Programming of central and peripheral insulin resistance by low birthweight and postnatal catch-up growth in male mice

This research has been regulated under the Animals (Scientific Procedures) Act 1986 Amendment Regulations 2012 following ethical review by the University of Cambridge Animal Welfare and Ethical Review Body. C57BL/6J mice were purchased from Charles River (Harlow, UK). Female mice were mated at 6 weeks old. Dams were randomised to either a control diet (20% protein) or an isoenergetic low-protein diet (8% protein) fed ad libitum during gestation and lactation. Diet assignment was performed by an animal technician who was not involved in the outcome assessments.

Control and low-protein diets were from Arie Blok (Woerden, the Netherlands) [16]. Cross-fostering was performed at 3 days of age to establish two groups: control (offspring born to and suckled by dams fed a control-diet) and recuperated (offspring of low-protein-diet-fed dams nursed by control-diet fed dams). To maximise the effects of maternal nutritional differences on offspring postnatal growth (and recapitulate low birthweight followed by accelerated postnatal growth observed in humans), recuperated litters were culled to four pups and control litters were culled to eight pups on day 3. Body weights of animals were recorded at 3, 7, 14 and 21 days (control n = 30; recuperated n = 26) and weekly, thereafter (control n = 11; recuperated n = 8). Animals were caged in groups of two littermates and maintained on a 12 h light–dark cycle at 21–22°C with ad libitum access to chow diet (LAD1; Special Diet Services, Crawley, UK) and water. Ad libitum food intake in the home cage was measured weekly by calculating food consumed from food hoppers and the measurements were used to calculate cumulative food intake per mouse from 4–10 weeks of age. Because of the known effects of female oestrogen levels on insulin sensitivity [20] and the technical difficulties of oestrous cycle staging in mice, only male offspring were studied. All assessments were conducted at 12 weeks ± 3 days of age. For both groups, one male mouse in each litter underwent a different experimental procedure: (1) hyperinsulinaemic–euglycaemic clamp and related plasma analyses (control, n = 9; recuperated, n = 8); (2) body composition analysis and plasma and tissue collection (control, n = 11; recuperated, n = 8); (3) intracerebroventricular (ICV) insulin injection (control, n = 7; recuperated, n = 5); (4) GTT (control, n = 11; recuperated, n = 6).

Mice were fasted for 16 h overnight and injected i.p. with a 10% (wt/vol.) glucose solution at 1 g/kg body weight. Blood glucose levels were measured from the tail vein using a hand-held glucose monitor before and 15, 30, 60, 120 and 180 min after glucose injection.

After an overnight fast, animals were anaesthetised with an i.p. injection (10 μl/g) of Vetranquil (Novartis, Camberley, UK), Dormicum (Roche, Burgess Hill, UK), fentanyl (Martindale Pharma, Wooburn Green, UK) (VDF; 1:2:10 in 3 parts water) and the tail vein was cannulated. Programmable ‘Aladdin’ syringe pumps (World Precision Instruments, Hitchin, UK) were used for automated infusions of glucose and insulin via the tail vein. Blood glucose measurements and baseline blood samples were taken at t = −10 and t = 0 min via tail bleed. The clamp began with a primed-continuous (50 μl/h; 3.5 mU kg⁻¹ min⁻¹) i.v. infusion of insulin (Actrapid, Novo Nordisk, Gatwick, UK). During the 90 min insulin infusion, blood glucose was measured every 5 min for the first 20 min and every 10 min thereafter. To maintain euglycaemia relative to basal blood glucose levels, a variable i.v. infusion of 12.5% (wt/vol.) d-glucose was administered accordingly. The glucose infusion rate (GIR; μl/h) was used as an indicator of peripheral insulin action. Steady-state blood samples were taken at 10 min intervals during the final 30 min (t = 70 to t = 90 min) of the clamp, for determination of plasma glucose, insulin and NEFA.

