Ovarian cancer patients frequently develop resistance to chemotherapy regiments using Taxol and carboplatin. One of the resistance factors that protects cancer cells from Taxol-based therapy is multidrug resistance 1 (MDR1). micro(mi)RNAs are small noncoding RNAs that negatively regulate protein expression. Members of the let-7 family of miRNAs are downregulated in many human cancers, and low let-7 expression has been correlated with resistance to microtubule targeting drugs (Taxanes), although little is known that would explain this activity. We now provide evidence that, although let-7 is not a universal sensitizer of cancer cells to Taxanes, it affects acquired resistance of cells to this class of drugs by targeting IMP-1, resulting in destabilization of the mRNA of MDR1. Introducing let-7g into ADR-RES cells expressing both IMP-1 and MDR1 reduced expression of both proteins rendering the cells more sensitive to treatment with either Taxol or vinblastine without affecting the sensitivity of the cells to carboplatin, a non-MDR1 substrate. This effect could be reversed by reintroducing IMP-1 into let-7g high/MDR1 low cells causing MDR1 to again become stabilized. Consistently, many relapsed ovarian cancer patients tested before and after chemotherapy were found to downregulate let-7 and to co-upregulate IMP-1 and MDR1, and the increase in the expression levels of both proteins after chemotherapy negatively correlated with disease-free time before recurrence. Our data point at IMP-1 and MDR1 as indicators for response to therapy, and at IMP-1 as a novel therapeutic target for overcoming multidrug resistance of ovarian cancer.
Chemotherapy is one of the most frequently used treatment modalities for human cancer. In ovarian cancer, front line treatment consists of surgical debulking followed by a combination therapy of paclitaxel and carboplatin yielding initial response rates of up to 80%.1 Unfortunately, the majority of these patients relapse within 18 months and recurrent disease is frequently much more resistant to chemotherapy treatment. Tumor cells acquire resistance to chemotherapeutic drugs through various mechanisms including upregulation of members of the adenosine triphosphate binding cassette transporters (ABC transporter family) that act by pumping drugs across the cell membrane to the extracellular space. A highly studied member of this family is multidrug resistance-1 (MDR1; ABCB1)2 which encodes for membrane transporter P-glycoprotein (P-gp). Substrates for MDR1 include a wide array of toxins and commonly used chemotherapeutic agents, including Taxanes3 and anthracyclines.
Micro(mi)RNAs are small noncoding RNAs that negatively regulate a large number of genes via either mRNA degradation or translational silencing.4 Several miRNAs are involved in acquired therapy resistance of cancer cells, with one class targeting survival pathways or pathways that regulate apoptosis sensitivity (reviewed in Ref.5), whereas a second class modulates therapy sensitivity by targeting genes that are important for drug transport or metabolism. Examples include miR-200c (targets TUBB3),6 miR-24 (targets DHFR),7 miR-451 and miR-27a (targets MDR1)8 and miR-328 and miR-519c (target ABCG2).9, 10
Let-7 comprises a family of 13 related miRNAs located on nine different chromosomes. Let-7 was identified as a tumor suppressor for many cancers and it is downregulated in almost all investigated human cancers (reviewed in Ref.5). Let-7 was also found to regulate stemness of breast cancer tumor-initiating cells.11 In addition, a number of studies demonstrated a correlation between loss of let-7 and resistance to either chemotherapeutic drugs or radiation.12–15 Although this activity has been documented, the let-7 targets responsible for this effect are largely unknown.
