Extracellular vesicle-enriched miRNA profiles across pregnancy in the MADRES cohort

About the Authors:

Helen Bermudez Foley

Roles Data curation, Formal analysis, Investigation, Methodology, Project administration, Writing – original draft, Writing – review & editing

* E-mail: [email protected]

Affiliation: Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California, United States of America

ORCID logo https://orcid.org/0000-0002-1314-6813

Caitlin G. Howe

Roles Formal analysis, Writing – review & editing

Current address: Department of Epidemiology, Geisel School of Medicine at Dartmouth, Lebanon, New Hampshire, United States of America

Affiliation: Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California, United States of America

ORCID logo https://orcid.org/0000-0002-4935-9298

Sandrah P. Eckel

Roles Formal analysis, Methodology, Writing – review & editing

Affiliation: Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California, United States of America

Thomas Chavez

Roles Data curation, Formal analysis

Affiliation: Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California, United States of America

Lili Gevorkian

Roles Investigation

Affiliation: Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California, United States of America

ORCID logo https://orcid.org/0000-0002-7588-6219

Eileen Granada Reyes

Roles Investigation

Affiliation: Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California, United States of America

ORCID logo https://orcid.org/0000-0002-4034-3767

Bethany Kapanke

Roles Investigation

Affiliation: Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California, United States of America

ORCID logo https://orcid.org/0000-0003-3340-9999

Danilo Martinez

Roles Investigation

Affiliation: Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California, United States of America

ORCID logo https://orcid.org/0000-0002-0543-8517

Shanyan Xue

Roles Investigation

Affiliation: Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California, United States of America

ORCID logo https://orcid.org/0000-0001-8318-1967

Shakira F. Suglia

Roles Formal analysis, Writing – review & editing

Affiliation: Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, Georgia, United States of America

ORCID logo https://orcid.org/0000-0002-5634-9290

Theresa M. Bastain

Roles Funding acquisition, Supervision, Writing – review & editing

Affiliation: Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California, United States of America

Carmen Marsit

Roles Supervision, Writing – review & editing

Affiliation: Gangarosa Department of Environmental Health, Rollins School of Public Health, Emory University, Atlanta, Georgia, United States of America

Carrie V. Breton

Roles Conceptualization, Funding acquisition, Supervision, Writing – review & editing

Affiliation: Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California, United States of America

ORCID logo https://orcid.org/0000-0002-6082-4066

Introduction

microRNA (miRNA) are small non-coding RNA (~22nt) which modulate gene expression [1, 2] and may be dysregulated in multiple diseases and conditions, including cancer, diabetes, and Alzheimer’s disease, as well as in normal development [3–5]. They are also sensitive to the environment, with altered profiles reported for a variety of different biofluids in response to toxicant exposures, including metals, bisphenol-A, and formaldehyde [6–8].

There is growing interest in the use of miRNA found in maternal plasma as markers of maternal and fetal health. Some miRNA likely have important roles in healthy pregnancies and normal fetal development, but most studies have focused on miRNA associated with gestational pathologies, such as preeclampsia, preterm birth, and infant outcomes after fetal alcohol exposure [9–11]. Inflammation and body mass index (BMI) are also associated with miRNA profile changes during pregnancy, and these miRNA profiles may be useful biomarkers for certain pregnancy complications [12, 13]. The shifting profiles of miRNA in maternal circulation likely reflect changes in miRNA released by the placenta [14]. For example, oncogenic and anti-apoptotic miRNA targets have been reported to dominate the first trimester, while cell differentiation targets were found to be prevalent in the third trimester [15].

Although miRNA are abundant intracellularly, miRNA are also present extracellularly, circulating encapsulated in exosomes (~50-150nm) and other extracellular vesicles (EVs, up to 1 μm), along with proteins and other biological molecules [16, 17]. During pregnancy, about 15–25% of circulating EVs are estimated to originate in placental tissues [13]. These EVs increase in maternal circulation throughout the first trimester [18] and have been hypothesized to play important roles in maternal-fetal communication [19]. However, the mediating “cargo” in these EVs have not been sufficiently investigated. miRNA are likely candidates for some of the effects of EVs: for example, exosomal miRNA from the placenta are capable of modulating gene expression in maternal target cells [20], are hypothesized to contribute to some immunosuppressive functions during pregnancy [21], and are known to assist in transfer of gene-modulatory information (miRNA and mRNA) between cells [16]. If these processes are altered by changes in miRNA expression, critical parts of a healthy pregnancy could be compromised. Indeed, altered circulating miRNA profiles have been associated with several pregnancy complications, including preeclampsia, gestational diabetes, and preterm birth [22–24]. Evidence for the roles of exosomal miRNA during pregnancy, including evidence for miRNA effects in mouse models, has been reviewed by Yang et al., suggesting that miRNA is not simply a biomarker, but is also an active participant with many regulatory roles [25].

Circulating miRNA profiles are also likely dynamic over the course of a normal pregnancy, due to the changing stages of placental development and pregnancy progression [15]. However, a limited number of studies have used repeated measures to investigate circulating EV-encapsulated miRNA as a potential mediator of the role exosomes and EVs play in normal pregnancies [19]. Previous research has established a global increase of circulating exosomes miRNA during pregnancy [18], but did not describe miRNA-level changes over time. Few studies have characterized maternal circulating EV-associated miRNA longitudinally in pregnancy and have not established miRNA profiles in pregnancy for health disparities groups. Therefore, in the current study we characterized changes in EV-associated miRNA across pregnancy in the MADRES cohort, a primarily low-income Hispanic population of women in Los Angeles, California [26].

Results

The majority of pregnant women included in the current study were in their mid-twenties, and about a third were primigravida (36%). (Table 1) Approximately 33% of participants were overweight and 35% were obese. Most participants’ preferred language was English (62%), and the majority of participants self-reported as Hispanic (75%). Among the self-identified Hispanic participants identifying country of birth, 55% were born abroad and 45% were born in the U.S. Maternal education was also an important characteristic of this population, with a large percentage completing no more than a high school education (58%).

[Figure omitted. See PDF.]

Table 1. Demographics of the miRNA pregnancy cohort.

https://doi.org/10.1371/journal.pone.0251259.t001

Covariate-adjusted linear regression models using all samples (n = 449 observations) identified 90 miRNA that were associated with gestational age (GA) after multiple testing correction (FDR P<0.05, Fig 1, S1 Table in S1 File). Samples with missing covariate data were excluded using the lme algorithm. GA was associated with an increase in miRNA counts during pregnancy for 64 of the miRNAs and with a decrease in miRNA counts for 26 of the miRNA, although individual participants may have had different trends (S5 Fig in S1 File).

