Cell Host & Microbe
Tousled-like Kinases Modulate Reactivation
of Gammaherpesviruses from Latency
Patrick J. Dillon,1,2Sean M. Gregory,1,2Kristen Tamburro,1,2,3Marcia K. Sanders,1,2Gary L. Johnson,1,4
Nancy Raab-Traub,1,2Dirk P. Dittmer,1,2,3and Blossom Damania1,2,3,*
1Lineberger Comprehensive Cancer Center
2Department of Microbiology and Immunology
3Curriculum in Genetics
4Department of Pharmacology
University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA
Kaposi’s sarcoma-associated herpesvirus (KSHV) is
linked to human malignancies. The majority of tumor
cells harbor latent virus, and a small percentage
undergo spontaneous lytic replication. Both latency
and lytic replication are important for viral pathogen-
esis and spread, but the cellular players involved in
the switch between the two viral life-cycle phases
are not clearly understood. We conducted a small
interfering RNA (siRNA) screen targeting the cellular
kinome and identified Tousled-like kinases (TLKs)
as cellular kinases that control KSHV reactiva-
tion from latency. Upon treatment of latent KSHV-
infected cells with siRNAs targeting TLKs, we saw
robust viral reactivation. Knockdown of TLKs in
latent KSHV-infected cells induced expression of
viral lytic proteins and production of infectious virus.
tivation from latency of another related oncogenic
gammaherpesvirus, Epstein-Barr virus. Our results
establish the TLKs as cellular repressors of gamma-
Kaposi’s sarcoma-associated herpesvirus (KSHV) is linked to
a number of human malignancies, including Kaposi’s sarcoma
(KS), primary effusion lymphoma (PEL), and multicentric
Castleman disease (MCD) (Chang et al., 1994, Songyang et al.,
1994; Stu ¨rzl et al., 1997; Cesarman et al., 1995; Soulier et al.,
1995). In the majority of infected cells, KSHV remains latent;
however, a small percentage of cells can undergo spontaneous
lytic replication at any given time (Zhong et al., 1996; Poyet et al.,
2001). This low level of viral reactivation is believed to be impor-
tant for persistence and tumorigenesis (Grundhoff and Ganem,
2004). Both latency and lytic replication are important phases
for viral pathogenesis and spread, but the cellular players
involved in the switch between the two phases of the viral life
cycle are not clearly understood.
Incellculture,reactivation ofKSHVoccursfollowing treatment
with chemical compounds such as the phorbol ester 12-O-
tetradecanoyl-phorbol-13-acetate (TPA/PMA) and the histone
deacetylase inhibitor sodium butyrate. We previously reported
that activation of Toll-like receptors 7 and 8 in PEL cells led to
KSHV reactivation (Gregory et al., 2009). In an overexpression
system, using transient transfection of kinase complementary
DNAs (cDNAs), other investigators showed that both the Pim
and Ras family kinases are involved in KSHV reactivation (Cheng
et al., 2009; Yu et al., 2007).
To determine the cellular kinases that control KSHV reactiva-
tion from latency, we performed a small interfering RNA (siRNA)
screen of the cellular kinome in the absence of any chemical
inducers. This allowed us to assess which kinases are needed
for the virus to maintain latency or to induce reactivation. A
siRNA library containing siRNAs against 720 different human
kinases was used for the screen. We identified the Tousled-
like kinases (TLKs) as cellular kinases involved in reactivation
Originally described in the plant Arabidopsis thaliana, the
Tousled gene encodes a nuclear serine/threonine kinase that is
essential for flower and leaf development (Roe et al., 1993,
1997). Two mammalian homologs, TLK1 and TLK2, show 84%
sequence similarity to each other (Takahata et al., 2009). The
TLKs are regulated by cell-cycle-dependent phosphorylation,
their activity is tightly linked to DNA replication with maximal
activity during S phase, and they are sensitive to DNA-damaging
agents and inhibitors of DNA replication (Takahata et al., 2009).
