Genetic and phenotypic studies on Ross River virus variants of enhanced virulence selected during mouse passage

VIROLOGY

172,399-407

(1989)

Genetic and Phenotypic

Studies on Ross River Virus Variants Selected during Mouse Passage

A. D. J. MEEK, S. G. FARAGHER,’ Biochemistry

Department,

Faculty Received

of Science, January

R. C. WEIR, Australian

25, 1989; accepted

AND

of Enhanced

Virulence

L. DALGARNO’

National

University,

May

25, 1989

Canberra,

Australta

We have passaged Ross River virus (RRV) in mice to generate variants with increased mouse virulence and attempted to relate changes in virulence to genome sequence changes. RRV NBO (zero passage in mice) is a plaque-purified clone of the mouse-avirulent strain RRV NB5092, and is of low virulence for day-old mice. During RRV NBO replication in infant mice, its virulence for day-old mice increased markedly with time. By 7 days postinfection the LDSo value of harvested virus (passage level one) was -104-fold less than that of the parental virus. No further decrease in LD,, followed 10 serial passages in infant mice. However, 10th passage level virus showed increased clinical effects in weekold mice by comparison with virus from passage levels one and two. The growth kinetics of RRV variants in mice suggested that the rate and extent of RRV replication in the brain tissue determined the enhanced mouse virulence of serially passaged virus. Seven out of eight independently passaged, 10th passage level variants had changes in the E2 gene leading to one or two amino acid substitutions. The changes were at residues 212, 232, 234, 251, 341, 27 and 172, and 72 and 134 in these variants; all changes except two were nonconservative. Residues 212,234, and 251 form part of a neutralization determinant in RRV. Changes in epitope b2 (which includes amino acids 246, 248, and 251) alter the kinetics of RRV entry into cells (P. Kerr, R. C. Weir, and L. Dalgarno, unpublished data). First and second passage level virus of enhanced virulence was unchanged in E2 or El gene sequences from RRV NBO. However, 1st 2nd, and 10th passage level virus induced higher levels of virus-specific RNA synthesis than did RRV NBO in cultured BHK cells. We propose a model for the mechanism of virulence enhancement on passaging RRV NBO in mice. 8 1989 Academic Press, Inc.

INTRODUCTION

velope glycoproteins and, of these, three were considered to be potential virulence determinants (Faragher et a/., 1988). In studying virulence determinants, the value of sequence comparisons between viral isolates which are not clonally related has obvious limitations. Another approach derives from studies by Taylor and Marshall (1975) who passaged Nelson Bay strains of RRV in mice; this led to enhanced virulence resulting from the overgrowth of variants of increasing virulence during passaging. Sequence comparisons between the genomes of avirulent parental strains and virulent passaged variants could be of value in defining mouse virulence determinants. The RRV genome is a positive-sense RNA molecule of approximately 1 1,674 (for RRV NB5092) or 11,853 (for RRV T48) nucleotides (Faragher et a/., 1988). The genome acts as a messenger RNA for the synthesis of nonstructural proteins (nsPs) required for viral RNA replication (see Strauss and Strauss, 1986). A subgenomic 26 S RNA is the template for synthesis of the viral structural proteins (Rice and Strauss, 1981; Dalgarno et a/., 1983) in the order 5’-C-E3-E2-6K-El-3’. C is the viral capsid protein, and E2 and E 1 are the envelope glycoproteins. which carry neutralization and hemagglutination activities, respectively (Dalrymple et a/.,

Ross River virus (RRV), an alphavirus, is the etiological agent of epidemic polyarthritis in humans (Doherty et a/., 1963; Marshall and Miles, 1983). In mice, the usual experimental host, RRV causes a flaccid paralysis of the hind limbs and in severe cases death (Mims et a/., 1973; Murphy et a/., 1973; Taylor, 1972). RRV exists as a number of antigenic (Gard et al., 1973; Marshall et al., 1980; Woodroofe et a/., 1977) and genetic (Faragher er al., 1985b) types. Differences occur between certain of these genetic types in their virulence for mice. The prototype T48 strain, isolated in northern Queensland (Doherty et a/., 1963), is highly virulent. In contrast, RRV isolates from Nelson Bay in New South Wales induce low mortality in infant mice and cause no symptoms in older mice (Gard et a/., 1973; Taylor and Marshall, 1975). The complete nucleotide sequences and the deduced amino acid sequences of the genomes of RRVT48 and RRV NB5092 (a Nelson Bay isolate) have been established (Faragher et al., 1988). Of the 48 deduced amino acid differences between the encoded proteins of the two isolates, 8 were in the en’ Present address: Plant Molecular Biology Group, School of Biological Sciences, A.N.U., Canberra, Australia. * To whom requests for reprints should be addressed.

Research

399

0042.6822189

$3.00

CopyrIght 0 1989 by Academic Press, Inc All rtghts of reuroduci~on in any form reserved

400

MEEK

1976; Vrati et a/., 1988). E3 and 6K are signal sequences for membrane insertion of E2 and E 1, respectively. The RRV genome also contains untranslated sequences at the S’and 3’ends and in the junction region between the nsP genes and structural protein genes (Faraghereta/., 1988; Ou eta/., 1982a,b; 1983). These regions are thought to contain sequences regulating transcription and translation (Strauss and Strauss, 1986). In this report, we describe changes in the biological properties of Ross River virus variants which are selected during replication in infant mice and relate them to nucleotide sequence changes which accumulate during this procedure.

