DNA was extracted from the FFPE placental blocks using a QIAamp DNA FFPE tissue kit (Qiagen, catalog no. 56404) according to the manufacturer’s protocol. Infinium HD FFPE Restore kit (Illumina, San Diego, CA) was used for DNA restoration (Siegel et al., 2014) followed by bisulfite conversion of 500 ng of DNA using the EZ DNA Methylation kit (Zymo Research, Irvine, CA) according to the manufacturer’s instructions. The methodology has been detailed earlier (Radhakrishna et al., 2016; Radhakrishna et al., 2021b).

Genome-wide DNA methylation profiling was performed using the Infinium Methylation EPIC BeadChip array (Illumina Inc., San Diego, CA) with bisulfite-treated genomic DNA according to the manufacturer’s protocol (Vishweswaraiah et al., 2019; Radhakrishna et al., 2021b). The Methylation EPIC BeadChip microarray provides quantitative measurement of 853,307 CpG sites, including the methylation of 9,961 CpG site regulators of miRNA-encoding genes present in the array (Moran et al., 2016). The sample placement in each chip was randomized to avoid confounding and to achieve successful microarray experiments (Verdugo et al., 2009). During the preprocessing of the methylation data, CpG-sites annotated to X and Y chromosomes and/or containing SNPs near or within the probe sequence (within 10 bp of the CpG site), probes that lacked beta values, and probes with minor allele frequency exceeding 0.05 were excluded from the analysis (Chen et al., 2013; Liu et al., 2013; Wilhelm-Benartzi et al., 2013; Daca-Roszak et al., 2015), followed by the analysis of the beta values for the remaining CpG sites. SNPs within 10 bp of the microRNA binding site may affect microRNA and mRNA interactions and reduce binding affinity (Li et al., 2016).

To establish associations between miRNAs and the biological processes they regulate in OUD/NOWs, we performed a gene ontology (GO) analysis. Pathway analysis was carried out using Ingenuity Pathway Analysis (IPA^®) using differentially expressed miRNAs at FDR p-value <0.05. MiRNAs were removed from the analysis if they were duplicated or unrecognized by IPA. The most statistically significant miRNAs identified in the NOWS were used in canonical pathways.

Gene ontology analysis found major dysregulated pathways affected by methylation changes on miRNAs associated with NOWS and OUD-related outcomes. A comparison was made between three groups i. Distinguishing NOWS from prenatal opioid-exposure without NOWS (+Opioids/+NOWS, versus + Opioids/-NOWS), ii. Distinguishing prenatal opioid use versus unexposed controls (OUD detection) (+Opioids/+NOWS and +Opioids/-NOWS versus -Opioids/-) and iii. distinguishing NOWS versus unexposed controls (+Opioids/+NOWS versus + Opioids/-NOWS). Significant functional biological processes regulated by these miRNAs were identified with the plausible molecular mechanism of differentially expressed miRNAs. Network interaction between miRNAs and their target genes was provided in Figure 1, Figure 2, and Figure 3.

MicroRNA Target Prediction and Functional Study Database (miRDB) server (http://miRdb.org) were used to predict putative targets of aberrantly methylated miRNAs. With a set target score of ≥60 (Wang, 2016), we searched the miRDB database for predictions of targets of some of these significant miRNAs. From the miRDB, we retrieved the target genes for 47 miRNAs. From this, we discovered multiple predicted targets for each miRNA (Adlakha and Saini, 2014). These predicted 47 miRNA targets were compared with the differentially methylated genes identified in our genome-wide methylation data of the same sample cohort (Radhakrishna et al., 2021b). The results showed that multiple predicted targets (genes) of these miRNAs were also found to be differentially methylated in our patients. Supplementary Tables S1 to S15 provide examples of miRNA targets discussed in the manuscript that are important and available in the miRDB database.

