A chemical-regulated inducible RNAi system in plants


  • Hui-Shan Guo,

    1. Laboratory of Molecular Cell Biology, Temasek Life Sciences Laboratory, National University of Singapore, 1 Research Link, 117604 Singapore, and
    Search for more papers by this author
  • Ji-Feng Fei,

    1. Laboratory of Molecular Cell Biology, Temasek Life Sciences Laboratory, National University of Singapore, 1 Research Link, 117604 Singapore, and
    Search for more papers by this author
  • Qi Xie,

    1. Laboratory of Molecular Cell Biology, Temasek Life Sciences Laboratory, National University of Singapore, 1 Research Link, 117604 Singapore, and
    Search for more papers by this author
  • Nam-Hai Chua

    Corresponding author
    1. Laboratory of Plant Molecular Biology, Rockefeller University, 1230 York Avenue, New York, NY 10021, USA
      For correspondence (fax +1 212 327 8327; e-mail chua@mail.rockefeller.edu).
    Search for more papers by this author

For correspondence (fax +1 212 327 8327; e-mail chua@mail.rockefeller.edu).


Constitutive expression of an intron-containing self-complementary ‘hairpin’ RNA (ihpRNA) has recently been shown to efficiently silence target genes in transgenic plants. However, this technique cannot be applied to genes whose silencing may block plant regeneration or result in embryo lethality. To obviate these potential problems, we have used a chemical-inducible Cre/loxP (CLX) recombination system to trigger the expression of an intron-containing inverted-repeat RNA (RNAi) in plants. A detailed characterization of the inducible RNAi system in transgenic Arabidopsis thaliana and Nicotiana benthamiana plants demonstrated that this system is stringently controlled. Moreover, it can be used to induce silencing of both transgenes and endogenous genes at different developmental stages and at high efficiency and without any detectable secondary affects. In addition to inducing complete silencing, the RNAi can be produced at various times after germination to initiate and obtain different degrees of gene silencing. Upon induction, transgenic plants with genetic chimera were obtained as demonstrated by PCR analysis. Such chimeric plants may provide a useful system to study signaling mechanisms of gene silencing in Arabidopsis as well as other cases of long-distance signaling without grafting. The merits of using the inducible CLX system for RNAi expression are discussed.


RNA-mediated gene silencing is a conserved mechanism that recognizes double-stranded RNA (dsRNA) as a signal to trigger sequence-specific degradation of homologous mRNA. dsRNA has been used as a powerful tool for the investigation of RNA silencing in a variety of organisms, such as RNA interference (RNAi) in Caenorhabditis elegans (Fire et al., 1998) and mammalian cells (Paddison et al., 2002; Sui et al., 2002) and post-transcriptional gene silencing (PTGS) in plants (Chuang and Meyerowitz, 2000; Waterhouse et al., 1998). In plants and C. elegans, RNA silencing involves two steps: (i) a local induced silencing, including an initial processing of the triggering dsRNA into short interfering RNA (siRNA) of 21–25 nt (Elbashir et al., 2001; Hamilton and Baulcombe, 1999) and (ii) a systemic spread of the silencing signal throughout the entire organism (Voinnet et al., 2000; Winston et al., 2002). The presence of a spliceable intron in the transgene encoding the dsRNA appears to enhance the silencing efficiency (Smith et al., 2000; Wesley et al., 2001). Constitutive expression of intron-containing self-complementary ‘hairpin’ RNA (ihpRNA) constructs can induce PTGS with almost 100% efficiency when directed against viruses or endogenous genes (Smith et al., 2000).

The completion of the sequencing of the Arabidopsis genome has uncovered a large number of genes with unknown functions. Potentially, the dsRNA-mediated gene silencing technique can be used to investigate the functions of these genes. The most effective silencing, brought about by intron-containing dsRNA, would produce phenotypes resembling those of the null alleles of the target genes. If the target gene is required for basic cell function or development, constitutive dsRNA-mediated silencing of the gene may produce detrimental effects or even cause plant lethality resulting in no recovery of transgenic plants for investigation. This problem can be circumvented somewhat by inducing gene silencing, using either Agrobacterium infiltration (Voinnet and Baulcombe, 1997) or virus-derived vectors (Dalmay et al., 2000; Peele et al., 2001; Ratcliff et al., 2001; Turnage et al., 2002). However, these two methods have limited application in functional genomics because of their transient nature, and their silencing effects are not heritable.

In this paper, we describe the development of an inducible gene silencing system using intron-containing, inverted-repeat RNA (referred to as inducible RNAi). The chemical-inducible system we use here is the CLX system, an XVE-based (XVE for LexA-VP16-ER) (Zuo et al., 2000), site-specific DNA recombination mediated by Cre/loxP (Zuo et al., 2001). The Cre expression is placed under the control of the chimeric transcription factor XVE, whose activity is strictly regulated by estrogens. Upon induction by 17β-estradiol, Cre/loxP-mediated recombination leads to activation of the RNAi transcription cassette by bringing it immediately downstream of the constitutive G10-90 promoter (Zuo et al., 2001; Figure 1). Compared to other inducible systems, the most significant feature of the CLX system is that it is strigently controlled, and upon induction, it produces DNA recombination with high efficiency (for a review, see Hare and Chua, 2002; Ow, 2001).

Figure 1.

Schematic diagrams of the inducible and constitutive RNAi constructs.