TLKs are also involved in chromatin assembly. TLK1 and TLK2
bind to and phosphorylate the human chromatin assembly
factors Asf1a and Asf1b (Sillje ´ and Nigg, 2001). The TLKs have
been implicated in numerous replicative and transcriptional
processes, including chromosome condensation and segrega-
tion (Sunavala-Dossabhoy et al., 2003; Hashimoto et al., 2008),
gene silencing (Wang et al., 2007), and DNA repair (Sunavala-
Dossabhoy et al., 2005; Canfield et al., 2009).
In this study, we identified TLKs as modulators of KSHV reac-
tivation. Our results show that depletion of TLK2 in KSHV latently
infected epithelial cells leads to robust viral reactivation. Knock-
infectious progeny virions. Depletion of TLK2, and to a lesser
extent TLK1, also leads to KSHV viral reactivation from latently
204 Cell Host & Microbe 13, 204–214, February 13, 2013 ª2013 Elsevier Inc.
infected B cells. Moreover, we found that knockdown of TLK1,
and to a lesser extent TLK2, results in reactivation of another
gammaherpesvirus family member, Epstein-Barr virus (EBV).
Our results indicate that the TLKs are key regulators of KSHV
and EBV reactivation, and their expression is required for the
maintenance of viral latency.
Cellular siRNA Kinome Screen
To determine which cellular kinases are important for KSHV re-
activation, we performed a siRNA screen targeting the human
cellular kinome. More than 720 siRNAs against all human protein
and lipid kinases were included in the screen. Each well con-
tained a pool of four siRNAs with different target sequences to
a single cellular kinase. For the screen, we used KSHV-infected
293 (KSHV-293) cells that harbor latent KSHV and constitutively
express green fluorescent protein (GFP), whereas red fluores-
cent protein (RFP) is under the control of a lytic promoter and
thus is only expressed upon viral reactivation (Vieira and
O’Hearn, 2004). A variety of confounding factors often lead to
high false-discovery rates in siRNA screens, which is an inevi-
table result of high-throughput investigations. To lessen the
false-discovery rate, we performed our primary screen in tripli-
cate and used robust statistical analysis. We reverse transfected
thecells withthe siRNApools,and 70hrpost-siRNAtransfection
acquired GFP and RFP images and fluorescence intensities
using a Cellomics ArrayScan VTI HCS Reader (Figure 1A). To
ensure efficient siRNA transfection, an siRNA against Ubiquitin
B (UBB) was included as a control siRNA in the screen, because
UBBknockdown isknownto leadtocelldeath(Tiedemannetal.,
statistical program environment (http://www.R-project.org). We
determined statistically significant changes in viral reactivation
(i.e., RFP intensity) using both the median and mean RFP values
for all the wells of the siRNA screen. Figure 1C shows a waterfall
plot of the Z scores of the median RFP value for each of the
siRNAs. A Z score of R2 was considered significant (Figure 1C).
Additionally, Table 1 lists the cellular kinases that when depleted
showed a R2 SD increase from the overall mean RFP intensity.
One isoform of TLK, TLK2, stood out in both of these analyses
(Figure 1C; Table 1) because knockdown of TLK2 showed a Z
score of 15 based on the median RFP value, and TLK2 knock-
down led to a level of RFP expression that was 13 SDs above
the mean RFP value for the screen. The next most significant
kinases were only 3 SDs above the mean. These comprised
seven outof 720(1%) of the siRNA targets, attesting to the spec-
ificity and stringency of our screen. It is possible that some of the
siRNAs in our screen did not sufficiently deplete their target
proteinandthuscouldpotentially provideafalse-negative result.
Except for the TLKs, we did not confirm any of the hits listed in
Table 1. Figure 1D shows representative GFP and RFP images
Figure 1. Design of siRNA Screen and Analysis of Data
(A) Schematic of cellular kinome siRNA screen. siRNAs from the Dharmacon
SMARTpool kinase siRNA library were loaded into 384-well plates in triplicate.
