Transcription Factor TFAP2C Regulates Major Programs Required for Murine Fetal Germ Cell Maintenance and Haploinsufficiency Predisposes to Teratomas in Male Mice

Abstract

Maintenance and maturation of primordial germ cells is controlled by complex genetic and epigenetic cascades, and disturbances in this network lead to either infertility or malignant aberration. Transcription factor TFAP2C has been described to be essential for primordial germ cell maintenance and to be upregulated in several human germ cell cancers. Using global gene expression profiling, we identified genes deregulated upon loss of Tfap2c in embryonic stem cells and primordial germ cell-like cells. We show that loss of Tfap2c affects many aspects of the genetic network regulating germ cell biology, such as downregulation of maturation markers and induction of markers indicative for somatic differentiation, cell cycle, epigenetic remodeling and pluripotency. Chromatin-immunoprecipitation analyses demonstrated binding of TFAP2C to regulatory regions of deregulated genes (Sfrp1, Dmrt1, Nanos3, c-Kit, Cdk6, Cdkn1a, Fgf4, Klf4, Dnmt3b and Dnmt3l) suggesting that these genes are direct transcriptional targets of TFAP2C in primordial germ cells. Since Tfap2c deficient primordial germ cell-like cells display cancer related deregulations in epigenetic remodeling, cell cycle and pluripotency control, the Tfap2c-knockout allele was bred onto 129S2/Sv genetic background. There, mice heterozygous for Tfap2c develop with high incidence germ cell cancer resembling human pediatric germ cell tumors. Precursor lesions can be observed as early as E16.5 in developing testes displaying persisting expression of pluripotency markers. We further demonstrate that mice with a heterozygous deletion of the TFAP2C target gene Nanos3 are also prone to develop teratomas. These data highlight TFAP2C as a critical and dose-sensitive regulator of germ cell fate.

Introduction

Germ cell cancers (GCCs) are usually diagnosed between the age of 20–40 years and are the most common cancer type of young men [1]. In infants and pre-pubertal adolescents, teratomas and yolk-sac tumors (Type I GCC) are detected in gonads, cranium or along the body midline. These tumors characteristically consist of tissues of all three germ layers and are generally benign in nature, with rare malignant transformation. It is assumed that the precursor cells of these tumors are primordial germ cells/gonocytes which fail to progress into spermatogonia [2];3;4] and transform into embryonal carcinoma (EC) cells which appear at embryonic day (E) 15.5 [5]. Knowledge of the regulatory network of germ cell specification, maintenance and differentiation is required to further understand the molecular basis of this malignancy and some of the molecular key players have been determined in the past years. Specification of murine primordial germ cells (PGCs) occurs at E6.75 and is mediated by BMP signaling (BMP4/BMP8b) [6]; [7]; [8], which leads to induction of Prdm1 and Prdm14. PRDM1 (BLIMP1) together with PRDM14 are viewed as the key regulators since they orchestrate the re-acquisition of pluripotency and repression of the somatic program in PGCs [9]; [10]; [11]. Single cell analyses in Blimp1 deficient PGCs suggest that the transcription factor TFAP2C is a downstream target of BLIMP1, as Tfap2c level was found to be dramatically reduced in Blimp1 deficient PGCs [10]. TFAP2C (Tcfap2c/AP-2γ) is a member of the activator protein-2 (AP-2) family, which comprises of five closely related members, namely Tfap2a-e. They are characterized by a highly conserved DNA-binding and dimerization motif at the C-terminus. AP-2 transcription factors bind to DNA as functional dimers and their activation is mediated by an N-terminal transactivation domain [12]. Tfap2c is expressed in murine PGCs shortly after their specification from E7.25 up to E12.5 after they have migrated and colonized the genital ridges [13]. After arrival in the genital ridges, PGCs initiate further differentiation indicated by downregulation of pluripotency markers Nanog, Oct3/4, Sox2 [14], exit of mitosis [15], upregulation of germ cell markers Mutvh and Dazl [16]; [17] and epigenetic remodeling (remethylation, upregulation of de novo methylases) [18]. Similar to Blimp1 [9], Tfap2c deficient PGCs are lost shortly after specification at E8.5. In Tfap2c^−/−PGCs, germ cell markers (like Stella/Dppa3, Nanos3, Dazl, Mutvh) are downregulated and the somatic gene program is upregulated (Hoxb1, Hoxa1, T) resulting in loss of PGC-fate and somatic differentiation of the cells [13]. In humans, TFAP2C expression is detected in fetal germ cells from week 12 to 37 of pregnancy and in a small subset of cells (gonocytes) in the infantile testes (until 4 months of postnatal age) [19]. Furthermore, high levels of TFAP2C protein were observed in precursor lesions of GCC (carcinoma in situ (CIS)) and classical seminomas [19]; [20].

In this study, we analyzed the consequences of lack of Tfap2c in PGC-like cells (PGCLCs). Global cDNA expression profiling revealed that Tfap2c deficiency affects cell cycle exit, epigenetic remodeling, germ cell differentiation and regulation of pluripotency. Using chromatin-immunoprecipitation (ChIP) analyses we show that Sfrp1, Dmrt1, Nanos3, c-Kit, Cdk6, Cdkn1a, Fgf4, Klf4, Dnmt3b and Dnmt3l are direct transcriptional target genes of TFAP2C. The data suggest that TFAP2C governs many aspects of PGC development, some of them being also involved in GCC formation. In line with this, we demonstrate that haploinsufficiency of Tfap2c or its target gene Nanos3 lead to high rate of GCC in 129S2/Sv mice, resembling human pediatric germ cell tumors.

Materials and Methods

Animals/Ethics Statement

All experiments were conducted according to the German law of animal protection and in agreement with the approval of the local institutional animal care committees (Landesamt für Natur, Umwelt und Verbraucherschutz, North Rhine-Westphalia (approval ID: #8.87-50.10.31.08.238). The experiments were conducted in accordance with the International Guiding Principles for Biomedical Research Involving Animals as announced by the Society for the Study of Reproduction.

