Sex differences in miRNA expression and cardiometabolic risk factors in Hispanic adolescents with obesity

Abstract

Objective:

To evaluate sex differences in microRNA (miRNA) expression, anthropometric measures and cardiometabolic risk factors in Hispanic adolescents with obesity.

Methods:

Cross-sectional study of 68 (60% male) Hispanic adolescents with obesity, aged 13–17 years, recruited from a pediatric weight management clinic. We used small RNA sequencing to identify differentially expressed circulating miRNAs. We used Ingenuity Pathway Analysis and David bioinformatic resource tools to identify target genes for these miRNAs and enriched pathways. We used standard procedures to measure anthropometric and cardiometabolic factors.

Results:

We identified five miRNAs (miR-24–3p, miR-361–3p, miR-3605–5p, miR-486–5p and miR-199b-3p) that differed between females and males. miRNA targets-enriched pathways included PI3K-AKT, AMPK, insulin resistance, sphingolipid, TGF-beta, adipocyte lipolysis regulation and oxytocin signaling pathways. In addition, there were sex differences in blood pressure, skeletal muscle mass, lean body mass and percent body fat.

Conclusion:

We have identified sex differences in miRNA expression in Hispanic adolescents relevant to cardiometabolic health. Future studies should focus on sex-specific mechanistic roles of miRNAs on gene pathways associated with obesity pathophysiology to support development of precision cardiometabolic interventions.

INTRODUCTION

The prevalence of pediatric obesity and associated metabolic syndrome has steadily increased in the United States and worldwide over the past three decades, especially in adolescents. The trend in obesity prevalence has increased more dramatically in adolescents who are Hispanic compared to non-Hispanic ¹. Obesity is a risk factor for associated comorbidities, including glucose intolerance, insulin resistance, type 2 diabetes mellitus, non-alcoholic fatty liver disease, and hypertension, particularly in adolescents ^{2, 3}. Importantly, obesity during childhood may persist into adulthood increasing the risk of these comorbidities and overt cardiovascular disease in adulthood, compared to adults who did not have obesity ^{4, 5}. All of these considerations increase the burden on healthcare systems in the short and long term, necessitating a mechanism to identify at-risk individuals to inform a targeted approach for early intervention.

Individual characteristics such as age, race, ethnicity, socioeconomic factors, genetic and environmental factors, including lifestyle and dietary habits, may contribute to the development of obesity. In addition, several studies have shown that microRNAs (miRNAs) are associated with pediatric obesity ^{6, 7}. miRNAs are well known biomarkers and post-transcription regulators of gene expression. ^8–11. Importantly, sex differences have been associated with the development of obesity and associated comorbidities, especially in adults ^12–16. However, knowledge of sex differences in miRNAs expression relative to cardiometabolic risk factors in children is lacking, especially in the Hispanic population.

The goal of our cross-sectional study was to identify sex differences in miRNA expression and cardiometabolic risk factors in Hispanic adolescents with obesity in the Child Obesity Study at Children’s Hospital of San Antonio. We hypothesized that males and females will differ in their miRNA expression and cardiometabolic risk factors.

METHODS

The Baylor College of Medicine Institutional Review Board approved the study (IRB-H-40940). We obtained signed informed consent from the parents or guardians and assent from the participants. The work described herein has been carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki) for experiments involving humans.

This was a cross-sectional analysis of baseline data collected from participants enrolled in the Child Obesity Study. Baseline data were collected from the initial visit at a pediatric weight management clinic for Hispanic adolescents with obesity at Children’s Hospital of San Antonio. Criteria for inclusion in the study were age 13–17 years, self-reported Hispanic ethnicity and a body mass index ≥ 95^th percentile for age and sex. Exclusion criteria included a diagnosis of type 2 diabetes mellitus, use of neurohormonal medications, concomitant chronic or acute illnesses by self-report and unavailability of blood samples for miRNA quantification.

