JOURNAL OF VIROLOGY, July 2008, p. 6139–6149
Copyright © 2008, American Society for Microbiology. All Rights Reserved.
Vol. 82, No. 13
Rotavirus Infection Accelerates Type 1 Diabetes in Mice with
Kate L. Graham,1†‡ Natalie Sanders,1†‡ Yan Tan,1§ Janette Allison,1,2
Thomas W. H. Kay,2and Barbara S. Coulson1*
Department of Microbiology and Immunology, The University of Melbourne, Victoria 3010, Australia,1and
St. Vincent’s Institute, Fitzroy, Victoria 3065, Australia2
Received 18 March 2008/Accepted 9 April 2008
Infection modulates type 1 diabetes, a common autoimmune disease characterized by the destruction of
insulin-producing islet ? cells in the pancreas. Childhood rotavirus infections have been associated with
exacerbations in islet autoimmunity. Nonobese diabetic (NOD) mice develop lymphocytic islet infiltration
(insulitis) and then clinical diabetes, whereas NOD8.3 TCR mice, transgenic for a T-cell receptor (TCR)
specific for an important islet autoantigen, show more rapid diabetes onset. Oral infection of infant NOD
mice with the monkey rotavirus strain RRV delays diabetes development. Here, the effect of RRV infection
on diabetes development once insulitis is established was determined. NOD and NOD8.3 TCR mice were
inoculated with RRV aged >12 and 5 weeks, respectively. Diabetes onset was significantly accelerated in
both models (P < 0.024), although RRV infection was asymptomatic and confined to the intestine. The
degree of diabetes acceleration was related to the serum antibody titer to RRV. RRV-infected NOD mice
showed a possible trend toward increased insulitis development. Infected males showed increased CD8?
T-cell proportions in islets. Levels of ?-cell major histocompatibility complex class I expression and islet
tumor necrosis factor alpha mRNA were elevated in at least one model. NOD mouse exposure to mouse
rotavirus in a natural experiment also accelerated diabetes. Thus, rotavirus infection after ?-cell auto-
immunity is established affects insulitis and exacerbates diabetes. A possible mechanism involves in-
creased exposure of ? cells to immune recognition and activation of autoreactive T cells by proinflam-
matory cytokines. The timing of infection relative to mouse age and degree of insulitis determines whether
diabetes onset is delayed, unaltered, or accelerated.
Type 1 diabetes results from an autoimmune process in
which pancreatic ? cells are selectively destroyed. An islet
lymphoid infiltrate develops that is described as “insulitis”
(49). Virus infections are proposed to play a role in type 1
diabetes development through ?-cell cytolysis or loss of self-
tolerance following pancreatic infection, bystander activation
of T cells, and molecular mimicry between ? cell autoantigens
and viral epitopes (19, 50, 53).
Rotaviruses are the major agents of severe acute gastroen-
teritis in children and have been implicated in exacerbation of
type 1 diabetes development (26). Antibody seroconversion to
rotavirus in Australian children was associated with increases
in autoantibodies to glutamic acid decarboxylase (GAD) and
insuloma-associated protein 2 tyrosine phosphatase (IA-2).
Amino acid sequence similarity between rotavirus protein VP7
and T-cell epitopes in human GAD and IA-2 led to the sug-
gestion of T-cell molecular mimicry as a possible mechanism
(26, 27). Although later studies in Finnish children did not
confirm the association between rotavirus infection and islet
autoimmunity (6, 34), increased antibody responses to dietary
bovine insulin were noted after rotavirus infection (35). Addi-
tional findings that might support links between rotavirus in-
fection and other autoimmunity-related diseases also have
been reported (33, 47, 57, 58).
The nonobese diabetic (NOD) mouse spontaneously devel-
ops a form of autoimmune diabetes similar to human type 1
diabetes (3, 46). Most mice show severe insulitis by 10 weeks of
age. By 30 weeks of age the diabetes incidence typically reaches
60 to 80% in NOD females and 10 to 20% in NOD males.
