Prolactin Variants in Human Pituitaries and Pituitary Adenomas Identified With Two-Dimensional Gel Electrophoresis and Mass Spectrometry

Human prolactin (hPRL) plays multiple roles in growth, metabolism, development, reproduction, and immunoregulation, which is an important protein synthesized in a pituitary. Two-dimensional gel electrophoresis (2DE) is an effective method in identity of protein variants for in-depth insight into functions of that protein. 2DE, 2DE-based PRL-immunoblot, mass spectrometry, and bioinformatics were used to analyze hPRL variants in human normal (control; n = 8) pituitaries and in five subtypes of pituitary adenomas [NF− (n = 3)-, FSH+ (n = 3)-, LH+ (n = 3)-, FSH+/LH+ (n = 3)-, and PRL+ (n = 3)-adenomas]. Six hPRL variants were identified with different isoelectric point (pI)-relative molecular mass (Mr) distribution on a 2DE pattern, including variants V1 (pI 6.1; 26.0 kDa), V2 (pI 6.3; 26.4 kDa), V3 (pI 6.3; 27.9 kDa), V4 (pI 6.5; 26.1 kDa), V5 (pI 6.8; 25.9 kDa), and V6 (pI 6.7; 25.9 kDa). Compared to controls, except for variants V2-V6 in PRL-adenomas, V2 in FSH+-adenomas, and V3 in NF−-adenomas, the other PRL variants were significantly downregulated in each subtype of pituitary adenomas. Moreover, the pattern of those six PRL variants was significantly different among five subtypes of pituitary adenomas relative to control pituitaries. Different hPRL variants might be involved in different types of PRL receptor-signaling pathways in a given condition. Those findings clearly revealed the existence of six hPRL variants in human pituitaries, and the pattern changes of six hPRL variants among different subtypes of pituitary adenomas, which provide novel clues to further study the functions, and mechanisms of action, of hPRL in human pituitary and in PRL-related diseases, and the potential clinical value in pituitary adenomas.

Human prolactin (hPRL) plays multiple roles in growth, metabolism, development, reproduction, and immunoregulation, which is an important protein synthesized in a pituitary. Two-dimensional gel electrophoresis (2DE) is an effective method in identity of protein variants for in-depth insight into functions of that protein. 2DE, 2DE-based PRL-immunoblot, mass spectrometry, and bioinformatics were used to analyze hPRL variants in human normal (control; n = 8) pituitaries and in five subtypes of pituitary adenomas [NF − (n = 3)-, FSH + (n = 3)-, LH + (n = 3)-, FSH + /LH + (n = 3)-, and PRL + (n = 3)-adenomas]. Six hPRL variants were identified with different isoelectric point (pI)relative molecular mass (M r ) distribution on a 2DE pattern, including variants V1 (pI 6.1; 26.0 kDa), V2 (pI 6.3; 26.4 kDa), V3 (pI 6.3; 27.9 kDa), V4 (pI 6.5; 26.1 kDa), V5 (pI 6.8; 25.9 kDa), and V6 (pI 6.7; 25.9 kDa). Compared to controls, except for variants V2-V6 in PRL-adenomas, V2 in FSH + -adenomas, and V3 in NF − -adenomas, the other PRL variants were significantly downregulated in each subtype of pituitary adenomas. Moreover, the pattern of those six PRL variants was significantly different among five subtypes of pituitary adenomas relative to control pituitaries. Different hPRL variants might be involved in different types of PRL receptor-signaling pathways in a given condition. Those findings clearly revealed the existence of six hPRL variants in human pituitaries, and the pattern changes of six hPRL variants among different subtypes of pituitary adenomas, which provide novel clues to further study the functions, and mechanisms of action, of hPRL in human pituitary and in PRL-related diseases, and the potential clinical value in pituitary adenomas. PRL variants contribute to an osmotic adjustment disorder. Bollengier et al. indicated four variants of rat PRL in the pituitary cell with two-dimensional gel electrophoresis (2DGE), and also found that those variants were derived from different PTMs (9). Most of publications are involved in the studies of non-human PRL variants with identity of only a few non-human PRL variants (10)(11)(12). Some studies also found hPRL variants (n = 3-4) (13). However, until now 2DGE has not been used to study the hPRL variants. 2DGE is an effective method to separate proteins according to different pI values in the isoelectric focusing (IEF) direction, and different M r values in the sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) direction (14). pI and M r are the basic properties of a protein variant. 2DGE-based Western blotting coupled with antibody of a given protein can also be used to detect the variants of that protein. Therefore, 2DGE is able to array PRL variants with different pI and M r . Mass spectrometry (MS) is an effect method to characterize the isolated PRL variants and identify PTMs with an analysis of amino acid sequence and determination of PTM sites. These types of MS have been used to study the 2DGE-separated pituitary proteins, including matrix-assisted laser desorption/ionization time-of-flight MS (MALDI-TOF MS) peptide fingerprint (PMF), MALDI-TOF-TOF MS and tandem mass spectrometry (MS/MS), and liquid chromatography-electrospray ionization-quadruple-ion trap MS (LC-ESI-Q-IT MS) MS/MS analysis (15)(16)(17)(18).

