Exploration of microRNAs in porcine milk exosomes

Background Breast milk contains complex nutrients and facilitates the maturation of various biological systems in infants. Exosomes, membranous vesicles of endocytic origin found in different body fluids such as milk, can mediate intercellular communication. We hypothesized that microRNAs (miRNAs), a class of non-coding small RNAs of 18–25 nt which are known to be packaged in exosomes of human, bovine and porcine milk, may play important roles in the development of piglets. Results In this study, exosomes of approximately 100 nm in diameter were isolated from porcine milk through serial centrifugation and ultracentrifugation procedures. Total RNA was extracted from exosomes, and 5S ribosomal RNA was found to be the major RNA component. Solexa sequencing showed a total of 491 miRNAs, including 176 known miRNAs and 315 novel mature miRNAs (representing 366 pre-miRNAs), which were distributed among 30 clusters and 35 families, and two predicted novel miRNAs were verified targeting 3’UTR of IGF-1R by luciferase assay. Interestingly, we observed that three miRNAs (ssc-let-7e, ssc-miR-27a, and ssc-miR-30a) could be generated from miRNA-offset RNAs (moRNAs). The top 10 miRNAs accounted for 74.5% (67,154 counts) of total counts, which were predicted to target 2,333 genes by RNAhybrid software. Gene Ontology and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses using DAVID bioinformatics resources indicated that the identified miRNAs targeted genes enriched in transcription, immunity and metabolism processes, and 14 of the top 20 miRNAs possibly participate in regulation of the IgA immune network. Conclusions Our findings suggest that porcine milk exosomes contain a large number of miRNAs, which potentially play an important role in information transfer from sow milk to piglets. The predicted miRNAs of porcine milk exosomes in this study provide a basis for future biochemical and biophysical function studies. Electronic supplementary material The online version of this article (doi:10.1186/1471-2164-15-100) contains supplementary material, which is available to authorized users.


Background
Milk, as the sole source of nutrition for infants, contains a potent mixture of diverse components such as milk fat globules (MFG) [1], immune competent cells and soluble proteins, for instance IgA, cytokines and antimicrobial peptides [2], which can provide protection against infections in newborns [3]. In addition, breast milk may have a role in tolerance induction [1] and may protect infants from developing allergies [4].
MiRNAs are widespread among eukaryotes and represent key components of a conserved system of RNA-based gene regulation [30][31][32][33]. Many studies have demonstrated that miRNAs are key post-transcriptional regulators of gene expression and play important roles in a wide range of physiological and pathological processes [34], including development, differentiation, proliferation and immune responses. It is believed that about 60% of mammalian genes are regulated by miRNAs [35][36][37][38][39].
Aside from being important farm livestock, pigs are also model animals for medical research. In the present study, we investigated miRNAs in milk exosomes of Landrace pigs in order to provide new information for investigations into the physiological functions of porcine milk.

Milk samples
Porcine milk samples were collected between day 1 to 5 after parturition from healthy lactating Landrace female pigs bred in the breeding farm of the Livestock Research Institute (Guangzhou, China). Milk samples were frozen immediately after milking and were kept at-80°C until use.

Preparation of exosomes from milk
Porcine milk samples were centrifuged first at 2,000 × g for 30 min at 4°C to remove MFGs as well as mammary glandderived cells. Defatted samples were then subjected to centrifugations at 4°C for 30 min at 12,000 × g to remove residual MFGs, casein and other debris. Subsequently, from the final supernatant (so-called whey or milk serum), the membrane fraction was prepared by ultracentrifugation at 110,000 × g for 2 h in an SW41T rotor (Beckman Coulter Instruments, Fullerton, CA) [40].

Transmission electron microscopy (TEM)
The final fraction obtained as described above was diluted with 0.01 M PBS and ultracentrifuged again to recover microvesicles as pellets. Following fixation in 2% glutaraldehyde, microvesicles were negatively stained with uranyl acetate and observed by TEM (JEOL JEM2000EX, Tokyo, Japan).

