Abstract
Regulatory T-cell (Treg) infiltration can be targeted as a cancer immunotherapy. Here, we describe therapeutic efficacy of this strategy in a canine model of bladder cancer. We used dogs with naturally occurring bladder cancer to study the molecular mechanism of Treg infiltration into bladder cancer tissues and the effect of anti-Treg treatment. Tumor-infiltrating Tregs were evaluated by immunohistochemistry, and their association with prognosis was examined in dogs with bladder cancer. The molecular mechanism of Treg infiltration was explored by RNA sequencing and protein analyses. Murine xenograft experiments and canine studies were used to explore the therapeutic potential of anti-Treg treatment for bladder cancer. We found that tumor-infiltrating Tregs were associated with poor prognosis in dogs bearing spontaneous bladder cancer. Treg infiltration was caused by interaction between the tumor-producing chemokine CCL17 and the receptor CCR4 expressed on Tregs. CCR4 blockade inhibited tumor growth and Treg infiltration into the tissues in a xenograft mouse model. Dogs with spontaneous bladder cancer responded to anti-CCR4 treatment with improved survival and low incidence of clinically relevant toxicities. In human patients with bladder cancer, immunohistochemistry showed that tumor-infiltrating Tregs expressed CCR4. Thus, anti-CCR4 treatment may be a rational approach to test in clinical trials for human patients with bladder cancer.
Introduction
Bladder cancer is a prevalent and aggressive malignancy related to nearly 168,000 deaths annually worldwide (1). This cancer is variegated and histologically categorized into two types: The low-grade, superficial type, is more prevalent (∼70%) and is related to a favorable prognosis. The high-grade, muscle-invasive type, is less prevalent (∼30%) and poses a high mortality risk due to metastasis (2). Platinum-based chemotherapy is typically used for advanced disease, although patient outcomes are poor, with overall survival ranging from 9 to 15 months (3, 4). The 5-year survival rate is approximately 5% in the metastatic setting. This therapeutic stalemate demands different treatments.
Foxp3-expressing regulatory T cells (Treg) play a role not only in suppression of immune response against self-antigens but also in tumor progression. In humans, Treg infiltration has been reported in certain tumor tissues and has been related to progression and prognosis (5, 6). High Treg infiltration into tumor tissues is associated with poor prognosis in patients with ovarian cancer, breast cancer, and hepatocellular carcinoma. In tumor-bearing mice, Treg depletion that follows administration of anti-CD25 augments antitumor immunity and leads to tumor eradication (7, 8). Anti-Treg therapy is under investigation for human malignant tumors such as melanoma, adult T-cell leukemia/lymphoma, lung cancer, and esophageal cancer (9–12). The role of Tregs and the therapeutic efficacy of its depletion in bladder cancer remain unclear.
Although rodent models are indispensable in cancer research, highly inbred rodents kept in controlled environments are an inadequate model for the diverse human setting. Mice-based preclinical studies have often failed to estimate outcomes of human clinical trials (13), with <10% successful translation from rodent models to human clinical cancer trials (14). Current models for bladder cancer include murine carcinogen-based models, engraftment models, and genetically engineered models. However, few mouse models of bladder cancer develop muscle-invasive or metastatic phenotypes (15) found in the human setting. More relevant animal models of muscle-invasive bladder cancer are needed to study the disease.
Naturally occurring bladder cancer in companion dogs resembles human bladder cancer with regard to genetic and environmental heterogeneity, clinical signs, histopathology, disease progression, metastatic behavior, and response to cisplatin-based chemotherapy (16–18). Therefore, dogs with bladder cancer could serve as a bridge between laboratory animal models and humans. Unlike humans, the high-grade muscle-invasive type is the leading canine bladder cancer (>90% of cases), whereas the low-grade superficial type is infrequent in dogs. Several studies reported metastases of lymph nodes or lungs in approximately 15% of dogs at diagnosis and in 40% to 50% at death (17). Studies using microarray and RNA sequencing (RNA-seq) have shown similarities in gene-expression profiles between canine and human bladder cancers (19, 20). Moreover, dog-based clinical trials can be conducted in a comparatively short duration because dogs have a shorter life span than humans. Thus, the canine bladder cancer model could be informative to study the pathogenesis of muscle-invasive bladder cancer and to assess biomarkers and therapeutics.
Here, we showed that tumor-infiltrating Tregs were associated with poor prognosis in dogs with spontaneous bladder cancer, and that Treg migration was mediated by C-C chemokine receptor 4 (CCR4). We also demonstrated the therapeutic potential of anti-CCR4 treatment in dogs with bladder cancer. This canine model of muscle-invasive bladder cancer paves the way for the translation of CCR4 blockade therapy to human patients with bladder cancer.
Materials and Methods
Ethical statements
The Animal Care and Clinical Research Committees of the Veterinary Medical Center of the University of Tokyo (VMC-UT) approved this study protocol of tumor tissue sampling from client-owned dogs and canine clinical trial (approval no. VMC2016-01). We obtained written informed consent from all dog owners. Use of experimental animals (mice and dogs) was approved by the Institutional Animal Care and Use Committee of the University of Tokyo (approval nos. P16-151 and P17-108). All experiments followed approved guidelines. All human tissue samples were collected per the U.S. Common Rule, and human clinical data were collected per the HIPAA guidelines.