ET AL.

2/7 is uncloned virus from passage level 2 of passage series 7; NB 2/7/P is an isolate plaque purified from NB 217.

Estimation

of virulence

in mice

RRV Nt35092 (Gard et a/., 1973) was supplied by Dr I. D. Marshall (Microbiology Department, John Curtin School of Medical Research, Australian National University). RRV NBO (zero passage in mice) was derived from RRV NB5092 by two plaque-purification steps on Vero cells. Working stocks were tissue culture supernatants from BHK cells infected at low multiplicity. Virus was assayed by plaque formation on Vero cells.

Viral virulence was assayed by titration in outbred Walter and Eliza Hall Institute (WEHI) mice or outbred Swiss white mice less than 24 hr old as described by Taylor and Marshall (1975). Identical results for virulence parameters of a given virus strain were obtained in the two lines of mice. Each sample (30 ~1) of 1 O-fold serial dilutions of virus stocks in HBSS was inoculated intraperitoneally (i.p.) into 9-l 1 litter mates. The incidence of clinical signs and death was recorded daily for 14 days. Clinical signs included waddling gait (moderate signs) or flaccid paralysis of the hind limbs (severe signs). Asymptomatic mice were challenged after 14 days with 2 x 1 O5 PFU of the virulent RRVT48 strain; failure to develop symptoms was taken as evidence of a prior infection. Fifty percent end points of lethality (LD&, infectivity (ID,,), and clinical signs (CD,,) were estimated by the method of Reed and Muench (1938) and are presented in terms of PFU as determined on Vero cells. To compare virulence estimates, standard errors were determined by the empirically derived method of Pizzi (1950).

Cells

Virus growth in mice

Monolayer cultures of BHK cells were grown in Eagle’s minimal essential medium (EMEM; Glasgow modification) supplemented with 8% heat-inactivated bovine serum and 2% fetal calf serum (FCS). Vero cells were grown in Ml 99-lactalbumin hydrolysate medium supplemented with 10% heat-inactivated bovine serum.

Day-old mice were inoculated S.C. in the lumbar region with 100 Vero PFU of virus. At 24-hr intervals, groups of three mice were anaesthetised with chloroform; blood was collected from the heart using heparinized capillary tubes and diluted 1: 10 in HBSS (pH 7.2). Brain and hind leg muscle tissue were removed and homogenized as a 10% (w/v) suspension in HBSS. The virus titer in each tissue was calculated as the average of titers for individual mice at each time point.

MATERIALS

AND METHODS

Virus

Passage of virus in mice The passaging procedure followed Taylor and Marshall (1975). Passaging was initiated by subcutaneous (s.c.) lumbar inoculation of groups of three to five mice less than 24 hr old; the inoculum was -1 O6 PFU of RRV NBO. For serial passage, eight independent passage series were carried out. When clinical symptoms became apparent at 5-7 days p.i., the lumbar region, pelvis, and hind limbs were pooled for each group of mice and made into a 10% (w/v) suspension in Hank’s balanced salt solution (HBSS; pH 7.2). A 1 :lO dilution of this suspension was used as the inoculum for the next passage and the blind passaging procedure was repeated nine times. lnocula contained 104-1 0” PFU. Working stocks of passaged virus were supernatants from infected BHK cells. Passaged virus is notated according to passage level and passage series. Thus NB

Viral RNA and protein synthesis Rates of viral RNA synthesis in BHK cells were estimated by incubation of infected cell monolayers for 2 hr with [5-3H]uridine (10 #.XmI) in EMEM containing 5 pg/ml of actinomycin D (Newton et al., 1981). Monolayers were dissociated with 1% (w/v) sodium dodecyl sulfate, precipitated with trichloroacetic acid (5%) on glass fiber disks, and counted for radioactivity. Proteins in infected BHK cells were labeled and cell extracts prepared and electrophoresed as described by Dalgarno

et a/. (1984). Extraction

of virion RNA

Virus was purified from clarified BHK cell supernatant fluids by precipitation with 7% (w/v) polyethylene

ROSS

RIVER

glycol and centrifugation through a 15-300/o sucrose gradient containing 0.2 M NaCI, 1 mM EDTA, 0.3% FCS, and 0.05 MTris (pH 7.4). Extraction of virion RNA from pelleted peak fractions was as described by Ou et al. (1981). cDNA

synthesis

and restriction

enzyme

sequencing

401

VARIANTS TABLE VIRULENCE

1

MICE OF VIRUS ISOLATED AFTER PASSAGE RIVER VIRUS NBO IN INFANT MICE

FOR INFANT OF Ross

Experiment

digestion

Double-stranded cDNA was synthesized following Rice et al. (1988). Restriction enzyme digestions were under conditions recommended by the manufacturer (New England Biolabs). Single-stranded cDNA synthesis and digestion with Haelll and Taql were as described by Faragher et al. (1985a). Restriction digests were electrophoresed on 5% polyacrylamide gels (acrylamide:bisacrylamide, 39: 1). Following electrophoresis, gels were transferred to Whatman 3MM paper and exposed at -20” to Fuji RX medical X-ray film. Dideoxy

VIRUS

Harvest (days

A

7 7 7

B

0.5 1.0 2 3 5 7

time p.1.1

LD,, value (Vera PFU)a 12.3 4.5 4.0 -h 2.3 x lo4 635 335 28 0.9

a LDSo value IS for virus obtained from the pooled hind leg muscle tissue of five Infected mice. * lncldence of death was too low for LD,, determlnatlons.

