SUMO pathway dependent recruitment of cellular repressors to herpes simplex virus type 1 genomes.

Abstract

Components of promyelocytic leukaemia (PML) nuclear bodies (ND10) are recruited to sites associated with herpes simplex virus type 1 (HSV-1) genomes soon after they enter the nucleus. This cellular response is linked to intrinsic antiviral resistance and is counteracted by viral regulatory protein ICP0. We report that the SUMO interaction motifs of PML, Sp100 and hDaxx are required for recruitment of these repressive proteins to HSV-1 induced foci, which also contain SUMO conjugates and PIAS2β, a SUMO E3 ligase. SUMO modification of PML and elements of its tripartite motif (TRIM) are also required for recruitment in cells lacking endogenous PML. Mutants of PML isoform I and hDaxx that are not recruited to virus induced foci are unable to reproduce the repression of ICP0 null mutant HSV-1 infection mediated by their wild type counterparts. We conclude that recruitment of ND10 components to sites associated with HSV-1 genomes reflects a cellular defence against invading pathogen DNA that is regulated through the SUMO modification pathway.

Full text

Page 1

SUMO Pathway Dependent Recruitment of Cellular
Repressors to Herpes Simplex Virus Type 1 Genomes
Delphine Cuchet-Lourenc ¸o, Chris Boutell, Vera Lukashchuk, Kyle Grant, Amanda Sykes, Jill Murray,
Anne Orr, Roger D. Everett*
MRC-University of Glasgow Centre for Virus Research, Glasgow, Scotland, United Kingdom
Abstract
Components of promyelocytic leukaemia (PML) nuclear bodies (ND10) are recruited to sites associated with herpes simplex
virus type 1 (HSV-1) genomes soon after they enter the nucleus. This cellular response is linked to intrinsic antiviral
resistance and is counteracted by viral regulatory protein ICP0. We report that the SUMO interaction motifs of PML, Sp100
and hDaxx are required for recruitment of these repressive proteins to HSV-1 induced foci, which also contain SUMO
conjugates and PIAS2b, a SUMO E3 ligase. SUMO modification of PML and elements of its tripartite motif (TRIM) are also
required for recruitment in cells lacking endogenous PML. Mutants of PML isoform I and hDaxx that are not recruited to
virus induced foci are unable to reproduce the repression of ICP0 null mutant HSV-1 infection mediated by their wild type
counterparts. We conclude that recruitment of ND10 components to sites associated with HSV-1 genomes reflects a cellular
defence against invading pathogen DNA that is regulated through the SUMO modification pathway.
Citation: Cuchet-Lourenc ¸o D, Boutell C, Lukashchuk V, Grant K, Sykes A, et al. (2011) SUMO Pathway Dependent Recruitment of Cellular Repressors to Herpes
Simplex Virus Type 1 Genomes. PLoS Pathog 7(7): e1002123. doi:10.1371/journal.ppat.1002123
Editor: Karen L. Mossman, McMaster University, Canada
Received January 24, 2011; Accepted May 3, 2011; Published July 14, 2011
Copyright: ? 2011 Cuchet-Lourenc ¸o et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was funded by the Medical Research Council. The funders had no role in study design, data collection and analysis, decision to publish, or
preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: roger.everett@glasgow.ac.uk
Introduction
Herpesvirus infections are controlled by acquired and innate
defences involving cellular, humoral and cytokine mediated
responses (for reviews, see [1]). In recent years a concept has
