点击显示 收起
Department of Pathology and Center for Biodefense and Emerging Diseases, University of Texas Medical Branch, Galveston
Special Pathogens Branch, Canadian Science Center for Human and Animal Health, Health Canada, and Departments of Immunology and Medical Microbiology, University of Manitoba, Winnipeg, Manitoba, Canada
The tickborne encephalitis (TBE) serocomplex of flaviviruses consists primarily of viruses that cause neurologic disease; these viruses include Omsk hemorrhagic fever virus (OHFV), a virus that is genetically related to other TBE serocomplex viruses but that circulates in an ecologically distinct niche and causes markedly different human disease. The objective of this study was to examine a potential small-animal model for OHFV and to compare the pathology of infection with that of the neurotropic tickborne flavivirus, Powassan virus (POWV). POWV-infected BALB/c mice demonstrated typical arboviral encephalitis, characterized by paresis and paralysis before death, and viral infection of the cerebrum, characterized by inflammation and necrosis. In contrast, lethal OHFV infection did not cause paralysis or significant infection of the cerebrum but showed marked involvement of the cerebellum. Distinct pathological results in the spleens suggest that the immune response in OHFV-infected mice is different from that in POWV-infected mice. This study demonstrates a clear pathological difference between OHFV-infected mice and POWV-infected mice and supports the use of the BALB/c mouse as a disease model for OHFV.
Omsk hemorrhagic fever (OHF) is a disease caused by the tickborne flavivirus, Omsk hemorrhagic fever virus (OHFV). Although OHFV is found only in a small region of Russia [1], this virus offers the unique opportunity to study a viscerotropic flavivirus in a small-animal model that does not require adaptation of the virus to the animal.
The flaviviruses are a genus of 70 viruses that are generally transmitted by either mosquitoes or ticks [2]. Mosquitoborne flaviviruses include the human pathogens dengue virus, Japanese encephalitis virus, West Nile encephalitis virus, and yellow fever virus (YFV). The tickborne flaviviruses are largely represented by viruses causing tickborne encephalitis (TBE) and are found primarily in Europe and Asia. Among the tickborne flaviviruses are 3Alkhurma virus (ALKV), Kyasanur Forest disease virus (KFDV), and OHFVthat cause diseases characterized by hemorrhagic manifestations; unlike ALKV and KFDV, OHFV causes a disease not typically characterized by neurological symptoms or sequelae [2]. Among the tickborne flaviviruses is Powassan virus (POWV), a neurotropic flavivirus found in North America and parts of Asia [3].
OHF is associated with occupational or recreational activities. OHFV is found in a small region near Omsk in Siberia, Russia. The natural mammalian host for this virus is the water vole (Arvicola terrestris) [1, 2, 4], but the virus has adapted to the nonnative muskrat (Ondata zibethica) [1, 2, 4]. OHFV has also been isolated from a number of other organisms native to the endemic region, and serosurveys have demonstrated the infection of a large number of species within this region, including many mammals, reptiles, and amphibians [1]. OHFV is transmitted via the bite of its primary tick vector, Dermacentor reticulates, although the virus has been isolated from D. marginatus [2]. It is not clear whether Ixodes species of ticks are involved in the transmission of OHFV; however, this seems unlikely, given the geographically limited endemic region of OHFV and the large, overlapping geographic range of I. persulcatus ticks that transmit other TBE viruses. Although ticks are the principal route of transmission of OHFV, the virus can also be transmitted via contact with bodily fluids or carcasses of infected animals [2], hence the frequent infection of muskrat trappers. OHFV, like other viruses causing TBE, can be transmitted via milk from infected goats or sheep [1], and the isolation of this virus from aquatic animals [1] and water [4] suggests that it is extremely stable in the environment.
OHF is an acute, febrile disease that, like the diseases caused by other tickborne flaviviruses, is frequently biphasic. Human OHFV infection results in clinical symptoms quite different from those caused by TBEV infection [3]. OHFV infection is associated with capillary toxicosis, papulovesicular rash on the soft palate, cervical lymphadenopathy, ocular suffusion, occasional neurological involvement, and often a distinct hemorrhagic syndrome that includes visceral hemorrhages in the nose, gums, uterus, and lungs [1, 3]. The incubation period for OHFV is 210 days, the febrile period is 512 days in 40% of patients, and a second febrile period occurs 1015 days after infection. The mortality rate for OHF is estimated to be 0.4%2.5% [1]. Histological examination of tissues from infected humans has not been described.
