![[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, January 2007, p. 575-587, Vol. 81, No. 2
0022-538X/07/$08.00+0 doi:10.1128/JVI.01725-06
Copyright © 2007, American Society for Microbiology. All Rights Reserved.
Molecular Biology Institute, Department of Microbiology, Immunology and Molecular Genetics, University of California, Los Angeles, California
Received 9 August 2006/ Accepted 23 October 2006
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Much information has accumulated that is relevant to the functions of E1B-55K and E4orf6. The two viral proteins associate in vivo (68) and colocalize in Ad5-infected cell nuclei at viral DNA replication-transcription centers, as well as other regions of the nucleoplasm and cytoplasm during the late phase of infection (59). Both E1B-55K and E4orf6 have nuclear export signals and nuclear localization signals required for shuttling between the nucleus and the cytoplasm (22, 32, 82). E1B-55K also has RNA binding activity in vitro, although no specificity was observed for binding either host cell or viral mRNA (26). These results suggested that E1B-55K and E4orf6 might be directly involved in redirecting both viral and cytoplasmic mRNA nuclear export in infected cells with the viral proteins directly binding viral mRNPs and transporting them through nuclear pore complexes.
A major advance
in understanding how E1B-55K and E4orf6 inactivate p53 during
adenovirus infection came from the discovery that these two proteins
associate with the cellular proteins elongins B and C,
cullin 5, RBX1, and additional cellular proteins to form an
800-kDa ubiquitin-protein ligase complex (often referred to as
E3) that polyubiquitinates p53
(37,
65). This adenoviral
ubiquitin-protein ligase complex (Ad-Ub protein ligase) is also
implicated in the degradation of MRN subunits since MRN complexes, in
addition to p53, are not degraded by proteasomes in cells infected with
a mutant altered in either E1B-55K or E4orf6
(78). Also, E1B-55K binds
the MRN complex even in the absence of E4orf6, indicating that, as for
p53, E1B-55K is the primary substrate-binding subunit of the viral
ubiquitin-protein ligase
(10,
37,
50).
In the absence of E4orf6 and E4orf3, the cellular MRN complex is recruited to viral DNA replication centers (10, 24). The linear double-stranded adenoviral genomes act as targets for the MRN complex (10, 24, 53, 75, 78, 81). This repair complex, when uninhibited, functions with additional nonhomologous end-joining DNA-double-strand-break repair proteins to concatemerize adenoviral genomes, making them too long to package into adenovirus capsids (24, 53, 78, 81). The MRN complex also interferes with viral DNA replication even when concatemerization of the viral genomes is blocked in mutant cells with defects in other cellular proteins required for double-strand break repair (24, 75). This may be because exo- and endonuclease activities associated with the MRN complex (19, 62) inhibit initiation of viral DNA synthesis at the termini of the viral genome. E1B-55K inactivates the MRN complex in two ways. First, it binds the nuclear MRN complex, causing it to be exported from the nucleus into cytoplasmic centers of protein ubiquitination and proteosomal degradation called aggresomes (50). Araujo et al. (1) reported that E4orf3 also induces nuclear export of the MRN complex to aggresomes. Second, the Ad-Ub protein ligase complex targets MRN subunits for degradation by 26S proteasomes (10, 50, 78). These recent results suggest alternative possible explanations for why E1B-55K and E4orf6 associate with viral DNA replication centers and shuttle between the nucleus and cytoplasm. The viral proteins may associate with viral DNA replication centers because of their association with MRN complexes recruited to viral DNA replication centers by binding to the ends of viral DNA. Nucleus-cytoplasmic shuttling (21, 22, 45) may be a consequence of their function in exporting MRN complexes and p53 from the nucleus to cytoplasmic aggresomes (1, 50).
The
observation that >95% of the E1B-55K in Ad5-infected cells
early in the late phase of infection is associated with the
800-kDa ubiquitin-protein ligase complex
(37) suggests that all of
the E1B-55K and E4orf6 functions, including stimulation of viral late
mRNA nuclear export and translation and inhibition of host mRNA nuclear
export, as well as degradation of p53 and MRN complex subunits, are due
to the viral ubiquitin-protein ligase activity. Consistent with this
hypothesis, treatment of Ad5-infected cells with a protease inhibitor
results in a decrease in viral late protein synthesis similar to that
observed during infection with an E1B-55K or E4orf6 mutant, although
the accompanying decrease in viral DNA replication complicated the
interpretation of these results
(14). To test this
hypothesis further, we analyzed the influence of a dominant-negative
mutant form of CUL5, the host cell cullin incorporated into the
SCF-like Ad-Ub protein ligase, during the course of adenovirus
infection. We find that expression of dominant-negative CUL5 under
conditions that inhibit the degradation of p53 and MRE11 produces a
phenotype for viral late mRNA nuclear export and translation similar to
that of an E1B-55K deletion mutant. These results strongly support the
hypothesis that E1B-55K and E4orf6 influence late mRNA nuclear export
and translation by functioning as an SCF-like ubiquitin-protein
ligase.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Construction of E1-substituted Ad5 vectors.
