The Role of Herpes Simplex Virus Type 1 Thymidine Kinase in Pathogenesis

Transcription

1 J. gen. Virol. (1989), 70, Printed in Great Britain 869 Key words: HSV/latency/thymidine kinase The Role of Herpes Simplex Virus Type 1 Thymidine Kinase in Pathogenesis By S. EFSTATHIOU, 1. S. KEMP, 2 G. DARBY 2 AND A. C. MINSON ~ Division of Virology, Department of Pathology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QP and 2Department of Molecular Sciences, Wellcome Research Laboratories, Langley Court, Beckenham, Kent BR3 3BS, U.K. (Accepted 9 December 1988) SUMMARY A genetically engineered herpes simplex virus type 1 (HSV-1) thymidine kinase (TK) deletion mutant has been constructed and used to investigate the role of this gene in pathogenesis. Inoculation of mice with the HSV TK deletion mutant resulted in the establishment of latent ganglionic infection as demonstrated by superinfection of explanted ganglia with wild-type (wt) virus bat not by routine explant culture suggesting that the virus-encoded TK is not essential for the establishment of latent infection but may be necessary for either reactivation or virus replication following reactivation. In addition, Southern blot hybridization has been used to demonstrate in vivo complementation of this mutant by wt virus in both peripheral and central nervous system tissues of mice during acute infection and to show that such complementation can result in the establishment and reactivation of latent TK- infection. INTRODUCTION Several studies have suggested that mutants of herpes simplex virus type 1 (HSV-1) with a defective thymidine kinase gene (TK-) have a reduced ability to establish latent infections in vivo, and that this may be related to a decrease in the ability of such mutants to replicate in tissues of the nervous system (Field & Wildy, 1978; Tenser & Dunstan, 1979; Field & Darby, 1980). More recent studies using genetically engineered TK- deletion mutants have shown that an HSV-2 mutant cannot establish either acute or latent ganglionic infection in mice (McDermott et al., 1984; Tenser & Edris, 1987). Similarly, an HSV-1 mutant was shown to be latency-negative in mice but to establish latent infection in at least a proportion of infected rabbits, suggesting that the host may also be a determinant in the establishment of latent infection (Meignier et al., 1988). Whether failure of TK- virus to cause latent infection is due to a block in establishment, maintenance or reactivation remains unclear since failure to reactivate virus by routine explant culture does not define the stage at which such a block occurs. In this study we have investigated the ability of an HSV-1 (SC16) TK deletion mutant to establish latent infection in mice and have used Southern blot hybridization to follow in vivo complementation of this mutant by wild-type (wt) virus in both peripheral (PNS) and central nervous system (CNS) tissues. We show by the reactivation of latent virus following superinfection of explanted ganglia with wt virus that the TK- deletion mutant is able to establish latent ganglionic infection. However, infection with the mutant was quantitatively different from that caused by wt virus in that virus DNA could not be detected in nervous system tissue during either acute or latent infection. METHODS Viruses. The virus used for both the construction of the TK- deletion mutant SC 16 TKDM21 and for the animal experiments was the HSV-1 strain SC 16 (Hill et al., 1975), an isolate with low passage history. Virus infectivity was measured by plaque titration using BHK-21 C13 cells grown in Eagle's medium containing 10~ (v/v) tryptose phosphate broth and 10~ (v/v) calf serum (ETC) SGM

