An antisense oligonucleotide may bind to its target nucleic acid either by Watson-Crick base pairing or Hoogsteen base pairing. In Watson-Crick base pairing, heterocyclic bases of an antisense oligonucleotides form hydrogen bonds with the heterocyclic bases of target single-stranded nucleic acids (RNA or single-stranded DNA), whereas in Hoogsteen base pairing, the heterocyclic bases of target are double-stranded nucleic acids. Both these models of binding by antisense oligonucleotides have the potential to regulate gene expression. Antisense oligonucleotides are an attractive potential alternative to conventional drugs as antiviral agents. A major advantage is the relatively simple rational design of oligonucleotides which should bind only to specific nucleic acid sequences, compared with conventional drugs which are frequently targeted against sites of unknown structure in proteins. Progress to date provides hope for the development of a new class of antiviral chemotherapeutics based on antisense oligonucleotides (Figure 3) [9].

Chemical modifications of antisense oligonucleotide

Although antisense technology can induce profound sequence-specific inhibition of gene expression in some settings, major problems include instability, delivery, and unwanted side effects of the oligonucleotides. Chemical modification of the oligonucleotides was found to improve the efficacy. These modifications affect one of the three main components of nucleotides: the bases, the sugar

(especially at the 2’ position of the ribose) or the phosphate backbone. Evaluations of some of these modifications, summarized as three “generations” of antisense oligonucleotides.

In first-generation modified oligonucleotides, one non-bonding oxygen atom in the phosphate group is substituted by a sulfur atom (phosphorothioate oligodeoxynucleotides). This modification confers higher stability to the molecules and thus increases their half-life. The advantage of the phosphorothioate-modified oligonucleotides is that it allows RNase H cleavage of the target sequence. A disadvantage is that the phosphorothioate modification introduces a negative charge that causes nonspecific interactions with other molecules, for example, cellular proteins. As a result, phosphorothioate-modified oligonucleotides are notorious for inducing severe side effects that are toxic for cells [10]. A second generation of modified oligonucleotides was obtained by the introduction of an alkyl group in the 2’ position of the ribose molecule of nucleotides. The generated molecules include 2’-O-methyl- and 2’-O-methoxyethyl-RNA. These molecular species are less toxic than phosphorothioate oligodeoxynucleotides [11]. The conformation changes induced by these modifications and, in particular, the presence of ribose, improve target binding, however they also abrogate the oligonucleotide: target mRNA duplex ability to activate RNase H, a crucial aspect of ASO activity. Thus, the inhibitory effect of these molecules is only exerted through the inhibition of mRNA translation and is thus lower than that of phosphorothioates. This drawback can be avoided by generating chimeric oligonucleotides with 2’-modified nucleotides placed only at the ends of the oligonucleotide, thereby leaving a central RNase-compatible DNA gap. These hybrid oligonucleotides are known as gapmers [12]. Third generation oligonucleotides are DNA and RNA analogues with extensive modifications of the phosphate backbone and the sugar ring. Examples are peptide nucleic acids (PNA), in which the backbone has been replaced by a peptide linker, morpholino phosphoroamidates (MF), and locked nucleic acids (LNA). These modifications provide further improved nuclease resistance and target affinity. In addition, the PNA and MF modifications allow enhanced cellular uptake. To combine all benefits, third generation modifications can be used in combination with unmodified residues in chimaeric gapmers [13] [14].

1) Antisense oligoribonucleotide

Antisense RNA-mediated gene inhibition is a widely used strategy not only for studying gene regulation and revealing gene function, but also for targeting disease states, especially cancer and viral infection. Antisense RNA has proven a potent inhibitor of gene expression and has the potential to inhibit retroviral replication at a number of stages in the virus life cycle by targeting both viral and cellular RNA sequences. To be an effective therapeutic agent, an antisense RNA must bind to a specified mRNA and prevent translation of the protein (Figure 4) [15].

The possibility of using an expression vector to produce an antisense RNA that suppresses a pathogenic condition has been examined. For example, antisense RNA complementary to three target regions in the 5’ leader/LTR of human immunodeficiency virus type-1 (HIV-1), the TAR region, the primer binding site and the splice donor (SD)-packaging signal (Ψ) region were stably expressed from the CMV IE promoter in Jurkat cells, and expression confirmed by RT-PCR. When challenged with HIV-1, cell lines expressing antisense RNA targeting the SD/ᴪ region showed significant inhibition of replication. These results suggest that the packaging signal (Ψ) of HIV-1 represents an attractive target for antisense RNA-based gene therapy, although the main mode of action of such molecules may well be through antisense effects at an earlier stage of replication than packaging [16].

Hepatitis B virus (HBV) is an infectious agent with about 200 million carriers worldwide. Many patients with HBV infection develop chronic hepatitis, cirrhosis and hepatocellular carcinoma (HCC). Unfortunately, treatment for chronic infection by this virus is far from satisfactory. Oligodeoxynucleotides can only temporarily suppress the pathogenic effect of integrated virus unless given continually,

whereas antisense RNA can be constitutively expressed in a stably transfected cell line, and thus, potentially provide long term protection against the pathogenic effect of integrated virus. Antisense RNAs derived from the HBV genome are promising candidates as antiviral agents and may serve as novel tools to identify functionally important regions of HBV transcripts. Antisense RNAs targeted to the HBsAg coding region can specifically inhibit envelope protein synthesis both in an in vitro translation system, and more importantly, in an HBsAg secreting cell line. The effects were long lasting, and remained demonstrable for more than 10 months post-transfection [17]. Respiratory syncytial virus is a leading cause of respiratory disease in infants, young children, immunocompromized patients, and the elderly. RNase L, an antiviral enzyme of the interferon system, can be recruited to cleave RSVgenomic RNA by attaching tetrameric 2’ 5’-linked oligoadenylates to an antisense oligonucleotide complementary to repetitive intergenic sequences within the RSV genome. RBI034, a 2’-O-methyl RNA-modified analogue of the 2 5A anti-RSV compound, was found to have enhanced antiviral activity in cell culture studies while also cleaving RSV genomic RNA in an RNase L- and sequence-specific manner. RBI034s efficacy in suppressing RSV replication in cell culture is 50 to 100 times better than ribavirin, the only approved drug for RSV infection [18].