A modified pAdlox shuttle plasmid and
Ad5 
5 vector were used to construct E1-substituted Ad5
recombinants by Cre-mediated recombination between loxP sites
in the 5 vector and the pAdlox plasmid
(38). To enhance the
expression of an inserted gene, a simian virus 40 (SV40) small-T intron
and woodchuck hepatitis virus posttranscriptional regulatory element
(51,
85) were inserted into
the pAdlox multiple cloning site of pAdlox. Inserted coding regions for
CUL5 NTD or CUL5 FL were inserted into the modified pAdlox (pAdlox*)
plasmid so that transcripts from the cytomegalovirus immediate-early
promoter include the FL or NTD coding region followed by an SV40
small-T intron, the woodchuck hepatitis virus posttranscriptional
regulatory element, and an SV40 poly(A) site. A 2-µg portion of
pAdlox*-CUL5 NTD or pAdlox*-CUL5 FL linearized with SfiI plus 1
µg 5 vector DNA was transfected into 293Cre4 cells
(38) with QIAGEN
Effectene. Recombinant 5 vectors were plaque purified and
propagated in 293 cells and assayed by plaque titration on 293
cells.
Dominant-negative cullin 5 (CUL5 NTD).
CUL5 cDNA
was cloned by reverse transcription (RT)-PCR from HeLa cytoplasmic RNA
with primers encoding a Flag epitope at the amino terminus, as well as
a SalI site preceding the start codon and a BamHI site following the
stop codon. For the CUL5 NTD, a second reverse primer was used
with a stop codon following amino acid 393. The full-length CUL5 (FL)
and CUL5 NTD sequences were TA cloned into pCR2.1 (Invitrogen) before
transfer to pAdlox*. The mutations V341R and L344D were introduced into
pAdlox*-NTD (Stratagene site-directed mutagenesis kit). The CUL5 coding
regions of pAdlox*-FL and pAdlox*-NTD were sequenced before
introduction into 5 vectors (Table
2).
|
|
5 vectors at 10 PFU/cell. At 12 h later,
the same cells were infected again (superinfected) with Ad5 or
dl1520 at 100 PFU/cell. At 18 h postsuperinfection,
cells were harvested by low-speed centrifugation, washed twice with
ice-cold PBS, and lysed with 50 mM Tris HCl (pH 7.4)-1%
NP-40-150 mM NaCl-1 mM EDTA (lysis buffer) with
rotation at 4°C for 30 min. Lysates were freeze-thawed several
times and centrifuged at maximum speed in a Beckman Coulter
microcentrifuge at 4°C for 15 min, and supernatants were
centrifuged at 100,000 x g in a VTi65 rotor for
1.5 h at 4°C. Protein lysates were quantified by
Bradford assay (Bio-Rad). A 2.4-mg portion of protein per
immunoprecipitation was precleared with Amersham Biosciences Sepharose
A beads (50 µl packed beads) and Santa Cruz Biotechnology mouse
immunoglobulin G (IgG)-conjugated beads (10 µl packed beads)
for 4 to 7 h. The suspension was centrifuged at 8,000 rpm in
a Beckman Coulter microcentrifuge for 30 s, and the
supernatant was removed and incubated on a Nutator overnight at
4°C with 2A6-Sepharose A, Sigma M2 anti-Flag-conjugated beads,
or mouse IgG-conjugated beads as a control (25 to 30 µl packed
beads). The suspension was centrifuged as before, and beads were washed
two times with lysis buffer and two more times with wash buffer (50 mM
Tris HCl, pH 7.4, 150 mM NaCl). M2 elution was performed as described
by the manufacturer (Sigma), with 3x Flag peptide for elution.
2A6-Sepharose A and mouse IgG-conjugated beads were eluted by
incubation at 100°C for 10 min in 3x Laemmli sample
buffer.
Analysis of viral nucleic acids and proteins in superinfected cells.
Subconfluent 10-cm HeLa plates were
infected with the respective 5 vectors at 50 PFU/cell
(preinfection). At 16 h later, the cells were infected again
(superinfected) with Ad5 or dl1520 at 40 PFU/cell. At 16, 24,
36, or 48 h postsuperinfection, cells were washed once with
PBS and trypsinized with GIBCO 0.5% trypsin. Trypsin was inactivated
with 10% FBS DMEM, and the cells were gently pelleted, washed again
with ice-cold PBS, resuspended in 1 ml ice-cold PBS, and aliquoted as
follows: 200 µl for DNA isolation, 400 µl for protein
extraction, and 400 µl for nuclear and cytoplasmic RNA
extraction.
DNA isolation. Cells were lysed in 1 ml NP-40 lysis buffer (10 mM Tris, pH 8.0, 150 mM NaCl, 1.5 mM MgCl2, 0.65% NP-40) for 30 min to 1 h on ice. Nuclei were pelleted at 2,000 rpm in a microcentrifuge for 2 to 3 min. Nuclei were washed once in 1 ml NP-40 lysis buffer and pelleted as before. DNA was isolated with the QIAGEN DNeasy kit with addition of RNase A in the first step of isolation.
Cytoplasmic and nuclear RNA isolation. Cells were lysed in buffer RLN (50 mM Tris, pH 8.0, 140 mM NaCl, 1.5 mM MgCl2, 0.5% NP-40, 1,000 U/ml RNase inhibitor) for 5 min on ice. Nuclei were pelleted at 2,000 rpm for 3 min at 4°C in a microcentrifuge. RNA was isolated from the supernatant (cytoplasmic fraction) with the QIAGEN RNeasy kit, and cytoplasmic RNA was eluted in 100 µl with the provided RNase-free water. Nuclei were washed with 1 ml RLN buffer and pelleted as before. The supernatant was carefully aspirated, and nuclear RNA was extracted with Invitrogen TRIZOL reagent as described by the manufacturer. The nuclear RNA pellet was dissolved in 30 µl diethyl pyrocarbonate-treated water.
