IGF-1 infusion to fetal sheep increases organ growth but not by stimulating nutrient transfer to the fetus

Abstract

Insulin-like growth factor-1 (IGF-1) is an important fetal growth factor. However, the role of fetal IGF-1 in increasing placental blood flow, nutrient transfer, and nutrient availability to support fetal growth and protein accretion is not well understood. Catheterized fetuses from late gestation pregnant sheep received an intravenous infusion of LR3 IGF-1 (LR3 IGF-1; n = 8) or saline (SAL; n = 8) for 1 wk. Sheep then underwent a metabolic study to measure uterine and umbilical blood flow, nutrient uptake rates, and fetal protein kinetic rates. By the end of the infusion, fetal weights were not statistically different between groups (SAL: 3.260 ± 0.211 kg, LR3 IGF-1: 3.682 ± 0.183; P = 0.15). Fetal heart, adrenal gland, and spleen weights were higher (P < 0.05), and insulin was lower in LR3 IGF-1 (P < 0.05). Uterine and umbilical blood flow and umbilical uptake rates of glucose, lactate, and oxygen were similar between groups. Umbilical amino acid uptake rates were lower in LR3 IGF-1 (P < 0.05) as were fetal concentrations of multiple amino acids. Fetal protein kinetic rates were similar. LR3 IGF-1 skeletal muscle had higher myoblast proliferation (P < 0.05). In summary, LR3 IGF-1 infusion for 1 wk into late gestation fetal sheep increased the weight of some fetal organs. However, because umbilical amino acid uptake rates and fetal plasma amino acid concentrations were lower in the LR3 IGF-1 group, we speculate that animals treated with LR3 IGF-1 can efficiently utilize available nutrients to support organ-specific growth in the fetus rather than by stimulating placental blood flow or nutrient transfer to the fetus.

NEW & NOTEWORTHY After a 1-wk infusion of LR3 IGF-1, late gestation fetal sheep had lower umbilical uptake rates of amino acids, lower fetal arterial amino acid and insulin concentrations, and lower fetal oxygen content; however, LR-3 IGF-1-treated fetuses were still able to effectively utilize the available nutrients and oxygen to support organ growth and myoblast proliferation.

INTRODUCTION

Many pregnancy complications that result in fetal growth restriction or fetal overgrowth have either low- or high-circulating fetal insulin-like growth factor (IGF-1), respectively (1, 2). In human pregnancies, multiple studies have shown that IGF-1 umbilical cord blood concentrations strongly correlate with birth weight (3–6). Humans with mutations in the IGF-1 receptor gene demonstrate intrauterine growth restriction (7). In rodent animal models, IGF-1 and IGF-1 receptor gene inhibition leads to fetal growth restriction (8, 9). Conversely, previous studies in sheep have shown that IGF-1 supplementation augments fetal growth. Infusion of an IGF-1 analog to the sheep fetus increased fetal weight (10), although this result has not been observed in all studies, despite similar infusion length (11, 12). However, even when whole body fetal weight does not significantly increase after infusion of IGF-1, select organs are larger in the IGF-1 infused groups such as heart, lungs, liver, kidneys, adrenal glands, and spleen (11, 12). IGF-1 given to growth-restricted fetal sheep either by intravenous or by intra-amniotic route improves fetal growth rates (13, 14). One of the primary organs impacted by IGF-1 in rodents is skeletal muscle; IGF-1 is implicated in both the proliferation of myoblasts and their differentiation into myotubes in postnatal muscle cells (15). IGF-1 receptor gene knockout mice have muscle hypoplasia with lower numbers of myocytes (9), while overexpression of IGF-1 in adult mice and infusion of recombinant human IGF-1 into adult rats leads to muscle hypertrophy (16, 17).

Despite abundant evidence that IGF-1 promotes fetal growth and skeletal muscle growth, the role of IGF-1 in increasing placental blood flow, nutrient uptake, and fetal nutrient availability to support faster fetal and skeletal muscle growth is not well understood. IGF-1 is known to stimulate uptake of glucose and alpha-(methylamino)isobutyric acid, an amino acid analog, in vitro in human placental trophoblast cells (18, 19). Previous studies have also investigated the effect of in vivo IGF-1 infusion on placental expression and activity of nutrient transporters, but the results have been variable. IGF-1 intra-amniotic infusion to the growth restricted sheep fetus yielded higher placental mRNA expression of amino acid transporters (14), which may have increased nutrient transport to the fetus to support growth. However, in normally growing late gestation fetal sheep, IGF-1 infusion reduced placental clearance of amino acid and glucose analogs (20). Within the fetus, infusion of IGF-1 over a period of several hours led to changes in maternal and fetal glucose and nitrogen concentrations that suggested increased fetal and placental uptake of glucose and amino acids (21). Infusions of recombinant human IGF-1 for several hours also resulted in lower protein breakdown and leucine oxidation rates (22, 23). However, the effects of a 1-wk infusion of LR-3 IGF-1, an IGF-1 analog, to the fetus on placental nutrient transport and fetal protein kinetics have not been studied.

Therefore, our objective was to determine the effect of fetal infusion of LR3 IGF-1 for 1 wk on net uterine and umbilical nutrient uptake rates, fetal nutrient availability, and fetal protein kinetic rates in late gestation fetal sheep. We hypothesized that LR3 IGF-1 infusion would increase placental transport of glucose and amino acids to the fetus therefore increasing nutrient availability to the fetus. We used chronically catheterized fetal sheep, allowing for infusion of LR3 IGF-1 directly to the fetus and subsequent measurement of uterine and umbilical blood flow; umbilical transport of substrates including glucose, amino acids, lactate, and oxygen to the fetus; circulating concentrations of fetal nutrients; and whole fetal protein accretion rates. We also measured skeletal muscle myoblast proliferation and myofiber area in a group of animals similarly treated with LR3 IGF-1.

METHODS

Ethical Approval

Two cohorts of animals were used for the following studies. Study protocols for the primary cohort of animals were approved by the Institutional Animal Care and Use Committee at the University of Colorado Anschutz Medical Campus [Protocol 77617(10)1E] and are in compliance with guidelines from AAALAC International. Experiments were performed at the Perinatal Research Facility on the Anschutz Medical Campus. Study protocols for the secondary cohort of animals were approved by the Institutional Animal Care and Use Committee (Protocol IP0070) at Oregon Health & Science University in Portland, OR, and are in compliance with guidelines from AAALAC International.