2 870 S. EFSTATHIOU AND OTHERS (a) (b) 1 2 Pvull 1.3 kb J'K transcript PvulI L 816 ai ~-pm13tki PvulI Pvull 2-7 kb 1 1 t 3 ptkdela 1"3 kb AluI AluI :'~: Fig. 1. (a) Summary of the construction of an 816 bp deletion within the TK-coding region. The insert containing the deleted version of the TK gene was purified following AluI digestion and used to transfect cells with HSV-1 DNA to generate the TK deletion mutant SC16 TKDM21. (b) Southern blot hybridization of BamHI-digested wt SC16 (lane i) and TKDM21 DNA (lane 2) with ptkdela showing the presence of the 2.7 kb deleted BamHI Q fragment in TKDM21 DNA. " Plasmids. The 3-5 kb BamHl Q fragment of HSV- 1 (SC 16) cloned into pbr322 has been described (Darby et al., 1986). A 2043 bp PvulI fragment containing the entire TK-coding sequence was introduced into the Sinai site of M 13mp8 and this clone was designated pm 13TK 1. pm 13TK 1 was digested with the restriction enzymes NruI and Sinai to remove 816 bp of TK-coding sequence. The resultant clone which contained a deleted version of the TK gene was designated ptkdela (Fig. 1 a). Recombination betweenptkdela and HSV-1 DNA. ptkdela was digested with AluI and a 1-3 kb AluI fragment containing the deleted TK gene was gel-purified and cotransfected into BHK cells with high Mr HSV-1 (SC16)- infected cell DNA by the calcium phosphate precipitation method (Graham & van der Eb, 1973) and dimethyl sulphoxide boost (Stow & Wilkie, 1976). Monolayers were left for 4 days and then harvested. Sonicated infected cell suspensions were then assayed on BHK cells maintained in Glasgow modified Eagle's medium and 10% newborn calf serum containing 1 p.g/ml acyclovir (ACV) to select TK- variants. One of 10 virus clones analysed with the restriction enzyme BamHI had the expected 816 bp deletion within the HSV-1 (SC16) BamHI fragment. This virus stock was re-cloned and shown by Southern blot hybridization to have the expected deletion (Fig. 1 b). Virus clone TKDM21 was used in subsequent experiments. Mice. Four to 5-week-old female BALB/c mice were obtained from Tucks U.K. Ltd and were infected at 5 to 6 weeks of age. Inoculation of mice. Twenty ktl of virus suspension containing p.f.u, was inoculated subcutaneously into the left ears of anaesthetized mice (Hill et al., 1975). Measurement of virus titres in mouse tissues. Tissue samples were stored at -70 C. Samples were later homogenized in 0.2 ml of ETC in small glass grinders. The presence of infectious virus was determined by assay on BHK cells. Reactivation of latent virus from dorsal root ganglia. Pooled CII, CIII and CIV cervical dorsal root ganglia were explanted directly into penny bottles containing 0.5 ml of Glasgow modified Eagle's medium supplemented with 1% calf serum. Cultures were then incubated for 5 days at 37 C in a 5% CO2 humidified atmosphere before homogenization and assay for infectious virus. In cases where superinfection of ganglia was used for the reactivation of mutant virus, explanted ganglia were cultured in medium containing l0 s p.f.u, ofwt HSV-1 (SC16) for 5 days before homogenization and assay. DNA extraction. High M r DNA was extracted from mouse tissues and BHK cells as described previously (Efstathiou et al., 1986). Blot hybridizalion. Restriction endonuclease-digested DNA was electrophoresed in 0.8% agarose gels containing 1 ktg/ml ethidium bromide and transferred to nitrocellulose (Schleicher& Schuell) by the technique of Southern (1975). DNA fragments to be used as hybridization probes were labelled with [ct-3zp]dctp (3000 Ci/mmol; Amersham) either by nick translation (Rigby et al., 1977) or by extension of a random primer (Feinberg & Vogelstein, 1983) to a specific activity of 2 x 108 to c.p.m./p.g. Prehybridization and hybridization of filters were performed as described previously (Efstathiou et al, 1986). Autoradiography was carried out at -70 C using preflashed Kodak X-Omat S film and intensifying screens for exposures of 1 to 10 days. One-step growth curves. BHK cell monolayers in 50 mm tissue culture dishes were infected at a multiplicity of 5 p.f.u./ceu. After an adsorption period of I h the cells were washed three times with ETC and incubated at 37 C. Cells were harvested in the medium at various times after infection and stored at - 70 C before sonication and assay on BHK-21 cells.

