Chronically Increased S6K1 Is Associated with Impaired IRS1 Signaling in Skeletal Muscle of GDM Women with Impaired Glucose Tolerance Postpartum

Women with gestational diabetes who fail to regain glucose tolerance postpartum demonstrate persistent S6K1 activation and a decrease in insulin signaling in skeletal muscle.

Abstract

Context:

The rapidly increasing prevalence of gestational diabetes mellitus (GDM) globally places a growing population at risk for developing type 2 diabetes mellitus (T2DM), particularly those with persistent impaired glucose tolerance (IGT) postpartum.

Objective:

We sought to 1) identify dynamic insulin signaling abnormalities in vivo in a prospective, longitudinal study of GDM women compared to weight-matched pregnant controls both antepartum and postpartum; and 2) determine abnormalities that might distinguish GDM women who normalize their glucose tolerance postpartum from those with persistent IGT.

Design:

Skeletal muscle biopsies were obtained before and after a 75-g glucose load in nine overweight to obese GDM women and 10 weight-matched pregnant controls antepartum and postpartum. Postpartum biopsies were collected in five weight-matched GDM women with IGT (GDM/IGT).

Results:

GDM women had decreased skeletal muscle insulin-stimulated insulin receptor and insulin receptor substrate 1 (IRS1) tyrosine activation and reduced IRS1, concomitant with increased basal IRS1 serine phosphorylation and basal p70 S6-kinase (S6K1) activation, which resolved postpartum. However, GDM/IGT subjects had a persistent impairment in IRS1 activation and increased S6K1 phosphorylation compared to GDM subjects with normal glucose tolerance.

Conclusions:

This study reveals that women with GDM demonstrate impaired IRS1 signaling associated with increased S6K1 activation in skeletal muscle in vivo. This defect is maintained postpartum in GDM/IGT subjects, despite similar body weights and cytokine levels. Given that GDM women with persistent IGT are at a high risk of developing T2DM, understanding how the nutrient-sensitive mammalian target of rapamycin/S6K1 pathway is chronically activated in GDM may lead to important therapies that could prevent the progression to T2DM.

The incidence of gestational diabetes mellitus (GDM) has doubled over the last 10 yr, placing a rapidly growing population of women at risk for developing type 2 diabetes mellitus (T2DM) postpartum (PP) (1). Increasing evidence also suggests that GDM can impart long-term obesogenic and diabetogenic effects on the offspring. This risk may extend to an even greater population of women, with up to 18% being diagnosed with GDM if the International Association of Diabetes in Pregnancy Study Group recommendations are adopted (2). Women with GDM have a 30–50% risk of developing T2DM in the subsequent 5–10 yr. Those with “prediabetes” PP, otherwise referred to as impaired glucose tolerance (IGT) and defined as either impaired fasting glucose (IFG; glucose ≥100 and <126 mg/dl) or a 2-h value of at least 140 and less than 200 mg/dl after a 75-g glucose load, are at an exceptionally high risk, estimated at 17% per year or 80% within 5 yr (3, 4). Women with GDM demonstrate decreased skeletal muscle glucose transport and greater whole body insulin resistance compared with weight-matched pregnant women with normal glucose tolerance (NGT) (5, 6), suggesting that GDM may represent an “unmasking” of the genetic or epigenetic predisposition to T2DM induced by hormonal changes of pregnancy. Very little is known about the mechanisms for excess insulin resistance that place GDM women with IGT or IFG at such a high risk of developing T2DM compared with those women whose glucose tolerance normalizes PP.

Women who develop GDM enter pregnancy with a background of insulin resistance further exacerbated by placental hormones and cytokines, including human placental lactogen, human placental GH, TNF-α, IL-6, IL-10, and reduced levels of adiponectin (7–10). They also exhibit reduced β-cell function (11), as do those who are at highest risk of developing T2DM. Our previous studies of mechanisms for insulin resistance in pregnancy demonstrated a 2- to 3-fold increase in the p85α-subunit of phosphatidylinositol (PI) 3-kinase in skeletal muscle of both normal pregnant and overweight GDM women that reversed 1 yr PP (5, 12–14). These findings are similar to those in transgenic mice engineered to overexpress human placental GH in skeletal muscle (15). Furthermore, both pregnant control (CON) and GDM women have a reduction in insulin receptor substrate 1 (IRS1) activation compared with nonpregnant overweight patients (5). This defect in IRS1 activation was persistent in the overweight GDM subjects who failed to return to their prepregnancy weight 1 yr PP (13) but not in normal-weight, non-GDM subjects (12). Importantly, these overweight GDM women showed no return to insulin sensitivity as measured by a hyperinsulinemic-euglycemic clamp.