NOD diabetes mainly depends on CD4?and CD8?T cells,
and most cells in the insulitic lesion are CD4?T cells. Auto-
reactive T cells are primed in the draining pancreatic lymph
node(s) (PLN) and then migrate to the islets (20, 24, 29). Like
humans, NOD mice produce autoantibodies and T cells to
GAD and insulin. In addition, CD8?T cells directed to the
islet-specific glucose 6-phosphatase catalytic subunit-related
protein (IGRP) are an important component of islet-infiltrat-
ing T cells in prediabetic NOD mice (2, 15, 31, 41). Circulating
IGRP-reactive T-cell numbers predict diabetes in NOD mice
and new-onset patients (36, 52). Expression of the IGRP-
specific T-cell receptor (TCR) in NOD mice (NOD8.3 TCR)
led to 8.3 TCR expression on ?90% of islet-infiltrating T cells
and a high diabetes incidence with rapid onset (54, 55).
NOD8.3 TCR mice provide a simplified and rapid mouse
model of spontaneous diabetes and a useful tool to study the
role of CD8?T cells.
Murine rotaviruses and the rhesus monkey rotavirus strain
RRV induce diarrhea in infant mice and infect intestinal cells
* Corresponding author. Mailing address: Department of Microbi-
ology and Immunology, Gate 11, Royal Parade, The University of
Melbourne, Melbourne, Victoria 3010, Australia. Phone: 61 3 8344
8823. Fax: 61 3 9347 1540. E-mail: firstname.lastname@example.org.
† K.L.G. and N.S. contributed equally to this study.
‡ Present address: St. Vincent’s Institute, Fitzroy, Victoria 3065,
§ Present address: Department of Dentistry, The University of Mel-
bourne, Victoria 3010, Australia.
?Published ahead of print on 16 April 2008.
without causing disease in adults, although the doses required
differ by several logs (8, 38, 56). Oral RRV infection of infant
NOD mice causes gastroenteritis and delays diabetes onset,
whereas infection in young adult NOD mice without estab-
lished insulitis is asymptomatic and diabetes is unaffected (21).
RRV infection in infant or young adult NOD mice does not
initiate insulitis (21). Infectious RRV spreads to the pancreas
in infant NOD mice, with viral antigen localized in macro-
phages outside islets. Although RRV replicates in islets and
pancreatic cells isolated from young adult NOD mice, infec-
tious virus is not detectable in the pancreases of orally inocu-
lated mice of this age (11, 21).
The effect of RRV rotavirus infection on insulitis and dia-
betes development in older prediabetic NOD and NOD8.3
TCR mice with established insulitis was examined in the
present study. RRV is shown here to accelerate diabetes onset
and incidence in older prediabetic mice in the absence of
detectable extraintestinal spread and pancreatic infection.
NOD mice exposed to mouse rotavirus in a natural experiment
also showed increased diabetes development. These data pro-
vide the first evidence that rotavirus infection can accelerate
diabetes development in an animal model.
MATERIALS AND METHODS
Mice. NOD and BALB/c mice were obtained from the Animal Resources
Centre (Canning Vale, Western Australia, Australia). NOD8.3 TCR mice, ex-
pressing the TCR?? rearrangements of the H-2Kd-restricted, islet ?-cell-reactive
CD8?T-cell clone NY8.3 on a NOD genetic background (54), were provided by
P. Santamaria (University of Calgary, Calgary, Alberta, Canada). Mice were bred
and housed in the animal facility of the Department of Microbiology and Im-
munology at the University of Melbourne in isolators under specific-pathogen-
free conditions as described previously (21). Offspring of transgenic mice were
screened for the transgene by PCR analysis of tail-tip DNA. All procedures were
conducted in accordance with protocols approved by the Animal Ethics Com-
mittee of The University of Melbourne.
Mouse inoculation with RRV. Monkey rotavirus RRV (serotype P5B, G3)
was cultivated in MA104 cells, purified by glycerol gradient ultracentrifugation,
and titrated for infectivity as described previously (23, 25). Oral inoculation was
used to mimic the natural route of infection, as is often done in studies of
rotavirus infection in rodents (12, 18, 37). Inoculation procedures were similar to
those used previously in our laboratory for NOD mice (21). In brief, after
administration of NaHCO3solution to reduce stomach acidity, mice were inoc-
ulated by gavage with 0.2 ml of 50 mM Tris-HCl buffer (pH 7.4), containing 150
mM NaCl and 5 mM CaCl2(TSC) as a virus diluent control, 1.8 ? 106fluores-
cent cell-forming units (FCFU) of RRV in TSC, or an extract of mock-infected
MA104 cells that had been harvested by freeze-thawing and clarified by low-
speed centrifugation as a control for possible cell component contamination of
purified RRV. This cell extract was diluted to the same degree as the RRV
inoculum. In some experiments (as indicated), mice were given 107FCFU of
RRV in TSC. Female and male NOD mice were inoculated at 12 and 15 weeks
of age, respectively (unless otherwise indicated) to be confident of their advanced
insulitis but avoid infection when diabetes could occur spontaneously. Similarly,
NOD8.3 TCR mice were inoculated aged 5 weeks, 1 week before naive mice first
begin to develop diabetes. Diarrhea was defined as described previously (21), and
its presence was evaluated daily for 8 days after inoculation.
