The gene encoding the gIII envelope protein of pseudorabies virus vaccine strain Bartha contains a mutation affecting protein localization.

(1)

0022-538X/89/010250-09$02.00/0

The Gene

Encoding the

glll

Envelope Protein of Pseudorabies

Virus Vaccine Strain Bartha Contains

a

Mutation

Affecting Protein Localization

A. K. ROBBINS, J. P. RYAN,t M. E. WHEALY, AND L. W. ENQUIST* Central Research & Development Department, E. I. du Pont de Nemours & Co.,Inc.,

Experimental

Station, E3281B31,

Wilmington,

Delaware 19898 Received 13July 1988/Accepted 9 September 1988

Pseudorabies virus(PRV)vaccine strain Bartha hasa diminishedcapacity tocausedisease andharborsa

varietyof mutationsaffecting virulence. It has beenreportedthatPRVBarthaproducesvirions withreduced amountsof the major envelope glycoproteingIH. Onehypothesiswasthatthisphenotypewasdue toreduced

expression of thegIIIgene. Inthis report,wedemonstrate that the reduced amount ofgIllinvirionswasnot mediated at the level of transcription, but rather reflected a defect in protein localization. We describe experimentswithgenereplacementtechnologytoprovethat theexpressiondefectwascloselylinked to thegIII gene itself. Using pulse-chase experiments, we found a defect similar to that observed for certain signal sequence mutations of PRV Becker gIII. The Bartha gIII protein was translated, but was inefficiently

introduced into the membrane proteinexportpathway. Consequently, onlyafraction of theprimaryBartha gIll translation product was glycosylated and matured. The remaining fraction stayed presumably in the cytoplasm,where itneverbecameglycosylatedorinserted into cellor virus membranes. The resultwasthat

Bartha-infected cells producedvirions with reduced amounts ofgIIIin theirenvelopes.Comparisonof theDNA sequenceof thepromoterand amino-terminalcoding regionsof Becker and BarthagIllgenesrevealedasingle

base pair difference inBartha, changing codon 14 of the signalsequencefrom aleucine (CTC)toa proline

(CCC) codon.We suggest thatthesignalsequencemutation isresponsiblefor theapparent reducedexpression

phenotype of this attenuated strain. This mutation represents, toourknowledge, the firstreportednatural signalsequencemutationinaherpesvirusglycoprotein.

Pseudorabies virus(PRV)isan alphaherpesviruscausing

anaturalinfection in swine similartothatofherpes simplex virus in humans (4, 8). Because the disease in swine is of

economic importance, there has been considerable effortto

control thisproblem through vaccinationprograms. Accord-ingly, there hasbeen an increased efforttodevelop

attenu-atedlive vaccine strains. TheseattenuatedPRVstrainshave

been studied in detailbyanumberof laboratoriesto under-standthegeneralmechanisms ofherpesvirus virulence (1-3,

5, 7, 10, 11, 13, 14, 20, 21).

PRV strain Bartha (PRV-Ba) is a well-knownattenuated

vaccine strain (1) and is thebasis for the Duvaxyn vaccine sold in Europe (21). Although this strain grows well in

culture, itharbors a number of mutations, some of which

have been shown to contribute to its avirulent phenotype. Forexample,PRV-Bacontainsadeletionintheuniqueshort

region removing coding sequences for glycoproteinsgI and

gp63 (11, 14). Mettenleiteretal. (13) proved that glycopro-teingI played amajorrole in virulence but that itdid soin

conjunctionwithatleast one otherundefined viral function. These workers also demonstrated that the absence of gI

affects virus release from certaincelltypes (12).Lomniczi et al. (10) noted that theBamHI4fragment of PRV-Ba

encod-ing fourgenesinvolved in capsid assembly contained defects

contributingtothe lack of virulence.

Thesubject of this report is the characterization of another

defect in PRV-Ba, the reduced amount of glycoproteingIll found in the virusenvelope (2, 20). To explain this

observa-*Correspondingauthor.

tPresent address: Department of Microbiology, The Health ScienceCenter, Universityof Tennessee, Memphis, TN 38163.

tion, Ben-Poratetal. (2) havesuggested that PRV-Ba gIII is expressed in lesseramounts. We describeexperimentsusing genereplacementtechnologytoprovefirst that the

expres-siondefect iscloselylinkedtotheglllgeneitself and second thatitisnotmediatedatthelevel oftranscription,but rather

reflectsadefect inproteinlocalization, mostlikelyduetoa signal sequence mutation. Given that the gIll protein plays animportantrolein virusadsorption (20)and release(18,20, 22), aswell asbeing a major immunogen (2, 9), it is likely that the defective localization phenotype is yet another

deficiency contributing totheavirulence of PRV-Ba. MATERIALS AND METHODS

Cells, virus, and DNA. The growth and properties of the Becker strainofPRV(PRV-Be) and theporcine kidneycell line PK15 have beenpreviously described(16). PRV-Ba and PRV strain Ka (PRV-Ka) were generous gifts from T. Ben-Porat.

Cloning of the PstI fragments containing the gIll gene from PRV-Be, PRV-Ba, and PRV-Ka intoEscherichia coli pBR322plasmidswasdoneasdescribed previously(17). The structure ofeachplasmid was verified by restriction

endo-nuclease analysis.

Construction ofisogenic setofPRV-Be strains containing

gIlI gene ofPRV-Ba and PRV-Ka.Recombinantviruseswere obtainedby the calcium phosphate cotransfection and gene replacement techniques described previously (17, 18). Fig-ure1outlines the gene replacement strategy. Briefly, E. coli

plasmids containingthe4.3-kilobasepair (kbp) PstI fragment

harboringthe

gIll

genewerecleavedwith PstI and cotrans-fected with PRV10 DNA. PRV10 is a virus lacking a func-tional gIll gene (17). Itcontains adeletion ofa 1,480-base-250

on November 10, 2019 by guest

http://jvi.asm.org/

(2)

gillmRNA

Pstl Xhol Xhol Sacl Sacl Xhol

I

(ClonedPRV-gIIl Gene) ` /

Pstl Xhol Xhol Xhol Psti

-/

I_I_I___

PRV10: aglll Deletion

FIG. 1. Construction of isogenic derivatives of PRV-Be carrying theglll gene of PRV-Ba and PRV-Ka. The topline indicates the cloned4.3-kbp PstI fragment from PRV-Be, PRV-Ba, orPRV-Ka.

