The Transgenic ICP4 Promoter Is Activated in Schwann Cells in Trigeminal Ganglia of Mice Latently Infected with Herpes Simplex Virus Type 1

Abstract

Herpes simplex virus type 1 (HSV-1) establishes a latent infection in neurons of sensory ganglia, including those of the trigeminal ganglia. Latent viral infection has been hypothesized to be regulated by restriction of viral immediate-early gene expression in neurons. Numerous in situ hybridization studies in mice and in humans have shown that transcription from the HSV-1 genome in latently infected neurons is limited to the latency-associated transcripts. In other studies, immediate-early gene (ICP4) transcripts have been detected by reverse transcription-PCR (RT-PCR) in homogenates of latently infected trigeminal ganglia of mice. We used reporter transgenic mice containing the HSV-1(F) ICP4 promoter fused to the coding sequence of the β-galactosidase gene to determine whether neurons in latently infected trigeminal ganglia activated the ICP4 promoter. Mice were inoculated via the corneal route with HSV-1(F). At 5, 11, 23, and 37 days postinfection (dpi), trigeminal ganglia were examined for β-galactosidase-positive cells. The numbers of β-galactosidase-positive neurons and nonneuronal cells were similar at 5 dpi. The number of positive neurons decreased at 11 dpi and returned to the level of mock-inoculated transgenic controls at 23 and 37 dpi. The number of positive nonneuronal cells increased at 11 and 23 dpi and remained elevated at 37 dpi. Viral proteins were detected in neurons and nonneuronal cells in acutely infected ganglia, but were not detected in latently infected ganglia. Colabeling experiments confirmed that the transgenic ICP4 promoter was activated in Schwann cells during latent infection. These findings suggest that the cells that express the HSV-1 ICP4 gene in latently infected ganglia are not neurons.

Herpes simplex virus type 1 (HSV-1) establishes a latent infection in peripheral sensory ganglionic neurons in humans and in the mouse model. Latent infection in ganglionic neurons is the means by which HSV-1 remains in infected humans for years. Periodic reactivation of latent infection results in spread of the virus to naive hosts. The mechanism of regulation of the latent viral genome in neurons has been intensively investigated in mouse models. Many studies utilizing in situ hybridization in mice and in humans have shown that transcription from the latent HSV-1 genome in neurons is limited to RNA that maps to the gene coding for latency-associated transcripts (LATs) (reviewed in references 17, 50, and 59). In contrast, lytic viral infection is characterized by an ordered cascade of viral gene expression: immediate-early (IE) genes are expressed first followed by early and late genes (43). Based on these findings, it has been hypothesized that latent infection of sensory neurons requires the absence of viral IE gene expression (reviewed in references 18 and 20). According to this hypothesis, reactivation of HSV-1 from latency is dependent upon activation or loss of repression of viral IE gene expression in neurons (20, 41, 43). IE genes would then activate the lytic cycle of viral gene expression, leading to the production of infectious virus. Some studies utilizing reverse transcription-PCR (RT-PCR) have demonstrated the presence of viral IE (ICP4) transcripts in homogenates of trigeminal ganglia from mice latently infected with HSV-1 (25, 26). These studies have raised the question of whether HSV-1 IE genes are transcribed in neurons during latent infection.

Trigeminal ganglia are comprised of multiple cell types, including Schwann cells, satellite cells, and different types of neurons (31, 45, 53). In order to understand the significance of the finding of ICP4 mRNA in latently infected sensory ganglia, it is important to know whether the ICP4 transcription originates from neurons. Presently available techniques such as in situ hybridization are not sensitive enough to localize the ICP4 transcripts to specific cells. It is possible to examine by very sensitive methods which specific cells activate the HSV-1 ICP4 promoter in latently infected trigeminal ganglia. Activation of the ICP4 promoter in neurons would be required for production of ICP4 transcripts in latently infected neurons.

