SEARCH

SEARCH BY CITATION

Keywords:

  • Germ cell tumors;
  • Embryonal carcinoma;
  • Pluripotent;
  • Yolk sac cell;
  • ATRA;
  • BMP-2

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Human embryonal carcinoma (EC) cell lines exhibit considerable heterogeneity in their levels of pluripotency. Thus, NT2/D1 cells differentiate into neural lineages upon exposure to all-trans retinoic acid (ATRA) and non-neural epithelial lineages upon exposure to bone morphogenetic protein-2 (BMP-2). In contrast, 27X-1 cells differentiate into extra-embryonic endodermal (ExE) cells upon treatment with either morphogen. To understand the molecular basis for the differential responses of the two cell lines, we performed gene expression profiling at the undifferentiated EC cell line state to identify constitutive differences in gene expression. NT2/D1 cells preferentially expressed transcripts associated with neurectodermal development, whereas 27X-1 cells expressed high levels of transcripts associated with mesendodermal characteristics. We then determined temporal expression profiles of 27X-1 cells during ExE differentiation upon treatment with ATRA and BMP-2 and compared the data with changes in gene expression observed during BMP-2- and ATRA-induced differentiation of NT2/D1 cells. ATRA and BMP-2 induced distinct sets of transcription factors and phenotypic markers in the two EC cell lines, underlying distinct lineage choices. Although 27X-1 differentiation yielded comprehensive gene expression profiles of parietal endodermal lineages, we were able to use the combined analysis of 27X-1 data with data derived from yolk sac tumors for the identification of transcripts associated with visceral endoderm formation. Our results demonstrate constitutive differences in the levels of pluripotency between NT2/D1 and 27X-1 cells that correlate with lineage potential. This study also demonstrates that EC cells can serve as robust models to investigate early lineage choices during both embryonic and extra-embryonic human development.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

The patterns of morphologic differentiation exhibited by human male germ cell tumors (GCTs) are similar to lineage differentiation undergone by human embryos during early development [1]. Embryonal carcinoma (EC) cell lines derived from GCTs provide a powerful in vitro model to investigate gene expression and lineage choice associated with embryonic [2, [3], [4]5] and extra-embryonic [6, [7]8] development. The value of EC cells as bona fide pluripotent cell systems was further underscored by recent global gene expression data that established common patterns associated with “stemness” and pluripotency between human embryonal stem (ES) cells and EC cells [9, 10]. Unlike ES cells that require defined feeder cells to sustain growth in culture, EC cells are amenable to propagation without feeder support. In addition, they offer the possibility of cross-validating in vitro derived gene expression patterns with expression patterns identified from profiles of in vivo tumor differentiation.

A number of EC cell lines are known to exhibit divergent in vitro responses to the same developmental morphogen. For instance, NT2/D1 cells differentiate into neural [2, 3] and epithelial [4, 5] lineages upon treatment with all-trans retinoic acid (ATRA) and bone morphogenetic protein-2 (BMP-2), respectively, whereas either morphogen induces extra-embryonic endodermal (ExE) differentiation in the parietal endoderm (PE) subcategory in 27X-1 cells [6, [7]8]. Interestingly, both NT2/D1 and 27X-1 express comparable levels of pluripotency-associated transcripts such as POU5F1 and NANOG. Highly distinct differentiation responses between EC cells lines may be because of many differences in the basal levels of expression of other stem cell and developmental markers, resulting in a predisposition of lineage choices. On the other hand, a given morphogen may elicit distinct gene expression changes during differentiation in EC cell lines that have otherwise comparable profiles in their undifferentiated state.

To address the molecular basis for the divergent in vitro responses of NT2/D1 and 27X-1 cells lines, we performed gene expression analyses at the undifferentiated stage to identify constitutive differences in gene expression between the two lines. We next investigated dynamic gene expression profiles during ExE lineage differentiation of 27X-1 cells in response to ATRA and BMP-2 and compared the profiles with those obtained during neural and epithelial differentiation of NT2/D1 cells in response to the same morphogens [2, 4]. Finally, we compared the temporal gene expression profiles associated with PE lineages in 27X-1 cells with expression profiles obtained from yolk sac tumors that included both parietal and visceral components. The combined analysis of 27X-1 differentiation in vitro and yolk sac tumors in vivo allowed us to dissect key transcriptional components distinguishing between PE and visceral endoderm (VE) differentiation. Our data indicate that basal differences in gene expression in EC cell lines predict lineage potential and that such differences are further amplified upon morphogen induced differentiation in vitro. We also demonstrate the use of EC cell lines as a powerful model system to understand both early embryonic and extra-embryonic human development.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Cell Culture and Time Course

27X-1 and NT2/D1 cells were cultured in Dulbecco's modified Eagle's medium (high-glucose) containing 15% heat-inactivated fetal bovine serum, 2 mM l-glutamine, and 100 units/ml penicillin and streptomycin as described previously [11]. For the time course analyses, 27X-1 cells were seeded at 3.0 × 106/10-cm plate and treated on the following day with ATRA (10 μM; Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) or BMP-2 (50 ng/ml; a gift from Wyeth Research, Cambridge, MA, http://www.wyeth.com/research), whereas the untreated cells received only 10 μl of dimethyl sulfoxide. Cells were collected at 2, 4, 8, 12, 18, 24, 48, and 72 hours after treatment for early time points. For the later times (144, 192, 240, 288, and 336 hours), cells were reseeded every 2–3 days in the continued presence of ATRA or BMP-2 and collected at the stated time points. To passage the cells, NT2/D1 and 27X-1 cells were collected by trypsinizing with 0.5% trypsin/sodium citrate solution. The entire time course program was carried out in triplicate, with matched untreated zero time cultures serving as controls.