Blood samples were drawn from the tail tip into heparin-coated capillary tubes, chilled on ice and centrifuged to separate out plasma. Commercially available kits were used to determine plasma levels of insulin (Ultra-sensitive mouse insulin ELISA, Crystal Chem, IL, USA) and NEFA (Wako NEFA HR kit, Alphalabs, Eastleigh, UK). TNFα was measured by the MRC MDU Mouse Biochemistry Laboratory (Addenbrookes Hospital, Cambridge, UK). All measurements were carried out as technical duplicates and results were verified with a CV of <5%.

Body composition was determined at 09:00 in fed animals using time-domain nuclear magnetic resonance (TDNMR; Minispec Plus, Bruker, Billerica, MA, USA). Body weight was recorded before TDNMR for calculation of relative fat mass and lean mass.

Mice were placed in a Meta-Trace (Creative Scientific, UK) home cage measurement system linked to a Minimox system developed in house (P. Murgatroyd, Cambridge Institute of Metabolic Science and NIHR/Wellcome Trust Clinical Research Facility, Addenbrooke's Hospital, Cambridge, UK) to measure oxygen consumption (\( {\overset{\cdot }{V\mathrm{O}}}_2 \)) and carbon dioxide production (\( \overset{\cdot }{V{\mathrm{CO}}_2} \)). Ventilation rates were 390 ml/min and samples were collected every 18 min over 48 h. Energy expenditure (EE) was analysed over a 24 h period, using the equation: EE (J) = 15.818 × \( {\overset{\cdot }{V\mathrm{O}}}_2 \) + 5.176 × \( \overset{\cdot }{V{\mathrm{CO}}_2} \) . The respiratory exchange ratio (RER) was calculated as \( \overset{\cdot }{V{\mathrm{CO}}_2} \)/\( {\overset{\cdot }{V\mathrm{O}}}_2 \).

Mice were single-housed for 5 days prior to surgery and body weight recorded daily. Mice were anaesthetised with a mix of inhaled isoflurane and oxygen via nose cone and placed in a stereotactic device. A 26-gauge stainless-steel guide cannula (C315G-SPC; Plastics One, Roanoake, VA, USA) was implanted into the right lateral ventricle at +1.0 mm lateral, −0.5 mm caudal and −2.0 mm ventral to the bregma, fixed to the skull with cyanoacrylate adhesive and secured with dental cement (Associated Dental Products, Swindon, UK). To prevent occlusion, a dummy cannula (C315DC-SPC; Plastics One) was fitted. All animals received subcutaneous analgesia and antibiotic (Rimadyl, 5 mg/kg; Terramycin LA, 60 mg/kg; both Pfizer, Kent, UK). Body weight was monitored daily following surgery and intracerebroventricular insulin injections, with subsequent monitoring of food intake over a 20 h period, were performed 6–7 days later, once pre-surgical body weight had been recovered. Correct cannula placement was confirmed using an angiotensin-II thirst test (human A9525, Sigma-Aldrich, Gillingham, UK; 200 ng dose in saline [154 mmol/l NaCl] vehicle) immediately after the intracerebroventricular insulin injection experiment. Only animals that showed a rapid dipsogenic response (drink onset within 15 s to 3 min post injection) following angiotensin-II injection were included in the analyses.

Two hours before the onset of the dark period, food was removed from the cage and insulin (Actrapid; diluted in saline to 0.2 μU/μl) was injected into the lateral ventricle of awake, free-to-roam mice using a Hamilton syringe (Sigma-Aldrich) attached to the indwelling cannula. An insulin dose of 0.4 μU was selected based on trial data and previous reports of insulin-induced inhibition of food intake in mice [21, 22]. At the start of the dark period (18:00, t = 0), a pre-weighed quantity of food was returned to the cage and food intake was measured at t = 2, 4, 12 and 20 h following food presentation. For comparison, baseline food intake measurements were taken at the same time points the day prior to insulin injection; the dummy cannula was manipulated (removed and replaced) while mice were handled in the same way as for the insulin injection process, except no injection was performed. Pre-surgery food intake of single caged animals was calculated for the 12 h period immediately prior to surgery.

