A CD3-bispecific molecule targeting P-cadherin demonstrates T cell-mediated regression of established solid tumors in mice

Abstract

Strong evidence exists supporting the important role T cells play in the immune response against tumors. Still, the ability to initiate tumor-specific immune responses remains a challenge. Recent clinical trials suggest that bispecific antibody-mediated retargeted T cells are a promising therapeutic approach to eliminate hematopoietic tumors. However, this approach has not been validated in solid tumors. PF-06671008 is a dual-affinity retargeting (DART^®)-bispecific protein engineered with enhanced pharmacokinetic properties to extend in vivo half-life, and designed to engage and activate endogenous polyclonal T cell populations via the CD3 complex in the presence of solid tumors expressing P-cadherin. This bispecific molecule elicited potent P-cadherin expression-dependent cytotoxic T cell activity across a range of tumor indications in vitro, and in vivo in tumor-bearing mice. Regression of established tumors in vivo was observed in both cell line and patient-derived xenograft models engrafted with circulating human T lymphocytes. Measurement of in vivo pharmacodynamic markers demonstrates PF-06671008-mediated T cell activation, infiltration and killing as the mechanism of tumor inhibition.

Electronic supplementary material

The online version of this article (doi:10.1007/s00262-017-2081-0) contains supplementary material, which is available to authorized users.

Introduction

Bispecific antibodies (bsAbs) have emerged in recent years as promising agents for immune-mediated tumor cell killing. The approval by the US Food and Drug Administration of the bispecific T cell engager (BiTE) blinatumomab [1] for the treatment of relapsed or refractory B cell precursor acute lymphoblastic leukemia, as well as preclinical studies with additional bispecific platforms including the dual-affinity retargeting (DART^®) scaffold [2], has demonstrated the potential role for bsAbs as potent cancer immunotherapies [3]. By co-engaging CD3 on the T cell and a cell surface antigen on the tumor, bsAbs circumvent the need for the recognition of the T cell receptor to bind antigen presented in the context of major histocompatibility class I on that target cell, and alternatively trigger potent T cell-mediated killing of the tumor cell. This expands the repertoire of T cells able to recognize the tumor and act as effector cells [4]. While bsAbs show great potential in opening up new therapeutic tumor targets, many challenges remain, including short half-life bsAb formats, and the translation of efficacy seen against hematological tumors into solid tumors in the clinic.

P-cadherin upregulation has been detected in various tumors. Overexpression is associated with invasiveness and/or poor prognosis in breast [5–7], endometrial [8], ovarian [9], colorectal cancers [10], intrahepatic cholangiocarcinoma and pancreatic cancer [11]. Low levels of P-cadherin gene expression has been detected on a diverse panel of normal tissues [12] and has been reported to be localized to the basal cell compartment of numerous normal epithelial tissues [13]. The observed difference in P-cadherin expression levels between tumor and normal tissues may offer a therapeutic window for a targeted therapy approach.

Here we report the preclinical pharmacological characterization of PF-06671008, a potent half-life extended anti-CD3/anti-P-cadherin LP-DART molecule previously described in Root et al. [14]. PF-06671008 was highly potent against a panel of tumor cell lines expressing P-cadherin co-cultured with T cells. Pharmacokinetic studies demonstrated a long half-life and sustained exposure over time in tumor-bearing mice. Weekly dosing resulted in regressions of established P-cadherin-expressing tumors in both cell line and patient-derived xenograft (PDX) in a dose- and P-cadherin expression-dependent manner. Additionally, pharmacodynamic markers of T cell infiltration and effector activity were measured and demonstrate PF-06671008-mediated T cell infiltration and killing as the mechanism of tumor inhibition.

Materials and methods

Cell culture

BT20, Chinese hamster ovary (CHO), Colo205, LoVo, HCC70, HCC1806, HCC1954, HCT116, HT29, Ls174T, NCI-H358 (H358), NCI-H1650 (H1650), NCI-H1975 (H1975), NCI-H2228 (H2228), SW480, and SW620 were obtained from the American Type Culture Collection. SUM-149PT (SUM149) cells were obtained from the Asterand Biorepository, and NCI-H322 (H322) was obtained from European Collection of Cell Cultures. All cell lines were obtained within the past 10 years and cultured according to the supplier-recommended conditions. Cell line authentication was completed for all lines by IDEXX Bioresearch.

Cell lines were transduced with pantropic retrovirus produced from pLPCX_LucSh, pMSCVpuro_LucSh, or pMSCVneo LucSh retroviral transfer vectors (Clontech) to introduce firefly luciferase followed by selection of drug-resistant pools using puromycin, geneticin, or zeocin (Gibco-Life Technologies). CHO cells were stably transfected to express human P-cadherin and firefly luciferase (CHO-pCad-Luc).