We used reporter transgenic mice containing the HSV-1(F) ICP4 promoter fused to the β-galactosidase coding sequence to assay for HSV-1 ICP4 promoter activation in specific cells in latently infected trigeminal ganglia. Promoter transgenic mice were inoculated by the intracorneal route with HSV-1(F) to determine whether neurons in trigeminal ganglia activated the ICP4 promoter during latency. Trigeminal ganglia were assayed for β-galactosidase-positive cells at 5, 11, 23, and 37 days postinoculation (dpi). Moderate numbers of neurons and nonneuronal cells in trigeminal ganglia were positive for β-galactosidase at 5 dpi. The number of positive neurons decreased at 11 dpi, and no positive neurons were detected at 23 and 37 dpi. In contrast, the number of positive nonneuronal cells increased at 11 dpi, and large numbers remained positive at 23 and 37 dpi. The β-galactosidase-positive nonneuronal cells were morphologically and immunohistochemically determined to be Schwann cells. These studies showed that the ICP4 promoter was activated in Schwann cells in latently infected trigeminal ganglia. This suggests that Schwann cells may be the source of the ICP4 transcripts detected in previous studies of latently infected trigeminal ganglia (25, 26).

MATERIALS AND METHODS

Animal infections.

HSV-1(F) was grown in Vero cells, and virus stocks were prepared as previously described (39). Transgenic mice (Tg6305) containing the HSV-1(F) ICP4 promoter sequence (29, 33, 57) fused to the Escherichia coli β-galactosidase coding sequence have been previously described (34). Transgenic mice were mated with wild-type (C57BL/6 × C3H) mice to generate heterozygous transgenic mice and nontransgenic littermates. Transgenic mice and nontransgenic control littermates were identified by PCR with tail DNA (35) and primers (5′- GCATCGAGCTGGGTAATAAGCGTTGGCAAT-3′ and 5′- GACACCAGACCAACTGGTAATGGTAGCGAC−3′) for the β-galactosidase sequence (34). Mice were anesthetized with methoxyflurane or isoflurane, and each cornea was scratched with a 26-gauge needle. Mice were either mock infected with 5 μl of minimal essential medium containing fetal calf serum on each cornea or infected with 5 μl of medium containing 10⁷ PFU of HSV-1(F) on each cornea. All animals were maintained and handled in accordance with a protocol approved by the University of Missouri Animal Care and Use Committee.

Explant reactivation.

After corneal inoculation, nontransgenic mice of the same genetic background as the transgenic mice were euthanized at 23 and 37 dpi, and trigeminal ganglia were removed. Each ganglion was placed in a single well of a six-well plate containing medium and a monolayer of Vero cells. Explanted ganglia were monitored daily for cytopathic effects in the indicator cells and were transferred to fresh cells every 4 to 5 days (36).

β-Galactosidase assays.

Transgenic mice were infected with HSV-1 or mock infected, and nontransgenic littermates were infected with virus as described above. Mice from each group were euthanized at 5, 11, 23, and 37 dpi, and trigeminal ganglia were removed (see Results for the number of mice at each time point in each experiment). Trigeminal ganglia were frozen on dry ice and stored at −70°C. Ganglia were fixed as whole tissues in 4% paraformaldehyde for 30 min immediately upon removal from −70°C. Ganglia were washed in phosphate-buffered saline (PBS) and incubated in substrate solution for 14 to 18 h at 37°C (34, 37). The substrate solution contained 20 mM potassium ferrocyanide, 20 mM potassium ferricyanide, 2 mM MgCl₂, and 1 mg of 5-bromo-3-indolyl-β-d-galactoside (X-Gal) per ml, as well as 120 μl of 10% Nonidet P-40 and 100 μl of 1% sodium deoxycholate per 20 ml. Trigeminal ganglia were then washed for 5 min with PBS and thinly sliced with a razor blade. All tissue sections were mounted in PBS, and the coverslips were sealed with Permount. The total numbers of β-galactosidase-positive neurons and nonneuronal cells in each ganglion in each group were counted. The means and standard errors of the means were computed with SigmaStat. Neurons were morphologically defined as having round-to-oval nuclei approximately 9 to 20 μm in diameter. Schwann cells were morphologically defined as having elongated nuclei approximately 10 to 20 μm in length and 2 to 5 μm in width; the length of the nuclei was at least two times the width.

Immunohistochemistry for HSV-1.

Paraffin-embedded sections of trigeminal ganglia from HSV-1-infected nontransgenic control mice were deparaffininzed with xylene and ethanol. Sections of virus-infected ganglia were labeled for HSV-1 antigen by the avidin-biotin-peroxidase method (Vector) with HSV-1 antiserum (DAKO) at a 1:1,000 dilution (35). Selected sections were lightly counterstained with hematoxylin. Ten ganglia from five HSV-1-infected animals were assayed at each time point (5, 11, 23, and 37 dpi). As controls, sections of mock-infected ganglia were reacted with HSV-1-specific antibody and adjacent sections of HSV-1-infected ganglia were reacted with control rabbit serum (DAKO) in the same assays.

