Development 127, 5449-5461 (2000) 5449 Printed in Great Britain © The Company of Biologists Limited 2000 DEV2556

Taube nuss is a novel gene essential for the survival of pluripotent cells of early mouse embryos

Anne K. Voss*,‡,§, Tim Thomas*,‡, Petros Petrou, Konstantinos Anastassiadis1, Hans Schöler2 and Peter Gruss Max-Planck-Institute of Biophysical Chemistry, Department of Molecular Cell Biology, Am Fassberg 11, 37077 Goettingen, Germany 1European Molecular Biology Laboratory, Gene Expression, Meyerhofstr. 1, 69117 Heidelberg, Germany 2University of Pennsylvania, New Bolton Center, Center for Animal Transgenesis and Germ Cell Research, 382 W. Street Rd, Kennett Square, PA 19348, USA *These contributed equally to this work ‡Present address: Development and Neurobiology, The Walter and Eliza Hall Institute of Medical Research, Royal Parade, Parkville, Victoria 3050, Australia §Author for correspondence (e-mail: [email protected])

Accepted 19 September; published on WWW 14 November 2000

SUMMARY The cells of the inner cell mass constitute the pluripotent the zonae pellucidae, implanted and induced decidual cell population of the early embryo. They have the potential reactions, but failed to develop beyond E4.0. At this time to form all of the tissues of the embryo proper and the trophoblast cells were viable, but inner cell masses were some extra-embryonic tissues. They can be considered a not detectable. At E3.75, massive TUNEL-positive DNA transient stem cell population for the whole of the embryo, degradation and chromatin condensation were visible and stem cells maintaining the same capacity can be within the inner cell masses, whereas the cell membranes isolated from these cells. We have isolated, characterised where intact. Caspase 3 was expressed in these cells. In and mutated a novel gene, taube nuss (Tbn), that is essential vitro, the inner cell mass of mutant embryos failed to for the survival of this important cell population. The proliferate and died after a short period in culture. These taube nuss protein sequence (TBN) was highly conserved results indicate that the novel protein, taube nuss, is between human, mouse, Xenopus laevis, Drosophila necessary for the survival of the inner cell mass cells and melanogaster, Caenorhabditis elegans and Arabidopsis that inner cell mass cells died of apoptosis in the absence thaliana, particularly in a domain that is not present in any of the taube nuss protein. As cell pruning by apoptosis published proteins, showing that TBN is the founding is a recognised developmental process at this stage of member of a completely new class of proteins with an development, the taube nuss protein may be one of the important function in development. The Tbn gene was factors regulating the extent of programmed cell death at expressed ubiquitously as early as E2.5 and throughout this time point. embryonic development. It was also expressed in adult brain with slightly higher levels in the hippocampus. The Tbn mutant embryos developed normally to the blastocyst Key words: Inner cell mass, Stem cells, Apoptosis, Programmed cell stage and contained inner cell masses. They hatched from death, Embryoblast, Blastocyst, Mouse, TBN

INTRODUCTION pruned from this population by programmed cell death (Smith and Wilson, 1971). This presumably occurs as a The first cell-lineage specification and restriction in counterbalance to excessive cell proliferation or to remove mammalian development occurs in preimplantation embryos, cells that started to differentiate into an undesirable direction when blastocysts form from compacted morulae. They consist (Pierce et al., 1989). In wild-type E3.5 embryos up to 10% of of an outer cell population, the trophectoderm, from which the inner cell mass cells have been observed to undergo apoptosis, trophoblast compartment of the extra-embryonic tissues but cell death is not observed in the trophectoderm lineage at develops and an inner cell population, the inner cell mass or this point (reviewed by Sanders and Wride, 1995, and by embryoblast, from which all tissues of the embryo proper, the Pampfer and Donnay, 1999). Electron microscopy, vital dye yolk sac, the amnion and the allantois, develop. The cells of exclusion and in situ end-labelling studies have shown that the the inner cell mass are, therefore, a transient pluripotent stem cells dying in the inner cell masses exhibit the typical signs of cell population of the embryo proper and some extra- apoptosis (Copp, 1978; Handyside and Hunter, 1986; Brison embryonic tissues (reviewed by Hogan et al., 1994). Cells are and Schultz, 1997). 5450 A. K. Voss and others

During hatching and implantation the inner cell mass MATERIALS AND METHODS proliferates. Then the proamniotic cavity is formed, and the cells that formerly formed the inner cell mass are organised Generation of the mutant Tbngt allele into an epithelial structure, the primitive ectoderm or epiblast, The promoter-less reporter construct pGT1.8geo (kindly provided by which lines the proamniotic cavity. The process of proamniotic W. Skarnes; Skarnes et al., 1995) was electroporated into the parental cavity formation involves the programmed cells death of inner murine embryonic stem (ES) cell line MPI-II (Voss et al., 1997) as cell mass cells in the centre of the inner cell mass and selective described previously (Voss et al., 1998b). survival of cells in the periphery of the inner cell mass Cloning and sequencing of the Tbn cDNA and the 5′ (Coucouvanis and Martin, 1995). The surviving cells are junction of pGT1.8geo and the Tbn locus thought to receive a survival signal from the basement Total RNA was isolated from ES cells heterozygous for the insertion membrane surrounding the inner cell mass, while the cells in of pGT1.8geo into the murine genome (clone F5). 5′ rapid the centre die, because they received and executed a death amplification of cDNA ends was performed as previously described signal. Both processes involving cell death, the pruning of (Voss et al., 1998a) using oligonucleotides complementary to the lacZ inner cell mass cells and the formation of the proamniotic gene of pGT1.8geo as gene-specific primers. RACE products were cavity, require careful regulation of the death process, as some isolated and sequenced. The sequences specific for the endogenous cells are entailed to die, while immediate neighbours are gene were amplified by PCR, cloned into pGemT (probe 1, Fig. 1A) required to survive. and used as a probe to screen an ES cell cDNA library. The ES cell + Programmed cell death occurs as a regular mechanism cDNA library was generated for this purpose. PolyA RNA from wild- during embryonic development in order to prune proliferating type ES cells (MPI-II) was reverse transcribed with random hexamere oligonucleotides and second-strand synthesis was carried out using a cell population of excess or aberrant cells or to shape tissues. cDNA synthesis kit (Pharmacia). After filling-in reaction, ligation to Well-known examples of developmental cell death during EcoRI/NotI adapters (Pharmacia), size selection on a Sephacryl mammalian development are the involution of the interdigital S400HR column, the cDNA was cloned into the EcoRI site of webs and of epithelial structures in areas where tissues fuse, λpExcell (Pharmacia) and packaged into Gigapack III Gold e.g. the fusion of the palate (reviewed by Sanders and Wride, (Stratagene). E. coli NM522 were transfected with the phages and 1995; Vaux and Korsmeyer, 1999). Initially, programmed cell plated. The cDNA library of 1,000,000 plaques was harvested without death during development was always thought to involve further amplification. The average insert size was 2 kb±0.8. 0.17% of apoptosis. More recently, it has become clear that other forms the clones contained β-actin cDNA as an insert. Five cDNA clones isolated from this library were sequenced. One clone contained the of programmed cell death that have features in common with ′ necrosis may also occur during development (Chautan et al., entire open reading frame, but not the 5 UTR, which was obtained by 5′ RACE, and only part of the 3′ UTR. The remaining 3′ 1999). untranslated sequences were obtained by 3′ RACE using cDNA Apoptosis is characterised by the internal disassembly of the generated from E12.5 embryos. Three independent PCRs were cell, particularly the DNA, while the cell membrane remains performed using oligonucleotide primers derived from the coding intact. Consequently, chromatin and other cellular material is sequence. Each reaction produced a band of the same size which was fragmented into cell membrane-enclosed portions that can cloned into pGemT and sequenced. easily be cleared by neighbouring cells. In this way adverse DNA was isolated from a tail biopsy of an animal heterozygous for effects of cell death on neighbouring cells and an inflammatory the pGT1.8geo insertion in the Tbn locus. The 5′ junction was reaction are avoided. Developmental cell death is thought to be amplified by PCR using and a lacZ specific primer (5′ GGCGATCGGTGCGGGCCTCTTCGC 3′) and an Tbn-specific regulated by extracellular death-inducing factors and survival ′ ′ factors. There is some evidence that tumour necrosis factor α primer (5 GGAGACACTGACAGAGATGCTGCAGAGC 3 ). The PCR product was cloned into pGemT (Promega) and sequenced. and transforming growth factor α may be, respectively, cell death-inducing and survival factor for the inner cell mass cells Sequence comparison of blastocysts (Brison and Schultz, 1997; Pampfer et al., 1997). The deduced protein sequence was encoded by the longest open Internally, apoptosis involves an antagonism of pro- and anti- reading frame, started with the first ATG, and was in frame with the apoptotic factors. Once apoptosis is induced, the activation of lacZ ORF. The cDNA and the protein sequence were compared with a cascade of proteolytic enzymes including caspases (cysteine entries into the EMBL data bases, SWISS-PROT, and GenBank by proteases that cleave their substrate after an aspartic acid blast and fasta analysis (Wisconsin Package, Genetics Computer residue) ultimately results in the disassembly of the cell Group). Similarities between protein sequences were analysed using (reviewed by Vaux and Korsmeyer, 1999). Caspases are ClustallW (Thompson et al., 1994) and depicted using MacBoxshade (http://www.ch.embnet.org/software/BOX_form.html). synthesised as inactive proenzymes and require cleavage for their activation. Chromatin condensation and DNA Generation of the mutant Tbngt mouse line fragmentation, major manifestations of apoptosis, can also Chimaeras were produced by a technique modified from Nagy and occur in a caspase-independent manner (Susin et al., 1999). Rossant (1993) omitting the sandwich technique (Voss et al., 1998b). Some of the genes coding for proteins involved in apoptosis Chimaeras were mated with wild-type females of the mouse strains are expressed in preimplantation embryos (Weil et al., 1996; C57Bl/6 and NMRI to produce animals heterozygous for the Tbngt Jurisicova et al., 1998; Exley et al., 1999). mutation. Noon of the day on which the vaginal plug was observed We have isolated a novel mouse gene, taube nuss (Tbn), the was termed day 0.5 of gestation (E0.5). founding member of a novel group of proteins, and have Genotyping of mice and embryos and Tbn RT-PCR generated Tbn mutant mice. These mice exhibit disturbances DNA was isolated from tail biopsies, extra-embryonic membranes of in the balance between cell death and cell survival in the early embryos (E9.5 to E15.5) and whole embryos (E8.5). For genotyping embryo, so that the pluripotent inner cell mass cells all die by of BamHI-digested DNA samples by Southern analysis, probe 1 (Fig. apoptosis while the trophectoderm cells survive. 1A,C) was generated by cloning the Tbn-specific sequences of the 5′ Tbn, a novel gene essential for the ICM 5451

