Stage-specific inhibition of MHC class I presentation by the Epstein-Barr virus BNLF2a protein during virus lytic cycle.

Abstract

The gamma-herpesvirus Epstein-Barr virus (EBV) persists for life in infected individuals despite the presence of a strong immune response. During the lytic cycle of EBV many viral proteins are expressed, potentially allowing virally infected cells to be recognized and eliminated by CD8+ T cells. We have recently identified an immune evasion protein encoded by EBV, BNLF2a, which is expressed in early phase lytic replication and inhibits peptide- and ATP-binding functions of the transporter associated with antigen processing. Ectopic expression of BNLF2a causes decreased surface MHC class I expression and inhibits the presentation of indicator antigens to CD8+ T cells. Here we sought to examine the influence of BNLF2a when expressed naturally during EBV lytic replication. We generated a BNLF2a-deleted recombinant EBV (DeltaBNLF2a) and compared the ability of DeltaBNLF2a and wild-type EBV-transformed B cell lines to be recognized by CD8+ T cell clones specific for EBV-encoded immediate early, early and late lytic antigens. Epitopes derived from immediate early and early expressed proteins were better recognized when presented by DeltaBNLF2a transformed cells compared to wild-type virus transformants. However, recognition of late antigens by CD8+ T cells remained equally poor when presented by both wild-type and DeltaBNLF2a cell targets. Analysis of BNLF2a and target protein expression kinetics showed that although BNLF2a is expressed during early phase replication, it is expressed at a time when there is an upregulation of immediate early proteins and initiation of early protein synthesis. Interestingly, BNLF2a protein expression was found to be lost by late lytic cycle yet DeltaBNLF2a-transformed cells in late stage replication downregulated surface MHC class I to a similar extent as wild-type EBV-transformed cells. These data show that BNLF2a-mediated expression is stage-specific, affecting presentation of immediate early and early proteins, and that other evasion mechanisms operate later in the lytic cycle.

Full text

Page 1

Stage-Specific Inhibition of MHC Class I Presentation by
the Epstein-Barr Virus BNLF2a Protein during Virus Lytic
Cycle
Nathan P. Croft1, Claire Shannon-Lowe1, Andrew I. Bell1, Danie ¨lle Horst2, Elisabeth Kremmer3, Maaike E.
Ressing2, Emmanuel J. H. J. Wiertz4, Jaap M. Middeldorp5, Martin Rowe1, Alan B. Rickinson1, Andrew D.
Hislop1*
1School of Cancer Sciences, University of Birmingham, Edgbaston, Birmingham, United Kingdom, 2Department of Medical Microbiology, Leiden University Medical
Center, Leiden, The Netherlands, 3Institute of Molecular Immunology, Helmholtz Zentrum Mu ¨nchen, Mu ¨nchen, Germany, 4Department of Medical Microbiology,
University Medical Centre Utrecht, Utrecht, The Netherlands, 5Department of Pathology, VU University Medical Centre, Amsterdam, The Netherlands
Abstract
The gamma-herpesvirus Epstein-Barr virus (EBV) persists for life in infected individuals despite the presence of a strong
immune response. During the lytic cycle of EBV many viral proteins are expressed, potentially allowing virally infected cells
to be recognized and eliminated by CD8+T cells. We have recently identified an immune evasion protein encoded by EBV,
BNLF2a, which is expressed in early phase lytic replication and inhibits peptide- and ATP-binding functions of the
transporter associated with antigen processing. Ectopic expression of BNLF2a causes decreased surface MHC class I
expression and inhibits the presentation of indicator antigens to CD8+T cells. Here we sought to examine the influence of
BNLF2a when expressed naturally during EBV lytic replication. We generated a BNLF2a-deleted recombinant EBV (DBNLF2a)
and compared the ability of DBNLF2a and wild-type EBV-transformed B cell lines to be recognized by CD8+T cell clones
specific for EBV-encoded immediate early, early and late lytic antigens. Epitopes derived from immediate early and early
expressed proteins were better recognized when presented by DBNLF2a transformed cells compared to wild-type virus
transformants. However, recognition of late antigens by CD8+T cells remained equally poor when presented by both wild-
type and DBNLF2a cell targets. Analysis of BNLF2a and target protein expression kinetics showed that although BNLF2a is
expressed during early phase replication, it is expressed at a time when there is an upregulation of immediate early proteins
and initiation of early protein synthesis. Interestingly, BNLF2a protein expression was found to be lost by late lytic cycle yet
DBNLF2a-transformed cells in late stage replication downregulated surface MHC class I to a similar extent as wild-type EBV-
transformed cells. These data show that BNLF2a-mediated expression is stage-specific, affecting presentation of immediate
early and early proteins, and that other evasion mechanisms operate later in the lytic cycle.
Citation: Croft NP, Shannon-Lowe C, Bell AI, Horst D, Kremmer E, et al. (2009) Stage-Specific Inhibition of MHC Class I Presentation by the Epstein-Barr Virus
BNLF2a Protein during Virus Lytic Cycle. PLoS Pathog 5(6): e1000490. doi:10.1371/journal.ppat.1000490
Editor: Bill Sugden, University of Wisconsin-Madison, United States of America
Received January 7, 2009; Accepted May 27, 2009; Published June 26, 2009
Copyright: ? 2009 Croft et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by a grant from the Medical Research Council (G9901249). ADH is funded by a Medical Research Council UK New Investigator
Award (G0501074); ADH and MR are supported by the Wellcome Trust. Additional support for DH, MER and EJHJW was from the Dutch Cancer Society (UL 2005-
3259), the M.W. Beijerinck Virology Fund of the Royal Academy of Arts and Sciences, and the Netherlands Organisation for Scientific Research (Vidi 917.76.330).
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: hislopad@adf.bham.ac.uk
Introduction
The detection and elimination of virally infected cells by the
host immune system relies heavily upon CD8+T cells recognizing
peptides endogenously processed and presented by HLA class I
molecules. Proteasomal degradation of endogenously synthesized
proteins provides a source of peptides which are delivered into the
endoplasmic reticulum by the transporter associated with antigen
processing (TAP), where they are loaded onto nascent HLA-class I
molecules. Peptide:HLA-class I complexes are then transported to
the cell surface where CD8+T cells examine these complexes with
their T cell receptors. Recognition of these complexes leads to the
killing of the infected cell by the CD8+T cell (reviewed in [1,2]).
As such, many viruses have developed strategies to evade CD8+
T cell recognition in order to aid their transmission and
persistence within hosts. This is particularly true for the
herpesviruses; large double-stranded DNA viruses characterized
by their ability to enter a latent state within specialized cells in
their respective hosts, with this itself a form of immune evasion due
to the transcriptional silencing of most if not all genes. However,
herpesviruses occasionally undergo reactivation into their lytic
cycle, where a large number of viral genes are expressed. Here
there is a sequential cascade of gene expression beginning with the
immediate early genes, followed by the early genes and finally the
late genes. Potentially then many targets for CD8+T cell
recognition are generated during lytic cycle replication. The
finding of immune evasion mechanisms in members of each of the
three a-, b- and c-herpesvirus subfamilies highlights the strong
immunological pressure these viruses are under. These evasion
strategies often subvert cellular processes involved in the
generation and presentation of epitopes to T cells (reviewed in
[3,4]). The importance of these processes is highlighted by the
PLoS Pathogens | www.plospathogens.org1 June 2009 | Volume 5 | Issue 6 | e1000490