Mouse inoculation with mouse rotavirus. Five-day-old BALB/c mice housed
with their rotavirus-seronegative dams in an isolation room in the animal facility
were inoculated by oral gavage as described previously (21) with 30 ?l of TSC
(controls; n ? 42) or clarified stool homogenate (10% [wt/vol] in TSC) from a
diarrheic mouse (n ? 42). In an enzyme immunoassay (EIA) using rabbit anti-
serum to RRV as capture antibody and monoclonal antibody RVA as a detector
antibody (21), this stool extract showed a mean specific optical density at 450 nm
(OD450) ? the standard deviation of 1.196 ? 0.066, indicating that a high level
of rotavirus antigen was present. According to the previously proposed nomen-
clature scheme for mouse rotaviruses, this virus is designated as the agent of
epizootic diarrhea of infant mice (EDIM)-Melbourne, abbreviated as EM here
(8). Inoculated mice were monitored for diarrhea as described above.
Detection of rotavirus in organs and samples from rotavirus-infected mice.
For analysis of RRV-infected NOD and NOD8.3 TCR mice, the small intestine,
pancreas, liver, spleen, serum, and blood cells were obtained from each animal
at days 3 to 8 after infection (n ? 4 per day). Stools were collected (one pellet
per mouse) at days 2 (n ? 20), 3 (n ? 20), 4 (n ? 20), 5 (n ? 16), 6 (n ? 12),
7 (n ? 8), and 8 (n ? 4) after infection. The procedures for collection and
processing of these tissues and stools have been described previously (21). In-
fectious RRV was detected by culture amplification of virus, followed by assay of
rotavirus antigen in cultures by capture EIA, and stools and pancreases were
examined for rotavirus antigen by EIA prior to culture, as described previously
The pancreas, liver, spleen, serum, and blood cells were collected from each
BALB/c mouse infected with mouse rotavirus EM and control BALB/c mice at
days 2 to 7 after infection (n ? 5 per day) and processed as described previously
(21). Intestinal cells were obtained by using established methods (14, 59). In
brief, the small intestine was placed in Hanks balanced salt solution containing
5% (vol/vol) fetal bovine serum (CSL, Ltd., Melbourne, Australia) and 1 mM
dithiothreitol (Sigma). Intestines were cut open lengthwise and then into 1-cm
pieces. Cells (including enterocytes) were detached by using an orbital shaker at
37°C for 40 min, filtered through a 70-?m-pore-size cell strainer, and pelleted by
centrifugation at 450 ? g for 5 min. Since EM was not culture adapted, rotavirus
antigen in organs and samples (except the small intestine) from infected BALB/c
mice was detected by capture EIA without culture amplification. The proportion
of detached intestinal cells expressing rotavirus antigen was determined by flow
cytometric analysis of methanol-fixed cells, as described previously (22).
Assay for rotavirus antibodies. Titers of rotavirus antibodies were determined
in the sera from all mice by EIA using RRV antigen, as previously described (21).
All mice were negative for serum antibodies to RRV prior to experimentation.
Seroconversion was defined as a fourfold increase in antibody titer.
Glucose homeostasis and diabetes monitoring. Glycosuria was monitored by
using Diastix reagent strips at 1 day prior to RRV or control inoculation and at
days 3, 5, 7, 9, and 11 after infection. In the weekly screening after inoculation,
urine testing was alternated with measurement of blood levels by using an
Accu-Check Advantage II blood glucose meter and strips. Blood glucose levels
were determined immediately in glycosuric mice. As described previously, con-
secutive blood glucose levels of ?240 mg/dl on two occasions 2 to 3 days apart
were considered to indicate diabetes development (21).