KSHV-293 cells were added to each well containing siRNA and incubated for
(B) Control siRNAs demonstrate efficient siRNA transfection in our screen.
GAPDH and UBB siRNAs were reverse transfected at 25 nM each into 2,500
KSHV-293 cells/well in a 384-well plate. At 70 hr posttransfection, bright-field,
GFP, and RFP pictures were taken on a fluorescent microscope.
(C) Statistical analysis of primary hits. A waterfall plot was used to analyze the
data acquired from the screen. It shows the Z score of the median RFP
intensity of each siRNA.
(D) Reactivation by TLK2 knockdown in KSHV-293 cells. GFP and RFP images
of a representative field taken during the screen are shown for the wells
containing siRNA targeting GAPDH, TLK1, and TLK2. See also Figure S1.
Cell Host & Microbe
TLKs Modulate Gammaherpesvirus Reactivation
Cell Host & Microbe 13, 204–214, February 13, 2013 ª2013 Elsevier Inc. 205
from the kinome screen. As can be seen in the RFP panels,
knockdown of TLK1 shows no change in RFP expression
compared with the GAPDH siRNA control in KSHV-293 cells,
whereas TLK2 knockdown results in a large increase in RFP-
positive cells indicativeof viralreactivation. Torule outthepossi-
bility that viral reactivation following TLK2 knockdown is due to
off-target effects of one or more of the TLK2 siRNAs, we trans-
fected KSHV-293 cells with each of the four individual TLK2
siRNAs that target different regions of the TLK2 transcript. The
cells were examined by microscopy (Figure S1A available online)
and cell lysates were subjected to western blots (Figure S1B).
Three of the four siRNAs showed significant knockdown of
TLK2 and robust viral reactivation.
TLK2 Knockdown Leads to KSHV Reactivation
in KSHV-293 Cells
To validateourscreenresults,wereversetransfected KSHV-293
cells with equivalent amounts of glyceraldehyde 3-phosphate
dehydrogenase (GAPDH) siRNA or a pool of four siRNAs against
TLK1 and TLK2. The cells were imaged at 70 hr posttransfection
for GFP and RFP expression, and again knockdown of TLK2
robustly induced RFP expression (Figure 2A). We also harvested
cell lysates and performed western blots to look at kinase
expression levels. As shown in Figure 2B, siRNA transfection
led to dramatic decreases in the expression of the target protein.
To validate this result, we also used a TLK2 siRNA from
a different source that targeted a different part of the TLK2
transcript and measured viral reactivation. We saw the same
reactivation phenotype as described above (Figures S2A and
S2B). To determine whether TLK2 levels were increased in
we performed western blots probing for TLK2, and found that
expression levels of TLK2 were similar between infected and
uninfected cells (Figure S2C). We also tested whether knock-
down of TLK2 affected the cell viability of HEK293 cells in the
absence of KSHV. We transfected 293 cells with a nontargeting
control (NTC) siRNA or siRNAs against TLK2 or UBB, and exam-
ined them by microscopy at 72 hr to gauge viability (Figure S2D).
death in the TLK2 knockdown cells was similar to that in the NTC
siRNA transfected control cells and much less pronounced than
that in the positive control cells transfected with UBB siRNA.
We also examined the effect of TLK2 knockdown on HEK293
salt (MTS) assay, and found that TLK2 knockdown did not cause
an overall loss in proliferation as was seen following UBB knock-
down over the time period examined (Figure S2E).
