Background
Malignant gliomas, the most lethal brain tumor in adults, account for approximately 13,000 deaths annually in the US [
1]. Long-term prognosis for glioblastoma patients remains poor despite surgery and chemoradiotherapy. Major reasons for treatment failure include its highly infiltrative nature and chemoresistance. Given the limitations of aggressive multimodality treatment, targeted cell therapy is an attractive therapeutic alternative.
Despite the paucity of studies, development of cell therapy for glioblastomas has been encouraging. Arming anti-CD3 activated T cells (ATC) with bispecific antibodies (BiAb) that target the T cell receptor on one hand and the tumor-associated antigen on the other can redirect the non-MHC restricted cytotoxicity of ATC to lyse tumors. Arming
ex vivo expanded T cells with BiAbs may not only improve clinical responses but also minimize toxicity by avoiding the cytokine storm that can occur by systemic infusion of BiAb alone [
2]. Arming ATC with HER2Bi or EGFRBi converts every ATC into a specific cytotoxic T cell [
3‐
7]. Our preclinical studies show that armed ATC can target breast [
6], prostate [
8], ovarian [
5] EGFR+ cancers (head & neck, colorectal, pancreatic, lung [
4], neuroblastomas [
9], and CD20
+ NHL [
7]. ATC armed with HER2Bi were not only able to lyse cancer cells that have high (3+) expression of HER2 but more importantly target and lyse MCF-7 cells that express low or nil HER2 expression [
6] Moreover, armed ATC can kill multiple times, secrete cytokines/chemokines and multiply after engaging tumor cells
in vitro[
10].
In vivo anti-tumor activity of armed ATC when co-injected with tumor cells to prevent the tumor development or when injected intratumorally into xenograft model of prostate cancer, armed ATC persist in Beige/SCID mice for 91 days in the spleen and bone marrow without interleukin-2 (IL-2) support [
8,
11]. Intravenous infusions of armed ATC inhibit tumor growth in the xenograft models in colon [
4] and ovarian cancer [
5]. In our phase I clinical trial involving stage IV breast cancer patients who received activated T cells (ATC) armed with anti-CD3×anti-Her2/
neu bispecific antibody (HER2Bi), high levels of circulating tumoricidal cytokines and specific cytotoxicity by PBMC were observed [
10]. In an earlier trial, using targeted therapy, lymphokine activated killer (LAK) cells armed with chemically heteroconjugated bispecific antibody (anti-CD3MAb x anti-glioma MAb) in 10 patients showed promising clinical results. In 10 patients, 4 patients had regression of tumor and another 4 patients showed histological eradication of remaining tumor cells post surgery with no recurrence in 10–18 months follow-up [
12]. ATC armed with HER2Bi and/or anti-CD3×anti-EGFR (EGFRBi) produced by chemical heteroconjugation of anti-CD3 (OKT3) with trastuzumab or cetuximab, respectively, offers a compelling choice for adjuvant immunotherapy following surgery and chemoradiotherapy.
Although immortalized glioma lines can provide useful biologic insights, cell lines from freshly-resected tumors may more accurately represent the behavior of glioma cells in vivo. In this study, we first established that primary glioma cells can be killed by armed ATC and then addressed further questions of therapeutic relevance: 1) Does dual targeting with BiAbs by mixing individual populations of EGFRBi- and HER2Bi-armed ATC or arming ATC with both BiAbs simultaneously enhance specific cytotoxicity? 2) Can CD133 enriched, CD133− and unfractionated tumor cells be killed differentially by armed ATC? 3) Will armed ATC eliminate a temozolomide (TMZ) resistant subline of U251MG? 4) Do armed ATC continue to kill after being irradiated and in the presence of TMZ? 5) Does binding of armed ATC to glioma cell lines induce the secretion of cytokines?