emerged of an additional antiviral defence mechanism that
operates within individual cells. Unlike cytokine-mediated re-
sponses, intrinsic antiviral resistance involves the actions of pre-
existing cellular proteins that, in the case of herpesviruses, act to
repress viral transcription [2,3,4]. This defence is counteracted by
viral regulatory proteins, for example the immediate-early (IE)
proteins ICP0 of herpes simplex virus type 1 (HSV-1) [5,6,7], ie1
(IE72) of human cytomegalovirus (HCMV) [8], and HCMV virion
component pp71 [9,10,11,12,13,14]. One aspect of intrinsic
resistance concerns cellular nuclear sub-structures known as
ND10 or promyelocytic leukaemia (PML) nuclear bodies, and a
number of their major components, namely PML itself, Sp100,
hDaxx and ATRX. In HSV-1 infections, ICP0 overcomes the
repressive properties of these proteins by inducing their degrada-
tion or dispersal [7,15,16,17]. ICP0 null mutant HSV-1 exhibits a
greatly reduced plaque forming efficiency, but this defect is
partially reversed in cells depleted of PML, Sp100, hDaxx or
ATRX [5,6,7].
A notable feature of PML and other ND10 components is their
recruitment to novel ND10-like foci that are closely associated
with parental HSV-1 genomes and early replication compartments
during the initial stages of infection [18,19]. The recruitment of
PML to the virus-induced foci is not dependent on de novo viral
protein expression and occurs extremely rapidly, indicating that
the cell responds to the entry of viral genomes into the nucleus
[18,20]. Although the effect can be seen in wild type (wt) HSV-1
infections, it is short lived as the recruited proteins are rapidly
degraded or dispersed through the effects of ICP0. During
infection with ICP0 null mutant HSV-1, however, the ND10
proteins remain in these novel sites in a much longer-lived
manner. The correlation between the biological activity of many
ICP0 mutant proteins and their ability to counteract this
recruitment process [21] suggests that this phenomenon reflects
an aspect of intrinsic antiviral resistance. This model proposes that
the recruited proteins generate a repressive environment that
impedes viral transcription. This paper concerns the molecular
mechanism by which ND10 components are recruited to the virus-
induced foci and tests the hypothesis that the recruitment process
contributes to intrinsic resistance to HSV-1 infection.
The formation of the virus-induced ND10-like structures is
distinct from that of normal ND10 in uninfected cells because it is
not dependent on PML or indeed any of the major ND10
components that have so far been studied [5,6,7,20]. Here, we
have used a depletion/reconstitution approach to analyze the
molecular requirement for the recruitment of PML, Sp100 and
hDaxx to HSV-1 genome-associated sites in newly infected cells.
We found that in all cases the presence of a SUMO interaction
motif (SIM) [22] is required for this property, and that the major
SUMO modification sites of PML, but not Sp100, are also
required. The virus-induced ND10-like foci also contain SUMO-
2/3 conjugates and PIAS2b, a SUMO E3 ligase, even in the
absence of PML. Unlike their wt counterparts, SIM deletion
mutants of PML isoform I and hDaxx are unable to repress ICP0
null mutant HSV-1 infection. These data demonstrate that
SUMO modification pathways play a key role in the recruitment
PLoS Pathogens | www.plospathogens.org1July 2011 | Volume 7 | Issue 7 | e1002123

