Molecular inflammation and adipose tissue matrix remodeling precede physiological adaptations to pregnancy

Abstract

Changes in adipose tissue metabolism are central to adaptation of whole body energy homeostasis to pregnancy. To gain insight into the molecular mechanisms supporting tissue remodeling, we have characterized the longitudinal changes of the adipose transcriptome in human pregnancy. Healthy nonobese women recruited pregravid were followed in early (8–12 wk) and in late (36–38 wk) pregnancy. Adipose tissue biopsies were obtained in the fasting state from the gluteal depot. The adipose transcriptome was examined via whole genome DNA microarray. Expression of immune-related genes and extracellular matrix components was measured using real-time RT-PCR. Adipose mass, adipocyte size, and cell number increased in late pregnancy compared with pregravid measurements (P < 0.001) but remained unchanged in early pregnancy. The adipose transcriptome evolved during pregnancy with 10–15% of genes being differently expressed compared with pregravid. Functional gene cluster analysis revealed that the early molecular changes affected immune responses, angiogenesis, matrix remodeling, and lipid biosynthesis. Increased expression of macrophage markers (CD68, CD14, and the mannose-6 phosphate receptor) emphasized the recruitment of the immune network in both early and late pregnancy. The TLR4/NF-κB signaling pathway was enhanced specifically in relation to inflammatory adipokines and chemokines genes. We conclude that early recruitment of metabolic and immune molecular networks precedes the appearance of pregnancy-related physiological changes in adipose tissue. This biphasic pattern suggests that physiological inflammation is an early step preceding the development of insulin resistance, which peaks in late pregnancy.

Maternal body composition reflects homeostatic changes.

extensive modifications of maternal body composition take place during pregnancy. Total adipose mass and maternal blood volume increase significantly, whereas there is little change in maternal lean tissue mass (28, 32). The dynamic role of adipose tissue in the metabolic adaptations to pregnancy is highlighted by longitudinal changes in adipose tissue mass. There is a wide individual variation in fat mass accumulation that can range from −2 to >10 kg/woman depending on race, ethnicity, and nutritional and metabolic factors (30, 40). In lean women, the total gain in fat mass represents about 30% of gestational weight gain (22), but women with pre-ravid obesity usually gain less total weight than lean women (18). In uncomplicated pregnancies with appropriate weight gain, most fat deposition occurs during the second trimester and remains constant until term (22, 54). The mechanisms underlying the early adjustments for increased fat deposition have yet to be defined in either normal or metabolically compromised pregnancies.

Storage and endocrine function.

The changes in body composition during pregnancy are driven primarily by adaptations of maternal metabolic homeostasis. The ultimate goal of pregnancy-induced metabolic changes is to meet the high-energy demands of fetal development. Glucose, the primary energy fuel used by fetal tissues, needs to be readily available for transplacental transfer, whereas maternal tissues can rely on other energy substrates, such as lipids (3). The adaptations of lipid metabolism follow a well-described biphasic pattern. The first half of pregnancy is centered on storing maternal energy as adipose tissue triglycerides, whereas in late pregnancy the stored lipids are mobilized to be used by peripheral tissues and in preparation for lactation (25). These sequential adaptations are facilitated by modifications of insulin secretion and action. In early pregnancy, the higher insulin sensitivity facilitates cellular anabolism through activation of lipogenesis. Then, insulin resistance develops progressively to culminate in the third trimester (7, 43). Insulin resistance allows adipose tissue to mobilize the lipids stored earlier and skeletal muscle to utilize less glucose. These changes in maternal metabolic homeostasis result in increased circulating levels of insulin and triglycerides in late pregnancy (23). In addition to alterations in its capacity for energy storage, the endocrine function of adipose tissue also evolves during pregnancy. The synthesis and plasma concentration of leptin and adiponectin, two major adipokines, exhibit longitudinal changes parallel to those of insulin sensitivity. Whereas leptin concentrations increase early, there is a significant decrease in plasma adiponectin during the third trimester (9, 26). The longitudinal modifications of adipokines in healthy pregnancy are further enhanced in the context of pregnancy with diabetes and obesity (10, 24).

Remodeling.

White adipose tissue displays a remarkable flexibility, with an important and reversible capacity for expansion throughout adult life. Remodeling of the adipose tissue mass requires the integration of cellular mechanisms to support an increase in size and/or number of adipocytes. The turnover of fat cells is highly sensitive to variations in the metabolic and hormonal milieu, i.e., development, aging, diet, or metabolic disorders occurring throughout the lifespan (16, 33, 52). In obese individuals, the enlargement of adipose tissue requires active turnover of adipocytes with chronic inflammation elicited by accumulation of macrophages in the stromal vascular compartment (21, 60). Previously, we have shown that macrophage infiltration and increased adipocyte death contribute to inflammation in pregnant women with pregravid obesity (2, 12).

