Purification of the Cellular C1 Factor Required for the Stable Recognition of the Oct-1 Homeodomain by the Herpes Simplex Virus a-Trans-induction Factor (VP16)”

The assembly of specific multiprotein complexes on the herpes simplex virus a/IE (immediate early) enhancer elements requires the interactions of the Oct-1 POU homeodomain, the viral aTIF (a-trans-induction factor) (VP16), and at least one additional cellular factor, the C1 factor. The C1 factor interacts directly with aTIF, likely forming an intermediate protein complex that recognizes the Oct-1 homeodomain-DNA complex. The biochemical purification of the mammalian C1 factor suggests that it is composed of multiple subunits of related, but heterogeneous, polypeptides. The interaction of a subset of these polypeptides with aTIF is stimulated by post-translational modifications of the C1 proteins, suggesting that this factor may be a critical target for the regulation of the herpes simplex virus a/IE transcription.

with aTIF, likely forming an intermediate protein complex that recognizes the Oct-1 homeodomain-DNA complex. The biochemical purification of the mammalian C1 factor suggests that it is composed of multiple subunits of related, but heterogeneous, polypeptides. The interaction of a subset of these polypeptides with aTIF is stimulated by post-translational modifications of the C1 proteins, suggesting that this factor may be a critical target for the regulation of the herpes simplex virus a/IE transcription.
As a class of DNA-binding factors, the homeodomain proteins represent significant regulatory determinants in eukaryotic development. These proteins are involved in the selective transcriptional activation of genes that determine cellspecific fate and promote cell type differentiation (1-4). However, the mechanisms by which distinct homeodomain proteins enact their selective biological functions remain unclear as members of this family contain conserved structures that bind to highly related DNA sequence elements (core motif: TAATNN (5-10)). Recently, analyses of the biology of homeodomain proteins have suggested that determinants within the homeodomain structure specify interactions with other regulatory factors, thus further defining the activity and DNA target specificity (11)(12)(13)(14)(15)(16).
The Oct-1 and Oct-2 proteins represent model examples of homeodomain proteins with multiple, distinct functional activities. These proteins are members of a subgroup of the homeodomain family (POU domain family) that contains a highly conserved motif (POU domain) consisting of the POUspecific subdomain and POU homeo-subdomain (17)(18)(19)(20). These two independent DNA-binding, subdomains are collec-3 TO whom correspondence should be addressed Center for Cancer tively responsible for the high affinity recognition of the octamer DNA element (ATGCAAAT (21-23)). Both Oct-1 and Oct-2 recognize this element with an equivalent DNA binding specificity and affinity. However, they have been implicated in the control of such diverse events as the transcriptional regulation of snRNA genes (24-31), the cell cycle regulation of the histone H1 and H2B genes (32-341, the expression of lymphoid-specific genes (35)(36)(37)(38)(39)(40)(41)(42)(43), and DNA replication (44)(45)(46). Thus, regulation by the octamer proteins exemplifies two questions of functional specificity in homeodomain biology: the discrimination between two proteins with highly related DNA-binding domains and the mechanism by which a given homeodomain protein can function in a number of distinctly regulated events. The regulation of herpes simplex virus (HSV)' a/IE gene expression by Oct-1, in conjunction with the potent viral transactivating protein aTIF (VP16, ICP25, VMW65), provides a biochemically accessible system for the study of the distinct role of homeodomain proteins. The lytic cycle of herpes simplex virus has been extensively investigated and is composed of a highly regulated cascade of gene expression (Ref. 47,reviewed in Ref. 48). The a/IE (immediate early) genes are expressed upon infection, and their expression is regulated by an HSV-encoded transactivating protein (aTIF, a-trans-induction factor (49)(50)(51)). This protein is packaged in the tegument structure of the virus, released into the cytoplasm upon infection, and transported to the nucleus where it induces the transcription of the five a/IE genes (52,53).
As depicted in Fig. 1, the HSV a/IE element is bipartite, consisting of a divergent octamer sequence (ATGCTAAT) and 3"flanking sequences that are highly conserved among the a/IE elements (55). In the C1 complex, the Oct-1 POUspecific domain recognizes ATGC and the POU homeodomain recognizes TAATGA (22, 23). The second DNA-binding protein, the viral aTIF, recognizes the conserved sequences that The abbreviations used are: HSV, herpes simplex virus; DTT, dithiothreitol; BSA, bovine serum albumin; PAGE, polyacrylamide gel electrophoresis; HPLC, high performance liquid chromatography; PC, phosphocellulose chromatography. extend 3' to the octamer site (22, 68). However, unlike Oct-1, the independent DNA binding affinity of aTIF is very low. Although the protein can cooperatively bind DNA via specific interactions with the Oct-1 homeodomain, this complex is not stable to gel electrophoresis (22, 68,69).
