Introduction
Traditionally, anticancer treatments have been tested in classical two-dimensional (2D) culture systems, but these models suboptimally reproduce the complex morphological and histopathological features characteristic of tumor microenvironments, particularly concerning tumor–host and tumor–immune cell interactions, which limits their value as models for cancer immunotherapy drug development [
1]. Currently, there is limited availability of in vitro models that adequately mimic in vivo conditions. Compared to 2D cultures, three-dimensional (3D) models better predict in vivo immune cell behavior [
2], thereby providing a more robust system to evaluate immunotherapeutic agents.
Cells cultured in 3D culture systems (spheroids) are phenotypically different to those grown in 2D culture systems (monolayers), as shown by differential gene expression [
3‐
10]. In particular, expression of genes encoding signal transduction proteins, and cell surface markers, and localization and oligomerization of these markers in 3D cancer spheroids were shown to be more similar to the in vivo environment [
11,
12]. The 3D microenvironment, including the extracellular matrix (ECM), is profoundly altered in tumors [
13,
14], and 3D spheroid models simulate these alterations [
15]. In 3D spheroids, the development of extracellular matrix (ECM), the distribution of oxygen and metabolites, and cell proliferation gradients are more similar to those of cancer tissues than those of 2D culture models and more closely resemble avascular or intervascular tumor microregions [
10,
16‐
20]. Therefore, 3D spheroid systems constitute a more physiological model for in vitro testing of drugs [
12,
21‐
23]. In addition to structural features, 3D cultures offer temporal advantages because they also allow kinetic analysis of therapeutic activity, which is suboptimally assessed in 2D cultures owing to fast and efficient immune cell activation and target cell lysis (unpublished data).
The behavior of the immune cells is also affected by the infiltration dynamics within a 3D scaffold due to physical constraints and chemical cues. Infiltration of immune cells was studied in benign and malignant cell spheroids, where it was found that stimulation of human peripheral blood mononuclear cells (PBMCs) by tumor cells or IL-2 was necessary for the infiltration of PBMCs into the spheroids [
24]. In 3D melanoma spheroids, tumor cells had a higher capacity to inhibit both T cell activation and proliferation than in 2D cultures, suggesting that spheroids support a more physiological immune-modulatory function than 2D cultures [
25]. Cell-killing assays in 3D models have shown comparable results to in vivo experiments [
26]. Tumor 3D homotypic spheroids were also shown to be a more appropriate model than classical 2D cultures for studying the cytotoxicity and infiltration of NK cells, allowing stable, reproducible, and long-term co-culture with immune cells and enabling the characterization of immune cell subpopulations [
27].
We explored the versatility of a newly developed 3D heterotypic spheroid model composed of tumor cells, fibroblasts, and immune cells by studying drug targeting and immune cell infiltration, activation, and cytotoxicity in response to novel cancer immunotherapy agents currently in Phase I clinical trials (IgG-IL2v and T cell bispecific antibodies [TCBs]) as monotherapy and in combination.
IgG-IL2v is a monomeric IL2 variant fused to an IgG1 antibody with FcγR-silent Fc that is either targeted to carcinoembryonic antigen (CEA)–expressing tumors (CEA-IL2v), fibroblast activation protein-α-expressing fibroblasts (FAP-IL2v), or untargeted (DP47-IL2v = IgG-IL2v). The IL2v mutations abolish binding to the alpha subunit of the IL2 receptor (IL2Rα, CD25) and result in an IL2 variant (IL2v) that lacks the preferential activation of regulatory T cells (T
regs) by normal IL-2, binding only to the IL2 receptor βγ complex and thus leads to activation and expansion of CD8+ effector, CD4+ helper T cells and NK cells in vitro and in vivo [
28,
29]. Phase I clinical trials with CEA-IL2v as well as FAP-IL2v are currently ongoing (NCT02350673 and NCT02627274) [
30,
31].
