![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Previous Article | Next Article 
Journal of Virology, March 2006, p. 2548-2565, Vol. 80, No. 5
0022-538X/06/$08.00+0 doi:10.1128/JVI.80.5.2548-2565.2006
Copyright © 2006, American Society for Microbiology. All Rights Reserved.
Ellen Cahir-McFarland,
Bo Zhao, and
Elliott Kieff*
Departments of Medicine and Microbiology and Molecular Genetics, Brigham and Women's Hospital and Harvard Medical School, Channing Laboratory, 181 Longwood Avenue, Boston, Massachusetts 02115
Received 16 August 2005/ Accepted 21 November 2005
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionAddendum in proofReferences |
|---|
5% before IgG cross-linking and increased to more than 50% 48 h after induction. Seventy-two hours after IgG cross-linking, gp350-positive cells excluded propidium iodide as well as gp350-negative cells. EBV RNA levels changed temporally in parallel with previously defined sensitivity to inhibitors of protein or viral DNA synthesis. BZLF1 immediate-early RNA levels doubled by 2 h and reached a peak at 4 h, whereas BMLF1 doubled by 4 h with a peak at 8 h, and BRLF1 doubled by 8 h with peak at 12 h. Early RNAs peaked at 8 to 12 h, and late RNAs peaked at 24 h. Hybridization to intergenic sequences resulted in evidence for new EBV RNAs. Surprisingly, latency III (LTIII) RNAs for LMP1, LMP2, EBNALP, EBNA2, EBNA3A, EBNA3C, and BARTs were detected at 8 to 12 h and reached maxima at 24 to 48 h. EBNA2 and LMP1 were at full LTIII levels by 48 h and localized to gp350-positive cells. Thus, LTIII expression is a characteristic of late EBV replication in both B lymphoblasts and epithelial cells in immune-comprised people (J. Webster-Cyriaque, J. Middeldorp, and N. Raab-Traub, J. Virol. 74:7610-7618, 2000). EBV replication significantly altered levels of 401 Akata cell RNAs, of which 122 RNAs changed twofold or more relative to uninfected Akata cells. Mitogen-activated protein kinase levels were significantly affected. Late expression of LTIII was associated with induction of NF-
B responsive genes including IB
and A20. The exclusion of propidium, expression of EBV LTIII RNAs and proteins, and up-regulation of specific cell RNAs are indicative of vital cell function late in EBV replication. |
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAddendum in proofReferences |
|---|
EBV replication in latently infected B lymphocytes is essential for persistent oropharyngeal replication. Prolonged acyclovir treatment effectively inhibits EBV production in the oropharynx. However, latent B-lymphocyte infection is unaffected, and EBV replication rapidly ensues when acyclovir treatment is stopped (105). Furthermore, genetically deficient humans, with X-linked agammaglobulinemia, lack mature B lymphocytes and do not have latent EBV infection in B lymphocytes or persistent oropharyngeal EBV replication (31, 53, 77). Since oropharyngeal EBV is essential for EBV transmission to uninfected people, EBV replication in latently infected B lymphocytes has a key role in EBV epidemiology and persistence in human populations. Also, Southern Chinese people with higher levels of EBV antibody are more likely to develop nasopharyngeal cancer (107), consistent with a role for high-level EBV replication in malignant conversion of oropharyngeal epithelial cells. Moreover, the induction of EBV replication in latently infected cells is being evaluated as a therapeutic approach to stop malignant cell proliferation (4).
The experiments described here were undertaken to investigate the ongoing interaction between EBV and cell gene expression following the induction of EBV replication in latently infected B lymphocytes. Since EBV-infected peripheral blood B lymphocytes in persistently infected people are frequently LTI infected and antigen activation of the B-cell receptor is a physiologically appropriate stimulus for EBV replication, we have studied the time course of EBV and cell gene expression following the induction of EBV replication following surface immunoglobulin (IgG) cross-linking in Akata Burkitt's lymphoma (BL) cell. Cross-linking of surface IgG in Akata cells results in synchronous induction of EBV replication in more than 50% of the cells (28, 95, 97, 98). EBV encoded RNAs and promoter start sites have been identified following the induction of EBV replication in latently infected B lymphoblast lines, including Akata. RNAs have been characterized by sensitivity to inhibitors of protein or viral DNA synthesis (30, 43). Herpesvirus immediate-early (IE) RNAs are transcribed by preexisting proteins, are unaffected by inhibition of cell protein synthesis, and encode inducers of early (E) RNA transcription. E RNAs encode proteins that modify nucleotides and replicate virus DNA but are not affected by inhibitors of virus DNA synthesis, whereas late (L) RNA transcription is inhibited by viral DNA synthesis inhibitors. L RNAs encode proteins necessary for virus morphogenesis and egress. IE, E, and L RNAs temporally overlap. Since we were most interested in the interaction of EBV and cell genes in transcript regulation, we focused primarily on the precise temporal sequence of changes in virus and cell RNA, using comprehensive oligonucleotide arrays to measure changes in virus and cell RNA levels (48). Similar arrays have been used to investigate virus or cell RNAs in cells permissive for herpes simplex virus (HSV), pseudorabies virus,cytomegalovirus (CMV), and Kaposi's sarcoma-associated herpesvirus (KSHV) replication (1, 40, 41, 49, 52, 54, 66, 71, 76, 86, 92-94, 104).
