Functionally distinct serine phosphorylation sites of p36, the cellular substrate of retroviral protein kinase; differential inhibition of reassociation with p11.

P36 was originally defined as the major cytoplasmic target of retrovirally coded tyrosine-kinases. While recently much has been learned about its biochemistry, the functional importance of its tyrosine and serine phosphorylation has not been approached. As p36 is now understood as a multi-ligand protein its in vitro phosphorylation by three different serine/threonine kinases was followed. Monomeric p36 is a much better substrate than the complex containing two copies each of p36 and p11 (protein I). All p36 phosphorylation sites occur within the amino-terminal 29 residues specifically released by mild proteolysis. As this region harbors an important interaction site for p11 the reduced phosphorylation of p36 in the protein I complex results most likely from a lowered accessibility. Phosphorylation of p36 is serine specific. Reconstitution experiments define at least two functionally distinct sites. One product of protein kinase C reconstitutes with p11 to protein I, while this complex formation normal for p36 is observed neither for the second phosphorylation product nor for the derivatives resulting from phosphorylation by calmodulin or cAMP dependent kinases. The results lend direct support to the hypothesis that phosphorylation of p36 can modulate one of its molecular functions. Obvious implications for other Ca2+-dependent lipid binding proteins are discussed.     

The regulatory chain in the p36-kd substrate complex of viral tyrosine-specific protein kinases is related in sequence to the S-100 protein of glial cells.

The major cytoplasmic target of various tyrosine-specific protein kinases is a 36-kd protein (p36). This protein can exist as a monomer or as a complex with a small subunit which seems to have a regulatory function. Amino acid sequence analysis of the small subunit from porcine intestine documents a unique polypeptide of 95 residues with a calculated mol. wt. close to 11 kd (p11). Since an immunologically related subunit of the same electrophoretic mobility is also found in the corresponding complex of chicken intestine p11 is well conserved across species. Unexpectedly, the sequence of p11 shows a high homology with the glia-specific protein S-100 whose biological function is not known. Although both proteins are dimers of rather small polypeptides we have not been able to detect in our preparations of p11 the moderate Ca2+ binding known for S-100. Certain implications of this sequence relation are discussed.     

The submembranous location of p11 and its interaction with the p36 substrate of pp60 src kinase in situ

p36, a major cytoplasmic substrate of pp60 src kinase, is present beneath the plasma membrane. It can be isolated either as a monomer or as a heterotetramer (protein I) containing two copies each of p36 and a unique p11 polypeptide. To compare the expression rules of p36 and p11 as well as their cellular distributions, monoclonal antibodies to the two porcine proteins were isolated. In tissue culture cells p11-specific antibodies decorated the same submembranous compartment previously seen with antibodies to p36 and fodrin or spectrin and followed the p36 images under all fixation/extraction conditions tested. Immunofluorescence microscopy on tissue sections showed coincident expression patterns of both proteins confirming and extending previous results with p36 antibodies. Antibodies with limited cross-species reaction have been used to trace the fate of porcine p11 and p36 injected into cultured cells. Both proteins are incorporated in the submembranous compartment, where they remain in Triton cytoskeletons prepared in the presence but not in the absence of Ca2+. The incorporation of p36 in vivo conforms with its Ca2+-dependent binding to actin, fodrin, and certain phospholipids in vitro. In contrast, the incorporation of p11 seems to depend on an in situ interaction with p36 or an exchange with endogenous p11 present on p36. The combined results indicate a strong coupling of p11 and p36 in cellular compartmentalization and tissue differentiation.     

Alkylation of cysteine 82 of p11 abolishes the complex formation with the tyrosine-protein kinase substrate p36 (annexin 2, calpactin 1, lipocortin 2)

p36 (annexin 2) is the major cytoplasmic target of the src tyrosine-kinase and forms in vitro and in vivo a stable tetrameric complex in which two p36 polypeptides interact with a dimer of a unique p11 polypeptide. p11 belongs into the superfamily of EF-hand proteins. Upon mild cysteine modification conditions, both cysteines (position 61 and 82) of the free p11 become substituted, and the ability to form the p36.p11 complex is lost. Under the same conditions, the 2 cysteines of p11 incorporated into the complex display differential reactivity. Here, cysteine 61 is fully substituted while cysteine-82 is protected. p11 derivatives substituted only on cysteine 61 retain binding activity for p36 unless cysteine 82 is substituted by a second cycle of modification of the isolated p11. Thus, the C-terminal extension protruding from the second EF-hand of the p11 molecule (residues 77-96) is important for the interaction with p36. As a consequence of our analysis, we report a new separation of p36 and p11 from the p36.p11 complex. This is based on a reversible cysteine modification and thus is an alternative to the denaturation and renaturation cycle used previously.     