RT. Two micrograms of RNA was first treated with Invitrogen amplification grade DNase I as described by the manufacturer. The final volume after addition of EDTA was 11 µl. Six microliters was used for transcription with the Invitrogen Superscript III kit, while 5 µl was saved for the RT negative control. To control for transcription efficiency, 4 x 1010 copies of the Promega 1.2-kb kanamycin polyadenylated RNA control (catalog no. C1381) was spiked into each sample prior to DNase I digestion. RT was primed with oligo(dT) to select for processed mRNA. RT was performed as described by the manufacturer with the kanamycin primer sequence provided by the manufacturer.
Quantitative PCR.
An Applied
Biosystems 7500 real-time PCR system was used for the quantitative
RT-PCR. ABI SYBR green PCR master mix was used. A 2.5-µmol
portion of each primer was added to the reaction mixture, followed by 5
µl of diluted template in a total reaction volume of 25
µl. Quantitative PCR was done in triplicate for each sample by
denaturing at 95°C for 10 min, followed by 40 cycles of
95°C for 15 s and 60°C for 45 s.
For absolute quantification, a standard curve was generated for the
cellular translation elongation factor EF2 promoter region cloned into
pCDNA3.1 (Invitrogen TA Cloning Kit). 5 vector DNA isolated
from purified virions was used for the standard curves of L2, L3, and
L5. Sets of primers (Table 2) for individual quantitativeRT-PCR of L2, L3, and L5 cDNAs were designed to amplify the sequence
from the common 3' end of all processed RNAs from each of these
late regions. Raw numbers were normalized to equal internal control
kanamycin RNA. Standard deviations were calculated by the best value of
the sum or quotient.
For each independent experiment, each treatment was done in duplicate. Transcripts are presented as the number per cell, while viral genomes are presented as the total number of adenovirus genomes per cellular EF2 gene.
In vivo 35S labeling.
Subconfluent 60
mM HeLa plates were infected as already described. At 2 h
before harvesting, cells were washed twice with PBS and then 770
µl of GIBCO (–)L-cysteine,
(–)L-methionine, (–)L-glutamine
medium was added to each plate, which was incubated for 15 min before
addition of 100 µCi of Tran35S-Label (MP, formerly
ICN Biomedicals, Inc.) and incubation for 2 h. Cells were
washed twice again in PBS, and protein lysates were harvested as
already described. Protein concentration was determined, and equal
amounts of protein were analyzed by SDS-PAGE in a 10% gel. Gels were
dried, counted, and imaged with a Molecular Dynamics
PhosphorImager.
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Using
an Ad5 E1A-E1B-deleted expression vector, 
5
(38), we expressed a
Flag-tagged N-terminal region of CUL5 that terminated following residue
384, homologous to residue 410 in CUL1. Pavletich and coworkers had
expressed residues 1 to 410 of CUL1 in Escherichia coli in
their work on the structure of human SCF
(90). In preliminary
immunoprecipitation experiments, we found that the residue 1 to 384
CUL5 N-terminal region did not interact well with an
E1B-55K/E4orf6/elongin BC complex in HeLa cells coinfected with WT Ad5
and that much of the truncated CUL5 expressed in HeLa cells appeared to
be insoluble. Pavletich and coworkers reported that the equivalent
N-terminal fragment of CUL1 expressed in E. coli also has poor
solubility (90). To
increase solubility sufficiently for purification and crystallography,
Zheng et al. (89)
constructed an N-terminal domain V367D L371D double mutant in residues
on the surface of the N-terminal domain that interact with hydrophobic
residues in the C-terminal domain at the interface between the two
domains (Fig. 1D).
Consequently, we constructed a V341D L344D double mutant of the CUL5
N-terminal domain at CUL5 residues that align best with equivalent CUL1
residues V367 and V370 (henceforth called CUL5 NTD or simply
NTD).
We reasoned that CUL5 NTD would have to be expressed at a
high concentration relative to endogenous CUL5 to compete for binding
to the viral ubiquitin ligase. To determine if this can be achieved
with these vectors, HeLa cells were infected with the NTD vector
(5-NTD) or, as controls, with the empty 5 vector
(5) or the same 5 vector expressing Flag-tagged
full-length WT human CUL5 (5-FL). Extracts were prepared at
36 h postinfection (hpi) and subjected to Western blotting
with an anti-CUL5 antiserum (Fig.
2A). The full-length CUL5 vector produced approximately 10 times the
endogenous level observed in the extract from empty-vector-infected
cells (Fig. 2A, a band of
full-length CUL5 in the vector lane was apparent at longer exposure).
The NTD was also expressed at a far higher level than endogenous CUL5.
It was observed as a doublet in the Western blot, probably because of
partial proteolysis at the N terminus since antibody to the N-terminal
Flag epitope detected only one NTD band. (The doublet probably is not
due to NEDD8 conjugation to the NTD, since NEDD8 is bound to the
C-terminal domains of cullins
[61].)
|
5-NTD or 5-FL for
12 h to allow expression of Flag-tagged NTD or FL CUL5, and
then the same cells were infected again (superinfected) with either WT
Ad5 or the E1B-55K null mutant dl1520
(5). Since the 5
vector is deleted for E1A and E1B, no E1B-55K was expressed during the
initial 12 h, and because of the absence of E1A, the
expression of other viral genes was expected to be greatly diminished
and delayed compared to infection with WT Ad5
(7,
42). Following
superinfection, E1A expressed from the superinfecting WT Ad5 or
dl1520 genome was expected to stimulate transcription from
both the 5 vector and the superinfecting genome, leading to
the replication of both viral genomes and expression of late viral
proteins from both the 5 vector and superinfecting genome.
