In our mouse model about 6000 neurons per TG are latently infected as defined by the presence of the viral genome (Sawtell 1997; Thompson and Sawtell 1997; Sawtell, Poon et al. 1998; Thompson and Sawtell 2001). Spontaneous reactivation occurs in one neuron out of 20 TG or about 1/120,000 latently infected neurons per day (Sawtell 1998). Stressful stimuli including hyperthermic stress (HS) increase the reactivation frequency about 50-fold, to ~1/2500 (Sawtell 1998; Sawtell, Poon et al. 1998; Thompson, Preston et al. 2009). Clearly the maintenance of latent infections is tightly controlled, but little is known about viral functions that maintain the latent state.

The latency associated transcript (LAT) locus has been extensively investigated. and shown to express several unusual forms of non-coding RNAs during latency including a primary transcript, stable introns spliced from this transcript, small RNAs and microRNAs, one of which is upstream of the LAT promoter and encoded on the opposite strand (Rock, Nesburn et al. 1987; Stevens, Wagner et al. 1987; Wechsler, Nesburn et al. 1988; Farrell, Dobson et al. 1991; Peng, Vitvitskaia et al. 2008; Umbach, Kramer et al. 2008). In addition, some evidence of viral proteins expressed during latency (most from the LAT region) has been reported (Green, Courtney et al. 1981; Doerig, Pizer et al. 1991; Thomas, Gough et al. 1999; Perng, Maguen et al. 2002; Thomas, Lilley et al. 2002; Henderson, Jaber et al. 2009; Jaber, Henderson et al. 2009), but the functions of these proteins are not known.

We and others have shown that HSV-1 LAT mutants can establish latent infections and reactivate from them (Sedarati, Izumi et al. 1989; Sawtell and Thompson 1992; Bloom, Devi-Rao et al. 1994; Perng, Dunkel et al. 1994; Thompson and Sawtell 1997; Thompson and Sawtell 2001) . Analysis at the single neuron level reveals that significantly fewer latent infections are established by LAT null mutants demonstrating a role for this locus during the establishment of latency (Sawtell and Thompson 1992; Thompson and Sawtell 1997; Perng, Slanina et al. 2000; Thompson and Sawtell 2001). In the mouse model this is likely to be due to a function which serves to suppress lytic viral protein production in TG neurons (Sawtell and Thompson 1992; Garber, Schaffer et al. 1997; Thompson and Sawtell 1997). In the absence of the LAT locus significantly more neurons expressing the LAT promoter enter into lytic infection and fewer neurons survive (Sawtell and Thompson 1992; Thompson and Sawtell 1997; Thompson and Sawtell 2001).

The LAT locus is expressed in a subset of neurons during latency (Deatly, Spivack et al. 1987; Rock, Nesburn et al. 1987; Stevens, Wagner et al. 1987; Arthur, Efstathiou et al. 1993). We hypothesize that regulators encoded in the LAT locus provide function(s) that repress entry into the lytic cycle and this in turn serves to moderate the frequency of reactivation and maintains the pool of latently infected neurons through time (Sawtell and Thompson 1992; Garber, Schaffer et al. 1997; Thompson and Sawtell 1997; Thompson and Sawtell 2001; Kang, Mukerjee et al. 2003). For example, there is a post transcriptional constraint on the production of ICP0 mediated by the LAT locus during in vivo reactivation (Thompson, Shieh et al. 2003). In this way the frequency of reactivation and or the number of neurons exiting latency per reactivation event could be limited. This would be important to minimize loss from the pool of latently infected neurons which, in turn, would increase the probability of transmission to new hosts throughout the life of the infected individual. If this hypothesis is true, then reactivation competent latent infections established by LAT null viruses should be more prone to reactivate than those established by viruses that have an intact LAT locus.