RNA Isolation and Affymetrix Array Hybridization

RNA isolation, preparation of cRNA, and hybridization were performed as described previously [2, 4] using human genome HG-U133 A and B oligonucleotide arrays (Affymetrix, Santa Clara, CA, http://www.affymetrix.com).

Statistical Analysis of Data

K-Means Clustering.

Criteria used for changes in expression levels were as previously defined and validated [2, 4]. In brief, transcripts were submitted to K-means temporal clustering analysis after satisfying the following criteria: exhibited ≥2-fold change in expression in all three replicates, designated as an increase or decrease, and classified as present in treated for an increase and present in untreated for a decrease. Comparisons between the expression profiles of triplicate untreated control samples yielded no transcripts that obeyed the above criteria. The total number of transcripts that obeyed these criteria along with their temporal clusters is presented as supplemental data (supplemental online Table 1a [ATRA] and Table 1b [BMP-2]). The K-means clustering analysis was performed as previously described by us [4]. In brief, the data were log-transformed, and a cubic B-spline was used for smoothing [12]. The smoothed data for transcripts were split into clusters using the method of K-means [13], which has been adapted for gene expression time series data [14].

For gene expression analysis between NT2/D1 and 27X-1 cells and the human EC and yolk sac tumors, microarray-derived Cel Files (Affymetrix HG133A+B) were background-corrected, and raw data were normalized and log-transformed using the robust multichip analysis (RMA) method [15]. Significance analysis for microarrays (SAM) [16] was performed between the RMA-derived triplicate gene expression data files of NT2/D1 and 27X-1 cells. SAM analysis was also performed between pure embryonal carcinoma versus yolk sac tumors that we had previously profiled (Gene Expression Omnibus accession number GSE 3218) [17]. Our previous studies of time course gene expression changes mediated by ATRA and BMP-2 in NT2/D1 were performed using Affymetrix HGU95Av2 chips [2, 4]. To compare the gene expression changes in NT2/D1 cells obtained by use of Affymetrix HGU95Av2 chips with those of 27X-1 cells obtained by use of Affymetrix HG133A+B chips, the HG133A+B equivalents of relevant probe sets on the HGU95Av2 chips was obtained from the official Web site of Affymetrix (http://www.netaffx.com).

Semiquantitative Reverse Transcription-Polymerase Chain Reaction and Immunocytochemical Analyses

Semiquantitative multiplex reverse transcription-polymerase chain reaction (RT-PCR) analysis was performed as reported previously by us [4] to validate the gene expression changes noted during the 27X-1 time course programs. RARB, ID2, and GATA6 from the time course program mediated by ATRA and TFAP2A from the time course program mediated by BMP-2 (with ACTB as an internal control) were chosen for validation using RNA samples from one of the time course replicates. For validation of gene expression differences between NT2/D1 and 27X-1 cells identified by microarray, semiquantitative RT-PCR of six transcripts that were differentially expressed between the two cell types was performed (CD44, ALCAM, and MEST were expressed higher in 27X-1; NEFH, GLI3, and PROM1 were expressed higher in NT2/D1) on the RNA samples obtained from both the total cell populations and the stage-specific embryonic antigen 3 (SSEA3)-positive cells. Expression levels of POU5F1 were also measured in these samples to serve as a control transcript with comparable levels of expression. Sequences of the primers used in the study are provided as supplemental information (supplemental online Table 2). Identity of the RT-PCR products was verified by sequencing analyses.

Immunocytochemical validation of gene expression alterations was performed as described previously [4]. In brief, cultured cells were washed in phosphate-buffered saline (PBS) before fixation with 4% paraformaldehyde/0.15% picric acid for 15 minutes at room temperature. Cells were then permeabilized with 0.3% Triton X-100/1% bovine serum albumin (BSA) and 5% normal goat serum (NGS) for 1 hour and were incubated with primary antibody for 3 hours in PBS/0.1% BSA/5% NGS. The primary antibodies used were monoclonal antibodies against mouse antigens Oct3/4 (POU5F1) (1:100; Santa Cruz Biotechnology, Santa Cruz, CA, http://www.scbt.com), Troma1 (keratin 8) (1:20; Developmental Studies Hybridoma Bank, Iowa City, IA, http://www.uiowa.edu/∼dshbwww/), monoclonal antibodies against human antigens keratin 7 and keratin 19 (1:100; Biocarta, San Diego, http://www.biocarta.com/), and polyclonal antibodies against nidogen (1:100; Calbiochem, San Diego, http://www.emdbiosciences.com). Anti-CD133/AC133-PE (PROM1) was prelabeled with phycoerythrin (1:300; Mitenyi Biotec Auburn, CA, http://www.miltenyibiotec.com); cells were not permeabilized with Triton X-100 when probed for the surface marker PROM1. Appropriate Alexa-fluorescent dye-labeled secondary antibodies were used for visualization (Invitrogen, Carlsbad, CA, http://www.invitrogen.com).