Generation of Blimp1mVenus/Tfap2c^−/− ESCs

The derivation of embryonic stem cells (ESCs) from blastocysts was performed by mating Blimp1mVenus (Tg^{(Prdm1-Venus1)Sait}, MGI:3805969) [21] with Tfap2c^flox/flox (Tfap2c^tm1Hsc, MGI:2176695) [22] mice. ES cells were derived according to published protocols to obtain Blimp1mVenus/Tfap2c^flox/flox ESCs [23]. Transient transfection with a pGK-Cre plasmid resulted in Cre-mediated excision of the floxed sites leading to Blimp1mVenus/Tfap2c^−/−ESCs. Genotyping of Tfap2c and Blimp1mVenus alleles was done as described previously [21]; [22].

Embryoid Body Formation

For gene expression analysis Blimp1mVenus/Tfap2c^ctrl and Blimp1mVenus/Tfap2c^−/−ESCs were cultured for embryoid body (EB) formation in DMEM/F12/Neurobasal media (1∶1) supplemented with Glutamax, N2-Supplement (1×), B27-Supplement (1×), L-Glutamine (1×) (all Gibco, Life Technologies, Darmstadt, Germany), BSA (2,5 mg/ml; Sigma Aldrich; Munich, Germany), ß-Mercaptoethanol (100 µM, PAA; GE Healthcare; USA), Insulin (1 mg/ml; Sigma Aldrich; Munich, Germany), PD0325901 (1 µM; Stemgent, San Diego, USA), CHIR99021 (3 µM; Stemgent, San Diego, USA) and Lif (1000 U/ml; Esgro, Merck Millipore, Darmstadt, Germany). Cells were dissociated with Accutase (PAA; GE Healthcare; USA). PGCLC differentiation was performed as described by Hayashi et al. [24]. Blimp1mVenus/Tfap2c^ctrl-PGCLCs and StellaGFP ESCs [25] (which were differentiated into EBs for 5 days as described by Young et al. [26]) were used for ChIP analyses. For StellaGFP-PGCLC in vitro differentiation DMEM supplemented with GlutaMAX, Sodium Pyruvat (Gibco, Life Technologies, Darmstadt, Germany), ß-mercaptoethanol (100 µM; PAA; GE Healthcare; USA), nonessential amino acids (1×; PAA; GE Healthcare; USA); essential amino acids (1×; PAA; GE Healthcare; USA); L-Glutamine (2 mM; PAA; GE Healthcare; USA); Penicillin/Streptomycin (1×; PAA; GE Healthcare; USA); LIF (1000 U/ml; Esgro, Merck Millipore, Darmstadt, Germany); 15% ES FCS (Hyclone; Thermo Scientific; USA) and BMP4 (100 ng; R&D Systems; Wiesbaden, Germany) was used.

Fluorescence Activated Cell Sorting (FACS)

Embryoid bodies were washed with PBS, dissociated with Accutase (PAA; GE Healthcare; USA) and filtered through a cell strainer (40 µm, BD Biosciences, Heidelberg, Germany). Analysis was performed using FACS Aria III analytical flow cytometer (BD Biosciences, Heidelberg, Germany). All Blimp1mVenus positive cells were recorded per experiment. Data were analyzed using BD FACSDiva 6.3.1 software. Significance was calculated by 2 paired t-test. A p-value of <0.05 was considered to be significant.

Transfection of TCam-2 Cells

siRNA mediated knockdown of TFAP2C in TCam-2 cells was performed as described by [27]. Scrambled and TFAP2C siRNA were obtained from Origene (TFAP2C SR304789, Origene Technologies, Rockville, USA).

Chromatin-immunoprecipitation

Aliquots of 1×10⁵ PGCLCs were cross-linked with 1% formaldehyde for 7 min and incubated with 0.1 M glycine for 5 min at room temperature. Cells were resuspended in 200 µl SDS Lysis Buffer and sonicated for 2 cycles (5 min) and 1 cycle (2 min) (30 sec “ON”/30 sec “OFF”). Chromatin-immunoprecipitation (ChIP) experiments were performed as described in [28]. For immunoprecipitation 3 µg antibody against TFAP2C (clone H77/sc-8977 X; SantaCruz, Santa Cruz, USA) and as control antibody against rabbit IgG (Merck Millipore, Darmstadt, Germany) on protein G coupled Dynabeads (Life Technologies, Darmstadt, Germany) were used. Immunoprecipitated DNA was amplified using GenomePlex® Single Cell Whole Genome Amplification Kit (WGA4) (Sigma Aldrich; Munich, Germany) according to manufacturer’s protocol. Amplified ChIP DNA was subjected to qPCR using a Maxima SYBR Green/ROX (Fermentas, Thermo Scientific; USA) in a ViiA™ 7 Real-Time PCR System (Applied Biosystemş Life Technologies, Darmstadt, Germany) according to the manufacturer’s specified parameters. Amplicons were normalized to the non-immunoprecipitated input DNA. Primers used for qPCR are listed in Table S4.

Semiquantitative RT-PCR

Total RNA from ESCs and sorted PGC-like cells were isolated using RNeasy Mini Kit (Qiagen, Hilden, Germany). First-strand cDNA were synthesized using RevertAid Premium reverse transcriptase (Fermentas, Thermo Scientific; USA). RNA from gonads was extracted using the Nucleo Spin FFPE RNA Kit (Macherey & Nagel, Düren, Germany); 100 ng was used for cDNA synthesis (SuperScriptIII; Invitrogen, Karlsruhe, Germany). RT-PCR-primers are listed in Table S5.

qRT-PCR

RNA from PGCLCs and gonads was isolated using RNeasy Mini/Micro Kit (Qiagen, Hilden, Germany), respectively. Total RNA from TCam-2 cells was extracted by Trizol (Sigma Aldrich; Munich, Germany) according to manufactureŕs instructions. For amplification of PGCLCs, gonadal and TCam-2 mRNA, SYBR GreenER Reagent Mix (Invitrogen, Karlsbad, Germany)/Maxima SYBR Green/ROX (Fermentas, Thermo Scientific; USA) was used. Housekeeping genes βActin and Gapdh were used for data normalization. The experiments were performed in biological triplicates. Primer sequences for qRT-PCR are given in Table S6.

