VIRAL HEPATITIS
`Activity of Stabilized Short Interfering RNA in a Mouse
`Model of Hepatitis B Virus Replication
`
`David V. Morrissey, Karin Blanchard, Lucinda Shaw, Kristi Jensen, Jennifer A. Lockridge, Brent Dickinson,
`James A. McSwiggen, Chandra Vargeese, Keith Bowman, Chris S. Shaffer, Barry A. Polisky, and Shawn Zinnen
`
`To develop synthetic short interfering RNA (siRNA) molecules as therapeutic agents for systemic
`administration in vivo, chemical modifications were introduced into siRNAs targeted to con-
`served sites in hepatitis B virus (HBV) RNA. These modifications conferred significantly pro-
`longed stability in human serum compared with unmodified siRNAs. Cell culture studies
`revealed a high degree of gene silencing after treatment with the chemically modified siRNAs. To
`assess activity of the stabilized siRNAs in vivo initially, an HBV vector-based model was used in
`which the siRNA and the HBV vector were codelivered via high-volume tail vein injection. More
`than a 3 log10 decrease in levels of serum HBV DNA and hepatitis B surface antigen, as well as
`liver HBV RNA, were observed in the siRNA-treated groups compared with the control siRNA-
`treated and saline groups. Furthermore, the observed decrease in serum HBV DNA was 1.5 log10
`more with stabilized siRNA compared with unmodified siRNA, indicating the value of chemical
`modification in therapeutic applications of siRNA. In subsequent experiments, standard sys-
`temic intravenous dosing of stabilized siRNA 72 hours after injection of the HBV vector resulted
`a 0.9 log10 reduction of serum HBV DNA levels after 2 days of dosing. In conclusion, these
`experiments establish the strong impact that siRNAs can have on the extent of HBV infection and
`underscore the importance of stabilization of siRNA against nuclease degradation. (HEPATOLOGY
`2005;41:1349-1356.)
`
`See Editorial on Page 1220
`
`RNA interference (RNAi) is a recently discovered
`
`cellular mechanism that detects and destroys dou-
`ble-stranded RNA1 and seems to play a role in the
`cell’s antiviral defense system.2 Short interfering RNA
`(siRNA) molecules are approximately 21-nucleotide,
`double-stranded RNA intermediates of the RNAi mech-
`anism that guide a unique RNAi protein complex termed
`RNA-induced silencing complex to target RNA, leading
`to its subsequent degradation. Although in the natural
`RNAi pathway, siRNAs are derived from long double
`strand RNA that is processed by the nuclease Dicer into
`discrete 21-mers,3,4 introduction of synthetic siRNAs
`
`Abbreviations: RNAi, RNA interference; siRNA, short interfering RNA; HBV,
`hepatitis B virus; HBsAg, hepatitis B surface antigen; HDI, hydrodynamic tail vein
`injection; siNA, short interfering nucleic acid.
`From Sirna Therapeutics, Inc., Boulder, CO.
`Received December 7, 2004; accepted February 26, 2005.
`Address reprint requests to: David V. Morrissey, Ph.D., Sirna Therapeutics, Inc.,
`2950 Wilderness Place, Boulder, CO 80301. E-mail: morrisseyd@sirna.com; fax:
`303-449-8829.
`Copyright © 2005 by the American Association for the Study of Liver Diseases.
`Published online in Wiley InterScience (www.interscience.wiley.com).
`DOI 10.1002/hep.20702
`Potential conflict of interest: Nothing to report.
`
`into the cell also leads to RNAi-mediated silencing of
`target gene expression.5 The use of synthetic siRNAs that
`use the endogenous cellular mechanism to downregulate
`the expression of disease-related or viral genes may lead to
`the development of a new therapeutic approach. In this
`report, we present evidence that synthetic siRNAs inhibit
`the replication of the hepatitis B virus (HBV) in cell cul-
`ture and in a mouse model of HBV replication.
`Although siRNA has become an effective research tool
`to downregulate gene expression in cell culture, the inher-
`ent instability of RNA limits its potential use for in vivo
`gene silencing. The development of siRNAs as therapeu-
`tic agents will likely require improvements in the stability
`of siRNAs and the efficiency and specificity of tissue-
`targeted delivery in vivo. Chemical modifications made to
`synthetic siRNAs for the purpose of stabilization not only
`must provide resistance to nuclease degradation, but also
`must be compatible with proper recognition and function
`of the siRNA by cellular RNAi protein machinery.