The formation of a stable C1 complex requires Oct-1, aTIF, and an additional cellular factor (the C1 factor; Fig. 1). The C1 factor does not appear to independently bind the a/IE element (22, 70), but rather, interacts with aTIF in the absence of the other components of the C1 complex (22, 71). Therefore, it is likely that the C1 factor interacts with aTIF to form a stable protein complex that subsequently recognizes the DNA-bound Oct-1 homeodomain. To analyze the nature of the C1 factor, we have biochemically purified this factor from mammalian cell extracts.

MATERIALS AND METHODS
Purification of the Mammalian C l Factor-The C1 factor was purified from nuclear extracts prepared from a 347-g frozen pellet of HeLa cells (a gift of R. Wobbe and S. Ludmerer of Merck Sharp and Dohme Corp.). Nuclear extracts were prepared as described (72), except that the extracted nuclei were pelleted at 20,000 X g for 30 min and subsequently reextracted with nuclear extraction buffer + 0.1% Nonidet P-40 for 30 min. The pooled extracts were clarified by centrifugation at 40,000 X g for 1 h in an SW 40 rotor prior to dialysis against buffer A (20 mM Hepes (pH 7.9), 0.5 mM EDTA, 20% (v/v) glycerol, 2.5 mM DTT, 1 mM phenylmethylsulfonyl fluoride) + 100 mM KCl. The protein concentration of the resulting nuclear extract was 8-12 mg ml-I.
The extract (1035 ml, 3620 mg of total protein) was loaded onto the phosphocellulose column at 250 ml h-'. The column was washed with buffer B + 50 mM KC1 at 500 ml h-', and the C1 factor was eluted with a 2 liters of linear gradient of 50-100 mM KC1 at 250 ml h-'. The eluate was monitored until it reached a conductivity equivalent to 85 mM KC1, at which time the column was step eluted with 1 liter of buffer B + 100 mM KC1 and 1.5 liters of buffer B + 750 mM KCl.
Preparation I1 consisted of four consecutive phosphocellulose column runs (PC-D, PC-E, PC-F, and PC-H).
1.5 liter of buffer C + 50 mM KCl. The phosphocellulose column fractions containing the peak of the C1 factor activity were diluted to buffer C + 50 mM KC1 (final of 160 mg in 1090 ml) and loaded onto the Q Sepharose FF column at 250 ml h-'. The column was washed with buffer C + 50 mM KC1 until the Amnm returned to the base line. The C1 factor activity was eluted with a 0.75-liter linear gradient of 50-250 mM KC1 in buffer C at 200 ml h-', followed by 0.3 liter of buffer C + 250 mM KC1 and 0.3 liter of buffer C + 750 mM KC1. Preparation I1 consisted of two consecutive Q Sepharose FF column runs (QFF-A and QFF-B).
A second Q Sepharose FF column (5 X 5 cm) was prepared as described above. The fractions containing the peak of the C1 factor activity were pooled, diluted to buffer C + 50 mM KC1 (final of 24.4 mg in 1300 ml) and loaded onto this column at 200 ml h-'. The activity was eluted by washing the column successively with 0.3 liter of buffer C + 200 mM KC1 and 0.3 liter of buffer C + 500 mM KC1 at 100 ml h".
The fractions containing the peak of the C1 factor activity were diluted to buffer C + 30 mM KC1 (final 19.8 mg in 210 ml) and applied to a Mono Q HR 10/10 column (Pharmacia LKB Biotechnology Inc.), equilibrated in the same buffer, at 0.8 ml rnin". The column was washed with 40 ml of buffer C + 30 mM KC1 and was developed with a 56-ml linear gradient of 30-125 mM KC1 in buffer C followed by 16 ml of buffer C + 125 mM KCl. The column was subsequently washed with 32 ml of buffer C + 250 mM KC1 and 24 ml of buffer C + 750 mM KCl. The C1 factor activity eluted from the Mono Q column in two defined peaks (pool I and pool 11).
