Diabetes Reseurch a& Clinical Practice. 16 (c; 1992 Elsevier
Science Publishers
( 1992) 75-84
B.V. All rights reserved 016%8227/92/$05.00
75
DIABET 00620
Evidence that the insulin resistance of pregnancy may not involve a post-receptor defect in human adipocytes C. Bruce”, J. Bumby”, D. Mangnallb “Deparrwer~t qf‘ Obstetrics arld Gwaeco1og.v. ‘Department
of Surgery.
Gerleral Hospital. Shefield
and R.B. Frasera
University of Shejfield, Clinical Sciences Centre, Northern S5 7AV, UK
(Received 21 April 1991) (Revision accepted 17 November 1991)
Summary
Insulin binding and the capacity of insulin to stimulate the conversion of glucose to carbon dioxide and lipid, and to activate the protein tyrosine kinase associated with the insulin receptor have been investigated in adipocytes isolated from pregnant and non-pregnant women. Insulin binding and the conversion of glucose to lipid were the same for both groups. However, conversion of glucose to CO, was higher in the non-pregnant group due to an elevated basal activity, and the increase produced by insulin was similar in both groups. The tyrosine kinase activity of the isolated receptor preparations was higher in the pregnant group due to an increase in the basal non-insulin dependent activity, and the increase produced by insulin was similar in both groups. These findings show the in vitro insulin responsiveness of isolated adipocytes is similar for both groups, and suggests that the in vivo insulin resistance of late pregnancy, as far as adipose tissue is concerned, is not due to any inherent defect in insulin action at the receptor or post-receptor level. In vivo insulin resistance may result from an increased level of circulating insulin antagonists. Key words: Insulin binding;
Pregnancy;
Adipocytes;
Introduction
Alterations in insulin sensitivity during pregnancy, particularly the third trimester, are well described [l-3] and have been reported to be associated with a post-receptor defect in insulin action [4,5]. Studies with human tissues are limited by Corresporldence to: Mrs C. Bruce, University Department of Obstetrics and Gynaecology, Clinical Sciences Centre, Northern General Hospital. Sheffield S5 7AU, UK.
Receptor
tyrosine kinase activity -~
ethical considerations and methodological difficulties. The use of the most readily accessible human tissue, blood cells, in studies of insulin sensitivity has been criticised since these may not respond in the same way as liver, muscle and adipose tissue [6]. Adipose tissue and skeletal muscle are considered to be more physiologically relevant, though less convenient to obtain. Although both are insulin sensitive, it cannot be assumed that they always show similar changes in responsiveness [ 71. Nonetheless, adipose tissue
76
has been widely used in metabolic studies in humans. The present studies have examined several in vitro biochemical features of adipocytes prepared from pregnant and non-pregnant women. Studies of insulin binding and insulin-dependent conversion of glucose to carbon dioxide, or incorporation into total lipid were undertaken to see to what extent the in vivo insulin-resistant state could be clearly demonstrated in vitro. Measurements were also made of the insulin-stimulated protein tyrosine kinase activity associated with the insulin receptor to see if there was any evidence for a receptor-linked defect in insulin responsiveness.
women (age range 19-44 years) undergoing gynaecological surgery (two tubal surgery, one laparotomy, 10 total hysterectomy and one subtotal hysterectomy) acted as control subjects. At the time of the study, 10 were in the proliferative phase of the menstrual cycle and four in the secretory phase as shown by histology and plasma progesterone levels. Tissue Adipose tissue was obtained at elective abdominal surgery for benign disease in the control subjects, and at elective caesarian section in the pregnant subjects. Tissue weighing 5-15 g was excised from the area above the rectus sheath and transported to the laboratory at 37°C in 0.90/, saline containing 5.0 mM glucose.
Materials and Methods Chemiculs Bovine albumin, Pepstatin A, leupeptin, phenylmethylsulphonyl fluoride (PMSF), bacitracin, benzamidine, N-acetylglucosamine, poly(glutamate/tyrosine, 4: 1) and wheat germ agglutininlinked Sepharose were from Sigma. Collagenase H (from Clostridiltm histol_l,ticum) was from Boehringer. Human insulin (100 IU/ml) was from Novo. D-[ U’“C]glucose (specific activity 270 mCi/mmol) and A 14 monoiodoinsulin (specific activity 2088 Ci/mmol) were obtained from the Radiochemical Centre, Amersham. Silicon oil (Silicon fluid Down Corning 200/5Ocs) was from BDH. Other reagents were the highest available quality. Informed consent and Ethical Committee upprovul This study was approved by the Ethical Committee of the Northern General Hospital, and was explained to all subjects who gave informed consent to the study. Patients Twenty pregnant women (age range 20-43 years) who delivered by caesarean section in weeks 38-40 of gestation participated in the study. Fourteen non-pregnant, non-diabetic, non-obese
Prepurution of udipocytes Adipocytes were isolated as described by Pedersen et al. [ 81. Aliquots of tissue (1 g) were suspended in 2 ml 10 mM Hepes pH 7.5 containing 25 mg/ml bovine albumin, 0.5 mg/ml collagenase H and 5.0 mM glucose. The procedure was performed aseptically to limit bacterial contamination. The tissue was digested at 37°C for 1 h and the cells washed four times in 10 mM Hepes pH 7.5 containing 50 mg/ml albumin and 0.5 mM glucose. The resulting cells were suspended to give a lipocrit of 0.3 and the cell diameter measured by a hanging drop technique, the image captured by videomicroscopy and measured with a Seescan Image analyser (Seescan Imaging Ltd., Cambridge, UK). We found that human adipose tissue was less readily digested than rat adipose tissue. Whereas rat adipose tissue could be digested to single cells without any major cell damage (as evidenced by the release of an oily fat) the same conditions applied to human tissue resulted in a lower yield of cells. We found human adipose tissue needed to be digested for at least 60 min to obtain a significant yield of cells, but incubating further did not increase the yield since the release of newly separated cells from the tissue was matched by an approximately equal rate of cell lysis. This rela-
77
tively low yield of cells limited the number of assays that could be performed for each cell preparation, and we therefore confined ourselves to studies of insulin concentrations within the normal physiological range (up to 5 nmol/l). Other workers have shown little effect at concentrations above this [9,10]. Methods for the assessment of viability and integrity of adipocytes are poorly documented, and the problem is largely ignored by most workers. We have attempted to assess viability by the standard vital stain approach (with trypan blue and erythrocin red) and found both stains were fully excluded from the preparations when freshly prepared, and also after 2-h incubation at 37°C. However, we also found that ethanoltreated adipocytes, although taking on a wrinkled appearance under the microscope, also excluded these dyes, whereas other tissue cells such as the HSN mouse fibrosarcoma cell line fully excluded the dyes in the live state but were readily stained after ethanol treatment. Acridine orange has also been used as a vital stain for adipocytes and yeast cells [ 11,121. In our experience, adipocyte preparations show green fluorescent nuclei when first prepared, suggesting that all the cells are fully viable [I I], and that the number of non-viable cells showing orange nuclei is of the order of 3-5”, ofthe total after 2 h at 37 “C. However, we have some reservations about the usefulness of this method since we have seen green fluorescing nuclei floating freely in preparations left overnight at 37°C which were clearly not associated with live viable cells. We therefore consider dye exclusion tests for viability to be problematic with adipocytes and we have adopted the approach used by Gliemann [ 131 for assessing cell integrity. This relies on the fact that, on cell lysis, the adipocytes release oil into the medium and this oil can be seen at the top of a lipocrit. We estimate from the oil released after 2-h incubation that lysis was at most S-100/, of the adipocyte preparation. We also examined dried smears of cells stained with toluidine blue, which clearly shows the cell nuclei, and found no evidence for anucleate cells at the end of 2-h incubation. This agrees with
