Molecular modeling and site-directed mutagenesis of an anti-phosphotyrosine antibody predicts the combining site and allows the detection of higher affinity interactions.

Bacterially expressed Fv regions of the anti-phosphotyrosine antibody Py20 have been shown to have an affinity similar to the parent IgG. Previous studies revealed a requirement for both heavy and light chain variable regions, VH and VK joining amino acids and the heavy chain D region for high affinity binding. In order to identify amino acids which might play a role in complexing phosphotyrosine by this antibody, a molecular model for the Py20 Fv was generated using the 3-D crystal structure coordinates of a closely related IgG (R19.9) and the molecular modeling programs CHARMm and CONGEN. The resulting model was tested by mutational analysis. Twelve amino acid residues in both the Py20 heavy and light chain variable regions were altered and the effects of these mutations on phosphotyrosine recognition were determined by affinity chromatography, a competitive ELISA and an antigen overlay assay. The results from this analysis revealed that light chain tryptophan 96 and heavy chain arginines 99 and 59, both of two tyrosines 105 and 106 and a histidine 35 were necessary for high affinity binding. Surprisingly, changing the heavy chain tyrosines 105 and 106 individually resulted in an FV that bound phosphotyrosine with an apparent affinity 10-fold greater than the wild type molecule. Even though the 105 and 106 FVs are monovalent, they recognize tyrosine-phosphorylated proteins displayed on a Western blot as well as the intact IgG. Five other amino acid changes had no obvious effect on antigen binding in any of the assays used.

Requirement for both H and L chain V regions, VH and VK joining amino acids, and the unique H chain D region for the high affinity binding of an anti-phosphotyrosine antibody.

Sequence analysis of a panel of antibodies to phosphotyrosine revealed predominant H and L chain V regions in the immune response and a unique D segment in the Py20 mAb, which exhibits a high affinity for phosphotyrosine. In order to determine the influence of somatic diversity on the high affinity binding of Py20, H and L chain V regions were expressed in Escherichia coli as an Fv dimer. Whereas the H or L chain V regions of Py20 alone were unable to bind phosphotyrosine, the Fv binds phosphotyrosine with an affinity comparable with the intact IgG as determined by fluorescence quenching experiments (1.55 x 10(-7) M vs 1.25 x 10(-7) M, respectively). Substitution of the Py20 V regions with other IgG V regions that differed greatly in sequence abolished binding. A high affinity Py20-combining site was dependent on the presence of the unique D-D segment. Replacement of the Py20 D-D region with a single homologous D region resulted in a decrease in affinity (5.9 x 10(-7) M). Substitution of this D-D region for the D region of another anti-phosphotyrosine antibody that is known to bind phosphotyrosine weakly (1 x 10(-3) M) conferred high affinity binding. Removal of three tyrosines from the first of the two D regions was accompanied by a fivefold reduction in affinity for phosphotyrosine. In addition, changing the VK and VH junctional amino acids resulted in a complete loss of binding. Therefore, the formation of the high affinity Py20 combining site requires both a H and L chain that are similar in sequence to those of Py20 including the unique D region and the junctional amino acids.

Complementary distributions of vinculin and dystrophin define two distinct sarcolemma domains in smooth muscle.

The sarcolemma of the smooth muscle cell displays two alternating structural domains in the electron microscope: densely-staining plaques that correspond to the adherens junctions and intervening uncoated regions which are rich in membrane invaginations, or caveolae. The adherens junctions serve as membrane anchorage sites for the actin cytoskeleton and are typically marked by antibodies to vinculin. We show here by immunofluorescence and immunoelectron microscopy that dystrophin is specifically localized in the caveolae-rich domains of the smooth muscle sarcolemma, together with the caveolae-associated molecule caveolin. Additional labeling experiments revealed that beta 1 integrin and fibronectin are confined to the adherens junctions, as indicated by their codistribution with vinculin and tensin. Laminin, on the other hand, is distributed around the entire cell perimeter. The sarcolemma of the smooth muscle cell is thus divided into two distinct domains, featuring different and mutually exclusive components. This simple bipartite domain organization contrasts with the more complex organization of the skeletal muscle sarcolemma: smooth muscle thus offers itself as a useful system for localizing, among other components, potential interacting partners of dystrophin.

