Interactions of insulin with bovine endothelium

I n t e r a c t i o n s of Insulin With Bovine Endothelium Michael L. Peacock, Robert S. Bar, and Jonathan Goldsmith Bovine endothelial cells have been isolated from pulmonary and systemic vessels and grown in culture as primary, passaged and passaged cloned-strains. The cultures w e r e shown to be endothelial in nature on the basis of several endothelial-specific and endothelial-associated traits. Endothelial cells from all sources had specific receptors for insulin in primary culture and after serial passage. Endothelial cells derived from pulmonary arteries and aortas bound 2.5 times more insulin than cells derived from the pulmonary vein. Each endothelial cell type maintained a specific complement of receptors through at least 25 passages in vitro. Coupled with previous findings of insulin receptors on endothelial cells from human umbilical vessels, these data suggest that insulin receptors may be an intrinsic component of all vascular endothelium.

ENDOTHELIUM forms the intimal surface T HEof blood vessels. It is composed of metabolically active cells that can be successfully cultured in vitro and identified by specific morphological and functional criteria, j We have recently demonstrated that primary cultures of endothelial cells derived from human umbilical vessels possess specific surface receptors for insulin.2 However, studies with the human cells were limited due to the difficulty and expense in preparing primary cultures, the relatively small number of cells obtained and the inability to subculture the cells on a routine basis. In addition, we were concerned that receptors for insulin might be limited to the umbilical vessel endothelium, a highly specialized form of fetal tissue. In the present study, we have prepared primary cultures of endothelial cells from bovine vessels of the pulmonary and systemic circulation and subcultured these cells in vitro. We found that (1) all bovine endothelial cells studied possessed specific surface receptors for insulin, (2) each source of endothelial cells retained a specific complement of insulin receptors through at least 25 passages in vitro and (3) endothelial cells from aortas and pulmonary arteries bound 2.5-fold more insulin per cell than cells from pulmonary veins, with the differences in binding maintained with serial subculture. MATERIALS A N D METHODS

Cell Culture Bovine cells. Endothelial cells were prepared from bovine pulmonary arteries, pulmonary veins and aortas. For each vascular

From the Department o f lnternal Medicine VA Hospital, Diabetes-Endocrinology Research Center, University o f Iowa, Iowa Oty, Iowa. Supported in part by Grant AM2542I from the National Institutes o f Health and by grants from the Veterans" Administration, the Iowa Affiliate o f the American Diabetes Association, the American Lung Association and the Iowa Heart Association. Received for publication February 26, 1981. Address reprint requests to Robert S. Bar, M.D., 3E-21, VA Hospital, Iowa Oty, Iowa 52242. ©1982 by Grune & Stratton, Inc. 00 26~94 95/8 2/3101~9009501.00/0 52

