Persistent West Nile Virus Infection in the Golden Hamster: Studies on Its Mechanism and Possible Implications for Other Flavivirus Infections

Abstract

Golden hamsters (Mesocricetus auratus) experimentally infected with West Nile virus (WNV) developed chronic renal infection and persistent shedding of virus in urine for up to 8 months, despite initial rapid clearance of virus from blood and the timely appearance of high levels of specific neutralizing antibodies. Infectious WNV could be recovered by direct culture of their urine and by cocultivation of kidney tissue for up to 247 days after initial infection. Only moderate histopathologic changes were observed in the kidneys or brain of the chronically infected hamsters, although WNV antigen was readily detected by immunohistochemistry within epithelium, interstitial cells, and macrophages in the distal renal tubules. Comparison of WNV isolates from serial urine samples from individual hamsters over several months indicated that the virus underwent both genetic and phenotypic changes during persistent infection. These findings are similar to previous reports of persistent infection with tickborne encephalitis and Modoc viruses

During recent studies of the pathogenesis of West Nile virus (WNV) infection in hamsters [1, 2], it was observed that some hamsters developed chronic viruria, despite the rapid appearance of high levels of specific neutralizing antibodies in serum and our inability to culture WNV directly from their tissues or blood. To investigate this apparent paradox, further experiments were performed to elucidate the mechanism by which WNV infection can persist in immunocompetent, healthy hamsters. The present article reports our findings, which indicate that some hamsters develop persistent renal infection and continue to shed infectious WNV in their urine for up to 8 months after experimental infection

Materials and Methods

HamstersHamsters used in the study were 6–8-week-old female golden hamsters (Mesocricetus auratus) obtained from Harlan Sprague Dawley. Hamsters were housed 3 or 4/cage and were cared for in accordance with the Guidelines of the Committee on Care and Use of Laboratory Animals (Institute of Laboratory Animal Resources, National Research Council, Washington, DC) under an animal use protocol approved by the University of Texas Medical Branch. All work was performed in biosafety level 3 facilities

VirusThe WNV strain used in all of our hamster studies was NY385-99, which was originally isolated from a snowy owl that died at the Bronx Zoo during the 1999 epizootic in New York City [1]. The virus used to initiate the present experiments had been passaged 3 times in Vero cells with 2 additional passages in hamsters. Sixty hamsters were inoculated intraperitoneally (ip) with ∼10^2.8 pfu of the virus stock

Experimental designAfter inoculation with WNV, blood samples (100 μL) were obtained daily from 6 of the hamsters (numbers 8533–8538), for 7 consecutive days, to determine the level and duration of viremia. On day 18 after infection, another blood specimen was obtained, to determine the antibody response. Beginning on day 26 after infection and continuing at approximately weekly intervals for 8 months, 1 or 2 hamsters in the group of 60 were killed and tested, as described below

To assay for viruria, fresh urine was also collected at 2–3-week intervals from hamsters 8533–8538 and less frequently from other individuals in the group. Urine was mixed with sufficient PBS (pH 7.4) that contained 30% fetal bovine serum (FBS; PBS diluent) to prepare a 1:10 dilution. The diluted fresh urine (200 μL) was inoculated into flask cultures of Vero cells for virus assay. Some of the WNV-positive urine samples were also titrated by plaque assay, to quantitate the amount of virus present

Direct cultureHamsters to be assayed for WNV were anesthetized with Halothane (Halocarbon Laboratories) and exsanguinated by cardiac puncture to obtain a serum sample. The skin was rinsed with 70% ethanol, and the abdominal and thoracic cavities were opened with sterile scissors and forceps. Any urine present in the urinary bladder was aspirated for culture. The lungs, liver, spleen, urinary bladder, kidneys, and adrenal glands were removed and placed in sterile plastic Petri dishes. The entire brain was also removed and divided longitudinally into halves. A portion of the liver and spleen, half of the brain, and 1 kidney and lung were placed in individual sterile 50-mL tubes with ∼40 mL of sterile buffered saline (pH 7.4) that contained penicillin (100 U/mL) and streptomycin (0.1 mg/mL). These tissue samples were used for culture. A duplicate set of the same tissues, plus the urinary bladder and 1 adrenal gland from each hamster, were placed in 10% buffered formalin solution for histologic examination

