Introduction
Osteosarcoma is the most common primary bone tumor in childhood and adolescence. With the introduction of multiagent chemotherapy, overall survival has improved to 60–70% [
1]. However, survival rates have remained stagnant, and the prognosis for patients with metastatic or relapsed disease remains poor, with a 5-year overall survival rate of 20% [
2‐
4]. Since the provocative observations of Dr. Coley on bacterial toxins inducing tumor regression [
5], many immunotherapy attempts have been made in soft tissue and bone sarcomas, but success has been very limited [
6,
7]. The EURAMOS-1 clinical trial, which incorporated IFN-α2b, failed to show the benefit [
8], and also, antibody-based immunotherapies have not succeeded in improving outcome including trastuzumab or dinutuximab in clinical trials. Recent whole-genome sequencing (WGS) and molecular profiling studies of osteosarcoma have shown high levels of chromosome structural variations, rearrangements, and mutation clusters that result in significant disease heterogeneity but few clinically approachable alterations [
9,
10]. These studies have yielded insights into aberrant signaling pathways such as PI3K/mTOR, IGF, and Wnt [
11‐
13], but the efficacy of these targeted therapies in unselected high-risk osteosarcoma patients has been limited [
14,
15].
On the other hand, exploiting cytotoxic T cells against osteosarcoma remains a viable alternative. Yet, upregulation of programmed cell death-1 receptor (PD-1) on CD8(+) tumor-infiltrating lymphocytes (TILs) and interaction with its ligands (PD-L1 and PD-L2) in tumor cells are proven immune escape routes to impede anti-tumor activity of T cells against osteosarcoma [
16,
17]. Although immune checkpoint inhibitors (ICIs) have yet to demonstrate their benefit in patients with osteosarcoma (NCT02406781), blockade of PD-1 and PD-L1 interactions showed their potential to improve anti-tumor response in preclinical studies [
18].
Here, we exploit bispecific antibody-directed T cell immunotherapy for osteosarcoma. We choose disialogangliosides (GD2) and human epidermal growth factor receptor-2 (HER2) as candidate target antigens because of their high expression across a number of osteosarcoma cell lines and their proven safety in IgG-mediated treatment of neuroblastomas and breast cancers using IgG monoclonal antibodies, respectively. We previously described T cell engaging bispecific antibodies (T-BsAbs) using sequences of anti-CD3 (huOKT3) and anti-disialoganglioside [GD2] (hu3F8) or anti-epidermal growth factor receptor-2 [HER2] (trastuzumab) antibody structured on IgG-[L]-scFv format with silenced Fc, exerting potent anti-tumor activities [
19,
20]. Anticipating T cells in cancer patients to be suboptimal in cell number and function [
21], arming ex vivo expanded T cells with T-BsAb should improve in vivo efficacy of BsAb treatment and minimize the risk of neurotoxicity or significant cytokine release syndrome (CRS), which was encountered by direct BsAb injection or CAR T cell treatment [
22‐
25]. Here, we test anti-tumor activities of anti-GD2-BsAb and anti-HER2-BsAb against osteosarcoma in vitro and in vivo. In addition, we generate ex vivo armed T cells (EATs) using the anti-GD2-BsAb (GD2-EATs) or anti-HER2-BsAb (HER2-EATs) and evaluate their antitumor efficacy. Furthermore, we incorporate immune checkpoint inhibitors (ICIs), anti-human PD-1 (pembrolizumab), or anti-human PD-L1 (atezolizumab) antibodies to GD2-EAT or HER2-EAT therapy and study the optimal inhibitor and combination schedule in order to improve their anti-tumor response.
Methods
Cell lines
Representative human osteosarcoma cell lines, 143B (ATCC Cat# CRL-8304, RRID:CVCL_3477), U-2 OS (ATCC Cat# HTB-96, RRID:CVCL_0042), MG-63 (ATCC Cat# CRL-1427, RRID:CVCL_0426), HOS (ATCC Cat# CRL-1543, RRID:CVCL_0312), and Saos-2 (ATCC Cat# HTB-85, RRID:CVCL_0548), and osteoblast cell line, hFOB 1.19 (ATCC Cat# CRL-11372, RRID:CVCL_3708), were purchased from ATCC (Manassa VA). All cells were authenticated by short tandem repeats profiling using PowerPlex 1.2 System (Promega, Cat# DC8942), and periodically tested for mycoplasma infection using a commercial kit (Lonza, Cat# LT07-318). The cells were cultured in RPMI1640 (Sigma) supplemented with 10% heat-inactivated fetal bovine serum (FBS, Life Technologies) at 37 °C in a 5% CO2 humidified incubator.
