Ontogeny and Thermogenic Role for Sternal Fat in Female Sheep

Abstract

Brown adipose tissue acting through a unique uncoupling protein (UCP1) has a critical role in preventing hypothermia in newborn sheep but is then thought to rapidly disappear during postnatal life. The extent to which the anatomical location of fat influences postnatal development and thermogenic function in adulthood, particularly following feeding, is unknown, and we examined both in our study. Changes in gene expression of functionally important pathways (i.e., thermogenesis, development, adipogenesis, and metabolism) were compared between sternal and retroperitoneal fat depots together with a representative skeletal muscle over the first month of postnatal life, coincident with the loss of brown fat and the accumulation of white fat. In adult sheep, implanted temperature probes were used to characterize the thermogenic response of fat and muscle to feeding and the effects of reduced or increased adiposity. UCP1 was more abundant in sternal fat than in retroperitoneal fat and was retained only in the sternal depot of adults. Distinct differences in the abundance of gene pathway markers were apparent between tissues, with sternal fat exhibiting some similarities with muscle that were not apparent in the retroperitoneal depot. In adults, the postprandial rise in temperature was greater and more prolonged in sternal fat than in retroperitoneal fat and muscle, a difference that was maintained with altered adiposity. In conclusion, sternal adipose tissue retains UCP1 into adulthood, when it shows a greater thermogenic response to feeding than do muscle and retroperitoneal fat. Sternal fat may be more amenable to targeted interventions that promote thermogenesis in large mammals.

In the majority of large mammals studied to date, birth is a critical period for the rapid recruitment of nonshivering thermogenesis in brown adipose tissue (BAT), and this coincides with the maximal appearance of uncoupling protein (UCP1) (1, 2). This is followed by a transformation of fat from a brown to a white appearance, although the rate and magnitude of this process can vary between depots (3). For example, in humans the supraclavicular (or neck) depot retains UCP1 into adulthood (4), whereas the peri-adrenal depot does not (5). Consequently, in adults the supraclavicular depot has the capacity to exhibit a major thermogenic response to both cold exposure (6) and diet (7). The extent to which the retention of UCP1 through the life cycle, and thus the associated thermogenic potential, is determined by a fat depot’s early development and/or anatomical location is currently unknown.

Studies in rodents initially suggested that brown and white adipocytes arise from different lineages and that brown adipocytes may originate from the same precursor as skeletal muscle (8). This relationship now appears to be more complex and depot specific because white adipocytes can have diverse and mixed origins (9, 10), potentially overlapping with brown adipocytes (11). Additional populations of adipocytes have been identified as beige or “brite” (i.e., brown-in-white) and are characterized as small populations of UCP1-expressing cells surrounded by large numbers of white adipocytes (12, 13). The thermogenic relevance of these cells remains to be established, as their UCP1 content is only 10% of that of classic BAT (14). To date, most studies investigating beige fat have been confined to adult rodents in which “almost everything” examined was able to “brown” white adipose tissue (15). Furthermore, a diverse range of molecular markers for beige adipocytes has been suggested, but their applicability across species (16), as well as the optimal conditions in which these classifications are defined (17), is being questioned.

Our study therefore had two aims. The first was to compare the gene-expression profiles for the primary thermogenic, metabolic, and functional markers of brown, beige, and white adipose tissue in retroperitoneal fat [the most abundant depot in fetal sheep and a “classic” adipose tissue depot (18)], sternal (or neck) fat, and hind limb muscle, a representative skeletal muscle. Sheep, like many large mammals, do not possess the interscapular BAT (19) that is present in rodents. The current comparison, undertaken in young sheep, spanned the period from birth to 1 month of postnatal life, examining three important time points that in retroperitoneal fat are coincident with the peak abundance of UCP1 after birth (i.e., 1 day of age), the age at which UCP1 has declined to basal amounts before the onset of rapid growth (i.e., 7 days of age), and the subsequent loss of UCP1 (i.e., 28 days of age (1). We also conducted detailed molecular analyses to establish whether each depot could have a different developmental origin and if adipose tissue in the sternal depot is more similar to skeletal muscle than to classic (i.e., retroperitoneal) adipose tissue. The second aim was to examine whether the sternal depot responds to the thermogenic stimulus of feeding in adulthood (20) and if this is comparable to responses in skeletal muscle rather than those in classic adipose tissue. In addition, we examined whether diet-induced obesity or low body weight modulated this response and thus whether sternal fat could be a potential target for promoting thermogenesis.

Methods

Animal experimentation

All animal work was approved by the relevant animal ethics committees at The University of Nottingham or Monash University. To avoid any confounding effects of sex and/or a disproportionate increase in muscle mass following overfeeding [as seen in males (21)], only females were studied.

Study 1. The effects of postnatal age on sternal adipose tissue development

Ten triplet-bearing sheep of mixed breed that all gave birth naturally to appropriately grown offspring at term (over a 4-week period) were entered into the study. Triplets were chosen for the study because this meant there would be no confounding maternal influences between each sampling date. One lamb from each mother was therefore randomly selected at either 1, 7, or 28 days of age, had blood sampled from the jugular vein, and then was euthanized by injection of sodium pentobarbital (0.5 mL/kg). The sternal and retroperitoneal adipose tissue depots were dissected and weighed, together with a representative sample from the hind limb muscle (vastus lateralis). Samples were immediately placed in liquid nitrogen and were then stored at −80°C until analyzed.