of viral RNA

Virion RNAs were sequenced by the dideoxynucleotide method using as primers synthetic 15-l 7 residue oligodeoxynucleotides complementary to appropriate regions of RRV NB5092 or RRVT48 RNAs (Faragher et

a/., 1988). RESULTS Effect on virulence for mice of a single passage of RRV NBO in mice RRV NBO was a plaque-purified isolate from the mouse-avirulent strain RRV NB5092. In virulence assays in day-old mice, RRV NBO gave an LDsO of 7.9 X 1 O4 PFU, a CD,, of 49 PFU, and an IDS0 of 1.1 PFU. The large difference in LDso and IDS,, values indicated a high subclinical infection rate, i.e., a virus of low virulence for infant mice. RRV NBO was subjected to three independent passages in day-old mice. In each passage, virus harvested from hind leg muscle tissue at 7 days p.i. showed a marked increase in virulence over the parental virus with LDsO values falling from 7.9 X 1O4 to 612.3 PFU (Table 1, Experiment A). For the one isolate examined, the IDS0 was equal to the LD,, (4.0 PFU), indicating that all infected mice died (100% mortality) even at the lowest levels of inoculum. To investigate the timecourse of virulence change during the -/-day replication period, day-old mice were infected with RRV NBO, and hind leg muscle tissue was harvested for virulence assay at 0.5, 1, 2, 3, 5, and 7 days p.i. Virus harvested at 0.5 and 1 day p.i. was of similar virulence to RRV NBO. Virus harvested at 2, 3, 5, and 7 days showed a progressive decrease in LDsO, indicating that selection and amplification of virulent mutants occurred from 2 days p.i. (Table 1; Experiment B).

Virulence passages

changes following 10 serial of RRV NBO in mice

Previous studies in which mouse-avirulent field strains of RRV have been serially passaged in infant mice have shown that virulence enhancement, assessed in terms of mortality and morbidity rates in mice of various ages, is close to maximum after 1O-l 2 passages (Gard et a/., 1973; Taylor and Marshall, 1975). We have examined virulence enhancement of RRV NBO in eight independent mouse passage series. After 10 passages, virulence assays were performed on each of the eight uncloned virus populations (harvested at 5-7 days p.i., depending on the severity of clinical signs), and on a single plaque-purified isolate derived from each lOth-passage stock (Table 2). For passage series 2, 3, and 7, virulence assays were also carried out at intermediate passage levels. In day-old mice, uncloned 10th passage virus did not differ significantly in virulence between the eight passage series: LDsO values ranged from 0.7 to 6.4 PFU. The LDsO of virus from series 2, 3, and 7 had not decreased significantly from the value seen after a single mouse passage (Table 2). In all but one passage series, plaque-purified isolates from the 10th passage gave LDsO values in day-old mice which were not significantly different from those of the 10th passage stocks from which they derived (Table 2). For passage series 2, the plaque-purified variant NB 10/2/P had an LDsO of 870 PFU compared with 2.4 PFU for NB 10/2. This indicated that passaged virus can contain subpopulations of differing virulence. For all 10th passage virus groups, the ratio LDsO: CD,,:ID,,, assayed in infant mice, was approximately 1: 1: 1 (data not shown). Although RRV NBO could be

402

MEEK

ET AL.

TABLE VIRULENCE FOR INFANT MICE OF VIRUS ISOLATEDAFTER

2

SERIAL PASSAGE OF Ross RIVER VIRUS NBO IN INFANT MICE Passage

Passage level

2

3

4

5

6

LDSO values 0 1 2 3 4 5 6 7 8 9

8X104 4.5 1.1 0.4

10

2.4

8X104 4.0 1.5 6.1 2.6 1.8 3.3 3.8 3.1 2.8

1.1 0.3

10 (plaque purified)

870

8X

10“

-

1.0

-

-

0.2 -

0.7

3.5

4.0

5.9

3.7

3.3

6.4

0.3

1.1

0.7

0.6

0.9

0.8

1.0

VIRULENCE FOR WEEK-OLD MICE OF VIRUS ISOLATEDAFTER SERIAL PASSAGE OF Ross RIVER VIRUS NBO IN INFANT MICE CDS, in week-old mice (Vero PFU) >7.2 X 105” >4.1 x loba 3.2 X lo4 >8.2 X 1 OS’ 3x104 1.2 x 1 o5 5.9 x 102

NBO NB 212 NB 1 O/2 NB 213 NB 10/3 NB l/7 NB 1 O/7 a Clinical signs were absent could not be determined.

or barely

PFU) 8X104 -

3

Virus designation

10

8X104 12.3 3.3

-

discernible

and CD,,, values

10“

8

-

-

8X

(Vero

7

-

-

distinguished from all 10th passage groups by LDSO and CDS,, , there was no significant difference between NBO and 10th passage virus in IDS0 values. Thus when expressed in terms of Vero PFU, virulent and avirulent viruses were equally capable of establishing an infection in day-old mice although the outcome of infection was different. In week-old mice the mortality rates of 10th passage RRV were too low to permit calculation of LDbO values (data not shown). However, CDS0 values for 10th passage virus from groups 2, 3, and 7 in week-old mice (5.9 X 102-3.2 X lo4 Vero PFU) were markedly less than the corresponding values for the 1st and 2nd passage level virus where incidence of clinical signs was generally too low for CDS0values to be determined (Table 3).

