Structural Characterization of the Cysteine-rich Domain of TFIIH p44 Subunit*

In an effort to understand the structure function relationship of TFIIH, a transcription/repair factor, we focused our attention on the p44 subunit, which plays a central role in both mechanisms. The amino-terminal portion of p44 has been shown to be involved in the regulation of the XPD helicase activity; here we show that its carboxyl-terminal domain is essential for TFIIH transcription activity and that it binds three zinc atoms through two independent modules. The first contains a C4 zinc finger motif, whereas the second is characterized by a CX 2CX 2–4FCADCD motif, corresponding to interleaved zinc binding sites. The solution structure of this second module reveals an unexpected homology with the regulatory domain of protein kinase C and provides a framework to study its role at the molecular level.

In an effort to understand the structure function relationship of TFIIH, a transcription/repair factor, we focused our attention on the p44 subunit, which plays a central role in both mechanisms. The amino-terminal portion of p44 has been shown to be involved in the regulation of the XPD helicase activity; here we show that its carboxyl-terminal domain is essential for TFIIH transcription activity and that it binds three zinc atoms through two independent modules. The first contains a C4 zinc finger motif, whereas the second is characterized by a CX 2 CX 2-4 FCADCD motif, corresponding to interleaved zinc binding sites. The solution structure of this second module reveals an unexpected homology with the regulatory domain of protein kinase C and provides a framework to study its role at the molecular level.
Transcription of protein coding genes in eukaryotes requires the formation at core promoters of a multiprotein complex composed of RNA polymerase II and the general transcription factors TFIIA, -IIB, -IID, -IIE, -IIF, and -IIH (reviewed in Refs. [1][2][3]. TFIIH, which possesses nine subunits, plays a critical role in transcription. CAK (the Cdk-activating kinase complex, composed of cdk7, cyclinH, and MAT1) phosphorylates the carboxyl-terminal domain of the RNA polymerase II largest subunit (4); the two DNA helicases, XPB and XPD, are involved in the opening of the DNA template around the start site to allow transcription by RNA polymerase II (5,6). The TFIIH role in nucleotide excision repair, a DNA repair pathway essential for maintaining the integrity of the genome, is also firmly established (7). Mutations in either XPB or XPD result in DNA repair defects, one of the phenotypes found in three genetic disorders: xeroderma pigmentosum (XP), Cockayne syndrome, or trichothiodystrophy (8,9).The roles of the four other TFIIH subunits, p62, p52, p44, and p34, are not well understood.
Of these four subunits of TFIIH, p44 (10) is the most characterized. Its yeast homologue, Ssl1, is thought to nucleate the formation of the core TFIIH with TFB1, the yeast counterpart of p62 (11,12). The human p44 protein interacts with human p34, XPD, XPB (13)(14)(15), as well as CSA, a WD repeat-containing protein (16). The p44 amino terminus shares sequence homology with S5a, a proteasome subunit, leading to the proposal that this homologous region could be a ubiquitous regulating motif for both transcription and translation (17). Additionally, the p44 subunit possesses a carboxyl-terminal cysteine-rich region (residues 252 to 395) (10). Its sequence analysis reveals relationships to three different zinc binding motifs: a C4 zinc binding motif followed by either a TFIIIA-like zinc finger or a RING-related sequence (see Ref. 18 and Fig. 2).
Here we demonstrate through biochemical, biophysical, and structural studies that the carboxyl-terminal portion of p44 is essential for the integrity of TFIIH transcriptional activity. This role is supported by two distinct zinc binding modules belonging to the C4 zinc finger family and to the interleaved zinc-binding protein family. The solution structure of this latter module reveals an unexpected homology with the regulatory domain of protein kinase C and provides a framework to study its role at the molecular level.