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAddendum in proofReferences |
|---|
Antibodies. Mouse monoclonal antibody (MAb) 2L10 was used to detect cell surface gp350 expression by fluorescence microcopy or fluorescence-activated cell sorter (FACS) analyses. S12 (60), A10 (62), and PE2 (106) MAbs specific for LMP1, EBNA3C, and EBNA2, respectively, were used in Western blots. Sheep anti-EBNA-3A antibody (Exalpha Biologicals) was used to detect EBNA3A. EBV immune human serum was used to detect EBNALP. Mouse MAb specific for EBNA1 was purchased from Advanced Biotechnologies.
FACS analysis and confocal analysis. For FACS analysis (FACSCalibur; Becton Dickinson), 106 cells were incubated with mouse anti-gp350 followed by goat anti-mouse IgG-fluorescein isothiocyanate (Santa Cruz Biotechnology, CA). Cell viability was determined by propidium iodide exclusion (30 µg/ml; Molecular Probes). For confocal microscopy (Nikon PCM2000), cells were mounted, fixed, stained for gp350, and counterstained with propidium iodine.
RNA preparation. Briefly, 108 cells were resuspended at 4°C in 10 ml of 20 mM Tris-HCl, pH 7.4, 0.25% Triton X-100, and 1.25% sucrose in a Dounce homogenizer. Cells were subjected to 10 strokes over 15 min at 4°C. Cytoplasm was separated from nuclei by centrifugation at 2,000 x g for 5 min (45). Cytoplasmic RNA was purified by the RNAzol B method (Tel-Test, TX) and then by an RNeasy column (QIAGENCorporation).
EBV arrays.
In all, 206 75-base oligonucleotides
were based on the complete EBV genome, GeneBank accession
no. AJ507799. Oligonucleotides were synthesized
for EBV open reading frames (ORFs) and to detect potential intergenic
transcripts. Oligonucleotides specific for ORFs were selected with a
3' bias, a G-C content of 55%, and absence of
self-complementary sequences. Of these oligonucleotides, 147 were
expected to detect 92 known ORFs, 113 were expected to detect specific
transcripts, 23 were expected to detect more than one coterminal RNA,
and 37 were expected to detect RNAs from putative intergene DNA.
Oligonucleotides specific for human beta tubulin (CTC AGG CTT
CTC AGT TCC CTT AGC CGT CTT ACT CAA CTG CCC CTT TCC TCT CCC TCA GAA TTT
GTG TTT GCT GCC), glyceraldehyde-3-phosphate dehydrogenase
(GAPDH) (CTC CAA ACA GCC TTG CTT GCT TCG AGA ACC ATT TGC TTC
CCG CTC AGA CGT CTT GAG TGC TAC AGG AAG CTG GCA and
GAC CCG ACC CCA AAG GCC AGG CTG TAA ATG TCA CCG GGA GGA TTG
GGT GTC TGG GCG CCT CGG GGA ACC TGC CCT), human beta actin
(GCT TCT AGG CGG ACT ATG ACT TAG TTG CGT TAC ACC CTT TCT TGA CAA AAC
CTA ACT TGC GCA GAA AAC AAG ATG), ribosomal protein S5 (TGG GAA GTG GAG
CAC CGA TGA TGT GCA GAT CAA TGA CAT TTC CCT GCA GGA TTA CAT TGC AGT GAA
GGA G), and ribosomal protein S9 (ACC CTT CGA GAA ATC TCG TCT CGA CCA
AGA GCT GAA GCT GAT CGG CGA GTA TGG GCT CCG GAA CAA ACG T) were
included as controls to allow for normalization of hybridization
efficiency among experiments. Oligonucleotides specific for
Arabidopsis ribulose bisphophate carboxylase small subunit
(AGC CAC CAA GCT TCA CCG GTT AAT TTC CCT TTG CTT TTG TGT AAA
CCT CAA AAC TTT ATC CCC CAT CTT TGA TTT and TGG CTT CCT CTA TGC TCT CTT
CCG CTA CTA TGG TTG CCT CTC CGG CTC AGG CCA CTA TGG TCG CTC CTT TCA
ACG) and pEARLI (GGC TAA CAT TCT TGG AAT CAA CTT
GAA CCT CCC AAT ATC TTT AAG TCT ACT CCT TAA TGT TTG CAG CAA ACA
ACT and CCC TAG CTC TGG CTC GAG CAA GTG CCC TAA
AGA CAC CCT CAA GCT CGG TGT CTG CGC TAA TGT GCT CAA CGG CCT)
were included as specificity controls. Five computer-generated
oligonucleotides without significant homology for human or EBV DNAs
were the primary specificity controls (ACA GGT GTC CTC AAA
CCA GCC TGA AAC GTT ACT AGG TGA AGA ATC ACC GCG GTT GTC GGT AGT TAA GCG
A, CCA TCC GGG CCA TAA GTT TAT AGT AGC GAT TGT TTT GCC CCT ACC AGC GAA
TCG CGC CCA GTT AGT AAT C, CCT TGG ATG GGT AAA TTC CCT CGT CTA CGC GTA
ACA ACT GAA CGC GTA GCG CGA CGG TCT CAG GAA ATT A, CGG CCA CAA CTC TCA
GGA CGC ATA TAA GAC GCG GAA AGG CAT ACA CGT CTA CTT AGA GAC ACC GAG ACT
T, and GAA TGG CAT CAA CGG CGC TGT ACA TAG TCT TCT
CGC CTA CAT AAT AGC GCT AGT TGA TAG GAA CCA GGG G).
