Proceedings of the National Academy of Sciences

Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) is foundational to life on Earth, catalyzing carbon dioxide (CO2) fixation to generate biomass. However, Rubisco is a slow and inefficient enzyme that has proven challenging to engineer. We applied the structure-informed machine learning (ML) model ESM-IF1 to identify plausible amino acid sites in the large subunit of Nicotiana tabacum Rubisco to target for directed evolution. ML-assisted library design followed by selection in Rubisco-dependent Escherichia coli identified multiple enriched variants displaying improved catalytic efficiency. Several improved variants carried amino acid changes not found in the evolutionary lineage of plants, despite being assembly competent in plant chloroplasts, demonstrating that ML-assisted protein design can explore functional sequence space beyond what is observed from natural sequence diversity. Most prominently, the T391I substitution improved carboxylation rate by 29% and aerobic carboxylation efficiency by 43%. Our findings demonstrate the utility of ML-assisted evolution for engineering Rubisco with improved carboxylation efficiency and potential for enhancing crop productivity.

Biophysical pleiotropy, the phenomenon in which a mutation affects multiple protein properties, underlies many genetic diseases and shapes protein evolution, yet remains poorly understood, limiting synthetic biology and therapeutic development. Although extensively studied in soluble proteins, little is known about how pleiotropic effects shaped the evolution of receptor protein functions. Here, we developed a cell-based assay designed to assess pleiotropic effects in G protein-coupled receptors (GPCRs), the largest receptor class in eukaryotes. Because our assay design allows for multiple GPCR molecular phenotypes (basal activity, ligand-dependent signaling, and cellular receptor abundance) to be simultaneously assessed, we applied it to study rhodopsin, a visual GPCR that evolved high light sensitivity by maximizing light-driven responses while minimizing thermally driven noise. To investigate this, we integrated our assay with deep mutational scanning of relevant rhodopsin domains to test the impacts of [~]2,000 single-residue substitutions in rhodopsin across all three molecular phenotypes ([~]6,000 measurements). By analyzing these data in the context of rhodopsins protein structures, we discovered complex pleiotropic effects that act asymmetrically to impose joint constraints, restricting mutational tolerance in rhodopsins retinal binding pocket and G protein interaction interface. A dose-response analysis of rhodopsin signaling revealed that these pleiotropic constraints arise from the dual functions of its chromophore, which acts as an agonist upon light-driven isomerization but also an inverse agonist in darkness. Because these constraints reflect mechanistically driven limitations in the evolution of receptor signaling, these findings reveal a role for biophysical pleiotropy in shaping the sensory capabilities of receptor proteins. Significance StatementUnderstanding how genetic variation impacts protein function is an important but challenging area of research as many mutations are pleiotropic, simultaneously affecting multiple protein properties, including structure, stability, and activity. Previously, genetic screens have systematically mapped pleiotropic effects in soluble proteins, but transmembrane receptor proteins embedded in cellular membranes are much more difficult to assay. These are key receptors that convert environmental stimuli into appropriate physiological responses. Here, we present a systematic study of pleiotropic effects within the receptor responsible for dim-light vision in vertebrates, revealing how light sensitivity can be mediated through differential conformational states of its chromophore. This is important not only for understanding how receptors evolved enhanced capabilities, but also for the development of tunable protein systems.

