Characterization of a Trimeric Complex Containing Oct-1, SNAPc, and DNA*

The human small nuclear (sn) RNA promoters contain a proximal sequence element (PSE), which recruits the basal transcription factor SNAPc, and a distal sequence element characterized by an octamer sequence, which recruits the POU domain transcription factor Oct-1. The Oct-1 POU domain and SNAPc bind cooperatively to probes containing a PSE and an octamer sequence, and this effect contributes to efficient transcription in vitro. In vivo, however, Oct-1 regions outside of the POU domain can activate snRNA gene transcription. Here, we have examined whether the role of these regions is to contribute to cooperative binding with SNAPc. We find that they indeed improve cooperative binding, but most of the effect is nevertheless mediated by just the POU domain. This suggests that Oct-1 activates transcription of snRNA genes in at least two steps, recruitment of SNAPc mediated primarily by the POU domain, and a later step mediated by regions outside of the POU domain. We also show that a PSE-binding complex observed in nuclear extracts consists of Oct-1 and SNAPc. Although Oct-1 cannot bind effectively to the PSE probe on its own, in the complex it contacts DNA. Thus, in a nuclear extract, SNAPc can recruit Oct-1 to a probe to which Oct-1 cannot bind on its own.

Transcriptional activators are key regulators of RNA polymerase II transcription, but their mode of action is still poorly understood. Activators often consist of a DNA-binding domain, whose role is to target the activator to the correct promoter, and of activation domains, whose role is to enhance transcription (1). The activation domains may help recruit members of the basal transcription machinery to the promoter, enhance transcription elongation, or perhaps trigger modifications of the basal machinery that result in enhanced transcription initiation (2-6) (see Refs. 7 and 8, for reviews).
The RNA polymerase II and III snRNA 1 gene promoters both contain an essential proximal sequence element (PSE), which recruits the basal transcription factor SNAP c (also called PTF) (9 -11), and a distal sequence element, which serves as a transcriptional enhancer and is characterized by the presence of an octamer sequence. The octamer constitutes a binding site for both the Oct-1 and Oct-2 POU domain transcription factors, but the distal sequence element is thought to recruit Oct-1. Indeed, like snRNA genes, Oct-1 is broadly expressed, whereas Oct-2 is a B cell-specific factor (see Ref. 12, for a review). Moreover, in vivo, the Oct-1 and Oct-2 activation domains display promoter specific activities; the Oct-1 activation domains preferentially activate snRNA promoters, whereas the Oct-2 activation domains preferentially activate transcription from mRNA promoters (13,14). This differential activation results from differences in the mRNA and snRNA basal promoter elements, suggesting that the Oct-1 and Oct-2 activation domains interact differentially with promoter-specific basal transcription factors (13).
Both the Oct-1 and Oct-2 POU domains bind cooperatively with SNAP c /PTF to a probe containing a PSE and an octamer sequence, and at least in the case of the Oct-1 POU domain, this cooperative binding promotes increased levels of transcription in vitro (9,15). The observation that in vivo, Oct-1 regions outside of the POU domain activate snRNA gene transcription, and do so much more efficiently than Oct-2 regions outside of the POU domain, suggests that the POU domain is not sufficient for transcription activation in vivo (16). How do Oct-1 regions outside of the POU domain contribute, then, to transcription activation?
Here we have tested whether Oct-1 regions outside of the POU DNA-binding domain play any role in cooperative binding with SNAP c to probes containing a PSE and an octamer sequence. We find that they do contribute to cooperative binding but most of the effect is mediated by the POU domain, suggesting that the Oct-1 activation domains play their primary role at a later step in the activation process. We also show that in crude nuclear extracts, a complex consisting of Oct-1 and SNAP c forms on a probe containing a PSE-binding site but lacking an octamer site. Formation of the complex is dependent on the ability of Oct-1 to bind DNA, and indeed Oct-1 contacts DNA in the complex. Thus, in the very complex mixture of proteins that constitutes a nuclear extract, SNAP c can recruit Oct-1 to a probe to which Oct-1 cannot bind on its own.