2) Antisense oligodeoxyribonucleotide

The antisense oligodeoxyribonucleotides (AODNs) are usually synthetic single stranded DNA oligonucleotides, generally 12 - 30 nt in length, which interact via perfect Watson and Crick base-pairing with the target mRNA. Once introduced into the cell, pairing of the AODNs with its target blocks ribosomal translation and stimulates degradation of the RNA:DNA hybrid by cellular RNase H enzymes. Predictably, overexpression of RNase H amplifies the effect of various AODNs, while conversely inhibiting RNase H reduces their effect. Certain AODNs are not capable of activation of RNase H; instead they inhibit gene translation by steric competition with the translational machinery. In addition, AODNs can target intron-exon junctions to disrupt splicing (if bound to pre-mRNA). Furthermore, AODNs can also disrupt RNA trafficking by occupying protein-RNA interaction sequences necessary for correct intracellular localization (Figure 5) [19].

AODNs have been widely used to interfere with the gene expression of a number of human viral pathogens, including influenza virus, vesicular stomatitis virus, respiratory syncytial virus (RSV), human papillomavirus (HPV), herpes simplex virus, human immunodeficiency virus (HIV), hepatitis, and others [9]. The first AODN drug approved by the US Food and Drug Administration was Vitravene (Fomivirsen) in 1998, which targets the cytomegalovirus (CMV) IE2 gene for the treatment of CMV induced retinitis. Fomivirsen is a phosphorothioate ASO (first-generation) and binds to the IE2 fragment of mRNA of CMV that is involved in viral replication. Fomivirsen is administered as an intravitreal injection and has been found to have long duration of action; plasma concentrations are negligible. Metabolism of fomivirsen is locally mediated by exonuclease enzyme in the eye. No significant drug interactions have been observed with fomivirsen except that it should not be administered within 4 weeks following administration of the antiretroviral agent cidofovir. The drug is administered intravitreally once a week and the median time for progression of retinitis in patients was found to be 71 - 90 days. No serious adverse drug reactions were observed during clinical trials for intravitreal fomivirsen. Reports of occurrence of reversible bull’s-eye maculopathy with fomivirsen administration have been documented without any effect on vision. Following the advent of highly active antiretroviral drugs for treating infections with human immunodeficiency virus, a decline in the rate of CMV retinitis in patients was observed. Hence, fomivirsen was withdrawn by the marketing authorization holder for commercial reasons [20].

An alternative antisense-based antiviral approach relies on a 937 nt RNA molecule complementary to the env gene of HIV. This antisense RNA, called VRX496, is inserted into a HIV-1-based lentiviral vector which retains the long LTRs of HIV, thus its expression is up-regulated upon infection with wild type HIV. When the virus starts to replicate in the host cell, the antisense RNA prevents the production of the envelope protein, thereby blocking the HIV replication cycle. VRX496 is currently being tested in phase II clinical trials [21]. Another promising approach to inhibit HIV was developed by Matzen et al. which is based on an asON complementary to the polypurine tract. The viral RNase H, activated by this asON/viral RNA heteroduplex, cleaves the viral genome and thus destroys it. In the case of positive-strand RNA viruses, such as

HCV, the replication takes place entirely in the cytoplasm of the host cell. Therefore, for those viruses an asON-mediated cleavage of the viral RNA by the nuclear enzyme RNase H is not possible [21]. ISIS developed a 20 nt PS-modified asON (ISIS-14803) which targets a region within the internal ribosome entry site (IRES) of HCV RNA and acts via a steric block mechanism. ISIS-14803 was tested in clinical trials, but it showed a very low anti-HCV activity in patients, thus studies were discontinued [22]. Recently, Warren et al. described the application of positively charged phosphorodiamidate morpholino-modified asONs in nonhuman primates infected with Zaire Ebola virus (ZEBOV) or Lake Victoria Marburg virus (MARV). The recovery rate of infected monkeys upon treatment with asONs was 60% and 100% after infection with ZEBOV and MARV, respectively. These asONs, which will soon be entering phase I clinical trials, sterically hinder the translation of VP24 and VP35 viral transcripts which are important for ZEBOV and MARV replication [23].

Ribozymes are promising tools for the specific inhibition of viral gene expression and replication. They represent one of the most attractive developments of antisense nucleic acids, which have been shown in the past few years to act as antiviral agents. Ribozymes complex with target sequences via complementary antisense sequences and hydrolyze the target site [23]. Ribozymes are single stranded RNA molecules (50 - 100 nt) having a well-defined tertiary structure allowing them to catalyze chemical reactions without the requirement of any protein. They are naturally occurring molecules originally discovered in Tetrahymena thermophilia by T.R. Cech who was awarded in 1989 the Nobel Prize for Chemistry [24]. Ribozymes catalyze reactions such as hydrolysis of phosphodiester bonds, peptide bond formation, ligation, as well as polymerization and can be subdivided into cis- (intramolecular catalysis) or trans-acting (intermolecular catalysis) molecules [25]. Of the different types of ribozymes, the hammerhead and hairpin ribozyme have been studied extensively (Table 1) [26].

