**4.3 Anti-CVB3 siRNAs**

RNAi-mediated antiviral strategies can achieve much higher efficiency than ASONs. Thus, recent studies have focused on the design and evaluation of anti-CVB3 siRNAs. This group of small double-stranded RNAs, as a silencer of target gene expression, can virtually inhibit any genes of virus and cell if the site of targeting within the gene is unique. Thus, the target search for anti-CVB3 siRNAs is not only concentrating on CVB3 genome but also extending to the host cellular genes required for viral infection or replication.

#### **4.3.1 Targeting the CVB3 genome**

CVB3 genome harbors many *cis*-acting sequence elements for viral transcription and translation, such as the 5' and 3' UTRs, the IRES and other segments for binding of transcription and translation initiation factors. In addition, the viral genome also encodes many essential enzymes for CVB3 multiplication, such as proteases 2A and 3C as well as the RNA-dependent RNA polymerase 3D. These structures are rationale targets for design of anti-CVB3 siRNAs. This hypothesis has been tested by a number of groups. The earlier selection of the siRNA targets was focused on CVB3 protease 2A. Almost at the same time, two groups independently found that inhibition of 2A protease by specific siRNAs significantly reduced CVB3 replication. The first group by Yuan et al., evaluated five siRNAs targeting the 5'UTR, AUG start codon, VP1, 2A and 3D, respectively and found that siRNA targeting 2A (nts 3543-3561) showed strongest anti-CVB3 activity in HeLa cells, resulting in 92% reduction of viral replication and siRNAs targeting VP1, 3D and the 5'UTR

Nucleic Acid-Based Strategies for the Treatment of Coxsackievirus-Induced Myocarditis 413

deliver the 3D to the replication complex (L. E. Leong et al., 1993). They also indicated that since the function of 3C is required prior to 3D, a down-regulation in 3C would have a detrimental effect on viral transcription as available 3D would not be able to carry out replication of CVB4 replication without the assistance of 3C. The authors' discussion seems to be reasonable; however, according to the order (timing) of action for these enzymes, 2A cleaves the polyprotein prior to 3C's cleavage. For this situation, it may be difficult to explain why the siRNAs targeting 2A did not achieve a more efficacious anti-CVB3 activity than siRNA targeting 3C. Obviously, many issues relating to the mechanisms of action need to be further studied. However, according to the present reports, one point is clear that 2A,

Viral structural protein VP1 was also a selected target for testing anti-CVB3 siRNAs; however, the data from literature often showed a lower effectiveness of the siRNA targeting this structural gene as compared to that targeting other genes (Ahn et al., 2005; J. Y. Kim et al., 2007; Yuan et al., 2005). Due to the absence of a proof-reading activity in 3D, the mutation rate for RNA viruses is as high as 10-3 -10-4 (Cann, 2005). Thus, in recent years, the discovery of the occurrence of escape mutants due to siRNA treatment of HCV, poliovirus and HIV infections (Boden et al., 2003; Gitlin et al., 2005; Wilson& Richardson, 2005) greatly encouraged researchers to search for new approaches to counteract the drug resistance. One direction is the application of multiple distinct siRNAs or a siRNA pool to target more than one target genes of the virus (Merl& Wessely, 2007; Nygardas et al., 2009). The other direction is the identification of conserved *cis-*acting replication elements (CRE) (van Ooij et al., 2006). Theoretically, the 5'- and 3'-UTRs are the ideal target regions for siRNAs as they harbor a number of conserved *cis*-acting elements. However, studies with poliovirus and CVB3 found that siRNA residing in these regions are less efficient than siRNAs targeting on other regions (e.g., the coding region and particularly the non-structural coding region) in inducing antiviral activity (Gitlin& Stone& Andino, 2005; Merl& Wessely, 2007; Saleh et al., 2004; Yuan et al., 2005). This low antiviral potency seems to be due to the highly ordered structure of the UTRs itself, as well as the formation of the protein-RNA complexes in the region, which may block the access of the RISC complexes to its target sequences. To address this issue, Lee and coworkers selected a CRE within the coding region of 2C. Evaluation in HeLa cells demonstrated the downregulation of virus replication and attenuation of cytotoxicity in various strains and clinic isolates. Cells treated with this siRNA were resistant to the emergence of viable escape mutants and showed sustained antiviral ability (Lee et al., 2007). Based on this study, a similar experiment using siRNA targeting CRE of CVA24 2C was conducted and the authors reported the similar observations (Jun et al., 2008). These findings from *in vitro* studies were further solidified by *in vivo* evaluation, in which recombinant lentivirus was employed to express shRNAs targeting the CRE of CVB3 2C. Mice injected intraperitoneally with recombinant lentiviruses had significant reductions in viral titers, viral myocarditis and proinflammatory cytokines as well as improved survival rate, after being challenged with CVB3 (Y. J. Kim et al., 2008). Recently, this CRE was further confirmed for a number of enteroviruses, by using

