Skip to main content
NCBI home page
Molecular Therapy logoLink to Molecular Therapy
. 2016 Aug 2;24(9):1581–1591. doi: 10.1038/mt.2016.138

Advanced Design of Dumbbell-shaped Genetic Minimal Vectors Improves Non-coding and Coding RNA Expression

Xiaoou Jiang 1, Han Yu 1, Cui Rong Teo 1, Genim Siu Xian Tan 1, Sok Chin Goh 1, Parasvi Patel 1,2, Yiqiang Kevin Chua 1, Nasirah Banu Sahul Hameed 3, Antonio Bertoletti 3, Volker Patzel 1,*
PMCID: PMC5113105  PMID: 27357627

Abstract

Dumbbell-shaped DNA minimal vectors lacking nontherapeutic genes and bacterial sequences are considered a stable, safe alternative to viral, nonviral, and naked plasmid-based gene-transfer systems. We investigated novel molecular features of dumbbell vectors aiming to reduce vector size and to improve the expression of noncoding or coding RNA. We minimized small hairpin RNA (shRNA) or microRNA (miRNA) expressing dumbbell vectors in size down to 130 bp generating the smallest genetic expression vectors reported. This was achieved by using a minimal H1 promoter with integrated transcriptional terminator transcribing the RNA hairpin structure around the dumbbell loop. Such vectors were generated with high conversion yields using a novel protocol. Minimized shRNA-expressing dumbbells showed accelerated kinetics of delivery and transcription leading to enhanced gene silencing in human tissue culture cells. In primary human T cells, minimized miRNA-expressing dumbbells revealed higher stability and triggered stronger target gene suppression as compared with plasmids and miRNA mimics. Dumbbell-driven gene expression was enhanced up to 56- or 160-fold by implementation of an intron and the SV40 enhancer compared with control dumbbells or plasmids. Advanced dumbbell vectors may represent one option to close the gap between durable expression that is achievable with integrating viral vectors and short-term effects triggered by naked RNA.

Introduction

Inefficient delivery and gene expression represents a major obstacle to applications of therapeutic nucleic acids ex vivo and in vivo. To support gene delivery, researchers use nonviral or viral vectors. While nonviral delivery systems comprising non-nucleic acid-based helper functions often are complex and toxic to the cells, viral vectors are efficient but costly and their clinical application is limited by the risk of uncontrolled vector integration as well as innate or pre-existing immune responses due to viral pre-exposition of the patients.1,2,3 The most straightforward naked nucleic acid-based strategies are advantageous in terms of safety and cost but applications suffer from low delivery efficiencies. While RNA-based vectors trigger only short-term expression due to the short half-life of RNA in living cells, DNA-based vectors are more stable and can trigger prolonged transient expression as well as long-term expression following genomic integration under certain conditions. Despite the advantages of cheap and easy manufacture, plasmid-based vectors harbor major disadvantages: Firstly, the bacterial backbone can trigger side effects including immunotoxicity, antibiotic resistance, and transgene silencing; secondly, the large size hampers cellular and nuclear delivery; and thirdly, plasmids are difficult to chemically modify. Thus, there is a pressing need to develop gene delivery vectors which are safe, efficient, and affordable. Two minimalistic DNA-based vector systems hold great promise in this regard: DNA minicircles and dumbbell-shaped DNA vectors (db-vectors), both of which were shown to improve recombinant gene expression in vitro and in vivo.4 Minicircles represent circular double-stranded gene expression cassettes, whereas the smaller db-vectors comprise a linear double-stranded expression cassette that is covalently closed at both ends with single-stranded loop structures.5,6 These vectors are devoid of unnecessary sequences including the bacterial plasmid DNA backbone or antibiotic resistance genes and the small vector size is assumed to be advantageous in terms of delivery and sensing by DNA length-dependent innate immune receptors.7,8,9,10,11 Both, minicircles and db-vectors circumvent the disadvantages associated with plasmid vectors and can trigger short- to medium-term expression even in human primary cells, in which plasmid-driven gene expression is rapidly silenced.12 Compared with minicircles which require a minimum size of at least 300 base pairs (bp) due to circular tension, db-vectors have no lower size limit and are just as large as the expression cassette of interest.13 Genes for the expression of small RNAs are about 150 bp in length which is far below the minimum size of plasmids or even the smallest minicircle DNA. This implies that, with regard to small RNA expression, these vectors are 2- to 20-fold larger than necessary. In addition, dumbbells but not minicircles or plasmids can easily be linked to various functional groups such as fluorophores, cell-penetrating peptides, nuclear localization signals, or immune stimulatory peptides via the loop structures.4,14 Dumbbell vectors for coding RNA expression and decoys have been investigated for several years in preclinical and clinical studies without showing adverse side effects or toxicity.15,16 Recent in vitro and in vivo studies indicated improved safety profiles for db-vectors as compared with plasmids. Moreover, the length of extragenic spacers, which can be found in most plasmids and minicircles but not in db-vectors, was described to correlate with transgene silencing in primary cells.17,18,19,20,21 Dumbbells have not been explored for small RNA expression and researchers adhere to the use of viral vectors which are considered to involve additional risks to achieve long-term expression; or, alternatively, use chemically synthesized modified RNA which is costly and may cause cell toxicity, to trigger short-term effects.22

Here we describe an advanced design of db-vectors for noncoding or coding RNA expression. The advantage of the db-vector system of having no lower size limit becomes most apparent for small RNA-expressing vectors. To fully exploit this advantage, we designed minimized hairpin template-transcribing dumbbells of only 130 or 151 bp in length for small hairpin RNA (shRNA) or precursor microRNA (pre-miRNA) expression, respectively. These vectors were generated with high conversion yields using a novel protocol that combines polymerase chain reaction (PCR) amplification with stem-loop ligation. The minimized dumbbells exhibited improved kinetics of delivery and transcription and triggered enhanced target gene knockdown as compared with plasmids or “linear” dumbbells, i.e., dumbbells harboring linear expression cassettes. The implementation of simian virus 40 (SV40)-derived enhancer elements which were reported to promote active nuclear import of DNA, significantly improved the activity of both, small RNA and protein-expressing dumbbells in a cell type-dependent manner.23,24 The integration of a spliceable intron known to facilitate RNA processing and nuclear RNA export enhanced gene expression in all investigated cells lines.25 Our data indicate that small size represents an important advantage of the db-vector system. However, distinct molecular features which at first glance seem to conflict with the paradigm of size minimization can strongly improve dumbbell-shaped genetic vectors.

Results

Design of minimized db-vectors for shRNA and pre-miRNA expression

We investigated db-vectors as a safe and cheap vector system to trigger transient expression of small noncoding RNAs. Size minimization represents a common theme in genetic vector development and is considered an important aspect with regard to improved delivery and reduced innate immune sensing. Therefore, we sought to minimize the dumbbell size as much as possible by employing three strategies: Firstly, shRNAs and pre-miRNAs form self-complementary hairpin structures. In the consequence, linear expression cassettes contain redundant sequences as the sense and antisense portions forming the stem are included in both strands of the DNA duplex. We eliminated redundant sequences and in the minimized dumbbells, part of the DNA vector structure resembles the structure of the transcribed hairpin RNA (Figure 1a). In such designed vectors RNA transcription goes around the hairpin template including one of the loops of the db-vector. “Hairpin template”-transcription allowed us to shorten shRNA-expressing vectors by about 30–40 bp and miRNA-expressing vectors by 60 bp or more. Secondly, we implemented the minimal human H1 (mH1) promoter.26 The mH1 promoter is 99 bp in length, 128 bp shorter than the full-length H1 promoter, and has not found applications in db-vectors until recently.27 We compared the activities of the mH1, the full-length H1, and the CMV promoters in the context of plasmid vectors. Both the real-time reverse transcription PCR (rtRT-PCR) quantification of mature miR-125b-1 and miR-125b-1-triggered knockdown of a firefly luciferase miR-125b-1 sensor construct indicated that the mH1 promoter was as active as the full-length H1 or CMV promoter (Supplementary Figure S1). Thirdly, we designed an integrated transcriptional promoter/terminator element. Therefore we replaced nonessential sequence positions at the 3ʹ end of the mH1 promoter by a restriction site needed for vector production and an inverted polymerase III transcriptional terminator, i.e., an adenosine pentamer (A5), to terminate hairpin template-transcription in the opposite strand, shortening the vectors by 5 bp (Figure 1a,b). Transcription of this modified promoter starts downstream of the inverted termination signal so that its sequence is not added to the 5ʹ end of the transcript leading to the formation of hairpin structures with UU 3ʹ overhangs. We cannot exclude that substitution of nonessential promoter sequences by an adenosine pentamer prior to the transcriptional start may affect the transcriptional start position; however, that would not matter in our constructs as the guide strand lies in the 3ʹ arm of the hairpin. Together, the novel features allow reducing the size of shRNA- or miRNA-expressing dumbbells by 171 bp (57%) or 191 bp (56%) enabling the production of small RNA-expressing dumbbells as short as 130 bp.