Colabeling of β-galactosidase-positive cells.

The identity of the β-galactosidase-labeled cells in latently infected ganglia was verified by colabeling with markers for either Schwann cells or neurons. Trigeminal ganglia from nine latently infected transgenic mice were collected at 23 dpi, frozen on dry ice, and stored at −70°C. Trigeminal ganglia were first processed for β-galactosidase labeling as described above and thinly sliced with a razor blade. Slices of trigeminal ganglia that contained positively labeled cells were paraffin embedded. Tissues were processed for embedding with Clear-Rite III. Seven-micrometer sections of latently infected trigeminal ganglia were deparaffinized with Clear-Rite III, xylene, and ethanol. Sections were labeled for glial fibrillary acidic protein (GFAP), a marker for Schwann cells (5, 12, 16, 40, 47, 62), or neurofilament protein, a marker for neurons (15), by the avidin-biotin-peroxidase method (Vector). Diaminobenzidine tetrahydrochloride (DAB) was used as the substrate. Rabbit anti-GFAP (DAKO) was used at a dilution of 1:250, and mouse anti-neurofilament protein (Sigma) was used at 1:40. The M.O.M. kit (Vector) was used to reduce background labeling with the mouse anti-neurofilament antibody. Sections were placed onto coverslips and then examined and photographed by light microscopy. The procedure described above was modified in a second set of experiments. Six transgenic mice were infected with HSV-1, and trigeminal ganglia were collected at 23 dpi. Ten-micrometer cryotome sections were cut from β-galactosidase-labeled ganglia that had been refrozen. These sections were stained for GFAP by the avidin-biotin-peroxidase method as described above.

RESULTS

HSV-1 antigen was detected in neurons and nonneuronal cells in trigeminal ganglia of acutely infected mice.

Viral antigen was detected in a variety of cell types, including neurons, Schwann cells, and satellite cells, in acutely infected trigeminal ganglia (Fig. 1A and B). At 5 dpi, all ganglia (10 of 10) contained multiple HSV-1-positive cells. At 11 dpi, 2 of 10 ganglia contained a single neuron that was positive for HSV-1 antigen (data not shown). At 23 and 37 dpi, all ganglia were negative by immunoperoxidase assays for HSV-1 antigen in all cell types (data not shown). Mock-infected control ganglia (reacted with HSV-1 antiserum) were negative for staining (data not shown). Adjacent sections of acutely infected ganglia incubated with normal rabbit serum were also negative for staining (Fig. 1C).

FIG. 1.

Immunohistochemical labeling of HSV-1 antigen in nonneuronal cells and neurons in trigeminal ganglia of mice 5 dpi. Mice were infected with HSV-1(F), and trigeminal ganglia were collected at 5 dpi. Sections of ganglia were reacted with HSV-1-specific antiserum (A and B) or control rabbit antibody (C). Bound antibody was detected with an indirect immunoperoxidase assay. Sections in panels A and B were lightly counterstained with hematoxylin. (A) Arrowheads indicate satellite cells positive for HSV-1 antigen. The arrow indicates a positively labeled neuron. (B) Schwann cell positive for HSV-1 antigen. (C) Section of an infected ganglion incubated with control rabbit antibody. Bar, 15 μm.

Latent infection of trigeminal ganglia in this model at 23 and 37 dpi was confirmed by explant cocultivation experiments. Reactivated virus was detected at 5 to 9 days postexplant from 10 of 10 ganglia (five mice) explanted at 23 dpi and at 7 to 10 days postexplant from 8 of 8 ganglia (four mice) explanted at 37 dpi. As expected, multiple cell types in trigeminal ganglia were infected with HSV-1 during acute infection, and a latent infection was established in trigeminal ganglia at 23 and 37 dpi in our model.

The ICP4 promoter was activated in neurons and nonneuronal cells in trigeminal ganglia of transgenic mice at 5 and 11 dpi with HSV-1.