3C and 3D are three important targets for design anti-CVB3 siRNAs.

a novel program and *in vitro* evaluation (Lee et al., 2009).

Another approach to fight the drug resistance caused by escape mutants is the selection of therapeutic targets within the host cellular genes that are necessary for virus entry or viral

**4.3.2 Targeting host cellular genes** 

showed modest antiviral effects, respectively. By mutational analysis of the mechanism of siRNA action, they found that siRNA functions by targeting the positive strand of the virus and requires a perfect sequence match in the central region of the target, but mismatches were more tolerated near the 3' end than the 5'end of the antisense strand (Yuan et al., 2005). This finding on the targeting of siRNA to positive strand of CVB3 was further supported by a later study using siRNA targeting the CVB3 3D gene (Schubert et al., 2007). The second group that studied siRNA targeting CVB3 2A by Merl and co-workers evaluated antiviral activity of siRNA-2A (nts 3637-3657) *in vitro* and in highly susceptible type I interferon receptor-knockout mice. They found that siRNA-2A led to significant reduction of viral tissue titers, attenuated tissue damage and prolonged survival of mice (Merl et al., 2005). It is very interesting to point out that although the two groups used different targeting sequences within the 2A RNA, they all achieved high efficiency of antiviral effects. However, the later work by Racchi et al., which used these two siRNAs together to transfect HeLa cells and then infect with CVB3 did not potentiate the anti-CVB3 effect compared with an equimolar concentration of either siRNA (Racchi et al., 2009).

CVB3 RNA polymerase 3D is probably the most frequently used target for design of anti-CVB3 siRNAs as it is the only viral enzyme involved in CVB3 RNA replication. To date, at least a half dozen of studies on 3D have been reported. The earlier *in vitro* investigations using either un-modified or LNA-modified siRNAs or plasmid vector-expressed shRNAs all achieved significant reduction of viral replication in CVB3-infected HeLa or Cos-7 cells (Ahn et al., 2005; Schubert et al., 2005; Schubert et al., 2007; Werk et al., 2005; Yuan et al., 2005). The *in vivo* evaluation using mouse models also showed very promising results. One study employing transient transfection for *in vivo* mouse models demonstrated that two of the six candidate siRNAs targeting 3D and VP1, respectively, exerted strong anti-CVB3 effects in viral replication, accompanied by attenuated pancreatic tissue damage (J. Y. Kim et al., 2007). Another *in vivo* study is the intravenous treatment of mice with an adeno-associated virus vector (AAV2.9) expressing a shRNA targeting 3D (Fechner et al., 2008). Intravenous injection of recombinant AAV2.9 significantly attenuated the cardiac dysfunction compared to vector-treated control mice on day 10 after CVB3 infection. Recently, a study by combination of soluble CAR receptor (sCAR-Fc) and siRNA targeting 3D achieved a synergistic effect in antiviral effect in human myocardial fibroblast cell culture (Werk et al., 2009).