Hammerhead ribozymes are typically ~40 nt long and have a secondary structure formed by three helical domains (domains I, II, and III), surrounding a junction containing the catalytic core, defined by the presence of specific nucleotides (Figure 6(A)) [26]. While helix II is formed intramolecularly by the ribozyme itself, helices I and III depend on pairing of the single-stranded ribozyme sequence with its complementary sequence in the substrate, and can thus be chosen at will to match the sequence of the target RNA [27]. Hairpin ribozymes consist of 4 paired helices (helices I - IV) and two internal loops (Figure 6(B)) [26]. Enzymatic cleavage occurs at the level of a bond in loop A. In the plant satellite RNAs, cleavage takes place within the RNA molecule itself; however the sequences of helices I and II can be modified in order to render them complementary to a desired nucleotide sequence of the target RNA. In contrast to hammerhead ribozymes, hairpin ribozymes do not require the presence of a metal ion in the catalytic core [28].

Generally, reactions mediated by ribozymes are multiple-turnover and sequence-specific, both characteristics make these molecules suited candidates for therapeutic application (Figure 7) [29]. For this purpose, usually trans-cleaving hammerhead ribozymes are used, which are characterized by a central catalytic domain and two flanking arms (6 - 12 nt) complementary to the target RNA. To improve stability and affinity for the target, modified nucleotides such as locked nucleic acids (LNA) or 2’-O-methyl (2’-OMe), are commonly introduced into the flanking arms of the ribozyme. Ribozymes can be endogenously transcribed from vectors or chemically synthesized and delivered to cells [30].

For exogenous delivery the ribozyme may be synthesized in vitro or even

chemically, and the resulting RNA then is delivered to the cell by standard transfection methods, e.g., by lipofectamine or electroporation. An exogenous synthesis bears the added possibility of modifying the RNA molecule during or after synthesis in a nonphysiological way, to probe structural modifications at individual positions or to provide an improved resistance against cellular nucleases, and therefore achieve a longer lifetime. Such modifications include substitution of nucleotides by 2’-fluorocytidine in the nonessential parts of the ribozyme, or insertion of phosphorothioate internucleotide bonds. Such modifications will increase the stability of ribozymes against nucleases without serious losses in cleavage activity. After exogenous transfection of such RNAs or RNA variants, the ribozyme-mediated inhibitory effect will inevitably be abolished when the ribozyme itself is finally destroyed in the cellular environment; otherwise it has to be supplied repeatedly. The other way to introduce ribozymes into the cell is by endogenous expression following transfection of a ribozyme-encoding cDNA. From an expression cassette, the ribozyme cDNA will be transcribed repeatedly by cellular RNA polymerases. By making the catalytic RNA part of a larger transcript (mostly mRNA or tRNA), it shares the fate of the RNA species in which it is inserted, in its promoter-controlled expression, and also its cellular localization, as well as decay rates. Endogenous expression includes the possibility of inserting a selection marker, e.g., an antibiotic-resistance gene, into the ribozyme-encoding plasmid DNA [31].

Human papillomaviruses (HPV) are double stranded DNA viruses (8 kb) that infect stratified epithelia (i.e. skin or genital tract). The relationship between cervical cancer and HPV is related to the production of two viral early proteins E6 and E7, which are sufficient and necessary to acquire and maintain a transformed phenotype. Two hammerhead ribozymes were directed to the translational start site of HPV-16 E6 and E7 genes. Both ribozymes (nt 110 and 558) were assayed in vitro and showed efficient cleavage of the target in different conditions of ionic strength, [Mg⁺⁺], and temperature. Evaluation of the simultaneous activity of both ribozymes led to the conclusion of a cooperative effect suggesting a dual use of this ribozyme to disrupt HPV-16 E6/E7 expression [32]. A comprehensive approach on 5’-GUC-3’ hairpin ribozyme target sites on HPV-16 E6/E7 genes resulted in six different hairpin ribozymes. However, only two ribozymes processed significantly (>90%) the target sequence in vitro. Ribozyme in vivo performance resulted in the cis-inhibition of E6/E7 expression and growth delay and suppression of immortalization on transfected human normal keratinocytes [33].

AIDS and its associated disorders are caused by the human immunodeficiency virus type 1 (HIV-1). The HIV-1 RNA genome (approximately 9.8 kb) contains replicative (gag, pol, env), regulatory (tat, rev, tev, vpr, nef) and accessory (vpu, vif) genes [34]. Because the HIV genome is RNA, all of these genes are potential targets for ribozyme therapy. Therefore, there is a wide variety of targets for ribozyme therapy including mRNAs from host cell receptors and viral elements. For the retrovirus HIV-1, a tat-vpr-specific anti-HIV ribozyme, called OZ1, is currently been tested in phase II clinical trials. OZ1 consists of a Moloney murine leukemia virus-based gammaretroviral vector (LNL6) containing a gene that encodes a ribozyme targeting the overlapping vpr and tat reading frames of HIV-1. This approach is based on a gene transfer by which the DNA of the OZ1 retroviral vector is integrated into hematopoietic stem cells. In vitro experiments showed long-term inhibition of HIV-1 replication by OZ1 in cell culture without occurrence of escape mutants. In HIV-infected patients treated with OZ1, the CD4+ lymphocyte counts were higher than in the group treated with placebo. Then again, over time a decline in the percentage of patients who maintained the integrated DNA from the OZ1 retroviral vector was observed [35]. Recently, a very promising hammerhead ribozyme (HHPol13) targeting the HIV-1 pol gene was also described. The antiviral effect of HHPol13 was compared to the one of an siRNA targeting the same region in the viral pol gene. Results showed that the ribozyme led to a drastic reduction of HIV-1 replication in cell culture experiments while the siRNA showed no inhibition [36]. The Hepatitis B Virus (HBV) is a 3.2 kb double-stranded DNA virus from the family hepadnaviridae. HBV X protein (HBx) is a transcriptional activator required for the establishment of HBV infection and implicated in hepatocarcinogenesis. The specific repression of HBx gene expression may be a good strategy to block the progression of liver diseases caused by HBV infection. Two ribozymes targeting HBV Surface and X transcripts were designed to cleave 3’ to nt 476 and 1649, respectively. The ribozyme target sites were selected on the basis that they were highly conserved amongst different viral isolates, they contained the 5’-GUC-3’ triplet preferred by HH ribozymes and were present in multiple transcripts. Ribozyme treatment caused a marked reduction in Surface mRNA compared to untransduced cells [37]. The HBV encapsidation signal (ε) is a highly conserved cis-element present twice on the HBV RNA. Effective in vitro cleavage inside ε has been achieved by hammerhead ribozymes containing flanking sequences complementary to an adjacent less structured region suggesting a potential target for in vivo ribozyme therapy [38].

