Engagement of the Lysine-Specific Demethylase/HDAC1/CoREST/REST Complex by Herpes Simplex Virus 1

Abstract

Among the early events in herpes simplex virus 1 replication are localization of ICP0 in ND10 bodies and accumulation of viral DNA-protein complexes in structures abutting ND10. ICP0 degrades components of ND10 and blocks silencing of viral DNA, achieving the latter by dislodging HDAC1 or -2 from the lysine-specific demethylase 1 (LSD1)/CoREST/REST repressor complex. The role of this process is apparent from the observation that a dominant-negative CoREST protein compensates for the absence of ICP0 in a cell-dependent fashion. HDAC1 or -2 and the CoREST/REST complex are independently translocated to the nucleus once viral DNA synthesis begins. The focus of this report is twofold. First, we report that in infected cells, LSD1, a key component of the repressor complex, is partially degraded or remains stably associated with CoREST and is ultimately also translocated, in part, to the cytoplasm. Second, we examined the distribution of the components of the repressor complex and ICP8 early in infection in wild-type-virus- and ICP0 mutant virus-infected cells. The repressor component and ultimately ICP8 localize in structures that abut the ND10 nuclear bodies. There is no evidence that the two compartments fuse. We propose that ICP0 must dynamically interact with both compartments in order to accomplish its functions of degrading PML and SP100 and suppressing silencing of viral DNA through its interactions with CoREST. In turn, the remodeling of the viral DNA-protein complex enables recruitment of ICP8 and initiation of formation of replication compartments.

The major tasks of the tegument proteins introduced during infection and those expressed immediately afterwards are to preclude cellular response to infection and at the same time suppress attempts by the host cell to silence the expression of viral DNA. Thus, the tegument protein product of U_L41, an endoribonuclease, selectively degrades RNAs to preclude the expression of cellular stress response genes (23, 24), whereas VP16 recruits cellular proteins to enhance the expression of α (immediate-early) genes. Of the α proteins, ICP27 contributes to the inhibition of host responses by blocking RNA splicing, ICP47 blocks presentation of antigenic peptides to the immune system, and ICP0 at the very least blocks the activation of antiviral response by interferon and suppresses the silencing of post-α gene expression (18). This report concerns the antisilencing activities of ICP0. In brief, in earlier reports we showed that ICP0 shares amino acid homology over a span of 80 amino acids and physically interacts with CoREST, a component of the complex containing HDAC1 or -2/CoREST/REST and a key repressor or neuronal genes in nonneuronal cells (6). We also reported that in transfected cells, ICP0 dislodges HDAC1 from the CoREST/REST complex and that in infected cells, HDAC1 and the CoREST/REST complex are exported from the nucleus to the cytoplasm by a process independent of ICP0 (6). Subsequently, we reported that a truncated form of CoREST lacking the HDAC1 binding site compensated in a cell-dependent fashion for the absence of ICP0. Thus, a mutant virus in which ICP0 was replaced with the truncated CoREST protein replicated 100-fold better than the parent ΔICP0 mutant in Vero cells and 10- to 50-fold better in other cell lines (7).

This report deals with two aspects of the interaction of herpes simplex virus 1 (HSV-1) with the complex containing HDAC1 or -2/CoREST/REST. The first aspect concerns lysine-specific demethylase 1 (LSD1), a key component of the complex. The question posed is whether the interaction of LSD1 with the HADC/CoREST/REST complex is altered in the course of productive infection. The second question posed in our studies involves the physical locations of the complex vis-à-vis ND10 structures with which incoming DNAs aggregate and ICP8, which signals the formation of replication compartments.

LSD1, a 110-kDa protein, was initially copurified with the HDAC/CoREST/REST complex (9, 27) and later characterized as the first true histone demethylase (19). LSD1 is a FAD-dependent amine oxidase homolog highly conserved in organisms ranging from yeasts to humans (3). In vitro, LSD1 specifically demethylates mono- and dimethyl histone H3 lysine 4 (H3K4me1/2) but does not act on H3K4me3 (trimethyl) (19). H3K4 methylation is generally associated with transcriptional activation (21, 22). The demethylation of H3K4 by LSD1 and its interaction with CoREST/REST/HDAC repressors suggest that LSD1 most likely coordinates with histone deacetylation to play a role in transcriptional repression. Knockdown of LSD1 by small interfering RNA indeed resulted in increased amounts of H3K4me2 (dimethyl) on neuron-specific genes and their transcription in nonneuronal cell lines (19).