Relevant restrictionenzymecleavage sites areindicated. The glll

gene ofPRV-Ka contains an additional Sacl cleavage site. The

shaded boxdelineates theglll-codingsequences.Thearrowabove thebox indicates the approximate startsite, direction of transcrip-tion, andstopsite of the gIll mRNA. The bottom line indicates the homologous regiononthe glll nullmutantvirus, PRV10 (17). The

gll mutation is a deletion of the indicated XhoI fragment. The

potentialcrossoversrequired forgenereplacement and

reconstruc-tion ofanactiveglllgene areindicated by largeXs.Themethod for

genereplacement is described in Materials and Methods.

pair (bp) XhoI fragment removing 230 bp of the upstream glll promoter region and 87% of the gIll coding sequence.

PRV10 produces no detectable glll protein and forms a

nonreactive orwhite plaque in the black-plaque assaywith

glll-specific antibodies followed byaperoxidase-linked

sec-ond antibody (17). In the same assay, Ml antibody gives reactiveorblack plaques for the parentalPRV-Be, PRV-Ba,

and PRV-Ka. After cotransfection, virus plaques were

screened in the black-plaque assay for production of the

parentalglll protein. With monoclonal antibody Ml, black plaques werefound atfrequencies of 5% or more from all three cotransfection lysates, and one such black-plaque

recombinant from each was picked for further analysis.

Viruseswere namedas follows: PRV20(glllgene of

PRV-Be), PRV21(gIIIgeneof PRV-Ka), and PRV22(glllgeneof

PRV-Ba). In these three recombinant viruses, we can be confidentthatatleastthe1,480-bpXhoIfragment lackingin PRV10 has been replaced by DNAprovided by the cloned PstI fragment. It is highly likely that more DNA upstream and downstream of the XhoI fragment also has been

re-placed.

DNA and RNA analysis. Viral DNA preparation and Southern blot analysis of viral DNA with a gIII-specific

32P-labeled probe were performed as described previously

(15, 17). ForDNA sequencing, a 1.1-kbp XhoI-KpnI

frag-ment containing the putative promoter and 5'-coding

se-quence of glll was cloned into M13mpl8. The dideoxy

method (19) was used to sequence the glll promoter and signal sequence coding region with asynthetic

oligonucleo-tideprimerthathybridized upstreamof thegene (6). Theisolation of totalcytoplasmicRNAfrom infected cells andsubsequentNorthern(RNA)and slot-blotanalysiswere

conducted aspreviously described (6, 17).

The invitrotranslation of totalcytoplasmicRNAobtained fromPRV-infectedcells has been describedpreviously (17). Translatedextractswereresolvedbysodiumdodecylsulfate (SDS)-polyacrylamide gel electrophoresis or used for the immunoprecipitation ofglll-specific proteins with 282 anti-serum(18).

Fractionation and immunoprecipitation of PRV glycopro-teins. Forglucosamine labelingofglycoproteins,PK15cells

were infected at a multiplicity of infection of 10 with the

indicated viruses and grown throughout a 16-h infection in mediumcontaining 100,uCi of [3H]glucosamine per ml. For

cysteine labeling of proteins, a similar protocol was used exceptthat 100,uCi of [35]cysteine per ml was present from 6 to 16 h postinfection. Infected cell, virion, and medium fractionation ofPRV-infected PK15 cells was performed as previously described (18). The preparation of infected cell extractsand the immunoprecipitation procedure have been described previously (18). As indicated in the text, some samples were denaturedby boiling priortoadding antibody

to improve 282 serumreactivity.

Antibody reagents. The antibodies used in these studies included mouse monoclonal antibodies Ml and M16 (reac-tiveonlyagainstthe nativeglycosylatedforms ofgIll; these

antibodies do not react with in vitro-translated gIll or intracellular nonglycosylated gIll [6, 9]), goat polyvalent

anti-gIll antiserum 282(reactivewith both native and dena-tured forms of glycosylated or nonglycosylated gIll [18]), and goatpolyvalentanti-gIIantiserum 284 and mouse

mono-clonalantibodyM3(bothreactive onlyagainstthe gIIfamily ofglycoproteins [9, 18]).

Pulse-chaseanalysis ofgIII protein. The pulse-chase pro-cedure used has beendescribedpreviously (18; J. P. Ryan, M. R.Whealy, A. K.Robbins,C. L. Keeler, Jr.,andL. W.

Enquist, UCLA Symp. Mol. Cell. Biol., in press). Briefly,

PK15cellswereinfectedatamultiplicity ofinfection of10, andat 4 hpostinfection, cells were pulse-labeled for2 min with 100 ,Ci of

[35S]cystine

per ml. The radiolabel was

removed, and the cells were incubated in the presence of

excess nonradioactive cysteine and methionine for various times. At the desired chase times, media (containing re-leasedvirus) andinfected cellfractions wereharvested and thegIIIspecieswasimmunoprecipitated with 282serum. In these experiments, the extracts were denatured prior to

immunoprecipitationwith 1.0%SDS and 10mM

dithiothrei-tol.

Black-plaque assay. The black-plaque assay has been

described previously(6).

RESULTS

Rationale.Wesoughttodetermine thecauseofthe appar-entlow

gIll

expression by attenuated vaccine strain PRV-Ba. Wereasonedthatsuchanexpressiondefect could either be linked to the gIII gene itselforbe caused byone ofthe manyknown defects ofPRV-Ba. We therefore deleted the gIII gene andits promoterfromPRV-Be,astraincontaining

awell-characterizedandfully expressedgIII gene(6, 15-18,

22; Ryanetal.,inpress) andreplacedit withthe

gIl-coding

region of PRV-Ba. As controls, we also replaced PRV-Be gIII gene with the parental PRV-Be

gIll

gene and the

gIll

gene of the widely used PRV-Ka strain. If the reduced expression phenotype was linkedto the gIII gene, itwould

manifestitselfin therecombinant virus.

If thedefectwaslinkedto

gIll,

it could be duetoreduced transcription or RNA

accumulation,

reduced translation efficiency,poor

immunoreactivity owing

toalterations inthe

gIll

proteinitself,or someposttranslationaldefectincluding proteolysis or improper protein localization. Using these

isogenic strains, we

performed

a setof

experiments

exam-iningthesteady-state levelsof

gIll-specific

RNA,the

ability

of this RNAtobetranslated in

vitro,

the

steady-state

level,

location, and immunoreactivity of

gIll

proteins,

and the

pulse-chasekinetics of the gIII

primary

translation

products.

