Reactivation of latent herpes simplex virus from explanted dorsal root ganglia

Transcription

1 Journal of General Virology (1994), 75, Printed in Great Britain 2017 Reactivation of latent herpes simplex virus from explanted dorsal root ganglia Marion Ecob-Prince* and Karen Hassan University of Glasgow, Department of Neurology, Institute of Neurological Sciences, Southern General Hospital, Glasgow G51 4TF, U.K. Reactivation was induced by explantation of dorsal root ganglia (DRG) from mice that were latently infected with herpes simplex virus type 1. Reactivation was first detected, using combined in situ hybridization and immunocytochemistry, at 2 or 3 days post-explantation (p.e.). Evidence of reactivation was found primarily in neurons that did not also contain latency-associated transcripts (LATs). Occasionally, mrna of immediate early gene 2 (IE2) or IE4/5, in the absence of other viral mrnas or antigen, was found in LAT + neurons. Thus, if reactivation was occurring in LAT + neurons, the LATs must have been lost as an early consequence; however we could detect neither a decrease in the percentages of LAT + neurons nor a reduction in the intensity of the LAT signal during the period of reactivation. However, the number of foci of reactivation was generally less than 2'9% of the estimated number of LAT + cells in the DRG; this may account for our failure to see such changes. A redistribution of the LATs into the cytoplasm was found in some cells but this could reflect the poor survival and consequent death of the explanted neurons. We conclude that the majority of LAT + neurons did not reactivate on explantation and that if reactivation occurred only in LAT + neurons, the LATs must have been removed from the nucleus as an early consequence of reactivation. Alternatively, there may be a population of latently infected cells that do not express LATs. Introduction Subcutaneous injection of herpes simplex virus type 1 (HSV-1) into the footpad of a mouse results in the establishment of latently infected neurons in dorsal root ganglia (DRG). The only virus-specific transcription known to occur in such latently infected cells is that of the latency-associated transcripts (LATs) but their function remains unclear (for review, see Fraser et al., 1992). The three collinear LATs (2-0 kb, l'5kb and 1.45 kb) accumulate in the nuclei of latently-infected neurons; the 2.0 kb LAT is also found at low levels during lytic replication in tissue culture cells (Spivack & Fraser, 1987, 1988; Wagner et al., 1988). The LATs are known to overlap the 3' terminus of the IE1 sequence and are transcribed from the complementary strand of DNA (Rock et al., 1987; Spivack & Fraser, 1987; Stevens et al., 1987). Several hypotheses have been advanced to suggest a function for the LATs. The 2 kb major LAT may be an intron (Dobson et al., 1989; Farrell et al., 1991) spliced from a larger 8-5 kb transcript (Mitchell et al., 1990; Zwaagstra et al., 1990). On the other hand, the 2 kb and 8"5 kb LATs may represent alternative transcription units, the 2 kb LAT itself being spliced into smaller forms (Wagner et al., 1988; Wechsler et al., 1988) thus revealing potential open reading frames (ORFs; Spivack et al., 1991). Antibody to this putative LAT-associated polypeptide recognizes antigen in latently-infected neurons in tissue culture (Doerig et al., 1991) but there is no report ofa LAT-associated protein in vivo. It was also postulated that LATs might function as an antisense inhibitor of IE1 (Stevens et al., 1987) and the 2 kb LAT has been shown to inhibit IE1 trans-activation in vitro (Farrell et al., 1991). Mutant viruses, unable to express LATs, are able to establish, maintain and reactivate from latent infections, so there is no absolute requirement for LAT expression. However LAT expression is generally regarded as a reliable marker for the presence of latent virus, particularly in peripheral neurons. Although viruses unable to express the 2 kb LAT were able to reactivate on explantation of the DRG (Block et al., 1990; Ho & Mocarski, 1989), some did so with delayed kinetics (Hill et al., 1990; Leib et al., 1989; Steiner et al., 1989; Trousdale et al., 1991). This led to the suggestion that expression of LATs potentiated reactivation. Using more discrete lesions, recent studies have shown that although lesions in the TATA element of the LAT promoter either reduced (Rader et al., 1993) or negated (Deshmane et al., 1993) the expression of SGM

2 2018 M. Ecob-Prince and K. Hassan LATs, reactivation was normal following explantation of trigeminal ganglia. Conversely, a deletion of the CRE element of the LAT promoter reduced reactivation without affecting expression of LATs (Rader et al., 1993). However, the importance of the LATs themselves may depend on the site of latency (Sawtell & Thompson, 1992) or whether reactivation was induced in vivo or alter explantation in vitro (Hill et al., 1990; Trousdale et al., 1991), Although explantation is an accepted method for studying reactivation of latent virus, the results are difficult to interpret. A DRG with fewer sites of reactivation will release less virus which, in turn, will take longer to be detected: it does not mean that reactivation is delayed. It is also unclear whether reactivation occurs in all latently infected cells. Such questions can be answered only at the level of the individual cell and only then can we determine whether explantation is a good model for reactivation of latent virus. In our previous studies, we attempted to investigate reactivation in vivo following neurectomy of the sciatic nerve (Ecob-Prince et al., 1993b). Although there was evidence of IE gene expression 2 to 4 days after the operation, virus-specific antigen was not detected in the cells and, on explantation, virus reactivation in the explanted DRG was similar to that of unoperated DRG. This raised the question of whether neurectomy had, in our hands, been of a sufficient stimulus as to cause reactivation of latent virus, resulting in lytic replication and subsequent loss of latent sites, as demonstrated by others (McLennan & Darby, 1980; Tenser et al., 1988). We have now investigated reactivation, at the level of individual cells, following explantation of latently infected DRG into culture. Explantation