The Latency Associated Transcripts

1.3.2.1Description of the LATs and their expression during latency

The LATs are the only viral transcripts abundantly transcribed during latency and have been the focus of a hefty amount of research. They comprise a series of colinear, predominantly nuclear transcripts. The minor LAT is transcribed antisense to the gene encoding ICP0 and extends to a polyadenylation signal in the short repeat region, making it 8.3 kb long (Zwaagstra et al., 1990). The major LAT is a highly abundant non- polyadenylated species of 2.0 kb and is derived by a splicing event from the less abundant minor LAT. Further splicing of the 2.0 kb major LAT RNA occurs within neurons to

produce the 1.5 kb LAT, which accumulates as a stable lariat and is thought to be important for the establishment of latency (Farrell et al., 1991; Rock et al., 1987; Spivack and Fraser, 1987; Wagner et al., 1988b; Zabolotny et al., 1997). While the LATs can bind to polyribosomes, this probably reflects a structural or regulatory role for the LATs in the ribosomal complex as the major LAT lacks polyadenylation (Ahmed and Fraser, 2001; Goldenberg et al., 1997; Wagner et al., 1988a). The overwhelming consensus is that the LATs do not encode a functional protein, despite some reports to the contrary (Doerig et al., 1991; Drolet et al., 1998; Henderson et al., 2009; Jaber et al., 2009; Lagunoff and Roizman, 1994; Naito et al., 2005; Thomas et al., 1999). One complicating factor in determining the exact role of the LATs during latency is the presence of other ORFs that overlap the LAT region, making the construction of deletion mutants problematic. Proteins that are coded for in the same region as the LATs include ICP34.5, ICP4, and ICP0, as well as other less studied ORFs (Bolovan et al., 1994; Jaber et al., 2009; Lagunoff and Roizman, 1994; Perng et al., 1996a; Perng et al., 1995; Wagner et al., 1988a).

The LATs are first detectable in ganglia during the lytic phase of HSV-1 infection, typically by about 48 to 72 hours p.i. in ocularly infected mice (Kramer et al., 1998). However, much higher transcript levels are detected in ganglia during latency (Margolis et al., 1992). The LATs have been detected in latently infected humans, as well as experimentally infected animal models including the guinea pig, rabbit and mouse (Deatly et al., 1987; Krause et al., 1988; Lyn Burke et al., 1991; Rock et al., 1987; Spivack and Fraser, 1987; Stevens et al., 1988; Stevens et al., 1987; Wang et al., 2005a).

1.3.2.2The importance of the LATs and their role in the establishment and

reactivation from latency

Despite decades of intensive research on the role of the LATs, they are not a critical part of the HSV-1 cycle of latency and reactivation, as the region of the genome encoding the LATs is not absolutely required for either the establishment, maintenance of or reactivation from latency (Fareed and Spivack, 1994; Hill et al., 1990; Izumi et al., 1989; Javier et al., 1988; Leib et al., 1989; Perng et al., 1994; Sedarati et al., 1989; Steiner et al., 1989). Early studies using techniques such as laser capture microdissection (LCM) followed by quantitative reverse-transcriptase PCR (qRT-PCR), in situ PCR and direct analyses of individual neurons by CXA found that LATs are transcribed in a fraction of neurons harbouring latent HSV-1 genomes ranging from about five to 30% of latently infected cells (Chen et al., 2002b; Ellison et al., 2000; Maggioncalda et al., 1996; Mehta et al., 1995; Sawtell, 1997; Wang et al., 2005a). However, this has been confounded by the limited sensitivity of these detection methods, the small population of infected neurons, and the

variable expression of the LATs. The variable expression of the LATs proved difficult to account for, as the use of the HSV-2 guinea pig reactivation model revealed that the frequency of spontaneous reactivation is not correlated with the level of LAT production (Bourne et al., 1994). Most recently it has been found that LATs are probably expressed in all latently infected neurons at some point during latency, but the consequences of this transient expression are unknown (Ma et al., 2014; Proença et al., 2008).

Determining the role of the LATs has been problematic, as it is difficult to dissect out the impact of expression during the lytic stage of infection and establishment of latency from the maintenance of and reactivation from latency. A further layer of complexity is added by the different animal models of HSV-1, particularly differences in the rabbit and mouse models of HSV-1 (Perng et al., 2001; as reviewed by Wagner and Bloom, 1997).