We recently identified a set of 12 let-7-regulated oncofetal genes (LOGs), eight of which were known to be upregulated during progression of various human cancers.16 The second most highly ranked LOG after HMGA2 was IMP-1 (IGF2BP1/CRD-BP), and we experimentally validated IMP-1 as a let-7 target.16 IMP-1 is an RNA binding protein that acts by stabilizing the mRNA of a number of target genes. Targets include the insulin-like growth factor II leader 3 mRNA species, c-myc, IGF2, tau, FMR1, semaphorin, and βTrCP1 mRNAs and H19 RNA.17–25 In addition, IMP-1 was shown to protect the mRNA of MDR1 from endonucleolytic attack in an in vitro RNA stability assay.26
Here, we demonstrate that let-7 does not function as a general regulator of drug sensitivity in cancer cell lines. However, we demonstrate that let-7 expression does function to alter drug sensitivity in situations of acquired resistance where the let-7 target IMP-1 is coexpressed with MDR1. Specifically, we show that let-7g selectively affects the sensitivity of a drug resistant ovarian cancer cell line to Taxanes by targeting IMP-1, which in turn causes destabilization of MDR1 at the mRNA and protein level, and this increases sensitivity of the multidrug resistant ovarian cancer cell to Taxanes. The association between IMP-1 and MDR1 was confirmed in ovarian cancer patients, where upregulation of both IMP-1 and MDR1 in the tumor after administration of chemotherapy correlated with an adverse prognosis, consistent with a function of IMP-1 in stabilizing MDR1 in human cancer.
Material and methods
NCI/ADR-RES, OVCAR-8, T47D, IGROV1 and LOX-IMVI cell lines were obtained from the National Cancer Institute as part of the NCI60 panel of cell lines and were cultured in RPMI-1640 supplemented with 10% fetal bovine serum (FBS) and penicillin/streptomycin. HEK-293 and Phoenix cells were obtained from the American Tissue Culture Centre and cultured in Dulbecco's Modified Eagle Medium supplemented with 10% FBS and penicillin/streptomycin. SKOV3ip1 cells expressing Lin28 or Lin28B were maintained in media supplemented with 3 μg/ml puromycin.
Western blots and antibodies
Western blot analyses were performed as previously described.16 Antibodies used were as follows: IMP-1 (sc-21026), MDR1 (sc-55510) from Santa Cruz (Santa Cruz, CA), HMGA2 (59170 AP) from BioCheck (Foster City, CA), actin (A5441) from sigma, Lin28 from Cell Signaling Technology, Danvers, MA, Lin28B from Abgent, San Diego, CA. All secondary horse radish peroxidase-conjugated antibodies were purchased from Southern Biotech (Birmingham, AL).
Plasmids, cloning and retroviral transductions
HeLa cDNA was derived using the SuperScript III First Strand DNA synthesis for RT-PCR kit (Invitrogen Life Technologies, Carlsbad, CA) following the manufacturer's protocol. The human IMP-1 ORF was amplified from HeLa cDNA with primers 5′-CCGCGGCCCCAAGCTTATGAAGCTTTAC-3′ and 5′-CGATGTTAACAAGCTTTCACTTCCTCCGTGC-3′ using Platinum Taq Pfx (Invitrogen Life Technologies). The IMP-1 ORF fragment was cloned into the pLHCX vector (Clontech, Mountain View, CA) using the In-Fusion 2.0 cloning kit (Clontech). Let-7g in the pLNCX vector was a kind gift of Dr. Sven Diederichs. Lin28 and Lin28B in the pMSCV vector were a kind gift of Dr. Scott Hammond.
Phoenix cells were plated in 10-cm dishes and the following day, 15 μg of plasmid along with 4 μg pVSV-G were transfected using 30 μl Lipofectamine 2000 (Invitrogen Life Technologies) per transfection following the manufacturer's protocol. Twenty-four hours later, media was collected, filtered using a 0.45-μm filter and transferred to target cells along with 8 μg/ml polybrene (Sigma, St. Louis, MO). Dishes were centrifuged at 2,700 rpm for 1 hr at room temperature. Media was concurrently replaced with the Phoenix cells and the same procedure was repeated 24 hr later. Forty-eight hours following the final infection, let-7g expression in ADR-RES, HEK-293 and OVCAR8 was selected using 500 μg/ml G418. IMP-1 expression in ADR-RES was selected using 300 μg/ml hygromycin.
