|Home | About | Journals | Submit | Contact Us | Français|
Bicistronic vectors are useful tools for exogenous expression of two gene products from a single promoter element; however, reduced expression of protein from the second cistron compared with the first cistron is a common limitation to this approach. To overcome this limitation, we explored use of dihydrofolate reductase (DHFR) complementary DNA encoded in bicistronic vectors to induce a second protein of interest by methotrexate (MTX) treatment. Previous studies have demonstrated that levels of DHFR protein and DHFR fusion protein can be induced translationally following MTX treatment of cells. We demonstrated that in response to MTX treatment, DHFR partner protein in a bicistronic construct is induced for longer periods of time when compared with endogenous DHFR and DHFR fusion protein, in vitro and in vivo. Using rapamycin pretreatment followed by MTX treatment, we also devised a strategy to modulate levels of two proteins expressed from a bicistronic construct in a cap-independent manner. To our knowledge, this is the first report demonstrating that levels of proteins in DHFR-based bicistronic constructs can be induced and modulated using MTX and rapamycin treatment.
The internal ribosome entry site (IRES)-based bicistronic vectors are important tools that allow coexpression of two or more gene products from a single promoter element.1 The encephalomyocarditis virus (ECMV) IRES is the most widely used IRES element in experimental and pharmaceutical applications to express proteins in eukaryotic cells or cell-free extracts.2 An IRES element can initiate translation by a cap-independent mechanism that does not require known initiation factors.3 Some viral and cellular messenger ribonucleic acids (mRNAs) use IRES elements to initiate translation in such a cap-independent manner. It is now well established that internal initiation mediated by most IRESs requires noncanonical translation factors.3–5 Several proteins that bind to the cellular IRES sequences have been identified as IRES trans-activating factors.6 Like the bacterial operon structure, a bicistronic vector consists of a cluster of genes under control of a single promoter. Theoretically, expression of a second gene could be comparable to that of the first gene because each translation sequence has the ribosome-binding site. However, in practice, IRES-dependent second gene expression is low compared with cap-dependent expression of the first gene.7,8
To develop a method to modulate levels of proteins expressed from bicistronic constructs, we exploited the translational autoregulation of dihydrofolate reductase (DHFR) by DHFR protein and induction of DHFR protein on methotrexate (MTX) treatment. MTX is a clinically important antifolate that is used in combination with other chemotherapeutic agents for treatment of acute lymphocytic leukemia; osteosarcoma; carcinomas of the breast, head, and neck; choriocarcinoma; and non-Hodgkin lymphoma.9 The primary target of MTX is DHFR that catalyzes the nicotinamide adenine dinucleotide phosphate (NADPH)-dependent formation of tetrahydrofolate from dihydrofolate.10 Previous studies from our laboratory have shown that DHFR protein levels increase on exposure to antifolate treatment; DHFR binds to its cognate RNA and regulates its own translation. Antifolates such as MTX relieve this translational control, allowing resumption of DHFR synthesis.11,12 Proteins fused to DHFR are similarly regulated and can be induced six- to eightfold on antifolate treatment.13 In this study, we explored the possibility that MTX treatment can induce proteins in bicistronic constructs separated by an IRES element from DHFR.
Exploiting translational regulation of DHFR to augment levels of a second protein by antifolate treatment may be useful for myeloprotection strategies using two or more drug resistance genes to protect bone marrow cells from the toxicity of cancer chemotherapy or for suicide gene therapy, where levels of gene products linked to DHFR can be increased by MTX treatment, making prodrug therapy more effective (eg, herpes simplex virus thymidine kinase [HSVTK] and ganciclovir or cytosine deaminase [CD] and 5-fluorocytosine [5-FC]). For the purposes of myeloprotection, we have previously shown that transduction of bone marrow or peripheral blood CD34 + progenitor cells with retroviral constructs containing DHFR fused to CD confer resistance to MTX and cytosine arabinoside.14,15 This concept has been demonstrated in proof-of-principle in vivo experiments and will be tested soon in clinical trials. A potential limitation to this approach is the possibility of generating an immune response to fusion proteins of DHFR. Therefore, we explored the use of bicistronic constructs for modulating the expression of two proteins translated separately, for the purposes of gene therapy, and possibly for other indications.
