Portal de Periódicos da CAPES

Structure-function analysis of the Z-DNA-binding domain Zalpha of dsRNA adenosine deaminase type I reveals similarity to the (alpha +beta ) family of helix-turn-helix proteins

1999; Springer Nature; Volume: 18; Issue: 2 Linguagem: Inglês

10.1093/emboj/18.2.470

1460-2075

Markus Schade,

RNA Research and Splicing

Article15 January 1999free access Structure–function analysis of the Z-DNA-binding domain Zα of dsRNA adenosine deaminase type I reveals similarity to the (α + β) family of helix–turn–helix proteins Markus Schade Markus Schade Department of Biology, Massachusetts Institute of Technology, Cambridge, MA, 02139 USA Forschungsinstitut für Molekulare Pharmakologie, 10315 Berlin, Germany Search for more papers by this author Christopher J. Turner Christopher J. Turner Francis Bitter Magnet Laboratory, Massachusetts Institute of Technology, Cambridge, MA, 02139 USA Search for more papers by this author Ky Lowenhaupt Ky Lowenhaupt Department of Biology, Massachusetts Institute of Technology, Cambridge, MA, 02139 USA Search for more papers by this author Alexander Rich Corresponding Author Alexander Rich Department of Biology, Massachusetts Institute of Technology, Cambridge, MA, 02139 USA Search for more papers by this author Alan Herbert Alan Herbert Department of Biology, Massachusetts Institute of Technology, Cambridge, MA, 02139 USA Search for more papers by this author Markus Schade Markus Schade Department of Biology, Massachusetts Institute of Technology, Cambridge, MA, 02139 USA Forschungsinstitut für Molekulare Pharmakologie, 10315 Berlin, Germany Search for more papers by this author Christopher J. Turner Christopher J. Turner Francis Bitter Magnet Laboratory, Massachusetts Institute of Technology, Cambridge, MA, 02139 USA Search for more papers by this author Ky Lowenhaupt Ky Lowenhaupt Department of Biology, Massachusetts Institute of Technology, Cambridge, MA, 02139 USA Search for more papers by this author Alexander Rich Corresponding Author Alexander Rich Department of Biology, Massachusetts Institute of Technology, Cambridge, MA, 02139 USA Search for more papers by this author Alan Herbert Alan Herbert Department of Biology, Massachusetts Institute of Technology, Cambridge, MA, 02139 USA Search for more papers by this author Author Information Markus Schade1,2, Christopher J. Turner3, Ky Lowenhaupt1, Alexander Rich 1 and Alan Herbert1 1Department of Biology, Massachusetts Institute of Technology, Cambridge, MA, 02139 USA 2Forschungsinstitut für Molekulare Pharmakologie, 10315 Berlin, Germany 3Francis Bitter Magnet Laboratory, Massachusetts Institute of Technology, Cambridge, MA, 02139 USA *Corresponding author. E-mail: [email protected] The EMBO Journal (1999)18:470-479https://doi.org/10.1093/emboj/18.2.470 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info RNA editing alters pre-mRNA through site-selective adenosine deamination, which results in codon changes that lead to the production of novel proteins. An enzyme that catalyzes this reaction, double-stranded RNA adenosine deaminase (ADAR1), contains two N-terminal Z-DNA-binding motifs, Zα and Zβ, the function of which is as yet unknown. In this study, multidimensional NMR spectroscopy was used to show that the topology of Zα is α1β1α2α3β2β3. Long-range NOEs indicate that β1 and β3 interact with each other. Site-directed mutagenesis was used to identify residues in α3, β3 and the loop connecting β2 to β3 that affect Z-DNA binding. Also identified were 11 hydrophobic residues that are essential for protein stability. Comparison with known structures reveals some similarity between Zα and (α + β) helix–turn–helix proteins, such as histone 5 and the family of hepatocyte nuclear factor-3 winged-helix–turn–helix transcription factors. Taken together, the structural and functional data suggest that recognition of Z-DNA by Zα involves residues in both the α3 helix and the C-terminal β-sheet. Introduction RNA editing modifies the linear flow of information from DNA to protein (Bass, 1993). It was discovered in the mitochondria of trypanosomes, where uridines are inserted in, or deleted from, certain RNAs (Benne et al., 1986). Numerous other examples of RNA editing, where