Surgical Procedure

For the primary cohort of animals, Columbia-Rambouillet mixed-breed sheep (Nebeker Ranch, Lancaster, CA) with singleton pregnancies (n = 25) underwent maternal laparotomy and hysterotomy under general anesthesia for fetal and maternal catheter placement at 119 ± 0 days gestational age (dGA, term = 147 dGA). Sheep were fasted for 24 hr and thirsted for 12 h before surgery. Pregnant sheep were given diazepam (0.2 mg·kg⁻¹) and ketamine (20 mg·kg⁻¹) through a superficial vein to induce general anesthesia, and they were maintained on isoflurane inhalation anesthesia (2%–4%) for the duration of the surgical procedure. Depth of anesthesia was determined and maintained in response to maternal corneal reflex, toe pinch, assessment of jaw tone, continuous pulse oximetry, heart rate monitoring, and exhaled CO₂ monitoring, and anesthetic effect in the fetus was assessed by muscle tone. Sheep were given penicillin G procaine (600,000 units intramuscularly) before surgery. A midline incision was made along the linea alba and the uterine horn containing the fetus was exposed. A catheter was placed into a uterine vein such that the tip of the catheter was positioned 2–3 cm proximal to the ovary. A hysterotomy was performed and the fetal hindlimb was exposed. The fetal hindlimb length was measured from the femoral head to the top of the hoof. Polyvinyl catheters were placed in the umbilical vein, bilateral fetal pedal arteries with the tip positioned in the external iliac artery, and bilateral saphenous veins with the tip positioned in the common femoral vein. Ampicillin (500 mg) was injected into the amniotic fluid before surgical closure of the abdomen. Catheters were then placed in the maternal femoral artery and vein. Maternal and fetal catheters were tunneled subcutaneously to the maternal flank. Before skin closure at maternal midline, Marcaine was applied for local analgesic (3 mL, 5 mg·mL⁻¹ Marcaine, 0.5% bupivacaine hydrochloride). Sheep received flunixin meglumine (2.2 mg·kg⁻¹ divided twice per day intramuscularly) on the day of surgery and flunixin meglumine (2.2 mg·kg⁻¹ per day intramuscularly) for the 2 following days. The sheep also received Probios (10 g by mouth twice per day) for 2 days postoperatively.

Infusion and Metabolic Study

Animals were allowed to recover following surgery for a minimum of 5 days before the start of experimental infusions. Four fetuses died after surgery and before the start of the infusion. Remaining fetuses were randomly assigned to receive an intravenous infusion of LR3 IGF-1 (Sigma-Aldrich, St. Louis, MO) at 6.6 µg·kg⁻¹·h⁻¹ based on an estimated fetal weight of 3.5 kg (LR3 IGF-1; n = 12) or 0.5% BSA in saline at 0.2 mL·h⁻¹ (SAL; n = 9) for 7 days. The dose of LR3 IGF-1 was based on previous LR3 IGF-1 infusion studies in fetal sheep showing growth-promoting effects (10, 12). Fetal and maternal blood gas analysis, glucose, and lactate concentrations were measured daily. Fetal plasma was also obtained from the fetus under baseline conditions before infusion start and on the final day of the infusion to measure insulin, endogenous IGF-1, cortisol, norepinephrine, and amino acid concentrations. One additional blood sample was obtained on day 3–4 of the infusion to measure insulin and endogenous IGF-1.

On the final day of the infusion period and with infusions still running, animals underwent a metabolic tracer study. Blood samples were obtained from the maternal femoral artery, uterine vein, fetal external iliac artery, and umbilical vein for baseline measurement of substrates including blood gas, glucose, lactate, and amino acids. The fetus was given a continuous IV infusion of l-[1-¹³C] leucine (Sigma-Aldrich) at a concentration of 6 mg·mL ^− 1; the fetus received a 2-mL bolus followed by 2 mL·h⁻¹ infusion for 180 min to measure fetal protein kinetic rates (24). ³H₂O was given intravenously to the fetus to measure both uterine and umbilical blood flow by the transplacental diffusion method (24). After equilibration, four steady-state blood draws were obtained every 15 min from the umbilical vein, fetal external iliac artery, maternal femoral artery, and uterine vein to measure blood gas, glucose, lactate, amino acid, and tracer concentrations. After the metabolic study was concluded, the pregnant sheep received diazepam (0.2 mg·kg⁻¹) and ketamine (20 mg·kg⁻¹) intravenously and fetuses were delivered via maternal laparotomy and hysterotomy. An overdose of intravenous pentobarbital sodium (Fatal Plus; Vortech Pharmaceuticals, Dearborn, MI) was administered to both the mother and the fetus, after which fetal whole body weight, fetal organ weights, fetal length, and uterine and placental weights were obtained.

One saline-infused and four LR3 IGF-1-infused fetuses died during the infusion periods (between days 4 and 6) and therefore were not included in the analysis. A total of 8 SAL (3 males, 5 females) and 8 LR3 IGF-1 (2 males, 6 females) animals completed the infusion. Due to catheter malfunction, umbilical blood flow measurements were completed in 6 SAL and 7 LR3 IGF-1 animals, and uterine blood flow measurements were completed in 7 SAL and 8 LR3 IGF-1 animals.

Fetal Blood Measurements

Fetal whole blood was analyzed for measurements of pH, Paco₂, Pao₂, oxygen saturation, oxygen content, and hematocrit using a blood gas analyzer (ABL 825, Radiometer, Copenhagen, Denmark), and fetal plasma was used to measure glucose and lactate concentrations (YSI 2900, YSI Inc., Yellow Springs, OH). Plasma insulin, endogenous IGF-1, and cortisol concentrations were measured by ELISA as previously described (25). Plasma amino acid and norepinephrine concentrations were measured using HPLC as previously described (26). Plasma enrichments of leucine and α-ketoisocaproic acid and whole blood enrichments of CO₂ were measured as previously described (27, 28). Fetal blood gas measurements and nutrient concentrations of glucose, lactate, and amino acids from the final day of the infusion are reported as the average of the measurements from the four steady-state blood draws described above.