Histology. Pancreases were fixed in Bouin’s solution and embedded in paraffin.
Sections (5 ?m) were cut 200 ?m apart at four levels and stained with hema-
toxylin and eosin as previously described (21). For each mouse a quantitative
insulitis score was determined by using an adaptation of a previously described
method (43). All islets present in each level of the pancreas (at least 20 per
mouse) were scored as 0 (no islet infiltrate), 1 (insulitis visible around the outer
edge of the islet), 2 (intra-islet infiltration into ?30% of total islet area), 3
(intra-islet infiltration into 30 to ?70% of total islet area), or 4 (infiltration into
70 to 100% of the islet).
Isolation and flow cytometric analysis of cells from islets and PLN. Purified
islets were obtained from pancreases by bile duct cannulation, collagenase di-
gestion, and density gradient separation, as described previously (32). Cells
released from islets with trypsin-EDTA were allowed to recover in culture
medium for 0.5 to 1.5 h before analysis, as described previously (13). PLN were
dissected from harvested pancreases, mechanically disrupted by using glass
slides, filtered through nylon mesh, and washed. Single-cell suspensions from
islets and PLN were labeled for T- and B-cell analysis with combinations of
monoclonal antibodies specific to mouse antigens (BD Biosciences). Cells were
labeled with fluorescein isothiocyanate-conjugated CD4 (GK1.5), phycoerythrin-
conjugated CD45R/B220 (RA3-6B2), and CyC-conjugated CD8a (53-6.7). In
addition, islet cells were labeled with allophycocyanin-conjugated anti-CD45
(30-F11) to isolate lymphocytes from other cell types. For detection of major
histocompatibility complex class I (MHC-I), islet cells were labeled with biotin-
ylated anti-H-2Kdand allophycocyanin-conjugated anti-CD45, followed by phy-
coerythrin-conjugated streptavidin. Propidium iodide at 50 ?g/ml was included at
the final step to allow the exclusion of dead cells. Cell data were acquired on a
FACScalibur flow cytometer and analyzed with CellQuest Pro software v5.2 (BD
Biosciences). Forward and side scatter gates for lymphocytes were set on CD45?
cells from islets and on all PLN cells. These lymphocytes were then analyzed for
CD4, CD8, and B220.
Islet expression of TNF-? mRNA. Total RNA was extracted from purified
islets by using TRIzol reagent (Invitrogen). Total RNA was reverse transcribed
by using random primers. Real-time PCR was conducted using Assays-on-De-
mand kits (Applied Biosystems) for tumor necrosis factor alpha (TNF-?) and ?
actin as a reference gene, in a Corbett Research Rotor-Gene 3000 sequence
6140GRAHAM ET AL. J. VIROL.
Statistical analysis. The Student paired t test, Mann-Whitney test, and one-
way analysis of variance (ANOVA) were used as appropriate. Other tests were
used as indicated. Diabetes curves were evaluated by Kaplan-Meier life-table
analysis and the log-rank test for trend.
Older NOD and NOD8.3 TCR mice infected with RRV
showed asymptomatic infection, limited intestinal replication,
and no viremia or extraintestinal spread. In order to examine
virological parameters of rotavirus infection in insulitic mice,
NOD mice aged 12 weeks (females) or 15 weeks (males) and
NOD8.3 TCR mice aged 5 weeks were infected by oral gavage
with RRV rotavirus. All RRV-inoculated mice seroconverted
to homologous rotavirus by 2 weeks after infection, whereas all
diluent-inoculated mice showed negative antibody titers to
RRV of ?1:50 (see Fig. 3; also data not shown). No RRV-
infected or control mice showed diarrhea. On day 4 after
infection, 5% (1/20) of stools from both male and female NOD
mice contained infectious RRV, and a stool from another
female mouse contained rotavirus antigen but not infectious
virus. On day 7 after infection, 12% (1/8) of stools from fe-
males also contained RRV antigen. Males did not excrete any
detectable rotavirus antigen at any time after infection. Over-
all, 23% of female NOD mice and 5% of males excreted
detectable infectious RRV and/or rotavirus antigen. On days 2
to 4 after infection of approximately equal numbers of female
and male NOD8.3 TCR mice, stools from 23% of females and
18% of males contained infectious RRV, and 25% (1/4) of the
small intestines collected on day 4 contained infectious RRV.
