The DOCK-AND-LOCK (DNL) method is a platform technology that combines recombinant engineering and site-specific conjugation to create multispecific, multivalent antibodies of defined composition with retained bioactivity. We have applied DNL to generate a novel class of trivalent bispecific antibodies (bsAb), each comprising an anti-CD3 scFv covalently conjugated to a stabilized dimer of different antitumor Fabs. Here, we report the further characterization of two such constructs, (E1)-3s and (14)-3s, which activate T cells and target Trop-2– and CEACAM5-expressing cancer cells, respectively. (E1)-3s and (14)-3s, in the presence of human T cells, killed target cells grown as monolayers at subnanomolar concentrations, with a similar potency observed for drug-resistant cells. Antitumor efficacy was demonstrated for (E1)-3s coadministered with human peripheral blood mononuclear cells (PBMC) in NOD/SCID mice harboring xenografts of MDA-MB-231, a triple-negative breast cancer line constitutively expressing Trop-2 and PD-L1. Growth inhibition was observed following treatment with (E1)-3s or (14)-3s combined with human PBMC in 3D spheroids generated from target cell lines to mimic the in vivo behavior and microenvironment of these tumors. Moreover, addition of an antagonistic anti–PD-1 antibody increased cell death in 3D spheroids and extended survival of MDA-MB-231-bearing mice. These preclinical results emphasize the potential of combining T-cell–redirecting bsAbs with antagonists or agonists that mitigate T-cell inhibition within the tumor microenvironment to improve immunotherapy of solid cancers in patients. They also support the use of 3D spheroids as a predictive alternative to in vivo models for evaluating T-cell functions. Cancer Res; 77(19); 5384–94. ©2017 AACR.
T cells play an undisputed role in cancer immunotherapy. T-cell–orchestrated antitumor immunity requires the recognition of the antigenic peptide/major histocompatibility complex by the T-cell antigen receptor (TCR) and the subsequent engagement of costimulatory receptors to sustain T-cell activation, which is negatively regulated by coinhibitory receptors. The increased knowledge of T-cell activation and modulation has resulted in four current approaches to harness T-cell–mediated immune responses against malignant tumors; namely, (i) adoptive transfer of autologous T cells that exhibit antigen-specific, antitumor activity (1); (ii) infusion of host T cells genetically engineered to express either TCR with altered specificity, or chimeric antigen receptors (CAR) with antibody-like recognition of a certain tumor (2); (iii) the exploration of immune checkpoint blockade to promote tumor regression by overcoming inhibition and increasing survival of tumor-reactive T cells (3); and (iv) redirection of T cells by T-cell activating and tumor-targeting, bispecific antibodies (bsAb) to destroy cancers (4). Moreover, ongoing preclinical and clinical studies to evaluate combinatorial strategies for further unleashing the antitumor immunity of T cells have yielded promising results, as exemplified by combinations of immune checkpoint inhibitors (5–7), combinations of immune checkpoint inhibitors either with immune checkpoint activators (8–10) or with genetically modified T cells (11), combinations of T-cell–redirecting bsAbs with PD-L1 antagonists (12) or CD28 agonists (13), and others (14).
The numerous bsAbs reported to date vary in formats, molecules they target, potential applications, as well as stages of development (4, 15–17). For example, bsAbs generated to redirect immune cells for killing cancer, besides their diverse tumor targets, can be Fc-bearing (12, 18, 19) or Fc-lacking (20–22); bivalent (20, 22), trivalent (18, 20,21, 23), or tetravalent (24–26); and activating T cells (18, 21, 22) or NK cells (23, 27).
