An effective strategy of human tumor vaccine modification by coupling bispecific costimulatory molecules

Claudia Haas,1 Christel Herold-Mende,2 Roswitha Gerhards,3 and Volker Schirrmacher1 1German Cancer Research Center, Tumor Immunology Program, Heidelberg, Germany; 2Department of Neurosurgery, University-Clinic, Heidelberg, Germany; and 3Marien-Hospital, Herne, Germany.

A new, generally applicable procedure is described for the introduction of defined costimulatory molecules into human cancer cells to increase their T-cell stimulatory capacity. The procedure involves infection with Newcastle disease virus to mediate the cell surface binding of costimulatory molecules (e.g., specially designed bispecific antibodies (bsAb)). The modification is independent of tumor cell proliferation and laborious recombinant gene technology and can be applied directly to freshly isolated and ␥-irradiated patient-derived tumor cells as an autologous cancer vaccine. Following the infection of tumor cells with a nonvirulent strain of Newcastle disease virus, the cells are washed and then further modified by coincubation with bsAbs, which attach with one arm to the viral hemagglutinin-neuraminidase (HN) molecule on the infected tumor cells. The second specificity of one bsAb (bs HN ϫ CD28) is directed against CD28 to augment antitumor T-cell responses by selectively channeling positive costimulatory signals via the CD28 pathway. A second bsAb (bs HN ϫ CD3) was produced to deliver T-cell receptor-mediated signals either alone (bsCD3 vaccine) or in combination with anti-CD28 (bsCD3 vaccine plus bsCD28 vaccine). In human T-cell stimulation studies in vitro, the bsCD28 vaccine caused an up-regulation of early (CD69) and late (CD25) T-cell activation markers on CD4 and CD8 T lymphocytes from either normal healthy donors or cancer patients (autologous system) and induced tumor cytostasis in nonmodified bystander tumor cells. In addition, in combination with the bsCD3 vaccine, augmented antitumor cytotoxicity and T-cell proliferative responses were observed. This tumor vaccine modification procedure is highly specific, quick, economic, and has a broad range of clinical applications.

Key words: Cancer vaccine; bispecific antibody; Newcastle disease virus; tumor-associated antigen; T-cell costimulation.

here is now much evidence that an antigen (Ag)- come immunogenic after CD80, CD86, and/or cytokine Tspecific, T-cell-mediated immune response requires, gene transfection, and mice that rejected such transfec- apart from the Ag, additional costimulatory signals.1,2 tants developed systemic protective immunity against These signals are normally provided by professional the nontransfected tumor.2–6 Ag-presenting cells (APCs) such as dendritic cells, mac- B7 transfection, however, often does not suffice to rophages, and activated B lymphocytes but not, for transfer immunogenicity to tumor cells such as human instance, by carcinoma cells, which represent the major- carcinoma cells, even when these cells are antigenic and ity of human tumors and are derived from the non-APC- susceptible to cytotoxic T lymphocytes (CTLs).7 This type cells that form the epithelial layers. In the absence may be due in part to the fact that B7 also interacts with of costimulation, encounters with a tumor-associated Ag a negative regulator of T-cell activation, CTLA-4.8,9 An (TAA) can lead to T-cell inactivation or death rather ideal tumor vaccine should selectively channel costimu- than to proliferation and differentiation into effector latory signals via the CD28 pathway. This is achieved in cells. Modern gene therapy strategies, therefore, aim at the human cancer vaccines described here, in which a transforming carcinoma cells into professional APCs by bispecific (bs) anti-CD28 antibody (Ab) that does not transfecting genes that code for T-cell costimulatory interact with CTLA-4 is used for tumor vaccine modifi- molecules, such as CD80 (B7-1),2,3 CD86 (B7-2), and/or 4,5 cation. The vaccine modification procedure does not cytokines. In support of this concept, some nonimmu- require recombinant gene transfection or long-term cell nogenic transplantable tumor cells were shown to be- culture and selection procedures. It is a new, quick, and economic procedure involving only two steps that are Received April 16, 1998; accepted August 16, 1998. easy to perform: virus infection and bsAb attachment. Address correspondence and reprint requests to Prof. Volker Schirrma- For virus infection, we used a low dose of Newcastle 10,11 cher, Deutsches Krebsforschungs-zentrum, Abteilung 0710, Im Neuenhei- disease virus (NDV), an antineoplastic agent that mer Feld 280, 69120 Heidelberg, Germany. E-mail address: V. was shown to potentiate tumor vaccine T-cell stimula- Schirrmacher@dkfz-heidelberg-de tory capacity without changing the tumor specificity of © 1999 Stockton Press 0929-1903/99/$12.00/ϩ0 the response.12 Postoperative vaccination with NDV-

254 Cancer Gene Therapy, Vol 6, No 3, 1999: pp 254–262 HAAS, HEROLD-MENDE, GERHARDS, ET AL: T-CELL ACTIVATION VIA BISPECIFIC CANCER VACCINE 255

