Pathogenicity of Hantaan Virus in Newborn Mice: Genetic Reassortant Study Demonstrating that a Single Amino Acid Change in Glycoprotein G1 Is Related to Virulence

Abstract

Two Hantaan virus strains, clone 1 (cl-1), which is virulent in newborn mice, and its attenuated mutant (mu11E10), were used to examine the pathogenesis of Hantaan virus infection in a mouse model and identify virus factors relating to virulence. After subcutaneous inoculation of newborn BALB/c mice, cl-1 caused fatal disease with high viral multiplication in peripheral organs, but mu11E10 produced nonfatal infection with a low level of virus multiplication. Intracerebral inoculation of either strain caused fatal disease. Histopathological changes in the dead animals were prominent in the brain, indicating that the brain is the target organ and produces the fatal outcome. These results indicate that mu11E10 has a generally less virulent phenotype, and because of decreased multiplication in peripheral tissues, neuroinvasiveness is also decreased. An experiment with genetic reassortant viruses showed that in newborn mice the M segment is the most related to virulence and the L segment is partly related. Sequence comparison detected a single deduced amino acid change (cl-1 Ile to mu11E10 Thr) at amino acid number 515 in glycoprotein G1. One nucleotide change, but no amino acid substitution, was observed in the noncoding region of the L segment. In mouse brain microvascular endothelial cells in vitro, viruses possessing a cl-1-derived M segment grew more rapidly than viruses containing a mu11E10-derived M segment. These results suggest that the single amino acid change in the glycoprotein alters peripheral growth, which affects invasion of the central nervous system in mice.

Hantaan virus is the prototype of the genus Hantavirus, which belongs to the family Bunyaviridae (44). Hantavirus includes the etiologic agents of two distinct disease syndromes in humans, hemorrhagic fever with renal syndrome (HFRS) and hantavirus pulmonary syndrome (HPS), which are transmitted from rodents to humans (40). Old World hantaviruses, such as the Hantaan (HTN), Seoul, Dobrava, Puumala, and related viruses cause HFRS, which is characterized by fever, renal failure, and, in severe cases, hemorrhagic manifestations. The severity of HFRS is dependent on the serotype of the causative virus. New World hantaviruses, Sin Nombre and related viruses, cause HPS, which is characterized by severe acute respiratory dysfunction and a mortality rate of 50% (10, 34, 62). Each hantavirus is primarily associated with a specific rodent species. Many field observations and epizootiological studies have shown that hantaviruses cause chronic infections without signs of disease in their reservoirs (40).

HTN virus causes a severe form of HFRS in humans. It is an acute prostrating febrile illness with renal failure, in which one-third of patients develop hemorrhagic manifestations and 10 to 15% develop shock. In addition, the mortality rate of severe HFRS is 5 to 10% (37, 49). The pathology of HFRS is multisystemic. The spread and tropism of the virus in a patient's body, and immune-related pathogenesis and responses, are thought to be important factors in an HFRS crisis (7, 18, 24, 54, 55). Despite many investigations, little is known about the pathogenesis of HTN virus infection in humans (HFRS), since a suitable animal model for the disease in humans has not been reported. In animal experiments, only neonatal animals and immunologically deficient animals, such as nude mice and severe combined immune deficient (SCID) mice, die after inoculation with HTN virus (32, 59). Moreover, in contrast to Puumala virus and Prospect Hill virus (15, 56), attempts to induce HFRS manifestations in monkeys using HTN virus have been unsuccessful. Previously, several groups studied the pathogenesis of HTN virus in newborn mice (23, 26, 29). Although newborn mice develop fatal illness, resistance to the disease increases with age (33). In newborn mice experimentally infected by any inoculation route, the virus causes a widespread infection with high titers of virus in almost all organs. Moreover, the site of viral multiplication is endothelial cells and monocytes/macrophages, just as in human infection (16, 26, 31, 36, 49, 57, 62). Similar to fatal and severe cases of human HFRS (1, 2, 6), infected mice develop inflammatory and destructive lesions in various organs and die within 2 to 4 weeks, with wasting and neurological signs (23, 26, 33). Several studies suggest that these lesions are immune mediated (29, 32, 59). These investigations suggest that a newborn mice model may be useful for studying the spread of HTN virus in a host body and the pathogenesis of acute systemic infection.

Similar to other bunyaviruses, the hantavirus genome consists of three negative-strand RNA segments, the large (L), medium (M), and small (S) segments, which encode RNA-dependent RNA polymerase (L protein), two surface glycoproteins (G1 and G2), and nucleocapsid protein (NP), respectively (43, 45, 46). The 3′ and 5′ termini of the hantavirus genome segments are complementary to each other and highly conserved. The complementary sequences at the 3′ and 5′ termini of each segment are capable of forming stable base-paired panhandle structures and probably are the basis of the noncovalently closed circular RNA structures. The panhandles of hantaviruses are thought to play a role in regulating viral transcription and replication and to serve as recognition sites for viral RNA polymerase (12, 41, 42).

Unlike other viruses that cause disease in animals, there are few studies of the molecular basis of HTN virus virulence. Tamura et al. reported two plaque-purified variant clones derived from strain HTN virus 76-118: a highly virulent clone 1 (cl-1) against newborn mice and an avirulent clone 2 (cl-2) (47). Sequencing analysis of these viruses suggested that a single amino acid substitution in the G2 glycoprotein determined virulence in mice (20). Several similar studies of other members of the Bunyaviridae (LaCrosse and Rift Valley fever virus) involving genetic reassortants between virulent and avirulent strains have been reported (11, 14, 21, 52). In segmented-RNA viruses, study of the genetic reassortment between different virulence strains can be used to determine the genome segment associated with their virulence. However, further characterization of the genetic determinants relating to pathogenesis is difficult when there are numerous nucleotide differences between the virulent and avirulent strains. Recently, reverse genetics systems were established for negative-stranded RNA viruses, allowing characterization of the molecular basis of viral pathogenesis and of their biological properties. However, no such system has yet been established for bunyaviruses except for Bunyamwera virus (5). Therefore, reassortment between monoclonal antibody (MAb)-resistant variants is also effective for genetic mapping of virulence. Recently, we generated a mutant virus (designated mu11E10) from HTN virus strain cl-1 by selection with neutralizing MAb 11E10 to a glycoprotein G2 (3). mu11E10 was highly attenuated against newborn mice. Sequencing of the M segment of mu11E10 showed that this involved a single amino acid substitution: isoleucine (cl-1) to threonine (mu11E10) in position 515 of the G1 glycoprotein (22). However, except for the coding region of the mutant M segment, the entire nucleotide sequences of the L, M, and S genome segments have yet to be determined. Therefore, the significance of the single amino acid change in glycoprotein G1 in virulence is unclear.