Other less frequently used CVB3 target genes are protease 3C, structural protein VP1 and non-structural protein 2C. Like protease 2A, protease 3C also plays an important role in the viral life cycle by processing CVB3 polyproteins to generate mature individual structural and non-structural proteins after initial cleavage by 2A (Chau et al., 2007; L. E. C. Leong, Cornell, C. T., Semler, B. L., , 2002). One study designed three siRNAs targeting genes encoding 3C, 2A and 3D of CVB4. Evaluation by transfection of rhabdomyosarcoma (RD) cells demonstrated that siRNA-3C was the most potent siRNA among these three in inhibition of CVB4 replication. This antiviral activity was followed by siRNAs targeting 3D and 2A (Tan, 2010). The difference in efficiency of these siRNAs was discussed by these authors and they proposed that this may be due to the differences in function of these viral enzymes, which are encoded by these regions: the 3C region encodes a protease 3C which is responsible for majority of the cleavage of the viral polyprotein (L. E. C. Leong, Cornell, C. T., Semler, B. L., , 2002) and 3C as well as its precursor 3CD also plays an important role at the level of viral transcription (Parsley et al., 1999). Protease 3C has been shown to be critical for interaction with the cloverleaf structures found at the 5'UTR of the viral genome to

showed modest antiviral effects, respectively. By mutational analysis of the mechanism of siRNA action, they found that siRNA functions by targeting the positive strand of the virus and requires a perfect sequence match in the central region of the target, but mismatches were more tolerated near the 3' end than the 5'end of the antisense strand (Yuan et al., 2005). This finding on the targeting of siRNA to positive strand of CVB3 was further supported by a later study using siRNA targeting the CVB3 3D gene (Schubert et al., 2007). The second group that studied siRNA targeting CVB3 2A by Merl and co-workers evaluated antiviral activity of siRNA-2A (nts 3637-3657) *in vitro* and in highly susceptible type I interferon receptor-knockout mice. They found that siRNA-2A led to significant reduction of viral tissue titers, attenuated tissue damage and prolonged survival of mice (Merl et al., 2005). It is very interesting to point out that although the two groups used different targeting sequences within the 2A RNA, they all achieved high efficiency of antiviral effects. However, the later work by Racchi et al., which used these two siRNAs together to transfect HeLa cells and then infect with CVB3 did not potentiate the anti-CVB3 effect compared with

CVB3 RNA polymerase 3D is probably the most frequently used target for design of anti-CVB3 siRNAs as it is the only viral enzyme involved in CVB3 RNA replication. To date, at least a half dozen of studies on 3D have been reported. The earlier *in vitro* investigations using either un-modified or LNA-modified siRNAs or plasmid vector-expressed shRNAs all achieved significant reduction of viral replication in CVB3-infected HeLa or Cos-7 cells (Ahn et al., 2005; Schubert et al., 2005; Schubert et al., 2007; Werk et al., 2005; Yuan et al., 2005). The *in vivo* evaluation using mouse models also showed very promising results. One study employing transient transfection for *in vivo* mouse models demonstrated that two of the six candidate siRNAs targeting 3D and VP1, respectively, exerted strong anti-CVB3 effects in viral replication, accompanied by attenuated pancreatic tissue damage (J. Y. Kim et al., 2007). Another *in vivo* study is the intravenous treatment of mice with an adeno-associated virus vector (AAV2.9) expressing a shRNA targeting 3D (Fechner et al., 2008). Intravenous injection of recombinant AAV2.9 significantly attenuated the cardiac dysfunction compared to vector-treated control mice on day 10 after CVB3 infection. Recently, a study by combination of soluble CAR receptor (sCAR-Fc) and siRNA targeting 3D achieved a synergistic effect in antiviral effect in human myocardial fibroblast cell culture (Werk et al.,

Other less frequently used CVB3 target genes are protease 3C, structural protein VP1 and non-structural protein 2C. Like protease 2A, protease 3C also plays an important role in the viral life cycle by processing CVB3 polyproteins to generate mature individual structural and non-structural proteins after initial cleavage by 2A (Chau et al., 2007; L. E. C. Leong, Cornell, C. T., Semler, B. L., , 2002). One study designed three siRNAs targeting genes encoding 3C, 2A and 3D of CVB4. Evaluation by transfection of rhabdomyosarcoma (RD) cells demonstrated that siRNA-3C was the most potent siRNA among these three in inhibition of CVB4 replication. This antiviral activity was followed by siRNAs targeting 3D and 2A (Tan, 2010). The difference in efficiency of these siRNAs was discussed by these authors and they proposed that this may be due to the differences in function of these viral enzymes, which are encoded by these regions: the 3C region encodes a protease 3C which is responsible for majority of the cleavage of the viral polyprotein (L. E. C. Leong, Cornell, C. T., Semler, B. L., , 2002) and 3C as well as its precursor 3CD also plays an important role at the level of viral transcription (Parsley et al., 1999). Protease 3C has been shown to be critical for interaction with the cloverleaf structures found at the 5'UTR of the viral genome to

an equimolar concentration of either siRNA (Racchi et al., 2009).

