Tumor endothelial cells (TEC) lining tumor blood vessels actively contribute to tumor progression and metastasis. In addition to tumor cells, TEC may develop drug resistance during cancer treatment, allowing the tumor cells to survive chemotherapy and metastasize. We previously reported that TECs resist paclitaxel treatment via upregulation of ABCB1. However, whether TEC phenotypes are altered by anticancer drugs remains to be clarified. Here, we show that ABCB1 expression increases after chemotherapy in urothelial carcinoma cases. The ratio of ABCB1-positive TEC before and after first-line chemotherapy in urothelial carcinoma tissues (n = 66) was analyzed by ABCB1 and CD31 immunostaining. In 42 cases (64%), this ratio increased after first-line chemotherapy. Chemotherapy elevated ABCB1 expression in endothelial cells by increasing tumor IL8 secretion. In clinical cases, ABCB1 expression in TEC correlated with IL8 expression in tumor cells after first-line chemotherapy, leading to poor prognosis. In vivo, the ABCB1 inhibitor combined with paclitaxel reduced tumor growth and metastasis compared with paclitaxel alone. Chemotherapy is suggested to cause inflammatory changes in tumors, inducing ABCB1 expression in TEC and conferring drug resistance. Overall, these findings indicate that TEC can survive during chemotherapy and provide a gateway for cancer metastasis. Targeting ABCB1 in TEC represents a novel strategy to overcome cancer drug resistance.
These findings show that inhibition of ABCB1 in tumor endothelial cells may improve clinical outcome, where ABCB1 expression contributes to drug resistance and metastasis following first-line chemotherapy.
Urothelial carcinoma, which occurs in the urinary tract, is the most common histologic type of bladder, ureter, and urethral cancers. Urological malignant tumors are the second most common cancers after prostate carcinoma (1), and the bladder is the most frequent site of urothelial carcinoma. Transurethral tumor resection is generally performed as curative surgery for superficial tumors, but in the case of advanced tumors, multimodality treatment is required. Cisplatin-based neoadjuvant chemotherapy and total cystectomy are performed for the treatment of muscle-invasive bladder cancers (2); however, its recurrence rate is about 40% (3), and once recurrence occurs, the prognosis is poor. In addition, in cases with inoperable locally advanced or metastatic bladder cancers, combination chemotherapy with gemcitabine and cisplatin is widely used as a first-line treatment (4); however, the duration of response is limited, and most patients become refractory to first-line treatment. Although second-line chemotherapy, including taxane, is performed, the prognosis in refractory patients remains poor (5).
In most cases, anticancer drugs become ineffective in cancer treatment, and this is thought to be caused by acquiring drug resistance in tumor cells (6) or existing cancer stem cells in the tumor microenvironment (7). Drug resistance remains an important clinical challenge for cancer treatment. It is generally known that tumor cells acquire anticancer drug resistance via phenotypic changes, such as increased drug transporter expression (8). Conversely, it has been reported that tumor stromal cells can be altered and are involved in tumor progression and chemoresistance (9, 10). For example, the recruitment of bone marrow–derived immunosuppressive cells or secreting factors from cancer-associated fibroblasts may contribute to tumor progression and chemoresistance (11, 12).
Tumor endothelial cells (TEC), which line tumor blood vessels, are also tumor stromal cells. In recent years, it has been reported that TECs have various abnormalities and diversities (13–15). Moreover, it has been realized that TECs contribute to the promotion of tumor malignancy and metastasis (16, 17). Some TECs have chromosomal abnormalities (18, 19) and characteristics of stem cells (15, 20). We have reported that TECs express high levels of a drug efflux transporter, ABCB1, which is encoded by the multi-drug resistance gene (MDR1), and acquire resistance to anticancer drugs such as paclitaxel, an ABCB1 substrate (21). Furthermore, we demonstrated that resistance to paclitaxel in TECs was diminished by combination therapy with an ABCB1 inhibitor, which enhanced the antiangiogenic effect and antitumor effect in the melanoma xenograft model (22). TECs remaining after chemotherapy can supply nutrients and oxygens to tumor cells, which causes regrowth and metastasis. Thus, it has been considered necessary to understand the mechanism of drug resistance in not only tumor cells but also TECs to overcome drug resistance in cancer therapy. However, it is still unknown whether TEC phenotypes are altered by anticancer drugs. If anticancer drugs can induce phenotypic changes in TECs and cause drug resistance, development of novel therapeutic strategies is imperative that target TECs or molecule(s) involved in the mechanism of TEC chemoresistance to avoid tumor metastasis. In this study, we focused on the changes in the expression of ABCB1 in tumor blood vessels during chemotherapy and elucidated the mechanisms of ABCB1 upregulation by chemotherapy.
Materials and Methods
Human tissue samples
Tumor tissues were surgically resected from patients who were clinically diagnosed with urothelial carcinoma at the Hokkaido University Hospital (Sapporo, Hokkaido, Japan) and related hospitals (Table 1). All protocols were approved by the Institutional Ethics Committee of Hokkaido University (Sapporo, Hokkaido, Japan), and written informed consent was obtained from each patient before surgery. Final pathologic diagnosis of the cases was confirmed by the examination of formalin-fixed surgical specimens.
