Principles of protein-protein interactions.

This review examines protein complexes in the Brookhaven Protein Databank to gain a better understanding of the principles governing the interactions involved in protein-protein recognition. The factors that influence the formation of protein-protein complexes are explored in four different types of protein-protein complexes--homodimeric proteins, heterodimeric proteins, enzyme-inhibitor complexes, and antibody-protein complexes. The comparison between the complexes highlights differences that reflect their biological roles.

Protein-protein interactions in the rigor actomyosin complex.

Since it has not been possible to crystallize the actomyosin complex, the x-ray structures of the individual proteins together with data obtained by fiber diffraction and electron microscopy have been used to build detailed models of filamentous actin (f-actin) and the actomyosin rigor complex. In the f-actin model, a single monomer uses 10 surface loops and two alpha-helices to make sometimes complicated interactions with its four neighbors. In the myosin molecule, both the essential and regulatory light chains show considerable structural homology to calmodulin. General principles are evident in their mode of attachment to the target alpha-helix of the myosin heavy chain. The essential light chain also makes contacts with other parts of the heavy chain and with the regulatory light chain. The actomyosin rigor interface is extensive, involving interaction of a single myosin head with regions on two adjacent actin monomers. A number of hydrophobic residues on the apposing faces of actin and myosin contribute to the main binding site. This site is flanked on three sides by charged myosin surface loops that form predominantly ionic interactions with adjacent regions of actin. Hydrogen bonding is likely to play a significant role in actin-actin and actin-myosin interactions since many of the contacts involve loops. The model building approach used with actomyosin is applicable to other multicomponent assemblies of biological interest and is a powerful method for revealing molecular interactions and providing insights into the mode of action of the assemblies.

A mean field model of ligand-protein interactions: implications for the structural assessment of human immunodeficiency virus type 1 protease complexes and receptor-specific binding.

We propose a general mean field model of ligand-protein interactions to determine the thermodynamic equilibrium of a system at finite temperature. The method is employed in structural assessments of two human immuno-deficiency virus type 1 protease complexes where the gross effects of protein flexibility are incorporated by utilizing a data base of crystal structures. Analysis of the energy spectra for these complexes has revealed that structural and thermo-dynamic aspects of molecular recognition can be rationalized on the basis of the extent of frustration in the binding energy landscape. In particular, the relationship between receptor-specific binding of these ligands to human immunodeficiency virus type 1 protease and a minimal frustration principle is analyzed.

Functional antagonism between the retinoic acid receptor and the viral transactivator BZLF1 is mediated by protein-protein interactions.

The Epstein-Barr virus-encoded protein BZLF1 is a member of the basic leucine zipper (bZip) family of transcription factors. Like several other members of the bZip family, transcriptional activity of BZLF1 is modulated by retinoic acid receptors (RARs). We present evidence that the RAR alpha and BZLF1 can reciprocally repress each other's transcriptional activation by a newly discovered mechanism. Analysis of RAR alpha mutants in transfection studies reveals that the DNA binding domain is sufficient for inhibition of BZLF1 activity. Analysis of BZLF1 mutants indicates that both the coiled-coil dimerization domain and a region containing the transcriptional activation domain of BZLF1 are required for transrepression. Coimmunoprecipitation experiments demonstrate physical interactions between RAR alpha and BZLF1 in vivo. Furthermore, glutathione S-transferase-pulldown assays reveal that these protein-protein interactions are mediated by the coiled-coil dimerization domain of BZLF1 and the DNA binding domain of RAR alpha. While RAR alpha is unable to recognize BZLF1 binding sites, the RAR alpha can be tethered to the DNA by forming a heteromeric complex with BZLF1 bound to DNA. Tethering RARs via protein-protein interactions onto promoter DNA suggest a mechanism through which RARs might gain additional levels of transcriptional regulation.

In vitro trans-splicing in Saccharomyces cerevisiae.

The interactions established at the 5'-splice site during spliceosome assembly are likely to be important for both precise recognition of the upstream intron boundary and for positioning this site in the active center of the spliceosome. Definition of the RNA-RNA and the RNA-protein interactions at the 5' splice site would be facilitated by the use of a small substrate amenable to modification during chemical synthesis. We describe a trans-splicing reaction performed in Saccharomyces cerevisiae extracts in which the 5' splice site and the 3' splice site are on separate molecules. The RNA contributing the 5' splice site is only 20 nucleotides long and was synthesized chemically. The trans-splicing reaction is accurate and has the same sequence, ATP, and Mg2+ requirements as cis-splicing. We also report how deoxy substitutions around the 5'-splice site affect trans-splicing efficiency.