Each pool was separately diluted to buffer C + 40 mM KC1 (2 mg in 32 ml, MSD column; and 1 mg in 10 ml, MSE column, respectively) and applied to a Mono S HR 5/5 column (Pharmacia LKB Biotechnology Inc.), equilibrated in buffer C + 40 mM KC1, at 0.2 ml rnin".
The columns were washed with 5 ml of buffer C + 40 mM KC1 and developed with an 8-ml linear gradient of 40-200 mM KC1 at 0.1 ml rnin". The columns were subsequently washed with 3 ml of buffer C + 200 mM KC1 and 3 ml of buffer C + 750 mM KCl.
In all cases, the KC1 concentrations of alternate chromatographic fractions were determined by plotting the measured conductivity against that of a linear titration of KC1 in the appropriate chromatographic buffer. The protein concentration of each chromatographic fraction was determined by Bio-Rad protein assays, using ?-globulin as a protein standard. The elution of the C1 factor, the determination of the C1 factor activity, and the calculation of the specific activities of the chromatographic fractions were done as described below.
Electrophoretic Mobility Gel Shift Assays-The production and purification of the Protein A-Oct-1 POU domain was as described aTIF protein from AcNPV-aTIF infected SF9 cells was as described The HSVaO DNA probe is equivalent to pRB608 (56). Unless otherwise stated, the reactions for the formation of the C1 complex were done as described (22) and included 0.8 ng (12 fmol) of HSVaO DNA probe, 200 ng of poly[dI-dC].poly[dI-dC], 20 mM Hepes (pH 7.9), 0.5 mM EDTA, 50-100 mM KC1, 2 mM DTT, 4% Ficoll 400, 500-800 pg ml" BSA, 30 ng of purified aTIF protein, 0.3 pl of bacterial lysate containing 10-20 ng of PA-Oct-1 POU protein, and the appropriate C1 factor chromatographic fraction in a total of 10 pl. The reactions were incubated at 30 "C for 20 min and were resolved in 4% nondenaturing polyacrylamide gels (acrylamide:bisacrylamide, 291) using 0.5 X Tris/glycine buffer (73,74). The gels were dried and exposed to Kodak X-Omat film or directly quantitated by using a Molecular Dynamics phosphorlmager with ImageQuant 3.0/3.1 software.
The C1 factor activity units were calculated by diluting the chromatographic fractions into a C1 complex formation assay such that 50% of the Oct-1-DNA complex was assembled into a C1 complex. The results of all of the column runs were standardized by the direct comparison and quantitation of the activity of each column load fraction. A standardization factor was calculated based upon the activity of the first phosphocellulose column run.
The quantity, in picomoles, of purified Cl factor was calculated by determining the amount of C1 factor that was required to shift 12 fmol of Oct-1-DNA complex into a C1 complex (1 saturation unit is a minimum of 12 fmol of C1 factor).
For the dephosphorylation of the C1 factor, 0.5 pl of HeLa cell nuclear extract (3-5 pg of total protein) or 40 saturation units of the purified C1 factor were incubated with 0.2 unit of potato acid phosphatase (Sigma) for 15 min at 25 "C. The reactions were then diluted into a C1 complex formation assay or were prepared for SDS-PAGE.
Gel Filtration Chromatography of Cl Factor-The production of a Protein A-aTIF protein affinity matrix and its use in the affinity selection of the HeLa cell C1 factor was as described (22). For the determination of the molecular mass of the C1 factor, a Superose 12 gel filtration column was equilibrated in buffer D (20 mM Hepes (pH 7.9), 0.5 mM EDTA, 2.5 mM DTT, 5-10% (v/v) glycerol, 200-500 mM KCI, 0.1% Nonidet P-40, and/or 0.05% Tween 20). The column was standardized by the fractionation of a mixture of molecular mass standards (dextran (void), ferritin (440 kDa), aldolase (158 kDa), bovine serum albumin (67 kDa), and chymotrypsin (25 kDa)) in buffer D at 0.25 ml min".
The peak C1 activity fraction from the aTIF affinity chromatography was diluted to buffer D (100-200 pl) and fractionated in an identical manner. The molecular mass of the C1 factor was then calculated on the basis of its retention volume in comparison to the linear plot of the retention volume uersus log molecular weight of the protein standards. The recovery of the C1 factor activity from several column fractionations was 6040% of the loaded activity.