Overall, RRV was found intestinally in 45% of NOD8.3 TCR
mice. The altered T-cell repertoire of NOD8.3 TCR mice
might have contributed to their increased rate of intestinal
RRV detection over NOD mice, through a reduced ability to
control RRV replication. The stool and intestinal extracts gen-
erally contained low levels of infectious RRV, showing OD460
values for RRV antigen by EIA after culture amplification of
?0.25, with one exception (OD460? 1.1). In our experience,
samples with OD460values of ?0.25 contain levels of infectious
RRV below the detection limit for direct titration in permissive
cells (4 ? 102FCFU/ml), so titration was not attempted (21).
Infectious RRV was not detected in the pancreases, livers,
spleens, sera, or blood cells of these NOD and NOD8.3 TCR
mice by culture amplification and then EIA. This method was
successfully used in our laboratory by the same experimenters
in the same year to detect infectious rotavirus at these sites in
RRV-infected infant NOD mice (21). In addition, RRV anti-
gen was not detected in the pancreas by EIA.
RRV infection of female and male NOD mice with estab-
lished pancreatic insulitis accelerated diabetes development.
To determine whether rotavirus infection modulates the tim-
ing and incidence of spontaneous diabetes onset, groups of 12-
to 15-week-old NOD mice were orally inoculated with RRV,
virus diluent, or cell extract and then monitored for glycosuria
and hyperglycemia (Fig. 1). These studies overlapped in time
and location with closely related experiments undertaken by
our group, which showed that oral RRV inoculation of NOD
mice delays diabetes onset in infants and has little effect in
adults aged 4 to 6 weeks (21). Mice inoculated with cell extract
were included to control for the effect of any contamination of
purified RRV with immunostimulatory components from cells
used to propagate RRV.
NOD mice did not show glucosuria in the 2 weeks after
RRV infection or control inoculations. However, hyperglyce-
mia and diabetes began to develop at 3 weeks (females) and 9
weeks (males) after RRV infection, 4 weeks (females) and 7
weeks (males) earlier than mice inoculated with virus diluent
Two female NOD mice inoculated with cell extract devel-
oped diabetes 1 week and 4 weeks earlier than any virus di-
luent-inoculated females (Fig. 1A). However, there was no
significant difference between the diabetes curves of females
inoculated with virus diluent or cell extract (P ? 0.73), indi-
cating that any MA104 cell components in the purified RRV
preparation would not materially affect diabetes development.
A significant acceleration in diabetes development was ob-
served from the diabetes curves of RRV-infected females com-
pared to females fed virus diluent (P ? 0.023) or cell extract
(P ? 0.019). The proportion of females with diabetes at 21
weeks of age increased from 14% (2/14) in those fed cell
extract to 60% (9/15) after RRV infection (P ? 0.021). How-
ever, by 34 weeks of age the diabetes proportions in control
and RRV-infected mice were similar (0.24 ? P ? 0.43).
The diabetes curves of cell extract- and RRV-inoculated
males showed no significant difference from those of diluent-
fed males (Fig. 1B; P ? 0.30 and P ? 0.18, respectively).
However, by survival analysis RRV-infected male mice showed
a significant acceleration in diabetes development compared to
mice fed cell extract (P ? 0.035). Compared to diluent- and
FIG. 1. Diabetes development was accelerated by RRV infection
of female (A) and male (B) NOD mice with established insulitis. Mice
were inoculated orally at 12 weeks (female) or 15 weeks (male) of age
with RRV, virus diluent, or cell extract and then monitored for dia-
betes until 34 weeks (female) or 45 weeks (male) of age. The data
provided are from a single experiment that is representative of two
independent experiments, each performed with similar numbers of
VOL. 82, 2008ROTAVIRUS ACCELERATION OF TYPE 1 DIABETES 6141
extract-fed mice, the proportion of diabetic males at 45 weeks
of age increased ?4-fold after RRV infection, from 7% (1/14)
and 0% (0/15), respectively, to 27% (4/15), although this was
not statistically significant (P ? 0.33 and P ? 0.09, respec-
RRV infection of female and male NOD8.3 TCR mice with
established pancreatic insulitis accelerated diabetes develop-
ment. The effect of RRV infection on the timing and incidence
of diabetes was examined in NOD 8.3 TCR mice, which spon-
taneously develop diabetes more rapidly than NOD mice. As
shown in Fig. 2A, the diabetes survival curves in diluent-inoc-
ulated and naive female NOD8.3 TCR mice were indistin-
guishable (P ? 0.45). Compared to diluent-inoculated and
naive females, RRV-infected females showed a highly signifi-
cant acceleration in timing of diabetes onset and increased
diabetes incidence (P ? 0.0052 and P ? 0.0072, respectively).