Gliemann’s observations [ 131 that the number of cells with stainable nuclei fell only after about 3 h, concomitant with an increase in oil droplet formation. Insulin-binding studies Aliquots (200 ~1) of cell suspension (lipocrit 0.3) were preincubated with and without unlabelled insulin (O-5.0 nmol/l) for 45 min at 37°C. The assay conditions of 37’ C and pH 7.5 were chosen to minimise problems of internalisation and aggregation and to achieve a steady state [ 141. Insulin tracer (25 ~1 of 150 pmol/l) was added to give a final volume of 250 ~1. Non-specific binding was estimated by adding an excess (25 ~1) of stock insulin (100 IU/ml) as the unlabelled ligand. The reaction was stopped after 1 h by the addition of 8 ml of ice-cold 0.9p/0 NaCl and 1.2 ml silicon oil. The suspension was centrifuged at 1500 x g for 2 min and the cells removed by aspiration into a disposable pipette tip which was counted for activity. Results are expressed as bound counts/total counts. Conversion of D-[V4C]glucose to / i4C]C02 Aliquots (200 ~1) of cell suspension (cytocrit 0.3) were preincubated in plastic tubes at 37°C with and without insulin (O-500 pmol/l) for 45 min as described by Rodbell [ 151. D-[ u 14C]glUCOSe (0.5 PCi) was then added to give a final volume of 250 ~1. Initial experiments in which the glucose concentration of the incubation mixture was 5 mM resulted in barely detectable levels of radioactivity in the collected CO,, and the experiments were therefore performed at 0.5 mM glucose in order to ensure a sufficiently high specific activity. The cells were incubated in a cylindrical tube containing a conical tube into which was placed a filter. The tube was sealed with a rubber bung and incubated for a further 90 min. The reaction was stopped by injecting 300 ,LJ 4 N H,SO, into the suspension and 300 ~1 25% w/v phenethylamine in MeOH onto the filter to trap the CO,. The tubes were kept at 4°C for 1 h and the filters removed and counted. Blank tubes containing washing buffer instead of cells were
78 carried through the procedure to check for bacterial contamination. Results are expressed as pmol glucose converted/105 cells/90 min. Conversion I$ D-(Lii”C]glucose to [‘4CJ-labelled lipid After CO, collection was complete, 5 ml of an acidified mixture of 2_propanol/heptane (2-propanol/heptane/concentrated H,SO,, 40 : 10 : 1) was added to the incubation tube and vortexed for 10 min. Heptane (3.3 ml) and water (3.0 ml) were added and the mixing continued for a further 10 min. Aliquots (1.0 ml) of upper phase were removed and counted. Results are expressed as pmol glucose incorporated/lo5 cells/90 min. Preparation of insulin receptors Fresh adipose tissue (lo-20 g) was homogenised with two volumes of cold homogenisation buffer (20 mM Hepes, 8 mM EDTA, 160 mM sodium fluoride, 10 mM sodium pyrophosphate, 2 mM dichloroacetate, 0.2 mM sodium orthovanadate, 1 mM levamisole containing 2 mg/ml bacitracin and 1 mg/ml benzamidine at a final pH of 7.4. Immediately prior to use, PMSF was added to a final concentration of 2.5 mM and leupeptin to 1 PM and pepstatin A to 1 PM). The homogenate was centrifuged at 20,000 rpm in an SS34 rotor of a Sorval RCSB centrifuge for 10 min at 4°C. The resulting upper fatty layer was removed with a spatula and the pellets rehomogenised by hand into the supernatant above, and then recentrifuged at 50,000 rpm in a T865 rotor of a Sorval OTD 65 Ultracentrifuge for 90 min. The supernatants were removed and the pellets resuspended in 10 ml of solubilisation media (homogenisation buffer containing Triton X-100 at a final concentration of l.O?; v/v) and gently homogenised by hand and then stirred on ice for 60 min. The solubilised material was centrifuged at 20,000 rpm for 30 min in an SS34 rotor in the RCSB at 4°C and the supernatants filtered through a 0.45 pm filter prior to loading onto a 1 x 14 cm column of wheat germ agglutinin linked to Sepharose 4B. The column was pre-equilibrated in columnloading buffer (20 mM Hepes, 100 mM NaCI,
2.5 mM KCI, 1 mM CaC&, 0.1 mM sodium vanadate, 100 mM sodium fluoride, 10 mM sodium pyrophospate, 4 mM EDTA containing 1O”b glycerol, 0.5 P’, Triton X-100, 2 mM PMSF, 1 FM leupeptin and 1 PM pepstatin A at a final pH of 7.4) and the solubilised extract was loaded at 0.5 ml/min. The eluate from the column was monitored at 280 nm, and when the absorbance began to rise the eluate was collected until all the sample had been applied. At this point the collected eluate was recycled through the column at 50 pI/min (usually for about 20 h). The column was then eluted with wash buffer (20 mM Hepes, 100 mM NaCI, 2.5 mM KCl, 10% glycerol, 0.5q; Triton X-100, 0.1 mM sodium vanadate, 2 mM PMSF, 1 ,LLMleupeptin and 1 PM pepstatin A. pH 7.4) at 0.5 ml/min until all the non-bound material had eluted (as judged by the return to basal absorption at 280 nm). Receptors were eluted with wash buffer containing 0.3 M N-acetylglucosamine. Wash buffer was passed through the column at 0.5 ml/min until the UV absorbence of the eluate began to increase. At this point the flow was stopped and the column clamped for 30 min, after which time the buffer flow was restored and the UV-absorbing material eluted and collected. The eluate volume was noted, a sample taken for protein assay by UV absorption, and to the rest bovine albumin was added to a final concentration of 5 mg/ml. This was then concentrated over a XM 100 membrane (Amicon) to 0.5-1.0 ml final volume for use in the kinase assay. Measurement of the receptor protein tyrosine kinase ac tivitv Receptor protein tyrosine kinase activity was measured by a modification of the method of Khan et al. [ 161. The capacity to phosphorylate polyglutamate/tyrosine (4 : 1) was measured over the range O-65 pmol/l insulin. This wide range of insulin concentrations was chosen because maximal kinase stimulation occurs at insulin concentrations 20-lOO-fold greater than that giving maximal 2-deoxyglucose uptake in adipocytes [ 171 and concentrations as high as 0.1 pmol/l or
79
1 Almol/l are widely used [ 18-221. The receptor preparation was incubated with the appropriate insulin concentration for 1 h at room temperature. The poly glu/tyr was added and the mix incubated for a further 30 min at room temperature before the addition of a reaction mixture containing “Pgamma-ATP, MgCl?, MnSO,, Mops and sodium vanadate. At this point the tubes contained 10 ~11 of receptor preparation, 3.75 ~1 15 mg/ml poly glu/tyr (4 : l), 2.5 ~1 insulin, 3.125 ~1 400 mM Mops/O.8 mM sodium vanadate pH 7.0, 3.125 ~1 64 mM MgC1,/32 mM MnSO,, and 2.5 ~1 1.0 mM “P-ATP. The tubes were incubated at 30°C for 30 min and 20-~1 samples spotted onto strips of 3MM chromatography paper and chromatographed for 16 h in 109b TCA. The origins were cut out with scissors, washed twice in acetone, air-dried and counted in 2.5 ml of scintillant. Results were expressed as pmol ‘?P incorporated/min/mg protein receptor protein. Statistical
tests
Statistical comparisons of data were by Student’s unpaired t-test or by regression analysis for between group, and by paired t-test for within group comparisons, and a P value of 0.05 or less was taken to be statistically significant.
Results
0.10
1
0.1 -
om-
:
I o.ooI
z ;
0.04
0.02 -
02 l.666S-64
1
0.01 0.1 INSULIN nmol/l
10
Fig. 1, Binding of Al4 monoiodoinsulin to adipocytes from 17 pregnant women (0) and 1 I non-pregnant women (0). Results are mean + SEM.
affinities or receptor numbers. The interpretation of insulin radioreceptor assays of this kind has been reviewed by Nattrass and Dodds [ 141. Transformation of this data to give a Scatchard plot is commonly performed in order to estimate receptor numbers and binding affinities. The linear portion of the Scatchard plot from which these parameters are calculated is derived from samples with low radioactivity, conditions which are associated with greatest error. For the present data we consider this extrapolation would be unreliable, and the numbers so obtained would be of little physiological significance. 7wr 6
Adipocytes prepared from tissue from pregnant subjects had a mean diameter of 101 ,um (n = 20, SD = 7.1 pm, range = 89-116 pm) which was not statistically significantly different from the control value of 99pm (M = 14, SD = 4.8 pm, range = 90-108 pm). The binding of insulin to adipocytes from 16 pregnant and 11 non-pregnant women is shown in Fig. 1 in the form of displacement curves. The percentage bound for the pregnant group at tracer concentration was 0.083 %, and for the non-pregnant group 0.082%. At increasing amounts of non-labelled insulin there was no statistically significant difference between the groups, suggesting no significant differences in the binding, receptor
l.OOOE-63
4
0.000
T
O.OOl
0.100
o.olo l”6”Ll”
1.000
nIbloll,
Fig. 2. D-[U’“C]glucose conversion to [“C]CO,. Adipocytes were from 20 pregnant (0) and 14 non-pregnant (0) women, The conversion rate is pmol glucose converted/lO’ cells/90 min, and the results are mean f SEM. Regression analysis showed no statistically significant difference between the pregnant and non-pregnant groups.