Superinfection with dl1520 should have resulted in expression
of the same viral genes as in cells superinfected with WT Ad5, except
for the absence of E1B-55K
(28). Extracts prepared
when the superinfected cells reached the late phase of infection (18
hpi with WT Ad5), were subjected to immunoprecipitation and Western
blotting. As expected, anti-E1B-55K monoclonal antibody 2A6
(
-55K) precipitated E1B-55K, E4orf6, elongin C, full-length
CUL5, and RBX1 from cells superinfected with 5-FL (Fig.
2B). None of these
proteins were immunoprecipitated from cells infected with 5-FL
and dl1520 because of the absence of E1B-55K. Significantly,
E4orf6, elongins B and C, and the CUL5 NTD but no detectable RBX1
coprecipitated with anti-E1B-55K from cells preinfected with
5-NTD (Fig. 2B,
-55K). When immunoprecipitation was done with antibody to the
Flag epitope on overexpressed FL or NTD CUL5 (-Flag), high
levels of Flag-FL CUL5 were immunoprecipitated along with high levels
of RBX1, which is intimately associated with the C-terminal domain of
cullins (Fig. 1A). As
expected, no RBX1 coimmunoprecipitated with the Flag-tagged NTD protein
(Fig. 2B). Both the FL and
NTD CUL5 proteins bind elongin C, E4orf6, and E1B-55K, although the
levels of all of these proteins were lower in the -Flag
immunoprecipitates from cells expressing NTD CUL5 compared to FL CUL5.
These results may suggest that FL CUL5 with bound RBX1 binds the
elonginBC/E4orf6/E1B-55K complex with higher affinity than the NTD
protein does. Finally, both FL and NTD CUL5 bound elongin C and E4orf6
in the absence of E1B-55K (cells superinfected with dl1520),
as observed in insect cells coinfected with baculovirus expression
vectors for these proteins
(65). These results are
precisely those expected if the NTD binds to the
elonginBC/E4orf6/E1B-55K complex observed earlier
(65; Fig.
1B); however, binding by
NTD CUL5 may be reduced compared to binding by FL CUL5.
To
determine if expression of CUL5 NTD inhibits the ubiquitin ligase
activity of the Ad5-Ub ligase, we investigated whether overexpression
of the NTD inhibited the degradation of host cell p53 and MRE11. p53 is
the one well-established substrate of the Ad5-Ub ligase because it is
polyubiquitinated by the purified complex in vitro
(37,
65). Proteasomal
degradation of the MRE11 and RAD50 subunits of the MRN complex requires
both E1B-55K and E4orf6, indicating that at least one of these proteins
is very likely a second substrate of the Ad5-Ub ligase
(78). Duplicate cultures
of HeLa cells were initially infected with the 5-NTD,
5-FL, or empty 5 vector. Sixteen hours later, the six
cultures were all superinfected with WT Ad5. At 48 hpi with WT Ad5,
extracts were prepared from these and uninfected cells cultured in
parallel and Western blotted for p53 and MRE11, as well as CAND1 (a
cellular protein that binds and inactivates cullins
(89) and ß-actin
as loading controls (Fig.
2C).
In HeLa cells, the endogenous level of p53 is low to undetectable by Western blotting because this human cervical carcinoma cell line expresses the human papillomavirus type 18 (HPV18) E6 protein from the integrated viral genome (72). E6 forms a ubiquitin-p53 ligase with host cell protein E6AP (55). However, upon infection with Ad12, p53 is initially stabilized and then degraded (34). In HeLa cells infected with an E4orf6 mutant that does not assemble the Ad5-Ub ligase, p53 is bound by E1B-55K and accumulates to a high level in both the nucleus and cytoplasmic aggresomes, as it does in Ad5-transformed cells that lack early region 4 (50, 88). This is probably because E1B-55K, HPV18 E6, and the principal cellular ubiquitin-p53 ligase MDM2 (8) compete for binding to the N-terminal region of p53 (43, 49). In the absence of E4orf6, E1B-55K binds and stabilizes p53 (48), probably by sequestering it from the HPV E6 and MDM2 ubiquitin-p53 ligases. p53 also accumulated to a high level in the cultures preinfected with the NTD vector (Fig. 2C). However, it was not detectable in cells preinfected with the empty vector or the FL vector. Presumably, p53 in cells infected with the empty and FL CUL5 vectors was polyubiquitinated by both the HPV E6- and Ad-Ub ligases and degraded by proteasomes. However, inhibition of the Ad-Ub protein ligase by overexpression of the NTD resulted in p53 stabilization rather than degradation, as in cells infected with an E4orf6 mutant that expresses E1B-55K but cannot assemble the Ad-Ub ligase in the absence of E4orf6. Similarly, MRE11 was degraded in cells preinfected with the empty and FL vectors but was only partially reduced in cells preinfected with the NTD vector (Fig. 2C). These results indicate that high-level expression of the CUL5 NTD inhibits Ad-Ub protein ligase activity.
Viral genome replication and early viral protein expression are not inhibited by CUL5 NTD.