Flow-Activated Cell Sorting Analysis

NT2/D1 and 27X-1 cells were collected after trypsinization in 0.5% trypsin/sodium citrate and were blocked in 2% BSA/PBS for 10 minutes at room temperature. Cells were then incubated with primary antibodies for 30 minutes at 4°C and were washed two times in PBS/0.1% BSA before incubation with Alexa-488-labeled secondary antibodies. After two washes in PBS/0.1% BSA, cells were analyzed for surface staining using the FACSCalibur system (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com). Primary antibodies used in the analyses were against human antigens SSEA3 (clone MC-631 [1:75]), SSEA4 (clone MC-813 [1:75]) (Developmental Studies Hybridoma Bank), and anti-TRA-1-60 and anti-TRA-1-81 antibodies (Chemicon, Temecula, CA, http://www.chemicon.com). Appropriate secondary antibodies labeled with Alexa-488 were used to stain the cells. To collect SSEA3-positive NT2/D1 and 27X-1 cells, trypsinized cells were washed with PBS/0.1% BSA and were incubated with primary antibody SSEA3 (clone MC-631; Developmental Studies Hybridoma Bank) in PBS/0.1% BSA for 30 minutes at 4°C. After two washes in PBS/0.1% BSA, cells were stained with Alexa-488-labeled secondary antibody (Invitrogen) for 30 minutes and washed with PBS/0.1% BSA. Cells were collected using Moflo Cell Sorter (DakoCytomation, Carpinteria, CA, http://www.dakocytomation.com) and stored in −70°C until further use.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Constitutive Differences in Gene Expression Predict In Vitro Differentiation Response of NT2/D1 and 27X-1 Cells

ATRA and BMP-2, respectively, mediate neural [2, 3] and epithelial [4, 5] differentiation in NT2/D1 cells, whereas both morphogens promote PE differentiation in 27X-1 cells [6, [7]8]. To understand whether constitutive differences in patterns of gene expression between NT2/D1 and 27X-1 cells in part account for the observed differences in morphogen-mediated differential lineage induction in these cells, we determined gene expression profiles of both cell lines in triplicate and compared the relative levels of gene expression by the SAM program. SAM analysis identified 283 and 759 transcripts (Q value cutoff = 0), respectively, that were expressed at ≥2-fold levels in NT2/D1 and 27X-1 cells (supplemental online Tables 3 and 4).

Functional annotation of the 50 most differentially expressed transcripts in NT2/D1 cells compared with 27X-1 cells (Table 1) revealed an abundance of transcripts associated with neurectodermal development. These included genes that function in transcription regulation and signaling (ID2, PBX1, OTX2, LDB2, SFRP1, SMCY, and DUSP23), neurite outgrowth (NEFH, MLLT11 [AF1Q], and SP8), axonal guidance (FEZ1 and NAV1), and neural development (SGNE1, NMA2, OPN3, GNG4, C5ORF13, BEX1, and C7ORF16). A list of remaining transcripts are provided as supplemental information (supplemental online Table 3), which includes several additional neurogenic transcription factors (PAX8, SOX13, GLI3, HKR1, and BHLHB3) and those involved in neural development (NELL2, ARHGEF9, PROM1, EPHA1, BEX2, BEXL1, FGF19, NBEA, ADCY1, SYT6, SYN2, and DCMKL1).

Table Table 1.. Fifty most highly overexpressed transcripts in NT2/D1 cells relative to 27X-1 cells
Thumbnail image of

On the other hand, functional annotation of the 50 most differentially overexpressed transcripts in 27X-1 cells compared with NT2/D1 (Table 2) identified a number of transcripts associated with non-neural developmental pathways such as mesodermal (MEST, FER1L3, CDH11, CD44, COL1A1, COL5A1, and ANXA3) and epithelial (CDH1, SFN, and KRT19) lineages. The remaining transcripts are presented as supplemental data (supplemental online Table 4), which include multiple additional genes involved in epithelial (EPPK1, DSP, PKP3, VCL, AHNAK, and CTNNA1) and mesendodermal (EOMES, DSC96, TAGLN, MYL9, TEAD4, GATA3, MTMR1, CNN1, MYADM, PDLIM1, PDLIM2, and CALD1) development.

Table Table 2.. Fifty most highly overexpressed transcripts in 27X-1 cells relative to NT2/D1
Thumbnail image of

Furthermore, NT2/D1 and 27X-1 cells also differed in the expression of a number of known stem cell markers. Whereas NT2/D1 cells expressed higher levels of SOX2, ECAT8, LEFTY2, and PROM1, 27X-1 cells expressed higher levels of DPPA3 (STELLA), ZFP42, DSC96, and EOMES. SOX2, LEFTY2, and ZFP42 are considered core markers associated with stemness or pluripotentiality in human ES cells [18], whereas PROM1 (CD133) is a known stem cell marker expressed in neural, hematopoietic, and embryonic stem cells [19]. DSC96 and EOMES expressed at higher levels in 27X-1 cells are characteristic of mesenchymal lineages [20]. The two cell types also differed in the expression of transcripts associated with Nodal and Activin signaling, which are known to play a significant role in early embryonic development [21]. Whereas 27X-1 cells expressed higher levels of both NODAL and INHBA (Activin A), NT2/D1 cells expressed a known nodal signaling inhibitor, LEFTY2. Thus, by virtue of constitutive gene expression, NT2/D1 and 27X-1 cells seem predisposed to respond with neural/epithelial and extra-embryonal lineage fates upon treatment with the respective morphogens.

