Chimeric cDNA studies of Theiler's murine encephalomyelitis virus neurovirulence.

Copyright X 1993, AmericanSocietyforMicrobiology

Chimeric

cDNA Studies of Theiler's Murine

Encephalomyelitis Virus Neurovirulence

LIANG ZHANG, ANDREASENKOWSKI, BENJAMIN SHIM, ANDRAYMONDP. ROOS*

Departmentof Neurology/MC2030, University of Chicago Medical Center, Chicago, Illinois 60637

Received 26January 1993/Accepted29March1993

Strain GDVII and other members of the GDVIIsubgroupof Theiler's murineencephalomyelitisvirusare

highly

neurovirulent and

rapidly

fatal, while strainDA and other members of the TO subgroup producea

chronic, demyelinating disease. GDVII/DA chimeric cDNA studies suggest that a major neurovirulence determinant is within the GDVII 1Bthrough 1D capsid protein coding region, althoughthe additionalpresence

of upstream GDVII sequences, includingthe 5' untranslatedregion, contributestofullneurovirulence. Our studies indicate that therearelimitations in

precisely

delineatingneurovirulence determinants with chimeric cDNAs between

evolutionarily

divergedviruses, suchas GDVII and DA.

The term Theiler's murine encephalomyelitis virus (TMEV) refers to picornavirus strains most closelyrelated to the cardiovirus genus that can be separated into two subgroupsonthe basis of theirbiologicalactivities (7, 8;for areview, seereference 13).After inoculation intoweanling

mice, strain GDVII and other members of the GDVII subgroupof TMEVareextremelyneurovirulent andrapidly fatal. Incontrast, strain DA and other members of the TO subgroup produce achronic persistent demyelinating infec-tion that resemblesmultiple sclerosis.

One ofourmaingoals istodefine the molecular determi-nants of TMEV-induced disease and the mechanisms in-volved. Despite the markedly different biological activities of strains from the two TMEVsubgroups, different strains haveapproximately 90% identityatthe nucleotide level and 95% identity at the amino acid level. With this in mind, infectious clones of DA (pDAFL3) and GDVII (pGDVII-FL2)have been prepared and chimeric cDNAs and recom-binant viruses have beengenerated (1-3, 9, 12, 14, 16, 18). The present study attempts a finer delineation of TMEV neurovirulence determinants.

Chimeric cDNAclones, constructed by replacingpart of pDAFL3 with segments from pGDVIIFL2, are named by firstlisting5' and 3' junctions of the GDVIIsegment, then addingaslash,andfinally addingDAFL3. Virus derivedby transfection of GDVIIFL2was fully neurovirulent, with a 50% lethal dose

(LD50)

of 0.7 PFU, while transfection-derived DAFL3 virus was not neurovirulent (LD50 > 106 PFU) (Fig. 1). Aspreviously published (2),all the recombi-nant viruses shown inFig. 1 failed to killmice unless they contained the 1B to 2C segment (1B-2C) from GDVII. Although GDVII 1B-2C was essential for neurovirulence, virus that contained only this GDVII segment did not achieve the full neurovirulence of the GDVIIparentalstrain (GD1B-2C/DAFL3 [LD5O = 102.7 PFU] versus GDFL2 [LD5Q = 0.7 PFU]). The addition of GDVIIsequence 5' to

1B-2C

(GD5'-2C/DAFL3

virus) conferred full neuroviru-lence on the virus

(LD50

= 1 PFU), demonstrating that

neurovirulence is multigenic and that certain regions have

more effect than others. Despite the fullneurovirulence of GD5'-2C/DAFL3 virus, the addition of GDVIIsequence 3' to 1B-2C(GDlB-3'/DAFL3 virus)also enhancedthe

neuro-*Correspondingauthor.

virulence of

GDlB-2C/DAFL3

virus, suggesting a

redun-dancy of neurovirulence determinants, i.e., severalregions of the genome canleadto equivalent degreesof neuroviru-lence. Redundancy is not unexpected given the likelihood that GDVII virus, under evolutionary pressure, might de-velop avarietyofmeans to kill a mouse. Thisredundancy

serves to complicate the identification of neurovirulence determinants bymeans ofa chimericGDVII/DA approach. To further delineate key neurovirulence sequences, we

divided GDVII 1B-2C intotwopieces and separately substi-tuted each segment intopDAFL3, constructingpGD1B-2A/ DAFL3 and pGD2A-2C/DAFL3. Surprisingly, neither re-combinant virus was significantly neurovirulent (Fig. 2A). Thesefindingswereobserved withmultiple clones, suggest-ing that the loss of virulence did not result from an unin-tended attenuatingmutation introduced during the genetic construction. As additional confirmation of this point, the DA 2A-2C segment in

GDlB-2A/DAFL3

chimeric cDNA (thathadgeneratedattenuatedGDlB-2A/DAFL3 virus)was substituted with theequivalent GDVII segment; the result-ant GD1B-2C/DAFL3viruswas as neurovirulent as proto-typicGDlB-2C/DAFL3virus.