Deletions in the SH2 domain of p60v-src prevent association with the detergent-insoluble cellular matrix.

p60v-src has been shown to associate with a detergent-insoluble cellular matrix containing cytoskeletal proteins, but p60c-src does not bind to this matrix. We analyzed the association of mutant src proteins with the matrix and found that mutants which lack an amino-terminal portion (residues 149 to 169) of the SH2 domain cannot bind to the matrix. Neither the SH3 region nor other portions of the SH2 region were required for association. We also tested protein kinase-defective mutants and chimeras of p60v-src and p60c-src. We found a strong correlation between the kinase activity of p60src and its association with the detergent-insoluble matrix. Double infection of kinase-defective and kinase-active mutants did not result in matrix binding of the kinase-defective src proteins. We also found that Tyr-416, the major site of autophosphorylation in p60v-src, was not required for matrix association.     

The cDNA sequence for the protein-tyrosine kinase substrate p36 (calpactin I heavy chain) reveals a multidomain protein with internal repeats

We have isolated and sequenced a full-length cDNA clone for the protein-tyrosine kinase substrate p36 (calpactin I heavy chain). This sequence predicts a 339 amino acid (Mr 38,493) protein containing an N-terminal region of 20 amino acids, known to interact with a 10 kd protein (light chain), and a C-terminal region, found to contain two Ca2+/phospholipid-binding sites, that can be aligned as four 70 amino acid repeats. A single p36 gene was detected in the mouse genome, and a major p36 mRNA of 1.6 kb was found to be expressed in different mouse tissues. Unexpectedly, p36 and the phospholipase A2 inhibitor lipocortin I were found to be 50% identical in sequence over the C-terminal 300 residues. The function of p36 and its relation to other proteins are discussed.     

A conserved Gbeta binding (GBB) sequence motif in Ste20p/PAK family protein kinases

Serine/threonine protein kinases of the Ste20p/PAK family are highly conserved from yeast to man. These protein kinases have been implicated in the signaling from heterotrimeric G proteins to mitogen-activated protein (MAP) kinase cascades and to cytoskeletal components such as myosin-I. In the yeast Saccharomyces cerevisiae, Ste20p is involved in transmitting the mating-pheromone signal from the betagamma-subunits of a heterotrimeric G protein to a downstream MAP kinase cascade. We have previously shown that binding of the G-protein beta-subunit (Gbeta) to a short binding site in the non-catalytic carboxy-terminal region of Ste20p is essential fortransmitting the pheromone signal. In this study, we searched protein sequence databases for sequences that are similar to the Gbeta binding site in Ste20p. We identified a sequence motif with the consensus sequence S S L phi P L I/V x phi phi beta (x: any residue; phi: A, I, L, S, or T; beta: basic residues) that is solely present in members of Ste20p/PAK family protein kinases. We propose that this sequence motif, which we have designated GBB (Gbeta binding) motif, is specifically responsible for binding of Gbeta to Ste20p/PAK protein kinases in response to activation of heterotrimeric G protein coupled receptors. Thus, the GBB motif is a novel type of signaling domain that serves to link protein kinases of the Ste20p/PAK family to G protein coupled receptors.     

Identification of a multifunctional binding site on Ubc9p required for Smt3p conjugation

Ubiquitin-like proteins (ub-lps) are conjugated by a conserved enzymatic pathway, involving ATP-dependent activation at the C terminus by an activating enzyme (E1) and formation of a thiolester intermediate with a conjugating enzyme (E2) prior to ligation to the target. Ubc9, the E2 for SUMO, synthesizes polymeric chains in the presence of its E1 and MgATP. To better understand conjugation of ub-lps, we have performed mutational analysis of Saccharomyces cerevisiae Ubc9p, which conjugates the SUMO family member Smt3p. We have identified Ubc9p surfaces involved in thiolester bond and Smt3p-Smt3p chain formation. The residues involved in thiolester bond formation map to a surface we show is the E1 binding site, and E2s for other ub-lps are likely to bind to their E1s at a homologous site. We also find that this same surface binds Smt3p. A mutation that impairs binding to E1 but not Smt3p impairs thiolester bond formation, suggesting that it is the E1 interaction at this site that is crucial. Interestingly, other E2s and their relatives also use this same surface for binding to ubiquitin, E3s, and other proteins, revealing this to be a multipurpose binding site and suggesting that the entire E1-E2-E3 pathway has coevolved for a given ub-lp.     