modified autologous tumor cell vaccine (ATV-NDV) Production of bsCD28 vaccine and bsCD3 vaccine was effective in vivo in various animal tumor models to ␥ 13,14 Tumor cells were inactivated with 200 Gy -irradiation via a prevent metastatic spread and to increase survival. 137Cs source (Gammacell 1000, Atomic Energy of Canada), The mechanism of the immunopotentiating effect of this washed, and resuspended in phosphate-buffered saline (1 ϫ 15–17 7 virus was analyzed in detail. Since 1988, a compara- 10 /mL). The cells were incubated for 1 hour at 37°C in a CO2 ble ATV-NDV was developed for clinical application.18,19 incubator in the presence of NDV Ulster (100 hemagglutinat- To obtain a high-quality product with potential clinical ing units) and washed. A total of 1 ϫ 106 NDV-modified tumor efficacy, a number of procedures were developed: mechan- cells (TuN) were incubated with a predetermined saturating ␮ ical dissection and enzymatic digestion to obtain single cells dose of 1 g/mL of the indicated bsAb or mAb and incubated from freshly operated tumors, removal of debris by Percoll for 20 minutes on ice. After washing, the bs vaccine was centrifugation, and removal of tumor-infiltrating leuko- adjusted to the required cell concentration. 20,21 cytes by immunomagnetic beads. Preparation of responder cells The present report summarizes Ͼ3 years of work toward the production, purification, and successful func- Human peripheral blood mononuclear cells (PBMCs) were tional testing in vitro of two new bsAbs. These bsAbs isolated from the heparinized blood of healthy donors or from allow for the production of a bsAb-modified, second- the HNO33, NCH92, and DaT45 patients by Ficoll-Hypaque generation vaccine (bs vaccine), because one arm of the (Pharmacia, Uppsala, Sweden) density gradient centrifugation. For enrichment of peripheral blood lymphocytes (PBLs), the bsAb is directed against the viral hemagglutinin-neura- isolated PBMCs were adjusted to 2 ϫ 106 cells/mL in RPMI minidase (HN) molecule that serves as a universal 1640 and incubated for 1 hour at 37°C in tissue culture flasks anchoring molecule of the ATV-NDV vaccine. The to remove adherent cells. The nonadherent lymphocytes were second specificity is directed either against CD28 to carefully washed twice with prewarmed (37°C) complete me- deliver costimulatory signals via CD28 or against CD3 dium. T cells were separated by the sheep red blood cell (as a positive control) to deliver T-cell receptor (TCR) rosetting technique. The purity of the isolated T cells was complex-mediated signals and to achieve maximal stim- Ͼ95% as determined by immunofluorescence. ulation in combination with anti-CD28. Our data dem- onstrate potent stimulatory capacity on T cells of human Mixed lymphocyte-tumor cell culture (MLTC) carcinoma cells that are cell surface-modified by agonis- PBMCs and bs vaccine were mixed at a 10:1 ratio and seeded tic anti-CD28 and/or anti-CD3 Abs. into 12- or 6-well plates. After the indicated time interval, cells were washed and stained for FACS analysis. For whole blood cultures, the heparinized blood of the HNO33 patient was MATERIALS AND METHODS diluted 1/5 with complete medium plus 2% FCS.

Cell lines and production of bsAbs Cytotoxicity test The parental hybridomas for the generation of quadromas For the generation of CTLs, purified T cells were mixed with were OKT3 (anti-CD3),22 15E8 (anti-CD28),22 and HN 1.4 c 23 the bs vaccine at a 10:1 ratio and incubated for 5 days in (anti-HN) . BsAb were produced by fusing 8-azaguanine- complete medium with 2% FCS without interleukin-2 (IL-2) in resistant hypoxanthine phosphoribosyltransferase-negative an- 24-well plates in a 37°C incubator. Next, T cells were washed ti-CD3 or anti-CD28 hybridomas with iodoacetamide-treated three times, counted, and used as effector cells in a 4-hour (5 mM, 30 minutes, 4°C) anti-HN hybridomas via polyethylene cytotoxicity assay. For target cells, NDV-modified Colo205 glycol and then by selecting quadromas in hypoxanthine/ tumor cells were incubated for 1 hour at 37°C with 100 ␮Ci aminopterin/thymidine medium. Supernatants were tested by 51 Na2 CrO4 (DuPont-New England Nuclear, Bad Homburg, fluorescence-activated cell sorter (FACS) analysis for the Germany). The cells were then washed twice in RPMI 1640 presence of both parental immunoglobulin (Ig) isotypes. and counted. The target cells were added to the preactivated Quadromas were grown in miniperm production modules lymphocytes at the indicated effector to target ratio in round- (Heraeus, Osterode, Germany), harvested Abs were purified bottom microtiter plates. The final volume was 200 ␮L. The over a protein G column, and bsAbs were separated from plates were incubated for 4 hours at 37°C in a CO2 incubator. parental monoclonal Abs (mAbs) by high-performance liquid The mean percentage of specific 51Cr release was calculated chromatography. The mAbs used in this study for FACS from triplicates as described previously.15 analysis were: anti-CD4 phycoerythrin (PE), anti-CD8 PE, anti-CD25 fluorescein isothiocyanate (FITC), and anti-CD69 Tumor cytostasis test FITC (all from PharMingen, Hamburg, Germany). Colon carcinoma cell lines Colo205 and HT29 were used. All cell For the tumor cytostasis assay, HT29 or HNO33 tumor target lines, including the primary cell lines HNO33 (larynx carcino- cells suspended in complete medium with 2% FCS were ma), NCH92 (glioblastoma), and DaT45 (colon carcinoma), pipetted into sterile 96-well, round-bottom plates (2000–3000 were cultured in RPMI 1640 medium (Life Technologies, cells in a volume of 10 ␮L centrally per well) and incubated at Eggenstein, Germany) supplemented with 2 mM L-glutamine, 37°C for 3–6 hours to promote central adherence. In the 50 international units/mL penicillin, and 50 ␮g/mL streptomy- meantime, bs vaccine and lymphocytes were prepared. Subse- cin complete medium plus 10% fetal calf sera (FCS). All tumor quently, bs vaccine and lymphocytes or whole blood were lines expressed major histocompatibility complex class I mol- added to the adherent tumor cells and cocultured at 37°C. In ecules as revealed by immunostaining using mAb W6/32 from the autologous test system (e.g., Fig 3), 2 ϫ 103 bs vaccine Dako (Glostrup, Denmark); for development, the avidin- stimulator cells from the tumors of patients were used in biotin complex Elite Vectastain kit (Boehringer Ingelheim, combination with 2 ϫ 104 PBLs from a whole blood sample of Germany) was used. the same patient that contained 106 PBLs/mL. In the alloge-