POWV was isolated in 1958, from a fatal case of encephalitis in Ontario, Canada [5], and is found in North America, including the northern United States and Canada, and parts of Russia [2, 3]. Although only a few cases of human infection have been reported, serosurveys in the United States and Canada suggest that the geographic distribution is wider than what would be anticipated on the basis of case reports [2]. The natural transmission cycle for POWV involves various Ixodes species of ticks, although the virus has also been isolated from D. andersoni [6]. Mammalian hosts include rodents, canines, raccoons, and skunks [7]. Human infection occurs via the bite of infected ticks and is associated with outdoor professional or recreational activity. POWV infection causes a typical arboviral encephalitis characterized by acute inflammation of the brain, spinal cord, and meninges [5, 8]; in severe cases, spastic paralysis, stupor, and coma can occur, and the case-fatality rate is up to 60% [3], and a large percentage of survivors have severe neurological sequelae [2].
The present study examines the histopathologic manifestations of OHFV infection in a mouse model and compares these findings with those in neurotropic POWV infection. The study's results show that, in BALB/c mice, OHFV causes a disease that is significantly different from that caused by POWV. Pathology data for OHFV infection in mice are similar to those described for OHFV infection in humans, and histologic examination of animal tissues shows a disease with mild meningoencephalitis and little cerebral involvement but significant cerebellar involvement. These results contrast to the severe disease caused by POWV, a meningoencephalitis characterized by significant inflammation and widespread necrosis of cerebral tissue. The present study offers the first description of a small-animal model for a flavivirus that causes a hemorrhagic disease in humansan animal model that, unlike that for YFV, does not require adaptation of the virus to the model system [911].
MATERIALS AND METHODS
Animals and viruses.
The animals used in the present study were 35 week-old female BALB/c mice supplied by Charles River Laboratories. OHFV strain Bogoluvovska was provided by the Centers for Disease Control in Atlanta, Georgia, and virus stocks were grown in Vero E6 cells. The virus stock is approximately passage 2 in suckling mouse brain, passage 24 in Vero E6 cells, and was derived from a stock produced on 3 November 1961. All work with live OHFV was performed under biosafety level (BSL)4 containment. A mouse-brainderived stock of POWV strain LB generated on 16 December 1965 was obtained from the World Arbovirus Reference Collection, University of Texas Medical Branch. The virus then was subjected to 3 passages in Vero cells and 3 passages in LLC-MK2 cells before being used in the present study.
Clinical and neurologic assessment.
The animals were observed daily for signs of illness, including malaise, ruffled fur, paralysis, and other general indications of illness in mice. Additionally, simple neurologic assessments were performed, including the following: landing tests, in which the healthy animals' ability to extend all 4 limbs in preparation for landing is assessed; balance tests, in which the animals' ability to negotiate the edge of the cage is assessed; and grasping tests, in which the healthy animals' ability to grab the cage lid when lifted by the tail is assessed. All studies were approved by local committees and were performed according to national guidelines for animal care.
Pathology.
Animals were divided into 2 groups; 20 of them were inoculated intraperitoneally (ip) with 1000 LD50 (1.25 × 104 pfu) of POWV/mL, 22 mice were inoculated ip with 1.3 × 103 TCID50 units of OHFV/mL, and, for each of the 2 types of virus, 4 mice were mock-inoculated to serve as controls. Beginning at 1 day postinfection (dpi) and ending at either 8 dpi (in the case of OHFV group) or 9 dpi (in the case of the POWV group), 2 inoculated mice were euthanized, and serum, liver, spleen, kidney, and brain were harvested. For each virus, 2 control animals were euthanized at dpi 1 and on the final day of the experiment, and similar tissues isolations were performed. The organs were fixed in 10% formalin, dehydrated, embedded in paraffin, and sectioned for histological studies. All procedures using the OHFV-infected tissue sections was performed in the BSL-4 facility in Winnipeg, and all procedures using POWV-infected tissue sections was performed in the BSL-3 facility in Galveston. Formalin was changed once before the fixed tissue sections were removed from biocontainment.
Histopathology.