SDS-PAGE-Extracts and chromatographic fractions, representing 20-50 saturation units of C1 activity, were resolved in denaturing SDS-polyacrylamide gels (16 cm2 X 0.75 mM, 7-12.5% (acrylamide:bisacrylamide, 300.8)) at 15 mA. Where appropriate, the gels were subsequently stained with silver and dried or transferred to nitrocellulose in a Bio-Rad transblot unit for 1.5 h at 125 V/0.6 A using a Tris/glycine transfer buffer (25 mM Tris/glycine, 20% methanol, 0.02% SDS). The quantity of the C1 factor polypeptides was estimated by SDS-PAGE of C1 chromatographic fractions in parallel with a titration of BSA. These gels were stained with silver, dried, and scanned with an LKB XL densitometer.
To renature the C1 factor, HeLa cell nuclear extract and purified C1 factor were precipitated with acetone, resuspended in buffered 6 M urea and renatured by dialysis to buffer E (20 mM Hepes (pH 7.9), 0.5 mM EDTA, 10% (v/v) glycerol, 5 mM DTT, 100 mM KC]) for 6-12 h. Alternatively, the polypeptides in the purified C1 factor preparations were resolved in a 9.5% SDS denaturing gel, eluted from the gel, precipitated with acetone, and renatured as described above (75). The resulting dialysates were tested for their ability to support the formation of a C1 complex based upon an anticipated 2-5% renaturation efficiency.
Tryptic Peptide Analysis of the Cl Factor Polypeptides-150 pmol of purified C1 factor activity were precipitated by the addition of 4 volumes of acetone, incubation at -20 "C for 16 h and centrifugation at 200,000 X g for 6 h in an SW 50.1 rotor. The protein pellet was resuspended in 40 pl of SDS-PAGE loading buffer and boiled for 5 min, and 100 pmol were loaded into a single 6-mm well of a 0.75-mm, 7% (acrylamide:bisacrylamide, 30:0.6) SDS denaturing gel. The resolved polypeptides were transferred to nitrocellulose in a Bio-Rad transblot unit at 125 V (0.6-A current limit) at 4 "C in an ice bath for 2.5 h using a Tris/glycine transfer buffer as described above. The nitrocellulose blot was stained with Ponceau S ,and protein bands representing the 68-, loo-, and 123-135-kDa polypeptides were excised, destained, and digested with trypsin as described (77)(78)(79). The tryptic peptides were eluted from the nitrocellulose and resolved in a narrow bore, reverse phase, C, column using an Applied Bioscience HPLC. The tryptic analysis was done by R. Cook in the MIT Center for Cancer Research Biopolymers Laboratory.

RESULTS
Purification of the Mammalian Cl Factor-Several preliminary observations influenced the development of the purification scheme for the mammalian C1 factor. The fractionation of protein extracts by phosphocellulose, DEAE-Sepharose, and S Sepharose chromatography revealed that the C1 activity did not fractionate homogeneously, but uniformly eluted from these columns within a 50-800 mM KC1 gradient (data not shown). This chromatographic behavior suggested that the C1 factor was heterogeneous or was associated with a variety of components in the nuclear extracts.
In several purifications, chromatographic buffers containing urea were used to dissociate protein complexes, thus allowing for the purification of the individual polypeptides (80)(81)(82). To test the hypothesis that the C1 factor might fractionate more homogeneously under similar conditions, aliquots of HeLa cell nuclear extract were incubated in the presence of 1-4 M urea at 4 "C for 30 min. These extracts were subsequently dialyzed to remove the urea or directly diluted into a C1 complex formation assay. In the presence of 1 or 2 M urea, the activity of the C1 factor was unaffected, as compared with untreated extracts. In contrast, the activity was increased 100% in the presence of 3 M urea and was decreased 30% in the presence of 4 M urea (data not shown).
Furthermore, upon fractionation of HeLa nuclear extract by phosphocellulose chromatography in the presence of 3 M urea, the C1 activity eluted from the column within a narrow KC1 concentration range (data not shown, refer to Figs. 2 and 3).
The partially purified C1 activity was stable in the presence of 3 M urea for extended periods of time at both 4 and -80 "C.
However, removal of the urea from these fractions by dialysis or buffer exchange chromatography resulted in a significant loss of the C1 activity and a reversion to the heterogeneous fractionation patterns (data not shown). Therefore, the entire purification of the C1 factor was done in the presence of 3 M urea.
