Herpesvirus latency confers symbiotic protection from bacterial infection.

Abstract

All humans become infected with multiple herpesviruses during childhood. After clearance of acute infection, herpesviruses enter a dormant state known as latency. Latency persists for the life of the host and is presumed to be parasitic, as it leaves the individual at risk for subsequent viral reactivation and disease. Here we show that herpesvirus latency also confers a surprising benefit to the host. Mice latently infected with either murine gammaherpesvirus 68 or murine cytomegalovirus, which are genetically highly similar to the human pathogens Epstein-Barr virus and human cytomegalovirus, respectively, are resistant to infection with the bacterial pathogens Listeria monocytogenes and Yersinia pestis. Latency-induced protection is not antigen specific but involves prolonged production of the antiviral cytokine interferon-gamma and systemic activation of macrophages. Latency thereby upregulates the basal activation state of innate immunity against subsequent infections. We speculate that herpesvirus latency may also sculpt the immune response to self and environmental antigens through establishment of a polarized cytokine environment. Thus, whereas the immune evasion capabilities and lifelong persistence of herpesviruses are commonly viewed as solely pathogenic, our data suggest that latency is a symbiotic relationship with immune benefits for the host.

Full text

Page 1

LETTERS
Herpesvirus latency confers symbiotic protection
from bacterial infection
Erik S. Barton1{, Douglas W. White1,5, Jason S. Cathelyn2, Kelly A. Brett-McClellan1, Michael Engle3,
Michael S. Diamond1,2,3, Virginia L. Miller2,4& Herbert W. Virgin IV1,2
All humans become infected with multiple herpesviruses during
childhood.Afterclearanceofacuteinfection,herpesvirusesentera
dormant state known as latency. Latency persists for the life of the
host and is presumed to be parasitic, as it leaves the individual at
risk for subsequent viral reactivation and disease1. Here we show
that herpesvirus latency also confers a surprising benefit to the
host. Mice latently infected with either murine gammaherpes-
virus 68 or murine cytomegalovirus, which are genetically highly
similar to the human pathogens Epstein–Barr virus and human
cytomegalovirus2, respectively, are resistant to infection with the
bacterial pathogens Listeria monocytogenes and Yersinia pestis.
Latency-induced protection is not antigen specific but involves
prolonged production of the antiviral cytokine interferon-c and
systemic activation of macrophages. Latency thereby upregulates
the basal activation state of innate immunity against subsequent
infections. We speculate that herpesvirus latency may also sculpt
the immune response to self and environmental antigens through
establishmentofapolarizedcytokineenvironment.Thus,whereas
the immune evasion capabilities and lifelong persistence of her-
pesviruses are commonly viewed as solely pathogenic, our data
suggest that latency is a symbiotic relationship with immune ben-
efits for the host.
Murine gammaherpesvirus 68 (cHV68) undergoes brief lytic rep-
lication in a number of cell types, and establishes lifelong latency
in memory B cells, macrophages and dendritic cells in vivo3. While
investigatingtheroleofinterferonsincontrollingcHV68reactivation
afterintranasalinfection,wenotedthatperitonealmacrophagesfrom
latently infected mice were uniformly activated, displaying cytoplas-
mic vacuolization, membrane ruffling, increased size and upregula-
tion of surface major histocompatibility complex (MHC) class II
molecules (Fig. 1a–d; Supplementary Fig. 1). Peritoneal macrophage
activationwasprominentforatleasttwomonthsaftermucosalinocu-
lation of cHV68 and clearance of lytic infection (data not shown). As
fewer than 1 in 500 peritoneal macrophages are latently infected4,
macrophage activation was not a direct consequence of infection. In
contrast to macrophages explanted from mock-infected mice (Fig.
1e),macrophagesfromlatentlyinfectedmicewerebactericidal,killing
L. monocytogenes rapidly after uptake (Fig. 1f, 6 h after infection).
Because activated macrophages are an important innate defence
against bacterial pathogens, we considered the hypothesis that her-
pesviruslatencymightbeassociatedwithhostresistancetoinfection.
