Persistence of Infectious Herpes Simplex Virus Type 2 in the Nervous System in Mice after Antiviral Chemotherapy

Abstract

Young adult mice were inoculated with herpes simplex virus type 2 (HSV-2) in the ear pinna. A relatively severe infection resulted, and 45% of the mice died by 11 days postinfection. Therapy at 1 mg/ml by means of the drinking water with either famciclovir for periods of 5 or 10 days or valaciclovir for 5, 10, 15, or 20 days decreased clinical signs and reduced mortality to 15% or less. Throughout a period of 27 days, mice were tested daily for the presence of infectious virus in the ear pinna, brain stem, and ipsilateral trigeminal ganglia. Virus was cleared from these tissues in surviving, untreated animals by 12 days postinfection, and no infectious virus was detected subsequently in any tissue. Furthermore, no infectious virus was detected after day 9 in mice that had been treated with famciclovir. In mice that had received valaciclovir therapy, however, infectious virus was repeatedly detected in the trigeminal ganglia and brain stem tissue samples up to 7 days after treatment was discontinued. To date, no specific mechanism to account for these results has been discovered; however, possible mechanisms for the persistence of potentially infectious virus in neural tissue of treated mice are discussed.

Since the first description of murine infection models for herpes simplex virus (HSV), it has been shown repeatedly that, following inoculation by means of a peripheral site, such as the skin or the cornea, virus replication occurs both at the local site and in the peripheral ganglia that innervate the site of local infection (20). This usually persists for 1 to 2 weeks by which time infectious virus is cleared from both local and neural tissue in mice that survive the acute infection. It has been repeatedly shown that ganglia or central nervous system (CNS) tissues removed after the acute infection has subsided and subsequently explanted, homogenized, and tested for infectious virus yield negative results (11).

It is very well known that the peripheral nervous system continues to harbor HSV in latently infected neurons (4, 17, 22) and that these tissues may be reactivated by explanting the ganglia and incubating the tissue in vitro (26, 28). CNS neurons also continue to harbor HSV DNA, although the reactivation of CNS tissues to yield infectious virus has proved more difficult (5, 21). It has also proved difficult to reactivate latent HSV in vivo in mice, although, over the years, several different methods have been employed to achieve this with limited success. For example, Sellotape stripping, UV irradiation of ear pinnae (11), and, more recently, use of the corneal infection model have enabled infectious virus to be detected both in CNS and ganglion tissue 1 to 2 days after transient hyperthermia (18).

In our previous published studies on the chemotherapy of HSV type 1 (HSV-1) in mice with prodrugs famciclovir (FCV) and valaciclovir (VACV) yielding the nucleoside analogues penciclovir and acyclovir (ACV), respectively, we have reported that mice treated with VACV produce transient recurrences of infectious virus in the nervous system 1 or 2 days after the cessation of VACV treatment (23). This was most marked when the mice were subjected to an immunosuppressive regimen during the period of chemotherapy, in which case both neural tissues and skin were found to yield infectious virus after treatment stopped (7, 25). However, similar results were obtained when no immunosuppression was applied, although, in this case, the recurrences were confined to neural tissues and no infectious virus was detected in skin samples (23).

Similarly, we reported that when experiments were carried out with HSV-2, recurrences were detected in immunocompetent mice (24). The fact that recurrences of infectious virus on cessation of treatment were observed only in mice that had been treated with VACV implied that there may be a difference between the mechanisms of action of FCV and VACV in vivo to account for the response. Alternatively, the apparent difference between the compounds might have resulted from a systematic error in the design or execution of experiments carried out to date. The objective of the present study was to confirm the observation that transient recurrence of infectious virus occurs in the nervous system of HSV-2-infected mice on cessation of VACV (but not FCV) therapy and to rule out the possibility of a trivial explanation for the data. This required a very large experiment with meticulous methods. We describe carefully the protocols employed to generate these data, and this attention to detail may help to explain the remarkably low mouse-to-mouse variation we obtained in the study. The central observation (recurrence of infectious virus in neural tissues on cessation of VACV therapy) was confirmed. The temporal pattern of infectious virus detected in the mice forms a basis on which to build hypotheses to explain the observations which may be subject to further experimentation. These results should provoke discussion as to whether or not these findings for mice have a wider significance.

