Magnúsdóttir et al 2013 a tripartite transcription factor network regulates primordial germ cell spe by magnusdottir

ARTICLES

A tripartite transcription factor network regulates primordial germ cell specification in mice Erna Magnúsdóttir1,2,6 , Sabine Dietmann3 , Kazuhiro Murakami1,2 , Ufuk Günesdogan1,2 , Fuchou Tang1,2,6 , Siqin Bao1,2,6 , Evangelia Diamanti4 , Kaiqin Lao5 , Berthold Gottgens4 and M. Azim Surani1,2,3,7 Transitions in cell states are controlled by combinatorial actions of transcription factors. BLIMP1, the key regulator of primordial germ cell (PGC) specification, apparently acts together with PRDM14 and AP2γ. To investigate their individual and combinatorial functions, we first sought an in vitro system for transcriptional readouts and chromatin immunoprecipitation sequencing analysis. We then integrated this data with information from single-cell transcriptome analysis of normal and mutant PGCs. Here we show that BLIMP1 binds directly to repress somatic and cell proliferation genes. It also directly induces AP2γ, which together with PRDM14 initiates the PGC-specific fate. We determined the occupancy of critical genes by AP2γ—which, when computed altogether with those of BLIMP1 and PRDM14 (both individually and cooperatively), reveals a tripartite mutually interdependent transcriptional network for PGCs. We also demonstrate that, in principle, BLIMP1, AP2γ and PRDM14 are sufficient for PGC specification, and the unprecedented resetting of the epigenome towards a basal state. PGCs in mice originate from the rapidly dividing post implantation epiblast cells that are primed for somatic fate, following repression of some pluripotency genes1 . They also exhibit an inactive X chromosome, histone H3 Lys 9 dimethylation (H3K9me2) and DNA methylation2,3 . A transcriptional network for PGC specification should reverse this trend by the time 30–40 founder PGCs are established at embryonic day 7.5 (E7.5). PGC fate is initiated by BMP4-induced expression of BLIMP1 in a few proximal epiblast cells at E6.25 (refs 4–8), which marks their divergence from somatic neighbours (see Fig. 3b). Indeed, BLIMP1 mutant cells fail as PGCs and resemble neighbouring somatic cells7,9–11 . BLIMP1 binds to a specific DNA sequence12–20 to either repress21–25 or activate26 its direct targets. Shortly after BLIMP1, there is induction of Prdm14 also by BMP4 (ref. 6), followed by Tcfap2c encoding AP2γ (ref. 27). Genetic experiments indicate that these factors are individually critical for PGC specification. It is important however to establish whether their combinatorial roles and precise targets are necessary and sufficient for PGC specification, and for the initiation of the unique epigenetic program28 . In this study we combined information from different experimental models to establish how BLIMP1, PRDM14 and AP2γ contribute to

PGC specification, both individually and combinatorially. We propose a tripartite transcriptional network that accounts for PGC specification and their unique properties. Indeed, co-expression of BLIMP1, AP2γ and PRDM14 in an in vitro model can substitute for cytokines in the direct induction of PGC-like cells (PGCLCs). Close scrutiny of the genetic network also provides a detailed view of how these genetic factors regulate the unique epigenetic program in germ cells, which might serve as a paradigm for wider applications in the context of tissue regeneration and experimental manipulation of cell fates. RESULTS We first sought an in vitro surrogate cell-culture system to examine the individual and cooperative roles of BLIMP1, PRDM14 and AP2γ, and to identify their direct targets by chromatin immunoprecipitation (ChIP) experiments, which requires large amounts of material. This is difficult with PGCs because they are relatively rare, and difficult to culture, transfect and manipulate. We therefore investigated BLIMP1 expression in several primary cell types, embryonic stem cells (mESCs), embryonic germ cells and epiblast stem cells, but none of them survived except for P19 embryonal carcinoma cells28 (P19EC; Fig. 1a). Indeed, P19EC cells are also appropriate for this purpose because they

Wellcome Trust/Cancer Research UK Gurdon Institute, University of Cambridge, Tennis Court Road, Cambridge CB2 1QN, UK. 2 Department of Physiology, Development and Neuroscience, University of Cambridge, Downing Street, Cambridge CB2 3DY, UK. 3 Wellcome Trust-Medical Research Council Stem Cell Institute, University of Cambridge, Tennis Court Road, Cambridge CB2 1QN, UK. 4 Cambridge Institute for Medical Research, Wellcome Trust–MRC Building, Hills Road, Cambridge CB2 0XY, UK. 5 Genetic Systems, Applied Biosystems, Part of Life Technologies, 850 Lincoln Centre Drive, Foster City, California 94404, USA. 6 Present address: Department of Biochemistry and Molecular Biology, University of Iceland, Vatnsmyrarvegi 16, 101 Reykjavík, Iceland (E.M.); College of Life Sciences, Peking University, Yiheyuan Road, Beijing, 100871, China (F.T.); Inner Mongolia University, College of Life Science, Hohhot 010021, China (S.B.). 7 Correspondence should be addressed to M.A.S. (e-mail: a.surani@gurdon.cam.ac.uk) Received 26 March 2013; accepted 3 June 2013; published online 14 July 2013; DOI: 10.1038/ncb2798

905

ARTICLES a

BLIMP1 AP2γ

PGCs

Transcriptional network

Expression analysis PRDM14 BLIMP1 ChIP-seq

Single-cell RNA-seq

P19ECs PGCs

AP2γ ChIP-seq

PRDM14 ChIP-seq 2.5 2.0 1.5 1.0 0.5 0

Eomes

8 T-Brachyury

1.5

1.0

Relative levels (normalized to Ppia)

Gata2

24 h 24 h 48 h HI LO

Gata6

100

20 50

0 24 h 24 h 48 h HI LO

Dppa3

24 h 24 h 48 h HI LO

50 40 30 20 10 0

24 h 24 h 48 h HI LO

0.5

Tcfap2c

Nanos3

1,500

Dnd1

24 h 24 h 48 h HI LO

Dazl

24 h 24 h 48 h HI LO

Prdm14

24 h 24 h 48 h HI LO

Ddx4

EGFP

BLIMP1

Rhox9

500

24 h 24 h 48 h HI LO

1,000

24 h 24 h 48 h HI LO

Nanog

1.0

0.5

2.0 1.5

24 h 24 h 48 h HI LO

150

Dnmt3b

Relative levels (normalized to Ppia)

Eomes

1.2 1.0 0.8 0.6 0.4 0.2 0

Myc

2.0 1.6 1.2 0.8 0.4 0

Prdm1

6 5 4 3 2 1 0

1.2 1.0 0.8 0.6 0.4 0.2 0

25 20 15 10 5 0

3.5 3.0 2.5 2.0 1.5 1.0 0.5 0

T-Brachyury

Nanos3

Ddx4

2.5 2.0 1.5 1.0 0.5 0

25 20 15 10 5 0

9 8 6 4 2 0

Dnmt3b

Dnd1

Dppa3

Control

PRDM14

AP2γ

AP2γ + PRDM14

BLIMP1

PRDM14 + BLIMP1

AP2γ + BLIMP1

PRDM14 + AP2γ + BLIMP1

0 24 h 24 h 48 h HI LO

24 h 24 h 48 h HI LO

Figure 1 BLIMP1, AP2γ and PRDM14 repress somatic regulators and induce PGC genes in P19ECs. (a) Design and overview of the experimental approaches towards transcriptional network for PGC specification. The arrow pointing to PGCs refers to EGFP-reporter-labelled PGCs in mouse embryos. (b) RT–qPCR analysis of differentially regulated genes downstream of BLIMP1 in P19ECs, showing induction of the PGC genes Nanos3, Rhox9, Dppa3, Tcfap2c, Dnd1, Prdm14, Dazl and Ddx4. HI, sorted cells

with high BLIMP1–EGFP or EGFP fluorescence; LO, sorted cells with low BLIMP1–EGFP or EGFP fluoresence. (c) RT–qPCR analysis showing the individual and combinatorial effect of BLIMP1, AP2γ and PRDM14 in P19ECs on the mesendodermal transcription factor genes Eomes and T -Brachyury, the DNA methyltransferase Dnmt3b, Myc and the PGC genes Nanos3, Dnd1, Prdm1, Ddx4-Vasa and Dppa3. The error bars in b,c represent mean ± s.d. for three independent cell cultures.

originate from E7.5 epiblasts29 , and share important properties of post implantation epiblasts, the precursors of PGCs in vivo30,31 .

DNA methyltransferase Dnmt3b (Supplementary Fig. S1b), which are amongst the key responses observed in PGCs2,32 . Importantly, the PGC genes Nanos3 and Rhox9 were induced. By lowering the statistical threshold to a false discovery rate (FDR) ≤0.05, we detected an induction of Dppa3–Stella and Tcfap2c (encoding AP2γ; Supplementary Fig. S1b and Table S1). Furthermore, quantitative PCR with reverse transcription (RT–qPCR) revealed an induction of Dnd1 and Prdm14 at 48 h, and PGC markers, Dazl and Ddx4 (Fig. 1b). Whereas Oct4–Pou5f1 expression continued (Supplementary Table S1), we noted repression of Nanog, which could explain the induction of Gata4 and FoxQ1 (ref. 33; Supplementary Fig. S1b and Table S1). Overall, and in important respects, the response of P19EC to BLIMP1 approximates that seen during PGC specification in vivo.

Repression of somatic program and induction of PGC genes in P19EC cells We examined P19EC cells for transcriptional response following ectopic expression of BLIMP1–EGFP fusion protein or EGFP alone after 24 h with low (24 h-LO) and high expression (24 h-HI), and all fluorescent cells at 48 h (Supplementary Fig. S1a). Whereas the 24 h-HI cells showed an apoptosis response due to a strong dose-dependent effect24 , this was not the case with 24 h-LO cells. We therefore focused mainly on 24 h-LO and 48 h cells (Supplementary Figs S1b,S2). BLIMP1 in P19EC cells induced the repression of genes including the mesendodermal factor Eomes, HoxA5, Evx1, Myc and the de novo

906

ARTICLES

BLIMP1

6 × 10–4

Inter-genic 2,480 2,566

1 35

Intra-genic

Id3

5 kb

Intergenic peaks

1 35

427 1,248

Myc

794

1 kb 5' TSS 1–10 kb 5' TSS 1 kb 3' TES Distant

4 × 10–4

0 × 100 –10 –5 0 5 10 Distance to nearest TSS (kb)

50 kb

2 × 10–4

Bits

1 35

1 31

Molecular function –log (P value)

a1a2

5730446D14Rik

2700086A05Rik

a9 a11 a13 a7 a10 a11as

100 kb

Input 1 50

BLIMP1

b13

0 20 40 60 80 Nucleic acid binding DNA binding Transcription factor activity RNA binding Chromatin binding Transcription cofactor activity

b8 b3 b7b6 b4 b2 b1 b5

–log (P value)

50 kb

0 204060 80100 1 43

BLIMP1 1

c13 c12 c11 c9 c8 c10

c6 c5

50 kb

Input 1 30

Nucleic acid metabolic process* Cellular process Cell cycle Developmental process System development Ectoderm development Nervous system development Signal transduction Mitosis Mesoderm development

BLIMP1 1

C C

C T

1 2 3 4 5 6 7 8 9 1011

P value: 1.2 × 10–388

0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00

EGFP BLIMP1

Biological process

Input

S Br nai3 ac M Br hP yc ac ro hD x i T st Eh af3 m t1 M Taf ed 7 Eo 3 m1 e Gs U rn h M rf ed 1 13

BLIMP1

Percentage of input

Input

700 600 500

Pre-immune serum BLIMP1 serum

100 30 20 10 0 Sn Dn ai3 m t3 Eo b m H es ox B2 M yc

Input

All peaks

Percentage of amplified input

BLIMP1

2 kb

Input

Density

d3 Evx2 d12 d10 d4 d13 d11 d9 d8

1700109F18Rik

Figure 2 BLIMP1 binding to gene promoters encoding transcription factors, and cell-cycle and developmental regulators. (a) Track view of BLIMP1 ChIP-seq density profile shown using the UCSC genome browser centred on the Id3 and Myc genes. (b) BLIMP1 peaks distributed roughly evenly between inter- and intragenic positions, with 1,248 of 2,480 intergenic peaks falling within 1 kb 50 of the TSS, and 527 peaks were more than 10 kb away from promoters (TSS, transcriptional start site; TES, transcriptional end site). (c) Distribution of BLIMP1 binding relative to promoters revealing a median distance of +171.5 bp from the TSS. (d) De novo motif analysis revealed high enrichment for the BLIMP1 consensus (P value of 1.2 × 10−388 ). (e) BLIMP1 binding profiles on Hox gene clusters are indicated in the views with their respective numbers. For example, Hoxa1

is indicated by a1, Hoxa2 is a2 and so forth. (f) Functional categories of genes bound by BLIMP1 showing the P value for the molecular function as well as biological process GO-terms. ∗ The full name of this GO-term is: nucleobase, nucleoside, nucleotide and nucleic acid metabolic process. (g) Validation of BLIMP1-binding regions by ChIP–qPCR in P19EC cells. The results are representative of two independent experiments. (h) A validation of BLIMP1-binding regions in PGCLCs by ChIP followed by whole-genome amplification of precipitated and input DNA assayed by qPCR. The y axis represents the percentage of signal from amplified input material at the same starting quantity of the immunoprecipitated DNA to ensure proportional amplification. The results are representative of two independent experiments.