Expression and Purification of Recombinant Proteins-Domains of the human p44 TFIIH subunit with either the wild type or a mutated amino acid sequence were expressed in the Escherichia coli BL21 (DE3) strain as His-tagged or GST fusion proteins from plasmids pET-15b-p44(252-395) or pGEX-p44(252-320) and pGEX-p44(321-395). In constructing these expression vectors, fragments were generated by polymerase chain reaction and introduced into the plasmids p15b (Novagen) and pGEX-NB, a modified version of pGEX-4T2 (Amersham Pharmacia Biotech), using the NdeI and BamHI restriction sites. Cells were grown at 37°C to an A 600 of 0.6, induced with 0.4 mM isopropyl-␤-D-thiogalactopyranoside, and harvested after 4 h of incubation at 25°C. Cells, resuspended in 20 mM Tris-HCl, pH 8, 100 mM NaCl, 5 mM 2-mercaptoethanol, were disrupted by sonication and centrifuged for 1 h at 4°C at 100,000 ϫ g. The p44(252-395) protein was purified with the TALON (CLONTECH TM ) cobalt affinity resin using a 10 to 250 mM imidazole gradient. The GST-p44(252-320) and GST-p44(321-395) proteins were purified using GSH-Sepharose (Amersham Pharmacia Biotech), and the GST fusion protein was cleaved with bovine thrombin (0.25 units/mg of fusion protein). All recombinant proteins were finally subjected to gel filtration chromatography using a Superdex 75 26/60 column (Amersham Pharmacia Biotech) in 20 mM Tris-HCl, pH 8, 50 mM NaCl, 5 mM 2-mercaptoethanol.
Atomic Absorption Spectroscopy-Protein samples were dialyzed against 20 mM Tris-HCl, pH 8, 250 mM NaCl, 5 mM 2-mercaptoethanol, 0.1 mM EDTA at 4°C for 16 h. Measurement of metal content was performed on a Varian AA75 spectrophotometer at the appropriate wavelength and deduced from a standard calibration.
Histidine Modification-Protein in 100 mM sodium phosphate, pH 7, was modified with a 180-fold excess of diethylpyrocarbonate (Sigma-Aldrich). The kinetics of modification were followed at 240 nm for 60 min at room temperature.
Mass Spectrometry Measurements-Before analysis, protein samples at 1 mg/ml were dialyzed for 16 h against 20 mM ammonium acetate, pH 7.4, 5 mM 2-mercaptoethanol. Measurements were performed on a VG-BioQ triple quadrupole mass spectrometer (Micromass) upgraded so that the electrospray ionization source has Quattro II performances. In all this work, ions were produced in the positive ion mode and were detected at the exit of the first analyzer (see Ref. 21 for a detailed description of the experimental setting). For measurement under denaturing conditions, p44 fragments were diluted to 5 M in a 1:1 wateracetonitrile mixture (v/v) containing 1% of formic acid. Samples were introduced into the ion source via a 10-l syringe loading injector at a flow rate of 5 l/min, and spectra were acquired from m/z 500 to m/z 1500. For measurements under non-denaturing conditions, proteins were diluted to 10 M in 25 mM ammonium acetate, pH 7.4, and mass data were recorded from m/z 1000 to m/z 3000.
NMR Spectroscopy and Calculations-The unlabeled and uniformly 15 N-labeled human p44(321-395) proteins were produced by growing BL21 (DE3) bacteria in LB or Bioexpress CGM-100-N (Cambridge Isotope Laboratories, Inc.) medium, respectively. Proteins were purified as described above, concentrated to 1-2 mM, and dialyzed against 50 mM deuterated Tris-HCl (pH 6.8 at 25°C), 20 mM NaCl, 1 mM dithiothreitol in H 2 O/ 2 H 2 O (9:1) or in 2 H 2 O. No addition of zinc was required throughout the sample preparation to obtain the native protein with two bound zinc atoms. Despite our efforts we could not obtain a 113 Cd-substituted sample suitable for NMR analysis.
Measurements were carried out on BRUKER AMX-500, DRX-600, and DRX-800 spectrometers. Two-dimensional experiments DFQ COSY, TOCSY (20-and 60-ms mixing time, DIPSI composite pulse cycle), and NOESY experiments at 30°C allowed resonance assignment. The line broadening of His-376 and His-380 protons did not allow determination of their protonation state using long range 15 N heteronuclear single quantum coherence. Slowly exchanging amide protons were identified in a series of NOESY experiments recorded after dissolution in 2 H 2 O.
NOE Assignment-After a first analysis of the NOESY spectra, 366 NOEs were unambiguously assigned. This set of NOEs was subsequently used to drive the iterative assignment of the full set of NOE cross-peaks using the ARIA (MSI Inc., San Diego, CA (22)) procedure. Peak picking was performed manually on NOESY spectra recorded at 30°C and 20°C and on a NOESY 15 N heteronuclear single quantum coherence. Cross-peak volumes were integrated using XEASY program (23). Nine ARIA iterations allowed the unambiguous assignment of 1242 NOE cross-peaks; 166 NOEs cross-peaks retained their ambiguous assignment at the end of the procedure. Inter-proton distance restraints were obtained by classifying peak volumes as strong, medium, weak, and very weak, corresponding to distances of 2.5, 3.2, 4.5, and 5 Å respectively. The standard pseudo-atom corrections were applied.