For si-IMP-1 and si-MDR1, 105 ADR-RES cells were seeded in 12-well plates and transfected 18 hr later with 50 nM si-IMP-1,27 25 nM si-MDR1 smart pool (Dharmacon, Lafayette, CO) or si-Scrambled control #1 (Dharmacon) using 6 μl INTERFERin™ transfection reagent per well (Polyplus, New York, NY). Cells were harvested in RIPA lysis buffer after 72 hr. In certain cases, the cells were retransfected at 72 hr and harvested 6 days post-initial transfection.
Lentiviral particles used for infecting LOX-IMVI cells were purchased from Sigma: PLKO.1 (SHC002V), shLin28B1 (NM_001004317.1-2895s1c1; TRCN0000122191; CCGGGCAGGCATAATAAGCAAGTTACTCGAGTAACTTGCTTATTATGCCTGCTTTTTTG), shLin28B2 (NM_001004317.1-482s1c1; TRCN0000122599; CCGGGCCTTGAGTCAATACGGGTAACTCGAGTTACCCGTATTGACTCAAGGCTTTTTTG), shLin28 in T47D cells (NM_024674.3-529s1c1; TRCN0000021803; CCGGTGCTACAACTGTGGAGGTCTACTCGAGTAGACCTCCACAGTTGTAGCATTTTT) shLin28 in IGROV1 cells (NM_024674.3-448s1c1; TRCN0000021800; CCGGCCTGGTGGAGTATTCTGTATTCTCGAGAATACAGAATACTCCACCAGGTTTTT). Cells were infected in following the manufacturer's instructions.
Quantitative real-time polymerase chain reaction
Total RNA was extracted using the miRNeasy Mini kit (Qiagen, Valencia, CA) following the manufacturer's protocol. Real-time polymerase chain reaction (PCR) of miRNAs was performed as previously described using specific Taqman primers for let-7a, let-7d, let-7g, let-7i, miR-200c or U6 purchased from Applied Biosystems (Carlsbad, CA). Gene expression analysis was performed as previously described16 using primers for IMP-1, MDR1 and c-MYC (ABI).
Cell cycle analysis
Paclitaxel (Taxol), vinblastine and carboplatin were purchased from Sigma. Paclitaxel and vinblastine were dissolved in DMSO and carboplatin was dissolved in PBS. For cell cycle analysis, ADR-RES cells were plated in 12-well plates at 105/well. The following day, cells were treated with paclitaxel, vinblastine, carboplatin or DMSO vehicle. Twenty-four hours later, cells were trypsinized and resuspended in 0.1% sodium citrate, pH 7.4, 0.05% Triton X-100, 50 μg/ml propidium iodide. After an incubation of 1 hr at 4°C, the cell cycle profile was read on a FACSCanto flow cytometer using the PE channel to detect propidium iodide (BD Biosciences, San Jose, CA).
For the sulforhodamine B (SRB) cytotoxicity assays, cells were plated in 96-well plates at a density such that they reached near confluency 96 hr post-plating. ADR-RES cells were plated in triplicate at 10,000/well, OVCAR-8 and SKOV3ip1 at 6,000/well and HEK-293, T47D, LOX-IMVI and IGROV1 at 8,000/well. Twenty-four hours later, cells were treated with a serial dilution of either paclitaxel or carboplatin (Sigma). Seventy-two hours later, cells were washed once with PBS and fixed using the CytoScan SRB Cell Cytotoxicity Assay (G Biosciences, Maryland Heights, MO) following the manufacturer's protocol. OD540 was determined using an iMark Microplate Reader (Qiagen). Percent cytotoxicity was determined using the following equation: % cytotoxicity = [100 × (cell control - experimental)]/cell control.