Purification of the mammalian C1 factor was accomplished according to the scheme depicted in Fig. 2 and as briefly described below. HeLa cell nuclear extracts, in the presence of 3 M urea, were applied to a column of phosphocellulose. As assayed by the ability to assemble Oct-1 and aTIF into a stable C1 complex on the HSVaO a/IE response element, the peak of the C1 activity eluted a t 80 mM KC1 (Figs. 2 and 3, top) and represented a 14-fold purification of 30% of the total activity (Table I). These fractions were applied to a column of Q Sepharose FF (Q Sepharose FF-l), and the C1 activity was eluted from the column at 100-110 mM KC1 (Fig. 2), resulting in a 14-fold purification of 108% of the chromatographed activity (Table I). The partially purified C1 activity was rechromatographed in a second column of Q Sepharose FF (Q Sepharose FF-2; Fig. 2) . Elution of the C1 factor in a single 200 mM KC1 step resulted in a 1.7-fold purification of 171% of the applied activity (Table I). The significant increase in the C1 activity during this fractionation stage probably reflects the removal of inhibitors of the C1 complex assembly. The activity recovered from the Q Sepharose FF-2 chromatographic step was applied to a Mono Q column. In this stage, the C1 activity was recovered in two distinct elution

TABLE I
Calculation of the activities and yields of the chromatographic purification of the mammalian CI factor The activities and chromatographic yields of the C1 factor preparations were calculated from the data obtained from quantitative C1 complex formation assays as described under "Materials and Methods." The total activity (in standard units), total protein (mg), and specific activity are listed for the material that was loaded onto the indicated chromatographic matrix (Load) and for the resulting chromatographic fractions containing the elution peak of the C1 activity (Elution Peak). The purification fold and ?6 yield of the C1 activity is indicated for the individual (Stage) and the cumulative chromatographic stages (Cumulative). The first and second rows of the Mono Q data represent the calculations of the pool I and pool I1 C1 factors, respectively. The cumulative purification of the C1 factor polypeptide(s) is an underestimation due to the loss of C1 activity, but not total protein, during a technical problem encountered with the Mono Q HR 10/10 column. This loss has been incomorated into the cumulative Durification and yield of the Q SeDharose FF-2 stage. Q chromatography resulted in a 17-fold purification of 165% (pool I) and a 9-fold purification of 44% (pool 11) of the total applied C1 activity (Table I). The increase in the C1 activity during this fractionation (209% of the applied activity) was similar to that which was observed in the previous Q Sepharose FF-2 chromatographic step.
The pool I and I1 activities were each individually chro-  (Table I).
As summarized in Table I, the total chromatographic fractionation resulted in an -9000-fold cumulative purification of 24% of the C1 activity that was detected in the initial HeLa cell nuclear extracts. The cumulative yield is likely to be an overestimation of the actual yield of the C1 activity, since the association of the C1 factor with Oct-1 and (uTIF appears to be partially inhibited in the nuclear extracts and the initial chromatographic fractions. However, the cumulative purification of the C1 factor polypeptide(s) is likely to be an underestimation (2-%fold; refer to the legend to Table I). The described protocol (MSD and MSE preparations) resulted in the purification of -235 pmol of the C1 factor.
The Preparations of Purified Cl Actiuity Contain Multiple Polypeptides-The fractions containing the peak of the C1 Purification of the Cl Factor activity from the various stages of this purification were analyzed by SDS-PAGE. Fig. 4 (left and center) illustrates the polypeptides present in these fractions from the phosphocellulose, Q Sepharose FF-2, Mono Q (pool I), and Mono S (MSD) chromatography. The purified C1 factor activity is represented by MSD fractions 9-13, which contain a polypeptide of 100 kDa and a cluster of three proteins of 123-135 kDa. Less abundant polypeptides of 68-, 180-, and 230-kDa proteins are also evident. Fig. 4 (right) shows the resolved polypeptides of the purified C1 activity from two independent fractionations, where MSA represents the pool I activity of preparation I, MSB is the pool I1 of preparation I, and MSD is the pool I of preparation 11. Clearly, all of these fractions contain the 100-and 123-135-kDa polypeptides. In addition, the pool I1 C1 factor of both preparations contain significantly higher levels of the 68-and 180-kDa polypeptides, whereas the pool I fractions contain the 155-and 230-kDa proteins (Fig. 4, right, and data not shown).
The picomoles of the polypeptides in the C1 factor fractions were estimated from densitometric scans of silver-stained, SDS denaturing gels. The comparison of the picomoles of protein that were required to account for the amount of C1 complex activity indicated that the activity is mediated by one or more of the common, abundant polypeptides.
