Assembly of SNAPc, Bdp1, and TBP on the U6 snRNA Gene Promoter in Drosophila melanogaster

The DmSNAPc-Bdp1 complex can recruit TBP to the U6 promoter.Recently published data from our laboratory revealed that DmSNAPc can recruit Bdp1 to form a stable DmSNAPc-Bdp1 complex on the U6 gene promoter in the absence of TBP and Brf1 (32). To follow up that observation, we have now examined whether the DmSNAPc-Bdp1 complex plays a direct role in the recruitment of TBP to U6 promoter DNA. To that end, electrophoretic mobility shift assays (EMSAs) were performed (Fig. 1A) in which increasing amounts of TBP were incubated with the U6 promoter probe DNA in the absence (lanes 9 to 11) or presence (lanes 3 to 5) of DmSNAPc or both DmSNAPc and Bdp1 (lanes 6 to 8). Two reactions (lanes 12 and 13) contained still larger amounts of TBP and demonstrated that these larger amounts of TBP could bind to the U6 probe in the absence of other proteins.

• Open in new tab
• Download powerpoint

FIG 1

DmSNAPc-Bdp1 complex efficiently recruits TBP to the U6 promoter. (A) EMSA reactions with a wild-type U6 promoter DNA probe and increasing amounts of TBP (0.25, 0.5, and 0.75 μl) added in the presence of 3 μl DmSNAPc (lanes 3 to 5), in the presence of both 3 μl DmSNAPc and 2.5 μl Bdp1 (lanes 6 to 8), or TBP with no additional proteins (lanes 9 to 11). Lanes 12 and 13 show reactions with larger amounts of TBP (1 μl and 1.5 μl, respectively) to reveal the binding of TBP alone to the probe. Lanes 1 and 15 show reactions with DmSNAPc alone, and lanes 2 and 14 show reactions that contained DmSNAPc and Bdp1 together but no TBP. Results from quantitation of the relative band intensities by densitometry are shown in the lower part of the figure. It is possible that the lowest shifted band in lanes 3 to 8 includes some contribution from TBP, but even so, the conclusions from the experiment are unaffected. (B) EMSAs with specific antibody supershifts are shown. Each reaction mixture contained 3 μl DmSNAPc; lanes 2 to 6 each contained 2.5 μl of Bdp1; and lanes 3 to 6 each contained 0.5 μl of TBP. The TBP-Bdp1-DmSNAPc complex (lane 3) was supershifted by 1.5 μl of antibody against either DmSNAPc (α-190, lane 4), Bdp1 (α-Myc, lane 5), or TBP (α-V5, lane 6). Free probe was run off the gel to improve the resolution among the shifted bands. Lanes 1 and 2, although nonadjacent, are both from the same gel as lanes 3 to 6.

Reactions that contained the smaller amounts of TBP alone exhibited only very weak gel shifts (Fig. 1A, lanes 9 to 11). When the same amounts of TBP used in these reactions were added to a constant amount of DmSNAPc (Fig. 1A, lanes 3 to 5), clear DmSNAPc bands were observed together with a very light smear of slower-migrating bands that likely correspond to DNA fragments cooccupied by DmSNAPc and TBP (32). Importantly, when Bdp1 was present together with DmSNAPc and TBP, even the smallest amount of TBP was sufficient to form a higher-order and stronger complex of slower mobility (Fig. 1A, lanes 6 to 8). Moreover, essentially all the DmSNAPc bandshift seen in lanes 3 to 5 was supershifted into the higher-order complex in lanes 6 to 8. The addition of Bdp1 to TBP without DmSNAPc produced no bandshift beyond that seen with TBP alone (data not shown). The quantitation of these results (Fig. 1A, lower section) indicates that TBP is more efficiently recruited to the U6 promoter DNA by DmSNAPc and Bdp1 than DmSNAPc alone (lanes 6 to 8 versus lanes 3 to 5). These results indicate that Bdp1 can cooperate with DmSNAPc to recruit TBP to the U6 promoter.

The presence of all three proteins in the higher-order complex was confirmed by antibody supershifts (Fig. 1B). The TBP-Bdp1-DmSNAPc complex migrated more slowly than the Bdp1-DmSNAPc complex (Fig. 1B, lane 3 versus lane 2), and this complex was further supershifted by antibodies against either DmSNAP190, the Myc epitope on Bdp1, or the V5 epitope on TBP (Fig. 1B, lanes 4, 5, and 6, respectively).