3 Pathogenicity of an HSV TK mutant I I I I I I x 8 I~ ~ ~ ~. 6 Q I I I I Time post-infection (h) Fig Incubation time (min) Fig. 3 Fig. 2. One-step growth curves ofwt HSV-1 (SC16)(O)andTKDM21 (O) following infection of BHK cells at a multiplicity of 5 p.f.u./cell. Fig. 3. Thymidine kinase assays performed on extracts of mock-infected BU-BHK cells (m), HSV-1 (SC16) (0)- or TKDM21 (O)-infected cells 18 h after infection at a multiplicity of 10 p.f.u./cell. Results are the average of duplicates for each time point. Preparation of HSV-infected cell extracts for TK assay. Confluent monolayers of 5-bromo-2'-deoxyuridineresistant baby hamster kidney (BHK) ceils (BU-BHK), which express no cellular TK, were infected with 10 p.f.u./cell HSV-1 strain SC16 or SC16 TKDM21 and incubated at 37 C. Cells were harvested after 18 h and extracts were prepared as previously described (Larder & Darby, 1982). The sonicates were assayed for TK activity at 37 C as described by Klemperer et al. (1967) using [14C]thymidine as the substrate at 30 gm (1.6 gci/ml). RESULTS Properties of HSV-1 (SC16) recombinant TKDM21 The TK deletion mutant TKDM21 was isolated as described in Methods. The plaque morphology of TKDM21 was identical to the non-syncytial wt SC16 parental virus in BHK cells. One-step growth curves of TKDM21 and of wt HSV-1 (SC16) are shown in Fig. 2. The virus growth curves had similar characteristics with a lag period of 5 h, similar rates of increase of infectious virus between 5 and 12 h and equivalent final virus titres. The TK activities of TKDM21 and wt HSV-1 (SC16) were assayed on extracts from infected TK- BU-BHK monolayers harvested 18 h post-infection. No TK activity was detected in crude extracts prepared from TKDM21-infected cells (Fig. 3). Since ACV can select DNA polymerase mutants (Field et al., 1980) the sensitivity of the TKDM21 virus-encoded DNA polymerase to inhibition by ACV triphosphate was determined and found to be similar to that of wt virus indicating that ACV had not resulted in the selection of a DNA polymerase mutant (data not shown). Pathogenicity studies Five- to 6-week old female BALB/c mice (120) were divided into three experimental groups and inoculated in the left ear pinna with either 5 x 106 p.f.u, ofwt HSV-1 (SC16), 5 x 106 p.f.u. of the deletion mutant TKDM21 or for in viva complementation studies with a 1:1 mixture

4 . 872 S. EESTATHIOU AND OTHERS )- I I I I I I I I! - i I (b) I l l l l l l l l l o o t. ~z 5 4 I I I I I I I l r l l l l t l ' l (c) (d) I I l l l l l! Time post-infection (days) Fig. 4. Detection of HSV in homogenates of (a) ear, (b) cervical dorsal root ganglia, (c) brainstem and (d) spinal cord removed from animals following ear inoculation with either HSV-1 (SC16) (O), TKDM21 (O) or a SC16/TKDM21 mixture (A)- All data are for inocula of 5 x 106 p.f.u, except for data points depicted by <> which refer to animals infected with 10 s p.f.u, of TKDM21. comprising p.f.u, ofwt HSV-1 (SC16) and p.f.u, oftkdm21. The relatively high doses resulted in a 10% mortality in animals inoculated with wt HSV-1 (SC16), a 5~ mortality in animals inoculated with a mixture of wt and mutant virus and no deaths in those animals inoculated with the TK deletion mutant TKDM21. Infectious virus assays were performed on homogenates of ear, left cervical ganglia (CII, CIII and CIV), spinal cord (0.5 cm 3 from the cervical region) and brainstem (pons medulla) from mice killed at 1, 3, 5 and 10 days post-infection (Fig. 4). Peak virus titres in cervical ganglia, brainstem and spinal cord tissue homogenates of animals inoculated with wt HSV-1 (SC 16) were observed 5 days after infection. A similar pattern of virus replication in the nervous system of mice inoculated with a mixture of wt and mutant virus was observed, with peak virus titres detected in cervical ganglia 3 days after infection and peak titres in brainstem and spinal cord tissue homogenates occurring 5 days after infection. In contrast, infectious virus could not be detected in cervical ganglia, brainstem or spinal cord tissue homogenates assayed from animals inoculated with TKDM21 at any time after infection. Comparison of virus titres in the inoculated ears of animals from each experimental group (Fig. 4a) showed lower titres in those animals inoculated with TKDM21 alone as compared to

