Introduction

The emergence of RNA interference (RNAi) as a central eukaryotic posttranscriptional regulatory mechanism has sparked continuous interest in the nucleases (ribonuclease III, RNase III) that both recognize and cleave double-stranded RNA (dsRNA). RNase III (transcribed and translated by the rnc gene in Escherichia coli) enzymes occur ubiquitously in different organisms. They have now been shown to occupy a central position in mediating dsRNA-dependent processes, including RNA maturation, RNA decay, gene silencing (responsible for generating microRNAs or short interfering RNA), and a range of other cellular activities (Apirion and Gitelman 1980; Dunn and Studier 1973; Elela et al. 1996; Fire et al. 1998; Krautz-Peterson and Skelly 2008, Zhang et al. 2004). The isolation of a point mutation in the chromosomal gene (rnc105) that abolished its enzyme activity (Kindler et al. 1973) revealed that the rnc gene is not an essential gene and can be knocked out.

With the development of molecular biology, an efficient bacteriophage λ-dependent Red recombination system has been developed for genetic engineering. Red recombination is used to substitute the exogenous linear polymerase chain reaction (PCR) fragments for the target gene in a chromosome by integrating a defective phage λ into the E. coli chromosome or a plasmid. The Red recombination system includes three genes: bet, exo, and gam, and their products are Bet, Exo, and Gam, respectively (Datsenko and Wanner 2000). These Red proteins enable the occurrence of recombination events between DNA species with as little as 40–60 bp homologous sequences (Baba et al. 2006; Yu et al. 2000; Datsenko and Wanner 2000; Poteete and Fenton 2000).

Studies show that complete cleavage of dsRNA by E. coli RNase III produces 12–15 bp duplex products on average (Amarasinghe et al. 2001), and these short dsRNAs could not induce gene silencing. Actually, in fungi, plants, and most invertebrates, longer dsRNA could be more effective in inducing RNA silencing and in the regulation of gene expression, in addition to its role as a tool in functional genomics analysis (Dykxhoorn et al. 2003; Meister and Tuschl 2004). Current approaches to producing long dsRNA rely on annealing sense and antisense single-stranded RNA strands. However, this may produce poor-quality dsRNA. The E. coli strain HT115 has a mutation in the rnc gene and could be used to produce long dsRNA for RNA-mediated virus resistance and studies in gene function (Tenllado and Díaz-Ruíz 2001; Tenllado et al. 2003, 2004; Timmons et al. 2001).

Numerous studies have reported protecting plants from virus infection by using dsRNA or hairpinRNA (hpRNA) derived from sequences of plant virus species (Liu et al. 2007; Smith et al. 2000; Waterhouse et al. 1998; Zhu et al. 2008). In 2001, Tenllado and his colleagues discovered that direct transformation of nontransgenic plant species with dsRNA produced by an in vitro expression system could efficiently interfere with virus infection. Thereafter, Tenllado et al. (2003) developed an in vivo expression system to produce large amounts of virus-derived dsRNAs in bacteria. The results showed that delivering dsRNA by a combination of lysing cells with the French Press onto plant surfaces could efficiently prevent plants from virus infections. In the antiviral response, the delivered dsRNA, transgenes, or endogenous genes are first processed by an RNase III-like nuclease, called Dicer in Arabidopsis, into 21–23 nt small interfering RNAs (siRNAs). Then, the siRNAs assemble into endoribonuclease-containing complexes known as RNA-induced silencing complexes (RISCs). The siRNA strands are then unwound to form activated RISCs and subsequently guide the RISCs to complementary RNA molecules and destroy the homologous target mRNAs (Hamilton and Baulcombe 1999; Hammond et al. 2000; Zamore et al. 2000). DsRNAs or self-complementary hairpin RNAs produced during the intermediate steps of genome replication were shown to be key triggers of RNA silencing (Bass 2000; Elbashir et al. 2000; Hamilton and Baulcombe 1999; Molnár et al. 2005; Vogler et al. 2007). In this paper, we constructed different prokaryotic bacterial expression systems to produce dsRNA derived from the Tobacco mosaic virus (TMV) coat protein (CP) gene. This paper aims to discover the best vector/host combinations for the highest output of dsRNA and explore the resistance of exogenous dsRNA or hpRNA for protecting the plant from virus infections.

