Abstract
Background
Obesity-associated dysglycemia is associated with metabolic disorders.
MicroRNAs (miRNAs) are known regulators of metabolic homeostasis. We
aimed to assess the relationship of circulating miRNAs with clinical
features in obese Qatari individuals.
Methods
We analyzed a dataset of 39 age-matched patients that includes 18
subjects with obesity only (OBO) and 21 subjects with obesity and
metabolic syndrome (OBM). We measured 754 well-characterized human
microRNAs (miRNAs) and identified differentially expressed miRNAs along
with their significant associations with clinical markers in these
patients.
Results
A total of 64 miRNAs were differentially expressed between
metabolically healthy obese (OBO) versus metabolically unhealthy obese
(OBM) patients. Thirteen out of 64 miRNAs significantly correlated with
at least one clinical trait of the metabolic syndrome. Six out of the
thirteen demonstrated significant association with HbA1c levels;
miR-331-3p, miR-452-3p, and miR-485-5p were over-expressed, whereas
miR-153-3p, miR-182-5p, and miR-433-3p were under-expressed in the OBM
patients with elevated HbA1c levels. We also identified, miR-106b-3p,
miR-652-3p, and miR-93-5p that showed a significant association with
creatinine; miR-130b-5p, miR-363-3p, and miR-636 were significantly
associated with cholesterol, whereas miR-130a-3p was significantly
associated with LDL. Additionally, miR-652-3p’s differential expression
correlated significantly with HDL and creatinine.
Conclusions
MicroRNAs associated with metabolic syndrome in obese subjects may have
a pathophysiologic role and can serve as markers for obese individuals
predisposed to various metabolic diseases like diabetes.
Keywords: miRNA, metabolic disorder, HbA1c, obesity, mirDIP, network
analysis
Introduction
The worldwide rise in obesity and its strong association with metabolic
diseases have elicited interest in the underlying mechanisms. According
to the WHO report 2021, worldwide obesity has nearly tripled since 1975
([43]1). In 2016, more than 1.9 billion adults, 18 years and older,
were overweight and over 650 million were obese ([44]1). The global
obesity epidemic is causing an alarming incidence of metabolic
disorders. Obesity can be considered a growing epidemic that is
associated with hyperglycemia (elevated blood glucose levels >7.0
mmol/L or hemoglobin that is glycosylated HbA1c > 6.5%), insulin
resistance, and dyslipidemia (characterized by elevated cholesterol,
low-density lipoproteins (LDL) and decreased serum high-density
lipoproteins (HDL)), collectively referred to as metabolic syndrome
([45]2). However, there are subjects with an elevated body mass index
(BMI) who do not progress to metabolic syndrome; they are generally
labeled as “Metabolically Healthy Obese” ([46]2–[47]7), they have
obesity only (OBO); but the protective mechanisms are unknown. Body fat
distribution is suspected to play an important role ([48]8). High liver
fat content and predominantly abdominal adiposity were shown to be
linked to the metabolically unhealthy obesity phenotype (obesity with
metabolic syndrome or OBM), whereas subcutaneous adiposity is
associated with the metabolic healthy obesity phenotype ([49]9,
[50]10). Over the past years, some biological mechanisms and phenotypic
characteristics have been identified that differentiate individuals
with OBO from OBM ([51]11). The concept of OBO may serve as a model to
better understand the pathways and mechanisms linking obesity to
metabolic diseases. Therefore, considering the potentially devastating
impact of obesity, there is urgency in elucidating underlying
mechanisms and identifying novel markers for risk stratification and
targeted early treatment.
Impaired adipose tissue metabolism and function are central to the
pathogenesis of obesity and associated metabolic disorders. MicroRNAs
(miRNAs) play a crucial role in regulating gene expression and are
likely to have an essential function in the pathogenesis of obesity and
metabolic disorders ([52]12). MicroRNAs are small non-coding RNAs
participating in the post-transcriptional regulation of genes by
negatively regulating them. Evidence is accumulating that circulating
miRNAs, released by many types of cells act as a new class of endocrine
factors. MiRNAs might serve as endocrine and paracrine messengers that
facilitate communication between donor cells and tissues with receptor
cells or target tissues, thereby potentially having important roles in
metabolic organ crosstalk ([53]13). In response to various
pathophysiological conditions, miRNAs can be released by cells into
their environment transported by different extracellular fluids,
including blood, and could serve as biomarkers of diverse diseases
including diabetes and related metabolic disorders. The role of miRNAs
as key regulators of metabolic homeostasis has been intensely explored
over the last decade. Brando et al. have collated the significant
circulating miRNAs that are altered in obese subjects, where microRNAs
such as miR-92a-3p, miR-122, miR-122-5p, miR-140-5p, miR-142-3p,
miR-151a, miR-155, miR-222, and miR-15a have been shown to be
upregulated. On the other hand miR-15a, miR-26a, miR-30b, miR-30c,
miR-125b, miR-126, miR-139-5p, miR-144-5p, miR-146a, miR-150, miR-223,
and miR-376a are reported to be downregulated in obese adults when
compared to healthy lean individuals ([54]14). Controversially, the
role of miR-15b remains obscure where it has been upregulated in one
study ([55]15) while another group reported downregulation of miR-15b
in obese subjects compared to lean counterparts ([56]16). Although
obesity is linked to differentially expressed miRNAs, they additionally
contribute to various metabolic disorders including hypertension,
hepatic steatosis, and insulin resistance by influencing the metabolism
of cholesterol, LDL (mir-26a and mir-15b), and elevation of circulating
glucose (mir-140-5p, miR-142-3p, miR-222, and mir125b) that eventually
glycosylates hemoglobin (HbA1c) respectively ([57]12, [58]14,
[59]17–[60]21).