HMVECs were purchased from Lonza and cultured in EGM-2MV Medium (Lonza). Human bladder cancer cell lines UMUC3, J82, T24, 5637, and tdTomato-Luc2 gene-transfected UMUC3 cells were kindly provided by Dr. Tanaka (Department of Cancer Pathology, Hokkaido University Graduate School of Medicine, Sapporo, Hokkaido, Japan). UMUC3 and J82 cells were cultured in DMEM (Sigma) supplemented with 10% heat-inactivated FBS. T24 and 5637 cells were cultured in RPMI1640 medium (Sigma) supplemented with 10% FBS. Gemcitabine-resistant UMUC3 cells (UMUC3-GEM-R) and cisplatin-resistant UMUC3 cells (UMUC3-CDDP-R) were generated by gradually increasing the concentration of anticancer drugs in the conditioned medium. UMUC3-GEM-R can grow in 25 nmol/L of gemcitabine, and UMUC3-CDDP-R can grow in 3 μmol/L of cisplatin. Cells were cultured at 37°C in a humidified atmosphere containing 5% CO2. The absence of Mycoplasma pulmonis was checked using PCR with TaKaRa PCR Mycoplasma Detection Set.
Six-week-old female nude mice (BALB/c AJcl-nu/nu, Clea Japan) were housed under specific pathogen-free conditions. All procedures for animal care and experimentation adhered to institutional guidelines and were approved by the Ethical Committee for Experimental Animal Care of Hokkaido University (Sapporo, Hokkaido, Japan).
Chemicals and antibodies
The following chemicals and antibodies were purchased: gemcitabine (Wako, #077-05671 or Tokyo Kasei Kogyo Co. Ltd, #G0367), cisplatin (Wako, #033-30091), paclitaxel (Pfizer), paclitaxel (Enzo, #BML-T104), verapamil hydrochloride (Vasoran, Eizai Co. Ltd.), anti-mouse CD31 microbeads (Miltenyi Biotec, #130-097-418), recombinant human IL8 protein (R&D Systems, #208-IL), NF-κB inhibitor BAY11-7082 (Calbiochem), anti-mouse CD31 microbeads (Miltenyi Biotec, #130-097-418), mouse anti-human ABCB1 antibody (Santa Cruz Biotechnology, #sc-13131), mouse anti-human CD31 antibody (Leica Biosystems, #NCL-CD31), mouse anti-human cytokeratin antibody (Nichirei #412811), mouse anti-human IL8 antibody (R&D Systems, #MAB208), rabbit anti-human IL8 antibody (Abcam, #ab7747), mouse IgG1 isotype control (R&D Systems, #MAB002), rabbit anti-mouse/human ABCB1 antibody (LifeSpan Biosciences Inc., #LS-B1448), rabbit anti-mouse/human MDR1 antibody (Bioss Antibody Inc., #bs-0563R), rabbit anti-mouse CD31 antibody (Abcam, #ab28364), rabbit anti-alpha smooth muscle actin (alpha-SMA) antibody (Abcam, #5694), Alexa Fluor 647 rat anti-mouse CD31 antibody (BioLegend, #102515), Alexa Fluor 594 goat anti-rabbit IgG (Life Technologies, #A-11037), rabbit anti-cleaved caspase-3 antibody (Cell Signaling Technology, #9664), rabbit anti-CA9 antibody (Novus Biologicals, #100-417), FITC-dextran (Sigma, #FD70S), HRP-conjugated goat-anti-rabbit IgG (Dako, #P0448), alkaline phosphatase-conjugated goat-anti-rabbit IgG (Dako, #D0487), anti-β-actin antibody (Cell Signaling Technology, #4970), anti-phospho-NFκB p65 antibody (Ser536) (Cell Signaling Technology, #3033), anti-NFκB p65 antibody (Cell Signaling Technology, #8242), and HRP-conjugated anti-rabbit IgG (Cell Signaling Technology, #7074).
Isolation of RNA and quantitative PCR
Total RNA was isolated using ReliaPrep RNA Cell Miniprep System (Promega, Z6012). Tissue RNA was isolated using an RNeasy Mini Kit (Qiagen). cDNA was synthesized using ReverTra-Plus (Toyobo) as described previously (16). qRT-PCR was performed using KAPA SYBR FAST qPCR Kit (Kapa Biosystems). Cycling conditions were set based on CFX Manager (Bio-Rad). mRNA expression levels were normalized to those of β-actin and analyzed using the ΔΔCt method. The primers used are listed in Supplementary Table S1.