Constructionofan isogenicsetof PRV-BestrainswithgIII

genes from PRV-Ka and PRV-Ba. The gene replacement

on November 10, 2019 by guest

http://jvi.asm.org/

[image:2.612.55.293.76.171.2]

(3)

strategyis describedinFig. 1and inMaterialsand Methods. PRV-Be, PRV-Ka, and PRV-Ba all expressed detectable amounts ofgIllon the surface of infected cells since these

viruses formed black plaques with glll-specific Ml mono-clonal antibody. We used this observation to screen for

isogenic viruses with the desiredgIII genereplacementafter

cotransfection ofPRV10(17) with cloned fragments

encod-ing theglllgenes ofPRV-Be, PRV-Ba, PRV-Ka. Thehigh frequency ofgene replacement

(>5%)

reflected the strong

growth advantage of viruses expressing gIII over viruses

lacking a functional gIII gene (22). The isogenic set of viruses derived was PRV20 (gIII geneofPRV-Be), PRV21

(gIllgene ofPRV-Ka), and PRV22 (gIll gene ofPRV-Ba). The viral DNA was digested with a variety of restriction

enzymesand analyzed by Southern blothybridizationwith

the

32P-labeled

4.3-kbpPstIfragment ofPRV-Be. Withthe

exception ofSacI, no restriction fragment polymorphisms

were detected (data not shown). BothgIII-Be (PRV-Be or

PRV20) and gIII-Ba (PRV-Ba or PRV22) contained two

apparently identicalSacl sites within thebody ofthe gene,

while gIII-Ka (PRV-Ka or PRV21) contained an additional

Sacl site. Fragments from all viruses hybridizing to the PRV-Be gIIIprobe appeared with the same intensity,

indi-catingno major sequence divergence. There were no

obvi-ousdeletionsorrearrangementsofthe4.3-kbpPstIfragment

amongtheviruses.

Analysis of total cytoplasmicRNAfrom PRV-infected cells. To determine whether transcription of the gIII gene was

altered in PRV-Ba or PRV22, we studied the quantity and

quality of

glll-specific

RNA by using Northern blot and

slot-blottechniques. Total cytoplasmicRNAwasextracted fromPK15cellsinfected withtheindicated viruses. Estima-tionof the quantity of

glll-specific

RNAwas accomplished with a slot-blot format. The purified RNA (10 ,ug) was

serially

dilutedtwofold, denatured,andtransferredto nitro-cellulose membranes inaslot-blotapparatusasdescribedin

Materials and Methods. The quality of

glll-specific

RNA was determined by Northern blotting. For these studies, 5 ,ug of each RNA sample was analyzed as described in

MaterialsandMethods.Theradioactiveprobeusedfor both

blots was an NcoI-BamHI fragment containing only gIII gene sequencesfrom PRV-Be.

Autoradiograms ofthe slotblots

(Fig.

2) are shown. The

slot-blot analysis indicatedthat the amount ofsteady-state gIII-specific RNA was nominally the same for all viruses tested. The experimenthas been repeatedusing RNA

pre-pared at different times after infection with similar results (datanotshown). The Northern blotanalysis indicatedthat anidentically sized species of gIII-specificRNAcomigrating with the 1.55-kilobase PRV-Be RNA was found for all

viruses (datanot shown).

In vitro translation of total RNA extracted from infected PK15cells.The sameRNAanalyzed by slot-blot and North-ern analysis was used for in vitro translation experiments (Fig. 3). RNA from uninfected cells was included as a control. The polypeptides translated in vitro were labeled

with [35S]methionine. Either in vitro-translated samples werefractionateddirectly on a 10% polyacrylamide-SDS gel

(Fig.3, left panel), or gIII-specific translation products were

immunoprecipitated with polyvalent 282 serum and then

fractionatedon a10% polyacrylamide-SDS gel (Fig. 3, right

panel). Translationproducts from uninfected cell RNA are in

lanesmarked C.

By examining the intensity of bands in the

nonimmuno-precipitatedsamples (seven lanes on the left), it is clear that all samples received approximately the same amount of

tt K,,i R i 20 2 1 22

mom_{_,} m .M~ mum

_ __

_.

-. - 4p1

FIG. 2. Slot-blotanalysis ofglll-specific RNA. Total cytoplas-mic RNA was extracted from PRV-Be-, PRV-Ka-, PRV-Ba-, PRV20-, PRV21-, or PRV22-infected PK15 cells as described in Materials and Methods. A10-p,gsample of eachRNAwasserially dilutedtwofold, denatured, and subsequently transferred to nitro-cellulosein aslot-blot apparatus. The probe used was a 1,400-bp NcoI-BamHIfragment from pALM3 (17)thathybridizedonlytothe body of the glll gene. Lanes: Be, PRV-Be; Ka, PRV-Ka; Ba, PRV-Ba; 20, PRV20; 21, PRV21;22,PRV22.

RNA, althoughthe pattern forPRV-Ba isdistinctlydifferent fromthat forPRV-Be orPRV-Ka. Fortheisogenic strains, PRV20, PRV21, and PRV22, the patternswere identicalto thatfound forPRV-Be,asexpected.Itisnoteworthythata novelprotein product tightly linked to gIII appeared to be alteredinbothPRV-KaandPRV-Ba.This is best seen in the left panel comparing the last three isogenic samples. The

profilesareidentical except for theindicated band(shownby adot) migrating slightly fasterfor PRV21 and PRV22, but notPRV20. The identity of this proteinis unknown.

Itis clear thatfor all viruses,282 antiserum immunopre-cipitated proteins of approximately 58-kilodaltons, charac-teristic of thepredicted 479-amino-acid gIII primary trans-lationproduct (15, 17). However,forPRV-Ka, PRV-Ba,and theisogenicstrainsPRV21 andPRV22,theprimary transla-tionproduct migrated slightlyfaster than that from PRV-Be orPRV20. This may reflect differencesin amino acid com-position or may indicate that the genes contain small in-frame deletions. Comparedwith the amountof PRV-BegIII

immunoprecipitated by282 serum,bothPRV-Ba and PRV22 RNA yielded about two- to threefold-reduced amounts of immunoreactive gIII. Given that theamountsofgIII-specific

RNA in these preparations were essentially identical (Fig. 2), this result suggests a lower translation efficiency ofthe PRV-Ba gIII message in the rabbitreticulocyte system.

Immunoprecipitation of glucosamine-labeled PRV-specific

proteins fromvirus particles. PK15 cells were infectedwith

PRV-Be, PRV-Ka, PRV-Ba, PRV20, PRV21, orPRV22 as described in Materials and Methods. [3H]glucosamine was added at the time of infection to label the glycoproteins.

After 16 h, medium containingextracellular virus particles

was collected and the virus was purified by centrifugation

and analyzed as described in Materials and Methods.