Materials and methods

Bacterial strains, plasmids, media, chemicals, and other reagents

Table 1 lists bacterial strains and plasmids used in this work. Glucose, l-arabinose, Taq DNA polymerase, agarose, and other chemicals came from TaKaRa (Dalian, China). The DpnI was from New England Biolabs (Ipswich, MA, USA); KOD-plus DNA polymerase was from Toyobo (Tokyo, Japan). Luria–Bertani, Super Optimal Broth, and Super Optimal Broth with catabolite repression media were prepared as described elsewhere (Sambrook et al. 1989). Taq DNA polymerase was used in all PCR tests. KOD-plus DNA polymerase was used to generate DNA for cloning and mutagenesis. Tiangen products (Tiangen Biotech, Beijing, China) were used to isolate plasmid DNAs, gel-purify fragments, or purify PCR products. The concentration and purity of all PCR products, extracted dsRNA, and DNA were determined with a NanoDrop ND-1000 UV spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA).

Table 1 E. coli strains and plasmids used in this study
Full size table

Construction of different vectors and amplifying the targeting fragments

Table 2 lists all primers used in this study. The TMV CP gene was amplified by real-time (RT)-PCR from extracted TMV mRNA using primers TMVCPII-5 and TMVCPII-3 and subcloned into the PstI and SalI sites of vector pUC19 to give pUC19CP (Fig. 1). The subcloned fragment consisted of a 480-bp cDNA presenting the entire coding region of the TMV CP gene and a 120-bp spacer (often called “loop” in hpRNA) representing sequences of the bacterial glucoronidase gene (Chen et al. 2003). Plasmid pUC19CP was used as a template for the construction of three different vectors (LCP480, pGEM-CP480, and pET-CP480) for the production of dsRNA in vivo after transformation into different E. coli strains (Fig. 1). We constructed vector LCP480 by inserting a 480-bp TMV CP gene digested with PstI and SalI into the multicloning sites of plasmid L4440 digested with the same restriction endonuclease sites. We constructed vectors pGEM-CP480 and pET-CP480 with inverted repeat of the 480-bp TMV CP gene and separated the two DNA units by the spacer sequence (a 120-bp glucuronidase gene).

Table 2 The primers used for gene knock-out, mutantscreening, and vector construction
Full size table
Fig. 1
figure 1

Map of three different vectors. Schematic representation of three different vectors carrying the TMV CP cDNA sequence: a LCP480, two T7 promoters; b pET-CP480, one T7 promoter, inverted-repeat TMV cDNA sequence; and c pGEM-CP480, one T7 promoter, inverted-repeat TMV cDNA sequence

Full size image

The primers RNaseIII50-5 and RNaseIII50-3 and LacY-5 and LacY-3 were used for amplifying the chloramphenicol resistance gene and kanamycin resistance gene with 50-bp homologous sequences with the rnc and LacY genes, respectively. The targeting PCR fragments with cat R or kan R were generated independently and were amplified in the Eppendorf Mastercycler (Eppendorf, Germany) according to the protocols suggested by KOD-plus instruction (Toyobo). The PCR products were digested with DpnI, ethanol precipitated, and resuspended in the appropriate TE buffer (10 mmol/L Tris–HCl, 1 mmol/L ethylenediamine tetraacetic acid (EDTA), pH 7.5). Afterwards, the PCR products were size-fractionated on a 1% agarose gel in Tris–acetate–EDTA buffer. The corresponding band was excised and purified with TIANgel Midi Purification Kit (Tiangen Biotech) according to the manufacturer’s protocol, except that in the last step, we suspended the amplified linear DNAs in sterile water.

Gene knock-out, mutant selection, elimination of pKD46, and the resistant marker gene

JM109 (DE3) and HMS174 (DE3) PLysS were transformed with plasmid pKD46 according to the method suggested by Sambrook et al. (1989). Knock-out of the rnc and LacY genes, mutant selection, and elimination of pKD46 and the antibiotic-resistant marker gene were performed according to the protocols suggested by Datsenko and Wanner (2000) with small modifications. The primers (RNaseIII50-5 and RNaseIII50-3; LacY-5 and LacY-3) and plasmids (pKD3 or pKD4) were used to knock out the rnc and LacY genes, respectively. Colony-direct PCR (CD-PCR) was used to show that all mutants had the correct structures.