An 87-amino-acid region of Bdp1 between residues 424 and 510 is sufficient for TBP recruitment to the DmSNAPc-U6 promoter complex.We next investigated whether a specific region of Bdp1 is involved in recruiting TBP to the DmSNAPc-U6 promoter complex. Six Bdp1 constructs that had truncations exclusively at either the N or C terminus first were examined for their ability to recruit TBP. Previously, we found that DmSNAPc was able to recruit each of these six constructs to U6 promoter DNA, whereas Bdp1 constructs with further truncations from either the N or C terminus were not recruited (32). The upper panels of Fig. 2 (lanes 1 to 20) show that each of these Bdp1 truncation mutants was able to recruit TBP to form a higher-order complex with DmSNAPc on the U6 promoter (bands indicated with small circles). Quantitation of the results (table at the bottom of Fig. 2) suggests that the deletion of Bdp1 sequences located N-terminal of residue 335 or C-terminal of 510 actually increases the ability of the remaining Bdp1 sequences [constructs Bdp1(335-695) and Bdp(1-510)] to recruit TBP. This prompted us to further investigate the role of Bdp1 residues 335 to 510 in TBP recruitment.

• Open in new tab
• Download powerpoint

FIG 2

Eighty-seven-residue region of Bdp1 just C-terminal of the SANT domain is sufficient for TBP recruitment in the presence of DmSNAPc. EMSAs are shown that utilized various Bdp1 truncation constructs together with DmSNAPc and full-length TBP. At the top of the figure, truncated Bdp1 constructs utilized in EMSAs are illustrated as long rectangles. The location of the highly conserved SANT domain is indicated by a gray box within each construct, and the region identified as important for DmSNAPc recruitment of Bdp1 (amino acid residues 424 to 510) (32) is indicated by the black box within each construct. The particular Bdp1 construct utilized in each lane is indicated above the respective lanes and varied between 2 μl and 7 μl. Reactions run in all lanes contained DmSNAPc (3 μl). TBP (0.5 μl) was added to the reaction mixtures loaded in the lanes indicated by plus signs. The circles indicate the positions of the bands that correspond to the complexes formed by DmSNAPc together with TBP and the various truncation constructs of Bdp1. The asterisks, on the other hand, indicate the bands that correspond to DmSNAPc-Bdp1 complexes on the DNA that do not include TBP. Lanes 33 to 39 are all from the same exposure of the same gel. Quantitation of the results of lanes 1 to 32 is shown at the bottom of the figure and represents the ratio (relative to the wild type, normalized to 1.0) of the TBP-Bdp1-DmSNAPc-U6 promoter complex to the Bdp1-DmSNAPc-U6 promoter complex in the same lane.

For the next experiments, we used three Bdp1 constructs that contained either residues 335 to 510, only residues 335 to 424 (the SANT domain), or only residues 424 to 510. The two constructs that contained Bdp1 residues 424 to 510 formed a stable complex with DmSNAPc that recruited TBP to the U6 promoter as well as or better than the wild type (Fig. 2, lanes 24 to 27, bands indicated by circles; also see quantitation at the bottom). In contrast, the construct that contained only the SANT domain had little, if any, ability to recruit TBP in concert with DmSNAPc (Fig. 2, lanes 28 and 29). These results, together with those previously published (32), indicate that residues 424 to 510 are sufficient both for Bdp1 to interact with DmSNAPc and to recruit TBP.

It is nonetheless entirely possible that the SANT domain partially contributes to the efficiency of TBP recruitment. For example, we noticed a reproducible reduction in the intensity of the TBP-Bdp1-DmSNAPc band in the case of constructs that lacked the SANT domain compared to similar constructs that contained the SANT domain [Fig. 2, construct Bdp1(424-695) in lane 12 versus Bdp1(335-695) in lane 10 and construct Bdp1(424-510) in lane 27 versus construct Bdp1(335-510) in lane 25; also see quantitation at the bottom]. This is not surprising, because crystal and cryoelectron microscopy (cryo-EM) structures indicate that the human and yeast Bdp1 SANT domains directly contact TBP (17, 35, 36).