Here, we aimed to unravel the associations between the metabolic
parameters in obese individuals with miRNA profiling. We identified 64
significantly differentially expressed miRNAs of which 36 were
down-regulated and 28 were up-regulated. By undertaking an association
discovery approach, we identified the expression of eleven out of the
36 down-regulated miRNAs and two of the 28 up-regulated miRNAs in our
patient dataset were significantly correlated with at least one
clinical trait of relevance to metabolic syndrome. The down regulated
miRNAs include miR-106b-3p, miR-103a-3p, miR-130b-5p, miR-153-3p,
miR-182-5p, miR-331-3p, miR-363-3p, miR-433-3p, miR-636, miR-652-3p and
miR-93-5p whereas miR-452-3p and miR-485-5p were upregulated. Several
of the miRNAs in OBM patients were significantly dysregulated and
associated with increased levels of HbA1c and cholesterol. These
include miR-130b-5p, miR-153-3p, miR-182-5p, miR-331-3p, miR-363-3p,
miR-433-3p, miR-452-3p, miR-485-5p and miR-636. [61]Figure 1 provides
an outline of our experimental design.
Figure 1.
[62]Figure 1
[63]Open in a new tab
The experimental study design. BMI, Body Mass Index; OBO, Obesity with
no metabolic disease; OBM, Obesity with metabolic diseases. To
determine the fold-changes of miRNA expression between the OBO vs OBM
patients, we used the Relative Quantification (RQ) measure. We
considered those miRNAs to be differentially expressed for which
|log2(RQ)| > 2 and significance threshold p< 0.05. This resulted in the
identification of 64 differentially expressed miRNAs. Out of the 64
miRNAs, there were 13 miRNAs whose expression correlated with at least
one clinical trait of relevance for metabolic syndrome (including
HBA1c, Creatinine, Cholesterol, LDL, and HDL).
Materials and methods
Study Design
The participants were recruited at the Qatar Metabolic Institute, Hamad
Medical Corporation, Doha, Qatar. The study protocol was approved by
the institutional review board (IRB) of Hamad Medical Corporation (HMC,
IRB protocol #16245/16) and all participants provided written informed
consent. Obesity was determined according to CDC guidelines. Both Class
1 (BMI of 35 to 40) and Class 2 (BMI > NA). Both Class 1 and Class 2
obesity were referred to as morbid obesity. A total of 120, male and
female participants aged between 18 to 65 years with morbid obesity
(BMI≥35 kg/m^2) were included. Individuals such as pregnant females and
those with identified chronic disease or terminal illness were excluded
from the study. The subjects were classified into two groups those
without metabolic syndrome (OBO) and with metabolic syndrome (OBM)
components of the metabolic syndrome; obesity PLUS any 2 of the
following: triglycerides ≥ 150 mg/dL (1.7 mmol/L), HDL< 40 mg/dL (1.03
mmol/L) in men or< 50 mg/dL (1.29 mmol/L) in women, blood pressure ≥
130/85 mmHg and fasting blood glucose ≥ 110 mg/dL (5.6 mmol/L)
([64]22). An additional, filter of age and BMI matching yielded 39
subjects that consisted of 18 OBO subjects and 21 OBM subjects. Among
the 18 OBO group subjects, none of the subjects had hyperglycemia, 2
individuals were identified with hypertension, 2 with mildly elevated
triglycerides, and 6 with a borderline decrease in HDL. Venous blood
samples were collected from these 39 subjects for total miRNA
isolation.
Participants Characteristics
Height and weight were measured in light clothing without shoes.
Fasting blood samples were taken between 7-9 AM after at least 12h of
fasting. For serum collection, whole blood was collected via BD
Vacutainer Serum Separation Tubes (BD Biosciences, Franklin Lakes, NJ,
USA). Blood samples were kept at room temperature for 30-60 minutes and
then centrifuges at 3000g for 10 minutes. Following centrifugation,
serum was separated and immediately stored at -80°C for further use.
Blood biochemistry was performed at the HMC clinical laboratory which
has been accredited by the College of American Pathologists (CAP).
Measurements included HbA1c with Turbidimetric Inhibition Immunoassay
(TINIA Roche Diagnostics, Mannheim, Germany), glucose by enzymatic
reference method with hexokinase (Cobas 6000, Roche Diagnostics
International, Switzerland), Total cholesterol, triglycerides, and
high-density lipoprotein (HDL) cholesterol levels were measured
enzymatically using a Synchron LX20 analyzer (Beckman-Coulter, High
Wycombe, UK).