We then examined the effects of all three factors in P19EC cells following stable expression of PRDM14 and AP2γ, both individually and together. We transfected control cells and the stable lines with BLIMP1–EGFP or EGFP alone for 24 h and examined sorted fluorescent cells. This showed repression of Eomes and T-Brachyury, whereas PRDM14 alone suppressed Dnmt3b (Fig. 1c), its direct target34 . Whereas BLIMP1 repressed Myc, PRDM14 in combination with AP2γ modestly induced Myc expression, an effect that was overcome by BLIMP1 expression. Thus, repression of somatic regulators is complex, and may not be attributable to BLIMP1 alone. The induction of PGC genes revealed cooperative effects of AP2γ and PRDM14, which induced Nanos3 and Prdm1 (encoding BLIMP1),

with a modest induction of Ddx4. Whereas Nanos3 induction was attenuated by BLIMP1, Dnd1 was induced by 15-fold when all three factors were present, but Dppa3 was strictly PRDM14 dependent (Fig. 1c). These observations show that PRDM14 and AP2γ cooperatively induce the germ cell program, with the additional effect of BLIMP1 on Dnd1 induction. The analysis of P19ECs shows a response to BLIMP1, PRDM14 and AP2γ individually and collectively, with features that are pertinent to PGCs, including the repression of somatic genes and induction of PGC genes. We posit that P19EC cells are appropriate for the identification of direct targets of the three key determinants of PGC specification.

907

E8.5 PGC

220

E7.5 PGC

240

E6.5 PGC

260

E7.5 soma

280

Read count (log2 (RPM))

300

E7.5 BLIMP1KO

Euclidean distance (log2 (RPM))

ARTICLES

14 12 10 8 6 4 2 0

E7.5 E7.5 E6.5 E7.5 E8.5 KO soma

PGCs

Read count (log2 (RPM))

22 20 18 16 14 12 10 8 6 4 2 0

Prdm1 Prdm14 Tcfap2c

BLIMP1 PGCs

Tcfap2c T-Brachyury Cbx7 Dnmt3b

1 35 BLIMP1

12 10 8 6 4 2 E7.5 E7.5 E6.5 E7.5 E8.5 KO soma

Myc Ccnd1 Hoxa1 Hoxb1 Hoxa2

E7.5 E7.5 E6.5 E7.5 E8.5 KO soma

PGCs

Nanos3 Dppa3 Dnd1 Dazl Ddx4

1 35 1 Tcfap2c 10 kb

35 Input 1 35

10 kb

BLIMP1

Input

PGCs

Pou5f1 Nanog Sox2 lfitm1

Dnmt3b 5 kb

PGCs

Input 35

20 18 16 14 12 10 8 6 4 2 0

E7.5 E7.5 E6.5 E7.5 E8.5 KO soma

20 kb

35 Input

E7.5 E7.5 E6.5 E7.5 E8.5 KO soma

18 16 14 12 10 8 6 4 2 0

BLIMP1 _

T-Brachyury

1 Cbx7

Figure 3 RNA-seq analysis of PGCs, and BLIMP1 binding to differentially regulated genes. (a) Cluster analysis of single-cell RNA-seq expression profiles of PGCs and somatic neighbours. The bar indicates the mean Euclidean distance between the samples. Note that E7.5 BLIMP1 mutant (KO) cells cluster next to E7.5 somatic cells, whereas the PGCs form distinct clusters following specification. (b) Expression levels of selected gene transcripts during PGC specification assayed by single-cell RNA-seq. The y axis represents read numbers per

million reads (RPM) sequenced. The region shaded in blue represents BLIMP1 mutant (KO) cells. The dotted line represents the onset of PGC specification. The results are the mean of two single cell RNA-seq profiles at each time point assayed. (c) Expression of Dnmt3b, T-Brachyury, Tcfap2c and Cbx7 during PGC specification, which are all bound by BLIMP1 during PGC specification. (d) Track views of BLIMP1 binding to the genomic loci of Dnmt3b, T-Brachyury and Tcfap2c (encoding AP2γ) and Cbx7.

Identification of BLIMP1 targets and their relevance for PGC specification To identify BLIMP1 targets, we transfected P19ECs with BLIMP1–EGFP, followed by ChIP sequencing (ChIP-seq) using an EGFP antibody. This revealed 5,046 BLIMP1 high-confidence binding peaks for 4,389 protein-coding and 313 non-coding genes (Supplementary Tables S2 and S3), including 8/11 known targets such as Myc and Id3 (Fig. 2a). We observed a peak distribution with a median position of +171.5 base pairs (bp) relative to transcriptional start sites (TSSs; Fig. 2b,c) consistent with BLIMP1 binding on promoters and an enrichment of the previously characterized consensus binding sequence for BLIMP1 (ref. 35; P = 1.1 × 10−388 ; Fig. 2d). Notably, BLIMP1 bound to T-Brachyury, Eomes and the entire Hox gene loci (Fig. 2e and Table S3) reflecting its role in PGC specification in vivo7,9 . Functional category analysis revealed a striking enrichment of BLIMP1 binding to genes encoding transcriptional regulators and of genes regulating developmental processes (Fig. 2f). Moreover, BLIMP1 was bound to Tcfap2c (encoding AP2γ), which is induced in PGCs (see later). We validated Myc and several of the BLIMP1-bound regions identified by ChIP–qPCR in P19ECs (Fig. 2g). We also validated BLIMP1 binding to Eomes, Dnmt3b, HoxB2 and Myc in PGCLCs

generated in vitro36 with a ChIP-grade BLIMP1 antibody35 (Fig. 2h), but comprehensive analysis in PGCLCs was technically not feasible owing to limited amounts of precipitated DNA that could be generated. To determine the significance of BLIMP1 targets, we scrutinized transcriptional changes in wild-type and Prdm1 (BLIMP1) mutant PGCs from E6.25–E8.5 embryos, including E7.5 somatic neighbours, which share a common ancestry (Fig. 3a). For this, we performed single-cell RNA-seq analysis and found that all three factors, Prdm1 (encoding BLIMP1), Tcfap2c (encoding AP2γ) and Prdm14, were fully induced by E6.5 in putative PGCs, with extensive repression of somatic and cell-cycle regulators, and induction of the PGC program2,7 (Fig. 3b,c and Supplementary Fig. S3a,b, Table S4). In contrast, we detected expression of somatic genes and a lack of expression of some germ cell genes in Prdm1 mutant cells (Fig. 3b,c). Overall, the E7.5 single-cell expression profiles of Prdm1 mutant cells clustered with E7.5 somatic cells and not with PGCs (Fig. 3a). We then carried out a global assessment of BLIMP1-bound genes in relation to the differentially expressed genes in PGCs (compared with the binding to the whole set of expressed genes). Indeed, BLIMP1 was bound to both repressed (Brachyury and Dnmt3b), and induced (Cbx7 ) genes. Notably, BLIMP1 bound to Tcfap2c (encoding AP2γ), which is induced (Fig. 3c,d and see below). We calculated the binding

908

0.1

–0.1

Fundc1 Gclm *Cxcl12 *Spata13 *Rbck1 Lrrfip2 *Zscan10 *Senp7 2310047B19Rik *Rab28 *Tmem43 Supv3l1 Pip4k2a *Myc *Meis2 Eps8 *Stk39 Stard5 *Klf9 Rnf219 Ttc39b *Rab8b *Ppp1r9a *Wnt5b Usp3 *Rragd *2210411K11Rik *Dnmt3b *Eomes Fabp5 *Rbpms2 Bzw2 *Etv1 Odz4 Plce1 Ppif *Has2 Prps1 Xkr5 *2310044G17Rik *Fst Gja1 BLIMP1 peak: * *Bcl9 *Rerg Log2 (RPM): Fkbp5 *Upp1 Heg1 10.00 Steap1 6.67 *Tmem98 3.34 *Hoxa2 0 *Leo1 Nrg4 Log2 (FC): *Zfp60 Arfip2 –5.00 *Phf6 *Acss2 –3.33 Hus1 –1.67 *Dbt 0 *Mpi

–0.1 Repressed

Induced

–0.2

Repressed

Induced

–0.2

–0.3

–0.3 –3 –2 –1

–3 –2 –1

Log2 (fold change)

b 70% 60%

Log2 (fold change)

Induced Repressed

50% 40% 30% 20% 10% 0%

Whole 24 h array HI

24 h LO

E7.5PGC versus soma B24LO versus E24LO B48 versus E48 BLIMP1 peak

0.2

BlLIMP1 KO

0.2

Peaks Scores

c E8.5 PGC

BLIMP1 target enrichment (log2)

Peaks Scores

E7.5 PGCs versus soma 0.3

E7.5 PGC

E7.5 PGCs versus BLIMP1 KO 0.3

E6.5 PGC

E7.5 soma

ARTICLES

48 h

Figure 4 BLIMP1 represses most of its direct targets. (a) Relative enrichment of BLIMP1-binding regions and the scores associated with genes differentially expressed between E7.5 PGCs and BLIMP1 mutant (KO) cells, and between E7.5 PGCs and E7.5 somatic cells, respectively. The x axes indicate the log2 (fold change) and the y axes indicate the log2 of the BLIMP1 target enrichment at each fold change interval of differentially expressed genes over the average target frequency of the whole expression data set. Peaks: the enrichment of peaks associated with genes in each interval at E7.5. Scores: the enrichment of binding scores calculated for genes in each

interval at E7.5. (b) Binding frequency of BLIMP1 to genes associated with features on the whole microarray as well as differentially regulated genes. (c) Heat map depicting the genes repressed by BLIMP1 in both P19ECs and during PGC specification. The first 4 (blue) columns refer to normalized gene expression levels in PGCs. The next 3 columns (red) reflect values of differential expression between the samples indicated. The asterisks in the final column indicate a high-confidence BLIMP1-binding region associated with a gene. BLIMP1 binding was detected in 34/59 repressed genes in both data sets. (RPM: read numbers per million. FC: fold change.)

scores for reads both inside and outside peak regions and their distance to promoters (see Methods), as well as the defined peak regions. This revealed that the repressed genes in E7.5–E8.5 PGCs, compared with either somatic cells or Prdm1 (BLIMP1) mutant cells, are enriched for BLIMP1 (Fig. 4a). The repressed genes that were bound by BLIMP1 had a greater enrichment for developmental, transcriptional and Wnt-signalling function when compared with the whole set of repressed genes (Supplementary Fig. S3a). As misregulated gene expression in the Prdm1 mutant cells from in vivo is a direct consequence of the lack of BLIMP1 in these cells, this result further shows the functional relevance of our analysis to the processes occurring during PGC specification in vivo.

Comparing the ChIP-seq data with the BLIMP1-induced changes in P19ECs revealed a predominant effect on repression of direct targets. Whereas BLIMP1 bound with 24.1% of the genes in the genome (Fig. 4b, Whole Array), nearly 50% of the repressed genes were bound by BLIMP1, but binding to induced genes was close to background (Fig. 4b). We further observed a striking enrichment of BLIMP1 binding on repressed genes in both P19ECs as well as in E7.5 PGCs when compared with soma (Fig. 4c and Supplementary Fig. S3e) where 34/59 repressed genes were bound by BLIMP1. This is statistically a very high over-representation, (χ 2 P-value = 1.8×10−10 ), which shows conclusively that the dominant effect of BLIMP1 in PGC specification is gene repression, including those required for the somatic fate.