Dot-blot hybridization and sequence analysis for knock-out of the rnc and LacY genes

Genomic DNA of E. coli was extracted using the Biospin Bacteria Genomic DNA Extraction Kit (Bioer Technology, Hangzhou, China) according to the instructions of the supplier. The E. coli rnc and lacY mutation strains were further verified by dot-blot hybridization using DIG High Prime DNA Labeling and Detection Starter Kit I (Roche, Germany) according to the instructions of the supplier. Dot-blot hybridization conditions and detection were performed according to Sambrook et al. (1989). The recombination fragments of different mutations were cloned into the vector pEASY-T1 and sequenced by TaKaRa for further verification. The correct mutant strains have been deposited at China General Microbiological Culture Collection Center (CGMCC). The collection numbers for M-JM109, M-JM109lacY, and M-HMS174 are CGMCC 1.8744, CGMCC 1.8745, and CGMCC 1.8746, respectively.

Induction and extraction of dsRNA

The induction of dsRNA from different prokaryotic bacterial expression systems was performed according to Tenllado et al. (2003) with small modifications. TRIZOL Reagent (Invitrogen, USA) was used for the isolation of total RNA from bacteria as described in the instructions in the supplier’s instructions. The nucleic acid pellet obtained from 1 mL culture of E. coli cells was resuspended in 20 μL TE buffer and used for 2% agarose gel electrophoresis. Accumulation of dsRNA in bacterial extracts and its purity were confirmed by treatment with RNaseA under high salt conditions (Tenllado et al. 2003).

Quantitative real-time PCR and resistance analysis of extracted dsRNA transcripts

One-step real-time RT-PCR of the dsRNA transcripts were carried out in a Bio-Rad iQ5 system using a Quant One Step RT-PCR (SYBR Green I) Kit (Tiangen Biotech) according to the supplier’s instructions with small modifications. All samples were run in triplicate, and relative amounts of dsRNA were determined using the 2−ΔΔCt method (Livak and Schmittgen 2001). One-way analysis of variance (ANOVA) analysis and Duncan’s multiple-range tests were performed with SAS 8.2 software (SAS Institute, Cary, NC, USA).

We performed resistance tests following methods proposed by Tenllado et al. (2003). The tested tobacco was kept in growth chambers with a 16-h light and an 8-h dark cycle at 25°C. We observed the development of TMV-infection symptoms (first, a light-green coloration between the veins of young leaves followed by a mosaic on the tobacco leaves) for as long as their life cycle (2 months post-inoculation). The content of the TMV virus in tested plants was detected by indirect enzyme-linked immunosorbent assay (ELISA), according to Liu et al. (2007).

Small RNA fraction extraction and Northern blot

We extracted small RNA fractions of the tested tobacco using PureLink™ miRNA Isolation Kit (Invitrogen) according to the instructions of the supplier. Electrophoresis, blotting, and EDC cross-linking of small RNA fractions were carried out sequentially according to Pall et al. (2007). Probe synthesis, siRNA hybridization, and chemiluminescent detection (DIG Northern Starter Kit, Roche) were performed according to Goto et al. (2003).

Results

Knock-out of the rnc and LacY genes

Plasmid pKD46 was transformed into JM109 (DE3) and HMS174 (DE3) PLysS, respectively. Knock-out mutants of strains JM109 (DE3) and HMS174 (DE3) PLysS with a precise deletion of a 458-nt segment in the rnc gene were produced as described earlier in the “Materials and methods” section. Representative kanamycin-resistant and chloramphenicol-resistant transformants were characterized after transformation of PCR-fusion products of the cat R gene of pKD3 and rnc sequences and of PCR-fusion products of the kan R gene of pKD4 and rnc sequences into the different E. coli hosts JM109 (DE3) and HMS174 (DE3), respectively. Both carried the pKD46 plasmid for Red-mediated recombination. PCR tests using locus-specific primers and cat- or kan-specific primers (Table 2) revealed that all mutants from the strains had new junctions and locus-specific fragments of the expected size (Fig. 2). For convenience, we named the rnc gene mutants of these strains M-JM109 and M-HMS174, respectively.