The IHC staining of CD31, ABCB1, and IL8 in human clinical samples was performed at Morphotechnology Co. The IHC staining of cytokeratin in human clinical samples was performed according to a previously described method (23). Formalin-fixed, paraffin-embedded tissue sections of mouse tumor tissues were prepared as described previously (24) and stained using anti-CD31 antibody, anti-IL8 antibody, anti-CA9 antibody, and anti-alpha-SMA antibody. The liquid DAB+ Substrate Chromogen System (Dako, K3468) was used for horseradish peroxidase (HRP) color development. We used the Vulcan Fast Red Chromogen Kit2 (Biocare Medical, BRR805AS) for alkaline phosphatase color development. Frozen sections of mouse lung tissues were prepared as described previously (15) and stained using anti-human cytokeratin antibody. Nuclei were counterstained with hematoxylin (Wako, 131-09665). Sample images were acquired using NanoZoomer (Hamamatsu Photonics). For double immunofluorescence staining, frozen sections were stained using anti-CD31 and anti-ABCB1 or cleaved caspase-3 antibodies followed by counterstaining with 4,6-diamidino-2-phenylindole (Dojin). Fluorescent staining images were acquired using an FV1000 Confocal Microscope (Olympus). The acquired images were processed using Fluoview FV10-ASM Viewer Software (Olympus).
Evaluation of IHC staining
ABCB1 or IL8 expression in tumors was analyzed by measuring the percentage of ABCB1-positive area and staining intensity and scored by the semi-quantitative scoring system (H score; ref. 25). ABCB1 expression in tumor blood vessels was analyzed by measuring the ABCB1-positive ratio of tumor blood vessels. Five fields of vascular hot spots were selected at low magnification (×100) by staining with CD31 antibody in serial sections. The numbers of ABCB1-positive and -negative blood vessels were counted, and the ABCB1-positive ratio of tumor blood vessels was determined using the average of five fields. In the noncancerous section, the ABCB1-positive ratio of blood vessels was determined using the average of the two fields. The IL8 staining area in UMUC3 tumors was determined in five fields of strongly stained areas at low magnification (×100; n = 4–5 mice per group). The average percentages of IL8-stained areas in five selected fields were measured using NIH ImageJ Software. ABCB1 expression in UMUC3-TECs was determined as the percentage of ABCB1-positive TECs in the total endothelial cell numbers. The cell number was counted in 50 randomly selected fields at high magnification (×600). ABCB1 expression in tumor blood vessels of IL8-knockout or Scr UMUC3 tumor tissues was calculated as the average ABCB1-positive ratio of blood vessels in five fields at low magnification (×100; n = 4 mice per group). Microvessel density (MVD) and cleaved caspase-3–positive proportions in blood vessels were quantified at low magnification (×100). MVD was determined as the percentage of CD31-stained areas in the total area. Cleaved caspase-3–positive proportions in blood vessels were determined as the percentage of cleaved caspase-3–stained areas in CD31-stained areas. The Kruskal–Wallis test was used for the statistical analysis of MVD and cleaved caspase-3–positive ratio in blood vessels in 40 randomly selected fields from all tumors in each group. The CA9-stained area was measured using NIH ImageJ Software and we calculated a ratio of the CA9-stained area by the total tissue area. CD31 and alpha-SMA were double stained for the analysis of pericyte-covered blood vessel. The number of blood vessels stained with CD31 or alpha-SMA was counted in vascular hot spots, and we also calculated the ratio of the double-positive vessels to total number of vessels. The average double-positive ratio was calculated for five fields of vascular hot spots in each tumor. The Kruskal–Wallis test was used for statistical analysis in each group (n = 5 or 6). We calculated the percentage of FITC-dextran–positive areas in the overall area to determine tumor vessel permeability. The Kruskal–Wallis test was used for the statistical analysis of tumor vessel permeability in 50 randomly selected fields from all tumors in each group. Quantitative analysis was performed using NIH Image J Software.
Tumor conditioned medium treatment and IL8 inhibitory assay
Tumor cells were grown to 70%–80% confluence in DMEM with 10% FBS. We replaced the conditioned medium with fresh medium containing 3 μmol/L of cisplatin, 25 nmol/L of gemcitabine, or control solution. The cells were exposed to anticancer drugs or control solution for 8 hours. After the conditioned medium was replaced with fresh medium, the cells were incubated for additional 24 hours (Supplementary Fig. S2A). Conditioned medium was collected from these cells and passed through a 0.22-μm filter (Merck Millipore Ltd., SLGS033SS) to remove the cells. For Western blot assay or FACS analysis, HMVECs were exposed to conditioned medium for 1 or 72 hours. For the IL8 inhibitory assay, IL8-neutralizing antibody was diluted using conditioned medium, and HMVECs were exposed to conditioned medium for 48 hours.
Cell survival assay
After 48 hours of exposure to UMUC3 conditioned medium (prepared as described above), HMVECs were treated with paclitaxel at the indicated concentrations in UMUC3 conditioned medium for 96 hours. Cell viability was assessed using a 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium Inner Salt Assay (Promega Corp).
Anticancer drug treatment and PCR array
For quantitative PCR, bladder cancer cells or HMVECs were treated with either 25 nmol/L of gemcitabine or 3 μmol/L of cisplatin for 24 hours. Then, the RNA (1.0 μg) was isolated from the cells. For the PCR array, UMUC3 cells were treated with 25 nmol/L of gemcitabine or 3 μmol/L of cisplatin for 24 hours. RNA (0.5 μg) was isolated from these cells, and mRNA expressions of various cytokines were analyzed with a PCR Array Kit (Qiagen, RT² Profiler PCR Array Human Cytokines & Chemokines PAHS-150Z).