909

ARTICLES a

c Bits

1 267

1 0

AP2γ

1 2 3 4 5 6 7 P value: 1.1 × 10–241

5 kb

270 Input

10 kb

Density (×10–5)

Mir181d Nanos3 Mir181c

Mir23a Mir27a Mir24-2 Mir3074-2

1 0

192

–20 –10 0 10 20 Distance to nearest TSS (kb)

Input

1 192

AP2γ

0.16%

1 94 AP2γ 1 Hoxa9

5 kb

Hoxa10 Mir196b

Hoxa11 Hoxa11as Hoxa11as

10 kb

55 Input

1 55

_ -

AP2γ

Hes7

Isotype control

10 kb

100

Dppa3

Input

1 247

AP2γ

1 Gdf3

Pou5f1

Input

d 0.18%

1 61

AP2γ

247

10 kb

100 Input

T-Brachyury

MGI expression: detected –log (FDR)

AP2γ ChIP

0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

TS21 testis; primordial germ cells TS17 forelimb bud; ectoderm TS20 primordial germ cells TS8 epiblast

0.14% 0.12% 0.10%

0.08%

Biological process

0.06% 0

0.04% 0.02%

DE 4 ct

Dp p

an N

–log (FDR) 2 4 6

Stem cell development Stem cell maintenance Head morphogenesis Stem cell differentiation Negative reg. of myeloid cell differentiation Face morphogenesis Body morphogenesis Head development PDGFR signaling pathway Epithelial cell development Somatic stem cell maintenance

Figure 5 AP2γ binds to germ cell genes and somatic regulators. (a) De novo motif analysis revealed high enrichment for the AP2γ-consensus-binding site with a P value of 1.1 × 10−241 . (b) Distribution of AP2γ binding relative to promoters revealing a mean distance of +53 bp from the TSS. (c) Track view of AP2γ binding on the Pou5f1 distal enhancer, to the PGC genes Dppa3 and Nanos3, and the somatic differentiation regulators HoxA10, HoxA11, Hes7 and T-Brachyury. (d) A ChIP–qPCR of AP2γ binding to

the promoter of Nanos3, Dppa3–Stella and the distal enhancer of Oct4 (Oct4-DE). The results are representative of three independent experiments. (e) Over-represented gene categories sorted by cell-type-specific expression at different Theiler stages (TS) during embryonic development; TS8 corresponds roughly to E6.0, TS17 to E10.5, TS20 to E17 and TS21 to E21-P0. (f) Over-represented functional categories of genes bound by AP2γ showing the P value for the enrichment of biological process GO-terms.

AP2γ binds to germ cell genes and somatic regulators Towards building a genetic network for PGC specification, we performed an unbiased scan of the BLIMP1-binding regions for transcription factor binding motifs (Supplementary Fig. S4a). We found a bimodal distribution of AP2 family motifs surrounding the BLIMP1 peak (Supplementary Fig. S4b). Similarly, PRDM14-binding sites were also highly enriched for AP2 motifs34 (Supplementary Fig. S4c,d). We therefore mapped PRDM14- and BLIMP1-binding sites to all of the genes in the PGC transcriptome, and then found AP2γ motifs within the binding regions37–39 . This revealed a preferential enrichment of genes regulated during PGC specification that were associated with BLIMP1- and PRDM14-binding sites that contained AP2γ motifs (Supplementary Fig. S4e). This strongly implies that the three factors cooperate both in gene induction and repression in PGCs. This prompted us to carry out ChIP-seq analysis for AP2γ. With P19ECs stably expressing AP2γ, we performed ChIP-seq with a previously validated antibody39 , and identified 3,191 high-confidence

AP2γ-binding regions that map to 1,393 genes (Supplementary Tables S5 and S6). The peaks were enriched with the AP2 consensus motif (Fig. 5a) (P = 1.1 × 10241 ). AP2γ binding was centred on promoters at a median position of − 53 bp relative to the TSS, albeit the peak distribution was much broader than that of BLIMP1 (Fig. 5b), perhaps implying binding to gene-distal sequence elements. Importantly, the Pou5f1 distal enhancer, which is bound by PRDM14 and pluripotency transcription factors, was amongst the most strongly bound regions34,40,41 (Fig. 5c). Notably, Pou5f1 expression in the post implantation epiblast and P19ECs is driven from the proximal enhancer, whereas the distal enhancer is used following PGC specification. Several somatic genes were also bound by AP2γ, including Hoxa11, HoxB13 and T-Brachyury, and regions distal to the TSS of Nanos3 and the first intron of Dppa3 (Fig. 5c,d and Supplementary Table S6). Analysis of functional categories of targets revealed highly significant binding to PGC genes (for example, Nanog, Dppa3 and Pou5f1), and

910

ARTICLES

Target enrichment (log2)

0.8

1.5

0.6

1.0

0.4 0.2

0.5

–0.2

–0.5

–0.4

–1.0 –5 –4 –3 –2 –1 0 Log2 (fold change)

AP2γ PRDM14 BLIMP1

PRDM14 BLIMP1

AP2γ PRDM14

PRDM14

1 31

267

BLIMP1

533

PRDM14 BLIMP1

0 1 2 3 4 5 6 Log2 (fold change)

BLIMP1/AP2γ PRDM14/AP2γ BLIMP1/PRDM14/AP2γ

PRDM14 P value < 0.001 (permutation test)

368 240

1740

1 21 1

T-Brachyury

8072

10 kb

156

PRDM14

1 35

P value < 1 × 10–299 (χ2 test)

Dnmt3b 5 kb

Pou5f1

BLIMP1 1058

53 AP2γ

628

5 kb

PRDM14

1 58

Uhrf1

AP2γ

195 35

25,788

5 kb

1 188 1

4,573

2,682 65

–5 –4 –3 –2 –1 0 Log2 (fold change)

Mir10a Hoxb4

1 33 1

BLIMP1 PRDM14 AP2γ

AP2γ

BLIMP1 AP2γ

10 kb

1 218

206

Induced

0 1 2 3 4 5 6 Log2 (fold change)

5 kb

Repressed

BLIMP1/AP2γ PRDM14/AP2γ BLIMP1/PRDM14/AP2γ

BLIMP1 PRDM14 AP2γ

Induced

Repressed

2.0

–log10 (enrichment P value)

AP2γ 44 1 PRDM14 70 1

60 1 81 1 Nanos3 Mir181c

Mir181d

2 kb

Dppa3

Migration

PRDM14

Germ cell genes

Epigenetic programme

AP2γ

Somatic regulators

BLIMP1

Proliferation

Figure 6 A transcription factor network for PGC specification. (a) Relative enrichment of BLIMP1, AP2γ and PRDM14 targets on differentially expressed genes between E7.5 PGCs and soma, and the combinatorial association of the peak regions to the differentially expressed genes. (b) The plots depict the hypergeometric P values for the corresponding enrichment of BLIMP1, AP2γ and Prdm14 targets on differentially expressed genes shown in a. (c) Venn diagrams showing the total number of genomic target sites overlapping by one base pair or more between BLIMP1, AP2γ and

PRDM14. The P value for the enrichment of overlap between the factors of P < 0.0001 was calculated using a permutation test. See details in Supplementary Fig. S8a. (d) Venn diagrams showing the total number of genes overlapping between BLIMP1, AP2γ and PRDM14. The P value of P < 1 × 10−299 was calculated using a χ 2 test. See details in Supplementary Fig. S8b. (e) Track view of BLIMP1, PRDM14 and AP2γ binding to selected repressed and induced PGC targets. (f) A transcriptional network model depicting the role of PRDM14, BLIMP1 and AP2γ during PGC specification.

E6.0 epiblast (for example, activin and FGF receptor genes Acvr1b, Fgfr1 and Fgfr2). This is consistent with AP2γ being involved in PGC specification from epiblast cells (Fig. 5e,f and Supplementary Table S6). Furthermore, AP2γ was enriched for genes involved in morphogenesis and development, indicating a relevance to PGCs and somatic gene repression.

Transcriptional network for PGC specification Next, we combined all available information for a detailed scrutiny of how the key regulators induce PGCs. With respect to gene expression in PGCs versus soma at E7.5, AP2γ was significantly enriched on both induced and repressed genes (Fig. 6a), as confirmed by the hyper-geometric distribution statistical test (Fig. 6b). The repressed

911

ARTICLES a

1,333

NELFA 13,250

BLIMP1

948

SIN3A

3,713

b BLIMP1

NELFA

BLIMP1

4,098

7,098

9,613 2,364

1,725

SIN3A

BLIMP1 1,366

5,625 727

703

N-MYC 6,063

BLIMP1 4,343

387

529

TBP

BLIMP1

1340 N-MYC

4,517

3,087

BLIMP1

711

SPT5 2,837

RING1B

4,659

442

4,685

1,751

4,855

SPT5

BLIMP1

2,582

BLIMP1

EZH2

4,604

BLIMP1

312

4,060

2,380

BLIMP1

2,768

881

2,210

EZH2 2,052

642

BLIMP1 2,449

5 kb

2,842 887 2,204

BLIMP1

RING1B

4,734

BLIMP1

TBP

NELFA

EZH2

1 25

1 60 SIN3A

RING1B 1 50

1_ 30 BLIMP1

BLIMP1 1 Hoxb13 AK078606 100 kb

1 AK078566

Hoxb8 Hoxb5

Mir196a-1 Hoxb7 Hoxb9

Hoxb3

Mir10a

Klf5

Hoxb1

Hoxb2

Hoxb6 Hoxb4

AK002860

20 kb

10 kb 46

59 NELFA

EZH2

1 26

1 31 SIN3A

RING1B

1 33

1 28 BLIMP1

BLIMP1 1_ Ccnd1

1 Wnt5a

Figure 7 BLIMP1 binds to targets of mESC self-renewal regulators and polycomb proteins. (a) Venn diagrams showing the total number of target sites overlapping by one base pair or more between BLIMP1 on one hand and the indicated transcription factors on the other. (b) Venn diagrams showing the total number of genes overlapping between BLIMP1 on one hand and

indicated transcription factors on the other hand. In a and b, the circles for the self-renewal cluster genes are indicated in green and the polycomb cluster genes in blue. (c) Track view of BLIMP1 binding on example genomic loci including the views for EZH2, RING1B and NELFA as well as SIN3A where appropriate.

genes were also co-bound by BLIMP1–AP2γ, and by PRDM14–AP2γ, and co-binding of all three factors showed enhanced enrichment with an increased degree of repression (Fig. 6a, left panel), more so than with either PRDM14 or BLIMP1 alone. In contrast, the induced genes were preferentially enriched for PRDM14 and AP2γ together (Fig. 6a,b right panels), where the enrichment of BLIMP1 alone or in combinations was statistically not significant (Fig. 6b). Although AP2γ clearly has a significant impact on gene expression, genes bound by AP2γ alone were depleted of association with specific functional gene categories (Supplementary Figs S5, S6, unless co-bound

by BLIMP1 and/or PRDM14. Hence, co-binding of BLIMP1 with the other factors mediates repression, whereas AP2γ and PRDM14 co-binding leads to gene induction, suggesting a high degree of co-regulation by these factors. Statistical testing of the overlap of binding sites for the three factors (P < 0.0001 permutation test; Fig. 6c and Supplementary Fig. S8a) and the overlap of genes bound by the three factors (P < 1 × 10−299 , χ 2 test; Fig. 6d and Supplementary Fig. S8b) further indicates an enrichment of co-bound genes over what would be expected by random chance, further supporting the functional relationship between the three factors.

912

ARTICLES a

GOF + BLIMP1, AP2γ, PRDM14

GOF + cytokines

ESCs

EpiLCs

+cytokines

–cytokines

–DOX

PGCLCs:

+DOX

PGCLCs: Day 2

Day 2 37.2%

44.9%

Day 4

Day 4 7.5%

47.4%

5.7%

38.0%

44.8%

6.3%

44.8%

Day 6

Prdm1 18 16 14 12 10 8 6 4 2 0

EpiLCs Day 2 Day 4

90 80 70 60 50 40 30 20 10 0

PGCLCs

Relative levels (normalized to Arbp1)

19.4%

1.2

Pou5f1

1.0 0.8 0.6 0.4 0.2 0

EpiLCs Day 2 Day 4

Kdm4b

2.0 1.5 1.0

EpiLCs Day 2 Day 4

Prdm14 500 450 400 350 300 250 200 150 100 50 0

PGCLCs

40 35 30 25 20 15 10 5 0

PGCLCs

2.5

Tcfap2c

Sox2

EpiLCs Day 2 Day 4

Dnmt3b

EpiLCs Day 2 Day 4

PGCLCs

180 160 140 120 100 80 60 40 20 0

PGCLCs

1.2

EpiLCs Day 2 Day 4

Dppa3 80 70 60 50 40 30 20 10 0

HoxB1

PGCLCs

4,500 4,000 3,500 3,000 2,500 2,000 1,500 1,000 500

EpiLCs Day 2 Day 4

Uhrf1

1.0

0.8

0.6

0.4

0.2

PGCLCs

EpiLCs Day 2 Day 4

25 20 15 10 5

PGCLCs

Figure 8 Co-expression of BLIMP1, AP2γ and PRDM14 induces PGC-like cell fate in vitro. (a) Bright-field and fluorescent microscopy images of the Oct4 reporter ESC line (GOF) in culture. The images show ESCs, EpiLCs (stage preceding PGCLC) and PGCLCs as indicated. The left panels show the induction of PGCLCs using cytokines, and the right panels show the induction of PGCLCs without cytokines using doxycycline-dependent

EpiLCs Day 2 Day 4

PGCLCs

Myc

1.2

0.5 EpiLCs Day 2 Day 4

PGCLCs

0.8

Nanog 35

EpiLCs Day 2 Day 4

1.0

EpiLCs Day 2 Day 4

PGCLCs

T-Brachyury

PGCLCs

1.2

Nanos3 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0

Cytokines BLIMP1 AP2γ PRDM14

EpiLCs Day 2 Day 4

PGCLCs

induction of BLIMP1, AP2γ and PRDM14 from GOF cells harbouring stable doxycycline-responsive constructs. The numbers on the figure panels indicate the ratio of fluorescent cells induced at each time point. (b) RT–qPCR analysis of sorted fluorescent PGCLCs on days 2 and 4 of either cytokine or doxycycline induction, as well as EpiLCs. a,b show a representative of two independent experiments. Scale bars, 100 µm.