Constructs
Constructs for PCR Probes-The plasmids containing the H2B octamer site and human U6 PSE were previously described (15). The plasmid containing the H2B octamer site was described previously (17). The plasmids AD (also referred to in the text as mouse U6 PSE probe) and the mouse U6 PSE probe with the ABC mutation were generated by annealing two oligos, filling in with the Klenow fragment of DNA polymerase, cutting with BamHI and HindIII, and inserting into pUC118. This resulted in plasmids containing inserts with the sequences GGATCCGAAACTCACCCTAACTGTAAAGTAATTGTGTTTCT-TGGCTTCTCGAGCCTTGTGGAAGCTTAAG and GGATCCGAAACT-CCCACTACCGGTCCAGTAATTGTGTTTCTTGGCTTCTCGAGCCTT--GTGGAAGCTTAAG for the AD plasmid and the plasmid containing the mouse U6 PSE with the ABC mutation, respectively. The N7 plasmid has been previously described (18). Probes were generated by PCR amplification of these constructs using the universal sequencing primer end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase and the re-verse sequencing primer. The probes AD-mutHD and AD-short were generated by PCR with the plasmid AD as a template and primers with the sequences TCACACAGGAAACAGCTATGACCATGACCACGAAT-TCG and AGCTCGGTACCCGGGGATCC, respectively, substituted for the reverse sequencing primer. All the probes were generated with the same radiolabeled primer and had, therefore, the same specific activity.

Sources of Proteins
SNAP c -The SNAP c used in these experiments was derived from a Mono Q peak fraction, which corresponds to the fourth step in the purification of SNAP c and is purified approximately 2,500-fold (10). The total protein concentration in the fraction is approximately 0.3 mg/ml.
Expression and Purification of Oct Proteins in Escherichia coli-All proteins were expressed in E. coli BL21 (DE3) cells using the T7 expression system (21) as described previously (20). The Oct-1 POU, Oct-1 POU R49A, and Pit-1 POU domains were expressed as GST fusion proteins and were purified with glutathione-agarose beads (Sigma). In some cases the GST moiety was removed by cleavage with thrombin and dialyzed against buffer D (20 mM HEPES, pH 7.9, 100 mM KCl, 0.5 mM EDTA, 20% glycerol, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride). Proteins were analyzed by SDS-polyacrylamide gel electrophoresis and visualized by Coomassie Blue staining. In all cases the proteins appeared to be greater than 90% pure. Protein concentrations were measured by the Bio-Rad protein assay (Bio-Rad).
Histidine-tagged proteins were produced by growing 1-liter cultures of E. coli BL21 (DE3) cells expressing histidine-tagged Oct-1 (H.Oct-1) or histidine tagged Oct-1.P.1 (H.Oct-1.P.1) as described above. The cells were lysed by sonication in OctQ buffer (20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 0.1% Tween 20, 5% glycerol, 5 mM 2-mercaptoethanol, 0.5 mM phenylmethylsulfonyl fluoride, 1 g/ml pepstatin A, 1 mM benzamidine, 2 g/ml aprotonin, 1 g/ml leupeptin). Cell lysate was centrifuged at 40,000 ϫ g for 30 min. Supernatant was collected and passed over a Mono Q 10/10 column (Pharmacia). The flow-through fractions were kept and dialyzed against OctQ buffer containing 1 M NaCl. Protein was applied to a 1.5-ml Ni-NTA column (Qiagen) and eluted with a gradient from 0 to 40 mM imidazole. Fractions containing octamer binding activity were then dialyzed against OctQ buffer containing 100 mM NaCl and applied to a Mono S 5/5 column (Pharmacia) and eluted with a salt gradient from 0.1 to 1 M NaCl. Fractions containing octamer binding activity were pooled and dialyzed against buffer D. Protein purity and concentration were assessed by the same method as with the POU domains above except that the protein gels were stained with silver. For both Oct-1 and Oct-1.P.1, a number of truncated proteins were visible below the full-length products. Oct-2 and the Oct-2 POU domain were a generous gift of Dr. Masafumi Tanaka (Cold Spring Harbor Laboratory).