Isolation and expansion of human CD3+ lymphocytes

EDTA-treated blood was freshly isolated from healthy donors. PBMC was isolated by density gradient centrifugation layered over Histopaque 1077 in ACCUSPIN conical tubes (Sigma-Aldrich). CD3+ lymphocytes were isolated via a CD3+-negative isolation kit (Stem Cell Technologies). CD3+ cells were resuspended with OpTmizer CTS T cell expansion serum-free media supplemented with GlutaMax-1, 1% PenStrep, and 20 ng/mL recombinant IL-2 (Life Technologies). Dynabeads T-Expander CD3/CD28 magnetic beads (Life Technologies) were added at a ratio of 2 beads per cell and cultured for 1 week.

PF-06671008 cell binding

CHO parental, CHO cells overexpressing P-cadherin, as well as freshly isolated CD3+ cells from four separate donors, were assayed for PF-06671008 cell binding. CHO cells and CD3+ cells were incubated with a limiting dilution of PF-06671008. Binding of PF-06671008 was detected by staining with allophycocyanin (APC)-labeled F(ab’)2 fragment of goat anti-human IgG-Fc antibody (Jackson ImmunoResearch). Cells were analyzed using a LSRII with FACS Diva software (BD Biosciences). CHO cell samples were treated with 10 ng/mL of propidium iodide to exclude dead cells from the analysis.

CTL assay

Luciferase-expressing CHO cells with or without human P-cadherin expression were mixed with in vitro-expanded human CD3+ lymphocytes at a 10:1 E:T ratio. Cells were treated with a limiting dilution of PF-06671008 or control LP-DART. Luciferase activity remaining from the viable CHO target cells on day 2 was assessed using OneGlo luciferase assay reagent (Promega). Relative light unit (RLU) values were collected with an Envision plate reader (PerkinElmer). The percent-specific cytotoxicity was determined by the following formula, 100−100*(RLU value of PF-06671008 treated samples/average RLU of control bispecific treated samples at the same concentration).

Tumor cells engineered to express luciferase were mixed with freshly isolated CD3+ cells at a 3:1 E:T ratio with a limiting dilution of PF-06671008. The remaining luciferase activity from the viable tumor target cells on day 3 was assessed. Percent cytotoxicity was determined by the following formula, 100−100*(RLU value of PF-06671008 treated samples/RLU of untreated samples). N = 4 separate donor CD3+ lymphocytes were used per cytotoxicity assay and the average EC₅₀ was calculated for each tumor target tested.

Tumor cell lines were stained with PE-labeled anti-P-cadherin mAb (Pfizer PF-03732010 [15]) conjugated at a ratio of 1:1 PE label to mAb (eBioscience), or with a PE-labeled mouse IgG₁ control mAb (Biolegend). Propidium iodide was added prior to acquisition. QuantiBRITE PE bead kit (BD Biosciences) was utilized according to the manufacturer-recommended procedure to calculate the number of PE-labeled antibodies bound per cell (ABC or Ab/cell).

PF-06671008 T cell activation

Blood from four donors was collected in BD vacutainer Cell Preparation tubes (BD Biosciences) and processed to isolate PBMC. PBMC and HCT116 tumor cells were mixed at a ratio of 10:1. PF-06671008 or control LP-DART was added to the cells including untreated controls. Cell samples were collected following 48-h incubation. Fc receptors were blocked with Trustain Fx (Biolegend), followed by a staining cocktail consisting of the following anti-human antibodies: Pacific Blue-labeled anti-CD3 (BD Biosciences), PE-labeled anti-CD4 (BD Biosciences), Alexa 647-labeled anti-CD8 (BD Biosciences), Alexa 488-labeled anti-CD56 (Biolegend), PerCP-Cy5.5-labeled anti-4-1BB (Biolegend), and PE-Cy7-labeled anti-CD69 (Biolegend).

Pharmacokinetic studies in HCT116 tumor-bearing mice

Six- to eight-week-old NOD.Cg-Prkdc ^scid IL2rg ^tm1Wjl/SzJ, NOD scid gamma (NSG) mice were obtained from the Jackson Laboratory. An i.v. dose of PF-06671008 at 0.5 mg/kg or 0.05 mg/kg was given to HCT116 tumor-bearing mice followed by serum and tumor collection ranging from 5 min to 240 h post-dose with n = 3 samples collected per time point. Tumors were homogenized in lysis buffer (Sigma). PF-06671008 was measured from serum and tumor homogenates using a ligand binding assay. Plates were coated with a polyclonal goat antibody recognizing the CD3 scFv domain and blocked with 0.5% BSA followed by sample addition. Goat anti-human IgG-biotin (Qualex) was added followed by streptavidin–HRP. Plates were developed using tetramethylbenzidine (KPL) and sulfuric acid stop reagent. Absorbance was read on a SpectraMax plate reader (molecular devices) at OD450 nm.

