Maxizymes and Small Hairpin-Type RNAs That Are Driven by a tRNA Promoter Specifically Cleave a Chimeric Gene Associated with Leukemia in Vitro and in Vivo¹

APL⁴ is characterized by the presence of the t(15;17) translocation, which generates a chimeric gene that encodes the oncoprotein PML-RAR α. RAR α is a nuclear hormone receptor whose target genes are involved in fundamental developmental processes and the terminal differentiation of myeloid hematopoietic progenitors (1). The PML gene encodes a putative tumor suppressor protein that is involved in the control of apoptosis. The fusion protein contains large portions of both the PML and RAR α proteins (2, 3, 4). It includes the DNA-binding domain of RAR α, the ligand-binding domain of RAR α, and the dimerization domain of RAR α and PML. The presence of these various domains apparently allows the PML-RAR α fusion protein to exert a dominant negative effect on the functions of RAR α and/or PML protein (2, 3, 5, 6, 7). Patients with APL can be treated effectively with an inducer of cell differentiation, namely, all-trans-retinoic acid, and complete remission is often achieved. However, such remission is transient, and relapsed APL often has all-trans-retinoic acid resistance. Therefore, consolidation (or intensification) therapy with high-dose chemotherapy or total-body irradiation followed by stem cell transplantation during the remission period is thought to be essential for a cure. However, it is impossible or risky to perform such transplantation in patients without a HLA-matched donor or in elderly patients. Thus, there is a need for an alternative therapy that is generally applicable and safe.

Catalytic RNAs, in particular, hammerhead ribozymes, have been used in attempts to suppress the expression of PML-RAR α, but tested ribozymes had cleavage ability that lacked the necessary specificity (8, 9, 10), probably because hammerhead ribozymes can only cleave their target RNAs at a NUX triplet, where N is A, G, C, or U and X is A, U, or C (11). When there is no NUX triplet near the junction site of the target chimeric mRNA, the ribozyme will cleave not only the chimeric mRNA but also the normal mRNA.

To overcome the problem of target specificity, we constructed an allosterically controllable ribozyme that we designated a maxizyme [minimized, active, x-shaped (heterodimeric), and intelligent (allosterically controllable) ribozyme] that functions as a dimer (12). The maxizyme cleaves its target RNA only when it recognizes a specific RNA sequence, which does not have to include a NUX triplet (Fig. 1). Thus, the maxizyme can, theoretically, cleave any RNA specifically.

The phenomenon known as RNAi was first recognized in the studies of antisense RNA in Caenorhabditis elegans, in which it was demonstrated that dsRNA was more effective than antisense RNA in the suppression of gene expression (13). Although dsRNA-induced RNAi has been shown to be a powerful tool for the silencing of gene expression in invertebrates, obstacles have been encountered in attempts to apply this technology to mammalian cells. In particular, dsRNAs of more than approximately 30 nts induce antiviral responses that result in the nonspecific degradation of RNAs (14). However, siRNA duplexes of 21 nts were found to trigger RNAi in a sequence-specific manner in human cells (15). Furthermore, it has recently been reported that shRNAs can also inhibit gene expression specifically (16, 17, 18, 19).

In this study, we examined the suppression of the expression of a chimeric gene for PML-RAR α by maxizymes and by shRNAs.

All RNA molecules used for assays in vitro were synthesized by a DNA/RNA synthesizer (model 394; PE Applied Biosystems, Foster City, CA). Reagents for RNA synthesis were purchased from Glen Research (Sterling, VA).