RACE product into pGemT. Tbn RT-PCR was performed using 5′ GG- previously (Voss et al., 1998b). For staining of preimplantation AGACACTGACAGAGATGCTGCAGAG C 3′ and 5′ ATCTGGTA- embryos 0.1% BSA was added to all solutions except the staining GTCAGACACTGGCTCACGG 3′ as primers. Radioactive RT-PCR solution to reduce stickiness, and embryos were fixed for 3 minutes on RNA isolated from preimplantation embryos was performed as only. described previously (Nichols et al., 1998). Explant and cell culture Whole embryo and histological analysis E3.5 embryos of Tbngt/+ heterozygous mutant intercrosses were Embryos were flushed from their mothers’ uteri, placed in M2, viewed isolated and plated onto gelatine coated 24-well plates into ES cell and photographed using a Labovert inverted microscope (Leitz). For derivation medium (Voss et al., 1997). Control cultures were histological analysis embryos or whole concepti in their mothers’ uteri established with embryos that were wild type at the Tbn locus. To were fixed in 4% paraformaldehyde, infiltrated and embedded in E3.5 embryos cultured in the same way, 2 ng/ml recombinant human paraffin. Serial sections (5-8 µm) were cut, deparaffinised and stained FGF4 (R&D Systems) and 1 µg/ml heparin were added. with Haematoxylin and Eosin by standard techniques. Slides were viewed and photographed with differential interference contrast optics Generation and processing of chimaeric embryos using a Zeiss Axiophot microscope. Photographs of serial sections of For chimaeric analysis of the mutant phenotype in vitro, embryos of mutant and control animals were compared in detail. Tbngt/+ heterozygous mutant intercrosses were recovered in the eight- cell stage, labelled with a lipophilic fluorescent dye (CellTracker CM- Northern analysis, in situ hybridisation and β- DiI, Molecular Probes) and aggregated with wild-type unlabelled galactosidase staining embryos, essentially as describe previously (Nagy and Rossant, 1993). Northern blots were prepared by standard techniques and hybridised The aggregates were cultured for 24 hours in M16 under oil and then with [α-32P]dCTP labelled Tbn-specific probe 1 described above (Fig. viewed and photographed with a fluorescent microscope (Axiophot 2, 1A). In situ hybridisation was performed essentially as described Zeiss). For chimaeric analysis of the mutant phenotype in vivo, (Hogan et al., 1994). Briefly, section were deparaffinised, rehydrated embryos of Tbngt/+ heterozygous mutant intercrosses were recovered through graded alcohol concentrations, incubated for 10 minutes (or at the eight-cell stage and aggregated with tetraploid embryos that 30 minutes for adult brains) at room temperature in 10 mg/ml were wild type at the Tbn locus, essentially as describe previously proteinase K, fixed in 4% paraformaldehyde for 10 minutes, then (Nagy and Rossant, 1993; Voss et al., 2000). Chimaeric concepti were dehydrated through graded alcohol concentrations. Sections were air recovered at E9.5. The embryos were photographed and then used for dried, then hybridisation solution containing 5×105 cpm/µl in vitro genotyping by Tbn-specific Southern analysis. The extra-embryonic transcribed cRNA probe (probe 2, Fig. 1A) was placed onto the membranes were used for genotyping. section. Slides were incubated overnight at 56°C and then washed as described (Hogan et al., 1994). Slides were autoradiographed at 4°C Myc tagging for 14 days, developed and counterstained with Haematoxylin. The coding sequence of the Tbn cDNA was cloned in frame with six Embryos were stained for β-galactosidase activity as described Myc epitopes (Roth et al., 1991) into pCS2+MT and sequenced to A probe 1 probe 2 NLS Fig. 1. Tbn mRNA and Southern analysis. S1 P S2 3’UTR (A) Schematic drawing of the Tbn mRNA, sequences coding for two regions of sequence similarity with H1 H2 related proteins (S1 and S2, grey) containing a putative α-helix each (H1 and H2, striped), the proline-rich domain (P, dotted), an area of 4 putative nuclear localization signals (NLS, black) and the position of the gene trap insertion at base 216 (arrowhead) are indicated. The 3′ UTR is shown truncated at base 1153, not showing the remaining 1430 bases. The probes used for northern (probe 1), Southern analysis (probe 1 and 2), and for in situ hybridisation (probe 2) are indicated. (B) Northern analysis of total RNA isolated from adult brain, E12.5 placenta, E12.5 embryo, ES cell and ES cells heterozygous for the gene trap insertion at the Tbn locus probed with a Tbn-specific probe (probe 1 in A). The positions of the 18S (1.9 kb), 28S (4.7 kb), 45S (13 kb) ribosomal RNA, the Tbn mRNA of 2.5 kb and the Tbn/pGT1.8geo fusion mRNA are indicated. The ethidium bromide image of the 18S rRNA on the gel is shown below. (C) Southern analysis of DNA isolated from tail biopsies of offspring of Tbngt/+ intercrosses cut with BamHI and probed with probe 1. The positions of the bands resulting from the wild-type (wt, Tbn+) and the mutant (mt, Tbngt) allele are indicated. Results show two wild-type (wt) and four heterozygous (ht) animals. 5452 A. K. Voss and others

....10....20....30....40....50....60 TBN_Mmusculus MADTAAGPGGSGTRPGSKQSTNPADNYHLARRRTLQVVVSSLLTEAGFESAEKASVETLT A PRODOS_Dmelanogaster MEKVEA...... VTVSNVDPY....RRILNKVVSQLLLDKGAGQASNHSLETLT

....70....80....90...100...110...120 TBN_Mmusculus EMLQSYISEIGRSAKSYCEHTARTQPTLSDIVVTLVEMGFNVDTLPAYAKRSQRMVITAP PRODOS_Dmelanogaster QMLQALIWEIGNSAHNYCELSGRTMPTVGDVSLALINMGISISNLDPYMRKETHVPIPLP

...130...140...150...160...170...180 TBN_Mmusculus PVTNQPVTPKALTAGQNRPHPPHIPSHFPEFPDPHTYIKTPTYREPVSDYQILREKAASQ PRODOS_Dmelanogaster PQQTQQRPLSLLQAGIKAPHPHYVPSYFPPMPDPHAYIRTPTHKQPVTEYEAIREKAACH

