Specific binding of IGF-binding protein (IGFBP)-3 was shown to be present in the isolated, beating rat heart. The uptake of perfused (125)I-labeled IGF-I in the beating heart was decreased to 9% by blocking IGF-I binding sites with the IGF-I analog Long R(3) (LR(3)) IGF-I. When LR(3) was perfused with complexes of (125)I-IGF-I. IGFBP-3, uptake of (125)I-IGF-I was decreased to 41%, which was significantly greater than LR(3) and (125)I-IGF-I (41 vs. 9%). These data suggest that both microvessel IGF-I and IGFBP-3 binding sites contribute to the transport of IGF-I in the perfused rat heart. This also suggests a novel and plausible mechanism whereby circulating IGFs reach sites of IGF bioactivity.

Insulin-like growth factor binding protein 6 (IGFBP-6) is a high-affinity IGFBP with substantially greater affinity for insulin-like growth factor-II (IGF-II) than IGF-I. IGFBP-6(3) is a chimera which has a 20 amino acidC -terminal portion of IGFBP-6 switched with the homologous area of IGFBP-3, P3. Unlike IGFBP-4(3), in which the P3 region was exchanged for the homologous region of IGFBP-4 (P4), IGFBP-6(3) does not bind to endothelial cells. Double mutations were made with the P3 region exchanged as well as a second area differing from IGFBP-3 to form IGFBP-6(3)A and IGFBP-6(3)B, by replacing this area with the homologous region of IGFBP-3. Neither [(125)I]IGFBP-6(3)A nor IGFBP-6(3)B specifically bound to endothelial cells. However, each double mutant competed for [(125)I]IGFBP-3 binding to cultured cells. In the perfused heart, transendothelial transport of IGFBP-6 and IGFBP-6(3) was only 25% of similar transendothelial transport of perfused IGFBP-3. We conclude that chimeras of IGFBP-6 and IGFBP-3(6) clearly differ from IGFBP-4(3) in their ability to bind specifically to endothelial cells and in their capacity to undergo transendothelial transportation in the perfused heart.

Insulin-like growth factor-binding protein (IGFBP)-3 contains a highly basic COOH-terminal heparin-binding region, the P3 region, which is thought to be important in the binding of IGFBP-3 to endothelial cells. IGFBP-3 and IGFBP-4, and their chimeras IGFBP-3(4) and IGFBP-4(3), were treated with plasmin and with thrombin, proteases known to cleave IGFBP-3. IGFBP-3 was highly susceptible to plasmin, whereas IGFBP-4 was less so. Substitution of the P3 region for the P4 region in IGFBP-4 (IGFBP-4(3)) increased the ability of the protease to digest IGFBP-4(3); substitution of the P4 region for the P3 region in IGFBP-3 (IGFBP-3(4)) decreased the digestion of IGFBP-3(4). When 125I-labeled IGFBP-3 or 125I-IGFBP-4(3) was first bound to vascular endothelial cells, subsequent proteolysis by either plasmin or thrombin was substantially inhibited. Proteolysis of 125I-IGFBP-3(4) was not inhibited in the presence of endothelial cells. The P3 peptide was cleaved by plasmin but not by thrombin. We conclude that the P3 region is central to proteolysis of IGFBP-3 by plasmin and thrombin, processes which were inhibited by association of IGFBP-3 with endothelial cells.