Study 2. The effects of altered body weight and fat mass on the temperature response to feeding in sternal adipose tissue

Manipulation of adult body weight

Fifteen adult female Corriedale sheep aged between 3 and 5 years and of normal body weight (∼55 kg) were all ovariectomized (to avoid any confounding effects of the reproductive cycle on tissue temperature) and were then randomly divided into three different body weight groups. Five animals were made lean (32.5 ± 1.5 kg) by feeding them a restricted diet of ∼500 g of lucerne chaff per day, and five became obese (79.9 ± 3.7 kg) after receiving a supplemented diet of lucerne hay ad libitum and ∼300 g daily of high-energy food supplement (i.e., oats and lupin grain) (22, 23). The remaining five “control” animals (52.6 ± 1.1 kg) were maintained on pasture. Differential body weights were then maintained for 1 year before experimentation. To assess the temperature response to feeding, the diets were standardized across all groups, and the high-energy supplementation of the obese group stopped. Visceral adiposity was determined at the time of euthanasia (lean: 0.04 ± 0.01 kg; control: 1.58 ± 0.18 kg; obese: 4.11 ± 0.19 kg).

Profiling postprandial changes in temperature and metabolites

For tissue-specific temperature recordings, customized dataloggers with 10-cm or 20-cm download leads (SubCue, Calgary, AB, Canada) were inserted into the skeletal muscle of the hind limb (vastus lateralis) and the sternal (midline) and retroperitoneal fat and were set to record temperatures at 15-minute intervals as previously published (20, 24). After surgery, the animals were housed indoors to enable precisely timed “meal” feeding and were exposed to natural variations in photoperiod and ambient temperature. To entrain postprandial thermogenesis, animals were placed on a temporal program-feeding regimen, in which they had access to food at set mealtimes between 1100 and 1600 hours each day (20, 24, 25). Animals were program-fed for 2 weeks before the onset of experimentation. To characterize changes in plasma metabolite and insulin levels with feeding, blood samples (6 mL) were collected into heparinized tubes at 30-minute intervals between 1000 and 1600 hours. The samples were centrifuged to obtain plasma, which was stored at −20°C until assayed.

Food intake was recorded after offering the obese and standard animals 2 kg of lucerne chaff and monitoring any refusals, whereas the lean animals received and consumed 500 g of chaff per day. Food intake, once corrected to body weight, was similar in the obese and lean animals, whereas the control group ate slightly more (P < 0.01) than the other two groups (data not shown).

Temperature data were downloaded, and the diurnal thermogenic pattern and the postprandial response were then analyzed in each tissue. After 3 weeks of program feeding, all animals were euthanized, as described previously, between 830 and 1030 hours. Representative fat and muscle samples were collected and stored as described previously for study 1. At this time, it was confirmed that each temperature probe was still located within the same anatomical position as at surgery.

Laboratory analysis

Gene expression

Total RNA was extracted, and following confirmation of RNA integrity, gene expression was determined using real-time polymerase chain reaction (PCR) (1). The specificity of each ovine primer was confirmed by classic PCR with ovine complementary DNA from suitable tissue samples, negative controls, and ovine genomic DNA and by analyzing the products by agarose gel electrophoresis. The primers were used only when there was a clear single band on the gel corresponding to the expected amplicon size, negative control lanes were clear, and any products from amplification of genomic DNA could be easily distinguished from the target. The PCR products of the selected primers were analyzed using high-sensitivity Sanger dideoxy sequencing, and the returned sequences were verified by alignment with the predicted on-line sequence to ensure that they were specific to the intended target. The primers used are summarized in Table 1. The amount of messenger RNA (mRNA) was calculated relative to the geometric mean of the most stable reference genes as determined by geNorm and/or NormFinder analysis. For the postnatal tissues, housekeeping genes included IPO8, KDM2B, RPLP0, and TBP, and cyclophilin, βactin, β₂-microglobulin, and malate dehydrogenase 1 were used for the adult tissue.

Table 1.

Summary of Specific Ovine Sequence of Forward and Reverse Oligonucleotide Primers Used for Real-Time PCR