In this study, we hypothesized that DHFR, when separated by an IRES element from a second gene product, would still be subject to similar antifolate-mediated “induction” and would lead to elevated levels of the second protein (the partner protein). We reasoned that DHFR, by binding to the bicistronic message, would regulate translation of both itself and the partner protein and both would be induced by MTX. Our data indicate that DHFR can regulate expression of a partner protein in a bicistronic construct separated by an ECMV-IRES element. MTX treatment can induce the expression of proteins expressed from bicistronic constructs in vitro and in tumors in vivo. Although DHFR or DHFR fusion protein is induced within 24 hours of antifolate treatment, induction is delayed for the reporter when present as part of the bicistronic construct. Further, DHFR-IRES-reporter induction is sustained over a longer period of time than DHFR or the fusion protein. Induction of a second protein in response to MTX is independent of DHFR position with respect to the IRES element. Rapamycin pretreatment followed by MTX administration turns off induction of the first protein and allows induction of the partner protein downstream of IRES. Using this strategy, levels of a desired protein can be selectively increased severalfold by MTX treatment when the protein is expressed from a bicistronic construct containing DHFR. Importantly, it also allows for a mild selection pressure to maintain sustained production of a second protein. This strategy may have applications in gene therapy and for experiments designed to modulate protein levels expressed from a DHFR-containing bicistronic construct.
RPMI 1640, Dulbecco's Modified Eagle's Medium (DMEM), fetal bovine serum (FBS), G418 sulfate (Geneticin), penicillin, streptomycin, trypsin, and pCDNA3.1 (+) vector were purchased from Invitrogen (Carlsbad, CA). The pIRES-2EGFR and pE2F-Luc were purchased from Clontech (Mountain View, CA). Oligonucleotide primers were synthesized by Integrated DNA Technologies, Inc. (Coralville, IA). ECL reagents for Western blotting were obtained from Amersham Biosciences (Piscataway, NJ). The antibody to enhanced green fluorescent protein (EGFP) was from Roche Applied Science (Nutley, NJ). The anti α-tubulin antibody was obtained from Sigma Chemical Co. (St. Louis, MO). Rabbit polyclonal antibody to human DHFR was custom produced by Research Genetics/Invitrogen (Carlsbad, CA). Secondary antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Restriction enzymes were from New England Biolabs (Ipswich, MA). Cloning cylinders were obtained from Specialty Media (Phillipsburg, NJ). Aprotinin, phenylmethylsulfonyl fluoride, and sodium orthovanadate were obtained from Sigma. d-Luciferin for in vitro luciferase assay was obtained from Promega (Madison, WI). The luciferin for in vivo animal imaging was purchased from Xenogen-Caliper Biosciences (Massachusetts, MD).
Three inherently different cancer cell lines were used in this study. The GP+envAm12 packaging cell (AM12) system (derived from NIH-3T3 cells) was used to generate retro-viruses for transductions and was maintained in DMEM and 10% FBS supplemented with 0.4 mg/mL of G418 (Calbiochem, La Jolla, CA). The human colorectal adenocarcinoma C85 cells (metastatic to liver) were cultured in RPMI 1640 medium supplemented with 10% FBS. 16 HT1080 (CCL-121, American Type Culture Collection, Manassas, VA), a human fibrosarcoma cell line, was cultured in DMEM and 10% FBS.
