Upregulation of miR-199a/b contributes to cisplatin resistance via Wnt/β-catenin-ABCG2 signaling pathway in ALDHA1⁺ colorectal cancer stem cells

Female athymic BALB/c nu/nu mice, 3–4 week-old, obtained from HFK Bioscience (China), were maintained at the Animal Core Facility at The First Affiliated Hospital of Xinxiang Medical University, under specific pathogen-free (SPF) condition. All studies on mice were conducted in accordance with the National Institutes of Health “Guide for the Care and Use of Laboratory Animals” and were approved by the ethical committee of The First Affiliated Hospital of Xinxiang Medical University.

Overall, 10 patients with primary CRC, who consecutively underwent chemotherapy at Department of Gastroenterology, The First Affiliated Hospital of Xinxiang Medical University, were enrolled into this study from January 2009 to June 2015. Informed consent for the additional core needle biopsy and experimental use of tumor samples was obtained from all patients, following a protocol approved by the Ethics Committee of The First Affiliated Hospital of Xinxiang Medical University. The tumor response was evaluated by the Response Evaluation Criteria in Solid Tumors (RECIST), which was defined in Supplemental Information.

Flow cytometry assay was done on single-cell suspensions obtained by enzymatic digestion of spheres which were derived from primary CRC samples and labeled with Alexa fluor@ 488 conjugated anti-ALDHA1 antibody (Abcam, USA) by using an Epics Altra flow cytometer (Beckman Coulter, USA).

Fluorescence activated cell sorting (FACS): Spheres which were derived from primary tumor samples were dissociated into single cells by enzymatic digestion. Single-cell suspensions were washed and incubated with staining solution 1% bovine serum albumin (BSA) and 2mM ethylenediaminetetraacetic acid (EDTA) with the specific antibodies at appropriate dilutions. The details are listed in the Supplemental Information.

Single-cell suspensions were obtained from CRC primary tissue samples. CRC primary tissue samples were shipped to laboratory in cold RPMI-1640 medium with penicillin/streptomycin within 1 h of removal from patients. Surgical specimens were washed with cold phosphate-buffered saline (PBS) supplemented with high doses of penicillin/streptomycin three times, chopped with a sterile blade, and incubated with 1 mg/mL collagenase II (Sigma-Aldrich, USA) for 30 min at 37°C. The details are listed in the Supplemental Information.

ALDHA1⁺ cells, ALDHA1⁻ cells, and differentiated adherent progeny of ALDHA1⁺ cells were seeded on 96-well plates at 2000 cells/well. Cells were then treated with increasing concentrations of doxorubicin from 0 to 20 µg/mL for 24 h. 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (Sigma-Aldrich) was used to determine relative cell growth every 24 h for cell growth curves, which was defined in Supplemental Information.

Total RNA was extracted using TRIzol (Invitrogen, USA) and treated with RNase-free DNase (QIAGEN, USA). Mature miRNA expression analysis was conducted using a TaqMan MicroRNA Assays (Applied Biosystems, USA). Quantitative real-time reverse transcription polymerase chain reaction (RT-PCR; qPCR) analysis was performed using a SYBR Green Reagents (Bio-Rad, USA) on the iQ5 Real-Time PCR Detection System (Bio-Rad). The relative expression levels of the miRNAs in tissue and cells were normalized to U6 and those in the serum and exosomes were normalized to the insert control according to the manufacturer’s protocol.

A pMIR-REPORT^™ luciferase reporter vector with a miR-199a/b target sequence cloned into its 3′ untranslated region (3′-UTR) (Invitrogen) was used. Luciferase activity was assayed using a Luciferase Assay Kit (Promega, USA).

For overexpression of miR-199a/b, the pcDNATM6.2-GW/EmGFP-miR vector was recombined into pLenti6/V5 using LR clonase (Invitrogen), resulting in the generation of two pLenti6/V5-miR-199a/b plasmids. Cells were transfected with viral supernatant, as described earlier.²²

To inhibit endogenous miR-199a/b, cells were transfected with the antagomir-199a/b lentivector (System. BioSciences (SBI), USA) or negative control (anti-nc lentivector).²³ Levels of antagomir-199a/b lentivirus expressing anti-miR-199a/b were determined using the synthetic miR-199a/b binding site–containing luciferase reporter gene. Western blotting is detailed in Supplemental Information.