[Figure omitted. See PDF.]

Fig 1. Volcano plot of miRNA associated with GA, scaled to 18 weeks of pregnancy.

Red vertical dotted lines indicate -10% and 10% change over 18 weeks. Red and purple horizontal dotted lines indicate FDR-adjusted and Bonferroni-adjusted significance thresholds, respectively.

https://doi.org/10.1371/journal.pone.0251259.g001

The same 90 miRNA were associated with GA (FDR P<0.05) when restricting the analysis to participants who provided plasma samples at both timepoints (n = 296 observations), and all 90 changed in the same direction in both analyses (S1 Fig and S2 Table in S1 File). An additional 14 miRNA were also found to be associated with GA that were not identified in the primary analysis (S1 Fig and S2 Table in S1 File). A second sensitivity study was performed excluding pre-term births (<37 weeks, n = 399 observations). Eighty-six of the 90 miRNA identified in the main analysis remained significant, including the top miRNA, and just 4 miRNA lost significance (hsa-miR-1197, hsa-miR-151a-3p, hsa-miR-26a-5p, hsa-miR-369-3p). Eight miRNA were associated with GA in the sensitivity study, but not in the main analysis (S4 Table in S1 File).

Differences by fetal sex were also explored. When analyses were stratified by fetal sex, 48 miRNA were associated with GA in females, and 96 miRNA were associated with GA in males (FDR P<0.05) (S3 Table in S1 File). Of those, 33 miRNA were shared between pregnancies with male and female fetuses. However, none of the interaction terms for GA and fetal sex were statistically significant (FDR P>0.05).

To investigate possible gene pathways related to the GA-associated miRNA, potential target genes were first identified using mirDIP using the 90 miRNA that were associated with GA in the primary analysis. Overrepresented and underrepresented PANTHER pathways and Gene Ontology categories (GOslim Biological Process and Molecular Function) are listed in Table 2. The top underrepresented categories in GOslim Biological Process were enriched in immune response categories. The PANTHER pathway analysis indicated enrichment of genes involved in signaling pathways, including Wnt, Ras, and EGF pathways. Large numbers of genes were associated with certain GO categories and pathways, especially those in metabolic processes, catalytic activity, binding and signaling pathways (Fig 2).

[Figure omitted. See PDF.]

Fig 2. Distribution of genes in Gene Ontology categories.

A. Number of genes in GO Biological Categories. B. Number of genes in GO Molecular Function. C. Number of Genes in PANTHER Pathways.

https://doi.org/10.1371/journal.pone.0251259.g002

[Figure omitted. See PDF.]

Table 2. Gene Ontology for predicted targets of miRNA significantly associated with GA.

The first ten results for gene ontology (GOSlims Biological Process, GOSlims Molecular Function, PANTHER pathway classification) tests for significantly over- and under-represented groups (PANTHER overrepresentation test, FDR P<0.05) are presented in Table 2. No named PANTHER pathways were significantly underrepresented (FDR P>0.05).

https://doi.org/10.1371/journal.pone.0251259.t002

Discussion

The data presented in this paper include a large number of participants with repeated measures during pregnancy, and focused on a predominantly lower-income population, many of whom have Hispanic heritage, and have historically been underrepresented in miRNA research. Thus, this study contributes a great deal to the diversity of currently available data for miRNA in pregnancy. A wide-range of pre-pregnancy BMIs, maternal education levels, and different racial groups were also represented in this study population. We identified 90 miRNA that were associated with GA, with the majority increasing over the course of pregnancy. Pathway analyses suggest that the gene targets of these miRNA are enriched in categories describing RNA and DNA regulation and are underrepresented in categories relating to immune response.

miRNA detection

Of nearly 800 miRNA assayed, 130 miRNA were detectable above the sample-specific background in 50% of maternal plasma samples. This is the largest characterization of miRNA from EV-enriched samples during pregnancy. Interestingly, the majority of miRNA detected were also significantly associated with GA at the time of sample collection (90/130), indicating that many miRNA may play some role during pregnancy, although few of them have been previously assessed in studies focusing on pregnancy complications.

miRNA associated with GA

Several studies have reported associations between miRNA, including EV-associated miRNA, and pregnancy related outcomes or conditions. However, much less is known about the role of EV-associated miRNA in normal development during pregnancy. Pregnancy is a complex and dynamic period. Maternal physiology must undergo substantial changes to meet the needs of the developing fetus, and it has been hypothesized that miRNA may contribute to coordinating these changes, in part by facilitating maternal-fetal communication during pregnancy [27]. miRNA profiles are therefore likely to differ between early and late pregnancy, and prior evidence suggests that oncogenic and anti-apoptotic miRNA targets may dominate the first trimester, while cell differentiation targets may be more prevalent in the third trimester [15].

In the current study, we identified a set of 90 EV-associated miRNA which were significantly associated with GA after adjustment for covariates, suggesting that these miRNA may change during pregnancy. In sensitivity analyses, we restricted participants with paired early and late pregnancy EV miRNA measures, and results were largely consistent with the main analysis.

While the majority of these miRNA (64) were positively associated with GA, which suggests that they increase over the course of pregnancy, we also identified a set of 26 miRNA that were negatively associated with GA. These associations may reflect some of the shifting responses and physiologic changes that occur during pregnancy. miRNA that decreased over the course of pregnancy may play a larger role at the beginning of pregnancy, while those that increase over the course of pregnancy may be important for supporting processes that occur in late pregnancy.

For example, GA was associated with lower levels of both miR-126-3p (raw P = 1.27E-09, FDR P = 1.65E-07) and miR-191-5p (raw P = 7.60E-06, FDR P = 9.87E-05). These two miRNA are known to regulate angiogenesis and may inhibit the development of new blood vessels or contribute to vascular remodeling critical to early placental growth [28–30]. Reduced maternal levels of miR-126-3p are hypothesized to contribute to vascular inflammation during pregnancy and have been reported at low levels in women with preeclampsia [31]. The miRNA analysis presented here did not exclude participants with pregnancy-induced hypertension or preeclampsia, however our results suggest that even in largely normal pregnancies, miRNA plays a role in vascular remodeling in early pregnancy and may also decrease in later pregnancy.