To test this hypothesis, we challenged mice with L. monocytogenes, a
Gram-positive intracellular pathogen controlled in vivo by activated
macrophages5. Acute cHV68 infection did not alter susceptibility to
thispathogen(Fig.2a).Incontrast,latentlyinfectedmicewerehighly
resistant to L. monocytogenes (Fig. 2b). Significant protection from
L.monocytogeneswasobservedupto3monthsaftercHV68infection
(Fig. 2c), the latest time point assessed. Protection correlated with
,100-fold reduction in bacterial burden in the spleen and liver
(Fig.2d,e).ApointmutantofcHV68thatiscapableoflyticinfection
but severely defective in establishment of latency (ORF73.stop)6did
not protect mice from L. monocytogenes infection and did not
stimulate chronic macrophage activation (Fig. 2d, e; Supplemen-
tary Fig. 1), demonstrating the importance of latent infection in this
1DepartmentsofPathologyandImmunology,2MolecularMicrobiology,3Medicine,and4Pediatrics,and5DivisionofRheumatology,WashingtonUniversityMedicalSchool,660South
Euclid Avenue, St. Louis, Missouri 63110, USA. {Present address: Purdue University, Department of Biological Sciences, 915 W. State Street, West Lafayette, Indiana 47907, USA.
0612
2
3
4
5
06
12
2
3
4
5
100
100
101
102
103
104
101
102
103
104
ba
ef
cd
F4/80
Class II MHC
100
100
101
102
103
104
101
102
103
104
F4/80
Class II MHC
P=0.67
P<0.0001
P<0.0001
Hours after infectionHours after infection
log10(C.F.U. ml–1)
log10(C.F.U. ml–1)
Figure 1 | Macrophages are activated during cHV68 latency.
a–f, Peritoneal macrophages from mock (a, c, e) or latently infected
(b,d,f)mice.a,b,Wright’sGiemsastain.Scalebar,25mm.c,d,Surfaceclass
II MHC levels on F4/801peritoneal macrophages (arrow) in mock-infected
mice (57614 (mean fluorescence intensity 6 s.e.m.); n523) compared
with latently infected mice (1,3636144; n518). e, f, Ex vivo bactericidal
activity of peritoneal macrophages at the indicated times. P values,
calculated using the Mann–Whitney rank sum test, compare latently
infected and mock-infected mice at the same time points. Data are
representativeoftwotofiveindependentexperiments.c.f.u,colony-forming
units.Horizontalbarsindicatethearithmeticmeanoflog-transformeddata.
Vol 447|17 May 2007|doi:10.1038/nature05762
326
Nature
©2007
PublishingGroup

Page 2

protection. Latency-induced protection and macrophage activation
were also observed in mice infected with murine cytomegalovirus
(MCMV), a betaherpesvirus (Fig. 2d, e; Supplementary Fig. 1). Pre-
vious infection with human herpes simplex virus type-1 (HSV-1) or
Sindbis virus (an RNA virus incapable of latent infection) triggered
neither protection from bacterial infection nor chronic macrophage
activation(Fig.2d;SupplementaryFig.1).Thus,latentinfectionwith
two divergent herpesviruses results in protection of the host from
L. monocytogenes, indicating that this phenotype is a general and
previously unappreciated consequence of herpesvirus latency.
To explore the possibility that herpesvirus latency enhanced anti-
bacterial resistance more generally, we challenged cHV68 latently
infected mice with the human and rodent pathogen Y. pestis, a
Gram-negative extracellular bacterium and the causative agent of
plague7. Using a pneumonic plague model, mock-infected mice dis-
playedrapidY.pestisreplicationinthelung,withspreadtothespleen
at later time points (Fig. 3a, b). In contrast, latently infected mice
were resistant, exhibiting decreased Y. pestis replication in the lung
(Fig. 3a) and decreased systemic spread (Fig. 3b). Latently infected
mice also exhibited resistance in a bubonic plague model, displaying
100- to 1,000-fold decreases in systemic Y. pestis replication (Fig. 3c,
d). The protection afforded by latency was not universal, as latently
infected mice succumbed to West Nile virus (WNV) infection with
kinetics indistinguishable from mock-infected mice (Fig. 3e).