MDR1 activity assays
ADR-RES cells (106) per test were suspended in 1 ml cold loading buffer [RPMI-1640, 10% BSA, 1 μg/ml DIOC2(3) (Millipore)] for 15 min. Cells were centrifuged and washed once in cold efflux buffer (RPMI-1640 and 10% BSA) before being resuspended to 106/ml in 1 ml of cold efflux buffer, warm efflux buffer (cold efflux buffer warmed to 37°C) containing DMSO or warm efflux buffer containing 25 μM vinblastine. After 15 min, 5 ml of cold efflux buffer was added to each tube and the tubes were placed on ice. The cells were centrifuged and washed twice in cold efflux buffer. Finally, the washed cells were resuspended in 500 μl of cold PBS and read on a FACSCanto (BD Biosciences) flow cytometer. DIOC2(3) retention was determined by measuring fluorescence on the Fluorescein isothiocyanate (FITC) channel.
FITC-labeled UIC2 antibody (ab66250) was purchased from Abcam (Cambridge, MA). ADR-RES cells were suspended in 100 μl of warm UIC2 binding buffer (PBS, 1% BSA) at a concentration of 106/100 μl. Either DMSO or 25 μM vinblastine was added to each tube. Tubes were incubated in a water bath at 37°C for 10 min. FITC-labeled UIC2 (10 μl) or isotype control was added to each sample and tubes were incubated for a further 30 min. Then, 1 ml of cold UIC2 binding buffer was added to each tube and cells were centrifuged and washed twice in cold UIC2 binding buffer. Cells were resuspended in 500 μl of cold UIC2 binding buffer and the FITC fluorescence was read using a FACSCanto flow cytometer (BD Biosciences).
Patients and tissues
The University of Chicago ovarian cancer database28, 29 with 500 patients was searched for patients who had a recurrence after first line chemotherapy treatment with carboplatin and paclitaxel and for whom tumor material was available for the primary tumor and the cancer after recurrence. Tissue blocks from these 20 patients with epithelial ovarian cancer who had undergone two surgeries by a gynecologic oncologist at the University of Chicago were included in the study after Institutional Review Board approval was obtained. Blocks from the first and second surgery were pulled from the archives, cut and stained with H&E. All pathology was verified by two gynecologic pathologists (A.M. and K.G.) at the University of Chicago. Follow-up information was gathered by reviewing hospital and outpatient clinic charts, data from the Illinois Cancer Registry, the U.S. Social Security Index, and by contacting physicians involved in the patient's care. Follow-up information was updated every 6 months through December 2009. The median follow-up time was 46 months.
Tumors were formalin-fixed, paraffin-embedded and sectioned. Antigen retrieval was performed by heating the tissue in 0.01 M sodium citrate, pH 5.0 for 20 min. All slides were blocked in avidin and biotin blocking solutions. Slides were stained with H&E, or incubated with primary antibodies against MDR1 (1:20; Sigma, St. Louis, MO) or IMP-1 (1:100; described in Ref.30). After washing with TBS, slides were incubated with the species-appropriate biotinylated secondary antibody (1:500) and detected with peroxidase-linked avidin using the Vectastain ABC kit (Vector Laboratories, Burlingame, CA) for 30 min. Slides were rinsed in TBS and stained with 3,3-diaminobenzidine chromogen.
Pearson correlation coefficients and related p-values were calculated between expression changes of MDR1 and IMP-1. An outlier patient was identified using Cook's distance analysis.31 The expression of MDR1 or IMP-1 was grouped by its mean value (high expression > mean value, low expression < mean value). Survival analysis (Kaplan–Meier) was performed to determine the correlation between the disease-free days (days between start of chemotherapy and recurrence), and the expression of MDR1 and IMP-1. The p-values of individual markers were calculated using parametric model with Weibull error distribution. To determine the effect of both markers (IMP-1 and MDR1) disease-free survival a log-likelihood test following chi square distribution was performed.
