RESEARCH ARTICLE Molecular Reproduction & Development 81:326–340 (2014)
Cell Surfactomes of Two Endometrial Epithelial Cell Lines That Differ in Their Adhesiveness to Embryonic Cells SONALI R. BHAGWAT, TEJASHREE REDIJ, KRUTTIKA PHALNIKAR, SUMEET NAYAK, SWATI IYER, SUSHAMA GADKAR, UDDHAV CHAUDHARI, SANJEEVA D. KHOLKUTE, AND GEETANJALI SACHDEVA* Primate Biology Laboratory, National Institute for Research in Reproductive Health, Indian Council of Medical Research, Mumbai, India
SUMMARY Adhesiveness of the endometrial epithelium to an embryo plays a critical role in the initiation of pregnancy. Loss or gain of adhesiveness also dictates the potential of endometrial epithelial cells to metastasize, an event that can result from certain genetic insults. A proteomics-based exploration of the ‘‘adhesiveness’’ these epithelial cells was employed that could identify targets that could disrupt embryoendometrium interactions and/or metastasis of endometrial cancer cells. The present study defined the surfactomes of two human endometrial epithelial cell lines known for their differential adhesiveness to embryonic cells. Comparative, two-dimensional electrophoretic analysis of the surfactomes of RL95-2 (exhibiting higher adhesiveness to the embryonic cell line JAr) and HEC-1A (exhibiting reduced adhesiveness to JAr cells) revealed 55 differentially enriched proteins. Of these, 10 proteins were identified by MALDI-TOF/TOF or LCMS/MS. TUBB2C, ADAMTS3, and elongation factor beta were more abundant on the HEC-1A cell surface whereas HSP27, HSPA9, GP96, CRT, Tapasin-ERP57, PDI, and b-actin were more abundant on the RL95-2 cell surface. Nano LCMS/MS was also employed to generate a more comprehensive surfactomes of RL95-2 and HEC-1A. The study also demonstrated a pro-adhesive role of CRT and HSPA9 and an anti-adhesive role of TUBB2C populations found on the cell surface. In brief, this study identifies the cell-surface protein complements of two human endometrial epithelial cell lines, and reveals the role of three proteins in heterotypic cell adhesion. Mol. Reprod. Dev. 81: 326340, 2014. ß 2014 Wiley Periodicals, Inc. Received 9 December 2013; Accepted 6 January 2014
INTRODUCTION The plasma membrane and cell surface together serve as an interface between the outer and inner environment of a eukaryotic cell. This interface receives a variety of external stimuli, including hormones, drugs, and infections, and transduces the resulting signals to the cytoplasm or nucleus through specific mediators (Wong and Gough, 2009; Bu and Callaway, 2011; Nussinov, 2012). The plasma membrane also facilitates the bulk transport of various substances into and out of a cell (Allison and Davies,
ß 2014 WILEY PERIODICALS, INC.
Corresponding author: National Institute for Research in Reproductive Health JM Street, Parel Mumbai-400012, India. E-mail: [email protected]
The authors declare that there is no conflict of interest. Grant sponsor: Indian Council of Medical Research and Department of Biotechnology, Government of India
Published online 5 February 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/mrd.22301
1974; Gray et al., 2012). Each of these functionsthat is, defense, selective permeability, transport, and signal transductionis executed by specific proteins and/or Abbreviations: 2D, two-dimensional; ADAMTS3, a disintegrin-like metalloprotease domain with thrombospondin type 1 motif 3; CRT, calreticulin; ERP27, endoplasmic-reticulum resident protein 27; GO, gene ontology; GP96, heatshock glycoprotein 96; HSP, heat-shock protein; HSPA9, heat shock 70-kD protein 9B; LCMS/MS, liquid chromatography mass spectrometry, in tandem; MALDI-TOF/TOF, matrix-assisted laser desorption ionization time-of-flight, in tandem; PDI, protein disulfide isomerase; TUBB2C, tubulin beta 2C.
SURFACTOMES
OF
protein complexes that are present on the cell surface, and the ultimate physiological activity of a cell is dictated by the exact cell-surface protein repertoire defined at its plasma membrane. The endometrial epithelium is the first maternal interface to participate in embryoendometrium interactions (Denker, 1993). The ability to adhere with the embryo is acquired by the endometrial epithelium only during the mid-secretory or receptive phase, which spans Days 1924 of the human menstrual cycle (Harper, 1992; Paria et al., 2000). Deficits in this interaction contribute significantly to implantation failures or infertility (Revel, 2012; Salamonsen et al., 2013). Various adhesion moleculessuch as integrins, selectins, cadherins, mucins (Achache and Revel, 2006), heparan sulfate proteoglycans (Kirn-Safran et al., 2008), ezrin/radixin/moesin (ERM), ERM-associated adhesion molecules (Matsumoto et al., 2004), and trophinin-tastin-bystin complex (Aoki and Fukuda, 2000) have been identified as differentially expressed factors during the receptive phase. We recently completed an integrated analysis of the available transcriptome-based datasets, and identified 179 genes as putative endometrial receptivity-associated. Interestingly 62.25% of these genes encode cell-surface and extracellular-matrix proteins (Bhagwat et al., 2013). Differential expression of cell-surface proteins during the uterine receptive phase can be easily rationalized in view of their indispensable functions in adhesion, signal transduction, cellcell communication, inflammation, and immune response, although some have been further implicated in pathologies such as endometriosis and endometrial cancer (Kao et al., 2003; Pfeifer et al., 2010). Thus, studies on the identification of the cell-surface protein complements of endometrial epithelial cells could lead to the discovery of proteins critical for embryoendometrium interactions during pregnancy. Another outcome of such investigations may be the identification of proteins that facilitate heterotypic cell adhesion, which is frequently observed during the metastasis of cancerous cells. The present study was undertaken to identify the cell surfactomes of two endometrial epithelial cell lines that are routinely used as in vitro models because of their differential affinity for embryonic-cell adhesion. RL95-2 is a human endometrial epithelial cell line that exhibits 80% adhesiveness to trophoblast cells whereas HEC-1A shows only 40% adhesiveness (John et al., 1993). Previous studies comparing these cells reported that RL95-2 cells have higher levels of both Plexin B1 and c-Met, which form a complex of membrane receptors (Harduf et al., 2007; Harduf et al., 2009), and Lewis Y, a difucosylated oligosaccharide, as well as fucosyltransferase 4 (FUT4), the enzyme required for Lewis Y synthesis (Liu et al., 2012), as compared to HEC1A. CD98 and CD147 were also found to be differentially expressed in RL95-2 and HEC-1A (Dominguez et al., 2010). Endeavors have not been made to generate the cell surfactomes of RL95-2 and HEC-1A, however. Thus, the present study used proteomic-based approaches to inventory the cell-surface proteome of RL95-2 and HEC-1A, and to identify some proteins that have a functional role in cell adhesion.
