PreImplantation factor (PIF*) promotes embryotrophic and neuroprotective decidual genes: effect negated by epidermal growth factor

The method was already described [25]. Briefly, synthetic PIF analog (MVRIKPGSANKPSDD) was produced using solid-phase peptide synthesis (Peptide Synthesizer, Applied Biosystems, Foster City, CA, USA) employing 9-fluorenylmethoxycarbonyl (Fmoc) chemistry. Final purification was carried out by reversed-phase high-pressure liquid chromatography, and peptide identity was verified by mass spectrometry as previously described [14].

Yale University School of Medicine review board approval was obtained for this study. Using our previously established cell culture method [14, 15], non-pregnant human endometrial stromal cells (HESC) and cells collected from healthy first trimester of pregnancy deciduas were studied. Discarded endometrial tissue from premenopausal women undergoing hysterectomies due to benign indications was used. Decidual specimens from the first trimester were obtained from women undergoing elective termination in weeks 6–12 of normal pregnancy. Endometrial and decidual cells were isolated and re-suspended in Roswell Park Memorial Institute (RPMI)-1640 medium, grown to confluence, and found to be leukocyte free (<1%). After reaching confluence, the cells were decidualized using 10^-8 mol/L estradiol and 10^-7 mol/L synthetic progestin analog (R5020) (DuPont/NEN, Boston, MA, USA), in both cases for 7 days. The cells were switched to a serum-free medium containing insulin, transferrin, and selenium (Collaborative Research Inc., Waltham, MA, USA); 5 μmol/L trace elements (GIBCO, Carlsbad, CA, USA); and 50 μg/mL ascorbic acid (Sigma-Aldrich, St. Louis, MO, USA), and treated for 24 h with or without synthetic PIF (100 ng/mL). In other experiments, 50 ng EGF was cultured together with 100 ng/mL PIF for 24 h in HESC and first-trimester decidua cultures (FTDC). The tissue culture was collected and frozen at -80°C The samples were tested in triplicate.

Genes that were significantly changed in expression by PIF (p < 0.05) using Student t test followed by a greater than twofold change were reported. Further results were divided into upregulated or downregulated lists and underwent new gene pathway analysis. MetaCore from GeneGo Inc., a Thomson Reuters business (Carlsbad, CA, USA), was used to identify and visualize the involvement of differentially expressed genes in specific cellular pathways. Enrichment analysis algorithms across several GeneGo ontologies were used to rank pathways, process, and diseases. To determine possible key regulators that may contribute to PIF-induced changes in gene expression to the human interactome (manually annotated interactions from peer-reviewed published experiments), the interactome tool in the MetaCore platform was used. To further understand the mechanisms and processes represented by the top connected hubs, networks were built using the shortest path algorithm followed by enrichment analysis using the disease and gene ontology processes.

Pathway analysis was performed using the Ingenuity Systems Inc. (Redwood, CA, USA) software, ranking by greatest number of genes in a given pathway and associated statistical significance.

HESC protein lysates (n = 3/group) from an independent experiment as compared to those of an mRNA study were homogenized in retinoic acid receptor alpha (RARA) buffer (100 mM Tris 250 mM NaCl, 0.1% Triton X-100, 0.1% sodium dodecyl sulfate (SDS)) and assayed for protein concentration. Twenty micrograms of each lysate was loaded in duplicate onto an SDS-poly-acrylamide gel Novex 4%–12% (Invitrogen, Carlsbad, CA, USA) followed by electrophoresis. Each lane was excised into 40 equal slices, digested with trypsin, and analyzed by nano-liquid chromatography/mass spectrometry (nano-LC/MS/MS) on an LTQ-Orbitrap XL™ tandem mass spectrometer (ThermoFisher, San Jose, CA, USA) carried out at NextGen Sciences (Ann Arbor, MI, USA). Data were searched against the Mascot concatenated forward-and-reversed v3.38 International Protein Index (IPI) database (Matrix Science Ltd., London, UK) and collated into non-redundant lists using Scaffold software (Proteome Software Inc., Portland, OR, USA). Using these combined forms of software creating ion identity score, and using the PeptideProphet provide high accuracy by determining true probabilities of proteins identified. Spectral counting was employed for relative quantitation, and a t test was utilized to show significant differences.

Neural development initiates shortly post-implantation, and therefore, PIF’s effect on pathways that lead to the secretion of neurotrophic factors for the embryo by the decidua was examined (Table 1). The highest ranking gene (13.8-fold increase) is TLX2—a nuclear receptor subfamily 2 group E that binds DNA. This protein plays a major role in anterior brain differentiation and vision development [25]. Also, highly ranking is EPHA10 (9 fold) whose protein encodes a receptor for tyrosine kinase targeted by ephrin-A family members [26]. This protein is important for cell-to-cell communication and neural cell mobility. On the other hand, RARA gene expression decreased (-9 fold) [27]. This gene encodes a receptor for retinoic acid which is involved in growth arrest.

