Engineering transcription factors with novel DNA-binding specificity using comparative genomics

Abstract

The transcriptional program for a gene consists of the promoter necessary for recruiting RNA polymerase along with neighboring operator sites that bind different activators and repressors. From a synthetic biology perspective, if the DNA-binding specificity of these proteins can be changed, then they can be used to reprogram gene expression in cells. While many experimental methods exist for generating such specificity-altering mutations, few computational approaches are available, particularly in the case of bacterial transcription factors. In a previously published computational study of nitrogen oxide metabolism in bacteria, a small number of amino-acid residues were found to determine the specificity within the CRP (cAMP receptor protein)/FNR (fumarate and nitrate reductase regulatory protein) family of transcription factors. By analyzing how these amino acids vary in different regulators, a simple relationship between the identity of these residues and their target DNA-binding sequence was constructed. In this article, we experimentally tested whether this relationship could be used to engineer novel DNA-protein interactions. Using Escherichia coli CRP as a template, we tested eight designs based on this relationship and found that four worked as predicted. Collectively, these results in this work demonstrate that comparative genomics can inform the design of bacterial transcription factors.

Figure 1.

Multiple sequence alignment of the DNA-recognition helix within candidate proteins from the CRP/FNR family of transcription factors. Residues in CRP known to make specific contacts within DNA bases (22) are highlighted in black. The specificity-determining residues from Rodionov and coworkers (20) correspond to columns 4, 5 and 10. Note that the specificity determining residues for PrfA are somewhat ambiguous due to an insertion in the alignment relative to CRPwt. The numbers in parentheses denote the region of the protein shown in the alignment and the mutant column designates the corresponding CRP mutation used in this study.

Figure 2.

(A) Expression of lacZ-gfp transcriptional fusion with mutated operator sites in wild-type and Δcrp cells. (B) Expression of lacZ-gfp transcriptional fusion with mutated operator sites in Δcrp cells when wild-type CRP is ectopically expressed from an atc-inducible promoter. (C) Expression of lacZ-gfp transcriptional fusion with wild-type operator site in Δcrp when the different CRP variants are ectopically expressed from an atc-inducible promoter. All expression values were normalized relative to the lacZ-gfp transcriptional fusion in a wild-type (crp+) background.

Figure 3.

Results obtained when pairing the lacZ-gfp transcriptional fusion with the cognate CRP mutation ectopically expressed from an atc-inducible promoter. All expression values were normalized relative to the lacZ-gfp transcriptional fusion in a wild-type (crp+) background.

Figure 4.

Expression with the CRP7–Om7 pair at varying levels of atc induction. As a reference, all other results involve atc induction at 20 ng/ml. All expression values were normalized relative to the lacZ-gfp transcriptional fusion in a wild-type (crp+) background.

Figure 5.

Results obtained when pairing all CRP mutations against all operator site mutations. (A) Results in the absence of atc; (B) Results in presence of atc. All values were normalized relative to the lacZ-gfp transcription fusion in a wild-type (crp+) background. To improve contrast, all values less than 10% the expression of the lacZ promoter in a wild-type background were displayed as having zero expression. The standard deviation for all pairs was less than ten percent (results not shown).

Figure 6.

Results obtained from screen of randomized operator sites. In these experiments, we screened for increased CRP5 activation of Om5 reporter by randomizing positions 9, 10 and 11 in the operator sequences. The resulting sequences isolated in the screen were as follows: Om5*1 (TCCGGT), Om5*2 (CAGTGA), Om5*2 (GCTGGA) and Om5 (CATATC).