Biochemical and structural studies of ASPP proteins reveal differential binding to p53, p63, and p73

ASPP1 and ASPP2 are activators of p53-dependent apoptosis, whereas iASPP is an inhibitor of p53. Binding assays showed differential binding for C-terminal domains of iASPP and ASPP2 to the core domains of p53 family members p53, p63, and p73. We also determined a high-resolution crystal structure for the C terminus of iASPP, comprised of four ankyrin repeats and an SH3 domain. The crystal lattice revealed an interaction between eight sequential residues in one iASPP molecule and the p53-binding site of a neighboring molecule. ITC confirmed that a peptide corresponding to the crystallographic interaction shows specific binding to iASPP. The contributions of ankyrin repeat residues, in addition to those of the SH3 domain, generate distinctive architecture at the p53-binding site suitable for inhibition by small molecules. These results suggest that the binding properties of iASPP render it a target for antitumor therapeutics and provide a peptide-based template for compound design.     

Cloning and expression of cDNA for human endonexin II, a Ca2+ and phospholipid binding protein

Endonexin II is a member of the family of Ca2+-dependent phospholipid binding proteins known as annexins. We cloned human endonexin II cDNA and expressed it in Escherichia coli. The apparent size and Ca2+-dependent phospholipid binding properties of purified recombinant endonexin II were indistinguishable from those of the placental protein. A single mRNA of approximately 1.6 kilobase pairs was found to be expressed in human cell lines and placenta and was in close agreement with the length of the cDNA clone (1.59 kilobase pairs). The cDNA predicted a 320-amino acid protein with a sequence that was in agreement with the previously determined partial amino acid sequence of endonexin II isolated from placenta. Endonexin II contained 58, 46, and 43% sequence identity to protein II, calpactin I (p36, protein I), and lipocortin I (p35), respectively. The partial sequence of bovine endonexin I was aligned with the sequence of endonexin II to give 63% sequence identity. Like these other proteins, endonexin II had a 4-fold internal repeat of approximately 70 residues preceded by an amino-terminal domain lacking similarity to the repeated region. It also had significant sequence identity with 67-kDa calelectrin (p68), a protein with an 8-fold internal repeat. Comparing the amino-terminal domains of these four proteins of known sequence revealed that, in general, only endonexin II and protein II had significant sequence identity (29%). Endonexin II was not phosphorylated by Ca2+/phospholipid-dependent enzyme (protein kinase C) even though it contained a threonine at a position analogous to the protein kinase C phosphorylation sites of lipocortin I, calpactin I, and protein II.     

The p36 substrate of pp60src kinase is located at the cytoplasmic surface of the plasma membrane of fibroblasts; an immunoelectron microscopic analysis

p36, a member of the family of Ca2+/lipid-binding proteins, is a major cellular substrate for the tyrosine kinase encoded by the src oncogene. It occurs in two distinct physical states, as either a monomer or a heterotetramer (protein I), which comprises two copies each of p36 and a p11 polypeptide. Immunofluorescence microscopy and cell fractionation studies suggest that p36 and p11 are located underneath the plasma membrane. To investigate whether p36 is indeed associated with the plasma membrane, we have examined its cellular distribution at the electron microscopic level with gold-labeled antibodies. In human fibroblasts, p36 is clearly associated with the cytoplasmic side of the plasma membrane and shows a uniform and regular distribution. Decoration with monoclonal antibodies against p11 reveals the same distribution, suggesting that the p36(2)p11(2) complex (protein I) occurs in the cell in a strict association with the plasma membrane. Titration experiments show that this association is Ca2+ dependent and still occurs at physiological Ca2+ concentrations (10(-7) M). Fodrin, a non-erythroid spectrin, known to bind p36 in vitro, shows a very similar distribution on the cytoplasmic side of the plasma membrane. The results suggest that in a resting and unstimulated cell p36 and p11 reside as a complex bound to the inner side of the plasma membrane.     