Female athymic BALB/c nu/nu mice at 3–4 weeks of age were injected subcutaneously into the right flank with 1 × 10⁴ of ALDHA1⁺ CCSCs and 1 × 10⁴ of ALDHA1⁻ cells. The xenografts were monitored until the tumor volumes reached ~100 mm³; each group was randomized into two subgroups that were either left untreated or received intraperitoneal injections of doxorubicin (4 mg/kg) every 5 days (three cycles). The tumor volume was measured with a caliper using the following formula: V (mm³) = 0.52 × length (mm) × width² (mm²).

Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining was performed by the DeadEnd^™ Fluorometric TUNEL system according to the manufacturer’s instructions (Promega) in xenograph tumor tissues derived from lenti-si-ABCG2-CCSCs, lenti-si-NC-CCSCs group, and untransduced ALDHA1⁺ CCSCs following treatment with or without three cycles of cisplatin. Cells were then observed under a fluorescence microscope (Olympus 200 Optical Co. Ltd, Germany), and a nucleus with bright green fluorescence staining was recorded as a TUNEL-positive event.

SPSS13.0 software was used. Each experiment was performed at least three times. The data were expressed as mean ± standard deviation (SD) and one-way analysis of variance (ANOVA). An unpaired Student’s t test was used to determine the significant differences of all the results. Significances are, ***p < 0.001; **p < 0.01; and *p < 0.05.

To identify whether ALDHA1 is the surface marker to sort CCSCs, we analyzed 10 fresh tumor samples derived from a series of CRC patients. Flow cytometry demonstrated the presence of a variable fraction of ALDHA1, ranging from 0.1% to 9.7%, in 9 out of 10 CRC specimens. CSCs are believed to be able to form spheres in serum-free cultivation. Therefore, we cultured two cases of primary CRC tissue samples (Table S1) to induce sphere formation in serum-free cultivation. After culturing for 2–4 weeks, although a large number of cells died, some tumor cells grew to form spheres (data were not shown). The tumor spheres were at least passaged 10 times, indicating the self-renewal of these tumor sphere cells. Because the definition of CSCs relies mainly on stem cell–like properties, we further fractionated ALDHA1⁺ and ALDHA1⁻ cells derived from clinical case CC051 and CC126 by FACS to corroborate the capacity of self-renewal, differentiation, and tumorigenesis in ALDHA1⁺ spheres. ALDHA1⁺ cells could form compact self-renewing spheres whereas ALDHA1⁻ cells could not (Figure 1(a)). Significantly, with increasing concentrations of cisplatin from 0.5 to 20 µg/mL, cell viability in both ALDHA1⁻ cells and differentiated adherent progeny of ALDHA1⁺ cells decreased at each concentration compared with that in ALDHA1⁺ cells (Figure 1(b)). In addition, we found that the expression of ALDHA1 decreased gradually when ALDHA1⁺ spheres were cultured in a serum-containing medium (Figure S1). To evaluate the tumorigenic potential of ALDHA1⁺ cells, the same numbers of ALDHA1⁺ and ALDHA1⁻ cells (10⁴ cells) were subcutaneously injected into the flank of nude mice. We found that tumor formation of ALDHA1⁺ cells was faster and resulted in increased tumor take compared with that observed in ALDHA1⁻ cells (Figure 1(c)). To investigate whether CRCs originating from ALDHA1⁺ cells possess higher long-term tumorigenic potential compared with the rarely observed tumors originating from the ALDHA1⁻ cells, we performed serial transplantation assays in nude mice of cells isolated from CC051 or CC126 tumor xenografts originally derived from ALDHA1⁺ or ALDHA1⁻ cells injection. Cells derived from ALDHA1⁺ tumors were able to generate tumors in primary, secondary, and tertiary transplantation, whereas cells from ALDHA1⁻ tumors lost tumorigenic potential during serial transplantations (Figure 1(c)). Therefore, these data indicated that ALDHA1⁺ cells from CRC patient–derived tumor spheres bear stem cell–like features.

To obtain the expression profile of miR-199a/b in ALDHA1⁺ CCSCs, we compared miR-199a/b expression in ALDHA1⁺ CCSCs with that in differentiated adherent progeny of ALDHA1⁺ CCSCs and ALDHA1⁻ cells. We found that the expression of miR-199a/b in ALDHA1⁺ CCSCs was significantly increased at messenger RNA (mRNA) level compared with that in differentiated adherent progeny of ALDHA1⁺ CCSCs and ALDHA1⁻ cells (Figure 2(a)). To further explore the targeting function of miR-199a/b, we evaluated the percentage of luciferase suppression by transduction of ALDHA1⁺ CCSCs with miR-199a/b antagomir and negative control (anti-nc). We found that luciferase activity of ALDHA1⁺ CCSCs (un) and ALDHA1⁺ CCSCs transduced with anti-nc was suppressed (Figure 2(b)). The suppression of luciferase activity in ALDHA1⁺ CCSCs transduced with antagomir-199a/b was significantly restored (Figure 2(b)). Thus, these findings indicate that the expression as well as the target function of miR-199a/b is enhanced in ALDHA1⁺ CCSCs.