The aim of this study was to characterize the mechanisms responsible for the increase in adipose tissue mass during normal human pregnancy. We have conducted a longitudinal analysis beginning before and continuing throughout pregnancy to characterize the morphological and molecular traits of the adipose tissue in nonobese women. We show that molecular changes supporting remodeling of gluteal adipose tissue precede phenotypic changes in fat accretion and lipid metabolism that culminate in late pregnancy.

METHODS

Subjects.

Eleven women with singleton pregnancies were recruited prepregnancy (P) and followed up longitudinally in both early (E; 8–12 wk) and late pregnancy (L; 36–38 wk). The protocol was approved by the Institutional Review Board and Clinical Research Unit Scientific Review Committee at Case Western Reserve University. Volunteers gave their written informed consent, in accordance with the MetroHealth Medical Center (Cleveland, OH) Guidelines for the Protection of Human Subjects.

Metabolic and anthropometric measurements.

Anthropometric measurements and body composition were estimated using hydrodensitometry, as described previously (8, 27). Insulin sensitivity was measured with the euglycemic hyperinsulinemic clamp technique (7). Plasma was separated by centrifugation and kept frozen at −20°C until being assayed. Glucose was assessed by the glucose oxidase method (YSI, Yellow Springs, OH). Plasma insulin was measured using the sensitive human insulin radioimmunoassay kit (Linco Research, St. Charles, MO). The limit of sensitivity for the insulin assay is 0.2 μU/ml, with intra-assay coefficients of variation (CV) of 4.4–6.8%. Leptin and adiponectin were measured using ELISA kits (R & D Systems, Minneapolis, MN) with CVs of 3.0–6.2 and 6.2–8.4%, respectively. Interleukin-6 and interleukin-8 were assayed by ELISA (QuantiGlo; R & D Systems) with CVs of 5.3–7.8 and 2.6–3.4%, respectively. All plasma samples were run in duplicate.

Adipose tissue analysis: measurements of cell number and size.

Biopsies of adipose tissue (800–1,200 mg) from 11 women were obtained via liposuction of the subcutaneous gluteal depot at each time point (P, E, and L). One fragment of tissue was fixed immediately for immunohistochemistry, one fragment was snap-frozen in liquid nitrogen for RNA processing, and the remaining tissue was processed for collagenase digestion. After digestion with 1 mg/ml collagenase (Worthington Biochemical, Lakewood, NJ) for 45 min at 37°C, the mixture was filtered through a 100-μm gauze mesh. The infranatant was removed, and the floating layer of adipocytes was brought up to 10 ml for counting and size measurements. Adipocyte diameter was determined using a ×10 W. F. objective lens containing a ruler (Olympus BH-2 microscope; Olympus, Hiroshima, Japan). Numbers from 10 fields were averaged, and cellularity was defined as the number of adipocytes per gram of digested adipose tissue (44).

Immunohistochemistry.

Adipose tissue sections were fixed in 10% formalin and included in paraffin. Five-micrometer sections were counterstained using eosin and hematoxylin gill II (Sigma-Aldrich). Digitized images were obtained using a Nikon E600 microscope equipped with a DXM200 camera.

Protein expression.

Lysates of whole adipose tissue were prepared using 10 mM Tris-NaCl, pH 8, and lysis buffer with 1% Triton and protease inhibitors (Sigma-Aldrich). Protein lysates (2–200 μg/ml) were separated by SDS-PAGE (7% polyacrylamide gels) and immunoblotted or directly blotted to nitrocellulose for microfiltration (Hoefer). HL-60 whole cell lysate (Santa Cruz, CA) was used as positive control. Primary antibodies against GAPDH (1:3,000), glucose transporter 5 (1:500), lipoprotein lipase (LPL; 1:500), and adiponectin (1:500) were from Abcam. Actin (1:2,000), CCAAT/enhancer-binding protein (C/EBP) (1:500), and Toll-like receptor 4 (TLR4) (1:300) were from Santa Cruz Biotechnology; lipolysaccharide-binding protein (1:500) was from Origene Technologies. Detection was performed using anti-mouse or anti-rabbit HRP-coupled secondary antibodies (1:5,000) and ECL Plus Western blot chemiluminescence reagents (Amersham, GE Heathcare Sciences). Signal intensity was quantitated by gel doc imager (Bio-Rad).

Gene expression analysis.

