The MDR1 (ABCB1) gene codes for P-glycoprotein, an ATP-dependent membrane transporter which pumps many cytotoxic drugs out of the cells and confers resistance to chemotherapy.1, 2 Efforts to circumvent the occurrence of drug resistance in the clinic have focused mainly on the development of P-glycoprotein modulators3 and MDR1 transcriptional repressors.4, 5 Transcription from MDR1 can start at 2 promoters: a major downstream promoter (DSP), which is used by most cell lines and tissues expressing the MDR1 gene,6 and a minor upstream promoter (USP).7, 8, 9 Although lacking a TATA-box, the DSP promoter has a CAAT box and an inverted CCAAT element/Y-box upstream from the transcriptional start site (tsp), where an initiator element (Inr) is found. Reporter gene expression studies have identified an array of transcription factor binding sites including, among others, 2 GC boxes, 2 p53 elements, an ETS-binding element, a heat shock element, several T-cell factor elements, a nuclear factor for interleukin-6 (NF-IL6) element and an activator protein-1 site.10 In addition, a region termed the enhanceosome has been identified in the MDR1 DSP at which different stress signals converge to upregulate MDR1 transcription.11 Several stimuli, including histone deacetylase inhibitors, differentiation agents, ultraviolet irradiation and the chemotherapeutic drug doxorubicin, mediate the binding of an enhanceosome complex containing the trans-acting factors, NF-Y, Sp1 and Sp3, the recruitment of the histone acetyltransferase P/CAF to this complex, histone acetylation and chromatin remodeling, thus promoting MDR1 transcription.12
In acute lymphoblastic leukemia patients overexpressing P-glycoprotein, the MDR1 USP represents the major promoter (and in some patients, the only promoter) used by mononuclear cells.13 In breast cancer, the presence of transcripts derived from the MDR1 USP correlates with metastatic invasion of lymph nodes, and can be detected in isolated carcinoma cells, from both the primary tumor and from invaded lymph nodes.14 However, the MDR1 USP has not been characterized in detail. Here we report the detection of tissue-specific MDR1 USP-derived transcripts in normal human tissues, the characterization of this promoter in drug-resistant cells and its activation by the interleukin-6 (IL6)/NF-IL6 pathway. Because of their low abundance, translation of P-glycoprotein from MDR1 USP-derived transcripts is insufficient to affect the response to first-line chemotherapy in patients with advanced breast cancer.
A total of 60 patients with classical, Philadelphia chromosome positive chronic myelogenous leukaemia (CML) seen at the Hematology Department, Hammersmith Hospital, were used in this study. Peripheral blood samples were taken from the patients, after informed consent, at presentation in chronic phase. Only 1 patient, who showed activation of the MDR1 USP, had been previously treated and was resistant to busulfan therapy. White blood cells were separated from the whole blood by red cell lysis.
A total of 137 primary breast cancer tissues from patients who developed recurrent disease, and were treated with first-line chemotherapy, were analyzed. After analysis of housekeeping mRNAs for normalization of quantitative PCR (see later), 22 were discarded (20 due to an aberrant RPLO:GAPDH ratio and other 2 due to an ambiguous chemotherapy response). The patient and tumor characteristics of the remaining 115 patients in association with the median MDR1 USP-derived mRNA levels and their interquartile range are shown in Table I. Of the 115 patients, 36 (31%) were treated with CMF (cyclophosphamide, methotrexate, 5-fluouracil) and 79 (69%) with FAC/FEC (5-fluouracil, adriamycin/epirubicin, cyclophosphamide). Response to first-line chemotherapy was: complete remission, 6 patients; partial remission, 41 patients; stable disease, 27 patients; and progressive disease, 41 patients. Three patients had a short stable disease and were considered as nonresponders. Therefore, 71 patients (62%) were calculated as responders. Seventy-nine patients (69%) experienced progression, and their median progression-free survival (PFS) was 4 months (range, 1–73 months). One hundred and two patients have died, with a median time to death of 11 months (range, 1–82 months) after the start of first-line chemotherapy. Median follow-up time of patients still alive is 81 months (range, 35–112 months).
Table I. Associations of MDR1 USP-Derived Transcript Levels with Clinicopathologic Factors
Number of patients
p for Kruskal–Wallis test, including a Wilcoxon-type test for trend when appropriate.
p for Mann–Whitney U test.
Cut point used for ER and PgR (2 missing values); ≥10 fmol/mg protein.
Human peripheral blood mononuclear cells (PBMCs) were isolated from 7 volunteer donors using Histopaque-1077 (Sigma, St. Louis, MO) as recommended by the manufacturers.
This research study was approved by The Hammersmith Hospital Research Ethics Committee and was conducted in accordance with the declaration of Helsinki. Samples were blinded for analysis and patients understood that the results would not be made available to them.