In this study, we compared the pathogenesis of HTN virus in newborn mice infected with virulent (cl-1) and attenuated (mu11E10) strains. In a recent study, Rodriguez et al. successfully generated genetic reassortment among viruses causing HPS (38), although no study of the generation of HTN virus reassortants in vitro has been reported. Therefore, this is the first report of the generation of genetic reassortant viruses between the two viruses, and it identifies the genetic determinant related to their virulence.

MATERIALS AND METHODS

Infectivity titration and cell culture of viruses.

Two strains of HTN virus that differ in virulence to neonatal mice, cl-1 (virulent) (47) and mu11E10 (22) (attenuated), were used in this study. Cl-1, which is derived from strain 76-118, was provided by K. Yamanishi, Osaka University Medical School, Osaka, Japan. Cl-1 was passaged two times in Vero E6 cells cultured in our laboratory and used as the stock virus. The entire RNA genome sequence of cl-1 used in this study was determined (Table 1) and compared with the genome sequence of cl-1 published by Isegawa et al., since the passage history of strain cl-1 propagated in our laboratory differed from the published one (20). There were seven and six nucleotide differences in the L and M segment sequences between the published cl-1 and the cl-1 used in this study, respectively. An amino acid substitution was observed in the L segment (data not shown). Strain mu11E10 was generated as a neutralizing MAb escape mutant from the cl-1 used in this study, using MAb to envelope protein G2 (11E10) (3), as described previously (22). Working stocks of both strains were prepared from the culture supernatants of Vero E6 cells infected with each virus.

TABLE 1.

Genetic determinants of HTN virus virulence in newborn mice

Virus	Genotypeb	Nucleotide and amino acid substitutiona				Virulence
		L			Mc	s.c.		i.c.
		2038	3373	6497 (NC)d	1584 (515e)	Mortalityf (%)	Mean survival time ± SD (days)	Mortality (%)	Mean survival time ± SD (days)
Parent
cl-1	c c	G	T	T	T (Ileg)	16/18 (88.8)h	24.3 ± 3.5	7/7 (100)	19.7 ± 0.5i
mu11E10	m m	C	C	C	C (Thr)	0/7 (0)h		7/7 (100)	24.0 ± 2.1i
Reassortant
RC-08	m c	C	C	C	T (Ile)	35/41 (85.4)h	24.4 ± 3.5j	6/6 (100)	22.5 ± 0.5i
RC-37	c m	G	T	T	C (Thr)	6/34 (17.6)h	26.8 ± 5.3j	5/5 (100)	23.2 ± 0.8i

^aNumbers correspond to those of cl-1 and mu11E10 beginning at the viral cDNA sequence. 

^bGenome segments derived from cl-1 (c) and from mu11E10 (m). 

^cNucleotide sequence of M segment of mu11E10 was reported by Kikuchi et al. (22). 

^dNC, noncoding region. 

^eAmino acid position. 

^fNumber dead/total number tested. 

^gDeduced amino acid. 

^hP < 0.001; cl-1 versus mu11E10, RC-08 versus RC-37, cl-1 versus RC-37, mu11E10 versus RC-08. 

ⁱP < 0.001; cl-1 versus mu11E10, cl-1 versus RC-08, cl-1 versus RC-37. 

^jP < 0.001; RC-08 versus RC-37. 

Virus infectivity titers (focus-forming units [FFU]) were measured by counting the number of infected cell foci detected by an indirect immunofluorescent antibody assay (IFA) (58) using previously described procedures with some modification. Serial 10-fold-diluted specimens were adsorbed on monolayers of Vero E6 cells grown in 96-well plastic plates for 1 h at 37°C in 5% CO₂. Then the cells were overlaid with growth medium containing 1.5% carboxymethyl cellulose sodium salt (Wako Pure Chemical, Osaka, Japan) and incubated for 4 to 5 days at 37°C in 5% CO₂. After incubation, the medium containing carboxymethyl cellulose was removed, and the cells were washed with phosphate-buffered saline (PBS) three times and fixed with acetone-methanol mixed 1:1 for 10 min at room temperature. To detect HTN virus antigen, we used anti-NP MAb c16D11 from mouse ascitic fluid (60). After c16D11 treatment, cells were incubated with fluorescein isothiocyanate-conjugated goat anti-mouse immunoglobulin G (IgG) [IgA + IgG + IgM (H + L); Zymed, South San Francisco, Calif.].

E6 clones of Vero cells (ATCC C1008 CRL1586) and a subcloned line of CV-1 African green monkey cells designated CV-7 were grown in Eagle's minimal essential medium (MEM; Nissui, Tokyo, Japan) supplemented with 5% fetal bovine serum (FBS), 2 mM l-glutamine, kanamycin (60 mg/liter), streptomycin (100 mg/liter), and penicillin (10⁵ U/liter) at 37°C.

Mice and animal experiments.

Specific-pathogen-free pregnant inbred BALB/c/slc mice were obtained from SLC (Hamamatsu, Japan). Within 24 h after birth, neonatal mice were inoculated subcutaneously (s.c.) or intracerebrally (i.c.) with 1,000 FFU of cl-1, mu11E10, or genetic reassortants between the two parental viruses. The mortality, clinical signs, and body weight of each of the 5 to 41 mice in each group inoculated were recorded for days 4 to 35 after inoculation.

All animals were treated according to the laboratory animal control guidelines of our institute, which conform to those of the U.S. National Institutes of Health. All animal experiments were carried out in a class P3 facility.

Histological studies and immunohistochemistry.

The organs were collected from two mice at 2- or 4-day intervals from 4 to 22 days after inoculation. The mice were anesthetized and killed by exsanguination. Brains, lungs, spleens, and kidneys were excised aseptically. Whole organ tissue was fixed in Bouin's solution for 24 h at 4°C; the specimens were embedded in paraffin, sectioned at 5 μm, and stained with hematoxylin and eosin.

To detect HTN virus antigens in tissue, paraffin sections were stained by the peroxidase-antiperoxidase (PAP) staining method or the avidin-biotin-peroxidase complex (ABC) method using a Vectastain elite ABC rabbit IgG kit (Vector Laboratories, Inc., Burlingame, Calif.). In brief, thin sections were deparaffinized in xylene, hydrated in ethanol, and washed in PBS. The sections were incubated for 30 min in 0.1% H₂O₂ in methanol to block endogenous peroxidase and then in normal goat serum to reduce background staining. For the PAP method, the sections were incubated at 37°C for 2 h with a mixture of three MAbs to NP (c16D11 from mouse ascitic fluid, diluted to 1:100; f23A1 from the culture supernatant of hybridoma cells; E5G6 from the culture supernatant of hybridoma cells) as the primary antibody. For the ABC method, anti-SR-11 (Seoul-type hantavirus) rabbit serum diluted 1:500 in 10% Block Ace (Yukijirushi, Sapporo, Japan) was used as the primary antibody. The sections were incubated with bovine anti-mouse IgG (Fab) diluted 1:100 and monoclonal mouse PAP complex (Zymed) for 1 h at 37°C, in succession. For the ABC method, successive incubations in biotinylated goat anti-rabbit IgG in 10% Block Ace and ABC complex in 10% Block Ace were performed for 30 min at 37°C. After each serum treatment, the sections were washed with PBS. Peroxidase activity was demonstrated using 3,3-diaminobenzidine tetrahydrochloride (3 mg/10 ml; Sigma) in 0.03% H₂O₂; staining was done under microscopic control. Sections were counterstained with hematoxylin. Uninfected tissues or specimens without the primary antibody served as negative controls.

