DNA:Protein Interaction Thermodynamics Part 2 (Molecular Biology)

5. Conformational Changes of the DNA upon Protein Binding

While many proteins recognize regular B-DNA through the formation of a number of hydrogen bonds and van der Waals interactions between amino acid side chains and functional groups of the bases, it has long been recognized in many cases that DNA can adopt a bent conformation when bound to a protein (121). A dramatic example of protein-induced DNA bending is provided by the structure of integration host factor (IHF) bound to the H ‘ -site of phage l, in which the 34-bp piece of DNA is literally wrapped around the protein, creating a buried protein-DNA interface of 4600 A2 (Fig. 3) (122, 123). The DNA is bent by more than 160°, thereby almost completely reversing the direction of the DNA within a short distance.

Figure 3. Van der Waals representation of the IHF-DNA complex (122). The drawing illustrates both the massive DNA bend induced by the protein and the large contact surface of ~ 4600 A2 between protein and DNA. The DNA sequence is indicated.

Other examples of strong protein-induced bends are provided by the structures of the DNA complexes of the purine repressor PurR (124), the sex-determining protein SRY (125), the pre-B-and T-lymphocyte-specific factor LEF-1 (126), and the human oncogene product ETS1 (127). In none of the above cases would it have been possible to predict the dramatic bends observed in the complexes from simple inspection of the DNA sequence.

Conversely, studies with 434 repressor indicated that its DNA-binding affinity was modulated by the flexibility of the noncontacted nucleobases at the center of the binding site (128, 129). A linear relationship was found between the free energy of the association reaction between 434 repressor and operator mutants and the flexibility of their central sequences (130). Similar results had been obtained for the Cro repressor from bacteriophage l (131). These studies indicated for the first time that intrinsic sequence-dependent properties are important for the formation of DNA-protein complexes.

Sequence-dependent bending of double-stranded DNA often occurs at the junctions between regions of G-C and A-T base pairs. DNA bending of 10° to 20° has been observed in the crystal structures of oligonucleotides containing an AT-core (132). While such sequences can be bent, however, they are not necessarily bent. The transition from G-C to A-T base pairs renders this region of the DNA flexible and capable of potentially undergoing a bend. Such facultative bending is illustrated by the X-ray structure of the dodecamer d(CGCGAATTCGCG) (Fig. 4), in which a bend of 18° occurs at the GC/AT junction at one end of the helix, while the helix remains unbent at the chemically equivalent junction at the other end of the helix (133, 134). The molecular mechanism of such facultative bending has been reviewed in detail elsewhere (135).

Figure 4. Facultative bending into the major groove at one end of the dodecamer duplex [d(CGCGAATTCGCG)]2 (134). Note the positive roll at the CpG and GpA steps and the negative propeller twisting in the central four base pairs. Overall and local helical axes are shown.

The recognition sequences of serum response factor (SRF) (136) (see MADS-Box Proteins), TATA-binding protein (TBP) (74-77), the restriction endonucleases EcoRI and EcoRV (58, 137), and EcoRI DNA methyltransferase (138) all have AT-rich core sequences, characterized by increased bendability. The crystal structure analyses of DNA complexes of these proteins revealed that they take advantage of the inherent bendability of their DNA targets. The DNAs adopt significantly bent conformations in all of these complexes.

Myocyte enhancer factor-2C (MEF-2C) interacts with DNA in a similar fashion to SRF (139, 140), and its consensus DNA-binding site was determined in polymerase chain reaction (PCR)-mediated binding site selection assays as CTA(A/T^TAG (141). A detailed analysis of the affinities of MEF- 2C for DNA of varying sequence revealed that mutations within the central four base pairs are tolerated so long as adenine is replaced with thymine or vice versa (139, 142). However, the replacement of the central bases with guanine and cytosine significantly diminished the affinity for MEF-2C (Table 1).