Calculations

Hindlimb linear growth rate was calculated by subtracting the length of the hindlimb measured at the time of surgery from the length of the hindlimb at necropsy and dividing by the number of days between surgery and necropsy. Uterine and umbilical plasma flow were calculated by multiplying uterine or umbilical blood flow, respectively, by 1 minus hematocrit. Net uterine nutrient uptake rates of oxygen, glucose, lactate, and summed amino acids were calculated by subtracting the mean uterine vein concentration from the mean maternal femoral artery concentration of the four steady-state blood draws and multiplying it by uterine plasma flow for glucose, lactate, and amino acids or by uterine blood flow for oxygen. Net umbilical nutrient uptake rates of oxygen, glucose, lactate, and summed amino acids were calculated by subtracting the mean fetal arterial concentration from the mean umbilical vein concentration of the four steady-state blood draws and multiplying it by umbilical plasma flow for glucose, lactate, and amino acids or by umbilical blood flow for oxygen. Umbilical blood flow and nutrient uptake rates were normalized to fetal weight. Umbilical nutrient/oxygen quotients of glucose, lactate, and summed amino acids were calculated by dividing the whole blood arterial-venous difference of the substrate by the whole blood arterial-venous difference of oxygen content and multiplying by the number of oxygen molecules needed to oxidize one molecule of each nutrient. Net fluxes of leucine tracer across the umbilical circulation were calculated as in Table 1 and as previously described (24, 27, 28). One LR3 IGF-1 animal was found to be an outlier for umbilical amino acid uptake as evaluated using externally studentized residual, Cook’s distance, and covariance ratio calculations (SAS Institute, Cary, NC) and therefore was excluded from the analysis of umbilical amino acid uptake and fetal amino acid concentrations. This animal was also excluded from analysis of the metabolic tracer study as the fetal arterial concentration of infused l-[1-¹³C] leucine was too low for analysis.

Table 1.

Fetal leucine (Leu) flux calculations and protein metabolic rates

Rate	Equation	SAL (n = 6)	LR3 IGF-1 (n = 6)	P Value
Net Leu uptake	LeuV−A × PF	4.56 ± 0.30	3.50 ± 0.30	0.03
Net KIC uptake	KICV−A × PF	0.58 ± 0.15	0.31 ± 0.05	0.07
LeuΦUmb	[1 − 13C]LeuA−V × PF/inf	0.24 ± 0.04	0.16 ± 0.03	0.13
Fetal Leu disposal rate (DR, Flux I)	(100 × inf/Leu MPE) − inf	8.63 ± 0.78	7.94 ± 0.35	0.44
Flux of Leu into placenta from fetal blood (Flux II)	DR × LeuΦUmb	2.04 ± 0.39	1.26 ± 0.21	0.11
Flux of Leu into fetal tissues from fetal blood (Flux III)	DR × (1 − LeuΦUmb)	6.59 ± 0.70	6.69 ± 0.32	0.90
Flux of Leu into fetal blood from placenta (Flux IV)	Net Leu umbilical uptake + Flux II	6.60 ± 0.53	4.75 ± 0.38	0.02
Fetal protein breakdown (Flux V)	DR − Flux IV	2.03 ± 0.55	3.19 ± 0.24	0.08
Fetal Leu oxidation (Flux VI)	[13CO2]A−V × BF/MPE KIC	3.62 ± 0.64	2.34 ± 0.29	0.10
Fetal protein accretion (Flux VII)	Net Leu umbilical uptake + net KIC uptake − Flux VI	1.51 ± 0.34	1.46 ± 0.36	0.92
Fetal protein synthesis (Flux VIII)	Flux VII + Flux V	3.55 ± 0.57	4.65 ± 0.28	0.11

Rates are µmol·min⁻¹·kg⁻¹. V and A represent umbilical vein and fetal artery, respectively. LeuΦUmb is the fraction of Leu taken up by the placenta. Values are means ± SE. P values from Student’s t tests or Mann–Whitney U tests are shown. BF, umbilical blood flow; DR, fetal leu disposal rate; IGF-1, insulin-like growth factor-1; Inf, [1-¹³C] leu infusion rate; KIC, α-ketoisocaproic acid; MPE, molar percent enrichment; PF, umbilical plasma flow; SAL, saline.

Skeletal Muscle Analysis

To measure myoblast proliferation and myofiber hypertrophy, we used muscle biopsy samples from a secondary cohort of animals from a similar study of LR3 IGF-1 infusion into late gestation fetal sheep. Full details of the experimental protocol are detailed elsewhere (10, 29) and included a brief, 1 h maternal exposure to acute hypoxia. This additional exposure was assumed to have a negligible effect on myoblast proliferation and hypertrophy given that the doubling time for fetal myoblasts in cell culture is greater than 24 h (25).

In brief, mixed Western breed sheep were obtained from Agna LLC (Salem, OR). Fetal catheters were placed with a similar protocol as described above with the following exceptions. Maternal and fetal vascular catheters were surgically placed in pregnant sheep (n = 33) at 120 ± 0 dGA. A tracheal catheter was also placed in the mother. Sheep were given intramuscular atropine (7.5 mg), anesthetized with intravenous ketamine (400 mg) and diazepam (10 mg), and ventilated with oxygen:nitrous oxide (2:0.7) and isoflurane (1.5%–2.0%) for the procedure. In the event of a twin pregnancy, the larger fetus (based on body size at the time of surgery) was chosen for the study. Penicillin G (1,000,000 U) and ciprofloxacin (2 mg) were instilled into the amniotic sac before surgical closure, and the sheep received subcutaneous penicillin G (1,000,000 U) following the surgical procedure. Sheep received subcutaneous buprenorphine (0.3 mg) immediately following surgery and then twice daily during surgical recovery for 2 days thereafter. Six fetuses died after surgery but before the start of the infusion. At the time of randomization, three fetuses were noted to have spontaneous hypoxia and were excluded from randomization. After 6 days of recovery, fetuses (n = 24) were randomized to receive either an infusion of LR3 IGF-1 (LR3 IGF-1) or saline (SAL) at the same doses as described above for 7 days. Fetal sex and litter size were equally distributed between the groups. Twenty fetuses completed the infusion; 1 SAL and 3 LR3 IGF-1 fetuses died during the infusion period. Following the infusion, the mother and fetus were exposed to acute hypoxia by infusion of nitrogen gas into the maternal tracheal catheter in order to decrease maternal oxygen saturations by 50% for ∼1 h for experiments previously described (29). Arterial blood gas, hormone, and substrate concentrations for this subset of fetuses were obtained during the infusion and have been previously reported (10).