Re-maining infected cellswere used in thesubsequent

experi-ment. Thepurifiedvirusparticles wereanalyzed directlyor

lysed and reactedwith gIII- orgll-specific antibodies. The

immunoprecipitates were collected and fractionated on a

10% polyacrylamide-SDS gel. The3H-labeled polypeptides detectedby fluorography aredisplayedin Fig. 4.

ThegIII defectof PRV-Ba is demonstrated inFig.4Awith the Ml monoclonal antibody. Reduced amounts of the mature, 92-kDa form ofgIll were observed in PRV-Ba as

on November 10, 2019 by guest

http://jvi.asm.org/

[image:3.612.337.547.76.217.2]

(4)

C Be Ka Ba 20 21 22 C Be Ka Ba 20 21 22

68

-46

-30

--68

-46

-30

FIG. 3. In vitro translation of total cytoplasmic RNA extracted from PRV-infected cells. Total RNA from mock-infected and PRV-infected cellswasharvested at 16 h postinfection, and equal amounts were translated in a rabbit reticulocyte system asdescribedin Materials and Methods.Aportion of these products wasimmunoprecipitated with polyvalent 282 serum. Both the total translation mixture (left panel) and theimmunoprecipitate (right panel) of each sample were fractionated on a 10% polyacrylamide-SDS gel.[35S]methionine-labeledpolypeptides werevisualized by fluorography. The dot adjacent to bands from the total translation mixture indicates a novel,non-glll-relatedtranslation productdiscussed in the text. Lanes: C, mock-infected control; Be, PRV-Be infected; Ka, PRV-Ka infected; Ba, PRV-Ba infected; 20, PRV20 infected; 21, PRV21 infected; 22, PRV22 infected. Numbers on the sides show molecular mass in kilodaltons.

compared with PRV-Be or PRV-Ka. The defect was

elabo-rated, if not more severe, in the isogenic derivative PRV22

(compare PRV22 with PRV20 or PRV21) and therefore is

tightly

linkedtothePRV-BagIl geneitself.Note that in this

experiment, fewer PRV-Be virions were recovered after purification. Thiscanbededucedby examining the intensity

of the bandsinFig.4D, the controlprofiles ofthegll protein immunoprecipitated fromthe samesample, or, from Fig. 4E,

thetotalglucosamine-labeled profilesofpurifiedvirus parti-cles.

It hasbeen reportedthat PRV-Ba-encoded gIII does not reactwithasecond

glll-specific

monoclonalantibody, M16

(2).This lack of reactivity isclearlydemonstratedinFig. 4B for both parental PRV-Baandthe isogenic strain, PRV22.

Figure 4C shows the reactivity with polyvalent 282 anti-serum. Note that PRV-Ba anditsisogenic derivativePRV22

reacted with thisserum,althoughthe amountofgIIIprotein precipitated wasreduced. Thisreductionisbest seen in the

comparison oftheisogenic strains.

Figure 4D is a control immunoprecipitation of the gIl family of glycoproteins from the same virion samples to

indicate efficiency of infection andsample recovery.

TheexperimentinFig.4Bindicated that thePRV-Ba gIII

protein lacked the M16 epitope. This observation could

indicate that the PRV-Ba glll protein, rather than being poorly expressed, was simply less immunoreactive in gen-eralwithgIII-specific antisera. However,the data inFig.4E

strongly suggested that the gIII protein of PRV-Ba was present inphysically lower amounts in purified virions. In this experiment, purified [3H]glucosamine-labeled virions werefractionated directlyon a10%polyacrylamide-SDS gel

withno immunoprecipitation. Comparethe last three lanes

containing the isogenic strains and note the reduction of

intensityof the92-kDaspecies for PRV22 and PRV-Ba.

We conclude that the apparent reduced expression phe-notype ofgIll seenfor PRV-Ba is tightlylinked to thegIII gene, and in addition, we corroborate the observations of Ben-Porat et al. (2) that the PRV-Ba gIII gene encodes a mutant glycoprotein with altered immunoreactivity to the M16monoclonal antibody.

Immunoprecipitation ofglucosamine-labeled, PRV-specific proteins from infected PK15 cells. In the previous experi-ment, we showed that PRV-Ba glll protein is reduced in virus particles. The next set of experiments examined the infected cells from the sameinfection. The experimentwas doneasdescribedabove. Afterremoval ofthemedium, the infected cells were lysed (see Materials and Methods) and reacted with gIII- orgll-specific antibodies. The immuno-precipitateswerecollected andsubsequently fractionatedon a10%polyacrylamide-SDS gel. The 3H-labeled polypeptides detectedby fluorography aredisplayedinFig.5.

Figure5A shows that allvirusesproduced matureglllin infected cells reactive with Ml monoclonal antibody, al-though PRV-Ba and PRV22 consistently gave lower amounts. The faster-migrating band of 62 kDa, more pre-dominant inPRV-Ba-infected cells(indicatedbyadot),was seenin theother infectionsaswell afterlongerexposures.It was not seen with M16 monoclonal antibody or with 282 polyvalent serum. It could also be precipitated from [35S]cysteine-labeled cell extracts by usingonly Ml mono-clonalantibody(seebelow). ItsrelationshiptogIll,ifany,is

currently notunderstood.

The M16 monoclonal antibody recognizes not only the mature92-kDa form ofgIII, but also the 74-kDaprecursor (18). Although PRV-Be, PRV-Ka, and theirisogenic deriv-ativesgaveindistinguishablepatternsofimmunoprecipitated

glllprotein,PRV-Ba andPRV22hadno detectable

reactiv-itywith thisantibody (shown inFig. SB).

so

.40

*R.

0

on November 10, 2019 by guest

http://jvi.asm.org/

[image:4.612.140.467.75.310.2]

(5)

13: Ka F:a:1 :" i

.... Ci a

f.;.1P

Be Ka Baz 20 21 22

--i10

_~~~~~~~~~~~~~- 1.

Wa! .WW.

::

....C..

.,Wm

74--92..

.2

-AA. kllis0 _1

t@,, :!

[image:5.612.70.310.76.373.2]

9-o

3 :iie-u

FIG. 4. Immunoprecipitation of [3H]glucosamine-labeled,

gIll-specific proteins from purified virions. PK15 cellswereinfected with

PRV-Be, PRV-Ka, PRV-Ba, PRV20, PRV21, or PRV22 in the

presence of[3H]glucosamine asdescribed in Materials and

Meth-ods.Extracellularvirus particleswerepurified by centrifugation and

analyzed with glll- and gll-specific antibodies. (A) Ml monoclonal antibody. (B) M16 monoclonal antibody. (C) 282 polyvalentanti-glll

serum.(D) M3 monoclonalantibody (gll specific, control). (E) Total

profile ofglucosamine-labeled virions priortoimmunoprecipitation. Immune complexes were collected by adsorption to Pansorbin (Staphylococcus aureus) and fractionated by electrophoresison a

10% polyacrylamide-SDS gel. 3H-labeled polypeptides were

de-tected by fluorography. Thevirus strain is indicated above the lanes in panels A and D and is in thesameorder in panels B, C, and E.