RNA isolation and Solexa sequencing
Total RNA was isolated from samples collected after ultracentrifugation using Trizol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. The quality of RNA was examined by 2% agarose gel electrophoresis and with a Biophotometer 6131 (Eppendorf, Germany), as well as further confirmed by using a Bioanalyzer (Agilent Technologies, Santa Clara, CA). Small RNAs (18-30 nt) were obtained from the total RNA, 5' and 3' adaptors were ligated to the small RNAs, then the adaptor-ligated RNAs were subsequently transcribed into cDNA by RT-PCR, and the samples were amplified by PCR using primers complementary to the two adaptors. The PCR products were purified and subjected to Solexa sequencing (Illumina, CA) at the Beijing Genomics Institute (BGI, Shenzhen, China).

Sequence data analysis
The raw reads obtained from Solexa sequencing were processed to obtain clean reads by summarizing data production, evaluating sequencing quality, calculating the length distribution of small RNA reads, removing low quality reads and adaptor sequences as described in previous paper [41]. All the clean reads were aligned against non-coding RNAs from the GenBank and Rfam (11.0) (ftp.sanger.ac.uk/pub/databases/Rfam) database to annotate and classify rRNA, tRNA, snRNA and other ncRNA sequences using tag2 annotation software (developed by BGI). Then the selected sequences were mapped to the pig genome (sscrofa9, www.ensembl.org/Sus_scrofa/) using SOAPv1.11 software [42] to analyze their expression and distribution. Subsequently, the miRNA candidates were further analyzed by miRDeep 2 against all known miRNAs and porcine miRNA precursors (miRBase 20.0). All remaining candidates who did not map to any miRNAs in miRBase 20.0 were considered as potential novel miRNAs. To further identify these potential novel miRNA candidates, software MIREAPv0.2 (http://sourceforge.net/projects/mireap) [43] developed by BGI was used to predict novel miRNA by exploring the secondary structure, the Dicer cleavage site and the minimum free energy of the annotated small RNAs which could be mapped to genome. In briefly, the sequence length should be between 18-26 nt, maximal free energy allowed for a miRNA precursor was −18 kcal/mol, maximal space between miRNA and miRNA* was 35 nt, and flank sequence length of miRNA precursor should be 10 nt. Finally, all remaining novel miRNA candidates were further subjected to MiPred (http://www.bioinf.seu.edu.cn/miRNA/) to filter out pseudo-pre-miRNAs. The minimum free energy must be > −20 kcal/mol or P-value was >0.05 [44], and their secondary structures were also checked using the Mfold3.2 software [45]. All data for analysis in this study have been deposited in https://mynotebook.labarchives. com/share/allinchen/MTkuNXwxMzMxMS8xNS0yL1R yZWVOb2RlLzE1NzEyODU2fDQ5LjU= with a DOI:10. 6070/H4DN432G.

PCR and qRT-PCR identification of known and novel miRNAs
Total RNA (identical sample to that of the Solexa sequencing sample) was first digested with DNase I (Invitrogen), and 2 μg of total RNA was reverse transcribed to poly (A) tail-added cDNA using the One Step Prime-Script miRNA cDNA Synthesis Kit (TaKaRa, Dalian) according to the manufacturer's instructions. Briefly, a 60-nt adaptor containing a poly (A) structure was added to the 3' sequence of miRNAs, which were then reverse transcribed to an 80-bp cDNA sequence [46]. The cDNA was diluted 5-fold with ddH 2 O, and PCR was performed on a Bio-Rad system (BIO-RAD,USA )in a final 20 μL volume reaction containing 2 μl PCR cDNA, 10 μL of 2× PCR Mix (Toyobo, Osaka, Japan) and 1 mM of each primer. The real-time PCR thermal profile was as follows: 5 min at 95°C, 40 cycles of 30 s at 94°C, 30 s at the corresponding annealing temperature (Tm) and 72°C for 30 s, followed by 72°C at 10 min. PCR products were examined on an agarose gel to confirm that a single PCR product was generated, and 5S ribosomal RNA was used as an internal control for the PCR. The miRNA forward primer was designed with Primer 5.0 (Table 1), and the reverse primer for miRNAs was the Uni-miR qPCR Primer offered by the kit One Step PrimeScript miRNA cDNA Synthesis Kit (TaKaRa, Dalian).

Bioinformatics analysis Chromosomal localization and cluster analysis of miRNAs
Pre-miRNAs of all miRNAs (known miRNAs and novel miRNAs) were mapped to the porcine genome (sscrofa9, www.ensembl.org/Sus_scrofa/) according to their positions on the chromosomes. Pre-miRNA positions less than 10 kb apart were considered to belong to the same miRNA cluster.