MATERIALS AND METHODS

Virus inoculum.

The virus used was HSV-2 strain Bry (9, 27). Working stocks were grown in BHK-21 cells inoculated at a low multiplicity of infection and stored at −70°C as previously described (8).

Mice and virus inoculation.

A total of 680 female BALB/c mice (Bantin and Kingman, Kingston, Hull, United Kingdom) aged 3 to 4 weeks were inoculated intradermally in the left ear pinna with 10 μl of medium containing HSV-2 (Bry) at a dose of 2 × 10⁴ PFU/mouse. Each cage initially contained 10 mice. Clinical signs (weight loss, ear swelling, erythema, neurological signs, and death) in separate groups of 20 animals (i.e., two cages for each treatment group) were monitored daily.

Antiviral therapy.

FCV and VACV were synthesized at the laboratory of SmithKline Beecham by previously published methods (2, 10). The compounds were administered by means of the drinking water starting from 22 h postinfection (p.i.) continuously for 5, 10, 15, or 20 days. FCV was dissolved in tap water at 0.2 and 1 mg/ml. VACV was dissolved at 1 and 5 mg/ml. The water consumption for each cage of 10 mice was measured each day to enable the dose to be calculated, and the supply was refreshed daily.

Sampling tissues.

Initially, all mice were weighed and placed in cages of 10 mice to give an even distribution of weights. Some cages were assigned for observation only, and others contained mice to be sampled on particular days. All therapy was initiated 22 h after inoculation. Since the inoculations took approximately 7 h, the starting times were staggered accordingly. A strict routine was then adhered to each day thereafter. First, all mice were checked for mortality. Their water bottles containing drugs were replenished, and water consumption was recorded. Mice in observation cages were scored for clinical signs, weight loss or gain, and ear thickness. At approximately the same time each day (2 p.m.) mice were selected at random from predetermined cages in groups of three. They were euthanized, and the left ear pinna, brain stem, and left trigeminal ganglia were removed from each mouse and placed immediately into a 1-ml aliquot of Eagle's minimum essential medium. These remained on ice for up to 3 h until all dissections were completed. Sampling continued according to this routine every day up to 27 days p.i.

Assay for infectious virus.

Immediately following completion of all the dissections, the tissues were homogenized individually. Thus, the earliest samples remained on ice for up to 5.5 h, with most less than 4 h. Each homogenized sample was subjected to sonication for 2 min and then low-speed centrifugation for 10 min at 2,000 rpm in a Beckman Chillspin. The supernatant was tested for infectious virus by plaque titration on monolayers of BHK cells previously prepared in 24-well plates. Two hundred microliters of each homogenate was inoculated onto just-confluent cell sheets at 10⁰, 10⁻¹, and 10⁻² dilutions in triplicate. Plates were examined each day from 48 h up to day 5 when the monolayers were stained and the results were finally scored. Each sample was coded at dissection, and the code was not broken until all samples had been tested and the plaques had been counted. This method was not intended to give highly quantitative results for virus titers in the tissues but was intended to be the most sensitive test for positive samples.

Reconstruction of infected tissues.

Because of the surprising pattern of results produced, a small experiment in which uninfected mouse brain stem tissues were prepared according to a schedule similar to that described above was carried out subsequently. At various points in the procedure tissues that had been obtained from uninfected mice were “spiked” by adding medium containing a range of titers of HSV-2 (10⁰ to 10³ PFU) that had been predetermined by plaque titration. In summary, we obtained approximately 50% recovery of the added infectious virus with absolutely no evidence of cross contamination between samples. No virus was detected in samples spiked with ≤10 pfu of virus, and 100% of the samples scored positive when tissues were spiked with ≥100 PFU. These results suggested that a small loss of infectious virus occurred during the processing of tissue homogenates, but the loss was not sufficient to account for the markedly different titers that were recorded from various infected mouse tissues in the main experiment.