Individual hypothalamic nuclei (ARC, PVH and VMH) were microdissected from 1 mm-thick coronal sections of frozen brains cut using a brain matrix (Braintree Scientific, Braintree, MA, USA). After microdissection, RNA was extracted using RNeasy MinElute columns with on-column DNase treatment according to the manufacturer’s instructions (Qiagen, Manchester, UK), and RNA was reverse transcribed using a High-Capacity cDNA Reverse Transcription kit (Thermo Fisher, Loughborough, UK). Insr, Irs1, P85α (also known as Pik3r1) and Pik3cb mRNA expression was measured in technical duplicates using SYBR quantitative PCR (Thermo Fisher). Irs2 and Ptpn1 (which encodes protein tyrosine phosphatase 1B [PTP1B]) mRNA was measured using TaqMan assay-on-demand (Irs2: Mm03038438_m1; Ptpn1: Mm00448427_m1; both Thermo Fisher). Target gene expression was normalised to the housekeeping gene Sdha (expression of which did not differ between groups). All quantitative PCR was performed using a StepOne Plus qPCR machine (Thermo Fisher).

ARC was microdissected as above. Protein was extracted in lysis buffer (50 mmol/l HEPES [pH 8], 150 mmol/l NaCl, 1% (wt/vol.) Triton X-100, 1 mmol/l Na₃VO₄, 30 mmol/l NaFl, 10 mmol/l Na₄P₂O₇, 10 mmol/l EDTA (all Sigma-Aldrich) and a 1:200 dilution of protease inhibitor cocktail set III [Merck-Millipore, Watford, UK]). Total protein concentration of lysates was determined using a bicinchoninic acid kit (Sigma-Aldrich) and samples diluted in Laemmli buffer. Samples (4 μg protein) were loaded onto 8% polyacrylamide gels for electrophoresis and transferred to a polyvinylidene difluoride (PVDF) membrane (Merck-Millipore). The membrane was washed, then stained with Ponceau red (Sigma-Aldrich). The stained membrane was imaged and bands measured for normalisation. Membranes were blocked in 5% (wt/vol.) BSA in Tris buffered saline with Tween 20 (TBST), followed by overnight incubation with the following primary antibodies: insulin receptor β subunit (sc-711, Santa Cruz Biotechnology, Dallas, TX, USA; 1:200 dilution), p-IRS1 Ser307 (07-247, Upstate Biotechnology, Lake Placid, NY, USA; 1:1000 dilution) and phosphoinositide 3-kinase (PI3K) p110β (ab151549, Abcam, Cambridge, UK; 1:1000 dilution). All antibodies were validated as cross-reactive to mouse. Following 3 × 5 min washes in TBST, membranes were incubated for 1 h in horseradish-peroxidase (HRP)-linked secondary antibody (Abcam) diluted 1:10,000 in 5% BSA in TBST. Antibody binding was detected using chemiluminescence substrate and the ChemiDoc Imaging system (Bio-Rad, Hemel Hempstead, UK). Protein abundance was quantified by band densitometry using Image Lab 5.1 software (Bio-Rad) and normalised to total transferred protein as visualised by Ponceau stain.

All data were analysed using Prism 7 software (GraphPad, La Jolla, CA, USA). Values are expressed as mean ± SEM. For data comparing one variable between offspring, an unpaired t test was used. In the ICV insulin injection experiments, a paired t test was used to compare measurements at baseline and after insulin injection from the same animals. For experiments measuring data over time, a two-way ANOVA with repeated measures was used to determine the overall effect of maternal diet and time. A Sidak post hoc test was carried out if the overall p value for two-way ANOVA was < 0.05.

Recuperated offspring showed accelerated postnatal growth and were significantly heavier than controls by the end of the first postnatal week (Fig. 1a). Recuperated offspring remained significantly heavier than controls throughout the remainder of the lactation period (p < 0.001). However, there was no difference in body weight between control and recuperated offspring from 4 to 12 weeks of age, or in whole-body fat mass as measured by TDNMR (as a percentage of body weight) at 12 weeks of age (Fig. 1b,c). Despite no change in overall adiposity, recuperated offspring had a significantly larger epididymal fat depot than controls when expressed as a percentage of whole body weight (p < 0.05; Fig. 1d).

