Proc. Natl. Acad. Sci. USAVol. 92, pp. 10980-10984, November 1995Medical Sciences

Reactivated and latent varicella-zoster virus in human dorsalroot ganglia

(latency/reactivation/in situ hybridization/PCR-ELISA)

OCTAVIAN LUNGU*, PAULA W. ANNUNZIATOt, ANNE GERSHONt, SUSAN M. STAUGAITISt, DEBORAH JOSEFSONt,PHILIP LARUSSAt, AND SAUL J. SILVERSTEIN*§Departments of *Microbiology, tPediatrics, and tPathology, Columbia University, College of Physicians and Surgeons, 701 West 168th Street,New York, NY 10032

Communicated by Harold S. Ginsberg, National Institute ofAllergy and Infectious Diseases, Rockville, MD, July 31, 1995

ABSTRACT Ganglia obtained at autopsy were examinedby in situ hybridization from one patient with zoster (alsocalled herpes zoster or shingles), two varicella-zoster virus(VZV)-seropositive patients without clinical evidence of zos-ter, one VZV-seronegative child, and one fetus. Ganglia pos-itive for VZV had a hybridization signal in both neuronal andnonneuronal satellite cells. Ganglia obtained from the fetusand from the seronegative infant were consistently negativefor VZV. Two striking observations were evident regarding thepresence ofVZV DNA in ganglia obtained from the individualwith zoster at the time of death. First, ganglia innervating thesites of reactivation and ganglia innervating adjacent sitesyielded strongly positive signals in neurons and satellite cells,whereas ganglia from distant sites were rarely positive. Sec-ond, VZV DNAwas found in both the nuclei and the cytoplasmof neurons innervating areas of zoster. However, in neuronsinnervating zoster-free areas, VZV DNA was found only in thenucleus of neurons and their supporting satellite cells. Im-munohistochemistry with a fluorescent monoclonal antibodyto the VZV glycoprotein gpI, a late virus protein, revealed apositive signal in the cytoplasm of ganglia with clinicalevidence of reactivation. These results illustrate that bothneuronal and satellite cells become latently infected followingprimary VZV infection. The presence of VZV DNA and gpl inthe cytoplasm of neurons demonstrates productive infectionfollowing reactivation at the site of latency.

Varicella-zoster virus (VZV) is an alphaherpes virus thatcommonly infects young individuals living in temperate cli-mates (1-3). Following primary infection, clinically presentedas chicken pox, latent virus persists in the host and mayreactivate at a later date to cause zoster (also called herpeszoster or shingles). Clinical reactivation occurs most frequentlyin immunocompromised and older individuals and places theaffected patient at risk for developing postherpetic neuralgia.This debilitating pain syndrome is associated with significantmorbidity that may not be relieved by antiviral and analgesictherapy (4). It is therefore desirable to prevent zoster andpostherpetic neuralgia in individuals harboring latent virus.Understanding the cellular and viral factors governing VZVlatency and reactivation is essential to achieve this goal.Determining the location of persistent and reactivated virus isessential for defining the molecular basis of VZV pathogen-esis.

Electron microscopy has demonstrated VZV in sensoryganglia of patients with zoster (5, 6), and polymerase chainreaction (PCR) and Southern blot hybridization have detectedVZV DNA in latently infected dorsal root ganglia (7, 8).Studies utilizing in situ hybridization methods have identifiedseveral different VZV transcripts in affected ganglia (7, 9-13).

These studies have led to controversy as to the nature of thecell harboring the latent VZV genome. VZV transcripts havebeen localized exclusively in neurons (10, 13, 14) or in non-neuronal satellite cells (9, 12) depending on the in situ hybrid-ization technique used.

In the current study, the cellular and intracellular location ofthe VZV genome in latently infected human dorsal rootganglia and in ganglia with clinical evidence of reactivatedvirus was determined by in situ hybridization. Evidence forvirus replication in affected ganglia was provided by poly-merase chain reaction (PCR) and immunohistochemistry.These results demonstrate that latent VZV resides in bothneurons and their supporting satellite cells and that latent viruscan reactivate in neurons.

