Endothelial function is preserved in pregnant women with well-controlled type 1 diabetes

BJOG: an International Journal of Obstetrics and Gynaecology June 2002, Vol. 109, pp. 699 707 Endothelial function is preserved in pregnant women with well-controlled type 1 diabetes Christine Ang a, Chris Hillier b, Fiona Johnston b, Alan Cameron c, Ian Greer a, Mary Ann Lumsden a, * Objective Pregnant women with diabetes mellitus have a higher incidence of adverse pregnancy outcomes. Vascular, and in particular, endothelial function may be significantly modified in diabetes resulting in impaired endothelium-dependent relaxation. This study aims to investigate endothelium-dependent relaxation in pregnant women with pre-existing type 1 diabetes mellitus. Methods Small arteries (mean luminal diameter f295 Am) were isolated from biopsies of subcutaneous fat from pregnant women with pre-existing type 1 diabetes mellitus, non-diabetic pregnant women, and nondiabetic non-pregnant women. Endothelial and smooth muscle function were determined using wire myography, and the contributions of nitric oxide, vasodilator prostanoid and endothelial hyperpolarisation were studied using specific inhibitors. Results Arteries obtained from the diabetic pregnant women did not demonstrate any difference in either endothelial or smooth muscle function when compared with non-diabetic pregnant women. The contribution of nitric oxide to endothelium-dependent relaxation was f20% in the pregnant women regardless of whether they were diabetic, and f11% in the non-pregnant women. Endothelial hyperpolarisation appeared to contribute largely to vasorelaxation in human subcutaneous arteries, and was at least twice that of nitric oxide in pregnant women and fivefold greater in non-pregnant women. Conclusions This study provides evidence that pregnant women with well-controlled pre-existing type 1 diabetes mellitus have both normal endothelial and smooth muscle function. Endothelium-dependent hyperpolarisation appears to play a large role in vascular relaxation in human subcutaneous resistance arteries. This study suggests that the problems associated with diabetic pregnancies are unlikely to be due to vascular dysfunction. INTRODUCTION Numerous studies have demonstrated impaired endothelial function in diabetes, implicating it as one of the mechanisms underlying the complications associated with the disorder 1,2. Abnormal vascular responsiveness in diabetics, due particularly to endothelial dysfunction, can deteriorate rapidly during pregnancy, when the ability of the maternal vasculature to respond quickly to changes in metabolic requirements may be further compromised. The increased incidence of complications such as pre-eclampsia and placental insufficiency, associated with increased a Department of Obstetrics and Gynaecology, University of Glasgow, UK b Vascular Assessment Unit, Glasgow Caledonian University, Glasgow, UK c Department of Fetal Medicine, The Queen Mother s Hospital, Glasgow, UK * Correspondence: Dr M. A. Lumsden, University Department of Obstetrics and Gynaecology, The Queen Mother s Hospital, Glasgow G3 8SJ, UK. D RCOG 2002 BJOG: an International Journal of Obstetrics and Gynaecology PII: S 1 47 0-0 3 2 8(0 2)0 1 3 53-8 maternal and fetal vascular resistance, is more common in diabetic compared with non-diabetic pregnancies 3,4. Nitric oxide has been one of the most widely studied endothelium-dependent relaxing factors, inducing vasodilatation via the guanosine 3V:5V cyclic monophosphate (cgmp) pathway. Its significant contribution to the control of vascular tone has been suggested by studies that have shown marked vasoconstriction following nitric oxide synthase inhibition with analogues of L-arginine, such as N N -nitro-l-arginine methyl ester (L-NAME) and N N -nitro- L-arginine (L-NOARG) 5. In most human vascular beds, the contribution of the vasodilator prostanoid to endotheliumdependent relaxation is negligible 6, but along with nitric oxide, is thought to be associated with the physiological vasodilatation seen in pregnancy 7,8. It mediates vasorelaxation through accumulation of adenosine 3V:5V cyclic monophosphate (camp), and can be assessed by inhibition of the prostaglandin converting enzyme, cyclo-oxygenase. Until recently, there has been little emphasis on the role of endothelial hyperpolarisation in vascular relaxation. This hyperpolarisation was initially thought to be due to an endothelium-derived hyperpolarising factor (EDHF) whose precise chemical identity remains uncertain, but has been shown to be insensitive to the combined effects of nitric oxide synthase and cyclo-oxygenase inhibition 9. www.bjog-elsevier.com

