Purpose:

Gastrointestinal cancers remain areas of high unmet need despite advances in targeted and immunotherapies. Here, we demonstrate potent, tumor-selective efficacy with PF-07062119, a T-cell engaging CD3 bispecific targeting tumors expressing Guanylyl Cyclase C (GUCY2C), which is expressed widely across colorectal cancer and other gastrointestinal malignancies. In addition, to address immune evasion mechanisms, we explore combinations with immune checkpoint blockade agents and with antiangiogenesis therapy.

Experimental Design:

PF-07062119 activity was evaluated in vitro in multiple tumor cell lines, and in vivo in established subcutaneous and orthotopic human colorectal cancer xenograft tumors with adoptive transfer of human T cells. Efficacy was also evaluated in mouse syngeneic tumors using human CD3ϵ transgenic mice. IHC and mass cytometry were performed to demonstrate drug biodistribution, recruitment of activated T cells, and to identify markers of immune evasion. Combination studies were performed with anti–PD-1/PD-L1 and anti-VEGF antibodies. Toxicity and pharmacokinetic studies were done in cynomolgus macaque.

Results:

We demonstrate that GUCY2C-positive tumors can be targeted with an anti-GUCY2C/anti-CD3ϵ bispecific, with selective drug biodistribution to tumors. PF-07062119 showed potent T-cell–mediated in vitro activity and in vivo efficacy in multiple colorectal cancer human xenograft tumor models, including KRAS- and BRAF-mutant tumors, as well as in the immunocompetent mouse syngeneic tumor model. PF-07062119 activity was further enhanced when combined with anti–PD-1/PD-L1 treatment or in combination with antiangiogenic therapy. Toxicity studies in cynomolgus indicated a monitorable and manageable toxicity profile.

Conclusions:

These data highlight the potential for PF-07062119 to demonstrate efficacy and improve patient outcomes in colorectal cancer and other gastrointestinal malignancies.

Translational Relevance

Colorectal cancer is an area of high unmet need and a leading cause of cancer-related deaths worldwide. The majority of colorectal cancers is microsatellite stable, frequently with mutated KRAS or BRAF oncogenes, resulting in poor prognoses, and shows a lack of response to currently approved immunotherapy. T-cell engaging CD3-bispecific antibodies hold potential as potent therapeutics against solid tumors. Here, we demonstrate T-cell–mediated antitumor activity with PF-07062119, a novel CD3 bispecific against tumors expressing Guanylyl Cyclase C (GUCY2C), a target expressed widely across colorectal cancer and other gastrointestinal tumors. PF-07062119 shows efficacy in multiple colorectal cancer models independent of their KRAS or BRAF mutational status, because bispecific activity is dependent on GUCY2C expression. We also show significant combination benefits with checkpoint blockade and antiangiogenesis therapy. PF-07062119 has a well-tolerated toxicity profile in cynomolgus monkeys. Accordingly, PF-07062119 warrants further clinical investigation for the treatment of colorectal cancer and other gastrointestinal cancers.

Gastrointestinal malignancies, including colorectal cancer, gastric cancer, and esophageal cancer, continue to be areas of high unmet medical need despite advances in targeted therapies (1, 2). Colorectal cancer alone is a worldwide issue affecting both men and women, and responsible for 9.2% of all cancer deaths. The lack of response to targeted therapy, such as anti-EGFR antibodies, has been correlated with mutations in the KRAS and BRAF oncogenes (3–6). In addition, immunotherapies, such as immune checkpoint inhibitors, have failed to show significant survival benefit in most patients with colorectal cancer, owing to low tumor mutational burden and reduced density of immune infiltration (7, 8).

The potential of redirected T-cell therapies has been demonstrated by the approval of blinatumomab in hematologic malignancies, and more recently by reports of early clinical activity with CD3-bispecific antibodies targeting solid tumors, such as colorectal and prostate cancers (8, 9). CD3 bispecifics hold potential as potent cancer therapeutics as they recruit and activate a broad repertoire of T cells against tumor cells expressing a tumor-associated cell surface antigen (10). They circumvent the need for T-cell receptor engagement with MHC Class I in complex with antigenic peptide, and instead recruit T cells to target cells expressing cell surface antigen. One arm of the bispecific binds to a tumor-associated cell surface antigen, and the other arm binds to the CD3ϵ protein on T cells, leading to a cytotoxic T lymphocyte (CTL) response against tumor cells.

Given the high potency of CD3 bispecifics, it is important to demonstrate tumor selective activity with an antigen that has minimal or restricted expression in normal tissues. Guanylyl Cyclase C (GUCY2C) is expressed across gastrointestinal malignancies including more than 90% of colorectal cancer across all stages and more than 50% of gastric or gastroesophageal junction cancer (11–13). GUCY2C is normally involved in maintaining intestinal homeostasis (14) and is activated by the peptide hormones guanylin and uroguanylin (15, 16), as well as by the Escherichia coli heat stable (ST) enterotoxin, which causes familial diarrhea (17, 18). GUCY2C expression in normal tissues is largely restricted to the apical side of intestinal epithelial tight junctions (19). Because tumors have disrupted tight junction architecture (20, 21), this may allow for preferential uptake of GUCY2C-targeted biologics, thereby making it an attractive antigen for targeted therapy approaches.

Here, we show that an anti-GUCY2C/anti-CD3ϵ bispecific has preferential uptake in tumors versus normal tissues in mouse tumor models of colorectal cancer, leading to antitumor efficacy in the absence of toxicity to target-expressing tissues. The therapeutic candidate, PF-07062119, demonstrates potent T-cell–dependent efficacy in vitro, as well as in vivo in several GUCY2C-expressing human cell line– and patient-derived xenograft (PDX) tumor models, a colon orthotopic xenograft model, as well as in a mouse syngeneic tumor model. Furthermore, the potential benefits of combining PF-07062119 therapy with immune checkpoint inhibitors as well as anti-VEGF therapy are described. An exploratory toxicity study in cynomolgus monkeys showed T-cell activation and elevations in cytokines, and in-life findings that were manageable with oral and/or subcutaneous fluids. No erosions, ulcerations, evidence of overt necrosis, or loss of overlying epithelium were observed in the intestinal tract, where GUCY2C is expressed. These studies collectively provide rationale for clinical evaluation of PF-07062119 for the treatment of gastrointestinal cancers.

Generation of anti-GUCY2C antibodies

Anti-GUCY2C antibodies were generated using hybridoma technology. Mouse lymphoma 300.19 cells overexpressing human GUCY2C (300.19/huGUCY2C) were injected as immunogens into 8-week-old female Balb/c mice for the generation of hybridomas. Balb/c mice were immunized with 5 × 106 300.19/huGUCY2C cells twice per week for 1 month without adjuvant intraperitoneally. Antibodies derived from mouse immunizations were humanized using conventional grafting methods (22).

Characterization of KRAS and BRAF mutational status of colorectal cancer cell line xenograft and PDX models

Mutational status of KRAS and BRAF oncogenes was determined for cell line xenograft (CLX) and PDX models used for evaluating antitumor activity of PF-07062119. The OASIS 3.0 database was queried for all cell lines and PDX models, and KRAS or BRAF mutations in these models were identified (23). KRAS and BRAF mutational status in LS174T has been described earlier (24, 25).

Peripheral blood mononuclear cell collection and isolation of human T cells for in vitro and in vivo studies

Whole blood was collected from healthy donors and immediately treated with the anticoagulant 10 mmol/L ethylenediaminetetraacetic acid (EDTA). All peripheral blood mononuclear cell (PBMC) and T-cell isolations were carried out at room temperature. Each 25 mL of blood was mixed with 10 mL of Dulbecco's Phosphate-Buffered Saline (DPBS) containing 2 mmol/L EDTA and layered above the frit of a 50 mL Accuspin tube preloaded with 15 mL of Histopaque-1077 density gradient media. Tubes were spun at 800 × g for 20 minutes, and the material above the frit was decanted into a fresh 50 mL conical tube and spun again at 800 × g for 10 minutes. Following centrifugation, the supernatant was discarded, and the pellet was resuspended in DPBS containing 2 mmol/L EDTA. PBMCs were spun at 200 × g for 10 minutes, the supernatant was discarded, and the cells were resuspended to 5 × 107 cells/mL in Robosep buffer. T cells were isolated using the EasySep human T-cell enrichment kit (Stem Cell Technologies) according to the manufacturer's protocol. T cells were used at this stage for in vitro experiments or expanded and activated further for in vivo studies. T cells were activated using a Human T-cell Activation/Expansion kit (Miltenyi) following the manufacturer's protocol using T-cell media containing X-Vivo 15 with 5% human serum AB, 1% Penn/Strep, and 0.01 mmol/L 2-mercaptoethanol. After 48 hours of T-cell activation, T cells were transferred to a G-Rex cell culture device (Wilson Wolf) for expansion, and human IL2 (Shenandoah Biotechnology) was added to the media at a final concentration of 5 ng/mL and replenished after 2 days. T cells were harvested 5 days after expansion. At the time of harvest, beads were removed with a magnet, and cells were resuspended in DPBS at 1 × 107 cells/mL for in vivo adoptive transfer studies.

Immunohistochemistry

Tumor samples were fixed in 10% neutral-buffered formalin for 48 to 96 hours, trimmed, and processed or dehydrated overnight through a series of xylene, graded alcohols (100%, 95%, 70% reagent alcohol), and paraffin. Samples were then embedded in paraffin and cooled prior to being sectioned on a microtome at 5 μm. Sections were floated in deionized water, placed on glass slides, and allowed to dry. Slides were placed in a baking oven at 60°C for approximately 30 minutes, then deparaffinized in xylene, and rehydrated through a graded series of alcohols to deionized water.

For hematoxylin and eosin (H&E) staining, sections were placed in CAT hematoxylin (Biocare Medical) for 1 minute followed by rinsing in tap water to blue. Sections were differentiated with 80% reagent alcohol, placed in Edgar Degas Eosin (Biocare Medical) for 90 seconds, dehydrated in 95% then 100% reagent alcohol, and cleared in xylene. Slides were coverslipped using Permount medium.