Human T cell engraftment in established tumor model

Tumor cells (5 × 10⁶ HCT116, 5 × 10⁶ SW480, 2 × 10⁶ Ls174T, or 5 × 10⁶ SW620) were implanted s.c. into the right flank of 6–8-week-old NSG female mice mixed with 4 mg/mL Cultrex basement membrane extract (Trevigen). Seven days prior to randomization, mice were inoculated with 5 × 10⁶ freshly isolated PBMC i.p. One week following PBMC implant, tumor volumes were measured and blood samples were collected for flow cytometry. Whole blood samples were lysed with BD Pharmlyse solution (BD Biosciences) and stained with anti-antibodies CD3 Pacific Blue, CD4 Alexa 488, and CD8 APC, and analyzed. The percent CD3+ of total lymphocytes was determined for each animal. Two-parameter randomization was used to establish treatment groups of n = 10 animals per group providing equal weighting between tumor volume and CD3+ cell engraftment. PF-06671008, control LP-DART, or PBS as vehicle was dosed i.v. Tumor measurements were collected twice weekly with continuous health monitoring.

HCT116 tumor-bearing mice (n = 3) engrafted with PBMC and dosed with PF-06671008 were euthanized 6 h or 6 days following a single dose to assess human cytokine levels and tumor-infiltrating CD3+ T cells. Tumor samples were collected into gentleMACS C tubes containing tumor cell dissociation buffer (Miltenyi Biotec) and processed using the manufacturer’s protocol for soft human tumors using the gentleMACS tissue dissociator (Miltenyi Biotec). One million live cells were stained with anti-CD3 FITC (BD Biosciences). Propidium iodide was applied prior to data collection. The percent CD3+ of total live cell events was determined by FACS. Serum levels of IL-2 and IFNγ were assessed via Milliplex MAP high-sensitivity T cell panel multiplex assay (Millipore).

Human T cell adoptive transfer in established tumor model

For tumor cell line xenograft studies, 6–8-week-old NSG mice were inoculated s.c. with 5 × 10⁶ HCT116 cells or 2 × 10⁶ H1975 cells in the flank or 5 × 10⁶ SUM149 cells in the mammary fat pad. HCT116 and H1975 cells were suspended in PBS, while SUM149 cells were suspended in growth media and mixed 1:1 with Matrigel Basement Membrane Matrix (BD Biosciences). For PDX studies, NSG mice were implanted in the flank with PDX-CRX-11260 (Asterand, 144460A1) or PDX-BRX-12326 (Moores Cancer Center, AA1126) tumor models via trocar with tumor fragments propagated in NOD scid mice. Mice were randomized at the tumor size of 150–200 mm³. An initial dose of PF-06671008 or vehicle was administered to animals on day 0 and 2–2.5 × 10⁶ cultured T cells were inoculated the following day. Mice were dosed weekly up to five times, and all compound and T cell administrations were i.v. Tumor measurements were collected twice weekly with continuous health monitoring.

Histology and immunohistochemistry

Formalin-fixed paraffin-embedded tumor xenografts were stained with H&E, co-stained using sequential IHC for granzyme B/CD3/P-cadherin or granzyme B/CD8/CD4, and single stained for cleaved caspase 3. Staining antibodies include anti-granzyme B clone 11F1 (Novocastra), anti-CD3 clone SP162 (Spring Bioscience), and anti-P-cadherin clone 56 (BD Bioscience) conjugated with the Alexa Fluor 488 protein labeling kit (Life Technologies), anti-CD4 clone EP204 (Cell Marque), anti-CD8 clone SP16 (Spring Bioscience), and anti-cleaved caspase 3 clone 5AE1 (Cell Signaling Technologies). Two independent investigators performed light microscopic evaluation of IHC and H&E-stained sections.

Whole slide digital images were opened in HALO image analysis software (Indica Labs) to evaluate the percent cell death. Two tumor sections were analyzed and n = 4–5 tumors per treatment group. Sums for total tumor area, total cell death area, and cell death tissue percent were calculated for each tumor. Cleaved caspase 3 stained slides were scanned at 20X using a Leica Biosystems AT2 whole slide scanner. A Tissue Studio Ruleset (Definiens) was optimized to evaluate the presence of cleaved caspase 3 within the viable tumor excluding regions of necrotic and irrelevant tissue. Percent cleaved caspase 3-positive cells relative to all cells within the viable regions of the tumor were calculated.

