Glutamine synthetase is essential in early mouse embryogenesis

RESEARCH ARTICLE

Glutamine Synthetase Is Essential in EarlyMouse EmbryogenesisYouji He,1 Theodorus B.M. Hakvoort,1 Jacqueline L.M. Vermeulen,1 Wouter H. Lamers,1* andMaria A. Van Roon1,2

Glutamine synthetase (GS) is expressed in a tissue-specific and developmentally controlled manner, andfunctions to remove ammonia or glutamate. Furthermore, it is the only enzyme that can synthesizeglutamine de novo. Since congenital deficiency of GS has not been reported, we investigated its role in earlydevelopment. Because GS is expressed in embryonic stem (ES) cells, we generated a null mutant byreplacing one GS allele in-frame with a �-galactosidase-neomycine fusion gene. GS�/LacZ mice have nophenotype, but GSLacZ/LacZ mice die at ED3.5, demonstrating GS is essential in early embryogenesis.Although cells from ED2.5 GSLacZ/LacZ embryos and GSGFP/LacZ ES cells survive in vitro in glutamine-containing medium, these GS-deficient cells show a reduced fitness in chimera analysis and fail to survivein tetraploid-complementation assays. The survival of heavily (>90%) chimeric mice up to at least ED16.5indicates that GS deficiency does not entail cell-autonomous effects and that, after implantation, GS activityis not essential until at least the fetal period. We hypothesize that GS-deficient embryos die when they movefrom the uterine tube to the harsher uterine environment, where the embryo has to catabolize amino acidsto generate energy and, hence, has to detoxify ammonia, which requires GS activity. Developmental Dy-namics 236:1865–1875, 2007. © 2007 Wiley-Liss, Inc.

Key words: glutamine synthetase; deletion; embryonic lethal; blastocyst; mouse

Accepted 8 March 2007

INTRODUCTION

Glutamine is the predominant aminoacid in mammalian plasma, account-ing for approximately 20% of its totalamino-acid content. Glutamine’s prev-alence corresponds to its central rolein nitrogen metabolism, where itserves not only as a temporary storageand transport form of toxic glutamate

and ammonia, but also as a donor ofthe amino moiety of amino and nucleicacids, and hexosamines. Despite themany metabolic pathways in whichglutamine serves as substrate, glu-tamine synthetase (GS; EC 6.3.1.2) isthe only enzyme that can synthesizeglutamine de novo. The importance ofGS can be read from the fact that it isan evolutionary very conserved en-

zyme and it is expressed in a tissue-specific and developmentally con-trolled manner (Meister, 1974; Lie-Venema et al., 1998; van Straaten etal., 2006). Furthermore, a completedeficiency of GS has not been reportedin either human or mouse. Very re-cently, however, two unrelated humanneonates with a considerable congen-ital loss of GS activity, brain malfor-

ABBREVIATIONS ED embryonic day ES cell embryonic stem cell GS glutamine synthetase

1AMC Liver Center and Dept. of Anatomy and Embryology, Academic Medical Center, University of Amsterdam, Amsterdam, TheNetherlands2Animal Research Institute, Academic Medical Center, Facility for Genetically Modified Mice, Amsterdam, The NetherlandsGrant sponsor: NWO; Grant number: SGC210009.*Correspondence to: Wouter H. Lamers M.D., Ph.D., AMC Liver Center and Dept. of Anatomy and Embryology, AcademicMedical Center, University of Amsterdam, Meibergdreef 69-71, 1105 BK, Amsterdam, The Netherlands.E-mail: [email protected]

DOI 10.1002/dvdy.21185Published online 6 June 2007 in Wiley InterScience (www.interscience.wiley.com).

DEVELOPMENTAL DYNAMICS 236:1865–1875, 2007

© 2007 Wiley-Liss, Inc.

mations, multiorgan failure, and in-fant death were described (Haberle etal., 2005), but whether these findingsare causally related remains to be es-tablished. In agreement with these vi-tal facts, we have established for thefirst time that the complete knockoutof the mouse GS gene results in earlyembryonic lethality. Although GS-de-ficient embryonic cells survive in vitroin glutamine-containing medium, GS-deficient embryonic stem cells exhibita reduced fitness if transplanted intowild-type blastocysts, and fail to sur-vive in aggregates with tetraploidwild-type blastomeres.

RESULTS

Generation of GS�/LacZ Mice

The GS gene is transcriptionally ac-tive in E14IB10 ES cells (Fig. 1D).Therefore, we were able to generateGS�/LacZ ES cells by a promoter trapstrategy, in which the GS coding se-quence and 3�UTR of one allele werereplaced in-frame with a �-galactosi-dase-neomycin-resistance fusion prod-uct via homologous recombination.Correct targeting was confirmed byRT-PCR using a forward primer in GSexon I upstream of the targeting con-struct and reverse primers in LacZand GS, respectively (Fig. 1D). Thesame strategy was used to produceGSGFP/� ES cells, in which the GScoding sequence is replaced by the Re-nilla green-fluorescent protein (notshown).

Two mouse lines, obtained from twoindependent GS�/LacZ ES clones, werephenotypically normal and fertile. Webackcrossed the mice to the FVB back-ground for �10 generations. Althoughthe reduction in activity did not reachsignificance in all organs, the elimina-tion of one functional allele (GS�/LacZ

mice) resulted on average in a 50%reduction of enzyme activity whencompared to wild-type (Table 1). Onlythe colon and stomach of GS�/LacZ

mice retained approximately 70% ofthe activity measured in GS�/� mice.

We did not observe any phenotypicabnormalities in GS�/LacZ mice. Or-gans of GS�/LacZmice were analyzedfor expression of �-galactosidase andthe endogenous GS protein. Figure 2shows overall co-localization of endog-enous GS and �-galactosidase in thepericentral hepatocytes in the liver

Fig. 1. Targeted disruption of the mouse GS gene. Schematic diagram is drawn to scale. A: Uponhomologous recombination, the LacZ targeting vector replaces exon II from the translation-start site toBamHI 800 bp downstream of exon VII with a �-galactosidase (LacZ)/neomycin phosphotransferase(NEO) fusion gene. B: In GSGFP/� ES cells, one GS allele was replaced in-frame by a GFP-laminA fusiongene. These ES cells were used to generate GSGFP/LacZ ES cells. GS exons are shown as red boxes. Thebent arrow indicates the translation-start site. 763, position in mouse GS gene relative to the transcrip-tion-start site; B and E, restriction enzymes BamHI and EcoRI; triangles 1–13, oligonucleotide-primersites listed in Table 4. C: The genotypes of ED3.5 embryos from GS�/LacZ intercrosses as determinedby PCR analysis (solid triangles, wild-type locus; open triangles, targeted locus. �/�, wild-typeembryos; �/�, GS�/LacZ embryos; �/�, GSLacZ/LacZ embryos. D: RT-PCR analysis of ES clones: �/�,wild-type ES cells; �/�, GS�/LacZ ES cells; �/�, GSGFP/LacZ ES cells.