RNA Isolation and Quality Control
Whole blood (2.5 ml) was collected into PaXgene Blood RNA Tubes
(PreAnalytix). The tubes were inverted 8-10 times then placed at room
temperature for at least 2 hours, frozen at -80°C, thawed overnight,
then total RNA was isolated with a PAXgene Blood RNA Kit including the
DNase Set (Qiagen). The concentrations and purity of the RNA samples
were evaluated spectrophotometrically (Nanodrop ND-1000, Thermo,
Wilmington, DE USA). The RNA isolation process was validated by
analyzing the integrity of several RNAs with the RNA 6000 Nano Chip Kit
(Agilent). The presence of the small RNA fraction was confirmed by the
Agilent Small RNA Kit (Agilent).
MicroRNA (miRNA) Profiling
The expression levels of 754 miRNAs were profiled using the TaqMan
OpenArray Human MicroRNA panels (PN: 4470189; Life Technologies Forster
City, CA, USA) on a QuantStudio 12K Flex instrument. For all
experimental groups, 3 µL (~10 ng) of total RNA was used for reverse
transcription (RT) reactions using MegaPlex RT Primers Human Pool Set
v3.0 (PN: 4444745; Pool A v2.1 and Pool B v3.0) according to the
manufacturer’s optimized protocol for low sample input for profiling
human microRNA using the OpenArray platform on BioRad c1000 Touch
thermal cycler. No-template controls were included. Pre-amplification
of RT products was performed using a 5 µL RT reaction combined with the
matching Megaplex PreAmp Primer Pool A v2.1 or B v3.0 and amplified
using the thermal cycler (Applied biosystems). The pre-amplified
products were diluted at 1:40 in 0.1x TE pH 8. For each experimental
set, 10 µL of the diluted products were combined to give a total of 40
µL pooled sample. For both Pool A and Pool B groups, 22.5 µL of the
pooled products were combined with an equivalent volume of TaqMan
OpenArray Real-Time Master Mix and aliquoted into a 96-well plate.
Then, 5 µL from each well were then transferred into a 384 well plate
for loading onto OpenArray plates using an AccuFill robotic system. The
OpenArray plates were run on a QuantStudio 12K Flex instrument (Life
Technologies) and the raw data files were imported and analyzed using
the DataAssist software (Life Technologies). Failed reactions were
excluded from analysis and undetermined CT values for samples sets
determined to have good amplifications were assigned a threshold value
of 40, defining low abundance or absence of miRNA expression. Global
mean normalization was used to calculate relative fold change for the
miRNA expression.
Statistics
Statistical characteristics of clinical measurements were calculated by
comparing the OBO and OBM samples using R v4.2.0 ([65]23). The
normality of the measurements was tested using Anderson-Darling test
using nortest v1.0.4 package ([66]24). The Student’s t-test was used to
calculate the p-value of the normally distributed measurements. For the
remaining measurements, Mann-Whitney test from the base package in R
was used. P-values were not corrected for false discovery rate (FDR)
owing to the small sample size.
The miRNA expression levels were measured via raw CR[T] values, which
are inversely proportional to miRNA expression i.e., the higher the
CR[T] value lower the expression of the circulating miRNA ([67]25).
However, current miRNA microarray platforms might not have enough
miRNAs which are stably expressed as indicated in ([68]26). Thus, to
measure fold-changes in miRNA expression, we determined the Relative
Quantification (RQ) values using the standard formula ([69]27). An RQ
value showcased the fold-change (FC) of a specific miRNA in two
populations. An RQ=1 indicated that a specific miRNA was not
differentially expressed in the OBO versus OBM samples. Otherwise, if
the |log2(RQ)| > 2 and significance threshold (p< 0.05), then the miRNA
was differentially expressed between the two groups as observed in
[70]Figure 2A .
Figure 2.
Figure 2
[71]Open in a new tab
(A) Volcano Plot highlighting the differentially expressed microRNAs.
The red-colored microRNAs are over-expressed in OBM versus OBO while
the blue-colored microRNAs are under-expressed. Here ‘RQ’ is equivalent
to the fold-change of a particular miRNA (Wang, Wang, and Xi 2011) and
is ∝ mean -ΔCR[T] values. (B) The mean -ΔCR[T] values for the
differentially expressed miRNAs for the OBM and OBO groups
respectively. The -ΔCR[T] values are ∝ to miRNA expression, where the
higher -ΔCR[T] value (or CR[T] value) corresponds to higher miRNA
expression levels. This is further reflected in the logRQ values which
are equivalent to fold-change in the expression of individual miRNA.
Here ‘logPval’ corresponds to -log10 (P-value).
Visualizations
The volcano and scatter plots were constructed using the ggplot2 v3.3.6
package in R. The visualization of the miRNA expression matrix was
performed using the ComplexHeatmap v2.12.0 package ([72]28) in R.
Correlation Analysis
We performed a set of correlation analyses, where we correlated the
expression (-CR[T] value when available) of each differentially
expressed miRNAs with the different clinical traits of relevance to
metabolic syndrome including HBA1c, Creatinine, Cholesterol, LDL, and
HDL. The correlations were estimated using the ‘cor.test’ function from
the stats package using the Pearson correlation method. The
correlations between differentially expressed miRNAs and clinical
traits were visualized using the corrplot v0.92 package
([73]https://github.com/taiyun/corrplot).
Additionally, we visualize the significantly correlated miRNAs’
expression versus individual clinical trait values for the OBO and OBM
patients through a scatter plot. We fit a linear regression line along
with confidence intervals using the ‘geom_smooth’ function and annotate
the Pearson correlation scores and p-values in the plot using the
‘stat_cor’ function from the ggplot2 package.