The hepatitis C virus (HCV) is an enveloped positive sense RNA virus. A ribozyme, called Heptazyme targeting the highly conserved IRES in the 5’-untranslated region (5’-UTR) of the HCV genome, was tested in Phase II clinical trials. Heptazyme led to a 10% reduction of the viral load in the serum of patients. However, toxicological tests for this ribozyme were not encouraging and led to discontinuation of the clinical studies [39]. Most recently, Levesque et al. coupled anti-HCV ribozymes with a specific on/off adaptor, called SOFA, which enable cleavage activity only in the presence of the desired RNA substrate increasing the selectivity of the ribozymes. They examined in vitro the ability of SOFA-coupled ribozymes to inhibit HCV replication using a luciferase-based replicon and observed up to 40% reduction in the luciferase activity [40].

A non-natural variant of ribozymes are the so-called deoxyribozymes or DNAzymes, in which the nucleic acid is DNA instead of RNA (Figure 8). Similar to ribozymes, the molecules also consist of a catalytic domain flanked by sequences pairing with the target RNA [41]. The discovery of these molecules has offered an answer to some of the limits of conventional ribozymes. First, deoxyribozymes, being made of DNA, are easier and less expensive to synthesise, while DNA is much more resistant than RNA to degradation both in vitro and in vivo. In fact, studies have shown that, in contrast to ribozymes, very stable deoxyribozyme constructs can be created via the introduction of a minimal number of modifications, reducing the possibility of nonspecific toxic effects. Another intrinsic advantage of deoxyribozymes lies in the fact that a DNA:RNA duplex tends to dissociate more easily than an RNA:RNA duplex; thus, when dealing with long-sequence targets, a deoxyribozyme should be more active—under multiple-turnover conditions—and potentially more selective than a traditional ribozyme (Figure 9) [29]. The superior chemical stability, lower potential toxicity,

and improved catalytic efficiency make deoxyribozymes substantially more attractive than ribozymes as candidates for exogenous applications. However, a deoxyribozyme cannot be produced continuously after the vector that encodes it is introduced into a particular tissue, because only ribozymes are produced from the DNA sequence. Therefore, deoxyribozymes must be delivered directly to affected cells [42] [43].

Deoxyribozymes can be considered as promising antiviral agents due to their ability to efficiently cleave RNA molecules with high specificity. Several studies have shown the usage of DNAzymes to inhibit influenza A and B virus replication in host cells. DNAzymes have also been shown to inhibit the replication of SARS coronavirus (SARS-CoV), Respiratory syncytial virus (RSV), human rhinovirus 14 and HCV. Influenza A virus genome segment 7 encodes protein M1, which is the matrix protein playing crucial role in the virus life cycle. In a study, several novel 10 - 23 DNAzymes and hammerhead ribozymes, specifically targeted to cleave at the conserved domains of the influenza virus M1 RNA are used. The Mg²⁺-dependent sequence-specific cleavage of M1 RNA was achieved by both the DNAzymes and hammerhead ribozymes in a dose-dependent manner. This combination of catalytic DNAzymes and hammerhead ribozymes with antisense molecules, in principle, resulted in more effective gene suppression, inhibited the whole virus replication in host cell, and thus could be exploited for therapeutic purposes [44].

The BM2 ion channel protein of influenza B virus is highly conserved and is essentially required during the uncoating processes of virus, thus an attractive target for designing antiviral drugs. A group screened several DNAzymes containing the 10 - 23 catalytic motif against the influenza B virus BM2 RNA. DNAzymes directed against the predicted single-stranded bulge regions showed sequence-specific cleavage activities. The DNAzyme209 not only showed significant intracellular reduction of BM2 gene expression in transient-expression system but also provided considerable protection against influenza B virus challenge in MDCK cells [44].

RSV has also been targeted using DNAzymes. A series of Dzs were designed and synthesized to silence RSV N, M2 and F genes. One Dz (DZn1133) cleaved the RSV RNA in vitro, inhibited the transcription and expression of F viral gene, reduced the RSV production by about 7 logs and protected more than 90% of RSV-infected Hep-2 cells from the cytopathic effect when used at the concentration of 8 µM. Additionally, 10 wild RSV strains (including both subgroups A and B) isolated from patients showed significant suppression by DZn1133 with higher anti-RSV activity than that for ribavirin (1-β-D-ribofuranosyl-1-H-1,2,4-triazole-3-carboxamide) or an antisense oligonucleotide (ODN) complementary to the same region of the N RNA [45].

SARS-CoV has also been targeted using DNAzyme molecules. The use of a DNAzyme to target the 5’-UTR of a highly conserved fragment in the SARS-CoV genome to suppress viral replication has been reported. A study showed the efficacy of a mono-DNA enzyme (Dz-104) possessing the 10 - 23 catalytic motif, both in vitro and in cell culture experiments. The Dz-104 showed an in vitro cleavage of the SARS-CoV RNA substrate and reduced the expression of the SARS-CoV 5’UTR-eGFP fusion RNA in the mammalian cells. The results provide a basis for the potential therapeutic usage of DNA enzymes to combat SARS infection [46].