For general histopathology examination, tissue sections were stained with hematoxylin/eosin (HE). For TUNEL assays and immunohistochemistry, slides were prepared by removal of the paraffin by Histochoice (Sigma), rehydration of sections by decreasing concentrations (100%, 95%, and 70%) of ethanol, and elimination of endogenous peroxidase by treatment, at room temperature (RT) for 10 min, with 3% H2O2 diluted in 20% methanol. Antigen retrieval was performed at RT for 20 min, by treatment with 20 g/mL proteinase K. TUNEL assays were performed by use of the In situ Cell DeathTMR kit (Roche), according to the manufacturer's instructions. Immunohistochemistry was performed at RT for 1 h, by incubation of the slides with a virus envelope protein domain IIIspecific polyclonal antiserum [12, 13] diluted in 0.5% bovine serum albumin/PBS and subsequent detection by use of a DAB detection system (DAKO Laboratories), which included an anti-rabbit biotinylated secondary antibody and horseradish peroxidaseconjugated avidin and was used according to the manufacturer's instructions. Slides containing tissue sections from control mice were included in all immunohistochemistry and TUNEL assays.
RESULTS
OHFV and POWV Infection in Mice
Both in mice inoculated ip with OHFV and in mice inoculated with POWV, the resulting infection was uniformly lethal, and the 2 groups had similar survival timesthat is, 68 dpi. However, the groups exhibited very different types of disease: the POWV-infected mice showed few signs of illness until 56 dpi, when they had general symptoms such as hunched posture, ruffled fur, and general malaise (table 1). Like mice infected with other neurotropic flaviviruses, the POWV-infected mice began to show indications of neurological illnesssuch as paresis, hind-limb paralysis, or total paralysisat dpi 67. The transition to this final stage of disease was rapid and very clear. In contrast, the OHFV-infected miceeven those that had been inoculated with undiluted stock virus at a titer of 1.3 × 105 TCID50 units/mLshowed no clear indications of neurological illness. The OHFV-infected mice, like the POWV-infected mice, had general signs of illness; however, when they were assessed by use of the landing, balance, and grasping tests described above, they showed no indication of a physical inability to perform the actions. At later stages of infection, the OHFV-infected mice were obviously very ill and were unable to complete the tests, although they clearly attempted to perform the tasks. Additionally, when examined postmortem, these mice had both conjunctival suffusion with crusting and extremely enlarged spleens, whereas none of the POWV-infected mice had either of these signs of illness. The differences in the gross manifestations of disease demonstrate that these viruses, although genetically quite similar, are physiologically very different and that these differences can be identified in the BALB/c mouse model. To gain a clearer understanding of the pathology of the diseases caused POWV and OHFV, comprehensive studies of individual organs were performed.
Histopathology
Mice were inoculated with either 1000 LD50 (1.25 × 104 pfu) of POWV/mL or 1.3 × 103 TCID50 units of OHFV/mL and then were euthanized daily, until the disease had run its course (e.g., 9 dpi for POWV and 8 dpi for OHFV). After euthanasia, serum, liver, spleen, kidney, and brain were harvested for histological examination. Significant histopathologic changes were noted, particularly in the spleen and brain.
Spleen.
Spleens of the POWV-infected mice showed discernible blastogenesis/activation in the periarteriolar lymphoid sheath (PALS) as early as 1 dpi. During the course of POWV infection, morphologic features of cell activation in the white pulp were prominent, and ill-defined germinal centers containing tingible body macrophages appeared at 6 dpi (figure 1C and 1D). At 8 dpi, the white pulp of the POWV-infected mice was populated almost entirely by large blastlike cells, with only focal apoptosis/tingible body macrophages. The red pulp also contained large macrophagelike cells. In contrast, OHFV-infected spleens showed less-robust cellular proliferation and activation in the PALS, beginning at 2 dpi. At all time points, the white pulp of the OHFV-infected mice appeared to be less organized and less activated than that of the POWV-infected mice. At 6 dpi, the white pulp of the OHFV-infected mice showed significant necrosis, with numerous nuclear fragments, tingible body macrophages, and confluent areas of cellular damage (figure 1A and 1B). The red pulp was congested, and the number of hematopoietic cells that it contained was somewhat increased. These changes continued until 8 dpi. Overall, these findings suggest early, strong, and sustained T and B cell blastogenesis in the POWV-infected mice, compared with that in the OHFV-infected mice; the white-pulp necrosis in the OHFV-infected mice was consistent with direct viral damage (see below).
Brain.