Table 1. DNA Binding Parameters for MEF-2C (2-117) and GG-MEF-2C(1-117)

		GG-MEF-2C(1-117)
DNA Sequence	MEF-2C(2-117)b	c
TGCTGC TATAAATA GAGTGA	110 (±20)	103 (±9)
TGCTGC TATATATA GAGTGA	130 (±25)	82 (±5)
TGCTGC TTTAAATA GAGTGA	118 (±11)	112 (±28)
TGCTGC TATTAATA GAGTGA	117(±23)	203(±14)
TGCTGC TAATAATA GAGTGA	108 (±20)	227 (±22)
TGCTGC AAAAAAAA GAGTGA	742 (±163)	1004 (±195)
TGCTGC TATGCATA GAGTGA	1072 (±200)	1780 (±370)
CTGCTGC TATA-ATA GAGTGA	129 (±26)	207 (±32)
CTGCTGC TATAAAT- GAGTGA	109 (±11)	135 (±12)
CTGCTGC -ATAAAT-	339(±73)	1896(±417)
GAGTGAC
CTGCTGC TAT—	126 (±21)	115 (±12)
ATAGAGTGAC

 

A correlation was observed between the affinity of MEF-2C for a mutant site and the intrinsic bendability of the site. MEF-2C binds to a DNA site with an alternating run of eight thymines and adenines with maximal affinity, while the affinity for an inherently rigid A-tract sequence is reduced by more than one order of magnitude (Table 1). Because of its poor stacking (Fig. 5), the deformability of a TA step is inherently much greater than that of an AT or an AA step (74, 121, 135, 143-152). A simple mechanical model to explain this observation has been provided by Finch and co-workers (135, 153). The proximity of the methyl groups of two successive thymines and the methyl groups and the phosphate backbone make the helix rather rigid (Fig. 5). In an AT step, the stacking of the methyl group with the adjacent adenine and the intervening sugar phosphate backbone again prevents bending by a roll mechanism. However, in a TA step the methyl group projects into the major groove without any significant stacking interactions with either the adjacent adenine or the phosphate backbone, and this step therefore displays a higher deformability than the AA and AT steps.

Figure 5. Schematic representations of thesteps indicating the sequence specific conformational differences between these steps. The DNA is viewed from the major groove indicated by Ma. The thymine methyl groups are indicated by small spheres. The methyl groups belonging to the right strand are shaded. The twist, slide, and roll movements that could create sterically unfavorable interactions are marked with x. The regions of the DNA which clash due to these movements are indicated. This part of the figure was adapted from Suzuki et al. (153). The differences in the nearest-neighbor stacking energies are given on the right, indicating that the most favorable stacking interaction is observed for Pu-Py dinucleotide steps (143, 147).

Reducing the length of the AT run from eight to six nucleosides by removing the first and the eighth base of the run (which generates a consensus site for SRF) reduced the stability of the MEF-2C complex almost 20-fold (Table 1). When the six AT bases consisted of alternating thymines and adenines, however, MEF-2C bound to the corresponding oligonucleotide again with almost the same affinity as to a MEF-2C site of length eight. This observation strongly supports the proposal that the inherent bendability of the DNA-binding site is a principal determinant of the DNA-binding specificity of MEF-2C (139, 140, 142). The affinities of the various AT-rich sequences for MEF-2C might be a measure of the relative bendability of the unbound DNA.

Similarly, a comparison of the DNA structure of the free uncomplexed Trp operator with that observed in the trp repressor-operator complex provided evidence that particular DNA sequences might be predisposed to adopt a non-B-form conformation in protein complexes (72).

Circular dichroism spectroscopy and bending analysis by circular permutation assays revealed that MEF-2C potentiates the natural tendency of its DNA target to adopt a bent conformation (142). DNA binding by MEF-2C is accompanied by DNA bending of approximately 70°, irrespective of the particular DNA sequence. This observation depends on the presence of the N-terminal methionine residue. In its absence, DNA containing a high-affinity binding site is bent by only 49°, while heterologous sequences remain unbent (139, 142). The differences in DNA-binding affinity are much less pronounced in the absence of the ^-terminal methionine (Table 1). The ^-terminal methionine appears to anchor MEF-2C to the ends of the AT run, thereby orientating the protein properly on the DNA. In the minor groove, A-T base pairs can be distinguished from G-C base pairs by the lack of a heterocyclic atom at C2 of adenine. Based on the SRF structure (see MADS-Box Proteins), it had been suggested that the ^-terminal methionine of MEF-2C is located over A4 and could specify the adenine by means of hydrophobic interaction with C2 of adenine (136). However, the observation that MEF-2C can bind with high affinity to a run of three alternating TA steps indicated that the ^-terminal methionine does not interact with C2 of adenine (154) of the MEF site. With the short TA run, the steric clash between the side chain of methionine and the amino group on C2 of guanine (154) would reduce the stability of the complex significantly (142). It is therefore more likely that the side chain of the ^-terminal methionine packs against the sugar ring of the nucleotide (154).