A chimeric RNA/DNA oligonucleotide (chimeraplasts) was developed to correct point or frameshift mutations in episomal or chromosomal copies of genes. It consists of 68 nucleotides (nt) that base-pair to give a double-stranded molecule, in which one strand is comprised entirely of DNA, and the other strand comprises of two runs of 10 nt of 2’-O-methyl (2’-O-Me) RNA, separated by a 5 nt stretch of DNA (Figure 10(A)) [47]. A short region predicted to have a high

melting temperature (a GC clamp) is located at the 3’ end, and four T residues separate the double-stranded regions, allowing the energetically favourable formation of tight hairpins. The base that is to be mutated lies at the centre of the stretch of five DNA nucleotides on the chimeric strand, and has its complement at the centre of the all DNA strand. This ensures that the chimeraplast is capable of intramolecular base pairing along its entire length (Figure 10(B)) [47]. It is presumed that invasion of the host genome is initiated and stabilised by the chimeric strand; the Tm of a hybrid between DNA and 2’-O-Me RNA is higher than that of either DNA-DNA or DNA-RNA, and will thus be more stable [48].

Severe acute respiratory syndrome (SARS) is a severe pulmonary infectious disease caused by a novel coronavirus. To develop an effective and specific medicine targeting the SARS-coronavirus (CoV), a chimeric DNA-RNA hammerhead ribozyme was designed and synthesized using a sequence homologous with the mouse hepatitis virus (MHV). The synthetic ribozyme cleaved the synthetic target MHV and SARS-CoV RNA into fragments of predicted length. The chimeric DNA-RNA hammerhead ribozyme targeting SARS-CoV significantly inhibited multiplication of MHV in DBT cells by about 60%. The chimeric DNA-RNA hammerhead ribozyme targeting SARS-CoV significantly inhibited the expression of SARS-CoV RNA in 3T3 cells transfected with the recombinant plasmid. The chimeric DNA-RNA ribozyme targeting SARS-CoV significantly inhibited MHV viral activity and expression of recombinant SARS RNA in vitro. These findings indicate that the synthetic chimeric DNA-RNA ribozyme could provide a feasible treatment for SARS [49].

RNA interference (RNAi), an ancient defense pathway, is a posttranscriptional mechanism of gene silencing through chromatin remodeling, inhibition of protein translation, or direct mRNA degradation. RNAi is a naturally occurring process of gene regulation present in plants and mammalian cells. This phenomenon was first observed in plants in the late 1980s, but its molecular mechanism remained unclear until it was discovered in 1998 by Fire et al., in the nematode C. elegans [50]. The process is triggered by the presence, in the cytoplasm, of small, double-stranded RNA (dsRNA) molecules of 21 - 26 bp, with two unpaired bases at the 3’ ends and phosphorylated 5’ ends. One strand of these dsRNA molecules (the guide strand) is assembled into a ribonucleoprotein complex called RISC (RNA-induced silencing complex), which selectively binds a target RNA, inducing its cleavage or blocking its translation. By pairing with the target mRNA, the guide strand confers binding specificity to RISC. The dsRNA molecules that activate RNAi can be generated by one of at least three pathways: the microRNA (miRNA), the short interfering RNA (siRNA) and short hairpin RNA (shRNA), also called expressed RNAi activator pathways [51].

1) MicroRNAs

MicroRNAs (miRNAs) are noncoding RNA molecules of 18 - 25 nucleotide in length that regulate at posttranscriptional level the expression of genes by binding to the 3’-UTR of target genes. A miRNA can regulate different mRNAs, because they are not specific to a single mRNA. They are generated by the cells from specific RNA transcripts forming stem-loop structures (i.e., palindromic regions interrupted by a loop of non-complementary nucleotides). The genes coding for these transcripts are localized within introns of protein-coding genes (40% of cases) or in intergenic regions (60% of cases); in the latter situation, the same genomic region often contains a cluster of different miRNA genes, one close to the other. Most miRNA genes are transcribed by RNA polymerase II as long (in some cases, over 1000 nt) RNA transcripts. These transcripts, known as primary miRNA (pri-miRNA), contain a CAP and a poly-A at their 5’ and 3’ ends respectively [52]. They are processed in the nucleus by a complex consisting of at least two components: the RNase III enzyme Drosha and a dsRNA-binding protein called Pasha in Drosophila or DGCR8 (DiGeorge critical region 8) in mammals. This complex, also known as the Microprocessor, processes the pri-miRNA to generate the pre-miRNAs, which consists of a short stem-loop structure of 65 - 75 nt. These are transported from the nucleus to the cytoplasm by exportin-5, and here are further processed by a complex of proteins that includes the RNase III Dicer and the protein Loquacious (Loqs) in Drosophila or TRBP in humans to generate the final miRNAs, presenting two characteristic overhang nucleotides at their 3’ ends. Processing by Dicer is coupled with loading of the guide strand into RISC, which is the final effector of RNAi [52].

2) Short interfering RNAs

Short Interfering RNAs (siRNAs) are short double-stranded RNA segments with 21 - 23 nucleotides and are completely complementary to the mRNA sequence of the protein whose transcription is to be blocked. Unlike miRNAs, siRNAs are generated from longer precursor dsRNAs, which can be endogenously introduced into the cells or produced endogenously. Similar to miRNAs, these dsRNA molecules are also processed by Dicer and the final, RISC-mediated RNAi pathway is in common with that of miRNAs [53]. The main advantage of synthesized siRNA is that these molecules do not need to reach the nucleus to induce the therapeutic effect. As a drawback, stability must be improved in order to optimize the efficacy [54].

Chemical modification of siRNA

Gene silencing efficiency of siRNA also depends on its stability. Unmodified double-stranded RNA is much more stable as compared to single-stranded RNA. Nonetheless, the chemically modified nts utilized for synthesized siRNA have been introduced to further enhance its stability and thereby prolong its inhibition effects. The widely used class of modification involves modification of 2’ position of the ribose ring. siRNA molecules with uniformly substituted 2’ moieties in antisense and/or sense strands were not able to induce RNAi, whereas those with 2’ modification at 3’ overhangs or partial 2’ modification in base-paired region improved their stability in serum without loss of RNAi activity [55]. Several studies have shown that some siRNAs with a phosphorothioate backbone are active in RNAi assay, but the activity is impaired with increasing degrees of modification [56]. An alternative backbone modification is the boranophosphate linkage, in which the non-bridging phosphodiester oxygen is replaced with an isoelectronic borane (-BH₃) moiety. Compared to phosphorothioate siRNAs, boranophosphate counterparts are more nuclease resistant and less toxic [57].