INTRODUCTION
This present study used 2DGE, 2DGE-based Western blotting, and MS to detect and identify hPRL variants in human pituitaries and differentially expressed profiles of hPRL variants among different subtypes of pituitary adenomas relative to controls, which will provide novel clues to further study the functions, and mechanisms of action, of hPRL in human pituitary and in PRL-related diseases, and the potential clinical value in pituitary adenomas.

Tissue Samples
Human control pituitary glands were post-mortem tissues obtained from the National Disease Research Interchange (n = 1) and the Memphis Regional Medical Center (n = 7), which were approved by University of Tennessee Health Science Center (UTHSC) Internal Review Board (IRB). Human pituitary adenoma tissues were obtained from the Emory University Hospital, which were approved by Emory University Hospital IRB, and Department of Neurosurgery of Xiangya Hospital, which were approved by the Xiangya Hospital Medical Ethics Committee of Central South University, China. Consent was obtained from each patient or the family of control pituitary subject after full explanation of the purpose and nature of all procedures used. All tissues were removed, frozen immediately in liquid nitrogen, and stored at −80 • C until processing. The clinical information of pituitary adenoma and control samples is summarized in Table 2.

Protein Extraction
Tissue processing and protein extraction of pituitary control and adenoma tissues have been previously described (17,19). Briefly, each tissue sample (∼600 mg) was washed with 0.9% NaCl (3 ml, 5×) to thoroughly remove blood, and was finely ground in liquid nitrogen. A volume (5 ml) of protein extraction buffer that contained 2 mol/L thiourea, 7 mol/L urea, 40 g/L CHAPS, 100 mmol/L dithiothreitol (DTT), 5 mol/L immobilized pH gradient (IPG) buffer pH 3-10 NL, and a trace of bromphenol blue was added, and the mixture was thoroughly mixed. The mixture was vortexed (2 h) on ice and was centrifuged (15,000 × g, 15 min, 4 • C). The supernatant was collected, and its protein concentration was determined with a Bio-Rad 2D Quant kit (Bio-Rad). The supernatant was the "protein sample solution."

2DGE and Western Blot
First Dimension-IEF IEF was performed with precast IPG strips (pH 3-10 NL, 180 × 3 × 0.5 mm) and 18 cm IPG strip holder on an IPGphor instrument (GE Healthcare). Before IEF, an aliquot (350 µl) of protein sample solution was loaded onto an IPG strips. The IPG strip was rehydrated about 18 h. IEF was performed at 25 • C with the following parameters: a gradient 250 V and 1 h for 125 Vh, a gradient at 1000 V and 1 h for 500 Vh, a gradient at 8000 V and 1 h for 4000 Vh, a step-and-hold at 8000 h and 4 h for 32000 Vh, a step-and-hold at 500 V and 0.5 h for 250 Vh to achieve a total of 36,875 Vh and ∼7.5 h analysis. The IPG strip was removed and laid on its plastic back to blot off mineral oil. After IEF, IPG strips were processed with SDS-PAGE.

