Thermodynamics-based models of transcriptional regulation by enhancers: the roles of synergistic activation, cooperative binding and short-range repression - PubMed (original) (raw)
Thermodynamics-based models of transcriptional regulation by enhancers: the roles of synergistic activation, cooperative binding and short-range repression
Xin He et al. PLoS Comput Biol. 2010.
Abstract
Quantitative models of cis-regulatory activity have the potential to improve our mechanistic understanding of transcriptional regulation. However, the few models available today have been based on simplistic assumptions about the sequences being modeled, or heuristic approximations of the underlying regulatory mechanisms. We have developed a thermodynamics-based model to predict gene expression driven by any DNA sequence, as a function of transcription factor concentrations and their DNA-binding specificities. It uses statistical thermodynamics theory to model not only protein-DNA interaction, but also the effect of DNA-bound activators and repressors on gene expression. In addition, the model incorporates mechanistic features such as synergistic effect of multiple activators, short range repression, and cooperativity in transcription factor-DNA binding, allowing us to systematically evaluate the significance of these features in the context of available expression data. Using this model on segmentation-related enhancers in Drosophila, we find that transcriptional synergy due to simultaneous action of multiple activators helps explain the data beyond what can be explained by cooperative DNA-binding alone. We find clear support for the phenomenon of short-range repression, where repressors do not directly interact with the basal transcriptional machinery. We also find that the binding sites contributing to an enhancer's function may not be conserved during evolution, and a noticeable fraction of these undergo lineage-specific changes. Our implementation of the model, called GEMSTAT, is the first publicly available program for simultaneously modeling the regulatory activities of a given set of sequences.
Conflict of interest statement
The authors have declared that no competing interests exist.
Figures
Figure 1. Statistical thermodynamic models of gene expression.
(A) All possible molecular configurations of a CRM with two binding sites (purple), that may or may not be bound by a transcription factor (green circle = activator, red circle = repressor). The statistical weight W of each configuration is shown to its right. Each occupied site makes a contribution to W in a multiplicative fashion. (B) Cooperative DNA-binding is modeled by introducing a multiplicative factor (ω) to the statistical weight of a configuration. The same configuration is shown along with its statistical weight W under a model with no cooperativity (top) and a model with self-cooperative DNA-binding (bottom). (C) Statistical weight contributions from TF-DNA interactions (W) and from TF-BTM interactions (Q) for each configuration, in the Direct Interaction model (blue circle = BTM). Each bound activator or repressor molecule contributes to the TF-BTM interaction term (Q) in a multiplicative fashion. The statistical weight also receives a contribution from BTM binding at the promoter; this term is not shown here. (D) Same as (C), but for the short range repression model. A bound repressor (red circle) does not have a direct interaction with the BTM. Also, there is one additional configuration allowed here, as compared to Direct Interaction: one where repressor is bound and “effective” in shutting down its neighborhood for binding at activator sites (bottom). The statistical weight (W) of this configuration is scaled by a factor of βR, reflecting the strength of the repressor to change the chromatin accessibility. (E) Two ways to model the action of multiple bound activators: “additive effect” (top 2 configurations) and “multiplicative effect” (bottom). The total statistical weight (W×Q) under each model is shown. In the former, only one bound activator may contact the BTM in any configuration, while the latter has no such restriction and leads to transcriptional synergy.
Figure 2. Effect of cooperative DNA-binding of TFs and the mode of transcriptional activation (multiplicative or not) on model performance.
(A,B) Predicted expression profiles of a DirectInt model with no cooperativity (“no-coop”, blue) and a model with self-cooperative binding for Bcd and Kni (“coop”, green) are shown for each CRM, with reference to the CRM's known readout (“Obs.”, red). The correlation coefficient between a model's prediction and the known readout is indicated in the top right corner of the panel. Each expression profile is on a scale of 0 to 1 (scaling does not affect correlation coefficient), and shown for bins 20 to 80 (i.e., 20% to 80% egg length) of the embryo. Shown are two CRMs for which one model was deemed better than the other (CC≥0.65, difference in CC≥0.05). (C,D) Predicted expression profiles of a DirectInt model with multiplicative activation (“synergy”, green) and one with additive activation (“no-synergy”, blue). Shown are two CRMs where the multiplicative model is better than the additive model (CC≥0.65, difference in CC≥0.05). Self-cooperative DNA-binding was not used in this evaluation.
Figure 3. Model predictions.
The predicted expression profile of the DirectInt model (with Bcd and Kni self-cooperativity) is shown (blue) in comparison to the known readout (red), for all CRMs in the data set. Each expression profile is on a scale of 0 to 1 (y-axis), and shown for bins 20 (left) to 80 (right) of the embryo. Labels in green indicate CRMs where the CC is greater than 0.65.
Figure 4. Evaluation of short range repression (SRR) model.
(A,B) Two of the four repressors (Kr and Hb) are evaluated separately, by comparing predictions from a model where one repressor is modeled through DirectInt (“DI”) to predictions from a model where that repressor acts through SRR (“_Kr-_SRR”and “_Hb-_SRR”, in panels A and B respectively). For each model, the average correlation coefficient (CC) of the K best predictions (CRMs) of that model is shown, as a function of K. Also shown for each model is the average CC (over the same K CRMs) when the repressor is “knocked down” (e.g., “DI-_Kr−_”, “_Kr-_SRR-_Kr−_” in panel A). (C,D) The SRR model for (C) Kr and (D) Hb (with range of repression dR = 250 bp) is compared to the corresponding model at dR = 10 bp, where the repressor can only affect overlapping or adjacent sites. Semantics of the plots are as in (A–B).
Figure 5. Effect of evolutionary filter on binding sites used in model.
The average CC over all 37 CRMs of the DirectInt model (without cooperative DNA-binding) is shown. The x-axis indicates the number of species in which conservation of a binding site was required for it to be included in the model's input. The red curve corresponds to the case where the conservation filter does not allow turnover, i.e., the sites used in the model must be fully conserved across all species considered. The blue curve represents a conservation filter that allows turnover, i.e., where a site may undergo lineage-specific changes. Thus for “number of species” = 6, a site used in the model may be conserved in six or fewer species, as long as the conservation is deemed significant by the procedure described in Text S1.
References
- Davidson EH. 2006. The Regulatory Genome: Gene Regulatory Networks in Development and Evolution: Academic Press.
- Arnosti DN, Barolo S, Levine M, Small S. The eve stripe 2 enhancer employs multiple modes of transcriptional synergy. Development. 1996;122:205–214. - PubMed
- Shea MA, Ackers GK. The OR control system of bacteriophage lambda. A physical-chemical model for gene regulation. J Mol Biol. 1985;181:211–230. - PubMed
Publication types
MeSH terms
Substances
LinkOut - more resources
Full Text Sources
Molecular Biology Databases