The importance of Bdp1 residues 424 to 510 for recruitment of Bdp1 and TBP by DmSNAPc was further investigated by using a construct [Bdp1Δ(424-510)] that contained an internal deletion of only the 87 residues from 424 to 510 (last construct illustrated in the upper section of Fig. 2). The wild-type and deletion constructs were normalized by Western blotting against the FLAG epitope and used in side-by-side assays together with DmSNAPc and TBP (Fig. 2, lanes 33 to 39). DmSNAPc was unable to recruit Bdp1Δ(424-510), whereas it did recruit the wild-type Bdp1 (lanes 34 and 35). Similarly, in contrast to the wild-type Bdp1, Bdp1Δ(424-510) was unable to cooperate with DmSNAPc to recruit TBP (lanes 36 and 37), even though the deletion construct contained all of the Bdp1 sequence except amino acid residues 424 to 510. Thus, the remaining 608 residues of Bdp1 (residues 1 to 423 and 511 to 695 together) cannot compensate for the loss of the 87 amino acids from 424 to 510 with respect to Bdp1 and TBP recruitment to the DmSNAPc-U6 promoter complex.

TBP recruitment by the DmSNAPc-Bdp1 complex requires a TATA box.We next investigated whether TBP recruitment under such circumstances was also dependent upon the presence of the TATA box in the U6 promoter. For this, a DNA EMSA probe was used that was identical to the wild-type probe, except the TATA sequence (TTTATATA) was mutated to GGGACCTC. As expected, TBP by itself was able to bind to the wild-type probe but did not bind to the mutant TATA probe (Fig. 3A, lanes 1 and 2). On the other hand, DmSNAPc bound equally well to both the wild-type and mutant TATA probes (Fig. 3A, lanes 3 and 4). Interestingly, the mutation of the TATA sequence had no effect on the recruitment of Bdp1 by DmSNAPc (Fig. 3A, lanes 5 and 6). In stark contrast, the DmSNAPc-Bdp1 complex effectively recruited TBP to the wild-type promoter but was unable to recruit TBP to the mutant TATA promoter (Fig. 3A, compare lanes 7 and 8). These results indicate that the recruitment of TBP, but not Bdp1, to the U6 promoter by DmSNAPc is dependent upon the presence of a TATA sequence. The results further strongly suggest that TBP directly interacts with the TATA sequence when it is recruited to the U6 promoter by DmSNAPc and Bdp1.

• Open in new tab
• Download powerpoint

FIG 3

TBP recruitment to the U6 promoter by DmSNAPc-Bdp1 requires a TATA box. (A) Shown is an EMSA that utilized either the wild-type U6 promoter probe (odd-numbered lanes) or a probe that was identical to the wild-type sequence with the exception of a mutated TATA box (even-numbered lanes). Each probe was incubated with either 3 μl of DmSNAPc alone (lanes 3 and 4), 3 μl of DmSNAPc together with 2 μl of Bdp1 (lanes 5 and 6), or 3 μl of DmSNAPc together with 2 μl of Bdp1 and 0.4 μl of TBP (lanes 7 and 8). Each probe was also incubated with 1 μl of TBP alone (lanes 1 and 2). The identities of the shifted bands and free probe are indicated to the left of the panel. (B) Unlabeled wild-type or mutant TATA-containing oligonucleotides were used as competitors in EMSAs to further examine the requirement of a TATA box for TBP recruitment to the U6 promoter. Each reaction mixture contained 1 μl DmSNAPc and 2 μl Bdp1. All other reaction mixtures also contained 1.2 μl of TBP, with the exception of the reaction mixtures shown in lanes 1 and 16. Unlabeled competitor oligonucleotides were added in increasing amounts (1.5, 5.0, 15, 50, 150, and 500 ng) to lanes 3 through 8 (mutant TATA competitor) and to lanes 10 through 15 (wild-type TATA competitor).

As an alternative method to examine the importance of the TATA box for TBP recruitment, EMSAs were performed with increasing amounts of unlabeled oligonucleotides that contained either the wild-type or the mutated TATA sequence (Fig. 3B). The competitor oligonucleotides lacked the PSEA sequence so as not to compete away binding of DmSNAPc (and Bdp1) to the labeled U6 promoter probe. The results indicate that the TBP-Bdp1-DmSNAPc band was competed away more effectively by the TATA-containing oligonucleotide than the oligonucleotide that contained a mutant TATA sequence (compare lanes 3 to 8 with lanes 10 to 15). Quantitation of the data (not shown) revealed that the oligonucleotide with the TATA sequence was a greater than 50-fold better competitor than the oligonucleotide with the mutated sequence, providing further evidence that the TATA sequence is essential for efficient TBP recruitment.

