|The Homepage of Dr. Edward K. Wagner|
Herpes simplex virus Research
Latent Infections by HSV
In a latent infection the viral genome is maintained intact in specific sensory neurons where it is genetically equivalent to that present in a viral particle, but the highly regulated productive cycle cascade of gene expression, so characteristic of herpesvirus infections, does not occur. As a consequence, any transcription during latent infection with most herpesviruses is from a very restricted portion of the viral genome, and this transcription is important in some aspect of the process itself. In a very general sense, then, herpesvirus latency is comparable to the lysogenic phase of infection engendered by bacteriophage lambda, but the parallel stops there.
The molecular and physiological details of the latent phase of infection by specific herpesviruses are quite varied and divergent, and indeed, the only common denominator appears to be latency itself. Latent infection with HSV can be viewed as having three separable phases: establishment, maintenance, and reactivation.
Establishment of Latent Infections
As shown in the illustrated experiment using the rabbit eye model, during the initial, acute infection, virus replicates to high levels at the peripheral site of infection. Infection resolves and virus is cleared--usually within two weeks. During this period, virus travels axonally to the sensory nerve ganglion ennervating the site of infection. There is a period of acute infection in the ganglion; however, this resolves as the acute infection does. A fraction of neurons are left with viral DNA present in episomal form. Thus, there must be a profound restriction of viral gene expression so that the cytopathic results of productive infection do not occur.
The precise mechanism by which HSV establishes latent infections is not
known, but recently a number of important lines of study of the
function of HSV regulatory proteins and a re-examination of the actual
mechanism of viral DNA replication have coalesced to suggest an
intriguing possibility. It has become very clear through the work of
While we have outlined HSV replication as occurring through a circular intermediate in the animation in this site, the actual nature of the replication intermediates are much more complex involving the initiation of replication at all viral origins of replication more or less at the same time, and re-initiation at these origins while replication proceeds, resulting in the formation of a large network or "tangle" of replicated and replicating DNA.
Jackson and DeLuca have used mutants of ICP0 and a method of separating such large networked DNA from genomic sized pieces developed by Gardella and associates some time ago to visualize the earliest stages of DNA structure following infection. They see that when ICP0 is not functional viral DNA does circularize--as is seen in latent infections as observed many years ago by Rock and Fraser. Further, circular DNA does not appear to replicate and is quite stable. Normally this circularization does not happen, but DeLuca postulates a model where the "decision" point whether virus infection proceeds on to replication or latency is a result of whether the viral DNA is driven to form a circle. If it does, viral DNA replication cannot proceed and latency follows naturally. The role of ICP0 in this model is to insure by targeted degradation of DNA repair enzymes in the ND10 structures that circularization does not normally occur. Of course, this model does not explain why ICP0 does not function normally in neurons where latency is being established, but it illuminates possible approaches for study of this problem.
The Maintenance of Latent Virus
During the latent phase, productive cycle genes are generally transcriptionally and functionally quiescent and only the latency associated transcript (LAT) is expressed. The promoter for the LAT contains neuron-specific cis-acting elements, but the a full understanding of why this weak promoter is favored where other stronger ones are quiescent is not yet at hand.
As to maintenance of the latent infection, operationally, viral genomes persisting in latently infected cells must provide a reservoir of potential infectious virus upon reactivation. Latent HSV genomes are harbored within the nucleus of a non-dividing sensory neuron and do not need to replicate, indeed the challenge arises from the need for the virus to reactivate from a transcriptionally quiescent, non-replicating cell.
The maintenance of the HSV genome in latently infected neurons appears to be entirely passive; i.e., it requires no viral gene expression or gene product at all. However, HSV DNA is maintained as a nucleosomal, circular episome in latent infections, and low levels of genome replication might occur or be necessary for the establishment or maintenance of a latent infection from which virus can be efficiently reactivated.
Reactivation and LAT
Successful reactivation of HSV results in the appearance of infectious virus at the site of initial infection in an immune host. In HSV infections it can be shown that the expression of LAT facilitates, but is not absolutely required for reactivation. Only a very limited region of LAT is involved in this facilitation, and levels of viral DNA in latently infected ganglia are the same following infection with LAT expressing or LAT(-) mutants.