By use of individual sterile scissors and forceps, the tissue samples were minced into small pieces. The minced tissues were homogenized in sterile 7-mL Ten Broeck glass tissue grinders that contained enough PBS diluent to make 10% (wt:vol) suspensions. After clarification by centrifugation (6000 g), 200 μL of each supernatant was inoculated into a flask culture of Vero cells. Cultures of tissue homogenates and urine were held at 37°C and observed daily for viral cytopathic effect (CPE). If CPE was observed, a sample (125 μL) of the culture fluid was tested by use of the VecTest WNV antigen assay kit (Medical Analysis Systems), according to the manufacturer’s instructions

CocultivationTissue samples from the experimentally infected hamsters were also assayed for WNV by cocultivation. The cocultivation technique used was a modification of methods described elsewhere [3, 4]. Tissue fragments were minced with sterile scissors and incubated at 37°C for 15–60 min in 50-mL conical tubes with a trypsin-EDTA solution, to dissociate the cells. After several washes, the cells were resuspended in 3–5 mL of medium; the entire mixture was inoculated into 2 culture flasks that contained monolayer cultures of Vero cells. Cocultures were maintained at 37°C for 10–14 days and were examined for evidence of viral CPE. If CPE was observed, the presence of WNV in the culture medium was confirmed by use of the VecTest West Nile antigen assay. The cocultivation medium used for brain cells was Dulbecco’s modified Eagle medium and F-12 nutrient medium (Gibco) with 10% FBS and gentamicin (100 μg/mL). The cocultivation medium for other tissues was Medium 199 with Earle’s salts, l-glutamine, NaHCO₃ (Gibco), 10% FBS, and penicillin-streptomycin

RNA extractionTissue fragments and urine from selected hamsters were also put into RNAlater (Ambion) to assay for WNV RNA. For RNA extraction, a Trizol/chloroform procedure was used [5]. Approximately 100 mg of tissue was homogenized at room temperature for 5 min in 750 μL of Trizol (Invitrogen) in a 1.5-mL Eppendorff tube. RNA was extracted by use of chloroform and isopropyl alcohol; the resulting RNA pellet was dissolved in 30 μL of RNase-free water

Reverse-transcription polymerase chain reaction (RT-PCR)A primer pair was designed to amplify a 570-bp segment of the WNV genome: forward (WNV 4137), 5′-CTTGGCTCTAGCCTCAACAGG-3′; reverse (WNV 4706), 5′-GCTCCTGCTTGATAACTGC-3′. For first-strand cDNA synthesis, the SuperScript II First-Strand Synthesis System (Invitrogen) was used, with the reverse primer, in a total volume of 20 μL, according to the manufacturer’s protocol. Each reaction contained 50 U of SuperScript II reverse transcriptase. One microliter of the cDNA was included in the PCR mixture that contained 2 mmol/L MgCl₂, in a total volume of 25 μL. After heating at 94°C for 5 min, the cycle was repeated for 35 times at 94°C for 1 min, 58°C for 1 min, and 72°C for 2 min, with an additional extension at 72°C for 10 min at the end of the final cycle. DNA was separated and visualized with electrophoresis in 1.5% agarose gel

Antibody determinationsAntibodies to WNV in serum from infected hamsters were measured by hemagglutination inhibition (HI), IgM-capture ELISA (IgM ELISA), and plaque reduction neutralization (PRN) tests, as described elsewhere [1, 6, 7]. Hamster serum samples were tested by HI by use of serial 2-fold dilutions of 1:20–1:5120. In the IgM ELISA, serum samples were screened at a single 1:40 dilution; results were recorded as optical density at 405 nm. Absorbance values ⩾0.20 OD₄₀₅ were considered to be positive. For the PRN test, serum samples were tested at 2-fold dilutions of 1:20–1:40,960; dilutions that produced ⩾90% plaque reduction were considered to be positive

Histologic and immunohistochemical examinationsAfter formalin fixation for 24 h, the tissue samples (lung, liver, spleen, kidney, adrenal gland, urinary bladder, and brain) were transferred to 70% ethanol for storage and subsequent embedding in paraffin [1, 8]. Immunohistochemical (IHC) straining for WNV antigen was performed as described elsewhere [1, 8]. WNV-infected mouse immune ascitic fluid was used as the primary antibody at a dilution of 1:100; it was directly labeled and detected by use of a commercially available ISO-IHC AEC kit (Inno-Genex). As negative controls, tissue from hamsters experimentally infected with Modoc virus (Flavivirus) were stained with anti–WNV antibody, and hamster tissues from the present study were stained with an anti–Pirital virus (Arenavirus) mouse antibody, by use of the same IHC methods

