Explore the vast range of titles available.

Prion-like low-complexity domains (PLCDs) have distinctive sequence grammars that determine their driving forces for phase separation. Here we uncover the physicochemical underpinnings of how evolutionarily conserved compositional biases influence the phase behaviour of PLCDs. We interpret our results in the context of the stickers-and-spacers model for the phase separation of associative polymers. We find that tyrosine is a stronger sticker than phenylalanine, whereas arginine is a context-dependent auxiliary sticker. In contrast, lysine weakens sticker–sticker interactions. Increasing the net charge per residue destabilizes phase separation while also weakening the strong coupling between single-chain contraction in dilute phases and multichain interactions that give rise to phase separation. Finally, glycine and serine residues act as non-equivalent spacers, and thus make the glycine versus serine contents an important determinant of the driving forces for phase separation. The totality of our results leads to a set of rules that enable comparative estimates of composition-specific driving forces for PLCD phase separation. The complex link between protein sequence and phase behaviour for a family of prion-like low-complexity domains (PLCDs) has now been revealed. The results have uncovered a set of rules—which are interpreted using a stickers-and-spacers model—that govern the sequence-encoded phase behaviour of such PLCDs and enable physicochemical rationalizations that are connected to the underlying sequence composition.

Prion-like domains (PLDs) can drive liquid-liquid phase separation (LLPS) in cells. Using an integrative biophysical approach that includes nuclear magnetic resonance spectroscopy, small-angle x-ray scattering, and multiscale simulations, we have uncovered sequence features that determine the overall phase behavior of PLDs. We show that the numbers (valence) of aromatic residues in PLDs determine the extent of temperature-dependent compaction of individual molecules in dilute solutions. The valence of aromatic residues also determines full binodals that quantify concentrations of PLDs within coexisting dilute and dense phases as a function of temperature. We also show that uniform patterning of aromatic residues is a sequence feature that promotes LLPS while inhibiting aggregation. Our findings lead to the development of a numerical stickers-and-spacers model that enables predictions of full binodals of PLDs from their sequences.

Proteins are marginally stable molecules that fluctuate between folded and unfolded states. Here, we provide a high-resolution description of unfolded states under refolding conditions for the N-terminal domain of the L9 protein (NTL9). We use a combination of time-resolved Förster resonance energy transfer (FRET) based on multiple pairs of minimally perturbing labels, time-resolved small-angle X-ray scattering (SAXS), all-atom simulations, and polymer theory. Upon dilution from high denaturant, the unfolded state undergoes rapid contraction. Although this contraction occurs before the folding transition, the unfolded state remains considerably more expanded than the folded state and accommodates a range of local and nonlocal contacts, including secondary structures and native and nonnative interactions. Paradoxically, despite discernible sequence-specific conformational preferences, the ensemble-averaged properties of unfolded states are consistent with those of canonical random coils, namely polymers in indifferent (theta) solvents. These findings are concordant with theoretical predictions based on coarse-grained models and inferences drawn from single-molecule experiments regarding the sequence-specific scaling behavior of unfolded proteins under folding conditions.