The stable LAT intron that accumulates in the nucleus is not required, nor is the expression of any protein from the primary LAT transcript. An experiment demonstrating this fact using a rabbit reactivation model is shown below. Individual recombinant viruses were generated that either contained a deletion of their region of the genome shown in red, or in which potential translation initiation codons in the critical region were modified. Note, only the deletion or substitution of the critical region [348(-)] or the removal of the LAT promoter [LAT(-)] leads to a reduce reactivation frequency.
The process of reactivation from latency is triggered by stress as well as other signals which is thought to transiently lead to increased transcriptional activity in the harboring neuron. The sensory nerve ganglia must survive repeated bouts of reactivation without losing function. A possible scenario might involve the ability of one or several latently infected neurons to replicate only a few viral genomes and generate only a few infectious virions during the initial reactivation event. This might happen with or without the extensive cytopathology associated with normal vegetative viral replication, or with the death of only a very few cells. This process may be augmented by viral genes shown to interfere with apoptosis, such as ICP34.5, which act to prevent neuronal death during reactivation where limited replication occurs.
In the experiment shown below, latently infected rabbits were induced to reactive by epinepherine induction, and trigeminal ganglia removed 16 or so hours later. These were then analyzed by PCR to detect viral transcripts. Unfortunately, rabbits infected with LAT negative viruses that do not reactivate efficiently also show this pattern. This means that the critical event takes place in a background where a number of cells express a few productive cycle transcripts, but do not proceed into reactivation. This complicates analysis of the process a great deal.
Animal Models for Studying HSV Latency
The broad host range of HSV has allowed the use of animal models for the study of viral latency. An ideal animal model would be able to recreate all aspects of the human disease, but, obviously, this is not attainable. Still, in terms of the ability of HSV to establish a localized initial infection followed by neuronal spread and establishment of latency, a number of very useful models exist, and again as noted above, have provided the basic tools for our understanding of the disease in humans.
The most appropriate model for latency must allow virus reactivation, and this reactivation should be similar to that seen in humans. Thus, in an immune competent animal it should occur spontaneously, be inducible by stress, and recovery from recrudescence should be complete. Also, initial infection should be mild enough to ensure that all or most experimental subjects survive infection with no sequelae.
Two animal models, the rabbit and the guinea pig, approximate this ideal situation, although both suffer from limitations, and both involve considerable expense. A third model animal, the mouse (the most reasonable in terms of cost), is being used extensively, but suffers from the lack of efficient in vivo reactivation. These three animal systems have provided the means for generating the vast majority of data now available concerning HSV latency and reactivation, but other models also exist.
Latency and Reactivation in Rabbits
Infection of rabbit eyes leads to a latent infection in which virus can be recovered from the trigeminal nerve ganglia only following explant and co-cultivation with indicator cells. In addition, virus can be sporadically recovered from the eye following periods of latency. Of particular use is the fact that the reactivation can be efficiently induced by the iontophoresis of epinephrine into the eye--a procedure perfected by Hill and colleague at the LSU Eye Center. This model has been very important in establishing the requirement for LAT expression for efficient reactivation.
Latency and Reactivation in Guinea Pigs
Vaginal inoculation of female guinea pigs with HSV-1 or HSV-2 results in obvious primary infection with some mortality. Following recovery, survivors of primary infection periodically display vesicular recrudescence in the vaginal area from which infectious virus and/or viral DNA can be recovered. Although reactivation cannot be reliably induced the fact that HSV-2 spontaneously reactivates with much higher frequency than HSV-1 makes this a very attractive system for comparative analysis of the influence of viral genes on reactivation and HSV-1xHSV-2 recombinant viruses are being investigated with an eye towards attempting to identify features important in this difference. The value of guinea pigs in studying vaccine efficiency and in other aspects of experimental pathogenesis makes this an extremely valuable and promising system. The system is under extensive use by L. Stanberry at the University of Cincinnati Medical School.
Murine Models for Latency and Reactivation.
1. The foot-pad/dorsal root ganglia model.
Direct demonstration of the ability of HSV to establish and maintain a latent infection in neuronal cells was accomplished by J.G. Stevens at UCLA using mouse foot-pad infection which is followed by latent infection of spinal ganglia. This model system is roughly analogous to genital infection of HSV in humans, and has been central to describing many of the parameters of HSV latent infection including the identification of the neuron as the site of latent infection, axonal transport of virus through the sciatic nerve, ability of non-replicating virus to establish latent infections, and the characterization of restricted transcription of the latency specific transcription unit during the latent phase of infection. Further, the model is useful with HSV-2, despite this virus' greater neuropathology.