(a) A schematic diagram of an intron-containing intermediate vector pSK-int with multiple restriction sites on both arms of the intron. DNA fragments encoding sense and antisense RNA (approximately 350 nt of 3′GFP or approximately 400 nt of 5′PDS, respectively) were cloned into pSK-int at 5′ and 3′ arms of the intron in appropriate restriction sites indicated by dotted lines, resulting in pSK-GFPi and pSK-PDSi.

(b) A schematic diagram showing structural features of the inducible pX7-RNAi construct and Cre/loxP-mediated DNA recombination (see Zuo et al., 2001 for details). XVE, a chimeric transactivator containing the regulator domain of an estrogen receptor (Zuo et al., 2000); HYG, a hygromycin-resistance marker; Cre, the bacteriophage P1 Cre recombinase with an intron (Zuo et al., 2001); loxP, specific recognition sites of Cre; OlexA-46, eight copies of the LexA DNA binding site fused to the −46 CaMV 35S promoter; G10-90, a strong, synthetic, constitutive promoter; int, actin 11 intron (ATU27981, nt 1957–2111); RNAi, DNA sequences encoding the intron-containing inverted-repeat RNAs; G10-90-RNAi, the reconstituted transcription unit derived from Cre/loxP-mediated DNA recombination after inducer treatment. Arrows inside transcription units indicate the direction of transcription. P1–4 denote primers used for PCR analysis shown in Figure 5(c).

(c) A fragment of PDS-int-PDS was cloned between the 35S promoter and the terminator in the binary plasmid pCAMBIA 1300 (AF234296) to give 35S-PDSi, a constitutive RNAi construct.

Here, we show that the inducible RNAi system can be used to silence, with high efficiency, the expression of a GFP transgene and an endogenous phytoene desaturase (PDS) gene in Nicotiana benthamiana as well as in Arabidopsis thaliana. Upon induction at seed germination or post-germination stages, the efficiency and effectiveness of PDS silencing are comparable to those obtained with a 35S-RNAi construct. A stable and reproducible inducible RNAi phenotype was obtained in subsequent transgenic generations. The merits of the inducible RNAi system for silencing of endogenous genes are discussed.


Silencing of a GFP transgene

To evaluate the function of the inducible RNAi system, the construct pX7-GFPi (Figure 1b) was introduced into a transgenic N. benthamiana line carrying a 35S-GFP transgene (GFP-16c) (Ruiz et al., 1998) by Agrobacterium-mediated transformation. T1 seeds from 12 transgenic lines (GFP-16c/pX7-GFPi, named as 16c-GFPi) were germinated on the inductive medium (MS + 2 µm 17β-estradiol) containing hygromycin (hyg). After 2 weeks, eight lines showed uniform red fluorescence under UV light, which was caused by chlorophyll autofluorescence in the absence of GFP accumulation. Some red seedlings stopped growing upon continued incubation on the selective medium, presumably because of the complete Cre/loxP-mediated excision of the hygromycin-resistance gene upon induction. The remaining four lines initially displayed a mixture of red/green fluorescence after 2 weeks of treatment with the inducer, but propagation of the red color to the entire plant was observed in these lines 2 weeks after they were transferred from the inductive medium to the soil (Figure 2a, left). No difference was observed between 16c-GFPi seedlings germinated on the selective medium in the absence of the inducer (MS + hygromycin) and GFP-16c control plants, as judged by the degree of green fluorescence (Figure 2b, right).

Figure 2.

17β-estradiol-inducible GFP silencing in Nicotiana benthamiana transgenic line 16c-GFPi and Arabidopsis thaliana line At-GFPi.

(a) 16c-GFPi plants were first germinated on an inductive (left) or selective (right) medium for 2 weeks, before being transferred to soil, and photographed under UV light at 2 weeks after transfer. Left, an induced 16c-GFPi plant exhibited uniform red fluorescence in the whole plant. Note that in this plant initiation of GFP silencing was indicated by a mixture of red/green fluorescence. Right, an untreated 16c-GFPi plant showing green fluorescence. Scale bar = 0.5 cm.

(b, c) Northern blot analysis of induced silencing in 16c-GFPi plants (b) and At-GFPi plants (c). Total high (top and middle panels) and low (lower panel) molecular weight RNAs were extracted from 2-week-old red fluorescent seedlings after 17β-estradiol induction (+), untreated green fluorescent seedlings (−), and GFP-16c control plants (0). Equal RNA loading (5 µg per leaf) was monitored by methylene blue staining of the 28S RNA (middle panel), and the blot was probed with a GFP-specific sequence (top panels). Bottom panels show the blot for siRNA, which was hybridized with a 32P-labeled transcript corresponding to the 3′-terminal region (350 nt) of GFP. Each lane contained 50 µg RNA. Arrow indicates the position of a 25-base DNA oligonucleotide.

Northern blots were analyzed with total RNA from treated red fluorescent 16c-GFPi seedlings, untreated green fluorescent 16c-GFPi seedlings and GFP-16c control seedlings. GFP mRNA was almost undetectable in red 16c-GFPi seedlings (Figure 2b, upper panel, +), as compared with GFP-16c control (0) or untreated 16c-GFPi seedlings (–). GFP-related siRNA, which is a key component of RNA silencing, was detected in red seedlings using a radiolabeled GFP-specific probe (Figure 2b, lower panel, +), whereas no GFP-related siRNA could be detected in GFP-16c control or untreated 16c-GFPi green seedlings (0 and –). Both fluorescence and RNA analyses indicated silencing of the GFP transgene in 16c-GFPi plants upon treatment with 17β-estradiol.