DHFR-IRES-Luc vector was generated by cloning DHFR complementary deoxyribonucleic acid (cDNA) in NheI/XhoI sites of the multiple cloning site (MCS) of a pCMV-IRES-Luc vector generated in our laboratory that contains a CMV promoter and an ECMV IRES element (584 bp), a firefly luciferase gene in a pCDNA3.1 vector backbone (Invitrogen). Briefly, the ECMV-IRES-Luc vector was generated by a three-step subcloning strategy. First, the firefly luciferase was sub-cloned from the pE2F-Luc vector (Clontech) into the NcoI and XbaI sites in the MCS of pIRES-2EGFP vector (Clontech), giving rise to a pIRES-Luc intermediate vector. Subsequently, IRES-Luc from the intermediate vector was subcloned into the pCDNA3.1 vector (Invitrogen) using the NheI and XbaI restriction sites. The resultant vector was called pCMV-IRES-luc. The DHFR-luciferase fusion construct was constructed by cloning a DHFR cDNA in the MCS of the previously published TK-GFP-Luc (TGL) vector. 17 The retroviral vector construct DHFR-TK-EGFP-Luc (DTGL) quadruple fusion gene was constructed by modifying the original TGL vector 17 to include a double- mutant (dm) (Phe22Ser31) cDNA of human DHFR. Dm DHFR is MTX resistant, and its use in gene therapy will minimize cell death after MTX treatment. Similar to wild-type (wt) DHFR, dm DHFR still autoregulates DHFR expression and is induced on MTX treatment. The functionality of all four genes in the quadruple fusion construct was determined as follows: for dm DHFR, resistance to MTX; HSVTK: 14C-FIAU uptake versus 3H-thymidine uptake; EGFP: green fluorescence of transduced cells; Luc: luciferase imaging following addition of d-luciferin. Other retrovirus-based vectors that we have used have been previously described, including the DHFR-GFP, DHFR-Luc, and DHFR-IRES-EGFP vectors containing an ECMV IRES element,7 the DHFR-IRES-TS and TS-IRES-DHFR containing an ECMV IRES element, a DHFR cDNA, and a TS cDNA.18 These vectors are retrovirus-based vectors and are based on the Moloney murine leukemia virus–based plasmid. 19 All recombinant DNA manipulations were confirmed by restriction digestions and verified by sequencing.
The vector transfections of DHFR-IRES-Luc (DIL) vector in HT1080, AM12, and C85 cells were performed using lipofectamine reagent (Invitrogen). At 75% confluency, the cells were transfected with 10 μg of plasmid according to the manufacturer's protocol. Fresh medium was supplied 24 hours posttransfection. Given that DIL vector contains the neomycin phosphotransferase gene, 20 the transfected culture was selected in medium containing 750 μg/mL G418 (Invitrogen). Fourteen days later, individual clones were isolated by ring cylinders and expanded into stable resistant cell lines. To get a homogeneous population of cells, three different clones, each from one cell line (HT1080, AM12, and C85) expressing DHFR and luciferase, were selected for the study. Cell pellets of monolayer cultures (70–80% confluent) were harvested by 2-minute trypsinization followed by centrifugation (850g for 5 minutes) and by washing three times in 1 × phosphate-buffered saline (PBS).
Transfections of parental GP+envAM12 cells with respective plasmids (DHFR-EGFP, DHFR-IRES-EGFP, DHFR-Luc, DHFR-IRES-TS, and TS-IRES-DHFR) were carried out using DOTAP transfection reagent (Invitrogen). The C85 cells were plated 48 hours before the infection in 10 cm plates and were transduced by previously described methods. 16 Virus exposures, each of 6 hours duration, were carried out; for each exposure, fresh AM12 supernatant supplemented with polybrene (8 μg/mL) was used.
C85, AM12, and HT1080 parental and retrovirus-infected cells were plated in six-well plates. The cells were exposed to MTX (100 nM) for various time points (24–96 hours), and the EGFP, luciferase, and DHFR expression was determined by fluorescence microscopy, luciferase assay, and Western blotting, respectively.
C85 cells expressing DHFR-EGFP and DHFR-IRES-EGFP were treated with a single dose of MTX (100 nM), and the induction of EGFP was monitored by capturing images at various time points using fluorescence microscopy. Fluorescent images of randomly selected areas were taken with a Nikon fluorescence microscope.
AM12, HT1080, and C85 cells expressing DHFR-Luc and DHFR-IRES-luciferase were plated in 12-well plates and were exposed to MTX (100 nM) for various time points (24–96 hours). Cell lysate was made, and luciferase assays were carried out using the luciferase assay system kit (Promega) as per the manufacturer's protocol. Briefly, cells were lysed using 80 μL lysis buffer, the cell lysates were centrifuged at 14,000 rpm, and the supernatant was collected. Twenty-microliter supernatant was mixed with 100 μL of d-luciferin, and luciferase activity was measured using the luminometer TD-20/20 (Turner Design, Sunnyvale, CA). The experiment was repeated four times.
HT1080, AM12, and C85 cells expressing DIL were plated in 12-well plates and were exposed to MTX (100 nM, 500 nM) for various time points (24–96 hours). Cell lysate was made and luciferase assay was carried out using a luciferase assay system kit (Promega) as per the manufacturer's protocol (via supra). The experiment was repeated four times.