Methods
Generation and expansion of activated T cells
Anti-CD3-activated ATC were expanded in culture from human peripheral blood mononuclear cells (PBMC) [
6]. Briefly, ATC were produced by activating PBMC with 20 ng/ml of soluble anti-CD3 (OKT3, Ortho Pharmaceutical, Raritan, NJ) and expanded in IL-2 (aldesleukin, Prometheus Laboratories Inc., San Diego, CA) (100 IU/ml) in RPMI 1640 medium supplemented with 10% fetal calf serum (FCS), 2 mM L-glutamine, and 1% penicillin-streptomycin for 14 days. After culture, ATC were harvested, washed, counted, and resuspended in RPMI 1640 for immediate use or cryopreserved.
In experiments where ATC were irradiated, cells received a single 2500 cGy dose using a blood bank cell irradiator (Nordion, Ottawa, Canada) to prevent lymphocyte proliferation [
10] and determine whether cytotoxic activity of BiAb-armed ATC is radioresistant.
Bispecific antibodies and arming of ATC
Preparation and characteristics of BiAbs have been described previously [
3,
6]. HER2Bi was prepared by chemical heteroconjugation of OKT3 and trastuzumab (Herceptin®, Genentech, South San Francisco, CA). EGFRBi was produced by chemical heteroconjugation of OKT3 and cetuximab (Erbitux®, Bristol-Myer Squibb, NY). Anti-CD3×anti-CD20 BiAb (CD20Bi) was made from heteroconjugation of OKT3 and rituximab (Rituxan®, Genentech, South San Francisco, CA) [
7]. ATC were armed with BiAbs at 50 ng/10
6 cells for 1 hour at 4°C, washed, and resuspended in complete RPMI 1640.
Ex vivoprimary glioblastoma lines
Tumor tissue was washed with PBS+EDTA (2 mM), chopped into fragments ≤1 mm, and enzymatically digested using Accumax (Innovative Cell Technologies, San Diego, CA). Fragments of undigested tissue were removed by low g sedimentation and cell clumps were removed by tissue sieves. Contaminating erythrocytes were removed by centrifugation over Ficoll-Hypaque. Viable single cells were counted using trypan blue exclusion. Culture of the
ex vivo adherent differentiated glioma cells was carried out in DMEM-F12 medium (Mediatech, Manassas, VA) supplemented with 10% FCS (Atlanta Biologicals, Atlanta, GA), L-glutamine, and gentamicin (10 μg/ml). Propagation of neurospheres containing cells with stem-like properties was performed in Neurobasal medium (Invitrogen, Carlsbad, CA) containing N-2 and B-27 supplements, human recombinant EGF, and human recombinant basic FGF (each at 20 ng/ml) (PeproTech, Rocky Hill, NJ) [
13].
Long-term glioblastoma lines
Glioma cell lines U87MG, U118MG, and U251MG were also cultured as adherent monolayers in the DMEM-F12-based medium. U87 and U251 cells were grown in 6-well plates in medium supplemented with TMZ over a range of concentrations (10—1000 μM). Medium was changed every 3 days, maintaining the original TMZ concentration. Over 2 weeks, growth of U87MG cells was unaffected, whereas loss of some U251MG cells was recognizable at 10 μM and progressively increased such that a few surviving cells were identified at 333 μM TMZ, but none at 1000 μM. The cells selected in 333 μM TMZ were subsequently propagated in medium containing TMZ (333 μM).
Antibodies, cell separation, and cellular phenotyping
Monoclonal antibodies (cetuximab, trastuzumab, and rituximab) were labeled with the N-hydroxysuccinimide ester of Alexa Fluor 488 (Invitrogen, Carlsbad, CA). These reagents were used for flow cytometry at 1—10 μg/ml. CD133+ cells were isolated using magnetic bead separation (Miltenyi Biotec, Auburn, CA). Frequency of CD133+ cells was determined by flow cytometry using phycoerythrin- (PE) conjugated monoclonal anti-CD133/2 antibodies (Miltenyi Biotec, Auburn, CA).