Nuclear Extracts-HeLa cell nuclear extracts were prepared as described (22).

EMSA
The binding reactions were performed in a total volume of 20 l containing final concentrations of 100 mM KCl, 20 mM HEPES, pH 7.9, 5 mM MgCl 2 , 0.2 mM EDTA, 10% glycerol, 20 g of fetal calf serum as a protein carrier, 2 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 0.4 g each of poly(dI-dC) and pUC118. The amounts of SNAP c and Oct proteins added are indicated in the figure legends. The reactions were incubated at room temperature for 20 min before addition of 25,000 cpm (50 -100 pg) of radiolabeled DNA probe followed by a 30-min incubation at room temperature. The reactions were electrophoresed through 5% nondenaturing polyacrylamide gels (acrylamide/bisacrylamide ratio, 39:1) in 1 ϫ TGE running buffer (50 mM Tris base, 380 mM glycine, 2 mM EDTA) at 150 V for 4.5 h at room temperature. The gels were dried and autoradiographed. The intensities of the signals were measured with a Fuji BAS1000 PhosphorImager.

OP-Cu Footprinting
EMSA reactions containing 16 l of SNAP c , O.4 g of His-Oct-1, and 100,000 cpm of DNA probe in a total volume of 80 l were performed as described above. In-gel footprinting reactions were performed essentially as described (23,24) except that the reactions were performed at 4°C. The gel was washed with 600 ml of 50 mM Tris, pH 8.0. The wash solution was removed and the gel was immersed in 400 ml of 50 mM Tris, pH 8.0, 40 ml of solution A (2 mM 1,10-phenanthroline, 0.45 mM CuSO 4 ), and 40 ml of solution B (58 mM 3-mercaptopropionic acid). The cleavage reaction was allowed to proceed for 13 min and stopped by the addition of 40 ml of 28 mM 2,9-dimethyl-1,10-phenanthroline. The gel was incubated for an additional 15 min at 4°C and 10 min at room temperature. The gel was subject to autoradiography and the bands corresponding to specific protein-DNA complexes, as well as free probe, were excised. Gel slices were eluted overnight in 0.4 ml of 0.1% SDS, 5 mM EDTA, 20 mM Tris, pH 8.0. Eluted DNA was extracted with phenol: chloroform (1:1), precipitated with ethanol, and analyzed on a 6% polyacrylamide (19:1), 8 M urea, 0.5 ϫ TBE sequencing gel. The gel was dried and autoradiographed.

Regions Outside of the Oct-1 and Oct-2 POU Domains Contribute Weakly to the Recruitment of SNAP c to the PSE in an
EMSA-The Oct-1 and Oct-2 POU domains can recruit the SNAP complex to the PSE in vitro (9,15). To determine (i) whether regions outside of the POU domain contribute to such recruitment of SNAP c to the PSE, and (ii) whether the Oct-1 regions outside of the POU domain are more active in this process than the Oct-2 regions outside of the POU domain, we compared the abilities of Oct-1, the Oct-1 POU domain (POU-1), Oct-2, and the Oct-2 POU domain (POU-2) to recruit SNAP c to a PSE in an EMSA. We used a DNA probe containing the H2B octamer sequence, a high affinity binding site for Oct-1 and Oct-2, and the human U6 PSE, a low affinity binding site for SNAP c . With this combination of sites, Oct proteins can substantially increase the binding of SNAP c to the low affinity PSE, and the effect is therefore easily visualized (15). As shown in Fig. 1A, when SNAP c was added to the probe alone, a weak protein-DNA complex was formed (lane 2). However, when SNAP c and increasing amounts of full-length Oct-1 (lanes 3-11), Oct-1 POU (lanes [12][13][14][15][16][17][18][19][20], full-length Oct-2 (lanes 21-29), or Oct-2 POU (lanes 30 -38) were added together to the probe, a slower migrating complex was formed, whose intensity increased with increasing amounts of Oct proteins to levels much higher than those observed in the absence of Oct proteins. The slower migration indicated that SNAP c and the respective Oct proteins occupy the same probe, and the enhanced binding shows that the Oct proteins recruit SNAP c to the PSE, as observed previously for the Oct-1 and Oct-2 POU domains (15).
