Random transposon mutagenesis of large DNA molecules in Escherichia coli.

R E V I E W S
TRENDS IN MICROBIOLOGY 190 VOL. 6 NO. 5 MAY 1998
Copyright © 1998 Elsevier Science Ltd. All rights reserved. 0966 842X/98/$19.00 PII: S0966-842X(98)01255-4
Cytomegaloviruses (CMVs)represent the prototypeviruses of the b subgroup
of the herpesvirus family. They
are characterized by their strict
species specificity, indicative
of a long pathogen–host co-
evolution, and are found in
most mammals. Phenotypically,
CMVs are distinguished by
their slow replication in a lim-
ited number of cell types and
their typical cytopathology.
Both human (HCMV) and
mouse (MCMV) CMVs have
240-kb double-stranded DNA
genomes encoding at least 200
open reading frames (ORFs);
this represents the highest
potential coding capacity within the herpesvirus
family1,2 (see Fig. 1). CMV replication is tightly regu-
lated in a multistep process and can be subdivided into
the immediate early (IE), early (E) and late (L) phases
of gene expression. Although MCMV and HCMV are
the best-studied CMVs, only a few of of their genes
have been characterized. Gene blocks located in the
central 100 kb of the HCMV and MCMV genomes are
closely related to each other, whereas the sequences
near the ends of the genomes harbour gene families
that are arranged in tandem arrays and encode
mostly glycoproteins (gps).
The primary infection by a CMV is usually efficiently
controlled by the immune system and does not cause ma-
jor illness. However, immune control does not achieve
the complete clearance of the virus. Instead, the CMV
genome persists in a nonproductive form at specific
sites in the infected host, with minimal viral gene ex-
pression. Viral DNA replication is frequently reacti-
vated from latency and results in recurrent infection and
virus shedding. Infection with HCMV is widespread
throughout most populations (50–95% seroprevalence).
Only in immunologically immature or immunocom-
promised individuals does primary or recurrent CMV
infection cause severe, disseminated and even fatal dis-
ease manifestations. Intra-uterine infection can result
in congenital defects, including permanent brain dam-
age. CMV retinitis is associated with AIDS and leads
to progressive retinal destruction with visual loss. Fol-
lowing allogeneic bone marrow
transplantation, CMV infection
is a leading cause of pneumonia.
CMV infection can also result
in graft dysfunction and loss fol-
lowing solid organ transplan-
tation. The notion that the sta-
tus of the host’s immune system
largely defines the outcome of
CMV infections is not only con-
sistent with clinical experience
but has also been confirmed by
experimental studies with CMV
infections in their natural hosts.
Cytomegaloviral strategies
to limit immune control
To achieve permanent coexist-
ence with their hosts, CMVs:
(1) establish a latent state of infection and restrict the
number of viral genes expressed in order to minimize
exposure to the immune system; (2) use specific
(‘privileged’) tissues for replication that have a less
stringent immune surveillance [e.g. epithelial cells of
the salivary glands do not express sufficient major
histocompatibility complex (MHC) class I molecules
to mediate virus clearance by CD8+ T cells] and allow
virus shedding into body fluids that are transmitted be-
tween individuals (e.g. saliva, breast milk, semen and
cervical fluids); and (3) compromise antiviral host
defence mechanisms by expressing distinct factors de-
signed to silence the host’s immune response, thereby
lengthening the period available for virus multipli-
cation. The evidence for elaborate molecular mecha-
nisms by which CMVs avoid detection and elimination
is rapidly growing.
Subversion of MHC class I functions by MCMV
In vivo depletion experiments in MCMV-infected mice
and infusion of transferred antiviral immune cells in
bone-marrow-transplanted humans identified CD8+
T cells as the most effective subset for the clearance of
acute infection in visceral organs and for protection from
otherwise lethal infection3,4. However, MCMV repre-
sents the first example of a herpesvirus that directly
interferes with MHC class I-restricted antigen presen-
tation to CD8+ T cells5. In the MHC class I pathway of
antigen presentation (see Fig. 2a), peptides are generated
Slowly replicating, species-specific and
complex DNA viruses, such as
cytomegaloviruses (CMVs), which code
for >200 antigenic proteins, should be easy
prey to the host’s immune system. Yet,
CMVs are amazingly adapted opportunists
that cope with multiple immune responses.
Frequently, CMVs exploit immune
mechanisms generated by the host. These
strategies secure the persistence of CMVs
and provide opportunities to spread to
naive individuals.