...190...200...210...220...230...240 TBN_Mmusculus RRDVERALTRFMAKTGETQSLFKDDVSTFPLIAARPFTIPYLTALLPS...... ELEIQ PRODOS_Dmelanogaster KRDIEKALTKFLCKTTETNNLFPTEDNMFPLIACKPAFPPYAAALNPTDQVFDFEELEYH

...250...260...270...280...290...300 TBN_Mmusculus QMEETDSSEQEEQTDTEN.NALHISTDDSGAEKESASVLQQSSSLSGSRNGEESVIDNPY PRODOS_Dmelanogaster YLVANRTEDEPSKDDGEEGDSENEEMDGDKSKEEKPELDIKPNSNTNKAILENPNIDNPY

...310...320...330...340.. TBN_Mmusculus LRPVKKPK...... IRRKKSLS...... PRODOS_Dmelanogaster LRAATLPKRSKNCPTPGTMPSRSLATTAPTIRTPSTLEITKTNL

....10....20....30....40....50....60 TBN_S1 LQVVVSSLLTEAGFESAEKASVETLTEMLQSYISEIGRSAKSYCEHTARTQPTLSDIVVT B Human_S1 LQVVVSSLLTEAGFESAEKASVETLTEMLQSYISEIGRSAKSYCEHTARTQPTLSDIVVT Xlaevis_S1 LQVVVSSLLTEAGFDSAEKAAVESLTEMLQSYLSEIGRSAKSYCEHTARTQPTLPDIVVT PRODOS_S1 LNKVVSQLLLDKGAGQASNHSLETLTQMLQALIWEIGNSAHNYCELSGRTMPTVGDVSLA Celegans_S1 LEQIVTAMCYKSGFDDIEEGALETLMLLFHSYIKRIGEQSRFACEVAGRTIVTPGDVWFG Athaliana_S1 AKAAVAQVCESVGYENFKDPALESLSGFALQYILQLGKTATSFANLTGRSQCNVFDIILA

.. TBN_S1 LVEMG mouse human xeno. dros. caen. arab. Human_S1 LVEMG Xlaevis_S1 LIEMG PRODOS_S1 LINMG TBN_S1 ---- 100.0 96.9 66.2 50.8 49.2 Celegans_S1 LVNMG Human_S1 100.0 ---- 96.9 66.2 50.8 49.2 Athaliana_S1 LDDLT Xlaevis_S1 90.8 90.8 ---- 64.6 52.3 50.8 PRODOS_S1 50.8 50.8 47.7 ---- 49.2 43.1 Celegans_S1 35.4 35.4 33.8 35.4 ---- 38.5 Athaliana_S1 24.6 24.6 24.6 26.2 21.5 ----

....10....20....30....40....50....60 TBN_S2 PHPPHIPSHFPEFPDPHTYIKTPTYREPVSDYQILREKAASQRRDVERALTRFMAKTGET C Xlaevis_S2 PHPAHIPSHFPEFPDPHTYIKTPTYREPVCDYQVLREKAASQRRDVERALTRXMAXTGET PRODOS_S2 PHPHYVPSYFPPMPDPHAYIRTPTHKQPVTEYEAIREKAACHKRDIEKALTKFLCKTTET Athaliana_S2 PPGKHIPLWLPAFPDPHTYKETPMWIERVSDPRGDKIEQARQRRKAERALLSLQRKLVCK Celegans_S2 PHPSYVYEWLPPLPDPHTYIKTEISEDIDFSYEKVREAMSQKKRNGITSLVNYMIRNYPS

.. TBN_S2 QSLF Xlaevis_S2 QSLF PRODOS_S2 NNLF mouse xeno. dros. arab. caen. Athaliana_S2 ISSR Celegans_S2 ICLF TBN_S2 ---- 93.8 76.6 46.9 43.8 Xlaevis_S2 92.2 ---- 71.9 43.8 40.6 PRODOS_S2 57.8 54.7 ---- 40.6 50.0 Athaliana_S2 43.8 40.6 26.6 ---- 34.4 Celegans_S2 32.8 32.8 35.9 21.9 ----