Numbers showmolecularmassinkilodaltons.

The polyvalent glll antiserum 282 reacts with both the precursor andthe matureform ofgIII fora variety ofgIll

mutants(6, 18; Ryanetal., inpress). The data given in Fig. 5C demonstrate that all viruses produced precursor and matureforms ofglll reactivewith thissera, although PRV-Ba and PRV22 clearly produced less 74-kDa protein. The criticalcomparisonis of theisogenic strainsinthe lastthree lanes of Fig. 5C. It is clear that PRV22 produced less immunoreactive precursor and mature glycosylated gIII proteinthan eitherPRV20orPRV21.

The control immunoprecipitations of gll glycoprotein (Fig. 5D) were similar, indicating that all infections had progressed essentially to the same state. The 100-kDa

pre-cursor form, not seen in purified virions (Fig. 4D), was

precipitated for all viruses. Wenotethat thegll profile from PRV-Ba, butnotPRV22,was slightly different qualitatively

andquantitatively from all other virusesasobserved

previ-ously (2), suggesting that PRV-Ba has mutations ingll as

well(see Discussion).

Immunoprecipitation of PRV-specific proteins from [35S] cysteine-labeled infected PK15cells. Asimilar setof

steady-FIG. 5. Immunoprecipitation of [3H]glucosamine-labeled,

gIll-specific protein from infected cellextracts.PK15 cellswereinfected withPRV-Be, PRV-Ka, PRV-Ba,PRV20, PRV21, orPRV22 in the

presence of[3H]glucosamine asdescribed in Materials and Meth-ods. The infected cell extracts were reacted with gIII- and

gIl-specific antibodies. (A) Ml monoclonal antibody. (B) M16

mono-clonal antibody. (C) 282 polyvalent anti-gIll serum. (D) M3 monoclonal antibody (gIl-specific control). Immune complexes

werecollectedby adsorptiontoPansorbin(S. aureus) and fraction-ated by electrophoresis on a 10% polyacrylamide-SDS gel.

3H-labeled polypeptides were detected by fluorography. The virus strain is indicated abovepanels A and D and is in thesameorderfor panels B and C. The 92- and 74-kDa forms ofgIll are indicated. Numbers on the right side of panel D show molecular mass in kilodaltons.

state labeling experiments was done with [35S]cysteine rather than [3H]glucosamine as the label. The experiment willdetect all stablegIlI-immunoreactive species, including those that are not glycosylated. Cells were infected with PRV-Be,PRV-Ba, orPRV22, labeled from 6to 16 h postin-fection as described in Materials and Methods, and subse-quentlyfractionated into released virions and infectedcells. Extracts of infected cells or virions were reacted with anti-gIII-specific antibodies, and immunoprecipitates were collected and then fractionated on a 10%

polyacrylamide-SDS gel. The 35S-labeled polypeptides were detected by

fluorography. The results with infected cell extracts are shownin Fig. 6.

Ml monoclonal antibodyimmunoprecipitated the mature 92-kDa form ofgIll from each infected cell extract. How-ever, both PRV-Ba- and PRV22-infected cells clearly con-tained lessimmunoreactiveprotein,consistent withprevious observationsfor[3H]glucosamine-labeled infected cells(Fig. 4). (The sharpband ofapproximately 92 kDapresentin Ml and M16 immunoprecipitates of infected cell extracts is a

coprecipitated, nonglycosylated, non-gIll-related protein [datanotshown].It is notseeniftheextracts aredenatured

priorto immunoprecipitation, aswasdone with 282 serum.

However, asnoted inMaterials andMethods, Ml andM16 monoclonal antibodies do not react with denatured gIll

protein, hence nondenatured extracts must be used in this experiment.)Abandof about 62 kDawas morepredominant

for PRV-Ba, but could be seen in other infections upon

longerexposures.Aspecies ofthesameapparentmolecular

mass was observed previously in [3H]glucosamine-labeled

-- 68

55 iii;; /.% 1. -)

Aik. .0":

...00

.4 40 A-m,

on November 10, 2019 by guest

http://jvi.asm.org/

[image:5.612.323.565.78.272.2]

(6)

___ .A --M

e Ba £2 Ile Ba 2121 Be Ba 22

94-74-

I.

58-A

2 ..

533

92

.74

..58

B 0 30 45 60 90 120 240 480 1N.0 I * ,-

-N

v.

.-..

FIG. 6. Immunoprecipitation of[35S]cysteine-labeled PRV gly-coproteins. PK15 cells were infected with PRV in the presence of [35S]cysteine as described in Materials and Methods. Extracts of PRV-Be-, PRV-Ba-, or PRV22-infected cells were reacted with monoclonal antibodies Ml and M16 or polyvalent 282 serum as indicated. Immune complexes were collected by adsorption to Pansorbin (StaphA) and fractionated by electrophoresis on a 10% polyacrylamide-SDS gel.I'S-labeledpolypeptides weredetected by fluorography. Extracts from infected cells were denatured prior to immunoprecipitation with polyvalent 282 serum as described in Materials and Methods. Numbers show molecular mass in kilodal-tons.

92-7

4-

58-C

0 30 45 60 90 120 240 480

i2

*w ^*

infected cells (Fig. 4). As discussed, its relationship, if any, togIII is under study.

M16 monoclonal antibody immunoprecipitated only the

92-kDamatureform ofgIIIand the 74-kDa precursorspecies fromPRV-Be-infectedcells. This same monoclonal antibody

did not immunoprecipitate any gIII-specific proteins from

either PRV-Ba- or PRV22-infected cells, consistent with

observations with [3H]glucosamine-labeled infected cells (Fig. 4).

282 serum immunoprecipitated only the 92-kDa mature

form and the 74-kDaprecursorform ofglllfrom

PRV-Be-infected cells. In contrast, a significant fraction of the

282-reactive

"S-labeled

polypeptides in PRV-Ba- or

PRV22-infected cells migrated with an apparent molecular mass of 58kDa. Thisspecieswasnotapparentin similarexperiments

with

[3H]glucosamine

labeling (Fig. 4). The normal 92-kDa matureform and the74-kDa precursorspecieswerepresent,

but atreduced levels.