GA was also associated with lower miR-181a-5p counts (raw P = 7.01E-08, FDR P = 4.55E-06), and previous research suggests that this miRNA may be involved in the suppression of labor until later stages of pregnancy [32]. Decreasing levels of miR-181a, as observed in this study, are consistent with its role in supporting the beginning of labor. Low levels of miR-181a have also been associated with increased insulin resistance in non-pregnant obese subjects [33]. Reductions in this miRNA across gestation may therefore contribute to the increases in insulin resistance observed during pregnancy, in a manner similar to obesity [34]. Interestingly, miR-181a is also significantly decreased in newborns born to obese mothers [35], suggesting the mother’s metabolic status may influence the metabolic status of the newborn. The MADRES cohort has a high prevalence of pre-pregnancy obesity: 24% meet BMI criteria for overweight and 25% of mothers meet BMI criteria for obesity. Future research on metabolic health of expectant mothers and their children in the MADRES study is therefore critical.

Sixty-four of the 90 miRNA significantly associated with GA were positively associated with GA, and 34 were also found to be significant using Bonferroni correction. These included miR-1323 (raw P = 3.28E-06, FDR P = 6.93E-05), a placenta-specific miRNA, and miR210-3p (raw P = 1.08E-05, FDR P = 1.24E-04), which may play important roles in maintaining a healthy fetal weight and supporting placental function during pregnancy [36, 37]. Maternal plasma levels of miR-1323 were found to be lower in cases of fetal growth restriction, however the target pathways are unknown [36], and miR-210-3p expression in plasma was found to be lower for women with preeclampsia, acting through target genes that regulate trophoblast invasion and migration [37].

Interestingly, few of the miRNA from the chromosome 19 miRNA cluster (C19MC) and the chromosome 14 miRNA cluster (C14MC) that are particularly expressed during pregnancy were identified as significantly associated with GA [38, 39]. Several of these have been associated with pre-eclampsia (miR-518c, miR 525-5p, miR-526a, miR-299, miR-495 [40–42]), fetal growth restriction (miR 525-5p, [43]) and gestational diabetes (miR-518d, miR-1323 [44, 45]). Most of these have strong expression in the placenta, but have also been identified in circulating maternal plasma and in exosomes[14, 43, 45–48].

Other miRNA identified in this study have also been previously examined in maternal plasma and placenta. miR15a-5p is upregulated in exosomes isolated from maternal plasma in women with preeclampsia [49, 50] and miR 199a-5p in placenta is downregulated in cases of severe fetal growth restriction [51]. However, miRNA have multifaceted roles. Plasma and serum levels of miR127-3p increased in women with recurrent miscarriage [52], but increased levels during second trimester were associated with lower odds of a small for GA infant [53]. Although these roles appear contradictory, they may instead indicate that miRNA actions differ by situation, underscoring that studies addressing dynamics of miRNA profiles are critical for understanding the complex roles played by miRNA.

Sex-specific associations for miRNA and GA

Previous research has largely overlooked differences in miRNA by fetal sex during pregnancy [11, 14, 15, 35–37, 54, 55]. In sex-stratified analyses, we found some evidence for possible differences by fetal sex for a subset of the miRNA. For example, 15 miRNA were associated with GA in males but not females, and another 15 miRNA were associated with GA in females but not males, while 33 were found to be similarly associated with GA in both males and females. However, cross-product interaction terms for GA and fetal sex were not statistically significant. Further research on interactions between fetal sex and miRNA during pregnancy is therefore critical for a more complete understanding of potential sex differences in miRNA changes during normal pregnancies.

Gene target analysis

A single miRNA can affect the transcription of multiple gene targets. In fact, more than 30% of human genes are estimated to be regulated by miRNA [56]. Thus, even small changes in miRNA may have large impacts on genes involved in a variety of different pathways. EV-associated miRNA may exert particularly widespread effects given their ability to reach distal target tissues [27]. Results from our pathway analysis suggest that EV-associated miRNA in pregnancy may impact genes involved in major signaling systems, such as the Wnt and EGF signaling pathways, which are critical for development [57, 58]. Aberrant signaling in these pathways has been associated with pregnancy complications, such as gestational diabetes, and are targets of disrupted miRNA in pregnancy [59, 60]. Other significant pathways identified in our analysis included the cadherin and PGDF signaling pathways, which are associated with cell differentiation and regulation of cellular growth, consistent with the major changes in maternal and fetal growth that occur during pregnancy.

Gene Ontology results were similar. The gene targets of miRNA associated with GA were found to be enriched in categories describing RNA and DNA regulation, suggesting that miRNA in pregnancy could affect transcription factors in feedback loops that may amplify downstream effects, given small changes in miRNA number [61]. Transferase and kinase activity may also indicate changing energy and ATP usage. Interestingly, the gene targets of these miRNA are underrepresented in categories relating to immune responses, meaning that fewer genes than expected for these miRNA are involved in immune responses as they change over pregnancy (GA). The importance of miRNA in supporting healthy placentation in early pregnancy and immunotolerance during pregnancy is well-supported [62–64]. In particular, miR-223, miR-155 and miR146a are known to support regulatory T-cells that reduce inflammation and reduce cell-mediated immunity in early pregnancy [65]. In this data, miR-223 and miR-146a were negatively associated with GA, suggesting that some miRNA involved in immune tolerance may change with GA although GA-associated miRNA are generally not involved with the maternal immune response.

Strengths and limitations

In this study we examined how EV-associated miRNA profiles in maternal plasma change during pregnancy in a cohort of predominantly low-income Hispanic women, contributing to the literature on miRNA in pregnancy. The data presented here reflect a large array of miRNA screened (nearly 800 unique miRNA), more than is feasible with standard qPCR techniques, and characterizes miRNA associated with GA.

Although the samples in this study were collected during both early and late pregnancy, reflecting changes in miRNA across the course of pregnancy, the data reported here did not have the temporal resolution to describe the trajectory of miRNA expression in finer detail. Identification of miRNA associated with fetal sex may have been limited by smaller sample sizes but warrants further investigation. While this paper does not establish a baseline for normal miRNA expression throughout pregnancy, the data suggest that caution should be used in interpreting miRNA at different timepoints of pregnancy. While the full cohort included 35 preterm births, a sensitivity analysis in which they were removed found little difference in miRNA associated with GA in this population.