A fluorescence spectroscopy study of the calpactin I complex and its subunits p11 and p36: calcium-dependent conformation changes

A fluorescence study of the calpactin I complex, a heterotetramer composed of two molecules of p36 and two molecules of p11, and its subunits, was performed to clarify their conformation. The analysis of the fluorescence characteristics of the single Trp of p36, in the absence of Ca(2+), shows that: (i) in the complex, Trp is buried within the protein matrix and subjected to static quenching from nearby groups; (ii) for p36 the results are similar, but Trp seems even more shielded than in the complex. Adding Ca(2+) to the calpactin I complex, or to p36, shifts the Trp emission maximum wavelengths, and increases the quantum yields which reflect a conformational change, burying the Trp in a more hydrophobic environment. In the presence and even in the absence of Ca(2+), the binding of phosphatidylserine liposomes induces a conformational change, detected by fluorescence measurements. The Ca(2+) dissociation constants, as determined by fluorescence titrations, are similar for the complex and p36 (KD approximately 0.5 x 10(-3) M). The affinity is enhanced a 1000-times in the presence of negatively charged phospholipids. In p11, both Try residues are located in a hydrophobic environment and the protein fluorescence does not change upon Ca(2+) addition.     

The N terminus of the peroxisomal cycling receptor, Pex5p, is required for redirecting the peroxisome-associated peroxin back to the cytosol

Most newly synthesized peroxisomal matrix proteins are transported to the organelle by Pex5p, a remarkable multidomain protein involved in an intricate network of transient protein-protein interactions. Presently, our knowledge regarding the structure/function of amino acid residues 118 to the very last residue of mammalian Pex5p is quite vast. Indeed, the cargo-protein receptor domain as well as the binding sites for several peroxins have all been mapped to this region of Pex5p. In contrast, structural/functional data regarding the first 117 amino acid residues of Pex5p are still scarce. Here we show that a truncated Pex5p lacking the first 110 amino acid residues (DeltaN110-Pex5p) displays exactly the peroxisomal import properties of the full-length peroxin implying that this N-terminal domain is involved neither in cargo-protein binding nor in the docking/translocation step of the Pex5p-cargo protein complex at the peroxisomal membrane. However, the ATP-dependent export step of DeltaN110-Pex5p from the peroxisomal membrane is completely blocked, a phenomenon that was also observed for a Pex5p version lacking just the first 17 amino acid residues but not for a truncated protein comprising amino acid residues 1-324 of Pex5p. By exploring the unique properties of DeltaN110-Pex5p, the effect of temperature on the import/export kinetics of Pex5p was characterized. Our data indicate that the export step of Pex5p from the peroxisomal compartment (in contrast with its insertion into the organelle membrane) is highly dependent on the temperature.     

A single amino acid substitution makes ERK2 susceptible to pyridinyl imidazole inhibitors of p38 MAP kinase.

Mitogen-activated protein (MAP) kinases are serine/threonine kinases that mediate intracellular signal transduction pathways. Pyridinyl imidazole compounds block pro-inflammatory cytokine production and are specific p38 kinase inhibitors. ERK2 is related to p38 in sequence and structure, but is not inhibited by pyridinyl imidazole inhibitors. Crystal structures of two pyridinyl imidazoles complexed with p38 revealed these compounds bind in the ATP site. Mutagenesis data suggested a single residue difference at threonine 106 between p38 and other MAP kinases is sufficient to confer selectivity of pyridinyl imidazoles. We have changed the equivalent residue in human ERK2, Q105, into threonine and alanine, and substituted four additional ATP binding site residues. The single residue change Q105A in ERK2 enhances the binding of SB202190 at least 25,000-fold compared to wild-type ERK2. We report enzymatic analyses of wild-type ERK2 and the mutant proteins, and the crystal structure of a pyridinyl imidazole, SB203580, bound to an ERK2 pentamutant, I103L, Q105T, D106H, E109G. T110A. These ATP binding site substitutions induce low nanomolar sensitivity to pyridinyl imidazoles. Furthermore, we identified 5-iodotubercidin as a potent ERK2 inhibitor, which may help reveal the role of ERK2 in cell proliferation.     