Page 2

of ND10 proteins to HSV-1 genome associated sites. We propose
that this SUMO-dependent cellular response is an important
component of intrinsic cellular defence against foreign DNA that
has entered the nucleus.
Results
Recruitment of PML isoforms to sites associated with
HSV-1 genomes
The recruitment of PML and other proteins to sites associated
with parental HSV-1 genomes and early replication compartments
can be detected by examination of cells at the periphery of
developing viral plaques. The viral genomes enter the nucleus of
newly infected cells in a directional and asymmetric manner, and
remain close to the nuclear envelope. The viral IE transcriptional
regulator ICP4 avidly binds to viral DNA, and therefore forms foci
that, in cells at the early stages of infection, are commonly
distributed along one interior edge of the nucleus. PML and other
ND10 proteins then accumulate at novel sites that are closely
associated with the viral genomes [18,19]. This phenotype allows
the unambiguous identification of proteins that have been
recruited to and accumulate at the virus-induced sites. A typical
example of endogenous PML recruitment in an ICP0-null mutant
HSV-1 (DICP0) infected HepaRG cell expressing a control
shRNA is shown in Figure 1A (left) in comparison with a similarly
infected PML depleted cell (Figure 1A, right).
We investigated the molecular characteristics of PML that are
required for this recruitment by adopting a depletion/reconstitu-
tion approach [23]. All the PML isoforms studied (Figure 1B),
expressed as EYFP fusions, were recruited to sites associated with
ICP4 foci in cells containing endogenous PML (Figure 1C). In cells
depleted of endogenous PML, however, while PML isoforms I to
V were recruited, PML.VI remained entirely in large foci that
were not associated with ICP4 (Figure 1D). The separated
channels of the images in Figures 1C and 1D are shown in
Figures S1 and S2. This assay is not amenable to precise
quantification as the extent of recruitment varies between cells and
is most obvious when the ICP4 foci are small. Nonetheless, when
recruitment occurs it is evident in all cells with ICP4 foci near the
nuclear periphery. A large number of cells were examined in each
experiment, and we have scored a protein as defective in
recruitment only when its behaviour was clearly different from
the appropriate control. As an example, compare PML.VI in the
control and PML depleted backgrounds in Figures 1C and D.
Although recruitment of PML proteins that we have scored as
recruitment positive occurred to some extent in all relevant cells,
we noted some differences in PML isoform behaviour. For
example, PML.I was recruited more extensively than PML.V
(Figure 1D). We have not investigated the basis of these
differences.
The defect in PML.VI recruitment to virus-induced foci also
occurred in analogous human fibroblast (HF) derived cells. PML.I
co-localized with Sp100 in both control and PML depleted HFs
(Figure S3), both PML.I and PML.VI were recruited to the virus-
induced in control HFs (Figure S4A), but PML.VI was not
recruited in PML depleted HFs (Figure S4B).
PML isoform VI includes all the conserved exons present in the
other isoforms, except exon 7a (Figure 1B). Therefore exon 7a
includes sequences that are required for recruitment of PML to the
virally induced foci. The defect in PML.VI recruitment is not
exhibited when endogenous PML is present because PML
interactions mediated through the coiled-coil motif [24,25] enable
the defective protein to be recruited in partnership with the
endogenous isoforms.
PML exon 7a is required for recruitment to sites
associated with HSV-1 genomes
PML exon 7a encodes an 18 amino acid segment containing a
SIM that has been implicated in proper ND10 assembly [26].
Mutants of PML.I and PML.IV that lack exon 7a (Figure 2A).
were expressed at levels similar to their wt versions, with similar
patterns of SUMO modification, and co-localized with Sp100 in
both control and PML depleted cells (Figure 2B and C). While
both mutants were recruited to virus-induced foci in the presence
of endogenous PML, both were defective in recruitment in PML
depleted cells (Figures 2D and S5A). This defect of PML.I.D7a was
confirmed in HF derived cells (Figure S5B and C; the localization
of this protein in uninfected HFs is shown in Figure S3). EYFP-
PML.I with substitution mutations in the SIM (PML.I.mSIM)
exhibited properties very similar to those of PML.I.D7a in
uninfected and ICP0-null mutant HSV-1 infected control and
PML depleted cells (Figure S6). These data confirm that the lack of
recruitment of PML.VI is due to the absence of the SIM in exon
7a.
The role of SUMO modification in the recruitment of PML
to virus-induced foci
SUMO modification of PML is essential for ND10 assembly in
uninfected cells [27]. Therefore we investigated whether this
modification, in addition to the SIM, plays a role in PML
recruitment to the virus-induced foci. Mutants of PML.I or
PML.IV with lysine to arginine substitutions at residues 160 and
490 (PML.I.KK and PML.IV.KK; Figure 3A) exhibit substantial
defects in SUMO modification, but some apparently modified
species remain, especially in cells containing endogenous PML,
and more so with PML.IV than PML.I. Modification of
PML.I.KK in PML depleted cells was almost completely
eliminated (Figure 3B).
PML.I.KK and PML.IV.KK localized to ND10 normally in
cells expressing endogenous PML, probably through interactions
with endogenous PML via their coiled-coil domains (Figure 3C,
Author Summary
Viruses encounter several different defences that impede
infection, including acquired immunity mediated by the
immune system and innate immunity that includes the
synthesis of antiviral proteins through the interferon
pathway. In recent years, a third arm of antiviral defence
has been described, named intrinsic immunity or intrinsic
resistance, that is conferred by constitutively expressed
cellular proteins. In the case of herpesviruses, intrinsic
resistance involves the action of cellular repressors that
restrict viral transcription once the viral genome enters the
nucleus. Several studies have presented evidence that one
aspect of intrinsic resistance involves cellular proteins that
form distinct nuclear structures known as ND10. Several
ND10 components are known to accumulate rapidly at
sites in close association with herpes simplex virus type 1
genomes. Here we report that this cellular response
requires the ability of several of the proteins in question
to interact with a small ubiquitin-like protein known as
SUMO. In two such examples of these proteins, we show
that their ability to interact with SUMO is required for their
roles in repressing viral infection. We suggest that this
SUMO-dependent pathway may underlie a more general
mechanism by which cells protect themselves from
invading foreign DNA.
SIM Dependent Recruitment of PML to HSV-1 Genomes
PLoS Pathogens | www.plospathogens.org2July 2011 | Volume 7 | Issue 7 | e1002123