For bispecific biodistribution staining, following euthanasia and prior to tissue collection, animals underwent cardiac perfusion with saline to remove systemic therapeutic, which might interfere with IHC detection. To immunohistochemically detect the bispecific molecule, sections underwent heat-induced epitope retrieval in Borg Decloaker (Biocare Medical) for 30 minutes, followed by peroxidase block in Peroxidazed 1 (Biocare Medical) for 10 minutes, and protein block with Background Punisher (Biocare Medical) for 10 minutes. Primary antibody rabbit mAb anti-Human IgG (EPR4421, Abcam ab109489) was applied at 1:200 (0.41 μg/mL) for 1 hour followed by MACH2 Rabbit HRP-Polymer (Biocare Medical) for 30 minutes. Color was developed using Betazoid DAB (Biocare Medical) chromogen for 5 minutes. After a rinse in water, the sections were counterstained for 30 seconds in Tacha's Hematoxylin (Biocare Medical), dehydrated in 100% reagent alcohol, and cleared in xylene. Slides were coverslipped using Permount medium.

For GUCY2C staining, sections underwent heat-induced epitope retrieval in Borg Decloaker for 30 minutes, followed by peroxidase block in Peroxidazed 1 for 10 minutes, and protein block with Background Punisher for 10 minutes. The primary antibody anti-GUCY2C (9H3-rabbit IgG) was applied at 2 μg/mL for 60 minutes followed by the labeled polymer MACH2 Rabbit HRP-Polymer for 30 minutes. Color was developed using Betazoid DAB chromogen for 5 minutes. After a rinse in water, the sections were counterstained for 10 seconds in Tacha's hematoxylin, dehydrated in 100% reagent alcohol, and cleared in xylene. Slides were coverslipped using Permount medium and examined under a microscope for Histo score (H-score) evaluation. The H-score was calculated by multiplying the estimated percentage of tumor cells with membrane (apical, complete circumferential, or incomplete circumferential) staining of each staining intensity (0 to 3; 0 = negative, 1 = low, 2 = medium, 3 = high) by the intensity value (0 to 3) and then adding together all 4 scores for a single score. The highest H-score obtainable by this method is 300, or 100% of cells staining at an intensity of 3. Colorectal tumors were also described by their grade of differentiation status as "well differentiated," "moderately differentiated," or “poorly differentiated” by examining the H&E-stained sections of all tumors. All slides were evaluated by a trained pathologist.

For CD3, Granzyme B, and PD-L1 expression in the LS0134 adoptive transfer efficacy study, tumors were harvested from mice treated with PF-07062119 at 1 mg/kg and 0.03 mg/kg, and from the vehicle-treated group. Tumors (n = 5/group) were formalin-fixed and paraffin-embedded as described above. Following epitope retrieval in Borg Decloaker, peroxidase block, and protein block as mentioned above, sections were stained with rabbit mAb anti-CD3 (SP162, Abcam ab135372) at 1:200 (0.075 μg/mL) or rabbit mAb anti–PD-L1 (SP142, Abcam ab228462) at 1:200 (0.385 μg/mL) for 1 hour followed by MACH2 Rabbit HRP-polymer for 30 minutes. Color was developed using Betazoid DAB chromogen for 5 minutes. For Granzyme b and CD3 sequential costaining, the tumor sections were incubated in mouse mAb anti-Granzyme B (11F1, Leica Biosystems) at 1:50 (0.94 μg/mL) for 1 hour, followed by MACH2 Mouse HRP-Polymer (Biocare Medical) for 30 minutes and Betazoid DAB chromogen for 5 minutes. Sections were rinsed in deionized water, and endogenous peroxidase activity was inactivated again with Peroxidazed 1 for 10 minutes. Nonspecific protein interactions were blocked again for 10 minutes with Background Punisher. Sections were incubated with anti-CD3 as described above, followed by MACH2 Rabbit HRP-polymer for 30 minutes and Vina Green (Biocare Medical) chromogen for 12 minutes. Immunostained sections were rinsed with deionized water, counterstained for 30 seconds in Tacha's hematoxylin, washed in tap water, dehydrated in alcohol, and cleared in xylene. All slides were coverslipped using Permount medium.

IHC analysis of CD3-positive T cells in the anti–VEGF-A and PF-07062119 combination study in PDX-CRX-11201 was done on tumors harvested on day 7 after the first bispecific dose, using the CD3 staining method described above.

Human T-cell adoptive transfer established tumor model

Female NOD-scid IL2Rγ-null (NSG) animals (Jackson Laboratory) were used for experiments under approval by the Institutional Animal Care and Use Committee, and all applicable animal care and use regulations, guidelines, and policies were followed. For xenograft studies, NSG mice were inoculated with LS1034, HT55, LS174T, or HCT116 cells in the flank in a total injection volume of 0.2 mL. For the LS1034, HT55, and HCT116 subcutaneous CLX models, 5 × 106 cells per mouse were administered in 50% Matrigel Basement Membrane Matrix. For the LS174T CLX model, 2 × 106 cells per mouse were administered in 50% Matrigel Basement Membrane Matrix. PDX-CRX-11201 (Asterand) tumor fragments (4 mm × 4 mm) were generated from 500 to 800 mm3 tumors, expanded, and then implanted subcutaneously in the flank of NSG mice.

Tumor measurements were collected using Vernier caliper, and volumes were calculated by use of the modified ellipsoid formula: ½ × length × width2. Mice were randomized and staged at tumor size of 150 to 200 mm3. An initial dose of PF-07062119, a nontargeted CD3-bispecific control, or PBS (vehicle) was administered to animals on day 0, and 2 × 106 cultured activated pan human T cells (containing CD8 and CD4 T cells) were inoculated the following day. Mice were dosed in 0.2 mL bolus injections weekly for 3 doses in all studies except in the single-agent efficacy study in PDX-CRX-11201, which was dosed weekly for 4 doses. In combination studies, combination agents were administered starting on day 0. Anti-human PD-L1 was administered at 10 mg/kg every 3 days for 6 doses, anti-human PD-1 was dosed at 5 mg/kg weekly for 3 doses, and anti–VEGF-A mAb (G6-31) that blocks both human and mouse VEGF (26) was dosed at 5 mg/kg every 3 days for 4 doses. All compound and T-cell administrations were intravenous via the lateral tail vein of each animal. Tumor measurements were collected twice weekly along with continuous monitoring for signs of a graft versus host response.

Tumor volume data were log-transformed after an adjustment was made for zero tumor size. A separate ANOVA analysis using all groups was performed for each day. From each analysis, a one-sided P value from a Student t pairwise comparison of each group with the control was reported. For observed differences between the combination agent treatments and their respective single-agent treatments, statistical analysis was performed to determine statistical significance (P values) for each difference. In the studies using anti–PD-L1 and anti–PD-1, the tumor size was log-transformed, and a one-sided t test of the log-transformed tumor sizes was used to assess significance. Because the difference observed varied with day, this assessment was performed separately for each day. For the study using anti–VEGF-A, many tumors had zero size, so a nonparametric Wilcoxon ranked sum test (one-sided with t-approximation) of tumor sizes was used in place of a parametric t-test. The P values reported in sections 7.4 and 7.5 are those for the last day of measurements because that P value is similar (in most case identical) across the last 5 days of measurements. A separate P value is computed for each single agent (when compared with the combination agent), and the more conservative (higher) of the two P values is reported for each combination being evaluated.

LS1034 colorectal orthotopic tumor model

Note that 5 × 106 LS1034 tumor cells expressing firefly luciferase were subcutaneously inoculated in female NSG donor mice. Three weeks after cell inoculation, the subcutaneous tumors (250 to 300 mm3) were harvested, and tumor fragments (4 × 4 mm) were prepared for orthotopic implants. Laparotomy was performed through a midline incision, and the cecum was exposed. Then LS1034-luc tumor fragment was sutured to the cecum adjacent to the ascending colon. The mice with orthotopic tumor implants were monitored weekly with the tumor growth assessed by In Vivo Imaging System (Perkin Elmer) scans of luciferin bioluminescence using Living Image version 4.3.1 software. Animals were staged to an average bioluminescence level of 24 to 26 × 106 photons/s and dosed weekly intravenously for 5 doses via tail vein injection with PF-07062119 at 0.3, 0.1, and 0.03 mg/kg or nontargeted CD3 bispecific at 0.3 mg/kg. Activated/expanded T cells were administered as described earlier for adoptive T-cell transfer, 24 hours after dosing with bispecifics.

CT26-mGUCY2C efficacy study in human CD3ϵ transgenic mice

To generate CT26-mGUCY2C cells, mouse Gucy2c nucleotide sequence (NM_001127318) encoding the mature extracellular domain, transmembrane domain, and the first three amino acids of the cytoplasmic domain (nucleotide sequence 194-1507 from GenBank accession # NM_001127318, corresponding to amino acid sequence V20-Y457 of GenBank accession # NP_001120790) was inserted into a mammalian expression vector downstream of the synthetic CAGGS promoter. The vector also contained the neomycin resistance gene for growth selection. CT26 cells (ATCC) were transfected with this expression construct, and a G418-resistant, single-cell clonal population was generated using limiting dilution.

CT26-mGUCY2C tumor cells were cultured in vitro with RPMI medium supplemented with 10% FBS in a humidified chamber at 37°C under 5% CO2 atmosphere. Cells were harvested during the exponential growth phase, and 1 × 106 cells were inoculated in each human CD3ϵ transgenic mouse (CrownBio) subcutaneously in the right rear flank region in 0.1 mL of PBS mixed with Matrigel (1:1) for tumor development. Mice were randomized (n = 5 mice/group) when mean tumor size reached approximately 68 mm3. On the day of randomization, mice were dosed with PF-07062119 at 3, 1, or 0.3 mg/kg, or with the nontargeted CD3-bispecific control at 3 mg/kg every 3 days for 6 doses. Tumor volumes were measured twice a week using calipers, and animals were monitored for changes in bodyweight, behavior, or signs of toxicity.

Pharmacokinetic measurements of GUCY2C(M)-CD3 in LS1034 adoptive transfer model

A qualified quantitative ligand-binding assay was used to measure GUCY2C(M)-CD3 in NSG mouse-diluted whole blood, using the Meso-Scale Discovery (MSD) assay platform. The assay used biotinylated goat anti-human IgG (Southern Biotech, 2049-08) coated on the surface of the MSD plate to capture GUCY2C(M)-CD3. After washing, the bound drug was detected with a mouse anti-human IgG Fc (Clone JDC10 Southern Biotech, 9040-01), which was labeled with ruthenium. After a final set of washes, tripropylamine was added and plates read on the MSD SECTOR Imager 6000. The electrochemiluminescent signal generated was proportional to the amount of bound drug. Sample concentrations were determined by interpolation from a standard curve that was fit using a 4-parameter logistic equation with 1/y^2 weighting. The range of quantitation in 100% diluted whole blood was 80.0 ng/mL to 5,120 ng/mL. The pharmacokinetic parameters were determined from individual animal data using noncompartmental analysis in Watson LIMS (Version 7.5, Thermo, Inc.). Concentrations below the limit of quantitation were not used in the calculations. In addition, pharmacokinetic (PK) data were also analyzed using a two-compartment PK model with linear elimination from the central compartment using Phoenix 64 Win Non Lin (Certara L.P.).

