KURIYAN LAB

UC BERKELEY

HOWARD HUGHES MEDICAL INSTITUTE

Structural Biology of Cell Signaling and DNA Replication

Our laboratory is interested in the structure and mechanism of the enzymes and molecular switches that carry out cellular signal transduction and DNA replication. We use x-ray crystallography to determine the three-dimensional structures of proteins involved in signaling and replication, as well as biochemical, biophysical, and computational analyses to figure out how they work. A major focus in the laboratory is understanding the allosteric mechanisms that enable proteins to be exquisitely sensitive to input signals. 

PUBLICATIONS FROM THE LABORATORY

Current Research

Cell Signaling

  

DNA Replication

Cell Signaling: The major class of signaling molecules that we study are the protein kinases, a large family of closely related enzymes that catalyze the addition of phosphate to serine, threonine, and tyrosine residues in proteins. We also study the mechanism by which the guanine nucleotide-binding protein Ras, a crucial signaling switch, is activated by hormone and growth factor receptors. A new project (in collaboration with Michael Marletta, UC Berkeley) involves the analysis of the guanylyl cyclases, enzymes that generate cGMP, an important mediator of cell signaling. Because of the importance of these signaling molecules in cancer, much of our work has implications for the development of new drugs.

 

Processive DNA Replication: DNA polymerases that replicate chromosomes achieve high speed by utilizing specialized proteins that allow the polymerase to move rapidly along DNA without letting go. These proteins include the "sliding DNA clamp" (the beta clamp in Escherichia coli, PCNA [proliferating cell nuclear antigen] in eukaryotes) and the clamp loader complex (gamma complex in E. coli, RFC [replication factor C] complex in eukaryotes) that couples ATP binding and hydrolysis to the opening of the beta clamp and its loading onto DNA. Rapid movement of the replication fork involves the coordinated action of two polymerase-exonuclease complexes, working with sliding clamps and a clamp loader complex.

Cell Signaling

Regulation of the Src and Abl tyrosine kinases: autoinhibition of c-Src and c-Abl.

The closely related tyrosine kinases c-Src and c-Abl are regulated by their SH2 (Src-homology 2) and SH3 (Src-homology 3) domains. SH2 and SH3 domains are small modules that recognize specific peptide motifs. The Src kinases are prominent in the history of cell signaling research and cancer because the retroviral gene v-Src was the first oncogene to be discovered.

The cellular c-Src protein is controlled by an interaction between its SH2 domain and a phosphorylated tyrosine residue (Tyr 527) in the carboxyl-terminal tail of the protein (this tail is lacking in the aberrant v-Src protein). The inhibitory action of the SH2-tail interaction involves the internal engagement of another peptide-binding module present in these proteins, the SH3 domain (first shown by our laboratory and by Stephen Harrison [HHMI, Harvard University] and his colleagues).

The intramolecular docking of the SH2 domain in the Src kinases helps orient and stabilize a polyproline type II helical conformation within the Src kinase, to which the SH3 domain is bound. Computer simulations of protein dynamics and mutagenesis experiments suggest that the coordinated internal binding of the SH2 and SH3 domain results in the formation of a rigid clamp on the surface of the kinase, reducing its conformational flexibility and preventing activating transitions.

A significant puzzle regarding the autoinhibition of c-Abl concerned the absence of a phosphorylation site corresponding to Tyr 527 in c-Src. Like c-Src, the major isoform of c-Abl is myristoylated at the amino terminus. In c-Src, the myristoyl modification is important for attachment of the protein to the plasma membrane. Surprisingly, we discovered that the myristoyl group in c-Abl plays a role in its autoinhibitory mechanism.

In collaboration with Giulio Superti-Furga (Center for Molecular Medicine, Vienna), we showed that the N-terminal myristoyl group of c-Abl binds to a deep hydrophobic cavity at the base of the kinase domain, the existence of which had not been suspected previously because the cavity is created by a conformational change in the kinase upon myristate binding.

The structure of autoinhibited c-Abl (excluding the C-terminal half of the protein, which has no corresponding element in c-Src) resembles closely that of the Src kinases. Instead of relying on phosphotyrosine binding for internal docking, the SH2 domain in c-Abl binds directly to the C-terminal lobe of the kinase domain. This interaction is gated by the conformational change induced in the C-terminal a-helix of the kinase domain by the binding of the myristoyl group. Support for the importance of a rigid conformation of the SH3-SH2 has been provided by the demonstration that a critical N-terminal segment, not seen in our earlier structures, forms a belt that interacts with the SH3-SH2 linker, in a manner that stabilizes the SH3-SH2 unit.

Freedom and Phosphorylation -- An Interview with Src



The inhibition of the Abl tyrosine kinase by imatinib.

In 2000, we showed that the recognition of an inactive conformation of Abl, in which a catalytically important Asp-Phe-Gly motif at the heart of the kinase domain is flipped by ~180o with respect to the active conformation, underlies the specificity of the cancer drug imatinib (Gleevec).