Second Dimension-SDS-PAGE
After IEF, an Ettan DALT II system (GE Healthcare, up to 12 gels at a time) was used. An Ettan DALTsix multiple casters (GE Healthcare) was used to cast the 12% PAGE resolving gel (250 × 215 × 1.0 mm). The resolving-gel solution was made by mixing 75 ml of 1.5 mol/L Tris-HCl pH 8.8, 90 ml of 400 g/L acrylamide/bis-acrylamide (29:1 = weight:weight, cross-linking ratio = 3.3%), 3 ml of 10% ammonium persulfate, 150 ml deionized distilled water (ddH 2 O), and 50 µl of tetramethylethylenediamine. The IPG strip was equilibrated in 15 ml of reducing equilibration buffer (15 min) that consisted of a trace of bromphenol blue, 375 mmol/L Tris-HCl pH 8.8, 2% w/v SDS, and 20% v/v glycerol. The IPG strip was equilibrated in 15 ml of alkylation equilibrium solution (15 min) that consisted of 2.5% w/v iodoacetamide instead of 2% w/v DTT. A boiled solution that consisted of 1% w/v agarose solution was used to seal the equilibrated IPG strip onto the top of the resolving gel. 2DGE was performed in 25 L of Tris-glycine-SDS electrophoresis buffer that consisted of 192 mmol/L glycine, 25 mmol/L Trisbase, and 0.1 % w/v SDS at 25 • C with a constant voltage (250 V, 360 min).

2DGE-Based Western Blot With Anti-hPRL Antibody
After electrophoresis, the 2D gel was removed, and proteins in the 2D gel were transferred to a polyvinylidene fluoride (PVDF) membrane (0.8 mA/cm 2 for 1 h 20 min) with an Amersham Pharmacia Biotech Nova Blot semi-dry transfer instrument. The proteins on the PVDF membrane were blocked for 1 h at room temperature with a solution (100 ml) of 0.3% bovine serum albumin/phosphate-buffered saline (BSA/PBS) with 0.2% Tween 20 and 0.1% sodium azide (PBST). The proteins on the PVDF membrane were incubated for 1 h at room temperature with rabbit anti-hPRL antibodies that were diluted (v:v = 1:1000) in a 0.3% BSA/PBST solution. After incubation with the primary antibody, the PVDF membrane was washed with 200 ml PBST solution (15 min × 4) and rinsed twice with ddH 2 O. The secondary antibody, goat anti-rabbit alkaline phosphtaseconjugated IgG was diluted (v:v = 1:4000) in a 0.3% BSA/PBST solution. The solution was added to the blots for 1 h at room temperature. The membrane was washed with 200 ml PBST solution (15 min × 4), and proteins were visualized with 5bromo-4-chloro-3-indolyl phosphate.

Protein Staining and Image Analysis
All 2DGE-separated proteins were visualized with silver-staining (20). The silver-stained 2D gel was digitized and analyzed with Discovery Series PDQuest 2D Gel Analysis software (15,21).

MALDI-TOF MS
MALDI-TOF MS was used to analyze each protein that was digested with trypsin. Experiments were performed on a Perseptive Biosystems MALDI-TOF Volyager DE-RP MS (Framingham, MA, USA). The parameters of the instrument were described (15). The protonated molecule ion [M+H] + was produced with MALDI-TOF MS. Data-processing software (DataExplore) was used to obtain accurate masses. Blank-gel experiments were conducted simultaneously to remove masses from known contaminants (usually keratin), matrix, trypsin, and other unknown contaminants. Each protein was identified by using the data obtained from MALDI-TOF MS PMF by searching the Swiss-Prot database 091215 (513877 sequences; 180750753 residues; January 19, 2015) with PeptIdent software.

LC-ESI-Q-IT MS
Proteins in 2D gel-spots that corresponded to each positive Western blot were excised and digested in the gel with trypsin. According to the manufacturer's method recommendations, the mixture of tryptic peptides was purified with a ZipTipC18 microcolumn. Purified tryptic peptides were analyzed with an LCQ Deca mass spectrometer (Thermo-Finnigan, San Jose, CA, USA). The instrument parameters were: electron multiplier-900 V, ESI voltage 2.0 kV, and capillary probe temperature 110 • C. The detailed experimental steps have been described (15). Data from LC-ESI-Q-IT MS were used to identify the protein with Swiss-Prot database.