3) Short hairpin RNA

Short hairpin RNA (shRNA), also called expressed RNAi activator, is a plasmid coded RNA that needs to be transcribed in the nucleus to downregulate the expression of a desired gene. It can be transcribed through either RNA polymerase II or III. The first transcript generates a hairpin-like stem-loop structure and is then processed in the nucleus by a complex containing the RNase II enzyme Drosha. The individual pre-shRNAs generated are finally transported to the cytoplasm by exportin 5. Once in the cytoplasm, the complex Dicer processes the loop of the hairpin to form a double-stranded siRNA. Since shRNA is constantly synthesized in the target cells, more durable gene silencing is achieved in comparison to other forms of RNAi. shRNAs represent an important tool in the assessment of gene function in mammals and are largely used as a research tool [58].

Mechanisms of gene silencing of interfering RNAs

RNAi leads to post-translational silencing of gene expression by multiple mechanisms, which include cleavage of target RNA, inhibition of translation, and sequestration into specific cytoplasmic compartments (Figure 11) [59]. The final effector of RNAi, from either the miRNAs or the siRNA or the shRNA pathway, is the ribonucleoproteins complex RISC, the catalytic component of which includes various members of the Argonaute (Ago) family of proteins. These proteins contain a PAZ domain, which binds miRNA/siRNA, and a PIWI domain, similar to that of RNases H, which is the effector of target RNA cleavage in correspondence of a specific nucleotide, corresponding to nucleotide at position 10 of the guide filament. This endonucleolytic activity (also known as Slicer activity, which is apparently exerted by Ago2 but not by other members of the Argonaute family within RISC) requires perfect pairing between the guide filament in RISC and the target RNA, a condition that is usually sought when siRNAs are administered exogenously to the cell. In this condition, RISC functions as an enzyme, progressively destroying one molecule of substrate after the other. In the case of most endogenously produced miRNAs, in contrast, pairing usually occurs with the 3’ UTR of the target mRNA and is not perfect, only requiring 7 nt at the 5’ end of the guide filament (from nucleotide in position 2 to nucleotide in position 8); these 7 nucleotidesnform the so-called seed sequence. In this case, silencing of gene expression is mainly due to the inhibition of translation of the target mRNA. Finally, a third mechanism of gene silencing is due to the ability of the RISC-associated Argonaute proteins to sequester the target mRNAs within specific cytoplasmic compartments, called P bodies, in which RNA degradation is believed to occur [59] [60].

RNAi application to viral infection

Because of the large therapeutic potential offered by RNAi against the pathogenic

viruses, many of them are being considered for medical use in the near future (Figure 12). It works against RNA viruses with negative- or positive-strand genome polarity, as well as against DNA viruses [6]. The RNAi, as a therapeutic tool, can be targeted against the various phases of the viral life cycle of DNA and RNA viruses including replication, transcription, assembly of new virions, and budding out of the target cells (Figure 13) [61].

1) RNAi prevents viral infection

The ability of synthetic siRNAs to inhibit viral production was determined in cultured cells by introducing siRNAs into the cells before viral infection and then assaying virus titers in the culture supernatants at different times after infection [62]. Influenza virus causes one of the most prevalent infections in humans. Introducing siRNAs and shRNA expression constructs into cultured cells and chicken embryos before virus infection was able to inhibit influenza virus production, especially those encoding NP, PA, and PB1. In addition, RNAi

inhibitory effects also occurred in cells that had been infected with virus prior to siRNA and shRNA introduction [63].

Poliovirus, a RNA virus, is able to invade the nervous system and cause paralysis. Gitlin et al. revealed that RNAi effectively prevented viral infection in mammalian cells by using siRNAs against the poliovirus genome [62]. Hepatitis C virus (HCV) infection is a major cause of chronic liver disease, which can lead to the development of liver cirrhosis and hepatocellular carcinoma. NS3 and NS5B siRNAs could specifically inhibit HCV RNA replication and protein expression in Huh-7 cells [64]. AcNPV (A. californica nucleopolyhedrovirus) is a member of baculoviruses family that naturally infects many different insect species. The introduction of dsRNA containing a sequence from ie1 or ie1-siRNAs into Sf21 cells and living insects could protect them against baculovirus infection. The inhibitory effects of siRNAs could be stable inside cells for over a week [65].

2) RNAi inhibits the expression of viral antigens

Many viral proteins don’t involve in viral RNA or protein synthesis, but take part in pre- or post-transcriptional aspects of the viral life cycle. A number of cellular membrane molecules may act as receptors for viruses. RNAi has widely been used in the inhibition of the expression of viral surface antigens in the research of viral diseases. CCR5, a human chemokine receptor protein, is a necessary coreceptor for infection by most strains of HIV-1. When siRNAs against CCR5 were introduced into human peripheral blood T lymphocytes, the expression of potent CCR5-siRNA would result in up to 10-fold inhibition of CCR5 synthesis on the cell surface over a period of 2 weeks [66].

Paramyxoviruses initiate infection by attaching to cell surface receptors and fusing viral and cell membranes. Three viral attachment proteins, glycoprotein (G) receptors, hemagglutinin-neuraminidase (HN) and fusion (F) proteins direct membrane fusion (formation of syncytia). The siRNA knockdown of F resulted in complete loss of syncytium without G or SH synthesis [67]. SARS-CoV (the SARS-associated coronavirus) has been classified as a novel member of the coronavirus family with a spike protein, which plays an important role in viral entry and pathogenesis. DNA vector-driven siRNA and shRNA plasmids against the spike protein of SARS-CoV could electively and specifically silence gene expression of the Spike protein in SARS-infected 293T cells [68].