(opioid-exposed infants who required pharmacologic management for NOWS versus opioid-exposed infants that did not require treatment for NOWS) (+Opioids/+NOWS versus + Opioids/-NOWS) (Table 2). The miR-130 family includes four members (miR-130a, miR-130b, miR-301a, and miR-301b) with the same source sequences and which perform similar biological functions (Wang et al., 2020). The miR-130 mature miRNA is known to increase the expression of PPARA mRNA (Papi et al., 2013). PPARA is a nuclear hormone receptor that responds to certain types of cannabinoids (O'Sullivan, 2016).

MiR-130 is regulated by TP53, HNF4A, AGO2, TNFRFS1B, and insulin, and was found to regulate the expressions of KDR, IL4, IFNG, TNF, and RELA. This regulation is likely dampened to a certain extent due to methylation, resulting in higher protein expressions and activations of the miRNA. MiR-130 regulates IL4, which in turn regulates the mu-opioid receptor (MOR), whose principal ligands include opioid peptides and analgesics. The MOR also has an important role in dependence on other drugs of abuse, such as nicotine, cocaine, and alcohol via its modulation of the dopamine system.

Interestingly, miR-130 putatively targets several other genes that are preferentially involved in brain disorders. For instance, miR-130, in combination with other miRNAs, jointly regulates ATXN1, which causes spinocerebellar ataxia type 1 (SCA1). SCA1 causes seizures, slurred speech, slowness of movement, and cognitive impairments. Similar neurological signs such as slurred speech or slowed movements are seen in OUD subjects, and conditions such as tremors and seizures are common in both OUD and infants with NOWS. Our genome-wide methylation analysis of NOWS revealed hypermethylation of the ATXN1 gene (Radhakrishna et al., 2021b).

Inhibition of miR-130 leads to decreased translocation of phosphorylated RELA protein to nuclei (Hazra et al., 2019). RELA protein (Chen et al., 2006) plays a significant role in the downstream activation of the OPRD1 (also known as DOR, δ-opioid receptor, or delta-opioid receptor) is a paralog of OPRM1 associated with opioid dependence gene that encodes the delta-opioid receptor (DOR) protein in humans. In the central nervous system, DOR and mu-opioid receptors (MOR) can interact and modulate each other’s and are involved in the regulation of pain (Beaudry et al., 2015; Olesen et al., 2018). DOR is responsible for reducing the intensity of pain signals, while MOR is responsible for reducing the frequency of pain signals. Heterodimerization of MOR and DOR can indeed impact the recruitment of beta-arrestin 2 (ARRB2) protein, which subsequently influences downstream intracellular signaling (Rozenfeld and Devi, 2007).

(opioid-exposed infants who required pharmacologic management for NOWS + opioid-exposed infants that did not require treatment for NOWS versus unexposed controls. (+Opioids/+NOWS plus + Opioids/-NOWS versus -Opioids/-NOWS). Table 3 describes the miRNA dataset from mothers with a history of opioid usage. This cohort was compared against normal unexposed controls and showed significant methylation differences in miR-34, miR-129, miR-181, and miR-548 among others.

The binding of the AGO2 protein was found to occur with miR-34 mature miRNAs (Wu et al., 2011). The presence of AGT proteins was found to increase expressions of mouse miR-34 and miR-129 mature miRNAs indicating positive regulation by AGT (Jin et al., 2012; Huo et al., 2019). AGT protein increases paracrine activation of CNR1 protein (Turu et al., 2009), followed by clustering of CNR1 with CNR2. Activated CNR2 protein increases the expression of OPRM1 (Borner et al., 2006) possibly resulting in an OPRM1-OPRD1-OPRL1-OPRK1 feedback loop.