The profiles for [35S]cysteine-labeled virions were also

examined (data not shown). The results were essentially identical to those obtained for [3H]glucosamine-labeled

vi-rions (Fig. 4) in that both PRV-Ba and PRV22 virions

contained lesscysteine-labeled gIII than PRV-Bedid. Weconclude that the [3H]glucosamine-and [35S]cysteine-labeled

gIll

profiles from infected cellsareessentially

iden-ticalwithonemajor exception.PRV-Ba-andPRV22-infected cells containanew,58-kDamajor specieslabeledonlywith

[35S]cysteine andreactive onlywith 282 serum. We suggest that this species is a product of the PRV-Ba glll gene

because it isproducedbyPRV22 and notPRV-Be.

Pulse-chaseanalysis ofgIllprotein. Thelong-term

labeling

experiment presented above defines stable gIII

species

whoseintensityoflabelingis dictated by synthesisrateand turnoverrate. Thefollowing experimentexamined therates of

synthesis

and

stability

of

[35S]cysteine-labeled glll

pro-FIG. 7. Pulse-chase analysis of gIII species produced by PRV-Be,PRV-Ba, andPRV22. PK15cellswereinfectedwith10 PFUof theindicated virus percell andincubatedat37°C. The pulse-chase protocol is described in Materials and Methods.At 4 h postinfec-tion, cellswerepulse-labeled with[35S]cysteinefor2min and then incubated in the presence ofexcess nonradioactive cystine and methionine. Atthe times indicated above each lane (in minutes), cells were harvested. glll species wereimmunoprecipitated from infected cell extracts with polyvalent 282 serum (A, B, and C). Immunoprecipitates were resolved on a10%)o polyacrylamide-SDS gelandvisualized byfluorography. (A)PRV-Be; (B) PRV-Ba; (C) PRV22.

teins produced by thevirus strains and determinedwhether anytransient, processedforms of theprotein existedin the

infected cell that might not be detected under steady-state radiolabeling conditions. Ofparticular interest was the 58-kDa gIll species made by PRV-Ba and PRV22 seen in

steady-state [35S]cysteine labeling (Fig. 6), but not in [3H] glucosamine labeling (Fig.4and5).PK15cellswereinfected with PRV-Be, PRV-Ba, and PRV22. A pulse-chase

experi-mentwith [35S]cysteine wasperformedat4 hpostinfection

asdescribed in Materials and Methods.gIII-specificproteins were immunoprecipitated with polyvalent 282 serum. The resultsareshown inFig. 7.

Asnotedpreviously (6, 18; Ryanetal.,inpress),the initial

product ofglll translation observed in a 2-min pulse for PRV-Be was the 74-kDa species known tobe modified by high-mannose, N-linked

glycosylation

characteristic of a

on November 10, 2019 by guest

http://jvi.asm.org/

[image:6.612.64.285.72.248.2] [image:6.612.359.511.78.437.2]

(7)

location in the endoplasmic reticulum. Export of PRV-Be gIIIfrom theendoplasmicreticulum tothe Golgiapparatus

wasrapid and efficient; the 92-kDamatureformappearedas earlyas30min after pulse-labeling, and onlyasmall fraction

of the74-kDaprecursor remained aftera1-h chase.

The pulse-chase profile obtained for PRV-Ba or PRV22

was strikingly different. First, the predominant primary

translation product observed after the 2-min pulse was a

58-kDa form. The 74-kDa form was present but barely detectable. The small quantity of 74-kDa protein chased to themature92-kDaspecies with essentially identical kinetics

as PRV-Be-encoded gIII. The 58-kDa species remained

essentially constant during the first 2 h of the chase period but showed significant loss of signal toward the end of the 8-h chase. We were struck by the similarity between the PRV-Baand PRV22 pulse-chase profiles and those found for definedglll signalsequencemutantsof PRV-Be constructed by Ryanetal. (in press). Certain signalsequence mutations partially blockcotranslationalexportof the gIII protein from the cytoplasm to the endoplasmic reticulum. The 58-kDa form is believed to be the authentic, nonglycosylated -pri-marytranslation productasidentified by in vitro translation (6, 15, 17; Ryan etal., in press).

For PRV-Ba and PRV22, the intense labeling of the 58-kDa form compared with other forms of gIII is notewor-thy. We note that the amount of cysteine available to the cellsduring the 2-min pulse was not in excess (addition of more radioactive cysteine resulted in more incorporation [unpublishedobservations]). Therefore, the intensity of label inthe58-kDa speciessuggeststhat itwasbeingsynthesized

atanincreasedrate compared with the 74-kDaglycosylated precursor.

Weconclude from theseexperiments that the gIIIgeneof PRV-Baand PRV22 contains adefect giving risetoanovel

primary product of translation not characteristic of the membrane-boundglycosylated 74-kDaprecursor.Themajor species of gIIIobserved ina2-minpulseat4hpostinfection is a 58-kDaprotein with the sameapparent massas the in

vitro-translatedgeneproduct.This species is translatedatan

increasedrate compared with the normal74-kDaprecursor.

The reduced amount of normal 74-kDa precursor made is converted to mature 92-kDa product at the same rate as

normal gIllandappearswith time in virusparticles (datanot shown). The PRV-Ba 58-kDa aberrant gIII precursor is a

dead-end product and slowly turns over. We cannot com-paredirectly theamountof 58-kDaprecursor seenin

steady-statelabeling (Fig. 6) and the apparent amount seen in the

pulse-chase experiment. In addition, wehave notmeasured the half-life of the aberrant precursor at late times of infection.

DNA sequence analysis of promoter and amino-terminal coding regions of PRV-Ba gIll. The datafrom the pulse-chase experimentswere characteristic of the profiles observed for

certain defined signal sequencemutantsin the PRV-Be gIll gene (6; Ryan et al., in press). We examined the DNA sequenceof thepromoter and 5'region ofthe PRV-Baglll gene to determine whether we could identify a potential

signal sequence mutation. The gIll signal sequence was

localizedtothe amino-terminal 22amino acidsbyEnquistet

al. (6) and Ryanet al. (inpress).The DNA sequenceof the

PRV-Ba gIII promoter region and signal sequence was

determined from -60to +93 (+1 is the A ofthe initiator ATG codon)asdescribed in Materials andMethods (datanot shown). Theupstreamregions ofPRV-Be andPRV-BagIll including the predicted promoter were identical. However, we notedasingle base changeofaTtoaCatnucleotide 41

ofthe

gIll

gene.This

change

in PRV-Ba convertedaleucine

codon

(CTC)

to a

proline

codon

(CCC)

in the

predicted

hydrophobic

coreofthe

glll signal

sequence. Thiswas the

only base

change

in the signal sequence. We have not

sequenced thewhole PRV-Ba

glll

gene,butnotethat there arefurther sequence

changes

in thebodyof the PRV-Ba

glll

geneoutsideof the

signal

sequence

region.

