Search

The first three members of the ErbB family of receptor tyrosine kinases activate a wide variety of signaling pathways and are frequently misregulated in cancer. Much less is known about ErbB4. Here we use tandem mass spectrometry to identify 19 sites of tyrosine phosphorylation on ErbB4, and protein microarrays to quantify biophysical interactions between these sites and virtually every SH2 and PTB domain encoded in the human genome. Our unbiased approach highlighted several previously unrecognized interactions and led to the finding that ErbB4 can recruit and activate STAT1. At a systems level, we found that ErbB4 is much more selective than the other ErbB receptors. This suggests that ErbB4 may enable ErbB2 and ErbB3 to signal independently of EGFR under normal conditions, and provides a possible explanation for the protective properties of ErbB4 in cancer.

In the course of screening for new small-molecule modulators of cell motility, we discovered that quinocarmycin (also known as quinocarcin) analog DX-52-1 is an inhibitor of epithelial cell migration. While it has been assumed that the main target of DX-52-1 is DNA, we identified and confirmed radixin as the relevant molecular target of DX-52-1 in the cell. Radixin is a member of the ezrin/radixin/moesin family of membrane-actin cytoskeleton linker proteins that also participate in signal transduction pathways. DX-52-1 binds specifically and covalently to the C-terminal region of radixin, which contains the domain that interacts with actin filaments. Overexpression of radixin in cells abrogates their sensitivity to DX-52-1's antimigratory activity. Small interfering RNA-mediated silencing of radixin expression reduces the rate of cell migration. Finally, we found that DX-52-1 disrupts radixin's ability to interact with both actin and the cell adhesion molecule CD44.

We have previously described a novel putative 47kDa substrate for the protein kinase Akt (designated AS47) in 3T3-L1 adipocytes. In the present study, we have found by co-immunoprecipitation that AS47 was associated with cyclic nucleotide phosphodiesterase 3B (PDE3B) in lysates of 3T3-L1 adipocytes. The patterns of expression of AS47 and PDE3B upon 3T3-L1 adipocyte differentiation, among mouse tissues, and in adipocytes with and without the transcription factor C/EBPalpha were virtually coincident. Partial knockdown of AS47 in 3T3-L1 adipocytes with shRNA resulted in a similar reduction in PDE3B protein. These results indicate that AS47 exists in a complex with PDE3B in adipocytes and that the amount of AS47 protein regulates the amount of PDE3B.

Raf kinase inhibitor protein (RKIP) is a modulator of cell signaling that functions as an endogenous inhibitor of multiple kinases. We demonstrate here a positive role for RKIP in the regulation of cell locomotion. We discovered that RKIP is the relevant cellular target of locostatin, a cell migration inhibitor. Locostatin abrogates RKIP's ability to bind and inhibit Raf-1 kinase, and it acts by disrupting a protein-protein interaction, an uncommon mode of action for a small molecule. Small interfering RNA-mediated silencing of RKIP expression also reduces cell migration rate. Overexpression of RKIP converts epithelial cells to a highly migratory fibroblast-like phenotype, with dramatic reduction in the sensitivity of cells to locostatin. RKIP is therefore the compound's valid target and a key regulator of cell motility.

The most prominent cell-surface integrin alpha4beta1 partner, a 70-kDa protein, was isolated from MOLT-4 T leukemia cells, using anti-alpha4beta1 integrin antibody-coated beads. By mass spectrometry, this protein was identified as EWI-2, a previously described cell-surface partner for tetraspanin proteins CD9 and CD81. Wild-type EWI-2 overexpression had no effect on MOLT-4 cell tethering and adhesion strengthening on the alpha4beta1 ligand, vascular cell adhesion molecule-1 (VCAM-1), in shear flow assays. However, EWI-2 markedly impaired spreading and ruffling on VCAM-1. In contrast, a mutant EWI-2 molecule, with a different cytoplasmic tail, neither impaired cell spreading nor associated with alpha4beta1 and CD81. The endogenous wild-type EWI-2-CD81-alpha4beta1 complex was fully soluble, and highly specific as seen by the absence of other MOLT-4 cell-surface proteins. Also, it was relatively small in size (0.5 x 10(6) Da to 4 x 10(6) Da), as estimated by size exclusion chromatography. Overexpression of EWI-2 in MOLT-4 cells caused reorganization of cell-surface CD81, increased the extent of CD81-CD81, CD81-alpha4beta1, and alpha4beta1-alpha4beta1 associations, and increased the apparent size of CD81-alpha4beta1 complexes. We suggest that EWI-2-dependent reorganization of alpha4beta1-CD81 complexes on the cell surface is responsible for EWI-2 effects on integrin-dependent morphology and motility functions.