Canine bladder cancer model, selection criteria, and sample collection
Characteristics of dogs used for histologic, gene-expression, and protein expression analyses are in Supplementary Table S1. For histologic analysis, we obtained archival formalin-fixed, paraffin-embedded urinary bladder cancer tissues from 26 dogs at the VMC-UT. All dogs underwent total cystectomy (21). Bladder cancer diagnosis was confirmed by histopathology. Normal canine bladder tissues were obtained from 13 routine necropsy cadavers. Sex, age, and breeds were not significantly different between bladder cancer cases and normal controls. Survival time and current status (alive, deceased, or lost) of all dogs were determined by medical record or interview. Overall survival (OS) was defined as the time from total cystectomy to the established cause of death of the animal at the end of the study (April 6, 2016). Disease-free survival (DFS) was defined as the time from total cystectomy to relapse or death at the end of the study.
Fresh urinary bladder tissues from 12 dogs with bladder cancer were analyzed for gene expression. Archival snap-frozen normal bladder tissues collected from six healthy beagles euthanized for another experimental purpose were used as a control. For enzyme-linked immunosorbent assay (ELISA), serum samples were collected from 16 dogs with bladder cancer and 10 healthy dogs (Supplementary Table S1). Fresh urine samples were collected by a urethral catheter from 16 dogs with bladder cancer and 14 healthy dogs. Freshly resected tumor tissues from 8 dogs with bladder cancer were assessed by flow cytometry for tumor-infiltrating lymphocytes (TIL).
Dogs with spontaneous bladder cancer (n = 14) were enrolled in a canine clinical trial of anti-CCR4 treatment, excluding dogs treated with chemotherapy or radiotherapy from this clinical trial. Before treatment, the tumor stage was defined per the World Health Organization (WHO) criteria for canine bladder cancer (22, 23). Characteristics of dogs used in the clinical trial summarized in Supplementary Table S2.
Human samples
We purchased paraffin-embedded sections of urinary bladder specimens from 3 healthy volunteers and 11 patients with bladder cancer (BioChain Institute and OriGene Technologies). Characteristics of human patients with bladder cancer were summarized in Supplementary Table S3.
Immunohistochemistry
The expression of Foxp3, CCR4, programmed cell death protein-1 (PD-1), and T-cell immunoglobulin, and mucin domain-3 (TIM-3) was examined by immunohistochemistry. Heat-induced antigen retrieval was performed by autoclaving the sections for 5 minutes at 121°C in 10 mmol/L sodium citrate buffer (pH 6.0). Endogenous peroxidase activity was blocked by incubation with REAL Peroxidase-Blocking Solution (Dako) at room temperature for 10 minutes. The sections were blocked with 5% skim milk in Tris-buffered saline with 0.1% Tween 20 (TBST) at room temperature for 60 minutes and then incubated with primary antibodies, a rat anti-Foxp3 (1:400 dilution, clone FJK-16s, eBioscience), a mouse anti-CCR4 (1:100 dilution, clone 1G1, BD Biosciences), a rabbit anti–PD-1 (1:200 dilution, NBP1-77276, Novus Biologicals), or a goat anti–TIM-3 (1:500 dilution, ab47997, Abcam) at 4°C overnight. Secondary antibodies were applied as follows: EnVision polymer reagent for mouse or rabbit (Dako) at room temperature for 45 minutes, or a biotin-labeled anti-rat or goat IgG antibody (Vector Laboratories) at 37°C for 30 minutes followed by HRP-labeled streptavidin (Dako) at room temperature for 30 minutes. The reaction products were visualized with 3,3′-diaminobenzidine (DAB).
For human bladder tissues, antigen retrieval, endogenous peroxidase blocking, and milk blocking were performed as described above. As primary antibodies, a rabbit anti-Foxp3 (1:100 dilution, clone 1054C, R&D Systems) or a mouse anti-CCR4 (1:100 dilution, clone 1G1, BD Biosciences) was applied at 4°C overnight. Immunohistochemistry was performed by using secondary antibodies, EnVision polymer reagent for rabbit or mouse (Dako) at room temperature for 45 minutes, followed by DAB detection. Immunofluorescence was performed by using secondary antibodies, Alexa Fluor 598 goat anti-rabbit IgG (1:500, Invitrogen), and Alexa Fluor 488 donkey anti-mouse IgG (1:500, Invitrogen).
Cells with clear lymphocyte morphology, distinct nuclear staining for Foxp3 or cytoplasmic staining for CCR4, PD-1, and TIM-3 were evaluated as positive. Foxp3+ or CCR4+ cells were quantified in 10 representative fields of each slide (40× magnification) using the ImageJ software (24).
Analysis of RNA-seq data set
We conducted RNA-seq as described (20). Raw sequence data for RNA-seq of canine tumor–normal pairs have been deposited in the DDBJ Sequenced Read Archive repository (http://trace.ddbj.nig.ac.jp/dra/index_e.html; accession no. DRA005844). We imported gene counts for each sample into R for the differential gene-expression analysis with EdgeR. Using the TMM normalization and Tagwise dispersion (individual dispersion for each gene), we adjusted abundance differences across samples and identified differentially expressed genes (DEGs). Of DEGs, Treg migration-related genes were extracted and imported into Cluster3.0 and Java TreeView (version 1.1.6r4) for hierarchical clustering analysis and visualization.