The sequence of human caveolin reveals identity with VIP21, a component of transport vesicles.

Caveolin is a protein present in membrane specializations termed caveolae where it may play a structural role. Previous studies on the cDNA sequence of chicken caveolin demonstrated that it is an integral membrane protein without significant homology to any known protein. The cDNA sequence of human lung caveolin is presented here. A striking sequence homology is observed with the chicken protein, as well as a very recently reported protein termed VIP21 [(1992) J. Cell Biol. 118, 1003-1014], a putative vesicle transport protein isolated from MDCK cells. Genomic Southern blots suggest that caveolin is present in a single copy, and Northern blots confirm that the caveolin mRNA is elevated in muscle.

Sequence and expression of caveolin, a protein component of caveolae plasma membrane domains phosphorylated on tyrosine in Rous sarcoma virus-transformed fibroblasts.

Caveolae are flask-shaped plasma membrane invaginations abundant in endothelium and muscle but may be present in all cells. They contain a filamentous coat material thought to be important in their structure and function. Recent studies have demonstrated that a 22-kDa protein (caveolin) phosphorylated on tyrosine in Rous sarcoma virus-transformed chicken fibroblasts is a component of the caveolae coat on the inner aspect of the membrane. We now report the deduced protein sequence of chicken caveolin derived from cDNA PCR products and genomic DNA clones. Caveolin is a unique protein of 178 amino acids and displays little sequence similarity to other proteins in the GenBank data base. Hydrophobicity predictions indicate an unusual 40-amino acid hydrophobic region near the C terminus that may be used to anchor the protein to the membrane. When chicken caveolin was expressed in mouse 3T3 cells and detected by immunofluorescence microscopy, the typical caveolae pattern was observed. This includes brightly fluorescent membrane patches in many cases concentrated at the margin of cells and in arrays. Caveolae may be distinct from other membrane domains due at least in part to caveolin.

Localization of a 215-kDa tyrosine-phosphorylated protein that cross-reacts with tensin antibodies.

Tyrosine phosphorylation of cytoskeletal proteins at adhesive junctions has been speculated to play a role in the regulation of cell signaling at these sites. Previously, monoclonal antibodies were generated against phosphotyrosine-containing proteins from Rous sarcoma virus-transformed chick embryo fibroblasts, resulting in two antibodies which recognized antigens of 76 and 215 kDa that localized to focal contacts. We have now localized the 215-kDa antigen to a number of adhesive junctions in vivo, including the zonula adherens, intercalated discs, and myotendinous and neuromuscular junctions. In sections of skeletal muscle and in isolated myofibrils, the 215-kDa protein was localized to the I-band. By immunoprecipitation and immunoblot analysis, we determined that the 215-kDa antigen cross-reacts with a polyclonal anti-tensin antibody.

Tyrosine phosphorylation of mitogen-activated protein kinase in cells with tyrosine kinase-negative epidermal growth factor receptors.

Epidermal growth factor (EGF) treatment of cells expressing the human EGF receptor (EGFr) results in rapid tyrosine phosphorylation of several cellular proteins including mitogen-activated protein (MAP) kinase. EGF treatment of cells expressing a tyrosine kinase-inactive EGFr failed to induce the tyrosine phosphorylation of endogenous substrates in response to EGF; however, the tyrosine phosphorylation and activation of MAP kinase did occur. This observation indicates that MAP kinase is activated in response to a signal other than the tyrosine kinase activity of the EGFr. Because EGF does not stimulate cells expressing the inactive EGFr to proliferate, phosphorylation of MAP kinase may not be sufficient for the EGF-dependent mitogenesis.

Caveolin, a protein component of caveolae membrane coats.

Caveolae have been implicated in the transcytosis of macromolecules across endothelial cells and in the receptor-mediated uptake of 5-methyltetrahydrofolate. Structural studies indicate that caveolae are decorated on their cytoplasmic surface by a unique array of filaments or strands that form striated coatings. To understand how these nonclathrin-coated pits function, we performed structural analysis of the striated coat and searched for the molecular component(s) of the coat material. The coat cannot be removed by washing with high salt; however, exposure of membranes to cholesterol-binding drugs caused invaginated caveolae to flatten and the striated coat to disassemble. Antibodies directed against a 22 kd substrate for v-src tyrosine kinase in virus-transformed chick embryo fibroblasts decorated the filaments, suggesting that this molecule is a component of the coat. We have named the molecule caveolin. Caveolae represent a third type of coated membrane specialization that is involved in molecular transport.