To quantitate the enhancement of SNAP c binding by the different Oct proteins, we first determined the amounts of DNA binding activity in the different Oct protein preparations by performing an EMSA identical to that shown in Fig. 1A except that SNAP c was omitted from the binding reactions. The results, shown in Fig. 1B, were quantitated with a PhosphorImager. When shorter than full-length products were present, as in the case of Oct-1 (lanes 2-10), they were included in the quantitation as detailed in the figure legend. The resulting values were used to normalize the amounts of SNAP c -DNA complexes formed in the presence of the different Oct proteins in Fig. 1A. The resulting data are graphed in Fig. 1C and show that at equal amounts of DNA binding activity, Oct-1 recruited SNAP c about 2-fold better than Oct-2 and 4-fold better than the Oct-1 and Oct-2 POU domains. In the case of Oct-1, this value may be an underestimate, since a fraction of the Oct-1 protein was not full-length and may have, therefore, been missing a domain required for efficient SNAP c recruitment. Nevertheless, most of the effect (about a 10-fold enhancement of SNAP c binding) was directed by the Oct-1 or Oct-2 POU domain.
To eliminate the possibility that the Oct-1 POU domain was masking the contributions of regions outside of the POU domain to recruitment of SNAP c to the PSE, we also tested a FIG. 1.Oct-1 is more active than Oct-1 POU in recruiting SNAP c to the PSE. A, EMSA with a probe containing the H2B octamer motif and a human U6 PSE in the absence of proteins (lane 1) or with 5 l of a SNAP c fraction (Mono Q fraction (10)) alone (lane 2) or 5 l of the SNAP c chimeric Oct-1 protein referred to as 1.P.1 in which the Oct-1 POU domain was swapped with the POU domain of the pituitary transcription factor Pit-1 (20). The Pit-1 POU domain, which is 50% identical to the Oct-1 POU domain, binds to the H2B octamer motif (20) but does not efficiently recruit SNAP c to the PSE (15). Any effect of Oct-1 regions outside of the POU domain on recruitment of SNAP c to the PSE should, therefore, be easily visualized with the 1.P.1 protein. The effects of 1.P.1. and the Pit-1 POU domain on SNAP c binding to the PSE are shown in Fig. 2A. The amounts of octamer binding activity in the different protein preparations in the absence of SNAP c were determined as described above for the Oct-1 and Oct-2 proteins (data not shown), and the quantitated data are shown in Fig. 2B. SNAP c on its own bound poorly to the probe (Fig. 2A,  lane 2), but SNAP c together with increasing amounts of Oct-1 bound up to 40-fold more efficiently to the PSE (Fig. 2, A, lanes  2-11, and B). When Oct-1 was replaced by 1.P.1, a greatly reduced but clearly detectable enhancement of SNAP c binding was observed (3.5-fold at a 1.P.1 concentration equivalent to the highest concentration of Oct-1: see Fig. 2, A, lanes 12-20,  and B). This enhancement was larger than that observed with just the Pit-1 POU domain (a maximum of less than 2-fold, see Fig. 2, A, lanes 21-29, and B). As with the Oct-1 protein in Fig.  1, the enhancement of SNAP c binding by the 1.P.1 protein may be larger than that shown in Fig. 2B, because a significant fraction of the 1.P.1 protein was not full-length and may have been missing a domain important for efficient SNAP c recruitment. Together, these results indicate that although the Oct-1 POU domain is primarily responsible for the recruitment of SNAP c to the PSE, Oct-1 regions outside of the POU domain also contribute to the effect in this assay.
