OBJECTIVE:Increased levels of the soluble adhesion molecule sE-selectin have been reported in normoglycaemic women with previous gestational diabetes but not in other groups at increased risk of future type 2 diabetes. To explore the basis for these discrepant findings, we studied the relationship between sE-selectin and glucose regulation in a large group of women with previous gestational diabetes. DESIGN:Comparison of sE-selectin levels between a study cohort ascertained on the basis of recent gestational diabetes and suitable control subjects. PATIENTS:One hundred and forty women with recent gestational diabetes (104 European, 20 South Asian and 16 Afro-Caribbean) and 125 normoglycaemic control women (90 European, 19 South Asian and 16 Afro-Caribbean). MEASUREMENTS:sE-selectin, fasting lipids, insulin and current glucose regulation status. RESULTS:There was no overall difference in sE-selectin levels between women with a history of gestational diabetes and control women among the 3 ethnic groups. European post-GDM women with abnormal glucose regulation postpartum (n = 30) had higher sE-selectin than control women (67 (54-91) ng/ml vs. 57 (43-75) ng/ml, P = 0.049). There was no difference in sE-selectin between normoglycaemic European women with a history of gestational diabetes (n = 74) and control women, even though the former displayed metabolic abnormalities predictive of diabetes. In those European post-GDM women with normal glucose regulation postpartum, sE-selectin levels were negatively correlated with time since delivery (r = -0.25, P = 0.04), suggesting that the previously described elevation of sE-selectin following GDM pregnancies slowly resolves postpartum. There was no correlation between sE-selectin levels and features of insulin resistance. CONCLUSIONS:In contrast to previous findings, this larger study does not support a role for sE-selectin as a marker, independent of prevailing glucose levels, of early metabolic abnormalities or future diabetes risk in women with previous gestational diabetes.

OBJECTIVE Women with a history of gestational diabetes mellitus (GDM) and women with polycystic ovary syndrome (PCOS) both demonstrate abnormalities in insulin action and secretion, and both are at increased risk of developing type 2 diabetes. To determine whether these similarities reflect a common pathophysiological basis, we examined the prevalence of ultrasound-based polycystic ovarian morphology in a large multiethnic group of women with a history of GDM and a group of women who had normal glucose tolerance during pregnancy.PATIENTS AND DESIGN We studied 91 women with previous GDM (48 European, 20 South Asian, 10 Afro-Caribbean and 13 of other or mixed ethnicity) and 73 normoglycaemic control women (56 European, one South Asian, 14 Afro-Caribbean and two of other or mixed ethnicity), a median (interquartile range) of 20 (11-36) and 29 (17-49) months postpartum, respectively. A detailed history was taken, and the prevalence of PCO morphology on ultrasound scan was assessed. Fasting lipids, insulin, glucose status, gonadotrophins and testosterone were measured. Estimates of beta-cell function (%B) and insulin sensitivity (%S) were derived using the HOMA algorithm.RESULTS Women with previous GDM had higher fasting glucose (5.4 (4.8-6.0) vs. 4.7 (4.4-5.0) mmol/l, P < 0.0001) and features reminiscent of syndrome X: higher BMI (26.4 (22.8-31.4) vs. 23.8 (21.0-27.5) kg/m(2), P = 0.002), waist/hip ratio (0.82 (0.79-0.88) vs. 0.77 (0.73-0.81), P < 0.0001), fasting insulin (165 (68-299) vs. 54 (24-156) pmol/l, P < 0.0001), triglycerides (1.1 (0.8-1.6) vs. 0.8 (0.6-1.1) mmol/l, P < 0.0001) and lower insulin sensitivity (%S) (27 (16-62) vs. 86 (34-139)%, P < 0.0001) compared to control women. The prevalence of PCO was higher in the previous GDM group than in the control subjects (47/91 (52%) vs. 20/73 (27%), chi(2) = 9.86, P = 0.002 overall, odds ratio 2.7, P = 0.007 by logistic regression allowing for ethnicity). There was no difference in any metabolic parameter between the post-GDM PCO group and the post-GDM normal ovaries group, but irregular cycles were more prevalent in the PCO group (22/47 (47%) vs. 9/44 (21%), chi(2) = 7.03, P = 0.008).CONCLUSIONS We found a higher prevalence of polycystic ovarian morphology in women with a history of gestational diabetes. Among the women with previous gestational diabetes, irregular cycles were more prevalent in the PCO group than in the women with normal ovarian morphology, but no other differences in endocrine or metabolic parameters were detected. These findings confirm an association between PCO and gestational diabetes and suggest that women with gestational diabetes display metabolic abnormalities irrespective of ovarian morphology.