IGF binding proteins-3 and -4, whether given in the perfused rat heart or given iv in the intact animal, cross the microvascular endothelium of the heart and distribute in subendothelial tissues. IGF binding protein-3, like IGF-I/II, localizes in cardiac muscle, with lesser concentrations in CT elements. In contrast, IGFBP-4 preferentially localizes in CT. In this study, chimeric IGF binding proteins were prepared in which a basic 20-amino-acid C-terminal region of IGF binding protein-3 was switched with the homologous region of IGF binding protein-4, and vice-versa, to create IGF binding protein-3(4) and IGF binding protein-4(3). Perfused IGF binding protein-3(4) behaved like IGF binding protein-4, localizing in connective tissue elements, whereas IGF binding protein-4(3) now localized in cardiac muscle at concentrations identical to perfused IGF binding protein-3. To determine whether these small mutations altered the affinity of the chimera for cells, the ability of (125)I-IGF binding protein-3(4) and (125)I-IGF binding protein-4(3) to bind to microvascular endothelial cells was determined and compared with IGF binding protein-3. IGF binding protein-3(4) retained 15% of the binding capacity of IGF binding protein-3, whereas IGF binding protein-4(3) bound to microvessel endothelial cells with higher affinity and greater total binding than that of IGF binding protein-3. We conclude that small changes in the C-terminal basic domain of IGF binding protein-3 and the corresponding region of IGF binding protein-4 can alter their affinity for cultured cells and influence their tissue distribution in the rat heart.

18 amino acid peptides from the C-terminal region of IGFBP-3, -5 (P3, P5), increased the incorporation of(35)SO(4)into proteoglycans in endothelial cells with greater stimulation in large vessel than microvessel cells. The homologous region of IGFBP-6 (P6) also stimulated sulfate uptake, but less potently than P3 and P5. P6 variants were synthesized with one or two amino acids changed to the basic amino acid in the equivalent position of P3. The P6 variants with one additional basic amino acid behaved similarly to P6. The P6 mutant with two altered amino acids was equipotent to P3. P3F, a scrambled version of P3 was less effective than P3. P3, P5, P6, P3F and all P6 variants all stimulated glucose uptake, which occurred only in microvessel cells. P1, P2, P4, and equimolar intact IGFBP-3 stimulated neither glucose uptake nor sulfate incorporation. Thus, C-terminal basic portions of IGFBP-3, -5 and -6 alter two specific functions of endothelial cells with sufficient differences to suggest mediation by distinct mechanisms.

Media from large vessel endothelial cells (pulmonary artery, aorta) contained intact connective tissue growth factor (CTGF) and a dominant 19-kDa band. N-terminal analysis of the 19-kDa band showed sequence corresponding to CTGF amino acid 181-190, suggesting that the 19-kDa band represented a proteolytic fragment of CTGF. Intact CTGF was increased by cAMP but not by transforming growth factor-beta (TGFbeta). CTGF messenger RNA (mRNA) was not changed by cAMP nor TGFbeta. In two microvessel endothelial cells, mRNA was found at low levels by PCR and Northern analysis, but no CTGF protein was seen on Western analysis. In the microvessel cells, TGFbeta increased and cAMP did not change CTGF mRNA levels, with neither TGFbeta nor cAMP increasing CTGF protein. The discordance between protein and mRNA levels in large vessel and microvessel endothelial cells was mostly explained by the effects of cAMP and TGFbeta on media proteolytic activity; in large vessel cells, cAMP inhibited degradation of CTGF, whereas in microvessel cells, TGFbeta and cAMP stimulated proteolytic activity against CTGF. We conclude that in large vessel endothelial cells, cAMP increased intact CTGF protein by inhibiting degradation of CTGF, whereas TGFbeta stimulated neither CTGF mRNA nor protein; in microvessel cells, TGFbeta increased CTGF mRNA, while both TGFbeta and cAMP stimulated CTGF degradation.