Figure 1.

Figure 1

Open in a new tab

Design, generation, and functional investigation of novel minimized hairpin template db-vectors for small RNA expression. (a) Structures of small hairpin RNA (shRNA) expression vectors. Upper two: conventional plasmid p-iPR-linear-s/as and dumbbell db-iPR-linear-s/as vectors with linear shRNA expression cassettes and integrated promoter-restriction endonuclease site element (iPR). Lower vector: minimized hairpin template (hp) dumbbell harboring an iPRT element. R indicates a restriction overhang ligation site. Loops L1 and L2 are (T)4 tetra loops. (b) Design of the modified promoter elements. mH1, wild-type minimal H1 promoter; iPR, integrative promoter/restriction element; iPRT, integrative promoter/restriction/terminator element. (c) Novel protocol for the generation of minimized small RNA-expressing dumbbells. The promoter is polymerase chain reaction-amplified using a 5ʹ-phosphorylated forward primer introducing a Nb.Bpu10I nicking enzyme cleavage site (NE) and a reverse primer introducing a sticky end producing endonucleolytic cleavage site (RE). After enzymatic cleavage, the upstream loop forms by intramolecular ligation of the overhang. Ligation of a chemically synthesized hairpin structure-forming oligonucleotide completes the dumbbell structure. (d) Agarose gel electrophoresis analyses of the ligated dumbbells before (−) and after (+) exonuclease treatment. Different variations of the protocol trigger different conversion yields. Variant (1): no purification step; variant (2): binding of an antisense oligonucleotide to the released single-stranded cleavage product suppresses religation; variant (3): small enzymatic cleavage products are removed by gel permeation chromatography; variant (4): combination of variants (2) and (3).

Novel protocol for efficient generation of novel minimized shRNA-expressing dumbbells

We developed a novel protocol to generate hairpin template-transcribing db-vectors (Figure 1c). Therefore, the mH1 promoter sequence is amplified by PCR using primers introducing upstream a Nb.Bpu10l nicking site and downstream a BamHI cleavage site (Supplementary Table S1). After Nb.Bpu10l/BamHI cleavage, the upstream loop consisting of four Ts is generated by intramolecular ligation of the Nb.Bpu10l 5ʹ overhang. The sequence coding for small RNA expression is concurrently added by intermolecular ligation of an oligomeric DNA hairpin structure. This new protocol combines features of two dumbbell production techniques that were reported earlier.28,29,30 However, with our novel universal protocol, changing the expression cassette does not involve any cloning since alternative hairpin templates can be ligated to the same dumbbell core structure. The conversion yields for the new protocol, i.e., the amount of the covalently closed dumbbell after exonuclease treatment divided by the amount of dumbbell before exonucleolytic digestion, approximate 90% when the small fragments resulting from endonucleolytic cleavage are functionally neutralized with a complementary sequence and/or removed by column purification prior to loop ligation (Figure 1c,d).

Minimized dumbbells exhibit improved kinetics of delivery, small RNA expression, and target gene knockdown

Earlier studies reported superior delivery and transcriptional activity triggered by db-vectors compared with plasmid DNA.31 First time we investigated the kinetics of cellular and nuclear delivery, transcriptional activity, and knockdown triggered by dumbbell-driven shRNA expression in comparison with equimolar amounts of equivalent plasmid vectors. HepG2 cells were co-transfected with the firefly luciferase reporter vector pGL3-Control and equimolar amounts of either a db-vector harboring a linear expression cassette, a minimized hairpin template-transcribing dumbbell, or a plasmid, all expressing the same luciferase-targeting shRNA (Figure 1a). After 10 minutes, 1, 6, and 24 hours, total episomal DNA, nuclear episomal DNA, or total small RNA were isolated and either the mH1 promoter DNA or the shRNA was quantified by rtPCR or rtRT-PCR, respectively (Figure 2a–c). To directly monitor transcription, the rtRT-PCR was designed to exclusively detect the shRNA precursor but not the processed antisense shRNA guide strand. Target gene expression was not detectable 10 minutes or 1 hour post-transfection and knockdown was instead monitored at 6, 12, and 24 hours (Figure 2d). The kinetics indicate a clear advantage of the dumbbells over the corresponding plasmid in terms of cellular delivery and nuclear targeting and consequently also in terms of shRNA transcription (Figure 2). While cellular dumbbell delivery was enhanced up to 5-fold (P < 0.05) (Figure 2a), nuclear delivery was enhanced up to 74- or 25-fold (P < 0.001) 10 minutes or 24 hours post-transfection compared with the plasmid (Figure 2b), pointing toward an accelerated rate of dumbbell diffusion from the cytoplasm into the nucleus. Accordingly, shRNA transcription increased 12- or 7-fold (P < 0.001) 10 minutes or 24 hours post-transfection (Figure 2c). Finally, the db-vectors triggered significantly stronger luciferase knockdown compared with the plasmid vector (Figure 2d). An advantage of the minimized dumbbell as compared with the “linear” dumbbell was most evident (3.7-fold enhancement; P < 0.001) in terms of the transcriptional activity 10 minutes after transfection and was also indicated by 11–14% improved knockdown activities. At 10 minutes post-transfection, cellular delivery levels had reached about 6% (plasmid and “linear” dumbbell) or 3% (minimized dumbbell) of the 24 hours levels, and the transcriptional levels about 5/2/8% (plasmid/“linear” dumbbell/minimized dumbbell) (Figure 2a,c). Nuclear delivery on the other hand increased by three orders of magnitude from 10 minutes to 24 hours (2.8 × 103/0.8 × 103/1.2 × 103-fold for plasmid/“linear” dumbbell/minimized dumbbell) (Figure 2b). Notably, the absolute vector copy numbers detected in the whole cell versus the nuclear extracts cannot be directly compared since we added feeder cells to isolate the nuclei and copy numbers refer to 10 ng of total RNA.

Figure 2.

Figure 2

Open in a new tab

Kinetics of dumbbell delivery and expression. (a) Cellular dumbbell delivery, (b) nuclear dumbbell delivery, (c) dumbbell-driven small hairpin RNA (shRNA) expression, and (d) shRNA-dumbbell-triggered knockdown of the firefly luciferase target gene. HepG2 cells were transfected with equimolar amounts (1 pmol/well) of either a db-vector, linear (db-iPR-linear-s/as) or hairpin (db-iPRT-hp-s/as) template, or plasmid p-iPR-linear-s/as, all expressing the same luciferase targeting shRNA. 10 minutes, 1, 6, and 24 hours post-transfection, total episomal DNA (a) or nuclear episomal DNA (b) was isolated. Vector copy numbers, i.e., copy numbers of the mH1 promoter sequence, in the cells were quantified using SybrGreen rtPCR (a) or a TaqMan-probe-based rtPCR (b). (c) Total RNAs were also extracted and the relative expression of the luciferase-targeting shRNA was quantitated using rtRT-PCR. (d) To monitor luciferase knockdown, HepG2 cells were co-transfected with the firefly luciferase reporter vector pGL3-Control (250 ng/well), and equimolar amounts (0.12 pmol/well) of dumbbell vectors or plasmid as described above and firefly luciferase activities were measured 6, 12, and 24 hours post-transfection. Error bars indicate the mean ± SEM (n = 2 (a); n = 3 (b–d)). Significance was tested using one-way analysis of variance with Newman–Keuls post hoc test. No significant difference was detected between the linear and the hairpin dumbbell groups at 1, 6, and 24 hours (a–c) and at 10 minutes (b). *P < 0.05, **P < 0.01, or ***P < 0.001. rtRT-PCR, real-time reverse transcription polymerase chain reaction.