IL8 treatment assay
HMVECs were treated with 12.5 ng/mL of recombinant IL8 for 48 hours to analyze the effect of IL8 on MDR1 mRNA expression. HMVECs were treated with 12.5 ng/mL of recombinant IL8 for 1, 2, and 4 hours to detect NF-κB signaling.
Cells were lysed using RIPA Buffer (Cell Signaling Technology). The total protein concentration was determined using a BCA Protein Assay Kit (Pierce). Western blotting was performed according to standard methods using antibodies specific for pNF-κB, NF-κB, and β-actin, and an HRP-conjugated secondary antibody (Cell Signaling Technology, #7074) as described previously (16). Quantitative analysis was performed using NIH ImageJ Software.
RNA isolation from UMUC3-TECs
TECs were isolated as described previously (15), with modifications. Briefly, xenografts of UMUC3 tumors were minced, after which, TECs were sorted using a MACS Cell Separation System (Miltenyi Biotec) with anti-CD31 microbeads. The isolated TECs were characterized by flow cytometry for the analysis of CD31 purity using FACS Aria II (Becton Dickinson; Supplementary Fig. S3). Data were analyzed using FlowJo Software (Tree Star Inc.). Another TEC fraction was washed with 1× PBS, and total RNA was isolated as described above. Student t test was used for comparisons of MDR1 expression in CD31-enriched cells between the two groups (n = 3 per group).
FACS analysis of single-cell preparations from UMUC3 tumor tissues
After three cycles of gemcitabine and cisplatin treatment, xenografts of UMUC3 tumors were minced and single cells were isolated as described previously (15). These cells were stained with anti-CD31 and anti-ABCB1 antibodies and characterized by flow cytometry. Student t test was used for comparisons of the ABCB1-positive ratio in CD31-postive cells between the two groups (n = 4 per group).
Serum IL8 concentration was determined using Quantikine ELISA Kits for human IL8 (R&D Systems, #D8000C).
IL8 knockout by CRISPR-Cas9
The clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 system targeting IL8 was purchased from Applied Biological Materials, Inc. (# K1077101). The sgRNA sequence targeting IL8 was 5′-CTAAGTTCTTTAGCACTCCT-3′. We used scrambled sgRNA CRISPR Lentivector (Applied Biological Materials, Inc., # K018) as a negative control. These self-inactivating lentiviral vectors, together with the packaging vector pCAG-HIVgp and the VSV-G- and REV-expressing construct pCMV-VSV-G-RSV-REV (from H. Miyoshi, Department of Physiology, Keio University School of Medicine, Tokyo, Japan) were introduced into HEK293T cells using FuGene HD (Promega), according to the manufacturer's recommendations. Lentivirus-mediated gene transfer was performed as described previously (16). After the selection of a knockout and confirmation using immunocytochemistry (Supplementary Fig. S5A), we used the cells for in vivo assay.
Mouse tumor xenograft model and treatment regimens
UMUC3 cells (4 × 106), tdTomato-Luc2-UMUC3 tumor cells (4 × 106), or IL8-knockout UMUC3 cells (4 × 106) were diluted with an equal volume of Matrigel and sterile Hank's Balanced Salt Solution (HBSS), and were subcutaneously injected into the right flanks of 7-week-old female nude mice (BALB/c AJcl-nu/nu), obtained from Clea Japan. After 9–14 days following tumor injection, gemcitabine/cisplatin treatment was initiated. Each mouse first received intraperitoneal injection of gemcitabine (60 mg/kg), followed by cisplatin (6 mg/kg) the next day. Each treatment regimen was administered once a week for two or three courses. Deionized distilled water or DMSO diluted with sterile HBSS was injected as a control for gemcitabine or cisplatin, respectively. The mice were monitored regularly. Tumor volume was measured using the following standard formula: (shortest diameter)2 × (longest diameter) × 0.5. After 4 days following the last treatment, blood was collected from anesthetized mice by cardiac puncture. Tumor tissues were resected for IHC analyses and RNA isolation. For second-line paclitaxel and verapamil treatment, daily intraperitoneal injections of paclitaxel (1.3 mg/kg) and verapamil (20 mg/kg) were administered 4 days following the third cisplatin injection. HBSS was injected as control. After 16 days of treatment, the intravascular blood, which contained FITC-dextran from either tumor tissues or lungs, was drained by cardiac perfusion with 1× PBS from anesthetized mice. The tumor tissues and lungs were resected, weighed, and examined with IHC analyses.
Unless otherwise indicated, data in the figures are presented as mean ± SD. Box and whisker plots represent median (center line), 25th and 75th percentiles (box), and minimum and maximum (whiskers). The Wilcoxon signed rank test was used to analyze the statistical difference between clinical cases before and after chemotherapy. The Spearman rank correlation test was used to analyze the correlation between IL8 expression in tumor tissues and ABCB1-positive ratio of tumor blood vessels. The Student t test or the Wilcoxon rank-sum test was used for comparisons between the two groups. The Kruskal–Wallis test, followed by the Wilcoxon test for paired comparisons were performed for multiple comparisons. Overall survival or disease-specific survival was defined as the time from the date of surgery before chemotherapy to the date of death due to any cause or urothelial carcinoma, and it was estimated using the Kaplan–Meier method and analyzed using the log-rank test. Statistical significance was taken as P < 0.05 and was representative of three independent experiments. Statistical analysis was performed using JMP version 13 (SAS Institute).
