Introduction

The fundamental question posed by virologists for more than seven decades is the mechanism by which herpes simplex viruses (HSV) periodically flare up and cause recrudescent lesions. In the intervening decades, we have learned that at the portal of entry into the body, HSV-1 causes a productive infection in which infected cells are destroyed and the virus multiplies and a latent infection in dorsal root and autonomic ganglia in which the virus is silenced and the neuron survives (Roizman et al. 2007). Thus, the original question doubled: How does a virus that is prone to infect and destroy cells at the portal of entry become silent in neurons and what makes it reactivate?

The two posed questions are not operationally defined and in fact are based on a false premise. The underlying assumption is that HSV finds resistance to replication in the neurons but not at the portal of entry, or to simplify the conceptualization of the problem, that it is sufficient for the viral DNA to present itself in the nucleus for viral replication to ensue in cells at the portal of entry. The hypothesis tested in this laboratory is based on an entirely different premise, namely that the virus encounters resistance to infection in cells at the portal of entry into the body and in the neurons, that it is able to overcome resistance at the portal of entry into the body but not in neurons (Roizman 2011). The corollary of this hypothesis is that, by a stochastic process, the silencing in some neurons ceases or is overcome and the virus replicates. Finally, following the principle of Occam’s razor, the tests of this hypothesis should focus at least initially on the assumption that the same cellular factors that operate at the portal of entry of the virus into the body also operate in neurons in which the viral DNA is silenced. The objective of this report is to provide a brief summary of the tests of this hypothesis.

The checkpoints of viral replication in productively infected cells

The fundamental hypothesis underlying these studies and stated above predicts that cells attempt to silence viral DNA on entry into the nucleus. Several lines of evidence indicate that this is indeed the case. On entry into the cells, HSV DNA is coated by histones and other proteins (Knipe and Cliffe 2008; Kutluay and Triezenberg 2009; Lacasse and Schang 2010). This finding per se does not prove resistance to infection if only a fraction of viral DNA actually initiates replication. The non-replicating DNA most likely ends as heterochromatin. The evidence that the bound proteins actually attempt to repress viral DNA comes from two series of experiments. Thus, nearly 30 years ago, it was shown that a viral protein, VP16 or α-TIF, activates transcription of α genes—the first set of genes expressed after infection (reviewed in Roizman et al. 2007; Roizman 2011). The current version of the events illustrated in Fig. 1 is that the complex VP16, Oct1, and the host transcriptional factor HCF1 binds to response elements in promoters of α genes and recruits LSD1 (lysine-specific demethylase) and the CLOCK histone acetyl transferase to remodel chromatin in order to initiate transcription (Liang et al. 2009; Zhou et al. 2010; Kalamvoki and Roizman 2010, 2011; Metzger et al. 2005). It is noteworthy that LSD1 is unstable in the absence of its partner, CoREST (Shi et al. 2005. Yang et al. 2006; Lee et al. 2005). CoREST in turn interacts with REST. In non-neuronal cells, CoREST/REST proteins form a complex with HDAC1 or HDAC2 and act as repressors of neuronal genes (reviewed in Ballas and Mandel 2005; Tapia-Ramírez et al. 1997; Gopalakrishnan 2009). The key recent evidence on the role of these proteins in initiation of transcription of viral genes rests on studies showing that even a partial depletion of LSD1, CoREST, REST, or CLOCK by corresponding siRNA resulted in a delay in the initiation of transcription of α genes (Zhou et al. 2010). The fundamental conclusion is that the host attempts to repress the viral genome from the moment of entry of the virus into cells and that a viral tegument protein, VP16, carried by the virion into the cell recruits the cellular factors essential to initiate transcription of α genes

Fig. 1
figure 1

A model of the events at the two key checkpoints in viral gene expression during productive infection. a Schematic representation of infected cell at the point of entry of viral DNA and VP16 into the nucleus. b–e Schematic representations of nuclei of infected cells. b HSV DNA is bound by histones and repressors including the HDAC-1/CoREST/REST complex. c Assembly of transcriptional machinery at response elements in promoter of α genes. VP16 recruits Oct1, HCF-1, and LSD1. Included in the complex are CLOCK histone acetyl transferase and additional transcriptional factors that enable the expression of α genes. d Newly made α protein ICP0 binds to CoREST and dislodges HDAC1 or 2 enabling the expression of β and γ genes