5 Pathogenicity of an HSV TK mutant 873 (a) I1 2 wt wt/tkdm21 TKDM U RcRcll I U Rc Rc q 3.5 kb kb (b) wt wt/tkdm21 TKDM21 I U RcRcll U RcRcll U Rc Rc Fig. 5. Detection of the wt 3.5 kb SC16 BamHI Q fragment or the 2.7 kb deleted TKDM21 BamHI fragment in (a) individual mouse brainstems (lanes 1 to 5) or (b) individual spinal cords (lanes 1 to 5) and pooled cervical ganglia (lane 6) from wt-, wt/tkdm21- or TKDM21-inoculated animals sampled 5 days after infection. Samples (10 ~tg) of DNA were digested in each case with BamHI, transferred to nitrocellulose and hybridized to ptkdela. U is uninfected whole mouse brain DNA. Rc is one copy/cell reconstructions of wt SC16 DNA or five copy/cell reconstructions of TKDM21 DNA. animals inoculated with either wt virus or a mixture ofwt and mutant virus and is likely to reflect poor replication of the mutant at the site of inoculation. The inability to detect virus in nervous system tissue of animals inoculated with TKDM21 was not related solely to the observed reduction in virus titres at the periphery since an additional group of animals inoculated with 108 p.f.u, of TKDM21 in the left ear pinna had peripheral virus titres equivalent to wt virusinoculated animals but infectious virus remained undetectable in nervous system tissue assayed 5 days after infection (Fig. 4). These results show that the TK deletion mutant TKDM21 is significantly less neurovirulent than wt virus. Since this block did not appear to be related to peripheral virus titres during acute infection, the inability to detect virus in the nervous system is more likely to reflect a true neuropathogenic attenuation than merely a block in the ability of virus to enter the nervous system. Southern blot hydridization was used next to determine whether the inability to detect infectious virus in the nervous system of mice inoculated with TKDM21 was reflected in the amount of virus DNA present in such tissues and to assess complementation of TKDM21 following infection of animals with a mixture of wt and mutant virus. Detection of HSV DNA in mouse neural tissue during acute infection DNA was extracted from brainstems, spinal cords and cervical ganglia of five mice from each experimental group at 5 days post-infection. The DNA was digested with BamHI and analysed by Southern hybridization using the ptkdela fragment as probe. Fig. 5 shows the detection of the 3-5 kb HSV-1 (SC16) BamHI Q fragment in brainstem, cervical ganglia and spinal cord DNA of animals inoculated with wt SC 16. The amount of HSV DNA detected in these samples varied from 0.1 to five copies/cell in individual mice assessed by comparison with the one

6 S. EFSTATHIOU AND OTHERS wt wt/tkdm21 TKDM U RcRcll U RcRcll U Rc Rc kb kb Fig. 6. Detection of the wt 3.5 kb SC16 BamHI Q fragment or the 2.7 kb deleted TKDM21 BamHI fragment in individual mouse brainstems (lanes 1 to 5) or pooled cervical ganglia (lane 6) from wt-, wt/tkdm21- or TKDM21-inoculated animals sampled 16 weeks post-infection. Samples (10 lag) of DNA were digested in each case with BamHI, transferred to nitrocellulose and hybridized to ptkdela. U is uninfected whole mouse brain DNA. Re is one copy/cell reconstructions of either wt SC16 or TKDM21 DNA. One brainstem DNA sample contained a high Mr fragment (O) the nature of which is not known but could be due to either a partial digestion product or an unrelated contaminating DNA species. copy/cell reconstruction. Southern blot analysis of brainstem DNA extracted from animals inoculated with a mixture of wt and mutant virus revealed the presence of both the wt 3.5 kb BamHI Q fragment and the 2-7 kb deleted BamHI Q fragment in approximately equimolar amounts, corresponding to one copy/cell. Analysis of spinal cord and cervical ganglia from this group of animals revealed the presence of both wt and deleted BamHI Q fragments although a reduction in the amount of the deletion mutant BamHI Q fragment was observed. This perhaps reflects a reduced efficiency in complementation in these tissues. Virus DNA was never detected in brainstem, cervical ganglia or spinal cord samples from mice inoculated with the mutant TKDM21. Similarly, we failed to detect HSV DNA in either brainstem or cervical ganglia in five animals that had been inoculated with 108 p.f.u, of TKDM21 (data not shown). This result is consistent with our inability to detect infectious virus in tissues from equivalent animals inoculated with TKDM21. Detection of HSV DNA in mouse neural tissue during latent infection Sixteen weeks after infection when latent infections had been established five animals from each experimental group were killed and DNA was extracted from brainstems of individual mice and from pooled cervical ganglia (CII, CIII and CIV). Ten ~tg of DNA from each sample was digested with BamHI and was analysed by Southern hybridization using the ptkdela fragment as probe. The 3-5 kb SC16 BamHI Q fragment was detected in all brainstem and cervical ganglia DNA samples from animals inoculated with wt virus. The amounts corresponded to approximately one copy/cell assessed by comparison with the reconstruction, with little mouse-to-mouse variation