To evaluate the cleavage activities and specificity of maxizymes in vitro, we chose recognition sites for maxizymes as follows: R_{PML4-RAR α mRNA} (5′-UUG-ACC-UGC-CAU-UGA-GAC-3′) and R_{PML6-RAR α mRNA} (5′-GGA-GGC-AGC-CAU-UGA-GAC-3′), corresponding to PML-RAR α chimeric mRNA, and R_{RAR α mRNA} (5′-CCA-UCC-CCA-GCC-ACC-AUU-GAG-AC-3′), corresponding to normal RAR α mRNA. The substrate, an oligomer that included a site for cleavage by maxizymes, was 5′-³²P-labeled S_{RAR α} (5′-UGC-UUU-GUC-UGU-CAG-GAC-3′), in which the NUX triplet GUC is underlined. The sequences of maxizymes were as follows: MzR, 5′-GUC-UCA-AUG-AUC-GAA-ACA-AAG-CA-3′; Mz4L, directed against PML(exon 4)-RAR α, 5′-GUC-CUG-ACA-CUG-AUG-AGA-UGC-AGG-UCA-A-3′; and Mz6L, directed against PML(exon 6)-RAR α, 5′-GUC-CUG-ACA-CUG-AUG-AGA-UGC-UGC-CUC-C-3′. Each reaction mixture contained 50 mm Tris-HCl (pH 8), 10 mm MgCl₂, 2000 cpm 5′-³²P-labeled substrate, each recognition site (1 μm), MzR (0.1 μm), and MzL (Mz4L or Mz6L; 0.1 μm). Each reaction was carried out under enzyme-saturating (single-turnover) conditions. Reactions were initiated by the addition of MgCl₂, and each resultant mixture was then incubated at 37°C. An aliquot of a solution that contained 7 m urea, 100 mm EDTA, 0.1% xylene cyanol, 0.1% bromphenol blue, and 30% glycerin that was equal in volume to the reaction mixture was added to the reaction mixture to stop the reaction. Incubation time was 20 min when MzR and Mz4L were added, and it was 5 min when MzR and Mz6L were added. Reaction mixtures were fractionated by electrophoresis on a 20% polyacrylamide/7 m urea gel, and relative levels were calculated from their radioactivities, measured with an imaging analyzer STORM 830 system (Molecular Dynamics, Sunnyvale, CA).

Chemically synthesized oligonucleotides encoding maxizymes and shRNAs were converted to double-stranded sequences by PCR. After digestion with appropriate restriction enzymes, maxizyme fragments and shRNA fragments were cloned downstream of the tRNA^Val promoter of pV (which contains the chemically synthesized promoter of a human gene for tRNA^Val between the restriction enzyme sites of the pMX puro vector; Ref. 20). The sequences of the constructs were confirmed by direct sequencing.

Cells that stably expressed human PML-RAR α mRNA were obtained by retroviral transduction of HeLa cells. The p105^{PML-RAR α} retroviral vector was produced by transfection of φNX-A packaging cells with pMX-neo-p105^{PML-RAR α} that encoded PML-RAR α mRNA. HeLa/p105^{PML-RAR α} cells were selected and maintained in MEM supplemented with 10% fetal bovine serum (Life Technologies, Inc., Rockville, MD) and 1 mg/ml G418 (Life Technologies, Inc.).

Cells were washed with PBS and resuspended in MEM, and then total RNA was isolated with ISOGEN (Nippon Gene, Toyama, Japan) according to the manufacturer’s protocol. RT-PCR was performed with a RNA PCR kit (AMV) Version 2.1 (TaKaRa, Kyoto, Japan) in accordance with the manufacturer’s protocol. The primers used were 5′-ATG-CTG-TGC-TGC-AGC-GCA-T-3′ as 5′ primer and 5′-CCA-TAG-TGG-TAG-CCT-GAG-GAC-3′ as 3′ primer. Products of PCR were visualized after electrophoresis in a 2% agarose gel.

Transfections with plasmid DNA were performed using Effectene (Qiagen, Hilden, Germany) in accordance with the manufacturer’s protocol. HeLa/p105^{PML-RAR α} cells (1 × 10⁵) were cultured in each well of 6-well plates. After 24 h, the cells were transfected with 2 μg of maxizymes or shRNAs using Effectene reagent (Qiagen). After 48 h of incubation, cell lysates were harvested.