Two helices plus a linker: a small model substrate for eukaryotic RNase P.

Using precursor tRNA molecules to study RNA-protein interactions, we have identified an RNA motif recognized by eukaryotic RNase P (EC 3.1.26.5). Analysis of circularly permuted precursors indicates that interruptions in the sugar-phosphate backbone are not tolerated in the acceptor stem, in the T stem-loop, or between residues A-9 and G-10. Prokaryotic RNase P will function with a minihelix consisting of the acceptor stem connected directly to the T stem-loop. Eukaryotic RNase P cannot use such a minimal substrate unless a linker sequence is added in the gap where the D stem and anticodon stem-loop were deleted.

Probing of tertiary interactions in RNA: 2'-hydroxyl-base contacts between the RNase P RNA and pre-tRNA.

A general method has been developed to analyze all 2' hydroxyl groups involved in tertiary interactions in RNA in a single experiment. This method involves comparing the activity of populations of circularly permuted RNAs that contain or lack potential hydrogen-bond donors at each position. The 2' hydroxyls of the pre-tRNA substrate identified as potential hydrogen bond donors in intermolecular interactions with the ribozyme from eubacterial RNase P (P RNA) are located in the T stem and T loop, acceptor stem, and 3' CCA regions. To locate the hydrogen-bond acceptors for one of those 2' hydroxyls in the P RNA, a phylogenetically conserved adenosine was mutated to a guanosine. When this mutant P RNA was used, increased cleavage activity of a single circularly permuted substrate within the population was observed. The cleavage efficiency (kcat/Km) of a singly 2'-deoxy-substituted substrate at this position in the T stem was also determined. For the wild-type P RNA, the catalytic efficiency was significantly decreased compared with that of the all-ribo substrate, consistent with the notion that this 2' hydroxyl plays an important role. For the P RNA mutant, no additional effect was found upon 2'-deoxy substitution. We propose that this particular 2' hydroxyl in the pre-tRNA interacts specifically with this adenosine in the P RNA. This method should be useful in examining the role of 2' hydroxyl groups in other RNA-RNA and RNA-protein complexes.

Protein crosslinking studies suggest that Rhizobium meliloti C4-dicarboxylic acid transport protein D, a sigma 54-dependent transcriptional activator, interacts with sigma 54 and the beta subunit of RNA polymerase.

Rhizobium meliloti C4-dicarboxylic acid transport protein D (DCTD) activates transcription by a form of RNA polymerase holoenzyme that has sigma 54 as its sigma factor (referred to as E sigma 54). DCTD catalyzes the ATP-dependent isomerization of closed complexes between E sigma 54 and the dctA promoter to transcriptionally productive open complexes. Transcriptional activation probably involves specific protein-protein interactions between DCTD and E sigma 54. Interactions between sigma 54-dependent activators and E sigma 54 are transient, and there has been no report of a biochemical assay for contact between E sigma 54 and any activator to date. Heterobifunctional crosslinking reagents were used to examine protein-protein interactions between the various subunits of E sigma 54 and DCTD. DCTD was crosslinked to Salmonella typhimurium sigma 54 with the crosslinking reagents succinimidyl 4-(N-maleimidomethyl)cyclohexane-1-carboxylate and N-hydroxysulfosuccinimidyl-4-azidobenzoate. Cys-307 of sigma 54 was identified by site-directed mutagenesis as the residue that was crosslinked to DCTD. DCTD was also crosslinked to the beta subunit of Escherichia coli core RNA polymerase with succinimidyl 4-(N-maleimidomethyl)cyclohexane-1-carboxylate, but not with N-hydroxysulfosuccinimidyl-4-azidobenzoate. These data suggest that interactions of DCTD with sigma 54 and the beta subunit may be important for transcriptional activation and offer evidence for interactions between a sigma 54-dependent activator and sigma 54, as well as the beta subunit of RNA polymerase.