913

ARTICLES BLIMP1, AP2γ and PRDM14 co-binding on repressed genes included developmental and signalling genes, particularly those of Fgf and Wnt signalling. BLIMP1 was preferentially bound to proliferation genes, and PRDM14 and AP2γ to cell-motility and cytoskeleton organization genes (Supplementary Figs S5a and S6a), indicating initiation of PGC migration. All three factors are over-represented on induced genes involved in actin cytoskeleton organization and intracellular signalling cascades (Supplementary Figs S5b,S6b), although PRDM14, either alone or with AP2γ, is predominant over BLIMP1. Critically, PRDM14 binds to PGC-specific genes, including Tcfap2c (encoding AP2γ), Sox2 and Kit as well as Dnd1, Nanos3, Dppa3, Prdm1 (encoding BLIMP1) and Prdm14 (Fig. 6e and Supplementary Fig. S6). Furthermore, AP2γ binds to Dppa3 as well as Nanos3, and acts cooperatively with PRDM14 in the induction of Nanos3 (Figs 1c, 6e and Supplementary Fig. S6). All three factors are involved in the induction of Dnd1 (Fig. 1c). Thus, after the induction of AP2γ by BLIMP1, it cooperates with PRDM14 to induce PGC genes. In the context of a compendium of transcription factors in mESCs (ref. 42), BLIMP1 clustered predominantly with self-renewal and polycomb factors, consistent with somatic gene repression (Fig. 7 and Supplementary Fig. S7). PRDM14 clustered with pluripotency transcription factors, but AP2γ associated weakly with factors in the compendium (Supplementary Fig. S7). This underlines the context-dependent combinatorial action of these three factors, which as expected is not fully captured in comparison with mESCs factors. A key role of the transcriptional network is in initiating epigenetic reprogramming in PGCs. Consistently, BLIMP1 and PRDM14 repress Kdm6b encoding a H3K27 demethylase, whereas PRDM14 mediates the induction of the H3K9 demethylase Kdm4b, and together with BLIMP1, induces Kdm3a (Supplementary Fig. S5a,b). This ensures the erasure of H3K9me2 that is a prerequisite for global DNA demethylation43 . The two factors also repress the de novo DNA methyltransferase Dnmt3b as well as Uhrf1, encoding an accessory protein for the maintenance DNA methyltransferase DNMT1; this facilitates global DNA demethylation in PGCs (Supplementary Fig. S6a). BLIMP1 and PRDM14 each bind to one of the two promoters to repress Uhrf1 in PGCs (Fig. 6e), whereas PRDM14 alone in mESCs is insufficient to do so44 . The regulation of histone and DNA methylases links PGC specification to the dynamic and genome-wide epigenetic reprogramming45 . BLIMP1 and PRDM14 bind extensively to differentially expressed genes during PGC specification, whereas AP2γ binds to a subset of them and perhaps acts as a facilitator of key events, except perhaps for chromatin modifications. The proposed transcriptional network is interdependent, because Tcfap2c (encoding AP2γ) expression is dependent on BLIMP1. Expression of Prdm14 and Prdm1 (encoding BLIMP1) expression is mutually interdependent as revealed by genetic experiments6,7 . PRDM14 also binds Tcfap2c and could enhance Tcfap2c expression after its induction by BLIMP1. This supports an obligatory functional relationship between BLIMP1, PRDM14 and AP2γ as critical regulators of PGC specification. Direct induction of PGC-like fate by BLIMP1, AP2γ and PRDM14 in vitro Next we investigated whether BLIMP1, AP2γ and PRDM14 are sufficient to directly induce PGC fate and used the in vitro induction of PGCLCs to test this premise28 . PGCLCs can be induced in response to cytokines (Fig. 8a, left panel), which we observed in 47.4–52.6%

914

of the cells after 4 days. Using the same reporter cells harbouring doxycycline-inducible constructs for the three transcription factors, we observed induction of PGCLCs in the absence of cytokines in ∼45–60% of the cells (Fig. 8a, right panel). Both the overall response and transcriptional analysis by qPCR using the sorted fluorescent PGCLCs were remarkably similar to those induced by cytokines with respect to the key PGC, epigenetic and somatic cell regulators. The response to the transcription factors was slightly enhanced (Fig. 8b and Supplementary Fig. S8c), as reflected in the higher induction of Dppa3 and Nanos3, and a more pronounced repression of HoxB1 and T-Brachyury. This observation establishes the principle that the proposed transcriptional network delineating specific and combined functions of BLIMP1, AP2γ and PRDM14 accounts for the necessary gene expression changes for PGC specification. Further work in the future will advance our knowledge of how these factors, both individually and in combination, induce PGC fate. DISCUSSION We present a comprehensive examination of the initiation of PGC specification by the combinatorial roles of BLIMP1, AP2γ and PRDM14, leading to a unique epigenetic program culminating in the epigenetic basal state45,46 . Co-expression of the three factors is by itself apparently sufficient to induce PGC-like fate in the absence of cytokines, and supports the proposed tripartite genetic network for PGC specification. BLIMP1, PRDM14 and AP2γ contribute to the repression of mesodermal genes in PGCs to set them apart from their somatic neighbours; until then, these cells are indistinguishable from each other. BLIMP1 has a dominant function in this respect, which is reinforced by PRDM14 and AP2γ. In contrast, PRDM14 and AP2γ together are associated with gene induction. Notably, Tcfapc2 (encoding AP2γ) is a direct target of BLIMP1 and induced by it in P19EC cells in vitro, and probably maintained thereafter by PRDM14 in PGCs. Tcfap2c fails to be induced in BLIMP1 mutant PGC-like cells in vivo and its induction by BLIMP1 is perhaps the vital link for the initiation of the PGC-specific genes. The high overlap of AP2γ targets with those of BLIMP1 and PRDM14 implies that it cooperates, augments or otherwise modulates the response of a subset of the targets. Furthermore, the distinct and predominant binding of BLIMP1 to promoter regions as opposed to the binding to gene-distal regulatory regions noted for PRDM14 (ref. 47) suggests parallel mechanisms in regulating transcription. The collaborative role of AP2γ is also reflected in its broad distribution that is centred on promoters, but potentially, encompasses distal elements. This comprehensive insight into PGC specification in mice may facilitate investigations on germ cells in other mammals, including humans, and enhance an understanding of context-dependent functions of transcription factors. For example, AP2γ has a role in trophectoderm differentiation in conjunction with CDX2 and EOMES, whereas it participates in the repression of Eomes in PGCs. BLIMP1 also drives cell fate commitment in several different lineages, whereas PRDM14 is crucial for pluripotency in ESCs47–49 . These differences are presumably linked to the molecular control of competence, which precedes and anticipates specific cell fate decisions. A fundamental property of early germ cells are the unique epigenetic changes that ensue following PGC specification, leading to global erasure of DNA methylation and acquisition of a basal epigenetic state.

ARTICLES The mechanism that regulates this unique resetting of the epigenome in germ cells could in principle be extended towards approaches for modifying epigenetic states of normal and diseased tissues. This study may help towards achieving wider objects of general interest in the field of regenerative medicine. METHODS Methods and any associated references are available in the online version of the paper. Note: Supplementary Information is available in the online version of the paper ACKNOWLEDGEMENTS We thank N. Miller for flow cytometry, and M. Trotter, C. Bradshaw and G. Allen for bioinformatics analysis. We thank S. Kim for help with figures. We thank R. Sengupta and J. Hackett for critically reading the manuscript. This work was supported by grants from the Wellcome Trust and HFSP to M.A.S. and the ERC to E.M. AUTHOR CONTRIBUTIONS E.M. and M.A.S. conceived of and designed the study; E.M., K.M., U.G., F.T. and S.B. performed the experiments; E.M., S.D. and E.D. performed computational analysis; K.L. performed RNA sequencing; B.G., E.M. and M.A.S carried out critical assessment of the data; M.A.S. and E.M. wrote the manuscript. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. Published online at www.nature.com/doifinder/10.1038/ncb2798 Reprints and permissions information is available online at www.nature.com/reprints 1. 2.

3. 4. 5. 6. 7.

9. 10.

11.

12.

13. 14. 15. 16. 17.

Fuhrmann, G. et al. Mouse germline restriction of Oct4 expression by germ cell nuclear factor. Dev. Cell 1, 377–387 (2001). Seki, Y. et al. Extensive and orderly reprogramming of genome-wide chromatin modifications associated with specification and early development of germ cells in mice. Dev. Biol. 278, 440–458 (2005). Borgel, J. et al. Targets and dynamics of promoter DNA methylation during early mouse development. Nat. Genet. 42, 1093–1100 (2010). McLaren, A. & Lawson, K. A. How is the mouse germ-cell lineage established? Differentiation 73, 435–437 (2005). Saitou, M., Barton, S. C. & Surani, M. A. A molecular programme for the specification of germ cell fate in mice. Nature 418, 293–300 (2002). Yamaji, M. et al. Critical function of Prdm14 for the establishment of the germ cell lineage in mice. Nat. Genet. 40, 1016–1022 (2008). Kurimoto, K. et al. Complex genome-wide transcription dynamics orchestrated by Blimp1 for the specification of the germ cell lineage in mice. Genes Dev. 22, 1617–1635 (2008). Robertson, E. J. et al. Blimp1 regulates development of the posterior forelimb, caudal pharyngeal arches, heart and sensory vibrissae in mice. Development 134, 4335–4345 (2007). Ohinata, Y. et al. Blimp1 is a critical determinant of the germ cell lineage in mice. Nature 436, 207–213 (2005). Vincent, S. D. et al. The zinc finger transcriptional repressor Blimp1/Prdm1 is dispensable for early axis formation but is required for specification of primordial germ cells in the mouse. Development 132, 1315–1325 (2005). Saitou, M., Payer, B., O’Carroll, D., Ohinata, Y. & Surani, M. A. Blimp1 and the emergence of the germ line during development in the mouse. Cell Cycle 4, 1736–1740 (2005). Turner, C. A., Mack, D. H. & Davis, M. M. Blimp-1, a novel zinc finger-containing protein that can drive the maturation of B lymphocytes into immunoglobulinsecreting cells. Cell 77, 297–306 (1994). Keller, A. D. & Maniatis, T. Identification and characterization of a novel repressor of beta-interferon gene expression. Genes Dev. 5, 868–879 (1991). Morgan, M. a J. et al. Blimp-1/Prdm1 alternative promoter usage during mouse development and plasma cell differentiation. Mol. Cell. Biol. 29, 5813–5827 (2009). Martins, G. A. et al. Transcriptional repressor Blimp-1 regulates T cell homeostasis and function. Nat. Immunol. 7, 457–465 (2006). Kallies, A. et al. Transcriptional repressor Blimp-1 is essential for T cell homeostasis and self-tolerance. Nat. Immunol. 7, 466–474 (2006). Chan, Y-H. et al. Absence of the transcriptional repressor Blimp-1 in hematopoietic lineages reveals its role in dendritic cell homeostatic development and function. J. Immunol. 183, 7039–7046 (2009).