Page 2

convergent evolution seen in herpesviruses, where members of the
different subfamilies target the same points involved in the
generation of CD8+T cell epitopes but use unrelated proteins to
do this.
Until recently, less evidence has been available on immune
evasion by the lymphocryptoviruses (LCV, c1-herpesviruses)
during lytic cycle. The prototypic virus of this genus, Epstein-
Barr virus (EBV), infects epithelial cells and B lymphocytes,
establishing latency in the latter cell type. Central to EBV’s biology
is its ability to expand the reservoir of latently infected B cells
through growth-transforming gene expression, independent of
lytic replication [5]. It was unclear then whether lytic immune
evasion mechanisms would be required by EBV to amplify the
viral reservoir within a host. However, during lytic cycle
replication, presentation of EBV epitopes to cognate CD8+T
cells falls with the progression of the lytic cycle, while B cells
replicating EBV have decreased levels of surface HLA-class I and
decreased TAP function [6–8]. These observations suggested that
EBV interferes with antigen processing during lytic cycle
replication. Targeted screening of EBV genes for immune evasion
function led to the identification of the early expressed lytic cycle
gene BNLF2a which functions as a TAP inhibitor [9]. This novel
immune evasion gene encodes for a 60 amino acid protein that
disrupts TAP function by preventing both peptide- and ATP-
binding to this complex. Consequently, cells expressing BNLF2a in
vitro show decreased surface HLA-class I levels and are refractory
to CD8+T cell killing when co-expressed with target antigens [9].
In the current study we analyze the influence BNLF2a has on
presentation of EBV-specific epitopes during lytic cycle replica-
tion, to determine whether BNLF2a acts alone or whether other
immune evasion mechanisms are present in EBV and how
BNLF2a affects antigen presentation during the different phases of
gene expression. The impact of BNLF2a was isolated through the
construction of a recombinant EBV lacking the gene and this virus
used to infect cells for antigen processing and presentation studies.
Cells replicating this BNLF2a-deleted virus were found to be better
recognized by immediate early and early antigen-specific CD8+T
cells but not late antigen-specific T cells. Consistent with this
finding, surface class I HLA expression was restored to normal
levels in cells expressing immediate early but not late expressed
EBV proteins. Our results suggest that immune evasion mecha-
nisms in addition to BNLF2a are operational during EBV lytic
cycle replication.
Results
Construction of a DBNLF2a mutant virus
We initially disrupted the BNLF2a gene of the B95.8 strain of
EBV contained within a BAC by insertional mutagenesis
(Figure 1A). A targeting plasmid was created in which the
majority of the BNLF2a gene was replaced with a tetracycline
resistance cassette which in turn was flanked by FLP recombinase
target (FRT) sites. This vector was recombined with the EBV BAC
and recombinants selected. Such clones, designated DBNLF2a,
had the tetracycline gene removed by FLP recombinase and were
screened for deletion of the BNLF2a gene by restriction
endonuclease analysis and sequencing (data not shown). DBNLF2a
BACs were then stably transfected into 293 cells and virus
replication induced by transfection of a plasmid encoding the EBV
lytic switch protein BZLF1. Virus was also produced from cells
transduced with the wild-type B95.8 EBV BAC and a B95.8 EBV
BZLF1-deleted BAC (DBZLF1) [10], encoding a virus unable to
undergo lytic cycle replication unless BZLF1 is supplied in trans.
The different recombinant EBVs derived from the 293 cells
were used to transform primary B cells, to establish lymphoblas-
toid cell lines (LCLs). To determine if expression of other viral
proteins was affected by the deletion of BNLF2a, western blot
analysis on lysates of LCLs generated from wild-type, DBNLF2a
and DBZLF1 viruses was performed. As a subset of cells in the
LCL culture will spontaneously enter lytic cycle replication, blots
were probed with antibodies specific for representative proteins
expressed during lytic cycle as well as latent cycle expressed
proteins. Figure 1B shows typical blots of lysates probed for the
immediate early proteins BZLF1 and BRLF1, the early proteins
BALF2, BNLF2a and BMRF1, the late protein BFRF3 and the
latent protein EBNA2. No difference in expression of these
proteins was observed between the wild-type and DBNLF2a virus
transformed LCLs, with the exception of BNLF2a protein which
was not present as expected in DBNLF2a LCLs. No lytic cycle
protein expression could be detected in DBZLF1 LCLs.
Deletion of BNLF2a confers an increase in immediate
early and early antigen recognition by cognate CD8+T
cells, but has no effect on late antigen recognition
A panel of different donor derived LCLs transformed with wild-
type, DBNLF2a, and DBZLF1 viruses were employed to study
lytic antigen recognition by EBV lytic phase-specific CD8+T cells.
Here we planned to incubate these LCLs with the different types
of lytic antigen-specific CD8+T cells and assay for T cell
recognition by IFN-c secretion. However, the percentage of LCLs
that spontaneously enter lytic cycle is variable. Initially then we
quantified the number of cells within the LCL cultures expressing
the lytic cycle marker BZLF1 by flow cytometry. Figures 2A and
2B show representative flow plots of wild-type, DBNLF2a and
DBZLF1 LCLs stained for BZLF1 expression using LCLs derived
from two donors. Typically we found between 0.5–3% of wild-type
and DBNLF2a LCLs expressed BZLF1 (upper and middle panels),
whilst none was observed in DBZLF1 LCLs (lower panels).
To ensure we used equivalent numbers of the different types of
lytic antigen positive cells in our T cell recognition experiments,
we developed a system to equalize the number of lytic antigen
positive cells in each assay. Here the proportion of BZLF1
expressing cells in each culture were equalized by making a
dilution series of the LCL with the highest percentage of BZLF1
Author Summary
Epstein-Barr virus (EBV) is carried by approximately 90% of
the world’s population, where it persists and is chronically
shed despite a vigorous specific immune response, a key
component of which are CD8+T cells that recognize and
kill infected cells. The mechanisms the virus uses to evade
these responses are not clear. Recently we identified a
gene encoded by EBV, BNLF2a, that when expressed
ectopically in cells inhibited their recognition by CD8+T
cells. To determine the contribution of BNLF2a to evasion
of EBV-specific CD8+T cell recognition and whether EBV
encoded additional immune evasion mechanisms, a
recombinant EBV was constructed in which BNLF2a was
deleted. We found that cells infected with the recombinant
virus were better recognized by CD8+T cells specific for
targets expressed co-incidently with BNLF2a, compared to
cells infected with a non-recombinant virus. However,
proteins expressed at late stages of the viral infection cycle
were poorly recognised by CD8+T cells, suggesting EBV
encodes additional immune evasion genes to prevent
effective CD8+T cell recognition. This study highlights the
stage-specific nature of viral immune evasion mechanisms.
EBV CD8+T Cell Evasion
PLoS Pathogens | www.plospathogens.org2June 2009 | Volume 5 | Issue 6 | e1000490