H. Hengel*, W. Brune and U.H. Koszinowski are in
the Max von Pettenkofer Institute, Dept of Virology,
Gene Center, Ludwig-Maximilians University,
81377 Munich, Germany.
*tel: +49 89 740 17202, fax: +49 89 740 17250,
e-mail: hengel@lmb.uni-muenchen.de
Immune evasion by cytomegalovirus –
survival strategies of a
highly adapted opportunist
Hartmut Hengel, Wolfram Brune and Ulrich H. Koszinowski

R E V I E W S
TRENDS IN MICROBIOLOGY 191 VOL. 6 NO. 5 MAY 1998
by proteolytic cleavage (e.g. through the multicatalytic
proteinase complex, the proteasome) of viral proteins
in the cytosol. Binding of peptides to MHC class I mol-
ecules requires their vectorial transport across the endo-
plasmic reticulum (ER) membrane by a specific peptide
transporter, TAP1/2 (transporter associated with anti-
gen processing). The TAP complex consists of two
MHC-encoded subunits, TAP1 and TAP2, which form
a heterodimer. The proteins are members of the ATP-
binding cassette (ABC) transporter family. After their
import into the ER, peptides can bind to MHC class I–
b 2-microglobulin (b 2m) heterodimers. The assembly of
MHC class I complexes is assisted by transient inter-
actions with molecular chaperones, which include cal-
nexin, calreticulin and tapasin6,7. Ternary MHC class I
complexes are exported from the ER, via the secretory
pathway, to the plasma membrane for T-cell recognition.
Apparently, both MCMV and HCMV avoid the
stringent control of MHC class I-restricted immunity8–11
by the expression of multiple, virus-specific gene func-
tions that affect antigen presentation by MHC class I
molecules (listed in Table 1). These viral genes are not
required for virus replication in vitro. Their com-
mon phenotype is the loss of MHC class I molecules on
the plasma membrane. However, the underlying mol-
ecular mechanisms differ considerably. In MCMV, three
genes (m152, m04 and m06), which are members of
two gene families (m02 and m145) located within the
flanking regions of the virus genome (see Fig. 1a), have
been identified and encode type I transmembrane gps
that interact with MHC class I complexes12,13 (U. Reusch
and U.H. Koszinowski, submitted). The m152 gene,
coding for a 37/40-kDa gp, accounts for the arrest and
accumulation of MHC class I molecules in the ER–
Golgi intermediate compartment (ERGIC)/cis-Golgi
compartment8,12 (Fig. 2b). Transcription of m152 starts
within 2 h postinfection, reaches a maximum at 4 h post-
infection and declines during the later phases of infec-
tion12. In contrast, m04 and m06 are expressed later
than m152, within a second set of early genes, but are
also active during the late phase of infection. These genes
encode gps of 34 and 48 kDa, respectively, and both pro-
teins attach tightly to folded and b 2m-associated MHC
class I molecules in the ER (Ref. 13; U. Reusch and
U.H. Koszinowski, submitted). The presence of gp34–
MHC class I complexes at the cell surface of MCMV-
infected cells has been interpreted as escape of MHC
class I complexes from the m152/gp40-mediated block-
ade of transport after gp34 association13 (Fig. 2b). It
has been speculated that gp34–MHC class I complexes
displayed at the cell surface may either serve as decoy
receptors for natural killer (NK) cells or prevent pep-
tide recognition by CD8+ T cells, or both. Although they
both belong to the same gene family and share MHC
Hindlll
(a)
(b)
Hindlll
MCMV
HCMV
A
m02
B M H D C G F K L J I O P E
USUL
Q X V W H
N
m03 m04 m05 m06 m07 m08 m09 m10 m11 m12 m13 m14 m15 m16 m17 m158m157m155m153 m154m152m151m150m146m145
US1 US2 US3 US4 US6 US7 US8 US9 US10 US11US5
Fig. 1. HindIII cleavage map of (a) the mouse cytomegalovirus (MCMV) and (b) the human cytomegalovirus (HCMV) wild-type strain AD169
genomes. Upper-case letters represent HindIII fragments. (a) The m02 and m145 gene families harbouring the major histocompatibility complex
(MHC) class I-regulating genes m04, m06 and m152 have been magnified. Note that the m17 gene is a member of the m145 gene family.
(b) The terminal boxes of the US (unique short) and UL (unique long) components indicate the joining regions of the HCMV genome. The US2 and
US6 gene families, which include the MHC I-regulating genes US2 and US3, and US6 and US11, respectively, are located within the US component
of the genome. The arrows represent the direction of transcription.

R E V I E W S
TRENDS IN MICROBIOLOGY 192 VOL. 6 NO. 5 MAY 1998
class I-binding properties, the m06/gp48 protein delivers
MHC class I complexes to a different destination to the
m04/gp34 protein: the late endosome and lysosome. The
diversion of MHC class I complexes from the physiologi-
cal export route to this compartment is probably me-
diated by dipeptide LL (double leucine) sorting motifs
present in the cytoplasmic domain of the protein. In the
lysosome, both MHC class I molecules, as well as gp48,
undergo rapid proteolytic degradation (U. Reusch and
U.H. Koszinowski, submitted) (see Fig. 2b). Whether
the MHC class I subversive genes act in a cooperative
and multistep process during the MCMV replication
cycle awaits further experimentation.
