Publications
click on papers for summaries, notes, and downloads…
Protein function emerges from the global pattern of cooperative interactions between amino acids, but for both conceptual and technical reasons, this pattern has been very difficult to experimentally define. In this work, we use high-throughput technologies to carry out a study of many thousands of pairwise amino acid couplings in several homologs of a protein family – a “deep coupling scan”. The data show how to estimate native amino acid couplings from mutational data, and suggest that a few positions are engaged in a two-state internal conformational equilibrium that underlies function. These data serve as a benchmark to evaluate coevolution methods that have been proposed to explain protein structure and function. This analysis shows that with different methods DCA and SCA, it is indeed possible to significantly predict both local structural contacts and functional couplings in proteins. [PDF]
A general concept that emerges from many studies is that allosterc mechanisms are often deeply conserved in protein families, even despite great divergence in cellular functions. In this paper, we show that mammalian cryptochrome (CRY) proteins share an allosteric mechanism with the distantly related photolyase enzymes that mediate DNA repair after UV damage. The regulatory site has been modified in the CRY proteins to differentially recognize proteins that modulate the mammalian circadian clock. [PDF] [SI]
An important technical goal is efficient, high-throughput construction of synthetic genes. Here, we report a useful resource towards this goal – a validated set of 166 20-nucleotide orthogonal PCR primers. Used in pairwise combination, it is possible to create 13,695 gene-specific primer sets, each capable of uniquely amplifying the sequences corresponding to individual genes from commercial mixed pool oligonucleotide syntheses. [PDF] [SI]
Many studies suggest that proteins are generally tolerant to mutation, enabling significant neutral variation during evolution. However, in some cases (e.g. the small G-protein Ras), phylogenetically distant orthologous proteins display high sequence conservation even though their paralogs have diverged extensively. What is the origin of high selective pressures in orthologs such as Ras? Here, we show that Ras in fact displays considerable neutral variation when studied in isolation (binding a partner protein), but is under global purifying selection when placed in the context of its natural biochemical network. Specifically, in the context of its guanine nucelotide exchange factor (GEF) and its GTP-ase accelerating protein (GAP). Thus, regulatory constraints that shape the behavior of cellular information processing pathways can powerfully influence selection pressures on inividual components. As a corrolory, the removal of regulation uncovers neutral and activating mutations that can drive the evolution of new phenotypes; in Ras, these include well-known oncogenic variations. [PDF]
A physical understanding of protein function requires direct visualization and analysis of intramolecular mechanics – the motions of amino acids that define the “reaction coordinate” and the forces constraining those motions. The relevant motions are expected to be subtle (sub-angstrom fluctuations), and occur over a very broad range of time-scales (picosecond to seconds), findings that frame the technical challenge in “seeing” proteins work. In this paper, we report a new method for directly visualizing motions with atomic resolution and for now, over time scales in the nanosecond to microseconds range. This method, called EFX, involves exciting subtle motions throughout protein crystals by applying large external electric pulses, and then reading out the effect by time-resolved X-ray diffraction. The initial results indicate that EFX samples motions that include those that contribute to the functional reaction corrdinate. This work represents a major new direction in our group in understanding the physical basis for protein function. [PDF] [SI]
Signaling downstream of the T-cell receptor requires the evolution of exquisite specificity in the action of the tyrosine kinases Lck and ZAP-80. For example, disallowed reactions are shown in the diagram at right with red crosses. In this work, we use a novel deep sequencing-based bacterial surface display assay to learn the underlying principles. We discover an electrostatic filter that explains both the distinct specificity of Lck and ZAP-70 kinases, and the inability of ZAP-70 to autophosphorylate. Thus, it seems that in this case specificity is achieved through simple rules that enable many degenerate solutions, a suggestion that many local specificity rules in kinase pathways can be achieved relatively easily through evolution. More generally, the experimental platform developed here provides the basic tools for probing the evolution of molecular recognition in the kinase family as a whole. [PDF]