in copy number (Fig. 6). Confirmation of the latent status of the virus DNA was made by re-probing filters with the BamHI K junction fragment which detected endless DNA characteristic of the latent state (Rock & Fraser, 1983 ; Efstathiou et al., 1986) (data not shown). Southern blot analysis of DNA extracted from brainstem and pooled cervical ganglia from animals inoculated with a mixture of wt and mutant virus revealed the presence of both the wt 3.5 kb BamHI Q fragment and the deleted 2-7 kb restriction fragment indicating that wt helper function had enabled mutant TKDM21 to establish latent infection. Virus DNA could not be detected in DNA extracted from individual brainstems, or pooled cervical ganglia from five mice inoculated with the mutant TKDM21, a result consistent with the inability to detect virus DNA in the nervous system of these mice during the acute disease. Explant culture of cervical ganglia (CII, CIII, CIV) of five mice from each experimental group resulted in virus reactivation from all five mice inoculated with wt virus, four of five mice inoculated with a mixture of wt and mutant virus and none of the 10 mice inoculated with either

7 Pathogenicity of an H S V TK mutant wt wt/tkdm I 3.5 kb ~ 2-7 kb Fig. 7. Southern blot analysis of reactivated virus from cervical ganglia of five mice inoculated with either wt SC16 or a wt/tkdm21 mixture (lanes 1 to 5). Explanted ganglia were incubated for 5 days and then homogenized and assayed on BHK cells. BHK cell DNA (10 ~tg)was in each case digested with BamHI and analysed by Southern hybridization using ptkdela as probe to detect both the wt 3-5 kb BamHI Q and deleted 2-7 kb BamHI fragment p.f.u, or 108 p.f.u, of mutant TKDM21. D N A extracted from monolayers infected with reactivated virus was analysed by Southern blot hybridization (Fig. 7). The wt 3.5 kb BamHI Q fragment was detected in monolayers infected with reactivated virus from five mice infected with wt SC16. Both the wt 3.5 kb BamHI Q fragment and deleted 2.7 kb fragment were detected in virus reactivated from four of five mice infected with a mixture of wt and mutant virus although a greater amplification of wt virus was reflected by the relative amounts of wt and mutant virus D N A, indicating considerable selection for wt virus during the reactivation procedure. Reactivation of mutant TKDM21 following superinfection of latently infected ganglia The inability to detect the T K - deletion mutant TKDM21 in nervous system tissue either by direct homogenization and assay during acute infection or by Southern blot hybridization during both acute and latent infection and the failure to reactivate virus from cervical ganglia by explant culture suggested either that this virus could not establish a latent infection or that the presence of small amounts of virus during both acute and latent infection failed to be detected. Superinfection of human ganglia with temperat.ure-sensitive (ts) mutants has been used in the detection of non-reactivable or defective HSV D N A (Brown et al., 1979; Lewis et al., 1984) presumably by a process of complementation and/or recombination. Using a similar approach, superinfection of explanted ganglia with wt SC16 was used in an attempt to reactivate virus from dorsal root ganglia of TKDM21-infected animals. Explanted cervical ganglia (CII, CIII, CIV) from five mice inoculated 16 weeks previously with mutant TKDM21 were placed in medium containing 105 p.f.u, of wt SC16. Following incubation at 37 C for 5 days tissue homogenates were used to infect B H K cell monolayers. D N A extracted from infected

8 876 (a) 1 S. EFSTATHIOU AND OTHERS Rc Rc (b) "5 2.7 Fig. 8. (a) Southern blot analysis of virus from the ganglia of TKDM21-inoculated animals removed during the latent stage of infection and superinfected with wt S C 16. Superinfected cervical ganglia from five mice (lanes 1 to 5) were homogenized and assayed on BHK cells. BHK cell DNA (10 gg) was in eacb case digested with BamHI and analysed by Southern hybridization using ptkdela as probe. Rc is a 10 copy/ceu reconstructions of either wt SC16 or TKDM21 DNA. (b) Virus obtained from the superinfected ganglia from one mouse (lane 5) was further analysed by Southern hybridization following one passage in medium containing 1 ~tg/ml ACV to reduce the amount of wt SC16 superinfecting virus (b, lane 3). Lane l, TKDM21; lane 2, SC16. monolayers was analysed by Southern blot hybridization using ptkdela as probe (Fig. 8). Although a faint band corresponding in size to the 2-7 kb deleted BamHI Q fragment was detected in each set of ganglia from five mice, the weak intensity of these bands made the unambiguous identification of TKDM21 virus DNA difficult. For this reason ACV enrichment