Oligonucleotides were synthesized and purified by Integrated DNA
Technologies(Coralville, IA) and were repurified before use.
Oligonucleotides were robotically deposited onto glass slides in
triplicate by the MGH Microarray Core Facility using GeneMachines
Omnigrid printer (Genomics Solution, MI). Oligonucleotide printing was
quality controlled by terminal deoxynucleotidyltransferase assays
(Amersham). EBV oligonucleotide array hybridizations were performed at
the MGH Microarray Core Facility on GeneTAC Hybridization Stations
(HA000019; Genomic Solutions, MI). Target RNAs were reverse transcribed
with a mixture of random and oligo(dT) primers and labeled with Cy3
(untreated) or Cy5 (anti-Ig treated). Labeled cDNAs were hybridized to
the EBV array at 70°C and stepped down 2°C hourly to
60°C. Microarrays were subsequently scanned by using a GenePix
4000B scanner (Axon Instruments, Inc., CA). Data were collected at a
maximum resolution of 10 mm/pixel with 16 bits of depth by using
ImaGene software (BioDiscovery).
Data were generated from at least two independent time course experiments over 72 h, with each time point hybridized to two slides with triplicate depositions of the full array set. The 2-h time point data are from multiple hybridizations, but only one time course experiment. Data were not analyzed if less than 2/3 arrays from at least two different slides were of acceptable quality. Except for the 2-h time point, where the median number of valid data points was 6, other time points had a median of 12 valid data points. The local background for each array element was subtracted. The signal intensity of each slide was normalized to the internal cell gene controls. Changes in EBV gene expression are expressed first as the calibrated ratio of the average normalized signal intensity of IgG cross-linked cells to mock-treated cells. The data are then translated to relative changes (calibrated ratio normalized to a value of 1 at time zero). The relative changes were converted to log base 2 and imported into the TIGR hierarchical clustering program, MeV.
Northern blots. Cytoplasmic RNA, 10 µg per lane, was separated by electrophoresis on a 1% agarose gel containing 2.2 M formaldehyde, transferred onto a nylon membrane (Zeta-probe GT; Bio-Rad), immobilized by heating at 80°C for 2 h, and hybridized with 32P-labeled EBV DNA probes.
Western blots. Total cell lysates were subjected to sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE), transferred to nitrocellulose membranes, and blotted with the indicated monoclonal antibody or EBV immune human serum followed by peroxidase-conjugated secondary antibody in blocking buffer.
Affymetrix chip analyses. Affymetrix U133 arrays were used to compare mRNA expression from three independent experiments with IgG cross-linked and control EBV+ Akata cells and two parallel independent experiments with EBV– Akata cells. Samples were taken at 0, 4, 8, 12, 24, and 48 h after treatment for both cell types. An additional time point at 36 h was included for EBV+ Akata cells. RNAs were prepared as for EBV RNAs. RNA labeling, hybridization, staining, and scanning used Affymetrix protocols. Expression data were subjected to global scaling in CGOS using a target TGT value of 500. The data were filtered to exclude data where the intensity was less than twice that of absent calls for the specific chip. Genes for which data were present in 2 of 3 inductions and 5 of 7 time points in EBV+ Akata cells (7,575 genes) or both inductions and 4 of 6 time points in EBV-negative cells were retained for statistical analysis (6,865 genes). Changes in gene expression in EBV+ Akata cells were identified by analysis of variance testing (P value of 0.01). The induction ratio of differentially expressed genes was normalized to the 0-h IgG cross-linked samples and twofold or greater changes at one or more time points after treatment compared to the 0-h treated samples were considered significant (134 array elements for 122 unique genes). The corresponding data from EBV– Akata cells for this set are also presented. The ratios of RNA changes following IgG cross-linking of EBV+ to that following IgG cross-linking EBV– Akata cells were calculated. Differences of twofold between EBV+ and EBV– Akata cell lines were considered to be due to EBV replication as opposed to surface IgG cross-linking. In addition, total RNA from BL41 B95-8, BL41, BL30 B95-8, BL30, and two B95-8-transformed lymphocyte cell lines (LCLs) were characterized on U133A chips. Data are presented as BL41 B95-8/BL41, BL30 B95-8/BL30, and average of LCLs/average of BL41 and BL30.
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAddendum in proofReferences |
|---|
|
75% viable before IgG
cross-linking. Whereas control, EBV+ Akata cells
increased from 75% to 85% viable cells at
72 h, cross-linked EBV+ Akata cells
decreased from 75% to 68% viable cells at
72 h after IgG cross-linking (Fig.
1C). Cell cycle analyses
revealed no significant change in the number of cells in G1,
S, or G2M in the first 24 h after IgG
cross-linking of EBV+ or EBV–
Akata cells (data not shown). Moreover, 70% of
EBV+ Akata cells that had progressed to EBV
replication with high-level plasma membrane gp350 expression at 24 to
72 h still excluded propidium iodide at 72 h (Fig.