LSD1 and CoREST stably interact via a long helical region of CoREST intertwining along the helices of the LSD1 tower domain (26). This interaction protects LSD1 from proteasomal degradation in vivo and also enhances both substrate recognition and demethylating activity (20). LSD1 requires the first 20 amino acids of the H3 tail for high demethylase efficiency (4). The requirement for a long stretch of the substrate peptide may allow LSD1 to sense the histone code embedded in the modification of surrounding amino acids. Indeed, HDAC inhibitors reduce LSD1 demethylation activity, suggesting that hypoacetylated nucleosomes are preferable substrates (20). Interestingly, LSD1 has also been implicated in H3K9 demethylation in cases of nuclear receptor-dependent gene activation (5, 15). The contrasting roles of LSD1 in different biological processes raised questions regarding its role in HSV gene expression.

In this report, we show that in wild-type virus-infected cells, LSD1 is partly degraded in a proteasome-dependent manner. Of particular interest is that in infected cells, HDAC1, CoREST, and LSD1 are initially dispersed throughout the cytoplasm but coaggregate with ICP8 in dense structures defined as replication compartments (17). All three proteins are exported at least in part to the cytoplasm, but comparisons of cells infected for 4 and 8 h suggest that LSD1 is translocated to the cytoplasm later than HDAC1 or CoREST. A most significant finding is that the HDAC1/LSD1/CoREST/REST complexes remaining in the nucleus colocalize at least initially with ICP8 in the immediate vicinity of the ND10 structures. Since it is likely that all these proteins interact with viral DNA but that ICP0 is detected primarily in the ND10 structures, we propose the hypothesis that ICP0 dynamically interacts with proteins in ND10 bodies and with DNA-protein complexes that abut the ND10 structures.

MATERIALS AND METHODS

Cells and viruses.

HeLa cells originally obtained from the American Type Culture Collection (ATCC) were grown in Dulbecco's modified Eagle's medium supplemented with 5% fetal bovine serum. Human embryonic lung (HEL) fibroblasts immortalized by transduction with telomerase, a kind gift from Thomas E. Shenk (Princeton University), were grown as described above but in medium supplemented with 10% serum. HSV-1(F) is the prototype HSV-1 strain used in this laboratory. ΔICP0 mutant viruses R7910 and R8501 were described elsewhere (7, 12).

Immunoprecipitation (IP).

HeLa cells grown in 25-cm² flasks were mock infected or exposed to virus. At the time points indicated in Results, the cells were lysed in 0.5 ml of lysis buffer (10 mM Tris, pH 8.0, 140 mM NaCl, 1.5 mM MgCl₂, 1 mM dithiothreitol, 0.5% Nonidet P-40, 0.1 mM NaVO4, 10 mM NaF, 10 mM glycerol phosphate, 1× protease inhibitor cocktail). After brief sonication and centrifugation, the cell debris was discarded or dissolved in 1× sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer as shown in Fig. 2. The supernatant fluids were reacted with polyclonal anti-CoREST antibody or polyclonal anti-LSD1 antibody overnight at 4°C. The lysate-antibody mixtures were then incubated with 50 μl of protein A Sepharose CL-4B 50% slurry (GE Healthcare, Pittsburgh, PA) at 4°C for 1 h. The beads were then rinsed four times with the same buffer and eluted by 1× SDS-PAGE loading buffer.

FIG. 2.

Dissociation of HDAC1 from CoREST/LSD1 after HSV-1 infection. HeLa cells were mock or HSV-1(F) infected at 10 PFU per cell for 24 h before being harvested. The cells were washed and lysed in 0.5 ml of lysis buffer by brief sonication. After centrifugation, the pellet of cell debris was dissolved in 0.5 ml of SDS-PAGE loading buffer and sonicated to decrease the viscosity. The cleared lysates were divided into two aliquots and incubated overnight with 3 μl of polyclonal anti-LSD1 antibodies (Ab). Precipitates captured with both protein A Sepharose (IP) and post-IP supernatant fluids were electrophoretically separated in a denaturing gel, transferred to a polyvinylidene difluoride membrane, and immunoblotted with antibodies as indicated. “Input” indicates 5% of the total lysate used in IP. Compared to the input, equal volumes of a sonicated insoluble debris pellet (lanes 1 and 2) and post-IP supernatant (lanes 5 and 6) were also loaded on the same gel. Panel A shows a lighter exposure of the images of anti-LSD1 antibody reacting with LSD1 antibody. Panel B shows a darker exposure designed to visualize LSD1 and CoREST in post-IP supernatant fluids.

Immunoblots.