involves a severe 'neurectomy' by which the neurons lose most of their axonal cytoplasm as well as trophic influences from both the peripheral target and the circulatory system. The stimulus is known to reactivate latent virus and should be the same for all cells in the DRG. Previously we had shown that the mutant in1814 (Ace et al., 1988) established a higher percentage of LAT + neurons than did either its rescuant (1814R) or the wild-type HSV-1 (wt HSV-1) from which it was derived (Ecob-Prince et al., 1993a). Nevertheless, despite the higher number of LAT + cells, the number of cells in which reactivation occurred either in vivo or in vitro at the level of explanted DRG was not increased accordingly (Ecob-Prince et al., 1993b). Therefore we used all three viruses in this study to investigate whether, at the cellular level, there was a correlation between the number of LAT + neurons and the number of cells in which virus could reactivate. We have investigated whether explantation stimulated reactivation in all latently infected cells, whether reactivation occurred in a synchronous fashion and whether reactivation was in LAT neurons. We have also considered whether explantation of latently infected DRG is itself a good model of reactivation of HSV. Methods Virus stocks and inoculation of mice. All virus stocks were produced and assayed in BHK-21 cells grown in Eagle's MEM supplemented with 10% newborn calf serum (NCS). The wt HSV-1 used in these studies was the Glasgow strain 17 + (Brown et al., 1973) taken from an elite stock with a low passage number. The insertion mutant in1814 (which lacks the trans-activation function of Vmw65) and its revertant 1814R have been described previously (Ace et al., 1989). Virus stocks were diluted to 4x 106p.f.u./ml in PBS containing 10% NCS and ml (105 p.f.u./mouse) was inoculated subcutaneously in the right footpad of 3 to 4-week-old male BALB/c mice. Reactivation in vitro. Six to 12 weeks after inoculation, DRG from thoracic level 13 (T13) to lumbar level 6 (L6) were dissected out and explanted separately into the wells of a 96-well plate. They were cultured for 1, 2, 3, 4, 5, 7 or 21 days at 37 C in BHK medium containing 10% NCS, 10% fetal calf serum (FCS), antibiotics and 2-5 gg/ml amphotericin B. Supernatants from DRG of 10 mice were removed at each time point to be assayed in the presence of 3 mmhexamethylene bisacetamide (HMBA; Sigma) for release of reactivated virus. In the presence of HMBA, in1814 plaques on BHK cells with the same efficiency as wt HSV-1 (McFarlane et al., 1992) but the drug has no effect on the efficiency of plaque formation by wt HSV-1 and so was used routinely in the assay, irrespective of the virus type. Alternatively, cultured DRG were also fixed at the daily intervals in 4% paraformaldehyde, embedded and serially sectioned (5 Ixm) onto slides coated with aminopropyltriethoxysilane (Maddox & Jenkins, 1987), as previously described (Ecob-Prince et al., 1993a), to detect virus-specific antigen and mrna produced by virus reactivating in the explanted DRG. For these studies, the DRG from two or three mice were used in each of two experiments for each of the early time points, giving a total of four to six mice for each virus at 0, 1, 2, 3 and 4 days postexplantation (p.e.). Two mice were used for each of the later time points. The results from observation of serially sectioned DRGs from a total of 85 mice are presented. Immunocytochemistry. Paraffin sections were rehydrated to PBS and reacted with a rabbit polyclonal antibody to HSV (Dakopatts) which was visualized using a peroxidase ABC reaction (Vectastain). All antisera were pretreated with RNase inhibitor to prevent loss of the signal detected in the subsequent in situ hybridization (ISH). Every section for each of three mice at 2 (wt HSV-1) or 3 (in1814 or 1814R) days p.e. was stained for virus antigen and used to count directly the number of single cells or loci of cells that were antigen-positive. Probes. Plasmids containing fragments of IE1/LAT and UL18 in pt7t3 19U were kindly provided by Dr Frazer Rixon. The riboprobe for LAT and IE1 mrna has been described before (Ecob-Prince et al., 1993a). The 2563 bp fragment contains the sequence between the Sail and BamHI restriction sites [nucleotides to according to Perry & McGeoch (1988)] of the BamHI B fragment of wt HSV-1 DNA. In the sense direction it recognizes the 2 kb LAT and in the antisense direction it recognizes IE1 mrna. A 1340 bp fragment containing the UL18 ORF was released from pmj521 by digestion with BamHI and Sail and ligated into a BamHI-SalI-digested pt3t7 19U vector to give pt318 (Nicholson, 1992). Plasmids containing fragments of IE2, IE4/5 and Vmw65 in pt3t7 18U (Pharmacia) were kindly provided by Dr Chris Preston. Plasmid pt7t3 IE2 contains the BamHI (position ) to Sspl (115246) fragment of IE2; pt7t3 IE5

3 Reactivation of latent HSV 2019 contains the BamHI (position ) to EeoRI (146693) fragment of IE5 but may also hybridize to the 3' end of IE4 and is therefore designated as an IE4/5 riboprobe; pt7t3 TIF contains the Vmw65 coding sequences cloned as an EcoRI-HindIII fragment from pmc 17 (Ace et al., 1988). Plasmids were linearized and transcribed in the presence of [3~S]UTP using the appropriate restriction enzyme and polymerase, treated by alkaline hydrolysis and diluted in 10 mm-dtt to give 1 ng RNA/kb/ktl. In situ hybridization (ISH). After immunocytochemistry, serial sections were each reacted with a different probe to determine the coincident expression of different genes in the same cell. The percentage of LAT + cells was calculated by counting the total number of LAT + nuclei and dividing it by the total number of neuronal nuclear profiles in at least 10 sections, involving at least 1000 neuronal nuclei. The methods used have been described previously (Ecob-Prince et al., 1993 a; Griffiths et al., 1989). Briefly, sections were rehydrated, refixed in 4% paraformaldehyde, treated with proteinase K (20 gg/ml for 10min at 37 C), followed by acetic anhydride (0.25% in 0.1 M- triethanolamine for 10 min at room temperature) and dehydrated again before adding the probe. The probe was diluted 10-fold in stock hybridization mix (50% formamide, 10% dextran sulphate, 20 mm- Tris-HCl ph 8-0, 0-3 M-NaC1, 