Full size image

The second set of experiments also has a long history. It has been known for many years that an α protein designated as the infected cell protein no. 0 (ICP0) is a promiscuous transactivator of genes introduced into cells by transfection or infection. At low multiplicities of infection, α genes are expressed but the transition from α to β (early) or γ genes did not ensue (reviewed in Roizman et al. 2007; Roizman 2011). The block in the transition from α to β and γ gene expression could be overcome at high multiplies of infection. ICP0 differed from ICP4, another regulatory α protein required for expression of viral genes (Roizman et al. 2007). Unlike ICP4, ICP0 does not bind to viral DNA. A clue to the role of ICP0 in the transition from α to β gene expression emerged from the observation that approximately 70 residues of ICP0 located close to the C terminus of the protein were also present at the N terminus of CoREST (Gu et al. 2005; Gu and Roizman 2007, 2009). Subsequent studies showed that CoREST and ICP0 interact that the binding site for CoREST in ICP0 is next to the sequence homologous to the N terminus of CoREST. In CoREST, the binding site for ICP0 is at the N terminus and corresponds to the sequence that is conserved in ICP0 (Gu and Roizman 2007). In addition, two findings solidified a functional role of the interaction between ICP0 and the HDAC1 or 2/CoREST/REST repressor complex. Thus, ICP0 dislodged HDAC1-1 from the CoREST/REST complex. In addition, at a slightly later time in infection, at least a fraction of the dislodged HDAC1, CoREST, and REST were exported to the cytoplasm (Gu et al. 2005). The key experiment designed to determine the role of the interaction was based on the expectation that if the role of ICP0 is to block repression by the HDAC-1 or 2/CoREST/REST complex, then a dominant negative CoREST incapable of binding HDAC1 should compensate for the absence of ICP0 in cells infected with a mutant virus at low multiplicities of infection (Gu and Roizman 2009). This was in fact the case: A mutant virus in which ICP0 was replaced by a dominant negative CoREST replicated to 10- to 100-fold higher titers than ΔICP0 mutant in a cell-type-dependent fashion.

The fundamental conclusions of the two series of studies are that the viral genome is repressed immediately upon entry into the nucleus that expression of all viral genes requires derepression of the genome in two stages: the first one for the expression of α genes and the second for the expression of β and γ genes. It is noteworthy that the same repressor components are involved at both checkpoints. An obvious question is why has the virus evolved a two-step process of derepression of the viral genome when in fact a single step would have sufficed?

Derepression of the HSV genome to enable post-α genes expression is but one of the functions of ICP0. This multifunctional protein localizes early in infection at ND10 nuclear bodies. Here it acts as an E3 ubiquitin ligase to degrade promyelocytic leukemia protein and SP100 (Boutell et al. 2002; Hagglund et al. 2002; Gu and Roizman 2003). A key objective accomplished by degrading ICP0 is to render the infected cell resistant to exogenous interferon (Chee et al. 2003). ICP0 also binds to Bmal1 and helps recruit Bmal1 and its partner, CLOCK, to the ND10 bodies (Kawaguchi et al. 2001; Kalamvoki and Roizman 2010). The various functions of ICP0 appear to be interdependent (Gu and Roizman 2009).

The latent state

There is universal agreement that in experimental animal systems, HSV infects nerve endings at the site of inoculation. The sequelae are less clear. The evidence indicates that the virus replicates at least in some cells that replication ultimately abates and the residual virus detected in ganglia is in latently infected neurons (Roizman et al. 2007). It is convenient to explore the hypothesis illustrated schematically in Fig. 2 that the virus replicates in some neurons that are ultimately destroyed but is silenced in other neurons. A key finding that bears on the silencing of viral DNA in neurons committed to the maintenance of latent state is the observation that VP16 and HCF1 are both retained in the neuronal cytoplasm and do not reach the nucleus (Boissiere et al. 1999; Kristie et al. 1999). Presumably in the absence of VP16 and HCF1 α gene would not be expressed. A central issue is whether the absence of HCF1 and VP16 alone accounts for the silencing of viral DNA in latently infected neurons.

Fig. 2
figure 2

A model of the establishment of latency in neurons. a Viral DNA enters nucleus of the neuron in the absence of VP 16 or HCF-1 or both. b Viral DNA is silenced by histones and repressors. c Viral DNA–host protein complex is remodeled into heterochromatin. LAT and miRNAs carryout surveillance to degrade accumulating low level transcripts of key viral genes