Results

P-cadherin-dependent T cell activation and cytotoxicity

PF-06671008 was constructed as a stable bispecific diabody fusion protein targeting both human CD3 expressed by T cells, and P-cadherin expressed by a variety of human tumors [14]. PF-06671008 demonstrated binding to isolated T lymphocytes measured by FACS with an average calculated EC₅₀ for binding of 12.8 ± 5.87 nM (Fig. 1a). To demonstrate PF-06671008 binding to P-cadherin, Chinese hamster ovary (CHO) cells expressing P-cadherin were tested for binding with PF-06671008 compared to parental CHO cells. Figure 1b demonstrates binding of the bispecific molecule to CHO cells expressing human P-cadherin with a calculated EC₅₀ for binding of 1.36 nM.

Fig. 1.

PF-06671008 binds to CD3 and P-cadherin and directs T cell cytotoxicity to P-cadherin-expressing cells. a Human CD3+ T lymphocytes and b P-cadherin-expressing (triangle) or non-expressing CHO cells (circle) were co-incubated with increasing concentrations of PF-06671008 and detected with APC-labeled anti-human IgG-Fc secondary by FACS. MFI values were plotted versus the PF-06671008 concentration Log₁₀. c P-cadherin-expressing and non-expressing CHO cells engineered to express firefly luciferase were mixed with expanded human CD3+ lymphocytes with increasing concentrations of PF-06671008 or control. Percent cytotoxicity of the P-cadherin-expressing CHO cells (triangle) and parental CHO cells (circle) was plotted against PF-06671008 concentration Log₁₀

To confirm PF-06671008 mediates P-cadherin-dependent killing of target cells by human CD3+ T cells, CHO cells and P-cadherin-expressing CHO cells were modified to express firefly luciferase and used as target cells in CTL assays. In vitro-expanded CD3+ T cells were incubated with luciferase-expressing target cells at a 10:1 E:T ratio with increasing concentrations of PF-06671008 normalized to the equivalent treatment with control LP-DART consisting of the CD3 binding domain paired with an irrelevant scFv against fluorescein. Following 48 h, the viability of the target cells was assessed by luciferase assay. PF-06671008-directed potent cytotoxic activity against the P-cadherin-expressing CHO cells, EC₅₀ = 0.41 pM. PF-06671008 failed to direct cytotoxicity against the parental cells (Fig. 1c).

PF-06671008 activity correlates with target cell P-cadherin expression

Quantitative FACS was used to assess the relative expression level of endogenous P-cadherin across a panel of tumor cell lines engineered to express firefly luciferase. The same panel was then used as target cells for CTL assays using freshly isolated T cells. The calculated P-cadherin receptor density values of the panel of 16 tumor target cell lines ranged from 802 ± 403.1 (Colo205) to 37582 ± 734 (H1650) ABC (Supplemental Table 1). T cells and tumor cells were mixed at a 3:1 E:T ratio and incubated with PF-06671008. The viability of the tumor target cells was assessed by luciferase assay and the average EC₅₀ value for tumor cell killing was calculated from four individual T cell donors for each tumor cell line. The cytotoxicity EC₅₀ values ranged from 0.11 pM (HCC1954) to 1.44 nM (HT29). Linear regression analysis was performed by comparing the average EC₅₀ value for CTL activity to the average P-cadherin receptor expression for each individual tumor cell line (Fig. 2a). While there was variation across the panel tumor cell lines tested (r ² = 0.4602), a significant relationship was found between CTL EC₅₀ potency values relative to receptor density values (p value = 0.0039).

Fig. 2.

PF-06671008 directs T cell cytotoxicity and activation in response to human tumor cells dependent upon P-cadherin expression. a A panel of human tumor cell lines were used as cytotoxicity targets to determine EC₅₀ for cytotoxicity and assessed for human P-cadherin receptor expression by quantitative FACS. The Log₁₀ average EC₅₀ for cytotoxicity was plotted in a linear regression curve fit versus the Log₁₀ average ABC value for P-cadherin binding. Histogram plots of b CD69 and c 4-1BB expression by the CD3+ CD56− subset of PBMC are shown. Dose–response curves of d percent CD69 positive and e percent 4-1BB positive of CD3+ CD4+ (triangle) or CD3+ CD8+ (circle) PBMC are shown

PF-06671008 mediates dose-dependent T cell activation in the presence of P-cadherin-expressing tumor cells