Subversion of MHC class I functions by HCMV
In vitro, presentation of endogenous peptides to CD8+
cytotoxic T lymphocytes (CTLs) is completely abro-
gated in HCMV-infected cells11. As in MCMV, the
MHC class I subversive genes of HCMV are arranged
as members of gene families located at a similar gen-
omic position to the MCMV m152 gene. Analysis of
HCMV deletion mutants lacking genes in the short com-
ponent of the HCMV genome has revealed two inde-
pendent gene regions associated with the downregu-
lation of MHC class I complex formation: one of these
is US11, and the other maps between US2 and US5
(Ref. 14; Fig. 1b). It is remarkable that HCMV controls
the MHC class I pathway of antigen presentation at ear-
lier checkpoints than MCMV does (Fig. 2b). Already
during the IE phase of infection, expression of US3, en-
coding a protein of 32–33 kDa with two glycosylation
forms, leads to the retention of peptide-loaded MHC
class I complexes in the ER (Refs 15,16). Transcription
of US3 ceases soon after infection but is followed by ex-
pression of a neighbouring gene, US2, which encodes a
gp of 24 kDa. The mode of action of gpUS2 on MHC
class I is similar to that elucidated for the 33-kDa gp
encoded by US11 (Ref. 17). In US11-expressing cells,
the half-life of MHC class I heavy chains is shortened
to 1–2 min. When US11 transfectants are treated with
proteasome inhibitors, a deglycosylated 40-kDa inter-
mediate of MHC class I heavy chains accumulates in
the cytosol. This finding indicates that, in the presence
of gpUS2 or gpUS11, MHC class I molecules are dis-
located back to the cytosol, where they are degraded
by the proteasome17,18 (Fig. 2b).
The export pathway for MHC class I molecules
has been resolved in more detail by studies using US2-
expressing cells18,19. Co-immunoprecipitation of a de-
glycosylated MHC class I heavy chain intermediate and
a deglycosylated 20-kDa product of US2 with the Sec61
complex suggests that the export of MHC class I mol-
ecules is mediated through the translocon. This find-
ing implies that the same channel that allows MHC
class I access to the ER lumen is used in reverse for the
export of gps from the ER. Indeed, genetic evidence from
yeast supports the notion that the translocon is linked
to a general retrograde protein transport pathway for
ER degradation20.
HCMV also inhibits TAP-mediated transport of cyto-
solic peptides into the ER lumen, an earlier stage of the
MHC class I antigen presentation pathway21. The in-
activation of TAP is accomplished by an ER-resident
20-kDa gp encoded by the HCMV US6 gene and occurs
despite the significantly augmented expression of TAP
molecules in HCMV-infected cells22–24 (Fig. 2b). Ex-
pression of US6 starts early after infection and reaches
maximum levels in the late phase of infection, when
other genes implicated in MHC class I interference be-
come almost silent25. The gpUS6 protein associates with
the recently identified transient assembly complex6, com-
posed of TAP1, TAP2, MHC class I–b 2m, calreticulin
and tapasin, and also binds to calnexin. However, the
blockade of TAP by gpUS6 is independent of the pres-
ence of the MHC class I heavy chain and tapasin and
does not involve peptide binding to TAP (Ref. 22).
The 72-kDa transcription factor encoded by the
HCMV IE1 transcription unit is abundantly expressed
in the IE and E phases of HCMV infection. In con-
trast, the frequency of CD8+ CTLs with a specificity for
this antigen is relatively low among polyclonal CMV-
specific CTLs (Ref. 26). Lysis of CMV-infected fibro-
blasts by IE-specific CTLs is low but is higher in fibro-
blasts infected with a mutant virus in which a matrix
protein (pp65, encoded by ORF UL83) has been de-
leted27. Recognition of the 72-kDa protein expressed
by a recombinant vaccinia virus is diminished by co-
expression of pp65 but restored when a pp65 construct
that lacks the pp65-associated kinase activity is used27.
From these data, it has been proposed that phosphoryl-
ation of the 72-kDa IE protein by pp65 would limit its
access to the processing machinery27 (Fig. 2b).
Counteracting viral immune evasion
Complete evasion from immune control would result in
the uncontrolled replication of the cytolytic virus, which
would kill the host and limit dissemination of the virus;
however, CMVs do not prevent host mechanisms to
restore antigen presentation of CMV-infected cells to
CD8+ T cells. Experimental evidence suggests that the
viral immune evasion mechanisms operate in vivo with
only a limited efficacy.
First, there is evidence that the degree of inhibition
by both MCMV and HCMV is subject to regulation
by distinct cytokines in vitro11,28. Of these, interferon g
(IFN-g ) is certainly the most potent and can efficiently
restore antigen presentation of infected fibroblasts, al-
though this has also been reported for tumour necrosis
factor a (TNF-a ) and type I IFNs (Ref. 11). Appar-
ently, the cytokines do not affect the expression of viral
genes, nor their inhibitory effect upon MHC class I
functions, but strongly increase the synthesis of MHC
class I proteins. Although many of the MHC class I mol-
ecules remain blocked by the viral effects, a surplus of
these molecules escapes the effects and provides effi-
cient surface presentation of viral peptides. Interest-
ingly, this effect has only been seen when fibroblasts
are preincubated with cytokines before infection and
not in fibroblasts infected before exposure to cyto-
kines11,28,29. Therefore, it is tempting to speculate that
CMVs also harbour mechanisms to interfere with the
host cell response to IFNs and TNF-a .
