Analysis of U1 Small Nuclear RNA Interaction with Cyclin H*

TFIIH is a general transcription and repair factor implicated in RNA polymerase II transcription, nucleotide excision repair, and transcription-coupled repair. Genetic defects in TFIIH lead to three distinct inheritable diseases: xeroderma pigmentosa, Cockayne syndrome, and trichothiodystrophy, with xeroderma pigmentosa patients being highly susceptible to skin cancer. Earlier data revealed that the cyclin H subunit of TFIIH associates with U1 small nuclear RNA, a core-splicing component. In addition to its role in RNA processing U1 small nuclear RNA also regulates diverse stages of transcription by RNA polymerase II both in vivo and in vitro, including abortive initiation and re-initiation. Here we identify structural components of U1 and cyclin H implicated in the direct interaction and show how they affect function. Because of unique features of cyclin H we have developed a new methodology for mapping RNA interaction with the full-length cyclin H polypeptide based on electrospray ionization tandem mass spectrometry. We also demonstrate the importance of U1 stem-loops 1 and 2 for the interaction with cyclin H. Functional assays implicate the identified interaction with U1 in regulation of the activity of the cyclin H associated kinase CDK7.

TFIIH is a multisubunit general transcription factor implicated in the regulation of RNA polymerase II (RNAPII) 3 transcription (1), as well as nucleotide excision repair and transcription-coupled repair (2,3). Functionally, TFIIH is involved in RNAPII transcription at the stages of initiation, promoter escape and early elongation (4,5), formation of the re-initiation scaffolding (6), and, potentially, the formation of gene loops (7). Structurally TFIIH is composed of nine subunits. There are two unidirectional DNA helicase activities in the six subunit core, and the remaining three subunits make up the CDK-activating kinase (CAK) complex (8,9). The CAK (CDK7/cyclin H/MAT1), in the context of TFIIH, phosphorylates the RNAPII large subunit carboxyl-terminal domain (CTD) (10,11). The phosphorylation of the CTD is involved in transcription initiation, elongation, and the coupling of transcription to pre-mRNA processing (4,5,12).
Earlier studies have described the modulation of P-TEFb kinase activity by an interaction with the non-coding RNA 7SK and the HEXIM (MAQ1) polypeptide (14,15). Following these observations we described a non-coding RNA interaction with TFIIH and the interacting RNA was identified as U1 snRNA (U1) (16). The interaction with U1 has been shown to modulate biochemical activities of TFIIH at the initiation and re-initiation stages of transcription in vitro and to correlate with in vivo observations of coupling transcription with the proximity of the first intron to the promoter (16 -19). U1 snRNA has been characterized as an important component of splicing machinery, implicated in early recognition of splice donor sites, and is also involved in the coupling of splicing with efficient transcription initiation and polyadenylation in vivo (12,20).
The CDK7-associated cyclin (cyclin H) was identified by UV crosslinking as the subunit that mediates the interaction between TFIIH and U1 (16). Here we present biochemical data defining the structural and functional interaction between cyclin H and U1. Taking into consideration the unique structural properties of cyclin H we have developed a novel method of mapping protein-RNA interactions based on electrospray ionization tandem mass spectrometry. Our approach enabled us to conduct mapping analysis on the full-length soluble folded polypeptide and to analyze the importance of the mapped interaction in the functional enzymatic assay.
For competition assays 1 l of a titration of non-labeled RNA (0.5, 50 fmol) was added to 100 fmol of cyclin H, 5 fmol of 32 P-labeled U1, 1 g of nonspecific competitor in 20 l of binding buffer on ice for 1 h. U1 and mutU1, a stem-loop 2 deletion mutant (⌬51-91) (22), were used as competitors. Wild-type U1 was in vitro transcribed from a PCR product derived from U1 plasmid (16) (deletion mutants were made using ExSite mutagenesis kit (Stratagene)).