RESULTS

Dose of antiviral compounds.

Mice showed no distaste for either of the compounds at any concentration up to and including VACV at 5 mg/ml, although toxic signs (see below) were observed at this dose. Their total intake was measured each day, and the consumption was used to calculate the average daily dose. Volumes consumed per day ranged from 2.0 to 3.0 ml/mouse with a mean value of 2.8 ± 0.1 ml/mouse for all the treatment groups. There were no significant differences among the levels of consumption of water containing either drug at any dose. The water consumption corresponded to a mean calculated dose of 150 to 200 mg/kg of body weight/day for the compounds when supplied at 1 mg/ml, with proportional doses at the higher and lower concentrations; thus, at 5 mg/ml, the mice consumed VACV at approximately 1g/kg/day.

Clinical signs.

Without therapy, all mice developed characteristic signs of HSV-2 infection during the 2 weeks after inoculation. All these mice developed erythematous and swollen ear pinnae, and approximately half the mice died from the infection as shown in Fig. 1. The infected, untreated mice that died all succumbed during the period 9 to 11 days after inoculation. No mice died at later times or, for treated mice, during the period when recurrences of infectious virus were detected in neural tissues.

FIG. 1.

Effects of oral therapy with FCV or VACV on survival of HSV-2-infected mice. Mice were inoculated intradermally via the ear pinna and treated by means of the drinking water from 22 h p.i. for the numbers of days indicated. Mortality (in percent) was determined from groups of 20 mice. (a) FCV therapy; (b) VACV therapy. The group of infected, untreated control mice was the same for panels a and b.

Effects of therapy.

Treatment with either FCV or VACV reduced the severity and duration of clinical signs (Fig. 1). Mortality was reduced from 45 to ≤5% for FCV and to 10 to 15% for VACV at 1 mg/ml. Higher mortality (60%) was observed in mice receiving 5 mg of VACV/ml, and this appeared to be associated with toxicity (Fig. 1b) (see below). The effects of therapy were reflected in the weights of mice, as shown in Fig. 2. Weight loss relative to that of infected, untreated mice was reduced by FCV therapy. VACV had less effect on weight loss, and at the higher dose of 5 mg/ml the effect was exacerbated. Furthermore, at 1 mg of VACV/ml, all four groups of treated mice weighed less than the infected, untreated group, providing further evidence of a toxic effect of therapy.

FIG. 2.

Effects of oral therapy with FCV or VACV on weight change of HSV-2-infected mice. Mice were inoculated intradermally via the ear pinna and treated by means of the drinking water for the numbers of days indicated. Mice were from different groups used in the experiment shown in Fig. 1. Mean weights were determined daily for groups of eight mice. The infected, untreated control group and the uninfected control group were the same for all panels.

In a small subsequent experiment, groups of uninfected mice were treated with VACV in the drinking water at 1, 2.5, or 5 mg/ml because toxicity was suspected. Severe weight loss in uninfected mice receiving VACV at 5 mg/ml was observed, although there was no weight loss in mice receiving 1 mg/ml. An intermediate effect was seen in the group receiving 2.5 mg/ml. On cessation of therapy after 12 days, uninfected mice that had received 5 mg/ml rapidly gained weight to match that of the untreated animals but no mortality was observed up to day 15 p.i. (Fig. 3). No weight loss or any other adverse signs were observed in uninfected mice treated with up to 5 mg of FCV/ml (data not shown).

FIG. 3.

Effects of increasing doses of oral VACV on the weight gain of uninfected mice. Uninfected female BALB/c mice (4 weeks old) were treated with VACV in the drinking water at 1, 2.5, or 5 mg/ml. Treatment was continued for 12 days; thereafter mice received normal tap water. Average weights were determined daily from groups of 10 mice (two groups of 10 mice were untreated controls and are shown as separate groups on the graph). This was a different experiment from that depicted in Fig. 1 and 2.