Gene	Accession Number	Forward Primer	Reverse Primer	Amplicon Length (bp)
ACSM5	XM_015469322.1	CCACCATATGATGTGCAGGT	TGTCTTCTCAGGGTTGTCCA	138
ADIPOQ	NM_174742.2	ATCAAACTCTGGAACCTCCTATCTAC	TTGCATTGCAGGCTCAAG	232
ADIRF	NM_001114513.2	CCACAGAAGCAGGGCAGA	AAACCCGAGAAAGCCTCA	100
ATF2	XM_004004570.1	TCCCACTTGTTCGACCAGTCA	TTGACAGTATCGCCGTTGGT	151
C/EBPα	XM_004015623.1	CTGGAGCTGACCAGTGACA	GGGCAGCTGACGGAAGAT	96
C/EBPβ	NM_176788.1	ACGACTTCCTCTCCGACCTC	CCCAGACTCACGTAGCCGTA	85
CIDEA	NM_001083449.1	AAGGCCACCATGTACGAGAT	GGTGCCCATGTGGATAAGACA	138
CPT1b	NM_001034349.2	TGATCACGTATCGCCGTAAA	GAGCACATCTGTGTCCTTCC	137
DIO2	NM_001010992.3	AGCCGCTCCAAGTCCACTC	TTCCACTGGTGTCACCTCCT	175
En-1	XM_003581845.4	AACCCGGCCATACTGCTAAT	TTCTTCTTCAGCTTCCTGGTG	152
Eva-1	XM_004016067.3	GGAATTTCCGTCCTCGAGAT	AGGATGGAGACGTCATACCG	139
FABP4	NM_174314.2	TGAAATCACTCCAGATGACAGG	TGGTGGTTGATTTTCCATCC	98
GPR120	XM_002698388.1	CCTGGGACGTGTCATTTGCTA	CTGGTGGCTCTCCGAGTAGG	140
HOXC8	XM_002704245.5	TGTAAATCCTCCGCCAACAC	TGATACCGGCTGTAAGTTTGC	140
HOXC9	XM_002704244.2	GACCTGGACCCCAGCAAC	GCTCGGTGAGGTTGAGAAC	175
INSR	XM_002688832.3	CTGCACCATCATCAACGGAA	CGTAACTTCCGGAAGAAGGA	162
LEP	NM_173928.2	CCAGGATGACACCAAAACC	TGGACAAACTCAGGAGAGG	140
LHX8	XM_004003563.1	AGAGCACGCCACAAGAAACA	AGGGCTGGAGTCCAAGAGTT	199
NR3C1	NM_001206634.1	ACTGCCCCAAGTGAAAACAGA	ATGAACAGAAATGGCAGACATTTTATT	151
PGC1α	NM_177945.3	GATTGGCGTCATTCAGGAGC	CCAGAGCAGCACACTCGAT	84
PPARγ	NM_181024.2	GACCCGATGGTTGCAGATTA	TGAGGGAGTTGGAAGGCTCT	145
PPARγ	NM_181024.2	GACCCGATGGTTGCAGATTA	TGAGGGAGTTGGAAGGCTCT	145
PRDM16	XM_003583245.1	TGGCAGCTGGCTCAAGTACA	CGGAACGTGGGCTCCTCATC	198
PRLR	NM_174155.3	CTCCACCCACCATGACTGAT	CAGCGAATCTGCACAAGGTA	169
RIP140	XM_002684642.2	CGAGGACTTGAAACCAGAGC	TCTTAGGGACCATGCAAAGG	179
RyR1	NM_001206777	GGGATATGGGTGACACGAC	TCTCAGCATCAGCTTTCTCC	158
Serca 1a	XM_004020863.3	GCTGCTGTGGGCAATAAGAT	GCCAGTACCCCACTCTTTGA	150
Serca 2a	XM_012097784.2	CAGGTGTACCCACATTCGAG	TTCCCGAATGACAGACATGA	85
SHOX2	NM_001205527.1	CGCCTTTATGCGTGAAGAAC	TTGGCTGGCAGCTCCTAT	142
SREBF1	XM_004013336.1	AGGGGGACAAGGAGTTCTCA	CTCCGGCCATATCCGAACAG	72
Tbx15	NM_001079775.1	AATGGACATTGTACCTGTGGAC	TGACCACCTGTCTCATCCAA	158
TCF21	XM_014480981.1	ATCCTGGCCAACGACAAGTA	TCAGGTCACTCTCGGGTTTC	94
UCP1	XM_003587124.1	GGGCTTTGGAAAGGGACTACT	CAGGGCACATCGTCTGCTAAT	128
UCP2	NM_001033611	AAGGCCCACCTAATGACAGA	CCCAGGGCAGAGTTCATGT	128
UCP3	NM_001308581.1	ACCTGCTCACCGACAACTTC	CATATACCGCGTCTTCACCA	107
β2-Microglobulin	AY549962	CCAGAAGATGGAAAGCCAAA	CAGGTCTGACTGCTCCGATT	117
β-Actin	U39357	GCAAAGACCTCTACGCCAAC	TGATCTTGATCTTCATCGTGCT	120
Cyclophilin	JX534530	GCATACAGGTCCTGGCATCT	CATGCCCTCTTTCACTTTGC	136
IPO8	NM_001206120.1	GCCCTTGCTCTTCAGTCATT	GTGCAACAGCTCCTGCATAA	93
KDM2B	XM_004017579.1	CGGTCCTACCTCACTCAGGA	CCGTCTATGCTGGGCTTTCT	74
Malate dehydrogenase 1	AF233351	CGTTGCAGAGCTGAAGGATT	GGTGCACTGAGAGATCAAGG	100
RPLP0	NM_001012682.1	CAACCCTGAAGTGCTTGACAT	AGGCAGATGGATCAGCCA	227
TBP	NM_001075742.1	CTTGGACTTCAAGATTCAGAACA	CCAGGAAATAACTCTGGCTCA	120
YWHAZ	NM_174814.2	CCGGACACAGAACATCCAGTC	TCAGCTCCTTGCTCAGTTACAG	125

Shaded rows at the bottom of the table designate housekeeping genes.