To exclude the possibility that the observed differences in gene expression patterns between the two cell types are due to a contaminating fraction of spontaneously differentiated cells, we first performed flow-activated cell sorting (FACS) analysis of NT2/D1 and 27X-1 cells for four of cell surface stemness markers such as TRA-1-61, TRA-1-80, SSEA3, and SSEA4. The large majority (∼93%–99%) of both cell types expressed all four markers. We next collected the SSEA3-positive populations by FACS, isolated the RNA, and performed semiquantitative RT-PCR analyses for six representative transcripts (GLI3, PROM1, NEFH, CD44, MEST, and ALCAM) that were indicated by microarray analyses as differentially expressed between the total populations of NT2/D1 and 27X-1 cells. The “pluripotentiality/stemness-associated” transcript POU5F1 was included as control, which was expressed at comparable levels between the two cell types based on our array data. The RT-PCR analyses confirmed higher levels of expression of GLI3, NEFH, and PROM1 in NT2/D1 cells. In contrast, MEST, ALCAM, and CD44 were expressed at higher levels in 27X-1 cells. These differences were observed for both the total cell populations and the SSEA3-positive population (supplemental online Fig. 1). The stemness marker POU5F1 was expressed at comparable levels between the two cell types in both total and SSEA3-positive fractions, consistent with the microarray data. Our data thus show that differences in gene expression between NT2/D1 and 27X-1 cells observed by microarray analysis reflect gene expression patterns in the undifferentiated cells and are not due to spontaneous differentiation. Our gene expression data were further corroborated by immunocytochemical data confirming stemness (POU5F1) and proliferation-associated (Ki67) marker expression in nearly all the cells in both cell types (supplemental online Fig. 2). As a representative marker differentially expressed in microarray and RT-PCR analyses, we performed immunocytochemistry for PROM1. Our data revealed higher expression of PROM1 in NT2/D1 cells compared with 27X-1 cells.

ATRA and BMP-2 Induce Distinct Sets of Transcription Factors and Phenotypic Markers in NT2/D1 and 27X-1 Cells

To understand whether ATRA and BMP-2 induce distinct sets of transcription factors and lineage-specific markers in NT2/D1 and 27X-1 cells to modulate respective embryonal versus extra-embryonal lineages in the two cell types, we performed time course gene expression profiling of 27X-1 cells in response to both morphogens and compared them with NT2/D1 expression profiles in response to ATRA [2] and in response BMP-2 [4]. Upon exposure of 27X-1 cells to ATRA and BMP-2, 3,496 transcripts (4,405 probe sets) and 3,371 transcripts (4,269 probe sets), respectively, exhibited a ≥2-fold change in expression in triplicate at least at one time point during each of the course. Notably, 1,671 transcripts (2,089 probe sets) were altered in both time courses, consistent with previous observations that both morphogens induce differentiation of 27X-1 cells towards a PE lineage [6, [7]8]. Early transcriptional targets were distinguished from those regulated subsequently by K-means temporal clustering, altered during the first 48 hours, and those altered during the entire time course. This analysis identified nine and 12 clusters comprising transcripts regulated during the first 48 hours of ATRA and BMP-2 treatment, respectively (supplemental online Fig. 3a and Fig. 3b), and 14 clusters each from the total time course for either morphogen (supplemental online Fig. 3c and Fig. 3d).

As expected, a number of known direct targets of ATRA (e.g., CYP26A1, RARA, RARB, and HOX family members) and BMP-2 (e.g., MSX1, MSX2, and TFAP2A) were induced in the early response clusters in the respective time course programs (Fig. 1A, 1B; supplemental online Fig. 3a [clusters 1 to 3]; supplemental online Fig. 3b [clusters I–III]). However, consistent with the reported role of BMP signaling in ATRA-mediated extra-embryonic endoderm differentiation [22], we noted induction of multiple BMP-responsive genes at later times, mediated by ATRA, which coincided with the induction of BMP7 (e.g., TFAP2A, GATA2, GATA3, ID2, and ID3, Fig. 1B). Furthermore, in agreement with the onset of a differentiation response, a number of key transcripts associated with stemness/pluripotentiality (e.g., NANOG, POU5F1, and UTF1; Fig. 1C) and cell proliferation (e.g., CCNA2, CCNE2, and CDC2; Fig. 1D) were downregulated, whereas multiple cell-cycle inhibitors such as CDKN1A, CDKN2A, and CDKN2B were induced by both morphogens (Fig. 1E).

thumbnail image

Figure Figure 1.. Heat maps of the representative gene expression alterations noted during the time course of ATRA and BMP-2-mediated parietal endoderm differentiation in 27X-1 cells. (A): ATRA-responsive transcripts, (B) BMP-2-responsive transcripts, (C) stemness/pluripotentiality-associated transcripts, (D) cell proliferation-associated transcripts, (E) cell-cycle inhibitors, and (F) transcripts associated with parietal endoderm differentiation. Color scale displayed at the bottom shows the average log fold change of the expression data in triplicate. Abbreviations: ATRA, all-trans retinoic acid; BMP-2, bone morphogenetic protein-2.