Fig. 2. Sequences comparison of TBN and related proteins. (A) Amino acid sequence comparison of the deduced TBN protein (TBN_Mmusculus) and an unpublished D. melanogaster protein (PRODOS_Dmelanogaster), both full length. Identical amino acid residues are shown in red, shaded black, similar residues are shown in grey, shaded grey, and different residues are not shaded. (B,C) Amino acid sequence comparison of two regions of particular sequence similarity (S1 and S2) between TBN and related proteins. (B) Similarity region 1 (S1) of TBN (TBN_S1), the deduced protein sequence of a human EST (Human_S1, GenBank Accession Number, AA641254), the deduced protein of a partial X. laevis cDNA (Xlaevis_S1, AC#AW199998), PRODOS (PRODOS_S1, GenBank Accession Number, Y15513), a hypothetical protein deduced from sequences of C. elegans chromosome II (Celegans_S1, GenBank Accession Number, Z46934) and a hypothetical protein deduced from genomic sequences of chromosome 4 of A. thaliana (Athaliana_S1, GenBank Accession Number, CAB36711). The percentage of sequence identity is shown in the lower half, and the percentage of sequence similarity in the upper half of the table inserted. (C) Similarity region 2 (S2) of TBN (TBN_S2), S2 of the X. laevis protein, S2 of PRODOS, S2 of the same C. elegans protein and of the same A. thaliana protein shown in B. The human sequence did not extent far enough 3′ to yield information about the human S2. Note that the human protein and TBN are 100% identical in S1. TBN and the Xenopus protein are more than 90% identical in S1 and S2. The degree of sequence identity in mouse and Arabidopsis in S2 is higher than that of mouse and Caenorhabditis. Tbn, a novel gene essential for the ICM 5453 confirm the successful cloning. The tagged protein was in vitro of the Tbn cDNA was cloned by 5′ RACE using translated. The Tbn/6myc containing plasmid was transfected into oligonucleotides specific for sequences of the gene trap COS7 or NIH3T3 cells using lipofectamine (Gibco BRL). Five, 8, 10 construct. RACE products of about 450 bases were isolated and or 24 hours after transfection the cells were fixed in 4% sequencing revealed no significant similarity to published paraformaldehyde, subjected to Myc immunofluorescence detection sequences in the databases. Using the sequences specific for using a mouse monoclonal anti-c-Myc (9E10, Santa Cruz, 1:100) and the endogenous gene, the Tbn sequence 3′ of the gene trap an Alexa 568 goat anti-mouse antibody (Molecular Probes, 1:500), and counterstained with DAPI. insertion was cloned from a random primed cDNA library made for this purpose from ES cells. The resulting Tbn cDNA Immunocytochemistry clones spanned 1594 bases of the 2.5 kb cDNA and contained Zona pellucidae were removed from preimplantation unhatched the entire open reading frame. The 5′ UTR and part of the 3′ embryos with acid Tyrode’s solution and hatched embryos were used UTR were cloned by 5′ and 3′ RACE. The Tbn mRNA was as such. For FGF4 and OCT4 detection, the embryos were fixed in 2583 bases in size (Fig. 1A,B; GenBank Accession Number, 2% paraformaldehyde in PBS and incubated with intervening and a AF222802). The 5′ UTR was 20 bases, the 3′ UTR 1638 bases final incubations in PBS plus 0.25% gelatine in 0.1% Triton X-100 in long. The open reading frame of 924 bases started at the first PBS, first antibody solution and second, fluorescent antibody solution. µ ATG which was a non-Kozak sequence (Kozak, 1989). This One PBS plus gelatine solution also contained 0.1 g/ml of the open reading frame coded for a protein of 308 amino acids. fluorescent nuclear stain Hoechst 33258 (Molecular Probes). Then the α embryos were mounted with an aqueous mounting medium (Mowiol) The deduced protein contained two regions predicted to be - and viewed and photographed using a fluorescence microscope helices (King and Sternberg, 1996) between amino acids 60 (Axiophot 2, Zeiss). For caspase 3 detection, the embryos were first and 70, and 170 to 192; four putative nuclear localisation mounted onto object slides, then fixed in acetone, and then treated signals (NLS, Psort II, proteomics tools, SwissProt), i.e. two with the antibody solutions. The first antibodies and dilutions used four-residue patterns at amino acids 297 and 302, and two were a goat anti-human FGF4 (Santa Cruz, 1:50), rabbit anti-murine seven-residue patterns at amino acids 295 and 299; and a OCT4 (Palmieri et al., 1994; 1:400), and rabbit anti-active human proline-rich domain between amino acids 120 and 166 (Fig. caspase 3 (Pharmingen, 1:100). The second antibodies used were 1A). Alexa 568 donkey anti-goat IgG and Alexa 568 goat anti-rabbit IgG The gene trap insertion occurred at base 216 with respect to (both Molecular Probes, 1:1000). In our hands, the anti-active caspase the cDNA that disrupted the protein-coding region after the 3 antibody showed strong binding to cells that were also exhibiting chromatin condensation. However, staining of all cells of the first 20%, i.e. at amino acids 66 (Fig. 1A). With the gene preimplantation embryo was visible at a lower level. trap construct we inserted a lacZ reporter/neomycin phosphotransferase selector gene into the Tbn locus. A fusion TUNEL and Trypan Blue exclusion test mRNA of the predicted size of about 4.7 kb was detectable in The zona pellucidae were removed from E3.75 embryos recovered RNA isolated from ES cells heterozygous for the gene trap from Tbngt/+ heterozygous mutant intercrosses or wild-type insertion at the Tbn locus (Fig. 1B). The 5′ genomic intercrosses. The embryos were fixed for 10 minutes in 1% junction between Tbn and pGT1.8geo apparently fused intron paraformaldehyde in PBS, washed in PBS and dried onto object sequences of the Tbn locus to intron sequences in the en2 splice slides. On the slides, they were processed with intervening PBS acceptor of pGT1.8geo. The sequence at the fusion is 5′ washes through 0.1% Triton X-100 15 minutes, then equilibration ATCACCACCCCCAATGCCCAACACTTGTATGG 3′ with buffer and terminal transferase reaction, incorporating digoxigenin- the Tbn sequences underlined. Southern analysis using probe labelled deoxynucleotide and fluorescein-labelled anti-digoxigenin ′ antibody (all ApoTag in situ apoptosis detection kit, Oncor 2 (Fig. 1A) showed that the 3 region of the endogenous locus Appligene). As positive controls, embryos were treated with 0.5 µg/ml was intact (data not shown). Use of the splice acceptor and DNase I after the permeabilisation step. The slides were covered with polyadenylation signal of the gene trap construct resulted in a an aqueous mounting medium (Mowiol) to which 0.5 µg/ml Hoechst protein truncated in the middle of the first region of similarity 33258 were added. The embryos were viewed and photographed using (S1) with related proteins (described below, Fig. 2) in the a fluorescence microscope (Axiophot 2, Zeiss). For Trypan Blue middle of the first putative α-helix and fused to a β- exclusion, embryos were recovered and incubated for 10 minutes in galactosidase/neomycin phosphotransferase fusion protein. 0.4% Trypan Blue in M16 culture medium, then washed in M2 and This fusion protein lacked one half of the first region of viewed immediately under an inverted microscope (Labovert, Leitz). sequence similarity of TBN and related proteins, the entire As positive controls, wild-type embryos were killed using sodium azide. second region of similarity, half of the first and all of the second putative α-helix, the proline-rich domain, and the putative Statistics nuclear localization signal. As the first region of similarity is Number of cells in morphologically abnormal and normal embryos 100% conserved between human and mouse, and 51% obtained were compared by Student’s t-test. Results are given as conserved between fly and mouse, it appears to be one of the mean±standard deviation. functional domains of the protein. Since this region is disrupted, a truncated protein would not be likely to confer residual function or a gain of function. The heterozygous RESULTS animals developed and lived normally indicating that fusion protein generated from one mutant allele had no adverse effect. Isolation and cloning of taube nuss and generation Owing to the early lethality of the Tbngt mutation we are unable of the mutant allele to establish if low levels of wild-type RNA were produced from In a gene trap screen for genes that are important in embryonic the mutant allele as discussed previously (Voss et al., 1998a). development (Voss et al., 1998b) we isolated a novel murine We called the fusion protein containing the first 66 amino acids gene which we called taube nuss, Tbn (empty nut). The 5′ end of TBN, TBN66/β-gal/neo. 5454 A. K. Voss and others TBN-related proteins The cDNA sequence and the deduced protein sequence had no similarities to published sequences, but the TBN protein sequence showed 45% sequence similarity to an unpublished SWISSPROT database entry, the PRODOS protein of D. melanogaster (GenBank Accession Number, Y15513), 32% sequence similarity to five non-continuous neighbouring regions on chromosome II of C. elegans (GenBank Accession Number, Z46934), and 28% sequence similarity to a hypothetical protein on chromosome 4 of A. thaliana (GenBank Accession Number, CAB36711). In addition to the sequences described here, several human and mouse ESTs were found to be similar to the Tbn cDNA. Finally an incomplete cDNA of X. laevis (GenBank Accession Number, AW199998) was found to encode a protein very similar to TBN. However, as none of these ESTs was a full-length sequence, they were not included in Fig. 2A. The two regions of highest similarity (S1 and S2) between the TBN-related proteins are depicted in Fig. 2B,C. Within these two regions, S1 and S2, TBN showed 100% sequence identity to a protein deduced from a human EST, 97% and 94% sequence similarity to the deduced Xenopus protein, 66% and 77% to PRODOS, 51% and 44% to the Caenorhabditis protein, and 50% and 47% to the Arabidopsis protein, respectively. Particularly conserved was a domain within the second similarity region in position 138 to 188 in our diagram (Fig. 2A, amino acids 1 to 50 in Fig. 2C), which we called the PDPH domain. Within this region the amino acids PDPH, two preceding prolines, and carboxy-terminal of the PDPH domain, a tyrosine, a threonine, an arginine and a leucine residue were conserved 100% between vertebrate, insect, worm and plant, both in amino acid identity and spacing. The sequences of all proteins were used to predict a putative common secondary structure. Both similarity domains were predicted to contain an α-helix each (H1 and H2; King and Sternberg, 1996). α-helices in these regions were also predicted by all other algorithms used to predict secondary structure. The PDPH domain at the beginning of the second similarity domain was predicted to form random coils. In addition, the mouse, Drosophila, Fig. 3. Expression pattern of Tbn. Embryos recovered from matings involving gt/+ β Caenorhabditis and the Arabidopsis proteins were 1 or 2 Tbn heterozygous parents stained for activity of the -galactosidase predicted to contain nuclear localization signals at the reporter gene (A,C,G), or paraffin sections of wild-type embryos (B,D-F,H,I) or adult brain (J,K) hybridised in situ with a Tbn-specific cRNA C terminus, in a region where the primary structure probe (probe 2 in Fig. 1A) or a sense control probe (F). Bright-field images contains only a few similar amino acid residues. The (B,D,I,K) and dark-field images (E,F,H,J). (A) E2.5 embryos, (B) E6.5 other sequences did not extend far enough C- embryo, (C) E8.5 embryo, (D,E) E8.5, Tbn-specific probe, (F) E8.5, sense terminally to include this region. At its N terminus, control, (G) E11.5 embryo, (H,I) E11.5, Tbn-specific probe, (J,K) Adult brain, TBN showed similarity to several transcription dentate gyrus and hippocampus, Tbn-specific probe. Note the uniform factors and histones, suggesting that this might be a expression of the Tbn gene as judged by in situ hybridisation, which was DNA-binding domain. In contrast, PRODOS and distinguishable from background (compare E with F). In contrast, the β- the two hypothetical proteins did not contain such galactosidase activity was not as uniform. It was higher in E11.5 heart sequences. (G). Asterisk marks blood, which causes nonspecific background in dark field, Searches through all available databases did not arrow marks β-galactosidase-positive embryos. al, allantois; dg dentate gyrus; drg, dorsal root ganglia; e, embryonic ectoderm; ee, extra-embryonic yield any other proteins with the PDPH domain. ectoderm; fb, forebrain; hb, hindbrain; he, heart, hf, headfolds; hi, Therefore, we conclude that the TBN-related proteins hippocampus; li, liver; pc, proamniotic cavity; so, somites. Scale bars: 63 µm are distinct from all previously described protein (A), 35 µm (B), 450 µm (C), 440 µm (D-F), 1.5 mm (G-I), 430 µm (J,K). families. Tbn, a novel gene essential for the ICM 5455

