The 2024 study by Beknazarov, Konovalov, Herbert and Poptsova associated Z-DNA-forming flipon sequences (Z-flipons) conserved between human and mouse with increased experimentally measured transcription reinitiation rates and proposed that Z-DNA formation actuates a promoter reset, but its evidence was aggregated across tissues and purely associative. The open problem is to determine which Z-flipons are actually formed in which tissues, developmental stages, cell-cycle phases and individual cells — including parent-of-origin-specific activity at imprinted genes — and to test causally whether Z-flipon conformational switching controls transcription reinitiation, supported by orthogonal in vivo Z-DNA assays free of detection artifacts.

Background

Most genomic DNA adopts the right-handed B-form double helix, but alternating purine–pyrimidine repeats, especially (CG)n, can flip into the left-handed Z-conformation (Z-DNA), a transition that absorbs the negative supercoiling generated ahead of RNA polymerase during transcription. Herbert coined the term flipon for sequence modules that reversibly switch conformation; Z-flipons are the Z-DNA-forming subset. Formation of Z-DNA in living cells is inferred from binding of Zalpha domains (protein modules specific for the Z-conformation, found in the editors ADAR1 and ZBP1), from anti-Z-DNA antibodies such as Z22 (used in ChIP-seq), and from chemical footprinting of the single-stranded DNA at B–Z junctions: KEx (potassium permanganate/S1 nuclease footprinting) and KAS-seq (kethoxal labeling of single-stranded DNA) map transiently unwound DNA on a seconds-to-minutes timescale in intact cells. CUT&Tag (cleavage under targets and tagmentation) is an in situ antibody-guided mapping assay that had not been applied to Z-DNA at the time of the source study. DeepZ (Beknazarov et al. 2020) is a deep-learning model that predicts Z-flipon locations genome-wide from DNA sequence, Z-Hunt energetics (a thermodynamic estimator of a sequence's propensity to form Z-DNA under supercoiling), and omics features (histone marks, transcription factors, DNase accessibility, RNA polymerase occupancy); Z-DNABERT is an orthogonal transformer-based predictor used for validation. Beknazarov, Konovalov, Herbert and Poptsova (2024) trained DeepZ separately on human and mouse data, extracted Z-flipons conserved between the two genomes, and validated them against chemical mapping and Z-DNABERT. They found conserved Z-flipons enriched in alternative and bidirectional promoters of neurogenesis and developmental genes, associated with topoisomerases and with GTF2E2 (the beta subunit of general Transcription Factor E), and — using genome-wide transcription-kinetics measurements from the Cramer laboratory — associated with increased transcription reinitiation rates, i.e. the rate at which a promoter re-assembles its preinitiation complex (the assembly of general transcription factors and RNA polymerase II required to start transcription) for successive rounds of initiation, without effects on elongation or pausing. They proposed a model in which Z-flipon switching actuates a promoter reset, capturing energy stored in negative supercoiling and enabling reassembly of the transcription machinery. Related studies have not yet addressed the conserved-flipon question: tissue-specific flipon prediction has been demonstrated only for G-quadruplex flipons (DeepGQ; Bashkatov et al. 2025), Z-DNA CUT&Tag has been realized only in rice (He et al. 2025), single-cell and imprinting studies of Z-DNA are absent, and the first direct mammalian perturbation evidence — engineered Z-DNA promoters in the MYC–FACT context (Wan et al. 2025) — does not address the conserved-flipon reinitiation model. Crucially, this model rests on omics data aggregated across tissues and on associations rather than perturbations.

