allen et al., 2007

8/6/2019 Allen et al., 2007

1/12

THE JOURNAL OF GENE MEDICINE R E S E A R C H A R T I C L EJ Gene Med 2007; 9: 287298.Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/jgm.1018

Development of strategies for conditional RNAinterference

Danny Allen*

Paul F. Kenna

Arpad Palfi

Helena P. McMahon

Sophia Millington-WardMary OReilly

Pete Humphries

G. Jane Farrar

Department of Genetics, Trinity

College Dublin, Dublin 2, Ireland

*Correspondence to: Danny Allen,

Department of Genetics, Trinity

College Dublin, Dublin 2, Ireland.

E-mail: [email protected]

Received: 19 September 2006

Revised: 22 December 2006

Accepted: 22 January 2007

Abstract

Background RNA interference (RNAi) represents a powerful tool with

which to undertake sequence-dependent suppression of gene expression.

Synthesized double-stranded RNA (dsRNA) or dsRNA generated endoge-nously from plasmid or viral vectors can be used for RNAi. For the latter,

polymerase III promoters which drive ubiquitous expression in all tissues

have typically been adopted. Given that dsRNA molecules must contain few

5 and 3 over-hanging bases to maintain potency, employing polymerase II

promoters to drive tissue-specific expression of RNAi may be problematic due

to potential inclusion of nucleotides 5 and 3 of siRNA sequences.

Methods To circumvent this, polymerase II promoters in combination with

cis-acting hammerhead ribozymes and short-hairpin RNA sequences have

been explored as a means to generate potent dsRNA molecules in tissues

defined by the promoter in use.

Results The novel constructs evaluated in this study produced functional

siRNA which suppressed the enhanced green fluorescent protein (eGFP) both

in vitro and in vivo (in mice). Additionally, the constructs did not appear

to elicit a significant type-1 interferon response compared to traditional

H1-transcribed shRNA.

Conclusions Given the potential off-target effects of dsRNAs, it would

be preferable in many cases to limit expression of dsRNA to the tissue of

interest and moreover would significantly augment the resolution of RNAi

technologies. Notably, the system under evaluation in this study could readily

be adapted to achieve this objective. Copyright 2007 John Wiley & Sons,

Ltd.

Keywords RNA interference; ribozymes; polymerase II promoters; gene therapy

Introduction

Use of double-stranded RNA (dsRNA) to modulate gene expression, or

RNA interference (RNAi), has been adopted extensively subsequent to the

breakthrough paper of Andrew Fire and colleagues in 1998 demonstrat-

ing potent sequence-specific gene silencing in Caenorhabditis elegans [1].

Observation of an interferon response in mammalian cells to long dsRNA

molecules precipitated the use of short dsRNA molecules or small inter-

fering RNA (siRNA) to circumvent this response [2]. Much of the initialresearch was focused on exploration of the utility of synthesized siRNAs,

which elicit transient suppression of a target gene. The transience of siRNA-

based suppression promoted the emergence of chemically protected siRNA

Copyright 2007 John Wiley & Sons, Ltd.

8/6/2019 Allen et al., 2007

2/12

288 D. Allen et al.

molecules incorporating modifications thereby providing

molecules with longer half-lives [3].

An alternative means of overcoming transient suppres-

sion involves engineering vector(s) from which functional

siRNAs may be expressed [4 9]. Short hairpin RNAs

(shRNAs) expressed from such vectors are processed

intracellularly into functional siRNAs [10]. Expression

of shRNAs can be achieved using various systems; e.g.

ubiquitously expressing polymerase III promoters such

as the H1 or U6 promoters which utilize short initia-

tion and termination signals have been employed [48].

However, the majority of mammalian promoters are poly-

merase II promoters and typically use more complex

initiation and termination signals than their polymerase

III counterparts. The cytomegalovirus (CMV) promoter, a

ubiquitous polymerase II promoter, has been employed to

drive expression of shRNAs; typically, shRNA sequences to

be expressed have been placed juxtaposed to the transcrip-

tional start site [11]. In contrast to CMV, transcriptionalstart sites for many polymerase II promoters are ill defined

or multiple transcriptional start sites may be utilized.

Nevertheless, there are clear advantages to targeting

RNAi-based suppression to specific tissues particularly

in the light of potential off-target effects associated

with RNAi-based suppression identified from microarray

studies [12]. Additionally, such tissue specificity would

increase the resolution of RNAi technology, in principle,

mirroring the resolution achieved with conditional gene

targeting.

Materials and methods

Construction of RNAi plasmids

Polymerase chain reactions (PCRs) were carried out

under the following conditions. The PCR reaction had

a final volume of 20 l containing 5 pmol of forward

and reverse primer, 0.5pmol of forward and reverse

template oligonucleotide, 0.2 M dATP, dTTP, dCTP, and

dGTP, 5 l 10 buffer and 1 unit of Taq polymerase

(10 buffer = 500 mM KCl, 100 mM Tris, pH 9, 0.1%

gelatin (v/v), 1% Triton X and 15 mM MgCl2

). The

standard PCR cycle run was; 94 C for 5 min followed

by 35 cycles of 94 C for 1 min, 55 C for 1 min and 72 C

for 1.2 min, ending with a final step of 72 C for 10 min.

All primer and template oligonucleotide sequences are

detailed in Table 1. PCR products were cloned into

pCDNA3.1 (Invitrogen) using restriction enzyme sites

incorporated into the forward and reverse primers. Colony

PCR screens were carried out to detect positive colonies.

DNA from positive colonies was sequenced to ensure

synthesis fidelity of the cloned PCR products.

Cell transfection and RNA extraction

HeLa cells were cultured in DMEM+ (500 ml Dulbeccos

modified Eagles medium (DMEM), 50 ml foetal calf

serum (FCS), 5 ml 100 mM sodium pyruvate, 5 ml

L-glutamine and 5 ml penicillin/streptomycin) using

standard procedures. All co-transfections (enhanced

green fluorescent protein (eGFP) plasmid + test/control

plasmid) were carried out in triplicate in six-well plates.

Twenty-four hours prior to transfection, 5 105 cells

were plated in each well of a six-well plate and

grown under standard conditions minus antibiotics.