We have previously described the construction of a novel class of trivalent, Fc-lacking, T-cell–redirecting bsAbs (21), each comprising a single chain variable fragment (scFv) of Okt3 (anti-CD3) site-specifically linked via the DOCK-AND-LOCK (DNL) method to a stabilized dimer of different antitumor Fab's. These DNL conjugates are designated (X)-3s, where the codes (X) and 3s denote the bivalent antitumor Fab dimer and the monovalent anti-CD3 scFv, respectively, as depicted in Fig. 1A. They have been shown to mediate the formation of immunological synapses between T cells and cognate target cells, induce T-cell activation and proliferation in the presence of target cells, kill target cells with subnanomolar IC50 when cocultured with T cells in vitro, and inhibit growth of human tumor xenografts in NOD/SCID mice reconstituted with human peripheral blood mononuclear cells (PBMC; ref. 21). We have also demonstrated in a follow-on study with (E1)-3s, a representative (X)-3s with specificity for Trop-2–expressing epithelial cancer cells, that the addition of interferon-α (IFN-α) further enhanced the potency of redirected T cells in vitro without a significant increase in cytokine production, and that the combination of (E1)-3s with peginterferon alfa-2a more effectively delayed the growth of Trop-2–expressing NCI-N87 human gastric cancer xenografts in vivo than single treatments with either (E1)-3s or peginterferon alfa-2a (28). In the current study, we provide evidence that the combination of a novel PD-1–blocking antibody (cPD-1) with (E1)-3s or the CEACAM5-targeting (14)-3s could potentiate the antitumor activity of redirected T cells against target human cancer cells grown in vitro as monolayer cultures or three-dimensional (3D) multicellular tumor spheroids (MCTS; ref. 29), or in vivo as xenografts in NOD-SCID mice.
Materials and Methods
Cell lines, antibodies, and reagents
Human cancer cell lines of breast (MDA-MB-231, HCC1954, HCC38, and BT-20), colon (HT-29, LS 174T, LoVo and COLO 205), and T lymphocytes (Jurkat), were purchased from the ATCC. Whereas MDA-MB-231, HCC38, HT-29, LS 174T, LoVo and COLO 205 were authenticated by short tandem repeat (STR) profiling in 2013 or 2014; HCC1954 (acquired in 2014), Jurkat (acquired in 2014), and BT-20 (acquired in 2010) have not been tested by STR profiling, but were never in culture for more than 50 passages or 6 months between collection or thawing and use in the described experiments. Each cell line was maintained according to the recommendations of ATCC and routinely tested for Mycoplasma using MycoAlert Mycoplasma Detection Kit (Lonza). For monolayer cultures, cells were grown in RPMI-1640 medium supplemented with l-glutamine (2 μmoles/L), 10% FBS, and 1% penicillin–streptomycin (Life Technology) at 37°C in a humidified atmosphere of 5% CO2. The establishment of MDA-MB-231-S120, a subline of MDA-MB-231 resistant to SN-38, has been described previously (30). The bispecific (E1)-3s and (14)-3s were produced as reported (21).
Sublines of MDA-MB-231–overexpressing Trop-2 were generated as follows. The cDNA of Trop-2 (GenBank: X77754.1) was synthesized, inserted into the HindIII-EcoRI sites of pcDNA3.1 (Thermo Fisher Scientific), and worked up to obtain the correctly assembled vector, which was transfected into MDA-MB-231 cells with lipofectamine using the Lipofectamine 2000 DNA Transfection Reagent Protocol (Thermo Fisher Scientific). At 72 hours post-transfection, cells were placed in fresh medium containing 1,000 μg/mL of G418 (Geneticin; Thermo Fisher Scientific) for 7 days, then serially diluted and dispensed into a 96-well plate. G418-resistant colonies were identified, picked, expanded in 6-well plates, stained with AF647-conjugated hRS7, and analyzed on a FACSCanto flow cytometer (BD Biosciences), from which seven clones with a similar or enhanced expression of Trop-2 relative to the parental MDA-MB-231 were selected for further experiments.
SpESF-X10-2D1, a PD-1–overexpressing clone, was selected in medium containing methotrexate (0.2 μmol/L) after transfection of SpESF-X10 (31) with PD-1-pdHL2, which was constructed as follows. The cDNA of PD-1 was PCR-amplified from Human PD-1/PDCD1 Gene cDNA ORF Clone (AcroBiosystems) using PD-1/XbaI left primer (TCTAGACACAGGACCTCACCATGCAGATCCCACAGGCGCC) and PD-1/EagI right primer (CGGCCGTCAGAGGGGCCAAGAGCAGTGTCC). The resulting amplimer was cloned into pGemT plasmid, from which the sequence of the XbaI/EagI insert was confirmed and ligated with the XbaI/EagI fragment of IFNα2b-DDD2-pdHL2 (32).