An overdose of intravenous pentobarbital sodium was then administered to the pregnant sheep after which the fetus was weighed. The tibialis anterior (TA) and flexor digitorum superficialis (FDS) muscles were quickly dissected from the fetal hindlimb and weighed. Biopsies from these muscles were placed on chipboard thinly coated with optimal cutting temperature media, frozen in liquid nitrogen-cooled isopentane for at least 60 s, and stored at −80°C.

Immunohistochemistry

We completed immunohistochemical analysis of muscles from 5 SAL and 7 LR3 IGF-1 fetuses from a subset of animals that had muscle available for analysis [SAL: 2 singletons (both male) and 3 twins (2 female and 1 male); LR3 IGF-1: 4 singletons (3 female and 1 male) and 3 twins (all male)]. TA and FDS muscle sections were removed from chipboard, and 10 µm sections were prepared on SuperFrost Plus microscope slides (Fisher Scientific, Hampton, NH) as previously described (25). To assess myoblast proliferation, sections were incubated with the following antibodies: anti-PAX7 mouse monoclonal IgG (1:100; Developed by Tokyo Institute of Technology and obtained from the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at the University of Iowa, Department of Biology, Iowa City, IA; RRID:AB_2299243), anti-Ki-67 rabbit monoclonal IgG (1:250; Cell Signaling Technology, Danvers, MA; RRID:AB_2687446), and anti-laminin rabbit polyclonal IgG (1:500, Sigma-Aldrich; RRID:AB_477163). Immunocomplexes were detected with polyclonal IgG goat anti-mouse conjugated to Alexa Fluor 594 (1:250; Thermo Fisher, Rockford, IL) and goat anti-rabbit conjugated to Alexa Fluor 488 (1:250; Thermo Fisher) and counterstained with DAPI (1:1,000; Sigma-Aldrich).

To identify myosin heavy chain (MHC) Type I and Type IIa, sections were incubated with anti-laminin rabbit polyclonal IgG (1:100), anti-myosin heavy chain (MHC) Type I (slow) mouse monoclonal IgG (1:1,000; Developmental Studies Hybridoma Bank; RRID:AB_2235587) and anti-MHC Type IIa mouse IgG (1:2; a gift from Dr. Leslie Leinwand, Boulder, CO). Immunocomplexes were detected with AMCA donkey anti-rabbit (1:250; Jackson ImmunoResearch Laboratories, Inc., West Grove, PA), and Cy2 donkey anti-mouse IgG (1:250 for Type I and 1:500 for Type IIa; Jackson ImmunoResearch Laboratories, Inc.).

Images were captured with an Olympus IX-83 microscope (Olympus; Waltham, MA). Morphometric analysis was performed with CellSense software (Olympus). For each muscle type, two sections were analyzed with a minimum of 1,000 fibers per section. For the assessment of myoblast proliferation, a minimum of 2,500 total nuclei and 250 myofibers per fetus were analyzed from a minimum of four representative sections from each muscle. For cross-sectional area and fiber proportion, four sections were analyzed with a minimum of 1,000 fibers per section. Cross-sectional area of Type I and IIa MHC myofibers was quantified.

Statistical Analysis

Student’s t test was used to compare SAL and LR3 IGF-1 groups (Graph Pad Prism 8, San Diego, CA). A Mann–Whitney U test was used when variances were unequal. Two-way ANOVA was used to determine the effect of LR3 IGF-1 (LR3 IGF-1, SAL) and muscle type (TA, FDS) on myoblast proliferation and myofiber hypertrophy; post hoc tests were performed using Bonferroni multiple comparison test. Values are reported as means ± SE. A P value of ≤0.05 was considered statistically significant.

RESULTS

Fetal and Placental Weights

Fetal whole body, organ, muscle, and placental weights are shown in Table 2. Fetal weight was not statistically different between SAL and LR3 IGF-1. Heart, adrenal gland, and spleen weights were 34%, 31%, and 100% larger, respectively, in LR3 IGF-1 compared to SAL (P < 0.05). Brain weight, individual muscle weights, and linear growth rate of the hindlimb were not different between groups. When individual organ weights were normalized to fetal weight, heart and spleen weights remained significantly higher in LR3 IGF-1; normalized brain weight was significantly lower in LR3 IGF-1 (Table 2). Uterine and uteroplacental weights were similar between groups.

Table 2.

Maternal and fetal characteristics

	SAL (n = 8)	LR3 IGF-1 (n = 8)	P Value
Maternal characteristics
Weight at surgery, kg	56 ± 4	57 ± 2	0.70
Fetal characteristics
Gestational age at end of  study, day	134 ± 1	134 ± 0	0.85
% Male	38%	25%
Fetal weight, g	3260 ± 211	3682 ± 183	0.15
Crown rump length, cm	48.2 ± 1.2	50.6 ± 0.8	0.14
Hindlimb length, cm	37.9 ± 1.1	38.9 ± 0.8	0.48
Hindlimb linear growth rate, cm·day−1	0.65 ± 0.08	0.68 ± 0.04	0.69
Brain, g	49.2 ± 1.4	47.0 ± 1.3	0.27
Per fetal weight, g·kg−1	15.6 ± 1.2	12.9 ± 0.5	0.04
Heart, g	25.2 ± 2.0	33.8 ± 2.0	0.009
Per fetal weight, g·kg−1	7.7 ± 0.3	9.2 ± 0.3	0.002
Lungs, g	125.1 ± 10.6	142.0 ± 11.2	0.29
Per fetal weight, g·kg−1	38.4 ± 2.2	38.5 ± 2.0	0.98
Liver, g	95.0 ± 10.3	116.8 ± 7.8	0.11
Per fetal weight, g·kg−1	28.7 ± 1.9	31.7 ± 1.6	0.24
Kidneys, g	20.8 ± 1.3	25.4 ± 2.1	0.08
Per fetal weight, g·kg−1	6.4 ± 0.3	6.9 ± 0.4	0.41
Adrenal glands, g	0.48 ± 0.03	0.63 ± 0.03	0.003
Per fetal weight, g·kg−1	0.16 ± 0.02	0.17 ± 0.01	0.44
Spleen, g	6.5 ± 0.5	13.0 ± 2.5	0.007
Per fetal weight, g·kg−1	2.0 ± 0.2	3.6 ± 0.8	0.01
Biceps femoris, g	15.3 ± 1.2	16.6 ± 1.1	0.47
Per fetal weight, g·kg−1	4.7 ± 0.2	4.5 ± 0.3	0.62
Gastrocnemius, g	8.0 ± 0.6	8.5 ± 0.5	0.53
Per fetal weight, g·kg−1	2.4 ± 0.1	2.3 ± 0.1	0.13
Soleus, g	0.30 ± 0.07	0.25 ± 0.05	0.64
Per fetal weight, g·kg−1	0.09 ± 0.02	0.07 ± 0.01	0.47
Tibialis anterior, g	3.3 ± 0.3	3.4 ± 0.2	0.80
Per fetal weight, g·kg−1	1.00 ± 0.05	0.92 ± 0.04	0.22
Flexor digitorum superficialis, g	2.5 ± 0.2	2.7 ± 0.2	0.40
Per fetal weight, g·kg−1	0.76 ± 0.03	0.73 ± 0.03	0.51
Uterus, g	658 ± 56	683 ± 56	0.76
Uteroplacental unit, g	2,389 ± 352	2,168 ± 240	0.61

Values are means ± SE. P values from Student’s t tests or Mann–Whitney U tests are shown.