one base is substituted for another, have been found in organisms of all eukaryotic phyla (Chen et al., 1987; Covello and Gray, 1989; Hoch et al., 1991) and their viruses (Polson et al., 1996). In one type of substitution RNA editing, adenosine is deaminated to inosine in a site-specific manner. Inosine is translated as guanosine. This process is used to modulate the activity of neural glutamate (Sommer et al., 1991) and serotonin receptors (Burns et al., 1997) and is altered during disease (Brusa et al., 1995). In these cases, editing is guided by an intron that folds back onto an exon to form a double-stranded RNA (dsRNA) editing substrate (reviewed in Maas et al., 1996). The use of introns in editing reactions requires that RNA modification must occur before pre-mRNAs are spliced. The site-specific editing of RNA is catalyzed both in vitro and in vivo by ADAR1 and ADAR2 (dsRNA adenosine deaminases type I and II) (Melcher et al., 1996). However, both ADAR1 and ADAR2 can be quite non-specific in their action. In vitro, any dsRNA longer than 30 bp can act as a substrate (Hurst et al., 1995; Maas et al., 1996), suggesting that auxiliary factors may normally be required to regulate activity. Consequently, the mechanisms for site-selection and regulation of editing efficacy in vivo are yet to be determined (Herbert, 1996). At the N-terminus, ADAR1 contains two left-handed DNA (Z-DNA)-binding motifs: Zα and Zβ (Herbert et al., 1997). These domains are conserved among the human, rat, bovine and frog enzymes and are predicted to belong to the helix–turn–helix (HTH) family of proteins, most likely the winged-helix variety (Herbert et al., 1997). The high affinity of Zα for linear poly(dCdG) stabilized in the Z-DNA conformation by chemical bromination and for supercoiled plasmids containing Z-DNA inserts (Herbert et al., 1997) links ADAR1-mediated RNA editing to Z-DNA. A possible role for Z-DNA may be to target editing by ADAR1 to sites of active transcription. Z-DNA can form in vivo in segments of alternating purine–pyrimidine sequences under conditions of torsional stress resulting from the movement of RNA polymerase through a gene (Singleton et al., 1982; Haniford and Pulleybank, 1983; Liu and Wang, 1987). Studies in both prokaryotes and eukaryotes show that Z-DNA formation in vivo is initiated by transcription (Lipps et al., 1983). Anti-Z-DNA antibody cross-linking experiments in metabolically active, agarose-embedded nuclei revealed transcription-dependent Z-DNA formation in segments of the c-myc gene (Wittig et al., 1992). Recognition of Z-DNA thus may allow ADAR1 to associate with pre-mRNA prior to splicing. Potential Z-DNA-forming sequences are present in many genes, predominantly in the 5′ regions where generation of negative superhelicity is expected to be greatest (Schroth et al., 1992). Z-DNA formation by these DNA segments may be one way of modulating the site selectivity and editing efficiency of ADAR1 in vivo (Herbert and Rich, 1996). In order to gain a molecular understanding of how the regulation of ADAR1-mediated RNA editing relates to the recognition of Z-DNA, we used a twin-tracked approach of structure/function analysis. We determined the secondary structure of Zα by multidimensional NMR spectroscopy, and showed that the topology is α1β1α2α3β2β3. By mutagenesis, we found that residues that are pivotal for Z-DNA binding cluster on one side of α3, making it likely that they form Z-DNA contacts. In addition, we identified residues in α1, α2 and α3 that probably form a hydrophobic core essential for protein stability. Zα shows some structural similarity to the core fold of hepatocyte nuclear factor (HNF)-3-like winged-HTH DNA-binding proteins, in which α3 is used as a recognition helix. Results Secondary structure and topology Multidimensional nuclear magnetic resonance (NMR) spectroscopy is a powerful technique used to determine the atomic structure of proteins with a mol. wt of <35 kDa. Here we determined the secondary structure of human Zα (residues 134–200; 9.4 kDa) by an NMR approach, using 15N-labeled protein to record well-resolved three-dimensional (3D) 15N-edited NMR experiments. A set of 3D 15N-edited TOCSY-HSQC, HNHA and NOESY-HSQC experiments (Marion et al., 1989; Vuister and Bax, 1993) allowed complete backbone assignment and yielded a continuous network of short- and medium-range nuclear Overhauser effect (NOE) distance restraints. These restraints were sufficient to elucidate the secondary structure of residues 134–200. The high quality of the spectra is exemplified by 2D strips extracted from the 3D 15N-edited NOESY-HSQC spectrum (Figure 1), illustrating the sequential and secondary structure assignment process for residues I172–L179. NOE intensities show characteristic patterns for α-helices and β-strands (Figure 2) (Wuethrich, 1986), yielding three α-helices and three β-strands. Coupling constants of HN–Hα are consistent with these secondary structure assignments (Figure 2). Figure 1.Sequential and secondary structure assignment in α3. Two-dimensional strips of α3 residues extracted from a 3D 15N-edited NOESY-HSQC spectrum of Zα depict sequential NOEs between amide protons (marked by double arrows in the 7–9.5 p.p.m. region), sequential Hα(i)→NH(i+1) NOEs (solid arrows in the 3.4–5.3 p.p.m. Hα region) and medium range Hα(i)→NH(i+3) NOEs (dashed arrows in the 3.4–5.3 p.p.m. Hα region) characteristic for α-helices. Backbone 15N chemical shifts are shown above each residue label at the bottom. Download figure Download PowerPoint Figure 2.The secondary structure of human Zα. The sequence of Zα from residues 134–200 is shown. dNN and dαN represent backbone NH–NH and Hα–NH NOEs and are indicated by strong (thick bar), medium (intermediate bar) and weak (thin bar) intensity. dαN(i,i+3) and dαN(i,i+4) NOEs are characteristic for α-helices, and strong dαN(i,i+1) NOEs are diagnostic of β-strands. The line J(HNHA) shows the quantative analysis of backbone HN–Hα coupling constants. Coupling constants 8 Hz suggest an extended chain conformation (marked with an e). Those residues with a coupling constant between 6 and 8 Hz are marked with an r. Download figure Download PowerPoint Analysis of long-range NH–NH and Hα–NH NOE restraints in 3D 15N-edited NOESY-HSQC spectra (70, 150 and 250 ms mixing time) revealed that residues 185–189 in β2 and residues 194–198 in β3 form an antiparallel β-sheet at the C-terminus (Table I; Figure 4). Backbone NOE restraints observed between T156 in β1 and L194 and W195 in β3 indicate that β3 also contacts β1, making the β-sheet triple-stranded. Figure 3.Site-directed mutagenesis of human Zα. (A) Summary of mutants tested. Asterisks (*) indicate substantial (>80%) proteolytic degradation. (B) Southwestern assay. Autoradiogram of Z-DNA affinity assays, identifying those mutants that cause diminished Z-DNA binding. The panel is a collage of five different membranes. The lanes marked + are positive controls using wild-type Zα protein. (C) Western blot. Filters used for Southwestern assays were probed using antibodies to GST in order to quantitate protein loadings. Comparable band intensities show that equal amounts of full-length protein were assayed in (B) (Southwestern). Mutants sensitive to proteolysis, e.g. E152A, L176A and L179A, show a smearing downwards of the band, which results from overloading the SDS gel in order to ensure equal amounts of full-length protein. Because of the need to avoid extensive overloading of the gel, some mutants that exhibit severe proteolysis were excluded from the Southwestern. Download figure Download PowerPoint Table 1. Long-range NOEs in Zα A β-sheet NOE B Fold restraining NOE β2 β3 β1 β3 L185.Hα A198.NH T156.NH L194.Hα W195.H6 Y177.H2/6 Q186.NH A198.NH T156.NH W195.NH W195.H5/7 L176.Hδ2 Q186.NH I197.Hα W195.Hβ1/2 T157.NH W195.H2 A158.NH Q186.NH K196.NH A158.Hβ L176.Hδ2 K187.Hα K196.NH T156.NH E152.Hα E188.NH W195.Hα L165.NH F146.H2/6 E188.NH L194.NH L165.Hδ1/2 L147.NH The backbone NOE restraints observed in the C-terminal β-sheet and between β-strand β1 and β3 are summarized in A, e.g. the Hα of L185 and the backbone NH of A198 are connected through an NOE. B shows other unambiguous long-range NOEs that determine the tertiary fold, obtained from 3D 15N-separated NOESY-HSQC and 2D NOESY (D2O) build-up experiments. For each NH–NH NOE, two independent cross-peaks were observed. Additional unambiguous long-range NOE constraints from the 3D 15N-edited