2009).

deliver the 3D to the replication complex (L. E. Leong et al., 1993). They also indicated that since the function of 3C is required prior to 3D, a down-regulation in 3C would have a detrimental effect on viral transcription as available 3D would not be able to carry out replication of CVB4 replication without the assistance of 3C. The authors' discussion seems to be reasonable; however, according to the order (timing) of action for these enzymes, 2A cleaves the polyprotein prior to 3C's cleavage. For this situation, it may be difficult to explain why the siRNAs targeting 2A did not achieve a more efficacious anti-CVB3 activity than siRNA targeting 3C. Obviously, many issues relating to the mechanisms of action need to be further studied. However, according to the present reports, one point is clear that 2A, 3C and 3D are three important targets for design anti-CVB3 siRNAs.

Viral structural protein VP1 was also a selected target for testing anti-CVB3 siRNAs; however, the data from literature often showed a lower effectiveness of the siRNA targeting this structural gene as compared to that targeting other genes (Ahn et al., 2005; J. Y. Kim et al., 2007; Yuan et al., 2005). Due to the absence of a proof-reading activity in 3D, the mutation rate for RNA viruses is as high as 10-3 -10-4 (Cann, 2005). Thus, in recent years, the discovery of the occurrence of escape mutants due to siRNA treatment of HCV, poliovirus and HIV infections (Boden et al., 2003; Gitlin et al., 2005; Wilson& Richardson, 2005) greatly encouraged researchers to search for new approaches to counteract the drug resistance. One direction is the application of multiple distinct siRNAs or a siRNA pool to target more than one target genes of the virus (Merl& Wessely, 2007; Nygardas et al., 2009). The other direction is the identification of conserved *cis-*acting replication elements (CRE) (van Ooij et al., 2006). Theoretically, the 5'- and 3'-UTRs are the ideal target regions for siRNAs as they harbor a number of conserved *cis*-acting elements. However, studies with poliovirus and CVB3 found that siRNA residing in these regions are less efficient than siRNAs targeting on other regions (e.g., the coding region and particularly the non-structural coding region) in inducing antiviral activity (Gitlin& Stone& Andino, 2005; Merl& Wessely, 2007; Saleh et al., 2004; Yuan et al., 2005). This low antiviral potency seems to be due to the highly ordered structure of the UTRs itself, as well as the formation of the protein-RNA complexes in the region, which may block the access of the RISC complexes to its target sequences. To address this issue, Lee and coworkers selected a CRE within the coding region of 2C. Evaluation in HeLa cells demonstrated the downregulation of virus replication and attenuation of cytotoxicity in various strains and clinic isolates. Cells treated with this siRNA were resistant to the emergence of viable escape mutants and showed sustained antiviral ability (Lee et al., 2007). Based on this study, a similar experiment using siRNA targeting CRE of CVA24 2C was conducted and the authors reported the similar observations (Jun et al., 2008). These findings from *in vitro* studies were further solidified by *in vivo* evaluation, in which recombinant lentivirus was employed to express shRNAs targeting the CRE of CVB3 2C. Mice injected intraperitoneally with recombinant lentiviruses had significant reductions in viral titers, viral myocarditis and proinflammatory cytokines as well as improved survival rate, after being challenged with CVB3 (Y. J. Kim et al., 2008). Recently, this CRE was further confirmed for a number of enteroviruses, by using a novel program and *in vitro* evaluation (Lee et al., 2009).