1D). The fraction of
gp350+ cells that took up propidium iodide was only
4% higher than the fraction of gp350– cells
that took up propidium iodide in the same culture. In sum,
EBV+ Akata cells were less viable than
EBV– Akata cells, and the disparity increased after
IgG cross-linking, but EBV replication over the ensuing 72 h
did not substantially increase cell membrane permeability to propidium
iodide. EBV RNA changes following IgG cross-linking. Changes in EBV cytoplasmic polyadenylated RNAs following IgG cross-linking of EBV+ Akata cells were assessed by hybridization of labeled cDNA to oligonucleotides representative of known or putative EBV RNAs or intergenic DNAs (29) (for updates, see http://www.med.ic.ac.uk/ludwig/ebv.htm). The levels of EBV RNAs in EBV+ Akata cells at each time after IgG cross-linking were normalized to levels from untreated EBV+ Akata cells that were processed in parallel. RNA levels are expressed relative to (n-fold over or under) the level of the RNA in EBV+ cross-linked Akata cells at time zero and are listed in Table 1. Normalization and ratio to levels at the start of the experiment diminish the effect of background spontaneous asynchronous EBV replication. In other experiments (n = 4), similarly determined RNA levels at 48 h after IgG cross-linking of EBV+ Akata cells (IgG) were compared to RNA levels at 48 h after IgG cross-linking of EBV+ Akata cells in the presence of PAA (IgGPAA) and to RNA levels in EBV+ Akata cells treated with PAA alone, so as to observe the relative effects of EBV DNA synthesis inhibition on EBV RNA levels in cross-linking-induced and background EBV replication.
|
After IgG cross-linking, BZLF1 IE RNA increased more than 3-fold by 2 h, while other RNAs were unchanged (Table 1 and Fig. 2A). Consistent with some previous results (5, 6, 50, 79), IE BMLF1 was the next RNA to increase with a threefold increase at 4 h. BRLF1 RNA did not reach twofold-increased levels until 8 h (Table 1 and Fig. 2A) (5). Collectively, most E RNAs levels were <2-fold increased at 4 h, >2-fold increased at 8 h, peaked at 12 h, and declined thereafter (Table 1 and Fig. 2B). Most L RNAs were also >2-fold increased at 8 h, peaked at 24 h, and declined thereafter (Table 1 and Fig. 2B). Thus, the principal temporal difference between E and L RNAs was the peak at 12 h for E versus 24 h for L RNAs.
|
|
In instances where 2 or more RNAs have the same polyadenylation [poly(A)] site, the array oligonucleotide specific for the 3' RNA also detected the coterminal upstream RNAs. For example, Bam a rightward reading frame 1 (BaRF1), BMRF1, and BMRF2 RNAs terminate at the same poly(A) site (Table 1). Consequently, the BMRF2 RNA array oligonucleotide detected BaRF1, BMRF1, and BMRF2 RNAs, complicating interpretation of the BMRF2 result.
Array oligonucleotides were uniformly deposited by control terminal deoxynucleotidyltransferase assays, and all but the least abundant RNAs were usually detected. Most array oligonucleotides also had good response ranges. However, relative array oligonucleotide inefficiencies were evident by comparisons among array oligonucleotides that detected the same or coterminal RNAs or by comparison with Northern blots, e.g., BFRF3, BKRF4, and BMRF1 (Table 1 and Fig. 4). BFRF1 and BFRF2 RNAs are coterminal with BFRF3, and the BFRF3 array oligonucleotide was half as efficient as the BFRF1 and 2 oligonucleotides (Table 1). This discrepancy is also evident at 12 h, when BFRF3 RNA is almost as abundant as BFRF1 and 2 RNAs (Table 1 and Fig. 4A and B). In Fig. 4 (middle and lower panels), the BKRF4 array oligonucleotide was less efficient than BKRF2 or BKRF3 oligonucleotides in detecting BKRF2, 3, and 4, and a BMRF1 oligonucleotide was less effective than the BaRF1 array oligonucleotide in detecting the BaRF1 RNA at 8, 12, and 24 h, particularly given the high abundance of BKRF4 at 12 and 24 h and of BMRF1 at 8 and 12 h (Table 1 and Fig. 4).
|
The clustering of EBV E RNAs among L RNAs or of L RNAs among E RNAs was usually due to detection of coterminal upstream RNA(s) of a different class. In Fig. 3, array oligonucleotides that detect upstream coterminal RNAs are indicated. For example, the L BMRF2 RNA array oligonucleotide detected the coterminal E BaRF1 and BMRF1 RNAs, both of which were at maximal levels at 8 to 12 h, causing BMRF2 to cluster with E RNAs (Table 1 and Fig. 3 and 4). Similarly, the E BLLF2 RNA array oligonucleotide detected the upstream L BLLF1 RNA and the E BKRF3 RNA array oligonucleotide detected the upstream L BKRF2 RNA, causing these E RNAs to cluster as L RNAs (Table 1 and Fig. 3). On the other hand, BXLF1, BHRF1, BHLF1, and LF3 E RNAs did not reach peak levels until 24 h and therefore clustered with the late genes.
LF1 and LF2 RNAs had not been characterized as E or L RNAs based on sensitivity to inhibition of virus DNA synthesis. LF2 was near peak at 8 h and peaked at 12 h, consistent with an E RNA, whereas LF1 peaked at 12 to 24 h, consistent with an E or L RNA. LF2 clustered with E RNAs, whereas LF1 clustered with L RNAs (Fig. 3).
