Two samples were collected from each patient: one from the central area of the leiomyoma and one from the unaffected myometrium. All the leiomyomas were confirmed histologically as benign ordinary leiomyomas. Samples were stored at −80°C until proteomic analysis was performed.

Clean leiomyoma and myometrium (300 mg each) were homogenized in 1.2 ml of dissolution TUC buffer [7 M urea, 2 M thiourea, 4% CHAPS, 40 mM Tris, 65 mM DTT and 0.24% Bio-Lyte (3–10)] with a protease inhibitor mix (2 mM PMSF, 1 mM benzamidine, 1 mM EDTA, 1 mM NaF) and a trace of bromophenol blue. The solutions were vortexed at maximum speed several times and kept at room temperature for 1 h and centrifuged at 10,000 × g at 4°C for 30 min. The protein content of the supernatant was determined using the Bradford assay. Eight hundred micrograms of proteins from each sample were used for the 2-DE analysis.

NL IPG Readystrips, 18-cm, pH 3-10 (Bio-Rad, Hercules, CA, USA) were rehydrated at 50 V for 12 h at 20°C. Isoelectric focusing (IEF) was performed in a Protein IEF cell (Bio-Rad) set to 170.000 Vh. After IEF, the IPG strips were equilibrated by serial incubation (20 min) in an equilibration buffer (6 M urea, 2% SDS, 50 mM Tris-HCl pH 8.8, 30% glycerol, and 1% DTT) and in an equilibration buffer containing 4% iodoacetamide instead of DTT. Equilibrated IPG strips were transferred onto a 12% polyacrylamide gel for the second dimension. After the second dimension, gels were fixed and stained for 48 h with colloidal Coomassie Blue, and excess dye was removed with distilled water. On average, three experimental replicates were performed per sample. Molecular masses were determined by precision protein standard markers (Bio-Rad). 2-DE gels were scanned with a Molecular Imager PharosFX system (Bio-Rad) and the quantitative analysis of the spots was carried out using the ProteomeWeaver 4 program (Bio-Rad).

Spot normalization was automatically performed by the software and was based on a normalization algorithm intended for numerical analysis, which did not require any internal standard. For each gel an intensity factor was computed to ensure all normalization factors were as close to one as possible. The matching produced a list of super spots, which represent certain protein species present in the gel. For the correct matching, each super spot was manually controlled prior to normalization. For each matched pair of gels, the quotient between the pair-matched spot was calculated. The normalization factor was the median of these quotients (ProteomeWeaver 4 program; Bio-Rad).

Spots from 2-DE were washed 4 times with 50 mM NH₄HCO₃ and acetonitrile (ACN; Sigma-Aldrich) alternatively and dried under vacuum in a SpeedVac system. Three microliters of 12.5 ng/µl sequencing grade modified trypsin (Promega, Madison, WI, USA) in 50 mM NH₄HCO₃ was added to each gel spot and samples were digested overnight at 37°C. Peptides were extracted with three changes of 50% ACN/0.1% formic acid (FA; Fluka), dried under vacuum and stored at −20°C until mass spectrometry (MS) analysis was performed.

Samples were dissolved in 10 µl of 0.1% trifluoroacetic acid (TFA; Riedel-de Haën). One microliter of each sample was mixed with 1 µl of matrix solution (α-cyano-4-hydroxycinnamic acid; Fluka). Five mg/ml in 70% ACN/0.1% TFA) and 0.8 µl of the resulting solution were spotted onto a stainless steel MALDI target plate for the MS analysis on the MALDI-TOF/TOF 4800 analyzer (AB Sciex, Framingham, MA, USA). The analysis was performed in a data dependent mode: a full MS scan was acquired from each sample, followed by MS/MS spectra of the ten most intense signals.

Few samples that could not be identified by MALDI-TOF/TOF analysis were further analyzed by LC-MS/MS on an LTQ-Orbitrap XL mass spectrometer (Thermo Fisher Scientific, Rockford, IL, USA), coupled with a nano-HPLC Ultimate 3000 (Dionex - Thermo Fisher Scientific). Samples were loaded onto an in-house pico-frit column packed with C18 material (ReproSil, 300Å, 3 µm; Dr. Maisch HPLC GmbH) and separated using a 20-min linear gradient of ACN/0.1% formic acid (from 3 to 40% ACN), at a flow rate of 250 nl/min. Capillary voltage was set at 1.3–1.5 kV and source temperature at 200°C. The analysis was performed in a data dependent mode, and a full scan at 60,000 resolution on the Orbitrap was followed by MS/MS fragmentation scans on the ten most intense ions acquired with CID fragmentation in the linear trap.

MS and MS/MS spectra obtained from MALDI-TOF/TOF analysis were converted into MGF (Mascot generic format) files to be elaborated with Proteome Discoverer 1.4 (Thermo Fisher Scientific), while raw data files from the LTQ-Orbitrap XL mass spectrometer were directly analyzed with the software. Proteome Discoverer was interfaced to a Mascot search engine, version 2.2.4 (Matrix Science, London, UK).