Mol. Reprod. Dev. 81:326–340 (2014)
ENDOMETRIAL CELL LINES DIFFERING
IN
EMBRYO ADHESIVENESS
RESULTS Two-Dimensional Cell Surface Proteomes of RL95-2 and HEC-1A Six cell-surface protein extracts (n ¼ 3 each for RL95-2 and HEC-1A) were used for developing the proteomes using two-dimensional (2D) gel electrophoresis. Cellsurface protein isolation and 2D electrophoresis were done simultaneously for RL95-2 and HEC-1A in each experiment. Analysis of the 2D proteomes revealed the presence of at least 194 protein spots in RL95-2 and at least 244 protein spots in HEC-1A. One-hundred-thirty-eight spots were paired in RL95-2 and HEC-1A (Fig. 1). Only those proteins that showed significant change (>2-fold) in abundance across all three pairs of RL95-2 and HEC-1A were considered for further analysis. Compared to HEC1A, RL95-2 had a higher abundance of 37 protein spots and reduced abundance of 18 spots in its cell- surface-proteinenriched fraction. Ten differentially abundant proteins were identified by matrix-assisted laser desorption ionization time-of-flight, in tandem (MALDI-TOF/TOF) or liquid chromatography mass specometry, in tandem (LCMS/MS) (Tables 1 and 2) (their peptide reports are shown as supplementary data). Heatshock protein-glycoprotein 96 (GP96) precursor, heat shock 70-kD protein 9B (Mortalin-2 or (HSPA9), calreticulin precursor (CRT), chain A tapasin endoplasmic-reticulum resident protein 57 (ERP57), protein disulfide isomerase (PDI), b-actin, and HSP27 were more abundant on the cell surface of RL95-2 whereas a disintegrin-like metalloprotease domain with thrombospondin type 1 motif 3 (ADAMTS3), TUBB2C, and elongation factor 1 beta were more abundant in the cell surface extract of HEC-1A.
Validation of Select Proteins Immunoprobing the cell-surface-protein-enriched fractions revealed higher abundance of HSPA9 and CRT in RL95-2 compared to HEC-1A (Fig. 2), which corroborated with the 2D electrophoresis data. HSPA9 migrated at the predicted 73 kDa. CRT, predicted to be a 48-kDa protein, migrated at 60 kDa, which has been attributed to a highly charged C-terminal domain and acidity (pI-4.1) (Fliegel et al., 1989). In addition to the band at the expected size, immunoreactivity for TUBB2C was observed at a lower molecular weight (45 kDa) in all three HEC-1A cell-surfaceprotein extracts, but not in any from RL95-2, suggestive of a different form of TUBB2C on the HEC-1A cell surface. Immunoprobing of total cell lysates further revealed significantly higher levels of CRT in RL95-2 compared to HEC-1A, whereas total HSPA9 and TUBB2C levels did not differ in the two cell lines (Fig. 3). This implicates differential trafficking of HSPA9 and TUBB2C to the cell surface in RL95-2 and HEC-1A.
Cell-Surface Localization by Immunofluorescence Immunofluorescence studies were carried out to investigate if CRT, HSPA9, and TUBB2C are indeed localized at the cell surface. We initially intended to conduct these
327
Molecular Reproduction & Development
BHAGWAT
ET AL.
Figure 1. 2D proteomes of cell-surface-protein extracts of RL95-2 and HEC-1A. Cell-surface proteins were first resolved by isoelectric point in the pH range of 47, and then by 10% SDS-PAGE. Images of the silver-stained gels were analyzed for detection of the differentially abundant proteins. Encircled spots indicate the proteins that were more abundant in the cell-surface extract of one cell line as compared to the other. [Color figure can be seen in the online version of this article, available at http://wileyonlinelibrary.com/journal/mrd]
experiments in non-permeabilized cells, but despite using various protocols, permeabilization of RL95-2 cells could not be prevented. This indirectly suggested that the molecular cell-surface architecture of the of two cell lines differs. Thus, we performed immunoflourescence studies
in permeabilized RL95-2 and HEC-1A cells (Fig. 4), and used confocal-microscopy-based optical sections to distinguish cell-surface versus intracellular immunoreactivity (Supplementary Fig. S1). Immunoreactivity of each candidate protein was evident in lateral and apical
TABLE 1. Identity of Proteins Displaying Higher Abundance in RL95-2 Cell-Surface Protein Extract compared to HEC-1A Cell-Surface Extract Spot number C19 B10
Identity Heat Shock Protein gp96 Precursor Heat Shock 70 kD Protein 9B (Mortalin-2) Fragment
Accession number
Mowse score
Observed molecular weight (Da)
pI
GI:15010550
381
90138.10
4.73
GI:21040386
258
73588.70
5.87
B9
Calreticulin precursor human
GI:117501
118
60000.00a
4.30
D4
GI:220702506
114
54541.00
5.60
GI:2507461
277
56746.80
5.98
C8
Chain A, TapasinERP57 Heterodimer Protein disulfide-isomerase (EC 5.3.4.1) ER60 precursor Actin beta
GI:46397333
290
40978.40
5.56
B11
Heat shock protein 27
GI:19855073
88
22768.50
5.98
C13
Previous reports indicating cell surface localization in primary tissues and cell lines Wiest et al. (1997); Robert et al. (1999); Dai et al. (2007) Sapozhnikov et al. (2002); Gastpar et al. (2004); Melendez et al. (2006); Sedlackova et al. (2009) White et al. (1995); Gardai et al. (2005); Gold et al. (2010) Sadasivan et al. (1996); Vigneron et al. (2009) Zai et al. (1999); Root et al. (2004); Bi et al. (2011) Miles et al. (2006); Moroianu et al. (1993) Bausero et al. (2004); Garcia-Arguinzonis et al. (2010)
a
Deviation from the expected molecular weight (48 kDa) was due to highly charged C-terminal domain and acidity of calreticulin.