Fig. 2. Co-expression of GS and LacZ in GS�/LacZ mice. Serial sections of organs of 2-month-oldGS�/LacZ mice were stained immunohistochemically for the presence of GS (A, C, E, G) and�-galactosidase protein (B, D, F, H). The expression of �-galactosidase and the endogenous GSshowed co-localization in the pericentral hepatocytes of the liver (A, B), in the astrocytes of thecerebellum (C, D), in the enterocytes of the crypts of the small intestine (E, F), and in the S3 segmentof the proximal contorted tubules of the kidney cortex (G, H). C, central vein; Co, cortex; OS, outerstripe; Me, medulla. Scale bar � 200 �m (A–F).

1866 HE ET AL.

(Fig. 2A and B), the astrocytes in thebrain (the cerebellum is shown in Fig.2C and D), the enterocytes in the in-testine (mainly in the crypts and gob-let cells; Fig. 2E and F), and in the S3segment of the proximal contorted tu-bules of the kidney cortex (Fig. 2G andH); see van Straaten et al. (2006) for amore extensive description of GS ex-pression. From these data, we con-clude that �-galactosidase expressionin GSLacZ mice can be used as amarker for GS expression.

GS Knockout Mice AreEmbryonically Lethal(Table 2)

In a total of 95 offspring (3 weeks old)from heterozygous intercrosses, 33(35%) were wild type and 62 (65%)were heterozygous, but no GSLacZ/LacZ

homozygotes were found (Table 2).Similarly, no mice with GSLacZ/LacZ

genotype were detected as neonates oras ED9.5 embryos. Interestingly, atED3.5 (blastocyst stage), 9 out of 125

embryos (7%) were genotyped asGSLacZ/LacZ (Fig. 1C), which is still sig-nificantly lower than the expected25% according to Mendelian inheri-tance (Chi-square test: P � 0.00002;Table 2). These results show thatGS-null embryos die prior to implan-tation. Figure 3A and B shows thestaining of wild-type blastocysts inmouse uterus for the presence of GS.GS is expressed both in the inner cellmass and in the trophectoderm. Theexpression pattern was consistentwith the results of whole mount X-galstaining for �-galactosidase expres-sion in GS�/LacZ/ GSLacZ/LacZ morulaeand blastocysts (Fig, 3D and E). Fig-ure 3F shows a blastocyst with thezona pellucida still intact, but withthe inner cell mass displaying decay-ing cells with a few remaining blue-staining, LacZ-positive spots. GS-nullmutants were never found after ED3.5(Table 2), which indicates GS expres-sion is required and essential in vivoat this early embryonic stage.

GSLacZ/LacZ Embryonic CellsCan Be Rescued byGlutamine Supplementation

To investigate if the early embry-onic lethality associated with theGSLacZ/LacZ genotype was confined toembryos and would not occur if em-bryonic GSLacZ/LacZ cells were givenaccess to external glutamine, we iso-lated embryos from heterozygous in-tercrosses at ED2.5. After removal ofthe zona pellucida, the specimenswere cultured in medium with 2 mML-glutamine for 6 days, prior to ge-notype analysis. A total of 48 em-bryos including 43 morulae and 5blastocysts were cultured. In thefirst two days of culture, most ofthem still kept their shapes asspheres and some of them developedinto blastocysts with the tropho-blasts lying outside. Thereafter, theyflattened out, attached to the cultureplates, and developed outgrowths,including distinct cell clumps on topof a monolayer of the trophoblast gi-

TABLE 1. GS Enzyme Activity in Wild-Type (GS�/�) and Heterozygous (GS�/LacZ) Micea

Organ

GS�/� GS�/LacZ

P valuen Mean � SEM n Mean � SEM

Epididymis 4 5623 � 271 7 2738 � 137 2.0E-07Brain (cerebrum) 8 1234 � 55 14 786 � 15 4.8E-09Brain (remainder) 8 910 � 54 14 526 � 21 1.8E-07White adipose tissue 8 989 � 83 14 482 � 52 2E-04Brown adipose tissue 8 794 � 22 14 435 � 32 1.6E-07Liver 8 717 � 21 14 397 � 12 1.4E-05Testis 4 337 � 12 7 130 � 6 5.8E-05Stomach-distal 8 328 � 37 14 228 � 17 0.033Kidney 8 179 � 6 14 99 � 9 7.9E-07Colon 8 62 � 11 14 45 � 3 0.177Muscle 8 18 � 4 14 8.6 � 22 0.063Jejunum 8 5.1 � 2.0 14 1.5 � 0.5 0.035

aEnzyme activity was measured as -glutamyltransferase activity and expressed as nmol per minute per mg protein at 37°C.

TABLE 2. Distribution of Genotypes of Offspring From GS�/LacZ Intercrossesa

Age Total no.

No. offspring with genotype (%)

�/� �/� �/�

At weaning* 95 33 (35) 62 (65) 0Neonates** 48 15 (31) 33 (69) 0ED 9.5*** 21 7 (33) 14 (67) 0ED 3.5**** 125 40 (32) 76 (61) 9 (7)

aThe chi-square test was used to examine if the observed number of offspring with the �/� genotype differed from the expectedMendelian inheritance distribution. *P � 0.0000001; **P � 0.0003; ***P � 0.03; ****P � 0.00002.

GS DEFICIENCY IN EARLY EMBRYOS 1867

ant cells (Fig. 4). For 29 of them, agenotype analysis was possible: 13were wild type, 12 GS�/LacZ, and 4GSLacZ/LacZ (14%). No growth or mor-phological differences were observedamong the three groups of embryoniccell cultures. The results demon-strate that ED2.5 GS-null mutantcells derived from the trophoblast(the giant cells) and from the em-bryoblast (the small cells) were ca-pable of proliferation and could notbe distinguished from either GS�/LacZ

or GS�/� embryos during in vitro cul-ture in the presence of glutamine.