The transmission disequilibrium test with use of trios (an affected proband with both parents) is a robust method for assessing the role of gene variants in disease that avoids the problem of population stratification that may confound conventional case/control studies and allows the detection of parent-of-origin effects. Trios have played a major role in defining genes in a number of polygenic conditions, including type 1 diabetes. We assessed the prevalence, clinical characteristics, and suitability for defining type 2 susceptibility genes of European type 2 diabetes trios. In a Caucasian population in the U.K., only 2.5% of type 2 patients had both parents alive. Using a nationwide strategy, we collected 182 trios defined by strict clinical criteria. Immunological and genetic testing resulted in the exclusion of 25 trios as a result of latent autoimmune diabetes (n = 13), inconsistent family relationships (n = 7), and maternally inherited diabetes and deafness (n = 5). The 157 remaining probands had similar treatment requirements to familial type 2 diabetic subjects but presented at a younger age, were more obese, and more frequently had affected parents. Using this resource, we have not found any evidence for linkage disequilibrium between type 2 diabetes and the glucokinase gene markers GCK1 and GCK2 and the chromosome 20 marker D20S197. We conclude that European type 2 diabetes trios are difficult to collect but provide an important additional approach to dissecting the genetics of type 2 diabetes.

Angiogenesis, the sprouting of new blood vessels from existing vessels, occurs in many physiological and pathological processes, including em bryonic development, wound healing, and tumor growth. It is required for tumor growth because new blood vessel formation is necessary for tumors to expand beyond a minimum volume. Several growth factor receptor tyrosine kinases have been implicated in angiogenesis, including receptors for epidermal, fibroblast, and platelet-derived growth factors, as well as the receptors Flk-1/KDR, Flt-1, Tek/Tie-2, and Tie-1. Endothelial cells in the vessels of tumors express Flk-1/KDR, a receptor for vascular endothelial growth factor. Flk-1 was previously shown to play a role in angio genesis and tumor formation of s.c. xenografts of C6 glioma cells using dominant-negative methodology. We now demonstrate that Flk-1 seems to be generally involved in the growth of a wide range of solid tumors, including mammary, ovarian, and lung carcinoma, as well as glioblastoma. Furthermore, survival times in rats bearing intracerebral tumors were prolonged using the same dominant-negative methodology. The involvement of Flk-1 in a variety of tumor types suggests an important role for Flk-1 in tumor angiogenesis.

A novel member of the eck/eph family of receptor tyrosine kinases (RTKs), termed mouse developmental kinase 1 (MDK1), was identified and shown to be closely related to the Eek, Ehk1/Cek7, Ehk2, Cek4/Mek4/hek and Sek/Cek8 subfamily. Northern blot analysis revealed MDK1 mRNA transcripts of 6.8, 5.7, 4.0, 3.2 and 2.6 kb that encode apparent splice variants. Sequence analyses of MDK1 cDNA clones from adult mouse brain predict the existence of at least five isoforms, including two truncated receptor variants lacking the kinase domain. Northern blot and in situ hybridization analysis indicate that in the adult mouse MDK1 RNA expression is restricted to brain, testes and spleen. The distinct patterns of MDK1 gene expression during mouse development suggest an important role in the formation of neuronal structures and possibly other morphogenic processes.