Early viral mRNAs,
proteins, and viral DNA replication are near WT Ad5 levels in E1B-55K
mutant-infected HeLa cells
(25). Similarly, we found
that E1A, the 72-kDa DNA-binding protein, and E1B-55K were expressed at
similar levels at early times after Ad5 infection in cells preinfected
with the 5-NTD, 5-FL, and empty 5 vectors
(Fig.
3A). In this experiment, E4orf6 accumulated to a higher level in cells
preinfected with the NTD vector, but in most experiments E4orf6 was
expressed at similar levels in cells preinfected with the empty
5 vector, 5-NTD, or 5-FL (data not shown).
Viral DNA also accumulated to similar levels in cells preinfected with
each of these 5 vectors and then analyzed at 16, 24, and
36 h post Ad5 superinfection, as determined by quantitative
PCR (Fig.
3B).
|
To
determine if the E1B-55K/E4orf6 function that stimulates viral late
protein synthesis in HeLa cells requires the ubiquitination activity of
the Ad-Ub protein ligase complex, duplicate cultures of HeLa cells were
infected with the empty 5 vector, 5-NTD, or
5-FL. At 16 hpi with these vectors, all cultures were
superinfected with WT Ad5 and extracts were prepared at 24 hpi with WT
Ad5 and subjected to Western blotting. Western blots were probed with a
rabbit antiserum raised against purified, disrupted Ad5 virions and
CAND1 as an internal control (Fig.
4A). For comparison, HeLa cells were infected with WT Ad5 or
dl1520 and extracts were prepared 24 h after
infection and Western blotted with the same anti-Ad5 virion serum and
anti-ß-actin as a loading control (Fig.
4B). The level of virion
proteins expressed in HeLa cells preinfected with the NTD vector to
inhibit the ubiquitin ligase activity of the Ad-Ub protein ligase
was reduced comparably to that observed in
dl1520-infected compared to WT Ad5-infected HeLa cells.
Similar results were also observed in A549 cells and in HeLa cells
preinfected at 32°C for 16 h with 5 vectors
and 32 h postsuperinfection with Ad5 (data not shown). In
multiple repeats of this experiment, the reduction of viral proteins
was quantitated by performing serial dilutions of the sample from cells
preinfected with the empty 5 vector on the same Western blot
as the sample from cells preinfected with the 5-NTD vector.
Western blots were also analyzed with a quantitative imaging system
(see Materials and Methods). Late viral proteins were reduced from
twofold to eightfold, depending on the viral late protein, pVII
generally being decreased the most and hexon the least. In some
experiments, preinfection with the 5-FL vector inhibited viral
late protein expression to a small extent compared to cells infected
with the empty vector (Fig.
4A), but this was not
observed in most experiments. It might be due to a modest disturbance
in elongin B- and C-based ubiquitin-protein ligase function resulting
from CUL5 overexpression and the resulting change in stoichiometry of
CUL5 to its interacting partners. But the reduction in viral late
protein expression was reproducibly much greater in cells preinfected
with the 5-NTD vector than in cells preinfected with the
5-FL vector.
|
5
vectors as before and then superinfected with either WT Ad5 or
dl1520. We reasoned that if the decreased virion protein
synthesis in cells preinfected with the NTD vector was due to a
generalized defect in cellular protein synthesis, we should observe a
decrease in the low level of virion protein synthesis in
dl1520-infected cells relative to cells preinfected with the
empty vector. Further, the magnitude of this difference would be
comparable to the decrease in late gene expression observed in WT Ad5
superinfection. Alternatively, if the decreased virion protein
synthesis in 5-NTD-preinfected cells superinfected with WT Ad5
was due to specific inhibition of the Ad-Ub ligase and not a
nonspecific inhibition of protein synthesis, we should observe no
significant further decrease in virion protein expression in cells
superinfected with dl1520 because the Ad-Ub ligase is not
expressed in cells infected with this E1B-55K null mutant.
In
fact, no significant further decrease in virion protein expression
occurred in HeLa cells preinfected with the 5-NTD vector
compared to the empty vector or 5-FL and superinfected with
dl1520 (Fig. 4C).
In the control empty vector preinfected cells, late viral protein
expression was decreased in dl1520-superinfected compared to
WT Ad5-superinfected cells, as expected, because of the absence of
E1B-55K. In the dl1520-superinfected cells, the level of viral
late protein was not significantly diminished further by preinfection
with 5-NTD as it was in cells superinfected with WT Ad5 (Fig.
4C). Similar results were
observed in three independent experiments. The same result was observed
again in the pulse-labeling experiment of Fig.
6. Consequently,
preinfection with the 5-NTD vector does not reduce viral late
gene expression through some nonspecific inhibitory effect of the CUL5
NTD. These results strongly support the model in which E1B-55K and
E4orf6 stimulate viral late protein synthesis by functioning as a
CUL5-based protein-ubiquitin
ligase.
|
5 vectors or the empty vector as before and
superinfected with WT Ad5 16 h later. Cytoplasmic RNA was
isolated at 24 hpi with WT Ad5 and analyzed with primers for the
3' ends of the L2, L3, and L5 mRNAs
(74) (Fig.
5A). For comparison, HeLa cells were also infected with WT Ad5 and
dl1520 without vector preinfection; RNA was isolated at 24 hpi
and analyzed in the same way (Fig.
5B).
|
5-NTD vector caused a similar reduction in viral
cytoplasmic late RNAs compared to control cells preinfected with the
empty 5 vector (Fig.