700 C. ANG ET AL. However, current literature suggests that endotheliumdependent hyperpolarisation is more likely due to electrical coupling through myoendothelial gap junctions, rather than a single chemical factor 10 12. Endothelial hyperpolarisation has been shown to be important in the resistance vasculature, where changes in vascular reactivity can significantly alter systemic blood pressure and tissue perfusion 13. Furthermore, 25 mm potassium chloride solution and the combined effects of apamin and charybdotoxin have been shown to inhibit its effects on vascular relaxation 14,15. In these small vessels, endothelium-dependent hyperpolarisation was able to mediate near normal relaxation following nitric oxide inhibition, and may be important in the regulation of vascular tone in diseases such as diabetes where either nitric oxide production or activity is impaired 16. Few studies have been published investigating the influence of pre-existing type 1 diabetes mellitus on maternal vascular responses, with particular emphasis on the contribution of endothelial hyperpolarisation. We have hypothesised that endothelium-dependent relaxation is impaired in pregnant women with pre-existing type 1 diabetes, but that compensatory hyperpolarisation is increased in these women compared with pregnant women without diabetes. The aim of this study, therefore, was to determine the three different components of endothelium-dependent relaxation, namely, nitric oxide, vasodilator prostanoid and endothelial hyperpolarisation, in pregnant women with and without type 1 diabetes mellitus. METHODS Study groups Local Ethical Committee approval was granted. Written informed consent was obtained prior to caesarean section in 11 healthy pregnant and nine diabetic pregnant women at term; and prior to elective laparotomy in seven healthy non-pregnant women. All women were normotensive and matched for maternal and gestational age. The diabetic women had pre-existing diabetes and were well controlled on insulin, not requiring significant manipulation of their regimes with advancing gestation. The non-pregnant women were age-matched, fit and well, and were not on any regular medication. Further patient characteristics including blood pressure, birthweight, glycosylated haemoglobin (Hb A1c ) levels and disease duration are summarised in Table 1. Standardising recruitment to include only women having abdominal surgery enabled the same vascular bed (abdominal subcutaneous resistance arteries) to be studied. The indications for caesarean section in the healthy pregnant women were previous caesarean section (seven), previous traumatic delivery (two), fetal distress (one) and breech presentation (one). The indications for caesarean section in the diabetic pregnant women were previous caesarean section (three), fetal distress (three) and failed induction of labour (three). The non-pregnant women were recruited from those admitted for elective gynaecological surgery and included those undergoing ovarian cystectomy (three), total abdominal hysterectomy (two), colposuspension (one) and myomectomy (one). Small vessel myography At laparotomy, biopsies of subcutaneous fat were obtained from the incision edge. The biopsies were transported to the laboratory in 0.9% saline in ice within half an hour. A total of 34 small arteries with a mean luminal diameter of f295 Am were dissected free and stored in physiological saline at 4jC. The mean time of storage of the vessels was 17 hours prior to experimentation (ranging from 0 to 19 hours). Overnight storage is an important point in the interpretation of vessel responses, however, there did not appear to be any difference between vessels experimented on immediately following dissection and those stored overnight. The relationship between storage time, maximal relaxation and tension development is shown in Table 1. Patient characteristics. Values are expressed as mean (SEM). Significance assumed if P < 0.05 using one-way ANOVA and Bonferroni s post hoc test. HbA 1c ¼ glycosylated haemoglobin. Healthy pregnant (n ¼ 11) Diabetic pregnant (n ¼ 8) Healthy non-pregnant (n ¼ 7) Age (years) 30 (2) 28 (2) 35 (3) Gestational age (weeks) 39 (0.2) 37.5 (0.4) Birthweight (g) 3542 (151) 3991 (242) Maternal weight (kg) 63 (3) 72 (4) 69 (6) Maternal body mass index (kg/m 2 ) 25 (1) 27 (1) 26 (2) Blood pressure (systolic) (mmhg) 114 (3) 127 (4) 123 (7) Blood pressure (diastolic) (mmhg) 70 (3) 79 (2) 80 (4) Smokers n/n 1/11 0/11 3/7 HbA 1c (%) 6.3 (0.3) Duration of diabetes 15.1 (2.1)