To investigate whether the upregulation of miR-199a/b expression in ALDHA1⁺ CCSCs contributes to cisplatin resistance, in vivo assay was taken. We found that xenograft tumors in antagomir-199a/b-CCSCs group were slightly smaller than that in anti-nc-CCSCs group and untransduced group (Figure 3(a)), although no significant difference was observed. Upon intratumor injection with cisplatin, the size of xenograft tumors in antagomir antagomir-199a/b-CCSCs group was significantly reduced in comparison with that in anti-nc-CCSCs group and untransduced group (Figure 3(a)). We, next, detected cisplatin-induced apoptosis by using Annexin V/propidium iodide (PI) staining in antagomir-199a/b-CCSCs, anti-nc-CCSCs, and untransduced CCSCs. In the presence of cisplatin, antagomir-199a/b-CCSCs showed increased percentage of Annexin V⁺PI⁺/Annexin V⁺PI⁻cells about 5- to 14- fold as compared with anti-nc-CCSCs and untransduced CCSCs (Figure 3(b)). However, in the absence of cisplatin, antagomir-199a/b-CCSCs showed increased percentage of apoptotic and necrotic cells about only 1- to 4- fold as compared with anti-nc-CCSCs and untransduced CCSCs (Figure 3(b)). Our data demonstrate that declined miR-199a/b expression by antagomir sensitizes ALDHA1⁺ CCSCs to cisplatin via apoptosis.

To explore the molecular mechanisms of miR-199a/b action on resistance to cisplatin in ALDHA1⁺ CCSCs, we integrated mRNA expression profiling findings with prediction algorithms of TargetScan, a miRBase algorithm. We found that the seed sequence of miR-199a/b had a perfect match with the 3′-UTR of the Gsk3β mRNA (Figure 4(a)). To validate whether Gsk3β is a direct target of miR-199a/b, wild sequence of Gsk3β 3′-UTR and its corresponding mutant counterparts were cloned into pMIR-REPORT luciferase vector. Reporter activities in wild type 3′-UTR (WT) were promoted by miR-199a/b antagomir but not in mutant 3′-UTR (MT) (Figure 4(b)). These results indicated that miR-199a/b downregulated Gsk3β expression by directly targeting its 3′-UTR. To further examine whether miR-199a/b affects Gsk3β expression, we transduced ALDHA1⁺ CCSCs with lenti-miR-199a/b or antagomir-199a/b. Western blot analysis revealed that miR-199a/b overexpression triggered a repressive effect on endogenous Gsk3β but stimulated β-catenin protein expression (Figure 4(c)). Conversely, in miR-199a/b depleted cells (antagomir-199a/b-CCSCs), expression of Gsk3β was increased whereas β-catenin expression was suppressive (Figure 4(c)). Therefore, our data indicated that miR-199a/b regulated Gsk3β expression by directly targeting their 3′-UTRs. The increase of miR-199a/b might result in the reduction of Gsk3β protein by relieving translational repression at the 3′-UTRs of Gsk3β.

To establish clinicopathological associations, Gsk3β expression patterns in CRC and adjacent normal tissues were examined with immunohistochemistry (IHC). The mean IHC scores of tumor samples versus adjacent normal tissues were 38.3 and 104.6, with significant differences (p < 0.01; Figure 4(d)). Moreover, 80% of CRC tumor samples showed higher levels of β-catenin than their adjacent normal counterparts (Figure 4(e)).