The binding of the TP53 response element from the miR-34 promoter and TP53 protein was found to occur followed by targeting of TP53 mRNA by miR-34 mature miRNA. TP53 protein further increases the transcription of the miR-34 gene and miR-34 mature miRNA was found to, in turn, upregulate the expression of acetylated (K382) p53 protein followed by increasing activity of p53 protein (Feng et al., 2011). Protein modification of AGO2 leads to an increased expression of miR-548 and miR-181 (Jin et al., 2012; Garibaldi et al., 2016). Homozygous experimental p53 gene deletion was found to decrease the expression of miR-548 mature miRNA (Shin et al., 2009). Experimental interference of human p53 mRNA by shRNA led to an increased expression of miR-519 mature miRNA (Garibaldi et al., 2016). Binding of 3′UTR from ZEB2 mRNA and miR-181A mature miRNA occurs, indicating probable regulation of ZEB2 mRNA. Negative regulation of TP53 in this pathway leads to activations of several genes in the pathway including PPARA which is a nuclear hormone receptor capable of taking certain cannabinoid ligands.

Women who used illicit or unprescribed opioids during pregnancy have a higher risk of fetal growth restriction and preterm delivery (Maghsoudlou et al., 2017). Recent studies have shown differential miR-548 expression profile was associated with spontaneous preterm births (Gray et al., 2017). Placental DNA methylation of miR-548 and WWTR1 genes influence insulin sensitivity during pregnancy, and preterm birth and is linked to insulin resistance (Hofman et al., 2004; Mathai et al., 2012). Interestingly, nine of the 47 dysregulated miRNAs belonged to the miR-548 family, including miR548G, miR548F1, miR548W, miR548H3, miR548AE2, miR548F5, miR548F3, miR548C, and miR548C, all of which were hypermethylated. Moreover, WWTR1 is also hypermethylated in the present study.

(opioid-exposed infants who required pharmacologic management for NOWS versus unexposed controls (+Opioids/+NOWS versus -opioids/-NOWS, control) (Table 4). MiR-515 is the largest miRNA gene cluster in humans (Zhang M. et al., 2016), the family members of miR-515 include miR-516 and miR-518 (Zhang M. et al., 2016). MiRNA-515 and miR-34 are found to be connected directly to opioid and cannabinoid receptors respectively. MiR-515 was found to regulate opioid genes OPRM1, and OPRD1, while miR-34 was found to regulate cannabinoid genes GPR19 and PPARA via Notch and TP53 proteins.

Mature miR-515 is known to decrease the activity of the human nuclear factor kappa B (NFkB) complex (Keklikoglou et al., 2012), however, in the absence of active miR-515 due to methylation as seen in the current cohort, the Nfkb complex is actively expressed leading to continued expression of OPRM1 mRNA. Thus, methylation of miR-515 results in the transcription of human OPRM1 mRNA which otherwise would be under the strict regulation of the NFkB complex (Kraus et al., 2003; Memet, 2006).

RELA mRNA is targeted by miR-515 mature miRNA leading to its decreased expression (Keklikoglou et al., 2012). Interference of Rela mRNA by siRNA decreases the expression of Oprd1 mRNA. RELA protein increases activation of the reporter gene with a promoter fragment (-262–1) from the OPRD1 gene (Chen et al., 2006). Methylation of miR-515 leads to the absence of regulation leading to an increased expression of RELA. Increased RELA subsequently increases its downstream partner OPRD1. The OPRM1 and OPRD1 genes were also found to be distinctively methylated in the previously described datasets. OPRM1 was found hypomethylated in mothers who used opioids during pregnancy and delivered NOWS newborns which required immediate medical care when compared to mothers who had used opioids during pregnancy and delivered newborns without NOWS. In contrast, OPRD1 was found hypermethylated in mothers with a history of opioid usage who delivered babies without NOWS when compared against unexposed controls. Both OPRD1 and OPRM1 were found to be distinctly methylated in the current dataset of mothers who used opioids and delivered NOWS babies when compared against unexposed controls. However, PPARA activation status differed across these three datasets.