Cell lines and transfection
The drug-resistant cell lines NCI/ADR-RES (formerly known as MCF7/Adr), KBV-1 and KD225 have been described previously.14 Breast carcinoma MCF-7 cells were obtained from the American Type Culture Collection (Manassas, VA). Where indicated, MCF-7 cells were treated with 50 ng/ml human IL-6 (BD Pharmingen, San Diego, CA) for up to 96 hr. IL-6 was added to the culture every 2 days.15 Cells were transiently transfected using an Amaxa Nucleofector (Amaxa GmbH, Cologne, Germany) following the manufacturers' recommendations with 4.5 μg of firefly luciferase test plasmid and 0.5 μg of either phRGTK (expressing Renilla luciferase) or pSV-β-galactosidase (both from Promega, Madison, WI) to normalize for transfection efficiency. Reporter gene expression was measured in extracts from cells 24 hr after transfection using the Dual-Luciferase reporter assay system (Firefly and Renilla luciferase) and the β-galactosidase enzyme assay system (Promega) following manufacturer's recommendations.
Total RNA was prepared from cell lines and frozen breast cancer biopsies (three 15-μm cryostat sections) by RNAzol extraction (Biogenesis, Poole, UK). For all breast cancer samples the number of tumor cells represented at least 50% of total nucleated cells, as judged by hematoxylin and eosin staining of an additional cryostat section. Total RNA from PBMCs from CML patients was isolated by the guanidinium thiocyanate method.16 Human Total RNA Master Panel II was purchased from BD Biosciences Clontech (Palo Alto, CA).
Determination of mRNA expression by reverse transcription-PCR
Total RNA was transcribed with Superscript II RNase H− reverse transcriptase (breast cancer samples) or avian myeloblastosis virus reverse transcriptase (all other samples) as described previously.2, 14 The nucleotide sequences of all primers used in this study are listed in Supplementary Table I. Estimation of total MDR1 mRNA (derived from both USP and DSP) by amplifying exons 6, 7 and 8, MDR1 USP-derived transcripts by amplifying exon –1, exon 1 and exon 2 (Fig. 1a) and GAPDH mRNA has been described previously.14 Quantitative real-time PCR was carried out to estimate RPLO mRNA with primers RPLOf and RPLOr by the comparative threshold cycle method, as described previously.2, 14
To compare expression among different tumor samples, expression from several genes potentially useful for normalization purposes was analyzed (18S rRNA, RPLP0, TBP, GAPDH, β-actin and β-glucoronidase). Expression of GAPDH and RPLP0 mRNAs showed the best correlation and were used to exclude tumor samples with abnormal housekeeper protein transcript levels from the analysis (aberrant RPLPO:GAPDH ratio).2 The relative levels of MDR1 USP-derived and GAPDH mRNA were determined as earlier by analyzing the amplified products in the linear range of amplification (usually between 20 and 30 cycles).
Ribonuclease protection assay
A cosmid human chromosome 7 genomic DNA library (Geneservice, Cambridge, UK) was screened with human MDR1 cDNA (cDNA clone 65704; American Type Culture Collection) using standard protocols17 and cosmid Y44G16 isolated. A 6.5-kb EcoRI fragment comprising the 5′-end of MDR1 was isolated from Y44G16 and cloned into pBluescript II KS+ (Stratagene, La Jolla, CA) generating pD18. A 1-kb DNA fragment, comprising exon 1 and flanking intervening sequences of MDR1, was isolated from pD18 by digestion with PstI and cloned into pBluescript II KS+ generating pD364. A uniformly labeled antisense RNA probe (∼1 kb) derived from EcoRI-digested pD364 was synthesized with T7 RNA polymerase using a MAXIscript in vitro transcription kit (Ambion, Austin, TX). RNase protection was carried out using a Ribonuclease Protection Assay Kit (Ambion). Briefly, 30 μg total RNA was hybridized to the MDR1 RNA probe for 16 hr at 42°C and then digested with RNase A/T1 essentially as described by the manufacturer. RNase-protected fragments were run in a denaturing urea–polyacrylamide gel and quantitated with a PhosphorImager (GE Healthcare, Piscataway, NJ).
Rapid amplification of cDNA ends
To determine the tsp of MDR1 USP, rapid amplification of cDNA ends was performed with RNA isolated from normal human brain and prostate (BD Biosciences Clontech), white blood cells from CML patients, NCI/ADR-RES, and KBV-1 cells using a 5′-/3′-RACE kit (Roche Diagnostics GmbH, Mannheim, Germany) following manufacturer's recommendations and oligonucleotides OLEY178 and OLEY179. PCR products were cloned in a Topo Zero Blunt vector (Invitrogen, Paisley, UK) and sequenced. At least 10 clones were sequenced for each RNA.
The 1.1-kb sequence of the MDR1 DSP from Sma I and Hin dIII-digested pMDR1 (−1,202)12 was cloned upstream of the luciferase gene in SmaI and HindIII-digested promoterless pGL3Basic (Promega), generating pGL1100DSP.