T cell activation via PF-06671008-mediated synapse formation between effector and target cells was assessed by the induction of cell surface CD69 and 4-1BB expression [16–20]. CD69 and 4-1BB expression levels were assessed by FACS following co-incubation of HCT116 tumor cells with human PBMC plus PF-06671008 or controls. The expression of CD69 and 4-1BB by the CD3 + fraction of PBMC, PBMC mixed with HCT116, PBMC with HCT116 plus control LP-DART, or PBMC with HCT116 plus PF-06671008 was assessed (Fig. 2b, c). Induction of both CD69 and 4-1BB receptor expression was noted with PF-06671008 treatment but not controls. Dose-dependent increases of both CD69 and 4-1BB receptor expression were observed for both CD4+ and CD8+ T cell subsets (Fig. 2d, e). The EC₅₀ values calculated for CD69 increase were 4.87 and 4.45 nM for CD8+ and CD4+ T cells, respectively, while 4-1BB responses were calculated at 17.9 and 13.7 nM for the CD8+ and CD4+ subsets.

Pharmacokinetic properties of PF-06671008 in mice with established tumors

Pharmacokinetic (PK) studies were performed to assess the duration of PF-06671008 exposure in vivo. An i.v. dose of PF-06671008 at 0.5 or 0.05 mg/kg was given to HCT116 tumor-bearing NSG mice followed by serum and tumor collection. There was an approximate dose linear increase in systemic exposure as determined by C _max and AUC following i.v. administration from 0.05 to 0.5 mg/kg (Supplemental Table 2). Half-life ranged from 3.7 to 6.0 days. The tumor to serum ratio was 17 and 26% for 0.05 and 0.5 mg/kg doses, respectively. The T _max occurred at 0.25 h in serum and 48 h in tumor. Serum PK was additionally assessed in HCT116 tumor-bearing mice with PBMC engraftment. The overall kinetic parameters were similar to those observed in mice without PBMC engraftment. Previous studies of unrelated DART molecules in several strains of mice have shown rapid clearance, with T _1/2 values ranging from 2.4 to 3.6 h [21].

PF-06671008 mediates tumor regressions in mice with established tumor xenografts and engrafted T cells

To assess the anti-tumor effects of PF-06671008 in vivo, s.c. tumor xenografts were implanted into NSG animals followed by i.p. injection of PBMC. CD3+ lymphocyte engraftment and expansion in the mouse [22] was monitored in peripheral blood by FACS 1 week post-PBMC implantation. Randomization of the groups, based upon tumor size and CD3+ cell engraftment, was performed 1 day prior to the initial treatments. Four colorectal tumor xenografts, HCT116, SW480, Ls174T and SW620, with a range of P-cadherin expression were selected for in vivo evaluation. HCT116, SW480, Ls174T and SW620 tumor cells express 41795 ± 2807, 12086 ± 2939, 3970 ± 1021, and 1066 ± 291 P-cadherin-specific ABC in vitro, respectively. In vivo tumor growth response to weekly doses of PF-06671008, control LP-DART, or vehicle was assessed (Fig. 3a–d). HCT116 xenografts demonstrated tumor regressions when dosed with PF-06671008 at 0.5 mg/kg and tumor growth stasis at 0.05 mg/kg. SW480 xenografts demonstrated tumor regressions when dosed with PF-06671008 at 0.5 mg/kg with modest tumor growth delay observed at 0.05 mg/kg. Ls174T xenografts showed modest tumor growth delay at 0.5 mg/kg PF-06671008, while SW620 showed no response to PF-06671008. Tumor regression or stasis was not observed in animals dosed with vehicle or the control.

Fig. 3.

PF-06671008 directs P-cadherin-dependent growth inhibition to established tumors in human PBMC-engrafted NSG mice. a HCT116, b SW-480, c Ls174T, and d SW620 xenograft tumors were subcutaneously implanted to NSG mice engrafted with human PBMC. Animals were weekly dosed with vehicle (circle), 0.5 mg/kg control LP-DART (square), 0.05 mg/kg PF-06671008 (vertical triangle), or 0.5 mg/kg PF-06671008 (inverted triangle). The average tumor volume (mm³ ± SEM) is plotted against the time in days post-tumor implant

PF-06671008 treatment induces dose-dependent increases of tumor-infiltrating CD3+ cells and serum cytokines

CD3 IHC staining of HCT116 tumor samples collected from PBMC-engrafted mice 6 days following treatment with PF-06671008 versus negative controls demonstrated CD3+ cell infiltration to tumors treated with PF-06671008 (Fig. 4a). To demonstrate quantitative dose-dependent T cell infiltration, tumors were digested and assessed for CD3+ lymphocyte infiltration by FACS. Significant increases of CD3+ lymphocytes as a percentage of the total live cell events were observed with increasing dose of PF-06671008. The percent CD3+ lymphocytes of live cell events ranged from 2.05 ± 0.85 in the untreated tumors to 8.83 ± 3.14, 30.13 ± 11.32, and 58.33 ± 7.04 in the 0.01, 0.05 and 0.5 mg/kg dose groups, respectively (Fig. 4b).