A trained medical assistant measured anthropometrics at the pediatric weight management clinic. Height was measured using a digital stadiometer (BSM170; InBody, Cerritos, CA). We used electric impedance (Scale 570; InBody, Cerritos, CA) to measure weight, percent body fat, skeletal muscle mass, body fat mass and lean body mass. We used a Dinamap oscillometer (GE Healthcare, Milwaukee, WI) to measure blood pressure (BP) with an appropriately sized cuff on the participant’s right upper extremity. If the initial measurement was high, the BP measurement was repeated twice manually and we recorded the average.

Blood samples (3 ml) were drawn in red-top tubes after fasting for 12 hrs. Samples were centrifuged at 2500 rpm for 20 min to obtain serum. We measured serum total cholesterol, fasting blood glucose, high-density lipoprotein cholesterol and triglycerides using standard enzymatic methods with a fully automated analyzer. We used Friewald’s equation ¹⁷ to measure low-density lipoprotein cholesterol and high-performance liquid chromatography to measure glycosylated hemoglobin A1C. Additional measurements included blood insulin, fasting blood glucose, aspartate aminotransferase, alanine amino transferase and gamma-glutamyl transferase.

Whole blood (1–3 ml) was collected in Tempus Blood RNA Tube (Thermo Fisher Scientific, Waltham, MA), containing 6 ml of RNA stabilizer according to manufacturer’s protocol. The blood sample was mixed with the stabilizer by agitating the collection tube 8 times. The samples were stored in −80°C freeze.

We used small RNA sequencing to identify all miRNAs expressed in the blood samples. Briefly, total RNA was isolated from whole blood using Direct-zol RNA Kit (Zymo Research, Irvine, CA) according to the manufacturer’s protocol with some modifications. An appropriate amount of phosphate-buffered saline was added to the sample for a final 1:1 ratio of sample and stabilization reagent. The diluted sample was vortexed vigorously at for 30 sec. We then followed the manufacturer’s protocol. RNA was quantified using Qubit Fluorimeter (DeNovix, Wilmington, DE), and the quality assessed using TapeStation RNA ScreenTape and HS reagents (Agilent, Santa Clara, CA). The RNA samples were stored at −80°C.

We used NextFlex Small RNA-Seq Kit v3 (PerkinElmer, Waltham, MA) and 500 ng of total RNA to generate cDNA libraries. Following library amplification and the post-PCR cleanup, we assessed the quality of the libraries using TapeStation DNA ScreenTape and HS D1000 reagents (Agilent), and the quantity using KAPA® Library Quantification Kit (Roche Diagnostics Corp., Indianapolis, IN), following the manufacturer’s protocols. After normalization, we generated library two pools, 10 nM/L each, for sequencing.

We used Illumina’s reagents and instruments, including HiSeq Rapid Duo cBot Sample Loading Kit for loading sample pools in the Rapid Flow Cell, with duo lanes, for template hybridization and first extension on the cBot 2. Subsequently, we used HiSeq Rapid Cluster v2 and HiSeq Rapid SBS Kit v2 (50 cycles) for cluster generation and sequencing on a HiSeq 2500 instrument. Sequence clusters containing base calls and quality scores were streamed into Illumina’s BaseSpace Sequence Hub, where they were demultiplexed and converted to fastq files using bcl2fastq2 software.

We used the miRDeep2 pipeline ¹⁸, utilizing default settings; and miRbase v22 to analyze the fastq-formatted sequence reads and to identify all (known and novel) expressed miRNAs and their read counts. Read counts were normalized holistically using reads per million mapped to miRNA.

We used Partek Genomic Suite (PGS; Partek, Inc.) together with Prism 8 (GraphPad) for statistical analysis. We used analysis of variance tool embedded in PGS to compare differences in variable distribution between males and females. We set our two-tailed alpha at <0.05. P values were corrected using false discovery rate (15%). To identify definitive outliers in the data, we used robust outlier tool in Prism 8.

We used miRNA Target Filter Tool implemented in Ingenuity Pathway Analysis v01–16 (Qiagen, Germantown, MD) to identify putative miRNA targets. Then, we used the output, including miRNA targets predicated with high confidence and experimentally validated targets, to identify enriched pathways using David Bioinformatic Resources 6.8 ¹⁹. P values were adjusted using Benjamini-Hochberg correction factor.