Generation of GSGFP/LacZ

Chimeras

The conclusion that glutamine supple-mentation allows GSLacZ/LacZ embry-onic cells to survive implies thatGSLacZ/LacZ cells should also survive inchimeric embryos. In other words, wepredicted that GS deficiency wouldnot confer a cell-autonomous inabilityto survive in vivo. In order to obtainGS�/� ES cells, we disrupted the re-maining GS allele in GSGFP/� ES cellswith the �-galactosidase-neomycin re-sistance fusion gene, that is, the LacZtargeting vector. Among 64 neomycin-resistant clones obtained, we found 34(55%) homologous recombinations, ofwhich 17 were of the GS�/LacZ and 17of the GSGFP/LacZ genotype (deter-mined by long PCR on both ends, datanot shown). The equal distribution ofthese 2 genotypes indicates thatGSGFP/LacZ ES and GS�/LacZ cloneshave the same viability. We continuedwith three GSGFP/LacZ ES clones, allwith no detectable GS mRNA by RT-PCR (Fig. 1D).

Wild-type blastocysts injected withGSGFP/LacZ ES cells were implantedinto pseudopregnant females and al-lowed to develop. GS�/LacZ ES cellsserved as controls. The LacZ reportergene in both GSGFP/LacZ and GS�/LacZ

ES cells was used to trace ES cell-derived cells in chimeric embryos. Ta-ble 3 shows that injection of 194 blas-tocysts with GSGFP/LacZ cells resultedin 36 (19%) live embryos, whereas injec-tion of 279 blastocysts with GS�/LacZ

cells resulted in 71 (25%) live offspring.Of the GSGFP/LacZ-injected blastocysts,only 9 (5%) contained recombinant cellsat harvest, whereas this number was 31(11%) for the GS�/LacZ-injected blasto-

cysts. The approximately 2-fold lowersurvival of GSGFP/LacZ cells compared toGS�/LacZ cells in blastocysts in vivo wasstatistically significant (P � 0.013; Ta-ble 3) and demonstrates that the capa-bility to generate viable chimeras is

lower for GSGFP/LacZ than for GS�/LacZ

ES cells. In other words, GSGFP/LacZ EScells have a 2-fold lower chance to sur-vive in host blastocysts than GS�/LacZ

ES cells. Figure 5A–F shows wholemount �-galactosidase (X-gal) staining

Fig. 4. In vitro cultivation of ED2.5 embryos. E2.5 embryos were isolated from GS�/LacZ inter-crosses and cultured in 96-well plates. Pictures were taken on a daily basis with a Leica DMIRE2inverted microscope. ICM, inner cell mass; TGC, trophoblast giant cells. Scale bars � 50 �m.

Fig. 3. LacZ and GS expression in blastocysts. A–C: Serial sections of wild-type blastocystscollected in the cavity of an adult uterus were stained for the presence of GS protein. GS is alsoexpressed in the endometrial epithelium (van Straaten et al., 2006). A1–B2: Higher magnificationsof the embryos indicated by arrows in A and B. C: Control incubation of serial section of B.D–F: ED3.5 embryos from GS�/LacZ intercrosses were analyzed by whole-mount staining for�-galactosidase activity. D: �-galactosidase activity in a morula stage. E: Staining in both the innercell mass and trophectoderm. F: A dying blastocyst. Red arrows, blastocysts; ICM, inner cell mass;TE, trophectoderm; ZP, zona pellucida. Scale bars � 100 �m (A–C); 25 �m (A1–F).

1868 HE ET AL.

of ED12.5 and ED16.5 embryos. Thechimeric embryos displayed a range of alow- to a high-degree of chimerism,judged by X-gal staining (Fig. 5A–C),and were comparable in GSGFP/LacZ andGS�/LacZ chimeric embryos. Histologi-cal analysis of ED12.5, ED16.5, andpostnatal day 4 GSGFP/LacZ chimerasdid not reveal any obvious morphologi-cal abnormalities, even when a high de-gree of chimerism was present (Fig. 6).Furthermore, the morphological fea-tures of the limb buds, gut, neuronalstructures, and vasculature were ap-propriate for the stage of development(Rugh, 1990). GSGFP/LacZ-derived LacZ-positive cells were found in nearly alltissues/organs examined, includingbrain, choroid plexus, spinal cord, sali-vary glands, heart, lung, liver, stomach,pancreas, small intestine, colon, kidney,and adrenal (Fig. 6). This distribution ofLacZ-positive cells is consistent withthe normal distribution of GS-positivecells in non-chimeric littermate em-bryos (e.g., Fig. 6H); see also previousstudies (Lie-Venema et al., 1997a,b).Immunohistochemistry on serial sec-tions of chimeric embryos showed thatthe GS-positive wild-type cells (GS�/�)had a complementary distribution pat-tern to LacZ-positive cells (GSGFP/LacZ)(e.g., Fig. 6B vs. C, E vs. F, and D vs. G).The contributions of mutant cells inmany different lineages clearly showedthat the consequences of GS deficiencywas either not detrimental in many or-gans or was compensated for by sur-rounding wild-type cells.


“ES” Embryos

After having demonstrated the survivalof GSGFP/LacZ chimeras to the neonatalstage as opposed to the early embryoniclethality of GS-null mice (ED3.5), wetested the viability of embryos gener-ated entirely from GS-deficient ES cells.For this, we used the ES-tetraploidcomplementation approach, in whichthe wild-type tetraploid cells are ex-cluded from the embryonic lineages, butsupport inner cell mass development ofcompromised ES cells (Nagy et al.,1993). Figure 7A shows that, judging bythe number of decidual reactions, thereis no difference in implantation effi-ciency between GSGFP/LacZ and GS�/�

ES-embryos at ED9.5 (16 out of 371 �4.3% and 8 out of 298 � 2.7%; P � 0.26).However, we did not recover any viableGSGFP/LacZ ES embryo at ED9.5 (0 outof 16), which is significantly lower thanof the control group (2 out of 8, 25%)(P � 0.037). We noticed that one im-plantation in the GS-deficient group re-sulted in a dead embryo, in which theheart had developed to ED8.5 (Fig. 7B1;staging according to Theiler, 1983).PCR genotype analysis of this hearttube and the corresponding placentaltissues showed contribution of both GS-GFP/LacZ and tetraploid cells to both or-gans (Fig. 7B2).