The investigation was conducted in accordance with the ethical standards, the Declaration of Helsinki, and national and international guidelines. All methods have been approved by the Institutional Ethics Committee of Hokkaido University (Sapporo, Hokkaido, Japan), and written informed consent was obtained from each patient before surgery. All procedures for animal care and experimentation adhered to institutional guidelines and were approved by the local animal research authorities.
The number of ABCB1-positive TECs increased after first-line chemotherapy for urothelial carcinoma
To assess the changes of ABCB1 expression in tumor cells during chemotherapy, immunostaining was performed using an anti-ABCB1 antibody in urothelial carcinoma specimens before and after first-line chemotherapy (Fig. 1A; Table 1). We confirmed tumor cells with cytokeratin staining (Supplementary Fig. S1A). ABCB1 expression in tumor cells varied among the cases. There were no significant differences in ABCB1 expression in the tumor cells before and after first-line chemotherapy (Fig. 1B). Next, we analyzed ABCB1 expression in tumor blood vessels, which were visualized with an anti-CD31 antibody (Fig. 1C). Only a few TECs were stained with the anti-ABCB1 antibody in the most tumor tissues collected before first-line chemotherapy. However, interestingly, there were more ABCB1-positive TECs in the tumor tissue after first-line chemotherapy (Fig. 1C). Quantitative analysis showed that the ratio of ABCB1-positive TECs was significantly increased after first-line chemotherapy (Fig. 1D). Conversely, in noncancerous areas, the blood vessels were hardly stained with anti-ABCB1 in both before and after first-line chemotherapy (Fig. 1E; Supplementary Fig. S1B). The ABCB1 expression change in tumor cells by chemotherapy has no tendency. Alternatively, it was suggested that chemotherapy induced ABCB1 expression in TECs.
MDR1 expression was induced in endothelial cells by chemotherapy-induced IL8 from tumor cells
We treated endothelial cells with anticancer drugs to elucidate the mechanism of MDR1/ABCB1 upregulation in TECs by chemotherapy. However, neither gemcitabine nor cisplatin induced MDR1 expression in endothelial cells (Supplementary Fig. S1C). Moreover, first-line chemotherapy did not change the ABCB1 expression of blood vessels in the noncancerous area, as described above (Fig. 1E). Then, we focused on the interaction between tumor cells and endothelial cells under anticancer drug treatment. For the systemic treatment of urothelial carcinoma, gemcitabine and cisplatin combination therapy is generally selected as first-line chemotherapy. Endothelial cells were treated with conditioned medium from UMUC3 or J82 cells after gemcitabine or cisplatin treatments. Conditioned medium collected from UMUC3 or J82 after treatment with each anticancer drug upregulated MDR1/ABCB1 expression in the endothelial cells (Fig. 2A; Supplementary Fig. S2B and S2C). In addition, conditioned medium from anticancer drug–treated UMUC3 cells induced resistance to paclitaxel in endothelial cells (Supplementary Fig. S2D). These data suggested that UMUC3 or J82 cells secreted factors after anticancer drug treatment, which induced MDR1/ABCB1 expression in the endothelial cells and caused drug resistance. Because several cytokines have been reported to upregulate MDR1 expression (26), we next examined the cytokine expression in UMUC3 with or without anticancer drug treatment using a PCR array. Both gemcitabine and cisplatin induced IL8 expression in UMUC3 cells (Fig. 2B). We confirmed secretion of IL8 in gemcitabine- or cisplatin-treated UMUC3 by ELISA (Supplementary Fig. S3A). Moreover, gemcitabine and cisplatin induced IL8 expression in other human bladder cancer cells (Supplementary Fig. S3B). In addition, gemcitabine- or cisplatin-resistant UMUC3 cells showed high IL8 expression (Supplementary Fig. S3C). These data suggested that gemcitabine and cisplatin induce constitutive IL8 expression in tumor cells. Then, we analyzed the effect of IL8 on endothelial cells. NF-κB has been reported as a transcription factor for MDR1 (27). As expected, recombinant IL8 treatment and conditioned medium from anticancer drug–treated UMUC3 cells activated NF-κB (Fig. 2C; Supplementary Fig. S3D) and induced MDR1 expression in endothelial cells (Fig. 2D). Neutralizing IL8 antibody treatment downregulated MDR1 expression in endothelial cells treated with conditioned medium from anticancer drug–treated UMUC3 cells (Fig. 2E and F). These data suggested that anticancer drugs induced MDR1 expression in endothelial cells via IL8 upregulation in tumor cells.