Problem Statement

Which Z-flipons — in particular those conserved between human and mouse — are actually formed in which biological contexts, and does their formation causally control transcription reinitiation as proposed? Concretely: (1) Cell-state resolution: define sets of tissue-specific and developmental-stage-specific Z-flipons (including embryonic and extraembryonic stages) from tissue- and stage-specific omics data rather than cross-tissue aggregates, and couple these sets to tissue-specific alternative promoter usage; determine how Z-flipon formation varies across cell-cycle phases and within individual cells; and determine whether Z-flipons act in a parent-of-origin-specific manner at imprinted genes. (2) Causality: test whether the association between conserved Z-flipons and increased transcription reinitiation rates is causal — whether Z-DNA formation at these promoters drives the proposed reset of preinitiation complexes (engagement of GTF2E2/Transcription Factor E, topoisomerase-tuned supercoiling) — distinguishing the effect of the Z-forming conformation from that of the underlying DNA sequence. (3) Enabling measurements: develop and apply orthogonal, high-resolution in vivo assays for Z-DNA that control for the artifactually inducing, stabilizing, or topology-altering effects of antibody- and Zalpha-domain-based detection. Together, these directions would convert an aggregate, associative flipon map into a cell-state-resolved, mechanistically validated account of Z-flipon function in transcription regulation. All three directions are jointly required to resolve the open problem.

References

  1. Beknazarov N, Konovalov D, Herbert A, Poptsova M (2024). Z-DNA formation in promoters conserved between human and mouse are associated with increased transcription reinitiation rates. Scientific Reports 14:17786. DOI 10.1038/s41598-024-68439-y. https://doi.org/10.1038/s41598-024-68439-y
  2. Beknazarov N, Jin S, Poptsova M (2020). Deep learning approach for predicting functional Z-DNA regions using omics data. Scientific Reports 10:19134. DOI 10.1038/s41598-020-76203-1. https://doi.org/10.1038/s41598-020-76203-1
  3. Bashkatov A, Andreasyan A, Konovalov D, Herbert A, Poptsova M (2025). Deep learning deciphers the related role of master regulators and G-quadruplexes in tissue specification. Scientific Reports 15:23119. DOI 10.1038/s41598-025-07579-1. https://doi.org/10.1038/s41598-025-07579-1
  4. Wan X, Fan H, Li Z, Du Y, Liu Y, Zeng W, Sun Z, Shao J, Jia J, Zhang Y, Fang D (2025). MYC drives left-handed Z-DNA formation to shape gene expression. Nature Communications 17:199. DOI 10.1038/s41467-025-66886-3. https://doi.org/10.1038/s41467-025-66886-3
  5. Shajahan S, Loe-Mie Y, Asselin L, Salmon-Legagneur M, Traboulsi T, Thierry A, Dejean A, Cossec J-C (2026). Z-DNA formation regulates the totipotent-like state and primes Zscan4-dependent chromatin compartmentalization. Nature Structural & Molecular Biology 33(4):677-690. DOI 10.1038/s41594-026-01751-5. https://doi.org/10.1038/s41594-026-01751-5
  6. He Z, Run Y, Feng Y, Yang Y, Tavakoli M, Ahmed A, Ariel F, Zhang W (2025). Global identification and functional characterization of Z-DNA in rice. Plant Biotechnology Journal 23(4):1277-1290. DOI 10.1111/pbi.14585. https://doi.org/10.1111/pbi.14585
  7. Run Y, Tavakoli M, Zhang Y, Vasquez KM, Zhang W (2026). Formation and biological implications of Z-DNA. Trends in Genetics 42(2):163-176. DOI 10.1016/j.tig.2025.07.006. https://doi.org/10.1016/j.tig.2025.07.006
  8. Sahayasheela VJ, Ooga M, Kumagai T, Sugiyama H (2025). Z-DNA at the crossroads: untangling its role in genome dynamics. Trends in Biochemical Sciences 50(3):267-279. DOI 10.1016/j.tibs.2025.01.001. https://doi.org/10.1016/j.tibs.2025.01.001
  9. Lee CC, Hsu SF, Chang YW, Chen YW, Ho MR, Sugiyama H, Wang WC, Wang AH (2026). Structural basis of B-to-Z DNA transition mediated by an anti-Z-DNA antibody. Nucleic Acids Research 54(5):gkag160. DOI 10.1093/nar/gkag160. https://doi.org/10.1093/nar/gkag160