Fig. 4.

PF-06671008 increases human CD3+ infiltration and human cytokine expression in mice bearing HCT-116 tumors with human PBMC engraftment. a Representative images of CD3+ IHC staining (brown) with H&E counterstain (blue) of untreated, 0.5 mg/kg control LP-DART, and 0.5 mg/kg PF-06671008 treated tumor samples are shown. Micron bars 100 µM. b The percent human CD3+ of all viable cells isolated from HCT116 tumor samples is plotted for each individual sample from each dose group compared via one-way ANOVA and Dunnett’s multiple comparisons test to determine significance. c Human IL-2 and d IFNγ cytokine expression (pg/mL) are plotted for each individual sample and dose groups compared via one-way ANOVA and Dunnett’s multiple comparisons test to determine significance

Circulating human cytokine levels from PF-06671008-treated tumor-bearing mice were measured. Mice bearing HCT116 tumors with confirmed engraftment of CD3+ T cells were treated with PF-06671008 or control LP-DART. Serum samples collected 6 h following treatment were assayed for human IL-2 and IFNγ (Fig. 4c, d). Both human cytokines were detected from mouse serum samples regardless of treatment condition with high levels of human IFNγ measured at baseline in untreated PBMC-engrafted mice consistent with developing GvHD. Significant elevations (p < 0.05) of both IL-2 and IFNγ were observed from samples collected from mice treated with 0.5 mg/kg PF-06671008 (3712.2 ± 1858.5 pg/mL IL-2 and 13456.2 ± 2661.8 pg/mL IFNγ) compared to vehicle-treated animals (161.7 ± 17.8 pg/ml IL-2 and 6658 ± 1298.1 pg/ml IFNγ). No elevation of serum cytokine was observed from animals treated with 0.5 mg/kg control over vehicle.

PF-06671008 mediates tumor regressions in mice with established tumor xenografts and adoptively transferred human T cells

PBMC engraftment of NSG animals enabled assessment of tumor growth inhibition of P-cadherin-expressing xenografts driven by the expanding human CD3+ lymphocyte subset. However, the in vivo expansion of CD3+ lymphocytes limited the evaluation window to roughly 3 weeks of tumor growth following initial dose before clinical signs of GvHD required euthanasia. A model was developed incorporating i.v. adoptive transfer of in vitro-expanded human CD3+ T cells to tumor-bearing NSG animals followed by bispecific treatment, which extended the length of the tumor growth studies beyond the length of time afforded by the PBMC engraftment studies.

HCT116, SUM149, and H1975 s.c. xenograft tumors were established in NSG animals. Tumor-bearing animals were distributed to dose groups based upon tumor size and dosed with the indicated treatments. 24 h following treatment, expanded T cells were administered i.v. For each tumor model tested, 0.5 mg/kg weekly dosing of PF-06671008 resulted in complete regressions of the established tumors, while tumor growth delay followed by relapse was observed with 0.05 mg/kg PF-06671008 treatment (Fig. 5a, c). Vehicle treatment combined with adoptive transfer of human CD3+ T cells failed to delay tumor growth. Treatment of P-cadherin-positive tumors with PF-06671008 in the absence of human CD3+ cell transfer or treatment of animals with control LP-DART plus CD3+ cells also failed to delay tumor growth (Supplemental Fig. 1).

Fig. 5.

PF-06671008 treatment of established tumors followed by adoptive transfer of human CD3+ T cells directs tumor growth inhibition. a HCT116, b SUM149, and c H1975 xenografts were dosed i.v. with vehicle (circle), 0.05 mg/kg PF-06671008 (square), or 0.5 mg/kg PF-06671008 (inverted triangle) followed by human T cell transfer. The average tumor volume (mm³ ± SEM) is plotted versus days post-treatment. d PDX-CRX-11260, adenocarcinoma of the large intestine, and e PDX-BRX-12326, ER⁻PR⁻Her2⁻ ductal carcinoma of the breast, were dosed with vehicle, 0.05 mg/kg PF-06671008, or 0.5 mg/kg PF-06671008 followed by human T cell transfer. Individual animal tumor growth plots versus time from tumor implant are shown

In vivo expansion of the CD3+ cells in response to PF-06671008 treatment of HCT116 tumor-bearing mice was examined by FACS. Significant expansion of the CD3+ cell subset as a percentage of the total circulating lymphocyte population was evident with both 0.05 and 0.5 mg/kg dose levels (Supplemental Fig. 2). No expansion of CD3+ T cells was detected from untreated mice or mice treated with control. Additionally, a separate group of five NSG mice without tumor was administered T cells and treated with 0.5 mg/kg PF-06671008. No expansion of human CD3+ T cells was detected confirming P-cadherin-expressing tumor was required to induce the proliferation of CD3+ T cells.