Insulin-like growth factor-binding protein-3 (IGFBP-3) was digested with plasmin, and the proteolytic fragments were isolated by HPLC and tested for bioactivity as measured by stimulation of glucose uptake in microvessel endothelial cells. Two of the pooled fractions of the digest stimulated glucose uptake. The major bioactive pool, at an estimated protein concentration <50 ng/ml, stimulated glucose uptake to 150% of control with greater stimulation and 220% of control at ∼250 ng/ml. Two fragments were present in the bioactive fraction, the dominant one migrating at ∼20,000 and the other at ∼8,000. Both fragments bound125I-labeled insulin-like growth factor and [3H]heparin. NH2-terminal amino acid analysis of the bioactive peak yielded two sequences. One, representing the majority of the material, had an NH2-terminal sequence identical to IGFBP-3; the second fragment began at amino acid 202 of IGFBP-3. In contrast to the bioactive fragments, intact IGFBP-3, at concentrations up to 130 μg/ml, had no bioactivity. These findings demonstrate that IGFBP-3 can be degraded into fragments that have potent bioactivities that are not present in the intact IGFBP-3 molecule.

Insulin-like growth factor binding protein (IGFBP) secretory profiles were determined for vascular smooth muscle cells (VSMC) derived from bovine aorta and human aorta, pulmonary artery, and coronary artery. The bovine cells produced IGFBP-4, IGFBP-3, and an IGFBP-3 protease. IGF-I stimulated messenger RNA (mRNA) and media levels of IGFBP-3. The human cells produced IGFBP-3, IGFBP-4, and IGFBP-3 and IGFBP-4 proteases. The three human cells also produced a 30K IGFBP, shown to be IGFBP-6, based on increased affinity for IGF-II vs. IGF-I, size decrease when treated with O-glycanase, but not N-glycanase, reactivity with IGFBP-6 antiserum, presence of a 1.3-kilobase pair mRNA that hybridized to IGFBP-6 specific complementary DNA, and N-terminal amino acid sequence corresponding to IGFBP-6. In the human cells, IGF-I increased media levels of IGFBP-3 through stimulation of IGFBP-3 mRNA and dissociation of cell bound IGFBP-3, and decreased IGFBP-4 via potentiation of IGFBP-4 proteolysis. Neither the bovine nor the human aorta VSMC produced sufficient IGFBP-2 or IGFBP-2 mRNA to be detected by ligand blot and Northern analysis, as previously reported for porcine and rat aorta smooth muscle cells. The variable expression of IGFBPs and IGFBP proteases by VSMC are likely to contribute to differential vascular reactivity to the IGFs in larger arterial blood vessels.

IGFBP-3 contains a carboxyterminal basic region which, when present as an isolated 18 amino acid peptide (P3), binds heparin, associates with cultured endothelial cells and stimulates glucose uptake. The P3 molecule has now been modified relative to charge, amino acid sequence and size to determine structure-function relationships relative to four properties of P3: affinity for heparin; inhibition of IGFBP-3 binding; stimulation of glucose uptake; and displacement of bFGF from the extracellular matrix of endothelial cells. Results indicate: (1) the presence or absence of heparin binding was concordant with the presence/absence of the other three properties; (2) the number of basic amino acids was an important, if not limiting, factor for each property; (3) the order of potency of the basic amino acids was arginine = lysine > > histidine; (4) the unrelated, basic protein, protamine, mimics all properties of P3; and (5) the putative consensus heparin-binding sequence of P3 was not essential for any of the P3 activities.

Insulin-like growth factor binding protein (IGFBP)-3 was exposed to plasmin, thrombin, and pregnancy serum, substances normally present at the endothelial surface in enriched concentrations. The NH2-termini of the proteolytic fragments were sequenced, and their ability to bind insulin-like growth factor (IGF) and heparin was assessed by ligand blotting. Plasmin generated at least five fragments, three beginning at the NH2-terminus of IGFBP-3 and two with NH2-termini corresponding to middle portions of IGFBP-3. The dominant fragment bound both IGF and heparin while NH2-terminal fragments bound only IGF. Thrombin generated three and serum five easily identified fragments; the dominant fragments, beginning at midportions of IGFBP-3, retained IGF and heparin affinity, whereas the remaining fragments had differential affinities for IGF and heparin. We suggest that such fragments, when generated at the endothelia surface, have the potential to alter regional vascular concentrations of IGF and thus influence both IGF and endothelial function.

