JOURNAL OF VIROLOGY, July 2011, p. 6212–6219
Copyright © 2011, American Society for Microbiology. All Rights Reserved.
Vol. 85, No. 13
Ubiquitin-Specific Peptidase 20 Targets TRAF6 and Human
T Cell Leukemia Virus Type 1 Tax To Negatively
Regulate NF-?B Signaling?
Junichiro Yasunaga,1Frank C. Lin,1Xiongbin Lu,2and Kuan-Teh Jeang1*
Molecular Virology Section, Laboratory of Molecular Microbiology, National Institute of Allergy and Infectious Diseases,
National Institutes of Health, Bethesda, Maryland,1and Department of Cancer Biology, The University of
Texas MD Anderson Cancer Center, Houston, Texas2
Received 12 January 2011/Accepted 11 April 2011
NF-?B plays a key role in innate and acquired immunity. Its activity is regulated through intricate signaling
networks. Persistent or excessive activation of NF-?B induces diseases, such as autoimmune disorders and
malignant neoplasms. Infection by human T cell leukemia virus type 1 (HTLV-1) causes a fatal hematopoietic
malignancy termed adult T cell leukemia (ATL). The HTLV-1 viral oncoprotein Tax functions pivotally in
leukemogenesis through its potent activation of NF-?B. Recent findings suggest that protein ubiquitination is
crucial for proper regulation of NF-?B signaling and for Tax activity. Here, we report that ubiquitin-specific
peptidase USP20 deubiquitinates TRAF6 and Tax and suppresses interleukin 1? (IL-1?)- and Tax-induced
NF-?B activation. Our results point to USP20 as a key negative regulator of Tax-induced NF-?B signaling.
Protein ubiquitination is an essential posttranslational mod-
ification that is implicated in many biological processes (14,
52). Ubiquitin is a small protein composed of 76 amino acids.
It contains 7 lysine residues (K6, K11, K27, K29, K33, K48, and
K63). Multiple ubiquitin monomers can become covalently
linked, and polyubiquitin molecules linked through the lysine
48 residue (K48) are known to modulate protein degradation.
In contrast, polyubiquitin molecules linked through the lysine
63 (K63) residue do not induce degradation but influence
protein localization, protein-protein interaction, protein func-
tional activation, and other activities (14, 44). The addition of
ubiquitin to or the removal of ubiquitin from protein substrates
can reversibly and dynamically change protein functions, and
these reactions are executed by ubiquitin ligases and deubiq-
uitinases (44). Accordingly, the ubiquitination process is me-
diated by the serial actions of E1 ubiquitin-activating enzyme,
E2 ubiquitin-conjugating enzyme, and E3 ubiquitin ligase (14,
52), and deubiquitination is through deubiquitining enzymes
(DUBs) that directly remove ubiquitin molecules from their
substrates (32). Based on sequence analyses, approximately 90
DUB genes have been identified in the human genome. These
DUBs are divided into 5 subclasses according to their protein
sequences: the ubiquitin C-terminal hydrolases (UCHs), ubiq-
uitin-specific proteases (USPs), Machado-Joseph disease pro-
tein domain proteases (MJDs), ovarian tumor proteases
(OTUs), and JAMM motif proteases (32). The specificities and
actions of these various deubiquitinases remain to be fully
In mammals, NF-?B signaling plays key roles in inflamma-
tion, cell proliferation, and apoptosis (36). Like phosphoryla-
tion, ubiquitination is an important posttranslational modifi-
cation that can regulate NF-?B activity (43). For example, the
ubiquitination of a regulatory subunit in the IKK complex,
IKK? which is also known as NEMO, has a central role in
signal transduction. It has been shown that K63-linked linear
conjugation of ubiquitin to IKK? positively regulates NF-?B
signaling (47, 53). In addition, the ubiquitination of tumor
necrosis factor receptor-associated factors (TRAFs) is also
important for IKK activation. TRAF6 is an E3 ubiquitin ligase,
and it performs self-ubiquitination through K63-linked chains
upon cellular activation through Toll-like receptors (TLRs)
and cytokine receptors (8). Moreover, in the canonical NF-?B
pathway, I?B?, which sequesters NF-?B proteins in the cyto-
plasm in an inactive state, is conjugated with K48-linked poly-
ubiquitin chains and is proteasomally degraded when cells are
stimulated to activate NF-?B (4). Similarly, cellular activation
induces the ubiquitination and proteosomal processing of NF-
?B2 p100 to p52, allowing NF-?B RelB/p52 dimers to translo-
cate into the nucleus.
