Cloning and mRNA expression of vascular endothelial growth factor in ischemic retinas of Macaca fascicularis.

PURPOSE: To identify and isolate cDNAs for the alternatively spliced vascular endothelial growth factor (VEGF) mRNAs present in retina and to compare the relative levels of the splice variants and localization of VEGF mRNA in nonischemic and ischemic adult simian retinas. METHODS: Retinas of cynomolgous monkeys were made ischemic by laser occlusion of the main branch retinal veins. Reverse transcription-polymerase chain reaction was used to amplify the VEGF coding region of RNA from ischemic and control retinas, and amplification products were analyzed by agarose gel electrophoresis, Southern blot, and nucleotide sequencing. Analysis of VEGF mRNA expression was accomplished by in situ hybridization. RESULTS: Control and ischemic retinas produce mRNAs that correspond to the 121 and 165 amino acid diffusible isoforms of VEGF, and relatively low levels of VEGF189, the heparin-binding isoform. There was no significant difference in the levels of the alternatively spliced VEGF transcripts between control and ischemic retinas. Cloning and sequencing revealed that simVEGF cDNAs are 99% identical to human VEGFs and are predicted to encode proteins identical to their respective human homologues. In situ hybridization of nonischemic retinas revealed a low level of VEGF mRNA in retinal ganglion cells and in the inner nuclear layer. VEGF mRNA levels were elevated in ischemic retinas as early as 1 day after laser vein occlusion, when there was a small but reproducible increase in signal. The expression peaked at approximately 13 days, coincident with maximal iris neovascularization, and was significantly reduced by 28 days, when the iris vessels largely regressed. CONCLUSIONS: The elevation of simVEGF121 and VEGF165 in ischemic retinas is consistent with a role for diffusible, retina-derived angiogenic factors in the development of ocular neovascularization.

Alterations in gene expression associated with changes in the state of endothelial differentiation.

The endothelium maintains a developmental plasticity which allows rapid phenotypic change in response to extracellular signals during normal processes, such as corpus luteum formation and wound healing, and in pathologic processes, such as tumor angiogenesis. Endothelial cells (EC) in culture have been very useful for investigating various aspects of endothelial growth and behavior. In spite of documented similarities between EC in vitro and the endothelium in vivo, many characteristics of the vessel endothelium are lost when the cells are placed into culture. We have undertaken to identify differences in gene expression between differentiated vessel endothelium and dedifferentiated EC. We utilized a new technique called differential display which compares polymerase chain reaction (PCR)-amplified mRNA from two (or more) cell populations. Endothelium scraped directly from freshly obtained aortas, and demonstrated to be free of contaminants, were used as the source of differentiated RNA, whereas proliferating, primary explanted EC grown for five days in the presence of basic fibroblast growth factor (bFGF) provided a pool of 'dedifferentiated' RNA. Using differential display, we have observed numerous reproducible differences in gene expression. To confirm that the expression differences visualized by differential display represented actual differences in gene expression, we isolated vessel-specific and culture-specific cDNA tags for additional analysis. Three cDNA tags specific to vessel endothelium were cloned and sequenced, and compared to nucleotide and protein databases. Two of the clones (A1 and 2.5) displayed no significant sequence similarity, whereas a third clone (A2) is nearly identical to a human expressed sequence tag (EST) and has significant sequence similarities to a plant and Xenopus ubiquitin-like protein. Northern and/or in situ hybridization analysis of the A1 and A2 genes confirmed their restricted expression to the vessel endothelium. The expression of A1 by the endothelium in vivo is not simply a function of growth state, as cultured cells did not express A1 even when grown to postconfluence. One other cDNA fragment, selected as a culture-induced gene, was identified by sequence analysis as the bovine homologue of laminin B1, and Northern analysis confirmed that expression was induced upon culturing of EC. Use of differential display to study endothelial gene expression will allow us to investigate the molecular mechanisms that underlie initiation and maintenance of endothelial differentiation.

