Adaptive Stress Response Through a Flexible Condensate Cascade

Cells must respond to a broad, ever-changing spectrum of environmental insults, yet cannot afford a dedicated signaling pathway for every possible stress. This dissertation asks how a single stress-response system achieves both flexibility and selectivity, using the yeast heat shock response (HSR) as a model.

At its core, this work establishes that HSR activation proceeds through a flexible con- densate cascade: heat shock drives sequential condensation of orphan ribosomal proteins, the co-chaperone Sis1, the chaperone Hsp70, the transcription factor Hsf1, the Mediator coactivator, and RNA polymerase II. Five mechanistically distinct proteotoxic stresses, heat shock, arsenite, the proline analog AZC, the ribosome biogenesis inhibitor DZA, and hy- perosmotic NaCl, each engage this cascade, but do so at different entry points and with different efficiencies, converging on Sis1 condensation as a shared, obligate bottleneck. De- spite broadly overlapping condensation of the chaperone-sensing machinery across all five stresses, only heat shock and arsenite productively activate HSR transcription, showing that condensation of the sensing arm is necessary but not sufficient for output. Transcriptional activation is instead gated downstream, at the level of Hsf1 and the transcriptional machin- ery itself. Nuclear depletion of Sis1 further demonstrates that, rather than acting as an activating relay, Sis1 functions as a rheostatic repressor of Hsf1 whose influence is titrated in proportion to stress severity — allowing the chaperone network to respond flexibly to diverse insults while insulating the core transcriptional program from spurious activation. This convergent-cascade architecture reframes the HSR not as a pleiotropic responder to any proteotoxic insult, but as a selectively gated program that grants heat and other proteostasis- collapsing stresses privileged access while still sensing a much broader range of chemically distinct challenges.

This convergent-cascade model builds on and extends a foundational finding: that the downstream, transcription-proximal half of the HSR condensate cascade itself assembles in an ordered, stepwise fashion rather than through concerted, all-or-none recruitment. Live- cell imaging and transcriptomics show that Hsf1 first clusters upon partial release from Hsp70; phosphorylation and further Hsp70 dissociation then permit the coactivator Me- diator to partition into these clusters; and Mediator-bound Hsf1 assemblies finally recruit RNA polymerase II. This hierarchical assembly produces graded, rather than switch-like, transcriptional output, and fully assembled condensates drive an adaptive three-dimensional reorganization of the yeast genome that is dispensable for transcription itself but confers a measurable fitness benefit — establishing condensate architecture, and not merely conden- sate presence, as a regulated determinant of cellular behavior, and providing the mechanistic scaffold onto which the upstream cascade is built.

Finally, this dissertation asks whether the logic of pathway-level prioritization scales up to how cells triage many simultaneous environmental demands. Profiling single-cell transcrip- tomes across twenty combinatorial environments — varying carbon source, nutrient rich- ness, osmolarity, temperature, and reactive oxygen species — and resolving this data with a spectral mutual-information method reveals that environmental information is organized hierarchically across the transcriptome: variation attributable to carbon source dominates the shallowest modes of transcriptional variation, followed in order by nutrient richness, osmolarity, temperature, and oxidative stress. This statistical hierarchy predicts, and is borne out by, physiological prioritization: under combined heat and osmotic stress, cells preferentially mount the osmotic-stress program and suppress the heat-shock response, a prioritization encoded within individual cells rather than emerging only at the population level. Mechanistically, this prioritization is implemented through differential translation ini- tiation — osmotic stress collapses polysomes more severely than heat shock, preventing the accumulation of orphan ribosomal proteins that would otherwise sequester Sis1 and Hsp70 away from Hsf1 — directly linking environmental priority routing to the same condensate cascade described above, and suggesting that a common molecular vocabulary underlies both pathway-level and organism-level stress prioritization.