The role of DUBs in the NF-?B pathway has also been
studied. For example, the familial cylindromatosis tumor sup-
pressor CYLD is one of the DUBs that have been found to
suppress NF-?B activity. CYLD has been shown to bind IKK?
and to reduce the ubiquitination of TRAF2, TRAF6, and
IKK? (3, 24, 48). A20 is a second well-studied DUB that
negatively regulates NF-?B activation by reducing the ubiq-
uitination of TRAF2, TRAF6, and RIP1 (17, 20, 41). A20 has
dual activities in ubiquitination and deubiquitination. Hence,
A20 with TAX1BP1 as a cofactor promotes the cleavage of
K63-linked polyubiquitin chains on RIP1, and A20 with E3
ligase Itch can conjugate K48-linked chains on RIP for pro-
teosomal degradation (39, 51). Recently, it was also found that
A20 can inhibit the ubiquitination of TRAF2 and TRAF6 by
dissociating complexes composed of TRAFs and E2 ubiquitin-
conjugating enzymes (41).
Infection by human T cell leukemia virus type 1 (HTLV-1)
causes a fatal hematopoietic malignancy, adult T cell leukemia
* Corresponding author. Mailing address: 9000 Rockville Pike,
Building 4, Room 303, Bethesda, MD 20892. Phone: (301) 496-6680.
Fax: (301) 480-3686. E-mail: firstname.lastname@example.org.
?Published ahead of print on 27 April 2011.
(ATL) (29), and one of its key regulatory proteins, Tax, plays
important roles in viral pathogenesis (2, 10, 11, 26, 37). Tax
(29) can potently activate NF-?B through both the canonical
and noncanonical pathways (15, 42, 45). Tax was recently
found to inhibit A20 function by disrupting its interaction with
TAX1BP1 and Itch (39). Currently, how other DUBs may
contribute to Tax-induced NF-?B signaling has not been stud-
ied. Here, we report on the characterization of USP20 for its
regulation of Tax- and TRAF6-mediated activation of NF-?B.
MATERIALS AND METHODS
Cells and reagents. Human embryonic kidney cell line HEK293T and human
cervical cancer cell line HeLa were cultured in Dulbecco’s modified Eagle’s
medium containing 10% fetal bovine serum (FBS) and antibiotics. Human T cell
lines MT1, MT2, MT4, ATL2, Jurkat, CEM, and H9 were maintained in RPMI
1640 with 10% FBS. Recombinant human interleukin-1? (IL-1?) (Peprotech)
Plasmids and siRNAs. The ?B-luc reporter plasmid was described previously
(19). Human USP20 coding sequences were amplified by PCR and subcloned
into the pcDNA3 (Invitrogen) and the pCAG-HA (50) expression vectors. A
FIG. 1. Expression of USP20 suppresses HTLV-1 Tax- and IL-1?-induced NF-?B activation. (A) HEK293T cells were transfected with a
Tax-expressing plasmid, Hpx-Tax (0.05 ?g in lanes 4 to 6 and 13 to 15; 0.15 ?g in lanes 7 to 9 and 16 to 18), and ?B-luc reporter (0.05 ?g) with
pcDNA3-USP20 (0.5 ?g in lanes 2, 5, and 8; 1 ?g in lanes 3, 6, and 9) or pCMV-USP33 (0.5 ?g in lanes 11, 14, and 17; 1 ?g in lanes 12, 15, and
18). In each case, a pCMV-?-Gal plasmid (0.01 ?g) was included, and ?-Gal values were used to normalize for transfection efficiency.
Immunoblotting (lower panels) was done to confirm the expression of transfected and control (actin) protein using the indicated antibodies.