A 2.7-kb region 5′- of the USP transcriptional start point was amplified with an Expand High Fidelity PCR kit (Roche Diagnostics GmbH) using the primers OLEY133 and OLEY137 from the RG060P12 BAC genomic clone (Invitrogen). The 5′-end of this sequence began at −2,502 and the 3′-end terminated at +187 (relative to the tsp determined from brain mRNA). The PCR product was digested with HindIII and NcoI and cloned upstream of the luciferase reporter gene in HindIII and NcoI-digested pGL3Basic (Promega) generating pGL2700USP.
pGL2700USP was used as a template for the generation by PCR (Expand High Fidelity PCR kit; Roche Diagnostics GmbH) of a series of deletion fragments using the sense oligonucleotides OLEY136, OLEY160, OLEY157, OLEY245, OLEY158, OLEY159 and antisense oligonucleotide OLEY137 that were digested with HindIII and NcoI and cloned upstream of the luciferase reporter gene in HindIII and NcoI-digested pGL3Basic (Promega) generating pGL1100, pGL900, pGL500, pGL400, pGL250 and pGL180, respectively.
pGL400 and pGL250 were used as templates for the generation of a series of mutations in which the recognition sites for transcription factors NF-Y (−214), ETS (−58), NF-IL6 (−58), PuF (−65) and the PuBox (−57) were disrupted. After amplification (Expand High Fidelity PCR kit; Roche Diagnostics GmbH) with sense oligonucleotides OLEY246, OLEY247, OLEY248, OLEY250 and OLEY249, respectively, and the antisense oligonucleotide OLEY137, the PCR products were digested with HindIII and NcoI and cloned upstream of the luciferase reporter gene in HindIII and NcoI-digested pGL3Basic (Promega) generating pGL400ΔNFY, pGL250ΔETS, pGL250ΔNFIL6, pGL250ΔPuF and pGL250ΔPuBox, respectively.
Recognition sites for transcription factors H-APF1 (−330) and NF-IL6 (−120) were mutated in pGL500 by site-directed mutagenesis using a QuikChange Site-Directed Mutagenesis Kit (Stratagene) with oligonucleotides OLEY197/198 and OLEY195/196 generating pGL500ΔAPF and pGL500ΔNFIL6, respectively.
Oligonucleotides corresponding to a single (OLEY211b/OLEY212b) and triplet repeat (OLEY213/OLEY214) of the −64to −47 MDR1 USP region were phosphorylated, annealed and ligated to HindIII and XhoI-digested pGL180 following standard procedures,17 originating pGL180NFIL6X1 and pGL180NFIL6X3, respectively.
The 5′-UTR of MDR1 mRNA transcribed from the USP was amplified from human brain RNA (BD Biosciences Clontech) by RT-PCR14 using oligonucleotides OLEY175 and OLEY25 and cloned into pCR4-TOPO (Invitrogen) generating pD136. The 5′-UTR of MDR1 mRNA transcribed from the USP was liberated from pD136 by digestion with HindIII and NcoI and cloned into HindIII and NcoI-digested pGL3 promoter (Promega) generating pD323. The sequence immediately upstream from the AUG in pD323 was mutated back to the same context as that found in MDR1 by site-directed mutagenesis (QuikChange Site-Directed Mutagenesis Kit; Stratagene) with oligonucleotides OLEY176 and OLEY177 generating pD337. The firefly luciferase gene with the 5′-UTR of MDR1 mRNA transcribed from the USP was liberated from pD337 with EcoRI and XbaI and cloned into EcoRI and XbaI-digested pCINeo (Promega) generating pCIUSP. Two upstream ATGs in the 5′-UTR of MDR1 mRNA transcribed from the USP were mutated by site-directed mutagenesis as earlier using the oligonucleotide pairs OLEY199/OLEY200 and OLEY201/OLEY202, generating pCIUSPΔ1 and pCIUSPΔ2, respectively.
An ∼1.6-kb DNA fragment carrying the human NF-IL6 cDNA was obtained from EcoRI, XhoI and AflIII digestion of IMAGE clone 3028673 (Geneservice), treated with the Klenow fragment of DNA polymerase17 and cloned into the SmaI site of pCINeo pGL3promoter (Promega) generating pCINFIL6.
In vitro transcription and translation
In vitro synthesis of transcripts from PCIUSP and derived plasmids and in vitro translation in rabbit reticulocyte lysates was performed using a TNT Quick Coupled Transcription/Translation System (Promega) following manufacturer's recommendations. In vitro translated firefly luciferase activity was measured directly in the lysates as earlier after 90 min at 30°C.