Regulated expression on human macrophages of endoglin, an Arg-Gly-Asp-containing surface antigen.

Endoglin is an endothelial homodimeric membrane antigen containing the tripeptide arginine-glycine-aspartic acid (RGD), which is a recognition motif for adhesion receptors of the integrin family. We have investigated the expression of endoglin by monocyte/macrophage cells from different tissue compartments and at different stages of cell differentiation. Although endoglin is absent from peripheral blood monocytes, it is expressed by in vitro differentiated monocytes as determined by flow cytometry using the endoglin-specific monoclonal antibody 44G4 and 8E11. Furthermore, Northern blot analyses revealed a correlation between the presence of endoglin mRNA and the surface expression of the antigen by in vitro differentiated monocytes. Immunostaining of frozen tissue sections with the 8E11 monoclonal antibody demonstrated the presence of endoglin not only in the endothelium of all the tissues studied, but also on the interstitial macrophages present in the red pulp of the spleen. Using as a model of macrophage differentiation monocytic cell lines treated with phorbol esters, we found that the reactivity of the 8E11 monoclonal antibody is greatly increased on U-937 and HL-60 cells during their PMA-induced differentiation. These findings clearly demonstrate for the first time the regulated expression of the putative adhesion molecule endoglin by macrophages.

Identification of distinct epitopes of endoglin, an RGD-containing glycoprotein of endothelial cells, leukemic cells, and syncytiotrophoblasts.

Endoglin is a glycoprotein expressed predominantly on human endothelial cells. It was first identified with mAb 44G4, produced against the pre-B acute lymphoblastic HOON cell line. We now report that four mAbs independently produced against human umbilical vein endothelial cells (HUVECs), chronic myelogenous leukemia in blast crisis, or U-937 pro-monocytic cells stimulated with phorbol myristate acetate also react with endoglin. High levels of reactivity of all mAbs were observed with HUVEC, while intermediate levels were seen with HOON and U-937 cells. By sequential immunoprecipitation from HUVEC and U-937 cell extracts, it was established that RMAC8, HEC-19, 8E11, and 1G2 mAbs react with the same protein as 44G4. Three distinct epitopes recognized by 44G4, RMAC8, and 1G2 mAbs were identified by competitive radioimmunoassay and flow cytometry. The HEC-19 epitope is spatially related to the 44G4 epitope, whereas the 8E11 epitope is most closely related to the 1G2 epitope. Western blot analysis showed that all antibodies react with the endoglin dimer (Mr = 170,000) purified from placenta. Immunostaining of sections of full-term placenta revealed reactivity not only with fetal vessels but also with the syncytiotrophoblast, the fetal cell layer which interfaces with maternal blood. When HUVEC monolayers were treated with the different mAbs to endoglin, prior to incubation with U-937 cells, a 5- to 10-fold stimulation of adhesion was observed. A fibronectin hexapeptide containing RGD, but not the corresponding RGE peptide, was capable of inhibiting the increased adhesion, when tested with mAb 44G4 and RMAC8. However, the same peptides had no effect on the binding of any of the five anti-endoglin mAbs to cells. Since 44G4 and RMAC8 recognize two distinct epitopes of endoglin, and since all five mAbs stimulated adhesion, the results suggest that a signal has been triggered through endoglin on HUVECs. Endoglin might be implicated either directly, by binding to a specific integrin-like ligand, or indirectly, by regulating the level of adhesion between certain integrins and their receptors.

Endoglin is expressed on a subpopulation of immature erythroid cells of normal human bone marrow.