MATERIALS AND METHODSTissue Specimens. Dorsal root ganglia from two VZV-

seropositive patients without clinical evidence of zoster, oneVZV-seronegative child and one fetus without a maternalhistory of varicella, were obtained at autopsy. Ganglia from an85-year-old man with malignant glioma of the thoracic spinalcord who had a zosteriform rash in the distribution of the rightTi1, T12, and Li sensory nerves at the time of death were alsoobtained. Prior to developing zoster, the patient had radiationtherapy to the spinal cord at the levels of T4 through Li. Atautopsy, he had a vesicular eruption in the right Tll and Lidistribution, separated by an area of recently healed epidermisin the right T12 distribution.

Ganglia were fixed in 10% formalin and embedded inparaffin. Six-micrometer-thick sections were mounted on Si-lanated slides (Digene Diagnostics, Beltsville, MD) for in situhybridization and immunohistochemistry studies or wereplaced in sterile Eppendorf tubes for solution PCR. Tissuesections were heat-fixed at 60°C. The microtome blade wascleaned with 70% ethanol between cutting each tissue block.

In Situ Hybridization. The sections were deparaffinized withxylene, rinsed twice with ethyl alcohol, and digested withproteinaseK (PK) diluted to 10 ,g/ml in PKbuffer (100mM Tris,pH 7.4/150 mM NaCl/12.5 mM EDTA) for 10 min at roomtemperature. After rinsing with in situ hybridization buffer (50mM KCL/10 mM Tris, pH 7.4/3 mM MgCl2), the slides wereplaced in this buffer containing 40 pmol of the fluorescein-labeled VZV oligonucleotide probe (Table 1) per ml anddenatured at 95°C for 10 min. The hybridization reaction wasperformed at 60°C for 1-2 hr. The slides were rinsed twice, andthe hybridized probe was detected by incubating the sectionswith a 1:200 dilution of alkaline phosphatase (AP)-conjugatedanti-fluorescein antibody (Boehringer Mannheim) in 5 x SSC

Abbreviations: AP, alkaline phosphatase; ORF, open reading frame;DAPI, 4',6-diamidino-2'-phenylindole dihydrochloride; VZV, varicel-la-zoster virus.§To whom reprint requests should be addressed.

10980

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. Natl. Acad. Sci. USA 92 (1995) 10981

(1 x = 0.15 M NaCl/0.015 M sodium citrate, pH 7) containing1% bovine serum albumin and 1% Triton X-100 for 30 min atroom temperature in a humidified chamber. The slides werethen washed in AP buffer (100 mM Tris, pH 9.5/100 mMNaCl/50 mM MgCl2), and the signal was visualized by lightmicroscopy after developing for 20 min with 9 ,ul of nitrobluetetrazolium chloride (Boehringer Mannheim) and 3.5 ,ul of5-bromo-4-chloro-3-indolyl phosphate (Boehringer Mann-heim) in 2 ml of AP buffer.

Immunohistochemistry. The slides were deparaffinized asfor in situ hybridization and treated with Serotec targetunmasking fluid (Harlan Bioproducts for Science, Indianap-olis) according to the manufacturer's recommendations. TheVZV glycoprotein gpl was detected by immunohistochemistryusing fluorescein-conjugated and anti-gpl antibody (OrthoDiagnostics) according to the following modifications of themanufacturer's protocol. The tissue sections were incubatedwith the anti-gpl antibody for 30 min and rinsed with phos-phate-buffered saline. AP-conjugated anti-fluorescein anti-body was used to enhance the signal as in the in situ hybrid-ization procedure. The signal was visualized by light micros-copy.

4',6-Diamidino-2'-phenylindole dihydrochloride (DAPI)DNA Staining. The tissue sections mounted on Silanated slideswere washed in methanol, and double-stranded DNA wasdetected by incubating the slides with DAPI (BoehringerMannheim) as recommended by the manufacturer.PCR and ELISA Detection of Amplified Products. A pre-

viously described PCR method (15) was modified to amplifyand detect VZV from paraffin-embedded tissue sections. Thesections were digested overnight with 100 ,ul ofPK (400 ,ug/ml)at 60°C. After the aqueous phase was separated from theparaffin and cellular debris by centrifugation, 5% Chelex 100(Sigma) was added to each specimen and incubated for 1 hr at60°C. The PK was inactivated by boiling the specimens for 30min.The sequences of the NLA and FOK primers used for PCR