18. Nishikawa, K. et al. Blimp1-mediated repression of negative regulators is required for osteoclast differentiation. Proc. Natl Acad. Sci. USA 107, 3117–3122 (2010). 19. Chang, D. H., Angelin-Duclos, C. & Calame, K. BLIMP-1: Trigger for differentiation of myeloid lineage. Nat. Immunol. 1, 169–176 (2000). 20. Kim, S. J., Zou, Y. R., Goldstein, J., Reizis, B. & Diamond, B. Tolerogenic function of Blimp-1 in dendritic cells. J. Exp. Med. 208, 2193–2199 (2011). 21. Yu, J., Angelin-Duclos, C., Greenwood, J., Liao, J. & Calame, K. Transcriptional repression by Blimp-1 (PRDI-BF1) involves recruitment of histone deacetylase. Mol. Cell. Biol. 20, 2592–2603 (2000). 22. Ren, B., Chee, K. J., Kim, T. H. & Maniatis, T. PRDI-BF1/Blimp-1 repression is mediated by corepressors of the Groucho family of proteins. Genes Dev. 13, 125–137 (1999). 23. Gyory, I., Wu, J., Fejér, G., Seto, E. & Wright, K. L. PRDI-BF1 recruits the histone H3 methyltransferase G9a in transcriptional silencing. Nat. Immunol. 5, 299–308 (2004). 24. Ancelin, K. et al. Blimp1 associates with Prmt5 and directs histone arginine methylation in mouse germ cells. Nat. Cell Biol. 8, 623–630 (2006). 25. Su, S-T. et al. Involvement of histone demethylase LSD1 in Blimp-1-mediated gene repression during plasma cell differentiation. Mol. Cell. Biol. 29, 1421–1431 (2009). 26. Cretney, E. et al. The transcription factors Blimp-1 and IRF4 jointly control the differentiation and function of effector regulatory T cells. Nat. Immunol. 12, 304–311 (2011). 27. Weber, S. et al. Critical function of AP-2 gamma/TCFAP2C in mouse embryonic germ cell maintenance. Biol. Reprod. 82, 214–223 (2010). 28. Hackett, J. a, Zylicz, J. J. & Surani, M. A. Parallel mechanisms of epigenetic reprogramming in the germline. Trends Genet. 28, 1–11 (2012). 29. Mcburney, M. Isolation of male embryonal carcinoma cells and their chromosome replication patterns*1. Dev. Biol. 89, 503–508 (1982). 30. Yeom, Y. I. et al. Germline regulatory element of Oct-4 specific for the totipotent cycle of embryonal cells. Development 122, 881–894 (1996). 31. Chambers, I. et al. Functional expression cloning of nanog, a pluripotency sustaining factor in embryonic stem cells. Cell 113, 643–655 (2003). 32. Yabuta, Y., Kurimoto, K., Ohinata, Y., Seki, Y. & Saitou, M. Gene expression dynamics during germline specification in mice identified by quantitative single-cell gene expression profiling. Biol. Reprod. 75, 705–716 (2006). 33. Mitsui, K. et al. The homeoprotein Nanog is required for maintenance of pluripotency in mouse epiblast and ES cells. Cell 113, 631–642 (2003). 34. Ma, Z., Swigut, T., Valouev, A., Rada-Iglesias, A. & Wysocka, J. Sequence-specific regulator Prdm14 safeguards mouse ESCs from entering extraembryonic endoderm fates. Nat. Struct. Mol. Biol. 18, 120–127 (2011). 35. Kuo, T. C. & Calame, K. L. B lymphocyte-induced maturation protein (Blimp)-1, IFN regulatory factor (IRF)-1, and IRF-2 can bind to the same regulatory sites. J. Immunol. 173, 5556–5563 (2004). 36. Hayashi, K., Ohta, H., Kurimoto, K., Aramaki, S. & Saitou, M. Reconstitution of the mouse germ cell specification pathway in culture by pluripotent stem cells. Cell 146, 519–532 (2011). 37. Woodfield, G. W., Chen, Y., Bair, T. B., Domann, F. E. & Weigel, R. J. Identification of primary gene targets of TFAP2C in hormone responsive breast carcinoma cells. Genes Chromosomes Cancer 962, 948–962 (2010). 38. Tan, S. K. et al. AP-2γ regulates oestrogen receptor-mediated long-range chromatin interaction and gene transcription. EMBO J. 30, 2569–2581 (2011). 39. Kidder, B. L. & Palmer, S. Examination of transcriptional networks reveals an important role for TCFAP2C, SMARCA4, and EOMES in trophoblast stem cell maintenance. Genome Res. 20, 458–472 (2010). 40. Loh, Y. H. et al. The Oct4 and Nanog transcription network regulates pluripotency in mouse embryonic stem cells. Nat. Genet. 38, 431–440 (2006). 41. Chen, X. et al. Integration of external signaling pathways with the core transcriptional network in embryonic stem cells. Cell 133, 1106–1117 (2008). 42. Martello, G et al. ESRRB is a pivotal target of the GSK3/Tcf3 axis regulating embryonic stem cell self-renewal. Cell Stem Cell 11, 491–504 (2012). 43. Nakamura, T. et al. PGC7 binds histone H3K9me2 to protect against conversion of 5mC to 5hmC in early embryos. Nature 486, 415–419 (2012). 44. Grabole, Nils et al. Prdm14 promotes germline fate and naïve pluripotency by modulating signalling and the epigenome. EMBO Rep. 14, 1–9 (2013). 45. Hajkova, P. et al. Epigenetic reprogramming in mouse primordial germ cells. Mech. Dev. 117, 15–23 (2002). 46. Hajkova, P. et al. Chromatin dynamics during epigenetic reprogramming in the mouse germ line. Nature 452, 877–881 (2008). 47. Ma, Z., Swigut, T., Valouev, A., Rada-Iglesias, A. & Wysocka, J. Sequence-specific regulator Prdm14 safeguards mouse ESCs from entering extraembryonic endoderm fates. Nat. Struct. Mol. Biol. 18 (2011) 120–117. 48. Kuckenberg, P. et al. The transcription factor TCFAP2C/AP-2gamma cooperates with CDX2 to maintain trophectoderm formation. Mol. Cell. Biol. 30, 3310–3320 (2010). 49. Chia, N-Y. et al. A genome-wide RNAi screen reveals determinants of human embryonic stem cell identity. Nature 468, 316–320 (2010).

915

METHODS

DOI: 10.1038/ncb2798

METHODS Plasmids. The cDNA encoding BLIMP1 was cloned into the multiple cloning site of pTRE-Tight from Promega. The tetracycline transactivator tTA cDNA, a kind gift from S. Jackson (Gurdon Institute, University of Cambridge, UK), was cloned into pCAC-IP. The PRDM14 and AP2γ expression constructs were described previously50 . For inducible expression of BLIMP1, AP2γ and PRDM14, the cDNAs were cloned into a piggy-bac vector downstream of a tetracycline response element (TRE). The vectors were co-transfected with a piggy-bac vector containing the reverse tetracycline transactivator rtTA, as well as a vector transiently expressing piggy-bac transposase.

Cells. The mouse P19ECs were obtained from ATCC and were cultured in DMEM, 10% fetal calf serum, 2 mM l-glutamine (Gibco) and 50 µg ml−1 penecillin/streptomycin (Gibco). P19ECs stably expressing PRDM14 and AP2γ were generated as previously described1 . Lipofectamine 2000 (Invitrogen) was used for transfections. PGCLCs were cultured as previously described36 , from Oct4 Distal Enhancer EGFP reporter ESCs (GOF), and collected on day 3 after PGCLC induction for ChIP, and on days 2 and 4 for RT–qPCR after flow cytometric sorting for EGFP.

ChIP-seq. Nuclear extracts from P19ECs co-transfected with TRE-Tight BLIMP1–EGFP, and pCAG-tTA-IP for BLIMP1, were subjected to ChIP, using previously described methods51 . A mouse anti-GFP (3E6) from Invitrogen was used (A11120). The same protocol was used for AP2γ, with a rabbit polyclonal from Santa Cruz (H-77). A total of 10 µg of antibody was used per immunoprecipitation for both antibodies above. Library preparation of the BLIMP1 ChIP was performed using Illumina ChIP-seq DNA Sample Preparation Kit (cat. No 1003473). The AP2γ ChIP-seq libraries were generated using the Nugene Ovation Ultralow Library Kit. Sequencing was performed on an Illumina Genome Analyser for the BLIMP1 ChIP, and an Illumina HiSeq 2000 sequencer for AP2γ, at the Cambridge Research Institute Genomics Facility, generating single end 36 bp reads. The data were deposited in the Array Express repository http://www.ebi.ac.uk/arrayexpress. For validation of BLIMP1 ChIP-seq results in PGCLCs, 5 × 106 cells from unsorted cultures with approximately 60% induction of PGCLCs were used for ChIP with a polyclonal antiserum against BLIMP1 (CLU267; ref. 35) or a corresponding pre-immune serum; the sera were used at a 1:100 dilution. The resulting immunoprecipitated DNA was processed as described before52 and qPCR was performed for target regions.

ChIP-seq analysis. Sequences corresponding to adaptors were removed from the sequence library, and the sequence tags subsequently mapped to the mouse genome (NCBI Mouse v37/mm9) using BWA (ref. 53). The BWA outputs were filtered for only unique hits and no more than 33 of the same base per read. Sequences mapping to the BLIMP1 ORF (chr 10:44, 156, 983–44, 178, 554) were removed from the BLIMP1 ChIP-seq data. Peaks were called for the BLIMP1 ChIP-seq with CisGenome54 , at FDR ≤ 0.001 and MACS (ref. 55) using a P-value cutoff of 1×10−9 . Peak regions for each individual experiment were then intersected using BED tools56 to only include peak regions identified by both algorithms. Subsequently the peak regions for the two experiments were intersected again to include only reproducible sites of binding in the final list of high-confidence peaks. For the AP2γ ChIP-seq, peaks were called using MACS using a P-value cutoff of 1 × 10−250 , tag count of 50 tags and fold change of ×20 to generate a list of high-confidence peaks. For assessing the distance of peaks to gene TSSs the ChIPpeakAnno package from Bioconductor was employed57 . Peaks were assigned to genes using a cutoff distance of 10 kb upstream and 2 kb downstream of genes, for functional category enrichment analysis of genes associated with peaks and the gene list analysed with Panther58 . Validation of new high-confidence peak regions was done in duplicate experiments using either EGFP- and BLIMP1–EGFP-transfected cells or triplicate experiments with AP2γ stables for subsequent ChIP–qPCR analysis. For the validation ChIP experiments, 10 µg total of the antibodies listed above was used.

Sequence motif analysis. De novo motif discovery was performed using MEME-ChIP (http://meme.sdsc.edu/meme/cgi-bin/meme-chip.cgi; ref. 59). For TRANSFAC motif scanning, we used the web-based tool Centdist57 . The mouse genome was scanned for the AP2γ motif using the HOMER (http://biowhat.ucsd. edu/homer/) software package’s ScanMotifGenomeWide.pl function with default parameters61 .

Microarray analysis. At 24 and 48 h post transfection, cells were dissociated from the culture plates and subjected to flow cytometric analysis and sorting using a MoFlo high-speed cell sorter (Dako Cytomation). At 24 h, 2.5–5 × 105 fluorescent cells were collected from each of 2 gates designated as high (HI) and low (LO). At 48 h, cells were collected from as inglegate of cells. The cells were suspended in Trizol (Invitrogen), total RNA isolated and cleaned up with Qiagen RNeasy microkit. The samples were analysed on a Bioanalyzer (Agilent Technologies)

before amplification and cRNA generation for hybridization onto Illumina Mouse WG-6v2.0 Expression BeadChip microarrays. cRNA generation, quality control and hybridization were performed by Cambridge Genomic Services at the University of Cambridge. The experiment and subsequent hybridizations were performed in triplicate, and microarray data were deposited in the ArrayExpress repository. Statistical analysis was performed as previously described50 , with the exception that differential expression between groups of two or more profiles was assessed using the output of a moderated t -test for each probe set. Differential expression was deemed significant at an FDR of 0.005. ChIP-bound regions, identified by ChIP-seq experiments, were assigned to microarray probe sets as follows. Transcripts and their genomic coordinates were related to probe sets using the Bioconductor:biomaRt interface to the Ensembl60 database (http://www.ensembl.org/). The Bioconductor:IRanges package was employed to identify ChIP-bound regions that occurred within −10 kb upstream and +2 kb downstream of each transcript. The over-representation of functional categories of differentially expressed genes was assessed using the online tool DAVID (ref. 62).

qPCR. qPCR reactions were performed with SybrGreen JumpStart Taq ReadyMix (Sigma) using 0.2 µM of gene specific forward and reverse primers (see Table S7 for primer sequences). In the case of cDNA, 5 ng of template was used per reaction. Quantification was performed as previously described50 . Mean threshold cycles were determined from two technical replicates unless otherwise stated using the comparative Ct method. RT–qPCR expression levels were normalized to Ppia. RNA-seq. E6.5 and E7.5 PGCs were originally selected as BLIMP1–EGFP-positive cells, the E7.5 somatic cells were BLIMP1–EGFP negative and the E8.5 PGCs expressed Oct4–EGFP. The BLIMP1−/− PGCs were picked at random from dissected embryos for downstream selection of PGC-like cells. Two cells were selected from each time point according to the expression of somatic or PGC markers. The PCGs selected were positive for the expression of the lineage markers Prdm1 (apart from Prdm1 mutant PGCs) and Dppa3, but negative for the early mesodermal marker HoxB1. The E7.5 somatic cells selected were negative for the above PGC lineage markers but HoxB1 positive. The cDNA of the selected cells was amplified and sequenced using high-throughput sequencing (RNA-seq) as previously published63 . Reads were aligned to the mouse genome (mm 9) using the Bowtiesoftware, and all reads that mapped at more than 10 genomic locations allowing one mismatch were discarded. Reads were mapped to the extracted genomic sequences allowing one mismatch, and subsequently to known splice junctions (±42 nt) allowing no more than three mismatches. Read counts were normalized using the TMM method and differential expression analysis was performed using the edgeR Bioconductor package. The data were deposited in the ArrayExpress repository (http://www.ebi.ac.uk/arrayexpress). For hierarchical clustering of the expression profiles, 6,838 genes with higher variability in expression (interquantile range >2.5) were selected, and time points clustered on the basis of the Euclidean distance of normalized expression values between genes using complete linkage clustering with Cluster3 software64 http://bonsai.hgc.jp/∼mdehoon/software/cluster/software.htm). The correlation of different expression profile comparisons, between the RNA-seq time points as well as microarray analysis was performed by selecting the 10% most highly induced or repressed genes and calculating the pair-wise Pearson correlation coefficients.