Fig. 5. Phenotype of Tbngt/gt mutant embryos. (Α) β-galactosidase Fig. 4. Subcellular localization of TBN protein. (A) β-galactosidase staining of E3.5 embryos recovered from Tbngt/+ heterozygous staining of Tbngt/+ ES cells. (B,D,F) Myc immunocytochemistry of intercrosses. The arrows point to two embryos showing twice the COS7 cells transfected with a construct coding for a TBN/6MYC intensity of β-galactosidase staining as heterozygous embryos. These fusion protein (B,D) or the 6MYC tags alone (F). (C,E,G) nuclear embryos were considered homozygous. (B-G) Haematoxylin and stain DAPI. Note the localization of the Myc immunofluorescence to Eosin stained sections of uteri (B,C, E4.0; D,E, E5.0) or deciduae the nuclei (arrowhead) or the cytoplasm (arrow) in B, and the nuclei (F,G, E5.5) of Tbngt/+ heterozygous intercrosses. Mutant morphology in D and the cytoplasm in F. The TBN66/β-gal/neo fusion protein is shown in B,D,F; normal morphology is shown in C,E,G. Note that was found in the nucleus of some cells (arrow, A) and in the the strongly staining homozygous embryos in A exhibit normal E3.5 cytoplasm of others (A). Scale bars: 30 µm (A), 24 µm (B,C,F,G), morphology including inner cell masses. In contrast, at E4.0 embryos 15 µm (D,E). without an inner cell mass were observed (B), although trophoblast cells are present. At E5.0 no structure resembling an embryo proper could be detected and trophoblast cells had collapsed (compare Expression pattern of taube nuss D with E). At E5.5 no embryo was visible (compare F with G). Northern analysis showed low levels of Tbn mRNA in ES cells e, ectoderm; ec, ectoplacental cone; ee, extra-embryonic ectoderm; icm, inner cell mass; te, trophectoderm. Scale bars: 64 µm (A), 20 and E12.5 embryo, and even lower levels in E12.5 placenta and µm (B,C), 50 µm (D,E), 45 µm (F,G). adult brain. The northern blot shown in Fig. 1B was exposed for 14 days. Low β-galactosidase activity was first detected in compacted morulae at E2.5 when all blastomeres were staining From E9.5 onwards, somewhat higher level of β-galactosidase uniformly (Fig. 3A). Comparison of β-galactosidase staining activity were observed in the developing heart (E11.5 shown and Tbn-specific in situ hybridisation showed that the reporter in Fig. 3G), which did not reflect an increase in mRNA levels activity generally reflected the endogenous expression pattern (compare Fig. 3G with 3H). Tbn mRNA was detectable at a (Fig. 3, compare C with F). At E3.5 the inner cell mass stained low levels in adult brain (Fig. 1B) and β-galactosidase activity more strongly than the trophectoderm (Fig. 5A). At E6.5, 7.5 was observed in adult brain primarily in the hippocampus and and 8.5, uniform low level Tbn expression and β-galactosidase not in tissues outside the brain. In situ hybridisation of adult activity were observed throughout all tissues of the embryos brain showed low level expression throughout the brain and (E6.5 shown in Fig. 3B, E8.5 shown in Fig. 3C,E). All slightly higher expression in the hippocampus (Fig. 3J). From radioactive in situ hybridisations shown in Fig. 3 were exposed this expression analysis we concluded that the Tbn gene was for 2 weeks. The level of expression of Tbn was low, but could expressed ubiquitously at very low levels throughout be distinguished from background (compare Fig. 3E with 3F). embryonic development. Some tissues like the inner cell mass, 5456 A. K. Voss and others

Fig. 6. The development of mutant embryos after aggregation with wild-type embryos and in explant cultures of E3.5 embryos. (A-C) embryos from Tbngt/+ heterozygous intercrosses were recovered at E2.5, labelled with DiI and aggregated with unlabelled wild-type embryos. (A) Schematic drawing of the experimental procedure; (B) fluorescence image and (C) bright-field image of a chimaeric embryo exhibiting the morphology seen in 98% of the cases. The genotype of the aggregation partner derived from the heterozygous intercrosses was not assessed. The uniform result suggested no differences between mutants and controls in this assay. Note the thorough mixing of labelled and unlabelled cells that was observed in all cases. (D-F) Embryos from Tbngt/+ heterozygous intercrosses were recovered at E3.5 and cultured in ES cell derivation medium. (D) Experimental approach. Normal (E) and mutant (F) morphology, both after 3 days in culture. Note the prominent inner cell mass outgrowth on a patch of trophoblast cells in E, the lack of an inner cell mass outgrowth and dead or dying cells in F, whereas the trophoblast cells are comparable with those in E. Scale bars: 25 µm (B,C), 100 µm (E,F). the developing heart and the adult hippocampus appeared to (Fig. 4B,D). In NIH3T3 cells TBN/6MYC was predominantly generate higher levels of the TBN66/β-gal/neo fusion protein, localized in the cytoplasm (data not shown). possibly reflecting differences in turnover in these tissues. Subcellular localization of the taube nuss protein Tbngt/+ ES cells were stained for β-galactosidase activity. The reporter gene activity was detected in the cytoplasm of most cells and in the nucleus of some cells (Fig. 4A). Nuclear localization of the TBN66/β-gal/neo fusion protein appeared to be correlated with specific cell types growing in monolayers, whereas cytoplasmic localization occurred in cells growing non-contact inhibited. This finding was unexpected, as the TBN66/β-gal/neo fusion protein did not contain the putative nuclear localization signal. As the β-gal/neo fusion protein alone was not transported into cell nuclei, the first 66 amino acids of TBN must have directed transport of the fusion protein into the nuclei possibly by association with a nuclear protein. To investigate where the full-length TBN protein localized, we cloned the protein coding region of Tbn in frame with a six Myc epitopes tag and transfected the fusion construct into COS7 or NIH3T3 cells. As was the case with the TBN66/β- gal/neo, the TBN/6MYC fusion protein was found in the nucleus in some COS7 cells and in the cytoplasm in others

Fig. 7. FGF4 immunocytochemistry of E3.5 and E4.0 embryos of Tbngt/+ heterozygous intercrosses. (A,D,G) Phase contrast, (B,E,H) nuclear stain Hoechst 33258 and (C,F,I) FGF4 immunofluorescence image of the same embryos. Mutant and control embryos were indistinguishable and stained strongly for FGF4 at E3.5 (C). (D-F) Normal morphology and FGF4 staining at E4.0. (G-I) Mutant morphology and FGF4 staining at E4.0. Note that the inner cell mass is clearly visible as a densely packed group of cells in B and E, which stains strongly for FGF4 in the cytoplasm, the lack of most of the inner cell mass in H and the reduced FGF4 staining in I. Scale bars: 27 µm. Tbn, a novel gene essential for the ICM 5457

Fig. 9. TUNEL and activated caspase 3 immunocytochemistry of Fig. 8. OCT4 immunocytochemistry of E3.5 and E4.0 embryos of gt/+ gt/+ E3.75 embryos of Tbn heterozygous intercrosses. (A,D,G) phase Tbn heterozygous intercrosses. (A,D,G) phase contrast, contrast, (B,E,H) nuclear stain Hoechst 33258 and (C,F) TUNEL of (B,E,H) nuclear stain Hoechst 33258, and (C,F,I) Oct4 the embryos in A,B and D,E, respectively. (I) caspase 3 immunofluorescence image of the same embryos Mutant and control immunofluorescence image of the embryo in G,H. Note the presence embryos stained strongly for OCT4 at E3.5 and were of cells staining prominently by TUNEL in F. A cell with condensed indistinguishable (A-C). (D-F) Normal morphology and OCT4 chromatin is indicated with an arrow in H. The same cell stained staining at E4.0. (G-I) Mutant morphology and OCT4 staining at strongly for activated caspase 3 in I (arrow). Scale bars: 15 µm. E4.0. Note that the inner cell mass is clearly visible as a densely packed group of cells in B and E, which stains strongly for nuclear OCT4, the lack of most of the inner cell mass in H and the reduced At E5.0 and E5.5 eight of 23 implantation sites were smaller Oct4 staining in I. Scale bars: 24 µm. and did not contain structures resembling an embryo proper (Fig. 5, compare D with E). Furthermore, at E5.5 they were We concluded from these results that TBN was primarily fragile and ruptured easily (Fig. 5, compare F with G). From localized in the cytoplasm and transported from the cytoplasm this we concluded that, although an inner cell mass was present to the nucleus in some cells, possibly depending on the in Tbngt/gt homozygous embryos at E3.5, it failed to develop functional or developmental state of the cell. beyond E3.5 in vivo. In contrast, trophoblast cells devoid of an The Tbngt/gt mutant phenotype in vivo inner cell mass were visible at E4.0 and elicited a decidual gt response. No differences in the mutant phenotype were The Tbn allele was transmitted through the male (n=50) observed on a mixed genetic background (inbred 129Sv, and the female (n=50) germ line in a 1 to 1 ratio, with the gt C57Bl6, and outbred NMRI) compared with a genetic wild-type allele indicating that both, Tbn spermatozoa and background enriched for the inbred strain C57Bl6. Tbngt oocytes developed normally, were viable, capable of fertilisation and able to contribute to the formation of Chimaeric analysis of the Tbn mutant phenotype heterozygous animals. The Tbngt/+ heterozygous animals were To investigate if Tbngt/gt cells were capable of a normal morphologically normal. Their differential blood cell counts interaction with wild-type cells, E2.5 embryos recovered from and blood chemistry, including their blood glucose levels were Tbngt/+ heterozygous intercrosses were labelled with the normal. fluorescent tracer DiI and aggregated 1:1 with unlabelled wild- A total of 183 offspring of heterozygous intercrosses were type E2.5 embryos (n=35, Fig. 6A). After 24 hours of in vitro analysed for the distribution of the Tbngt allele. Homozygous culture, 98% of the aggregates had formed chimaeric animals or embryos were not detected by Tbn-specific blastocysts, exhibiting thorough mixing of the two aggregation Southern analysis (Fig. 1C) in 89 offspring of Tbngt/+ partners (Fig. 6B,C). The remaining 2% had failed to aggregate heterozygous intercrosses at 3 weeks of age, E15.5, E12.5, so that both embryos formed blastocysts independently. From E9.5 and E8.5 (Table 1). At E6.5 and E7.5, no embryos staining this we concluded that at E2.5 and E3.5 Tbngt/gt cells were not twice the intensity for β-galactosidase activity were observed developmentally compromised and did not segregate from (n=31). At E2.5 and E3.5 21% of the embryos exhibited double wild-type cells. the intensity of β-galactosidase staining as heterozygous To investigate, if the death of the inner cell mass cells was embryos (Fig. 5A, n=63). At E4.0 and E4.5, six of 29 embryos only secondary to a defect in the trophectoderm, we generated were devoid of an inner cell mass (Fig. 5, compare B with C). chimeric concepti consisting of 91 embryos recovered from 5458 A. K. Voss and others