How do proteins adapt to new functions as conditions of selection change? Using new techniques for massive-scale mutagenesis, we describe a critical role for a special class of intermediate mutations – termed “conditionally neutral” (CN). CN mutations have the unique property of opening up new functions while retaining the existing function, and can therefore pre-exist in populations as standing genetic variation before the new functions are required. This can facilitate evolution by temporally decorrelating environmental fluctuations (that change selection pressures) from the generation of favorable mutations. Structurally, we find that all CN mutations are allosteric; thus, in addition to traditional roles in mediating signaling and regulation, allostery may also mediate evolvability in proteins. A natural implication is the perhaps surprising idea that the origin of allostery might lie not in function, but simply in the capacity of proteins to evolve. This work is one piece of a body or work in the lab that began with Lee et al. (2008), and works through Reynolds et al. (2011), McLaughlin et al. (2012). and Stiffler et al. (2015)…see below for these papers. [PDF] [SI]
Deep mutagenesis – the process of making and quantitatively assaying many thousands of mutations in a protein – is emerging as a useful general method in biochemistry. In this paper, we describe the development of deep mutagenesis protocols, using the TEM-1 beta-lactamase as a model. This enzyme confers resistance in E.coli to a subset of pennicilin-class antibiotics, and is a widely used model system for understanding catalysis and evolution. The principles of assay and library construction (as well as cautions) discussed here may be useful for devleoping similar protocols for other protein systems. This paper is a technical accompaniment to Stiffler et al. Cell (2015) below. [PDF]
Epistasis is defined as the non-independence of mutational effects, a matter of central importance to relating genotype to phenotype (and fitness). However, the concept of epistasis remains murky, with different definitions in different fields and lack of practical approaches for determining the extent and pattern of epistasis. Here, we show that different definitions of epistasis are versions of a single mathematical formalism – the weighted Walsh-Hadamard transform (VH). This transform expresses the phenotypic effects of specific variants as a hierarchical sum of epistatic interactions between the underlying mutations. The number of significant epistatic terms defines a measure of the internal complexity of the genotype to phenotype map – the degree to which phenotypes of all variants can be captured in a compressed form. We discuss the principles of practical stategies for learning epistasis. [PDF]
The statistical coupling analysis (or SCA) is an approach for analyzing the collective evolution of amino acids within proteins. This approach exposes a functional architecture within proteins in which the basic evolutionary units are sparse, physically contiguous, long-range networks of residues (termed sectors). Sectors represent various biochemical properties of proteins, and correspond to functional and/or phylogenetic divergences in a protein family. This paper represents a technical explanation of the SCA method, with annotated examples. [PDF]
The year 2015 saw the passing of one of the great scientists in biological research – Al Gilman. This essay is an attempt to explain his unique and profound impact as a scientist, administrator, and mentor from the perspective of two of his junior colleagues. [PDF]
Mutational robustness and evolvability (the capacity to generate beneficial heritable variation), and two important characteristics of proteins that contribute to fitness. Using the TEM-1 beta-lactamase as a model, we show that robustness and evolvability are not fixed quantities, but instead depend on the strength of selection, with both properties collapsing as selection pressures increase. This result due to a steep non-linear relationship between fitness and catalytic efficiency that shifts as selection pressure varies. Spatially, the pattern of constraints radiates nearly uniformly outward from the active site as selection strength increases. In contrast, the pattern of adaptive mutations comprises physically contiguous “wires” that link the active site to a few distantly positioned surface sites through the protein core. In summary, robustness and evolability depend on an excess of cellular enzymatic activity relative to the selection threshold. A logical implication is that high catalytic activity in enzymes could be driven not by direct selection, but by the need to be adaptive under fluctuating enviornments. This work is one part of a body of work that includes Lee et al. (2008), Reynolds et al. (2011), McLaughlin et al. (2012). and Raman et al. (2016). [PDF]
The cyanobacterial clock is a minimal model system for undestanding mechanism and evolution at the systems level. Clock period and temperature compensation are encoded by the dynamics of just three proteins (Kai ABC), can be reconstituted in vitro, and are reported to be directly connected to organismal fitness. Here, we apply dynamical systsems analysis to understand the core mechanism of circadian oscillations in the Kai system. We show that the clock mechanism is effectively a relaxation oscillator, with one state of KaiC encoding the critical non-linearity. Parameter sensitivity analyses provide hypotheses for the constraints in this model, a starting point for further experimental tests. [PDF]
