Background
Prostate cancer (PCa) remains a leading cause of death amongst men, with an expected 191,930 new cases diagnosed in the United States in 2020 [
1], representing 7.1% of global cases [
2]. In the United States alone, 33,300 deaths are forecast for 2020 [
1]. Currently, depending on the stage of disease, treatment options include active surveillance (sometimes for many years), prostatectomy, androgen deprivation therapy, chemotherapy, radiotherapy or a combination of these [
3]. Despite therapy many patients ultimately develop androgen-insensitive PCa metastatic castrate resistant prostate cancer (mCRPC). At this late stage, median survival is less than 3 years [
4]. There is a crucial need for novel effective therapies for PCa.
Bispecific T cell engaging antibodies (BiTEs) show great promise in cancer therapy. BiTEs are able to redirect tumour ignorant T cells to specifically lyse tumour cells in the absence of antigen presentation, therefore overcoming tumour cell evasion tactics such as down-regulation of MHC expression. Blinatumomab, targeting CD19 and CD3, has seen clinical success in the treatment of Philadelphia chromosome-negative relapsed/refractory B cell precursor Acute Lymphocytic Leukemia and received FDA approval in 2014. Since then, there has been a wave of bispecific T cell-engaging constructs in pre-clinical and clinical development, varying in their structure, function and target antigen, designed both for haematological and solid tumours [
5,
6]. In prostate cancer, a PSMA targeting BiTE, Pasotuxizumab, showed good efficacy in a preclinical animal model of PCa [
7,
8]. The molecule is now in Phase I for mCRPC, where some clinical efficacy has been demonstrated (NCT01723475) [
9]. Despite these exciting results demonstrating the clinical potential of a BiTE construct in mCRPC, additional therapies are needed as PSMA is not expressed by all PCa patients [
10]. Moreover, PSMA is expressed in a number of normal tissues, including moderate expression in the salivary glands, which has led to side effects such as xerostomia in PSMA-targeted radiotherapy studies [
11]. These data suggest there is a need for novel BiTE targets for PCa that exhibit an improved expression profile.
Glypican-1 (GPC-1) is a heparan sulfate proteoglycan consisting of a core protein linked to several heparan sulfate chains and attached via a GPI anchor to the cell surface [
12]. Playing a critical role in tumour cell growth, invasion and metastasis as a growth factor co-receptor, GPC-1 is overexpressed in a variety of solid tumours, including PCa [
12,
13]. In line with the role of GPC-1 in tumour progression, expression of GPC-1 increases with increasing Gleason Grade [
13], suggesting its potential as a target for aggressive, difficult to treat lesions, including in mCRPC. Importantly, GPC-1 is not expressed in normal adult tissue [
13,
14], indicating that targeting of GPC-1 may be associated with an improved side effect profile as compared to other antigens. Indeed, several preclinical studies have examined the effect of targeting GPC-1 in animal models showing no adverse effects [
15,
16]. Moreover, a first in human clinical trial using the chimeric anti-GPC-1 antibody Miltuximab
® (Glytherix Ltd) was shown to be safe and well tolerated, exhibiting no drug related adverse events [
17].
Given the strong rationale for the targeting of GPC-1 using a BiTE construct in PCa, we engineered Miltuximab® as a CD3 engaging BiTE. This BiTE demonstrated effective tumour cell killing in PCa cell lines DU-145 and PC3. The release of pro-inflammatory cytokines from T cells activated by MIL-38-CD3 was also observed. An up-regulation of PD-1 on the surface of the activated T cell population suggested the potential for synergy of MIL-38-CD3 therapy with checkpoint inhibition.
Methods
Cell lines
DU-145 and PC3 cells were purchased from ATCC. Jurkat and Raji cells purchased from ATCC, were a generous gift from Dr. Andrew Hutchinson. All cell lines were tested for lineage and mycoplasma contamination quarterly (CellBank Australia, Sydney). DU-145 and Jurkat cells were cultured in Roswell Park Memorial Institute medium (RPMI; Life Sciences) supplemented with 10% v/v Foetal Bovine Serum (FBS; Interpath). PC3 cells were cultured in RPMI supplemented with 20% v/v FBS. Raji cells were cultured in RPMI supplemented with 10% v/v heat inactivated FBS (HI FBS; 56 °C, 30 min).