Results

Pattern of viremia and antibody responseTable 1 shows the pattern of viremia in 6 hamsters, after inoculation with the WNV stock. The level (highest mean titer, 10^4.7 pfu/mL) and duration (4 days) of viremia were similar to those observed in hamsters infected with the parent NY 385–99 strain of WNV by the ip route and by mosquito bite [1, 9]. None of the 60 hamsters in this experiment became ill or died; the hamsters appeared well, remained active, and gained weight throughout the duration of the study

Table 1

Pattern of viremia in hamsters inoculated intraperitoneally with West Nile virus (102.9 pfu) from a kidney homogenate of an infected hamster

WNV was not detected in the hamsters’ blood after day 6 after infection (table 1), and IgM and early HI antibodies were present in their serum when tested on day 7 after infection (table 2). By day 18 after infection, the IgM antibody reading (OD₄₀₅) was lower, and the HI titer was higher. By day 95 after infection, the IgM ELISA was negative, but high levels of HI and neutralizing antibodies persisted for the duration of the experiment (8 months). A similar antibody response was observed in hamsters after infection with the parent NY385-99 strain [6, 10]. In summary, the pattern of viremia in the hamsters after WNV infection and their subsequent antibody response conformed to the expected host responses (paradigm) for WNV and other arthropodborne flavivirus infections [11]

Table 2

Results of West Nile virus antibody determinations (IgM ELISA, hemagglutination-inhibition test [HI], and plaque reduction neutralization test [PRNT]) done on serum collected from hamsters at various times after infection

Persistent shedding of WNV in urineBeginning on day 26 after infection, urine was collected intermittently from the 60 hamsters and was cultured for WNV. During the first 2 or 3 months, almost all of the hamsters had infectious WNV in their urine; but, over the next 5 or 6 months, 50%–60% of the hamsters apparently cleared their infection—WNV could no longer be detected by direct culture of their urine or cocultivation of their kidneys

Table 3 illustrates the culture results on serial urine samples collected from 5 representative hamsters (8533, 8534, 8535, 8537, and 8538) during an 8-month period. Hamsters 8533 and 8534 shed WNV virus in their urine for ∼4 months (for 137 and 136 days, respectively); their subsequent urine cultures were all negative, as was cocultivation of their kidneys on days 201 and 214 after infection. In contrast, hamsters 8535, 8537, and 8538 continued to shed infectious WNV in their urine until the experiment was terminated after 8 months. The amount (titer) of infectious virus in their urine remained fairly constant, according to the results of plaque assay

Table 3

Chronic viruria in hamsters after experimental infection with West Nile virus (WNV)

Chronic renal infectionTable 4 shows results of urine and organ cultures performed on 19 persistently infected hamsters at the time of death. Urine was assayed for virus by direct culture in Vero cells; brain, liver, spleen, lung, and kidney were tested by both direct culture and by cocultivation. Although 18 of 19 urine samples yielded WNV on direct culture, virus was rarely isolated from tissue samples tested by this method after day 21 after infection. By direct culture, WNV was isolated only from spleen, lung, and kidney of hamster 8536 on day 26; from lung of hamster 2001 on day 54; and from kidneys of hamsters 2087, 8537, and 8538 on days 223, 244, and 247 (data not shown). In contrast, 16 of 19 hamsters yielded WNV from ⩾1 of their organs by cocultivation (table 4). The most common tissue source of virus was the kidney

Table 4

Isolation of infectious West Nile virus from selected tissues and urine of chronically infected hamsters

The serum neutralizing antibody titer for each of the hamsters at the time of death is also shown in table 4. All of the hamsters tested had high levels of WNV neutralizing antibodies when they were killed. This probably explains why the direct cultures rarely yielded virus, whereas cocultivation of the same tissues did. Homogenization of tissue releases intracellular virus, but it also allows the virus to come into contact with antibodies that may be present in blood and interstitial fluids in the tissue. The washing and digestion of the minced tissue that occurs during treatment with trypsin-EDTA presumably eliminates most or all of the neutralizing antibodies, without destroying intracellular virus, which is subsequently released and amplified on cocultivation with the Vero cell monolayer

Histologic findingsNo significant microscopic abnormalities were consistently present in the lungs, liver, adrenal glands, spleen, or urinary bladder. The liver occasionally showed nonspecific lymphocytic infiltration of the portal tracts and multiple microgranulomas in the lobular parenchyma. A few granulomas with multinucleated giant cells were also observed. However, these were rare and were not consistent findings. For this reason, and the fact that amyloidosis and spontaneous hepatic cirrhosis are relatively common findings in older hamsters [12], the observed liver changes in the hamsters were interpreted as probably being unrelated to WNV infection