Temperature-dependent tyrosine phosphorylation of microtubule-associated protein kinase in epidermal growth factor-stimulated human fibroblasts.

Treatment of normal human fibroblasts with epidermal growth factor (EGF) results in the rapid (0.5 min) and simultaneous tyrosine phosphorylation of the EGF receptor (EGFr) and several other proteins. An exception to this tyrosine phosphorylation wave was a protein (42 kDa) that became phosphorylated on tyrosine only after a short lag time (5 min). We identified this p42 kDa substrate as the microtubule-associated protein (MAP) kinase using a monoclonal antibody to a peptide corresponding to the C-terminus of the predicted protein (Science 249, 64-67, 1990). EGF treatment of human fibroblasts at 37 degrees C for 5 min resulted in the tyrosine phosphorylation of 60-70% of MAP kinase as determined by the percent that was immunoprecipitated with antiphosphotyrosine antibodies. Like other tyrosine kinase growth factor receptors, the EGFr is activated and phosphorylated at 4 degrees C but is not internalized. Whereas most other substrates were readily tyrosine phosphorylated at 4 degrees C, MAP kinase was not. When cells were first stimulated with EGF at 4 degrees C and then warmed to 37 degrees C without EGF, tyrosine phosphorylation of MAP kinase was again observed. Treatment of cells with the protein kinase C activator phorbol myristate acetate (PMA) also resulted in the tyrosine phosphorylation of MAP kinase, and again only at 37 degrees C. Tryptic phosphopeptide maps demonstrated that EGF and PMA both induced the phosphorylation of the same peptide on tyrosine and threonine. This temperature and PMA sensitivity distinguishes MAP kinase from most other tyrosine kinase substrates in activated human fibroblasts.

Heavy and light chain variable region sequences and antibody properties of anti-phosphotyrosine antibodies reveal both common and distinct features.

Phosphotyrosine and similar analogs have been used to elicit antibodies that have found widespread use in the study of cellular tyrosine phosphorylation. In order to better understand the anti-phosphotyrosine immune response and to elucidate the details of the specific association between a tyrosine phosphate and an antibody combining site, we have undertaken a detailed comparison of antibody stability, specificity, apparent affinity, and primary structure for eight different anti-phosphotyrosine antibodies derived from immunizations with three different antigens. Two of these, 2G8 and 1G2, were derived from an immunization using azobenzylphosphonate conjugated to carrier, and five others, Py2, Py20, Py42, Py54, and Py69, were the products of an immunization with phosphotyrosine conjugated to carrier. Each of these anti-hapten antibodies was an IgG. One antibody, 129, an IgM, was the result of an immunization with a mixture of tyrosine-phosphorylated proteins which had been purified from growth factor treated cells. We found that antibody binding was significantly inhibited by millimolar levels of divalent cations or high concentrations of monovalent salt, with the exception of the antibody 129 where binding was significantly enhanced by both. Under optimal conditions, the highest apparent affinities for phosphotyrosine were observed for antibodies Py69 and Py20 (10(-6)-10(-7) M) and the lowest for 129 and 1G2 (10(-3)-10(-4). The heavy and light chain variable regions of seven of these antibodies were cloned and sequenced and a predominant anti-phosphotyrosine response was observed. The light chains of these antibodies could be assigned to one of two major VK groups, VK10 and VK19, with sequence identity between the different light chains of each class ranging from 65 to 100% at the amino acid level. Similar sequence identity was found among the heavy chain sequences (89-98% identity at the amino acid level) with the exception of one antibody, 2G8, which was only distantly related to the others (61-64% amino acid identity). These heavy chains belong to the same heavy chain family, J558. Two of the antibodies, Py20 and Py69, were clearly derived from the same progenitor cell since both share a highly unusual apparent V-D-D-JH organization. However, a significant level of somatic mutation has occurred between the two antibodies resulting in subtle changes in their apparent affinity and specificity.

