Portal de Periódicos da CAPES

Invariance of the zinc finger module: A comparison of the free structure with those in nucleic‐acid complexes

2007; Wiley; Volume: 67; Issue: 2 Linguagem: Inglês

10.1002/prot.21289

1097-0134

Duo Lü, Aaron Klug,

RNA modifications and cancer

Since the classical C2H2 zinc finger motif was predicted in 1985,1 3% of the known and predicted genes in the human genome have been found encoding zinc finger proteins.2 Many structures have been reported of zinc finger–DNA interactions and show a common feature of zinc finger–DNA recognition. Zinc fingers read out the sequence information from the DNA major groove by forming specific hydrogen bonds between certain residues on the α-helix of zinc finger motif and the DNA bases.3 Most commonly, as in the canonical structure, each finger recognizes four bases on two overlapping three base pair subsites, but there are variants found in special cases.3 Owing to the zinc finger protein binding, the DNA is slightly unwound and has 11.3 bp/turn.4 The TFIIIA-5S RNA system has been extensively studied for zinc finger–nucleic acids recognition. TFIIIA (Transcription Factor IIIA) binds to the internal control region of 5S RNA gene (5S DNA) as the first step in the transcription carried out by RNA polymerase III. Its DNA binding domain is composed of nine tandem linked zinc finger motifs located at the N-terminal end.1 TFIIIA binds not only the 5S DNA but also the 5S RNA, the product of the gene. Biochemical data suggested that the central fingers 4–6 contribute most to the TFIIIA-5S RNA interaction.5 The crystal structure of a complex of 61 bases truncated 5S RNA, embodying the central region of 5S RNA, interacting with Fingers 4, 5, and 6, revealed two modes of recognition.6 In the case of Finger 5, the zinc finger module recognizes a region of regularly folded RNA double helix through its phosphate backbone. The other two fingers, Fingers 4 and 6 recognize specific bases presented accessibly by the intricately folded RNA loop structures. The general conclusion is that RNA is recognized by zinc fingers through a combination of its different kinds of structural elements. As well as having determined the crystal structure of the complex of the 61n RNA and the three finger peptide TFIIIA(4–6), we have also crystallized the peptide on its own and solved the structure at the resolution of 1.65 Å. This is reported here for the first time. Taking advantage of the high resolution, we here compare its individual finger modules with those in another high resolution structure and the complex of the Zif268 three finger peptide with its DNA binding site at 1.6 Å resolution,7 to examine if there is any significant conformational change in the zinc finger module. The protein was expressed and purified as described in Ref. 8. The purified protein was dialyzed into the buffer containing 20 mM HEPES pH 7.0, 2 mM MgCl2, 150 mM KCl, 5 mM DTT, 0.3 mM ZnSO4, and was then concentrated in a vivaspin unit. Protein concentrated to 7 mg/mL was used for hanging drop equilibrium evaporation. The crystals were grown in a buffer solution of 1.6M (NH4)2SO4, 25 mM MgSO4, 50 mM Tris·HCl pH 8.5. For X-ray diffraction, crystals were frozen in the presence of 25% ethylene glycol. The crystallographic data and analysis are summarized in Table I. The MAD data sets were collected at the ID29 ESRF at a resolution of 2.0 Å. The 1.65-Å data set was collected on the same beam line. The space group is P6122 with a unit cell of a = b = 50.95 Å and c = 173.43 Å. The heavy atom sites were found by Shelx,9 and refined by Sharp.10 The model was built using ARP/wARP followed by some manual modification with O,11 and then refined in Refmac5 to a Rfree = 0.23, Rwork = 0.22. The comparisons of structures were done with the Superpose Molecules program in the CCP4 package.12 Model figures were made in Ribbons.13 Coordinates are deposited in the Protein Data Bank, access code 2J7J. To examine the conformational flexibility of the overall zinc finger motif, we superimposed the fingers in our new structure onto those in an earlier high resolution structure, the Zif268–DNA complex at 1.6 Å.7 The fingers in the two peptides vary in their sequences (Fig. 1) and can be subdivided into three groups. (i) Fingers 4 and 5 in TFIIIA and Finger 1 in Zif268, which have four amino acids between the two Cys, can be grouped together; (ii) Fingers 2 and 3 in Zif268 form another group with only two amino acids between the two Cys; (iii) Finger 6 in TFIIIA is atypical in having five amino acids between the two Cys and four amino acids between the two His. We exclude Finger 6 from the comparison, and leave it for a separate discussion. Comparison of Fingers 4–6 of TFIIIA with the three fingers of Zif268. (a) Sequences of TFIIIA(4–6) and Zif268. (b) Superimposition of the structures of TFIIIA Finger 5 (purple) (this study) and Zif268 Finger 1 (cyan)7. The skeleton shows only the main chains and the two Cys, two His, the three conserved hydrophobic residues, and the Pro preceding the first hydrophobic residue. We first compare the fingers within the same peptide (Table II), and in this way set up the standard of similarity between fingers whether the nucleic acid is bound or not. In the comparison carried out between the subgroups listed earlier, the amino acids between the two Cys were omitted for their obvious differences in length. The results show that the RMSD difference between the finger modules is around 0.4–0.9 Å, no matter whether they bind the nucleic acid or not. These differences, 0.9 Å and less, can be attributed to the different sequences and imply no noticeable change in conformation. Next we compare the finger modules of the two peptides from the unbound