Interaction of Cell and Virus Proteins with DNA Sequences Encompassing the Promoter/Regulatory and Leader Regions of the Herpes Simplex Virus Thymidine Kinase Gene*

During the course of a productive infection with herpes simplex virus (HSV), gene expression is coordinately regulated in a cascade fashion. Three major kinetic classes of genes, termed alpha, beta, and gamma, are sequentially activated. The mechanism responsible for repression and subsequent activation of beta and gamma genes is not known. A mobility-shift electrophoresis assay was used to examine DNA fragments containing the promoter/regulatory and the mRNA leader regions of the thymidine kinase gene (TK, a model beta gene) for their ability to bind proteins present in nuclear extracts prepared from uninfected and infected cells. Specific complexes unique to each extract were formed. Using a monoclonal antibody specific for ICP4 (the major regulatory protein of HSV) we demonstrated that this protein is present in the complexes formed between probes encompassing either the promoter/regulatory or leader sequence DNAs and proteins in infected-cell extracts. These complexes formed despite the lack of a high affinity binding site for ICP4 in either of these regions. The stability of complexes formed in infected-cell extracts with DNA probes containing the promoter/regulatory, leader region, and a high affinity ICP4-binding site were compared by dissociation analysis. The relative kd(obs) for these DNA-protein complexes was in the order: TK-leader region much greater than TK-promoter/regulatory region greater than or equal to high affinity ICP4-binding site. Cu+/1,10-phenanthroline footprinting revealed that infected-cell complexes which form on a probe containing a high affinity ICP4-binding site generate a protection pattern, whereas those formed on a probe containing the TK-leader sequence do not. In contrast, complexes formed with the latter probe in extracts from uninfected cells are kinetically stable and refractile to cleavage. A model for activation of the TK gene which incorporates these results is presented.


Interaction of Cell and Virus
classes of genes, termed (Y, fi, and y, are sequentially activated.
The mechanism responsible for repression and subsequent activation of fl and y genes is not known.
A mobility-shift electrophoresis assay was used to examine DNA fragments containing the promoter/regulatory and the mRNA leader regions of the thymidine kinase gene (TK, a model j3 gene) for their ability to bind proteins present in nuclear extracts prepared from uninfected and infected cells. Specific complexes unique to each extract were formed. Using a monoclonal antibody specific for ICP4 (the major regulatory protein of HSV) we demonstrated that this protein is present in the complexes formed between probes encompassing either the promoter/regulatory or leader sequence DNAs and proteins in infected-cell extracts.
These complexes formed despite the lack of a high affinity binding site for ICP4 in either of these regions.
The stability of complexes formed in infectedcell extracts with DNA probes containing the promoter/regulatory, leader region, and a high affinity ICP4-binding site were compared by dissociation analysis.
The relative kacobsJ for these DNA-protein complexes was in the order: TK-leader region >> TKpromoter/regulatory region 2 high affinity ICP4-binding site. Cu+/l,lO-phenanthroline footprinting revealed that infected-cell complexes which form on a probe containing a high affinity ICP4-binding site generate a protection pattern, whereas those formed on a probe containing the TK-leader sequence do not. In contrast, complexes formed with the latter probe in extracts from uninfected cells are kinetically stable and refractile to cleavage. A model for activation of the TK gene which incorporates these results is presented.
The linear, double-stranded genome of herpes simplex virus (HSV)' has the capacity to encode at least 75 genes (1, 2). Three temporally regulated gene families, termed cy, p, and y on the basis of their temporal order of synthesis, are expressed during the course of a lytic infection (3,4). Regulated expression of these genes results from an interactive control network comprised of cell-and virus-specified tram-acting factors and &-acting elements present in the virus chromosome (5-12). However, the relationships between these factors and the sequences with which they interact are not well understood.
The five members of the cy gene family are the first to be expressed (3,13). During the infectious cycle, transcription from all (Y genes except a0 (14,15) is turned off in response to the synthesis of LY gene products (11,(16)(17)(18)(19). One of these cy gene products, ICP4, is the major regulatory protein of the virus (17,20,21). /3 genes are transcriptionally silent until functional ICP4 is synthesized (15,22). Finally, after the initiation ofvirus DNA replication, the y genes are expressed (23). Their transcription also requires functional ICP4 and is regulated by the presence of ICP27, another cr gene product (24)(25)(26)(27)(28).
The promoter for the thymidine kinase (TK) gene, a prototype of the 6 kinetic class, has been extensively mutated and these mutants have been exhaustively studied. These analyses revealed that the promoter is composed of overlapping elements that affect both constitutive and regulated expression of this gene (29)(30)(31). Extensive transcriptional analyses of mutants with alterations in the sequences which comprise the promoter/regulatory and adjacent downstream region of the gene have shown a requirement for conservation of the nucleotides within this region for both basal level expression and transcriptional activation by virus-specified proteins (30)(31)(32)(33)(34).