A monoclonal antibody (1G2) was raised by immunization of a Balb/c mouse with the leukemic blasts from a patient suffering from chronic myelogenous leukemia blast crisis (CML-BC). Sequential immunoprecipitation of the protein from human umbilical vein endothelial cells with 1G2 and the endoglin-specific monoclonal antibody 44G4 indicated that both antibodies react with the same molecule, a homodimer of molecular mass 180,000. This protein was first identified on acute lymphoblastic leukemia and was shown to be primarily associated with endothelial cells. In addition, 1G2 and 44G4 identified the same subpopulation of human bone marrow mononuclear cells (BMMNC), as established by two colour immunofluorescence analysis. By cell sorting and colony assays it could be demonstrated that endoglin is not expressed on hemopoietic precursor cells (CFU-G, CFU-GM, CFU-GEMM, BFU-E). May-Grünwald-Giemsa staining of sorted BMMNC revealed that 1G2 recognized immature proerythroblasts and double-fluorescence analysis showed that endoglin is present on a subset of glycophorin A-positive BMMNC. 1G2 was not reactive on bone marrow B-cells (CD19, CD20), T-cells (CD3, CD7), natural killer cells (CD56), myeloid cells (CD13, CD14, CD15, CD33), and on CD34-positive cells. Endoglin contains an arginine-glycine-aspartic acid sequence, a feature generally associated with extracellular matrix proteins which interact with integrins. It is suggested that proerythroblasts may utilize endoglin to interact with integrins in cell-cell adhesion events in the stromal or hemopoietic compartment of the bone marrow.

Primary structure of endoglin, an RGD-containing glycoprotein of human endothelial cells.

Endoglin is a major glycoprotein of human vascular endothelium. As observed with monoclonal antibody 44G4, the distribution of endoglin is restricted to endothelial cells in all tissues except bone marrow. cDNA clones were isolated from an endothelial cell lambda gt11 cDNA library using a rabbit antibody prepared against endoglin purified from placenta. Eleven antibody-positive and cross-hybridizing clones were obtained; reactivity with endothelial cell 3.4-kilobase mRNA transcript was observed. The N-terminal sequence of placental endoglin was determined and found within the deduced protein sequence, thus confirming the identity of the cDNA and revealing a partial signal peptide. Endoglin is a type I integral membrane protein of Mr = 68,051 with an extracellular region of 561 amino acids, a hydrophobic transmembrane domain, and a 47-residue cytoplasmic tail. There are four potential N-linked glycosylation sites in the N-terminal domain and a probable O-glycan domain rich in Ser and Thr residues proximal to the membrane-spanning domain. Data base searches revealed that endoglin is a novel protein. The sequence contains an RGD tripeptide (374-376), the first identified on a surface protein of endothelium. The presence of RGD, a key recognition structure in cellular adhesion, suggests a critical role for endoglin in the binding of endothelial cells to integrins and/or other RGD receptors.

Quantitative Phenotyping of Childhood Leukemia Identifies Variable and Invariable Cell Surface Antigens.

Cells obtained from 75 cases of childhood leukemia were subjected to flow cytometry analysis to estimate the density of several cell surface antigens and derive a quantitative immunological phenotype. Sixty-five cases of acute lymphoblastic leukemia (ALL) including 10 T-ALL, 6 non-T ALL designated groups I and II (HLA-DRCALLA), 48 non-T ALL termed groups III and IV (HLA-DRCALLA) and one B-ALL were studied; 10 cases of acute myeloblastic leukemia (AML) were also analysed. The estimation of the relative fluorescence index (RFI) on leukemic blasts led to the derivation of mean values for each marker in the leukemia subgroups. We have quantitated the levels of the antigens generally used in the classification of these leukemias (CALLA, CD5, CD20, CD13, HLA-DR and CD19) and of other cell surface antigens associated with leukemic cells. For example, CALLA (CD10) level was high (mean RFI value of 26.4) on the leukemic cells of non-T ALL groups III and IV. The CD5 antigen was present on T-ALL, as expected, with an RFI value of 4.5; however, low levels were observed on the more immature non-T ALL of groups I and II (RFI = 2.3 on only 27% of blast cells). The quantitative analysis of the cell surface antigens associated with non-T ALL has revealed molecules such as CALLA, HLA-DR, CD9 and CD44 present at high and variable levels and others such as CD19, CD38, 44G4, 44D7, 44H9 and 44H6 generally of lower intensity, less variable from one patient to another, and with similar mean levels of expression in the different subgroups. These invariable antigens are not altered by the lineage or stage of differentiation of the leukemic cells. The variable antigens could be correlated with the functional and/or differentiation status of the cells and could also be modified by the alterations of regulatory processes associated with malignancy. The quantitation of multiple leukemia-associated antigens, whose structure and function are becoming rapidly established, should help in elucidating the function of these molecules in leukemogenesis and/or disease progression.