TFIIIA and the Zif268–DNA complex. The RMSDs are listed in Table III. From the result, it is clear that the zinc finger modules have essentially the same conformation and do not undergo any further changes on nucleic acid binding. This is visualized in Figure 1(b). The new structure reported here is of the unbound form of TFIIIA(4–6). The structures of these three finger modules have been previously determined in the RNA bound form.6 To examine whether the interactions have induced any conformational changes in the standard zinc finger modules, we superimposed the bound fingers on those in this new structure. The RMSDs are listed in the Table IV. The structures of RNA bound forms were solved only to 3.1-Å resolution, so that the model building may not be as accurate as that for the 1.65-Å unbound form. Despite the resolution difference, the zinc finger modules do not show any noticeable conformational change. In contrast to the relatively rigid zinc finger module conformation, the linkers between fingers affect the conformation of the whole three-finger protein and are determined by the intermolecular interactions in the crystal. We compared the RNA bound and unbound forms of the TFIIIA(4–6) by superimposing them at Finger 5 (Fig. 2). Although the linker between Fingers 5 and 6 is only two amino acids long, it shows great flexibility. Thus Finger 6 in the two structures points in different directions as shown in Figure 2. Similarly, Finger 4 points in very different directions in two structures, and therefore the linker between Fingers 4 and 5 is flexible too. Flexibility of the linkers; two X-ray structures of the three-finger peptide, 4–6, from TFIIIA, aligned by superimposition on Finger 5. The unbound form is in purple and the RNA bound form in cyan. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.] These comparisons of 3D structures illustrate the flexibility of the linkers between the zinc finger modules, which was first established from 2D NMR studies of a two-finger peptide from the transcription factor SWI5.14 The two linkers in our structure are different in sequence from those in SWI5, but have the similar feature of flexibility. It was however shown in an earlier study of another zinc finger structure that the linker contributes to the affinity in the DNA–protein interaction15 but does not have a main role in the nucleic acid recognition. Among the nine fingers in the TFIIIA, the sequence of Finger 6 is atypical in that the standard location of the second of the three conserved hydrophobic residues1 (lying four amino acids after the second Cys) is not preserved (Fig. 1). An earlier NMR study showed a different arrangement of the hydrophobic residues in a similar case.16 From our structure we can indeed see that the hydrophobic residues take on an alternative arrangement. In the absence of the hydrophobic residue at the standard second position four amino acids away after the second Cys, the Phe (F172) residue located two amino acids after the second Cys stacks its aromatic ring on top of the first His. This atypical location of the second hydrophobic amino acid is found in several other zinc finger sequences, and Figure 3 therefore represents an alternative hydrophobic core arrangement. The atypical Finger 6. Only the main chain atoms are shown apart from the two Cys, two His, and the three hydrophobic residues. A green dashed line indicates the stacking between the second hydrophobic residue Phe and the first His. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.] The other obvious sequence differences between Finger 6 and the others in TFIIIA do not lead to any significant changes in conformation. The chain conformation of the four amino acids between the two histidines follows a distorted helical path. Similarly, the five amino acids between the two Cys make a larger loop, but do not disturb the two Cys positions for the tetrahedral zinc ion binding. In the course of this analysis, we have also re-examined the role of the first of the three large conserved hydrophobic residues (usually aromatic) originally noted by Miller et al.1 This residue is located two amino acid positions away on the N-terminal side of the first cysteine. When the first NMR and X-ray structures appeared, the whole residue was considered as part of the hydrophobic cluster stabilizing the interior of the finger structure. In our high resolution structure, we find that the Cβ atom of this residue does come close to the third hydrophobic residue, and so contributes to the hydrophobic cluster as originally postulated,14 but the aromatic part of the residue is now seen to have another role. In both Zif268 and F5 of our structure and in a number of other structures,17, 18 it is close in van der Waals distance to the well conserved proline19 three positions N-terminal to the first cysteine and also frequently to the backbone amide of the residue in helical position 1. There is thus a bridge across the 'top' of the finger [as can be seen at top left in Fig. 1(b)], which helps stabilize the compact structure of the finger. The relative rigidity of the finger can therefore be ascribed to three cross-struts linking the β-sheet to the recognition helix: first, the van der Waals bridge just mentioned; second, the interaction between the three conserved hydrophobic residues; third, and of course the foremost, the ligation by the zinc ion of the two cysteines and two histidines, which pins the two secondary structure elements together. It is therefore not surprising that the zinc finger module resists cleavage by proteases, which cut between the modules of TFIIIA, as shown originally by Miller et al.1 and later by Nakaseko et al.,14 for a different zinc finger sequence.