Complex signaling pathways organize cell expansion and proliferation across cells to pattern tissues and organs in plants. The sulfotyrosine peptide hormone family, PLANT PEPTIDE CONTAINING SULFATED TYROSINE (PSY), contributes to these processes. We identified two plasma membrane-localized receptors, PSYR1 and PSYR2, that are necessary for PSY signaling and regulate growth in Physcomitrium patens. Membrane-associated PSYRs accumulate to high levels in a mutant lacking TYROSYL PROTEIN SULFOTRANSFERASE (TPST). Given that a tpst null mutant ({Delta}tpst) is impaired in sulfation, this suggests that in the absence of sulfated peptides, PSYRs accumulate on the membrane. A null mutant of the PSY receptors, {Delta}psyr1/2, showed increased growth and was epistatic to {Delta}tpst, suppressing defects in gametophore formation and early senescence. The transcriptional profiles comparing wild type to {Delta}psyr1/2 and {Delta}psyr1/2/{Delta}tpst showed 25 to 30 differentially expressed genes between the receptor null mutants and wild type, with a common signature of cell wall remodeling and stress responses. Similarly, a PSYR1 kinase-inactive mutation rescued {Delta}tpst and relieved the accumulation of membrane-associated PSYRs. In contrast, overexpression of PSYRs inhibited plant growth, with phenotypic severity correlating with the amount of overexpression. These data are consistent with a constitutive activation model in which membrane-associated PSYRs unbound to PSY serve to inhibit growth through an active kinase. In the presence of the PSY peptide, the kinase is inactivated, promoting growth and driving PSY expression. The relationship between growth-repressive PSYR kinase activity and growth-promoting PSYR kinase inactivation in P. patens serves as a model for optimizing plant growth and development.

It is an open question whether variation in the genetic load of unconditionally deleterious mutations contributes substantially to the variability in human disease. Here, we solve for mutation-selection-drift balance and predict variation in genetic load given a realistic human genome-wide deleterious mutation rate, U, and a distribution of fitness effects (DFE). Empirical estimates of U come from sequence constraint, which fails to count slightly deleterious mutations that nevertheless fix. We use the inferred DFE to correct for this and conclude that total human U>3.8. Two humans typically differ in ancestral fitness by 17-33% given uncertainty in U, or by 6-49% when we consider a broad range of alternative DFEs. Results are similar for other species with larger mean selection coefficients, such as other mammals. Most variation in load comes from rare variants with frequencies below 1%, with a substantial fraction coming from ultra-rare variants below 0.01%. This could help explain why some of the heritability observed in pedigree studies is missing from genome-wide association studies. Accounting for rare and ultra-rare variants, e.g., via variant-effect prediction of unique mutations from whole-genome sequencing rather than via polygenic risk scores, could help identify individuals at high risk of disease. SignificanceMany human mutations mildly disrupt molecular function, e.g., by destabilizing proteins. Having too many of these mutations would have reduced fitness in ancestral human environments and might contribute to disease today. Here, we mathematically derive how much variation in fitness such mutations cause, using estimated human parameter values. Rare variants with larger fitness effects contribute the most. Identifying individuals with high disease risk likely requires methods capable of scoring rare variants.

Cohesin is a conserved multiprotein complex (SMC1, SMC3, RAD21, and either STAG1 or STAG2) that organizes three-dimensional genome architecture and regulates chromosome segregation, gene expression, and DNA damage repair (1-4). Following double-strand breaks (DSBs), cohesin is recruited to sites of DNA damage - a process considered essential for efficient homologous recombination (5-15). Yet how DSB signaling elicits cohesin recruitment and subsequent cohesion establishment remains poorly understood. Here we show that telomere replication stress activates de novo STAG2-cohesin loading, thereby promoting break-induced telomeric DNA repair. We demonstrate that this DNA break-elicited cohesin recruitment is strictly controlled by the BRCT domain-containing DNA damage recognition factor PAXIP1 and its functional partner PAGR1. Cryo-electron microscopy structure reveals that PAGR1, together with PAXIP1, physically binds to a composite interface formed by the STAG2-RAD21 cohesin subcomplex. Complementary mutational and biochemical analyses define the molecular basis of this interaction and establish its essential role in break-induced cohesion establishment. Furthermore, we show that PAXIP1-PAGR1-enacted STAG2-cohesin recruitment complements with the PML body-associated pathway in orchestrating break-induced alternative lengthening of telomeres (ALT). Concurrent depletion of PML together with PAXIP1, PAGR1 or STAG2 disrupts ALT-mediated telomere maintenance, leading to end-to-end chromosomal fusion and mitotic cell death. Collectively, these findings uncover a distinctive molecular mechanism through which DSB signaling directs de novo cohesion establishment, and highlight its critical importance in break-induced telomere repair.