Cancer Gene Therapy, Vol 6, No 3, 1999 256 HAAS, HEROLD-MENDE, GERHARDS, ET AL: T-CELL ACTIVATION VIA BISPECIFIC CANCER VACCINE

Figure 1. New strategy of tumor vaccine design: virus infection and coupling of bsAbs. Through such tumor vaccines, T lymphocytes receive microenvironmental signs and signals via (a) the Ag receptor, (b) costimulatory molecules, and (c) cytokines. These signals then have to be interpreted by the T cell in terms of potential danger and response. During the primary response to virus-infected cells, the local production of type I IFNs acts as an adjuvant and augments the intensity of the response.29 In modifying an autologous vaccine, this strategy introduces costimulatory molecules together with TAA on the same cell and thus aims at the most efficient T-cell response to costimulation.26 TCR, TCR complex; IFN-R, type I IFN receptor; TA, tumor-associated Ag; HN, HN molecule.

neic test system (e.g., Fig 2), the numbers of PBL effector cells introduction of the viral HN molecule into the tumor 3 2 (2 ϫ 10 ) and stimulator cells (2 ϫ 10 ) were lower to avoid cell plasma membrane. NDV was used for a virus allostimulation that was too strong. All determinations were modification of live ATV, because of its good cell performed eight times. As soon as the control wells (tumor binding and immunopotentiating properties.10,11 In a cells without lymphocytes) were confluent, the microtiter plates were washed four times with phosphate-buffered saline murine tumor model, immunization with ATV-NDV did to remove effector cells. Next, 200 ␮L of medium and 1 ␮Ci of not change the specificity of the T-cell-mediated antitu- 12 [3H]thymidine were added per well. The incorporated radio- mor response, although it did selectively increase the activity was measured after 16–24 hours, and values from frequency of tumor-specific CTL precursors12 and po- experimental and control wells were compared to calculate the tentiate tumor-specific CTL activity via the induction of percentage of growth inhibition of tumor cells. type I interferons (IFNs)16 and via the HN molecule.15,17 There was no nonspecific mitogenic or adjuvant effect of Immunofluorescence staining and FACS analysis NDV.17 Virus potentiation of tumor vaccine T-cell stim- For double staining, 1 ϫ 106 lymphocytes from MLTCs were ulatory capacity required cell surface binding but not incubated with PE-conjugated anti-CD4 or anti-CD8 mAb and infection.17 We have shown previously that NDV can with FITC-labeled anti-CD69 or anti-CD25 mAb (PharMin- infect freshly isolated noncultured human tumor cells of gen) for 30 minutes on ice and analyzed with a FACScan different origins,18–20 as well as established tumor cell (Becton Dickinson, Heidelberg, Germany) after washing. lines.24 For step 2, two different bsAbs, produced by Dead cells were excluded from the analysis by staining with respective quadromas and purified by high-performance propidium iodide. The mouse IgG2a isotype of anti-CD3 and anti-HN was detected with goat anti-mouse IgG2a-PE; the IgG1 liquid chromatography, were used. While one arm is isotype of anti-CD28 was detected by goat anti-mouse IgG1- directed against the HN molecule, the second arm is PE (Southern Biotechnology Associates, Birmingham, Ala). directed either against the CD3-associated (TCR) com- plex or against CD28. The bsAbs were shown to express the two parental hybridoma derived isotypes and to be RESULTS able to cross-link in the range of 0.01–10 ␮g/mL bsAb 51Cr-labeled TuN with IL-2 preactivated human T lym- Strategy of tumor cell modification phocytes, thereby leading to T-cell-mediated target cell The concept of generating bs tumor vaccines is illus- lysis. In the functional in vitro assays, the bsAbs (1 trated in Figure 1. It consists of two steps: Tumor cells ␮g/mL) were coupled to NDV-infected tumor cells, are first infected according to a previously standardized which were inactivated by ␥-irradiation. After washing, protocol13,15,18 with a low dose of NDV to generate this product, which can be directly applied in vivo,is NDV-infected ATV-NDV or TuN. This leads to the further addressed as bs vaccine (bsCD28 vaccine and

Cancer Gene Therapy, Vol 6, No 3, 1999 HAAS, HEROLD-MENDE, GERHARDS, ET AL: T-CELL ACTIVATION VIA BISPECIFIC CANCER VACCINE 257

Figure 2. Bs vaccine induces an up-regulation of T-cell activation markers (A,B), antitumor cytotoxicity (C), and antitumor cytostasis (D) in normal PBLs of healthy donors. A,B: A total of 107 PBL responder cells were cocultured with 106 Colo205 tumor cells that had been modified as indicated. The up-regulation of T-cell activation markers was determined at different time intervals by respective double staining and FACS analysis. C: Allo PBL-derived effector cells, generated in a 5-day MLTC with the indicated bs vaccine (IV-VI) or controls (I-III), were tested at a 50:1 ratio to Colo205-NDV target cells in a 4-hour 51Cr release assay % specific lysis was calculated after the subtraction of spontaneous 51Cr release. We have shown previously that the cytotoxicity is directed against the tumor and not against NDV Ags.12 *, nonmodified Colo205 cells did not induce significant allocytotoxicity. D: Tumor cytostasis in adherent bystander HT29 colon carcinoma cells was determined after coculture with allo PBLs and bs HT29 vaccine (IV-VI) or controls (I-III) as described in Materials and Methods. The ratio of responder PBLs to bystander tumor cells was 1:1.