TABLE

series

8x

10“

-

0.1

8x

lo4

-

Growth of RRV NBO and 10th passage level virus in tissues of day-old mice Day-old mice were inoculated S.C. with 100 PFU of RRV NBO or with 100 PFU of the plaque-purified 10th passage isolates NB10/2/P (LD,, N 870 PFU), 10/3/P (LD5,, N 0.3 PFU), and 10/7/P (LDSOI: 0.9 PFU). Virus titers in the blood, hind leg muscle and brain were assayed at daily intervals (Fig. 1). Maximum titers were highest in the hind leg muscle for each virus strain, consistent with previous data on the tissue distribution of RRV (Murphy et a/., 1973; Mims et a/., 1973). Both NB 10/3/P and NB 10/7/P reached higher maximum titers than RRV NBO in blood, brain, and muscle. NB 10/3/P and 10/7/P could be detected in the brain at 1 day p.i. whereas no virus was detected at this time for NBO and NB 10/2/P. Overall the data suggest that the extent of RRV replication is an important determinant of mouse virulence, and that the rate and time course of replication in brain tissue may be of crucial significance.

Virus growth and viral macromolecule BHK cells infected with RRV NBO and passsaged variants

synthesis

in

During growth in cultured BHK cells, NBO, 10/3/P and 10/7/P at m.o.i.s of ~0.001 and -10 showed no significant difference in latent period, in peak titers, or in the ratio of extracellular virus to cell-associated virus (data not shown). NB 10/2/P grew to lower titers than NBO at both m.o.i.s (data not shown). Rates of actinomycin-resistant RNA synthesis were greater in BHK cells infected (m.o.i. N 10) with NB 10/2/P, 10/3/P, and

ROSS

RIVER

VIRUS

403

VARIANTS

in the kinetics of either viral protein synthesis or shutdown of host protein synthesis between the two viruses (data not shown). Sequence of the structural protein genes of 10th passage variants

0

I2

Time

3

(days

4

.i

6

7

8

9

post-infection)

FIG. 1. Titers of RRV NBO and RRV passaged variants in the muscle, brain, and blood during growth in infant mice. Litters of day-old mice were inoculated (s.c.) with 100 PFU of RRV NBO, NB 10/2/P, NM 10/3/P, or NB 10/7/P. At intervals, three mice were killed, and muscle, brain, and blood harvested for virus assay. The figure represents the average of two experiments. The vertical line at three days p.1. represents the standard error (determined using ANOVA analysis on a GENSTAT computer program) which incorporates data from all time points for all viruses. This program could not be used after 5 days p.i. because of the death of mice infected with NB 10/3/P and NB 10/7/P; from 6 days, standard errors were calculated separately for each virus. Asterisks indicate time of death of mice infected with NB 10/3/P and NB 10/7/P. Cl, RRV NBO; n , NB 10/2/P; 0, NB 10/3/ P; l , NB 10/7/P. a, blood; b, brarn; c, muscle.

10/7/P than in NBO-infected cells (Fig. 2). RRV NB 1O/ 2/P induced levels of viral RNA synthesis which were intermediate between those of NBO and the other 10th passage variants. Viral RNA synthesis in BHK cells infected with first passage virus (NB l/3/9, LDso N 0.4 PFU) and second passage virus (NB 2/7/2, LDso N 1.5 PFU) also showed an increase in viral RNA synthesis (approximately 309/oover the period 9-l 3 hr p.i.) compared with that for RRV NBO (data not shown). BHK cells infected with RRV NBO or NB 10/3/P (m.o.i. N 10) were labeled with [3H]amino acids at various times after infection and cell extracts electrophoresed on polyacrylamide gels. No differences were detected

Since virion structural proteins are potential determinants of alphavirus virulence, the complete 26 S region of the virion RNA was sequenced for RRV NBO and for three representative 10th passage variants, NB 1O/2/ P, NB 10/7/P, and NB 10/3. Sequencing was by the dideoxy method using viral RNA as template. The data obtained for these four strains covered the entire 26 S region with the exception of 18 nucleotides adjacent to the poly(A) tail. Nucleotides at 44 positions could not be determined due to cross-banding in the sequence ladders. RRV NB 10/2/P and NB 10/7/P 26 S RNAs each differed from RRV NBO at a single nucleotide in the 26 S region; NBl O/3 differed from RRV NBO at two nucleotides (Table 4). All four of these substitutions were in the E2 gene and each led to a predicted amino acid change. Only the E2 gene of the remaining 10th passage variants was sequenced (Table 4). NB lo/51 P, 10/8/P, and 1O/l O/P differed from NBO at a single amino acid in E2; NB 10/6/P differed from NBO at two amino acids. No change was detected in the El gene of any of the three variants sequenced in El, nor in the E2 gene of NB 10/4/P. Only one silent nucleotide substitution (in the E2 gene of NB 10/3/P) was detected in the nine variants sequenced. All but two of the amino acid changes in E2 (Phe27 + Tyr in NB 10/3/P; Leu341 + Met in NB 10/8/P) were nonconservative (Table 4). Five of the variants had predicted changes

1 4

6

Time

8

(hours

10

12

14

If

18

post-infection)

FIG. 2. Virus-specrfic RNA syntheses In BHK cells infected with RRV NBO or with passaged variants. BHK cell monolayers were mockinfected or infected (m.o.i. N 10) with RRV NBO, NB 10/2/P. NB 1 O/ 3/P, or NB 10/7/P. Cells were labeled at intervals with or without actinomycin D (Materials and Methods). Two infected cultures were labeled at each time and averages are plotted. Time points represent the midpoint of each labeling period. Cl, RRV NBO; l , NB 10/2/P; 0, NB 10/3/P; 0, NB 10/7/P; c), mock-Infected, no AMD; +, mockinfected, with AMD.