amplification are published (Table 1) (15). Primers weresynthesized with a biotin moiety incorporated at the 5' end ofthe FOK primers (Operon Technologies, Alameda, CA). VZVDNA was amplified in 100-,ul reaction volumes containing 10,ul of template, 0.5 ,uM of each primer, 0.2 ,uM of eachdeoxynucleotide, and 0.5 unit of Taq polymerase (Perkin-Elmer) in 50 mM KCl/1.5 mM MgCl2/10 mM Tris HCl, pH8.3. The reactions were amplified in a Perkin-Elmer GeneAmp 9600 thermocycler for 38 cycles as follows: 30 sec at 94°Cfollowed by 40 sec at 55°C and 1 min at 72°C. An initialdenaturing step of 3 min at 95°C and a final elongation step of3 min at 72°C were included. Human fibroblast cells infectedwith the Ellen strain of VZV, monkey kidney cells infectedwith herpes simplex viruses 1 and 2 (HSV 1 and HSV 2), andwater were included with each amplification as positive andnegative controls. To ensure that the samples containedamplifiable DNA, the human ,B-globin gene was amplifiedfrom each VZV-negative specimen (Perkin-Elmer). Amplifi-cation products were detected by ethidium bromide stainingafter electrophoresis through a 1.8% agarose gel or by ELISA

Table 1. Oligonucleotide primers and probes used for in situhybridization and PCR-ELISA

Oligo-nucleotide Nucleotide sequenceNLA GGAACCCCTGCACCATTAAAFOK Bio-TCCCTTCATGCCCGTTACATProbe CCFIUGATGCAACFIuGTTGGFIuCCAGAGGAGCFIuGG

with a VZV-specific fluorescein-labeled oligonucleotide probe(Table 1).ELISA detection of the PCR products was performed as

described (16). Briefly, the 222-bp amplification product witha 5' biotin moiety was captured on streptavidin-coated poly-styrene plates (Pierce) and hybridized with 10 pmol of thecomplementary fluorescein-labeled oligonucleotide probe(Table 1). PCR product-probe hybrids were detected withAP-conjugated anti-fluorescein antibody (Boehringer Mann-heim). One hundred microliters of AP substrate solution (1mg/ml dilution of p-nitrophenyl phosphate; Sigma) was pre-pared and added to each well as recommended by the man-ufacturer. The absorbance at 405 nm was determined after a1-hr incubation. An absorbance greater than 0.4 indicated thepresence of amplified product. This PCR-ELISA methoddetects 1-5 fg of DNA.

RESULTSOligonucleotide in situ hybridization probes may detect bothDNA and RNA species. The fluorescein-labeled VZV probe(Table 1) is complementary to 27 bp from ORF 54, a genecontained within the BamHI D fragment of VZV, which is notknown to be transcribed during latency (9, 14). In situ hybrid-ization of ganglia obtained from two patients without evidenceof acute VZV infection or reactivation revealed that both thefluorescent and antibody-mediated AP hybridization signalswere detected in the nucleus of affected cells. However,greater resolution of detail and less background staining wereobtained when the signal was amplified with anti-fluoresceinantibody conjugated to AP. Positive sections exhibited hybrid-ization signal in both neurons and nonneuronal satellite cells(Fig. 1A). The proportion of positive cells differed among the

FIG. 1. In situ hybridization of VZV-positive and -negative ganglia.(A) A ganglion harboring VZV illustrates in situ hybridization APsignal in the nuclei of neurons and satellite cells. (B) Ganglion froma VZV-seronegative infant lacks AP in situ hybridization signal.

Bio denotes conjugated biotin and Flu denotes conjugated fluores-cein. The probe is complementary to 27 bp within the VZV openreading frame (ORF) 54 region amplified by the NLA/FOK primers.