RESULTS

Sixty-eight participants (60% males) were included in the study. Mean age was 15 ±1.3 years with the mean body mass index percent of the 95^th percentile being 137% (Table 1). Males had significantly higher mean skeletal muscle mass, lean body mass, systolic BP and diastolic BP and lower percent body fat compared to females (Figure 1). There were no other between-sex differences in the other cardiometabolic risk factors, demographics and anthropometric measures.

Table 1:

Factors showing gender differences in adolescents with obesity

Factors	P-value	q-value*	Fold-Change (F vs. M)
hsa-miR-24–3p	0.01840	0.03600	−1.80
hsa-miR-486–5p	0.02370	0.04200	1.06
hsa-miR-199b-3p	0.02680	0.04800	1.65
hsa-miR-3605–5p	0.03820	0.05400	−1.61
hsa-miR-361–3p	0.04330	0.06000	−1.45
Age	0.65596	0.09141
SMM (Kg)	0.00002	0.00600	−1.25
LEAN MASS (Kg)	0.00002	0.01200	−1.23
Diastolic BP (mmHg)	0.00016	0.01800	−1.11
Percent BF	0.00018	0.02400	1.15
Systolic BP (mmHg)	0.00020	0.03000	−1.08
FBG	0.07093	0.06600	2.81
TG	0.19698	0.07200	−2.07
LDL-C	0.22645	0.07800	−2.05
TC	0.24412	0.08400	−1.65
VO2MAX	0.27275	0.09000	1.49
HDL	0.28104	0.09600	−3.29
Ins	0.35796	0.10200	−2.30
GGT	0.49709	0.10800	−1.82
CRP	0.60658	0.11400	1.50
nHDL	0.74726	0.12600	−1.18
A1C	0.74938	0.13200	1.52
ALT	0.75742	0.13800	1.27
AST	0.79870	0.14400	−1.28
BMI PERCENTILE	0.86143	0.15000	1.01

^*FDR=0.15

Figure 1:

Box plots showing the distribution and mean of A. SMM (skeletal muscle mass), B. lean body mass, C. systolic BP, D. diastolic BP, and E. percent body fat for female compared to male adolescents with obesity. The boxes represent interquartile range and whiskers represent minimum and maximum values. * represents p values < 0.05.

Using small RNA sequencing, we identified 203 miRNAs expressed in blood with a minimum of 250 read counts. There was disparity in expression of five miRNAs between males and females (Table 1). The expression of miR-24–3p, miR-361–3p and miR-3605–5p was downregulated in females and upregulated in males while miR-486–5p and miR-199b-3p was upregulated in females and downregulated in males. After p value correction, the expression of miR-24–3p, miR-486–5p and miR-199b-3p remained significantly different while miR-361–3p and miR-3605–5p were statistically not significance. According to miRbase version 22, miR-361–3p is transcribed on the X-chromosome, miR-486–5p and miR-24–3p on chromosome 8 and miR-199b-3p on chromosome 2.

We identified a number of putative targets of the miRNAs that were differentially expressed between females and males (Table 2). Interestingly, we observed that there were no experimentally validated targets for two of the miRNAs (miR-3605–5p and miR-361–3p). Table 3 (online only) shows the gene symbols of the miRNA targets.

Table 2:

miRNA targets

Candidate MicroRNAs	All targets (with low and high prediction confidence and with experimental evidence)	Targets with high prediction confidence and with experimental evidence	Targets with experimental evidence alone
miR-24–3p	1,118	104	22
miR-199b-3p	623	66	11
miR-486–5p	423	37	4
miR-3605–5p	866	46	0
miR-361–3p	1,471	158	0

Table 3.