(B) ATL2 cells were transfected with ?B-luc reporter (1 ?g) with pcDNA3-USP20 (2 ?g in lane 2; 4 ?g in lane 3) or pCMV-USP33 (2 ?g in lane
5; 4 ?g in lane 6). pCMV-?-Gal (0.2 ?g) was included in each transfection as an internal normalization control. (C) HEK293T cells were
transfected with ?B-luc reporter (0.05 ?g) with pcDNA3-USP20 (0.5 ?g in lanes 2 and 5; 1 ?g in lanes 3 and 6) or pCMV-USP33 (0.5 ?g in lanes
8 and 11; 1 ?g in lanes 9 and 12) and were treated with recombinant IL-1? (10 ng/ml) for 8 h (lanes 4 to 6 and 10 to 12). pCMV-?-Gal (0.01 ?g)
was included in each transfection as an internal control. Immunoblotting was performed to confirm the expression of transfected and control (actin)
protein using the indicated antibodies. Total amounts of transfected DNA were equalized in each sample by the addition of vector DNA. Cell
lysates were assayed for luciferase. The results from three independent experiments are shown as average values ? standard deviations (SD).
VOL. 85, 2011USP20 DEUBIQUITINATES TRAF6 AND HTLV-1 Tax6213
human USP33 cDNA clone (I.M.A.G.E. 3874822) was purchased from ATCC,
and the coding region was inserted into pCAG-HA. For ectopic expression of
Tax, Hpx-Tax (18) and pCAG-Flag-Tax were used. Flag-TRAF6 has been de-
scribed elsewhere (20). The pcDNA-HA-ubiquitin expression vector was used in
immunoblotting experiments to detect ubiquitinated Tax and TRAF6. For the
knockdown of USP20, synthesized small interfering RNAs (siRNAs) (Qiagen)
were used. The target sequence of USP20 is 5?-TCGAGTGACACGGATGAG
AAA-3?. Sequence-nonspecific siRNA was purchased from Qiagen and used as
a negative control.
Antibodies. Mouse monoclonal anti-Tax (NIH AIDS Research and Reference
Reagent Program) was used to detect Tax protein in immunoblotting. Anti-Flag
monoclonal antibody (M2; Sigma), anti-Flag polyclonal antibody (Sigma), anti-
hemagglutinin (anti-HA) monoclonal antibody (HA-7; Sigma), anti-HA poly-
clonal antibody (Sigma), anti-USP20 polyclonal antibody (Bethyl Laboratories),
anti-USP33 polyclonal antibody (Bethyl Laboratories), anti-tubulin monoclonal
antibody (DM1A; Sigma), and anti-actin monoclonal antibody (AC-40; Sigma)
Luciferase assay. Expression plasmids or siRNAs were transfected into cells
using Lipofectamine 2000 (Invitrogen). Cells were transfected with Tax and USP
expression vectors or were transfected with siRNAs first, followed 48 h later with
?B-Luc and pCMV-?-galactosidase (pCMV-?-Gal) plasmids. At 24 h after
transfection of the reporters, cell lysates were subjected to luciferase assay. Total
amounts of DNA and RNA to be transfected were adjusted by the addition of
empty vectors or nonspecific siRNAs. To detect luciferase and ?-Gal activity,
luciferase substrate (Promega) and the Galacto-Star assay system (Applied Bio-
systems) were used. Relative values of luciferase activity were calculated using
?-Gal activity as an internal control for transfection.
Coimmunoprecipitation assay. At 48 h after transfection, cells were lysed with
radioimmunoprecipitation assay (RIPA) buffer (50 mM HEPES, 0.3% NP-40,
150 mM NaCl, 2 mM EDTA, 20 mM ?-glycerophosphate, 0.1 mM sodium
orthovanadate, 1 mM sodium fluoride, 0.5 mM dithiothreitol [DTT], and 1?
protease inhibitor cocktail from Roche). Cell extracts were subjected to immu-
noprecipitation with anti-Flag or anti-HA antibodies, and the captured proteins
were detected by immunoblotting using the indicated antibodies.
Immunofluorescence. Cells were cultured on glass coverslips, and fixed in 4%
paraformaldehyde at 24 h after transfection. After blocking of nonspecific reac-
tions with 1% bovine serum albumin (BSA), cells were then incubated with the
indicated primary antibodies, followed by a subsequent incubation with the
secondary antibodies conjugated with Alexa Fluor 488 or 594 (Molecular
Probes). DNA was counterstained with 0.1 ?g/ml Hoechst 33342. Coverslips
were mounted in Prolong Antifade (Molecular Probes), and cells were visualized
with a Leica TCS SP2 confocal microscope.