Total RNA was isolated from intact adipose tissue using the Trizol (Invitrogen, Carlsbad, CA) extraction method. Gene expression was analyzed via whole genome microarray profiling, using U133 Affymetrix arrays and platforms as described (41). Selection of the significantly modified transcripts was performed by a multistep filter strategy. Among the genes that showed an absolute call of present according to MAS 5.0 algorithm, we selected the genes with a difference in signal detection of ≥4.5 times the average background minus the scaled noise. The hybridization intensities of significantly modified genes were examined by hierarchic cluster analysis with Gene-Spring GX (Agilent Technologies, Santa Clara, CA) and Treeview software (free). The genes that satisfied these criteria were then selected on the basis of a fold change >1.52 or <1.52 that was consistent in at least two comparisons. Genes that related to energy metabolism were identified according to the function of their putative encoded proteins from public databases. (IPA; Ingenuity Systems, Redwood City, CA). Changes in gene expression were validated by real-time PCR (Roche Thermocycler; Roche Applied Science, Indianapolis, IN) with LightCycler FastStart DNA SYBR Green 1 master mix and primers from Integrated DNA Technologies (Coralville, IA). Primers for specific target sequences were designed within the 3′ coding region of the genes.

Statistics.

All values are presented as means ± SE. Differences between dependent variables were examined with one-way or two-way repeated-measures analysis of variance. Significant mean differences among the three time points were identified with Fisher's protected least significant difference post hoc test. The data were analyzed using the StatView II statistical package (Abacus Concepts, Berkeley, CA). Statistical significance was set at P < 0.05.

RESULTS

Longitudinal phenotypic and metabolic changes.

Longitudinal measurements were performed prepregnancy (P) and in both early (8–12 wk) (E) and late pregnancy (36–38 wk) (L). Metabolic and anthropometric parameters are listed in Table 1. Both plasma cholesterol and triglyceride concentrations were increased in L, whereas plasma free fatty acids were decreased compared with pregravid measurements. Insulin sensitivity decreased significantly in L compared with both E and P. Leptin concentrations increased significantly in E and remained elevated until L. Measurements of body composition showed that total body fat mass was increased in late gestation (P < 0.001; Fig. 1). An increase in the size and number of adipocytes was observed in L compared with pregravid but was not detected at early stages (P < 0.001; Fig. 1).

Table 1.

Anthropometric and metabolic characteristics of women before and during pregnancy

	Prepregnancy	Early Pregnancy	Late Pregnancy
Weight kg	70.2 ± 5.6	72.1 ± 5.2	84.3 ± 4.8*
BMI, kg/m2	24.0 ± 2.0	24.7 ± 1.8	28.9 ± 1.7*
Total body fat mass, kg	19.9 ± 4.5	20.4 ± 4.1	25.9 ± 3.8*
Lean body mass, kg	50.3 ± 1.8	51.6 ± 1.5	58.3 ± 2.0*
Triglycerides, mg/dl	58 ± 10	63 ± 5	195 ± 55**
Cholesterol, mg/dl	156 ± 11	154 ± 18	240 ± 18*
FFA basal, μmol/ml	740 ± 21	610 ± 13	560 ± 12*
FFA clamp, μmol/ml	140 ± 19	120 ± 40	202 ± 39*
Glucose, mg/dl	90 ± 2	82 ± 3	79 ± 3*
Insulin, μU/ml	12 ± 1.2	10 ± 1.8	16.2 ± 2.8
ISI, mg·zmd·min−1·zmd·kg−1	8.6 ± 0.9	8.0 ± 0.8	5.2 ± 0.7*
Leptin, ng/ml	12.6 ± 4.8	22.0 ± 8.1	26.5 ± 8.3*

Results are means ± SE of n = 6 subjects whose adipose tissue was analyzed for gene expression.

BMI, body mass index; FFA, free fatty acids; ISI, insulin sensitivity index.

^**P < 0.001 and

^*P < 0.01, statistical significance compared with prepregnancy.

Fig. 1.

Morphological changes in adipose tissue in human pregnancy. Top: total adipose tissue mass, adipocyte size, and number of adipocytes prior to and during healthy uncomplicated pregnancy. Early: 8–11 wk of gestation; late: 34–36 wk of gestation. Statistical significance: *P < 0.05 vs. pregravid. Results are means ± SD for 11 women. Bottom: immunohistochemistry of subcutaneous gluteal adipose tissue sections. Initial magnification, ×20. P, pregravid; E, early pregnancy (8–12 wk); L, late pregnancy (34–36 wk).

Molecular characteristics of adipose tissue.

