Introduction
Malignant pleural mesothelioma (MPM) is a rare solid organ tumor, which originates from malignant transformed cells of the mesothelium [
1]. A causal link to environmental factors has been established as at least 70% of MPM patients have a definite record of chronic asbestos exposure [
2]. MPM is considered an incurable disease with a median survival of 2 years even when intensive multi-modality treatment is performed [
3‐
5]. Therefore, new therapeutic options are desperately needed.
Adoptive transfer of modified (re-directed) autologous T cells is a promising therapeutic strategy and objective responses were observed in preliminary clinical trials [
6]. Re-directed T cells are (autologous) T cells that are retrovirally transduced to express a chimeric antigen receptor (CAR) specific for a target on the cell surface. CARs contain a single-chain Fv (scFv)-based molecule, which is specific for the target and coupled to T cell-specific signaling moieties such as CD28 and CD3ζ. The general functionality of these CARs is documented widely by recent literature [
7]. The most striking success was observed in B cell lymphoma patients when CD19 was used as an immunological target [
8]. However, also in solid cancer types such as ovarian cancer [
9] and neuroblastoma [
10] adoptive T cell transfer was tested. Recently, we successfully generated peptide-specific re-directed T cells targeting NY-ESO-1 [
11]. We tested different signaling domains of the chimeric antigen receptors (CARs) and could observe
in vivo and
in vitro immunological functionality [
12].
To treat MPM with re-directed T cells, we set out to identify a surface protein that is universally expressed by the majority of MPM subtypes (epithelioid, sarcomatoid and biphasic). Fibroblast activation protein (FAP) was suggested to be a potential target antigen since FAP is widely expressed by various epithelial and mesenchymal cancer types [
13]. FAP expression has been studied extensively by immunohistochemistry in the past [
14] and is known to differ between cell types and even within the tumor tissue. Two patterns of expression are most frequently found: 1) FAP expression by cancer associated fibroblasts (CAFs) of the tumor stroma only (e.g. breast or colorectal cancer [
15]) or 2) by both the tumor stroma and the tumor cells (e.g. sarcoma [
16]). Altogether, FAP is expressed in about 90% of most common cancer types like breast, lung and colorectal cancer [
17]. Its expression is also associated with chronic inflammation, tissue remodeling [
18] and immune modulation in the tumor tissue [
19]. We show here that FAP is expressed in all three major MPM histotypes, namely the epithelioid, sarcomatoid and the intermediate called biphasic.
FAP has been validated as target antigen in oncology by a monoclonal antibody called F19 (humanized version: sibrotuzumab) in different phase I/II clinical trials [
20,
21]. The antibody recognizes exclusively non-degraded human FAP. F19 accumulated specifically in the tumor tissue [
22]; however, the clinical effect was marginal. The results indicated that the sole use of an antibody was not sufficient to induce a meaningful immunological anti-tumor response. Therefore, F19 was not further developed for clinical use [
21].
We developed re-directed T cells with a CAR consisting of a scFv of the FAP-specific F19 antibody, a CD28 signaling domain lacking the lck binding moiety [
23] and a CD3ζ signaling domain. Our rational to develop FAP-specific re-directed T cells based on the F19 antibody was to utilize its already clinically proven specificity to target FAP positive tumor tissue combined with the immunological effector function of T cells. As observed by others our previous results clearly indicated increased antigen-specific function of re-directed T cells when the CAR contained a CD28 signaling domain [
12,
24]. Therefore, we decided to generate a second generation CAR with a co-stimulating signal provided by the CD28 domain. For the first time we show here that re-directed T cells specific for FAP are cytotoxic towards FAP positive targets
in vitro and control xenografted human FAP positive tumors
in vivo.
Material and methods
Cell lines
293T and MSTO-211H were purchased from ATCC (Manassas, VA). HT1080FAP are HT1080 cells stably transfected with human FAP and HT1080PA are mock-transfected HT1080 cells [
25]. T2-1B cells are HLA-A*02:01-positive T2 cells stably transfected with the HLA-A*02:01-restricted NY-ESO-1 peptide 157–165 [
26]. Tumor cell lines were cultivated in standard R10 media (RPMI1640 GlutaMax supplemented with 10% fetal bovine serum (FBS) (v/v), 50 U/ml penicillin and 50 μg/ml streptomycin; all obtained from Invitrogen (Karlsruhe, Germany)). For the culture of transfected HT1080 and T2 cells, 200 μg/ml G418 (Sigma-Aldrich Chemie GmbH, Buchs, Switzerland) or 2.5 μg/ml Hygromycin B (Invitrogen, Karlsruhe, Germany) was added, respectively.