Anticancer drug treatment stimulates IL8 production in tumors in vivo
To analyze whether anticancer drug treatment induces IL8 expression in tumors in vivo, gemcitabine and cisplatin were administered to UMUC3-bearing mice. Similar to clinical cases (Fig. 1D), MDR1 expression was upregulated in CD31-enriched cells (UMUC3-TECs) isolated from UMUC3 tumor tissues treated with gemcitabine/cisplatin compared with that in TECs from control UMUC3 tumor tissues (Fig. 3A; Supplementary Fig. S4A and S4B). In addition, flow cytometric analysis data showed that ABCB1 expression in TECs was increased in UMUC3 tumor tissues treated with gemcitabine/cisplatin compared with control UMUC3 tumor tissues (Fig. 3B and C). These data encouraged us to use this experimental model to mimic human urothelial cancers treated using anticancer drugs.
IL8 mRNA and protein levels in UMUC3 tumor tissues were analyzed after gemcitabine/cisplatin treatment by real-time PCR and immunostaining, respectively (Fig. 3D–F). Both human IL8 mRNA levels (Fig. 3D) and the staining area of human-specific IL8 (Fig. 3E and F) increased after anticancer drug treatment. Furthermore, these treatments increased the serum human IL8 levels (Fig. 3G). These results suggested that the treatment of anticancer drugs stimulates IL8 production and secretion in tumor cells in vivo. To confirm the contribution of IL8 from tumor cells to MDR1/ABCB1 upregulation in endothelial cells, IL8-knockout UMUC3 cells were xenografted, and the tumor tissues were collected after anticancer drug treatment (Supplementary Fig. S5A and S5B). The number of ABCB1-positive TECs decreased in IL8-knockout UMUC3 tumor tissues after anticancer drug treatment (Fig. 3H and I). These data indicated that gemcitabine/cisplatin treatment caused ABCB1 upregulation in TECs via IL8 production by tumor cells in the tumor microenvironment.
ABCB1 upregulation in TECs was correlated with poor prognosis in patients with urothelial carcinoma
The association between IL8 expression in tumor tissues and the prognosis in patients with urothelial carcinoma was analyzed using the PrognoScan database (28). High IL8 expression was correlated with poor prognosis in patients with urothelial carcinoma (Fig. 4A). The above data suggests that chemotherapy induces ABCB1 upregulation in TECs via IL8 secreted from tumor cells. In our clinical cases, to analyze the association between IL8 expression in tumor cells and ABCB1 expression in TECs, IL8 expression was analyzed with immunostaining (Fig. 4B). ABCB1 expression in TECs was moderately correlated with IL8 expression in tumor cells after first-line chemotherapy (Fig. 4C; r = 0.3913). When the patients were divided into two groups according to ABCB1 expression in TECs (Supplementary Fig. S6A), no significant differences were noted in terms of IL8 expression levels in the tumor cells in cases with a low ratio of ABCB1-positive TECs (ABCB1-low group) before and after first-line chemotherapy. Conversely, in cases with a high ratio of ABCB1-positive TECs (ABCB1-high group), IL8 expression in the tumor cells significantly increased after first-line chemotherapy. Indeed, the IL8 expression level in the tumor cells increased in 16 of 24 cases (66.7%) within the ABCB1-high group (Fig. 4D), consistent with in vitro and in vivo data. Furthermore, ABCB1-high group displayed shorter overall and disease-specific survival compared with the ABCB1-low group (Fig. 4E). Among the subgroup of low ABCB1 expression in tumor cells, the ABCB1-high group also displayed shorter overall and disease-specific survival compared with the ABCB1-low group (Supplementary Fig. S6B, red line). These data suggested that chemotherapy induced ABCB1 expression in TECs, which led to drug resistance, and resulted in poor prognosis in patients with urothelial carcinoma.
Combination of ABCB1 inhibitor with second-line chemotherapy enhanced antitumor and antiangiogenic effects and reduced lung metastases
If MDR1/ABCB1 upregulation in TECs induced by first-line chemotherapy causes drug resistance in second-line chemotherapy, we speculated that a combination of ABCB1 inhibitor with an anticancer drug could improve the therapeutic effect of anticancer drugs. We previously reported that combination therapy using the ABCB1 inhibitor, verapamil, with metronomic dose paclitaxel, a substrate of ABCB1, inhibited cell proliferation in TECs with a high MDR1/ABCB1 expression (21). In addition, paclitaxel and ABCB1 inhibitor treatment enhanced antitumor and antiangiogenic effects and reduced lung metastases in the highly metastatic melanoma xenograft model (22). In this study, we used the ABCB1 inhibitor with metronomic dose paclitaxel to analyze the efficacy of this combination therapy in the bladder cancer model. We selected low-dose metronomic paclitaxel treatment, because it was reported that low-dose metronomic chemotherapy targets the tumor blood vessels rather than tumor cells. UMUC3 tumor cells were xenografted into nude mice and treated with gemcitabine/cisplatin according to clinical practice as first-line chemotherapy. The combination of verapamil with second-line paclitaxel chemotherapy was administered to mice after the first-line chemotherapy (Fig. 5A; Supplementary Fig. S7A). The tumor growth rate was lower in the paclitaxel + verapamil group than in the paclitaxel single-administration group (Fig. 5B). There was a decrease in the MVD (Supplementary Fig. S7B), whereas the numbers of apoptotic TECs had increased by combination therapy, as shown with cleaved caspase-3 staining (Fig. 5C). Other studies have reported that metronomic chemotherapy induces tumor vessel normalization (29). When pericytes were visualized using an anti-alpha-SMA antibody, there was an increase in the pericyte-covered mature blood vessels due to metronomic paclitaxel when compared with the control group (Supplementary Fig. S7C). Although the hypoxia area was not different between the three groups, blood vessel integrity, such as reduction of vessel permeability, had occurred with combination therapy (Supplementary Fig. S7D–S7F). The metastatic tumor cells in lung tissues decreased with combination therapy (Fig. 5D). These data suggested that the combination of ABCB1 inhibitor with second-line paclitaxel chemotherapy enhanced the antitumor and antiangiogenic effects and reduced lung metastases. Therefore, ABCB1 inhibitor might be useful to improve the therapeutic effect of second-line chemotherapy.