MIL-38-CD3 BiTE
A gene sequence encoding the MIL-38-CD3 bispecific antibody was generated by Geneart (Life Technologies). The MIL-38-CD3 construct was designed as a tandem scFv consisting of anti-GPC-1 scFv (Miltuximab
® sequence; GTL Ltd.) linked to anti-CD3 scFv (derived from Blinatumomab sequence; Micromet;
http://www.drugbank.ca/drugs/DB09052) by a glycine serine linker. A signal peptide was included to enable secretion of the construct from Chinese Hamster Ovary (CHO) cells into the cell culture medium [
18]. A 6xHis tag was included at the N-terminal of the sequence for purification, and a c-myc tag at the C-terminal (EQKLISEEDLN) was also included as a detection tag for Western blotting, ELISA and flow cytometry.
Additional control constructs were designed including
1.
CD3-PEG as a positive control for CD3 binding. The construct targets CD3 and polyethylene glycol (PEG) moeity as a target not expressed in human cells
2.
J591-PEG as a negative control for CD3 binding with one arm specific for PSMA (J591 sequence), and one arm specific for PEG, thus not binding CD3.
Both constructs contained anti-PEG gene sequences from patent US20120015380A1 which are used by our UQ collaborators for targeted delivery of PEGylated nanomaterials to cancer and immune cells [
18]. CD3 sequences were derived from Blinatumomab sequence and J591 (anti-PSMA) Vh and Vl gene sequences were derived from the patent WO2003024388A2.
The MIL-38-CD3 and other construct genes were cloned into the pcDNA 3.1 mammalian expression vector using HindIII and Not1 restriction sites and standard ligation based cloning methodology and confirmed by DNA sequencing at the Australian Genome Research Facility (AGRF). Larger quantities of DNA required for transient transfections were prepared from 400 ml of transformed E. coli (top10) using DNA Midiprep kits (Macherey Nagal). Final precipitated DNA was resuspended in 1 ml of sterile water, concentration determined at A260 using Nanodrop 1000 and stored at − 20 °C for transient transfections.
Suspension-adapted CHO cells were transiently transfected with DNA encoding MIL-38-CD3 (and other BiTE controls) using PeiMax transfection reagent (a total of 3 × 108 cells in 100 ml; 3 × 106 cells/ml). Transfected cells were cultured for 12–14 days supplemented with specialised feed medium (starting density 1-2 × 106 cells/ml, final density 6 × 106 cells/ml), at a temperature of 32 °C, 7.5% CO2 with shaking at 130 rpm. The expression and secretion of proteins of interest in the medium was confirmed by western blotting using the HRP labelled anti-myc antibody (Miltenyi Biotec; clone SH1-26E7.1.3) to detect the C-terminal myc tag. This also confirmed expression of full length protein. Transfections were stopped when cell viability, measured by trypan blue staining was below 60%. The cells were then centrifuged at 4750 rpm for 10mins and the supernatant was collected and filtered through a 0.2 μm PES filter (Millipore). The filtered supernatant was loaded at 5 ml/min over a 5 ml HisTrap Excel column (GE Healthcare; Australia), equilibrated in 20 mM sodium phosphate, 500 mM NaCl pH 7.4. The bound protein was eluted in 15 ml of 20 mM sodium phosphate, 500 mM NaCl 500 mM Imidazole pH 7.4 buffer, and buffer exchanged into 20 mM sodium phosphate, 500 mM NaCl pH 7.4 using a HiPrep 26/10 Column (GE Healthcare; Australia). Protein concentration was determined by NanoDrop at Abs 280 nm and the extinction coefficient of the protein.
In vitro characterization of MIL-38-CD3
Flow cytometry
Cells were collected from culture and incubated with MIL-38-CD3 (10 μg/ml) for 45 min on ice. Cells were washed 3 times with PBS/2% HI FBS (Flow cytometry staining wash; FSW) then stained with anti-cmyc-FITC (Miltenyi Biotec, Australia; clone SH1-26E7.1.3) or anti-his FITC or PE (Miltenyi Biotec, Australia; clone GG11-8F3.5.1) for 45 min on ice. Cells were washed 3 times in FSW and then acquired on the BD Fortessa X20 (BD, Australia) using FACS DIVA software (BD, Australia). The viability dye topro-3-iodide (Life Technologies; Australia) was used to restrict analysis to viable cells. This dye was added to a final concentration of 1 μM immediately prior to acquisition. Data were analysed in FCS Express V5 (De Novo Software, USA).