Virus titration in organs of mice infected with cl-1 and mu11E10.

Two mice were sacrificed at 2- to 4-day intervals from days 4 to 22 after inoculation. The mice were anesthetized and killed by exsanguination, and blood samples were collected at the same time. Brains, lungs, spleens, and kidneys were excised aseptically. The organs were frozen and thawed once, and 10% homogenates were prepared with PBS. The 10% homogenate was centrifuged at 6,000 rpm for 5 min at 4°C, and the supernatant was stored at −80°C until assayed. The infectivity titers of the supernatants were assayed by measuring focus formation (see above) in serial 10-fold dilutions on Vero E6 cell monolayers in 96-well plastic plates.

Detection of serum viremia in infected mice by semiquantitative RT-PCR.

To optimize the conditions of reverse transcription-PCR (RT-PCR) for semiquantitation of the virus titer in the serum of infected mice, we first determined the numbers of cycles of PCR required to amplify viral RNA (vRNA) corresponding to the virus titers. RNA was extracted from serially diluted 250-μl culture supernatants including 10⁵ to 10 FFU of viruses (cl-1) using Isogen-LS (Nippon Gene, Tokyo, Japan) according to the manufacturer's instructions, and the extracted RNA was resuspended in 5 μl of diethylpyrocarbonate treated water. To synthesize the cDNA of the S segment genome RNA, we used a specific oligonucleotide primer (cm-S459F; 5′-GACAACAAGGGGGAGGCAAACTACCAAGG-3′) complementary to the HTN virus S segment vRNA. vRNA isolated from culture supernatants (virus titer, 10⁵ to 10 FFU) and 2.5 μM cm-S459F in 20 μl of solution, containing First-Strand Buffer (50 mM Tris-HCl [pH 8.3], 75 mM KCl, 3 mM MgCl₂; Gibco BRL), 20 mM dithiothreitol, and 0.5 mM deoxynucleoside triphosphates, was incubated with 200 U of Superscript II RNase H⁻ reverse transcriptase (Gibco BRL) at 55°C for 1 h, 70°C for 10 min, and then at 99°C for 1 h. Specific oligonucleotide primers for S segment cm-S459F and cm-S675R (5′-CCGTGCCTTAATCTGTGCAGGGTAGAGCCC-3′) were used to amplify a 216-bp PCR product of the S segment. PCR amplification was carried out with AmpliTaq Gold DNA polymerase (Perkin-Elmer, Branchburg, N.J.) using a RoboCycler gradient 96 temperature cycler (Stratagene, La Jolla, Calif.) for 40 to 18 cycles in one-cycle steps, each consisting of 94°C for 30 s, 55°C for 30 s, and 72°C for 1 min. The reaction mixture was made up of PCR buffer II (Perkin-Elmer), 1.6 mM MgCl₂, 1 mM deoxynucleoside triphosphates, and 0.5 U of AmpliTaq Gold. The PCR products were electrophoresed on 2% agarose gels containing ethidium bromide, and the optimal number of PCR cycles for detecting PCR products corresponding to each virus titer was determined. The optimum numbers of PCR cycles for detecting vRNA were 35, 33, 31, 29, and 23 cycles for 10, 10², 10³, 10⁴, and 10⁵ FFU, respectively. The relationship between the number of PCR cycles and the viral titer did not change in two independent experiments. To confirm whether these conditions for RT-PCR would detect vRNA in mouse sera, we performed the same RT-PCR study using a 1:1 mixture of culture supernatants (viral titers, 10 to 10⁵ FFU) and normal mouse serum. RNA was extracted from untreated mixtures and mixtures that were heat inactivated at 56°C for 30 min. No change in the relationship between the number of PCR cycles and the virus titers of the mixtures was observed with the different conditions. Thus, we were able to detect serum viremia in infected mice using a semiquantitative RT-PCR method.

The blood samples collected from two mice (see above) were pooled. RNA was extracted from 60 μl of serum using Isogen-LS. The virus titer in the serum was then determined by the semiquantitative RT-PCR procedure just described.

Detection of serum IFA antibodies in infected mice.

IFA was carried out using acetone-fixed smears of Vero E6 cells infected with HTN virus as antigen. Fluorescein isothiocyanate-conjugated goat anti-mouse IgG [IgA + IgG + IgM (H + L); Zymed] was used as the second antibody. The IFA titers were expressed as the reciprocal of the highest dilution of the antisera that resulted in a specific immunofluorescence in the cytoplasm of the infected cells.

Sequencing analysis of cl-1 and mu11E10.

Total cellular RNA was extracted from Vero E6 cells infected with each stock virus (cl-1 and mu11E10) using Isogen (Nippon Gene) according to the manufacturer's instructions. cDNA representing the complete HTN virus L, M, and S segments was synthesized using a synthetic oligonucleotide complementary to 14 bases of the 3′ end of the vRNA sequence conserved in all segments (14-Primer; 5-TAGTAGTAGACTTC-3′) (46). The reverse transcription procedure was as described above. Primers designed from the published sequence of cl-1 were used to amplify several overlapping cDNA fragments of the L, M, and S genome segments by PCR. The PCR conditions were those described above. The PCR products were separated on 1% agarose gels, and the fragments were excised and purified using a GeneClean III kit (Bio 101, La Jolla, Calif.) according to the manufacturer's instructions. Direct sequencing of the purified PCR products was then performed using the same or newly synthesized primers used to amplify the fragments. Purified PCR products were sequenced by the dideoxy-chain termination method using an ABI PRISM dye terminator cycle sequencing kit (Perkin-Elmer, Applied Biosystems Division, Foster City, Calif.). The samples were sequenced on a model 373A or 377 DNA sequencing system (Perkin-Elmer, Applied Biosystems Division). Sequence data were analyzed using GENETYX-MAC sequence analysis software (Software Development Co., Ltd.) on a Macintosh computer. Two independent sequencing analyses were done.