Three Ca2+-binding proteins from porcine liver and intestine differ immunologically and physicochemically and are distinct in Ca2+ affinities

Intestinal brush-border-derived membrane vesicles contain, after demembranation in the presence of Ca2+, a subset of polypeptides that are specifically solubilized by the addition of Ca2+ chelators. As described previously, this fractionation scheme leads to the enrichment of two major proteins (I and II), one of which has been shown to be identical to the cellular p36K target of Rous sarcoma virus-encoded tyrosine-specific protein kinase (Gerke, V., and Weber, K., (1984) EMBO J. 3, 227-233). We have applied a similar protocol to membrane vesicles from porcine liver and purified a third Ca2+-binding protein (III). All three proteins had wide tissue distributions, and were absent from brain, red blood cells, and cardiac and skeletal muscle. Relative amounts varied between tissues, with protein I low in liver and protein III very low in intestine. Despite their similar extractability the three proteins (I, II, and III) are clearly distinct as far as immunological, biochemical, and physicochemical properties are concerned. They also show characteristic differences in their affinities for Ca2+ ions. The association constants of Ca2+ binding for proteins I and III have been estimated by means of indirect methods to be 10(4) M-1 (protein I) and 10(6) M-1 (protein III), while the direct Hummel-Dreyer method reveals Ca2+ binding to protein II, characterized by an association constant of 0.4 X 10(5) M-1 in the absence and 0.2 X 10(5) M-1 in the presence of 2 mM MgCl2. Conformational changes upon binding Ca2+ are described for protein II using circular dichroism, fluorescence emission, and UV difference spectra. These alterations could be attributed to an increased exposure of tyrosine and tryptophan residues to a more aqueous environment, and led to increased hydrophobicity of protein II that would explain the observed Ca2+-dependent interaction with hydrophobic matrices like phenyl-Sepharose.     

Predicted structure for the calcium-dependent membrane-binding proteins p35, p36, and p32

A new family of proteins (annexins) that bind to membranes at micromolar free Ca2+ has been recognized. Its members include an EGF-receptor kinase substrate (p35), a retroviral tyrosine kinase substrate (p36), the liver protein endonexin (p32) and an electric ray protein, calelectrin. Each protein contains four sequence repeats with a further 2-fold internal homology. Using the predicted secondary structure and pattern of conserved hydrophobic residues in each repeat, we have built a three-dimensional model that is largely isostructural with the known molecular conformation of bovine intestinal calcium-binding protein. The final (energy-refined) model had a core formed from the conserved hydrophobic residues. It differed from ICaBP principally in the length of the two Ca2+-binding loops with only one loop being able to bind. The model suggests a mechanism for interaction of these new Ca2+-binding proteins with phospholipid bilayers.     

A novel copper-binding fold for the periplasmic copper resistance protein CusF

We have determined the crystal structure of apo-CusF, a periplasmic protein involved in copper and silver resistance in Escherichia coli. The protein forms a five-stranded beta-barrel, classified as an OB-fold, which is a unique topology for a copper-binding protein. NMR chemical shift mapping experiments suggest that Cu(I) is bound by conserved residues H36, M47, and M49 located in beta-strands 2 and 3. These residues are clustered at one end of the beta-barrel, and their side chains are oriented toward the interior of the barrel. Cu(I) can be modeled into the apo-CusF structure with only minimal structural changes using H36, M47, and M49 as ligands. The unique structure and metal binding site of CusF are distinct from those of previously characterized copper-binding proteins.     

Divergent RNA binding specificity of yeast Puf2p

PUF proteins bind mRNAs and regulate their translation, stability, and localization. Each PUF protein binds a selective group of mRNAs, enabling their coordinate control. We focus here on the specificity of Puf2p and Puf1p of Saccharomyces cerevisiae, which copurify with overlapping groups of mRNAs. We applied an RNA-adapted version of the DRIM algorithm to identify putative binding sequences for both proteins. We first identified a novel motif in the 3' UTRs of mRNAs previously shown to associate with Puf2p. This motif consisted of two UAAU tetranucleotides separated by a 3-nt linker sequence, which we refer to as the dual UAAU motif. The dual UAAU motif was necessary for binding to Puf2p, as judged by gel shift, yeast three-hybrid, and coimmunoprecipitation from yeast lysates. The UAAU tetranucleotides are required for optimal binding, while the identity and length of the linker sequences are less critical. Puf1p also binds the dual UAAU sequence, consistent with the prior observation that it associates with similar populations of mRNAs. In contrast, three other canonical yeast PUF proteins fail to bind the Puf2p recognition site. The dual UAAU motif is distinct from previously known PUF protein binding sites, which invariably possess a UGU trinucleotide. This study expands the repertoire of cis elements bound by PUF proteins and suggests new modes by which PUF proteins recognize their mRNA targets.