ELISA
The MIL38-CD3 construct alongside other controls were tested for binding to recombinant human GPC-1 (Rnd Systems) and human CD3 delta and CD3 epsilon (Abcam) using ELISA. For ELISA, Maxisorp plates (Nunc) were coated with 1 μg/ml of rGPC-1 or hCD3 diluted in PBS for 20 h at 4 °C. Each well was blocked with 200 μl 2% skim milk in PBST (PBS+ 0.05% Tween 20) for 1 h at room temperature. The blocker was then decanted and 100 μl of test construct samples (supernatant or purified sample) were added to wells. Purified J591-PEG (binds to PSMA and PEG) was used as a negative control. CD3-PEG (binds hCD3 and PEG) was used as a positive control for CD3 binding. The samples and controls were incubated for 2 h at room temperature, then washed 3 times with PBST (200 μl per wash), then 100 μl HRP labelled anti-c-myc antibody (Miltenyi Biotech, diluted 1/5000 in block solution) was added to plates for 1 h, then the plate was washed 3 times with PBST. The substrate TMB (Sigma; 100 μl per well) was added to wells and incubated for 15mins in the dark to yield a colorimetric reaction. The reaction was stopped with 2 M sulphuric acid (100 μl per well) and absorbances were recorded at A450nm using the SpectraMax plate reader.
Western blot
Protein samples were heated at 95 °C for 5 mins in NuPAGE sample loading buffer (Invitrogen) with DTT reducing agent (Invitogen) and run on a 4–12% Bis-Tris PAGE (Invitrogen) in 1xMES Buffer for 30mins at 200 V. Proteins were transferred to PVDF membranes using Transblot System (Biorad) set on a mixed size program for 7 mins. Membranes were blocked in 2%MILK-PBST for 1 h at room temp. Ten milliliter of HRP labelled anti-c-myc antibody (Miltenyi Biotech, diluted 1/5000 in block solution) was added to membranes for 1 h, then membranes were washed in PBST 3x5mins. Ten milliliter of ECL reagent (Invitrogen) was added to the membrane for 2 mins and chemiluminesence visualised using the Biorad Gel Doc System.
BiaCore analysis
The determination of binding and affinity of MIL38-CD3 construct for GPC1 and CD3 was determined using the Biacore T200 platform in collaboration with the National Biologics Facility (NBF).
Immobilisation pH scouting for GPC-1
For initial immobilisation pH scouting experiments recombinant human GPC-1 (R&D Systems) was diluted 1 in 50 in acetate buffer, pH 4, 4.5, 5, 5.5 to 100 μg/ml and tested for binding to CM5 chip surface. Binding was similar under all conditions, and so pH 5.5 was chosen since it is a milder buffer. 3700RU of GPC-1 was immobilized on flowcell and compared to blank flowcell. MIL38-CD3 construct was 2x serially diluted from 50 μg/ml to 0.375 μg/ml (8 dilutions in total) and injected for 30s at 30 μl/min. 2000 RU bound to the GPC-1 derived surface when 50 μg/ml test construct was injected. Both 3 M MgCl2 and 10 mM Glycine-HCL pH 1.7 were tested as regeneration buffer and it was found that 10 mM glycine was able to regenerate the chip surface well so this was used for the kinetics experiments.
MIL-38-CD3 to GPC-1 kinetics
For the kinetics study, 100RU of GPC-1 was immobilized on the flowcell, and compared to a second flow cell as blank control. Five concentrations of MIL-38-CD3 were prepared for injection, 50 μg/ml, 16.67 μg/ml, 5.55 μg/ml, 1.85 μg/ml, 0.62 μg/ml. Chips were regenerated with 10 mM glycine at pH 1.7. Data analysis was generated using 1:1 model.