Fig. 2
figure 2

PCR analysis to verify the successful construction of E. coli mutants. PCR analysis to verify mutants M-JM109 and M-HMS174 (a, left and right, respectively), the elimination of the chloramphenicol marker gene in M-JM109 (b) and in the M-JM109lacY mutant strain (c). M 1 Gene Ruler™ DNA Ladder Mix, M 2 100 bp DNA-Ladder Marker. aLeft, lane 1 PCR performed on wild-type JM109 (DE3) strain with the primers JD-5 and JD-3, expected fragment size 879 bp; lane 2 the targeting linear DNA fragment performed on pKD3 with the primers RNaseIII50-5 and RNaseIII50-3, expected fragment size 1,152 bp; lane 3 PCR performed on M-JM109 with the primers JD-5 and JD-3, expected fragment size 1,473 bp; lane 4 PCR performed on M-JM109 with the primers JD-5 and Cat-3, expected fragment size 1,201 bp; lane 5 PCR performed on M-JM109 with the primers Cat-5 and JD-3, expected fragment size 1,322 bp. aRight, lane 1 PCR performed on wild-type HMS174 (DE3) PlysS strain with primers JD-5 and JD-3, expected fragment size 879 bp; lane 2 the targeting linear DNA fragment performed on pKD4 with the primers RNaseIII50-5 and RNaseIII50-3, expected fragment size 1,615 bp; lane 3 PCR performed on M-HMS174 with the primers JD-5 and JD-3, expected fragment size 1,936 bp; lane 4 PCR performed on M-HMS174 with the primers JD-5 and Cat-3, expected fragment size 1,664 bp; lane 5 PCR performed on M-HMS174 with the primers Cat-5 and JD-3, expected fragment size 1,787 bp. bLane 1 PCR performed on M-JM109 with the primers JD-5 and JD-3, expected fragment size 1,473 bp; lane 2 PCR performed on chloramphenicol eliminated mutant with the primers JD-5 and JD-3, expected fragment size 543 bp; lane 3 PCR performed on chloramphenicol eliminated mutant with the primers JD-5 and Cat-3, expected fragment size 271 bp; lane 4 PCR performed on chloramphenicol eliminated mutant with the primers Cat-5 and JD-3, expected fragment size 394 bp. clane 1 PCR performed on M-JM109 with primers LacY-JD5 and LacY-JD3, expected fragment size 1,692 bp; lane 2 the targeting linear DNA fragment PCR performed on pKD4 with the primers LacY-5 and LacY-3, expected fragment size 1,615 bp; lane 3 PCR performed on M-JM109lacY with the primers LacY-JD5 and LacY-JD3, expected fragment size 2,296 bp; lane 4 PCR performed on M-JM109lacY with the primers LacY-JD5 and Cat-3, expected fragment size 2,155 bp; lane 5 PCR performed on M-JM109lacY with the primers Cat-5 and LacY-JD3, expected fragment size 1,656 bp

Full size image

The lac permease (lacY) mutation (Tuner™ strains) allows uniform entry of isopropyl-β-d-thio-galactoside (IPTG) into all cells in the population, which produces a concentration-dependent, homogeneous level of induction (Minea et al. 2005). In response to this, we knocked out the LacY gene of strain M-JM109, which produces more dsRNA than other E. coli strains (Figs. 3 and 4). To avoid using more antibiotics, we eliminated the resistant-marker gene with plasmid pCP20. On elimination of the chloramphenicol resistance-marker gene, resulting sensitive clones in PCR tests using locus- and cat R-specific primers (Table 2) produced new fragments, as expected from the loss of the cat R gene. Subsequently, we used PCR-fusion-products of the kan R gene of pKD4 and of LacY sequences (for details, see the “Materials and methods” section) in transformation of one of the chloramphenicol-sensitive descendants of mutant M-JM109 in order to knock out precisely a 911-nt segment in the LacY (Fig. 2) gene. For convenience, we gave the LacY gene mutant of M-JM109 the name M-JM109lacY. We further verified accuracy of all mutations by dot-blot hybridization analysis and sequence analysis (data not shown).