In this study, we focused on ABCB1 expression in tumor blood vessels rather than on those in tumor cells as the potential mechanism of drug resistance leading to poor prognosis of patients with cancer. After chemotherapy, tumor cells resistant to anticancer drugs persist and metastasize via tumor blood vessels. We hypothesized that TECs survive during chemotherapy and acquire drug resistance in the tumor microenvironment to facilitate tumor cell metastasis.
In this study, we demonstrated that the number of ABCB1-positive TECs increased after chemotherapy in urothelial carcinoma. High ABCB1 expression in tumor blood vessels was correlated with poor prognosis in patients with urothelial carcinoma. We revealed that the underlying mechanism might be that treatment with gemcitabine and cisplatin in UMUC3-bearing mice induced IL8 secretion in tumor cells, which subsequently elevated MDR1/ABCB1 expression in TECs. A combination therapy of metronomic dose paclitaxel with the ABCB1 inhibitor as second-line chemotherapy enhanced the antitumor effect and reduced lung metastasis in vivo (Fig. 6).
ABCB1 is a member of the ATP-binding cassette transporter superfamily. ABCB1 functions as a drug efflux pump to transport a variety of compounds. ABCB1 overexpression confers strong resistance to cytotoxic drugs, including various anticancer drugs such as vinca alkaloids and taxanes (30). ABCB1 is known to be highly expressed in cancer stem cells, which exhibit drug resistance (8, 31). High ABCB1 expression is correlated with tumor malignancy and poor prognosis (32). Furthermore, it is reported that ABCB1 expression in tumor cells is elevated by anticancer drug treatment. For example, anticancer drugs can induce epigenetic changes in the promoter region of the MDR1 gene, which leads to high ABCB1 expression in tumor cells (33). NF-κB, which is a transcription factor of MDR1, is also activated by anticancer drug treatment (34). In fact, in this study, several cases exhibited upregulation of ABCB1 in tumor cells after chemotherapy, even though there was no significant difference between before and after chemotherapy. It is important to elucidate the mechanism underlying the manner in which tumor cells acquire drug resistance in those patients, to provide personalized therapy. Conversely, there are few studies that report ABCB1 expression in the stromal cells that constitute the tumor microenvironment. In normal tissues, ABCB1 expression is observed in organs excreting metabolic products such as the bile canaliculi and kidney tubules (35). The vascular endothelial cells that constitute the blood–brain barrier also express ABCB1 to protect the brain from harmful substrates (36); however, this can inhibit drug transfer to the brain and become a cause of treatment resistance in brain metastasis (37, 38).
TECs, constituting tumor blood vessels, were thought to be genetically homogeneous, as well as normal endothelial cells. However, it is now known that TECs are heterogeneous depending on their surrounding tumor microenvironment (14, 15). Several studies reported that TECs have various abnormalities. For example, we previously reported that TECs possess stem cell characteristics, showing upregulation of stem cell markers such as MDR1 and ALDH (22), and Naito and colleagues reported that TECs contained a side population cell fraction (21, 39). We have found that TECs have acquired resistance to anticancer drugs via the upregulated expression of drug efflux transporters, similar to cancer stem cells (21, 40). Furthermore, such stem cell–like endothelial cells have high angiogenic potential (39). From these facts, TECs that have acquired stem cell–like characteristics in the tumor microenvironment are likely to survive even after chemotherapy and maintain tumor cell growth. The increase of ABCB1-positive TECs is considered to be one of the mechanisms for tumor recurrence after anticancer drug treatment and a cause of drug resistance. The prevention of anticancer drug entry into tumor tissues is also reported to be one of the mechanisms of their drug resistance (41). ABCB1 may prevent the delivery of anticancer drugs into TECs by efflux from the tumor microenvironment. However, further studies are required to analyze whether ABCB1 in TECs prevents drug delivery into tumor tissues.
