For any protein to function properly, the polypeptide chain produced by the ribosome, the protein-synthesizing factory of the cell, has to fold into the correct three-dimensional structure. Protein folding in the cell starts while the growing peptide chain is still bound to the ribosome. The information on the sequence of amino acids of the protein is stored in the sequence of codons in the messenger RNA (mRNA). The ribosome reads the mRNA codons one by one and translates them into the sequence of amino acids of the protein. Not all codons are read with the same speed: periods of rapid translation are separated by pauses. The role of such pauses is not known, but interfering with a fine-tuned rhythm of translation causes incorrect folding of proteins which can lead to disease. Translation rates may be changed by synonymous mutations in the mRNA, i.e. mutations that do not change the meaning of a codon in terms of which amino acid to be incorporated, but affect the speed by which the codon is translated. The goal of this project is to understand how nascent proteins fold while being synthesized on the ribosome and how the speed of translation affects their folding. Fluorescent reporters introduced at defined positions in the nascent protein will allow us to distinguish the unfolded, partially folded, and fully folded protein as it is produced by the ribosome. By determining, in real time, the hierarchy by which the structural elements in the protein are formed, we will delineate the protein´s co-translational folding pathway and compare it to the folding properties of the completed protein when it is released from the ribosome. By introducing synonymous mutations in the mRNA, we will examine how translational pausing affects folding. We will also determine the structure and dynamics of independent folding units in the growing peptide. The work requires an interdisciplinary approach that combines the expertise in molecular biology, biochemistry, biophysics, and structural biology. The results will provide important insight into the mechanism of protein folding in the cell. Furthermore, the work will contribute to better understanding the origin of human diseases caused by incorrect protein folding and give a tool to upscale the production of functional proteins for medical and biotechnological purposes.

Many clinically important antibiotics are peptide natural products of great structural complexity made by bacteria or fungi. The biosynthesis is achieved by nonribosomal peptide synthetases (NRPS), in which catalytic subunits and carrier domains form an assembly line-like mega-enzyme and interact in a coordinated way to achieve a directed product assembly via covalently bound intermediates. As such, a fundamental mechanistic understanding of these remarkable molecular machines will have direct implications in the rational design of novel synthetic pathways by biocombinatorial approaches. Large-amplitude conformational transitions have been identified and are likely to play a central role in the apparent directionality of the stepwise peptide assembly within the NRPS. Two conceptually distinct hypotheses have been proposed for triggering the conformational changes: They could be driven by chemical-reaction directed interactions between the individual domains with well-defined protein conformations or, alternatively, they could be promoted by Brownian random movements that eventually lead to productive interactions. Using these two still highly speculative models as guide, this proposal is intended to dissect the multi-step enzymatic reactions and clarify the mechanistic roles of large-amplitude conformational transitions. Specifically, we aim (1) to monitor the sub-domain dynamics in the adenylation domain, (2) to monitor the domain-domain interactions between the adenylation and the peptidyl carrier domain, and (3) to monitor the dynamical domain coordination of a minimal NRPS in real time. We will apply single-molecule FRET spectroscopy to this kind of multi-domain mega-synthetases for the first time. We will also develop new techniques for site-specific protein labeling and new data-driven statistical analysis methods to interpret the data. An international research consortium consisting of leading research groups in chemical biology (Germany), single-molecule spectroscopy (US), and mathematical physics (Japan) is formed to tackle this challenging yet fundamentally important problem. This will be a new research direction for all three participating labs and it will be impossible for any one of the groups to achieve the aims working alone.

Object recognition is a difficult computational task solved by the nervous system. For a wide range of organisms object recognition occurs predominantly using the chemical senses. Behaviorally significant odor sources must be faithfully recognized despite being composed of an often complex chemical signature and being embedded in a complex and dynamic chemical background. The present proposal aims to understand the neural mechanisms that make such computations possible.