These findings leave several important points yet to be addressed. First, the mechanisms for insulin resistance in skeletal muscle of GDM women compared with weight-matched CON have not been studied dynamically after in vivo insulin stimulation. Second, the pattern of IRS1 serine phosphorylation and relevant serine kinases that might dampen both the insulin receptor (IR) and IRS1 signaling cascades are unknown (16–18). Third, the cellular mechanisms that distinguish insulin resistance in GDM women who normalize their glucose tolerance PP compared with those who maintain IGT are unknown. In this prospective, longitudinal investigation, we studied changes in skeletal muscle insulin signaling in overweight to obese pregnant CON and GDM women. Vastus lateralis muscle biopsies were taken before and after a 75-g glucose challenge in the third trimester and PP. Furthermore, the design included a third group of weight-matched GDM women with persistent IGT PP, primarily due to an increase in fasting glucose (IFG), allowing us to dissect out those changes unique to GDM women who maintain IGT PP. Our hypotheses were that: 1) women with GDM compared with weight-matched pregnant CON will demonstrate insulin signaling abnormalities during pregnancy that may or may not resolve PP; and 2) women who maintain IGT PP have greater insulin resistance and may therefore demonstrate persistent abnormalities that underlie their greatly increased risk for developing T2DM.

Subjects and Methods

Study design and enrollment

Paired vastus lateralis skeletal muscle biopsies were obtained in the fasted state and after a 75-g oral glucose challenge to simulate more typical insulin secretion including gut incretin responses. Studies took place both antepartum (AP; 30–32 wk gestation) and PP (9–10 wk after delivery) in 10 overweight to obese pregnant women without GDM [CON, body mass index (BMI), 26–35 kg/m²] and nine BMI-matched, diet-controlled GDM women who normalized their glucose tolerance PP (PP-GDM/NGT). Due to the mild nature of the GDM in the AP-GDM group (diet-controlled alone), all of these patients normalized their glucose tolerance PP. In addition, we prospectively enrolled five weight-matched GDM women to participate in the PP arm of the study after the standard of care 75-g oral glucose tolerance test (OGTT) PP indicated persistent IGT or IFG in the absence of any medical therapy (PP-GDM/IGT) since delivery. These five PP-GDM/IGT women required medical therapy during their pregnancy, making them ineligible for AP enrollment, and therefore, AP data were not prospectively collected in this group. As shown in Table 1, these women met the criteria for prediabetes or IGT based primarily on IFG rather than their 2-h postglucose value (108 vs. 88 mg/dl in the GDM/IGT and GDM/NGT groups, respectively). However, for simplicity, the nomenclature “IGT” is used to designate all GDM women who failed the 75-g OGTT PP by either an IFG or abnormal 2-h postglucose value. All three groups of women were closely matched for age, ethnicity, BMI, and gestational age. Subjects with overt diabetes [fasting blood glucose (FBG) >125 mg/dl; random glucose >200 mg/dl; and/or hemoglobin A1C >6.5%] were excluded. Moreover, those who smoked or had hypertension, renal disease, thrombophilias, preeclampsia, steroid use, or evidence of intrauterine growth restriction were excluded.

Table 1.

AP and PP characteristics of women in CON, GDM/NGT, and PP-GDM/IGT groups

	CON	GDM/NGT	PP-GDM/IGT
n	10	9	5
Age (yr)	31 ± 2	30 ± 1	29 ± 1
% Caucasian	80	89	80
Gravity/parity	1/0	3/2	2/1
BMI (kg/m2)	30 ± 1	32 ± 1	34 ± 1
Weight (kg)	82 ± 2	85 ± 2	89 ± 1
AP
Gestational age at study visit (wk)	31 3/7	31 2/7
Hemoglobin A1C (%)		5 ± 0.2
50-g glucola, mg/dl (mmol/liter)	105 ± 6 (5.8 ± 0.3)	158 ± 5 (8.8 ± 0.3)a	160 ± 2 (8.8 ± 0.1)
3-h 100-g OGTT (FBG), mg/dl (mmol/liter)		90 ± 1 (5.0 ± 0.07)	102 ± 2 (5.7 ± 0.1)c
3-h 100-g OGTT (@ 1 h), mg/dl (mmol/liter)		192 ± 5 (10.7 ± 0.3)	198 ± 9 (11 ± 0.5)c
3-h 100-g OGTT (@ 2 h), mg/dl (mmol/liter)		178 ± 5 (9.9 ± 0.3)	c
3-h 100-g OGTT (@ 3 h), mg/dl (mmol/liter)		125 ± 12 (6.9 ± 0.6)	c
PP
Gestational age at delivery (wk)	40	39	38
Days PP at study visit	70 ± 3	66 ± 4	68 ± 5
Cesarean rate (%)	20	22	20
Infant birth weight (g)	3246 ± 113	3460 ± 143	3247 ± 119
Large for gestational age (#, >90 percentile)	0	1	0
Breast-feeding at PP visit (%)	90	55	20
BMI (kg/m2)	27 ± 1	29 ± 1	31 ± 1a
Weight (kg)	74 ± 2	76 ± 2	82 ± 3
Taking oral contraceptive (%)	30	22	20
75-g 2-h OGTT (FBG), mg/dl (mmol/liter)		88 ± 2 (4.9 ± 0.1)	108 ± 3 (6.0 ± 0.2)b
75-g 2-h OGTT (@ 2 h), mg/dl (mmol/liter)		93 ± 7 (5.1± 0.4)	108 ± 7 (6.0 ± 0.4)

Data are expressed as mean ± sem. Women in the PP-GDM/IGT group participated in a PP study visit only, and their AP weights at 30–32 wk were extracted from the medical record.