Allosteric regulation can provide fine control over enzyme activities; thus it is attractive to selectively target allosteric sites for develpment of pharmaceuticals. Here, we use the SCA method to predict surface sites that connect to allosteric networks (sectors) with the cathepsin K protease, and carried out screens for small molecules that can bind to these sites. A number of predicted compounds in fact modified cathepsin K activity. We characterize one such compound in depth, showing that it indeed binds to a predicted allosteric site and inhibits the activity of cathepsin K for both synthetic and natural substrates. This proof-of-principle experiment builds on the concept introduced in Lee et al. (2008) and Reynolds et al. (2011) that sector connected surface sites act as hotspots for the introduction of new regulation. [PDF]
Statistical analysis of protein sequences indicates an architecture for natural proteins in which most amino acids act nearly independently, and a few are engaged in a dense pattern of coevolution and form physical networks with the tertiary structure (sectors). This architecture might be a key and distinguishing feature of evolved proteins—a design principle providing not only for foldability and high-performance function but also for robustness to perturbation and the capacity for rapid adaptation to new selection pressures. Here, we review a general computational protein design approach for systematically testing these hypotheses. We provide some key new data that illustrates how targeted mutation of sectors and their surroundings influnce protein folding and function. [PDF]
The sector model for proteins suggests that the main units of fitness are sparse, physically connected networks of amino acid residues that connect distant functional surface sites through the protein core. Here, we develop a bacterial two-hybrid assay for PDZ function coupled to deep sequencing to globally test this model through saturation mutagenesis. We show that the pattern of mutational effects is in fact well described by the evolutionary conservation and correlations between sequence positions. However, a more specific test of correlations is to examine second-order mutational effects – the non-independence of pairs of mutations. Indeed, we show that the context-dependence of mutational effects on the background of a variation at the primary specificity site on the PDZ ligand is exclusively localized within sectors. This work sets the stage for the findngs reported in Raman et al. (2016) and the spatial patterns should be compared to the findings in Stiffler et al. (2015). [PDF] [SI]
A potentially attractive model system to study design constraints on protein systems is the core oscillator of the cyanobacterial clock, encoded by three gene products – Kai A,B, and C. A free-running circadian clock with temperature compensation has been reconstituted in vitro from ATP + Kai A,B, and C, and good mathematical models have been proposed to explain the origin of the clock properties from the inter- and intra-molecular dynamics of these proteins. Traditional assays for clock period are based on following the phosphorylation status of KaiC. Here, we describe the developement of a novel FRET-based assay that follows binding between KaiB and the so-called S-state of KaiC, a key feedback reaction in the clock mechanisms. This confirms predictions from theory that there is a ~ 8h phase lag between the two, and provides a new tool to observe the dynamics of KaiC. [PDF]
Sectors are coevolving groups of amino acids within the tertiary structure of proteins that typically form physically-connected networks that link a few distantly positioned surface sites to the main functional site. In some cases, this architecture seems to explain known allosteric mechanisms, but interestingly, it is also found in proteins with no known allostery. In these cases, the prediction is that sector-connected surface sites should represent cryptic allosteric sites – capable of accepting regulatory input by simply linking an input signaling module. In this work, we carry out a systematic “domain-insertion scan” as a deep test of this hypothesis. Specifically, we insert a plant light-sensing domain (as-LOV2) into every surface-exposed site in E. coli dihyrdrofolate reductase (72 total), and examined each site-specific fusion for light-dependent DHFR activity in vivo. The bottom line is that 14 out of 72 sites show statistically significant light-dependence, and every one of these 14 sites occurs at the surface edge of the protein sector. Thus, we propose that sector connected surface sites are “hot spots” for the emergence of new regulation in proteins, with sectors themselves serving as the communication mechanism. We synthesize these results with prior work to produce a model for the de novo evolution of communication between proteins. [PDF] [SI]A short preview of a paper reporting a large-scale proteomic screen for proteins that interact with the core cellular machinery mediating autophagy. The main interesting aspect of such unbiased screens is the possibility of using the data to adjust or even re-draw our picture of the network of components and reactions that make up a complex cellular process. [PDF]
