Predicting the impact of promoter variability on regulatory outputs

Abstract

The increased availability of whole genome sequences calls for quantitative models of global gene expression, yet predicting gene expression patterns directly from genome sequence remains a challenge. We examine the contributions of an individual regulator, the ferrous iron-responsive regulatory element, BqsR, on global patterns of gene expression in Pseudomonas aeruginosa. The position weight matrix (PWM) derived for BqsR uncovered hundreds of likely binding sites throughout the genome. Only a subset of these potential binding sites had a regulatory consequence, suggesting that BqsR/DNA interactions were not captured within the PWM or that the broader regulatory context at each promoter played a greater role in setting promoter outputs. The architecture of the BqsR operator was systematically varied to understand how binding site parameters influence expression. We found that BqsR operator affinity was predicted by the PWM well. At many promoters the surrounding regulatory context, including overlapping operators of BqsR or the presence of RhlR binding sites, were influential in setting promoter outputs. These results indicate more comprehensive models that include local regulatory contexts are needed to develop a predictive understanding of global regulatory outputs.

Figure 1. The BqsR binding motif in the genome and its impact on global expression.

(A) The response regulator BqsR activates gene expression. The BqsR operator sequence contains a repeated pentamer (TTAAG) separated by 6 bp. (B) A comparison of the predicted operator strength with the observed experimental fold change in expression measured using RNA-seq. Operator strength was predicted with Equation 1 using best-fit values for parameters a and b. The curve shows y = x. For operator regions containing multiple BqsR binding sites, only the site with the highest score was plotted.

Figure 2. Variability of the BqsR operator’s repeated pentamer sequence and its influence on gene regulation.

(A) The reference operator found in the PA14_04180 promoter region contains two pentamers separated by a 6-bp spacer region. (B) T he graph shows the PWM score and sequence logo for the upstream and downstream pentamer sequences calculated using the operators from Fig. 1. (C) Expression measurements of synthetic constructs quantified the influence of each point mutation for all nucleotides in the upstream pentamer on gene expression. Error bars show standard error of biological triplicates.

Figure 3. Variability of the spacer sequence and its influence on gene regulation.

(A) The number of operators containing spacers of length 5, 6, or 7 bp is plotted for potential operators containing either 0 or 1 mutations in the pentamer regions, revealing a preference for spacer regions of length 6 bp. (B) Experimental measurements of synthetic operators confirm that a 6-bp spacer length is critical for BqsR-mediated regulation, as spacer lengths of 5 or 7 bp resulted in expression levels similar to the negative control in which the upstream pentamer was deleted. (C) Operators from (A) containing a 6-bp spacer region were analyzed for sequence preference by calculating the PWM score and sequence logo. (D) Gene expression measurements of 6 synthetic spacer sequences showed that the sequence of the spacer region modulates the level of gene regulation up to a factor of 2.

Figure 4. Promoter regions with multiple operators.

(A) The number of operators within each promoter region of the genome as compared to randomly distributing the same number of operators throughout the genome. Operators are sequences 16 bp in length and contain up to 2 mutations in the pentamer regions. (B) Several of these clustered promoters are spaced by 6 bp, the same spacing between the two pentamer regions of an individual promoter, creating arrays of overlapping binding sites as shown in the schematic. (C) The promoter for PA14_04180 is one such promoter containing 4 repeated pentamer regions. (D) Gene expression measurements on synthetic versions of the PA14_04180 promoter, in which individual or pairs of pentamer repeats were removed, revealed that each pentamer repeat contributes to the overall level of gene regulation, although not all the BqsR binding sites contribute equally to the regulatory output.

Figure 5. Creating a binding energy matrix for the BqsR operator sequence.

(A) The reference operator sequence from gene PA14_04180. (B) Energy matrix derived from the experimental measurements of regulation by synthetic operators. (C) For comparison, the energy matrix derived from the position weight matrix used for prediction shown in Fig. 1B, which was constructed using qPCR expression data for 12 genes. Binding energies in k_BT units are relative to the binding strength to the reference operator in (A). Matrix positions in dark red indicate bp changes that cause large reductions in expression levels, whereas matrix positions in dark blue indicate bp that increase expression levels.

Figure 6. Comparison of global predictions of gene expression using the energy matrices in Fig. 5 to gene expression measured using RNA-Seq.

(A) Predicted vs. experimentally measured BqsR-mediated fold change in expression levels for 75 genes. Black x’s show predicted expression based on the energy matrix shown in Fig. 5B. Blue circles show predicted expression based on the PWM shown in Fig. 5C. (B) Both PWM model and our model made similar predictions of fold change in expression. (C) Error in prediction, defined as the ratio of predicted to experimentally measured fold change in gene expression, plotted as a function of measured fold change in expression.

Figure 7. The influence of operator position and secondary transcription factors on the ability to predict BqsR-mediated gene regulation.

(A) For the genes predicted in Fig. 6, the error in prediction, defined as the ratio of predicted over experimentally measured expression levels, is shown as a function of operator positions relative to the transcription start site. A prediction error of 1 indicates the measurement exactly matched the prediction. (B) Error in prediction plotted as a function of the number of BqsR binding sites found in the promoter region. (C) Venn diagrams showing regulon overlaps between the predicted genes and selected transcriptional regulators with statistically significant overlap. See section “Comparison to other operons” in Supplemental Materials. (D) Position weight matrices for 12 transcription factors were used to identify which of the predicted genes were likely to be coregulated by an additional transcription factor. The upper graph shows the number of promoter regions containing a potential binding site for a second transcription factor. Bars on the bottom graph show the average prediction error for each set of genes containing a secondary transcription factor; the larger the ratio, the greater the error.

Figure 8. Fold change in expression due to ferrous iron shock for promoters coregulated by both BqsR and RhlR.

For genes containing both BqsR and RhlR binding (bqsP, PA14_01240, PA14_04180, and PA14_07000), the fold change in expression due to Fe(II) shock was decreased upon overexpression of RhlR, demonstrating the ability of RhlR to modulate the effect BqsR has on expression at these promoters. Expression from the oprH promoter, which does not contain an RhlR binding site, was not significantly influenced by RhlR overexpression. The inset shows the bqs promoter, which contains overlapping operators. WT response is shown in white, ∆bqsR is shown in red, and WT overexpressing rhlR is shown in blue. *indicates a p-value < 0.05.