328
Mol. Reprod. Dev. 81:326–340 (2014)
SURFACTOMES
OF
ENDOMETRIAL CELL LINES DIFFERING
IN
EMBRYO ADHESIVENESS
TABLE 2. Identity of Proteins Displaying Higher Abundance in the Cell Surface of HEC-1A as Compared to that in the RL95-2 Spot number D12
Identity
Accession number
Mowse score
Molecular weight (Da) a
pI
Previous reports indicating cell surface localization in primary tissues and cell lines
GI:13183078
33
88897.00
5.63
Hall et al. (2003)
D9
A disintegrin-like and metalloprotease domain with thrombospondin type I motifs-like 3 Tubulin beta-2c chain
GI:74728419
552
49840.00
4.83
D14
Elongation factor 1-beta
GI:4503477
52
24919.00
4.50
Rubin et al. (1982); Bernier-Valentin et al. (1983); Quillen et al. (1985) Li et al. (2011); Itagaki et al. (2012)
a Deviation from the theoretical molecular weight (MW) or pI, predicted from the full-length amino acid sequence of the respective protein. These deviations may be attributed to posttranslational modifications.
membranes, thus confirming the presence of CRT, HSPA9, and TUBB2C on the cell surface of these endometrial epithelial cell lines.
In Vitro JAr Spheroid Attachment Assay Spheroids of trophectoderm-derived JAr cells were next used to assess the contribution of these cell-surface proteins in the adhesivity of each endometrial epithelial cell line. As expected, JAr spheroids preferentially attached to RL95-2 (80 5% as compared to 40 6% attachment to HEC-1A) (Fig. 5A,B, no-pretreatment controls). Pretreatment of RL95-2 with antibodies against HSPA9 or CRT significantly (P < 0.0001) reduced the percent of spheroids attached, whereas pre-treatment with antibodies
against TUBB2C significantly (P < 0.00001) increased the percent of spheroids attached to HEC-1A (Fig. 5). Treatment with non-immune rabbit or mouse IgGs, at the same concentration as used for blocking antibodies against CRT, HSPA9, or TUBB2C, did not affect the ability of JAr spheroids to attach to RL95-2 or HEC-1A cells, These observations thus hint at a pro-adhesive function of the cell-surface-localized CRT and HSPA9, and an antiadhesive function of the cell-surface-localized TUBB2C.
Nano-LCMS/MS-Generated Cell-Surface-Protein Repertoires Nano LCMS/MS analysis of the cell-surface-proteinenriched fractions identified approximately 487 proteins,
Figure 2. Detection of CRT (A), HSPA9 (B), and TUBB2C (C) in the cell-surface-protein enriched fractions of RL95-2 and HEC-1A cells. Cellsurface proteins were extracted from RL95-2 and HEC-1A growing at three different passage numbers (R1R3, H1H3). A representative blot stained with Coomassie blue after the chemiluminescent detection is shown in panel D to display total cell-surface-protein load in each lane. Ratio of the intensity of the band of interest to the total-protein load was calculated for all three proteins to obtain relative abundance (E). For TUBB2C, intensities were calculated for the band detected at the expected size as well as of the band detected at lower molecular weight, indicated by an arrow. P < 0.01, P < 0.001, P < 0.0001.
Mol. Reprod. Dev. 81:326–340 (2014)
329
Molecular Reproduction & Development
BHAGWAT
ET AL.
Figure 3. Detection of CRT (A), HSPA9 (B), and TUBB2C (C) in the total-cell lysates of RL95-2 and HEC-1A. Cell lysates were isolated from RL952 and HEC-1A growing at three different passage numbers (R1R3, H1H3). Blots were reprobed for GAPDH, as a loading control. Ratio of the intensity of the band of interest to that of GAPDH was calculated for each sample to obtain the relative abundance (D). CRT levels were significantly different in the total protein lysates between RL95-2 and HEC-1A. P < 0.001.
including the proteins present in either RL95-2 or HEC-1A or both (Supplementary Table 1). One-hundred ninety-five proteins were common between the cell-surface-protein repertoires of RL95-2 and HEC-1A (Supplementary Table 2). Peak-area quantification of the identified proteins in RL95-2 and HEC-1A revealed the presence of 245 proteins only in RL95-2 cell surface extract (Supplementary Table 3) and 47 only in HEC-1A surface extract (Supplementary Table 4). Interestingly, when compared with the cell-surface-enriched fraction of DU145 (a prostate cancer cell line), 178 (137 present only in RL95-2 and 21 only in HEC-1A) proteins were found exclusively in the endometrial epithelial cell lines (Supplementary Tables 5A and 5B). Twenty proteins common to RL95-2 and HEC-1A, were not detected in the DU-145 cell surface extract, thus it is likely that some of these cell-surface proteins have an endometrium-specific function. Out of the 487 proteins detected on the cell surface of either of both endometrial epithelial cell lines, 120 proteins were found to have a signal sequence; 130 had extracellular transmembrane domains; and 276 had helical transmembrane domains (Supplementary Table 1), as predicted by TMPred (http://www.ch.embnet.org). Gene ontology (GO) analysis of the proteins found only in RL95-2 (and not in HEC-1A) was carried out using the Database for Annotation, Visualization and Integrated Discovery (DAVID), revealing that the enriched proteins were predicted to reside at the plasma membrane or were integral to/ intrinsic to membranes (Fig. 6A), and to function in cell adhesion, intracellular transport, and protein transport (Fig. 6B).