When a protein binds to a more rigid DNA sequence, it often does not bend DNA, or a substantial amount of the binding free energy must be used to bend the DNA target. Even in these cases, however, proteins often target the most deformable base step within their target sequence. The purine repressor, PurR, which is involved in the biosynthesis of purines and pyrimidines, binds to runs of four consecutive adenines and thymines, interrupted by the dinucleotide CG (155). PurR bends the DNA into the major groove by 45° through the insertion of a hinge into the minor groove and intercalation of the side chains of two leucine residues into the CG step (124). Calorimetric analysis suggested that purine-pyrimidine steps stack more stably than do purine-purine or pyrimidine-pyrimidine steps by approximately 0.52 kcal/mol, while the stacking energy of pyrimidine-purine steps is reduced by another 0.39 kcal/mol (Fig. 5) (143, 147). Therefore, CpG is the step that can be unstacked most easily within the pur operator. Similarly, the bending of 22° of the PRDI site within the interferon-b promoter in the complex with interferon regulatory factor (IRF) occurs at the CpA step which is the most easily unstacked dinucleotide in the PRDI sequence, ACTTTCACTTCTC (156). The cocrystal structure of the DNA complex of the chromosomal protein Sac7d from the hyperthermophile Sulfolobus acidocaldarius revealed that the DNA is kinked by 61° through the intercalation of two amino acid side chains, namely Val26 and Met29 between the bases of the dinucleotide C(2)G(3) (157). This is the base step that is most easily unstacked in the DNA sequence GCGATCGC. All these examples demonstrate that proteins recognize DNA not only through specific contacts with the nucleobases, but also through inherent, sequence-dependent properties of the DNA. Therefore, considering the properties of both the protein and the DNA makes possible an understanding of the sequence specificity and the observed DNA bend in DNA-protein complexes (135, 158).

The binding specificity of a protein is determined by the difference in binding affinities of the protein for the specific and the nonspecific sites. DNA bending is an energy costly process, even for sequences of enhanced bendability (Fig. 6). Some of the binding free energy must be used to bend a DNA molecule which, in the absence of protein, adopts either an unbent or an only slightly bent conformation. Therefore, protein-induced DNA bending can contribute to the overall specificity of a DNA-binding protein (142, 159-161). MEF-2C, for example, must use some of the binding free energy from specific interactions to induce bending at the specific site. In the case of a MEF-2C mutant lacking the N-terminal methionine, however, no binding free energy appears to be used for the bending of the nonspecific complex (139, 142). Consequently, the difference between the free energies of the specific and the nonspecific complexes is reduced because of the unfavorable contribution from bending the specific DNA. On the other hand, MEF-2C(1-117), which contains the N-terminal methionine, bends both specific and nonspecific DNA, thereby increasing the difference in the free energy between the specific and the nonspecific complexes. The increased DNA-binding specificity of MEF-2C(1-117) over that of MEF-2C(2-117) relies therefore in part on the fact that MEF-2C(1-117) bends DNA irrespective of the particular sequence. MEF-2C(1-117) appears to select DNA-binding sites that are characterized by increased bendability, because for these DNA sequences less binding free energy must be expedited to force them into the bent conformation necessary for the optimal shape complementarity observed in the complexes. Interestingly, the CD spectra of the DNA complexes of MEF-2C(1-117) indicated that the conformation of the DNA was independent of its sequence, at least within the resolution of CD spectroscopy (142), while the CD spectra of the DNA in the MEF-2C(2-117) complexes varied strongly with the DNA sequence (139, 142). Scanning tunneling microscopy has indicated that the specific and nonspecific DNA complexes of l Cro are characterized by similar bend angles. The bend angle induced by binding of a Cro dimer to DNA was determined as 69° ± 11° and 62° ± 23° for specific and non-specific DNA, respectively (159).