Histology

Tissue sections were prepared as previously published (1) and were stained using hematoxylin and eosin and for UCP1. The number of adipocytes was counted in randomly positioned grids using a counting frame area of 62,500 µm². The Schaffer method was used, and ∼120 adipocytes were examined, with the coefficient of variation <2%. Adipocyte size was measured using the nucleator method (26, 27), and cell area was calculated by using orthogonal lines originating from the midpoint of the cell, which was taken as the center of the lipid droplet within complete adipocytes. For UCP1 immunohistochemistry, adjacent sections were collected. Sections were deparaffinized, and endogenous peroxide activity was blocked with 0.3% hydrogen peroxide in methanol. Sections were then washed and blocking serum (normal goat serum in 0.1 M phosphate-buffered saline) was added; rewashed; and incubated with primary antibody (1:100 rabbit anti-UCP1) for 24 hours at room temperature. Slides were then washed and incubated for 1 hour with secondary antibody (1:200 biotinylated anti-rabbit antibody; Antibodies Australia, Melbourne, Australia). Immunostaining was revealed using 3,3′-diaminobenzidine color reagent. Rat BAT was used as a positive control (primary antibody 1:1000), and staining without primary antibody was used to determine staining specificity.

Mitochondrial content and immunoblotting

The relative abundance of UCP1 and the total mitochondrial protein content were determined in the postnatal samples as previously described (28). In the adult samples, the relative abundance of UCP1, UCP3, SERCA1, and SERCA2a was determined using antibodies as previously published (20, 29, 30) and is summarized in Table 2. All data were corrected against the density of staining for total protein. Each antibody gave a signal at the correct molecular weight (see Supplemental Fig. 1), and the specificity of binding for each antibody was confirmed using nonimmune rabbit serum.

Table 2.

Antibodies Used

Peptide/ Protein Target	Antigen Sequence (if known)	Name of Antibody	Manufacturer, Catalog No., or Name of Source	Species Raised in Monoclonal or Polyclonal	Dilution Used	Research Resource Identifier
UCP1		Anti-UCP1	Antibodies Australia	Rabbit	0.1111111111	AB_2304253
UCP3	RALMKVQVLRESPF	Anti-UCP3	Abcam	Rabbit	0.7361111111	Ab34677
SERCA1		CaF2-5D2	Developmental Studies Hybridoma Bank	Mouse	0.7361111111	AB_531812
SERCA2a		CaS/C1	Developmental Studies Hybridoma Bank	Mouse	3.5138888889	AB_2061452

Plasma metabolite and hormone analyses

Plasma glucose and lactate were analyzed using an auto-analyzer (YSI, Inc., Yellow Springs, OH), and nonesterified fatty acid (NEFA) enzymatically (31). Plasma insulin (32) and irisin (33) were analyzed by enzyme-linked immunosorbent assay (Kit no. EK-067-29; Phoenix Pharmaceuticals, Inc., Burlingame, CA) in single assays.

Statistical analyses

Differences in gene expression and protein abundance between depots with age and/or different body weights were analyzed using Kruskal-Wallis nonparametric tests with Bonferroni correction for multiple analyses. In study 2, longitudinal data for temperature and plasma analysis were analyzed by repeated-measures analysis of variance. Differences in the temperature response to feeding and in adipocyte size were analyzed by one-way analysis of variance using Fisher’s least significant differences for post hoc analyses.

Results

Changes in UCP1 and gene expression profile between depots during early development

As expected, the abundance of the UCP1 gene and a key regulator of BAT function, DIO2, was highest in both fat depots examined at 1 day of age and then declined (Fig. 1). This adaptation in gene expression occurred as white adipose tissue mass increased substantially, with growth up to 28 days of age being greater in terms of relative body weight in the retroperitoneal depot (sternal 4.7 ± 0.5 g/kg; retroperitoneal 9.6 ± 1.2 g/kg; P < 0.05). At 1 day of age, the total mitochondrial protein content was greater in sternal fat than in retroperitoneal fat, which resulted in the total amount of UCP1 protein being higher in the sternal depot (sternal 1.4 ± 0.5 arbitrary units (au) per depot; retroperitoneal 0.3 ± 0.1 au per depot; P < 0.05). However, the rate of decline in UCP1 was greater between 1 and 7 days of age in the sternal depot (i.e., 7 days: sternal 0.7 ± 0.1 au per depot; retroperitoneal 0.5 ± 0.1 au per depot). UCP1 protein was undetectable in muscle at any time point.

Figure 1.

Summary of the changes in gene expression for putative markers of brown, beige, or white adipose tissue or skeletal muscle in the sternal and retroperitoneal fat depots and hind limb muscle over the first 28 days of life in young sheep. Values are means with their standard errors; n = 4 to 6 per age group. Significant differences between age groups are indicated by *P < 0.05.