Differences assessed using t tests for independent groups: ^a P < 0.05 vs. CON;

^bP < 0.05 vs. GDM/NGT.

^cThe 3-h OGTT was discontinued in subjects as soon as the diagnostic criteria for GDM of two abnormal glucose values manifested, resulting in the absence of 2- and 3-h glucose concentrations.

Each of the women gave their informed consent for participation in the study, which was approved by both the Colorado Multiple Institutional Review Board and the Adult Clinical Translational Research Center. All pregnant women received a screening 50-g glucose challenge (50 g glucola) at 24–28 wk, and those with glucose of at least 135 mg/dl received a 100-g diagnostic 3-h OGTT. Diagnosis of GDM was based on the Coustan and Carpenter criteria adopted by the American Diabetes Association (ADA) and the Fifth International Workshop on GDM (1, 19). All GDM women who were biopsied AP were controlled with diet alone (A1GDM) and were placed on standard diet therapy composed of 50% carbohydrate (primarily complex carbohydrate), 30% fat, and 20% protein (20). Both the GDM and CON groups were given diet recommendations by a nutritionist to limit daily calories to approximately 1800–2000 and weight gain in pregnancy to 11.4 kg. Postpartum, all GDM women received a standard 2-h 75-g OGTT, as recommended by the ADA to determine their glucose tolerance (19).

Skeletal muscle biopsies

Three days before the biopsies, all subjects were counseled by a dietitian to consume a diet similar to the diet prescribed to the GDM women. AP subjects underwent the first biopsy in the overnight-fasted state and again in the opposite limb 60 min after a 75-g oral glucose load. Using a Bergstrom needle, skeletal muscle biopsies (200–300 mg) were obtained from the vastus lateralis under local anesthesia using 1% lidocaine. All CON and GDM subjects returned 9–10 wk PP for a second paired set of biopsies. Postpartum biopsies were performed fasting and 30 min after a 75-g glucose load. The different AP and PP OGTT time points were chosen because maximal insulin levels have been shown to occur 60 min after an oral glucose load while pregnant, but only 30 min after a glucose load while nonpregnant (21). Therefore, PP blood samples and biopsies were collected at baseline and 30 min only.

Plasma analyses

Fasting and post-glucose-load blood samples for insulin, glucose, C-peptide, free fatty acids (FFA), IGF-I, and cytokine panels were collected at both AP and PP biopsy visits. Serum insulin was measured by RIA (Linco Research, St. Charles, MO), glucose using a hexokinase method (Olympus America, Center Valley, PA), and FFA (Wako Chemicals, Richmond, VA) and glycerol (R-Biopharm, Darmstadt, Germany) using enzymatic methods. IGF-I was measured using ELISA (Diagnostic Systems Laboratories, Webster, TX). Inflammatory markers [monocyte chemoattractant protein-1 (MCP-1), IL-6, IL-8, leptin, adiponectin, resistin, and TNF-α) were measured using a Millipore Multiplex Human Cytokine Panel (Millipore, Billerica, MA).

Immunoblot analysis

Frozen skeletal muscle biopsies were homogenized (1 mg:10 ml) in ice-cold lysis buffer as previously described (22), and protein concentration was determined using the Non-Interfering Protein Assay (EMD, Darmstadt, Germany). Insulin signaling proteins were analyzed by immunoblot assay using 50 μg of whole tissue homogenate as described (22). Antibodies for detection of phospho-S6K1(T421/S424), phospho-S6K1(T389), S6K1, phospho-PKCθ(T538), total protein kinase Cθ (PKCθ), phospho-JNK (T183/Y185), phospho-Akt(S473), and total Akt were from Cell Signaling Technology (Danvers, MA); IR, IRS1, p110, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were from Santa Cruz Biotechnology (Santa Cruz, CA); phospho-IRS1(S307), p85α, and phospho-tyrosine were from Millipore; c-Jun N terminal kinase (JNK) was from R&D Systems (Minneapolis, MN); and phospho-IRS1(Y612) was from Invitrogen (Carlsbad, CA). Immunoreactive products were detected using Western Lightning (PerkinElmer, Boston, MA) and exposed to X-OMAT film. Protein levels were quantified using Quantity-One software (Bio-Rad, Hercules, CA) and expressed relative to an internal pooled human muscle standard. GAPDH was measured as a loading control.

