A system for quantifying dynamic protein interactions defines a role for Herceptin in modulating ErbB2 interactions.

The orphan receptor tyrosine kinase ErbB2 is activated by each of the EGFR family members upon ligand binding. However, difficulties monitoring the dynamic interactions of the membrane receptors have hindered the elucidation of the mechanism of ErbB2 activation. We have engineered a system to monitor protein-protein interactions in intact mammalian cells such that different sets of protein interactions can be quantitatively compared. Application of this system to the interactions of the EGFR family showed that ErbB2 interacts stably with the EGFR and ErbB3, but fails to spontaneously homooligomerize. The widely used anti-cancer antibody Herceptin was found to effectively inhibit the interaction of the EGFR and ErbB2 but not to interfere with the interaction of ErbB2-ErbB3. Treatment of cells expressing EGFR and ErbB2 with Herceptin results in increased EGFR homooligomerization in the presence of EGF and a subsequent rapid internalization and down-regulation of the EGFR. In summary, the protein interaction system described here enabled the characterization of ErbB2 interactions within the biological context of the plasma membrane and provides insight into the mechanism of Herceptin action on cells overexpressing ErbB2.

Epitope recognition of antibodies that define the sialomucin, endolyn (CD164), a negative regulator of haematopoiesis.

Endolyn (CD164) is a sialomucin that functions as an adhesion molecule and a negative regulator of CD34+ CD38- human haematopoietic precursor cell proliferation. The 105A5 and 103B2/9E10 CD164 monoclonal antibodies (mAbs), which act as surrogate ligands, recognize distinct glycosylation-dependent classes I and II epitopes located on domain I of the native and recombinant CD164 proteins. Here, we document five new CD164 mAbs, the 96 series, that rely on conformational integrity, but not glycosylation, of exons 2- and 3-encoded CD164 domains, thereby resembling the class III mAbs, N6B6 and 67D2. Although all the 96 series class III mAbs labelled both the 105A5+ and 103B2/9E10+ cells, cross-competition and immunoblotting studies allow them to be categorized into two distinct class III subgroups, i.e. the N6B6-like subgroup that only recognizes 80-100 kDa proteins and the 67D2-like subgroup that also recognizes a higher molecular weight (>220 kDa) form. To more closely define the reactivity patterns of mAbs to the classes I and II epitopes, the global glycosylation patterns of the soluble human (h) CD164 proteins were determined using lectin binding, high-performance liquid chromatography (HPLC) and mass spectrometry. hCD164 recombinant proteins bound to the lectins, Galanthus nivalis agglutinin, Datura stramonium agglutinin, Sambucus nigra agglutinin, Maackia amurensis agglutinin and peanut agglutinin, indicating the presence of high mannose and complex N-glycans, in addition to core 1 O-glycans (the Tn antigen) and alpha2-3 and alpha2-6 sialic acid moieties. Our HPLC and mass spectrometry results revealed both high mannose and complex N-glycosylation with various numbers of branches increasing the complexity of the glycosylation pattern. Most O-glycans were small, core 1 or 2 based. High levels of sialylation in alpha2-3 and alpha2-6 linkages, without sialyl-Lewis X, indicate that the majority of these hCD164 recombinant proteins are unable to bind to selectins in our assay system, but may interact with Siglec molecules.

Homophilic adhesion of human CEACAM1 involves N-terminal domain interactions: structural analysis of the binding site.