`A potential clinical use of siRNAs is to combat chronic
`infection by HBV. Chronic HBV afflicts approximately
`350 million people worldwide and accounts for 1.2 mil-
`lion deaths annually. In the United States alone, approx-
`imately 1.25 million individuals (0.05%) are chronic
`carriers of HBV.6 The natural progression of chronic
`HBV infection over 10 to 20 years leads to liver cirrhosis
`
`1349
`
`Alnylam Exh. 1065
`
`

`

`1350 MORRISSEY ET AL.
`
`HEPATOLOGY, June 2005
`
`in 20% to 50% of patients, and progression of HBV
`infection to hepatocellular carcinoma has been well doc-
`umented.7 Because of the overlapping nature of the HBV
`pregenomic RNA and the four major transcripts, an
`siRNA targeted to a single site is capable of cleaving more
`than one HBV RNA species, and thus has the potential to
`interfere with numerous processes in the viral life cycle.
`To test the potential of synthetic siRNA molecules as
`therapeutic agents for systemic administration in vivo,
`chemical modifications designed to inhibit degradation
`by nucleases were introduced into chemically synthesized
`siRNAs targeted to conserved sites in HBV RNA. We
`have developed a novel combination of siRNA chemical
`modifications that permit effective gene silencing in the
`complete absence of 2⬘-OH residues. The combination of
`modifications include 2⬘-fluoro, 2⬘-O-methyl, and 2⬘-de-
`oxy sugars, phosphorothioate linkages, and terminus cap-
`ping chemistries. This combination of modifications
`conferred significantly prolonged stability to siRNAs in
`serum, and studies in an HBV cell culture system showed
`effective gene silencing by the modified siRNAs. To test
`the in vivo potency of the stabilized siRNA, an HBV
`vector-based mouse model was used with coadministra-
`tion of the siRNAs with a HBV vector via high-pressure
`tail vein injection, resulting in a 3 log10 decrease in serum
`HBV DNA levels 3 days after a single administration. In
`subsequent experiments, standard systemic intravenous
`dosing of stabilized siRNA 72 hours after injection of the
`HBV vector resulted in an approximately 1 log10 reduc-
`tion of serum HBV DNA levels after 2 days of dosing.
`This first demonstration of in vivo activity of a systemi-
`cally delivered, modified siRNA in an HBV mouse model
`illustrates the therapeutic potential of siRNAs.
`
`Materials and Methods
`
`Oligonucleotide Synthesis and Characterization.
`RNA oligonucleotides were synthesized at Sirna Thera-
`peutics, Inc. (Boulder, CO), using ABI 394 synthesizers
`and standard phosphoramidite chemistry. Modified RNA
`oligonucleotides were synthesized, deprotected, and puri-
`fied as previously described.8 The integrity and purity of
`the final compounds were confirmed by standard HPLC,
`capillary electrophoresis , and MALDI-TOF mass spec-
`trometry methodologies. Oligonucleotides used were
`more than 80% full length. The siRNA sequences for site
`263 were: sense strand, 5⬘-GGACUUCUCUCAAUU-
`UUCUTT-3⬘; antisense strand, 5⬘-AGAAAAUUG-
`AGAGAAGUCCTT-3⬘. The negative control sequences
`are inverted from 5⬘ to 3⬘.
`siRNA Annealing. siRNA strands (20 ␮mol/L each
`strand) were annealed in 100 mmol/L potassium acetate, 30
`
`mmol/L HEPES-KOH at pH 7.4, and 2 mmol/L magne-
`sium acetate. The annealing mixture was first heated to 90°C
`for 1 minute and then transferred to 37°C for 60 minutes.
`Annealing was confirmed by nondenaturing PAGE and
`melting temperature assessment in 150 mmol/L NaCl.
`Serum Stability Assay. Oligonucleotides were de-
`signed such that standard ligation methods would gener-
`ate full-length sense or antisense strands. Before ligation,
`standard kinase methods were used with [␥-32P]ade-
`nosine triphosphate to generate an internal 32P label. Li-
`gated material was gel purified using denaturing PAGE.
`Trace internally labeled sense (or antisense) was added to
`unlabeled material to achieve a final concentration of 20
`␮mol/L. The unlabeled complementary strand was
`present at 35 ␮mol/L. Annealing was performed as de-
`scribed above. Duplex formation was confirmed by un-
`modified PAGE and subsequent visualization on a
`Molecular Dynamics Phosphoimager (Sunnyvale, CA).
