|Home | About | Journals | Submit | Contact Us | Français|
Although human islet transplantation is a promising approach for treating type I diabetes, its success is limited due to the poor survival rate of transplanted islets. Expression of a growth factor gene to promote revascularization and silencing of proapoptotic genes before transplantation may improve the outcome of islet transplantation.
In this study, we constructed bipartite plasmid vectors to co-express a VEGF cDNA and shRNA targeting iNOS gene. Firstly, we screened shRNA sequences against human iNOS by transfecting plasmids encoding shRNA targeting different start sites of human iNOS. Then, the effect of different promoters (such as H1, U6 and CMV) and miRNA backbones on gene silencing was determined. No statistical difference in iNOS gene silencing was observed for the shRNA with H1, U6 and CMV promoters. In addition, a conventional shRNA showed better silencing of iNOS gene, compared to shRNA containing mir375 and mir30 backbones. A bipartite plasmid was also constructed with mir30-shRNA and a VEGF cDNA controlled by a single CMV promoter. This plasmid showed a better silencing effect compared with plasmid without VEGF cDNA.
In this study, we have successfully constructed bipartite vectors co-expressing a VEGF cDNA and a shRNA against iNOS gene. These vectors could be an attractive candidate to improve the survival of transplanted islets.
RNA interference (RNAi) is an evolutionarily conserved biologic process that regulates gene expression by small interfering double stranded RNA (siRNA) mediated sequence-specific, post-transcriptional gene silencing [1-3]. Several steps are involved in RNAi: 1) long double-stranded RNA (dsRNA) is processed by DICER into 19-23 base pair siRNA duplex; 2) siRNA duplex is incorporated into a complex named RNA-induced silencing complex (RISC); 3) RISC is activated by eliminating passenger strand of siRNA duplex, and results in mRNA degradation or translational repression .
RNAi has been extensively used as a tool for gene silencing in cell lines and animal models [5, 6]. In addition, it is also a promising approach for treating various acquired and genetic diseases [7, 8]. Recently, we have tested the effects of siRNAs against inducible nitric oxide synthase (iNOS) genes on insulin producing β-cell line (INS-1E) and on human islets [9, 10]. iNOS gene silencing with chemically synthesized siRNA decreased NO production in INS-1E rat β-cells and human islets, reduced pro-inflammatory cytokine induced β-cell death, and partially protected the human islet function . We also observed in another study that adenovirus-based shRNA against caspase-3 gene (Adv-caspase-3-shRNA) efficiently silenced caspase-3 gene and its gene silencing effect lasted beyond five days, which resulted in the protection of islets from cytokine-induced apoptosis .
As an alternative cure for type I diabetes, human islet transplantation has made a great progress, especially after the success of Edmonton protocol [12, 13]. However, a large number of the patients returned to insulin dependent within a year after islet transplantation . The main reason for the failure of islet transplantation is that less than 30% of the transplanted islets are survived in the early days post-transplantation due to several reasons, including (1) poor revascularization, which results in insufficient supply of oxygen and nutrition; and (2) proinflammatory cytokines induced islet β-cell death including apoptosis and necrosis . In our previous study, we investigated the use of bipartite plasmid or adenoviral vectors which co-express one gene for reducing inflammatory response (e.g. hIL-1Ra) and another for facilitating revascularization (e.g. HGF, VEGF) [16, 17]. This combinatorial approach showed synergistic effect by working on two independent therapeutic targets. Therefore, it is possible to observe improved therapeutic efficacy with bipartite vectors that co-express a vascular endothelial growth factor (VEGF) cDNA and shRNA targeting an apoptotic gene such as iNOS or caspase-3.