Sequencing of the 3′ and 5′ termini of all segments was carried out by an RNA ligase-based method (48). vRNA was isolated from 1 to 2 ml of culture supernatant using Isogen-LS (Nippon Gene). The free vRNA genome ends were then ligated overnight with T4 RNA ligase (Takara, Tokyo, Japan). After RNA ligation, the ligated RNA was again extracted with Isogen-LS (Nippon Gene). After reverse transcription using oligonucleotide-specific primers (for the L and S segments) or random hexamers (for the M segment), nested PCR was done using primers specific for each segment that flanked the 5′ and 3′ termini of cDNA synthesized from ligated vRNA. The nucleotide sequences of the 3′ and 5′ termini of each segment were confirmed with six clones of each segment, after cloning the PCR products into plasmid pCR2.1 (TA cloning kit; Invitrogen, San Diego, Calif.). Sequencing was then performed using the same or newly synthesized primers to amplify the fragments, and the results were analyzed as described above.

Generation and recovery of genetic reassortants between cl-1 and mu11E10.

Genetic reassortants between cl-1 and mu11E10 were generated by coinfection of Vero E6 cells (2.5 × 10⁶ cells/T-25 flask) at multiplicties of infection (MOIs) of 0.004 for cl-1 and 0.012 for mu11E10. At 72 h after coinfection, the supernatants were harvested and plaque purified on CV-7 cells. Briefly, the supernatants were adsorbed on a monolayer of CV-7 cells grown in six-well plastic plates (2 × 10⁵ cells) for 1 h at 37°C. After adsorption, cells were overlaid with 2 ml of growth medium supplemented with 10% FBS, 2 mM l-glutamine, and 0.1 mM MEM nonessential amino acids solution (Gibco BRL) containing 0.75% SeaKem ME agarose (FMC BioProducts, Rockland, Maine) and incubated for 9 days at 37°C in 5% CO₂. After the first overlay incubation, the monolayers were stained by the addition of a 2-ml layer of the same agarose-medium solution containing 5% neutral red solution (Gibco BRL). Plaques were visualized 2 to 3 days later, and well-separated plaques were harvested by plunging a pipette directly into the agarose. The viruses were inoculated on Vero E6 cells grown in a 96-well plastic plate for 7 days and then passaged to Vero E6 cells in 24-well plates. Since the titers of plaque-purified viruses were low, we used soybean agglutinin (SBA), which is an N-acetylgalactosamine-specific lectin, to enhance virus adsorption to cells, as previously described (35). The mixture of culture supernatant harvested from Vero E6 cells in the 24-well plates and 500 μg of SBA (Vector Laboratories) per ml was incubated for 30 min and adsorbed for 60 min on Vero E6 cells in T-25 flasks at 37°C in 5% CO₂. After incubation for 7 days on Vero E6 cells in T-25 flasks, the culture supernatants were harvested, treated with SBA, and passaged into Vero E6 cells in T-150 flasks. The culture supernatants of Vero E6 cells in the T-150 flasks were used as the working stock virus.

The genotype of each virus was analyzed by restriction endonuclease digestion of RT-PCR products amplified from total cellular RNA isolated from infected cells. Reverse transcription was primed using 14-Primer. Then RT-PCR was carried out using primer pairs to amplify the regions of the L (nucleotides [nt] 1878 to 2098) and M (nt 1411 to 2030) segments containing the strain-specific mutation. The restriction endonucleases DdeI and SspI were used to detect strain-specific mutations on the L and M segments, respectively (see Fig. 6B). Each fragment digested with restriction endonucleases was analyzed by 2.4% agarose gel electrophoresis.

FIG. 6.

Determination of the genotype of reassortants between cl-1 and mu11E10 by restriction endonuclease digestion (A) and a schematic diagram of genotype screening of reassortants by restriction endonuclease digestion (B). (A) L and M, L and M RNA segments, respectively; c and m, genome segments derived from cl-1 and mu11E10, respectively. Lanes 1 through 4 show the restriction enzyme digestion patterns of the parent viruses (DdeI, SspI, DdeI, and SspI, respectively). Lanes 5 through 8 show genetic crosses between cl-1 and mu11E10. Lanes 5 (DdeI digested) and 6 (SspI digested) show the patterns for RC-08, which is a representative clone of genotype m c reassortants. Lanes 7 (DdeI digested) and 8 (SspI digested) show the patterns for RC-37, which is a clone of a genotype c m reassortant. Lane M, marker DNA fragments with sizes as indicated. (B) Genotypes of reassortants were determined by restriction endonuclease digestion of the RT-PCR products (L [220 bp] and M [619 bp]). Restriction endonucleases DdeI and SspI recognized strain-specific mutations on the L (2,038 nt) and M (1,584 nt) segments, respectively.

Viral growth kinetics in putative target cells in vitro.

(i) Mouse brain microvascular endothelial cells (MBMEC; Cell Applications, Inc.) were grown in MBMEC growth medium with growth supplement (Cell Applications). MBMEC (3.0 × 10⁶ cells in T-75 flasks) were infected at an MOI of 0.03. Culture supernatants were harvested at 1- to 2-day intervals on days 1 to 8 after infection. The growth medium was exchanged at 4-day intervals.

(ii) Peritoneal exudate cells (PEC) were collected from adult BALB/c mice without previous stimulation by washing the peritoneal cavity with Hanks' balanced salt solution (Nissui) supplemented with kanamycin, streptomycin, and penicillin. The cells were washed twice by repeated centrifugation in Dulbecco's modified Eagle's medium (Nissui). After incubation for 24 h, nonadherent cells were removed. Adherent cells were plated at 7 × 10⁵ cells in T-25 flasks and infected with viruses at an MOI of 0.1. Culture supernatants were harvested at 1- to 2-day intervals from days 1 to 8 after infection.

(iii) NA (mouse neuroblastoma) cells were grown in Eagle's MEM (Gibco BRL) supplemented with 10% FBS, 2 mM l-glutamine, kanamycin (60 mg/liter), streptomycin (100 mg/liter), and penicillin (10⁵ U/liter) at 37°C. NA cells (2.0 × 10⁶ cells in T-25 flasks) were infected at an MOI of 0.03. Culture supernatants were harvested at 12-h intervals from 12 to 96 h after infection.

Statistics.

The significance of differences in mortality and mean day of death was determined using Fisher's exact probability test and Student's t test.

RESULTS

Comparison of virulence of cl-1 and mu11E10.