A PSE-binding Complex Present in Crude Nuclear Extracts Contains SNAP c and Oct-1-To search for any putative PSEbinding factors other than SNAP c , we performed EMSAs with a probe carrying the mouse U6 PSE alone, a high affinity binding site for SNAP c , and nuclear extracts. As a control, we used a probe carrying six point mutations in the PSE that debilitate U6 snRNA transcription both in vivo and in vitro (Ref. 18, and data not shown). As shown in Fig. 3, lane 5, we observed, in addition to the SNAP c ⅐DNA complex, a second complex of slower mobility that did not form with the mutant probe (lane 12). The migration of both complexes was retarded after incubation with polyclonal antibodies directed against the SNAP43 (10) or the SNAP45 (25) subunits of SNAP c (lanes 6 and 7) but not after incubation with preimmune antibodies (lane 10), suggesting that both complexes contain SNAP c . In addition, the migration of the upper complex was retarded after incubation with two different anti-Oct-1 monoclonal antibodies (26) (lanes 8 and 9), but not with an irrelevant monoclonal antibody (12CA5) directed against an epitope in the influenza hemagglutinin protein (27) (lane 11). This suggested that the upper complex consists of SNAP c and Oct-1 bound to the probe, even though the probe lacks an octamer sequence and is not bound by Oct-1 alone (lane 2). Indeed, a complex of nearly identical mobility was obtained when purified SNAP c and re-combinant Oct-1 protein were incubated with the probe (lane 4). Thus, incubation of a probe containing a high affinity PSE but no independent Oct-1-binding site with a crude nuclear extract, where the relative concentrations of different factors have not been manipulated, results in the formation of a complex containing both SNAP c and Oct-1. This suggests that this complex is physiologically relevant, and that SNAP c can recruit Oct-1 to a probe devoid of octamer site.
The Complex Containing SNAP c and Oct-1 Does Not Form with a DNA-binding Defective Mutant of Oct-1-The SNAP c ⅐Oct-1 complex can assemble on a probe to which Oct-1 alone cannot bind. We therefore asked whether SNAP c and Oct-1 coexist as a complex in the absence of DNA. Attempts to co-immunoprecipitate Oct-1 with an anti-SNAP c antibody and vice versa were unsuccessful, suggesting that this is not the case (data not shown). In another approach, we asked whether the complex could form with an Oct-1 POU domain containing a single alanine substituted for an arginine at position 49 of the POU domain (R49A (12)). This Oct-1 POU domain mutant does not bind to an octamer site efficiently but can assemble into a VP16-induced complex with VP16 and HCF (12). As shown in Fig. 4, addition of either a histidine-tagged Oct-1 protein or a GST-Oct-1 POU domain fusion protein retards the migration of the SNAP c ⅐PSE complex (lanes 3 and 4), suggesting that both proteins can form a complex with SNAP c on a PSE probe. Indeed, these complexes can be supershifted with anti-Oct-1 antibodies (lanes 9 and 10). In contrast, addition of a GST-Oct-1 R49A mutant protein does not retard the complex (lanes 5 and 6) nor render it reactive to anti-Oct-1 antibodies (lanes 11 and 12). Thus, an Oct-1 mutant defective for binding DNA cannot assemble with SNAP c on the PSE probe.