5 mm-edta, 10 ram-sodium phosphate buffer ph 8-0, 0.5 mg/ml yeast trna and 1 x Denhardt's solution) and then heated at 80 C for 2min before quenching on ice. Hybridization was done at 50 C for 18 to 21 h. Slides were then washed at 65 C in 50% formamide in 2 x SSC (0'3 M-NaC1, 0.03 M- trisodium citrate) with 0.1 M-DTT for 20 min, digested with RNase A (20 gg/ml in 0-5 M-NaC1 with 10mM-Tris-HCl ph 7-5 and 5 mm- EDTA) for 30 min at 37 C, washed in 50% formamide again and then washed in decreasing concentrations of SSC before being dehydrated in alcohols containing 0-3 M-ammonium acetate. Dried slides were coated with Ilford K5 nuclear emulsion and exposed for 6 days at 4 C before being developed in D19 and fixed in Hypam. Slides were then counterstained with haematoxylin, dehydrated and mounted. Histology. Stains for cresyl fast violet and Palmgren's silver impregnation were done as routine preparations (Disbrey & Rack, 1970). Results DRG from latently-infected animals were explanted into culture and the onset of reactivation was investigated in two ways. At daily intervals, supernatant medium was removed and assayed for the presence of cell-free virus. In addition, also at daily intervals, DRG were fixed, embedded in paraffin and serially sectioned to investigate the location and distribution of virus-specific antigen (Fig. 1) and mrna in individual neurons. Although the presence of virus-specific antigen could be assessed in every section, it was not possible to determine a full profile of the mrnas present in each cell. An individual neuron could occasionally be followed through as many as seven sections but was usually present in three to five sections, so it was not possible to hybridize all the riboprobes to each cell. Moreover, because LATs are confined to the nucleus in latently-infected neurons, a cell could not be scored for the presence of LATs unless the section hybridized with the LAT riboprobe went through the nucleus of that particular cell. If other mrnas were in similarly restricted areas, their presence could also be missed. An absence of hybridization, therefore, did not always indicate an absence of mrna in that particular cell. This meant that we were able to investigate only positive associations of one RNA with another and were usually unable to determine whether all six RNAs were present in one cell although their individual associations with antigen could be assessed in every section. There was no evidence of spontaneous reactivation in any of the latently infected mice examined and no evidence of reactivation during the first day in culture of DRG from animals infected with wt HSV-1, in1814 or 1814R (Table 1). Reactivation was first detected in DRG latently-infected with wt HSV-1 where virus-specific mrna with or without antigen was found in sections at 2 days p.e., most often in L4 or L5 DRG. This involved mainly single neurons but also involved spread to closely associated non-neuronal cells such as satellite cells around the neurons and Schwann cells in the nerve root. By 3 days p.e., such foci were larger and more numerous involving adjacent neurons, and cell-free virus was often detected in the supernatant of the DRGs which contained them. Reactivation of virus in DRGs from animals latently infected with in 1814 or 1814R was qualitatively indistinguishable from that of animals latently infected with wt HSV-1, but was delayed, being detected first in the sections at 3 days p.e. and in the supernatants at 4 days p.e. By 10 days p.e., 90 % or 85 % of DRG latently infected with in1814 or 1814R, respectively, had reactivated, compared to 80% of DRG with wt HSV-1. Reactivation (judged by the expression of virus-specific mrna and antigens associated with lytic replication) at either 2 (wt HSV-1) or 3 (in1814, 1814R) days p.e. was found primarily in neurons that did not contain LATs (Fig. 2a, b). Most (99 %) of the LAT + neurons (Fig. 2a) showed no evidence of reactivation. In these LAT neurons, various mrnas could be found, sometimes in association with virus antigen accumulation. The mrna for IE2 was expressed alone, particularly in antigennegative cells, but could otherwise be found in antigenpositive cells that might also contain IE4/5 mrna. The mrna of IE4/5 could be found in isolation or with IE2 mrna, but in about half of the cells it was co-expressed with mrna for Vmw65, IE1 or ULI8, particularly in antigen-positive single cells, in foci of antigen-expressing cells, IE2 or IE4/5 mrna were found only occasionally and the signals for IE1 and UL18 were of greatest intensity. Rarely, evidence of reactivation was also found in cells that contained LATs (Fig. 2c, d). The signal for LAT was located in neuronal nuclei and gave a heavy, medium or light intensity of grains, as described before (Ecob-Prince et al., 1993a). In a few such cells (about 10 in 1000 LAT + nuclei), mrna of IE2 or IE4/5 was detected, but these

4 2020 M. Ecob-Prince and K. Hassan Fig. I (left). Section through a D R G latently infected with 1814R, 3 days p.e., which has been stained by immunocytochemistry for virus antigen. A single neuron contains the brown deposit indicating the presence of virus-specific antigen. The section was counterstained with haematoxylin. Bar marker represents 25 jam. pi 0 Fig. 2 (above). Pairs (a and b, c and d) of serial sections from a D R G latently infected with in1814, 3 days p.e. Sections were all stained by immunocytochemistry for virus-specific antigens before being hybridized with a probe for either LATs (a and c) or IE2 m R N A (b and d). The LAT neuron in (a) shows no evidence of reactivation as judged by virus antigen or IE2 m R N A (b), whereas another neuron (arrow in a) contains both. Occasionally, evidence of reactivation was found in LAT neurons. For example, the LAT + neuron in (c) contains IE2 m R N A (arrow in d) in the absence of virus antigen or other virusspecific mrnas. Bar marker represents 25 jam. Fig. 3 (left). Section o f a D R G latently infected with wt HSV-I, 2 days p.e. Although the LAT signal was usually confined to the nuclei of latently infected cells (thick arrow), it became dispersed into the cytoplasm of some neurons. One such neuron, in which the nucleus is also visible, is indicated by the thin arrow. Bar marker represents 25 jam.