330
DISCUSSION Embryo-endometrial adhesion is considered a rate-limiting step in the success of human pregnancy. Ethical constraints in obtaining human gestational tissues limit the opportunities to identify the factors that mediate embryo-endometrium adhesion in vivo, so researchers tend to use either animals or in vitro embryo-endometrium co-cultures as experimental models (Tan et al., 2005; Teklenburg and Macklon, 2009). The transformed, human endometrial epithelial cell lines RL95-2 and HEC-1A, which differ in their adhesiveness towards trophoblastic cells, have also been used to simulate the response of highand low-adhesive endometrial epithelium, respectively (Sato et al., 1992; John et al., 1993; Thie et al., 1995). Indeed, there are caveats inherent in employing these transformed cell lines as models for adhesive or lessadhesive endometrial epithelium, but it was reassuring to note that both endometrial epithelial cell lines, especially RL95-2, share the molecular phenotype of receptive-phase primary endometrial epithelium (Dominguez et al., 2010; Bhagwat et al., 2013). Few investigations have been conducted to identify the factors contributing to differential adhesiveness of RL95-2 and HEC-1A lines; of these studies, all focused on select factors based on a priori knowledge about their adhesive roles in other cell types. For example, integrin alpha 6, plexin, moesin, and ezrin receptor were found to be differentially abundant in the two cell lines (Thie et al., 1995; Martin et al., 2000; Harduf et al., 2007). CD98, CD36, and thrombospondin were also more abundant in the
Mol. Reprod. Dev. 81:326–340 (2014)
SURFACTOMES
OF
ENDOMETRIAL CELL LINES DIFFERING
IN
EMBRYO ADHESIVENESS
Figure 4. Immunofluorescence detection of CRT, HSPA9, and TUBB2C in RL95-2 and HEC-1A cell lines. In each panel, (a) represents immunofluorescent detection with the respective antibody, (b) represents DAPI-stained nuclei, (c) is a phase-contrast image, and (d) is the merged images of DAPI with immunostaining. Negative controls, stained with rabbit or mouse IgGs, are shown in the insets. Scale bars, 20 mm.
Mol. Reprod. Dev. 81:326–340 (2014)
331
Molecular Reproduction & Development
BHAGWAT
ET AL.
Figure 5. In vitro JAr spheroid-attachment assay assessing the role of cell-surface localized CRTand HSPA9 in RL95-2 (A) and of TUBB2C in HEC1A (B). Cells were pre-treated with antibodies for 2 hr, and then tested for the ability of JAr spheroids to adhere to the monoloayer. Cells without any pre-treatment (control) or treatment with the same concentration of irrelevant rabbit IgG (as control for polyclonal antibodies against CRTor HSPA9) or mouse IgG (as control for monoclonal antibodies against TUBB2C) were also investigated for their ability to bind with spheroids. P < 0.00001, P < 0.0001.
more-adhesive RL95-2 line compared to HEC-1A, as well as more abundant in receptive-phase (more adhesive phase) as compared to pre-receptive-phase (a nonadhesive phase) human endometrial tissues (Dominguez et al., 2010; Bhagwat et al., 2013). This molecular similarity prompted us to employ proteomics approaches to identify cell-surface proteins that differ qualitatively or quantitatively in RL95-2 and HEC-1A cell lines. The investigation was expected to generate a catalogue of the cell-surface proteins of endometrial epithelium, but considering that RL952 (a grade-2, moderately differentiated adenosquamous endometrial carcinoma) and HEC-1A (stage 1A endometrial adenocarcinoma) are of different tumor grades with differences in their growth and metastatic potential, the catalog of cell-surface proteins could also facilitate the design/discovery of novel and efficacious antimetastatic targets in endometrial cancer, the seventh most common cancer worldwide among women (Banno et al., 2012).
The Surfactomes of RL95-2 and HEC-1A In the present study, approximately 487 cell-surface proteins were identified among the two endometrial epithelial lines using a label-free and gel-free Nano-LCMS/MS approach. GO classification revealed enrichment of 100 cell-surface proteins in categories such as the plasma membrane, cell surface, and apical part of the cell. More intriguing, however, was the presence of some intracellular proteinssuch as histones, ribosomal proteins, and nuclear pore proteinssince it is feasible that subset of these proteins are localized on the cell surface, probably through exocytosis or their association with ecto- or cytoplasmic domains of some cell-surface proteins.
332
RL95-2 and HEC-1A shared 195 proteins on their cell surface. These common proteins may perform housekeeping functions, required for the upkeep of endometrial epithelial cells. Among the differentially expressed populations, 245 proteins were detected only in RL95-2 and 47 only in HEC-1A. Nano LCMS/MS data showed the enrichment of calcium-transporting ATPase 4 isoforms XD, XA, ZA, XK, ZK, XB, ZB, and ZD; ephrin receptor; and protocadherin on the RL95-2 cell surface versus a predominance of integrins alpha 5 and beta 3 on the HEC-1A cell surface. Catenin delta 1, neural cell adhesion molecule L1, cathepsin B, HSP70A/B, plexin-A1, sec61 subunit alpha, b2 microglobulin were all highly enriched on the surface of RL95-2 cells, which was made more conspicuous by their absence in the HEC-1A cell-surface protein fraction. Our 2D proteomic data revealed differential abundance of HSPA9, HSP27, tapasin ERP27 heterodimer, glycoprotein 96 (GP96), CRT, PDI, b-actin, TUBB2C, ADAMTS3, and elongation factor 1 beta in the cell surface fractions of RL95-2 and HEC-1A. The majority of these proteins (e.g., CRT, HSP27, HSPA9, PDI, and TUBB2C) were also detected in the cell-surface-enriched fractions by Nano LCMS/MS. Of the 10 proteins identified by 2D electrophoresis, three (i.e., HSPA9, HSP27, and GP96) are members of the heat-shock protein (HSP) family.