We considered twopossible explanations for these find-ings. (i)The presenceofchimeric 2Aattenuates

GD1B-2A/

DAFL3 and

GD2A-2C/DAFL3

viruses. (ii) Interactions be-tween GDVII genes or gene products that are critical for neurovirulence and intact in GD1B-2C/DAFL3 virus are

disruptedin the attenuated constructs.To investigatethese issues, we constructed pGDL-2A/DAFL3 with the same AatIIjunction in the 2Acoding region asinthe attenuated

viruses. GDL-2AIDAFL3 virus was as neurovirulent as

GDlB-2C/DAFL3 (Fig. 2B), indicating that chimeric 2A is nottheexplanationfor attenuation of

GD1B-2A/DAFL3

and GD2A-2C/DAFL3viruses. The decreasedgrowthkinetics of GDlB-2A/DAFL3 and GD2A-2C/DAFL3 viruses (see be-low) supported the hypothesis that interactions of GDVII

genes or gene products critical for neurovirulence were

disruptedin theseviruses.

Wepresumedthat neurovirulencedeterminants in GDVII L-2A (and GDVII 1B-2A) were present upstream from P2 and not in 2A since the part of 2Acontributed by GDVII containedonlytwoamino acidsdifferent from DA. Tomore

carefully investigate this,weengineeredasilent XhoI siteat

the 3' end of the 1D coding region of GDVII and DA to

generate chimeras witha"perfect" GDVII/DA 1D/2A junc-4404

on November 9, 2019 by guest

http://jvi.asm.org/

5' UTR IL P1

Plasmid Construct L Al 1B 7 1C I1D

pGDVIIFL2

pDAFL3

NcoI

pGD1 B-3'/DAFL3

2A | P2

I2A 12B 2C

3B

3D I

Neurovirulence L.50(pfu)

++ 0.7

6

_ >10

1.4

++ 1 0

pGD5'-1B/DAFL3

pGD1B-2C/DAFL3

pGD5'-2C/DAFL3

pGD2C-3C/DAFL3

pGD3C-3'/DAFL3

NcoI

NcoI StuI

Sn'!

StuI StuI

II

I 1000 2000I3000II4000 5000 6000 7000 8000

58 3

1000 2000 3000 4000 5000 6000 7000 8000

4.9

_ >10

2.7

+ 10

++ 1.0

5.9 _ >10

4.5

_ >10

NucleotideLength(bp)

FIG. 1. Aseries ofTMEVcDNAs andtheir neurovirulencephenotypes. Thepositionof theTMEVcodingareais shown atthetop, and the nucleotidepositioninthegenome is shown atthe bottom. The GDVIIgenomeandsegmentsfromit areshownassolid bars, while the DAgenome is shown as open bars.Restrictionenzymesites at thejunctions of chimeric constructs arenoted. Each of thechimericcDNAs waslinearizedwithXbaI(whichcuts3' to the genome) and transcribedinvitro with T7 RNApolymerase, and the transcriptswerethen transfectedinto L or BHK-21cellstogenerate virus(18). Thirty microliters of undilutedor10-fold-dilutedtransfection-derivedrecombinant viruses wasinoculatedintracerebrally into five3- to4-week-old SJVJ (Jackson Laboratories)orDBA/2(Charles RiverLaboratories)mice

todeterminetheLD50forweanlingmice(4).Inaddition, neurovirulence,which weoperationally definedasthedeath ofaweanlingmouse within4weeksofinoculation,wasscoredas-(attenuated), +(slightlyneurovirulent),or + + (neurovirulent). The results demonstratethat the presenceofaGDVII1B-2Csequence isessential forthedeath of miceand TMEVneurovirulence.

tion. Thisenabledusto exactly transferasegmentfromthe GDVII5' terminustotheXhoIsitetoproduceGD5'-XhoI/ DAFL3. The latter virus is as neurovirulent as parental GDVII virus (Fig. 3A), indicating that GDVII sequences

upstreamfrom P2aresufficient for full neurovirulence. The GDVII 5' untranslated region (5' UTR) had a clear effecton neurovirulencewhen certain GDVII downstream

sequenceswerealso present, since

GD5'-XhoI/DAFL3

virus

(LD5O =0.7 PFU,likeGDFL2 virus) is 20to100 timesmore neurovirulent than themost neurovirulent recombinant that doesnotcontain the 5'UTR,