The EV miRNA extraction method we used included a size- selection step which enriched for exosomes, including placenta-derived exosomes, so these measures may also include miRNA from other circulating EV subtypes. The methods used did not specifically focus on placenta-derived EVs and miRNA, but it is likely that miRNA assessed in this study include both placenta-derived miRNA (estimated at 15–25% of all maternal circulating EVs [13]) as well as encapsulated maternal miRNA. Additionally, because this study lacks pre-pregnancy miRNA data on these participants, it is possible that some miRNA are not expressed exclusively during pregnancy, but generally present in women and may be affected by pregnancy status and GA. Combining EV classes and size fractions may have reduced the signal of maternal-fetal crosstalk or sex-specific miRNA expression that could be limited to a particular size class of EVs. Further investigation into the functional roles miRNA selected to different EV size classes may yield more details about maternal-fetal crosstalk and the mechanisms of miRNA action during pregnancy. Additional data on promising miRNA for use as biomarkers could be studied with more timepoints by RT-qPCR, establishing in a better baseline for future clinical use.

Finally, pathway analysis is also a challenging area for studies of miRNA in pregnancy, often relying on cancer-related data that share some mechanisms, but are somewhat unclear in interpretation [66, 67]. In order to address the complexity of miRNA predictive analysis, we limited analysis to GOslims categories, to reduce the number of associations with cancer-related pathways, and still found that the predicted genes reflected many characteristics that match with pathways hypothesized to change during pregnancy.

Conclusion

EV-associated miRNA that change in expression during pregnancy may be involved in important cell signaling and transcriptional regulation pathways that are critical to supporting a healthy pregnancy. A greater understanding of the biological role of miRNAs during pregnancy may help to develop biomarkers for pregnancy and birth complications through the use of blood or other non-invasive biofluids that would help to improve both maternal and children’s health trajectories [12, 50, 68, 69]. Previous studies on miRNA during pregnancy have been limited by small sample sizes, the quantification of a few select miRNA, and the omission of data of racial or ethnic background, and other key demographics of the participants, such as pre-pregnancy BMI.

Data from diverse populations are critical for understanding the normal roles of miRNA in pregnancy and how differences in underlying health conditions may affect pregnancy. Certain populations are at increased risk for metabolic disorders, such as obesity and diabetes, due to socio-cultural and economic backgrounds. Metabolic conditions are known to alter miRNA profiles [33, 70, 71], and thus associations between miRNA and underlying conditions must be continue to be studied in these at-risk groups. EV-enriched miRNA profiles during pregnancy in diverse study populations will help to improve the development of early and minimally invasive biomarkers that can be used to improve maternal and child health outcomes.

Materials and methods

Participants

The Maternal and Developmental Risks from Environmental and Social Stressors (MADRES) study is an ongoing cohort of more than 800 pregnant women and their children receiving prenatal care in Los Angeles, CA (USA). The full study design and cohort description is available in an earlier publication [26]. Pregnant women were recruited to the study at less than 30 weeks of gestation, were over 18 years of age, and spoke English or Spanish. Participants with the following conditions were excluded: 1. HIV-positive status, 2. physical, mental, or cognitive disability preventing participation, 3. current incarceration, and 4. multiple pregnancy. Study procedures were approved by the University of Southern California Institutional Review Board (USC IRB) and written informed consent was obtained for each participant (Fig 3).

[Figure omitted. See PDF.]

Fig 3. Recruitment and study flow chart.

https://doi.org/10.1371/journal.pone.0251259.g003

An early-pregnancy visit was conducted in the first or early second trimester, during which participants completed administered questionnaires, and were measured for height and weight. Height was measured using a Perspectives Enterprises Model PE-AIM-101 stadiometer and two measurements are averaged for each participant. Weight was measured using a Tanita digital scale accurate to 5g and two measurements are averaged for each participant. For both height and weight, a third measurement was taken if the first two measurements differed by more than 1cm or 0.05kg, according to PhenX protocol [26].

At this time, a plasma sample was also collected from each consenting participant. A visit in the third trimester was conducted in a similar manner. miRNA analyses were competed after delivery to ensure that miRNA measures from paired early- and late- pregnancy samples from the same participant were measured in the same batch.

GA at birth was determined using a hierarchy of the available data: (1) ultrasound measurement of crown-rump length in first trimester (<14 weeks GA, n = 208); (2) trimester ultrasound measurement of fetal biparietal diameter in the second trimester (<28 weeks GA, n = 96). (3) obstetric clinical estimate abstracted from the participant’s medical records (n = 45). (4) date of the participant’s self-reported last menstrual period (n = 2). Participants with no available GA at birth information were excluded from this study (n = 6). GA at biospecimen collection was back-calculated using estimated GA at birth and the difference between date of collection and date of birth.

miRNA extraction

Blood samples were collected in EDTA vacutainers, transported on ice, and were processed within 2 hours. Plasma aliquots (500μl) were stored at -80°C. Aliquots for miRNA extraction were selected from early (mean = 13.6 weeks, sd = 4.2) and late (mean = 31.7 weeks, sd = 2.0) pregnancy, as available. EV-associated miRNA was extracted from each plasma aliquot using the Qiagen ExoRNeasy kit following the manufacturer’s instructions, with a second addition of 90μL chloroform and an additional round of phase separation. The purified miRNA was frozen at -80°C until quantification.

Quantification

Samples were initially quantified by BioAnalyzer with the small RNA kit (Agilent Technologies, Inc. USA). If the sample contained a sufficient amount of miRNA (>100 pg/μl), 3 μl of the purified miRNA was prepared using the NanoString NCounter with Human v3 miRNA expression assay (Nanostring Technologies, Inc.) with a modified protocol. Rather than the 10:1 dilution called for after addition of the hybridized probes, samples were diluted 1:2 and quantified following the rest of the recommended NanoString protocol. Replicate samples (n = 6) were run to address possible technical variation across chips. However, no significant differences were observed in miRNA counts across replicates (Wilcoxon Rank Sum, P = 0.49, S1 Fig in S1 File).

Statistical analyses

RCC files measuring 800 miRNA for each sample were converted to text files using NSolver (v. 4.0.70, Nanostring Technologies) for downstream processing. miRNA counts were log2 transformed, and counts from both early and late pregnancy were normalized together in R (v. 4.0.0) using the “NanoString Norm” package (v. 1.2.1). Machine-read values were used for counts that fell below sample-specific limits of detection. Participants with high and low total miRNA counts (higher and lower than the mean total log-transformed count +/- 3 standard deviations) were dropped (8 participants), as well as those without estimated GA at birth (6 participants). For the duplicated samples used in quality control, the samples with smaller total miRNA counts were dropped (6 samples). If normalization factors flagged by NanoStringNorm were >6, individual samples were dropped (1 sample). A total of 491 total samples representing 337 unique participants were retained. Sample-specific background correction was conducted, calculated using the mean +1.5SD of negative controls. One hundred thirty miRNA had values above this sample-specific background threshold in at least 50% of samples (in either the first or third trimester) and were selected for further analysis. A Principal Components Analysis was conducted using the stats package in R [72] to identify a minimal set of covariates correlated with the first 4 miRNA PCs and to prioritize inclusion in hierarchical linear regression models, based on correlations (FDR P<0.05) (S2 Fig in S1 File).