Fig. 3
figure 3

Production of dsRNA by different E. coli strains defective for RNaseIII. Approximately 10 μg of total RNA extracted from different E. coli strains was loaded into each lane. Lane M dsRNA Marker; CK total RNA extracted from the wild-type strain JM109 (DE3) containing LCP480; lanes1–4 dsRNA extracted from M-HMS174/LCP480, HT115/LCP480, M-JM109/LCP480, and M-JM109lacY/LCP480, respectively; lanes 5–8 dsRNA extracted from HT115/pGEM-CP480, M-JM109/pGEM-CP480, M-JM109lacY/pGEM-CP480 and M-HMS174/pGEM-CP480, respectively; lanes 9–12 dsRNA extracted from M-HMS174/pET-CP480, HT115/pET-CP480, M-JM109/ pET-CP480 and M-JM109lacY/pET-CP480, respectively

Full size image
Fig. 4
figure 4

Quantitative analysis of dsRNA production in different E. coli expression systems. Averages and standard deviations were calculated from n = 3

Full size image

Production of dsRNA by different E. coli strains defective for RNase III

For comparison, different prokaryotic expression systems were designed to express dsRNA or hpRNA containing TMV CP cDNA sequence under the control of the T7 promoter. To simplify the descriptions of the bacterial strains used in this paper, we used the following terminology: HT115 is an RNase III-mutant bacterial host harboring a λDE3 lysogen (source of T7 polymerase), and HT115/LCP480, HT115/pGEM-CP480, and HT115/pET-CP480 are lysogenic strains harboring the plasmids LCP480, pGEM-CP480, and pET-CP480, respectively. M-JM109/LCP480, M-JM109/pGEM-CP480, M-JM109/pET-CP480, M-HMS174/LCP480, M-HMS174/pGEM-CP480, M-HMS174/pET-CP480, M-JM109lacY/LCP480, M-JM109lacY/pGEM-CP480, and M-JM109lacY/pET-CP480 are named analogously to the plasmid-harboring HT115 derivatives. Inducing the T7 promoter with IPTG, we found that the HT115, M-JM109, M-HMS174, and M-JM109lacY strains accumulated significantly higher levels of dsRNA, while the accumulation of dsRNA by the RNase III-expressing strains was not detectable by gel electrophoresis (Fig. 3).

As shown by the agarose gel in Fig. 3, there is a clear band about 660-bp extracted from strains HT115/LCP480, M-HMS174/LCP480, M-JM109/LCP480, and M-JM109lacY/LCP480, respectively. LCP480 has two convergent T7 polymerase promoters in opposite orientation, and both can be induced by IPTG. LCP480 therefore produces (+) and (−) ssRNA, which then leads to the formation of dsRNAs inside bacterial cells. Because the transcribed dsRNA contains (next to the TMV CP gene sequence) a 180-bp sequence with the multicloning region of plasmid L4440, the dsRNA produced by plasmid LCP480 is 660-bp dsRNA long. There are two obvious dsRNA bands extracted from strains containing either pGEM-CP480 or pET-CP480. The reason for this phenomenon is a 120-bp hairpin loop in the self-complementary RNAs of pGEM-CP480 and pET-CP480, respectively. RNAs with a hairpin loop move slowly in agarose gels compared to dsRNA without a loop region. Thus, the upper band in Fig. 3 localizes at a place corresponding to a sequence size of 550-bp dsRNA, while the other band was found below the regular position of 500-bp dsRNA likely because a 480-bp dsRNA without a loop was formed due to degradation of the loop by RNaseA.