Cell culture

T84, LS1034, LS174T, and HCT116 cells were obtained from the ATCC. HT55 cells were obtained from the European Collection of Authenticated Cell Cultures (Sigma-Aldrich). T84 cells were cultured in a 1:1 mixture of Ham's F12 medium and DMEM with Glutamax and 10% FBS. LS1034 cells were cultured in RPMI-1640 with 10% FBS. HT55 cells were cultured in modified Eagle's medium with 20% FBS. LS174T cells were cultured in modified Eagle's medium with Glutamax and 10% FBS. HCT116 cells were grown in McCoy's 5a medium with 10% FBS. HCT116-hGUCY2C cells were generated by transducing HCT116 cells with human GUCY2C lentiviral particles (Origene, RC213901L3V) and selecting puromycin-resistant clonal populations.

T84, LS1034, LS174T, HT55, and HCT116 tumor cells were transduced with pantropic retrovirus produced from pMSCVpuro_LucSh or pMSCVneo_LucSh retroviral transfer vectors to introduce firefly luciferase followed by selection of drug-resistant pools puromycin or geneticin. HCT116 and T84 cells expressing luciferase were maintained under 1 μg/mL of puromycin drug selection. LS174T and LS1034 luciferase cells were maintained with 200 μg/mL, and HT55 luciferase cells were maintained with 500 μg/mL of Geneticin. All cells were grown in a humidified chamber at 37°C under 5% CO2 atmosphere.

GUCY2C receptor density measurements

Tumor cell lines were dissociated using enzyme free cell dissociation buffer (CDB; Gibco) for saturation flow cytometry. GUCY2C receptor density in cell lines was measured using an anti-GUCY2C (9H3-human IgG, Pfizer) conjugated at a 1:1 ratio to phycoerythrin (PE; Thermo Scientific). Cells (1 × 105) were stained with 10 μg/mL of the PE-labeled anti-GUCY2C mAb on ice for 1 hour. Cells were washed and resuspended in FACS buffer (PBS, 3% BSA) with 4′,6-diamidino-2-phenylindole (DAPI) and acquired using BD LSRII Fortessa with FACS Diva software version 8.0.1. QuantiBRITE PE-labeled beads were used in the same acquisition PE voltage settings to calculate the number of PE-labeled antibodies per cell (ABC) based on the background-corrected PE geometric median fluorescence intensity.

Characterization of PF-07062119 binding to human T cells and GUCY2C-expressing tumor cells

Freshly isolated human T cells and tumor cells (HCT116, HCT116-human GUCY2C, CT26, CT-26-mouse GUCY2C) dissociated with CDB (1 × 106 cells per sample) were blocked and washed in cold FACS buffer containing PBS and 3% BSA. Cells were stained in flow buffer containing PF-07062119 (0.25 μmol/L) on ice for 45 minutes. Cells were washed with cold PBS and resuspended in 100 μL of cold FACS buffer containing 1:100 dilution of goat anti-human PE (Jackson Immunoresearch, 109-115-098) and incubated in the dark on ice for 30 minutes. Cells were washed with cold PBS and resuspended in 200 μL of cold PBS with 3 μmol/L DAPI for data acquisition using a BD Fortessa X-20 or BD LSR II Fortessa with FACS Diva software (BD Biosciences). Data analysis was performed using Flowjo software v10.4.1 (Treestar Inc.).

CTL assay

To test PF-07062119 sensitivity on a panel of colorectal tumor cell lines endogenously expressing GUCY2C, monolayer cultures of firefly luciferase–expressing tumor cell lines were isolated from cell culture flasks using TrypLE dissociation reagent. Tumor cells were resuspended in R10 medium (RPMI, 10% FBS, 3 mL of 45% glucose). Five thousand tumor cells per well were plated into clear-bottom white wall 96-well plates (Costar). Human T cells were also resuspended in R10 media and added to tumor cells at an effector to target ratio (E:T ratio) of 5:1. The cells were treated with serial dilutions of PF-07062119 or non–targeted-CD3 control bispecific and incubated at 37°C under 5% CO2 for 48 hours. The luciferase signal in viable cells was measured in relative lights units using the neolite reagent read on a Victor plate reader (Perkin Elmer) at 0.1 seconds/well. EC50 values were calculated in Graphpad PRISM v7.04 using variable slope four parameter nonlinear regression analysis of percent cytotoxicity versus Log10 concentration of bispecifics. For each tumor cell line, 3 separate donor CD3+ lymphocytes were used in cytotoxicity assays, and the average EC50 for each cell line was calculated. PF-07062119 activity was also measured in GUCY2C-expressing tumor cells in the absence of T cells.

IFNγ induced in vitro by PF-07062119

To test PF-07062119–induced cytokine release, supernatants were collected at the end of the CTL assays described above and analyzed using a Milliplex MAP Human Cytokine/Chemokine Magnetic Bead Panel according to the manufacturer's guidelines, read on a Luminex 200 (Luminex Xmap Technology) with Luminex xPONENT software 3.1, and analyzed using MILLIPLEX ANALYST v5.1.0.0 to measure human IFNγ levels.

Mass cytometry deep phenotyping

The mass cytometry panel included the following metal-conjugated antibodies from Fluidigm: anti–CD45-89Y (3089003B), anti–CD127-176Yb (3176004B), anti–FoxP3-162Dy (3162011A), anti–PD-1-175Lu (3175008B), anti–CCR4-158Gd (3158006A), anti–Tim-3-153Eu (3153008B), anti–Lag3-150Nd (3150030B). In addition, the following antibodies were conjugated using Maxpar X8 antibody labeling kits (Fluidigm) for their corresponding metal conjugate—anti–CD3-154Sm (Tonbo Biosciences, 70-0038-U100), anti–CD4-145Nd (Tonbo Biosciences, 70-0049-U100), anti–CD8-168Er (Tonbo Biosciences, 70-0088-U100), anti–CD25-149Sm (BD Biosciences, 340739), anti–PD-L1-159Tb (Pfizer, Avelumab), anti–41-BB-161Dy (Pfizer proprietary, PF-05082566), anti–OX40-163Dy (Pfizer proprietary, PF-8600), anti–CD27-167Er (Tonbo Biosciences, 70-0279-U100), anti–CD45RA-171Yb (Tonbo Biosciences, 70-0458-U100), and anti–GITR-170Er (Pfizer proprietary, 10H2). These metal-conjugated antibodies were used in an antibody mastermix to label input T cells, as well as tumors harvested from the LS1034 adoptive transfer efficacy study in groups treated with PF-07062119 at 1 mg/kg and 0.03 mg/kg, 7 days after the first bispecific dose. Tumors were dissociated into single-cell suspensions using a tumor dissociation kit (Miltenyi 130-096-730) according to the manufacturer's guidelines. Viability staining was performed with 5 μmol/L final concentration of Cisplatin (Fluidigm) for 5 minutes in 1 mL of PBS (Ca/Mg free). Cell suspensions were washed and then stained with the antibody mastermix with 1 to 3 million cells in 100 μL of staining buffer. Cells were incubated at room temperature for 30 minutes before being washed and resuspended in nuclear antigen staining buffer (Maxpar Nuclear Antigen Stain Buffer Set, Fluidigm). Cells were washed in Maxpar Nuclear Antigen Staining Perm buffer before staining for intracellular antigens. After washing in PBS, cells were fixed in 2% paraformaldehyde (Electron Microscopy Sciences) overnight. Cells were pelleted and resuspended in freezing media containing 90% FBS and 10% DMSO and stored at -80°C until acquisition. Prior to acquisition, samples were thawed at 37 °C and washed in Maxpar Staining Buffer. Cell pellets were resuspended in Maxpar Fix/Perm buffer with 0.05 μmol/L Cell-ID Intercalator-Ir to stain DNA for cell-ID. After incubation at room temperature for 30 minutes, Cell-ID 20-Plex Pd Barcoding Kit (Fluidigm) was added to corresponding samples and mixed by pipetting up and down and incubated for an additional 30 minutes. Cells were washed in PBS twice, then washed in water prior to resuspending in water with 1:10 EQ Four Element Calibration Beads (Fluidigm). The concentration of the cell suspension was adjusted so samples are acquired at a rate of 250 to 500 events per second. A CyTOF 2 instrument with Helios upgrade was tuned using Maxpar tuning solution and quality controlled using EQ Four Element Calibration Beads. Samples were acquired, normalized, concatenated, and debarcoded using Fluidigm's CyTOF software version 6.7.1014. Debarcoded samples were uploaded to the Cytobank cloud-based analysis program for manual gating. CD45+ populations were exported from Cytobank and uploaded to a Pfizer cytometry analysis portal. This portal utilizes Cytofkit as the base for Rphenograph analysis and resulting tSNE maps. Data generated through the portal were then used for statistical proportion differential to determine clusters that differentially shift with treatment when compared with a control group. Phenotyping of these clusters was determined using heatmaps of each cluster for all markers used in the Rphenograph clustering. Resulting data from the clustered samples were used to generate heatmaps and tSNE maps in Tibco Spotfire software. The heatmap depicts expression of Arcsinh-transformed median intensities of each marker to represent the phenotypes of each cluster.

GUCY2C expression in gastrointestinal cancer and normal tissues

Although GUCY2C expression in colorectal, gastric, esophageal, and pancreatic cancers has been evaluated previously (11, 12), both membrane and cytoplasmic expressions of the target have been observed across these indications. Because our bispecific targeting modality relies on binding to cell surface GUCY2C, we sought to specifically characterize membrane expression of GUCY2C in primary gastrointestinal tumors and in normal tissues (Indivumed, Proteogenix, Avaden, University of Michigan, and Cornell University) using IHC with an anti-GUCY2C mAb (Fig. 1; Supplementary Table S1). GUCY2C was expressed in the majority of colorectal adenocarcinomas across all stages, including liver metastases (Fig. 1AF). It was also expressed in gastric (Fig. 1A and G; Supplementary Table S1) and esophageal adenocarcinomas (Fig. 1A and H; Supplementary Table S1). Moderately to well differentiated tumors tended to be associated with higher membrane expression of GUCY2C compared with those that were poorly differentiated (Supplementary Table S1). Relatively heterogeneous and low incidence of membrane expression was observed in pancreatic adenocarcinomas (Fig. 1A; Supplementary Table S1). In normal tissues, GUCY2C expression was primarily observed on the apical side of the colon (Fig. 1I) and small intestinal epithelium (Fig. 1J). A lower level of expression was present in prostatic apical epithelium, and all other tissues evaluated were negative (Supplementary Fig. S1).