In collaboration with Charles Sawyers (HHMI, University of California, Los Angeles), we are analyzing the structural basis for imatinib resistance. Surprisingly, only a minority of the imatinib-resistance mutations result in steric occlusion of the drug-binding site. A larger class of mutations appear to destabilize the specific conformation of Abl to which imatinib binds. A third class of mutations cannot be explained in this manner; most of these mutations occur in regions of the structure that have essentially identical conformation in the imatinib-bound and active conformations.

Imatinib binds to c-Src about 10,000-fold less tightly than to c-Abl. The inability of imatinib to inhibit c-Src is puzzling, since every residue that makes contact with the drug in c-Abl is either the same or replaced with a similar residue in c-Src, and it is not clear why the Src kinase domain cannot adopt the conformation required for imatinib binding. The problem of understanding the specificity of imatinib for Abl over Src is related closely to the problem of understanding the origin of patient resistance in BCR-Abl, and a thorough study of both phenomena is needed. The only way to obtain definitive insights into these conformational transitions is likely to be through the use of nuclear magnetic resonance, and this is a major goal of our future work (in collaboration with David Wemmer, UC Berkeley).


Activation of Ras by nucleotide exchange.

Ras is a small membrane-bound guanine nucleotide–binding protein that transmits signals when bound to GTP. GTP bound to Ras is hydrolyzed to form GDP, and the Ras·GDP complex is inactive as a signaling molecule. GDP is bound so tightly to Ras that the conversion of Ras·GDP to Ras·;GTP requires the action of proteins known as nucleotide exchange factors.

The activation of growth factor receptors results in the recruitment of the Ras-specific nucleotide exchange factor SOS (son-of-sevenless) to the membrane, where it encounters Ras and catalyzes the exchange of bound GDP for GTP. In collaboration with Dafna Bar-Sagi (New York University Medical School), we are analyzing the structural basis for the SOS mechanism. In 1998, we determined the structure of SOS bound to Ras, thereby explaining how SOS causes nucleotide release from Ras. Our recent serendipitous discovery that Ras, which is the substrate of SOS, is itself required for SOS activity is helping us understand a complex mechanism that regulates the activation of Ras by SOS. Ras·GTP is a potent mediator of cell behavior, morphology, and fate, and the Ras activation mechanism that we are uncovering appears to minimize the risk of accidental activation of Ras.

The Rem (Ras exchanger motif) and Cdc25 (named for the Ras activator protein in yeast) domains of SOS, located in the C-terminal half of SOS, are the minimal elements required for nucleotide exchange activity. We discovered that SOS contains a second distal binding site for Ras·GTP, between the Rem and Cdc25 domains, and that Ras binding to this site stimulates nucleotide exchange activity. More surprising was the realization that the binding of Ras·GDP to the distal allosteric site is required for basal SOS activity. Disruption of the distal site by mutation results in essentially complete shutdown of SOS activity because Ras can no longer bind allosterically.

SOS is normally autoinhibited in cells, and we showed that the Dbl homology–pleckstrin homology (DH-PH) unit of SOS, located in the N-terminal half of the protein, blocks the site of allosteric Ras binding, thereby inactivating the protein. Dafna Bar-Sagi has demonstrated that the N-terminal domain of SOS, which we showed contains two histone domains, is required for the activation of Ras. We have found that the histone domains dock onto the interfacial region between the DH-PH and Rem domains, and most likely help organize the structure at the membrane.

The major focus of our work in this area is to understand the biological consequences of the dependence of SOS on Ras, including the nature of the switch that releases the blockage of the distal binding site by the DH-PH unit. In collaboration with Jay Groves (UC Berkeley), we are setting up assay systems for SOS activity in which Ras is localized either to small unilamellar vesicles or to supported lipid bilayers (our previous work was carried out with soluble forms of Ras and SOS).

At the workplace -- An interview with Sos1



Autoregulation of Ca2+/calmodulin-dependent protein kinase II.

CaMKII is unique among protein kinases for its dodecameric assembly and its complex response to Ca2+. As shown by others, particularly Howard Shulman (formerly at Stanford University) and his co-workers, the enzyme has the ability to respond to the frequency of calcium pulses, an attribute that might underlie the importance of the protein in long-term potentiation in neurons. A two-step activation process makes it possible for CaMKII to retain a "memory" of prior activation by Ca2+/CaM. First, Ca2+/CaM removes an autoinhibitory segment located C-terminal to the kinase domain. The second step is the transphosphorylation of kinase domains, which prevents the rebinding of the regulatory segment and increases the affinity of CaM for the enzyme by ~13,000 fold.

We have determined the crystal structure of the association domain of CaMKII, the central hub of the assembly. We have also determined the crystal structure of the kinase domain of CaMKII, which reveals an unexpected dimeric organization in which the calmodulin-responsive regulatory segments form a coiled-coil strut that blocks peptide and ATP binding to the otherwise intrinsically active kinase domains. The structure of the kinase dimer, when combined with small-angle x-ray scattering data for the holoenzyme, suggests that inactive CaMKII forms tightly packed autoinhibited assemblies. Although the structures of the component pieces represent a big step toward the generation of models for the activation process, a crystal structure of the fully assembled dodecameric holoenzyme is our most important immediate goal for the future. (Our work on CaMKII is in collaboration with Angus Nairn, Yale University.)