Quantitative RT-PCR
Using 2-step real-time RT-PCR (Thermal Cycler Dice Real-Time System; Takara Bio), we quantified mRNA expression of chemokines identified by the RNA-seq analysis. The TATA-box binding protein (TBP) was used as reference. Supplementary Table S4 presents the primer pair sequences.
Flow cytometry
For Foxp3 intracellular staining, fixation and permeabilization of canine peripheral blood mononuclear cells (PBMC) were with the Fixation/Permeabilization Solution Kit (BD Biosciences). We used the following monoclonal antibodies (mAb) for flow cytometry of PBMCs: CD4 (clone CA13.1E4; Leukocyte Antigen Biology Laboratory), CD8 (clone CA9.JD3; Leukocyte Antigen Biology Laboratory), CCR4 (clone 1G1; BD Biosciences), and Foxp3 (clone FJK-16s; eBioscience). All analyses were performed with FACSVerse (BD Biosciences).
We mechanically divided fresh tumor tissues into 1-mm3 pieces followed by digestion with collagenase (0.5 g/L) at 37°C for 30 minutes to examine TILs by flow cytometry. Digested tissue was passed through a 100-mm sieve. Isolated cells were fixed, permeabilized, and stained by the following mAbs: CD4, CD8, CD45 (clone CA12.10C12; Leukocyte Antigen Biology Laboratory), CCR4, and Foxp3.
Cell culture
Three canine bladder cancer cell lines (TCCUB, Sora, and Love) were kindly provided by Drs. K. Saeki, T. Nakagawa, and R. Nishimura (The University of Tokyo; ref. 25) in 2015 and were used for experiments within three months. We also used a cell line derived from a spontaneous canine B-cell leukemia, namely, GL-1, and a canine T-cell lymphoma cell line, namely, CL-1 (26), for CCR4 expression and chemotaxis assays. These cells were grown in RPMI-1640 (Sigma), supplemented with 10% FBS (Gibco) and 100 μg/mL penicillin and streptomycin (Sigma) at 37°C under 5% CO2. Cell lines were not reauthenticated or tested for cross-contamination between cell lines in the past year. All cell lines were negative for Mycoplasma.
Gene-expression analysis
For the chemokine gene-expression analysis by RT-PCR, we extracted total RNA from canine bladder cancer cells using the RNeasy Mini Kit (Qiagen). RT was accomplished using a ReverTra Dash Kit (Toyobo), per the manufacturer's protocol. In addition, PCR was performed over 30 cycles using Taq polymerase (AmpliTaq Gold; Thermo Fisher Scientific) under a three-step thermal cycling of 94°C for 30 seconds, 60°C for 30 seconds, and 72°C for 30 seconds. Supplementary Table S4 presents the primer pair sequences.
For CCR4 gene-expression analysis, total RNA was extracted from two canine lymphoid cell lines, and RT-PCR was performed described above with primers specific to canine CCR4: 5′-CCC TAA GCC TTG CAC CAA AGA-3′ (forward) and 5′-TGT ACT TGA ACA GGA CCA CAA CCA-3′ (reverse). TBP was used as reference.
ELISA
For ELISA, we cultured the three cell lines until confluent and then incubated bladder cancer cells in RPMI-1640 without FBS at 37°C in 5% CO2 for 24 hours. After incubation, the supernatants of the culture media were collected and immediately frozen at −80°C until subsequent ELISA.
We measured canine CCL17 protein in serum, urine, and culture medium samples using the Canine TARC/CCL17 ELISA Kit (Cusabio), per the manufacturer's protocol. Urinary creatinine concentration was evaluated using the LabAssay Creatinine Kit (Wako), and urinary CCL17 was expressed as pg/mg of creatinine (Cre).
Chemotaxis assay
To evaluate the cross-reactivity of mogamulizumab to canine CCR4, we performed a transwell chemotaxis assay using a 96-well Boydenchamber assay with a polycarbonate filter (3-μm pores, Chemotaxicell, Kurabo). The recombinant CCL17 was diluted in RPMI-1640 medium from 1 to 100 nmol/L, and 120 μL of the diluted solution was placed in the lower chamber. GL-1 and CL-1 cells were resuspended in RPMI-1640 at 2 × 106 cells/mL, and 70 μL of each cell suspension was added to the upper chamber. In some experiments, anti-human CCR4 (mogamulizumab, Kyowa Hakko Kirin; 1, 10, or 100 μg/mL) or CCR4 antagonist (C-021, Tocris; 1, 10, or 100 nmol/L) was added to the upper chamber. After incubating for 3 hours at 37°C under 5% CO2, the membrane filters were removed, fixed, and stained with Wright–Giemsa. Migrated cells were counted in five randomly selected high-power fields (×400). The assay was performed in triplicate and repeated 3 times in separate experiments.