RNA extraction and real-time quantitative PCR analysis

Approximately 5–10 mg of crushed, mixed skeletal muscle tissue preserved in RNAlater (QIAGEN, Valencia, CA) was homogenized in 400 μl TRIzol (Invitrogen). Total RNA was extracted using QIAGEN′s RNeasy Mini Kit with DNase treatment and reverse transcribed with iScript cDNA synthesis kit (Bio-Rad). Quantitative PCR was performed using primers for IRS1 with iQ SYBR Supermix (Bio-Rad) following the manufacturer's protocol. Reactions were run in duplicate on an iQ5 Real-Time PCR Detection System (Bio-Rad) along with a no-template control per gene. RNA expression data were normalized to levels of reference gene cyclophilin using the comparative threshold cycle method. Primer sequences were: IRS1-F, 5′-ACC GTC AGT AGC TCA ACT GGA CAT-3′; IRS1-R, 5′-GGG TAC CCA TGA GTT AGA AGA GGA-3′; cyclophilin-F, 5′-AGG GTT TAT GTG TCA GGG TGG TGA-3′; and cyclophilin-R, 5′-ATT TGC CAT GGA CAA GAT GCC AGG-3′.

Statistical analyses

Power

The primary endpoint for the power analysis was tyrosine phosphorylation of the IR (pY-IR) based on our previous studies. Cross-sectional data published by the senior author were used to derive a difference between pregnant CON and GDM women, where 10 women per group was estimated to allow for power of 0.89 with α = 0.05 (5, 23).

Insulin signaling analyses

A repeated-measures two-way ANOVA for main effects of time (AP vs. PP or basal vs. insulin-stimulated) and group (CON vs. GDM) was used for longitudinal comparisons of insulin signaling proteins detected by immunoblot assay. A Tukey post hoc test was performed when significant differences in main effects or interactions were detected. A Student's t test was used for PP comparisons between PP-GDM/NGT and PP-GDM/IGT insulin signaling proteins detected by immunoblot assay. A two-sided P value <0.05 was considered significant.

Plasma analyses

Differences between groups were assessed using a Student's t test for independent groups (AP, CON vs. GDM; PP, CON vs. GDM/NGT and GDM/NGT vs. GDM/IGT; PASW Statistics v17.0; SPSS, Inc., Chicago, IL). A two-sided P value <0.05 was considered significant.

Results

Subjects

Characteristics of CON pregnant and GDM women, and their offspring, were well-matched (Table 1). The mean BMI was in the obese range (BMI ≥ 30 kg/m²) at the time of the 30- to 32-wk biopsy. There was no difference in BMI between GDM/NGT and GDM/IGT subjects PP. More women in the CON group breastfed their infants compared with either GDM group. Similar numbers of women in all three groups had very recently (within 2–3 wk) started oral contraceptives by the time of the PP biopsy (20–30%).

Metabolic variables

AP plasma analyses

There were no significant differences in fasting AP plasma concentrations of glucose, insulin, C-peptide, FFA, IGF-I, TNF-α, or any of the cytokines measured between AP-CON and AP-GDM subjects (Table 2 and Fig. 1). However, in response to a 75-g glucose load, glucose concentrations in AP-GDM were significantly higher at 60 min compared with CON (184 ± 5 vs. 140 ± 8 mg/dl; P < 0.0001; Table 2 and Fig. 1A). In AP-CON, insulin increased 8-fold, compared with 4-fold in AP-GDM subjects at 30 min (P = 0.034), suggesting a defective and delayed insulin response in GDM women, but manifested equivalent peaks at 60 min at the time AP biopsies were taken when insulin signaling studies were performed (Fig. 1B). This pattern was similar in plasma C-peptide measurements (Table 2).

Table 2.

Metabolic parameters for AP and PP study visit days for women in CON and GDM/NGT groups

	AP		PP
	CON	GDM/NGT	CON	GDM/NGT
n	10	9	10	9
IGF-I (ng/ml)	366 ± 46	326 ± 45	189 ± 17	188 ± 17
FFA (μEq/liter)	589 ± 35	683 ± 63	748 ± 113	601 ± 96
TNF-α (pg/ml)	5.1 ± 0.4	5.4 ± 0.5	5.7 ± 0.5	6.3 ± 0.6
MCP-1 (pg/ml)	149 ± 27	187 ± 20	206 ± 41	254 ± 44
IL-6 (pg/ml)	3.3 ± 1.3	1.8 ± 0.2	1.9 ± 0.5	2.4 ± 0.4
IL-8 (pg/ml)	1.6 ± 0.2	1.4 ± 0.2	2.6 ± 0.3	2.5 ± 0.4
Leptin (ng/ml)	1.1 ± 0.1	1.5 ± 0.2	0.6 ± 0.1	1.2 ± 0.2a
Adiponectin (μg/ml)	2172 ± 273	1653 ± 255	1704 ± 183	1716 ± 275
Resistin (μg/ml)	2.5 ± 0.4	2.1 ± 0.4	1.3 ± 0.3	1.6 ± 0.2
Fasting glucose, mg/dl (mmol/liter)	72 ± 3 (4.0 ± 0.1)	77 ± 2 (4.3 ± 0.1)	80 ± 3 (4.4 ± 0.2)	82 ± 2 (4.6 ± 0.1)
Glucose (30 min), mg/dl (mmol/liter)	139 ± 8 (7.7 ± 0.4)	154 ± 6 (8.6 ± 0.3)	123 ± 7 (6.8 ± 0.4)	147 ± 7 (8.2 ± 0.4)a
Glucose (60 min), mg/dl (mmol/liter)	140 ± 8 (7.8 ± 0.4)	184 ± 5 (10.2 ± 0.3)a
Fasting insulin, μU/ml (pmol/liter)	12 ± 1 (71 ± 9)	11 ± 2 (63 ± 10)	4 ± 1 (26 ± 3)	6 ± 1 (39 ± 7)
Insulin (30 min), μU/ml (pmol/liter)	97 ± 20 (583 ± 120)	47 ± 5 (283 ± 31)a	65 ± 18 (391 ± 109)	51 ± 11 (303 ± 63)
Insulin (60 min), μU/ml (pmol/liter)	67 ± 7 (399 ± 45)	67 ± 10 (401 ± 58)
Fasting C-peptide (ng/ml)	2.0 ± 0.1	2.4 ± 0.2	1.6 ± 0.2	1.8 ± 0.2
C-Peptide (30 min) (ng/ml)	8.6 ± 1.0	5.9 ± 0.4a	7.4 ± 1.5	5.7 ± 0.4
C-Peptide (60 min) (ng/ml)	9.4 ± 1.0	9.1 ± 0.8