Heat-Shock Proteins on the Cell Suface HSPs are not considered conventional integral surface proteins as they lack a transmembrane domain. Cyto:: plasmic HSPs are known to facilitate the folding of naıve proteins and the refolding of denatured proteins (Lancaster and Febbraio, 2005). Yet, some of the HSPs such as GP96, HSP70, BiP, and HSP90 have been reported to be present
Mol. Reprod. Dev. 81:326–340 (2014)
SURFACTOMES
OF
ENDOMETRIAL CELL LINES DIFFERING
IN
EMBRYO ADHESIVENESS
Figure 6. Gene ontology analysis of the cell-surface proteins detected only in RL95-2 (not detected in HEC-1A), on the basis of (A) cellular component and (B) biological processes. Only those GO annotations which had a significant P-value
Cell Surfactomes of Two Endometrial Epithelial Cell Lines That Differ in Their Adhesiveness to Embryonic Cells SONALI R. BHAGWAT, TEJASHREE REDIJ, KRUTTIKA PHALNIKAR, SUMEET NAYAK, SWATI IYER, SUSHAMA GADKAR, UDDHAV CHAUDHARI, SANJEEVA D. KHOLKUTE, AND GEETANJALI SACHDEVA* Primate Biology Laboratory, National Institute for Research in Reproductive Health, Indian Council of Medical Research, Mumbai, India
SUMMARY Adhesiveness of the endometrial epithelium to an embryo plays a critical role in the initiation of pregnancy. Loss or gain of adhesiveness also dictates the potential of endometrial epithelial cells to metastasize, an event that can result from certain genetic insults. A proteomics-based exploration of the ‘‘adhesiveness’’ these epithelial cells was employed that could identify targets that could disrupt embryoendometrium interactions and/or metastasis of endometrial cancer cells. The present study defined the surfactomes of two human endometrial epithelial cell lines known for their differential adhesiveness to embryonic cells. Comparative, two-dimensional electrophoretic analysis of the surfactomes of RL95-2 (exhibiting higher adhesiveness to the embryonic cell line JAr) and HEC-1A (exhibiting reduced adhesiveness to JAr cells) revealed 55 differentially enriched proteins. Of these, 10 proteins were identified by MALDI-TOF/TOF or LCMS/MS. TUBB2C, ADAMTS3, and elongation factor beta were more abundant on the HEC-1A cell surface whereas HSP27, HSPA9, GP96, CRT, Tapasin-ERP57, PDI, and b-actin were more abundant on the RL95-2 cell surface. Nano LCMS/MS was also employed to generate a more comprehensive surfactomes of RL95-2 and HEC-1A. The study also demonstrated a pro-adhesive role of CRT and HSPA9 and an anti-adhesive role of TUBB2C populations found on the cell surface. In brief, this study identifies the cell-surface protein complements of two human endometrial epithelial cell lines, and reveals the role of three proteins in heterotypic cell adhesion. Mol. Reprod. Dev. 81: 326340, 2014. ß 2014 Wiley Periodicals, Inc. Received 9 December 2013; Accepted 6 January 2014
INTRODUCTION The plasma membrane and cell surface together serve as an interface between the outer and inner environment of a eukaryotic cell. This interface receives a variety of external stimuli, including hormones, drugs, and infections, and transduces the resulting signals to the cytoplasm or nucleus through specific mediators (Wong and Gough, 2009; Bu and Callaway, 2011; Nussinov, 2012). The plasma membrane also facilitates the bulk transport of various substances into and out of a cell (Allison and Davies,
ß 2014 WILEY PERIODICALS, INC.
Corresponding author: National Institute for Research in Reproductive Health JM Street, Parel Mumbai-400012, India. E-mail: [email protected]
The authors declare that there is no conflict of interest. Grant sponsor: Indian Council of Medical Research and Department of Biotechnology, Government of India
Published online 5 February 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/mrd.22301
1974; Gray et al., 2012). Each of these functionsthat is, defense, selective permeability, transport, and signal transductionis executed by specific proteins and/or Abbreviations: 2D, two-dimensional; ADAMTS3, a disintegrin-like metalloprotease domain with thrombospondin type 1 motif 3; CRT, calreticulin; ERP27, endoplasmic-reticulum resident protein 27; GO, gene ontology; GP96, heatshock glycoprotein 96; HSP, heat-shock protein; HSPA9, heat shock 70-kD protein 9B; LCMS/MS, liquid chromatography mass spectrometry, in tandem; MALDI-TOF/TOF, matrix-assisted laser desorption ionization time-of-flight, in tandem; PDI, protein disulfide isomerase; TUBB2C, tubulin beta 2C.
SURFACTOMES
OF
protein complexes that are present on the cell surface, and the ultimate physiological activity of a cell is dictated by the exact cell-surface protein repertoire defined at its plasma membrane. The endometrial epithelium is the first maternal interface to participate in embryoendometrium interactions (Denker, 1993). The ability to adhere with the embryo is acquired by the endometrial epithelium only during the mid-secretory or receptive phase, which spans Days 1924 of the human menstrual cycle (Harper, 1992; Paria et al., 2000). Deficits in this interaction contribute significantly to implantation failures or infertility (Revel, 2012; Salamonsen et al., 2013). Various adhesion moleculessuch as integrins, selectins, cadherins, mucins (Achache and Revel, 2006), heparan sulfate proteoglycans (Kirn-Safran et al., 2008), ezrin/radixin/moesin (ERM), ERM-associated adhesion molecules (Matsumoto et al., 2004), and trophinin-tastin-bystin complex (Aoki and Fukuda, 2000) have been identified as differentially expressed factors during the receptive phase. We recently completed an integrated analysis of the available transcriptome-based datasets, and identified 179 genes as putative endometrial receptivity-associated. Interestingly 62.25% of these genes encode cell-surface and extracellular-matrix proteins (Bhagwat et al., 2013). Differential expression of cell-surface proteins during the uterine receptive phase can be easily rationalized in view of their indispensable functions in adhesion, signal transduction, cellcell communication, inflammation, and immune response, although some have been further implicated in pathologies such as endometriosis and endometrial cancer (Kao et al., 2003; Pfeifer et al., 2010). Thus, studies on the identification of the cell-surface protein complements of endometrial epithelial cells could lead to the discovery of proteins critical for embryoendometrium interactions during pregnancy. Another outcome of such investigations may be the identification of proteins that facilitate heterotypic cell adhesion, which is frequently observed during the metastasis of cancerous cells. The present study was undertaken to identify the cell surfactomes of two endometrial epithelial cell lines that are routinely used as in vitro models because of their differential affinity for embryonic-cell adhesion. RL95-2 is a human endometrial epithelial cell line that exhibits 80% adhesiveness to trophoblast cells whereas HEC-1A shows only 40% adhesiveness (John et al., 1993). Previous studies comparing these cells reported that RL95-2 cells have higher levels of both Plexin B1 and c-Met, which form a complex of membrane receptors (Harduf et al., 2007; Harduf et al., 2009), and Lewis Y, a difucosylated oligosaccharide, as well as fucosyltransferase 4 (FUT4), the enzyme required for Lewis Y synthesis (Liu et al., 2012), as compared to HEC1A. CD98 and CD147 were also found to be differentially expressed in RL95-2 and HEC-1A (Dominguez et al., 2010). Endeavors have not been made to generate the cell surfactomes of RL95-2 and HEC-1A, however. Thus, the present study used proteomic-based approaches to inventory the cell-surface proteome of RL95-2 and HEC-1A, and to identify some proteins that have a functional role in cell adhesion.