Transcriptome Microarray Analysis

For the microarray experiments GeneChip Mouse Genome 430 2.0 Arrays (Affymetrix, Santa Clara, CA, USA) were used. RNA was extracted from ESCs and sorted PGCLCs using the RNeasy Mini Kit (Qiagen, Hilden, Germany). RNA quality was tested using the Agilent 2100 Bioanalyzer (Agilent Technologies, Böblingen, Germany). For the Affymetrix array, 100 ng of total RNA were used. The arrays were washed and stained according to the manufacturer’s recommendations and scanned in a GeneChip scanner 3000 (Affymetrix, Santa Clara, CA, USA). The normalization was calculated with the RMA (Robust Multi-array Analysis) algorithm [29]. Data post-processing and graphics were performed with in-house developed functions in Matlab. Hierarchical clustering of genes and samples was performed with one minus correlation metric and the unweighted average distance (UPGMA) (also known as group average) linkage method. The gene ontology analysis based on AMIGO gene ontology database [30]. The significance of the gene ontology terms of the differentially expressed genes was analyzed using an enrichment approach based on the hypergeometric distribution. All sets of gene ontology terms were backpropagated from the final term appearing in the gene annotation until the root term of each ontology. The significance (p-value) of the gene ontology terms enrichment was calculated using the hypergeometric distribution. The multitest effect influence was corrected through controlling the false discovery rate using the Benjamini-Hochberg correction at a significance level α = 0.05. The data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO Series accession number GSE45941.

Laser Microdissection

Paraffin-embedded tissue was sectioned to 10 µm thickness and mounted on P.A.L.M. Frame slides (Carl-Zeiss Micro Imaging, Göttingen, Germany) and stained with SSEA-1 antibody (R&D Systems, Minneapolis, USA). SSEA-1 positive areas were dissected using P.A.L.M. Microlaser Technology (Carl-Zeiss Micro Imaging, Göttingen, Germany). Dissected tissue was collected in lysis buffer (1 mM EDTA pH 8, 10 mM Tris-HCl pH 8, 0.5% Tween20, 100 mg/ml proteinase K) and incubated at 37°C for 3 hrs, heated to 85°C for 10 min. DNA was extracted with QiaAmp DNA Micro Kit (Qiagen, Hilden, Germany). RNA was extracted with Nucleo Spin FFPE RNA Kit (Macherey & Nagel, Düren, Germany).

Histology and Immunohistochemistry

Sections and cells were fixed and stained as described by Weber et al. [13]. The following antibodies were used: TFAP2C (1∶300–1∶500, SantaCruz, Santa Cruz, USA), OCT3/4 (1∶100, SantaCruz, Santa Cruz, USA), SSEA1 (1∶100, R&D Systems, Minneapolis, USA), anti-rabbit biotinylated (1∶500; DAKO, Hamburg, Germany), anti-rat biotinylated (1∶200; DAKO, Hamburg, Germany), anti-rabbit Alexa fluor 594 (1∶1000; Life Technologies, Darmstadt, Germany); anti-mouse Alexa fluor 488 (1∶1000; Life Technologies, Darmstadt, Germany). Nuclei were stained using 1∶500 33342 Hoechst (1 mg/ml; Life Technologies, Darmstadt, Germany).

Results

Tfap2c Deficient ESCs Generate Less PGCLCs

In order to analyze the molecular consequences of Tfap2c deficiency in PGCs we took advantage of protocols allowing PGCLC derivation from embryonic stem cells through an epiblast-like cell (EpiLC) intermediate [24]. Blimp1mVenus reporter mice [21] were bred into the Tfap2c^flox/flox background [22] and two Blimp1mVenus/Tfap2c^flox/flox embryonic stem cell (ESC) lines were established. Next, the ESC-lines were transiently transfected with a plasmid encoding the Cre-Protein (pgk-Cre) mediating the deletion of the Tfap2c^flox/flox alleles resulting in Blimp1mVenus/Tfap2c^−/− ESCs (#1-Tfap2c^−/− and #2-Tfap2c^−/−) (Fig. S1 A, B, C). For all following experiments, the parental Blimp1mVenus/Tfap2c^flox/flox ESC-lines served as control (ctrl). Under serum and feeder-free conditions ESCs were differentiated for two days into EpiLCs, which display a flattened epithelial-like structure (Fig. S2 A). The expression of marker genes in #1-ctrl, #1-Tfap2c^−/−, #2-ctrl and #2-Tfap2c^−/− cells was analyzed by semiquantitative RT-PCR to address the question, whether loss of TFAP2C affects differentiation from ESCs to EpiLCs. Upon differentiation into EpiLCs, marker genes for ESCs, Klf4 and Prdm14, were downregulated and epiblast associated genes Fgf5 and Dnmt3b were upregulated (Fig. S2 B). The results suggest that the generation of EpiLCs is not affected by lack of TFAP2C.

Next, PGCLC differentiation was induced by cultivating EpiLCs in medium supplemented with BMP4, BMP8b, SCF, LIF and EGF [24]. Induction of the Blimp1mVenus reporter indicated the generation of PGCLCs, which were quantified using fluorescence activated cell sorting (FACS) (Fig. S1 C/S2 C). Using Tfap2c^−/− ESC-lines, we obtained less Blimp1mVenus positive PGCLCs (#1∶9.3% vs. 2.7% (significant); #2∶3% vs. 1.5% (non-significant)) (Fig. S2 D). RT-PCR demonstrated that lack of Tfap2c did not affect the expression of the early germ cell marker Prdm14, but lead to a reduction of late markers (Stella, Nanos3) (Fig. S2 E). Expression of Prdm14 and the Blimp1mVenus reporter indicate that PGCLCs are specified in the absence of TFAP2C. However, the reduction of the number of PGCLCs strongly suggests a role of TFAP2C in the maintenance of the PGC-fate. The fact, that the expression of late germ cells markers is impaired argues against the possibility that the Tfap2c^−/−PGCLCs undergo premature differentiation.