Total cell lysates or immunoprecipitates were electrophoretically separated on denaturing polyacrylamide gels and transferred to preequilibrated poly (vinylidene difluoride) membrane (Millipore, Billerica, MA). The membranes were blocked in TBST (20 mM Tris, pH 7.5, 150 mM NaCl, 0.5% Tween 20) containing 5% nonfat dry milk, reacted at 4°C overnight with a primary antibody in TBST-5% dry milk, rinsed, and reacted with alkaline phosphatase-conjugated secondary antibodies (Bio-Rad, Hercules, CA) or peroxidase-conjugated secondary antibodies (Sigma, St. Louis, MO). The membranes were rinsed and developed with either BCIP (5-bromo-4-chloro-3-indolylphosphate) plus nitroblue tetrazolium (Denville Scientific, Inc., Metuchen, NJ) or enhanced chemiluminescence Western blotting detection reagents (GE Healthcare, Pittsburgh, PA) according to the manufacturer's instructions. In experiments involving immune precipitation, the amount of the input represented 5% of the amount of lysate that was reacted with the antibody.

Confocal microscopy.

HEL cells seeded on four-well slides (Thermo Scientific, Waltham, MA) were mock infected or virus infected as described in Results. The cells were then fixed with 4% paraformaldehyde for 10 min, permeabilized with 0.2% Triton X-100 for 10 min, and blocked in phosphate-buffered saline containing 1% bovine serum albumin and 5% horse serum. The slides were then reacted with primary antibodies, stained with fluorescein isothiocyanate-conjugated goat anti-rabbit (Sigma, St. Louis, MO) and Texas Red-conjugated goat anti-mouse (Invitrogen, Carlsbad, CA) secondary antibodies, and then mounted with VectaShield mounting medium (Vector Laboratories, Burlingame, CA). The images were taken with a Zeiss LSM410 confocal microscope.

Antibodies.

Rabbit polyclonal anti-LSD1 ab17721 was initially obtained from Thomas M. Kristie (NIH) and subsequently purchased from Abcam (Cambridge, MA). Polyclonal anti-CoREST was from Millipore (Billerica, MA). Polyclonal anti-HDAC1 and anti-HDAC2 were from Sigma (St. Louis, MO). Polyclonal anti-Daxx and mouse monoclonal anti-PML were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Mouse monoclonal anti-ICP8 and anti-ICP27 were purchased from Rumbaugh-Goodwin Institute for Cancer Research, Inc. (Plantation, FL). Rabbit polyclonal anti-U_L38 has been described elsewhere (25).

RESULTS

Only a small fraction of LSD1 is available for pulldown with anti-CoREST antibody in infected cells.

In uninfected cells, LSD1 forms a complex with CoREST. The objective of this series of experiments was to determine whether LSD1 is bound to CoREST in HSV-1(F)-infected cells. HeLa cells grown in 25-cm² flask cultures were harvested at 4 or 15 h after exposure to 5 PFU of HSV-1(F) per cell, lysed, and reacted with antibody against CoREST as described in Materials and Methods. The immune precipitate was solubilized, subjected to electrophoresis in a denaturing gel, and probed with antibodies against LSD1, HDAC1, or HDAC2. Input controls consisting of 5% of the amount reacted with antibody were processed concurrently in the same manner. The results (Fig. 1) were as follows.

1. LSD1, HDAC1, and HDAC2 were detected in input samples from mock-infected cells and cells infected for 4 h (Fig. 1, lanes 1 to 4). HDAC1 and HDAC2 were detected in lysates of cells harvested 15 h after infection. LSD1 was readily detected in overexposed images (not shown) but was barely detectable (lane 4) in the lighter exposure shown in Fig. 1. It is noteworthy that fractions of HDAC1 from lysates of cells harvested at 4 and 15 h after infection migrated more slowly, consistent with an earlier report that HDAC1 is phosphorylated in infected cells (16).

2. Antibody to CoREST pulled down LSD1 from lysates of mock-infected cells and cells harvested 4 h after infection. LSD1 was also pulled down from lysates of infected cells, but these amounts of LSD1 pulled down from lysates at 4 and 15 h after infection were smaller than those pulled down from corresponding mock-infected cells.

3. Both HDAC1 and HDAC2 were coprecipitated down from lysates of mock-infected cells and lysates of cells infected for 4 h. Neither HDAC1 nor HDAC2 was pulled down by lysates of cells harvested at 15 h after infection. This result is congruent with those published earlier (6).

FIG. 1.

