 105 The authors wish to thank Dr A Aurias (Institut Curie, labeled cell targets were then added to the tubes. The cell Paris, France) for providing access to fluorescence imag- mixture was incubated at 4°Cor37°C for 4 h. Macro- ing system, Mrs C Chevalier (Institut Curie) for animal phages were then stained with FITC-CD14/CD64 mAbs handling, Dr O Cochet (Exonhit, Paris, France) and Pr R (Beckman Coulter, Brea, CA, USA). Hawkins (Christie CRC Research Centre, Manchester, Fluorescence cell analysis for double staining was per- UK) for their help in the generation of the anti-CD16 scFv formed using a FACSCalibur (Becton Dickinson). FITC- cDNA, Dr A Coronel (Institut Curie) for IFNγ-activated CD14+/CD64+ macrophages also exhibiting red fluor- preparations, Pr F Belardelli (Istituto Super- escence were considered as phagocytic cells or tumor iore di Sanita, Rome, Italy) for advice on SCID mice, and cell/macrophage doublets. Percentage of phagocytosis Mrs Qi Cui for careful English reading of the manuscript. was calculated using the following formula:38,39 (FITC+ This work was supported by grants from the Institut PKH26+ cells (37°C) Ϫ FITC+ PKH26+ cells (4°C)) × Curie, the INSERM, the Fondation pour la Recherche 100/FITC+PKH26- cells (37°C) + FITC+PKH26+ cells Me´dicale (FRM) and the Association pour la Recherche (37°C) Ϫ FITC+PKH26+ cells (4°C). contre le Cancer (ARC).

Gene Therapy Antitumor immunity with surface anti-FcR scFv N Gruel et al 1728 References 20 Vivier E et al. Signaling function of reconstituted CD16:ζ:γ recep- tor complex isoforms. Int Immunol 1992; 4: 1313–1323. 21 Espevik T, Nissen-Meyer J. A highly sensitive cell line, WEHI 1 Hulett MD, Hogarth PM. Molecular basis of func- 164 clone 13, for measuring cytotoxic factor/tumor necrosis fac- tion. Adv Immunol 1994; 57:1–127. tor from human monocytes. J Immunol Meth 1986; 95:99–105. 2 Gosselin EJ et al. Enhanced antigen presentation using human 22 Boyer A et al. Generation of phagocytic MAK and MAC-DC for Fcγ receptor (monocyte/macrophage)-specific immunogens. J therapeutic use: characterization and in vitro functional proper- Immunol 1992; 14: 3477–3481. ties. Exp Hematol 1999; 27: 751–761. 3 Anegon I, Cuturi MC, Trinchieri G, Perussia B. Interaction of Fc 23 Liao KW, Lo YC, Roffler SR. Activation of by anti- receptor (CD16) ligands induces transcription of interleukin 2 CD3 single-chain antibody dimers expressed on the plasma receptor (CD25) and lymphokine genes and expression of their membrane of tumor cells. Gene Therapy 2000; 7: 339–347. products in human natural killer cells. J Exp Med 1988; 167: 24 Huls G et al. Antitumor immune effector mechanisms recruited 452–472. by phage display-derived fully human IgG1 and IgA1 mono- 4 Weiner LM et al. A human tumor xenograft model of therapy clonal antibodies. Cancer Res 1999; 59: 5778–5784. with a bispecific monoclonal antibody targeting c-erbB-2 and 25 Michon J et al. In vitro killing of neuroblastoma cells by neutro- CD16. Cancer Res 1993; 53:94–100. phils derived from colony-stimulating factor- 5 Hombach A et al. A CD16/CD30 bispecific monoclonal antibody treated cancer patients using an anti-disialoganglioside/anti- induces lysis of Hodgkin’s cells by unstimulated natural killer FcγRI bispecific antibody. Blood 1995; 86: 1124–1130. cells in vitro and in vivo. Int J Cancer 1993; 55: 830–836. 26 Regnault A et al.Fcγ receptor-mediated induction of dendritic 6 Ball ED et al. Initial trial of bispecific antibody-mediated immun- cell maturation and major histocompatibility complex class I- otherapy of CD15-bearing tumors: cytotoxicity of human tumor restricted antigen presentation after immune complex internal- cells using a bispecific antibody comprised of anti-CD15 (MoAb ization. J Exp Med 1999; 189: 371–380. PM81) and anti-CD64/FcγRI (MoAb 32). J Hematother 1992; 1: 27 Wallace PK et al. Exogenous antigen targeted to FcγRI on 85–94. myeloid cells is presented in association with MHC class I. J 7 Weiner LM et al. Phase I trial of 2B1, a bispecific monoclonal Immunol Meth 2001; 248: 183–194. antibody targeting c-erbB-2 and FcγRIII. Cancer Res 1995; 55: 28 Van Vugt MJ et al. The FcγRIa (CD64) ligand binding chain trig- 4586–4593. gers major histocompatibility complex class II antigen presen- 8 Hartmann F, Renner C, Jung W, Pfreundschuh M. Anti- tation independently of its associated FcRγ-chain. Blood 1999; 94: CD16/CD30 bispecific antibodies as possible treatment for 808–817. refractory Hodgkin’s disease. Leuk 1998; 31: 385–392. 