Mol. Reprod. Dev. 81:326–340 (2014)
ENDOMETRIAL CELL LINES DIFFERING
IN
EMBRYO ADHESIVENESS
RESULTS Two-Dimensional Cell Surface Proteomes of RL95-2 and HEC-1A Six cell-surface protein extracts (n ¼ 3 each for RL95-2 and HEC-1A) were used for developing the proteomes using two-dimensional (2D) gel electrophoresis. Cellsurface protein isolation and 2D electrophoresis were done simultaneously for RL95-2 and HEC-1A in each experiment. Analysis of the 2D proteomes revealed the presence of at least 194 protein spots in RL95-2 and at least 244 protein spots in HEC-1A. One-hundred-thirty-eight spots were paired in RL95-2 and HEC-1A (Fig. 1). Only those proteins that showed significant change (>2-fold) in abundance across all three pairs of RL95-2 and HEC-1A were considered for further analysis. Compared to HEC1A, RL95-2 had a higher abundance of 37 protein spots and reduced abundance of 18 spots in its cell- surface-proteinenriched fraction. Ten differentially abundant proteins were identified by matrix-assisted laser desorption ionization time-of-flight, in tandem (MALDI-TOF/TOF) or liquid chromatography mass specometry, in tandem (LCMS/MS) (Tables 1 and 2) (their peptide reports are shown as supplementary data). Heatshock protein-glycoprotein 96 (GP96) precursor, heat shock 70-kD protein 9B (Mortalin-2 or (HSPA9), calreticulin precursor (CRT), chain A tapasin endoplasmic-reticulum resident protein 57 (ERP57), protein disulfide isomerase (PDI), b-actin, and HSP27 were more abundant on the cell surface of RL95-2 whereas a disintegrin-like metalloprotease domain with thrombospondin type 1 motif 3 (ADAMTS3), TUBB2C, and elongation factor 1 beta were more abundant in the cell surface extract of HEC-1A.
Validation of Select Proteins Immunoprobing the cell-surface-protein-enriched fractions revealed higher abundance of HSPA9 and CRT in RL95-2 compared to HEC-1A (Fig. 2), which corroborated with the 2D electrophoresis data. HSPA9 migrated at the predicted 73 kDa. CRT, predicted to be a 48-kDa protein, migrated at 60 kDa, which has been attributed to a highly charged C-terminal domain and acidity (pI-4.1) (Fliegel et al., 1989). In addition to the band at the expected size, immunoreactivity for TUBB2C was observed at a lower molecular weight (45 kDa) in all three HEC-1A cell-surfaceprotein extracts, but not in any from RL95-2, suggestive of a different form of TUBB2C on the HEC-1A cell surface. Immunoprobing of total cell lysates further revealed significantly higher levels of CRT in RL95-2 compared to HEC-1A, whereas total HSPA9 and TUBB2C levels did not differ in the two cell lines (Fig. 3). This implicates differential trafficking of HSPA9 and TUBB2C to the cell surface in RL95-2 and HEC-1A.
Cell-Surface Localization by Immunofluorescence Immunofluorescence studies were carried out to investigate if CRT, HSPA9, and TUBB2C are indeed localized at the cell surface. We initially intended to conduct these
327
Molecular Reproduction & Development
BHAGWAT
ET AL.
Figure 1. 2D proteomes of cell-surface-protein extracts of RL95-2 and HEC-1A. Cell-surface proteins were first resolved by isoelectric point in the pH range of 47, and then by 10% SDS-PAGE. Images of the silver-stained gels were analyzed for detection of the differentially abundant proteins. Encircled spots indicate the proteins that were more abundant in the cell-surface extract of one cell line as compared to the other. [Color figure can be seen in the online version of this article, available at http://wileyonlinelibrary.com/journal/mrd]
experiments in non-permeabilized cells, but despite using various protocols, permeabilization of RL95-2 cells could not be prevented. This indirectly suggested that the molecular cell-surface architecture of the of two cell lines differs. Thus, we performed immunoflourescence studies
in permeabilized RL95-2 and HEC-1A cells (Fig. 4), and used confocal-microscopy-based optical sections to distinguish cell-surface versus intracellular immunoreactivity (Supplementary Fig. S1). Immunoreactivity of each candidate protein was evident in lateral and apical
TABLE 1. Identity of Proteins Displaying Higher Abundance in RL95-2 Cell-Surface Protein Extract compared to HEC-1A Cell-Surface Extract Spot number C19 B10
Identity Heat Shock Protein gp96 Precursor Heat Shock 70 kD Protein 9B (Mortalin-2) Fragment
Accession number
Mowse score
Observed molecular weight (Da)
pI
GI:15010550
381
90138.10
4.73
GI:21040386
258
73588.70
5.87
B9
Calreticulin precursor human
GI:117501
118
60000.00a
4.30
D4
GI:220702506
114
54541.00
5.60
GI:2507461
277
56746.80
5.98
C8
Chain A, TapasinERP57 Heterodimer Protein disulfide-isomerase (EC 5.3.4.1) ER60 precursor Actin beta
GI:46397333
290
40978.40
5.56
B11
Heat shock protein 27
GI:19855073
88
22768.50
5.98
C13
Previous reports indicating cell surface localization in primary tissues and cell lines Wiest et al. (1997); Robert et al. (1999); Dai et al. (2007) Sapozhnikov et al. (2002); Gastpar et al. (2004); Melendez et al. (2006); Sedlackova et al. (2009) White et al. (1995); Gardai et al. (2005); Gold et al. (2010) Sadasivan et al. (1996); Vigneron et al. (2009) Zai et al. (1999); Root et al. (2004); Bi et al. (2011) Miles et al. (2006); Moroianu et al. (1993) Bausero et al. (2004); Garcia-Arguinzonis et al. (2010)
a
Deviation from the expected molecular weight (48 kDa) was due to highly charged C-terminal domain and acidity of calreticulin.