Transfections were carried out using Lipofectamine 2000

as per the manufacturers instructions (Invitrogen).

RNA extraction

Twenty-four hours post-transfection total RNA was

isolated from cells using TRI reagent (MRC) as per the

manufacturers instructions. RNA was then treated with

RNase-free DNase (Promega). RNAs extracted from HeLa

cells were resuspended in 100 l of nuclease-free water

and stored at 80 C. When enriching for small RNAs the

mirVana miRNA isolation kit was used, instead of TRI

reagent, as per the manufacturers instructions (Ambion).

RNA extraction and real-timereverse-transcription (rt)PCR

RNA levels were analyzed using real-time rtPCR carried

out on a Lightcycler (Roche Diagnostics) using the

Table 1. Primer and template oligonucleotide sequences used in this study

tsRNAi Primer Forward CTAGCTAGCTCTAGAGGAtsRNAi Primer Reverse CCCAAGCTTGAATTCCACtsRNAi A eGFP Oligo CTAGCTAGCTCTAGAGGATCCAGTAGCTGATGAGTCCGTGAGGACGAAACGGTACCCGGTAGCAAGCTGACCCTGAAGTTCATCAGA

AGAGAACTTCAGGGTCAGCTTGCCGGTCGACGGATCATGATCCGTCCTGATGAGTCCGTGAGGACGAAACAACGAATTCAAGCTTtsRNAi B eGFP Oligo CCTAGCGGATCCAGTAGCTGATGAGTCCGTGAGGACGAAACGGTACCCGGTACCGTCGCAAGCTGACCCTGAAGTTCATGACGGATC

TAGATCCGTCCTGATGAGTCCGTGAGGACGAAACTGGGTCGCTAAAGCCCACCCAGCTGATGAGTCCGTGAGGACGAAACGGTACCGTCGAACTTCAGGGTCAGCTTGCCGGACGGATCATGATCCGTCCTGATGAGTCCGTGAGGACGAAACAACCACCAACGAATTCAAGCTT

tsRNAi C eGFP Oligo CCTAGCGGATCCAGTAGCTGATGAGTCCGTGAGGACGAAACGGTACCCGGTACCGTCCTACTGCAAGCTGACCCTGAAGTTCATCAGAAGAGAACTTCAGGGTCAGCTTGCCGAATAAAGGAAATTTATTTTCATTGCAATAGTGTGTTGGTTTTTTGTGTGAAGCTTCCTAGC

tsRNAi A non Oligo CTAGCTAGCTCTAGAGGATCCAGTAGCTGATGAGTCCGTGAGGACGAAACGGTACCCGGTATTCTCCGAACGTGTCACGTTTCAGAAGAACGTGACACGTTCGGAGAATTGTCGACGGATCATGATCCGTCCTGATGAGTCCGTGAGGACGAAACAACGAATTCAAGCTT

tsRNAi B non Oligo CCTAGCGGATCCAGTAGCTGATGAGTCCGTGAGGACGAAACGGTACCCGGTACCGTCTTCTCCGAACGTGTCACGTTTGACGGATCTA

GATCCGTCCTGATGAGTCCGTGAGGACGAAACTGGGTCGCTAAAGCCCACCCAGCTGATGAGTCCGTGAGGACGAAACGGTACCGTCACGTGACACGTTCGGAGAATTGACGGATCATGATCCGTCCTGATGAGTCCGTGAGGACGAAACAACCACCAACGAATTCAAGCTT

tsRNAi C non Oligo CCTAGCGGATCCAGTAGCTGATGAGTCCGTGAGGACGAAACGGTACCCGGTACCGTCCTACTTTCTCCGAACGTGTCACGTTTCAGAAGAACGTGACACGTTCGGAGAATTAATAAAGGAAATTTATTTTCATTGCAATAGTGTGTTGGTTTTTTGTGTGAAGCTTCCTAGC

Copyright 2007 John Wiley & Sons, Ltd. J Gene Med 2007; 9: 287298.

DOI: 10.1002/jgm

8/6/2019 Allen et al., 2007

3/12

Polymerase II Promoter-Driven RNAi 289

Table 2. Real-time rtPCR primer sequences

eGFP Forward TTCAAGGAGGACGGCAACATCCeGFP Reverse CACCTTGATGCCGTTCTTCTGCActin Forward TCACCCACACTGTGCCCATCTACGAActin Reverse CAGCGGAACCGCTCATTGCCAATGGGAPDH Forward GAAGGTGAAGGTCGGAGTC

GAPDH Reverse GAAGATGGTGATGGGATTTCPPIA Forward CCCACCGTGTTCTTCGACATPPIA Reverse CCAGTGCTCAGAGCACGAAAOAS1 Reverse CAGCTTCGTACTGAGTTCGCOAS1 Reverse TAGTTCTGTGAAGCAGGTGGIFITM1 Forward CAACACCCTCTTCTTGAACTGGIFITM1 Reverse AGATGTTCAGGCACTTGGCGISGF3 Forward AAGTACCATCAAAGCGACAGCISGF3 Reverse CATTATTGAGGGAGTCCTGGEIF2AK2 Forward ACCTCAGTGAAATCTGACTACCEIF2AK2 Reverse CAGATGATGATTCAGAAGCG

QuantiTect SYBR Green RT-PCR kit (Qiagen). Samples

were run under the following conditions; an rt step (50C,

20 min), followed by an inactivation step (95

C, 15 min)and 35 cycles of an amplification step (94 C, 15 s;

55 C, 20 s; 72 C, 4 s). Mean values, standard deviations

and t-tests were calculated using DataDesk v6.0 (see

below). Differences in levels of expression observed

between samples were deemed statistically significant

at p < 0.05. All real-time rtPCR primer sequences are

detailed in Table 2. The cycle-threshold method of real-

time rtPCR analysis was used for both mRNA relative

quantification and interferon-related gene expression. The

2CT method of determining mean fold changes in gene

expression was used to analyze the interferon-related

gene expression.

Hydrodynamic tail-vein injection

Endotoxin-free plasmid DNA was diluted in sterile

phosphate-buffered saline (PBS). Total injection volumes

per mouse (in ml) were calculated by dividing the weight

of the mouse (in g) by 10. Unanesthetized mice were

restrained in a 50 ml syringe barrel with rubber bungs.