IL8 is an inflammatory mediator that mobilizes neutrophils and is an important cytokine in innate immunity (42). Several cell types including macrophages, fibroblasts, and endothelial cells secrete IL8 (43, 44). IL8 is also well-known as a cytokine that promotes tumor angiogenesis (45). IL8 expression in tumor cells has been reported in various cancers. High IL8 expression in tumors is correlated with poor prognosis (46, 47). In this study, we demonstrated that anticancer drug treatment induced IL8 expression in tumor cells. IL8 expression in endothelial cells was also upregulated by gemcitabine treatment (Supplementary Fig. S8). The mobilization of tumor-associated macrophages and fibrosis in tumor tissues increased with the upregulation of IL8 after anticancer drug treatment (48, 49). Some reports elucidated the effects of IL8 on tumor stroma cells; tumor-derived IL8 signaling promotes tumor cell proliferation and invasiveness via cancer-associated, fibroblast-derived CCL2 and CXCL12 (50). In this study, we showed that IL8 upregulated ABCB1 expression in endothelial cells and induced drug resistance. From the above, the changes in the tumor microenvironment caused by anticancer drugs can induce IL8 production and promote tumor malignancy. This study indicated that the high ratio of ABCB1-positive TECs was correlated with poor prognosis. Inflammatory changes such as IL8 induction in the tumor microenvironment might be involved in the poor outcome. In addition, because IL8 induced the upregulation of ABC transporters in TECs, inflammatory changes might have occurred even in TECs. We previously reported that biglycan, one of the damage-associated molecular patterns secreted from TECs in highly metastatic tumors, promotes metastasis (16). Anticancer drugs can induce phenotypic changes in TECs and promote tumor cell metastasis via secretion of inflammatory cytokines.
For treatment of advanced urothelial carcinoma, anticancer drug treatment has been general. Various chemotherapies have been attempted for advanced urothelial carcinoma in preclinical or clinical studies; however, there are few studies that show good outcomes. In addition, particularly for gemcitabine/cisplatin-resistant urothelial carcinoma, standard chemotherapy as a second-line treatment has not been established. In addition, the therapy for targeting tumor blood vessels has not been thoroughly studied in urothelial carcinoma, and antiangiogenic agents have not been adapted, unlike in renal cell carcinoma. However, from the findings of this study, it is suggested that targeting TECs or preventing the acquisition of abnormalities in TECs can be a novel important therapeutic strategy for urothelial carcinoma. Indeed, a recent phase III study indicated that ramucirumab, a human IgG1 VEGFR2 antagonist, plus docetaxel prolonged progression-free survival compared with docetaxel plus placebo in patients with platinum-refractory advanced urothelial carcinoma (51). This study encouraged us to target tumor blood vessels for treatment of urothelial carcinoma. Our study suggests that IL8 inhibition during chemotherapy can be one of the strategies to inhibit the acquisition of abnormalities such as ABCB1 upregulation in TECs. Apart from this finding, we have reported that VEGF signaling is also involved in ABCB1 upregulation in TECs (21). Multiple factors in the tumor microenvironment may induce ABCB1 upregulation in TECs in a complicated manner. Therefore, the single inhibition of IL8 may be insufficient to suppress the acquisition of abnormalities in TECs. In addition, besides verapamil, which we used in our study, it is reported that combination therapy of anticancer drugs with antiangiogenic tyrosine kinase inhibitors enhanced the antitumor effect by suppressing ABCB1 expression in TECs (52). Taken together, TECs that acquire abnormalities during cancer therapy are another target to overcome drug resistance in cancer therapy. Targeting ABCB1 in TECs might be an attractive novel strategy to increase the effectiveness of anticancer drugs for treating urothelial carcinoma.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
H. Kikuchi: Data curation, formal analysis, investigation, visualization, methodology, writing-original draft, writing-review and editing. N. Maishi: Conceptualization, formal analysis, validation, writing-original draft, writing-review and editing. D.A. Annan: Formal analysis, validation, investigation. M.T. Alam: Investigation. R.I.H. Dawood: Investigation. M. Sato: Investigation. M. Morimoto: Investigation. R. Takeda: Investigation. K. Ishizuka: Investigation. R. Matsumoto: Resources, methodology. T. Akino: Resources. K. Tsuchiya: Resources. T. Abe: Resources. T. Osawa: Resources. N. Miyajima: Resources. S. Maruyama: Resources. T. Harabayashi: Resources. M. Azuma: Resources. K. Yamashiro: Resources. K. Ameda: Resources. A. Kashiwagi: Resources. Y. Matsuno: Resources. Y. Hida: Conceptualization, supervision, validation, project administration. N. Shinohara: Supervision, validation. K. Hida: Conceptualization, supervision, project administration, writing-review and editing.
We would like to thank Drs. S. Tanaka and M. Tsuda for providing the UMUC3, J82, T24, 5637, and tdTomato-Luc2 gene-transfected UMUC3 cells. We also thank Ms. M. Sasaki, Ms. Y. Suzuki, and Ms. T. Takahashi for their technical assistance with the experiments. This research was funded by JSPS Grants-in-Aid for Scientific Research Innovative Areas on integrated analysis and regulation of cellular diversity (to K. Hida, JP18H05092), JSPS Grants-in-Aid for Scientific Research (to N. Maishi, JP18K09715; H. Kikuchi, JP19K18549; Y. Hida, JP18H02891; and K. Hida, JP18H02996), and Grants from Japan Agency for Medical Research and Development (to N. Maishi, JP18ck0106198h0003 and K. Hida, JP19ck0106406h0002).
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.