Canine tumor–engrafted murine model
We established a canine tumor–bearing mouse model with canine immune cells to evaluate the effect of anti-CCR4 as described previously (27) with a slight modification. We isolated PBMCs from two healthy dogs to use as effector cells in immunodeficient mice. Male severe immunodeficient NOD/Shi-scid IL2Rγ-null (NOG) mice (In-Vivo Science) were inoculated with 2 × 106 canine bladder cancer cells (Love) subcutaneously at the left flank. Next, we divided tumor-bearing mice into four groups for treatment with (i) saline control (vehicle), (ii) canine PBMCs, (iii) anti-CCR4 (mogamulizumab), and (iv) canine PBMCs + mogamulizumab. Mogamulizumab (100 μg/mouse; Kyowa Hakko Kirin) or vehicle injections into the tail veins of mice were started 4 days after tumor inoculation, when the mean tumor volume reached approximately 40 mm3, and continued every other day. In addition, canine PBMCs (5 × 106) or vehicle injections into the tail veins were initiated 4 days after tumor inoculation and continued every fourth day. We determined tumor volume using a caliper and applying the following formula to approximate the volume of a spheroid: 0.52 × (width)2 × (length). After 30 days of tumor inoculation, mice were sacrificed and tumor tissues were collected for histologic analysis and flow cytometry.
Mogamulizumab injection into healthy dogs
To examine the efficacy and safety of anti-CCR4 treatment in dogs, mogamulizumab (0.01, 0.1, or 1 mg/kg) was intravenously injected into four healthy beagles (two males and two females). Before and after injection (1, 3, 7, 14, 21, and 28 days), EDTA blood samples were collected and PBMCs were isolated. The following mAbs were used for flow cytometry of PBMCs: CD4 (clone CA13.1E4) and CCR4 (clone 1G1). Adverse events were assessed and classified according to the Veterinary Cooperative Oncology Group (VCOG) criteria (28). A 6-week washout period was carried out among the dose escalation mogamulizumab trials.
Canine clinical trial design
Anti-CCR4 (mogamulizumab; 1 mg/kg; Kyowa Hakko Kirin) was administered by 30-minute intravenous infusions once every 3 weeks. We selected a dosage of 1 mg/kg. The administration interval was based on preliminary data using experimental dogs. The treatment was continued for 1 year or until dogs experienced disease progression, had unacceptable toxicity, or their owners stopped adhering to the study protocol. Piroxicam (Pfizer, 0.3 mg/kg), the standard drug used to treat canine bladder cancer, was administered every 24 hours starting at the time of mogamulizumab treatment. We used age-, sex-, and tumor stage–matched dogs with bladder cancer, treated with only piroxicam, as matched controls. No placebo control, blinding, or randomization was performed in this study.
Clinical assessment
Dogs were examined for clinical responses and toxicity at least once every 3 weeks by owner observations, physical exam, complete blood counts, serum chemical profiles, 3-view thoracic radiography, and abdominal ultrasonography. A single ultrasound operator measured bladder masses following a standardized mapping procedure (29, 30) and recorded estimated tumor volume. We used ultrasound for imaging because it could be conducted without general anesthesia (as would be required for computed tomography in dogs). We defined the tumor response as follows (31, 32): complete remission (no tumor lesions detected), partial response (PR; ≥50% decrease in tumor volume and no new tumor lesions), stable disease (SD; <50% change in tumor volume and no new tumor lesions), and progressive disease (PD; ≥50% increase in tumor volume or the development of new tumor lesions). Progression-free survival (PFS) was defined as the time from the start of treatment until PD or death at the end of the study (July 10, 2018), and OS as the time from the start of treatment until death of the animal at the end of the study. Toxicity was assessed by VCOG criteria (28).
Biomarkers
For biomarker assessment, we collected fresh urine samples pretreatment to use for CCL17 measurements by ELISA. Blood samples were collected for PBMC isolation and analyzed for the CD4, CD8, Foxp3, and CCR4 expression by flow cytometry.
Statistical analysis
All data presented as bar or line graphs are presented as mean ± SEM. We used JMP Pro version 11.2 (SAS Institute) for statistical analyses. Between-group comparisons were done with the Mann–Whitney U test. The Kruskal–Wallis test, followed by the Dunn test, was used for multiple comparisons. The two-way ANOVA with Bonferroni correction compared tumor volumes in canine tumor-engrafted mice. Correlation between two variables was assessed using the Spearman rank correlation coefficient. Survival curves were generated by the Kaplan–Meier method and compared using the log-rank test. We set the statistical significance as P < 0.05.
Results
Tumor-infiltrating Tregs correlate with poor prognosis in canine bladder cancer
In certain tumors, Treg infiltration correlates with a poor prognosis. We assessed tumor-infiltrating Tregs and the correlation between their number and prognosis in canine naturally occurring bladder cancer. Tissue samples were assessed from 26 dogs with bladder cancer (Supplementary Table S1); all these dogs underwent total cystectomy (21). Based on the WHO TNM classification for canine bladder cancer (22, 23), 7 of 26 (27%) tumors were classified as T1 (superficial papillary tumor), 14 of 26 (54%) as T2 (muscle-invasive), and 5 of 26 (19%) as T3 (tumor invading neighboring organs). We detected nodal and distant (to the lung) metastases in 2 (8%) and 1 (4%) dog, respectively. After total cystectomy, 6 dogs received no treatment, 14 received nonsteroidal anti-inflammatory drugs (NSAID: piroxicam or firocoxib), and 6 were treated with chemotherapy (carboplatin, cisplatin, mitoxantrone, or cyclophosphamide) combined with NSAIDs.