80
1500
GLUCOSE INCORPORATED 1
T
INSULIN
I
nmolll
Fig. 3. Incorporation of u-[U ‘Clglucose into [ “Cllipid. Adipocytes were obtained from 20 pregnant (0) and 14 non-pregnant (0) women, and the incorporation rate is expressed as pmol glucose incorporated/IO5 cells/90 mm, and the results are expressed as mean f SEM.
The rates of oxidation of D-[U’JC]glucose to [ ‘“C]C02 are shown in Fig. 2. Although the basal glucose oxidation rate of the pregnant group (117 pmol COJ 10’ cells/90 min) was lower than that of the control group (399 pmol CO,/lO’ cells/90 min) this difference was not statistically significant. Furthermore, the capacity of the adi-
ACTIVITY
pocytes from either pregnant or non-pregnant tissue to respond to physiological concentrations of insulin was small, although these increases were statistically different from the basal values (P < 0.01). but not statistically different from each other. The non-pregnant group increased by only 10 ‘;, to 440 pmol CO,/ 10’ cells/90 min, and
5o [ 1
401
30
I-” I
20
INSULIN
nmol/l
Fig. 4. Receptor protein tyrosine kinase activity in wheat germ agglutinin preparations from six pregnant (0) and five nonpregnant women (0). Activities are expressed as pmol “P incorporated/min/mg protein, and results are mean + SEM. Regression analysis showed the pregnant group to be statistically significantly higher than the non-pregnant group (P < 0.03).
81
the pregnant group increased by 44 “//,to 2 10 pmol C0,/105 cells/90 min. This rate of oxidation is 5-50 times lower than that expected for adipocytes from rats [23,24]. The conversion of D-[Ui4C]glucose to total lipid is shown in Fig. 3. There was no statistically significant difference between the basal values (821 and 975 pmol/105 cells/90 min for the pregnant and non-pregnant groups respectively) or in the response to increasing insulin concentration (at maximal insulin concentration the pregnant group had increased by 24:~ to 1020 pmol/105 cells/90 min and the control group increased by 13% to 1100 pmol/105 cells/90 min). These increases are again statistically significantly different from the basal values (P < O.Ol), but not from each other. The conversion rates are roughly an order of magnitude smaller than those of rodent adipose tissue [25]. The protein tyrosine kinase activities of the isolated receptors are shown in Fig. 4. The basal activity of the non-pregnant group (9.1 pmol 32P/min/mg protein) was significantly less than that of the pregnant group (22.0 pmol 32P/min/mg protein, P = 0.01). Both groups showed modest, statistically significant increases in activity in response to increasing insulin concentrations. The non-pregnant group increased by 92% and the pregnant group by 7 1 y0 at 10 pmol/l insulin (for both, P < 0.05 compared to 0 insulin), but these increases are not statistically significantly different from each other. Thus, although the regression analysis showed the pregnant group to be statistically significantly higher than the non-pregnant group, this was due to an increase in the basal non-insulin-stimulated activity, and the response to insulin was the same in both groups.
Discussion There is a large volume of literature documenting the in vivo resistance to insulin action which accompanies the third trimester of pregnancy, but the underlying mechanisms are poorly understood. Increases in the circulating concentra-
tions of hormones antagonistic to insulin (such as placental lactogen, prolactin, cortisol and progesterone) are probably major causes of antagonism to insulin action. Exposure of adipocytes in vitro to hormones antagonistic to insulin can mimic the insulin-resistant state [21,26], leading Ryan and Enns [26] to postulate a postbinding defect in insulin action during pregnancy, a conclusion which had also been reached by Puavilai et al. [ 51. Alterations in insulin binding to adipocytes in pregnancy have also been reported, but the results are conflicting. In humans, decreases in binding have been reported [ 271, which contrasts with the present results showing no changes in binding, whilst in rats binding increases during pregnancy and is then reduced just prior to parturition [25,28]. Several studies have documented a relationship between adipocyte size and glucose oxidation and lipid metabolism, although there is disagreement as to whether these are positively or negatively correlated [23,24,29-331. Furthermore, Andersen and Kuhl [32] reported an increased adipocyte size in pregnancy, but concluded nonetheless that it was unlikely that adipocytes contributed in any major way to the in vivo insulin resistance of pregnancy. The present studies are not entirely in accord with these observations, although we agree with the general conclusion of Andersen and Kuhl [ 321 that a major role for adipose tissue in the in vivo insulin resistance of pregnancy is unlikely. Firstly, we find that the metabolic activity of human adipose tissue is very low, certainly when compared to the more widely used rodent adipose tissue, with which most of the experimental methodologies were developed. This is not an unexpected finding since human adipose tissue has been considered to function much more as a simple store for triglyceride synthesised by the liver and transported to the adipose tissue as very low density lipoprotein, rather than as a site for active de novo triglyceride synthesis as occurs in rats and mice. It does however mean that the contribution to glucose uptake made by the adipose tissue following a glucose load may be proportionately less in the human than in rodents, and thus as a tissue
82
for the study of insulin resistance it may be less relevant in humans. An important consequence of this relatively inert metabolism is the difficulty of accurate metabolic measurements. With such a low activity, the methods which are routinely employed are being used at the limits of their sensitivity. We found the very small differences provoked by insulin with both the non-pregnant and pregnant tissues difficult to measure. It may well be that, in humans, skeletal muscle would prove to be a much more useful tissue with which to address the questions of the site of insulin resistance, and at least one study [ 341 has shown that the inherent difficulties of obtaining enough tissue may not be insurmountable. Secondly, we find that for all the parameters we examined, the cells from pregnant women were as responsive to insulin as the control cells from non-pregnant women. Insulin binding and the conversion of labelled glucose to lipid were not statistically significantly different in the two groups. The rates of conversion of glucose to CO, were higher in the non-pregnant group than in the pregnant group. However, this was due to an increased oxidation rate in the absence of insulin, and the increase due to insulin was the same in both groups. It is perhaps relevant that Andersen et al. [34] found a reduction in the activity of pyruvate kinase in adipose tissue from pregnant women compared to non-pregnant women, and inferred from this a reduced glycolytic flux in pregnancy. Any reduction in glycolysis may also constrain the conversion of glucose to CO,, as observed in the present experiments. The enzyme data of Andersen et al. [34] further suggest that, in adipose tissue, the flux-limiting enzyme of glycolysis may be pyruvate kinase rather than phosphofructokinase or hexokinase, and that the reverse is true in muscle. This suggests that in the pregnant state a reduction in the activity of pyruvate kinase may ensure an adequate supply of triose phosphates and glycerol phosphate for esterification to triglyceride under conditions where insulin-stimulated glucose uptake in vivo may be restricted. This inferred requirement for a sustained capacity for esterification may be
related to the substrate cycling between triglycerides and fatty acids which occurs in human adipocytes [33]. Although the basal activity of the tyrosine protein kinase was higher in the pregnant group than the control group, a similar increase in activity in response to insulin was observed for both groups, and there was no evidence for any defect in insulin-dependent kinase activity in the pregnant group compared to the control nonpregnant group. The reason for the difference in basal rather than insulin-stimulated activity between the two groups is not clear. Receptor preparations produced by wheat germ agglutinin chromatography are known to be heterogeneous, with insulin receptor representing less than 1O. of the total [22]. However, at present we have no evidence as to the nature of the non-insulindependent tyrosine protein kinase associated with the preparations from the pregnant group, although an increase in some other receptor-associated kinase is the most exciting possibility. In none of our studies did we see clear evidence for an in vitro insulin resistance associated with the pregnant state. In contrast to Andersen and Kuhl [ 321 we found no differences in size between adipocytes from pregnant or non-pregnant subjects, so considerations of the confused relationship between cell size and metabolism become irrelevant in the context of the present findings. It is perhaps worth noting that although Ryan and Enns [26] showed freshly isolated adipocytes from pregnant rats to have a reduced capacity for 3-methylglucose transport compared to controls, the capacity to respond to insulin was still maintained. On a percentage basis the increase with cells from pregnant rats was the same as that for controls. Thus, even with rat adipose tissue, the evidence for an in vitro difference in insulin sensitivity may be questioned. A third consideration which arises from these observations is why a well described and documented resistance to insulin action in vivo may not be manifest once the tissue is removed from the body and examined in vitro. We consider the simplest explanation is that the resistance is not an inherent feature of the tissue in the pregnant
83 state, but is imposed on the tissue by the presence in vivo of high concentrations of circulating hormones with actions antagonistic to insulin. In this state a post-receptor defect in insulin action would not be anticipated, and there is no evidence from our studies that any defect per se in insulin sensitivity, post-receptor or otherwise, occurs in human adipose tissue in the pregnant state.