Page 3

upper row). In contrast, and consistent with previous work [27],
PML.I.KK exhibited a drastically altered localization in PML
depleted cells, forming a reduced number of foci of increased size,
some of which were in the cytoplasm. The nuclear foci contained
greatly reduced amounts of Sp100. The aberrant PML.IV.KK foci
were predominantly nuclear, again with little co-localization with
Sp100 (Figure 3C, lower row). Residual co-localization with Sp100
of these PML mutants may be influenced by potential SUMO
modification at other lysine residues. However, it is also affected
by cell type since in the corresponding PML depleted HFs co-
localization of PML.I.KK with Sp100 was more marked (Figure
S9A).
To eliminate any residual SUMO modification of PML.I.KK
we introduced additional mutations at either lysine 65 [28] or at
lysine 616, which lies in a good match (LKID) to the consensus
SUMO modification site (YKXE). Mutant PML.I.K123 (K65R,
K160R, K490R) exhibited similar residual SUMO modification to
that of PML.I.KK in cells expressing endogenous PML, whereas
the K616R mutation virtually eliminated all modified bands in
mutants PML.I.K234 (K160R, K490R, K616R) and PML.I.4KR
(all indicated lysine residues substituted with arginine) (Figure S7).
We conclude that PML.I is not detectably modified at lysine 65,
but that lysine 616 is likely to be a SUMO modification site that is
specific for PML.I and PML.IV (because it lies in exon 8a which is
present in these two isoforms only). We note that lysine 65 is very
close to zinc coordinating cysteine residues in the RING finger, so
SUMO modification here would likely affect both the overall
architecture oftheRING and
PML.I.4KR co-localized with Sp100 in control but not PML
depleted cells, in both HepaRG and HF backgrounds (Figures S8A
and S9A). These data support the hypothesis that the sporadic co-
localization of PML.I.KK with Sp100 in PML depleted cells
(Figure 3C) is due to the remaining potential for SUMO
modification.
As in the case of the SIM mutants, both PML.I.KK and
PML.IV.KK could be recruited to virus-induced foci in cells
expressing endogenous PML (Figure 3D, upper row), but
recruitment was either absent or at greatly reduced levels in
PML depleted cells (Figure 3D, lower row, and Figure S8B).
Similar results were obtained in the equivalent HF derived cells
(Figure S9B), and with the 4KR mutant in both HepaRG and HF
backgrounds (Figures S8C and S9B). Therefore, in addition to the
role of the SIM, mutation of the major SUMO modification sites
of PML severely impedes the capacity of the protein to be
recruited to sites associated with viral genomes, providing
endogenous PML isoforms are absent.
itsinteractioninterfaces.
Absence of the SIM does not affect the intrinsic mobility
of PML
The defect of mutant PML proteins in recruitment to virus-
induced foci could be explained by a decrease in intrinsic mobility.
The dynamics of PML isoforms and SUMO modification deficient
mutants in both the presence and absence of endogenous PML
have been reported [18,29,30,31], but the role of the SIM has not
been investigated before. We analyzed the dynamics of all EYFP-
linked PML isoforms, and the D7a and KK mutants of PML.I and
PML.IV, in both control and PML depleted HepaRG cells.
Figure 1. The major nuclear isoforms of PML and their
recruitment to sites associated with HSV-1 genomes. A. Typical
examples of recruitment of endogenous PML (green) to sites associated
with HSV-1 genomes (ICP4: red) in cells at the edges of ICP0 null mutant
(DICP0) plaques in control and PML depleted cells. B. PML isoforms I to
VI, noting the included exons and the translated product length. The
bracketed exons indicate the use of alternative reading frames for these
sequences. (+8) at the end of PML VI indicates an additional 8 residues
following exon 6 as a result of alternative splicing that deletes exon 7a.
Adapted from [23]. C and D. Recruitment of EYFP-PML isoforms in
control (panel C) and PML depleted cells (panel D). Scale bars indicate
5 mm. Analysis of the localization of these proteins in uninfected cells
and their expression levels is presented elsewhere [23].
doi:10.1371/journal.ppat.1002123.g001
SIM Dependent Recruitment of PML to HSV-1 Genomes
PLoS Pathogens | www.plospathogens.org3 July 2011 | Volume 7 | Issue 7 | e1002123