To define the mechanism whereby herpesvirus latency protects
against two distinct bacterial pathogens, we measured cytokine
levels in the serum of latently infected mice. We found that latent in-
fection triggeredelevatedlevelsofinterferon-c(IFN-c)andtumour-
necrosis factor-a (TNF-a), soluble mediators of macrophage
activation and resistance to bacterial infection (Fig. 4a). IFN-c was
required for latency-induced protection from bacterial challenge
(Fig. 4b), and consistent with a critical role for latent infection in
this process, we did not observe IFN-c upregulation in mice infected
with the latency-defective ORF73.stop mutant (data not shown).
Infection of H–2Kb2/2H–2Db2/2B2M2/2mice indicated that clas-
sical class I MHC-restricted CD81T-cells, an important source
of IFN-c during cHV68 infection8, were completely dispensable
for protection (Fig. 4c). Notably, both Ifng2/2and H–2Kb2/2
H–2Db2/2B2M2/2mice clear acute infection with cHV68 and
05 101520
0
25
50
75
100
Mock
γHV68
05101520
0
25
50
75
100
0
1051520
0
25
50
75
100
O73.stop
MCMV
HSV
Sindbis
Listeria
0
1
2
3
4
5
6
7
8
Mock
Mock
γHV68
γHV68
O73.stop
MCMV
0
1
2
3
4
5
6
7
8
Survival (%)Survival (%)Survival (%)
Days after Listeria challengeDays after Listeria challengeDays after Listeria challenge
Mock
γHV68
γHV68
Mock
P<0.0001
P=0.0038
log10 (C.F.U. per sample)
log10 (C.F.U. per sample)
P<0.0001
P<0.0001
P=0.0008
P<0.0001
P<0.0001
a
de
bc
Figure 2 | Latent herpesvirus infection renders mice resistant to
L.monocytogenes. a–c,SurvivalofmiceinfectedwithcHV681(a),4(b)or12
(c) weeks before challenge with L. monocytogenes. P values were calculated
usinglogrankanalysis(twotothreeexperiments,n58–15micepercondition).
d,e,Bacterialtitresinthespleen(d)andliver(e)3daysafterchallengeinmice
infected 28–42 days previously with the indicated virus or a sublethal dose of
L.monocytogenes.PvalueswerecalculatedusingMann–Whitneyranksumtest
compared to mock (three to five independent experiments). Horizontal bars
indicate the arithmetic mean of log-transformed data. The dashed line
represents the limit of detection. Where not indicated, P.0.3.
3
4
5
6
7
8
4
5
6
7
8
9
10
0
1
2
3
4
5
0
2
4
6
8
10
0510152025 30
0
20
40
60
80
100
Mock
Latent
Mock
Latent
Mock
Latent
Mock
Latent
Lung Spleen
Lung Spleen
Lung Spleen
Mock
Lung Spleen
Latent
ab
c
d
e
XX
log10 (C.F.U. g–1)
log10 (C.F.U. g–1)
log10 (C.F.U. g–1)
log10 (C.F.U. g–1)
P<0.0001
P<0.0001
P=0.0007
P<0.0001
Survival (%)
Days after WNV challenge
Figure 3 | Latently infected mice are resistant to Y. pestis but not WNV.
a–d,Bacterialtitresinmock(blacksymbols)orlatentlyinfected(cHV68,red
symbols) mice challenged intranasally (a, 24 hours past infection (h.p.i.),
lung, b, 48 h.p.i.) or subcutaneously (c, 36 h.p.i., spleen, d, 60 h.p.i.) with
Y. pestis. P values, calculated using the Mann–Whitney rank sum test,
compare latently infected and mock-infected mice (two independent
experiments). X indicates death before harvest. Horizontal bars indicate the
arithmetic mean of log-transformed data. The dashed line represents the
limit of detection. Where not indicated, P.0.08. e, Survival of mock or
latently infected mice challenged by footpad inoculation with WNV. The
P value was calculated using logrank analysis with ten mice per group.
NATURE|Vol 447|17 May 2007
LETTERS
327
Nature
©2007
PublishingGroup

Page 3

establish latent infection, which is associated with persistent replica-
tion below the level of detection of standard plaque assays8–10. This
indicates that latency-induced protection does not require classical
CD81T-cell-derived IFN-c,andmay bemediatedbyother celltypes
that have been demonstrated to produce IFN-c during gammaher-
pesvirus latency, including activated CD41T cells3,11–13.