Fetal Blood Gas, Substrate, and Hormone Measurements

Fetal arterial blood gas, substrate, and hormone concentrations at the end of the experimental infusion period are shown in Table 3. Endogenous IGF-1 concentrations were 45% lower (P < 0.05), and insulin concentrations were 58% lower in LR3 IGF-1 compared to SAL (P < 0.05). Plasma glucose concentrations were similar between groups. Blood gas parameters, including pH, Paco₂, Pao₂, and hematocrit, were similar between groups. Oxygen content was 21% lower in LR3 IGF-1 (P < 0.05), and oxygen saturations were 14% lower in LR3 IGF-1 (P < 0.05). Cortisol, norepinephrine, and lactate concentrations were not different between groups. There was a trend toward lower summed total fetal amino acid concentrations in LR3 IGF-1 (P = 0.065). When the amino acid concentrations were considered individually, glutamate and isoleucine were lower in LR3 IGF-1 (P < 0.05), and alanine, leucine, serine, taurine, and valine tended to be lower (P < 0.1).

Table 3.

Fetal blood gas, substrate, and hormone measurements at end of infusion period

	SAL (n = 7)	LR3 IGF-1 (n = 7)	P Value
Blood gas measurements
pH	7.35 ± 0.01	7.33 ± 0.01†	0.12
PaCO2, mmHg	53.7 ± 1.2	54.2 ± 1.1†	0.76
PaO2, mmHg	19.9 ± 0.4	20.5 ± 0.8†	0.49
SaO2, %	43.5 ± 2.2	37.3 ± 1.8†	0.045
O2 content, mmol·L−1	2.8 ± 0.2	2.2 ± 0.1†	0.008
Hematocrit, %	33.3 ± 1.8	30.3 ± 0.8†	0.14
Plasma substrates
Glucose, mg·dL−1	20.2 ± 1.2	17.9 ± 1.6†	0.28
Lactate, mmol·L−1	2.76 ± 0.16	2.74 ± 0.28†	0.94
Total amino acids, µmol·L−1	3582 ± 175	2844 ± 318	0.07
Alanine, µmol·L−1	313.9 ± 12.8	252.4 ± 52.5	0.07
Arginine, µmol·L−1	98.3 ± 11.9	77.6 ± 9.7	0.20
Asparagine, µmol·L−1	39.2 ± 3.5	35.0 ± 7.3	0.61
Aspartate, µmol·L−1	22.7 ± 1.1	20.5 ± 2.2	0.38
Cysteine, µmol·L−1	15.6 ± 1.2	13.0 ± 1.0	0.12
Glutamate, µmol·L−1	45.0 ± 2.1	26.5 ± 3.3	0.0005
Glutamine, µmol·L−1	400.6 ± 21.8	336.0 ± 45.2	0.22
Glycine, µmol·L−1	310.2 ± 49.4	254.9 ± 49.0	0.44
Histidine, µmol·L−1	40.3 ± 2.4	35.6 ± 4.8	0.40
Isoleucine, µmol·L−1	105.3 ± 6.3	77.9 ± 7.4	0.02
Leucine, µmol·L−1	156.5 ± 13.1	121.0 ± 12.2	0.07
Lysine, µmol·L−1	61.4 ± 6.6	44.9 ± 7.0	0.11
Methionine, µmol·L−1	82.0 ± 8.2	67.0 ± 5.4	0.15
Ornithine, µmol·L−1	40.1 ± 2.8	35.7 ± 5.4	0.48
Phenylalanine, µmol·L−1	109.6 ± 7.9	100.0 ± 7.9	0.41
Proline, µmol·L−1	114.8 ± 10.6	99.8 ± 17.6	0.48
Serine, µmol·L−1	724.0 ± 53.6	571.2 ± 61.7	0.09
Taurine, µmol·L−1	99.5 ± 24.3	43.6 ± 10.4	0.06
Threonine, µmol·L−1	252.2 ± 33.0	171.1 ± 34.2	0.11
Tryptophan, µmol·L−1	40.3 ± 2.7	34.4 ± 4.0	0.25
Tyrosine, µmol·L−1	112.2 ± 9.7	103.7 ± 9.7	0.55
Valine, µmol·L−1	398.2 ± 24.7	322.8 ± 29.2	0.07
Plasma hormones
Insulin, ng·mL−1	0.31 ± 0.05†	0.13 ± 0.03†	0.008
IGF-1, ng·mL−1	113.6 ± 11.0†	62.0 ± 8.5†	0.002
Cortisol, ng·mL−1	30.7 ± 8.7†	34.1 ± 9.1†	0.79
Norepinephrine, pg·mL−1	682 ± 102†	1093 ± 339†	0.38

Values are means ± SE. P values from Student’s t tests or Mann–Whitney U tests are shown. †Indicates n = 8. IGF-1, insulin-like growth factor-1; SAL, saline.

Uterine and Umbilical Blood Flow, Nutrient Uptake Rates, and Nutrient/Oxygen Quotients

Uterine and umbilical blood flow rates were similar between groups (Fig. 1). Uterine oxygen, glucose, and total amino acid uptake rates were similar between SAL and LR3 IGF-1, and uterine lactate output rate was similar between groups (Fig. 2). In respect to each individual amino acid, phenylalanine and arginine uterine uptake rates were lower in LR3 IGF-1 (P < 0.05), and lysine uterine uptake rate was higher in LR3 IGF-1 (P < 0.05; Fig. 3). Umbilical oxygen, glucose, and lactate uptake rates were similar between groups (Fig. 4). The umbilical uptake rate of the total amino acids was lower in LR3 IGF-1 (P < 0.05). In respect to each individual amino acid, the umbilical uptake rates of asparagine, glutamine, isoleucine, leucine, phenylalanine, threonine, tryptophan, and tyrosine were lower in LR3 IGF-1 (P < 0.05), as was the umbilical output of glutamate (Fig. 5). The glucose/oxygen and lactate/oxygen quotients were not different between groups; however, the total amino acid/oxygen quotient and total nutrient/oxygen quotient were lower in LR3 IGF-1 (P < 0.05; Fig. 6).