ENDOTHELIAL RELAXATION IN DIABETIC PREGNANT WOMEN 701 Table 2. The relationship between vessel storage time and vascular function. Values are mean (SEM). n ¼ number of vessels. Maximal relaxation refers to % relaxation to 3-AM carbachol in vessels preconstricted with 10-AM noradrenaline, while tension refers to vessel contraction with 10-AM noradrenaline. Storage time (hours) Healthy pregnant Diabetic pregnant Healthy non-pregnant Max. relaxation (%) Tension (mn/mm) Max. relaxation (%) Tension (mn/mm) Max. relaxation (%) Tension (mn/mm) 0 83.9 (9.4) (n ¼ 4) 14 94.4 (0.8) 3.6 (0.7) 94.6 (n ¼ 2) (n ¼ 2) (n ¼ 1) 19 92.6 (5.3) 4.1 (0.4) 95.7 (1.0) 3.4 (0.3) 95.9 (0.7) (n ¼ 14) (n ¼ 14) (n ¼ 10) (n ¼ 10) (n ¼ 3) 3.7 (1.0) (n ¼ 4) 3.7 (n ¼ 1) 3.3 (0.3) (n ¼ 3) Table 2. The isolated arteries were then studied in vitro using conventional small vessel wire myography, which allows the assessment of isometric responses of vessels to a chosen agonist. Isometric tension was measured (Myodaq/Myodata 2.01, Danish Myotech, Aarhus, Denmark) using standard methodology 17. In brief, the vessels were allowed to equilibrate for one hour in physiological saline at 37 F 0.1jC. They were exposed to a combination of 95% O 2 /5% CO 2 to achieve a ph of 7.4 at 37jC. The vessels then underwent a normalisation process, which determined the passive tension characteristics of each individual vessel. This allowed the internal circumference of each vessel to be set to a value 90% of that it would have, if relaxed in vivo under a transmural pressure of 100 mmhg. This value has been shown to be the internal circumference at which the vessels will produce a maximum contractile response 17. Before commencing experimentation, vessel viability (contractile and endothelial function) was confirmed by contraction to 123 mm potassium chloride solution and relaxation to carbachol (3 AM in vessels pre-constricted with 10 AM noradrenaline). Vessels achieved a mean relaxation of f 84 F 3% in healthy pregnant women, f 82 F 3% in diabetic pregnant women, and f 82 F 3% in non-pregnant women. Arteries that did not achieve at least 60% relaxation were considered to be damaged and unsuitable for studies of endothelial function. Mechanical damage in our experience usually significantly reduces endothelial relaxation responses to below 15%. On the other hand, the literature suggests that the observed relaxation to acetylcholine in both pregnant and non-pregnant women, and patients with type 1 diabetes and gestational diabetes, will be approximately equal to or greater than 60% 2,6,18. The use of a 60% cutoff therefore ensured that only mechanically damaged vessels would be excluded, and functionally impaired (by the disease process) vessels would be included. A total of 39 arteries were isolated, 5 were rejected, leaving 34 that were used in the study. There was no difference in the percentage of rejected vessels per patient group. Finally, to provide a measure of maximal vasoconstriction, the vessels were constricted with 10 AM noradrenaline in 123 mm potassium chloride solution, were allowed to relax to baseline, and were equilibrated for a further hour. Pharmacological protocols Four pharmacological protocols were used. All experiments were carried out in arteries pre-constricted with 10 AM noradrenaline, which produced a sustained maximal contraction during the duration of the response curve. However, in studies where vessels were incubated with 25 mm potassium chloride, the concentration of noradrenaline was titrated to achieve a similar pre-constriction tension to previous cumulative concentration response curves. 1. In order to assess endothelium-dependent relaxation and quantify the contribution of hyperpolarisation to this, both nitric oxide and prostacyclin were inhibited. Cumulative concentration response curves were performed to carbachol (1 nm to 30 AM), before and after a 30-minute stepwise incubation with: L-NAME (nitric oxide synthase inhibitor, 0.1 mm) L-NAME and indomethacin (cyclo-oxygenase inhibitor, 30 AM) L-NAME, indomethacin and 25 mm potassium chloride (inhibitor of endothelial hyperpolarisation) 14 2. The second protocol aimed to study the effectiveness of camp and cgmp pathways in the vascular smooth muscle, independent of endothelial modulation. Cumulative concentration response curves were performed to prostacyclin (1 nm to 30 AM) and sodium nitroprusside (1 nm to 30 AM), endothelium-independent agonists dependent on camp and cgmp, respectively. 3. To ensure that any residual relaxation following nitric oxide synthase inhibition was not due to an incomplete inhibition of the pathway, in a different set of vessels obtained from the same healthy pregnant women, vessels were incubated simultaneously for 30 minutes with L-NAME (0.1 mm), indomethacin (30 AM) and oxadiazoloquinoxalin (ODQ, 1 AM), an inhibitor of soluble guanylate cyclase, before studies were performed with carbachol (1 nm to 30 AM) and sodium nitroprusside (1 nm to 30 AM). 4. The fourth protocol aimed to confirm the contribution of endothelial hyperpolarisation to the residual relaxation