source, cells were prepared as primary cultures, passaged cultures and cloned cell strains.4,5 (1) Primary and passaged cultures: Pulmonary arterial and aortic cell cultures were initiated from vessels obtained from freshly slaughtered steers and heifers. The vessels were clamped, and excised, then rinsed with .01M phosphate buffered saline (PBS) and filled with 0.1% crude collagenase (Worthington Biochemical Co., Freehold, N.J.). After 20 min, the cell suspension was collected and pooled with cells obtained by a subsequent PBS rinse of the vessels. The cell suspensions were centrifuged at 250 × g for 10 rain. Cell pellets were then resuspended in M-199 with 17% fetal bovine at a density of ~25,000 cells/cm2 of dish surface. Cells from bovine pulmonary veins were obtained by scraping vascular walls with a sterile cotton-tipped applicator. The endothelial cells were then handled as previously described for the bovine arterial endothelium. For experimental purposes, cells were plated on 35 mm dishes, and allowed to grow to conflueney in a 5% CO2 atmosphere at 37°C. Medium was changed at 24 hr and then every 48-72 hr. The cells had an apparent doubling time of 24 hr, and reached confluence in 72-96 hr. To subculture cells, the medium was removed and replaced with PBS containing 0.1% trypsin and 0.05% EDTA until cell detachment occurred. These suspensions were centrifuged under the previously detailed conditions (vide supra). Cells were replated at a 1:5 dilution. Confluent monolayers were used for study 4-6 days after seeding. (2) Cloned strains: Cloned strains for pulmonary arterial, pulmonary venous and aortic endothelial cells were initiated by diluting primary suspensions until fewer than 20 cells per milliliter were present. These cells were plated in 100/~1 wells (Falcon micro test II culture plates, Falcon, Oxnard, Ca). Wells containing only single endothelial cells were allowed to divide and then passed onto 35 mm dishes, and subsequently passed every 5 to 7 days. Insulin binding studies were performed on 2 separate clones of cells from pulmonary arteries and veins and on one clone of aortic cells. The endothelial nature of the primary, passaged and cloned cells was established on the basis of several criteria that are either specific for or consistent with an endothelial cell origim 1'4 The cells had a uniform appearance and demonstrated an epithelioid growth pattern. Cells grew to confluent monolayers and obeyed density dependence, even after 30 serial passages. Cells were uniformly positive for factor VIII antigen, synthesized and released prostacyclin, angiotensin converting enzyme and plasminogen activator. Immunofluorescence for factor VIII antigen was assessed using rabbit antibody to bovine factor VIII and FITC-conjugated goat anti-rabbit IgG (Cappel Labs, Cochranesville, PA.); bovine fibroblasts and smooth muscle cells of vascular origin were similarly tested and found to lack fluorescence. Prostacyclin was determined as its metabolite, 6-keto-PGF~ using a specific radioimmunoassay that has been previously described for 6-keto-PGFl~. 4'6 Angiotensin converting enzyme (kininase II) was determined by radioassay (Ventrex Laboratories Laboratories, Portland, Me.). 7 Plasminogen activator was assayed by both a soluble-phase system based on the Metabolism, Vol. 31, No. 1 (January), 1982

INSULIN AND BOVINE ENDOTHELIUM

BOVINE AORTA 2

53

~

2.4

/

/,-"~

SPECIFIC

~-~

~3 Z --~ ~ 1.6

//"

.-1 ffl

Z

f

~ O.~

/'~/f •

"NO.-SPEe~F,C",

...~. .......

,, ZX

...'"

¢..

~.b

o15

~.5

zlo

61o

""

HOURS Fig. 1. [~2Sl]-insulin binding to passaged bovine aortic endothelium at 22°C. Binding is expressed per 10 e cells. time required to lyse a standard fibrin clot in the presence of excess plasminogen8 and also in a two-stage assay measuring hydrolysis of N-a-CBZ-L-Lysine-P-Nitrophenylester, monitored spectrophotometrically at 340 nM and compared to a solid phase urokinase standard. 9 Human cells. Primary cultures of endothelial cells from h u m a n umbilical veins were prepared and characterized as previously described, zJ° After m a n y unsuccessful attempts to passage these cells, two separate groups of cells were subcultured using trypsin/ E D T A to detach cells and M-199 plus 17% fetal bovine serum as medium. In both groups, after two passages, the doubling time slowed to 4-10 days and increased numbers of larger, vacuolated cells which were factor VIII antigen positive were present. Binding studies were performed through passage 10. Insulin binding studies. Binding studies were performed on adherent, monolayer cultures as previously described. 2'3 Data are expressed as binding per cell with each d a t u m point representing the mean of 3-5 dishes. Prior to binding experiments, cell counts were performed on monolayer cells using a calibrated ocular recticle, with 1000-1500 cells being counted per dish. [125I]-iodo-insulin (150-200 u C i / u g insulin) was prepared by step-wise chloramine T oxidation.H'~2 Proinsulin and desoctapeptide insulin were kind gifts of Drs. R. Chance and F. Carpenter.