Figure 1.

Uterine and umbilical blood flow. Uterine blood flow in SAL (white bar, n = 7) and LR3 IGF-1 (gray bar, n = 8) (A) and umbilical blood flow in SAL (white bar, n = 6) and LR3 IGF-1 (gray bar, n = 7) (B). Umbilical blood flow is normalized to fetal weight. Individual animal data and means ± SE are shown. Open circles (○) indicate a male fetus, and closed circles (●) indicate a female fetus. Student’s t test was performed. IGF-1, insulin-like growth factor-1; SAL, saline.

Figure 2.

Uterine nutrient uptake rates. Uterine uptake rates of oxygen (A), glucose (B), lactate (C), and total amino acids (D) in SAL (white bar, n = 7) and LR3 IGF-1 (gray bar, n = 8). Individual animal data and means ± SE are shown. Open circles (○) indicate a male fetus, and closed circles (●) indicate a female fetus. Student’s t test or Mann–Whitney U test was performed. IGF-1, insulin-like growth factor-1; SAL, saline.

Figure 3.

Uterine uptake rates of individual amino acids. Uterine uptake rates of essential (A) and nonessential (B) amino acids in SAL (white bars, n = 7) and LR3 IGF-1 (gray bars, n = 8). Means ± SE are shown. *Indicates P < 0.05 by Student’s t test. IGF-1, insulin-like growth factor-1; SAL, saline.

Figure 4.

Umbilical nutrient uptake rates. Umbilical uptake rates of oxygen (A), glucose (B), lactate (C) in SAL (white bar, n = 6) and LR3 IGF-1 (gray bar, n = 7), and total amino acids (D) in SAL (white bar, n = 6) and LR3 IGF-1 (gray bar, n = 6) normalized to fetal weight. Individual animal data and means ± SE are shown. Open circles (○) indicate a male fetus, and closed circles (●) indicate a female fetus. *Indicates P < 0.05 by Student’s t test. IGF-1, insulin-like growth factor-1; SAL, saline.

Figure 5.

Umbilical uptake rates of individual amino acids. Umbilical uptake rates of essential (A) and nonessential (B) amino acids in SAL (white bars, n = 6) and LR3 IGF-1 (gray bars, n = 6) normalized to fetal weight. Means ± SE are shown. *Indicates P < 0.05 by Student’s t test or Mann–Whitney U test. IGF-1, insulin-like growth factor-1; SAL, saline.

Figure 6.

Nutrient/oxygen quotients. Nutrient/oxygen quotients for glucose and lactate are shown for SAL (white bar, n = 6) and LR3 IGF-1 (gray bar, n = 7) and summed amino acids and total of all nutrients for SAL (white bar, n = 6) and LR3 IGF-1 (gray bar, n = 6). Individual animal data and means ± SEM are shown. Open circles (○) indicate a male fetus, and closed circles (●) indicate a female fetus. *Indicates P < 0.05 by Student’s t test or Mann–Whitney U test. IGF-1, insulin-like growth factor-1; SAL, saline.

Fetal Protein Metabolism

Fetal protein metabolic rates are shown in Table 1. Net leucine uptake by the fetus was 23% lower in LR3 IGF-1 (P < 0.05), and flux of leucine into the fetal blood from the placenta was 28% lower in LR3 IGF-1 (P < 0.05). Fetal protein accretion, synthesis, breakdown, and oxidation rates were similar between SAL and LR3 IGF-1.

Myoblast Proliferation and Myofiber Cross-Sectional Area

The fetal and biochemical characteristics of this secondary cohort of animals from which these samples were collected are previously published and similar to the biochemical characteristics of the animals described above including lower insulin and oxygen content in LR3 IGF-1. LR3 IGF-1 and SAL had similar lactate, cortisol, and norepinephrine concentrations at the end of the LR3 IGF-1 infusion (10). Fetal sex and litter size were not different between groups—57% of LR3 IGF-1 and 60% of SAL animals were male, and 43% of LR3 IGF-1 and 60% of SAL animals were a result of a twin pregnancy. LR3 IGF-1 fetuses in this cohort were 9% heavier than SAL, but this did not reach statistical significance (P = 0.34). Individual muscle weights for the FDS and TA were similar between groups. The ratio of PAX7⁺ nuclei (a marker of myoblasts) to total nuclei (Fig. 7A) and the number of PAX7⁺ nuclei per myofiber (Fig. 7B) were higher in LR3 IGF-1 compared to SAL (P < 0.05). The ratio of Ki-67⁺ nuclei (a marker of active cellular proliferation) to total nuclei was higher in LR3 IGF-1 compared to SAL (P < 0.05; Fig. 7C). Similarly, the ratio of PAX7⁺ myoblasts that also expressed Ki-67⁺ was higher in LR3 IGF-1 compared to SAL (P < 0.05; Fig. 7D). Cross-sectional areas of Type I and Type IIa myofibers were similar between groups, as were the relative proportions of the fiber types (Fig. 8).

Figure 7.

Myoblast proliferation in tibialis anterior (TA) and flexor digitorum superficialis (FDS) muscles. The ratio of PAX7⁺ nuclei to total nuclei (A), PAX7⁺ nuclei per myofiber (B), Ki-67⁺ nuclei to total nuclei (C), and PAX7⁺ myoblasts that were also Ki-67⁺ to PAX7⁺ nuclei (D) were quantified in TA and FDS muscles for SAL (white bar, n = 5) and LR3 IGF-1 (gray bar, n = 7). Individual animal data and means ± SE are shown. Open circles (○) indicate a male fetus, and closed circles (●) indicate a female fetus. Significant effects from two-way ANOVA (LR3 IGF-1, muscle type) are indicated; Bonferroni post hoc test was performed. IGF-1, insulin-like growth factor-1; SAL, saline.

Figure 8.