To co-express two genes in a single vector, we can use two separate expression cassettes driven by two promoters or using internal ribosome entry site (IRES) sequence [18, 19]. The use of IRES is usually not preferred, since the gene after IRES usually showed lower expression level . Co-expression of a shRNA and a gene with two separate promoters is similar to that of two different genes, but it might be different for co-expression of a shRNA and a gene with a single promoter. Unlike gene expression plasmid which utilizes a RNA polymerase II (Pol II) promoter, shRNA expression can utilize both Pol II promoters (such as CMV promoter) and RNA polymerase III (Pol III) promoters (such as H1, U6 promoters). IRES sequence is not necessary for co-expressing multiple shRNA or combination of a shRNA and a cDNA within a single promoter [21-23]. It is has also been reported that co-expression of shRNA and cDNA can be realized by inserting a promoterless shRNA within the intron of a gene .
A bipartite vector, which can effectively silence a target gene and expresses a therapeutic gene properly, is in great need for treating various diseases including diabetes, cancer and viral infection. siRNA targeting sequences and the type of promoters used will greatly influence gene silencing efficiency. A shRNA and cDNA could be expressed by two different promoters or by a single promoter. It was also reported that the miRNA based shRNA showed improved gene silencing than conventional shRNA [25, 26]. In this study, we have systemically investigated these factors to find a most optimal bipartite vector and indentify parameters defining a potent shRNA and gene co-expression vector.
Fetal bovine serum (FBS) was purchased from Mediatech, Inc (Herndon, VA). Penicillin/streptomycin, phosphate-buffered saline (PBS), 0.25% (w/v) trypsin-EDTA and DMEM medium were purchased from GIBCO-BRL (Gaithersburg, MD). All oligonucleotides used for shRNA cloning were obtained from Integrated DNA Technology (Coralville, IA). All the enzymes used in cloning were purchased from New England Biolabs (Ipswich, MA).
Negative control shRNA plasmid and Human iNOS cDNA plasmid (p-iNOS) were purchased from OriGene (Rockville, MD). Plasmids encoding shRNA targeting five different regions of iNOS gene were purchased from Open Biosystems (Huntsville, AL). To generate pH1-shiNOS-CMV-GFP, p-U6-shiNOS, and pCMV-shiNOS, two shRNA oligonucleotides (synthesized by IDT DNA) were annealed and cloned into pRNAT-H1.1/shuttle (BamHI, HindIII), pSIREN-Shuttle (BamHI, EcoRI) and p-shuttle2(XbaI+AflII), respectively. pH1-shiNOS-CMV-VEGF was generated by replacing GFP gene with VEGF PCR fragment from pCMV-VEGF165. Briefly, pH1-shiNOS-CMV-GFP was digested with PflmI, followed by treating with Klenow enzyme. After purification, it was digested with NheI and then purified. VEGF gene was amplified with PCR using following primer: Forward (Nhe I): GCCTAGCTAGCTAGATGAACTTTCTGCTGTCTTG; Reverse (DraI): CGCTATTTAAATCACCGCCTCGGCTTGTCACATC. To make pU6-shiNOS-CMV-VEGF, pSIREN-shuttle was digested with I-ceu and Bam HI, and the fragment with U6 promoter was sub-cloned into I-Ceu and Bam HI site in pH1-shiNOS-CMV-VEGF. pU6-mir375-shiNOS was generated by cloning annealed mir375 oligonucleotides into pSIREN-shuttle (BamHI, EcoRI). To make pU6-mir30-shiNOS, mir30-shiNOS sequence was synthesized and sub-cloned into pSIREN-Shuttle (BamHI, EcoRI). mir30-shiNOS sequence:GGATCCGTGCTCGCTTCGGCAGCACATATACTAGTCGACTAGGGATAACAGGGTAATTGTTTGAATAGGCTTCAGTACTTTACAGAATCGTTGCCTGCACATCTTGGAAACACTTGCTGGGATTACTTCTTCAGGTTAACCCAACAGAAGGGCGGCCGCAAGGTATATTGCTGTTGACAGTGAGCGCGTGTATTTAACTGCCTTGTGTAGTGAAGCCACAGATGTACACAAGGCAGTTAAATACACATGCCTACTGCCTCGTCTAGAAAGGGGCTACTTTAGGAGCAATTATCTTGTTTACTAAAACTGAATACCTTGCTATCTCTTTGATACATTTTTTGaattc. To make pCMV-mir30-shiNOS, mir30-shiNOS sequence was amplified with PCR using the following primer: forward (Dra I): GTATTTAAAGGATCCGTGCTCGCTTCGGC; reverse (Afl II): CGCCTTAAGAATGTATCAAAGAGATAGCA and PCR product was cloned into p-shuttle2 (DraI, AflII) after restriction enzyme digestion. pCMV-VEGF-mir30-shiNOS was made by inserting VEGF PCR fragment between CMV promoter and mir30 shiNOS sequence in pCMV-mir30-shiNOS using NheI and DraI sites. shRNA sequences are listed in supplementary information. All the plasmids were purified by Promega mini-prep kit and confirmed by DNA sequencing.