Oct-1 Contacts DNA When Complexed with SNAP c on the PSE Probe-The observation that a mutant Oct-1 defective for binding DNA could not assemble with SNAP c on the PSE probe raises the possibility that Oct-1 interacts with DNA in the complex. To examine this possibility directly, we performed orthophenanthroline-Cu (OP-Cu) footprinting (23, 24) on protein-DNA complexes fractionated by EMSA. We used three probes, designated AD, N7, and H2B Octa. As shown in Fig. 5A, the AD and N7 probes both contain the mouse U6 PSE, but the flanking sequences are different. The H2B Octa probe contains the H2B octamer motif, a high affinity site for Oct-1. The three probes were incubated with SNAP c , Oct-1, or both, and the binding reactions were fractionated by EMSA. The gel was then exposed to OP-Cu and the DNA fragments corresponding to free probe, or to probe complexed with SNAP c , or to probe complexed with SNAP c and Oct-1 were eluted from the gel and fractionated on a sequencing gel. The results are shown in Fig.  5B. DNA from the Oct-1/H2B Octa probe complex displayed a footprint over the octamer motif, as expected (lane 8). DNA from the SNAP c /AD probe and SNAP c /N7 probe complexes showed a clear footprint on the PSE (lanes 2 and 5). Significantly, on both the AD and N7 probes, DNA from the complexes containing SNAP c and Oct-1 displayed, in addition to the footprint on the PSE, a footprint higher up in the gel (lanes  Fig. 5A, the distance between this additional footprint and the PSE footprint varies on the two probes, but in each case the same sequence is protected. The footprint corresponds to an ATT sequence within a region, ATGATTACGAA, with limited similarity to an octamer motif in either orientation (see Fig. 5C). These results suggest that Oct-1 contacts DNA in the complex, and that this point of contact is remarkably flexible relative to the location of the PSE.

6). As shown in
Oct-1 Recognizes a Specific DNA Sequence in the SNAP c •Oct-1 Complex with DNA-The footprinting results above suggested that although Oct-1 can contact DNA at various distances from the PSE in the SNAP c ⅐Oct-1⅐DNA complex, the sequences recognized by Oct-1 are specific. To confirm this possibility, we tested the ability of Oct-1 to supershift a SNAP c ⅐PSE complex formed on a probe containing a mutation within the ATT sequence contacted by Oct-1 (AD-mutHD probe, see Fig. 5A), or a truncated probe missing this sequence altogether (AD-Short probe, see Fig. 5A). As shown in Fig. 5D, Oct-1 could "supershift" the SNAP c ⅐DNA complex on the AD and N7 probes, as before, but very weak complexes, or no retarded complexes, were observed with the probe mutated within the ATT sequence (lanes [11][12][13][14][15], or the truncated probe (lanes 16 -20), respectively. These results show that Oct-1 requires sequence-specific interactions with the DNA to form the SNAP c ⅐Oct-1⅐DNA complex.

DISCUSSION
The snRNA promoters contain an enhancer, the distal sequence element, which is nearly always characterized by the presence of an octamer sequence as well as, in some cases, an Sp1-binding site (28). This contrasts with the enhancers of mRNA promoters, which differ from one promoter to the next and can consist of a wide variety of protein-binding sites. Consistent with the uniformity of the distal sequence element, basal transcription from snRNA promoters is activated by Oct-1 activation domains and a glutamine-rich activation domain derived from Sp1 (13), but not, in general, by activation domains derived from other activators. For example, the VP16 or Oct-2 activation domains do not activate snRNA gene transcription (13,14,16,29). This selectivity suggests that the Oct-1 activation domains exert their action on a snRNA promoter-specific transcription factor, such as SNAP c .
The Oct-1 DNA-binding domain (POU domain) has been shown before to bind cooperatively with SNAP c to a probe containing a PSE and an octamer motif (9, 15). Thus, a possi-bility is that the role of the Oct-1 activation domains is to reinforce this effect. Indeed, we find that the Oct-1 regions outside of the POU domain contribute to cooperative binding with SNAP c , and the effect is significantly larger than that observed with Oct-2 regions outside of the POU domain. There is, therefore, a correlation between the ability of the Oct-1 and Oct-2 regions outside of the POU domain to recruit SNAP c to the PSE and to enhance snRNA gene transcription in vivo (14). Nevertheless, most of the SNAP c recruitment effect is contributed by the POU domain, suggesting that the main function of the Oct-1 activation domains is different. They may be involved either in recruiting other members of the basal machinery, perhaps other snRNA-promoter specific factors, or, for example, in inducing conformational changes in the basal machinery that result in more efficient transcription.