We visualized Foxp3 with immunohistochemistry and examined the localization and number of tumor-infiltrating Tregs. We observed only a few Foxp3+ Tregs in the normal canine bladder, whereas Foxp3+ Tregs were present abundantly in both the intratumoral area and peritumoral stroma of canine bladder cancer (Fig. 1A). Compared with normal tissues, both intratumoral and peritumoral Tregs were more frequent in dogs with bladder cancer (Fig. 1B). In addition, the immune-checkpoint molecules PD-1 and TIM-3 were expressed on some mononuclear cells with a lymphoid morphology (Supplementary Fig. S1). Expression of immunosuppressive cytokine IL10 was elevated in canine bladder cancer compared with normal controls (Fig. 1C), whereas no significant difference was noted in TGFβ expression between normal and tumor tissues.
During follow-up, 19 of 26 dogs died (18 dogs of progression of bladder cancer; 1 dog of perioperative complications). At 1,095 days, 7 dogs were alive. The median DFS and OS in dogs with bladder cancer were 135 (range, 3–1,026) days and 278.5 (range, 3–1,026) days, respectively. Based on the median number, we classified each case as having a high or low density of Foxp3+ Tregs, then assessed the correlation between tumor-infiltrating Tregs and prognosis. The DFS and OS for cases with high intratumoral Treg infiltration were shorter than those for cases with low intratumoral Tregs (Fig. 1D). Moreover, peritumoral Treg infiltration correlated with shorter DFS and OS (Fig. 1E). These findings suggest that tumor-infiltrating Tregs alter the clinical outcome for dogs with bladder cancer.
CCL17 expression is elevated in canine bladder cancer
To identify molecules guiding Treg infiltration in canine bladder cancer, we explored DEGs of chemoattractants for Tregs using a previously generated RNA-seq data set (20). We found 12 chemokine genes to be differentially expressed in canine bladder cancer compared with normal tissues (Fig. 2A). Quantitative PCR showed gene expression of CCL17 in bladder cancer increased by approximately 150-fold compared with normal canine bladder (Fig. 2B). Urinary CCL17 protein concentration was increased in dogs with bladder cancer (Fig. 2C). However, no significant difference was noted in serum CCL17 concentration between normal dogs and dogs with bladder cancer (Fig. 2D).
Tumor-infiltrating Tregs express CCR4 in canine bladder cancer
CCL17 induces chemotaxis through the chemokine receptor CCR4 (33). We performed immunohistochemistry to determine whether CCR4-expressing cells infiltrate canine bladder cancer tissues. As expected, CCR4+ cells with mononuclear lymphoid morphology were observed in bladder cancer but not in normal tissues (Fig. 2E). More CCR4+ cells were evident in dogs with bladder cancer (Fig. 2F), their density positively correlated with the number of tumor-infiltrating Foxp3+ Tregs (Fig. 2G). Flow cytometry of TILs confirmed that CD45+CD4+Foxp3+ Tregs highly expressed CCR4 compared with CD45+CD4+Foxp3− T cells (Fig. 2H and I). Furthermore, the number of CCR4-expressing CD45+CD4+Foxp3+ Tregs inversely correlated with CD45+CD8+ TILs (Fig. 2J). These findings suggest that Tregs infiltrate into the tumor tissue through the CCL17/CCR4 axis in canine bladder cancer.
Anti-CCR4 inhibits Treg infiltration and tumor growth in canine cancer–engrafted mouse model
A humanized anti-human CCR4, mogamulizumab, is commercially available for the treatment of CCR4+ adult T-cell leukemia/lymphoma (12). To confirm the cross-reactivity of mogamulizumab to canine CCR4, we used two canine lymphoid cell lines, CL-1 and GL-1. RT-PCR and flow-cytometric analyses revealed that CCR4 mRNA and protein were expressed in CL-1 but not in GL-1 (Supplementary Fig. S2A and S2B). Recombinant canine CCL17 induced dose-dependent migration of CL-1, which express CCR4, but not GL-1, which do not express CCR4 (Supplementary Fig. S2C). Anti-CCR4 treatment, using mogamulizumab or CCR4 antagonist C-021, inhibited the migration of CL-1 (Supplementary Fig. S2D), suggesting cross-reactivity of mogamulizumab to canine CCR4.