Generation of 5G9.G1.B11 and cPD-1
The PD-1–blocking mAb and its chimeric counterpart were generated as follows. BALB/c mice were immunized with recombinant human PD-1–Fc fusion protein (AB Biosciences), resulting in the isolation of a positive clone (5G9) by hybridoma technology. To ensure monoclonality, 5G9 was subcloned twice, yielding 5G9.G1.B11. Subsequently, VK and VH sequences of 5G9.G1.B11 were determined, combined with their respective human kappa and human IgG1 constant domains, and expressed in SpESF-X10 to obtain cPD-1.
Isolation of PBMCs
Buffy coats of healthy donors were purchased from the NJ Blood Center. PBMCs were obtained by density gradient centrifugation on Ficoll-Paque Premium (GE Healthcare Biosciences) following the manufacturer's protocol with minor modifications. Briefly, the buffy coat was diluted 3-fold with sterile PBS, and 35 mL of this suspension was slowly poured over 15 mL separation medium. After 30 minutes of centrifugation at 1,700 rpm (Sorvall RT Plus) with brakes turned off, the interphase was moved to a new reaction tube, washed twice with PBS, and centrifuged for 10 minutes at 1,500 rpm. The resulting cell pellets were suspended in complete RPMI-1640 medium, adjusted to a concentration of 1 × 107/mL, and used directly. Alternatively, the suspended PBMCs were adjusted to a concentration between 2 and 6 × 107/mL, mixed with an equal volume of 2 x freezing medium (20% DMSO and 80% FBS), from which 1-mL aliquots were dispensed into cryogenic vials and stored at −80°C until needed.
Isolation of CD8+ T cells from whole blood
Whole blood of healthy donors was added to Uni-SepMAXI tubes (Novamed) and fractionated by centrifugation. The buffy coat was collected, washed 2 to 3 times with PBS and centrifuged at 250 × g to further remove platelets. The resulting PBMCs were counted and 1 × 108 cells were pelleted for isolation of CD8+ T cells by negative selection using the kit and procedure of Miltenyi Biotech.
Cytotoxicity assays for tumor cells grown in monolayer
Adherent tumor cells were dissociated with Trypsin-EDTA (0.25%), washed 1 time with complete medium, counted, combined with CD8+ T cells at an effector to target ratio of 6:1, and centrifuged at 400 × g. The resulting pellet was suspended in medium such that the addition of 100 μL would provide 1 × 104 tumor cells and 6 × 104 T cells (effector-to-target ratio of 6) in each well of a 96-well plate. (E1)-3s or (14)-3s, starting at 6 nmol/mL, was 10-fold serially diluted in the medium specified for the target tumor cell line. After adding 200 μL of each dilution to each well, the plate was incubated for 24 to 72 hours. At the end of incubation, media were removed from wells and replaced with fresh, warm complete media to flush any nonadherent cells remaining in the wells. Media containing MTS reagent were then introduced and OD490 of the plate was read when sufficient color appeared in the untreated wells. The assay was done in triplicate.
Formation of MCTS for cytotoxicity evaluation
Ninty-six–well plates were coated with 80 μL of 1% agarose (Sigma) per well, which was then filled with 10,000 tumor cells detached from monolayer in 200 μL culture medium. After 4 days, the single spheroid formed in each well was transferred to non–tissue-culture–treated 6-well plates, with each well containing up to 6 spheroids in 2 mL culture medium and treated with testing agents for 2 to 3 days as indicated. At the end of the treatment, propidium iodide (PI) was added to stain dead cells and the fluorescent images were taken with a microscope camera.
Quantitation of surface antigens
Expression of Trop-2 or CEACAM5 on the cell surface was determined by flow cytometry (33). Briefly, cells were harvested with Accutase Cell Detachment Solution (BD Biosciences) and assayed for Trop-2 or CEACAM5 expression using QuantiBRITE PE beads (BD Biosciences) and a PE-conjugated anti–Trop-2 or anti-CEACAM5 antibody (eBiosciences,) following the manufacturer's instructions. Data were acquired on a FACSCanto flow cytometer (BD Biosciences) with CellQuest Pro software. Staining was analyzed with Flowjo software (Tree Star).