Overall, 37 oligonucleotides between genes were arrayed, and 33 yielded potentially significant hybridizations, which could be indicative of new EBV RNAs (Table 1). A rightward RNA (ITG6R) with a 10-fold peak increase at 36 h was detected by two array oligonucleotides just upstream of BHRF1, possibly reflecting a temporally L or LTIII promoter for BHRF1 (Table 1). A second rightward RNA with a 14- to 15-fold peak increase at 12 to 24 h was detected by two array oligonucleotides (ITG10R, ITG11R) between BOLF1 and BPLF1 (Table 1). A third rightward RNA with an 8- to 12-fold peak increase at 24 h was detected by two array oligonucleotides (ITG13R, ITG14R) between the BSLF1 promoter and the BMLF1 promoter. This RNA peak is substantially later than the upstream BMRF2 RNA and is therefore unlikely to be transcribed from the BMRF2 promoter. A fourth rightward RNA with a four- to eightfold peak increase at 24 h was detected by two array oligonucleotides (ITG15R, ITG16R) between BBLF2 and BBLF3. Since the temporal sequence of detection of this RNA is similar to that of the upstream BBRF2 RNA, the RNA detected by these intergenic array oligonucleotides may be transcribed from the BBRF2 promoter. Another leftward RNA, with a 3- to 4.5-fold peak increase at 24 to 36 h, was detected by two array oligonucleotides (ITG17L, ITG18L) downstream of the BDLF3 poly(A) site. This RNA is temporally similar to BDLF3 and may be a longer transcript from the BDLF3 promoter. A rightward RNA with a three- to fourfold peak at 24 to 36 h was detected by two array oligonucleotides (ITG21R, ITG22R), was similar in temporal sequence to the upstream BTRF1, and may be transcribed from the BTRF1 promoter. A leftward RNA with a three- to fourfold peak at 24 to 36 h was detected by two array oligonucleotides (ITG23L and ITG24L). Another leftward RNA peaked at 12 to 24 h and was detected by an oligonucleotide (ITG25L) that is predicted to hybridize to the putative BVLF1.5. Lastly, rightward intergenic array oligonucleotides (ITG27R, ITG29R, ITG31R, and ITG33R) are downstream of the BART promoter (138,350) (Table 1) (24), are similar in temporal sequence of RNA detection to BART exons, and may be detecting partially spliced BART RNAs (Table 1).
EBV LTIII is induced by EBV replication in Akata cells. A previous study noted LMP1 and LMP2 RNA and protein in the absence of EBNA2 protein expression in a fraction of Akata cells that had been induced to replicate following IgG cross-linking (80). Since the EBV array detected most LTIII RNAs with peak levels at 24 to 36 h after IgG cross-linking, LTIII protein expression was investigated (Fig. 5A and B). Full-length LTIII LMP1 was detected by Western blotting as early as 8 h after IgG cross-linking. LMP1 increased in abundance through 48 h. Similar to earlier findings, D1 LMP1, an N-terminally truncated L LMP1 was not detected (12, 80). EBNALP was first detected at 12 h, whereas EBNA2 and EBNA3C were not detected until 24 h. By 48 h, EBNALP, EBNA2, EBNA3A, EBNA3C, EBNA1, and LMP1 proteins were at levels similar to IB4, a LTIII LCL (Fig. 5A and B). EBV LTIII protein levels by Western blotting did not precisely correlate with RNA levels assayed by array, probably due to the low abundance of most LTIII RNAs, differences in stability among LTIII RNAs and proteins, and the different sensitivities of Western blotting and array assays.
|
Two-color immunofluorescence assays were used to determine if LTIII EBV proteins were expressed in EBV+ Akata cells that had progressed to EBV replication. Cross-linked cells that had bright plasma membrane reactivity with antibody to gp350 at 36 to 48 h after IgG cross-linking also had nuclear reactivity with antibody to EBNA2 and patched plasma membrane reactivity for LMP1, whereas gp350-negative cells were negative for EBNA2 and LMP1 (data not shown). EBV gp350 induction following IgG cross-linking varied from 50% to more than 80%, and LMP1 and EBNA2 reactivity correlated with gp350 reactivity (data not shown). Thus, LTIII EBV proteins are expressed late in EBV replication in Akata cells.
EBV replication effects on cell gene expression. Changes in cell cytoplasmic RNA abundances in EBV+ and EBV– Akata cells after IgG cross-linking were assessed using Affymetrix U133A gene chips. Analysis of variance testing identified 446 array elements representing 401 unique RNAs that significantly changed in abundance (P < 0.01) after IgG cross-linking of EBV+ Akata cells and 134 array elements representing 129 unique RNAs that significantly changed (P < 0.01) after IgG cross-linking of EBV– Akata cells. Only 26 unique genes changed after IgG cross-linking of both EBV+ and EBV– Akata cells. Thus, EBV replication and LTIII gene expression affected cell RNA levels differently and more extensively than IgG cross-linking.
Of 401 unique RNAs that changed significantly after IgG cross-linking, 122 changed at least twofold at one or more time points after IgG cross-linking of EBV+ Akata cells relative to the same time after IgG cross-linking of EBV– Akata. Cluster analysis of the 122 genes grouped most as induced or repressed at several continuous time points (Fig. 6). Although some RNA changes in EBV+ Akata cells relative to EBV– Akata cells after IgG cross-linking were simply magnified or prolonged, most RNA changes were very different in magnitude, duration, or direction (Fig. 6).
|
Some of the 4-, 8-, or 12-
to 24-h IgG cross-linking-induced EBV+ Akata cell
RNAs were also increased in abundance relative to
EBV– Akata cells at 24 to 48 h. These
included TCF20, PLEK, LRMP, MS4A1, PTPN6, TNFAIP3, NFKBIA, DUSP5, SGK,
CD83, TRAF1, IL21R, CD22, WBSCR5, STAT3, CD72, M6PR, and INSIG1 RNAs.
