doi:10.1016/j.jmb.2004.12.062 
Copyright © 2005 Elsevier Ltd All rights reserved.
Modulating the DNA Affinity of Elk-1 with Computationally Selected Mutations
Sheldon Parka, Eric T. Boderb and Jeffery G. Savena, 
, 
aMakineni Theoretical Laboratories, Department of Chemistry, University of Pennsylvania, 231 South 34th Street, Philadelphia, PA 19104, USA
bDepartment of Chemical and Biomolecular Engineering, University of Pennsylvania, 220 South 33rd Street, Philadelphia, PA 19104, USA
Received 22 September 2004;
revised 13 December 2004;
accepted 16 December 2004.
Edited by J. E. Ladbury.
Available online 2 March 2005.
References and further reading may be available for this article. To view references and further reading you must
purchase this article.
In order to regulate gene expression, transcription factors must first bind their target DNA sequences. The affinity of this binding is determined by both the network of interactions at the interface and the entropy change associated with the complex formation. To study the role of structural fluctuation in fine-tuning DNA affinity, we performed molecular dynamics simulations of two highly homologous proteins, Elk-1 and SAP-1, that exhibit different sequence specificity. Simulation studies show that several residues in Elk have significantly higher main-chain root-mean-square deviations than their counterparts in SAP. In particular, a single residue, D69, may contribute to Elk's lower DNA affinity for Pc-fos by structurally destabilizing the carboxy terminus of the recognition helix. While D69 does not contact DNA directly, the increased mobility in the region may contribute to its weaker binding. We measured the ability of single point mutants of Elk to bind Pc-fos in a reporter assay, in which D69 of wild-type Elk has been mutated to other residues with higher helix propensity in order to stabilize the local conformation. The gains in transcriptional activity and the free energy of binding suggested from these measurements correlate well with stability gains computed from helix propensity and charge–macrodipole interactions. The study suggests that residues that are distal to the binding interface may indirectly modulate the binding affinity by stabilizing the protein scaffold required for efficient DNA interaction.
Keywords: protein–DNA interactions; molecular dynamics simulation; Ets proteins; transcription factor; helix propensity
Abbreviations used: MD, molecular dynamics; DBD, DNA-binding domain; SRF, serum response factor; NLS, nuclear localization sequence; AD, activation domain; EGFP, enhanced green fluorescent protein
Figure 1. (a) The structure of Elk bound to DNA containing the PE74 sequence (PDB 1DUX). The image was generated using SwissPDB viewer21 and POV-ray v. 3.5. (b) Sequence alignment of the ETS domains of Elk (residues 5–90) and SAP (residues 5–89). Identical residues are indicated with vertical bars, high sequence similarity with two dots, and weak sequence similarity with one dot. The secondary structures of Elk shown above the sequence correspond to helices H1–H3 (bars) and strands S1–S4 (arrows). Residue 69 is in bold. The residue numbering is for Elk. (c) The PE74 and Pc-fos sequences used in the study. The two flanking bases that are different are highlighted.
Figure 2. (a) The main-chain RMSD (Å) of Elk and SAP from MD simulations. (b) The residue-specific RMSD of Elk (diamond) and SAP (square), time-averaged over the period of 1–4 ns of simulation. The three regions (R1–R3) where the Elk RMSD significantly exceeds that of SAP are indicated. The configurations from two independent 4 ns simulations were analyzed. (c) The RMSD of Elk (diamond) and SAP (square) computed from the B-factors in the Elk-PE74 and SAP-PE74 structures using B=(8/3)π2RMSDB2.
Figure 3. (a) The reporter activity of SAP against reporter plasmids containing either nine copies of PE74 (filled) or Pc-fos (light) in the promoter. The background (dark) corresponds to yeast transformed with the Pc-fos-EGFP reporter alone. (b) The reporter activity of Elk against PE74-EGFP and Pc-fos-EGFP. Coloring is the same as in (a).
Figure 4. The reporter activity of various point mutants: wild-type (black), D69V (cyan), D69A (pink), D69S (green), and background (orange).
Figure 5. D69S was analyzed by MD to determine if an increase in DNA affinity correlates with a decreased RMSD. The residue-specific RMSD of D69S (square) was computed by averaging over the 1–4 ns interval of simulation. Ensemble average of two independent simulations.
Figure 6. An example of configurations showing hydrogen bonds formed near the carboxy terminus of the recognition helix. (a) SAP, (b) Elk, (c) D69S.
Figure 7. ΔΔGexp versus ΔΔGtheo. The linear fit through the data is ΔΔGexp=1.147 ΔΔGtheo−0.012 kcal/mol.
Table 1.
The normalized fluorescence of single point Elk mutants and the corresponding experimental free energy gain 
The theoretical free energy differences (ΔΔGtheo) were computed from helix propensities and the macrodipole–side-chain interactions (see the text)
a n=7 for S, A, Q, V, N, E, H;
n=4 for T;
n=3 for SAP.
Abstract
Purchase PDF (434 K)
E-mail Article
Add to my Quick Links
Cited By in Scopus (3)
Purchase PDF (159 K)
