In cellular systems, biophysical interactions between macromolecules underlie a complex web of functional interactions. How biophysical and functional networks are coordinated, whether all biophysical interactions correspond to functional interactions, and how such biophysical-versus-functional network coordination is shaped by evolutionary forces are all largely unanswered questions. Here, we investigate these questions using an "inter-interactome" approach. We systematically probed the yeast and human proteomes for interactions between proteins from these two species and functionally characterized the resulting inter-interactome network. After a billion years of evolutionary divergence, the yeast and human proteomes are still capable of forming a biophysical network with properties that resemble those of intra-species networks. Although substantially reduced relative to intra-species networks, the levels of functional overlap in the yeast-human inter-interactome network uncover significant remnants of co-functionality widely preserved in the two proteomes beyond human-yeast homologs. Our data support evolutionary selection against biophysical interactions between proteins with little or no co-functionality. Such non-functional interactions, however, represent a reservoir from which nascent functional interactions may arise.

Hsp104, a yeast protein-remodeling factor of the AAA+ (ATPases associated with various cellular activities) superfamily, and its homologs in bacteria and plants mediate cell recovery after severe stress by disaggregating denatured proteins through a poorly understood mechanism. Here, we present cryo-electron microscopy maps and domain fitting of Hsp104 hexamers, revealing an unusual arrangement of AAA+ modules with the prominent coiled-coil domain intercalated between the AAA+ domains. This packing results in a greatly expanded cavity, which is capped at either end by N- and C-terminal domains. The fitted structures as well as mutation of conserved coiled-coil arginines suggest that the coiled-coil domain plays a major role in the extraction of proteins from aggregates, providing conserved residues for key functions in ATP hydrolysis and potentially for substrate interaction. The large cavity could enable the uptake of polypeptide loops without a requirement for exposed N or C termini.

The efficacy of hormonal therapies for advanced estrogen receptor-positive breast cancers is limited by the nearly inevitable development of acquired resistance. Efforts to block the emergence of resistance have met with limited success, largely because the mechanisms underlying it are so varied and complex. Here, we investigate a new strategy aimed at the very processes by which cancers evolve resistance. From yeast to vertebrates, heat shock protein 90 (HSP90) plays a unique role among molecular chaperones by promoting the evolution of heritable new traits. It does so by regulating the folding of a diverse portfolio of metastable client proteins, many of which mediate adaptive responses that allow organisms to adapt and thrive in the face of diverse challenges, including those posed by drugs. Guided by our previous work in pathogenic fungi, in which very modest HSP90 inhibition impairs resistance to mechanistically diverse antifungals, we examined the effect of similarly modest HSP90 inhibition on the emergence of resistance to antiestrogens in breast cancer models. Even though this degree of inhibition fell below the threshold for proteotoxic activation of the heat-shock response and had no overt anticancer activity on its own, it dramatically impaired the emergence of resistance to hormone antagonists both in cell culture and in mice. Our findings strongly support the clinical testing of combined hormone antagonist-low-level HSP90 inhibitor regimens in the treatment of metastatic estrogen receptor-positive breast cancer. At a broader level, they also provide promising proof of principle for a generalizable strategy to combat the pervasive problem of rapidly emerging resistance to molecularly targeted therapeutics.

The heat shock protein 90 (Hsp90) chaperone functions as a protein-folding buffer and plays a role promoting the evolution of new heritable traits. To better understand how Hsp90 can affect mRNA translation, we screen more than 1,600 factors involved in mRNA regulation for physical interactions with Hsp90 in human cells. The mRNA binding protein CPEB2 strongly binds Hsp90 via its prion domain. In a yeast model, transient inhibition of Hsp90 results in persistent activation of a CPEB translation reporter even in the absence of exogenous CPEB that persists for 30 generations after the inhibitor is removed. Ribosomal profiling reveals that some endogenous yeast mRNAs, including HAC1, show a persistent change in translation efficiency following transient Hsp90 inhibition. Thus, transient loss of Hsp90 function can promote a nongenetic inheritance of a translational state affecting specific mRNAs, introducing a mechanism by which Hsp90 can promote phenotypic variation.

The ε4 allele of apolipoprotein E (APOE4) is a genetic risk factor for many diseases, including late-onset Alzheimer's disease (AD). We investigate the cellular consequences of APOE4 in human iPSC-derived astrocytes, observing an endocytic defect in APOE4 astrocytes compared with their isogenic APOE3 counterparts. Given the evolutionarily conserved nature of endocytosis, we built a yeast model to identify genetic modifiers of the endocytic defect associated with APOE4. In yeast, only the expression of APOE4 results in dose-dependent defects in both endocytosis and growth. We discover that increasing expression of the early endocytic adaptor protein Yap1802p, a homolog of the human AD risk factor PICALM, rescues the APOE4-induced endocytic defect. In iPSC-derived human astrocytes, increasing expression of PICALM similarly reverses endocytic disruptions. Our work identifies a functional interaction between two AD genetic risk factors-APOE4 and PICALM-centered on the conserved biological process of endocytosis.

Parkinson disease (PD) has useful symptomatic treatments that do not slow the neurodegenerative process, and no significant disease-modifying treatments are approved. A key therapeutic target in PD is α-synuclein (αS), which is both genetically implicated and accumulates in Lewy bodies rich in vesicles and other lipid membranes. Reestablishing αS homeostasis is a central goal in PD. Based on previous lipidomic analyses, we conducted a mouse trial of a stearoyl-coenzyme A desaturase (SCD) inhibitor ("5b") that prevented αS-positive vesicular inclusions and cytotoxicity in cultured human neurons.