Myofiber cross-sectional area (CSA) and proportion in tibialis anterior (TA) and flexor digitorum superficialis (FDS) muscles. The CSA of myofibers expressing myosin heavy chain (MHC) Type I (A) and Type IIa (B), and the relative proportion of MHC Type I (C) and Type IIa (D) to total myofibers in TA and FDS muscles for SAL (white bar, n = 5) and LR3 IGF-1 (gray bar, n = 7). Individual animal data and means ± SE are shown. Open circles (○) indicate a male fetus, and closed circles (●) indicate a female fetus. Significant effects from two-way ANOVA (LR3 IGF-1, muscle type) are indicated; Bonferroni post hoc test was performed. IGF-1, insulin-like growth factor-1; SAL, saline.

DISCUSSION

In this study, we aimed to understand the role of LR3 IGF-1, an IGF-1 analog, in regulating umbilical uptake of nutrients and fetal nutrient availability to support fetal growth and protein accretion. We demonstrated that LR3 IGF-1 infusion did not affect uterine or umbilical blood flow, and it did not increase fetal uptake rates of glucose, lactate, or oxygen from the uterine or umbilical circulation. In fact, the umbilical uptake rate of amino acids was significantly lower in LR3 IGF-1 animals, which was consistent with a lower net leucine uptake rate and lower flux of leucine from the placenta into fetal blood measured using a leucine tracer. Fetal arterial concentrations of several amino acids were lower in LR3 IGF-1 animals. Oxygen content and saturations were lower in LR3 IGF-1 animals, as was insulin. Protein accretion, synthesis, breakdown, and oxidation rates were similar in LR3 IGF-1 and SAL animals. However, we did not observe a significant increase in fetal body weight; LR3 IGF-1 fetuses were 13% heavier than SAL, but this was not statistically higher (P = 0.15). Despite similar body weights, LR3 IGF-1 fetuses did have higher heart, adrenal gland, and spleen weights. Skeletal muscle from LR3 IGF-1 animals had higher myoblast proliferation but not myofiber hypertrophy. In summary, a 1-wk infusion of LR3 IGF-1 into late gestation fetal sheep promoted organ-specific growth, but the primary physiological mechanism was not by stimulation of nutrient uptake rates by the uteroplacental unit or the fetus.

We hypothesized that LR3 IGF-1 infusion would increase placental transport of glucose and amino acids to the fetus therefore increasing nutrient availability. This was based on prior literature showing that IGF-1 increased placental mRNA expression of amino acid transporters (14) and increased uptake of glucose and an amino acid analog in placental trophoblast cells (18, 19), and studies showing that an infusion of IGF-1 for several hours increased fetal and placental uptake of glucose and amino acids (21). Contrary to our hypothesis, umbilical uptake rates of glucose and lactate to the fetus were similar to the SAL group after 1 wk of LR3 IGF-1 infusion, and the umbilical uptake rate of amino acids was actually lower in LR3 IGF-1. Lower umbilical uptake of amino acids, net leucine uptake, leucine flux to the fetal blood, and concentrations of several fetal arterial amino acid concentrations indicate that there is decreased placental transport of amino acids to the fetus after a week-long infusion of LR3 IGF-1. Further studies are needed to examine the mechanisms responsible for this difference. Amino acid transport across the placenta to the fetus is an energy-dependent process. Insulin, an important regulator of fetal growth (30) and muscle protein synthesis (31), has been shown to stimulate transport of an amino acid analog in human placental trophoblasts (32) and increase activity of system A amino acid transporters in placental fragments in vitro (33). In our study, LR3 IGF-1-treated fetuses had significantly lower insulin concentrations compared to SAL, which is consistent with previous studies of fetal IGF-1 infusion (10, 11, 22, 23). The mechanism responsible for lower insulin is currently under investigation, but it may be a direct effect of IGF-1 on the fetal beta cell or an indirect effect of hypoxia or low fetal arterial amino acid concentrations. We have previously shown that both fetal oxygen and amino acid concentrations impact fetal insulin concentrations (34, 35). Regardless, we speculate that lower fetal insulin concentrations may have contributed to lower umbilical transport of amino acids to the fetus, but this requires further investigation.

We evaluated fetal protein kinetic rates using a leucine tracer and determined that whole fetal protein accretion, synthesis, breakdown, and oxidation rates were similar in SAL and LR3 IGF-1 despite lower arterial concentration of amino acids. Previous studies demonstrated that an infusion of recombinant IGF-1 for 3 h into fetal sheep decreased protein breakdown and protein oxidation, ultimately increasing fetal protein accretion (22). Similarly, an infusion of recombinant IGF-1 and insulin for 7 h increased protein synthesis and accretion in fetal sheep when fetal blood amino acid concentrations were held constant (36). However, our study demonstrates that with a 1-wk LR3 IGF-1 infusion, particularly in the setting of lower fetal blood concentrations of amino acids, insulin, and oxygen content, protein metabolism was similar in SAL and LR3 IGF-1. We also determined that the amino acid/oxygen quotient was lower in LR3 IGF-1, and despite normal glucose/oxygen and lactate/oxygen quotients, the total nutrient/oxygen quotient was lower in LR3 IGF-1, suggesting lower oxidative metabolism of amino acids. Fetal oxidation of leucine, measured via our tracer method, was also lower in LR3 IGF-1, although this did not reach statistical significance (P = 0.1). In late gestation fetal sheep, amino acids, insulin, and oxygen are important drivers for protein anabolism; amino acid infusion to the fetus increased protein synthesis and accretion (37), and insulin infusion increased protein accretion and amino acid utilization when amino acid concentrations were maintained (38, 39). In neonatal pigs, infusion of either insulin or amino acids for several hours stimulated skeletal muscle protein synthesis (31). In the growth-restricted fetus, which is known to have reduced muscle mass, amino acid supplementation increased protein accretion, even in the setting of hypoinsulinemia (26). The hypoxic sheep fetus has also been shown to have decreased protein synthesis, breakdown, and protein accretion (40). In this study, the LR3 IGF-1-treated fetus had decreased arterial oxygen content, which is consistent with previous literature (10, 12, 22, 23, 41), despite normal umbilical uptake rates of oxygen; the mechanism responsible is unknown. We did not find significant differences in hemoglobin, cortisol, norepinephrine, or lactate in LR3 IGF-1-treated animals to explain decreased oxygen content. Despite lower fetal insulin, amino acid concentrations, and oxygen content, fetal protein accretion in LR3 IGF-1 was maintained similar to SAL. Euaminoacidemia, normoinsulinemia, and/or normal oxygen content may be required to optimally stimulate fetal protein accretion during LR3 IGF-1 infusion. Further research is needed to study the effects of LR3 IGF-1 infusion on fetal protein accretion when amino acid, insulin, and/or oxygen concentrations are maintained.