Download figure to PowerPoint

Accompanying these changes, we also noticed a robust induction of several transcripts associated with PE differentiation by both morphogens (Fig. 1F). These transcripts belonged to multiple functional categories such as transcription factors and signaling molecules (GATA6, SOX7, CITED2, EPAS1, and DAB2), structural components associated with extracellular matrix (COL4A5, COL4A6, LAMA1, LAMA5, LAMB1, LAMC1, LAMC2, NID, FN1, and HSPG2), post-translational modification and protein processing associated with extracellular matrix (P4HA2, PLOD2, KDELR3, and NAGLU), and molecules secreted by parietal endoderm (PLAT, LCN7, and MDK) (Fig. 1F) [23, [24], [25], [26]27]. In contrast to PE lineage markers, several characteristic transcripts related to VE derivation (e.g., AFP, APOA1, APOB, TTR, HNF4A, HNF3B, and FOXA2) remained unaltered in both morphogen programs, in agreement with previous observations [6]. Phase-contrast images of both the untreated control and the morphogen-induced differentiated 27X-1 cells are provided as supplemental information (supplemental online Fig. 4). We performed semiquantitative RT-PCR and immunocytochemical analyses to validate the array expression data. Whereas RT-PCR confirmed the induction of RARB, ID2, GATA6 (by ATRA), and TFAP2A (by BMP-2) during each of the time course programs, (supplemental online Fig. 5), immunocytochemical analyses detected the loss of POU5F1 expression, accompanied by the induction of several phenotypic markers (keratin-7, keratin-19, keratin-8 [Endo A], and Nidogen) modulated by both morphogens (supplemental online Fig. 6). Single channel images depicted significant POU5F1 levels in nearly all the untreated cells. (supplemental online Fig. 7).

We next compared the gene expression changes associated with ATRA and BMP-2-mediated ExE differentiation in 27X-1 cells with those of neural and epithelial cell fates in NT2/D1 cells, mediated by the same morphogens [2, 4]. Using the same criteria as those used in analyzing changes in 27X-1, ATRA and BMP-2 were found to modulate 915 and 1,197 transcripts, respectively, in NT2/D1 cells during the course of neural [2] and epithelial [4] differentiation.

ATRA induced a number of transcripts associated with neural development in NT2/D1 but not in 27X-1 cells. These transcripts belonged to diverse functional categories such as transcription factors (PAX6, HOXB3, HOXC4, HOXC5, HOXD4, ARNT2, ASCL1 [MASH1], BTEB1, ZIC1, BHLHB2 [DEC1], BACH1, NHLH2, EGR3, NR2F1, SMARCD3, and ZFHX1B), developmental growth factors/receptors (PDGFRA, NRG1, EPHA2, FGF9, DLK1, WNT13, NTRK2, NELL1, TMEFF1, FYN, and BDNF), factors associated with neurite outgrowth/axonal guidance (DCX, NFM, AMPH, RIT1, WASF1, WASF3, PFN2, NRP2, ENPP2, SLIT2, CRMP1, and TMSNB), and miscellaneous other genes involved in neural development (PCP4, BAI1, PENK, NCALD, SH3GL3, and CBLN).

Likewise, BMP-2 induced a number of transcripts characteristic of epithelial (CDH1, TJP3, CTNNA2, and CLDN3) and smooth muscle differentiation (ACTA2, SMTN, TPM2, CRIP1, CRIP2, SSPN, CNN1, SGCG, FLNC, MYH11, MYH6, SILV, TNNI1, TNNT2, DMN, TLN2, MYL4, MYL1, TMOD, TTN, and CALD1) in NT2/D1 but not in 27X-1 cells. Conversely, both ATRA and BMP-2 induced key transcripts associated with ExE lineage such as GATA6 and SOX7 only in 27X-1 but not in NT2/D1. Thus, the specific response of these two EC cell lines appears to be linked to the level of pluripotency defined by their constitutive gene expression pattern.

Comparison of Gene Expression Between PE Differentiation In Vitro and Yolk Sac Tumors In Vivo Identifies Markers Associated with VE Development

Differentiation of EC tumors towards a yolk sac phenotype involves both PE and VE lineages [1], unlike the in vitro differentiation of 27X-1 cells that involves only the PE lineage [6, [7]8]. We reasoned that by comparing the gene expression patterns between PE differentiation in vitro and yolk sac tumor development in vivo, the pathways associated with development of the VE and PE components of yolk sac tumors could be distinguished.

To this end, we first identified differentially expressed transcripts between 10 pure yolk sac and 15 pure EC tumors using SAM analysis [16]. A total of 1,570 transcripts were differentially expressed (Q value cutoff = 0.08) in yolk sac tumors in comparison with EC tumors. Of these 1,570 transcripts, 492 were found to alter during PE differentiation of 27X-1 cells and thus represented the PE component of in vivo yolk sac tumor differentiation (supplemental online Table 5A). This left 1,078 transcripts, of which 481 were expressed at twofold or higher in yolk sac tumors compared with EC tumors (supplemental online Table 5B). Functional annotation of the 50 most abundantly expressed transcripts among these (sixfold and above, Table 3) indicated that a number of well-characterized (AFP, TTR, BMP-2, HNF4A, HNF3B, APOA1, APOB, and SOX17) and putative (C5, OTX2, CER1, HP, TF, APOC3, FOXA2, and VIL1) transcripts associated with VE lineage were induced during yolk sac tumor differentiation [6, 26, 28, [29]30]. None of these transcripts were induced during PE differentiation of 27X-1 cells. We also noted that several MAGE (MAGEA3, MAGEA6, MAGEA12, and MAGEA2) and RAGE family members were expressed sixfold or higher in yolk sac tumors (Table 3), suggesting that these transcripts may be associated with VE differentiation.