Another group of RNAs increased in abundance at 24 to 48 h
after IgG cross-linking of EBV+ Akata cells relative
to cross-linking of EBV– Akata cells. These RNAs
included OASL, IF135, PHF11, MX2, and IFIT4. The higher levels of these
24- to 48-h RNAs temporally correlated with EBV L and LTIII gene
expression (Fig. 6). For
comparison, array profiles of levels of the IgG cross-linking-induced
or -repressed RNAs in EBV LTIII-infected BL cells or LCLs are compared
with uninfected BL cells in the last 4 lanes of Fig.
7 so as to identify RNAs associated with LTIII gene expression. Many
LTIII RNAs are specifically up-regulated by LMP1 through NF-B
activation.
|
Innate immune response genes that were specifically up- or down-regulated following induction of EBV replication included genes encoding interferon responsive (as noted in the preceding paragraph) and mitogen-activated protein (MAP) kinase RNAs. MAP2K1 (MEK1) RNA was transiently up-regulated at 8 to 12 h in EBV+ and unchanged or repressed in EBV– Akata cells. In contrast, RNAs that encode the p38 activators MAP2K3 (MEK3) and MAP2K6 (MEK6) were up-regulated at 4 to 12 h and down-regulated at 24 to 48 h in IgG cross-linked EBV+ Akata cells but were continuously activated from 4 to 48 h in IgG cross-linked EBV– Akata cells. Further, MAP2K5 (MEK5), an ERK5 regulator, was strongly down-regulated in EBV+ Akata cells at 8 to 12 h and less affected in EBV– Akata cells. Thus, EBV replication destabilizes or represses MAP2K3, MAP2K5, or MAP2K6 RNAs at specific times after IgG cross-linking of EBV+ Akata cells.
RNA differences between EBV+ and EBV– Akata cells. Comparison of time zero control data for EBV-infected and uninfected Akata cells revealed differences between the EBV+ and EBV– Akata cells, which could be related to LTI EBV infection in 95% of the EBV+ Akata cells, to spontaneous EBV replication in 5% of the EBV+ Akata cells before IgG cross-linking, or to differences in cell passage history. Normalization of duplicate array data from 3 EBV+ Akata cell RNAs to the average of duplicate array data from 2 EBV– Akata cell RNAs identified 130 array elements up-regulated and 95 down-regulated at least twofold in EBV-infected versus uninfected Akata cells (Fig. 7). One group of up-regulated RNAs included SP110, SLC1A4, IRF7, OAS2, ISGF3G, C1orf38, TNFRSF7, ISG20, BST2, AICDA, TRIM22, TRPV2, IFI35, FLJ22457, RAC2, MYL6, PSMB9, CTSH, and IFI30, which are induced in LTIII EBV-infected BL41 or BL30 cells and in LCLs, compared to BL cells. These RNAs could be up-regulated in the EBV+ Akata cells because of EBV LTI or because 5% of EBV+ Akata cells are spontaneous replicating EBV, and LTIII is a feature of late EBV replication. LTIII has been associated with higher levels of gamma interferon receptor and gamma interferon up-regulated RNAs, including OAS2, IRF7, ISGSF3G, IFI30, and IFI35 RNAs (16).
A substantial
fraction of the RNAs that were down-regulated in
EBV+ versus EBV– Akata cells were
also down-regulated in LTIII EBV infection. These included TCF3,
STXBP1, LEF1, GPR18, TFIP11, TP73L, OPRS1, SPINT2, TGDS, CDC16, Cullin
4a, FLJ11305, ATF5, PSAT1, ING1, a p53 interacting tumor suppressor,
HK2, KPNB3, NARS, GTF21, EIF4B, E2F5, TOMM20, LRMP, VAMP3, IKBKAP,
MAPRE2, SDCCAG3, SFRS14, and LOXL2. Only a few RNAs that were
down-regulated in EBV+ versus
EBV– Akata cells were up-regulated in LTIII. These
included MS4A1 (CD20), TFRC (transferin receptor, CD71), WSB1 (WD40 and
SOCS-box 1), and MKNK2 (S6k1).
Most RNAs that were up-regulated in EBV+ Akata cells in comparison with EBV– Akata cells were not substantially up- or down-regulated by LTIII infection. TP53 (p53) and SH2D1A (XLP/SAP) RNAs were up-regulated in EBV+ Akata cells and were unaffected by LTIII. These RNAs could be up-regulated by spontaneous EBV replication in 5% of the EBV+ Akata cells, by LTI infection in 95% of EBV+ Akata cells, or by secondary cytokine effects of EBV infection in EBV+ Akata cells. Notably, no gene was induced (or repressed) in both EBV+ and IgG cross-linked Akata cells relative to uninfected and non-cross-linked Akata cells. RNAs that were different in abundance in EBV+ Akata cells or IgG cross-linked Akata cells relative to uninfected and non-cross-linked Akata cells changed in different directions. ISG20, IL10RA, SP110, and OAS2 RNAs were more abundant in EBV+ Akata cells and less abundant in anti-IgG cross-linking of EBV– Akata cells, whereas LRMP, DC12, MS4A1, TCF20, PPP3CC, and ID2 RNAs were less abundant in EBV+ Akata cells and more abundant in IgG cross-linked Akata cells.
