Quantitative assessment of the antiviral potencies of 21 shRNA vectors targeting conserved, including structured, hepatitis B virus sites

Background & Aims

RNA interference (RNAi) may offer new treatment options for chronic hepatitis B. Replicating via an RNA intermediate, hepatitis B virus (HBV) is known to be principally vulnerable to RNAi. However, beyond delivery, the relevant issues of potential off-target effects, target site conservation in circulating HBV strains, and efficacy of RNAi itself have not systematically been addressed, nor can the different existing data be quantitatively compared. The aim of this study was to provide such information.

Methods

To focus on the intracellular RNAi process itself and minimise other variables affecting overall RNAi efficacy, we used a robust co-transfection system to quantitatively assess the relative potencies of 21 small-hairpin (sh) RNA vectors, targeting conserved sites throughout the HBV genome, against viral RNAs, proteins, nucleocapsids, and secreted virions under standardised conditions.

Results

The approach enabled a distinct efficacy ranking, with the six most potent shRNAs achieving ⩾95% reductions in virion formation, sequence-specifically and without detectable interferon induction, yet by differentially affecting different steps. Efficacy correlated poorly with predictions and was not principally abolished by target structure. Sequence comparisons suggest that truly conserved, RNAi-targetable sequences comprise less than 500 nucleotides of the circulating HBV genomes.

Conclusions

The HBV genome can harbour only a finite number of optimal target sites, but current predictions are poorly suited to constrain the number of possible candidates. However, the small size of the highly conserved sequence space suggests experimental identification as a viable option.

Introduction

RNA interference (RNAi) is induced by double-stranded (ds) RNAs that are processed into ∼21- to 25-mer ds small interfering RNAs (siRNAs) and loaded on the RNA-induced silencing complex (RISC). Upon sense strand removal, the antisense (guide) strand can pair to a complementary region in a target RNA, leading to its cleavage (‘slicing’) and degradation. Less extensive complementarity may induce translational suppression and/or slicing-independent inactivation (for a recent review: [1]).

Therapeutic applications of RNAi, via synthetic siRNAs [2] or DNA vectors encoding small-hairpin (sh) RNAs [3], pose several extra challenges [4], prominently effective and specific in vivo delivery, and avoidance of adverse side effects. Such off-target effects may be caused by complementarity to non-target genes in only six or seven nucleotides of the guide strand (‘seed sequence’) [5]; by immune stimulation, e.g., interferon induction [6]; and by the delivery method itself [7]. Generally, side effects should be minimal for RNAi effectors with maximal specific efficacy. RNAi efficacy is determined on multiple levels, including cell delivery, precursor processing, and incorporation into the RISC of the desired strand. Equally important is considered target accessibility [8], which is affected by RNA structure and protein binding [9].

The targeting of infectious agents faces the additional problems of preexisting genetic variability and, for viruses in particular, rapid acquisition of resistance [10]; both are best accounted for by targeting evolutionarily conserved sites [11]. A major focus of this study was therefore to define efficient target sites on the hepatitis B virus (HBV) genome that are highly conserved.

HBV is a small DNA virus that replicates via reverse transcription [12], [13]. Despite its tiny genome (∼3.2 kb), it is one of the most successful human pathogens, causing severe liver disease [14]. Current therapies, i.e., type-I interferons (IFNs) and inhibitors of the viral reverse transcriptase, P protein, are only partially effective [15], [16].

Key features of the HBV replication cycle are outlined in Fig. 1A. Though HBV amplification in cell culture via infection is not feasible, some human hepatoma cell lines, such as Huh7, support the formation of progeny virions upon transfection with HBV vectors. Infection or transfection induces four classes of 3′ terminally colinear viral transcripts (Fig. 1B). Several subgenomic (sg) RNAs (∼2.4, 2.1, and ∼0.8 kb) serve as mRNAs for the surface proteins L (PreS1/PreS2/S), M (PreS2/S) and S, and the HBx protein. Of the two ∼3.5 kb transcripts, the longer precore mRNA encodes the precursor of hepatitis B e antigen (HBeAg). The pregenomic (pg) RNA acts as mRNA for core protein and P protein and, after encapsidation, as template for new viral DNA [13]. Encapsidation and reverse transcription require the binding of P to a ∼60-nt stem-loop, ε, near the 5′ end of the pgRNA [12], [13].

The direct dependence on the pgRNA for replication makes HBV an attractive target for RNAi, and several studies (reviewed in [17]) have reported supposedly efficient interference with viral gene products and/or replication. However, quantitative comparisons of RNAi efficacy are precluded by the multitude of different systems used to generate the HBV target; by differences in RNAi effectors, i.e., siRNA or differently designed shRNA vectors; and by the use of nonidentical conditions and/or read-out systems. Sequence specificity and general off-target effects have rarely been addressed; this is particularly relevant for viruses that, like HBV, are sensitive to interferon-mediated cell responses [18]. Target site conservation, if considered, is usually based on a few representatives of the eight recognised HBV genotypes A through H (reviewed in [19], [20]).

The aims of this study were to identify, under standardised conditions encompassing a minimal number of variables, conserved sites on the HBV genome that are particularly vulnerable to shRNA-mediated interference. To this end, we generated 21 shRNA vectors targeting 19 sites along the length of the HBV genome (Fig. 1B), including sites allowing us to address sequence specificity and target structure dependency of RNAi effects. For correlation with existing literature data, we included a site in the S gene that repeatedly [21], [22], [23], [24] has been reported to be a superior target. We refer to this site as ‘1740’, reflecting the position of its 5′ end in the core gene start site-based numbering system [25]; for comparison, relevant coordinates are also given in the alternative system [26] where the fourth nt of an EcoRI site starting at position 1280 marks the origin (indicated below by numbers separated by a slash). The full sequence of the infectious HBV2 isolate [27] used as target (genotype D, serotype ayw; Accession No. J02203) is provided in Supplementary Fig. S6.

To minimise factors influencing antiviral activity other than basic RNAi efficacy, we used a simple yet robust co-transfection protocol, with all shRNA vectors having the same design. Quantitative comparison of the levels of viral RNAs, proteins, and replication products enabled us to establish a distinct ranking of shRNA efficacies, and identify several shRNAs that massively inhibited progeny virus production in the absence of detectable off-target effects. Finally, a comparison of several hundred HBV sequences revealed a sequence space of at most 500 highly conserved nucleotides on which future efforts towards improved RNAi-based anti-HBV therapies may now be focused.