CD3-fc to MIL-38-CD3 kinetics
MIL-38-CD3 was captured on GPC-1 immobilised on the flowcell and 5 concentrations of human CD3-fc were prepared for injection, 50 μg/ml, 16.67 μg/ml, 5.55 μg/ml, 1.85 μg/ml, 0.62 μg/ml. 10 mM glycine at pH 1.7 was used as regeneration buffer. Data was analysed using both a bivalent analyte model and a steady state affinity model.
Preparation of purified peripheral blood T cells
Human buffy coats were obtained from the Australian Red Cross Life Blood under a Material Supply Agreement. The Australian Red Cross Life Blood obtained consent for the use of donor samples, and the use of these samples for the current study was approved by the Australian Red Cross Life Blood (approval number 17-05NSW-13). Human ethics approval was obtained from Macquarie University Human Ethics Committee (approval number 5201700069). To obtain the peripheral blood mononuclear cell (PBMC) fraction, buffy coats were diluted 1:1 in room temperature (RT) RPMI and then 4 ml of diluted buffy coat was overlayed on 3 ml Ficoll Paque (RT; GE Healthcare, Australia). Overlays were centrifuged at 400×g, 35 min, with brakes and acceleration off. The plasma layer at the top was removed by pipetting using a transfer pipette, and the leukocyte layer at the interface of the ficoll and the plasma was removed to a new 15 ml falcon tube. Cells were topped up with 10x volume of RT RPMI and centrifuged at 400×g, 20 min to collect. Cell pellets from the same donor were combined and cells were then resuspended in 10 ml RPMI and centrifuged at 150×g for 10 min to remove platelets. The supernatant was carefully poured off and PBMCs were counted. PBMCs were resuspended in HI FBS (90% v/v) + DMSO (10% v/v) at 1-3 × 107 cells/ml/tube for storage and aliquoted into cryovials. Cryovials were frozen in CoolCells or isopropanol at − 80 °C and then moved to liquid nitrogen at 24 h post freezing. To resuscitate PBMCs, cells were thawed rapidly in a water bath and immediately rinsed with pre-warmed (37 °C) RPMI supplemented with HI FBS (10% v/v; PBMC media) at 300×g, 5 min. To enrich for T cells by negative selection, the human Pan T cell isolation kit II (Miltenyi Biotec, Australia) was used according to the manufacturer’s instructions.
In vitro T cell activation assays and phenotyping
Enriched T cells were prepared at densities of 2 × 106 cells/ml in PBMC media. DU-145 cells were collected by rinsing with PBS and then incubation with PBS/EDTA 2 mM for 15 min at 37 °C/5%CO2. Cells were then prepared at a density of 2 × 105 cells/ml in PBMC media. Constructs (MIL-38-CD3 or control BiTE CD3-PEG, were prepared at 11x final concentration for addition to the plate in PBMC media. For positive control of T cell activation, 96 well flat bottomed tissue culture plates (Falcon; FAL353072) were coated with anti-CD3 on some wells. Anti-human CD3 functional grade (EBioscience; OKT3; 16–0037-85; lot 4,308,895; 5 μg/ml diluted in sterile PBS) was coated on wells of the plate for 2 h at 37 °C or O/N at 4 °C. The wells were then washed twice (200 μl) with sterile PBS. Construct (MIL-38-CD3 or control BiTE) were plated first (10 μl), followed by test cell line (50 μl; 1 × 104 cells) and PBMCs or T cells (50 μl; 1 × 105 cells) to a final Effector:Target ratio of 10:1. Cells were incubated for 24 h for measurement of CD69 and CD25 levels or 72 h for cytokine release assays, in culture (37 °C/5%CO2).
To measure CD69 and CD25 levels, cells were collected by gentle scraping and stained with a cocktail of antibodies against CD3 (BD; UCHT1), CD4 (BD; RPA-T4), CD8 (BD; RPA-T8), CD25 (BD; M-A251), CD69 (BD; FN50) for 45 min on ice, protected from light. Cells were then washed three times with FSW and resuspended in FSW before addition of viability dye Topro-3-Iodide (Life Technologies, Australia).
To measure cytokine release, culture supernatants were collected after centrifugation of the plate at 72 h (300×g, 5 min). Supernatants were frozen at − 20 °C until analysis. For analysis, 50 μl cell supernatant was measured using the human TNF ELISA kit (BD, Australia) or human IFN-γ ELISA it (BD, Australia), according to manufacturer’s instructions.