miRNA in situ hybridization
miRNA in situ hybridization on Formalin-fixed, Paraffin-embedded (FFPE) fixed samples was performed using the miRCURY LNA™ microRNA ISH Optimization Kit (FFPE; Exiqon (Woburn, MA) #90005) and the instruction manual v1.3. (Exiqon) based on a published protocol.32 The Roche (Basel, Switzerland) DIG Wash and Block Buffer Set (Roche, Cat. # 11 585 762 001) was used. The procedure was followed according to the instruction manual with minor alterations. Slides with matched samples from patients before and after chemotherapy were processed in parallel. For the in situ hybridization, the LNA™ scrambled microRNA probe, double-DIG labeled (5′-GTGTAACACGTCTATACGCCCA-3′; Exiqon, Cat. # 99004-15) was used as negative control. The LNA™ hsa-let-7d microRNA probe, double-DIG labeled (5′-AACTATGCAACCTACTACCTCT-3′; Exiqon, Cat. # 38459-15) was used to stain for let-7d in the FFPE fixed tissue. The let-7d miRCURY LNA™ microRNA detection probe was used at a final concentration of 60 nM. Twenty-five microliters of probe mixture was used for an 18 × 18 mm2 hybridization area. When the hybridization area was larger the volume was increased accordingly. The alkaline phosphatase reaction took place at 30°C for 120 min, or at 4°C overnight to allow for more time flexibility to develop the dark-blue Nitroblue tetrazolium-formazan precipitate. Before mounting, slides were counter stained with Nuclear Fast Red™ (Vector Laboratories, Burlingname, CA, Cat. # H-3403) at room temperature at a reduced time for 20 sec. Images were taken on an Olympus BX41 microscope equipped with a digital camera and images were saved in 16 bit tiff format.
Let-7 does not universally regulate drug sensitivity of cancer cells
To study drug sensitivity conferred by let-7, we generated cell lines with different let-7 expression levels. The cell lines used for these studies were not selected for drug resistance, did not display a multidrug resistant phenotype and did not express MDR1. In each case, we demonstrated the gain or loss of at least one specific let-7 target gene at the protein level to demonstrate a functional change in let-7 activity as a result of the manipulation. It is important to note that let-7 low cell lines do not gain expression of all let-7 target genes on loss of let-7 expression, and the expression of let-7 target genes is dependent on many additional factors including the transcriptional activity of the target gene promoter. The let-7 low cell line OVCAR-8, for instance, expresses IMP-1 but not HMGA2, whereas LOX-IMVI cells express only HMGA2.
To initially test our hypothesis, we stably expressed let-7g in the let-7 low cell lines OVCAR-8 and HEK-293. Stable expression of let-7g resulted in downregulation of the let-7 targets HMGA2 and/or IMP-1. However, in neither case did this result in a change in the sensitivity to Taxol (Fig. 1a) or carboplatin (data not shown), suggesting that these cell lines do not express let-7 targets capable of modulating drug sensitivity. As this result was obtained by modulating only let-7g, we next tested whether altering multiple let-7 family members in the same cell line had an effect on drug sensitivity. Lin28 and Lin28B are master regulators of let-7 expression and suppress expression of all let-7 family members without affecting expression of other miRNAs.33, 34 To modulate expression of multiple let-7 family members, we knocked down Lin28 or Lin28B in let-7 low cell lines or stably overexpressed Lin28 and Lin28B in the let-7 high cell line SKOV3ip1 (Figs. 1b and 1c and Supporting Information Fig. S1). ShRNA lentiviruses reduced the expression of Lin28B in the let-7 low expressing cell lines LOX-IMVI (Fig. 1b) and T47D, and reduced expression of Lin28 in the let-7 low expressing ovarian cancer cell line IGROV1 (Supporting Information Figs. S1a and S1b). The reduction of Lin28/Lin28B caused an increased expression of let-7 species in each cell line, which in LOX-IMVI cells resulted in a reduction in the expression of the let-7 target HMGA2 (Fig. 1b). Similarly, in SKOV3ip1 cells, overexpression of either Lin28 or Lin28B caused a reduction in let-7 levels paralleled by an increase in HMGA2 expression (Fig. 1c) confirming that the changes in Lin28 and Lin28B expression had functional consequences. Similar to the results of overexpression of let-7g in OVCAR-8 and HEK-293, increasing the expression of multiple let-7 family members by reducing Lin28/Lin28B levels did not sensitize cells to Taxol-induced cell death, and reducing let-7 levels by expressing Lin28 or Lin28B did not increase resistance to Taxol. In summary, neither let-7 nor the master regulators of let-7 expression Lin28/Lin28B appear to be universal regulators of drug sensitivity, suggesting that loss of let-7 confers drug resistance through specific mechanisms that are not general characteristics of all let-7 low cancer cell lines.