To compare the virulence of cl-1 and mu11E10, newborn BALB/c mice were inoculated s.c. or i.c. with 1,000 FFU of cl-1 or mu11E10 within 24 h after birth. The mortality, clinical signs, and body weight of each of the inoculated mice were recorded from days 4 to 35 after inoculation (Fig. 1). The survival rates at 35 days for cl-1 and mu11E10 after s.c. inoculation (Fig. 1A) were 11 and 100%, respectively (P < 0.001). Thus, we confirmed the significant difference in the virulence of cl-1 and mu11E10 described by Kikuchi et al. (22). The clinical features of mice infected with cl-1 first became apparent 2 weeks after s.c. inoculation. By days 13 to 14, the coats of the mice began to appear ruffled, and the animals were hyperexcitable. Paralysis of the hind limbs, hunched posture, and progressively diminishing mobility appeared by days 15 to 18. The first deaths occurred 3 weeks after inoculation. In contrast, mice inoculated with mu11E10 did not exhibit clinical signs, except for the appearance of ruffled fur by days 13 to 14. The mice recovered from their illness by day 20 to 25. Furthermore, the respective body weights of mice infected with cl-1 and mu11E10 were 50 and 80% that of control mice (Fig. 1C). Therefore, the difference in body weights of the mice infected with each virus was also consistent with their clinical course and mortality.

FIG. 1.

Survival and body weights of newborn mice infected s.c. and i.c. with cl-1 or mu11E10 and of uninfected (control) mice. Newborn BALB/c mice were inoculated s.c. or i.c. with 1,000 FFU of cl-1 (n = 18 [s.c.] and 7 [i.c.]) or mu11E10 (n = 7 [s.c. and i.c.]) within 24 h of birth. Mortality (A and B) and body weight (C and D) were recorded for days 4 to 35 (body weight, until day 30) after inoculation. Crosses indicate that all mice infected with HTN virus died. Each point is the mean weight of two mice. Vertical bars indicate the range of values. The range of values for 4 to 22 days after inoculation is ±0.3 g (C and D).

In contrast, mortality was 100% in mice inoculated with either virus i.c. Infected mice showed the same clinical signs and had body weights 50% less than those of control animals (Fig. 1D). However, the mean day of death (±standard error of the mean) of cl-1-infected mice (19.7 ± 0.5) was significantly earlier than that (24.0 ± 2.1) of mu11E10-infected mice (P < 0.001) (Fig. 1B).

Further, to exclude the possibility that the appearance of the revertant of mu11E10 contributed to the death of mice inoculated i.c., we used the s.c. route to inoculate five newborn mice with brain homogenate (1,000 FFU) recovered from mice infected i.c. with mu11E10. None of these mice developed severe clinical signs, and all survived (data not shown).

Relationship between histopathological changes and pathogenicity.

To explain the difference in the pathogenesis of mice infected with cl-1 and mu11E10, we examined the pathological changes in the organs of mice infected s.c. with either virus. Necrosis of neurons, infiltration of inflammatory cells in the cerebral cortex, and meningoencephalitis were observed to be more intense in the brains of mice infected with cl-1 (Fig. 2A) than in those of mu11E10-infected mice (Fig. 2B). Necrosis of neural cells was evident after day 14 in cl-1- and mu11E10-infected mice. These histopathological changes in the brain seemed to correlate to the clinical course of mice infected with HTN virus. No marked histopathological changes were observed in the lung, kidney, or spleen of cl-1- and mu11E10-infected mice (data not shown).

FIG. 2.

Histopathological changes of mouse brain after s.c. inoculation of cl-1 (A) or mu11E10 (B) and normal control mice (C). Brains were removed, fixed, sectioned, and stained with hematoxylin and eosin as described in Materials and Methods. (A) Cl-1-infected mouse brain showing intense necrosis of neurons and infiltration of inflammatory cells in the cerebral cortex and meningoencephalitis. (B) Mutant 11E10-infected mouse brain showing mild sclerotic lesions and necrosis of neural cells in the brain, which was less intense than that observed in cl-1-infected mice. (C) No changes were observed in the normal mouse brain. These photographs were taken at an original magnification of ×50.

Difference in the in vivo level of viral multiplication and tropism in tissues between cl-1 and mu11E10.

To characterize the growth kinetics and tropism of cl-1 and mu11E10, we compared the time of detection and the level of viral antigen in several organs of mice inoculated s.c. with either virus (Fig. 3). In addition, the virus titers were determined in tissues of mice infected s.c. and i.c. (Fig. 4).

FIG. 3.

Distribution of NP in the tissues of mice infected with cl-1 or mu11E10. The brains, lungs, kidneys, and spleens were removed from two mice infected s.c. with each virus, fixed, sectioned, and immunostained with anti-HTN virus MAbs and serum, using the PAP or ABC method as described in Materials and Methods. The brain and lung sections obtained on day 14 after inoculation and the kidney and spleen sections obtained on day 10 after inoculation are shown. The antigen-positive foci rapidly spread widely in every tissue of cl-1-infected mice compared with the tissues of mu11E10-infected mice. Viral antigen was detected in the neurons and glial cells from the brains of infected mice (A and B). In the lung, viral antigen was detected in alveolar epithelial cells and endothelial cells (C and D). In the kidney, viral antigen was detected in tubulointerstitial endothelial cells, fibroblasts (E and F), and mesangial cells (F). In the spleen, viral antigen was detected in splenic sinus endothelial cells and macrophage-like cells (G and H). Organs from mock-infected mice showed no positive staining for viral antigen (data not shown). The magnification is shown in each photograph.

FIG. 4.

Virus titers in the tissues of newborn mice infected s.c. and i.c. with cl-1 and mu11E10. Newborn mice were inoculated s.c. or i.c. with 1,000 FFU of cl-1 or mu11E10. Two mice were sacrificed at 2- to 4-day intervals from days 4 to 22 after inoculation. The brain (A) lung (B), kidney (C), and spleen (D) were collected and stored at −80°C. Crosses show that all mice infected with HTN virus died. Each point is the mean titer for two samples. Vertical bars indicate the range of values.

After s.c. inoculation, viral cl-1 antigen was detected in the brain beginning 6 days after inoculation. In contrast, mu11E10 antigen was first detected in the brain 10 days after inoculation, although in other organs viral antigens were first detected 6 days after inoculation (data not shown). Furthermore, cl-1 spread faster than mu11E10 in every organ (Fig. 3).

As shown in Fig. 4, the cl-1 virus titers were 10- to 100-fold higher than the mu11E10 titers in every tissue tested, by both s.c. and i.c. inoculation. With s.c. cl-1 inoculation, infectious virus was detected in the brain and kidney beginning on day 8. In mice infected s.c. with mu11E10, detection of the infectious virus was delayed by 4 days (Fig. 4A and C). Particularly, the kinetics of growth in the brain were consistent with the antigen detection study.

With i.c. inoculation, in contrast to s.c. inoculation, the growth kinetics of both viruses in the brain were similar, but titers of cl-1 were three- to seven-fold higher than those of mu11E10 (Fig. 4A). In other tissues, as with s.c. inoculation, cl-1 multiplied more than mu11E10 (Fig. 4B to D).