5 Reactivation of latent HSV 2021 Table 1. Percentage of DRG in which reactivation, judged by the presence of virus in the culture supernatant or of virus-specific antigen or mrna (ABC/ISH) in sections, can be detected at different times after explantation DRG positive for reactivation (%) DRG at different days p.e. with LATs Measure of Virus (%) reactivation wt HSV-1 56% (25/45) Supernatant virus* ABC/ISHt in % (35/45) Supernatant virus ABC/ISH R 71% (32/45) Supernatant virus ABC/ISH * Cell-free virus in supernatant medium of T12 to L6 of approximately 10 mice (up to 80 DRG per point). 5" ABC/ISH on sections of T12 to L6 of at least four mice (more than 32 DRG per point). cells always lacked expression of any other virus-specific mrna or protein (Fig. 2d). Neither was there an association with a particular intensity of LAT signal. Of the thousands of LAT + neurons observed in the study, none were associated unequivocally with the expression of virus-specific antigen or early/late mrnas. At late times after explantation (4 days p.e. for DRG infected with wt HSV-1 or 5 days p.e. for in1814 or 1814Rinfected DRG) the LAT probe hybridized to a signal dispersed over the cytoplasm and nuclei of cells that also expressed high levels of virus antigen and early/late mrna. We presume this to be the expression of the 2 kb LAT, previously shown to be expressed during lytic replication (Spivack & Fraser, 1987, 1988; Wagner et al., 1988). This was quite distinct from the appearance of LATs in latently infected cells, some of which were still present in the sections of DRG even at these later times after explantation. It therefore appeared that virus reactivated in very few of the cells that still contained LATs. So, the remainder of the reactivation must either have been from latently infected cells in which LATs had been removed as a prerequisite or early consequence of reactivation, or reactivation was from another population of latently infected cells that did not express LATs. If reactivation was occurring exclusively in at least a proportion of LAT + neurons, the presumed downregulation of the LATs should result in a decrease in the percentage of LAT + cells over the period of reactivation. Therefore we took groups of animals that were latently infected with each of the viruses and estimated the percentage of LAT + neurons present in the DRG either before or at 1, 2 or 3 days p.e. Table 2 shows the percentage of LAT + neurons in L4 DRGs of latently infected animals, all of which had evidence of reactivation at 2 (wt HSV-1) or 3 (in1814, 1814R) days p.e. There was considerable variation in the values between animals within each group and so the individual values are shown. However, there was no consistent or statistically significant decrease in the numbers of LAT + cells around the time of reactivation in DRG infected with any of the viruses. Similar tables have been constructed (results not shown) for each DRG from T12 to L6 for each of the viruses, some of which would have no evidence of reactivation at these earlier times of 2 or 3 days p.e. The tables did not indicate a difference between DRG in which reactivation was occurring from those in which it was not, and similarly revealed no evidence that there was a reduction in the percentage of LAT + neurons around the time of reactivation following explantation. We then considered whether there would be a shift in the proportion of LAT + cells from those with a heavy intensity of grains to those with a light intensity, indicating a possible down-regulation and/or removal of LATs, around the time of reactivation. We have previously described (Ecob-Prince et al., 1993a) and defined the different levels of grain intensity over LAT + cells as light (< 50 grains), medium (50 to 120 grains) or heavy (those with so many grains as to be uncountable). It can be seen from Table 3 that the LAT + nuclei prior to explantation of DRG from animals latently infected with wt HSV-1, in1814 or 1814R can be divided into about 29 % with light, 34 % with medium and 37 % with heavy grain intensities. Although there are a few animals in all groups in which the percentage of neurons with a light intensity rises to 60 to 70 % (and heavy intensity reduces to under 10 %), this did not correlate with the periods of reactivation and was not consistent between animals in the same group. Therefore, there was no evidence of a shift in the intensity of grains over LAT + neurons from a heavy to a light distribution. However we had noted that the signal for LAT could appear, in a few neurons, to be 'scattered' throughout the cell (Fig. 3). This appearance was found occasionally in neurons at 2 days p.e. (where the nuclear profiles could also be seen) but was found more frequently in DRG explanted for more than 4 days (irrespective of the virus with which the mice had been inoculated) where nuclear profiles were less obvious. The total lack of viral antigen or other mrnas in these cells demonstrated that this was not the expression of LAT associated with lytic replication. It might suggest that LATs were being transported to the cytoplasm, possibly after splicing and/or prior to translation. Alternatively, the scattered LATs could indicate there was a breakdown of the

6 2022 M. Ecob-Prince and K. Hassan Table 2. Percentages of LAT + neurons in L4 DRG of latently infected mice either before, or at different times after, explanlation to induce reactivation* LAT neurons in L4DRG(%) at different days p.e. Virus wt HSV-I 3.2, 41.6, 6.1, 10-2, 10.2, 6-0, 21.0, 26.3, 21.9, 8.3, 4.1, 3.3, , "2t "8 6-0 in "1, 11"9, 12'7, 4"2, 2'6, 2'3, 9"9, 10"7, 9"9, 11"9, 9"6, 2"8, 6"6, 10"3, 13"7 10'9, 4"4 5'7 3"0, 5"6 12"6 4"9 9'1 6'6 1814R 13"7, 1"5, 15"8, 17"3, 2"4, 1"1, 2'3, 4"6, 1'8, 15"6, 22'7, 5"9, 19"6, 7"0, 27"2 12"3 4"3 4"9, 1"8, 4"9 14'1 8"3 3"3 9"3 * Reactivation was first detected in DRG from animals infected with wt HSV-1 at 2 days p.e., and in those infected with in1814 or 1814R at 3 days p.e. t Numbers in bold are mean values. Table 3. Percentage of LAT + neurons in DRG from T12 to L6 which have a light (L), medium (M) or heavy (H) intensity of grains over their nuclei, either before explantation or at different times after explantation-induced reactivation LAT + neurons in T12 to L6DRG(%) at different days p.e Virus L M H L M H L M H L M H wthsv-i " m R , * Numbers in bold are mean values. nuclear membrane associated with death of the explanted neurons. We therefore investigated whether the neurons were healthy in the explanted DRG. Serial sections from DRG of the mock-infected control mice which had been fixed at 1, 2, 3, 4, 5, 7 and 10 days p.e. were stained with cresyl fast violet to demonstrate the Nissl substance (RNA and polyribosomes of the rough endoplasmic reticulum) characteristic of neuronal cytoplasm, or with Palmgren's silver impregnation to demonstrate the presence and organization of neurofilament. Only the appearance of sections stained for Nissl substance are presented (Fig. 4) but results of the silver stains gave similar conclusions. The DRG prior to explantation contained densely packed neuronal cell bodies (Fig. 4a), each with a central nucleus and prominent nucleolus,