Table Table 3.. Fifty most highly overexpressed transcripts in pure yolk sac (YS) tumors relative to pure embryonal carcinoma (EC) tumors after a subtraction of all-trans retinoic acid and bone morphogenetic protein-2-mediated time course data associated with parietal endoderm lineage establishment in 27X-1 cells
Thumbnail image of

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Human EC cell lines exhibit a divergent differentiation behavior in vitro in response to given morphogens as illustrated by the neural and ExE lineage development in response to ATRA by NT2/D1 and 27X-1 and by epithelial versus ExE lineage development in response to BMP-2. Indeed, some EC cell lines, such as 2102EP, have limited differentiation potential in response to inputs by developmental morphogens such as ATRA [31]. The molecular basis for these divergent differentiation responses in EC cell lines is poorly understood. In the present study, we first sought to determine whether undifferentiated NT2/D1 and 27X-1 cells differed in the expression levels of genes that govern pluripotency and early embryonic development. Our analysis showed that NT2/D1 cells predominantly expressed transcripts associated with neurectodermal differentiation pathways, whereas 27X-1 cells expressed transcripts characteristic of non-neural and mesendodermal differentiation pathways, in accordance with their differentiation potential. A number of stem cell associated transcripts and transcripts associated with Nodal signaling were also differentially expressed between the two cell lines.

A higher-level expression of SOX2 and PROM1 may underlie the neurectodermal characteristics displayed by NT2/D1 cells since both markers are highly expressed not only in human ES cells but also in human neural stem cells [19]. The expression of SOX2 and PROM1 is suppressed when NT2/D1 cells undergo terminal differentiation [2, 4], suggesting immature neurectodermal-like characteristics of NT2/D1 cells at the undifferentiated state. Recently, we also observed higher-levels of expression of the stem cell marker SOX2 in pluripotent human EC tumors that possess the ability to undergo differentiation as compared with seminomas that lack such ability [17]. 27X-1 cells, on the other hand, expressed higher level of NODAL, which has been reported to both inhibit the differentiation of pluripotent cells towards a neurectodermal lineage as well as favor the establishment of an ExE lineage [21]. Furthermore, nodal signaling has also been found to be essential for mesendodermal patterning during early embryogenesis [32], consistent with our characterization of 27X-1 cells as mesendodermal. Higher-level expression of NODAL in 27X-1 cells and the known nodal signaling inhibitor EBAF (LEFTYA) in NT2/D1 cells therefore likely play an important role in determining ExE and neurectodermal differentiation responses in vitro.

In addition to the reported constitutive differences between EC lines that predict differentiation potential, we also observed differential induction of transcriptional programs and lineage markers in response to morphogens. Morphogens acting on EC cell lines typically execute the first part of the process by inducing expression of cell-cycle inhibitors and downregulation of genes associated with cell proliferation as noted in our present and previous investigations [2, 4]. However, the final lineage specification is dependent upon the differential induction cell-fate specific transcription factors and phenotypic markers. Our analysis revealed that ATRA and BMP-2 induce different sets of transcription factors to mediate neural/epithelial differentiation in NT2/D1 cells and ExE differentiation in 27X-1 cells. Thus, a number of known targets of ATRA signaling that function as neurogenic transcription factors (PAX6 [33] HOXB3 [34], HOXC4 [35], HOXD4 [36], ASCL1 [mash1] [37] BHLHB2 [Dec1/stra13] [38], NHLH2 [39], and NR2F1 [COUP-TF1] [40]) and developmental growth factors (FGF9 [41] and WNT13 [42]) were induced selectively during neural differentiation of NT2/D1 cells but not during ExE differentiation of 27X-1 cells, in response to ATRA. Conversely, two ATRA targets, GATA6 [43] and SOX7 [44], associated with ExE lineage differentiation in pluripotent cells, were induced by ATRA in 27X-1 but not in NT2/D1 cells. Similar divergence in responses was also noted when BMP-2 induced epithelial and smooth muscle-associated transcripts in NT2/D1 and ExE differentiation in 27X-1 cells.

Derivation of extra-embryonic endoderm from inner cell mass cells is an early developmental event that is essential for proper growth and orientation of the embryo. The VE and PE components of the ExE not only assist in the uptake and processing of nutrients but also provide various molecular cues that direct body axis plan and embryonic patterning [45, 46]. Thus far, ATRA-mediated differentiation of murine F9 EC cells was the main in vitro platform available to characterize the biochemical and molecular processes associated with ExE differentiation [47, 48]. Our analysis comprises a first effort to both catalog the gene expression alterations associated with ExE differentiation and demarcation of the transcriptional differences associated with VE and PE differentiation in vitro and in vivo in pluripotent human cells. We reasoned that a comparison of in vitro 27X-1 differentiation into a PE lineage with in vivo yolk sac tumor differentiation into both PE and VE lineages would allow us to dissect out both PE and VE lineage specific genes. This novel approach resulted in the generation of comprehensive gene lists that should aid in our understanding of the molecular events associated with in vivo PE and VE differentiation.