Two different experiments were performed. In the first experiment, trigeminal ganglia were collected at 5, 11, 23, and 37 dpi (Fig. 2A). In the second experiment, ganglia were collected at 11, 23, and 37 dpi (Fig. 2B). The data shown at 5 dpi (Fig. 2A) include two groups of mice infected independently. At 5 dpi, moderate numbers of neurons (average of 50 cells per ganglion [n = 13]) and nonneuronal cells (average of 57 cells per ganglion [n = 13]) were labeled for β-galactosidase (Fig. 2A and 3A and C). No labeled nonneuronal cells and 0 to 3 labeled neurons were present in trigeminal ganglia (n = 8) from mock-infected transgenic mice (Fig. 2A and 3B and D). No nonneuronal cells or neurons were labeled for β-galactosidase in trigeminal ganglia (n = 4) of infected nontransgenic mice at 5 dpi (data not shown).

FIG. 2.

Number of β-galactosidase-positive neurons and Schwann cells in trigeminal ganglia of infected transgenic mice. Transgenic mice were infected with HSV-1(F) or mock infected. Trigeminal ganglia were collected at 5, 11, 23, and 37 dpi and assayed for β-galactosidase labeling. The numbers of positive neurons and Schwann cells in each ganglion were counted as described in Materials and Methods. Results are reported as means ± standard errors. The results at 5 dpi (A) represent the combined counts from two independently infected groups of mice. An asterisk indicates that mock-infected transgenic mice were not used at this time point.

FIG. 3.

The transgenic ICP4 promoter is activated in both neurons and nonneuronal (Schwann) cells in trigeminal ganglia of transgenic mice at 5 dpi. Transgenic mice were infected with HSV-1(F) (A and C) or mock infected (B and D). Trigeminal ganglia were collected at 5 dpi and processed for β-galactosidase labeling. The medium-to-large round nuclear profiles in panel A are neurons, and the elongated nuclear profiles in panel B are nonneuronal (Schwann) cells. Bar, 10 μm.

At 11 dpi, the number of β-galactosidase-positive neurons was substantially less than that at 5 dpi, but still greater than that in mock-infected transgenic mice (average of 14 positive cells/ganglion [n = 2] [Fig. 2A] and 12 positive cells/ganglion [n = 7] [Fig. 2B]). In contrast to the decreased numbers of positive neurons, the number of β-galactosidase-positive nonneuronal cells at 11 dpi in trigeminal ganglia of infected transgenic mice had increased to an average of 175 positive cells per ganglion (n = 2) (Fig. 2A) and 175 positive cells per ganglion (n = 7) (Fig. 2B and 4A).

FIG. 4.

The transgenic ICP4 promoter is activated in Schwann cells in trigeminal ganglia of transgenic mice at 11, 23, and 37 dpi. Transgenic mice were infected with HSV-1(F) (A, B, and C) or mock infected (D). Trigeminal ganglia were collected at 11, 23, and 37 dpi and assayed for β-galactosidase labeling. Positively labeled elongated nuclear profiles characteristic of Schwann cells are visible at 11 dpi (A), 23 dpi (B), and 37 dpi (C). (D) Ganglia from a mock-infected transgenic mouse at 37 dpi. Bar, 10 μm.

The ICP4 promoter was activated in nonneuronal cells in trigeminal ganglia of transgenic mice during latent infection with HSV-1.

At 23 dpi, β-galactosidase-positive nonneuronal cells in virus-infected transgenic mice (Fig. 4B) remained significantly elevated above those in mock-infected mice at an average of 163 cells per ganglion (n = 8) (Fig. 2A) and 263 positive cells/ganglion (n = 8) (Fig. 2B). By 37 dpi, the number of β-galactosidase-positive nonneuronal cells (Fig. 4C) had decreased, but remained significantly elevated above that in mock-infected mice at an average of 47 cells/ganglion (n = 8) (Fig. 2A) and 32 cells per ganglion (n = 8) (Fig. 2B). Surprisingly for each of the above listed groups (n = 8) at 23 and 37 dpi, the numbers of β-galactosidase-positive neurons in trigeminal ganglia of virus-infected transgenic mice were not significantly different from those of mock-infected transgenic mice (usually 0, but always less than 3) (Fig. 2A and B). Trigeminal ganglia taken from HSV-1-infected nontransgenic mice did not show any β-galactosidase-positive nonneuronal cells at any time point (data not shown). Four mock-infected transgenic mice and two infected nontransgenic mice were used at each time point in each experiment.