We previously suggested that ICP4 may regulate transcription of HSV genes by multiple independent pathways (35). This model envisioned ICP4 interacting with sequences other than the defined high affinity binding site (36)(37)(38), or working via interactions with other cellular regulatory factors and/or virus-specified proteins. Recently, ICP4 was shown to be able to interact and bind sequences which lack or contain degenerate homologues of the high affinity binding site (36,(39)(40)(41). Despite several elegant studies, no specific target sequences in the TK promoter have been identified as ICP4-responder elements. Binding experiments, where sequences spanning the TK promoter/regulatory and the mRNA leader regions were incubated in crude nuclear extracts from infected cells, suggested that ICP4 is a component of a complex formed under these conditions (42). In this study, we demonstrate that sequences which compose both the promoter/regulatory (nucleotides -197 to -11) and leader regions (nucleotides -16 to +56) of the HSV-1 TK gene form specific complexes in extracts from both uninfected and infected cells, and that ICP4 is a component of the latter complexes. Analysis of the stability of complexes formed in extracts from infected cells reveals that those formed with the promoter/regulatory region are stable, whereas those formed with the leader region are kinetically extremely labile. In support of these results, we demonstrate that the complexes 9402 formed with the leader region of the TK gene generate a chemical nuclease protection pattern that is indistinguishable from that obtained with unbound DNA. These  Maxam-Gilbert sequencing ladder of the labeled DNA, were loaded immediately on a standard 15% polyacrylamide, 8.3 M urea sequencing gel in TBE buffer and electrophoresed at constant power (90 watts). Following electrophoresis, the gel was fixed in a 10% acetic acid, 10% methanol solution for 20 min, transferred to two pieces of Whatman 3MM paper, vacuum-dried, and exposed to Kodak XAR-5 film at -70 "C with an intensifying screen for 3-7 days.

RESULTS
Recognition of TK DNA by Cell and Infected-cell Factors-A DNA fragment containing the promoter/regulatory and leader sequences from the TK gene was previously shown to form a complex after incubation in extracts from infected cells (42). To investigate the sequence requirements for complex formation, the 5' 33 bp of the TK gene were cloned as three separate regions after dissection at conveniently located restriction endonuclease sites. These regions, termed TK A, B, and C, contain 181, 72, and 84 bp, respectively, and constitute the known promoter/regulatory domain, the mRNA leader, and the translation initiation site for the gene (Fig. 1). Each region was cloned into pIBI31, end-labeled at the Hind111 site, and then excised by digestion with either BamHI (TK A) or EcoRI (TK B and C) and tested for its ability to interact with proteins present in crude nuclear extracts prepared from either uninfected or infected cells.  TK A), MluI to BglII (-16 to +56, TK B), and BglII to MluI (+53 to +140, TK C) fragments into suitable sites within the polylinker region of the vector pIBI31. Each sequence was excised and used as probe in a mobility-shift electrophoresis assay. The TK B region (a) was further subdivided as indicated (c-f), and each of these fragments or methylated (g-i) TK B DNAs were employed as probes in DNAbinding reactions. The TK B LS+16/+36 fragment (b) has been described (31); the remainder of the fragments (c-f) were used as cleavage products derived from wild-type (W7') TK B DNA. The methylated TK B DNAs (g-i) were prepared as described in the text. The boxes on the top schematic locate sites within the TK promoter to which known cellular transcription factors can bind: OTF-1 (ATGCAAAT), Spl (GGGCGG), TFIID (ATA box), and C/EBP (CCAAT box). The right-angle arrow at position +1 denotes the start of transcription. The AUG box (+llO) denotes the position of the initiat,or AUG for thymidine kinase synthesis.
DNA-protein interactions were monitored by assaying for the presence of discrete complexes in native polyacrylamide gels. The results of these analyses are shown in Fig. 2. TK A and B probes each formed complexes after incubation with either extract, whereas the C fragment failed to form a detectable complex under the same conditions.
The TK A and B complexes which form in extracts from uninfected cells can be differentiated on the basis of their salt sensitivity. Thus, formation of the complexes with the A region probe occurred at both 5 (Fig. 2, lane c) and 50 (Fig. 2, lane b) mM NaCl, whereas formation of complexes with the B probe was sensitive to the lower salt concentration (Fig. 2, lanes f and g). By contrast, complex formation in infected-cell extracts with either fragment only occurred at 5 mM NaCl (Fig. 2, lanes d and h) (51). The middle band which formed with the TK B probe in extracts from uninfected cells at 5 mM NaCl (Fig. 2, laneg) was not reproducible and was deemed to be nonspecific. Complexes formed in extracts from infected cells migrated with a different mobility from those detected after incubation with extracts from uninfected cells. This experiment demonstrated that the promoter/regulatory region and DNA sequences from which the mRNA leader is transcribed are recognized and bound by both cell and infected-cell proteins. Moreover, there are distinct differences in the rate of migration of the complexes which form after incubation in the two extracts and the TK A and B sequences are differentially recognized by cell factors as determined by their salt requirements for binding. Although there was precedence for a role of leader sequences in regulation of HSV y genes (53-55), complex formation with DNA from this region of a /3 gene was unexpected.