AD-293 cells were cultured in DMEM supplemented with 10% FBS, 100 U/ml penicillin, and 100 mg/ml streptomycin. In transfection experiments, cells were seeded in a 48- well plate at a density of 20,000 cells/well 24 hours before transfection. Then, 0.05μg iNOS cDNA plasmid and 0.3 μg shRNA plasmid were co-transfected into cells with Fugene HD transfection reagent (Roche applied science, Indianapolis, IN).
Human iNOS and VEGF expression were determined at mRNA level by real time RT-PCR. Following treatment, total RNA was extracted with RNeasy® Mini KIT and treated with DNase by on-column digestion (RNase-Free DNase Set). RNA concentration was determined by spectrophotometer (NanoDrop®). One hundred and seventy nanograms total RNA was converted into cDNA using multiscribe reverse transcriptase reagents and random hexamers at a 10 μl reaction system. Two microliters of cDNA were used as a template and analyzed by SYBR Green-I dye universal PCR master mix on LightCycler® 480 Instrument. The primers used for real-time PCR were as follows: human iNOS: Forward 5′-ACGTGCGTTACTCCACCAACA-3′; Reverse 5′-CATAGCGGATGAGCTGAGCA-3′ (amplicon size 102bp); human ribosomal protein S19 (human S19); Forward 5′ GCTTGCTCCCTACCGATGAGA-3; Reverse 5′-ACCCCGGAGGTACAGGTG-3′ (amplicon size 73bp) . The primer for VEGF gene was the same as previously described . To assess the specificity of the amplified PCR product, the Melting Curve Analysis was performed on a LightCycler® 480 Instrument. The results at iNOS mRNA level were compared by calculating the CP value and normalized by the reference genes (human S19).
Nitric oxide (NO) is rapidly oxidized in culture medium into nitrite, which accumulates in the sample and can be easily correlated with NO production. Therefore, nitrite concentration was determined using the Griess assay (Promega, Madison, MI). Fifty microliters of cell culture supernatant was added to a 96 well-plate and mixed with 50 μL of 1% sulfanilamide in 5% phosphoric acid solution and incubated for 5 min at room temperature in the dark. Then, 50 μL of 0.1% N-1-naphthylethylenediamine dihydrochloride (NED) aqueous solution was added to each well. The plate was incubated for an additional 10 min, and absorbance was measured at 560 nm using a microplate reader. To determine the nitrite concentration in each sample, a standard curve was prepared using nitrite standard solution and culture medium as matrix.
At 36 hours after transfection of AD-293 cells with 0.3μg/well plasmids, culture medium of AD 293 cells were collected. The VEGF concentration in the culture medium was measured by ELISA according to the manufacture's protocol (R& D systems, Minneapolis, MN).
Targeting sequence is one of the most important elements for effective gene silencing. Therefore, we first screened five plasmids encoding shRNA targeting different regions of human iNOS gene. The targeting sequences of different plasmids are listed in Figure 1A. To find the most potent targeting sequence, plasmids expressing shRNA against five different iNOS mRNA regions were co-transfected with piNOS into AD 293 cells. Plasmid with scrambled sequence was used as a control. At 24 hours after transfection, NO production from iNOS gene expression was measured as nitrite by Griess Assay. As demonstrated in Figure 1B, all five shRNA expression plasmids were able to silence iNOS expression and among them the 4060 showed 66% reduction of NO production (p<0.01). Therefore, the targeting sequence 4060 was used in the following experiments for shRNA or miRNA based shRNA construction.