Electrophoretic mobility shift assay
Nuclear extracts were prepared from NCI/ADR-RES cells using the NucBuster protein extraction kit (Novagen, Darmstadt, Germany), according to manufacturer's instructions. Oligonucleotide OLEY220 (comprising the NF-IL6 binding site at −58) was labeled by phosphorylation with [γ-32P]-ATP with T4 polynucleotide kinase (Promega) for 1 hr at 37°C and annealed to complementary oligonucleotide OLEY221. Unincorporated label was removed by chromatography through a Sephadex G-25 spin column (GE Healthcare, Chalfont St. Giles, UK). Two probes with mutations in the recognition sequence for NF-IL6 were generated by annealing unlabeled oligonucleotides OLEY226 and OLEY227 (NF-IL6 mutated 1) and OLEY228 and OLEY229 (NF-IL6 mutated 2). An irrelevant probe was generated by annealing OLEY224 and OLEY225. The labeled probe was incubated with the NCI/ADR-RES nuclear extract (7 μg protein) in binding buffer (100 mM KCl, 20 mM HEPES, 0.2 mM EDTA, 20% (v/v) glycerol, pH 8.0), in a total volume of 20 μl. Nonspecific competitor poly(dI⋅dC) and single-stranded DNA (EMSA Accessory Kit; Novagen) was included in the binding reaction to minimize the binding of nonspecific proteins to the labeled target DNA, where the reaction mixture contained an excess of unlabeled, nonspecific or mutated probes to compete for transcription factors. Reactions were performed for 30 min at room temperature before the samples were resolved in a 6% nondenaturing polyacrylamide gel, run at 100 V for 2–3 hr. Wet gels were exposed in a PhosphorImager screen overnight.
Sucrose gradient density centrifugation and detection of RNA across polysome profiles
NCI/ADR-RES cells were grown to 80% confluence and treated with 200 μM cyclohexamide at 37°C for 5 min prior to harvesting by trypsinization in the presence of cyclohexamide. The cell pellets (2 × 107 cells) were resuspended in 0.5 ml lysis buffer [0.3 M NaCl, 15 mM MgCl2, 15 mM Tris-HCl, pH 7.6, 1% (v/v) Triton X-100, 100 U RNAsin (Promega), 1 mg/ml of heparin, 200 μM cycloheximide] and lysed on ice for 10 min. Nuclei were pelleted at 10,000g for 10 min, and the resulting supernatant was layered onto a 10.5 ml continuous sucrose gradient [15–50% (w/v) sucrose in detergent-free lysis buffer] and centrifuged at 150,000g for 90 min in an SW41-Ti rotor (Beckman Coulter, Fullerton, CA) at 4°C. Fractions of ∼500 μl were collected from the top using a Teledyne ISCO (Lincoln, NE) density gradient fractionation system with continuous monitoring at 254 nm as a function of gradient depth. RNA from sucrose gradient fractions was isolated essentially as described16 and modified as follows: an equal volume of 8 M guanidine thiocyanate, 25 mM sodium citrate, 0.5% (w/v) N-lauryl sarcosine, 0.1 M β-mercaptoethanol, 3 M sodium acetate (pH 5.2) and 50% (v/v) water-saturated phenol buffer was added to the fractions, vortexed and 1/10 volume of chloroform and 1 μg/ml glycogen (Ambion) was added. Samples were mixed by vortexing at room temperature and centrifuged at 10,000g for 10 min at 4°C. An equal volume of isopropanol was added to the aqueous phase and was left at 4°C overnight. After centrifugation at 20,000g for 15 min, RNA pellets were washed with 75% (v/v) ethanol and resuspended in RNAse-free water. Detection of MDR1 mRNA was by semiquantitative RT-PCR as described earlier.
Differences in expression levels of MDR1 USP-derived mRNA were assessed with the Mann–Whitney U test or Kruskal–Wallis test. Univariate logistic regression analysis was used to associate MDR1 USP-derived mRNA levels with the type of response to first-line chemotherapy. The overall response (OR) and its 95% confidence interval (CI) were calculated. The likelihood ratio test in logistic regression models was used to test for differences. The Cox proportional hazard model was used to calculate the hazard ratio (HR) and 95% CI in the analysis of PFS and post-relapse overall survival (PRS). Survival curves were generated using the method of Kaplan and Meier, and the log-rank test was used to test for differences. All p values are two-sided and p < 0.05 was considered statistically significant. Computations were done with the use of STATA statistical package, release 9 (STATA, College Station, TX).
MDR1 USP-derived transcripts are detected in drug-resistant cells and human tissues
We have previously developed a highly sensitive PCR method for detecting transcripts derived from the MDR1 USP. This method uses a sense primer in the first exon from MDR1 (annealing only to MDR1 USP exon –1 derived transcripts) and an antisense primer annealing to the first nucleotides of the P-glycoprotein coding region, followed by hybridization of the PCR products with a labeled oligonucleotide spanning the first intron between exons –1 and 1a. Thus, only bona fide mRNA derived from the MDR1 USP, and with the capacity to code for P-glycoprotein, is detected (Fig. 1a).