MALDI-TOF-TOF MS
Proteins in the 2D gel-spots that corresponded to the positive Western blot were excised and digested in the gel with trypsin. For Perspective Biosystems MALDI-TOF-TOF MS analysis, the tryptic peptide extraction was eluted directly from a liquid chromatography microcolumn onto a MALDI plate with a matrix that contained 3 mg/ml saturated α-cyano-4-hydroxycinnamic acid solution. The purified tryptic peptide mixture was analyzed (n = 5). Tryptic peptides were analyzed with MALDI-TOF-TOF MS that operated in the reflective mode at an acceleration voltage of 25 kV over m/z 800-4000. Precursor ions close to the theoretical m/z were selected for TOF-TOF [UltraFlex III MALDI-TOF-TOF (Bruker Daltonics)] MS analysis. After automatic analysis, any remaining unidentified ions were manually analyzed. In the results obtained from MS, data compared to the theoretical values from the database were used to determine whether these peptides had undergone PTMs.

Statistical Analysis
The Chi-square test included in SPSS 22 software was used to analyze the difference in proportional ratio of PRL variants among five subtypes of pituitary adenomas, with a significance level of p = 0.05.
In the non-hormone expressed nonfunctional pituitary adenoma (NF-NFPA) group relative to controls ( Table 3), hPRL was downregulated by 8.3-fold in spot v1 and 4.9-fold in spot v2, was not changed in spot v3, and was lost in spots v4, v5, and v6. In the leuteinizing hormone (LH)-positive NFPAs relative to controls, hPRL was downregualted by 12.6-fold in spot v1, 4.1-fold in spot v2, 26.2-fold in spot v3, 20.1-fold in spot v4, 36.7-fold in spot v5, and 33.6-fold in spot v6. In the follicle-stimulating hormone (FSH)-positive NFPAs relative to controls, hPRL was not changed in spot v2, lost in spot v5, and was downregulated by 46.2-fold in spot v1, 14.6-fold in spot v3, 17.6-fold in spot v4, and 11.3-fold in spot v6. In FSH/LH-positive NFPAs relative to controls, hPRL in all spots was downregulated by 99.9-, 3.8-, 12.3-, 19.0-, 19.7-, and 32.6fold, respectively. However, for prolactinomas (PRL-adenomas) relative to controls, hPRL was downregulated in only spot v1 by 3.4-fold, and not changed in the other spots. Moreover, the overall proportional ratio of six PRL variants was significantly different among the five subtypes of pituitary adenomas (NF − , FSH + -, LH + -, FSH + /LH + -, and PRL + -adenomas) analyzed here with the Chi-square test performed with SPSS 22 software (p < 0.05) (Figure 4). The Chi-square tests carried out between every two subtypes of pituitary adenomas indicated that the proportional ratio of six hPRL variants was significantly different between every two subtypes of pituitary adenomas (p < 0.05), except for no significant difference between subtypes FSH + /LH + and PRL + , and between subtypes FSH + and PRL + .

Validation of hPRL Variants With 2DGE-Based PRL Immunoaffinity Blot
2DGE-based Western blot coupled with anti-hPRL antibody and MS was an effective method to validate hPRL variants in human pituitaries. Four hPRL variants in human pituitary, including variants v1, v4, v5, and v6, were detected with an hPRL    2DGE-based Western blot coupled with anti-hPRL antibody and MS clearly confirmed hPRL variants that were contained in spots v1, v4, v5, and v6 (Figure 1 vs. Figure 5). However, the hPRL variants in spots v2 (pI 6.3; 26.4 kDa) and v3 (pI 6.3; 27.9 kDa) were not confirmed; that finding might be due to several factors: (i) the PRL antibody was a general antibody, not specific to a given variant, and (ii) the M r of hPRL in the spots v2 and v3 was larger than the M r (25.9 kDa) of hPRL, obviously some adducts were added to the hPRL, which could be result in its non-reaction with the hPRL antibody. Whether the hPRL was detected with an immunoblot or not the hPRL in spots v2 and v3 was unequivocally identified with MS and MS/MS (Figure 1). Therefore, hPRL in spots v2 and v3 is still recognized as the hPRL variants in human pituitaries.