Background: Aminoglycoside antibiotic resistance is largely the result of the production of enzymes that covalently modify the drugs including kinases (APHs) with structural and functional similarity to protein and lipid kinases. One of the most important aminoglycoside resistance enzymes is AAC(6')-APH(2"), a bifunctional enzyme with both aminoglycoside acetyltransferase and kinase activities. Knowledge of enzyme active site structure is important in deciphering the molecular mechanism of antibiotic resistance and here we explored active site labeling techniques to study AAC(6')-APH(2") structure and function., Results: AAC(6')-APH(2") was irreversibly inactivated by wortmannin, a potent phosphatidylinositol 3-kinase inhibitor, through the covalent modification of a conserved lysine in the ATP binding pocket. 5'-[p-(Fluorosulfonyl)benzoyl]adenosine, an electrophilic ATP analogue and known inactivator of other APH enzymes such as APH(3')-IIIa, did not inactivate AAC(6')-APH(2"), and reciprocally, wortmannin did not inactivate APH(3')-IIIa., Conclusions: These distinct active site label sensitivities point to important differences in aminoglycoside kinase active site structures and suggest that design of broad range, ATP binding site-directed inhibitors against APHs will be difficult. Nonetheless, given the sensitivity of APH enzymes to both protein and lipid kinase inhibitors, potent lead inhibitors of this important resistance enzyme are likely to be found among the libraries of compounds directed against other pharmacologically important kinases.

The membrane glycoprotein CD36 is involved in platelet aggregation, inhibition of angiogenesis, atherosclerosis, and sequestration of malaria-parasitized erythrocytes. In this study, immunoprecipitations with anti-CD36 antibodies were performed to identify proteins that associate with CD36 in the platelet membrane. Platelets were solubilized in 1% Triton X-100, 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS), Brij 96, or Brij 99, and the proteins that coprecipitated with CD36 were identified by peptide mass spectrometry and Western blotting. The tetraspanin protein CD9 and the integrins alphaII(b)beta3 and alpha6beta1 specifically coprecipitated with CD36 from platelets that were solubilized in CHAPS and Brij 99 but not from platelets that were solubilized in Triton X-100. Only CD9 is coprecipitated with CD36 from platelets that were solubilized in Brij 96. Reciprocal immunoprecipitations with antibodies to CD9, alpha6, alphaIIb, or beta3 from Brij 99-solubilized platelets coprecipitated CD36. Coprecipitation of CD36, CD9, and alpha6beta1 was also observed on platelets from a patient with Glanzmann thrombasthenia, indicating that alphaII(b)beta3 is not required for the other proteins to associate. Colocalization of alpha6 and CD36, of CD9 and CD36, and of alpha6 and CD9 was observed on intact platelets prior to solubilization, using double immunofluorescence microscopy. These data indicate that CD36 associates with CD9 and integrins on human blood platelets. These associated proteins may mediate or participate in some of the diverse biological functions of CD36.

Background: Myriocin is a natural product that potently induces apoptosis of a murine cytotoxic T lymphocyte cell line (CTLL-2) and inhibits a serine palmitoyltransferase (SPT) activity that has been detected in cell extracts and is thought to initiate sphingolipid biosynthesis. Because SPT has never been biochemically purified and a comprehensive appraisal of myriocin-binding proteins has not been conducted, we isolated specific targets using myriocin-based affinity chromatography., Results: Myriocin derivatives were synthesized and evaluated using CTLL-2 proliferation and SPT activity assays. Guided by these results, affinity chromatography matrices were prepared and two specific myriocin-binding proteins were isolated from CTLL-2 lysates. Analyses of these polypeptides establish conclusively that they are murine LCB1 and LCB2, mammalian homologs of two yeast proteins that have been genetically linked to sphingolipid biosynthesis., Conclusion: The ability of myriocin-containing matrices to bind factors that have SPT activity and the exclusive isolation of LCB1 and LCB2 as myriocin-binding proteins demonstrates that the two proteins are directly responsible for SPT activity and that myriocin acts directly upon these polypeptides.