The β-galactosidase-positive nonneuronal cells had very elongated nuclei (Fig. 3B and Fig. 4A, B, and C) that were approximately 2 to 4 times greater in length (10 to 20 μm) than width (2 to 5 μm). These β-galactosidase-positive nonneuronal cells were often found in rows in nerve roots. The morphological features of the labeled cells corresponded to those of Schwann cells (55). The identity of the β-galactosidase-positive cells in latently infected ganglia was confirmed by colabeling experiments. Transgenic mice were infected and trigeminal ganglia were collected at 23 dpi and assayed for β-galactosidase labeling. β-Galactosidase-positive cells were colabeled for GFAP, a marker for Schwann cells, by using antibody to GFAP in an indirect immunoperoxidase assay. In multiple experiments, the β-galactosidase-positive cells at 23 dpi were colabeled for GFAP (Fig. 5A and B). Control rabbit antibody did not label any β-galactosidase-positive cells (Fig. 5C). Colabeling experiments with anti-neurofilament antibody were performed on sections of latently infected (23 dpi) trigeminal ganglia containing β-galactosidase-positive cells. In multiple experiments, none of the β-galactosidase-positive cells were labeled with neurofilament antibody (Fig. 5D). Neurons and neuronal processes, which did not contain β-galactosidase, were positively labeled with the neurofilament antibody in the same sections (Fig. 5D). Control mouse monoclonal antibody did not label neurons or neuronal processes (Fig. 5E). Taken together, these results confirm that the ICP4 promoter was activated in Schwann cells in trigeminal ganglia of latently infected mice.

FIG. 5.

The transgenic ICP4 promoter is activated in Schwann cells but not neurons in trigeminal ganglia of transgenic mice latently infected with HSV-1. Transgenic mice were infected with HSV-1(F), and trigeminal ganglia were collected at 23 dpi. Sections of ganglia were colabeled for β-galactosidase and GFAP. (A and B) β-Galactosidase-positive Schwann cells were colabeled for GFAP. Arrows in panel B indicate the positively labeled cytoplasm surrounding the β-galactosidase-positive nucleus. (C) No GFAP labeling in the section adjacent to panel A, which was incubated with control rabbit antibody. (D) Section of ganglia was colabeled for β-galactosidase and neurofilament. Arrowheads indicate neuronal processes labeled for neurofilament. Arrows indicate β-galactosidase-positive cells negative for neurofilament label. (E) No neurofilament labeling in section adjacent to panel D that was incubated with control mouse monoclonal antibody. Arrows indicate β-galactosidase-positive cells. Bars, 8 μm in panels A to C and 10 μm in panels D and E.

DISCUSSION

In the mouse model of HSV-1 infection, intracorneal or footpad inoculation results initially in an acute infection of sensory ganglia (reviewed in references 4, 18, 50, and 59). This stage of infection is characterized by the presence of all classes of viral gene transcripts (IE, early, and late), viral antigens, and infectious virus in the sensory ganglia (10, 23, 28, 51, 52). Acute infection in immunocompetent mice is generally restricted to approximately 3 to 7 dpi (25, 41). Latent infection of sensory neurons persists after clearance of the acute infection (40, 51). During this phase, infectious virus and viral antigen cannot be detected, and viral transcription is limited. The latent stage of infection is established by 3 weeks postinfection (4, 18, 28, 32).

During acute infection of sensory ganglia, multiple cell types in addition to neurons contain viral antigen and viral particles. Electron microscopic examination of acutely infected dorsal root and trigeminal ganglia has shown virus particles in Schwann cells and satellite cells from 4 to 8 dpi (10, 23, 56). Immunohistochemical and immunofluorescent examinations of acutely infected ganglia have demonstrated the presence of viral antigen in nonneuronal cells at 2 to 6 dpi (10, 23, 28). Our study of HSV-1 antigen in trigeminal ganglia confirms these earlier findings. A few studies have suggested that nonneuronal cells may be able to maintain virus during the latent period of infection. Nonneuronal cells, including epithelial cells of the cornea and skin, have been previously demonstrated to contain viral DNA (30, 37, 44, 49) and in some cases to yield infectious virus following long-term infection (2, 8, 9, 21, 46, 48).