Transcription factor association scores and enrichment analysis. ChIP-seq data sets were aligned using Bowtie and peaks were called with MACS. Transcription factor association scores were calculated on the basis of the MACS-generated coverage profiles per genomic coordinates. For each gene i the transcription factor association scores of BLIMP1 were calculated as described previously65 ; sum (gk ) ∗ e(−d/d0), where gk is the RPM-normalized intensity; that is, the number of reads aligned to the coordinate k per million reads in the ChIP-seq data set, dk is the distance (number of nucleotides) between the TSS of gene i and the coordinate k, and dk is a constant (500 bp). These sums were generated up to ±100 kb. When calculating the transcription factor target enrichment on sets of differentially expressed genes the enrichment of high-confidence peaks, the number N1 of BLIMP1, PRDM14 or AP2γ peaks in the set of N differentially expressed genes (log2 FC <X ; log2 (FC) > X ), was compared with the average number of peaks in the whole data set for each factor. Target enrichment is given as log2[(N1 /N )/(B/A)], where B is the total number of peaks and A is the total number or genes in the RNA-seq expression data set. In analogy, the average BLIMP1 TF score in the set of N differentially expressed genes was compared to the average total BLIMP1 TF score. For the P19EC intersection enrichment the set of N differentially expressed genes in E7.5 PGCs versus E7.5 Soma (log2FC < X ,log2(FC) > X ; P value <0.01) was intersected by genes differentially expressed (logFC < −0.3 (10% quantile), logFC > 0.38 (90% quantile); P value <0.01), after BLIMP1 transfection to P19ECs.

METHODS

DOI: 10.1038/ncb2798 Transcription network analysis. Genes repressed and induced in E7.5 PGCs versus E7.5 soma were classified into categories based on whether they were bound by BLIMP1, PRDM14 or both, and whether there were AP2γ motifs present in the binding regions or whether they were bound by AP2γ. Subsequently, GO-analysis of genes in each category was performed and the GO-term profiles, consisting of the gene count for each term with 4 genes or more in at least one of the binding categories, were clustered using Cluster 3. We used Cytoscape to construct network binding profiles of the induced and repressed genes belonging to over-represented GO-term categories66 . To assess the statistical significance of overlap between genomic binding sites of the three factors, 10,000 random permutation trials were performed where in each trial, BLIMP1, AP2γ and PRDM14 peak regions were randomly re-assigned and complete peak regions overlapping by 1 bp or more between pairs were counted. Accession numbers. The primary accession numbers are E-MTAB-1600 for the AP2γ ChIP-seq data, E-MTAB-1599 for the BLIMP1 ChIP-seq data, E-MTAB-865 for the microarray data and E-MTAB-1178 for the RNA-seq data. The referenced accession numbers are GSE25409 for the PRDM14 ChIP-seq, GSE24841 for the SIN3a ChIP-seq, GSE20530 for the NELFA ChIP-seq, GSE26680 for the RING1B ChIP-seq and GSE13084 for EZH2 ChIP-seq. 50. Gillich, A. et al. Epiblast stem cell-based system reveals reprogramming synergy of germline factors. Cell Stem. Cell 10, 425–439 (2012). 51. Magnúsdóttir, E. et al. Epidermal terminal differentiation depends on B lymphocyte-induced maturation protein-1. Proc. Natl Acad. Sci. USA 104, 14988–14993 (2007). 52. Ng, J-H. et al. In vivo epigenomic profiling of germ cells reveals germ cell molecular signatures. Deve. Cell 24, 324–333 (2013). 53. Li, H. & Durbin, R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25, 1754–1760 (2009).

54. Ji, H. et al. An integrated software system for analyzing ChIP-chip and ChIP-seq data. Nat. Biotech. 26, 1293–1300 (2008). 55. Zhang, Y. et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137 (2008). 56. Quinlan, A. R. & Hall, I. M. BEDTools: A flexible suite of utilities for comparing genomic features. Bioinformatics 26, 841–842 (2010). 57. Zhu, L. J. et al. ChIPpeakAnno: A Bioconductor package to annotate ChIP-seq and ChIP-chip data. BMC Bioinformatics 11, 237 (2010). 58. Thomas, P. D. PANTHER: A browsable database of gene products organized by biological function, using curated protein family and subfamily classification. Nuclic Acids Res. 31, 334–341 (2003). 59. Machanick, P. & Bailey, T. L. MEME-ChIP: Motif analysis of large DNA datasets. Bioinformatics 27, 1696–1697 (2011). 60. Zhang, Z., Chang, C. W., Goh, W. L., Sung, W-K. & Cheung, E. CENTDIST: Discovery of co-associated factors by motif distribution. Nucli. Acids Res. 39, W391–W399 (2011). 61. Heinz, S. et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol. Cell 38, 576–589 (2010). 62. Huang, D. W., Sherman, B. T. & Lempicki, R. a Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat. Protocols 4, 44–57 (2009). 63. Tang, F. et al. Tracing the derivation of embryonic stem cells from the inner cell mass by single-cell RNA-Seq analysis. Cell Stem Cell 6, 468–478 (2010). 64. De Hoon, M. J. L., Imoto, S., Nolan, J. & Miyano, S. Open source clustering software. Bioinformatics 20, 1453–1454 (2004). 65. Ouyang, Z., Zhou, Q. & Wong, W. H. ChIP-Seq of transcription factors predicts absolute and differential gene expression in embryonic stem cells. Proc. Natl Acad. Sci. USA 106, 21521–21526 (2009). 66. Shannon, P. et al. Cytoscape: A software environment for integrated models of biomolecular interaction networks. Genome Res. 13, 2498–2504 (2003).

S U P P L E M E N TA R Y I N F O R M AT I O N DOI: 10.1038/ncb2798

Figure-S1 (Surani) a

24h 100

EGFP BLIMP1-EGFP

100

0 0

BLIMP1-EGFP

EGFP 24h LO

48h

24h LO

48h Eras En2 Tcfap2a Wnt8a En1 Eomes Hoxa5 Hopx Chek1 Taf7 Etv1 Chek2 Polr1b Myc Wnt5b Nanog Dnmt3b Fgf5 Hoxa2 Jmjd5 Chd1 Setdb2 Evx1 Gli2 Jmjd1a Rhox6 Ifitm1 Dnmt3l Col18a1 Gadd45b Kit Olig1 Nanos3 Krt18 Krt8 Wnt6 Snai3 Gata6 Spo11 Comp Igf2 Gata2 Rhox9 Prdm1

Positive

48h

0 10

Induced Genes 0

-log(p-value) 2

Extracellular matrix Extracellular region part Proteinaceous extracellular matrix Extracellular region Positive regulation of development Calcium ion binding Regulation of cell motion Neuron development Cell motion Positive regulation of cell differentiation Cell projection organization Neuron differentiation Regulation of neurogenesis Exocrine system development Regulation of nervous system development

Repressed

Repressed Genes Transcription factor activity Pattern specification process Transcription regulator activity Embryonic morphogenesis Sequence specific DNA binding Embryonic organ morphogenesis Embryonic organ development DNA dependent regulation of transcription Regulation of RNA metabolic process Neuron differentiation DNA binding Chordate embryonic development Embryonic development ending in birth*tching tRNA metabolic process Transcription factor complex

−1

-log(p-value) 2

Induced

Row Z−Score 6

Figure S1 Expression profiling of P19ECs with ectopic expression of BLIMP1. (a). Flow cytometric plot showing the fluorescence intensity of EGFP and BLIMP1-EGFP transfected cells. Gates employed for cell sorting are indicated. (b) Heat map showing differentially regulated genes

in P19ECs upon BLIMP1 expression. Each column represents a timepoint assayed in triplicate. The colours indicate the z-score for differential expression. (c). Gene ontology analysis of the top 10% of genes repressed and induced upon BLIMP1 expression in P19ECs.

WWW.NATURE.COM/NATURECELLBIOLOGY

S U P P L E M E N TA R Y I N F O R M AT I O N

Figure-S2 (Surani) a

24 LO Down 0

24 0LO UP1

transcrip�on regula�on GO:0003700~transcrip�on factor ac�vity GO:0031981~nuclear lumen GO:0045941~posi�ve regula�on of … GO:0060429~epithelium development GO:0016331~morphogenesis of … GO:0009615~response to virus gtp-binding GO:0016481~nega�ve regula�on of …

24 HI Down 0

GO:0031981~nuclear lumen transcrip�on regula�on zinc GO:0016481~nega�ve regula�on of transcrip�on GO:0030324~lung development GO:0045944~posi�ve regula�on of transcrip�on… GO:0046872~metal ion binding GO:0030326~embryonic limb morphogenesis GO:0042054~histone methyltransferase ac�vity GO:0001707~mesoderm forma�on GO:0048864~stem cell development GO:0001944~vasculature development atp-binding GO:0004386~helicase ac�vity GO:0008138~protein tyrosine/serine/threonine… Aminoacyl-tRNA synthetase mitosis GO:0019866~organelle inner membrane GO:0002520~immune system development GO:0009982~pseudouridine synthase ac�vity GO:0055123~diges�ve system development GO:0006839~mitochondrial transport GO:0048667~cell morphogenesis involved in… GO:0014020~primary neural tube forma�on ubl conjuga�on pathway GO:0016310~phosphoryla�on GO:0034062~RNA polymerase ac�vity GO:0004725~protein tyrosine phosphatase… GO:0060324~face development GO:0046131~pyrimidine ribonucleoside… GO:0006306~DNA methyla�on GO:0016477~cell migra�on

48 Down 0

24 HI UP 0

GO:0046872~metal ion binding GO:0005773~vacuole GO:0043068~posi�ve regula�on of programmed… GO:0016310~phosphoryla�on serine/threonine-protein kinase GO:0043067~regula�on of programmed cell death GO:0051085~chaperone mediated protein… GO:0030336~nega�ve regula�on of cell migra�on GO:0005624~membrane frac�on GO:0005765~lysosomal membrane GO:0005261~ca�on channel ac�vity GO:0031410~cytoplasmic vesicle GO:0051270~regula�on of cell mo�on GO:0031669~cellular response to nutrient levels nucleo�de-binding GO:0006897~endocytosis GO:0045860~posi�ve regula�on of protein… GO:0048002~an�gen processing and… GO:0050801~ion homeostasis GO:0043492~ATPase ac�vity, coupled to… GO:0060113~inner ear receptor cell diﬀeren�a�on GO:0031228~intrinsic to Golgi membrane GO:0006914~autophagy gtp-binding GO:0001568~blood vessel development GO:0042325~regula�on of phosphoryla�on GO:0004713~protein tyrosine kinase ac�vity GO:0022843~voltage-gated ca�on channel ac�vity GO:0051339~regula�on of lyase ac�vity GO:0007276~gamete genera�on GO:0070663~regula�on of leukocyte prolifera�on GO:0048640~nega�ve regula�on of… GO:0006635~fa�y acid beta-oxida�on

48h UP 0

lysosome extracellular matrix GO:0031982~vesicle GO:0001568~blood vessel development GO:0001942~hair follicle development GO:0000904~cell morphogenesis … GO:0051270~regula�on of cell mo�on GO:0006643~membrane lipid metabolic … GO:0060284~regula�on of cell … GO:0042325~regula�on of … GO:0048667~cell morphogenesis … GO:0016477~cell migra�on GO:0001501~skeletal system development GO:0050770~regula�on of axonogenesis GO:0043433~nega�ve regula�on of … GO:0007276~gamete genera�on GO:0007272~ensheathment of neurons GO:0008406~gonad development GO:0014031~mesenchymal cell… GO:0050804~regula�on of synap�c … GO:0016310~phosphoryla�on GO:0050771~nega�ve regula�on of … GO:0007276~gamete genera�on GO:0001701~in utero embryonic …

nuclear lumen tRNA processing ribosome biogenesis nucleo�de-binding amino acid biosynthesis mitochondrial envelope RNA methyltransferase ac�vity atp-binding dna-directed rna polymerase GO:0022411~cellular component … GO:0000278~mito�c cell cycle gtp-binding GO:0006220~pyrimidine nucleo�de … GO:0007067~mitosis GO:0055029~nuclear DNA-directed … GO:0030097~hemopoiesis GO:0016481~nega�ve regula�on of …

Figure S2 Functional categories of differentially expressed genes upon BLIMP1 expression in P19ECs. (a-f). Gene ontology analysis of all genes with significant

GO:0051270~regula�on of cell mo�on GO:0031012~extracellular matrix GO:0051960~regula�on of nervous system… GO:0001942~hair follicle development GO:0042325~regula�on of phosphoryla�on GO:0007272~ensheathment of neurons GO:0050770~regula�on of axonogenesis GO:0043392~nega�ve regula�on of DNA binding GO:0016477~cell migra�on GO:0008219~cell death nucleo�de phosphate-binding region:GTP GO:0030335~posi�ve regula�on of cell migra�on GO:0050768~nega�ve regula�on of neurogenesis GO:0050801~ion homeostasis GO:0001501~skeletal system development GO:0043068~posi�ve regula�on of programmed… GO:0045860~posi�ve regula�on of protein… GO:0001701~in utero embryonic development calcium transport GO:0005765~lysosomal membrane GO:0048812~neuron projec�on morphogenesis GO:0006811~ion transport GO:0000904~cell morphogenesis involved in… GO:0035272~exocrine system development

expression changes (FDR < 0.005) in P19ECs upon BLIMP1 expression for both induced and repressed genes for each of the 3 comparisons performed.