Figure 1.

GUCY2C expression in gastrointestinal tumors and normal tissues. A, H-scores depicting GUCY2C cell surface expression across colorectal, pancreatic, gastric, and esophageal tumors (red line, median H-score). IHC showing GUCY2C expression (brown) in (B) colorectal adenocarcinoma stage I; (C) colorectal adenocarcinoma stage II; (D) colorectal adenocarcinoma stage III; (E) colorectal adenocarcinoma stage IV; (F) liver metastasis of colorectal adenocarcinoma (H, hepatocytes; T, tumor; dashed line, tumor/normal boundary); (G) gastric adenocarcinoma; (H) esophageal adenocarcinoma; (I) normal colon; and (J) normal small intestine. Scale bar, 200 μm.

Figure 1.

GUCY2C expression in gastrointestinal tumors and normal tissues. A, H-scores depicting GUCY2C cell surface expression across colorectal, pancreatic, gastric, and esophageal tumors (red line, median H-score). IHC showing GUCY2C expression (brown) in (B) colorectal adenocarcinoma stage I; (C) colorectal adenocarcinoma stage II; (D) colorectal adenocarcinoma stage III; (E) colorectal adenocarcinoma stage IV; (F) liver metastasis of colorectal adenocarcinoma (H, hepatocytes; T, tumor; dashed line, tumor/normal boundary); (G) gastric adenocarcinoma; (H) esophageal adenocarcinoma; (I) normal colon; and (J) normal small intestine. Scale bar, 200 μm.

Close modal

Tumor selective targeting with an anti-GUCY2C/anti-CD3ϵ bispecific

To test the ability to target GUCY2C-expressing tumors with an anti-GUCY2C/anti-CD3ϵ bispecific, we developed a heterodimeric diabody Fc-fusion protein comprised of two single-chain Fv (scFv) domains, one targeting GUCY2C and the other targeting CD3ϵ. The VH of one binding domain pair is joined with the VL of the other binding domain pair such that when the two constructs are coexpressed, the result is the formation of a diabody. These scFv domains are fused to the Fc domain of human IgG1 (Fig. 2A). Mutations (L234A, L235A, and G237A EU numbering) were introduced in the CH2 region of the Fc to reduce binding to Fc gamma receptor (FcγR; ref. 27). In addition, knob-in-hole mutations in the CH3 region of the Fc domain were used to facilitate correct heterodimeric pairing and bispecific purification (28, 29). The Fc domain fusion was also designed to extend the bispecific half-life and allow for less frequent dosing than antibody fragment–based bispecifics such as BiTEs (e.g., blinatumomab), which require constant i.v. infusion into patients via a pump. BiTEs contain only two tandem antibody variable domains (scFv-scFv) and no Fc domain. The diabody-Fc format is similar to the BiTE format in that it contains two scFv domains. However, the variable domains in the diabody are more tightly packed together than those of the BiTE, allowing for more potent killing activity (30), and introduction of the Fc further extends half-life from 1 hour to approximately 1 week (31).

Figure 2.

An anti-GUCY2C/anti-CD3ϵ bispecific selectively targets tumors versus normal tissue. A, Schematic of an anti-GUCY2C/anti-CD3ϵ bispecific human IgG1 FcγR- diabody. B, Pharmacokinetics of a murine GUCY2C cross-reactive bispecific GUCY2C(M)-CD3 at 1 and 0.3 mg/kg showing an estimated half-life of 11 days in LS1034 tumor–bearing female NSG mice with human adoptive T-cell transfer. C, IHC showing GUCY2C expression (left plot, brown staining) in LS1034 CLX tumor (top) and in the colon from an LS1034 tumor–bearing mouse (bottom). Middle plot shows biodistribution of the bispecific GUCY2C(M)-CD3 (brown) in the same LS1034 tumor (top) and lack of bispecific uptake in the colon (bottom), 7 days after dosing in an adoptive transfer study. Right plot shows infiltration of human CD3-positive T cells (brown) in the bispecific-treated tumors (top) but not in colons from the same mice (bottom). Scale bar, 200 μm. D, GUCY2C(M)-CD3 showed antitumor activity in the LS1034 human T-cell adoptive transfer model at 1 and 0.3 mg/kg (n = 10 mice/group, Q3DX3 dosing, P value < 0.0001).

Figure 2.

An anti-GUCY2C/anti-CD3ϵ bispecific selectively targets tumors versus normal tissue. A, Schematic of an anti-GUCY2C/anti-CD3ϵ bispecific human IgG1 FcγR- diabody. B, Pharmacokinetics of a murine GUCY2C cross-reactive bispecific GUCY2C(M)-CD3 at 1 and 0.3 mg/kg showing an estimated half-life of 11 days in LS1034 tumor–bearing female NSG mice with human adoptive T-cell transfer. C, IHC showing GUCY2C expression (left plot, brown staining) in LS1034 CLX tumor (top) and in the colon from an LS1034 tumor–bearing mouse (bottom). Middle plot shows biodistribution of the bispecific GUCY2C(M)-CD3 (brown) in the same LS1034 tumor (top) and lack of bispecific uptake in the colon (bottom), 7 days after dosing in an adoptive transfer study. Right plot shows infiltration of human CD3-positive T cells (brown) in the bispecific-treated tumors (top) but not in colons from the same mice (bottom). Scale bar, 200 μm. D, GUCY2C(M)-CD3 showed antitumor activity in the LS1034 human T-cell adoptive transfer model at 1 and 0.3 mg/kg (n = 10 mice/group, Q3DX3 dosing, P value < 0.0001).

Close modal

To illustrate both tumor selective targeting and efficacy, we generated an anti-GUCY2C/anti-CD3ϵ bispecific (GUCY2C(M)-CD3), which had equivalent binding affinity to human and mouse GUCY2C (Supplementary Table S2). NOD-scid IL2Rγnull (NSG) mice bearing LS1034 colorectal cancer CLX tumors were treated with GUCY2C(M)-CD3 at two different dose levels of 1 and 0.3 mg/kg, along with adoptive human T-cell transfer. Pharmacokinetic analyses demonstrated that the Fc domain fusion design of GUCY2C(M)-CD3 led to a systemic estimated half-life of 11 days in these tumor-bearing mice, which enabled weekly dosing with the bispecific (Fig. 2B). Even though GUCY2C expression was observed in both the LS1034 tumors and the normal intestinal tract epithelium of the same mice (Fig. 2C, left plot), only tumors showed uptake of GUCY2C(M)-CD3 (Fig. 2C, middle plot) and infiltration of human CD3–positive T cells (Fig. 2C, right plot), whereas no drug or human CD3 T-cell infiltration could be detected in the intestinal tract. Importantly, antitumor efficacy was observed with GUCY2C(M)-CD3 at 1 and 0.3 mg/kg but not with a nontargeted CD3 bispecific (Fig. 2D). No changes in body weight were noted in response to bispecific treatment (Supplementary Fig. S2). These observations support the hypothesis that GUCY2C can be selectively targeted on tumors with an anti–GUCY2C-targeting agent.

PF-07062119 elicits target-dependent CTL activity in vitro

After establishing the ability to preferentially target GUCY2C-expressing tumors, we developed a fully humanized and lead-optimized anti-GUCY2C/anti-CD3ϵ therapeutic bispecific (described herein as PF-07062119). PF-07062119 was generated in the diabody Fc fusion format shown in Fig. 2A and characterized in vitro for target binding and T-cell–mediated cytotoxicity. PF-07062119 showed cell surface binding to human T cells, as well as HCT116 tumor cells overexpressing GUCY2C, but not to HCT116 cells that are GUCY2C negative (Fig. 3A; Supplementary Fig. S3). To understand the range of GUCY2C expression capable of eliciting T-cell–mediated cytotoxicity, cell surface receptor density was quantitatively measured across a panel of colorectal tumor cell lines. These cells showed receptor densities ranging from 875 to 8,067 receptors per cell (Fig. 3B). Treatment of these cells with PF-07062119 and human T cells showed dose-dependent and GUCY2C expression–dependent cytotoxicity, as measured by tumor cell survival (Fig. 3C). Consistent with cytotoxicity measurement, supernatants harvested from these cytotoxicity assays showed release of IFNγ in a PF-07062119 dose-dependent manner (Fig. 3D). No change in tumor viability or IFNγ release was observed in GUCY2C-negative HCT116 cells, confirming the need for GUCY2C expression for induction of PF-07062119–mediated T-cell effector function.

Figure 3.

PF-07062119 demonstrates in vitro CTL-mediated killing in GUCY2C-positive tumor cells. A, Flow cytometry assay shows binding of PF-07062119 to naïve human T cells and to HCT116-hGUCY2C cells, but not to GUCY2C-negative HCT116 parental cells. B, Receptor density of GUCY2C in tumor cells measured using a flow cytometry–based QuantiBrite assay using an anti-GUCY2C mAb labeled 1:1 with phycoerythrin. C, PF-07062119 shows CTL-mediated killing in GUCY2C-positive tumor cell lines (E:T ratio = 5:1, 48-hour assay). D, PF-07062119 elicits dose-dependent IFNγ release in GUCY2C-positive tumor cells in the presence of T cells (E:T ratio = 5:1, 48 hours after treatment). NB, no binding; NR, not reached.

Figure 3.

PF-07062119 demonstrates in vitro CTL-mediated killing in GUCY2C-positive tumor cells. A, Flow cytometry assay shows binding of PF-07062119 to naïve human T cells and to HCT116-hGUCY2C cells, but not to GUCY2C-negative HCT116 parental cells. B, Receptor density of GUCY2C in tumor cells measured using a flow cytometry–based QuantiBrite assay using an anti-GUCY2C mAb labeled 1:1 with phycoerythrin. C, PF-07062119 shows CTL-mediated killing in GUCY2C-positive tumor cell lines (E:T ratio = 5:1, 48-hour assay). D, PF-07062119 elicits dose-dependent IFNγ release in GUCY2C-positive tumor cells in the presence of T cells (E:T ratio = 5:1, 48 hours after treatment). NB, no binding; NR, not reached.