MiRNA-mRNA Interaction Network
We used the microRNA Data Integration Portal, mirDIP v4.1
([74]http://ophid.utoronto.ca/mirDIP/), which provides nearly 152
million human microRNA–target predictions collected from 30 different
resources ([75]21). The mirDIP integrative score was constructed by
taking a statistical consensus from the predictions available through
myriad resources and was assigned to each unique miRNA-target
interaction to provide a unified measure of confidence. The integrated
scores range, 0 to 1, was used; higher scores correspond to stronger
evidence of potential interaction between miRNA and target gene; the
target genes were thus identified.
Pathway Enrichment Analysis
The mRNAs which were identified to be regulated by the differentially
expressed miRNAs were then utilized in an overexpression analysis
framework. We used the ConsensusPathDB ([76]29) web portal
([77]http://cpdb.molgen.mpg.de/) as utilized in ([78]30–[79]34) to
identify significantly enriched pathways choosing the PID
([80]http://pid.nci.nih.gov/) and KEGG
([81]https://www.genome.jp/kegg/pathway.html) database. We also used
the ConsensusPathDB web-portal to determine the significantly enriched
GO terms. The significantly enriched pathways and GO terms were
determined using a hypergeometric test.
The hypergeometric test was performed as described below. Let the total
number of genes associated with our differentially expressed miRNAs be
n. Out of these n, say k genes are part of a pathway (p). This pathway
(p) consists of a total of K genes. The total number of background
genes (or all protein-coding genes in humans) be N. Then, the
probability of significance of the pathway can be determined by the
hypergeometric test as follows:
[MATH:
P(p)= (
Kn)
(N−Kn−k)<
mrow>(Nn) :MATH]
where
[MATH:
(Nn)
:MATH]
represents the combination function ([82]35).
Results
Clinical Characteristics
The clinical characteristics of the study subjects are summarized in
[83]Table 1 . The clinical traits that were significantly different
between the two groups include HbA1c (p=0.002), triglycerides
(p=0.001), high-density lipoprotein (HDL, p=0.008), glucose (p=0.009),
and insulin (p=0.05). Other important clinical traits which were not
significantly different between the two sets include clinical variables
such as creatinine, low-density lipoprotein (LDL), and cholesterol.
Table 1.
Clinical and biochemical traits of the study subjects.
Feature OBO OBM P Value
Age (years) 38.06 ± 4.21 40.52 ± 7.26 0.283
Females (N) 11 9
Males (N) 7 12
Height (cm) 167.4 ± 11.9 170.8 ± 9.6 0.370
Weight (kg) 113.4 ± 19.6 110.9 ± 27.6 0.782
BMI (kg/m^2) 40.0 ± 4.5 39.6 ± 3.0 0.746
Smoking (%) 6.0 33.0
HbA1c (%) 5.5 ± 0.27 7.02 ± 1.9 0.002
TG (mmol/L) 1.39 ± 0.48 2.65 ± 1.52 0.001
Cholesterol (mmol/L) 4.9 ± 1.1 4.8 ± 1.1 0.855
LDL (mmol/L) 2.8 ± 1.3 2.6 ± 1.1 0.728
HDL (mmol/L) 1.5 ± 0.7 1.0 ± 0.3 0.008
Glucose (mmol/L) 5.2 ± 0.6 7.4 ± 3.4 0.009
Creatinine (mmol/L) 67.5 ± 14.1 65.3 ± 14.1 0.563
Insulin (miU/mL) 19.0 ± 13.3 27.6 ± 13.2 0.053
CRP (mg/L) 12.8 ± 12.5 7.1 ± 4.5 0.064
ALT (U/L) 20.7 ± 11.6 36.5 ± 35.1 0.063
AST (U/L) 18.8 ± 9.6 23.6 ± 15.0 0.251
[84]Open in a new tab
OBO (obesity only), and OBM (obesity with metabolic syndrome).
Significance was determined by the Student’s t-test.
Differential Expression Analysis
We identified a total of 64 miRNAs to be differentially expressed
between the OBO and OBM groups ([85] Figures 2A, B and [86]Supplement
Table 1 ) of which 36 miRNAs were down-regulated and 28 were
up-regulated in the OBM patients when compared to the metabolically
healthy obese (OBO) patients ([87] Figure 2A ). Specific miRNAs;
miR-873-5p (-ΔCR[T] = 1.62), miR-9-3p (-ΔCR[T] = 1.91), mir-708-5p
(-ΔCR[T] = 1.96) were significantly up-regulated in OBM (had higher
mean -ΔCR[T]) in comparison to OBO patients ([88] Figure 2B and
[89]Supplementary Table 1 ). On the contrary, miRNAs; miR-100-3p
(-ΔCR[T] = -9.18), miR-486-5p (-ΔCR[T] = -5.16) and miR-92a-3p (-ΔCR[T]
= -4.76), were among the most significantly downregulated miRNAs in OBM
versus OBO patients.