Determination of Signal Peptide (Position 1-28) Contained in Each hPRL Variant in Human Pituitaries
PRL in human Swiss-Prot database contains 227 amino acids (position 1-227) with a signal peptide (position 1-28) ( Table 1), which is a prohormone of hPRL with a molecular weight of 25.9 kDa, whereas mature hPRL contains 199 amino acids (position 29-227) with a removal of the signal peptide (position 1-28), with a molecular weight of 22.9 kDa. Therefore, it is necessary to determine whether those identified six hPRL variants are derived from hPRL prohormone or mature hPRL. With the theoretical calculation using PeptideMass Cleavage software ( Table 5), there are seven characteristic tryptic peptides (positions 1-4, 1-9, 1-38, 5-9, 5-38, 10-38, and 10-44) derived from hPRL prohormone   (Figure 2). Moreover, from the 2DGE pattern of six hPRL variants (Figure 1), it is clear that variants v1, v4, v5, and v6 had the corresponding M r of 26.0 kDa, 26.1 kDa, 25.9 kDa, and 25.9 kDa, respectively, which are very close to the molecular weight of hPRL prohormone 25.9 kDa; and that the variants v2 and v3 have the corresponding M r of 26.4 kDa, and 27.9 kDa, which is a bit larger than the molecular weight of hPRL prohormone 25.9 kDa. Therefore, according to MS data and 2DGE pattern, the identified six hPRL variants in human pituitary tissue should be derived from hPRL prohormone (position 1-227), but not from mature hPRL (position 29-227).

Bioinformatic Prediction of Potential Factors to Result in hPRL Variants
Alternative splicing and PTMs are two main factors to result in protein variants. From the analysis in the section determination of signal peptide (position 1-28) contained in each hPRL variant in human pituitaries, six identified hPRL variants were not from splicing and truncation. Thus, PTMs might be the main factor to result in six hPRL variants. In the 2DGE map of hPRL variants (Figure 1), variants v1, v4, v5, and v6 had the very similar M r , but with obviously measurably different pI. The deamidation of an asparagine (N) residue to aspartate (D) and of glutamine (Q) FIGURE 6 | The PRL-driven signaling pathways via the short or long PRL receptors, which was achieved from KEGG pathway database. residue to glutamate (E) is an effect of protein aging, and is often observed in 2-D gels (26)(27)(28)(29). Deamidation results in a series of spots with the same M r values but measurably different pI values (30), with an increase of 1 Da in peptide mass and a decrease in the apparent pI from the uncharged amide to a negatively charged carboxylate anion at pH 7.4. Moreover, deamidation might be produced under the basic conditions for storing samples (30).
In the 2DGE map of hPRL variants (Figure 1), variants v2 and v3 had the same apparent pI but obviously measurably different M r , and their M r 26.4 kDa and 27.9 kDa were a bit larger than hPRL prohormone (25.9 kDa). This phenomenon might arise from some PTMs with a relatively larger chemical group, such as glycosylation or phosphorylation. Glycosylation means that the protein contains one or more covalently linked carbohydrates of various types from monosaccharides to branched polysaccharides. Phosphorylation means that the protein is modified by the attachment of either a single phosphate group, or of a complex molecule such as 5'-phospho-DNA through a phosphate group at the targeted serine, threonine, or tyrosine residues. The annotation page of hPRL in the UniProt database (https://www.uniprot.org/uniprot/P01236); P01236, or PRL_HUMAN) clearly demonstrates the hPRL is a glycoprotein or phosphoprotein. Moreover, multiple phosphorylated sites at the serine (S), threonine (T), and tyrosine (Y) residues were significantly predicted by NetPhos 3.1 Server, including 14 pS sites, 5 pT sites, and 3 pY sites in the hPRL ( Table 6). Ten significantly N-glycosylated sites were predicted by NetNGlyc 1.0 Server in the hPRL ( Table 7), and six significantly O-glycosylated sites were predicted by NetOGlyc 4.0 Server in the hPRL ( Table 8).