Structure Calculations-Two runs of structure calculation were performed using the restrained simulated annealing protocol implemented in the program X-PLOR 3.851 (24,25). A first set of 50 structures was generated using only NOE distances as experimental restraints. This set of structures allowed the identification of secondary structure elements, hydrogen bonds acceptors compatible with slowly exchanging amide protons and the four cysteines involved in the first zinc coordination site (ZNI). For the second zinc binding site (ZNII), six possible coordinating ligands (Cys-360, Cys-363, His-376, His-380, Cys-382, and Cys-385) were located in the vicinity of the zinc ion. This set of structures pointed out few inconsistencies in the NOE-derived distances between ZNII-possible ligands. Two experimentally observed NOE distances between cysteine residues Cys-360 and Cys-363 and residues Cys-382 and Cys-385 had to be discarded to build a Cys-2-His-2 coordination scheme that was based on eight long range NOEs between residues Cys-360 and Cys-363 and residues His-376 and His-380.
A second set of 50 structures was then generated using 1082 NOE distance restraints, 12 hydrogen bond constraints, and 6 S␥-Zn distance restraints. The structures were further refined using 25 dihedral angle restraints and geometric constraints around zinc according to tetrahe-FIG. 1. Transcriptional (A) and helicase (B) activities of endogenous and recombinant TFIIH, either wild type or harboring deletion in the carboxyl terminus of the p44 subunit. Equivalent amounts of recombinant TFIIH complexes, either wild type (rIIH9-wt) or harboring deletion in carboxyl terminus of the p44 subunit (rIIH9-p44⌬(252-395), rIIH9-p44⌬(321-395)) were tested for their ability to allow the transcription of a 309 nucleotide sequence under the control of adenovirus major late promoter or displace an oligonucleotide. Endogenous TFIIH, purified from HeLa cell extracts (HeLa TFIIH) was used as a positive control. nt, nucleotides.
dral coordination values (26). From this latter set, 23 structures were then chosen based on the selection criteria: low total energy, no NOE violations greater than 0.5 Å, and no angle violations greater than 5°.
Sequence and Structure Analysis-General data base searches were performed with the PSI-BLAST program using default parameters (27). Multiple alignments were generated with ClustalX (28) and optimized manually with the Seqlab editor of the Wisconsin package from the Genetics Computer Group (Madison, WI) (29).
The search for proteins structurally related to p44(321-395) was performed using Dali (30) and Sarf2 (31). The best scoring entries were superimposed using lsq-man (32) and analyzed individually using the program O (32). Figures were prepared using DINO, 2 Grasp (34), and Pov-ray. 3 The Protein Data Bank codes for the proteins discussed in the text are listed below together with the number of equivalent C␣ atoms and the root mean square deviation. PKC cysteine-rich domain (1ptr)

RESULTS AND DISCUSSION
The amino-terminal portion of the p44 subunit of TFIIH is involved in the regulation of XPD helicase activity (15). To investigate the function of the carboxyl-terminal portion of p44, we generated, based on limited proteolysis, two recombinant TFIIH complexes in which the p44 subunit was deleted. In parallel, the corresponding carboxyl-terminal domains were cloned, expressed, and purified for biophysical and structural studies.