Specificity of Complex Formation-Competition experiments were performed to ask if complex formation with either region resulted from sequence-specific binding. Only homologous competitor DNA was effective at inhibiting TK A complex formation in extracts from uninfected cells ( TK gene. TK A, B, and C regions were cleaved at the EcoRI site within the polylinker of the vector pIBI31, end-labeled, digested with Hind111 to release the virus insert, and 0.5-l ng of DNA was incubated with 5 pg of crude nuclear extracts prepared from uninfected (lanes b, c, f, g, j, and k) or 5-h-infected (lanes d, h, and 1) HeLa cells, as described under "Materials and Methods." Binding reactions contained either 50 mM NaCl (lanes b, f, and j) or 5 mM (contributed by the extract) NaCl (lanes c, d, g, h, k, 1). Lanes a, e, and i contain probe DNAs incubated with 5 pg of bovine serum albumin. The reaction mixtures were loaded onto native, 4% low ionic strength polyacrylamide gels, and analyzed for complex formation by the mobility-shift electrophoresis assay. The TK B sequence was studied in an identical manner. As before, homologous DNA was an effective competitor in extracts from uninfected cells (Fig. 3B, lanes b and c), whereas heterologous sequences, with the exception of the TK B LS+16/+36 and ICP4 BS fragments, did not compete for complex formation (Fig. 3B, lanes f, h, and i). When the B probe was analyzed in infected-cell extracts, the results were very similar to those found with the A probe (Fig. 3B, lanes k-t). That is, both homologous (TK B, TK B LS+16/+36) and heterologous (TK A, ICP4 BS) fragments competed, whereas TK C or the Spl-binding sites fragments did not. These experiments demonstrate that binding to the A and B probes is specific and reveal that there are distinct differences in recognition of these sequences by extracts prepared from uninfected or infected cells. These differences may represent the interaction of unique factors specific for each region. Moreover, the demonstration that A region DNA competes with the B probe and vice versa only in extracts from infected cells suggests that a novel, infected-cell specific factor is utilized. Competition by a high affinity ICP4-binding site in extracts from uninfected cells suggests that common factors recognize the TK B and ICP4 BS regions. Inhibition of A and B complex formation after incubation in infected-cell extracts by DNA containing a well characterized ICP4-binding site indicates that ICP4 participates in complex formation with both of these regions. ZCP4 is a Component of the Complexes-The observation (42) that ICP4 is present in a complex formed with a 253-bp fragment spanning the TK promoter and cap site (-197 to +56), and the ability of a sequence containing a high affinity ICP4-binding site to inhibit complex formation after incubation of either the A or B probe with extracts from infected cells prompted us to further define the regions required for ICP4 to participate in complex formation. Accordingly, the sequences within the TK gene were probed using a monoclonal antibody specific for ICP4 to tag ICP4containing complexes and the mobility-shift electrophoresis assay to separate complexes containing ICP4 from those that do not. This "supershift" assay was first used to explore complex formation with the A probe. Incubation of the A probe with uninfected-cell extracts in the presence or absence of antibody reveals no difference in the mobility of the complexes which form (Fig. 4A, lanes a and b). When this probe is incubated with extracts from infected cells complexes with altered mobilities form (Figs. 2,lane d,and 4A,lane c). Addition of antibody specific for ICP4 after these complexes are formed results in the loss of bands and the appearance of novel, slower-migrating species (Fig. 4A, lane d). The fidelity of this reaction was tested by probing with an antibody specific for gC (a virus glycoprotein) which has no effect on the rate of migration of these complexes (Fig. 4A, lane e).
We next asked if ICP4 participated in complex formation with the TK B DNA probe. Two B region probes were used in this experiment. The wild-type probe contains the sequences from -16 to +56, while the other was derived from the # gene linker-scanning mutant described by McKnight et al. (31). In the latter probe (TK B LS+16/+36, Fig. lb) 20 bp of TK DNA between +16 and +36 were deleted and replaced with a decanucleotide containing a BamHI restriction site. These two probes allowed us to ask if specific sequences within the TK B region were required to form complexes which A.

DNA-Protein
Interactions at the HSV TK Locus contained ICP4. Each probe formed specific complexes with identical mobility patterns in extracts from uninfected cells and, as expected, their mobility was unaltered after incubation with antibody to ICP4 (Fig. 4B, lanes a, b, f, and g). Both probes formed again complexes with indistinguishable migration profiles when incubated with extracts from infected cells, and several of these migrated with a reduced mobility after incubation with antibody specific for ICP4 (Fig. 4B, lanes c and d, h, and i), but not when reacted with the gC-specific antibody (Fig. 4B, lanes e and j). Thus, ICP4 is present in the infected-cell complexes formed when either the A or B region is used as probe. The data also reveal that the sequence between +16 and +36 in the B region is not required for complex formation with uninfected-cell extracts or for interaction of ICP4, which was previously suggested by the competing behavior of the TK B LS+16/ +36 DNA (Fig. 3B, lanes f and p). However, much less of the slowly-migrating species is detected after incubation of the TK B LS+16/+36 probe in either extract.