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAddendum in proofReferences |
|---|
Several aspects of these data provide unexpected insights into EBV replication. First, following IgG cross-linking, IE BRLF1 RNA was substantially delayed in up-regulation relative to IE BZLF1 or IE/E BMLF1 RNAs. BRLF1 RNA was first detected at 8 h and did not reach peak levels until 12 to 24 h, whereas IE BZLF1 RNA was expressed at 2 h and peaked at 4 h and IE/E BMLF1 was expressed at 4 h and peaked at 8 to 12 h. Earlier BZLF1 expression is consistent with the IgG response elements in the BZLF1 promoter (20, 23, 37, 57, 64, 84). The BRLF1 promoter has both BZLF1 and BRLF1 response elements, and rapid up-regulation of BRLF1 RNA was anticipated (56). Indeed, both BRLF1 and BZLF1 are implicated in BMLF1 up-regulation (10, 14, 32, 39, 50, 61, 64, 90). However, BMLF1 expression preceded that of BRLF1. Although BMRF1, BHLF1, BaRF1, and BLRF2 E RNAs are known BRLF1 targets (5, 6, 26), only BHLF1 RNA levels increased temporally consistent with dependence on BRLF1 expression; BMLF1, BMRF1, and BaRF1 E RNA up-regulation preceded detectable change in BRLF1 RNA levels. However, both BRLF1 array oligonucleotides detected significant BRLF1 RNA from non-cross-linked EBV+ Akata cells and at time zero in the cross-linking experiments. The high level of spontaneous BRLF1 expression may have obscured early increases in BRLF1 RNA. Physiologically significant BRLF1 protein may have been expressed prior to 8 h and contributed to E RNA up-regulation.
Second, cluster analyses of EBV RNA array data resulted in an overall excellent correlation with previous E and L designations based on sensitivity to inhibitors of infected cell protein and viral DNA synthesis. The exceptions were BHRF1, BHLF1, LF3, and BXLF1, which are E RNAs that clustered with L RNAs. BHLF1 and LF3 are part of the EBV oriLyt duplication, share regulatory elements with BHRF1 (35, 83), had peak levels at 36 h, and are likely to be BRLF1 up-regulated. In contrast, BHRF1 and BXLF1 peaked at 12 h. Their clustering among L RNAs was probably driven by significant PAA inhibition. Despite the use of PAA at 300 µg/ml, and almost all EBV RNAs were significantly inhibited, despite adequate specificity to differentiate most E and L RNAs were differentially affected.
Third, although read-through transcription was evident with several intergenic probes, others likely detected novel RNAs. Thus, ITG13R and ITG14R detected RNA similar in abundance to upstream BMRF1 and BMRF2 RNAs, and ITG15R and ITG26R detected RNAs similar in abundance to upstream BBRF2 and BVRF1 RNAs. However, ITG17L detected an RNA that peaked at 24 h similar to upstream BDLF3, but an intervening oligonucleotide, ITG18L, did not detect this RNA, consistent with an alternative promoter or a downstream exon of a spliced RNA. Similarly, ITG29R, ITG31R, and ITG33R are downstream of BdRF1, but the intervening oligonucleotides ITG27R and ITG28R did not detect RNAs, most consistent with a novel RNA. Moreover, oligonucleotide ITG25L detected an RNA with doubling time of 12 h and peak time of 24 h, consistent with an L RNA. ITG25L is downstream of BILF2, which is similar in temporal regulation, but the poly(A) site for BILF2 is more than 2 kb upstream of ITG25L. ITG25L is likely detecting the putative BVLF1.5 RNA, which is part of the Genemark-predicted human herpesvirus 4 (EBV) gp128 homolog of Marmoset lymphocrytovirus ORF10 (78) and KSHV and herpesvirus saimiri ORF18 (3, 82). Indeed hybridization to ITG25L is the first direct evidence that EBV has a BVLF1.5 homologue.
Fourth, IgG cross-linking induction of EBV replication in Akata cells resulted in these arrays detecting most LTIII RNAs. RNAs encoding LMP1 and LMP2A were detected at 8 to 12 h and peaked at 24 to 36 h, whereas EBNALP, EBNA2, EBNA3A, and EBNA3C were detected at 12 to 24 h, with peak expression at 24 to 36 h. EBNALP and EBNA2 expression persisted at 72 h, whereas EBNA3A and EBNA3C were declining, consistent with the possibility that these transcripts may not be initiated from the same promoter. However, the effect is perhaps more likely due to differential stability of the EBNA2/LP mRNAs, during lytic infection. Similarly, LMP1 protein was first detectable at 8 h and was substantially increased through 48 h, whereas EBNALP, EBNA2, and EBNA3C proteins were first detected at 24 h and increased substantially through 48 h. By 48 h, LMP1, EBNA2, and EBNA3C protein levels were at least as high as levels in EBV-transformed lymphoblastoid cells. The detection of LMP2 RNA and of LMP1 protein prior to EBNA2 and EBNALP protein is consistent with the possibility that LMP1 and LMP2 expression at early times is independent of EBNA2 and EBNALP. Overall, LTIII protein levels were markedly inhibited by PAA treatment, consistent with LTIII gene expression being dependent on EBV DNA replication. Additional studies with other inhibitors of DNA synthesis are necessary to confirm that LTIII gene expression is stringently dependent on viral DNA synthesis, given the significant inhibitory effect of PAA on E gene expression in these experiments. Almost all Akata cells that expressed gp350 at 48 h after IgG cross-linking also expressed LMP1 and EBNA2 by immune fluorescence, and almost all LMP1- and EBNA2-positive cells expressed high levels of gp350, physically associating LTIII protein expression with cells in late-stage virus replication. Thus, the data indicate that LTIII gene expression is a feature of late-stage EBV replication in Akata cells.