Characterizing the structure and dynamics of protein unfolded states is essential for describing protein folding and aggregation mechanisms. The unfolded state and the initial stages of protein folding and aggregation under conditions which favor folding are most relevant but still poorly understood. The N-terminal domain of the ribosomal protein L9 (NTL9) has been used as a model system for much of this dissertation in order to investigate the collapse transition of protein folding and properties of the unfolded state under refolding conditions. The unnatural amino acid p-cyanophenylalanine (FCN) was used as a FRET partner with Trp, and was incorporated into NTL9 recombinantly. Seven donor/acceptor pairs were prepared to probe different regions within NTL9. Time-resolved FRET experiments interfaced with microfluidic mixing were performed and provided direct evidence for unfolded state compaction upon dilution from 10 M to 1 M urea prior to the folding transition. Continuous-flow small-angle X-ray scattering (SAXS), performed under the same conditions, indicates partial collapse of the unfolded state in 1 M urea. Molecular dynamics (MD) simulations of ensembles with observables that match those determined experimentally show that although the unfolded state under refolding conditions has global dimensions resembling a polymer chain in a theta solvent, it contains extensive native and non-native long-range contacts and residual secondary structure. The C-terminal domain of L9 (CTL9) was used as a second model system to investigate the protein folding collapse transition and the pH-dependent properties of the unfolded state. A simple biorthogonal labeling strategy was developed and optimized that allows attachment of donor and acceptor fluorophores with site-specificity and in very high yield. Time-resolved FRET probed the unfolded state as a function of pH and revealed that the unfolded state is expanded at pH 2 but not as expanded as an excluded volume chain. The unfolded state becomes more compact as the pH is increased and the net charge on the chain is reduced. An analysis of the Rg and sequence properties of CTL9 and other unfolded proteins at low pH documented in the literature indicates that the knowledge of the net charge per residue and mean hydrophobicity is not sufficient to predict the relative compactness of the acid unfolded state. Time-resolved FRET interfaced with microfluidic mixing showed that the unfolded state at neutral pH is more compact than the unfolded state at low pH and revealed that the collapse transition occurs at least 2,000 times faster than folding. FCN and Trp are quenched by selenomethionine (MSe) via electron transfer and provide a promising tool for protein structural studies. The MSe-FCN pair and the MSe-Trp pairs were introduced at residues i and i+4 of a designed 21-residue α-helix and at the same relative position within an α-helix of the 36-residue villin headpiece subdomain (HP36). Steady-state and time-resolved fluorescence studies show that these pairs provide a sensitive probe of helical structure. The MSe-Trp pair was also introduced recombinantly into CTL9 to demonstrate that the it can be used to monitor β-sheet formation. Unlike for the commonly employed Trp-His pair, quenching of F CN and Trp by MSe is pH independent. FCN is a conservative substitution for Tyr and Phe while MSe is a conservative substitution for hydrophobic amino acids. Given that methods exist to incorporate both FCN and MSe recombinantly and by solid phase peptide synthesis into the same protein, these fluorophore-quencher pairs will be useful for studies of protein structure, dynamics, association and aggregation. Infrared spectroscopy is widely used in studies of protein structure and dynamics. A biorthogonal labeling approach was developed to incorporate a metal-carbonyl infrared probe into CTL9 in high yield. Experiments showed that this probe is non-perturbing when attached to a solvent exposed position of CTL9. Fourier transform infrared spectroscopy (FTIR) revealed that the probe is sensitive to the dielectric of the solvent. 2D IR was used to measure lifetimes of the symmetric and asymmetric stretches in H2O and D2O and the difference in lifetime between the two solvents indicates that this method can be exploited to probe solvent accessibility, which will be useful in studies of protein-protein and protein-ligand interactions.

Evidence accumulated over the past decade provides support for liquid-liquid phase separation as the mechanism underlying the formation of biomolecular condensates, which include not only membraneless organelles such as nucleoli and RNA granules, but additional assemblies involved in transcription, translation and signaling. Understanding the molecular mechanisms of condensate function requires knowledge of the structures of their constituents. Current knowledge suggests that structures formed via multivalent domain-motif interactions remain largely unchanged within condensates. Two different viewpoints exist regarding structures of disordered low-complexity domains within condensates; one argues that low-complexity domains remain largely disordered in condensates and their multivalency is encoded in short motifs called stickers, while the other argues that the sequences form cross-beta structures resembling amyloid fibrils. We review these viewpoints and highlight outstanding questions that will inform structure-function relationships for biomolecular condensates.

Abstract Phase separation of intrinsically disordered prion-like low-complexity domains (PLCDs) derived from RNA-binding proteins enable the formation of biomolecular condensates in cells. PLCDs have distinct amino acid compositions, and here we decipher the physicochemical impact of conserved compositional biases on the driving forces for phase separation. We find that tyrosine residues make for stronger drivers of phase separation than phenylalanine. Depending on their sequence contexts, arginine residues enhance or weaken phase separation, whereas lysine residues weaken cohesive interactions within PLCDs. Increased net charge per residue (NCPR) weakens the driving forces for phase separation of PLCDs and this effect can be modeled quantitatively. The effects of NCPR also weaken known correlations between the dimensions of single chains in dilute solution and the driving forces for phase separation. We build on experimental data to develop a coarse-grained model for accurate simulations of phase separation that yield novel insights regarding PLCD phase behavior. Competing Interest Statement R.V.P is a member of the scientific advisory board of Dewpoint Therapeutics Inc and T.M. is a consultant of Faze Medicines, Inc. The work reported here has not been influenced by either of these affiliations.

This work introduces Gemma, a family of lightweight, state-of-the art open models built from the research and technology used to create Gemini models. Gemma models demonstrate strong performance across academic benchmarks for language understanding, reasoning, and safety. We release two sizes of models (2 billion and 7 billion parameters), and provide both pretrained and fine-tuned checkpoints. Gemma outperforms similarly sized open models on 11 out of 18 text-based tasks, and we present comprehensive evaluations of safety and responsibility aspects of the models, alongside a detailed description of model development. We believe the responsible release of LLMs is critical for improving the safety of frontier models, and for enabling the next wave of LLM innovations.