Sequence Requirements
for TK B Complex Formation-To determine whether discrete sequences within the B region were required for complex formation, probes encompassing nucleotides -3 to +56, -16 to +14, -16 to +38, and +19 to +56 (Fig. 1, c, d, e, and f) were prepared and analyzed for their ability to form complexes after incubation in the two extracts. Each probe gave rise to specific complexes after incubation in extracts from uninfected cells, although their abundance was low and variable and the slower-migrating species were difficult to detect (Fig. 5, lanes c, e, g, and i). The most abundant complexes were detected with the +19 to +56 probe (Fig. 5, lane i). After incubation in extracts from infected cells these probes, with the exception of that spanning the -16 to +14 sequence, formed novel migrating species that were more abundant than those formed in extracts from uninfected cells (Fig. 5, lanes d, f, h, andj). However, the yield was uniformly greater with those probes which contain the 3' half of the B sequence. These results suggest that complex formation occurs as a consequence of recognition of multiple sites scattered throughout the B region. To further probe these interactions, the cytosine residues at -6 in the noncoding and -7 in the coding strand and the adenines at +18 in the noncoding and +15 in the coding strand were methylated using Hue111 and PstI methylases, respectively. These DNAs were then extensively digested with the corresponding restriction enzyme, and the uncleaved fragments representing a homogenious population of fully methylated DNAs were isolated from nondenaturing polyacrylamide gels. Complex formation with each of the three methylated probes in uninfected-cell extracts was greatly decreased (Fig. 5, lanes k, m, and o), whereas it was only marginally reduced when infected-cell extracts were used for the binding reactions (Fig. 5, lanes 1, n, and p).
From these two experiments we infer that nucleotides throughout the B sequence are required to generate abundant specific complexes with extracts from uninfected cells. Methylation of even 2 bases at divergent sites results in decreased complex formation. Although the 5' region (-16 to +14) did not form abundant complexes in either extract, it seems that its presence promotes complex formation about the entire TK B region (perhaps by providing stabilizing nonspecific contacts) because, perturbation of the sequence by the addition of methyl residues at either -6 and -7 (both of which protrude in the major groove of the double helix) decreases complex formation in extracts from uninfected cells. In marked distinction to these results we find that the methylated DNAs barely affected complex formation in extracts from infected Labeled TK B region cleavage products were incubated with extracts prepared from uninfected or 5-h-infected HeLa cells as described under "Materials and Methods," and reaction mixtures were analyzed for complex formation by the mobility-shift electrophoresis assay. Lanes a, c, e, g, i, 12, m, and o contain complexes formed in extracts from uninfected cells, whereas lanes b, d, /, h, j, 1, n, and p contain complexes formed in extracts from infected cells. The complexes were formed with the following probes; Lanes: a and b contain wild-type TK B DNA (Fig. la); c and d contain the -3/+56 sequence (Fig. lc); e and f contain the -16/+14 sequence (Fig. Id); g and h contain the -16/+38 sequence (Fig. le); i and j contain the +19/+56 sequence (Fig. l/J; k and 1 contain the TK B LS+16/+36 probe methylated at the HaeIII site (Fig. lg); m and n contain the wild-type TK B probe methylated at the HaeIII site (Fig. lh), and o and p contain the wild-type TK B probe methylated at the PstI site (Fig. li).
cells. Thus, the interactions which occur in infected-cell extracts are distinct from those that take place in uninfectedcell extracts and, the presence of infected-cell proteins alters the sequence requirements for complex formation.
To further verify that the sequence requirements for complex formation in the two extracts were different, competition experiments with DNA fragments encompassing various parts of the B region were performed (Fig. 6). Complex formation in extracts from uninfected cells, with the wild-type B probe, was more efficiently competed by DNA encompassing the 3' portion of this sequence (+19 to +56) than by its 5' DNA (-16 to +14) counterpart (Fig. 6, lanes a, b, and c). However, this was not true when the TK B LS+16/+36 probe was tested (Fig. 6, lanes d, e, and f). Although binding was weak to begin with, each competitor inhibited complex formation. Because the yield of complex with the 5' sequence is exceedingly low, we can only say that the rapidly migrating complex is not competed by the 3' sequence (Fig. 6, lanes g and h). By comparison, complex formation with the 3' sequence is readily competed by the 5' sequence (Fig. 6, lanes i and j). These data demonstrate that both the 5' and 3' regions of the TK B sequence are recognized by some shared cellular factors, whose binding to the corresponding sequences leads to cooperative interactions that enhance TK B complex formation resulting in its increased abundance (Fig. 6, lane a).