Identification of nucleotide positions where Bdp1 cross-links to the U6 snRNA gene promoter as part of a DmSNAPc-Bdp1-TBP-DNA complex.In previous work, we used site-specific protein-DNA photo-cross-linking assays to map the nucleotide positions of the U6 promoter that are in close proximity to the three different subunits of DmSNAPc when DmSNAPc alone is bound to the DNA (33). In a more recent study, we similarly mapped the nucleotide positions where each of the three subunits of TFIIIB map to the U6 promoter DNA when TFIIIB alone is bound to the DNA (28). To extend these data, we have now investigated the binding pattern of the DmSNAPc-Bdp1-TBP ternary protein complex on the U6 promoter DNA.

To perform the site-specific protein-DNA photo-cross-linking assays, we prepared 28 different double-stranded DNA probes that contained the sequence of the promoter of the D. melanogaster U6:96Ab gene (Fig. 4A). Each probe contained a photo-cross-linking agent at a unique position in the phosphate backbone together with an adjacent radiolabel. The phosphate positions individually assayed by this procedure are indicated by the asterisks in Fig. 4A. Probes were coincubated with DmSNAPc, Bdp1, and TBP, irradiated with UV light, and subjected to nuclease digestion as previously described (33). Cross-linked proteins were identified by SDS-PAGE followed by autoradiography.

• Open in new tab
• Download powerpoint

FIG 4

Site-specific protein-DNA photo-cross-linking of Bdp1 in a complex with DmSNAPc and TBP to the U6 snRNA gene promoter. (A) DNA probe used for protein-DNA photo-cross-linking assays. The cross-linker was placed individually at 28 different phosphate positions (14 on each strand), as indicated by the asterisks. The negative numbers indicate the nucleotide distance from the transcription start site. The PSEA is indicated with blue color, and the TATA box is indicated with red color. (B) Autoradiography of cross-linked Bdp1, DmSNAPc, and TBP to 28 nucleotide positions within the U6 snRNA gene promoter. The identities of the protein bands are indicated to the left of each panel (except for TBP, which cross-linked to nucleotides within and closely flanking the TATA box). The apparent cross-linking of DmSNAP190 to positions −42 and −40 was not seen when the reactions were done with DmSNAPc that contained FLAG-tagged DmSNAP43 rather than DmSNAP190 (data not shown). This indicates that there is an excess of FLAG-tagged DmSNAP190 in the experiments shown here due to the use of anti-FLAG resin for the purification of DmSNAPc.

Figure 4B (upper) shows that Bdp1, when assembled with DmSNAPc and TBP, cross-linked to the nontemplate (upper) strand at positions −46 to −42 and −36 to −30 with various intensities. On the template strand (Fig. 4B, lower), Bdp1 cross-linked to phosphate positions −47, −45, and −39 to −35, again with various intensities. The upper and lower strand patterns are highly consistent with each other under the assumption that Bdp1 interacts primarily with a single face of the DNA over the region examined in these experiments. The cross-linking of Bdp1 that occurred on both DNA strands at the positions near the TATA box was not surprising based on results we previously obtained with TFIIIB alone (28) and results observed by others from yeast TFIIIB cryo-EM structures (17).

Interestingly, the data also revealed that nucleotides farther upstream (phosphates −46, −44, −42, −47, and −45) also cross-linked to Bdp1. These phosphate positions are located within the 3′ half of the PSEA itself at nucleotides that also cross-link to one or more of the DmSNAPc subunits. From these protein-DNA photo-cross-linking results, it can be surmised that Bdp1 is near nucleotides in the 3′ portion of the U6 PSEA. This strongly suggests that a region of Bdp1 must lie in very close physical proximity to one or more subunits of DmSNAPc. This is consistent with the existence of direct protein-protein contacts that permit DmSNAPc to recruit Bdp1 to the U6 promoter (32).

With respect to the cross-linking of DmSNAPc to the DNA, the patterns of cross-linking of the DmSNAPc subunits in Fig. 4B is very similar to the cross-linking patterns of those previously reported by Wang and Stumph (33). The only noticeable difference is that DmSNAP190 cross-linked more strongly to position −36 in the presence of Bdp1 than DmSNAPc alone (14). This suggests that DmSNAP190 is more “locked down” on the DNA in the presence of Bdp1 than with DmSNAPc alone.

