Evolution of DNA replication origin specification and gene silencing mechanisms

Abstract

DNA replication in eukaryotic cells initiates from replication origins that bind the Origin Recognition Complex (ORC). Origin establishment requires well-defined DNA sequence motifs in Saccharomyces cerevisiae and some other budding yeasts, but most eukaryotes lack sequence-specific origins. A 3.9 Å structure of S. cerevisiae ORC-Cdc6-Cdt1-Mcm2-7 (OCCM) bound to origin DNA revealed that a loop within Orc2 inserts into a DNA minor groove and an α-helix within Orc4 inserts into a DNA major groove. Using a massively parallel origin selection assay coupled with a custom mutual-information-based modeling approach, and a separate analysis of whole-genome replication profiling, here we show that the Orc4 α-helix contributes to the DNA sequence-specificity of origins in S. cerevisiae and Orc4 α-helix mutations change genome-wide origin firing patterns. The DNA sequence specificity of replication origins, mediated by the Orc4 α-helix, has co-evolved with the gain of ORC-Sir4-mediated gene silencing and the loss of RNA interference.

The human origin recognition complex is essential for pre-RC assembly, mitosis, and maintenance of nuclear structure

“…In humans and Drosophila , this N-terminal region is an unstructured IDR, required for ATP-independent chromatin recruitment of ORC and drives protein phase-separation with DNA in vitro ( Bleichert et al, 2018 ; Hossain et al, 2019 ; Parker et al, 2019 ). In yeast, this region of Orc1 has been shown to be critical for interacting with ARS DNA as well, but is completely structured unlike metazoan ORC1 ( Hu et al, 2020 ; Kawakami et al, 2015 ; Li et al, 2018 ). Another example is ORC6 in which the N-terminus harbors a structured TFIIB domain, but we noticed sgRNA depletions in the remaining regions, especially the C-terminal extremity, which is a septin-binding-region essential for cytokinesis and mutations in a conserved C-terminal motif have also been linked to Meier-Gorlin Syndrome ( Akhmetova et al, 2015 ; Balasov et al, 2020 ; Balasov et al, 2015 ; Balasov et al, 2007 ; Bicknell et al, 2011a ; Prasanth et al, 2002 ; Xu et al, 2020 ).…”

Conservation of a DNA Replication Motif among Phylogenetically Distant Budding Yeast Species

“…( C ) Alignment of a portion of Orc4 from the same species. The DNA-binding alpha-helix is highlighted gray, and the tyrosine (Y) that contacts nucleotides 14 and 15 in S. cerevisiae and likely nucleotides 12 and 13 in T. delbrueckii (red boxed in 3B) ( Hu et al 2020 ) is highlighted red.…”

“…Cryo-EM studies of the S. cerevisiae prereplication complex indicate that a loop in Orc2 and an alpha-helix in Orc4 make direct contacts with DNA in the extended ACS ( Yuan et al 2017 ). Moreover, genetic studies pinpoint tyrosine 486 in ScOrc4 as conferring specificity for nucleotides 14 and 15 in the ACS ( Hu et al 2020 ). These positions harbor A/T and G/T in S. cerevisiae , C. glabrata , and T. delbrueckii , but no particular nucleotide is found at these positions in the KLE ORC-binding sites ( fig.…”

MAVE-NN: learning genotype-phenotype maps from multiplex assays of variant effect

“…MAVE-NN supports two distinct implementations of our unified conceptutal approach: GE regression and measurement process agnostic (MPA) regression. GE regression is modeled after the approach of [46], while MPA regression resembles previously reported mutual information maximization analyses [34, 33, 17, 60]. In the following sections, we demonstrate the utility of MAVE-NN on previously published DMS and MPRA datasets.…”

“…Parallel work in the biophysics literature has focused on developing ways to infer G-P maps from highthroughput data in a manner that is fully agnostic to the quantitative form of the measurement process used to read out latent phenotypes [54, 55, 56, 17]. This approach, which focuses on the use of mutual information as an objective function, arose from techniques in sensory neuroscience [57, 58] that were elaborated and adapted for the analysis of microarray data [56, 59], then later applied to MPRAs [34, 33, 17, 18] and other MAVE experiments [60]. However such analyses of MAVE data have relied on Metropolis Monte Carlo, which (in our experience) is too slow to support widespread adoption.…”