The database used for protein identification was UniProt Human (version 20140709, 88993 sequences), while enzyme specificity was set to trypsin with 1 missed cleavage. The mass tolerance window was set to 10 ppm for parent mass and to 0.6 Da for fragment ions for the files from LTQ-Orbitrap XL, while the tolerances were 50 ppm (parent) and 0.3 Da (fragment ions) for the MALDI-TOF/TOF data. Carbamidomethylation of cysteine residues was set to 'fixed modification' and methionine oxidation to 'variable modification'.

Proteome Discoverer calculates a false discovery rate (FDR) based on the parallel search against a randomized database. Proteins were considered as positive hits if at least two independent peptides were identified with medium (95%) or high (99%) confidence.

Possible connections among identified cytoskeletal proteins with significant variations as compared to normal myometrium versus leiomyoma were analyzed by a protein and gene network software. For each protein, related gene names were acquired in UniProtKB and used for network generation by the use of STRING 9.0 (http://www.string-db.org/).

Differential proteins distributed in the biological function data were obtained from Gene Ontology (http://amigo.geneon-tology.org/rte).

Western blot analysis was performed as previously described (12). Protein extracts (30 µg) used for 2-DE were separated by 10% SDS-PAGE, and then transferred to a nitrocellulose membrane in a blotting chamber. The residual binding sites on the membrane were blocked by treatment with defatted dry milk proteins, and incubated overnight at 4°C with 1:1,000 diluted primary rabbit polyclonal antibody against myosin regulatory light polypeptide 9 (Sigma-Aldrich). After washing, membranes were incubated with an HRP-conjugated anti-rabbit IgG (Sigma-Aldrich) in a dilution of 1:3,000. At the end the protein expression was visualized by chemiluminescence (SuperSignal West Pico Chemiluminescent; Thermo Fisher Scientific) according to manufacturer's instructions. The intensity of the signals was quantified by VersaDoc Imaging system (Bio-Rad). The intensities of the immunostained bands were normalized with the total protein intensities measured by Coomassie brilliant blue G-250 from the same blot.

Statistical analyses were carried out with the non-parametric Wilcoxon sign-rank test for paired samples for both 2-DE and western blot data. A p-value <0.05 was considered as statistically significant. Analyses were conducted with Stata/IC 12.2 for Windows (StataCorp LP, College Station, TX, USA).

Using the 2-DE coupled with MS, a comparative proteomics analysis was performed between uterine leiomyoma and myometrium tissues. Correlation analysis of gel-pairs performed well, with average matching efficiency of about 75%. An average of 2,200 spots was detected on gels for both types of proteomes. In this study 10 protein spots, belonging to cytoskeletal proteins with several biological functions, were found to be significantly dysregulated in leiomyoma samples compared to the myometrium. Eight spots were significantly upregulated (>1.5-fold) and two were significantly downregulated (<0.6 fold) (Fig. 1), and corresponded to 10 proteins identified by MALDI-TOF/TOF and LTQ-Orbitrap XL by searching the MS/MS data against the human section of the UniProt database (Table I). We found a significant increase in the leiomyoma of tubulin β (TUBB), myosin regulatory light polypeptide 9 (MYL9), desmin (DES), four and a half LIM domains protein 1 (FHL1), keratin, type II cytoskeletal 1 (KRT1), keratin, type I cytoskeletal 9 (KRT9), LIM and SH3 domain protein 1 fragment (LASP1), actin α cardiac muscle 1 (ACTC1). Transgelin (TAGLN), prelamin A/C (LMNA) were found to be significantly downregulated in the leiomyoma compared to the myometrium.

Three of these proteins are involved in the promotion of cell migration (FHL1, LASP1 and MYL9), while TAGLN is involved in replicative senescence.

In this study we decided to validate myosin regulatory light polypeptide 9 (MYL9) because it is considered to be a cell migration marker, in several cancers and could play a key role in leiomyoma development. MYL9 expression in seven leiomyomas was compared to the expression in matched normal myometrial tissue samples by western blot analysis. MYL9 expression was significantly higher in the leiomyoma with respect to the myometrium, confirming results obtained from the 2-DE analysis. In Fig. 2 we report the quantitative analysis of MYL9 expression (P=0.031). We opted for a Coomassie staining protocol because, as described by Lv et al (13), and according to our results, the two housekeeping proteins, β-actin and tubulin, are upregulated in leiomyoma, and thus not adequate as controls. We do not have information on the expression of the other housekeeping proteins.

The dysregulated cytoskeletal related proteins identified in this study were loaded on STRING 9.0 to generate a prediction of protein-protein interaction networks. The strongest interactions were between: LASP1 and ACTC1 (combined score, 0.774), and KRT1 and KRT9 (combined score, 0.76) (Fig. 3A). We further analyzed the interaction between transforming protein RhoA (a key protein in the regulation of the signal transduction pathway involved in leiomyoma growth) and the proteins identified in our study (Fig. 3B).

The strongest evidence was the interaction between RHOA and MYL9 (combined score, 0.99). Another interesting result based on action prediction was the activation of MYL9 by RHOA (Fig. 3C).