GDL-2A/DAFL3

(LD5O

= 101.2

PFU). In addition, even a single nucleotide change in the TMEV5' UTRcandramaticallyattenuatethe virus(11),as is true for other picornaviruses (10). It should be noted, however, that most of the difference in neurovirulence

among GDVII and TO subgroup strains depends on the

presenceofGDVIIsequencedownstream fromthe5' UTR

rather than the presence of the GDVII 5' UTR, i.e., the

differences in the LD50s of neurovirulent recombinant vi-rusesthatcontain theGDVII5' UTR

(LD5O

=0.7PFU) and

lack it

(LD5O

= 101.2PFU) areminimal compared with the

difference between the

LD5Os

ofGDVII (LD5O = 0.7PFU)

andDA(LD5O > 106PFU) strains.

Tomorefinelydelineatesequences downstream fromthe 5' UTR critical forneurovirulence, we prepared additional

constructs with the downstream

engineered

XhoI

P1-P2

junction

site

and

with

decreasing

upstream

margins

of the

GDVII

segments

(Fig. 3B). GDL-XhoI/DAFL3 virus,

which

contains GDVII sequences from Lto the

engineered

XhoI,

hadanLD50

of

103i4PFU.

TheLD50 of

GDIA-XhoI/DAFL3,

which contains a

perfect

junction

from

GDVII

1A(VP4)

through

XhoI

and

contains

only

GDVII

P1 coding

sequence,

was

approximately

102.8

PFU,

suggesting

that the GDVII

capsid

proteins

are not

sufficient

to

produce

full

neuroviru-lence when

substituted

into DA. The viruses

generated

from

GDlB-XhoI/DAFL3

and

GDlD-XhoI/DAFL3

wereaboutas

attenuated

as

GD1B-2A/DAFL3

virus.

Despite

this latter

result, the LD50s of

GDlB-2C/DAFL3, GDL-XhoI/DAFL3,

and

GDlA-XhoI/DAFL3

suggest that

significant

neuroviru-lence

determinants

(presumably

disrupted

in

GDlB-XhoI/

DAFL3 and

GDlD-XhoI/DAFL3 viruses)

are included in GDVII

regions

1B

through

1D.

We suspect that therearevaried

mechanisms by

whicha

disruption

of interactions between GDVII genes or gene

products

thatarecritical for neurovirulence canleadto the

apparentattenuation ofsomerecombinant viruses that may actually contain

neurovirulence

determinants. Attenuation ofsomerecombinant viruses which contain parts of

GDVII

P1

(e.g.,

GDlB-XhoI/DAFL3)

mayresult from interference

with virionassemblyorreceptor

binding

becauseofchimeric

1 77::=

on November 9, 2019 by guest

http://jvi.asm.org/

[image:2.612.85.548.72.373.2]

A.

I

5LUTR

I

-Plasmid Construct

P1 P2

1C 1D 2A12B| 2C

3A P3

3AB

kc

T

3B

I

3D eeI

Neurovirulence

LD-50(PfU)

6

>10

pDAFL3

pGDFL2

pGD1 B-2C/DAFL3

pGD1B-2A/DAFL3

pG D2A-2C/DAFL3

B.