(Online only): MicroRNAs predicted targets

hsa-miR-199b-3p	hsa-miR-24–3p	hsa-miR-3605–5p	hsa-miR-361–3p	hsa-miR-486–5p
ADAM22	ABCB9	ABLIM1	AATK	ARID4B
ADAMTS3	ACVR1B	ADORA1	ABI1	BCAS2
AK4	ADD2	ADRB3	ADGRG3	CD247
AKAP9	ADORA1	AK4	AJUBA	CDKN2B
BCAR3	AGPAT1	ALOXE3	AP2A1	CRLS1
CBLB	AGPAT3	AMHR2	APLN	CSPG5
CD2AP	ARHGEF15	ATP5F1C	APOL3	CSTF1
CD44	ATG4A	BHMT2	APOL5	CTSK
CDK17	ATP6V0E2	CACNG3	AQP7	DHFR2
CDK7	AURKB	CHMP1A	ARF3	DNAJC19
CELSR2	BAX	CHRM1	ARHGEF1	EHHADH
CFL2	BBC3	CTRC	ARPC5L	FGF7
CHKA	BCL2L11	CX3CL1	ASH2L	FOXO1
CHRM3	BRCA1	CXCL9	ATP2A3	GPR78
CNOT7	CALCR	CYGB	BAK1	GPX8
COL4A5	CCNA2	CYP19A1	CACNG6	KLK2
COPS2	CDK1	CYP3A4	CACNG8	LILRB4
CREB1	CDK4	DDT	CAMK2B	LMTK2
CREBZF	CDKN1B	DHRS3	CAMKK1	MDH1
CXCL11	CDKN2A	ECHS1	CAVIN1	MRAS
EIF3M	CHST4	EIF2AK1	CBX7	NEK2
FCGR3A/FCGR3B	CISH	ENTPD3	CCL18	PARP2
FGF7	COX6B2	FAS	CCN2	PDGFC
FN1	CRH	FFAR3	CD40	PIM1
GIP	CTSD	GJB3	CD79A	PLA1A
GNA12	CXCR2	GNG2	CDKN2D	PTEN
GPAT3	CYP46A1	GSTA4	CEBPA	PTGDR
GREM1	DHFR	HACD2	CELSR3	PYCR2
GRK3	DNAJB12	HCRTR1	CEMIP2	RCOR3
ITGB8	DNAJB2	HIBADH	CHD4	RFC2
ITPK1	DNAJC16	HMGA1	CHMP1A	SELENOT
KTN1	DOT1L	IGBP1	CLDN1	SLC38A1
LPAR4	DUSP1	KLK5	CXCR5	SNRPD1
LPAR5	E2F2	NDRG1	CYP19A1	TOB1
MAP3K4	EIF3I	NRG3	DDR1	TWF1
MET	ENTPD6	PARD6G	DLL3	VTI1A
MTOR	FASLG	PPM1A	DNAJB2	ZNF331
NLK	FBLIM1	PRKACG	DNAJB5
PAK4	FEN1	RHOA	DTX4
PAWR	FGF11	RPS16	DUSP15
PIK3CB	FMO1	SRR	DUSP2
PLEKHA3	FST	STXBP1	EEF1AKMT3
PNRC1	FURIN	TEAD1	EFNA4
PPP1R1C	GJD2	UCKL1	EFNA5
PPP2R2A	GNE	UGT2B11	ELK1
PPP2R5E	GPAT3	ZWILCH	ENO2
PRDX6	GSTM5		ENSA
PTGS2	H2AX		EPHA10
RNGTT	HIP1R		ETV4
RORB	HNF1B		EXTL3
RPS6KA6	HNRNPA1		FADS6
RUNX1	IFNG		FGF4
SCD	IL10RB		FKBP4
SDC2	IL1A		FOXH1
SERPINE2	INMT		FRAT1
SMAD1	IRF1		GAB2
SUZ12	LAMB3		GGCT
SYT16	LILRA6		GGT1
TFAM	LIPT2		GGT5
TMSB10/TMSB4X	LPAR6		GJC2
UBE2W	MAP2K4		GJD2
UBQLN1	MAPK14		GLI1
UPRT	MOGAT3		GNA12
VAMP3	MPI		GPR37L1
VCAN	MT-ND4L		GPR89A/GPR89B
YES1	MT1E		GUCA1A
	MT1M		H3–3A/H3–3B
	MYC		HAP1
	NDST1		HBB
	NEFM		HEXD
	NFKBIE		HOXA10
	NOTCH1		HOXB1
	PCYOX1		IHH
	PDGFRB		IL13
	PER2		IL1RL2
	PIM2		IL21R
	PMAIP1		IL31
	POLR3D		IL36RN
	PPARG		IP6K2
	PPCS		ISY1-RAB43
	PRKCH		ITGB6
	PRSS8		ITPR2
	PSMB8		IVD
	RAB4B		KCNJ4
	RAP1A		KCNJ6
	RAP1B		KHK
	RASD2		KLK2
	RNF2		KRT2
	RPL36		LCN6
	SCP2		LIMK2
	SIRPA		LPIN3
	SMAD3		MAFK
	SMAD4		MAP3K10
	SMAD5		MEF2D
	SPN		MEX3B
	SSR1		MKNK2
	TNFSF9		MMP24
	TPSAB1/TPSB2		MPRIP
	TPSD1		MRM2
	TRIB3		MTNR1B
	UBE2D4		NCOA2
	UCK1		NDST1
	USP18		NDUFS7
	WNT8B		NFIC
			NLRC5
			P2RX6
			P2RY2
			PADI1
			PALM2AKAP2
			PAX5
			PEBP1
			PHF1
			PIAS4
			PIM2
			PIP4K2C
			PLA2G2C
			PLK5
			PODXL
			POLR2D
			PPP2R5B
			PRSS33
			PSAP
			PSMD2
			PSPN
			PTAFR
			RAD9A
			RAP2B
			RELA
			RGS21
			RHO
			RORC
			RPS21
			RPS6KB2
			SDR9C7
			SH3GL3
			SNRPD3
			SOST
			SRF
			SSTR5
			SYMPK
			TFE3
			TFR2
			TFRC
			THRA
			TNF
			TNNC1
			TNNI1
			TOMM20
			TRAF3
			TSPAN1
			U2AF2
			UBASH3B
			UBE2V1
			VASP
			WASF2
			ZBP1
			ZBTB16
			ZFP36