Hierarchical linear regression models with random intercepts for subject ID and Nanostring chip were used to investigate the associations between log2-counts for each of the 132 miRNA and GA. Multiple testing correction was completed with False Discovery Rate (FDR) threshold of 0.05. Valid samples were required to have an associated GA at time of collection, following the selection criteria listed above. Other covariates were selected based on either a priori criteria, including socioeconomic status (SES) variables, or significant correlation with the first 4 PCs. Final covariates included: maternal age at 12 weeks of pregnancy, recruitment site, study entry time (before or after 20 weeks gestation), ethnicity/race (US-born White Hispanic, Foreign-born white Hispanic, other Hispanic, Black non-Hispanic, other non-Hispanic), birth order (1st baby or 2nd and higher), fetal sex, preferred language (English or Spanish), maternal education (less than high school, graduated high school, some college or completed college), pre-pregnancy BMI (continuous), with random effects for participant ID and chip, to account for participants providing two samples and possible batch effects. Maternal age at 12 weeks was determined using the best available GA at birth information, birth date, and maternal date of birth. Race and ethnicity were combined to better reflect the wide range of Hispanic participants.

All statistical analyses were conducted in R (4.0.0) [72] using the NLME package (3.1–143) [73]. Samples with missing values for covariates were excluded for all analyses. The main analysis used all available data (n = 449 observations). A sensitivity analysis was also conducted restricting to the subset of participants who contributed plasma samples at each timepoint (n = 296 observations). A second sensitivity analysis excluded preterm births (<37 weeks, n = 399 observations). Analyses stratified by fetal sex were also performed (n = 173 male, 161 female).

Pathway analyses

miRNA found to be significantly (FDR P<0.05) associated with GA were selected for downstream pathway analysis. Some Nanostring miRNA targets included multiple miRNAs (e.g. hsa-miR-106a-5p+hsa-miR-17-5p), so all unique miRNA IDs were used for these cases. MirDIP version 4.1.11.1, Database version 4.1.0.3 [74] was used to predict gene targets and restricted to genes with a minimum average score >0.6. Pathway enrichment was examined using PANTHER [75–77] and all human genes in the database. Overrepresentation tests for Gene Ontology Biological Process (GOslim BP) and Molecular Function (GOslim MF) as well as PANTHER pathways were performed using Fisher’s exact test and p-values were FDR-adjusted. Number of genes in each GO category were exported from PANTHER and plotted in R with package ggplot2 [78].

Supporting information

S1 File.

https://doi.org/10.1371/journal.pone.0251259.s001

(DOCX)

Acknowledgments

The authors would like to thank the MADRES participants, study staff and community clinic partners for their contributions to this work, as well as the SCEHSC lab and USC Genomics Core for their work on miRNA samples.

Citation: Foley HB, Howe CG, Eckel SP, Chavez T, Gevorkian L, Reyes EG, et al. (2021) Extracellular vesicle-enriched miRNA profiles across pregnancy in the MADRES cohort. PLoS ONE 16(5): e0251259. https://doi.org/10.1371/journal.pone.0251259

References

1. Ambros V. The functions of animal microRNAs. Nature. Nature Publishing Group; 2004 Sep 15;431(7006):350–5. pmid:15372042

2. Bartel DP. MicroRNAs: Target Recognition and Regulatory Functions. Cell. 2009 Jan;136(2):215–33. pmid:19167326

3. Armstrong DA, Green BB, Seigne JD, Schned AR, Marsit CJ. MicroRNA molecular profiling from matched tumor and bio-fluids in bladder cancer. Molecular Cancer. Molecular Cancer; 2015 Nov 12;14(1):1–9. pmid:26576778

4. Garcia-Contreras M, Shah SH, Tamayo A, Robbins PD, Golberg RB, Mendez AJ, et al. Plasma-derived exosome characterization reveals a distinct microRNA signature in long duration Type 1 diabetes. Sci Rep. Springer US; 2017 Jul 12;7(1):1–10. pmid:28127051

5. Lugli G, Cohen AM, Bennett DA, Shah RC, Fields CJ, Hernandez AG, et al. Plasma Exosomal miRNAs in Persons with and without Alzheimer Disease: Altered Expression and Prospects for Biomarkers. Zhang B, editor. 2015 Oct 1;10(10):e0139233. pmid:26426747

6. Marsit CJ, Eddy K, Kelsey KT. MicroRNA responses to cellular stress. Cancer Research. American Association for Cancer Research; 2006;66(22):10843–8. pmid:17108120

7. Avissar-Whiting M, Veiga KR, Uhl KM, Maccani MA, Gagne LA, Moen EL, et al. Bisphenol A exposure leads to specific microRNA alterations in placental cells. Reproductive Toxicology. 2010 Jul;29(4):401–6. pmid:20417706

8. Rager JE, Smeester L, Jaspers I, Sexton KG, Fry RC. Epigenetic Changes Induced by Air Toxics: Formaldehyde Exposure Alters miRNA Expression Profiles in Human Lung Cells. Environ Health Perspect. 2nd ed. 2011 Apr;119(4):494–500. pmid:21147603

9. Balaraman S, Schafer JJ, Tseng AM, Wertelecki W, Yevtushok L, Zymak-Zakutnya N, et al. Plasma miRNA Profiles in Pregnant Women Predict Infant Outcomes following Prenatal Alcohol Exposure. Ryabinin AE, editor. PLoS ONE. Public Library of Science; 2016 Nov 9;11(11):e0165081–20. pmid:27828986

10. Gray C, McCowan LM, Patel R, Taylor RS, Vickers MH. Maternal plasma miRNAs as biomarkers during mid-pregnancy to predict later spontaneous preterm birth: a pilot study. Sci Rep. Springer US; 2017 Apr 4;7(1):1–7. pmid:28127051

11. Salomon C, Guanzon D, Scholz-Romero K, Longo S, Correa P, Illanes SE, et al. Placental Exosomes as Early Biomarker of Preeclampsia: Potential Role of Exosomal MicroRNAs Across Gestation. The Journal of Clinical Endocrinology & Metabolism. 2017 May 22;102(9):3182–94. pmid:28531338