A transgenic line of A. thaliana (ecotype C24), which showed constitutive expression of a 35S-GFP transgene (Dalmay et al., 2000) was transformed with the pX7-GFPi construct to test the inducible RNAi system. Twenty-nine T1 independent lines (named as At-GFPi) were obtained by floral dip transformation. Upon germination on the inductive medium, all 29 At-GFPi lines displayed red fluorescence, indicating silencing of the GFP transgene. By contrast, all seedlings germinated on the medium without the inducer showed uniform green fluorescence, indicating GFP expression (data not shown). Northern blots were analyzed with total RNA (Figure 2c, upper panel) and GFP-related siRNA (Figure 2c, lower panel) from treated, red fluorescent At-GFPi seedlings (Figure 2c, +) and untreated, green fluorescent At-GFPi seedlings (Figure 2c, –). Similar results were obtained as with the N. benthamiana 16c-GFPi plants, providing molecular evidence for silencing of the GFP transgene in At-GFPi plants upon 17β-estradiol treatment.

We used 15 At-GFPi lines for further investigation of 17β-estradiol-induced silencing at post-germination stage. Two-week-old T1 seedlings germinated on the selective medium in the absence of the inducer were transferred to fresh MS medium. All seedlings at this stage continued to display green fluorescence. However, when seedlings were transferred to the inductive medium, 10 lines displayed strong inducible GFP silencing as indicated by their uniform red fluorescence after 1-week induction (data not shown). The remaining five lines showed varying initiation of GFP silencing after 1-week induction. Further incubation with the inducer up to 2 weeks resulted in complete GFP silencing as reflected by the uniform red fluorescence in all these plants.

Efficient inducible silencing of an endogenous PDS gene in Arabidopsis thaliana

We chose the phytoene desaturase (PDS) gene of A. thaliana and N. benthamiana to test the ability of the inducible RNAi system to silence endogenous genes. The PDS gene was selected because loss of the phytoene desaturase enzyme blocks carotenoid synthesis culminating in a photobleaching phenotype because of photo-oxidation of chlorophylls (Ruiz et al., 1998). This visible phenotype facilitated visual monitoring of the induction process of PDS silencing. The constructs pX7-PDSi(At) and pX7-PDSi(Nb) (Figure 1b) were transformed into A. thaliana (ecotype Columbia) and N. benthamiana, respectively. In addition, pCAMBIA-PDSi(At) (Figure 1c) containing a 35S-PDSi(At) was also transformed into A. thaliana (ecotype Columbia).

We tested 35 lines of putative transgenic A. thaliana carrying the 35S-PDSi(At) transgene by virtue of their ability to grow on the selective medium. Thirty-two lines (35S-PDSi(At)) displayed the photobleaching phenotype. Most lines appeared near-white and stopped growing (data not shown) after 4–6 weeks on the culture medium. Only two lines that displayed varying green patches in their bleached leaves survived. Seedlings of these two lines exhibited abnormal development and poor fertility, and produced only a small amount of seeds.

Eighty-one independent A. thaliana transgenic lines transformed with pX7-PDSi(At) were obtained. In the absence of the inducer, all transgenic T1 lines (At-PDSi) displayed normal development and fertility. T2 seeds from 12 independent T1 lines were germinated on the selective medium in the absence or presence of the inducer. All T2 seedlings grew with normal phenotype on the medium in the absence of the inducer (Figure 3a). However, in the presence of the inducer, seedlings of all the 12 T2 At-PDSi lines showed uniform photobleaching phenotype in the cotyledons at 6–9 days post-induction (Figure 3b,c). Similar to the 35S-PDSi(At) lines, most of these seedlings stopped growing within 4–6 weeks (Figure 3d). Northern analysis showed that endogenous PDS mRNA levels were significantly reduced in bleached leaves of treated At-PDSi lines (Figure 3f, upper panel, +), but readily detected in untreated lines (Figure 3f, upper panel, –) and WT control seedlings (Figure 3f, upper panel, 0), indicating that the photobleaching phenotype resulted from silencing of the endogenous PDS gene. PDS-related siRNA was detected in treated, bleached At-PSDi seedlings (Figure 3f, lower panel, +) using a PDS-specific probe, whereas no PDS-related siRNA could be detected in either WT control seedlings (Figure 3f, lower panel, 0) or untreated At-PDSi seedlings (Figure 3f, lower panel, –).

Figure 3.

Inducible PDS silencing in transgenic Arabidopsis thaliana plants (At-PDSi) at seed germination stage.

(a–d) At-PDSi seeds of line 2 were germinated on selective media in the absence (a) or presence of an inducer (b, c, d).

(b) Seedlings with induced PDS silencing showed a uniform photobleaching phenotype in the cotyledons at 6 days.

(c,d) PDS silencing was seen in the first pair of true leaves as well as the leaves that developed later (c, 9 days, d, 28 days). Scale bars = 0.5 cm in (a, b, d) and 0.1 cm in (c).

(e) Seeds of At-PDSi line 1 (a single transgenic locus) were germinated on MS in the presence of 17β-estradiol without the selective antibiotic hygromycin. Fifteen out of 19 plants showed induced PDS silencing. The photograph was taken at 29 days post-induction. Scale bar = 0.5 cm.