Correlations With Metabolic Syndrome Relevant Clinical Markers
We next performed a set of correlation analyses, where we correlated
the expression (-CR[T] value when the measurement was available) of
each differentially expressed miRNAs with clinical lab traits of
relevance to metabolic syndrome including HbA1c, creatinine,
cholesterol, LDL, and HDL. 11 out of 36 down-regulated miRNAs and 2 out
of 28 up-regulated miRNAs correlated significantly (p< 0.05) with at
least one of the clinical lab traits ([90] Figure 3A ). The correlation
values across these 13 miRNAs and 5 clinical traits are summarized in
[91]Table 2 . As depicted in [92]Figure 3A the miRNAs miR-153-3p,
miR-182-5p, and miR-433-3p correlated negatively, while miR-331-3p,
miR-452-3p, and miR-485-5p demonstrated a positive correlation with
HbA1c. Interestingly, the trend for miRNAs: miR-153-3p, miR-182-5p, and
miR-433-3p, the -CR[T] values decreased linearly with higher
(dysregulated) levels of HbA1c ([93] Figure 3B ). This trend was
distinct for the OBM patients, suggesting the loss of expression of
these miRNAs in OBM patients was significantly related to increased (↑)
HbA1c levels. Similarly, from [94]Figure 3B for HbA1c, we could also
observe another trend for miRNAs: miR-331-3p, miR-452-3p, and
miR-485-5p. The -CR[T] values of these miRNAs went significantly up
i.e., these miRNAs were significantly over-expressed in OBM patients
with increased HbA1c levels.
Figure 3.
[95]Figure 3
[96]Open in a new tab
(A) Pearson correlation between clinical traits relevant to metabolic
syndrome and miRNA expression (-CR[T] values). The ‘x’ represents that
the correlation coefficient is not significant. The darker the
correlation coefficient (‘red’ or ‘blue’) the stronger the correlation
(more towards +1 or more towards -1). Significant correlations (p<
0.05) of clinical traits with relevance to metabolic syndrome with the
differentially expressed miRNAs. (B) Correlation with HBA1c; (C)
Correlation with Cholesterol; (D) Correlation with Creatinine; (E)
Correlation with HDL, and (F) Correlation with LDL.
Table 2.
Pearson correlation coefficients of the clinical traits associated with
metabolic syndrome with the miRNA expression of relevant differentially
expressed miRNAs.
Diff MiRNAs HBA1c Creatinine CHOLESTROL HDL LDL
miR-106b-3p 0.161 -0.436 -0.0776 -0.182 0.001
miR-130a-3p -0.0913 -0.161 -0.309 0.0669 -0.351
miR-130b-5p -0.111 -0.026 -0.388 -0.0154 -0.28
miR-153-3p -0.372 0.219 -0.115 0.168 -0.152
miR-182-5p -0.327 -0.0811 -0.225 0.0626 -0.229
miR-331-3p 0.484 -0.0371 0.0247 -0.197 0.0653
miR-363-3p 0.0575 -0.0793 -0.338 -0.064 -0.268
miR-433-3p -0.395 0.0775 -0.13 -0.0304 -0.113
miR-452-3p 0.456 -0.059 0.0267 -0.183 0.0224
miR-485-5p 0.555 0.0405 0.221 -0.0822 0.152
miR-636 -0.282 -0.103 -0.393 -0.0614 -0.337
miR-652-3p 0.0627 -0.383 -0.00376 -0.342 0.0976
miR-93-5p -0.0617 -0.437 -0.033 -0.231 0.0828
[97]Open in a new tab
The bold values represent strong correlations i.e. |correlation| > 0.3.
We further identified that miRNAs: miR-106-3p, miR-652-3p, and
miR-93-5p were significantly correlated with creatinine levels of
patients in our dataset, and miRNAs: miR-130b-5p, miR-363-3p, and
miR-636 were significantly associated with the cholesterol levels of
patients as observed in [98]Figure 3 . However, the ability to
distinguish OBM patients from OBO patients through the -CR[T] values of
these miRNAs was not as stark as of those miRNAs associated with HBA1C
(see [99]Figures 3D, C respectively). This can also be attributed to
the fact that creatinine and cholesterol levels were not significantly
different between the two groups as indicated in [100]Table 1 . We also
identified a miRNA, mir-652-3p, that was significantly negatively
correlated with LDL (R = -0.34, see [101]Figure 3E ). Interestingly,
the majority of OBM patients had lower LDL values as well as higher
expression of mir-652-3p, and the majority of OBO patients had higher
LDL values with lower expression of this miRNA. Lastly, we observe a
significant negative correlation between mir-130a-3p expression and the
clinical trait HDL (R = -0.35, see [102]Figure 3F ) with no clear
distinction between the OBM and OBO groups.
We performed a Student’s t-test to determine whether the expression
values of the 13 miRNAs of interest were significantly different
between the males (Gender = 0) and females (Gender = 1) or smokers
(Smoking 1 = Yes) versus non-smokers (Smoking 0 = No) in our dataset as
illustrated in [103]Figure 4 . From [104]Figure 4A , we observed that
miR-106b-3p and miR-652-3p had significantly different expressions in
males versus females, where both these miRNAs had lower expression in
males when compared to females. Hence the difference in the -CR[T]
values are positive (Δ-CR[T] > 0) as indicated in [105]Figure 4A .