12. Gilad S, Meiri E, Yogev Y, Benjamin S, Lebanony D, Yerushalmi N, et al. Serum MicroRNAs Are Promising Novel Biomarkers. PLoS ONE. 2008 Aug 20;:1–7. pmid:18773077

13. Elfeky O, Longo S, Lai A, Rice GE, Salomon C. Influence of maternal BMI on the exosomal profile during gestation and their role on maternal systemic inflammation. Placenta. Elsevier Ltd; 2017 Feb 1;50:60–9. pmid:28161063

14. Kotlabova K, Doucha J, Hromadnikova I. Placental-specific microRNA in maternal circulation—identification of appropriate pregnancy-associated microRNAs with diagnostic potential. Journal of Reproductive Immunology. Elsevier Ireland Ltd; 2011 May 1;89(2):185–91. pmid:21513988

15. Fu G, Brkić J, Hayder H, Peng C. MicroRNAs in Human Placental Development and Pregnancy Complications. IJMS. 2013 Mar;14(3):5519–44. pmid:23528856

16. Valadi H, Ekström K, Bossios A, Sjöstrand M, Lee JJ, Lötvall JO. Exosome-mediated transfer of mRNAs and microRNAs is a novel mechanism of genetic exchange between cells. Nat Cell Biol. Nature Publishing Group; 2007 May 7;9(6):654–9. pmid:17486113

17. Tannetta D, Masliukaite I, Vatish M, Redman C, Sargent I. Update of syncytiotrophoblast derived extracellular vesicles in normal pregnancy and preeclampsia. Journal of Reproductive Immunology. 2017 Feb;119:98–106. pmid:27613663

18. Sarker S, Scholz-Romero K, Perez A, Illanes SE, Mitchell MD, Rice GE, et al. Placenta-derived exosomes continuously increase in maternal circulation over the first trimester of pregnancy. J Transl Med [Internet]. 2014;12(1):204–19. Available from: http://translational-medicine.biomedcentral.com/articles/10.1186/1479-5876-12-204 pmid:25104112

19. Adam S, Elfeky O, Kinhal V, Dutta S, Lai A, Jayabalan N, et al. Review: Fetal-maternal communication via extracellular vesicles—Implications for complications of pregnancies. Placenta. Elsevier Ltd; 2017 Jun 1;54:83–8. pmid:27939894

20. Ouyang Y, Mouillet JF, Coyne CB, Sadovsky Y. Review: Placenta-specific microRNAs in exosomes—Good things come in nano-packages. Placenta. Elsevier Ltd; 2014 Feb 1;35(S):S69–S73. pmid:24280233

21. Bullerdiek J, Flor I. Exosome-delivered microRNAs of “chromosome 19 microRNA cluster” as immunomodulators in pregnancy and tumorigenesis. Molecular cytogenetics. 2012 Jun 15;:1–4. pmid:22214275

22. Barchitta M, Maugeri A, Quattrocchi A, Agrifoglio O, Agodi A. Review Article The Role of miRNAs as Biomarkers for Pregnancy Outcomes: A Comprehensive Review. International Journal of Genomics. Hindawi; 2017 Aug 13;2017(3):1–11. pmid:28884117

23. Mavreli D, Lykoudi A, Lambrou G, Papaioannou G, Vrachnis N, Kalantaridou S, et al. Deep Sequencing Identified Dysregulated Circulating MicroRNAs in Late Onset Preeclampsia. In Vivo. 2020 Aug 31;34(5):2317–24. pmid:32871756

24. Cook J, Bennett PR, Kim SH, Teoh TG, Sykes L, Kindinger LM, et al. First Trimester Circulating MicroRNA Biomarkers Predictive of Subsequent Preterm Delivery and Cervical Shortening. Sci Rep. Springer US; 2019 Apr 3;9(1):1–9. pmid:30626917

25. Yang H, Ma Q, Wang Y, Tang Z. Clinical application of exosomes and circulating microRNAs in the diagnosis of pregnancy complications and foetal abnormalities. J Transl Med. BioMed Central; 2020 Jan 21;18(1):1–9. pmid:31900168

26. Bastain TM, Chavez T, Habre R, Girguis MS, Grubbs B, Toledo-Corral C, et al. Study Design, Protocol and Profile of the Maternal And Developmental Risks from Environmental and Social Stressors (MADRES) Pregnancy Cohort: a Prospective Cohort Study in Predominantly Low-Income Hispanic Women in Urban Los Angeles. BMC Pregnancy Childbirth. BMC Pregnancy and Childbirth; 2019 May 29;19(1):1–16. pmid:30606156

27. Lee B, Saadeldin I, Oh HJ. Embryonic-maternal cross-talk via exosomes: potential implications. SCCAA. 2015 Jul;:103–5. pmid:26185458

28. Li P, Wei J, Li X, Cheng Y, Chen W, Cui Y, et al. 17β-Estradiol Enhances Vascular Endothelial Ets-1/miR-126-3p Expression: The Possible Mechanism for Attenuation of Atherosclerosis. The Journal of Clinical Endocrinology & Metabolism. 2016 Nov 21;102(2):594–603.

29. Gu Y, Ampofo E, Menger MD, Laschke MW. miR‐191 suppresses angiogenesis by activation of NF‐kB signaling. The FASEB Journal. 2017 Sep 19;31(8):3321–33. pmid:28424351

30. Welten SMJ, Goossens EAC, Quax PHA, Nossent AY. The multifactorial nature of microRNAs in vascular remodelling. Cardiovasc Res. 2016 Mar 18;110(1):6–22. pmid:26912672

31. Xu J, Gu Y, Lewis DF, Cooper DB, McCathran CE, Wang Y. Downregulation of vitamin D receptor and miR‐126‐3p expression contributes to increased endothelial inflammatory response in preeclampsia. American Journal of Reproductive Immunology. 2019 Jul 30;82(4):1–10. pmid:31323164

32. Gao L, Wang G, Liu W-N, Kinser H, Franco HL, Mendelson CR. Reciprocal Feedback Between miR-181a and E2/ERα in Myometrium Enhances Inflammation Leading to Labor. The Journal of Clinical Endocrinology & Metabolism. 2016 Oct 5;101(10):3646–56.