We further investigated whether mogamulizumab treatment inhibits Treg infiltration into the tumor tissue in canine bladder cancer–engrafted mouse model. For the development of the xenograft mouse model, we used three canine bladder cancer cell lines, namely, TCCUB, Sora, and Love (25). Consistent with the tissues from clinical cases, high mRNA expression of CCL17 was detected in all the cell lines (Fig. 3A). In addition, these bladder cancer cells produced CCL17 protein (Fig. 3B). As Love cells exhibited the highest production of CCL17 (Fig. 3B), we used this cell line for the xenograft mouse model in subsequent experiments. Canine bladder cancer cells were implanted into severe immunodeficient NOG mice, and then canine PBMCs, including both effector T cells and Tregs, and/or mogamulizumab were injected into mice (Fig. 3C). The injection of mogamulizumab, together with canine PBMCs, inhibited tumor growth of canine bladder cancer cells implanted in mouse flanks (Fig. 3D). At 30 days after the implantation, the tumor size was almost one third the size compared with that of the vehicle group. The injection of only canine PBMCs or mogamulizumab did not affect tumor growth (Fig. 3D). The histologic analysis revealed that a necrotic layer spread throughout the tumor tissues (Fig. 3E). The treatment with mogamulizumab in combination with canine PBMCs increased the necrotic area in the tumor (Fig. 3E and F). In addition, immunohistochemistry showed that Foxp3+ Tregs were observed in tumor tissues of NOG mice injected with canine PBMCs (Fig. 3G). Mogamulizumab treatment reduced the number of tumor-infiltrating Tregs in mice with canine PBMCs (Fig. 3G and H). Flow cytometry confirmed the decline of CCR4+Foxp3+ tumor-infiltrating Tregs in NOG mice with canine PBMCs and mogamulizumab (Fig. 3I). These observations suggest that anti-CCR4 treatment inhibits Treg infiltration and tumor growth in canine bladder cancer–engrafted mice.
Anti-CCR4 treatment improves survival in spontaneous canine bladder cancer
Next, we assessed the efficacy and safety of mogamulizumab in healthy dogs. Single intravenous administration of mogamulizumab induced a dose-dependent reduction of circulating CCR4+CD4+ T cells, and the effect lasted about 3 weeks (Supplementary Fig. S2E and S2F). We observed no adverse effect in dogs with mogamulizumab in a dose range 0.01–1 mg/kg. Hence, we selected the dosage of 1 mg/kg in a subsequent canine clinical trial.
To evaluate the clinical activity of anti-CCR4 treatment in spontaneous canine bladder cancer, we compared 14 dogs that received mogamulizumab (1 mg/kg every 3 weeks) and piroxicam (0.3 mg/kg every 24 hours) with a cohort of 14 age-, sex-, and tumor stage-matched dogs that received only piroxicam (Supplementary Table S2). Urethra involvement was observed in 3 (22%) dogs treated with mogamulizumab/piroxicam and 2 (14%) dogs with piroxicam only. All dogs that received mogamulizumab treatment in combination with piroxicam had a reduction in the tumor burden (Fig. 4A and B). In 14 dogs with mogamulizumab/piroxicam, 10 (71%) obtained a PR, 4 (29%) had SD, and none had PD. In 14 dogs with only piroxicam, 2 (14%) obtained PR, 9 (64%) had SD, and 3 (22%) had PD (Fig. 4B). Dogs with only piroxicam deteriorated during the first 1- to 5-month period, and all dogs discontinued treatment 160 days after the start of treatment because of disease progression (Fig. 4C, left). In contrast, mogamulizumab treatment in combination with piroxicam induced a clinical response at a median of 21 (range, 21–105) days from starting treatment (Fig. 4C, right). At the cutoff time for the study data (July 10, 2018), 6 (43%) dogs that received mogamulizumab treatment were still alive. The median PFS in dogs treated with piroxicam only and dogs with mogamulizumab/piroxicam were 76 (range, 21–161) days and 189 (range, 91–397) days, respectively. The median OS in dogs treated with piroxicam only and dogs with mogamulizumab/piroxicam were 241 (range, 108–516) days and 474 (range, ≥259) days, respectively. The PFS and OS in dogs with mogamulizumab treatment in combination with piroxicam were longer than those with only piroxicam (Fig. 4D).
In 14 dogs with mogamulizumab/piroxicam, seven (50%) had an adverse event (Table 1). All treatment-related adverse events were grade 1 or 2, and many were transient. The leading adverse events were vomiting (29%), rash (14%), diarrhea (14%), and infusion-related reaction (14%). There were no grade 3–5 treatment-related adverse events, and no dog exhibited an event leading to treatment withdrawal. We observed lymphopenia in some cases (14% with grade 1 and 14% with grade 2), which was considered as the pharmacologic effect of mogamulizumab.