HT1080 DIL cells were plated in 12-well plates. After 24 hours, the cells were exposed to MTX (100 nM) or pretreated with rapamycin (25 nM) for 2 hours followed by MTX (100 nM) treatment for various time points (24–96 hours). Luciferase assay was carried out using a luciferase assay system kit (Promega) as per the manufacturer's protocol (via supra). The experiment was repeated three times.
Cells expressing DHFR fusion genes and bicistronic constructs were treated with MTX for various time points (24–96 hours), collected by trypsinization after appropriate incubations, washed with PBS, and resuspended in RIPA buffer containing protease inhibitors (BD Biosciences Pharmingen, San Diego, CA). Cells were sonicated (Vibra cell, Sonics and Materials Inc, Newtown, CT), using 20-second bursts and centrifuged at maximum speed in a microcentrifuge (model l5417R, Eppendorf), to collect the cell-free supernatant. Western blot analysis was carried out following separation of proteins by SDS-PAGE on a 10% polyacrylamide gel. The blots were probed with DHFR (BD Biosciences Pharmingen), thymidylate synthase (TS) (F. Maley, Wadsworth Center, New York State Department of Health, Albany, NY) and EGFP antibodies (Santa Cruz Biotechnology). The blots were stripped and probed for α-tubulin (Sigma Chemical Co), which served as a control for protein loading. The data were analyzed using the Kodak Image Station 2000MM Multimodal Imaging System (IS2000MM) (Kodak, Rochester, NY) and Image-J software from the National Institutes of Health (Bethesda, MD).
DHFR and hamster glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (control) mRNA levels on MTX treatment (0–96 hours) were examined by reverse transcription–polymerase chain reaction (RT-PCR) using the SuperScript One-Step RT–PCR with Platinum Taq Kit (Invitrogen). Total RNA was isolated from the cell pellets using TRIzol reagent (Invitrogen). The following primers were used for the RT-PCR reaction: DHFR forward primer 5′-ATGGTTGGTTCGCTAAACTG-3′ and reverse primer 5′-TTAATCATTCTTCTCATATAC-3′ and GAPDH forward primer 5′-TCCACCCATGGCAAATTCC-3′ and reverse primer 5′-AGCATCGCCCCACTTGATT-3′. The RT-PCR reaction was performed according to the manufacturer's protocol.
GST-DHFR fusion proteins were cloned into the GST Gene Fusion system using pGEX4t-1 vector (Amersham Biosciences), as described previously. 12 Briefly, the GST-DHFR proteins were expressed and purified using glutathione-sepharose 4B (GST) according to the manufacturer's protocol. The purity of the GST-DHFR proteins was determined using SDS-PAGE. The activity of the purified DHFR protein was determined using enzyme kinetic studies using a spectrophotometer. Conversion of NADPH and dihydrofolate to NADP + and tetrahydrofolate was determined from initial velocity measurements of DHFR as described previously. 12
The DIL construct contains a T7 promoter site upstream of the DHFR. The vector was linearized using Bgl II restriction enzyme (New England Biolabs) purified using a PCR cleaning kit (Qiagen, Valencia, CA). In vitro transcription was performed using T7 RNA polymerase as per the manufacturer's instructions (Promega). The nuclease-treated rabbit reticulocyte lysate system (Promega) was used to translate the in vitro transcribed RNA samples in the presence of purified GST-DHFR protein (5–50 μg). Two microliters of each transcript (1 μg/mL), 35 μL of rabbit reticulocyte lysates, 2 μL of amino acid mixture containing all amino acids, 40 units of ribosomal RNAs in inhibitor and an aliquot of nuclease-free water to a final volume of 50 μL were added, and the reaction mixture was incubated at 30°C for 90 minutes. Twenty microliters of the cell-free translation products was added to 100 μL of d-luciferin, and luciferase activity was measured using the luminometer.