Minimized pre-miRNA-expressing dumbbells trigger improved target gene knockdown

The minimized db-design along with the new protocol for db-generation is applicable for shRNA- and pre-miRNA-expressing dumbbells. We designed a human pre-miR-125b-1-expressing minimized dumbbell (db-hp-miR-125b-1), a “linear” dumbbell (db-linear-miR-125b-1), and a plasmid vector (p-linear-miR-125b-1), the latter two both harboring linear pre-miR-125b-1 expression cassettes (Figure 3a and Supplementary Figure S2a). miR-125b-1 was reported to function as tumor-suppressor miRNA.32,33 All three vectors expressed the miR-125b-1 precursor structure and to achieve that, various mismatches and bugles had to be implemented into the design of db-hp-miR-125b-1. To efficiently sense miR-125b-1 expression, we constructed a sensor plasmid by replacing the firefly luciferase 3ʹUTR of plasmid pMIR-Report with a repeat of three miR-125b-1 binding sites. The sensor vector detected both endogenous as well as plasmid- and dumbbell-driven miR-125b-1 overexpression in HepG2 cells (Supplementary Figure S2b). miR-125b-1 overexpression was additionally monitored in HEK293T and CL48 cells (Supplementary Figure S2c,d). Next, we compared luciferase knockdown triggered by equal amounts of DNA in terms of mass (equimass amounts) of the different miR-125b-1-expressing vectors. The db-vectors, in particular the minimized dumbbell, triggered 3.9-fold (P < 0.05) stronger target gene knockdown as compared with the plasmid vector (Figure 3b). Compared with the linear dumbbell, the minimized dumbbell triggered about twofold stronger knockdown. When comparing equimolar amounts however, the dumbbells were only slightly better than the plasmid and we did not observe a difference anymore between the linear and the minimized dumbbells. A possible explanation might be that the mismatches within the minimized dumbbell slightly impair the transcriptional activity of the hairpin-structured expression cassette though mismatches per se can render a higher vector flexibility which possibly facilitates trafficking through the nuclear pores as evidenced recently.27 Both, our new protocol and the conventional enzymatic ligation assisted by nucleases (ELAN) method produced functionally equivalent minimized dumbbells (Figure 3b).30

Figure 3.

Figure 3

Open in a new tab

Design and functional investigation of novel minimized db-vectors for microRNA (miRNA) expression. (a) Pictorial diagram showing the design of miR-125b-1-expressing dumbbells and the transcribed miR-125b-1 precursor structure. The DNA template coding for mature miR-125b-1 is highlighted in magenta, the template for miR-125b-1* in cyan. db-linear-miR-125b-1, conventional linear dumbbell; db-hp-miR-125b-1, minimized hairpin template dumbbell; db-mEnh-hp-miR-125b-1, minimal SV40 enhancer containing hairpin template dumbbell; db-fEnh-hp-miR-125b-1, full-length SV40 enhancer containing hairpin template dumbbell. Loops that are not further specified represent (T)4 tetra loops. (b) Sensing of plasmid versus dumbbell-driven miR-125b-1 overexpression in HepG2 cells. Cells were transfected in 24-well plates either with 250 ng sensor plasmid pMIR-125b-1-Sensor alone or together (+) with 250 ng miR-125b-1-expressing plasmid (p-linear-miR-125b-1), equimass amounts of dumbbells (db-linear-miR-125b-1 or db-hp-miR-125b-1) or equimolar amounts of dumbbells, the latter which were topped up with feeder DNA for transfection to 250 ng. Dumbbell db-hp-miR-125b-1 was generated either using the conventional enzymatic ligation assisted by nuclease (ELAN) method or our (New) protocol. (c) The implementation of a minimal SV40 enhancer (mEnh) and more pronounced of a full-length SV40 enhancer (fEnh) significantly improves hairpin dumbbell-driven miR-125b-1-triggered luciferase knockdown in HepG2 cells transfected with equimolar amounts (1.5 pmol) of db-vectors. (b,c) Luciferase knockdown was monitored 48 hours post-transfection; values were standardized relative to the sensor control. (d) rtPCR detection of vector/trigger copy numbers in primary human T cells at different time points after nucleofection of hairpin dumbbell db-hp-miR-125b-1, linear dumbbell db-linear-miR-125b-1, plasmid p-linear-miR-125b-1, miR-125b-1-mimic, or luciferase-targeting control dumbbell db-iPRT-hp-s/as. Indicated are vector copy numbers/10 ng isolated nucleic acids. Percentages indicate relative intracellular copy numbers at day 10 compared to day 1 (100%). (e) miR-125b-1-triggered knockdown of the core-binding factor beta subunit mRNA in primary human T cells determined using rtRT-PCR. Day 1 refers to 24 hours post-transfection (d,e) and fold change refers to buffer transfected cells (1.0) at the respective days. Error bars indicate the mean ± SEM (n = 3–6 (b,d,e); n = 4 (c)). Significance was tested using one-way analysis of variance with Newman–Keuls post hoc test. *P < 0.05, **P < 0.01, or ***P < 0.001. rtRT-PCR, real-time reverse transcription polymerase chain reaction.

SV40 enhancer elements significantly improve target gene knockdown triggered by minimized miRNA-expressing dumbbells

Fast passive diffusion from the cytoplasm to the nucleus represents a key feature of db-vectors. We investigated whether db-vector-driven gene expression can be further improved by adding an active nuclear import component. Distinct promoter or enhancer sequences were reported to significantly enhance plasmid-driven gene expression. Examples are the SV40 enhancer sequence, the smooth muscle γ-actin promoter, and the origin of replication of the Epstein–Barr virus (oriP), whereby the latter's activity depends on the expression of the viral nuclear antigen 1.23,24,34,35,36 The proposed mechanism is that transcription factors harboring nuclear localization signals bind to these sequences mediating active nuclear co-import of the bound DNA via the protein import machinery.37 We implemented either the 237 bp full-length SV40 enhancer (fEnh) or a 72 bp minimal version (mEnh) of it into the minimized miR-125b-1-expressing dumbbell (Figure 3a).23 In HepG2 cells, strongest miR-125b-1-mediated luciferase knockdown (93%; P < 0.001) was triggered by the dumbbell harboring the fEnh, followed by the dumbbell with the mEnh (87%; P < 0.001), and the enhancer-negative control vector (69%; P < 0.001) (Figure 3c). These data demonstrate that SV40 enhancer elements can enhance db-driven miRNA expression 2.3-fold (mEnh) or 4.5-fold (fEnh) though their implementation triggers a substantial increase in vector size (47 or 156%).

Minimized miRNA-expressing dumbbells reveal higher stability and trigger stronger target gene knockdown in primary human T cells as compared with plasmids or miRNA mimics

While targeted delivery in vivo remains a difficult task, the ex vivo manipulation of primary cells including stem cells prior to reimplantation into a patient currently appears to be a more tangible approach. We investigated the kinetics of vector stability and miRNA-triggered target gene knockdown in primary human T cells ex vivo. Equimolar amounts (each 0.2 pmol) of the miR-125b-1-expressing dumbbell db-hp-miR-125b-1 (20 ng), plasmid p-linear-miR-125b-1 (400 ng), or a miR-125b-1 RNA mimic (2.8 ng) were delivered into 106 primary human T cells via nucleofection. With respect to the plasmid we also tested equimass amounts (400 ng/4 pmol) of dumbbell db-hp-miR-125b-1 and control dumbbell db-iPRT-hp-s/as as well as each 4 pmol (56 ng) of the miR-125b-1 mimic or a control mimic. The chosen DNA amount of 400 ng reflected the optimal amount recommended by the suppliers; the used amount of 4 pmol for the miRNA mimics was within the recommended range of 0.6–6 pmol RNA. At day 1, 3, 7, and 10 postdelivery, DNA vector or miRNA mimic copy numbers were quantified and knockdown of the Core-Binding Factor Beta subunit gene, a validated target of miR-125b-1, was monitored using rtPCR or rtRT-PCR, respectively (Figure 3).38 Notably, nucleofection only reaches a fraction of the targeted cells and a complete target gene knockdown is not to be expected. While the intracellular abundance of the plasmid vector or the miRNA mimic decreased about 29- or 15-fold from day 1 to day 10 post-transfection, the dumbbell appeared to be more stable showing an about 2-fold decay (Figure 3d). In primary T cells, both the plasmid and the miRNA mimic, the latter only at the highest test dose, triggered a weak but not significant target gene knockdown of 19% or 24% at day 1 which might be partly due to low intracellular stability. In addition, nucleofection, though being a very efficient delivery method, delivers a substantial amount of the miRNA mimic directly into the nuclei where RISC loading is not expected to occur. In case of the plasmid, transgene silencing might have impaired the knockdown as well. On the other hand, equimass amounts of the dumbbell vector were found to trigger significant core-binding factor beta subunit knockdown (56%; P < 0.05) at day 10 (Figure 3e).