bsCD3 vaccine, respectively). Virus infection and bsAb ␣-chain) was examined via respective double staining of binding require minute amounts of reagents, do not re- the T lymphocytes and FACS analysis (Fig 2, A and B). quire cell proliferation or cell culture, take Ͻ2 hours, and The colon carcinoma cell line used (Colo205) caused can be applied to autologous freshly isolated tumor cells. only weak T-cell stimulation (5–16%) with or without NDV modification, although it did express major histo- Bs vaccine-mediated stimulation of normal allogeneic compatibility complex class I alloantigens. The attach- human T cells ment of bsAb led to a strong increase in the T-cell stimulatory capacity of the tumor cells. For instance, We first tested the ability of the bs vaccine to stimulate bsCD28 vaccine (V) caused an up-regulation of CD69 peripheral blood T lymphocytes from normal healthy on 70% of CD4 and 43% of CD8 T cells and an people. Following different times of coincubation, ex- up-regulation of CD25 on 57% of CD4 and 53% of CD8 pression of the early T-cell activation marker CD69 and T cells. BsCD3 vaccine (IV) was only slightly more of the late T-cell activation marker CD25 (IL-2 receptor effective. When used in combination (VI), nearly all T

Cancer Gene Therapy, Vol 6, No 3, 1999 258 HAAS, HEROLD-MENDE, GERHARDS, ET AL: T-CELL ACTIVATION VIA BISPECIFIC CANCER VACCINE cells expressed CD69 and ϳ70% expressed CD25. The gous bs vaccine. As shown in Figure 3, bsCD28 vaccine specificity controls (I-III) reveal the necessary prerequi- (V) was very potent in stimulating the patient-derived T sites for the bs vaccine: (a) virus infection (bsAb did not cells. Both T-cell activation markers (CD69 and CD25) bind to noninfected Tu (I)) and (b) specific binding of were up-regulated on the majority of CD4 T cells (Fig bsAb via HN (the binding of parental mAbs to TuN (III) 3B). CD69 was also up-regulated of the majority of CD8 did not increase T-cell stimulatory properties). T cells, whereas CD25 was expressed by Ͻ50% of cells. ϩ Next, we investigated the ability of the bs vaccine to A stronger stimulatory effect could be seen for CD8 ϩ induce T-cell-mediated antitumor cytotoxicity and tu- CD25 T cells when the bsCD28 vaccine was combined mor cytostasis in MLTC. Purified T cells from healthy with the bsCD3 vaccine (Fig 3A, VI). volunteers were incubated for 5 days with bs vaccine As a further functional assay, we tested the capability ␥ prepared from the Fc receptor-negative colon carci- of the bs vaccine to induce T-cell proliferation (Fig 3C). noma line Colo205. Figure 2C shows that only a combi- This assessment was performed by autologous MLTC nation of both bs vaccine types was able to induce and using whole blood cultures that mimick more closely cell-mediated antitumor cytotoxicity in such cultures the in vivo situation. Patient-derived blood was incu- that was significantly above the controls. Each bs vaccine bated for 3 days with each bs vaccine alone, with a alone could not induce an augmented cytotoxicity. This combination of both vaccines, or with a control vaccine finding fits quite well with the T-cell costimulation as indicated previously. Cell proliferation was then de- hypothesis and demonstrates that coupling bsAb via HN termined by [3H]thymidine uptake. The strongest prolif- directly to the tumor cells was functional in inducing eration was induced when using a combination of both cell-mediated cytotoxicity. We have shown previously bs vaccines. BsCD28 vaccine (V) alone induced a weak that the low dose of NDV used for tumor cell modifica- response and bsCD3 vaccine (IV) a stronger one, but all tion did not lead to antiviral T-cell responses, increase other stimulators tested were ineffective. As shown in the frequency of antitumor CTL precursors, or augment Figure 3D, a tumor cytostatic effect could be induced by CTL, but not NK or lymphokine-activated killer cell, 12 each of the bs vaccines in autologous MLTCs, even with activity. We further tested whether T-cell activation via the bs patient-derived whole blood containing the responder vaccine could lead to a growth inhibitory effect on cells. The use of whole blood cultures could become of nonmodified tumor cells as bystanders in the MLTC practical importance, because it reduces the amount of system. Live HT29 colon carcinoma cells were seeded blood to be taken from a patient for performing func- into a microtiter plate, and allo PBL responder cells and tional analysis. There is hope that the augmented cyto- bs vaccine stimulator cells were added as indicated in static effect against autologous nonmodified tumor cells Figure 2D. After 3–4 days, when the cells in the control that is seen when stimulating with bs vaccine might wells (with tumor cells alone) were grown to confluency, translate into vaccination effects in the clinic. the MLTC experiment was terminated, the nonadherent In Table 1, we have summarized the results we cells (mainly lymphocytes) were removed, and the pro- obtained with lymphocytes from three different cancer liferation rate of the remaining tumor cells was deter- patients upon stimulation in MLTC with bs autologous mined by [3H]thymidine incorporation. Figure 2D shows vaccine: A larynx carcinoma (a), a glioblastoma (grade that in those wells in which the lymphocytes were IV) (b), and a colon carcinoma (Duke’s D stage; ad- stimulated with the bs vaccine, tumor growth inhibition vanced disease with distant metastases) (c). In the latter was significantly higher in comparison with the control case, the bs vaccine was prepared from a surgically wells. No significant differences were seen between removed liver metastasis. As can be seen, no stimulation groups in which either one of the bs vaccines or a was observed when nonmodified autologous tumor cells combination of both was used. Consequently, the re- were tested for their T-cell stimulatory capacity when quirements for the induction of tumor cytostasis were testing for an up-regulation of CD69 or CD25 on either less stringent than those for the induction of antitumor CD4 or CD8 cells by double immunofluorescence flow cytotoxicity. No tumor-inhibitory effects were seen with cytometry (FACS). When NDV-infected autologous tu- stimulator cells modified by bsAb without virus infection mor cells were used, T-cell activation was observed in (I), by virus infection only (II), or by attachment of ϳ25% of the CD4 or CD8 T cells of all three patients. parental mAb (III). When the three autologous NDV-infected vaccines were further modified by the attachment of bs HN ϫ CD28, the lymphocytes from the larynx carcinoma patient were ϩ ϩ Bs vaccine-mediated stimulation of patient-derived stimulated strongly (e.g., 100% CD4 CD25 ), those from the glioblastoma patient somewhat less so (e.g., autologous human T cells ϩ ϩ 75% CD4 CD25 ), and the ones from the colon- Analogous experiments were performed in strictly au- carcinoma patient with liver metastasis to ϳ50%. These tologous MLTCs with tumor cells and PBLs from the results demonstrate the effect of the new bsAb and same tumor patient. Primary tumor cells (HNO33) from suggest that the stimuli are strong enough to overcome a patient suffering from a larynx squamous carcinoma potential suppressive influences (mediated via trans- ␤ (T3N0M0) were isolated, cultured briefly, and modified forming growth factor- , IL-10, prostaglandins, or other according to the standard protocol to produce autolo- factors) from tumor cells such as glioblastoma.