IGFBP-3 and IGFBP-5, but not the other 4 IGF binding proteins, specifically bound to endothelial cell (EC) monolayers. Charged compounds, such as heparin and heparan sulfate, competed for this binding. Of the 6 IGFBPs, IGFBP-3 and IGFBP-5 had the greatest heparin affinity. Peptides of 18 amino acids were synthesized, corresponding to a common basic region of IGFBP-3 (P3), IGFBP-5 and IGFBP-6 (P6) which contained a heparin binding sequence. P3 and P6 inhibited IGFBP-3 and -5 binding to endothelial cell monolayers and the peptides bound directly to EC extracellular matrix. This suggested that the C-terminal basic segment of IGFBP-3/-5 is important for the association of the binding protein with the EC monolayer.

Insulin-like growth factor binding protein 4 (IGFBP4) was purified to homogeneity from conditioned media of bovine pulmonary artery endothelial cells and shown to have the N-terminal amino acid sequence DEAIHCPPCS, a sequence unique to IGFBP4. The IGFBP4 was separated into predominantly glycosylated and nonglycosylated fractions, with each fraction separately perfused through isolated, beating rat hearts. Both forms of IGFBP4 crossed the capillary boundary of the heart and distributed primarily in subendothelial connective tissue components with a connective tissue/cardiac muscle distribution ratio of 20:1 for the glycosylated fraction and 27:1 for the nonglycosylated fraction. Perfused IGFBP1, 2, 3, and IGF-I also crossed the capillary boundary but in contrast to IGFBP4, preferentially localized in cardiac muscle with a connective tissue/muscle ratio of approximately 1:3. We conclude that the connective tissue distribution previously reported for IGFBPs in conditioned media of pulmonary artery endothelial cells is due to IGFBP4.

Cultured endothelial cells have been shown to produce insulin-like growth factor-binding proteins (IGFBPs); however, the identity of these BPs has not been defined. We now demonstrate that cultured bovine endothelial cells produce IGFBP2, IGFBP3, and IGFBP4 and have mRNA specific for IGFBP2, -3, -4, -5 and -6. DNA probes for bovine IGFBP2-6 were obtained by polymerase chain reaction (PCR) amplification of cDNA from bovine large vessel pulmonary artery and aortic endothelial cells as well as omental and periaortic fat microvessel cells, using oligonucleotide primers whose sequences were based on the reported cDNA sequences of IGFBP2-6. The PCR-derived probes were labeled with 32P and used for Northern blot analysis of RNAs obtained from the four bovine endothelial cell types. Transcripts corresponding to IGFBP2-6 were found in RNA from large vessel endothelial cells (bovine pulmonary artery and bovine aorta) and microvessel cells (periaortic and omental fat). The PCR-derived probe for IGFBP4 was used to screen a bovine pulmonary artery cDNA library for a full-length bovine IGFBP4 cDNA clone. One positive clone, containing a single EcoRI insert of approximately 2.0 kilobases, was selected for further characterization by DNA sequence analysis. This clone contained an open reading frame encoding a 258-amino acid protein that was 97% identical to human IGFBP4, 268 basepairs of 5'-untranslated region, and a longer 1044 basepairs of 3'-untranslated region. IGFBP4 protein was purified from bovine pulmonary artery-conditioned medium, shown to have N-terminal amino acid sequence DEAIHCPPCSEEKLARCR (identical to human IGFBP4) and to be secreted in glycosylated and nonglycosylated forms. Immunoblots further demonstrated that microvessel cells, at early passage, secrete predominantly IGFBP2 and IGFBP3, while large vessel cells, at early and late passages, secrete IGFBP3 and IGFBP4. Thus, cultured bovine endothelial cells synthesize and secrete IGFBP2, IGFBP3, and IGFBP4 and have mRNA encoding IGFBP2-6. The production of specific IGFBPs by endothelial cells raises the interesting possibility that the vascular endothelium contributes to circulating and tissue levels of specific IGFBPs in vivo.

