Three subspecies of Sigmodon hispidus (cotton rat) were used in this study: the Harlan colony, the Texas population, and the Florida population. The Harlan colony consisted of 6- to 8-week-old female rats purchased from Harlan (Indianapolis, IN, USA) from an established colony. Because the exact geographic origin was unknown, cytochrome B mitochondrial gene sequences were amplified by PCR and compared phylogenetically with those from cotton rats from various locations in North America (27). The sequences from the Harlan colony grouped with those of animals collected in east Texas, Louisiana, and Tennessee but were outside of the clade from southern Florida (data not shown), which indicated that these rats originated from a nonenzootic region. (Florida is the only VEE complex alphavirus–enzootic region in the United States, aside from the Rocky Mountains, which are outside of Sigmodon distribution [8,23]). The Texas population consisted of 4- to 12-month-old wild-caught male and female S. hispidus berlandieri trapped in Galveston Island State Park, Texas (29.27°N, 94.83°W) (25). The Florida population consisted of 3- to 21-week-old male and female cotton rats (S. hispidus spadicipygus) trapped in southern Florida (25) and used to rear F1 rats for experimental studies. Before inoculation, all rats were tested for antibodies against VEEV, EVEV, and Eastern equine encephalitis virus (EEEV) and acclimated for 3 days in a Biosafety Level 3 animal facility. All experiments included >2 animals as negative controls. All studies were approved by the University of Texas Medical Branch Animal Care and Use Committee.

VEEV strain Co97-0054 (enzootic ID subtype), isolated in Colombia in 1997 from a sentinel hamster, and strain 68U201 (enzootic IE), isolated in Guatemala from a sentinel hamster and derived from a cDNA clone (28), were used for experimental infections. These strains were selected because they had low passage histories and represent the 2 major enzootic VEEV subtypes; strain Co97-0054 has also been used for experimental infection of spiny rats (22). Cotton rats were inoculated subcutaneously in the left footpad or left thigh with 3–4 log₁₀ PFU of virus, a dose consistent with the maximum amount of VEEV in mosquito saliva (29). To determine whether an increase in virus inoculum could change the outcome of the disease, 4 Florida rats were also inoculated with 5–6 log₁₀ PFU of VEEV strain 68U201. All animals were observed for signs of illness once a day for 15 days.

To determine the neurovirulence of VEEV in the Florida population, 3 rats (2 months of age) were inoculated intracranially with 3 log₁₀ PFU/mL of strain 68U201. Blood samples were collected 24 h after inoculation, and the rats were observed for signs of disease.

Blood samples were collected from the retroorbital sinus for <10 days after inoculation. Serum samples were diluted 1:10 in Eagle minimal essential medium supplemented with 20% fetal bovine serum, gentamicin, and L-glutamine and stored at –80°C. Viremia titers were determined by plaque assay with Vero cells (30).

Plaque reduction (80%) neutralization (PRNT) and hemagglutination inhibition (HI) tests were performed (30). To detect specific VEEV immunoglobulin M (IgM) antibodies, an IgM-capture ELISA was performed (31). Briefly, microplates were coated with anti-rat IgM, diluted 1:500 in carbonate-bicarbonate buffer pH 9.6 (Kirkegaard and Perry Laboratories, Gaithersburg, MD, USA), and incubated at 4ºC for at least 16 h. Subsequently, the following were sequentially added: test serum, mouse immune ascitic fluid prepared against VEEV antigens, anti-mouse conjugate (Kirkegaard and Perry), and ABTS (2,2′-azino-bis[3-ethylbenzthiazoline-6-sulphonic acid])–peroxidase substrate (Kirkegaard and Perry). Test serum samples were diluted at 1:40 in 0.5% bovine serum albumin in phosphate-buffered saline at pH 7.4, and the plates were read by using a spectrophotometer with a 405-nm wavelength filter. The cut-off value was calculated as the mean optical density (OD)_{405 nm} of negative control samples plus 3 standard deviations, or 0.200. Linear regression, the Student t test, and analysis of variance were used to analyze data.

