During early development, human breast milk and bovine milk-based formulas are the best and frequently the only source of infant nutrition. Casein, a class of proteins that makes up roughly 80 % of the protein found in milk, is broken down during digestion into a number of smaller peptides. One of these breakdown products is the opioid-like peptide beta-casomorphin-7 (BCM7) [1, 2], and a structurally similar opioid-like peptide is released during digestion of wheat-derived gluten [3, 4]. Canonically, food-derived opioid peptides are reported to act on the μ-opioid receptor (μ-OR) and they are able to reach the bloodstream [5] and cerebrospinal fluid [6] with the potential to exert effects in the brain [7]. However, the downstream effects of this food-derived opioid peptide exposure, especially in terms of gene expression and regulation, are still largely unknown. Beta-casein variants are of two major sub-types; namely A1 and A2 type of beta-casein. It is noteworthy to mention that BCM-7, the opioid peptide is only derived from A1-type of beta-casein and not A2 upon digestion.

We recently showed that casein and gluten-derived opioid peptides decrease excitatory amino acid transporter 3 (EAAT3)-mediated cysteine uptake by human GI epithelial cells and human neuronal cells in a dose-dependent manner, resulting in redox and epigenetic consequences via their capacity for μ-OR activation [8]. Epigenetic regulation of gene expression is of particular importance during early development when changes in DNA or histone methylation status can exert lifelong and even transgenerational effects [9–11]. Thus milk-derived opioid peptides may exert an important epigenetic influence with extended consequences. It is noteworthy to mention that the human form of BCM-7 was far less potent as compared to the bovine form of BCM-7 in inhibition of cysteine uptake, inducing oxidative stress as well as decreased SAM levels. 

With regard to neural development, epigenetic regulation can influence neural stem cell differentiation as well as neuronal plasticity and/or neuronal maturation, which are particularly prominent during fetal and early postnatal development [12]. We previously highlighted the critical role of prenatal and postnatal epigenetic programming in the redox-based linkage between GI, brain, and immune systems, which could potentially contribute to the etiology of autism and other disorders [10]. These redox-based metabolic and epigenetic changes have not only been identified in early developmental and neurological diseases [13–15], but also characterized in diseases arising across the lifespan, including immunological [10, 16] and neurodegenerative disorders [17–19].

A gluten-free (GF) casein-free (CF) diet has been reported to improve intestinal, autoimmune and neurological symptoms in celiac disease [20, 21], autism [22] and schizophrenia [23, 24]. However, the mechanism by which a GF/CF diet improves these symptoms is not fully understood, and in the absence of definitive evidence, the potential benefit of these dietary interventions remains unexplained or considered to be poor [25]. In the current study we investigated the functional effects of milk-derived opioid peptides using pathway analysis of genome wide changes in global DNA methylation patterns, measured via methyl-CpG binding domain (MBD) protein-enriched genome sequencing (MBD-seq) in cultured human neuroblastoma cells, as well as changes in transcription status using a genome-wide microarray expression assay. The observed changes may contribute to adverse effects of milk-derived opioid peptides in sensitive individuals.

Our previous results reported global DNA methylation changes under the influence of opioid peptides. However, the hBCM-7 was far less potent as compared to the bovine form of BCM-7 in altering the global DNA methylation status as well as changes in GSH and SAM levels [9]. In the current study, we investigated the genome-wide epigenetic changes under the influence of these opioid peptides. To investigate functional pathway and gene network changes induced by hBCM7, bBCM7 and morphine, the DNA methylation MBD-seq and DNA microarray data were collected. For this purpose, control SH-SY5Y neuroblastoma cells and cells treated for 4 h with 1 μM hBCM7, bBCM7 or morphine were used, the latter serving as a positive opioid effect control. The dose and time-points were selected based on our previous studies [8]. Whole genome DNA MBD-seq revealed differentially methylated promoter transcripts (DMTs), as defined by FDR < 0.1 and microarray data revealed differentially expressed transcripts (DETs), defined by fold change ≥ 1.5 and raw p-value ≤ 0.05, which included differentially methylated/ transcribed genes from both genic and non-coding regions.

Treatment with the prototypical opiod morphine (positive control) resulted in 4950 DETs, while bBCM7 treatment yielded 1467 and hBCM7 treatment resulted in 581; hence the rank order in gene expression changes was morphine > bBCM7 > hBCM7 (Fig. 1). Morphine treatment resulted in a distinct bias towards gene up-regulation, with 4785/4950 DETs (96.7 %) reflecting increased expression, while for bBCM7 1259/1467 (85.8 %) of DETs were associated with increased expression. By contrast, only 209/581 DETs (36.0 %) for hBCM7 treatment reflected up-regulation.