DIL-expressing HT1080 cells (5 × 106) were injected subcutaneously into 6-week-old male NCR nu/nu mice (Taconic Farms Inc, NY). Once tumors reached approximately 5 mm in diameter, baseline images were captured. Mice were then randomized into two groups, following which one group was injected with a single dose of MTX (5 mg/kg), and mice were imaged at various times (0, 24, 48, 72, and 96 hours). Bioluminescence imaging was carried out on anesthetized animals using a Kodak Image Station 2000MM Multimodal Imaging System. Luciferin (Xenogen-Caliper Biosciences) was injected intraperitoneally, 150 mg/kg body weight, 5 minutes prior to imaging. Bioluminescence imaging was performed with acquisition time 60 seconds, binning 2 × 2, field of view 150, and f/stop 5.6. Fluorescence imaging was carried out for 60 seconds with 2 × 2 binning with an excitation wavelength set at 435 nm and an emission wavelength set at 535 nm. Kodak image analysis software (Kodak MI) was used to quantitate the signal intensities from regions of interest (ROI).
For statistical analysis, Student t-test was used, and p < .05 was considered significant.
Endogenous DHFR levels translationally induced after MTX treatment of the C85 parental colon cancer cells peak at 48 hours and drop back to basal levels at 72 hours (Figure 1A). To determine if MTX can augment levels of a second protein whose cDNA is linked to DHFR cDNA, EGFP was expressed in C85 human colon cancer cells as a fusion gene (DHFR-EGFP) and in a bicistronic construct (DHFR-IRES-EGFP). Schematic representation of various vectors used in the study is shown in Figure 1B. EGFP alone was transfected as a control. Of interest, in cells expressing the bicistronic construct, DHFR-IRES-EGFP, levels of EGFP remain upregulated for longer periods of time in response to MTX, similar to EGFP expressed as a fusion protein (Figure 1, C–E). Because the EGFP antibody is very strong, the DHFR-EGFP fusion band looks very prominent. However, if looked at carefully, it seems that the fusion protein levels induce at 48 hours and drop at 72 hours (see Figure 1D). This observation is further confirmed or quantitated using luciferase as a second reporter (via infra). EGFP alone was transfected as a control to demonstrate that EGFP is not induced on MTX treatment (Figure 1F). DHFR and GAPDH (control) mRNA levels were determined using RT-PCR. No significant differences in DHFR mRNA levels in response to MTX treatment were observed (Figure 1G). The observation is consistent with the notion that an MTX-mediated increase in DHFR protein level is at the translational level and not at the transcriptional level. 5, 12, 21
To demonstrate that this event was not cell or vector restricted, luciferase was used as a second reporter in another bicistronic construct, DIL. AM12 cells were transfected with DIL to express DHFR and luciferase (see Materials and Methods). AM12 cells were also transfected with DHFR-Luc fusion cDNA. Induction of both DHFR-Luc fusion protein and DIL in a bicistronic construct was assayed by luciferase assay. Similar to EGFP in DHFR-IRES-EGFP (see Figure 1B), luciferase levels in cells containing the bicistronic construct DIL remain upregulated for longer periods of time (at least 96 hours) in response to MTX compared with the DHFR-Luc fusion protein (Figure 2A).
In addition to AM12 cells, the DIL bicistronic construct was also transfected in two other human cancer cell lines, HT1080 and C85, representing human fibrosarcoma and metastatic colorectal cancer, respectively (see Materials and Methods). Luciferase levels were significantly induced on MTX treatment in a dose-dependent manner in HT1080, AM12, and C85 cells (Figure 2B). The induction of Luc in three cells expressing the bicistronic construct is similar to induction of luciferase (Figure 2C), EGFP, and DHFR proteins (Figure 2D) in AM12 cells expressing DTGL fusion protein. These studies demonstrate that marked increases in the second protein could be translationally induced by MTX in cells expressing DHFR and a second protein from a bicistronic construct containing DHFR cDNA and a second cDNA, regardless of the cell type.
We next examined the induction pattern of proteins upstream and downstream to IRES in cells transfected with the bicistronic construct DIL on MTX treatment. HT1080 cells expressing DIL were treated with MTX (100 nM), and cell pellets were collected at 0, 24, 48, 72, and 96 hours following MTX treatment. DHFR and α-tubulin protein levels were determined by Western blotting. The bands were quantitated using the Kodak image station 2000MM; the sum intensity of DHFR bands was normalized to α-tubulin. DHFR protein upstream of IRES element was induced, peaked at 48 hours, and dropped at 72 to 96 hours. Luciferase placed downstream of the IRES element was induced and remained elevated for 96 hours (Figure 3A). In contrast, when expressed as a DHFR fusion protein, the levels of the reporter protein peak at 48 hours and drop to preinduction levels at 72 hours following MTX treatment (see Figure 2A). Thus, increased levels of reporter protein are sustained over at least 96 hours after MTX treatment when expressed in the context of a bicistronic construct containing DHFR (see Figure 3A).