Groups of outbred 5-6 week old Swiss Webster male mice were infected with 1×10⁵ pfu of the mutant 17AH and 3×10⁴ pfu of the genomically restored isolate 17AHR. The amount of 17AHR employed to infect the mice was adjusted so that equivalent levels of latency were obtained in both groups as previously detailed (Sawtell, Thompson et al. 2006; Thompson and Sawtell 2006; Thompson, Preston et al. 2009). The mice were maintained for 30 days post inoculation at which time the 17AH and 17AHR infected mice were each divided into two groups. One half of the latently infected mice were maintained an additional 294 days. Additional mice from these same groups were subjected to the hyperthemic stress procedure an average of 2.5 times per week during 280 days after the initial 30 day period. After week 40 the mice receiving the multiple HS treatments were rested for two weeks, and then all of the mice were subjected to a final HS procedure to induce reactivation in vivo. Two measures were employed to quantify the outcome in TG at 22 hrs post HS, (i) the quantification of infectious virus and (ii) quantification of the number of neurons exiting latency using whole ganglion immunohistochemistry (WGIHC) for viral proteins. With these methods we can reliably detect a two fold difference in the number of neurons that enter the lytic cycle or the number of TG positive and the number of pfu they contain with 95% confidence (e.g. p≤0.05 and a power of ≤0.80 as detailed previously (Thompson and Sawtell 1997; Sawtell 2003; Sawtell 2005; Sawtell, Thompson et al. 2006; Thompson, Preston et al. 2009). The results are shown in fig 2.

In mice receiving only a single HS at 42 weeks pi, neither the frequency (% TG reactivating) nor the number of neurons exiting latency was different between mice latently infected with 17AH or 17AHR (60% and 66% of TG positive, average=3.6 and 3.7 neurons reactivating per positive TG, p=1.0) (fig.2). In the 17AHR infected mice that received either one or multiple HS treatments, no difference in either the number of neurons that exited latency, the amount of virus detected or the frequency of detection of virus was observed (p=0.9 and 0.5 respectively, Mann-Whitney test). The results with 17AHR were not different than those obtained in prior experiments with the parental strain 17Syn+ in which 68% (15/22) of latently infected mice receiving a single HS treatment were positive for reactivated virus in TG vs. 66% (14/21) of those receiving multiple HS treatments through five months were positive and the number of pfu detected in TG was also not different (p≥0.7). However, there were striking decreases in these measures in TG from mice latently infected with the LAT null virus after multiple HS treatments compared to a single treatment (fig. 2; number of neurons positive p=0.025, and number of pfu detected p=0.0007, Mann-Whitney test). Interestingly, these results indicate that the level of spontaneous reactivation in these mice was not sufficient to reduce reactivation frequency in either LAT+ or LAT null infected animals. Previous studies predict that approximately 30 single neuron reactivation events would occur spontaneously per mouse latently infected with wild type virus during this period (Sawtell 2003). The multiple HS applied through 40 weeks would be expected to induce about 300 reactivation events (including multiple neurons) with wild type virus per mouse (Sawtell 2003; Sawtell and Thompson 2004; Thompson and Sawtell 2006; Thompson, Preston et al. 2009). As noted above, the mouse infections were adjusted to generate equivalent latent infections with 17AH and 17AHR. In addition in the single HS treated groups both the LAT null and genomic rescue infected mice had the same reactivation frequency (as measured by infectious virus production), and equivalent frequency of entry into lytic protein production (measured by WGIHC). Because these frequencies are directly and highly correlated with the number of latent infections established (Sawtell 1998) there was equivalent latency was present in these groups at 42 weeks pi. Therefore these data clearly demonstrate that the LAT locus provides one or more functions that serve to maintain the frequency of reactivation following stress in the TG over time. Further, this reduced frequency of reactivation (infectious virus production) is the result of fewer neurons exiting latency (viral protein positive) which rules out the possibility that a downstream block in the replication cycle was underlying the reduced frequency of reactivation.