PULMONARY VEIN

E .E I00

E t~ Z

~ 50 _z

._1

Nz_ z5 ¢ 10 II

The passaged bovine cells demonstrated characteristics of [12sI]-iodo-insulin binding that were similar to each other and similar to other, more classical receptors for insulin. The binding of [125I]-iodo-insulin was dependent on the pH, time and temperature of incubation. A sharp pH optimum was found at 7.8. Maximal steady-state binding was reached at 24-48 hr at 4°C, at 1.5-2 hr at 22°C and in less than 0.75 hr at 370C. At steady-state, the maximal specific binding of [~25I]iodo-insulin was in the order 4 ° > 22 ° > 37 °. In the subsequent studies the binding assays were performed at pH 7.8, 22°C for 2 hr of incubation. Under these conditions specific binding was always >80% of total binding, and degradation of [~25I]-iodo-insulin was <10% as determined by precipitability in 5% trichloroacetic acid (Fig. 1). All passaged cells demonstrated specificity of binding. In general, biologically active insulins and insulin derivatives competed with [~25I]-iodo-insulin in proportion to their biological activity, as determined by in vitro stimulation of glucose oxidation in isolated fat cellsJ 2 Thus, for the pulmonary cells, porcine insulin was more potent than porcine proinsulin which was more potent than desoctapeptide (DOP) insulin in competing for [~25I]-iodo-insulin binding (Fig. 2). In the single experiment performed with passaged aorta cells, DOP insulin was nearly as potent as proinsulin. Further evidence of the specificity of insulin binding was indicated by the interaction of the endothelial cells with anti-insulin receptor antibodies contained in the serum of a patient (B-2) with autoimmune insulin resistance and acanthosis nigricans) 3 After brief exposure to serum B-2 (1:500 dilution), insulin binding was decreased by 65%, 78% and 91% in the pulmonary artery, aorta and pulmonary vein cells, respectively. Although qualitative aspects of insulin binding were

PULMONARY ARTERY

AORTA

.~.:.-~......... ~.........

~, z5

RESULTS

o ...... -/~--------o

.:~DOP

"\x

\

• \x

~PR~NSUUN'X

\

Io -'°

o~PROINSULIN ~ \

""

",

...... \

-,,,):...

o

=o-'

""~

*\ INSULINI~~, INSULIN

ox ~suL~-~

%,~',..

io-'

Io-'

Io-s Io-"

.. .....,.,

\\.

"-~-~."~ ..

~

PR~N~N~,....

Nsu_~x,. ~ "",2£... x-",

I

16~°

16-'

16-s

1(3-7

I0"6 I0-"

'o

I0-I°

I0-9

[HORMONE] Moles/Liter Fig. 2. Competition for [1251]-insulin binding by porcine insulin, procine proinsulin and desoctapeptida (DOP) insulin in passaged bovine endothelial cells.

I0-a

I0-r

54

PEACOCK, BAR, AND GOLDSMITH

similar among the three sources of bovine endothelium, differences in quantitative binding were observed. For the passaged cells, maximal binding of [~25I]-iodo-insulin was 2.5% 10 6 cells for the aortic cells (N-5, SEM-0.2, passage 6-20), 2.7% for the pulmonary arterial cells (N-10, SEM-0.2, passage 7-25 of 3 separate groups of passaged cells) and 1.1% for the pulmonary venous ceils (N-9, SEM-0.2, passage 12-25 of 3 separate groups of passaged cells). Thus, in the cells from pulmonary vessels, arterial endothelial cells maximally bound 2.6 times more insulin than cells from the venous circulation (Fig. 3). The cloned strains demonstrated similar findings. Maximal binding averaged 3.0% (N-6) and 1.0% (N-4) for 2 separate clones of pulmonary arterial and pulmonary venous cells. In all cases, the differences in insulin binding appeared to reflect changes in receptor concentration without change in receptor affinity (as determined by Scatchard analysis and similar 150s or concentrations of insulin required to decrease maximal binding of [125I]-iodo-insulin by 50% [average 150 for each type of cell culture = 2 - 4 × 10 -1° M]. In all three types of bovine endothelial cells, maximal binding of [I2~I]-iodo-insulin as well as the binding capacity for insulin varied little throughout 25 passages. As the cells approached senescence (usually passage 25-35), the doubling time increased to 4-7 days, progressively greater numbers of larger vacuolated cells appeared and insulin binding decreased (Fig. 4, left). This phenomenon of decreasing insulin BOVINE PULMONARY =E