The brains were also unremarkable, except for occasional psammoma bodies (small foci of calcifications) in the basal nuclear region or in the brainstem. The occasional psammoma bodies in the brain may have been the result of prior death of neurons due to a possible viral effect, or they might have represented focal cerebral mineralization, which also occurs in older hamsters [12]

No significant pathologic changes were observed in kidneys of hamsters killed during the first few months after WNV infection. Later, however, scattered foci of light blue–colored amorphous deposits were observed in the interstitial area between the tubules in the renal papillae. Some were calcified and formed concentric laminated configurations (figure 1A); these most likely represented absorbed proteins from the filtrate in the tubular lumen. Starting at week ∼20 after infection and thereafter, the hamsters began to show dilatation of the renal tubules, with atrophy and flattening of the tubular epithelia, mostly in the cortical regions, with clustering (figure 1B). Although the tubular dilatation and atrophy seen in the kidneys of our hamsters appeared to be associated with chronic WNV infection, there was some uncertainty, given that similar changes (progressive glomerulonephropathy and tubular atrophy) have been reported to occur spontaneously in older hamsters, especially females [12, 13]

Figure 1

Photomicrographs of kidneys from hamsters chronically infected with West Nile virus (WNV). A Proteinaceous deposits in the interstitium of the renal papillae (light blue) (magnification, ×100). B Dilatation and atrophy of the renal tubules in clusters, in the cortex (magnification, ×100). C WNV antigen–positive cells in the peritubular regions of kidneys (immunohistochemical stain; magnification, ×200). D Higher magnification showing cells positive for WNV antigens to be mostly macrophages and, rarely, vascular endothelia (immunohistochemical stain; magnification, ×400)

Immunohistochemically, the liver, spleen, lungs, urinary bladder, and adrenal glands were all negative for specific WNV antigens. Similarly, no residual antigen was found in the brain. In contrast, moderate to strong positive staining was observed focally in kidney tissues of many hamsters. During the first few months, IHC-positive staining was mostly in the interstitium, mainly in macrophages and rarely in vascular endothelial cells (figure 1C and 1D). Later (⩾20 weeks after infection), the antigen positivity was limited mainly to the interstitial blue proteinaceous or calcified deposits. No IHC staining was observed in the negative control hamsters

RT-PCRUrine and tissues from some infected and control hamsters were also subjected to RT-PCR for detection of WNV genomes. The objective was to compare the sensitivity of RT-PCR with direct culture and cocultivation for detecting WNV. A second goal was to determine whether WNV RNA could be detected in tissues that were culture negative. Tissues and urine from noninfected hamsters were included as controls. Table 5 summarizes the results with brain, liver, spleen, lung, kidney, and urine samples from 7 chronically infected hamsters. WNV was isolated from urine by direct culture from all of the hamsters (the criterion for selection). By cocultivation and RT-PCR, WNV or viral RNA was detected in the kidneys from most of these same hamsters. However, the results of cocultivation and RT-PCR were not always in agreement, and neither of the latter 2 assay systems appeared to be more sensitive than direct culture of urine in identifying WNV-infected hamsters (tables 4 and 5). The RT-PCR results confirmed the cocultivation and IHC results, which indicated that persistent WNV infection in the hamsters was confined largely to the kidneys

Table 5

Comparison of results obtained by direct culture, cocultivation, and reverse-transcription polymerase chain reaction (RT-PCR) of tissue and urine samples from hamsters persistently infected with West Nile virus

Phenotypic and genotypic changes in WNV accompanying persistent infectionThe WNV isolates from urine from persistently infected hamsters showed both phenotypic and genotypic differences over time. Plaque morphology of the isolates changed, as did their virulence for hamsters. In addition, isolates from serial urine samples from the same hamsters showed significant and persistent point mutations, which may be associated with renal tropism or persistence. Characterization of these phenotypic and genotypic changes will be the subject of another report now in preparation (S.-Y.X. and R.B.T, unpublished data)

Discussion

Results of previous experiments [1, 2, 9, 10] showed that hamsters infected with WNV develop a brief viremia, lasting ∼5 days, which is followed by rapid development of IgM and IgG antibodies. If the hamsters survive infection, IgM antibodies wane, but IgG antibodies (as measured by HI and PRN tests) persist at high levels for many months or possibly the lifetime of the hamsters. However, many of the convalescent hamsters have a persistent infection and continue to shed infectious WNV in their urine for variable periods of time (months), despite a vigorous humoral antibody response. WNV infection in these hamsters appears to be localized to tubular epithelial cells and interstitia in the papillae of kidneys. The persistently infected hamsters do not manifest overt illness, and only moderate histologic changes (proteinaceous deposits in the interstitium between the infected tubules) can be seen. The chronically infected hamsters do not have detectable proteinuria, and the urine pH (8.0–8.5) is similar to that in normal hamsters with the same age and diet (R.B.T., unpublished data)