The competition profile in extracts from infected cells were analyzed for complex formation by the mobility-shift electrophoresis assay, and dried gels were exposed to x-ray film. Only the regions of the gels containing complexes are shown. The following probes were used; wild-type TK B region (lanes a, b, c, k, 1, and n), LS+16/+36 TK B region (lanes d, e, f, n, o, and p), -16/+14 DNA (lanes g, h, y, and r), and +19/+56 DNA (lanes i, j, s, and t).
revealed that the 3' portion of the B region (+19 to +56) inhibited binding of both the wild-type and the TK B LS+16/ +36 probes (Fig. 6, lanes h and p), whereas the 5' sequence (-16 to +14) failed to inhibit complex formation (Fig. 6, lanes 1 and 0). Furthermore, complex formation with the 5' sequence was efficiently competed by DNA containing the 3' region ( Fig. 6, lanes q and r). In contrast, complex formation with the 3' sequence was refractile to inhibition by the 5' portion of the B region (Fig. 6, lanes s and t). These data support a hypothesis in which common and disparate proteins recognize the same sequences within the B region and virus gene expression modifies interactions between the B region and cellular proteins. They also emphasize the importance of the 3' sequence for complex formation in extracts from infected cells.
Sequences Required for Participation of ICP4 in TK B Complex Formation-The supershift assay revealed that ICP4 is present in complexes formed after incubation of the TK B probe with extracts from infected cells. Complexes formed with the subregion probes described in Fig. 1 were examined for the presence of ICP4 by this assay. This analysis demonstrated that infected-cell complexes formed with the -3 to +56, -16 to +38, and +19 to +56 probes all contained ICP4 (Fig. 7, lanes d, n, and s). Although the slower migrating of the low abundance complexes formed after incubation with the -16 to +14 probe appears to be disrupted after incubation with the antibody, no novel migrating species was detected (Fig. 7, lanes h and i). Both the -16 to +38 and +19 to +56 probes efficiently form ICP4-containing complexes (Fig. 7,  lanes n and s). Control experiments with nonspecific antibody demonstrated the fidelity of this assay (Fig. 7, lanes e, o, and t). As expected, incubation of the complexes formed in uninfected-cell extracts with the antibody to ICP4 resulted in no change in their mobility (Fig. 7, lanes a, b, k, 1, p, q). These results, together with the observation that ICP4 is present in the TK B LS+16/+36 complex (Fig. 4B, lane i) suggest that ICP4 interactions occur at the 3' end of the B region and that there are at least two distinct domains within this region that can promulgate these interactions. Moreover, there appears to be redundancy built into the B sequence and only a portion of it is required for ICP4 to be present in the complexes.

Complex Formation with a Sequence Containing a High Affinity ICP4-binding
Site-ICP4 binds to various regions of the HSV genome with different affinities (36,(38)(39)(40)56). A high affinity binding site overlapping the cap site of the a4 gene inhibits complex formation with the B region probe in extracts from both uninfected and infected cells (Fig. 3B lanes h and r). Furthermore, our results indicate that common host factors recognize and form complexes with the ICP4binding sequence (ICP4 BS) and the TK B region. The ICP4 BS sequence inhibited TK A complex formation only in extracts from infected cells (Fig. 3A, lanes h and i versus r and s). Therefore, we asked if complexes could be formed using the ICP4 BS probe in both extracts and, if either TK A or B DNAs inhibited formation of these putative complexes. Therefore, a high affinity ICP4-binding site probe (ICP4 BS) was incubated with extracts from uninfected and infected cells under conditions that elicited the A and B complexes. Complexes which differed in their mobility and abundance were formed in each extract (Fig. 8). Multiple species were observed in extracts from uninfected cells, whereas only a single abundant complex was present in extracts from infected cells. Moreover, complexes formed in each extract at both 5 and 50 mM NaCl (Fig. 8, lanes a, b, k, and 1) distinguishing them from those formed with the TK B probe (51). Competition experiments revealed that these interactions were specific (Fig. 8, lanes i and s), and demonstrated that the TK B region was an efficient competitor in either extract (Fig. 8, lanes g, h, q, and r). As predicted, TK A DNA failed to inhibit complex formation in extracts from uninfected cells (Fig. 8, lanes e and fl, but did so in infected-cell extracts (Fig. 8, lanes  o and p). Furthermore, incubation of the complex formed in extracts from infected cells with the anti-ICP4 monoclonal antibody revealed the presence of ICP4 in the complex (Fig.   8, lane t).