Second, adoptive transfer experiments have demon-
strated that the antiviral function of CD8+ T cells has a

R E V I E W S
TRENDS IN MICROBIOLOGY 193 VOL. 6 NO. 5 MAY 1998
CMV
Virion
IE/E/L
phase
IE
E genes
L
Proteasome
ER
Golgi
ERGIC
Lysosome
MHC I
MHC I
Cytosol
CD3–
TCR CD8+
CTLTAP1/2
CMV
Exo
Virion
IE/E/L
phase
IE
E genes
L
Endo
Proteasome
ER Golgi
MHC I
CD3–
TCR CD8+
CTLTAP1/2
MHC I
Pe
ne
tra
tio
n
US11, US2
UL83
US6
US3
m152
m06
m04
(a)
(b)
Fig. 2. (a) The major histocompatibility complex (MHC) class I pathway of antigen processing and presentation. Cytoplasmic viral
proteins [either endogenous proteins synthesized de novo (endo) or exogenous proteins derived from infecting virions (exo)] are
degraded by the proteasome to produce peptides (filled and unfilled circles, respectively). Peptides bind to the cytosolic face of the
peptide transporter TAP1/2 (transporter associated with antigen processing), which transports the peptides into the endoplasmic
reticulum (ER). There, the peptides bind to MHC class I– b 2-microglobulin (b 2m) heterodimers to form a trimeric complex. MHC class I
complexes exit the ER, pass through the Golgi complex and reach the cell surface to present the peptide to CD8+ cytotoxic T lympho-
cytes (CTLs). (b) Mouse cytomegalovirus (MCMV) gene functions responsible for interference with the MHC class I presentation path-
way are indicated by blue boxes. The m152/gp40 protein retains MHC class I complexes in the ER–Golgi intermediate compartment
(ERGIC). Complexes of m04/gp34–MHC class I are formed in the ER and are exposed on the cell surface for an unknown function.
The m06/gp48 protein associates with MHC– b 2m and targets the complex to the endosome/lysosome for degradation. The human
cytomegalovirus (HCMV) gene products responsible for interference with antigen processing and presentation in the MHC class I
pathway are indicated by red boxes. The UL83-encoded pp65 matrix protein prevents peptide generation from the 72-kDa IE protein.
The US6-encoded glycoprotein blocks the import of peptides into the ER. The US2- and US11-encoded glycoproteins export membrane-
inserted MHC class I heavy chains back to the cytosol for degradation by the proteasome. The gpUS3 protein blocks MHC class I
complex export from the ER. Abbreviations: E, early; IE, immediate early; L, late; CD3–TCR, CD3–T cell receptor complex.

R E V I E W S
TRENDS IN MICROBIOLOGY 194 VOL. 6 NO. 5 MAY 1998
Table 1. Cytomegalovirus genes and mechanisms that counteract host defencesa
Virus Gene Protein Phaseb Mechanism Refs
MHC class I
presentation
MCMV Inhibition of antigen presentation to CD8+ T cells 5,8
m152 gp37/40 E Retention of MHC class I complexes in the ERGIC/cis-Golgi 12
m04 gp34 E Binding to MHC class I complexes on the cell surface 13
m06 gp48 E Targeting of MHC class I complexes into the endosome/ c
lysosome for degradation
HCMV Inhibition of antigen presentation to CD8+ T cells 10,11
US3 gp32/33 IE Retention of MHC class I complexes in the ER 15,16
US11 gp33 E Dislocation of MHC class I heavy chains into the cytosol 17
US2 gp24 E Export of MHC class I heavy chains from the ER via Sec61 18
US6 gp21 E/L Inhibition of TAP-mediated peptide translocation into the ER 22–24
Antigen
processing
HCMV UL83 pp65 L Inhibition of 72-kDa IE antigen presentation to CD8+ T cells 27
NK cell
function
MCMV m144 ? ? MHC class I homologue: inhibition of NK function in vivo 47
HCMV UL18 gp69 ? MHC class I homologue: inhibition of NK function in vitro 44
Complement
HCMV ? ? ? Increased expression of CD46 and CD55 48
HCMV ? ? ? Incorporation of CD55, CD59 and CD46 into the virion 49,50
Impairment of
macrophages:
MHC class II
HCMV ? ? ? Downregulation of MHC class II surface expression in 52
macrophages
Cytokines and
growth factors
HCMV ? ? ? Induction of TGF-b 1, suppression of T-cell responses, 31,33
stimulation of CMV replication
RCMV ? ? ? Induction of TGF-b 1, suppression of T-cell responses 32
MCMV m131 ? L Expression of a b chemokine homologue: chemotactic 36
agonist or antagonist?