(f) Northern analysis of inducible PDS silencing in At-PDSi plants. Total high (top and middle panels) and low (lower panel) molecular weight RNAs were extracted from WT Columbia seedlings (0), uninduced At-PDSi control seedlings (–), and a mixture of induced At-PDSi seedlings, which showed a photobleaching phenotype (+). The blot (top panel) was probed with a PDS-specific sequence corresponding to the 3′ region for detection of the endogenous PDS mRNA. Each lane contained 5 µg RNA, and 28S RNA visualized by methylene blue staining was used as a loading control. Bottom panel shows the blot for siRNA, which was hybridized with a 32P-labeled transcript corresponding to the 5′-terminal region (400 nt) of PDS. Each lane contained 50 µg RNA. The arrow indicates the position of a 25-base DNA oligonucleotide.

To rule out any toxic or non-specific secondary physiologic effects because of 17β-estradiol-induced PDS silencing, T2 seeds of At-PDSi line 1, which showed a 3 : 1 segregation ratio for HygR:HygS, were germinated on an inducer-containing medium without hygromycin, the selective antibiotic. After 3 weeks of incubation, 15 out of 19 T2 seedlings showed the photobleaching phenotype and ceased to grow (Figure 3e), whereas the remaining four seedlings were normal. These four seedlings were sensitive to hygromycin as they ceased to grow after being transferred to a hygromycin-containing medium. The approximately 3 : 1 segregation pattern of both the selection marker and the photobleaching phenotype suggested that the four hygromycin-sensitive plants were WT, in agreement with the Mendelian segregation ratio for a single transgenic locus. These four plants showed no response to 17β-estradiol and did not exhibit any morphological alteration, indicating that the inducer had no secondary non-specific physiological effect on WT plants. Our results suggest that the inducible RNAi system is able to silence the endogenous PDS of Arabidopsis at the seed germination stage with comparable efficiency and effectiveness as the constitutive 35S-PDSi transgene.

We also examined post-germination induction of PDS silencing. Two- or four-week-old T2 At-PDSi seedlings on the culture medium in the absence of the inducer were transferred to the inductive medium, and similar results were obtained from the seedlings of both age groups. We observed two photobleaching phenotypes. In the first group which includes At-PDSi lines 2, 5, 7, and 8, a strong PDS silencing was seen, 1 week after induction, with newly emerged leaves showing uniform bleaching surrounding the central area of the leaves. The bleaching was subsequently propagated to the entire leaf. Figure 4(a–d) shows results from one representative line 2. Most of these plants with a strong photobleaching phenotype stopped growing. The second group includes At-PDSi lines 1, 3, 4, 6, and 9–12, and these lines showed varying photobleaching phenotypes. After 2 weeks of induction, photobleaching was limited to the areas near the veins (Figure 4e, a plant of line 12) or to the white/green patchy regions in the entire leaf (Figure 4f, a plant of line 1); however, the leaves became near-white over the next 2 weeks (Figure 4g,h, a plant of line 1). Although the PDS silencing extended to most rosette leaves and some cauline leaves (Figure 4h), these plants with varying degree of PDS silencing could still develop normally and were fertile after transfer to the soil.

Figure 4.

Post-germination induction of PDS silencing in At-PDSi Arabidopsis thaliana transgenic plants.

(a–d) Strong induction of PDS silencing was seen in 2-week-old seedlings of line 2 after transfer to the inductive medium for 6 days (a, b) or 10 days (c). Note that (b) and (c) are from the same plant. Four-week-old seedlings of line 2 were transferred to the inductive medium and the picture was taken at 2 weeks post-induction (d).

(e–h) Delayed onset of PDS silencing. Four-week-old seedlings from different At-PDSi transgenic lines 12 (e) and 1 (f–h) were transferred to the inductive medium. Photobleaching was limited to the perivascular regions (e) or white/green patchy regions (f). The plants were photographed at 2 weeks after induction. Photobleaching progressed with time in the old leaves and extended to new rosette and cauline leaves when plants were photographed at 3 weeks (g) or 4 weeks (h) after induction. All scale bars represent 0.5 cm.

We collected seeds from six T2 plants of At-PDSi line 1 with induced silencing. When germinated on the selective medium (MS + Hyg), none of the T3 progeny showed PDS silencing. When the seeds were germinated on MS medium with neither the selective antibiotic nor the inducer, four T2 lines showed no PDS silencing. On the other hand, more than 10% of the progeny seedlings of the other two T2 lines showed a constitutive photobleaching phenotype. The constitutive bleached plants presumably derived from some converted germ line cells in the L2 layers of T2 plants that have undergone 17β-estradiol-induced Cre/loxP DNA recombination (Zuo et al., 2001), and therefore expressed G10-90-PDSi constitutively (Figure 1b).

Molecular analyses of inducible PDS silencing in At-PDSi plants

The delayed onset of PDS silencing in the second group prompted us to analyze the relationship between the photobleaching phenotype, endogenous PDS mRNA levels, and dsRNA induction upon 17β-estradiol-induced Cre/loxP DNA excision. T2 progenies of At-PDSi line 1 were analyzed in detail. First, dsRNA corresponding to the PDSi transcript region (Figure 1) was analyzed. Two-week-old seedlings of line 1-1 (heterozygous) and line 1-2 (homozygous) were transferred to the inductive medium. RNA was extracted from a portion of the seedlings at 42 h post-induction. Total RNA was digested with Rnase1™ (Promega, USA), and dsRNA was analyzed by hybridization with a PDS 5′-terminal probe containing sequences corresponding to the PDSi region (Figure 1). Figure 5(b) shows that signals of the expected size were detected in the treated seedlings of both lines 1-1 and 1-2 (lanes 2 and 3), but not in the untreated seedlings of line 1-1 (lane 1).