pGDL-2A/DAFL3

++

Ncol Stul

I

Ncol Aatil

I

Aatll Stul I

Aatll AatlI

L

~~~~~~~~~~~~~~~~~~~~~~~~~~~I

+

+

5' I3'

1000 2000 3000 4000 5000 6000 7000 8000

[image:3.612.60.542.92.410.2]

NucleotideLength(bp)

FIG. 2. ChimericcDNAsgeneratedtodelineatesequencesimportantintheneurovirulence of the GDVII 1B-2C segment.(A)TheGDVII

1B-2C segment (in GD 1B-2C/DAFL3 virus) contains a major neurovirulence determinant; however, there is surprisingly little or no

neurovirulence when eitherGDVII1B-2AorGDVII 2A-2Cisseparately substituted intotheDAgenome.(B)GDL-2A/DAFL3virus issimilar

inneurovirulence toGD1B-2C/DAFL3,indicatingthat theattenuationofGD1B-2A/DAFL3virusisnotduetoachimeric2A.

(disrupted) capsid regions.Insomecases,chimeric L-P1-2A andchimeric P2mayleadtoimpaired protein processingand attenuation of other recombinants. Occasional chimeric cDNA constructs, such as

GDlC-XhoI/DAFL3,

were not infectiousatall invitrodespiterepeatedattempts toprepare

them,probablybecausetheseconstructshadamoredrastic disturbance in genome interactions that are critical for

infectivity.

Weinvestigatedthe in vitrogrowthpropertiesofselected recombinant viruses.GDVIIFL2virusgrewtohigher levels thanDAFL3 virusat7 and 9 hpostinfection (Table 1), and,

aspreviously reported (12), GDlB-2C/DAFL3grewtolevels between those seen with GDVIIFL2 and DAFL3 viruses.

GDlB-2A/DAFL3

andGD2A-2C/DAFL3 viruses,however, grew to levels even lower than DAFL3 virus at both time points,perhaps reflectingadisruption of interactions critical forefficient in vitro growth (as well as neurovirulence). In these cases, neurovirulence seemed to correlate with effi-cient in vitro growth. On the other hand, recombinant viruses GDlB-XhoI/DAFL3 and GDlD-XhoI/DAFL3 grew

to reasonably high levels, approaching those seen with GDVIIFL2 virus(at 7 hpostinfection) despite their

attenu-atedphenotype.Theneurovirulencephenotype has similarly been found to correlate imperfectly with tissue culture growthin thecase ofotherpicornaviruses and

nonpicorna-viruses (5, 10); neurovirulence may correlate better with nervoussystemvirus growth,aspreviously suggested (12). Difficulties relatedtochimeric cDNA studies appeartobe

less of a

problem

with respect to

poliovirus intratypic

recombinant viruses than for GDVII and TO, probably because Sabin strains were derived after relatively few passages of the

parental

strain

(with

thesubsequent genera-tion ofalimited number of

point mutations),

while there is

substantial evolutionary

divergence

of the GDVII and TO

TABLE 1. Growthof virusat7 and 9 h after infection of BHK-21cells

Virus titer(%ofDAFL3)" at:

Virus

7hpostinfection 9hpostinfection

DAFL3 108-3(100) 1079 (100)

GDVIIFL2 108.8(276) 108-8 (718)

GDlB-XhoI/DAFL3 108.7(224) 108.1 (167)

GDlD-XhoI/DAFL3 108.6(195) 1080(141)

GD2A-2C/DAFL3 1077 (26) 1075 (44)

GDlB-2A/DAFL3 107-3(9) 107i5(35)

aExpressedasPFU permilliliter and percentage of the titer of DAFL3at

thespecifictimepointstudied.

0.7

1

02.3

>106

6 >10

1.2

1 0

I

on November 9, 2019 by guest

http://jvi.asm.org/

[image:3.612.315.550.615.704.2]

A.

5'UTR LI Pi

ILAl 1B 1C 1D

Plasmid Construct

I

P2

2A I12B 2C

I P3

3B

pDAFL3

Neurovirulence LA50(PfU)

_ _6

>10

++ 0.7

pGDFL2

pGD5'-Xhol/DAFL3

B.

pGDL-Xhol/DAFL3

pGDl A-Xhol/DAFL3

pGD1 B-Xhol/DAFL3

pGD1 D-Xhol/DAFL3

Xhol

Aatl Xhol

Saul Xhol

I

PfIMI Xhol

I

++ 0.7

3.4

+ 10

2.8

+ 10

6.2

- 10

5.0 >10

5,

1000 2000

3,

I8 0

3000 4000 5000 6000 7000 8000

NucleotideLength (bp)

FIG. 3. A series of chimeric cDNAs with a "silent" engineered XhoI at the 3' end of P1. (A) GD5'-XhoI/DAFL3 virus is fully neurovirulent,suggesting thatsequences upstreamfromP2 aresufficientforneurovirulence.(B)GDL-XhoI/DAFL3andGDlA-XhoI/DAFL3 viruses areneurovirulent,butGDlB-XhoI/DAFL3 andGDlD-XhoI/DAFL3viruses are not. Note thatGDlD-XhoI/DAFL3 virus is listedas attenuatedbecausenomice died after inoculation withundiluted virus; however, the LD50 could notbe designated >106because of the relatively low titerof thevirus stock.