Real-time PCR. USP20 transcripts were quantified by real-time PCR. The
sequences of the primers for USP20 were 5?-TCACAGAAGTCCACGAGAC
G-3? (sense) and 5?-TTGTCCTTCCCCTTGACGAA-3? (antisense). To com-
pare the relative expression levels between samples, we quantified mRNA levels
of GAPDH (glyceraldehyde-3-phosphate dehydrogenase) as the internal control
using the primers 5?-GCTCACTGGCATGGCCTTCCGTGT-3? (sense) and 5?-
TGGAGGAGTGGGTGTCGCTGTTGA-3? (antisense). Power SYBR green
PCR Master Mix (Applied Biosystems) was used for preparation of the PCR
mixtures, and the reaction was carried out with an Applied Biosystems 7500
real-time PCR system. All samples were measured in triplicate and analyzed with
Cell proliferation assay. ATL2 cells were transfected with USP20, USP33, or
control expression vector using Amaxa Nucleofector II (Human T Cell Nucleo-
fector kit, protocol X-001). Using a green fluorescent protein (GFP)-expressing
plasmid, we analyzed the transfection efficiency of ATL2 cells with this method
and found that routinely ?60% of cells were visibly GFP positive at 48 h after
transfection. Cell proliferation was evaluated with a 3-(4,5-dimethylthiazol-2-yl)-
2,5-diphenyltetrazolium bromide (MTT) colorimetric assay.
USP20 suppresses HTLV-1 Tax- and cytokine-induced
NF-?B activation. Activation of NF-?B contributes to cellular
transformation by HTLV-1, and Tax is a potent activator of
NF-?B (15, 29, 42). Protein ubiquitination plays an important
role in signal transduction; indeed, individual NF-?B pathway
constituents are frequently ubiquitinated when cells are stim-
ulated (4, 14, 36). We reasoned that to maintain homeostasis,
de novo-ubiquitinated proteins must eventually be deubiquiti-
nated. To identify the cellular DUBs that might affect Tax-
induced NF-?B activation, we tested 12 discrete DUBs and
found that USP20 can suppress Tax-induced luciferase expres-
sion in a dose-dependent manner (Fig. 1A, lanes 1 to 9). By
comparison, a related DUB, USP33 (and 10 other DUBs [data
not shown]), failed to influence NF-?B activation by Tax (Fig.
1A, lanes 10 to 18). Next, we checked the activities of USP20
and USP33 in HTLV-1 transformed ATL2 leukemic cells
which were directly derived from the leukemic cells of an ATL
patient (46). We found that USP20, but not USP33, inhibited
NF-?B activity in ATL2 cells (Fig. 1B). Collectively, the results
suggest that the NF-?B activating pathways in HEK293T (Fig.
1A) and ATL2 (Fig. 1B) cells are both negatively regulated by
In order to characterize the roles of USP20 and USP33 in
regard to other activators of NF-?B, we investigated their
FIG. 2. Knockdown of cell-endogenous USP20 enhances HTLV-1
Tax- and IL-1?-induced NF-?B activation. (A) Sequence-specific
siRNA was used to knock down USP20. The knockdown of cell-
endogenous USP20 by siRNA was confirmed with real-time RT-PCR.
USP20-specific siRNAs at 20 nM (bar 2) or 40 nM (bar 3) were
transfected into HEK293T cells, from which RNA was extracted.
USP20 mRNA levels were quantified by real time RT-PCR. (B) The
competent knockdown of cell-endogenous USP20 by siRNA was
checked by immunoblotting. USP20-specific siRNAs at 30 nM (lane 3)
or 300 nM (lane 4) were transfected into Jurkat cells, and cell-endog-
enous USP20 protein was assayed by immunoblotting. (C) HEK293T
cells were transfected with Hpx-Tax (bars 4 to 6) or treated with IL-1?
(50 ng/ml) at the indicated times (bars 9 to 12) with or without the
cotransfection of siRNA for USP20 (20 nM in lanes 2 and 5; 40 nM in
lanes 3, 6, 8, 10, and 12) (0 h of IL-1? means that no treatment
occurred). Total amounts of transfected DNAs and siRNAs were
equalized by the addition of either vector DNA or control siRNA,
respectively. Cell lysates were assayed for luciferase. Immunoblotting
was done to confirm the expression of transfected and control (actin)
protein using the indicated antibodies. The results from three inde-
pendent experiments are shown as average values ? SD.