702 C. ANG ET AL. response following inhibition of nitric oxide synthase and cyclo-oxygenase. Carbachol and sodium nitroprusside studies were repeated following 30 minutes incubation with charybdotoxin (100 nm) and apamin (100 nm), which inhibit both large and small conductance calciumdependent potassium channels and endothelium-dependent hyperpolarisation 15. Drugs and solutions All drugs, except prostacyclin (prostaglandin I 2 ), were obtained from Sigma, Poole, UK. Prostacyclin was obtained from CN Biosciences, Nottingham, UK. This preparation of prostacyclin is stable in water with activity for 5 to 7 minutes and was therefore suitable for these in vitro studies. Distilled H 2 O was used to prepare the drug solutions, except indomethacin, which had to be dissolved in dimethyl sulphoxide (DMSO). Physiological saline solution had the following composition: NaCl 118 mm, NaHCO 3 25 mm, KCl 4.5 mm, KH 2 PO 4 1.0 mm, CaCl 2 2.5 mm, MgSO 4 1.0 mm and C 6 H 12 O 6 (D-glucose) 6.0 mm, 0.023 mm EDTA and 0.03 mm ascorbic acid. Potassium chloride solutions (25 and 123 mm) had similar compositions to physiological saline, and were made by substituting NaCl for KCl on an equimolar basis. Statistical analysis Relaxation is expressed as a percentage relative to the maximum tension generated following arterial preconstriction, mean F standard error of the mean (SEM). n represents the number of patients recruited for each protocol. When more than one vessel per patient was used, the mean vessel response was calculated. The maximum response for each protocol was the greatest mean relaxation achieved, and not the percentage relaxation attained at the highest concentration of carbachol. Sensitivity was calculated using the PRISM statistical software package (Graphpad Software). Maximum relaxation and sensitivity (pd 2 negative log of the concentration required to produce 50% of the maximum response) were compared using one-way ANOVA for repeated measures and Bonferroni s post hoc test for multiple comparisons. A P value <0.05 was accepted as being statistically significant. Table 3. Vessel constriction to potassium (123 mm), noradrenaline (10 AM) and potassium (123 mm), and noradrenaline (10 AM) alone, between the patient groups. Values are mean (SEM). NA ¼ noradrenaline, NAK ¼ noradrenaline and K þ. K þ (mn/mm) NA (mn/mm) NAK (mn/mm) Healthy pregnant 2.9 (0.3) 3.4 (0.4) 3.2 (0.2) Diabetic pregnant 2.7 (0.4) 3.2 (0.3) 3.1 (0.5) Healthy non-pregnant 3.3 (0.7) 3.1 (0.5) 3.4 (0.3) found in either the maximum responses or the pd 2. Following incubation with L-NAME, the maximum relaxation was attenuated in all groups indicating a contribution of nitric oxide to endothelium-dependent relaxation. Figures 1a, b and c show the cumulative concentration response curves to carbachol following sequential inhibition of nitric oxide, vasodilator prostanoid and endothelial hyperpolarisation in each of the three study groups. The maximum relaxation and pd 2 data between the three groups are summarised in Table 4. The contribution of nitric oxide was approximately 20% in the pregnant women regardless of whether they were diabetic, and approximately 11% in the non-pregnant women. Endothelial hyperpolarisation appeared to contribute greatly to vasorelaxation in all three groups of women. There was no difference in maximum relaxation to either prostacyclin or sodium nitroprusside between the three patient groups (Fig. 2). The small degree of residual relaxation following incubation with L-NAME, indomethacin and 25 mm potassium chloride was not due to incomplete blockade of nitric oxide synthase as ODQ was unable to inhibit the response further. Maximum relaxation to carbachol in the presence of L-NAME þ indomethacin þ 25 mm potassium chloride: f16 F 5% vs L-NAME þ indomethacin þ 25 mm potassium chloride þ ODQ: f19 F 4%, n ¼ 6. The efficacy of ODQ was tested by observing negligible relaxation to sodium nitroprusside. The maximum relaxation to sodium nitroprusside in the presence of L-NAME þ indomethacin þ 25 mm potassium chloride þ ODQ was f3 F 1%, n ¼ 6 (Fig. 3). The combination of L-NAME, indomethacin, and the toxins (apamin and charybdotoxin) was