Cell lysates were subjected to SDS-PAGE on a 10% polyacrylamide gel. Rabbit polyclonal antibodies against RAR α (Santa Cruz Biotechnology, Santa Cruz, CA) and PML (MBL, Nagoya, Japan) were used to detect RAR α, PML, and PML-RAR α in HeLa/p105^{PML-RAR α} cells. To examine the toxicity of the tRNA-shRNA, we used an antibody against phosphorylated eIF2 α (United Biomedical, Inc., CA, USA) and eIF2 α (Cell Signaling Technology, MA, USA). As a control for phosphorylation of eIF2 α, cells were treated with poly(I)-poly(C) solution (100 μg/ml) for 8 h. Blocking and detection were performed as described previously (21).

We chose the promoter of a human gene for tRNA^Val (12, 22, 23, 24), which is transcribed by RNA polymerase III to generate expression plasmids for maxizymes and for shRNAs. This promoter allows the maxizymes and shRNAs to be expressed constitutively and at high levels in vivo, as requested for potential exploitation of maxizymes and shRNAs as therapeutic agents. We embedded sequences that encoded individual monomeric units downstream of this promoter. The resultant high-level expression of these RNAs promotes the dimerization of maxizymes, and, moreover, maxizymes and shRNAs attached to a tRNA can be transported efficiently from the nucleus to the cytoplasm, where RNAi appears to occur exclusively (19, 25). Moreover, most of the target mRNAs should also be located in the cytoplasm.

There are three types (bcr1, bcr2, and bcr3) of chimeric gene for PML-RAR α as a result of disruption of chromosome 15 within three clusters of breakpoints in the gene for PML. Each chimeric gene for PML-RAR α is expressed as differently spliced PML-RAR α transcripts because of alternative splicing of the PML portion (26). In this study, the target was the bcr1 form of the chimeric gene for PML-RAR α, in which intron 6 of the gene for PML is fused to intron 2 of the gene for RAR α (Fig. 2). The chimeric gene generates a variety of transcripts, but it is the sequence of the junction site of these transcripts that is critical to our experiments.

Pandolfi et al. demonstrated that the bcr1 variant generates two junction sites (Fig. 2; Ref. 26): at one site, exon 4 of the gene for PML is fused to exon 3 of the gene for RAR α; and at the other, exon 6 of the gene for PML is fused to exon 3 of the gene for RAR α. We designed maxizymes and shRNAs that targeted these two transcripts, respectively. The constructs encoded maxizyme right directed against RAR α region (MzR); maxizyme left directed against PML exon 4-RAR α exon 3 (Mz4L); maxizyme left directed against PML exon 6-RAR α exon 3 [Mz6L (note that MzR was common to the maxizymes directed against both of the two targets)]; shRNA directed against the junction of PML exon 4-RAR α exon 3 (sh 4); and shRNA directed against the junction of PML exon 6-RAR α exon 3 (sh 6).

The maxizymes were designed to adopt an active conformation only in the presence of the abnormal PML-RAR α junction, whereas they should remain inactive in the presence of the normal RAR α, mRNA and in the absence of the abnormal PML-RAR α junction (Fig. 1 A). These features were designed to maximize substrate specificity.

We also constructed vectors for the expression of shRNAs with a double-stranded region of 30 nts in length and a loop of 5 nts (Fig. 1 B). Sequences corresponding to junction sites in mRNAs for PML-RAR α were located in the middle of the double-stranded region.