However, for each of these miRNAs, there is no clear segregation of the
expression of the miRNA between the OBO versus OBM male patients
(Gender = 0) or female patients (Gender = 1) as observed in
[106]Figure 4B . This suggests that gender does not really have an
impact on the differential expression of these miRNAs (miR-106b-3p and
miR-652-3p) between the OBO and OBM patient groups.
Figure 4.
[107]Figure 4
[108]Open in a new tab
(A) Comparison of the expression pattern of the 13 differentially
expressed miRNAs for Gender and Smoking status of patients using a
Student’s t-test. Here ‘*’ represents a significant association (p<
0.05). (B) Boxplot illustrating the significant difference in
expression of miR-106b-3p and miR-652-3p between males and females. (C)
Boxplot highlighting the significant difference in expression of
miR-106-3p between patients who smoke versus those who don’t.
From [109]Figure 4C , we observe that miR-106b-3p has higher expression
in patients who don’t smoke (0 = No) when compared to patients who
smoke (1 = Yes) and the majority of the smokers (4 out of 6) belong to
the OBM category. While miR-106b-3p is differentially expressed w.r.t.
smoking status, there is no clear segregation of its expression between
OBO and OBM groups, for patients who don’t smoke. Moreover, owing to
the small sample size of patients with a positive smoking status (6
patients only), it is imperative not to draw strong conclusions.
However, for a larger population size smoking would be a covariate to
regress out when determining differentially expressed miRNAs for the
phenotype of interest (i.e. OBO vs OBM patients).
Mechanistic Insights from miRNA-mRNA Networks
We used the mirDIP database to extract information about target mRNAs
which can be regulated by the differentially expressed miRNAs with
significant associations with clinical traits. We use stringent cutoffs
including a minimum of 10 resources and an integrated score of at least
0.75 to retain a potential interaction between miRNA and the target
gene. This resulted in a total of 398 interactions between the seven
(out of the 13) differentially expressed miRNAs and 378 target genes.
Interestingly, we observed from [110]Figure 5 , that each of the seven
differentially expressed miRNAs forms its own cluster of target genes
with small overlaps amidst their interactomes. We then performed
downstream pathway enrichment using overexpression analysis through
ConsensusPathDB to identify significantly enriched pathways associated
with each of these miRNAs. We could determine the enriched pathways and
GO terms for four of these seven miRNAs. The top five significantly
enriched pathways and top three biological processes, cellular
components, and molecular functions for each of these miRNAs were
detailed in [111]Supplementary Tables 2 and [112]3 respectively.
Figure 5.
[113]Figure 5
[114]Open in a new tab
Top differentially expressed miRNAs with strong known interaction
(coming from >=10 resources and interaction score>=0.75 from mirDIP)
with target genes. Here we highlight only those miRNAs which are
significantly correlated with at least one clinical trait relevant to
metabolic syndrome.
For example, the miRNA, miR-153-3p, is differentially downregulated
with a ΔCʀтт of -3.4 (p=0.02, [115]Supplementary Table 1 ); the target
genes for this miRNA are SPHK2, GNAI3, ROCK1, and PLCB1; which are
essential for the Sphingolipid signaling pathway ([116] Supplementary
Table 2 ). The Sphingolipid signaling pathway has been shown to play an
important role in the regulation of obesity and type 2 diabetes
([117]36). Similarly, the PDGFR-beta signaling pathway was
significantly enriched based on the target genes of both miR-182-5p
(USP6NL; CTTN; RASA1; ACTR2; YWHAG) and miR-363-3p (MAP2K4; WASL;
ITGAV; RAP1B) respectively. The PDGFR-beta signaling pathway is known
to play a role in the regulation of adipose progenitor maintenance and
adipocyte-myofibroblast transitions ([118]37). We identified BMP
receptor signaling as one of the enriched pathways for miR-93-5p target
genes (BAMBI; RGMB; RGMA). It has previously ([119]38) been
demonstrated that BMP signaling was relevant for both the white and
brown adipogenesis and plays an important role when interconnecting
obesity with metabolic and cardiovascular diseases. Finally, we
identified the cellular senescence pathway from KEGG as a significantly
enriched pathway for miR-93-5p target genes (TGFBR2; E2F5; E2F1;
PPP3R1; CDKN1A; RBL2). Interestingly, recently Smith, Ulf et al.
([120]39) reviewed that white adipose tissue cells are highly
susceptible to becoming senescent both with aging, obesity and type 2
diabetes, independent of the chronological age. The white adipose
tissue senescence is associated with the inappropriate expansion of
adipocytes, insulin resistance, and dyslipidemia i.e., metabolic
syndrome, a finding in line with our phenotype.
Additionally, we identified several different significantly enriched GO
terms based on the target genes for each of the top four miRNAs (see
[121]Supplementary Table 3 ). These include GO terms associated with
biological processes such as positive regulation of metabolic process
(GO:0009893), cellular developmental process (GO:0048869), cellular
protein modification process (GO:0006464), mitotic cell cycle process
(GO:1903047) for miR-153-3p, miR-182-5p, miR-363-3p and miR-93-5p
respectively ([122] Supplementary Table 3 ).