404

MEEK TABLE

4

SEQUENCE CHANGES INTHE E2 GENE OF Ross RIVER VIWJS TENTH PASSAGE VARIANTS

Virus

Nucleotide

changea

Deduced amino acid change*

NB 10/2/P

A9262

+ G

E2 His 232 + Arg

NB 10/3

U8377 C9801

+ A + U

E2 Phe 27 + Tyr E2 Pro 172 + Ser

NB 10/3/P

U8377 C9081 c9200

+ A + U + u

E2 Phe 27 + Tyr E2 Pro 172 + Ser -

-

NB 10/4/P

-

NB 10/5/P

C9261

+ U

E2 His 232 + Tyr

NB 10/6/P

G8781 A9268

+ A + U

E2 Gly 72 + Ser E2 Lys 234 + lie

NB 10/7/P

A9320

+ U

E2 Lys 251 + Asn

NB 10/8/P

C9588

+ A

E2 Leu 341 + Met

NB 1 O/l O/P

G9202

+ U

E2Ser212+Ile

a Numbering is from the 5’ end of the RRV NB5092 genome RNA sequence (Faragher et al., 1988). For RRV NB 10/2/P, NB 1 O/3, and NB 10/7/P, the sequence of the entire structural protein region of the viral genome (including 5’ and 3’ untranslated sequences) was determined; for the other variants listed, the E2 gene was sequenced. * Numbering of amino acid residues is from the N-terminus of RRV NB5092 E2 (Faragher et al., 1988).

(at residues 212, 232, 234, and 251) which were in a region of E2 which constitutes a major neutralization determinant (Vrati eta/., 1988). None of the changes in E2 generated or removed glycosylation sites.

Time course of sequence during passaging

changes in E2

The passage level at which sequence changes in E2 could first be detected was determined for passage series 2,3, and 7. For series 2 and 7, sequence heterogeneity at nucleotides 9262 and 9320 (Table 4) was first seen at passage levels IO and 9 respectively (data not shown). For both passage series, the variant genotype constituted about half the viral population (as judged from sequence ladders) at passage level 10. For series 3 (Table 4), the nucleotide substitution which changed E2 residue 27 was first seen at level 3 but had not displaced the parental nucleotide until passage level 9 (data not shown). The change for residue 172 was first detected at level 8 and had displaced the parental nucleotide by passage level 10. Thus for passage series 2, 3, and 7, sequence changes in the E2 gene were detected at later stages in passaging than the level at which virulence enhancement was first observed (level 1).

ET AL.

Fourteen plaque-purified isolates from passage levels 1 and 2 of series 3 and 7 were sequenced in the region of previously demonstrated sequence change in the E2 gene (Table 4) and assayed for virulence in infant mice. All isolates examined were of the parental

genotype

at the relevant positions.

In preliminary

viru-

lence titrations, all were of greater virulence than RRV NBO (data not shown). Two plaque-purified isolates, NB l/3/9 and NB 2/7/2, obtained, respectively, from level 1 of series 3 and from level 2 of series 7, were selected for more detailed study. The El and E2 genes of these isolates were identical in sequence to those of NBO (data not shown). The infant mouse LDSOvalues for NB l/3/9 and 2/7/2 (0.44 and 1.5 PFU, respectively) were not significantly different from those of 10th passage variants with sequence changes in the E2 genes (Tables 1, 2).

Virulence of a monoclonal-antibody-selected variant of RRV NBO

E2

An mAb-selected variant of RRV NBO (NB ~103) with a His232 + Arg change in E2 has been isolated (Vrati eta/., 1988). This is the same change as that in NB 101 2/P(LDS0 N 870 PFU) which is of enhanced virulence compared with NBO. The virulence of NB ~103 was as-

sayed in day-old mice. No mice were killed at the highest dose used (1.6 X 1 O4 PFU). Thus the His232 + Arg change alone is not able to significantly enhance NBO virulence as determined by LD,, values in infant mice.

Restriction digest analysis of single-stranded and double-stranded cDNA from RRV NBO and passaged variants To estimate the extent of change in the NBO nsP genes after 10 mouse passages, single-stranded cDNA to virion RNA from NBO and to RNA from each of the eight plaque-purified 10th passage variants was digested with Haelll or Taql and electrophoresed on polyacrylamide gels (Faragher et al., 1985a). The restriction digest profiles for the passaged variants were identical to those of RRV NBO (data not shown), indicating that for each passage series very limited genetic change (estimated as <0.2Ob) occurred over the genome during 10 passages. The above method scans approximately 4% of the genome (Faragher et a/., 1985a). In order to increase the proportion of the genome scanned, NBO and NBl O/ 3 double-stranded cDNAs, labeled with [a-32P]dCTP, were digested with 41 restriction enzymes. From the NBO genome sequence (Faragher et al., 1988), these 41 enzymes should detect a sequence divergence among the genomes of approximately 0.049/o or greater. Restriction digest profiles for RRV NBO and NB