MDR1 overexpression has been observed in several leukemias, and in CML it is associated with the accelerated and blastic phase of the disease.18 In addition, some lymphoblastic leukemia patients overexpressing P-glycoprotein have activated the MDR1 USP.13 We asked whether the MDR1 USP was also activated in PBMCs from CML patients at presentation. Transcripts derived from the MDR1 USP were only detected in 2 out of 60 patients analyzed. None of the 7 PBMC controls showed activation of the MDR1 USP (Supplementary Fig. 1). Thus, MDR1 USP activation occurs very infrequently in CML at presentation.
To determine if the MDR1 USP is activated in any healthy human tissues, we used this USP-specific RT-PCR method to detect MDR1 expression in a panel of RNAs isolated from healthy human tissues including PBMCs (Fig. 1b).The MDR1 USP-expressing drug-resistant cell line NCI/ADR-RES (formerly known as MCF/Adr) served as a positive control. The highest levels of MDR1 USP-derived transcripts, which were comparable to those found in NCI/ADR-RES cells, were detected in adult and fetal brain, prostate, testis, thyroid gland and uterus. RNA derived from adult or fetal liver, heart, kidney, skeletal muscle, trachea and peripheral blood mononuclear cells did not express MDR1 USP-derived transcripts (or they were below the detection level of this assay). Previously, we have demonstrated a low frequency of MDR1 USP activation in normal breast tissue.14 All other tissues showed intermediate levels of MDR1 USP-derived transcripts (Fig. 1b). We also assayed for total amounts of MDR1mRNA, transcribed from either MDR1 USP or DSP, by amplifying MDR1 mRNA from the coding region (exons 6–8). The highest expressing tissues, with levels of mRNA comparable to those of the highly drug-resistant NCI/ADR-RES cells, were adrenal gland, placenta, uterus and PBMCs. Interestingly, tissue-specific promoter usage was observed in some tissues such as adrenal gland and PBMCs, where no MDR1 USP-derived transcripts, but high levels of DSP-derived MDR1 mRNA, were detected. Conversely, USP-derived MDR1 mRNA was not detected in the absence of exons 6–8 MDR1 transcripts, indicating that the former is a bona fide mRNA (Fig. 1b). While these sensitive RT-PCR methods demonstrate the presence of MDR1 USP-derived transcripts in a given tissue, and identify tissues expressing DSP-only initiated transcripts, they do not inform on the relative proportion of USP to DSP-derived MDR1 mRNA in those tissues expressing transcripts from both promoters. RNase protection analysis was used to differentiate USP from DSP-derived MDR1 transcription. By this assay we determined that there are between 5.5 and 7.4 times more transcripts derived from the MDR1 DSP than from the USP in 3 drug-resistant cell lines, NCI/ADR-RES, KBV-1 and KD225 (Fig. 1c), indicating that the DSP is the dominant promoter, at least in these cell lines. Because of the large amount of RNA necessary to carry out the RNase protection assays, we could not perform a similar analysis in human tissues, although with this technique no MDR1 USP-derived transcripts have previously been detected.8
Thus, bona fide transcripts derived from the MDR1 USP are detected in drug-resistant cell lines and normal human tissues. In the former, they represent ∼15% of total MDR1 mRNA. In the latter, a direct assessment of their relative abundance was not possible using RNase protection assays, although they can be detected by PCR. This may indicate a very low abundance.
Identification of a 350-bp region upstream of MDR1 exon −1 with promoter activity in drug-resistant cells
To gain insights into the mechanisms by which the MDR1 USP is activated, we characterized this promoter. We isolated a genomic DNA fragment of ∼2.7 kbp upstream from MDR1 exon –1 comprising several DNA repetitive elements (LINE2 and TATAA repeats) as well as a human endogenous retroviral long terminal repeat (LTR) (Fig. 2a). When inserted into a promoterless luciferase reporter vector, this 2.7-kb genomic DNA fragment (pGL2700USP) had promoter activity in NCI/ADR-RES cells (Fig. 2b). Consistent with the RNase protection data, luciferase activity from a reporter carrying the MDR1 DSP (pGL1100DSP) was ∼10-fold higher than that from the MDR1 USP, indicating that in NCI/ADR-RES cells MDR1 USP is weaker than the DSP.
Since the promoter activity associated with many LTRs can drive the expression of nearby loci,19 we asked whether the luciferase reporter activity from pGL2700USP could be derived solely from the LTR. Deletion analysis indicated that a DNA fragment ∼350 bp upstream from exon −1 (pGL500), in which the repetitive elements and the LTR are absent, had the same promoter activity as the 2.7 kbp in pGL2700USP. Thus, MDR1 USP activity is not due to the retroviral LTR (Fig. 2c).