Allosteric communication between protein domains represents an elementary step in building up complex, regulated cellular processes. A classic model is the Hsp70 family of molecular chaperones, in which ATP binding at an N-terminal nuceoltide-binding domain (NBD) switches ligand binding within a C-termal substrate-binding domain (SBD). Here, we apply the SCA method for mapping amino acid interactions through statistical analysis of homologous sequences to discover the allosteric mechanism linking the ATP- and substrate-binding sites in Hsp70. We make use of an interesting property of the Hsp70 family – namely the existence of an annotated subfamily that lacks allostery – to mathematically identify the network of coevolving amino acids that underlies this sequence divergence. That network represents a physically contiguous pathway of residues that leads from the ATP binding site in the NBD to the interdomain interface to substrate-binding site on the SBD, and includes amino acids that are essential for the allosteric coupling. This work shows how known functional divergences within a protein family can be exploited to discover the intramolecular mechanism(s) underlying those divergences. [PDF] [SI]
Proteins are traditionally described by their primary, secondary, and tertiary structural properties, a decomposition into units that forms our language for discussing protein function. Thus, we speak of individual amino acid residues, alpha helices, beta-strand, and loops, or tertiary structure motifs comprised of these elements. In. this work, we show that the natural decomposition of protein structure by the pattern of evolutionary correlations between amino acid residues does not follow any of these structure-based classifications. Instead, the evolutionary units are one or more sparse networks of amino acids (termed “sectors”) that arise from various regions of primary structure and various portions of secondary structure elements, and sharing only the structural principle that they are physically contiguous in the tertiary (or quaternary) structure. We show that individual proteins can have multiple, independent sectors, with each sector corresponding to a distinct biochemical activity and corresponding to a distinct mode of functional divergence(s) in a protein family. We propose that sectors represent a new decomposition of protein structure that corresponds to function. [PDF] [SI]
The “wire”-like topology of sectors within protein molecules – linking active sites to distantly positioned surface sites – leads to a simple idea for the engineering of new allosteric communication between proteins. Specifically, can communication between the active sites of two otherwise unrelated proteins simply amount to physically approximating the distant sector-connected surface sites? Here, we show that indeed, naive connection of a plant light-sensing domain (LOV2) and a bacterial metabolic enzyme (DHFR) through their sector surfaces is sufficient to create weak but significant light-dependent enzymatic activity. In contrast, connection at a non-sector-connected surface site is not. This work initiated a line of work that is followed and recapitulated in Reynolds et al. (2011), Stiffler et al. (2015), and Raman et al. (2016). [PDF] [SI]
The compound eye of the fruit fly Drosophila melanogaster represents an ideal model system to pursue a complete systems-level understanding of a cellular signaling pathway. In photoreceptor cells of the retina, a G-protein mediated signaling system transduces the absorption of photons into brief depolarizations of the cell membrane that comprise the input to downstream processing centers in the brain. The elementary signaling event is the “quantum bump” (QB), a transient (~30ms) all-or-nothing opening of about 25 ion channels in response to a single photon. Here, we show how the size, shape, and probability to occur of the QB arises from the stochastic nonlinear dynamics encoded by the underlying chemical reactions. The basic understanding is that the visual signaling machinery operates as a light-driven stochastic relaxation oscillator, converting photons into randomly distributed, cooperative bursts of channel openings. [PDF] [SI]In biological systems ranging from proteins to neural tissues, global behaviors emerge from nonlinear interactions between many constituent parts (amino acids or neurons, respectively). Empirical results suggest the sufficiency of pairwise correlations in capturing the essential global behaviors. Here, we show that the Monte Carlo sampling procedure used to demonstrate this result in proteins and the maximum entropy model used to demonstrate the result in neural systems are, in certain limits, mathematically identical. These procedural similarities likely speak to a deeper similarity in the nature of the general class of problems that include multicomponent interacting systems in biology. [PDF]
A short review of a few papers reporting the first high-resolution structures and detailed functional characterization of the beta-2 adrenergic receptor, in complex with a ligand and in the context of an engineered domain in an intracellular loop. Understanding the mechanism of allosteric regulation of activity in G protein-coupled receptors is of major importance in biology, and these studies represent a significant step towards that goal. [PDF]