Once in position, a lamp was used to warm the mouse-tail

and thereby dilate the tail vein facilitating visualization

and injection with a 3-ml syringe, fitted with a 0.75-cm

26-gauge needle. The syringe needle was placed into the

dilated tail vein and nearly the full length of the needle

inserted into the vein. Once inserted, the complete volume

of solution was injected into the tail vein within 35 s.

Notably, hydrodynamic injection appears to result in a

temporary minor heart failure in the mouse, resulting in

retrograde blood flow which pushes the injected sample

back into the liver.

Direct inferior vena cava injection

Endotoxin-free plasmid DNA was diluted in sterile

Hartmanns solution. Total injection volume per mouse was 500l. Mice were anesthetized prior to injection

using ketamine/xylazine. In anesthetized mice the inferior

vena cava (IVC) was exposed using a midline incision. A

1-ml syringe, fitted with a 30-gauge needle, was used

to deliver the DNA solution. Once the needle had been

inserted, the complete volume of solution was injected

into the IVC within 10 s.

RNA extraction from mouse organs

Forty-eight hours post injection, mice were sacrificed

using CO2 asphyxiation. Organs collected from mice were

homogenized in 1 ml of TRI reagent (MRC) using sterile

polypropylene pellet pestles (Sigma Aldrich). Total RNA

was isolated as per the manufacturers instructions. RNA

was then treated with RNase-free DNase (Promega). RNA

extracted from mouse liver was resuspended in 500 l of

nuclease-free water and stored at 80 C.

Tissue fixation and sectioningOrgans harvested for sectioning were preserved in 4%

paraformaldehyde and stored at 4 C in the dark until

embedded in 6% agar for sectioning. Sections of 100 m

thickness were cut on a vibratome (Leica VT1000s) at

a blade vibrating frequency of 70 Hz and a speed of

0.75 mm/s. Analysis of sections was performed with a

fluorescence microscope (Axiophot; Zeiss Ltd., UK) at

12.5 magnification.

Statistical analysis

Statistics were performed using DataDesk v6.0.

Unpaired two-sample t-tests were performed on the data

to compare the differences between sample means.

Animal handling

All animals used in these experiments complied with the

ARVO statement for the Use of Animals in Ophthalmic and

Vision Research and were also passed by an intuitional

ethics committee and under licence. Mice were held in the

Barrier Mouse Facility in the Smurfit Institute of Genetics

prior and subsequent to experimental procedures. All micewere sacrificed by CO2 asphyxiation.

Results

Expression cassettes have been generated in which a

combination of polymerase II promoters in conjunction

with cis-acting hammerhead ribozymes have been used

to express functional shRNAs targeting eGFP and eGFP

suppression; these have been evaluated initially in cell

culture and subsequently in mice. Three constructs

A C with unifying features, most notably cis-actinghammerhead ribozymes, have been explored. In essence,

cis-acting ribozymes are employed to cleave nucleotides

5 and 3 of antisense and sense sequences comprising


DOI: 10.1002/jgm

8/6/2019 Allen et al., 2007

4/12

290 D. Allen et al.

dsRNAs thereby promoting generation of functional

siRNAs. In each case, the sequence of one arm of the

cis-acting ribozyme is in part complimentary to either

the sense or antisense sequence used to form siRNAs.

Construct A comprises an shRNA sequence cleaved 5

and 3 by cis-acting hammerhead ribozymes. In contrast,

four cis-acting hammerhead ribozymes are employed in

construct B to generate sense and antisense strands

separately which must hybridize within the cell to

produce functional siRNAs. Construct C comprises a

single hammerhead ribozyme 5 of the sense portion

of the shRNA and a minimal poly-A signal at the

3 end placed juxtaposed to sequence coding for the

antisense portion of the shRNA. In all three eGFP targeting

constructs (AC) the shRNA target sequence was

CGGCAAGCTGACCCTGAAGTTCAT and for the three non-

targeting constructs (A C) the shRNA target sequence

was AATTCTCCGAAC GTGTCACGT. Both target sites are

based on Qiagen control siRNA target sites with thenon-targeting shRNA sequence having no homology to

any known mammalian genes. Both the eGFP-targeting

constructs and the non-targeting constructs were placed

under the control of a cytomegalovirus (CMV) promoter

(Figures 1A1C).

In principle, upon expression of the CMV-driven AC

constructs, the cis-acting hammerhead ribozymes should

self-cleave releasing shRNAs in the case of constructs

A and C and sense and antisense RNA in the case of

construct B. Given that the constructs described above

comprise multiple elements, to ensure that constructs

A C were indeed capable of expressing siRNAs, HeLa cells were transfected with these constructs. Total RNA was

subsequently extracted and enriched for small RNAs using

the mirVana miRNA isolation kit. RNA concentrations

were calculated by absorbance at 260 nm and RNA purity

analyzed using the A260 to A280 ratio. RNA preparations,

enriched for siRNAs, were run on 4% TBE NuSieve3 : 1

agarose gels using a dsRNA ladder (New England

Biolabs) to size RNAs. RNA bands in these samples were

compared to bands present in cells transfected with either

the positive control (H1-shRNA(eGFP)), empty vector

(pCDNA3.1) or untransfected cells. As shown in Figure 2,

the results suggest that constructs AC and the positive

control are all capable of producing siRNA as suggestedby the presence of bands of the correct size in the

RNA samples from these transfections. In contrast no

such bands were observed in the negative controls. It is

notable that whilst design B produces separate sense and

antisense strands of siRNA that have to anneal in vitro

to form siRNAs, it appears that this does indeed occur

(Figure 2).