Page 4

Typical FRAP recovery plots of PML.I and PML.I.KK in control
cells are presented in Figure 4A, confirming previous conclusions
that SUMO modification is required for the normal mobility of
PML [30,32]. The data for all PML isoforms and mutants in both
cell backgrounds are presented as their degree of fluorescence
recovery at the 112-second time point (Figure 4B). The results are
broadly in agreement with previous work, with the exception that
the relatively slow recovery of PML.V [30] was not reproduced
here. The only major differences were between the SUMO
modification deficient mutants and the wt forms. Importantly, the
defect in recruitment to virus-induced foci of PML.VI and the
PML.I and PML.IV mutants lacking exon 7a cannot be explained
simply by reduced mobility, whereas in the case of the SUMO
modification mutants this explanation cannot be excluded.
The role of elements within the TRIM in recruitment of
PML to virus-induced foci
Control and PML depleted cell lines that express EYFP-PML.I
with a deletion of the coiled-coil or point mutations in the RING
finger, B-Box 1 or B-Box 2 (Figure 5A) have been described
previously [23]. Briefly, the RING finger mutant (DRING) co-
localized with endogenous ND10 in control cells but was mostly
nuclear diffuse in cells lacking endogenous PML, the coiled-coil
deletion mutant (DCC) was nuclear diffuse in both cell
backgrounds, the B-Box 2 mutant (DBB2) formed foci in both
types of cell but in neither case were these co-localized with Sp100,
and the B-Box 1 mutant (DBB1) co-localized with Sp100 in
aberrant ND10-like structures in control cells and formed foci in a
proportion of PML depleted cells, none of which co-localized with
Sp100. SUMO modification of all tripartite motif (TRIM) mutants
was highly compromised in both cell backgrounds [23].
The DRING and DBB2 mutants were recruited efficiently to the
virus-induced foci in cells expressing endogenous PML, whereas
the DBB1 and DCC mutants were not (Figures 5B and S10A). The
DBB2 mutant was also efficiently recruited in PML depleted cells
(Figure 5B), even though this mutant was not well SUMO-
modified and did not efficiently co-localize with Sp100 in
uninfected cells in either cell background [23]. Therefore the
characteristics of PML required for nucleating an ND10-like
structure in uninfected cells are distinguishable from those
involved in recruitment to the novel virus-induced structures. As
in cells expressing endogenous PML, the DBB1 and DCC mutants
were not recruited into virus-induced foci in PML depleted cells
(Figure 5B). The most variable results were obtained with the
DRING mutant in PML depleted cells. In most cells infected with
ICP0 null mutant HSV-1 this mutant remained nuclear diffuse,
but in some cells it was recruited with variable efficiencies
(Figure 5B). We conclude that whereas B-Box 1 and the coiled-coil
are essential for PML recruitment to virus induced foci, B-Box 2 is
dispensable, and inactivation of the RING greatly diminishes but
does not always eliminate recruitment.
The SIM of Sp100 is required for recruitment to HSV-1
induced foci
Given that recruitment of PML to sites to HSV-1 induced foci is
dependent on its SIM, we investigated whether recruitment of
other ND10 proteins is also SIM dependent. Sp100A is the
smallest and most abundantly expressed Sp100 isoform. It includes
a SIM (residues 323–326), a major SUMO modification site at
Figure 2. SIM deletion mutants of PML are defective in
recruitment to sites associated with HSV-1 genomes. A. Maps
of PML isoforms I and IV and derivatives lacking exon 7a. B. Western
blot of the EYFP-linked proteins in control and PML depleted cells using
an anti-EGFP antibody. The major unmodified bands of the EYFP-PML
proteins are indicated by asterisks. C. Localization of EYFP-PML proteins
(green) and endogenous Sp100 (red) in uninfected control and PML
depleted cells. D. Typical assays of recruitment of EYFP-proteins (green)
to sites associated with HSV-1 genomes (ICP4; red) in ICP0 null mutant
(DICP0) infected control and PML depleted cells. Scale bars indicate
5 mm.
doi:10.1371/journal.ppat.1002123.g002
SIM Dependent Recruitment of PML to HSV-1 Genomes
PLoS Pathogens | www.plospathogens.org 4July 2011 | Volume 7 | Issue 7 | e1002123