Close modal

Because the GUCY2C pathway regulates intestinal fluidity, PF-07062119 was also characterized for its potential to modulate GUCY2C signaling by measuring production of the downstream effector, cyclic guanosine monophosphate (cGMP). Although the GUCY2C pathway agonist bacterial enterotoxin STp increased cGMP in a dose-dependent manner in GUCY2C-expressing T84 tumor cells, cGMP production was not enhanced by the addition of increasing concentrations of PF-07062119 (Supplementary Fig. S4). A nontargeted CD3-bispecific control also did not show cGMP production. In addition, PF-07062119 did not affect cGMP production induced by STp. These findings indicate the PF-07062119 is not a GUCY2C pathway agonist and does not neutralize GUCY2C pathway function in the presence of ligand.

PF-07062119 demonstrates antitumor efficacy in GUCY2C-positive human xenograft and mouse syngeneic tumors

After characterizing the in vitro range of activity of PF-07062119, we next evaluated in vivo efficacy with this bispecific in CLX and PDX models of colorectal cancer. GUCY2C expression was characterized in four CLX models and one PDX model by IHC to generate an H-score reflective of cell surface expression of GUCY2C (Table 1; Fig. 4A). The tumors were also characterized for their mutational status of KRAS and BRAF oncogenes, as well as their differentiation status (Table 1). Among the models tested, LS1034 and HT55 CLX models showed relatively high GUCY2C expression, reflective of human colorectal tumors with high GUCY2C expression (Fig. 1A; Supplementary Table S1). LS1034 and HT55 were mutant for KRAS and BRAF, respectively (Table 1). PDX-CRX-11201 also had relatively high GUCY2C expression and was mutant for KRAS. The LS174T CLX model was mutant for KRAS and showed lower GUCY2C expression, representative of human tumors expressing GUCY2C below the median H-score level in colorectal cancer (Table 1; Fig. 1A). The HCT116 model was negative for GUCY2C. Both LS174T and HCT116 were poorly differentiated tumors. Apically enhanced expression of GUCY2C was noted in all tumors with high GUCY2C expression, and higher differentiation state.

Table 1.

Characterization of colorectal cancer xenograft models tested for in vivo efficacy.

TumorH-Score (GUCY2C)KRAS mutationBRAF mutationDifferentiation status
LS1034 210 A146T WT Moderately differentiated 
HT55 180 WT N581Y Moderately differentiated 
LS174T 105 G12D WT Poorly differentiated 
HCT116 G13D WT Poorly differentiated 
PDX-CRX-11201 185 G12V WT Moderately differentiated 
TumorH-Score (GUCY2C)KRAS mutationBRAF mutationDifferentiation status
LS1034 210 A146T WT Moderately differentiated 
HT55 180 WT N581Y Moderately differentiated 
LS174T 105 G12D WT Poorly differentiated 
HCT116 G13D WT Poorly differentiated 
PDX-CRX-11201 185 G12V WT Moderately differentiated 
Figure 4.

PF-07062119 mediates antitumor efficacy in multiple in vivo models. A, GUCY2C expression in LS1034 (CLX), HT55 (CLX), PDX-CRX-11201 (PDX), LS174T (CLX), and HCT116 (CLX) tumor models (scale bar, 200 μm). B, PF-07062119 induced antitumor efficacy in LS1034, HT55, PDX-CRX-11201, and LS174T, and lack of efficacy in GUCY2C-negative HCT116 subcutaneous established models using adoptive human T-cell transfer. Mice with established tumors of 150 to 200 mm3 were dosed with PF-07062119, a nontargeted CD3-bispecific control or PBS (vehicle) weekly up to 3 times for all models, and up to 4 times for PDX-CRX-11201. Human T cells were administered intravenously 24 hours after the first bispecific or vehicle dose. C, LS1034 colon orthotopic established tumor model using adoptive human T-cell transfer described in (B) with bispecific dosed weekly up to five doses. GUCY2C expression (brown) in LS1034 orthotopic tumor and normal adjacent tissue. H&E staining (purple) shows tumor and normal tissue boundary. Bioluminescence imaging showing complete tumor regression with PF-07062119 at 0.3 mg/kg. D, Flow cytometry assay showing PF-07062119 binding to CT26-mGUCY2C and lack of binding to CT26 cells (110 nmol/L Ab 9H3-hIgG used for binding). E, CT26-mGUCY2C tumors show membrane expression of GUCY2C (brown) by IHC (scale bar, 200 μm). F, PF-07062119 mediated efficacy in CT26-mGUCY2C tumors in hCD3ϵ mice. Tumors staged to 50 to 80 mm3 were treated with PF-07062119 or nontargeted CD3 every 3 days up to 6 doses.

Figure 4.

PF-07062119 mediates antitumor efficacy in multiple in vivo models. A, GUCY2C expression in LS1034 (CLX), HT55 (CLX), PDX-CRX-11201 (PDX), LS174T (CLX), and HCT116 (CLX) tumor models (scale bar, 200 μm). B, PF-07062119 induced antitumor efficacy in LS1034, HT55, PDX-CRX-11201, and LS174T, and lack of efficacy in GUCY2C-negative HCT116 subcutaneous established models using adoptive human T-cell transfer. Mice with established tumors of 150 to 200 mm3 were dosed with PF-07062119, a nontargeted CD3-bispecific control or PBS (vehicle) weekly up to 3 times for all models, and up to 4 times for PDX-CRX-11201. Human T cells were administered intravenously 24 hours after the first bispecific or vehicle dose. C, LS1034 colon orthotopic established tumor model using adoptive human T-cell transfer described in (B) with bispecific dosed weekly up to five doses. GUCY2C expression (brown) in LS1034 orthotopic tumor and normal adjacent tissue. H&E staining (purple) shows tumor and normal tissue boundary. Bioluminescence imaging showing complete tumor regression with PF-07062119 at 0.3 mg/kg. D, Flow cytometry assay showing PF-07062119 binding to CT26-mGUCY2C and lack of binding to CT26 cells (110 nmol/L Ab 9H3-hIgG used for binding). E, CT26-mGUCY2C tumors show membrane expression of GUCY2C (brown) by IHC (scale bar, 200 μm). F, PF-07062119 mediated efficacy in CT26-mGUCY2C tumors in hCD3ϵ mice. Tumors staged to 50 to 80 mm3 were treated with PF-07062119 or nontargeted CD3 every 3 days up to 6 doses.

Close modal

To assess PF-07062119 activity in vivo, NSG mice were implanted subcutaneously with the characterized colorectal cancer cell lines and PDX-CRX-11201 fragments. An initial dose of PF-07062119, nontargeted CD3-negative control bispecific, or vehicle was administered to animals with established tumors, along with adoptive transfer of human T cells, followed by weekly dosing with PF-07062119 or control agents. In the LS1034 model, which had the highest expression of GUCY2C, PF-07062119 treatment at 0.1 mg/kg or higher doses led to complete tumor regressions (P value < 0.0001; Fig. 4B). The antitumor activity was dose dependent, because partial reduction in tumor volumes was observed at 0.03 mg/kg of PF-07062119. The HT55 CLX model and PDX-CRX-11201 (Fig. 4B), which both had similar H-scores, showed complete tumor regressions at doses as low as 0.15 mg/kg of PF-07062119 (P value < 0.0001). The LS174T CLX model, which had low GUCY2C expression, showed a small but significant reduction in tumor burden only at the high dose of PF-07062119 treatment (1 mg/kg, P value < 0.0001; Fig. 4B). The GUCY2C-negative HCT116 colorectal cancer model was not responsive to PF-07062119 at any dose tested (Fig. 4B). In summary, tumors with the highest H-scores showed complete tumor regressions with 0.1 to 0.15 mg/kg or higher doses of PF-07062119, whereas tumors with lower GUCY2C tumors had a moderate antitumor response with a higher dose of 1 mg/kg. In all models, treatment with the PBS vehicle control, or with a nontargeted CD3 bispecific at the same or higher doses that were efficacious with PF-07062119, did not result in any antitumor activity. These data demonstrate GUCY2C expression–dependent antitumor efficacy with PF-07062119, independent of the KRAS or BRAF mutational status of the tumor.

We also examined PF-07062119 activity in a colon orthotopic model to evaluate the ability of the drug to target tumors located within the gastrointestinal tract. LS1034 cells expressing luciferase were generated to monitor tumor growth in vivo using bioluminescence. GUCY2C expression in this model was confirmed by IHC analyses and found to be relatively comparable between LS1034 tumors and the adjacent mouse colon epithelium (Fig. 4C, top right). Similar to subcutaneous models, treatment of these established orthotopic tumors with PF-07062119 following adoptive human T-cell transfer led to complete tumor regressions at 0.3 mg/kg, as measured by bioluminescence imaging scans (P value < 0.0001; Fig. 4C, top left). In addition, necropsy findings at the end of the study indicated that no tumors were observed at secondary sites in mice treated with PF-07062119 at doses above 0.1 mg/kg. However, in animals treated with PF-07062119 at 0.03 mg/kg, 2 of 9 mice had tumors spread to the liver and 2 other mice had secondary spread at the abdominal wall. In the nontargeted CD3 treatment group, 2 of 9 animals had tumor spread to the liver. No body weight loss was observed in the bispecific-treated animals (Supplementary Fig. S5). These data indicate that PF-07062119 has the ability to target primary tumors located in the intestinal tract and could potentially prevent tumor spreading to secondary sites.

Although the above experiments demonstrate the ability of PF-07062119 to target tumors in immunodeficient mice using adoptive human T-cell transfer, we also sought to examine PF-07062119–driven antitumor activity in an immunocompetent mouse model. Although PF-07062119 binds to mouse GUCY2C albeit with lower affinity than to human GUCY2C (Supplementary Table S2), it does not bind to mouse CD3. Therefore, CT26 mouse syngeneic tumor cells were engineered to express mouse GUCY2C (CT26-mGUCY2C), and a transgenic mouse model expressing human CD3ϵ on mouse T cells was used for efficacy studies. PF-07062119 showed specific binding to CT26-mGUCY2C cells but not to GUCY2C-negative CT26 cells (Fig. 4D). GUCY2C expression on the tumor cell surface was confirmed by IHC in CT26-mGUCY2C subcutaneous tumors established in the human CD3ϵ transgenic model (Fig. 4E). Similar to NSG models with CLX and PDX tumors, PF-07062119 treatment showed reduction of tumor burden at all doses tested and resulted in complete regressions at doses above 1 mg/kg (Fig. 4F). No body weight loss was observed with PF-07062119 treatment (Supplementary Fig. S6), suggesting that PF-07062119 has tumor selective activity in an immunocompetent model, thereby showing the ability to harness T-cell effector function in the setting of cancer immune escape.