RNA-Protein Cross-link and RNA Purification-50 pmol of recombinant cyclin H and 50 pmol of U1 were mixed in 84 l of 25 mM HEPES, 10% glycerol and incubated for on ice. After 1-h incubation 9.3 l of 30% formaldehyde was added. After 5-min incubation at room temperature 24 l of 1 M ammonium bicarbonate and 123 l of 1 M glycine (pH 8.3) were added. After 2-h incubation at room temperature the mixture was diluted 200-fold in 100 mM ammonium bicarbonate, and the volume was reduced to 15 l on a 3-kDa cut-off MicroSpin column (Millipore). 30 g of trypsin was added and the protein digested at 37°C for 16 h. 40 l of PHOS-Select beads (Sigma) were prepared in sample solution (25 mM acetic acid (pH 2.5-3)). The digested sample was added to the beads in 500 l of sample solution. After 16-h incubation at 10°C the beads were washed twice in 500 l of sample solution, the beads were then washed twice in water. 500 l of 400 mM ammonium hydroxide, 25% acetonitrile was added for 1 h at room temperature. The supernatant was removed and incubated for 2 h at 70°C before the sample volume was reduced to 10 l using a vacuum centrifuge. The sample was analyzed by electrospray ionization tandem mass spectrometry (Q-Tof Micro, Micromass) and identified HWTFSSEEQLAR (9 -21 aa) and IALTDAYLLYTPSQIALTAILSSASR (198 -223 aa) as cyclin H peptides specific for the cyclin H-U1 cross-linked complex. Non-specific peptides that interact with PHOS-Select beads were identified by purifying a tryptically digested sample of cyclin H in the absence of U1. The only peptide detected under those conditions was GYEDDDYSKK (296 -306 aa).
Recombinant CDK7/Cyclin H Kinase-Recombinant CDK7 and cyclin H (expressed as above) from pET28b in E. coli BL21 strain. Both polypeptides were mixed together and bound to a 1-ml nickel-Sepharose HiTrap column (Amersham Biosciences) in the presence of 8 M urea, renatured under a decreasing 10-ml gradient of urea, and eluted with a gradient of imidazole.
Purification of Native Kinase-20 l of protein G-conjugated Sepharose beads (Sigma) and 15 l of antibodies (cyclin H (C-18) sc-609 (Santa Cruz Biotechnology) or preimmune rabbit serum for the negative control) were bound in 1 ml of binding buffer (50 mM Tris (pH 8), 150 mM NaCl, 0.05% Triton X-100). After 1-h incubation 20 l of HeLa cell extract was added. After 2-h incubation at 10°C, the beads were washed three times for 5 min in 1 ml of binding buffer and once in 1 ml of kinase buffer (20 mM Tris (pH 7.8), 100 mM NaCl, 10 mM MgCl 2 ).
Kinase Assay-15 l of kinase buffer, 1 l of 1 mM ATP, 1 l of [␥-32 P]ATP (Amersham Biosciences), and 1 l of CTD peptide (10 mg/ml) were added to the recombinant or native kinase preparations. After 1-h incubation at 30°C 20 l of 2 ϫ SDS loading dye was added, and the samples were analyzed on a 16% SDS-polyacrylamide Trisglycine gel, which was visualized as above. For competition assays with the native kinase, after the final wash in binding buffer, a titration (10, 50, and 100 pmol) of peptides (peptides 1 and 2 or [Glu 1 ]fibrinopeptide B (as above)) was added to the beads in 20 l of water. After 1-h incubation at 4°C the beads were washed in 1 ml of kinase buffer, the wash was removed and the kinase assay carried out as above.

Formation of a Specific Complex between Cyclin H and U1-We have
confirmed the specificity of the reported interaction between cyclin H and U1 (16). First, we have used a newly developed RIP assay to test this interaction in vivo (21). As shown in Fig. 1A, immunoprecipitation with cyclin H antibodies from cultured HeLa cells following the short range formaldehyde cross-linking results in specific isolation of U1 RNA, detected by RT-PCR (Fig. 1A).
The specificity of the in vitro interaction between cyclin H and U1 has been tested by two independent approaches: RNA bandshift and affinity co-precipitation. First, we analyzed complex formation between the affinity-purified soluble recombinant cyclin H and synthetic 32 P-labeled U1 (Fig. 1B) by RNA bandshift (electrophoretic mobility shift assay). As shown in Fig. 1C, a stable reversible complex is observed in the presence of the nonspecific competitor poly(dI-dC)⅐poly(dI-dC) (Amersham Biosciences). It is important to emphasize that an excess of free probe was present in the binding reaction and that we have observed only one single complex under those conditions. The 32 P-labeled U1 probe was also specifically competed out of the complex by an excess of non-labeled U1 (Fig. 1C, lanes 4 and 5). In the second approach, we have taken advantage of the His 6 -tag added to the cyclin H polypeptide. Following the conditions for formation of the complex between cyclin H and 32 P-labeled U1, we were able to affinity purify the U1 on nickelprimed Sepharose beads via the His 6 -tag of the interacting cyclin H. Controls confirmed that the association with U1 was specific and dependent on cyclin H in this reconstituted pull-down assay (Fig. 1D).