Figure 6. Free energy diagram illustrating the effect of DNA bending on the stabilities of the complexes of protein P with DNA D. The DNA-binding specificity DDG is defined as the difference of the binding free energies for the formation of the specific and nonspecific complexes: DDG = DGSP – DGNS. The total free energy for the formation of the specific and the nonspecific complexes can be partitioned into the contributions from the (favorable) interactions between the DNA and the protein, D G(Int), and the (unfavorable) free energy change required to bend the DNA, DG (Bend).

In order to bend DNA, proteins use primarily two mechanisms. In the first, kinks are introduced through the intercalation of amino acid side chains between adjacent base pairs. In the IHF-DNA complex, two intercalating proline residues projecting from the tip of each b-hairpin arm introduce large kinks at symmetrically displaced ApA steps. SRY (125) and LEF-1 (126), which are structurally similar proteins, use nonaromatic, hydrophobic amino acids to intercalate into base steps from the minor grove side of the double helix, creating a widened minor grove and a dramatically bent DNA. TATA box-binding protein induces strong bends of approximately 45° at either end of the TATA box through the intercalation of two phenylalanine rings between two adjacent base pairs (74-77), while the hyperthermophile chromosomal proteins Sac7d bends DNA by 72°, primarily through the intercalation of Val(26) and Met(29) at a single CpA base step (157).

The second mechanism for the introduction of bends into DNA was originally proposed by Rich (162). Asymmetric neutralization of the negative charges of the phosphodiesters by cationic amino acid residues could result in unbalanced Coulombic repulsions between the negative charges on DNA, causing DNA to collapse toward the bound protein (162-164). For example, the X-ray structure of serum response factor bound to its target DNA sequence indicated that asymmetric charge neutralization is at least partly responsible for the DNA bend angle of 72° (136). Several positively charged amino acid residues of SRF bind the phosphate groups on only one side of the SRF-site (Fig. 7). The three positively charged amino acids that interact with the distal ends of the DNA provide the hands through which SRF pulls up the far ends of the DNA. In addition, Arg143 lies in an extended conformation along the floor of the minor grove and stabilizes the relatively large propeller twists at the central base steps, which are characteristic of AT-rich regions. These interactions also facilitate the bending of DNA. The members of the MEF-2 family of proteins most probably follow a similar mechanism to induce bends into DNA (142). However, the bending of the AT-rich target sequences of SRF and of the MEF-2 proteins is helped by the intrinsically high bendability of these DNA sequences (135). The X-ray structure of the complex between the bZ protein GCN4 and an ATF-site-containing DNA revealed that the DNA was unbent (165). On the other hand, the heterodimer of the bZ proteins Jun and Fos appears to induce DNA bending when bound to an AP-1 site (166-168), although some controversy exists on this point (169-173). The identity of the amino acids just N-terminal to the basic region appears to be important for DNA bending, in that cationic amino acids in these positions contact the negatively charged phosphate diesters on only one face of the DNA, thereby inducing the DNA to bend away from the leucine zipper (174-178). The corresponding amino acids in GCN4 are the uncharged Pro-Ala-

Ala, consistent with the observation that GCN4 does not bend AP1-site-containing DNA. Analysis of the bending properties of GCN4 mutants revealed that cationic amino acids in these positions induce DNA bending toward the neutralized surface, while anionic amino acids induce bending in the opposite direction (179).

Figure 7. Interactions between positively charged amino acid side chains and DNA phosphates in the SRF-DNA complex (136). Note that these contacts occur only on one side of the duplex, causing the DNA to bend towards the protein. The histidines and lysines on either end of the duplex serve as handles to further bend the DNA (the DNA used is too short to allow formation of the His-phosphate contact on the left-hand side).

While DNA bending is the most dramatic change observed upon DNA binding, many other more subtle conformational rearrangements of the DNA have been observed. While a slight bending of the DNA of approximately 25° has been observed in the crystal structure of the 434 repressor-DNA complex (180, 181), the major characteristic is the significantly overwound DNA in the region of the central four base pairs of the binding site. Relative to canonical B-DNA, the net overtwisting is approximately 20°. Although these four base pairs are not in direct contact with the repressor, operators with A-T or T-A base pairs at these positions are bound more strongly than those bearing C-G or G-C (182). A relationship between the intrinsic twist of an operator, as determined by the sequence of its central bases, and its affinity was observed: Operators with lower affinity are undertwisted relative to operators with higher affinity (128).