5A), although the decrease
in viral late cytoplasmic RNA expression was not as great
in dl1520-infected as in WT Ad5-infected cells (compare Fig.
5A with B). This may be
because degradation of target proteins is not completely blocked by
overexpression of the CUL5 NTD, as observed from the partial
degradation of MRE11 compared to mock-infected cells (Fig.
2C). In contrast, there is
no activity of the absent Ad-Ub ligase in dl1520-infected
cells. In dl1520-infected cells, where there is no
E1B-55K, L2 cytoplasmic RNA was not decreased relative to that in WT
Ad5-infected cells as much as the L3 and L5 cytoplasmic RNAs (Fig.
5B and D). We do not know
why the inhibition of late mRNA expression is greater for L3 and L5
than for L2 (Fig. 5B) and
L4 (data not shown; also evident in reference
64). However, the
decreased effect of the NTD on L2 expression is consistent with the
phenotype of an E1B-55K mutant, although less severe, presumably
because of residual Ad-Ub ligase activity. Earlier work has shown that the decrease in viral late mRNAs in E1B-55K mutant-infected HeLa cells is due at least in part to decreased mRNA nuclear export compared to that observed in WT Ad5-infected cells (6, 25, 35, 64). This phenotype can be observed as a decrease in the ratio of cytoplasmic viral RNA to nuclear viral RNA in E1B-55K mutant-infected HeLa cells (30), although interpretation of this simple ratio is complicated by the fact that fully processed nuclear mRNAs that are not exported are subject to degradation (23).
To determine if a similar defect in viral late mRNA nuclear export results from overexpression of the CUL5 NTD, nuclear RNA was also isolated from the same cells from which cytoplasmic RNA was analyzed in Fig. 5A and B. Nuclear L2, L3, and L5 viral RNAs were assayed by quantitative RT-PCR, and the number of cytoplasmic RNA molecules per cell was divided by the number of nuclear RNA molecules per cell. In WT Ad5-infected HeLa cells at 24 hpi, this ratio was 2 to 3. In HeLa cells infected with the E1B-55K mutant dl1520, the ratio in L3 and L5 dropped to 0.5 to 1 but no significant change in the ratio of cytoplasmic to nuclear viral late RNA was observed for L2 RNA (Fig. 5D). This difference for L2 compared to L3 and L5 has not been described previously but may be due to differences in details of the experimental conditions used (e.g., 32°C versus 37°C) that may affect processes such as the rate of degradation of nuclear RNAs that are not exported (23) or the rate of degradation of mRNAs in the cytoplasm, both of which would influence the ratio of cytoplasmic to nuclear RNA.
Precisely the same
pattern of changes in the ratio of cytoplasmic to nuclear viral late
RNA was observed in HeLa cells preinfected with the 5-NTD
vector compared to the empty and 5-FL vectors (Fig.
5C and D). The ratio was
lower for L3 and L5 but not significantly changed for L2 (Fig.
5D). These results
strongly support the hypothesis that the stimulation of viral late mRNA
nuclear export produced by E1B-55K and E4orf6 requires the
ubiquitin-protein ligase activity of the Ad-Ub protein ligase built
from basic cellular components of a major class of cellular
ubiquitin-protein ligases (the CUL5 class), with E4orf6/E1B-55K
substituted for cellular substrate-targeting subunits such as Skp2/Cks1
in SCF (Fig. 1A and
B).
Host protein synthesis inhibition is only partially dependent on the viral ubiquitin-protein ligase.
E1B-55K mutants fail
to inhibit host protein synthesis late in infection, as is observed
with WT Ad5 (25). To
analyze the influence of the Ad-Ub ligase activity on this E1B-55K
function, the same infection scheme used in Fig.
5 was repeated, except
that cells were pulse-labeled with [35S]methionine-cysteine
34 to 36 h postsuperinfection. Total cell
protein was isolated, the same amount of protein from each
infection was subjected to SDS-PAGE, and the dried gel was counted with
a phosphorimager (Fig.
6A). Consistent with analysis of late virion protein expression by Western
blotting (Fig. 4A and data
not shown for times earlier and later postinfection), pulse-labeling
showed that preinfection with 5-NTD inhibited the rate of
viral late protein synthesis compared to cells preinfected with the
empty 5 vector (Fig.
6A, compare lanes 4 and
5). Overexpression of FL CUL5 in 5-FL-preinfected cells (lane
6) caused much less inhibition. Also, in cells superinfected with the
dl1520 E1B-55K mutant instead of WT Ad5, viral late protein
synthesis was inhibited only minimally and equally by overexpression of
either CUL5 FL or NTD (Fig.
6A, lanes 7 to 9). These
results once again strongly support the model in which Ad5
ubiquitin-protein ligase activity stimulates viral late protein
synthesis (by an unknown mechanism; see Discussion).
Preinfection
of HeLa cells with the 5 empty vector (Fig.