Alteration of components of the complex containing CoREST/LSD1/HDAC1 or -2 after HSV-1 infection. HeLa cells were mock or HSV-1(F) infected with 5 PFU per cell for 4 or 15 h. Total cell lysates were immunoprecipitated with rabbit polyclonal anti-CoREST antibody (Ab) overnight. The precipitates captured with protein A Sepharose were electrophoretically separated, transferred to a polyvinylidene difluoride membrane, and immunoblotted with antibodies as indicated.

We conclude that late in infection, the amount of total LSD1 in infected cells decreases but that CoREST retains the ability to bind LSD1 although it is no longer able to bind HDAC1 or HDAC2.

Anti-LSD1 antibody pulls down CoREST from lysates of infected cells.

The results of the experiments described above suggested that some of the CoREST protein is no longer partnered with LSD1. The experiments described here were done to determine whether LSD1 in infected cells is degraded or modified to the point where it no longer interacts with CoREST. In this series of experiments, HeLa cells were harvested 24 h after mock infection or infection with 10 PFU of HSV-1 per cell. The cells were rinsed, solubilized, cleared of cell debris by centrifugation, and reacted with antibody to LSD1. The debris collected after the first centrifugation, consisting of insoluble materials, the immune complexes, and the supernatant fluids saved after the immune complexes, were collected, solubilized, subjected to electrophoresis in denaturing gels, and probed with antibody against HDAC1, CoREST, or LSD1. The image in Fig. 2A is a lighter exposure of the bands reacting with the anti-LSD1 antibody. Fig. 2B is a darker exposure designed to visualize the small amounts of LSD1 and CoREST in the supernatant fluid after removal of the immune complexes. The results (Fig. 2) were as follows.

1. Approximately equivalent amounts of LSD1, CoREST, and HDAC1 were recovered in pellets containing cell debris that were collected after clarification of lysates from mock-infected or infected cells by centrifugation (Fig. 2, lanes 1 and 2).

2. The input fraction of lysate from infected cells harvested at 24 h after infection contained significantly less LSD1 than the input fraction of lysate from mock-infected cells (Fig. 2A). This result is consistent with the results shown in Fig. 1.

3. The antibody to LSD1 pulled down slightly reduced amounts of LSD1 and CoREST from lysates of infected cells, compared to the amounts pulled down from lysates of mock-infected cells (Fig. 2B, lanes 7 and 8).

4. The anti-LSD1 antibody pulled down significantly less HDAC1 from lysates of infected cells than from lysates of mock-infected cells (Fig. 2B, compare lanes 7 and 8).

5. The amounts of LSD1 and CoREST remaining in the supernatant fluids saved after removal of the immune complexes are shown in Fig. 2B, lanes 5 and 6. A more slowly migrated band of CoREST was observed in the post-IP supernatant of HSV-1(F)-infected lysate (lane 6), which was not present in the mock-infected lysate (lane 5). Equivalent amounts of HDAC1 were also recovered from supernatant fluids for mock-infected and infected cells.

We conclude as follows. (i) The total amount of LSD1 recovered at 24 h from infected-cell lysates was lower than that recovered from mock-infected cells. The results are consistent with partial degradation of LSD1. (ii) The anti-LSD1 antibody pulled down CoREST from infected-cell lysates, indicating that it remains bound to CoREST after infection. The levels of CoREST do not change during infection (6). And therefore, the fraction of CoREST not bound by LSD1 is contained in the post-IP supernatant. (iii) The results also indicate that the anti-LSD1 antibody did not pull down HDAC1 from infected-cell lysates. HDAC1 was present and could be detected in the supernatant fluid saved after the immune complexes were collected.

LSD1 is partially degraded in infected cells.

The results of the experiments described above suggested that LSD1 decreases with time after infection. To determine whether LSD1 is degraded in a proteasome-dependent manner, HeLa cells were mock infected or exposed to 5 PFU of HSV-1(F) or ΔICP0 mutant virus (R7910) per cell. The cells were either mock treated (Fig. 3, lanes 1 to 3) or pretreated with MG132 (5 μM) 1 h (lanes 4 to 6) before infection or 2 h (lanes 7 to 9) or 5 h (lanes 10 to 12) after infection. The cells were harvested 20 h after infection, solubilized, subjected to electrophoresis in a denaturing gel, and reacted with antibody to LSD1, CoREST, ICP27, or U_L38. The results (Fig. 3) were as follows.

FIG. 3.