Ocean warming and marine heatwaves threaten seagrasses, which are considered highly vulnerable to rising temperatures. We compared thermal stress responses of the tropical seagrass Halophila stipulacea collected from the Gulf of Aqaba in 2017 and 2022. Under mesocosm-applied thermal stress, 2017 plants showed reduced growth, impaired photochemistry, and extensive transcriptomic changes. In contrast, 2022 plants maintained normal growth and photochemistry with minimal transcriptional responses. The enhanced thermal stress tolerance of 2022 plants was associated with a "stress-ready" transcriptome, with many stress-response genes already up/downregulated under control conditions. These changes coincided with local ocean warming and a shift from episodic to chronic thermal stress. Our findings provide the first longitudinal dataset in a seagrass that tracks physiological and molecular changes occurring concurrently with documented ocean warming.

High pressure (HP) can trigger bacterial spore germination, acting either through germinant receptors (GRs) or the SpoVA channel. However, the mechanism by which HP activates these membrane-embedded proteins remains elusive. Here, using Bacillus subtilis, we demonstrate that the GerA germinant receptor (GR) is the primary target of moderate HP (50-300 MPa). Mutagenesis reveals that pore-lining residues within the GerA ion channel are essential for the pressure response, whereas canonical ligand-binding and intramembrane signaling residues are dispensable. We then propose a <underline>s</underline>tretch-<underline>t</underline>o-<underline>o</underline>pen (STO) model, in which HP differentially compresses the more compliant inner membrane (IM) relative to the rigid spore core, generating lateral membrane tension that promotes opening of the GerA channel. In situ membrane tension measurements indicate HP-induced compression of IM phospholipids and elevated membrane tension. This tension-dependent gating is further supported by the pressure-dependent phenotypic rescue of GerA channel mutants. Consistently, HP increases IM permeability to water-soluble and membrane-impermeable agents (propidium iodide and formaldehyde), an effect potentiated by GerA, indicating concomitant opening of GerA by HP. Furthermore, modulating IM fluidity via heat activation or decoating altered membrane physical properties and delayed HP-induced germination, establishing the IM as the critical mechanical transducer. Additionally, computational modeling and calculations support faster compression of the IM than of the core under HP, rationalizing the source of tensile stress. Together, our findings establish a novel mechanism of HP-induced GerA activation via the STO model: HP compresses the IM, generates lateral tension, and promotes opening of the GerA ion channel to trigger bacterial spore germination.

The NAD(P)H dehydrogenase-like complex, NDH-1MS, is a constitutively expressed CO2-concentrating mechanism critical for sustaining CO2 fixation in cyanobacteria. This complex has been proposed to function as a vectorial carbonic anhydrase and accumulates intracellular bicarbonate against chemical equilibrium by coupling CO2 hydration to the photosynthetic cyclic electron flow. Using cryo-electron microscopy, we determined the structure of NDH-1MS from Thermosynechococcus vestitus. Contrary to its inducible counterpart NDH-1MS, our data suggest that a candidate CO2 hydration site may reside in the carbon-concentrating subunit CupB instead of the CupB-NdhF4 interface previously proposed. The active-site region partially resembles metal ion-independent iota-carbonic anhydrases and may bind CO2 or bicarbonate. We further observe features consistent with a Grotthuss-type proton transfer network connecting the active-site region to the antiporter-like subunits NdhF4 and NdhD4; however, unlike complex I, the proton export channel in NdhF4 appears to be blocked by bulky hydrophobic amino acids. Structural comparison between the oxygenic photosynthesis-specific subunit NdhV-bound state and dissociated state further reveals a correlation between plastoquinone stability and NdhV association. Taken together, we propose a revised working model in which NDH-1MS may function as a metal ion-independent vectorial carbonic anhydrase, with CO2 hydration coupled to proton transfer events in the antiporter-like subunits across the thylakoid membrane. SignificancePhotosynthetic complex I-like CO2-uptake systems are central to cyanobacterial carbon fixation, yet the molecular basis for coupling CO2 hydration to electron transfer has remained unresolved. Here, structures of NDH-1MS in NdhV-bound and NdhV-free states revise the architectural framework for catalysis and coupling in the constitutive CO2-uptake module. The data identify a CupB-centered candidate active-site region, reveal a non-canonical terminal antiporter-like subunit with an apparently blocked P-side exit, and link NdhV association to altered plastoquinone stability. Together, these findings refine current models for vectorial carbonic anhydrase activity in cyanobacterial NDH-1 and provide a structural basis for understanding how photosynthetic redox chemistry may be coupled to inorganic carbon concentration.