Receptors for
chemokines and
growth factors
HCMV US28, ? E/L b chemokine-binding proteins 34,35,
UL33 gp43 38,39
HCMV US27 ? L Potential b chemokine-binding proteins 34
HCMV UL78 ? L Putative GCR homologue 1
MCMV M33 ? E/L b chemokine-binding proteins 35
MCMV M78 ? E/L Putative GCR homologue 2
HCMV UL144 ? ? Putative TNF receptor homologue 37
Antibodies
MCMV m138 gp86/ E Expression of a viral Fc receptor, binding of murine IgG 55,57
88/105
HCMV ? ? ? Expression of a viral Fc receptor, binding of human IgG 56
Adhesion
molecules
HCMV ? ? ? Downregulation of a 1 b 1 integrin expression 54
Apoptosis
HCMV IE1/IE2 IE Inhibition of apoptosis 58
aAbbreviations: CMV, cytomegalovirus; gp, glycoprotein; E, early; ER, endoplasmic reticulum; ERGIC, endoplasmic reticulum–Golgi intermediate
compartment; GCR, G protein-coupled receptor; HCMV, human CMV; IE, immediate early; IgG, immunoglobulin G; L, late; MCMV, mouse CMV;
MHC, major histocompatibility complex; NK, natural killer; RCMV, rat CMV; TAP, transporter associated with antigen processing; TGF-b , trans-
forming growth factor b ; TNF, tumour necrosis factor; ?, unknown.
bPhase of viral gene expression.
cU. Reusch et al., submitted.

R E V I E W S
TRENDS IN MICROBIOLOGY 195 VOL. 6 NO. 5 MAY 1998
strict requirement for IFN-g (Ref. 28), suggesting that
this cytokine also regulates antigen presentation of in-
fected cells in vivo. In addition, the quantitative assess-
ment of antigenic viral peptides from MCMV-infected
organs has demonstrated the pivotal role of IFN-g for
antigen processing (i.e. the efficient generation of anti-
genic peptides from viral proteins in vivo)30.
One final line of evidence for counteraction by the
host immune system points to cell types that are resist-
ant to the subversive effects of CMVs on MHC class I.
These cells constitutively present viral peptides to CD8+
T cells, even during permissive infection. This cellular
compartment is represented by professional antigen-
presenting cells such as macrophages (H. Hengel, un-
published). The resistance of macrophages to MHC
class I subversion may, at least in part, result from
highly efficient processing of viral proteins into finally
trimmed antigenic peptides. Extraction of viral peptides
from mice during acute MCMV infection indicates that
macrophages are a primary source of MHC class I pep-
tides processed in vivo (G. Geginat, unpublished). These
features favour macrophages, and perhaps dendritic
cells, as essential inducers of the CMV-specific CD8+
T cell response (see Fig. 3).
Manipulation of cytokine responses
Evidence exists that CMVs exploit cytokine responses
to the advantage of virus replication and for the inhi-
bition of immune responses. Transforming growth fac-
tor b (TGF-b ) is a pleiotropic cytokine that has con-
siderable effects on inflammatory and immunological
responses: for example, it suppresses the generation of
T helper type 2 (Th2), CTL and NK cells, counteracts
interleukin 2 (IL-2), TNF and IL-1, and depresses poly-
clonal antibody production. Notably, CMVs use TGF-b
for several purposes: CMV infection induces TGF-b syn-
thesis in infected fibroblasts, as demonstrated in vitro
for HCMV and in vivo for rat CMV (RCMV)31,32. Con-
sequently, TGF-b simultaneously stimulates virus rep-
lication31,33 and suppresses T-cell proliferation32.
Both HCMV and MCMV express G protein-coupled
receptor (GCR) homologues34,35. In addition, MCMV
codes for a viral chemokine36 (virokine), and HCMV
may code for a TNF receptor homologue37. This indi-
cates that CMV exploits the chemokine system to the
disadvantage of the host through molecular mimicry.
Members of the GCR superfamily of receptors trans-
duce signals through the cell membrane by activating
G proteins and are triggered by chemokines (small pro-
inflammatory peptides that are best known for their
leukocyte chemoattractant activities). The known hu-
man chemokines are divided into two subfamilies (a or
CXC, and b or CC). Most a chemokines attract neutro-
phils but not monocytes, whereas b chemokines at-
tract monocytes but not neutrophils. The HCMV genes
US27, US28, UL33 and UL78, which code for GCRs,
are dispensable for virus replication in vitro and are
transcribed in the E and L phases of lytic infection35,38.