ROSS

RIVER

I O/3 were indistinguishable with all 41 enzymes except /-/phi (data not shown). With /-/@?I, a fragment of approximately 450 bp in the NBO profile was replaced by two fragments of approximately 360 and 90 bp in the NB 10/3 profile. Thus a mutation in the NB 1013 genome generated a new Hphl site. Since neither of the two nucleotide differences identified between NBO and NB 10/3 in the 26 S region of the genome (Table 4) were in Hphl sites, and since the untranslated sequences of the NBO and NB 1 O/3 genomes are identical (see below), we conclude that the mutation is located in the nsP gene region. The Hphl restriction profiles for both NB l/3/9 and NB 2/7/2 were identical to the NBO profile (data not shown), indicating that initial virulence enhancement was due to sequence change elsewhere in the genome. DISCUSSION Passage of plaque-purified clones of a mouse-attenuated field strain of RRV in infant mice led to a dramatic increase in virulence for infant mice. LDsO values, measured in day-old mice, declined 1 03-1 05-fold during a 7-day replication period in infant mice. No significant change in LDbO took place during a further nine serial passages in mice although an increase in virulence on passaging was indicated by a fall in CD50 for week-old mice. The methods used for passaging were similar to the procedures of Taylor and Marshall (1975), who reported that the virulence of the unadapted RRV NB6024 strain increased progressively through 10 passages in infant mice. Although LDso values of early passage virus were not determined by Taylor and Marshall (1975), a rapid increase in mortality rates for dayold mice occurred between passage level 1 (39% mortality) and passage level 2 (95-100% mortality). Later passage levels showed a progressive increase in mortality and morbidity rates in older mice (Taylor and Marshall, 1975). A number of studies have reported the rapid amplification of virulent subpopulations during virus replication in laboratory animals, Taylor and Marshall (1975) estimated that an enrichment of at least 105-fold occurred in the proportion of a virulent, small plaque variant of RRV NB5092, relative to the avirulent parent virus, after a single mouse passage of a 10: 1 06.6 dose ratio (in day old mouse ID5J of virulent:avirulent virus. A large decrease in LD5,, for adult mice of an avirulent strain of Semliki Forest virus (SFV) was observed after passaging in hamsters (Bradish et a/., 1972). The original field isolate of SFV was of low virulence for adult mice and rapidly increased in virulence after a single passage in mice (Smithburn and Haddow, 1944). The pathogenesis of RRV in mice is not identical to that of SFV and Sindbis (SIN) virus, both of which are

VIRUS

VARIANTS

405

neurotropic (see Atkins et al., 1985; Griffin, 1986). In day-old mice infected with RRV NB5092 and RRV T48, maximum titers reached in muscle were l-l .5 log units higher than in brain (Taylor, 1972). RRV NB5092 infection in the central nervous system was primarily in non-neuronal cells, leading to the conclusion that paralysis of RRV infection was due to viral damage to muscle myofibrils (Mims et al., 1973; Murphy et al., 1973). However, Kingston (1983) concluded that hind leg paralysis was due to necrosis of motor neurons in the anterior horn of the spinal cord. Thus in certain situations RRV gives rise to a neurotropic infection similar to that of other alphaviruses. It was therefore of interest to establish whether the increase in virulence of passaged RRV could be related to virus growth in the brain. Our data (Fig. 1) suggest a relationship between the rapidity with which replication is established in the brain, the titers reached at 2-4 days p.i., and the outcome of infection in day-old mice. The reduced virulence of an E2 deletion mutant of RRV T48 for 7 day old mice has also been related to decreased replication in brain tissues (Vrati et al., 1986). Analysis of restriction digest profiles of singlestranded and double-stranded cDNAs generated from passaged variants indicated that only limited genetic change took place during passaging. Two sets of data indicated that the profound increase in virulence which took place during a single -/-day replication period in infant mice resulted from changes in the nsP genes. First, plaque-purified virulent variants from passage levels one and two had no changes in the El or E2 genes. Second, there were no differences in the 5’ or 3’ untranslated regions or in the junction region of 10th passage variants NB 10/2/P, 10/7/P, or 1 O/3 as compared with parental virus (data not shown). This, together with data on viral RNA synthesis, indicated that changes had taken place in the nsP genes: rates of viral RNA synthesis were higher in BHK cells infected with NB l/3/9 and 2/7/2 than in NBO-infected cells and similar increases were observed with 10th passage variants Studies with other alphaviruses indicate that nsPs can be involved in determining levels of virulence (see Atkins et al., 1985; Atkins and Sheahan, 1982; Barrett and Atkins, 1979; Mecham and Trent, 1983). During serial passage in infant mice, RRV virulence continued to increase as judged by CD50 estimates in 7 day old mice. There was no further decrease in LD5, values (in day-old mice) from that seen for first passage virus. Passaging was accompanied by the selection of strains with a small number of changes in the E2 gene, most of which resulted in nonconseNative amino acid substitutions. Residues 212, 232, 234, and 251, at which such changes took place, correspond to epitopes in a neutralization domain of RRV E2 (Vrati el a/.,

406

MEEK ET AL.