Staphylococcus aureus is a species of Gram-positive bacteria that uses the accessory gene regulator (agr) quorum sensing system to regulate many aspects of cell behavior such as virulence and toxin production. Quorum sensing is activated in dense S. aureus populations by the extracellular accumulation of a signaling molecule called the autoinducing peptide (AIP). While the biosynthetic enzymes that produce AIP have been elucidated in certain subgroups of S. aureus, the mechanism of AIP secretion is unknown. Here, we show that the biosynthetic intermediate known as AgrD (1-32) thiolactone is actively exported into the surrounding environment in certain S. aureus subgroups. Using a high-throughput transposon mutagenesis screen, we identify the integral membrane transporters, AbcA and PmtCD, as being involved in the biosynthesis of certain AIPs. Subsequent genetic complementation and biochemical experiments reveal that these proteins actively transport AgrD (1-32) thiolactone out of the cell. This enables the production of AIP directly on the cell surface following the action of MroQ, an integral membrane protease demonstrated here to possess an active site facing the extracellular milieu. Furthermore, we show that media supplementation with catalytically-active MroQ nanodiscs rescues AIP production in an mroQ knockout strain. This observation suggests the possibility that some proportion of AgrD (1-32) thiolactone exported by one cell may travel to neighboring cell surfaces for further processing into the diffusive AIP molecule. Collectively, our studies shed light on a poorly understood step in the agr quorum sensing pathway with clear implications for the design of novel therapeutic strategies.

Hydathodes at leaf margins mediate guttation of xylem-derived fluids and serve as primary entry sites for adapted vascular bacterial pathogens such as Xanthomonas campestris. Infection of Arabidopsis mutants with fewer hydathodes resulted in spontaneous mesophyll water-soaking and revealed unexpectedly-large pathogen populations explained by direct infection of the mesophyll niche through stomata and subsequent proliferation. Physical blockage of hydathodes also induced mesophyll water-soaking and enhanced bacterial growth in both Arabidopsis and cauliflower leaves. These findings reveal a dual role for hydathodes as primary sites of infection while being essential to restrict pathogen proliferation in nonvascular tissues. More broadly, our study identifies leaf hydraulics and guttation as central components of water immunity in vascular plants.