To enable in vivo imaging, HT1080FAP and HT1080PA cells were stably transfected with a D-firefly luciferase encoding plasmid (pGL4.26 plasmid, Promega, Dübendorf, Switzerland which was kindly provided by Martin Pruschy, University Hospital Zurich, Switzerland) using Fugene transfection reagent (Roche Diagnostics GmbH, Mannheim, Germany) according to the manufacturer’s protocol. Forty-eight hours after transfection cells were submitted to selection using 150 μg/ml Hygromycin B. Cells were cloned by limited dilution and luciferase expression was monitored using Bright-Glo™ Luciferase Assay System and a GloMax Microplate Luminometer (both Promega, Madison, WI) according to the manufacturer’s protocol. Clones with luciferase activity underwent two more rounds of limited dilution followed by further testing for luciferase activity. Finally, a stable HT1080FAP-luc and HT1080PA-luc clone were selected that exhibited high luciferase activity and maintained this over months even when cultured in the absence of Hygromycin B. The resultant HT1080FAP-luc and HT1080PA-luc cells were cultivated in standard R10 media supplemented with 200 μg/ml G418 and 150 μg/ml Hygromycin B.
Rheumatoid arthritis synovial fibroblasts originating from tissues obtained during joint replacement surgery (Schulthess Clinic, Zurich, Switzerland) were isolated for cell cultures as described previously [
27]. Briefly, synovial tissues were digested with dispase at 37°C for 60 minutes. After washing, cells were grown in Dulbecco’s MEM NUT MIX-F12 (Invitrogen, Karlsruhe, Germany) supplemented with 10% FBS, 50 U/ml penicillin, 50 μg/ml streptomycin and 10 mM HEPES (Invitrogen, Karlsruhe, Germany).
Antibodies and reagents
Antibodies for flow cytometry were purchased from eBioscience (San Diego, CA) (anti-human CD8a-FITC), Invitrogen (Karlsruhe, Germany) (LIVE⁄DEAD Fixable Aqua Dead Cell Stain Kit) and Southern Biotech (Birmingham, AL) (anti-human IgG-PE). Analysis of FAP expression was performed using the humanized anti-F19 antibody [
20] (kindly provided by Andrew Scott, Ludwig Institute for Cancer Research, Australia), whereas MabThera (Rituximab; anti CD20 mAb; Roche Pharma AG, Reinach, Switzerland) was used as a negative control antibody. Both antibodies were directly labeled with Alexa Fluor 647 (Alexa Fluor
® 647 Antibody Labeling Kit; Invitrogen, Karlsruhe, Germany) according to the manufacturer’s protocol. Murine F19 antibody [
25] (kindly provided by Andrew Scott, Ludwig Institute for Cancer Research, Australia) was used as primary antibody for immunohistochemistry staining, and the murine HD6 monoclonal antibody served as a control antibody [
28]. A biotin-labeled goat anti mouse IgG (Fcγ fragment specific) was used as secondary antibody (Jackson ImmunoResearch, Suffolk, UK). Recombinant human FAP was kindly provided by Andrew Scott, Ludwig Institute for Cancer Research, Australia.
Generation of the FAP-specific chimeric antigen receptor (CAR)
The variant heavy and the variant light chain of humanized F19 [
21] have been converted into a scFv fragment and were flanked with NcoI and BamHI restriction sites. This construct was cloned into the pBullet vector [
29] containing a human ∆CH2/CH3 domain, a ∆CD28 [
23] and a CD3ζ signaling domain (kindly provided by Hinrich Abken, University of Cologne, Germany). The human CH2/CH3 domain has been modified to reduce FcγR binding and thereby minimizing the risk of off-target T cell activation by CAR binding to FcγR
+ cells [
30]. Furthermore, the CD28 binding domain for lck was corrupted by site directed mutagenesis to avoid IL-2 release and subsequent persistence of T
reg cells [
23].
The resulting chimeric antigen receptor construct was termed anti-FAP-F19-∆CD28/CD3ζ. Anti-NY-ESO-1-T1-∆CD28/CD3ζ recognizing the HLA-A*02:01/ NY-ESO-1
157-165 peptide complex served as a control construct [
12]. It displays the same genetic modifications in the downstream CH2/CH3 and CD28 lck domains.