HUMAN UMBILICAL

i00

<~ tL 0 E3 z 50

"i

~

0

A

v

A

V

Fig. 3. Comparison of insulin binding in the venous cells IV, open bars) of the bovine pulmonary and human umbilical circulation to insulin binding in the corresponding arterial circulation (A, hatched bars), Maximal [12Sl]-iodo-insulin binding was 2.7%/108 cells (100% mean of 10 separate experiments) for the bovine pulmonary arterial cells, 1.1% (mean of 9 experiments) for the bovine venous cells, 6,6% (100% mean of 8 separate experiments) for the human umbilical arterial cells and 2 . 5 % (mean of 8 experiments) for the human umbilical venous cells. Data are presented as the mean _+1 S.E,M. Data for human umbilical vessels w e r e derived from reference 3.

BOVINE PULMONARYARTERY ¢o4

HUMAN UMBILICALVEIN

o

a

_z _J

O

•

0

.:"

,

•

12 15

.7

.

30

6

5

iO

~S~GENt.IVIBER Fig. 4. The effect of serial subculture on maximal [lzsI]-insulin binding in endothelial cells obtained from bovine pulmonary arteries (left) and human umbilical veins (right). Cultures w e r e prepared as described in Materials and Methods.

binding with apparent senescence of the cell cultures was observed in two separate groups of pulmonary arterial endothelial cells. The loss of insulin binding with senescence was not limited to the bovine pulmonary cultures. Subculturing of endothelial cells from human umbilical veins was associated with a similar series of events, i.e. the appearance of larger, vacuolated cells, a slowing of cell doubling and a loss of insulin binding (Fig. 4, right). For the human umbilical cells, apparent senescence and loss of insulin binding occurred as early as passage 2-3. In contrast, senescence in the bovine cells was not observed until prolonged subculture in vitro (Fig. 4). The data presented thus far have been obtained in the passaged and cloned cultures of bovine endothelium. Insulin binding was also assessed in primary cultures of the bovine pulmonary arterial cells and aorta cells. These cells were initially plated at subconfluent densities (-40% of confluency), allowed to grow to confluence and then assayed for [125I]-iodo-insulin binding. The primary cultures demonstrated appropriate specificity of binding (Fig. 5) and had quantitative binding that was similar to their subcultured counterparts; for primary pulmonary arterial cells, maximal binding of [125I]-iodo-insulin was 3.45%/106 cells (N = 2) versus 2.7% for subcultured cells (passage 7-25, N = 10), for primary aorta cells 2.5%/106 cells (N = 1) versus 2.6% for subcultured cells (passage 6-20, N = 5). DISCUSSION

In this report, we demonstrate the presence of insulin receptors on bovine endothelial cells obtained from the pulmonary and systemic circulation. Bovine aortic and pulmonary arterial cells had 2.5-fold more receptors/cell than the pulmonary venous cells, with each type of vascular endothelial cell retaining the specific number of receptors through 25 passages in

INSULIN AND BOVINE ENDOTHELIUM

~

~E I00:3

55

o .~ AORTA

I~="-'--'--Ti~'~ ~ ~ , . .

--.

• .l~ PULMONARY ARTERY

x

~ 80I1_ 0

~2 60.