Following infection of the foot-pad, local pathology is observed with clear evidence of involvement of the central nervous system (CNS), the mice that recover are evidently physiologically normal. When dorsal root ganglia are dissected and cultured (either whole or following disruption) on feeder cells, HSV-induced cytopathology can be detected within 4--12 days. This explant recovery of HSV from such latently infected spinal ganglia has been an extremely useful and relatively inexpensive means of assaying the presence of viral genomes within the tissue in question. It has been termed "reactivation," but in actuality this term should be reserved for the process in animals in which virus can be recovered from peripheral tissue, not from nervous tissue itself.
2. The mouse eye/trigeminal ganglia model.
A second murine model for HSV-1 and HSV-2 latency involves the infection of the cornea which is followed by virus latency in the trigeminal ganglia. As in the foot pad model, latent HSV genomes express LAT (the latency associated transcript) in a portion of those neurons maintaining them, and virus can be recovered by co-cultivation of explanted ganglia. An interesting variation on this method which comes closer to an in vivo method has been developed by Sawtell and Thompson at the University of Cincinnati. Here, latently infected mice are transiently exposed to hyperthermia, and then trigeminal ganglia are excised, sectioned, and assayed for the presence of observable virus by immunohistochemical methods or, if recombinant virus is used in which an expressible marker has been included in the genome using genetic engineering methods, by localization of such reporter gene activity.
Although such a model is not equivalent to recovering virus at the site of initial infection, and implicitly assumes that virus recovery in the nerve ganglion is equivalent to reactivation as assayed at a peripheral site, it does provide a second method for modeling reactivation in the mouse.
How does LAT influence HSV-1 latency and reactivation?
Experimental studies using viral regulatory mutants and cell activation have shown that the need for; aTIF and a0 can be abrogated to some degree by the induction of the early processes of cell division and metabolic stress in non-replicating cells. Thus, latency can be viewed as a dynamic balance where one or a few latently infected cells sporadically enter the early stages of viral gene expression as a response to stress, and this process usually aborts without cell death returning the cell to latent infection. Alternatively any limited virus produced is eliminated by innate and adaptive immunity.
In order for successful reactivation to occur, then, not only must a latently infected cell allow limited productive viral replication, but also the host must be at an immunological "low point" where virus recrudescence can proceed. This view correlates well with the fact psychological and physiological stress, known to be immuno-suppressive, are potent inducers of HSV-1 reactivation.
Despite this relatively straightforward model, the role of LAT in the process is not at all clear. The "latency-active" portion of the large LAT transcript does not encode a protein, and this, as well as the difficulty in generating mutants of LAT that do not affect other genes has made analysis by traditional molecular genetic approaches difficult.
Recently, Wechsler and colleagues have reported that there is an increased amount of cells entering apoptosis in rabbit neuronal ganglia infected with LAT-negative virus. And, despite controversy concerning specifics, this effect has been essentially confirmed by Thompson and Sawtell working with the thermo-induced mouse model described above and by workers with Fraser at the University of Pennsylvania. Consistent with this suggestion, very small differences in the levels of viral DNA genomes per latently infected neurons have been reported by the latter groups for infections with LAT(+) vs LAT(-) virus. While our own work with rabbits has shown no such effect (perhaps as an artifact of the high levels of initial infection of rabbit corneas following low moi infection], we have seen a marginally significant difference in mouse latent infections using the foot-pad model.
At this juncture, then, it appears that LAT expression is correlated with a marginal difference in the fate of the acutely infected neuron during the earliest stages of establishment of latency. But the question of how LAT mediates this effect is still very much in the dark. It is possible that the expression of the transcript itself has a role in maintaining a sub-set of latent genomes in a transcriptionally competent mode, and Bloom and colleagues have recently reported that the patterns of acetylation of histones associated with latent viral genomes is consistent with such a model. It has also been suggested by Fraser and especially Weschler that other minor transcripts "perhaps encoding reactivation-mediating proteins" may underlie the LAT active region. Clearly, much further experimental work with the very expensive and difficult animal models currently available will be necessary to sort out this problem.