The US28-encoded GCR has been shown to bind sev-
eral b chemokines but not a chemokines, and binding
of the ligands stimulates intracellular calcium mobi-
lization39. The fact that the GCR homologues UL33 (of
HCMV) and M33 (of MCMV) possess a spliced amino
terminus that results in an improved homology with
chemokine receptors suggests that they do indeed inter-
act with chemokines35. In addition, M33 of MCMV
might bind the putative viral b chemokine36. Disrup-
tion of the MCMV M33 gene, which is conserved and
co-linear with HCMV UL33, has no influence on the
growth kinetics of fibroblasts in vitro but results in a
clearly restricted replication in vivo35. This finding pro-
vides the first evidence that viral GCR homologues func-
tion in vivo, but the understanding of their role in CMV
biology remains poor. We speculate that CMVs use
chemokines to recruit uninfected host cells to the site of
infection, and viral GCRs may alter the trafficking of
infected cells (e.g. monocytes and macrophages) and
thereby affect virus dissemination.
Interference with innate immunity: NK cells and
complement
CMVs avoid attack by CD8+ T cells by interfering with
the expression of MHC class I molecules. NK cells kill
target cells that do not express class I molecules. The
stealth strategy, therefore, should render CMV-infected
cells more susceptible to NK-mediated lysis. MCMV
and HCMV contain the ORFs m144 and UL18, re-
spectively, which encode proteins homologous to MHC
class I proteins2,40. The UL18-encoded protein binds
both b 2m and peptides41,42, whereas the MCMV class I
homologue forms a heavy chain–b 2m complex that
is devoid of endogenous peptides43. In cells highly
susceptible to NK lysis, the expression of UL18 re-
sults in strong protection against NK attack, indicating
that the viral homologue can serve as a decoy recep-
tor44. Antibodies to CD94, which is expressed on the
cell surface of most NK cells, abolish the target cell
protection mediated by UL18, suggesting that CD94
might act as a receptor for UL18 (Ref. 44). Based on
its binding affinity, UL18 has been used to identify
a novel immunoreceptor, designated leukocyte im-
munoglobulin-like receptor (LIR-1), which also binds
human MHC class I molecules and is likely to mediate
inhibitory signals45. Notably, expression of UL18 in
HCMV-infected cells has not yet been detected in
HCMV-infected cells that are susceptible to NK lysis46.
The deletion of gene m144 from the MCMV genome
is associated with reduced virus growth in vivo47.
Depletion of NK cells in mice restores the virulent
phenotype of the mutant, indicating that the expres-
sion of the class I homologue renders the virus partially
resistant to NK cells47.
The complement system is a first-line immunological
defence and includes proteolytic enzymes, inflammatory
proteins, cell surface receptors and proteins that cause
cell death through osmotic lysis after insertion into bio-
logical membranes. Antibody-activated complement-
mediated cytolysis and virolysis are general immune ef-
fector mechanisms. Accordingly, a broad range of host
cells expresses several complement-inhibiting proteins.
These protect tissues from complement-mediated lysis
by the inhibition and dissociation of C3 convertases
that are central for the amplification of the comple-
ment system. The cell surface expression of two

R E V I E W S
TRENDS IN MICROBIOLOGY 196 VOL. 6 NO. 5 MAY 1998
complement control proteins (CCPs), CD46 and
CD55, is drastically enhanced following HCMV infec-
tion48. The increased CD55 expression enhances the
capacity of infected cells to regulate C3 deposition and
thus protects infected cells from complement-medi-
ated lysis49. The incorporation of CD55, CD59 and
CD46 into CMV virions may also point to potential
escape mechanisms from host defences49,50.
Infection of macrophages
CMVs also infect macrophages. In
contrast to fibroblasts, the infection
of monocyte-derived macrophages
by HCMV results in a productive
but nonlytic infection, suggesting
that these cells represent a source
of viral amplification and dissemi-
nation51. In macrophages, HCMV
is strongly cell associated, and
infectious virus accumulates in
discrete cytoplasmic vacuoles, the
formation of which has been at-
tributed to the prolonged survival
and persistence of the virus in this
cell type. Notably, HCMV induces
several changes in macrophages,
which may promote intracellular
persistence and immune evasion.
These include the disruption of the
microtubule network, allowing the
formation of vacuoles, and a reduc-
tion in surface-expressed MHC class
II molecules52. In addition, HCMV
infection results in a selective down-
regulation of CD14 gene expres-
sion53. CD14 is a specific receptor
for lipopolysaccharide (LPS), a
major constituent of Gram-negative
bacterial cell walls, and for the LPS-
binding protein. The significance of
this finding to CMV biology is not
yet clear.