\ XL,N

Z

orn 4 0

',',

z

~m ffl z

_~

o

i0-11

I0-I°

10.9

IO C

iO-T

10-6

10-5

[HORMONE] MOLES/LITER Fig. 5. Specificity of insuling binding to primary cultures of bovine aortic and pulmonary arterial endothelial cells, Cells were initially seeded at 40% of confluence, allowed to grow to confluent densities, then tested for insulin binding, In these experiments, maximal binding was 3.45%/106 cells for the pulmonary arterial cells and 2 . 5 % / 1 0 e cells for the aorta cells,

vitro. These data indicate that the insulin receptor is likely to be an intrinsic component of the plasma membrane of bovine endothelial cells. We have previously reported that primary cultures of human umbilical endothelial cells also have specific receptors for insulin. 2'3 Further studies with the human umbilical cells were limited by the small numbers of cells available and the expense and difficulty encountered in preparing primary cultures. In addition, we have been unable to subculture the human cells on a routine basis, and when subculture was successful the cells had prolonged doubling times and, most importantly, did not retain many properties of the primary cultures, including the presence of insulin receptors. In direct contrast to the human umbilical cells, the bovine endothelial cultures described in this report have many advantages_ for the detailed study of the interaction of insulin with endothelial cells. Bovine endothelial cells are easily established in cell culture, can be subcultured with rapid doubling times thereby yielding large numbers of cells, and they retain insulin receptors, morphology, defined growth pattern and a host of endothelial-specific and endothelial-associated traits despite repeated subculture in vitro. Senescence, which usually does not occur until >25 passages, is accompanied by a rapid loss of insulin receptors and other endothelial traits. Among the various bovine endothelial cells, there were differences in the ability to bind insulin. Within the pulmonary circulation, arterial endothelial cells bound 2.5-fold more insulin per cell than venous cells. This difference was observed in several distinct groups of passaged cells and was present after 25 passages in culture. Although the method of initially obtaining the arterial and venous cells differed (collagenase diges-

tion versus mechanical scraping) we do not believe that this would explain the arterial-venous difference in binding since we have demonstrated that all cultures are endothelial in nature and retain both endothelial traits and cell specific differences in insulin binding throughout serial subculture. However, it is important to note that the binding data have been expressed per cell. Although, at confluence, the mean cell radii and the concentration of cells per dish were similar in pulmonary arterial and venous cell cultures, it remains a possibility that the 2 types of endothelial cells manifest differences in cell surface area which could account for our findings. In addition to the A-V binding difference of the bovine pulmonary cells, we have previously noted that endothelial cells from human umbilical arteries also bind >2.5 times more insulin than comparable cells from the venous circulation. 3 Since the human cells could only be studied in short term, primary cultures, and since the umbilical arterial and venous cells were exposed to different ambient environments in the fetus, it was possible that the differences in arterialvenous insulin binding were influenced or induced by in vivo factors. However, when the data from the umbilical vessels are considered in the context of the bovine pulmonary studies, such explanations become less tenable. The adult pulmonary endothelium and the fetal umbilical endothelium are exposed to different environmental conditions relative to ambient oxygenation, insulin levels and nutrients. More importantly, the arterial venous differences in insulin binding persisted throughout repeated subculture of the bovine cells. These findings suggest that the concentration of insulin receptors at the endothelial surface may be controlled at a fundamental level of cellular differentiation. Obviously, careful study of other vascular beds is required to determine whether arterial-venous differences in endothelial cell binding of insulin are found in the entire circulation. As we have discussed elsewhere, 2'3 the potential role of the endothelial receptors for insulin could be: (1) altering intrinsic functions of the endothelial cells, (2) mediating the transport of insulin from the vascular compartment to certain tissue sites of action, and (3) serving as an additional "storage" compartment capable of modulating regional blood levels of insulin. All of these functions depend on the presence of insulin receptors in the intact capillary endothelium. Two recent radioautographic studies suggest that insulin receptors are indeed present in vivo on vascular endothelium. ~4"15 After pulse labeling with [125I]-iodoinsulin van Houten et al. demonstrated specific insulin binding to endothelial cells of rat brain capillaries with distinct regional variation of capillary binding of