Depletion of TLK2 Leads to Expression of KSHV Lytic
To determine whether TLK2 knockdown resulted in increased
viral lytic gene expression, we transfected KSHV-293 cells with
an NTC siRNA or siRNAs against TLK1 or TLK2. We also treated
mock-transfected KSHV-293 cells with 12-O-tetradecanoyl-
phorbol-13-acetate (TPA), a potent chemical inducer of KSHV
lytic reactivation, as a positive control. RNA was harvested
54 hr posttransfection and real-time quantitative PCR (qPCR)
was performed for three key viral lytic mRNAs: vIL-6, ORF57,
and vGPCR (Figure 2C). TLK2 knockdown cells induced expres-
sion of all three viral lytic mRNAs, to levels fairly similar to those
observed for the TPA-treated sample. The corresponding
western blots that show the protein levels of the targeted genes
are depicted in Figure 2D. To investigate whether depletion of
TLK2 led to an increase in viral genomes, we transfected
KSHV-293 cells with siRNAs against GAPDH, TLK1, or TLK2,
and extracted the total DNA 94 hr posttransfection. Viral
genomes were quantitated by qPCR (Figure 2E). TLK2 siRNA
transfected cells displayed an ?16-fold increase in viral genome
copy number compared with the GAPDH-depleted cells. Knock-
analysis (Figure 2F). This demonstrates that knockdown of TLK2
results in increased viral lytic gene expression and increased
viral genome replication.
The Major Viral Lytic Switch Protein, ORF50/RTA,
Is Activated Upon TLK2 Knockdown
The KSHV replication and transcription activator protein (RTA),
encoded byORF50,playsanessential rolein the initiation of viral
lytic gene expression. Since knockdown of TLK2 led to the
expression of viral lytic genes that were regulated by KSHV
ORF50, we wanted to determine whether ORF50 levels were
also increased. To determine whether ORF50 promoter activity
was increased following TLK2 knockdown, we performed a
reporter gene assay using a luciferase reporter gene under the
control of the ORF50 promoter. We transfected 293 cells with
siRNA against GAPDH, TLK1, or TLK2, and then transfected
the cells 24 hrlater with anORF50-promoter luciferase construct
(Damania et al., 2004). Luciferase expression was measured
48 hr posttransfection (Figure 3A). Depletion of TLK2 resulted
Table 1. List of Transfected siRNAs that Yield Increased RFP
Intensities at Least 2 SDs from the Mean when Transfected into
RFP Intensity2 SD 3 SD13 SD
Cell Host & Microbe
TLKs Modulate Gammaherpesvirus Reactivation
206 Cell Host & Microbe 13, 204–214, February 13, 2013 ª2013 Elsevier Inc.
in an ?10-fold increase in ORF50 promoter activity compared
with the GAPDH- and TLK1-depleted cells. Knockdown of the
targeted proteins was confirmed (Figure 3B). This indicates
that TLK2 knockdown leads to activation of the ORF50 pro-
moter, even in the absence of other viral proteins. Next, we
examined ORF50 transcript levels. KSHV-293 cells were trans-
fected with anNTC siRNAor siRNAs againstTLK1 or TLK2. Total
RNAwas harvested54 hrposttransfection,and qPCR forORF50
mRNA was performed. As can be seen in Figure 3C, there was
a 12-fold increase in ORF50 mRNA levels when TLK2 siRNA
was transfected into KSHV-293 cells compared with the control
siRNA. Additionally, lytic gene expression of two other viral
genes, K1 and ORF36, was also increased in TLK2 siRNA
transfected KSHV-293 cells (Figure 3D).
Depletion of TLK2 Leads to Complete Viral Reactivation
and Production of Infectious Virus
To test the effect of TLK2 knockdown on overall viral gene
expression, we performed genome-wide viral profiling using
our KSHV qPCR array (Dittmer, 2003). KSHV-293 cells were
transfected with the NTC siRNA or a siRNA targeting TLK1 or
TLK2. As a positive control for viral reactivation, we treated un-
transfected cells with 0.1 mM sodium butyrate, a known inducer
of KSHV reactivation. Cells were incubated for 96 hr and then
KSHV genome-wide transcription was measured. Similar to
the case with sodium-butyrate-treated cells, when TLK2 was
depleted from the KSHV-293 cells, nearly all of the viral genes
were upregulated, which is indicative of complete viral reactiva-
tion (Figure 4A; see also Figure S3). This was in contrast to the
control siRNA and TLK1 siRNA-treated cells, which showed
minimal levels of upregulated viral transcripts in KSHV-293 cells.