Figure 5.

Molecular characterization of delayed onset of PDS silencing in A. thaliana transgenic plants.

(a) Two-week-old seedlings of At-PDSi line 1-1 were transferred to the inductive medium, and the plant was photographed at 18 days post-induction. Scale bar represents 0.5 cm. Leaves with increasing degrees of photobleaching were labeled with numbers 1–4, and those with a similar phenotype were labeled with the same number. Leaves designated 1 were formed before seedlings were transferred to the inductive medium.

(b) Seedlings of At-PDSi line 1-1 and 1-2 were either untreated (–) or treated (+) for 42 h with 17β-estradiol, RNA was analyzed for the presence of pre-silencing dsRNA. Lane 1, line 1-1, untreated, 100 µg RNA; lane 2, treated line 1-1, 50 µg RNA; lane 3, treated lane 1-2, 100 µg RNA. RNAs were digested with Rnase1™ before being loaded onto formaldehyde-containing gels. The filter was hybridized with a PDS-specific 5′ sequence corresponding to the PDSi region (Figure 1a). Molecular weight standards are shown on the left.

(c) PCR analysis of genomic DNA prepared from WT Columbia (lane 1), untreated line 1-1 (lane 2), 18-day 17β-estradiol-treated line 1-1 (lanes 3–6) or line 1-2 (lanes 7–10). Leaves with similar phenotype (panel a) were pooled from several plants, and the group numbers indicated on top of the blot. Primers P1–4 were used as illustrated in Figure 1(b). The expected PCR products from different combinations of primer pairs were indicated on the right, M, DNA molecular markers.

(d) Northern analysis of endogenous PDS mRNA levels. RNAs were extracted from the same samples as for DNA preparation (panel c), and the corresponding group numbers are indicated on the top. Each lane contained 5 µg RNA.

(e) Showing one plant of At-PDSi line 1-1 treated with 17β-estradiol for 18 days as described in (a) one week after been transferred to high light intensity. Note that the silenced leaves including the old ones (one indicated by arrow) became strongly bleached. Compare this plant to those shown in Figure 4(g,h). Scale bar = 0.5 cm.

(f) Molecular analysis of inducer-treated transgenic plants 1 week after inducer withdrawal and transfer to the soil. Some seedlings of At-PDSi line 1-1 treated with 17β-estradiol for 18 days as described in (a) were transferred to the soil. After 1 week, leaves belonging to group 4 as well as two younger leaves that displayed photobleaching phenotype were pooled for PCR (left panel) and Northern blot analysis (right panel). Untreated line 1-1 was used as controls (–). For right panel, each leaf contained 5 µg RNA.

For the remaining seedlings on the inductive medium, new leaves that emerged after 1-week induction also became patchy in appearance, and photobleaching progressed with time in both the heterozygous T2 line 1-1 and the homozygous T2 line 1-2, similar to that in the parental T1 line 1. To facilitate further analysis, leaves with varying degrees of photobleaching were numbered as shown in Figure 5(a).

PCR analysis was performed using primers specific for the excised sequences and flanking non-excised sequences (see Figure 1b). Based on a previous study (Zuo et al., 2001), the P1/P2 and P3/P4 primer pairs were expected to yield PCR fragments of 696 and 1331 bp, respectively, from a non-recombinant T-DNA. By contrast, following Cre/loxP-mediated recombination and reconstitution of the G10-90-PDSi transcription unit, the P1/P4 primer pair produced a PCR product of 992 bp (Figure 1b). Leaves with similar phenotype were pooled from several plants for DNA preparation at 18 days post-induction. No fragment was amplified from WT Columbia (Figure 5c, lane 1), whereas P1/P2 and P3/P4 fragments were detected in untreated line 1-1 (Figure 5c, lane 2), indicating no DNA recombination. The P1/P4 fragment was detected in bleached (group 2 leaves) and patchy leaves (group 3 leaves) of both line 1-1 (Figure 5c, lanes 4 and 5) and line 1-2 (Figure 5c, lanes 8 and 9), indicating complete DNA excision in these leaves after 2 weeks of inducer treatment. In the near-green young leaves (group 4 leaves), however, varying amounts of the DNA excision were found (Figure 5c, lanes 6 and 10). There was no DNA excision in the sample collected from group 4 leaves of line 1-1 because only P1/P2 and P3/P4 fragments were amplified (Figure 5c, lane 6). By contrast, P1/P2, P3/P4, and P1/P4 fragments were detected in the samples collected from group 4 leaves of line 1-2 (Figure 5c, lane 10), indicating that DNA recombination occurred in some of the leaves.

We also found that DNA excision occurred in the old leaves (group 1 leaves) that were already developed before plants were transferred to the inductive medium. P1/P4 fragment was amplified from both line 1-1 (incomplete excision; Figure 5c, lane 3) and line 1-2 (complete excision; Figure 5c, lane 7), consistent with the highly effective inducer-dependent DNA recombination, probably as a result of penetration of 17β-estradiol to almost all cells in the lower leaves.