In vivo studies
Female NOD/SCID mice, 5-week-old, were purchased from Charles River Laboratories. All animal studies were approved by the Rutgers School of Biomedical and Health Sciences Institutional Animal Care and Use Committee. Mice were injected subcutaneously with a mixture of MDA-MB-231 tumor cells (5 × 106) and purified human T cells (2.5 × 106) combined with an equal volume of Matrigel. Therapy began one hour later. Treatment regimens, dosages, and number of animals are described in the figure legend. Animals were monitored daily for signs of tumor out-growth. Once tumors appeared, they were measured twice weekly. Tumor volume was determined by measurements in two dimensions using calipers, with volumes defined as: L × w2/2, where L is the longest dimension of the tumor and w the shortest.
Statistical analysis for the tumor growth data was based on AUC (area under curve) and survival time. Profiles of individual tumor growth were obtained through linear curve modeling. An F-test was employed to determine equality of variance between groups before statistical analysis of growth curves. A two-tailed t test was used to assess statistical significance between all the various treatment groups and controls except for the saline control, in which a one-tailed t test is used. As a consequence of incompleteness of some of the growth curves (due to deaths), statistical comparisons of AUC were only performed up to the time at which the first animal within a group was sacrificed. Survival was analyzed using Kaplan–Meier plots (log-rank analysis), using the Prism GraphPad Software (v6.05) purchased from Advanced Graphics Software. Survival surrogate endpoint was the time for tumor progression to 1.0 cm3. Significance was considered at P ≤ 0.05.
Characterization of 5G9.G1.B11, an antagonistic murine anti–PD-1 monoclonal antibody, and its chimeric counterpart (cPD-1)
The 5G9.G1.B11 mAb was purified by Protein A to homogeneity, as shown by SE-HPLC (Supplementary Fig. S1A) and SDS-PAGE (Supplementary Fig. S1B). The reactivity of 5G9.G1.B11 for PD-1 was confirmed by SE-HPLC with its binding to recombinant PD-1-His (Supplementary Fig. S1C), by ELISA with its binding to either PD-1-His or PD-1-Fc (Supplementary Fig. S1D), and by flow cytometry with its binding to PD-1-expressed on activated Jurkat T cells (Supplementary Fig. S2). Importantly, the blocking activity of 5G9.G1.B11 was demonstrated by a dose-dependent increase of IL-2 secreted by T cells in a mixed lymphocyte assay (Supplementary Fig. S3). cPD-1, a chimeric version of 5G9.G1.B11, comprising the VH and VK of 5G9.G1.B11 and human Fc of IgG1, was generated, and the mAb from the lead clone (2G9) was purified and shown to be homogeneous by SE-HPLC (Supplementary Fig. S4A) and SDS-PAGE (Supplementary Fig. S4B). The binding of cPD-1 to recombinant PD-1-His was demonstrated by ELISA (Supplementary Fig. S4C; EC50 = 49 ± 16 pmol/L based on three separate experiments), and further confirmed by flow cytometry (Supplementary Fig. S5A) with SpESF-X10-2D1, a subline of SpESF-X10 (34) transfected to overexpress PD-1 (Supplementary Fig. S5B). The blocking activity of cPD-1 was comparable with that of 5G9.G1.B11 or EH12.2H7 (a commercially available mouse anti-human PD-1 blocking mAb), as demonstrated by inhibiting the binding of biotinylated PD-1 (B*-PD-1) to the endogenous PD-L1 expressed on MDA-MB-231 (Supplementary Table S1). In this experiment using flow cytometry, the median fluorescence intensity (MFI) determined for the binding of biotinylated PD-1 (B*-PD-1) to PD-L1 on MDA-MB-231 cells with phycoerythrin-labeled streptavidin (PE-SA) decreased with the addition of increasing amounts cPD-1, 5G9.G1.B11, or EH12.2H7, and was observed at a similar level for each of the three tested concentrations (1.5, 15, and 150 μg/mL), further supporting the antagonistic properties of cPD-1.