Although we demonstrated lower amino acid availability to the fetus, similar fetal protein metabolism, and lower fetal insulin concentrations and oxygen content, select fetal organs in LR3 IGF-1 animals, including heart, adrenal glands, and spleen, were larger than SAL, and LR3 IGF-1 had higher myoblast proliferation. Organ-specific growth after IGF-1 infusion is supported in previous literature in sheep. Fetuses that received a 10-day infusion of recombinant IGF-1 had larger organs including heart, lungs, liver, adrenal glands, spleen, and kidneys compared to controls, although whole fetal weights were not statistically significant between groups (11). IGF-1 is known to be an important regulator of cardiac growth; in animal models, overexpression or infusion of recombinant IGF-1 promotes cardiac cell hyperplasia but not hypertrophy of cells (12, 42). In contrast, infusion of recombinant IGF-1 to fetal sheep resulted in hypertrophy of cells in the adrenal gland, without an increase in plasma cortisol concentrations or mRNA expression of enzymes necessary for the synthesis of steroid hormones or catecholamines (43). The authors of that study speculate that IGF-1 may initiate growth pathways such as MAP kinase (MAPK) or phosphoinositide 3-kinase (PI3K), leading to cellular hypertrophy without changes in hormone production. The mechanism causing spleen growth is also unknown. However, IGF-1 is known to regulate hematopoiesis (44), an important function of the fetal spleen, which may impact splenic growth.

Similar to cardiac muscle, our study showed higher myoblast proliferation but not hypertrophy. This is corroborated by knockout studies in which IGF-1 receptor gene knockout fetal mice have lower myocyte number, but the muscle cells were similar in size and morphology as wild type (9). Binding of IGF-1 to IGF-1 receptor activates one of two signaling pathways—MAPK or PI3K (45). In muscle, MAPK is important in myoblast proliferation, whereas PI3K is important for myoblast fusion and resistance to apoptosis (15). PI3K is also important for glucose uptake and protein synthesis in the myotube. Further research to evaluate these signaling pathways in organs and skeletal muscle from IGF-1-treated animals is ongoing.

In this study, LR3 IGF-1 fetuses were not significantly larger than SAL. LR3 IGF-1 fetuses were 13% heavier than SAL fetuses; however, this did not reach statistical significance. This is in contrast to a recent study using a similar protocol, in which LR3 IGF-1 animals were 15% heavier than controls (P < 0.05) (10). However, despite the correlation between birth weight and umbilical cord concentrations of IGF-1, higher fetal weight gain following infusion of IGF-1 is not a consistent finding in animals that are not growth restricted. Thus, there may be limitations to fetal growth in the previously normally growing animal, or limitations in substrate availability may preclude excessive growth. Additionally, we measured umbilical nutrient uptake rates, fetal protein kinetics, and fetal weight once after 1 wk of LR3 IGF-1 infusion; we were unable to measure these repeatedly during the infusion. Therefore, it is unknown if the trajectory of fetal weight gain and organ growth initially increased but then slowed over time. Similarly, it is unknown if umbilical uptake of amino acids was consistently lower throughout the infusion or if it also changed over time.

Although we measured endogenous IGF-1 plasma concentrations in the fetal sheep, we were unable to measure circulating concentrations of LR3 IGF-1. Endogenous IGF-1 was lower in LR3 IGF-1 animals, likely due to suppression of endogenous IGF-1 production during the infusion of exogenous LR3 IGF-1. However, the dose we used was replicated in a previous study and demonstrated measurable LR3 IGF-1 blood concentrations in the fetus and lower concentrations of endogenous IGF-1 (12). We were also unable to analyze for sex differences or the effect of twin pregnancy (for skeletal muscle histological analyses) due to the small sample size; however, males, females, and twins were equally distributed between groups. There is evidence that intrauterine IGF-1 treatment may alter body composition and glucose metabolism in a sex-specific manner (46). Larger studies will be needed in the future to determine if there are sex-specific effects of IGF-1 on fetal protein metabolism.

In conclusion, we determined that a 1-wk infusion of LR3 IGF-1 into late gestation fetal sheep resulted in lower umbilical uptake rates of amino acids, lower fetal arterial amino acid and insulin concentrations, and lower fetal oxygen content; however, LR3 IGF-1 promoted the growth of several organs and proliferation of myoblasts, while fetal protein metabolism was maintained. Thus, LR3 IGF-1-treated fetuses were still able to effectively utilize the available nutrients and oxygen to support growth in an organ-specific manner. Further studies are needed to examine the specific metabolic pathways responsible for decreased placental transport of amino acids, increased organ growth, and increased myoblast proliferation by LR3 IGF-1 and to determine the importance of maintaining insulin, amino acid, and oxygen concentrations during LR3 IGF-1 infusion in order to maximally support growth and protein accretion.

GRANTS

This work was supported by the National Institute of Health Grants R01DK088139 (to P.J.R.), R01HD093701 (to P.J.R.), T32HD007186 (to J.S. and A.W. trainees, P.J.R., P.I.), R01DK108910 (to S.R.W.), R01HD071068 (to S.S.J.), R01HL142483 (to S.S.J.), R01HD079404 (to L.D.B.), and S10OD023553 (to L.D.B.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.S., A.W., S.R.W., S.S.J., P.J.R., and L.D.B. conceived and designed research; J.S., S.H., A.W., E.I.C., S.C.S., S.R.W., S.S.J., P.J.R., and L.D.B. performed experiments; J.S., S.H., A.W., E.I.C., S.R.W., S.S.J., P.J.R., and L.D.B. analyzed data; J.S., S.H., A.W., E.I.C., S.R.W., S.S.J., P.J.R., and L.D.B. interpreted results of experiments; J.S., S.H., and L.D.B. prepared figures; J.S., S.H., and L.D.B. drafted manuscript; J.S., S.H., A.W., E.I.C., S.C.S., S.R.W., S.S.J., P.J.R., and L.D.B. edited and revised manuscript; J.S., S.H., A.W., E.I.C., S.C.S., S.R.W., S.S.J., P.J.R., and L.D.B. approved final version of manuscript.