Scaffolding proteins are thought to be passive elements of cellular signaling systems, serving to pre-organize the active signaling machinery to achieve speed, efficiency, and specificity. Here, we show that the InaD scaffold that operates in photoreceptor cells of Drosophila compound eye, is actually a dynamic element, undergoing conformational changes on the time scale of signaling (~10’s of milliseconds) to shape the visual response. The conformational change is driven by a light-dependent transient oxidation of a disulfide bond within one of the PDZ domains (PDZ5) that makes up InaD. This oxidation perturbs the conformation of the PDZ5 binding pocket, occurs in vivo, contributes to fast visual kinetics, and is required for an evolutionarily conserved behavior in daytime-active flies called the “escape jump response”. This work connects a single atomic level feature (the PDZ5 disulfide bond) to organismal behavior, and introduces the notion of dynamic scaffolding as a potential mechanism for rapid evolution of signaling behaviors. [PDF] [SI] [Hardie NV] [Montell NV] [Zhang NV]An interesting empirical observation is the capacity of some lineages to undergo very rapid morphological variation in response to selective pressure. An example is in dogs, where dramatic changes in morphology have occurred compared to related species of Carnivora. What might be the mechanistic origin of this rapid evolutionary variation? Here, based on genome sequencing in many species representing 10 carnivore clades, we find evidence for selectively elevated slippage rates in tandem repeat regions of various proteins in the dog and rodent clades that show fast morphological change. Since changes in repeat lengths has been shown to influence function in proteins controlling developmental morphology, we propose that variation in basal slippage rates across clades may underlie variations in rate of morphological change. [PDF]
Ferric citrate binding to an externally facing site in its outer membrane receptor (FecA) triggers an allosteric signal relayed through FecA to periplasmic regions that initiate events leading to regulated transcription of the FecA system. Here, we use the SCA method to discover the network of amino acids in FecA that is responsible for transmitting the allosteric signal. This network comprised a sparse but physically contiguous system of atoms that links the ferric citrate binding site to the periplasmic loops, and targeted mutagenesis of residues demonstrates the role of the deduced network in allosteric communication. Based on a comparative study of homologous outer membrane proteins in bacteria, we suggest that the allosteric mechanism is conserved in the family of TonB-dependent transporters. [PDF]A short review of “knowledge-based potentials” – statistical parameters derived from databases of known proteins to empirically capture useful aspects of the physical chemistry of protein folding and function. Such statistical potentials have played a major role in enhancing the performance of computational protein design by both improving the accuracy of physics-based models of the interatomic forces and by limiting the complexity of searching the sequence space. [PDF]
The cellular signaling machinery comprises a large network of molecular pathways that transform external inputs into biological responses – including changes in metabolism, transcription, and secretion. How complex is the signal machinery? Are responses to combinations of inputs a simple, weighted linear sum of the responses to individual inputs or are there nonlinear “crosstalk” interactions between pathways that fundamentally complicate the response to combinatorial inputs? Here, we analyze a very large-scale dataset in RAW 264.7 macrophages in which a broad range of quantitative cellular outputs were measured in response to single and pairwise combinations of 21 receptor-specific ligands. The data expose a large number of non-linear interactions between signaling cascades, only a few of which are mechanistically characterized. Nevertheless, the number of cross-talk interactions is such fewer than theoretically possible, possibly indicating strong physical or evolutionary limits on the computational complexity of cellular information processing. [PDF]
Protein sequences evolve through random mutagenesis with selection for optimal fitness. Cooperative folding into a stable tertiary structure is one aspect of fitness, but evolutionary selection ultimately operates on function, not on structure. In an accompanying paper (Socolich et al., 2005), we proposed a model for the evolutionary constraint on a small protein interaction module (the WW domain) through application of the SCA, a statistical analysis of multiple sequence alignments. Construction of artificial protein sequences directed only by the SCA showed that the information extracted by this analysis is sufficient to engineer the WW fold at atomic resolution. Here, we demonstrate that these artificial WW sequences function like their natural counterparts, showing class- specific recognition of proline-containing target peptides. Consistent with SCA predictions, a distributed network of residues mediates functional specificity in WW domains. The ability to recapitulate natural-like function in designed sequences shows that a relatively small quantity of sequence information is sufficient to specify the global energetics of amino acid interactions. [PDF] [Kelly NV] [Gierasch NV]