Retroviral transduction of peripheral blood CD8+ T cells
Retroviral transduction of human peripheral CD8
+ T cells was performed as previously described [
12] with following minor modifications. CD8
+ T cells were purified from healthy donor buffy coats using anti-CD8 labeled magnetic beads and MACS technology (Miltenyi, Bergisch Gladbach, Germany). Positive selection with anti-human CD8 microbeads typically resulted in a ≥ 95% pure CD8
+ population. CD8
+ T cells were subsequently activated with CD3/CD28 human T cell expander beads (Invitrogen, Germany) at a bead to cell ratio of 1:5 and 100 IU/ml IL-2 (ImmunoTools, Friesoythe, Germany) in standard R10 media for 48 h. Activated CD8
+ T cells were retrovirally transduced for 48 hours in the presence of 100 IU/ml IL-2 by co-cultivation with 293T cells that were transiently producing high titers of infectious retrovirus carrying the genomic information of the chimeric antigen receptors. CD8
+ T cells were expanded for 2 more days in R10 plus 100 IU/ml IL-2 before harvest and subsequent application in experiments.
Flow cytometry
Flow cytometry was carried out as described previously [
12]. Live/dead staining was performed using the LIVE/DEAD fixable Aqua Dead Cell Stain Kit according to the manufacturer’s protocol. BrdU Proliferation assay were performed with the FITC BrdU flow kit (BD Pharmingen, San Diego, CA) according to the manufacturer’s instruction. Flow cytometric measurements were performed using a FACSCanto II or FACSCalibur machine (BD Biosciences, San Diego, CA). Data were analyzed using FlowJo software (Tree Star, Ashland, OR).
Cytokine assays
Cytokine production was assessed by sandwich ELISA assays. Supernatants of co-cultivated effector and target cells were collected after 24 hours of incubation. IFNγ and IL-2 levels were detected using BD OptEIA set human IFNγ and BD OptEIA set human IL-2 kits, respectively, according to the manufacturer’s instruction (BD Biosciences, San Diego, CA).
Immunohistochemistry
To study potential off-target sites of FAP re-directed T cells, we purchased frozen tissue arrays that contained human adult normal tissue (BioChain, Newark, CA) and stained them for FAP expression. To study expression of FAP in MPM, a total of 9 fresh-frozen samples from malignant pleural mesothelioma patients were retrieved from the biobank of the Institute of Surgical Pathology, University Hospital Zurich.
To perform immunohistochemistry acetone-fixed tissue slides were developed employing the Vectastain ABC Kit Peroxidase Mouse IgG (VECTOR LABORATORIES, Burlingame, CA) according to the manufacturer’s protocol with minor modifications. Briefly, slides were additionally incubated with an avidin and a biotin block (both VECTOR LABORATORIES, Burlingame, CA) for 10 min at RT. Two μg/ml murine F19 or murine HD6 antibody as negative control in 20% of blocking buffer (10% rabbit serum (VECTOR LABORATORIES, Burlingame, CA), 5% milk powder (Roth, Karlsruhe, Germany) in PBS) were added as primary antibody overnight at 4°C. The secondary biotin-goat anti mouse Fc antibody was diluted 1:2000 in 20% of blocking buffer. ImmPACT DAB, SK-4105 (VECTOR LABORATORIES, Burlingame, CA) was used as peroxidase substrate kit according to the manufacturer’s protocol.
The slides were analyzed with a high-speed wide field fluorescence microscope Leica LX and pictures were recorded using a Leica DFC 350 FX camera system (both Leica Microsystems, Heerbrugg, Switzerland).
