Payne and Wagner (Science, 2014) used protein-binding microarray data for hundreds of mouse and yeast transcription factors to construct the first large-scale empirical genotype networks of transcription factor binding sites, concluding that high-affinity sites are the most mutationally robust and that robustness and evolvability are antagonistic at the genotypic level but synergistic at the phenotypic level. These inferences rest on methodological choices — a hard binding threshold, E-scores as an in vitro proxy for binding affinity, and restriction to contiguous 8-nucleotide sites — and on a context that ignores chromatin and in vivo regulatory architecture; the same authors' subsequent work acknowledges these limitations, and comprehensive empirical study of entire regulatory circuits has since been identified as an open direction (Aguilar-Rodríguez et al. 2018; Santos-Moreno et al. 2023). The problem is therefore twofold and linked by one objective: to determine, quantitatively and empirically, (i) how these methodological choices and biological-context factors alter the structure of binding-site genotype networks and the robustness and evolvability conclusions drawn from them, and (ii) how exhaustive genotype-to-phenotype mapping, so far achieved only for individual binding-site repertoires and small synthetic circuits, with none yet for native or large-scale regulatory circuits, can be extended to whole regulatory circuits so that robustness and evolvability can be characterized at the circuit scale. Later work has delivered partial progress — synthetic three-node circuits based on CRISPR interference (CRISPRi, a programmable gene-repression method) with experimentally mapped genotype networks, and massively parallel reporter assays yielding in vivo adaptive landscapes of bacterial binding sites — but no systematic quantification of the original caveats' effects on genotype-network statistics, and no exhaustive mapping of native or large-scale regulatory circuits, exists.

Background

Transcription factors (TFs) are proteins that bind short, specific DNA words (transcription factor binding sites, TFBSs) in regulatory regions to control gene expression. A genotype-phenotype (GP) map assigns DNA sequences (genotypes) to the molecular phenotypes they produce, such as the expression level of a gene. A genotype network is the set of all sequences that produce the same phenotype, connected by single small mutations (e.g., point mutations); because many neighbors of any sequence preserve the phenotype, the phenotype is mutationally robust, and because different regions of the network have different mutant neighbors, the network lets evolving populations access new phenotypes, a property called evolvability. Genotype networks were first studied theoretically, for RNA folding maps and, closer to this problem, in computational models of gene regulatory circuits by Ciliberti, Martin and Wagner (2007), who showed computationally that circuits producing a given expression pattern form large, connected genotype networks that facilitate evolutionary innovation.

Payne and Wagner (2014) provided the first large-scale empirical version of this analysis for TFBSs. Using universal protein-binding microarray (PBM) data — arrays on which a labeled TF is probed against all possible short DNA words — they defined binding affinity through the E-score, a PBM-derived enrichment score, treated each contiguous 8-nucleotide sequence as a genotype, and classified a sequence as a binding site when its E-score exceeded a fixed threshold. Across hundreds of mouse and yeast TFs they found that binding sites of every TF form vast, connected genotype networks; that high-affinity sites are the most tolerant of nucleotide substitutions; and that mutational robustness and evolvability, which are antagonistic when comparing individual genotypes, are synergistic at the level of phenotypes, because the most robust phenotypes occupy the largest genotype networks and thus touch the most novel phenotypes.

The same authors' later work flags what these inferences rest on. The analysis uses a single hard binding threshold, an in vitro PBM proxy for affinity measured at one TF concentration, and contiguous 8-nucleotide sites, whereas real sites vary in length, overlap each other, reside in chromatin, and function inside regulatory circuits where multiple TFs, cooperative interactions, and genomic context determine occupancy and expression. Whether the structure of the inferred genotype networks — their size, connectivity, and the robustness/evolvability conclusions — survives these idealizations has not been systematically assessed. In the other direction, scaling up remains an open goal: exhaustive empirical study of entire regulatory circuits, which high-throughput technologies are beginning to enable, to understand how network architecture at larger regulatory scales shapes robustness and evolvability. Subsequent empirical work by the same and adjacent groups refined the empirical GP map of TF-DNA recognition (Aguilar-Rodríguez et al. 2018) and compared the evolvability of transcriptional versus RNA-mediated regulation (Payne, Khalid and Wagner 2018), and massively parallel reporter assays (MPRAs) — experiments that measure the regulatory output of tens of thousands of DNA variants pooled in one assay — have begun to deliver in vivo binding-site landscapes (Westmann, Goldbach and Wagner 2024, 2026; Westmann and Wagner 2026), while Santos-Moreno et al. (2023) built the first experimentally mapped genotype networks of small synthetic regulatory circuits based on CRISPR interference (CRISPRi), a programmable gene-repression method. None of these, individually or jointly, discharges the two-part question, which a 2026 review (Mascolo et al.) still frames as an open challenge on the way to a global quantitative GP map for gene regulation.