Truncation (pGL250) or deletion (pGL180) of the genomic DNA fragment upstream from MDR1 exon –1 diminished and abolished, respectively, luciferase reporter activity (Fig. 2c). This indicates that the 350-bp genomic fragment in pGL500 contains all the elements necessary for promoter activity present in pGL2700USP. Several putative transcription factor binding elements were located in this region, including H-APF1, NF-Y, NF-IL6 and ETS, and there was a high degree of conservation between the human and mouse MDR1 USP regions (Fig. 3). Interestingly, an 18-bp region (−64 to –47) containing putative overlapping elements for PuF, ETS, NF-IL6 and a PuBox was fully conserved between human and mouse promoters.
The tsp determined in mRNA isolated from prostate cells, the drug-resistant cell lines NCI/ADR-RES and KBV-1, and from white blood cells from CML patients, varied within a region of 16 bp. This variation in tsp has also been observed in the MDR1 DSP.8 However, MDR1 USP-derived transcripts from healthy human brain initiated ∼60 bp further upstream (Fig. 3). As for the MDR1 DSP, no TATA box was apparent immediately upstream from the tsp. Two Inr elements (consensus PyPyA+1N(T/A)PyPy) were identified close to the 2 regions where transcription initiates: +4 to +10 and +52 to +58. No other core promoter elements were present in the vicinity of the MDR1 USP tsp.20
The presence of an upstream open reading frame in MDR1 5′-UTR modulates translation of MDR1 mRNA
Two upstream AUGs are present in the 5′-UTR at positions +11 and +97 relative to the MDR1 USP tsp. The first translational initiation codon is immediately followed by a stop codon, whereas the second is at the beginning of an upstream open reading frame (uORF) and has the potential to translate a peptide of 13 amino acids (Fig. 3). Since the presence of uORFs in the 5′-UTR can affect translation of the downstream ORF,21 we cloned the cDNA corresponding to the entire 5′-UTR from MDR1 mRNA transcribed from the USP upstream from a luciferase reporter gene (pCIUSP). To assess how the presence of the uORF would affect luciferase translation, either uATGs or just the uATG from the uORF were mutated (pCIUSPΔ2 and pCIUSPΔ1, respectively). Transfection of pCIUSP into cells would hamper the assessment of the effect of the uORF on luciferase expression, since this reporter could also be translated from shorter transcripts initiated in the MDR1 DSP, which overlaps exon 1a (Fig. 1a), and have no uORF. Therefore we used the T7 RNA polymerase promoter present in the plasmid to drive the expression from the above vectors in vitro and used rabbit reticulocyte extracts to translate the in vitro transcribed RNAs into protein and measured luciferase activity. Translation of the luciferase reporter was detected from pCIUSP-derived transcripts. However, when the uORF ATG was mutated, luciferase expression increased ∼50%. Mutagenesis of the first uATG did not increase luciferase expression further (Fig. 4a). Thus, the uORF located in the 5′-UTR of MDR1 mRNA transcribed from the USP downregulates, but does not abolish, translation of the downstream ORF.
This was further confirmed when we assessed the translational status of transcripts derived from the MDR1 USP in NCI/ADR-RES cells by velocity sedimentation of cell extracts on continuous sucrose gradients.22 The gradient A254nm profile showed a large peak in the top fractions where tRNA and other small RNAs migrate (data not shown). The next 3 peaks represent the 40S and 60S ribosomal subunits and the 80S monosomes (fractions 3–7). RNA molecules associated with 2 or more ribosomes, and therefore engaged in translation, migrated approximately in the middle of the gradient (fractions 8–10), and with increasing number of polyribosomes they migrated further towards the bottom of the gradient (fractions 10–22) (Fig. 4b). As we have demonstrated previously for P-glycoprotein-expressing leukemia cells,22MDR1 transcripts (exons 6–8) from NCI/ADR-RES cells migrated to the lower part of the gradient which is associated with translating polyribosomes. MDR1 USP-derived transcripts from NCI/ADR-RES cells also migrated at the bottom of the gradient (Fig. 4c). When ribosomes were dissociated by including 15 mM EDTA in the extraction buffer and sucrose gradients,22 the position of both transcripts shifted toward the lighter (nonpolysome) fractions of the gradient (data not shown). This indicates that these transcripts, despite the uORF in the 5′-UTR, are heavily loaded with ribosomes and are being translated into P-glycoprotein.
A NF-IL6 element is responsible for the activation of both MDR1 promoters in drug-resistant cells
To determine the elements important for promoter activity in NCI/ADR-RES cells we performed mutagenesis analysis. Mutation of the −330 H-APF1 element (the nomenclature of the elements indicates the nucleotide position relative to the tsp followed by the transcription factor name) did not change promoter activity of pGL500, indicating that this site is not important for MDR1 USP activation in NCI/ADR-RES cells (Fig. 5a). However, there was a decrease in promoter activity between pGL500 and pGL400 reporter plasmids, indicating that some regulatory elements must reside between –350 and –217 and contribute to the activation of the MDR1 USP (Fig. 5a). Mutagenesis of −214 NF-Y and −120 NF-IL6 elements did not change the promoter activity of the corresponding parental vectors, implying that these elements are not recognized by these transcription factors. Moreover, there was no significant difference between pGL400 and pGL250, indicating that the 152-bp region upstream from pGL250 does not contribute to promoter activity (Fig. 5a).
