Altered cellular interactions between endothelial cells and nonenzymatically glucosylated laminin/type IV collagen.

Laminin and type IV collagen are two major basement membrane glycoproteins. In previous studies it has been shown that nonenzymatic glucosylation induces structural alterations of these macromolecules and also reduces their ability to self-associate. In the present study, endothelial cells were tested for their ability to adhere and spread on nonenzymatically glucosylated laminin and type IV collagen. Adhesion and spreading were reduced when glucosylated macromolecules were used as substrates. Glucosylation-induced changes in adhesion and spreading may be an important initial event signaling other phenotypic modifications of cells in the microvasculature and may be a crucial factor in order to understand the pathogenesis of diabetic microangiopathy at the molecular level.

It is by now well established that nonenzymatic glucosylation is one of the main mechanisms by which hyperglycemia affects structurally and functionally various macromolecules (1)(2)(3)(4)(5)(6)(7)(8)(9)(10). The major targets in this process are macromolecules with relatively long half-lives; among them, basement membrane components are of major interest because their alterations may be a key factor contributing to diabetic microangiopathy.
We have examined in the past the effect of nonenzymatic glucosylation on two major basement membrane glycoproteins, laminin and type IV collagen (11,12). We have observed that their structural alterations due to high glucose concentration can lead to defective interactions that may have a profound influence on the molecular architecture of the basement membrane (11,12). In the vasculature, however, basement membrane components interact not only with each other but also with the endothelial cells that line every vessel, and these interactions are crucial for various endothelial cell prop-* This work was supported by a grant from the Juvenile Diabetes Foundation (to E. C. T.), grants from the American Heart Association, Minnesota Affiliate (to A. S. C. and E. C. T.), and National Institutes of Health Grants DK43569 (to A. S. C.) and DK39216 and DK43574 (to E. C. T.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
7 To whom correspondence and reprint requests should be addressed: Dept. of Laboratory Medicine and Pathology, University of Minnesota Medical School, Box 609 UMHC, 420 Delaware St. S.E., Minneapolis,MN 55455. erties and ultimately for the structural and functional integrity of the vascular wall.
Recently, using bovine aortic endothelial cells as a model system, we have started to explore the effect of nonenzymatic glucosylation of basement membrane components on various endothelial cell properties. In the present report we have focused on two of the earliest types of interactions between endothelial cells and laminin/type IV collagen: cell adhesion and cell spreading.

EXPERIMENTAL PROCEDURES
Laminin and type IV collagen were isolated from the Engelbreth-Holm-Swarm tumor matrix according to previously described protocols (11,13,14).
Nonenzymatic Glucosylation of Laminin and Type IV Collagen-Laminin was dialyzed against phosphate-buffered saline (PBS)' pH 7.4, containing 10 mM EDTA, 50 pg/ml pbenylmethylsulfonyl fluoride, 50 pg/ml N-ethylmaleimide (Sigma), and 0.02% NaN3. Type IV collagen was dialyzed against the same buffer which also contained 0.5 M NaCl. Both proteins were centrifuged at 20,000 rpm for 30 min to remove large aggregates. The nonenzymatic glucosylation of both proteins started by the addition of glucose in the same buffers, so that the final concentration of laminin and type IV collagen was 250 pg/ml and those of glucose were 0 (control), 50, and 500 mM. The nonenzymatic glucosylation was performed at 29 "C for 60 h in the dark with occasional shaking. At the end of the incubation samples were dialyzed against PBS at 4 "C. The protein concentration was measured, and after appropriate adjustments, the samples were immediately used for coating plates. Nonenzymatically glucosylated laminin and type IV collagen were subjected to SDS-polyacrylamide gel electrophoresis in 6% gels under reducing conditions (in the presence of 2 mM P-mercaptoethanol) using well established techniques. Similar gels were run using proteins labeled with lz5I and then incubated in the absence or presence of glucose. The gels were then dried and subjected to autoradiography.