Neurological involvement of the brains of the POWV-infected mice began to appear at 6 dpi, which, in some animals, correlates with the progression of the disease and the appearance of paresis or paralysis. The POWV-infected mice developed clear meningoencephalitis, with infiltration of mononuclear cells into the subarachnoid space and the perivascular region around small parenchymal vessels (i.e., "perivascular cuffing") (figure 2). In all of the POWV-infected mice, mononuclear-cell infiltration was patchy, with some areas being severely involved and others being relatively uninvolved. In more severely involved areas, necrosis in the form of acute neuronal injury, edema, and karyorrhexis was evident. These lesions were found in all areas of the cerebrum, including the superficial cortex (in association with meningeal inflammation), the deep cortex, and the white matter, hippocampus, corpus striatum, and thalamus; however, no lesions were found in the cerebellar parenchyma, and there was only mild, patchy inflammation of the cerebellar meninges.
In the OHFV-infected mice, minimal signs of encephalitis or meningitis were evident at 4 dpi, with mild inflammation of the meninges and some evidence of perivascular cuffing but no evidence of necrotizing lesions (figure 2). Subsequently, the OHFV-infected mice had fully developed meningoencephalitis that, like that in the POWV-infected mice, was patchy in character but that had increased in both severity and extent at dpi 57. Although the types of lesions present were not significantly different from those in POWV-infected mice, the localization was different, with a striking difference in the involvement of the cerebellum: in the late stage of infection, POWV-infected mice had essentially normal cerebella, whereas OHFV-infected mice had significant pathology in their cerebella, including complete necrosis of Purkinje cells, necrosis of the granular layer of the cerebellum, with infiltration of mononuclear cells and, within the deep cerebellar nucleus, necrotic foci with hypereosinophilia of neurons and neuropil (figure 3). TUNEL assays indicated that many cells within the granular layer of the cerebellum were apoptotic (figure 3), whereas the Purkinje cells were not.
Liver and kidney.
The livers of the POWV-infected mice were essentially normal, except for minimal inflammation; the livers of the OHFV-infected mice appeared to have more severe inflammation, with evidence of mononuclear-cell infiltration and prominent Kupffer cells. In neither case was the pathology particularly striking. In both the POWV-infected mice and the OHFV-infected mice, the kidneys were essentially normal.
Virus Localization
Immunohistochemical studies were performed to determine virus localization in tissues from the OHFV-infected mice and from the POWV-infected mice. OHFV antigen was identified in the spleen and brain, whereas POWV antigen was seen in the brain but not in the spleen. OHFV antigen was also detected in endothelial cells of the liver; however, this staining was inconsistent and therefore inconclusive. Viral antigen was not detected in the kidneys of either the OHFV-infected mice or the POWV-infected mice.
Virus in the spleens of the OHFV-infected mice was scattered predominately throughout the red pulp, although some was seen in the white pulp, suggesting viral replication in the spleen (figure 4). Viral antigen was found in the liver of the OHFV-infected mice. OHFV antigen appeared to be present in the sinusoids of the liver, but this could not be absolutely confirmed by light microscopy. Viral antigen also appeared to be present in vascular endothelial cells and adjacent cells, although the amount was not striking, suggesting that, if OHFV infects vascular endothelial cells, it is not a highly productive infection (figure 4). Immunohistochemical staining for viral antigen in the liver of the POWV-infected mice did not indicate the presence of virus in this organ (data not shown).
Virus localization in the brains of the POWV-infected mice was significantly different from that in the brains of the OHFV-infected mice, largely reflecting histological differences between the two viruses. The distribution of neuronal immunoreactivity to viral antigen was less uniform in the OHFV-infected mice than in the POWV-infected mice. In the OHFV-infected mice, at dpi 7, the superficial cerebral cortexes had less immunoreactivity toward viral antigen than did deep structures (thalami and basal ganglia) and cerebella, where both the cortex and the deep cerebellar nucleus showed significant immunoreactivity. Also at dpi 7, focal areas in the brain showed immunoreactive cells with morphology consistent with astrocytes of the fibrillary or Bergmann type. At the late stage of infection (i.e., 58 dpi), OHFV antigen was largely isolated to the cerebellum, where Purkinje cells appeared to be the primary target, although antigen was also seen in neurons in the deep cerebellar nucleus of the cerebellum (figure 5). These findings contrast significantly from those in the POWV-infected mice, in which the staining of brain sections at 7 dpi showed extensive immunoreactivity in neurons of both the cerebral cortex and the deep cerebellar nucleus (figure 6), as well as more-focal patchy reactivity in the granular and pyramidal cell layers of the hippocampus (figure 6) and in the Purkinje and granule cells of the cerebellum. Generally, neuronal immunoreactivity was much more prevalent than were actual lesions, and most of the immunoreactive neurons were morphologically normal. No immunoreactive cells with glial morphology were identified in the POWV-infected mice.