Page 5

lysine 297, and a region near the N-terminus that is required for
localization to ND10 the HSR domain (Figure 6A) [33,34].
Control and Sp100-depleted HepaRG cells [6] were transduced
with lentivirus vectors expressing EYFP-linked Sp100A and
mutants lacking the HSR domain (DHSR), the SUMO modifica-
tion site (K297R) or the SIM (mSIM), or a combination of both
K297R and mSIM mutations (DSSIM). Consistent with previous
studies [33,34], none of the mutant proteins exhibited the SUMO
modification pattern characteristic of wt Sp100A (Figure 6C). The
mSIM mutant of Sp100A at least partially co-localized with PML
in control cells, and also in Sp100 depleted cells expressing higher
levels of the recombinant protein (Figure 7B, leftmost panels),
although it was nuclear diffuse when weakly expressed in Sp100
depleted cells (not shown).
Both wt and K297R mutant EYFP-Sp100A were recruited to
the virus-induced foci in both control and Sp100 depleted cells
Figure 3. SUMO modification mutants of PML are defective in
recruitment to sites associated with HSV-1 genomes. A. Maps of
PML isoforms I and IV and derivatives with lysine to arginine
substitutions at residues K160 and K490. B. Western blot of these
proteins in control and PML depleted cells, using an anti-EGFP antibody.
The major unmodified bands of the EYFP-PML proteins are indicated by
asterisks. C. Localization of the EYFP-PML proteins (green) and
endogenous Sp100 (red) in uninfected control and PML depleted cells.
D. Typical assays of recruitment of these PML proteins (green) to sites
associated with HSV-1 genomes (ICP4; red) in ICP0 null mutant (DICP0)
infected control and PML depleted cells. Scale bars indicate 5 mm.
doi:10.1371/journal.ppat.1002123.g003
Figure 4. FRAP analysis of wt PML isoforms and SUMO
modification and SIM deletion mutants of PML.I and PML.IV
in control and PML depleted cells. A. Examples of the original data
of PML.I and PML.I.KK in control HepaRG shLuci cells.B. The means and
standard deviations of the percentage recoveries at the final time point
for each of the proteins in both cell backgrounds. Statistical significance
between the indicated data sets was calculated by Student’s two-tailed
t-Test, p=,0.0001.
doi:10.1371/journal.ppat.1002123.g004
SIM Dependent Recruitment of PML to HSV-1 Genomes
PLoS Pathogens | www.plospathogens.org5 July 2011 | Volume 7 | Issue 7 | e1002123