wild type at the Tbn locus is unable to rescue the Tbngt/gt 16 mutant phenotype. The Tbngt/gt mutant phenotype in vitro 14 wild type tbn mutant intercrosses E3.5 embryos recovered from Tbngt/+ heterozygous 12 intercrosses were cultured in ES cell derivation medium (n=97, Fig. 6D). All 97 embryos were either blastocysts at the time of recovery or developed to blastocysts within 12 hours of culture. 10 They all hatched from the zona pellucida and attached to the substrate. 75 embryos (77%) formed prominent inner cell mass 8 outgrowth on a flat patch of trophectoderm cells typical for such cultures (Fig. 6E). The remaining 22 (23%) failed to form 6 an inner cell mass outgrowth. Instead, the inner cell mass cells initially present from the E3.5 embryos plated died over a 4

Number of embryos Number period of 3 days in culture, rounded up and detached from the flat patch of trophectoderm present in all cultures (Fig. 6F). 2 Addition of FGF4 to the culture medium in similar cultures did not result in cell proliferation from Tbn mutant embryos. In 0 cultures of control embryos performed in parallel, all 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 blastocysts formed inner cell mass outgrowths (n=189). Southern analysis showed that the viable inner cell mass Number of TUNEL positive cells per embryo outgrowths were either heterozygous for the gene trap insertion Fig. 10. Numerical distribution of TUNEL-positive cells among Tbn or wild type at the Tbn locus. Heterozygotes and wild type mutant embryos. Numbers of TUNEL-positive cells in 32 embryos were present in a ratio of about 2:1 among the viable cultures recovered at E3.75 from Tbngt/+ heterozygous intercrosses (black) (n=63, 40 and 23, respectively). From these findings in vitro and 36 embryos of wild-type matings (grey). Note that the majority and from the observed phenotypic abnormalities in vivo we (90%) of wild type embryos had 0 to 3 TUNEL positive cells, concluded that TBN is essential for the survival of the whereas the embryos of mutant crosses segregated into three groups, pluripotent inner cell mass cells, but not for the survival of 0 to 3, 5 to 9, and 11 to 15 TUNEL-positive cells per embryo. trophectoderm cells.