Data are expressed as mean ± sem.

Differences assessed using t tests for independent groups: ^a P < 0.05.

Fig. 1.

A and B, AP and PP insulin and glucose levels in response to OGTT in CON and GDM subjects at 31 wk AP and 10 wk PP. AP plasma glucose (mg/dl) (A) and insulin (μU/ml) (B) concentrations were measured in CON (open bar) and GDM (black bar) women before (t = 0) and 30 and 60 min after a 75-g glucose drink. C and D, PP plasma glucose (C) and insulin (D) values were measured in CON and GDM women who normalized glucose tolerance (PP-GDM/NGT; black bar) or had IGT (PP-GDM/IGT; gray bar) before (t = 0) and 30 min after a 75-g glucose drink. Plasma data were analyzed using a t test for independent groups. *, P ≤ 0.05 for CON vs. GDM; †, P ≤ 0.05 for PP-GDM/NGT vs. PP-GDM/IGT.

PP plasma analyses

Postpartum fasting glucose concentrations were similar between the PP-CON and PP-GDM/NGT groups (Table 2). However, fasting glucose was elevated at the time of PP biopsies for PP-GDM/IGT compared with PP-GDM/NGT (91 ± 2 vs. 82 ± 2 mg/dl; P = 0.009; Table 3 and Fig. 1C). Fasting insulin concentrations were similar in PP-GDM/NGT compared with PP-CON (Table 2) and were also not different in PP-GDM/NGT vs. PP-GDM/IGT (Table 3 and Fig. 1D). Postpartum fasting FFA, IGF-I, TNF-α, MCP-1, IL-6, IL-8, adiponectin, and resistin were not different between CON and GDM/NGT groups (Table 2), or between GDM/NGT and GDM/IGT (Table 3). In response to a 75-g glucose load, plasma glucose at 30 min was significantly increased in PP-GDM/NGT compared with PP-CON (147 ± 7 vs. 123 ± 7 mg/dl; P = 0.023; Table 2 and Fig. 1C). There were no significant differences in insulin concentrations achieved at 30 min between PP groups (Table 2 and Fig. 1D) when insulin signaling studies were performed in skeletal muscle. No blood samples were taken in the PP state at 60 min.

Table 3.

Metabolic parameters for PP study visit days for GDM women with NGT compared to IGT groups

	GDM/NGT	GDM/IGT
n	9	5
IGF-I (ng/ml)	188 ± 17	164 ± 26
FFA (μEq/liter)	601 ± 96	569 ± 142
TNF-α (pg/ml)	6.3 ± 0.6	5.1 ± 0.6
MCP-1 (pg/ml)	254 ± 44	207 ± 9
IL-6 (pg/ml)	2.4 ± 0.4	2.5 ± 0.6
IL-8 (pg/ml)	2.5 ± 0.4	2.3 ± 0.5
Leptin (ng/ml)	1.2 ± 0.2	1.4 ± 0.4
Adiponectin (μg/ml)	1716 ± 275	1446 ± 494
Resistin (μg/ml)	1.6 ± 0.2	2.3 ± 0.5
Fasting glucose, mg/dl (mmol/liter)	82 ± 2 (4.6 ± 0.1)	91 ± 2 (5.0 ± 0.1)a
Glucose (30 min), mg/dl (mmol/liter)	147 ± 7 (8.2 ± 0.4)	155 ± 8 (8.6 ± 0.4)
Fasting insulin, μU/ml (pmol/liter)	6 ± 1 (39 ± 7)	9 ± 2 (54 ± 11)
Insulin (30 min), μU/ml (pmol/liter)	51 ± 11 (303 ± 63)	42 ± 9 (251 ± 56)
Fasting C-peptide (ng/ml)	1.8 ± 0.2	2.5 ± 0.2
C-Peptide (30 min) (ng/ml)	5.7 ± 0.4	7.4 ± 1.3

Data are expressed as mean ± sem.