3) RNAi suppresses the transcription of viral genome

Viruses need to transcribe its genome after entering the host cells. In addition to using cellular factors, viruses can generate their transcriptional activators such as E6 and E7 proteins of human papillomaviruses (HPVs). E6 siRNAs and shRNAs were transfected into HPV16-related cervical cancer cells. Experiments demonstrated that E6 siRNAs and shRNAs were potent in the suppression of viral oncogene expression. For RNA viruses, especially in retroviruses, gag, env and pol are essential for genome transcription. Avian Sarcoma Leucosis Virus (ASLV) is a positive-RNA virus. siRNAs matching two sequences in the gag gene of ASLV were introduced into cultured chicken DF-1 cells. After DF-1 cells were co-transfected by ASLV and siRNAs, the production of the RCAS virus in culture supernatants was dramatically reduced (90%) 2 day after treatment [69]. Moreover, the inhibition of transcription by anti-gag siRNAs and anti-env siRNAs against HIV-1 virus genome was also observed in CD4-positive cells [70].

4) RNAi blocks viral replication

Once a virus has entered a cell, it needs to replicate its genome. It has been shown that RNAi is induced by virus replication in animals. Activation of RNA silencing is thought to be involved in the inhibition of viral replication. West nile virus (WNV) is an enveloped arbovirus with a single-stranded, 11-kb positive-polarity RNA genome. siRNAs specific for WNV capsid or NS5 ORF were constructed in pHH21 and then transfected into 293T cells. Subsequently, the treated cells were infected with WNV. It was showed that WNV protein expression, genomic RNA synthesis, and infectious virus production were all dramatically reduced by siRNAs targeting these two distinct viral sequences [71]. The HBV genome is a double-stranded DNA molecule and generates four viral RNAs that encode the core protein/HBeAg and polymerase-reverse transcriptase, HBsAg, and X protein, respectively. Human hepatoma cells were co-transfected with HBV and siRNAs against HBV pre-genome RNA. The results also indicated that siRNA-mediated gene silencing could inhibit HBV replication through suppression of viral RNA [72]. shRNAs targeting HBV sequences also effectively inhibited viral replication and the production of core antigen in hepatocytes [73].

5) RNAi silences viral accessory genes

There are many accessory genes in viruses. These genes take part in the viral pathogenesis such as latency and persistence and regulate the expression of other genes. Silencing these genes should be good therapeutic strategies for the treatment of viral diseases. Epstein-Barr virus (EBV) is a prototype gamma herpesvirus with a dsDNA genome. Latent membrane protein-1 (LMP-1), encoded by EBV, is thought to be the only oncoprotein playing an essential role in cell transformation as well as nasopharyngeal carcinoma (NPC) metastasis. Li et al. showed that the stable suppression of LMP-1 by shRNA plasmid significantly altered EBV-positive NPC cell (C666) motility, substratum adhesion, and transmembrane invasion ability [74]. Application of RNAi to the study of accessory genes provides a chance to learn more about molecular mechanisms underlying viral infection so that people could develop more effective drugs to treat viral diseases.

6) RNAi hinders the assembly of viral particles

Assembly of viral particles varies with different virus families. The formation of the virus caspid has been well studied. Many caspid proteins are involved in the assembly of virus particles including VP4 and matrix proteins. Rotavirus causes severe diarrhea in infants and children worldwide. To combat this virus, Dector et al., utilized siRNA directed to the VP4, a viral structural protein that is essential for the attachment of the infecting virus to the cell surface. They showed a significant reduction in the number of viral particles produced in MA104, an infected monkey kidney cell line [75]. Vesicular stomatitis virus (VSV) is a member of non-segmented negative-stranded RNA viruses. VSV matrix gene (M) mRNA was shown to produce three proteins: M1, M2 and M3. M protein(s) can inhibit host gene expression and cause apoptosis of the host cell. In an attempt to determine which product of the M gene mRNA is involved in apoptosis and cell-rounding, siRNAs targeting to the region of target mRNA that is unique to M1 and upstream of M2 and M3 were transfected into HEp-2 cells. Results showed that these siRNAs abrogated the expression of all three proteins, and the cell detachment was concomitantly inhibited [76].

7) RNAi displays roles in virus-host interactions

Viruses can control both viral and cellular gene expression while host cells can also protect themselves against virus infection through immune response. The suppression of these genes by siRNAs can be also used to address the role of such genes in host-virus interaction. HCV belongs to the family Flaviviridae with a positive RNA genome. To determine whether HCV expression is interfered with RNAi, siRNAs against cellular (LaminA/C) and viral (HCV) RNAs were introduced into Huh-7.5 cells containing replicating HCV. Both cellular and viral RNAs were effectively silenced [77]. Bitko and his colleagues recently applied siRNAs-knockdown technology to determine the role of profilin in the RSV life cycle. It was showed that profilin stimulated RSV transcription in vitro, but was not absolutely necessary for virus replication in the previous studies. The knockdown of profilin showed that it had a minor effect on viral transcription but strongly inhibited the formation of actin stress fibers [67]. Nodaviral infection can trigger an RNA silencing-based antiviral response (RSAR) in Drosophila RSAR, which is a rapid virus clearance in the absence of expression of a virus-encoded suppressor. Li et al. showed that Drosophila RSAR was mediated by the RNA interference (RNAi) pathway, as the viral suppressor of RSAR inhibited experimental RNAi initiated by exogenous double-stranded RNA. Specific RNA silencing was induced by IFN. Antagonist proteins encoded by viruses are suppressors of the Drosophila S2 cells in response to viral RNA replication. Furthermore, this mosquito antiviral immunity is RISC dependent and sensitive to suppression evoked by the B2 protein encoded with either FHV or nodamura virus (NoV) [78]. The study of virus-host interactions can promote the understanding of RNAi machinery as well as the application of RNAi in the viral diseases, which will extremely draw more researchers to dedicate to it.