Endogenous PDS mRNA levels of T2 line 1-2 were assessed by Northern analysis 18 days after induction. Consistent with the photobleaching phenotype and DNA excision, PDS mRNA levels were significantly decreased in groups 2 and 3 leaves (Figure 5d, lanes 4 and 5), and slightly decreased in group 4 leaves compared to that in untreated line 1-1 control seedlings (lane 2) and WT Columbia control seedlings (lane 1). As expected, PDS mRNA was also degraded in group 1 old leaves (Figure 5d, lane 3), consistent with the DNA excision assay. Although PDS silencing in old leaves was indeed induced after treatment, the persistence of the green color presumably resulted from the continued presence of carotenoids that were synthesized before the T2 seedlings were transferred to the inductive medium. This residual amount of pre-formed carotenoids was able to protect chlorophylls from photo-oxidation under low light intensity (our assay condition: 35 µmol sec−1 m−2). However, when these plants were transferred to high light intensity (70 µmol sec−1 m−2), all leaves including the old ones became strongly bleached (Figure 5e).

Some line 1-1 seedlings treated with the inducer for 18 days were transferred to the soil, and the phenotype of group 4 leaves as well as that of the newly emerged rosette leaves was followed. Group 4 leaves became patchy in appearance, and they became progressively bleached until near-white in appearance in 1 week. Similar results were observed in another two newly grown rosette leaves, which displayed limited areas of photobleaching. DNA and RNA were extracted from new photobleached leaves (including group 4 leaves), and PCR analysis showed that only P1/P2 and P3/P4 fragments were amplified (Figure 5f), indicating no DNA excision in these leaves. However, RNA analysis demonstrated a severe reduction in PDS mRNA levels, indicating that the PDS gene was silenced presumably via signals generated by the lower leaves.

Inducible silencing of endogenous PDS gene in transgenic Nicotiana benthamiana

Similar results were obtained in N. benthamiana transformed with the pX7-PDSi(Nb) construct. Inducible PDS silencing was observed when the inducer was given at germination (Figure 6a) or post-germination stages (Figure 6b). Northern analysis of endogenous PDS mRNA levels (data not shown) confirmed the visual photobleaching phenotype, suggesting that the inducible RNAi system is capable of inducing endogenous gene silencing in N. benthamiana and probably other plant species as well.

Figure 6.

Inducible PDS silencing with uniform photobleaching phenotype in one Nicotiana benthamiana transgenic (Nb-PDSi) line.

(a) Seeds were germinated on the selective medium in the presence of the inducer (MS + Hyg + 17β-estradiol), and seedlings were photographed at 2 weeks post-germination.

(b) Two-week-old seedlings were transferred to the inductive medium without hygromycin, and seedlings were photographed at 2 weeks post-induction. The green plant is presumably a WT plant that exhibited no secondary effect upon 17β-estradiol treatment. Scale bar = 0.5 cm.


In this work, we report the development of an inducible dsRNA-mediated silencing (inducible RNAi) system for conditional gene silencing in transgenic plants. This system contains two steps: (i) the inducible expression of dsRNA resulting from 17β-estradiol-induced DNA recombination and (ii) induction of target gene silencing by the dsRNA. A detailed characterization of the 17β-estradiol-inducible silencing system in transgenic A. thaliana and N. benthamiana plants demonstrated that this system is stringently controlled and can produce, at high efficiency, conditional silencing of both a GFP transgene and an endogenous PDS gene and without any detectable secondary affect. At seed germination stage, all tested At-PDSi lines showed a uniform photobleaching phenotype similar to that obtained with constitutively silenced 35S-PDSi lines. This suggests that the efficiency and effectiveness of the inducible RNAi system against endogenous genes are comparable to those obtained with constitutive expression of dsRNA (constitutive RNAi).

The highly efficient gene silencing obtained by intron-containing dsRNA expressed from a constitutive promoter would produce loss-of-function transgenic plants similar to null mutants. If the target gene is essential for basic cell function and development, constitutive RNAi would probably prevent shoot regeneration, cause plant lethality or embryogenesis defect, and block the ability of the transformed plants to produce subsequent generations. These problems can be avoided by using the inducible RNAi system described here. Moreover, the inducible RNAi system provides the possibility to induce gene silencing at different stages of plant development post-germination. This is demonstrated by the At-PDSi lines described here, in which a portion of the tested 2- or 4-week-old seedlings showed strong photobleaching after induction similar to that shown by transgenic plants with 35S-PDSi phenotypes. Moreover, the varying initiation and degrees of inducible gene silencing generate plants with a range of loss-of-function phenotype much like an allelic series, facilitating functional analysis.

Both heterozygous (line 1-1) and homozygous (line 1-2) progeny of At-PDSi line 1 retain the ability to reproduce inducible PDS silencing as the parental line, indicating that the inducible RNAi can be transmitted to the next generation. Having stable and reproducible RNAi transgenic lines would allow genetic crosses to be made and investigations of gene functions to be carried out with subsequent generations. Other silencing induction systems, such as Agrobacterium infiltration (Voinnet and Baulcombe, 1997) and infection with vectors derived from RNA viruses (Dalmay et al., 2000; Ratcliff et al., 2001) or DNA viruses (Peele et al., 2001; Turnage et al., 2002), have limited utility for functional genomics because of their transient nature, and the silencing effects are not heritable. Moreover, Agrobacterium infiltration cannot be applied to Arabidopsis because of its small plant size and clumpy rosette leaves. In the case of virus-induced gene silencing, some infected plants display stunted growth and they often do not produce inflorescences, flowers, or seeds (Turnage et al., 2002). Moreover, disease symptoms caused by viral infection could confound the interpretation of the phenotype as a result of silencing of the target gene. By contrast, the inducible RNAi system described here can be triggered by simply treating plants with the inducer for a certain period of time. Treated plants display a specific silencing phenotype without any non-specific effects (see Figures 2a, 4e–h and 6b).