Transcriptome Analysis of Tfap2c Deficient ESCs and in vitro Differentiated PGCLCs

To gain insight into the consequences of lack of Tfap2c on a genome-wide level, gene expression analysis using ESCs and PGCLCs (#1-ctrl; #1-Tfap2c^−/−; #2-ctrl and #2-Tfap2c^−/−) was performed with Affymetrix Mouse 430 2.0 microarrays. Hierarchical clustering showed that Tfap2c^−/−PGCLCs cluster close together and distant from control samples indicating the high similarity of gene expression profiles within the Tfap2c^−/− and ctrl pairs (Fig. 1 A). Heat map shows differentially expressed transcripts between the Tfap2c^−/− and ctrl PGCLCs (Fig. 1 B). Scatter plotting of Tfap2c^−/− vs. ctrl PGCLCs identified 455 significantly deregulated genes (Fold-change >1.5 in log2 scale) (Fig. 1 C; Table S1). Of note, TFAP2C has only moderate effects in ESCs (albeit Tfap2c is expressed in 10–15% of ESCs) [23] as indicated by hierarchical clustering, heat mapping and scatter plotting (Fig. 1 A, B, D). In ESCs, only 26 genes were deregulated after loss of TFAP2C (Table S2). Comparing the PGCLC with ESC datasets revealed an overlap of 13 genes (Fig. 1 E; Table S3). These results indicate that TFAP2C controls a diverse set of genes in ESCs and PGCLCs and that the lack of this protein in PGCLCs affects the expression of far more genes compared to ESCs.

Figure 1. Genes regulated in PGCLCs and ESCs by TFAP2C.

Affymetrix microarray gene expression analysis performed with RNA extracted from #1-ctrl; #1-Tfap2c^−/− and #2-ctrl; #2-Tfap2c^−/− ESCs and PGCLCs. (A) Hierarchical clustering. The red bars cluster the ESCs, green bars the Tfap2c^−/−PGCLCs and blue bars ctrl PGCLCs, respectively. The shorter the horizontal bar that connects two branches the closer are the populations. (B) Heat map was performed with the probes whose range or variation across all samples was at least 3. Color bar on top codifies the gene expression in log2 scale. Red and blue indicate higher and lower relative expression. (C-D) Pairwise scatter plot of global gene expression in ctrl versus Tfap2c^−/−PGCLCs (C) and ESCs (D). Black lines indicate 1.5 fold-change in log2 scale of gene expression levels between paired PGCLCs and ESCs. Color bars on the side display the scattering density with light blue indicating lower and blue higher scatter density. Genes upregulated in Tfap2c^−/− samples are shown as red dots; genes downregulated are shown as green dots. R² = Fisher’s correlation coefficient. (E) Venn diagram; in PGCLCs 455 genes are deregulated; in ESCs 26 genes are deregulated. The intersection part show the commonly deregulated genes (n = 13) by TFAP2C in PGCLCs and ESCs (Fold-change >1.5 in log2 scale).

Next we performed gene ontology analysis [30] (Data S1) and categorized the genes deregulated in Tfap2c^−/−PGCLCs in relation to their biological function. Here, it became evident that in Tfap2c^−/−PGCLCs many genes associated with somatic differentiation (category I) are upregulated, whereas genes associated with germ cell maintenance and maturation (category II) are downregulated (Table 1). Of note, we also included few genes in Table 1 with relevance to the respective category displaying a fold-change rate below our initial cutoff (≤1.5 in log2 scale, see Table 1). These data suggest an impairment of PGC to gonocyte progression and derepression of somatic differentiation upon loss of TFAP2C. These results confirm earlier data, where markers indicative for somatic differentiation were induced in Tfap2c deficient EBs [13]. In addition, we found upregulation of genes associated with cell cycle regulation (category III), pluripotency (category IV) and epigenetic modification (category V) in Tfap2c^−/−PGCLCs. The data indicate that in migrating PGCs, TFAP2C represses the cell cycle, pluripotency program and DNA methylation.

Table 1. Categorized deregulated genes in Tfap2^−/−PGCLCs based on gene ontology analysis.

Category	Upregulatedgenes	FC	Function	Downregulated genes	FC	Function
I:	Bcl11a	1.93	leukemogenesis and hematopoiesis	Gata2	−1.87	hematopoiesis
Somatic	Cited2	1.87	CNS development	Lhx1	−1.59	mesodem formation
differentiation	Chd7	1.67	CNS development	Slit2	−2.34	neurogenesis
	Cobl	2.43	neural tube closure	Kctd15	−3.20	inhibit neural crest formation
	Ehna	2.01	neural tube closure	Pecam1/CD31	−1.50	mesoderm development
	Epha2	1.53	neural tube development
	Gbx2	2.36	gastrulation-brain specific
	Gli2	3.48	neural tube development
	Hck	2.83	hemopoietic cell kinase
	Hhex	1.52	hematopoietically expressed homeobox gene
	Hoxa5	1.97	mammary gland- lung development
	Hoxa1	1.13	central nervous system neuron differentiation
	Hoxa3	1.13	cartilage- and thymus development
	Hoxb1	1.03	CNS development
	Mtap1b	1.70	CNS development
	Ncam1	2.42	neuron-neuron adhesion
	Nefl	3.56	neurogenesis
	Neurod1	1.57	neurogenesis
	Pou4f2	1.58	expressed in brain
	Pten	1.63	CNS development
	Robo1	1.79	neurogenesis
	Sfrp1	3.12	neural tube closure and development
	Sfrp2	1.66	neural tube development
	Smarca1	2.06	neuron differentiation, brain development
	Tcf7l2	2.56	neural tube development
	Zic2	3.71	CNS development; neural tube closure
	Zic5	3.14	neural tube closure
II:	Cited2	1.87	sex determination	Aurkc	−2.06	key regulator mitosis; meiosis
Germ cell	Pten	1.63	tumor suppressor; apoptosis	Bik	−2.20	spermatogenesis, apoptosis
maintenance	Sprink3	2.54	inhibits calcium uptake in spermatozoa	Cdk16	−1.00	required for spermatogenesis
and				Ceacam10	−2.98	sperm surface protein
maturation				Chdh	−4.80	localized in mitochondrial
				Cxcr4	−1.94	germ cell migration
				Dazl	−1.40	gametogenesis; spermatogenesis
				Dmrt1	−2.68	male sex determination and differentiation
				Hk1	−3.04	expressed in meiosis and postmeiotic germ cells
				Immp2l	−1.57	spermatogenesis
				Kit	−1.91	cell survival, migration and proliferation
				Phf13/Spoc1	−1.85	spermatogenesis
				Rhox5	−4.29	spermatogenesis, sperm maturation
				Rhox4a	−2.70	in fetal germ cells expressed
				Rhox6	−6.11	germ cell lineage determination
				Rhox9/Psx2	−5.59	oogenesis
				Sept4	−1.63	sperm capacitation; mitochondrion organization
				Slco4c1	−2.32	sperm maturation
				Spa17	−2.30	sperm surface protein
				Stella/Dppa3	−4.27	PGC specific protein
				Stk31	−2.40	male germ cell specific expression
				Tes	−2.04	cell adhesion, cell proliferation
III:	Ccnd1	2.33	interacts with Cdk4/Cdk6	Urgcp	−3.17	regulation of Ccnd1
Cell cycle	Ccnt2	1.88	interacts with Cdk9
	Cdc14a	1.57	cell cycle; cell division
	Cdk6	1.50	promotes G1/S transition
	Cdk7	1.17	cell cycle; cell division
	Cdkn1a/p21	1.02	bind and inhibit cyclin-dependent kinase
	Plk2	1.68	involved in G1/S phase transition
	S100a10	2.15	cell cycle progression and differentiation
IV:	Eras	3.52	maintenance of ESC pluripotency
Pluripotency	Fgf4	2.94	regulation cell proliferation, differentiation
	FoxD3	1.03	early direct response gene of Oct3/4
	Gbx2	2.36	cell pluripotency and differentiation
	Jam2	2.47	early direct response gene of Oct3/4
	Klf4	2.22	maintenance of ESC pluripotency
	Msi2	2.31	RNA binding protein; self-renewal of ESC
	Nr0b1	1.62	maintenance of ESC pluripotency
	Tdgf1	1.36	early direct response gene of Oct3/4
V:	Dnmt3b	1.34	genome-wide de novo methylation	Pcgf5	−2.40	chromatin remodeling; histone modification
Epigenetics	Dnmt3l	3.21	DNA methyltransferase like enzyme
	Smarca6/Hells	1.09	de novo or maintenance of DNA methylation
	Mbd2	1.54	recruits Hdacs and Dnmts; gene silencing
	Tet2	1.27	conversion of 5mC to 5hmC
	Uhrf1	1.45	major role in the G1/S transition