Both in vitro assays of bandshift and cyclin H pull down provided us with conditions and concentrations for formation in the solution of a stable, specific complex between cyclin H and U1. These conditions were used as guidelines for the experiments described below.
Mapping the Interaction on Cyclin H-A standard approach to mapping the interaction site on a polypeptide involves assaying a set of deletion mutants for their binding capacity. However, cyclin H constitutes an exceptional case because of its unique structural features. Previous work by Andersen et al. (24) has shown that the ␣-helices at the NH 2 and COOH termini of cyclin H come into a proximal position and interact as part of the structural organization. Deletion of these two helices has a drastic effect on the polypeptide structure and the activity of the kinase (24). Therefore we were interested in developing a method of mapping the protein-RNA interaction in solution without introducing any terminal deletions.
Our strategy was to form a U1-cyclin H complex in solution, crosslink the protein to RNA, and follow it by tryptic digestion of the protein and purification of RNA with the cross-linked peptide (Fig. 2). In the final stage the peptide was identified by electrospray ionization tandem mass spectrometry. From several choices of cross-linking techniques we employed chemical cross-linking with formaldehyde, which had the advantage of reversibility (25), allowing us to identify unmodified peptides.
By using formaldehyde cross-linking and affinity purification of RNA-containing products with PHOS-Select iron affinity gel (Sigma) we isolated and identified two peptides (peptide 1 (HWTFSSEEQLAR (9 -21 aa)) and peptide 2 (IALTDAYLLYTPSQIALTAILSSASR) (198 -223 aa)); these peptides were specific for the U1-cyclin H cross-linked complex (Fig. 3A, marked in red). Importantly, despite the fact that the two specific peptides are distant from each other, when superimposed on the three-dimensional structure of cyclin H they form a single interface (Fig. 3B, marked in red) (PDB ID: 1KXU (www.pdb.org)) (26 -28).
To test whether the identified peptides are functionally implicated in the complex formation with U1, we tested their effect on the interaction. The 32 P-labeled U1-cyclin H complex was formed and affinitypurified on nickel-primed Sepharose via the His 6 -tag on the cyclin H polypeptide as described earlier (Fig. 1D). The U1-cyclin H complex was disrupted by the presence of the identified peptides in the reconstituted pull-down assay (Fig. 3C). This strongly suggests that these peptides are indeed involved in the interaction of cyclin H with U1.

Structure-Function Analysis of U1-Cyclin H Interaction
To further support the evidence for stem-loop 2 involvement in the interaction, we repeated RNA bandshifts (see Figs. 1C and 4C) in the presence of various specific competitors. With the titration range of 0.5 to 50 fmol, wild-type U1, but not the stem-loop 2 deletion mutant mutU1 (22), was able to compete out the complex (Fig. 4C, compare  lanes 4 and 6). This further implicated stem-loop 2 in the specific interaction with cyclin H.
To further delineate the properties of the U1 interaction with cyclin H, we also employed RNase footprinting. 5Ј-End 32 P-labeled U1 was digested with titrated amounts of RNase A and RNase T1 after formation of the complex with cyclin H. The resulting cleavage pattern revealed specific protected and hypersensitive sites. As shown in Fig. 5, footprinting analysis identifies protected areas within stem-loops 1 and 2 (Fig. 5, compare lane 2 with lanes 3-5) as well as a hypersensitive site within stem-loop 3 (Fig. 5, compare lanes 2 and 3 with lanes 4 and 5).