We have chosen the inducible Cre/loxP DNA excision system (CLX) rather than the XVE transient inducible system (Zuo et al., 2000) to produce the inducible RNAi for several reasons. The mechanism for RNA silencing involves an initial induction process followed by a systemic spread of the silencing signal (for review, see Mlotshwa et al., 2002). The local initiation of gene silencing by dsRNA is equally effective for both transgene and endogenous genes (Smith et al., 2000; Wesley et al., 2001). However, unlike the widespread, persistent silencing observed for a GFP transgene, systemic silencing of endogenous genes was transient and limited (Palauqui and Vaucheret, 1998; Voinnet et al., 2000). Amplification of the signal is necessary for efficient systemic silencing (Palauqui and Vaucheret, 1998; Voinnet et al., 1998). It has been proposed that the silencing signal is perpetuated by transgenes, but not by endogenous genes (Fagard and Vaucheret, 2000; Mlotshwa et al., 2002). The CLX system can rescue the weak amplification of the silencing signal by endogenous genes. Once the Cre/loxP-mediated DNA excision occurred upon inducer treatment, expression of the downstream intron-containing dsRNA would be permanently activated. This situation mimics the expression of an RNAi using a constitutive promoter, which is the most efficient, effective, and high-throughput system for gene silencing, and the systemic spread of silencing signal is therefore not required. Expression of dsRNA from the inducible XVE system (Zuo et al., 2000) without DNA recombination would require repeated applications of 17β-estradiol to the transgenic plants for sustained expression, and this may not be practical for plants growing in the soil.

The strong photobleaching induction in At-PDSi lines (lines 2, 5, 7, and 8) at post-germination stages is believed to be a result of complete DNA excision, which would reconstitute the G10-90-PDSi transcription unit to produce PDSi transcripts constitutively throughout the entire plant. In this case, no silencing signal amplification was required. This interpretation is consistent with our PCR analysis of At-PDSi lines (second group), displaying a delayed onset of gene silencing. In both line 1-1 and line 1-2 leaves, complete excision of DNA within the loxP sites showed strong photobleaching or patchy PDS-silencing phenotype. In near-green leaves where incomplete or no DNA excision was detected, only a weak systemic spread (weak signal amplification) of PDS silencing to upper rosette leaves was seen during the 18-day incubation with the inducer (Figure 5a–d). However, upon an additional week of growth after inducer withdrawal, strong PDS gene silencing was also detected in group 4 leaves and in the next two younger leaves (Figure 5f). These results suggest a limited systemic translocation of the silencing signal to two to three upper leaves.

Because of 17β-estradiol instability (Zuo et al., 2000), a second or even multiple treatment with fresh inducer may be needed to fully reactivate this RNAi system in some transgenic lines. With appropriate improvement of the induction conditions, a higher DNA excision efficiency, and therefore a higher proportion of lines showing strong induction, may be obtained. Nevertheless, the incomplete Cre/loxP DNA excision, which results in genetic chimera in transgenic plants, may provide a useful system to study mechanisms of long-distance signal transduction in gene silencing in Arabidopsis, which is difficult to graft (Turnbull et al., 2002). The mechanisms involved in systemic RNA silencing in plant systems are being actively investigated using grafting and transient expression approaches with N. benthamiana or N. tabacum (Guo and Ding, 2002; Mallory et al., 2001; Voinnet et al., 2000). No mutations specific to systemic silencing have yet been reported in plant systems. Because of the small plant size and the clumpy rosette leaves, it is impossible to carry out localized infiltration of Arabidopsis with Agrobacterium, and grafting manipulation in Arabidopsis is also a challenging task. For these reasons, the ability to generate genetic chimera in transgenic Arabidopsis producing RNAi only from treated tissues would be very useful for future investigations. As shown by PCR analysis in At-PDSi line 1-1 and line 1-2, DNA excision occurred in lower rosette leaves, but not in upper rosette leaves. Gene silencing resulting from local RNAi induction (complete excision, Figure 5c,d) or signal-mediated long-distance (no excision) induction (Figure 5f) can be predicted by simple PCR analysis. The ability to generate genetic chimera in Arabidopsis may also find useful applications in research on other types of long-distance signaling (e.g. flowering time) in plants.

Experimental procedures

Plasmid construction

DNA manipulations and cloning were carried out using standard procedures (Sambrook et al., 1989). The third intron of Arabidopsis actin gene 11 (ATU27981, nt 1957–2111) was selected for the intron-containing intermediate construct (pSK-int). This intron was amplified by PCR using two primers: Pint5′, 5′-TACGTAAGTAGATCTTCAACACC-3′; and Pint3′, 5′-GGAATTCTGCAAACACACAAGACAAT-3′. The primers were designed such that their border sequences contained the consensus sequence (bold letters) for plant introns: AG//GTAAGT…TGCAG//G (Shapiro and Senapathy, 1987). Two restriction sites SnaBI and EcoRI (underlined) were added for cloning purposes. A PCR fragment of 155 bp was digested with SnaBI/EcoRI and cloned into EcoRV/EcoRI-digested pBluscript II SK+ to yield the intermediate construct pSK-int (Figure 1a).