Differences assessed using t tests for independent groups: ^a P < 0.05.

GDM is associated with decreased insulin-stimulated IRS1 tyrosine phosphorylation but increased IRS1 serine phosphorylation

There were no differences in pY-IR between GDM and CON skeletal muscle after a 75-g OGTT (Fig. 2, A and C). However, in AP-GDM women, there was a significant increase in basal pY-IR compared with CON. The increase in AP-GDM basal pY-IR results in an overall decrease in insulin-stimulated pY-IR (insulin − basal pY-IR; Δ) compared with all other groups (P < 0.05; Fig. 2, A and C). Downstream of IR, we found that insulin-stimulated IRS1(Y612) phosphorylation significantly increased 2-fold over basal in AP-CON muscle, whereas AP-GDM muscle again showed no significant increase in insulin-stimulated phosphorylation of IRS1 at Y612 [insulin − basal pIRS1(Y612); Fig. 2, A and D]. The lack of insulin activation at IRS1 may be partially due to reduced activation of IR even in the presence of similar insulin concentrations achieved at 60 min. The defects in both pY-IR and IRS1(Y612) activation in GDM muscle was reversed PP. Reduced IRS1 activation was also accompanied by a 50% reduction in IRS1 abundance in AP-GDM muscle (P < 0.05; Fig. 2F). Analysis of skeletal muscle IRS1 mRNA expression revealed a significant increase in IRS1 expression in AP-GDM compared with AP-CON (P < 0.05), suggesting that the reduced protein abundance is not caused by reduced gene expression (Fig. 2G). In addition, there was a significant 2-fold increase in basal IRS1(S312) phosphorylation after correcting for IRS1 abundance in AP-GDM vs. AP-CON that returned to CON levels PP (Fig. 2H). There were no differences in PI 3-kinase p85α regulatory or p110α catalytic subunit abundance between AP-CON and AP-GDM (data not shown), although both groups demonstrated a significant reduction in p85α abundance PP as demonstrated previously (12, 13). We also found no significant difference in basal or OGTT-stimulated Akt phosphorylation between AP and PP CON and GDM subjects at both the S473 site (Fig. 2, A and E) and the T308 site (Fig. 2A).

Fig. 2.

Insulin activation of IR and IRS1 is reduced in skeletal muscle from GDM women during pregnancy and restored PP. A, Representative immunoblots for basal- and OGTT-stimulated phosphorylation of IR, IRS1, and Akt are shown. M designates molecular weight marker lane. B, Representative immunoblots for basal IRS1 abundance and IRS1(S312) phosphorylation are shown. M designates molecular weight marker lane. C–E, pY-IR (C), IRS1(Y612) phosphorylation (D), and pAkt(S473) activation (E) were quantitated from immunoblot assay in skeletal muscle homogenates from CON and GDM women both AP and PP before (open bar) and after (black bar) a 75-g glucose drink. F and G, IRS1 protein abundance (F) and IRS1 mRNA expression (G) were measured in CON (open bar) and GDM (black bar) women AP and PP. H, Immunoblot analysis of IRS1(S312) phosphorylation is shown expressed relative to IRS1 abundance in basal skeletal muscle homogenates. GAPDH was used as a loading control. Data were analyzed by repeated measures two-way ANOVA with a Tukey post hoc analysis. *, P ≤ 0.05 indicates significant differences for basal vs. OGTT and AP vs. PP; §, P ≤ 0.05 for CON vs. GDM.

Increased activation of S6K1 in GDM skeletal muscle

IRS1 can be serine phosphorylated through numerous pathways, including the JNK, IκB kinase, PKCθ, and S6K1 pathways (24, 25). In our weight-matched populations, no differences were detected in the phosphorylation or abundance of inflammatory kinases PKCθ [basal pPKCθ(T538)/total, AP-CON 1.06 ± 0.15, PP-CON 1.00 ± 0.10, AP-GDM 0.99 ± 0.19, PP-GDM 0.90 ± 0.24] or JNK [basal and OGTT pJNK(T183/Y185)/total, basal AP-CON 1.32 ± 0.34, OGTT AP-CON 1.00 ± 0.25, basal PP-CON 1.00 ± 0.38, OGTT PP-CON 0.80 ± 0.23, basal AP-GDM 1.35 ± 0.36, OGTT AP-GDM 1.54 ± 0.25, basal PP-GDM 2.21 ± 0.52, OGTT PP-GDM 1.77 ± 0.78] in skeletal muscle (Fig. 3A) or in plasma levels of TNF-α (Table 2) between AP-GDM and AP-CON. In our previous studies using a mouse model of spontaneous GDM, GDM pregnancy induced serine phosphorylation of IRS1 through activation of S6K1 (15). We therefore measured S6K1 phosphorylation at two sites. Phosphorylation of S6K1(T389), the insulin-responsive site, was stimulated 2- to 3-fold after OGTT with no difference between CON and GDM muscle (Fig. 3, A and B). There was, however, a significant approximately 2-fold increase in skeletal muscle basal S6K1(T421/S424) phosphorylation in AP-GDM vs. AP-CON (Fig. 3, A and C), with no change in total S6K1. Furthermore, basal S6K1(T421/S424) phosphorylation trended (P = 0.06) to return to CON levels in PP-GDM muscle.