DISCUSSION
The development of an animal model for OHFV infection offers the unique opportunity to compare very distinct yet genetically closely related arboviruses. OHFV and POWV are members of the TBE serocomplex of flaviviruses [14], yet OHFV causes a human disease that rarely results in neurological manifestations, whereas POWV infection in humans causes severe encephalitis and neurological sequelae in survivors [8]. In 35 week-old female BALB/c mice, the disease caused by OHFV infection was distinct from that caused by POWV infection. Although, at the doses used in the present study, inoculation ip with either POWV or OHFV was uniformly lethal, death in all of the POWV-infected mice occurred after paresis and/or paralysis, whereas OHFV caused a very different disease in mice. Compared with the POWV-infected mice, the OHFV-infected mice had no indication of neurological problems, had conjunctival suffusion that has also been reported in human cases, and had very distinct histological properties. The OHFV-infected mice had significantly enlarged spleens and few indications of viscerotropic disease in either the kidney or the liver, although there was some indication that endothelial cells within the liver were infected with the virus. The lack of viral antigen in the kidneys of BALB/c mice indicates that viral replication is not occurring in this organ and, in this particular animal model, reduces the possibility of transmission via urine.
The most significant histological differences between the OHFV-infected mice and the POWV-infected mice occurred in the spleen and the brain. The spleens of the POWV-infected mice appeared to be very reactive, suggesting early onset of robust proliferation of lymphocytes in T-cell areas of the spleens; it is not known whether this reaction correlates with cell-mediated immune responses against the virus. Spleens of the OHFV-infected mice were enlarged, probably because of red-pulp congestion rather than white-pulp cellular proliferation, which was less striking than that in the POWV-infected mice. During the later stages of infection, there was prominent white-pulp necrosis in the OHFV-infected mice, which may either reflect a direct viral cytopathogenic effect or be the result of a brisk immune response. In either case, splenic morphology at 68 dpi suggested ongoing cellular activation and an immune response in POWV-infected mice and decreased proliferation of lymphocytes in OHFV-infected mice, perhaps adversely affecting the immune response. This significant difference in splenic response to infection suggests that the immune response to OHFV infection differs from the immune response of POWV infection, a difference that may be manifested through variation in either cell tropism or cytokine response to infection in peripheral cells, with OHFV infection activating a more dramatic inflammatory response.
Compared with the POWV-infected mice, the OHFV-infected mice showed no behavioral indications of having an encephalitic disease. However, histological examination of the OHFV-infected mice found meningoencephalitis, severe pathology, and viral antigen in their cerebella; the pathology in these cerebella would suggest severe neurological dysfunction, although overt neurological problems (e.g., paralysis) were not apparent. In contrast, the POWV-infected mice had both a more severe meningoencephalitis and focal necrosis, albeit primarily in the cerebrum. There are a few possible explanations for these dramatic differences in virus localization and pathology. The most obvious explanation is a specific cell tropism that targets the virus to certain cell types within the brain. The localized accumulation of OHFV antigen in the cerebellum suggests that there are specific cells in this region of the brain that are particularly susceptible to infection. This tropism may be due to such factors as cellular receptors or virus-replication efficiency in particular cell types. The widespread distribution of POWV antigen throughout the cerebra of POWV-infected mice indicates that the cell-susceptibility pattern in the latter is different from that in OHFV-infected mice. Additionally, the finding that POWV-infected cells were morphologically normal suggests that necrosis in the brain is an indirect result of infectionand that it is perhaps due to inflammation and cellular cytokine responses.