bold: genes with fold-change >1.5 in log2 scale; regular: genes with fold-change ≤1.5 in log2 scale.

Subsequently, differential expression seen in the cDNA-microarray of selected marker genes was validated using qRT-PCR in PGCLCs (Fig. 2 A) (downregulated: Nanos3 (−4.78), Rhox5 (−4.26), Stella (−3.84), Dmrt1 (−3.07), c-Kit (−2.78) and Dazl (−0.88) and upregulated: Fgf4 (4.84), Dnmt3l (3.79), Ccnd1 (3.36), Dnmt3b (2.07) and Klf4 (1.95) [fold-changes in log2 scale]). Furthermore, expression of several markers was determined after siRNA mediated knockdown of TFAP2C in the human seminoma-like cell line TCam-2 [31] (Fig. 2 B). Here, after reduction of TFAP2C (to ∼30% compared to the control level) c-KIT (−2.22) and DMRT1 (−1.58) were downregulated and CCND1 (1.60), SOX2 (1.34), DNMT3L (0.98), HOXA5 (0.88), and DNMT3B (0.39) were upregulated (all fold-changes in log2 scale). So, in both systems all selected genes were deregulated in agreement with the array analysis. This suggests a conserved function of TFAP2C/Tfap2c in humans and mice.

Figure 2. Validation of cDNA microarray data.

(A) Quantitative RT-PCR of a subset of markers with RNA isolated from ctrl and Tfap2c^−/−PGCLCs. Expression levels were normalized to βActin and expression level of ctrl PGCLCs were set to 1. qRT-PCR was performed in biological triplicates. Error bars indicate standard deviation. (B) Quantitative RT-PCR of a subset of markers with RNA isolated from TCam-2 cells after siRNA mediated knockdown of TFAP2C. Expression levels were normalized to GAPDH and expression levels of scrambled-siRNA-transfection were set to 1. qRT-PCR was performed in biological triplicate. Error bars indicate standard deviation. (C) ChIP/qPCR analysis for Tfap2c was performed with four biological replicates of PGCLCs. The qPCR results were calculated with the percentage input method and ChIP analyses with IgG antibody served as control and were set to 1. Error bars indicate standard deviation. ChIP analysis demonstrates increased binding of TFAP2C at indicated loci. (A) – (C) Marker genes were grouped in categories: somatic differentiation (category I), germ cell maintenance and maturation (category II), cell cycle regulation (category III), pluripotency (category IV) and epigenetic modification (category V).

ChIP Analysis Reveals Direct Transcriptional Targets

To evaluate direct binding of TFAP2C to promotor regions of deregulated genes we performed TFAP2C specific ChIP-analysis using PGCLCs derived from Stella-GFP ESCs [25] and Blimp1mVenus/Tfap2c^ctrl ESCs. Putative TFAP2C binding sites identified by the rVista algorithm (based on TRANSFAC) [32] (sequence information Table S4) in genomic regions of Sfrp1, Dmrt1, Nanos3, c-Kit, Cdk6, Cdkn1a, Fgf4, Klf4, Dnmt3b and Dnmt3l were enriched for TFAP2C protein (vs. IgG control) (Fig. 2 C), implicating those genes as direct target genes of TFAP2C.

Male Mice Heterozygous for the Tfap2c Deletion Develop Teratomas

Our analyses revealed, that TFAP2C is not only involved in suppression of the somatic program but also controls cell cycle, epigenetic modifications and pluripotency. Since these pathways are discussed to be deregulated in GCC, we wondered whether reduction of Tfap2c could contribute to the initiation of these tumors. The Tfap2c^tm1Hsc (MGI: 93391) allele [22] was bred into the 129S2/Sv genetic background and screened for effects of haploinsufficiency. To determine whether loss of one allele of Tfap2c affected the overall levels of Tfap2c-mRNA, E12.5 genital ridges of control and Tfap2c^+/− embryos were dissected, RNA isolated and qRT-PCR analysis performed. Deletion of one Tfap2c-allele leads to reduction of Tfap2c-mRNA (Fig. 3 A) and to upregulation of the direct target Cdkn1a/p21 (Fig. 3 A). Starting from the seventh generation in 129S2/Sv 82% (n = 51) of all heterozygous males developed testicular tumors (Fig. 3 B) (35% presented as bilateral cases), whereas wildtype littermates (n = 22) remained tumor-free. Tumors presented as teratomas (Fig. 3 C) and displayed a variety of tissues like immature glia (Fig. 3 D) and mature cartilage (Fig. 3 E), muscle (Fig. 3 E), respiratory epithelium (Fig. 3 F) as well as squamous epithelium (Fig. 3 G). Of note, the tumors did not have a negative effect on life expectancy of affected animals, as selected mice were observed for over 1.5 years without apparent impairment of their general well-being.