In FTDC, the major upregulated gene was SMAD1 (53.4-fold increase) (Table 1) [28]. Its encoded protein is TGFβ signaling protein, a major modulator of the bone morphogenetic protein—highly relevant for embryo neural development [29]. There is also an increase in SMAD6 and SMAD2 (2.7-fold and 2.5-fold, respectively). EPHA5 increased by 7.8-fold; this protein is a receptor tyrosine kinase that binds ephrin-A family ligands, which are highly relevant to neural development, especially axonal guidance and synaptogenesis [30]. On the other hand, PIF downregulates PPP2R2C, which is involved in negative control of cell growth and division [31]. Collectively, this set of data indicates that PIF is involved in pathways that are critical for neural development.

PIF also appears to affect several neural-related pathways that could contribute to neurotrophic effects on the embryo. The highest correlation was found with axon guidance signaling followed by neuroregulin, which is involved in EGF signaling; the specific genes involved are listed in Table 2. Effective synaptic transmission is recognized to be critical for effective neural development. The regulatory effect of PIF in this pathway reflects the peptide’s significant protective role against altered neural development. The lowest ranking—but still relevant—gene pathway affected by PIF is the oxidative phosphorylation pathway, which is also critical for embryo survival.

Upon analyzing the FTDC data (Table 3), an interesting pattern emerged since the data showed that the highest ranking neural pathways affected by PIF are actually related to adult neurological disorders. Among them are amyloid processing genes that can lead to Alzheimer’s disease, amyotrophic lateral sclerosis, and Huntington’s chorea. This is followed by effects on neural signaling pathways. PIF’s effect on individual genes and their expression in the different pathways is shown in Table 3. This correlation with adult disease implies an evolution in PIF targeting of the maternal milieu with advancing gestation which is different from earlier development phases which appear to be more involved in early childhood diseases.

Since PIF exhibits trophic effects mainly on neural development, it was of interest to further examine the genes involved in embryo development as well, possibly a long-term protective strategy to prevent adverse post-natal disorders. The MetaCore program enabled such analysis. The full description of the role of PIF in multiple cellular signaling pathways is however beyond the scope of this manuscript. Hence, this information is available in detail in Additional file 1 (pages 1–42), and only key findings are referenced herein. For such a comprehensive analysis, the data had to be analyzed as pathway maps, process networks, followed by the specific physiologic processes involved, and finally the disease association, when these processes are perturbed which lead to disease during childhood or in adulthood (Table 4). Data was analyzed as the highest association in each category, with associated statistical significance.

PIF’s highest ranking pathway in HESC is the angiotensin signaling via beta-arrestin, which is coupled with neural transmission or nerve impulse associated with nervous system development and, when affected, leads to autism. The second ranking pathway affected by PIF was genes involved in the epithelial-mesenchymal transition (EMT), which is associated with muscle contraction, is needed for neural functionality, and if affected, can lead to child development disorders. Interestingly, PIF was found to have significant associations with pathways relating to mood disorders and stress, TGFβ, WNT, CDK5, and BMP.

Among the significantly upregulated processes for which PIF increased the gene expression more than twofold are (Table 5) angiotensin—via beta arrestin, G-protein signaling RhoA/B, and CDC42, showing that the effects on neurogenesis and immune response are dominant. In contrast, the processes downregulated by PIF are EMT, EGF activation, and axon growth repulsion. The balance among upregulated and downregulated processes implies a dynamic adaptive effect on these pathways.

In FTDC (Table 6), on the other hand, different maps to diseases were identified. PIF’s highest ranking pathway was acetaminophen metabolism (i.e., xenobiotics) which relates to reproduction/hormone signaling and, when affected, can lead to autoimmune connective tissue disorders (Table 5). The second ranking pathway was steroid metabolism related to vasculogenesis and, when altered, leads to development of skin disorders. Of note, connective tissue disorders and neoplasms occur mostly in adults. The associated upregulated processes are cell cycle role in 14-3-3, ERK5, and mitosis, and neuronal survival and development. The downregulated processes are xenobiotics related to acetaminophen, 2-napthalamine, benzo (a) pyrene, and steroid metabolism. The emphasis on protection against xenobiotics is critical for PIF’s protective role during embryogenesis.