DISCUSSION

We report the generation of a consti-tutive GS deletion in the mouse and

established that GS already becomesessential in the blastocyst stage of em-bryonic development. If this findingcan be extrapolated to human devel-opment, it explains why a completedeficiency of GS has not yet been re-ported.

To facilitate the analysis of GS de-ficiency in developing mice, we re-placed the coding sequence of the GSgene with that of the LacZ reportergene. Both our histochemical and bio-chemical data showed that this re-porter gene accurately and reliablymimicked the spatio-temporal expres-sion pattern of the GS gene. Immuno-histochemical analysis of GS�/LacZ

mice showed overall co-localization of�-galactosidase and endogenous GS,which shows that introns 2–6 and theGS 3� UTR do not affect the pattern ofexpression of GS. GS�/LacZ miceshowed on average a 50% reduced GSenzyme activity in different organswhen compared to wild-type mice,showing that both GS alleles are tran-scriptionally active. This finding is inapparent contrast with that in two hu-man infants, in whom mutations inthe C-terminus of GS strongly de-creased specific enzyme activity andincreased cellular GS protein content(Haberle et al., 2005). This differenceprobably reflects the fact that we com-pletely eliminated expression fromone allele, whereas the point muta-tions in the human patients allowedexpression of the affected allele andperhaps even stabilized the resulting

TABLE 3. Summary of GSGFP/LacZ and GS�/LacZ Mouse Chimera Productiona

Implanted Recovered

ES clones No. blastocystsEmbryonicstage

No. non-chimericembryos

No. chimericembryos

GSGFP/LacZ 30 ED 12.5 2 3130 ED 16.5 21 434 Postnatal 4 2

Total 194 27* 9**

GSGFP/LacZ 15 ED 12.5 0 659 ED 16.5 8 6205 Postnatal 32 19

Total 279 40* 31**

aA Chi-square test was used to compare the recovery of blastocysts injected with GSGFP/LacZ or GS�/LacZ ES cells and containing(“chimeric”) or not containing (“non-chimeric”) descendants of these cells at recovery. *P � 0.9 (27 out of 194 and 40 out of 279);**P � 0.013 (9 out of 194 and 31 out of 279).


protein. In any case, our findings donot appear to support the hypothesisthat a decrease in cellular GS activityinduces an increased expression of thegene to compensate for the loss in en-zyme activity.

Viability of GS-DeficientCells

We did show that ED2.5 GSLacZ/LacZ

embryonic cells survived in vitro, ifthey were provided with glutamine(Fig. 4). Similarly, the equal numberof GS�/LacZ and GSGFP/LacZ ES clonesthat were obtained during targeting

Fig. 5.

Fig. 6.

Fig. 7.

Fig. 5. Whole-mount staining of GSGFP/LacZ

chimeras for �-galactosidase activity. Embryoswere recovered at ED12.5 (A–D) and ED16.5 (E,F). A–C and E: Embryos containing GSGFP/LacZ

cells. D and F: Non-chimeric littermates. Anincision into the anterior abdominal wall wasmade to allow penetration of the fixative andX-gal staining solution. Scale bars � 2 mm.

Fig. 6. The in vivo distribution of GSGFP/LacZ

cells in ED 16.5 chimeras. Serial cryostat sec-tions of an ED16.5 GSGFP/LacZ chimera (A–G)and a non-chimeric littermate (H) were stainedfor �-galactosidase enzyme activity and GSprotein. In the chimera (A, D), GSGFP/LacZ cells(blue) were present in brain, spinal cord, sali-vary gland, heart, lung, liver, stomach, pan-creas, small intestine, colon, kidney, adrenal,and muscle. B,C: A magnification of the boxedarea in A showing the choroid plexus hanging inthe 4th brain ventricle. E,F: A magnification ofthe boxed area in D showing the liver. Thechimeric embryo shows a complementarystaining pattern of �-galactosidase and GS (Bvs. C; E vs. F; D vs. G), whereas the non-chimeric embryo (H) shows a normal distribu-tion of GS-positive cells. A, adrenal; C, cerebel-lum; Co, colon; H, heart; I, small intestine; K,kidney; L, liver; Lu, lung; M, muscle; Me, mes-encephalon; My, myelencephalon; P, pancreas;R, rhinencephalon; S, salivary gland; St, stom-ach; SC, spinal cord; T, tongue; Te, telenceph-alon; V, vertebrae. Scale bars � 500 �m(A,D,G,H); 100 �m (B,C); 25 �m (E,F).

Fig. 7. Survival of GSGFP/LacZ “ES” embryos atED9.5. Aggregates of GSGFP/LacZ or GS�/� EScells and tetraploid 4-cell stage B6D2F1/J@Rjembryos were introduced into the uterus ofpseudopregnant mice. A: Implantation fre-quency estimated as the number of decidualreactions in the uterus and survival at ED9.5. Achi-square test was used to compare the effi-ciency of implantation and survival: “a”: P �0.26 (16 out of 371 and 8 out of 298) and “b”:P � 0.037 (0 out of 16 and 2 out of 8). B: Heartof dead GSGFP/LacZ ES embryo (B1; develop-mental stage: ED8.5) that contained both GS-GFP/LacZ cells (LacZ-positive) and tetraploidwild-type cells (GS-positive; B2). H, heart DNA;p, placental DNA of the same embryo.

the second GS allele in GS�/GFP EScells was in agreement with the pre-diction that the LacZ-targeting con-struct had the same chance of target-ing the wild-type or GFP-knockinallele if both GS heterozygous andknockout cells would survive in glu-tamine-containing medium. More-over, we observed that GS�/� did notbehave in a cell-autonomous mannerin chimeric embryos (Figs. 5 and 6).Our finding that GS is dispensable forcells that have access to external glu-tamine differs from the apparent in-ability to culture fibroblasts derivedfrom two congenitally GS-deficienthuman infants (Haberle et al., 2005).Whatever the reason for this latterobservation, our culture data do indi-cate that GS-deficient embryos diedue to the metabolic effects of GS ac-tivity, that is, insufficient access toglutamine or toxicity of ammonia orglutamate.