PF-07062119 treatment increases tumor-infiltrating lymphocyte infiltration and activation, and enhanced efficacy with checkpoint blockade

To better understand the mechanism of action of PF-07062119–mediated human T-cell activity, the LS1034 subcutaneous adoptive transfer model was used to evaluate changes in tumors and tumor-infiltrating lymphocytes (TIL) in vivo. Tumors treated at 1 mg/kg of PF-07062119 showed increased T-cell infiltration compared with tumors treated at the minimally efficacious dose of 0.03 mg/kg, whereas tumors in the vehicle treatment group did not show any T-cell infiltration (Fig. 5A, top row). TILs in PF-07062119–treated tumors showed expression of granzyme B that was polarized toward tumor cells, suggesting that these TILs were poised to form immune synapses and kill tumor cells (Fig. 5A, second row). Tumors treated with PF-07062119 also showed upregulation of PD-L1 at both 1 mg/kg and 0.03 mg/kg doses (Fig. 5A, third row). The upregulation of PD-L1 was specific to PF-07062119–treated tumors, and was not observed in vehicle treated, indicating that checkpoint mechanisms that could dampen T-cell activity were being induced with bispecific treatment. These PF-07062119–induced changes in LS1034 tumors were concomitant with tumor necrosis (Fig. 5A, bottom row). In addition, tSNE analyses on CYTOF data from TILs from these tumors identified two T-cell clusters that were unique to the PF-07062119 treatment groups. These clusters were CD4- and CD8-positive T cells that were PD-1+ 41-BB+ Tim-3+ Lag-3+, indicating that these TILs were activated, but also expressed activation-induced checkpoint markers, which if chronically expressed could negatively regulate T-cell effector function (Fig. 5B and C). Although these data imply that PF-07062119 treatment induced T-cell infiltration into tumors in a dose-dependent manner, the upregulation of PD-L1 on tumors and PD-1 on TILs suggests that a PD-1/PD-L1 axis that could dampen T-cell activity is induced with PF-07062119 treatment.

Figure 5.

PF-07062119 recruits TILs to mediate cytotoxicity and shows combination benefit with checkpoint blockade. A, PF-07062119 treatment shows dose-dependent infiltration of CD3+ T cells (top row; brown; scale bar, 300 μm). Granzyme B (second row; brown; scale bar, 200 μm) was observed in these CD3+ T cells (second row; green; scale bar, 300 μm) and was polarized toward adjacent tumor cells (inset). PF-07062119 dose-dependent upregulation of PD-L1 was observed on tumors (third row, brown), concomitantly with tumor cell necrosis (bottom row, H&E; scale bar, 200 μm). B, tSNE of CYTOF analyses from TILs at day 7 after dosing, identified unique clusters 11 and 12 in PF-07062119 tumors. C, Heatmap depicts expression of Arcsinh-transformed median intensities of each marker to represent the phenotypes of each TIL cluster shown in (B). Expression values range from 0 to 8 with max expression represented in red, median expression in white, and min expression in black. D, Combination of a minimally active dose of PF-07062119 with anti–PD-1 or anti–PD-L1 shows combination benefit compared with single agent alone.

Figure 5.

PF-07062119 recruits TILs to mediate cytotoxicity and shows combination benefit with checkpoint blockade. A, PF-07062119 treatment shows dose-dependent infiltration of CD3+ T cells (top row; brown; scale bar, 300 μm). Granzyme B (second row; brown; scale bar, 200 μm) was observed in these CD3+ T cells (second row; green; scale bar, 300 μm) and was polarized toward adjacent tumor cells (inset). PF-07062119 dose-dependent upregulation of PD-L1 was observed on tumors (third row, brown), concomitantly with tumor cell necrosis (bottom row, H&E; scale bar, 200 μm). B, tSNE of CYTOF analyses from TILs at day 7 after dosing, identified unique clusters 11 and 12 in PF-07062119 tumors. C, Heatmap depicts expression of Arcsinh-transformed median intensities of each marker to represent the phenotypes of each TIL cluster shown in (B). Expression values range from 0 to 8 with max expression represented in red, median expression in white, and min expression in black. D, Combination of a minimally active dose of PF-07062119 with anti–PD-1 or anti–PD-L1 shows combination benefit compared with single agent alone.

Close modal

Based on the rationale that PF-07062119 treatment induced expression of PD-1/PD-L1 on TILs and tumors, we tested if checkpoint blockade with anti–PD-1 or anti–PD-L1 antibodies could enhance antitumor efficacy of PF-07062119. LS1034 subcutaneous tumors treated with a suboptimally efficacious PF-07062119 of 0.03 mg/kg showed significant improvement in efficacy in combination with either an anti–PD-L1 or an anti–PD-1 antibody (P value < 0.0001 and 0.0004, respectively), whereas these agents were not efficacious as single agents in this model (Fig. 5D). These findings indicate that although checkpoint mechanisms may limit efficacy of PF-07062119, the bispecific activity can be further enhanced in combination with checkpoint blockade agents.

Anti-VEGF blockade enhances PF-07062119 antitumor activity

Antiangiogenesis treatment is currently used for the treatment of colorectal cancer and is also reported to improve responses to T-cell–mediated therapy by enhancing T-cell infiltration into the tumor. Because suboptimal doses of PF-07062119 showed reduced T-cell infiltration, we tested the effects of combining anti–VEGF-A blockade with PF-07062119 treatment in vivo. PDX-CRX-11201 was used for this evaluation because this tumor model is only partially responsive to G6-31, an anti–VEGF-A mAb that blocks murine VEGF-A. A combination of anti–VEGF-A treatment with PF-07062119 administered at a suboptimal dose of 0.05 mg/kg led to a significant combination benefit leading to complete tumor regressions (P value < 0.0001; Fig. 6A). In addition, IHC analyses showed that tumors that received the combination treatment had notably increased T-cell infiltration compared with tumors treated with either PF-07062119 or anti–VEGF-A alone (Fig. 6B). Therefore, these data suggest that apart from the combination benefit observed with checkpoint blockade agents, antiangiogenesis treatment could further enhance the efficacy observed with PF-07062119.

Figure 6.

Combination study of PF-07062119 with anti–VEGF-A mAb. A, Combination of a minimally efficacious dose of PF-07062119 in PDX-CRX-11201 tumor, which is partially responsive to anti–VEGF-A mAb treatment, results in a significant combination benefit compared with single agents. B, Increased CD3+ T cells (brown) were observed in the PF-07062119 + Anti–VEGF-A combination treatment group compared with single agents or nontargeted CD3 treatment.

Figure 6.

Combination study of PF-07062119 with anti–VEGF-A mAb. A, Combination of a minimally efficacious dose of PF-07062119 in PDX-CRX-11201 tumor, which is partially responsive to anti–VEGF-A mAb treatment, results in a significant combination benefit compared with single agents. B, Increased CD3+ T cells (brown) were observed in the PF-07062119 + Anti–VEGF-A combination treatment group compared with single agents or nontargeted CD3 treatment.

Close modal

Toxicology and pharmacokinetics of PF-07062119 in cynomolgus monkey

Cynomolgus monkeys were selected as relevant species for toxicity studies after confirming binding of PF-07062119 to cynomolgus T cells and GUCY2C protein (Supplementary Table S2; Supplementary Fig. S7). Nonclinical safety and pharmacokinetics of PF-07062119 were evaluated in cynomolgus monkeys in an exploratory intra-animal dose escalation (2 doses, once weekly) toxicity study at 30 μg/kg followed a week later by 100 μg/kg, and 60 μg/kg followed a week later by 180 μg/kg (Supplementary Fig. S8A). The main in-life effects observed with PF-07062119 treatment included emesis, a slight increase in body temperature, decreased activity, hunched posture, reduced appetite, dehydration, body weight loss (≤ 7%), and soft or liquid feces. However, animals in both cohorts tolerated the treatment with administration of oral and/or subcutaneous fluids. Importantly, histopathologic evaluations of the intestinal tract, which is the main site of GUCY2C expression in normal tissues, showed only minimal to mild infiltrates and crypt hyperplasia in multiple segments of the large and small intestine, along with typically minimal to mild villous atrophy in the small intestine (Supplementary Fig. S8B). However, there were no erosions, ulcerations, or other evidence of overt necrosis or loss of overlying epithelium.

PF-07062119 treatment induced pharmacodynamic effects in the monkeys, as measured by increases in the percentage of activated (CD69+) CD8+ T cells (Supplementary Fig. S7C), and in peripheral cytokines, including IFNγ, IL2, IL10, and IL6 (Supplementary Fig. S7D). Cytokine induction observed after the first dose in each cohort was dampened after the second dose administration of PF-07062119, indicative of a priming effect. Toxicokinetic analysis of PF-07062119 in both treatment cohorts, assessed by maximum observed concentration (Cmax) and area under the concentration–time curve, showed dose-proportional systemic exposure, as well as a linear PK profile (Supplementary Fig. S7E). No sex-related differences in systemic exposure were noted in this study.

Despite the benefit and approval of checkpoint blockade agents for the treatment of microsatellite instability–high colorectal tumors, immunotherapy has had limited success in the majority of gastrointestinal cancers. Attempts to enhance T-cell activity against these cancers with CD3 bispecifics have shown early signs of activity in the clinic but have been limited by target expression on normal tissues (32, 33). Here, we have described the expression of GUCY2C across gastrointestinal cancers, particularly in moderately to well-differentiated colorectal, gastric, and esophageal adenocarcinomas, and provided support for the hypothesis that GUCY2C-expressing tumors can be selectively targeted with an anti-GUCY2C/anti-CD3ϵ bispecific, owing to the restricted expression of the target in the apical intestinal epithelium. For this evaluation, a bispecific with equivalent affinity to mouse and human GUCY2C was used to show selective biodistribution to xenograft tumors expressing human GUCY2C versus the intestinal tract of the host mice expressing mouse GUCY2C. In further support of this hypothesis, antitumor efficacy was observed in this study in the absence of any tissue damage to the intestinal tract (as evaluated by histopathology), suggesting that GUCY2C-positive tumors can be preferentially targeted with an anti-GUCY2C/anti-CD3ϵ bispecific T-cell redirection approach.

Based on these data, PF-07062119, a fully humanized and lead-optimized anti-GUCY2C/anti-CD3ϵ bispecific, was developed and characterized in vitro and in several in vivo tumor models for pharmacologic activity. Both immunodeficient mice with human T-cell adoptive transfer and immunocompetent mice were used in a complementary manner to characterize in vivo efficacy with PF-07062119. Notably, PF-07062119–mediated antitumor activity was observed in all models in a dose-dependent manner and was associated with the level of GUCY2C expressed in human xenograft models, suggesting a potential to obtain activity through dose modulation in the clinic. The human xenograft tumors evaluated for in vivo efficacy included those with mutations in the KRAS or BRAF oncogenes, which are clinically associated with a lack of response to currently approved EGFR-targeted therapies, suggesting that PF-07062119 can target tumors regardless of the mutational status of these genes.