The majority of studies using mouse models of HSV-1 infection where the expression of LAT is abrogated conclude these viruses reactivate much less efficiently following explant reactivation or other methods of in vivo reactivation (Devi-Rao et al., 1994; Leib et al., 1989; Sawtell and Thompson, 1992a; Steiner et al., 1989). However, reactivation by viruses that fail to express LAT is influenced by the route of infection, the strain of HSV-1 used or the site of latency establishment (Izumi et al., 1989; Nicoll et al., 2012; Perng et al., 2001). A LAT null virus was deficient for reactivation when latency was established in the in the TG but not in the lumbosacral ganglia (Sawtell and Thompson, 1992a). Similarly, a mutant virus lacking the TATA box and promoter function of LAT on the strain 17syn+ background was deficient for explant reactivation, but a comparable virus constructed on the less virulent KOS background was not (Devi-Rao et al., 1994; Thompson et al., 1986). In the ocular rabbit model of HSV-1 infection, viruses that lack expression of the LATs establish latency to similar levels but show a reduced frequency of spontaneous reactivation. They also fail to reactivate as efficiently following induced reactivation, in either the iontophoresis-epinephrine or other models of reactivation (Bloom et al., 1994; Hill et al., 1996b; Hill et al., 1990; Perng et al., 1994; Trousdale et al., 1991). Only the first 1.5 kb of the 8.3 kb minor LAT is required for a normal reactivation phenotype (Bloom et al., 1996; Perng et al., 1996b). A smaller 348 bp deletion was shown to be associated with a decrease in spontaneous reactivation frequency (Bloom et al., 1996). However, a similar, though not entirely overlapping, 371 bp deletion in LAT constructed using the McKrae strain had no impact on reactivation in the rabbit ocular model, when the same mutant was constructed using strain 17syn+, both spontaneous and induced reactivation were reduced relative to wildtype virus (Hill et al., 1996b; Loutsch et al., 1999; Perng et al., 1996c). Further, this same region is not crucial for the recovery of virus by explant induced reactivation of latently infected mice (Bloom et al., 1996; Maggioncalda et al.,

1994). Given the lack of consistency of the behaviour of LAT mutant viruses in mouse and rabbit models, it is unclear whether any of these findings would be relevant when considering human infection.

The observed reactivation phenotype is based on the assumption that latency is established at equivalent levels in the absence of LAT expression. Most studies show no difference in viral replication during the acute infection or the maintenance of latency, as manifest by the stability of the viral genome over time. Unfortunately, this was often measured by relatively insensitive methods like slot blot hybridisation on whole ganglia (Bloom et al., 1994; Hill et al., 1990). Thompson and Sawtell (1997) found a 75% reduction in the number of cells in which latency is established in mice infected with a mutant lacking either the basal LAT promoter or 5’ end of the LAT gene relative to wildtype virus as indicated by the presence of viral DNA detected by CXA. Mice infected with this LAT mutant were impaired for reactivation following hyperthermia. Results from other murine and rabbit models have confirmed that LAT expression seems to dictate the number of neurons in which latency is established, with mutant viruses with reduced LAT expression exhibited a reduce ability to reactive (Devi-Rao et al., 1994; Maggioncalda et al., 1996; Perng et al., 2000a; Sawtell and Thompson, 1992a). However, altering the inoculum dose of the poorly reactivating LAT deletion virus 17ΔPst in rabbit showed that the poor reactivation phenotype cannot be solely accounted for by a failure to establish wildtype levels of latency as evidenced by low viral genome copy number (O'Neil et al., 2004). It is generally accepted that the abrogation of LAT expression has little impact on the maintenance of latency. However, given the broad viral genome copy number distribution across latently infected cells, qPCR will only reveal substantial differences in the size of the latent reservoir. By using a virus that fails to express LAT due to a deletion in the promoter region in a model that allows for historical marking of all neurons latently infected with HSV-1, it was revealed that the latent reservoir was more unstable in the absence of LAT expression. This was coupled with a slight decrease in the efficiency of latency establishment (Nicoll et al., 2012). Therefore, the LATs may still have an influence on the maintenance of latency.