Infection-induced ‘cross protection’ between unrelated bacterial
pathogens was the foundational observation defining cell-mediated
immunity14. Classicalcross protection occurswhen effector lympho-
cytes responding to the initial infection secrete IFN-c, thereby activ-
ating bystander macrophages and generating a heightened state of
innate immunity to secondary infection. This classical form of cross
protection wanes rapidly once the primary pathogen is cleared and
responding lymphocytes differentiate from effector to memory
phenotype15, and has not previously been observed during latent
virus infection.
Incontrast, twomechanisms ofcrossprotection that aremediated
by memory lymphocytes have been recently described. In the first,
termed heterologous immunity, cross-reactive antigenic epitopes
between the primary and secondary pathogens result in antigen-
specific memory lymphocyte activation after secondary infection16.
In the second mechanism, CD81memory T cells generated after
primary infection can become activated in an antigen-independent
manner by cytokines (including interleukin (IL)-12 and IL-18)
expressed during the early stages of subsequent infections. In both
cases, memory T-cell activation leads to secretion of IFN-c and an
enhanced response to the secondary infection17. In contrast to clas-
sical cross protection, memory-mediated cross protection does not
involve heightened innate immunity before secondary challenge.
Rather, the second pathogen activates an innate cytokine response
that triggers bystander memory T-cell activation.
The chronic IFN-c secretion and macrophage activation we
observe in latently infected mice suggest that latency induces a pro-
longed state of classical cross protection. However, these observa-
tions do not exclude a role for the T-cell-dependent mechanisms
described above. To determine whether latency-induced cross pro-
tection requires activation of memory T cells at the time of bacterial
challenge, we depleted both CD41and CD81lymphocytes one day
before L. monocytogenes challenge (Fig. 4d). T-cell depletion was
effective, as indicated by elimination of L. monocytogenes-specific
memory in depleted L. monocytogenes-vaccinated mice (Fig. 4d). In
contrast, the resistance of latently infected mice to L. monocytogenes
remainedwhenCD41andCD81Tcellsweredepleted.Weconclude
that viral latency generates a state of broadly effective innate cross
protection that does not require the effector functions of lympho-
cytes at the time of secondary challenge.
Latency-induced cross protection differs from classical cross pro-
tection in at least two aspects. First, IFN-c-dependent cross protec-
tion can occur between latent viral infection and acute bacterial
infections.Second,thisprotectionisquiteprolonged,lastingmonths
after clearance of acute infection. The durability of cross protection
that we observe during gammaherpesvirus latency leads us to pro-
pose that latent infection represents a true symbiotic relationship,
protecting the host from subsequent infection by means of heigh-
tenedinnateimmuneactivation.AlthoughourdatasuggestthatIFN-
c-activated macrophages mediate latency-induced cross protection
during both beta- and gammaherpesvirus infection, other cell types
may contribute to the protective effects observed.
Are our findings with murine herpesviruses relevant for other
latently infected hosts? We are unaware of studies that directly
address the role of herpesvirus infection in human susceptibility to
secondary bacterial infection. However, the observation of latency-
induced cross protection in two subfamilies of the Herpesviridae
suggests that this is an evolutionarily conserved aspect of the virus–
hostrelationship.Furthermore,accumulatingevidenceindicatesthat
latency with all three herpesvirus subfamilies in humans involves
chronic, low-level immune activation accompanied by IFN-c and
TNF-a secretion in response to frequent but subclinical viral react-
ivation18–22. Although we do not observe latency-induced cross pro-
tection in mice infected with the alpha-herpesvirus HSV-1, this may
reflect the tight anatomical restriction of HSV-1 latency in neurons,
in contrast to latent infection of circulating haematopoietic cells by
beta- andgammaherpesviruses. Aplausible hypothesis forthe mech-
anism of latency-induced cross protection involves the chronic, low-
level presentation of viral antigens as a result of viral reactivation,
resulting in prolonged T-cell activation and secretion of IFN-c. This
effect may be particularly important at mucosal and epithelial sites,
where herpesvirus reactivation occurs and most pathogenic chal-
lengesinitiate.OurdataclearlyindicatethatIFN-cexpressionduring
latency can have beneficial effects beyond control of viral reactiva-
tion, and suggest the need for further epidemiological studies of the
immune consequences of herpesvirus infection in humans. These
studies should also take into account the possible deleterious con-
sequences of chronic inflammation elicited during latency, such as
the development of autoimmune disease or cancer23.
