The possible origin of miRNAs differentially regulated in leiomyoma progenitors

Background: Leiomyoma are the most common indication for hysterectomy in the world and have a strong economic impact on health care systems; many different mechanisms have been considered for their aetiology, such as inammation, dysregulated progenitor cells or different regulation of miRNAs. After performing a whole genome miRNA proling in progenitor cells (PCs) derived from healthy myometrium (MPCs) and from leyomioma (LPCs), only 15 miRNAs were identied as differentially expressed between MPCs and LPCs. Progenitor cells from Amniotic Fluid (AFPCs) are considered the most undifferentiated cells after the embryonic ones. Here we try to clarify if the miRNAs differently regulated between leiomyoma and myometrium cells arise as a conversion of MPCs along the differentiation process or if they may originate from a divergent cell commitment. To track the origin of the dysregulation, miRNA expression was analyzed in AFPCs (considered as surrogate for embryonic cells), MPCs and LPCs. MPCs, LPCs and AFPCs were isolated and subjected to whole genoma miRNA proling; the expression of the 15 miRNAs previously identied as differentially regulated in MPCs and LPCs was compared to that detected in AFPCs. Results: Clustering analysis sub-grouped the 15 miRNAs into 4 major clusters that converge to the KEGG pathways: Adherens junction, ECM-receptor interaction, TGFβ signaling and cell cycle. miRNAs are differentially regulated in MPCs and LPCs compared to AFPCs and 10/15 of them show statistically signicant variations between MPCs and LPCs. Conclusion(s): Our results seem to point that a linear physiological differentiation axis exists from AFPCS to MPCs that, under insults, pathologically continues toward LPCs. Our results linear AFPCS that, antibodies (Becton-Dickinson) HLA-DR, CD14, CD19, CD45, osteocytes,

Today it is note that a population of progenitor cells occurs in almost all human tissues [6]. They are referred to as mesenchymal stem cells and, according to the criteria by Dominici [7], must be plastic adherent, positive for CD73, CD90, CD105 and negative for HLA-DR, CD14, CD19, CD34, CD45 and able to differentiate towards oste-, chondro-and adipogenic lineages. While satisfying these criteria, mesenchymal cells deriving from adult tissues show tissue-speci c features that during differentiation become even more characterizing [8]. Among mesenchymal stem cells, amniotic uid progenitor cells (AFPCs) are of particular interest, since they express both adult and embryonic cell markers, indicating them as cells within an intermediate stage between embryonic and adult phenotype [9]. In a theoretical line of increasing differentiation, AFPSCs should be before PCs of myometrium and broids, having a lower degree of differentiation than latters. miRNAs are switchers able to modulate the cell fate by turning on/off speci c gene targets and their aberrant expression can proportionally affect these critical processes leading to various pathological outcomes [10,11]. We recently demonstrated that 15 miRNAs are differentially expressed between PCs from leyomioma and healthy myometrium4, but we still don't know the mechanisms underlying these differences and its origin. In particular, when do these alterations occur? Do they arise during differentiation from embryonic cells to PCs (ie, differential miRNA expression leads to different PCs in normal myometrium and broids, hypothesis A, Fig. 1a) or later during the differentiation process (ie, miRNA dysregulation affects PCs from myometrium initially and subsequently exacerbates thus determining the acquisition of PCs with broids' features, hypothesis B, Fig. 1b)? Since the use of embryonic cells to answer these research questions is forbidden, AFPSCs that maintain a lower degree of differentiation than PCs were used as surrogate of the embryonic ones, with the aim to follow the onset of these alterations during the differentiation process.

Results
MPCs, LPCs and AFPCs respectively isolated from healthy myometrial, broid tissue and amniotic uids met the three criteria established by the International Society for Cellular Therapy (ISCT) for the identi cation of human mesenchymal stem cells [7]. Cells were plastic-adherent in standard culture conditions, strongly positive for CD105, CD73, and CD90 and negative for CD45, CD34, CD14, CD19, and HLA-DR surface molecules and able to differentiate towards mesenchymal lineages in vitro (data not shown) [12][13][14].

