Designed proteins to modulate cellular networks

Abstract

A major challenge of protein design is to create useful new proteins that interact specifically with biological targets in living cells. Such binding modules have many potential applications, including the targeted perturbation of protein networks. As a general approach to create such modules, we designed a library with approximately 10(9) different binding specificities based on a small 3-tetratricopeptide repeat (TPR) motif framework. We employed a novel strategy, based on split GFP reassembly, to screen the library for modules with the desired binding specificity. Using this approach, we identified modules that bind tightly and specifically to Dss1, a small human protein that interacts with the tumor suppressor protein BRCA2. We showed that these modules also bind the yeast homologue of Dss1, Sem1. Furthermore, we demonstrated that these modules inhibit Sem1 activity in yeast. This strategy will be generally applicable to make novel genetically encoded tools for systems/synthetic biology applications.

Figure 1

TPR framework and Dss1 structures. a) Structure of the TPR2A domain, used as our scaffold, where each repeat and the solvating helix are colored differently. The seven residues that we designated for randomization are highlighted in blue. All of the residues selected for randomization (K229, N233, Y236, N264, K301, R305, and N308 of Hsp Organizing protein, Hop, where the TPR2A domain is amino acids 222–352) interact with Hsp90s C-terminal peptide. Five of these residues, known as a “carboxylate clamp” (K229, N233, N264, K301, and R305), are critical for TPR2A’s ability to bind Hsp90 (13, 15). b) Mammalian breast cancer type 2 (BRCA2) protein’s DNA-binding domain complexed with ssDNA and Dss1 (30). Dss1 is shown in ribbon representation, highlighted in orange. BRCA2 is shown in grey and the DNA is shown in green. The figure below shows a close-up of the area highlighted in the black box, showing the C-terminal residues of the Dss1 protein that were targeted in the selection. Dss1-C19 19-mer peptide sequence, the 12 underlined residues, are seen in the X-ray structure interacting with BRCA2; the last 7 C-terminal residues are not seen in the structure.

Figure 2

T-Mods in vitro binding by ITC. ITC binding isotherms for the interaction of Dss1 19-mer C-terminal peptide with two Dss1-binding T-Mods: T-Mod(Dss1A) (a) and T-Mod(Dss1B) (b). Dss1-C19 peptide was titrated at 2.66 mM into 130 µM T-Mod(Dss1A) solution, and 1.52 mM Dss1-C19 was titrated in 114 µM T-Mod(Dss1B) solution. The data was fit to a 1:1 binding model to calculate the stoichiometry (N) and the binding affinity (K_d) of the interactions using Origin 7.0 (T-Mod(Dss1A) K_d = 20.2 µM, N = 0.8; T-Mod(Dss1B) K_d = 18.6 µM, N = 1.1).

Figure 3

T-Mods-Dss1 binding mode. a) Circular dichroism spectra of Dss1-C19 peptide free in solution at 50 µM (solid line) and bound to T-Mod(Dss1A) (dashed line). The spectrum of the bound peptide corresponds to the subtraction of the T-Mod(Dss1A) spectrum (50 µM) from the T-Mod(Dss1A) in complex with Dss1-C19 peptide (50 µM each). b) Structural model of the T-Mod(Dss1)-Dss1 peptide binding interface. The first helix of the third repeat of the TPR module is shown as a green ribbon. The three randomized residues that reached a consensus in the Dss1-binding modules are shown as gray sticks. The C-terminal part of Dss1 from the Dss1-BRCA2 complex structure is shown in orange. The amino acid side chains are shown in sticks. The key residues for the Dss1-C19-TPR interaction, R57 and L60, are labeled. c) ITC binding isotherms for the interaction of Dss1-C19 peptide mutant R57A with T-Mod(Dss1A). The Dss1-C19 peptide was titrated at 1.37 mM concentration into 100 µM T-Mod(Dss1A) solution. The binding enthalpies were integrated and fit to a 1:1 binding model to calculate the stoichiometry (N) and the binding affinity (K_d) of the interactions (T-Mod(Dss1A) K_d = 157 µM, N = 0.8). d) ITC binding isotherms for the interaction of Dss1-C19 peptide mutant L60A with T-Mod(Dss1A). The Dss1-C19 peptide was titrated at 1.32 mM into 100 µM T-Mod(Dss1A) solution.

Figure 4

In vivo binding activity of TPR domains. a) Sequence alignment of human Dss1 protein and the yeast ortholog Sem1, where “*” means identical residues, and “:” means similar residues. The black square highlights the 19 C-terminal residues of Dss1 protein that were targeted during the selection. The schematic shows the results of the secondary structure prediction in the C-terminal region by Agadir (40). b) Temperature-sensitive growth phenotype of S. cerevisiae transformants. Cell cultures were diluted to the same concentration, and serially diluted cells were spotted onto galactose plates and incubated at 25 or 37 °C. The strains of S. cerevisiae, plated from top to bottom, are the following: wild-type (WT), sem1Δ (cells lacking the chromosomal SEM1 gene), sem1Δ transformed with the plasmid pGPD416-Dss1 (sem1Δ+Dss1), WT expressing T-Mod(Dss1A) fused to a N-terminal nuclear localization signal (SV40-NLS) from a p424-GAL1 vector (WT+TPR), sem1Δ expressing T-Mod(Dss1A) fused to SV40-NLS from the p424-GAL1 vector (sem1Δ+TPR), sem1Δ complemented with pGPD416-Dss1 and expressing T-Mod(Dss1A) with a N-terminal NLS from the p424-GAL1 vector (sem1Δ+Dss1+TPR). c) Hydroxyurea-sensitive growth phenotype of S. cerevisiae transformants. Cell cultures were diluted to the same concentration, and serially diluted cells were spotted into SD galactose plates and SD galactose plates with 0.1 M hydroxyurea (HU) and incubated at 30 °C. The strains of S. cerevisiae, plated from top to bottom, are the following: wild-type (WT), sem1Δ (cells lacking the chromosomal SEM1 gene), WT expressing T-Mod(Dss1A) fused to a N-terminal nuclear localization signal (SV40-NLS) from a p424-GAL1 vector (WT+TPR), rpn10Δ (cells lacking the chromosomal RPN10 gene), sem1Δ rpn10Δ (cells lacking both the chromosomal SEM1 and RPN10 genes), rpn10Δ expressing T-Mod(Dss1A) fused to SV40-NLS from the p424-GAL1 vector (rpn10Δ+TPR).