Residues 424 to 510 of Bdp1 cross-link to the U6 snRNA gene promoter upstream of the TATA box and within the PSEA.To identify the region(s) of Bdp1 that cross-linked to the U6 snRNA gene promoter upstream of the TATA box, we utilized several truncated Bdp1 proteins together with DmSNAPc and TBP in photo-cross-linking experiments. Because amino acid residues 424 to 510 of Bdp1 were implicated in both Bdp1 and TBP recruitment by DmSNAPc (Fig. 2) (32), we performed photo-cross-linking experiments with N-terminal and C-terminal truncations of Bdp1 that included or lacked this region (Fig. 5A). For these protein-DNA photo-cross-linking assays, we examined five nucleotide positions from each strand (phosphates −44, −42, −36, −34, and −32 on the nontemplate strand and phosphates −47, −45, −39, −37, and −35 on the template strand), positions where full-length Bdp1 had been observed to cross-link (Fig. 4).

• Open in new tab
• Download powerpoint

FIG 5

Site-specific protein-DNA photo-cross-linking of truncated Bdp1 constructs, DmSNAPc, and TBP at select nucleotide positions within the U6 snRNA gene promoter. (A) Truncation constructs used to identify a region of Bdp1 that cross-links to the U6 snRNA gene promoter in the presence of DmSNAPc and TBP. (B) Autoradiography of protein-DNA cross-linking reactions at ten nucleotide positions where full-length Bdp1 (Bdp1 FL) cross-linked (Fig. 4). In each panel, constructs that contained residues 424 to 510 [Bdp1 FL, Bdp1(424-695), and Bdp1(1-510)] cross-linked to each position examined (bands indicated by closed circles). On the other hand, constructs missing residues 424 to 510 did not cross-link or cross-linked extremely weakly. The positions where bands should appear for these constructs (determined by Western blotting) are indicated by squares. (C) Cross-linking of full-length Bdp1 (Bdp1 FL) and construct Bdp1(424-510) to the same 10 nucleotide positions as in panel B.

Figure 5B shows that full-length Bdp1 as well as two Bdp1 constructs that contained amino acid residues 424 to 510 [Bdp1(424-695) and Bdp1(1-510)] each cross-linked strongly to all ten of the assayed positions. However, the two constructs that lacked these residues either did not detectably cross-link [Bdp1(511-695)] or yielded an extremely weak signal [Bdp1(1-423)].

Those results indicate that a region of Bdp1 between amino acid residues 424 and 510 is involved in the cross-linking interaction with the U6 snRNA gene promoter. To further examine this possibility, the construct Bdp1(424-510) was employed in cross-linking assays with the same ten phosphate positions used for Fig. 5B. Due to the small size of the construct, the cross-linked samples were run on higher-percentage gels (Fig. 5C). We observed that amino acid residues 424 to 510 of Bdp1, in a complex with DmSNAPc and TBP, cross-linked to each of these nucleotide positions (Fig. 5C, lanes 6 to 10 and lanes 16 to 20) where full-length Bdp1 cross-linked (lanes 1 to 5 and 11 to 15). Moreover, the pattern of cross-linking intensities with Bdp1(424-510) was very similar to that obtained with the full-length Bdp1. In conclusion, these findings indicate that an 87-amino-acid region of Bdp1 that is sufficient for Bdp1 recruitment by DmSNAPc is also in close proximity to the U6 promoter DNA upstream of the TATA box, apparently positioned there by an interaction with DmSNAPc.

When bound to U6 promoter DNA, Bdp1 residues 424 to 510 form an interface with DmSNAP43 and DmSNAP190.To further investigate regions of Bdp1 that may interact with DmSNAPc on the U6 promoter, we next used cross-linking mass spectrometry (CXMS). DmSNAPc and Bdp1 were coincubated together with U6 promoter DNA. The complexes then were treated with the chemical cross-linker BS3, which can cross-link lysines that are in proximity to each other. No cross-links between Bdp1 and DmSNAP50 were identified (data not shown). However, we were able to identify 23 lysine-lysine cross-links between Bdp1 and either DmSNAP43 or DmSNAP190 (Fig. 6A).

• Open in new tab
• Download powerpoint

FIG 6

BS3 cross-linking map for the DmSNAPc-Bdp1-U6 promoter complex (A) and the DmSNAPc-Bdp1-TBP-U6 promoter complex (B). The green dots indicate the positions of lysine residues along the protein sequences. The red dots and lines between the dots indicate cross-linked lysine residues that were identified by mass spectrometry. The gray rectangle indicates the region of Bdp1 required for Bdp1 and TBP to be recruited to the U6 promoter by DmSNAPc (Fig. 2). Only cross-links involving Bdp1 and/or TBP are shown.