The present study has described a mouse model for the tickborne flavivirus OHFV and has used this model to compare infection by this viscerotropic virus versus infection by the neurotropic POWV, demonstrating that the pathologies caused by infection with these 2 viruses are different. Although the study has examined only 1 strain of each virus, both the genetic similarity of virus strains [15, 16] and the consistency of disease in humans suggest that minimal differences in disease would be found in the mouse model. However, studies examining this question are planned. Additionally, the present study used a large dose of virus, with the objective of examining the course of a lethal infection. Given the limited availability of data regarding human infection with either OHFV or POWV, it is difficult to evaluate either the relationship between the dose given to mice and the infectious dose for humans or the relationship between histopathology in mice and that in humans. Although perhaps not the ideal model, the mouse model demonstrates that the differences between the infections caused by these viruses, although subtle, are distinct and may define the progression and resolution of disease. The response to infection with sublethal doses of virus may be key to an understanding of both the progression of disease and the role that immune response plays in the regulation of the pathogenesis of disease. Studies to examine these questions are currently underway.
Acknowledgments
We thank Friederike Feldmann and Daryl Dick, for support in BSL-4 training and animal care; Drs. Thomas Ksiazek and Pierre Rollin, for providing OHFV; and Dr. Robert Tesh, for providing POWV.
References
1. Kharitonova NN, Leonov YA. Omsk hemorrhagic fever [in Russian]. New Delhi: Amerind Publishing, 1985. First citation in article
2. Burke DS, Monath TP. Flaviviruses. In: Knipe DM, Howley PM, eds. Field's virology. 4 ed. Philadelphia: Lippincott Williams & Wilkins, 2001:1043125. First citation in article
3. Gritsun TS, Lashkevich VA, Gould EA. Tick-borne encephalitis. Antiviral Res 2003; 57:12946. First citation in article
4. Busygin FF. Omsk hemorrhagic fevercurrent status of the problem [in Russian]. Vopr Virusol 2000; 45:49. First citation in article
5. McLean DM, Donohue WL. Powassan virus: isolation of virus from a fatal case of encephalitis. Can Med Assoc J 1959; 80:70811. First citation in article
6. Thomas LA, Kennedy RC, Elkund CM. Isolation of a virus closely related to Powassan virus from Dermacentor andersonii collected along north Cache la Poudre river, CO. Proc Soc Exp Biol Med 1960; 104:3559. First citation in article
7. Artsob H, Spence L, Th'ng C, et al. Arbovirus infections in several Ontario mammals, 19751980. Can J Vet Res 1986; 50:426. First citation in article
8. Gholam BI, Puksa S, Provias JP. Powassan encephalitis: a case report with neuropathology and literature review. CMAJ 1999; 161:141922. First citation in article
9. Tesh RB, Guzman H, da Rosa AP, et al. Experimental yellow fever virus infection in the golden hamster (Mesocricetus auratus). I. Virologic, biochemical, and immunologic studies. J Infect Dis 2001; 183:14316. First citation in article
10. Xiao SY, Zhang H, Guzman H, Tesh RB. Experimental yellow fever virus infection in the golden hamster (Mesocricetus auratus). II. Pathology. J Infect Dis 2001; 183:143744. First citation in article
11. McArthur MA, Suderman MT, Mutebi JP, Xiao SY, Barrett AD. Molecular characterization of a hamster viscerotropic strain of yellow fever virus. J Virol 2003; 77:14628. First citation in article
12. Holbrook MR, Shope RE, Barrett A. Use of recombinant E protein domain IIIbased enzyme-linked immunosorbent assays for differentiation of tick-borne encephalitis serocomplex flaviviruses from mosquito-borne flaviviruses. J Clin Microbiol 2004; 42:140110. First citation in article
13. Bhardwaj S, Holbrook MR, Shope RE, Barrett ADT, Watowich SJ. Biophysical characterization and vector-specific antagonist activity of domain III of the tick-borne flavivirus envelope protein. J Virol 2001; 75:40027. First citation in article
14. Lin D, Li L, Dick D, et al. Analysis of the complete genome of the tick-borne flavivirus Omsk hemorrhagic fever virus. Virology 2003; 313:8190. First citation in article
15. Ebel GD, Spielman A, Telford SR 3rd. Phylogeny of North American Powassan virus. J Gen Virol 2001; 82:165765. First citation in article
16. Li L, Rollin PE, Nichol ST, Shope RE, Barrett AD, Holbrook MR. Molecular determinants of antigenicity of two subtypes of the tick-borne flavivirus Omsk haemorrhagic fever virus. J Gen Virol 2004; 85:161924. First citation in article