In conclusion, our study provides novel insights into the molecular differences underlying pluripotency and distinct morphogen responses between NT2/D1 and 27X-1 EC cell lines. Our data also present an extensive repertoire of signaling molecules and lineage markers associated with the establishment of ExE lineages and allow the distinction of PE and VE fates. In vitro differentiation of 27X-1 cells constitutes a powerful model system to address questions of extra-embryonic fate choice during early human development.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

We thank the Genomics Core Facility of Memorial Sloan-Kettering Cancer Center (MSKCC) for hybridization and scanning of the arrays. We thank the Flow Cytometry Core facility of MSKCC for the analysis of the surface markers and cell sorting. This work was supported by grants from NIH and the Byrne Research Fund.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information
  • 1
    Chaganti RS, Houldsworth J. Genetics and biology of adult human male germ cell tumors. Cancer Res 2000; 60:14751482.
  • 2
    Houldsworth J, Heath SC, Bosl GJ et al. Expression profiling of lineage differentiation in pluripotential human embryonal carcinoma cells. Cell Growth Differ 2002; 13:257264.
  • 3
    Freemantle SJ, Kerley JS, Olsen SL et al. Developmentally-related candidate retinoic acid target genes regulated early during neuronal differentiation of human embryonal carcinoma. Oncogene 2002; 21:28802889.
  • 4
    Chadalavada RS, Houldsworth J, Olshen AB et al. Transcriptional program of bone morphogenetic protein-2-induced epithelial and smooth muscle differentiation of pluripotent human embryonal carcinoma cells. Funct Integr Genomics 2005; 5:5969.
  • 5
    Caricasole A, Ward-van Oostwaard D, Zeinstra L et al. Bone morphogenetic proteins (BMPs) induce epithelial differentiation of NT2D1 human embryonal carcinoma cells. Int J Dev Biol 2000; 44:443450.
  • 6
    Roach S, Schmid W, Pera MF. Hepatocytic transcription factor expression in human embryonal carcinoma and yolk sac carcinoma cell lines: Expression of HNF-3 alpha in models of early endodermal cell differentiation. Exp Cell Res 1994; 215:189198.
  • 7
    Pera MF, Cooper S, Mills J et al. Isolation and characterization of a multipotent clone of human embryonal carcinoma cells. Differentiation 1989; 42:1023.
  • 8
    Pera MF, Herszfeld D. Differentiation of human pluripotent teratocarcinoma stem cells induced by bone morphogenetic protein-2. Reprod Fertil Dev 1998; 10:551555.
  • 9
    Sperger JM, Chen X, Draper JS et al. Gene expression patterns in human embryonic stem cells and human pluripotent germ cell tumors. Proc Natl Acad Sci U S A 2003; 100:1335013355.
  • 10
    Andrews PW, Matin MM, Bahrami AR et al. Embryonic stem (ES) cells and embryonal carcinoma (EC) cells: opposite sides of the same coin. Biochem Soc Trans 2005; 33:15261530.
  • 11
    Houldsworth J, Reuter VE, Bosl GJ et al. ID gene expression varies with lineage during differentiation of pluripotential male germ cell tumor cell lines. Cell Tissue Res 2001; 303:371379.
  • 12
    A Practical Guide to Splines. In: de BoorC, ed. Applied Mathematical Sciences. New York: Springer-Verlag,1978;Vol 27.
  • 13
    Lloyd SP. Least square quantization in PCM. IEEE Trans Inf Theor 1982; 28:129137.
  • 14
    Soukas A, Cohen P, Socci ND et al. Leptin-specific patterns of gene expression in white adipose tissue. Genes Dev 2000; 14:963980.
  • 15
    Irizarry RA, Hobbs B, Collin F et al. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 2003; 4:249264.
  • 16
    Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A 2001; 98:51165121.
  • 17
    Korkola J, Houldsworth J, Chadalavada RS et al. Down-regulation of stem cell genes, including those in a 200kb gene cluster at 12p1331, is associated with in vivo differentiation of human male germ cell tumors. Cancer Res 2006; 66:820826.
  • 18
    Brivanlou AH, Gage FH, Jaenisch R et al. Stem cells. Setting standards for human embryonic stem cells. Science 2003; 300:913916.
  • 19
    Uchida N, Buck DW, He D et al. Direct isolation of human central nervous system stem cells. Proc Natl Acad Sci U S A 2000; 97:1472014725.
  • 20
    Russ AP, Wattler S, Colledge WH et al. Eomesodermin is required for mouse trophoblast development and mesoderm formation. Nature 2000; 404:9599.
  • 21
    Vallier L, Reynolds D, Pedersen RA. Nodal inhibits differentiation of human embryonic stem cells along the neuroectodermal default pathway. Dev Biol 2004; 275:403421.
  • 22
    Rogers MB, Rosen V, Wozney JM et al. Bone morphogenetic proteins-2 and -4 are involved in the retinoic acid-induced differentiation of embryonal carcinoma cells. Mol Biol Cell 1992; 3:189196.
  • 23
    Hayashi Y, Emoto T, Futaki S et al. Establishment and characterization of a parietal endoderm-like cell line derived from Engelbreth-Holm-Swarm tumor (EHSPEL), a possible resource for an engineered basement membrane matrix. Matrix Biol 2004; 23:4762.
  • 24
    Futaki S, Hayashi Y, Yamashita M et al. Molecular basis of constitutive production of basement membrane components. Gene expression profiles of Engelbreth-Holm-Swarm tumor and F9 embryonal carcinoma cells. J Biol Chem 2003; 278:5069150701.