10
II
12
13
Acknowledgements This work was supported by a grant from Birthright, the research charity of the Royal College of Obstetricians and Gynaecologists. We are grateful to Professor I. Dunsmore, Department of Probability and Statistics, for advice on the statistical analysis.
14
References
17
Kalkhoff, R., Schalch, D.S.. Walker, J.L., Beck, P., Kipnis. D.N. and Daughaday, W.H. (1964) Diabetogenic factors associated with pregnancy. Trans. Assoc. Am. Physicians 77, 270-280. Spellacy, W.N. and Goetz, F.C. (1963) Plasma insulin in normal late pregnancy. N. Engl. J. Med. 268, 988-991. Bleicher, S.J., O’Sullivan, J.B. and Freinkel, N. (1964) Carbohydrate metabolism in pregnancy. N. Engl. J. Med. 27 1. 866-872. Ryan, E.A., O’Sullivan, M.J. and Skyler. J.S. (1985) Insulin action during pregnancy. Studies with the euglycemic clamp technique. Diabetes 34. 380-389. Puavilai, G., Drobny, E.C.. Domont, L.A. and Baumann, G. (1982) Insulin receptors and insulin resistance in human pregnancy: evidence for a post receptor defect in insulin action. J. Clin. Endocrin. Metab. 54, 247-253. Taylor, R.. Proctor, S.J., James, O., Clark, F. and Alberti, K.G.M.M. (1984) The relationship between human adipocyte and monocyte insulin binding. Clin. Sci. 67, 139-142. Taylor, R. (1986) Insulin receptors and the clinician. Br. Med. J. 292. 919-922. Pedersen. 0.. Hjollund, E., Beck-Nielsen, H.O., L.indsilov, 0.. Sonne, 0. and Gliemann, J. (1981) Insulin receptor binding and receptor mediated insulin dein human adipocytes. Diabetologia 20, gradation 636-641. Hjollund. E.. Pedersen, O., Espersen. T. and Klebe, J.G. (1986) Impaired insulin receptor binding and postbinding
15
16
18
19
20
21
22
23
24
25
defects of adipocytes from normal and diabetic pregnant women. Diabetes 35, 598-603. Pedersen, O., Hjollund, E. and Lindskov, H.O. (1982) Insulin binding and action on fat cells from young healthy females and males. J. Am. Physiol. 243. E158-E167. Larch. E. and Rentsch, G. (1969) A simple method for staining and counting isolated adipose tissue fat cells. Diabetologia 5, 356-357. Schwartz, D.S., Larsh, H.W. and Bartels, P.A. (1977) Enumerative fluorescent vital staining of live and dead pathogenic yeast cells. Stain Technol. 52. 203-210. Gliemann. J. (1967) Assay of insulin-like activity by the isolated fat cell method. 1. Factors influencing the response to crystalline insulin. Diabetologia 3, 382-388. Nattrass, M. and Dodds, K.E. (1987) Interpretation of insulin radioreceptor assays. Ann. Clin. Biochem. 24, 13-21. Rodbell, M. (1964) Metabolism of isolated fat cells. I. Effects of hormones on glucose metabolism and lipolysis. J. Biol. Chem. 239, 375-380. Khan, M.N., Baquiran, G., Brule, C., Burgess, J., Foster, B., Bergeron, J.J.M. and Posner, B.I. (1989) Internalisation and activation of the rat liver insulin receptor kinase in vivo. J. Biol. Chem. 264, 12931-12940. Klein, H.H., Ciaraldi, T.P., Freindenberg, G.R. and Olefsky, J.M. (1987) Adenosine modulates insulin activation of insulin receptor kinase in intact rat adipocytes. Endocrinology 120, 2339-2345. Rosen, O.M., Herrara, R., Olowe, Y., Petruzzelli, L.M. and Cobb, M.H. (1983) Phosphorylation activates the insulin receptor tyrosine protein kinase. Proc. Nat]. Acad. Sci. 80, 3237-3240. Zick, Y., Kasuga. M., Kahn, C.R. and Roth, J. (1983) Characterisation of insulin-mediated phosphorylation of the insulin receptor in a cell-free system. J. Biol. Chem. 258,75-80. Ridge, K.D.. Hofmann, K. and Finn, F.M. (1988) ATP sensitises the insulin receptor to insulin. Proc. Natl. Acad. Sci. 85, 9489-9493. Haring, H., Kirsch, D., Obermaier, B., Ermel, B. and Machicao, F. (1986) Decreased tyrosine kinase activity of insulin receptor isolated from rat adipocytes rendered insulin-resistant by catecholamine treatment in vitro. Biochem. J. 234, 59-66. Fujita-Yamaguchi, Y., Choi, S., Sakamoto, Y. and Itakura, K. (1983) Purification of insulin receptor with full binding activity. J. Biol. Chem. 258, 5045-5049. Craig, B.W. and Treadway, J. (1986) Glucose uptake and oxidation in fat cells of trained and sedentary pregnant rats. J. Appl. Physiol. 60, 1704-1709. Craig, B.W., Garthwaite, S.M. and Holloszy, J.O. (1987) Adipocyte insulin resistance: effects of aging, obesity, exercise and food restriction. J. Appl. Physiol. 62,95-100. Flint, D.J.. Clegg, R.A. and Vernon, R.G. (1980) Regulation of adipocyte insulin receptor number and metabo-
84
26
27
28
29
30
lism during late pregnancy. Mol. Cell. Endocrinol. 20, 101-111. Ryan, E.A. and Enns, L. (1988) Role of gestational hormones in the induction ofinsulin resistance. J. Clin. Endocrinol. Metab. 67, 341-347. Pagano, G., Cassader, M., Massobrio, M., Bozzo, C., Trossarelli, G.F., Menato, G. and Lenti, G. (1980) Insulin binding to human adipocytes during late pregnancy in healthy, obese and diabetic state. Horm. Metab. Res. 12, 177-181. Flint, D.J., Sinnett-Smith, P.A., Clegg, R.A. and Vernon, R.G. (1979) Role of insulin receptors in the changing metabolism of adipose tissue during pregnancy and lactation in the rat. Biochem. J. 182, 421-427. Foley, J.E., Laursen, A.L., Sonne, 0. and Gliemann, J. (1980) Insulin binding and hexose transport in rat adipocytes. Diabetologia 19, 234-241. Vinten. J. and Galbo, H. (1983) Effect ofphysical training
31
32
33
34
on transport and metabolism of glucose in adipocytes. Am. J. Physiol. 244, E129-134. Savard, R., Deshaies, Y., Despres, J.P., Marcotte, M., Bukowiecki, L., Allard, C. and Bouchard, C. (1984) Lipogenesis and lipoprotein lipase in human adipose tissue reproducibility ofmeasurement and relationships with fat cell size. Can. J. Physiol. Pharmacol. 62, 1448-1452. Andersen, 0. and Kuhl, C. (1988) Adipocyte insulin receptor binding and lipogenesis at term in normal pregnancy. Eur. J. Clin. Invest. 18, 575-581. Hammond, V.A. and Johnston, D.G. (1987) Substrate cycling between triglyceride and fatty acid in human adipocytes. Metabolism 36, 308-313. Andersen. O., Falholt, K. and Kuhl, C. (1989) Activity of enzymes of glucose and triglyceride metabolism in adipose and muscle tissue from normal pregnant women at term. Diabetic Med. 6, 131-136.