Beneficial immune modulation during herpesvirus latency may
extend beyond protection from pathogens. Substantial epidemiolo-
gical data support the ‘hygiene’ hypothesis, in which prior infection
protectsagainstdevelopmentofallergy24.Fewstudieshaveaddressed
therole ofherpesvirus latencyin skewingimmune responses, despite
the increasing age of seroconversion to herpesvirus infection in
populations with increased atopy and autoimmunity25. One study
reports a protective effect of Epstein–Barr virus infection in the
reduction of IgE sensitization to environmental allergens, an effect
thatwasmorepronouncedonco-infection withhumancytomegalo-
virus and Epstein–Barr virus26. Our findings suggest a mechanistic
0
10
20
30
40
MockMockLatentLatent
28 d.p.i. 41 d.p.i.
05101520 25
0
25
50
75
100
Mock
γHV68
0510 152025
0
25
50
75
100
Diluent
0
1
2
3
4
5
6
7
8
a
Cytokine levels (pg ml–1)
b
d
c
*
*
*
*
Survival (%)
Survival (%)
Mock
γHV68
Days after Listeria challenge
Days after Listeria challenge
IL-12
α-TNF
γ-IFN
IL-12
α-TNF
γ-IFN
IL-12
α-TNF
γ-IFN
IL-12
α-TNF
γ-IFN
P<0.0001
log10 (C.F.U. per sample)
P=0.573
P=0.0079
P=0.243
Isotype
Anti-
CD4/CD8
Diluent
Isotype
Anti-
CD4/CD8
Diluent
Isotype
Anti-
CD4/CD8
Mock
γHV68
latent
Listeria
immune
Figure 4 | Mechanism of enhanced innate immunity in latently infected
mice. a, Cytokine levels (mean6s.e.m.) following infection with cHV68.
Asterisk,P#0.0002,relativetomock,Mann–Whitneyranksum.Wherenot
indicated, P.0.4. b, c, Survival of mock or latently infected Ifng2/2(b) or
H–2Kb2/2H–2Db2/2B2M2/2(c) mice, challenged with L. monocytogenes.
The P value is relative to mock, logrank analysis (two to three experiments,
n510–20 mice per condition). d, Bacterial titres in spleen 3 days after
challenge with L.monocytogenes in mock, latently infected or L.
monocytogenes-immune mice. CD41and CD81T cells were depleted 1 day
before challenge. Horizontal bars indicate the arithmetic mean of log-
transformed data. P-values, calculated using the Mann–Whitney rank sum
test, compare depleted to isotype controls (two experiments).
LETTERS
NATURE|Vol 447|17 May 2007
328
Nature
©2007
PublishingGroup

Page 4

explanation for this observation, whereby prolonged secretion of
T-helper-1-type cytokines during latency may inhibit the develop-
ment of T-helper-2-driven immune pathology. Accordingly, poten-
tial benefits of herpesvirus latency should be considered when
evaluating vaccines against herpesvirus infection, as decreased infec-
tion rates may be associated with unintended negative consequences
for vaccinated individuals.
Our current understanding of cellular and molecular immunity is
founded largely on studies in specific-pathogen-free mice that have
no known herpesvirus infections. We now demonstrate that herpes-
virus infection triggers systemic, profound immune modulation,
with the potential to alter significantly the kinetics and nature of
host response to foreign antigens. As there are few herpesvirus-free
humans, we speculate that an accurate understanding of ‘normal’
immunity may require a re-evaluation of immune function in the
presence ofthesesymbionts, which have beenexerting selective pres-
sures on the mammalian immune system since before the vertebrate
radiation27.
METHODS SUMMARY
Mice, viruses and infections. C57BL/6J mice were infected with virus between
8 and 20 weeks of age. Ifng2/2or H–2Kb2/2H–2Db2/2B2M2/2mice are
described elsewhere8,10. Mice received wild-type or ORF73.stop cHV68
(WUMS strain, intranasal), HSV-1 (strain 17, intraperitoneal), MCMV (strain
Smith, intraperitoneal), or Sindbis virus (AR339, intraperitoneal). Mice were
consideredlatentlyinfectedat28daysafterinfection.Listeriamonocytogeneswas
administered intraperitoneally. Yersinia pestis was administered intranasally7or
subcutaneously.WNVwasadministeredbyfootpadinoculation28.Lymphocytes
were depleted by administration of anti-CD4 (YTS-191.1), anti-CD8 (H35), or
isotype control SFR-DR5 rat monoclonal antibodies intraperitoneally29.