To examine whether the siRNAs being produced by con-

structs AC were indeed functional, suppression studies

were undertaken in HeLa cells using a CMV promoter

to drive expression of the sequences. To determine

the optimal experimental conditions, HeLa cells wereco-transfected with an eGFP reporter plasmid (pEGFP-

C1, Clontech) and varying concentrations of the design

C expression plasmid (between 0.1 and 10 g). eGFP

transcript levels were measured by real-time rtPCR,

normalized using glyceraldehyde-3-phosphate dehydro-

genase (GAPDH), actin and cyclophilin-A expression

and compared to eGFP expression in cells transfected


DOI: 10.1002/jgm

8/6/2019 Allen et al., 2007

5/12


with the non-targeting design C construct. As shown in

Figure 3, the change in eGFP transcript level was sig-

nificant (p < 0.05) using 1 g of plasmid resulting in

a 53% decrease in expression. At 5 and 10 g of plas-

mid the fold change was also significant (p < 0.05) with

a 73% and 74% decrease in expression, respectively.Constructs AC were then co-transfected with an eGFP

reporter plasmid into HeLa cells as above (5 g of con-

struct with 1 g of eGFP reporter plasmid) (Figure 4).

Constructs AC resulted in approximately 71%, 70% and

73% suppression of eGFP expression, respectively, while

the H1-shRNA(eGFP) positive control resulted in 58%

suppression of the target (when compared to the appro-

priate non-targeting shRNA control). The low level of

suppression with the H1-shRNA(eGFP) may possibly be

attributed to co-transfection of the two plasmids the

target and the suppressor plasmids. Additionally, these

constructs are driven by different promoters, i.e. a CMV-

driven eGFP reporter and a H1-driven shRNA. The data

provide a clear indication that functional siRNAs can be

generated by exploiting hammerhead ribozymes to trim 5

and 3 sequences flanking core elements (sense/antisense

sequences) comprising a dsRNA molecule. Indeed sup-

pression of the eGFP target was achieved by either

expressing shRNA sequences (constructs A and C) or

sense and antisense siRNA sequence (construct B). In prin-

ciple, the same approach could be utilized subsequently

incorporating tissue-specific polymerase II promoters to

provide tissue specificity of RNAi. Trimming of sequences

5 and 3 of the RNAi molecule allows for flexibility in

choice of polymerase II promoter and hence would enabletissue-specific RNAi.

Figure 1. Diagrammatic representation of constructs AC.

(A) Construct A utilizes two cis-acting hammerhead ribozymes

to cleave 5 and 3 of an shRNA sequence. (B) In contrast,

construct B utilizes four cis-acting hammerhead ribozymes,

two that cleave 5 and 3 of the siRNA sense strand and two

that cleave 5 and 3 of the siRNA antisense strand. Sense

and antisense strands must anneal intracellularly to generate

functional siRNAs. (C) Construct C utilizes a single cis-acting

ribozyme to cleave 5 of the shRNA sequence with a minimal

polyadenylation signal immediately 3 of the shRNA sequence.

All three constructs are expressed using a CMV promoter.

Additionally, constructs AC have been generated carryingeither an shRNA sequence targeting eGFP or a non-targeting

control shRNA sequence

In some instances siRNAs and shRNAs have been found

to elicit a type-1 interferon response. It was important

to determine if the presence of additional elements in

the constructs, e.g. cis-acting hammerhead ribozymes,

may potentially result in an increased type-1 interferon

response, which clearly would not be desirable. Todetermine whether constructs A C stimulate a type-1

interferon response the expression of several interferon-

stimulated genes (OAS1, IFITM1, ISGF3g and EIF2AK2)

was compared to their expression levels in cells co-

transfected with either the H1-shRNA(eGFP) positive

control (directed towards the same eGFP target) or

the non-targeting shRNA control. Transcript levels were

measured by real-time rtPCR and normalized using

GAPDH, actin and cyclophilin-A expression. For the

interferon response the 2CT method was used to

determine mean fold changes in gene expression [13].

As shown in Figure 5, the resulting data indicated that

there was no significant change in expression of IFITM1

(p = 0.56) and EIF2AK2 (p = 0.1) with constructs AC

when compared to the non-targeting control construct.

Although OAS1 expression varied, e.g. a 2.3-fold increase

in OAS1 expression was observed with construct A, there

was no significant (p = 0.25) change in OAS1 expression

compared to the H1-shRNA(eGFP) positive control.

Furthermore, ISGF3g expression increased between 2.3-

and 3.5-fold with constructs A C; however, this increase

in ISGF3g expression was not significant (p = 0.24)

compared to the H1-shRNA(eGFP) positive control. In

conclusion, the results obtained suggest that, despite

the inclusion of additional components in constructs A C such as cis-acting hammerhead ribozymes, these

constructs operate in a similar manner to standard H1-

driven shRNA constructs in terms of the immune response

they elicit.

Given that from the in vitro work described above

it is clear that cis-acting ribozymes can act in concert

with shRNA sequences to produce functional siRNAs

and thereby elicit suppression of a target gene in cell

culture, it was timely to explore if this strategy would

also be effective in vivo. Hence eGFP-expressing mice,

C57BL/6-Tg(ACTB-eGFP)1Osb/J (Jackson Laboratories),

were used to explore if functional siRNAs couldbe generated from constructs A C in vivo. C57BL/6-

Tg(ACTB-eGFP)1Osb/J mice express an eGFP gene driven

Figure 2. The four top panels depict photographs of 4% TBE NuSieve3 : 1 agarose gels (under UV light) with a dsRNA ladder and

approx. 2 g of RNA enriched for small RNAs from HeLa cells transfected with the empty vector (pCDNA3.1) (A), positive control

(H1-shRNA(eGFP)) (B), construct A (C), construct B (D), or construct C (E)


DOI: 10.1002/jgm

8/6/2019 Allen et al., 2007

6/12

292 D. Allen et al.

Figure 3. Graph of eGFP transcript levels in HeLa cells co-transfected with construct C at various concentrations and an eGFP

reporter plasmid. To determine the optimal plasmid concentration for further experiments, increasing quantities of construct Cwere co-transfected with an eGFP reporter plasmid. eGFP transcript levels were measured by real-time rtPCR and normalized using

GAPDH, actin and cyclophilin-A expression. The eGFP expression obtained was compared to eGFP expression in cells transfected

with non-targeting constructs. The reduction in eGFP transcript levels was significant (p < 0.05) using 1 g of the construct C

plasmid (53% decrease in mRNA levels). At 5 g and 10 g of plasmid the fold change was also significant (p < 0.05) with a 73%

and 74% decrease in eGFP mRNA, respectively. Error bars represent standard deviation