Glucose Incorporation in Nonenzymatically Glucosylated Type IV Collagen and Laminin-In order to determine the number of glucose molecules incorporated under our experimental conditions in type IV collagen and laminin, we used [6-3H]glucose (from ICN Radiochemicals, CA) to glucosylate these proteins. The commercially available solution of ~- [6-~H]glucose was first dried under a stream of nitrogen. The dried radiolabeled glucose was then diluted in PBS containing 10 mg/ml bovine serum albumin (BSA from ICN ImmunoBiologicals, Costa Mesa, CA), 10 mM EDTA, pH 7.4 and incubated for 3 days at 37 "C. In order to separate the radiolabeled glucose from unknown contaminants that usually coexist in the commercial batches of radioactive glucose (15), the above mixture was loaded on a 25-ml Sephadex G-25 column previously equilibrated with PBS, 10 mM EDTA, pH 7.4, and eluted in a flow rate of 1 ml/min with the same buffer; 0.5-ml fractions were collected and counted in a scintillation counter, and the fractions containing free radioactive glucose were pooled and lyophilized. Unlabeled, lyophilized glucose was then added to a final concentration of 50 or 500 mM, and the pooled peak was combined with type IV collagen and/or laminin as above described. The nonenzymatic glucosylation was stopped with dialysis at 4 'C against 0.1 M phosphate buffer, pH 7.0, and the glycated proteins were incubated with 200 molar excess of NaBHl (sodium borohydride diluted in 0.2 M NaOH, from Sigma) for 10 min at room temperature and, then, for another 50 min at 4 "C. All the samples were dialyzed against 0.1 M phosphate buffer, pH 7.0, then against 1 M urea in 0.1 M phosphate buffer, pH 7.0, to remove nonspecifically bound radioactivity and, finally, against PBS containing 10 mM EDTA, pH 7.4. The protein content of each sample was measured, and aliquots from The abbreviations used are: PBS, phosphate-buffered saline; SDS, sodium dodecyl sulfate; BSA, bovine serum albumin; HEPES, N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid; HBSS, Hanks' balanced salt solution; DMEM, Dulbecco's modified Eagle's medium. each sample were precipitated with acetone and counted in the scintillation counter (Beckman LS 5000TD). In order to assess the extent of cross-link formation, aliquots were examined in a Perkin Elmer LS3B fluorescence spectrometer at an excitation wavelength of 370 nm and an emission wavelength of 440 nm.
Cell Cultures-Bovine aortic endothelial cells were isolated from calf aortas after the protocol described previously (16) with a few modifications. Briefly, aortas were obtained from 6-month-old calves at a local slaughterhouse and processed within 2 h from the time of death. The vessels were washed with a sterile solution of Hanks' balanced salts (HBSS with Ca2+ and M e ) , 50 mM HEPES (Sigma), 1000 units/ml penicillin G, 1000 pg/ml streptomycin, and 500 pg/ml gentamycin sulfate, pH 7.4, and digested with a sterile solution of 2 mg/ml of collagenase CSLIII (from Worthington) in HBSS, 0.35 g/ liter NaHC03, 200 units/ml penicillin, 200 pg/ml streptomycin, and 100 pg/ml gentamycin sulfate, pH 7.4. The cell suspension was centrifuged at 1,500 rpm for 10 min, and the pellet was resuspended in Dulbecco's modified Eagle's medium (DMEM from Sigma) containing 10% fetal calf serum and the above antibiotics. The endothelial cells were identified by their characteristic cobblestone morphology, their contact inhibition upon reaching confluency and the positive staining for factor VI11 antibody, as reported elsewhere (16). The endothelial cells were routinely grown in 75-cm2 flasks in DMEM containing 10% calf serum and 10% fetal calf serum and were used until passage 15 to avoid phenotypic drift and senescence with continuous subpassaging.
Cell Adhesion Assays-96-well plastic plates were routinely coated with control and nonenzymatically glucosylated type IV collagen and laminin by the addition of 50 pl of solution/well. Each macromolecule was used for coating at five different concentrations: 100,50,25, 12.5, and 1 pg/ml in PBS in quadruplicates. The plates were left overnight at 29 "C to dry and were kept at 4 "C, until used during the following 7-8 days.
Subconfluent (60-70% confluent) bovine aortic endothelial cells were labeled with 500 mCi of [35S]methionine (from ICN Radiochemicals) for 18 h. Cells were washed with an HBSS solution containing Ca2+ and Mg+, then with HBSS containing 1 mM EDTA, and were detached by the addition of a 0.05% trypsin, 0.75 mM EDTA solution. The cells were then washed three times with 10 ml of DMEM by centrifugation at 1200 rpm for 5 min and finally diluted in 4 ml of a DMEM solution containing 25 mM HEPES, 2 mg/ml BSA, pH 7.4 (binding buffer). The number of cells was calculated using a hemocytometer. Plates coated with nonenzymatically glucosylated proteins were incubated with PBS containing 2 mg/ml BSA, pH 7.4, for 60 min at 37 "C in a humidified incubator. Subsequently, five thousand cells were applied on each well for various time intervals up to 60 min. The non-adherent cells were aspirated at the end of the incubation, and the plates were washed three times with 200 pl of binding buffer/well. Finally, the plates were incubated for 30 min at 60 "C with 100 pllwell of a solution of 0.5 N NaOH, 1% SDS (lysis buffer). This solution was then transferred in vials with scintillation fluid, and the radioactivity was counted in a scintillation counter. The number of cells adherent was variable depending on the length of the incubation, but it was always in the range of 25-40%. In all the experiments, control wells were used where no laminin or type IV collagen were present. Each value reflects the specific adhesion, which is the adhesion on the macromolecule minus the adhesion on BSAcoated wells. Control values never exceeded 5% of the experimental values.