To examine conformational changes and cleavage activities of maxizymes in vitro, we prepared a short 18-mer substrate derived from the mRNA for RAR α (nt 364–381) that corresponded to the target (cleavage) site. We incubated the synthetic maxizymes with the 5′-³²P-labeled short 18-mer substrate in the presence and absence of the 18-mer effector molecule derived from PML-RAR α mRNA (the junction sequence) or in the presence of a 23-mer that corresponded to normal RAR α mRNA. In this reaction, the 18-mer PML-RAR α effector molecule that corresponds to the junction sequence in PML-RAR α mRNA should act in trans. The maxizyme must recognize the 18-mer PML-RAR α effector molecule via interactions of the sensor arms of maxizyme left (Mz4L or Mz6L) and MzR to adopt the conformation of the active dimer (Fig. 1,A). The other recognition arms of the maxizyme must recognize the 18-mer RAR α substrate RNA sequence that includes a NUX triplet. Then the catalytically indispensable Mg²⁺ ions can be captured in the Mg²⁺-binding pocket, and specific cleavage should occur (Fig. 1,A). As shown in Fig. 3, products of cleavage (a 9-mer) were clearly detected in the presence of the sequence that corresponded to the PML-RAR α junction. According to the relative levels of their radioactivities of substrates, 44% of the substrates were cleaved within 20 min at 37°C, under single-turnover conditions, in the presence of 0.1 μm each of Mz4L and MzR. Similarly, 55% of substrates were cleaved in the presence of 0.1 μm each of Mz6L and MzR within 5 min at 37°C. By contrast, products of cleavage were almost undetectable in the absence of sequence that corresponded to the PML-RAR α junction (no effector) or in the presence of the sequence derived from the normal RAR α mRNA, demonstrating a high level of specificity, as anticipated. It should be emphasized that, according to our previous studies, the maxizyme is more active and more specific in vivo (in culture cells) than in vitro (27, 28).

Our results demonstrated the allosteric control of the activity of the maxizyme in vitro, in accordance with the predicted conformational changes (depicted in the bottom panels in Fig. 3) that should occur in the presence of the effector molecule (the PML-RAR α junction).

We next examined the activity of the maxizyme against an exogenous PML-RAR α mRNA target and compared this activity with the activity of tRNA-shRNA directed against the same target mRNA. RNAi is a process of dsRNA-mediated sequence-specific gene silencing in animals and plants. Particularly, siRNAs and shRNAs can induce RNAi-mediated gene silencing efficiently in mammalian cells. For expression of shRNAs in mammalian cells, we used the tRNA promoter that promotes cytoplasmic localization of transcripts because it has been known that RNAi occurs in the cytoplasm (19). In addition, our tRNA-shRNA is processed by RNase III Dicer efficiently and shows significant suppression of expression of target mRNAs in the cytoplasm (19).

At first, we established stable transformants of HeLa cell line (HeLa/p105^{PML-RAR α}) that expressed human PML-RAR α mRNA and confirmed the expression of PML-RAR α mRNA by RT-PCR. As noted above, the splicing variants of PML-RAR α transcripts give rise to a variety of products of RT-PCR. We anticipated the formation of 420-mer and/or 679-mer and/or 823-mer products of RT-PCR with our primers. As shown in Fig. 4, the bands corresponding to the 420-mer product and 679-mer product were clearly detected in the lane labeled HeLa/p105^{PML-RAR α} plus (+) RT. By contrast, no bands were detected in the lane labeled HeLa plus (+) RT and in lanes labeled RT minus (−), demonstrating that HeLa/p105^{PML-RAR α} expressed PML-RAR α transcripts specifically.

Next, we examined whether expression of the gene for PML-RAR α could be silenced by the maxizyme or by shRNA in HeLa/p105^{PML-RAR α} cells. We transfected these cells with plasmids that encoded only tRNA^Val (mock), the tRNA maxizyme (Mz4L, Mz6L, and MzR), or tRNA-shRNA (sh 4 and sh 6), and then we performed Western blotting analysis with RAR α (Fig. 5,A) or PML (Fig. 5,B)-specific antibodies. As evidenced clearly from Fig. 5, the level of PML-RAR α protein in cells that expressed either the maxizyme or shRNA was significantly lower than that in cells that expressed only the gene for tRNA^Val. Neither the maxizyme nor the shRNA affected expression of the normal gene for RAR α and PML. These results indicated that both the maxizyme and shRNA were able to suppress the expression of the chimeric gene for PML-RAR α specifically.