The p44 Carboxyl-terminal Domain Is Essential for Transcription Activity-Two recombinant TFIIH complexes in which the p44 subunit was deleted at its carboxyl-terminal end from amino acid 252-395 (p44⌬(252-395)) and from 321-395 (p44⌬(321-395)) were generated. After being overexpressed in the baculovirus expression system, rIIH9-wt, rIIH9-p44⌬(252-395) and rIIH9-p44⌬(321-395) were purified by immunoprecipitation using a monoclonal antibody directed toward the amino-terminal end of p44. The composition of rIIH9-p44⌬(321-395) complex assayed by Western blotting was similar to that of rIIH9-wt, whereas the p34 subunit was missing in rIIH9-p44⌬(252-395) (data not shown). These complexes were tested in an in vitro transcription system containing all the basal transcription factors in addition to the RNA polymerase II and the adenovirus major late promoter template. Both rIIH9-p44⌬(252-395) and rIIH9-p44⌬(321-395) are deficient in transcription activity (Fig. 1A, compare lanes 6 -11 with lanes 3-5) but keep the same XPD helicase activity as rIIH9-wt (Fig. 1B). The deficient TFIIH transcription activity of rIIH9-p44⌬(321-395) is due to the truncated p44 protein itself, since its subunit composition is similar to that of rIIH9-wt, whereas the deficient TFIIH transcription activity of rIIH9-p44⌬(252-395) is due to the absence of p34 subunit, as it was shown (19) that the absence of a single subunit abolishes TFIIH transcription activity. Together, these data demonstrate a crucial in- Conserved residues (more than 80% identity) are indicated in white with black shading, whereas conservatively substituted residues (more than 80% of the residues belong to the same class: Pro, Ala, Gly, Ser, Thr; Phe, Tyr, Trp, Ile, Leu, Met, Val; Asp, Glu, Gln, Asn; Arg, Lys, His; Cys) are drawn in white with gray shading. Above the alignment, conserved blocks of residues are delimited by dashed lines. Below the alignment, conserved histidines (q) and cysteines (*) discussed in the text are labeled. The p44 signature and the consensus motif of classical C4 zinc finger, TFIIIA zinc finger, and RING finger according to Klug and Schwabe (36) and Borden and Freemont (35)  The p44 Cysteine-rich Domain Binds Three Zinc Atoms-A sequence similarity search using human p44(252-395) as the query sequence retrieved eight homologous proteins that were aligned. As shown in Fig. 2, 13 of the 14 cysteines (* residues) and 4 of the 9 histidines (q residues) present in the human p44 are conserved among all species from Homo sapiens to Saccha-  1, 2, and 3,  respectively). B-H, mass spectra of the p44 fragments under native (ammonium acetate) and denaturing conditions (HCOOH, formic acid). B and C, mass spectra of p44(252-395) recorded under denaturing (B) and native conditions (C). D and E, same as B and C for p44(252-320). F, G, and H, mass spectra acquired in native conditions of the same p44(321-395) sample before treatment (F), after EDTA treatment (G), or after EDTA treatment followed by back addition of an excess of zinc acetate (H). Peaks marked * and ࡗ correspond to sodium and 2-mercaptoethanol adducts, respectively. romyces cerevisiae. Two blocks of conservation can be defined: block I (Gly-288 to Leu-328) and block II (Cys-345 to Cys-385) are separated by an insertion of 12 to 23 residues. Block I contains five cysteines and one histidine and possesses a CX 2 CX 10 CX 2 C amino acid sequence that corresponds to a C4 zinc binding motif (36). Block II contains eight cysteines and two histidines and possesses a typical sequence CX 2 CX 2-4 FCADCD (A is an aliphatic residue) that is found only in human p44 and its orthologues. Block II also exhibits similarities with a consensus TFIIIA zinc finger (37) and with a derived C3HC4 RING finger motif (38) (Fig. 2).
To determine whether the p44 carboxyl-terminal portion binds a metal through the detected motifs, three expression vectors that enable the production of p44(252-395) (Blocks I and II) as well as p44(252-320) (Block I) and p44(321-395) (Block II) were constructed (Fig. 3A, lanes 2 and 3; see also "Experimental Procedures"). The p44(252-320) and p44(321-395) fragments were chosen on the basis of limited proteolysis of p44(252-395). The metal content of the purified recombinant polypeptides was analyzed. Atomic absorption measurements revealed the presence of zinc atoms within the protein fragments, whereas no trace of metal such as Ca 2ϩ , Ni 2ϩ , Co 2ϩ , Li ϩ , Mn 2ϩ , or Mg 2ϩ could be detected. The accurate zinc stoichiometry was determined from mass spectrometry measurements by comparing the mass measured for the denatured protein (Fig. 3, B and D) with that of the native one (Fig. 3, C and E) for p44(252-395), p44(252-320), p44(321-395) (see also Table I). Apart from the primary ion peaks, minor peaks belonging to sodium (Fig. 3, *) or 2-mercaptoethanol (Fig. 3, ࡗ) adducts were detected. The mass spectrometry data obtained under denaturing conditions fit the theoretical values. Mass spectrometry measurements of the native protein fragments show additional molecular masses of 191.1, 63.8, and 127.9 Da for p44(252-395), p44(252-320), and p44(321-395), respectively (Fig. 3, C, E, and F). These correspond to 3, 1, and 2 zinc atoms associated with the three fragments, respectively (Table I).