LSD1 is partly degraded during HSV-1 replication. HeLa cells were mock infected or exposed to 5 PFU of HSV-1(F) or R7910 (ΔICP0) per cell. The cells were either pretreated with 5 μM MG132 at 1 h before infection (lanes 4 to 6) or treated with 5 μM MG132 at 2 h (lanes 7 to 9) or 5 h (lanes 10 to 12) after HSV-1 infection. The infected cells were incubated in medium with (lanes 4 to 12) or without (lanes 1 to 3) MG132 for 20 h before being harvested. Total cell lysates were electrophoretically separated, transferred to a polyvinylidene difluoride membrane, and immunoblotted with antibodies as indicated.

The amounts of LSD1 recovered from untreated, infected cells were smaller than those recovered from cells treated with MG132 added at any time. Furthermore, the degradation was lower in ΔICP0 mutant virus-infected cells than in cells infected with wild-type virus (Fig. 3, compare lanes 2 and 3). Wild-type and ICP0 mutant virus-infected cells accumulated equivalent amounts of ICP27. On the other hand, the late protein U_L38 accumulated in smaller amounts in ΔICP0 mutant virus-infected cells than in wild-type virus-infected cells.

The results suggest that the degradation of LSD1 occurs after 5 h of infection inasmuch as the amount of LSD1 recovered from cells exposed to MG132 at 5 h after infection was greater than that of untreated, infected cells.

Localization of LSD1, CoREST, HDAC, and Daxx in wild-type virus-infected cells.

Elsewhere, we reported that CoREST and HDAC1 are exported from nuclei of infected cells (6). In this series of experiments, we examined the localization of LSD1, CoREST, HDAC1, and Daxx, a marker of ND10 bodies, relative to that of ICP8. In this series of experiments, HEL cells grown in four-well slides were infected with 10 PFU of HSV-1(F) per cell. The cells were fixed at 4 h (Fig. 4) or 8 h (Fig. 5) after infection and reacted with antibodies to ICP8 and either LSD1, CoREST, HDAC1, or Daxx. In these experiments, ICP8 served to identify infected cells and also identify the stage in the evolution of the replication compartment inasmuch as ICP8 serves at an indicator of that compartment and has been extensively characterized in the literature (13, 17). It should also be noted that notwithstanding the relatively high multiplicities of infection, the evolutions of the replication compartment were relatively asynchronous in both cultures of cells infected for 4 h and those of cells infected for 8 h. The results (Fig. 4 and 5) were as follows.

FIG. 4.

Localization of CoREST/LSD1/HDAC components at 4 h after HSV-1 infection. HEL cells were mock or HSV-1(F) infected with 10 PFU per cell. At 4 h after infection, the cells were fixed, permeabilized, and reacted with anti-LSD1 plus anti-ICP8 (A, panels a to l), anti-CoREST plus anti-ICP8 (B, panels a to l), anti-HDAC1 plus anti-ICP8 (C, panels a to s), or anti-Daxx plus anti-ICP8 (D, panels a to r) antibodies. The cells were then stained with fluorescein isothiocyanate-conjugated anti-rabbit or Texas Red-conjugated anti-mouse secondary antibodies and mounted with mounting medium for confocal images. Solid arrows point to dense bodies containing ICP8 at early infection. Dashed arrows point to enlarged bodies of aggregated ICP8. Arrowheads point to cytoplasmic LSD1, CoREST, or HDAC1.

FIG. 5.

Localization of CoREST/LSD1/HDAC components at 8 h after HSV-1 infection. HEL cells were mock or HSV-1(F) infected at 10 PFU per cell for 8 h and reacted with anti-LSD1 plus anti-ICP8 (A, panels a to c), anti-CoREST plus anti-ICP8 (B, panels a to c), anti-HDAC1 plus anti-ICP8 (C, panels a to c), or anti-Daxx plus anti-ICP8 (D, panels a to c) antibodies. Arrows point to aggregated ICP8.

(i). In uninfected cells, CoREST, LSD1, and HDAC1 formed fine granular structures dispersed throughout the nucleus but not in the nucleoli. (Fig. 4A, panel c; B, panel c; and C, panel c).

(ii). The nuclei of cells infected for 4 h exhibited various stages of progression of infection, from very early accumulations of ICP8 to extensive nuclear remodeling. In cells in early stages of the evolution of the replication compartment (e.g., Fig. 4A, panel f; B, panel f; and C, panels f and g), ICP8 was dispersed throughout the nuclei. In addition, in some cells, small, dense bodies containing ICP8 were dispersed throughout the nucleus (e.g., Fig. 4A, panel e; B, panel f; and C, panel g). At this stage, there was no specific nuclear redistribution of LSD1, CoREST, or HDAC1.