Fig. 3.

Increased S6K1 phosphorylation in skeletal muscle from GDM subjects during pregnancy. A, Representative immunoblots for phosphorylation of S6K1, PKCθ(T538), and JNK(T183/Y185) are shown. M designates molecular weight marker lane. B and C, S6K1 phosphorylation at amino acid sites T389 (B) and T421/S424 (C) were measured in basal (open bar) and glucose-stimulated (closed bar) skeletal muscle homogenates from vastus lateralis biopsies taken from CON and GDM both AP and PP. Data were analyzed by repeated measures two-way ANOVA with a Tukey post hoc analysis. *, P ≤ 0.05 indicates significant differences for basal vs. OGTT; §, P ≤ 0.05 for CON vs. GDM.

IRS1 serine phosphorylation remains elevated PP in skeletal muscle from IGT GDM women

Our second goal was to compare cellular mechanisms for skeletal muscle insulin resistance between GDM subjects who normalized glucose tolerance PP (PP-GDM/NGT) vs. GDM women with IFG or IGT (PP-GDM/IGT). PP-GDM/IGT subjects had higher fasting glucose concentrations than PP-GDM/NGT both at the time of PP biopsy (Table 3 and Fig. 1C) and during their routine 75-g OGTT (Table 1). Despite similar insulin levels at 30 min (Table 3 and Fig. 1D), skeletal muscle from PP-GDM/NGT but not PP-GDM/IGT had a 2-fold increase in levels of insulin-stimulated IRS1(Y612) phosphorylation (P < 0.05; Fig. 4, A and C). Reduction in insulin-stimulated IRS1 activation led to a significant decrease in insulin-stimulated Akt(S473) phosphorylation in PP-GDM/IGT compared with PP-GDM/NGT muscle (Fig. 4, B and C). The reduction in Akt activation was not related to a difference in the abundance of p85α regulatory or p110 catalytic subunit of PI 3-kinase between PP-GDM/IGT and PP-GDM/NGT (data not shown).

Fig. 4.

Increased IRS1 serine phosphorylation remains elevated in GDM women with IGT PP. A and B, IRS1(Y612) phosphorylation (A) and Akt(S473) phosphorylation (B) were measured by immunoblot assay in skeletal muscle homogenates from vastus lateralis biopsies taken before (open bar) and 30 min after (closed bar) a 75-g glucose drink in GDM women that were diagnosed with NGT (PP-GDM/NGT) or IGT (PP-GDM/IGT). D and E, IRS1(S312) phosphorylation (D) and S6K1(T421/S424) phosphorylation (E) were measured by immunoblot in basal muscle from PP-GDM/NGT (black bar) and PP-GDM/IGT (gray bar). C and F, Representative immunoblots are shown for glucose-stimulated pIRS1(Y612) and pAkt(S473) (C) and for basal phosphorylation of IRS1(S312), S6K1(T421/S424), PKCθ(T538), and JNK(T183/Y185) (F). Data for A and B were analyzed by repeated measure one-way ANOVA; *, P ≤ 0.05 for basal vs. OGTT; §, P ≤ 0.05 for NGT vs. IGT. For D and E, *, P ≤ 0.05 for significant differences by t test.

Similar to the AP-GDM group, we found that PP-GDM/IGT subjects had significantly increased basal IRS1(S312) phosphorylation (Fig. 4D). There were no differences detected in skeletal muscle IRS1 abundance between PP-GDM/NGT and PP-GDM/IGT subjects (data not shown). We also found no differences in abundance or phosphorylation of inflammatory kinases PKCθ [basal pPKCθ(T538)/total, PP-NGT 0.89 ± 0.15, PP-IGT 0.87 ± 0.14] and JNK [basal pJNK(T183/Y185)/total, basal PP-NGT 0.81 ± 0.17, PP-IGT 0.58 ± 0.15] in PP-GDM/IGT vs. PP-GDM/NGT muscle (Fig. 4F). By contrast, skeletal muscle from PP-GDM/IGT subjects had a near 2-fold increase in basal S6K1(T421/S424) phosphorylation compared with PP-GDM/NGT subjects (P = 0.05; Fig. 4, E and F).