Target prediction and Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses
Porcine miRNA targets were obtained at the genome level. In brief, miRNA targets were predicted using the RNAhybrid software algorithm (http://bibiserv.techfak. uni-bielefeld.de/rnahybrid/) in 3'-UTR sequences of transcripts from the whole pig genome obtained from Ensembl Gene 66 database (sscrofa9, www.ensembl.org/ Sus_scrofa/). Strict criteria (perfect match of 2-8 nt in the seed region; no more than −25 kcal/mol low free energy of miRNAs-mRNA binding) were applied to the target prediction procedure. GO and KEGG pathway analyses were performed using DAVID bioinformatics resources (http://david.abcc.ncifcrf.gov/).

Identification of exosomes
Exosomes were obtained from porcine milk by ultracentrifugation. After negative staining, approximately round-shaped porcine milk exosomes with diameters of~50-100 nm were observed by TEM, showing a greater density at the center than at the boundary ( Figure 2A, B).

Porcine milk exosomes contain RNA
To further ascertain whether the porcine milk exosomes collected by ultracentrifugation contained RNA, we extracted the samples using Trizol reagent and then examined the recovered product by electrophoresis on a 2% agarose gel. To exclude the possibility of DNA contamination, total RNA was incubated at 37°C for 30 min with DNase I or RNase. The extracted RNA could not be digested by DNase I, while it could be degraded by RNase ( Figure 3A). These results demonstrated that the porcine milk exosomes contained RNA. More interestingly, total RNA of porcine milk exosomes were enriched with 5S rRNA ( Figure 3B), consistent with previous studies [27,28,47].

Solexa sequencing and analysis Solexa sequencing
The sRNAs were enriched from porcine exosomes to construct a library for Solexa sequencing. We obtained 9,033,167 raw reads and 6,013,724 high qualities reads after removal of low quality and contaminant reads. Among the high quality reads, there were 4,964,542 clean reads (82.55%), representing 1,691,655 unique sRNAs. The majority of the sRNAs in porcine milk exosomes were 18-25 nt in length (74.89%, Figure 4), with 2,458,894 reads (49.53%) representing 872,096 unique sRNAs (51.55%), including miRNAs and other sRNAs, such as rRNA, tRNA, snRNA, snoRNA, scRNA, small recognition particle RNA (srpRNA), repetitive sequence elements and unannotated sequences,  which could be mapped to the pig genome. BLAST searching with the miRbase 20.0, identified a total of 230,216 reads, representing 1,555 unique known miRNAs. Due to RNA bias editing, 5' modifications and 3' modifications, many pre-miRNAs produce multiple mature miRNA isoforms, namely isomiRNAs, as described in many studies [48][49][50]. In the subsequent analysis, all isomiRNAs generated from the same precursor were considered one type of miRNA. Consequently, these 1,555 unique miRNAs corresponded to 176 known mature miRNAs (205 pre-miRNAs, all the detail were listed in Additional file 1: Table S1). In addition, we identified 315 novel mature miRNAs (generated from 366 pre-miRNAs, detail in Additional file 2: Table S2). Among the 315 novel miRNAs, 18 have not been deposited as porcine miRNAs in miRbase 20.0, but had very high similarity with miRNA sequences of other species. These 18 miRNAs are labeled "PS" (Table 2), while 298 miRNAs that have not been deposited in miRbase 20.0 for any organism are labeled as "PC" and presented in Additional file 2: Table S2. There were 73 miRNAs with more than 100 counts and 264 miRNAs with less than 10 counts. The top 10, top 20, top 50 and top 100 miRNAs accounted for  "sub", "delete", "add" represents nucleotide substitution, deletion, addition (at 5-end of miRNAs), respectively perfect stands for perfect match with reference miRNAs.