Lisowski et al. [1] recently proposed that the pigeon's magnetic compass resides in the liver and depends upon ferritin-containing macrophages operating through a superparamagnetic mechanism. Here we evaluate this hypothesis using behavioral, neurobiological, evolutionary, and biophysical evidence together with three original analyses. First, we present a quantitative reanalysis of the classic pigeon homing experiments of Walcott and Green [2], calculating for the first time the magnetic fringe-field distribution produced by the head-mounted coil apparatus. These calculations demonstrate that the liver would have experienced a maximum shift of about 1.4% of the ambient geomagnetic field, far too small to account for the observed orientation effects, thereby localizing the relevant magnetoreceptive structures to the region of the head and excluding the liver. Second, we quantify the magnetic alignment expected for ferritin cores under geomagnetic conditions and show that ferrihydrite nanoparticles are approximately five orders of magnitude too weak to resist thermal noise, while their superparamagnetic moments fluctuate on timescales roughly eleven orders of magnitude faster than those relevant to navigation. Third, we analyze the neurobiological requirements of magnetic navigation and show that mobile macrophages are unlikely to provide stable directional information because compass signals must be continuously cross-calibrated against vestibular and visual inputs within a body-centered reference frame. Comparative evidence from fish, birds, and turtles further associates magnetically responsive structures with craniofacial tissues and trigeminal sensory pathways. We suggest the disruption of the magnetic response is due to induced anemia from clodronate treatment that triggers dissolution and absorption of iron in the magnetite-containing cells involved in magnetoreception. Together these findings indicate that a liver-based ferritin compass is inconsistent with behavioral, neurobiological, and physical evidence and instead support cephalized magnetite-based receptor systems as the basis of avian magnetoreception.

Host-generated flow is expected to oppose bacterial migration by sweeping cells downstream, yet some bacteria can migrate upstream against the direction of flow. However, it is unclear how the magnitude of shear force influences upstream bacterial migration. Here, we use microfluidics to examine upstream migration of the human pathogen Pseudomonas aeruginosa under host-relevant shear forces (0.1-10 pN). By independently varying flow rate and solution viscosity, we discover that increasing shear force counterintuitively enhances population-level upstream migration. Single-cell tracking reveals that force-enhanced migration is driven by an increase in twitching motility speed. Using a microfluidic-based trigonometry approach, we link the increase in speed to the cell-surface angle. At low shear forces, type IV pilus retraction generates a torque that tilts cells toward a vertical orientation, geometrically constraining forward movement and decreasing twitching speed. In contrast, higher shear forces push cells toward a horizontal orientation, increasing twitching speed. Collectively, our results reveal how shear force can enhance bacterial motility and upstream migration, providing a framework to understand how host shear forces may promote the spread of bacterial infections.

Heterotrophic bacteria utilize organic carbon as a source of energy and material for growth. The balanced allocation of carbon between these two processes, termed growth efficiency, is a key physiological property of microbes because it links energy and biomass production, and determines the fraction of carbon lost as CO2. However, we do not understand what controls growth efficiency in microbes or how it correlates with taxonomy and resource identity. Here, we develop a quantitative high-throughput method for measuring CO2 production during bacterial growth. For 23 bacterial strains spanning three phyla, grown on two carbon substrates, we quantify growth efficiency by measuring dynamic CO2 production and carbon accumulation in biomass. Intra-phylum comparisons show that glycolytic substrates yield higher efficiency growth (less CO2 produced per biomass carbon) than gluconeogenic substrates. However, growth efficiency varies as much across resources as it does across phyla. A physiological model shows that variation in the growth efficiency depends on the ATP produced per respired CO2 and the ATP needed per biomass on a given substrate, suggesting phylum level differences in energy supply and demand dictate differences in growth efficiency. This theory predicts no global correlation between growth rate and growth efficiency, a finding our data support. Finally, we report phylum-level variation in the dynamics of CO2 production, which we link to the presence of overflow metabolism on glycolytic sub-strates. This study revises our understanding of how taxonomy and substrate identity impact carbon allocation during growth and sets the stage for understanding CO2 production in communities. SignificanceBacteria are responsible for roughly half of all biological CO2 production on the planet, much of it in soils. How organic carbon flows through microbial metabolism sets the rate at which soils return CO2 to the atmosphere. Bacterial growth efficiency, defined by the fraction of consumed carbon that is retained as biomass rather than released as CO2, determines the fluxes of carbon into the atmosphere. At present, we do not understand what controls growth efficiency in bacteria. We develop a new quantitative measurement of growth efficiency which we apply to 23 soil bacteria from three phyla on two substrates. We show that efficiency is structured hierarchically, at the highest level substrate identity defines efficiency which then varies substantially across taxa. So both the identity of species involved and the compounds they consume determine the rate of carbon loss.