Medical Sciences: Lungu et aL

10982 Medical Sciences: Lungu et al.

Table 2. In situ hybridization results of ganglia obtained from fourpatients without evidence of acute VZV infection

Neurons Satellite cellsPatients Specimen positive, % positive, %

Adult 1 1 ganglion 20-30 20-302 ganglia 5-10 5-104 ganglia None None

Adult 2 1 ganglion 15-20 20-301 ganglion 3-8 5-101 ganglion None None

Infant None NoneFetus None None

At least four sections were examined from each ganglion.

ganglia examined; some had many positive cells and others hadonly a few positive cells (Table 2). Five percent to 30% ofneurons and/or satellite cells in ganglia from patients withzoster or a history of varicella were positive for VZV DNA byin situ hybridization.To determine the specificity of the hybridization probe, two

ganglia obtained from a stillborn fetus with no maternalhistory of recent varicella or herpes zoster infection and threeganglia from an infant with no history of varicella infectionwere examined. In situ hybridization studies of these specimenswere consistently negative (Fig. 1B and Table 2). Studies withother control fluorescein-labeled oligonucleotide probes forseveral human papillomaviruses and Mycobacterium tubercu-losis were negative. Thus, the probe is specific for VZV DNA,and the fluorescein moiety does not nonspecifically labelganglionic tissue.The dorsal root ganglia of an 85-year-old man with malig-

nant glioma of the thoracic spinal cord who had a zosteriformrash in the distribution of the right Tll, T12, and Li sensorynerves at the time of death were studied. On histologicexamination, the right T12 ganglion showed severe necrosisand inflammation, while the right Tll and Li ganglia weremildly inflamed. Histology of ganglia not associated withclinical zoster was unremarkable. In situ hybridization andPCR results of individual ganglia from this case are summa-rized in Table 3. Ganglia innervating the sites of reactivationand adjacent ganglia yielded strongly positive signals. Distantganglia were rarely positive. The right T 1 ganglion had manyneurons and satellite cells that were positive for VZV DNA byin situ hybridization. The right T12 ganglion was diffuselynecrotic, and cellular definition was obscured in all but a fewneurons that were mostly positive forVZV DNA. The right T8,

T9, T10, Li, and left L4 ganglia had few positive cells. APsignal representing VZV DNA was found in both the nucleiand the cytoplasm of some affected Tll neurons and mostaffected T12 neurons (Fig. 2A). In contrast, VZV-positiveganglia from dermatomes without corresponding evidence ofreactivation, such as the left L4 ganglion, had detectable viralDNA in the nuclei but not in the cytoplasm of affected neurons(Fig. 2D). Although the patient's rash also involved the rightLi dermatome, this ganglia had in situ hybridization signalsonly in nuclei of VZV-positive cells.The presence of VZV DNA in the cytoplasm of neurons

from the right Tii and T12 ganglia indicates productiveinfection. During VZV replication, late genes such as thoseencoding structural glycoproteins are expressed. To demon-strate that late gene products were being produced in gangliawith cytoplasmicVZV signals, these ganglia were examined forexpression of gpI. gpI was detected in the cytoplasm of neuronsfrom the right Tii and T12 ganglia (Fig. 2B). The left L4ganglion did not have detectable gpl (Fig. 2E). When theseganglia were stained with DAPI, a fluorochrome that interactswith double-stranded DNA, the neuronal cytoplasm of theright T 1 and T12 ganglia exhibited punctate fluorescence ina pattern indistinguishable from the pattern seen by in situhybridization (Fig. 2C). No cytoplasmic DAPI fluorescencewas detected in the left L4 ganglion (Fig. 2F).The results from PCR amplification confirmed the in situ

hybridization findings. These results correlated, with the ex-ception that five ganglia positive by in situ hybridization werenegative by PCR. Four of the five ganglia had only a fewpositive cells by in situ hybridization, and it is likely that theamount of VZV DNA present in these specimens was belowthe level of detection of the PCR-ELISA assay. The right T10ganglion had many positive cells by in situ hybridization but wasnegative by PCR-ELISA. The VZV PCR-negative tissuesections contained amplifiable ,B-globin DNA.