Potential mechanisms
Viral evasion from immune detec-
tion and elimination is certainly
more complex than believed in the
past and involves further mecha-
nisms (Table 1). For example: (1)
HCMV infection of fibroblasts re-
sults in a preferential loss of a 1 b 1
integrin expression on the cell sur-
face54; (2) both MCMV and HCMV
express gps that specifically bind to
the Fc part of homologous immuno-
globulin G (IgG)55,56. Deletion of
m138, the gene encoding the
MCMV Fc receptor, does not im-
pair viral replication in vitro but
results in an attenuated phenotype
in vivo. However, this phenotype
has also been observed in B-cell-
deficient mice lacking IgG. This
finding excludes the possibility that at least the pri-
mary function of this molecule is connected with anti-
body control57; (3) the HCMV IE1 and IE2 proteins
have been demonstrated to block apoptosis mediated by
TNF-a or the adenovirus E1A protein58; (4) HCMV
may encode a putative superantigen (SAg) present on
infected monocytes. SAgs are a group of microbial
proteins that are defined by their potent ability to
CD8+ CTL
CD8+ CTL CMV
Macrophage
CD8+ CTL
CD8+
CTLp
CD8+
CTLp
CMV
CMV
CMV
Fibroblast
Fibroblast
IFN-γ
IFN-α
IFN-β
TNF-β
TNF-α
Evasion
Control
Fibroblast
Fig. 3. Host factors that counteract cytomegalovirus (CMV) evasion from major histocompatibili-
ty complex (MHC) class I presentation. CMV genes prevent antigen presentation during permis-
sive infection of fibroblasts (the efficacy of MHC class I presentation in CMV-infected cells is rep-
resented by the size of MHC class I molecules shown on the cell surface). Unfilled arrows indicate
presentation resulting in lysis of CMV-infected cells by CD8+ cytotoxic T lymphocytes (CTLs). In-
terferon b (IFN- b ) secreted from infected fibroblasts, CD8+ T-cell-derived cytokines [IFN- g and tu-
mour necrosis factor b (TNF- b )] and cytokines released from CMV-infected macrophages (IFN- a ,
TNF- a ) restore antigen presentation to CD8+ T cells if they act upon fibroblasts before the cells
become infected by CMV. CMV-infected macrophages induce CD8+ T-cell precursors (CTLp) to de-
velop into CMV-specific CD8+ effector cells. Dotted arrows represent horizontal spread of CMV
virus progeny to uninfected cells.

R E V I E W S
TRENDS IN MICROBIOLOGY 197 VOL. 6 NO. 5 MAY 1998
activate T cells (CD4+ and sometimes CD8+) in a poly-
clonal manner depending on the usage of specific Vb
gene segments by the T cell receptor b chain. The
HCMV-expressed SAg selectively expands the Vb 12 T
cell receptor-bearing lymphocyte subset59. Alterna-
tively, this SAg could be a human gene product, the
expression of which is transcriptionally activated by
HCMV expression. What effect would the polyclonal
activation of CD4+ T cells have? One possible answer
comes from recently published work showing that
HCMV can be reactivated from latently infected
monocyte-derived macrophages by allogeneic stimu-
lation, i.e. after activation of T cells60. This implies
that CMVs are able to use T cell–macrophage inter-
actions and/or soluble T-cell-derived factors for the
induction of recurrent infection from latency.
Conclusions and perspective
For decades, studies on the immune biology of CMVs
have described the antiviral immune responses of the
host. Recent work, however, shows that CMVs are not
merely conventional antigens but represent a paradigm
for persistent viruses that actively manipulate and ex-
ploit several aspects of the immune response. The iden-
tification and analysis of individual viral gene functions
have identified the targets for modulation and eluci-
dated the specific immune control mechanisms involved
(work that is still far from being completed). These stud-
ies need to be complemented by the analysis of their
biological relevance in vivo during the life cycle of CMV,
i.e. in primary, latent and recurrent infection. To date,
the potential interactions of these immune-subversive
viral genes and their consequences for immune con-
trol are unknown. For example, the downregulation of
MHC class I should affect the susceptibility of CMVs
to NK effector mechanisms. The number of genes af-
fecting immunity that have been identified so far prob-
ably represent the tip of an iceberg. However, we have
stumbled on them either because of their sequence hom-
ology or because they have a clear functional pheno-
type. The coevolution of CMVs and their hosts over
millions of years must have selected the regulatory fac-
tors that modulate immune and non-immune functions
of cells and promote virus survival. Herpesviruses are
promising sources for the search for highly adapted
molecular regulators of various cellular functions.
Acknowledgements
This work was supported by the Deutsche Forschungsgemeinschaft
through SFB 352, project C8 and HE 2526/3-1.