Page 3

expressing cells with the antigen negative DBZLF1 LCL derived
from that donor. T cell recognition of the different LCL
transformants was then measured by incubating these LCLs with
CD8+T cells specific for epitopes derived from proteins expressed
in immediate early, early and late phases of the EBV lytic cycle
and measuring IFNc release by the T cells. We have previously
shown that CD8+T cells in these assays directly recognize lytically
infected cells and not cells which have exogenously taken up
antigen and re-presented it [6]. Figure 2C shows results of a T cell
recognition experiment using LCL targets derived from donor 1.
In this case the more lytic wild-type LCL was diluted with the
DBZLF1 LCL to give equivalent numbers of lytic targets in the
assay. When CD8+T cells specific for the immediate early HLA-
B*0801 restricted BZLF1 RAK epitope were incubated with the
different LCLs, a 6-fold increase in recognition of the DBNLF2a
LCL was observed compared to the wild-type LCL as measured
by secretion of IFNc. Similar results were obtained using LCLs
derived from donor 2 (Figure 2D). In this case the more lytic
DBNLF2a LCL was diluted with the antigen-negative DBZLF1
LCL. When the cultures were equalized for BZLF1 expression a 3-
fold increase in recognition of the DBNLF2a LCL was seen when
compared to recognition of the wild-type LCL.
A similar trend was observed for recognition of epitopes derived
from the other immediate early protein BRLF1. Here CD8+T
cells specific for the HLA-C*0202 restricted epitope IACP
(Figure 2E) and the HLA-B*4501 restricted epitope AEN
(Figure 2F) were used to probe antigen presentation by the LCL
sets derived from donors 3 and 4 respectively. As shown in
Figures 2E and 2F, the DBNLF2a LCLs from both donors were
recognized more efficiently than the wild-type LCL using both T
cell specificities. The IACP clones showed a 50-fold increase and
the AEN clones showed a 4–5-fold increase in IFNc secretion
upon challenge with the LCLs.
We next measured recognition of the different LCL types using
CD8+T cells specific for two early antigens; the HLA-B*2705
restricted ARYA epitope from BALF2 and the HLA-A*0201
restricted TLD epitope from BMRF1. Here we tested multiple T
cell clones derived from three donors against three different donor
derived sets of LCLs. Figure 3 shows representative results using
ARYA- and TLD-specific T cell clones against LCLs derived from
donor 3. Similar to what was seen for the immediate early antigens,
Tcellrecognitionoftheearlyantigenswasincreaseduponchallenge
with the DBNLF2a LCL compared to the wild-type LCL, with the
most potent increase in recognition observed using the BALF2-
specific clones which showed a 20-fold increase in recognition
(Figure 3A). The TLD epitope from BMRF1 was found to be
recognized the poorest in these assays, never the less a two-fold
increase in recognition of the DBNLF2aLCL compared to the wild-
type was consistently observed using independently derived T cell
clonesandLCLsderivedfromdifferentdonors(Figure3B).Multiple
clonesofathirdearlyspecificity,HLA-A*0201BMLF1,alsoshowed
increased recognition of the DBNLF2a LCL (see below).
We next turned to study recognition of late-expressed antigens
using T cells specific for the HLA-A*0201 restricted FLD epitope
from BALF4 and the HLA-B*2705 restricted RRRK epitope from
BILF2. We have found that these two epitopes are processed
independently and dependently of the proteasome respectively,
with the BALF4 epitope presented independently of TAP (data not
shown). We would predict from our previous studies of TAP
Figure 1. Generation of a mutant Epstein-Barr virus deleted for
BNLF2a (DBNLF2a). (A) Schematic drawing of the BNLF2a-containing
region of the EBV genome, before and after disruption of the BNLF2a
open reading frame. Removal of the tetracycline resistance cassette by
flp recombinase leaves one flp recombinase target (FRT) site intact. (B)
LCLs transformed with either the wild-type (wt), DBNLF2a (D2a) or
DBZLF1 (DBZ) viruses were analysed by Western blot for expression of
BNLF2a, several representative lytic cycle antigens, and the latent cycle
expressed protein EBNA2. Antibodies specific for b-actin were used to
ensure equal protein loading. Lat, latent; IE, immediate early; E, early; L,
late.
doi:10.1371/journal.ppat.1000490.g001
EBV CD8+T Cell Evasion
PLoS Pathogens | www.plospathogens.org3 June 2009 | Volume 5 | Issue 6 | e1000490

Page 4

dependence of peptide-epitopes that the hydrophilic BILF2
peptide RRRK would be processed in a TAP dependent manner
[11]. Figures 4A and 4B show representative results of experiments
using two FLD-specific clones and one RRRK-specific clone
assayed against two different donor derived LCLs. T cell
recognition of late-expressing DBNLF2a and wild-type LCLs
was found to be low but of an equivalent level. This pattern of
recognition was seen using LCL sets derived from three other
donors (data not shown).
To confirm the above results and minimize any variability
between assays, we tested the recognition of the different LCL types
in parallel by CD8+T cell clones specific for epitopes that were
presented by the same HLA molecule but produced at different
phases in the replication cycle. Initially we compared recognition of
thedonor1setofLCLsbytheHLA-A*0201restrictedCD8+T cells
specific for the YVL epitope from the immediate early protein
BRLF1, the GLC epitope derived from the early expressed protein
BMLF1 and the FLD epitope from the late expressed BALF4
protein. In LCLs made with the BNLF2a-deleted virus there was a
clear increase in the ability of YVL- and GLC-specific CD8+T cells
to recognize these targets in comparison to the wild-type LCLs,with
these specificities showing a 20- and 6-fold increase in IFN-c
secretion respectively (Figure 5A left panels). We also checked
recognition in parallel with the HLA-A*0201 restricted TLD-
specific clones which showed an increase in recognition similar to
what we observed above (data not shown). By contrast, no apparent
difference in recognition was observed using the CD8+T cells
specific for the late-derived FLD epitope. In parallel we also
estimated the functional avidity of these T cell clones by IFNc
secretion in response to DBZLF1 LCLs loaded with 10-fold
dilutions of epitope peptide (Figure 5A right panels). The 50%
optimal recognition of the late effector FLD c21 was similar to that
of the immediate early effector YVL c10, both being in the 1028–
1029M range of peptide avidity, whilst the early effector GLC c10
was less avid with a 50% optimal recognition of 1026M.
In a second series of experiments we compared the ability of the
donor 3 set of LCLs to be recognized by HLA-B*2705 restricted
CD8+T cells. Here we used clones specific for the ARYA epitope
derived from the early protein BALF2 and the RRRK epitope
derived from the late protein BILF2. Again we found that the
LCLs made using the BNLF2a-deleted virus were well recognized
by the early antigen-specific effector compared to the wild-type
transformed LCLs with a 14-fold increase in recognition (Figure 5B
left panels), but both LCL types were recognized at an equivalent
low level by the late-specific cells. In peptide titration assays the
50% optimal CD8+T cell recognition values for the ARYA and
RRRK clones were similar, at 461027and 261027respectively
(Figure 5B right panels).
Figure 2. Estimation of DBNLF2a and wild-type LCLs express-
ing lytic antigens; recognition by immediate early antigen-
specific CD8+T cells. The proportion of LCLs spontaneously
reactivating into lytic cycle was assessed by intracellular BZLF1 staining
and analysis by flow cytometry, with representative examples shown for
LCLs derived from two different donors: (A) donor 1 and (B) donor 2.
Immediate early lytic cycle CD8+T cell recognition of wild-type (wt),
DBNLF2a (D2a) and DBZLF1 (DBZ) LCLs using HLA-B*0801-restricted
RAK (BZLF1) clones against appropriately HLA matched donor 1 and 2
LCLs (C and D respectively) was measured by IFNc ELISA. Results using
wild-type or DBNLF2a cells diluted with DBZLF1 cells as appropriate are
shown, where arrows indicate equivalent numbers of lytic antigen
expressing cells. Experiments were also conducted using HLA-C*0202-
restricted IACP (BRLF1) clones against donor 3 LCLs (E), and HLA-
B*4501-restricted AEN (BRLF1) clones against donor 4 LCLs (F). For
donor 4, both the wild-type and DBNLF2a LCLs were diluted with
DBZLF1 LCL (wild-type-LCL titration data not shown). Data are
represented as mean+/2SEM.
doi:10.1371/journal.ppat.1000490.g002
EBV CD8+T Cell Evasion
PLoS Pathogens | www.plospathogens.org4June 2009 | Volume 5 | Issue 6 | e1000490

Page 5

To confirm that the increased recognition of the DBNLF2a LCLs
by the immediate early and early T cells seen in these experiments
was due to the absence of BNLF2a and not to a secondary mutation
within the DBNLF2a virus, we re-expressed BNLF2a in the
DBNLF2a LCLs and conducted recognition assays on these cells.
DBNLF2a LCLs were transfected with a BNLF2a expression vector
which co-expressed the truncated nerve growth factor receptor
(NGFR) and cells expressing this receptor selected with magnetic
beads. These BNLF2a expressing cells and were used as targets in
standard recognition assays alongside NGFR negative BNLF2a
negative cells from the transfection, wild-type LCLs,unmanipulated
DBNLF2a LCLs and DBZLF1 LCLs. T cells specific for the
immediate early epitope AEN and early epitope ARYA were used
as effectors in parallel assays. Figure S1 shows representative results
of two independent transfection experiments. For both CD8+T cell
clones, re-expression of BNLF2a in the DBNLF2a LCLs decreased
recognition of these LCLs to low levels relative to the unmanipu-
lated DBNLF2a LCL, suggesting the increased recognition of the
DBNLF2a LCLs observed in the previous experiments is due to the
absence of BNLF2a.
EBV BNLF2a is expressed during lytic cycle concomitant
with peak immediate early and early gene expression
An unexpected outcome of the recognition experiments was the
increased detection of immediate early antigens in the DBNLF2a
Figure 3. Recognition of DBNLF2a LCLs and wild-type LCLs by early antigen-specific CD8+T cells. LCLs from donor 3 were measured for
lytic antigen expression and the percentage positive indicated. The proportion of lytic antigen positive wild-type (wt) and DBNLF2a (D2a) cells were
equalised by dilution with DBZLF1 (DBZ) LCL and recognition assays performed as described in Figure 2. Recognition of early lytic antigen targets was
assessed using CD8+T cells specific for the HLA-B*2705-restricted ARYA (BALF2) epitope (A) and the HLA-A*0201-restricted TLD (BMRF1) epitope (B).
Arrows indicate equivalent numbers of lytic antigen expressing cells. Data are represented as mean+/2SEM.
doi:10.1371/journal.ppat.1000490.g003
EBV CD8+T Cell Evasion
PLoS Pathogens | www.plospathogens.org5 June 2009 | Volume 5 | Issue 6 | e1000490