S U P P L E M E N TA R Y I N F O R M AT I O N

Figure-S3 (Surani)

Repressed Genes log2 (Fold Enrichment)

log2 (Fold Enrichment) 0

All BLIMP1 targets

Middle ear morphogenesis Motor axon guidance Axon guidance Cell morphogenesis involved in differentiation Axonogenesis Neuron development Neuron projection morphogenesis Cell morphogenesis involved in neuron differentiation Neuron differentiation Cell morphogenesis Cell projection organization Neuron projection development GTP binding Cellular component morphogenesis Guanyl nucleotide bindingg Guanyl ribonucleotide binding Transcription factor activity Transcription regulator activity Regulation of transcription, DNA-dependent Regulation of RNA metabolic process DNA binding Transcription Regulation of transcription

E8.5 PGCs vs BLIMP1 KO

Blimp1 target enrichment (log ) 2 −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3

Embryonic skeletal system development Odontogenesis of dentine-containing tooth Negative regulation of canonical Wnt receptor signaling pathway Embryonic skeletal system morphogenesis Anterior/posterior pattern formation Regulation of Rho protein signal transduction Canonical Wnt receptor signaling pathway Positive regulation of cell migration Positive regulation of gene expression Vasculogenesis Homophilic cell adhesion Axon guidance Positive regulation of protein phosphorylation Heart development Wnt receptor signaling pathway Endocytosis Post-embryonic development In utero embryonic development Kidney development Organ morphogenesis

All BLIMP1 targets

peaks scores

Repressed

Induced

−3 −2 −1 0 1 2 3

log2 (Fold Change)

E7.5 PGCs vs E7.5 soma

peaks scores intersect.

0.0 0.2 0.4 0.6 0.8

−0.2

Induced Genes

BLIMP1 target enrichment (log )

−4

−2

log 2(Fold Change)

50 kb

30 -

Meis2

Wnt5b

Fbxl14

20 kb

28 -

Has2as

50 kb

Spata13

AK019124

Figure S3 RNAseq analysis during PGC specification; integrative analysis of PGC transcriptome, BLIMP1 induced changes in P19EC profiles, and BLIMP1targets. (a and b). Gene ontology analysis showing functional categories of genes repressed and induced respectively, between E7.5 PGCs and somatic cells and the genes from the comparison that are bound by BLIMP1. (c). Relative enrichment of BLIMP1 binding regions and the scores associated with genes differentially expressed between E8.5 PGCs and E7.5 Prdm1 (encoding BLIMP1)-KO PGC–like cells. (d). Correlation analysis of differentially expressed genes during PGC specification and upon BLIMP1 expression in P19ECs. The Pearson correlation coefficients are indicated in the bottom half for each pair-wise comparison and each point on the plot indicates the differential expression of a gene in the comparisons indicated on the x- and y-axes respectively by the juxtaposition to the squares along

Klf9

Mir1192

50 kb

22 1_

AK012325 AK144485 2810405F15Ri k

10 kb

30 -

Has2

30 -

100 kb

30 -

Etv1

the diagonal. (e). Relative enrichment of BLIMP1 binding regions associated with genes that are differentially expressed both upon BLIMP1 expression in P19ECs as well as during PGC specification. The x-axis indicates the log2 (fold change) and the y-axis indicates the log2 of the BLIMP1 target enrichment at each fold change-interval of differentially expressed genes over the average target frequency of the whole expression data set. Peaks: the enrichment of peaks associated with genes in each interval differential expression expression level interval Scores; the enrichment of binding scores calculated for genes in each interval. Intersect: The enrichment of peaks associated with genes differentially expressed in both comparisons, in each interval of differential expression. (f). ChIP sequencing tracks from the UCSC browser of genes showing example genomic loci of genes bound by BLIMP1 and repressed in both PGCs and P19ECs upon BLIMP1 expression.

WWW.NATURE.COM/NATURECELLBIOLOGY

S U P P L E M E N TA R Y I N F O R M AT I O N

Figure-S4 (Surani) a 0

BLIMP1

44.02

IRF2

27.00

2.40E-10

SP1-Q6

23.94

2.06E-8

AP2

16.72

9.17E-5

NRF1

16.17

6.57E-5

Motif

81.05

PRDM14

80.67

SP1 (CACD)

56.81

AP2

56.59

SOX

45.25

AP2

AP2g

Distance From Peak Centre

Score p-Value

OCT4

BLIMP1

PRDM14

AP2

AP2g

Frequency per 50 bp

Score p-Value

Frequency per 50 bp

Motif

Distance From Peak Centre

0.8 0.2 0.0 −0.2

−0.2

0.0

0.2

0.4

0.6

Target enrichment (log )

0.8

−4

−2

log2 (FC) BLIMP1 BLIMP1/AP2g BLIMP1/AP2g/PRDM14

−4

−2

log2 (FC)

PRDM14 PRDM14/AP2g

Figure S4 AP2γ motif analysis on BLIMP1 and PRDM14 binding regions. (a). TRANSFAC motif scanning of the BLIMP1 binding regions. The enrichment scores and p-values for the enrichment of each motif are indicated. (b). The distribution of the BLIMP1, AP2 alpha- and the AP2γ motifs on the BLIMP1 binding regions. (c). TRANSFAC motif scanning of the PRDM14 binding regions. d). The distribution of the

PRDM14 motif, AP2 alpha and AP2g motifs around the centre of the binding regions. (e). Relative enrichment of BLIMP1, and PRDM14 targets on differentially expressed genes between E7.5 soma and PGCs filtered by the association of an AP2γ motif in the peak region and the combinatorial association of the peak regions to the differentially expressed genes.

S U P P L E M E N TA R Y I N F O R M AT I O N

Figure-S5 (Surani) a

PRDM14

BLIMP1.PRDM14

PRDM14.noBLIMP1.noAP2g

AP2g

BLIMP1

PRDM14.AP2g

BLIMP1.AP2g.PRDM14

BLIMP1.AP2g

BLIMP1.noPRDM14.noAP2g

AP2g.noBLIMP1.noPRDM14

planar cell polarity pathway involved in neural tube closure glucose homeostasis Rho guanyl−nucleotide exchange factor activity regulation of Rho protein signal transduction peptidyl−serine phosphorylation guanyl−nucleotide exchange factor activity GTPase activator activity protein kinase binding intracellular protein kinase cascade dendrite development protein glycosylation MAPKKK cascade negative regulation of MAP kinase activity wound healing transferase activity zinc ion binding response to DNA damage stimulus protein polyubiquitination actin cytoskeleton organization transcription coactivator activity ubiquitin−protein ligase activity dephosphorylation intracellular signal transduction protein kinase activity protein serine/threonine kinase activity transferase activity, transferring phosphorus−containing groups protein phosphorylation phosphorylation kinase activity neuromuscular junction development Rho protein signal transduction histone deacetylase binding endocytosis ATPase activity, coupled to transmembrane movement of ions, phosphorylative mechanism negative regulation of epithelial cell proliferation regulation of endocytosis DNA−dependent ATPase activity zinc ion transport Rab GTPase activator activity regulation of Rab GTPase activity damaged DNA binding lipid catabolic process microtubule motor activity microtubule−based movement SH3 domain binding germ cell development T cell receptor signaling pathway somatic stem cell maintenance Rho GTPase activator activity Rho GTPase binding cell death response to ethanol Rab GTPase binding

Color Key

−4

−2

Value

Log2 Fold Enrichment

BLIMP1

PRDM14

BLIMP1.PRDM14

AP2g

PRDM14.noBLIMP1.noAP2g

BLIMP1.AP2g.PRDM14

BLIMP1.noPRDM14.noAP2g

BLIMP1.AP2g

PRDM14.AP2g

AP2g.noBLIMP1.noPRDM14

phosphatidylinositol binding skeletal system morphogenesis negative regulation of transcription, DNA−dependent receptor binding Wnt receptor signaling pathway small GTPase mediated signal transduction regulation of transcription from RNA polymerase II promoter sequence−specific DNA binding positive regulation of cell proliferation positive regulation of neuron differentiation anterior/posterior pattern formation cell−cell signaling organ morphogenesis positive regulation of cell migration protein complex binding regulation of catalytic activity angiogenesis developmental growth embryonic hindlimb morphogenesis Wnt receptor signaling pathway, calcium modulating pathway mesoderm formation palate development calcium ion binding odontogenesis of dentine−containing tooth lung development embryonic limb morphogenesis kidney development actin binding chromatin binding heart development cell proliferation vasculogenesis positive regulation of cell−substrate adhesion integrin binding in utero embryonic development positive regulation of apoptosis induction of apoptosis by extracellular signals blood vessel development skeletal system development axon guidance negative regulation of transcription from RNA polymerase II promoter sequence−specific DNA binding transcription factor activity multicellular organismal development positive regulation of transcription from RNA polymerase II promoter positive regulation of transcription, DNA−dependent positive regulation of gene expression transcription, DNA−dependent regulation of transcription, DNA−dependent protein binding zinc ion binding DNA binding transferase activity dorsal/ventral pattern formation cell migration guanyl−nucleotide exchange factor activity GTPase activator activity endocytosis positive regulation of ERK1 and ERK2 cascade smoothened signaling pathway branching involved in ureteric bud morphogenesis homophilic cell adhesion proximal/distal pattern formation hair follicle morphogenesis protein tyrosine kinase activity positive regulation of epithelial cell proliferation positive regulation of canonical Wnt receptor signaling pathway embryonic digit morphogenesis carbohydrate metabolic process forebrain development determination of left/right symmetry ureteric bud development heart looping regulation of Rho protein signal transduction Rho guanyl−nucleotide exchange factor activity platelet activation post−embryonic development actin filament capping substrate adhesion−dependent cell spreading phosphoprotein binding central nervous system development cytoskeleton organization embryonic skeletal system morphogenesis regulation of cell cycle embryonic skeletal system development positive regulation of protein phosphorylation dephosphorylation protein serine/threonine kinase activity adult locomotory behavior mammary gland development canonical Wnt receptor signaling pathway negative regulation of canonical Wnt receptor signaling pathway axonogenesis phosphoprotein phosphatase activity kinase activity phosphorylation protein phosphorylation transferase activity, transferring phosphorus−containing groups protein kinase activity dorsal/ventral neural tube patterning cellular protein localization neural tube closure cell fate commitment double−stranded DNA binding protein tyrosine phosphatase activity transcription corepressor activity positive regulation of cell death negative regulation of BMP signaling pathway neuron fate commitment positive regulation of peptidyl−serine phosphorylation negative regulation of fat cell differentiation protein kinase C binding protein tyrosine/serine/threonine phosphatase activity cellular response to protein stimulus regulation of gene expression phosphatase activity neural crest cell migration somitogenesis phospholipid biosynthetic process response to protein stimulus

−4

−2

Log2 Fold Enrichment

Figure S5 GO-Term clustering of genes differentially associated with BLIMP1, AP2γ and Prdm14 during PGC specification. Clustering analysis of the GOterms associated with genes that are repressed (a). or induced (b). in PGCs

compared to neighbouring somatic cells at E7.5 and bound by different combinations of either BLIMP1, AP2γ or Prdm14. The dark blue colour indicates that fewer than 4 genes were mapped to the indicated GO-term.