Similar to C85-DIL cells, we next tested in HT1080-DIL cells whether induction in DHFR and partner protein was due to an increase in the DHFR mRNA levels. DHFR and GAPDH mRNA levels were determined using RT-PCR (see Materials and Methods). Similar to our previous observation in C85-DIL cells (see Figure 1G), DHFR mRNA levels do not change on MTX treatment in HT1080-DIL cells (Figure 3B). Hence, induction of DHFR and the partner protein on MTX treatment cannot be attributed to an increase in the bicistronic mRNA levels (see Figure 1G and Figure 3B). The data corroborate previous observations that induction of DHFR protein level observed on antifolate treatment is at the translational level and not at the transcriptional level. 12
Using in vitro transcription and translation methods, we further tested whether the addition of DHFR protein can inhibit translation from the partner cistron located downstream of the ECMV IRES element. GST-DHFR protein was purified using the GST Gene Fusion system (see Materials and Methods) (Figure 4A). The purified GST-DHFR protein was found to be of the correct size and catalytically active as assayed by DHFR enzyme kinetic studies. 12 Given that the DIL vector contains a T7 promoter site upstream of DHFR, the vector was linearized, and an in vitro transcription reaction was performed to generate a bicistronic mRNA pT7-DIL. The mRNA was translated in vitro in the presence of increasing concentrations of purified GST-DHFR protein (5–50 μg) using the rabbit reticulocyte lysate system. The effect of DHFR protein on the translation of the partner protein (luciferase) was measured by luciferase assay (Figure 4B).
The addition of DHFR protein significantly inhibited translation of the partner protein luciferase. The addition of MTX to the above reaction relieved inhibition and restored luciferase levels, suggesting that DHFR protein not only autoregulates translation from DHFR mRNA but also regulates expression of partner protein separated by ECMV IRES element (see Figure 4B). As a control, we have shown before that overexpression of GST alone has no effect on DHFR expression. 12 Hence, DHFR inhibits expression from the second cistron, probably by inhibiting the CMV-IRES activity in bicistronic vectors. Hence, an increase in expression of the second protein (induction), specific for DHFR-containing bicistronic constructs, is due to the loss of DHFR-mediated repression of the bicistronic transcript on MTX treatment.
We next examined whether induction patterns of DHFR partner protein are DHFR position dependent and whether they are cap dependent or independent. To determine whether induction of partner protein is DHFR position dependent, we created two bicistronic constructs by altering positions of DHFR and TS: DHFR-IRES-TS and TS-IRES-DHFR (see Figure 1B). These vectors were transfected into C85 cells. Cells expressing DHFR-IRES-TS and TS-IRES-DHFR were treated with MTX (100 nM). Cells were lysed at various time points following MTX treatment, and DHFR and TS protein levels were analyzed by Western blotting. On MTX treatment, TS protein levels were induced in both DHFR-IRES-TS- and TS-IRES-DHFR-expressing C85 cells. Notably, increased expression of a second protein (TS or DHFR) was sustained over a longer period of time when placed downstream of IRES (Figure 4, C and D), whereas levels of the first protein were induced at 48 hours and dropped to preinduction levels at 72 hours after MTX treatment. In a control experiment, we demonstrated that levels of endogenous DHFR but not endogenous TS are induced on MTX treatment (Figure 4E). Our results indicate that the presence of DHFR in bicistronic constructs (DHFR-IRES-TS and TS-IRES-DHFR) facilitates MTX-mediated induction of partner protein TS; however, induced levels of the partner protein TS were sustained for longer periods of time when placed after the IRES element.