Next, about 501 DETs were shared by both morphine and bBCM7, 178 overlapped between morphine and hBCM7, 76 overlapped between hBCM7 and bBCM7, while 33 overlapped between all three treatments. Morphine and bBCM7 shared a higher number of DETs, as compared to morphine and hBCM7 (Fig. 1). Hierarchical clustering for top 50 DETs showed that changes in gene expression for hBCM7 and bBCM7 are more related to each other than to morphine (Fig. 2).

Morphine treatment resulted in 3673 promoter region DMTs (defined as transcription start site ± 500 bp), bBCM7 ~ 197 and hBCM7 ~ 346 DMTs; with the rank for differential promoter region methylation as morphine > hBCM7 > bBCM7. Of these DMTs, morphine and bBCM7 shared 159, morphine and hBCM7 had 25 common and only 1 DMTs overlapped between all three treatments (Fig. 1).

Next, promoter region methylation of all 53,561 genes was evaluated at their transcription start sites (TSS) and relative DNA methylation was calculated for 1000 bp in either direction, normalized to the level at the control at TSS. A fluctuating methylation density was observed in the TSS region in untreated control cells, featuring a peak of increased methylation at +65 to +80, flanked by two valleys of decreased methylation at −200 to −300 and +600 to +800 bp (Fig. 3). While the overall pattern of TSS DNA methylation was unchanged, a 4 h opioid treatment increased genome-wide TSS methylation. Morphine caused a 22 % increase in DNA methylation above the control level at the + 65 to +80 peak, followed by bBCM7 (17 %) and hBCM7 (11 %); (Inset in Fig. 3). The relative efficacy of these opioids to alter DNA promoter methylation mirrors their efficacy in their inhibition of cysteine uptake as previously reported by this lab, which also had an efficacy order of morphine > bBCM7 > hBCM7 [28]. Thus bBCM7 exhibits a higher efficacy for increasing TSS methylation than hBCM7 (p < 0.05).

Lastly, we investigated if there were any underlying opioid-mediated changes in the epigenetic status of enzymes and proteins involved in the transulfuration and glutathione synthesis pathway. We observed differences in DNA methylation levels of several genes from the transsulfuration pathway and the methionine cycle (Fig. 4). Both genic and intragenic methylation changes were observed. More importantly these changes observed in the epigenetic patterns were also correlated the gene expression changes we reported in our previous studies. For example, morphine and bBCM7 both induced elevated levels in the DNA methylation patterns in the promoter region of the gene coding for the excitatory amino acid transporter (EAAT3). This corresponded with our previous results wherein we reported that morphine and bBCM7 treatment was associated with decreased levels of EAAT3 transporter as measured using qPCR [8]. Similar association between changes in mRNA levels and the DNA methylation levels were also observed with other enzymes and proteins involved in the transulfuration pathway [8]. Thus, morphine, hBCM7 and bBCM7 produce changes in DNA methylation, which may affect metabolic pathways regulating GSH and SAM levels.

Epigenetic regulation not only allows a pluripotent cell to differentiate into alternative cell types containing the same genetic information, but it also provides a mechanism for plastic and adaptive responses to changes in the cellular environment arising from physiological substances, xenobiotics, or food, representing a form of genetic metadata. We previously outlined a possible metabolic link between redox and methylation, and a potential mechanism by which changes in redox status translate into epigenetic outcomes [14, 19] and lastly, the potential for nutritional factors to affect prenatal and postnatal epigenetic programming [10]. Redox and epigenetic changes have been implicated in early developmental and neurological diseases [13–15], as well as immunological [10, 16] and neurodegenerative disorders [17–19]. Recently, we showed that redox effects of food-derived peptides with opioid activity exert powerful control over DNA methylation [8]; however, a synthetic peptide constituted with a scrambled sequence of the BCM7, did not show any effect on the global DNA methylation (Additional file 2: Figure S3), acting as a negative control, not activating the mu opioid receptor and inducing downstream epigenetic changes. Thus, the current results are an extension and insightful investigation of our previous work, clarifying the epigenetic effects of food-derived opioids, especially milk-derived opioids.