Flow cytometry
For flow cytometric analysis of antigen expression by human osteosarcoma cell lines, cells were harvested, and cell viability was determined. 1 × 106 cells were stained with 1 μg of antigen specific mAbs in a total volume of 100 μL for 30 min at 4 °C. Anti-CD20 chimeric mAb, rituximab, or mouse IgG1 monoclonal antibody was used as isotype control. After washing with PBS, cells were re-incubated with 0.1 µg PE-conjugated goat anti-human IgG Ab (SouthernBiotech Cat# 2040-09, RRID:AB_2795648). For each sample, 20,000 live cells were analyzed using a BD FACS Calibur TM (BD Biosciences, Heidelberg, Germany). Data were analyzed with FlowJo V10 software (FlowJo, RRID:SCR_008520) using geometric mean fluorescence intensity (MFI). The MFI for isotype control antibody was set to 5, and the MFIs for antibody binding were normalized based on isotype control.
Effector cell preparation
Effector peripheral blood mononuclear cells (PBMC) were separated by Ficoll from buffy coats purchased from the New York Blood Center. T cells were purified from PBMC using Pan T cell isolation kit (Miltenyi Biotec, Cat# 130096535). These T cells were activated by CD3/CD28 Dynabeads (Gibco™, Cat# 11132D) for 7 to 14 days in the presence of 30 IU/mL of IL-2 according to manufacturer’s protocol. PBMCs and ATCs were analyzed by FACS for their proportion of CD3(+), CD4(+), CD8(+), and CD56(+) cells.
Cytotoxicity assays (51chromium release assay)
Antibody-dependent T cell-mediated cytotoxicity (ADTC) was assessed by 51Cr release assay, and EC50 was calculated using Sigma Plot software. Tumor cells were labeled with sodium 51Cr chromate (Amersham, Arlington Height, IL) at 100 mCi/106 cells at 37◦C for 1 h. After two washes, tumor cells were plated in a 96-well plate before mixing with activated T cells (ATCs) at decreasing concentrations of T-BsAb. Effector-to-target cell ratio (ET ratio) was 10:1, and cytotoxicity was analyzed after incubation at 37◦C for 4 h. The released 51Cr was measured by a gamma counter (Packed Instrument, Downers Grove, IL). Percentage of specific lysis was calculated using the formula: 100% (experimental cpm—background cpm)/(total cpm—background cpm), where cpm represented counts per minute of 51Cr released. Total release of 51Cr was assessed by lysis with 10% SDS (Sigma, St Louis, Mo, Cat# 71736), and background release was measured in the absence of effector cells and antibodies.
Antibodies
For each BsAb, scFv of huOKT3 was fused to the C-terminus of the light chain of human IgG1 via a C-terminal (G4S)3 linker [
26]. N297A and K322A on Fc were generated with site-directed mutagenesis via primer extension in polymerase chain reactions [
27]. The nucleotide sequence encoding each BsAb was synthesized by GenScript and was subcloned into a mammalian expression vector. Each BsAb was produced using Expi293™ expression system (Thermo Fisher Scientific, Cat# A14635) separately. BsAbs were purified with protein A affinity column chromatography. GD2-BsAb was linked to the carboxyl end of the anti-GD2 hu3F8 IgG1 light chain [
19], and HER2-BsAb linked to the anti-HER2 trastuzumab IgG1 light chain [
20]. Anti-CD33/anti-CD3 BsAb or anti-GPA33/anti-CD3 BsAb was used as a control BsAb for ADTC assay and in vivo animal experiments [
28,
29]. The other BsAbs used in this study were previously described (US patent# 62/896,415). The purity of BsAbs was evaluated by size-exclusion high-performance liquid chromatography (SE-HPLC), and they had high levels of purity (> 90%). The BsAbs remained stable by SDS-PAGE and SEC-HPLC after multiple freeze–thaw cycles. The biochemical data of BsAbs used are summarized in Additional file
1: Table S1.