MDR1 is involved in the Taxol resistance of let-7 low NCI/ADR-RES cells
We recently validated IMP-1 as a target of let-7 with cancer relevance.16 As IMP-1 stabilizes MDR1 mRNA in vitro,26 and because Taxol is a substrate for MDR1, we hypothesized that let-7 regulates sensitivity to Taxanes through targeting of IMP-1, and that this mechanism would only be active in cells coexpressing IMP-1 and MDR1. Among the NCI60 cell collection, we found a derivative of the ovarian cancer cell line OVCAR-8 (an MDR1 negative cell line), designated NCI/ADR-RES (ADR-RES) (http://dtp.nci.nih.gov/docs/misc/common_files/NCI-ADRres.html), that had been selected for drug resistance by culturing in increasing concentrations of adriamycin (an MDR1 substrate). This cell line displays a well-defined multidrug resistant phenotype and has been routinely used as a tool to investigate mechanisms of the multidrug resistant phenotype. ADR-RES was the only cell line of the NCI60 found to express significant amounts of MDR1 protein, and had the highest expression of IMP-1 among 59 of the 60 NCI60 cell lines while expressing very low amounts of let-7 (data not shown).
Treatment of parental OVCAR-8 and ADR-RES cells with Taxol confirmed a much higher resistance of ADR-RES cells to Taxol-induced cell death when compared with OVCAR-8 cells (Fig. 2a). This difference was not found for carboplatin, which is not a substrate for MDR1 (Fig. 2b), suggesting that MDR1 is mainly responsible for the resistance of ADR-RES cells to Taxol. An siRNA-mediated knockdown of MDR1 confirmed that this transporter contributes to the high resistance of ADR-RES cells to Taxol (Fig. 2c).
Let-7g expression in ADR-RES causes downregulation of both IMP-1 and MDR1, selectively increasing the sensitivity to Taxol
On the basis of our data, we hypothesized that IMP-1 stabilized expression of MDR1 mRNA and contributed to the increased resistance to Taxol observed in the ADR-RES cell line. To test this hypothesis, let-7g was stably expressed in ADR-RES cells. As shown in Figures 3a and 3b, a clear reduction of IMP-1 mRNA and protein was detected in these cells coinciding with inhibition of MDR1 expression (but not of c-MYC) at both the mRNA and protein levels (Figs. 3a and 3b). To determine whether the reduction of MDR1 was due to the reduction in IMP-1 and was not an IMP-1-independent effect of let-7g, we specifically knocked down IMP-1 in these cells. This knockdown also caused a reduction in MDR1 (Fig. 3b). The let-7g-induced reduction of MDR1 resulted in a reduced ability of the cells to retain the dyes calcein AM and DiOC2(3), two established substrates for MDR135, 36 (Fig. 3c and data not shown). Let-7g overexpression also led to reduced surface expression of MDR1 as well as a reduced vinblastine-mediated activation of the MDR1 transporter,3 as evidenced using the conformation specific antibody UIC2 which binds to surface MDR1 more efficiently once MDR1 has been activated by the presence of a substrate37 (Fig. 3d).