NOESY-HSQC spectra and also from a 2D NOESY build-up in D2O restrain the tertiary fold of Zα further (Table I; Figure 4). Close contact between α3 and β3 is shown by the well-resolved side chain NOEs from the aromatic ring protons of W195 to the methyl protons of L176 and also to the aromatic protons of Y177. In addition, α2 folds against α1 to produce NOEs between the amide proton of L165 and the H2,6 protons of F146, as well as between the methyl protons of L165 and the amide proton of L147. Zα thus consists of a triple-stranded β-sheet in which β3 folds back against α3 and the N-terminus of α2. Helix α1 folds against α2. These results provide a coarse view of the tertiary structure and suggest that Zα has a topology similar to (α + β) HTH proteins such as HNF-3γ (Clark et al., 1993) (Figure 4). Mutagenesis The NMR-derived topology was set in a functional context by site-directed mutagenesis, a technique allowing high-resolution mapping of protein interactions with DNA (Cunningham and Wells, 1989; Brown et al., 1994). In many cases, alanine was substituted for the native residue. Alanine substitutions remove side chains beyond the β carbon, thereby eliminating hydrophobic, salt-bridged and hydrogen-bridged side chain interactions, while maintaining the peptide backbone. In addition, other amino acids were used to replace alanines present in Zα, and also to test the effect of more conservative substitutions on Zα function. A total of 44 amino acids between residues 139 and 198 of the human Zα domain were systematically replaced using PCR-based site-directed mutagenesis (Figure 3A). Mutant proteins were expressed as GST fusion proteins in Escherichia coli and purified by affinity chromatography, allowing recovery of full-length and proteolysed mutant protein. Mutant proteins were resolved by SDS–PAGE, blotted to a nitrocellulose membrane and probed for Z-DNA binding with radioactively labeled Z-DNA (Southwestern assay) (Figure 3B). Only protein with a fully intact Zα domain bound Z-DNA (data not shown). Subsequent probing of the blot with anti-GST made it possible to measure the extent of C-terminal proteolytic degradation, an indirect measure of how a mutation affects protein stability in vivo (Figure 3C). Each mutant was also tested in a bandshift assay using the same DNA probe. The bandshift assay allows more accurate quantitation and can be used over a 100-fold greater dilution range than the Southwestern. Dissociation constants (Kds) for the most deleterious mutants were measured with a BIAcore instrument under equilibrium conditions using an immobilized Z-DNA ligand as a substrate. Concordant results between assays were found, except for some mutants discussed below. Figure 4.The topology and candidate Z-DNA contacts of Zα. The data show that the topology and location of candidate DNA contacts of Zα are in some respects similar to those of histone H5 and HNF-3γ. The positions of α-helices and β-strands are indicated with boxes connected by thick lines. Numbers correspond to amino acid residues. Long-range NOEs given in Table I are indicated with thin lines and show the interactions between the C-terminal β-sheet and the α-helices of Zα. By analogy with HNF-3, β1 is shown as antiparallel to β3 because their relative orientation could not be determined unambiguously with the NOEs currently known. Residues (K169, N173, Y177, K181) on the face of α3 that contacts Z-DNA are indicated, as are the extensive contacts made by W195. W195A was the most deleterious mutant characterized. Download figure Download PowerPoint Using the Southwestern assay, we identified four groups of mutants in Zα that have effects on Z-DNA binding and protein stability; first, a cluster of mutants that lie in the C-terminal β-sheet; secondly, a collection of mutants in α3 that alter Z-DNA binding but not protein stability; thirdly, a series of mutants that affect hydrophobic residues present in α1, α2 and α3 and disrupt protein stability; and fourthly, a number of mutants present in the turns between structural elements defined by NMR that reduce Z-DNA binding. A number of these mutants affect residues that are highly conserved between human, mouse, rat, bovine and Xenopus Zα sequences. Each