Characterizing the equilibrium ensemble of folding pathways, including their relative probability, is one of the major challenges in protein folding theory today. Although this information is in principle accessible via all-atom molecular dynamics simulations, it is difficult to compute in practice because protein folding is a rare event and the affordable simulation length is typically not sufficient to observe an appreciable number of folding events, unless very simplified protein models are used. Here we present an approach that allows for the reconstruction of the full ensemble of folding pathways from simulations that are much shorter than the folding time. This approach can be applied to all-atom protein simulations in explicit solvent. It does not use a predefined reaction coordinate but is based on partitioning the state space into small conformational states and constructing a Markov model between them. A theory is presented that allows for the extraction of the full ensemble of transition pathways from the unfolded to the folded configurations. The approach is applied to the folding of a PinWW domain in explicit solvent where the folding time is two orders of magnitude larger than the length of individual simulations. The results are in good agreement with kinetic experimental data and give detailed insights about the nature of the folding process which is shown to be surprisingly complex and parallel. The analysis reveals the existence of misfolded trap states outside the network of efficient folding intermediates that significantly reduce the folding speed. D iscovering the mechanism by which proteins fold into their native 3D structure remains an intriguing problem (1, 2). Essential questions are: How does an ensemble of denatured molecules find the same native structure, starting from different conformations? Are there particular sequences in which the structural elements of a protein are formed (3-5)? Are there multiple parallel routes by which protein structure formation can proceed (6, 7)?Full answers to these questions require one to characterize the ensemble of folding pathways, including their relative probabilities. In principle, this detailed information is accessible via molecular dynamics (MD) simulations which, when used in concert with experimental evidence, are becoming an increasingly accepted tool to understanding structural details that are not easily accessible via the experimental observables (8). MD simulations with atomistic models of proteins have been used to study the dynamics of small proteins with folding times in the microsecond range (9-13). However, even though MD simulations make the full spatiotemporal detail accessible to observation, the characterization of the pathway ensemble is computationally difficult: A brute-force approach would start simulations from an equilibrium of unfolded structures, say A, and simulate until they relax into a set of folded state B. The analysis would then only be comprised of those trajectory segments that leave A and relax to B without...

Proteins of the Bcl-2 family either enhance or suppress programmed cell death and are centrally involved in cancer development and resistance to chemotherapy. BH3 (Bcl-2 homology 3)-only Bcl-2 proteins promote cell death by docking an α-helix into a hydrophobic groove on the surface of one or more of five pro-survival Bcl-2 receptor proteins. There is high structural homology within the pro-death and pro-survival families, yet a high degree of interaction specificity is nevertheless encoded, posing an interesting and important molecular recognition problem. Understanding protein features that dictate Bcl-2 interaction specificity is critical for designing peptide-based cancer therapeutics and diagnostics. In this study, we present peptide SPOT arrays and deep sequencing data from yeast display screening experiments that significantly expand the BH3 sequence space that has been experimentally tested for interaction with five human anti-apoptotic receptors. These data provide rich information about the determinants of Bcl-2 family specificity. To interpret and use the information, we constructed two simple data-based models that can predict affinity and specificity when evaluated on independent data sets within a limited sequence space. We also constructed a novel structure-based statistical potential, called STATIUM, which is remarkably good at predicting Bcl-2 affinity and specificity, especially considering it is not trained on experimental data. We compare the performance of our three models to each other and to alternative structure-based methods and discuss how such tools can guide prediction and design of new Bcl-2 family complexes.

Uncovering the relationships between peptide and protein sequences and binding properties is critical for successfully predicting, re-designing and inhibiting protein-protein interactions. Systematically collected data that link protein sequence to binding are valuable for elucidating determinants of protein interaction, but are rare in the literature because such data are experimentally difficult to generate. Here we describe SORTCERY, a high-throughput method that we have used to rank hundreds of yeast displayed peptides according to their affinities for a target interaction partner. The procedure involves fluorescence-activated cell sorting (FACS) of a library, deep sequencing of sorted pools, and downstream computational analysis. We have developed theoretical models and statistical tools that assist in planning these stages. We demonstrate SORTCERY’s utility by ranking 1026 BH3 peptides with respect to their affinities for the anti-apoptotic protein Bcl-xL. Our results are in striking agreement with measured affinities for 19 individual peptides with dissociation constants ranging from 0.1 to 60 nM. High-resolution ranking can be used to improve our understanding of sequence-function relationships, and to support the development of computational models for predicting and designing novel interactions.

Many applications in protein engineering require optimizing multiple protein properties simultaneously, such as binding one target but not others or binding a target while maintaining stability. Such multistate design problems require navigating a high-dimensional space to find proteins with desired characteristics. A model that relates protein sequence to functional attributes can guide design to solutions that would be hard to discover via screening. In this work, we measured thousands of protein–peptide binding affinities with the high-throughput interaction assay amped SORTCERY and used the data to parameterize a model of the alpha-helical peptide-binding landscape for three members of the Bcl-2 family of proteins: Bcl-xL, Mcl-1, and Bfl-1. We applied optimization protocols to explore extremes in this landscape to discover peptides with desired interaction profiles. Computational design generated 36 peptides, all of which bound with high affinity and specificity to just one of Bcl-xL, Mcl-1, or Bfl-1, as intended. We designed additional peptides that bound selectively to two out of three of these proteins. The designed peptides were dissimilar to known Bcl-2–binding peptides, and high-resolution crystal structures confirmed that they engaged their targets as expected. Excellent results on this challenging problem demonstrate the power of a landscape modeling approach, and the designed peptides have potential uses as diagnostic tools or cancer therapeutics.

Molecular diversity surveys have demonstrated that aquatic fungi are highly diverse, and that they play fundamental ecological roles in aquatic systems. Unfortunately, comparative studies of aquatic fungal communities are few and far between, due to the scarcity of adequate datasets. We combined all publicly available fungal 18S ribosomal RNA (rRNA) gene sequences with new sequence data from a marine fungi culture collection. We further enriched this dataset by adding validated contextual data. Specifically, we included data on the habitat type of the samples assigning fungal taxa to ten different habitat categories. This dataset has been created with the intention to serve as a valuable reference dataset for aquatic fungi including a phylogenetic reference tree. The combined data enabled us to infer fungal community patterns in aquatic systems. Pairwise habitat comparisons showed significant phylogenetic differences, indicating that habitat strongly affects fungal community structure. Fungal taxonomic composition differed considerably even on phylum and class level. Freshwater fungal assemblage was most different from all other habitat types and was dominated by basal fungal lineages. For most communities, phylogenetic signals indicated clustering of sequences suggesting that environmental factors were the main drivers of fungal community structure, rather than species competition. Thus, the diversification process of aquatic fungi must be highly clade specific in some cases.The combined data enabled us to infer fungal community patterns in aquatic systems. Pairwise habitat comparisons showed significant phylogenetic differences, indicating that habitat strongly affects fungal community structure. Fungal taxonomic composition differed considerably even on phylum and class level. Freshwater fungal assemblage was most different from all other habitat types and was dominated by basal fungal lineages. For most communities, phylogenetic signals indicated clustering of sequences suggesting that environmental factors were the main drivers of fungal community structure, rather than species competition. Thus, the diversification process of aquatic fungi must be highly clade specific in some cases.