7 Reactivation of latent H S V 2023 (a) (c) (f) (e),,~ ~ "~t~gg"w. "" If ece e.a ~ Fig. 4. Sections of D R G either before explantation (a, b), or at 1 (c, d) or 3 ( e, f ) days p.e. Stained by cresyl fast violet to demonstrate Nissl substance. D R G before explantation contain closely packed neurons, each with a central nucleus and single prominent nucleolus (a, b). After 1 day in culture, neurons in the centre of the D R G stain poorly (c) and those at the periphery have eccentric nuclei (d). By 3 days p.e. centrally located neurons have either been lost or are grossly abnormal (e), leaving a peripheral rim of neurons that have eccentric nuclei but which still contain Nissl substance (f). Bar markers represent I00 I~m (a, c, e) or 30 lam (b, d,f). seen more clearly at a higher magnification (Fig. 4b). By 1 day p.e., neurons in the central portion of the DRG stained poorly (Fig. 4c) whereas those in the peripheral rim still appeared relatively normal except for the eccentric nuclei (Fig. 4d). However, after 3 days p.e., centrally located neurons had either been lost or were grossly abnormal (Fig. 4e) and even those in the peripheral region had eccentric nuclei and poor Nissl substance (Fig. 4f). Neurofilament organization was lost from all but a few of the peripherally located neurons by 1 day p.e. and was disrupted or absent from most neurons by 3 days p.e. Therefore the neurons do not survive the effects of excision and explantation and by 3 days p.e. many had already been lost, particularly in the

8 2024 M. Ecob-Prince and K. Hassan Table 4. A comparison of the number of sites of reactivation (as judged by the presence of virus-specific antigen) with the estimated number of LAT + neurons in the same DRG, when reactivation was first detected at 2 or 3 days p.e. Virus T11 T12 T13 LI L2 L3 L4 L5 L6 Total per mouse wt HSV-1 in R 2 days p.e. (a) Antigen 2 s* 3 s 4Sl F 8s7 F 12s8 F 22s5 F 20 s 5 ~ ls4 ~ 104 LAT (b) Antigen s 22s5 F 5sl F - 34 LAT (c) Antigen 0 6 s 2 s 0 1 s 11 s 23 s 8 s - 51 LAT days p.e. (a) Antigen s I s 3 s 5 s 3s2 F lsl F 19 LAT (b) Antigen 1 s Sl 1F1L 12s14 F 9s6 F 0 75 LAT (c) Antigen ls6 F lsl ~ 0 19 LAT days p.e. (a) Antigen s 9s8 F 16s5F3 L 0 42 LAT (b) Antigen 0 1 s s 2 s 6Sl ~ - 12 LAT (c) Antigen 0 0 ls 9 s5 F 3 s 1F 8 s 1F 9 s 1F 7 s IF 1 s 47 LAT * S, single cell; F, small focus of neuron and closely associated cells; L, large focus of cells. centre of the DRG. Therefore we concluded that the scattered pattern of LATs was probably a reflection of the breakdown of the nuclear membrane, rather than a reorganization of LATs prior to reactivation although this latter explanation cannot be excluded, particularly at times before 3 days p.e. In an attempt to prolong the survival of the neurons, we explanted DRG into media containing either 50 % NCS, 50 % FCS or 50 % horse serum (HS); a mixture of 25 % NCS and 25 % FCS or HS, or a mixture of 10% HS and 10% FCS. None of these media prolonged the survival of the DRG neurons over that in the usual 10% NCS plus 10% FCS. In sections observed during the study, the number of LAT + neurons always appeared to out-number the sites of reactivation. We therefore stained, for antigen, all the serial sections of DRG from each of three mice (for which the percentage of LAT + cells was known) for each of the viruses at either 2 (wt HSV-1) or 3 (in1814 and 1814R) days p.e. By microscopic observation of the slides, the actual number of antigen-positive areas was counted (Table 4). These were either single cells which were found in one to seven serial sections (superscript S in the Table), small foci which were found through five to 10 sections (superscript F in the Table) or large foci continuous through up to 15 or 26 sections (superscript L in Table 4). These counts do not include the small fraction of cells with IE2 or IE4/5 mrna in the absence of antigen. Table 4 re-emphasizes the point that reactivation was most often associated with L4 DRG, and also with L3 and L5 DRG. Moreover, small foci of antigen-positive cells, and not just single cells, were found in these DRG, suggesting that reactivation had begun earlier and had already spread to adjacent cells. Reactivation was not synchronous. The total number of LAT + neurons per DRG has been estimated from the calculated percentage of LAT + cells, using the assumption that DRG T12 to L2 contain approximately 3000 neurons, that L3 and L6 contain approximately 3500 neurons and that L4 and L5 contain approximately 4000 neurons (M. Ecob-Prince, personal observation). The number of sites of reactivation was between 1-2 % and 2.9 % of LAT + neurons in the same DRG, irrespective of the virus type. One exception was found in the DRG of an animal which was latently infected with in 1814 (line a, in Table 4), where the number of foci of reactivation was 6.3% of the LAT + neurons present. However, these DRG also contained large foci of antigen-positive cells so the actual number of sites may be increased by the presence of secondary foci. Therefore there may be some correlation between the number of LAT + neurons and the number of foci of reactivation in each DRG. Sections of DRG at 4, 5, 7 or 21 days p.e. were also investigated by combined ABC/ISH. Antigen foci increased in size and number with increasing time, involving more non-neuronal cells, particularly in the root of the DRG. Foci of reactivation in DRG from levels other than those of L3, L4 and L5 were seen more frequently. The larger foci contained little or no IE2 or