release of infectious virions from the cell that can subsequently
infect naive cells. To determine whether TLK2 knockdown led
to the production of infectious virions, we transfected KSHV-
293 cells with siRNAs targeting GAPDH, TLK1, or TLK2. Both
the cell lysates and supernatants were collected 96 hr posttrans-
fection. The lysates were used to perform western blots exam-
ining the expression of two viral lytic proteins (an early lytic
protein [vIL-6] and a late lytic protein [K8.1A]) that were highly
expressed only in the TLK2 siRNA-transfected cells (Figure 4B)
and not in the TLK1 or GAPDH siRNA-transfected cells.
Figure 2. TLK2 Plays a Role in KSHV Reactivation
four siRNAs against TLK1 or TLK2 was used. At 70 hr posttransfection, images were taken on a fluorescent microscope.
(B) siRNAs efficiently knock down the target. Cellular lysates from the samples imaged in Figure 2A were harvested and western blots were performed for TLK1,
TLK2, GAPDH, and the loading control, tubulin.
(C) Viral lytic mRNAs are expressed. KSHV-293 cells were either mock transfected or reverse transfected with 50 nM of the NTC siRNA or the pooled TLK1 or
TLK2 siRNAs. The mock-transfected sample was treated with 25 ng/ml of TPA at the time of transfection. Levels of the viral lytic transcripts vIL-6, Orf57, and
vGPCR were measured by qPCR at 54 hr posttransfection. Values are normalized to the control siRNA.
(D) Western blots were performed for the indicated proteins for the experiment described in (C).
(E) KSHV-293 cells were reverse transfected as described in (C). At 96 hr posttransfection, DNA was harvested and viral load was determined by qPCR. Primers
for Orf57 were used as the indicator of viral genome copies.
(F) Western blots were performed for the indicated proteins to confirm knockdown in the experiment described in (E). In Figure 2 asterisks denote nonspecific
bands. Error bars represent SD from the mean. Data were analyzed using a two-tailed type II Student’s t test for significance. See also Figure S2.
Cell Host & Microbe
TLKs Modulate Gammaherpesvirus Reactivation
Cell Host & Microbe 13, 204–214, February 13, 2013 ª2013 Elsevier Inc. 207
The supernatants collected from the KSHV-293 transfected
cells were clarified to remove cellular debris and then used to
infect naive Vero cells to determine whether infectious virus
was present. The Vero cells were monitored by fluorescence
microscopy for GFP-positive cells, which are indicative of viral
infection. Pockets of GFP-positive cells can clearly be seen in
the Vero cells treated with the supernatants from the TLK2
knockdown KSHV-293 cells (Figure 4C), indicating that infec-
tious virus was produced. We harvested these newly infected
Vero cells and extracted the total intracellular DNA to perform
a viral load assay. In agreement with our other data, naive Vero
cells incubated with the supernatants from the TLK2 knockdown
KSHV-293 cells displayed much higher levels of viral genomes
than cells incubated with the supernatants from the GAPDH
knockdown cells (Figure 4D).
TLK Depletion Can Also Reactivate KSHV from PEL
To examine the effect of TLK knockdown in KSHV-infected B
cells, we transfected BCBL-1 PEL cells with an NTC siRNA or
siRNAs against TLK1 or TLK2. At 120 hr posttransfection, the
cells were harvested and protein lysate was extracted and sub-
jected to western blot analysis against the viral lytic proteins,
vIL-6 and K8.1A. As can be seen in Figure 5A, BCBL-1 cells
transfected with the TLK2 siRNA showed viral reactivation, as
indicated by the increased expression of vIL-6 and K8.1A lytic
proteins compared with the control siRNA-transfected cells.