Mutagenesis of the ETS, NF-IL6, PuF sites and PuBox within the conserved −64 to −47 region indicated that only the element recognized by NF-IL6 affected MDR1 USP activity. When this region was cloned, either singly or in triplicate, upstream from MDR1 exon −1 in pGL180, it activated the downstream luciferase reporter. Interestingly, another important element(s) must reside in the −47 to −1 region because mutagenesis of the −58 NF-IL6 element did not reduce the luciferase reporter activity to a level comparable to that of pGL180, and 1 copy of the −58 NF-IL6 element cloned into pGL180 did not restore the reporter luciferase activity to that found in pGL250 (Fig. 5a).
Electrophoretic mobility shift assays confirmed that extracts from NCI/ADR-RES cells bind to a probe corresponding to the conserved −64 to –47 region. When the recognition sequence for NF-IL6 was mutated, there was no difference in the binding between a non-specific probe and the NF-IL6 mutated probes (Fig. 5b). Thus, mutagenesis and biochemical analyses have identified the −58 NF-IL6 element as important for MDR1 USP activity.
Breast carcinoma MCF7 cells express the IL-6 receptor and are capable of mediating IL-6 signals.15 Treatment of MCF7 cells with IL-6 activates transcription from the DSP, P-glycoprotein is produced and the cells become drug resistant.6, 15 In addition treatment with IL-6 activated the MDR1 USP. Transcripts derived from the MDR1 USP were detected 1 hr after IL-6 addition, and accumulated at longer incubation times (24–96 h). As expected, total MDR1 transcripts (exons 6–8) increased also in response to IL-6 (Fig. 5c). Equally, when the MDR1 USP-luciferase reporter construct (pGL2700USP) was cotransfected into MCF7 cells with a NF-IL6 expressing plasmid (pCINFIL6), there was an increase in the reporter activity (Fig. 5d). Importantly, the MDR1 DSP responded strongly whereas the response of the USP-driven luciferase reporter increased only 2-fold when cotransfected with pCINFIL6. The USP activation due to the expression of NF-IL6 did not even reach that of DSP in the absence of NF-IL6 (Fig. 5d). Thus, IL-6 activates transcription from both MDR1 promoters through binding of NF-IL6. However, the contribution of the USP to the P-glycoprotein produced in response to IL-6 is minimal.
Expression of MDR1 USP-derived transcripts in primary breast tumors is not associated with the efficacy of first-line chemotherapy
Previously, we have determined that the presence of MDR1 transcripts (exons 12 and 13, and therefore unable to differentiate whether DSP- or USP-derived) in primary breast tumors is inversely related with the efficacy of first-line chemotherapy, and high expression level is a significant predictor of poor prognosis for patients with advanced disease.2 We asked whether levels of MDR1 transcripts specifically transcribed from the USP in primary operable breast tumor tissues of patients who, upon relapse, were treated with systemic first-line chemotherapy, correlated with clinical outcome. The associations of MDR1 USP transcript levels with patient and clinicopathological characteristics are shown in Table I. Only lymph node involvement at the time of surgical removal of the primary tumor was significantly related to increased MDR1 USP-derived transcripts (p = 0.046), in agreement with our previous data obtained with a different cohort of unselected patients.14 All other relationships were not statistically significant. In univariate logistic regression analysis for the type of response, the MDR1 USP transcript level analyzed as a log-continuous variable showed no statistically significant relationship with the efficacy of chemotherapy (OR = 1.21, p = 0.24). In addition, Cox univariate regression analysis for PFS and PRS as a function of continuous MDR1 USP-derived transcript levels, did not show a significant relationship. The Kaplan–Meier curves for PFS and PRS as a function of the level of MDR1 USP-derived transcript dichotomized at the median level are shown in Figure 6.
The existence of additional or alternative promoters is a common feature in mammalian genomes, and an estimated 18% of all genes show evidence of differential promoter usage.23 These include genes for transporters such as ABCG1, ABCG2 and ABCA1.24, 25, 26 In most cases, transcripts from alternative promoters have identical ORFs and do not give rise to different proteins, their role being tissue- and/or developmental stage-specific, and provide a means to implement transcriptional and regulatory complexity.27 Although we have shown that the uORF in the 5′-UTR of transcripts derived from the MDR1 USP reduces the translation of a downstream reporter, the P-glycoprotein sequence is identical irrespective of the promoter used to transcribe it.