Requirements for the Formation of a C l Complex-The C1 factor is defined as the activity that is required for the assembly of Oct-1 and aTIF into a specific multiprotein complex on the a/IE response element (65). T o determine if the purified C1 polypeptide(s) retained the characteristics of the defined activity, MSD fraction 13 was added to a C1 complex formation reaction in the presence and absence of the other protein components. As illustrated in Fig. 5 , the addition of the Oct-1-POU protein to the DNA binding reaction containing the HSVaO DNA probe resulted in the formation of the Oct-1-DNA complex (lane 1 ). In contrast, the addition of aTIF, MSD-13, or both to the DNA binding reaction did not result in a detectable DNA-protein complex (lanes 2, 3, and 6). The addition of aTIF to the reaction containing the Oct-1 POU protein resulted in a slight retardation of the Oct-1-DNA complex, due to the formation of an electrophoretically unstable ternary complex (lane 4 ) . In contrast, the addition of the C1 factor to a similar reaction did not affect the migration of the Oct-1-DNA complex (lane 5). As expected, the addition of the purified factor to the reaction containing Oct-1 POU and aTIF generated the characteristic C1 complex (lane 7). Therefore, the characteristics of the purified C1 factor are consistent with the defined activity present in nuclear extracts.
The Molecular Mass of the C1 Factor-The molecular mass of the C1 factor was determined by gel filtration chromatography under several conditions, as well as with a number of independent C1 factor preparations. Fig. 6 illustrates the chromatographic sizing of the C1 activity that was partially purified by aTIF protein affinity chromatography. The activity was fractionated by Superose 12 gel filtration chromatography, and the chromatographic fractions were assayed for their ability to support the formation of a C1 complex in the presence of Oct-1 and aTIF. As shown in Fig. 6, the C1 factor eluted near the void volume of the column with a molecular mass that ranged from approximately 0.5-1.5 x lo6 Da. Furthermore, the heterogeneous chromatographic size of the C1 factor is visibly evident by the formation of a series of distinctly migrating C1 complexes in the gel shift assay (Fig. 6; compare fractions 26-30).
The Individual Polypeptides of the C l Factor Preparations Do Not Contain C1 Actiuity-In an attempt to determine if a particular polypeptide in the purified C1 factor preparations was solely responsible for the activity, nuclear extracts of HeLa cells and C1 factor chromatographic fractions were precipitated, resuspended in 6 M urea, and renatured by dialysis. Under these conditions, 5% of the precipitated C1 activity could be recovered (data not shown). However, when the polypeptides of the purified factor were resolved by SDS-PAGE, eluted from the gel, and treated in a similar manner, no single polypeptide was capable of assembling Oct-1 and aTIF into a C1 complex (data not shown). It is possible that the recovery of C1 activity in the precipitated extracts was due to the presence of stable protein aggregates that rapidly renatured. However, due to the inefficient renaturation of the C1 factor and the limited sensitivity of the assay, it was not possible to test the hypothesis that the C1 activity was dependent upon the renaturation of specific combinations of SDS-PAGE-resolved polypeptides.
High Affinity Binding of the C l Factor Polypeptides to aTIF-A significant characteristic of the C1 factor is its high affinity interaction with aTIF in the absence of the other C1 complex components (22,71). To identify the polypeptide(s) that directly associate with aTIF, the purified C1 factor preparations were resolved by SDS-PAGE, transferred to a nitrocellulose filter, and incubated with ["2P]aTIF protein. As illustrated in Fig. 7 (right panel), the aTIF probe specifically bound the 123-, 130-135-, and 230-kDa polypeptides in the active fraction from the initial phosphocellulose chromatography (PC). Similarly, in each of the purified C1 factor preparations (MSD, MSA, and MSB), aTIF bound the 123and 130-135-kDa polypeptides. The protein additionally associated with the 155-and the 180-kDa proteins that are uniquely represented in the pool I (MSD and MSA) and pool I1 (MSB) preparations, respectively. In contrast, the 100-kDa polypeptide, common to all of the C1 factor pools, and the 68-kDa protein of the pool I1 preparation did not bind to aTIF in this assay (compare Fig. 7 , left and right panels). The specificity of these interactions is evident from the lack of association of the labeled aTIF protein with the major polypeptides of the crude phosphocellulose chromatographic fraction (PC) and SDS-PAGE-resolved Escherichia coli lysates ( Fig. 7 and data not shown).