One of the obvious questions about the results of our study is whether persistent WNV infection in the hamster is simply a laboratory artifact or whether it is a more general phenomenon indicative of the behavior of other flaviviruses in their vertebrate hosts. Review of the literature suggests that other flaviviruses also produce a chronic infection in mammalian hosts. The ability of some members of the genus Pestivirus (classical swine fever and bovine viral diarrhea virus 1) and of the genus Hepacivirus (hepatitis C viruses) of the family Flaviviridae to produce persistent or chronic infection in their vertebrate hosts is well recognized [14, 15]. Less well known is evidence indicating that some members of the genus Flavivirus can also produce persistent infection, as is illustrated in the present experiments. Two other examples of flaviviruses that cause persistent infection are Modoc virus (MODV) and tickborne encephalitis virus (TBEV)

MODV naturally infects deer mice (Peromyscus maniculatus) in the western United States and Canada. Available information suggests that MODV is not arthropodborne but is maintained in deer mouse populations by horizontal and possibly vertical transmission [16–18 ]. Laboratory studies [16–18 ] have demonstrated that both deer mice and hamsters develop persistent MODV infection with chronic viruria. Davis and Hardy [17] showed that hamsters experimentally infected with MODV developed a brief viremia lasting 2–6 days, followed by the development of HI and neutralizing antibodies by day 7 after infection. Despite a brisk antibody response, MODV was chronically shed in the hamsters’ urine for at least 12 weeks after infection. During this chronic phase, virus could not be isolated directly from cultures of organ homogenates; but, by cocultivation of the tissues, it was possible to recover MODV from the kidneys, spleens, and lungs of the experimentally infected hamsters for up to 32 weeks after infection [17]. Similar results were obtained with naturally and experimentally infected deer mice [16, 18]. These results are analogous to our findings in the WNV-infected hamsters, although pathologic or IHC studies were not done on the MODV-infected hamsters

Similar results were reported by Russian investigators who studied persistent WNV and TBEV infection in the rhesus macaque (Macaca mulatta) [19–20]. Pogodina et al. [19] reported that WNV could be isolated by cocultivation from the brain, lymph nodes, and kidneys of experimentally infected rhesus macaques for at least 5.5 months after infection

In a series of articles published between 1981 and 1984, Pogodina et al. [4, 20–22 ] reported results of experiments done in rhesus macaques infected with the Far Eastern subtype of TBEV. The authors reported that TBEV could be isolated from organs (brain, liver, spleen, lymph nodes, and kidneys) of rhesus macaques inoculated subcutaneously or intracerebrally with the virus, for up to 2 years after infection. TBEV could be recovered from the tissues only by cocultivation and not by direct culture of organ homogenates [4]. Furthermore, virus recovered from tissues of the persistently infected rhesus macaques lost its pathogenicity for mice and, in some cases, the ability to produce CPE in vertebrate cell cultures [21]. As with MODV and WNV, TBEV persisted in some tissues despite high levels of specific neutralizing antibodies in the rhesus macaques’ convalescent serum samples [22]

Less-well-documented reports also have been published suggesting persistent infection of bat salivary glands with Rio Bravo virus [23], chronic renal infection and viruria with Omsk hemorrhagic fever virus in muskrats [24], and persistent Japanese encephalitis virus infection in mice [25] and humans [26]. In addition, we have recently demonstrated that hamsters experimentally infected (subcutaneously) with St. Louis encephalitis virus also shed the virus in their urine for at least 4 months after infection (R.B.T. and M.S., unpublished data)

Collectively, these results suggest that persistent infection is not an uncommon occurrence with viruses in the genus Flavivirus and that this phenomenon may play an important role in their natural history (maintenance). Its clinical and epidemiologic significance for humans remains to be determined; but certainly, further in-depth investigation of this aspect of flavivirus ecology seems to be warranted

Acknowledgments

We thank Patrick Newman, for technical assistance in preparing the histologic sections; and Dora Salinas, for assistance in preparing the manuscript

Presented in part: Fall meeting of the Texas Branch of the American Society for Microbiology, Houston, 4–6 November 2004

Financial support: National Institutes of Health (contracts NO1-AI25489 and NO1-AI30027)