33. Lozano-Bartolomé J, Llauradó G, Portero-Otin M, Altuna-Coy A, Rojo-Martínez G, Vendrell J, et al. Altered Expression of miR-181a-5p and miR-23a-3p Is Associated With Obesity and TNFα-Induced Insulin Resistance. The Journal of Clinical Endocrinology & Metabolism. 2018 Feb 1;103(4):1447–58. pmid:29409019

34. Hodson K, Dalla Man C, Smith F, Thelwall P, Cobelli C, Robson S, et al. Mechanism of Insulin Resistance in Normal Pregnancy. Horm Metab Res. 2013 Jul 16;45(08):567–71. pmid:23549674

35. Méndez-Mancilla A, Lima-Rogel V, Toro-Ortíz JC, Escalante-Padrón F, Monsiváis-Urenda AE, Noyola DE, et al. Differential expression profiles of circulating microRNAs in newborns associated to maternal pregestational overweight and obesity. Pediatric Obesity. John Wiley & Sons, Ltd; 2017 Oct 17;13(3):168–74. pmid:29045034

36. Higashijima A, Miura K, Mishima H, Kinoshita A, Jo O, Abe S, et al. Characterization of placenta-specific microRNAs in fetal growth restriction pregnancy. Prenat Diagn. 2013 Jan 27;33(3):214–22. pmid:23354729

37. Nejad RMA, Saeidi K, Gharbi S, Salari Z, Saleh-Gohari N. Quantification of circulating miR-517c-3p and miR-210-3p levels in preeclampsia. Pregnancy Hypertension. Elsevier; 2019 Apr 1;16:75–8. pmid:31056162

38. Prieto DMM, Markert UR. MicroRNAs in pregnancy. Journal of Reproductive Immunology. Elsevier Ireland Ltd; 2011 Mar 1;88(2):106–11. pmid:21353310

39. Morales-Prieto DM, Ospina-Prieto S, Chaiwangyen W, Schoenleben M, Markert UR. Pregnancy-associated miRNA-clusters. Journal of Reproductive Immunology. Elsevier Ireland Ltd; 2013 Mar 1;97(1):51–61. pmid:23432872

40. Ishibashi O, Ohkuchi A, Ali MM, Kurashina R, Luo S-S, Ishikawa T, et al. Hydroxysteroid (17-β) Dehydrogenase 1Is Dysregulated by Mir-210and Mir-518cThat Are Aberrantly Expressed in Preeclamptic Placentas. Hypertension. 2012 Feb;59(2):265–73. pmid:22203747

41. Gao Y, She R, Wang Q, Li Y, Zhang H. Up-regulation of miR-299 suppressed the invasion and migration of HTR-8/SVneo trophoblast cells partly via targeting HDAC2 in pre-eclampsia. Biomedicine et Pharmacotherapy. Elsevier; 2018 Jan 1;97:1222–8. pmid:29145147

42. Li X, Song Y, Liu D, Zhao J, Xu J, Ren J, et al. MiR-495 Promotes Senescence of Mesenchymal Stem Cells by Targeting Bmi-1. Cell Physiol Biochem. 2017 Jul 10;42(2):780–96. pmid:28628915

43. Hromadnikova I, Dvorakova L, Kotlabova K, Krofta L. The Prediction of Gestational Hypertension, Preeclampsia and Fetal Growth Restriction via the First Trimester Screening of Plasma Exosomal C19MC microRNAs. IJMS. 2019 Jun;20(12):2972–18. pmid:31216670

44. Zhao C, Zhang T, Shi Z, Ding H, Ling X. MicroRNA-518d regulates PPARα protein expression in the placentas of females with gestational diabetes mellitus. Mol Med Report. 2014 Mar 17;9(6):2085–90. pmid:24639097

45. Gillet V, Ouellet A, Stepanov Y, Rodosthenous RS, Croft EK, Brennan K, et al. miRNA Profiles in Extracellular Vesicles From Serum Early in Pregnancies Complicated by Gestational Diabetes Mellitus. The Journal of Clinical Endocrinology & Metabolism. https://academic.oup.com/jcem; 2019 Sep 23;:1–13. pmid:31058973

46. Miura K, Higashijima A, Murakami Y, Fuchi N, Tsukamoto O, Abe S, et al. Circulating Levels of Pregnancy-Associated, Placenta-Specific microRNAs in Pregnant Women With Placental Abruption. Reprod Sci. 2016 Dec 16;24(1):148–55. pmid:27297699

47. Donker RB, Mouillet JF, Chu T, Hubel CA, Stolz DB, Morelli AE, et al. The expression profile of C19MC microRNAs in primary human trophoblast cells and exosomes. Molecular Human Reproduction. 2012 Jul 4;18(8):417–24. pmid:22383544

48. Miura K, Higashijima A, Murakami Y, Tsukamoto O, Hasegawa Y, Abe S, et al. Circulating chromosome 19 miRNA cluster microRNAs in pregnant women with severe pre-eclampsia. J Obstet Gynaecol Res. 2015 Jul 30;41(10):1526–32. pmid:26224638

49. Wang Y, Du X, Wang J. Transfer of miR-15a-5p by placental exosomes promotes pre-eclampsia progression by regulating PI3K/AKT signaling pathway via CDK1. Molecular Immunology. 2020 Dec;128:277–86. pmid:33197745

50. Houshmand-Oeregaard A, Schrölkamp M, Kelstrup L, Hansen NS, Hjort L, Thuesen ACB, et al. Increased expression of microRNA-15a and microRNA-15b in skeletal muscle from adult offspring of women with diabetes in pregnancy. Human Molecular Genetics. 2018 Mar 8;27(10):1763–71. pmid:29528396

51. Hromadnikova I, Kotlabova K, Hympanova L, Krofta L. Cardiovascular and Cerebrovascular Disease Associated microRNAs Are Dysregulated in Placental Tissues Affected with Gestational Hypertension, Preeclampsia and Intrauterine Growth Restriction. Jeyaseelan K, editor. PLoS ONE. 2015 Sep 22;10(9):e0138383–27. pmid:26394310

52. Yang Q, Gu W-W, Gu Y, Yan N-N, Mao Y-Y, Zhen X-X, et al. Association of the peripheral blood levels of circulating microRNAs with both recurrent miscarriage and the outcomes of embryo transfer in an in vitro fertilization process. J Transl Med. BioMed Central; 2018 Jul 3;:1–14. pmid:29316942

53. Rodosthenous RS, Burris HH, Sanders AP, Just AC, Dereix AE, Svensson K, et al. Second trimester extracellular microRNAs in maternal blood and fetal growth: An exploratory study. Epigenetics. Taylor & Francis; 2017 Nov 29;12(9):1–8. pmid:28758828