To clone sequences encoding the inverted-repeat RNA into the pSK-int intermediate vector, a 357 bp fragment corresponding to nucleotides (nt) 360–716 of the GFP 3′-terminal sequence (Voinnet and Baulcombe, 1997) was cloned into the 5′ and 3′ arms of the intron (Figure 1a), and the resulting plasmid was named as pSK-GFPi. The 5′-terminal sequences of PDS of A. thaliana and N. benthamiana were obtained by RT-PCR amplification with specific primers. For A. thaliana PDS, the primers were: Pat5′, 5′-GACTAGTATGGTTGTGTTTGGGAATG-3′; and Pat3′, 5′-GATATCCTTCCATGCAGCTATC-3′. These primers were used to obtain a fragment of 405 bp corresponding to nt 128–532 of the A. thaliana PDS cDNA (L16237), and SpeI and EcoRV restriction sites (underlined) were added to the cDNA fragment. For PCR amplification of the N. banthamiana PDS sequence, the primers were: Pnb5′, 5′-GACTAGTATGCCTCAAATTGGACTTGT-3′; and Pnb3′, 5′-CAGCTGTAGACAAACCACCCAAAC-3′ homologous to regions of the tomato PDS cDNA (M88683) nt 318–337 and nt 676–696, respectively. These primers were designed with the addition of SpeI and PvuII restriction sites (underlined). Using N. benthamiana RNA as templates, a 386 bp fragment was obtained with RT-PCR, whose sequence exhibited high homology with the tomato PDS cDNA. RT-PCR fragments derived from A. thaliana (At) and N. benthamiana (Nb) were cloned into the pCR-Blunt vector (Invitrogen, USA) to give pCR-PDS(At) and pCR-PDS(Nb), respectively. Fragments of SpeI–EcoRI and HindIII–XhoI were inserted into both arms of the intron of pSK-int digested with the appropriate restriction enzymes as shown in Figure 1(a) to obtain pSK-PDSi(At) and pSK-PDSi(Nb), respectively.

For inducible dsRNA transformation constructs, the kanamycin-resistance gene in pX6-GFP (Zuo et al., 2001) was replaced with a hygromycin-resistance gene, and the derivative called pX7-GFP. To create an inducible expression of intron-containing dsRNA, fragments of XhoI–XbaI from pSK-GFPi, pSK-PDSi(At), and pSK-PDSi(Nb) were subcloned into pX7-GFP digested with XhoI/SpeI (XbaI and SpeI are compatible), resulting in pX7-GFPi, pX7-PDSi(At), and pX7-PDSi(Nb), respectively (Figure 1b).

In addition, the PstI–SacI fragment from pSK-PDSi(At) was cloned into a modified binary vector pCAMBIA-1300 (AF234296), which contained a 35S promoter and a 35S terminator, to give pCAMBIA-PDSi(At), which is a constitutive RNAi construct (Figure 1c).

Upon request, vectors described in this paper are available to academic researchers for non-commercial projects.

Plant materials, transformation, and growth conditions

A transgenic N. benthamiana line (GFP-16c) carrying a 35S-GFP transgene with a kanamycin-selectable marker at a single locus in homozygous condition (Ruiz et al., 1998) and a transgenic A. thaliana ecotype C24 line (Dalmay et al., 2000) carrying a similar transgene were used for pX7-GFPi transformation. A. thaliana ecotype Columbia was used for pX7-PDSi(At) and pCAMBIA-PDSi(At) transformation. WT N. benthamiana was used for pX7-PDSi(Nb) transformation. N. benthamiana transformation was carried out by co-culture with Agrobacterium, whereas A. thaliana was transformed by the floral dip method (Clough and Bent, 1998). The selective medium contained MS medium plus hygromycin (20 mg l−1 for A. thaliana and 40 mg l−1 for N. benthamiana), whereas the inductive medium contained, in addition, 17β-estradiol (2 µm). GFP fluorescence was examined using a 100 W hand-held long-wavelength UV lamp.

Analyses of RNA and DNA

Total RNA was isolated from plant tissues by LiCl precipitation (Verwoerd et al., 1989). The LiCl supernatant fraction was precipitated with 3 volumes of ethanol to obtain genomic DNA and low molecular weight RNA (siRNA). dsRNA was obtained by digesting total RNA with Rnase1™ (Promega, USA) (0.5 U Rnase µg−1 total RNA) at 37°C for 3 h. For Northern analysis, total RNA or dsRNA was separated on 1.2% agarose formaldehyde gels, transferred to Hybond-N+ membranes, and hybridized with 32P-labeled cDNA probes specific for the respective RNA. Low molecular weight RNA analysis was done as described (Hamilton and Baulcombe, 1999; Llave et al., 2000). The probes for GFP and PDS siRNA were 32P-labeled 3′-terminal 356 nt of GFP or 5′-terminal 400 nt of PDS antisense RNA, respectively, transcribed by T7 RNA polymerase. The PCR analysis with approximately 200 ng of genomic DNA was subjected to 94°C for 20 sec, 50°C for 20 sec, and 72°C for 2 min for 30 cycles. Primers for PCR analysis were: P1, 5′-GCCGCCACGTGCCGCCACGTGCCGCC-3′; P2, 5′-CTCGTCAATTCCAAGGGCATCGGT-3′; P3, 5′-CTGGACACAGTGCCCGTGTCGGA-3′; P4, identical to Pint3′ for intron amplification (see Results).


We thank Dr David Baulcombe for seeds of transgenic N. benthamiana (line 16c) and transgenic A. thaliana carrying a 35S-GFP transgene.