Discussion

We have previously shown that insulin resistance in skeletal muscle in vitro from pregnant obese GDM women was associated with impaired insulin signaling due to reduced activation of the key docking protein, IRS1 (5, 13), and that this was associated with increased TNF-α mRNA in skeletal muscle (13). In agreement with our previous studies in rectus abdominis skeletal muscle studied in vitro after stimulation with maximal insulin concentration (5), we show here that pregnant women with GDM have impaired IR and IRS1 tyrosine phosphorylation and lower IRS1 abundance in vastus lateralis muscle in response to an oral glucose load compared with weight-matched NGT pregnant subjects. GDM subjects in the present study had a 50% IRS1 depletion in skeletal muscle. We also found that serine phosphorylation of IRS1(S312) was significantly increased in AP-GDM compared with AP-CON during late pregnancy and returned to normal levels as insulin resistance improved after pregnancy.

Increased IRS1 serine phosphorylation has been shown to interfere with both tyrosine phosphorylation of IRS1 (26) and the association between IRS1 and PI 3-kinase (27, 28). In addition, increased phosphorylation of IRS1 at S312 as well as other IRS1 serine sites has been shown to promote IRS1 protein degradation (29, 30). The increase in IRS1 mRNA in AP-GDM vs. AP-CON in this study supports the hypothesis that increased serine phosphorylation may trigger IRS1 protein degradation and suppress IRS1 tyrosine phosphorylation and may be a major mechanism for skeletal muscle insulin resistance in GDM women. Whether this is due to the increase in phosphorylation of IRS1(S312) or a combination of serine sites in GDM subjects is not proven here. Measurement of IRS1(S312) phosphorylation alone does not necessarily imply causality for IRS1 degradation.

In our diet-controlled GDM women, IRS1 levels and serine phosphorylation returned to normal PP along with improved glucose tolerance, suggesting that women with mild GDM have increased susceptibility to activation of one or more serine kinases during pregnancy. We found that the nutrient-responsive serine kinase S6K1, downstream from mammalian target of rapamycin (mTOR), was increased in basal skeletal muscle from GDM subjects, compared with weight-matched women with NGT, and trended to return to normal PP. Both mTORC1 and S6K1 are key elements of the negative feedback loop that inhibit insulin-induced PI 3-kinase activity through phosphorylation of specific serine residues on IRS1 (31). This loop acts as a nutrient sensor integrating a number of environmental signals such as glucose, insulin, and amino acids (32, 33). S6K1 activation has previously been shown to be elevated by hyperglycemia (34), by hyperinsulinemia in liver and fat (35), by chronic high-fat diet in muscle and adipose tissue (36), and by amino acid infusion in humans (37). However, how nutrients drive the mTOR/S6K pathway is not completely understood.

Because S6K1 is activated both in response to insulin and as an inducer of insulin resistance, it is hard to tell which comes first, skeletal muscle insulin signaling defect(s) or hyperinsulinemia. One attractive hypothesis is that mTOR/S6K1 is active in both the islet and skeletal muscle to cause abnormal insulin sensitivity and reduced insulin secretion. Recently, transgenic mice were engineered to overexpress S6K1 in pancreatic islets alone (38). These experiments demonstrated that in vivo activation of mTORC1/S6K down-regulated IRS signaling in the islet leading to β-cell insulin resistance that ultimately resulted in β-cell failure. Thus, under conditions of nutrient overload, both β-cells and skeletal muscle insulin signaling could be failing simultaneously via the same mechanism(s).

We did not analyze the levels of IκB kinase-β or suppressor of cytokine signaling 3, both of which have been implicated in inflammation-linked serine phosphorylation of IRS1 in rodents (39). It is important to note that the GDM subjects studied here had similar levels of inflammatory adipokines compared with CON, suggesting, albeit indirectly, that these inflammation-linked serine kinases, such as JNK, may not be a major cause of the insulin resistance. Although our data allow us to imply that S6K1 is involved in insulin resistance in GDM, it remains possible that other IRS protein kinases could potentially contribute to even further worsening of insulin resistance. Furthermore, Garvey et al. (40) demonstrated a unique cellular translocation defect in adipocytes from GDM subjects compared with lean control pregnant subjects, associated with insulin-responsive glucose transporter type 4 depletion. In skeletal muscle of insulin-resistant subjects, glucose transporter 4 levels are normal; however, there is evidence for defects in the trafficking and translocation (41).

Women with GDM who do not revert to NGT PP are at an extraordinarily high risk for conversion to T2DM. Longitudinal PP studies examining these high-risk women have demonstrated that in addition to insulin resistance, a β-cell defect plays a major role in subsequent development of T2DM (11, 42). The results presented in this prospective study comparing insulin signaling in weight-matched GDM subjects with or without persistent IGT validates previous in vitro human studies (5, 12) and in vivo studies in GDM animal models (15) and suggests that the nutrient-sensing S6K1 pathway may be especially important for reduced skeletal muscle insulin signaling in IGT subjects, and possibly reduced insulin secretion under conditions of nutrient overload (38). Understanding the nutrient signatures in GDM women that may regulate the S6K1 pathway may direct therapeutic targets to improve insulin sensitivity and prevent the development of T2DM and its associated cardiovascular morbidity.