Detection of infectious virus during and after cessation of therapy. (i) Ear pinna.

Infected, untreated mice showed biphasic virus growth in the ears with peak titers on day 2 and then again on day 8 p.i., but infectious virus was cleared to below the level of detection (<0.5 log₁₀ PFU/tissue) on day 10 p.i., and no ears were positive for infectious virus at later times (Fig. 4). Mice receiving FCV or VACV at 1 mg/ml cleared virus from the ear pinna on day 3 p.i., while those receiving 0.2 mg of FCV/ml cleared the virus on day 5 p.i. There was no recurrence of infectious virus in the ear pinna in any of the treatment groups on cessation of therapy (Fig. 4).

FIG. 4.

Detection of infectious virus in the skin and neural tissues of individual mice with or without oral therapy. Mice were inoculated intradermally in the left ear pinna and treated by means of the drinking water starting from 22 h p.i. and continuing for the periods shown. The mice were taken from different groups from the same experiment shown in Fig. 1 and 2. Each day from day 1 to 27 p.i. three untreated mice and three mice from each treatment group were euthanized and their left ear pinnae (E), left trigeminal ganglia (TG), and brain stems (BS) were removed, homogenized, and tested for infectious virus by plaque titration. Red square, tissue scored positive with ≥2.1 log₁₀ pfu/tissue; orange square, tissue was positive with ≤2.0 log₁₀ pfu/tissue; 0, no infectious virus was detected; blank, no mice were sampled. The periods of treatment are indicated by blue (VACV) and green (FCV) rectangles.

(ii) Brain stem.

Virus was first detected in the brain stems of untreated mice on day 6 p.i., with the highest titers on day 7 or 8 p.i. (approximately 3.0 log₁₀ PFU/brain stem). Virus was cleared to below the level of detection by day 12 p.i. in surviving untreated animals. No infectious virus was detected in the brain stems of mice receiving 1 mg of FCV/ml either during or after therapy. At the lower dose of 0.2 mg/ml, virus was detected during therapy on days 7 to 9 p.i. (Fig. 4) (approximately 2.0 log₁₀ PFU/brain stem) or on days 7 and 8 p.i. only after cessation of therapy on day 5 (approximately 2.5 log₁₀ PFU/brain stem). For mice receiving VACV up to day 5 or 10 p.i., virus was detected sporadically during the period from day 6 to 16 p.i.

Only in VACV-treated mice was infectious virus detected in the brain stem on several distinct occasions after cessation of therapy. This included mice that had received 5 mg of VACV/ml for 5 or 10 days and mice that had received VACV at 1 mg/ml for 15 or 20 days continuously (Fig. 4). The titers of virus were in the range of 2.0 to 3.5 log₁₀ PFU/brain stem.

(iii) Trigeminal ganglia.

Virus was first detected in the trigeminal ganglia of infected, untreated mice on day 6 p.i. and was cleared to below the level of detection by day 11 p.i. No infectious virus was detected in the ganglia of mice receiving 1 mg of FCV/ml during therapy; however, at the lower dose of 0.2 mg/ml, infectious virus was detected on days 6 and 7 p.i., when therapy terminated on day 5, with titers of approximately 2.0 log₁₀ PFU/trigeminal ganglion. For mice receiving VACV therapy, infectious virus was detected at various times during the period of acute infection as shown in Fig. 4 (1 to 3 log₁₀ PFU/trigeminal ganglion). However, infectious virus was still detectable in some infected, untreated mice up to day 10 p.i.

Persistence of virus on cessation of therapy.