Using whole genome microarray analysis, the adipose tissue transcriptome, encompassing 12,897 ± 417 genes, remained quantitatively stable during pregnancy, with 13,510 ± 197 genes expressed during E and 13,097 ± 298 genes expressed during L. Principal component analysis revealed a large overlap in expression pattern between the pregravid, E, and L stages (Fig. 2). The genes differentially expressed in E (n = 1,286) and L (n = 1,111) represented 10–15% of the adipose tissue transcriptome. Comparison of global gene expression also revealed a distinct expression pattern in E and L (Fig. 2). Analysis of differentially modified genes based on their biological function indicated that 65% of pregnancy-associated changes were related to tissue remodeling (39%) and lipid metabolism (26%) (Tables 2 and 3 and Fig. 3). Genes involved in tissue remodeling were further separated into three main functional categories: mediators of the immune response, extracellular matrix components, and angiogenic factors (Fig. 3). Marked changes in genes regulating pathways for the inflammatory response and metabolism were noted in E compared with P. Innate immune responses elicited through the TLR4/NF-κB signaling pathway led to early increases in gene and protein expression of TLR4 as well as TNFα, IL-6, and small chemokines of the C-X-CL family (Fig. 4, A and B). Metabolic genes facilitating lipid accumulation through adipocyte differentiation and lipogenesis were most activated in E (Fig. 5, A and B, and Table 4). High expression of transcription factors of the forkhead box O (FOXO), peroxisome proliferator-activated receptor-γ (PPARγ), cAMP response element-binding protein, and sterol regulatory element-binding protein families were indicative of adipocyte differentiation. Increased VLDL receptor and LDL-related protein 1, fatty acid synthase, acetyl-CoA carboxylase, and stearoyl-CoA desaturase catalyzing fatty acid uptake, synthesis, and elongation were evidence of active lipogenesis. Extracellular matrix components with structural (elastin, collagen) and adhesive properties (laminin, fibronectin) were activated predominantly in E and included upregulation of metalloproteinases to enhance tissue plasticity through the release of insulin-like growth factor I (IGF-I). The concomitant activation of several angiogenic factors [VEGF, TGFβ, angiotensin, fibroblast growth factor receptor 2 (FGFR2)] suggested that neovascularization was associated with matrix remodeling (Table 2).

Fig. 2.

Three-dimensional scatter plot principal component analysis (PCA) of the adipose tissue transcriptome. Gluteal adipose tissue obtained longitudinally in P, E, and L from 6 of the 11 women. Full cohort was processed for microarray analysis. A: analysis of full data set (n = 22, 277 genes) showed overlap in transcriptome profiles before and during pregnancy. B: analysis of the genes modified significantly in E and L (n = 1,807 genes) compared with pregravid showed a distinctive pattern at each stage of pregnancy. Each knot represents 1 microarray data set.

Table 2.

Genes differentially modified in adipose tissue in early pregnancy

NCBI Accession No.	Gene Name	Fold Change	Description
Lipid metabolism
NM_006415	SPTLC1	1.5	Serine palmitoyltransferase long chain
NM_004863	SPTLC2	1.6	Serine palmitoyltransferase long chain