6A, lane 4) resulted in a
slightly higher rate of translation of viral late proteins in
Ad5-superinfected HeLa cells than observed in cells that were mock
preinfected and then superinfected with WT Ad5 (lane 2). This increase
was even greater comparing dl1520-superinfected cells that
were mock preinfected or preinfected with the empty 5 vector
(lane 3 versus lane 7). This may have resulted from early entry and
uncoating of the E1A-E1B deleted 5 vector genome in the
preinfected cells. This vector DNA would be available for transcription
and replication as soon as E1A was expressed in cells superinfected
with either WT Ad5 or dl1520. This may have led to a somewhat
faster time course of infection when empty-vector-preinfected cells
were superinfected compared to cells that did not contain the genome of
a vector with E1A-E1B deleted in their nuclei before infection with WT
Ad5 or dl1520. Indeed, we also observed a slightly lower viral
DNA concentration in mock-preinfected cells than in empty 5
vector-infected cells at 16, 24, and 36 h postsuperinfection
(data not shown). However, this effect of preinfection with 5
vectors on viral late protein synthesis was much less than the effect
of NTD expression observed by comparing cells preinfected with
5-NTD and the most appropriate control, cells preinfected with
the empty 5 vector (Fig.
6A, compare lanes 4
and5).
The inhibition of host protein synthesis by WT Ad5 at 36
hpi is evident in the greatly decreased number of counts in host
proteins that migrate to positions throughout the length of the lane in
the mock-infected cell lane compared to the neighboring Ad5-infected
cell lane (Fig. 6A). To
quantitate host protein synthesis, total counts were determined in the
region of each lane between the band at the top of the resolving gel
and hexon at 120 kDa, the largest and most abundantly
synthesized viral protein (bracket in Fig.
6A, counts in Fig.
6B). By this assay, WT Ad5
inhibited cellular protein synthesis to 20% of the rate in
mock-infected cells. dl1520 infection inhibited cellular
protein synthesis less than WT Ad5 infection (to 40 to 50% of that in
mock-infected cells), but significant inhibition was nonetheless
evident in the absence of E1B-55K. In cells preinfected with 5
vectors (Fig. 6B, lanes 4
to 9), host protein synthesis was reduced to 20 to 25% of that in
mock-infected cells under all conditions and overexpression of CUL5 NTD
had little effect. As discussed below, these results suggest that the
decreased inhibition of host cell protein synthesis in cells infected
with E1B-55K mutants (Fig.
6B) is an indirect effect
of decreased expression of late proteins, including 100K, the Ad5 late
protein responsible for inhibition of host protein synthesis
(16,
84). Indeed,
L4 cytoplasmic RNA, which encodes 100K, is reduced in
dl1520-infected cells to 25% of the concentration observed in
WT Ad5-infected HeLa cells (data not
shown).
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
800 kDa
that polyubiquitinates p53 in vitro
(37). This complex also
likely polyubiquitinates subunits of the MRN complex
(78) required
for initiating a cellular DNA damage response after infection by Ad5
mutants that fail to inactivate the MRN complex
(10). During the late
phase of infection, both E1B-55K and E4orf6 are also required to
stimulate viral late protein synthesis in HeLa cells
(5,
18,
35,
83). This is due in part
to the stimulation of viral late mRNA nuclear export at the expense of
host cell mRNA nuclear export
(3,
4,
6,
25,
35,
64).
How can the
E1B-55K and E4orf6 proteins influence processes as diverse as cell
cycle control, apoptosis, mRNA nuclear export, and translation? The
results presented here strongly support the model in which, as for the
degradation of p53 and the MRN complex during Ad5 infection, the
influence of E1B-55K and E4orf6 on mRNA nuclear export requires the
ubiquitin-protein ligase activity of the Ad-Ub protein ligase complex
they form. An earlier paper reached the same conclusion in studies with
a proteasome inhibitor
(14). However, the
inhibition of viral DNA synthesis by the proteasome inhibitor
complicated the interpretation of these experiments. In our study, a
more specific inhibition of the Ad5 ubiquitin ligase was achieved by
overexpression of a dominant-negative CUL5 mutant. Under these
conditions, there was no inhibition of viral DNA replication (Fig.
3B). We also analyzed
viral late nuclear and cytoplasmic RNAs. The results strongly support
the model in which the viral E1B-55K/E4orf6 E3 ubiquitin ligase
stimulates viral late mRNA nuclear export. The conclusion that all
E1B-55K functions derive from its function in the Ad-Ub protein ligase
complex also fits with the observation that >95% of the
cellular E1B-55K present in HeLa cells during the late phase of
infection is in the 800-kDa viral ubiquitin ligase complex
(37). Also, the Ad-Ub
protein ligase complex complex and its substrates MRN and p53 (to which
it is presumably bound) are exported out of the nucleus to cytoplasmic
aggresomes (1,
50), cellular degradation
centers for misfolded proteins where components of the
ubiquitin-proteasome system are enriched
(27,
44).
How might the degradation of a cellular (or viral) protein stimulate viral mRNA nuclear export and inhibit host cell mRNA export? In one possible model, the viral ubiquitin ligase directs the degradation of a protein required for export of most host cell mRNPs but not viral mRNPs. This would inhibit host mRNP export and might stimulate viral mRNP export by reducing competition for export factors by host mRNPs. For example, RNA export factor binds most mRNPs and facilitates binding of the principal mRNP transporter, the NXF1 (TAP)-Ntr1 (p15) heterodimer (79). But some retroviruses contain a constitutive transport element, a structured region in the 3' untranslated region of the viral mRNAs that binds NXF1 directly (17). The 201-base 5' tripartite leader associated with the >20 mRNAs transcribed from the major late promoter stimulates Ad5 late mRNA nuclear export (41). Perhaps, in addition to its function in viral late protein synthesis (16), the tripartite leader also functions similarly to retroviral constitutive transport elements. In that case, inactivation of the RNA export factor might stimulate viral mRNA export (Fig. 7, model 1).
|
What is the substrate of the Ad5 ubiquitin-protein ligase whose degradation or, potentially, monoubiquitination (77) stimulates viral late mRNA nuclear export at the expense of cellular mRNA export in HeLa cells? Two substrates are known, p53 and the MRN complex. The critical target cannot be p53 because the defect in viral late gene expression is also observed in p53–/– cells (31, 36, 67, 80).