These results complement those obtained with the TK A and B regions and demonstrate that common cellular proteins recognize and interact with both the ICP4 BS and the TK B DNAs, despite any apparent homology between these sequences. In addition, complex formation with these two regions is differentially affected by changes in ionic strength. Moreover, it is clear that some of the host factors bound to the TK A region are distinct from those recognizing ICP4 BS and TK B region sequences. In all instances ICP4 was identified as a complex participant in extracts from infected cells. A 49-bp AuaI/BamHI fragment from the 04 gene (schematic) was end-labeled at the BarnHI site and incubated in reaction mixtures containing extracts from uninfected (lanes a-j) or 5-h-infected (lanes k-t) HeLa cells, in the presence or absence of various unlabeled competitor DNAs. The binding reactions contained either 5 mM (contributed by the extract) NaCl (lanes a, k, and m-t), or 50 tnM NaCl (lanes 6-j and 1). The competitor DNAs and their molar ratios relative to probe were: lanes c, n and d, n, ICP4 BS at 5-and lo-fold molar excess, respectively; lanes e, o and f, p, TK A region DNA at lo-and 20.fold molar excess, respectively; lanes g, q and h, r, TK B region DNA at lo-and 20-fold molar excess, respectively; and lanes i and s, each contain a 50-fold molar excess of the TK C region DNA. Samples in lanes i and t are reactions incubated in the presence of monoclonal antibody specific for ICP4 after complexes were allowed to form. The reaction mixtures were analyzed for complex formation by the mobility-shift electrophoresis assay, and dried gels were exposed to x-ray film. The region of the gel containing the free DNA is not shown.
amined their relative dissociation rates. For each of the three probes (TK A, TK B, and ICP4 BS), complexes were allowed to form for 30 min (51) and then chased with a 250-fold molar excess of the ICP4 BS fragment for various periods of time before being analyzed by the mobility-shift electrophoresis assay. The high affinity ICP4 BS DNA was used as a sink to efficiently deplete any free or dissociated ICP4. Cell-and virus-specific complexes formed with the A region and ICP4 BS probes were almost equally stable throughout the chase period (Fig. 9, A and C). B-region complexes formed in extracts from uninfected cells began to dissociate between 5 and 10 min after the chase, but were still detectable after 20 min (Fig. 9B). However, only a small percentage of the B-region complexes formed in extracts from infected cells persisted after a l-min chase and they were no longer detected after 2 min. Thus, the TK A and ICP4 BS probes both formed kinetically stable complexes when incubated in either uninfected-or infected-cell extracts. In contrast, the B region complexes formed in extracts from uninfected cells were only stable relative to those formed in infected-cell extracts. Although competition analyses suggest that some components of the complexes which form about both the B region and ICP4 BS probes are shared (Figs. 3B, lanes h and i, and 8, lanes g and h), the difference in the relative dissociation rates of the complexes formed in uninfected-cell extracts (Fig. 9B, lanes d and e, and SC, lanes d and e) revealed that these complexes were distinguishable. Differences in &b,) suggest that even though each probe forms complexes which contain Complexes were allowed to form under standard reaction conditions (30 min) in extracts prepared from uninfected (lanes a-e) or 5-h-infected (lanes f-j) HeLa cells with either the TK A, TK B, or ICP4 BS probes (A, B, and C, respectively). At the end of the incubation period, a 250fold molar excess of unlabeled ICP4 BS DNA was added to each reaction and samples were withdrawn at 1 (lanes b and g), 5 (lanes c and h), 10 (lanes d and i), and 20 (lanes e andj) min after the chase, and applied directly to a running 4% low ionic strength polyacrylamide gel.
plexes that contain ICP4, and that a probe containing a high affinity ICP4-binding site formed complexes with extracts from uninfected and infected cells. Therefore, to identify the precise nucleotide sequences within the 3' B-region (nucleotides +19 to +56) and ICP4 BS probes that were involved in complex formation, we probed (in situ) the complexes using the nuclease activity of Cu'/l,lO-phenanthroline (52) and analyzed them as described under "Materials and Methods." The electrophoretic pattern of the cleaved DNA isolated from the most abundant complex formed with 3' TK B region in extracts from uninfected cells, demonstrated that the sequence from +30 to +56 was protected from nuclease activity (Fig. lOA, lanes b and c). It was not possible to determine if sequences between +I9 and +29 were also protected because they were equally refractile to cleavage in the free DNA fraction (Fig. lOA, lane b). However, the cleavage pattern of this DNA derived from a complex containing ICP4 revealed no differences between it and the corresponding free DNA (Fig. lOA, lanes d and e). This result is consistent with the extremely rapid dissociation of the complexes formed in extracts from infected cells (Fig. 9B, lanes f and g). On the other hand, the gel pattern of the cleaved DNA, present in the complex formed with the ICP4 BS probe in extracts from infected cells, showed that a 15-bp sequence was protected from cleavage (Fig. lOB, lane c) as described previously (37,56). Although the DNA in the most abundant complex formed with this probe in extracts from uninfected cells was preferentially protected starting at the same site, the footprint extended to the 3' end of the sequence (Fig. lOB, lanes f and   g). To demonstrate cleavage specificity, the nuclease digestion profile of DNA isolated from a rapidly migrating nonspecific complex was examined and shown to be identical to the pattern of free DNA (Fig. lOB, lane d).