DISCUSSIONThe establishment of latent VZV infection is thought toinvolve either axonal transport of virus from skin lesions orhematogenous seeding of virus to the dorsal root gangliaduring primary infection. As the result of interactions betweenthe virus, host immune responses, and the cellular environ-ment, VZV establishes and maintains a dormant state insensory ganglia. Reactivation is thought to occur when alter-ations in the balance between virus and host factors allow local

Table 3. Comparison of results of in situ hybridization (ISH) and PCR-ELISA analyses for the presence of VZV DNA in left and rightganglia obtained from an individual with zoster involving the T1l, T12, and Li dermatomes

Left Right

VZV DNA presence VZV DNA presence

ISH ISH

Neurons Satellite cells Neurons Satellite cellsGanglion Rash positive, % positive, % PCR Rash positive, % positive, % PCR

T8 - 0 0 - - 5-10 5-10T9 - 0 0 - - 10-15 10-15T10 - 0 0 - - 30-40 25-30Tll - 3-6 3-6 - + 60-70*t 50-70 +T12 - NA NA + + 70-80 Necrotic +Li - 0 0 - + 5-10 10-15 +L3 - 0 0 - - 0 0L4 - 4-8 5-10: - - 0 0L5 - 0 0 - - 0 0

NTA, not avaIlabe1-11 br study.-1-NA, not available for study.*gpl+.tNeuronal cytoplasm was positive by ISH.tgpI-.

Proc. Natl. Acad. Sci. USA 92 (1995)

Proc. Natl. Acad. Sci. USA 92 (1995) 10983

i ,..: _ 1t W-~~u .t . ,50 .

FIG. 2. In situ hybridization, immunohistochemistry for gpl, and DAPI staining of the right Tll and left L4 ganglia. (A) In situ hybridizationof right Ti 1 illustrates AP signal in the nucleus and cytoplasm of neurons and in the nucleus of satellite cells. (B) Immunohistochemistry detectingVZV gpl in the cytoplasm of right T1l neurons. (C) DAPI-stained particles are apparent in the neuronal cytoplasm of right Tll. (D) In situhybridization of left L4 illustrating AP signal in the nucleus of neurons and satellite cells but not in the cytoplasm of neurons. (E)Immunohistochemistry of left L4 is negative for gpL. (F) No DAPI-stained particles are present in the neuronal cytoplasm of left L4.

replication of virus in the ganglion and axonal transport to theskin, resulting in zoster (reviewed in ref. 2).Determining the cellular location of persistent virus is

essential to understanding the regulation of latency and reac-tivation. In situ hybridization methods have detected VZVnucleic acids exclusively in the satellite cells of latently infectedtrigeminal ganglia (9, 12) or in the neurons (10, 14). Thesedifferences in latency sites may occur because the probes usedin these studies hybridized to different VZV transcripts andgenes. It is possible that unique regions of the viral genome aretranscribed in different cell types.

In this study, ganglia from adults that were positive for VZVexhibited signal in both neurons and satellite cells. The ques-tion arises whether virus underwent reactivation prior toobtaining the tissue. This seems unlikely because no in situhybridization signal was found in the cytoplasm of positiveneurons from patients without active zoster.These data are consistent with those obtained from animal

models and in vitro studies. The latent virus is not entirelyquiescent; several VZV genes have been shown to be expressedduring latency including ORFs 4, 21, 29, 62, and 63 (9, 10, 12,14, 17). Ganglia harboring latent VZV obtained from a ratmodel contained the product of ORF 63, a VZV-transactivat-ing protein, in neurons and satellite cells (18). Additionally,investigators were able to recover VZV from an in vitro modelof latency and reactivation only when both neurons andsatellite cells were included in the system (19). These results

and those presented here lead to the conclusion that latentVZV is maintained in neurons and satellite cells of dorsal rootganglia and that both cell types are essential to the processesof VZV latency and reactivation.

Ganglia obtained from a patient with zoster at the time ofdeath had intense in situ hybridization signals in the nuclei andcytoplasm of many neurons and in the nuclei of satellite cells.VZV DNA was present in the neurons and satellite cellspredominantly in the ganglia corresponding to the affecteddermatomes and to a lesser extent in the adjacent ipsilateralganglia without corresponding dermatologic signs of reactiva-tion. Ganglia innervating asymptomatic dermatomes hadfewer VZV-positive neurons and satellite cells.The presence of cytoplasmic VZV DNA and gpI illustrates

replication ofVZV in neurons during reactivation. These dataare consistent with electron microscopic studies (5, 6). Thelocation of virus in the satellite cells could not be determinedbecause of their large nucleus-to-cytoplasm ratio. However,the absence of gpI in these cells suggests that during reacti-vation, replication of VZV occurs predominantly in neurons.