Mammary carcinoma kinase 10 (MCK-10) and colon carcinoma kinase 2 (CCK-2) constitute a subclass of receptor tyrosine kinases characterized by a discoidin I motif in the extracellular domain and a large cytoplasmic juxtamembrane (JM) region. While the ectodomain structure suggests a common role in cell aggregation, the JM domains of MCK-10 and CCK-2 are structurally most divergent and display features that suggest an involvement in signal generation and definition. MCK-10 occurs in at least three isoforms, which contain alternatively spliced consensus sequences for internalization and SH3 domain interaction. The presence of the 37 amino acid insert affects receptor autophosphorylation and changes ectodomain glycosylation. Proteolytic cleavage within the extracellular domain of MCK-10 generates a membrane-anchored kinase domain and releases a soluble ectodomain fragment including the discoidin I homology domain. CCK-2 and MCK-10 expression was found in connective and epithelial tissues, respectively, which in cancers of epithelial origin results in mutually exclusive expression in stroma and tumor cells, indicating a possible involvement of this class of RTKs in tumor invasion.

ANGIOGENESIS, the sprouting of capillaries from pre-existing blood vessels, is a fundamental process in the formation of the vascular system during embryonic development. In adulthood, angiogenesis takes place during corpus luteum formation and in pathological conditions such as wound healing, diabetic retinopathy, and tumorigenesis. Vascularization is essential for solid tumour growth and is thought to be regulated by tumour cell-produced factors, which have a chemotactic and mitogenic effect on endothelial cells1-4. Vascular endothelial growth factor (VEGF), a homodimeric glycoprotein of relative molecular mass 45,000, is the only mitogen, however, that specifically acts on endothelial cells, and it may be a major regulator of tumour angiogeness in vivo5,6. Its expression has been shown to be upregulated by hypoxia, and its cell-surface receptor, Flk-1, is exclusively expressed in endothelial cells7,8. Here we investigate the biological relevance of the VEGF/Flk-1 receptor/ligand system for angiogenesis using a retrovirus encoding a dominant-negative mutant of the Flk-1/VEGF receptor to infect endothelial target cells in vivo, and find that tumour growth is prevented in nude mice. Our results emphasize the central role of the Flk-1 /VEGF system in angiogenesis in general and in the development of solid tumours in particular.

Examination of flk-1 receptor tyrosine kinase mRNA expression by in situ hybridization analysis revealed specific association with endothelial cells at all stages of mouse development, including the blood islands in the yolk sac of day 8.5-10.5 embryos, in which the early progenitors of this lineage originate. flk-1 transcripts were abundant in proliferating endothelial cells of vascular sprouts and branching vessels of embryonic and early postnatal brain, but were drastically reduced in adult brain, where proliferation has ceased. Identification of the angiogenic mitogen, vascular endothelial growth factor (VEGF), as the high affinity ligand of Flk-1 and correlation of the temporal and spatial expression pattern of Flk-1 and VEGF suggest a major role of this ligand-receptor signaling system in vasculogenesis and angiogenesis.