CEA-targeted TCB (CEA TCB) is a novel tumor-targeted T cell bispecific antibody that binds simultaneously to CEA overexpressed on many solid tumors (colorectal, pancreatic, gastric, lung, breast, and ovarian carcinomas) and to CD3 (epsilon chain, CD3e) present on T cells. In the presence of CEA-expressing target cells and T cells, treatment with CEA TCB leads to rapid formation of immunological synapses and T cell activation followed by T cell-mediated killing of tumor cells, T cell proliferation and cytokine release [
32‐
34]. Phase I clinical trials with CEA TCB are currently ongoing (NCT02324257) [
35]. FAP TCB is a fibroblast-targeted TCB that binds to FAP and the CD3e antigen.
Materials and methods
Cell lines and 3D heterotypic spheroid cultures
Human colon adenocarcinoma cells LS174T (ECACC 87060401) and LoVo (ECACC 87060101) were cultured in Eagle’s Minimum Essential Media (EMEM; Gibco®, Life Technologies, Zug, Switzerland) and DMEM (Gibco®, Life Technologies, Zug, Switzerland), respectively; MRC-5 normal fetal lung fibroblast cells (ATCC CCL-171) and CCD-18Co a normal colon fibroblast cell line (ATCC CRL-1459) were cultured in EMEM or fibroblast growth medium-2 (FGM™-2; Lonza, Basel, Switzerland). Media were supplemented with 10% FBS (Gibco®, Life Technologies, Zug, Switzerland), 2 mM GlutaMax (Gibco®, Life Technologies, Zug, Switzerland), 1% nonessential amino acid solution (NEAA, 100×, Sigma-Aldrich Chemie GmbH, Buchs, Switzerland), and 1 mM sodium pyruvate (Sigma-Aldrich Chemie GmbH, Buchs, Switzerland). Cells were cultured in a humidified atmosphere of 5% CO2 at 37 °C and split every 2–3 days at a ratio of 1:5 (LS174T and LoVo) and 1:2 (MRC-5, CCD-18Co).
3D heterotypic spheroid cultures were generated by hanging drop method using the GravityPLUS™ kit (InSphero AG, Schlieren, Switzerland) at a ratio of 1:50 (tumor cells to fibroblasts). Spheroids were cultured in FGM™-2 supplemented with 10% FBS. Cultures were incubated in a humidified atmosphere of 5% CO2 at 37 °C for 3 days. Spheroid compaction was confirmed by observation under a microscope, before transferring spheroids to GravityTRAP™ assay plates (InSphero AG, Schlieren, Switzerland).
PBMCs were isolated from healthy donors by standard Histopaque density gradient centrifugation and added to the supernatant.
For IL2 (Proleukin [aldesleukin], Novartis) and IgG-IL2v antibody monotherapy assays, heterotypic spheroids were incubated with PBMCs and IgG-IL2v antibody (100 nM) for 24 h. For TCB monotherapy assays, heterotypic spheroids were incubated for 24, 48, and 72 h with PBMCs (PBMCs to fibroblasts 10:1) and TCBs (50 nM). For IgG-IL2v and TCBs combination therapy assays, heterotypic spheroids were pre-incubated for 16 h with untargeted (DP47) IgG-IL2v (10 nM) and PBMCs (PBMCs to fibroblasts 10:1). On the following day, the heterotypic spheroids and PBMCs were washed with fresh media and then cultured for 24, 48, and 72 h in medium containing TCBs (50 nM). Non-treatment control assays were performed in the presence and absence of PBMCs.
Targeting of IgG-IL2v in the heterotypic spheroid model
Fluorescently labeled CEA-targeted, FAP-targeted, and untargeted IgG-IL2v cytokines were used to monitor targeting in 3D heterotypic spheroids. 3D heterotypic spheroids were incubated for 16 h in media containing 100 nM labeled IgG-IL2v. Media were replaced to remove unbound IgG-IL2v before rinsing the cells in Dulbecco’s phosphate-buffered saline (DPBS) and fixing in 4% paraformaldehyde for 1 h at room temperature. Cells were embedded in mounting medium containing DAPI (InSphero AG, Schlieren, Switzerland) and transferred onto glass slides prior to 2-photon confocal microscopy. Two-photon confocal microscopy provided high-resolution imaging of the inner spheroid core and detection of the bound fluorescently labeled IgG-IL2v. Tumor and fibroblast compartments were identified based on morphology.