Numerous in situ hybridization studies have determined that HSV transcription during latency in trigeminal ganglia is limited to LATs in neurons (11, 14, 38, 52, 61). Infectious virus and viral antigen have not been detected during latent infection in sensory ganglia of mice in the absence of reactivation. RT-PCR studies have detected low levels of IE (ICP4) gene expression in homogenates of sensory ganglia during latency in mice (25, 26). In these studies, it was not determined which cells in the ganglia produced the ICP4 transcripts.

It has been hypothesized that neuronal regulation of IE gene expression controls the latent infection and reactivation of HSV-1 (18, 20, 54). In this model, the absence of IE gene expression in neurons is necessary for maintenance of latent infection with HSV-1. According to this hypothesis, reactivation of HSV-1 would be triggered by loss of repression or activation of viral IE genes, which would lead to virus replication. The change in IE gene expression in neurons might result from altered neuronal gene expression in response to physiologic stimuli, such as UV light, trauma, or heat stress. This model predicts that viral IE gene promoter activation should be absent in neurons during latent infection. The above-stated hypothesis for regulation of latent HSV-1 infection in neurons has been questioned based on the detection of ICP4 transcripts in homogenates of latently infected trigeminal ganglia by RT-PCR. This model for the regulation of HSV-1 latency will require modification if viral IE gene expression occurs in neurons during latent infection.

Given the cellular complexity of sensory ganglia, it is necessary to know whether neurons or another cell type are the source of the ICP4 transcripts detected in homogenates of latently infected ganglia. If cells other than neurons produce ICP4 transcripts in ganglia during the time frame of latent infection, then the IE transcription may not be related to regulation of latent infection in neurons. We examined trigeminal ganglia from acutely (5 dpi) and latently (23 dpi) infected mice by in situ hybridization for ICP4 transcripts with a digoxigenin-labeled ICP4 DNA fragment containing nucleotides +19 to +119 relative to the transcription start site of the ICP4 gene. ICP4 transcripts were detected in acutely infected ganglia, but were not detected in latently infected ganglia (N. S. Taus and W. J. Mitchell, unpublished data). Sections of trigeminal ganglia from acutely and latently infected mice were also examined by in situ PCR for viral DNA. The primers and conditions of the PCR were as previously described (30, 37). HSV-1 DNA was detected in a variety of cells, including Schwann cells and satellite cells in acutely infected ganglia. Several cells morphologically defined as Schwann cells in 2 of 10 ganglia appeared to be positive for viral DNA in latently infected trigeminal ganglia (N. S. Taus and W. J. Mitchell, unpublished data.).

In the absence of a direct assay that is sensitive enough to detect IE transcripts in individual cells in latently infected ganglia, we have used an assay to measure activation of the HSV-1 ICP4 promoter in situ. This assay utilizes transgenic mice that contain the ICP4 promoter fused to the bacterial β-galactosidase coding sequence (34). It allows sensitive detection of ICP4 promoter activity in individual cells during acute and latent infection with HSV-1. Assay systems containing chimeric reporter genes (heterologous promoter fused to the coding sequence of an indicator protein) are widely used for studying gene expression. These assay systems include reporter transgenes integrated into mouse chromosomal DNA, chimeric reporter genes inserted into plasmid DNA and transfected into cells, and reporter transgenes inserted into novel positions in the viral genome (1, 3, 6, 7, 13, 19, 22, 24, 27, 42, 43, 58, 60). All of these approaches have possible problems; however, much has been learned about gene regulation from the use of reporter genes.

Our results show that the transgenic ICP4 promoter was activated in large numbers of nonneuronal (Schwann) cells in the trigeminal ganglia of mice at 11, 23, and 37 days after corneal inoculation. Activation of the transgenic ICP4 promoter in Schwann cells during latent HSV-1 infection suggests that viral IE gene transcription occurs in Schwann cells. It is possible, but less likely, that the ICP4 reporter transgene is expressed in Schwann cells as a result of latent infection of neurons. Transgenic ICP4 promoter activation was not detected in neurons in latently infected trigeminal ganglia (23 and 37 dpi). Moderate numbers of neurons were positive at 5 dpi, but these decreased to the level of uninfected transgenic ganglia (0 to 3 positive cells/ganglion) in latently infected mice. The absence of detectable transgenic ICP4 promoter activation in neurons during latency suggests that IE gene transcription during latency is unlikely to occur in neurons.