Upon treatment with EDTA, we observed that the 9015.9-Da protein peak (Fig. 3F) gives rise to two different species of 8888.0 Da and 8950.8 Da corresponding to the loss of two and one zinc atoms, respectively (Fig. 3, F and G). The addition of a large excess of zinc acetate restores the native state of the protein, leading to a species of 9015.7 Da, which corresponds to the back addition of the two zinc atoms (Fig. 3H). The zinc binding to the protein is specific since no additional species is observed. Together, these data demonstrate that p44 carboxylterminal domain contains three zinc atoms coordinated by two independent domains that correspond to p44(252-320) and p44(321-395).
In a second set of experiments, we overexpressed and purified p44 variants where conserved cysteines and histidines were mutated to alanine. Their metal content was then analyzed using mass spectrometry and through chemical modification experiments (Fig. 2 and Table II It was also found that the non-conserved Cys-381 did not substitute for zinc binding (data not shown), nor did the conserved His-324. Together, these data suggest that Cys-345, Cys-348, Cys-360, Cys-363, Cys-368, Cys-371, and at least one histidine (His-376 or His-380) are involved in the coordination sphere of the two zinc atoms in p44(321-395).
Three-dimensional Solution Structure of p44(321-395)-The three-dimensional structure of p44(321-395) was determined by NMR spectroscopy using both unlabeled and uniformly 15 Nlabeled protein ("Experimental Procedures"). Backbone and side chain protons are assigned for 72 residues out of the 75. One unique set of resonance is observed for all residues; how-  (252-320), and p44(321-395) Measurements were performed in ammonium acetate for native conditions and in acetonitrile for denaturing conditions. We calculated the ⌬ mass from denaturing and native condition measurements and divided by the molecular mass of zinc (33 4. Solution structure of p44(321-395). A, the number of short-range (i, iϩ1, cyan bars), medium range (i, iϩ (2)(3)(4), yellow bars), and long range (orange bars) inter-residue NOE-deduced distance restraints is represented as a function of residue number. The average root mean square deviation (Rmsd) value from the mean structure for 23 structures of p44(321-395) is indicated as a solid line. B, overlay of the C␣ backbones from the 23 simulated annealing structures in stereo for residues 328 to 386. The mean positions of the Zn 2ϩ atoms are represented by orange (ZNI) and blue (ZNII) sphere.C, ribbon diagram of the average solution structure. In the left panel, residues 328 to 375, which form the well defined core as well as side chains from residues involved in zinc chelation (yellow) and buried hydrophobic residues (green), are represented. Zinc atoms from ZNI and ZNII are drawn in orange and blue, respectively. In the right panel, the distribution of charges at surface of the molecule is illustrated for the face covered by the ␣-helix (conserved charged residues are labeled). No charge was assigned to zinc. The color scheme shows negatively charged atoms as red, positively charged atoms areas as blue, and neutral areas as white. Left and right panels represent 180°-rotated views of the molecule.
Cys-382 to Cys-385, respectively). This dynamic behavior affects the structure precision of the tail residues of p44(321-395) but not those located in the core of the structure (Fig. 4A). Structure calculations were performed using dynamic simulated annealing using 456 intra-residual, 382 medium, and 244 long range restraints (Fig. 4A) to generate a set of 50 structures, 23 being retained and shown in the stereo view ( Fig. 4B) (See "Experimental Procedures" and Table III).
The p44(321-395) core domain consists of a three-stranded anti-parallel ␤-sheet packed against a small ␣-helix, the carboxyl terminus being spatially close to the amino terminus (Fig. 4C, left). Residues Gln-332 to Ile-334, Val-357 to Val-359, and Val-366 to Cys-368 form strands ␤1, ␤2, and ␤3, respectively. One face of the ␤-sheet is partially covered by the ␣-helix, the other by a loop that begins with one turn of a 3 10 helix. The domain is stabilized by two small clusters of buried and partially buried hydrophobic residues located on both sides of the ␤-sheet. The first cluster includes the conserved aromatic residues Phe-331, Tyr-346, Tyr-358, and Phe-367, stabilizing the face covered by the ␣-helix. The second cluster located on the other face consists of Tyr-339, Val-357, Val-359, and Val-368. The well defined compact core of p44(321-395) domain partially exposes conserved hydrophobic residues including Phe-331 and Ile-334, located on ␤1. The three-dimensional distribution of charges at the protein surface reveals that the conserved negatively charged residues Glu-333, Asp-370, Asp-372, and Asp-377 are grouped on one side of the molecule (Fig. 4C, right). Two of them, Asp-370 and Asp-372, are located on the ␣-helix and are part of the conserved motif CX 2 CX 2-4 FCADCD that is a defining feature of p44.