For each miRNA, we identified pathways, annotated in Kyoto Encyclopedia of Genes and Genomes that were enriched in the list of miRNA targets that were predicted with high confidence and experimentally validated (Tables 4, 5 and 6; all online only). For miR-199b-3p, enriched pathways for phosphatidylinositol 3-kinase-protein kinase B (PI3K-AKT) signaling pathway and pathways associated with cancer had the greatest statistical significance and remained significant after p-value correction. Other enriched pathways for miR-199b-3p targets included the 5’ AMP-activated protein kinase (AMPK), insulin resistance and sphingolipid signaling pathways. For miR-24–3p targets, we observed the following enriched pathways: cell cycle, cancer, hepatitis B and TGF-beta signaling pathways. Enriched pathways for miR-3605–5p and miR-361–3p targets included adipocyte lipolysis regulation and oxytocin signaling pathways, respectively.

Table 4.

(online only): Enriched KEGG pathways associated with miR-199b-3p

miR-199b-3p
KEGG pathways	# Genes involved	Fold Enrichment	P-Value	Benjamini correction
PI3K-Akt signaling pathway	12	5.20	0.00001	0.00139
Pathways in cancer	12	4.57	0.00003	0.00236
Regulation of actin cytoskeleton	8	5.70	0.00038	0.01848
AMPK signaling pathway	6	7.29	0.00116	0.04209
Bacterial invasion of epithelial cells	5	9.59	0.00160	0.04635
Proteoglycans in cancer	7	5.23	0.00177	0.04281
Focal adhesion	6	4.36	0.01059	0.20159
Small cell lung cancer	4	7.04	0.01779	0.28258
ECM-receptor interaction	4	6.88	0.01892	0.26957
ErbB signaling pathway	4	6.88	0.01892	0.26957
Insulin resistance	4	5.54	0.03316	0.39289
Sphingolipid signaling pathway	4	4.98	0.04322	0.44816

Table 5.