Quantitative real-time PCR of dsRNA

To compare dsRNA produced by different dsRNA prokaryotic expression systems, we performed quantitative real-time (qRT)-PCR and measured the absorbance of extracted dsRNA at 260 nm. To further corroborate differences in dsRNA production by the different bacterial expression systems, we used the qRT-PCR. ANOVA analysis and multi-comparison analysis were carried out from the average results of three independent experiments. According to a Duncan all-pairwise comparison test, the three vectors LCP480, pET-CP480, and pGEM-CP480 show differences to each other in dsRNA production (P < 0.01). The vector pGEM-CP480 is best, the pET-CP480 is second best, and the vector LCP480 is third. Two of the four mutations, M-JM109 and M-JM109lacY, show differences to M-HMS174 and HT115 (P < 0.01). However, M-JM109 showed no difference to M-JM109lacY, and M-HMS174 showed no difference to HT115. The differences are as follows: M-HMS174/LCP480 (1.00 ± 0.01) < HT115/LCP480 (1.34 ± 0.08) < M-JM109/LCP480 (1.72 ± 0.40) < M-JM109lacY/LCP480 (2.14 ± 0.10), M-HMS174/pET-CP480 (1.87 ± 0.13) < HT115/pET-CP480 (2.22 ± 0.04) < M-JM109/ pET-CP480 (3.85 ± 0.36) < M-JM109lacY/pET-CP480 (4.31 ± 0.06), and M-HMS174/pGEM-CP480 (3.10 ± 0.13) < HT115/pGEM-CP480 (3.72 ± 0.04) < M-JM109/ pGEM-CP480 (6.50 ± 0.69) < M-JM109lacY/ pGEM-CP480 (7.28 ± 0.56). In total, strain M-JM109 or the M-JM109lacY strain and the vector pGEM-CP480 are the best choices for dsRNA production (Fig. 4).

Resistance analysis of extracted dsRNA transcripts

To prove whether the bacterial-produced dsRNA could interfere with TMV infection, we tested 20 tobacco plants for each vector. The tobacco plants were inoculated with mixtures of TMV plus bacterial-dsRNA extracts obtained from E. coli strain M-JM109lacY. Per tested plant, 300  μg dsRNA was applied regardless of the vector used for dsRNA production. The titer of TMV was 2 μg/mL of purified virus. The plants were inoculated with equal volume mixtures (100 μL per tobacco leaf) of TMV plus dsRNA produced by three different vectors (LCP480, pET-CP480, and pGEM-CP480). At 10 days post-inoculation (dpi), all wild-type tobacco showed the characteristic TMV mosaic on their leaves. The tested tobacco was assessed approximately every 10 days. The final classification refers to the state of the plants after the last assessment date. We discerned two major categories of plants after simultaneous treatment of TMV and bacterial dsRNA, resistant and susceptible. The susceptible tobacco showed varying degrees of symptoms; some had very mild symptoms (very mild green mosaics compared with the TMV-infected control tobacco). In all three cases, 20% of the total tested plants showed very mild symptoms, and 30% of the total tested plants showed the characteristic TMV mosaic on their leaves. The resistant plants remained symptom-free, even beyond 2 months post-inoculation. The dsRNA produced by the three vectors LCP480, pGEM-CP480, and pET-CP480 could all protect tobacco from TMV infection. They showed no great difference in resistance to TMV infection (regardless of whether the produced dsRNA contained a hairpin loop or not). In all three cases, 50% of the tested plants were resistant. The TMV-ELISA values of resistant tobacco show no great difference with those of the negative control tobacco, and the virus content in the susceptible tobacco was obviously lower than in the positive control tobacco but higher than in the negative control tobacco (Table 3).

Table 3 ELISA detection of resistance mediated by dsRNA transcripts
Full size table

Northern blot of siRNA extracted from the tested tobacco

Virus resistance in transgenic plants containing a virus-derived transgene is frequently caused by posttranscriptional gene silencing of the transgene (Smith et al. 2000; Vargas et al. 2008; Waterhouse et al. 1998). To prove whether the employment of exogenous dsRNA could induce virus resistance, we investigated the presence or absence of siRNA in tested tobacco plants at the end of the experiment. We also analyzed susceptible tobacco and wild-type tobacco nontreated with TMV of the same age as positive and negative controls, respectively. TMV RNA signals were detected in the susceptible tobacco but not in the resistant tobacco and TE buffer-treated wild-type tobacco (data not shown). We detected siRNA signals in the resistant and susceptible tobacco but not in the wild-type tobacco (Fig. 5). Northern blot results, showing a decreased accumulation of TMV RNA in tested resistant and susceptible tobacco (data not shown) and the presence of siRNAs (Fig. 5) in tested resistant and susceptible tobacco, strongly support the conclusion that resistance to TMV is an RNA-mediated virus resistance.