Potent killing of target cells grown in monolayer
The cytolytic activity of (E1)-3s via engagement of human T cells with Trop-2–expressing tumor cells has been demonstrated in vitro for Capan-1 (pancreatic cancer) and NCI-N87 (gastric cancer) to display IC50 (the concentration to achieve 50% lysis) of 29 and 0.85 pmol/L, respectively (21). In addition, the IC50 of (14)-3s was determined to be 2 pmol/L for LS 174T (a human colonic cancer cell line). We now report that (E1)-3s also mediated potent killing, via T cells at an effector-to-target ratio of 6, of four breast cancer cell lines (MDA-MB-231, HCC1954, HCC38, and BT-20), all of which express endogenous Trop-2 on the cell surface (Supplementary Table S2); as well as four Trop-2-transfected sublines of MDA-MB-231 (designated as 231-C13, 231-C29, 231-C36, and 231-C39), each having a higher expression of Trop-2 than the parental MDA-MB-231 (Supplementary Table S3). Based on the 48-hours viability assays using MTS [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium)], the IC50 of (E1)-3s was 4.4 ± 1.5 pmol/L (n = 4) for MDA-MB-231 and between 1 and 10 pmol/L for the other cell lines (Fig. 1B and Supplementary Table S3). In addition, the IC50 of (E1)-3s determined for the SN-38-resistant MDA-MB-231-S120 (33) from the 24-hours assay was 11.5 pmol/L, similar to the IC50 of 12.4 pmol/L determined for MDA-MB-231 (Supplementary Table S4), indicating the overexpression of ABCG2, a multidrug-resistant transporter belonging to the ATP-binding cassette sub-family G, in MDA-MB-231-S120 has little impact on the antitumor effect of T cells redirected by (E1)-3s.
For colonic cancer cell lines (Fig. 1C), (14)-3s was confirmed to display a subnanomolar IC50 (19 ± 13 pmol/L; n = 2) in LS 174T, while showing a similar activity in LoVo (IC50 = 19 ± 17 pmol/L; n = 2), but a much reduced strength in either COLO 205 (IC50 = 671 ± 311 pmol/L; n = 2) or HT-29 (IC50 > 3,000 pmol/L; n = 2). Because the surface expression of CEACAM5 is relatively high on LS 174T or LoVo cells (binding sites for hMN-14 per cell in excess of 20,000; Supplementary Table S2) and considerably lower on COLO 205 cells (binding sites for hMN-14 per cell about 7,000 to 8,000; Supplementary Table S2), these results reflect a likely correlation between CEACAM5 expression and (14)-3s potency, as observed for two other bsAbs, CEA TCB (18) and MEDI-565 (34) that redirect T cells to kill CEACAM5-positive colorectal cancer cell lines. However, the much higher IC50 observed for HT-29 also suggests that the expression levels of CEACAM5 might not be the only determinant for the potency of (14)-3s.
Cytolysis of MCTS by bsAb-redirected T cells
Single spheroids of MDA-MB-231 were grown for 5 days in 96-well plates coated with agarose, transferred to 6-well plates, and evaluated for cytolysis by T cells [labeled intracellularly with carboxyfluorescein succinimidyl ester (CFSE)] in the presence of (E1)-3s at 100 pmol/L. As shown in Fig. 2A, cells in spheroids were effectively killed at 24 hours only in the presence of both (E1)-3s and T cells. Under these conditions, cell death was detectable at 24 hours with 1.5 × 106 of T cells (Fig. 2B) and most, if not all, cells in the spheroids treated with the highest concentration of T cells (6 × 106) were dead at 48 hours, but not the surrounding T cells (data not shown). With PBMCs as effector cells and after a 48-hours incubation, (14)-3s at 100 pmol/L likewise killed the spheroids of HT-29 (Fig. 3A) and LS 174T (Fig. 3B). Similar cytotoxicity in 3D spheroids of HT-29 and LS 174T was also demonstrated for (14)-3s using purified T cells, as shown Supplementary Fig. S6.