`Internally labeled, duplexed or single-stranded siRNA
`was added to human serum to achieve final concentra-
`tions of 90% serum (Sigma, St. Louis, MO) and 2
`␮mol/L siRNA duplex with a 1.5 ␮mol/L excess of the
`unlabeled single-stranded siRNA. Samples were incu-
`bated at 37°C. Aliquots were removed at specified time
`points and quenched using a 5-second Proteinase K (20
`␮g) digestion (Amersham, Piscataway, NJ) in 50 mmol/L
`Tris-HCl pH 7.8, 2.5 mmol/L ethylenediaminetetraace-
`tic acid, 2.5% SDS, followed by addition of a 6⫻ volume
`of
`formamide loading buffer (90% formamide, 50
`mmol/L ethylenediaminetetraacetic acid, 0.015% xylene
`cyanol and bromophenol blue, 20 ␮mol/L unlabeled
`chase oligonucleotide of the same sequence as the radio-
`labeled strand). Samples were separated by denaturing
`PAGE and were visualized on a Molecular Dynamics
`Phosphoimager.
`ImageQuant
`(Molecular Dynamics)
`software was used for quantitation.
`Cell Culture Studies. The human hepatoblastoma
`cell lines Hep G2 was grown in minimal essential Eagle
`media supplemented with 10% fetal calf serum, 2
`mmol/L glutamine, 0.1 mmol/L nonessential amino ac-
`ids, 1 mmol/L sodium pyruvate, and 25 mmol/L HEPES.
`Replication-competent cDNA was generated by excising
`and relegating the HBV genomic sequences from the
`psHBV-1 vector. Hep G2 cells were plated (3 ⫻ 104
`cells/well) in 96-well microtiter plates and were incubated
`overnight. A cationic lipid–DNA–siRNA complex was
`formed containing (at final concentrations) cationic lipid
`(11-15 ␮g/mL), religated psHBV-1 (4.5 ␮g/mL), and
`siRNA (25 nmol/L) in growth media. After a 15-minute
`incubation at 37°C, 20 ␮L of the complex was added to
`the plated Hep G2 cells in 80 ␮L of growth media minus
`antibiotics. The media was removed from the cells 72
`
`

`

`HEPATOLOGY, Vol. 41, No. 6, 2005
`
`MORRISSEY ET AL.
`
`1351
`
`hours after transfection for hepatitis B surface antigen
`(HBsAg) analysis. All transfections were performed in
`triplicate.
`HBsAg Enzyme-Linked Immunosorbent Assay.
`Levels of HBsAg were determined using the Genetic Sys-
`tems/Bio-Rad (Richmond, VA) HBsAg enzyme-linked
`immunosorbent assay kit, according to the manufactur-
`er’s instructions. The absorbance of cells not transfected
`with the HBV vector was used as background for the
`assay, and thus subtracted from the experimental sample
`values.
`HBV Vector-Based Mouse Model. To assess the ac-
`tivity of chemically stabilized siRNAs against HBV, sys-
`temic dosing of
`the siRNA was carried out after
`hydrodynamic tail vein injection (HDI) of HBV vector in
`mouse strain C57BL/J6. The HBV vector used, pWTD,
`is a head-to-tail dimer of the complete HBV genome.9 For
`a 20-g mouse, a total injection of 1.6 mL containing
`pWTD in saline, was injected into the tail vein within 5
`seconds. For a larger mouse, the injection volume was
`scaled to 140% of the blood volume of the mouse. For
`studies in which the HBV vector and siRNA were coin-
`jected, 1 ␮g of vector and 0.03 to 1 ␮g of siRNA were
`used. In experiments in which systemic dosing of the
`siRNA by conventional intravenous injection followed
`the HDI of the HBV vector, 0.3 ␮g of vector was used.
`Systemic dosing of siRNAs was at 0.3 to 30 mg/kg thrice
`daily. To allow recovery of the liver from the disruption
`caused by HDI, systemic dosing was started 72 hours after
`HDI.