As a result of the reduction of MDR1, cells were more sensitive to the two MDR1 substrates Taxol and vinblastine but not to carboplatin (Fig. 4a). Consistently, the ability of both Taxol and vinblastine to induce G2/M arrest was significantly increased in the let-7g expressing cells (Fig. 4b). The effect of carboplatin on the cell cycle was unchanged. This result is in stark contrast to our results with the parental OVCAR-8 cell line, where overexpression of let-7g and significant downregulation of IMP-1 produced no significant effect on Taxol sensitivity (Fig. 1a), suggesting that IMP-1 does not function to modulate drug sensitivity in the absence of MDR1 expression.
Let-7g regulates sensitivity to Taxol via IMP-1-mediated regulation of MDR1
To directly test whether let-7 regulates the expression of MDR1 through targeting of IMP-1, we reconstituted let-7g expressing ADR-RES cells with exogenous IMP-1 that was resistant to let-7 regulation by virtue of lacking the 3′-UTR (Fig. 5a). These reconstituted cells showed restored expression of MDR1 (Fig. 5a) and maintained high let-7g expression (Fig. 5b). Importantly, the IMP-1 reconstituted cells were more resistant to Taxol-induced cell death (Fig. 5c). These cells completely lost the increased sensitivity to Taxol-induced G2/M arrest normally exhibited by let-7g expressing cells (Fig. 5d). In summary, the data presented here indicate that let-7 regulates the expression of IMP-1 causing destabilization of MDR1 mRNA and increased sensitivity to Taxane-induced cell cycle arrest and cell death in ADR-RES cells.
Both IMP-1 and MDR1 are upregulated in ovarian cancer patients in response to chemotherapy while let-7d is downregulated
To test whether the link between IMP-1 and MDR1 was also found in ovarian cancer patients, we tested 20 patients with epithelial ovarian cancer for whom surgical material was available from the initial surgery before chemotherapy with Taxol and carboplatin (first surgery) and at the time of recurrence (second surgery). Tumor tissue was stained for both IMP-1 and MDR1. Two pathologists (A.M and K.G.) independently scored the staining intensities in the tumor areas. The staining intensity of IMP-1 and MDR1 in a portion of the samples was elevated in samples that were collected after chemotherapy at the time of recurrence (Fig. 6a). Consistent with a molecular function of IMP-1 on MDR1, both proteins were expressed in the same areas of the tumor in the patient samples (Fig. 6a). The average scores of the analyses of both pathologists were taken to determine the change in staining intensity between the two surgeries. In parallel tumor material was punched out from the paraffin blocks of 18 of the 20 patients at the first and second surgery and the let-7d expression was determined using real-time PCR. In ten of the 18 samples (56%), the expression of let-7d was reduced more than 10%, at the same time it was increased in only four samples (22%; Fig. 6b). This difference was statistically significant (p = 0.043). The inverse correlation between the let-7d and the MDR1 expression was almost statistically significant (p = 0.07). The mean increase in score of MDR1 expression (scale of 0–3) in the ten patients with reduced let-7d was 0.3, whereas the mean reduction in score among the four patients with increased let-7d expression was only −0.3. In addition, in situ hybridization for let-7d on the tumors from patient #2 was performed. Let-7d expression was reduced in the tumor at the second surgery when compared with the first surgery (Fig. 6c).
As MDR1 is an established resistance factor for treatment with Taxol, patients were sorted into two groups according to their change in expression of MDR1: group I with clear upregulation of MDR1 and group II with no change or downregulation of MDR1 (see Supporting Information Table S1). Group I patients had a shorter disease-free survival than group II patients (Fig. 6d). Consistent with the function of IMP-1 to stabilize MDR1 expression, we found a significant co-upregulation of both proteins in all but one patient (Fig. 6e). We hypothesized that chemotherapy causes upregulation of MDR1. The downregulation of let-7 which results in upregulation of its target IMP-1 generates a permissive environment for the stabilization of MDR1. To test this hypothesis, we tested whether the increase in the expression of both proteins correlated with the number of disease-free days between the beginning of chemotherapy and the time of recurrence. Indeed, patients in whom both proteins were upregulated in response to chemotherapy had a poorer prognosis compared with patients in whom both proteins were downregulated (Fig. 6f). The data support a model where the let-7 target IMP-1 stabilizes the mRNA of MDR1, a drug transporter that is often upregulated in response to treatment with Taxanes.