NM_030791	SGPP1	17.5	Sphingosine-1 phosphate phosphatase
D85181	SC5DL	1.7	Sterol-C5-desaturase
AF116616	SCD	1.5	Stearoyl-CoA desaturase
S69189	ACOX1	−1.6	Acyl coenzyme A oxidase
NM_013402	FADS1	1.7	Fatty acid desaturase
NM_001995	ACSL1	1.5	Acyl-CoA synthetase long-chain
NM_004104	FASN	1.7	Fatty acid synthase
NM_198834	ACACA	1.5	Acetyl-CoA carboxylase-α
NM_023928	AACS	1.5	Acetoacetyl CoA synthetase
NM_024090	ELOVL6	6.2	Elongation of long chain fatty acids 6
NM_022821	ELOVL1	1.6	Elongation of very long chain fatty acids 1
U18197	ACLY	1.6	ATP cytrate lyase
NM_002317	LOX	2.0	Lysyl oxidase
NM_002979	SCP2	1.5	Sterol carrier protein
NM_006227	PLTP	1.5	Phospholipid transfer protein
NM_000237	LPL	1.7	Lipoprotein lipase
NM_015869	PPARG	1.6	Peroxisome proliferator-activated receptor
NM_001455	FOXO3B	2.1	forkhead box O3
NM_002015	FOXO1	1.5	forkhead box O1
NM_003355	UCP2	2.5	Uncoupling protein 2
NM_000098	CPT2	−1.5	Carnitine palmitoyltransferase 2
NM_000017	ACADS	−1.5	Acyl-coenzyme A dehydrogenase short chain
NM_001608	ACADL	−1.7	Acyl-CoA dehydrogenase long chain
NM_005327	HADHA	−1.5	Hydroxyacyl-CoA dehydrogenase
NM_000041	APOE	−1.7	Apolipoprotein E
NM_207630	CREM2	1.5	cAMP-responsive element modulator
NM_204203	C/EBP	1.8	CCAAT/enhancer binding protein-γ
NM_201988	CREB	1.7	cAMP responsive element binding protein
NM_221147	WWOX	12.1	WW domain containing oxidoreductase
Growth factors and matrix remodeling
NM_003376	VEGFA	1.7	Vascular endothelial growth factor A
NM_001078	VCAM1	1.8	Vascular cell adhesion molecule
NM_000612	INS-IGF2	4.6	Insulin-like growth factor 2
M37484	IGF1	2.0	Insulin like growth factor 1
NM_022970	FGFR2	1.5	Fibroblast growth factor receptor 2
NM_012098	ANGPTL2	1.6	Angiopoietin like 2
AB000889	PPAP2B	1.9	Phosphatidic acid phosphatase 2B
NM_004995	MMP-14	2.0	Matrix metalloproteinase 14
NM_001846	COL4A2	3.2	Collagen type IV α2
NM_003873	NRP1	1.6	Neuropilin 1
NM_001069	TUBB2B/2A	2.8	Tubulin-β 2A and 2B
NM_000224	KER18	3.5	Keratin 18
NM_002964	S100A8	1.5	S100 calcium binding protein A8
AF019888	ARPC4	3.2	Arp complex 20 kDa subunit
Immune responses
NM_002184	IL6ST	3.4	Interleukin-6 signal transducer
NM_000878	IL2RB	1.9	Interleukin-2 receptor-β
NM_004515	ILF2	1.5	Interleukin enhancer-binding factor
M97935	STAT1	2.5	Signal transducer and activator of transcription
NM_002982	CCL2	2.2	Chemokine (C-C motif) ligand
NM_002089	CXCL2	2.0	Chemokine (C-X-C motif) ligand
NM_005409	CXCL11	2.4	Chemokine (C-X-C motif) ligand
NM_002416	CXCL9	1.6	Chemokine(C-X-C motif) ligand
NM_005408	CCL13	1.5	Chemokine (C-C motif) ligand
NM_001565	CXCL10	1.5	Chemokine C-X-C motif ligand
NM_014452	TNFRSF21	2.0	Tumor necrosis factor receptor
NM_002546	TNFRSF11B	1.7	Tumor necrosis factor receptor
NM_000416	IFNGR1	1.5	Interferon-γ receptor
NR_024168	TLR4	1.5	Toll-like receptor 4
NM_000591	CD14	1.5	CD14 molecule
NM_004139	LBP	5.2	Lipopolysaccharide-binding protein
NM_001556	IKBKB	1.8	Inhibitor k light polypeptide enhancer
NM_003998	NFKB1	1.5	Nuclear factor k light polypeptide enhancer
NM_014294	TRAM1	1.7	Translocation associated membrane protein 1
BC003388	TANK	1.6	TRAF family member associated NFKB activator
NM_020056	HLA-DQA1	1.6	Major histocompatibility complex
NM_006512	SAA4	1.5	Serum amyloid A4
NM_030754	SAA1	1.6	Serum amyloid A1
NM_208703	SAAP	3.7	Amyloid β (A4) precursor
NM_001212	C1QBP	1.5	Complement component 1q binding protein