able to fully inhibit all endothelium-dependent relaxation. Maximum relaxation to carbachol in the presence of L-NAME þ indomethacin þ apamin þ charybdotoxin: f2 F 1%, n ¼ 6 (Fig. 4). RESULTS No differences in vessel constriction to potassium, noradrenaline and potassium, and noradrenaline alone were observed between the patient groups. This is shown in Table 3. Vasorelaxation responses to carbachol were similar in all the three groups of patients with no significant differences DISCUSSION This study provides evidence that pregnant women with well-controlled pre-existing type 1 diabetes mellitus have normal endothelial and smooth muscle function in subcutaneous resistance vessels taken from the anterior abdominal wall. Furthermore, we have shown that endothelial

ENDOTHELIAL RELAXATION IN DIABETIC PREGNANT WOMEN 703 Fig. 1. (A) Relaxation to carbachol (n) in healthy pregnant women (n ¼ 11) following incubation with L-NAME (.); L-NAME and indomethacin (E); and L-NAME, indomethacin and 25 mm potassium chloride (y). (B) Relaxation to carbachol (5) in diabetic pregnant women (n ¼ 9) following incubation with L-NAME (o); L-NAME and indomethacin (4); and L-NAME, indomethacin and 25 mm potassium chloride ( w ). (C) Relaxation to carbachol (n) in healthy non-pregnant women (n ¼ 7) following incubation with L-NAME (.); L-NAME and indomethacin (E); and L-NAME, indomethacin and 25 mm potassium chloride (y). hyperpolarisation contributes largely to endotheliumdependent relaxation in pregnant women with and without diabetes. We have not shown any impairment of endothelial function in pregnant women with type 1 diabetes, which supports previous studies suggesting that endothelial function is maintained in patients with uncomplicated type 1 diabetes 19,20. Our observations in isolated subcutaneous resistance arteries obtained from the anterior abdominal wall may be due to one of two reasons. Firstly, the assessment of endothelial function is determined by the degree of agonist-mediated vasorelaxation following preconstriction. However, once a vessel reaches complete relaxation, any further increases in the arterial luminal diameter cannot be achieved. Therefore, if present, a small degree of endothelial dysfunction may be unobservable in the diabetic women due to the physiological increase in vasodilatation known to be associated with pregnancy.

704 C. ANG ET AL. Table 4. Relaxation responses of isolated arteries obtained from healthy pregnant, diabetic pregnant and healthy non-pregnant women. One-way ANOVA and Bonferroni s post hoc test were carried out between the three patient groups. Values are expressed as mean (SEM), and refer to relaxation to carbachol with the sequential addition of inhibitors. Significance assumed if P < 0.05 using one-way ANOVA and Bonferroni s post hoc test. L-NAME ¼ N N -nitro- L-arginine methyl ester; INDO ¼ indomethacin; KCl ¼ potassium chloride solution. Healthy pregnant Diabetic pregnant Healthy non-pregnant Internal arterial diameter (Mm) 294 (22) 296 (23) 293 (26) Mean tension prior to carbachol (mn/mm) 4.1 (0.4) 3.4 (0.3) 3.5 (0.5) Carbachol Max. relaxation (%) 92.6 (2.1) 95.5 (0.9) 89.9 (5.2) pd 2 6.8 (0.1) 7.0 (0.1) 7.0 (0.2) þl-name Max. relaxation (%) 74.3 (3.9) 76.1 (3.4) 80.4 (5.4) pd 2 6.6 (0.1) 6.7 (0.2) 6.5 (0.2) þl-name þ INDO Max. relaxation (%) 74.4 (4.5) 74.5 (3.6) 74.6 (6.3) pd 2 7.3 (0.2) 7.2 (0.3) 6.9 (0.3) þl-name þ INDO þ25 mm KCl Max. relaxation (%) 34.2 (7.4) 28.2 (5.7) 28.2 (7.3) pd 2 6.6 (0.2) 6.2 (0.2) 6.0 (0.3) Secondly, the patients recruited were young women with a relatively shorter disease duration and tighter glycaemic control (due to closer supervision at joint medical/antenatal clinics). This is reflected by the mean HbA 1c level, birthweight and delivery of the fetus at term ( > 37 weeks) in our study population. We have demonstrated a preservation of smooth muscle function in pregnant women with type 1 diabetes, as responses to both prostacyclin and sodium nitroprusside were similar in all three groups of women. However, the effect of diabetes on the vascular smooth muscle has also been shown to differ between different study populations. Assessed by exposure to sodium nitroprusside (an endothelium-independent nitric oxide donor), vascular smooth muscle function has been shown to be preserved in patients with type 1 diabetes mellitus and impaired in patients with Fig. 2. Endothelium-independent relaxation to prostacyclin and sodium nitroprusside in isolated small arteries obtained from healthy pregnant (P, n ¼ 11), diabetic pregnant (IDDM, n ¼ 9) and healthy non-pregnant women (NP, n ¼ 7).