Figure 3. Teratoma development in Tfap2c heterozygous mice in 129S2/Sv genetic background.

(A) Quantitative RT-PCR with RNA isolated from E12.5 genital ridges of wt and Tfap2c^+/− embryos was performed. Expression levels of Tfap2c and p21 were normalized to Gapdh. qRT-PCR was performed in biological duplicates. Error bars indicate standard deviation. (B) Percentage of total tumor incidence in Tfap2c heterozygous 129S2/Sv male mice. The seventh generation in 129S2/Sv shows 82% (n = 51) testicular tumors. Red bar: bilateral cases (35%), blue bar: unilateral tumors (47%). (C) Gross pathology of testicular teratoma in Tfap2c^+/− male mice. (D–G) HE-staining of testicular teratomas of 3–6 month old mice. Tumors show immature glia (D), mature cartilage, muscle (E), respiratory epithelium (F) and squamous epithelium (G). Scale bars: 200 µm.

Pluripotency is Retained in Foci of Tfap2c Heterozygous Testes

Tumors were reminiscent of early childhood germ cell cancer (Type I) in humans. These tumors initiate early during germ cell development and are characterized by failure of PGCs to downregulate pluripotency markers. The lesions represent hallmarks of embryonal carcinomas [1]. Here, testes of E15.5 and E16.5 Tfap2c^+/− embryos were dissected, paraffin embedded and sectioned. Immunohistochemical staining of Tfap2c^+/− gonads revealed foci of cells which maintain strong SSEA1 (Fig. 4 A) and OCT3/4 (Fig. 4 B) protein levels. The cells of these foci were still localized within the confines of the seminiferous tubule at this time point. Serial sections suggest that OCT3/4 positive foci were negative for TFAP2C (Fig. 4 C). SSEA-1 positive foci of E16.5 testes were micro-dissected and subjected to RT-PCR and adjacent (SSEA-1 negative) testes tissue served as control. In SSEA1-positive foci Nanog and Sox2 mRNA was detectable (Fig. 4 D). The fact, that pluripotency genes are expressed and cells within the foci possess large, highly condensated nuclei as well as clearly visible nucleoli (Fig. 4 B) indicate that these cells resemble embryonal carcinoma (EC) cells [33]. The absence of c-Kit expression further corroborates the EC-like nature of this lesion (Fig. 4 D).

Figure 4. Pluripotency maintaining and differentiation block show EC like nature of lesion.

(A–C) IHC staining of E16.5 embryonal testis. (A) SSEA1, (B) OCT3/4 and (C) TFAP2C. Scale bars: 50 µm. (B) Black arrow show cells with large, highly condensated nuclei as well as clearly visible nucleoli. (D) RT-PCR from RNA of microdissected SSEA-1 positive foci of E16.5 testes detecting Nanog, Sox2 and c-Kit. Comparison of microdissected SSEA-1 positive tissue (T) and SSEA-1 negative tissue (wt). Gapdh served as control.

Teratoma Development in Male Mice Heterozygous for Nanos3

Since we identified Nanos3 as a direct transcriptional target of TFAP2C we addressed the question, whether Nanos3 heterozygous mice also develop tumors. Therefore, we bred the Nanos3-Cre knock-in allele (Nanos3^{tm2.1(cre)Ysa}, MGI:4358405), which represents a null allele, [34] into the 129S2/Sv genetic background. Interestingly, from the fourth generation in 129S2/Sv 45% of Nanos3 heterozygous males showed teratoma formation (n = 20), with 5% showing bilateral tumors (Fig. 5 A), whereas control animals (n = 17) remained tumor-free. The histology of teratomas found in Nanos3 heterozygous animals is comparable to those of Tfap2c heterozygous mice (Fig. 5 B mesoderm; Fig. 5 C ectoderm; Fig. 5 D endoderm). These data suggest that teratoma susceptibility mediated by Tfap2c in 129S2/Sv genetic background is at least in part due to reduction of Nanos3. However, tumor incidence in Nanos3 heterozygous mice was lower and tumor progression was slower compared to Tfap2c heterozygous animals, arguing for additional targets of TFAP2C involved in GCC pathogenesis.

Figure 5. Teratoma development in Nanos3 heterozygous mice in 129S2/Sv genetic background.

(A) Percentage of total tumor incidence in Nanos3 heterozygous 129S2/Sv male mice. 45% (n = 20) showed testicular tumors. Red bar: bilateral cases (5%); blue bar: unilateral cases (40%). Testes of control (wt) and Nanos3 heterozygous (cre/+) mice (inset in A). (B–D) HE-staining of testicular teratomas of 4–8 month old mice. Tumors show tissues of all three germ layers: muscle (B), squamous epithelium (C) and respiratory epithelium (D). Scale bars: 50 µm.

Discussion

In this study, we analyzed the network of genes regulated by TFAP2C during germ cell development. Using an in vitro PGC differentiation protocol we observed deregulation of genes controlling cell cycle, somatic differentiation, epigenetic modification, pluripotency and germ cell maintenance and maturation upon loss of Tfap2c. ChIP analysis indicates direct binding of TFAP2C to Sfrp1, Dmrt1, Nanos3, c-Kit, Cdk6, Cdkn1a, Fgf4, Klf4, Dnmt3b and Dnmt3l. Since some of the genes deregulated upon loss of Tfap2c are known to induce GCC in mice, we bred the Tfap2c mutation onto the 129S2/Sv genetic background. We demonstrate that reduction of TFAP2C protein level leads to persisting expression of pluripotency and cell cycle markers in foci representing precursor lesions. Consequently, incidence of GCC increases up to 85% in Tfap2c^+/− males. Tumor susceptibility seems at least in part to be mediated by the TFAP2C target gene Nanos3 since Nanos3 heterozygous mice also develop GCC.