1988). Changes were also found in E2 at residues 27, 72, 172, and 341. What is the selection pressure responsible for these changes? It seems unlikely that they result from immunological selection for two reasons. First, mice acquire the capacity to produce a mature antibody response only after -7 days (Murgita and Wigzell, 1981); in the procedure used, virus was harvested when the mice were 5-7 days of age. Second, if a monoclonal or oligoclonal response led to variant selection, variants selected in one passage could well be selected against in later passages. The changes in E2 may be selected because they alter the conformation of E2 and change cell tropisms. In support of this, changes in amino acids 248-251, which are part of a neutralization domain of RRV E2, alter RRV penetration kinetics: further, variants of RRV NB5092, selected by serial passage in cells in which unadapted virus replicates poorly, are altered in a second epitope within the same region (P. Kerr, R. C. Weir, L. Dalgarno, unpublished data). Studies with SIN indicate that single amino acid changes at a number of positions in E2 and El can induce changes in neurovirulence for mice (Lustig eT a/., 1988; Polo eta/., 1988). The effects of each change can be different; further, genetic background and strain of mice are important in determining the phenotypic change. The best documented of the changes, Arg 114 + Ser in E2, leads to three phenotypes which cosegregate: attenuation in neonatal mice, sensitivity to anti-E2 monoclonal antibodies, and accelerated penetration in BHK cells (Davis et al., 1986). We propose the following model as the mechanism for virulence enhancement on passaging RRV NBO in mice. Early in passaging, spontaneous mutants with alterations in the nsP genes leading to increased rates of viral RNA replication are selected. Increased mouse virulence results from the enhanced capacity of these variants to spread to and replicate in target tissues before intervention of host defense mechanisms. Mutations in the E2 gene which are selected later in passaging lead to a more rapid penetration of virus into target cells or a change in cell tropism, providing an advantage over variants which have changes only in the nsP genes. ACKNOWLEDGMENTS We thank Dr. 1. D. Marshall for valuable discussions. This work was supported by a grant from the National Health and Medical Research Council of Australia.

REFERENCES ATKINS, G. J., and SHEAHAN, B. J. (1982). Semliki ulence mutants have altered cytopathogenicityfor system cells. Infect. lmmun. 36, 333-341.

Forest

virus neurovircentral nervous

ATKINS, G. J., SHEAHAN. B. J., and DIMMOCK, N. J. (1985). Semliki Forest virus infection of mice: A model for genetic and molecular analysis of viral pathogenicity. J. Gen. Viral. 66, 395-408. BARRETT, P. N., and ATKINS, G. J. (1979). Virulence of temperaturesensitive mutants of Sindbis virus in neonatal mice. Infect. Immun. 26,848-852. BRADISH, C. J., ALLNER, K., and MABER, H. B. (I 972). Infection, interaction and the expression of virulence by defined strains of Semliki Forest virus. J. Gen. Iho/. 16, 359-372. DALGARNO, L., RICE, C. M., and STRAUSS, J. H. (1983). Ross RiverVirus 26s RNA: Complete nucleotide sequence and deduced sequence of the encoded structural proteins. Virology 129, 170-l 87. DALGARNO, L., SHORT, N. J., HARDY, C. M., BELL, J. R., STRAUSS, J. H., and MARSHALL, I. D. (1984). Characterization of Barmah Forest virus: An alphavirus with some unusual properties. Virology 133, 416-426. DALRYMPLE, J. M., SCHLESINGER, S.. and RUSSELL, P. K. (1976). Antigenie characterisation of two Sindbis envelope glycoproteins separated by isoelectric focusing. Virology69, 93-l 03. DAVIS, N. L., FULLER, F. J., DOUGHERTY, W. G., OLMSTED, R. A., and JOHNSTON, R. E. (1986). A single nucleotide change in the E2 glycoprotein gene of Sindbis virus affects penetration rate in cell culture and virulence in neonatal mice. froc. Nat/. Acad. Sci. USA 83, 6771-6775. DOHERTY, R. L., WHITEHEAD, R. H., GORMAN, B. M., and O’GOWER, A. R. (1963). The isolation of a third group A arbovirus in Australia, with preliminary observations on its relationship to epidemic polyarthritis. Ausf. 1. Sci. 26, 183-l 84. FARAGHER, S. G., HUTCHISON. C., and DALGARNO, L. (1985a). Analysis of Ross River virus genomic RNA using Haelll digests of singlestranded cDNA to infected-cell RNA and virion RNA. Virology 141, 248-256. FARAGHER, S. G., MARSHALL, I, D., and DALGARNO, L. (1985b). Ross River virus genetic variants in Australia and the Pacific Islands. Aust. J. Exp. Biol. Med. Sci. 63, 473-488. FARAGHER, S. G., MEEK, A. D. J., RICE, C. M., and DALGARNO, L. (1988). Genome sequences of a mouse-avirulent and a mouse-virulent strain of Ross River virus. Virology 163, 509-526. GARD, G., MARSHALL, I. D., and WOODROOFE, G. M. (1973). Annually recurrent Epidemic Polyarthritis and Ross River Virus activity in a coastal area of New South Wales. II. Mosquitoes, viruses, and wildlife. Amer. J. Trop. Med. Hyg. 22, 551-560. GRIFFIN, D. E. (1986). Alphavirus pathogenesis and immunity. In “The Togaviridae and Flaviviridae” (S. Schlesinger and M. J. Schlesinger, Eds.), Chap. 8, pp. 209-250. Plenum, New York. KINGSTON, R. S. (1983). “Pathogenesis of Experimental ExoticAlphaviruses. Infections in Laboratory Rodents, Horses and Sheep,” Ph.D. thesis. Colorado State University, Fort Collins, CO. LUSTIG, S., JACKSON, A. C., HAHN, C. S., GRIFFIN, D. E., STRAUSS, E. G., and STRAUSS, J. H. (1988). Molecular basis of Sindbis virus neurovirulence in mice. /. Viral. 62, 2329-2336. MARSHALL, I. D., and MILES, J. A. R. (1983). Ross River virus and epidemic polyarthritis. Curr. Top. Vector Res. 2,31-56. MARSHALL, I. D., WOODROOFE, G. M.. and GARD, G. P. (1980). Arboviruses of coastal south-eastern Australia. Aust. /. Exp. Biol. Med. Sci. 58, 91-102. MECHAM, J. O., and TRENT, D. W. (1983). A biochemical comparison of the in vitro replication of a virulent and an avirulent strain of Venezuelan Encephalitis Virus. /. Gen. Virol. 64, 1 1 1 l-l 1 19. MIMS, C. A., MURPHY, F. A., TAYLOR, W. P.. and MARSHALL, I. D. (1973). Pathogenesis of Ross River Virus infection in mice. I. Ependymal infection, cortical thinning, and hydrocephalus./. infect. Dis. 127,121-128.