The expression of adenosine triphosphate–binding cassette (ABC) transporters, which are putative to mediate the inherent drug efflux capability, including ABCG1, ABCA1, ABCG2, ABCA2, ABCC1, and ABCC2, was detected in antagomir-199a/b-CCSCs compared with that in anti-nc-CCSCs and untransduced CCSCs. We found that the expression of ABCG1, ABCA1, and ABCG2 was decreased in antagomir-199a/b-CCSCs compared with that in anti-nc-CCSCs and untransduced CCSCs by using qPCR (Figure 5(a)). To further characterize the association between Wnt/β-catenin pathway and ABC transporters, we used XAV-939 (the inhibitor of Wnt/β-catenin pathway) to block the activity of Wnt/β-catenin pathway. We found that the expression of β-catenin and ABCG2 was significantly decreased in the presence of XAV-939 (Figure 5(b)), which suggested that Wnt/β-catenin pathway was one of the upstream mechanisms to regulate the expression of ABCG2 in ALDHA1⁺ CCSCs. However, the expression of ABCG1 and ABCA1 was almost unchanged at protein level with and without the presence of XAV-939 (Figure 5(b)).

To further investigate the link between the cisplatin resistance and the expression of miR-199a/b and ABCG2, a lentiviral-based approach was used to knockdown ABCG2 with lenti-si-ABCG2. Without cisplatin, the cell viability of ALDHA1⁺ CCSCs was not significantly changed upon transduction with antagomir-199a/b, anti-nc, or lenti-si-ABCG2 (Figure 5(c)), suggesting that the upregulation of miR-199a/b and ABCG2 were not required for ALDHA1⁺ CCSCs survival. With increasing concentration of cisplatin, the percentage of viable cells in antagomir-199a/b-CCSCs and lenti-si-ABCG2-CCSCs decreased more rapidly than that in anti-nc-CCSCs and transduced CCSCs (Figure 5(c)). At each concentration, cisplatin induced the highest growth inhibition in ALDHA1⁺ CCSCs with the transduction of lenti-si-ABCG2 (Figure 5(c)).

Moreover, we built xenograft tumor model by using ALDHA1⁺ CCSCs, lenti-si-nc-CCSCs, or lenti-si-ABCG2-CCSCs. We found that xenograft tumors in lenti-si-ABCG2-CCSCs group were slightly smaller (Figure 6(a)), although no significant difference was observed. Upon intratumor injection with cisplatin, the size of xenograft tumors in lenti-si-ABCG2-CCSCs group was significantly reduced in comparison with that in lenti-si-nc-CCSCs and untransduced ALDHA1⁺ CCSCs (Figure 6(a)). In addition, to detect whether cisplatin-mediated cell death is generated by apoptosis, TUNEL apoptosis index was examined in every tumor group. Intratumor injection with cisplatin in xenograft tumors derived from ALDHA1⁺ CCSCs resulted in an increased TUNEL index compared with intratumor injection with glucose (control) (Figure 6(b)). Significantly, with the treatment of cisplatin, xenograft tumors in lenti-si-ABCG2-CCSCs group showed an obviously higher TUNEL index than those in lenti-si-NC-CCSCs group and untransduced ALDHA1⁺ CCSCs (Figure 6(b)). Taken together, our data suggested that upregulation of miR-199a/b in ALDHA1⁺ CCSCs contributes to cisplatin resistance via Wnt/β-catenin-ABCG2 signaling.

To test the hypothesis, we characterized the associations among miR-199a/b, Wnt/β-catenin pathway, and ABCG2. We found that the expression of β-catenin and ABCG2 were reduced, while Gsk3β was upregulated in ALDHA1⁺ CCSCs transduced with antagomir-199a/b (Figure 7(a)). With the presence of XAV-939, the expression of β-catenin and ABCG2 in ALDHA1⁺ CCSCs was downregulated (Figure 7(a)). The expression of ABCG2 was reduced in ALDHA1⁺ CCSCs transduced with lenti-si-ABCG2, whereas the expression of Gsk3β and β-catenin was unchanged compared with those in untreated ALDHA1⁺ CCSCs (Figure 7(a)). Significantly, with the presence of XAV-939, the expression of ABCG2 and β-catenin in ALDHA1⁺ CCSCs transduced with lenti-miR-199a/b restored compared with those in untransduced ALDHA1⁺ CCSCs with the presence of XAV-939 (Figure 7(a)).

To investigate whether miR-199a/b expression is related to chemoresistance in CRC patient cohort, 10 CRC patients who exhibited CR (chemosensitive), PR (chemosensitive), or PD (chemoresistant) according to RECIST criteria were enrolled. We found that miR-199a/b expression in PD tissues was the highest among the different chemoresponsive CRC tissues (Figure 7(b)). It implies a link between miR-199a/b expression and tumor response in CRC patients. Survival curve showed that patients with low miR-199a/b expression survived obviously longer than patients with low miR-199a/b expression (Figure 7(c)).