Isolation of tyrosine-phosphorylated proteins and generation of monoclonal antibodies.

The methods used above allow one to generate the tools to study individual phosphotyrosine-containing proteins. While all of the substrates we have identified in this way contain phosphotyrosine, this cannot be assumed. Some polypeptides may bind to the anti-phosphotyrosine column because of association with other phosphotyrosine-containing proteins or by nonspecific binding to the anti-phosphotyrosine affinity column. The antibodies generated to these substrates allow one to assess potential questions directly. At a minimum, phosphoamino acid analysis must be performed on the candidate substrate labeled in intact cells with 32PO4 and precipitated with the antibody to the substrate. Methods for this can be found in Cooper and Hunter.

Immunodetection of the ligand-activated receptor for epidermal growth factor.

Many receptors for cellular growth factors are known to be protein tyrosine kinases which become activated upon ligand binding at their extracellular domain. We describe here a method to detect the activation state of Epidermal Growth Factor receptor (EGFr) with a monoclonal antibody (mAb74). This antibody was found to preferentially recognize the ligand-activated EGFr as detected by immunoprecipitation, Western blotting and immunocytochemical techniques. mAb74 did not recognize other tyrosine-phosphorylated proteins and was not inhibited by phosphotyrosine, suggesting that it is recognizing an epitope specific for the ligand-activated EGF receptor. The reactivity of mAb74 towards EGFr was closely correlated with the EGF-dependent tyrosine phosphorylation of endogenous substrates. This antibody allows one to detect the activated EGF receptor in vitro or in vivo even in a complex mixture of other tyrosine kinases and substrates.

Paxillin: a new vinculin-binding protein present in focal adhesions.

The 68-kD protein (paxillin) is a cytoskeletal component that localizes to the focal adhesions at the ends of actin stress fibers in chicken embryo fibroblasts. It is also present in the focal adhesions of Madin-Darby bovine kidney (MDBK) epithelial cells but is absent, like talin, from the cell-cell adherens junctions of these cells. Paxillin purified from chicken gizzard smooth muscle migrates as a diffuse band on SDS-PAGE gels with a molecular mass of 65-70 kD. It is a protein of multiple isoforms with pIs ranging from 6.31 to 6.85. Using purified paxillin, we have demonstrated a specific interaction in vitro with another focal adhesion protein, vinculin. Cleavage of vinculin with Staphylococcus aureus V8 protease results in the generation of two fragments of approximately 85 and 27 kD. Unlike talin, which binds to the large vinculin fragment, paxillin was found to bind to the small vinculin fragment, which represents the rod domain of the molecule. Together with the previous observation that paxillin is a major substrate of pp60src in Rous sarcoma virus-transformed cells (Glenney, J. R., and L. Zokas. 1989. J. Cell Biol. 108:2401-2408), this interaction with vinculin suggests paxillin may be a key component in the control of focal adhesion organization.

Microinjection of antibodies to the calpactin I light chain in MDBK cells causes precipition of the cytoskeletal calpactin I complex without affecting the distribution of related proteins.

The calpactin I complex is composed of two heavy chain (39K) and two light chain (11K) subunits. The heavy chain is a member of a protein family that includes lipocortins, endonexin and chromobindins while the light chain is a member of the S100 family (7 distinct members are known). We have found that the kidney epithelial cell line MDBK expresses four members of the heavy chain family and two members of the light chain protein family. Antibodies to the light chain of calpactin I were found to cause the precipitation of injected antibody together with the associated heavy chain without apparent effect on the distribution of related proteins. This suggests a differential targeting of various members of the calpactin heavy and light chain families even within the same cell.

Tyrosine phosphorylation of a 22-kDa protein is correlated with transformation by Rous sarcoma virus.