Figure 4. Graph of eGFP transcript levels in HeLa cells co-transfected with constructs A C and an eGFP reporter plasmid. eGFP

transcript levels were measured by real-time rtPCR and normalized using GAPDH, actin and cyclophilin-A expression. Expression

was compared to eGFP expression in cells transfected with non-targeting constructs. Cells transfected with the positive control

construct (H1-shRNA(eGFP)) showed 58% suppression of eGFP transcript levels. Cells transfected with constructs A, B and C

showed 71%, 70% and 73% suppression of eGFP transcript levels, respectively. Notably, all four eGFP-targeting constructs resulted

in significant eGFP suppression when compared to the appropriate non-targeting shRNA control construct (p < 0.05). Error bars

represent standard deviation

by a chicken beta-actin promoter with a CMV enhancer

[14]. Sixty-four transgenic mice were treated twice over

a 24-h period with 20 g of constructs AC (either the

eGFP-targeting or the non-targeting versions), the H1-

shRNA(eGFP) positive control or the non-targeting H1-shRNA, administered via hydrodynamic tail-vein injection

(total 40 g). Liver, kidney and spleen were harvested

48 h after the second injection. RNA extracted from tissues

was analyzed for eGFP expression by real-time rtPCR and

normalized using GAPDH and ribosomal 18s expression.

eGFP expression in mice injected with eGFP-targeting

constructs AC was compared to eGFP expression in mice

injected with the non-targeting control constructs AC.Results obtained in vivo were similar to those observed

in the experiments undertaken in cell culture (Figure 6).

In liver, the positive control construct H1-shRNA(eGFP)


DOI: 10.1002/jgm

8/6/2019 Allen et al., 2007

7/12


Figure 5. Graphs of OAS1 (A), IFITM1 (B), ISGF3g (C), and EIF2AK2 (D) transcript levels in HeLa cells co-transfected with constructs

AC, the H1-shRNA(eGFP) or the non-targeting shRNA control and an eGFP reporter plasmid. Transcript levels were measured by

real-time rtPCR and normalized using GAPDH, actin and cyclophilin-A expression. Expression was compared to cells transfected

with either the H1-shRNA(eGFP) positive control (directed towards the same eGFP target) or the non-targeting shRNA control.

The resulting data indicated no significant change in expression of IFITM1 (p = 0.56) and EIF2AK2 (p = 0.1) compared to the

non-targeting control. Although OAS1 expression varied, e.g. a 2.3-fold increase in OAS1 expression was observed with design A,

there was no significant (p = 0.25) change in OAS1 expression compared to the H1-shRNA(eGFP) positive control. Furthermore,ISGF3g expression increased between 2.3- and 3.5-fold with constructs AC; however, this increase in ISGF3g expression was not

significant (p = 0.24) compared to the H1-shRNA(eGFP) positive control. Error bars represent standard deviation

resulted in a 45% decrease in eGFP transcript levels.

Similarily, constructs A C resulted in 43%, 40% and

54% decreases in eGFP transcript levels respectively in

liver. Notably, eGFP suppression with constructs A C

in liver did not differ significantly from the positive

control (p = 0.57). In contrast, all four eGFP-targeting

RNAi constructs demonstrated a significant difference, in

terms of eGFP suppression, from the non-targeting shRNA

negative control construct (p < 0.05). Furthermore, inaddition to liver, eGFP suppression was evaluated in

kidney and spleen. Whilst no eGFP suppression was seen

in kidney or spleen it was established, via the injection

of an eGFP reporter plasmid, that hydrodynamic injection

resulted in preferential delivery to liver with no significant

expression in other organs such as kidney and spleen

(discussed below).

In addition to evaluation using real-time rtPCR,

liver samples from all 64 injected mice were also

taken and preserved in paraformaldehyde for fluorescent

microscopy. The four panels shown in Figure 7A depict

representative 100-m vibratome liver sections (12.5

),under UV light and a GFP filter, from C57BL/6-Tg(ACTB-

EGFP)1Osb/J mice tail vein injected with a total of

40 g of a non-targeting H1-shRNA and non-targeting

constructs A C (A D, respectively). A significant level

of fluorescence can be observed in each of the sections.

In these sections it is possible to view an entire liver

cross-section demonstrating even distribution of eGFP

fluorescence (Figure 7A). In contrast, the four panels in

Figure 7B depict representative 100-m vibratome liver

sections (12.5), under UV light and a GFP filter, from

C57BL/6-Tg(ACTB-EGFP)1Osb/J mice tail vein injected

with a total of 40g of eGPF targeting H1-shRNAand eGFP-targeting constructs AC (AD, respectively).

Notably, a significant reduction in green fluorescence

can be observed in each of the sections in Figure 7B

compared to those shown in Figure 7A. In these sections it

is possible to view an entire liver cross-section in this case

demonstrating the even distribution of eGFP suppression

(Figure 7B). In summary, the results from in vivo delivery

of constructs AC in mice mirror those obtained in cell

culture and suggest that these constructs incorporating

cis-acting hammerhead ribozymes and shRNA sequences

may be used to generate potent siRNAs in vivo from

polymerase II promoters.To determine whether constructs A C stimulate a

type-1 interferon response in vivo the expression of several

interferon-stimulated genes (OAS1, IFITM1, ISGF3g and


DOI: 10.1002/jgm

8/6/2019 Allen et al., 2007

8/12

294 D. Allen et al.

Figure 6. Graph of eGFP transcript levels in liver, kidney and spleen from C57BL/6-Tg(ACTB-EGFP)1Osb/J mice. The