Science Publishers
( 1992) 75-84
B.V. All rights reserved 016%8227/92/$05.00
75
DIABET 00620
Evidence that the insulin resistance of pregnancy may not involve a post-receptor defect in human adipocytes C. Bruce”, J. Bumby”, D. Mangnallb “Deparrwer~t qf‘ Obstetrics arld Gwaeco1og.v. ‘Department
of Surgery.
Gerleral Hospital. Shefield
and R.B. Frasera
University of Shejfield, Clinical Sciences Centre, Northern S5 7AV, UK
(Received 21 April 1991) (Revision accepted 17 November 1991)
Summary
Insulin binding and the capacity of insulin to stimulate the conversion of glucose to carbon dioxide and lipid, and to activate the protein tyrosine kinase associated with the insulin receptor have been investigated in adipocytes isolated from pregnant and non-pregnant women. Insulin binding and the conversion of glucose to lipid were the same for both groups. However, conversion of glucose to CO, was higher in the non-pregnant group due to an elevated basal activity, and the increase produced by insulin was similar in both groups. The tyrosine kinase activity of the isolated receptor preparations was higher in the pregnant group due to an increase in the basal non-insulin dependent activity, and the increase produced by insulin was similar in both groups. These findings show the in vitro insulin responsiveness of isolated adipocytes is similar for both groups, and suggests that the in vivo insulin resistance of late pregnancy, as far as adipose tissue is concerned, is not due to any inherent defect in insulin action at the receptor or post-receptor level. In vivo insulin resistance may result from an increased level of circulating insulin antagonists. Key words: Insulin binding;
Pregnancy;
Adipocytes;
Introduction
Alterations in insulin sensitivity during pregnancy, particularly the third trimester, are well described [l-3] and have been reported to be associated with a post-receptor defect in insulin action [4,5]. Studies with human tissues are limited by Corresporldence to: Mrs C. Bruce, University Department of Obstetrics and Gynaecology, Clinical Sciences Centre, Northern General Hospital. Sheffield S5 7AU, UK.
Receptor
tyrosine kinase activity -~
ethical considerations and methodological difficulties. The use of the most readily accessible human tissue, blood cells, in studies of insulin sensitivity has been criticised since these may not respond in the same way as liver, muscle and adipose tissue [6]. Adipose tissue and skeletal muscle are considered to be more physiologically relevant, though less convenient to obtain. Although both are insulin sensitive, it cannot be assumed that they always show similar changes in responsiveness [ 71. Nonetheless, adipose tissue
76
has been widely used in metabolic studies in humans. The present studies have examined several in vitro biochemical features of adipocytes prepared from pregnant and non-pregnant women. Studies of insulin binding and insulin-dependent conversion of glucose to carbon dioxide, or incorporation into total lipid were undertaken to see to what extent the in vivo insulin-resistant state could be clearly demonstrated in vitro. Measurements were also made of the insulin-stimulated protein tyrosine kinase activity associated with the insulin receptor to see if there was any evidence for a receptor-linked defect in insulin responsiveness.
women (age range 19-44 years) undergoing gynaecological surgery (two tubal surgery, one laparotomy, 10 total hysterectomy and one subtotal hysterectomy) acted as control subjects. At the time of the study, 10 were in the proliferative phase of the menstrual cycle and four in the secretory phase as shown by histology and plasma progesterone levels. Tissue Adipose tissue was obtained at elective abdominal surgery for benign disease in the control subjects, and at elective caesarian section in the pregnant subjects. Tissue weighing 5-15 g was excised from the area above the rectus sheath and transported to the laboratory at 37°C in 0.90/, saline containing 5.0 mM glucose.
Materials and Methods Chemiculs Bovine albumin, Pepstatin A, leupeptin, phenylmethylsulphonyl fluoride (PMSF), bacitracin, benzamidine, N-acetylglucosamine, poly(glutamate/tyrosine, 4: 1) and wheat germ agglutininlinked Sepharose were from Sigma. Collagenase H (from Clostridiltm histol_l,ticum) was from Boehringer. Human insulin (100 IU/ml) was from Novo. D-[ U’“C]glucose (specific activity 270 mCi/mmol) and A 14 monoiodoinsulin (specific activity 2088 Ci/mmol) were obtained from the Radiochemical Centre, Amersham. Silicon oil (Silicon fluid Down Corning 200/5Ocs) was from BDH. Other reagents were the highest available quality. Informed consent and Ethical Committee upprovul This study was approved by the Ethical Committee of the Northern General Hospital, and was explained to all subjects who gave informed consent to the study. Patients Twenty pregnant women (age range 20-43 years) who delivered by caesarean section in weeks 38-40 of gestation participated in the study. Fourteen non-pregnant, non-diabetic, non-obese
Prepurution of udipocytes Adipocytes were isolated as described by Pedersen et al. [ 81. Aliquots of tissue (1 g) were suspended in 2 ml 10 mM Hepes pH 7.5 containing 25 mg/ml bovine albumin, 0.5 mg/ml collagenase H and 5.0 mM glucose. The procedure was performed aseptically to limit bacterial contamination. The tissue was digested at 37°C for 1 h and the cells washed four times in 10 mM Hepes pH 7.5 containing 50 mg/ml albumin and 0.5 mM glucose. The resulting cells were suspended to give a lipocrit of 0.3 and the cell diameter measured by a hanging drop technique, the image captured by videomicroscopy and measured with a Seescan Image analyser (Seescan Imaging Ltd., Cambridge, UK). We found that human adipose tissue was less readily digested than rat adipose tissue. Whereas rat adipose tissue could be digested to single cells without any major cell damage (as evidenced by the release of an oily fat) the same conditions applied to human tissue resulted in a lower yield of cells. We found human adipose tissue needed to be digested for at least 60 min to obtain a significant yield of cells, but incubating further did not increase the yield since the release of newly separated cells from the tissue was matched by an approximately equal rate of cell lysis. This rela-
77
tively low yield of cells limited the number of assays that could be performed for each cell preparation, and we therefore confined ourselves to studies of insulin concentrations within the normal physiological range (up to 5 nmol/l). Other workers have shown little effect at concentrations above this [9,10]. Methods for the assessment of viability and integrity of adipocytes are poorly documented, and the problem is largely ignored by most workers. We have attempted to assess viability by the standard vital stain approach (with trypan blue and erythrocin red) and found both stains were fully excluded from the preparations when freshly prepared, and also after 2-h incubation at 37°C. However, we also found that ethanoltreated adipocytes, although taking on a wrinkled appearance under the microscope, also excluded these dyes, whereas other tissue cells such as the HSN mouse fibrosarcoma cell line fully excluded the dyes in the live state but were readily stained after ethanol treatment. Acridine orange has also been used as a vital stain for adipocytes and yeast cells [ 11,121. In our experience, adipocyte preparations show green fluorescent nuclei when first prepared, suggesting that all the cells are fully viable [I I], and that the number of non-viable cells showing orange nuclei is of the order of 3-5”, ofthe total after 2 h at 37 “C. However, we have some reservations about the usefulness of this method since we have seen green fluorescing nuclei floating freely in preparations left overnight at 37°C which were clearly not associated with live viable cells. We therefore consider dye exclusion tests for viability to be problematic with adipocytes and we have adopted the approach used by Gliemann [ 131 for assessing cell integrity. This relies on the fact that, on cell lysis, the adipocytes release oil into the medium and this oil can be seen at the top of a lipocrit. We estimate from the oil released after 2-h incubation that lysis was at most S-100/, of the adipocyte preparation. We also examined dried smears of cells stained with toluidine blue, which clearly shows the cell nuclei, and found no evidence for anucleate cells at the end of 2-h incubation. This agrees with