Biological Functions of PRL in Different Physiological and Pathological Conditions
PRL plays an extremely important role in different vertebrate species and organ systems, such as growth, metabolism, development, reproduction, immunoregulation (31), behavior and brain (6), water and balance (32), stress response, and suppression of (33-35) and regulation of firing of oxytocin neurons (36,37). As it is named, PRL exerts multiple effects that include development (mammogenesis) and stimulation of mammary gland growth, synthesis of milk (lactogenesis), and maintain the secretion of milk (galactopoiesis) on the mammary gland (38). Moreover, PRL affects reproduction of  the mammalian species. During pregnancy, PRL concentrations are high. During pregnancy, low immunity might be associated with PRL. Decreased serum PRL might cause anxiety in the first trimester of pregnancy (39). PRL induces and maintains maternal behaviors in rabbit, rat, hamsters, and sheep (40). These maternal behaviors include cleaning, nesting, grouping, and nursing of the baby by the mother (41,42). Aside from the effects of PRL on reproductive processes, PRL also plays a significant role to maintain homeostasis of the body's environment by development of blood vessels (angiogenesis), regulation of the immune systems, and osmotic balance (3). PRL is also involved in the synthesis and secretion of milk, and the major ions in milk are sodium, potassium, and chloride ions. The levels of chloride and sodium ions in milk are low, and potassium is high (43). PRL might maintain low levels of sodium in milk because it can promote sodium retention in the mammary gland. Moreover, PRL is involved in some endocrine-related cancers. PRL plays a vital role in the regulation of the immune responses; it interferes with inhibiting apoptosis, induces B-cell tolerance, increases antibody secretion, enhances antigen presentation, and upregulates cytokine production (44). PRL has different functions in the immune system of healthy people, including Treg cell inhibitory factor and inhibition of cell secretion (45). PRL contributes to the inflammatory microenvironment by reducing the inhibitory effect of Treg cells on T cells (45,46). The proteolytic fragments of native PRL inhibit angiogenesis. Anti-angiogenic activity is achieved by the interaction between the 16 kDa N-terminal fragment of PRL and a specific receptor (47). Under normal conditions, PRL can promote the growth and development of the mammary gland, and the 16 kDa fragment of PRL can inhibit angiogenesis. Dilatation and regression of blood vessels seriously affect the normal growth and degeneration of breast cancer. However, angiogenesis disorders are the characteristics of growth and metastasis of breast cancer. That is not given physiological significance. It seems possible that the 16-kDa fragment of PRL plays a pathological role to inhibit angiogenesis or a local inhibitor of tumorigenesis.  The bold values mean statistical significantly positive results.

Determination of the Number of hPRL Variants in Human Pituitaries
The pituitary proteome was separated with 2DGE, followed by MS identification. Six 2D gel spots were found to contain hPRL with MS-identification, which represents six hPRL variants in human pituitaries. Of them, four hPRL variants (v1, v4, v5, and v6) were confirmed with 2DGE-based Western blotting in combination with anti-hPRL antibodies and followed by MS identification. However, two hPRL variants (v2 and v3) was not detected with 2DGE-based Western blotting in combination with anti-hPRL antibodies, which is due to the factors: (i) anti-hPRL antibody was not a variant-specific antibody, and (ii) those two hPRL variants might have different properties including different PTMs and unknown factors to cause its no-reaction with anti-hPRL antibody. Anyway, this present study directly MS-identified the hPRL in six different 2D gel spot with different pI and M r . Therefore, the present study still used the six hPRL variants to study their differential expression profile among different subtypes of pituitary adenomas.

Formation of hPRL Variants in Human Pituitaries
There are many factors that might produce protein variants: (i) Estimates of DNA level variation. Major sources of human protein variation include the encoding of single nucleotide polymorphisms and mutations. (ii) The main source of RNA level variation. Alternative splicing is a key factor in transcriptome complexity and regulating complex human characteristics. (iii) Errors in translation. Errors in protein translation provide a very large potential source of proteome expansion, especially in stressed or aging cells. (iv) PTMs. Due to PTMs, the potential number of proteoforms can increase exponentially (48). PTMs might have a certain influence on the structure and function of the protein; for example, PTMs of α-synuclein are major regulators of their function, structure, degradation, and toxicity (49). PTMs of the FUS protein might affect its associated pathology and serve as a therapeutic target (50).
Currently, different types of PTMs have been found in the human, including glycosylation, phosphorylation, acetylation, ubquitination, methylation, deamidation, and nitration (27,28,51). These factors could also produce hPRL variants. The present study found that six PRL variants that are derived from the same PRL gene in human pituitaries were not due to alternative splicing, but rather to PTMs, including deamidation, glycosylation, and phosphorylation. The hPRL variants v1, v4, v5, and v6 on the 2DGE map had a very similar M r but obviously different pI, which might be mainly due to deamidaton. The hPRL variants v2 and v3 with a slightly larger M r relative to hPRL (position 1-227) might be mainly due to the glycosylation and phosphorylation. Therefore, PTMs might be the main reason to produce hPRL variants. Furthermore, we confirmed that the amino acid sequence of six identified hPRL variants in human pituitaries contained the signal peptide (position 1-28), which is the hPRL prohormone (position 1-227), but not the mature PRL (position 29-227).