Efforts to grow GS-deficient em-bryos by maintaining them in an en-vironment that would presumablytake care of their glutamine supply ortheir ammonia detoxification showedthat the environment in the blastocystdiffers profoundly from that in thePetri dish. Thus, we noted thatGSGFP/LacZ ES cells had an approxi-mately 2-fold reduced chance to gen-erate viable chimeras compared toGS�/LacZ ES cells (Table 3). Further-more, we noted that none of the ESembryos made by fusing tetraploidwild-type blastomeres and diploid GS-deficient ES cells survived to ED9.5.In such “ES” embryos, the trophecto-derm and hypoblast derive from wild-type tetraploid cells, whereas thesetetraploid cells should not thrive inthe epiblast lineage (Nagy et al.,1993). Together, these data indicatethat embryos with a GS-deficient in-ner cell mass do not survive, even ifGS is present in the surrounding (tet-raploid) trophectoderm. Because wewere able to produce embryos withboth a high and a low degree of chi-merism for GS-deficient cells—oneembryo with �90% blue cells in GS-expressing tissues, including thebrain, survived to ED16.5 and did notdevelop the brain malformations thatwere found in GS-deficient human in-fants (Haberle et al., 2005) (Fig. 6)—only a few wild-type cells were appar-ently necessary to allow a chimeric

embryo to survive. We, therefore, at-tribute the recovery of a single, se-verely retarded ED8.5 ES embryo tothe fact that its inner cell mass didharbour tetraploid wild-type cells(Fig. 7). The participation of wild-typetetraploid cells in the formation of ES-embryos has been reported and canamount to 10–50% of the embryoscells in liver, lung, and heart ofED18.5 embryos (Wang et al., 1997).

What Causes Death in GS-Deficient Embryos at ED3.5?

Table 2 clearly shows that GS�/� em-bryos are dying off at ED3.5, that is,when the embryo moves from the ovi-duct into the uterine lumen and tran-sits to the blastocyst stage of develop-ment. This finding demonstrates thatGS enzyme activity becomes essentialin vivo at this early embryonic stage.In accordance, our immunohistochem-ical data show that GS is expressed inboth the inner cell mass and the tro-phectoderm of mouse blastocysts (Fig.3). The question, therefore, ariseswhether substrate toxicity (ammoniaor glutamate) or product shortage(glutamine) causes embryonic death.Major changes in the environment ofthe embryo when it moves from theoviduct into the uterus are a 2-folddecline in the concentration of glucoseand most amino acids (Harris et al.,2005). In view of the fact that glucoseis the major energy source of the blas-tocyst (Martin and Leese, 1999), thedecline in ambient glucose concentra-tion in passing from the oviduct to theuterus is striking and suggests thatalternate energy sources are used,with amino acids as obvious candi-dates. In this respect, the 3-fold de-cline in concentration of ambient glu-tamine (Harris et al., 2005) isremarkable, because it is reportedthat the addition of amino acids, inparticular glutamine, to the culturemedium of embryos in vitro improvestheir development (Rezk et al., 2004).

The blastocyst consumes almost ex-clusively glutamate, aspartate, serine,arginine, and glycine (together ap-proximately 6.5 pmol/embryo.hr) andproduces alanine and glutamine (to-gether approximately 2 pmol/em-bryo.hr) (Houghton, 2006). Assuming10 ng protein/embryo and 2.5 celldivisions/day, it can be calculated that

approximately 3 pmol/embryo.hr isused for protein synthesis in the blas-tocyst. Furthermore, ammonia pro-duction in mouse blastocysts amountsto approximately 1 pmol/embryo.hr(Gardner and Lane, 1993). This pro-duction of ammonia does not affectembryo development, but further ac-cumulation has detrimental effects onmany aspects of embryo physiology,including reduced blastocyst cell num-ber, decreased development of the in-ner cell mass, perturbed metabolism,impaired intracellular pH regulation,and altered gene expression (Laneand Gardner, 2003). Because glu-tamine and alanine are well-knownamino-carriers, their net productionimplies that both alanine aminotrans-ferase and glutamine synthetase ap-parently function to avoid the accu-mulation of toxic amounts of freeammonia due to amino-acid catabo-lism. The data, therefore, indicatethat glutamine is used as an amino-carrier rather than a fuel in the mouseblastocyst and that the alanine ami-notransferase route alone does notsuffice to detoxify all the ammoniaproduced. In aggregate, these consid-erations favor ammonia accumulationas the cause of death of the GS-defi-cient blastocysts. Our findings, there-fore, question the benefits of addingextra glutamine to embryo culturemedia (Biggers et al, 2004; Rezk et al.,2004).

Conclusion

GS-deficient mice die at ED3.5, un-equivocally demonstrating that GS car-ries out an essential function at thatdevelopmental stage. We hypothesizethat the transition from the tube intothe harsher uterine environment forcesthe embryo to catabolize amino acids togenerate energy and that the resultingammonia production requires GS activ-ity for detoxification. The survival ofheavily chimeric mice up to at leastED16.5 suggests that after implanta-tion, GS activity is not essential until atleast the fetal period.

EXPERIMENTALPROCEDURES

Mice

Mice were maintained on a 12-hrlight/12-hr dark cycle with free access


to water and food. The duration ofpregnancy was counted from the de-tection of a vaginal plug, which wasset at 0.5 days. The studies were car-ried out in accordance with the Dutchguidelines for the Care and Use ofLaboratory Animals. Handling of em-bryonic stem (ES) cells and embryoswas done as described (Hogan et al.,1994).

Construction of the LacZTargeting Vector

Fourteen kilobase of the full GS ge-nome (Glu1, MGI: 95739) of mousestrain 129/Ola, from position 763 (rel-ative to transcription start site) in in-tron 1 to the EcoRI site 5 kb down-stream of exon VII, was subclonedfrom a genomic clone (no. 7026 in P1plasmid; Genome Systems Inc., St.Louis, MO). Because GS is expressedin ES cells (see Results section), wedesigned a targeting vector with pro-moter trap selection to disrupt themouse GS gene (Fig. 1A) (Skarnes etal., 1992, 1995; Evans et al., 1997;Friedrich and Soriano, 1991). PlasmidpB�geopA, containing a fusion prod-uct of �-galactosidase and the neomy-cin resistance gene driven by the hu-man �-actin promoter (Skarnes et al.,1995; Friedrich and Soriano, 1991),

was modified by replacing the human�-actin promoter with that of mouseGS, using the NcoI site for a correctin-frame fusion of the GS and theLacZ gene. Furthermore, the SV40pAsite was replaced by the bovinegrowth-hormone polyadenylation sig-nal, taken from pcDNA3 (XbaI andPvuII) (Invitrogen, Breda, The Neth-erlands). For correct targeting, theSalI-NcoI (2.2 Kb) upstream and theBamHI-EcoRI (4.3 Kb) downstreamfragments of the GS-gene were used.