Two PDX models were selected for PF-06671008 efficacy evaluation based upon positive P-cadherin IHC staining (Supplemental Fig. 3). Subcutaneous implants of PDX-CRX-11260, adenocarcinoma of the large intestine, and PDX-BRX-12326, triple negative (ER⁻ PR⁻ Her2⁻) ductal carcinoma of the breast were implanted to NSG animals. All animals received CD3+ T cell adoptive transfer 24 h following the initial dose of PF-06671008 (Fig. 5d, e). Both PDXs demonstrated complete regressions with 0.5 mg/kg weekly dosing of PF-06671008 and mixed responses observed to 0.05 mg/kg weekly dosing.

PF-06671008 treatment induces T cell infiltration and markers of effector function in tumors

HCT116 xenografts were harvested for IHC after combined adoptive transfer of T cells plus a single 0.5 mg/kg dose of PF-06671008. Granzyme B, a lymphocyte serine protease, was selected as a marker of tumor cytotoxic T cell activity. Granzyme B+ CD3+ lymphocytes were observed in close proximity to P-cadherin-expressing tumor cells (Fig. 6a). Granzyme B displayed a polarized appearance in many cells suggesting localization at a bispecific-mediated immune synapse. Granzyme B immunostaining and polarization were observed not only on CD8+ cells, but also on CD4+ cells within the xenografts treated with PF-06671008 (Fig. 6b).

Fig. 6.

IHC illustrates the formation of the PF-06671008-mediated immune synapse and tumor killing. a, b HCT116 or c, d SUM149 tumor samples were examined using IHC at 6 and 7 days, respectively, following treatment with PF-06671008 or vehicle control. a Granzyme B/CD3/P-cadherin triple IHC demonstrates the relationship between P-cadherin + tumor cells (green), CD3+ infiltrating T lymphocytes (pink), and expression of Granzyme B (brown) in a PF-06671008-treated HCT116 tumor. Granzyme B+ CD3+ tumor-infiltrating lymphocytes are shown surrounded by P-cadherin+ tumor cells (insets). b Granzyme B/CD8/CD4 triple stain shows Granzyme B expression (brown) in CD8+ (green) and CD4+ (pink) cells in a PF-06671008- treated HCT116 tumor. c Cleaved Caspase-3 IHC, and d H&E staining of 0.5 mg/kg PF-06671008-treated (left panels) or control-treated (right panels) SUM149 tumors demonstrating increased cell death in response to treatment with PF-06671008. a Micron bars 200 µM, b–d 100 µM

A single dose of 0.5 mg/kg PF-06671008 to mice bearing SUM149 xenograft tumors with adoptively transferred human T cells was able to induce 9% increase of cleaved caspase 3 immunoreactivity (Fig. 6c) compared to controls 7 days after dosing (p < 0.0001, Supplemental Fig. 4). Widespread tumor cell death and minimal viable tumor cells were observed in H&E-stained SUM149 tumor sections treated with PF-06671008 (Fig. 6d). Tumors treated with the control LP-DART or vehicle had minimal cell death (< 1%) and the tumor cells appeared morphologically viable. Increases of 44% cell death were seen after PF-06671008 treatment as compared to controls (p < 0.0001, Supplemental Fig. 4).

Discussion

PF-06671008 is a novel bsAb designed to target P-cadherin-expressing solid tumors. This unique half-life extended bsAb is differentiated from previously described DART bsAbs [2, 21, 23, 24].

P-cadherin has been extensively studied as a target for cancer therapy, and while its role may vary across different cancer types, increased P-cadherin expression is well documented in multiple tumor types and in certain indications correlates with poor survival [5–11]. We have shown co-engagement of P-cadherin and CD3 by a LP-DART results in potent T cell activation and target cell lysis. Our in vitro analysis of bsAb-mediated effector function on a panel of tumor cell lines demonstrates a significant relationship between the level of P-cadherin cell surface expression and CTL activity.

Many CD3 bsAbs targeting solid tumors have been, or are being, tested in early phase 1 clinical trials [25–30]. To date, there is a noted lack of reports that have clinically validated the efficacy of CD3 bsAbs in solid tumor indications. We set out to explore the efficacy of PF-06671008 on a variety of solid tumor mouse models. Previously we tested PF-06671008 in a co-implantation model where both tumor and T cells are mixed at a set E:T ratio, implanted in mice and dosed daily for a limited number of days[14]. Using this model, we have demonstrated very potent inhibition of tumor growth at doses comparable to what has been reported by others [31]. While one can screen many candidates in a rapid manner using the co-implantation model, it has limitations. The lack of an established tumor and its associated immunosuppressive microenvironment may overestimate the potency and efficacy of the T cell retargeting bsAb, and does not require T cell infiltration into the tumor, a known mechanism by which tumors can resist immune recognition [32].