29 Weiner LM et al. Human interactions of a bispecific 9 Valone FH et al. Phase Ia/Ib trial of bispecific antibody MDX- monoclonal antibody targeting tumor and human FcγRIII. 210 in patients with advanced breast or ovarian cancer that over- Immunol Immunother 1996; 42: 141–150. expresses the proto-oncogene HER-2/neu. J Clin Oncol 1995; 13: 30 Tomasello E, Blery M, Vely F, Vivier E. Signaling pathways 2281–2292. engaged by NK cell receptors: double concerto for activating 10 Gansbacher B et al. Retroviral vector-mediated γ-interferon gene receptors, inhibitory receptors and NK cells. Semin Immunol transfer into tumor cells generates potent and long lasting anti- 2000; 12: 139–147. tumor immunity. Cancer Res 1990; 50: 7820–7825. 31 Mingari MC, Moretta A, Moretta L. Regulation of KIR 11 Guarini A et al. IL-2 gene-transduced human HLA-A2 mela- expression in human T cells: a safety mechanism that may noma cells can generate a specific antitumor cytotoxic T-lym- impair protective T-cell responses. Immunol Today 1998; 19: phocyte response. Cytok Mol Ther 1995; 1:57–64. 153–157. 12 Armstrong TD et al. Major histocompatibility complex class II- 32 Segal DM, Weiner GJ, Weiner LM. Bispecific antibodies in can- transfected tumor cells present endogenous antigen and are cer therapy. Curr Opin Immunol 1999; 11: 558–562. potent inducers of tumor-specific immunity. Proc Natl Acad Sci 33 McCall AM et al. Isolation and characterization of an anti-CD16 USA 1997; 94: 6886–6891. single chain Fv fragment and construction of an anti- 13 Townsend SE, Allison JP. Tumor rejection after direct co-stimu- HER2/neu/anti-CD16 bispecific scFv that triggers CD16-depen- lation of CD8+ T cells by B7-transfected melanoma cells. Science dent tumor cytolysis. Mol Immunol 1999; 36: 433–445. 1993; 259: 368–370. 34 Arndt MAE et al. A bispecific diabody that mediates cytotoxicity against xenotransplantated human Hodg- 14 Imro MA et al. Human melanoma cells transfected with the B7- + kin’s tumors. Blood 1999; 94: 2562–2568. 2 co-stimulatory molecule induce tumor-specific CD8 cytotoxic 35 Moutel S et al. Generation and characterization of a mouse sin- T lymphocytes in vitro. Hum Gene Ther 1998; 9: 1335–1344. gle-chain antibody fragment specific for disialoganglioside 15 Hurvitz AA et al. Enhancement of the antitumor immune (GD2). Hybridoma 1997; 16: 335–346. response using a combination of interferon-γ and B7 expression 36 Chesnut JD et al. Selective isolation of transiently transfected in an experimental mammary carcinoma. Int J Cancer 1998; 77: cells from a mammalian cell population with vectors expressing 107–113. a membrane anchored single-chain antibody. J Immunol Meth 16 Fleit HB, Wright SD, Unkeless JC. Human neutrophil Fc gamma 1996; 193:17–27. receptor distribution and structure. Proc Natl Acad Sci USA 1982; 37 Teillaud JL et al. Natural and recombinant soluble low affinity 79: 3275–3279. Fc␥R: detection, purification and functional activities. Immuno- 17 Cochet O et al. Intracellular targeting of oncogenes: a novel methods 1994; 4:48–64. approach for cancer therapy. In: Marasco WA (ed.). Intrabodies. 38 Ely P et al. Bispecific-armed, IFNγ-primed macrophage- RG Landes Company, Chapman & Hall: New York, 1998, pp mediated phagocytosis of malignant non-Hodgkin’s lymphoma. 129–142. Blood 1996; 87: 3813–3821. 18 Teillaud C et al. Soluble CD16 binds peripheral blood mono- 39 Munn DH, Cheung NK. Phagocytosis of tumor cells by human nuclear cells and inhibits pokeweed-mitogen-induced monocytes cultured in recombinant macrophage colony-stimul- responses. Blood 1993; 882: 3081–3090. ating factor. J Exp Med 1990; 172: 231–237. 19 Vivier E, Ackerly M, Rochet N, Anderson P. Structure and func- 40 Thibaut E et al. Software for the quantitative evaluation of in tion of the CD16:ζ:γ complex expressed on human natural killer vitro monoclonal antibody production from ELISA data. J Immu- cells. Int J Cancer 1992; 7:11–14. nol Meth 1987; 104:15–24.

Gene Therapy