Discussion
Individuals with a persistently high BMI are at the risk of developing
metabolic syndrome, a medical condition characterized by obesity,
insulin resistance, dyslipidemia, and hypertension, with an
accompanying risk of type 2 diabetes mellitus and cardiovascular
disease ([123]22). There is abundant literature that has investigated
the metabolic differences underpinning lean and obese subjects
([124]14). Obese individuals have been the focus of health care in
recent years since the reversal of obesity by lifestyle, medical or
surgical intervention protects them from metabolic syndrome ([125]40).
However, clinical observations identify a proportion of individuals
with elevated BMI who led an active and healthy life relatively free of
metabolic complications. This population is of particular interest and
intensely investigated to elucidate the underpinning mechanisms and
gene regulation that confer protection against the development of
metabolic syndrome. Moreover, differentiation between OBO and OBM as
well as early detection is paramount for clinical management of these
individuals.
Several studies have identified the essential role of differentially
expressed miRNA in obesity where a cluster of miRNAs; miR-92a-3p,
miR-122, miR-122-5p, miR-140-5p, miR-142-3p, miR-151a, miR-155,
miR-222, and miR-15a are upregulated and a group of miRNA; miR-15a,
miR-26a, miR-30b, miR-30c, miR-125b, miR-126, miR-139-5p, miR-144-5p,
miR-146a, miR-150, miR-223 and miR-376a are downregulated in obese
adults ([126]14). In this study, we focus particularly on morbidly
obese individuals with a BMI>35kg/m^2 who are metabolically protected
and susceptible. Our results demonstrate that miR-106b-3p, miR-130a-3p,
miR-130b-5p, miR-153-3p, miR-182-5p, miR-331-3p, miR-363-3p,
miR-433-3p, miR-636, miR-652-3p, and miR-93-5p were significantly
downregulated whereas miR-452-3p, and miR-485-5p were significantly
upregulated in morbidly obese patients with metabolic diseases compared
to obese patients without any metabolic disease in a dataset of Qatari
population ([127] Figures 2 and [128]3 ). These miRNAs significantly
correlated with at least one clinical trait of relevance to metabolic
syndrome like increased levels of HbA1c, creatinine, cholesterol, LDL,
and HDL in our dataset ([129] Figure 3A ). The differentially expressed
miRNAs correlate significantly with HbA1c (downregulated miR-153-3p,
miR-182-5p and miR-433-3p; upregulated miR-331-3p, miR-452-3p and
miR-485-5p), creatinine (downregulated miR-106b-3p, miR-652-3p and
miR-93-5p), cholesterol (downregulated miR-130b-5p, miR-363-3p and
miR-636), LDL (downregulated miR-130a-3p), and HDL (downregulated
miR-652-3p) in our dataset in the context of OBO and OBM.
Interestingly, we identify differential expression of miR-92a-3p,
miR-122-5p, miR-15a, miR-125b, and miR-146a in our data which have
previously been reported to be relevant for obesity ([130] Figure 2B
and [131]Supplementary Table 1 ) ([132]14).
Our unique dataset with the morbidly obese individuals (OBO and OBM)
highlights various differentially expressed miRNAs which have been
previously reported in obesity ([133]14) conferring confidence in our
study. Although identified in prior studies, these miRNAs (miR-92a-3p,
miR-122-5p, miR-15a, miR-125b, and miR-146a) do not associate
significantly with clinical lab traits, nor are they upregulated or
downregulated in alignment with previous studies. This can be a result
of our focus on metabolically healthy and unhealthy obese subjects
compared to prior investigations that analyze lean and obese groups.
Our findings further demonstrate that miR-153-3p, miR-182-5p, and
miR-433-3p are downregulated in the OBM group and negatively correlated
with HbA1c. Among these miRNAs, miR-153-3p has been reported to be
overexpressed in lupus nephritis patients ([134]28). Through miRNA-mRNA
network analysis, we have shown that miR-153-3p regulates sphingolipid
signaling. The sphingolipid pathway is known to be extensively involved
in obesity and obesity-induced hyperglycemia ([135]36). In agreement
with our results, miR-182-5p has been reported to be suppressed in
diabetes patients. Interestingly, it was shown that the expression of
miR-182-5p is high in newly diagnosed patients compared to healthy
control. However, its expression decreased with the increasing duration
of T2DM ([136]41). Another miRNA downregulated in the OBM group and
positively correlated with HbA1c is miR-433-3p, which has been reported
to be overexpressed in serum of hepatocellular carcinoma patients
([137]42), pediatric beta-thalassemia patients, and needs further
evaluation in the context of changes to hemoglobin ([138]43). Moreover,
miR-331-3p, which was downregulated in metabolically unhealthy obese
patients and positively correlated with HbA1c has been reported as a
biomarker for HCV-related hepatocellular carcinoma ([139]44), and
non-small cell lung cancer ([140]45). Among the upregulated miRNAs in
our results, miR-485-5p has been reported earlier to be associated with
atherosclerosis ([141]46), and lung and oral cancer ([142]47, [143]48),
whereas, miR-452-3p has been not reported earlier and might be a novel
biomarker. Overall, these differentially expressed miRNAs,
significantly correlated with HbA1c in obese patients with metabolic
diseases and seem to regulate the glycemic pathways. The mechanisms
behind the observed correlations of these miRNAs with HbA1c are still
unclear and need to be investigated further.