Cell Spreading Assays-96-well plastic plates were coated with control and glucosylated type IV collagen and laminin as described above. For these assays, cells were used without any metabolic labeling and were incubated in matrix-coated wells as described in cell adhesion assays. At the end of the incubation time non-adherent cells were aspirated. The wells were washed 3 times with PBS and fixed with 2% glutaraldehyde in PBS for 15 min at 20 "C. The fixative was then aspirated, the wells were washed with PBS, stained with 0.1% crystal violet (Fisher Scientific) for 30 min, washed with water, and dried.

RESULTS
Laminin and type IV collagen were nonenzymatically glucosylated in vitro using conditions that would minimize as much as possible degradation or other structural changes not related to elevated glucose concentrations. For that purpose, we exposed these macromolecules to glucose for a very short incubation time, in the presence of protease inhibitors, an-tioxidants, and agents preventing macromolecular self-association. At the end of the incubation period, unincubated macromolecules as well as macromolecules incubated in the presence of 0,50, and 500 mM glucose were examined by SDSpolyacrylamide gel electrophoresis under reducing conditions. The polypeptide chains of both laminin and type IV collagen appeared intact and did not exhibit any degradation products or shift in their electrophoretic mobility.
This finding was confirmed using '251-labeled macromolecules, where after incubation in the presence or absence of glucose structural integrity was monitored by autoradiography (data not shown).
The extent of glucose incorporation was determined as described above. In order to stabilize glucose adducts, sodium borohydride was used at the end of the incubation period followed by extensive dialysis against urea. At 50 mM glucose, an average of 0.9 or 1.35 nM of glucose were incorporated per nM of laminin or type I v collagen. At 500 mM glucose, an average of 11.04 nM glucose per nM of laminin and 12.02 nM of glucose per nM of type IV collagen were found incorporated.
The extent of cross-links formed was measured spectrophotometrically and found to be minimal (data not shown); this is not surprising in view of the very short incubation time. Gel electrophoresis profiles confirmed this finding.
The ability of control and nonenzymatically glucosylated laminin and type IV collagen to promote cell adhesion and spreading was then examined. For cell adhesion experiments, plastic plates were coated with these macromolecules after incubation in the absence of glucose or in the presence of 50 and 500 mM glucose, as described. It was found that under the conditions used, approximately 40% of the control macromolecules are retained on the plastic. Nonenzymatic glucosylation slightly increased this number, and therefore calculations of cell adhesion and spreading were made after adjusting for the mass of coated protein. Glucosylation in 50 mM glucose increased the coating of laminin and type IV collagen by 1% and 8%, respectively; in 500 mM glucose these increases were 11% and 30%.
Bovine aortic endothelial cells were isolated, cultured, and characterized as described (16). Subconfluent cells were metabolically labeled with [35S]methionine, detached, washed, and allowed to adhere to coated substrates, as described above. Adhesion was examined after 15 min for type IV collagen and 20 min for laminin. The differences in time intervals used in this and subsequent assays were dictated by the fact that bovine aortic endothelial cells exhibit different affinities for the two substrates; they adhere more avidly to type IV collagen compared with laminin. The results are shown in Fig. lA for type IV collagen and Fig. 1B for laminin. In the case of type IV collagen, a reduction of cell adhesion in the range of 30-40% was observed between control and glucosylated samples, in all coating concentrations used (100 to 1 pg/ml); differences in the samples glucosylated at low or high glucose concentrations were smaller and became statistically significant only at the highest coating concentration used (Fig. lA).
In the case of laminin, a reduction of cell adhesion in the range of 20-30% was observed in all but the lowest coating concentration. However, samples glucosylated under different glucose concentrations did not exhibit any statistically significant difference when grown on laminin (Fig. 1B). We cannot satisfactorily explain this somewhat puzzling finding. It could be due to several factors: severe steric changes induced even at low glucose concentrations, differential adherence to plastic of heavily modified macromolecules, or lack of sensitivity of the technique used. These experiments were performed at least three times in quadruplicates and the data were analyzed statistically using the Mann-Whitney nonparametric test (17). Similar results were obtained when the adhesion assay was allowed to proceed for longer time intervals: 45 min for type IV collagen and 60 min for laminin (data not shown). Adhesion experiments were performed in the absence of serum and were not allowed to proceed beyond 60 min for two reasons: first, in order to minimize the secretion of matrix proteins by endothelial cells, a factor that may interfere with the assays used in these studies; and second, in order to avoid unknown stimulatory or inhibitory effects due to serum factors.