VP39 is a bifunctional mRNA-modifying protein that acts as both an mRNA cap-specific 2'-O-methyltransferase and a processivity factor for VP55, the vaccinia poly(A) polymerase catalytic subunit. Although regions of the protein surface required for methyltransferase function are well defined, it has been unclear whether the protein polyadenylylation function requires direct RNA contact and, if so, where the contact site(s) might be located on the protein surface. Here, we show that the VP55-VP39 heterodimer forms a stable complex with a 50mer oligonucleotide bearing a U2-N25-U motif, as opposed to the U2-N15-U motif that is optimal for stable complex formation with VP55 alone. An oligonucleotide bearing a U2-N25-U motif in which the downstream U residue is replaced with 4thioU can be efficiently photocrosslinked to VP39, but only in the context of the VP55-VP39 heterodimer. By partial proteolysis of end-labeled VP39, the site of oligonucleotide photocrosslinking was localized to the region of VP39 between residues Lys90 and Arg122. Peptide microsequencing and confirmatory mutagenesis identified the side-chain of Arg107 as the photocrosslinking site. Substitution of this residue with lysine abolished photocrosslinking entirely, consistent with the established RNA binding role of arginine in other RNA-binding proteins. This study provides clear evidence for a polyadenylylation-specific RNA-contact site on the surface of VP39, which is distinct from the RNA-binding methyltransferase "cleft" characterized in recent crystallographic and biochemical studies., (Copyright 1999 Academic Press.)

The transactivator protein Tat stimulates transcriptional elongation from the HIV-1 LTR. One mechanism by which Tat increases HIV-1 transcription is by interacting with RNA polymerase II and TFIIH to increase phosphorylation of the polymerase C-terminal domain. Recent studies indicate that specific elongation factors may also be required to modulate Tat function. Here, we used biochemical analysis and in vitro transcription assays to identify cellular factors required for Tat activation. This analysis resulted in the purification of a cellular factor Tat-CT1 which is a human homolog of the yeast transcription factor SPT5. Immunodepletion of Tat-CTl from HeLa extract demonstrated that this factor was involved in transcriptional activation by Tat. However, the absence of this factor from HeLa extract did not prevent transcriptional activation by VP16. These findings are consistent with a model in which Tat-mediated effects on transcriptional elongation are mediated in part by the action of the human homolog of the yeast transcription factor SPT5., (Copyright 1998 Academic Press Limited.)

Background: Base-excision DNA repair (BER) is the principal pathway responsible for the removal of aberrant, genotoxic bases from the genome and restoration of the original sequence. Key components of the BER pathway are DNA glycosylases, enzymes that recognize aberrant bases in the genome and catalyze their expulsion. One major class of such enzymes, glycosylase/lyases, also catalyze scission of the DNA backbone following base expulsion. Recent studies indicate that the glycosylase and lyase functions of these enzymes are mechanistically unified through a common amine-bearing residue on the enzyme, which acts as both the electrophile that displaces the aberrant base and an electron sink that facilitates DNA strand scission through imine (Schiff base)/conjugate elimination chemistry. The identity of this critical amine-bearing residue has not been rigorously established for any member of a superfamily of BER glycosylase/lyases., Results: Here, we report the identification of the active-site amine of the human 8-oxoguanine DNA glycosylase (hOgg1), a human BER superfamily protein that repairs the mutagenic 8-oxoguanine lesion in DNA. We employed Edman sequencing of an active-site peptide irreversibly linked to substrate DNA to identify directly the active-site amine of hOgg1 as the epsilon-NH2 group of Lys249. In addition, we observed that the repair-inactive but recognition-competent Cys249 mutant (Lys249-->Cys) of hOgg1 can be functionally rescued by alkylation with 2-bromoethylamine, which functionally replaces the lysine residue by generating a gamma-thia-lysine., Conclusions: This study provides the first direct identification of the active-site amine for any DNA glycosylase/lyase belonging to the BER superfamily, members of which are characterized by the presence of a helix-hairpin-helix-Gly/Pro-Asp active-site motif. The critical lysine residue identified here is conserved in all members of the BER superfamily that exhibit robust glycosylase/lyase activity. The ability to trigger the catalytic activity of the Lys249-->Cys mutant of hOgg1 by treatment with the chemical inducer 2-bromoethylamine may permit snapshots to be taken of the enzyme acting on its substrate and could represent a novel strategy for conditional activation of catalysis by hOgg1 in cells.