`HBV DNA Analysis. Viral DNA was extracted from
`50 ␮L mouse serum using QIAmp 96 DNA Blood kit
`(Qiagen, Valencia, CA), according to the manufacture’s
`instructions. HBV DNA levels were analyzed using an
`ABI Prism 7000 sequence detector (Applied Biosystems,
`Foster City, CA). Quantitative real-time polymerase
`chain reaction was carried out using the following primer
`and probe sequences: forward primer, 5⬘-CCTGTAT-
`TCCCATCCCATCGT (HBV nucleotides 2006-2026);
`reverse primer, 5⬘-TGAGCCAAGAGAAACGGACTG
`(HBV nucleotide 2063-2083); and probe FAM, 5⬘-
`TTCGCA AAATACCTATGGGAGTGGGCC (HBV
`nucleotide 2035-2062). The psHBV-1 vector containing
`the full-length HBV genome was used as a standard curve
`to calculate HBV copies per milliliter of serum.
`HBV RNA Analysis. Total cellular RNA was isolated
`from approximately 200 mg liver sections using Tri-Re-
`agent (Sigma) according to the manufacture’s instruction.
`The extracted RNA was digested with DNAse I (TURBO
`DNAse, Ambion, Austin, TX) before analysis. Twenty
`micrograms of total RNA was separated on a 1% agarose-
`formaldehyde gel and was transferred to a MaxSense hy-
`
`Inc.,
`(Enzo Diagnostics,
`bridization membrane
`Farmingdale, NY). A full genome length HBV probe was
`generated from the psHBV-1 vector. Mouse ␤-actin
`probe DNA was obtained from Ambion. DNA probes
`were labeled with 32P-dCTP using a random primer mix
`(DECAPrime II, Ambion). The blot was hybridized over-
`night at 55°C in PerfectHyb Plus hybridization buffer
`(Sigma) with a 1 ⫻ 106 cpm/mL probe. The blots were
`exposed to phosphor screens and were scanned on Molec-
`ular Dynamics Phosphoimager. Band intensity was ana-
`lyzed using ImageQuant software, and liver HBV RNA
`levels were expressed as a ratio of HBV to ␤-actin RNA.
`
`Results
`
`Modification and Stability of Synthetic siRNAs. To
`develop an siRNA molecule with suitable serum stability
`for therapeutic application, siRNAs were synthesized with
`chemical modifications. Through a series of sequential
`rounds of modification and testing for stability and activ-
`ity, siRNA duplexes were developed in which all 2⬘-OH
`groups were substituted. These siRNA duplexes lacking
`2⬘-OH groups are termed a “short interfering nucleic
`acid” (siNA). Several novel combinations of siNA substi-
`tutions were generated in which the sense and antisense
`strands lack 2⬘-OH groups and are differentially modi-
`fied. One of these modified siNA duplexes was used in the
`experiments in this report. This siNA duplex is composed
`of a sense strand containing 2⬘-fluoro substitutions on all
`pyrimidine positions, deoxyribose in all purine positions,
`with 5⬘ and 3⬘ inverted abasic end caps. The antisense
`strand contains 2⬘-fluoro substitutions in all pyrimidine
`positions, all purines are 2⬘-O-methyl substituted, and the
`3⬘ terminal linkage is a single phosphorothioate linkage.
`The siNA and unmodified siRNA duplexes were as-
`sessed for their resistance to degradation in human serum.
`Throughout the course of the experiments, constant levels
`of ribonuclease activity were verified. The extent and pat-
`tern of unmodified siRNA degradation (3-minute time
`point) did not change after preincubation of serum at
`37°C for up to 24 hours. Figure 1 shows the stability of
`the unmodified siRNA and the siNA to nucleases as a
`function of time. The unmodified siRNA single strands
`or duplexes were extremely unstable in 90% serum, with
`a half-life of less than 1 minute for the single strands and
`3.3 to 5 minutes for the duplexes. In contrast, the chem-
`ical modifications of the siNA provided dramatic im-
`provement in stability. As single strands, the siNA sense
`and antisense strand exhibited half-lives in human serum
`of 18 and 15.5 hours, respectively. Within the siNA du-
`plex, the sense strand demonstrated a half-life of 2.1 days,
`whereas the antisense strand had a half-life of 3 days. The
`
`

`

`1352 MORRISSEY ET AL.
`
`HEPATOLOGY, June 2005
`
`Fig. 2. Screen of conserved short interfering RNA (siRNA) sites in
`hepatitis B virus (HBV) RNA. Secreted hepatitis B surface antigen
`(HBsAg) levels were assayed by enzyme-linked immunosorbent assay
`from Hep G2 cells transfected with HBV expression vector and siRNAs to
`conserved sites in HBV RNA, or an irrelevant control at final concentra-
`tions of 25 nmol/L. HBsAg levels were assayed 3 days after transfection
`and are expressed as optical density units at 450 nm. Mean levels
`(⫾SD) were calculated from three replicate transfections.