Detection of Transient In Vivo Interactions between Substrate and Transporter during Protein Translocation into the Endoplasmic Reticulum

The split-ubiquitin technique was used to detect transient protein interactions in living cells. Nub, the N-terminal half of ubiquitin (Ub), was fused to Sec62p, a component of the protein translocation machinery in the endoplasmic reticulum of Saccharomyces cerevisiae. Cub, the C-terminal half of Ub, was fused to the C terminus of a signal sequence. The reconstitution of a quasi-native Ub structure from the two halves of Ub, and the resulting cleavage by Ub-specific proteases at the C terminus of Cub, serve as a gauge of proximity between the two test proteins linked to Nub and Cub. Using this assay, we show that Sec62p is spatially close to the signal sequence of the prepro-α-factor in vivo. This proximity is confined to the nascent polypeptide chain immediately following the signal sequence. In addition, the extent of proximity depends on the nature of the signal sequence. Cub fusions that bore the signal sequence of invertase resulted in a much lower Ub reconstitution with Nub-Sec62p than otherwise identical test proteins bearing the signal sequence of prepro-α-factor. An inactive derivative of Sec62p failed to interact with signal sequences in this assay. These in vivo findings are consistent with Sec62p being part of a signal sequence-binding complex.

Interactions of protein antigens with antibodies.

There are now several crystal structures of antibody Fab fragments complexed to their protein antigens. These include Fab complexes with lysozyme, two Fab complexes with influenza virus neuraminidase, and three Fab complexes with their anti-idiotype Fabs. The pattern of binding that emerges is similar to that found with other protein-protein interactions, with good shape complementarity between the interacting surfaces and reasonable juxtapositions of polar residues so as to permit hydrogen-bond formation. Water molecules have been observed in cavities within the interface and on the periphery, where they often form bridging hydrogen bonds between antibody and antigen. For the most part the antigen is bound in the middle of the antibody combining site with most of the six complementarity-determining residues involved in binding. For the most studied antigen, lysozyme, the epitopes for four antibodies occupy approximately 45% of the accessible surface area. Some conformational changes have been observed to accompany binding in both the antibody and the antigen, although most of the information on conformational change in the latter comes from studies of complexes with small antigens.

The sulfur controller-2 negative regulatory gene of Neurospora crassa encodes a protein with beta-transducin repeats.

The sulfur regulatory system of Neurospora crassa is composed of a set of structural genes involved in sulfur catabolism controlled by a genetically defined set of trans-acting regulatory genes. These sulfur regulatory genes include cys-3+, which encodes a basic region-leucine zipper transcriptional activator, and the negative regulatory gene scon-2+. We report here that the scon-2+ gene encodes a polypeptide of 650 amino acids belonging to the expanding beta-transducin family of eukaryotic regulatory proteins. Specifically, SCON2 protein contains six repeated G beta-homologous domains spanning the C-terminal half of the protein. SCON2 represents the initial filamentous fungal protein identified in the beta-transducin group. Additionally, SCON2 exhibits a specific amino-terminal domain that potentially defines another subfamily of beta-transducin homologs. Expression of the scon-2+ gene has been examined using RNA hybridization and gel mobility-shift analysis. The dependence of scon-2+ expression on CYS3 function and the binding of CYS3 to the scon-2+ promoter indicate the presence of an important control loop within the N. crassa sulfur regulatory circuit involving CYS3 activation of scon-2+ expression. On the basis of the presence of beta-transducin repeats, the crucial role of SCON2 in the signal-response pathway triggered by sulfur limitation may be mediated by protein-protein interactions.

Structure of the Ets-1 pointed domain and mitogen-activated protein kinase phosphorylation site

The Pointed (PNT) domain and an adjacent mitogen-activated protein (MAP) kinase phosphorylation site are defined by sequence conservation among a subset of ets transcription factors and are implicated in two regulatory strategies, protein interactions and posttranslational modifications, respectively. By using NMR, we have determined the structure of a 110-residue fragment of murine Ets-1 that includes the PNT domain and MAP kinase site. The Ets-1 PNT domain forms a monomeric five-helix bundle. The architecture is distinct from that of any known DNA- or protein-binding module, including the helix-loop-helix fold proposed for the PNT domain of the ets protein TEL. The MAP kinase site is in a highly flexible region of both the unphosphorylated and phosphorylated forms of the Ets-1 fragment. Phosphorylation alters neither the structure nor monomeric state of the PNT domain. These results suggest that the Ets-1 PNT domain functions in heterotypic protein interactions and support the possibility that target recognition is coupled to structuring of the MAP kinase site.