(iii). At later stages of the evolution of the replication compartments (Fig. 4A, panel i; B, panel i; and C, panels m, r, and s), the dense structures containing ICP8 increased in size (e.g., Fig. 4A, panel i; B, panel i; and C, panel m). Concurrently, there was a redistribution of CoREST, LSD1, and HDAC1 in that these proteins formed dense structures that colocalized with ICP8. In addition, we noted the accumulation of CoREST and HDAC1 in the cytoplasm (Fig. 4B, panel i, and C, panels m, r, and s).

(iv). In uninfected cells, Daxx formed fine, dense bodies characteristic of ND10 nuclear structures (Fig. 4D, panels a and c). In addition, Daxx was distributed throughout the nucleus (Fig. 4D, panels a and c). The ND10 structures disappeared at relatively early stages of viral replication, before extensive accumulation of ICP8 (Fig. 4D, panels h and i). At later stages, we noted the formation of denser structures of Daxx that coincided with the dense structures containing ICP8 (Fig. 4D, panels o and r).

A representative sample of cells present in HEL cell cultures 8 h after infection is shown in Fig. 5. For the most part, the cells are more uniform in appearance. The small, dense bodies containing ICP8 (Fig. 5B, panels a and b, and D, panel b) were less frequent. Thus, ICP8-containing bodies tended to coalesce into much larger structures (e.g., Fig. 5A, panels a and b, and C, panel c, etc.). In all instances, intranuclear LSD1, CoREST, and HDAC1 formed dense structures that colocalized with ICP8 (Fig. 5A, panel a; B, panel c; and C, panel c). A common feature of infected cells was the presence of LSD1 protein in the cytoplasm. Thus, LSD1 was present in the cytoplasms of 17% of 104 ICP8-positive cells at 4 h after infection and in 86% of 131 ICP8-positive cells at 8 h after infection.

In stark contrast, Daxx colocalized with ICP8 in some cells (Fig. 5D, panel a). In most cells, however, ND10 bodies that colocalized with ICP8 were significantly diminished or disappeared altogether (Fig. 5D, panels b and c).

We conclude from these studies that at early stages of infection, ICP8 is dispersed in nuclei in the form of fine granules somewhat coarser than those formed by LSD1, CoREST, or HDAC1. Concurrently, with the aggregation of ICP8 into dense bodies presumed to be replication compartments, all three proteins colocalize with the dense bodies. In addition, CoREST and HDAC1 are exported relatively early in infection. Export of LSD1 to the cytoplasm was detected in large numbers of infected cells at 8 h postinfection, that is, after the translocation of CoREST and HDAC1. Daxx initially colocalized with ICP8. At later stages in infection, Daxx became dispersed and the amounts of Daxx colocalizing with the replication compartment were grossly diminished.

CoREST and ICP8 localize initially in the immediate vicinity of ND10 bodies.

ND10 structures are dispersed relatively rapidly in wild-type virus-infected cells. To study the possible interaction of the components of the CoREST/REST complex and ICP8 with ND10 structures, it was necessary to study cells productively infected with ΔICP0 mutant virus. In the studies described below, HEL cells grown in four-well slides were exposed to 20 PFU of R8501 (ΔICP0) mutant virus per cell. The cells were fixed and reacted with antibodies to CoREST, PML, or ICP8 at 8 h after infection. The time after infection selected for study was based on observations that the replication of ΔICP0 mutant virus is delayed and, in comparison with events in wild-type virus-infected cells, highly asynchronous. Fig. 6A and D show confocal images of uninfected cells. Fig. 6B, C, and E show images of infected cells. The results may be summarized as follows.

FIG. 6.

Localization of PML and CoREST in cells infected with ΔICP0 mutant virus. HEL cells were mock or R8501 infected at 20 PFU per cell for 8 h. The cells were fixed, permeabilized, and reacted with anti-CoREST plus anti-PML (A and B, panels a to c), anti-CoREST plus anti-ICP8 (C, panels d to j), or anti-PML plus anti-ICP8 (D, panels k and l, and E, panels m to q) antibodies. Arrows point to structures containing PML or ICP8 as indicated in Results.

In uninfected cells, CoREST was dispersed throughout the nucleus (Fig. 6A). Early in infection, CoREST formed a small punctate structure that localizes in the immediate vicinity of ND10. Fig. 6B, panel a, shows a punctuate CoREST body in the close proximity of an enlarged ND10 structure. As replication progressed, CoREST and ND10 bodies become enlarged, amorphous in shape, and partially overlapped (Fig. 6B, panels b and c).