subgroups. It may be that other approaches besides GD-VII/DA chimeric cDNA studies will be valuable in the identification of disease determinants, such as the use of intrastrain chimeras between awild-type parentand a

pas-saged strain with an altered disease phenotype. We are

presently studying chimeras between tissue culture-passed strainDA(whichisnormallyattenuated)andsucklingmouse

brain-passed DA(which has heightened neurovirulence [6, 8, 15, 17]).

We thank Lee Baksas for secretarial assistance and Teresa

Bodwelland Alfonso Rivera fortechnicalhelp.

This research was supported by NIH grants P01NS21442 and P01NS24575. Liang Zhang is a recipient ofa National Multiple

SclerosisSocietypostdoctoral fellowship. REFERENCES

1. Calenoff, M. A., K. S. Faaberg, and H. L. Lipton. 1990.

Genomicregionsofneurovirulence andattenuation in Theiler's murine encephalomyelitis virus. Proc. Natl. Acad. Sci. USA

87:987-982.

2. Fu,J.,M.Rodriguez,and R. P. Roos. 1990. Strains from both

Theiler'svirussubgroupsencodeadeterminant for

demyelina-tion. J. Virol.64:6345-6348.

3. Fu, J., S. Stein, L. Rosenstein, T. Bodwell, M. Routbort, B.

Semler,and R. P. Roos. 1990. Neurovirulence determinants of

genetically engineered Theilerviruses. Proc. Natl. Acad. Sci. USA87:4125-4129.

4. Karber,G. 1931. Beitragzurkollektiven Behandlung

pharma-koligischerReihenveruche.Arch.Exp. Pathol. Pharmakol.162: 480-483.

5. Kuhn, R.J.,D. E. Griffin,H.Zhang, H. G. M. Niesters, and J. H. Strauss. 1992.Attenuation ofSindbis virus neurovirulence by using defined mutations in nontranslated regions of the genome RNA. J. Virol.66:7121-7127.

6. Lehrich,J.R., B. G. W. Arnason, and F. H. Hochberg. 1976. Demyelinative myelopathyinmice inducedbythe DA virus. J. Neurol.Sci. 29:149-160.

7. Lipton, H. L. 1980. Persistent Theiler's murine encephalomy-elitis virus infection in mice depends on plaque size. J. Gen. Virol.46:169-177.

8. Lipton, H. L., and M. C. Dal Canto. 1978. The TO strainsof Theiler'svirus cause "slowvirus-like" infectionin mice. Ann. Neurol. 6:25-28.

9. McAllister, A., F. Tangy, C. Aubert, and M. Brahic. 1990. Geneticmapping of the abilityofTheiler's virustopersistand demyelinate.J.Virol.

64:4252-4257.

10. Minor,P.D.,G.Dunn, A.John,A.Phillips,G. D.Westrop,K. Wareham, and J. W. Almond. 1989.Attenuationand reversion of theSabintype 3vaccinestrain,p.307-318. In B. L. Semler and E. Ehrenfeld (ed.), Molecular aspects of picornavirus infection and detection. American Society for Microbiology, Washington, D.C.

11. Pritchard, A. E., M.A. Calenoff, S. Simpson, K.Jensen, and H. L.Lipton. 1992. Asinglebasedeletionin the 5'noncoding region ofTheiler's virus attenuates neurovirulence. J. Virol. 66:1951-1958.

12. Rodriguez, M., andR.P. Roos.1992.Pathogenesisofearlyand late disease in mice infected with Theiler'svirus, using

intra-I

.M

on November 9, 2019 by guest

http://jvi.asm.org/

[image:4.612.84.552.93.387.2]

typic recombinant GDVII/DA viruses. J. Virol. 66:217-225. 13. Roos, R. P., andN.Casteel. 1992.Determinants of neurological

disease induced byTheiler's murineencephalomyelitis virus,p.

283-318. In R. P. Roos (ed.), Molecular neurovirology: patho-genesis of central nervous system virus infections. Humana Press, Totowa, N.J.

14. Roos,R. P.,S. Stein, Y. Ohara,J. Fu, and B. L.Semler. 1989. Infectious cDNA clones of the DA strain ofTheiler's murine encephalomyelitis virus. J. Virol. 63:5492-5496.

15. Roos, R.P., and P. J. Whitelaw. 1984. Biochemicalanalysis of DAstrain of Theiler's murineencephalomyelitis virus obtained directly from acutely infected mouse brain. Infect. Immun.

44:642-649.

16. Stein, S. B., L. Zhang, and R. P. Roos. 1992. Influence of Theiler's murineencephalomyelitis virus 5' untranslated region

ontranslation and neurovirulence. J. Virol.66:4508-4517.

17. Stroop, W. G., M. Brahic, and J.R.Baringer. 1982. Detection of tissueculture-adapted Theiler's virus RNA in spinal cord white

mattercellsthroughout infection. Infect. Immun. 37:763-770.

18. Tangy, F., A. McAllister, C. Aubert, and M. Brahic. 1991. Determinants ofpersistence of demyelination of the DA strain of Theiler's virus are found only in the VP1 gene. J. Virol.

65:1616-1618.

on November 9, 2019 by guest

http://jvi.asm.org/