Abnormal neurotransmission in mice lacking synaptic vesicle protein 2A (SV2A)

Synaptic vesicle protein 2 (SV2) is a membrane glycoprotein common to all synaptic and endocrine vesicles. Unlike many proteins involved in synaptic exocytosis, SV2 has no homolog in yeast, indicating that it performs a function unique to secretion in higher eukaryotes. Although the structure and protein interactions of SV2 suggest multiple possible functions, its role in synaptic events remains unknown. To explore the function of SV2 in an in vivo context, we generated mice that do not express the primary SV2 isoform, SV2A, by using targeted gene disruption. Animals homozygous for the SV2A gene disruption appear normal at birth. However, they fail to grow, experience severe seizures, and die within 3 weeks, suggesting multiple neural and endocrine deficits. Electrophysiological studies of spontaneous inhibitory neurotransmission in the CA3 region of the hippocampus revealed that loss of SV2A leads to a reduction in action potential-dependent γ-aminobutyric acid (GABA)ergic neurotransmission. In contrast, action potential-independent neurotransmission was normal. Analyses of synapse ultrastructure suggest that altered neurotransmission is not caused by changes in synapse density or morphology. These findings demonstrate that SV2A is an essential protein and implicate it in the control of exocytosis.

Identification of a Human VPF/VEGF 3′ Untranslated Region Mediating Hypoxia-induced mRNA Stability

Hypoxia is a prominent feature of malignant tumors that are characterized by angiogenesis and vascular hyperpermeability. Vascular permeability factor/vascular endothelial growth factor (VPF/VEGF) has been shown to be up-regulated in the vicinity of necrotic tumor areas, and hypoxia potently induces VPF/VEGF expression in several tumor cell lines in vitro. Here we report that hypoxia-induced VPF/VEGF expression is mediated by increased transcription and mRNA stability in human M21 melanoma cells. RNA-binding/electrophoretic mobility shift assays identified a single 125-bp AU-rich element in the 3′ untranslated region that formed hypoxia-inducible RNA-protein complexes. Hypoxia-induced expression of chimeric luciferase reporter constructs containing this 125-bp AU-rich hypoxia stability region were significantly higher than constructs containing an adjacent 3′ untranslated region element without RNA-binding activity. Using UV-cross-linking studies, we have identified a series of hypoxia-induced proteins of 90/88 kDa, 72 kDa, 60 kDa, 56 kDa, and 46 kDa that bound to the hypoxia stability region element. The 90/88-kDa and 60-kDa species were specifically competed by excess hypoxia stability region RNA. Thus, increased VPF/VEGF mRNA stability induced by hypoxia is mediated, at least in part, by specific interactions between a defined mRNA stability sequence in the 3′ untranslated region and distinct mRNA-binding proteins in human tumor cells.

DNA annealing by Rad52 Protein is stimulated by specific interaction with the complex of replication protein A and single-stranded DNA

Homologous recombination in Saccharomyces cerevisiae depends critically on RAD52 function. In vitro, Rad52 protein preferentially binds single-stranded DNA (ssDNA), mediates annealing of complementary ssDNA, and stimulates Rad51 protein-mediated DNA strand exchange. Replication protein A (RPA) is a ssDNA-binding protein that is also crucial to the recombination process. Herein we report that Rad52 protein effects the annealing of RPA–ssDNA complexes, complexes that are otherwise unable to anneal. The ability of Rad52 protein to promote annealing depends on both the type of ssDNA substrate and ssDNA binding protein. RPA allows, but slows, Rad52 protein-mediated annealing of oligonucleotides. In contrast, RPA is almost essential for annealing of longer plasmid-sized DNA but has little effect on the annealing of poly(dT) and poly(dA), which are relatively long DNA molecules free of secondary structure. These results suggest that one role of RPA in Rad52 protein-mediated annealing is the elimination of DNA secondary structure. However, neither Escherichia coli ssDNA binding protein nor human RPA can substitute in this reaction, indicating that RPA has a second role in this process, a role that requires specific RPA–Rad52 protein interactions. This idea is confirmed by the finding that RPA, which is complexed with nonhomologous ssDNA, inhibits annealing but the human RPA–ssDNA complex does not. Finally, we present a model for the early steps of the repair of double-strand DNA breaks in yeast.