The Interaction of the C l Factor Polypeptides with aTIF Is
Modulated by Their Phosphorylation State-Treatment of the C1 activity, in nuclear extracts or in chromatographic fractions, with potato acid phosphatase generated a C1 activity that formed a more rapidly migrating C1 complex, suggesting that one or more of the polypeptides of this factor are phosphoproteins (Ref. 22 and data not shown). Therefore, the purified C1 factor preparation was similarly treated with phosphatase, and the reaction was resolved by SDS-PAGE. In comparison with untreated fractions, treatment of the MSD preparation resulted in an alteration in the migration of the 100-and 123-135-kDa polypeptides (Fig. 7,left panel,MSD and MSD*).
To determine if the phosphorylation status of these polypeptides might affect their ability to interact with aTIF, the MSD fraction was treated with potato acid phosphatase and incubated with the aTIF protein probe. As shown in Fig. 7 (right panel, MSD and MSD*), this treatment eliminated the interaction of the 130-135-and 155-kDa polypeptides and significantly (2.7-fold) reduced the interaction of the 123-kDa protein with aTIF, compared with the untreated factor.
Partial Tryptic Digestion of the C1 Factor Polypeptides-The interaction of aTIF with multiple polypeptides (123-180 kDa) in the purified C1 factor preparations suggested that several of the polypeptides might be variant forms of a single protein. This possibility was addressed by comparing the patterns of tryptic peptides derived from the partial digest of the 68-, loo-, and 123-135-kDa proteins (Fig. 8). The fractions containing the purified C1 factor were resolved by SDS-PAGE and transferred to nitrocellulose. Under these gel conditions, three major protein bands (68-, loo-, and the 123-135-kDa proteins (Fig. 4, MSD fraction 13)) were detected after staining the membrane with Ponceau S. The excised protein bands were incubated with trypsin, and the eluted peptides were resolved by HPLC reverse phase chromatography. 'the C1 Factor cation of a complex pattern of submolar peptides. Most surprisingly, the 100-and 123-135-kDa bands generated profiles that were nearly identical but distinct from that of the 68-kDa protein (Fig. 8 and data not shown). However, although they are nearly superimposable, there are two regions of the 100-and 123-135-kDa profiles that differ significantly (denoted A and B in Fig. 8). Finally, the 210 nm absorbance profile (Fig. 8) indicates that the 100-kDa and the group of 123-135-kDa proteins are present in a 1:l stoichiometric ratio in the purified C1 factor preparations.

DISCUSSION
Purification of the Cl Factor-The C1 factor is required for the stable assembly of aTIF and the Oct-1 homeodomain on the HSV a/IE element. Previous characterization of the C1 component had indicated that it was an evolutionarily conserved factor, being found in both insect and human cells (65). The factor did not independently bind the a/IE enhancer element but interacted directly with aTIF to form an aTIF-C1 factor protein complex. This protein complex is probably responsible for the stable recognition of the Oct-l-a/IE element complex.
The development of a scheme for the purification of the C1 factor from mammalian cell extracts required consideration of the nonhomogeneous chromatographic behavior of this protein(s). This observation suggested that the protein might be composed of multiple subunits and/or be associated with a heterogeneous group of proteins in the nuclear extracts. Therefore, the factor was chromatographically purified in the presence of 3 M urea, which resulted in a consistent and homogeneous fractionation. Approximately 235 pmol of C1 factor was thus purified -9000-fold from nuclear extracts of -1 X 10" HeLa cells, with a final yield of -25% of the initial C1 factor activity. Thus, the C1 factor is present a t approximately 5000 molecules/cell.
The purified C1 factor preparations contained a set of polypeptides of 100-and 123-135-kDa, with less abundant species of 68-and 155-kDa. In later stages of the purification, the factor eluted from the Mono Q column in two chromatographically distinct peaks of activity: pool I and pool 11. Both of these pools contained the 100-and 123-135-kDa polypeptides. In addition, the various C1 factor preparations contained only these polypeptides in a sufficient stoichiometric quantity to account for the C1 activity. Therefore, it is likely that the C1 factor is composed of one or more of these proteins.