Forexample,the nextchange identified inoursequencewas anA-to-C

change

at nucleotide89ofthe

glll

gene. This would

change

codon 30 from an

asparagine

codon

(GAC)

to a

glycine

codon

(GCC).

DISCUSSION

PRV-Ba isanattenuated vaccinestrain

harboring

multiple

mutations. Many ofthese

changes

and their effecton PRV

virulence have been documented (3, 5, 7,

8,

10, 11, 13,

14,

20,21).Onewell-knownbutpoorly characterized

phenotype

ofthePRV-Bastrain isthe apparentlow-level

expression

of

glycoprotein

gIll

(2).

We used the

technique

ofgene

replacement

to compare the PRV-BagIIIgenewithtwonon-vaccine strainalleles in an

isogenic

background.

Inthis way, wewere ableto

study

only those mutations

tightly

linkedto thegIII gene. Three PRV-BagIII defectswere definedin this manner. The

first,

a lack of

reactivity

with the M16 monoclonal

antibody,

is well documented (2). Clearly, there are mutations in the

bodyof the PRV-Ba

gIll

gene

affecting

the M16

epitope.

We know little about the natureofthesemutations. The second

defectis anapparenttwo- tothreefold-reduced

efficiency

of

in vitro translation ofPRV-Ba

glll

mRNA. At the moment, it is unclear whether this defect is manifested in vivo or is

peculiar

to

purified

RNA translated in vitro. We note that DNA sequence

analysis

revealed no alterations in the pro-moter

region

and 5'

region

ofthe

glll

mRNA excepta

single

base

change

at codon 14.

The most

striking

defect of PRV-Ba

glll,

however, was novel precursor

synthesis

and

subsequent

effectson

protein

localization.

The

protein

had been shown

previously

to be present in reduced amounts in virus

envelopes

(2).

We

confirmed these observations and also demonstrated that there was no apparent defect in

expression

of

gIII-specific

mRNA. A

key

observation was that the

predominant

PRV-Ba

gIll

translation

product

detectedina2-min

labeling

with

[35S]cysteine

was58

kDa,

the

predicted

size ofthe

primary,

unmodifiedgene

product.

This 58-kDa

species

was

reason-ably

stable and could

easily

be detectedaftera

9-h,

steady-state

labeling experiment

with

[35S]cysteine.

In contrast,the

58-kDa

species

was notdetected

during

a

16-h, steady-state

labeling

experiment

with

[3H]glucosamine.

Moreover, the

pulse-chase analysis

indicated that the 58-kDa

species

was translated at a

higher

rate than the normal 74-kDa

glycosy-lated precursor. The 58-kDa

gIll

proteinistherefore

nongly-cosylated

and

synthesized by

moreefficienttranslation

ma-chinery.

This phenotype can be best explained by another mutation in the PRV-Ba

gIll

gene, the leucine-to-proline

change

at

position

14 ofthe

signal

sequence. We noted in

pulse-chase experiments (Fig.

7)that theprofile forPRV-Ba

gIll

corresponded remarkablywelltothatseenforadefined

signal

sequence mutantdescribed by Ryanetal. (inpress). This mutation, 12R, altered theleucine atposition 12to an

arginine.

Codons12and 14areboth in the

hydrophobic

core of the

glll signal

sequence. Both mutations wereleaky, in the sense that a fraction of the

gIll

protein entered the

protein

export pathway to be processed and matured like normal

glll

protein. However,themajority of pulse-labeled

on November 10, 2019 by guest

http://jvi.asm.org/

(8)

12RorPRV-Ba

gIll

protein remained unglycosylated within the cell, most likely inthe cytoplasm. The small amount of maturegIIIproduced appeared on the cell surface, and some was incorporated into envelopes of egressing virions. There-fore, there is indeed a lower amount of gIll protein in PRV-Ba and PRV22 virus envelopes and cell membranes, but the cause is not reduced gene expression. Rather, because the gIII signal sequence is mutated, the primary gene product is localized improperly, and reduced amounts ofmature

gIll

protein are available for virion assembly. This mutation represents, to our knowledge, the first reported natural signal sequence mutation ina herpesvirus glycopro-tein.

It is important to note that the site-directed gIll signal sequence mutation at codon 12 described by Ryan et al. (in press)affects the hydrophobic domain of the signal sequence by introducing a charged amino acid. The change in PRV-Ba

gIll

at codon 14 introduces proline, a neutral amino acid that will alter the predicted alpha-helix structure of this domain, but should have little effect on the hydrophobic content. Work is in progress to determine the role of the hydrophobic region ingIII signal sequence function (J. P. Ryan, M. E. Whealy, A. K. Robbins, and L. W. Enquist, manuscript in preparation).

Why would such a specific defect occur during the process of making the attenuated vaccine strain? Is the signal se-quence mutation due to chance occurrence or could there be some rationale for its appearance? Most work must be done tounderstand the role ofgIIIin the life cycle of PRVbefore these questions can be answered. Nevertheless, it is clear that a signal sequence mutation represents one simple way to lower, but not totally deplete, the amount ofgIll protein in virus envelopes or on the surface of infected cells without affecting the mature protein structure. Reducing the amount of intracellulargIll that can enter the protein export pathway clearly leads to production of virions with reduced gIll in their envelopes. With defined signal sequence mutations, we can determine whether simply lowering the amount of ma-turegIII is sufficient for decreased virulence. By correcting the signal sequence mutation in PRV-Ba gIll, it will be possible todetermine whether the additional mutation(s) in the body of the gene affecting antibody recognition plays a role in virus release and virulence.

Schreurs et al. (20) suggested that roles in virus adsorption and release from infected cells constituted separate func-tions of thegIll protein. One line of evidence supportingthis hypothesis was that PRV-Ba, despite its reduced comple-ment of

gIll

in virions, absorbed normally to cells yet was defective in release of virus. The implication was that the PRV-Ba gIII protein was defective in one domain affecting release, while the adsorption domain remained intact. Our finding that the PRV-BagIII gene harbors a signal sequence mutation suggests an alternative testable hypothesis. Per-haps the processes of adsorption and virusreleasedifferwith respect to their concentration dependence on mature gIII. For example, the process of efficient virus absorption to cells could require very little gIll protein per virion, while the process of virus release could be more dependent on gIll concentration or could involve moregIll-dependent steps. A test of this hypothesis would be that reversion of the codon 14 mutation in PRV-Ba gIll (if it restored normal levels of gIII in the virus envelope) would result in a virus with normal kinetics of virus release.