The histopathological study indicated that the brain was the major target tissue of HTN virus infection in newborn mice, which suggests that the kinetics of viral growth in the brain are the key to the pathogenesis of viral virulence in mice inoculated s.c.

Relationship between serum virus titer and antibody response with stage of infection.

To further characterize the mechanism of viral spread in vivo, we measured vRNA in the sera as a viral titer, using a semiquantitative RT-PCR system (described in Materials and Methods) and IFA antibody titers, in mice infected either s.c. or i.c. In the late stage of infection, direct measurement of serum virus titers in infected mice by neutralization of virus with increasing serum antibody is difficult. Therefore, we used the level of vRNA measured by semiquantitative RT-PCR system as an indicator of the virus titer.

In the early stage of infection (days 1 to 6 after inoculation), cl-1 vRNA was detected in the sera 4 days sooner and with higher titer (>10⁴ FFU) than mu11E10 vRNA in mice infected s.c. (Fig. 5A). This difference in the time of detection was consistent with the kinetics of viral growth in the tissues. With i.c. inoculation, a similar tendency was observed. IFA antibody was not detected in the sera in the early stage of infection, regardless of virus or inoculation route (Fig. 5C and D). After s.c. inoculation with mu11E10, vRNA could not be detected on day 6.

FIG. 5.

vRNA and IFA antibody titers in sera of mice infected s.c. (A and C) or i.c. (B and D) with cl-1 (circles) or mu11E10 (squares). The serum was collected from the infected mice described for Fig. 4. (A and B) Correlation of the amount of viral RNA to virus titer; (C and D) IFA antibody titers. Optimization of the RT-PCR conditions for semiquantitation of virus titers in infected mice sera is described in Materials and Methods. IFA antibody was measured by assay using acetone-fixed smears of Vero E6 cells infected with HTN virus as antigen.

In the mid to late stage of infection after s.c. inoculation (days 8 to 22 after inoculation), the level of vRNA in the sera reached a plateau (Fig. 5A). IFA antibody was detected in the sera beginning 2 weeks after inoculation (Fig. 5C). With increasing serum antibody titers in infected mice, the virus titers in the peripheral organs started to decrease, except in the brain. In contrast, vRNA did not clear from the sera of infected animals. The virus titer plateau was slightly higher with i.c. inoculation than with s.c. inoculation.

Molecular differences between cl-1 and mu11E10.

To further ascertain the possible molecular basis for the difference in the virulence of the two viruses, we determined the complete nucleotide sequences of all the genome segments (L, M, and S) of both cl-1 and mu11E10. Table 1 shows the nucleotide and deduced amino acid substitutions for cl-1 and mu11E10. Nucleotides are indicated by reference to the cDNA sense sequence. There are three nucleotide differences in the L segment and one in the M segment. There were no nucleotide or amino acid substitutions in the S segment. In the L segment, two of three nucleotide changes located in the open reading frame were synonymous substitutions (position 2038, G in cl-1 and C in mu11E10; position 3373, T in cl-1 and C in mu11E10). Another nucleotide substitution was located in the noncoding region in the 5′ termini of the vRNA (position 6497, T in cl-1 and C in mu11E10). In the M segment, there was only one nucleotide substitution (position 1584, T in cl-1 and C in mu11E10); this resulted in an amino acid substitution at position 515 in glycoprotein G1 (isoleucine in cl-1 and threonine in mu11E10).

Furthermore, we determined the 3′ and 5′-terminal sequences of L, M, and S genome segments of cl-1 and mu11E10 by the RNA ligation method (48). We obtained six clones of ligated 3′ and 5′ termini of each segment for sequencing analysis. It is generally reported that HTN viruses (76-118, cl-1 published by Isegawa et al. [20], and other strains) have conserved cDNA sequences at the 5′ and 3′ termini of the L, M, and S segments. The 5′ sequence is 5′-TAGTAGTAG and the 3′ sequences are CTACTACTA-3′ in the L and M segments and ATACTACTA-3′ in the S segment. In contrast to these viruses, heterogeneity of complementary sequences at the 5′ and 3′ termini was observed in the L and M segments of cl-1 and mu11E10. In the L segment of cl-1, three of the six cDNA clones of the 5′ terminus had only two TAG repeats (5′-…TAGTAG). In the M segment, four of the six clones of the cl-1 M segment 5′ terminus had two TAG repeats (5′-…TAGTAG). In mu11E10, two of the six clones had 5′ termini of 5′-…TAGTAG and 5′-…AGTAG, respectively. In contrast to the 5′ termini, the 3′ termini of all cDNA clones of the L and M segments of both viruses changed from CTACTACTA-3′ to ATACTACTA-3′. No heterogeneity of the complementary sequence was observed in the S segment of either virus (data not shown).

Generation of genetic reassortants between cl-1 and mu11E10.

Since there were molecular differences between cl-1 and mu11E10 in the L and M segments, we generated genetic reassortants between the two viruses to identify the segment associated with virulence in newborn mice.

We picked up 21 plaque clones from supernatants of coinfected Vero E6 cells (Fig. 6A). The genotype of the reassortants was screened using the restriction endonuclease digestion pattern of RT-PCR products amplified from total cellular RNA isolated from infected cells (Fig. 6B). Of the 21 clones, 15 were reassortants between cl-1 and mu11E10. Fourteen reassortants possessed the mu11E10-derived L segment/cl-1-derived M segment (genotype m c), and one possessed the cl-1-derived L segment/mu11E10-derived M segment (genotype c m) (Fig. 6A). We selected the clones designated RC-08 (m c) and RC-37 (c m) as representative of the reassortants for further biological characterization.

Furthermore, we confirmed the antigenicity of the reassortants by IFA using MAb 11E10 (3) against glycoprotein G2, since mu11E10 was a neutralizing MAb 11E10 escape mutant (22). RC-37 (c m) had no reactivity against MAb 11E10 but was reactive against other MAbs to glycoproteins G1 and G2 (data not shown).

Determination of the genome segment associated with virulence.

To identify the genome segment associated with virulence in mice, the reassortants described above were inoculated s.c. and i.c. into newborn mice. The virulence of reassortants was determined using the same procedure described above. The survival rates at 35 days for RC-08 (m c) and RC-37 (c m) after s.c. inoculation (Fig. 7A) were 15 and 78%, respectively (P < 0.001). The mortality rate and clinical course of mice infected with RC-08 (m c) were similar to those of mice infected with parental cl-1 (Fig. 1A and 7A). RC-37 (c m) showed a significantly attenuated phenotype against mice compared with RC-08 (m c) but was slightly more virulent than parental mu11E10. These results indicate that the M genome segment is the most related to virulence, and the L segment is partly related in peripheral route infection.