Very low amounts of UCP1 and DIO2 mRNA were detected in skeletal muscle (Fig. 1). As each fat depot lost UCP1, there was a parallel decrease in gene expression for PDK4 [which is present in murine BAT (34)], a finding that was also apparent in muscle. In contrast, the gene that encodes for irisin (i.e.,FNDC5) was highly abundant in all three tissues examined, with a clear peak in muscle at 7 days of age. Although there were no changes in FNDC5 expression with age in retroperitoneal adipose tissue, it decreased in sternal fat. No change in plasma irisin concentration was observed with age (7 days: 117 ± 7 ng/mL; 28 days 129 ± 11 ng/mL). The lipid droplet protein CIDEA exhibited high transcript expression in both fat depots and did not change with age, whereas expression was very low in muscle. RIP140 was equally abundant in both fat and muscle, and although it showed a clear rise at 28 days in both fat depots, there were no significant age-related changes in muscle.

To establish a clearer overview of the differences in relative gene expression of a range of markers previously considered to be indicative of brown, beige, and white adipose tissue or skeletal muscle, a more exhaustive analysis was undertaken on the samples obtained at 7 days of age (Fig. 2). This demonstrated that other BAT-related genes (e.g., Eva1) were highly expressed in both fat depots, as were those genes primarily involved in either lipid metabolism (i.e.,FABP4) or adipogenesis (i.e.,PPARγ and CEBPα and β), whereas ADIRF was more abundant in sternal fat than in retroperitoneal fat. None of these genes were present in muscle. Other genes thought to potentially regulate adipose development, such as HOXC8 and HOXC9, and the “white fat marker” gene TCF21 were also highly expressed in retroperitoneal adipose tissue but not in muscle or, more surprisingly, in sternal adipose tissue. mRNA for En-1 was highly abundant in sternal fat but not in the other two tissues. In contrast, SHOX2 mRNA was abundant in sternal fat and muscle but was barely detectable in the retroperitoneal depot. Specific muscle marker genes CPT1b and Tbx15, which have been reported to be expressed during the differentiation or induction of BAT (35, 36), were highly expressed in muscle but minimally expressed in fat. Finally, ACSM5 mRNA was more abundant in muscle than in retroperitoneal fat but was hardly detectable in sternal fat.

Figure 2.

Summary heat map comparison of quantitative PCR gene expression between sternal and retroperitoneal adipose tissue and skeletal muscle from six individual sheep sampled at 7 days of age. All normalized data were made relative to the highest expressing sample and are given in arbitrary units between 0 and 1. Each column contains data from an individual animal for the three tissues examined, and each row contains data for a specific gene. Red squares represent the highest expression (1) and blue squares the lowest (0).

To further understand the development of sternal fat, the expressions of additional genes were examined (Table 3) and divided into four categories: developmental genes previously considered to be markers of brown or beige fat and those that regulate thermogenesis, metabolism, or adipogenesis. For thermogenic genes, we observed a reduction in PRLR and PGC1α expression between 7 and 28 days, whereas ATF2 expression increased. Expressions of the beige/white marker gene HOXC9 and the classic BAT marker gene LHX8 both rose substantially at each age, whereas expression of the BAT fate-determining gene PRDM16 showed a small decrease by 28 days and the beige marker gene SHOX2 was unchanged. Surprisingly, changes in the expression profiles of adipogenic genes varied. The mRNA abundance of both PPARγ and NR3C1 increased, whereas that of CEBPα transiently increased at 7 days of age and that of SREBF1 declined. A majority of other metabolic genes (i.e. adiponectin, leptin, and GPR120) also showed increased expression with age. However, mRNA abundance for FABP4 and the INSR was unchanged.

Table 3.

Changes in Gene Expression of Putative Markers of Thermogenic, Developmental, Adipogenic, and Metabolic Pathways in Sternal Adipose Tissue Over the First Month of Postnatal Life in Sheep

	Animal Age
	1 Day	7 Days	28 Days
Thermogenesis
PRLR	400 ± 80a	390 ± 50b	100 ± 10a,b
PGC1α	11,270 ± 1970a	10,800 ± 1540c	2220 ± 650a,c
ATF2	4770 ± 280d	4300 ± 180b	7090 ± 510b,d
CIDEA	304 ± 56	400 ± 30	310 ± 70
Development
HOXC9	65 ± 10a	110 ± 10	190 ± 30a
LHX8	10 ± 1d,e	120 ± 40d	210 ± 70e
PRDM16	120 ± 40d	80 ± 10	60 ± 10d
SHOX2	1190 ± 420	990 ± 90	1420 ± 210
Adipogenesis
PPARγ	21,810 ± 4 40a	34,670 ± 1890	46,480 ± 6110a
C/EBPα	12,300 ± 3070a	27,130 ± 1920a,c	12,820 ± 1180c
NR3C1	17,820 ± 2590d	15,870 ± 68b	26,450 ± 910b,d
SREBF1	8850 ± 2860a	4440 ± 270d	3350 ± 260a,d
Metabolism
LEP	110 ± 40b,d	1970 ± 470d	6100 ± 1200b
ADIPOQ (×103)	130 ± 30a,d	370 ± 30d	530 ± 80a
FABP4 (×103)	1450 ± 390	1730 ± 110	2090 ± 300
GPR120	10 ± 5b	60 ± 20a	560 ± 150a,b
INSR	5150 ± 1960	7240 ± 860	5290 ± 580

Values are mean copy number with their standard errors; n = 5 or 6 per group.

Significant differences with age indicated by similar superscripts:

^a

P < 0.01.

^b

P < 0.001.

^c

P < 0.01.

^d

P < 0.05.

