Identification of key residues involved in adrenomedullin binding to the AM1 receptor.

BACKGROUND AND PURPOSE: Adrenomedullin (AM) is a peptide hormone whose receptors are members of the class B GPCR family. They comprise a heteromer between the GPCR, the calcitonin receptor-like receptor and one of the receptor activity-modifying proteins 1-3. AM plays a significant role in angiogenesis and its antagonist fragment AM22-52 can inhibit blood vessel and tumour growth. The mechanism by which AM interacts with its receptors is unknown. EXPERIMENTAL APPROACH: We determined the AM22-52 binding epitope for the AM1 receptor extracellular domain using biophysical techniques, heteronuclear magnetic resonance spectroscopy and alanine scanning. KEY RESULTS: Chemical shift perturbation experiments located the main binding epitope for AM22-52 at the AM1 receptor to the C-terminal 8 amino acids. Isothermal titration calorimetry of AM22-52 alanine-substituted peptides indicated that Y52, G51 and I47 are essential for AM1 receptor binding and that K46 and P49 and R44 have a smaller role to play. Characterization of these peptides at the full-length AM receptors was assessed in Cos7 cells by cAMP assay. This confirmed the essential role of Y52, G51 and I47 in binding to the AM1 receptor, with their substitution resulting in ≥100-fold reduction in antagonist potency compared with AM22-52 . R44A, K46A, S48A and P49A AM22-52 decreased antagonist potency by approximately 10-fold. CONCLUSIONS AND IMPLICATIONS: This study localizes the main binding epitope of AM22-52 to its C-terminal amino acids and distinguishes essential residues involved in this binding. This will inform the development of improved AM receptor antagonists.

Autoinhibition and activation mechanisms of the Wiskott-Aldrich syndrome protein.

The Rho-family GTPase, Cdc42, can regulate the actin cytoskeleton through activation of Wiskott-Aldrich syndrome protein (WASP) family members. Activation relieves an autoinhibitory contact between the GTPase-binding domain and the carboxy-terminal region of WASP proteins. Here we report the autoinhibited structure of the GTPase-binding domain of WASP, which can be induced by the C-terminal region or by organic co-solvents. In the autoinhibited complex, intramolecular interactions with the GTPase-binding domain occlude residues of the C terminus that regulate the Arp2/3 actin-nucleating complex. Binding of Cdc42 to the GTPase-binding domain causes a dramatic conformational change, resulting in disruption of the hydrophobic core and release of the C terminus, enabling its interaction with the actin regulatory machinery. These data show that 'intrinsically unstructured' peptides such as the GTPase-binding domain of WASP can be induced into distinct structural and functional states depending on context.

Structure of Cdc42 in complex with the GTPase-binding domain of the 'Wiskott-Aldrich syndrome' protein.

The Rho-family GTP-hydrolysing proteins (GTPases), Cdc42, Rac and Rho, act as molecular switches in signalling pathways that regulate cytoskeletal architecture, gene expression and progression of the cell cycle. Cdc42 and Rac transmit many signals through GTP-dependent binding to effector proteins containing a Cdc42/Rac-interactive-binding (CRIB) motif. One such effector, the Wiskott-Aldrich syndrome protein (WASP), is postulated to link activation of Cdc42 directly to the rearrangement of actin. Human mutations in WASP cause severe defects in haematopoletic cell function, leading to clinical symptoms of thrombocytopenia, immunodeficiency and eczema. Here we report the solution structure of a complex between activated Cdc42 and a minimal GTPase-binding domain (GBD) from WASP. An extended amino-terminal GBD peptide that includes the CRIB motif contacts the switch I, beta2 and alpha5 regions of Cdc42. A carboxy-terminal beta-hairpin and alpha-helix pack against switch II. The Phe-X-His-X2-His portion of the CRIB motif and the alpha-helix appear to mediate sensitivity to the nucleotide switch through contacts to residues 36-40 of Cdc42. Discrimination between the Rho-family members is likely to be governed by GBD contacts to the switch I and alpha5 regions of the GTPases. Structural and biochemical data suggest that GBD-sequence divergence outside the CRIB motif may reflect additional regulatory interactions with functional domains that are specific to individual effectors.

hnRNP A1 binds promiscuously to oligoribonucleotides: utilization of random and homo-oligonucleotides to discriminate sequence from base-specific binding.