Cancer Gene Therapy, Vol 6, No 3, 1999 HAAS, HEROLD-MENDE, GERHARDS, ET AL: T-CELL ACTIVATION VIA BISPECIFIC CANCER VACCINE 259

Figure 3. Bs vaccine induces an up-regulation of T-cell activation markers (A,B), T-cell proliferation (C), and antitumor cytostasis (D) in autologous tumor patient-derived lymphocytes. A,B: A total of 107 PBLs from a larynx carcinoma patient were cocultured with 106 autologous HNO33 tumor cells prepared from primary tumor cultures and modified as indicated. The up-regulation of T-cell activation markers was determined as described in the legend to Figure 2. C: A total of 20 ␮L of heparinized blood per well from the same patient (containing 2 ϫ 104 PBLs) was cocultured in a 200-␮L volume with 2 ϫ 103 autologous vaccine stimulator cells (I-VI) for 3 days; next, [3H]thymidine uptake was determined. D: Tumor cytostasis in adherent bystander HNO33 primary tumor cells was determined after coculture with autologous whole blood and bs HNO33 vaccine (IV-VI) or controls (I-III) as described in Materials and Methods. These results were obtained using 30 mL of blood (A,B) and 5 mL of blood (C,D). For ethical reasons, no more blood was taken from the patient; therefore, in A and B, only the most relevant experimental groups were tested.

DISCUSSION kines, which can recruit a broad antitumor response in vivo. A large variety of human tumor cells (carcinomas The attachment of defined bsAb to the first generation of different origin, melanoma, glioblastoma, lymphoma, vaccine ATV-NDV represents a new technology that 33 established human tumor cell lines, 40 primary cul- leads to various second-generation vaccines and to new tures, and Ͼ400 freshly isolated patient-derived tumor perspectives for clinical application. The preparation of cells) could be infected by NDV Ulster. In a recent ATV-NDV follows an approved and clinically estab- editorial,24 a new age for therapeutic viruses was pro- lished standard protocol that involves the isolation of posed, and NDV was considered as a very promising and autologous tumor cells from freshly removed tumor safe agent. The RNA virus replicates in the cytoplasm of tissue by mechanical and enzymatic dissociation,18 puri- infected tumor cells without integrating into the DNA. fication,21 and cryopreservation, followed by ␥-irradia- Viral replication was found to be independent of host tion and infection with NDV, avirulent strain Ulster.21 cell proliferation and thus to work perfectly well in The advantages of using NDV are its efficient, selective, noncultured, freshly isolated, and ␥-irradiated cells. Fur- and safe manner of infecting human tumor cells and the thermore, we demonstrated that new NDV Ulster par- concomitant local induction of cytokines and chemo- ticles (virions) produced by infected tumor cells were

Cancer Gene Therapy, Vol 6, No 3, 1999 260 HAAS, HEROLD-MENDE, GERHARDS, ET AL: T-CELL ACTIVATION VIA BISPECIFIC CANCER VACCINE

Table 1. Stimulation of the T Lymphocytes of Cancer Patients with Bs Autologous Tu-NDV Vaccine: Up-regulation of CD69 and CD25

Early (CD69; Day 1) Late (CD25; Day 3)