Notably, 14 of the 23 Bdp1 cross-links are within the region of Bdp1 (residues 424 to 510) that is required for Bdp1 recruitment by DmSNAPc, and three additional cross-links involved lysines close to this region (424 to 510) (two from within the C-terminal part of the SANT domain). These results indicate that this region of Bdp1 is in close proximity to DmSNAP43 and DmSNAP190 and, together with results presented in Fig. 2 and 5, suggest that this region forms an interface with DmSNAP43 and DmSNAP190. The only other part of Bdp1 that cross-links to DmSNAPc in our experiments (6 cross-links) is a more N-terminal region that includes lysines between amino acids 203 and 231 (Fig. 6). These results indicate that a second region of Bdp1 N-terminal of the SANT domain is also in close proximity to DmSNAP43 and DmSNAP190; however, this region is not essential for a stable interaction with DmSNAPc on the U6 promoter (Fig. 2).

The DmSNAP190 cross-links to Bdp1 occur mostly within the regions N-terminal (eight cross-links) and C-terminal (seven cross-links) of the Myb domain, compared to only two cross-links within the Myb domain itself. Similarly, the majority of the cross-links of DmSNAP43 to Bdp1 occur mainly within the nonconserved C-terminal portion of DmSNAP43. Together, these CXMS results suggest that the nonconserved N- and C-terminal regions of DmSNAP190 and the C-terminal region of DmSNAP43 are in close proximity to Bdp1 and may be involved in recruiting Bdp1. DmSNAP50, on the other hand, likely lies more distant from Bdp1 than DmSNAP190 and DmSNAP43. These results are in very good accord with the previously modeled locations of the DmSNAPc subunits on the U6 promoter that were based primarily upon protein-DNA photo-cross-linking (14, 29, 31, 33).

DmSNAP43 forms an interface with TBP in the DmSNAPc-Bdp1-TBP-U6 promoter complex.We next performed CXMS with TBP in the reaction together with DmSNAPc, Bdp1, and the U6 promoter. We identified 19 intersubunit links between DmSNAPc, Bdp1, and TBP. The majority of these were cross-links that involved the lysine-rich FLAG epitope tags (data not shown); however, eight cross-links involved residues within the natural coding regions of these proteins (Fig. 6B). Although fewer cross-links were identified in the sample containing TBP (Fig. 6B) than without TBP (Fig. 6A), the pattern of cross-linking between DmSNAP190 and Bdp1 is similar. The cross-link between TBP Lys:250 and Bdp1 Lys:455, within the region of residues 424 to 510, is consistent with the ability of this region of Bdp1 to recruit TBP to the U6 promoter (Fig. 2).

The cross-link of TBP to a lysine within Bdp1(203-231) suggests that this region of Bdp1 is close not only to DmSNAP190 and DmSNAP43 (Fig. 6A) but also to TBP. Finally, the cross-link between the Bdp1 SANT domain and TBP is consistent with the known interaction of the SANT domain with TBP (17).

Perhaps the most interesting aspect of the results shown in Fig. 6B is the two cross-links that indicate the proximity of DmSNAP43 to TBP. This is not unexpected based on the ability of DmSNAPc and TBP to form a complex on the U6 promoter (Fig. 1) (32). When taken together with the site-specific protein-DNA photo-cross-linking data, the results strongly suggest that DmSNAP43 is an elongated molecule that can both interact with the 3′ half of the PSEA and reach TBP on a downstream TATA box.

When TBP was present in the sample (Fig. 6B), no cross-links were observed between DmSNAP43 and Bdp1; this contrasts with the results obtained in the absence of TBP (Fig. 6A). Based on the available data, the explanation for this difference currently is not clear. While the change in the cross-linking pattern could be due to a conformational change in the complex or occlusion by TBP, it also could result from undersampling due to the complexity or the limited availability of the sample.

A model of DmSNAPc, Bdp1, and TBP on a D. melanogaster U6 snRNA gene promoter.The data presented here (from EMSAs, site-specific protein-DNA photo-cross-linking, and CXMS) allow us to develop a more complete and detailed model that builds upon a series of previous models of DmSNAPc bound to the U6 promoter (14, 29, 31, 33). Together, the data permit us to better integrate DmSNAPc with Bdp1 and TBP. Our updated understanding is shown in Fig. 7. This model contains nearly all of DmSNAPc, a portion of Bdp1, and the conserved region of TBP.