Autoinhibition and activation mechanisms of EGFR

The EGFR family receptor tyrosine kinases (EGFR/Erb1, ErbB2/HER2, ErbB3/HER3 and ErbB4/HER4) play critical roles in regulating many important cellular processes such as cell proliferation, differentiation and migration. Abnormal activation of EGFR family members due to mutation or overexpression are associated with a variety of cancers, and to date more than ten drugs targeted towards these receptors are being tested or are in clinic use for cancer therapy. The EGFR family members share a common architecture: an extracellular ligand-binding region, a single transmembrane helix, an intracellular tyrosine kinase domain and a C-terminal tail. The well accepted mechanism for EGFR activation is that ligand-engagement induces dimerization of the receptor and activation of the intracellular kinase domain, which then phosphorylates several tyrosine residues in the C-terminal tail. These phospho-tyrosines in the C-terminal tail recruit downstream PTB- or SH2-containing signaling molecules which relay the signal to downstream signaling pathways.

Despite extensive studies, the mechanisms of how the EGFR kinase domain is kept inactive prior to ligand engagement and how it is activated upon ligand-induced receptor dimerization had eluded definition. We have recently shown that the EGFR kinase domain is monomeric and autoinhibited by adopting a Src/CDK-like inactive conformation. Induced dimerization by attaching the kinase domain to the surface of lipid vesicles increases kinase activity by ~15-fold, suggesting that some kind of inter-molecular interaction converts the kinase to its active state. Our analysis of the crystal structures of the EGFR kinase domain in the active conformation revealed that the kinase domain forms an asymmetric dimer, with the bottom of one kinase monomer docking onto the top of the other. This interaction mimics strikingly the binding of cyclin to cyclin-dependent kinase, suggesting that one kinase monomer in the asymmetric dimer activates the other by acting as a cyclin-like activator, although the kinase domain is structurally unrelated to cyclin. This model of activation is confirmed by our biochemical and cell biological experiments. A sequence analysis of the EGFR family members suggests that other members in the family use the same mechanism for activation through homo- or hetero-dimerization.

We are currently studying additional layers of regulation of EGFR, including both mechanisms that are built into the receptor molecule itself and mechanisms by which other proteins regulate the receptor activity. This work is in collaboration with Phil Cole (Johns Hopkins University) and Dane Wittrup (Massachusetts Institute of Technology). This work is supported in part by a grant from NCI.

Processive DNA Replication

We work closely with Michael O'Donnell (HHMI, Rockefeller University) to explore the molecular mechanism of high-speed DNA replication. We had shown early on that the sliding DNA clamps are ring-shaped proteins that encircle duplex DNA, providing a mobile platform to which the polymerases are tethered. We gained insight into clamp loading in E. coli by determining the structure of the clamp loader γ complex, and also that of one of the subunits bound to an open form of the β clamp. Clarification of the clamp-loading mechanism was obtained by determination of the structure of the eukaryotic clamp-loader complex (RFC) bound to the sliding clamp (PCNA) and to an ATP analog. This structure revealed that a key aspect of the mechanism is the ATP-dependent formation of a spiral structure by the clamp loader, which results in the recognition of DNA and the formation of hydrolysis-competent interfacial catalytic sites.

Our fluorescence experiments confirmed essential aspects of the model for DNA recognition, as did a subsequent electron microscopic reconstruction done by Kosuke Morikawa and co-workers (Biomolecular Engineering Research Institute, Osaka, Japan). We have used computer simulations to show that PCNA spontaneously adopts right-handed spiral structures when one of its interfaces is opened, consistent with the formation of a spiral, rather than flat, ATP- and DNA-dependent organization by the activated clamp loader. We are now working toward a crystal structure of a DNA complex of the clamp loader–clamp assembly. A new direction for the project is the structural analysis of alternative clamp loader–clamp complexes that are involved in DNA repair. (The work on DNA replication is supported in part by the National Institutes of Health.)

More recently we have determined the structure of the bacterial replicative DNA polymerases (DNA polymerase III), which shares no sequence similarity with other polymerases. The crystal structure reveals a unique chain fold with localized similarity in the catalytic domains to DNA polymerase beta and related nucleotidyltransferases. The structure of Pol III is strikingly different from those of members of the canonical DNA polymerase families, which include eukaryotic replicative polymerases, suggesting that the DNA replication machinery in bacteria arose independently. A structural element near the active site in Pol III that is not present in nucleotidyltransferases but which resembles an element at the active sites of some canonical DNA polymerases suggests that, at a more distant level, all DNA polymerases may share a common ancestor. The structure also suggests a model for interaction of Pol III with the sliding clamp and DNA.