Blockade of PD-1 by cPD-1 enhanced lysis of target cells grown in monolayer or as MCTS
The constitutive expression of PD-L1 on MDA-MB-231 (Supplementary Fig. S7 and ref. 35) and its Trop-2–overexpressing sublines (Supplementary Table S3) suggests they would be suitable for evaluating the potential of cPD-1 to enhance the potency of T cells by blocking the interaction of PD-1 with PD-L1. The results shown in Supplementary Table S4 indicate that a 2- to 3-fold decrease of IC50 was observed for MDA-MB-231 and MDA-MB-231-S120 with the addition of 10 to 15 μg/mL of cPD-1 to the monolayer culture in which CD8+ T cells were present at an effector-to-target ratio of 6, whereas the concentrations of (E1)-3s were varied. Moreover, when the corresponding EC50 (half maximal effective concentration) values were analyzed with the Prism software for each paired samples, the enhanced potency by cPD-1 was highly significant (P < 0.0001; Supplementary Table S4). Further evidence for cPD-1 to enhance the potency of T cells redirected by (E1)-3s was provided by the demonstration of more dead cells in the spheroids of MDA-MB-231 treated with both (E1)-3s and cPD-1 than those treated with only (E1)-3s, using PBMCs (Fig. 4).
The effect of cPD-1 also was investigated for HCC1954, which can be induced by IFN-γ to express PD-L1, as shown by flow cytometry of individual cells obtained from monolayer culture (Supplementary Fig. S8A) or by immunostaining of clustered cells in a spheroid (Supplementary Fig. S8B). The method to evaluate the response of HCC1954 spheroids, stimulated with IFN-γ or not, to combined (E1)-3s and PBMCs, is illustrated in Fig. 5A. Whereas the addition of cPD-1 clearly increased the extent of cell death in HCC1954 spheroids pretreated with IFN-γ (Fig. 5B) or not (Fig. 5C), PD-1 blockade with cPD-1 evidently resulted in more T-cell–mediated death in pretreated (Fig. 5D, bottom) than in untreated samples (Fig. 5D, top). Using T cells purified from PBMCs, the addition of cPD-1 to (E1)-3s also induced more cell death in the spheroids formed from IFN-γ–stimulated MDA-MB-231, when compared with (E1)-3s without the addition of cPD-1 (Supplementary Fig. S9A). The purified T cells also induced (E1)-3s–mediated cell death in the spheroids formed from MDA-MB-231, but the effect of cPD-1 was not as prominent as that observed in the spheroids formed from IFN-γ–stimulated MDA-MB-231 (Supplementary Fig. S9B and S9C). These results thus provide a basis for using the 3D spheroids to study T-cell functions as an alternative to humanized murine models. The presence of PD-1 on T cells isolated from PBMCs as well as on T cells in the PBMC population was shown by flow cytometry in Supplementary Fig. S10.
Combination of (E1)-3s and human T cells with cPD-1 improved the prolonged survival of NOD/SCID mice bearing MDA-MB-231 xenografts
Because MDA-MB-231 (a triple-negative human breast cancer cell line) is shown to express PD-L1 and that in vitro treatment with (E1)-3s and cPD-1 effects a significant increase in T-cell killing of target cells mediated by (E1)-3s, an animal model was established to determine whether these results would translate to an in vivo setting. The study consisted of five groups: three treatment groups, one untreated control group, and one out-growth control group, with 5 NOD/SCID mice in each group. To begin the therapy, the out-growth control group was injected with MDA-MB-231 cells only, and the other four groups were each injected subcutaneously with a mixture of MDA-MB-231 and human T cells. One hour later, (E1)-3s, cPD-1, or a combination of (E1)-3s and cPD-1 was administered to each of the three treatment groups intravenously (i.v.), whereas saline was given to the untreated control group as well as the out-growth control group.
Untreated control mice injected with the mixture of tumor cells and T cells produced tumors that progressed at a slower rate when AUC was compared with that of outgrowth control animals injected with only MDA-MB-231 cells (Fig. 6A; P = 0.0378), suggesting that these T cells may have some activity on their own to slow tumor growth and improve survival (Fig. 6B; P = 0.0031). However, treatment with only cPD-1 was not significant to provide an antitumor benefit to the animals and was no different from untreated control mice. Conversely, both (E1)-3s alone and (E1)-3s plus cPD-1 significantly slowed tumor growth in these mice (P < 0.0121; AUC), with a resulting significant improvement in overall survival compared with controls (Fig. 6B; P < 0.0031). Importantly, the addition of cPD-1 therapy with (E1)-3s did provide an even greater antitumor effect compared with mice treated with only (E1)-3s (P = 0.0121; AUC). At the last assessable time-point for both groups (day 42), mice treated with (E1)-3s plus cPD-1 had tumors that were 2.3-fold smaller in volume than in those treated with only (E1)-3s (0.432 ± 0.264 cm3 vs. 0.979 ± 0.126 cm3; P = 0.0031). The combination therapy likewise provided a significant survival benefit compared with (E1)-3s monotherapy (median survival time = 49 days vs. 42 days, respectively; P = 0.008). These data suggest that in the absence of (E1)-3s, T cells alone may exhibit some antitumor activity in this model, but it was so low that the addition of a checkpoint inhibitor was not enough to produce an improved immune response. On the other hand, the redirection of these T cells by (E1)-3s to MDA-MB-231 would better activate them, and the addition of cPD-1 would block PD-L1–mediated inhibitory signaling of these same T cells, resulting in an overall significantly improved therapeutic effect.