set of mutants is discussed below. The C-terminal β-sheet shows two distinct sides The C-terminal β-sheet is composed of strands β2 and β3, connected by a four residue loop. The network of backbone NOEs between β2 and β3 shows that the side chains of K187, W195 and I197 are next to each other, projecting to one side of the β-sheet (side 1). The hydrophilic residues Q186, E188 and K196 project to the other side (side 2). Side chain NOEs show that side 1 of the β-sheet is in contact with α3. Mutations to residues on side 1 of the β-sheet have profound effects. W195A is the most deleterious mutation of all Zα mutants made, regardless of the assay. Binding was diminished beyond detection and substantial proteolytic degradation was observed. These results confirm an important role for W195 in stabilizing Zα, and are consistent with the extensive long-range NOEs demonstrated for this residue by NMR (Table I; Figure 4). The more conservative W195Y mutant improved protein stability, but still caused an ∼4-fold reduction in binding (bandshift assay, data not shown). In contrast, mutation of the other hydrophobic residue in β3, I197, to alanine had no effect on binding or protein stability. Changing the non-conserved residue K187 in β2 to alanine caused a mild but significant reduction in binding, raising the possibility that this flexible, positively charged side chain contacts DNA directly. When residues on side 2 of the β-sheet were mutated, it was found that K196A enhanced Z-DNA binding while Q186A and E188A had wild-type protein stability and Z-DNA binding. Thus it appears that residues on this side of the β-sheet do not affect protein stability, but can influence DNA binding indirectly, for example by modifying the dipole moment of the β-sheet. The loop (L1) between β2 and β3 contains two highly conserved residues: P192 and P193. Mutant P192A had significantly diminished binding, while P193A showed only a slight reduction. As prolines often bend the protein backbone to form loops, the effects of these proline mutants may arise indirectly by disrupting β-sheet folding crucial for interaction with Z-DNA. Thus, P192 could be important in stabilizing the fold while P193 only makes minor contributions. However, a direct interaction between these residues and Z-DNA is also possible and cannot be excluded by this study. Mutation of the non-conserved L1 residue, T191, to alanine also diminished binding, suggesting that the T191 side chain stabilizes loop structure, while L194G caused only a mild diminution in binding. The effects of mutating the other loop residue, G190, were not tested. Thus, of the four L1 residues mutated, all have some effect on DNA binding by Zα. α3 has characteristic properties of a recognition helix Comparing results from NMR studies and alaninescanning mutagenesis shows that a striking cluster of deleterious mutations occurs in α3 (residues 169–182). An α-helical net diagram (Figure 5) segmenting α3 into three faces is useful in interpreting these mutants. Face 1 contains charged and polar residues. Mutations to these residues diminish Z-DNA binding but do not significantly affect protein stability. Face 2 is composed of hydrophobic residues. Changes to these amino acids are poorly tolerated, and cause increased protein degradation within E.coli. Face 3 is occupied by residues that do not reduce DNA binding or affect protein stability. In some cases, mutation to these residues enhances Z-DNA binding. The three functionally distinct faces are discussed separately. Figure 5.α-helical net diagram of α3. The putative DNA-contacting face 1, the hydrophobic face 2 and the proposed solvent-exposed face 3 are colored dark gray, light gray and white, respectively. Mutations to K169, N173, Y177 and K181 affect DNA binding without affecting protein stability. Changes to I172, L176 and L179 greatly diminished protein stability. Both E171A and S178D increased Z-DNA binding. Download figure Download PowerPoint Residues K169, K170, N173, Y177 and K181 form face 1 (Figure 5). Residues K169, N173 and Y177 show a high degree of conservation among the human, rat, bovine and frog Zαs, suggesting that they have an important role in Zα function. Residue 181 (K in humans) is