9 Reactivation of latent HS V 2025 IE4/5 mrna but the signals for UL18 and IE1 were very strong. LAT expression typical of that seen in lytic replication (signal scattered over the cytoplasm and nucleus of antigen-positive areas) was also strong and, together with the changes associated with increasing death of neurons, made interpretation of the observations very difficult. Scattered LATs over antigennegative areas were also fairly common after 5 days p.e., although some cells still appeared to contain nuclear LATs typical of latently infected cells. By 7 days p.e., the level of viral mrna associated with antigen-positive areas was reduced and by 21 days p.e., there was very little evidence of mrna except in the few remaining non-neuronal cells that were still alive in the DRG which were now full of viral antigen. Discussion The results indicated that although the percentage of LAT neurons in animals latently infected with in1814 was similar to that described earlier (Ecob-Prince et al., 1993a), mice latently infected with either wt HSV-1 or 1814R also contained large numbers of LAT + neurons. Infection with these viruses also resulted in fewer deaths of animals than before (results not shown) suggesting that these mice were more able to survive the infection. This emphasizes our earlier conclusion that the number and spread of LAT + neurons was not unique to in1814 but could also occur in mice that survived infection with wild-type or rescuant viruses. The results presented also suggest that the rescuant had a secondary mutation because of the heterogeneity of the wild-type parent stock (C. Preston, personal communication) and is not directly comparable to the wild-type virus used in this study. Reactivation was judged by the presence of virusspecific mrnas and antigens in serial sections of DRG, from which release of infectious virus into the supernatant had also been assessed. There were no qualitative differences between the events involved in reactivation of the different viruses. The virus in1814 which lacks the trans-activation function of Vmw65 appeared to reactivate as efficiently as the rescued revertant 1814R in the explanted DRG. This confirms previous findings (Steiner et al., 1990) and suggests that reactivation following explantation does not involve Vmw65, at least in the initial stages. The subsequent lytic replication may be affected but as the virus appears to spread only to adjacent cells, it will be infecting these at a high m.o.i, at which the requirement for Vmw65 is less important. Reactivation was detected at the cellular level before virus was detectable in the supernatant. At these earliest times of 2 or 3 days p.e., reactivation was mainly in single neuronal cells or in small foci that included a neuron and its associated satellite or Schwann cells. The positive associations of virus-specific mrnas with each other and with LATs as well as antigen were assessed from serial sections of DRG from at least four animals for each of three viruses for each time point of 0, 1, 2, 3 and 4 days p.e. There are obvious dangers in attempting to derive a dynamic interpretation from static observations. The sensitivity of detection will vary between different techniques and different riboprobes, reactivation did not occur in a synchronous manner and a full profile of all the different mrnas could not be determined for every cell. Nevertheless, it was not possible to perform serial observations on a single animal as reactivation occurred, so we would like to propose the following scheme as a possible series of events involved in reactivation of latent virus following explantation of latently infected DRG. There was certainly a progression from antigen (Ag) being present in single cells, then in small foci and then in large foci as replication spread. The association of different RNAs with these stages is summarized as follows. IE2 IE2/ IE4/5 IE4/5/ Vmw65/ IE4/5 IE1/ UL18/ Vmw65/ IE 1/LATs UL18 Ag- Ag -/+ Ag + Ag + Ag ++ Single Single Single Small Large cell cell cell loci loci According to this scheme, the presence of IE2 mrna with or without IE4/5 mrna, in the absence of virus antigen, would indicate that virus was in the early stages of reactivation in that cell. Evidence of reactivation was found primarily in neurons which did not contain LATs in their nuclei. Rarely, expression of IE2 and/or IE4/5 was found in a neuron that did contain LATs in the absence of other virus-specific mrnas or antigen, suggesting that these neurons were probably in the early stages of virus reactivation. If the LATs are downregulated or removed as a very early event in reactivation, it would explain why reactivation was found primarily in neurons that did not contain LATs. For example, IE3 mrna (Vmw175), for which we did not have a riboprobe, negatively regulates expression from LAT constructs in co-transfection assays (Batchelor & O'Hare, 1990) and could be involved in the early stages of down-regulation of LATs in reactivation. Their active removal from the nucleus into the cytoplasm (for degradation or translation into the putative LAT-related polypeptide) may also be involved prior to the onset of lyric replication. We noted a dispersal of the LAT signal into the cytoplasm of some neurons in which there was no evidence of virus lytic replication. However, in-

10 2026 M. Ecob-Prince and K. Hassan terpretation of our results is complicated by the fact that the neurons were dying around this time, and the dispersal even at early times after explantation may be the result of the breakdown of the nuclear membrane. Moreover, we were unable to demonstrate either a reduction in the percentage of LAT + neurons or a reduction in the intensity of the LAT signal from one of heavy to light intensity, either prior to or around the time of reactivation. If reactivation was from neurons which had previously contained LATs, then we might have expected to detect both phenomena. We then compared, at early times after explantation, the total number of sites of reactivation with the estimated number of LAT + neurons for a particular DRG. The number of sites of reactivation was a slight underestimation because it did not include those few cells in which there was mrna without antigen. Generally, DRG with the highest levels of LAT + cells had the highest number of sites of reactivation. There also appeared to be a relatively reproducible proportion of LAT + cells involved in reactivation, assuming that all reactivation was from LAT + cells. However such correlations do not show that reactivation was exclusively from LAT + neurons. Despite all neurons receiving a similar stimulus, reactivation involved very few cells, possibly less than 3 % of those which expressed LATs. The poor survival of the explanted neurons may contribute to the low numbers because virus can reactivate only in a lb)ing host cell. These small numbers of successful reactivations may also have contributed to our failure to detect an overall decrease in the numbers of LAT + cells or a shift from heavy to light intensity of the LAT signal around the time of reactivation. If the neurons had survived longer, reactivation may have occurred eventually in a higher proportion of neurons. Nevertheless, the virus which did reactivate was found only rarely in LAT + neurons. However, if the LATs are removed as a prerequisite or early event in reactivation, and if it happens in such a short time frame that the chances of catching the event at any one moment (e.g. fixation) are