The SV40 enhancer and/or a spliceable intron improves dumbbell-driven gene expression

Most db-vectors reported in the literature were designed to express functional proteins, and hence, much larger in size compared with the small RNA-expressing dumbbells discussed above. Large protein-expressing dumbbells—even more so than faster diffusing minimized small RNA-expressing dumbbells—are expected to benefit from an active nuclear import mechanism. We investigated as to whether the SV40 enhancer would be suitable to enhance protein expression driven by db-vectors of >2 kbp in length. As a second molecular feature to enhance protein expression, we implemented a spliceable intron. While the use of introns was reported earlier, SV40 enhancer elements have not found applications in dumbbells yet.39 Using the ELAN method, we generated a db-vector comprising a SV40 promoter-driven firefly luciferase gene and the SV40 polyadenylation site (db-luc) as well as variations of this parental vector harboring either the full-length SV40 enhancer (db-luc-enh), the human β-globin gene chimeric intron (db-int-luc), or both (db-int-luc-enh) (Figure 4a).30,40 For comparison we tested four plasmid vectors with the identical expression cassettes and features but which in addition contained the 2,855 bp pGL3-Control backbone. HEK293T or HepG2 cells were transfected with equimass amounts of the plasmid and db-vectors, and luciferase gene expression was monitored 48 hours post-transfection. In HEK293T cells, dumbbells triggered comparable or up to twofold higher levels of gene expression compared with the plasmids; in HepG2 cells, db-triggered expression was three- to sevenfold (P < 0.001) stronger (Figure 4b,c). While implementation of the intron enhanced gene expression of all constructs unconditionally in both cell lines, the SV40 enhancer was active in HepG2 cells but not in HEK293T cells. In HepG2 cells, implementation of the SV40 enhancer triggered a remarkable 16-fold (P = 0.005) or 27-fold (P < 0.001) enhancement of gene expression in the dumbbell group compared with the respective enhancer-negative dumbbells db-luc or db-int-luc. Highest levels of luciferase expression were triggered by dumbbell db-int-luc or db-int-luc-enh in HEK293T or HepG2 cells, respectively. In HepG2 cells, db-int-luc-enh-triggered gene expression was 7-fold higher (P < 0.001) than for the equally featured plasmid p-int-luc-enh, 56-fold higher (P < 0.001) than for the standard dumbbell db-luc, and 160-fold higher (P < 0.001) compared with the enhancer-negative pGL3-Control vector (p-luc). In HepG2 cells, the advantages of the db-vectors over the plasmid vectors are noticeable even when considering the ~twofold higher molar dumbbell amounts tested.

Figure 4.

Figure 4

Open in a new tab

Enhancement of dumbbell versus plasmid-driven luciferase expression by the β-globin gene chimeric intron and/or the full-length SV40 enhancer. (a) Design of dumbbell vectors. db-luc, parental vector; db-luc-enh, SV40 enhancer dumbbell; db-int-luc, intron dumbbell; db-int-luc-enh, intron-enhancer dumbbell. All dumbbell loops are (T)4 tetra loops. Luciferase expression triggered by equimass amounts of dumbbell vectors and equivalent plasmids in HEK293T (b) or HepG2 (c) cells 48 hours post-transfection. Error bars indicate the mean±SEM (n = 3 (b); n = 5 (c)). Significance was tested using one-way analysis of variance with Newman–Keuls post hoc test. *P < 0.05, **P < 0.01, or ***P < 0.001.

Discussion

We explored dumbbell-shaped DNA minimal vectors for small RNA expression, described the design of novel minimized db-vectors for shRNA or miRNA expression, and introduced a new method to generate such vectors. The novel hairpin template-transcribing dumbbells can be as short as 130 bp which is significantly shorter than any alternative genetic vector including dumbbells harboring linear expression cassettes. With the novel protocol, we achieved dumbbell yields of up to 90% and the generation of multiple vectors targeting different mRNAs does not require any cloning. Notably, the shortest of our vectors may also be assembled at a comparable cost by ligating two synthetic oligonucleotides without the need of PCR amplification.

Our study revealed significant advantages of dumbbell vectors over plasmids which were found to depend on the mode of delivery and the type of target cells. The experiments in which we transfected human tissue culture cells with equimolar amounts of vector DNA indicated an inverse correlation between the sizes of equally featured vectors and the kinetics of delivery and gene expression, i.e., minimized dumbbells were better than dumbbells with linear expression cassettes, and dumbbells were superior to the corresponding plasmid controls (Figure 2). It is difficult to understand why dumbbells were delivered more efficiently than plasmids under transfection conditions that were optimal for the amount of transfected plasmid DNA; however, that might be explained by different efficiencies of liposome formation. The observation that a 1,000-fold increase in nuclear delivery from 10 minutes to 24 hours post-transfection of any of the vectors only triggered an about 10-fold augmentation of the detectable RNA levels could be due to either (i) transcriptional vector inactivation, (ii) enhanced RNA degradation, (iii) codetection of DNA associated with the nuclei, or (iv) cytotoxicity and cell death in the consequence of an oversaturation of the cellular silencing machinery.41 Pathway oversaturation and cytotoxicity would be dose-dependent and more pronounced in case of the transcriptionally more active db-vectors. However, such a correlation was not observed indicating that cytotoxicity might not explain the disparity between the kinetics of nuclear vector delivery and transcriptional activity. As compared with the “linear” dumbbell, the minimized dumbbell exhibited a significantly enhanced transcriptional activity already 10 minutes after transfection started, though in the case of the minimized dumbbell, transcription has to go around the DNA hairpin structure which one would not consider advantageous. Hence, the enhanced transcriptional activity can most likely be assigned to accelerated nuclear diffusion assuming nuclear diffusion of the minimized dumbbell reaches its steady state in <10 minutes, and that the difference between the minimized and the “linear” dumbbell was not captured anymore even by the earliest investigated 10 minutes time point. In tissue culture cells, the advantage of the dumbbells over the plasmids could mainly be assigned to a higher transfection efficiency (Figure 2). Conversely, in human primary cells, the plasmid seemed to have an advantage over the dumbbells with regard to the efficacy of nucleofection as a substantially higher copy number of plasmid DNA could be detected at day 1 after nucleofection of equimolar DNA amounts (Figure 3d). On the other hand, dumbbells were much more stable in primary cells than plasmids as about half of the dumbbell DNA but <5% of the plasmid could be detected at day 10 after nucleofection. We explored minimized hairpin template-transcribing dumbbells for the expression of miRNA precursor structures. To achieve accurate pre-miRNA expression, we first time designed dumbbells harboring multiple mismatches and bulges which did not adversely affect the yields of vector production, vector activity, or stability. In primary human T cells, the mismatched miR-125b-1-expressing dumbbell vector was found to be substantially more stable as compared with a plasmid or a miRNA mimic. While it is not surprising that unmodified RNA harboring free 5ʹ and 3ʹ ends is relatively fast degraded, a high incidence of nicks or DNA damage may induce exonucleolytic degradation or cytotoxicity in the case of the large plasmid DNA vector. In this regard, the difference between the dumbbell and the plasmid becomes even more pronounced considering that dumbbells but not plasmids are exonuclease treated in the course of production yielding only covalently-closed exonuclease-resistant DNA. Notably, the same strategy of considering mismatches within the dumbbell structure does also allow expression of shRNAs that are extended with miRNA stems for improved processing. This idea was supported by an experiment in which we extended the hairpin template expression cassette of a luciferase-targeting shRNA by the stem of hsa-miR-30.42 The dumbbell harboring the miRNA stem triggered significantly stronger target gene knockdown in HEK293T or in HepG2 cells as compared with the control dumbbell (Supplementary Figure S3).