1.3.2.3Inhibition of lytic viral gene transcription by the LATs

Since the discovery of the LATs, it has been posited that they serve to inhibit lytic viral gene expression, enhancing the stability of latency (Sawtell and Thompson, 1992a). This was first demonstrated by performing in situ hybridisation (ISH) on TG taken from acutely infected mice to show that there is an earlier increase of ICP4, VP16 and glycoprotein H transcripts in the absence of LAT (Garber et al., 1997). Similarly, the detection of rare lytic

transcripts during latency showed that ICP0 transcripts were differentially expressed compared to LAT (Maillet et al., 2006). Further, cultured neuroblastoma cells transformed to express the 2 kb LAT had reduced permissiveness to HSV-1 infection and a reduction in the levels of all immediate early mRNAs, including ICP0 (Farrell et al., 1991; Mador et al., 1998). Further, Chen and colleagues found greater accumulation of ICP4 and TK transcripts during latency following infection with a LAT null virus compared to wildtype virus, suggesting a potential role for LAT in silencing viral gene expression (Chen et al., 1997). By contrast, an opposing phenotype was observed in rabbits ocularly infected with the LAT deletion virus, 17ΔPst, with a significant decrease in accumulation of ICP4, TK or glycoprotein C (gC) transcripts during latency compared to wildtype virus (Giordani et al., 2008). However, despite the palpable differences in the rabbit and mouse models of HSV-1 latency, LAT clearly plays a role in modulating lytic viral gene expression during latency. This was originally thought to be mediated by antisense inhibition of ICP0 or ICP4 by LAT, leading to increased virus shutdown and establishment of latency (Rock et al., 1987; Stevens et al., 1987). However, this is not the case (Burton et al., 2003a; Chen et al., 2002a; Shen et al., 2009; Steiner et al., 1989). This is reinforced by the observation that sequences that are responsible for the spontaneous reactivation of HSV-1 McKrae in the rabbit ocular model do not overlap ICP0 (Perng et al., 1996b). Further, adding the first 1.5 kb of the primary LAT transcript into the dLAT2903 virus at an ectopic locus was able to restore the reactivation phenotype to this virus, despite the absence of expression of the remainder of LAT (Drolet et al., 1999; Perng et al., 1996b).

As an alternative means of regulating viral gene expression, LAT serves as a microRNA (miRNA) precursor (described in greater detail in Section 1.3.3; Umbach et al., 2008). Additionally, two small RNAs that are 62 and 36 nucleotides long have also been identified that are expressed in mice, named LAT sRNA1 and LAT sRNA2 respectively (Peng et al., 2008; Shen et al., 2009). Following cotransfection of sRNA1 or sRNA2 with HSV-1 genomic DNA into Neuro2A cells, they can inhibit cold shock induced apoptosis and the production of infectious virus. They may do this inducing IFN-β promoter activity in the presence of the receptor retinoic acid-inducible gene 1, but they also act by inducing herpes virus entry mediator expression in latently infected mice (Allen et al., 2014; da Silva and Jones, 2013; Peng et al., 2008; Shen et al., 2009).

1.3.2.4Promotion of cell survival by the anti-apoptotic activity of LAT

LAT has an anti-apoptotic activity that results in increased neuronal survival, increasing the establishment of latency (Perng et al., 2000b; Thompson and Sawtell, 2001). Initial experiments with LAT deletion viruses resulted in increased apoptosis in infected mice

and rabbits during acute infection (Perng et al., 2000b). Subsequent analysis revealed that the region associated with this anti-apoptotic activity in vivo mapped to the first 1.5 kb following the LAT promoter, at the 3’ end of exon 1 and 5’ end of the stable 2kb intron (Ahmed et al., 2002; Branco and Fraser, 2005; Inman et al., 2001; Perng et al., 2000b). Similarly, insertion of other inhibitors of apoptosis into such LAT null viruses is able to restore the ability of these viruses to reactivate in both rabbits and mice, although this was not measured in a highly quantitative way (Jin et al., 2008; Jin et al., 2005; Perng et al., 2002).