Identification of a human endothelial cell antigen with monoclonal antibody 44G4 produced against a pre-B leukemic cell line.

Staining of a variety of human tissue sections (lymph node, tonsil, spleen, thymus, kidney, lung, and liver) by the indirect immunoperoxidase method indicated that mAb 44G4, produced against a human pre-B leukemic cell line, was strongly reactive with vascular endothelium. All other cell types observed in these tissues were unreactive. Immunofluorescence staining of endothelial cells isolated from umbilical cord vein and grown in culture confirmed that mAb 44G4 recognized a surface membrane component of vascular endothelium. Granulocytes, monocytes, B and T lymphocytes, and T lymphocytes cultured in the presence of PHA for 72 h did not express the 44G4 Ag. mAb 44G4 reacted weakly with leukemic cells from 28 of 41 patients with non-T cell acute lymphocytic leukemia and 4 of 7 patients with acute myelocytic leukemia, whereas 8 of 10 cases of T cell acute lymphocytic leukemia were negative. Moderate reactivity with leukemic cell lines of pre-B and myelomonocytic origin was also observed. The level of 44G4 Ag on umbilical endothelial cells was three to five times that of leukemic cell lines and 25 times the average levels observed on leukemic cells isolated from patients. Immunoprecipitation of lysates prepared from surface-iodinated endothelial cells and the immunizing pre-B leukemic cell line revealed that the 44G4 Ag from both cell types was composed of two subunits of apparent m.w. 95,000 linked by disulfide bond(s). Comparison of the cellular localization and subunit structure of 44G4 to that of known Ag suggests that it represents a previously undescribed marker of endothelial cells.

Biochemical characterization of the 44G4 antigen from the HOON pre-B leukemic cell line.

The 44G4 antigen is expressed in high amounts on human endothelial cells and at low levels on leukemic cells of pre-B and myelomonocytic origin. Its level of expression on the pre-B leukemic HOON cell line used for derivation of the corresponding mAb is intermediate but sufficient to permit the purification of the Ag. The molecule isolated by immunoaffinity from HOON is a glycoprotein since it bound to Ricinus communis agglutinin, wheat germ agglutinin, and peanut agglutinin lectins. The Ag was purified 2400-fold from a soluble taurocholate extract of HOON cells by affinity to wheat germ agglutinin-agarose and 44G4-IgG-Sepharose. The purified glycoprotein is likely a homodimer as it migrated on SDS-PAGE with an apparent m.w. of 170,000, nonreduced, and 95,000, reduced. Removal of N-linked oligosaccharides by endoglycosidase F led to a decrease in m.w. of 25,000; if neuraminidase and O-glycanase were also present, the total decrease in m.w. was 33,000 suggesting a polypeptide chain of 62,000 and 8,000 in O-linked substitutions. The glycoprotein digested with N-glycanase, neuraminidase, or O-glycanase could still be immunoprecipitated with the 44G4 mAb indicating that the antigenic epitope resides in the polypeptide. By Western blot analysis, the dissociated but nonreduced protein was reactive with 44G4, whereas the reduced and alkylated protein was not. Therefore, the epitope is dependent on the presence of intact disulfide bond(s). Sequential immunoprecipitation with OKT9 and 44G4 antibodies indicated that these epitopes are present on two distinct molecules and that 44G4 is distinct from the transferrin receptor despite a similar subunit structure.