ENDOTHELIAL RELAXATION IN DIABETIC PREGNANT WOMEN 705 Fig. 3. Relaxation to carbachol (y) in healthy pregnant women (n ¼ 6) following incubation with L-NAME, indomethacin and 25 mm potassium chloride (n); L-NAME, indomethacin, 25 mm potassium chloride and ODQ (E). Relaxation to sodium nitroprusside following incubation with L-NAME, indomethacin, 25 mm potassium chloride and ODQ (.). type 2 diabetes mellitus, compared with age- and sexmatched control subjects 1,21. Our data suggest that diabetes does not cause a generalised reduction in smooth muscle sensitivity to endogenous vasodilators. The storage of subcutaneous vessels overnight has been previously shown to be acceptable with no measurable change in function observed. McIntyre et al. 22 showed that vascular function was preserved in third-order rat mesenteric arteries following cold storage for up to four days, when exposed to the vasoconstrictors noradrenaline phenylephrine, potassium chloride and endothelin, and the vasodilators acetylcholine and 3-morpholinosydnonimine. In our study, although there did not appear to be any difference between vessels experimented on immediately following dissection and those stored overnight in terms of vascular relaxation or constriction, overnight storage is an important point in the interpretation of this dataset. The literature has suggested an increased role of endothelium-dependent relaxing factors during pregnancy, contributing to significant vasodilatation 7,8. However, previous studies have concentrated mainly on contributions by nitric oxide and prostacyclin with little emphasis on endotheliumderived hyperpolarisation. The data from this study appear to demonstrate an increased contribution of nitric oxide of approximately 20% in pregnant women and approximately 11% in non-pregnant women. However, the contribution of endothelial hyperpolarisation is, qualitatively, at least twice that of nitric oxide in pregnant women, and fivefold greater in non-pregnant women, supporting other studies which suggest that endothelium-dependent hyperpolarisation is more prominent in resistance arteries 9,13,14. The absence of a group of young, non-pregnant women with type 1 diabetes mellitus does not allow the effect of pregnancy on endothelial relaxation in diabetic women to be studied. However, there would have been significant difficulty in recruiting a sufficient number of young, type 1 diabetic women undergoing laparotomy, and it would have been necessary to obtain the vessels from gluteal biopsies. Following inhibition with L-NAME, indomethacin and 25 mm potassium chloride, we observed a residual relaxation of f30%, which was possibly either due to incomplete inhibition of nitric oxide or endothelial hyperpolarisation. Earlier studies suggested that L-NAME may not fully inhibit the nitric oxide synthase pathway 23, and Gerber et al. 14 reported that pre-incubation with 25 mm potassium chloride abolished the effects of endothelium-derived hyperpolarisation in rat mesenteric arteries. We confirmed the efficacy of L-NAME by incubating the vessels with L-NAME, indomethacin, 25 mm potassium chloride and ODQ (an inhibitor of cgmp production). In the presence of ODQ, no further inhibition of the carbachol response was observed. We tested the efficacy of ODQ by repeating that protocol with the nitric oxide donor, sodium nitroprusside. As sodium nitroprusside demonstrated only f3% relaxation in the presence of ODQ, we were confident that the remaining relaxation was not due to the incomplete inhibition of nitric oxide synthase activity. We then substituted 25 mm potassium chloride with apamin (100 nm) and % Relaxation 0 20 40 60 80 100 Agonist (Log M) -9-8 -7-6 -5 Fig. 4. Relaxation to carbachol (y) in healthy pregnant women (n ¼ 6) following incubation with L-NAME, indomethacin, apamin and charybdotoxin (n). Relaxation to sodium nitroprusside following incubation with L-NAME, indomethacin, apamin and charybdotoxin (.).