For measurement of PD-1 expression, T cells were co-cultured with DU-145 tumour cells and MIL-38-CD3 for 48 h, before T cells were harvested by gentle scraping and pelleted (300×g, 5 min). Cells were incubated with anti-PD-1-PE (BD; clone M1H4) for 45 min on ice, then washed three times and resuspended in FSW, before addition of viability dye topro-3-iodide (Life Technologies, Australia). Expression of PD-1 was then measured by acquisition on the BD Fortessa X20 (BD, Australia).
In vitro cell killing assays
Target tumour cells were labelled with Vybrant DIO (Life Technologies, Australia) for 18 min in RPMI at 37 °C in a 15 ml falcon tube. Cells were then centrifuged (180×g, 5 min for DU-145 and C3 cells, or 300×g, 5 min for Raji) and the supernatant was discarded. Enriched T cells were resuspended in 10 ml PBMC media and centrifuged. Cells were then prepared at a density of 4 × 105 cells/ml in PBMC media. Test antibody construct was plated in 96 well flat bottomed tissue culture plates in a volume of 10 μl at the appropriate concentration, followed by 50 μl tumour cell suspension, followed by 50 μl T cell suspension. Cells were incubated for 72 h. The plate was centrifuged (300×g, 5 min), and the supernatant removed by pipetting and discarded. The plate was then washed twice with PBS (200 μl/well; 300×g, 5 min) and each time the supernatant was discarded. Cells were detached from the plate by incubation with Trypsin 0.25% EDTA in RPMI (Sigma Aldrich, Australia) for 3 min at 37 °C/5%CO2. Detachment of cells was confirmed by microscopy. Trypsinisation was stopped by addition of an equivalent amount of R10 containing Topro-3-iodide (Life Technologies, Australia; 0.3 μM; final concentration per well 0.15 μM) to the plate. The plate was then kept on ice until analysis. Cells were then acquired on the BD Fortessa X20 (BD, Australia) and analysis was performed in FCS Express Version 5 (De Novo Software, USA). For analysis, cells were first gated on forward and side scatter to exclude debris, but to include dead cells. Single cells were identified using forward scatter height and area. Analysis was then restricted to FITC+ tumour cells, and then Topro-3-Iodide (APC) signal within the tumour cell gate was measured as a proportion of tumour cells.
Statistics
All statistical tests between two groups were performed using a two-tailed unpaired t test in GraphPad Prism 5.0.
Discussion
The overexpression of GPC-1 in PCa, and its restricted expression in normal adult tissue, make GPC-1 an attractive therapeutic target in an indication where there is an unmet need for novel therapies. Clinical stage anti-GPC-1 antibody Miltuximab
® has proven safe in a first in human study. There is clinical rationale for the therapeutic use of a BiTE construct in PCa [
9]. Thus, we engineered a BiTE construct using the GPC-1 binding sequence of Miltuximab
® and demonstrated effective and specific T cell activation and PCa cell lysis in two phenotypically distinct GPC-1
high PCa cell lines, namely DU-145 and PC3 (derived from a PCa brain metastasis, and bone metastasis, respectively). Importantly, the affinity of MIL-38-CD3 for CD3 was 1000 fold lower than for GPC-1, an important feature for a BiTE, to minimise T cell binding in the absence of tumour cell recognition, and resultant toxicity [
24]. These data support the clinical development of MIL-38-CD3 for PCa. Potentially promoting an inflammatory environment in the tumour, we report the release of inflammatory cytokines TNF and IFN-γ from the T cell population. This activated population up-regulated PD-1, suggestive of potential for combination therapies with checkpoint inhibition.