Generation of GS�/LacZ ESCells

To generate the GS�/LacZ ES cell line,the linearized lacZ targeting construct(25 �g, Fig. 1A) obtained by restrictionenzyme digestion with Swa I was elec-troporated into 1.3*107 ES cells (lineE14IB10 derived from the 129P2/OlaHsd mouse strain). Out of a total of39 neomycin-resistant clones, 29 clones(74%) were correctly recombined as de-duced for their expression of the GS-LacZ fusion gene by RT-PCR.


(GS-Null) ES Cells

The GSGFP/LacZ ES cell line, whichdoes not contain any functional GS

allele, was generated from GS�/GFP

ES cells (line E14IB10; the construc-tion of GS�/GFP ES cells will be de-scribed in detail elsewhere), in whichone GS allele was replaced in-frameby a GFP-laminA fusion product viahomologous recombination (Fig. 1B).The second allele was targeted withthe LacZ targeting construct as de-scribed above. A total of 64 neomycin-resistant ES cell clones were obtained.All colonies were genotypicallyscreened by PCR for the presence ofGS, GFP, and LacZ sequences. Clonespositive for both GFP and LacZ, andnegative for GS, were subsequentlyscreened for proper recombination bymeans of a long PCR of the 5� and 3�ends of the construct (Expand longtemplate PCR kit, Roche, Almere, TheNetherlands). For the 5�-end, a for-ward primer in GS intron 1, upstreamof the targeting construct, and reverseprimers in GFP and LacZ gene wereused. For the 3�-end, forward primersin laminA and LacZ, and reverseprimers in the GS gene downstream ofthe corresponding targeting con-structs were used. All primer se-quences are listed in Table 4. The ESclones containing the correct homolo-gous recombination were thenchecked for expression of GS-LacZ and

TABLE 4. Primer Sequences, Annealing Temperature, and Expected PCR Product Lengths

Product Primers Sequence 5� 3 3�

AnnealingTemperature (°C)

Fragment(bp)

GS 1) GS-F 9154 CTTACTCCACACACGAGTT 53 4012) GS-R 9574 ACACAAACAACCAGAAAACT

LacZ 3) LacZ-F GCATCGAGCTGGGTAATAAGCGTTGGCAAT 69 2554) LacZ-R ACTGCAACAACGCTGCTTCGGCCTGGTAAT

GFP 5) GFP-F GTGAGCAAGCAGATCCTGAA 58 1366) GFP-R GTTGCCGAACAGGATGTTGC

GS-GFP 7) GS-F 722 TCTACCCTCGCCGGGGTA 61 2,4226) GFP-R GTTGCCGAACAGGATGTTGC

5�GS-LacZ 7) GS-F 722 TCTACCCTCGCCGGGGTA 63 5,3814) LacZ-R ACTGCAACAACGCTGCTTCGGCCTGGTAAT

LaminA-3�GS 8) Lam-F CCATCTCCTCTGGCTCTTCT 60 3,9609) GS-R 10748 GATACGCAGAGATACCATCA

LacZ-3�GS 3) LacZ-F GCATCGAGCTGGGTAATAAGCGTTGGCAAT 50 5,65010) GS-R 14851 GTGTGATGGCTCATGGAATG

GS-LacZ 11) GS-F 54 TCCTCTCCGCCTCGCTCTC 56 12412) LacZ-R Xi ACAACGTCGTGACTGGGAAAACC

GS-GFP 11) GS-F 54 TCCTCTCCGCCTCGCTCTC 58 2076) GFP-R GTTGCCGAACAGGATGTTGC

GS cDNA 11) GS-F 54 TCCTCTCCGCCTCGCTCTC 57 18413) GS-R 3115 CCGGTACCATCAACCCAG

GAPDH GAPDH-F TTCCTACCCCCAATGTGTC 57 255GAPDH-R AGCCGTATTCATTGTCATACC

1872 HE ET AL.

GS-GFP fusion mRNA and lack of GSmRNA by RT-PCR.

Generation of GS�/LacZ Mice

After karyotyping, the GS�/LacZ ESclones were injected into C57BL/6Jblastocysts, which were then trans-ferred to pseudopregnant B6D2F1/OlaHsd (Harlan, The Netherlands) fostermothers. To generate GS�/LacZ mice,chimeric male mice were bred withFVB/NHanHsd females (Harlan, TheNetherlands). Germ-line transmissionwas achieved for two independentclones. The GS�/LacZ mice werecrossed into the FVB background for10 generations. Offspring were geno-typed by PCR, using primers for theGS and LacZ genes (Table 4 and Fig.1). Embryos at embryonic day (ED) 3.5(blastocyst stage) and ED9.5 were iso-lated from heterozygous GS�/LacZ in-tercrosses and genotyped. Blastocystswith a minimal amount of adheringmedium were sampled individuallyinto 200-�L PCR reaction tubes, towhich 5 �L 50 mM KOH was added.The samples were lysed at 95°C for 5min, cooled to room temperature, neu-tralized with 5 �L 50 mM Tris-HCL(pH 4.0). Ten microliters of the mix-ture was used for PCR genotype anal-ysis.


Chimeric Embryos

After karyotyping, the GSGFP/LacZ ESclones were injected into C57BL/6Jblastocysts, which were then trans-ferred to pseudopregnant B6D2F1/OlaHsd foster mothers. ChimericGSGFP/LacZ embryos were collected atED12.5, ED16.5, and postnatal day 4.They were identified by whole-mountX-gal staining or by PCR-analysisboth for LacZ and GFP. The distribu-tion of GSGFP/LacZ and wild-type EScells in organs and tissues was furtherexamined by a combination of histo-chemistry (�-galactosidase stainingwith X-gal) and immunohistochemis-try (anti-GS primary antibody) on cry-ostat sections or by immunohisto-chemistry with anti-�-gal and anti-GSprimary antibodies on paraffin-em-bedded sections (as described below).