To understand the range of possible and probable A1 functions in pre-mRNA biogenesis, it is important that we quantify the relative ability (or inability) of A1 to bind high affinity RNA target sequences and/or structures. Using a fluorescence competition assay we have determined apparent binding affinities for a wide range of 20mer oligos containing putative and possible A1 targets including the high affinity 'winner' sequence identified by selection/amplification [Burd,C.G and Dreyfuss,G. (1994) EMBO J. 13, 1197-1204], AUUUA sequences found in 3'-UTRs of labile mRNAs, 5'- and 3'-splice sites and telomeric sequences. With the exception of a 20mer 'winner' sequence, all other 20mers examined bind A1 with a narrow, approximately 10-fold range of affinities extending from 3.2 x 10(6) to 4.2 x 10(7) M(-1). Studies with homo-oligomers suggest this range reflects nucleotide base rather than sequence specificity and hence, it was possible to predict reasonably accurate affinities for all other 20mers examined except for the 'winner', whose unusually high affinity of 4.0 x 10(8) M(-1) results from a unique higher order structure and sequence. Since there is no known physiological role for the 'winner' 20mer sequence, these data suggest A1 generally binds indiscriminately to all available pre-mRNA sequences. Both the large abundance of A1 in vivo and its binding properties are thus consistent with it playing a structural role in pre-mRNA biogenesis.

Origins of binding specificity of the A1 heterogeneous nuclear ribonucleoprotein.

The A1 heterogeneous nuclear ribonucleoprotein (hnRNP) is the best studied of the "core" hnRNP proteins that are tightly associated with heterogeneous nuclear RNA (hnRNA) within eukaryotic nuclei. Previous studies suggested that hnRNP A1 preferentially binds (under nonequilibrium conditions) to the pyrimidine-rich span of sequence at the 3'splice site of most introns [Swanson, M.S., & Dreyfuss, G. (1988) EMBO J. 11, 3519-3529; Buvoli et al. (1990) Nucleic Acids Res. 18, 6595-6600; Ishikawa et al. (1993) Mol. Cell. Biol. 13, 4301-4310]. Recently, Burd and Dreyfuss [(1994) EMBO J. 13, 1197-1204] used selection/amplification from pools of random sequence RNA to uncover an even higher-affinity A1 oligo that contained two copies of a high-affinity consensus sequence, UAGGGU/A. We have extended these studies by using a fluorescence assay to characterize the equilibrium binding properties of A1 to each of these oligonucleotides. By also characterizing the binding of A1 to sequence-randomized control oligonucleotides, we have been able to better evaluate the inherent "sequence-specific" binding properties of A1. Although these studies indicate that under equilibrium conditions A1 cannot specifically recognize the beta-globin, 3'-splice site DNA oligo analogue studied by Buvoli et al. (1990), they confirmed the high-affinity binding to the "winner" 20-mer RNA that was uncovered via selection/amplification and that has the sequence UAUGAUAGGGACUUAGGGUG (Burd & Dreyfuss, 1994). In 0.1 M NaCl, we found that A1 has approximately 100-fold higher affinity for this winner sequence sequence than it does for either a randomized version of this sequence or a 20-mer oligo corresponding to an unrelated beta-globin intron sequence. This winner RNA oligo aggregates in solution to form an apparent dimer that may represent a G-quartet resulting from dimerization of two Hoogsteen base-paired hairpins. On the basis of salt sensitivity studies carried out with various fragments of A1, the ability of A1 to discriminate the winner sequence from its randomized control results primarily from increased ionic interactions with the glycine-rich, COOH terminal domain of A1 that extends from residue 196 to 319. Nonetheless, most of the overall energy of binding for the A1 winner complex results from determinants that are resident within the first 195 residues of A1. The unique ability of the winner sequence (but not its sequence-randomized control) to form a higher-order aggregate, which may correspond to a G-tetrad, appears to facilitate the additional ionic interactions with the COOH terminal domain. Taken together, these data suggest the need to reevaluate possible and probable functions of A1 in vivo.

Multiple RNA binding domains (RBDs) just don't add up.

One of the most common motifs for binding RNA in eukaryotes is the RNA binding domain (RBD) or RNA Recognition Motif (RRM). One of the more intriguing aspects of these proteins is their modular nature. Proteins have been found containing from one to four RRMs. In most instances, these domains have some basal level of non-sequence specific RNA binding affinity. In addition, many also have a higher affinity for a specific structure or sequence of RNA. In the cases of heterogenous nuclear ribonucleoprotein A1 (hnRNP A1), yeast poly-A binding protein and splicing factor U2AF65, the individual free energy of binding of the RBDs for RNA are not strictly additive. By invoking a model in which the amino acids connecting adjoining RBDs are considered to be flexible linkers with an interresidue spacing of about 3.5 A, it is possible to predict the apparent association constants for at least some multi-RBD proteins to single-stranded RNA. We have surveyed the literature and found that individual RBDs are separated by 'linker' sequences of highly variable length. These linkers provide a critical determinant of binding affinity and may modulate cis versus trans binding. A clearer understanding of multi-RBD binding is essential to critically evaluating the role of these proteins in RNA splicing, packaging and transport.