^e

P < 0.05.

Differences in tissue temperature and metabolites in response to feeding and altered fat mass in adults

In adult sheep, the temperature of retroperitoneal fat was consistently higher than that of sternal fat and skeletal muscle (Fig. 3), which might reflect its deep-body location. Nevertheless, retroperitoneal fat and skeletal muscle displayed comparable temperature responses to feeding. The greatest feeding-induced temperature rise was in sternal fat (Fig. 4). There was no effect of increased fat mass on the temperature of the three tissues studied (Figs. 3 and 4). Although low body weight/adiposity was associated with reduced temperature of both adipose tissue and skeletal muscle, this effect was less pronounced in retroperitoneal fat (Fig. 3). Plasma glucose and insulin levels were lower in the lean animals, but there was little effect of altered adiposity on plasma NEFA and lactate levels (Fig. 5).

Figure 3.

Summary of the effect of altered body weight and fat mass on changes in the temperature of sternal and retroperitoneal (RP) adipose tissue and skeletal muscle as measured continuously over a 24-hour period. Black arrow indicates when food was first made available. Gray boxes indicate the time during which food was available. Obese animals are shown as white squares, controls as black squares, and lean animals as black triangles. Values are means with standard errors of the mean, and n = 4 or 5 per age group. **P < 0.01 lean compared with obese and control; ^aP < 0.05 lean compared with obese.

Figure 4.

Summary of the effect of altered body weight and fat mass on changes in postprandial rise in tissue temperature of sternal and RP adipose tissue and skeletal muscle as measured after feeding. This was calculated as the amplitude of the temperature change that occurred within the feeding window (1100 hours to 1600 hours). Values are means with standard errors of the mean, and n = 4 or 5 per group. Significant differences between depots for each body weight group of sheep are indicted by *P < 0.05; **P < 0.01 sternal fat compared with muscle and RP fat.

Figure 5.

(a–d) Summary of the effect of increased body weight and fat mass on changes in plasma metabolite and insulin levels with feeding (represented by shaded regions). Values are means with their standard errors, and n = 4 or 5 per group. Significant differences between depots for each body weight group of sheep are indicted by **P < 0.01; ***P < 0.001 lean compared with control and obese.

Differences in gene profile with altered fat mass in adults

The abundance of UCP1 mRNA was very low in all adult tissues, and UCP1 was consistently detectable by immunohistochemistry only in sternal fat (Fig. 6). UCP3 was highly abundant in skeletal muscle but was very low in sternal and retroperitoneal fat, being expressed 100-fold more in muscle than in fat (i.e., skeletal muscle 5 ± 1 au; adipose tissue 0.05 ± 0.1 au). Gene expression for UCP3 was increased in sternal adipose tissue in lean animals compared with obese animals (lean 18 ± 4 au; obese 0.3 ± 0.2 au; P < 0.05), whereas protein abundance was reduced (Table 4). In contrast, neither UCP1 mRNA nor protein were altered by changes in adiposity in either fat depot (data not shown). UCP2 mRNA was lower in sternal and retroperitoneal fat of lean animals than in control animals (i.e., retroperitoneal: lean 1.0 ± 0.3 au; control 3.3 ± 0.5 au; P < 0.05). There were also no effects of body weight on UCP3, RyR1, or SERCA2a in skeletal muscle. Expression of SERCA1 mRNA was lower in skeletal muscle of lean animals, but protein concentrations were again unaffected by fat mass. Finally, as expected, adipocyte cell size was influenced by altered body weight and adiposity. In retroperitoneal fat, adipocyte size changed in proportion to increased body weight. On the other hand, adipocyte size was decreased in the sternal fat of lean animals, but there was no effect of obesity on adipocyte size in this depot (Fig. 6).

Figure 6.

The effects of body weight on adipocyte size and histological appearance in sternal and RP fat of adult sheep. (a) Mean adipocyte cell size with sternal fat represented by black bars and RP fat by white bars. (b–g) Representative photomicrographs of hematoxylin and eosin–stained sections. (h) Example of UCP1 immunostaining from sternal adipose tissue. (i) Rat BAT was used as the positive control, and (j) staining specificity was determined with rat BAT in the absence of primary antibody. Scale bar represents 50 µm. Values are means with their standard errors, and n = 4 or 5 per group. -ve control indicates the negative control. Significant differences between depots for each body weight group of sheep are indicted by *P < 0.05 compared with sternal fat; ^aP < 0.01 compared with obese animals (within a fat depot).

Table 4.

Effects of Altered Adult Body Weight on Protein Abundance of UCP1 and UCP3 in Fat and Skeletal Muscle and of Potential Thermogenic Proteins (i.e., SERCA) in Muscle

	Lean	Control	Obese
Skeletal muscle
UCP3	0.63 ± 0.30	1.0 ± 0.26	1.00 ± 0.25
SERCA1	1.26 ± 0.16	1.0 ± 0.14	1.06 ± 0.17
SERCA2a	1.18 ± 0.11	1.0 ± 0.08	1.06 ± 0.10
Sternal adipose tissue
UCP1	1.22 ± 0.16	1.0 ± 0.18	0.99 ± 0.20
UCP3	0.68 ± 0.17a	1.0 ± 0.08	1.35 ± 0.10b
Retroperitoneal adipose tissue
UCP1	1.05 ± 0.16	1.0 ± 0.09	1.04 ± 0.10
UCP3	0.73 ± 0.08	1.0 ± 0.16	1.02 ± 0.10

All results expressed in arbitrary units relative to controls and corrected against the density of staining for total protein.