Insulin-like growth factor (IGF) binding proteins (IGFBP) were purified from conditioned media of cultured bovine endothelial cells (ECBP) and from human amniotic fluid (IGFBP-1), and then labeled by radioiodination. 125I-ECBP and 125I-IGFBP-1 were perfused through isolated beating rat hearts for 1 and 5 min, and the hearts fixed and analyzed for 125I-BP content and distribution. One to 4% of the perfused 125I-ECBP and 125I-IGFBP-1 crossed the capillary boundary. The ECBPs predominantly localized as intact 125I-BP in connective tissue elements of the heart with less 125I-BP in cardiac muscle. The ratio of 125I-ECBP in connective tissue: muscle (normalized to percent vol of these compartments) was greater than or equal to 10:1. In contrast, the IGFBP-1 had a greater affinity for cardiac muscle with ratios of 125I-IGFBP-1 in connective tissue:muscle of approximately 1:2. When 125I-IGF-I, in the absence of any BPs, was perfused through the hearts approximately 3-5% left the microcirculation and was found in subendothelial tissues. 125I-IGF-I localized primarily to cardiac muscle with a distribution of connective tissue:cardiac muscle of about 1:3. The findings in the isolated perfused heart were confirmed in intact animals. After 125I-IGFBP-1 was injected into anesthetized rats and allowed to circulate for 5 min, substantial radioactivity was associated with the heart. As in the isolated heart, the IGFBP-1 preferentially localized to cardiac muscle with a connective tissue:cardiac muscle ratio of 1:3. We conclude that IGFBPs produced by endothelial cells and the IGFBP-1 contained in amniotic fluid can cross the capillary boundaries of the rat heart, and that the ECBPs preferentially localize in connective tissue elements of the myocardium, whereas IGFBP-1 predominantly localizes in cardiac muscle.

Perfused endothelial cell IGF binding proteins (ECBP) have been previously demonstrated to leave the microcirculation of the rat heart and distribute primarily in connective tissue elements of the heart. In the present study, ECBP have been crosslinked to IGF-I and the biologically inactive [1-27,gly4,38-70]-hlGF-I, an analog of IGF-I lacking the type I IGF receptor domain. The crosslinked ECBPs were perfused through the isolated rat heart and their tissue distributions determined. Both [ECBP-Analog] and [ECBP-IGF-I] left the microcirculation of the heart. [ECBP-Analog] preferentially localized in connective tissue elements with a muscle:connective tissue ratio of approximately 1:6, similar to the tissue distribution of perfused ECBP. In contrast, the [ECBP-IGF-I] complexes localized in cardiac muscle with a muscle to connective tissue ratio of approximately 3:1, virtually identical to the tissue distribution of IGF-I when the IGF-I is perfused through the heart in the absence of any binding proteins. We conclude that 1) ECBP in the presence of IGF will cross capillary boundaries and 2) the tissue distribution of [ECBP-IGF-I] is dictated by the IGF-I molecule.

Insulin-like growth factor binding-proteins 1 and 2 (IGFBP-1, IGFBP-2) and endothelial cell IGF binding proteins (ECBP) were individually perfused through isolated beating rat hearts in the absence and presence of insulin. Insulin caused an increased movement of IGFBP-1 from the vascular space to tissues of the heart. Subendothelial content of IGFBP-1 was 110%, 126% (p less than .01) and 132% (p less than 0.05) of control hearts when perfused with 1, 10 and 100 ng/ml insulin, respectively. . In contrast, insulin treatment was associated with a decrease in ECBP content in cardiac tissue, being 83%, 62% (p less than 0.005) and 73% (p less than 0.05) of control when perfused with 1, 10 and 100 ng/ml insulin. The efflux of IGFBP-2 from the intravascular space was unaffected by insulin. The subendothelial tissue distribution of the transported binding proteins was not changed by insulin perfusion, with IGFBP-1 and IGFBP-2 localizing predominantly in cardiac muscle and ECBP having greater affinity for connective tissue elements. We conclude that in the perfused rat heart, insulin can differentially alter transcapillary movement of IGFBP-1, IGFBP-2 and endothelial cell IGF-binding proteins. Such insulin-facilitated changes could potentially mediate nutrient-dependent transport of IGF-I and IGF-II to peripheral tissues.