Recent studies from this laboratory have identified novel cytoskeletal proteins that are phosphorylated on tyrosine in vivo in Rous sarcoma virus-transformed chick fibroblasts (Glenney, J. R., Jr., and Zokas, L. (1989) J. Cell Biol. 108, 2401-2408). In the present report, the phosphorylation of these proteins was examined in cells expressing the nonmyristylated mutants of src that are not transformed. A good correlation was found between transformation and the tyrosine phosphorylation of a 22-kDa protein. Tyrosine phosphorylation of the 22-kDa protein was reduced more than 95% in cells expressing the nonmyristylated mutants of src. Size fractionation revealed that the 22-kDa phosphoprotein in transformed chick fibroblasts is found in a Mr 150,000 complex. Monoclonal antibodies were used to screen various chicken tissues where the 22-kDa protein was found at high levels in muscle and lung with low levels in epithelial cells and brain. The 22-kDa protein becomes an excellent candidate for a mediator of transformation by the tyrosine kinase class of oncogenes.

Novel tyrosine kinase substrates from Rous sarcoma virus-transformed cells are present in the membrane skeleton.

We have previously reported the production of monoclonal antibodies directed against phosphotyrosine, which is the modification induced by many oncogene products and growth factor receptors. One of these antiphosphotyrosine antibodies (py20) was used in affinity chromatography to isolate phosphotyrosine (PY)-containing proteins from Rous sarcoma virus-transformed chick embryo fibroblasts (RSV-CEFs). Mice were immunized with these PY-proteins for the production of monoclonal antibodies to individual substrates. Fifteen antibodies were generated in this way to antigens with molecular masses of 215, 76, 60, and 22 kD. Antibodies to individual substrates were used to analyze the subcellular location in normal and RSV-CEFs. Antibodies to the 215- and 76-kD antigen stained focal contacts when used in immunofluorescence microscopy while anti-22-kD protein antibodies resulted in punctate staining concentrated in the margins of cells and in parallel arrays. Both distributions were altered in transformed cells. When cells were extracted with nonionic detergent under conditions that stabilize the cytoskeleton, 50% of the 76-kD protein and greater than 90% of the 22-kD protein fractionated with the cytoskeleton. This study offers a new approach to both the identification of membrane skeletal proteins in fibroblasts and changes that occur upon transformation by an activated tyrosine kinase.

Isolation of a new member of the S100 protein family: amino acid sequence, tissue, and subcellular distribution.

A low molecular mass protein which we term S100L was isolated from bovine lung. S100L possesses many of the properties of brain S100 such as self association, Ca++-binding (2 sites per subunit) with moderate affinity, and exposure of a hydrophobic site upon Ca++-saturation. Antibodies to brain S100 proteins, however, do not cross react with S100L. Tryptic peptides derived from S100L were sequenced revealing similarity to other members of the S100 family. Oligonucleotide probes based on these sequences were used to screen a cDNA library derived from a bovine kidney cell line (MDBK). A 562-nucleotide cDNA was sequenced and found to contain the complete coding region of S100L. The predicted amino acid sequence displays striking similarity, yet is clearly distinct from other members of the S100 protein family. Polyclonal and monoclonal antibodies were raised against S100L and used to determine the tissue and subcellular distribution of this molecule. The S100L protein is expressed at high levels in bovine kidney and lung tissue, low levels in brain and intestine, with intermediate levels in muscle. The MDBK cell line was found to contain both S100L and the calpactin light chain, another member of this protein family. S100L was not found associated with a higher molecular mass subunit in MDBK cells while the calpactin light chain was tightly bound to the calpactin heavy chain. Double label immunofluorescence microscopy confirmed the observation that the calpactin light chain and S100L have a different distribution in these cells.

Differential localization of calmodulin, the 67kDa calcimedin and calpactin II in secretory ameloblasts.

By using specific antibodies, we have shown by Western blots that dental tissues contain calmodulin, the 67 kDa calcimedin and calpactin II. Moreover, by immunogold electron microscopy, we were able to compare the intracellular distribution of these three calcium-binding proteins in secretory ameloblasts. They were all found in the cytosol of these cells but only calcimedin was detected in the mitochondria. Calpactin II was the only one present in secretory vesicles. Twice as much calmodulin and calpactin II were detected in cell bodies as in Tomes' processes, but calcimedin was more abundant in the latter. The presence of these calcium mediators in well defined areas of ameloblasts may indicate differential ways for these cells to regulate different calcium-dependent processes during enamel formation.