The above experiment revealed a previously unknown role for the LAT locus in the long term maintenance of a stable reactivation frequency. We next sought to confirm this effect of the LAT locus using an alternate strategy to reduce the time required for the experiment. We previously showed statistically significant HS induced reactivation from latency as early as 9 days pi, but the results were complicated by the presence of viral protein positive neurons in non-induced TG (Sawtell 2003). Several characteristics of in vivo infection with viruses lacking a functional thymidine kinase suggested that this background could be a suitable system to examine the role of the LAT locus early after infection. First, in the absence of TK function, HSV does not efficiently replicate in the TG, and there is no evidence of cell to cell spread of virus within the TG, consistent with the absence or extremely low levels of infectious virus recovered (Field and Wildy 1978; Tenser, Miller et al. 1979; Jacobson, Ruffner et al. 1993; Sawtell, Thompson et al. 2006). Despite this, latent infections are established and exit from latency (expression of viral proteins) can be detected following HS (Sawtell, Thompson et al. 2006; Thompson, Preston et al. 2009).

LAT null + thymidine kinase (TK) null double mutants were generated to test in the mouse model. Briefly, the TK knock out cassette we previously used to make the virus tBTK− (Pyles and Thompson 1994) was inserted into our prototypical LAT null mutant 17AH (Thompson and Sawtell 1997). In this mutation cassette a copy of the Escherichia coli LacZ gene driven by the SV40 promoter is inserted at 47563 bp in the viral genome completely eliminating TK activity while not perturbing other viral genes including UL24 (Pyles and Thompson 1994). Three independently derived dual mutants were purified to homogeneity (isolates 4-1-23, 5-2-1, and 6-9-1) as indicated by genomic structures, PCR analysis, resistance to acyclovir, and “blue plaque” assays in which ≥ 1,000 plaques were viewed for beta galactosidase (β-gal) activity (not shown).

Groups of mice were infected on scarified corneas with 1×10⁶ pfu of either 17-AHTK− or 17LAT+TK− as detailed (Thompson and Sawtell 2001; Sawtell, Thompson et al. 2006; Thompson and Sawtell 2006). All three independent isolates displayed similar replication kinetics in vivo (not shown). These viruses achieved similar titers in the eye. However in the TG, titers of infectious virus were very low (about 3 to 4 orders of magnitude less than would be found in wild type infected TG), although levels detected in mice infected with 17AHTK− (LAT null) were greater than those detected in the 17LAT+TK− infected animals (fig 3A; isolate 4-1-23). The severe replication defect in TG is a phenotype common to most true TK null mutants (Horsburgh, Chen et al. 1998). Virus recovered from mouse TG at 4 days pi was resistant to acyclovir and retained β-gal activity. No virus was detected in any of the animals on days 6 and 8 pi (not shown). On day 9 pi the remaining mice were subjected to hyperthermic stress, which induces reactivation of HSV in vivo (Sawtell and Thompson 1992). This experiment was performed with two different 17AHTK− isolates (4-1-23 and 5-2-1) with similar results and shown are the results obtained with isolate 17AHTK− (4-1-23). Whole ganglion immunohistochemistry (WGIHC) was employed to detect neurons positive for viral proteins at 22 hours post HS as previously described (Thompson and Sawtell 1997; Sawtell 2003; Thompson, Preston et al. 2009). In the untreated group, TG neurons expressing viral proteins were not detected (0/7 and 0/10 17AHTK− and 17LAT+TK− respectively) (fig 3B). Following HS, 3/8 TG infected with the LAT+TK− mutant contained a single neuron in each of two TG and two neurons in a single TG positive for viral proteins. In contrast, 12/15 TG were positive (range= 1 to 9 positive neurons) in the LAT null TK null infected mice and this difference was significant (p=0.015 unpaired T test) (fig 3B). Thus using this experimental system, it is possible to detect a repressive effect conferred by the LAT locus on the activation of the lytic cycle in neurons as early as day 9 pi. These findings are consistent with the function of this locus in enhancing the number of latent infections (Sawtell and Thompson 1992; Thompson and Sawtell 1997; Perng, Slanina et al. 2000; Thompson and Sawtell 2001). However, whether this represents a single or multiple mechanisms is not known.