(online only): Enriched KEGG pathways associated with miR-24–3p

miR-24–3p
KEGG pathway	# genes involved	Fold Enrichment	P-Value	Benjamini correction
Cell cycle	9	5.61	0.0002	0.0290
Pathways in cancer	15	2.95	0.0004	0.0351
Hepatitis B	9	4.80	0.0005	0.0281
TGF-beta signaling pathway	7	6.44	0.0007	0.0286
p53 signaling pathway	6	6.92	0.0016	0.0530
Chronic myeloid leukemia	6	6.44	0.0022	0.0607
Epstein-Barr virus infection	7	4.43	0.0046	0.1063
MAPK signaling pathway	10	3.06	0.0048	0.0988
HTLV-I infection	10	3.04	0.0050	0.0905
Osteoclast differentiation	7	4.13	0.0064	0.1052
Signaling pathways regulating pluripotency of stem cells	7	3.86	0.0088	0.1294
Pancreatic cancer	5	5.95	0.0094	0.1260
Melanoma	5	5.44	0.0127	0.1553
Bladder cancer	4	7.54	0.0152	0.1715
Neurotrophin signaling pathway	6	3.86	0.0184	0.1922
Small cell lung cancer	5	4.55	0.0231	0.2221
Measles	6	3.49	0.0274	0.2448
FoxO signaling pathway	6	3.46	0.0282	0.2388
PI3K-Akt signaling pathway	10	2.24	0.0315	0.2517
Glycerolipid metabolism	4	5.33	0.0377	0.2814
Chagas disease (American trypanosomiasis)	5	3.72	0.0438	0.3069
Colorectal cancer	4	4.99	0.0446	0.2998

Table 6.

(online only): Enriched KEGG pathways associated with miR-3605–5p, miR-361–3p and miR-486–5p

miR-3605–5p
KEGG pathway	# genes involved Count	Fold Enrichment	P-Value	Benjamini correction
Chemokine signaling pathway	5.00	5.78	0.0092	0.7048
Drug metabolism - other enzymes	3.00	14.02	0.0180	0.6982
Regulation of lipolysis in adipocytes	3.00	11.52	0.0260	0.6866
Steroid hormone biosynthesis	3.00	11.12	0.0278	0.6054
Retinol metabolism	3.00	10.08	0.0333	0.5913
Renin secretion	3.00	10.08	0.0333	0.5913
Drug metabolism - cytochrome P450	3.00	9.48	0.0372	0.5661
Metabolism of xenobiotics by cytochrome P450	3.00	8.71	0.0434	0.5671
miR-361–3p
KEGG pathway	# genes involved Count	Fold Enrichment	P-Value	Benjamini correction
Oxytocin signaling pathway	8	3.67	0.0057	0.6539
Fc gamma R-mediated phagocytosis	6	4.91	0.0070	0.4808
Acute myeloid leukemia	5	6.14	0.0084	0.4060
Transcriptional misregulation in cancer	8	3.30	0.0100	0.3747
MAPK signaling pathway	10	2.72	0.0103	0.3202
HIF-1 signaling pathway	6	4.30	0.0122	0.3155
Inflammatory bowel disease	5	5.37	0.0133	0.2987
Hypertrophic cardiomyopathy	5	4.41	0.0256	0.4527
Regulation of actin cytoskeleton	8	2.62	0.0312	0.4808
Dilated cardiomyopathy	5	4.09	0.0325	0.4586
NF-kappa B signaling pathway	5	3.95	0.0363	0.4644
miR-486–5p
KEGG pathway	# genes involved Count	Fold Enrichment	P-Value	Benjamini correction
Melanoma	3	10.77	0.0291	0.9138
Prostate cancer	3	8.69	0.0431	0.8394

DISCUSSION

We observed sex differences in miRNA expression pattern, cardiometabolic risk factors and anthropometric measures in a cohort of Hispanic adolescents with obesity. miRNA expression patterns can inform our knowledge of the diverse mechanisms involved in cellular metabolism and obesity pathophysiology as well as genetic factors that may contribute to obesity and related sex differences.