Besides our attempts to minimize db-vector size, we identified molecular features that improve the biological activity of dumbbells though they enlarge the vector size. One of these features were SV40 enhancer-derived sequences. The minimal or full-length SV40 enhancer increased the size of the minimized miR-125b-1-expressing dumbbell by 47% or 146% but enhanced target gene knockdown 2.3- or 4.5-fold (Figure 3c). In principle, SV40 enhancer sequences could account for both, transcriptional activation or active nuclear DNA import and retention. However, in this example it appears to be more likely that the SV40 enhancer mediates nuclear DNA import rather than enhancing the transcriptional activity of the heterologous polymerase III mH1 promoter.

In this study, we also designed advanced dumbbell vectors for coding RNA expression. The implementation of the full-length SV40 enhancer into large SV40 promoter-driven luciferase-expressing dumbbells and plasmids improved gene expression in HepG2 cells by more than an order of magnitude but had no effect in HEK293T cells (Figure 4b,c). This observation is consistent with earlier publications reporting that the activity of the SV40 enhancer, though it is not limited to one cell type, may vary between different cell lines depending on the availability of cell type-specific transcription factors.43,44 Our data even evidence that the implementation of the SV40 enhancer triggered a slight impairment of db-driven luciferase expression in HEK293T cells which could either be assigned to slowed diffusion of the larger enhancer-positive vectors; or alternatively, indicate that cell type-specific trans-acting factors in combination with the promoter sequence can turn the properties of the SV40 enhancer into those of a silencer as suggested earlier.45 As a second feature that enlarges the vector size we investigated a functional intron. The implementation of the β-globin intron enhanced gene expression of all dumbbells and plasmid vectors in both investigated cell lines (Figure 4b,c). That was expected as splicing facilitates nuclear mRNA export and subsequent gene expression in metazoan cells.25 In addition, the enhancement of gene expression in HepG2 cells indicates that the SV40 enhancer and the β-globin intron enhance gene expression synergistically. Whereas the enhancer or the intron alone improved gene expression of the dumbbell 16-fold or 2-fold, both together triggered an about 56-fold enhancement of gene expression. Splicing mainly occurs co-transcriptionally, and our observation is in line with reports suggesting a link between the polymerase II transcription machinery and the spliceosome.46

In this study, advantages of the dumbbells over the plasmids were found to be highly evident even when delivering equimolar amounts of vector DNA (Figure 2) but, in terms of target gene knockdown, were most pronounced when equimass amounts were tested (Figure 3b,e). Considering limitations associated with most delivery strategies with regard to maximally deliverable volumes and DNA masses or cytotoxicity triggered by liposomal or other complexing compounds, it can be regarded as a strong advantage of the db-vector system being low in molecular weight as higher copy numbers can be delivered as compared with larger minicircles or plasmids. In our in vitro system, small dumbbell size was highly advantageous in terms of improved nuclear delivery and it remains to be tested whether that goes along with a higher risk of genomic vector integration. In vivo, genetic vectors additionally have to manage extracellular transport including extravasation, diffusion through the extracellular matrix network, target cell binding, and internalization. To overcome these physical barriers, the small dumbbell vector size can be beneficial.

Another advantage of dumbbell vectors may be that they are less sensed by the innate immune system for the following reasons: Firstly, due to the small size noncoding RNA expressing dumbbells may remain less detected or undetected by length-dependent immune sensors or pathways including the DNA-dependent activator of IFN-regulatory factors,7,47 RIG-I-dependent sensing of RNA polymerase III-transcripts,8 absent in melanoma 2,48,49 or the cyclic GMP-AMP synthase (cGAS) which acts via the adaptor protein stimulator of interferon genes (STING).10,11 In addition, the banishment of bacterial DNA limits the frequency of unmethylated CpG dinucleotides to levels characteristic for eukaryotic DNA suppressing recognition by Toll-like receptor 9.50 Secondly, dumbbell vectors are purified by exonuclease treatment lacking of free ends and nicks. For this reason, together with the small size, dumbbells are expected to be rather inert in terms of type III interferon activation via the cytosolic DNA damage sensor Ku70.9 Thirdly, our miRNA-precursor/stem expressing dumbbells harbor mismatches within the stem of the dumbbell which would reduce sensing by the cGAS-STING pathway. This pathway is also activated by unpaired guanosines in Y-shaped or hairpin loop structures but all nontranscribed standard loops of our dumbbells were chosen to be (T)4 tetra loops.51 Notably, most of the DNA sensing pathways are cell-type dependent and we would not expect strong immune sensing in activated human T cells or the tissue culture cell lines we used for our studies.10,52,53 However, immune sensing would be relevant and has to be studied in the context of future in vivo applications where the vectors get in contact with myeloid immune cells.

Our data demonstrate that db-vectors trigger accelerated, enhanced, and prolonged transient RNA expression. Thus, in terms of small RNA and coding RNA delivery, the use of newly designed db-vectors may represent one option to close the existing gap between short-term effects triggered by naked RNA including siRNAs, miRNA mimics, or mRNAs and durable effects that can be achieved with integrating lentiviral vectors thus positioning these vectors for therapeutic applications that require profound transient expression in primary cells ex vivo or in vivo but in which vector integration is neither required nor desirable. Examples of such applications are RNA-guided genome editing, miRNA-triggered somatic cell reprogramming, suicide gene therapy, and genetic vaccination. An advanced molecular design may facilitate preclinical and clinical investigations of db-vectors. With regard to the delivery problem in vivo, it appears to be the most reasonable strategy to deliver db-vectors into human primary cells including stem cells ex vivo, e.g., via nucleofection prior to reimplantation of the edited cells into a patient.

Materials and Methods

Cell cultivation and transfection. Human HEK293T, HepG2, or CL48 cells were cultured in Dulbecco's Modified Eagle's Medium (Invitrogen, Waltham, MA) supplemented with 10% v/v heat-inactivated Fetal Bovine Serum (Hyclone, South Logan, UT) and 1% penicillin–streptomycin solution (Invitrogen). Cells were kept in humidified incubator with 5% CO2 and were passaged at 80–90% confluence. All transfection assays were performed using Lipofectamine 2000 (Invitrogen) following the manufacturer's recommendations.

Plasmids. To replace the 227 bp full-length H1 promoter of pSuper (Oligoengine, Seattle, WA) by the 99 bp minimal H1 promoter, both DNA strands resembling the mH1 sequence mH1-plus and mH1-minus were chemically synthesized, annealed, purified, and inserted into the pSuper plasmid using EcoRI and BglII restriction sites to generate plasmid pSuper-mH1 (Supplementary Table S1).26 Subsequently oligodeoxyribonucleotides shR-luc-plus and shR-luc-minus encoding a firefly luciferase targeting shRNA were annealed and inserted into pSuper-mH1 using the BglII and XhoI restriction sites to generate p-iPR-linear-s/as. To generate the hsa-miR-125b-1 expression plasmid pMIR-125b-1, the miR-125b-1 precursor gene was PCR amplified using hp-125b-1 as the template and primers Fw-pre-miR-125b-1 and Rv-pre-miR-125b-1; the PCR product was then inserted into pSuper-mH1 using the BglII and HindIII sites. The 72 bp minimal SV40 enhancer sequence was synthesized by gene synthesis (GeneArt/Applied Biosystems, Regensburg, Germany), and the 237 bp full-length SV40 enhancer was amplified by PCR from plasmid pGL3-Control (Promega, Madison, WI) using primers Fw-SV40 and Rv-SV40. Both enhancers were cloned into pMIR-125b-1 using the SacI and EcoRI sites to generate plasmids pMIR-mEnh-125b-1 and pMIR-fEnh-125b-1. To generate the hsa-miR-125b-1 luciferase reporter sensor plasmid, DNA sequences resembling three tandem repeats of the fully complementary miRNA binding sites BD-miR-125b-1-plus and BD-miR-125b-1-minus were chemically synthesized and inserted into pMIR-Report (Promega) using the SacI and HindIII sites. To study the function of an intron and the SV40 enhancer, the 132 bp chimeric human beta-globin mini-intron was generated by gene synthesis (GeneArt/Applied Biosystems) and cloned into the pGL3-Control plasmid using HindIII and NcoI sites. To create enhancer negative constructs, both enhancer and poly(A) signal of pGL3-Control were deleted by XbaI and BamHI digestion. Subsequently, the SV40 late poly(A) signal was PCR amplified using Fw-polyA and Rv-polyA and cloned back into the vector using the same restriction sites.