In this report, the pancornulins are identified as members of the spr (small, proline-rich) multigene family by amino acid sequence and mass spectrometry analyses. One of the pancornulins (14.9 kDa) is identical to the protein predicted by spr-1 clone 128. The other pancornulins (16.9 kDa and 22 kDa) are novel members of the spr family. Immunoelectron microscopy of purified cornified envelopes with a pancornulin-specific antibody established these proteins more definitively as cornified envelope precursors. In addition, two-dimensional electrophoretic analyses of keratinocyte extracts labeled enzymatically with dansylcadaverine (to identify amine acceptors) or dansylPGGQQIV (to identify amine donors) showed that both glutamine and lysine residues within the pancornulins participate in the isopeptide linkage characteristic of cornified envelope formation. These results contrasted with those obtained using involucrin, a prominent cornified envelope protein shown capable of acting only as an amine acceptor in this system. Novel partial cDNAs obtained after reverse transcription and polymerase chain reaction amplification of total messenger RNA with pancornulin-specific primers suggest that the spr multigene family may be even larger than previously described. The bifunctional reactivity of the pancornulins in cross-linking and the large number of family members identified to date suggest that the pancornulins and other spr-1-related proteins may be more important in cornified envelope formation than previously considered, perhaps functioning as "bridge" molecules during the early phases of cornified envelope assembly.

Background: Members of the Rel family of transcription factors are important in regulating the inflammatory, acute phase, and immune responses in mammals. The structural basis for sequence-specific binding by Rel proteins is poorly understood, however. In the studies reported here, a new photoaffinity labeling procedure has been used to probe DNA contacts established by a Rel protein, the p50 homodimer of NF-kappa B., Results: Using a novel post-synthetic modification method, 8-azido-2'-deoxyadenosine (N3dA), a photoactive analog of 2'-deoxyadenosine, was introduced at a specific site within a consensus DNA binding site for the NF-kappa B p50 homodimer. Upon irradiation with ultraviolet light, the N3dA-substituted DNA was efficiently photocrosslinked to p50. The crosslinked amino acid of p50 was identified as K244, which lies in the carboxy-terminal half of the Rel homology region (RHR). Mutation of K244 exerts strong effects on DNA binding, confirming the importance of this residue for p50-DNA interactions., Conclusions: We have used N3dA-containing DNA to identify a residue of NF-kappa B p50 that contacts DNA illustrating the value of this approach in studies of protein-DNA interactions. K244 appears not to contact a DNA base, but rather a phosphate moiety that lies nearby. The segment of protein sequence in which K244 resides has been implicated in dimerization. The results presented here suggest that the DNA-binding domain extends farther toward the carboxy-terminus than previously thought.

A tumor cell-derived, collagenase stimulatory factor (TCSF), previously isolated and purified from LX-1 human lung carcinoma cells and judged by immunoblotting and SDS-PAGE to contain a single protein of approximately 58 kDa, has been further analyzed for its biological activity and composition. Three significant new findings have been made. First, the biological activity of TCSF preparations was shown definitively to reside in the 58-kDa protein. This was achieved in two ways: (a) a polyclonal antibody was raised against the 58-kDa protein, after excision from an SDS-PAGE gel, and shown to inhibit the stimulation of fibroblast collagenase production by TCSF preparations; (b) the 58-kDa protein was eluted from a transblot of purified TCSF and shown to stimulate fibroblast collagenase production. Second, partial sequencing of the 58-kDa protein revealed no significant homologies with other known collagenase stimulatory factors. Third, purified TCSF was found, on transblotting to Immobilon, to contain a doublet of 58 kDa (TCSF1) and 54 kDa (TCSF2) proteins; the former was present in higher concentration than the latter. N-terminal amino acid sequencing of the two intact proteins and of four corresponding pairs of tryptic peptides derived from the two proteins showed identity in each case, indicating that TCSF1 and TCSF2 are very similar in composition. However, TCSF1 but not TCSF2 stimulated fibroblast collagenase production, confirming that the 58-kDa protein is the major active component of TCSF preparations.