References
1 Chee, M.S. et al. (1990) Curr. Top. Microbiol. Immunol. 154,
125–169
2 Rawlinson, W.D. et al. (1996) J. Virol. 70, 8833–8849
3 Reddehase, M. et al. (1985) J. Virol. 55, 264–273
4 Riddell, S.R. et al. (1992) Science 257, 238–241
5 Del Val, M. et al. (1989) Cell 58, 305–315
6 Sadasivan, B. et al. (1996) Immunity 5, 103–114
7 Williams, D.B. et al. (1996) Trends Cell Biol. 6, 267–273
8 Del Val, M. et al. (1992) J. Exp. Med. 176, 729–738
9 Campbell, A.E. et al. (1992) J. Virol. 66, 3011–3017
10 Warren, A.P. et al. (1994) J. Virol. 68, 2822–2829
11 Hengel, H. et al. (1995) J. Gen. Virol. 76, 2987–2997
12 Ziegler, H. et al. (1997) Immunity 6, 57–66
13 Kleijnen, M. et al. (1997) EMBO J. 16, 685–694
14 Jones, T.R. et al. (1995) J. Virol. 69, 4830–4841
15 Ahn, K. et al. (1997) Proc. Natl. Acad. Sci. U. S. A. 93,
10990–10995
16 Jones, T.R. et al. (1997) Proc. Natl. Acad. Sci. U. S. A. 93,
11327–11333
17 Wiertz, E.J.H.J. et al. (1996) Cell 84, 769–779
18 Wiertz, E.J.H.J. et al. (1996) Nature 384, 432–438
19 Jones, T.R. and Sun, L. (1997) J. Virol. 71, 2970–2979
20 Plemper, R. et al. (1997) Nature 388, 891–895
21 Hengel, H. et al. (1996) J. Gen. Virol. 77, 2287–2296
22 Hengel, H. et al. (1997) Immunity 6, 623–632
23 Ahn, K. et al. (1997) Immunity 6, 613–621
24 Lehner, P.L. et al. (1997) Proc. Natl. Acad. Sci. U. S. A. 94,
6904–6909
25 Jones, T.R. and Muzithras, V.P. (1991) J. Virol. 65, 2024–2036
26 Borysiewicz, L.K. et al. (1988) J. Exp. Med. 168, 919–931
27 Gilbert, M.J. et al. (1996) Nature 383, 720–722
28 Hengel, H. et al. (1994) J. Virol. 68, 289–297
29 Lucin, P. et al. (1994) J. Gen. Virol. 75, 101–110
30 Geginat, G. et al. (1997) J. Immunol. 158, 3303–3310
31 Michelson, S. et al. (1994) J. Virol. 68, 5730–5737
32 Haagmans, B.L. et al. (1997) J. Gen. Virol. 78, 205–213
33 Alcami, J. et al. (1991) J. Gen. Virol. 72, 2765–2770
34 Chee, M.S. et al. (1990) Nature 344, 774–777
35 Davis-Poynter, N.J. et al. (1997) J. Virol. 71, 1521–1529
36 MacDonald, M.R. et al. (1997) J. Virol. 71, 1671–1678
37 Cha, T-A. et al. (1996) J. Virol. 70, 78–83
38 Margulies, B.M. et al. (1996) Virology 225, 111–125
39 Neote, K. et al. (1993) Cell 72, 415–425
40 Beck, S. and Barrell, B.G. (1988) Nature 331, 269–272
41 Browne, H. et al. (1990) Nature 347, 770–772
42 Fahnestock, M.L. et al. (1995) Immunity 3, 583–590
43 Chapman, T.L. and Bjorkman, P.J. (1998) J. Virol. 72, 460–462
44 Reyburn, H.T. et al. (1997) Nature 386, 514–517
45 Cosman, D. et al. (1997) Immunity 7, 273–282
46 Borysiewicz, L.K. et al. (1985) J. Immunol. 134, 2695–2701
47 Farrell, H.E. et al. (1997) Nature 386, 510–514
48 Spiller, B.O. et al. (1996) Eur. J. Immunol. 26, 1532–1538
49 Spear, G.T. et al. (1995) J. Immunol. 155, 4376–4381
50 Spiller, B.O. et al. (1997) J. Infect. Dis. 176, 339–347
51 Ibanez, E.C. et al. (1991) J. Virol. 65, 6581–6588
52 Fish, K.N. et al. (1996) J. Virol. 70, 1855–1862
53 Hopkins, H.A. et al. (1996) J. Infect. Dis. 174, 69–74
54 Warren, A.P. et al. (1994) J. Gen. Virol. 75, 3319–3325
55 Thäle, R. et al. (1994) J. Virol. 68, 7757–7765
56 Mac Cormac, L.P. and Grundy, J.G. (1996) J. Infect. Dis. 174,
1151–1161
57 Crnkovic-Mertens, I. et al. (1998) J. Virol. 72, 1377–1382
58 Zhu, H. et al. (1995) J. Virol. 69, 7960–7970
59 Dobrescu, D. et al. (1995) Cell 82, 753–763
60 Söderberg-Naucler, C. et al. (1997) Cell 91, 119–126
Questions for future research
• What is the genetic basis for the regulatory phenotypes observed
in cytomegalovirus (CMV)-infected cells (for example, Refs 48,
52,54)? Which phenomenon is caused on the transcriptional level,
and what is post-transcriptionally regulated?
• How do major histocompatibility complex (MHC) class I subversive
CMV genes interact? When are these genes expressed during
the viral life cycle in vivo?
• Which human CMV (HCMV) gene encodes a superantigen59?
• Do CMV genes that manipulate interferon responses exist?
• Are there cell-type-specific immune evasion functions?