FIG. 7.

Survival curve of newborn mice infected with reassortants (A) and viral multiplication in MBMEC (B). (A) Newborn BALB/c mice were inoculated s.c. or i.c. with 1,000 FFU of RC-08 (n = 41 [s.c.] and 6 [i.c.]) or RC-37 (n = 34 [s.c.] and 5 [i.c.]) within 24 h after birth, and mortality was recorded from days 4 to 35 after inoculation. (B) MBMEC were infected at an MOI of 0.03 with cl-1, mu11E10, RC-08, or RC-37. Culture supernatants were harvested 1, 2, 3, 4, 6, and 8 days after infection, and the virus titer was determined by focus-forming assay on Vero E6 cells. Points are the mean titer for two samples. Vertical bars indicate the range of values.

With i.c. inoculation, the mortality of RC-08- and RC-37-infected mice was 100%, similar to mice infected with the parental viruses (Fig. 7A).

Relationship between the genetic determinant of virulence in vivo and viral multiplication in putative target cells in vitro.

The in vivo analysis of virulence suggested that the M segment from mu11E10 contributed to the attenuation of virulence in newborn mice inoculated s.c. To investigate the mechanism for viral spread from peripheral tissues to the central nervous system (CNS), MBMEC (Fig. 7B), PEC, and neuroblastoma (NA) cells (data not shown) were infected with the reassortants and parental viruses, and their patterns of growth were compared. In MBMEC, the titer of viruses possessing the M segment from cl-1 (cl-1 and RC-08) was approximately 10-fold (10³ FFU) higher than that of viruses containing the M segment from mu11E10 (mu11E10 and RC-37) (Fig. 7B). In PEC collected from BALB/c mice, cl-1 and RC-08 grew with titers 4- to 10-fold higher than those of mu11E10 and RC-37 in the early period of growth. From day 4 after inoculation, however, only cl-1 maintained a higher titer (data not shown). In NA cells, the highest virus titers of all the viruses were too low to compare (<10² FFU) (data not shown). These results indicate that viruses that possessed the M segment from cl-1 grew with higher titers in MBMEC than attenuated viruses containing the M segment from mu11E10. Therefore, there was a correlation between viral multiplication in peripheral target cells and virulence associated with the M genome segment in vitro and in vivo.

DISCUSSION

In this study, newborn mice infected with virulent and attenuated HTN viruses were compared to examine the mechanism of pathogenesis in the mouse model. In addition, we generated genetically reassorted viruses from virulent and attenuated viruses to examine the genetic determinants related to virulence.

Pathogenesis of strains cl-1 and mu11E10 in newborn mice.

Several groups have reported that HTN virus causes neurological disease in newborn mice and that the brain is the major target organ (26, 29). Our study confirmed these findings by comparing the pathogenesis of a virulent virus and its attenuated mutant. When the viruses were inoculated via a peripheral route (s.c.), only the virulent cl-1 virus caused lethal infection, while all the mice infected with mu11E10 survived. The brains of cl-1-infected mice showed more severe histopathological changes than did those of mu11E10-infected mice. No apparent histopathological difference was observed in other organs. On the other hand, after i.c. inoculation, both viruses produced severe histopathological changes (data not shown) and all mice died. Therefore, the severity of lesions in the brain is thought to be the critical factor determining mortality in mice.

After s.c. inoculation, the virulent cl-1 virus grew faster and in higher titers than mu11E10 in all the organs examined (Fig. 4). This difference in the growth ability of cl-1 and mu11E10 was also observed in cells cultured in vitro: MBMEC (10-fold) (Fig. 7B) and Vero E6 cells (2- to 3-fold) (data not shown). Essentially, the higher growth ability of cl-1 in vivo and in vitro contributes to its higher virulence in newborn mice. Therefore, in the brain, cl-1 virus appeared 4 days earlier than mu11E10. These results indicate that mu11E10 has a generally less virulent phenotype, and because of decreased multiplication in peripheral tissues, its neuroinvasiveness is also decreased. Combining these observations, with s.c. inoculation the higher growth ability of cl-1 in peripheral tissues determines its virulence.

The invasion of the virus into the CNS with growth at high titers until 8 days after birth seems to be important in causing a fatal outcome. Several groups have reported the age-dependent susceptibility of hantavirus infection (32, 33). Although the results differed slightly depending on the mouse strain and the passage history of the virus strain, the survival rate of mice inoculated when they were at least 7 days old began to increase, even after i.c. inoculation. Passive transfer of immune spleen cells and antibody into infected newborn mice confers protection from a lethal challenge dose (4, 33, 61). Therefore, invasion of the CNS by the virus before the maturation of host resistance is a direct factor in fatal infection.

In our preliminary study, mice inoculated i.c. with virulent virus (cl-1) at ages <24 h, 2 days, and 4 days died at 20 to 28, 20 to 24, and 20 days of age, respectively. There was an inverse relationship between the mean number of days until the death of infected mice and the age at which they were inoculated. This suggests that a lethal outcome is dependent on host factors that are present 3 to 4 weeks after birth. Nakamura et al. reported a more rapid fatal course in immunocompetent (nu/+) mice than in T-cell-deficient (nu/nu) littermates (32). Moreover, Yoshimatsu et al. reported that the neurological disease characteristics of moribund newborn mice were uncommon in SCID mice (59). The combination of immune spleen cell-mediated protection and age-dependent resistance suggests that cell-mediated immunity might be responsible both for enhancing the disease and for recovery from infection. Therefore, an immunopathologic mechanism for the pathogenicity of HTN virus in newborn mice should be considered. Rapid invasion of the virulent virus into the CNS may induce an immune response in the CNS that is destructive rather than protective.

The mechanism of the age-dependent susceptibility of Sindbis virus infection in a mouse model has been studied. Antiapoptosis activity with maturation of neurons and a decrease in Sindbis virus receptors with maturation of neural cells were suggested as mechanisms (13, 50). In reovirus, differences in the cytokine profiles in the brain of infected neonatal and adult mice were also reported (8). However, these mechanisms have not been studied in animal models of hantavirus infection.

The newborn mouse model is frequently used for studies of hantavirus pathogenesis, although the resulting disease does not parallel human hantavirus infections. Unlike the rodents that are the natural reservoir of hantavirus, newborn mice develop a fatal illness with acute systemic infection. In addition, the long incubation period (typically 2 to 3 weeks), site of viral multiplication, and immune-mediated destructive inflammatory lesions in the CNS and various tissues observed in newborn mice are similar to the findings in severe human infection (16, 26, 31, 36, 49, 57, 62). Therefore, the newborn mouse model may provide useful information on fundamental aspects of acute systemic infection in humans.