• Open in new tab
• Download powerpoint

FIG 7

Model of a partial preinitiation complex assembled on the U6 snRNA gene promoter consistent with EMSAs, site-specific protein-DNA photo-cross-linking, and CXMS. DmSNAP190 is shown in yellow and gray, DmSNAP50 in green, DmSNAP43 in blue, Bdp1 in orange, and TBP in red. At the top, the linear structure of DmSNAP190 is shown with DmSNAP190 residues 1 to 456 and 528 to 721 indicated in yellow and gray, respectively, to correspond to the same two regions of DmSNAP190 in the panels below. The entire DmSNAP50 and DmSNAP43 are shown in green and blue, respectively. The naked B-form DNA structures shown above the protein-DNA structure are based on the U6-96Ab gene sequence between positions −73 and −23. The nontemplate strand is shown in light gray and the template strand in dark gray. The U6 PSEA is indicated by raspberry-colored bases, except for the five bases (light orange) that differ from the U1-95.1 gene PSEA, and are important for Pol III versus Pol II specificity (32, 45–47). The TATA box bases are shown in cyan. The naked DNAs also show the phosphate positions to which DmSNAP43 (blue), DmSNAP50 (green), and DmSNAP190 (yellow) cross-linked in this and previous work. Many of the phosphate positions cross-linked to more than one DmSNAPc subunit; in that case, individual spheres contain more than one color. Shown separately in orange are the phosphate positions to which Bdp1 cross-linked. The portion of the DmSNAP190 structure shown in yellow is based upon that of the fission yeast Reb1 protein; flexible loops between Myb repeats were used to position the Myb repeats in concert with previous data (31). The gray portion of the DmSNAP190 structure represents the C-terminal 179 amino acid residues of DmSNAP190 modeled as a ligand-binding domain of a nuclear hormone receptor placed in accordance with CXMS DmSNAP190-Bdp1 SANT domain cross-links. DmSNAP50 and DmSNAP43 are structures generated as described in the text. and placed onto the DNA in locations consistent with previous and current site-specific protein-DNA photo-cross-linking. Constraints from CXMS data together with protein-DNA cross-linking indicate that DmSNAP43 has an extended structure stretching from the PSEA to the upper surface of TBP. The placement of TBP and Bdp1 into the combined structure is described in the main text.

DmSNAP190 residues 1 to 441 (61% of the protein) could be modeled surprisingly well by the RaptorX structure prediction program (37) by using the crystal structure of the Schizosaccharomyces pombe Reb1 DNA-binding protein as a template (P value of 5.10e−05). DmSNAP190 is predicted to have structural similarity to Reb1 not only in its 4.5 Myb repeat region (residues 183 to 441, known as Myb repeats Rh, Ra, Rb, Rc, and Rd) but also in the DmSNAP190 N-terminal domain (residues 1 to 182). The latter domain of DmSNAP190 interacts extensively with the DNA (29–31), a feature shared with the homologous region of Reb1 (38). Therefore, we used the Reb1-templated structure of DmSNAP190 to model the first 441 residues of DmSNAP190 on the U6 PSEA (Fig. 7, protein residues shown in yellow).

Residues 1 to 182, previously modeled only as an unstructured oval, are positioned to interact with the 3′ half of the PSEA, in accordance with previous site-specific protein-DNA photo-cross-linking data (29–31). The 4.5 Myb repeats (residues 183 to 441) are placed on the DNA similarly to previous models that also relied upon protein-DNA cross-linking experiments in which the location of each individual Myb repeat was mapped on the U6 promoter DNA (31).

The structure of DmSNAP50 is unknown, and there seems to be no template sufficient to model its structure. Previously we mapped specific segments of these proteins that photo-cross-link to specific nucleotides in the U6 promoter (29). We used RaptorX to generate a priori globular structures of segments of DmSNAP50 and placed these in position to cross-link to the proper nucleotides of the DNA, as previously mapped (29). We do not presume that the a priori model represents the actual secondary and tertiary structures of DmSNAP50, but we believe that the globular, correctly sized structures obtained from the modeling program can serve as proxies to localize segments of the protein on the DNA based upon our previous protein-DNA photo-cross-linking. The whole of DmSNAP50 was modeled and is shown in green in Fig. 7. The use of B-form DNA is reasonable, as previous work indicated that the DNA was only slightly bent by the binding of DmSNAPc (39).