6214 YASUNAGA ET AL.J. VIROL.
effects on IL-1?-mediated activation. As shown in Fig. 1C,
NF-?B activation by IL-1? was suppressed when either USP20
(Fig. 1C, lanes 5 and 6) or USP33 (Fig. 1C, lanes 11 and 12)
was expressed in HEK293T cells. These findings suggest that
both USP20 and USP33 can negatively regulate IL-1?-medi-
ated NF-?B signaling, while USP20, but not USP33, can mod-
ulate Tax-induced NF-?B activation (Fig. 1A).
To check for physiological relevance and to rule out that the
results were simply from nonphysiological overexpression, we
knocked down cell-endogenous USP20 in HEK293T cells us-
ing siRNAs. Using real-time reverse transcription-PCR (RT-
PCR), we documented an siRNA dose-dependent knockdown
of USP20 mRNA in HEK293T (Fig. 2A, bars 2 and 3). To
check the knockdown effect of the siRNA at the protein level,
we transfected the siRNA into Jurkat cells and confirmed by
immunoblotting that USP20 expression was reduced as ex-
pected (Fig. 2B). Next, at 48 h after USP20 siRNA transfection
into HEK293T cells, we transfected a ?B-luc reporter plasmid
and checked for its induction by Tax or IL-1?. With either Tax
(Fig. 2C, compare bar 4 to bars 5 and 6) or IL-1? (Fig. 2C,
compare bar 9 to 10 and bar 11 to 12), the activity of the ?B-luc
reporter was higher in USP20-knocked-down cells than in con-
trol siRNA-treated cells. Collectively, the findings in Fig. 1 and
2 are consistent with USP20 serving to negatively modulate
cellular NF-?B activity.
TRAF6 interacts with USP20 and USP33. To our knowl-
edge, the above observations on USP20 and USP33 represent
the first demonstration of their effects on NF-?B signaling. We
were thus interested in searching for potential USP20/33 cel-
lular substrates among known NF-?B-signaling proteins.
TRAF6 is an NF-?B-signaling protein that has been reported
to be regulated by ubiquitination (8). We next asked if TRAF6
would interact with USP20 or USP33 in coimmunoprecipita-
tions. We transfected Flag-tagged TRAF6 and either HA-
tagged USP20 or HA-tagged USP33 into HEK293T cells. Cell
lysates were immunoprecipitated using anti-Flag or anti-HA,
and the presence of HA- or Flag-tagged proteins in the respec-
tive immunoprecipitates was assessed by immunoblotting. As
shown in Fig. 3A, TRAF6 bound USP20 and USP33. As a
control, we checked if another NF-?B-signaling protein also
known to be ubiquitinated, IKK? (47, 53), would interact with
USP20 or USP33; however, the latter interaction was not ob-
served (data not shown).
USP20 and USP33 have been reported to localize in peri-
nuclear regions and in small vesicles throughout the cytoplasm
(1, 7), and TRAF6 has been shown to locate diffusely in the
cytoplasm (49). Next, we immunostained HeLa cells trans-
fected with Flag-TRAF6 and either HA-USP20 or HA-USP33
(Fig. 3B and C). Although not all TRAF6 and USP20/USP33
colocalized together, we did observe partial colocalization of
these proteins in the cytoplasm (see yellow costaining in mag-
nified views of the “Merge” panels in Fig. 3B and C), consistent
with the potential for intracellular protein-protein interaction.
FIG. 3. TRAF6 interacts with USP20 and USP33. (A) HEK293T cells were transfected with the indicated plasmids, and immunoprecipitation
was done with either anti-Flag or anti-HA antibody as indicated. The immunoprecipitates and total cell extracts were resolved by SDS-PAGE and
immunoblotted. The blots were probed with either anti-Flag or anti-HA. (B and C) Flag-TRAF6 plasmid with either HA-USP20 or HA-USP33
was cotransfected into HeLa cells. Subcellular localizations of TRAF6, USP20, or USP33 were visualized with anti-Flag (green) or anti-HA (red).
The nuclei were stained with DAPI (4?,6?-diamidino-2-phenylindole) (blue). The higher magnification of the “Merge” is shown in the lower right
panels of panels B and C.