Gliemann’s observations [ 131 that the number of cells with stainable nuclei fell only after about 3 h, concomitant with an increase in oil droplet formation. Insulin-binding studies Aliquots (200 ~1) of cell suspension (lipocrit 0.3) were preincubated with and without unlabelled insulin (O-5.0 nmol/l) for 45 min at 37°C. The assay conditions of 37’ C and pH 7.5 were chosen to minimise problems of internalisation and aggregation and to achieve a steady state [ 141. Insulin tracer (25 ~1 of 150 pmol/l) was added to give a final volume of 250 ~1. Non-specific binding was estimated by adding an excess (25 ~1) of stock insulin (100 IU/ml) as the unlabelled ligand. The reaction was stopped after 1 h by the addition of 8 ml of ice-cold 0.9p/0 NaCl and 1.2 ml silicon oil. The suspension was centrifuged at 1500 x g for 2 min and the cells removed by aspiration into a disposable pipette tip which was counted for activity. Results are expressed as bound counts/total counts. Conversion of D-[V4C]glucose to / i4C]C02 Aliquots (200 ~1) of cell suspension (cytocrit 0.3) were preincubated in plastic tubes at 37°C with and without insulin (O-500 pmol/l) for 45 min as described by Rodbell [ 151. D-[ u 14C]glUCOSe (0.5 PCi) was then added to give a final volume of 250 ~1. Initial experiments in which the glucose concentration of the incubation mixture was 5 mM resulted in barely detectable levels of radioactivity in the collected CO,, and the experiments were therefore performed at 0.5 mM glucose in order to ensure a sufficiently high specific activity. The cells were incubated in a cylindrical tube containing a conical tube into which was placed a filter. The tube was sealed with a rubber bung and incubated for a further 90 min. The reaction was stopped by injecting 300 ,LJ 4 N H,SO, into the suspension and 300 ~1 25% w/v phenethylamine in MeOH onto the filter to trap the CO,. The tubes were kept at 4°C for 1 h and the filters removed and counted. Blank tubes containing washing buffer instead of cells were
78 carried through the procedure to check for bacterial contamination. Results are expressed as pmol glucose converted/105 cells/90 min. Conversion I$ D-(Lii”C]glucose to [‘4CJ-labelled lipid After CO, collection was complete, 5 ml of an acidified mixture of 2_propanol/heptane (2-propanol/heptane/concentrated H,SO,, 40 : 10 : 1) was added to the incubation tube and vortexed for 10 min. Heptane (3.3 ml) and water (3.0 ml) were added and the mixing continued for a further 10 min. Aliquots (1.0 ml) of upper phase were removed and counted. Results are expressed as pmol glucose incorporated/lo5 cells/90 min. Preparation of insulin receptors Fresh adipose tissue (lo-20 g) was homogenised with two volumes of cold homogenisation buffer (20 mM Hepes, 8 mM EDTA, 160 mM sodium fluoride, 10 mM sodium pyrophosphate, 2 mM dichloroacetate, 0.2 mM sodium orthovanadate, 1 mM levamisole containing 2 mg/ml bacitracin and 1 mg/ml benzamidine at a final pH of 7.4. Immediately prior to use, PMSF was added to a final concentration of 2.5 mM and leupeptin to 1 PM and pepstatin A to 1 PM). The homogenate was centrifuged at 20,000 rpm in an SS34 rotor of a Sorval RCSB centrifuge for 10 min at 4°C. The resulting upper fatty layer was removed with a spatula and the pellets rehomogenised by hand into the supernatant above, and then recentrifuged at 50,000 rpm in a T865 rotor of a Sorval OTD 65 Ultracentrifuge for 90 min. The supernatants were removed and the pellets resuspended in 10 ml of solubilisation media (homogenisation buffer containing Triton X-100 at a final concentration of l.O?; v/v) and gently homogenised by hand and then stirred on ice for 60 min. The solubilised material was centrifuged at 20,000 rpm for 30 min in an SS34 rotor in the RCSB at 4°C and the supernatants filtered through a 0.45 pm filter prior to loading onto a 1 x 14 cm column of wheat germ agglutinin linked to Sepharose 4B. The column was pre-equilibrated in columnloading buffer (20 mM Hepes, 100 mM NaCI,
2.5 mM KCI, 1 mM CaC&, 0.1 mM sodium vanadate, 100 mM sodium fluoride, 10 mM sodium pyrophospate, 4 mM EDTA containing 1O”b glycerol, 0.5 P’, Triton X-100, 2 mM PMSF, 1 FM leupeptin and 1 PM pepstatin A at a final pH of 7.4) and the solubilised extract was loaded at 0.5 ml/min. The eluate from the column was monitored at 280 nm, and when the absorbance began to rise the eluate was collected until all the sample had been applied. At this point the collected eluate was recycled through the column at 50 pI/min (usually for about 20 h). The column was then eluted with wash buffer (20 mM Hepes, 100 mM NaCI, 2.5 mM KCl, 10% glycerol, 0.5q; Triton X-100, 0.1 mM sodium vanadate, 2 mM PMSF, 1 ,LLMleupeptin and 1 PM pepstatin A. pH 7.4) at 0.5 ml/min until all the non-bound material had eluted (as judged by the return to basal absorption at 280 nm). Receptors were eluted with wash buffer containing 0.3 M N-acetylglucosamine. Wash buffer was passed through the column at 0.5 ml/min until the UV absorbence of the eluate began to increase. At this point the flow was stopped and the column clamped for 30 min, after which time the buffer flow was restored and the UV-absorbing material eluted and collected. The eluate volume was noted, a sample taken for protein assay by UV absorption, and to the rest bovine albumin was added to a final concentration of 5 mg/ml. This was then concentrated over a XM 100 membrane (Amicon) to 0.5-1.0 ml final volume for use in the kinase assay. Measurement of the receptor protein tyrosine kinase ac tivitv Receptor protein tyrosine kinase activity was measured by a modification of the method of Khan et al. [ 161. The capacity to phosphorylate polyglutamate/tyrosine (4 : 1) was measured over the range O-65 pmol/l insulin. This wide range of insulin concentrations was chosen because maximal kinase stimulation occurs at insulin concentrations 20-lOO-fold greater than that giving maximal 2-deoxyglucose uptake in adipocytes [ 171 and concentrations as high as 0.1 pmol/l or
79
1 Almol/l are widely used [ 18-221. The receptor preparation was incubated with the appropriate insulin concentration for 1 h at room temperature. The poly glu/tyr was added and the mix incubated for a further 30 min at room temperature before the addition of a reaction mixture containing “Pgamma-ATP, MgCl?, MnSO,, Mops and sodium vanadate. At this point the tubes contained 10 ~11 of receptor preparation, 3.75 ~1 15 mg/ml poly glu/tyr (4 : l), 2.5 ~1 insulin, 3.125 ~1 400 mM Mops/O.8 mM sodium vanadate pH 7.0, 3.125 ~1 64 mM MgC1,/32 mM MnSO,, and 2.5 ~1 1.0 mM “P-ATP. The tubes were incubated at 30°C for 30 min and 20-~1 samples spotted onto strips of 3MM chromatography paper and chromatographed for 16 h in 109b TCA. The origins were cut out with scissors, washed twice in acetone, air-dried and counted in 2.5 ml of scintillant. Results were expressed as pmol ‘?P incorporated/min/mg protein receptor protein. Statistical
tests
Statistical comparisons of data were by Student’s unpaired t-test or by regression analysis for between group, and by paired t-test for within group comparisons, and a P value of 0.05 or less was taken to be statistically significant.
Results
0.10
1
0.1 -
om-
:
I o.ooI
z ;
0.04
0.02 -
02 l.666S-64
1
0.01 0.1 INSULIN nmol/l
10
Fig. 1, Binding of Al4 monoiodoinsulin to adipocytes from 17 pregnant women (0) and 1 I non-pregnant women (0). Results are mean + SEM.
affinities or receptor numbers. The interpretation of insulin radioreceptor assays of this kind has been reviewed by Nattrass and Dodds [ 141. Transformation of this data to give a Scatchard plot is commonly performed in order to estimate receptor numbers and binding affinities. The linear portion of the Scatchard plot from which these parameters are calculated is derived from samples with low radioactivity, conditions which are associated with greatest error. For the present data we consider this extrapolation would be unreliable, and the numbers so obtained would be of little physiological significance. 7wr 6
Adipocytes prepared from tissue from pregnant subjects had a mean diameter of 101 ,um (n = 20, SD = 7.1 pm, range = 89-116 pm) which was not statistically significantly different from the control value of 99pm (M = 14, SD = 4.8 pm, range = 90-108 pm). The binding of insulin to adipocytes from 16 pregnant and 11 non-pregnant women is shown in Fig. 1 in the form of displacement curves. The percentage bound for the pregnant group at tracer concentration was 0.083 %, and for the non-pregnant group 0.082%. At increasing amounts of non-labelled insulin there was no statistically significant difference between the groups, suggesting no significant differences in the binding, receptor
l.OOOE-63
4
0.000
T
O.OOl
0.100
o.olo l”6”Ll”
1.000
nIbloll,
Fig. 2. D-[U’“C]glucose conversion to [“C]CO,. Adipocytes were from 20 pregnant (0) and 14 non-pregnant (0) women, The conversion rate is pmol glucose converted/lO’ cells/90 min, and the results are mean f SEM. Regression analysis showed no statistically significant difference between the pregnant and non-pregnant groups.