`
`siRNA at 25 nmol/L. An example of the siRNA screen is
`shown in Fig. 2. Subsequent dose response analysis of the
`active siNAs led to the identification of a highly active site
`located at starting 5⬘ nucleotide 263 in the HBV genome
`(data not shown). The 263 site lies within the S-region of
`the HBV RNA, with the siNA to the site predicted to
`cleave three of four major transcripts. The sequence and
`chemical modifications of the HBV site 263 siNA is
`shown in Fig. 3.
`Potency of Stabilized siRNAs in HBV Vector-Based
`Mouse Model. Because the siNA to HBV site 263 dem-
`onstrated a high level of stability in serum and exhibited
`effective gene silencing activity in cell culture, the ques-
`tion of how the differences in stability between the un-
`modified siRNA and siNA duplexes would impact
`
`Fig. 3. Sequence and chemical modifications of the hepatitis B virus
`site 263 short interfering nucleic acid.
`
`interfering RNA (siRNA) and chemically
`Fig. 1. Stability of short
`modified siRNAs in human serum. (A) The top gel panel shows radiola-
`beled sense strand, either single stranded (ss) or duplexed (ds) to the
`antisense strand, after
`incubation in human serum at 37°C. Both
`single-stranded and duplexed time points are as follows: 1, 5, 15, 30,
`60, 120, 300, and 1,310 minutes. Marker lanes from left to right are:
`single strand incubated in water for 0 or 1,310 minutes followed by
`duplex at 0 and 1,310 minutes. The subsequent gel panels follow the
`same loading pattern with the radiolabel present in the RNA antisense,
`the chemically modified sense strand, and the modified antisense strand,
`respectively. (B) The fractions of full-length siRNA as a function of time
`were quantitated and fit to a first-order exponential. The half-lives for
`single strands and duplexes are shown.
`
`resulting increase in stability of the siNA compared with
`the unmodified siRNA duplex was approximately 900-
`fold.
`siRNA Target Site Selection in HBV Cell Culture
`System. Potential siRNA sites conserved in more than
`95% of available HBV sequences (GenBank) were iden-
`tified by Blast sequence comparisons. To test siNAs tar-
`geted to the HBV RNA for activity in a cell culture model,
`a replication-competent HBV cDNA, derived from the
`psHBV-1 vector, was cotransfected along with duplexed
`siRNA into Hep G2 cells. The initial screening of siNAs
`to 40 sites in the HBV RNA resulted in the identification
`of 7 active sites demonstrating more than a 70% decrease
`in HBsAg levels as compared with an unrelated control
`
`

`

`HEPATOLOGY, Vol. 41, No. 6, 2005
`
`MORRISSEY ET AL.
`
`1353
`
`ence in serum HBV DNA between the active or inverted
`siRNA and the saline-treated groups. Both the unmodi-
`fied siRNA and siNAs displayed a dose-dependent reduc-
`tion of serum HBV DNA in the mouse model. A 2.2 log10
`reduction of serum HBV DNA was observed with the
`unmodified siRNA duplex after a 1-␮g dose, whereas a
`3.7 log10 reduction was observed with the siNA at the
`same dose. No difference was observed between the RNA
`or siNA inverted control groups and the saline group at
`any siRNA dose level. The analysis of the serum HBsAg
`levels correlated with the HBV DNA levels in each group
`(Fig. 5B). There was little or no difference in the activities
`of the unmodified siRNA and the siNA at the lower dose
`levels, as seen with either the HBV DNA or HBsAg end
`points. The pattern of the dose responses and magnitude
`of HBV reductions for the unmodified siRNA and siNA
`were highly reproducible, with an average difference of
`0.2 log10 between corresponding data points in two inde-
`pendent studies.