As inferred from spatial localization of conserved cysteines and histidines, p44(321-395) possesses two independent zinc binding sites that are separated by 14 Å and located on the face of the ␤ sheet covered by the ␣-helix. The first zinc binding site (ZNI), organized into a C 4 type, is composed of the first pair and the third pair of conserved cysteines (Cys-345/Cys-348 and Cys-368/Cys-371). When mutated, these residues prevent zinc binding. The second zinc binding site (ZNII) involves the cysteine residues Cys-360 and Cys-363 together with His-376 and His-380, which are the best candidates for the zinc binding ligands on the basis of the observed NOE. The possible presence of a minor conformation involving one of the two cysteines Cys-382 or Cys-385 could not be discarded and may be responsible of the localized line broadening.
p44(321-395) and PKC Exhibit a Similar Three-dimensional Fold-A data base search for structural homologues of p44(321-395) using the DALI program revealed significant homologies (Z-score Ͼ 3) with other proteins that bind two zinc atoms: the cysteine-rich domain of protein kinase C (PKC) (39,40), Raf-1 (41), and the Vps27p FYVE domain (42). The domains of PKC and Vps27p are involved in binding phorbol ester and phosphatidylinositol 3-phosphate, respectively. The serine/ threonine kinases PKC and Raf-1 play a role in signal transduction; Vps27p is involved in membrane trafficking within the cell.
The p44(321-395) domain and the cysteine-rich domains of PKC or Raf-1 share the same topology with three anti-parallel ␤ strands covered by an ␣ helix (Fig. 5A). When their C␣ atoms are superimposed, both the secondary structure elements and the two zinc atoms match. The structural homology between p44(321-395) and PKC/Raf-1 spreads over 39 and 36 C␣ atoms with a root mean square deviation value of 1.41 and 1.94 Å, respectively; the ligand binding loop of PKC is not included. The homology between p44(321-395) and Vps27p includes the ␤2/␤3 hairpin and the ␤1/␤2 hairpin, respectively (Fig.  5B). This substructure also exists within RING finger domains, GATA-1, the second unit of LIM domains, and the first zinc binding site of nuclear receptors (43). All these domains have a ␤ hairpin that corresponds to the ␤2/␤3 strands of p44(321-395) as well as a zinc binding site equivalent to the ZNI site in p44(321-395). The p44(321-395) domain, the cysteine-rich domains of PKC/Raf-1, the Vps27p FYVE domain, and RING domains all possess two interleaved zinc binding sites separated by approximately 14 Å, which are built on a common scaffold. ZNI site of the p44(321-395) domain is conserved with fully superimposable zinc ligands, whereas the ZNII site remains specific for each protein family (Fig. 5B).
Conclusion-In the present study, we establish that the carboxyl-terminal cysteine-rich region of the p44 TFIIH subunit is necessary for transcription. This domain binds exactly three zinc atoms via specific cysteine and histidine residues through two independent modules. The first contains a C4 zinc binding motif, whereas the second is characterized by a CX 2 CX 2-4 FCADCD motif. In addition, the determined solution structure of the second module reveals homologies with interleaved zinc binding domains.
How is the cysteine-rich carboxyl-terminal domain of p44 involved in TFIIH transcriptional activity? Zinc binding motifs allow accurate folding of domains (43) whose functions include DNA binding (44), ligand binding (40,42,45), and proteinprotein interactions (46,47). The p44 protein is a subunit of TFIIH, a stable macromolecular assembly known to interact with both proteins and DNA. The charge distribution of p44(321-395) rules out a possible role in the recognition of DNA via the ␣-helix as observed in the structurally related GATA-1 zinc fingers. The conserved solvent-accessible hydrophobic conserved residues in p44(321-395) three-dimensional structure suggest a role in protein-protein interactions. This is supported by the fact that the p44 subunit likely nucleates the formation of TFIIH complex through interaction with p34, p62, XPB (14), and XPD (15) and that the p44 carboxyl-terminal region was also found to interact with SCL, a hematopoietic transcription factor containing a basic helix-turn-helix motif (48). Ϫ14.9 Ϯ 28.3