Abbreviation: SERCA, sarcoplasmic reticulum calcium-transporting ATPase.

Significant differences between body weight groups indicated by different superscript letters: ^a vs ^b, P < 0.01.

Discussion

We have shown that during postnatal development in sheep, the sternal and retroperitoneal fat depots exhibit contrasting gene expression profiles that could be indicative of divergent prenatal origins. These differences potentially contribute to the enhanced temperature responses seen in sternal fat compared with retroperitoneal fat and skeletal muscle following feeding in adulthood. Although the abundance of UCP1 within the sternal depot declines with increased fat mass during both postnatal and adult life, it does not appear to compromise the ability of this depot to increase its temperature in response to feeding in adulthood. Furthermore, adipocyte cell size in the sternal fat depot appears unresponsive to increased adiposity, suggesting that it serves a function other than storing surplus lipids during nutrient excess and obesity.

Divergent patterns of development between fat and muscle in early life

It is becoming apparent that identifying functional markers in BAT and/or beige adipocytes is a complex process that is influenced by the depot and whether in vivo or in vitro methodologies are used (17, 37, 38). As we have shown previously in ovine retroperitoneal fat, there are at least three distinct phases of postnatal development (1). The major functionally related changes are seen between 1 and 28 days of postnatal age, coincident with the loss of UCP1 and the transition of brown to white adipose tissue (1). In the current study, we confirmed that this critical stage of development extends to both sternal fat and muscle and is coincident with rapid growth and functional changes within each tissue (39, 40). These findings are in accord with those recently described within epicardial fat during development of humans who have undergone heart surgery (41). Consideration of each gene or group of genes examined, the accepted function of each, and the known developmental ontogeny in other species is presented in the following sections, thereby providing insights into the pronounced differences in the molecular signatures of fat and muscle.

The most notable characteristic of sternal fat compared with both retroperitoneal fat and muscle was the very high abundance of En-1 mRNA. Lineage-tracing studies in mice indicate that cells showing early expression of En-1 during development give rise to dermis and epaxial muscle, but not to other muscles, and interscapular BAT “bundles” (42), which is in accordance with our findings postnatally. Anatomical location during early development is determined along three axes: anterior-posterior, proximal-distal, and dorsal-ventral (43), but the primary regulators are not fully elucidated, with a variety of gene families such as Wnt (44, 45), HOX (46), and Pax playing roles. HOX genes are important regulators of development (47, 48), for which expression of specific combinations of the paralogous HOX gene sets specify a particular anatomical location along the anterior-posterior axis. Our data show that HOXC8 and HOXC9 are highly expressed in the more posteriorly located retroperitoneal depot but not in the more anterior sternal depot and that this pattern of gene expression persists to 28 days. In mice, HOXC8 and HOXC9 gene expression is higher in retroperitoneal than in interscapular adipose tissue (49).

TCF21 was originally proposed as a marker for white preadipocytes (50), but its gene expression differs between fat depots (51, 52). Therefore, our finding that TCF21 mRNA was abundant in only retroperitoneal fat supports the hypothesis that tissue-specific patterns of TCF21 gene expression are indicative of fundamental differences between depots that are dependent on anatomical location. Further indirect evidence of depot-specific rates of development comes from examining SHOX2. Its pattern of gene expression was opposite to that of the HOXC genes measured, being expressed in both muscle and sternal fat, which is anteriorly located to the retroperitoneal depot, where there is little if any detectable expression. SHOX2 is able to interact with CEBPα to modulate ADRB3 and, by extension, to regulate lipolysis in adipose tissue (53). In addition, ablation of SHOX2 promotes lipolysis in mice (53). The low expression of this gene within retroperitoneal adipose tissue could therefore be indicative of a more rapid mobilization of NEFAs as well as a capacity for greater growth through adulthood than in sternal fat. Differences in metabolic capacity with respect to medium-chain fatty acid synthesis between tissues could also explain the much higher mRNA abundance of ACSM5 in retroperitoneal adipose tissue than in sternal adipose tissue. ACSM5 is also considered a characteristic of white fat rather than BAT (34), and its paucity reflects the retention of UCP1 within sternal fat. Our finding of greater gene expression for ACSM5 within muscle contrasts with findings in adult rodents (54) and could be indicative of the significant changes seen within muscle during early postnatal development. In sheep, this is coincident with the recruitment of shivering thermogenesis (40), which could be accompanied by the increased utilization of intramuscular fat as suggested in pigs (55). Gene expression for FNDC5 also peaked at 7 days in muscle, coincident with maximal recruitment of shivering thermogenesis as UCP1 declined (40). There was no parallel change in plasma concentrations of irisin at this stage. This was not entirely unexpected given the current controversy regarding the measurement of irisin (56) and its potential functionality or existence (57).