Moreover, to examine whether the tRNA-shRNA shows toxicity in cells, we examined whether the tRNA-shRNA induces an IFN-mediated antiviral response. It is known that long dsRNAs induce IFN-mediated antiviral responses that lead to the shutdown of protein synthesis (29). In this response, PKR is activated by dsRNAs, and then activated PKR phosphorylates eIF2 α, which promotes the repression of a general translation. Then we examined the level of phosphorylation of eIF2 α using Western blotting analysis with a specific antibody against phosphorylated eIF2 α and eIF2 α in HeLa/p105^{PML-RAR α} cells. As shown in Fig. 6, phosphorylation of eIF2 α was hardly detected in HeLa/p105^{PML-RAR α} cells and cells that expressed tRNA-shRNAs (sh4 and sh6). By contrast, the eIF2 α was phosphorylated in cells that were treated with poly(I)-poly(C). These results suggest that tRNA-shRNA does not induce nonspecific repression of a general translation and can possibly be used as specific anti-PML-RAR α agents.

In this study, to suppress the expression of a PML-RAR α chimeric gene, we constructed a tRNA-driven maxizyme and tRNA-driven shRNA directed against the chimeric mRNA.

The maxizyme is indeed useful due to the sensor function, and it has been demonstrated that maxizymes can be designed for specific cleavage of several sequences of interest (12, 30, 31, 32). In addition, the expected conformational change was demonstrated by a chemical probing of the complex (33). In this study, we were able to demonstrate both the specificity and the strong cleavage activity of maxizymes targeted against the junction site in vitro and in vivo.

To utilize RNAi as a therapeutic agent, it is important to develop a vector system that can express dsRNAs efficiently. At present, many groups have developed RNAi expression systems that utilize the U6 or H1 promoter in human cells (16, 17, 18, 34, 35, 36, 37, 38). In this study, we constructed a tRNA^Val-based system for expression of dsRNA directed against PML-RAR α mRNA to ensure high-level expression and transport of the tRNA-linked product to the cytoplasm. Our previous research demonstrated that a hammerhead ribozyme that was embedded downstream of a human tRNA^Val promoter was transported to the cytoplasm and had a high cleavage activity (23, 39, 40). It appears that RNAi occurs only in the cytoplasm, which is also the site of most target mRNAs. The results shown in Fig. 5 demonstrate that the maxizymes and shRNAs that were driven by tRNA^Val promoter dramatically inhibited the expression of the chimeric protein. Theoretically, some bands for chimeric proteins in Fig. 5 should be detected because of splicing variants, but in practice, only one band for PML-RAR α can be detected (26).

In connection with this, the reason the 823-mer product of RT-PCR in Fig. 4 was not detected is probably because the splicing variant that corresponded to the 823-mer band was not expressed as much compared with the other splicing variants.

If this tRNA^Val-based system for expression of dsRNA is applied to gene therapy, lentivirus vectors might be a promising candidate for gene delivery because they can deliver genes at high efficiencies to a wide variety of human blood cells and can integrate the transgenes into the host genome, resulting in stable expression (41, 42, 43, 44).

Many leukemias are caused by chromosomal translocations. The bcr-abl fusion gene is the hallmark of chronic myeloid leukemia. Recently, it has been reported that chemically synthesized siRNAs reduce the expression of this chimeric gene (45). We showed here that not only shRNA but also the maxizyme was able to inhibit the expression of a chimeric oncogenic protein, whereas normal proteins were unaffected. To our knowledge, this is the first example of the suppression of the expression of a chimeric gene through DNA vector-based RNAi. Our research demonstrates that both the maxizyme and shRNA can silence the expression of a chimeric gene when encoded by vectors with a tRNA^Val promoter, suggesting that it might be possible to apply these systems to gene therapy in a clinical setting.

We thank Dr. Iichiro Takata for the design of maxizymes and technical assistance.