While characterizing a PSE-binding complex observed in crude nuclear extracts, we found that it consists of SNAP c and Oct-1. This was unexpected, because Oct-1 on its own did not bind effectively to the probe, and our previous results had indicated that cooperative binding of Oct-1 and SNAP c to DNA requires the presence of both a PSE and an octamer motif (15). However, we find that within the complex, Oct-1 contacts DNA in a sequence-specific manner, and that this contact is required for formation of the complex. Thus, as we had observed previ- ously, cooperative binding of SNAP c and Oct-1 to DNA requires Oct-1-DNA contacts. Interestingly, the location of these contacts relative to the location of the PSE is flexible and changes on different probes. This is consistent with the observation that neither the distance between the octamer sequence and the PSE (9) nor the orientation of the octamer (15), are critical for cooperative binding.
The POU domain consists of two helix-turn-helix-containing DNA-binding structures, the POU homeodomain (POU H ) and the POU-specific domain (POU S ), joined together by a flexible linker (30 -34). Cooperative binding of the Oct-1 POU domain and SNAP c can be disrupted by a single amino acid change within the POU S domain, which maps to the surface of helix 1 away from the DNA and has no effect on DNA binding (15). This suggests that the POU S domain is involved in direct protein-protein interactions with SNAP c , and that its position In the AD probe, the spacing between the PSE and the protected TTA region is shorter than in the N7 probe. In the AD-muHD probe, the protected TT in the TTA region have been mutated to CC. The AD-short probe is truncated at the 5Ј end. B, the AD and N7 probes, or the Octa probe which contains the H2B octamer motif, were incubated with the proteins indicated above the lanes and the binding reactions were fractionated by EMSA. The free probe and resulting complexes were then treated in situ with OP-Cu, and the DNA fragments corresponding to free probe, or to probe complexed with SNAP c , or to probe complexed with SNAP c and Oct-1 were eluted from the gel and fractionated on a sequencing gel. The locations of the PSE, the ATT sequence, and the octamer sequence are indicated. The regions protected on the AD probe are indicated by black boxes, the regions protected on the N7 probe by gray boxes, and the region protected on the Octa probe by a white box. C, the sequence of the H2B octamer is shown on top, and a comparison between the H2B octamer sequence and both strands of the region protected in the N7 and AD probes is shown on the bottom. D, EMSA performed with the probes and the proteins indicated above the lanes. The location of the free probe, free single-stranded probe (ssProbe), and the complexes containing SNAP c , or SNAP c together with His-Oct-1, are indicated. may, therefore, be fixed relative to SNAP c and the PSE. We show here that cooperative binding can also be disrupted by a single amino acid mutation within the POU S domain that maps to the surface of helix 3 pointing toward the DNA and that affects DNA binding (12). This suggests that the POU S domain also contacts the DNA in the complex. How, then, can there be so much flexibility in the spacing between the PSE and the sequences contacted by Oct-1 POU? Perhaps the POU S domain-DNA contact is transient, occurring only during formation of the trimeric complex, whereas the POU H domain remains bound to DNA in the formed complex. Or perhaps the location of the POU S domain-DNA contact is dictated more by proteinprotein interactions with SNAP c than by specific DNA sequence. In contrast, the position of the POU H may be much more dependent on local DNA sequences than on the position of SNAP c and the PSE. Indeed, the sequence ATT (or AAT on the other strand) constitutes part of the AAAT sequence recognized by the POU H domain on a histone H2B-octamer site (33). Thus, perhaps on different probes, the relative locations of the Oct-1 POU S and POU H domains changes, the first being dictated mainly by the location of SNAP c , and the second by the local DNA sequence. Alternatively, SNAP c may itself be flexible, allowing different positionings of Oct-1 on the DNA while maintaining protein-protein contacts.