CEACAM1 on leukocytic, endothelial, and epithelial cells functions in homophilic adhesion, tumor suppression, regulating cell adhesion and proliferation, and in heterophilic adhesion as a receptor for E-selectin and Neisseria meningiditis, Neisseria gonorrhoeae, Haemophilus influenzae, and murine coronaviruses. The 8 transmembrane isoforms of human CEACAM1 possess an extracellular N-terminal IgV domain, followed by variable numbers of IgC2 domains. To establish which key amino acids contribute specifically to CEACAM1 homophilic adhesion, exposed amino acids in the N-terminal domain of a soluble form of CEACAM1 were subjected to mutagenesis. Analyses of mutant proteins with conformationally dependent antibodies indicated that most mutations did not substantially affect the structural integrity of CEACAM1. Nevertheless, decreased adhesion was observed for the single mutants V39A or D40A (single-letter amino acid codes) in the CC' loop and for the triple mutants located in the GFCC'C" face of the N-terminal domain. Interestingly, whereas single mutations in R64 or D82 that are predicted to form a salt bridge between the base of the D and F beta strands close to the critical V39 and D40 residues also abolish adhesion, an amino acid swap (R64D and D82R), which maintains the salt bridge was without significant effect. These studies indicate that the CC' loop plays a crucial role in the homophilic adhesion of CEACAM1. They further predict that specific hydrophobic amino acid residues on the nonglycosylated GFCC'C" face of CEACAM1 N-terminal domain are not only involved in heterophilic interactions with Opa proteins and H influenzae, but are also critical for protein-protein interactions between 2 CEACAM1 molecules on opposing cells.

Anuria, omphalocele, and perinatal lethality in mice lacking the CD34-related protein podocalyxin.

Podocalyxin is a CD34-related sialomucin that is expressed at high levels by podocytes, and also by mesothelial cells, vascular endothelia, platelets, and hematopoietic stem cells. To elucidate the function of podocalyxin, we generated podocalyxin-deficient (podxl(-/)-) mice by homologous recombination. Null mice exhibit profound defects in kidney development and die within 24 hours of birth with anuric renal failure. Although podocytes are present in the glomeruli of the podxl(-/)- mice, they fail to form foot processes and slit diaphragms and instead exhibit cell--cell junctional complexes (tight and adherens junctions). The corresponding reduction in permeable, glomerular filtration surface area presumably leads to the observed block in urine production. In addition, podxl(-/)- mice frequently display herniation of the gut (omphalocele), suggesting that podocalyxin may be required for retraction of the gut from the umbilical cord during development. Hematopoietic and vascular endothelial cells develop normally in the podocalyxin-deficient mice, possibly through functional compensation by other sialomucins (such as CD34). Our results provide the first example of an essential role for a sialomucin in development and suggest that defects in podocalyxin could play a role in podocyte dysfunction in renal failure and omphalocele in humans.

Relationship between novel isoforms, functionally important domains, and subcellular distribution of CD164/endolyn.

Functional analyses have indicated that the human CD164 sialomucin may play a key role in hematopoiesis by facilitating the adhesion of human CD34(+) cells to the stroma and by negatively regulating CD34(+)CD38(lo/-) cell proliferation. We have identified three novel human CD164 variants derived by alternative splicing of bona fide exons from a single genomic transcription unit. The predominant CD164(E1-6) isoform, encoded by six exons, is a type I transmembrane protein containing two extracellular mucin domains (I and II) interrupted by a cysteine-rich non-mucin domain. The 103B2/9E10 and 105A5 epitopes, which specify ligand binding characteristics, are located on the exon 1-encoded mucin domain I. Three human CD164(E1-6) mRNA species, exhibiting differential polyadenylation site usage, are differentially expressed in hematopoietic and non-hematopoietic tissues. This study provides additional evidence that human CD164(E1-6) represents the ortholog of murine MGC-24v and rat endolyn. Comparative analysis of murine MGC-24v/CD164(E1-6) with human CD164(E1-6) revealed two potential splice variants and a similar genomic structure. Whereas the human CD164 gene is located on chromosome 6q21, the mouse gene occurs in a syntenic region on chromosome 10B1-B2. By confocal microscopy, human CD164 in CD34(+)CD38(+) hematopoietic progenitor (KG1B) and epithelial cell lines appears to be localized primarily in endosomes and lysosomes, with low concentrations at the cell surface. However, in a minority of KG1B cells, CD164 is more prominently expressed at the plasma membrane and in the recycling endosomes, suggesting that its distribution is regulated in cells of hematopoietic origin.

CD164 monoclonal antibodies that block hemopoietic progenitor cell adhesion and proliferation interact with the first mucin domain of the CD164 receptor.