ROSS

RIVER

MURGITA, R. A., and WIGZELL, H. (1981). Regulation of immune functions in the fetus and newborn. Prog. Allergy 29, 54-l 33. MURPHY, F. A., TAYLOR, W. P., MIMS, C. A., and MARSHALL, I. D. (1973). Pathogenesis of Ross RiverVirus infection in mice. Il. Muscle, heart, and brown fat lesions. J. Infect. Dis. 127, 129-l 38. NEWTON, S. E., SHORT, N. J., and DALGARNO, L. (1981). Bunyamwera virus replication in cultured Aedes albopictus (mosquito) cells: Establishment of a persistent viral infection. J. Viral. 38, 1015-l 024. Ou, J.-H., RICE, C. M., DALGARNO, L.. STRAUSS, E. G., and STRAUSS, J. H. (1982a). Sequence studies of several alphavirus genomic RNAs in the region containing the start of the subgenomic RNA. Proc. Nat/. Acad. Sci. USA 79, 5235-5239. Ou, J.-H., STRAUSS, E. G., and STRAUSS, J. H. (1981). Comparative studies of the 3’.terminal sequences of several alphavirus RNAs. Virology 109, 28 l-289. Ou. J.-H., STRAUSS, E. G., and STRAUSS, J. H. (1983). The 5’.terminal sequences of the genomic RNAs of several alphaviruses. J. Mol. Biol. 168,1-15. Ou, J.-H., TRENT, D. W., and STRAUSS, J. H. (1982b). The 3’-non-coding regions of alphavirus RNA contain repeating sequences. J. Mol. Biol. 156,719-730. PIZZI, M. (1950). Sampling variation of the 50% end-point, determined by the Reed-Muench (Behrens) method. Hum. Biol. 22, 151-l 90. POLO, J. M.. DAVIS, N. L.. RICE, C. M., HUANG, H. V., and JOHNSTON, R. E. (1988). Molecular analysis of Sindbis virus pathogenesis in neonatal mice by using recomblnants constructed in vitro. J. Viral. 62,2124-2133. REED, L. J., and MUENCH. H. (1938). A simple method of estimating 50% endpoints. Amer. J. Hyg. 27, 493-497.

VIRUS

VARIANTS

407

RICE, C. M., DALGARNO, L.. GALLER, R., HAHN, Y. S., STRAUSS, E. G., and STRAUSS, J. H. (1988). Molecular cloning of flavivirus genomes for comparative analysis and expression. In “Modern Trends in Virology” (H. Bauer, H. D. Klenk, and C. Scholtissek, Eds.), pp. 8397. Springer-Verlag, Berlin/New York. RICE, C. M., and STRAUSS, J. H. (1981). Nucleotide sequence of the 26s mRNA of Sindbis virus and deduced sequence of the encoded viral structural proteins. Proc. Nat/. Acad. Sci. USA 78, 20622066. SMITHBURN, K. C., and HADDOW, A. J., (1944). Semliki Forest virus I. Isolation and pathogenic properties. J. Immunol. 49, 141-l 57. STRAUSS, E. G., and STRAUSS, J. H. (1986). Structure and replication of the alphavirus genome. ln “The Togaviridae and Flaviviridae” (S. Schlesinger and M. J. Schlesinger, Eds.), Chap. 3, pp. 35-90. Plenum, New York. TAYLOR, W. P. (1972). “Studies of Ross RiverVirus during Laboratory Adaptation,” Ph.D. thesis. Australian National University, Canberra, Australia. TAYLOR, W. P., and MARSHALL, I. D. (1975). Adaptation studies with Ross River Virus: Laboratory mice and cell cultures. J. Gen. Viral. 28,59-72. VRATI, S.. FARAGHER, S. G., WEIR, R. C., and DALGARNO, L. (1986). Ross River virus mutant with a deletion in the E2 gene: Properties of the virion, virus-specific macromolecule synthesis, and attenuation of virulence for mice. Virology 151, 222-232. VRATI, S., FERNON, C. A., DALGARNO, L., and WEIR, R. C. (1988). Location of a major antigenic site involved in Ross River virus neutralization. Virology 162, 346-353. WOODROOFE, G., MARSHALL, I. D., and TAYLOR, W. P. (1977) Antigenitally distinct strains of Ross Rivervirus from north Queensland and coastal New South Wales. Aust. J. Exp. Biol. Med Sci. 55. 79-87.