Antibody immunotherapy is revolutionizing modern medicine. Monoclonal antibodies are successfully used in the clinic. Antibodies are a crucial part of the body’s specific defense against infectious diseases and have considerable potential as therapeutic and prophylactic agents in humans and animals. The field has advanced dramatically over the past 40 years, driven in part by major advances in isolation and manufacturing technologies that have brought these important biologics to the forefront of modern medicine. However, the global uptake of monoclonal antibody biologics is impeded by biophysical and biochemical liabilities, production limitations, the need for cold-chain storage and transport, as well as high costs of manufacturing and distribution. Some of these hurdles may be overcome through transient in vivo gene delivery platforms, such as non-viral synthetic plasmid DNA and messenger RNA vectors that are engineered to encode optimized monoclonal antibody genes (Table 2). An alternative approach would be to leave the production of therapeutic antibodies to the body itself. These approaches turn the body into a biological factory for antibody production, eliminating many of the steps involved in bioprocesses and providing several other significant advantages, and differ from traditional gene therapy approaches. In general, the antibody heavy- and light-chain complementary DNA (cDNA) are examined and then optimized or designed for encoding and expression specifically for the type of vector chosen for in vivo delivery (Figure 14) [79]. The development of recombinant single-chain antibodies allows a genetic application strategy for prevention of infectious diseases. Genetic resistance to viruses would be a desirable alternative to vaccination [79].

Close to 4 billion people are estimated to be at risk for dengue virus infection worldwide, mostly in developing countries and in resource-poor settings. Flingai et al. demonstrated the first generation of DNA-encoded monoclonal antibody

engineering and delivery of anti-dengue virus antibodies. The anti-dengue virus DNA-encoded monoclonal antibody achieved peak expression levels of around 1 μg/mL for at least 19 weeks in mice and demonstrated neutralization of dengue virus serotypes 1 - 3. The anti-dengue virus LALA-DNA encoded monoclonal antibody demonstrated protection against primary dengue infection [80]. Advancing monoclonal antibody delivery could have an important impact on lowering overall influenza disease burden during seasonal influenza epidemics and against potential pandemic viruses. Yamazaki et al. were one of the first to show in vivo expression of a pDNA-monoclonal antibody anti-influenza A virus hemagglutinin (HA) mouse IgG1 antibody. They demonstrated initial expression of approximately 1 μg/mL anti-HA antibody in mice. The pDNA-monoclonal antibody expressed anti-HA antibody was detectable in lung bronchioalveolar lavage and protected mice against lethal A/H1N1/Puerto Rico/8 virus challenge [81].

The 2013-2016 Ebolavirus disease (EVD) epidemic was the largest to date, with over 28,000 confirmed cases and over 11,000 deaths. Antibodies against the Zaire ebolavirus GP are demonstrating promising therapeutic protection against EVD. The three-monoclonal antibody ZMapp cocktail was the first antibody combination to demonstrate promise in an EVD outbreak setting. Andrews et al. encoded the ZMapp cocktail antibodies into their pDNA-mAb platform, showing delivery of each individual mouse IgG2a antibody to a maximum of 10 μg/mL and 30 μg/mL for combined delivery and 13 weeks of in vivo expression. The pDNA-ZMapp was delivered to mice, followed by challenge 28 days later with mouse-adapted Ebolavirus, demonstrating protection against lethal infection [82].

There are approximately 37 million people living with HIV, with the envelope GP (Env) being the primary target for immune responses. Due to the high diversity of Env proteins, vaccine development and therapy have been challenging. Wise et al. engineered a panel of 16 broadly neutralizing HIV-1 antibodies as DNA-encoded monoclonal antibodies and delivered them in mice and non-human primates.

The term “decoy” refers to small DNA or RNA molecules (30 - 100 nt in length) containing the same binding site of a DNA or RNA binding protein. They are designed according to the nucleic acid consensus sequence recognized by a particular protein, therefore they possess a structure which mimics or is identical to the real ligand, thus acting as ligand antagonists. As a consequence, these molecules are able to compete, once present in high concentration inside the cells, with the endogenous nucleic acids for binding to this protein, thus diverting it from its natural target. Decoy oligonucleotides usually mimic the consensus binding site of a transcription factor. Gene expression controlled by this transcription factor is effectively prevented, thereby effectively silencing gene expression and preventing the protein from being produced (Figure 15) [50]. Therefore, being less specific in comparison with the small interfering RNA or antisense oligonucleotide method, the decoy oligonucleotide technique can be considered as a gene silencing approach. Nevertheless, decoys have been shown to be highly specific for their target transcription factor [83].

An applicative example of a decoy is found in gene therapy for infectious disorders. Replication of HIV-1 is highly dependent on binding of two of its proteins, the Tat trans activator and the regulatory Rev protein, to specific cis-acting

sequences on viral RNA, the TAR (trans-acting response) and RRE (Rev-responsive element) respectively. In the absence of Tat, transcription of proviral DNA does not occur, while the absence of Rev blocks export of the viral unspliced mRNAs into the cytoplasm. The intracellular expression of high amounts of short RNA sequences corresponding to either TAR (TAR decoy) or RRE (RRE decoy) diverts the respective binding proteins from their natural targets and thus inhibits viral replication [84] [85]. One of these decoy sequences (multimeric TAR) is present in one of the lentiviral vectors that have recently entered a clinical trial [86].

The success of the decoy-based therapeutic approaches is usually limited by the high concentration at which these molecules must be continuously present in the cells to act in a competitive manner. Such a concentration is usually difficult to obtain by extracellular administration. At least for the RNA decoys, this problem can be overcome by the intracellular expression of these molecules upon delivery of the respective genes. In addition, the DNA decoys are usually less efficient than the natural promoter DNA in binding a transcription factor, since, in the latter case, factor binding is stabilized by chromatin structure and interaction with the other proteins involved in transcription [87].