H1-shRNA(eGFP) positive control, and both eGFP-targeting and non-targeting constructs AC, were evaluated in vivo subsequent to

hydrodynamic tail-vein injection of C57BL/6-Tg(ACTB-EGFP)1Osb/J mice. A total of 64 mice were treated twice over a 24-h period

with 20 g of plasmid (total of 40 g). eGFP transcript levels were measured by real-time rtPCR and normalized using GAPDH and

ribosomal 18s expression. eGFP expression in mice injected with eGFP-targeting constructs was compared to eGFP expression from

mice injected with non-targeting control constructs. In liver, administration of the positive control construct (H1-shRNA(eGFP))

resulted in a 45% decrease in eGFP transcript levels. Similarly, in liver, constructs AC resulted in 43%, 40% and 54% reductions

in eGFP transcript levels, respectively. Notably, eGFP suppression with constructs A C did not differ significantly from that

achieved with the positive control (p = 0.57). All four targeting constructs resulted in significant eGFP suppression compared to

the non-targeting shRNA negative control (p < 0.05). Due to a lack of plasmid delivery outside of the liver, eGFP expression in

kidney and spleen with all constructs did not differ significantly from the negative controls. A total of eight mice were injected per

construct. Error bars represent standard deviation

EIF2AK2) was compared to their expression levels in

mice transfected with either the H1-shRNA(eGFP) positive

control (directed towards the same eGFP target) or

the non-targeting shRNA control. Transcript levels were

measured by real-time rtPCR and normalized using

GAPDH, actin and cyclophilin-A expression. For the

interferon response the 2CT method was used to

determine mean fold changes in gene expression [13].

As shown in Figure 8, the resulting data indicated that

there was no significant change in expression of IFITM1

(p = 0.33) and EIF2AK2 (p = 0.11) with constructs AC

when compared to the non-targeting control construct.Although OAS1 expression varied between 2.3- to 3.0-fold

compared to the non-targeting control construct, there

was no significant (p = 0.49) change in OAS1 expression


Furthermore, ISGF3g expression increased between 1.3-

and 2.0-fold with constructs AC; however, this increase

in ISGF3g expression was not significant (p = 0.87)


Notably, these results are similar to the expression

patterns found in vitro. In conclusion, the results

obtained suggest that, despite the inclusion of additional

components in constructs AC such as cis-actinghammerhead ribozymes, these constructs operate in a

similar manner to standard H1-driven shRNA constructs

in terms of the immune response they elicit in vivo.

As discussed above, in contrast to the suppression

obtained in liver, no significant eGFP suppression was

obtained in kidney or spleen, after tail-vein injection,

irrespective of the construct used (Figure 6). Delivery

of naked DNA has been explored extensively for many

tissues but typically DNA is subject to rapid degradation

by nucleases [15]. However, results from previous studies

suggest that hydrodynamic tail-vein injection in mice may

be used to deliver plasmid DNA effectively to liver [16].

Absence of eGFP suppression in organs other than liver in

the current study may be due to inadequate delivery of

plasmid DNA to those organs. To explore this hypothesisa CMV promoter-eGFP reporter gene construct (CMV-

eGFP) was delivered to wild-type CD-1, C57BL/6 and

129 mice by hydrodynamic tail-vein injection and liver,

kidney, spleen, heart and lung tissues analyzed for eGFP

expression 48 h post-administration of DNA by real-time

rtPCR and fluorescent microscopy. eGFP expression was

found to be almost exclusively limited to liver tissue (data

not shown).

In summary, while high-pressure tail-vein administra-

tion of naked DNA can provide effective in vivo delivery to

mouse liver, delivery to other organs is restricted or absent

using this approach. A prerequisite for in vivo explo-ration of specific RNAi is a means of delivering tsRNAi

constructs to other target organs. Methodologies which

provide delivery of DNA to organs other than liver have


DOI: 10.1002/jgm

8/6/2019 Allen et al., 2007

9/12


Figure 7. The four panels shown in (A) represent 100-m vibratome liver sections (12.5), under UV light and a GFP filter,

from C57BL/6-Tg(ACTB-EGFP)1Osb/J mice tail vein injected with a total of 40 g of a non-targeting H1-shRNA and non-targeting

constructs AC (AD, respectively). A total of 64 mice were treated twice over a 24-h period with 20 g of plasmid (total of 40 g).

A significant level of green fluorescence can be observed in each of the sections. In these sections it is possible to view an entire liver

cross-section demonstrating the even distribution of eGFP fluorescence. The four panels in (B) represent 100-m vibratome liver

sections (12.5), under UV light and a GFP filter, from C57BL/6-Tg(ACTB-EGFP)1Osb/J mice tail vein injected with a total of 40g

of eGPF targeting H1-shRNA and eGFP-targeting constructs AC (AD, respectively). A significant reduction in green fluorescence

can be observed in each of the sections in (B) compared to those in (A). Again in these sections it is possible to view an entire

liver cross-section demonstrating the even distribution of eGFP suppression achieved subsequent to hydrodynamic injection of the

eGPF-targeting design C construct. A total of eight mice were injected per construct. Scale bars: 500 m

been employed. For example, Wu and colleagues demon-

strated plasmid-based delivery of a luciferase reporter

gene to mouse kidney subsequent to a single injection

into the inferior vena cava (IVC) [17]. Similarly, in the

current study, a CMV-driven eGFP plasmid (40 g) has

been injected into mouse IVC and eGFP expression eval-

uated using real-time rtPCR and fluorescent microscopy.

eGFP expression was observed at both RNA and protein

levels (data not shown) confirming that IVC injection

may be used to deliver plasmid DNA to mouse kid-

ney.Given that constructs A C incorporating cis-acting

hammerhead ribozymes and shRNA sequences have been

found to generate potent siRNAs in vivo from polymerase

II promoters, together with methods for delivery of

DNA to multiple organs, it was timely to engineer a

liver-specific promoter sequence into these constructs.

The CMV promoter in constructs AC was replaced by

2.3 kb of the rat albumin promoter/enhancer sequence,

a promoter previously shown to drive liver-specific gene

expression in vivo [18]. Subsequently, eGFP-targeting and

non-targeting construct C driven by an albumin promoter

were administered by hydrodynamic tail-vein injection. A total of 24 mice were injected. Liver from injected

mice were harvested 48 h post-injection and analyzed

for eGFP expression using real-time rtPCR (Figure 9).

eGFP expression levels were compared to levels in mice

injected with non-targeting control plasmids. Significant

suppression (p < 0.05, DataDesk v6.0) of eGFP expression

was observed approximately 56% suppression of eGFP

expression as evaluated by real-time rtPCR.