Several mechanisms have been proposed for the evolution of alternative promoters, including gradual mutations, local duplication, transposable element insertion and genomic rearrangements.23 Since early attempts to detect MDR1 USP-derived transcripts by RNase protection in normal human tissues were unsuccessful, it was suggested that activation of the MDR1 USP in drug-resistant cell lines was due to gene rearrangements during the generation of these lines.8 More recently, we have detected activation of the MDR1 USP in the drug-resistant line KD225 and in breast cancer tissues where no rearrangement of the MDR1 locus had occurred.14 Here we demonstrate that a nearby retroviral LTR is not responsible for the activation of the MDR1 USP in drug-resistant NCI/ADR-RES cells and that the MDR1 USP is active in a variety of normal human tissues. Moreover, the MDR1 USP genomic region is highly conserved between human and mouse, where transcripts from an USP have also been detected (A.E. Trezise, The University of Queensland, Australia, personal communication). Together, these data indicate that the MDR1 USP is a bona fide promoter which is expressed in a variety of normal and malignant human tissues and in some drug-resistant cell lines.
P-glycoprotein is expressed in several human normal human tissues, and our data measuring the relative abundance of total MDR1 transcripts (detected with primers targeting exons 6–8 and therefore independently of the promoter used) in a panel of normal human tissues is in agreement with previous observations.28, 29, 30 However, we found a clear difference in the tissue specificity of each promoter. The most striking difference was in the adrenal gland where most transcripts are derived from the DSP. Inspection of the minimal genomic DNA region with promoter activity in drug-resistant NCI/ADR-RES cells does not show any tissue-specific transcription factor binding elements that could explain this or other transcript tissue distribution. Tissue-specific regulatory elements such as locus control regions might be located far upstream or downstream to the region with promoter activity in drug-resistant NCI/ADR-RES cells described here.31 Equally, mRNA stability is known to regulate P-glycoprotein in both normal human tissues and drug-resistant lines,22, 32, 33 and could result in the different tissue specific levels of MDR1 USP-derived transcripts.
The relative abundance of MDR1 USP-derived transcripts in drug-resistant lines is relatively modest (∼15% of all MDR1 mRNA). In drug-resistant NCI/ADR-RES cells, transcripts derived from the MDR1 USP are associated with polyribosomes, and therefore translated into P-glycoprotein, contributing to some extent to drug resistance in these cells. Considering that RNase protection assays have not previously detected these longer MDR1 transcripts in normal human tissues,8 we can infer that their proportion is even smaller in nonmalignant cells. In addition, the presence of MDR1 transcripts in the primary tumor of breast cancer patients inversely correlates with the efficacy of first-line chemotherapy for advanced disease,2 whereas the expression of MDR1 USP-derived transcripts does not. When considered together, these data suggest that in these breast cancer cases the in vivo levels of MDR1 transcripts derived from the USP are too low to translate into significant amounts P-glycoprotein to influence the response to chemotherapeutic drugs. In contrast, in acute lymphoblastic leukemia, the USP represents the major MDR1 promoter used by mononuclear cells from some patients over-expressing P-glycoprotein and in these P-glycoprotein translated from USP-derived transcripts could contribute to overall survival.13
We have previously reported that in a statistically significant high proportion of breast cancer patients, the primary tumors show a coactivation of MDR1 USP and the MDR1-overlapping gene, RPIP9 (Rap2 Interacting Protein 9).34 Although the role of the Rap2 group of proteins remains largely unknown, they might participate in adhesion and migration, which are pivotal for metastasis.35, 36 In addition, IL-6 is the only treatment, including chemotherapeutic drugs (data not shown), able to activate the MDR1 USP in breast cancer MCF7 cells. IL-6 is a pleiotropic cytokine involved in pro-inflammatory and acute phase responses which can be produced by tumor, stromal and inflammatory cells, displaying both tumor-promoting and tumor-inhibitory effects.37, 38, 39, 40 Primary cultures of breast carcinomas produce IL-6 and high levels of serum IL-6 have been observed in breast cancer patients at the preoperative stage,41 which could lead to activation of the MDR1 USP. Thus, although not associated with chemotherapy response in advanced breast cancer, activation of the MDR1 USP may represent a surrogate to mark other biological processes such as metastatic invasion or IL-6 activation.
In conclusion, MDR1 USP is a bona fide promoter active in many human normal tissues and drug-resistant cells which is activated by IL-6 via the transcription factor NF-IL6. Its activation in breast cancer patients does not seem to alter significantly the amount of P-glycoprotein produced by the DSP in the tumors and consequently does not alter the chemotherapy response. However, it might represent a marker for metastatic invasion or inflammatory processes. In addition, one should keep in mind that the measurement of the MDR1 USP-derived transcripts were carried out on the primary tumor tissue extracts, while the response to treatment was assessed on metastatic deposits that became evident years after surgical removal of the primary tumor. In this respect, a study in the neo-adjuvant setting might give more direct evidence of a potential relationship of MDR1 USP-derived transcripts with the efficacy of chemotherapy.