NCBI, National Center for Biotechnology Information. Changes are expressed in fold change compared with pregravid. GSE37215 approval (NCBI tracking system no. 16555206).

Table 3.

Genes differentially modified in adipose tissue in late pregnancy

NCBI Accession No,	Gene Name	Fold Change	Description
Immune-related genes
NM_004139	2	LBP	Lipopolysaccharide-binding protein
NM_138554	1.6	TLR4	Toll like receptor 4
NM_003265	2.5	TLR3	Toll-like receptor 3
NM_003326	2.8	TNFSF4	Tumor necrosis factor ligand
NM_014452	1.9	TNFRSF21	Tumor necrosis factor receptor 21
NM_014452	1.7	TNFRSF20	Tumor necrosis factor receptor 20
NM_002389	1.6	CD46	CD46 molecule complement regulatory protein
NM_000331	1.6	SAA1	Serum amyloid A 1
NM_006512	3.7	SAA4	Serum amyloid A4
BC020795	2.3	SAA2	Serum amyloid A2
NR_026576	9.1	SAA3P	Serum amyloid A3 pseudogene
NM_004887	1.6	CXCL14	Chemokine (C-X-C motif) ligand 14
NM_003150	1.5	STAT3	Signal transducer and activator of transcription
NM_002163	2.3	IRF8	Interferon regulatory factor 8
NM_001161529	2.3	CSF2RA	Colony-stimulating factor 2 receptor-α
NM_002852	2	PTX3	Pentraxin-related gene
NM_018643	2	TREM1	Triggering receptor expressed on myeolid cells
NM_001040059	1.9	CD68	CD68 molecule
NM_000757	1.9	CSF1	Colony-stimulating factor 1
NM_001161529	1.9	CSF2RA	Colony-stimulating factor 2, receptor A
NM_000395	1.5	CSF2RB	Colony-stimulating factor 2, receptor B
AY312956	2.5	TRADV3	T cell receptor-δ chain
NM_005384	1.7	NFIL3	Nuclear factor interleukin-3
NM_000585	1.5	IL15	Interleukin-15
NM_001190981	1.9	IL6ST	Interleukin-6 signal transducer
NM_000877	2.5	IL1R1	Interleukin-1 receptor type 1
NM_001560	3.1	IL13RA1	Interleukin 13 receptor α1
NM_030968	1.6	C1QTNF1	C1q and tumor necrosis factor related protein 1
NM_016546	1.9	C1RL	Complement component 1 receptor like
NM_001115131	3.2	C6	Complement component 6
NM_002982	2.5	CCL2	Chemokine (C-C motif) ligand 2
NM_001168298	1.7	CXCR2	Chemokine (C-X-C motif) receptor 2
NM_002089	2.3	CXCL2	Chemokine (C-X-C motif) ligand 2
NM_006271	2.8	S100A1	S100 calcium-binding protein A1
NM_002964	3.5	S100A8	S100 calcium-binding protein A8
NM_002965	3.5	S100A9	S100 calcium-binding protein A9
Glucose and lipid metabolism
NM_001135585	24.2	SLC2A5	Facilitated glucose/fructose transporter
NM_014668	6.9	GREB1	GREB1 protein
NM_013402	5.7	FADS1	Fatty acid desaturase 1
NM_004265	4.9	FADS2	Fatty acid desaturase 2
NM_004104	4.3	FASN	Fatty acid Synthase
NM_000384	3.5	APOB	Apolipoprotein B
NM_005063	3.05	SCD	Stearoyl-CoA desaturase-δ9
NM_025225	2.5	PNPLA3	Patatin-like phospholipase domain 3
NM_020376	1.7	PNPLA2	Patatin like phospholipase domain 2
NM_001161587	2.5	GYS1	Glycogen synthase 1
NM_021957	4.6	GYS2	Glycogen synthase 2
NM_001018056	2.5	VLDLR	Very-low-density lipoprotein receptor
NM_001161504	2.3	ALDH4A1	Aldehyde dehydrogenase 4 family A1
NM_005589	1.7	ALDH6A1	Aldehyde dehydrogenase 6 family A1
NM_001608	2	ACADL	Acyl-coenzyme A dehydrogenase long chain
NM_001012727	1.9	AGPAT2	Acylglycerol-3 phosphate acyltransferase 2
NM_018441	1.9	PECR	Peroxisomal trans 2 aneoyl CoA reductase
NM_000284	1.7	PDHA1	Pyruvate dehydrogenase α1
NM_005164	1.6	ABCD2	ATP-binding cassette subfamily D 2
NM_001184705	1.6	HADH	Hydroxylacyl CoA dehydrogenase
NM_000925	1.5	PDHB	Pyruvate dehydrogenase
NM_024420	1.5	PLA2GA4	Phospholipase A 2 group IV
Growth factors
NM_000141	3.0	FGFR2	Fibroblast growth factor receptor 2
NM_000618	2	IGF-I	Insulin-like growth factor I
NM_000599	1.5	IGFBP5	Insulin-like growth factor-binding protein 5
NM_001552	1.6	IGFBP4	Insulin-like growth factor-binding protein 4
NM_005544	2.5	IRS1	Insulin receptor substrate-1
NM_005542	1.9	INSIG1	Insulin-induced gene 1
NM_016133	2.3	INSIG2	Insulin-induced gene 2

Levels of expression are expressed in fold change compared with pregravid. GSE37215 approval (NCBI tracking system no. 16555206).

Fig. 3.

Transcriptional changes in adipose tissue during pregnancy. Functional analysis of the genes whose expression was modified during pregnancy compared with pregravid identified 4 main gene clusters based on biological annotation of their DNA sequences. ECM, extracellular matrix.

Fig. 4.

Pregnancy-related changes in adipose tissue Toll-like receptor 4 (TLR4) network. Genes implicated in TLR4 signaling pathways were selected from the lists in Table 2. A: mRNA levels were measured by quantitative RT-PCR analysis of adipose tissue from early pregnancy. Real-time threshold cycle (C_T) values were normalized to actin and expressed in fold change vs. pregravid. B: immnublot longitudinal analysis of TLR4 protein content in adipose tissue at P, E, and L. C: diagram of the molecular pathways related to TLR4 activation identified with gene ontology analysis. Statistical significance: *P < 0.05. Gray bars, pregravid; black bars, early pregnancy; n = 6 independent determinations. LPS, lipopolysaccharide; LBP, lipopolysaccharide-binding protein; FFA, free fatty acid; TRAM, TRIF-related adaptor protein; IKBKB, inhibitor of κ-light polypeptide gene enhancer.

Fig. 5.

Pregnancy-related changes in adipose tissue metabolic pathways. mRNA levels of adipose-specific enzymes and transcription factors selected from Table 2 were measured by quantitative RT-PCR analysis. Real-time C_T values were normalized to actin and expressed in fold change vs. pregravid. B: diagram of the molecular pathways identified with gene ontology analysis of differentially regulated genes in early pregnancy. Statistical significance: *P < 0.05. Open bars, pregravid; filled bars, early pregnancy for n = 6 independent determinations. ELOVL6, elongation of very long chain fatty acids protein 6; ACS, acyl-CoA synthetase; FASN, fatty acid synthase; DGAT, diacylglycerol acyltransferase; ACACA, acetyl-CoA carboxylase-1α; SCD, stearoyl-CoA desaturase; GPAT, glycerol phosphate acyltransferase; LPGAT, lisophosphatidylglycerol acyltransferase; ACoA, acetyl-CoA; FADS, fatty acid desaturase; LPL, lipoprotein lipase; FOXO, forkhead box O; SREBP, sterol regulatory element-binding protein; C/EBP, CCAAT/enhancer-binding protein; PPARγ, peroxisome proliferator-activated receptor-γ.