Potential Molecular Mechanisms of Action of hPRL Variants Elucidated by Signaling Pathway Networks
The study of the hPRL signaling pathway network is helpful to understand its functions; explore molecular mechanism of PRL-related tumors such as pituitary adenomas, including prolactinomas; develop effective therapeutic drugs; and discover tumor biomarkers (52). Only after the interaction between PRL and its receptors can PRL play a corresponding function. For example, for the secretion of milk, the binding of PRL and PRL receptor (PRLR) does not activate cell membrane binding enzyme, but phospholipase A. The signaling pathway map of hPRL is complex (Figure 6). There are two types of PRLRs: long-PRLR and short-PRLR. PRL binding to different receptors can cause different physiological functions. The signaling pathways activated by PRL binding to different receptors are specific and exhibit crossover. PRL binding to short receptors activates the PI3K/Akt signaling pathway, whereas PRL binding to long receptors activates the JAK2/STAT signaling pathway. The crosssignaling pathway activated by the binding of PRL to long or short receptors is a MAPK signaling pathway that activates cell cycle regulators and thus affects cell proliferation. The different short or long PRLRs should bind different hPRL variants v1-v6 to activate different signaling pathways and produce different biological effects in different physiological and pathological conditions. Furthermore, the significantly differential expression of six hPRL variants ( Table 3) might result in different signaling pathway alteration. The further studies on the association of different hPRL variants and PRLR signaling pathway might have important scientific merits and clinical significance.

STRENGTH AND LIMITATION
The present study clearly demonstrated the existence of six hPRL variants in human pituitaries with 2DGE and MS analyses, and its differential expression profile of six hPRL variants among different subtypes of pituitary adenomas; and confirmed that those six hPRL variants in human pituitary tissues were derived from hPRL prohormone with 227 amino acids (position 1-227), but not from mature hPRL with 199 amino acids (position 29-227).
However, mature PRL is secreted by prolactotroph cells concentrated on both sides of the posterior pituitary gland and enters the bloodstream after secretion. After blood transport, PRL will bind to its specific receptor before it can play a corresponding physiological role. PRL in pituitary tissue is difficult to detect with routine examinations, but PRL in blood is relatively easy to detect. Under normal circumstances, the level of PRL in the blood plays its role in a certain range. The abnormal secretion of PRL is related to many diseases, such as hyperprolactinemiaemia, tumor, and autoimmune diseases. Studies have shown that PRL has a direct or indirect relationship with many diseases. Whether these diseases are associated with serum PRL variants remains unknown. Therefore, it is necessary in our future studies to indepth investigate the mechanisms and function of different PRL variants and the status of serum PRL variants. The present study provides the basis for us to in-depth study serum PRL variants and their functions. CONCLUSION 2DGE, 2DGE-based Western blot coupled with anti-hPRL antibody, MS, and bioinformatics were used to identify hPRL variants in human pituitaries, and their differential expression profiles among different pituitary adenomas. Six hPRL variants in human pituitaries were identified, including two hPRL variants that were not detected by Western blotting. The differential expression patterns of six hPRL variants were significantly different among five subtypes of pituitary adenomas (NF − , FSH + -, LH + -, FSH + /LH + -, and PRL + -adenomas). Moreover, six hPRL variants in human pituitaries derived from hPRL prohormone (position 1-227) with different PTMs such as deamidation, glycosylation, and phosphorylation, but not from mature hPRL (position 29-227). Those findings provide novel insights into the functions, and mechanisms of action, of hPRL in human pituitary and in PRL-related diseases, and into the potential clinical value in pituitary adenomas. Furthermore, the hPRL variants involved signaling pathways that might clarify the biological functions of different PRL variants and their potential clinical significance, and contribute to the development of drugs that block the PRL signaling pathway for clinical treatment.

AUTHOR CONTRIBUTIONS
SQ analyzed data, prepared figures and tables, designed, and wrote the manuscript. YY collected tissue samples, performed clinical explanation, and carried out bioinformatics analysis and partial revision of manuscript. NL, TC, XW, and JL participated in experiments and partial data analysis. XL collected tissue samples and performed clinical diagnosis. DD participated in design, instructed, and critically revised manuscript. XZ conceived the concept, designed experiments, and manuscript, instructed experiments and data analysis, supervised results, coordinated, critically revised/wrote manuscript, and was responsible for its financial supports and the corresponding works. All authors approved the final manuscript.