Increasing evidence suggests that object recognition can be understood as a process of probabilistic (Bayesian) inference. Although this idea is already being exploited in the visual system, it has yet had little impact on the field of olfaction. Our central hypothesis is that odor stimuli are represented using a probabilistic population code, and that odor representations are processed in Bayesian optimal way by the nervous system. This problem requires at least two computational processes: integration, which underlies recognition of multi-component odor sources, and marginalization, which is essential for invariance to irrelevant background odors.

Using this probabilistic framework, we will develop computational models for integration and marginalization in the olfactory system and test them using parallel experiments in rats and in Drosophila larvae. The rat will provide a model system that can be probed using psychophysics and multi-electrode recordings, allowing us to study the neural basis of marginalization in populations of olfactory cortex neurons. The Drosophila larva will provide a model system with a nervous circuit amenable to genetic manipulations: using chemotaxis as a behavioral paradigm, we will using molecular perturbations to decompose olfactory input into elementary parts mediated by single odorant receptors.

The probabilistic framework will be used to design and interpret experiments which will themselves inform refinement of the theoretical hypotheses. By comparing across species, we will test for the existence of conserved, and possibly optimal, solutions to ubiquitous problem of the odor recognition in naturalistic olfactory environments. This project is based on computational and systems neuroscience and includes strong contributions of the fields of mathematics and statistics (Pouget), biophysics and fly molecular biology (Louis), psychology and physiology (Mainen).

Two key challenges in biology are to understand how biological forms are generated, and to elucidate the basis for their diversity. We propose to address these problems by studying the development and evolutionary diversification of leaf forms. To this end, we will integrate developmental genetics and computational modeling to compare, contrast, and put in an evolutionary context the mechanisms that shape the simple leaf of the model plant Arabidopsis thaliana and the compound subdivided leaf of its close relative Cardamine hirsuta. In addition to being taxonomically related, these plants show exceptional genetic tractability, thus providing excellent opportunities for comparative studies of leaf development and diversity.

We will characterize the spatiotemporal expression of key leaf-shape regulating genes and the activity of auxin during leaf development, and the influence of these genes on cell division, tissue growth and vein patterning. On this basis, we propose to devise models of A. thaliana and C. hirsuta leaf development. These models will integrate molecular-level signaling, the pattern of growth and division of individual cells, mechanical interactions between cells, and vein pattern formation. We will iteratively test, improve and validate the models by comparing them with experimental data for A. thaliana, C. hirsuta, and their mutants and transgenic variants. To conceptualize the diversity and evolution of leaf forms we will construct a morphospace with coordinates reflecting the activity of selected genes.

We expect to gain an understanding of the development, diversification, and evolution of leaf shapes in mechanistic terms including genetic regulation, developmental signaling, cell division and differentiation, and tissue growth. We will also advance the state of the art in plant modeling by introducing the propagating-boundary method to simulate development of organs, devising a representation of cellular structures developing in 3D, and modifying finite-element methods to efficiently capture the dynamics of growing 2D and 3D tissues and structures. All aspects of the proposed research have a collaborative, interdisciplinary character, bridging developmental biology and computational modeling.

Adaptation from standing genetic variation is central to the evolution of phenotypes, yet its dependence on the molecular details that connect genotype, phenotype, and fitness is a major unsolved problem in genetics. The effect of recombination, in particular, is expected to depend on the extent of genetic interaction among alleles. While higher rate of recombination facilitates combinatorial exploration of the genotypic space, in the presence of interaction it also has a detrimental effect of breaking up synergistic combinations of alleles. We propose a multidisciplinary investigation of the roles of recombination and genetic interactions. We will study laboratory adaptation of C. elegans populations that start with high genetic variability and are engineered to have different rates of inbreeding and outcrossing. The project integrates experimentally determined genome-wide genotype distributions, multidimensional phenotype (gene expression) measurements, and fitness data across 100 generations of culture in laboratory populations that differ in their rates of outcrossing. This integration forms the empirical test bed for theoretical models of the role of genomic architecture in adaptation. In particular, we will test the predicted positive correlation among non-additive genetic effects and levels of linkage disequilibrium across the genome. The proposed work combines population genetics and high-throughput molecular biology approaches with statistical physics modeling methods, and aims to bridge population genetics with systems biology.