The increased diabetes incidence was evident 1 week after
infection and maintained to 17 weeks of age (Table 1). These
findings indicate that the kinetics of diabetes onset was accel-
erated by RRV infection.
Male NOD8.3 TCR mice showed a lower diabetes incidence
than females, as expected from previous studies and the NOD
genetic background of these mice (54). The diabetes survival
curves in diluent-inoculated and naive male NOD8.3 TCR
mice were indistinguishable (Fig. 2B; P ? 0.85). RRV infection
significantly accelerated diabetes onset and incidence in males
compared to diluent inoculation (P ? 0.038), and a trend for
infected males to show accelerated diabetes over naive males was
evident (P ? 0.06). RRV-infected and control mice showed sim-
ilar diabetes incidences at 17 weeks of age (Table 1).
Relatively high proportions of naive female (57%; 16/28)
and male (36%; 8/22) NOD8.3 TCR mice became diabetic by
17 weeks of age, and the sexes showed similar overall patterns
of diabetes development following diluent and rotavirus inoc-
ulation (Fig. 2A and B). This allowed combination of the sexes
for overall analysis, in contrast to NOD mice (Fig. 2C). The
rates of diabetes observed irrespective of sex in diluent-inocu-
lated and naive NOD8.3 TCR mice were indistinguishable (P ?
0.16). However, diabetes onset was more rapid in RRV-in-
fected mice, particularly in the first 3 weeks after infection
(Table 1). After this time the slopes of the diabetes curves of
RRV-inoculated, diluent-inoculated, and naive mice were sim-
ilar, again showing that the diabetes acceleration induced by
RRV manifested predominantly in the first 3 weeks. Overall,
RRV infection significantly accelerated the onset of diabetes
over diluent-inoculated and naive mice (P ? 0.008 by survival
analysis). The proportions of diabetic mice at 17 weeks of age
did not differ between test and control mice (Table 1). How-
ever, a subset of these mice (13 females and 19 males) was
monitored until 19 to 20 weeks of age. The diabetes incidence
in this subset increased from 47% (7/15) in diluent-inoculated
FIG. 2. Diabetes development was accelerated by RRV infection of
female (A), male (B), or female and male (C) NOD8.3 TCR mice with
or virus diluent and monitored for diabetes until 17 weeks of age. The
data were compiled from four independent experiments performed over
a 1-year period, each of which showed similar patterns of diabetes devel-
opment. The naive NOD8.3 TCR mouse curves represent spontaneous
diabetes development in the animal facility during this time.
TABLE 1. Effects of RRV infection on the timing of diabetes onset in NOD8.3 TCR mice
Time (wk) after
Proportion (%) of diabetic mice after the indicated inoculation
RRV DiluentNoneRRV Diluent NoneRRV Diluent None
aP ? 0.0005 (Fisher exact test).
bP ? 0.0035 (Fisher exact test).
c0.25 ? P ? 0.38 (Fisher exact test).
d0.059 ? P ? 0.066 (chi-square test).
6142GRAHAM ET AL. J. VIROL.
mice to 88% (15/17) in RRV-infected mice (Fisher exact test,
P ? 0.021).
Relation between serum antibody titers to RRV and degree
of diabetes acceleration. Diabetic and nondiabetic mice inoc-
ulated with RRV showed similar geometric mean titers of
serum anti-rotavirus antibodies (NOD, 1:4,300 and 1:3,200,
respectively; NOD8.3 TCR, 1:1,200 and 1:1,500, respectively;
P ? 0.05). The reduced NOD8.3 TCR mouse titers probably
resulted from their altered T-cell repertoire. Stratification of
RRV-infected mice by antibody titer demonstrated that the
extent of diabetes acceleration was significantly greater in mice
showing higher titers of anti-rotavirus antibody in serum (Fig.