Discussion
The expression of 15 miRNAs previously identi ed as differentially regulated between myometrial PCs (MPCs) and leiomyoma PCs (PCs) [5] was compared to that observed in Amniotic Fluid PCs (AFPCs) to evaluate if the dysregulation of the 15miRNAs occurs: i) as the result of an early commitment of undifferentiated cells (Fig. 1A) or ii) as an exacerbation of the physiological differentiation from AFPCs to MPCs (Fig. 1B). The regulation of the expression level of the miRNAs across the cell type analysed does not allow giving unique suggestions.
Firstly, cells isolated from healthy and brotic uterine tissues as well as from amniotic uids were characterized according to the minimal criteria established by the International Society for Cellular Therapy (ISCT) for the identi cation of human mesenchymal stem cells7and were able to satisfy them.
After characterization, the rst issue to address was the validation of the choice of AFPCs as the most undifferentiated cells to be used as a surrogate of the embryonic ones [9,15]. Consistent with the fact that AFPCs are the most undifferentiated cells among the three cell types analyzed, they cluster separately from MPCs and LPCs that are instead closer to each other.
The heat map and the histogram show that the expression of the majority of 15 miRNAs is very different between AFPCs and LPCs/MPCs (mean of absolute FC 2.6 ± 2.8). Clustering analysis followed by DIANA mir-Path investigation allowed identifying four KEGG pathways related to the miRNA-clusters de ned based on expression: Adherens junction, ECM-receptor interaction, TGFβ signaling and cell cycle.
To address the question of which hypothesis was more consistent (A, early commitment; B, exacerbation), the trend of expression across the cell types of each identi ed pathway was evaluated.
Only adherens junction and TGFβ signaling pathways show a step-wise manner along the differentiation process.
To investigate what is the most probable hypothesis explaining the origin of LPCs, (A, early commitment; B, exacerbation), we evaluated in details the pattern of miRNA expression across the three cell-types. Only the 4 miRNA linked to regulation of adherens junction and one liked to TGFβ signaling pathways (miR-122-5p) show a progressive decreased or increased expression (respectively) from AFPCs to LPCs, ie along the hypothetical differentiation process.
Interestingly, LPCs expressed the lowest values of miRNAs related to adherens junction pathway and, most-likely as consequence, also the highest level of targeted proteins. As expected, the expression of adhesion molecules is higher in connective tissues where broblasts/myo broblasts develop strong bounds to ECM than in AFPCs [16]. While the increase in the expression of proteins related to adherens junction from AFPCs to MPCs has to be considered physiological, its further weighting towards LPCs may be connected to the leiomyoma onset [17,18].
Notably MED12, implicated as an oncogene in about 70% of uterine leiomyoma [19,20], is a target of hsa miR-10a-5p, whose expression is gradually reduced from AFPCS up to LPCs, possibly explaining MED12 upregulation in leiomyoma.
While the expression of the three miRNAs converging in TGFβ signalling pathway (has-miR-122-5p; has-miR-576-3p; has-miR-595) is weakly increased in MPCs compared to AFPCs, they are strongly upregulated in LPCs. The TGFβ pathway has been reported as deeply dysregulated in leiomyoma [21][22][23]; the genes targeted by these miRNAs belong to a cascade that involves SMAD 2 and SMAD4 and leads to apoptosis and G1 cell cycle arrest; the high expression of the selected miRNAs in LPCs may intervene by suppressing these proteins. This is in line with previous researches [23] reporting that leiomyomas are refractory to the antiproliferative effects of TGFβ1 and TGFβ3 observed in normal myometrium.
For the other two pathways, the expression of miRNAs across the cell type does not show a linear trend that could reinforce the hypothesis of exacerbate differentiation from MPCs to LPCs. Previous studies about miRNA pro ling from tissues samples of broid and myometrium identi ed a list of involved pathways that encloses the four here considered [24].
Interestingly, even if our pathways are enclosed in the list, the identi ed miRNAs are different.
It is not surprising that more miRNAs target the same genes but that the same pathways have been found differentially regulated in progenitor as well as in differentiated cells con rms the involvement of the selected pathways in the onset of leiomyoma. Indirectly, it also strengthens the hypothesis that the modulation of the expression of miRNAs during differentiation is a key mechanism in the pathogenesis of leiomyoma.