T cell arming
Ex vivo activated T cells were harvested between day 7 and day 14 and armed with each BsAb for 20 min at room temperature. After incubation, the T cells were washed with PBS twice. Properties of ex vivo bispecific antibody-armed T cells (EATs) were tested for cell surface density of BsAb using flow cytometry and in vitro cytotoxicity against target antigens. BsAb bound to T cell was measured with anti-idiotype antibody for GD2-EATs and anti-human IgG Fc antibody (BioLegend, Cat# 409303, RRID:AB_10900424) for HER2-EATs.
In vivo experiments
All animal experiments were approved by the Memorial Sloan Kettering’s Institutional Animal Care and Use Committee (IACUC) and were executed according to the IACUC guidelines. For in vivo experiments, BALB-
Rag2−/
−IL-2R-
γc-KO (BRG) mice (Taconic Biosciences) were used [
30]. In vivo experiments were performed in 6–10-week-old male mice. Tumor cells were suspended in Matrigel (Corning Corp, Tewksbury MA) and implanted in the flank of BRG mice. Besides tumor cell line xenografts, 3 different patient-derived tumor xenografts (PDXs) positive for both GD2 and HER2 were established from fresh surgical specimens with MSKCC IRB approval. Tumor size was measured using TM900 scanner (Piera, Brussels, BE), and treatment was started when tumor size reached 100 mm
3. Before treatment, mice were randomly assigned to each group. Tumor growth curves and overall survival were analyzed, and the overall survival was defined as the time from the start of treatment to when tumor volume reached 2000 mm
3. To define well-being of mice, CBC analyses, changes in body weight, general activity, physical appearance, and GVHD scoring were monitored. All animal experiments were repeated twice more with different donor’s T cells to ensure that our results were reliable.
Cytokine release assays
Human Th1 cell-released cytokines were analyzed after EAT injection using LEGENDplexTM Human Th1 Panel (Biolegend, Cat# 741035). Five human T cell cytokines including IL-2, IL-6, IL-10, IFN-γ, and TNF-α were analyzed using mouse serum 4 h, 12 h, and 24 h after EAT injection.
Single cell suspension for flow cytometry analysis of tumor
Surgically resected tumor samples were transported in PBS at room temperature and transferred to 50-mL conical tubes with warm medium (RPMI1640 + 10% FBS). Tissues were dissociated to 1–3 mm3 pieces using scalpels with blade and followed by 1-h enzymatic dissociation using 20X Collagenase II (ThermoFisher Scientific, Cat# 17101015), 100X DNase I (ThermoFisher Scientific, Cat# EN0521). Samples were filtered with 70-µm and 40-µm cell strainers, and red blood cells were eliminated using ACK lysis buffer (ThermoFisher Scientific, Cat# A1049201). After centrifugation, cells were resuspended in warm medium and counted to quantify viable cells using Trypan blue.
Flow cytometry of blood and tumor
Peripheral blood and tumors were collected and analyzed by flow cytometry. Anti-human antibodies against CD3 (BioLegend, Cat# 300308, RRID:AB_314044), CD4 (BioLegend, Cat# 357410, RRID:AB_2565662), CD8 (BioLegend, Cat# 300912, RRID:AB_314116), and CD45 (BioLegend, Cat# 304012, RRID:AB_314400) were used to define T cell engraftment and subpopulation, and anti-human PD-1 (BioLegend Cat# 367410, RRID:AB_2566680) and PD-L1 antibodies (BioLegend Cat# 329706, RRID:AB_940368) were used to quantify their expression by T cells and osteosarcoma tumor cells. Stained cells were processed with BD LSRFORTESSA (BD Biosciences, Heidelberg, Germany) and analyzed with FlowJo software (FlowJo, LLC, Ashland, OR).