Another set of miRNAs: miR-106b-3p, miR-652-3p, and miR-93-5p, which
were downregulated in OBM subjects, negatively correlated with the
creatinine levels in these patients. It has been reported earlier that
elevated serum creatinine levels are associated with late stages of
diabetic nephropathy or renal damage ([144]49). The role of these
miRNAs is either a cause or consequence of renal damage or possible
existing hypertension in the OBM cohort. The miR-652-3p has been
reported to be relevant for insulin resistance ([145]50) and in
polycystic ovary syndrome (PCOS) patients, its expression has been
shown to be downregulated in the context of creatinine and HDL and is
most likely associated with hepatic involvement in cases of insulin
resistance ([146]51). miR-106b-3p has been earlier reported to be
downregulated in dengue infection ([147]52); its significance in
metabolic disorders is unknown. Given the significant association of
miR-106b-3p with both gender (higher expression in females in
comparison to males) and smoking status (higher expression in
non-smokers compared to smokers) of patients in our dataset, this miRNA
needs a more detailed mechanistic investigation as its significant
correlation with creatinine might be conditioned on the patient’s sex
and smoking status. Our results indicate that miR-652-3p is not only
negatively correlated with creatinine but also with high-density
lipopolysaccharides (HDL), indicating its possible role in dyslipidemia
in obese patients and warrants more investigation. The results from our
study indicate decreased expression of miR-130a-3p, miR-130b-5p,
miR-363-3p, miR-636, and miR-652-3p respectively in the OBM subjects.
Among these miRNAs, miR-363-3p, miR-130b-5p, and miR-636 correlated
with cholesterol, and miR-130a-3p correlated with LDL. It has been
reported previously that miR-130a-3p levels were elevated in the
pancreatic islets of hyperglycemic subjects ([148]53) as well as
progressive cardiac failure ([149]54).
In line with previous studies, we report a significant differential
expression of miRNAs that play critical roles in insulin resistance,
sensitivity, and release. For example, miR-122-5p, miR-221-3p,
miR-126-3p, miR-223-3p, and miR-93-5p, which are downregulated in OBM
versus OBO, have been described within the context of insulin
sensitivity and resistance ([150]55). In addition, miR-34b-3p,
miR-9-3p, miR-375, miR-146a-3p, and miR-30e-5p, which are upregulated
in OBM versus OBO have been involved in insulin release in pancreatic
β-cells and regulate β-cell fate ([151]56–[152]59). Interestingly,
IGF1R, a receptor tyrosine kinase that mediates actions of insulin-like
growth factor 1 and one of the factors that are altered in obesity is a
key target of differentially expressed miRNAs identified by our
framework including miR-182-5p. Another important set of targets for
the differentially expressed miRNAs were the MAPK genes (MAP3K2 and
MAP2K4) which belong to the family of mitogen-activated protein kinase
(MAPK). MAPK genes and their interactors have been reported to protect
against adverse effects of high-fat feeding in a murine model,
demonstrating a decreased weight gain, improved glucose tolerance, and
insulin sensitivity, with markedly diminished adipose tissue
inflammation ([153]60).
In conclusion, our data show that subjects with morbid obesity and
metabolic syndrome compared to individuals with obesity without
metabolic syndrome show differential levels of several miRNAs which can
regulate multiple genes and metabolic pathways relevant to glycemic
regulation, lipid metabolism, and cellular regeneration. However, the
cause or consequence merits further studies. The miRNA group associated
with metabolic syndrome in morbidly obese subjects may have a
pathophysiologic role that warrants further elucidation. Regardless of
their role in disease pathogenesis these groups of miRNAs can serve as
additional markers to segregate OBM and OBO that can aid divergent
management strategies of treatment. To the best of our knowledge, this
is the first study of its kind that addresses the role of miRNAs in
morbidly obese healthy versus obese metabolic syndrome adults for a
population indigenous to Qatar. We do acknowledge that our dataset is
small and further studies are warranted in additional larger cohorts to
corroborate the importance of the identified differentially expressed
miRNAs.
Data Availability Statement
All the data including processed microRNA and anonymized patient
profiles with clinical characteristics as well as the code required to
generate the results are available at:
[154]https://github.com/raghvendra5688/OBH_vs_OBO_MiRNA_HMC.
Ethics Statement
The studies involving human participants were reviewed and approved by
HMC, IRB protocol #16245/16. The patients/participants provided their
written informed consent to participate in this study.
Author Contributions
FM, RM, FF, and A-BA-S conceived the study. FM, AI, TS, FC, AP, MA, and
IA collected, purified, and harmonized the biological samples. RM and
EU built the computational pipeline and performed bioinformatics
analysis. FF and A-BA-S supervised the analysis. FM, RM, and EU wrote
the manuscript. All authors proofread the manuscript. All authors
contributed to the article and approved the submitted version.
Conflict of Interest
Authors FM, AI, TS, AP, MA, IA, and A-BA-S were employed by HMC.
The remaining authors declare that the research was conducted in the
absence of any commercial or financial relationships that could be
construed as a potential conflict of interest.
Publisher’s Note
All claims expressed in this article are solely those of the authors
and do not necessarily represent those of their affiliated
organizations, or those of the publisher, the editors and the
reviewers. Any product that may be evaluated in this article, or claim
that may be made by its manufacturer, is not guaranteed or endorsed by
the publisher.
Acknowledgments