Flow cytometric analysis. Peritoneal cells were obtained by lavage, blocked
in 10% normal mouse serum, and stained for cell-surface MHC class II (I–Ab/
I–Eb; Phyroerythrin-conjugated, Pharmingen) and F4/80 (allophycocyanin-
conjugated, Caltag) and analysed on a Beckton-Dickinson FACScalibur.
L.monocytogenes growth in macrophages. Adherent peritoneal macrophages
wereinfectedwithL.monocytogenes;unboundbacteriawereremoved.At30min
after-adsorption,gentamicin(5mgml21)wasaddedtokillextracellularbacteria,
permitting the direct assessment of macrophage capacity to kill phagocytosed
L.monocytogenes.Viableintracellularbacteriawerequantifiedbyhypotoniclysis
of peritoneal cells and plating onto BHI agar plates30.
Quantification of bacterial growth in vivo. Spleens, livers and lungs were
removedfromL.monocytogenes-orY.pestis-infectedmiceattheindicatedtimes
after bacterial challenge. For quantification of L. monocytogenes, spleens were
homogenized in 10ml PBS/0.05% Triton X-100 and plated on BHI-agar plates.
Organs from Y. pestis-infectedmice were homogenized in sterile PBS and plated
onto BHI agar plates.
Quantificationofcytokineexpression.Cytokinesweredetectedinwholeserum
usingthemouseinflammationcytometricbeadarray(BDPharmingen)accord-
ing to the manufacturer’s instructions.
Full Methods and any associated references are available in the online version of
the paper at www.nature.com/nature.
Received 7 February; accepted 16 March 2007.
1. Rickinson, A. B. & Kieff, E. Fields Virology (eds Knipe, D. M. & Howley, P.)
2655–2670 (Lippincott-Raven, Philadelphia, 2007).
Virgin, H. W. et al. Complete sequence and genomic analysis of murine
gammaherpesvirus 68. J. Virol. 71, 5894–5904 (1997).
McClellan,K.B.,Gangappa,S.,Speck,S.H.&Virgin,H.W.Antibody-independent
control of gamma-herpesvirus latency via B cell induction of anti-viral T cell
responses. PLoS Pathog. 2, e58 (2006).
Weck, K. E., Kim, S. S., Virgin, H. W. & Speck, S. H. Macrophages are the major
reservoir of latent murine gammaherpesvirus 68 in peritoneal cells. J. Virol. 73,
3273–3283 (1999).
Edelson, B. T. & Unanue, E. R. Immunity to Listeria infection. Curr. Opin. Immunol.
12, 425–431 (2000).
2.
3.
4.
5.
6. Moorman, N. J., Willer, D. O. & Speck, S. H. The gammaherpesvirus 68 latency-
associated nuclear antigen homolog is critical for the establishment of splenic
latency. J. Virol. 77, 10295–10303 (2003).
Cathelyn,J.S.,Crosby,S.D.,Lathem,W.W.,Goldman,W.E.&Miller,V.L.RovA,a
global regulator of Yersinia pestis, specifically required for bubonic plague. Proc.
Natl Acad. Sci. USA 103, 13514–13519 (2006).
Braaten, D. C. et al. Effective control of chronic gamma-herpesvirus infection by
unconventional MHC Class Ia-independent CD8 T cells. PLoS Pathog. 2, e37
(2006).
Sarawar, S. R. et al. Gamma interferon is not essential for recovery from acute
infection with murine gammaherpesvirus 68. J. Virol. 71, 3916–3921 (1997).
10. Tibbetts, S. A., Van Dyk, L., Speck, S. H. & Virgin, H. W. Immune control of the
number and reactivation phenotype of cells latently infected with a gamma-
herpesvirus. J. Virol. 76, 7125–7132 (2002).
11.Christensen,J.P.&Doherty,P.C.Quantitativeanalysisoftheacuteandlong-term
CD41T-cellresponse toapersistentgammaherpesvirus.J.Virol.73,4279–4283
(1999).