of one reactivant (SIJ; Fig. 8 a, lane 5) was carried out. Fig. 8 clearly shows the presence of the mutant virus 2.7 kb deleted BamHI Q fragment with a concomitant decrease in the wt virus 3.5 kb BamHI Q indicating the successful reactivation of mutant virus TKDM21. The ability to reactivate the TK deletion mutant TKDM21 following superinfection with wt virus indicates that this mutant is able to establish and maintain latency in the absence of detectable virus replication during acute infection. This observation is in agreement with previous studies which have shown that HSV ts mutants unable to replicate at the body temperature of the mouse are able to establish latent infection (McLennan & Darby, 1980; Watson et al., 1980). Although the inability to reactivate virus following routine explant culture suggests a block in reactivation, it is possible that superinfection is a more sensitive method by which to detect reactivation from low levels of latent virus. DISCUSSION This paper describes studies on the pathogenesis of an HSV-1 (SC16) TK deletion mutant, TKDM21. As expected, this engineered deletion mutant, lacking 816 bp of TK-coding sequence, expresses no TK activity, grows normally in tissue culture and has the same nonsyncytial plaque morphology as the parental SC16 virus. It is of interest that two HSV-1 (strain 17) TK deletion mutants previously described by Sanders et al. (1982) had an observed change in plaque morphology from non-syncytial to a small syncytial phep.otype and grew poorly in tissue culture suggesting that a second gene function was affected in these mutants. This is most likely due to the fact that these mutants had large deletions covering much of the TK gene and its promoter which would have disrupted the overlapping reading frame UL24 (McGeoch et al., 1988) the product of which has been shown to be important but not essential for virus growth in tissue culture (D. M. Coen, personal communication). Mutant TKDM21 was constructed such

9 Pathogenicity of an HSV TK mutant 877 that 292 bp downstream of the initiating ATG of the HSV TK gene was retained. Of the two in vitro and in vivo initiation transcription sites that could potentially encode UL24 (Read et al., 1984) the transcript initiating proximal to the first initiating methionine of UL24 is maintained in mutant TKDM21 and is therefore most likely to be involved in the expression of this gene product. Mutant TKDM21 replicated less efficiently in the ears of infected mice and could not be detected in either sensory ganglia or brainstem tissue homogenates by conventional virus assay techniques, a result consistent with previous studies of TK- mutants (Field & Wildy, 1978; Tenser & Dunstan, 1979; Klein et al., 1981; Price & Khan, 1981). These had suggested that virus-encoded TK is required for replication in non-dividing neuronal cells and for efficient replication in epidermal cells involved early in the infection. Southern blot hybridization analysis failed to detect TKDM21 virus DNA during either acute or latent infection and attempts to reactivate virus from sensory ganglia by routine explant culture proved unsuccessful, an observation that has been made for both HSV-1 and HSV-2 TK deletion mutants following experimental infection of mice (McDermott et al., 1984; Tenser & Edris, 1987; Meignier et al., 1988). Recovery of TKDM21 from explanted ganglia following superinfection with wt virus indicated that the TK deletion mutant used in this study had established a latent infection and supports the view that virus-encoded TK is not an essential requirement for the establishment of latent infection. Whether the virus-encoded TK plays an essential role in the reactivation process is more difficult to determine and further work is needed to show whether this function is required for the reactivation process or for virus replication following reactivation. Clearly at a quantitative level the inability to detect virus DNA during either acute or latent infection contrasts with the results obtained with wt-inoculated animals in which virus DNA was readily detected. In vivo complementation of TKDM21 was followed through the PNS and CNS by Southern blot hybridization following infection of mice with a 1:1 mixture of wt TK + virus. The deleted 2.7 kb BamHI Q fragment present in the TKDM21 genome provided a marker which allowed the separation of the TK + and TK- components necessary for the evaluation of TK complementation. The presence of TK- virus during mixed infection had little effect on virus titres at the periphery or in the PNS or CNS of mice during the acute stage of disease and Southern blot analysis of brainstem, cervical ganglia and spinal cords of animals revealed the presence of both wt