“ES” Embryos

The tetraploid complementation ap-proach was used to generate GSGFP/LacZ

“ES” embryos (Eggan et al., 2001; Nagyet al., 1990). Tetraploid morulae werederived from one-cell tetraploidB6D2F1/J@Rj embryos by culturingthese for 24–30 hr after electrofusion.Preparation of GSGFP/LacZ ES cells andaggregation were performed as de-scribed (Nagy et al., 1990). Aggregateswere implanted into the uteri of pseu-dopregnant B6D2F1/J@Rj fosters.E14IB10 wild-type (GS�/�) ES cellswere used as controls. The “ES” em-bryos were collected at ED9.5. The pla-centa (extraembryonic) and the heart(embryo proper) were dissected andused to test for the presence of GS,LacZ, or GFP alleles.

Cultivation of ED 2.5Embryos

ED 2.5 embryos (morula stage) werecollected by flushing oviducts anduteri from female GS�/LacZ mice afterheterozygous mating. The zona pellu-cida was removed by a short incuba-tion in pH 2.5 acidic Tyrode’s solution.The embryoid bodies were sampled in-dividually into 0.1% gelatin-coated 96-well plates (Nunc, Wiesbaden, Ger-many) and cultured in 60% BuffaloRat Liver (BRL) conditioned medium(Smith and Hooper, 1987), supple-mented with 2 mM L-glutamine(GIBCO BRL Life Technologies,Breda, The Netherlands), 0.1 mM�-mercaptoethanol, and 1,000 U/mlleukemia-inhibiting factor (ESGRO;Chemicon International Ltd; Harrow,UK). Individual cultures were main-tained for 6 days. Cultures were in-spected daily and photographed in or-der to monitor the development ofoutgrowths. The genotype was deter-mined at the end of the cultivationperiod by PCR.

Detection of �-GalactosidaseExpression

Whole-mounts.

ED3.5, ED12.5, and ED16.5 embryoswere fixed in 4% buffered formalde-hyde for 5, 15, and 45 min at roomtemperature, respectively, and washed3� in PBS followed by whole-mount

staining with X-gal (Franco et al.,2001).

Cryosections.

Embryos were fixed in 4% bufferedformaldehyde, exposed to an ascend-ing gradient of sucrose solutions (10,20, 30%), embedded in Tissue-TekO.C.T (Sakura Finetek USA, Inc., Tor-rance, CA), frozen, and stored at�80°C until sectioning. Cryostat sec-tions of 10 �m were fixed in a 0.5%glutaraldehyde/PBS solution for 10min at room temperature, followed byX-gal staining as described (de Langeet al., 2004).

Immunohistochemistry

Cryosections.

Ten-micrometer-thick cryosectionswere briefly washed in PBS to removethe O.C.T. compound, postfixed with4% buffered formaldehyde for 10 minat room temperature, washed again inPBS, and boiled for 5 min in 10 mMNa-citrate (pH 6.0) to retrieveepitopes (Shan-Rong et al., 2000) andto inactivate endogenous alkalinephosphatase. After cooling to roomtemperature, sections were blocked inTeng-T (10 mM TrisHCl, 5 mM EDTA,150 mM NaCl, 0.25% (w/v) gelatin,and 0.05% (v/v) Tween-20, pH 8.0) for30 min at room temperature and incu-bated overnight with the first anti-body diluted in Teng-T (monoclonalanti-GS, 1: 1,500; Transduction Labo-ratories, Lexington, KY) and monoclo-nal anti-�-galactosidase (anti-�-gal),1:500; Promega Benelux, Leiden, TheNetherlands). Incubations from whichthe first antibody was omitted servedas negative controls. After washing,the sections were incubated with a1:100 dilution in Teng-T of goat anti-mouse IgG covalently coupled to alka-line phosphatase (Sigma, Zwijn-drecht, The Netherlands), washedagain and incubated with alkalinephosphatase substrate (NBT/BCIP,Roche, Almere, The Netherlands)(Franco et al., 2001) for 30 min.

Paraffin-embedded sections.

Organs and tissues were dissectedfrom chimeric neonates and 2-month-old mice, including liver, brain, intes-tine, and kidney, fixed in 4% bufferedformaldehyde overnight, embedded in


paraffin, and sectioned at 7 �m(Franco et al., 2001). To stain for thepresence of GS in wild-type 129P2/OlaHsd mouse blastocysts, 30 blasto-cysts were implanted into a 0.5-cmsegment of adult mouse uterus, ofwhich both ends were tied, followed bystandard fixation procedure. Thereaf-ter, sections were deparaffinized, hy-drated in graded ethanol solutionsand PBS, subjected to antigen re-trieval, and further processed as de-scribed above.

Glutamine SynthetaseActivity Assay

Tissues and organs were isolated from2-month-old mice as described (vanStraaten et al., 2006). The protein con-tent of the lysates (mg/ml) was mea-sured with the bicinchoninic acid re-agent of Pierce (Rockford, IL), usingbovine serum albumin as a standard.The enzymic activity of GS was deter-mined with the -glutamyltransferaseassay} as recently described (vanStraaten et al., 2006), and expressedas nmol product per minute per mgtotal protein at 37°C.

RNA Extraction and RT-PCR

Total RNA was extracted with Trizol(Sigma, Zwijndrecht, The Nether-lands) from 10 cm2 of cultured undif-ferentiated ES cells. The total RNAconcentration was determined spec-trophotometrically at 260 nm. First-strand cDNA was synthesized with200 U Superscript II according to themanufacturer’s instructions (Invitro-gen, Breda, The Netherlands). A totalof 3.5 �g of RNA and 40 pmol oligo-dT14VN was used in a 20-�L reactionvolume (Lekanne dit Deprez et al.,2002). After a 1:1 dilution with 2 mMTris, 0.2 mM EDTA (pH 8.0), the first-strand cDNA (0.5 �L) was used forPCR. The forward primer was locatedupstream of the targeting construct inGS exon I, whereas the reverse prim-ers were located in the LacZ, GFP, orendogenous GS-coding sequence inexon II. Glyceraldehyde 3-phosphatedehydrogenase (GAPDH) was used asinternal control (Table 4). When re-verse transcriptase was omitted in thecDNA synthesis reaction, no productwas formed.

Statistical Analysis

The genotypes of the offspring of het-erozygous GS�/LacZ intercrosses wereanalyzed with a Chi-square test perage group with Mendelian inheritanceas expected distribution. The datafrom the experimental and controlgroups of chimeric and tetraploid em-bryos were analyzed with a Chi-square test. A difference was consid-ered statistically significant at P �0.01. GS enzyme activity was testedfor gender and genotype effects with atwo-way analysis of variance(ANOVA) for each of 12 organs. ABonferroni correction was applied andthe effects were, therefore, consideredstatistically significant at P � 0.004.Since no significant gender effectswere observed (see also our earlierdata in van Straaten et al., 2006), wepooled the data of males and femalesfor both genotypes. Data are pre-sented as mean � SEM per group.