Both RNA-binding domains in heterogenous nuclear ribonucleoprotein A1 contribute toward single-stranded-RNA binding.

Heterogenous nuclear ribonucleoproteins (hnRNPs) such as hnRNP A1 are tightly associated with heterogenous nuclear RNAs (hnRNAs) within eukaryotic nuclei and are thought to be involved in hnRNA processing and splice site selection. The NH2-terminal two-thirds of hnRNP A1 contains two 92-amino acid RNA binding domains (RBDs) that are arranged in tandem and are more than 30% homologous with each other. Following this region is a flexible glycine-rich COOH-terminal domain. We have studied the nucleic acid binding properties of the two isolated RBDs (residues 1-92 and 93-184, respectively) and of A1 fragments corresponding to residues 1-184 and 1-196 (i.e., the latter fragment is called UP1) in order to evaluate their relative contributions to A1 binding. We have determined that the individual RBDs of A1 bind poly[r(epsilon A)], a fluorescent single-stranded RNA (ssRNA), with a surprisingly low apparent association constant of only 1.5 x 10(4) M-1 (1-92) and 4.5 x 10(4) M-1 (93-184), respectively. We hypothesize that this low affinity represents a basal level of binding that is common to most RBD-containing proteins. Oligonucleotide binding studies suggest the interaction site size for the 93-184 fragment is approximately 4 nucleotides or less and salt sensitivity studies indicate that only about 27% of the free energy of binding of this RBD derives from ionic interactions. Since the affinity of the 1-184 fragment is at least 10-fold above that of either of its component RBDs, both must contribute to binding. This conclusion is further supported by the increased occluded site size of 1-184 (n = 14 +/- 2), as compared to its 93-184 RBD (n = 6 +/- 1), and by the biphasic binding that was observed for the UP1:poly(U) interaction at pH 6.0. Our finding that the affinity of the 1-184 fragment is 1000-fold less than the product of the affinities of its 1-92 and 93-184 RBDs is consistent with these domains being joined by a flexible linker. By comparing the affinities of the 1-184 fragment with that for A1, we conclude that together the two RBDs in A1 account for only 53% of the free energy of A1 binding. Comparative binding studies with UP1 demonstrate that the short region spanning residues 185-->195 represents an important determinant of the binding affinity of A1 and, since this region contains a site of dimethylation, it may provide a mechanism for regulating the affinity of A1 for specific nucleic acid targets.

Conformation states of gramicidin A along the pathway to the formation of channels in model membranes determined by 2D NMR and circular dichroism spectroscopy.

Gramicidin A incorporated into SDS (sodium dodecyl sulfate) micelles exists as a right-handed, N-to-N-terminal beta 6.3 helical dimer [Lomize, A. L., Orechov, V. Yu., & Arseniev, A.S. (1992) Bioorg. Khim. 18, 182-189]. In the incorporation procedure to achieve the ion channel state of gramicidin A in SDS micelles, trifluoroethanol (TFE) is used to solubilize the hydrophobic peptide before addition to the aqueous/micelle solution. The conformational transition of gramicidin A to form ion channels in SDS micelles, i.e., in TFE and 10% TFE/water, has been investigated using 2D NMR and CD spectroscopy. In neat TFE, gramicidin A was found to be monomeric and may possibly exist in an equilibrium of rapidly interconverting conformers of at least three different forms believed to be left- and/or right-handed alpha and beta 4.4 helices. It was found that the interconversion between these conformers was slowed down in 55% TFE as evident by the observation of at least three different sets of d alpha N COSY peaks although CD gave a net spectrum similar to that in neat TFE. In 10% TFE gramicidin A spontaneously forms a precipitate. The precipitated species were isolated and solubilized in dioxane where gramicidin conformers undergo very slow interconversion and could be characterized by NMR. At least seven different gramicidin A conformations were found in 10% TFE. Four of thes are the same types of double helices as previously found in ethanol (i.e., a symmetric left-handed parallel beta 5.6 double helix, an unsymmetric left-handed parallel beta 5.6 double helix, a symmetric left-handed antiparallel beta 5.6 double helix, a symmetric right-handed parallel beta 5.6 double helix); the fifth is possibly a symmetric right-handed antiparallel beta 5.6 double helix. There is also evidence for the presence of at least one form of monomeric species. Previous observation on the solvent history dependence in the ease of channel incorporation may be explained by the presence of several different folding pathways to channel formation. To test this proposal, the conformation of gramicidin A in 10% DMSO and 10% methanol was studied. In the former environment, the major form was a random coil with a minor population of double-stranded helices, while in the latter, NMR spectra indicate the presence of the same double-helical conformers as found in neat methanol.