New model integrates innate responses, PML‐NB formation, epigenetic control and reactivation from latency

The dynamic nature of interactions between invading viral pathogens and their hosts has fascinated scientists for several decades. The well‐known capacity of herpes simplex virus (HSV) to establish life‐long infections in humans reflects a dynamic balance between maintaining a latent state in which viral genomes are silenced and re‐entry into the lytic phase during reactivation. Silencing of the viral genome has been shown to be a function of innate immune signalling, intrinsic cellular antiviral mechanisms and epigenetic repression. Thus, although many important observations have been made identifying cellular processes that contribute to the repression of the viral genome and latency, the field has lacked an understanding of how these factors work together. In this issue of EMBO Reports, Suzich et al (2021) present convincing evidence that brings together individual observations into a cohesive model that explains many of these outstanding mysteries. Here, we will review the background data that lead to this outstanding piece of work.

Life‐long HSV infections are characterized by a dynamic balance of latency and lytic reactivation. A study in this issue shows that viral genomes possess a memory of the IFN response during primary infection that restricts their ability to reactivate.

The delicate balance between latent and lytic infection is influenced by an intricate cellular network of overlapping genes and pathways. Silencing of the viral genome has been shown to be a function of innate immune signalling, intrinsic cellular antiviral mechanisms and epigenetic repression; however, the mechanisms by which these pathways work together to repress viral gene expression and control reactivation from the latent state are poorly understood. It was recognized in the late 1980s that interferon treatment could inhibit the expression of HSV immediate‐early proteins during lytic infection of mouse and human cells (Mittnacht et al, ¹⁹⁸⁸). In addition to innate immune responses, intrinsic cellular proteins have also been shown to be antiviral and exert potent effects on gene expression and reactivation from latency. In pioneering work by Gerd Maul, one of the first recognized intrinsic cellular restriction proteins was identified as the promyelocytic leukaemia protein (PML), an interferon‐inducible protein found in nuclear domains called PML‐NBs (Ascoli & Maul, 1991). PML‐NBs were shown to be antiviral and to associate with viral genomes. Subsequent work from several laboratories led to the discovery that the association with PML and other PML‐NB components with viral genomes is responsible for repression of immediate‐early viral gene expression (reviewed in Cuchet‐Lourenco et al, ²⁰¹¹). Repression was subsequently shown to be at least in part due to the entrapment of viral genomes within PML‐NBs during lytic infection in non‐neuronal cells and in latently infected trigeminal ganglia neurons (Catez et al, ²⁰¹²; Alandijany et al, ²⁰¹⁸). Cellular chromatin is the predominant means by which cellular transcription is controlled. The latent HSV genome is bound by nucleosomes, and both H3K9me3 and H3K27me3 post‐translational marks are found on latent genomes, with H3K27 predominating (reviewed in Watson et al, ²⁰¹³). These studies indicate that overlapping cellular processes are able to exert control over viral gene expression and reactivation from latent infections; however, the relationship between PML‐NB formation, innate immune signalling, epigenetic repression of viral gene expression and reactivation from the latent state has remained elusive.

Suzich et al (2021) now elegantly demonstrate that the latent HSV genome can exist in multiple states with respect to reactivation that are a function of the presence of PML‐NBs. PML‐NBs have been shown to result in the deposition of histone H3.3 modified on lysine 9 by trimethylation (H3K9me3), and in the absence of PML, there is a shift versus H3K27me3 (Delbarre et al, ²⁰¹⁷). H3K9me3 is referred to as constitutive heterochromatin, which is more restrictive to transcription than facultative heterochromatin bearing H3K27me3 marks. Facultative heterochromatin represents a more reversible state of repression. As summarized above, both marks have been found on latent viral genomes. In systems where PML‐NBs are prominent, repressed viral genomes have been found to contain H3K9me3 marks (Cohen et al, ²⁰¹⁸). Interestingly, however, Suzich et al (2021) make the observation that PML‐NBs are absent from primary sympathetic and sensory neurons and that the addition of type I interferon (IFN) induces the formation of PML‐NBs. Viral genomes localize to the induced PML‐NBs and establish a latent state that is more restrictive for reactivation. It has previously been reported that quiescent genomes exist in multiple states with respect to activation of viral genes and that there is a relatively large proportion of genomes in primary mouse trigeminal neurons that is not stringently repressed (Terry‐Allison et al, ²⁰⁰⁷). Suzich and co‐workers now show that HSV latency represents multiple states with respect to repression and the ability to reactivate, and they provide a model to explain this phenomenon (Fig 1) (Suzich et al, ²⁰²¹). They suggest a scenario in which infection of a naïve sensory neuron can establish latency that is more reactivatible, possibly maintained in the latent state by the polycomb repressor complexes PRC1/PRC2 as previously described (reviewed in Watson et al, ²⁰¹³). On the contrary, the presence of IFN at the time of infection results in the formation of PML‐NBs, thereby establishing a state in which HSV genomes are entapped, possibly acquiring more H3K9me3 marks (Cohen et al, ²⁰¹⁸).

Figure 1. Multiple types of HSV latency programmes are maintained epigenetically depending on the state of the cells at the time of infection.

(A) Top: When neurons are infected in the presence of interferon, the latent viral chromatin is enriched for H3K9m3 (constitutive heterochromatin). Under these conditions, the viral genome is encapsulated within PML‐NBs, which leads to deep repression of viral gene expression. According to the model presented in this paper (Suzich et al, ²⁰²¹), viral genomes possess a memory of the IFN response during de novo infection, resulting in differential subnuclear positioning of the genomes and their entrapment within PML‐NBs. Viral genomes in this state are severely restricted for reactivation. Bottom: According to this model, if IFN is not present at the time of infection, a different type of latency is established in which viral gene expression is controlled by PRC1/2. Under these conditions, the latent chromatin is enriched for H3K27m3 resulting in “facultative heterochromatin”. This latent state is more easily reversed leading to more frequent derepression and reactivation. (B) High‐resolution Airy scan‐based 3D confocal microscopy of INFa‐treated neurons revealed that viral DNA foci were entrapped or encapsulated within PML‐NBs, consistent with a previous report of viral genomes in latently infected TG in vivo (Catez et al, ²⁰¹²).

The data by Suzich et al (2021) raises the possibility that infected non‐neuronal cells or professional immune cells provide the source of IFN in the course of an in vivo infection that render uninfected neurons more refractory to productive infection, which results in a more restrictive form of latency. Remarkably, the PML‐NBs that are induced by IFN remain after IFN signalling has waned, providing a memory of reactivation restriction. Additionally, removing PML after the induction of latency resulted in a more reactivatable state (Suzich et al, ²⁰²¹).

In summary, this paper nicely integrates many observations made in this field over the last several decades and supports a model for how the delicate balance between latent infection and re‐entry into the lytic phase may be regulated. The notion that primary neurons have a memory of prior IFN exposure characterized by the persistence of PML‐NBs goes a long way towards explaining the roles of IFN‐ and PML‐associated proteins in the epigenetic control of the latent/lytic equilibrium.

EMBO reports (2021) 22: e53496.