Immunohistochemistry and quantification
For IHC analysis, heterotypic spheroids were embedded in paraffin wax. Staining was performed on 3-µm thick serial sections of the heterotypic spheroids mounted onto poly-L-lysine-coated glass slides. The morphology of the sectioned heterotypic spheroids was evaluated after H&E staining. IHC was performed by SophistoLab® (Muttenz, Switzerland) using the horseradish peroxidase Bond Polymer Refine Detection kit (Leica Biosystems AG, Muttenz, Switzerland) and analyzed using the BOND-MAX instrument (Leica Microsystems Newcastle Ltd, Newcastle, UK) as per manufacturer’s guidelines. Primary antibodies used for staining were: anti-human CD3 (clone SP7, NeoMarkers Inc., Fremont, US); anti-human CEA (produced in-house), anti-human CD45 (clone 2B11 & PD7/26; Cell Marque Lifescreen Ltd, Rocklin, US), anti-human CD14 (ab45870, Abcam Ltd, Cambridge, UK), and antihuman Seprase monoclonal antibody (Clone D8) (Vitatex Inc., Stony Brook, US).
CEA+ (tumor cells), CEA− (fibroblasts), and CD3+ (infiltrated T cells) areas were quantified by digital immunostained slides using ImageJ software (National Institutes of Health). CEA+ and CEA− areas were quantified by normalization to the total heterotypic spheroid area. The CD3+ areas within tumor and fibroblast compartments were quantified by normalization to the total tumor and fibroblast-rich areas, respectively. Tumor and fibroblast compartments were identified by morphology. Statistical analysis was performed by Student’s t test.
Cytokine/chemokine release by cytometric bead array
Cytokine/chemokine secretion in the supernatant was measured by flow cytometry, using the Cytometric Bead Array (CBA, BD Biosciences, Franklin Lakes, NJ, USA), according to the manufacturer’s guidelines. Supernatants from individual heterotypic spheroids were collected and stored at −20 °C. Supernatants were subsequently thawed, and 1 well (50 µL) and 5 pooled wells (150 µL) were analyzed under each treatment condition for IgG-IL2v monotherapy and TCB monotherapy/combination therapy experiments, respectively. The following CBA kits (BD Biosciences, Franklin Lakes, NJ, USA) were used: CBA human IFNγ Flex Set, CBA human Granzyme B Flex Set, CBA human RANTES Flex Set (D4), CBA human MIP-1β Flex Set (E4), CBA human TNF Flex Set, CBA human IL-1β Flex Set (B4), and CBA human IL-6 Flex Set. Samples were measured using the BD FACS Canto II, and analyses were performed using the Diva Software (BD Biosciences, Franklin Lakes, NJ, USA). Assays were performed in triplicate.
Flow cytometry
Tumor/fibroblast (ratio 1:50) heterotypic spheroids were incubated with 5 × 104 PBMCs per well. Following treatment, the external tumor layer of the heterotypic spheroids was dissociated by 5 min incubation at room temperature in enzyme-free, phosphate-buffered saline-based cell dissociation buffer (Gibco®, Life Technologies, Zug, Switzerland). The remaining central core of fibroblasts with residual tumor cells was dissociated by 10–20 min incubation with 0.64 mg/mL Dispase II and 1 mg/mL Collagenase D (Roche Diagnostics, Mannheim, Germany). The single-cell suspension was washed with DPBS and resuspended in the DPBS containing antibody mixture for cell staining. For each condition, 32 heterotypic spheroids were pooled and tested in triplicate.