Some changes in gene expression with age were similar in both muscle and fat depots (e.g.,PDK4), which may reflect the overall decline in basal metabolic rate (40), loss of UCP1, and pronounced fat deposition up to 1 month of age (39). At the same time, there is a transition from lipid to glucose metabolism (58) that would be facilitated by a decline in PDK4 activity (59), whereas increases in leptin, adiponectin, RIP140, and GPR120 gene expression are indicative of increased adiposity. In adults, however, fat mass and plasma adiponectin and its gene expression are normally negatively correlated (60). A different type of relationship during early life, coincident with the rapid growth of fat, is not unexpected, as has been seen for plasma leptin and the loss of its positive correlation with fat mass (61). At the same time, both LHX8 and ATF2 gene expression increased with age and fat mass, which was not expected given their putative “BAT identity marker roles,” as described by others in ovine retroperitoneal adipose tissue with development (62). As suggested by rodent studies, both ATF2 and GRP120 may therefore have greater roles in stimulating adipogenesis than thermogenesis (63, 64), whereas raised RIP140 would facilitate the loss of UCP1 (65). In summary, sternal fat development shares some characteristics with skeletal muscle that may also affect the retention of UCP1 and its thermogenic capacity in adulthood.

Functional consequences of UCP1 in sternal fat

The contribution of BAT to diet-induced thermogenesis in rodents remains contentious (66), although it does appear to have a role in young sheep (58), children (67), and adults (7). There is good evidence from developmental studies in both rodents and young sheep that muscle is recruited to generate heat when UCP1 is absent (68) and/or nonshivering thermogenesis is compromised (40). In adult sheep, temperature excursions in skeletal muscle in response to feeding and central infusion of leptin are consistent with increased thermogenesis (24, 29). The basal temperature of muscle also appears to be more sensitive to total fat mass than that of either fat depot studied. This could reflect its anatomical position and/or the impact of an increase in the surrounding fat and its insulating properties. It should also be noted that in large mammals, such as sheep, the thermogenic response to feeding is an entrained response (20), whereas in rodents, there is an influence of circadian rhythm (69). Furthermore, rodents are normally active in the dark phase, and the sensitivity of UCP1 to further stimulation is modulated by these diurnal activity patterns (70). These are dependent in part on both light exposure and activity of the sympathetic nervous system (71). A range of other factors may be critical in determining the thermogenic role of muscle in rodents, including sarcolipin (72) and UCP3 (73). In adult sheep, gene expression of UCP3 is increased in skeletal muscle after central infusion of leptin and is associated with increased heat production and a switch toward uncoupled respiration in isolated mitochondria (24). In addition, increased expression of RyR1 mRNA and SERCA2a protein in skeletal muscle coincides with dietary-induced thermogenesis (29). However, we found no difference in protein abundance for either SERCA1 or 2a or UCP3 in the muscle of animals of differing body weights.

In contrast to the acute effects of feeding, prolonged food restriction caused a marked decrease in temperature in skeletal muscle and sternal adipose tissue, an effect attenuated in the retroperitoneal fat. Consistent with decreased temperature in muscle, UCP3 gene expression declined markedly in the skeletal muscle of lean animals, but there was no associated change in protein. On the other hand, altered adiposity had no effect on UCP1 gene or protein abundance in sternal fat, but UCP3 mRNA was reduced in the lean group. Because of the much larger mass of muscle than BAT, its contribution to metabolic homeostasis is appreciably greater in sheep and humans (74), especially when plasma glucose concentrations are raised. Notably, glucose concentrations were considerably lower in the lean group in the current study. The reduced temperature in both skeletal muscle and sternal adipose tissue of lean animals is indicative of homeostatic reduction in thermogenesis in response to chronic food restriction and weight loss. This may be a mechanism to reduce energy expenditure in order to maintain body weight in states of negative energy balance and/or in the lean condition and may be mediated by lower thyroid hormone secretion (75).

In summary, sternal and retroperitoneal fat depots have distinct developmental profiles that are different from those seen in muscle. The different developmental profiles are associated not only with early adipose growth but also with thermogenesis in these tissues later in life. The extent to which sternal fat expansion and particularly UCP1 abundance can be modulated in early life may inform new strategies to manipulate energy balance, especially following feeding or in response to chronic food restriction during adulthood.

Abbreviations:

 
• au

  arbitrary unit

 
• BAT

  brown adipose tissue

 
• mRNA

  messenger RNA

 
• NEFA

  nonesterified fatty acid

 
• PCR

  polymerase chain reaction

 
• UCP1

  uncoupling protein.

Acknowledgments

The authors thank Dr. Robyn Murphy (La Trobe University), who gifted the SERCA antibodies. In addition, we thank Bruce Doughton, Lynda Morrish, and Elaine Chase for animal husbandry and assistance with animal work. We thank Professor Matthew Watt and Dr. Ruth Meex (Monash University) for advice and guidance with regard to Western blotting.

This work was supported by National Health and Medical Research Council Project Grant 1005935 (B.A.H).

Disclosure Summary: The authors have nothing to disclose

Author notes

Address all correspondence and requests for reprints to: Michael E. Symonds, PhD, Academic Division of Child Health Obstetrics & Gynaecology, School of Medicine, Queen’s Medical Centre, The University of Nottingham, Nottingham NG7 2UH, UK. E-mail: [email protected].