Molecular innovation and coevolution play major roles in the evolution of biological diversity, yet we have relatively few instances in which we understand the biological and molecular bases of coevolution. Rattlesnakes and other vipers have evolved resistance to their own novel venom metalloproteinase (MP) toxins through the invention and coevolution of serum-borne inhibitors (FETUAs) derived from the ancestral, non-inhibitory, serum glycoprotein Fetuin-A. However, how FETUAs acquired their function and their mechanism of action are unknown. Here, we use a combination of structure prediction, site-directed mutagenesis, functional analyses, and molecular dynamics to elucidate how the rattlesnake FETUA-3 protein evolved to inhibit three classes of venom MP toxins. We show that FETUA-3 uses a combination of structural motifs to bind multiple MPs with high-affinity and acts as a potent noncompetitive inhibitor by inserting its N-terminus into MP active sites and directly disrupting substrate cleavage. We also find strong selective constraints on key N-terminal residues that maintain FETUA-3 specificity for multiple MP targets. These results reveal that FETUA-3 evolved a unique mechanism of action distinct from other vertebrate metalloproteinase inhibitors and how a small number of FETUA proteins controls the activities of a large family of venom toxins.

The ability of enzymes to sense signals and respond by changing their structure and activity underlies cellular processes from signaling to metabolic control. While there have been recent advances in the de novo design of enzymes and conformationally switching proteins, combining these to achieve allosteric regulation of enzymatic activity in a fully designed system remains an outstanding challenge. Here, we show that denoising diffusion models enable the design of compact de novo enzymes whose activity can be allosterically activated or repressed by designed protein effectors. Our results establish a general route to allosteric control over a wide range of catalytic and other protein functions.

Cellulose is a common component of bacterial biofilms where it interacts with other biopolymers to form a 3-dimensional matrix enclosing the bacteria. Synthesized and secreted by the synthase-dependent biosynthesis pathway common to many bacterial exopolysaccharides, its surface exposure depends on the presence of the periplasmic cellulase BcsZ. During export across the periplasm, E. coli and other Enterobacteriaceae modify cellulose with lipid-derived phosphoethanolamine (pEtN). How BcsZ hydrolyzes pEtN-cellulose in the periplasm is unknown and so is the native distribution pattern of pEtN on cellulose. Here, we used carbohydrate synthesis, X-ray crystallography, native mass spectrometry, and super-resolution MINFLUX nanoscopy to delineate the role of BcsZ during cellulose biosynthesis. Crystal structures of BcsZ bound to chemically synthesized pEtN cello-oligosaccharides identify how the enzyme recognizes pEtN-modified glucosyl units. Comparing mono and double substituted cellohexaoses, we identify varying binding poses that are determined by two pEtN coordination sites within the BcsZ catalytic pocket. Combined, our structural analyses reveal an ideal BcsZ cellohexaose ligand containing two pEtN modified units separated by an unmodified cellotriosyl unit. The enzyme binds and hydrolyzes this compound with substantially increased affinity and efficiency. Further, BcsZ digestion of native pEtN cellulose combined with native mass spectrometry analyses reveals the stochastic distribution of pEtN on biofilm cellulose. Additionally, MINFLUX co-localization of BcsZ with other components of the biosynthetic complex demonstrates the random distribution of BcsZ across the periplasm. Our data suggest BcsZ functions independently of the biosynthetic complex to clear mislocalized pEtN cellulose from the periplasm.