Furthermore, a colon orthotopic xenograft model showed that PF-07062119 was effective at selectively eliminating primary tumors in the intestinal tract without damaging the normal adjacent tissue (as assessed by histopathology) and had the potential to prevent tumor spreading to secondary sites, such as the liver. Although these T-cell adoptive transfer studies highlighted PF-07062119–mediated redirected T-induced cell killing, we also demonstrated that dose-dependent efficacy could be achieved in a mouse syngeneic tumor model in the presence of a functional immune system. Even with the caveat that both mouse GUCY2C and human CD3ϵ proteins are overexpressed in the syngeneic system, this model allows for long-term evaluation of antitumor responses, whereas the duration of adoptive transfer tumor models is limited by the eventual onset of graft versus host disease.

We further explored the mechanisms of PF-07062119 activity using the T-cell adoptive transfer model and showed that activity was mediated through recruitment of TILs into GUCY2C-expressing tumors. These T cells had granzyme B expression, which was polarized toward target-expressing tumor cells, suggesting that they are poised to form immune synapses with the tumor cells and initiate cell killing. Such analyses of T-cell infiltration and activity could be informative in the clinic as pharmacodynamic biomarkers of PF-07062119–mediated activity. Further evaluations showed that tumors treated with PF-07062119 also upregulated PD-L1, and TILs from these tumors had expression of activation markers including PD-1, indicating initiation of checkpoint mechanisms that could dampen T-cell–mediated cytotoxicity. Based on these observations, PF-07062119 was combined with checkpoint blockade via anti–PD-1 and anti–PD-L1 antibodies. Combinations of anti–PD-1 or anti–PD-L1 blocking antibodies with a suboptimal dose of PF-07062119 showed improved efficacy compared with single agents, indicating that checkpoint blockade combinations will be important to consider during clinical evaluation of PF-07062119. A limitation of murine models is that due to their lack of sensitivity to systemic cytokine release, the sequencing of combinations and their effects on cytokine release syndrome will need to be evaluated in the clinic.

Recent preclinical and clinical studies with antiangiogenesis agents have been reported to improve responses to immunotherapy through various mechanisms, including enhancing T-cell infiltration into tumors (34–36). For example, the FDA recently approved combinations with axitinib (VEGFR inhibitor) and avelumab (anti–PD-L1 mAb) in renal cell carcinoma (37), and with bevacizumab (anti-VEGF mAb) and atezolizumab (anti–PD-L1 mAb) in non–small cell lung cancer (38), due to significant improvements in progression-free survival and overall response rate. Based on these observations, we tested a combination of PF-07062119 with an anti-VEGF mAb, which induced complete tumor regressions, whereas single agents were only minimally active in the PDX model tested. Although increased T-cell infiltrate was observed in the combination treatment group, further mechanistic studies will elucidate whether this increase in TILs was due to changes in the tumor vasculature or was a consequence of increased PF-07062119 uptake following anti-VEGF blockade. A limitation of the adoptive transfer model is that T-cell engraftment in the tumor is dependent on bispecific-mediated recruitment of T cells, thereby resulting in minimal T-cell engraftment with antiangiogenesis treatment alone. Therefore, additional combination studies in immunocompetent mice will be pursued to further elucidate the mechanisms by which VEGF blockade improves T-cell infiltration in PF-07062119–treated tumors.

Finally, the exploratory toxicity study with PF-07062119 in cynomolgus monkeys suggests a clinically monitorable and manageable toxicity profile. PF-07062119 treatment did not induce any erosions, ulcerations, or evidence of overt necrosis or loss of overlying epithelium, and the overall in-life findings were tolerated with administration of oral and/or subcutaneous fluids. Furthermore, the gastrointestinal findings of minimal to mild immune infiltrates, crypt hyperplasia, and villous atrophy would all be expected to resolve following cessation of treatment. Pharmacologic activity in the monkeys was shown by CD69 upregulation on CD8+ T cells and increases of systemic cytokines (Supplementary Fig. S7); although there are polymorphisms in macaque CD3ϵ that abolish activity of a specific anti-CD3ϵ antibody (FN18) in a subset of animals (39–41), PF-07062119 has shown binding and pharmacologic activity in all monkeys tested (Supplementary Figs. S7 and S8; Supplementary Table S2). In addition, pharmacokinetic analyses in cynomolgus monkeys estimated the terminal half-life of PF-07062119 to be similar to a regular mAb. A longer time-course PK study would be needed to calculate the precise terminal half-life.

Collectively, these studies demonstrate that GUCY2C-positive tumors can be preferentially targeted with an anti-GUCY2C/anti-CD3ϵ bispecific. Our lead clinical candidate, PF-07062119, has shown potent single-agent antitumor efficacy with PF-07062119 in multiple in vitro and in vivo models. This activity can be further enhanced with immune checkpoint blockade agents, which prevent T-cell exhaustion, as well as with antiangiogenesis agents, which could increase T-cell infiltration (Supplementary Fig. S9). Further evaluation of PF-07062119 along with these combination agents will be pursued in clinical studies to examine their ability to reduce tumor burden in patients with gastrointestinal cancers.

D. Mathur, A.R. Root, B. Bugaj-Gaweda, W. Fang, J.C. Kearney, C.M. Rohde, C. Stevens, C. Kamperschroer, K. Kelleher, E. Upeslacis, E.R. LaVallie, D.R. Fernandez, B.S. Buetow, E. Rosfjord, L. Bloom, and L. Tchistiakova are employees/paid consultants for Pfizer. P. Sapra is an employee/paid consultant for and holds ownership interest (including patents) in Pfizer. No potential conflicts of interest were disclosed by the other authors.

Conception and design: D. Mathur, A.R. Root, M. Guffroy, J. Golas, C.M. Rohde, C. Kamperschroer, J. Yao, L. Bloom, L.E. King, L. Tchistiakova, A. Nguyen, P. Sapra

Development of methodology: D. Mathur, B. Bugaj-Gaweda, S. Bisulco, X. Tan, W. Fang, J. Lucas, J. Golas, C.M. Rohde, R.F. Lawrence-Henderson, J. Yao, E.R. LaVallie, B.S. Buetow, E. Rosfjord, L.E. King, L. Tchistiakova

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.R. Root, B. Bugaj-Gaweda, S. Bisulco, X. Tan, W. Fang, J.C. Kearney, J. Lucas, J. Golas, C.M. Rohde, C. Stevens, R.F. Lawrence-Henderson, E. Upeslacis, J. Yao, E.R. LaVallie, B.S. Buetow, E. Rosfjord, L.E. King

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D. Mathur, A.R. Root, B. Bugaj-Gaweda, S. Bisulco, X. Tan, J. Lucas, M. Guffroy, J. Golas, C.M. Rohde, C. Stevens, C. Kamperschroer, R.F. Lawrence-Henderson, J. Yao, J. Narula, D.R. Fernandez, B.S. Buetow, L.E. King, L. Tchistiakova

Writing, review, and/or revision of the manuscript: D. Mathur, A.R. Root, S. Bisulco, J.C. Kearney, J. Lucas, M. Guffroy, J. Golas, C.M. Rohde, C. Kamperschroer, R.F. Lawrence-Henderson, E. Upeslacis, J. Yao, J. Narula, D.R. Fernandez, B.S. Buetow, E. Rosfjord, L. Bloom, L.E. King, L. Tchistiakova, P. Sapra

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D. Mathur, B. Bugaj-Gaweda, S. Bisulco, W. Fang, J.C. Kearney, K. Kelleher, E. Upeslacis, E. Rosfjord

Study supervision: J.C. Kearney, C. Kamperschroer, E. Upeslacis, J. Yao, E. Rosfjord, P. Sapra

We would like to thank Patrick Gaffney for support with statistical analyses, Andrew Gifford for pharmacokinetics analyses support, Susan Benard for BIAcore analysis, Dawn DeThomas for graphical illustrations, John Kreeger for pathology support, and John Hill, David Schaer, and Steven Pirie-Shepherd for discussions on the study design and article. All funding for this study was provided by Pfizer Worldwide R&D.

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.