Problem Statement

Determine, quantitatively and empirically, how the mutational robustness and evolvability of transcriptional regulation inferred from binding-site genotype networks depend on the choices and context of the mapping, and how they are organized at the scale of whole regulatory circuits. The question has two linked parts. Part (i) concerns how the structure of transcription-factor binding-site genotype networks — network size, connectivity, accessibility of novel phenotypes, and the resulting robustness and evolvability conclusions, including the synergism between the two — changes under each of the following four factors, quantified under one uniform standard (measurement, or a rigorous quantitative bound where measurement is infeasible): variation of the binding threshold; replacement of the E-score in vitro affinity proxy by direct affinity or in vivo occupancy measurements; modeling of sites as variable-length or overlapping rather than contiguous 8-nucleotide words; and inclusion of chromatin context and native in vivo regulatory architecture. Part (ii) concerns how the mapping is extended from individual TF binding-site repertoires to whole regulatory circuits by an exhaustive or systematically sampled genotype-to-phenotype map, and how circuit-scale network architecture — the number and arrangement of sites, TFs, and regulatory interactions — shapes the robustness and evolvability of gene expression phenotypes. The two parts serve one objective: a validated, context-aware empirical account of regulatory robustness and evolvability at circuit scale, of which the 2014 single-site networks are the provenance and benchmark.

References

  1. Payne JL, Wagner A. The Robustness and Evolvability of Transcription Factor Binding Sites. Science 343:875-877 (2014). DOI: 10.1126/science.1249046. https://doi.org/10.1126/science.1249046
  2. Ciliberti S, Martin OC, Wagner A. Innovation and robustness in complex regulatory gene networks. Proceedings of the National Academy of Sciences USA 104:13591-13596 (2007). DOI: 10.1073/pnas.0705396104. https://doi.org/10.1073/pnas.0705396104
  3. Payne JL, Wagner A. Mechanisms of mutational robustness in transcriptional regulation. Frontiers in Genetics 6:322 (2015). DOI: 10.3389/fgene.2015.00322. https://doi.org/10.3389/fgene.2015.00322
  4. Tuğrul M, Paixão T, Barton NH, Tkačik G. Dynamics of Transcription Factor Binding Site Evolution. PLOS Genetics 11:e1005639 (2015). DOI: 10.1371/journal.pgen.1005639. https://doi.org/10.1371/journal.pgen.1005639
  5. Jiménez A, Cotterell J, Munteanu A, Sharpe J. Dynamics of gene circuits shapes evolvability. Proceedings of the National Academy of Sciences USA 112:2103-2108 (2015). DOI: 10.1073/pnas.1411065112. https://doi.org/10.1073/pnas.1411065112
  6. Aguilar-Rodríguez J, Peel L, Stella M, Wagner A, Payne JL. The architecture of an empirical genotype-phenotype map. Evolution 72:1242-1260 (2018). DOI: 10.1111/evo.13487. https://doi.org/10.1111/evo.13487
  7. Payne JL, Khalid F, Wagner A. RNA-mediated gene regulation is less evolvable than transcriptional regulation. Proceedings of the National Academy of Sciences USA 115:E3481-E3490 (2018). DOI: 10.1073/pnas.1719138115. https://doi.org/10.1073/pnas.1719138115
  8. Schaerli Y, Jiménez A, Duarte JM, Mihajlovic L, Renggli J, Isalan M, Sharpe J, Wagner A. Synthetic circuits reveal how mechanisms of gene regulatory networks constrain evolution. Molecular Systems Biology 14:e8102 (2018). DOI: 10.15252/msb.20178102. https://doi.org/10.15252/msb.20178102
  9. Santos-Moreno J, Tasiudi E, Kusumawardhani H, Stelling J, Schaerli Y. Robustness and innovation in synthetic genotype networks. Nature Communications 14:2454 (2023). DOI: 10.1038/s41467-023-38033-3. https://doi.org/10.1038/s41467-023-38033-3
  10. Khetan S, Carroll BS, Bulyk ML. Multiple overlapping binding sites determine transcription factor occupancy. Nature 646:1001-1011 (2025). DOI: 10.1038/s41586-025-09472-3. https://doi.org/10.1038/s41586-025-09472-3
  11. Westmann CA, Goldbach L, Wagner A. Entangled adaptive landscapes facilitate the evolution of gene regulation by exaptation. bioRxiv preprint (2024). DOI: 10.1101/2024.11.10.620926. https://doi.org/10.1101/2024.11.10.620926
  12. Westmann CA, Goldbach L, Wagner A. The adaptive landscapes of three global Escherichia coli transcriptional regulators. eLife 14:RP103774 (2026). DOI: 10.7554/eLife.103774. https://doi.org/10.7554/eLife.103774
  13. Westmann CA, Wagner A. Regulatory scope shapes the adaptive landscapes of bacterial transcription factor binding sites. bioRxiv preprint (2026). DOI: 10.64898/2026.08.11.744152. https://doi.org/10.64898/2026.08.11.744152
  14. Mascolo E, Borbély R, Borst NO, Barton NH, Crocker J, Tkačik G. Long-term evolution of regulatory DNA sequences. Part 2: theory and future challenges. Current Opinion in Genetics and Development 98:102472 (2026). DOI: 10.1016/j.gde.2026.102472. https://doi.org/10.1016/j.gde.2026.102472