Inoculation of the Florida rats with 3 log₁₀ PFU of VEEV strain 68U201 (IE) or Co97-0054 (ID) resulted in no detectable signs of illness and survival rates of 100% and 87.5%, respectively (Figure 1). One Florida rat inoculated with strain Co97-0054 died on day 10 postinoculation without exhibiting any signs of illness. These findings contrasted with the results of VEEV infections of the Texas and Harlan populations. Although the Harlan rats were inoculated with only the subtype ID strain, signs of severe illness developed in all of the Harlan and Texas rats beginning on day 5. Signs included ruffled coats, lack of grooming, lethargy, and for many, signs of encephalitis (incoordination and instability when walking and erratic movements of the head and limbs), dehydration (measured by lack of skin turgor), and anorexia. Most animals that died before day 5 postinoculation showed no prior signs of illness. The average survival time for the Texas population was 6.8 days after inoculation with the subtype ID strain and 8.2 days with the IE strain; for the Harlan colony, it was 5 days after inoculation with the subtype ID VEEV. None of the animals that survived past day 15 died. The 2 sham-inoculated and the 2 noninoculated rats survived without signs of disease.

To determine whether the absence of disease in the Florida population was due to the inability of the virus to penetrate the central nervous system, 3 rats were inoculated intracranially with 3 log₁₀ PFU/mL of subtype IE VEEV. Viremia titers at 24 h postinoculation were 6.3, 6.2, and 6.8 log₁₀ PFU/mL (mean 6.5). By day 3 postinoculation, all of these rats started showing signs of illness, including ruffling of the fur and lack of movement. By day 9 postinoculation, 1 rat was dead and the other 2 exhibited instability and difficulty in walking, uncoordinated and erratic movements of the head and limbs, dehydration, and anorexia; these animals were euthanized because of the severity of disease. Histopathologic examination of the brains showed clear signs of encephalitis, focal meningoencephalitis (Figure 2, panels A, B) and associated perivascular mononuclear cell infiltration (Figure 2, panel C), and neurophagia, which led to the conclusion that the cause of death was from the viral infection and not from the injection or manipulation of the animals.

To determine whether the Florida population’s apparent resistance to VEE after peripheral infection was dose-dependent, 8 additional animals were inoculated with 5 or 6 log₁₀ PFU (4 animals per dose). In each population, 3 (75%) of the rats survived infection (data not shown). One rat inoculated with 6 log₁₀ PFU of virus succumbed to disease on day 3, whereas another animal inoculated with 5 log₁₀ PFU of virus died on day 11, which suggests that the dose could have affected disease progression. None of the other inoculated animals showed any clinical signs of illness.

Viremia profiles for the Florida rats were similar after inoculation (3 log₁₀ PFU) with either subtype ID or IE VEEV (p>0.05). Peak viremia titers of 6.2 and 5.4 log₁₀ PFU/mL, respectively, occurred at 24–48 h postinoculation, then became undetectable by days 4–5 postinoculation (Figure 3). The Texas population inoculated with 3–4 log₁₀ PFU of subtype IE virus and 3 log₁₀ PFU of subtype ID VEEV had similar viremia profiles, with peak titers of 6.1 and 6.6 log₁₀ PFU/mL, respectively, at 24–48 h.

We found in the Florida population a correlation between age and peak viremia titers. Younger animals inoculated with the IE virus had higher peak titers on days 1 and 2 postinoculation than did older animals (Figure 4, Table 1). In addition, we observed a significant difference in mean viremia titers on day 2 between 3- and 8-week-old animals and on day 3 between 5- and 8-week-old animals and between 3- and 8-week-old animals (p<0.05).