Our studies support and extend previous observations indicating that other alterations exist in PRV-Ba. Ingeneral, we found that PRV22 was more defective than the parental

PRV-BaorPRV-Bestrains in theglll phenotypesmeasured.

This suggests that the PRV-Ba glll protein functions well

only in the PRV-Ba background, perhaps indicating that

PRV-Ba has developed compensatory mechanisms to deal with its alteredgIIIprotein.As suggested by Ben-Porat etal.

(2),ourobservationsontheapparentincreasein theamount

of gIl present in PRV-Ba virions but not in PRV22 may reflect a compensatory alteration of the gIl protein to ac-commodate the lower level ofgIIIprotein (and the absence ofgI and gp63).

These observations are circumstantial evidence that PRV

glycoproteins are involved in multiple protein-protein inter-actions. Perhaps some of the alterations made specifically

affecting virulence in PRV-Ba resulted in compensatory

changesinotherproteinstomaintain functionalglycoprotein complexes. The gene replacement technology used in this report and the virulence rescue techniques used by Ben-Porat and co-workers (13) provide tools to test this idea.

ACKNOWLEDGMENTS

We thank the members of the Molecular Genetics group for advice and discussion; in particular, Logan Buckley provided criti-cal advice and encouragement.

LITERATURE CITED

1. Bartha, A. 1961. Experimental reduction of virulence of Au-jeszky's disease virus. Magy. Allatorv. Lapja 16:42-45. 2. Ben-Porat, T., J. M. DeMarchi, B. Lomniczi, and A. S. Kaplan.

1986. Role of glycoproteins of pseudorabies virus in eliciting neutralizing antibodies. Virology 154:325-334.

3. Ben-Porat, T., J.DeMarchi, J. Pendrys, R. A. Veach, and A.S. Kaplan. 1986. Proteins specified by the short unique region of the genome of pseudorabies virus play a role inthe release of virions from certain cells. J. Virol. 57:191-196.

4. Ben-Porat, T., and A. S. Kaplan. 1985. Molecular biology of pseudorabies virus, p. 105-173. In B. Roizman (ed.), The herpesviruses. Plenum Publishing Corp., New York.

5. Berns, A., A.Van Den Ouweland, W. Quint, J. van. Oirschot, and A. Gielkens. 1985. Presence of markers for virulence inthe unique short region or repeat region or both of pseudorabies hybrid viruses. J. Virol. 53:89-93.

6. Enquist, L. W.,C. L. Keeler, Jr., A.K. Robbins, J. P. Ryan, and M. E. Whealy. 1988. An amino-terminal deletion mutation of pseudorabies virus glycoprotein gIll affects protein localization and RNA accumulation. J. Virol. 62:3565-3573.

7. Geilkens, A. L., J. T. van Oirschot, and A. J. M. Berns. 1985. Genome differences among field isolates and vaccine strainsof pseudorabies virus. J. Gen. Virol. 66:69-82.

8. Gustafson, D. P. 1975. Pseudorabies, p. 391-410. In H. D. Dunne and A. D. Leman (ed.), Disease ofswine,4thed. Iowa State University Press, Ames.

9. Hampl, H., T. Ben-Porat, L. Ehrlicher, K. 0. Habermehl, and A. S. Kaplan. 1984. Characterization oftheenvelopeproteinsof pseudorabies virus. J. Virol. 52:583-590.

10. Lomniczi, B., S. Watanabe, T. Ben-Porat, and A. S. Kaplan. 1987. Genome location and identification of functions defective in the Bartha vaccine strainofpseudorabies virus. J.Virol. 61: 796-801.

11. Mettenleiter, T. C., N. Lukacs, andJ.-J. Rziha. 1985. Pseudo-rabies virus avirulent strains fail to express amajor glycopro-tein. J. Virol. 56:307-311.

12. Mettenleiter, T. C., C. Schreurs, F. Zuckermann, and T. Ben-Porat. 1987. Role of pseudorabies virusglycoprotein gI in virus release from infected cells. J. Virol. 61:2764-2769.

13. Mettenleiter, T. C., L. Zsak, A. S.Kaplan, T.Ben-Porat, andB. Lomniczi. 1987. Role of a structuralglycoprotein of pseudora-bies in virus virulence. J. Virol. 61:4030-4032.

14. Petrovskis, E. A., J. G.Timmins, T. M.Gierman,and L. E. Post. 1986. Deletions in vaccine strains of pseudorabies virus and their effect onsynthesisof glycoproteingp63. J. Virol.

on November 10, 2019 by guest

http://jvi.asm.org/

(9)

1169.

15. Robbins,A.K., R. J. Watson, M. E.Whealy,W. W.Hays, and L. W. Enquist. 1986. Characterization ofa pseudorabies virus

glycoproteingenewith homologytoherpessimplexvirustype1 andtype2glycoprotein C.J. Virol.58:339-347.

16. Robbins, A.K.,J. H. Weis,L. W. Enquist, andR.J.Watson. 1984. Construction of E. coli expression plasmid libraries: localization ofapseudorabies virus glycoproteingene.J. Mol.

Appl. Genet.2:485-496.

17. Robbins, A.K., M. E.Whealy,R. J. Watson,and L. W. Enquist. 1986. Pseudorabies virusgeneencodingglycoproteingIIIisnot

essential forgrowth in tissue culture. J.Virol.59:635-645. 18. Ryan, J. P., M.E.Whealy,A. K. Robbins, and L. W. Enquist.

1987.Analysis of pseudorabies virus glycoproteingIll

localiza-tion and modification by using novel infectious viral mutants carryinguniqueEcoRI sites. J. Virol. 61:2962-2972.

19. Sanger, F., S. Nicklen, and A. R. Coulson. 1977.DNA

sequenc-ing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74:5463-5467.

20. Schreurs, C., T.C. Mettenleiter, F. Zuckermann, N. Sugg, and T. Ben-Porat. 1988. Glycoprotein gIII ofpseudorabies virus is multifunctional. J. Virol. 62:2251-2257.

21. vanOirschot, J. T., and A. L. J. Gielkens. 1984. Some

charac-teristics of four attenuatedvaccinevirus strainsandavirulent

strain ofAujeszky'sdisease virus.Vet.Q. 6:225-229. 22. Whealy, M. E., A. K. Robbins, and L. W. Enquist. 1988.

Pseudorabies virus glycoprotein gIII is required for efficient virus growthin tissue culture.J. Virol. 62:2512-2515.