Human cancers demonstrate differential sensitivity to chemotherapeutic agents that is related to the inherent sensitivity of their tissues of origin. Cancer patients, however, frequently develop resistance to chemotherapy via acquisition of distinct mechanisms that alter the cellular response to a particular agent or class of agents.40 The loss of let-7 is a widely observed phenomenon in human cancer,5 and loss of particular let-7 family members has been correlated with drug resistance in multiple tumor types. There is to date, however, only one example in the literature of a clear mechanistic link between expression of let-7 family members and decreased chemotherapeutic sensitivity. It was recently demonstrated by Shimizu et al. that let-7 specifically targets the 3′-UTR of Bcl-xL in hepatocellular carcinoma cell lines, and that overexpression of let-7 sensitizes these cell lines to the chemotherapeutic agent sorafenib via this mechanism.40
Most let-7 family members are coregulated. We focused on let-7d and let-7g in this study because we previously showed for ovarian cancer that let-7d was the best and let-7g the second best marker for earlier more differentiated forms of cancer.41 In patient samples, we monitored expression of let-7d. We chose let-7g for overexpression in cancer cell lines because it could be expressed at higher levels than let-7d for reasons not known (data not shown).
Our data demonstrate that altering let-7 levels in cells that have not been selected for drug resistance and do not express MDR1 has little effect on the sensitivity of the cells to Taxanes. Once MDR1 is expressed, however, potentially as a result of exposure to chemotherapeutic agents, the let-7 target IMP-1 can function to stabilize the MDR1 mRNA and cause a more sustained expression of MDR1 protein. Loss of let-7 therefore provides a permissive environment for the acquisition of drug resistance. This interpretation is supported by data from ovarian cancer patients, where we found a good correlation between the downregulation of let-7d, and the upregulation of both IMP-1 and MDR1 in relapsed patients following initial chemotherapy. Of note, no patient was identified with high MDR1 expression in the absence of IMP-1 expression, further suggesting that high MDR1 expression requires the presence of IMP-1 for stabilization of its message.
The identification of MDR1 as an indirect target of let-7 through downregulation of IMP-1 suggests that let-7-regulated drug resistance may occur mostly in patients with recurrent disease who have undergone chemotherapy. This hypothesis is supported by the finding that breast cancer tumor-initiating cells, enriched in patients that have received chemotherapy due to their inherent drug resistance, are characterized by the absence of let-7 expression and expression of drug transporters such as ABCG2.11 Accordingly, we demonstrate downregulation of let-7d and co-upregulation of IMP-1 and MDR1 occur in distinct areas of ovarian tumors from patients who had received a Taxol-based chemotherapeutic regiment. MDR1 has been shown in multiple instances to affect treatment outcome by conferring intrinsic or acquired resistance to a variety of drugs.42 Drugs inhibiting MDR1 P-gp activity have been coadministered during chemotherapy of certain tumors, and the identification of novel means to inhibit MDR1 function is an active area of research.42 Although our findings need further confirmation, they could affect the criteria by which patients are selected for the treatment. Furthermore, they suggest that let-7-based treatment modalities could potentially function to increase drug sensitivity in patients with recurrent treatment refractory disease.
The authors would like to thank Drs. Susan Holbeck, Sven Diederichs and Scott Hammond and Stefan Hüttelmeier for providing the gene of let-7g, the Lin28 and Lin28B plasmids, NCI60 cells, and the anti-IMP-1 antiserum, respectively. This work was supported by the Robert H. Lurie Comprehensive Cancer Center (to M.P.) and the NCI (to E.L.).