WWW.NATURE.COM/NATURECELLBIOLOGY

S U P P L E M E N TA R Y I N F O R M AT I O N

Figure-S6 (Surani) a Wnt receptor signaling pathway

Positive regulation of ERK1 and ERK2 cascade Negative regulation of BMP signaling pathway Phosphatidylinositol binding Cell−cell signaling Cell-cell junction

Gli2

Pbx2

Positive regulation of cell proliferation

Fgf4

Ptpru

Ctnn bip1

Pcdh Wnt8a ga12

AP2g

Src

Snai1

Lama5

Chromatin modification St5

Hip1r

Prkd1 Snx15 Fgf 15

Cd44 Etnk2 Tnik

Pggt 1b

Fgf3

Irs2

Arnt2

Nphp3

Hhex Dsc2 Tyro3

Mertk

Antx r1

Pvrl3

Epc1

Smo

Satb2

Tnks

Tbx20

Hipk2 Tgfb1

BLIMP1

Wbp7

Hipk1 Hoxb3

Igf2

Lrp4

Akt2 Clstn3 Itgav Sp6

Hoxb2

Barx1 Hoxa1

Hoxb1

Hoxa2

Hoxb8

Hoxb7 Hoxb9 Cd47 Fzd6 Fzd1 Pdf Tax1 bp3 Sox4 Snx33 Timp2 Snx7 Tial1 Suv420 Creb3 Hira h1

Sox9

*C230081A13Rik

Dnmt 3b

In utero embryonic development

Odf2

Ldb2

Bre Dtx3l

Smarc al1

Hdac6

Spermatogenesis Foxa3

Prmt7

Homophilic cell adhesion Actin filament organization

Banp

Kdm 4b

Microtubule−based movement Cell cell signaling

Kit

Cdh4

Itgb3

Thbs4

Ophn1

Tns3

Kif 18b

Srga p1

Kif 16b

MAPKKK cascade BMP signaling pathway Rho protein signal transduction Intracellular protein kinase cascade Negative regulation of MAP kinase activity

Map2 Bhlh k6 a15

Cer cam Nlgn2

Usp3 3

Pik3 cb Col9 a1

Gsk 3a

Lrrc16a

Fhod3

Ninj1

Rps6 ka3

Ptpn6

Map3 k1

Zfp36 Tesk2

Dok1 Col18 a1

Ppp1 r9b

Prdm1

Sirt1

Nf1

Cntn ap2

Oxsr1

Vangl2 Dab1

Kif27

Tube1 Clasp2

Npnt Diap3 Spg7

Hist1 h1a

Pvr

Kif15

Apoe

Ppm1l

Bcl2l1

Azi1

Rad Tcp11 Acox1 51c

Map3 k3

Lrrk2

Gna13

Hbegf Ache

Pten Pdgfa

Tgf br3

Diap1

Phf21a Rbl1

Suz12 Cbx7

Ift81 Rgmb

Ash1l

Rps6 ka2

Zfp39

BLIMP1

Celsr1

Rps6 ka1

Arh gap17

Figure S6 A transcription factor network for PGC specification. BLIMP1, AP2γ and Prdm14 binding to differentially expressed genes at E7.5 between PGCs and soma. (a). Repressed genes. (b). Induced genes. The yellow nodes indicate the BLIMP1, PRDM14 and AP2γ peak and

Map4 k1

Fgfr2

Tcfa p2c

Map 4k2

Parva

Gal3 st1

Cblc

Mknk1

PRDM14

Sorl1

Dlg1

Cdh13

Jak1

Sirpa Aplp1 Diap2

Ttyh1 Itga9

Ss18

Twist1

AP2g

Gpr56

Prdm 14

Ednra

Btbd9

Gpld1 Sdk1

Kif5a

Mtap7

Gna12 Lats2 Dync 2h1

Ncor2 Ift88

Myb

Mtor

Bmp6

Mcf2

Gdf3

Kif24

Dnd1

Mark1

Dlg ap5

Gpx4 Esrrb

Tnc

Zfx

Limk2

Dab2 ip

Ptprm

Nr0b1

49*

Crem

Mll5

Sigle Kif21a c5

Actin cytoskeleton organization

Agfg1

Eya2

Hdac4

Hmg Hdac7 n5

Cell migration

Klf2

Dppa3

Imm p2l

Mea1

Dzip1

Nanos 3

Morc1

Klf10

Col4a 3bp

Hes3

Cell adhesion

D1 Pas1

Sox2

Chromatin modification

Germ cell development

Gsr

Nanog

Somatic stem cell maintenance

Hoxa3

Hoxb5

Hoxa4

Hoxd9 Hoxb6

Pax3 Rara Bmi1 L3mb Prdm6 tl3 Rcor1 Uhrf1 Chd9 Kdm6b

Dab2 Dvl1

Dusp3

Dll1

Dnmt 3a

Stat1

Cels r3

Pdg Zmiz1 Hoxb4 fra

Bmp4 Gas1 Bcl9 Wnt3a Fzd2 Dixdc1 Wnt6 Tiam1 Spata Bmper Fstl3 Wnt 10a 13 Cyr61 Pdzd2 Pard3 Ccnd1 Myc Trip6

Dsg2 Podxl

Smarc c2

Tpm1

Vil1 Myh10

Itga5

Erbb 2ip

Cxcl 12

Myo 1e

Hoxa10Tbx3

PRDM14

Crkl Cdx4

Pik3 r1 Itgb1 Robo2 bp1

Usp3

Itpr3

Mllt3

Cited2 Pbx1 Gja1 Dact1

Pcdh19 Pcdh8

*C23

Pth1r

Tshz3 Rpg rip1l

Panx1

Itgb5 Clstn1

Itga6

Notch1

Hip1

Jub

Pcdh gc3

Snx24 Wnt3

Heg1 Cdx1

Bcl2 l11

Dlk1

Arrb1

Lifr

Sbf2

Cfc1 Gata4

Ada

Twsg1

Htra1

Wnt5b

Macf1

Clstn2 Fbn2

Snx8

Embryonic skeletal system development Anterior/posterior pattern formation Proximal/distal pattern formation In utero embryonic development Embryonic limb morphogenesis Dorsal/ventral pattern formation Organ morphogenesis

Substrate adhesion dependent cell spreading Positive regulation of cell migration Regulation of cell migration Homophilic cell adhesion Integrin binding

Ss18l1

Pcgf2

Kdm3a

Fgf8

Fam 175a

Sfrp1

Taok2

Vav3 Pdlim7 Mib1

Dusp7 Pkn1

Suv 420h1

Rdh10

Wdtc1 Keap1

Robo1

*4932438A13Rik

associations. The smaller nodes indicate genes bound by the factors, with the binding associations indicated by lines connected to the yellow nodes. The colours of the gene-nodes indicate functional categories as shown.

S U P P L E M E N TA R Y I N F O R M AT I O N

Figure S7 (Surani)

Prdm14

AP2g

Blimp1

Correlation Coefficient

Figure S7 A compendium of mESC transcription factor integrated profiles. (a). A full hierarchical clustering analysis of the genome-wide BLIMP1 and AP2γ binding patterns together with binding patterns of transcriptional regulators from mESCs. The red lines above the heat-maps indicate the main clusters, showing the pluripotency cluster (Oct4 etc), polycombcluster (Ring1b etc), self-renewal/proliferation cluster (N-Myc etc), and a genome-architectural cluster (CTCF etc). The combinatorial binding pattern analysis was performed by generating a unified data matrix based

on 165,607 unique peak regions, indicating for each factor whether it was bound or not. Subsequently the hierarchical clustering and Pearson’s correlation coefficients shown in the heat map were used to investigate global relationships. The colours indicate level of correlation for all pairwise comparisons as indicated on the figure. Note that BLIMP1 associates most strongly with the self-renewal cluster and has high correlation with polycomb factors. PRDM14 associates with the pluripotency cluster whereas AP2γ binding does not correlate highly with mESC transcriptional regulators.

WWW.NATURE.COM/NATURECELLBIOLOGY

S U P P L E M E N TA R Y I N F O R M AT I O N

Figure-S8 (Surani) a

# of Bound Regions Expected Observed BLIMP1 and AP2γ 17.00 100 BLIMP1 and PRDM14 146.90 230 PRDM14 and AP2γ 162.10 568 BLIMP1 and AP2γ and PRDM14 1.60 35

p-value < 0.0001 (Permutation test, after 10.000 permutations)

# of Bound Genes Expected Observed BLIMP1 and AP2γ no PRDM14 80.02 53 BLIMP1 and PRDM14 no AP2γ 926.95 1740 PRDM14 and AP2γ no BLIMP1 365.44 628 BLIMP1 and AP2γ and PRDM14 36.22 240 BLIMP1 alone 2047.80 1058 PRDM14 alone 9351.39 8072 AP2γ alone 807.31 368 Unbound 20658.86 22115 Total 34274 34274

100 200 300 400 500 600

Observed

Expected

1 Ap2g/Blimp1 2 Blimp1/Prdm14

Number of common peaks

p-value = 0 (Chi-square test)

2e+07 4e+07 6e+07 8e+07

3 Ap2g/Prdm14

TF1-peaks * TF2−peaks

Prdm1

12 10 8 6 4

140

120

100

5 EpiLCs

Day 2

Day 4

EpiLCs

Relative levels / Arbp1

PGCLCs

1.4

Pou5f1

Day 2

Sox2

0.8

150

0.6

100 50

0.2 0

EpiLCs

Day 2

Day 4

EpiLCs

PGCLCs

Kdm4b

1.2 1

0.8

0.6

0.4

Day 2

Day 4

Dnmt3b

Day 2

Day 4

PGCLCs

EpiLCs

Day 2

Day 4

HoxB1

1.4

140

1.2

120

100

0.8

0.6

0.4

0.2

EpiLCs

Day 2

Day 4

Uhrf1

1.4

T-Brachyury

Day 4

PGCLCs

Day 4

PGCLCs

Nanog

8 6 4 2 EpiLCs

Day 2

Day 4

PGCLCs

EpiLCs

Day 2

Day 4

PGCLCs

Myc

1.2

Day 2

Cytokines

0.2 Day 2

EpiLCs

0.4

EpiLCs

0.6

Figure S8 Statistical testing for binding overlap between BLIMP1, Ap2g and PRDM14 and co-expression of BLIMP1, AP2γ and PRDM14 induces PGC-like cell fate in vitro. (a). Number of observed and expected overlap in genomic binding sites of BLIMP1, AP2γ and PRDM14, and a scatterplot showing the observed against expected overlap in genomic binding sites of: 1. AP2γ and BLIMP1, 2. BLIMP1 and PRDM14, and 3. AP2γ and PRDM14. The p-value for the enrichment of overlap in binding sites is p < 0.0001 for all comparisons. (b). A contingency table for the calculation of a chi-

Day 4

0.8

0.2 0

Day 2

0.4

PGCLCs

EpiLCs

PGCLCs

0.6

Day 4

180 160

Day 2

PGCLCs

1.6

0.8

EpiLCs

PGCLCs

1.2

0.2 EpiLCs

Nanos3

30 25

PGCLCs

1.8

0.4

Dppa3

Day 4

200

Prdm14

PGCLCs

250

1.2

160

2 0

Tcfap2c

EpiLCs

Day 2

Day 4

Blimp1 AP2g Prdm14

PGCLCs

square p-value for the overlap of genes bound by different combinations of BLIMP1, AP2γ and PRDM14. The calculated p-value based on this table was p < 1x10-299. The total number of genes (34274) represents the number of unique gene identifiers in Ensembl that were a basis for the gene annotation of the respective transcription factor binding sites. (c). RT-qPCR analysis of sorted fluorescent PGCLCs on Day2 and 4 of either cytokine or doxycycline induction, as well as EpiLCs. The experiment is the second of two experiments performed. The first experiment is shown in Figure 8.

S U P P L E M E N TA R Y I N F O R M AT I O N

Supplementary Tables Legends Table S1 Microarray Expression Data. This table shows the genome wide differential gene expression in P19EC cells upon transient transfection of BLIMP1. The first four columns show the Entrez Gene ID, the gene symbol and the Illumina Probe ID as well as the gene definition. The next 3 columns give the false discovery rate (FDR) for each comparison and the last 3 columns show the log2 fold change (LFC) for each comparison. Table S2 BLIMP1 Peaks. A list of the high-confidence binding regions of BLIMP1. Table S3 BLIMP1 Peaks with Gene Annotation. Gene annotations for the high-confidence binding regions of BLIMP1. Table S4 Single Cell PGC RNAseq Expression Data. This table shows the genome wide differential gene expression in nascent PGCs (E6.5, E7.5 and E8.5) and somatic neighbours (E7.5 soma) as well as Blimp1 null PGC-like cells (E7.5 Blimp1KO), assayed by single cell RNAseq. The first column shows the RefSeqID, the second column the gene symbol, the third column shows the gene localization and strand. The values indicate the expression levels in log2(reads pr. million). Table S5 AP2g Peaks. A list of the high-confidence binding regions for AP2g. Table S6 AP2g Peaks With Gene Annotation. Gene annotations for the high-confidence binding regions of AP2g. Table S7 Primer Sequences. The table shows the primer sequences of primers used for RT-qPCR as well as ChIP-qPCR experiments in 3’ to 5’ direction.