The novel sialomucin, CD164, functions as both an adhesion receptor on human CD34+ cell subsets in bone marrow and as a potent negative regulator of CD34+ hemopoietic progenitor cell proliferation. These diverse effects are mediated by at least two functional epitopes defined by the mAbs, 103B2/9E10 and 105A5. We report here the precise epitope mapping of these mAbs together with that of two other CD164 mAbs, N6B6 and 67D2. Using newly defined CD164 splice variants and a set of soluble recombinant chimeric proteins encoded by exons 1-6 of the CD164 gene, we demonstrate that the 105A5 and 103B2/9E10 functional epitopes map to distinct glycosylated regions within the first mucin domain of CD164. The N6B6 and 67D2 mAbs, in contrast, recognize closely associated and complex epitopes that rely on the conformational integrity of the CD164 molecule and encompass the cysteine-rich regions encoded by exons 2 and 3. On the basis of their sensitivities to reducing agents and to sialidase, O-sialoglycoprotease, and N-glycanase treatments, we have characterized CD164 epitopes and grouped them into three classes by analogy with CD34 epitope classification. The class I 105A5 epitope is sialidase, O-glycosidase, and O-sialoglycoprotease sensitive; the class II 103B2/9E10 epitope is N-glycanase, O-glycosidase, and O-sialoglycoprotease sensitive; and the class III N6B6 and 67D2 epitopes are not removed by such enzyme treatments. Collectively, this study indicates that the previously observed differential expression of CD164 epitopes in adult tissues is linked with cell type specific post-translational modifications and suggests a role for epitope-associated carbohydrate structures in CD164 function.

Functionally defined CD164 epitopes are expressed on CD34(+) cells throughout ontogeny but display distinct distribution patterns in adult hematopoietic and nonhematopoietic tissues.

Three distinct classes of epitopes on human CD164 have been identified. Two of these, recognized by the monoclonal antibodies 105A5 and 103B2/9E10, are the CD164 class I and class II functionally defined epitopes, which cooperate to regulate adhesion and proliferation of CD34(+) cell subsets. In this article, we demonstrate that these 2 CD164 epitopes are expressed on CD34(+) cells throughout ontogeny, in particular on CD34(+ )cell clusters associated with the ventral floor of the dorsal aorta in the developing embryo and on CD34(+) hematopoietic precursor cells in fetal liver, cord blood, and adult bone marrow. While higher levels of expression of these CD164 epitopes occur on the more primitive AC133(hi)CD34(hi)CD38(lo/-) cell population, they also occur on most cord blood Lin(-)CD34(lo/-)CD38(lo/- )cells, which are potential precursors for the AC133(hi)CD34(hi)CD38(lo/-) subset. In direct contrast to these common patterns of expression on hematopoietic precursor cells, notable differences in expression of the CD164 epitopes were observed in postnatal lymphoid and nonhematopoietic tissues, with the class I and class II CD164 epitopes generally exhibiting differential and often reciprocal cellular distribution patterns. This is particularly striking in the colon, where infiltrating lymphoid cells are CD164 class I-positive but class II-negative, while epithelia are weakly CD164 class II-positive. Similarly, in certain lymphoid tissues, high endothelial venules and basal and subcapsular epithelia are CD164 class II-positive, while lymphoid cells are CD164 class I-positive. It therefore seems highly likely that these CD164 class I and II epitopes will mediate reciprocal homing functions in these tissue types.

CD164, a novel sialomucin on CD34(+) and erythroid subsets, is located on human chromosome 6q21.