Salient features of (X)-3s, a novel class of T-cell–redirecting bsAbs
BsAbs capable of redirecting T cells to cancers and thereby activating such T cells to kill malignant tumors have shown promise in both preclinical and clinical studies. However, their therapeutic benefits on late-stage patients with solid cancers remain to be established. Previously, we reported highly effective T-cell–redirected therapy of Trop-2–expressing pancreatic and gastric tumors in xenograft models using (E1)-3s, and noted that the relatively low level of cytokine release induced by (E1)-3s would be a potential advantage over other types of T-cell–redirected bsAbs, in particular, BiTE (28).
In the current study, we explored the potential utility of (E1)-3s for therapy of Trop-2–expressing breast cancers, including triple-negative breast cancer (TNBC), as well as that of (14)-3s for CEACAM5-expressing colonic cancers. Both (E1)-3s and (14)-3s were highly potent for their target tumor cells, exhibiting largely subnanomolar IC50 against diverse cell lines grown in monolayer cultures. Moreover, the addition of cPD-1, a novel checkpoint inhibitor to PD-1, could significantly enhance (E1)-3s–mediated T-cell killing of MDA-MB-231 cells, which constitutively express PD-L1 (35). Additional evidence supporting the therapeutic benefits of PD-1 blockade with cPD-1 was highlighted in treatments of NOD/SCID mice bearing MDA-MB-231 xenografts with a combination of cPD-1, (E1)-3s, and human T cells, which in comparison with treatments without the addition of cPD-1, significantly inhibited tumor growth, as assessed by AUC (P = 0.0121), and improved the median survival time by one week (P = 0.008). Although the in vivo data were obtained from only one xenograft model, appeared to be modest, and might not be observed in other tumor models, these encouraging results were demonstrated semi-quantitatively for (E1)-3s and (14)-3s with the cognate 3D spheroids, which have been widely recognized to better reflect the physiology and microenvironment of tumor cells in vivo, and are being used increasingly in cancer research to complement monolayer cultures for drug testing, as exemplified by the pharmacological evaluation of T-cell–redirecting, anti-EpCAM x anti-CD3, calutumomab (36) and other antitumor agents, including antibody–drug conjugates (37) and a monospecific anti-EGFR antibody (38).
Comparison with other T-cell–redirecting bsAbs
Addition of a PD-L1-blocking antibody to HER2-TDB, a T-cell–engaging and HER2-targeting bsAb produced in bacteria using the knob-into-hole approach (39), was reported to enhance the therapeutic outcomes of hCD3 transgenic mice carrying syngeneic CT26-HER2 tumors (12). Despite the difference from HER-TDB in molecular components and target tumor antigen, (E1)-3s displays antitumor potency in Trop-2-expressing TNBC cells grown in monolayer cultures with low picomolar IC50 comparable with HER2-TDB in HER2-expressing cancer cells, and like HER2-TDB, was effective in killing target cancer cells resistant to established drug and antibody–drug conjugates. Moreover, the potential benefits of combination therapy with (E1)-3s and anti–PD-1 for augmenting T-cell immunity was demonstrated in a humanized mouse model and further shown in 3D spheroids to complement the in vivo results.