Cells transfected with TLK1 siRNA also showed a certain degree
of viral reactivation in PEL, but not to the same extent as TLK2
depletion, suggesting that depending on the particular cell
type, one of the two TLK genes may be predominantly involved
in modulating viral latency and suppressing reactivation. To
investigate the effect of TLK2 depletion on a variety of PEL cell
lines, we infected a panel of PEL cells (BC-3, JSC-1, and VG-1)
with either a control lentivirus or one targeting TLK2 for knock-
down. Infections proceeded for 96 hr and then western blots
for vIL-6 were performed on the harvested lysates (Figure 5B).
Viral reactivation was seen in each of the examined PEL cell
lines. These data show that depletion of TLKs in natural KSHV-
infected PEL cells leads to KSHV reactivation.
TLK2 Knockdown Leads to Decreased Phosphorylated
Histone H3 Bound to the ORF50 Promoter
The expression and repression of genes, both cellular and viral,
aretightly controlled bychromatin structureand modifications. It
was previously shown that TLKs can phosphorylate histone H3
was shown to modulate transcription (Burkhart et al., 2007; Goto
et al., 1999, Van Hooser et al., 1998; Lefebvre et al., 2002; Maha-
devan et al., 1991). In latently infected cells, the KSHV ORF50
promoter is associated with repressive histone, which is re-
leased upon reactivation (Gu ¨nther and Grundhoff, 2010; Toth
et al., 2010). It is possible that TLK2 depletion leads to viral reac-
tivation in KSHV-293 cells, in part because the histone H3 asso-
ciated with the ORF50 promoter is dephosphorylated, leading to
activation and expression of ORF50/RTA.
To determine whether depletion of TLK2 led to less phosphor-
ylated histone H3 associated with the ORF50 promoter,
we performed a chromatin immunoprecipitation (ChIP) assay.
KSHV-293 cells were transfected with either an NTC siRNA or
the TLK2 siRNA and incubated for 96 hr. ChIP analysis was per-
formed using an anti-phospho histone H3 (Ser10) or control
immunoglobulin G (IgG) antibody, and the amount of phosphor-
by qPCR. Following TLK2 depletion, there was an ?5.5-fold
reduction in the amount of serine 10-phosphorylated histone
H3 associated with the ORF50 promoter compared with the
control siRNA (Figure 5C). We also ran the qPCR reactions on
an agarose gel to visualize the PCR products (Figure 5D).
Because it is possible that the decreased association of phos-
pho-histone H3 with the ORF50 promoter is due to a global
Figure 3. The Major Lytic Switch Protein,
KSHV ORF50/RTA, Is Activated following
(A) The KSHV ORF50 promoter is activated by
TLK2 knockdown. HEK293 cells were transfected
with GAPDH siRNA or a pool of TLK1 or TLK2
siRNAs (50 nM) for 24 hr, followed by transfection
with an ORF50 luciferase reporter construct.
Lysate was harvested 48 hr after ORF50-lucif-
erase transfection and a luciferase assay was
performed. Values are normalized to protein levels
determined by Bradford assay.
(B) Western blots were performed for TLK1, TLK2,
and tubulin to confirm knockdown.
(C) Increase in ORF50 mRNA. Levels of ORF50
viral transcript were measured by qPCR as de-
scribed in (C).
(D) Induction of other viral lytic genes. KSHV-293
cells were reverse transfected with GAPDH siRNA
or apool of TLK1 or TLK2 siRNAs (50 nM)for 96 hr.
Total RNA was harvested and RT-PCR was per-
formed on the RNA (±RT) and PCR products were
run out on agarose gels. In Figure 3 error bars
represent SD from the mean. Data were analyzed
using a two-tailed type II Student’s t test for
significance. M, marker lane.
Cell Host & Microbe
TLKs Modulate Gammaherpesvirus Reactivation
208 Cell Host & Microbe 13, 204–214, February 13, 2013 ª2013 Elsevier Inc.