80
1500
GLUCOSE INCORPORATED 1
T
INSULIN
I
nmolll
Fig. 3. Incorporation of u-[U ‘Clglucose into [ “Cllipid. Adipocytes were obtained from 20 pregnant (0) and 14 non-pregnant (0) women, and the incorporation rate is expressed as pmol glucose incorporated/IO5 cells/90 mm, and the results are expressed as mean f SEM.
The rates of oxidation of D-[U’JC]glucose to [ ‘“C]C02 are shown in Fig. 2. Although the basal glucose oxidation rate of the pregnant group (117 pmol COJ 10’ cells/90 min) was lower than that of the control group (399 pmol CO,/lO’ cells/90 min) this difference was not statistically significant. Furthermore, the capacity of the adi-
ACTIVITY
pocytes from either pregnant or non-pregnant tissue to respond to physiological concentrations of insulin was small, although these increases were statistically different from the basal values (P < 0.01). but not statistically different from each other. The non-pregnant group increased by only 10 ‘;, to 440 pmol CO,/ 10’ cells/90 min, and
5o [ 1
401
30
I-” I
20
INSULIN
nmol/l
Fig. 4. Receptor protein tyrosine kinase activity in wheat germ agglutinin preparations from six pregnant (0) and five nonpregnant women (0). Activities are expressed as pmol “P incorporated/min/mg protein, and results are mean + SEM. Regression analysis showed the pregnant group to be statistically significantly higher than the non-pregnant group (P < 0.03).
81
the pregnant group increased by 44 “//,to 2 10 pmol C0,/105 cells/90 min. This rate of oxidation is 5-50 times lower than that expected for adipocytes from rats [23,24]. The conversion of D-[Ui4C]glucose to total lipid is shown in Fig. 3. There was no statistically significant difference between the basal values (821 and 975 pmol/105 cells/90 min for the pregnant and non-pregnant groups respectively) or in the response to increasing insulin concentration (at maximal insulin concentration the pregnant group had increased by 24:~ to 1020 pmol/105 cells/90 min and the control group increased by 13% to 1100 pmol/105 cells/90 min). These increases are again statistically significantly different from the basal values (P < O.Ol), but not from each other. The conversion rates are roughly an order of magnitude smaller than those of rodent adipose tissue [25]. The protein tyrosine kinase activities of the isolated receptors are shown in Fig. 4. The basal activity of the non-pregnant group (9.1 pmol 32P/min/mg protein) was significantly less than that of the pregnant group (22.0 pmol 32P/min/mg protein, P = 0.01). Both groups showed modest, statistically significant increases in activity in response to increasing insulin concentrations. The non-pregnant group increased by 92% and the pregnant group by 7 1 y0 at 10 pmol/l insulin (for both, P < 0.05 compared to 0 insulin), but these increases are not statistically significantly different from each other. Thus, although the regression analysis showed the pregnant group to be statistically significantly higher than the non-pregnant group, this was due to an increase in the basal non-insulin-stimulated activity, and the response to insulin was the same in both groups.
Discussion There is a large volume of literature documenting the in vivo resistance to insulin action which accompanies the third trimester of pregnancy, but the underlying mechanisms are poorly understood. Increases in the circulating concentra-
tions of hormones antagonistic to insulin (such as placental lactogen, prolactin, cortisol and progesterone) are probably major causes of antagonism to insulin action. Exposure of adipocytes in vitro to hormones antagonistic to insulin can mimic the insulin-resistant state [21,26], leading Ryan and Enns [26] to postulate a postbinding defect in insulin action during pregnancy, a conclusion which had also been reached by Puavilai et al. [ 51. Alterations in insulin binding to adipocytes in pregnancy have also been reported, but the results are conflicting. In humans, decreases in binding have been reported [ 271, which contrasts with the present results showing no changes in binding, whilst in rats binding increases during pregnancy and is then reduced just prior to parturition [25,28]. Several studies have documented a relationship between adipocyte size and glucose oxidation and lipid metabolism, although there is disagreement as to whether these are positively or negatively correlated [23,24,29-331. Furthermore, Andersen and Kuhl [32] reported an increased adipocyte size in pregnancy, but concluded nonetheless that it was unlikely that adipocytes contributed in any major way to the in vivo insulin resistance of pregnancy. The present studies are not entirely in accord with these observations, although we agree with the general conclusion of Andersen and Kuhl [ 321 that a major role for adipose tissue in the in vivo insulin resistance of pregnancy is unlikely. Firstly, we find that the metabolic activity of human adipose tissue is very low, certainly when compared to the more widely used rodent adipose tissue, with which most of the experimental methodologies were developed. This is not an unexpected finding since human adipose tissue has been considered to function much more as a simple store for triglyceride synthesised by the liver and transported to the adipose tissue as very low density lipoprotein, rather than as a site for active de novo triglyceride synthesis as occurs in rats and mice. It does however mean that the contribution to glucose uptake made by the adipose tissue following a glucose load may be proportionately less in the human than in rodents, and thus as a tissue
82
for the study of insulin resistance it may be less relevant in humans. An important consequence of this relatively inert metabolism is the difficulty of accurate metabolic measurements. With such a low activity, the methods which are routinely employed are being used at the limits of their sensitivity. We found the very small differences provoked by insulin with both the non-pregnant and pregnant tissues difficult to measure. It may well be that, in humans, skeletal muscle would prove to be a much more useful tissue with which to address the questions of the site of insulin resistance, and at least one study [ 341 has shown that the inherent difficulties of obtaining enough tissue may not be insurmountable. Secondly, we find that for all the parameters we examined, the cells from pregnant women were as responsive to insulin as the control cells from non-pregnant women. Insulin binding and the conversion of labelled glucose to lipid were not statistically significantly different in the two groups. The rates of conversion of glucose to CO, were higher in the non-pregnant group than in the pregnant group. However, this was due to an increased oxidation rate in the absence of insulin, and the increase due to insulin was the same in both groups. It is perhaps relevant that Andersen et al. [34] found a reduction in the activity of pyruvate kinase in adipose tissue from pregnant women compared to non-pregnant women, and inferred from this a reduced glycolytic flux in pregnancy. Any reduction in glycolysis may also constrain the conversion of glucose to CO,, as observed in the present experiments. The enzyme data of Andersen et al. [34] further suggest that, in adipose tissue, the flux-limiting enzyme of glycolysis may be pyruvate kinase rather than phosphofructokinase or hexokinase, and that the reverse is true in muscle. This suggests that in the pregnant state a reduction in the activity of pyruvate kinase may ensure an adequate supply of triose phosphates and glycerol phosphate for esterification to triglyceride under conditions where insulin-stimulated glucose uptake in vivo may be restricted. This inferred requirement for a sustained capacity for esterification may be
related to the substrate cycling between triglycerides and fatty acids which occurs in human adipocytes [33]. Although the basal activity of the tyrosine protein kinase was higher in the pregnant group than the control group, a similar increase in activity in response to insulin was observed for both groups, and there was no evidence for any defect in insulin-dependent kinase activity in the pregnant group compared to the control nonpregnant group. The reason for the difference in basal rather than insulin-stimulated activity between the two groups is not clear. Receptor preparations produced by wheat germ agglutinin chromatography are known to be heterogeneous, with insulin receptor representing less than 1O. of the total [22]. However, at present we have no evidence as to the nature of the non-insulindependent tyrosine protein kinase associated with the preparations from the pregnant group, although an increase in some other receptor-associated kinase is the most exciting possibility. In none of our studies did we see clear evidence for an in vitro insulin resistance associated with the pregnant state. In contrast to Andersen and Kuhl [ 321 we found no differences in size between adipocytes from pregnant or non-pregnant subjects, so considerations of the confused relationship between cell size and metabolism become irrelevant in the context of the present findings. It is perhaps worth noting that although Ryan and Enns [26] showed freshly isolated adipocytes from pregnant rats to have a reduced capacity for 3-methylglucose transport compared to controls, the capacity to respond to insulin was still maintained. On a percentage basis the increase with cells from pregnant rats was the same as that for controls. Thus, even with rat adipose tissue, the evidence for an in vitro difference in insulin sensitivity may be questioned. A third consideration which arises from these observations is why a well described and documented resistance to insulin action in vivo may not be manifest once the tissue is removed from the body and examined in vitro. We consider the simplest explanation is that the resistance is not an inherent feature of the tissue in the pregnant
83 state, but is imposed on the tissue by the presence in vivo of high concentrations of circulating hormones with actions antagonistic to insulin. In this state a post-receptor defect in insulin action would not be anticipated, and there is no evidence from our studies that any defect per se in insulin sensitivity, post-receptor or otherwise, occurs in human adipose tissue in the pregnant state.