Effect of chronic pentobarbital treatment on the development of cross-tolerance to ethanol and barbital.

Recently, we reported that a chronic regimen of ethanol by intubation, which produced clear tolerance to ethanol-induced hypothermia, ataxia and sleep, produced only a marginal degree of cross-tolerance to these effects of pentobarbital. The present experiments were designed to test the reverse process by examining cross-tolerance to pentobarbital after chronic pretreatment with ethanol, chronic pentobarbital treatment by gavage conferred clear cross-tolerance to both barbital- and ethanol-induced hypothermia, ataxia and sleep. In a separate experiment, cross-tolerance to barbital- and ethanol-induced hypothermia and ataxia was demonstrated over a wide range of test doses. Determination of ethanol blood levels as well as a complete time course of absorption, distribution and elimination of ethanol suggested that pharmacokinetic alterations may play a role in the development of cross-tolerance to ethanol in pentobarbital-treated subjects. The asymmetry of cross-tolerance raises the possibility that pentobarbital and ethanol invoke tolerance by mechanisms that are not wholly identical. This possibility requires further exploration. Conceivably the actions of ethanol which mediate the measured effects form a subset of a larger range of pentobarbital actions that could provide a stronger stimulus to tolerance development.

Tolerance to ethanol and cross-tolerance to pentobarbital and barbital.

A chronic regimen of ethanol by intubation, which produced clear tolerance to ethanol-induced hypothermia, ataxia and narcosis, produced only a marginal degree of cross-tolerance to these effects of pentobarbital. The lack of appreciable cross-tolerance to pentobarbital-induced hypothermia and ataxia was also observed over a wide range of test doses. However, cross-tolerance to barbital was observed after chronic treatment with ethanol. Increased rate of drug biotransformation did not contribute significantly to the observed tolerance and cross-tolerance. The difference in the extent of cross-tolerance between ethanol and the two barbiturates is consistent with the hypothesis that there is a degree of specificity in the sites of action of ethanol and other sedative-hypnotic drugs.

Differential localization within human kidney of five membrane proteins expressed on acute lymphoblastic leukemia cells.

The reactivity of a panel of monoclonal antibodies (MAb) produced against non-T, non-B acute lymphoblastic leukemia cells was investigated by immunoperoxidase staining of sections of normal human kidney. The antigens of kidney reactive with the MAb were isolated by immunoaffinity chromatography and were purified further by immunoprecipitation. Two MAb, 44D7 and 44H9, reacted with determinants found exclusively on the basolateral membranes of proximal convoluted tubules. The 44D7 antigen isolated from kidney was biochemically similar to that isolated from leukemic cells. It was resolved as a multimeric complex with an apparent m.w. of 120,000 when analyzed by SDS-PAGE under nonreducing conditions. The 44H9 antigen has not yet been purified from kidney. MAb 50B4 reacted with components of the interstitium and with the mesangium of glomeruli. It immunoprecipitated a polypeptide chain of apparent m.w. 85,000, similar to that of the 50B4 antigen isolated from leukemic cells. MAb 44G4 also reacted with the mesangium of glomeruli and with the interstitium of the kidney. However, the endothelium of glomerular capillaries and of interstitial blood vessels has also reacted with MAb 44G4. The kidney antigen recognized by MAb 44G4 was characterized as a major polypeptide band, 95,000 m.w. (reduced) and 125,000 m.w. (nonreduced), a subunit structure analogous to the 44G4 antigen isolated from leukemic cells. MAb 44E3 reacted with all cellular elements of glomeruli, tubules, blood vessels, and interstitium. Two polypeptide chains of apparent m.w. 94,000 and 90,000 were immunoprecipitated from kidney by MAb 44E3, while a single polypeptide chain of 94,000 m.w. was precipitated from leukemic cells. Our results describe five new antigens with distinctive cellular distributions within kidney.