Promoter is an important regulatory element in gene expression. Among several promoters, H1, U6 and CMV promoters are most widely used for shRNA mediated gene silencing. Since some discrepancies have been reported for the efficiency of these promoters [26, 27], we compared the efficiency of these three promoters for shRNA mediated iNOS gene silencing by constructing iNOS shRNA expression plasmids with H1, U6 and CMV promoters (Figure 2A). After co-transfection with piNOS plasmid into AD 293 cells, we found that compared to the control group, iNOS shRNA expression plasmids could effectively (p<0.01) reduce the NO production by 69, 62, and 60% for H1, U6 and CMV, respectively (Figure 2B). No statistical difference in iNOS gene silencing was observed among the shRNA expression vectors driven by these promoters. This suggests that there is no significant difference among these three promoters in iNOS gene silencing.
Co-expression of shRNA and therapeutic proteins is a promising combinatorial RNAi strategy , which will have synergistic effect by acting on different targets. However, it is not sure whether the over-expression of VEGF gene under a strong CMV promoter will interfere with the expression and processing of shRNA. As illustrated in Figure 3A, a bipartite plasmid has been constructed to co-express VEGF cDNA and iNOS-shRNA. VEGF cDNA was under the control of CMV promoter, while iNOS-shRNA was driven by U6 promoter. As shown in Figure 3B, compared to the control plasmid group, pU6-shiNOS-CMV-VEGF and pU6-shiNOS reduced the NO production by 59% and 62% (p<0.01), respectively. In contrast, the plasmid expressing VEGF did not show significant reduction of NO production. It was also noticed that there was no significant difference between pU6-shiNOS and pU6-shiNOS-CMV-VEGF for NO production. Therefore, the co-expression of VEGF cDNA and iNOS-shRNA will not have much interference with the shRNA-mediated gene silencing.
Efficient gene silencing has been reported when shRNA was embedded in miRNA backbone [26, 29]. To find out the possibility of increasing iNOS gene silencing effect by shRNA with miRNA backbone, two shRNA were designed based on mir375 and mir30 structures (Figure 4A). As shown in Figure 4C, significant reduction of NO production were observed in all of three vectors tested, pU6-shiNOS (reduced by 62%, p<0.01), pU6-miR375-shiNOS (reduced by 39%, p<0.01) and pU6-miR30-shiNOS (reduced by 30%, p<0.01). In addition, pU6-shiNOS is significantly more efficient than pU6-miR375-shiNOS or pU6-miR30-shiNOS (p<0.01). Gene silencing efficiency of these three shRNA vectors was also determined at mRNA levels by real-time PCR (Figure 4B). All of these three vectors reduced NO mRNA expression significantly (p<0.01, compared with control). In addition, pU6-shiNOS (reduced by 56%) is significantly more efficient than pU6-miR375-shiNOS (reduced by 30%) or pU6-miR30-shiNOS (reduced by 28%) (p<0.05, compared with pU6-shiNOS).
We have shown the successful co-expression of iNOS shRNA and VEGF cDNA under two expression cassettes. To further investigate the possibility of co-expressing VEGF cDNA and iNOS shRNA under a single promoter, a plasmid with mir30-shiNOS under the control of a CMV promoter was constructed and a VEGF cDNA was inserted between CMV promoter and mir30-shiNOS (Figure 5A). As shown in Figure 5B, a 39% reduction in NO production was achieved with pCMV-mir30-shiNOS (p<0.01, compared with control plasmid), which is close to that of pU6-mir30-shiNOS (30%). However, the gene silencing effect was significantly enhanced by inserting a VEGF cDNA between CMV promoter and mir30-shRNA. In contrast, 61% reduction in NO production was achieved with p-CMV-VEGF-mir30-shiNOS (p<0.01, with p-mir30-shiNOS).