TK + and TK- virus DNA in all tissues (Fig. 5). Similarly, during latent infection both wt TK and TK- virus DNA could be detected. However, by comparison to reconstructions it appears that the ratio of mutant to wt virus DNA was greater in acute than in latent infection (compare Fig. 5 a to Fig. 6) suggesting that complementation between TK and TK- virus may be less efficient during the establishment of latent infection. Successful reactivation of TKDM21 from ganglia containing both wt TK and TK- virus was possible although the small amounts of TKDM21 virus DNA detected following reactivation suggested a reduced efficiency of complementation during this process. One possibility which may explain this observation is that efficient complementation is multiplicity-dependent in vivo in that initial infection involving high doses of both wt TK + and TK- results in efficient complementation at the site of inoculation but that the lower titres detected during acute infection of both the PNS and CNS could influence the numbers of cells co-infected with wt TK + and TK- virus and thus the number of cells destined to become latently infected of which only a proportion may be co-infected with these viruses. It is at present impossible to know the proportion of latently infected cells containing both wt TK + and TK- virus although it seems likely that latent infection of neurons with both TK + and TK- viruses does occur. A number of previous studies demonstrating in vitro complementation/recombination of either TK+/TK - virus mixtures or between avirulent viruses have been published (Tenser et al., 1981 ; Field & Lay, 1984; Javier et al., 1986; Tenser & Edris, 1987; Sedarati et al., 1988) however, the results presented here are the first to follow complementation of a deletion mutant within the CNS and PNS during both acute and latent infection by Southern blot hybridization. It is of interest that studies on the sensitivity of clinical isolates to ACV from early clinical trials with this compound found a number of the isolates to contain mixtures of resistant and sensitive viruses (Dekker et

10 878 S. EFSTATHIOU AND OTHERS al., 1983) and that similar observations have been made in experimental infections of mice (Field & Lay, 1984). These results suggest that in vivo complementation between TK + and TKviruses may be a mechanism by which TK- drug-resistant virus could be maintained in the population. In conclusion we have demonstrated by superinfection of explanted ganglia with wt virus that a TK- deletion mutant is able to establish latent ganglionic infection suggesting that the virusencoded TK is not essential for the establishment of latent infection. However, it may be necessary for either reactivation or virus replication following reactivation. In addition we have followed the in vivo complementation of a TK- deletion mutant by Southern hybridization and have confirmed previous reports (Tenser et al., 1985; Tenser & Edris, 1987) which had shown that such complementation can result in the establishement and reactivation of latent TK- virus infection. This work was supported by the Medical Research Council, U.K. REFERENCES BROWN, S. M., SUBAK-SHARPE, J. H., WARREN, K. G., WROBLEWSKA, A. & KOPROWSKI, H. (1979). Detection of complementation of defective or uninducible (herpes simplex type 1) virus genomes latent in human ganglia. Proceedings of the National Academy of Sciences, U.S.A. 76, DARBY, G., LARDER, B. A. & INGLIS, M. M. (1986). Evidence that the 'active centre' of the herpes simplex virus thymidine kinase involves an interaction between three distinct regions of the polypeptide. Journal of General Virology 67, DEKKER, C., ELLIS, M. N., McLAREN, C., HUNTER, G., ROGERS, J. & BARRY, D. W. (1983). Virus resistance in clinical practice. Journal of Antimierobial Agents and Chemotherapy 12, supplement B, EFSTATHtOU, S., MINSON, A. C., FIELD, H. J., ANDERSON, J. R. & WILDY, P. (1986). Detection of herpes simplex virusspecific DNA sequences in latently infected mice and in humans. Journal of Virology 57, 44~455. FEINBERG, A. D. & VOGELSTEIN, B. (1983). A technique for radiolabeling DNA restriction endonuclease fragments to high specific activity. Analytical Biochemistry 132, FIELD, H. J. & DARBY, G. (1980). Pathogenicity in mice of strains of herpes simplex virus which are resistant to acyclovir in vitro and in vivo. Antimicrobial Agents and Chemotherapy 17, FIELD, H..L & LAY, E. (1984). Characterization of latent infections in mice inoculated with herpes simplex virus which is clinically resistant to acyclovir. Antiviral Research 4, FIELD, ~. J. & WILDY, P. (1978). The pathogenicity of thymidine kinase-deficient mutants of herpes simplex virus in mice. Journal of Hygiene 