At very early stages of infection, ICP8 was diffused throughout the nucleus (Fig. 6C, panel d). As replication progressed, ICP8 formed small punctuate structures that resembled ND10 structures (Fig. 6C, panel e). In most cells at early stages of infection, CoREST and ICP8 appeared to overlap, but the distributions of the two proteins were not equal (see enlargements of Fig. 6C, panels f and g), and more-detailed analyses indicate that ICP8 and PML do not overlap (Fig. 6E, panel m). Examination of these and other images suggests that initially ICP8 and CoREST are in close proximity but that with time ICP8 increases in amount and spreads beyond the mass of CoREST. This pattern is evident in Fig. 6C, panels h, i, and j.

In ΔICP0 mutant virus-infected cells, ND10 structures become highly enlarged and amorphous in shape (compare Fig. 6D, panels k and l, with E, panels o, p, and q). As the replication compartment evolved, there was little if any overlap between ND10 structures and ICP8 (Fig. 6E, panels o, p, and q).

DISCUSSION

The focus of this report is on the components of the complex containing HDAC1 or -2/LSD1/CoREST/REST and the interaction of this complex with HSV-1. In uninfected cells, this complex acts as a repressor. The term “engagement” in the title reflects the evidence that the interaction is antagonistic. This report focuses on two aspects of this interaction. Specifically, elsewhere we reported that ICP0 physically interacts with CoREST, a component of the repressor complex containing HDAC1/LSD1/CoREST/REST. The consequences of this interaction are that HDAC proteins are displaced from the repressor complex and that at a later time, both HDAC1 and the CoREST/REST complex are translocated to the cytoplasm. The translocations are the function of a late viral gene(s) and do not directly involve ICP0. Thus, CoREST is also translocated as a complex with HDAC1 and REST in cells infected at high numbers of ΔΙCP0 mutant virus PFU/cell (6). Additional evidence that the interaction of ICP0 with CoREST is a critical event in cells infected at low multiplicities of infection stems from experiments showing that a truncated form of CoREST that does not bind HDAC1 compensates for the absence of ICP0 in a cell-type-dependent manner (7). A key component of the CoREST/REST repressor system is LSD1. Here, we report that in infected cells, in contrast to what occurs in uninfected cells, the bulk of CoREST is not associated with LSD1, whereas the anti-LSD1 antibody pulls down CoREST; that LSD1 is partially degraded in the course of infection, in contrast to the stable levels of CoREST (6); and, finally, that LSD1 is translocated to the cytoplasm.

It is also of interest that the accumulation of LSD1 in cytoplasm was detected in a large fraction of cells fixed at 8 h after infection but that the presence of HDAC1 or CoREST was detected in a large fraction of cells fixed 4 h after infection. Our current data therefore suggest that the CoREST/REST complex devoid of bound LSD1 is translocated first and that the LSD1/CoREST/REST complex is translocated later in infection. The significance of this delay is unclear. The broader question is why only a fraction and not all of the components of the repressor complex are translocated to the cytoplasm, in contrast to the rather complete translocation of ICP0. The observation that the residual components of the repressor complex form aggregates that abut the initial aggregates of ICP8, signaling the formation of a replication compartment, raises the possibility that the HSV subverts components of the repressor complex for its own needs.

The second issue addressed in this report stems from the observation that in early stages of infection, ICP0 is localized in ND10 structures intermingled with other components of this structure (14). The general consensus is that DNA entering the nucleus localizes in the vicinity of ND10 but not internally in the ND10 structures (2, 10). The question posed in this report involves the localization of the members of the repressor complex vis-à-vis the ND10 bodies, the site of localization of ICP0. The issue is relevant for two reasons. First, the site of interaction of ICP0 with CoREST is near the C terminus of ICP0, distant from the RING domain responsible for the degradation of PML. Although the RING finger mutant contains the CoREST binding site, the replication of this mutant is retarded relative to that of the wild-type virus. Similarly, the mutant lacking the binding site for CoREST but containing an intact RING domain is partially defective in its ability to degrade PML (8). The data suggest that ICP0 expresses the two functions in tandem. The second reason stems from the observation that in cells transfected with DNA before infection, ICP0 is retained in the nucleus in a DNA concentration and cell-type-dependent manner but not in a DNA-type-dependent manner (11). Impairment in the degradation of PML, dispersal of ND10, and expression of post-α genes are associated with nuclear retention of ICP0. The fundamental conclusion of that report is that the transfected DNA competes with HSV DNA for “processing” by ICP0. Processing, in this instance, would involve interaction of ICP0 with proteins bound to the DNA. The question posed in this report involves the location of the HDAC1/LSD1/CoREST/REST complex during the early stages of infection.