Dumbbell production Generation of hairpin template-transcribing dumbbells. To produce hairpin template-transcribing db-vectors, the mH1 promoter was PCR amplified using pSuper-mH1 template, 5ʹ-phosphorylated primer Fw-Bpu-mH1 introducing a Nb.Bpu10I site and primer Rv-BamHI-mH1 or Rv-BamHI-mH1-iPR (for db-iPR-hp-s/as and db-iPR-miR-hp-s/as) introducing a BamHI site at 5ʹ and 3ʹ ends, respectively (Supplementary Table S1). For miR-125b-1-expressing dumbbells with enhancer elements, either Fw-mEnh or Fw-fEnh was used as the forward primer. The PCR product was then cleaved using Nb.Bpu10I and BamHI and incubated at 37 °C for 4 hours to release the oligonucleotide resulting from Nb.Bpu10I cleavage in the presence of a neutralizing oligonucleotide to avoid reannealing before oligonucleotides were removed using a PCR purification kit (QIAgen, Hilden, Germany). T4 DNA ligase was used to ligate the phosphorylated Nb.Bpu10I 5ʹ-overhang to the recessive 3ʹ-hydroxyl group to form one loop of the dumbbell and to ligate a 5ʹ phosphorylated DNA hairpin structure comprising a BglII overhang to the compatible BamHI sticky end to covalently close the dumbbell from the other side. DNA hairpin hp-s/as was used to generate db-iPRT-hp-s/as and db-iPR-hp-s/as. DNA hairpins hp-miR-s/as or hp-125b-1 were used to generate db-iPR-miR-hp-s/as or db-hp-miR-125b-1, respectively. Ligation was performed in the presence of BamHI and BglII to suppress the formation of alternative dumbbells, including misligated homodimers shifting the equilibrium toward the correctly ligated dumbbell. Resulting dumbbells were subjected to T7 DNA polymerase (Fermentas, Thermo Scientific, Waltham, MA) treatment to destroy nonligated and misligated by-products.

Nicking enzyme method. For nicking-enzyme-based production we followed the protocol described by Taki et al.28,29 performing two rounds of PCR. db-iPR-linear-s/as and db-linear-miR-125b-1 were produced using this method. To produce the luciferase-targeting shRNA-expressing dumbbell db-iPR-linear-s/as, or the miR-125b-1-expressing dumbbell db-linear-miR-125b-1, p-iPR-linear-s/as or pMIR-125b-1 plasmid was used as the PCR template, respectively. The sequences of forward and reverse primers for the first PCR reaction were Fw-linear and Rv-linear. For the second PCR reaction we used primers Fw-2nd and Rv-2nd. Exonuclease treatment was done as described above.

ELAN method. For ELAN-based production of dumbbell db-hp-miR-125b-1 and luciferase-expressing dumbbells, we followed the protocol by Cost.30 For production of db-hp-125b-1, 2,000 ng mH1 PCR product was digested with each 2 U FD BamHI and FD EcoRI and each 300 pmol of the loop sequences L1 and 30 pmol of hp-125b-1 were ligated using 20 U of T4 DNA ligase in the presence of 1 U of FD BamHI, FD EcoRI, FD BglII, and FD MfeI. For production of luciferase-expressing dumbbells, 2,000 ng parental plasmid was digested with each 2 U FD BamHI and FD XhoI and each 50 pmol of the loop sequences Bam-loop and Xho-loop were ligated using 20 U of T4 DNA ligase in the presence of 1 U of FD BamHI, FD XhoI, and FD SalI. Exonuclease treatment was done as described above. Taq DNA polymerase, restriction enzymes, T4 DNA ligase, and T7 DNA polymerase, if not specified otherwise, were purchased from Life technologies (Singapore).

Nuclear extraction. To investigate the nuclear import of different vectors, 5 × 105 HepG2 cells were seeded in six-well plates and transfected with 1 pmol of dumbbell or plasmid DNA using Lipofectamine 2000 (Invitrogen). pVAX1 (Life Technologies, Waltham, MA) plasmid was used as control and to top-up DNA for dumbbell transfection. Twenty-four hours post-transfection, cells were harvested, washed twice with cold PBS, incubated in hypotonic buffer (20 mmol/l Tris-Cl, pH 7.4, 10 mmol/l NaCl, 3 mmol/l MgCl2) for 15 minutes on ice, and lysed by 20 pestle strokes in a dounce homogenizer in hypotonic buffer. After centrifugation at 3,000 rpm for 10 minutes at 4 °C, the supernatant (cytoplasmic fraction) was subsequently removed and the pellet (nuclear fraction) was further lysed by four freeze-thaw cycles using liquid nitrogen and a water bath. Lysed nuclei were centrifuged for 30 minutes at the maximum speed at 4 °C to separate the nuclear membranes from the nucleoplasm. Total nuclear nucleic acids were extracted from the supernatant using TRIzol (Invitrogen), and the absolute abundance of transfected vector DNA was determined by TaqMan qPCR quantification of the copy number of the minimal H1 promoter sequence using the 7900HT Fast real-time PCR system (Applied Biosystems, Waltham, MA).

Nuclei acids isolation, reverse transcription and real-time PCR. To investigate the kinetics of delivery and transcription of the luciferase-targeting shRNA, 5 × 105 HepG2 cells were transfected with 1 pmol of either a db-vector (linear or hairpin designs) or plasmid p-iPR-linear-s/as. Ten minutes, 1, 6, and 24 hours post-transfection RNA and episomal DNA were isolated using TRIzol (Invitrogen) following the manufacturer's protocol. Vector copy numbers were determined by real-time PCR. The amplicon was located within the minimal H1 promoter that was shared by all the constructs. Since db-vectors and plasmids have different PCR amplification efficiencies, we used individual rtPCR standard curves for the absolute quantification of each of the respective vectors. For reverse transcription and real-time PCR quantification of the luciferase-targeting shRNA transcript, we used a universal stem-loop RT primer-based TaqMan RT-PCR protocol as reported recently.54 The fold change was determined by ▵▵Ct quantification using β-actin RNA as an internal standard. All real-time PCR reactions were performed using 1xTaqMan Universal PCR Master Mix (Applied Biosystems) following the manufacturer's instructions.

Preparation of primary human peripheral blood mononuclear cells and nucleofection. Peripheral blood mononuclear cells were collected under informed consent from healthy donors. Peripheral blood mononuclear cells were stimulated with 600 IU/ml interleukin-2 (rIL-2; R&D Systems, Minneapolis, MN) and 50 ng/ml anti-CD3 (OKT-3; eBioscience, San Diego, CA) in AIM-V 2% human AB serum for 8 days. rIL-2 was increased to 1,000 IU/ml 1 day before nucleofection. For nucleofection, 400 ng (0.2 pmol) of p-MIR-125b-1 plasmid or either equimass (4 pmol) or equimolar amounts of db-hp-miR-125b-1, or 0.2 pmol or 4 pmol of miR-125b-1 mimic (Dharmacon, Lafayette, CO) were delivered into 106 stimulated primary human T cells. As control, cells were nucleofected with 4 pmol of miRIDIAN microRNA mimic negative control #1 (Dharmacon). Nucleofection was performed using the Lonza (Basel, Switzerland) 4D-Nucleofector system. Episomal DNA and total RNA were extracted at day 1, 3, 7, and 10 post-nucleofection using TRIzol (Invitrogen) following the manufacturer's recommendations. Vector copy numbers and target gene knockdown were determined by real-time PCR.