CD164 is a novel 80- to 90-kD mucin-like molecule expressed by human CD34(+) hematopoietic progenitor cells. Our previous results suggest that this receptor may play a key role in hematopoiesis by facilitating the adhesion of CD34(+) cells to bone marrow stroma and by negatively regulating CD34(+) hematopoietic progenitor cell growth. These functional effects are mediated by at least two spatially distinct epitopes, defined by the monoclonal antibodies (MoAbs), 103B2/9E10 and 105A5. In this report, we show that these MoAbs, together with two other CD164 MoAbs, N6B6 and 67D2, show distinct patterns of reactivity when analyzed on hematopoietic cells from normal human bone marrow, umbilical cord blood, and peripheral blood. Flow cytometric analyses revealed that, on average, 63% to 82% of human bone marrow and 55% to 93% of cord blood CD34(+) cells are CD164(+), with expression of the 105A5 epitope being more variable than that of the other identified epitopes. Extensive multiparameter flow cytometric analyses were performed on cells expressing the 103B2/9E10 functional epitope. These analyses showed that the majority (>90%) of CD34(+) human bone marrow and cord blood cells that were CD38(lo/-) or that coexpressed AC133, CD90(Thy-1), CD117(c-kit), or CD135(FLT-3) were CD164(103B2/9E10)+. This CD164 epitope was generally detected on a significant proportion of CD34(+)CD71(lo/-) or CD34(+)CD33(lo/-) cells. In accord with our previous in vitro progenitor assay data, these phenotypes suggest that the CD164(103B2/9E10) epitope is expressed by a very primitive hematopoietic progenitor cell subset. It is of particular interest to note that the CD34(+)CD164(103B2/9E10)lo/- cells in bone marrow are mainly CD19(+) B-cell precursors, with the CD164(103B2/9E10) epitope subsequently appearing on CD34(lo/-)CD19(+) and CD34(lo/-)CD20(+) B cells in bone marrow, but being virtually absent from B cells in the peripheral blood. Further analyses of the CD34(lo/-)CD164(103B2/9E10)+ subsets indicated that one of the most prominent populations consists of maturing erythroid cells. The expression of the CD164(103B2/9E10) epitope precedes the appearance of the glycophorin C, glycophorin A, and band III erythroid lineage markers but is lost on terminal differentiation of the erythroid cells. Expression of this CD164(103B2/9E10) epitope is also found on developing myelomonocytic cells in bone marrow, being downregulated on mature neutrophils but maintained on monocytes in the peripheral blood. We have extended these studies further by identifying Pl artificial chromosome (PAC) clones containing the CD164 gene and have used these to localize the CD164 gene specifically to human chromosome 6q21.

The N-domain of the human CD66a adhesion molecule is a target for Opa proteins of Neisseria meningitidis and Neisseria gonorrhoeae.

Using COS (African green monkey kidney) cells transfected with cDNAs encoding human cell surface molecules, we have identified human cellular receptors for meningococcal virulence-associated Opa proteins, which are expressed by the majority of disease and carrier isolates. These receptors belong to the immunoglobulin superfamily of adhesion molecules and are expressed on epithelial, endothelial and phagocytic cells. Using soluble chimeric receptor molecules, we have demonstrated that meningococcal Opa proteins bind to the N-terminal domain of biliary glycoproteins (classified as BGP or CD66a) that belong to the CEA (CD66) family. Moreover, the Opa proteins of the related pathogen Neisseria gonorrhoeae, responsible for urogenital infections, also interact with this receptor, making CD66a a common target for pathogenic neisseriae. Over 95% of Opa-expressing clinical and mucosal isolates of meningococci and gonococci were shown to bind to the CD66 N-domain, demonstrating the presence of a conserved receptor-binding function in the majority of neisserial Opa proteins.

Identification of alpha v beta 3 as a heterotypic ligand for CD31/PECAM-1.

CD31 (PECAM-1) is a member of the immunoglobulin gene superfamily (IgSF) and has an important role in a number of endothelial cell functions including angiogenesis, inflammation, integrin activation and cell-cell adhesion. CD31 has both homotypic and heterotypic adhesive properties and in common with other IgSF members contains multiple functional domains. Using chimaeric fusion proteins of CD31 and a panel of haematopoietic cell lines we show that CD31 can bind cells in a predominantly homotypic or heterotypic manner depending on the cell line used. Heterotypic binding was found to be cation and temperature dependent and enhanced by Mn2+: all features of integrin mediated binding. Using a panel of anti-CD31 and anti-integrin antibodies we show that alpha v beta 3 is a ligand for CD31 on the monocytic cell line U937. The specificity of the interaction between alpha v beta 3 and CD31 was further confirmed by solid phase binding assays and the use of alpha v beta 3 transfected cells which bound CD31 specifically. Furthermore, we have mapped the binding site for alpha v beta 3 to domains 1 and 2 of CD31. The interaction of CD31 with alpha v beta 3 may be important in many aspects of endothelial function including leukocyte-endothelial transmigration and angiogenesis.