  • 25
    Futaki S, Hayashi Y, Emoto T et al. Sox7 plays crucial roles in parietal endoderm differentiation in F9 embryonal carcinoma cells through regulating Gata-4 and Gata-6 expression. Mol Cell Biol 2004; 24:1049210503.
  • 26
    Fujikura J, Yamato E, Yonemura S et al. Differentiation of embryonic stem cells is induced by GATA factors. Genes Dev 2002; 16:784789.
  • 27
    Harris TM, Childs G. Global gene expression patterns during differentiation of F9 embryonal carcinoma cells into parietal endoderm. Funct Integr Genomics 2002; 2:105119.
  • 28
    Thomas T, Southwell BR, Schreiber G et al. Plasma protein synthesis and secretion in the visceral yolk sac of the fetal rat: gene expression, protein synthesis and secretion. Placenta 1990; 11:413430.
  • 29
    Yasunaga M, Tada S, Torikai-Nishikawa S et al. Induction and monitoring of definitive and visceral endoderm differentiation of mouse ES cells. Nat Biotechnol 2005; 23:15421550.
  • 30
    Duncan SA, Nagy A, Chan W. Murine gastrulation requires HNF-4 regulated gene expression in the visceral endoderm: tetraploid rescue of Hnf-4(−/−) embryos. Development 1997; 124:279287.
  • 31
    Bahrami AR, Matin MM, Andrews PW. The CDK inhibitor p27 enhances neural differentiation in pluripotent NTERA2 human EC cells, but does not permit differentiation of 2102Ep nullipotent human EC cells. Mech Dev 2005; 122:10341042.
  • 32
    Osada SI, Wright CV. Xenopus nodal-related signaling is essential for mesendodermal patterning during early embryogenesis. Development 1999; 126:32293240.
  • 33
    Gajovic S, St-Onge L, Yokota Y et al. Retinoic acid mediates Pax6 expression during in vitro differentiation of embryonic stem cells. Differentiation 1997; 62:187192.
  • 34
    Faiella A, Zappavigna V, Mavilio F et al. Inhibition of retinoic acid-induced activation of 3′ human HOXB genes by antisense oligonucleotides affects sequential activation of genes located upstream in the four HOX clusters. Proc Natl Acad Sci U S A 1994; 91:53355339.
  • 35
    Otero JJ, Fu W, Kan L et al. Beta-catenin signaling is required for neural differentiation of embryonic stem cells. Development 2004; 131:35453557.
  • 36
    White JC, Highland M, Kaiser M et al. Vitamin A deficiency results in the dose-dependent acquisition of anterior character and shortening of the caudal hindbrain of the rat embryo. Dev Biol 2000; 220:263284.
  • 37
    Bain G, Ray WJ, Yao M. Retinoic acid promotes neural and represses mesodermal gene expression in mouse embryonic stem cells in culture. Biochem Biophys Res Commun 1996; 223:691694.
  • 38
    Boudjelal M, Taneja R, Matsubara S et al. Overexpression of Stra13, a novel retinoic acid-inducible gene of the basic helix-loop-helix family, inhibits mesodermal and promotes neuronal differentiation of P19 cells. Genes Dev 1997; 11:20522065.
  • 39
    Itoh F, Nakane T, Chiba S. Gene expression of MASH-1, MATH-1, neuroD and NSCL-2, basic helix-loop-helix proteins, during neural differentiation in P19 embryonal carcinoma cells. Tohoku J Exp Med 1997; 182:327336.
  • 40
    Jonk LJ, de Jonge ME, Pals CE et al. Cloning and expression during development of three murine members of the COUP family of nuclear orphan receptors. Mech Dev 1994; 47:8197.
  • 41
    Seo M, Noguchi K. Retinoic acid induces gene expression of fibroblast growth factor-9 during induction of neuronal differentiation of mouse embryonal carcinoma P19 cells. FEBS Lett 1995; 370:231235.
  • 42
    Wakeman JA, Walsh J, Andrews PW. Human Wnt-13 is developmentally regulated during the differentiation of NTERA-2 pluripotent human embryonal carcinoma cells. Oncogene 1998; 17:179186.
  • 43
    Zhuang Y, Faria TN, Chambon P et al. Identification and characterization of retinoic acid receptor beta2 target genes in F9 teratocarcinoma cells. Mol Cancer Res 2003; 1:619630.
  • 44
    Murakami A, Shen H, Ishida S et al. SOX7 and GATA-4 are competitive activators of Fgf-3 transcription. J Biol Chem 2004; 279:2856428573.
  • 45
    Lu CC, Brennan J, Robertson EJ et al. From fertilization to gastrulation: Axis formation in the mouse embryo. Curr Opin Genet Dev 2001; 11:384392.
  • 46
    Kimura C, Yoshinaga K, Tian E et al. Visceral endoderm mediates forebrain development by suppressing posteriorizing signals. Dev Biol 2000; 225:304321.
  • 47
    Rochette-Egly C, Chambon P. F9 embryocarcinoma cells: A cell autonomous model to study the functional selectivity of RARs and RXRs in retinoid signaling. Histol Histopathol 2001; 16:909922.
  • 48
    Thompson JR, Gudas LJ. Retinoic acid induces parietal endoderm but not primitive endoderm and visceral endoderm differentiation in F9 teratocarcinoma stem cells with a targeted deletion of the Rex-1 (Zfp-42) gene. Mol Cell Endocrinol 2002; 195:119133.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
Suppl_Fig_1.pdf190KSupplemental Figure 1
suppl_Fig_2.pdf117KSupplemental Figure 2
suppl_fig_3.pdf39KSupplemental Figure 3
suppl_Fig_4.pdf108KSupplemental Figure 4
Suppl_Fig_5.pdf85KSupplemental Figure 5
Suppl_Fig_6.pdf220KSupplemental Figure 6
Suppl_Fig_7.pdf129KSupplemental Figure 7
Supplemental_Figure_Legends_Chaganti.pdf69KSupplemental Legends

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.