#### **4.3.2 Targeting host cellular genes**

Another approach to fight the drug resistance caused by escape mutants is the selection of therapeutic targets within the host cellular genes that are necessary for virus entry or viral

Nucleic Acid-Based Strategies for the Treatment of Coxsackievirus-Induced Myocarditis 415

Paroo, Z., 2010). Thus, endogenous miRNAs are important targets for gene therapy and artificial miRNAs (AmiRNA) are useful tools for inhibiting disease-causing gene expression (Z. Liu et al., 2008; Sall et al., 2008). In this regard, although numerous studies have been documented for treatment of cancers, cardiovascular diseases, genetic diseases and other viral infections, only one study on anti-CVB3 from our group has been published so far. Here, I briefly summarize our work on design and evaluation of anti-CVB3 AmiRNAs (Ye et al., 2011). We constructed three short hairpin AmiRNAs (AmiR-1, -2 and -3) targeting the stem-loop of the 3'UTR of CVB3 with mismatches at the middle region of the target. Transfection of HeLa cells showed over-expression of these mature AmiRNAs as determined by real time quantitative RT-PCR. After these AmiRNA-expressing cells were infected with CVB3, the viral titers were reduced ~100 folds in cell cultures treated with AmiR-1 or AmiR-2 but not that treated with AmiR-3, at 24 h post infection. Mutational analysis of the targeting sites of AmiRNAs demonstrated that the central region but not the seed region of AmiRNAs is more tolerant to target mutation. In this study we also performed targeted delivery of the AmiRNAs to host cells through ligand-receptor

NA-based agents are inefficiently taken up by mammalian cells and would therefore benefits from additional vehicles or modifications that facilitate drug delivery. Depending on transient delivery or long-term treatment, the delivery approaches can be divided into non-viral delivery of chemically synthesized agents and viral delivery of drug-expressing cassettes (Table 1). For the non-viral delivery measures, they can be further divided into unspecific and cell-type specific delivery. The former method has been widely used for delivery of many chemically synthesized ASONs, ribozymes and siRNAs as well as plasmids encoding shRNAs through transfection of tissue culture cells, hydrodynamic transduction/transfection of mice or intravenous (IV) injection of mice via tails using cationic lipoplexes or liposomes. The successful examples for inhibition of viral pathogens of myocarditis includes deliveries of i) ASONs targeting CVB3 IRES (Yuan et al., 2006) and both ends of the CVB3 genome (A. Wang et al., 2001), ii) siRNAs targeting CVB3 2A (Merl et al., 2005; Yuan et al., 2005) and 3D (Ahn et al., 2005; Schubert et al., 2005; Schubert et al., 2007), iii) plasmids expressing shRNAs targeting 3D and VP1 (J. Y. Kim et al., 2007) and vi) ribozymes targeting HCV RNA (Gonzalez-Carmona et al., 2006). For the cell-type specific method, proper modification and conjugation of 'naked' raw therapeutic molecules are required to achieve targeted delivery. The different chemical modifications described earlier for ASONs are applicable to all NA-based agents. The conjugation of these antivirals can be achieved by covalent linkage of a ligand to the molecules, which enables the drug internalization via specific interactions between the ligand and its receptor. The ligands can be an antibody, vitamin, short peptide, RNA aptamer, folic acid, etc. The details were reviewed elsewhere (X. Ye & Yang, 2009). By this strategy, we have specifically delivered siRNAs targeting CVB3 2A to HeLa (cancer) cells, a cell line susceptible to CVB3, through interactions between folate and its receptor highly expressed on the surface of all cancer cells but not the normal cells (Zhang et al., 2009). This study was carried out by covalent linkage of the siRNA targeting CVB3 2A to a bacterial phage-29 packaging RNA (pRNA). This small pRNA (~170 nts) can form dimer, trimer and hexamer by base pairing through its

interactions, which will be discussed in next section.

**5. Drug delivery** 

replication. In this regard, the CAR receptor which is shared by CVB3 and adenovirus is an attractive candidate since both CVB3 and adenovirus are considered as the common causal agents of myocarditis. To date, two studies have been reported to silence CAR expression with specific siRNAs. One study reported that transfection of HeLa cells with siRNAs, siCAR2 or siCAR9, almost completely silenced the expression of CAR and that further analysis by viral plaque assay revealed ~60% reduction of CVB3 particle formation (Werk et al., 2005). Another study using cardiac-derived HL-1 cell line and primary neonatal cardiomyocytes (PNCMs) demonstrated that treatment with recombinant adenoviruses expressing shRNAs against CAR resulted in almost completely silencing of CAR expression in both HL-1 cells and PNCMs. CAR knockout resulted in inhibition of CVB3 infections by up to 97% in HL-1 and up to 90% in PNCMs. Adenoviruses were inhibited by only 75% in HL-1, but up to 92% in PNCMs (Fechner et al., 2007).