poorly conserved. In the Southwestern and bandshift assays, mutant N173A displays the most dramatic decrease in binding of all α3 mutations. Mutants K169A and Y177A show only a minor effect on binding in the Southwestern assay, but show significant loss of activity in the bandshift assay, while mutant K181A lacks activity in both the Southwestern and bandshift assays. CD spectra (data not shown) of these mutants are similar to wild-type, suggesting that secondary structure is also similar. When analyzed using BIAcore measurements, N173A reduces binding 160-fold, K169A 37-fold, Y177A 26-fold and K181A 4-fold, relative to the wild-type (Table II). The Kd for K181A is not reduced as much as would be predicted from bandshift and Southwestern assays. A possible reason for this discrepancy is suggested by inspection of the BIAcore sensorgram shown in Figure 6. During the dissociation phase of the measurement, K181A shows a faster drop in signal than wild-type protein. The rapid breakdown of protein–DNA complexes during electrophoresis may account for loss of bandshifting activity of K181A, while a similar breakdown during washing may explain the diminished binding seen in the Southwestern assay. The alignment of K169A, N173A, Y177A and K181A mutants in a row on one face of a helix is reminiscent of DNA contacts commonly found in the recognition helices of HTH DNA-binding proteins (Pabo and Sauer, 1992; Suzuki et al., 1995). Therefore, the face 1 residues are good candidates to form direct or water-mediated DNA contacts. Figure 6.BIAcore sensorgrams of wild-type and mutant K181A. For wild-type Zα (dashed lines), the curves represent from top to bottom injections of 100, 75, 50, 25 and 10 nM protein. For K181A, the curves (solid line) show injections of 100, 80, 60, 40 and 10 nM protein. Response units (RU) are an arbitrary measurement that is proportional to the mass of protein bound to Z-DNA polymer immobilized on the chip surface. The >2-fold lower steady-state response (plateau region) of mutant K181A reflects its lower affinity for Z-DNA. The rapid drop of response signal in the dissociation phase of K181A suggests a rapid breakdown of the DNA–protein complex. Download figure Download PowerPoint Table 2. Binding affinities and rates relative to wild-type Protein Relative Kd Wild-type 1 K169A 37 N173A 168 L176A 43 Y177A 26 K181A 4 P192A 13 W195A >10 000 Engrailed 286 The Kd was determined using BIAcore measurements to determine the amount of protein bound to a Z-DNA-coated chip surface after steady state was reached. Engrailed protein was used as a negative control for non-specific protein binding. The binding by Engrailed exceeded that of the most deleterious mutant W195A, for which a Kd could not be determined. Additional mutants were made to examine the nature of the interaction of K181 and Y177 with DNA. A K181R mutant binds indistinguishably from wild-type, suggesting that K181 recognizes Z-DNA in a flexible, non-restrained mode, e.g. through hydrogen bonds with the phosphate backbone. This result perhaps explains why K181 is replaced by a charged residue such as arginine or glutamic acid in other family members. A Y177K mutant was studied because the side chain of lysine is flexible enough to reach any putative hydrogen-bonding partner that interacts with the hydroxyl of tyrosine. Mutant Y177K bound better than Y177A, but still worse than wild-type, showing that lysine can mimic some but not all the binding contributions mediated by the tyrosine side chain. In addition to hydrogen bonding via its hydroxyl group, Y177 is also likely to make essential contacts through its rigid aromatic ring system. The importance of hydrophobic contacts is suggested by the substitution of Y177 with another hydrophobic residue, isoleucine, in Zβ. Taken together, the mutational analysis revealed that residues K169, N173, Y177 and K181 are pivotal to binding, suggesting that face 1 of α3 interacts directly with Z-DNA and thus is a recognition helix. The hydrophobic residues on face 2, I172, L176 and L179, are also highly conserved, with L176 being absolutely conserved between Zα and Zβ of all species. The result of mutating each of these residues was determined. A I172P mutation disrupted