`
`Fig. 5. Activity of stabilized short interfering nucleic acid (siNA) verses
`unmodified RNA short interfering RNA (siRNA) in a hepatitis B virus (HBV)
`co-hydrodynamic tail vein injection model. A hydrodynamic tail vein
`injection (HDI) containing 1 ␮g of the pWTD HBV vector and 0, 0.03, 0.1,
`0.3, or 1.0 ␮g siRNA was performed on C57BL/J6 mice. Active siRNA
`duplexes and inverted sequence controls in both unmodified RNA and
`stabilized chemistry were tested. The levels of serum (A) HBV DNA and
`(B) hepatitis B surface antigen (HBsAg) were measured 72 hours after
`injection and are expressed as mean log10 copies/mL (⫾SEM) and mean
`log10 pg/mL (⫾SEM), respectively, with n ⫽ 6 per treatment group. A
`dose-dependent reduction in both HBV DNA and HBsAg levels was
`observed with both the unmodified RNA and stabilized siRNAs. However,
`the magnitude of the reduction observed in the stabilized siRNAs-treated
`groups was 1.5 log10 (P ⬍ .0001) greater for both endpoints at the
`high-dose level.
`
`Fig. 4. Reduction of liver hepatitis B virus (HBV) RNA levels after
`co-hydrodynamic tail vein injection (HDI) administration of stabilized
`short interfering nucleic acid (siNA). An HDI containing 1 ␮g of the pWTD
`HBV vector and 1.0 ␮g of active or inverted control siNA was performed
`on C57BL/J6 mice. The levels of liver HBV RNA were determined by
`Northern blot analysis 72 hours after injection. The HBV RNA levels were
`normalized to actin messenger RNA (mRNA) and are reported as a ratio
`of HBV mRNA/actin mRNA (⫾SD). n ⫽ 3 per treatment group.
`
`activity in an in vivo model of HBV replication was ad-
`dressed.
`To examine the in vivo potencies of the siRNAs, a
`mouse model was used that uses HDI of a replication-
`competent HBV vector. To reduce the complicating fac-
`tors of systemic administration and to focus on only
`inherent in vivo potency of the siRNAs, initial experi-
`ments were performed in which the siRNA duplexes were
`codelivered with the HBV vector via hydrodynamic in-
`jection.10,11 To investigate in vivo activity of the stabilized
`siNAs, 1 ␮g of the pWTD vector and 1 ␮g of active or
`inverted control siNAs were codelivered to mice by HDI.
`Three days after HDI, levels of liver HBV RNA were
`determined by Northern blot analysis. As shown in Fig. 4,
`a significant reduction in HBV RNA was observed in the
`active siNA group compared with either the saline-treated
`control group (71%; P ⬍ .00126) or the inverted control
`group (40%; P ⬍ .0086).
`Dose-dependent silencing of HBV expression in the
`mouse model was examined by injection of 1 ␮g of the
`pWTD HBV vector along with 1, 0.3, 0.1, or 0.03 ␮g of
`either the unmodified siRNA or siNA duplex. Both active
`siRNAs and inverted control duplexes were examined in
`each siRNA chemistry. The animals were killed 72 hours
`after the injection, and levels of HBV DNA and HBsAg in
`the serum were assayed. Figure 5A shows the log10 differ-
`
`

`

`1354 MORRISSEY ET AL.
`
`HEPATOLOGY, June 2005
`
`The more limited knockdown of HBV RNA levels in
`the liver as compared with the reductions in serum titers is
`likely the result of a population of the pregenomic RNA
`within immature capsids that is sequestered and protected
`from siRNA-mediated degradation. Because the siRNAs
`are able to target the 2.1- and 2.4-kb transcripts as well the
`full-length transcript, it is likely that reductions in HBV
`protein production would have a negative impact on cap-
`sid maturation and release, thus accounting for the more
`dramatic decreases in serum titers.
`This result shows that the improved stability of the
`modified siRNA results in a more effective level of silenc-
`ing in vivo. Most notably, this is the first demonstration of
`in vivo activity of a modified siRNA completely lacking
`2⬘-OH residues.