Tbngt/+ heterozygous intercrosses and 91 tetraploid embryos Expression of FGF4 and OCT4 in Tbn mutant that were wild type at the Tbn locus. Tetraploid embryos, with embryos and Tbn expression in Oct4 mutant rare exceptions, are incapable of forming the embryo proper, embryos but are capable of forming the extra-embryonic tissues. In a The Tbn mutant phenotype was similar to that found in chimaeric embryo proper, 15% or less of the cells are of embryos lacking the POU domain transcription factor OCT4 tetraploid origin (Nagy and Rossant, 1993), whereas tetraploid (POU5F1 – Mouse Genome Informatics), which fail to form cells can rescue lethal functional insufficiencies of the an inner cell mass. Oct4 mutants show a reduction in Fgf4 trophoblast (Guillemot et al., 1994). If the Tbn mutant expression. Therefore, we investigated if the FGF4 and the trophoblast was the primary site of the defect we would expect OCT4 proteins were present at normal levels in embryos of to be able to generate homozygous E9.5 embryos, if we Tbngt/+ heterozygous intercrosses. In this experiment we supplied them with tetraploid trophoblast. We recovered 25 looked at a mixed population of embryos that were embryos after tetraploid aggregation. None of these, or their homozygous, heterozygous or wild type for the Tbn mutation. yolk sacs were homozygous for the Tbngt allele. Heterozygous At E3.5 we did not observe any changes in immunostaining for and wild type were present in a ratio of about 2:1 (16 and 9, either FGF4 (Fig. 7C) or OCT4 (Fig. 8C). In contrast, at E4.0 respectively). We concluded from this that trophoblast that is one quarter of the embryos showed a reduction in both FGF4 (Fig. 7, compare I with F) and Oct4 (Fig. 8, compare I with F) Table 1. Distribution of the mutant Tbngt allele among immunoreactivity. However, the levels of immunoreactivities offspring of heterozygous intercrosses were normal at E3.5, and the major site of FGF4 and OCT4 production, namely the inner cell mass, was almost absent from Age +/+ L/+ L/L total the E4.0 embryos (see Fig. 7, compare H with E; and Fig. 8, Weaning 19 39 0 58 compare H with E). Therefore, we conclude that the lack of E 15.5 2 4 0 6 E 12.5 4 4 0 8 staining is not due to a regulatory link between Tbn and Oct4 E 9.5 4 7 0 11 or Fgf4, but due to the rather substantial reduction in the E 8.5 2 4 0 6 number of cells producing these factors. E 7.5* 9 9 0 18 In order to investigate if Tbn expression was regulated by E 6.5* 7 6 0 13 Sum≥E 6.5 47 (39%) 73 (61%) 0 120 (100%) OCT4 we analysed expression of Tbn in E3.5 embryos of +/- E 3.5* 9 16 7 32 Oct4 heterozygous intercrosses by radioactive RT-PCR. We E 2.5* 6 19 6 31 amplified β-actin mRNA in the same reaction (n=76). We Sum≤E 3.5 15 (24%) 35 (56%) 13 (21%) 63 (100%) observed that Oct4 mutant embryos expressed Tbn. Those Oct4 *Genotyping by β-galactosidase staining, otherwise by Tbn-specific mutant embryos that already appeared retarded in development Southern analysis at E3.5 showed reduced levels of Tbn mRNA. However, as the Tbn, a novel gene essential for the ICM 5459 changes in morphology were observed at the same time as the a novel gene, taube nuss, and its loss-of-function phenotype in reduction in Tbn mRNA it was not likely that OCT4 regulated mice. The TBN66/β-gal/neo and TBN/6MYC fusion proteins Tbn. The potential presence of maternal Tbn mRNA appeared to be primarily localized in the cytoplasm, although complicates the analysis of a possible regulatory interplay nuclear translocation was observed in specific cell types in between TBN and OCT4. vitro. As the TBN66/β-gal/neo fusion protein had lost its nuclear localization signal, nuclear translocation did not appear The Tbn mutant inner cell mass cells die of to be dependent on the putative nuclear localization signal. It apoptosis may potentially occur through interaction with another protein. To elucidate the mode of death of the inner cell mass cells of TBN was essential for early embryonic development, Tbn mutant embryos we performed terminal deoxynucleotide specifically for the survival of the inner cell mass of E3.75 transferase-mediated dUTP-digoxigenin nick-end labelling blastocysts. The inner cell mass cells of mutant embryos died (TUNEL) on E3.75 embryos recovered from Tbngt/+ of apoptosis. As inner cell mass cells are a transient pluripotent heterozygous intercrosses. In this experiment, we looked at a stem cell population for the embryo proper and some extra- mixed population of embryos that were homozygous, embryonic tissues, an interesting possibility is that TBN might heterozygous or wild type for the Tbn mutation. We observed also be important for the survival of other stem cell populations an increase in TUNEL-positive cells in 22 of 32 embryos, as later in development. However, the early embryonic lethality compared with embryos recovered from intercrosses of animals precluded the study of a requirement of TBN at later stages. wild type at the Tbn locus (n=36). 90% of the embryos from TBN mutant cells participated in the formation of chimaeric wild-type intercrosses had between zero and three TUNEL- preimplantation embryos. The observation that aggregation positive cells. Ten embryos from heterozygous mutant with tetraploid wild-type embryos did not rescue the mutant intercrosses had between zero and three, 15 had five to nine and phenotype indicates that the defect resides in the inner cell seven had 11 to 15 TUNEL-positive cells (Fig. 9F versus 9C; mass and not in the trophoblast. Fig. 10). We interpret the three populations showing zero to Only two proteins, OCT4 and FGF4, have so far been three, five to nine and 11 to 15 apoptotic cells per embryo to identified as being essential for the development of the inner represent wild-type, heterozygous, and homozygous mutant cell mass. The Oct4 mutant embryos fail to develop an inner embryos. The putative group of heterozygous embryos (with cell mass beyond E3.5. However, the Oct4 mutant embryo five to nine apoptotic cells per embryo) showed the tendency of cells are able to survive in culture when supplemented with an increase in rate of mitosis (2.8-fold), which possibly FGF4 and to give rise to extra-embryonic ectoderm-like cells compensated for the increase in apoptosis. The positive (Nichols et al., 1998). In contrast, the Tbn mutant inner cell fluorescence by TUNEL coincided with condensed chromatin mass cells did not survive under the same culture conditions. visualised by Hoechst 33258 staining. A total of 43 embryos At E3.5 the Tbn mutant embryos were morphologically recovered from heterozygous mutant intercrosses were indistinguishable from control embryos, whereas the Oct4 examined for condensed chromatin using the Hoechst stain mutant embryos already appeared to be delayed in between E3.5 and E4.0. While the rate of cells with visible development as they rarely formed expanded blastocysts. This chromatin condensation was normal at E3.5, at E4.0 three of 14 indicates that either the OCT4 protein is required earlier than embryos did not contain a visible inner cell mass, exhibited the TBN protein or that the maternal mRNA and/or protein of massive cell death and only had half the number of total cells Tbn is longer lived than maternal Oct4 mRNA or protein. of the controls (66±9.7 versus 136±13.1; P≤0.01). Furthermore, OCT4 protein is expressed normally in Tbn mutant embryos at fragmentation of cells into membrane-enclosed bodies E3.5, indicating that TBN is not required for Oct4 expression containing DNA, resembling apoptotic bodies, was observed. and, as TBN is expressed in Oct4 mutants, this suggests there To investigate if the Tbn mutant embryos fulfilled other may not be a functional relationship between the two proteins. criteria of death by apoptosis we examined the integrity of the OCT4 appears to be necessary for the maintenance and, cell membrane by Trypan Blue exclusion and found that at probably, also for the establishment of the inner cell mass E3.75 the mutant embryos excluded the dye as did the wild- lineage, but not for cell survival. In contrast, TBN is essential type controls (data not shown). for the survival of the inner cell mass cells. Oct4 mutant To examine the apoptotic pathway involved in the death of embryos were found to express a reduced amount of Fgf4 the Tbn mutant inner cell mass cells, we stained E3.75 embryos mRNA. Fgf4 mutant embryos, similar to Oct4 and the Tbn recovered from Tbngt/+ heterozygous intercrosses with an mutants, fail to develop an embryo proper. However, some of antibody (67341A; Pharmingen) that preferentially recognises the Fgf4 mutant embryos show small disorganised embryonic the activated form of caspase 3. We clearly detected caspase 3- compartments at E5.5 (Feldman et al., 1995). Some Fgf4 positive cells and strong staining coincided with cells showing mutant embryos exhibited some proliferation of the inner cell chromatin condensation (Fig. 9H,I). In addition, we also saw masses in vitro even if severely impaired, whereas Tbn mutant low levels of immunoreactivity in all other cells. This suggests inner cell masses died in culture. This indicates that either that caspase 3 is activated in E3.75 embryos and that Tbn FGF4 is essential for the embryo proper about a day later than mutant embryos may die by apoptosis using a caspase 3- TBN or that maternal FGF4 is available longer than maternal mediated pathway. TBN. FGF4 is expressed in Tbn mutant mice, showing that TBN is not required for FGF4 expression. Apoptosis has been reported previously to occur at a DISCUSSION frequency of 10% or less of the cells in the inner cell masses of blastocysts (Smith and Wilson, 1971; Copp, 1978; In this study we reported the isolation and characterization of Handyside and Hunter, 1986; Pierce et al., 1989; Brison and 5460 A. K. Voss and others