Flow cytometry was performed using anti-CEA labeled with Alexa 488 (produced in-house), PerCPCy5.5 anti-human CD45 (Biolegend, San Diego, CA, USA), Brilliant Violet 421 anti-human CD56 (Biolegend, San Diego, CA, USA), Brilliant Violet 605 anti-human CD69 (Biolegend, San Diego, CA, USA), Brilliant Violet 605 Mouse IgG1, κ (kappa) Isotype (Biolegend, San Diego, CA, USA), and LIVE/DEAD® Fixable Aqua Dead Cell Stain Kit (Life Technologies, Zug, Switzerland). Samples were measured using the BD FACS Fortessa. Analyses were performed using the Diva Software (BD Biosciences, Franklin Lakes, NJ, USA). Assays were performed in triplicate.
Statistics
The statistical analysis was performed using GraphPad PRISM software version 6. Error bars represent the standard deviation in all graphs. Two-tailed, unpaired parametric t tests were performed by setting the confidence intervals to 95% (definition of statistical significance: p < 0.05).
Discussion
The 3D heterotypic spheroid model presented here provides a novel and versatile platform for testing of tumor- and fibroblast-targeted cancer immunotherapy agents.
The effects of a TCB antibody in a 3D homotypic spheroid model were recently reported by Hirschhaeuser et al. [
36]. However, this homotypic 3D model lacked the simultaneous presence of tumor cells and fibroblasts. Therefore, despite providing valuable information, this homotypic model is suboptimal for the assessment of therapies targeted to either tumor or fibroblasts (stroma) and does not allow the analysis of immune cell infiltration and activity directed to one of the two compartments. Here, we report for the first time a novel 3D heterotypic spheroid model composed of tumor cells, fibroblasts, and immune cells. This allows for cell–cell and cell–matrix interactions in a 3D microenvironment where immune cells can infiltrate and migrate toward both tumor cells and fibroblasts, thereby improving the predictability of in vitro testing of immunotherapy agents. This novel 3D heterotypic spheroid model more closely resembles the complexity of tumors than either homotypic 3D spheroid models or 2D culture systems. Two-dimensional models are useful for large screening assays of candidate therapeutics to select potential drug candidates as killing of tumor target cells or fibroblasts in 2D is very efficient. However, the 3D model can offer a more realistic picture of the in vivo activity of a potential drug candidate.
The 3D spheroids presented in the current manuscript represent a novel heterotypic model; however, limitations include the use of tumor and fibroblast cell lines and PBMCs derived from healthy individuals added at high E:T ratios. Such 3D models could be further optimized by using patient-derived tumors containing TILs with physiological E:T ratios. Alternatively, infiltration and activation of TILs, rather than healthy PBMCS could be investigated, given that TILs are frequently suppressed and show suboptimal activity as compared to PBMCs derived from healthy donors.
Specific targeting and retention to tumor and fibroblast spheroid areas was demonstrated using tumor- and fibroblast-targeted IgG-IL2v (CEA-IL2v and FAP-IL2v, respectively), along with enhanced leukocyte infiltration, activation, and subsequent elimination of tumor cells. Treatment with tumor- or fibroblast-targeted TCB antibodies (CEA TCB and FAP TCB, respectively) led to efficient target-specific cross-linking and retention of T cells to tumor cells or fibroblasts, respectively. This resulted in effective elimination and reduction in the corresponding spheroid areas. The combination of both TCBs with IgG-IL2v was more efficacious than either monotherapy alone, as shown by an increase of immune cell infiltration and activation, faster elimination of target cells, and enhanced IFNγ and Granzyme B secretion.
Taken together, the 3D heterotypic model presented here enables a comprehensive in vitro evaluation of immune cell infiltration and target cell elimination after therapeutic intervention, which would not be possible using 2D systems. The study also demonstrates that tumor- and fibroblast-targeted IgG-IL2v and TCBs are promising candidates for combination treatments in cancer immunotherapy.