Immunohistochemical (IHC) staining
Formalin-fixed paraffin-embedded tumor sections were made, and immunohistochemical (IHC) staining for human CD3, CD4, and CD8 T cells was done to confirm T cell infiltration inside tumors. The IHC staining was performed at Molecular Cytology Core Facility of MSKCC using Discovery XT processor (Ventana Medical Systems). Paraffin-embedded tumor sections were deparaffinized with EZPrep buffer (Ventana Medical Systems), antigen retrieval was performed with CC1 buffer (Ventana Medical Systems), and sections were blocked for 30 min with background buffer solution (Innovex). Anti-CD3 (Agilent, Cat# A0452, RRID:AB_2335677, 1.2 μg/mL), anti-CD4 (Ventana Medical Systems Cat# 790-4423, RRID:AB_2335982, 0.5 μg/mL), and anti-CD8 (Ventana Medical Systems Cat# 790-4460, RRID:AB_2335985, 0.07 μg/mL) were applied, and sections were incubated for 5 h, followed by 60-min incubation with biotinylated goat anti-rabbit IgG (Vector Laboratories, Cat# BA-1000, RRID:AB_2313606) at 1:200 dilution. For PD-L1 staining, the sections were pre-treated with Leica Bond ER2 Buffer (Leica Biosystems) for 20 min at 100 °C, stained with PD-L1 rabbit monoclonal antibody (cell signaling, Cat# 29122, 2.5 mg/mL) for 1 h on Leica Bond RX (Leica Biosystems). Control antibody staining was done with biotinylated goat anti-rat IgG (Vector Laboratories, Cat# BA-9400, RRID:AB_2336202). All images were captured from tumor sections using Nikon ECLIPSE Ni-U microscope and NIS-Elements 4.0 imaging software. Antigen-positive cells were counted with Qupath 0.1.2.
GD2 expression by IHC
Fresh-frozen tumor sections were made using Tissue-Tek OCT (Miles Laboratories, Inc, Elkhart, IN) with liquid nitrogen and stored at − 80 °C. The tumor sections were stained with mouse IgG3 mAb 3F8 as previously described [
31]. Stained slides were captured using a Nikon ECLIPSE Ni-U microscope and analyzed, and the tissue staining intensity was compared with positive and negative controls and scored from 0 to 4 according to two components: staining intensity and percentage of positive cells. Each sample was assessed and graded by two independent observers.
Statistics
In vivo anti-tumor effect and cytokine release analyses were compared using area under curve (AUC) and survival curves calculated using GraphPad Prism 8.0. Differences between samples indicated in the figures were tested for statistical significance by two-tailed Student’s t-test for two sets of data, while one-way ANOVA was used to among three or more sets of data. All statistical analyses were performed using GraphPad Prism V.8.0 for Windows (GraphPad Prism, RRID:SCR_002798). P < 0.05 was considered statistically significant. Asterisks indicate that the experimental P-value is statistically significantly different from the associated controls at * P < 0.05; ** P < 0.01; *** P < 0.001, **** P < 0.0001.
Discussion
Osteosarcoma tissues overexpress GD2 and HER2 on their surface, and these antigens targeting strategies have been a subject of great attention. However, clinical trials of anti-HER2 trastuzumab or anti-GD2 dinutuximab for metastatic or refractory osteosarcoma were not successful [
38,
39]. This failure attributed to relatively low expression levels of GD2 or HER2 on osteosarcoma tumor tissues [
40], or inherent insensitivity of this tumor to Fc-dependent immune cytotoxic mechanisms [
41,
42]. In this study, we targeted these antigens using T cell engaging BsAb and studied the anti-tumor effect of GD2-BsAb and HER2-BsAb against osteosarcoma. Both BsAbs successfully directed T cells into tumor tissues and exerted a significant anti-tumor effect. T cells armed with GD2-BsAb or HER2-BsAb showed potent tumor-suppressive effect in a variety of osteosarcoma xenograft mouse models with minimal in vivo toxicities. Moreover, osteosarcoma PDX-bearing mice showed long-term cures after GD2-EATs and HER2-EATs treatment, consistent with their high potency, although xenogeneic GVHD effect or epitope spread among long-term memory T cells in vivo cannot be ruled out [
37,
43,
44]. The use of bispecific murine antibodies in syngeneic mouse models will help address these potential mechanisms of tumor control.
To improve therapeutic efficacy of GD2-EATs and HER2-EATs, combination with PD-1 blockades was tested. CD8(+) TILs in metastatic osteosarcoma tissues overexpressed PD-1, and PD-1/PD-L1 blockades partially improved T cell function, resulting in longer survival with fewer pulmonary metastases in previous studies [
18]. However, how to optimally combine ICIs with other immunotherapies has yet to be determined, given the potential negative impact of the concurrent use of immunotherapeutics [
45,
46]. Our data also showed that concurrently administered anti-PD-1 or anti-PD-L1 had no benefit. Sequentially continuous anti-PD-L1 only did improve the anti-tumor effect of GD2-EATs against osteosarcoma. It suggests that continuous neutralization of PD-L1 may be necessary for optimal synergy with BsAb and T cell immunotherapy.