Tumor and stimulus CD4ϩ CD69ϩ CD8ϩ CD69ϩ CD4ϩ CD25ϩ CD8ϩ CD25ϩ

(a) Larynx carcinoma HN033 (T3N0M0) Tu ϪϪϪϪ TuN ϪϩϪϪ TuN-bsCD28 ϩϩϩϩ ϩϩϩ ϩϩϩϩ ϩϩϩ TuN-bsCD28 ϩ TuN-bsCD3 ϩϩϩϩ ϩϩϩϩ ϩϩϩϩ ϩϩϩϩ (b) Glioblastoma NCH92 (grade IV) Tu ϪϪϪϪ TuN ϪϩϩϪ TuN-bsCD28 ϩϩϩ ϩϩϩ ϩϩϩ ϩ TuN-bsCD28 ϩ TuN-bsCD3 ϩϩϩ ϩϩϩ ϩϩϩϩ ϩϩϩ (c) Colon carcinoma DaT45 (liver met.) (Duke’s D) Tu ϪϪϪϪ TuN ϩϩϩϪϪ TuN-bsCD28 ϩϩ ϩϩϩ ϩϩ ϩϩ TuN-bsCD28 ϩ TuN-bsCD3 ϩϩϩ ϩϩϩϩ ϩϩϩϩ ϩϩϩϩ T lymphocyte-expressing marker: ϩ, Ͻ25%; ϩϩ, 25–50%; ϩϩϩ, 50–75%; ϩϩϩϩ, 75–100%. noninfectious in three different assays, thus demonstrat- vaccine quality and effectivity.21 This finding, taken ing safety.25 together with recent data that also demonstrate a direct The ATV-NDV vaccine was applied intradermally in priming and costimulation2,3 of T cells by B7-transfected clinical phase I and II studies, and the effects of such tumor cells, supports the concept of costimulation of T active specific immunotherapy treatments on postoper- cells by tumor cells as a viable strategy for the generation ative survival were evaluated in high-risk cancer pa- of potent cancer vaccines. tients. Such vaccinations started in Germany as early as All findings that suggest that cell surface binding of 1988 and now involve several hundred patients. Promis- the two reagents, virus and bsAb, is a prerequisite for the ing results were obtained in adjuvant studies of operated functional activity support the hypothesis outlined in colorectal carcinoma20,26 and breast carcinoma.21 For Figure 1. When various noninfectious NDV prepara- advanced disease stages, additional clinical protocols tions (NDV-UV, NDV-Fo, and NDV-HI)17 were tested, were developed for breast and ovarian carcinomas21,27 as a correlation was seen between their CTL-augmenting well as for renal carcinoma,19 in which ATV-NDV function and cell surface binding.17 When nonbinding vaccination was combined with cytokines (recombinant NDV-HI was added to the cultures, no unspecific adju- IL-2 and recombinant IFN-␣) and conventional treat- vant effect was seen.17 Also, bsAb had to bind specifically ment modalities. to the stimulator cells via HN (Figs 2 and 3) so that they The bs vaccine (Fig 1) enables tumor Ag and costimu- could not be washed off during vaccine preparation. lators to be expressed on the same cell surface. This was Many strategies using bsAb for tumor therapy have found to be a prerequisite for an optimal T-cell re- designed reagents that recognize a TAA with one arm sponse28 and suggests that the signals (1–3 in Fig 1) (such as c-erbB-2,32 GD2,33 or CD1934) and the trigger- interact at or near the membrane of the responding cell ing molecules of immune effector cells with the other rather than as autonomous cascades mediating separate arm.32–34 Because most TAAs that are targeted are also effects. In addition, the NDV-induced type I IFN was expressed on some normal cells, unwanted side effects shown to induce bystander T-cell proliferation29 and to may occur, particularly when the bsAb reagents are play an important role in CTL activation.16 The bs applied systemically.35,36 There are several advantages to vaccine may be comparable in immunogenic strength the decision to bind a bsAb first to a tumor vaccine for with tumor cells double transfected with a cytokine gene local application and to the choice of the foreign viral and CD80, which were reported to be strong vaccines HN molecule, which is easily introduced ex vivo into and to induce CD25 on tumor-infiltrating lymphocytes.6 human tumor cells, as a tumor target: (a) the HN is not Although several studies have demonstrated indirect expressed on any normal cells of the cancer patient, (b) priming of T cells by tumor cells via host-derived targeting the bsAb to HN does not interfere with the APCs,30 the activation of primed T cells to tumor- tumors’ TAA expression and presentation, and (c) a specific CTL effector cells was shown to occur directly common anchoring molecule avoids the need to produce via intact viable tumor stimulator cells.31 Thus, tumor bsAb separately for every type of cancer. The specificity cell viability was an important parameter for CTL acti- controls revealed that the bsAb did not bind to nonvirus- vation in vitro and in vivo.31 In recent clinical studies, modified tumor cells and that the parental mAbs could tumor cell viability behaved as an independent factor for not increase vaccine immunogenicity.

Cancer Gene Therapy, Vol 6, No 3, 1999 HAAS, HEROLD-MENDE, GERHARDS, ET AL: T-CELL ACTIVATION VIA BISPECIFIC CANCER VACCINE 261