54. Murakami Y, Miura K, Sato S, Higashijima A, Hasegawa Y, Miura S, et al. Reference values for circulating pregnancy-associated microRNAs in maternal plasma and their clinical usefulness in uncomplicated pregnancy and hypertensive disorder of pregnancy. J Obstet Gynaecol Res [Internet]. 2018 Mar 8;44(5):840–51. Available from: pmid:29516574

55. Miura K, Miura S, Yamasaki K, Higashijima A, Kinoshita A, Yoshiura K-I, et al. Identification of Pregnancy-Associated MicroRNAs in Maternal Plasma. Clinical Chemistry. 2010 Oct 15;:1767–71. pmid:20729298

56. Baek D, Villén J, Shin C, Camargo FD, Gygi SP, Bartel DP. The impact of microRNAs on protein output. Nature. 2008 Jul 30;455(7209):64–71. pmid:18668037

57. Tepekoy F, Akkoyunlu G, Demir R. The role of Wnt signaling members in the uterus and embryo during pre-implantation and implantation. J Assist Reprod Genet. 2014 Dec 24;32(3):337–46. pmid:25533332

58. Forbes K, Westwood M. Maternal growth factor regulation of human placental development and fetal growth. Journal of Endocrinology. BioScientifica; 2010 Sep 10;207(1):1–16. pmid:20817666

59. Zhu Y, Tian F, Li H, Zhou Y, Lu J, Ge Q. Profiling maternal plasma microRNA expression in early pregnancy to predict gestational diabetes mellitus. International Journal of Gynecology & Obstetrics. 2015 Mar 23;130(1):49–53. pmid:25887942

60. Li J, Song L, Zhou L, Wu J, Sheng C, Chen H, et al. A MicroRNA Signature in Gestational Diabetes Mellitus Associated with Risk of Macrosomia. Cell Physiol Biochem. 2015 Aug 1;37(1):243–52. pmid:26302821

61. Martinez NJ, Walhout AJM. The interplay between transcription factors and microRNAs in genome-scale regulatory networks. BioEssays. 2009 Apr;31(4):435–45. pmid:19274664

62. Kamity R, Sharma S, Hanna N. MicroRNA-Mediated Control of Inflammation and Tolerance in Pregnancy. Frontiers in Immunology. 2019 Apr 4;:1–11. pmid:30723466

63. Bidarimath M, Khalaj K, Wessels JM, Tayade C. MicroRNAs, immune cells and pregnancy. Cellular and Molecular Immunology. Nature Publishing Group; 2014;11(6):538–47. pmid:24954225

64. Schjenken JE, Zhang B, Chan H, Sharkey DJ, Fullston T, Robertson SA. miRNA Regulation of Immune Tolerance in Early Pregnancy. American Journal of Reproductive Immunology. 2016 Feb 8;:1–9. pmid:26804209

65. Robertson SA, Zhang B, Chan H, Sharkey DJ, Barry SC, Fullston T, et al. MicroRNA regulation of immune events at conception. Mol Reprod Dev. 2017 Sep 23;:1–12. pmid:28135053

66. Holtan SG MD, Creedon DJ MD PhD, Paul Haluska MD P, Svetomir N Markovic MD P. Cancer and Pregnancy: Parallels in Growth, Invasion, and Immune Modulation and Implications for Cancer Therapeutic Agents. Mayo Clinic Proceedings. Mayo Foundation for Medical Education and Research; 2011 Sep 10;84(11):985–1000.

67. Vidal DO, Ramao A, Pinheiro DG, Muys BR, Lorenzi JCC, de Padua Alves C, et al. Highly expressed placental miRNAs control key biological processes in human cancer cell lines. Oncotarget. 2018 May 8;9(34):23554–63. pmid:29805755

68. Roberts JM, Hubel CA. Pregnancy: A Screening Test for Later Life Cardiovascular Disease. Women’s Health Issues. Jacobs Institute of Women’s Health; 2010 Sep 1;20(5):304–7. pmid:20800765

69. Dayan N, Schlosser K, Stewart DJ, Delles C, Kaur A, Pilote L. Circulating MicroRNAs Implicate Multiple Atherogenic Abnormalities in the Long-Term Cardiovascular Sequelae of Preeclampsia. American Journal of Hypertension. 2018 May 23;31(10):1093–7. pmid:29800045

70. Tagoma A, Alnek K, Kirss A, Uibo R, Haller-Kikkatalo K. MicroRNA profiling of second trimester maternal plasma shows upregulation of miR-195-5p in patients with gestational diabetes. Gene. Elsevier; 2018 Jun 13;672:137–42. pmid:29879500

71. Prats-Puig A, Ortega FJ, Mercader JM, Moreno-Navarrete JM, Moreno M, Bonet N, et al. Changes in Circulating MicroRNAs Are Associated With Childhood Obesity. The Journal of Clinical Endocrinology & Metabolism. 2013 Oct;98(10):E1655–60. pmid:23928666

72. R Core Team. R: A Language and Environment for Statistical Computing [Internet]. Vienna, Austria; 2013. Available from: http://www.R-project.org/

73. Pinheiro J, Bates D, DebRoy S, Sarkar D, R Core Team. nlme: Linear and Nonlinear Mixed Effects Models [Internet]. 2020. Available from: https://CRAN.R-project/package=nlme

74. Tokar T, Pastrello C, Rossos AEM, Abovsky M, Hauschild A-C, Tsay M, et al. mirDIP 4.1—integrative database of human microRNA target predictions. Nucleic Acids Research. Oxford University Press; 2017 Nov 29;46(D1):D360–70.

75. Thomas PD, Campbell MJ, Kejariwal A, Mi H, Karlak B, Daverman R, et al. PANTHER: A Library of Protein Families and Subfamilies Indexed by Function. Genome Research. Cold Spring Harbor Lab; 2003 Sep 1;13(9):2129–41. pmid:12952881

76. Thomas PD, Kejariwal A, Guo N, Mi H, Campbell MJ, Muruganujan A, et al. Applications for protein sequence-function evolution data: mRNA/protein expression analysis and coding SNP scoring tools. Nucleic Acids Research. 2006 Jul 1;34(Web Server):W645–50. pmid:16912992

77. Mi H, Huang X, Muruganujan A, Tang H, Mills C, Kang D, et al. PANTHER version 11: expanded annotation data from Gene Ontology and Reactome pathways, and data analysis tool enhancements. Nucleic Acids Research. 2017 Jan 3;45(D1):D183–9. pmid:27899595

78. Wickham H. ggplot2: Elegant Graphics for Data Analysis [Internet]. Springer-Verlag New York; 2016. Available from: https://ggplot2.tidyverse.org