Schultz, 1997). Criteria used to judge apoptosis in blastocyst cell death occurs in the developing heart and is though to be a cells are chromatin condensation and DNA fragmentation. mechanism of shaping this organ. The dying cells exhibit Chromatin condensation and DNA fragmentation can occur via apoptotic morphology (reviewed by Sanders and Wride, 1995). a caspase-dependent or a caspase-independent pathway (Susin An intriguing idea is that TBN may be involved in the et al., 1999). In the caspase-dependent pathway, caspase 3 is regulation of this process. one of last the effector components. It inactivates the inhibitor In conclusion, we report here the identification of a novel of caspase activated DNase (ICAD), which leads to an gene, Tbn, essential for the survival of the inner cell mass cells activation of caspase-activated DNase (CAD) and the of murine blastocysts. TBN is the founding member of a new fragmentation of DNA (Enari et al., 1998). Caspase 3 also group of proteins which show domains highly conserved activates acinus which results in chromatin condensation between chordata, arthropoda, nematoda and streptophyta, (Sahara et al., 1999). We show here that at least some of the suggesting a similar importance of the TBN-related proteins in dying cells appear to contain activated caspase 3. Therefore, these phyla. apoptosis occurring in the inner cell mass may involve activation of caspase 3. Consequently, the caspase-dependent We gratefully appreciate the excellent technical assistance of pathway of apoptosis occurs in blastocysts. This, however, Marion Stäger, Victor Diaz, Gudrun Weinrich and Andrea Conrad. We does not rule out other pathways of cell death also being thank W. Skarnes for providing pGT1.8geo. A. K. V. was supported important. by a fellowship of the Deutsche Forschungsgemeinschaft. T. T. was supported by an EMBO fellowship. This research was funded by As the Tbn mutant inner cell masses died by apoptosis the AmGen Inc. and the Max Planck Society. question arises of whether TBN is directly involved in the regulation of apoptosis or if these embryos die of apoptosis as a consequence of some severe defect directly caused by the REFERENCES absence of TBN, for example, a metabolic defect. As TBN mutant trophectoderm cells survive in culture for at least one Brison, D. R. and Schultz, R. M. (1997). Apoptosis during mouse blastocyst week after the death of the inner cell masses, we can exclude formation: evidence for a role for survival factors including transforming the possibility that TBN is essential for some general cell growth factor alpha. Biol. Reprod. 56, 1088-1096. function. We cannot exclude the possibility that TBN is Chautan, M., Chazal, G., Cecconi, F., Gruss, P. and Goldstein, P. (1999). essential for a process specific for the inner cell mass lineage Interdigital cell death can occur through a necrotic and caspase-independent pathway. Curr. Biol. 9, 967-970. and not required in the trophoblast lineage. Increased incidents Copp, A. J. (1978). Interaction between inner cell mass and trophectoderm of of apoptosis in the inner cell mass have been reported in the mouse blastocyst. I. A study of cellular proliferation. J. Embryol. Exp. blastocysts that developed under suboptimal conditions, for Morphol. 48, 109-125. example in blastocysts isolated from diabetic rats (Pampfer et Coucouvanis, E. and Martin, G. R. (1995). Signals for death and survival: a two-step mechanism for cavitation in the vertebrate embryo. Cell 83, 279- al., 1997) or cultured in vitro (Brison and Schultz, 1997). 287. Alternatively, TBN could be a negative regulator of Duckett, C. S., Nava, V. E., Gedrich, R. W., Clem, R. J., Van Dongen, J. apoptosis. As noted above, apoptosis occurs as a normal part L., Gilfillan, M. C., Shiels, H., Hardwick, J. M. and Thompson, C. B. of preimplantation embryonic development (Smith and Wilson, (1996). A conserved family of cellular genes related to the baculovirus iap 1971; Copp, 1978; Handyside and Hunter, 1986). If apoptosis gene and encoding apoptosis inhibitors. EMBO J. 15, 2685-2694. Enari, M., Sakahira, H., Yokoyama, H., Okawa, K., Iwamatsu, A. and occurs in some cells and not in others, there is presumably a Nagata, S. (1998). A caspase-activated DNase that degrades DNA during mechanism that protects those cells that do not undergo apoptosis, and its inhibitor ICAD. Nature 391, 43-50. apoptosis. TBN could possible be involved in such a Exley, G. E., Tang, C., McElhinny, A. S. and Warner, C. M. (1999). mechanism. Amino acid sequence comparison did not reveal Expression of caspase and BCL-2 apoptotic family members in mouse preimplantation embryos. Biol. Reprod. 61, 231-239. any similarity of TBN or the TBN-related proteins to known Feldman, B., Poueymirou, W., Papaioannou, V. E., DeChiara, T. M. and anti-apoptotic proteins of the Bcl2 family, the inhibitors of Goldfarb, M. (1995). Requirement of FGF-4 for postimplantation mouse apoptosis (IAP, Duckett et al., 1996) or the ICADs. Therefore, development. Science 267, 246-249. in the case of an involvement of TBN in the regulation of Guillemot, F., Nagy, A., Auerbach, A., Rossant, J. and Joyner, A. L. (1994). apoptosis, the TBN-related proteins would constitute a new Essential role of Mash-2 in extraembryonic development. Nature 371, 333- 336. group of proteins involved in this important process. It is Handyside, A. H. and Hunter, S. (1986). Cell division and death in the mouse noteworthy in this context that those of the known anti- blastocyst before implantation. Wilhelm Roux’s Arch. Dev. Biol. 195, 519- apoptotic proteins that have been studied in mutant mice do not 526. appear to be required for the regulation of apoptosis in early Hogan, B., Beddington, R., Constantini, F. and Lacy, E. (1994). Manipulating the Mouse Embryo: A Laboratory Manual. New York: Cold stages of development in blastocysts or during the formation Spring Harbor Laboratory Press. of the proamniotic cavity. Deletion of Bcl2 is compatible Jurisicova, A., Latham, K. E., Casper, R. F. and Varmuza, S. L. (1998). with life after birth at least for some time (Veis et al., 1993). Expression and regulation of genes associated with cell death during murine Bclx−/− mutant embryos die at E13 and exhibit massive cell preimplantation embryo development. Mol. Reprod. Dev. 51, 243-253. death in neural tissues (Motoyama et al., 1995). Bclw-/- mutant King, R. D. and Sternberg, M. J. (1996). Identification and application of the concepts important for accurate and reliable protein secondary structure mice are viable, but show fertility defects (Print et al., 1998; prediction. Protein Sci. 5, 2298-2310. Ross et al., 1998). If TBN was involved in the regulation of Kozak, M. (1989). The scanning model for translation: an update. J. Cell Biol. apoptosis, Tbn would so far be the first gene to be necessary 108, 229-241. to avoid apoptosis in development. Motoyama, N., Wang, F., Roth, K. A., Sawa, H., Nakayama, K., Negishi, I., Senju, S., Zhang, Q., Fujii, S. and et al. (1995). Massive cell death of Interestingly, activity of the reporter gene was high not only immature hematopoietic cells and neurons in Bcl-x-deficient mice. Science in the inner cell mass, where TBN protein is clearly required 267, 1506-1510. for cell survival, but also in the developing heart. Programmed Nagy, A. and Rossant, J. (1993). Production of completely ES cell-derived Tbn, a novel gene essential for the ICM 5461

fetuses. In Gene Targeting. A Practical Approach (ed. A. L. Joyner), pp. Sanders, E. J. and Wride, M. A. (1995). Programmed cell death in 147-178. Oxford: IRL press. development. Int. Rev. Cytol. 163, 105-173. Nichols, J., Zevnik, B., Anastassiadis, K., Niwa, H., Klewe-Nebenius, D., Skarnes, W. C., Moss, J. E., Hurtley, S. M. and Beddington, R. S. (1995). Chambers, I., Scholer, H. and Smith, A. (1998). Formation of pluripotent Capturing genes encoding membrane and secreted proteins important for stem cells in the mammalian embryo depends on the POU transcription development. Proc. Natl. Acad. Sci. USA 92, 6592-6596. factor Oct4. Cell 95, 379-91. Smith, M. S. and Wilson, I. B. (1971). Histochemical observations on early Palmieri, S., Peter, W., Hess, H. and Schöler, H. (1994). Oct-4 transcription implantation in the mouse. J. Embryol. Exp. Morphol. 25, 165-174. factor is differentially expressed in the mouse embryo during establishment Susin, S. A., Lorenzo, H. K., Zamzami, N., Marzo, I., Snow, B. E., of the first two extraembryonic cell lineages involved in implantation. Dev. Brothers, G. M., Mangion, J., Jacotot, E., Costantini, P., Loeffler, M. et Biol. 166, 259-267. al. (1999). Molecular characterization of mitochondrial apoptosis-inducing Pampfer, S. and Donnay, I. (1999). Apoptosis at the time of embryo factor. Nature 397, 441-446. implantation in mouse and rat. Cell Death Differ. 6, 533-545. Thompson, J. D., Higgins, D. G. and Gibson, T. J. (1994). CLUSTAL W: Pampfer, S., Vanderheyden, I., McCracken, J. E., Vesela, J. and De improving the sensitivity of progressive multiple sequence alignment Hertogh, R. (1997). Increased cell death in rat blastocysts exposed to through sequence weighting, position-specific gap penalties and weight maternal diabetes in utero and to high glucose or tumor necrosis factor-alpha matrix choice. Nucleic Acids Res. 22, 4673-4680. in vitro. Development 124, 4827-4836. Vaux, D. L. and Korsmeyer, S. J. (1999). Cell death in development. Cell 96, Pierce, G. B., Lewellyn, A. L. and Parchment, R. E. (1989). Mechanism of 245-254. programmed cell death in the blastocyst. Proc. Natl. Acad. Sci. USA 86, Veis, D. J., Sorenson, C. M., Shutter, J. R. and Korsmeyer, S. J. (1993). 3654-3658. Bcl-2-deficient mice demonstrate fulminant lymphoid apoptosis, polycystic Print, C. G., Loveland, K. L., Gibson, L., Meehan, T., Stylianou, A., kidneys, and hypopigmented hair. Cell 75, 229-240. Wreford, N., de Kretser, D., Metcalf, D., Kontgen, F., Adams, J. M. and Voss, A. K., Thomas, T. and Gruss, P. (1997). Germ line chimeras from Cory, S. (1998). Apoptosis regulator bcl-w is essential for spermatogenesis female ES cells. Exp. Cell Res. 230, 45-49. but appears otherwise redundant. Proc. Natl. Acad. Sci. USA 95, 12424-31. Voss, A. K., Thomas, T. and Gruss, P. (1998a). Compensation for a gene trap Ross, A. J., Waymire, K. G., Moss, J. E., Parlow, A. F., Skinner, M. K., mutation in the microtubule associated protein 4 locus by alternative Russell, L. D. and MacGregor, G. R. (1998). Testicular degeneration in polyadenylation and alternative splicing. Dev. Dyn. 212, 258-266. Bclw-deficient mice. Nat. Genet. 18, 251-256. Voss, A. K., Thomas, T. and Gruss, P. (1998b). Efficieny assessment of the Roth, M. B., Zahler, A. M. and Stolk, J. A. (1991). A conserved family of gene trap aproach. Dev. Dyn. 212, 171-180. nuclear phosphoproteins localized to sites of polymerase II transcription. J. Voss, A. K., Thomas, T. and Gruss, P. (2000). Mice lacking HSP90beta fail Cell Biol. 115, 587-96. to develop a placental labyrinth. Development 127, 1-11. Sahara, S., Aoto, M., Eguchi, Y., Imamoto, N., Yoneda, Y. and Tsujimoto, Weil, M., Jacobson, M. D., Coles, H. S., Davies, T. J., Gardner, R. L., Raff, Y. (1999). Acinus is a caspase-3-activated protein required for apoptotic K. D. and Raff, M. C. (1996). Constitutive expression of the machinery for chromatin condensation. Nature 401, 168-173. programmed cell death. J. Cell Biol. 133, 1053-1059.