No differences in viremia profile were observed between the Texas and Harlan rats, for which VEEV infection with subtype ID was generally fatal (p>0.05). Viremia in the Texas and Harlan rats peaked between 24 and 48 h postinoculation, with mean titers of ≈6.7 log₁₀ PFU/mL, and was undetectable by day 8 postinoculation. In contrast, significant differences were observed in peak viremia titers between the Florida and Texas rats inoculated with either the ID or IE VEEV subtype (p<0.05). In all cases, Texas rats produced higher titers (24 h postinoculation) and had a longer duration of viremia than the Florida rats, in which no disease was apparent (Figure 3, Tables 2, 3). Similar results occurred when Florida and Harlan rats inoculated with subtype ID were compared.

To determine whether the difference in disease outcome in Florida versus Texas rats was due to a difference in antibody responses, serum was tested by PRNT. In the Texas population inoculated with subtype IE VEEV, low titers of neutralizing antibodies (NAb) were produced by days 5–6 postinoculation (mean titer 20), and mean titers never exceeded 40 by day 10 (Figure 5, panel A); however, detectable NAb were not produced in all animals that died. In contrast, some Florida rats inoculated with the same virus strain had detectable NAb titers by day 4, and all had detectable titers by day 6. NAb titers were significantly lower in the Texas rats on days 5, 6, and 7 (p = 0.02, 0.04, 0.04, respectively).

When Texas rats were inoculated with the subtype ID VEEV strain, even lower NAb titers were produced, despite the development of higher viremia titers during the later stages, compared with those inoculated with subtype IE. Peak NAb titers occurred on day 8 postinoculation (mean 25), and 2 of the surviving animals had no detectable NAb titers (<19) late during infection (day 32). In contrast, in all Florida rats, detectable NAb developed by day 4 and peaked on day 6 (mean 57), when they were significantly higher (p = 0.004).

To determine the duration of the antibody response in the Florida population, we measured NAb titers in 2 rats for 7 months postinoculation. Titers peaked at 4 months and then gradually decreased to the detection limit by 7 months, when the experiment was terminated (Figure 5, panel B).

Because NAb were not detected in all infected animals, serum samples were further tested by HI and IgM ELISA. All rats had detectable HI antibodies by days 5–6 postinoculation. Although the titers were relatively low (<200) on day 6, titers increased constantly over the 10 days tested; no differences in titers were noted between VEEV subtype ID or IE infections in the Texas population. As was found by PRNT, HI antibody titers were higher for the Florida population than the Texas population, and titers were higher in animals inoculated with the ID than with the IE VEEV strain (Figure 5, panel C).

Because of volume limitations of daily blood collection, IgM titrations were performed only on blood samples from euthanized rats (2 rats per day per group). Both Florida and Texas populations had similar IgM titers during the first 7 days postinoculation, regardless of the virus used. Mean titers ranged from OD 0.3 on day 4 to OD 0.8 by day 7 in the Texas population and 0.5 in the Florida population (Figure 5, panel D).

In summary, NAb titers were higher in Florida rats inoculated with subtypes ID and IE than in Texas rats inoculated with the same viruses. This finding suggests that these animals may have mounted a more rapid and effective immune response that protected against severe VEE infection.

S. hispidus, a main reservoir host of VEE complex alphaviruses, comprises >22 subspecies in North America alone (26) that differ by up to 5% in their cytochrome b mitochondrial DNA sequences (27). Because some but not all North American populations occur in regions enzootic for VEE complex alphaviruses (e.g., EVEV in Florida), we attempted to better understand the host-VEEV interactions by inoculating 3 different populations with enzootic VEEV strains. Cotton rats from the enzootic area of southern Florida (sympatric with EVEV) responded to VEEV infection as expected: moderate viremia titers, seroconversion by days 4–5 postinoculation coincident with viremia cessation, and few deaths and little detectable disease. This apparently commensal relationship could reflect long-term selection for cotton rat resistance to EVEV in Florida. Although EVEV is a relatively benign virus in laboratory rodents, the ancestral form of EVEV, believed to be a subtype ID VEEV strain from Panama or South America, is more virulent (32).