The results presented in this report suggest that the components of the complex abut and are in close proximity to ND10 during the early stages of viral replication. The confocal microscopy studies suggest abutment but not intermingling of components, consistent with the hypothesis that the complex containing HDAC1 or -2/LSD1/CoRET/REST is associated with the compartment containing viral DNA rather than the compartment containing PML and ICP0.

The hypothesis that we would like to propose is that ICP0 dynamically interacts with both the components of ND10 and the DNA-protein complexes that abut the ND10 structure. ICP0 performs specific functions in both compartments. In the ND10 compartment, at least one of its functions is to degrade PML and SP100, resulting in the dispersal of ND10 components (1). In the abutting compartment containing viral DNA-protein complexes, ICP0 remodels the complex to enable acetylation of histones bound to viral DNA (6). At low multiplicities of infection, both functions must be performed in order for viral replication to reach completion. Neither function is required at high multiplicities of infection, suggesting that at least some of the viral DNA escapes silencing or that the remodeling of DNA-protein complexes at the ND10 structure can take place in the absence of ICP0. Furthermore, while ICP0 must perform both functions to enable productive infection at low numbers of PFU/cell, a truncated CoREST lacking the HDAC1 binding site and interacting solely with the LSD1/CoREST/REST complex compensates for the absence of ICP0 in a cell-type-dependent manner (7). While many details of the functions of ICP0 in ND10 structures are still missing, a paramount question involves the component of ND10 to which ICP0 is anchored. This component is unlikely to be PML, since ICP0 does not bind this protein and, furthermore, ICP0 lacking the RING domain and unable to degrade PML accumulates and remains resident in ND10 bodies (8). Current data suggest that in the DNA-protein compartment, ICP0 may be anchored to or dynamically interact with CoREST. The results are not consistent with the hypothesis that the two functions of ICP0, degradation of PML and suppression of silencing of viral DNA, are performed by different subpopulations of ICP0, since the two processes are totally interdependent (8).

Consistent with earlier reports (13, 17), ICP8 is initially dispersed throughout the nucleus. It then begins to accumulate in the structures abutting the ND10 bodies. This is clearly evident in cells infected with ΔICP0 mutant virus. In retrospect, this is not surprising. ICP8 does not bind to just any DNA, inasmuch as it does not associate with heterochromatin. Recent studies suggest that DNA is “processed” at the ND10 structure and that HSV-1 DNA competes with other DNAs for “processing” (11). One explanation of the accumulation of ICP8 at the ND10 structure is that this is the site of accumulation of DNA receptive to binding ICP8 and, in consequence, the site of initiation of formation of replication compartments. Early in infection, ICP8 forms several punctuate bodies. With time, these enlarge and coalesce to a point where they fill the nucleus.

On the basis of studies of cells productively infected with wild-type and ΔICP0 mutant viruses, we can now begin to explore the sequence of the early events in the infected cells. A proposed sequence of events is illustrated in Fig. 7 and detailed in the legend to the figure. The key feature of the model that we present is that all of the events leading to the initiation of a replication compartment take place in the juxtaposition of the ND10 structure or once PML is degraded adjacent to its erstwhile position. Components of the repressor complex remain in the vicinity of the ICP8 aggregate but are ultimately overwhelmed by the expansion of the replication compartment. Not shown in this model is that the events depicted in this schematic presentation can occur independently at several ND10 structures simultaneously. Ultimately, the replication compartments formed at each post-ND10 position coalesce.

FIG. 7.

Proposed sequence of events from entry of viral DNA into the nucleus to emergence of viral replication compartments. (Panel 1) DNA aggregates with ND10. At this time, the repressor complex (containing HDAC1 or -2/LSD1/CoREST/REST) is dispersed throughout the nucleus. (Panels 2 and 3) The α proteins are made and ICP0 colocalizes with PML in ND10 structures. Concurrently, aggregates of the repressor complex are formed and abut ND10 structures. (Panel 4) ICP0 begins to fill the nucleus and is also exported to the cytoplasm, as are components of the repressor complex (containing HDAC1 or -2/CoREST/REST or LSD1/CoREST). At this time, ICP8 begins to accumulate in the nucleus. (Panel 5) ICP8 aggregates at or adjacent to the residual repressor complex. (Panel 6) ICP8 complexes increase in size, coalesce with other complexes, and ultimately occupy a large fraction of the nucleus. The crosshairs mark the position of a ND10 structure. The sequence of events shown in panels 4 to 6 is based on studies of cells infected with the ΔICP0 mutant. In these cells, the PML is not degraded and serves as a position marker.