Not only potent silencing of pro-apoptotic gene, but also sufficient VEGF expression is essential. Therefore, we measured VEGF gene expression from two bipartite plasmids, pU6-shiNOS-CMV-VEGF and pCMV-VEGF-mir30-shiNOS. We measured VEGF gene expression by ELISA of the cell culture medium. VEGF concentration was around 65ng/ml for all above plasmids and there is no significant difference among these two plasmids (Figure 7B). The VEGF gene expression from these two bipartite plasmids was also measured at mRNA levels with real-time PCR using VEGF gene specific primers (Figure 6A). The results indicated VEGF mRNA levels in these two plasmids treated groups were 914 ± 18 and 922 ± 35 folds high than that in control group, respectively.
The selection of potent targeting sequence that leads to effective gene silencing still remains the key issue for practical application of RNAi technique for disease therapy. Web-based computer programs with different algorithms for designing siRNA and shRNA are available. To increase the chance of finding a potent sequence, empirical rules usually incorporated in these algorithms. These empirical rules includes: thermodynamic property; length of siRNA target; GC content; and RNA secondary structure [30-32]. However, the selection of targeting sequence is still an empiric process and depends on experimental screening of potential targets. This is because our understanding of RNAi mechanism is still insufficient. In addition, most of the algorithms for shRNA sequence design are actually that of siRNA design and convert the siRNA into shRNA sequence. In our study, we have converted potent siRNA targeting sequence into shRNA sequence with success [7, 11]. However, we also observed loss of silencing effect after converting potent siRNA sequence into shRNA (data not show). Therefore, a better understanding of RNAi mechanism is required for improved design of shRNA. In this study, five shRNA expression plasmids were purchased and screened to find the potent shRNA targeting sequence. In practical, this might be an efficient way to find a potent shRNA for certain application.
Promoter is another important element that determines the duration, intensity and specificity of gene expression [33, 34]. The activity of several promoters including tRNA, H1, U6, CMV, LTR, CMV enhancer/H1 has been investigated [23, 26, 35]. In this study, we did not observe significant difference in the levels of iNOS gene silencing when CMV, U6, and H1 promoters were used for driving shRNA-iNOS. Similar results have been reported by other research groups, where they found that when a less efficient luciferase shRNA sequence was used, there were some differences in luciferase gene silencing among different promoters; with H1 promoter being less efficient than CMV and U6 promoters. However, when a potent luciferase shRNA sequence was used, the difference in gene silencing effect among different promoters became minimal .
The use of miRNA-shRNA in siRNA expression vector is an attractive strategy, which mimics the structure of natural occurring miRNA and thus can be processed by cellular miRNA machinery more efficiently compared with conventional shRNA . Almost 80% more effective in reducing HIV p24 antigen production was achieved with TAT shRNA using miR-30 backbone compared with conventional shRNA . shRNA is expressed in high level and yields an abundance of precursor, while miRNA-based shRNA is expressed at low levels but processed more efficiently . To further improve the silencing effect, we designed two shRNAs based the mir375 and mir30 sequences. The gene silencing potency of mir375-shRNA and mir30-shRNA was compared with that of conventional shRNA. However, the shRNA showed best gene silencing effect (62% reduction of NO production), while mir375 or mir30 shRNA are less potent with 39% and 30% reduction of NO production, respectively. Similar findings have also been reported by other groups [27, 37]. Li et al. have compared conventional shRNA and miR30-based shRNAs against luciferase gene or mouse tyrosinase. Among 14 different targeting sequences against luciferase tested, in 11 sequences of the conventional shRNA showed better silencing effect than miR30-based shRNA. All the conventional shRNA with 10 different targeting sequences against mouse tyrosinase showed significantly better silencing effect than miR30-based shRNA counterpart. These authors explained that the shRNA structure used previously for comparison  which has a 29-nt stem and a 4-nt loop is not an optimal design for shRNA. When a 4-nt loop sequence was used in shRNA, an insufficient processing of shRNA by Dicer was observed which resulted in poor gene silencing . However, when a 9-nt loop (UUCAAGAGA) and 19-nt stem structure was used in shRNA design, a better gene silencing was observed in conventional shRNA compared with mir30-based shRNA. Therefore, it is reasonable to observe better gene silencing effect in shRNA used in our study which has a 21-nt stem and 6-nt loop (CTCGAG). Boudreau et al. also demonstrated in their work that optimized shRNAs are more potent than mir30-based shRNAs for silencing of three genes including GFP, SCA 1 and HD, when the variables were minimized in the comparison . For our study, the ultimate goal was to develop a vector for therapeutic purpose, thus the conventional shRNA which showed better iNOS gene silencing effect will be used in our future study.