Results from phase II clinical studies19,21,26,27 suggest high-risk cancer patients to protect against disseminated a benefit for certain cancer patients receiving treatment disease. with ATV-NDV with respect to postoperative recur- Other interesting tools to enhance Ag presentation in rence-free and overall survival. On the basis of the tumor vaccines have recently been introduced and dis- results obtained with bs vaccine in comparison with cussed.45 In one study for instance, clinical responses ATV-NDV in vitro, an augmentation of clinical effectiv- were seen when using tumor lysate-pulsed dendritic ity is to be expected when applying second-generation bs cells, although the identity of the relevant tumor Ags was 46 vaccine instead of first-generation vaccine. The three not known. In our strategy, in contrast with cancer types of autologous tumor vaccines studied thus far gene therapy approaches, the gene product of interest (ATV-NDV, bsCD3 vaccine, and bsCD28 vaccine) can (i.e., the costimulatory molecule) is being generated be applied separately, in combination, or sequentially. once as a universal bs reagent that can be easily attached Depending upon the antigenicity and immunogenicity of to a multitude of different tumor cells via the NDV- the respective tumor type and the expression or lack of derived HN molecule serving as a general membrane- expression of costimulatory signals, different sequences anchoring molecule. of vaccination can be envisaged. If tumor cells are weakly immunogenic, the ATV-NDV vaccine should be used first to expand the respective relevant T-cell clones REFERENCES before starting with the more immunogenic bsCD28 vaccine. If tumor cells are antigenic but lack costimula- 1. Janeway CH, Bottomly K. Signals and signs for lymphocyte tory signals, the bsCD28 vaccine may be used for the first responses. Cell. 1994;76:275–285. 2. Chen L, Ashe S, Brady WA, et al. Costimulation of rounds of immunization. Furthermore, the extent of deliv- antitumor immunity by the B7 counterreceptor for the T ery of costimulatory signals can be adjusted as required by lymphocyte molecules CD28 and CTLA-4. Cell. 1992;71: varying the amount of attached bsAb HN ϫ CD28. 1093–1102. If antigenicity is lacking, the bsCD3 vaccine could be 3. Townsend SE, Allison, JP. Tumor rejection after direct ϩ used first, which would stimulate T cells to express costimulation of CD8 T cells by B7-transfected mela- CD28.37 Subsequently, bsCD28 vaccine could be used noma cells. Science. 1993;259:368–370. for full activation of a T-cell response. In such a situa- 4. Dranoff G, Jaffee E, Lanzenby A, et al. Vaccination with tion, T cells may secrete cytokines, in particular T helper irradiated tumor cells engineered to secrete murine gran- 38 ulocyte-macrophage colony-stimulating factor stimulates type 1 cytokines, which could affect other host cells potent, specific, and long-lasting anti-tumor immunity. (for instance macrophages and endothelial cells) to Proc Natl Acad Sci USA. 1993;90:3539–3543. secrete nitric oxide and thereby contribute to increased 5. Khazaie K, Prifti S, Beckhove P, et al. Persistence of resistance to metastases.39,40 The application of bsCD3 dormant tumor-cells in the bone marrow of tumor-cell- and bsCD28 vaccines could also be useful for reversing vaccinated mice correlates with long-term immunological defective TCR ␨-chain expression in cancer patients, as protection. Proc Natl Acad Sci USA. 1994;91:7430–7434. seen in Hodgkin’s disease.41 In a preclinical severe 6. Cayeux S, Beck C, Aicher A, Do¨rker B, Blankenstein T. combined immunodeficient mouse Hodgkin’s tumor Tumor cells cotransfected with interleukin-7 and B7.1 genes induce CD25 and CD28 on tumor-infiltrating T xenograft model, a cure of disseminated disease was lymphocytes and are strong vaccines. Eur J Immunol. reported recently via an application of CD30 ϫ CD3 and ϩ 1995;25:2325–2331. CD30 ϫ CD28 bsAb in combination with isolated CD4 ϩ 7. Habicht A, Lindauer M, Galumbacher P, et al. Develop- and CD8 human T cells.42 A further application of the ment of immunogenic colorectal cancer cell lines for bs vaccine, which is supported by the data of this study, vaccination: expression of CD80 (B7.1) is not sufficient to could consist of an ex vivo activation and expansion of restore impaired primary T cell activation in vitro. Eur J patient-derived T cells for later reinfusion for adoptive Cancer. 1995;31A:2396–2402. cellular immunotherapy.43 8. Walunas TL, Lenschow D, Bakker CY, et al. CTLA-4 can function as a negative regulator of T cell activation. Recently, it was demonstrated in an animal tumor 44 Immunity. 1994;1:405–413. model that bsCD28 vaccine was better than a B7- 9. Leach DR, Krummel MF, Allison JP. Enhancement of transfected vaccine. It was able to induce tumor-specific antitumor immunity by CTLA-4 blockade. Science. 1996; T-cell responses in vitro and had immunotherapeutic 271:1734–1736. effects in vivo, leading to a regression of established 10. Cassel WA, Garret R. Newcastle disease virus as an tumors. In this study, the human T cells activated with bs antineoplastic agent. Cancer. 1965;18:863. vaccine exerted cytotoxic and cytostatic antitumor func- 11. Csatary L, Eckhardt S, Kukosza I. Attenuated veterinary tions against virus-modified and nonmodified tumor virus vaccine for the treatment of cancer. Cancer Detect cells (Figs 2 and 3). The fine specificity of these reactiv- Prev. 1993;17:619–627. ities at a clonal level will be evaluated in a separate 12. von Hoegen P, Weber E, Schirrmacher V. Modification of tumor cells by a low dose of Newcastle disease virus: study. The strategy of using autologous tumor vaccines, augmentation of the tumor-specific T cell response in the infected by virus and modified by defined bsAbs without absence of an anti-viral response. Eur J Immunol. 1988;18: interfering with TAA expression, thus may represent an 1159–1166. alternative to B7 gene transfection and may provide a 13. Heicappell R, Schirrmacher V, von Hoegen P, Ahlert T, broad spectrum of new applications for the treatment of Appelhans B. Prevention of metastatic spread by postop-

Cancer Gene Therapy, Vol 6, No 3, 1999 262 HAAS, HEROLD-MENDE, GERHARDS, ET AL: T-CELL ACTIVATION VIA BISPECIFIC CANCER VACCINE