A semiquantitative RT-PCR method successfully detected persistence of the virus in sera in which the virus could not be detected with existing antibodies. The persistent circulation and replication of virus with an antibody response in mu11E10-infected mice that exhibited transient symptomatic infection is similar to an experimental infection of Black Creek Canal virus causing HPS in its reservoir (Sigmodon hispidus) (19). Although the hantaviruses are maintained in persistently infected rodents in nature, the mechanisms by which a persistent infection is established remain unclear. Application of this method, using an attenuated strain in the mouse model, may prove a useful system for studying persistent infection in this field. However, it remains to be determined why vRNA levels corresponding to the high virus titers (10⁴ to 10⁵ FFU) are detected in sera without the detection of spleen virus titers. Since this RT-PCR system could not distinguish vRNA derived from infectious virus and plasma, the effect of trace RNA in plasma must be considered.

Genetic determinants of HTN virus virulence.

In this study, we produced genetically reassorted viruses from the parental virulent virus and its attenuated mutant, to examine the genetic determinants related to the virulence of HTN virus in the mouse. The survival rates after peripheral inoculation (s.c.) showed the M segment to be the most related to virulence and the L segment to be partly related. A single deduced amino acid substitution was detected at amino acid 515 of G1 protein, changing Ile in cl-1 to Thr in mu11E10. This amino acid is thought to be a major determinant of the greater viral multiplication in peripheral tissues in newborn mice. Since the survival of mice infected via the i.c. route with reassortant virus that possessed the mu11E10-derived L segment was significantly longer than that of the parental cl-1 (Table 1, P < 0.001), the L segment is thought to be a determinant of neurovirulence. Since the only nucleotide substitution found was in the 5′ noncoding region, the difference in neurovirulence might be related to the efficiency of L RNA synthesis rather than to altered function of the L protein.

A similar study involving genetic reassortants of the LaCrosse virus of the family Bunyaviridae showed the major determinants of neurovirulence and neuroinvasiveness to be located in the L and M segments, respectively (11, 14, 21). Therefore, polygenic control of virulence might be a common characteristic of viruses in the family Bunyaviridae. However, these reports did not define nucleotide or amino acid changes related to virulence.

The exact role of the M segment in virulence is unclear. As shown in Fig. 7B, higher growth activity in cultured MBMEC was related to the cl-1-derived M segment. A similar tendency was observed in PEC during the early period of growth (data not shown). This may induce higher virus replication in endothelial cells and macrophages/monocytes, which have been reported as target cells for hantavirus infection (26, 31), and ultimately contribute to faster spread in the CNS in the early stage of infection than with the attenuated virus.

Single amino acid changes in the envelope glycoprotein affect the pathogenicity and biological properties of various viruses in vivo and in vitro (9, 28, 39, 48). Tamura et al. reported that virulent (cl-1) HTN virus induced pH-dependent fusion of the infected cell membrane more efficiently than avirulent (cl-2) HTN virus (47). Furthermore, Isegawa et al. suggested that a single amino acid change of Ser (cl-1) to Gly (cl-2) in the predicted transmembrane domain of glycoprotein G2 is associated with virulence and membrane fusion ability (20). The position of the amino acid substitution of attenuated mu11E10 differed from that of cl-2. Interestingly, in our preliminary study, the virulent strains (cl-1 and RC-08 [Ile 515 on G1]) induced stronger pH-dependent cell fusion than that induced by the attenuated viruses (mu11E10 and RC-37 [Thr 515 on G1]) (data not shown). mu11E10 was generated as a neutralizing MAb (11E10) escape mutant and lacks the ability to bind the MAb. Although MAb 11E10 immunoprecipitates the G2 protein of HTN virus, the amino acid substitution is located in G1. Furthermore, an epitope mapping study using recombinant truncated glycoproteins expressed by recombinant vaccinia virus and baculovirus mapped the epitope of MAb 11E10 at the carboxyl terminus of the G2 protein, not in the transmembrane region (53). On the other hand, G1 is predicted to be a type 1 transmembrane protein (42). Although the single change in G1 at amino acid 515 is located in the hydrophobic region in the C terminus, this amino acid change contributes to a lack of ability to bind MAb 11E10. These results suggest that the MAb 11E10 epitope is a discontinuous epitope that is located in the carboxyl-terminal hydrophobic region proximal to the transmembrane domain and interacts with both G1 and G2. Amino acid changes related to the virulence of the two mutants (mu11E10 and cl-2) are located in the C-terminal hydrophobic regions of G1 and G2, respectively. Therefore, a single amino acid substitution in the hydrophobic region is responsible for changing the conformation of the envelope protein, which alters virulence and fusion activity. However, the relationship between fusion activity and virulence is still unclear.

Heterogeneity of the complementary sequences at the 5′ and 3′ termini was detected in the L and M segments of cl-1 and mu11E10 compared with the original sequences of HTN viruses. The variability seen between virulent and avirulent HTN viruses in this study may not affect their virulence; a similar lack of the complementary sequence at the 3′ end and a nucleotide substitution in the noncoding region of the termini of both L and M vRNA was observed in a comparison of the cl-1 used in this study and the published cl-1 sequence, both of which have similar virulence (20). On the other hand, Meyer and Schmaljohn proposed accumulation of deletions of the vRNA termini as one of the mechanisms for persistent Seoul virus infection in cultured cells; in that study, the amount of deleted L vRNA equaled the full-length vRNA in Vero E6 cells 7 days after infection (30). A similar phenomenon was reported for Tula virus (25). Deletion and heterogeneity of the terminal region of vRNA may be a general event for hantavirus infection in vitro.

Genetic reassortment of HTN virus.

Recently, Rodriguez et al. reported genetic reassortment among viruses causing HPS in an in vitro system (38). Several sets of genetic epidemiological data also suggest naturally occurring genetic reassortment in the field (17, 27). There are no reports of genetic reassortment of HTN virus in nature or in vitro. Our paper is the first report of artificial genetic reassortment by coinfection in vitro.

Unexpectedly, there was a clear predominance of the reassortants in progeny viruses compared with parental viruses. Furthermore, all except one reassortant possessed the M segment derived from cl-1. Rodriguez et al. reported a similar observation (38). Since cl-1 and RC-08 grew similarly in Vero E6 cells (data not shown), the actual mechanism for the predominance of a particular combination remains unclear. Furthermore, unlike other bunyaviruses, including HPS-related viruses (38, 51), no diploid virus was detected in our study. Virus activity (low stability of the diploid and lost ability to form the diploid) and the effect of passage in Vero E6 cells are possible causes of the absence of a diploid virus.

In this study, we determined the genetic factors relating to the pathogenicity of HTN virus in the newborn mouse model. However, to provide additional direct evidence and detailed information, it is necessary to establish a reverse genetics system for hantavirus.