Identification of miRNAs by PCR and sequencing
To verify the deep sequencing results, we selected randomly 13 known miRNAs and 15 predicted novel miRNAs for PCR amplification ( Figure 6A, B). Subsequently, the 15 newly predicted miRNAs were cloned and sequenced. The results showed that 8 sequences were fully matched, while 7 sequences had one or two mismatched nucleotides (Table 3). However their seed sequences remained unchanged. Simultaneously, the abundance of some novel miRNAs predicted by Solexa sequencing was confirmed by quantitative real-time PCR. The abundance of most miRNAs observed by qPCR of the sample pool and by sequencing were generally consistent (Figure 7).
Target verification of miR-PC-86 and miR-PC-263 against 3'UTR of IGF-1R using luciferase report assay To investigate whether the predicted miR-PC-86 and miR-PC-263 ( Figure 1) were functional novel miRNAs, target genes were predicted, and miR-PC-86/ miR-PC-263 were found to directly target IGF-1R 3'UTR sequence. The full-length 3'UTR of IGF-1R mRNA was inserted downstream of the luciferase gene in the pmir-GLO Dual-Luciferase miRNA Target Expression Vector reporter plasmid, and the seed sequence was also delete to disrupt miR-PC-86/ miR-PC-263 binding ( Figure 1B). The wild-type (pGLO-IGF-1R-3'UTR) or delete (pGLO-IGF-1R-3'UTR-delete) plasmid was co-transfected with the miR-PC-86 and miR-PC-263 mimics into IPEC-J2 cells. Forty-eight hours after transfection, the luciferase activity of the miR-PC-86 and miR-PC-263 group were significantly lower than that of the NC group (P < 0.05) respectively, and the reduction was rescued in the delete group ( Figure 1C). Thus, IGF-1R was initially confirmed as the target of miR-PC-86 and miR-PC-263.

Mature miRNAs
It is well accepted that only one of two strands generated from a precursor is preferentially incorporated into RNA-induced silencing complexes (RISC), whereas the complementary strand (miR*) may be degraded. Closer examination of mature miRNAs generated from pre-miRNAs showed that precursor miRNAs could be divided into three groups (Table 4): pre-miRNAs only with the left-arm sequence (5p), pre-miRNAs only with the right-arm sequence or both. Most pre-miRNAs seemed to be single-arm miRNAs (5p or 3p), while 50 pre-miRNAs possessed both 5p and 3p sequences (40 coupled mature miRNAs, all details were listed in Additional file 3: Table S3). Further analysis of the 40 coupled mature miRNAs indicated that most of the pre-miRNAs had no  significant difference in abundance between the 5p-arm and 3p-arm sequences (Figure 9). Some miRNAs had different expression levels between the two strands. For example, ssc-miR-193a-3p had 67,154 counts, ranking first among all miRNAs, while ssc-miR-193a-5p had 2,538 counts. Conversely, miR-423-5p had 22,588 counts, while its complementary strand, miR-423-3p, had only 654 reads, and which was shared by miR-22, miR-30a, miR-339, miR-17, miR-24, miR-331, miR-27b and let-7d (detail in Additional file 3: Table S3).
As described in other studies [51], some small RNAs were generated from the loop or the region between the loop and stem (ssc-let-7e, ssc-miR-27a and ssc-miR-30a) ( Figure 10A-C). In addition, we detected another interesting type of small RNAs known as miRNA-offset RNAs, or "moRNAs", which are derived from the ends of pre-miRNAs, predominantly from the 5' end, independent of the mature miRNA. A good example of moRNAs and small RNAs generated from the loop was pre-miR-30a ( Figure 10C). At the 5' end of pre-miR-30, a 18 nt RNA sequence was found to be generated from the loop, downstream of ssc-miR-30a-5p ( Figure 10C). The findings suggest that these RNAs may be only byproducts of Drosha and Dicer processing, or these small RNAs alternatively may take part in other important regulatory functions different from those of miRNAs.