More recently, several examples of Fc-bearing bsAbs generated by CrossMab technology (40) have been reported, including the T-cell-redirecting CEA TCB (18), which as (14)-3s, binds monovalently to CD3 and bivalently to the membrane proximal domain of human CEACAM5. Despite apparent differences in constituent antibodies and respective architecture, the resemblance of CEA TCB to (14)-3s in target specificity and valency is noteworthy. Yet a comparison of the EC50 determined under similar conditions in LS 174T for CEA TCB (1,410 pmol/L; ref. 41) with that of (14)-3s (2 to 28 pmol/L; ref. 21 and this study) indicates at least a 50-fold increase in potency for (14)-3s. The anti-CEACAM5 component of (14)-3s is derived from hMN-14 (42), which binds to a region of the A3B3 domain in the vicinity of the cell membrane. Because the effectiveness of a T-cell–redirecting bsAb could be inversely correlated with the distance of its cognate epitope to the target cell surface (43), whether the higher potency of (14)-3s might result from the closer binding of hMN-14 to the cell surface of LS 174T than that of PR1A3 (44) used in CEA TCB, remains to be investigated.
Cancer immunotherapy using bsAb-redirected T cells versus CAR-engineered T cells
Whereas T cells can be redirected to kill cancer cells using novel CARs or modified TCRs, with impressive results against B-cell malignancies as reported in clinical studies for acute lymphoblastic leukemia (45) and relapsed refractory chronic lymphocytic lymphoma (46), the alternative approach to redirecting T cells with bsAbs for cancer therapy is believed to be more amenable to commercial development and has resulted in two regulatory approvals, catumaxomab in 2009 for the treatment of malignant ascites (47) and blinatumomab in 2014 for the treatment of B-precursor acute lymphoblastic leukemia (48). Similar to the continuing evolution of novel CAR-T cells with the goal to improve safety and retain efficacy (49), numerous efforts are underway to develop optimal bsAbs that would redirect T cells to kill cancer cells with picomolar potency and IgG-like serum half-life, among which HER2-TDB appears to be a promising lead (12). Encouragingly, (E1)-3s is comparable to HER2-TDB with regard to several in vitro properties, including the enhanced potency in combination with antagonistic anti–PD-1 or anti–PD-L1 antibodies. However, recent reports have revealed the blockade of PD-1 could enhance A2A adenosine receptor (A2AR) expression on tumor-infiltrating CD8+ T cells, making them more susceptible to suppression by adenosine produced by CD73 on cancer cells (50). Thus, combining antagonists of A2AR with checkpoint inhibitors of the PD-1/PD-L1 pathway should further augment the antitumor activity of bsAb-redirected T cells.
In conclusion, the T-cell–redirecting bsAbs generated with the DNL method, which are distinguishable by their high potency and low production of PD-L1–stimulating IFN-γ, offer a promising new format to improve the combination immunotherapy of solid cancers with antagonists or agonists that alleviate T-cell inhibition within the tumor microenvironment. The current study also provides a basis for using 3D spheroids as an alternative to in vivo models for evaluating T-cell functions, with future experiments to address the effect of exhausted T cells, which are the predominant phenotype present in the tumor stroma. Although the cytolytic efficiency of (14)-3s, (E1)-3s, and other T-cell–redirecting bsAbs evidently requires the presence of cognate antigens on target cells, their ultimate potency in a certain cancer may depend on additional factors, such as the mutational status, epitope specificity and density on surface, and expression of intrinsic and induced immune evasion molecules, all of which remain to be explored (depicted in Fig. 7).
Disclosure of Potential Conflicts of Interest
D.M. Goldenberg has ownership interest (including patents) in Immunomedics, Inc. No other potential conflicts of interest were disclosed by the authors.
Conception and design: C.-H. Chang, E.A. Rossi, D.M. Goldenberg
Development of methodology: C.-H. Chang, D. Liu, E.A. Rossi
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Wang, R. Li, D.L. Rossi, D. Liu, T.M. Cardillo
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.-H. Chang, Y. Wang, D.L. Rossi, D. Liu, E.A. Rossi, T.M. Cardillo, D.M. Goldenberg
Writing, review, and/or revision of the manuscript: C.-H. Chang, Y. Wang, D. Liu, E.A. Rossi, T.M. Cardillo, D.M. Goldenberg
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D.M. Goldenberg
Study supervision: C.-H. Chang, D.M. Goldenberg
We thank John Kopinski, Diana Chereches, and Ali Mostafa for excellent technical assistance.
This work was supported by Immunomedics, Inc. There was no external financial or grant support received.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.