Page 6

transformed LCLs by the cognate CD8+T cells. Immediate early
genes are expressed prior to when the early gene BNLF2a would be
expected to be expressed and so epitopes derived from immediate
early proteins would not likely be well protected from presentation
to CD8+T cells. To clarify when BNLF2a is transcribed and
expressed relative to the other genes of interest, we studied the
transcription and protein expression kinetics of this gene and
others that were used in our T cell recognition assays by qRT-
PCR and western blot analysis during lytic replication. Here we
used the EBV-infected AKBM cell line in which lytic EBV
replication can be induced by cross-linking surface IgG receptors
with anti-IgG antibodies [8] as a source of RNA and protein for
analysis.
Following induction of EBV replication in the AKBM cells,
RNA samples were harvested over 48 hours post-induction (pi).
qRT-PCR analysis was conducted on the two immediate early
genes (BZLF1 and BRLF1), two representative early genes (BMLF1
and BNLF2a) and two representative late genes (BLLF1 (encoding
gp350) and BALF4 (encoding gp110)). Upon induction, immediate
early gene expression (BZLF1 and BRLF1) occurred very rapidly
with an increase in transcripts observed 1 hr pi, followed by peak
expression at 2–3 hours pi (Figure 6A, upper panel). Transcripts
for these two immediate early genes did not disappear completely
after their peak expression, however BZLF1 decreased quickly to
low levels consistent with previous findings [12]. There were still
more than 40% of the maximal BRLF1 transcripts present
Figure 4. Recognition of DBNLF2a LCLs and wild-type LCLs by late antigen-specific CD8+T cells. LCLs from donors 3 and 5 were
measured for lytic antigen expression and the percentage positive indicated. The proportion of lytic antigen positive wild-type (wt) and DBNLF2a
(D2a) cells were equalised by dilution with DBZLF1 (DBZ) LCL and recognition assays performed as described in Figure 2. Recognition of late lytic
antigen targets was assessed using CD8+T cells specific for the HLA-A*0201-restricted FLD (BALF4) epitope (A) and the HLA-B*2705-restricted RRRK
(BILF2) epitope (B). Arrows indicate equivalent numbers of lytic antigen expressing cells. Data are represented as mean+/2SEM.
doi:10.1371/journal.ppat.1000490.g004
EBV CD8+T Cell Evasion
PLoS Pathogens | www.plospathogens.org6 June 2009 | Volume 5 | Issue 6 | e1000490

Page 7

24 hours pi compared to only 5% of the maximal BZLF1
transcripts at the same time point. Early gene message was
expressed rapidly after induction with both BMLF1 and BNLF2a
reaching their peak expression at 4 hours pi (Figure 6A, middle
panel). However, BMLF1 message decreased quickly over the next
8 hours almost to its final levels, while high relative levels of
BNLF2a message were maintained over the next 20 hours from
peak expression dropping to 40% of the maximal level by 48 hours
pi. As expected, induction of the late gene BALF4 and BLLF1
transcripts was slower, with peak expression at 12 hours and
24 hours, respectively (Figure 6A, lower panel).
We next turned to examine the protein expression kinetics in
lytically induced AKBM cells by western blot analysis, employing
antibodies specific to proteins used in our recognition assays where
available (Figure 6B). Protein from each of the genes that had been
measured by qRT-PCR was detected shortly following the
expression of the corresponding transcript. Thus BZLF1, BRLF1
and BMLF1 protein were clearly detected at 2 hours pi as was
another early protein BALF2. BNLF2a protein was also weakly
detected at this point and clearly detected at 3 hours pi. BMRF1
showed delayed protein expression kinetics, being detected at 3–
4 hours pi. Expression of the protein levels remained mostly stable
for the duration of the time course, with the exception of BNLF2a
which was lost from the cells at 12–48 hours pi. The late protein
BALF4 was expressed by 6 hours and increased with time, while a
second representative late protein, BFRF3, showed much delayed
expression kinetics.
Surface HLA class I levels remain unaltered in the
immediate early/early phases of lytic cycle in DBNLF2a
LCLs, yet are downmodulated during late lytic cycle
The results from our recognition experiments indicated that
the deletion of BNLF2a did not lead to any increase in
recognition of late antigens by their cognate CD8+T cells.
Interestingly these late proteins were expressed when protein
levels of BNLF2a were declining to low levels. Potentially other
immune evasion proteins may be active at these later time
points, preventing efficient presentation of epitopes to CD8+T
cells. To explore this possibility we performed flow cytometric
Figure 5. Comparative CD8+T cell recognition of immediate early, early and late antigens expressed by DBNLF2a versus wild-type
LCLs. (A) LCLs from donor 1 were measured for lytic antigen expression and the percentage positive indicated. The proportion of lytic antigen
positive wild-type (wt) and DBNLF2a (D2a) cells were equalised by dilution with DBZLF1 (DBZ) LCL and recognition assays performed as described in
Figure 2. Recognition of immediate early (IE), early (E) and late (L) lytic antigen targets was assessed in parallel using representative CD8+T cells
specific for the HLA-A*0201 restricted epitopes YVL (BRLF1), GLC (BMLF1) and FLD (BALF4) (left panels). Simultaneously, the functional avidity of these
clones was measured by challenging the CD8+T cells with DBZLF1 LCLs sensitized with 10-fold dilutions of the peptide epitope and the dose of
peptide giving 50% maximal recognition determined (dashed line, right panels). (B) LCLs from donor 3 were measured for lytic antigen expression
and the percentage positive indicated. The proportion of lytic antigen positive cells were equalised by dilution with DBZLF1 LCL and recognition
assays performed as described in Figure 2. Recognition of early and late lytic antigen targets was assessed in parallel using representative CD8+T cells
specific for the HLA-B*2705 restricted epitopes ARYA (BALF2), and RRRK (BILF2) (left panels). Functional avidity of these clones was measured
simultaneously as in (A). Arrows indicate equivalent numbers of lytic antigen expressing cells. Data are represented as mean+/2SEM.
doi:10.1371/journal.ppat.1000490.g005
EBV CD8+T Cell Evasion
PLoS Pathogens | www.plospathogens.org7 June 2009 | Volume 5 | Issue 6 | e1000490

Page 8

Figure 6. RNA and protein expression kinetics of BNLF2a relative to immediate early, early and late genes. AKBM cells containing latent
virus were stimulatedtoinducelytic cyclereplication, samples harvested attheindicatedtimesandselectedviral transcriptandprotein levelsestimated.
Samples were harvested from 0 to 48 hours post induction (pi), and RNA was harvested and subjected to qRT-PCR detection of BZLF1, BRLF1, BMLF1,
BNLF2a, BALF4 and BLLF1 transcripts (A). Values shown are represented as expression relative to their maximum. Protein samples harvested from the
same time points were subjected to western blot analysis, where samples were probed with antibodies to the indicated lytic cycle antigens (B).
doi:10.1371/journal.ppat.1000490.g006
EBV CD8+T Cell Evasion
PLoS Pathogens | www.plospathogens.org8 June 2009 | Volume 5 | Issue 6 | e1000490