Fifth, LTIII gene expression late in
lytic infection may be important for EBV replication in B lymphocytes
and other cell types. LMP1 induces NF-B
(38,
44,
55,
65,
70,
81) and has antiapoptotic
effects (18,
55,
81) that likely enhance
cell viability and virus production late in infection
(15-17).
Indeed, LMP1 expression in IgG cross-linked Akata cells was temporally
associated with TRAF1 and A20 induction, both up-regulated by LMP1
activation of NF-B
(25,
34). Moreover, LMP1 has
recently been directly implicated in enhancing virus egress from Akata
cells (2). Most
importantly, LTIII gene expression has been described, in vivo, in foci
of EBV replication in epithelial cells
(102,
103). Thus, LTIII gene
expression may be a general feature of EBV late replication in
lymphocytes and epithelial cells. This may contribute to the antigenic
stimulus that maintains life-long high-level T-cell reactivity to
LTIII-infected cells (7,
51,
58,
74).
Overall, latency-associated RNA expression late in replication is reminiscent of HSV latency-associated transcripts, which are expressed late in HSV replication (59, 88, 91). These data are therefore consistent with an orthologous strategy of L and LT gene linkage to provide enhance lytic and latent cell survival, similar to that of HSV latency-associated transcripts (47, 72, 73).
Sixth, EBV
gene products had profound effects on cell gene expression at early and
late times after activation of virus replication in Akata cells. Of 401
cell genes that changed significantly following IgG
cross-linking-mediated induction of replication, only 26 gene changes
were attributable to IgG cross-linking. Most of the EBV
replication-related changes in cell RNA abundance are likely due to
virus-encoded transcription factors or virus-encoded activators of cell
signal transduction pathways. EBV replication up-regulated YKT6, VAMP1,
and RCP RNAs, which encode proteins involved in vesicle transport and G
protein signaling. These cell proteins may be important for EBV
envelopment or egress. Also, RNAs encoding 3 serpins (APLP2, SERPINB1,
and SERPINB9) that inhibit serine proteases found in the cytotoxic
granules of T lymphocytes and neutrophils were induced; these serpins
may contribute to immune evasion during lytic replication.
Interestingly, murine gammaherpesvirus 68 encodes for its
own serpin, M1 (100),
while other gammaherpesviruses do not. EBV replication was also
associated with the induction of RNAs encoding antiapoptotic proteins
that regulate the NF-B and ubiquitin pathways, including A20,
SIAH2, TRIM38, and SQSTM1, as well as regulators of cell gene
transcription, STAT3, NFATC1, ATF2, EGR1, and TCF20.
In part, EBV effects on cell RNA abundance may be secondary to the induction of MAP kinases that affect message stability. Lytic replication was associated with increased MAP2K1, MAP2K3, MAP2K6, and MAP3K12 RNAs 4 to 12 h after induction. In contrast, MAP2K5 decreased at early times and increased with LTIII. Further, EBV LMP1 activates p38 (27), which can result in stabilization of RNAs that have A/U-rich elements (101) and contribute to up-regulation of TNFAIP3, CCL3, and CCL4 RNA levels (33, 63).
Seventh, most
cell RNAs that change significantly in abundance during EBV replication
in Akata cells have not been noted to be affected by other
herpesviruses in other cells. Much of the difference may be due to the
different cell types in which replication has been studied. However,
EBV, HSV, and CMV replication all activate NF-B and induce
TNFAIP3 (A20) and NFKBIA (IKB)
(13,
19,
86,
94,
108). Further, EBV and
KSHV replication induce DUSP5
(36,
67). Moreover, CMV and
EBV replication up-regulate interferon-stimulated genes MxB, OAS1,
OAS2, OASL, and IFIT4
(13,
19,
108); EBV replication
initially down-regulates these genes but up-regulates them late in
infection when EBV LTIII genes are expressed. Also, EBV and HSV
regulate POLG (DNA polymerase 
) and E2F-1 RNAs in opposite
directions; EBV down-regulates and HSV up-regulates both RNAs
(94).
Eighth, although EBV replication is similar to that of other herpesviruses in profound reorganization of cell chromatin, massive synthesis of viral DNA, RNA, and protein, and changes in cell membranes (53), a most surprising aspect of this study is the extent to which Akata cells very late in EBV replication are viable by dye exclusion and responsive to changes in EBV gene expression with up and down-regulated changes in cell RNAs. The infected cell replicating EBV remains highly functional late in EBV replication.
|
ADDENDUM IN PROOF |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAddendum in proofReferences |
|---|
|
ACKNOWLEDGMENTS |
|---|
We thank Lynn Enquist, Eric Johannsen, Patrick Moore, Bernard Roizman, and Tom Shenk for helpful discussions. We thank Kenzo Takada for the Akata cell line. We thank Xiaowei Wang for selecting oligonucleotides for the EBV array and Glen Short and Jocelyn Burke for technical support with the EBV array.
|
FOOTNOTES |
|---|
These authors contributed equally to this research.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAddendum in proofReferences |
|---|