Full size image

Parsimony requires that we consider that the response of neuronal cells in dorsal root ganglia to the entry of HSV DNA into the nuclei is the same as that of cells at the portal of entry. According to this hypothesis in latently infected neurons, HSV fails to initiate transcription at two rather than solely at one checkpoint. Thus, the lack of VP16 and HCF1 would preclude the breaching of the first checkpoint. The absence of α proteins and in particular ICP0 would preclude the expression of β and γ genes. The test of this hypothesis rests on the demonstration that the second checkpoint exists in neurons. The data in support of the hypothesis are a mutant virus consisting of the wild-type genome into which a dominant negative REST genes driven by the SV40 promoter multiplied to higher levels and was more lethal than the wild-type parent virus (Du et al. 2010). In mice infected with this mutant by the corneal route, the trigeminal ganglia were totally destroyed. The results of the experiment suggest that a larger fraction of neurons initially committed to silencing of DNA are induced to replicate the virus through inactivation of checkpoint 2 by the dominant negative REST. The conclusion to be drawn from this study is: HSV does not invariably overcome checkpoints 1 or 2 consistent with the prediction that unchecked replication with increased morbidity and mortality is detrimental to virus spread in nature.

The second hypothesis that silencing of viral DNA in latently infected neurons is due to synthesis of LATs and associated miRNAs is fundamentally untenable as stated. The LAT is a stable intron derived from a very much longer short-lived precursor RNA (Wu et al. 1996). Most of the miRNAs identified to date derive from the domain of LAT or the larger RNA from which it is derived (Cui et al. 2006; Tang et al. 2008; Umbach et al. 2008, 2009). Notwithstanding the loss of neurons associated with infections with ΔLAT viruses, these mutants do establish latent infections albeit not as efficiently as wild-type viruses (Javier et al. 1988; Wang et al. 2001.) A more appropriate role of LATs and miRNAs is surveillance and degradation of low level transcription of viral genes from the silenced DNA. The failure of such surveillance and rapid degradation would result in a low level spontaneous reactivation and destruction of neurons.

It has been reported that in latently infected neurons, HSV DNA is in heterochromatin and that the fraction of remodeled DNA correlates with the expression of LATs (Chen et al. 2002; Maillet et al. 2006; Neumann et al. 2007; Kwiatkowski et al. 2009; Wang et al. 2005). Conceptually, sequestration of HSV DNA in heterochromatin could silence viral gene expression. There are two fundamental issues: Foremost, viral DNA would be expected to interact with appropriate repressors and is silenced in the course of its remodeling into heterochromatin. The key questions are what precipitates and are the key steps in the remodeling of viral DNA. Second, a vexing problem that threatens to remain unresolved for some time is that only a fraction of latent HSV DNA reactivates. Is the state of heterochromicity uniform for all HSV DNA?

As noted in the beginning of this section, the unfolded scenario of events taking place in infected ganglia makes the assumption that both productive and latent infection occur simultaneously and independently of each other. Direct evidence in support of this conclusion is scant. The conclusion is based on the observation that in cell culture expression of α genes invariably leads to cell death even under conditions in which viral replication does not ensue. We cannot exclude the hypothesis that the initial events in infected neurons are similar if not identical and that a stochastic event forces a lytic rather than silent infection.

Reactivation: principles and problems

In humans, clinical reactivations more appropriately called recrudescences of herpetic vesicles follow certain stimuli such as fever, exposure to UV light, and emotional or hormonal stimulation. It is now abundantly clear that reactivations of virus replication are far more frequent than the recrudescence of herpetic vesicles (Roizman and Sears 1987). The prevailing thought is that the immune system precludes the reactivated virus from causing recrudescent lesions. The fundamental question is the mechanism of reactivation. There are two interrelated issues.

The first issue stems from the abundant evidence that to initiate gene expression, VP16/OCT1/HCF1 must bind to promoters of α genes and induce the synthesis of α protein. VP16 is unlikely to be present to initiate replication of latent virus in neurons that have been harboring viral DNA for many decades. Hence, either (a) a host factor substitutes for VP16 to initiate α gene expression, (b) in neurons VP16 is expressed first followed by activation of α genes as suggested by Thompson et al. 2009 or (c) the promoter of the gene encoding ICP0 contains elements that enable it to be expressed at a low but sufficient level in the absence of VP16. Current evidence does not fully discriminate between the possibilities enumerated above.

The second issue is what triggers reactivation regardless of the pathway by which it takes place. An intriguing possibility is that in neurons HCF1 resides in the cytoplasm but is transported to the nucleus following stress (Kristie et al. 1999). This hypothesis would support a role for a newly synthesized VP16, but it would also imply that transcription of the VP16 genes and entry of HCF1 are functionally linked.

Conclusions

Deconstruction of the HSV latent state presents interesting challenges not insuperable by current methodology. The central issue is whether the models used to dissect the latent state embraces all elements of the human model.