In phototrophic organisms, pigment-protein membrane complexes are densely packed to form photosynthetic units (PSUs) that capture solar energy and convert it into chemical energy. Although the structures of many individual photosynthetic complexes have been resolved, how they are arranged and interact with others within photosynthetic membranes to enable efficient excitation energy transfer (EET) remains poorly understood. Here, we report cryo-electron microscopy structures of PSU supercomplex assemblies from the phototrophic a-proteobacterium Rhodovulum viride, including an RC-LH1 core associated with one or two peripheral LH2 complexes and a curved LH2 tetramer. These membrane-derived assemblies define the relative positions and orientations of neighboring photosynthetic complexes and place their pigment arrays in proximity across antenna-antenna and antenna-core interfaces. Structure-based simulations identify potential EET pathways within the PSU assemblies and reveal rapid energy transfer across both LH2-LH2 and LH2-LH1 interfaces. Collectively, these findings provide insights into the assembly and structural modularity of bacterial PSUs and elucidate how the lateral organization of membrane protein complexes facilitates efficient energy transfer. This work extends structural studies of bacterial photosynthesis from individual complexes to their native higher-order assembly, providing a framework for understanding how photosynthetic supercomplex organization shapes energy migration and for guiding the design of artificial photosynthesis.

The earliest stages of protein evolution remain a mystery: the nature of the first protein forms, their roles in emergent biological systems, and the forces that shaped them are largely unknown. Here, we combine insights from metabolic modelling, the organization of protein structure space, and protein folding mechanisms to probe the emergence of two ubiquitous cofactor- binding folds: Rossmanns and P-loop NTPases. While both folds are essential for contemporary life, we show that Rossmanns catalyze reactions deeper within the metabolic core and are more central in structure space than P-loop NTPases. Folding mechanism analysis further reveals that, whereas P-loop NTPases may require non-local interactions to fold, Rossmann folding can be nucleated by a structural module at the heart of the fold that contains a nucleotide-binding motif. Because this motif also directly mediates biochemical activity, this result suggests how early proteins may have compactly satisfied both folding and biochemical activity. Our results imply that folding constraints favored early enzymatic forms with compact binding motifs and modest catalytic roles. We conclude that the early emergence of the Rossmann fold reflects the chemical and physical constraints of protein folding, explaining both its profound antiquity and sustained longevity.

Long-range correlated motions couple distant regions of a protein, providing a physical basis for allosteric communication, cooperative conformational change, and the balance between structural stability and sensitivity to perturbations. Yet membrane proteins operate within a strongly anisotropic lipid bilayer, and how this environment reshapes such system-spanning coordination remains unclear. Using an implicit-membrane anisotropic network model, we perform a proteome-wide analysis of more than 3,000 human transmembrane proteins. We find that long-range correlations remain scale free under membrane constraints but become strongly direction dependent. Across protein sizes and topologies, their correlation lengths continue to scale with the corresponding molecular dimensions, while increasing membrane anisotropy extends in-plane correlations and shortens those along the membrane normal. Because spontaneous correlations and perturbation responses arise from the same underlying mechanics, we further resolve residue-level responses into in-plane and normal components. The resulting directional mechanical susceptibility provides new predictions of mutation-sensitive sites in GPCRs beyond those captured by conventional scalar flexibility measures. Together, these results show how environmental symmetry breaking can organize protein mechanics across scales, linking collective dynamics to the functional sensitivity of individual residues and connecting a general physical mechanism to experimentally measurable protein function.

Dense, well-ordered material states of proteins form the amyloid fibers that are a hallmark of neurodegenerative disease in the brain. Beyond forming amyloid fibers, some of these proteins can also adopt other material states termed condensates which are initially liquid-like but evolve to a soft, glassy phase. Fiber growth requires a large supply of monomers and, thus, it is often speculated that fibers emerge from within a dense condensate as it ages and its microscopic dynamics slow into a glassy state. Here, we use the well-established model system Fused in Sarcoma (FUS) to directly observe, quantify and theoretically describe fiber growth and its interplay with condensates. We report the discovery that fibers grow overwhelmingly in the dilute phase surrounding the condensates while the condensates concurrently evolve to a glassy arrested solid. The resulting protein fibers and glassy condensates are both distinct solid-like phases that coexist but do not directly interconvert. Taken together, these findings reveal that there are two competitive aging pathways in FUS condensation that are linked through phase separation kinetics.