Learning the dynamic organization of a replicating bacterial chromosome from time-course Hi-C data

Bacterial chromosomes are in continual motion as they undergo concurrent transcription, replication, and segregation. Time-course Hi-C experiments hold promise for studying chromosome organization across the cell cycle, but interpreting Hi-C data from dynamic systems remains challenging. Here, we develop a fully data-driven 4D Maximum Entropy approach to extract a model for the dynamic organization of a replicating bacterial chromosome directly from time-course Hi-C and microscopy data. After validating our 4D data-driven model model forCaulobacter crescentusagainst independent microscopy data, we infer quantitative information about changes in chromosome organization across the bacterial replication cycle. Our model reveals a sustained global linear organization of theC. crescentuschromosome during replication, as well as dynamic patterns of local chromosome extension induced by the replication forks. We use these data-driven inferences to constrain a mechanistic model for a replicating bacterial chromosome. Our model demonstrates that origin-pulling by a ParABS-like system, together with loop extrusion by condensin, can explain our inferred large-scale chromosome segregation patterns. The inferred replication-induced local changes in chromosome compaction, however, require additional mechanisms, which we attribute to replication-induced NAP unbinding and positive supercoiling. Overall, our work introduces a rigorous data-driven framework for quantitatively interpreting time-course Hi-C data, and offers new mechanistic insights into bacterial chromosome organization across the cell cycle.