3; survival analysis and log-rank test for trend, P ? 0.023 and
P ? 0.0074 for NOD, respectively, and P ? 0.042 and P ?
0.0059 for NOD8.3 TCR, respectively). High-responding NOD
and NOD8.3 TCR mice developed diabetes significantly ear-
lier than diluent-inoculated mice (Table 2; 0.017 ? P ?
0.028). NOD mouse inoculation with a fivefold-higher dose
of RRV (1.0 ? 107FCFU) did not materially affect the
diabetes curves or seroconversion antibody titers obtained
(data not shown).
Rotavirus infection of NOD mice at 10 to 20 weeks of age in
natural experiments. Several older adult NOD mice were in-
advertently exposed to EM mouse rotavirus infection when
coquarantined with mice imported from a rotavirus-positive
facility. Since EM exposure was detected during monitoring of
antibodies to RRV by EIA in sera collected fortnightly or
monthly, its timing could be determined to within a 2- to
4-week interval. No gastroenteritis was observed in the im-
ported mice or the exposed older adult NOD mice during this
period, suggesting EM was of low virulence for older adults. The
NOD mice were monitored for diabetes until 35 weeks old.
Most (6/9) of the adult male NOD mice that were exposed
to EM seroconverted to rotavirus, aged 13 to 20 weeks, and 5
of 6 (81%) of these EM-exposed mice developed diabetes aged
a mean ? the standard deviation of 25 ? 2 weeks (Fig. 4). In
contrast, none (0/3) of the EM-exposed male mice that did not
seroconvert developed diabetes (Fisher exact test, P ? 0.04).
The male diabetes incidence in this facility is ?10% at 30
weeks (Fig. 1), highlighting the extent of the increased diabetes
incidence after rotavirus seroconversion. Females that had se-
roconverted to rotavirus after oral RRV inoculation at 4 weeks
of age also were inadvertently exposed to EM. Many of these
females (6/14; 43%) seroconverted to rotavirus a second time
at 10 to 14 weeks of age. These mice all developed diabetes, at
18 ? 2 weeks of age (Fig. 4). In contrast, only 2 of 8 (25%) of
the RRV-inoculated females that did not seroconvert a second
time developed diabetes, at 16 and 25 weeks of age (P ? 0.01).
The diabetes incidence in female NOD mice in this facility was
?60% at 30 weeks (Fig. 1). These findings of diabetes accel-
eration in highly insulitic NOD mice after EM rotavirus expo-
sure are consistent with the results obtained from the RRV
infection experiments described above that were conducted in
NOD mice under controlled conditions.
All BALB/c mouse pups inoculated with a stool extract con-
taining high levels of EM antigen showed diarrhea that re-
TABLE 2. Relation between serum antibody titers to RRV and
mouse age at diabetes onset
Mean age (wk) ? SD at diabetes onset of mice with
antibody titers to RRV of:
24.2 ? 5.1
11.5 ? 2.9
22.9 ? 4.3
9.7 ? 3.6
19.3 ? 3.2c
6.5 ? 1.0d
10.6 ? 2.3
bRRV-infected NOD mice did not show titers in this range.
cP ? 0.027 compared to diluent-fed mice.
dP ? 0.018 compared to diluent-fed mice.
FIG. 3. Diabetes acceleration was of greater extent in mice showing
higher titers of anti-rotavirus antibody (Ab) in serum. The diabetes
curves for RRV-infected female NOD mice from Fig. 1 (A) and
NOD8.3 TCR mice from Fig. 2 (B) were stratified into mice with low
(1:100 to 1:400), medium (1:800 to 1:3,200), and high (1:6,400 to
1:25,600) serum antibody titers to RRV at 2 weeks after infection.
Serum from one NOD8.3 TCR mouse was unavailable for antibody
testing, so the results from that mouse were excluded.
FIG. 4. Effect of exposure to mouse rotavirus EM in a natural
experiment on diabetes development in NOD mice. Mice are stratified
into those that seroconverted to rotavirus after possible exposure to
EM (?serocon) and those that did not seroconvert (?serocon). Fe-
males had been inoculated with RRV at 4 weeks of age prior to EM
exposure, whereas males had been inoculated at 4 weeks of age with
control cell extract but had no previous rotavirus exposure.
VOL. 82, 2008 ROTAVIRUS ACCELERATION OF TYPE 1 DIABETES6143