Conditioned medium from cultured vascular endothelial cells contains material capable of stimulating acute metabolic processes in endothelial cells. The bioactivity of the conditioned medium is not caused by the copurification of known growth factors produced by the cells, in particular platelet-derived growth factor, basic fibroblast growth factor, or insulin-like growth factor (IGF)-I/II. We now demonstrate that the bioactivity is directly due to an IGF-binding protein(s) (ECBP) and, further, that the bioactive domain of the binding protein differs from the IGF-binding domain. Binding proteins (BPs) from cultured pulmonary artery endothelial cells were purified by sequential passage over sizing, multiplication-stimulating activity affinity, and hydrophobic columns. BP fractions were separated into those with and those without biological activity. The bioactive binding protein(s) was cross-linked with disuccinimidyl suberate to IGF-I or the recombinant IGF analog [1-27,Gly4,38-70]IGF-I (Analog). The IGF-I Analog, by itself, had minimal interaction with the type I IGF receptor in cultured microvessel endothelial cells and no intrinsic bioactivity, but did bind with high affinity to ECBP. All free BP and free IGF-I/Analog were removed from the cross-linked mixture by passage over gel filtration and IGF affinity columns. The cross-linked BP-IGF-I complex did not bind to the type I receptor of cultured endothelial cells, but did stimulate glucose and alpha-aminoisobutyric acid uptake in endothelial cells (approximately 2-fold increase); the magnitude of the response was nearly equal to the effect of ECBP or IGF-I alone. The BP-Analog complex also stimulated glucose and alpha-aminoisobutyric acid uptake, with the magnitude of the response approaching the effect of ECBP alone. The BP-Analog complex also did not react with type I IGF receptors on the cultured endothelial cells. We conclude 1) IGF-BP produced by endothelial cells possess intrinsic biological activity; 2) bioactivity of the BP(s) is retained when the IGF-binding domain of the BP is occupied by IGF-I or an inactive IGF-I analog; and 3) IGF-I bound to the bioactive BP does not react with its receptor and possesses minimal, if any, bioactivity in vitro.

The effects of platelet-derived growth factor (PDGF) on thymidine incorporation into DNA and glucose and neutral amino acid uptake were studied in endothelial cells cultured from macrovessels (bovine aorta and pulmonary artery) and microvessels (bovine fat and mouse brain). Similar to previous studies, PDGF did not bind to macrovessel cells, nor did it influence their metabolic function. In contrast, PDGF bound specifically to the two types of microvessel cell culture and in these cells also stimulated the uptake of glucose and neutral amino acids as well as the incorporation of thymidine into DNA. Stimulatory effects of PDGF occurred at concentrations of 2 ng/ml, with maximal stimulation up to 5-fold of the control value for amino isobutyric acid and glucose uptake and up to 8- to 10-fold for thymidine incorporation. The maximal effects of PDGF were additive to those of insulin-like growth factor, I, a known stimulator of all three metabolic processes in microvessel endothelial cells. The binding of PDGF to the endothelial cells was, in general, equivalent to PDGF binding to human foreskin fibroblasts, both in the magnitude of tracer binding and in the affinity of binding. Similar effects were found with recombinant and platelet-derived PDGF. We conclude that these two cultured microvessel endothelial cells not only produce PDGF-like material, but are capable of binding and responding to PDGF.