706 C. ANG ET AL. charybdotoxin (100 nm), which have been shown to inhibit the effects of endothelium-dependent hyperpolarisation 15. In the presence of L-NAME, indomethacin, apamin and charybdotoxin, the vessels failed to relax, suggesting that the previous residual relaxation had been due to the incomplete inhibition of endothelial hyperpolarisation. The non-nitric oxide, non-prostanoid effect appears to be sensitive to the combined effects of apamin and charybdotoxin, suggesting an involvement of large and small conductance calcium-dependent potassium channels. However, potassium itself can induce a direct hyperpolarisation and relax smooth muscle either by stimulating ouabainsensitive sodium potassium-atpase, inward-rectifying potassium channels, or by stimulating the release of a vasodilator mediator(s) from perivascular nerves that induce endothelium-dependent vasodilatation 24 26. Pascoal and Umans 27 explored endothelium-dependent vasodilatation in human omental resistance arteries, focussing on the action of nitric oxide-independent relaxing factors in both pregnant and non-pregnant women. They showed that vessels obtained from pregnant women relaxed to bradykinin in the presence of nitric oxide synthase, cyclooxygenase and non-selective potassium channel blockade, while vessels obtained from non-pregnant women showed minimal (12%) relaxation in the presence of the same three inhibitors. In contrast to our findings, they also suggested that endothelium-dependent vasodilatation pregnancy may be associated with a hyperpolarising vasodilator, which may not be a potassium channel factor. The difference in the effect of 25 mm potassium chloride on endothelium-dependent hyperpolarisation between human subcutaneous and rat mesenteric arteries may be due to the fact that the hyperpolarising effect now seems more likely to be due to electrical coupling, rather than a single identifiable chemical factor 10 12. In our study, the contribution of hyperpolarisation to endotheliumdependent relaxation may therefore be more important than previously thought, responsible for approximately 80% of the total relaxation response to carbachol, with no observable difference between the patient groups. However, this assumes that the mechanisms underlying vascular function act independently of each other. It is possible that inhibiting nitric oxide production may in fact enhance the potency of endothelial hyperpolarisation. Kilpatrick and Cocks 28 showed that in porcine coronary arteries, endothelium-dependent hyperpolarisation was able to mediate near-normal relaxation following nitric oxide inhibition, which may be important in the regulation of vascular tone in diseases when nitric oxide production or activity is impaired. The role of endothelial hyperpolarisation may therefore only appear to be increased compared with that of nitric oxide. Although the vasodilator prostanoid contributes more to relaxation in nonpregnant compared with pregnant women, when compared with endothelial hyperpolarisation is considerably less important. CONCLUSIONS This study provides evidence that pregnant women with well-controlled pre-existing type 1 diabetes have both normal endothelial and smooth muscle function in isolated subcutaneous abdominal resistance arteries. Furthermore, endothelial hyperpolarisation appears to play a large role in endothelium-dependent relaxation in human subcutaneous arteries, and is not altered by pregnancy or the presence of diabetes in pregnancy. The physiological increase in peripheral vasodilatation in pregnancy may contribute to the maintenance of vascular function and tissue perfusion in pregnant diabetic women. Although underlying diabetes does have some effect on the pregnant mother, fears of endothelial dysfunction leading to damaging vascular disorders are probably unfounded in well-controlled diabetics. Acknowledgements We thank the patients and staff of the Western Infirmary, Glasgow Royal Maternity and Queen Mother s Hospitals for their help in this study. This work was made possible through the financial support of The Wellcome Trust and the Glasgow Royal Infirmary Research Endowment Fund. References 1. Johnstone MT, Creager SJ, Scales KM, Cusco JA, Lee BK, Creager MA. Impaired endothelium-dependent vasodilatation in patients with insulin-dependent diabetes mellitus. Circulation 1994;88(6): 2510 2516. 2. McNally PG, Watt PAC, Rimmer T, Burden AC, Hearnshaw JR, Thurston H. Impaired contraction and endothelium-dependent relaxation in isolated resistance vessels from patients with insulin-dependent diabetes mellitus. Clin Sci 1994;87:31 36. 3. Garner PR, D Alton ME, Dudley DK, Huard P, Hardie M. Preeclampsia in diabetic pregnancies. Am J Obstet Gynecol 1990; 163:505 508. 4. Redman CWG. Current topic: preeclampsia and the placenta. Placenta 1991;12:301 308. 5. Bennett MA, Watt PAC, Thurston H. Endothelium-dependent modulation of resistance vessel contraction: studies with N G -nitro-larginine methyl ester and N G -nitro-l-arginine. Br J Pharmacol 1992;107:616 621. 6. McCarthy AL, Taylor P, Graves J, Raju SK, Poston L. Endotheliumdependent relaxation of human resistance arteries in pregnancy. Am J Obstet Gynecol 1994;171:1309 1315. 7. Williams DJ, Vallance PJT, Neild GH, Spencer JAD, Imms FJ. Nitric oxide-mediated vasodilation in human pregnancy. Am J Physiol 1997;272:H748 H752. 8. Greer IA, Walker JJ, McLaren M, et al. Immunoreactive prostacyclin and thromboxane metabolites in normal pregnancy. Br J Obstet Gynaecol 1985;92(6):581 585. 9. Coats P, Johnston F, MacDonald J, McMurray JJV, Hillier C. Endothelium-derived hyperpolarizing factor: identification and mechanisms of action in human subcutaneous resistance arteries. Circulation 2001;103(12):1702 1708.