Page 9

analysis of surface HLA class I levels on wild-type and
DBNLF2a LCLs from different donors, which had been co-
stained for viral proteins expressed at different phases of lytic
cycle. Wild-type LCLs stained for BZLF1 expression showed a
decrease in surface HLA class I levels by around 1/3 of the level
in latent (lytic antigen negative) cells, yet BZLF1 expressing
DBNLF2a LCLs showed little to no decrease in surface HLA
class I levels (Figure 7A and B upper panels). However, when
cells were stained for the late lytic cycle protein BALF4, surface
HLA class I levels in both the wild-type and DBNLF2a LCLs
were decreased by around half of the level of that seen in latent
cells (Figure 7A and B lower panels).
Discussion
In this study we have shown that CD8+T cell recognition of
immediate early and early lytic cycle antigens is dramatically
increased in LCLs transformed with a mutant EBV lacking the
immune evasion gene BNLF2a compared to the recognition of
wild-type EBV transformed LCLs. This increase in recognition
was conserved across different HLA-class I backgrounds and these
effects were seen using multiple different CD8+T cell specificities,
reinforcing the role of BNLF2a in active immune evasion during
EBV lytic cycle replication. No observable difference in recogni-
tion of late lytic cycle antigens was observed, and peptide titration
analysis of the late-specific CD8+T cell clones ruled out the
possibility that these effectors were simply less avid than those
specific for the immediate early and early phases.
The observed increase in recognition of immediate early
antigens was not anticipated when considered in the light of
BNLF2a’s previously described expression kinetics, where BNLF2a
transcripts were not found to peak until at least 4 hours after
immediate early gene expression [13]. By performing detailed
analysis of the transcription and protein expression kinetics of
BNLF2a and the immediate early genes in an EBV-infected B cell
line in which lytic replication could be induced, we found that
although immediate early protein expression was initiated prior to
that of BNLF2a, there was a substantial increase in the expression
immediate early proteins coincident with the expression of
BNLF2a at 3 hours post induction. Epitopes derived from the
first wave of immediate early protein synthesis will have no
protection from being processed and presented to CD8+T cells.
However given that the major source of epitopes feeding the class I
antigen processing pathway is now thought to be from de-novo
synthesized proteins in the form of short-lived defective ribosomal
products (DRiPs) rather than long lived protein (reviewed in [14]),
expression of BNLF2a during this second wave of expression of the
immediate early proteins would restrict the supply of epitope
peptides at this time.
Analysis of the sequence of early protein expression using the
inducible lytic replication system showed that BNLF2a was
expressed with the first wave of early proteins, BALF2 and
Figure 7. Surface HLA-class I expression in wild-type and DBNLF2a LCLs expressing immediate early or late antigens. Wild-type and
DBNLF2a LCLs were stained for surface HLA-class I and expression levels measured by flow cytometry on cells co-stained for lytic antigens: either the
immediate early antigen BZLF1 (upper panels), or the late antigen BALF4 (lower panels). The panels show histograms and MFI values of cell surface
HLA-class I expression gated on cells with latent virus (lytic antigen negative, shaded histogram) or lytic virus (lytic antigen positive, open histogram).
Staining data is presented from (A) Donor 1 LCLs and (B) Donor 2 LCLs.
doi:10.1371/journal.ppat.1000490.g007
EBV CD8+T Cell Evasion
PLoS Pathogens | www.plospathogens.org9June 2009 | Volume 5 | Issue 6 | e1000490

Page 10

BMLF1. Similar to what is seen with the immediate early proteins,
BNLF2a’s expression was upregulated coincident with the
increasing expression of these early proteins, again at a time
when epitope production from these proteins is likely to be
maximal. T cell recognition experiments using effectors specific for
these proteins showed that deletion of BNLF2a from the targets
caused clear increases in recognition of epitopes derived from
these proteins compared to those expressed in wild-type targets.
This indicates that although BNLF2a is expressed coincidently
with these proteins, it can afford a substantial degree of protection
from T cell recognition at this stage. Consistent with this finding
was the observation that BNLF2a-deficient cells expressing
BZLF1, and thus including those cells progressing through to
early stages of the replicative cycle, showed an increase in class I
MHC levels relative to wild type transformed cells, confirming
BNLF2a’s role in inhibiting antigen presentation at this time.
When different CD8+T cell specificities were assayed for their
ability to recognize their cognate antigen presented by the
DBNLF2a LCLs as compared to the wild-type LCLs, variable
levels of increased recognition were seen for the different T cell
specificities. In some cases why this variability occurs is not clear.
The abundance of the source protein does not appear to play a
role as T cells specific for the three epitopes derived from BRLF1
namely AEN, YVL and IACP show quite different levels of
increased recognition of the DBNLF2a LCL. The TAP depen-
dence of the epitopes studied where determined does not appear to
correlate with recognition. Furthermore as the hydrophobicity of
peptides broadly correlates with the TAP independence [11], no
clear correlation is seen between the hydrophobicity or likely TAP
independence and the increase in recognition. The HLA C
presented epitope IACP was consistently more greatly recognized
when presented by the DBNLF2a LCL compared to other
epitopes presented from these LCLs. Some immune evasion
proteins have been described to have allele specificity, such as the
cytomegalovirus encoded US3 protein [15], however whether
BNLF2a shows allele-specificity requires further investigation.
When the expression profile of the early protein BMRF1 was
examined it showed a delayed pattern of expression relative to
BNLF2a and the other early proteins studied. T cell recognition
assays with clones specific to epitopes derived from BMRF1
consistently showed the lowest increase in recognition by T cells in
BNLF2a-deficient targets, indicating that BNLF2a has some but
perhaps a lesser effect on presentation of epitopes from this
protein. This raises the possibility that other mechanisms are
preventing effective antigen presentation during this later phase of
early gene expression. More compelling evidence for other EBV-
encoded class I evasion mechanisms comes from the study of the T
cell recognition and expression kinetics of late phase protein
targets. The expression of the best characterized late protein,
BALF4, was seen to increase in the inducible cell line from 6 hours
post induction, with heightened expression occurring at 8–
12 hours. At this stage BNLF2a protein levels were decreasing
in these cells, yet T cell recognition experiments using late-specific
effectors to BALF4 and BILF2 show very poor recognition of wild-
type LCL targets. Importantly however, when using the same late-
specific effectors in recognition assays of BNLF2a-deleted targets,
no increase in detection is seen compared to wild-type targets.
Given that the target of BNLF2a is the TAP complex and we have
shown previously that this complex is not degraded during EBV
lytic cycle replication, at least at 24 hours post-induction of lytic
cycle [8], this would suggest that EBV-encoded mechanisms other
than BNLF2a are operating to block antigen presentation during
the late phase of replication. Supporting this idea is the
observation that BNLF2a-deficient LCLs expressing the late
antigen BALF4 show decreased levels of surface class I MHC
molecules similar to wild-type virus transformed cells.
Evasion of CD8+T cell recognition is likely to be most efficient
when multiple points of the antigen processing pathway are
targeted, with BNLF2a being one of potentially several immune
evasion proteins. Other proteins potentially involved in this
process include the early-expressed gene BGLF5 which functions
as an alkaline exonuclease and a host protein synthesis inhibitor.
BGLF5’s inhibition of global protein synthesis, including that of
class I MHC, can inhibit effective CD8+T cell recognition of
cognate targets [16,17]. A second candidate recently identified in
modulating surface class I levels is the early phase expressed gene
BILF1, whose product acts to promote turnover of surface class I
molecules [18]. Conceivably these proteins may act in a
complementary manner to BNLF2a at early time points, initially
by BILF1 clearing class I complexes containing immediate early
epitopes from the surface of the cell that were produced before
BNLF2a function was established and then BGLF5 acting to
prevent effective class I synthesis.
As to BNLF2a’s function in vivo, it is difficult to draw direct
inferences from animal herpesvirus models in which immune
evasion genes have been disrupted since the viruses used, either
the b-herpesvirus murine cytomegalovirus (MCMV) or the c-2
herpesvirus MHV-68, have different in vivo infection biology
compared to EBV. Nevertheless, recent work on the b-herpesvirus
MCMV has indicated that deletion of viral regulators of antigen
processing either has no effect on immunodominance hierarchies
or virus loads [19,20], or surprisingly, decreases the size of at least
some CD8+T cell reactivities [21]; perhaps as a consequence of
increased antigen clearance. In the case of MHV-68 which has a
similar cellular tropism to EBV, deletion of the immune evasion
gene mK3, which is expressed during latency establishment and
also during lytic replication, led to increased CD8+T cell
responses to lytic proteins yet had little effect on levels of virus
undergoing lytic replication. It did however decrease latent viral
loads, suggesting a role for mK3 in amplifying the latent virus
reservoir [22]. By contrast, BNLF2a is not expressed during
latency and EBV’s mechanism of amplifying the latent viral load
may come more from its growth transforming ability, by directly
expanding latently infected B cells when first colonizing the B cell
system. Ultimately, the impact BNLF2a has on immunodomi-
nance, viral loads and transmission may be best addressed using
the closely related rhesus macaque lymphocryptovirus (Cerco-
pithicine herpesvirus 15) model. This virus has a similar biology to
EBV and the same repertoire of genes [23], including a BNLF2a
homologue which has the ability to cause surface class I MHC
downregulation when expressed in rhesus cell lines [9].
Overall, these results indicate that BNLF2a functions to protect
the immediate early and early proteins from being efficiently
processed and presented to CD8+T cells. We would expect then
that in vivo BNLF2a would function to shield virus reactivating
from latency or initiating lytic cycle replication. Such stage-specific
expression of immune evasion genes is a feature of several
herpesviruses. Perhaps the clearest example comes from CMV
where multiple proteins involved in disrupting CD8+T cell
recognition of infected cells have been described. During CMV
replication the US3 gene, whose product retains class I complexes
in the endoplasmic reticulum, is abundantly expressed during the
immediate early phase [24–26], while the gene US11, whose
product dislocates class I molecules from the endoplasmic
reticulum into the cytosol, is expressed predominantly during
early phase replication, and the TAP inhibitor US6 is transcribed
in early and late phases [27]. The differential expression of these
genes then may be in part why these viruses utilize multiple
EBV CD8+T Cell Evasion
PLoS Pathogens | www.plospathogens.org 10June 2009 | Volume 5 | Issue 6 | e1000490