Europium release assay
Specific cytotoxicity of re-directed T cells was analyzed by a europium release assay (Perkin Elmer, Waltham, MA) as previously described [
12]. HT1080FAP-luc, HT1080PA-luc, MSTO-211H, T2-1B and primary rheumatoid fibroblasts served as target cells for anti-FAP-F19-∆CD28/CD3ζ and anti-NY-ESO-1-T1-∆CD28/CD3ζ re-directed T cells. All cell lines were labeled with 2,2’:6’,2”-terpyridine-6,6”- dicarboxylic acid acetoxymethylester (BATDA) for 30 min at 37°C. Target cells were seeded at a density of 10
4 cells per 96-round bottom-well and co-cultured with effector cells at different effector:target ratios for 4 h at 37°C in R10 media followed by supernatant analyses in a time-resolved Victor
2 flourometer (Perkin Elmer, Waltham, MA). The specific cytolysis of target cells (%) was calculated by the equation: 100x[experimental release(counts) - spontaneous release(counts)]/[maximum release(counts) - spontaneous release(counts)]
Therapy of xenografted FAP+ human tumors
NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ mice, commonly known as NOD.scid.γcKO (NSG), were originally obtained from Jackson Laboratories (Bar Harbor, ME, USA), bred and maintained at the University Hospital Zurich under specific pathogen-free conditions. HT1080FAP-luc tumor cells (106 cells per mouse) were co-injected intraperitoneally with re-directed T cells at an effector to target ratio of 5:1. Tumor burden was monitored by body weight measurements and in vivo bioluminescence imaging. For that purpose, mice were anesthetized with 2% isoflurane and i.p. injected with 150 mg/kg D-Luciferin (Caliper Life Sciences, Hopkinton, MA) and signals were visualized using IVIS®200 Caliper (Caliper Life Sciences, Hopkinton, MA). Monitoring of mice was started immediately after D-Luciferin injection and lasted up to 20 min to obtain the peak photon emission of each animal. Bioluminescence signals were collected and converted to photons/second/cm2/steradian in order to normalize each setting for F-stop, exposure time, binning and animal size using Living Image 3.2 (Caliper Life Sciences, Hopkinton, MA). A constant region of interest was designated around the torso of each animal in order to avoid any incoherence. For imaging purposes, a pseudocolor map representing light intensity was superimposed over a whole-body image. Age-matched mice were used for all experiments. Tumor-bearing mice were euthanized if their body weight increased or decreased more than 15% if compared to the weight of non-diseased, age-matched mice or if mice were confined in their normal behavior by tumor growth. Animal experiments were performed according to Swiss federal and cantonal laws on animal protection.
Statistical analysis
Data were analyzed with Graph Pad Prism version 5.00 for Windows (GraphPad Software, San Diego, CA). Student’s unpaired 2-tailed t-tests were performed between two groups of interest. Survival analysis was performed employing Kaplan Meier survival curves and significance between two variables was calculated using the log-rank test.
Discussion
MPM is still a disease in which even the combination of all three conventional branches of oncologic treatment, surgery, chemotherapy and radiotherapy have to be considered a palliative attempt and other therapeutic options have to be evaluated [
32]. Beside therapies targeting molecular pathways of the malignant cells [
33], immunotherapy is a novel therapeutic strategy to fight MPM [
34]. Most of the so far tested approaches focus on the induction of an active-specific immune response. As an alternative approach, passive adoptive immune therapy could be proposed. Therefore, we investigated the potential therapeutic effect of FAP-specific re-directed T cells as a new option for treatment of MPM. All three histological subtypes of MPM demonstrated FAP expression by the tumor stroma and by most tumor cells. Garin-Chesa and colleagues had demonstrated FAP expression in 7 MPM samples in the tumor stroma as well as on tumor cells with variable FAP expression levels [
13]. T cells expressing anti-FAP CARs at high level were generated by standard procedures and these re-directed T cells were antigen-specific as characterized by their specific
in vitro killing of FAP positive mesothelioma cells. To examine the therapeutic efficacy of FAP-specific re-directed T cells
in vivo, we adoptively transferred re-directed T cells into the peritoneal cavity of mice bearing FAP + tumors. In this setting, FAP-specific re-directed T cells showed strong therapeutic potential, preventing the growth of FAP + tumors and increasing survival of mice.
MPM derives from the mesothelial cell lineage of the serous membranes of the pleura. This process causes a chronic stimulation of the serous membranes resulting in pleural effusion in the majority of clinical cases. Since our first planned clinical protocol (manuscript is published: Petrausch U, Schuberth PC, Hagedorn C, Soltermann A, Tomaszek S, Stahel R, Weder W, Renner C: Re-directed T cells for the treatment of fibroblast activation protein (FAP)-positive malignant pleural mesothelioma (FAPME-1). BMC Cancer 2012, 12:615.) is designed to test the safety of FAP-specific re-directed T cells after injection in the pleural effusion, we developed an intra-peritoneal model for the adoptive transfer of T cells to model the function of these cells in a body cavity. As shown by June and coworkers for mesothelin-specific re-directed T cells [
35], the application of FAP-specific re-directed T cells resulted in the specific lysis of the respective target cells. Since the chosen model system tested human CD8 positive cells in a murine xenograft model with human tumor cells, the potential for allogeneic effects is a major concern. To control for this allogeneic effect, we used our previously, extensively characterized HLA-A*02:01/NY-ESO-1
157-165 peptide-specific re-directed T cells because these cells displayed some allogeneic reactivity [
12]. Importantly, HLA-A*02:01/NY-ESO-1
157-165 peptide-specific re-directed T cells could not control the growth of FAP + tumor cells demonstrating the antigen-specific therapeutic effect of our re-directed FAP-specific T cells.