At various times after cessation of VACV therapy, infectious virus was detected in either the brain stem, the left trigeminal ganglia, or both. In many cases two or three of the three mice tested yielded positive results simultaneously, especially for the brain stem samples, notwithstanding the fact that the mice were drawn from up to three different cages according to a predetermined random-selection procedure. In most samples the level of virus detected was well above the level of sensitivity (0.5 log₁₀ PFU/tissue). Virus was frequently detected in both the brain stem and trigeminal ganglia on the same day or in the brain stem only followed by detection in the trigeminal ganglia only on the next day. The temporal relationship between detection of virus in the two sites is shown in Table 1. All the recurrences occurred between 1 and 7 days after cessation of therapy. Recurrences were observed after high-dose VACV therapy and when therapy was continued for 15 or 20 days. No similar recurrence was observed on cessation of FCV therapy except in the group treated at 0.2 mg/ml when treatment stopped on day 5. Thus, these recurrences occurred during the period at which virus growth in the infected, untreated controls was at maximum levels due to the acute phase of the infection.

TABLE 1.

Relationship between detection of infectious virus in neural tissues and cessation of VACV therapy in individual mice^a

^aThe data were derived from the virus-positive tissues shown in Fig. 4 obtained from VACV-treated mice. 

^bConcentration of VACV in drinking water. 

^cTG, trigeminal ganglion; BS, brain stem. 

^dParentheses indicate that virus-positive tissue was detected during the period of acute infection. Diagonal and vertical lines indicate that the brain stem was virus positive one day before or the same day as, respectively, the trigeminal ganglia in the same group of mice. 

^eTotal number of mice for which virus-positive tissue was detected of the 21 mice tested. 

DISCUSSION

The term recurrence is normally used to describe the production of infectious HSV following reactivation from latency (30). We are using the term in the present paper in a more general way to describe our observation of a transient reappearance of infectious virus in the tissue or “rebound” of infectivity following cessation of a period of therapy. Observations of infectious virus recurring following termination of VACV therapy have been reported previously for HSV-1- (7, 23) and HSV-2-infected (24) mice. However, for HSV-1, recurrences were observed on a single occasion after treatment ceased, while for HSV-2, infectious virus was detected on multiple occasions (24). The purpose of the present study was to confirm and extend the observations with HSV-2. The most important findings were as follows. (i) There was a distinctly biphasic pattern of infection in the ears of infected, untreated mice. Infectious virus peaked in the ear on day 2 p.i., with the second peak occurring at the stage when virus growth was observed in neural tissue (day 8 p.i.). The second peak may reflect a centrifugal flow of virus to the skin from the peripheral nervous system neurons by fast axonal transport (15). (ii) Following clearance of infectious virus from the tissues of infected, untreated mice on day 11, no infectious virus was detected in any tissue for more than 2 weeks, during which time groups of three mice were examined each day. During this period a total of 144 tissue samples from 48 mice were rigorously tested for the presence of infectious virus and none was detected. Similarly negative results were obtained for 216 tissues from 72 mice treated with FCV up to and including day 17 p.i. (iii) In contrast to the above, when mice had been treated with VACV, many brain stem and trigeminal ganglion samples tested positive for virus during the same period of observation. These results leave no doubt that the detection of infectious virus posttherapy was exclusively associated with VACV.

The resurgence of infectious virus described in this paper is not readily explained. The pattern of infectious virus recovery described here is, however, broadly similar to that described in a previous paper (24). Recently we have carried out two further experiments using the same strain of HSV-2. In one case mice were inoculated in the ear pinna and in the other in the neck by means of scarification. Infectious virus was cleared from control mice by day 10 in both cases. Neural tissues were tested daily, and the recurrence of infectious virus was detected in both models in VACV-treated mice on days 13 and 14 p.i., thus confirming that the phenomenon is reproducible.