Table 4.

Adipose tissue protein expression quantitated by immunoblot

Protein name	Prepregnancy	Early Pregnancy	Late Pregnancy
LBP	1	1.9 ± 0.6*	2.2 ± 0.6*
SLC2A5 (GLUT5)	1	3.6 ± 1.1*	5.4 ± 1.2*
Adiponectin	1	1.8 ± 0.2*	2.6 ± 0.6*
C/EBP	1	2.2 ± 0.4*	3.2 ± 0.7*
TLR4	1	2.9 ± 0.4*	3.8 ± 1.1*
LPL	1	1.5 ± 0.4ns	3.3 ± 0.1*
GAPDH	1	0.8 ± 0.1ns	0.8 ± 0.1ns

Results are means ± SD of n = 3 subjects; data are expressed as fold change over prepregnancy levels.

LBP, lipopolysaccharide-binding protein; SLC2A5, solute carrier family 2 (facilitated glucose transporter), member 5; GLUT5, glucose transporter 5; C/EBP, CCAAT/enhancer binding protein; LPL, lipoprotein lipase.

^*P = 0.05; ^nsnonsignificant statistical significance compared with prepregnancy.

DISCUSSION

Our study demonstrates an enlargement of adipose tissue during pregnancy, combining a higher cell number (hyperplasia) and a larger cell size (hypertrophy). This is in agreement with data for lower body depots in women (17, 50) but differs from the mechanisms described in obesity with hypertrophy occurring prior to hyperplasia. Hyperplasic growth appears to be genetically regulated, whereas hypertrophy is more nutritionally regulated (29). Growth is also dependent on each depot-unique gene signature, which defines the proliferation of adipocyte precursors (57). The longitudinal changes were characterized in the gluteal subcutaneous adipose depot. This depot is preferentially expanding in pregnancy (18), accessible to biopsy, and more prone to inflammatory changes than mesenteric or omental visceral depots (19). Further studies performed at visceral sites will help to achieve a global picture of adipose tissue changes.

Expansion and remodeling.

Our study reveals that adaptations of adipose tissue to healthy pregnancy develop in a temporal manner, with an array of molecular changes preceding the anthropometric expansion of adipose mass. Adipose tissue enlargement relies on molecular cross-talks between distinct cell types of the stromal-vascular fraction surrounding the adipocytes (45, 48). In agreement with this concept, our study suggests that pregnancy-induced adipose tissue expansion involves a combination of signaling pathways shared by preadipocytes and immune and endothelial cells (Fig. 6).

Fig. 6.

Model of functional networks contributing to adipose tissue remodeling in human pregnancy. Transcriptional activation patterns suggest that molecular networks from several adjacent cell types cooperate with the adipose tissue remodeling during pregnancy. ECM components and angiogenic factors are needed for vascular and adipocyte growth. Lipogenic genes and transcription factors are required for cell differentiation and lipid storage. Macrophages infiltrated between the stromal cells produce proinflammatory cytokines IL-6, IL-8, and TNF, which may 1) enhance neovascularization and 2) facilitate the development of insulin resistance. ACC, acetyl-CoA carboxylase; MCP-1, monocyte chemoattractant protein-1; ACLY, ATP citrate lyase; AGPAT, acylglycerol-3-phosphate acyltransferase; MMP-14, metalloproteinase-14; COL1A2, collagen type 1 α2; PAI-1, plasminogen activator inhibitor-1; VCAM, vascular cell adhesion molecule-1, ANGPTL2, angiopoietin-like 2; FGFR2, fibroblast growth factor receptor 2.

The flexibility of the cellular matrix surrounding the adipocytes plays a central role in the regulation of tissue expandability. Remodeling of extracellular matrix components is accelerated in a number of metabolic diseases that are associated with changes in the hormonal milieu (12, 48). Increases in metalloproteinases, fibronectin, and laminin accelerate preadipocyte differentiation in hypertrophic adipose tissue of obese individuals (20, 47). On the other hand, angiogenesis, which is also required for adipose tissue remodeling, may be facilitated by metalloproteinases and collagenases (11). VEGF-A and its most important functional partner, angiopoietin, are essential for initiation of the angiogenic program (59, 61). Hence, the combined increase in growth factors (IGF-II, FGFR2) and angiogenic factors (VEGF, TGFβ) is in line with the contribution of angiogenic mechanisms to the overall adipose tissue remodeling of early pregnancy (49).