Prior to modeling DmSNAP43, we inserted TBP and D. melanogaster Bdp1 into the structural model. We used a published yeast TFIIIB structure (17) and overlapped the nearly B-form yeast DNA that extends upstream of the TATA box with the B-form DNA that contains the PSEA. We aligned the sequences with the proper separation (i.e., 12 bp) between the PSEA and the TATA box (40). This alignment of the DNA allowed the visualization of yeast TBP (red in Fig. 7) at the correct distance and rotational position relative to DmSNAPc. Because this same yeast structure included the very highly conserved SANT domain of Bdp1, we used the yeast Bdp1 SANT domain as a three-dimensional template to insert the corresponding domain of D. melanogaster Bdp1 (orange in Fig. 7) into the structure, replacing the SANT domain of the yeast protein with the SANT domain of fly Bdp1. The globular SANT domain that lies just behind TBP can be seen in the lower image of Fig. 7. Although the structural models of the SANT domains from the two species overlapped very well, RaptorX revealed no structural homologies between yeast and fruit fly Bdp1 outside the SANT domains (data not shown).

Residues 424 to 510 of D. melanogaster Bdp1, which lie just C-terminal of the SANT domain, are required for the recruitment of Bdp1 and TBP by DmSNAPc (Fig. 2) (32). Moreover, they cross-linked to specific nucleotides located between and within the PSEA and the TATA box (Fig. 5). Finally, this region, together with residues immediately adjacent to it, cross-linked by CXMS to both the DmSNAP190 and DmSNAP43 subunits (but not to DmSNAP50) (Fig. 6A). Residues 424 to 510 of fly Bdp1 are predicted to be disordered and are absent from the RaptorX modeling of fly Bdp1. However, just beyond that region, there was an unstructured loop that we utilized as a proxy to model the region that contains residues 424 to 510. Although these are different amino acids, we sculpted this region to fit into our model near DmSNAP190 and DmSNAP43, where it could also cross-link to the DNA in a manner that would be consistent with our site-specific protein-DNA photo-cross-linking results (Fig. 4 and 5). Much of this elongated loop, which represents Bdp1(424-510), is obscured behind DmSNAP43 in the upper image of Fig. 7.

Returning to DmSNAP190, we found that the last 179 residues at the C terminus of DmSNAP190 (residues 543 to 721) could be modeled by RaptorX using the ligand binding domain of mammalian nuclear hormone receptors (e.g., human lipid X receptor beta, thyroid hormone receptor, and mouse estrogen-related receptor gamma) as templates (P value of 1.39e−02). More importantly, we identified two crucial CXMS lysine cross-links (Fig. 6A) that indicated that residues 618 and 673 of DmSNAP190 reside in proximity to Bdp1 residues 391 and 402, respectively. The latter two Bdp1 residues are within the SANT domain of fly Bdp1. Therefore, these CXMS data indicate that the C-terminal region of DmSNAP190 must lie reasonably close to the Bdp1 SANT domain in the DmSNAPc-Bdp1-DNA complex. Figure 7 shows the C-terminal domain of DmSNAP190 (residues 543 to 721) modeled in gray color as a nuclear hormone receptor ligand binding domain. This domain of DmSNAP190 was placed in proximity to the Bdp1 SANT domain in an orientation and at a distance that would allow the respective lysines to be cross-linked by BS3.

Finally, DmSNAP43 (blue in Fig. 7) was modeled into the structure. Figure 6B shows that two pivotal BS3 cross-links between DmSNAP43 and TBP were identified. DmSNAP43 residues Lys:284 and Lys:132 cross-linked to TBP residues Lys:257 and Lys:341, respectively. These two TBP lysines lie on the upper surface of TBP in the first and second conserved repeats at positions equivalent to those of yeast TBP residues Lys:145 and Ala:229. To meet those constraints, as well as the constraints of the protein-DNA photo-cross-linking of DmSNAP43 to the PSEA, DmSNAP43 was modeled as an elongated molecule in which more than one region of the protein extended from the PSEA to the upper surface of TBP. The best template identified by RaptorX for the first 265 residues of DmSNAP43 was a family of synthetic helix-loop-helix-loop proteins (41) (P value of 3.3e−05); a priori modeling was used to generate a structure for the last 98 residues (266 to 363). In each case, the structures were refolded using the flexible loops between the alpha helices. The resulting DmSNAP43 structure is shown in blue in Fig. 7. This structure preserves the protein-DNA cross-linking proximity to the PSEA as well as the proximities of DmSNAP43 Lys:284 and Lys:132 to the respective upper surface residues of TBP.