erative immunotherapy with virally modified autologous cell proliferation by viruses and type I interferon in vivo. tumor cells. Int J Cancer. 1986;37:569–577. Science. 1993;272:1947–1950. 14. Plaksin D, Progador A, Vadai E, Feldman M, Schirrma- 30. Huang AYC, Golumbek P, Ahmadzadeh M, Jaffee E, cher V, Eisenbach L. Effective antimetastatic melanoma Pardoll D, Levitsky H. Role of bone marrow-derived cells vaccination with tumor cells transfected with MHC genes in presenting MHC class I-restricted tumor antigens. Sci- and/or infected with Newcastle disease virus (NDV). Int J ence. 1994;264:961–965. Cancer. 1994;59:796–801. 31. Schirrmacher V, von Hoegen P. Importance of tumor cell 15. Ertel C, Millar NS, Emmerson PT, Schirrmacher V, von membrane integrity and viability for CTL activation by Hoegen P. Viral hemagglutinin augments peptide-specific cancer vaccines. Vaccine Res. 1993;2:183–196. cytotoxic T-cell responses. Eur J Immunol. 1993;23:2592– 32. Weiner LM, Clark JI, Ring DB, Alpaugh RK. Clinical 2596. development of 2B1, a bispecific murine monoclonal anti- 16. von Hoegen P, Zawatzky R, Schirrmacher V. Modification body targeting c-erbB2 and Fc␥RIII. J Hematother. 1995; of tumor cells by a low dose of Newcastle disease virus: 4:453–456. potentiation of tumor specific cytolytic T cell activity via ␣ ␤ 33. Michon J, Perdereau B, Brixy F, Moutel S, Fridman WH, induction of interferon- / . Cell Immunol. 1990;126:80–90. Teillaud JL. In vivo targeting of human neuroblastoma 17. Schirrmacher V, Haas C, Ertel C. Virus potentiation of xenograft by anti-GD2/anti-Fc␥RI (CD64) bispecific anti- tumor vaccine T cell stimulatory capacity requires cell body. Eur J Cancer. 1995;31 A:631–636. surface binding but not infection. Clin Cancer Res. 1997;3: 34. De Gast GL, van Houten AA, Haagen IA, et al. Clinical 1135–1148. experience with CD3ϫCD19 bispecific antibodies in patients 18. Liebrich W, Schlag P, Manasterski M, et al. In vitro and with B cell malignancies. J Hematother. 1995;4:433–437. clinical characterization of a Newcastle disease virus- 35. Tibben JG, Boerman OC, Claessens RA, et al. Cytokine modified autologous tumor cell vaccine for treatment of release in an ovarian carcinoma patient following intrave- colorectal cancer patients. Eur J Cancer. 1991;27:703–710. Ј 19. Pomer S, Schirrmacher V, Thiele R, Lo¨hrke H, Brkovic D, nous administration of bispecific antibody OC/TR F(ab )2. Staehler G. Tumor response and 4 year survival data of J Natl Cancer Inst. 1993;85:1003–1004. patients with advanced renal cell carcinoma treated with 36. Kroesen BJ, Janssen RA, Buter J, et al. Bispecific mono- autologous tumor vaccine and subcutaneous rIL-2 and clonal antibodies for intravenous treatment of carcinoma IFN-␣. Int J Oncol. 1995;6:947–954 . patients: immunological aspects. J Hematother. 1995;4: 20. Ockert D, Schirrmacher V, Beck N, et al. Newcastle 409–414. disease virus-infected intact autologous tumor cell vaccine 37. Linsley PS, Ledbetter JA. The role of the CD28 receptor for adjuvant active specific immunotherapy of resected during T-cell responses to antigen. Annu Rev Immunol. colorectal carcinoma. Clin Cancer Res. 1996;2:21–28. 1993;11:191–212. 21. Ahlert T, Sauerbrei W, Bastert G, et al. Tumor cell 38. Paul WE, Seder RA. Lymphocyte responses and cytokines. number and viability as quality and efficacy parameters of Cell. 1994;76:241–251. autologous virus modified cancer vaccines. J Clin Oncol. 39. Umansky V, Rocha M, Kru¨ger A, von Hoegen P, Schirr- 1997;15:1354–1366. macher V. In situ activated macrophages are involved in 22. Knapp-Dorkin B, Rieber EP. Leukocyte Typing III: White host resistance to lymphoma metastasis by production of Cell Differentiation Antigens. New York: Oxford University nitric oxide. Int J Oncol. 1995;7:33–40. Press; 1989. 40. Rocha M, Kru¨ger A, van Rooijen N, Schirrmacher V, 23. Iorio RM, Nicholson PM, Bratt MA. Genetic variation Umansky V. Liver endothelial cells participate in T cell- within a neutralizing domain of the haemagglutinin-neur- dependent host resistance to lymphoma metastasis by aminidase glycoprotein of Newcastle disease virus. J Gen production of nitric oxide in vivo. Int J Cancer. 1995;63: Virol. 1986;67:1393–1402. 405–411. 24. Kenney S, Pagano IS. Viruses as oncolytic agents: a new 41. Renner C, Ohnesorge S, Held G, et al. T cells from age for “therapeutic” viruses. J Natl Cancer Inst. 1994;86: patients with Hodgkin‘s disease have a defective T cell 1185–1186. receptor ␨ chain expression that is reversible by T cell 25. Schirrmacher V, Haas C, Bonifer R, Ahlert T, Gerhards R, stimulation with CD3 and CD28. Blood. 1996;88:236–241. Ertel C. Human tumor cell modification by virus infection: 42. Renner S, Bauer S, Sahin U, et al. Cure of disseminated an efficient and safe way to produce cancer vaccine with xenografted human Hodgkin’s tumors by bispecific mono- pleiotropic immune stimulatory properties when using clonal antibodies and human T cells: the role of human Newcastle disease virus. Gene Ther. 1999;6:63–73. T-cell subsets in a preclinical model. Blood. 1996;87:2930– 26. Schlag P, Manasterski M, Gerneth TH, et al. Active 2937. specific immunotherapy with Newcastle-disease-virus- 43. Schirrmacher V, Beckhove P, Kru¨ger A, et al. Effective modified autologous tumor cells following resection of immune rejection of advanced metastasized cancer. Int J liver metastases in colorectal cancer: first evaluation of Oncol. 1995;6:505–521. clinical response of a phase-II trial. Cancer Immunol 44. Guo YJ, Che XY, Shen F, et al. Effective tumor vaccines Immunother. 1992;35:325–330. generated by in vitro modification of tumor cells with 27. Mo¨bus V, Horn S, Sto¨ck M, Schirrmacher V. Tumor cell cytokines and bispecific monoclonal antibodies. Nat Med. vaccination for gynecological tumors. Hybridoma. 1993;12: 1997;3:451–455. 543–547. 45. Lord EM, Frelinger JG. Tumor immunotherapy: cytokines 28. Liu Y, Janeway C, Jr. Cells that present both specific and antigen presentation. Cancer Immunol Immunother. ligand and costimulatory activity are the most efficient 1998;46:75–81. inducers of clonal expansion of normal CD4 T cells. Proc 46. Nestle FO, Alijagic S, Gilliet M, et al. Vaccination of Natl Acad Sci USA. 1992;89:3845–3849. melanoma patients with peptide- or tumor lysate-pulsed 29. Tongh DF, Borrow P, Sprent J. Induction of bystander T dendritic cells. Nat Med. 1998;4:238–332.

Cancer Gene Therapy, Vol 6, No 3, 1999