Page 11

evasion mechanisms. In the case of EBV replication, as BNLF2a
acts in a stage-specific manner we suggest that it will act in concert
with other EBV encoded immune evasion genes to reduce efficient
T-cell surveillance of reactivating or productively infected host
cells.
Materials and Methods
Ethics statement
All experiments were approved by the South Birmingham Local
Research Ethics Committee (07/Q2702/24). All patients provided
written informed consent for the collection of blood samples and
subsequent analysis.
Recombinant EBV strains
Wild-type and DBZLF1 recombinant EBV BACs used have
been previously described [10].The generation of a recombinant
EBV BAC deleted for BNLF2a was performed as follows: a
targeting vector containing the BNLF2a region was used to delete
BNLF2a from the wild-type B95.8 EBV BAC genome. The
introduction of a tetracycline cassette, flanked by FLP recombi-
nase target sites (FRT), between a unique XhoI site (26 bp from
the BNLF2a open reading frame ATG initiation codon) and AatII
site (108 bp downstream of the BNLF2a initiation codon) allowed
for the insertional mutagenesis of the BNLF2a ORF. This left a
66 bp 39 BNLF2a sequence fragment intact that was lacking an
initiation codon. Homologous recombination of the target vector,
via flanking sequences either side of the truncated BNLF2a,
allowed for the introduction of the mutation into the wild-type
EBV B95.8 BAC sequence. Successfully recombined clones were
doubly selected on tetracycline and chloramphenicol (the latter
resistance cassette present in the wild-type backbone sequence),
followed by removal of the tetracycline cassette through
transformation of an FLP recombinase. Bacterial clones that
survived this selection process were screened with several
restriction enzymes and also sequenced to confirm successful
disruption of BNLF2a (data not shown).
Wild-type, DBNLF2a and DBZLF1 recombinant virus prepa-
rations were generated by stably transfecting 293 cells with the
corresponding EBV BAC genome and inducing lytic cycle
replication, as previously described [10,28].
Generation of target cell lines and T cell clones
B lymphoblastoid target cell lines (LCLs) were generated by
transformation of laboratory donor B lymphocytes (isolated by
positive CD19 DynabeadH (Invitrogen) selection, as per the
manufacturer’s instructions) with the following recombinant
EBV viruses: wild-type, DBNLF2a and DBZLF1. LCLs were
maintained in standard medium (RPMI-1640, 2 mM glutamine,
and 10% [vol/vol] FCS). Effector CD8+T cells were generated as
previously described [6,29]. CD8+T cell clones used in this study
were specific for the following epitopes derived from the respective
EBV gene products: RAKFKQLL from BZLF1 presented by
HLA-B*0801 [30], AENAGNDAC from BRLF1 presented by
HLA-B*4501 [6], IACPIVMRYVLDHLI from BRLF1 presented
by HLA-C*0202 [6], ARYAAYYLQF from BALF2 presented by
HLA-B*2705 [6], TLDYKPLSV from BMRF1 presented by
HLA-A*0201 [31], FLDKGTYTL from BALF4 presented by
HLA-A*0201 [6], RRRKGWIPL from BILF2 presented by HLA-
B*2705 [6], YVLDHLIVV from BRLF1 presented by HLA-
A*0201 [32], GLCTLVAML from BMLF1 presented by HLA-
A*0201 [29,33].
CD8+T cell recognition experiments
The capacity of lytic-specific CD8+T cell clones to recognize
lytically replicating cells within LCLs of the relevant HLA type was
measured by IFNc ELISA (Endogen). Briefly, target LCLs (56104
cells/well) were co-cultured in triplicate with effector CD8+T cells
(56103cells/well) in V-bottomed 96-well plates in a total of 200 ml
standard media/well and incubated overnight at 37uC with 5%
CO2. After 18 hours 50 ml of culture supernatant from each well
was used for IFNc detection by ELISA
Reactivation of AKBM cells into EBV lytic cycle
AKBM cells and their use have been described previously [8].
Briefly, this EBV infected cell line contains a reporter GFP-rat
CD2 construct under the control of an early EBV promoter to
allow identification of cells in lytic cycle. Prior to induction,
AKBM cells were sorted by FACS to exclude any GFP+ve cells
that had spontaneously entered lytic cycle. The GFP-ve fraction
was then induced into lytic cycle by crosslinking of surface IgG
molecules as previously described [8]. Cells were then harvested at
the indicated timepoints post induction for western blotting and
qRT-PCR analysis.
Western blot assays
Total cell lysates were generated by denaturation in lysis buffer
(final concentration: 8 M urea, 50 mM Tris/HCl pH 7.5,
150 mM sodium 2-mercaptoethanesulfonate) and sonicated.
Protein concentration was determined using a Bradford protein
assay (Bio-Rad), and 20 mg of protein for each sample was
separated by SDS-polyacrylamide gel electrophoresis (SDS-
PAGE) using a Bio-Rad Mini Gel tank. Proteins were blotted
onto nitrocellulose membranes and blocked by incubation for 1 hr
in 5% skimmed-milk powder dissolved in PBS-Tween 20
detergent (0.05% [vol/vol]). Specific proteins were detected by
incubation with primary antibodies for BZLF1 (murine monoclo-
nal antibody (MAb) BZ.1, final concentration 0.5 mg/ml, [34]),
BRLF1 (murine MAb clone 8C12, final concentration 2.5 mg/ml,
Argene, cat. # 11-008), BMLF1 (rabbit serum to EBV BSLF2/
BMLF1-encoded SM, clone EB-2, used at 1/6000 [35]), BMRF1
(murine MAb clone OT14-E, used at 1/2000 [36]), BALF2
(murine MAb clone OT13B, used at 1/5000, [37]), BNLF2a
(clone 5B9, used at 1/100, a rat hybridoma supernatant directed
to the N-terminal region of BNLF2a generated by E. Kremmer
through immunization of Lou/C rats with KLH-coupled BNLF2a
peptides, followed by fusion of rat immune spleen cells with the
myeloma cell line P3X63-Ag8.653), BALF4 (murine Mab clone
L2, used at 1/100, [38]) BFRF3 (rat MAb clone OT15-E, used at
1/250, J. M. Middeldorp, [39]) and EBNA2 (murine MAb clone
PE-2, used at 1/50, [40]) for 2 hrs at room temperature, followed
by extensive washes with PBS-Tween. Detection of bound primary
antibodies was by incubation for 1 hr with appropriate horserad-
ish peroxidase (HRP)-conjugated secondary antibodies (goat anti-
mouse IgG:HRP (Sigma, cat. #A4416), goat anti-rat IgG:HRP
(Sigma, cat. #A9037), and goat anti-rabbit IgG:HRP (Sigma, cat.
#A6154). Bound HRP was then detected by enhanced chemilu-
minescence (ECL, Amersham).
Quantitative real-time reverse transcription PCR
Total RNA was extracted from 0.56106
NucleoSpinH RNA II kit (Machery-Nagel) followed by Turbo
DNA-freeTM(Ambion/Applied Biosystems) treatment to remove
any residual DNA contamination, as per the manufacturers’
instructions. 500 ng of RNA was reverse transcribed into cDNA
using a pool of primers specific for BZLF1, BRLF1, BMLF1,
cells using a
EBV CD8+T Cell Evasion
PLoS Pathogens | www.plospathogens.org11June 2009 | Volume 5 | Issue 6 | e1000490