MiRNA families
It is widely believed that the members of a given miRNA family regulate very similar sets of target genes. Apart from miRNA clusters, miRNA families were also recognized in the miRNAs of exosomes. Based on seed sequences, 35 miRNA families were identified (26 known and 9 novel miRNA families) to contain at least two members and the identification of novel miRNAs added  new members to 5 known families (Table 6). In our study, 8 miRNA families (let-7, mir-1, mir-17, mir-181, mir-148, mir-30, mir-92 and mir-99) were found with at least 3 members among all exosome miRNAs. The let-7 family had 9 members, miR-181 family had 4 members (miR-181a/b/c/d) and miR-30 family had 5 members (miR-30a/b/c/d/e). Most importantly, these miRNAs were highly expressed, and let-7 family members (let-7a/ f/c), miR-181a and miR-30a-3p were enriched among the top 10 miRNAs. However, members in the same family were highly differentially expressed. In the miR-181 family, miR-181a and miR-181b were dominant types with 13,345 reads and 3,333 reads, respectively. Similarly, miR-30a was the most abundant in the miR-30 family. The differential expression of members in the same family may be partly due to regulation of their precursors [52]. On the other hand, combined with the cluster analysis, we also observed that some miRNAs shared not only the same cluster but also the same families. These miRNAs included 181a/b, let-7f/miR-98, 181c/d, let-7a/f-5p/d-5p, 30b/d, 30c/e and miR-221/222. More interestingly, family members of the same cluster seemed to share expression patterns ( Figure 12A-E).
As mentioned above, miR-17-5p, miR-363, miR-106a, miR-18a, miR-19b, miR-92a, miR-20b and miR-92b formed a complex cluster and family network, and they also showed different expression patterns. MiR-92a, miR-19b and miR-363 were found to be highly expressed, while miR-17-5p, miR-18a, miR-20b and miR-106a were lowly expressed. The difference in abundance of the homologous or paralogous clusters may be attributed to the copy number of miRNA precursor itself or to the post-transcriptional regulation of the process of generating a mature miRNA from the precursor miRNA.
In addition, many miRNA families showed low expression (count number <100) in milk exosomes, such as the miR-1, miR-130, miR-17, miR-10, miR-29, miR-374, mir-9, miR-15 and miR-491 families ( Figure 12F), which are routinely expressed in specific tissues [53][54][55][56]. Interestingly, all 9 novel miRNAs families showed extremely low expression levels ( Figure 12G, count number <50), indicating that these miRNAs may only be expressed in certain physiology processes and may be the reason for why these miRNAs have not been detected until now. MiR-148a was reported to be an important biomarker for milk exosome miRNAs [28,57]. In this study, three members of this family (miR-148a, miR-148b and The ssc-let-7e precursor produced a 5p-arm sequence, a 3p-arm miRNA sequence and a loop-derived small RNA. (C) The ssc-miR-30a precursor produced a 5p-arm sequence, a 3p-arm miRNA sequence and a loop-derived small RNA. In addition, ssc-moRNA-3, belonging to new type of miRNA termed moRNA, was found at the 5' end of pre-miR-30.
miR-152, Figure 12H) showed modest expression levels, suggesting that miR-148 may be a stably expressed miRNA in exosomes of most mammals including pigs.

GO and KEGG pathway analyses
To better understand the role of miRNAs in pig milk exosomes, potential targets of miRNAs at the pig genome level were explored by using our previous method [58]. We selected the top 10 miRNAs for target prediction using RNAhybrid, which found 2,333 potential transcripts. GO analysis showed that those targets were enriched in many processes (Table 7), including antigen processing and presentation, MHC class II protein complex and transcription. Pathways analysis revealed these targets were enriched in 7 KEGG pathways (Table 8), which are essential for piglet metabolism (glycerophospholipid metabolism and citrate cycle), immunity (cell adhesion molecules, asthma and intestinal immune network for IgA production, antigen processing and presentation) and development (Notch signaling pathway). The results suggest that these milk miRNAs likely take part in regulation of important protective functions in piglets, including the intestinal immune network for IgA production and antigen processing and presentation.