328
Mol. Reprod. Dev. 81:326–340 (2014)
SURFACTOMES
OF
ENDOMETRIAL CELL LINES DIFFERING
IN
EMBRYO ADHESIVENESS
TABLE 2. Identity of Proteins Displaying Higher Abundance in the Cell Surface of HEC-1A as Compared to that in the RL95-2 Spot number D12
Identity
Accession number
Mowse score
Molecular weight (Da) a
pI
Previous reports indicating cell surface localization in primary tissues and cell lines
GI:13183078
33
88897.00
5.63
Hall et al. (2003)
D9
A disintegrin-like and metalloprotease domain with thrombospondin type I motifs-like 3 Tubulin beta-2c chain
GI:74728419
552
49840.00
4.83
D14
Elongation factor 1-beta
GI:4503477
52
24919.00
4.50
Rubin et al. (1982); Bernier-Valentin et al. (1983); Quillen et al. (1985) Li et al. (2011); Itagaki et al. (2012)
a Deviation from the theoretical molecular weight (MW) or pI, predicted from the full-length amino acid sequence of the respective protein. These deviations may be attributed to posttranslational modifications.
membranes, thus confirming the presence of CRT, HSPA9, and TUBB2C on the cell surface of these endometrial epithelial cell lines.
In Vitro JAr Spheroid Attachment Assay Spheroids of trophectoderm-derived JAr cells were next used to assess the contribution of these cell-surface proteins in the adhesivity of each endometrial epithelial cell line. As expected, JAr spheroids preferentially attached to RL95-2 (80 5% as compared to 40 6% attachment to HEC-1A) (Fig. 5A,B, no-pretreatment controls). Pretreatment of RL95-2 with antibodies against HSPA9 or CRT significantly (P < 0.0001) reduced the percent of spheroids attached, whereas pre-treatment with antibodies
against TUBB2C significantly (P < 0.00001) increased the percent of spheroids attached to HEC-1A (Fig. 5). Treatment with non-immune rabbit or mouse IgGs, at the same concentration as used for blocking antibodies against CRT, HSPA9, or TUBB2C, did not affect the ability of JAr spheroids to attach to RL95-2 or HEC-1A cells, These observations thus hint at a pro-adhesive function of the cell-surface-localized CRT and HSPA9, and an antiadhesive function of the cell-surface-localized TUBB2C.
Nano-LCMS/MS-Generated Cell-Surface-Protein Repertoires Nano LCMS/MS analysis of the cell-surface-proteinenriched fractions identified approximately 487 proteins,
Figure 2. Detection of CRT (A), HSPA9 (B), and TUBB2C (C) in the cell-surface-protein enriched fractions of RL95-2 and HEC-1A cells. Cellsurface proteins were extracted from RL95-2 and HEC-1A growing at three different passage numbers (R1R3, H1H3). A representative blot stained with Coomassie blue after the chemiluminescent detection is shown in panel D to display total cell-surface-protein load in each lane. Ratio of the intensity of the band of interest to the total-protein load was calculated for all three proteins to obtain relative abundance (E). For TUBB2C, intensities were calculated for the band detected at the expected size as well as of the band detected at lower molecular weight, indicated by an arrow. P < 0.01, P < 0.001, P < 0.0001.
Mol. Reprod. Dev. 81:326–340 (2014)
329
Molecular Reproduction & Development
BHAGWAT
ET AL.
Figure 3. Detection of CRT (A), HSPA9 (B), and TUBB2C (C) in the total-cell lysates of RL95-2 and HEC-1A. Cell lysates were isolated from RL952 and HEC-1A growing at three different passage numbers (R1R3, H1H3). Blots were reprobed for GAPDH, as a loading control. Ratio of the intensity of the band of interest to that of GAPDH was calculated for each sample to obtain the relative abundance (D). CRT levels were significantly different in the total protein lysates between RL95-2 and HEC-1A. P < 0.001.
including the proteins present in either RL95-2 or HEC-1A or both (Supplementary Table 1). One-hundred ninety-five proteins were common between the cell-surface-protein repertoires of RL95-2 and HEC-1A (Supplementary Table 2). Peak-area quantification of the identified proteins in RL95-2 and HEC-1A revealed the presence of 245 proteins only in RL95-2 cell surface extract (Supplementary Table 3) and 47 only in HEC-1A surface extract (Supplementary Table 4). Interestingly, when compared with the cell-surface-enriched fraction of DU145 (a prostate cancer cell line), 178 (137 present only in RL95-2 and 21 only in HEC-1A) proteins were found exclusively in the endometrial epithelial cell lines (Supplementary Tables 5A and 5B). Twenty proteins common to RL95-2 and HEC-1A, were not detected in the DU-145 cell surface extract, thus it is likely that some of these cell-surface proteins have an endometrium-specific function. Out of the 487 proteins detected on the cell surface of either of both endometrial epithelial cell lines, 120 proteins were found to have a signal sequence; 130 had extracellular transmembrane domains; and 276 had helical transmembrane domains (Supplementary Table 1), as predicted by TMPred (http://www.ch.embnet.org). Gene ontology (GO) analysis of the proteins found only in RL95-2 (and not in HEC-1A) was carried out using the Database for Annotation, Visualization and Integrated Discovery (DAVID), revealing that the enriched proteins were predicted to reside at the plasma membrane or were integral to/ intrinsic to membranes (Fig. 6A), and to function in cell adhesion, intracellular transport, and protein transport (Fig. 6B).