10
II
12
13
Acknowledgements This work was supported by a grant from Birthright, the research charity of the Royal College of Obstetricians and Gynaecologists. We are grateful to Professor I. Dunsmore, Department of Probability and Statistics, for advice on the statistical analysis.
14
References
17
Kalkhoff, R., Schalch, D.S.. Walker, J.L., Beck, P., Kipnis. D.N. and Daughaday, W.H. (1964) Diabetogenic factors associated with pregnancy. Trans. Assoc. Am. Physicians 77, 270-280. Spellacy, W.N. and Goetz, F.C. (1963) Plasma insulin in normal late pregnancy. N. Engl. J. Med. 268, 988-991. Bleicher, S.J., O’Sullivan, J.B. and Freinkel, N. (1964) Carbohydrate metabolism in pregnancy. N. Engl. J. Med. 27 1. 866-872. Ryan, E.A., O’Sullivan, M.J. and Skyler. J.S. (1985) Insulin action during pregnancy. Studies with the euglycemic clamp technique. Diabetes 34. 380-389. Puavilai, G., Drobny, E.C.. Domont, L.A. and Baumann, G. (1982) Insulin receptors and insulin resistance in human pregnancy: evidence for a post receptor defect in insulin action. J. Clin. Endocrin. Metab. 54, 247-253. Taylor, R.. Proctor, S.J., James, O., Clark, F. and Alberti, K.G.M.M. (1984) The relationship between human adipocyte and monocyte insulin binding. Clin. Sci. 67, 139-142. Taylor, R. (1986) Insulin receptors and the clinician. Br. Med. J. 292. 919-922. Pedersen. 0.. Hjollund, E., Beck-Nielsen, H.O., L.indsilov, 0.. Sonne, 0. and Gliemann, J. (1981) Insulin receptor binding and receptor mediated insulin dein human adipocytes. Diabetologia 20, gradation 636-641. Hjollund. E.. Pedersen, O., Espersen. T. and Klebe, J.G. (1986) Impaired insulin receptor binding and postbinding
15
16
18
19
20
21
22
23
24
25
defects of adipocytes from normal and diabetic pregnant women. Diabetes 35, 598-603. Pedersen, O., Hjollund, E. and Lindskov, H.O. (1982) Insulin binding and action on fat cells from young healthy females and males. J. Am. Physiol. 243. E158-E167. Larch. E. and Rentsch, G. (1969) A simple method for staining and counting isolated adipose tissue fat cells. Diabetologia 5, 356-357. Schwartz, D.S., Larsh, H.W. and Bartels, P.A. (1977) Enumerative fluorescent vital staining of live and dead pathogenic yeast cells. Stain Technol. 52. 203-210. Gliemann. J. (1967) Assay of insulin-like activity by the isolated fat cell method. 1. Factors influencing the response to crystalline insulin. Diabetologia 3, 382-388. Nattrass, M. and Dodds, K.E. (1987) Interpretation of insulin radioreceptor assays. Ann. Clin. Biochem. 24, 13-21. Rodbell, M. (1964) Metabolism of isolated fat cells. I. Effects of hormones on glucose metabolism and lipolysis. J. Biol. Chem. 239, 375-380. Khan, M.N., Baquiran, G., Brule, C., Burgess, J., Foster, B., Bergeron, J.J.M. and Posner, B.I. (1989) Internalisation and activation of the rat liver insulin receptor kinase in vivo. J. Biol. Chem. 264, 12931-12940. Klein, H.H., Ciaraldi, T.P., Freindenberg, G.R. and Olefsky, J.M. (1987) Adenosine modulates insulin activation of insulin receptor kinase in intact rat adipocytes. Endocrinology 120, 2339-2345. Rosen, O.M., Herrara, R., Olowe, Y., Petruzzelli, L.M. and Cobb, M.H. (1983) Phosphorylation activates the insulin receptor tyrosine protein kinase. Proc. Nat]. Acad. Sci. 80, 3237-3240. Zick, Y., Kasuga. M., Kahn, C.R. and Roth, J. (1983) Characterisation of insulin-mediated phosphorylation of the insulin receptor in a cell-free system. J. Biol. Chem. 258,75-80. Ridge, K.D.. Hofmann, K. and Finn, F.M. (1988) ATP sensitises the insulin receptor to insulin. Proc. Natl. Acad. Sci. 85, 9489-9493. Haring, H., Kirsch, D., Obermaier, B., Ermel, B. and Machicao, F. (1986) Decreased tyrosine kinase activity of insulin receptor isolated from rat adipocytes rendered insulin-resistant by catecholamine treatment in vitro. Biochem. J. 234, 59-66. Fujita-Yamaguchi, Y., Choi, S., Sakamoto, Y. and Itakura, K. (1983) Purification of insulin receptor with full binding activity. J. Biol. Chem. 258, 5045-5049. Craig, B.W. and Treadway, J. (1986) Glucose uptake and oxidation in fat cells of trained and sedentary pregnant rats. J. Appl. Physiol. 60, 1704-1709. Craig, B.W., Garthwaite, S.M. and Holloszy, J.O. (1987) Adipocyte insulin resistance: effects of aging, obesity, exercise and food restriction. J. Appl. Physiol. 62,95-100. Flint, D.J.. Clegg, R.A. and Vernon, R.G. (1980) Regulation of adipocyte insulin receptor number and metabo-
84
26
27
28
29
30
lism during late pregnancy. Mol. Cell. Endocrinol. 20, 101-111. Ryan, E.A. and Enns, L. (1988) Role of gestational hormones in the induction ofinsulin resistance. J. Clin. Endocrinol. Metab. 67, 341-347. Pagano, G., Cassader, M., Massobrio, M., Bozzo, C., Trossarelli, G.F., Menato, G. and Lenti, G. (1980) Insulin binding to human adipocytes during late pregnancy in healthy, obese and diabetic state. Horm. Metab. Res. 12, 177-181. Flint, D.J., Sinnett-Smith, P.A., Clegg, R.A. and Vernon, R.G. (1979) Role of insulin receptors in the changing metabolism of adipose tissue during pregnancy and lactation in the rat. Biochem. J. 182, 421-427. Foley, J.E., Laursen, A.L., Sonne, 0. and Gliemann, J. (1980) Insulin binding and hexose transport in rat adipocytes. Diabetologia 19, 234-241. Vinten. J. and Galbo, H. (1983) Effect ofphysical training
31
32
33
34
on transport and metabolism of glucose in adipocytes. Am. J. Physiol. 244, E129-134. Savard, R., Deshaies, Y., Despres, J.P., Marcotte, M., Bukowiecki, L., Allard, C. and Bouchard, C. (1984) Lipogenesis and lipoprotein lipase in human adipose tissue reproducibility ofmeasurement and relationships with fat cell size. Can. J. Physiol. Pharmacol. 62, 1448-1452. Andersen, 0. and Kuhl, C. (1988) Adipocyte insulin receptor binding and lipogenesis at term in normal pregnancy. Eur. J. Clin. Invest. 18, 575-581. Hammond, V.A. and Johnston, D.G. (1987) Substrate cycling between triglyceride and fatty acid in human adipocytes. Metabolism 36, 308-313. Andersen. O., Falholt, K. and Kuhl, C. (1989) Activity of enzymes of glucose and triglyceride metabolism in adipose and muscle tissue from normal pregnant women at term. Diabetic Med. 6, 131-136.