VOL. 85, 2011USP20 DEUBIQUITINATES TRAF6 AND HTLV-1 Tax6215
TRAF6 is a substrate of USP20. Previously, USP20 was
reported to interact with HIF-1? (27), type2 iodothyronine
deiodinase (7), and ?2 adrenergic receptor (?2AR) (1). Our
finding of an interaction between USP20 and TRAF6
prompted us to ask if TRAF6 might be a functional USP20
substrate. We therefore checked for the effect of USP20 on
IL-1?-induced ubiquitination of TRAF6. We compared the
levels of ubiquitinated TRAF6 in the absence or presence of
transfected USP20 (Fig. 4A). IL-1? treatment of cells for 8 h
increased the amount of ubiquitinated TRAF6 (Fig. 4A, com-
pare lane 1 to 3), and this increase was diminished by the
expression of USP20 (Fig. 4A, compare lane 3 to 4). In addi-
tion, siRNA knockdown of cell-endogenous USP20 enhanced
the level of ubiquitinated TRAF6 in IL-1?-treated cells (Fig.
4B, compare lane 3 to 4). Altogether the results support an
interpretation that USP20 promotes the deubiquitination of
IL-1?-induced ubiquitinated TRAF6.
In our experiments, Tax expression also increased intracel-
lular ubiquitinated TRAF6, as reported previously (Fig. 4C,
compare lane 2 to 4) (12). The Tax-induced ubiquitination of
TRAF6 was also sensitive to deubiquitination by USP20 (Fig.
4C, compare lane 4 to 5). Thus, TRAF6 appears to be a bona
fide USP20 substrate, since USP20 can efficiently deubiq-
uitinate otherwise ubiquitinated TRAF6 that is induced by
diverse stimuli such as Tax and IL-1?.
Ubiquitinated Tax is a substrate for USP20. Ubiquitina-
tion, sumoylation, acetylation, and phosphorylation of Tax
have been shown to influence its transcriptional function (5,
9, 22, 23, 25, 28, 34, 35, 40). Tax ubiquitination has been
reported to be required for its activation of NF-?B, and it
has been found that the Tax-polyubiquitin chains are pre-
dominantly K63 linked (40). Because it was observed that
USP20 expression inhibited Tax-induced NF-?B activation
(Fig. 1A), we wondered if this finding might also have re-
FIG. 4. IL-1?- or Tax-induced ubiquitinated TRAF6 is deubiquitinated by USP20. (A) HEK293T cells were transfected as indicated with
Flag-TRAF6, HA-Ub, and USP20. Cells without IL-1? treatment (lanes 1 and 2) and cells treated with IL-1? for 8 h (lanes 3 and 4) were subjected
to immunoprecipitation (IP) followed by immunoblotting (IB) to detect ubiquitinated TRAF6 (anti-HA; top panel). WCE, whole-cell extract.
(B) HEK293T cells were transfected as indicated with Flag-TRAF6, HA-Ub, and siRNA specific for USP20. Cells without treatment (0 h; lanes
1 and 2) and cells treated with IL-1? for 8 h (lanes 3 and 4) were immunoprecipitated and detected by immunoblotting for ubiquitinated TRAF6
(anti-HA; top panel). (C) HEK293T cells were transfected as indicated with Tax, Flag-TRAF6, HA-Ub, and USP20 (lanes 2 to 5). Cell lysates were
immunoprecipitated and immunoblotted for detection of ubiquitinated TRAF6 (anti-HA; top panel). Total amounts of transfected DNA or
transfected siRNA were equalized by the addition of empty vector or control siRNA, respectively. Immunoprecipitations were performed using
anti-Flag antibody. Ubiquitinated proteins were detected by immunoblotting with anti-HA antibody. Anti-Flag (for TRAF6), anti-USP20, anti-Tax,
anti-tubulin, and anti-actin antibodies were used to verify the respective proteins in immunoblottings. To normalize the amount of ubiquitinated
TRAF6, signals of ubiquitinated TRAF6 and total TRAF6 in the immunoblots were quantified by densitometry, and relative levels of normalized
ubiquitinated TRAF6 were calculated. The results are graphed in the middle of each panel.
6216YASUNAGA ET AL. J. VIROL.