330
DISCUSSION Embryo-endometrial adhesion is considered a rate-limiting step in the success of human pregnancy. Ethical constraints in obtaining human gestational tissues limit the opportunities to identify the factors that mediate embryo-endometrium adhesion in vivo, so researchers tend to use either animals or in vitro embryo-endometrium co-cultures as experimental models (Tan et al., 2005; Teklenburg and Macklon, 2009). The transformed, human endometrial epithelial cell lines RL95-2 and HEC-1A, which differ in their adhesiveness towards trophoblastic cells, have also been used to simulate the response of highand low-adhesive endometrial epithelium, respectively (Sato et al., 1992; John et al., 1993; Thie et al., 1995). Indeed, there are caveats inherent in employing these transformed cell lines as models for adhesive or lessadhesive endometrial epithelium, but it was reassuring to note that both endometrial epithelial cell lines, especially RL95-2, share the molecular phenotype of receptive-phase primary endometrial epithelium (Dominguez et al., 2010; Bhagwat et al., 2013). Few investigations have been conducted to identify the factors contributing to differential adhesiveness of RL95-2 and HEC-1A lines; of these studies, all focused on select factors based on a priori knowledge about their adhesive roles in other cell types. For example, integrin alpha 6, plexin, moesin, and ezrin receptor were found to be differentially abundant in the two cell lines (Thie et al., 1995; Martin et al., 2000; Harduf et al., 2007). CD98, CD36, and thrombospondin were also more abundant in the
Mol. Reprod. Dev. 81:326–340 (2014)
SURFACTOMES
OF
ENDOMETRIAL CELL LINES DIFFERING
IN
EMBRYO ADHESIVENESS
Figure 4. Immunofluorescence detection of CRT, HSPA9, and TUBB2C in RL95-2 and HEC-1A cell lines. In each panel, (a) represents immunofluorescent detection with the respective antibody, (b) represents DAPI-stained nuclei, (c) is a phase-contrast image, and (d) is the merged images of DAPI with immunostaining. Negative controls, stained with rabbit or mouse IgGs, are shown in the insets. Scale bars, 20 mm.
Mol. Reprod. Dev. 81:326–340 (2014)
331
Molecular Reproduction & Development
BHAGWAT
ET AL.
Figure 5. In vitro JAr spheroid-attachment assay assessing the role of cell-surface localized CRTand HSPA9 in RL95-2 (A) and of TUBB2C in HEC1A (B). Cells were pre-treated with antibodies for 2 hr, and then tested for the ability of JAr spheroids to adhere to the monoloayer. Cells without any pre-treatment (control) or treatment with the same concentration of irrelevant rabbit IgG (as control for polyclonal antibodies against CRTor HSPA9) or mouse IgG (as control for monoclonal antibodies against TUBB2C) were also investigated for their ability to bind with spheroids. P < 0.00001, P < 0.0001.
more-adhesive RL95-2 line compared to HEC-1A, as well as more abundant in receptive-phase (more adhesive phase) as compared to pre-receptive-phase (a nonadhesive phase) human endometrial tissues (Dominguez et al., 2010; Bhagwat et al., 2013). This molecular similarity prompted us to employ proteomics approaches to identify cell-surface proteins that differ qualitatively or quantitatively in RL95-2 and HEC-1A cell lines. The investigation was expected to generate a catalogue of the cell-surface proteins of endometrial epithelium, but considering that RL952 (a grade-2, moderately differentiated adenosquamous endometrial carcinoma) and HEC-1A (stage 1A endometrial adenocarcinoma) are of different tumor grades with differences in their growth and metastatic potential, the catalog of cell-surface proteins could also facilitate the design/discovery of novel and efficacious antimetastatic targets in endometrial cancer, the seventh most common cancer worldwide among women (Banno et al., 2012).
The Surfactomes of RL95-2 and HEC-1A In the present study, approximately 487 cell-surface proteins were identified among the two endometrial epithelial lines using a label-free and gel-free Nano-LCMS/MS approach. GO classification revealed enrichment of 100 cell-surface proteins in categories such as the plasma membrane, cell surface, and apical part of the cell. More intriguing, however, was the presence of some intracellular proteinssuch as histones, ribosomal proteins, and nuclear pore proteinssince it is feasible that subset of these proteins are localized on the cell surface, probably through exocytosis or their association with ecto- or cytoplasmic domains of some cell-surface proteins.
332
RL95-2 and HEC-1A shared 195 proteins on their cell surface. These common proteins may perform housekeeping functions, required for the upkeep of endometrial epithelial cells. Among the differentially expressed populations, 245 proteins were detected only in RL95-2 and 47 only in HEC-1A. Nano LCMS/MS data showed the enrichment of calcium-transporting ATPase 4 isoforms XD, XA, ZA, XK, ZK, XB, ZB, and ZD; ephrin receptor; and protocadherin on the RL95-2 cell surface versus a predominance of integrins alpha 5 and beta 3 on the HEC-1A cell surface. Catenin delta 1, neural cell adhesion molecule L1, cathepsin B, HSP70A/B, plexin-A1, sec61 subunit alpha, b2 microglobulin were all highly enriched on the surface of RL95-2 cells, which was made more conspicuous by their absence in the HEC-1A cell-surface protein fraction. Our 2D proteomic data revealed differential abundance of HSPA9, HSP27, tapasin ERP27 heterodimer, glycoprotein 96 (GP96), CRT, PDI, b-actin, TUBB2C, ADAMTS3, and elongation factor 1 beta in the cell surface fractions of RL95-2 and HEC-1A. The majority of these proteins (e.g., CRT, HSP27, HSPA9, PDI, and TUBB2C) were also detected in the cell-surface-enriched fractions by Nano LCMS/MS. Of the 10 proteins identified by 2D electrophoresis, three (i.e., HSPA9, HSP27, and GP96) are members of the heat-shock protein (HSP) family.
Heat-Shock Proteins on the Cell Suface HSPs are not considered conventional integral surface proteins as they lack a transmembrane domain. Cyto:: plasmic HSPs are known to facilitate the folding of naıve proteins and the refolding of denatured proteins (Lancaster and Febbraio, 2005). Yet, some of the HSPs such as GP96, HSP70, BiP, and HSP90 have been reported to be present
Mol. Reprod. Dev. 81:326–340 (2014)
SURFACTOMES
OF
ENDOMETRIAL CELL LINES DIFFERING
IN
EMBRYO ADHESIVENESS
Figure 6. Gene ontology analysis of the cell-surface proteins detected only in RL95-2 (not detected in HEC-1A), on the basis of (A) cellular component and (B) biological processes. Only those GO annotations which had a significant P-value