The effect of nonenzymatic glucosylation on the ability of laminin and type IV collagen to promote cell spreading was examined next. Two methods were used to assess the extent of cell spreading. In the first method, the percent of cells exhibiting "long processes" was determined. A long process was operationally defined as a protrusion from the cell body longer than the maximal diameter of the cell under examination. The number of cells with long processes decreased by 43% between control and type IV collagen incubated in the presence of 50 mM glucose, and by 62% between control and type IV collagen incubated in the presence of 500 mM glucose. In the case of laminin, no reduction in the number of cells with long processes was observed when control was compared with glucosylated laminin at 50 mM glucose, but a 45% decrease in cell processes was seen when cells were allowed to spread on glucosylated laminin at 500 mM glucose. These data were obtained at coating concentrations of 50 pg/ml for each macromolecule. Lower coating concentrations tended to minimize these differences (data not shown).
As a complementary and more objective approach, the perimeter of cells was measured. Bovine aortic endothelial cells allowed to spread on control and glucosylated type IV collagen for 15 min or on control and glucosylated laminin for 30 min were examined using a Nikon Diaphote microscope and their perimeter was traced by using an Optomax camera connected to an Apple IIe computer. Again, the difference in time interval in which spreading was stopped by fixation was dictated by differences in affinity for these macromolecules observed with bovine aortic endothelial cells. In these experiments, three different coating concentrations were used (50, 25, and 12.5 pg/ml). At least six randomly chosen fields were analyzed per well, and at least 10 wells were used for every permutation. The data shown in Fig. 2A (for type IV collagen) and Fig. 2B (for laminin) demonstrate that nonenzymatic glucosylation reduced the ability of cells to spread. In this case, coating with heavily glucosylated macromolecules resulted in a pronounced reduction in cell perimeter compared with cells examined on lightly glucosylated type IV collagen and laminin. Because totally round cells exhibit a perimeter in the range of 70-80 pm, the magnitude of the observed reductions constitutes a major phenotypic change. Similar results were obtained when cell spreading was assessed for longer time intervals (up to 60 min). Fig. 3 shows representative fields of cells allowed to adhere and spread on type IV collagen for 15 min. It can be appreciated that compared with control ( Fig. 3A), cells on glucosylated macromolecules (Fig.   3, B and C) exhibit less extensive processes. Also, more extensive glucosylation results in reduction in the number of existing processes (Fig. 3, compare panels B and C). thelia1 and possibly other cell types. Differential adhesion and spreading under diabetic conditions may lead to differences in growth, proliferation, and secretory activity of specific cell types. It is well documented that in diabetic retinopathy retinal microvascular endothelial cells may have an increased proliferation whereas microvascular pericytes are reduced in number (24). In diabetic nephropathy a sizable increase in the mesangial matrix is observed (25). In uitro studies have suggested that nonenzymatically glucosylated mesangial matrix resulted in reduced ability of mesangial cells to proliferate (26). Thus, diverse effects could be caused on various aspects of cell behavior as a result of matrix nonenzymatic glucosylation.

In this study, evidence is provided that endothelial cells undergo changes in their interactions with nonenzymatically
Many factors are likely to be involved in diabetic microangiopathy. It has been suggested that endothelial cell behavior may be altered by exposure to a high glucose concentration, even for a short time interval (27,28) and/or by excessive sorbitol formation via the poly01 pathway (29). This report provides evidence for yet another important mechanism which may contribute to altered endothelial cell properties in diabetic conditions, suggesting that nonenzymatic glucosylation of laminin and type IV collagen may eventually influence cellular phenotype. glycosylated laminin and type IV collagen, both important components of all basement membranes. These altered interactions resulted in a impaired adhesion and spreading to each of these basement membrane glycoproteins.
Basement membrane macromolecules and matrix components in general, are important elements in the regulation of cellular phenotype (18). Matrix mediated changes of cell shape interfere with growth and differentiation. Extracellular matrix components are crucial in determining cell shape and induction of specific gene expression in several cells, including cultured adipocytes (19), mammary epithelial cells (20), pheochromocytoma cells (21), and hepatocytes (22). Adhesion and spreading should be considered as very early events in the interaction between cells and extracellular matrices; therefore, these early binding events may be very critical in determining the extent and the nature of such interactions.
Alterations in the ability to adhere and spread have been very well documented in the case of transformed cells and associated with the expressed malignant phenotype (23). The data presented in this study provide support to the hypothesis that in diabetes, hyperglycemia-induced matrix modifications may to some extent influence the cellular properties of endo-