Luciferase reporter assay. To study the kinetics of the shRNA-triggered luciferase target gene knockdown, 5 × 104 HepG2 cells seeded in 24-well plates were co-transfected using Lipofectamine 2000 with luciferase reporter vector pGL3-Control (250 ng/well) and equimolar amounts (0.12 pmol/well) of either a db-vector (two different designs) or p-iPR-linear-s/as plasmid. 6, 12, and 24 hours post-transfection, shRNA-triggered luciferase knockdown was measured. To study the knockdown of firefly luciferase expression triggered by different shRNA-expressing minimized dumbbells, 5 × 104 HEK293T cells or HepG2 cells were co-transfected in 24-well plates with 400 ng pGL3-control luciferase reporter and 100 ng dumbbell DNA, and luciferase knockdown was monitored 48 hours post-transfection. Functionality of the miR-125b-1 sensor plasmid was tested by transfection of HepG2 cells with the sensor plasmid alone and compared with co-transfection of sensor plasmid plus feeder DNA or the miR-125b-1-expressing plasmid. Twenty-four hours post-transfection, luciferase expression was measured and standardized relative to the expression of the sensor-negative reporter vector. To study the knockdown of firefly luciferase-miR-125b-1, sensor reporter gene expression triggered by hsa-miR-125b-1-expressing dumbbells, 5 × 104 HepG2 cells were transfected in 24-well plates with 250 ng sensor plasmid alone or together with 250 ng miR-125b-1-expressing dumbbells or plasmid, and luciferase knockdown was monitored 48 hours post-transfection. To study the target gene knockdown triggered by hsa-miR-125b-1-expressing dumbbells with SV40 enhancer sequences, 5 × 104 HepG2 cells seeded in 24-well plates were (co-)transfected with 100 ng pMIR-125b-1-Sensor alone or together with 1.5 pmol dumbbell DNA, and luciferase knockdown was monitored 48 hours post-transfection. To study the functions of the chimeric intron and the full-length SV40 enhancer in the context of the luciferase-expressing dumbbells, 5 × 104 HEK293T or HepG2 cells seeded in 24-well plates were transfected with equimass amounts (400 ng) of dumbbell vectors or corresponding parental plasmids. Luciferase activity was measured 48 hours post-transfection using the Luciferase Assay System (Promega) following the manufacturer's protocol.

Statistical analysis. Error bars are standard errors of the mean (±SEM) of two to six independent experiments. Unpaired student t-test was used to determine significance when comparing two groups. For the comparison of more than two groups of data, one-way analysis of variance with Newman–Keuls post hoc test was used. Prism 5 Graphpad software was used for the statistical analysis. * represents P values <0.05, ** represents P values <0.01, and *** represents P values <0.001.

SUPPLEMENTARY MATERIAL Figure S1. Comparison of promoter strengths with regard to miR-125b-1 expression. Figure S2. Design and investigation of a firefly luciferase reporter vector for sensing of miR-125b-1 activity. Figure S3. Design and functional investigation of novel mismatched minimized hairpin template db-vectors for the expression of shRNA extended by a miRNA stem. Table S1. Oligonucleotides.

Acknowledgments

This work was supported by the National University of Singapore (NUS-Cambridge Start-up, grant number R-182-000-163-646 and Bridging Grant NUHSRO/2015/091/Bridging/02), the National Medical Research Council of Singapore (New Investigator, grant number NMRC/NIG/1058/2011), and the Ministry of Education of Singapore (Academic Research Fund (AcRF) Tier 1 Faculty Research Committee (FRC), grant numbers T1-2011Sep-04 and T1-2014Apr-02 and Seed Fund for Basic Science Research, grant number T1-BSRG 2015-05), all to V.P. The authors declare competing financial interests. A patent application covering major parts of the work is pending.

Supplementary Material

Supplementary Material
Click here for additional data file. (399.5KB, doc)