The absence of active miR-130 in the +Opioid/+NOWS group may have led to decreased expression of PPARA mRNA, therefore, leading to reduced production of oxidation enzymes to effectively scavenge for free radicals generated by oxidative stress from opioid use. This may have resulted in mothers delivering newborns with NOWS that required medical intervention. Interestingly, when the Opioid+/NOWS- group was compared against unexposed controls, PPARA was found to be under the positive regulation of TP53 leading to its activation. This protective role of PPARA was absent in the +Opioids/+NOWS cohort of mothers when compared against unexposed controls. This is possibly due to the active status of miR-134 and/or to a different set of regulatory partners Notch and TP53 whose role in this context is not known and can only be speculated. MiR-34 was found to regulate cannabinoid genes including PPARA via Notch and TP53 proteins. Though the anti-oxidation activity of PPARA was probably negated by the extensive and consistent activations of opioid genes OPRM1, and OPRD1, unlike seen in any of the other two cohorts and thus resulting in mothers delivering NOWS newborns that required medical care. Moreover, miR-34 has a functional role in the regulation of stress-induced anxiety, depression with suicidal ideation (Haramati et al., 2011). Anxiety and depression are common and normal phenomena, in mothers of infants experiencing NOWS (Corr et al., 2020).

SIRT1 which is hypomethylated in the current cohort study (FDR) p ≤ 0.05) (Radhakrishna et al., 2021b), is a nicotinamide adenosine dinucleotide (NAD)-dependent deacetylase that removes acetyl groups from various proteins. SIRT1 normally functions to limit the expression of miR-134, which targets the critical activity-dependent transcription factor (CREB) essential for learning and memory (Lv et al., 2013; O'Neal-Moffitt et al., 2014; Liu et al., 2017). SIRT1 exerts neuroprotection against ischemic injury and various neurodegenerative disorders (Xu et al., 2018). Dysregulation of brain SIRT1 activity can have devastating brain consequences including neurological dysfunction (Xu et al., 2018).

Experimental interference of human p53 mRNA by shRNA led to an increased expression of miR-519 mature miRNA (Garibaldi et al., 2016). Homozygous experimental p53 gene deletion was found to decrease the expression of miR-548 mature miRNA (Shin et al., 2009). MiR-10 which belongs to the miR-125A family member (Tehler et al., 2011) was found to decrease the activation of p53 (Joo et al., 2013). Studies on mouse mmu-miR-10 mature miRNA were found to decrease the expression of mouse p53 protein (Sen et al., 2014). Targeting of LIN28 mRNA by miR-10 and miR-34 mature miRNAs was found to occur leading to decreased translation of LIN28 (Zhong et al., 2010; Jain et al., 2012). p53 (TP53) protein was found to decrease the binding of BRF1 protein and RNA polymerase iii complexes (Crighton et al., 2003). Experimental inhibition of active RNA polymerase iii complexes was found to result in increased expression of miR-1909 while decreasing the expression of miR-1178 mature miRNA (Koo et al., 2015) demonstrating the role of transcriptional regulation by RNA polymerase iii.

The study has some limitations, namely, a lack of racial or ethnic diversity, given that it focused on individuals of European origin. A second limitation is that we have not replicated our findings in an independent study cohort, and therefore, we will continue to carry out in vitro and in vivo experiments to validate the conclusions in the future.

To our knowledge, this is the first report about miRNA methylation levels of NOWS on a previously well-documented large cohort of placental tissue specimens with genome-wide methylation profiles and gene expression profiles. NOWS is associated with an altered miRNA methylation pattern in the placenta, suggesting that miRNA deregulation is involved in the pathogenesis of NOWS. The current experimental data suggest that epigenetic variations in the placental tissue can serve as surrogate markers for brain health at birth, and thus infant micro-RNA “signatures” can predict the severity of NOWS even before withdrawal symptoms begin. The differential methylation of several miRNAs has been found in association with NOWS. The findings have biological plausibility, and they include several known biological pathways and genes with possible mechanisms associated with NOWS development. The identified dysregulated pathways and genes may therefore provide important opportunities for the development of novel future miRNA-based therapeutic approaches for NOWS. Several preclinical and clinical trials have been reported for miRNA-based therapeutics and few reports indicate that miRNA expression levels can be modified by modulating the miRNA processing pathway. However, validation with large NOWS cohorts and using robust techniques for DNA methylation analysis of miRNA genes are warranted before the application in clinical practice.