Viability and cytotoxicity assays
MTT assay
Target cells (4x104 in a volume of 0.1 ml) were plated in 96-well flat bottom microtiter plates (Corning Inc., Corning, NY) and allowed to adhere. Effector cells were added in a volume of 0.1 ml to achieve the effector:target ratio (E:T) indicated and the plates incubated overnight, allowing 16 hours for killing to occur. All groups were performed in triplicate. Non-adherent effector cells were removed and viability of the remaining target cells determined using MTT assay. Metabolism of MTT to the formazan product is a measure of residual viable cells, in contrast to 51Cr release, which is a direct measure of cytotoxicity. We validated the use of the former by confirming a linear relationship between viable cell number and MTT signal and also by performing MTT and 51Cr release in parallel and determining the correlation coefficient of the two data sets.
51Cr Release assay
This was performed in 96-well flat-bottom microtiter plates [
6]. Target cells (4x10
4 cells/well) were plated and allowed to adhere overnight, labeled with
51Cr at 37°C for 4 hours, and then washed
in situ to remove unincorporated isotope. Subsequently, effectors were added to achieve a given E:T.
51Cr release was measured after 18 hours and percent cytotoxicity calculated as follows: (experimental cpm – spontaneous cpm) / (maximum cpm – spontaneous cpm) × 100. Triplicate determinations were performed and the means and standard errors of the triplicates calculated.
Bio-Plex assay for the measurement of cytokine secretion
Cytokines were quantitated in culture supernatants using 25-plex human cytokine Luminex Assay (Invitrogen, Carlsbad, CA) in the Bio-Plex System (Bio-Rad Lab., Hercules, CA). The multiplex panel includes interleukin-1β (IL-1β), IL-1 receptor antagonist (IL-1Ra), IL-2, IL-2R, IL-4, IL-5, IL-6, IL-7, IL-8, IL-13, IL-17, tumor necrosis factor (TNF)-α, interferon (IFN)-α, IFN-γ, granulocyte-macrophage colony-stimulating factor (GM-CSF), macrophage inhibitory protein (MIP)-1α, MIP-1β, interferon-inducible protein (IP)-10, monokine induced by IFN-γ (MIG), eotaxin, regulated on activation normal T cell expressed and secreted (RANTES), and monocyte chemotactic protein (MCP)-1. The limit of detection for these assays is <10 pg/mL based on detectable signal of >2 fold above background. Cytokine concentration was calculated by the Bio-Plex Manager Software using a standard curve derived from recombinant cytokine standards.
Statistical analysis
Calculations of means and standard error of the mean (±SEM), non-parametric correlation tests, and 1- and 2-way ANOVA were performed using Prism5 (GraphPad Software, San Diego, CA). All experiments were repeated at least 3 times.
Discussion
High-grade gliomas are aggressive tumors and respond poorly to all treatment modalities. Since these tumors are highly infiltrative, surgical resection leaves a residuum of cells responsible for recurrence. Malignant gliomas are almost uniformly lethal, with a median survival of about 15 months. This underscores the need for effective, non-toxic strategies that can eliminate the residual tumor cells and also, perhaps, immunize the endogenous immune system against the tumor.
Our immunotherapy approach originates from a series of studies, in which T cells have been redirected to EpCam on adenocarcinomas [
3]; HER2/
neu on prostate [
7], breast [
6], and ovarian cancer [
5]; EGFR on a variety of tumor types [
4]; and CD20 on malignant B cells [
7]. These studies were performed
in vitro, in small animal models, and have progressed to clinical trials [
16]. In a recent study, we have shown the involvement of Granzyme B (GrzB) and IFN-γ signaling pathways in BiAb armed ATC mediated cytotoxicity of target cells [
17].
This study shows that both long-term glioma lines and primary cultures of freshly-resected glioblastoma are efficiently killed by ATC armed with either HER2Bi or EGFRBi, indicating that both antigens are potential targets. We also showed that killing is not enhanced by using both BiAbs simultaneously. It may be that, in future studies, individual gliomas will show differential expression of HER2/neu and EGFR and that it will be prudent to both phenotype and functionally test each tumor as a target, in order to choose the best BiAb.