. | Number of cases (%)b . | |||
---|---|---|---|---|
Eventa . | Grade 1 . | Grade 2 . | Grades 3–5 . | Total . |
Any event | 7 (50.0) | 5 (35.7) | 0 | 7 (50.0) |
Nonhematologic event | ||||
Vomiting | 3 (21.4) | 1 (7.1) | 0 | 4 (28.6) |
Rash | 2 (14.3) | 0 | 0 | 2 (14.3) |
Diarrhea | 2 (14.3) | 0 | 0 | 2 (14.3) |
Infusion-related reaction | 1 (7.1) | 1 (7.1) | 0 | 2 (14.3) |
Increased CRP | 0 | 2 (14.3) | 0 | 2 (14.3) |
Lethargy/fatigue | 1 (7.1) | 0 | 0 | 1 (7.1) |
Anorexia | 1 (7.1) | 0 | 0 | 1 (7.1) |
Urticaria | 1 (7.1) | 0 | 0 | 1 (7.1) |
Increased ALP | 0 | 1 (7.1) | 0 | 1 (7.1) |
Hematologic event | ||||
Lymphopenia | 2 (14.3) | 2 (14.3) | 0 | 4 (28.6) |
. | Number of cases (%)b . | |||
---|---|---|---|---|
Eventa . | Grade 1 . | Grade 2 . | Grades 3–5 . | Total . |
Any event | 7 (50.0) | 5 (35.7) | 0 | 7 (50.0) |
Nonhematologic event | ||||
Vomiting | 3 (21.4) | 1 (7.1) | 0 | 4 (28.6) |
Rash | 2 (14.3) | 0 | 0 | 2 (14.3) |
Diarrhea | 2 (14.3) | 0 | 0 | 2 (14.3) |
Infusion-related reaction | 1 (7.1) | 1 (7.1) | 0 | 2 (14.3) |
Increased CRP | 0 | 2 (14.3) | 0 | 2 (14.3) |
Lethargy/fatigue | 1 (7.1) | 0 | 0 | 1 (7.1) |
Anorexia | 1 (7.1) | 0 | 0 | 1 (7.1) |
Urticaria | 1 (7.1) | 0 | 0 | 1 (7.1) |
Increased ALP | 0 | 1 (7.1) | 0 | 1 (7.1) |
Hematologic event | ||||
Lymphopenia | 2 (14.3) | 2 (14.3) | 0 | 4 (28.6) |
aCRP, C-reactive protein; ALP, alkaline phosphatase.
bToxicity grade based on published criteria (32).
Overall, these results suggest that anti-CCR4 treatment leads to clinical responses and improves survival without severe adverse events in dogs with spontaneous bladder cancer.
Circulating CCR4+ Tregs and urinary CCL17 are biomarkers for responses to anti-CCR4
During mogamulizumab treatment, circulating CCR4+ Tregs were monitored. A reduction in circulating CCR4+ Tregs was observed after treatment (Fig. 5A). The reduction persisted during treatment (Fig. 5B). The pretreatment numbers of circulating CCR4+ Tregs in each case were not related to the maximum tumor volume reduction (Fig. 5C, left). In contrast, the reduced numbers of Tregs after the mogamulizumab treatment were significantly correlated with the reduced tumor volumes (Fig. 5C, right), further indicating that Treg depletion is needed for the clinical benefit of an anti-CCR4 treatment. However, reduced numbers of circulating CCR4+ Tregs after mogamulizumab treatment were not related to the PFS (P = 0.351) and OS (P = 0.257). In tumor tissues of a dog collected before and after therapy, we confirmed that mogamulizumab treatment decreased tumor-infiltrating Foxp3+ and CCR4+ cells (Supplementary Fig. S3).
We further examined the association of pretreatment urinary CCL17 with response. Dogs with PR had more urinary CCL17 than did dogs with SD (Fig. 5D, left). Similarly, urinary CCL17 correlated with tumor volume reduction (Fig. 5D, right) but not with PFS (P = 0.135) and OS (P = 0.662). These findings suggest that circulating CCR4+ Tregs and urinary CCL17 were useful biomarkers for predicting the clinical response to mogamulizumab treatment.
Tumor-infiltrating Tregs express CCR4 in human bladder cancer
Given the promising outcomes and favorable clinical responses of anti-CCR4 treatment in the comparative canine trial, we examined whether tumor-infiltrating Tregs express CCR4 in human bladder cancer. We observed a number of Foxp3+ Tregs and CCR4+ cells in human bladder cancer tissues but not in the normal bladder (Fig. 6; Supplementary Table S3). Immunohistochemistry in serial sections of human bladder cancer revealed mononuclear lymphoid cells stained positive for Foxp3 and CCR4 in the tumor tissues (Fig. 6A). Double-labeling immunofluorescence confirmed that Foxp3+ tumor-infiltrating Tregs expressed CCR4 (Fig. 6B). These findings suggest that anti-CCR4 may have therapeutic value for the treatment of human bladder cancer.
Discussion
Evidence suggests that tumor-infiltrating Foxp3+ Tregs are a potential target for cancer immunotherapy. Various molecules, including CD25, CD15s, CD134, cytotoxic T-lymphocyte associated protein 4 (CTLA-4), glucocorticoid-induced TNF receptor (GITR), lymphocyte-activation gene 3 (LAG3), and CCR4, are good candidates for Treg depletion or functional modulation (6, 10, 34). In this study, we have shown that Foxp3+ Tregs infiltrate into tumor tissues through the CCL17/CCR4 pathway, and CCR4 blockade exerts an antitumor effect in a canine tumor–engrafted mouse model as well as in dogs with spontaneous muscle-invasive bladder cancer. In our canine clinical trial, objective response rates (ORR) of mogamulizumab in combination with piroxicam treatment were 71% (10 of 14 dogs), and the median PFS and OS were 189 days and 474 days, respectively, although the sample size is relatively small. The results compare well with tests of first-line cisplatin-based regiments (ORR, 11%–73%; median PFS, 78–186 days; median OS, 179–329 days) or mitoxantrone in combination with piroxicam (ORR, 35%; median PFS, 194 days; median OS, 350 days) in dogs with bladder cancer (35–38). We also found that tumor-infiltrating Foxp3+ Tregs express CCR4 in human bladder cancer tissues. These observations collectively suggest the potential of mogamulizumab for the treatment of patients with bladder cancer.