In addition, the CMV-driven and albumin-driven eGFP-

targeting and non-targeting construct C plasmids were

injected into the IVC. Approximately 45% suppression

(p < 0.05, DataDesk v6.0) of eGFP expression as

evaluated by real-time rtPCR was observed in mouse

kidney (Figure 10) when the CMV-driven eGFP-targetingconstruct was injected andcompared to that obtained with

a non-targeting control. In contrast, the albumin-driven

construct elicited no suppression of eGFP expression

in mouse kidney. In summary, the albumin promoter

in concert with hammerhead ribozymes and shRNA

sequences can be used to drive expression of functional

siRNAs in liver.

Discussion

In this study a method of generating functional siRNAs, inprinciple from any polymerase II promoter, involving use

of cis-acting ribozymes has been developed and evaluated

both in vitro and in vivo (in mice). As demonstrated such


DOI: 10.1002/jgm

8/6/2019 Allen et al., 2007

10/12

296 D. Allen et al.

Figure 8. Graphs of OAS1 (A), IFITM1 (B), ISGF3g (C), and EIF2AK2 (D) transcript levels in liver from C57BL/

6-Tg(ACTB-EGFP)1Osb/J mice tail vein injected with constructs AC, the H1-shRNA(eGFP) or the non-targeting shRNA control

and an eGFP reporter plasmid. A total of 40 mice were treated twice over a 24-h period with 20 g of plasmid (total of 40 g).

Transcript levels were measured by real-time rtPCR and normalized using GAPDH, actin and cyclophilin-A expression. Expression

was compared to cells transfected with either the H1-shRNA(eGFP) positive control (directed towards the same eGFP target) or the

non-targeting shRNA control. The resulting data indicated that there was no significant change in expression of IFITM1 ( p = 0.33)

and EIF2AK2 (p = 0.11) with constructs AC when compared to the non-targeting control construct. Although OAS1 expression

varied between 2.3- to 3.0-fold compared to the non-targeting control construct, there was no significant (p = 0.49) change in OAS1expression compared to the H1-shRNA(eGFP) positive control. Furthermore, ISGF3g expression increased between 1.3- and 2.0-fold

with constructs AC; however, this increase in ISGF3g expression was not significant (p = 0.87) compared to the H1-shRNA(eGFP)

positive control. A total of eight mice were injected per construct. Error bars represent standard deviation

an approach may readily be adapted to provide tissue-

specific RNAi by incorporating tissue-specific polymerase

II promoters into constructs AC. Potential advantages

of tissue-specific gene silencing are evident given

results from microarray-based expression profiling studies

highlighting that off-target effects can frequently arise

as a result of RNAi-mediated gene silencing [12].

Additionally, RNAi has been proposed as a means ofgenerating knockout or knockdown transgenic animals

[19]. Tissue-specific RNAi would overcome potential

embryonic lethality in such transgenics and moreover

would provide enhanced resolution for RNAi technology

enabling gene silencing in individual tissue types. Given

these significant advantages a number of approaches to

achieve tissue-specific expression of functional siRNAs

have been proposed. The approaches typically utilize H1

or U6 promoter-driven shRNA constructs carrying loxP

sites in combination with tissue-specific delivery of Cre

recombinase to induce expression of functional siRNAs

[20 23]. Furthermore, tetracycline-based systems havebeen used to demonstrate conditional suppression in vitro

and in vivo [24,25]. In one such approach T7 recombinase

was used in conjunction with T7 promoter-driven shRNAs

and a 3 ribozyme to process shRNAs to achieve

inducible RNAi [25]. In addition, artificially generated

microRNAs (miRNAs) have recently been explored in vitro

as an alternative means of achieving gene silencing

and potentially could be used in combination with a

variety of polymerase II promoters to provide tissue

specificity [26,27]. The strategy explored in vitro and

in vivo in the current study represents an alternativemeans of achieving siRNA-based tissue-specific RNAi

and, in contrast to many of the approaches referred

to above, does not require delivery of Cre recombinase

or doxycycline to elicit tissue-specific expression and

would simply utilize a single cassette to achieve tissue-

specific RNAi. In essence, the approach developed

incorporates the use of polymerase II promoters to drive

tissue-specific expression in conjunction with cis-acting

hammerhead ribozymes to trim the sense and antisense

strands of the dsRNA to produce functional siRNA or

shRNA. Given the almost ubiquitous application of RNAi

in many areas of molecular biology, it is clear thatthere will be multiple applications for systems such as

that described which augment the resolution of RNAi

technology.


DOI: 10.1002/jgm

8/6/2019 Allen et al., 2007

11/12


Figure 9. Graph of eGFP transcript levels in liver of

C57BL/6-Tg(ACTB-EGFP)1Osb/J mice. The albumin promoter-

driven eGFP-targeting and non-targeting construct C were eval-

uated in vivo subsequent to hydrodynamic tail-vein injection. A

total of 16 mice were treated twice over a 24-h period with 20 g

of plasmid (totalof 40 g). eGFP transcript levels were measured

by real-time rtPCR and normalized using GAPDH and riboso-

mal 18s expression. eGFP expression in mice injected with the

eGFP-targeting construct was compared to eGFP expressionfrom

mice injected with the non-targeting control construct. In liver,

administration of the albumin promoter-driven eGFP-targeting

construct resulted in a 56% decrease in eGFP transcript levels.

Delivery of this construct resulted in significant eGFP suppres-

sion compared to the non-targeting shRNA negative control

(p < 0.05). A total of eight mice were injected per construct.

Error bars represent standard deviation

Acknowledgements

We would like to thank Sylvia Mehigan and Caroline Woods

who assisted with animal work. The research was supported

by Enterprise Ireland, DEBRA Ireland and the British Retinitis

Pigmentosa Society.

References

1. Fire A, Xu S, Montgomery MK, Kostas SA, Driver SE, Mello CC.Potent and specific genetic interference by double-stranded RNAin Caenorhabditis elegans. Nature 1998; 391: 806811.

2. Elbashir SM, Harborth J, Lendeckel W, Yalcin A, Weber K,Tuschl T. Duplexes of 21-nucleotide RNAs mediate RNA

interference in cultured mammalian cells. Nature 2001; 411:494498.