Cells with the stem-like property of self-renewal and the ability to differentiate into the bulk population of tumor cells have been identified in a number of different solid tumor types, amongst the best characterized of which are breast [
18] and gliomas [
19]. CD133 was the initial marker identified as characterizing the glioma cancer stem cell, although there are subsequent reports of CD133– cells with similar behavior [
20]. The important additional features of stem cells are that they have been postulated to be both chemo- and radioresistant and to be responsible for the extensive infiltration seen in gliomas [
21]. The ability to kill CD133+ and CD133− cells, as shown above, indicates that these stem-like cells may be susceptible to killing in this system.
We showed that TMZ-resistant U251MG cells are also susceptible to targeted killing and that armed ATC still kill in the presence of TMZ. TMZ-resistant U251MG cells were <1% CD133+. This may be due to the length of time this line has been in culture, but their susceptibility does indicate that glioma cells that do not have stem-like properties but acquire chemoresistance are also suitable targets for BiAb-armed ATC. We also demonstrated the radioresistance of armed ATC effector function and an indication that irradiation of ATC may cause an increase in cytotoxicity. One possible interpretation is that there is a radiosensitive population of cells in the ATC that suppresses cytotoxic activity. Whether this involves active suppression or merely reflects death of the radiosensitive cells remains unclear. These results suggest that patients undergoing conventional chemoradiation may be suitable candidates for treatment with armed ATCs.
Some tumor types have been shown to produce soluble factors that inhibit immune effectors [
22] and others to express membrane molecules, such as FasL that actually kill effectors [
23‐
29]. We ruled-out the former by showing that long-term incubation of ATC in culture supernatants from immortalized malignant glioma lines and
ex vivo gliomas and non-neoplastic astrocytes do not inhibit killing activity. We tested the latter, using long-term tumor lines and showed that these, at least, do not diminish the cytotoxic activity of armed ATC. That is, co-culture of armed ATC with glioma cells permits repeat killing [
10].
Finally, we also showed that when armed ATC are incubated with targets, there is increased secretion of three Th1 cytokines (IFN-γ, GM-CSF, and TNF-α) and one Th2 cytokine (IL-13). The armed ATC are the presumed source of the Th1 cytokines and their secretion would serve to activate microglia (IFN-γ), act as an adjuvant for immunization of endogenous lymphocytes (GM-CSF) and potentially augment the killing of target tumor cells (TNF-α). The cellular origin of the IL-13 is probably also the ATC, since CD8+ T cells have been shown to secrete this cytokine [
30]. However, an IL-13 receptor is expressed on some glioma cells [
31] and IL-13 has been shown to be an autocrine growth factor for both Reed-Sternberg cells in Hodgkin’s disease [
32‐
35] and pancreatic cancer [
36]. The potential contribution of the glioma cells to the increased IL-13 is under investigation.
Several studies have shown promising results in glioblastoma using various immunotherapeutic approaches [
37]. Lymphokine-activated killer (LAK) cells generated from PBMC by co-culture with IL-2 have been reported to selectively kill glioma cells
in vitro and when placed into the resection cavity with minimal systemic or neurological side effects [
38,
39]. In the first study in which cells were targeted to WHO grade III/IV gliomas, LAK cells were treated with a conjugate of anti-CD3 cross-linked to the NE-150 monoclonal antibody which recognizes an epitope of NCAM [
40]. In the control group which received untreated LAK cells, 9/10 patients experienced recurrence within 1 year and 8/10 died within 4 years. In the experimental group, 2/10 showed no response, 4/10 showed regression and 4/10 had complete response. No recurrences occurred during 10—18 months of follow-up in the 8 patients showing partial or complete response. Subsequently, others have reported the use of different BiAbs both
in vitro and early-stage clinical trials [
41‐
46].
Competing interests
LGL is founder of Transtarget, Inc. All other authors report no conflict of interest.
Authors’ contributions
IMZ, AT, and ON acquired the data and performed the statistical analysis. IMZ, GRB, LGL, and SM designed the study. All authors interpreted the data and helped draft the manuscript. AT, LGL, and SM revised the manuscript. All authors read and approved the final manuscript.