Could the failure of an E1B-55K mutant to stimulate viral late mRNA nuclear export be due to a failure to inactivate the MRN complex and hence a failure to inhibit the DNA damage response (10)? Of potential relevance to this question, two components of the nonsense-mediated mRNA decay (NMD) pathway were recently shown also to participate in the DNA damage response (2). Components of the exon junction recognition complex that function in NMD also participate in mRNA nuclear export (54). Consequently, NMD components may function in both mRNA and DNA damage surveillance (2), providing a possible mechanism for the DNA damage response to influence nuclear mRNA export. SMG1, required for NMD, is an ATM/R-related protein kinase that phosphorylates ATM/R sites in p53 and is required for a full p53 response (9). In response to DNA damage, SMG1 and ATR phosphorylate UPF1, a component of the exon junction recognition complex proposed to induce mRNA association with cytoplasmic P bodies, where translation is suppressed, and the mRNA is degraded (76). UPF1 is both an RNA- and a DNA-helicase whose depletion by shorthairpin RNA induces an ATR-dependent DNA damage response and cell cycle arrest in early S phase (2). Perhaps a branch of the DNA damage response resulting from SMG1 and ATM/R phosphorylation of NMD components leads to inhibition of viral late mRNA nuclear export in cells infected with E1B-55K or E4orf6 mutants. Further studies are necessary to determine the extent to which the DNA damage response contributes to the inhibition of viral nuclear mRNA export in HeLa cells infected with E1B-55K and E4orf6 mutants.
While two substrates of the Ad5 ubiquitin-protein ligase are known currently, the example of HPV E6 raises the possibility that there may be additional, as-yet-unknown, substrates. HPV-E6 was first identified as a component of an E3 ubiquitin ligase by its ability to stimulate polyubiquitination of p53 (55). However, subsequent studies suggest that it has multiple cellular substrates (55). In addition to the DNA damage response, infection with E1B-55K and E4orf6 mutants may induce other cellular antiviral responses that inhibit viral mRNA nuclear export. Perhaps a component of an as-yet-undiscovered antiviral response is also targeted for proteasomal degradation by the Ad-Ub protein ligase.
Shutoff of host cell protein synthesis during the late phase of infection was only partially dependent on E1B-55K function (Fig. 6A and B, compare lanes 1 to 3) and was not significantly inhibited by overexpression of the dominant-negative CUL5 NTD (Fig. 6A and B, compare lanes 4 to 6). These results suggest that inhibition of host protein synthesis is not a direct function of E1B-55K. Rather, E1B-55K may influence host cell protein synthesis shutoff indirectly by stimulating viral late protein synthesis. One of the late viral proteins is 100K, a multifunctional protein that (i) acts as a chaperone required for proper folding of the hexon trimers that make up the faces of the virion icosahedron (40), (ii) stimulates translation of mRNAs with a 5' tripartite leader common to viral mRNAs transcribed from the major late transcription unit, and (iii) inhibits host protein synthesis (15, 16, 84). 100K binds to the eIF4G scaffold subunit of the eIF4 translation initiation factor, displacing the MNK1 kinase. This leads to hypophosphorylation of the eIF4E cap-binding subunit of the eIF4 complex and inhibition of cap-dependent mRNA translation (15). However, 100K also contains RNA-binding domains that bind specifically to mRNAs containing the 5' tripartite leader, stimulating their translation in the face of hypophosphorylated eIF4E by a specialized translational initiation mechanism called ribosome shunting (87). Mutation of E1B-55K may partially reverse inhibition of host cell protein synthesis indirectly by causing decreased 100K expression (25, 60).
McCormick and colleagues have proposed that dl1520 (also known as ONYX015) might be an effective anticancer therapy by replicating in p53–/– tumor cells while being blocked from replication in p53+ cells (52). However, several studies indicated that the ability of E1B-55K mutants to replicate in various tumor cell lines in culture does not correlate well with p53 status (31, 36, 67, 80). Nonetheless, results of clinical trials indicate that ONYX015 is partially effective in treating head and neck squamous cell carcinomas in some patients but not in others. Perhaps tumor cells that support high-level replication by E1B-55K mutants have inactivated or repressed DNA and RNA surveillance mechanisms, in addition to the p53 pathway, that are inactivated by the Ad5 ubiquitin-protein ligase but not by E1B-55K mutants. Understanding how the Ad5 ubiquitin-protein ligase contributes to the stimulation of viral late mRNA export in HeLa cells may allow determination of which tumors might be treated effectively by adenovirus E1B-55K mutants.
|
ACKNOWLEDGMENTS |
|---|
This research was supported by grant CA64799 from the NIH, USPHS. J.L.W. was the recipient of a Genetics Mechanisms training grant from NIH USPHS National Research Service award GM07104.
|
FOOTNOTES |
|---|
Published
ahead of print on 1 November 2006.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
), high similarity (:), and similarity (.). Black arrows
point to the conserved hydrophobic residues changed. (D) Cul1
NTD-CTD interface: Val367 and Leu371 side chains shown in yellow
(