Mogamulizumab treatment was well tolerated in dogs. Most treatment-related adverse events were of grade 1 or 2, and vomiting was manageable with systemic antiemetic treatment alone. Allergic reactions such as rash and urticaria were observed but mild, and intervention was not required in this canine clinical trial. Consistent with the pharmacologic effect of mogamulizumab, 29% of dogs had a reduction of lymphocyte counts from pretreatment counts; however, there was no increased risk of clinically threatening infections. In humans, the most common adverse events associated with mogamulizumab treatment were infusion reactions (21%–89%), skin rashes (63%), chills (23.8%–59%), and nausea (19%–31%), which were manageable and reversible (12, 39). These safety data, including a lack of renal toxicity, suggest that patients with bladder cancer, who are often elderly and predisposed to having renal impairment, may be better able to tolerate mogamulizumab treatment than chemotherapy.
The present study showed that the likelihood of mogamulizumab response (tumor volume reduction) can be increased by determining urinary CCL17 concentration or the reduction of peripheral blood CCR4+ Tregs. By contrast to the circulating CCR4+ Tregs testing that needs to be performed before and after administration, assessment of urinary CCL17 concentration can predict the therapeutic effect of mogamulizumab before treatment. However, these biomarkers were not associated with PFS and OS. This discrepancy may be explained by the small sample size or lack of correlation between clinical response and survival in dogs with bladder cancer. At present, the prognostic value of urinary CCL17 and circulating CCR4+ Tregs is unknown. Further studies will be necessary to examine the utility of these biomarkers for predicting outcomes of mogamulizumab treatment.
Systemic immunotherapy targeting immune-checkpoint proteins, such as PD-1 and the ligand PD-L1, is clinically efficacious in various stages of bladder cancer (40, 41). As of May 2019, five inhibitors of the PD-1/PD-L1 pathway were approved by the FDA. A number of ongoing trials are evaluating targets in combination with anti–PD-1/PD-L1 therapy (40, 41). Immunotherapy can be effective for the treatment of bladder cancer. Whether CCR4 blockade can enhance the clinical efficacy of immune-checkpoint inhibitors warrants investigation.
Over the last few decades, human tumor cell line–based or patient-derived xenografts in immunocompromised mice have been applied in the preclinical testing of anticancer agents; however, there are several limitations in these murine models (42). First, the immunocompromised nature of the mice makes it impossible to examine therapeutic agents targeting the immune system. Secondly, subcutaneous or orthotopic implantation of the tumor does not accurately reflect the natural tumor microenvironment. Thirdly, tumor transplantation into the mouse may result in alterations of genetic, histopathologic, and molecular characteristics that may not mirror the human tumor accurately. On the other hand, dogs with spontaneous cancer better resemble humans with regard to heterogeneity, clinical sign, histopathology, disease progression, metastatic behavior, and immunologic phenotypes (16–18, 43, 44). The present study showed that CCR4 mediates Treg infiltration into the tumor tissues in dogs, mirroring a commonly observed feature in humans (10, 45). Examples of canine malignancies that share genotypic and phenotypic similarities with human counterparts include non-Hodgkin lymphoma, osteosarcoma, mammary carcinoma, melanoma, and soft-tissue sarcoma (43, 44, 46). A variety of emerging therapies, such as demethylating agents, folate analogues, Bruton tyrosine kinase inhibitor, HER2/neu-targeting cancer vaccine, heat shock protein 90 inhibitor, and topoisomerase I inhibitors, have demonstrated clinical activity for the treatment of these cancers in dogs (31, 32, 47–50), thereby accelerating human clinical trials. Studies of canine spontaneous cancer provide an opportunity to evaluate therapeutic drugs in translational studies linking basic studies with human clinical trials.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: S. Maeda, N. Matsuki
Development of methodology: S. Maeda
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Maeda, K. Murakami, A. Inoue
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Maeda, A. Inoue
Writing, review, and/or revision of the manuscript: S. Maeda, N. Matsuki
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T. Yonezawa
Study supervision: T. Yonezawa, N. Matsuki
Acknowledgments
We thank Drs. M. Tsuboi, J.K. Chambers, and K. Uchida for the histopathologic review of the canine samples and Dr. H. Tomiyasu for the technical assistance of RNA-seq analysis. Canine TCC cell lines were kindly provided by Drs. K. Saeki, T. Nakagawa, and R. Nishimura; the recombinant canine CCL17 was a generous gift from Nippon Zenyaku Kogyo Co., Ltd. We would like to acknowledge and thank the canine patients, the owners, and the clinical care team at the Veterinary Medical Center of the University of Tokyo. This work was supported by JSPS KAKENHI Grant-in-Aid for Young Scientists (A; grant no. JP16H06208 to S. Maeda) and Anicom Capital Research Grant (EVOLVE to S. Maeda).
The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.
Note: Supplementary data for this article are available at Cancer Immunology Research Online (http://cancerimmunolres.aacrjournals.org/).