Although cytotoxic CD8(+) T cells mediate direct tumor cell killing, CD4(+) T helper (TH) cells are also important in tumor cell eradication [
47], as CD4(+) CAR T cells exert significant cytotoxicity comparable to CD8(+) CAR T cells [
48]. According to recently published single-cell analysis data, both CD4(+) TH cells and CD8(+) cytotoxic T cells are equally effective in direct tumor cell killing, and their cytotoxicity is associated with both TH1 and TH2 cytokines, e.g., IFN-γ, TNF-α, IL-15, and IL-13, as confirmed by the expression of master transcription factor genes TBX21 and GATA3 [
49,
50]. In addition, rather than stringent TH1 or TH2 subtypes, the predominant anti-tumor response is dependent on a highly mixed TH1/TH2 function in the same cell, suggesting the activation of BsAb-directed T cells or CAR T cells is a canonical process that leads to a mixed response combining both TH1 and TH2 cytokines together with GM-CSF [
49]. This is consistent with the finding that polyfunctional CAR T cells are highly correlated to objective response of patients [
51]. On the other hand, the ratio of CD4(+) to CD8(+) T cells does have an effect on the anti-tumor activity of CAR T cells [
52,
53]. Furthermore, balanced ratio of CD4(+) and CD8(+) CAR T cells (CD4:CD8 ratio 1:1) seemed to be important for high remission rates in B-ALL [
51]. CD4(+) T cells help CD8(+) T cells differentiate, recruit and expand through IL-2, IL-21 and other cytokines to perform their tumoricidal functions [
54,
55]. CD4(+) T cells in tumorous condition or chronic infection are skewed toward the T follicular helper (TFH) phenotype [
56,
57]. While the majority of exhausted T cells in tumors express intermediate levels of PD-1, TFH cells express high levels of PD-1, not predictive of patient survival [
58,
59]. The in vivo anti-tumor effect was likely dependent on both CD4(+) T cells and CD8(+) T cells, as our serial IHC data (Fig.
1d) did suggest a temporal sequence where the initial arrival of CD4(+) T cells was followed by subsequent infiltration of CD8(+) T cells. In this regard, temporal and spatial distributions of PD-1(+) CD4(+) T cells might be susceptible to concurrently administered anti-PD-1.
On the other hand, the benefit of anti-PD-L1 was predicted by the expression of tumor-associated PD-L1 (B7-H1) by osteosarcoma cell lines. When confronted by tumor targets, EATs produce pro-inflammatory cytokines such as IFN-γ, upregulating PD-L1, which induces T cell apoptosis and inhibits T cell cytotoxicity [
60]. Activation-induced T cell death (AICD), associated with IL-10 and Fas/FasL interaction [
61], is a component of PD-L1-mediated T cell apoptosis and can be prevented by anti-PD-L1, but not by anti-PD-1 [
60].
In addition, surface proteomic analysis of osteosarcoma has identified a wide range of proteins with differential abundance on osteosarcoma cells and human primary osteoblasts including ephrin type-A receptor (EPHA2) [
62]. Using bioinformatics to assess the expression of surface target antigens on osteosarcoma could provide alternative promising strategy to discover new target antigens for T cell immunotherapies including BsAb and CAR [
63,
64].
In conclusion, targeted T cell therapy using GD2-BsAb or HER2-BsAb enabled effective T cell infiltration into tumors and exerted potent anti-tumor activity against osteosarcoma. GD2-EATs and HER2-EATs were also effective to treat osteosarcoma xenografts with reduced toxicity. When GD2-BsAb and HER2-BsAb were combined with anti-PD-L1, tumors had more T cells inside, especially when anti-PD-L1 was continued post-GD2-BsAb treatment. These data strongly support the clinical applicability of GD2- and HER2-BsAbs and the sequentially continuous combination of anti-PD-L1 antibody for the treatment of osteosarcoma.
Acknowledgements
We would like to especially thanks to Irene Cheung, Alan W. Long, Brian Santich, Madelyne Epinosa-Cotton, Mao Wang, See Liang Ng, Tsung-Yi Lin, and Xu Hong for their valuable comments on earlier drafts. Xu Hong designed and validated the anti-HER2 BsAb and analyzed BsAb affinity to CD3 and each target antigen. Hong-fen Guo did HPLC and SDS-PAGE and confirmed the purity and stability of each BsAb. Yi Feng stained fresh-frozen tumor sections.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.