3. Czauderna F, Fechtner M, Dames S. et al. Structural variationsand stabilising modifications of synthetic siRNAs in mammaliancells. Nucleic Acids Res 2003; 31: 27052716.

4. Brummelkamp TR, Bernards R, Agami R. A system for stableexpression of short interfering RNAs in mammalian cells. Science2002; 296: 550553.

5. Miyagishi M, Taira K. U6 promoter-driven siRNAs with foururidine 3 overhangs efficiently suppress targeted geneexpression in mammalian cells. Nat Biotechnol 2002; 20:497500.

6. Paul CP, Good PD, Winer I, Engelke DR. Effective expression ofsmall interfering RNA in human cells. Nat Biotechnol 2002; 20:505508.

7. Sui G, Soohoo C, Affar el B, et al. A DNA vector-based RNAi

technology to suppress gene expression in mammalian cells.Proc Natl Acad Sci U S A 2002; 99: 55155520.8. Yu JY, DeRuiter SL, Turner DL. RNA interference by expression

of short-interfering RNAs and hairpin RNAs in mammalian cells.Proc Natl Acad Sci U S A 2002; 99: 60476052.

Figure 10. Graph of eGFP transcript levels in kidney of

C57BL/6-Tg(ACTB-EGFP)1Osb/J mice. Both the CMV and

albumin promoter-driven eGFP-targeting and non-targeting

construct C were evaluated in vivo subsequent to direct inferior

vena cava injection. A total of 32 mice were treated with40 g of plasmid. eGFP transcript levels were measured by

real-time rtPCR and normalized using GAPDH and ribosomal

18s expression. eGFP expression in mice injected with the

eGFP-targeting construct was compared to eGFP expression from

mice injected with the non-targeting control construct. In kidney,

administration of the CMV promoter-driven eGFP-targeting

construct resulted in a 46% decrease in eGFP transcript

levels. Delivery of this construct resulted in significant

eGFP suppression compared to the non-targeting shRNA

negative control (p < 0.05). In contrast, administration of the

albumin promoter-driven eGFP-targeting construct resulted in

no significant decrease in eGFP transcript levels. A total of eight

mice were injected per construct. Error bars represent standard

deviation

9. Kawasaki H, Taira K. Short hairpin type of dsRNAs that arecontrolled by tRNA(Val) promoter significantly induce RNAi-mediated gene silencing in the cytoplasm of human cells. Nucleic

Acids Res 2003; 31: 700707.10. Paddison PJ, Caudy AA, Hannon GJ. Stable suppression of gene

expression by RNAi in mammalian cells. Proc Natl Acad Sci U S A2002; 99: 14431448.

11. Xia H, Mao Q, Paulson HL, Davidson BL. siRNA-mediated genesilencing in vitro and in vivo. Nat Biotechnol 2002; 20:10061010.

12. Jackson AL, Bartz SR, Schelter J, et al. Expression profilingreveals off-target gene regulation by RNAi. Nat Biotechnol 2003;21: 635637.

13. Livak KJ, Schmittgen TD. Analysis of relative gene expression

data using real-time quantitative PCR and the 2(-delta deltaC(T)) method. Methods 2001; 25: 402408.

14. Okabe M, Ikawa M, Kominami K, Nakanishi T, Nishimune Y.Green mice as a source of ubiquitous green cells. FEBS Lett1997; 407: 313319.

15. Pouton CW, Seymour LW. Key issues in non-viral gene delivery.Adv Drug Deliv Rev 2001; 46: 187203.

16. Lewis DL, Hagstrom JE, Loomis AG, Wolff JA, Herweijer H.Efficient delivery of siRNA for inhibition of gene expressionin postnatal mice. Nat Genet 2002; 32: 107108.

17. Wu X, Gao H, Pasupathy S, Tan PH, Ooi LL, Hui KM. Systemicadministration of naked DNA with targeting specificity tomammalian kidneys. Gene Ther 2005; 12: 477486.

18. Postic C, Shiota M, Niswender KD, et al. Dual roles forglucokinase in glucose homeostasis as determined by liverand pancreatic beta cell-specific gene knock-outs using Cre

recombinase. J Biol Chem 1999; 274: 305315.19. Tiscornia G, Singer O, Ikawa M, Verma IM. A general method forgene knockdown in mice by using lentiviral vectors expressingsmall interfering RNA. Proc Natl Acad Sci U S A 2003; 100:18441848.


DOI: 10.1002/jgm

8/6/2019 Allen et al., 2007

12/12

298 D. Allen et al.

20. Fritsch L, Martinez LA, Sekhri R, et al. Conditional gene knock-down by CRE-dependent short interfering RNAs. EMBO Rep2004; 5: 178182.

21. Tiscornia G, Tergaonkar V, Galimi F, Verma IM. CRErecombinase-inducible RNA interference mediated by lentiviral

vectors. Proc Natl Acad Sci U S A 2004; 101: 73477351.22. Ventura A, Meissner A, Dillon CP, et al. Cre-lox-regulated

conditional RNA interference from transgenes. Proc Natl AcadSci U S A 2004; 101: 10380 10385.23. Coumoul X, Shukla V, Li C, Wang RH, Deng CX. Conditional

knockdown of Fgfr2 in mice using Cre-LoxP induced RNAinterference. Nucleic Acids Res 2005; 33: e102.

24. Szulc J, Wiznerowicz M, Sauvain MO, Trono D, Aebischer P. A versatile tool for conditional gene expression and knockdown.Nat Methods 2006; 3: 109116.

25. Hamdorf M, Muckenfuss H, Tschulena U, et al. An inducibleT7 RNA polymerase-dependent plasmid system. Mol Biotechnol2006; 33: 1321.

26. Dickins RA, Hemann MT Zilfou JT, et al. Probing tumor

phenotypes using stable and regulated synthetic microRNAprecursors. Nat Genet. 2005; 37(11): 12891295Oct 2.27. Zeng Y, Cai X, Cullen BR. Use of RNA polymerase II to transcribe

artificial microRNAs. Methods Enzymol 2005; 392: 371380.


DOI: 10.1002/jgm

allen et al., 2007

Documents