Discussion
In the present study, a comprehensive miRNA expression profile of porcine breast milk exosomes was explored via a deep sequencing approach. We found in total 176 known miRNAs (miRBase 20.0) and 366 pre-miRNAs producing 315 mature miRNAs. Luciferase reporter assay was used to explore the targets of two predicted novel miRNAs in this study. Results indicated both of them down-regulated the luciferase expression by targeting 3'UTR of IGF-1R. All these pre-miRNAs were distributed in 30 clusters (11 novel and 19 known clusters), and the mature miRNAs could be assigned to 35 families (26 known and 9 unknown families). GO and KEGG pathway analyses show that those miRNAs may participate in many different immune-related processes. An analysis of the top 20 miRNAs showed that 14 of them may be involved in many regulatory aspects of the IgA immune network.
A recent study of exosome miRNAs in Yorkshire sow milk discovered 180 pre-miRNAs, including 140 known porcine pre-miRNAs and 40 novel pre-miRNAs, which encode 237 mature miRNAs (234 unique miRNAs) [29] In the current study, we discovered 205 known porcine pre-miRNAs (176 mature miRNAs) and 366 novel pre-miRNAs (315 mature miRNAs), approximately 254 more mature miRNAs than were revealed in the former report [29]. Therefore, our results substantially supplement the known pig miRNAs, particularly milk exosome miRNAs. Interestingly, most of the novel miRNAs were low in abundance (312 miRNAs with less than 100 reads and only 3 miRNAs with >100 reads), which is possibly the reason for why these miRNAs were not detected in a previous study by Gu et al. [29]. Further comparison revealed that miR-191 and let-7a, which potentially play a vital role in immunity, were found in both that study and in the top 10 miRNAs of our study. Other miRNAs identified previously (miR-30a-5p, miR-25-3p, miR-182-5p, miR-200c-3p and miR-375-3p) were not detected in our study. Furthermore, miR-148a, a potential biomarker for quality control in bovine milk [57] and human milk [28], which was found to be highly expressed throughout the lactation period of Yorkshire sows [29], was only detected at a moderate level in Landrace pigs in our study. MiR-148a has been reported to be a tumor metastasis suppressor in gastric cancer [59], and ectopic expression of miR-148a was shown to induce apoptosis and silence Figure 11 MiR-363~20b~106a homologous or paralogous cluster and expression level. (A) The miR-363/92a/19b/20b cluster and its homologous cluster miR-363/92a/19b/20b/106a on chromosome X were separated by a 33.5 kb DNA fragment. (B) The miR-17/18a/19/92a cluster was located on chromosome 11. In the genome, miR-92a/19b showed three copies; miR-363 and miR-20b had two copies; while miR-17, miR-18a and miR-106a had one copy. (C) Expression of mature miRNAs produced from the miR-363~20b~106a cluster and miR-17~92 cluster.
Bcl-2 in colorectal cancer cells [60]. By bioinformatics analysis, miR-148a was determined to be possibly related to immunity and gastrointestinal health, but the underlying regulatory mechanism remains unclear.
IgA is a major immunoglobulin in milk [75]. Expression of the polymeric IgA receptor (pIgR) in mammary epithelial cells contribute much to the development of the immune system at the early stage of lactation [76]. In the present study, some miRNAs were predicted to target genes (CD40, SLA, SLA-DMA, SLA-DOA, SLA-DOB, SLA-DRA, SLA-DRB1 and TGFB3) involved in processes of the intestinal immune network for IgA production in porcine milk. CD40 is a B-cell antigen activated during immune responses [77]. CD40 and CD40 ligand (CD40L) expressed on activated T cells are essential to B cell proliferation [78] and secretion of IgG, IgA and IgE [79]. SLA Class I were found to be expressed in the epithelial and lamina propria cells of the intestine in adult pigs and to be involved in mother-newborn interactions [80]. A study in humans showed that TGF-β acts as a specific switch for IgA present at early stages of development of B cells [81].   In the present study, the top 20 miRNAs were used for IgA network analysis. APRIL was the predicted target of miR-193a-5p, which is essential to triggering IgA 2 class switch in human B cells. Intestinal epithelial cells (IECs) release APRIL after sensing bacteria through Toll-like receptors, and mucosal vaccines activate IECs to induce more effective IgA 2 responses [82]. The let7 family and miR-423-5p were predicted to target CCL25, a potent and selective chemoattractant for IgA antibody-secreting cells [83]. CCL25 is known to selectively modulate immune responses, specifically the localization of T lymphocytes to the small-intestinal mucosa [84]. CD80 and CD86, which are costimulators of T lymphocytes [85], were identified as possible targets of five miRNAs in our study. Let-7a, let-7c, miR-181b, miR-185, miR-378 and miR-423-5p were predicted to target the inducible costimulatory molecule (ICOS), which plays a key role in regulating T-cell differentiation, T-cell proliferation, and secretion of lymphokines, providing effective help for antibody secretion by B cells [86]. We hypothesize that some miRNAs identified here in porcine milk regulate IgA production in the intestine of piglets, which may play an important role in mucosa immunity. However, their regulatory mechanisms warrant further study.

Conclusions
In conclusion, the present study revealed 176 known miRNAs and 366 (315 mature miRNAs) novel pre-miRNAs in porcine milk, most of which were predicted to be involved in regulation of digestive tract development and immunity of newborn piglets. These findings contribute to an increased understanding of the roles of miRNAs in porcine (S. scrofa) milk exosomes and to building the foundation for understanding their physiological functions and regulatory mechanisms.