Page 12

BNLF2a, BALF4 and BLLF1, with GAPDH included as an internal
control, followed by subsequent quantitative-PCR (q-PCR). EBV
lytic gene primers were as follows (primer sequences in
parenthesis):BZLF1(cDNA
F 59ACGACGCACACGGAAACC39, R 59CTTGGCCCGG-
CATTTTCT39, probe 59GCATTCCTCCAGCGATTCTGG-
CTGTT39), BRLF1 (cDNA 59CAGGAATCATCACCCG39,
F 59TTGGGCCATTCTCCGAAAC39, R 59TATAGGGCAC-
GCGATGGAA39,
probe
59AGACGGGCTGAGAATGCC-
GGC39), BMLF1 (cDNA 59GAGGATGAAATCTCTCCAT39,
F 59CCCGAACTAGCAGCATTTCCT39, R 59GACCGCTTC-
GAGTTCCAGAA39, probe 59AACGAGGATCCCGCAGA-
GAGCCA39), BNLF2a(cDNA
TGG39, F 59TGGAGCGTGCTTTGCTAGAG39, R 59GG-
CCTGGTCTCCGTAGAAGAG39,
CTGCGGCCTGCC39),BALF4
AGGCCCTC39, F 59CCAGCTTTCCTTTCCGAGTCT 39,
R 59ACACTGGATGTCCGAGGAGAA39, probe 59TCCA-
GCCACGGCGACCTGTTC39), and BLLF1 (cDNA 59ACTG-
CAGTACTAGCATGG39,
F
GACGTT39, R 59ACATGGAGCCCGGACAAGT39, probe
59AGCCCACCACAGATTACGGCGGT39). cDNA and for-
ward/reverse primers were synthesised by Alta Bioscience
(University of Birmingham). Probes were synthesised by Euro-
gentec S.A and labelled with 59 FAM fluorophore and 39 TAMRA
quencher. Data was normalised to GAPDH expression, and
expressed as relative to the maximal level of transcript for each
gene.
59GCAGCCACCTCACG39,
59GTCTGCTGACGTC-
probe
(cDNA
59CCTCTGC-
59CCATCAAC-
59AGAATCTGGGCTGG-
Flow cytometry
LCLs were assayed for the percentage of cells spontaneously
reactivating into lytic cycle by intracellular staining for BZLF1.
Cells were first fixed using 100 ml of Ebiosciences Intracellular (IC)
Fixative (cat. # 00-8222-49) for 1 hr on ice, followed by
permeabilisation through the addition of 100 ml Triton X-100
(final concentration 0.2%) and a further 30 minute incubation on
ice. After extensive washing with PBS, cells were incubated with
1 mg/ml of either MAb BZ.1 (anti-BZLF1) or with an IgG1isotype
control antibody for 1 hr at 37uC. Cells were washed twice in PBS
and then incubated with 1:20-diluted R-phycoerythrin-conjugated
goatanti-mouseIgG1
antibody
STAR132PE) for 1 hr at 37uC. Following further washes cells
(AbDSerotec,cat.
#
were resuspended in IC fixative and analysed on a Dako Cyan
flow cytometer (Dako, Denmark).
LCL surface HLA class I and intracellular lytic-cycle EBV
antigens were detected simultaneously by first staining viable cells
with 1:15-diluted allophycocyanin-conjugated-anti-human HLA-
A,B,C (Biolegend, cat. # 311410) antibody for 30 minutes on ice.
Cells were then washed extensively in PBS and fixed and
permeabilised as above, followed by incubation for 1 hr at 37uC
with 1 ug/ml of either MAb BZ.1 (immediate early antigen
BZLF1) or L2 (late antigen BALF4), or IgG1isotype control. After
several washes in PBS cells were incubated for 1 hr with 1:20-
diluted R-phycoerythrin-conjugated goat anti-mouse IgG1anti-
body as above. Cells were washed and fixed as above, followed by
analysis on a Dako cytometer (Dako, Denmark). All flow data was
analyzed using FlowJo software (Tree Star).
Supporting Information
Figure S1
is expressed in these cells. DBNLF2a LCLs were transfected by
electroporation with a plasmid which co-expressed BNLF2a and
the truncated nerve growth factor (NGFR) gene. After 48 hours,
BNLF2a expressing cells were purified by selecting NGFR
expressing cells. These cells were used in standard T cell
recognition assays in parallel with the NGFR-negative cells from
the transfection, wild-type virus transformed LCLs, the unmanip-
ulated DBNLF2a LCL and the DBZLF1 knock out LCL. CD8+T
cells specific for the immediate early epitope AEN and early
epitope ARYA were used as effectors in parallel assays. One
representative assay of two transfection experiments is shown.
Found at: doi:10.1371/journal.ppat.1000490.s001 (0.71 MB PDF)
T cell recogntion of DBNLF2a LCLs when BNLF2a
Acknowledgments
We thank Daphne van Leeuwen for excellent technical support.
Author Contributions
Conceived and designed the experiments: MER EJHJW MR ABR ADH.
Performed the experiments: NPC CSL AIB DH ADH. Analyzed the data:
NPC MER EJHJW MR ABR ADH. Contributed reagents/materials/
analysis tools: CSL AIB EK JMM. Wrote the paper: NPC ADH.
References
1. Stinchcombe JC, Griffiths GM (2003) The role of the secretory immunological
synapse in killing by CD8+ CTL. Semin Immunol 15: 301–305.
2. Groothuis TA, Griekspoor AC, Neijssen JJ, Herberts CA, Neefjes JJ (2005)
MHC class I alleles and their exploration of the antigen-processing machinery.
Immunol Rev 207: 60–76.
3. Vossen MT, Westerhout EM, Soderberg-Naucler C, Wiertz EJ (2002) Viral
immune evasion: a masterpiece of evolution. Immunogenetics 54: 527–542.
4. Lilley BN, Ploegh HL (2005) Viral modulation of antigen presentation:
manipulation of cellular targets in the ER and beyond. Immunol Rev 207:
126–144.
5. Rickinson A, Kieff E (2007) Epstein-Barr virus. In: Knipe DM, Howley PM, eds.
Fields Virology Philadelphia Walters Kluwer/Lippincott, Williams & Wilkins.
pp 2655–2700.
6. Pudney VA, Leese AM, Rickinson AB, Hislop AD (2005) CD8+ immunodo-
minance among Epstein-Barr virus lytic cycle antigens directly reflects the
efficiency of antigen presentation in lytically infected cells. J Exp Med 201:
349–360.
7. Keating S, Prince S, Jones M, Rowe M (2002) The lytic cycle of Epstein-Barr
virus is associated with decreased expression of cell surface major histocompat-
ibility complex class I and class II molecules. J Virol 76: 8179–8188.
8. Ressing ME, Keating SE, van Leeuwen D, Koppers-Lalic D, Pappworth IY, et
al. (2005) Impaired transporter associated with antigen processing-dependent
peptide transport during productive EBV infection. J Immunol 174: 6829–
6838.
9. Hislop AD, Ressing ME, van Leeuwen D, Pudney VA, Horst D, et al. (2007) A
CD8+ T cell immune evasion protein specific to Epstein-Barr virus and its close
relatives in Old World primates. J Exp Med 204: 1863–1873.
10. Feederle R, Kost M, Baumann M, Janz A, Drouet E, et al. (2000) The Epstein-
Barr virus lytic program is controlled by the co-operative functions of two
transactivators. Embo J 19: 3080–3089.
11. Lautscham G, Mayrhofer S, Taylor G, Haigh T, Leese A, et al. (2001)
Processing of a multiple membrane spanning Epstein-Barr virus protein for
CD8(+) T cell recognition reveals a proteasome-dependent, transporter
associated with antigen processing-independent pathway. J Exp Med 194:
1053–1068.
12. Takada K, Ono Y (1989) Synchronous and sequential activation of latently
infected Epstein-Barr virus genomes. J Virol 63: 445–449.
13. Yuan J, Cahir-McFarland E, Zhao B, Kieff E (2006) Virus and cell RNAs
expressed during Epstein-Barr virus replication. J Virol 80: 2548–2565.
14. Yewdell JW, Nicchitta CV (2006) The DRiP hypothesis decennial: support,
controversy, refinement and extension. Trends Immunol 27: 368–373.
15. Park B, Kim Y, Shin J, Lee S, Cho K, et al. (2004) Human cytomegalovirus
inhibits tapasin-dependent peptide loading and optimization of the MHC class I
peptide cargo for immune evasion. Immunity 20: 71–85.
16. Rowe M, Glaunsinger B, van Leeuwen D, Zuo J, Sweetman D, et al. (2007) Host
shutoff during productive Epstein-Barr virus infection is mediated by BGLF5
and may contribute to immune evasion. Proc Natl Acad Sci U S A 104:
3366–3371.
EBV CD8+T Cell Evasion
PLoS Pathogens | www.plospathogens.org12June 2009 | Volume 5 | Issue 6 | e1000490