We have recently shown that vascular endothelial growth factor (VEGF) is produced by human malignant glioma cells and acts on tumor endothelial cells, which express VEGF receptors, suggesting that VEGF is a regulator of tumor angiogenesis. To investigate the feasibility of antiangiogenic brain tumor therapy, we developed an intracerebral (i.c.) rat glioma model. We used two transplantable rat glioma cells lines, C6 and GS-9L, to analyze VEGF regulation in vitro and expression of VEGF and its high affinity tyrosine kinase receptors, flt-1 and flk-1, in vivo. Glioma cells were transplanted i.c. or s.c. into syngeneic rats. C6 gliomas exhibit morphological characteristics of human glioblastoma multiforme such as necroses with palisading cells. Immunocytochemistry with von Willebrand factor showed that C6 gliomas are highly vascularized and therefore show another prominent feature of human glioblastoma. GS-9L gliosarcomas were less vascularized. In situ hybridization showed that VEGF is expressed in vivo in rat glioma cells which reside along necrotic areas and therefore closely mimicks the expression pattern of VEGF observed in human glioblastoma. flt-1 and flk-1 are specifically expressed in endothelial cells in the tumor and at the border between tumor and normal brain but are absent from endothelial cells in the normal brain proper. The action of VEGF may therefore be restricted to tumor endothelium. Upregulation of VEGF, but not acid fibroblast growth factor, basic fibroblast growth factor, and platelet-derived growth factor B messenger RNA was observed in hypoxic C6 and GS-9L cells in vitro. These observations are consistent with a role for VEGF in tumor- and hypoxia-induced angiogenesis. Since the expression pattern of VEGF and its receptors in rat glioma appears to be indistinguishable from human glioblastoma multiforme, this model provides an excellent tool to study anti-angiogenic therapy.

Chimeric receptors composed of the human epidermal growth factor receptor (EGF-R) extracellular domain fused to wild-type and truncated platelet-derived growth factor receptor (PDGF-R) intracellular sequences were stably expressed in NIH 3T3 cells devoid of endogenous EGF-Rs. This experimental system allowed us to investigate the biological activity of PDGF-R cytoplasmic-domain mutants in PDGF-R-responsive NIH 3T3 cells by activating PDGF-specific signaling pathways with EGF. Deletion of 74 carboxy-terminal amino acids severely impaired the ability of the PDGF-R cytoplasmic domain to associate with cellular substrates in vitro. This deletion also inhibited receptor and substrate phosphorylation, reduced the receptor's mitogenic activity, and completely abolished its oncogenic signaling potential. Surprisingly, removal of only six additional amino acids, including Tyr-989, restored substantial receptor and substrate phosphorylation capacity as well as transforming potential and yielded a receptor with wild-type levels of ligand-induced mitogenic activity. However, the ability of this chimera to bind phospholipase C gamma was severely impaired in comparison with the ability of the wild-type receptor, while the association with other cellular proteins was not affected. Further deletion of 35 residues, including Tyr-977, nearly abolished all PDGF-R cytoplasmic-domain biological signaling activities. None of the three C-terminal truncations completely abolished the mitogenic potential of the receptors or had any influence on ligand binding or receptor down regulation. Together, these data implicate the 80 C-terminal-most residues of the PDGF-R, and possibly Tyr-989, in phospholipase C gamma binding, while receptor sequences upstream from Asp-988 appear to be essential for specific interactions with other cellular polypeptides such as ras GTPase-activating protein and phosphatidylinositol 3-kinase. Thus, the mutants described here allow the separation of distinct PDGF-activated signaling pathways and demonstrate that phospholipase C gamma phosphorylation is not required for mitogenesis and transformation.

A chimeric receptor consisting of an epidermal growth factor (EGF) receptor ligand-binding domain and platelet-derived growth factor (PDGF) receptor transmembrane and cytoplasmic signalling domains has been constructed and shown to be fully functional in phosphorylation, mitogenesis, transformation, Ca2+ release, and pH change assays. Expression of this receptor in EGF receptor-deficient, PDGF-responsive NIH 3T3 cells allows the activation of PDGF signalling pathways by EGF. This system was used to examine the function of kinase insertion sequences (KIS). While a mutant with a KIS deletion of 83 amino acids displayed a significant but reduced ability to induce mitogenic, transforming, and Ca2+ release responses in transfected cells, deletion of 20 additional amino acids resulted in abolishment of such activities. This differential loss of signalling potential correlated with the reduced or abolished potential of these receptor mutants to phosphorylate cellular substrates such as PLC gamma. Our results suggest an integral role for KIS in PDGF receptor cytoplasmic domain conformation and an involvement in substrate interaction, but provide no evidence for an exclusive role of KIS in the mediation of biological signals.