The identification and characterization of a novel human differentiation-inhibiting protein that selectively blocks erythroid differentiation.

We have isolated a novel inhibitor of erythropoietic differentiation from the plasma of a patient suffering from idiopathic pure red cell aplasia. This differentiation-inhibiting protein (DIP) specifically blocked the differentiation of human burst-forming unit-erythroid (BFU-E), but not colony-forming unit-erythroid (CFU-E) cells. DIP also blocked the maturation of murine BFU-E cells, but not CFU-E or CFU-granulocyte-macrophage cells, and it inhibited the dimethyl sulfoxide (DMSO)-induced differentiation of Friend murine erythroleukemia cells (FLC) at levels between 10(-10) and 10(-12) mol/L. DIP activity was not detectable in the plasma of normal, healthy subjects. Unlike other known inhibitors of hematopoiesis, DIP appears to directly inhibit erythropoietic differentiation, because it did not affect the proliferation of untreated FLC and it effectively blocked FLC hemoglobinization without affecting the ability of the blocked cells to proliferate. DIP blocked FLC differentiation only when added to the culture medium within 1 hour of inducing the cells with DMSO, suggesting that the protein inhibited an early, but critical, DMSO-induced cellular process. DIP appears to be at least partially responsible for the patient's anemia, and its unique activity suggests a role in the early development of some erythroleukemias.

Characterization of a novel erythropoiesis-inhibiting human protein.

We have isolated an erythropoiesis-inhibiting protein, DIP (differentiation-inhibiting protein), from the blood of a 60-year-old woman suffering from pure red cell aplasia. This protein inhibits the growth and differentiation of normal human and murine BFU-E, but not CFU-E, cells as well as dimethyl sulfoxide-induced hemoglobin synthesis by Friend murine erythroleukemia cells. It appears that DIP primarily affects differentiation rather than proliferation, because it does not inhibit the proliferation of untreated Friend erythroleukemia cells. DIP seems to function like a recently described 45-kDa autocrine differentiation-inhibiting protein factor (ADIF) which is secreted by tsAEV-transformed chicken erythroleukemia cells. Both proteins selectively block the differentiation of normal human and murine BFU-E cells as well as the differentiation (but not the proliferation) of Friend murine erythroleukemia cells. However, the human DIP is not an autocrine product of the patient's bone marrow cells, nor does it affect chicken erythroid cells.

The cytotoxicity of Pseudomonas exotoxin A, inactivated by modification of the cell-binding domain I, is restored when conjugated to an erythroid cell-specific targeting agent.

To be capable of selective killing of tumor cells, the non-selective Pseudomonas aeruginosa exotoxin A must have its cell-binding domain inactivated or removed and then be chemically linked to, or genetically fused with, a specific targeting agent. In the present study, epsilon-NH2 groups of lysine residues of the cell-binding domain of exotoxin A were extensively propionylated with N-succinimidyl-3-propionate (NSP). The NSP-treated exotoxin retained its cytocidal ADP-ribosyltransferase activity, but it could no longer bind to, and inhibit the proliferation of, Friend murine erythroleukemia cells. Cytotoxicity (i.e., the ability to inhibit proliferation) for the Friend erythroid cells was restored completely to the NSP-inactivated exotoxin by conjugating it to ADIF, an autocrine factor secreted by chicken erythroleukemia cells which selectively inhibits the differentiation of erythroid cells such as Friend erythroleukemia cells without inhibiting their proliferation.