Our study identified five miRNAs (miR-199a-3p, miR-486–5p, miR-361–3p, miR-3605–5p and miR-24–3p) that were expressed differentially between Hispanic adolescent males and females with obesity. Previous studies have reported sex differences in miRNA gene regulation ²⁰. Other studies have suggested that miRNAs transcribed on the X-chromosome may escape inactivation leading to suppression of the genes involved in lipid metabolism ^{21, 22}. Moreover, studies have shown that the number of activated X-chromosomes in cells may explain mechanisms underlying sex differences in pathophysiologic processes ^{23, 24}. One of the five miRNAs identified in our study (miR-361–3p) is transcribed on the X-chromosome. Future studies will explore whether this miRNA escapes inactivation and the implications for obesity in adolescents. We conclude that sex differences in expression of miRNAs observed in our study may be one mechanism underlying sex differences in cardiometabolic disorders in adolescents with obesity.

We have identified genes predicted to be targeted by the miRNAs exhibiting sex differences. Many of these predicted miRNA targets have been validated experimentally, according to miRTarBase, TarBase and miRecords databases. However, we did not find statistically significant validated targets for miR-3605–5p, miR-361–3p and miR-486–5p. Future studies will focus on validation of these targets and their mechanistic effect on sex differences in obesity. This is important because miRNAs fine-tune gene expression, and understanding their effects on cell physiology may have clinical implications that can uncover novel therapeutic targets for prevention and treatment in individuals at high risk of obesity and its cardiometabolic complications.

We identified several intriguing miRNA targets-enriched pathways relevant to obesity and development of hypertension, diabetes and cardiovascular disease, including PI3K-AKT and AMPK pathways. The PI3K-AKT pathway is enriched in miR-199b-3p targets and is involved in promotion of cell proliferation, cell survival, growth and angiogenesis in response to extracellular stimuli such as insulin. Interestingly, there is a large body of evidence linking miR-199b-3p with obesity and associated comorbidities ^25–28. This miRNA was upregulated in fat-exposed hepatocytes derived from human fetal brains in females compared to males. In addition, the PI3K-AKT pathway was found to be enriched in targets of differentially expressed miRNAs ²⁵. A previous study revealed that miR-26b modulates PI3K-AKT pathway by targeting the phosphatase and tension homolog gene to promote glucose uptake by adipocytes and insulin sensitivity ²⁹.

Enriched pathways associated with miR-361–3p included the oxytocin pathway. Previous studies have reported that plasma oxytocin concentration is associated with obesity and diabetes ^{30, 31}, and animal and cellular studies have unraveled the protective effect of oxytocin on metabolic outcomes ³². However, estrogen influences the effects of oxytocin-mediated reduction in chow diet intake to the extent that the effect is more pronounced in male than female rats ³³.

Previous studies in animal models have investigated the mechanisms underlying the influence of sex differences on body fat. It is well documented that sex hormones and their receptors, including estrogen progesterone and androgens have profound effects on adipose tissue and exhibit sex differences ¹³. A previous study described the mechanistic role of miR-22 in modulating sex-specific lipid metabolism and body fat ³⁴. MiR-22 suppressed expression of estrogen receptor alpha leading to decreased lipid metabolism and fatty acid oxidation and ultimately visceral white fat accumulation in male mice. In females, estrogen receptor alpha promotes self-activation by binding to the miR-22 precursor to inhibit mature miRNA processing. This observation indicates that miRNAs have the ability to modulate sex-specific pathophysiology processes.

There are limitations of the present study that should be considered. First, we did not investigate the association of sex hormones with the observed sex differences, in order to confirm their role in this relationship. Second, our findings are based only on Hispanic adolescents with obesity and we did not recruit adolescents without obesity, younger children, or children from other racial and ethnic groups for comparison, which limits our finding’s generalizability. Third, our sample size is small and may have limited our power to detect differences in other traits. Even though multiple testing correction was performed, type-I error is plausible. Finally, we did not fully account for several sources of bias, including confounding bias. There is a need to validate our findings in a longitudinal study of a larger multiethnic cohort of adolescents with obesity with a comparator group without obesity and using analytic methods to account for bias. In conclusion we identified sex differences in miRNA expression that are relevant to key cardiometabolic pathways in this study of Hispanic adolescents with obesity.