the protein fold so severely that no full-length protein could be recovered from E.coli. A more conservative L176A mutation also caused extreme proteolytic degradation, allowing recovery of only 5% full-length protein. Even after correcting for the reduced yield of full-length protein, Z-DNA binding by this mutant was diminished 50-fold (Table II). Replacement of the third conserved residue, L179, with alanine results in slightly less protein degradation than L176A (∼10% is full-length protein) and Z-DNA binding assayed by Southwestern and bandshift assays was somewhat better. These results show that mutation of hydrophobic residues on face 2 of α3 causes tremendous proteolytic degradation, suggesting that they form part of a hydrophobic protein core stabilizing Zα. Face 3 of α3 carries three charged and one polar amino acid, none of which are conserved. Alanine mutants of the two positively charged residues R174 and K182 were indistinguishable from wild-type. The mutants E171A and S178D revealed a slightly enhanced Z-DNA affinity. Thus, these four mutants on face 3 are dispensable for Z-DNA binding and are probably solvent exposed. α1 and α2 have essential hydrophobic faces Mutation of α1 and α2 revealed conserved residues that form a hydrophobic face on each helix. In α1, the mutations I143A and L144T disrupt both binding and protein stability, while the L150A mutation increases proteolytic degradation without affecting binding. Together with L147, which was not investigated, these three residues make up a hydrophobic face on α1, suggesting that they are involved in contacts that correctly align this face within the protein. In contrast, mutation of α1 residues that lie on other faces of the helix, such as R142A, K145A and E148A, had no effect. A similar face of hydrophobic residues is present in α2. Mutants A158L, L161G and L165P were found to be deleterious to both binding and protein stability. Replacement of A158 with a bulky leucine residue caused a milder phenotype than the severely destructive L161G and L165P mutations, suggesting that the environment of A158 at the joint of β1 and α2 is less constrained than that of the two leucines. In summary, the hydrophobic faces of α1 and α2 probably contribute to a hydrophobic core essential for correct folding of Zα, similar to that found in other HTH proteins (Pabo and Sauer, 1992; Suzuki et al., 1995). Essential turn residues Another class of mutants maps to the turns deduced from the NMR secondary structure of Zα. T1 connects α1 and β1 and contains residues E152 and K154. Mutant E152A clearly has diminished binding and protein stability, suggesting that contacts formed by the glutamate side chain are essential for correct folding of Zα. NOEs that show contact between this residue and T156, which lies in β1 (Table I; Figure 4), are consistent with such a role for E152. The mutant K154A binds Z-DNA almost as tightly as wild-type. T2, between α2 and α3, consists of residues G166, T167 and P168, of which only the latter showed a minimal reduction in binding upon mutation to alanine, indicating that the main chain bend of P168 makes some contribution to the turn. Direct or water-mediated contact of these side chain residues with DNA, as seen with other (α + β) HTH proteins (Kaufmann and Knoechel, 1996), is unlikely. T3 residues G183 and K184 link the antiparallel β-sheet to α3. G183 is absolutely conserved in Zα and Zβ, while K184 is variable. The NMR data suggest that T3 must form a sharp turn to enable the β-sheet to fold back against α3. Mutagenesis supports this suggestion. The introduction of a bulky leucine residue, which reduces the main chain flexibility, in place of G183 diminished Z-DNA binding substantially. Circular dichroism (CD) analysis of Y177A and K181A To understand further the effect of Y177 and K181 on the interaction of Zα with Z-DNA, CD experiments were performed to test the effects of these mutants on the B–Z transition of poly(dCdG). The spectra collected after 10 min of incubation at 30°C for wild-type, K181A and Y177A at a molar ratio of one peptide to one base pair are shown in Figure 7. Also shown is the reference spectrum for the polymer in the B-DNA conformation obtained in buffer without