The microarray analysis identified a set of 455 deregulated genes in Tfap2c^−/−PGCLCs compared to the control. Interestingly, the deregulated genes in Tfap2c^−/−PGCLCs could be categorized in five groups representing diverse biological functions; somatic differentiation (category I), germ cell maintenance and maturation (category II), cell cycle (category III), pluripotency (category IV) and epigenetics (category V). In addition, ChIP analyses demonstrated that TFAP2C binds to regulatory elements of genes of all these categories. Thus, aside from the role as inhibitor of somatic differentiation [13], [35], TFAP2C seems to orchestrate all aspects of germ cell development instead of controlling only one category which, after deregulation would affect all others as a secondary effect.

Genes related to neural tube development/closure and central nervous system development are the GO terms significantly enriched in Tfap2c^−/−PGCLCs. Interestingly, we observed that tumors in Tfap2c^+/− and Nanos3^+/− testes predominantly consist of undifferentiated neuronal tissue (glia cells). In general, Stevens described that teratomas induced by grafting genital ridges consist predominantly of neural tissue [36]. This suggests that TFAP2C specifically inhibits neural tube development/closure and central nervous system development during germ cell development. Here, the direct transcriptional target Sfrp1 (Secreted frizzled-related protein 1) might be involved. It acts by antagonizing BMP-signaling in the caudal neural tube, resulting in neural tube closure defects [37]. This further strengthens the notion, that loss of TFAP2C leads to derepression of somatic programs, specially neuronal cell fate, and loss of germ cells in Tfap2c knockout mice [13].

In addition, loss of TFAP2C severely impairs the expression of genes involved in germ cell maintenance and maturation such as Stella, Dazl, Immp2l, Rhox4a, Rhox5, Rhox6, Rhox9, Dmrt1, c-Kit, Cxcr4 and Nanos3. Among them, Dmrt1, c-Kit and Nanos3 are direct target genes as determined by ChIP analyses. These data indicate that the lack of germ cell maturation markers is not simply a consequence of the derepressed somatic program. TFAP2C directly orchestrates the expression of the genes relevant for maintenance and maturation of PGCs/gonocytes up to E12.5, the last day expression of Tfap2c can be detected [13].

Lack of TFAP2C led to upregulation of cell cycle and pluripotency markers as well as epigenetic regulators. This indicates a role of TFAP2C to retain the capacity of germ cells to repress pluripotency, controlling the cell cycle and epigenetic reprogramming in the germ line. Here, Cdk6, Cdkn1a/p21, Fgf4, Klf4, Dnmt3b and Dnmt3l could be shown to be direct TFAP2C targets, demonstrating that TFAP2C acts as a repressor for these genes. Upon loss of TFAP2C, cell cycle control genes such as Ccnd1 and Cdk6 are upregulated. Ccnd1 (Cyclin D1) is expressed in proliferating gonocytes and spermatogonia, suggesting a role for Ccnd1 in spermatogonial proliferation [38]. Ccnd1 expression is downregulated at E16.5. Interestingly, the percentage of germ cells, which continue to express Ccnd1 correlates with the risk of the respective mouse strain to develop teratoma [39]. It is known, that upregulation of Ccnd1 and its partner molecule Cdk6 lead to phosphorylation and inactivation of the retinoblastoma gene product (pRB), inducing expression of genes necessary for G1 to S-phase transit [40]. Furthermore, ectopic expression of Ccnd1 is sufficient to promote tumor formation by mediating growth factor independence and forces quiescent cells to reenter the cell cycle [40], [41]. Of note, a rapid progression from the G1 to S phase of the cell cycle facilitates the maintenance of pluripotency [42], [43]. Hence, we believe that Tfap2c^−/− germ cells might fail to initiate the mitotic arrest and as a consequence do not differentiate further into gonocytes. Interestingly, the direct transcriptional target Cdkn1a/p21 is upregulated in Tfap2c^−/−PGCLCs and heterozygous E12.5 genital ridges. Direct binding to and repression of p21 has been described before using MCF-7 breast cancer cells [44]. While upregulation of the cell cycle repressor p21 seems counterintuitive with regard to the above reported upregulation of Ccnd1/Cdk6, it has to be noted that p21 is not only associated with cell cycle arrest [45], [46], [47] but also described as a regulator of p53-dependent and p53-independent apoptosis [48], [49]. Using Dead End ^−/− (Dnd1^−/−) mice, testicular tumors were observed after repression of Bax-mediated apoptosis [50]. Here, we speculate that Tfap2c^+/− germ cells are less prone to apoptosis due to elevated level of p21.

Loss of TFAP2C caused upregulation of the pluripotency associated genes Eras, Fgf4, Klf4 and Nr0b1. Fgf4 and Klf4 are direct transcriptional targets. In addition, it has been reported, that loss of the TFAP2C target gene Dmrt1 leads to upregulation of pluripotency markers Sox2, Nanog, Oct3/4 and E-cadherin in testes of E13.5 mice. Sox2 is a direct transcriptional target of DMRT1 [51]. In addition, we found three OCT3/4 dependent genes [52] Jam2, FoxD3 and Tdgf1 upregulated in Tfap2c^−/−PGCLCs. Interestingly, expression of Eras (ES-expressing ras) a gene related to teratoma formation in ESCs [53] was also upregulated. Again, such an upregulation is also found upon deletion of Dmrt1 [51], indicating that Dmrt1 as a downstream target of TFAP2C cooperates in balancing pluripotency and germ cell differentiation. We conclude that TFAP2C represses the transcription of pluripotency-associated genes in PGCs.

During germ cell migration loss of global DNA-methylation is observed [54]. Loss of DNA methylation in migrating PGCs occurs from approximately E8.0– E11.5 [55], [56]. TFAP2C seems to contribute to this process by direct binding to regulatory regions and repression of the methyltransferases Dnmt3b, Dnmt3l, and ubiquitin-like, containing PHD and RING finger domain 1 (Uhrf1). These genes are essential components of the de novo methylation machinery [11], [57], [58], [59]. After arrival at the genital ridges (E12.5), when Tfap2c is downregulated, the methylation levels are reinstalled [60]. Interestingly, embryonal carcinomas display high levels of DNMT3B and DNMT3L [61], [62], [63], [64]. Hence, the failure to repress the de novo methylation indicates, that these cells are prone to transform to EC-like cells. There is an association between DNA methylation and GCC susceptibility due to the fact that the global loss of DNA methylation process overlaps with the critical period for GCC formation [65].