56

PEACOCK, BAR, AND GOLDSMITH

insulin. Bergeron et al has also presented data suggesting the presence of "low affinity" receptors on the lumenal surface of hepatic endothelial cells. The quantitative aspects of insulin binding to the endothelium was difficult to determine from these 2 in vivo studies. There is little doubt that bolus injections of [1251]iodo-insulin are rapidly cleared from the bloodstream by hepatic and renal tissue. 16'17However, with chronic exposure to plasma insulin, as occurs in vivo, the role of the endothelial receptors for insulin may be more significant. Thus, if a segment of capillary endothelium had surface receptors for insulin with binding

properties similar to the insulin receptors of cultured bovine endothelial cells, the capillary endothelium would have the ~potential to bind an amount of insulin that is several fold higher than the concentration of insulin in capillary blood) If this relative extraction of insulin did occur at the capillary level, it would be relevant in the transfer and/or storage of insulin.. At present, such considerations must be considered highly speculative and await direct in vivo and in vitro testing. The endothelial cultures described in this study should provide reasonable in vitro systems to test these hypotheses.

REFERENCES

1. Gimbrone MA, Jr: Culture of vascular endothelium, in Spaet T (ed): Progress in Hemostasis and Thrombosis, vol 3. New York, Grune and Stratton, 1976, pp 1-28 2. Bar RS, Hoak JC, Peacock ML: Insulin receptors in human endothelial cells: identification and characterization. J Clin Endocrinol Metab 47:699-702, 1978 3. Bar RS, Peacock ML, Spanheimer R G , et al: Differential binding of insulin to human arterial and venous endothelial cells in primary culture. Diabetes 29:991-995, 1980 4. Goldsmith JC, Jafvert CT, Lollar P, et al: Prostacyclin release from cultured and e x vivo bovine vascular endothelium: studies with thrombin, arachidonic acid and ionophore A23187. Lab Invest (in press) 5. Bar RS, Rechler MM, Goldsmith JC, et al: Interactions of multiplication stimulating activity (MSA) with cultured bovine endothelium (submitted for publication) 6. Czervionke RL, Smith JB, Hoak JC, et al: Use of a radioimmunoassay to study thrombin-induced release of PGI2 from cultured endothelium. Thromb Res 14:781-786, 1979 7. Ryan JW, Chung A, Ammons C, et al: A simple radioassay for angiotensin-converting enzyme. Biochem J 167:781-786, 1979 8. Johnson A J, Kline DL, Alkjaersig N: Assay methods and standard preparations for plasmin, plasminogen and urokinase in purified systems, 1967-1968. Thromb Diath Haemorrh 21:259272, 1969 9. Silverstein RM: The determination of human plasminogen

using na-CBZ-L-Lysin p-Nitrophenyl ester as substrate. Anal Biochem 65:500-506, 1975 10. Jaffe EA, Nachman RL, Becket CG, et al: Culture of human endothelial cells derived from umbilical veins: Identification by morphologic and immunologic criteria. J Clin Invest 52:2745-2756, 1973 11. Hunter W M , Greenwood FC: Preparation of iodine-131 labeled human growth hormone of high specific activity. Nature (Lond.) 194:495496, 1962 12. Freychet P, Roth J, Neville DM, Jr: Insulin receptors in the liver: specific binding of (125I)-insulin to the plasma membrane and its relation to insulin bioactivity. Proc Natl Acad Sci USA 68:18331837, 1971 13. Kahn CR, Flier JS, Bar RS, et al: The syndromes of insulin resistance and acanthosis nigricans: insulin receptor disorders in man. N Engl J Med 294:739-745, 1976 14. Van Houten M, Posner BI: Insulin binds to brain blood vessels in vivo. Nature 282:623-625, 1979 15. Bergeron JJM, Sikstrom R, Hand AR, et al: Binding and uptake of ~25I-insulin into rat liver hepatocytes and endothelium: an in vivo radioantographic study. J Cell Biol 80:427~,43, 1979 16. Zeleznik AJ, Roth J: Demonstration of the insulin receptor in vivo in rabbits and its possible role as a reservoir for the plasma hormone. J Clin Invest 61:1363-1374, 1978 17. Sodoyez JC, Sodoyez-Goffaux FR, Moris YM: ~25I-insulin: kinetics of interaction with its receptors and rate of degradation in vivo. Am J Physiol 239:E3-E11, 1980