Prostate cancer is considered an “immunologically cold” i.e. not T cell-inflamed, tumour, with few infiltrating immune cells and a paucity of inflammatory cytokines in the local tumour environment [
25]. A down-regulation in MHC I expression has been reported in PCa cell lines as well as clinical specimens [
26,
27]. Thus, checkpoint inhibitor molecules, have, historically been ineffective in the therapy of PCa in a majority of patients [
23]. The use of a BiTE overcomes these obstacles, driving T cell activation without requirement for tumour specific antigen priming [
28], while increasing T cell derived local inflammatory cytokine levels such as TNF and IFN-y to support an inflammatory environment and drive further T cell homing [
21], local immune cell activation and increased MHC I expression [
22]. Here, we demonstrate the potential of a GPC-1 targeted BiTE to achieve this goal. There is clear potential for the use of BiTE constructs as therapeutics for PCa [
9], but one wonders whether the combination of checkpoint inhibition with BiTE therapy could offer improved efficacy, and provide some clinical utility for checkpoint inhibitors in this traditionally non-responsive cancer type. This idea of combining checkpoint inhibition with BiTE therapy has been tested preclinically in many studies, demonstrating synergy between checkpoint inhibition and BiTE therapy [
28]. In the current study, we found an up-regulation of PD-1 on the surface of MIL-38-CD3 activated T cells, suggesting that a PD-1 inhibitor could extend duration of T cell activation, ultimately improving efficacy. Moreover, the inflammatory environment induced by MIL-38-CD3 at the tumour site may enhance the activity of a checkpoint inhibitor.
The current study aimed to evaluate the ability of MIL-38-CD3 to mediate PCa cell death in vitro. For these studies, two phenotypically distinct PCa cell lines were chosen as standard PCa cell lines used in the field to model PCa in vitro, the DU-145 and PC3 cell lines [
19]. Both are androgen independent lines, the DU-145 cell line is derived from a brain metastatic site considered similar in phenotype to adenocarcinoma, while the PC3 line is derived from bone metastasis, similar to the more aggressive prostatic small cell neuroendocrine carcinoma [
20,
29]. The comparison was also of interest as the two lines differ significantly in GPC-1 expression, with the highest expression in DU-145, and moderate expression in PC3. We demonstrated the efficacy of MIL-38-CD3 in both cell lines, in terms of T cell activation, cytokine release and cell killing, with significant cell killing achieved at 100 ng/ml for both lines. Of course, the use of primary prostate tumour cells in these in vitro assays would also be informative, particularly given the potential heterogeneity of GPC-1 expression in primary tumour. Moreover, being an in vitro study, one must note the potential influence of the in vivo tumour microenvironment on the apoptosis achieved using MIL-38-CD3 redirected T cells, notably any immunosuppressive factors present, and stromal cell support of cancer cell survival. Indeed, amongst the milieu of influencing factors present in the TME, it is known that prosate stromal cells may support the growth and inhibit apoptosis of prostate tumour cells [
30,
31]. This question could be addressed in part using a co-culture of primary stromal cells with primary prostate cells or prostate organoids, isolated from clinical samples [
30,
32].
To address the question of the effect of tumour microenvironment on the efficacy of MIL-38-CD3 as well as the next step in clinical development of MIL-38-CD3, we plan to test the construct in an animal model of PCa. A specific challenge in immunotherapeutic validation is the development of a clinically-relevant model. The CD3 arm of the construct does not cross-react with mouse CD3, so human T cells are required. The MIL-38 arm of MIL-38-CD3 does not cross-react with mouse GPC-1. Thus, the simplest model will be engraftment of enriched human T cells together with human tumour cells to establish a xenograft, followed by delivery of MIL-38-CD3 [
33‐
35]. Perhaps more clinically relevant, however, would be a mouse with humanised immune system and subcutaneous PCa xenograft. These studies are currently in planning. Despite not yet having demonstrated efficacy in an in vivo preclinical model, the in vitro validations described in this report strongly suggest the potential of MIL-38-CD3 BiTE as a promising new option in the therapy of PCa, either alone or in combination with checkpoint inhibitor therapy.
The expression of GPC-1 is not limited to PCa, rather, expression of GPC-1 has been described in a variety of solid tumours potentially responsive to BiTE therapy. Protein expression of GPC-1 has been described in breast [
36], pancreatic [
37,
38], bladder [
13], cervical [
39] and oesophageal [
16] cancers and glioblastoma [
40]. We envisage the potential use of MIL-38-CD3 in several solid tumours where it may prove to be efficacious in making known “cold” tumors like prostate and pancreas “hot” and, so, potentially responsive to therapeutic checkpoint inhibition.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.