`In Vivo Activity of Systemically Dosed Modified
`siRNA. As soon as the in vivo activity of the siRNAs
`against HBV was demonstrated by coadministration by
`HDI, the level of anti-HBV activity of unmodified siR-
`NAs and siNAs was examined by dosing via conventional
`intravenous injection. Because there have been reports of
`initial liver damage after hydrodynamic injection,10 dos-
`ing of siNAs was begun 72 hours after HDI. Our own
`examination of liver alanine aminotransferase and aspar-
`tate aminotransferase levels and histopathological analysis
`after HDI confirmed reports in the literature that the liver
`returns to near normal status by 72 hours after the initial
`HDI-induced injury (data not shown). To assess the in
`vivo activity of systemically dosed unmodified siRNAs
`and siNAs, the pWTD HBV vector was administered by
`HDI. Seventy-two hours later, the active siRNAs or in-
`verted controls were dosed via standard intravenous injec-
`tion thrice daily for 2 days. The animals were killed 18
`hours after the last dose, and levels of serum HBV DNA
`were examined. Although no statistically significant activ-
`ity was observed with the unmodified siRNAs at any dose
`(data not shown), a dose-dependent reduction in serum
`HBV DNA levels was observed in the siNA-treated
`groups in comparison with the inverted control or saline
`groups (Fig. 6). A statistically significant (P ⬍ .0006)
`reduction of 0.91 log10 was observed in the 30-mg/kg
`group compared with the saline group, whereas a 0.83
`log10 reduction (P ⬍ .0015) was seen in the 10-mg/kg
`group. These results demonstrate in vivo activity of a stan-
`dard intravenously administered modified siNA and
`highlight the need for chemical stabilization of siRNAs
`for in vivo applications.
`Discussion
`The potential use of siRNAs as therapeutic agents is
`likely to require improvements in siRNA stability and
`efficiency of delivery to specific target tissues. As a first
`
`Fig. 6. Activity of systemically administered short interfering RNA
`(siRNA) in a hepatitis B virus (HBV) mouse model. A hydrodynamic
`tail vein injection (HDI) was administered that contained 0.3 ␮g of
`the pWTD HBV vector. Stabilized active short interfering nucleic acid
`(siNA) and inverted control were administered via standard intrave-
`nous injection beginning 72 hours after HDI. The siRNAs were dosed
`at 30, 10, or 3 mg/kg thrice daily for 2 days. The animals were killed
`18 hours after the last dose was administered, and the levels of
`serum HBV DNA were determined by quantitative real-time polymer-
`ase chain reaction. Serum HBV DNA levels are expressed as mean
`log10 copies/mL (⫾SEM), with n ⫽ 6 per
`treatment group. A
`dose-dependent decrease in serum HBV DNA levels was observed
`with the stabilized siRNAs in comparison with the saline or inverted
`control groups. In the high-dose group, a reduction of 0.91 log10 (P ⬍
`.0006) was observed compared with the saline group.
`
`step in this process, we have developed a unique combi-
`nation of chemical modifications of synthetic siRNAs
`that provide greatly increased stability in serum as well as
`maintain effective levels of gene silencing.
`Given that RNAi requires recognition of siRNAs by
`the RNA-induced silencing complex protein complex,
`it is not surprising that siRNA function is sensitive to
`chemical modification. The interplay between inter-
`nucleotide linkage changes, base and sugar modifica-
`tions, and effects on helical character are likely to be
`complex and may preclude simple rules about which
`modifications are permissible. In addition, modifica-
`tions are likely to have dissimilar impact on different
`primary sequences. Recently, several studies have ex-
`plored some of these variables. Global replacement of
`ribose residues with either 2⬘-deoxyribose or 2⬘-O-
`methyl substitution in either strand has been reported
`to block or to reduce silencing activity.12,13 It has been
`reported that when all uridine positions were present as
`2-fluoro modifications, activity is
`significantly re-
`duced, and when additional 2⬘-fluoro substitutions are
`added, further severe reduction in activity is seen.14 In
`contrast, we and others13,15 have found no reduction in
`activity when 2⬘-fluoro substitutions were made on all
`pyrimidine residues on both the sense and antisense
`strands. Layzer et al.16 observed no difference in activ-
`ity between unmodified RNA and 2⬘-flouro pyrimi-
`dine modified siRNAs in a co-HDI mouse model
`targeting a luciferase reporter gene. This is in contrast
`to our results reported here in which there is a marked
`
`

`

`HEPATOLOGY, Vol. 41, No. 6, 2005
`
`MORRISSEY ET AL.
`
`1355
`
`improvement in the in vivo activity of the fully stabi-
`lized siNAs compared with unmodified siRNAs. In ad-
`dition to 2⬘-fluoro substitution on pyrimidines, we
`found that the remaining purine positions were ame-
`nable to substitutions of 2⬘-deoxyribose or 2⬘-O-
`methyl with minimal reduction in silencing activity.
`These differences between our results and the previ-
`ously reported effects of modifications on silencing ac-
`tivity may be because of our use of unique
`combinations of modifications and the addition of ter-
`minal capping structure.