Adoptive transfer of T cells with a new specificity generated by gene transfer is associated with two major concerns. First, the new specificity can cause on-target off-site effects resulting in T-cell mediated organ damage as documented by recent publications [
36]. Therefore, the exploration of off-site target expression is of crucial interest. In healthy tissue, low-level expression of FAP could only be detected in pancreas, placenta, cervix and uterus. We have not tested tissue from chronically inflamed tissue. Chronic inflammation of tissues causes activation of fibroblasts as part of the remodeling process which occurs during inflammation [
18]. To estimate the effect of re-directed FAP-specific T cells in chronically inflamed tissue, we used activated fibroblasts from patients with rheumatoid arthritis out of the effusion from inflamed joints. As reported, the herein used activated fibroblasts isolated from different donors expressed FAP [
37]. Therefore, the strong and antigen-specific lysis of fibroblasts derived from inflamed tissue by re-directed FAP-specific T cells was expected. This observation indicates the potential for adverse off-site effects in tissues with activated fibroblasts. The F19 CAR cannot be used in murine models of inflammation since the scFv only recognizes the human version of FAP. Since no other model for toxicity estimation is available with the F19 CAR we performed cytotoxicity assay with human fibroblast and implemented strict exclusion criteria for our planed clinical study. All patients with chronic inflammatory diseases have been excluded. Patients with coronary heart disease (CHD), stroke or peripheral vascular disease (PVD) also have to be excluded since expression of FAP was detected in atherosclerotic vessels [
31,
38,
39].
To further minimize potential long-term off-target toxicity we evaluated the therapeutic potential of a second generation CAR with a CD28 co-stimulatory domain. We and other have seen, that re-directed CD8+ T cells with CAR containing a CD28 moiety eliminated target cells more effectively than re-directed T cells with a CD3 domain alone [
12,
40]. CD28 signaling also induces IL-2 production, which leads to expansion of cells [
41]. The herein used CAR lacks the lck domain, avoiding IL-2 induction [
23]. The co-stimulation by 4-1BB also enhances persistence and clonal expansion of re-directed T cells, whereas cytotoxic activity is less impacted [
42]. Based on the most recent literature we designed a FAP-specific CAR causing cell lysis by CD8+ T cells and minimal long-term persistence. Based on the existing experience this allows a clinical phase I trial to first test a CAR recognizing FAP, for which no clinical data are available so far [
6].
The second concern is the gene transfer itself. The clinical course of patients with X-linked immunodeficiency after gene transfer in hematopoietic stem cells resulted in the development of leukemia in 4 out of 9 cases [
43]. Most recently, the safety of retroviral gene transfer in T cells in clinical settings was confirmed by measuring 540 patient years of follow up [
44]. During this time, retroviral gene transfer resulted in no malignant transformation even in cases with very high numbers of integrations per cell (2x10
11 integration site per cell). These recent data suggest that T cells can be safely transduced by retroviral transfer. Additionally, our re-directed FAP-specific T cells showed limited
in vitro survival in the absence of exogenous IL-2 suggesting that retroviral transduction did not result in altered survival control. The limited life span implies a predominant effector phenotype of terminally differentiated T cells as was previously shown for HLA-A*02:01/NY-ESO-1
157-165 re-directed T cells [
12]. These data also suggest that gene transfer did not alter this differentiation program of T cells allowing them to develop in functional effector T cells.
Acknowledgements
We would like to thank Sascha Kleber and Stephan Malzacher for excellent technical assistance. Furthermore, we thank Hinrich Abken and Markus Chmielewski (University of Cologne, Germany) for the pBullet plasmid. We are indebted to Martin Pruschy, Martina Zimmermann and Katrin Orlowski (University Hospital Zurich, Switzerland) for providing the pGL4.26 plasmid and excellent support with in vivo imaging. The here presented pre-clinical experiments for a phase I clinical study testing of FAP-re-directed T cells in MPM were partly planned and designed at the 12th joint ECCO-AACR-EORTC-ESMO Workshop ‘Methods in Clinical Cancer Research’, Waldhaus Flims, Switzerland, 19 – 25 June 2010.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
PCS, SMJ, CR, UP designed the research plan. PCS, UP wrote the manuscript. PCS, CH, PG, OMB performed IHC, killing assays, ELISAs, animal experiments. PCS, UP analyzed data, created the figures, performed statistics. AS performed histology reports. AJ, AM, AS, RS, MvB provided vital reagents/materials/animals. All authors read and approved the final manuscript.