Next, we investigated whether the prolonged induction of downstream protein is due to enhanced cap-dependent or cap-independent translation. To explain the differences in induction patterns of the first and second proteins in a bicistronic construct, 5′ cap–dependent translation was inhibited by pretreatment with rapamycin. It is known that rapamycin inhibits 4E-BP1 phosphorylation, leading to increased association between 4E-BP1 and eIF-4E (cap-binding subunit of initiation complex), inhibiting 5′ cap–dependent translation; rapamycin has been used as a specific inhibitor of cap-dependent translation. 1,22,23 HT1080 cells expressing DIL were pretreated with rapamycin (25 nM, a nontoxic dose) and then treated with MTX (100 nM). Cells were collected following 0, 24, 48, 72, and 96 hours of MTX treatment, and levels of DHFR were determined by Western blotting. As expected, levels of the upstream DHFR protein were not increased in the cells pretreated with rapamycin and then treated with MTX, indicating that cap-dependent translation was inhibited (Figure 5A). Of interest, luciferase levels were still induced in the cells and were increased for at least 96 hours, indicating that the induction was mediated by cap-independent translation. In untreated control cells, there was no significant induction of luciferase activity (Figure 5B).
These results indicate that antifolate-induced translational upregulation of protein upstream to IRES is 5′ cap dependent, whereas translational upregulation of protein downstream IRES is 5′ cap independent in a bicistronic construct containing DHFR. Given that cap-dependent translation is stringently regulated in the cell, levels of upstream protein are induced and drop back to normal. It is more likely that the regulation either is or appears to be less stringent as there are many fewer IRES-containing mRNAs (estimated to be about 3% of the total mRNA population). 24 Hence, downstream protein can remain induced for longer periods, and the induction is not inhibited by rapamycin pretreatment (see Figure 5, A and B).
We injected the C85 cells expressing DHFR-Luc fusion protein and HT1080 DIL cells subcutaneously in nude mice. After tumors became palpable, tumor-bearing animals were treated with saline or MTX (5 mg/kg), and noninvasive bio-luminescence imaging was carried out. Signal intensities from ROI were quantitated using Kodak MI (Figure 6, A–C). Similar to the in vitro observations, in DHFR-Luc-expressing tumors, on MTX treatment, the luciferase levels peaked at 48 hours and dropped back to normal, whereas in the untreated group, there was no significant change in luciferase intensity (see Figure 6A). Imaging the DIL tumor at 0, 24, 48, 72, 96, and 120 hours following MTX injection indicated that luciferase was induced and luciferase intensity increased progressively up to 120 hours after drug treatment in all of the animals of the MTX-treated group (see Figure 6, B and C). To test the feasibility to control a “second” gene expression in a bicistronic construct, beyond 96 hours, we included an additional time point up to 120 hours. In tumors of animals imaged at the 120-hour time point after the MTX induction, the induction in luciferase activity was sustained up to the 120-hour time point (see Figure 6, B and C). ROI was quantitated using the Kodak image station 2000MM. Luciferase expression was increased in all of the animals treated with MTX compared with the untreated animals. The control animals were randomized to control for luciferase and tumor size without MTX treatment up to 120 hours (see Figure 6, B and C). This additional piece of evidence, which further strengthened the results from the in vitro studies, shows the feasibility to control a “second” gene expression for longer periods of time, up to 120 hours, and further strengthens the biologic and clinical significance of this study.
Here we present a strategy not only to induce expression of proteins downstream of IRES element in a bicistronic construct containing DHFR cDNA but also to modulate the induction by MTX and rapamycin treatment. We demonstrate that the expression of the partner protein can be selectively induced and not DHFR using rapamycin pretreatment followed by MTX treatment.
Given that bicistronic vectors containing IRES elements allow coexpression of multiple gene products, 25 they hold promise for correction of multiple gene disorders as well as complex disorders such as cancer and infectious diseases by augmentation of two or more proteins. 26 However, a major limitation with available vector systems using IRES elements is that expression of the downstream gene is significantly less efficient than that of the upstream one. 7,8,25 Use of DHFR-based bicistronic vectors provides a tool to induce any protein of interest in a bicistronic construct. This may also be an alternate strategy to express and induce more than one drug resistance protein exogenously. It has been previously shown that coexpression of a multidrug resistance (MDR1) and HSVTK gene in a bicistronic retroviral vector allows selective killing of MDR1-transduced cells. Similarly, suicide gene products linked to DHFR via an IRES element can be increased by MTX treatment, making suicide gene-prodrug therapy more effective (eg, HSVTK and ganciclovir or CD and 5FC).