This chemical nuclease cleavage analysis further differentiates the interactions which occur when the 3' end of the TK B region probe is incubated in uninfected-or infected-cell extracts. Specifically, these results would be expected given the kinetically labile nature of the complexes formed in infected-cell extracts and suggest that host factors stably interact with this sequence in the absence of virus-specified proteins. In contrast, the ICP4 BS complexes which form in Complexes were formed with the TK B region (A) or ICP4 BS (B) DNAs and nuclear extracts prepared from uninfected or S-h-infected HeLa cells in standard binding reactions that were scaled up lo-fold. Reaction products were then fractionated by a preparative mobility-shift electrophoresis assay.  (21,35,57,58). To identify regions within the TK gene that interact with cell-and virus-specified proteins, we prepared nuclear extracts from uninfected and 5h-infected HeLa cells and, using a mobility-shift electrophoresis assay and footprinting with a chemical nuclease, examined their ability to form complexes with the DNA sequences that compose the promoter/regulatory and leader regions of the gene, in comparison with the binding properties of a DNA fragment which overlaps the cap site for the (~4 gene and contains a well characterized high affinity binding site for ICP4.
In this study, we report that all of these regions form specific complexes in extracts from uninfected cells which differ in their sensitivity to ionic strength and kinetic stabil-ity. These interactions are differentially altered after expression of at least one virus-specified protein, ICP4, as determined by both the above criteria as well as chemical nuclease cleavage analysis.
To date, the association of ICP4 with a high affinity binding site or a number of different DNA sequences that deviate from it has been demonstrated in five HSV genes encoding gD (16,38,39,59), ICPs 0 (19, 36, 42, 60), 4 (37, 41, 42, 56), 25 and 42 (40). Previous attempts to map an ICP4-binding site within the TK promoter did not localize the interaction. Our present analyses demonstrate that complexes which form with the TK sequences in either extract are specific and in each instance their mobility is altered by the presence of at least ICP4. The complexes formed with the A and B region probes in extracts from uninfected cells were competed by low concentrations of homologous DNAs. Furthermore, the B but not the A probe was inhibited from forming specific complexes in both extracts by the ICP4 BS. The complexes which formed with the A and B probes in extracts from infected cells were each competed by the other DNA and by the fragment containing the ICP4-binding site. Most of the complexes formed in extracts from infected cells with each of the TK probes contained ICP4, as determined by supershift experiments. Thus, the adjacent A and B regions appear to interact with both common and unique factors. The complexes which they form in the two extracts are differentiated by their physical and biochemical properties and are distinct from one another. The TK A region includes sequences recognized by well characterized cellular transcription factors (61-63) and gives rise to specific complexes in binding reactions with extracts from HSV-l-infected cells. Some of these complexes may represent the interaction of ICP4 and/or other LY gene products with the cellular transcription factors that occupy this region. In addition, CCAAT, ATA, and perhaps Spl binding activities might be modified after infection, leading to altered mobilities of complexes that either contribute to trans-induction per se or increase promoter strength, perhaps by stabilizing or making more accessible the transcription machinery. There are now several examples from other systems where trans-induction appears to be mediated via modulation of known binding activities. Abmayer et al. (64) have reported that an immediate-early protein from another herpes virus, pseudorabies, facilitates the interaction of the ATA box binding factor TFIID with promoters. Wu et al. (65) and Simon et al. (66) have implicated ATA box recognition in the induction of the adenovirus Elb and the hsp70 genes by adenovirus Ela.
The efficient competition by A region DNA for the TK B probe only in extracts from infected cells, implies that the two DNA fragments bind a common infected-cell specific factor(s). The binding protein(s) might contain independent DNA-binding domains, each capable of interacting with a different sequence element or, a single domain of the protein(s) could be capable of recognizing divergent DNA sequences as demonstrated for the octamer-binding proteins (67). This may be accomplished by recognition of a common DNA structure or conformation, as proposed for the yeast transcription activator, HAPl, which interacts with different sequence elements upstream of the CYCl and CYC7 genes (68). Alternatively, diversity may be generated by the participation of other cell-and/or virus-specified non-DNA-binding proteins in the complex. The biologic consequences of these multiple interactions are at present unknown, but they might potentiate the induction or repression of TK gene expression.
The TK B region was more thoroughly analyzed. The binding properties of subsets of sequences that encompassed this region were examined. Our results demonstrate that this region is bound by cell factors and that common cellular factors recognize this sequence and the ICP4 BS. The appearance of multiple species in the mobility-shift analysis predicts that, despite its small size, the TK B region is probably bound by more than one cell factor. The 5' end of this region (-16 to +14) does not efficiently form complexes in either extract, while probes containing all or portions of the 3' half (+19 to +56) do. Moreover, it is the complexes formed with the 3' end of the B region with which ICP4 strongly interacts. The TK B LS+16/+36, -16/+38, and +19/ +56 probes all form complexes that contain ICP4. Therefore, we posit that ICP4 can interact with complexes formed at either of two or both sets of sequences between +19 to +38 and +38 to +56 within the TK B region. The dissociation analysis and competition experiments with the high affinity ICP4-binding site indicate that this interaction is mediated through protein-protein and not DNA protein contacts (see also Ref. 51).