The microenvironment surrounding the adipocyte may also be shaped by the production of adipokines that exert both paracrine and endocrine effects. TNFα and IL-6 exhibit increased expression in both the early and late stages of pregnancy, and therefore, they may regulate the expression of other genes that contribute to the negative action of insulin (21). The increase in leptin receptor expression is in agreement with a stimulatory role of leptin on angiogenesis, whereas the lack of leptin gene regulation is consistent with the rise in plasma leptin levels in pregnancy being contributed mainly by the placenta (34).

Adipocyte biology.

The 29% increase in fat mass measured in late pregnancy was associated with active cellular remodeling of adipose tissue. The activation of endothelial LPL suggested accelerated hydrolysis of circulating triglycerides into free fatty acids available for cellular uptake. Our findings bring support to a crucial role of LPL for increasing lipid storage in adipose tissue of pregnant rats (42). The upregulation of many other genes directly affecting lipogenesis and adipocyte differentiation indicated that those pathways are recruited early in pregnancy to enhance lipid storage in both mature and newly synthesized adipocytes. The forkhead transcription factor FOXO1 isoform, a master regulator of energy homeostasis in white adipose tissue, was one of the early induced genes. Activation of FOXO1 may contribute to the regulation of adipocyte size in response to excessive calorie intake (36, 37). Because FOXO1 is a direct PPARγ ligand, it may also contribute to the regulation of adipose size and adipogenic pathways (15, 53). The nutrient-sensitive genes C/EBPα, sterol regulatory element-binding protein-1, and PPARγ, which are mandatory regulators of both early and late programs of adipose differentiation (6, 14, 56), were also activated early in pregnancy. As a whole, the changes in the metabolic machinery indicated that enhanced adipocyte storage and increased differentiation of new adipose cells were initiated at early pregnancy stages. These findings emphasize the notion that the molecular mechanisms of the adipogenic process come into place well before the appearance of classical phenotypic markers. Our data also suggest a similarity between the mechanisms of adipose tissue physiological expansion in pregnancy and pathological expansion in obesity. One example is the early increase in secreted frizzled-related protein 1, an endogenous modulator of Wnt/β-catenin signaling that peaks in patients with mild obesity and gradually falls in morbidly obese subjects (Table 2 and Ref. 31).

Inflammation.

Signs of molecular inflammation were detected in adipose tissue from early stages throughout the end of pregnancy. The functional cluster of genes related to immune regulation represented 30% of all modified genes in pregnancy compared with the pregravid state (Fig. 2). Increased expression of the macrophage markers CD14, CD68, HLA-DR, and HLA-DQ and the mannose receptor was in line with a recent report of inflammatory changes in adipose tissue in mice in late pregnancy (60). mRNA of TLR4 and downstream modulators were increased in early pregnancy. TLR4 belongs to a family of membrane receptors first recruited in innate sensing through binding of the lipid A moiety of lipopolysaccharide released from gram-negative bacteria (46).

The environmental factors that may trigger the recruitment of innate immune pathways in early pregnancy are not known. The activation of TLR4 signal transduction has been proposed as a molecular link between diet-induced obesity and increased insulin resistance (5, 54). Nutritional changes through either maternal hyperphagia or dysphagia are potential candidates to impact adipose tissue receptors and function (4). Along this line, changes in microbiota have been documented in relation to gestational weight gain and may impact TLR4 signaling in pregnancy (13). Bacterial recognition by the serum amyloid A (SAA) receptor CD36 may also initiate inflammation in response to fat supply (5, 59). The acute-phase proteins of the SAA family are markers of low-grade inflammation and obesity and promote endothelial remodeling partly through TLR4-mediated pathways (39, 58, 1).

Conclusion.

Whereas substantial literature has established inflammation as an obligatory component of adipose tissue remodeling in obesity and other diseases, this report is the first indication of recruitment of inflammatory pathways in healthy human pregnancy. Enhancement of adipose immune response preceded the appearance of maternal phenotypic changes in body composition and insulin action. This biphasic pattern reveals physiological inflammation as an early step toward the development of physiological insulin resistance, which peaks later in pregnancy.

GRANTS

This study was supported by National Institute of Child Health and Human Development Grant RO1-HD-022965 to P. Catalano and S. Hauguel-de Mouzon.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

V.R. analyzed the data; V.R. and S.H.-dM. interpreted the results of the experiments; V.R. and M.H. prepared the figures; V.R. and S.H.-dM. drafted the manuscript; V.R., S. Basu, M.H., L.P., S. Bernard, P.M.C., and S.H.-dM. approved the final version of the manuscript; S. Basu, M.H., L.P., J.M., B.K., and S. Bernard performed the experiments; P.M.C. and S.H.-dM. did the conception and design of the research; P.M.C. and S.H.-dM. edited and revised the manuscript.