Further work, predominantly based on in vitro transfection assays with the 2 kb LAT intron, revealed that LAT can block the extrinsic (caspase 8-dependent) apoptosis pathway, as well as less efficiently blocking the intrinsic (caspase 9-dependent) apoptosis pathway, protecting cells from death (Ahmed et al., 2002; Carpenter et al., 2007; Jin et al., 2003; Peng et al., 2004). It has also been shown that the 2 kb LAT can protect neuronal Neuro2A and C1300 cells against granzyme B (gzmB)-mediated caspase3-induced apoptosis and protect against CD8+ _{T cell killing in vitro (Jiang et al., 2011). Using LAT} deletion viruses in Neuro2A cells, it was revealed that LATs can prevent apoptosis by inducing preferential accumulation of the anti-apoptotic Bcl-XL over the pro-apoptotic Bcl- XS transcripts(Peng et al., 2003). However, the exact mechanism by which LAT mediates protection against apoptosis remains to be fully elucidated.

1.3.3

Role of miRNAs in the regulation of the establishment and

maintenance of latency

Briefly, miRNAs are approximately 22 nucleotide RNAs derived from longer primary transcripts that specifically recognize target mRNAs and inhibit their translation or promote their degradation (as reviewed by Bartel, 2009). Given the relatively small coding capacity of viral genomes, miRNAs represented an attractive means for regulating viral gene expression, particularly during latency. Recently, it was determined by deep- sequencing approaches of animal and cell culture based models of HSV-1 infection, as well as latently infected human samples, that HSV-1 encodes a set of 17 miRNAs (Held et al., 2011; Jurak et al., 2014; Umbach et al., 2008; Umbach et al., 2009). Up to 27 miRNAs have been identified based on bioinformatics-based approaches (Cui et al., 2006; Jurak et al., 2010; Munson and Burch, 2012; Pfeffer et al., 2005). Host miRNAs can also play a role in regulating HSV-1 infection. For example, the host miRNA miR-23a binds to the Interferon Regulatory Factor 1 (IRF1), downregulating signaling through the IRF1-mediated innate antiviral signaling pathway and augmenting viral replication (Ru et al., 2014).

Many of these HSV-1 miRNAs accumulate during lytic infection, with at least one miRNA (miR-H1) being expressed only during this time (Cui et al., 2006; Jurak et al., 2010; Munson and Burch, 2012; Umbach et al., 2009). While dispersed across the genome, there is a concentration of miRNAs encoded within the regions around the origin of replication (Jurak et al., 2010). Most of the virally derived miRNAs are persistently expressed throughout latency, with the majority derived from the LAT precursor or encoded in the LAT region (Jurak et al., 2010; Umbach et al., 2008; Umbach et al., 2009). These miRNAs are not essential but they may have some role in regulating the establishment and maintenance of latency in mice, as shown using LAT deletion viruses (Kramer et al., 2011). For example, miR-H2 is found within the LAT region and is antisense to ICP0, and knocking out miR-H2 leads to increased accumulation of ICP0 protein. This virus reactivates slightly faster following explant cultivation, but there was no effect on the establishment of latency as determined by viral DNA load in the TG (Jurak et al., 2014; Umbach et al., 2008). Downregulating this miRNA in transient assays did not have any effect on other immediate early HSV-1 genes (Umbach et al., 2009). Similarly, miR-H6 has been shown to downregulate ICP4 expression, and given its role in stimulating its expression of the viral lytic genes, it has been hypothesised that it may be involved in the maintenance of latency (Umbach et al., 2008).

So far, determining the biological role of most miRNAs during HSV-1 infection has been difficult (Du et al., 2015). An investigation into which of the miRNAs were loaded onto the RNA-induced silencing complex (RISC), and therefore are likely to be biologically relevant, revealed that only a fraction of some of the most abundant viral-encoded miRNAs, such as miR-H1-5p and miR-H6-3p, are associated with RISC. Additionally, some miRNAs were not bound to the RISC at all, and are not likely to be functionally relevant (Du et al., 2015; Flores et al., 2013). Further, while some miRNAs are transcribed on the opposite strand to known viral RNAs that serve as their target, for most miRNAs the target, likely of host cell origin, remains unidentified (Du et al., 2015; Jiang et al., 2015; Jurak et al., 2010; Munson and Burch, 2012). Finally, there are substantial differences in the accumulation of RNAs in animal-based versus cell culture models of HSV-1 latency, making dissection of the role of miRNAs in regulating HSV-1 latency a challenging task (Du et al., 2015; Jurak et al., 2014).