12. Flano, E., Woodland, D. L., Blackman, M. A. & Doherty, P. C. Analysis of virus-
specific CD41T cells during long-term gammaherpesvirus infection. J. Virol. 75,
7744–7748 (2001).
13. Polic, B. et al. Hierarchical and redundant lymphocyte subset control precludes
cytomegalovirus replication during latent infection. J. Exp. Med. 188, 1047–1054
(1998).
14. Mackaness, G. B. The immunological basis of acquired cellular resistance. J. Exp.
Med. 120, 105–120 (1964).
15. Mackaness, G. B. The influence of immunologically committed lymphoid cells on
macrophage activity in vivo. J. Exp. Med. 129, 973–992 (1969).
16. Selin, L. K. et al. Memory of mice and men: CD81T-cell cross-reactivity and
heterologous immunity. Immunol. Rev. 211, 164–181 (2006).
17. Berg, R. E., Crossley, E., Murray, S. & Forman, J. Memory CD81T cells provide
innate immune protection against Listeria monocytogenes in the absence of
cognate antigen. J. Exp. Med. 198, 1583–1593 (2003).
18. Tan,L.C.etal.Are-evaluationofthefrequencyofCD81TcellsspecificforEBVin
healthy virus carriers. J. Immunol. 162, 1827–1835 (1999).
19. Dunn, H. S. et al. Dynamics of CD4 and CD8 T cell responses to cytomegalovirus
in healthy human donors. J. Infect. Dis. 186, 15–22 (2002).
20. Amyes,E.etal.CharacterizationoftheCD41TcellresponsetoEpstein-Barrvirus
during primary and persistent infection. J. Exp. Med. 198, 903–911 (2003).
21. Hislop, A. D. et al. Tonsillar homing of Epstein-Barr virus-specific CD81 T cells
and the virus-host balance. J. Clin. Invest. 115, 2546–2555 (2005).
22. Theil, D. et al. Latent herpesvirus infection in human trigeminal ganglia causes
chronic immune response. Am. J. Pathol. 163, 2179–2184 (2003).
23. Balkwill, F. & Mantovani, A. Inflammation and cancer: back to Virchow? Lancet
357, 539–545 (2001).
24. Schaub,B.,Lauener,R.&vonMutius,E.Themanyfacesofthehygienehypothesis.
J. Allergy Clin. Immunol. 117, 969–977 (2006).
25. Takeuchi, K. et al. Prevalence of Epstein-Barr virus in Japan: trends and future
prediction. Pathol. Int. 56, 112–116 (2006).
26. Nilsson, C. et al. Does early EBV infection protect against IgE sensitization?
J. Allergy Clin. Immunol. 116, 438–444 (2005).
27. Roizman, B. & Pellett, P. E. Fields Virology (eds Knipe, D. M. & Howley, P. M.)
2475–2499 (Lippincott-Raven, Philadelphia, 2007).
28. Shrestha, B. et al. Gamma interferon plays a crucial early antiviral role in
protection against West Nile virus infection. J. Virol. 80, 5338–5348 (2006).
29. McClellan, J. S., Tibbetts, S. A., Gangappa, S., Brett, K. A. & Virgin, H. W. Critical
roleofCD4Tcellsinanantibody-independentmechanismofvaccinationagainst
gamma-herpesvirus latency. J. Virol. 78, 6836–6845 (2004).
30. Edelson, B. T. & Unanue, E. R. Intracellular antibody neutralizes Listeria growth.
Immunity 14, 503–512 (2001).
7.
8.
9.
Supplementary Information is linked to the online version of the paper at
www.nature.com/nature.
Acknowledgements This research was supported by grants from the National
Institutes of Health (E.S.B., V.L.M. & H.W.V.), an Abbott Scholar Award (D.W.W.)
and a Cancer Research Institute postdoctoral fellowship (E.S.B.).
Author Contributions The original hypothesis of the article was formulated by
E.S.B. and H.W.V. E.S.B. and D.W.W performed all experiments except those
characterizing Y. pestis infection (J.S.C. and V.L.M.) and WNV infection (M.E. and
M.S.D.). K.A.B.-M. made the initial observation of chronic IFNc secretion during
latency.ThemanuscriptwaswrittenbyE.S.B.,D.W.W.,andH.W.V.andallauthors
commented on data and conclusions prior to submission.