Another host gene, the tissue inhibitor of matrix metalloproteinase-1 (TIMP-1), has been suggested to be a potential target for siRNA to ameliorate CVB3-induced myocarditis. This suggestion is based on the investigation of Crocker and colleagues on a new role of TIMP-1 in exacerbating CVB-induced myocarditis (Crocker et al., 2007). They found that TIMP-1 expression was induced in the myocardium by CVB3 infection. Surprisingly, TIMP-1 knockout mice exhibited a profound attenuation of myocarditis, with increased survival. The amelioration of disease in TIMP-1 knockout mice was not attributable to either an altered T-cell response to the virus or to reduced viral replication. These data allowed the authors to propose and prove a novel function for TIMP-1: its highly localized up-regulation might arrest the matrix metalloproteinase (MMP)-dependent migration of inflammatory cells at the sites of infection thereby anatomically focusing the adaptive immune response. Finally, the benefits of TIMP-1 blockage in treating CVB myocarditis were confirmed by administration of siRNAs targeting TIMP-1, which diminished CVB3-induced myocarditis. However, this improvement of the disease is not due to the changes of viral titers, as demonstrated by viral plaque assay (Crocker et al., 2007).

Recently, the active investigations on CVB3-induced signal transduction pathways have provided new avenues for the search of therapeutic targets for the treatment of myocarditis. Since CVB3, like other picornaviruses, requires the activation of certain signal pathways for initiating their life cycle, inactivation of some signal molecules in the signal cascade with specific siRNAs would block CVB3 replication. Such kind of studies that have been documented thus far include i) the knockdown of ubiquitin expression by siRNAs to downregulate the ubiquitination and subsequent alteration of protein function and/or degradation (Si et al., 2008); ii) silencing of proteosome activator REGγ to inhibit the REGγmediated degradation of several important intracellular proteins (Gao, 2010), such as cyclindependent kinase inhibitors p21, p16 and p19 and tumor suppressor p53; and iii) knockdown of genes critical important for autophagy formation, these genes include ATG7, Beeclin-1 and VPS34 (J. Wong et al., 2008). Although these target genes mentioned above have been tested *in vitro* using specific siRNAs in signal transduction studies and showed promising outcomes, their potential serving as a therapeutic target for treatment of CVB3 infection needs a further evaluation by pharmacological study in animal models.

#### **4.4 Anti-CVB3 artificial miRNAs**

miRNAs are a group of recently discovered new regulators of gene expression. These endogenous regulators control one third of human gene expression (Bartel, 2009; Q. Liu, Paroo, Z., 2010). Thus, endogenous miRNAs are important targets for gene therapy and artificial miRNAs (AmiRNA) are useful tools for inhibiting disease-causing gene expression (Z. Liu et al., 2008; Sall et al., 2008). In this regard, although numerous studies have been documented for treatment of cancers, cardiovascular diseases, genetic diseases and other viral infections, only one study on anti-CVB3 from our group has been published so far. Here, I briefly summarize our work on design and evaluation of anti-CVB3 AmiRNAs (Ye et al., 2011). We constructed three short hairpin AmiRNAs (AmiR-1, -2 and -3) targeting the stem-loop of the 3'UTR of CVB3 with mismatches at the middle region of the target. Transfection of HeLa cells showed over-expression of these mature AmiRNAs as determined by real time quantitative RT-PCR. After these AmiRNA-expressing cells were infected with CVB3, the viral titers were reduced ~100 folds in cell cultures treated with AmiR-1 or AmiR-2 but not that treated with AmiR-3, at 24 h post infection. Mutational analysis of the targeting sites of AmiRNAs demonstrated that the central region but not the seed region of AmiRNAs is more tolerant to target mutation. In this study we also performed targeted delivery of the AmiRNAs to host cells through ligand-receptor interactions, which will be discussed in next section.