On its own the result of the footprinting experiment could not distinguish between the direct effect of cyclin H interaction and the indirect effect, i.e. a particular structural change within U1, induced by interaction with cyclin H. Nevertheless, protection of stem-loop 2, as observed in the footprinting experiment (Fig. 5, lanes 3 and 5), implicates the loop in the direct/indirect interaction with cyclin H and supports our earlier conclusion for the affinity purification of U1-cyclin H complexes (Fig. 4B) and for mutant competition of the U1-cyclin H complex in the bandshift assay (Fig. 4C). Overall, stem-loop 2 of U1 snNA is clearly implicated in the interaction with cyclin H.
Effect of U1 Interaction on the CDK7/Cyclin H Kinase Activity-Having identified the sites of interaction between cyclin H and U1, we proceeded to analyze the effect of the interaction on the enzymatic activity of the kinase, of which cyclin H is a regulatory component. Within the framework of in vitro biochemical study the relevance of the identified interaction could only be justified by its effects on the catalytic activity involved.
Previous studies have demonstrated that the CDK7/cyclin H kinase forms a tripartite complex (CAK) with the assembly factor MAT-1, which in turn stabilizes the in vitro CDK7/cyclin H complex and renders it more active, even independently of T-loop phosphorylation (30 -32). We therefore tested the effects of U1 on recombinant bipartite, tripartite, and native CDK7 kinases. Titration of U1 with recombinant (E. coli expressed) and renatured bipartite complex of CDK7 and cyclin H showed a consistent increase in the kinase activity when assayed against the CTD substrate (Fig. 6, A and B, lanes 2 and 3). Titration of U1 stem-loop 2 mutant mutU1 (22) did not affect the kinase activity (Fig.  6B, lanes 5-7). Bearing in mind the issue of low stability of the bipartite complex, we suspected that the presence of U1 may stabilize the interaction between the two subunits of the kinase. If that was the case, we  However, the eukaryotic background of baculovirus expression system also alerted us to the potential presence of contaminants, either RNA or protein in nature, that could already be modulating the kinase activity. We therefore tested native CDK7 kinase that was immunoprecipitated from HeLa cell extracts that, as shown before, contained U1 (16) as well as a number of other potential regulatory factors. Immunoaffinity-purified CDK7 kinase, but not the control purifications, displayed very high specific kinase activity when assayed against the CTD substrate. We then disrupted the interaction between cyclin H and U1 with titrated amounts of competing peptides. As shown in Fig. 6, titration of the specific peptides, which have previously been shown to disrupt the interaction with U1 (Fig. 3C), dramatically impaired the activity of the native kinase (Fig. 6C, lanes 6 -8). The control peptide showed no such effect. We concluded that in the context of the native immunoaffinity-purified CDK7 complex, interaction with U1 played an important role in supporting the kinase activity. This could be attributed to the impact of the additional regulatory factors recruited to the native complex via the cyclin H or the U1 side of the interaction.

DISCUSSION
Cross-linking Methodology-Transcriptional cyclin-dependent kinases CDK7 and CDK9 appear to be subject to regulation mediated by the non-coding RNAs U1 and 7SK (14 -16, 19, 33-35). Here we have conducted a biochemical analysis of the interaction between cyclin H and U1. A critical step in this analysis is based on a novel mass spectrometry based approach for mapping RNA-protein interactions. This method allows us to analyze the full-length polypeptide and to avoid deletion mutant analysis, which in the case of the cyclin H structure leads to the loss of structure and function of the peptide and the associated kinase activity (24). This new method allowed us to identify two peptides, which form a single interface on a free surface of cyclin H, away from the predicted CDK7 binding site (24, 26 -28). Unambiguously, when added as competitors, these peptides specifically disrupt the interaction of U1 with cyclin H in the reconstituted complex.
Questions remain concerning technical issues with the choices of cross-linking techniques. The widely used technique of UV-mediated cross-linking (36) can be carried out with high efficiency in the presence of nucleotide derivatives, such as 5-bromo-UTP and 5-iodo-UTP (36). We took the precaution to monitor whether the introduction of these derivatives had an effect on the specificity of the U1-cyclin H complex formation. Following earlier detection of the complex using RNA bandshift assays (Fig. 1C), we tested the formation of the same complex with 32 P-labeled U1 probes, synthesized with the incorporation of 5-bromo-UTP and 5-iodo-UTP. The bandshift assays revealed that the presence of the UTP derivatives in the RNA interfered with complex formation and overall reduced the yield of cross-linked product (supplemental Fig.  1). This result ruled out efficient UV-mediated cross-linking. The crosslink of choice, a reversible short range aldehyde cross-link, on the other hand proved successful. However, it is important to mention that within the mapping procedure the efficiency of tryptic digestion was badly affected by the cross-linking modification. Also, one should take into consideration potential masking of the tryptic digest site by bound RNA. Thus, we believe that our methodology benefited from having a highly pure two component system with high concentrations of both cyclin H and U1. We are now analyzing alternative reversible crosslinking techniques, which will allow us to obtain similar results in more complicated, low concentration ribonucleoprotein complexes.