To overcome these limitations, we explored established tumor models with both engrafted PBMC and adoptively transferred T cells to identify a model that would better evaluate the extended half-life format in vivo and allow for the measurement of tumor regressions. First, we tested the efficacy of PF-06671008 in the human PBMC-engrafted and established tumor model. This model involves the engraftment and expansion of PBMC in mice with implanted tumors. With this approach, we were able to observe dose-dependent bsAb-mediated regressions of tumors expressing moderate-to-high levels of P-cadherin when dosed with 0.5 mg/kg PF-06671008 weekly. This in vivo model allowed us to measure dose-dependent T cell infiltration into the tumors. Our data suggest once a threshold of bsAb-mediated T cell infiltration is reached, the ensuing cytolytic effector activity is sufficient to regress established tumors. The limited duration of dosing in this model, due to the progressing T cell-mediated graft versus host response [22] did not allow us to observe sustained regressions; therefore, we chose to pursue a second in vivo established tumor model, in which we would be able to follow individual mice out further in time.

In the second established tumor model evaluated, dosing of PF-06671008 was followed by the adoptive transfer of expanded T cells into tumor-bearing mice. The weekly dosing regimen resulted in regression of tumors expressing moderate-to-high levels of P-cadherin. Moreover, this model allowed us to observe and analyze the mice for an extended period of time with limited graft versus host responses. The success of this model also led us to pursue efficacy studies of PDX models. It was noted, with the limited number of models we tested, that mice bearing PDXs were more responsive to the lower 0.05 mg/kg dose compared with those bearing cell line-derived xenografts.

Recent reports suggest that pharmacology studies conducted with PDX tumors are more predictive for clinical outcome compared to conventional, cell line-derived xenograft models [33]. However, since bsAbs have yet to be clinically validated for efficacy in solid tumor indications, it is very challenging to translate preclinical data toward predicting patient responses. Additionally, in vivo tumor models that require implantation of a limited number of human T cells into highly immunodeficient mice may overestimate the dose and exposure required to achieve efficacy with a bsAb approach in an immunocompetent patient. Still, our goal was to increase the stringency by which we tested our bsAbs preclinically, to instill greater confidence that PF-06671008 may achieve clinical efficacy. The results of the in vitro and in vivo studies reported here directly supported our decision to test PF-06671008 in a Phase 1 dose escalation study in solid tumors (NCT02659631).

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

ABC

Antibodies bound per cell

APC

Allophycocyanin

BiTE

Bispecific T cell engager

bsAb

Bispecific antibody

CHO

Chinese hamster ovary

DART^®

Dual-affinity retargeting

LP-DART

Half-life extended dual-affinity retargeting

NSG

NOD scid gamma

PDX

Patient-derived xenograft

PK

Pharmacokinetic

RLU

Relative light unit

Author contributions

Timothy S. Fisher: writing, editing, experimental design, execution and interpretation of results. Andrea T. Hooper: writing, editing, experimental design and interpretation of results. Justin Lucas: writing, experimental design and execution. Tracey H. Clark: writing, experimental design and execution. Allison K. Rohner: experimental design and execution. Bryan Peano: experimental design and execution. Mark W. Elliott: experimental design and execution. Konstantinos Tsaparikos: experimental design and execution. Hui Wang: editing, experimental design and interpretation of results. Jonathon Golas: experimental design and execution. Maria Gavriil: experimental design and execution. Nahor Haddish-Berhane: experimental design and interpretation of results. Lioudmila Tchistiakova: experimental design and interpretation of results. Hans-Peter Gerber: experimental design and interpretation of results. Adam R. Root: experimental design and interpretation of results. Chad May: writing, editing, experimental design, and interpretation of results.

Compliance with ethical standards

Funding

No relevant funding.

Conflict of interest

The authors declare they have no conflict of interest.

Ethical approval and ethical standards

All animal experimental procedures complied with the Guide for the Care and Use of Laboratory Animals (Institute for Laboratory Animal Research, 1996) and were approved by the Pfizer Global Research and Development Institutional Animal Care and Use Committee. All human blood samples from healthy donors were collected through the Pfizer Global Occupational Health and Wellness research support program under review of the Institutional Review Board provided with informed consent compliant with guidelines set forth by the International Conference on Harmonization on Good Clinical Practice, as well as local regulatory and legal requirements.