Classical studies show that for many proteins, the information required for specifying the tertiary structure and the thermodynamic stability of the folded state is contained within the amino acid sequence. Here, we attempt to deduce the essential sequence rules for specifying a protein fold by computationally creating artificial protein sequences using only statistical information encoded in a multiple sequence alignment and no direct tertiary structure information. Experimental testing of libraries of artificial WW domain sequences shows that a simple statistical energy function capturing coevolution between amino acid residues is necessary and sufficient to specify sequences that fold into native structures. The artificial proteins show thermodynamic stabilities similar to natural WW domains, and structure determination of one artificial protein shows excellent agreement with the WW fold at atomic resolution. The relative simplicity of the information used for creating sequences suggests a marked reduction to the potential complexity of the protein-folding problem. [PDF] [Kelly NV] [Gierasch NV]
The ligand-binding domains (LBDs) of the nuclear hormone receptors represent a remarkable model system for allosteric regulation. This proteins have the capacity to couple four distinct functional surfaces – the ligand-binding pocket, the heterodimerization interface, a co-factor binding surface that modulates transcriptional activity, and the so-called AF2 helix, which mediates ligand-dependent transactivation. Here, we use the statistical coupling analysis (SCA) to deduce the intramolecular network of amino acids that mediates the allosteric coupling of these surfaces, and carry out dense mutational analysis to test the predictions. A particularly interesting result is that allosteric regulation through the heterdimerization interface can be dependent on the identify of the ligand, suggesting differential engagement of the allosteric network by different ligands operating on the same LBD. Such differential engagement of allostery has profound functional consequences in terms of the transcriptional responses induced by ligand-binding. [PDF] [Kuriyan NV]
Atomic resolution structures of proteins indicate that the core is typically well packed, suggesting a densely connected network of interactions between amino acid residues. The combinatorial complexity of energetic interactions in such a network could be enormous, a problem that limits our ability to relate structure and function. Here, we use high-resolution structure determination and functional measurements to examine the complexity of amino acid interactions in a local region within the core of the green fluorescent protein (GFP). Mutations at three sites within the chromophore-binding pocket display an overlapping pattern of structural change and are thermodynamically coupled, seemingly consistent with the dense network model. However, crystallographic and energetic analyses of higher-order coupling between mutations show that pairs of mutations couple through independent mechanisms within the region of structural overlap. The data indicate that, even in highly stable proteins, the core contains sufficient plasticity in packing to uncouple high-order energetic interactions of residues, a property that may be general in proteins. [PDF]
G proteins comprise a superfamily of guanine nucleotide-dependent allosteric switches in which exchange of GTP for GDP regulates binding to specific “effector” proteins. Using the statistical coupling analysis (SCA) method, we identified a network of amino acid residues ( a “sector” in updated language) that links the nucleotide binding pocket with known effector binding surfaces – a hypothesis for the allosteric mechanism. We carried out a mutations scan to test the predictions, and show that perturbations at sector positions are selectively associated with changes in allosteric coupling between nucleotide exchange and effector binding affinity. Furthermore, the data are consistent with a two-state allosteric model in which mutation of sector residues control the equilibrium between GDP- and GTP-bound conformations, causing the ratio of affinities of these species the effector to vary in a reciprocal fashion. Consistent with its allosteric role, the physical architecture of the sector in the tertiary structure is state-dependent – compact in the GTP-bound state and more loosely organized in the GDP-bound state. We propose that the network identified by SCA is a conserved allosteric mechanism conferring nucleotide-dependent switching in the G proteins, with the specific features of different family members built on this functional core. [PDF]In addition to their well-known role in visual signal transduction, photoreceptor cells of the retina also carry out another light-mediated process – programmed cell death, or apoptosis. The initial steps of the mechanism of light-dependent photoreceptor apoptosis was worked out in Kiselev et al. (2001), but this small review describes work on the later events in the process. Basically, this work shows that ceramide, a sphingolipid involved in many aspects of cellualar signaling, plays an essential role. Since diverse cellular processes control ceramide production, this molecule could be an integrator of many external signals that collaborate to determine the probability and timing of apoptosis. [PDF]