`The hydrodynamic delivery of nucleic acids in the
`mouse was described by Liu et al.,10 who showed that
`the vast majority of the injected nucleic acid is deliv-
`ered to the liver by this technique. Yang et al.11 first
`demonstrated that hydrodynamic injection of a repli-
`cation-competent HBV vector resulted in high levels of
`HBV replication in the livers of injected mice. In the
`vector-based model, HBV replicates in the liver of im-
`munocompetent mice for 7 to 10 days, resulting in
`detectable levels of HBV RNA and antigens in the liver
`and of HBV DNA and antigens in the serum. Several
`reports have documented the use of the HBV vector
`model to examine the in vivo activity of co-HDI ad-
`ministered HBV-targeted unmodified siRNAs17,18 or
`vector-expressed short hairpin RNAs19 in silencing
`HBV gene expression.
`We have optimized this model further by greatly
`reducing the amount of HBV vector injected per
`mouse, permitting the examination of dose-dependent
`siRNA activity at levels that are potentially more clin-
`ically relevant. Whereas previous reports used either
`unmodified RNA siRNAs or vector-expressed shRNAs,
`we have demonstrated the importance of stabilizing
`siRNA for in vivo activity. With co-HDI administra-
`tion of the siRNAs, we have demonstrated a 1.5 log10
`greater reduction in HBV replication with the stabi-
`lized siNA compared with unmodified siRNA. The re-
`ductions we have observed in serum titers with the
`stabilized siNAs are far more than the 70% to 88%
`decreases previously reported with unmodified RNA
`siRNAs or expressed siRNAs.
`Most notably, we have demonstrated in vivo activity
`of stabilized siNAs via standard intravenous adminis-
`tration. This is the first demonstration of siRNA in vivo
`activity in a hepatitis animal model with a clinically
`viable route of administration. Although we have seen
`a nearly 1 log10 reduction of serum HBV levels with
`siNAs in an HBV mouse model, the high dose level and
`frequency of administration required to produce this
`antiviral effect clearly indicates that improved effi-
`ciency of delivery will be required to develop siNAs
`
`into a clinically viable treatment approach. This is con-
`sistent with recently reported work showing reduced
`expression from an endogenous gene in the livers of
`mice treated with high doses of an siRNA.20 Although
`the results show the reductions in HBV titers to be
`siRNA sequence specific, the potential contribution of
`toxicity from the stabilized siRNAs has not been ruled
`out strictly and will need to be investigated further.
`With the advent of effective siRNA delivery strategies,
`characterization of potential toxicity and immuno-
`stimulation from siRNAs will be critical for develop-
`ment of the molecules into a clinically viable strategy.
`In conclusion, based on our analysis of the pharmaco-
`kinetic properties of unformulated siNAs, less than 1% of
`a 30-mg/kg intravenously administered dose reached the
`liver (data not shown). This suggests that we are achieving
`significant reduction in HBV replication with extremely
`small amounts of the siNAs in hepatocytes. Progress cur-
`rently is being made in the delivery of siNAs to hepato-
`cytes, which is likely to result in increased antiviral activity
`with clinically viable dose levels and frequency.
`
`Acknowledgment: We thank Roger Aitchison for
`statistical analysis and Aleem Siddiqui of the Univer-
`sity of Colorado Health Sciences Center for the gift of
`vectors.
`
`References
`1. Fire A, Xu S, Montgomery MK, Kostas SA, Driver SE, Mello CC. Potent
`and specific genetic interference by double-stranded RNA in Caenorhab-
`ditis elegans. Nature 1998;391:806-811.
`2. Zamore PD. Ancient pathways programmed by small RNAs. Science
`2002;296:1265-1269.
`3. Hammond SM, Bernstein E, Beach D, Hannon GJ. An RNA-directed
`nuclease mediates post-transcriptional gene silencing in Drosophila cells.
`Nature 2000;404:293-296.
`4. Zamore PD, Tuschl T, Sharp PA, Bartel DP. RNAi: double-stranded
`RNA directs the ATP-dependent cleavage of mRNA at 21 to 23 nucleotide
`intervals. Cell 2000;101:25-33.
`5. 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;411:494-498.
`6. Centers for Disease Control (CDC). Viral Hepatitis B Fact Sheet. Be-
`thesda, MD: CDC, 2002.
`7. Beasley RP. Hepatitis B virus. The major etiology