The C1 Factor Is Likely to Be a Multiprotein Complex of Polypeptides of 100-and 123-135-kDa-Several experiments strongly suggest that the polypeptides present in the various C1 factor preparations represent the tightly associated components of a multiprotein C1 factor. First, the cofractionation of these proteins in the presence of 3 M urea is unusual. Second, gel filtration chromatography of the C1 factor under relatively stringent conditions indicates that the factor has a molecular mass of approximately 0.5-1.5 X IO6 Da. Third, although an interpretation of a negative result, it was not possible to recover C1 activity by renaturation of any single polypeptide in the purified C1 factor preparations. Finally, the tryptic peptide analysis indicated that the 100-kDa polypeptide is present a t a 1:l stoichiometric ratio with the set of 123-135-kDa proteins in the purified C1 factor preparations. Collectively, the results are consistent with the hypothesis that the C1 factor is composed of multiple subunits.
Characterization of the purified C1 factor further suggests that it is heterogeneous. As detected in the gel shift assay, the C1 complex band is actually a set of tightly migrating com-plexes that can be resolved upon extended electrophoresis. The distinct mobilities of these complexes is partially a result of the phosphorylation status of the C1 factor polypeptide(s), since treatment of these fractions with phosphatase reduced the observed heterogeneity (data not shown). However, this modification alone does not account for the heterogeneity. It is likely that the heterogeneity reflects the presence of either the 123-135-, 150-(pool I), and/or 180-kDa (pool 11) polypeptides in the C1 factor. All of these polypeptides specifically bound the labeled aTIF probe in the protein blot analysis, suggesting that they are related. In addition, the relatively simple tryptic peptide profile of the 123-135-kDa proteins suggests that this set consists of variants of a single protein.
Surprisingly, although the 100-and the 123-135-kDa polypeptides differ in their ability to independently interact with aTIF, the tryptic peptide analysis of these proteins indicate that they are highly related. However, these patterns do differ significantly in two regions (refer to Fig. 8). Furthermore, as noted above, the quantitation of the amounts of these peptides indicates that the 100-kDa and the group of 123-135-kDa proteins are present in an approximate 1:l stoichiometric ratio in the purified C1 factor preparations. These data suggest that the C1 factor consists of a complex of two related subunits: the lOO-kDa, in combination with the various forms of the 123-135-kDa polypeptide.
The Interaction of the Cl Factor Polypeptides with aTIF: Regulation by Posttranslational Modification-The most significant functional characteristic of the C1 factor is its high affinity interaction with aTIF in the absence of the other components of the C1 complex. This interaction could account for the role of the C1 factor in the C1 complex by stabilizing a conformation of aTIF that would bind with a higher affinity to the Oct-1-DNA complex. It is also possible that the C1 factor interacts directly with the Oct-1 protein and/or DNA in the context of the assembled complex. Additional cellular factors have also been identified that appear to associate with the C1 complex (C2 factor) but have not yet been purified or characterized (65). Thus, the C1 complex is probably part of a larger complex in the cell.
Analysis of the interaction of aTIF with the resolved polypeptides of the C1 factor preparations demonstrates that the 123-135-kDa polypeptides, but not the 100-and 68-kDa polypeptides, specifically bind to the protein probe. Thus, the 123-135-kDa species, present in all of the C1 factor preparations, is probably the critical polypeptide(s) that mediates the interaction with aTIF in the formation of the C1 complex. Significantly, phosphorylation of these polypeptides appears to be important for their independent association with aTIF, as treatment with potato acid phosphatase significantly reduced this interaction. Thus, the phosphorylation status of the C1 factor polypeptides may be a critical point in the regulation of the assembly of the C1 complex.
The regulation of these protein-protein interactions by post-translational modification is reminiscent of the regulated interactions of the retinoblastoma protein, as well as the formation of the p67SRF-p62TCF-~-fos promoter complex. In the case of the Rb polypeptide, the association of the protein with viral oncoproteins and E2F is regulated by cell-cycle dependent phosphorylation of the Rb protein (83; for reviews, see Refs. 84 and 85). Similarly, the formation of the ternary c-fos promoter complex is significantly enhanced by the phosphorylation of the ~6 2~'~ polypeptide, apparently by regulating its interaction with ~6 7 '~~ (86). Of particular significance, the high affinity interaction of the C1 factor with aTIF suggests that the formation of this protein complex is the initial assembly step, being followed by the recognition of the Oct-l-a/IE DNA element complex. Thus, the modulation of the affinity of this interaction may constitute the determining step in the formation of a C1 complex. Although alternative mechanisms may exist to regulate HSV a/IE gene expression (87), the regulated assembly of this complex may be significant for the activation of the lytic cycle of HSV gene expression.