rare, then reactivation will be detected in cells that no longer express LATs. However we cannot exclude the possibility that reactivation is occurring in latently infected cells that do not express LATs, such as those described in autonomic ganglia (Rodahl & Stevens, 1992). The death of neurons in the explanted DRG raises several interesting issues. It was always assumed that the stimulus for reactivation following explantation was indirectly the result of either the loss of cytoplasm, or the loss of neurotrophic factor from either the peripheral target or from the circulation. In either case, it would result in a switch in RNA transcription patterns in the neuron, from one of specialized function (e.g. neurotransmitter release) to one orientated to the repair and synthesis of membranes and cytoplasm (Wong & Oblinger, 1991). The change in transcription factors/activity which this involved would, possibly inadvertently, also stimulate reactivation of the virus. Our finding that the neurons die as a result of explantation suggests that an alternative stimulus for reactivation, not mutually exclusive with the first, could be operating: that death of the cell stimulates reactivation or releases the virus from the effects of a repressor present in the living cell. If cell death is the stimulus for reactivation in explanted DRG, the question arises as to whether this is an adequate model of reactivation in vivo. There is no evidence that natural reactivation in a patient who undergoes some form of' stress' is the result of sensory nerve cell death. However cell death may occur following section of the trigeminal nerve to relieve neuralgia, and it may also be true of the animal models in which nerve section was thought to stimulate reactivation. If too much of the cytoplasm is lost when the axon is severed, the cell may be unable to recover from the trauma and will die (Arvidsson et al., 1986; Ygge, 1989). Similarly, neurons unable to find a peripheral target (by reinnervation following neurectomy) also die, possibly as a result of failure to receive neurotrophic factor(s) supplied by the target tissue. Reactivation in animal models using neurectomy has been most successful when the operation is most likely to have caused neuronal cell death, for example when the nerve has been severed as close to the cell bodies as possible (McLennan & Darby, 1980; Walz et al., 1974), or when the DRG have been left neurectomized for more than 10 days (Tenser et al., 1988) (which is the usual time for re-innervation to be complete in the mouse). In fact, there can be as much as a 33% reduction in the number of neurons in an uninfected L5 DRG at 21 to 24 days post-neurectomy (Tenser et al., 1993). When we investigated reactivation in DRG that had been similarly neurectomized, but left for less than 10 days, we did not find evidence of reactivation resulting in lytic replication (Ecob-Prince et al., 1993b). In conclusion, explantation of latently infected DRG causes reactivation of virus in only a small number of neurons possibly because the majority of the neurons are killed by the explantation and virus can reactivate only in a living host cell. We could not detect an overall reduction in the percentage of LAT + neurons nor a redistribution in the LAT signal from either one of heavy to light intensity or nuclear to cytoplasmic. However our failure to detect such changes could reflect the small numbers of cells involved and non-specific changes due to neuron cell death. If reactivation does occur exclusively in LAT + cells, our evidence would suggest that

11 Reactivation of latent HSV 2027 LATs are removed from the nuclei either as a prerequisite or very early consequence of reactivation. We cannot exclude the possibility that reactivation may also occur in latently infected cells that do not express LATs. We would like to thank Professor P. G. E. Kennedy, in whose department the work was done, and Professor J. H. Subak-Sharpe, for their interest and support. Mr L. Miller, in the Department of Neuropathology, gave us invaluable advice on histological technique. The work was supported by a grant from the BUPA Medical Foundation (to M.S.E.-P. and P.G.E.K.) and the MRC (M.S.E.-P.). References ACE, C.I., DALRYMPLE, M.A., RAMSAY, F.H., PRESTON, V.G. & PRESTON, C. M. (1988). Mutational analysis of the herpes simplex virus type 1 trans-inducing factor Vmw65. Journal of Virology 69, ACE, C. I., MCKEE, T. A., RYAN, J. M., CAMERON, J. M. & PRESTON, C. M. (1989). Construction and characterization of a herpes simplex virus type 1 mutant unable to transinduce immediate-early gene expression. Journal of Virology 63, ARVIDSSON, J., YGGE, J. & GRANT, G. (1986). Cell loss in lumbar dorsal root ganglia and transganglionic degeneration after sciatic nerve resection in the rat. Brain Research 373, BATCHELOR, A. H. & O'HARE, P. (1990). Regulation and cell typespecific activity of a promoter located upstream of the latencyassociated transcript of herpes simplex virus type 1. Journal of Virology 64, BLOCK, T. M., SPIVACK, J. G., STEINER, I., DESHMANE, S., MCINTOSH, M.T., LIRETTE, R. P. & FRASER, N. W. (1990). A herpes simplex virus type 1 latency-associated transcript mutant reactivates with normal latencies from latent infection. Journal of Virology 64, BROWN, S. M., RITCHIE, D. A. & SUBAK-SHARPE, J. H. (1973). Genetic studies with herpes simplex virus type 1. The isolation of temperature-sensitive mutants, their arrangement into complementation groups and recombination analysis leading to a linkage map. Journal of General Virology 18, DESHMANE, S.t., NICOSIA, M., VALYI-NAGY, T., FELDMAN, L.T., DILLN~R, A. & FRASER, N. W. (1993). An HSV-1 mutant lacking the LAT TATA element reactivates normally in explant cocultivation. Virology 196, DISBREY, B. D. & RACK, J. H. (1970). Histological Laboratory Methods. Edinburgh & London: Churchill Livingstone. DOBSON, A.T., SEDARATI, F., DEvI-RAO, G., FLANAGAN, W.M., FARRELL, M. J., STEVENS, J. G., WAGNER, E. K. & FELDMAN, L. T. (1989). Identification of the latency-associated transcript promoter by expression of rabbit beta-globin mrna in mouse sensory nerve ganglia latently infected with a recombinant herpes simplex virus. Journal of Virology 63, 3844~3851. DOERIG, C., PIZER, L. I. & WILCOX, C. L. (1991). An antigen encoded by the latency-associated transcript in neuronal cell cultures latently infected with herpes simplex virus type 1. Journal of Virology 65, 2724~2727. ECoB-PRINCE, M.S., PRESTON, C.M., RIXON, F.J., HASSAN, K. & KENNEDY, P. G. E. (1993a). Neurons containing latency-associated transcripts are numerous and widespread in dorsal root ganglia following footpad inoculation of mice with herpes simplex virus type 1 mutant in Journal of General