In the present study the recurrence phenomenon did not appear to be dependent on the dose or duration of VACV therapy. Moreover, it occurred at the highest dose of VACV employed (5 mg/ml). Uninfected mice given 5 mg of VACV/ml in the drinking water (corresponding to approximately 1g/kg/day) showed clear evidence of toxicity. They lost weight and only regained control weights after cessation of antiviral administration. A similar, though smaller, weight loss was observed at 2.5 mg/ml, while 1 mg/ml had no effect. The nature of this toxicity was not established; however, it raises the possibility that the toxic effects of VACV at all doses may have a bearing on the response to therapy in this model. One hypothesis, therefore, to explain our data is that a toxic effect of VACV in the infected mice interfered with the immune response to infection during the acute phase and that this allowed a transient survival of infectious virus in the period after withdrawing therapy. Others have reported that ACV may have a subtle effect on the immune response to HSV antigens in mice when administered from early times during the acute phase of virus replication (16). Furthermore, it has been suggested that ACV treatment causes a smaller rise in antibody production than placebo treatment, and this has been interpreted as an immunosuppressive effect of ACV treatment (3, 6, 12).

In contrast to the results with VACV, it was notable that no recurrences of infectious virus were observed on cessation of treatment in mice given FCV at any of the doses employed, including the lowest dose (0.2 mg/ml), for 10 days. After this suboptimal dose was applied for 5 days, infectious virus was detected in nervous tissues during the period 6 to 11 days p.i. However, this was coincident with the presence of infectious virus in the nervous system tissues of infected, untreated control mice.

The nature of the recurrent infection following cessation of VACV therapy is unknown, but observation of tissue sections obtained from mice infected with a HSV-1 recombinant strain which expresses the lacZ reporter gene suggests that neurons are the source of infectious virus (A. M. Thackray and H. J. Field, unpublished data). Whether or not these observations have any relevance to the chemotherapy of HSV infection in humans is open to question. It is possible that, in the treated murine infection, a form of latency has been established due to the presence of the antiviral drug analagous to that described in the cell culture findings by Wigdahl et al. (29). The authors reported that HSV-1 can be maintained in vitro by the treatment of HSV-infected human cells with various antiviral agents in combination with human leukocyte alpha interferon. However, the latent DNA and pattern of transcription in these infection models were found to differ from those observed in conventional latency in vivo in the absence of antiviral compounds (19).

A difficult observation to explain in the present study is the concordance of positive results obtained from three mice sampled on a particular day, notwithstanding the fact that mice were randomly sampled (according to a predetermined protocol), sometimes from different cages. Cross contamination may be absolutely ruled out by the rigorous controls and coding of samples. There was no consistent relationship between the length of time from dissection to homogenization and inoculation of tissue cultures, and no other systematic factor could be identified. Therefore, it seems likely that the ability to detect infectious virus in the assay must depend on a host factor which is common to the majority of the experimental animals. One such factor could be the circadian cycle. It is possible that at certain times of day all the mice may be more or less susceptible to stress depending on the stage of their circadian rhythm, which runs with a period near to, but usually less than, 24 h (13). Fluctuations in hormonal responses occur during the cycle, and these are reflected in cortisol levels which may have pathophysiological significance (14). In previous work we have shown that cortisone administered topically to HSV-infected mice in the form of 0.5% hydrocortisone cream has a marked effect on virus replication and impaired the clearance of infectious virus from the tissues (1). Although the mice in the present study were on a fixed diurnal light schedule and although we went to much trouble to sample the tissues at the same time and in the same manner each day, the sampling could have been at different points in their circadian cycles, and this might have been reflected in their abilities to clear infectious virus from the tissues on particular days.

The results of the present study add to the body of evidence previously published suggesting that there are significant differences between the effects on HSV pathogenesis of the two nucleoside analogue prodrugs FCV and VACV when the compounds are tested in a particular murine infection model where the ear pinna is the site of primary infection. The mechanisms which underlie these differences remain the subject of intense interest, and we are continuing to pursue further factors that may have a bearing on these differences, including the route of virus inoculation, the age of experimental animals, and possible toxic effects of the compounds. Only when the mechanisms are fully elucidated will it be possible to judge the extent to which our results may be extrapolated to humans.