In contrast, rats from 2 nonenzootic locations (the Harlan and Texas populations) had dramatically different outcomes: severe disease often culminated in clinical signs of encephalitis and high mortality rates. This difference in disease and survival was more pronounced than that reported in other studies of VEEV-reservoir host interactions, some of which suggested that cotton rats die because of experimental manipulations or anesthesia rather than from a viral cause (19–21,24). Our results indicate that VEEV was the cause of death for most or all of our rats, and signs of encephalitis were consistent with those described in VEEV-infected mice or guinea pigs (8,33,34). The rats infected in previous studies had several different geographic origins (Arizona, North Carolina, Florida, and Panama). Of these, only Panama and Florida are enzootic for VEE complex alphaviruses. North Carolina is enzootic for another closely related alphavirus, EEEV, for which birds are thought to be the main reservoirs (35). These allopatric rat populations should be reexamined to further test the hypothesis that lack of exposure to VEE complex or other alphaviruses has resulted in no selection for resistance. On the basis of previous susceptibility studies, the peak viremia titers in all infected cotton rats were sufficient to infect known enzootic vector mosquitoes (36,37).

Reservoir hosts play an important role in the maintenance and spread of zoonotic viruses. They generally show little or no disease after infection with VEEV and most other zoonotic viruses, which presumably reflects long-term selection for resistance (1,2). This resistance is little studied and poorly understood, yet it might provide insight into improved treatments for arbovirus infections in humans. Our findings may also have implications for VEEV ecology, especially in the event of virus introduction into a nonenzootic region, as occurred during the 1971 Texas VEEV epizootic (38,39). During such a scenario, virus-induced deaths might deplete cotton rat populations, depending on the VEEV transmission levels and the length of the outbreak.

The differences in viremia profiles exhibited by the VEE-sympatric versus VEE-allopatric cotton rat populations could explain the different disease outcomes. Although peaks of viremia titers were similar for both subspecies, durations of viremia differed. The prolonged viremia observed in the Texas and Harlan animals may reflect a poor or delayed innate or adaptive immune response, which led to sustained viral replication and death. This could be an indirect effect of replication in lymphatic tissues leading to immunosuppression. Although peak viremia titers appeared to be age-dependent in the Florida population inoculated with the subtype IE VEEV strain, disease outcomes between age groups, which might reflect maturation of the immune system, did not differ significantly. Antibody titers in both populations of cotton rats after inoculation with either virus strain were relatively low.

These findings contrast with results of a previously published study of EVEV infection of cotton rats from the same 2 geographic regions (25). In that study, both subspecies of rats survived infection, exhibited similar peak viremia titers, and had high antibody titers within 9 days postinoculation; it was suggested that the innate immune response was involved. EVEV, enzootic in South Florida, is less virulent in laboratory rodents than in most other viruses in the VEE complex, including the subtypes we used (19,40). Presumably due to this lack of virulence, Florida and Texas strains of cotton rats tested produced protective antibodies and survived EVEV infection (25).

Our data from the same 2 rat populations but inoculated with more virulent stains of VEEV present a different picture. Although the innate immune response may be involved as well, antibody detection correlated with the disappearance of viremia. The capability of Florida cotton rats to produce antibodies against VEEV early may allow them to better control replication and survive. Antibodies against VEEV persist for at least 6 months in laboratory-infected cotton rats (25). The average lifespan of cotton rats in nature is estimated to be ≈6–8 months (41).

To identify protective mechanisms in the Florida population, additional studies focusing more on the innate immune responses of enzootic and nonenzootic cotton rat populations are needed. This could be approached by cross-breeding the Texas and the Florida rats and studying the offspring. Elucidation of protective mechanisms may be useful for developing new strategies for treating human or equine infections.