Transcription is regulated at the molecular scale by highly dynamic events, including the binding of transcription factors (TFs) to regulatory genomic regions, fluctuations in chromatin conformation, and the assembly of macromolecular complexes. Nevertheless, our understanding of transcription has largely been driven by population and time-averaged snapshots. Though this has been successful in identifying the key players involved and their interaction networks, it has resulted in transcription and genome organization being likened to an assembly line, with a linear recruitment of key factors dictated by stable and hierarchical interactions. The ability to measure molecular-scale kinetics in living systems has revealed that transcription is far more akin to a chaotic marketplace, where dynamic transactions are critical.

Three key observations have emerged from live imaging, which have challenged our textbook models. First, live imaging of chromatin dynamics has provided counterintuitive and highly variable results. For example, a clear relationship between the proximity of enhancers and the level of activity at the promoters they regulate has yet to be established 1, 2, 3, 4. Single-molecule tracking has revealed that TFs generally bind their targets transiently on the order of just tens of seconds (Dynamics Database, URL: www.mir-lab.com/dynamics-database). These binding times are similar across a diverse range of TFs, suggesting that genomic occupancy is regulated by modulating binding frequency (on-rates), rather than duration (off-rates). Finally, mounting evidence suggests that proteomic microenvironments (condensates, hubs, clusters, etc. see definitions below) can tune the frequency of protein–DNA, protein–protein, and intrachromatin interactions to mediate regulatory processes. Microenvironments are thought to trap molecules through repetitive, weak, but selective interactions or through the creation of a physical boundary. The precise mechanisms of formation are likely diverse and perhaps even gene-specific.

We use the term ‘proteomic microenvironments’ here as a general term for distinct regions characterized by the local enrichment of proteins. While condensates, hubs, and clusters are all used to describe such local enrichments, we use microenvironments here for several reasons. The term ‘condensate’ has become synonymous with liquid–liquid phase separation and thus implies specific physical characteristics. While ‘hub’ is agnostic to physical properties and is often used to distinguish from phase separation, it does imply a functional role as a focal point for activity. Although ‘cluster’ is an agnostic term defined as a higher local density than expected by chance as characterized using statistical methods, it does not capture the complexity of the nuclear microenvironments we discuss here. In contrast, ‘microenvironment’ is agnostic to both form and function, and it implies a distinct and compositionally complex region within the nucleoplasm. Finally, although nucleic acids are ubiquitous in the nucleus and are almost certainly present within most microenvironments, we use ‘proteomic’ here to direct attention to how protein–protein interactions specifically shape these local environments. In this paper, we will, however, use ‘condensate’, ‘hub’, ‘cluster’, and other terms when describing specific examples to match their original use in the primary literature.

A large amount of recent work suggests that proteomic microenvironments generate conditions in which transient molecular interactions occur with high-enough frequency to generate functional outputs and may resolve some of the confusion that has arisen from the discovery that nuclear processes are not stable and hierarchical. However, the observations above highlight that we have only begun to understand the forces in living systems that establish the right local proteomic and genomic microenvironments to regulate genes. Here, we discuss the technologies that have enabled these discoveries and summarize recent findings. Finally, we propose how physical properties of gene-specific microenvironments, not just the differential recruitment of TFs, may give rise to distinct functional outcomes.