To assess energy balance, ad libitum cumulative food intake per mouse in the home cage was measured weekly for the period of 4–10 weeks of age. There was no difference in cumulative food intake between control and recuperated offspring (Fig. 1e). At 12 weeks, EE was lower in recuperated offspring compared with controls (p < 0.05; Fig. 1f). The RER was also significantly lower in recuperated offspring compared with controls (p < 0.05; Fig. 1g).

Control and recuperated offspring underwent an i.p. GTT at 12 weeks of age. There was an overall effect of maternal diet on glucose levels during the test (p < 0.001; Fig. 2a). Blood glucose was significantly higher in recuperated animals 120 min following i.p. glucose injection compared with controls (p < 0.05). Furthermore, recuperated offspring had a significantly higher AUC over the course of the GTT compared with controls (p < 0.05; Fig. 2b).

To determine whole-body insulin sensitivity, hyperinsulinaemic–euglycaemic clamps were carried out in offspring. Compared with baseline, in both groups, plasma insulin levels increased significantly following insulin infusion (control: 75.86 ± 6.90 to 901.70 ± 96.55 pmol/l; recuperated: 68.96 ± 10.34 to 665.50 ± 110.34 pmol/l; p < 0.01 compared with baselines). Plasma glucose levels were not significantly different at baseline and hyperinsulinaemic steady states in control or recuperated groups (Fig. 2c). There was a significant effect of maternal diet on GIR throughout the clamp (Fig. 2d), with GIR lower in recuperated offspring (p < 0.01). Furthermore, mean GIR during steady state was lower in recuperated offspring compared with controls (p < 0.05; Fig. 2e). Compared with baseline, plasma NEFA levels were significantly lower during hyperinsulinaemia in both groups (control: 0.44 ± 0.05 to 0.20 ± 0.04 mmol/l; recuperated: 0.27 ± 0.04 to 0.16 ± 0.04 mmol/l; p < 0.001 compared with baseline); the percentage drop in NEFA levels from baseline to hyperinsulinaemic steady state was not significantly different between groups (control: 56.2% ± 5.3%; recuperated: 42.8% ± 6.6%).

There was no difference in pre-surgery baseline food intake over a 12 h period between control and recuperated offspring (Fig. 3a). ICV insulin injection resulted in a significant reduction in food intake in control mice during the first 4 h compared with baseline (p < 0.01; Fig. 3b). This was followed by hyperphagia compared with baseline food intake from 12 to 20 h after food presentation (p < 0.05). In contrast to control animals, there was no effect of insulin on food intake compared with baseline in recuperated offspring (Fig. 3c). There was no effect of insulin injection on body weight in response to ICV insulin injection in either group (data not shown).

Gene expression of components of the insulin signalling pathway were measured in distinct hypothalamic nuclei capable of sensing and integrating insulin signalling into an alteration of food intake. There was no difference in the expression of Insr (encoding the insulin receptor), Irs1 (encoding IRS1) or Irs2 (encoding IRS2) in the ARC, VMH or PVH between control and recuperated offspring (Fig. 4a–c). There was also no difference in the expression of Pik3cb and P85α, which encode the catalytic and regulatory subunits of PI3K, respectively, in any of the nuclei examined (Fig. 4a–c). We next examined the expression of PTP1B (encoded by Ptpn1), which is known to negatively regulate insulin signalling in the hypothalamus [19]. There was a significant increase in Ptpn1 expression in the ARC (p < 0.01; Fig. 4a) but not PVH or VMH (Fig. 4b,c). Previous studies have shown that PTP1B expression is increased by TNFα, the levels of which are often associated with increased inflammation seen with obesity [23]. However, there was no difference in serum TNFα levels between control and recuperated offspring (control: 7.84 ± 0.52 ng/ml; recuperated: 9.06 ± 0.57 ng/ml) in the present study.

There was no difference in the expression of the insulin receptor in the ARC in recuperated offspring (Fig. 5a). However, recuperated offspring showed increased levels of IRS1 phosphorylated at Ser307 (P-IRS1; p < 0.01; Fig. 5b) and decreased levels of the p110β catalytic subunit of PI3K (p < 0.01; Fig. 5c) in the ARC. Antibody staining for each protein, along with loading control, are shown in Fig. 5d.