A key finding of this study is that DHFR not only regulates its own expression but also expression of partner protein separated by an IRES element in a bicistronic construct. Addition of DHFR protein inhibited IRES-mediated induction of luciferase gene in an in vitro translation reaction. It is known that DHFR binds its cognate mRNA and autoregulates its expression. From our data, one can speculate that DHFR may inhibit the CMV-IRES activity in bicistronic vectors, resulting in the inhibition of expression from the second cistron. One can also speculate that DHFR, by binding to DHFR mRNA in the first cistron, can influence IRES-mediated translation of the partner protein located in the second cistron. Hence, it seems that in bicistronic constructs, DHFR suppresses the expression of the partner protein. However, this suppression can be overcome by MTX treatment, and this can be a useful strategy to modulate and/or induce the expression of the partner protein. By altering the position of DHFR in bicistronic constructs (DHFR-IRES-TS and TS-IRES-DHFR), we also demonstrated that levels of partner protein in a bicistronic construct can be induced regardless of the position of DHFR. This also suggests that the interaction of DHFR with IRES element could be due to DHFR being located next to the IRES element. However, further studies are required to better understand this interaction.
The m7G-cap-dependent translation of cellular mRNAs is initiated by recruitment of ribosomes to the 5′ end of mRNAs via eukaryotic translation initiation factor 4F (eIF4F), a cap-binding protein (eIF4E), and an RNA helicase (eIF4A) bridged by a scaffolding molecule (eIF4G). Internal translation initiation mediated by IRESs bypasses the requirement for the cap and eIF4E (Figure 7). Thus, it is counterintuitive to propose that inhibition of translation, imposed by DHFR proteins binding to its cognate mRNA in the bicistronic construct, will inhibit IRES-mediated translation of the second protein. However, if one considers a circular form for the translation complex, then it is conceivable that by physically tying up the translation complex either upstream or downstream of IRES, translation of the bicistronic mRNA comes to a halt. Addition of MTX removes the bound DHFR protein and allows resumption of translation, both cap dependent and independent. The availability of eIF4E controls the switch between cap-dependent and IRES-mediated translation. 27 Rapamycin pretreatment followed by MTX treatment inhibited induction of the first protein, which is cap dependent. However, the second protein downstream of IRES was still induced and remained upregulated for up to 96 hours even when cap-dependent translation was inhibited in the cell. It has been shown that cellular IRESs allow the expression of certain mRNAs in conditions in which cap-dependent translation initiation is severely impaired. 28 Hence, we conclude that antifolate-mediated translational upregulation of the second protein in the bicistronic construct is cap independent and driven by the IRES element.
It has also been shown that IRES elements are less regulated in the cell 24 and levels of downstream protein poly (A)-binding protein (PABP) interaction with elF4G stimulate IRES-dependent translation. 29 The interaction between PABP to elF4G may also lead to a higher level of a protein downstream to IRES. This may explain our observation that the downstream protein remains induced for a longer period of time on MTX treatment compared with the upstream protein.
As DHFR levels in the cell can be induced multiple times on MTX treatment, we propose that the expression of DHFR partner protein in bicistronic construct can be induced multiple times on MTX treatment. However, further studies are required to validate this point. In this proof-of-principle study, owing to the limitations of our xenograft model (owing to the increasing tumor size, the animals were euthanized after 5 days of MTX treatment), we were unable to follow the induction of the second cistron beyond 5 days. However, if the induction of the cistron can be extended to greater times, beyond 5 days, this approach to modulate levels of protein in DHFR-based bicistronic vectors may find application in the field of gene therapy to augment levels of any desired protein.
We propose to use the DHFR/MTX system to induce multiple genes in polycistronic vectors. Shorter synthetic IRES elements have been reported, some as short as 30 nucleotides. 30,31 Moreover, shorter IRES elements may allow detailed investigation into the role of individual bases in the IRES element in the antifolate-induced translational regulation. Multiple genes can be placed in polycistronic vectors using shorter IRES elements to circumvent size restrictions in vectors (ie, DHFR-IRES-A-IRES-B-IRES-C-IRES-D). Once characterized, these polycistronic vectors loaded with multiple genes may be used to augment levels of multiple defective genes to treat complex disorders such as cancer and infectious diseases and may also have applications in myelo-protection and for suicide gene therapy.
We thank Anna Jurkiewicz for her technical assistance.
Financial disclosure of authors: This work was supported by grants, National Institutes of Health and National Cancer Institute (NIH/NCI), CA 86438 and CA 08010.
Financial disclosure of reviewers: None reported.