81, FIELD, H. J., DARBY, G. & WILDY, P. (1980). Isolation and characterization of acyclovir-resistant mutants of herpes simplex virus. Journal of General Virology 49, GgAn~, F. L. & VAN DER EB, A. J. (1973). A new technique for the assay of infectivity of human adenovirus 5 DNA. Virology 52, HILL, T. J., FIELD, H. J. & BLYTH, W. A. (1975). Acute and recurrent infection with herpes simplex virus in the mouse: a model for studying latency and recurrent disease. Journal of General Virology 28, JAVlErt, R. T., SEDARATI, F. & STEVENS, J. G. (1986). Two avirulent herpes viruses interact in the animal to generate virulent recombinants and a lethal infection. Science 234, KLEIN, R. J., FRIEDMAN-KILN, A. E. & DE STEFANO, E. (1981). Pathogenesis of experimental skin infections induced by drug resistant herpes simplex virus mutants. Infection and Immunity 34, KLEMPERER, H. G., HAYNES, G. R., SHEDDEN, W. I. H. & WATSON, D. H. (1967). A virus-specific thymidine kinase in BHK-21 ceils infected with herpes simplex virus. Virology 31, LARDER, B. A. & DARBY, G. (1982). Properties of a novel thymidine kinase induced by an acyclovir-resistant herpes simplex virus type 1 mutant. Journal of Virology 42, LEWIS, M. E., BROWN, S. M., WARREN, K. G. & SUBAK-SHARPE, J. H. (1984). Recovery of herpes simplex virus genetic information from human trigeminal ganglion cells following superinfection with herpes simplex virus type 2 temperature-sensitive mutants. Journal of General Virology 65, McDERMOTT, M. R., SMILEY, J. R., BRAIS, L. J., RUDZROGA, H. E. & BIENENSTOCK, J. (1984). Immunity in the female genital tract after intravaginal vaccination of mice with an attenuated strain of herpes simplex virus type 2. Journal of Virology 51, MC'GEOCH, D. J., DALRYMPLE, M. A., DAVISON, A. J., DOLAN, A., FRAME, M. C., McNAB, D., PERRY, L. J., SCOTT, J. E. & TAYLOR, P. (1988). The complete DNA sequence of the long unique region in the genome of herpes simplex virus type 1. Journal of General Virology 69, McLENNAN, J. L. & DARBY, G. (1980). Herpes simplex virus latency: the cellular location of virus in dorsal root ganglia and the fate of the infected cell following virus activation. Journal of General Virology 51, MEIGNIER, B., LONGNECKER, R., MAVROMARA-NAZOS, P., SEARS, A. E. & ROIZMAN, B. (1988). Virulence of and establishment of latency by genetically engineered deletion mutants of herpes simplex virus 1. Virology 162,

11 Pathogenicity of an HSV TK mutant 879 PRICE, R. W. & KHAN, A. (1981). Resistance of peripheral autonomic neurons to in vivo productive infection by herpes simplex virus mutants deficient in thymidine kinase activity. Infection and Immunity 34, READ, G. S., SHARP, J. A. & SUMMERS, W. C. (1984). In vitro and in vivo transcription initiation sites on the TKencoding BamH1 Q fragment of HSV-1 DNA. Virology 138, RIGEY, P. W. J., DIECKMANN, M., RHODES, C. & BERO, P. (1977). Labelling deoxyribonucleic acid to high specific activity in vitro by nick translation with DNA polymerase I. Journal of Molecular Biology 113, ROCK, D. L. & FRASER, N. W. (1983). Detection of HSV-1 genome in central nervous system oflatently infected mice. Nature, London 302, SANDERS, P. G., WILKIE, N. M. & DAVISON, A. J. (1982). Thymidine kinase deletion mutants of herpes simplex virus type 1. Journal of General Virology 63, SEDARATI, F., JAVIER, R. T. & STEVENS, J. G. (1988). Pathogenesis of a lethal mixed infection in mice with two nonneuroinvasive herpes simplex virus strains. Journal of Virology 62, SOUTHERN, E. M. (1975). Detection of specific sequences among DNA fragments separated by gel electrophoresis. Journal of Molecular Biology 98, STOW, N. D. & WILKIE, N. M. (1976). An improved technique for obtaining enhanced infectivity with herpes simplex virus type 1 DNA. Journal of General Virology 33, TENSER, R. B. & DUNSTAN, M. E. (1979). Herpes simplex virus thymidine kinase expression in infection of the trigeminal ganglion. Virology 99, TENSER, R. B. & EDRIS, W. A. (1987). Trigeminal ganglion infection by thymidine kinase-negative mutants of herpes simplex virus after in vivo complementation. Journal of Virology 61, TENSER, R. B., RESSEL, S. & DUNSTAN, M. E. (1981). Herpes simplex virus thymidine kinase expression in trigeminal ganglion infection: correlation of enzyme activity with ganglion virus titre and evidence of in vivo complementation. Virology 112, TENSER, R. B., JONES, J. C. & RESSEL, S. J. (1985). Acute and latent infection by thymidine kinase mutants of herpes simplex virus type 2. Journal of Infectious Diseases 151, WATSON, K., STEVENS, J. G., COOK, M. L. & SUBAK-SHARPE, J. H. (1980). Latency competence of thirteen HSV-1 temperature-sensitive mutants. Journal of General Virology 49, (Received 2 November 1988)