Previous in situ hybridization studies found only 0.01-0.3%of ganglion cells harboring VZV nucleic acids (9, 10). Ourresults illustrate that VZV infects a larger population ofganglion cells than previously recognized. VZV DNA wasdetected by in situ hybridization in 5% to 30% of neurons andsatellite cells in affected ganglia obtained from individualswithout signs of active VZV infection. In the individual with

Medical Sciences: Lungu et al.

10984 Medical Sciences: Lungu et al.

zoster, the majority of neurons and satellite cells in gangliawith corresponding clinical evidence of reactivation werepositive for VZV DNA. Results of in situ hybridization of theremaining ganglia studied from this individual were similar tothe results of ganglia obtained from individuals with latentvirus. It appears that initially only a few cells within a ganglionharbor latent VZV. Upon reactivation the virus replicates andinterneuronal spread occurs, leading to clinical disease. De-termining the role each ganglion cell type plays in maintainingnonproductive VZV infection and its involvement in theprocess of reactivation may suggest novel strategies to preventthe clinical manifestations of zoster and its sequelae.

Studies in the authors' laboratories were supported by Grant AI124021 from the U.S. Public Health Service to A.G. and S.J.S. and byfunds from Columbia Innovation Enterprise to S.J.S.O.L. was sup-ported by a grant from the VZV Research Foundation, Inc.

1. Gelb, L. (1990) in Varicella-Zoster Virus, eds. Fields, B. N. &Knipe, D. M. (Raven, New York), pp. 2011-2054.

2. Ostrove, J. M. (1990) Adv. Virus Res. 179, 834-844.3. Weller, T. H. (1983) N. Engl. J. Med. 309, 1362-1368.4. Straus, S. E. (1993) J. Am. Med. Assoc. 269, 1836-1839.5. Esiri, M. M. & Tomlinson, A. (1972) J. Neurol. Sci. 15, 35-48.6. Ghatak, N. & Zimmerman, H. (1973) Arch. Pathol. 95, 411-415.

7. Gilden, D., Vafai, A., Shtram, Y., Bercker, T., Devin, M. &Wellish, M. (1983) Nature (London) 306, 478-480.

8. Mahalingham, R., Wellish, M., Wolf, W., Dueland, A. N., Cohrs,R., Vafai, A. & Gilden, D. (1990) N. Engl. J. Med. 323, 627-631.

9. Croen, K., Ostrove, J., Dragovic, L. & Straus, S. (1988) Proc. Natl.Acad. Sci. USA 85, 9773-9777.

10. Hyman, R. W., Ecker, J. R. & Tenser, R. B. (1983) Lancet ii,814-816.

11. Meier, J. L. & Straus, S. E. (1992) J. Infect. Dis. 166, S13-S23.12. Meier, J. L., Holman, R. P., Croen, K. D., Smialek, J. E. &

Straus, S. E. (1993) Virology 193, 193-200.13. Tenser, R. B. & Hyman, R. W. (1987) Yale J. Bio. Med. 60,

159-167.14. Gilden, D., Rozenman, Y., Murray, R., Devlin, M. & Vafai, A.

(1987) Ann. Neurol. 22, 337-380.15. LaRussa, P., Lungu, O., Hardy, I., Gershon, A., Steinberg, S. &

Silverstein, S. (1992) J. Virol. 66, 1016-1020.16. Lungu, O., Sun, X. W., Wright, T. C., Ferenczy, A., Richart,

R. M. & Silverstein, S. (1995) Obstet. Gynecol. 85, 337-341.17. Cohrs, C., Mahalingam, R., Dueland, A. N., Wolf, A., Wellish, M.

& Gilden, D. H. (1992) J. Infect. Dis. 166, S24-S29.18. Debrus, S., Sadzot-Delvaux, C., Nikkels, A. F., Piette, J. &

Rentier, B. (1995) J. Virol. 69, 3240-3245.19. Somekh, E., Tedder, D. G., Vafai, A., Assouline, J. G., Straus,

S. E., Wilcox, C. L. & Levin, M. J. (1992) Pediatr. Res. 32,699-703.

Proc. Natl. Acad. Sci. USA 92 (1995)