Specific Interaction of CDK7 Kinase with U1-The described interaction of cyclin H with U1 has an important precedent. A second transcriptional kinase of the same class, CDK9, has been implicated in transcription elongation by phosphorylating the CTD of RNAPII (37,38). Importantly P-TEFb (CDK9/cyclin T) is regulated by interaction with non-coding RNAs. The RNA stem-loop structure of the human immunodeficiency virus transcript, TAR, and the Tat protein are implicated in the activation of the P-TEFb kinase activity (39,40). The 7SK RNA and HEXIM1 are implicated in the inactivation of this activity (14, 15, 33-35). More importantly, in both cases it appears that the cyclin sub-   1-3; B, lanes 2-4). B, titration of stem-loop 2 mutant mutU1 (22) (lanes 5-7) fails to stimulate the recombinant CDK7/cyclin H kinase activity. C, the activity of the native kinase is ablated in presence increasing amounts of peptides 1 and 2, interfering with U1 interaction (lanes 6 -8, see also Fig. 3C, lanes 5 and 6). The kinase was immunopurified using cyclin H antibodies ((C-18; sc-609 Santa Cruz Biotechnology). Preimmune serum was used for immunoprecipitation in lane 1 as a negative control. The carboxylterminal peptide recognized by the cyclin H antibodies does not overlap with peptides 1 or 2.
U1 snRNP-The observed coupling between the promoter proximal position of the first intron and promoter strength in vivo (17)(18)(19), and high levels of abortive initiation and re-initiation in vitro (16), raises interesting questions about the role of U1 snRNA. Assembly of the U1 snRNP involves recognition of U1 snRNA by the SMN complex (44,45). This is followed by recruitment of the Sm heptameric protein ring, hypermethylation of the snRNA 5Ј cap, 3Ј trimming, and nuclear import (20). Once in the nucleus the U1 snRNP recruits three more proteins: the U1-70K protein interacts with stem-loop 1, the U1-A protein interacts with the loop of stem-loop 2, and the U1-C protein is probably attached via protein-protein interactions (20). The three-dimensional structure of the U1 snRNP has been determined by singleparticle electron cryomicroscopy (20).
Despite the functional biochemical data concerning the interaction of TFIIH and U1 snRNA our current understanding of their association in vivo is limited. Determining how they interact in vivo will require functional analysis.
Here we have shown by footprinting assay that stem-loop 1 of U1 snRNA is implicated in interaction with cyclin H (Fig. 5). Interestingly this stem-loop has been shown to be involved in the recognition of U1 snRNA by the SMN complex (44). Importantly, analysis of the reconstituted re-initiation scaffold, after a U1-dependent second round of efficient transcription (16), reveals recruitment of the U1-70K subunit of U1 snRNP. 4 The U1-70K subunit has also been identified in yeast two-hybrid analysis as a partner of Ioc4, which is responsible for release of RNA Polymerase II into transcription elongation (46).
Preliminary analysis of TFIIH 3Ј-5Ј helicase activity, essential in promoter opening during initiation (4,(47)(48)(49), indicates that recruitment of U1 snRNP components, via U1 snRNA, can stimulate the helicase activity. 4 We are interested in analyzing the step-by-step assembly of the TFIIH-U1 based complex in vivo. In our initial study we have isolated by conventional purification the most stable and abundant form of TFIIH associated with U1 snRNA, potentially missing important co-factors. Of particular interest to us is the interaction with the SMN complex, which has also been shown to functionally interact with RNA Polymerase II (50). We are currently employing electrospray ionization mass spectrometry in the analysis of the components of the U1 snRNP that could be recruited and retained in the transcriptional pre-initiation and re-initiation complexes via the TFIIH interaction with U1 snRNA.