Conclusion
In conclusion, the regulation of miRNA expression across the cell type analysed does not lead to a unique conclusion: for adherens junction and TGFβ signalling pathways, it seems that the differences previously detected between leiomyoma and myometrium cells could arise in step-wise manner along the differentiation process from AFPCs, whereas results from cell cycle and ECM-receptor interactions pathways suggest that the different expression of miRNAs may originate from a divergent cell commitment. Nevertheless, by combining our results with others previously reported, data seem to point toward hypothesis B, namely that a linear physiological differentiation axis exists from AFPCS to MPCs that, under particular insults, pathologically continues toward LPCs.
These results may support and integrate previous hypotheses about leiomyoma development: miRNAs play a pivotal role acting directly at the rst steps of PCs differentiation and, as last hypothesis, creating the observed (epigenetic) predisposition.

Tissue collection
Amniotic uid (AF) samples (n° 9), 3 ml, were obtained from amniocentesis after the 16th week of pregnancy for routine prenatal diagnosis. The gestation age (GA) was determined by ultrasonic biparietal diameter and femur length measurements of the foetus. AF was collected by ultrasound-guided transabdominal puncture. The indications were advanced maternal age (34-36) and the cytogenetic analyses revealed normal karyotypes.
All patients provided their written informed consent to participate in the study, which was approved by the institutional ethics committees (2016-0360OR) and was conducted in accordance with the Declaration of Helsinki.
Cells were isolated, cultured and characterized as previously described [3,25,26]. After samples collection, cells were characterized by testing the minimal criteria identi ed by Dominici [7] for mesenchymal de nition.
Osteogenic differentiation was assessed by Von Kossa and Alkaline phosphatase (ALP) stainings; adipogenic differentiation was tested by Oil Red staining.
For chondrogenesis, cells were cultured in pellet culture system and sections were exposed to a solution of Safranin-O [27].

Bioinformatics analysis
Starting from raw FASTQ les, quality of reads obtained from each sample was assessed using FastQC software (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/), adapters were trimmed and reads with length < 18bp or > 27bp were ltered out with cutadapt (1.16). Filtered reads were aligned to the miRBase 22 using SHRiMP2 (2.2.3). Differential miRNA expression analysis between LPCs, MPCs and AF-MSCs was performed with DESeq2 (1.16.1). The expression of microRNAs was normalized as CPM (counts per million reads mapped) and subsequently the 15 miRNAs previously identi ed as differentially expressed between MPCs and LPCs were clustered by an unsupervised hierarchical clustering using spearman rank correlation and the average linkage method.

miRNAs targets analysis
To identify molecular pathways potentially altered by the expression of single or multiple miRNAs, Diana mir-Path Software was used [28]. This web-based application performs enrichment analysis of numerous miRNA target genes comparing each set of miRNA targets to all known KEGG (Kyoto Encyclopedia of Genes and Genomes, Kyoto, Japan) pathways.
The input dataset enrichment in each KEGG pathway is represented by the negative natural logarithm of the P value (− lnP).    Heat map and expression fold change of 15 miRNAs previously identi ed as differentially expressed between MPCs and LPCs based on next generation sequencing analysis. (A) Each row represents a different miRNA, and each column represents one sample from the AFPC, MPC or the LPC RNA pool of 12 specimens. Clusters were obtained by unsupervised hierarchical clustering using spearman rank correlation and the average linkage method. Colour key illustrates normalized expression levels (z-scores of normalized counts -cpm) of miRNAs across all samples. Gradient from yellow to blue indicates higher to lower expression. In the right column, the KEGG pathways identi ed by DIANA-miRPath software as targets of the analysed miRNAs. In brackets the number of targeted genes per pathways. (B) Expression fold changes (log2-transformed) of the 15 selected miRNAs in MPCs and LPCs as compared to AFPCs. adjusted P-value <0.05: *** both comparisons, ** only LPCs, * only MSCs.