Fig. 5
figure 5

Northern blot analysis of siRNA from tobacco plants. Two fully expanded tobacco leaves were mechanically inoculated per assay using carborundum as an abrasive. Northern blot analysis of small RNA fractions extracted from the tested plants at about 15 days postinoculation (about six-leaf stage). Approximately 15 μg of small RNA fractions were loaded into each lane. aLane 1 small RNA fractions extracted from wild-type tobacco; lanes 2–5 small RNA fractions extracted from resistant tobacco; lane 6 small RNA fractions extracted from susceptible tobacco. b 5 S rRNA was used to show that an equal amount of small RNA fraction was loaded in lanes 1–6

Full size image

Discussion

In this work, using Red-mediated recombination, we generated the RNase III-defective E. coli mutant strains M-JM109, M-JM09lacY, and M-HMS174 for producing great quantities of dsRNA. These mutants were proved to be efficient in producing dsRNA by lack of dsRNA-specific RNases, just as the previously described strain HT115 does (Tenllado et al. 2003).

In this study, we discovered that the strains M-JM109 and M-JM109lacY are the best choices for producing great quantities of specific dsRNA. The main reasons might be as follows. The genotype of strain JM109 (DE3) is endA1and recA1 (Table 1). The endA1 gene encodes endonuclease I, which has an important role in genomic DNA and plasmid replication and recombination, whereas the recA1 gene expresses an adenosine triphosphate-dependent DNA recombinase. The endA and recA1mutation will make the exogenous plasmids or the transcribed dsRNA more stable, and then the degradation of dsRNA may be lower (Kowalczykowski et al. 1994; Sambrook et al. 1989; Yanisch-Perron et al. 1985).

The output of dsRNA produced by different bacterial-expression systems was greatly different due to the different vectors in our study. In the end, we discovered that the output of dsRNA produced by pGEM-CP480 was the highest of the three vectors. Compared to pET-CP480 and LCP480, pGEM-CP480 has a high copy number plasmid, which can reach several hundred copies per cell (Bao and Cagan 2006; Sambrook et al. 1989). Secondly, although pGEM-CP480 as pET-CP480 has only one T7 promoter and from which the (+) and (−) ssRNA are transcribed in a single unit linked together by a loop, they might easily and quickly form dsRNA after transcription. The two T7 promoters of LCP480 separately transcribe (+) and (−) ssRNA; the (+) and (−) ssRNA are not linked by a loop after transcription. Accordingly, they might be vulnerable and easily degraded by the RNA endonucleases (such as RNaseA, RNaseB) in E. coli before forming the dsRNA (Kowalczykowski et al. 1994; Sambrook et al. 1989).

TMV causes serious disease in tobacco plants, with systemic mosaic symptoms and great losses in yield and quality of tobacco leaves (Lucas 1975; Zhu et al. 2008). The RNAi technique has been proved to be a powerful tool for pathogen-derived resistance. In this paper, the TMV CP gene was used for bacterial RNAi vectors, and the exogenous dsRNA produced by E. coli has been shown to be very effective in inhibiting TMV infection.

At present, researches need a safe and efficient strategy for antiviral gene engineering. Based on RNAi technology, delivery of dsRNA or hpRNA derived from plant virus sequences may be an important new strategy for virus resistance in transgenic or nontransgenic plants (Duan et al. 2008; Liu et al. 2007; Tenllado and Díaz-Ruíz 2001; Tenllado et al. 2003; Vargas et al. 2008). Compared to acquiring transgenic plants, using dsRNA transcripts provided by this strategy for RNAi has apparent advantages. First, this method avoids time- and labor-consuming, cost-intensive production of transgenic plants. Transgenic plants have increasingly raised concern for having potential negative ecological effects, such as heterologous encapsidation, complementation, synergy, and genetic flow between organisms (Tenllado et al. 2004). Second, the method offered here can deliver multiple virus dsRNAs to disrupt several virus species at once and may achieve multiple virus resistances at one time. Third, this method is safer than pesticides, which potentially cause chemical residue accumulation in the environment. Finally, dsRNA produced by this method remains are stable in vitro for several days after inoculation, even after exposure of the plants to a rain storm (Tenllado and Díaz-Ruíz 2001). This method will make RNAi technology widely available and applicable for protecting crops from virus infection. In this study, we have established a simple, fast, safe, and inexpensive strategy to produce large amounts of dsRNA derived from viral sequences using a bacterial-expression system. We hope this strategy will widely be used in the near future.