We demonstrated here that Dmrt1 is a direct target gene of TFAP2C in PGCLCs. Interestingly, mice deficient for Dmrt1 are sterile and 90% develop GCC [51]. A recent study by Krentz et al. investigated teratoma susceptibility in Dmrt1^fl/fl : Nanos3^Cre/+ male mice [66]. Interestingly, the authors found Nanos3 haploinsufficiency to modify the susceptibility of Dmrt1 deficient male mice by elevating the tumor incidence, clearly demonstrating a role of Nanos3 in germ cell tumor development. However, in contrast to our results, the study of Krentz et al. could not demonstrate increased teratoma formation by Nanos3 heterozygosity on its own. As discussed there, this fact could be due to insufficient outbreeding to the 129Sv genetic background. Further, 129S6/Sv : 129S1/Sv mice were used instead of 129S2/Sv as used in our study. Strain specific variation in tumor susceptibility may have an influence on the experimental outcome.

Mice deficient for Pten and Dnd1 also display increased teratoma formation [67], [68], [69]. Since, we did not see downregulation of these genes in Tfap2c^−/−PGCLCs we hypothesize, that these genes act in different pathways. The observed upregulation of Pten in Tfap2c^−/−PGCLCs is most likely due to induction of somatic differentiation, since Pten has been associated with development of neuronal and synaptic structures [70].

In Tfap2c^+/− male gonads foci with cells displaying continued expression of pluripotency markers Oct/4, Ssea1, Sox2 and Nanog were observed. In fact, this represents a hallmark critical for transformation of PGCs into embryonal carcinoma (EC) cells between E11.5–13.5 in mice [65]. Usually postmigratory PGCs downregulate pluripotency markers and enter mitotic arrest. However, EC cells show a block of germ cell differentiation and continued proliferation and expression of pluripotency markers also seen in human pediatric cancer.

In conclusion we provide evidence that the transcription factor TFAP2C orchestrates many aspects of the genetic network regulating germ cell development (Fig. 6). Maintenance and maturation of PGC are controlled by TFAP2C by repression of somatic differentiation, cell cycle, epigenetic remodeling, and pluripotency associated genes (Fig. 6). Furthermore, mice heterozygous for Tfap2c or its downstream target Nanos3 display an increase in GCC susceptibility.

Figure 6. Schematic of the genes and programs regulated by TFAP2C in primordial germ cells.

Black arrows indicate pathways transactivated and induced by TFAP2C and black lines with terminal bars indicate pathways repressed by TFAP2C during development of primordial germ cells. Genes listed in respective pathways indicate direct regulation as demonstrated by ChIP analyses.

Supporting Information

Figure S1

Generation of Tfap2c^ctrl and Tfap2c^−/− ESCs. (A) Immunofluorescence staining against TFAP2C and SSEA1 protein in ctrl and Tfap2c^−/−ESCs (cell line #1 and #2). Nuclei are stained with Hoechst. Scale bars: 100 µm. (B) Genotyping of Blimp1mVenus/Tfap2c^flox/flox and Blimp1mVenus/Tfap2c^−/− ESC lines by PCR. (C) FACS analysis to identify Blimp1mVenus positive PGCLCs. Population P3 showing mVenus positive cells. 8.1% Blimp1mVenus positive cells in ctrl whereas 3.2% positive cells measured in Tfap2c^−/− cells.

(TIF)

Figure S2

In vitro differentiation of Tfap2c^−/− and ctrl PGCLCs. (A) Brightfield pictures of #1-ctrl ESCs and after 1 and 2 days of in vitro differentiation into EpiLCs. EpiLCs show a flattened epithelial-like structure (A, middle). Scale bars: 100 µm (B) ESC and Epiblast markers were analysed by RT-PCR of RNA from #1-ctrl; #1-Tfap2c^−/− and #2-ctrl; #2-Tfap2c^−/− ESCs and EpiLCs. β-Actin served as control. (C) Brightfield pictures show #1-ctrl and #1-Tfap2c^−/− EBs. PGCLC induction is indicated by Blimp1mVenus fluorescence. Scale bars: 100 µm (D) Graph showing the percentage and standard deviation of PGCLCs generation efficiency, measured by FACS for #1-ctrl (9.3%) and #2-ctrl (3%). Efficiency of PGCLC formation is lower in Tfap2c^−/− cells (#1∶2.7% and #2∶1.5%). Significance * P≤0.05. (E) RT-PCR from RNA of #1-ctrl; #1-Tfap2c^−/− and #2-ctrl; #2-Tfap2c^−/− PGCLCs. Expression of early germ cell markers (Stella/Dppa3, Prdm14, Nanos3) is reduced in Tfap2c^−/−PGCLCs. As expected no signal was detected in Tfap2c^−/−PGCLCs. β-Actin served as control.

(TIF)

Table S1

Deregulated genes in Tfap2c^−/−PGCLCs (Fold-change >1.5 in log2 scale).

(XLSX)

Table S2

Deregulated genes in Tfap2c^−/−ESCs (Fold-change >1.5 in log2 scale).

(XLSX)

Table S3

Commonly deregulated genes in Tfap2c^−/−ESCs and -PGCLCs (Fold-change >1.5 in log2 scale).

(XLSX)

Table S4

Location of TFAP2/TFAP2C binding sites and primer sequences.

(XLSX)

Table S5

Primer sequences for RT-PCR.

(DOCX)

Table S6

Primer sequences for qRT-PCR.

(DOCX)

Table S7

List of abbreviations.

(DOCX)

Data S1

Gene Ontology analysis for >1.5 fold-change in log2 scale for Tfap2c^−/−PGCLCs in comparison to ctrl PGCLCs.

(PDF)

Funding Statement

This work was supported by DFG Grant Scho-503/9 to HS, and Deutsche Krebshilfe #108249 to HS. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.