To document the dynamic gestational age-dependent effect of PIF in early pregnancy, similarities and differences between those critical time periods were examined. Common pathways affected by PIF include muscle contraction and angiotensin signaling via four independent pathways: STATs, ERK, PYK2, and CREB pathways. Neurophysiologic processes affected were melatonin signaling and receptor-mediated growth axon repulsion. Global processes were transmission of nerve impulses, axonal guidance, and cell adhesion. The similarity between the two time periods with respect to neural development illustrates PIF’s continuing role in this critical process.

To better delineate a network of genes associated with PIF activity, an interactome analysis was carried out. This enables us to relate a large number of genes that are linked in a network. The hub enables to define the number of interactions among genes. Table 7 illustrates the leading genes and their interaction. In HESC, FUS was dominant [32]. This gene controls genome integrity and RNA processing and has the highest (n = 68) interactions, followed by HIPK2, whose protein is a kinase involved in p53-regulated cellular apoptosis [33].

The FTDC interactions were much more extensive (Table 7). The highest ranking was STAT3 (n = 603) [34]. The protein plays a central role in JAK/STAT signaling of several cytokines. The next ranking was CDK1 (n = 389), a kinase that controls cell cycle regulation [35]. Of note, the decidua genes are more connected through two signaling steps than the genes found in HESC, in which three steps are involved.

Table 8 illustrates the top biological processes affected by PIF in HESC. The highly prominent DNA-dependent regulation is required during rapid cellular development. In contrast, the top process in FTDC is the response to chemical stimulus, i.e., protection against adverse environment during embryogenesis. This further substantiated the evolving role of PIF from supporting neural development to protection against xenobiotics.

Using an independent experiment, the effect of PIF on a number of protein levels was determined by using semi-quantitative mass spectrometry (14). To provide a more detailed analysis, we have analyzed, in addition, the effect of PIF on a total of >1,300 HESC proteins demonstrating that several of them were affected by PIF (Additional file 2). We found that PIF affected several proteins involved in the control of oxidative stress and protection against abnormal protein synthesis: PDIA3, P4HB (PDI), TXND, PRD6, HSPB1, SDHA, and BAG1. This was clearly in line with the microarray data. In addition, specifically focusing on proteins involved in neural function, the data revealed that PIF affects several additional proteins among them. PPIC is affected in ataxia and in degeneration of Purkinje cells [36]. OCRL is affected in Lowe syndrome oculo-cerebral defects [37]. AP3D1 is involved in budding vesicle transmission to neurons [38]. SLC25A1 and SLC16A1 are proteins involved in mitochondrial transport widely expressed in the central nervous system [39]. LDHA is involved in substania nigra development and is deregulated in glioma tumors [40]. Thus, PIF also affects not only genes but also decidual proteins.

EGF is reported to block decidua formation, and PIF was shown to regulate EGF-related pathways, promoting amphiregulin and epiregulin while inhibiting betacellulin [15]. Therefore, whether EGF can affect PIF’s action was examined. Addition of EGF abolished PIF’s effect on the global genome. Only a limited number of genes (N = 15) were affected, mostly of which were downregulated. The EGF treatment-related pathway is shown (Figure 1). The decrease noted in MMP metalloproteinase expression may interfere with decidual function [41]. This is further amplified by the decrease in prolactin receptor (Figure 2), where the ligand is crucial for decidua formation. The most downregulated gene was BMPER, an inhibitor of BMP (-10 fold) followed by NIF3LBP1 (-7.4 fold), which binds NIF3 (an amyotrophic lateral sclerosis candidate gene), MCC1 (-7.4 fold) (a tumor suppressor), and PRLR (-4.2) (a prolactin receptor important for decidua function) [42‐46]. The gene polymorphism is associated with multiple sclerosis. The only gene which was increased is Nov (2 fold), which is a negative regulator of cell growth in choriocarcinoma cells [47].

PIF significantly reduces EGF receptor (-25.6 fold) in FTDC, therefore blocks the ligand’s activity [14]. Whether EGF interferes with PIF’s action on the decidua was analyzed. The most downregulated gene was ZHX2 (-8.4 fold), which was shown to promote AFP secretion leading to activation in liver cancer [48]. Figure 3 shows that EGF reduced IL8 expression and decreased PDE3A (-5.6 fold), involved in learning and memory [49]. There was also a decrease in PEG3, which blocks inflammation-induced apoptosis and is a tumor suppressor in glioma cells (-6.3 fold). Interestingly, PIF alone on FTDC negated both PEG3 (-5 fold) and PDE7B (-5.7 fold) [50, 51]. These observations document the negative relationship between PIF, a promoter of decidual function, and EGF, which acts as a pro-proliferative agent with anti-decidual properties.