1.
Bray
F
,
Ferlay
J
,
Soerjomataram
I
,
Siegel
RL
,
Torre
LA
,
Jemal
A
. 
Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries
.
CA Cancer J Clin
2018
;
68
:
394
424
.
2.
Jemal
A
,
Siegel
R
,
Ward
E
,
Hao
Y
,
Xu
J
,
Thun
MJ
. 
Cancer statistics, 2009
.
CA Cancer J Clin
2009
;
59
:
225
49
.
3.
Allegra
CJ
,
Jessup
JM
,
Somerfield
MR
,
Hamilton
SR
,
Hammond
EH
,
Hayes
DF
, et al
American Society of Clinical Oncology provisional clinical opinion: testing for KRAS gene mutations in patients with metastatic colorectal carcinoma to predict response to anti-epidermal growth factor receptor monoclonal antibody therapy
.
J Clin Oncol
2009
;
27
:
2091
6
.
4.
Di Nicolantonio
F
,
Martini
M
,
Molinari
F
,
Sartore-Bianchi
A
,
Arena
S
,
Saletti
P
, et al
Wild-type BRAF is required for response to panitumumab or cetuximab in metastatic colorectal cancer
.
J Clin Oncol
2008
;
26
:
5705
12
.
5.
Pratilas
CA
,
Hanrahan
AJ
,
Halilovic
E
,
Persaud
Y
,
Soh
J
,
Chitale
D
, et al
Genetic predictors of MEK dependence in non-small cell lung cancer
.
Cancer Res
2008
;
68
:
9375
83
.
6.
Solit
DB
,
Garraway
LA
,
Pratilas
CA
,
Sawai
A
,
Getz
G
,
Basso
A
, et al
BRAF mutation predicts sensitivity to MEK inhibition
.
Nature
2006
;
439
:
358
62
.
7.
Ganesh
K
,
Stadler
ZK
,
Cercek
A
,
Mendelsohn
RB
,
Shia
J
,
Segal
NH
, et al
Immunotherapy in colorectal cancer: rationale, challenges and potential
.
Nat Rev Gastroenterol Hepatol
2019
;
16
:
361
75
.
8.
Tabernero
J
,
Melero
I
,
Ros
W
,
Argiles
G
,
Marabelle
A
,
Rodriguez-Ruiz
ME
, et al
Phase Ia and Ib studies of the novel carcinoembryonic antigen (CEA) T-cell bispecific (CEA CD3 TCB) antibody as a single agent and in combination with atezolizumab: preliminary efficacy and safety in patients with metastatic colorectal cancer (mCRC)
.
J Clin Oncol
2017
;
35
(
15_suppl
):
3002
.
9.
Hummel
H-D
,
Kufer
P
,
Grüllich
C
,
Deschler-Baier
B
,
Chatterjee
M
,
Goebeler
M-E
, et al
Phase 1 study of pasotuxizumab (BAY 2010112), a PSMA-targeting Bispecific T cell Engager (BiTE) immunotherapy for metastatic castration-resistant prostate cancer (mCRPC)
.
J Clin Oncol
2019
;
37
(
15_suppl
):
5034
.
10.
Lameris
R
,
de Bruin
RC
,
Schneiders
FL
,
van Bergen en Henegouwen
PM
,
Verheul
HM
,
de Gruijl
TD
, et al
Bispecific antibody platforms for cancer immunotherapy
.
Crit Rev Oncol Hematol
2014
;
92
:
153
65
.
11.
Birbe
R
,
Palazzo
JP
,
Walters
R
,
Weinberg
D
,
Schulz
S
,
Waldman
SA
. 
Guanylyl cyclase C is a marker of intestinal metaplasia, dysplasia, and adenocarcinoma of the gastrointestinal tract
.
Hum Pathol
2005
;
36
:
170
9
.
12.
Danaee
H
,
Kalebic
T
,
Wyant
T
,
Fassan
M
,
Mescoli
C
,
Gao
F
, et al
Consistent expression of guanylyl cyclase-C in primary and metastatic gastrointestinal cancers
.
PLoS One
2017
;
12
:
e0189953
.
13.
Bashir
B
,
Merlino
DJ
,
Rappaport
JA
,
Gnass
E
,
Palazzo
JP
,
Feng
Y
, et al
Silencing the GUCA2A-GUCY2C tumor suppressor axis in CIN, serrated, and MSI colorectal neoplasia
.
Hum Pathol
2019
;
87
:
103
14
.
14.
Lin
JE
,
Snook
AE
,
Li
P
,
Stoecker
BA
,
Kim
GW
,
Magee
MS
, et al
GUCY2C opposes systemic genotoxic tumorigenesis by regulating AKT-dependent intestinal barrier integrity
.
PLoS One
2012
;
7
:
e31686
.
15.
Currie
MG
,
Fok
KF
,
Kato
J
,
Moore
RJ
,
Hamra
FK
,
Duffin
KL
, et al
Guanylin: an endogenous activator of intestinal guanylate cyclase
.
Proc Natl Acad Sci U S A
1992
;
89
:
947
51
.
16.
Kita
T
,
Smith
CE
,
Fok
KF
,
Duffin
KL
,
Moore
WM
,
Karabatsos
PJ
, et al
Characterization of human uroguanylin: a member of the guanylin peptide family
.
Am J Physiol
1994
;
266
(
2 Pt 2
):
F342
8
.
17.
Bolin
I
,
Wiklund
G
,
Qadri
F
,
Torres
O
,
Bourgeois
AL
,
Savarino
S
, et al
Enterotoxigenic Escherichia coli with STh and STp genotypes is associated with diarrhea both in children in areas of endemicity and in travelers
.
J Clin Microbiol
2006
;
44
:
3872
7
.
18.
Schulz
S
,
Green
CK
,
Yuen
PS
,
Garbers
DL
. 
Guanylyl cyclase is a heat-stable enterotoxin receptor
.
Cell
1990
;
63
:
941
8
.
19.
Kim
GW
,
Lin
JE
,
Waldman
SA
. 
GUCY2C: at the intersection of obesity and cancer
.
Trends Endocrinol Metab
2013
;
24
:
165
73
.
20.
Mullin
JM
,
Laughlin
KV
,
Ginanni
N
,
Marano
CW
,
Clarke
HM
,
Peralta Soler
A
. 
Increased tight junction permeability can result from protein kinase C activation/translocation and act as a tumor promotional event in epithelial cancers
.
Ann N Y Acad Sci
2000
;
915
:
231
6
.
21.
Soler
AP
,
Miller
RD
,
Laughlin
KV
,
Carp
NZ
,
Klurfeld
DM
,
Mullin
JM
. 
Increased tight junctional permeability is associated with the development of colon cancer
.
Carcinogenesis
1999
;
20
:
1425
31
.
22.
Jones
PT
,
Dear
PH
,
Foote
J
,
Neuberger
MS
,
Winter
G
. 
Replacing the complementarity-determining regions in a human antibody with those from a mouse
.
Nature
1986
;
321
:
522
5
.
23.
Fernandez-Banet
J
,
Esposito
A
,
Coffin
S
,
Horvath
IB
,
Estrella
H
,
Schefzick
S
, et al
OASIS: web-based platform for exploring cancer multi-omics data
.
Nat Methods
2016
;
13
:
9
10
.
24.
Paranavitana
CM
. 
Non-radioactive detection of K-ras mutations by nested allele specific PCR and oligonucleotide hybridization
.
Mol Cell Probes
1998
;
12
:
309
15
.
25.
Harada
S
,
Yanagisawa
M
,
Iwai
T
,
Aida
S
,
Kurasawa
M
,
Yorozu
K
, et al
Antitumor activities of bevacizumab for KRAS, BRAF, and PIK3CA mutated human colorectal cancer xenograft models
.
J Clin Oncol
2013
;
31
(
4_suppl
):
362
.
26.
Liang
WC
,
Wu
X
,
Peale
FV
,
Lee
CV
,
Meng
YG
,
Gutierrez
J
, et al
Cross-species vascular endothelial growth factor (VEGF)-blocking antibodies completely inhibit the growth of human tumor xenografts and measure the contribution of stromal VEGF
.
J Biol Chem
2006
;
281
:
951
61
.
27.
Kasaian
MT
,
Raible
D
,
Marquette
K
,
Cook
TA
,
Zhou
S
,
Tan
XY
, et al
IL-13 antibodies influence IL-13 clearance in humans by modulating scavenger activity of IL-13Ralpha2
.
J Immunol
2011
;
187
:
561
9
.
28.
Atwell
S
,
Ridgway
JB
,
Wells
JA
,
Carter
P
. 
Stable heterodimers from remodeling the domain interface of a homodimer using a phage display library
.
J Mol Biol
1997
;
270
:
26
35
.
29.
Ridgway
JB
,
Presta
LG
,
Carter
P
. 
‘Knobs-into-holes’ engineering of antibody CH3 domains for heavy chain heterodimerization
.
Protein Eng
1996
;
9
:
617
21
.
30.
Moore
PA
,
Zhang
W
,
Rainey
GJ
,
Burke
S
,
Li
H
,
Huang
L
, et al
Application of dual affinity retargeting molecules to achieve optimal redirected T-cell killing of B-cell lymphoma
.
Blood
2011
;
117
:
4542
51
.
31.
Moore
PA
,
Shah
K
,
Yang
Y
,
Alderson
R
,
Roberts
P
,
Long
V
, et al
Development of MGD007, a gpA33 x CD3-bispecific DART protein for T-cell immunotherapy of metastatic colorectal cancer
.
Mol Cancer Ther
2018
;
17
:
1761
72
.
32.
Fisher
TS
,
Hooper
AT
,
Lucas
J
,
Clark
TH
,
Rohner
AK
,
Peano
B
, et al
A CD3-bispecific molecule targeting P-cadherin demonstrates T cell-mediated regression of established solid tumors in mice
.
Cancer Immunol Immunother
2018
;
67
:
247
59
.
33.
Tabernero
J
,
Ohtsu
A
,
Muro
K
,
Van Cutsem
E
,
Oh
SC
,
Bodoky
G
, et al
Exposure-response analyses of ramucirumab from two randomized, phase III trials of second-line treatment for advanced gastric or gastroesophageal junction cancer
.
Mol Cancer Ther
2017
;
16
:
2215
22
.
34.
Allen
E
,
Jabouille
A
,
Rivera
LB
,
Lodewijckx
I
,
Missiaen
R
,
Steri
V
, et al
Combined antiangiogenic and anti-PD-L1 therapy stimulates tumor immunity through HEV formation
.
Sci Transl Med
2017
;
9
.
35.
Chen
DS
,
Hurwitz
H
. 
Combinations of bevacizumab with cancer immunotherapy
.
Cancer J
2018
;
24
:
193
204
.
36.
Roland
CL
,
Lynn
KD
,
Toombs
JE
,
Dineen
SP
,
Udugamasooriya
DG
,
Brekken
RA
. 
Cytokine levels correlate with immune cell infiltration after anti-VEGF therapy in preclinical mouse models of breast cancer
.
PLoS One
2009
;
4
:
e7669
.
37.
Motzer
RJ
,
Penkov
K
,
Haanen
J
,
Rini
B
,
Albiges
L
,
Campbell
MT
, et al
Avelumab plus axitinib versus sunitinib for advanced renal-cell carcinoma
.
N Engl J Med
2019
;
380
:
1103
15
.
38.
Reck
M
,
Mok
TSK
,
Nishio
M
,
Jotte
RM
,
Cappuzzo
F
,
Orlandi
F
, et al
Atezolizumab plus bevacizumab and chemotherapy in non-small-cell lung cancer (IMpower150): key subgroup analyses of patients with EGFR mutations or baseline liver metastases in a randomised, open-label phase 3 trial
.
Lancet Respir Med
2019
;
7
:
387
401
.
39.
Liu
YY
,
Wang
Z
,
Thomas
J
,
Goodwin
KJ
,
Stavrou
S
,
Neville
DM
 Jr
. 
Polymorphisms of CD3epsilon in cynomolgus and rhesus monkeys and their relevance to anti-CD3 antibodies and immunotoxins
.
Immunol Cell Biol
2007
;
85
:
357
62
.
40.
Uda
A
,
Tanabayashi
K
,
Mukai
R
,
Terao
K
,
Yamada
A
. 
Identification of an amino acid responsible for the CD3 polymorphism in cynomolgus monkeys (Macaca fascicularis)
.
J Med Primatol
2003
;
32
:
105
10
.
41.
Uda
A
,
Tanabayashi
K
,
Mukai
R
,
Yachi
M
,
Nam
KH
,
Yamada
A
. 
CD3 polymorphism in cynomolgus monkeys (Macaca fascicularis)
.
J Med Primatol
2001
;
30
:
141
7
.