Using SH-SY5Y human neuroblastoma cells, we found that bBCM7 and hBCM7, as well as the prototypical opiate agonist morphine, caused epigenetic as well as transcriptional changes that could be linked to pathways contributing to gastrointestinal disease and inflammation in the GI tract. Although, it is noteworthy to mention that the bBCM-7 released from A1-type of beta casein was far more potent in inducing these epigenetic changes as compared to the hBCM-7. It should be noted that the primary endpoint for the current study was not to model GI diseases in SH-SY5Y cells, but it is highly intriguing that the pathway analysis identified GI function and diseases to be the most highly affected. This result may not be surprising since we have previously showed that EAAT3 is present at high levels in human intestinal epithelial cells (Caco2 cells), and food-derived opioid peptides can inhibit EAAT3-mediated cysteine uptake in these cells, resulting in altered redox and methylation capacity [8]. Thus, these molecular effects of opiates on antioxidant levels and methylation status can occur in different cell types and can result in overlapping epigenetic consequences.

Several previous studies have indicated the use of SH-SY5Y cells as models for neuronal cells for research related to central nervous system [8] as well as enteric nervous system [29]. Furthermore, oxidative stress has recently been recognized as a significant factor in the pathogenesis of gastrointestinal complications [30], and treatment with GSH or the GSH released from glial cells resulted in alleviation of oxidative stress in neuronal cells [29] as well as gastrointestinal symptoms. Mu-opioid receptors are highly expressed in the gastrointestinal tract including in neurons of the enteric nervous system. Hence, although we did not use SH-SY5Y as a model for induction of gastrointestinal diseases, we observed that induction of oxidative stress resulted in epigenetic changes which corresponded to gastrointestinal diseases. These results provide a potential explanation for the observed high prevalence of GI dysfunction co-morbidity in patients with neurological disorders including autism spectrum disorders.

Both wheat- and milk-derived opiate peptides inhibit the uptake of cysteine, resulting in a decrease in GSH synthesis [8]. This decrease in antioxidant capacity leads to inhibition of the redox sensitive enzyme methionine synthase and a decrease in the availability of the methyl donor S-adenosylmethionine (SAM). Oxidative stress and decreased levels of SAM and GSH have been associated with autism by multiple studies [31–41]. The decrease in SAM translates into effects on global DNA methylation, with epigenetic consequences, which can contribute to neurological and neurodevelopmental disorders such as autism [42–47].

The changes in DNA methylation and gene expression we observed were associated with inflammation and gastrointestinal disease-related metabolic pathways. One way to interpret this result is a dysfunctional positive feed-back loop in which food-derived opiate peptides decrease cysteine uptake from the GI tract, causing inflammation and gastrointestinal disease which further limits the ability to take up cysteine and other important nutrients. Avoiding milk-derived BCM-7 might be beneficial to prevent this inflammation in sensitive individuals; however, supporting clinical evidence is not yet reported and hence any clinical extrapolation of these in vitro studies should be done with caution. In some, but not all studies, dietary intervention with a gluten-free casein-free diet has been shown to improve symptom presentation in autistic subjects [22, 48, 49]. BCM7 has been linked to delayed psychomotor development in infants [50]. Interestingly, a genetic variant in dipeptidylpeptidase IV (DPP-IV), the enzyme responsible for hydrolysis of BCM7, was found to be significantly overrepresented in autistic male subjects [51], and autoantibodies against DPP-IV are found in autistic subjects [52]. Thus diminished activity of DPP-IV may contribute to increased activity of BCM7 in autism. Our findings also highlight a difference in the efficacy of human vs. bovine-derived BCM7 in altering DNA promoter methylation and mRNA transcription inflammation. The ability of BCM7 to alter gene expression has been previously reported [12]. Here we show that bBCM7 acts more like morphine than hBCM7, in the both magnitude of DNA promoter methylation changes, a similar bias toward mRNA up-regulation and in the extent of overlapping discrete DMTs and DETs. Thus the epigenetic effects of bBCM7 may be quantitatively and qualitatively different from those of hBCM7, which could potentially also provide a molecular rationale for the recommendation of breastfeeding vs. formula feeding during early development.

Results from the current study indicate that epigenetic effects of milk-derived opiate, especially bovine BCM-7 peptides may possibly contribute to GI dysfunction and to development of inflammation. Thus eliminating these opioid peptides from the diet may reverse these effects in sensitive individuals, although future clinical studies will be required to further investigate this mechanism. While preliminary, our findings indicate a “gut-brain” epigenetic relationship.

The data set supporting the results of this article are included as additional files for supplementary Figure S1, S2 and Additional file 2: Figure S3 on the journal website.