The kinetic experiments with extracts from infected cells and the TK B region probe suggest a biochemical basis for our inability to generate a chemical nuclease cleavage pattern that differed between free and bound DNAs. The very short lifetimes of these complexes (52 min) predict that bound proteins can rapidly detach from the protected DNA fragment, allowing the chemical nuclease to gain access and cleave it. Thus, "stable" virus-specific complexes should have the same cleavage pattern as the corresponding free DNA. It is the exclusion volume of the gel which increases the local concentration of the reactants shifting the equilibrium toward reassociation that permits these kinetically labile complexes to be detected as stable species by the mobility-shift electrophoresis assay (69,70).
The experiments with the ICP4 BS were designed to monitor the fidelity of ICP4 binding. We demonstrated that this sequence was hound by extracts from uninfected cells, and that the complexes which formed with this probe utilized the same, or a subset of core proteins used to form complexes with the TK B-region probe. These sequences originate from genes which are expressed at two different times postinfection. An efficient mechanism for differential regulation of the 01 and /3 kinetic classes of HSV genes might utilize the same or a subset of core truns-acting cell factors which are subsequently altered by virus-specified or induced proteins, thus assuring temporal regulation of the various kinetic classes. A potential advantage to using common factors and multiple sites to control transcription is that small changes in the concentration of transcription factors can be amplified, leading to large differential effects on RNA synthesis. The regulatory effects could be positive or negative in nature. Thus, fluctuations in the concentrations of ICP4 and other a gene products could have major effects on factor and/or RNA polymerase II binding to viral genes and the effectiveness of the DNAs as templates for RNA synthesis. Recent studies showed that the virus-associated transactivator a-TIF (71-73) and the Ela protein from adenovirus (74) recognize DNAbound proteins. How they are targeted to the appropriate sequences and activate transcription is still unclear. If ICP4 uses this alternative mechanism of activation then it alone, or in conjunction with other (Y gene products, could differentially affect the arrangement of cell factors along the corresponding promoter/regulatory domains of a virus gene.
The complexity of the interactions we observed suggest that sequences throughout the promoter/regulatory and mRNA leader regions of the TK gene interact with an array of polypeptides. Therefore, recognition results from the summation of multiple independent protein-DNA contacts rather than solely from essential contacts within a core binding site. Accordingly, modification to or removal of any sequence, except a core element that is obligatorily required for transcription is unlikely to completely inhibit activation or repression of the TK gene.
To accommodate the experimental data we propose that virus genes are regulated as a consequence of a series of discrete interactions which initially are between c&acting regulatory elements and truns-acting cell factors. This interaction results in activation or repression of transcription depending on the kinetic class of the gene. Thus, (Y genes are transcriptionally active at early times postinfection and their level of activity is enhanced by the interaction of (u-TIF with OTF-1 (72, 75). Subsequently, when sufficient ICP4 is synthesized, it can bind to high affinity sites which initially are occupied by cell factors. Substitution of cell factors by ICP4 can now result in repression of transcription (19). Regulation of p gene expression, which is of necessity different, is readily accommodated by the model. Early after infection there is no detectable expression from the TK gene (14,15,22,30). We envision that at these very early times the promoter/regula-   Fig. 11A. Following synthesis of virusspecified (Y gene products, the interaction of cell proteins with the DNA template and/or each other is altered, resulting in the formation of complexes that migrate with novel mobilities. ICP4 is a component of these complexes. Synthesis of virusspecified proteins also changes the competition profile, so that now A region DNA competes with a B region probe and vice versa. This may reflect the presence of a newly synthesized protein or, one which is modified after infection resulting in recognition of sequences in each region (infected-cell specific factor). A consequence of this alteration is that infectedcell proteins have a lower affinity for their target sequence and now readily dissociate to allow RNA polymerase II to pass through the leader region and transcribe the gene (Fig.  11B). In support of this model we note that B region complexes which form in infected-cell extracts dissociate much more readily than their counterparts which form in uninfected-cell extracts (Fig. 9B). Moreover, Arsenakis and Roizman (76) have described a post-a-function which altered the ability of a host protein to bind virus DNA. They suggested this resulted from modification by virus gene products, In conclusion, our data suggest that ICP4, the major regulatory protein of herpes simplex virus, interacts with cell proteins that cover the defined promoter of the virus TK gene and a region adjacent to and downstream of it. The biological effect of this interaction is unknown. We have constructed a model wherein cellular factors interacting with the A and B promoter-leader sequences maintain or repress basal level