ENDOTHELIAL RELAXATION IN DIABETIC PREGNANT WOMEN 707 10. Coleman HA, Tare M, Parkington HC. EDHF is not K þ but may be due to spread of current from the endothelium in guinea pig arterioles. Am J Physiol 2001;280(6):H2478 H2483. 11. Coleman HA, Tare M, Parkington HC. K þ currents underlying the action of endothelium-derived hyperpolarizing factor in guinea-pig, rat and human blood vessels. J Physiol 2001;531(Pt. 2):359 373. 12. Sandow SL, Hill CE. Incidence of myoendothelial gap junctions in the proximal and distal mesenteric arteries of the rat is suggestive of a role in endothelium-derived hyperpolarizing factor-mediated responses. Circ Res 2000;86(2):341 346. 13. Shimokawa H, Yasutake H, Fujii K, et al. The importance of the hyperpolarizing mechanism increases as the vessel size decreases in endothelium-dependent relaxations in rat mesenteric circulation. J Cardiovasc Pharmacol 1996;28:703 711. 14. Gerber RT, Anwar MA, Poston L. Enhanced acetylcholine induced relaxation in small mesenteric arteries from pregnant rats: an important role for endothelium-derived hyperpolarizing factor (EDHF). Br J Pharmacol 1998;125:455 460. 15. Edwards G, Dora KA, Gardener MJ, Garland CJ, Weston AH. K þ is an endothelium-derived hyperpolarizing factor in rat arteries. Nature 1998;396:269 272. 16. Kilpatrick EV, Cocks TM. Evidence for differential roles of nitric oxide (NO) and hyperpolarization in endothelium-dependent relaxation of pig isolated coronary artery. Br J Pharmacol 1994;112: 557 565. 17. Mulvany MJ, Halpern W. Contractile properties of small arterial resistance vessels in spontaneously hypertensive and normotensive rats. Circ Res 1977;41(1):19 26. 18. Knock GA, McCarthy AL, Lowy C, Poston L. Association of gestational diabetes with abnormal maternal vascular endothelial function. Br J Obstet Gynaecol 1997;104:229 234. 19. McIntyre CA, Hadoke PWF, Williams BC, Lindsay RM, Elliott AI, McKnight JA. Selective enhancement of sensitivity to endothelin-1 despite normal endothelium-dependent relaxation in subcutaneous resistance arteries isolated from patients with Type 1 diabetes. Clin Sci 2001;100:311 318. 20. Smits P, Kapma JA, Jacobs MC, Lutterman J, Thien T. Endothelium-dependent vascular relaxation in patients with type 1 diabetes mellitus. Diabetes 1993;42(1):148 153. 21. Morris SJ, Shore AC, Tooke JE. Responses of the skin microcirculation to acetylcholine and sodium nitroprusside in patients with NIDDM. Diabetologia 1995;38:1337 1344. 22. McIntyre CA, Williams BC, Lindsay RM, McKnight JA, Hadoke PWF. Preservation of vascular function in rat mesenteric resistance arteries following cold storage, studied by small vessel myography. Br J Pharmacol 1998;123(8):1555 1560. 23. Chauhan SD, MacAllister RJ, Clapp LH, Ahluwalia A. Evidence that NO may contribute to EDHF-like responses in rat mesenteric and hepatic small arteries [abstract]. Br J Pharmacol 2000; 129(Suppl):1P. 24. McCarron JG, Halpern W. Potassium dilates rat cerebral arteries by two independent mechanisms. Am J Physiol 1990;259:H902 H908. 25. Knot HJ, Zimmermann PA, Nelson MT. Extracellular K þ -induced hyperpolarizations and dilatations of rat coronary and cerebral arteries involve inward rectifier K þ channels. J Physiol 1996;492(2): 419 430. 26. Rubanyi GM, Vanhoutte PM. Potassium-induced release of endothelium-derived relaxing factor from canine femoral arteries. Circ Res 1988;62:1098 1103. 27. Pascoal IF, Umans JG. Effect of pregnancy on mechanisms of relaxation in human omental microvessels. Hypertension 1996; 28(2):183 187. 28. Kilpatrick EV, Cocks TM. Evidence for differential roles of nitric oxide and hyperpolarization in endothelium-dependent relaxation of pig isolated coronary artery. Br J Pharmacol 1994;112:557 565. Accepted 11 March 2002