Allosteric communication is the process by which signals originating at one site in a protein propagate to regulate distant functional sites. Here, we extend the statistical coupling analysis (SCA), a sequence-based method, to carry out a global analysis of amino acid interactions within a protein. Application of this method for three classic model systems for allostery (G protein–coupled receptors, the chymotrypsin class of serine proteases and hemoglobins) reveals a surprisingly simple architecture for amino acid interactions in each protein family: a small subset of residues forms physically connected networks that link known distant functional sites in the tertiary structure. Although small in number, residues comprising the network show excellent correlation with the large body of mechanistic data available for each family. The data suggest that evolutionarily conserved sparse networks of amino acid interactions represent structural motifs for allosteric communication in proteins. In later work, we refer to these coevolving networks as “protein sectors”. [PDF]In 2001, a large consortium of scientists from throughout the United States convened to work on a global understanding of cellular function. The motivation was the growing sense that the more traditional approach of studying components of cells one at a time was potentially never going to tell us how a cell (or any complex biological system) operates. This summary article describes the broad scientific questions and strategies taken on by this consortium, and outlines an administrative strategy to work collaboratively towards a collective understanding of cell behavior. [PDF]
Designing protein sequences that fold into specific native three-dimensional structures is a problem of great potential complexity. This short review describes a new approach to this problem – not based on direct modeling of the physical chemistry amino acid interactions, but based on modeling the statistics of evolution, as represented in extant protein sequences and structure databases. These ‘knowledge-based’ potentials combined with computational optimization algorithms provides a powerful set of tools for designing proteins. More deeply, the study of these empirical potentials might help direct a fundamental physical understanding of the energetic principles of protein structure and function. [PDF]
Green fluorescent protein (GFP) is a standard tool for investigating a variety of cellular activities, and has served as a model system for understanding spectral tuning in chromophoric proteins. Distant homologs of GFP in reef coral and anemone display two new properties of the flu- orescent protein family: dramatically red-shifted spectra, and oligomerization to form tetramers. In this work, we report the 1.9 Å crystal structure of DsRed, a red fluorescent protein from Discosoma coral. DsRed monomers show similar topology to GFP, but additional chemical modification to the chromophore extends the conjugated π-system and likely accounts for the red-shifted spectra. Oligomerization of DsRed occurs at two chemically distinct protein interfaces to assemble the tetramer. The DsRed structure reveals the chemical basis for the functional properties of red fluorescent proteins and provides the basis for rational engineering of this new family of GFP homologs. This work represents perhaps the only thematically unlinked “random” project in the history of our research program. [PDF]
Light-induced photoreceptor apoptosis occurs in many forms of inherited retinal degeneration resulting in blindness in both vertebrates and invertebrates. Here, we uncover a pathway by which activation of rhodopsin in Drosophila mediates apoptosis through a G protein–independent signaling mechanism. This process involves the formation of membrane complexes of phosphorylated, activated rhodopsin and its inhibitory protein arrestin, and subsequent clathrin-dependent endocytosis of these complexes into a cytoplasmic compartment. Remarkably, apoptosis occurs reliably after a long (days) wait time after initial light exposure. This work provides a fascinating basis to understand the design of long-delay reliable switches in biology. This work represented one of three research directions we began in 1997 when we started our laboratory. Due to focus on the protein evolution problem and on the design of signaling cascades, we dropped this line of work. [PDF]
For mapping energetic interactions in proteins, we developed a technique that uses evolutionary sequence data for a protein family to measure statistical interactions between amino acid positions. For the PDZ domain family, this analysis predicted a set of energetically coupled positions for a binding site residue that includes unexpected long-range interactions. Mutational studies confirm these predictions, demonstrating that the statistical energy function is a good indicator of thermodynamic coupling in proteins. Sets of interacting residues form connected pathways through the protein fold that may be the basis for efficient energy conduction within proteins. This work represents the starting point in our laboratory of the statistical coevolution strategy for analyzing amino acid interactions in proteins. [PDF]A short review of molecular mechanisms of specificity in cellular signaling systems.
InaD is a multi-PDZ domain scaffolding protein that assembles a specific membrane-associated macromolecular complex in Drosophila photoreceptor neurons, and is essential for vision. The power of efficient genetic manipulation in flies, together with quantitative electrophysiological and imaging methods makes InaD a beautiful model system to study the general principles of organized signaling in cells. This short review summarizes work to date on this system, and advances the key next questions in this line of work.
Older papers….
Center for Physics of Evolution
Biochemistry & Molecular Biology The Institute for Molecular Engineering The University of Chicago 929 E. 57th Street Chicago, IL 60637