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The authors declare no competing financial interests.
Correspondence and requests for materials should be addressed to H.W.V.
(virgin@wustl.edu).
NATURE|Vol 447|17 May 2007
LETTERS
329
Nature
©2007
PublishingGroup

Page 5

METHODS
Mice and infections. C57BL/6J mice (Jackson and internal breeding colonies)
were infected with virus between 8 and 20 weeks of age. Ifng2/2or H–2Kb2/2
H–2Db2/2B2M2/2mice are described elsewhere8,10and were from internal
breeding colonies. cHV68 infections (104plaque-forming units (p.f.u.)) were
performedintranasallyinavolumeof40mlDMEM/10%FBS(DMEM/10).Lytic
infection of ORF73.stop virus after intranasal inoculation was confirmed by
enzyme-linked immunosorbent assay (ELISA) against total cHV68 antigen,
which demonstrated equivalent serum levels of cHV68-specific antibody after
wild-type cHV68 or ORF73.stop infection (data not shown). HSV1 (104p.f.u.),
MCMV (105p.f.u.), and Sindbis virus (104p.f.u.) were administered intraper-
itoneally in 0.5 ml DMEM/10. Listeria monocytogenes (nonlethal dose, 103c.f.u;
lethaldose106c.f.u)wasadministeredintraperitoneallyin0.5mlsaline.MCMV
infected mice displayed no serological evidence of cHV68 infection by ELISA
(datanot shown).Yersiniapestis usedforsubcutaneous infectionswere cultured
at 26 uC in BHI for 16–18 h, and for intranasal infections bacteria were cultured
at 37 uCinBHIsupplementedwith 2.5mM CaCl2for 16–18h.For analysisofY.
pestis colonization and dissemination mice were infected intranasally with 1 3
104CFU essentially as described7or subcutaneously with 1 3 102c.f.u. in 100 ml
atthebaseofthetail.WNV(100p.f.u.)wasadministeredbyfootpadinoculation
as described28. In some experiments, lymphocytes were depleted by administra-
tion of a single 1-mg dose of anti-CD4 (YTS-191.1), anti-CD8 (H35), or isotype
controlSFR-DR5(AmericanTypeCultureCollection)ratmonoclonalantibody
intraperitoneally29.
Viral and bacterial stocks. cHV68 (strain WUMS) stocks were prepared and
titrated in murine 3T12 fibroblasts. The latency-deficient cHV68 mutant
ORF73.stop was obtained from the laboratory of S. H. Speck6. MCMV strain
Smith was obtained from the laboratory of M. Colonna. HSV-1 strain 17 and
Sindbis virus strain AR339 are laboratory stocks. Listeria monocytogenes strain
EGD stocks were amplified from infected splenic lysate and were obtained from
thelaboratoryofE.R.Unanue.Thevirulentwild-typeY.pestisstrainCO927was
obtainedfromtheUSArmy,Ft.Detrick,Maryland.WNV(strain3000.0259)was
isolatedinNewYorkin2000andpassagedonceinC6/36Aedesalbopictuscells28.
Quantification of L. monocytogenes growth in cultured peritoneal macro-
phages. This assay has been described in detail previously30. Briefly, peritoneal
cells were removed by lavage with 10 ml ice-cold DMEM/10 and allowed to
adhere to polystyrene 24-well plates overnight. Non-adherent cells were
removed by washing with agitation in warm DMEM/10. Monolayers were
infected with 13105c.f.u. of fresh-standing overnight culture of L. monocyto-
genesbycentrifugaladsorption.Unboundbacteriawereremovedbyrinsingwith
warm PBS and thencultured in thepresence of gentamicin (5 mg ml21, added at
30minafterbacterialadsorption)tokillextracellularbacteriathathavenotbeen
phagocytosed, thereby permitting the direct assessment of macrophage capacity
to kill phagocytosed L. monocytogenes before bacterial escape from the endo-
some. At various times after adsorption, viable intracellular bacteria were quan-
tified by hypotonic lysis of macrophage monolayers in water and inoculation of
serial dilutions onto BHI agar plates.
doi:10.1038/nature05762
Nature
©2007
PublishingGroup