Another important feature of miRNA based shRNA is that they are more amendable to pol II transcription and polycistronic strategies, allowing delivery of multiple siRNA sequences (or a siRNA sequences and a cDNA) with a single promoter [21, 22, 38]. We first designed a miRNA30 based shRNA plasmid p-CMV-mir30-shiNOS, but it was not as potent as we expected. However, when we inserted a VEGF cDNA between CMV promoter and mir30-shiNOS sequence (p-CMV-VEGF-mir30-shiNOS), the gene silencing effect was significantly increased. This result is consistent with the work of Stegmeier et al., who reported that the insert of GFP, dsRED and Neo genes between CMV promoter and mirR30-shRNA cassette increased the knock down of Rb gene. The exact reason for this is still unknown. Probably certain space between CMV promoter and mir30-shRNA is necessary for efficient gene silencing. The p-CMV-VEGF-mir30-shiNOS could be another bipartite vector for therapeutic application, which co-expresses shRNA and a cDNA with a single promoter.
Co-transfection approach for evaluating gene silencing effect has been extensively reported [25, 27, 36, 37], which allowed researchers to measure the gene silencing effect with convenience and reliability. Therefore, a co-transfection of piNOS plasmid was used to determine the gene silencing effect of our newly constructed vectors encoding shRNA targeting different start sites of iNOS, promoters and backbones. The use of a stably transfected cell line has also been reported . A C6 cell stably expressing firefly luciferase gene has been used to study the gene silencing effect of shRNAs vectors against firefly luciferase gene. However, this is not appropriate for iNOS gene silencing, since iNOS is up-regulated only after stimulation with cytokines or other reagents. By using co-transfection approach, the iNOS mRNA was expressed from piNOS plasmid after transfection. This could mimic the scenario of cytokine induced iNOS gene much better than iNOS stable expressing cell lines, in which iNOS gene was constantly over-expressed. In addition, there are several limitations for the use of stable cell lines: 1) the establishment of a stable cell line usually takes a long time (at least several weeks); 2) a cDNA expression vector with an antibiotics resistance gene (e.g., neomycin, puromycin); 3) only a gene that is not toxic or will not interfere with the vital cellular process could be used to make a stable cell line. Because of above mentioned reasons, a co-transfection method would be a good approach for evaluation gene silencing effect during the development of gene silencing vectors.
The purpose of this study was to construct bipartite plasmids that could co-express VEGF gene for promoting revascularization and shRNA against iNOS gene to reduce human islet β-cell death. This combinatorial strategy will help to improve the survival and function of islet graft by promoting revascularization and inhibiting apoptosis . The reason for using bipartite plasmids is that it will reduce the use of total plasmid backbone, thus minimizing the immunogenic effect caused by bacterial plasmids . In addition, the bipartite plasmids were specially designed as a shuttle plasmid for constructing adenoviral vectors. Moreover, two restriction enzyme sites have been preserved flanking shRNA sequence in our bipartite vector. This feature will allow us to replace the shRNA sequence with a new one through directional cloning. We could conveniently construct pU6-shCas3-CMV-VEGF by replacing shRNA sequence in pU6-shiNOS-CMV-VEGF and caspase-3 gene silencing was observed (data not shown.).
In conclusion, we have constructed bipartite plasmids which co-express VEGF cDNA and shRNA against iNOS. VEGF cDNA and shRNA were driven by two different promoters or by one single promoter. These plasmids are properly designed, which allows us to conveniently change shRNA sequence. In addition, all of these plasmids could be used as a shuttle plasmid for producing adenoviral vectors.
We would like to thank the National Institute of Health (NIH) for financial support (RO1 DK69968).