Page 13

17. Zuo J, Thomas W, van Leeuwen D, Middeldorp JM, Wiertz EJ, et al. (2008)
The DNase of gammaherpesviruses impairs recognition by virus-specific CD8+
T cells through an additional host shutoff function. J Virol 82: 2385–2393.
18. Zuo J, Currin A, Griffin BD, Shannon-Lowe C, Thomas WA, et al. (2009) The
epstein-barr virus g-protein-coupled receptor contributes to immune evasion by
targeting MHC class I molecules for degradation. PLoS Pathog 5: e1000255.
doi:10.1371/journal.ppat.1000255.
19. Munks MW, Pinto AK, Doom CM, Hill AB (2007) Viral interference with
antigen presentation does not alter acute or chronic CD8 T cell immunodo-
minance in murine cytomegalovirus infection. J Immunol 178: 7235–7241.
20. Gold MC, Munks MW, Wagner M, McMahon CW, Kelly A, et al. (2004)
Murine cytomegalovirus interference with antigen presentation has little effect
on the size or the effector memory phenotype of the CD8 T cell response.
J Immunol 172: 6944–6953.
21. Bohm V, Simon CO, Podlech J, Seckert CK, Gendig D, et al. (2008) The
immune evasion paradox: immunoevasins of murine cytomegalovirus enhance
priming of CD8 T cells by preventing negative feedback regulation. J Virol 82:
11637–11650.
22. Stevenson PG, May JS, Smith XG, Marques S, Adler H, et al. (2002) K3-
mediated evasion of CD8(+) T cells aids amplification of a latent gamma-
herpesvirus. Nat Immunol 3: 733–740.
23. Rivailler P, Jiang H, Cho YG, Quink C, Wang F (2002) Complete nucleotide
sequence of the rhesus lymphocryptovirus: genetic validation for an Epstein-Barr
virus animal model. J Virol 76: 421–426.
24. Biegalke BJ (1995) Regulation of human cytomegalovirus US3 gene transcrip-
tion by a cis-repressive sequence. J Virol 69: 5362–5367.
25. Tenney DJ, Colberg-Poley AM (1991) Human cytomegalovirus UL36-38 and
US3 immediate-early genes: temporally regulated expression of nuclear,
cytoplasmic, and polysome-associated transcripts during infection. J Virol 65:
6724–6734.
26. Jones TR, Wiertz EJ, Sun L, Fish KN, Nelson JA, et al. (1996) Human
cytomegalovirus US3 impairs transport and maturation of major histocompat-
ibility complex class I heavy chains. Proc Natl Acad Sci U S A 93: 11327–11333.
27. Jones TR, Muzithras VP (1991) Fine mapping of transcripts expressed from the
US6 gene family of human cytomegalovirus strain AD169. J Virol 65:
2024–2036.
28. Delecluse HJ, Hilsendegen T, Pich D, Zeidler R, Hammerschmidt W (1998)
Propagation and recovery of intact, infectious Epstein-Barr virus from
prokaryotic to human cells. Proc Natl Acad Sci U S A 95: 8245–8250.
29. Steven NM, Annels NE, Kumar A, Leese AM, Kurilla MG, et al. (1997)
Immediate early and early lytic cycle proteins are frequent targets of the Epstein-
Barr virus-induced cytotoxic T cell response. J Exp Med 185: 1605–1617.
30. Bogedain C, Wolf H, Modrow S, Stuber G, Jilg W (1995) Specific cytotoxic T
lymphocytes recognize the immediate-early transactivator Zta of Epstein-Barr
virus. J Virol 69: 4872–4879.
31. Hislop AD, Annels NE, Gudgeon NH, Leese AM, Rickinson AB (2002) Epitope-
specific evolution of human CD8(+) T cell responses from primary to persistent
phases of Epstein-Barr virus infection. J Exp Med 195: 893–905.
32. Saulquin X, Ibisch C, Peyrat MA, Scotet E, Hourmant M, et al. (2000) A global
appraisal of immunodominant CD8 T cell responses to Epstein-Barr virus and
cytomegalovirus by bulk screening. Eur J Immunol 30: 2531–2539.
33. Scotet E, David-Ameline J, Peyrat MA, Moreau-Aubry A, Pinczon D, et al.
(1996) T cell response to Epstein-Barr virus transactivators in chronic
rheumatoid arthritis. J Exp Med 184: 1791–1800.
34. Young LS, Lau R, Rowe M, Niedobitek G, Packham G, et al. (1991)
Differentiation-associated expression of the Epstein-Barr virus BZLF1 transacti-
vator protein in oral hairy leukoplakia. J Virol 65: 2868–2874.
35. Buisson M, Manet E, Trescol-Biemont MC, Gruffat H, Durand B, et al. (1989)
The Epstein-Barr virus (EBV) early protein EB2 is a posttranscriptional activator
expressed under the control of EBV transcription factors EB1 and R. J Virol 63:
5276–5284.
36. Zhang JX, Chen HL, Zong YS, Chan KH, Nicholls J, et al. (1998) Epstein-Barr
virus expression within keratinizing nasopharyngeal carcinoma. J Med Virol 55:
227–233.
37. Zeng Y, Middeldorp J, Madjar JJ, Ooka T (1997) A major DNA binding protein
encoded by BALF2 open reading frame of Epstein-Barr virus (EBV) forms a
complex with other EBV DNA-binding proteins: DNAase, EA-D, and DNA
polymerase. Virology 239: 285–295.
38. Kishishita M, Luka J, Vroman B, Poduslo JF, Pearson GR (1984) Production of
monoclonal antibody to a late intracellular Epstein-Barr virus-induced antigen.
Virology 133: 363–375.
39. van Grunsven WM, van Heerde EC, de Haard HJ, Spaan WJ, Middeldorp JM
(1993) Gene mapping and expression of two immunodominant Epstein-Barr
virus capsid proteins. J Virol 67: 3908–3916.
40. Young L, Alfieri C, Hennessy K, Evans H, O’Hara C, et al. (1989) Expression of
Epstein-Barr virus transformation-associated genes in tissues of patients with
EBV lymphoproliferative disease. N Engl J Med 321: 1080–1085.
EBV CD8+T Cell Evasion
PLoS Pathogens | www.plospathogens.org13June 2009 | Volume 5 | Issue 6 | e1000490