References

  1. Yin, H, Kanasty, RL, Eltoukhy, AA, Vegas, AJ, Dorkin, JR and Anderson, DG (2014). Non-viral vectors for gene-based therapy. Nat Rev Genet 15: 541–555. [DOI] [PubMed] [Google Scholar]
  2. Biasco, L, Baricordi, C and Aiuti, A (2012). Retroviral integrations in gene therapy trials. Mol Ther 20: 709–716. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Zaiss, AK, Machado, HB and Herschman, HR (2009). The influence of innate and pre-existing immunity on adenovirus therapy. J Cell Biochem 108: 778–790. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Rodríguez, EG (2004). Nonviral DNA vectors for immunization and therapy: design and methods for their obtention. J Mol Med (Berl) 82: 500–509. [DOI] [PubMed] [Google Scholar]
  5. Mayrhofer, P, Schleef, M and Jechlinger, W (2009). Use of minicircle plasmids for gene therapy. Methods Mol Biol 542: 87–104. [DOI] [PubMed] [Google Scholar]
  6. Schakowski, F, Gorschlüter, M, Junghans, C, Schroff, M, Buttgereit, P, Ziske, C et al. (2001). A novel minimal-size vector (MIDGE) improves transgene expression in colon carcinoma cells and avoids transfection of undesired DNA. Mol Ther 3(5 Pt 1): 793–800. [DOI] [PubMed] [Google Scholar]
  7. Stetson, DB and Medzhitov, R (2006). Recognition of cytosolic DNA activates an IRF3-dependent innate immune response. Immunity 24: 93–103. [DOI] [PubMed] [Google Scholar]
  8. Ablasser, A, Bauernfeind, F, Hartmann, G, Latz, E, Fitzgerald, KA and Hornung, V (2009). RIG-I-dependent sensing of poly(dA:dT) through the induction of an RNA polymerase III-transcribed RNA intermediate. Nat Immunol 10: 1065–1072. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Zhang, X, Brann, TW, Zhou, M, Yang, J, Oguariri, RM, Lidie, KB et al. (2011). Cutting edge: Ku70 is a novel cytosolic DNA sensor that induces type III rather than type I IFN. J Immunol 186: 4541–4545. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Sun, L, Wu, J, Du, F, Chen, X and Chen, ZJ (2013). Cyclic GMP-AMP synthase is a cytosolic DNA sensor that activates the type I interferon pathway. Science 339: 786–791. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Li, X, Shu, C, Yi, G, Chaton, CT, Shelton, CL, Diao, J et al. (2013). Cyclic GMP-AMP synthase is activated by double-stranded DNA-induced oligomerization. Immunity 39: 1019–1031. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Chen, ZY, He, CY, Meuse, L and Kay, MA (2004). Silencing of episomal transgene expression by plasmid bacterial DNA elements in vivo. Gene Ther 11: 856–864. [DOI] [PubMed] [Google Scholar]
  13. Fogg, JM, Kolmakova, N, Rees, I, Magonov, S, Hansma, H, Perona, JJ et al. (2006). Exploring writhe in supercoiled minicircle DNA. J Phys Condens Matter 18: S145–S159. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Moreno, S, López-Fuertes, L, Vila-Coro, AJ, Sack, F, Smith, CA, Konig, SA et al. (2004). DNA immunisation with minimalistic expression constructs. Vaccine 22: 1709–1716. [DOI] [PubMed] [Google Scholar]
  15. Kapp, K, Kleuss, C, Schroff, M and Wittig, B (2014). Genuine immunomodulation with dSLIM. Mol Ther Nucleic Acids 3: e170. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Kobelt, D, Aumann, J, Schmidt, M, Wittig, B, Fichtner, I, Behrens, D et al. (2014). Preclinical study on combined chemo- and nonviral gene therapy for sensitization of melanoma using a human TNF-alpha expressing MIDGE DNA vector. Mol Oncol 8: 609–619. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Lu, J, Zhang, F, Xu, S, Fire, AZ and Kay, MA (2012). The extragenic spacer length between the 5' and 3' ends of the transgene expression cassette affects transgene silencing from plasmid-based vectors. Mol Ther 20: 2111–2119. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lu, J, Zhang, F and Kay, MA (2013). A mini-intronic plasmid (MIP): a novel robust transgene expression vector in vivo and in vitro. Mol Ther 21: 954–963. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Chen, ZY, Riu, E, He, CY, Xu, H and Kay, MA (2008). Silencing of episomal transgene expression in liver by plasmid bacterial backbone DNA is independent of CpG methylation. Mol Ther 16: 548–556. [DOI] [PubMed] [Google Scholar]
  20. Riu, E, Chen, ZY, Xu, H, He, CY and Kay, MA (2007). Histone modifications are associated with the persistence or silencing of vector-mediated transgene expression in vivo. Mol Ther 15: 1348–1355. [DOI] [PubMed] [Google Scholar]
  21. Gracey Maniar, LE, Maniar, JM, Chen, ZY, Lu, J, Fire, AZ and Kay, MA (2013). Minicircle DNA vectors achieve sustained expression reflected by active chromatin and transcriptional level. Mol Ther 21: 131–138. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kole, R, Krainer, AR and Altman, S (2012). RNA therapeutics: beyond RNA interference and antisense oligonucleotides. Nat Rev Drug Discov 11: 125–140. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Dean, DA (1997). Import of plasmid DNA into the nucleus is sequence specific. Exp Cell Res 230: 293–302. [DOI] [PubMed] [Google Scholar]
  24. Dean, DA, Dean, BS, Muller, S and Smith, LC (1999). Sequence requirements for plasmid nuclear import. Exp Cell Res 253: 713–722. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Luo, MJ and Reed, R (1999). Splicing is required for rapid and efficient mRNA export in metazoans. Proc Natl Acad Sci USA 96: 14937–14942. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Myslinski, E, Amé, JC, Krol, A and Carbon, P (2001). An unusually compact external promoter for RNA polymerase III transcription of the human H1RNA gene. Nucleic Acids Res 29: 2502–2509. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Yu, H, Jiang, X, Tan, KT, Hang, L and Patzel, V (2015). Efficient production of superior dumbbell-shaped DNA minimal vectors for small hairpin RNA expression. Nucleic Acids Res 43: e120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Taki, M, Kato, Y, Miyagishi, M, Takagi, Y, Sano, M and Taira, K (2003). A direct and efficient synthesis method for dumbbell-shaped linear DNA using PCR in vitro. Nucleic Acids Res Suppl 3: 191–192. [DOI] [PubMed] [Google Scholar]
  29. Taki, M, Kato, Y, Miyagishi, M, Takagi, Y and Taira, K (2004). Small-interfering-RNA expression in cells based on an efficiently constructed dumbbell-shaped DNA. Angew Chem Int Ed Engl 43: 3160–3163. [DOI] [PubMed] [Google Scholar]
  30. Cost, GJ (2007). Enzymatic ligation assisted by nucleases: simultaneous ligation and digestion promote the ordered assembly of DNA. Nat Protoc 2: 2198–2202. [DOI] [PubMed] [Google Scholar]
  31. Schakowski, F, Gorschlüter, M, Buttgereit, P, Märten, A, Lilienfeld-Toal, MV, Junghans, C et al. (2007). Minimal size MIDGE vectors improve transgene expression in vivo. In Vivo 21: 17–23. [PubMed] [Google Scholar]
  32. Liang, L, Wong, CM, Ying, Q, Fan, DN, Huang, S, Ding, J et al. (2010). MicroRNA-125b suppressed human liver cancer cell proliferation and metastasis by directly targeting oncogene LIN28B2. Hepatology 52: 1731–1740. [DOI] [PubMed] [Google Scholar]
  33. Gong, J, Zhang, JP, Li, B, Zeng, C, You, K, Chen, MX et al. (2013). MicroRNA-125b promotes apoptosis by regulating the expression of Mcl-1, Bcl-w and IL-6R. Oncogene 32: 3071–3079. [DOI] [PubMed] [Google Scholar]
  34. Vacik, J, Dean, BS, Zimmer, WE and Dean, DA (1999). Cell-specific nuclear import of plasmid DNA. Gene Ther 6: 1006–1014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Miller, AM and Dean, DA (2008). Cell-specific nuclear import of plasmid DNA in smooth muscle requires tissue-specific transcription factors and DNA sequences. Gene Ther 15: 1107–1115. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Längle-Rouault, F, Patzel, V, Benavente, A, Taillez, M, Silvestre, N, Bompard, A et al. (1998). Up to 100-fold increase of apparent gene expression in the presence of Epstein–Barr virus oriP sequences and EBNA1: implications of the nuclear import of plasmids. J Virol 72: 6181–6185. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Miller, AM and Dean, DA (2009). Tissue-specific and transcription factor-mediated nuclear entry of DNA. Adv Drug Deliv Rev 61: 603–613. [DOI] [PubMed] [Google Scholar]
  38. Lin, KY, Zhang, XJ, Feng, DD, Zhang, H, Zeng, CW, Han, BW et al. (2011). miR-125b, a target of CDX2, regulates cell differentiation through repression of the core binding factor in hematopoietic malignancies. J Biol Chem 286: 38253–38263. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Schirmbeck, R, König-Merediz, SA, Riedl, P, Kwissa, M, Sack, F, Schroff, M et al. (2001). Priming of immune responses to hepatitis B surface antigen with minimal DNA expression constructs modified with a nuclear localization signal peptide. J Mol Med (Berl) 79: 343–350. [DOI] [PubMed] [Google Scholar]
  40. Brinster, RL, Allen, JM, Behringer, RR, Gelinas, RE and Palmiter, RD (1988). Introns increase transcriptional efficiency in transgenic mice. Proc Natl Acad Sci USA 85: 836–840. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Grimm, D, Streetz, KL, Jopling, CL, Storm, TA, Pandey, K, Davis, CR et al. (2006). Fatality in mice due to oversaturation of cellular microRNA/short hairpin RNA pathways. Nature 441: 537–541. [DOI] [PubMed] [Google Scholar]
  42. Zeng, Y, Wagner, EJ and Cullen, BR (2002). Both natural and designed micro RNAs can inhibit the expression of cognate mRNAs when expressed in human cells. Mol Cell 9: 1327–1333. [DOI] [PubMed] [Google Scholar]
  43. de Villiers, J, Olson, L, Tyndall, C and Schaffner, W (1982). Transcriptional ‘enhancers' from SV40 and polyoma virus show a cell type preference. Nucleic Acids Res 10: 7965–7976. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Laimins, LA, Khoury, G, Gorman, C, Howard, B and Gruss, P (1982). Host-specific activation of transcription by tandem repeats from simian virus 40 and Moloney murine sarcoma virus. Proc Natl Acad Sci USA 79: 6453–6457. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Yamaguchi, M, Obata, Y and Matsukage, A (1989). Paradoxical effect of Simian virus 40 enhancer on the function of mouse DNA polymerase beta gene promoter. Nucleic Acids Res 17: 3725–3734. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Corden, JL and Patturajan, M (1997). A CTD function linking transcription to splicing. Trends Biochem Sci 22: 413–416. [DOI] [PubMed] [Google Scholar]
  47. Takaoka, A, Wang, Z, Choi, MK, Yanai, H, Negishi, H, Ban, T et al. (2007). DAI (DLM-1/ZBP1) is a cytosolic DNA sensor and an activator of innate immune response. Nature 448: 501–505. [DOI] [PubMed] [Google Scholar]
  48. Hornung, V, Ablasser, A, Charrel-Dennis, M, Bauernfeind, F, Horvath, G, Caffrey, DR et al. (2009). AIM2 recognizes cytosolic dsDNA and forms a caspase-1-activating inflammasome with ASC. Nature 458: 514–518. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Jin, T, Perry, A, Jiang, J, Smith, P, Curry, JA, Unterholzner, L et al. (2012). Structures of the HIN domain:DNA complexes reveal ligand binding and activation mechanisms of the AIM2 inflammasome and IFI16 receptor. Immunity 36: 561–571. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Hemmi, H, Takeuchi, O, Kawai, T, Kaisho, T, Sato, S, Sanjo, H et al. (2000). A Toll-like receptor recognizes bacterial DNA. Nature 408: 740–745. [DOI] [PubMed] [Google Scholar]
  51. Herzner, AM, Hagmann, CA, Goldeck, M, Wolter, S, Kübler, K, Wittmann, S et al. (2015). Sequence-specific activation of the DNA sensor cGAS by Y-form DNA structures as found in primary HIV-1 cDNA. Nat Immunol 16: 1025–1033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Hagmann, CA, Herzner, AM, Abdullah, Z, Zillinger, T, Jakobs, C, Schuberth, C et al. (2013). RIG-I detects triphosphorylated RNA of Listeria monocytogenes during infection in non-immune cells. PLoS One 8: e62872. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Berg, RK, Rahbek, SH, Kofod-Olsen, E, Holm, CK, Melchjorsen, J, Jensen, DG et al. (2014). T cells detect intracellular DNA but fail to induce type I IFN responses: implications for restriction of HIV replication. PLoS One 9: e84513. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Jung, U, Jiang, X, Kaufmann, SH and Patzel, V (2013). A universal TaqMan-based RT-PCR protocol for cost-efficient detection of small noncoding RNA. RNA 19: 1864–1873. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material
Click here for additional data file. (399.5KB, doc)

Articles from Molecular Therapy are provided here courtesy of The American Society of Gene & Cell Therapy

RESOURCES