Virology 74, ECoB-PRINCE, M.S., RIXON, F.J., PRESTON, C. M., HASSAN, K. & KENNEDY, P. G. E. (1993b). Reactivation in vivo and in vitro of herpes simplex virus from mouse dorsal root ganglia which contain different levels of latency-associated transcripts. Journal of General Virology 74, FARRELL, M.J., DOBSON, A.T. & FELDMAN, L.T. (1991). Herpes simplex virus latency-associated transcript is a stable intron. Proceedings of the National Academy of Sciences, U.S.A. 88, FRASER, N. W., BLOCK, T. M. & SPIVACK, J. G. (1992). The latencyassociated transcripts of herpes simplex virus: RNA in search of function. Virology 191, 1-8. GRIEFITHS, I. R., MITCHELL, L.S., MCPHILEMY, K., MORRISON, S., KYRIAKIDES, E & BARRIE, J. A. (1989). Expression of myelin protein genes in Schwann cells. Journal of Neurocytology 18, HILL, J. M., SEDARATI, F., JAVIER, R. T., WAGNER, E. K. & STEVENS, J.G. (1990). Herpes simplex virus latent phase transcription facilitates in vivo reactivation. Virology 174, Ho, D. Y. & MOCARSKI, E. S. (1989). Herpes simplex virus latent RNA (LAT) is not required for latent infection in the mouse. Proceedings of the National Academy of Sciences, U.S.A. 86, LEIB, D. A., BOGARD, C. L., KOsZ-VNENCHAK, M., HICKS, K. A., COEN, D. M., KNIPE, D. M. & SCHAEFER, P. A. (1989). A deletion mutant of the latency-associated transcript of herpes simplex virus 1 reactivates from the latent state with reduced frequency. Journal of Vitvlogy 63, MCFARLANE, M., DAKSIS, J.I. & PRESTON, C.M. (1992). Hexamethylene bisacetamide stimulates herpes simplex virus immediate early gene expression in the absence of trans-induction by Vmw65. Journal of General Virology 73, MCLENNAN, J. L. & DARBV, 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, MADDOX, P. H. & JENKINS, D. (1987). 3 Aminopropyltriethoxysilane (APES): a new advance in section adhesion. Journal of Clinical Pathology 40, MITCHELL, W. J., LIRETTE, R. P. & FRASER, N. W. (1990). Mapping of low abundance latency-associated RNA in the trigeminal ganglia of mice latently infected with herpes simplex virus type 1. Journal of General Virology 71, NICHOLSON, P. (1992). Analysis of Jbur capsid protein genes of HSV-1. Ph.D. thesis, University of Glasgow. PERRY, L. J. & McGEocH, D. J. (1988). The DNA sequences of the long repeat region and adjoining parts of the long unique region in the genome of herpes simplex virus type 1. Journal of General Virology 69, RADER, K. A., ACKLAND-BERGLUND, C. E., MILLER, J. K., PEPOSF, J. S. & LEIB, D.A. (1993). In vivo characterization of site-directed mutations in the promoter of the herpes simplex virus type 1 latencyassociated transcripts. Journal of General Virology 74, ROCK, D.L., NESBURN, A.B., GHIASI, H., ONG, J., LEWIS, T.L., LOr, ENSGARO, J. R. & WECHSLER, S. L. (1987). Detection of latencyrelated viral RNAs in trigeminal ganglia of rabbits latently infected with herpes simplex virus type 1. Journal of Virology 61, RODAHL, E. & STEVENS, J.G. (1992). Differential accumulation of herpes simplex virus type 1 latency-associated transcripts in sensory and autonomic ganglia. Virology 189, SAWTELL, N. M. & THOMVSON, R. L. (1992). Herpes simplex virus type 1 latency-associated transcription unit promotes anatomical sitedependent establishment and reactivation from latency. Journal of Virology 66, SVlVACK, J. G. & FRASER, N. W. (1987). Detection of herpes simplex virus type 1 transcripts during latent infection in mice. Journal of Virology 61, SPIVACK, J. G. & FRASER, N. W. (1988). Expression of herpes simplex virus type 1 (HSV-1) latency-associated transcripts and transcripts affected by the deletion in avirulent mutant HFEM: evidence for a new class of genes. Journal of Virology 62, SVIVACK, J. G., WOODS, G. M. & FRASER, N. W. (1991). Identification of a novel latency specific splice donor signal within the herpes simplex virus type kilobase latency-associated transcript (LAT) : translation inhibition of LAT open reading frames by the intron within the 2.0 kilobase LAT. Journal of Virology 65, STEINER, I., SPIVACK, J. G., LIRETTE, R. P., BROWN, S. M., MACLEAN, A. R., SUBA~-SHARPE, J. H. & FRASER, N. W. (1989). Herpes simplex virus type 1 latency-associated transcripts are evidently not essential for latent infection. EMBO Journal 8, 50~511. STEINER, I., SPIVACK, J. G., DESHMANE, S. L., ACE, C.l., PRESTON, C. M. & FRASER, N. W. (1990). A h~rp& simplex virus type 1 mutant containing a non-transducing Vmw65 protein establishes latent

12 2028 M. Ecob-Prince and K. Hassan infection in vivo in the absence of viral replication and reactivates efficiently from exptanted trigeminal ganglia. Journal of Virology 64, STEVENS, J.G., WAGNER, E.K., DEvI-RAO, G.B., COOK, M.L. & FELDMAN, L. T. (1987). RNA complementary to a herpesvirus ctgene mrna is prominent in latently infected neurons. Science 235, 1056-I059. TENSER, R. B., EDRIS, W. A. & HAY, K. A. (1988). Herpes simplex virus latent infection: reactivation and elimination of latency after neurectomy. Virology 167, TENSER, R. B., EDRIS, W. A. & HAY, K. A. (1993). Neuronal control of herpes simplex virus latency. Virology 195, TROUSDALE, M.D., STEINER, I., SPI[VACK, J.G., DESHMANE, S.L., BROWN, S. M., MACLEAN, A. R., SUaAK-SHARVE, J. H. & FRASER, N.W. (1991). In vivo and in vitro reactivation impairment of a herpes simplex virus type 1 latency-associated transcription variant in a rabbit eye model. Journal of Virology 65, WAGNER, E.K., FLANAGAN, M.W., DEvI-RAo, G., ZHANG, Y.F., HILL, J. M., ANDERSON, K. P. & STEVENS, J. G. (1988). The herpes simplex virus latency-associated transcript is spliced during the latent phase of infection. Journal of Virology 62, WALZ, M. A., PRICE, R. W. & NOTKINS, A. U (1974). Latent ganglionic infection with herpes simplex virus types 1 and 2: viral reactivation in vivo after neurectomy. Science 184, WECHSLER, S. L., NESBURN, A. B., WATSON, R., SLANINA, S, GHIASI, H. (1988). Fine mapping of the major latency-related RNA of herpes simplex virus type 1 in humans. Journal of General Virology 69, WONG, J. & OBLINGER, M. M. (1991). NGF rescues substance P but not neurofilament or tubulin gene expression in axotomized sensory neurons. Journal of Neuroscience 11, YGGE, J. (1989). Neuronal loss in lumbar dorsal root ganglia after proximal compared to distal sciatic nerve resection: a quantitative study in the rat. Brain Research 478, ZWAAGSTRA, J., GHIASl, H., SLANINA, S.M., NESBURN, A.B., WH~ATEEV, S.C., LILLVCROP, K., WOOD, J., LATCHMAN, D.S., PATEL, K. & WECRSLER, S. L. (1990). Activity of herpes simplex type 1 latency associated transcript (LAT) promoter in neuron-derived cells : evidence of neuron specificity and for a large LAT transcript. Journal of Virology 64, (Received 29 November 1993; Accepted 17 February 1994)