ACKNOWLEDGMENTSWe thank Drs. Rosa S. Beddington†

and William C. Skarnes (MRC, Lon-don) for making plasmid pB�geopAavailable, Dr. Etsuko Yasuda and Ms.Wilhelmina T. Labruyere for the gen-eration of GSGFP/LacZ ES cells, Dr. JanM. Ruijter for statistical analysis, andIrena van Herk, who contributed as astudent to the study. We are indebtedto the personnel of the animal facilityfor their care of the many mice used inthe study.

†Deceased.

REFERENCES

Biggers JD, McGinnis LK, Lawitts JA.2004. Enhanced effect of glycyl-L-glu-tamine on mouse preimplantation em-bryos in vitro. Reprod Med Online 9:59–68.

de Lange FJ, Moorman AFM, AndersonRH, Manner J, Soufan AT, de Gier-deVries C, Schneider MD, Webb S, van denHoff MJB, Christoffels VM. 2004. Lin-eage and morphogenetic analysis of thecardiac valves. Circ Res 95:645

Eggan K, Akutsu H, Loring J, Jackson-Grusby L, Klemm M, Rideout WM,Yanagimachi R, Jaenisch R. 2001. Hy-brid vigor, fetal overgrowth, and viabil-ity of mice derived by nuclear cloningand tetraploid embryo complementa-tion. Proc Natl Acad Sci USA 98:6209 –6214.

Evans MJ, Carlton MBL, Russ AP. 1997.Gene trapping and functional genomics.Trends Genet 13:370–374.

Franco D, de Boer PAJ, de Gier-de Vries C,Lamers WH, Moorman AFM. 2001.Methods on in situ hybridization, immu-nohistochemistry and b-galactosidase re-porter gene detection. Eur J Morphol 39:3–25.

Friedrich G, Soriano P. 1991. Promotertraps in embryonic stem cells: a geneticscreen to identify and mutate develop-mental genes in mice. Genes Dev 5:1513–1523.

Gardner DK, Lane M. 1993. Amino acidsand ammonium regulate mouse embryodevelopment in culture. Biol Reprod 48:377–385.

Haberle J, Gorg B, Rutsch F, Schmidt E,Toutain A, Benoist JF, Gelot A, Suc AL,Hohne W, Schliess F, Haussinger D,Koch HG. 2005. Congenital glutaminedeficiency with glutamine synthetasemutations. N Engl J Med 353:1926–1933.

Harris SE, Gopichandrian N, Picton HM,Leese HJ, Orsi NM. 2005. Nutrient con-centrations in murine follicular fluid andthe female reproductive tract. Theriog-enology 64:992–1006.

Hogan B, Beddington R, Costantini F, LacyE. 1994. Manipulating the mouse em-bryo: a laboratory manual. Plainview,NY: Cold Spring Harbor LaboratoryPress.

Houghton FD. 2006. Energy metabolism ofthe inner cell mass and trophectoderm ofthe mouse blastocyst. Differentiation 74:11–18.

Lane M, Gardner DK. 2003. Ammoniuminduces aberrant blastocyst differentia-tion, metabolism, pH regulation, geneexpression and subsequently alters fetaldevelopment in the mouse. Biol Reprod69:1109–1117.

Lekanne dit Deprez RH, FijnvandraatAC, Ruijter JM, Moorman AFM. 2002.Sensitivity and accuracy of quantita-tive real-time polymerase chain reac-tion using SYBR green I depends oncDNA synthesis conditions. Anal Bio-chem 307:63–69.

Lie-Venema H, de Boer PAJ, MoormanAFM, Lamers WH. 1997a. Organ-spe-cific activity of the 5� regulatory regionof the glutamine synthetase gene in de-veloping mice. Eur J Biochem 248:644 –659.

Lie-Venema H, de Boer PAJ, MoormanAFM, Lamers WH. 1997b. Role of the 5�enhancer of the glutamine synthetasegene in its organ-specific expression. Bio-chem J 323:611–619.

Lie-Venema H, Hakvoort TBM, vanHemert FJ, Moorman AFM, LamersWH. 1998. Regulation of the spatiotem-poral pattern of expression of the glu-tamine synthetase gene. Prog NucleicAcid Res 61:243–308.

Martin KL, Leese HJ. 1999. Role of devel-opmental factors in the switch frompyruvate to glucose as the major exoge-nous energy substrate in the preimplan-

1874 HE ET AL.

tation mouse embryo. Reprod Fertil Dev11:425–333.

Meister A. 1974. Glutamine synthetase ofmammals. In: Boyer PD, editor. The en-zymes. New York: Academic Press. p699–754.

Nagy A, Gocza E, Diaz EM, Prideaux VR,Ivanyi E, Markkula M, Rossant J. 1990.Embryonic stem cells alone are able tosupport fetal development in the mouse.Development 11:815–821.

Nagy A, Rossant J, Nagy R, Abramow-Newerly W, Roder JC. 1993. Derivationof completely cell culture-derived micefrom early-passage embryonic stemcells. Proc Natl Acad Sci USA 90:8424 –8428.

Rezk Y, Huff C, Rizk B. 2004. Effect ofglutamine on preimplantation mouse

embryo development in vitro. Am J Ob-stet Gynecol 190:1450–1454.

Rugh R. 1990. The mouse: its reproductionand development. New York: OxfordUniversity Press.

Shan-Rong S, Jiang G, Taylor CR. 2000.Antigen retrieval techniques: imunohis-tochemistry and molecular morphology.Natick, MA: Eaton Publishing.

Skarnes WC, Auerbach BA, Joyner AL.1992. A gene trap approach in mouseembryonic stem cells: the lacZ reportedis activated by splicing, reflects endoge-nous gene expression, and is mutagenicin mice. Genes Dev 6:903–918.

Skarnes WC, Moss JE, Hurtley SM, Bed-dington RSP. 1995. Capturing genes en-coding membrane and secreted proteinsimportant for mouse development. ProcNatl Acad Sci USA 92:6592–6596.

Smith AG, Hooper ML. 1987. Buffalo ratliver cells produce a diffusible activitywhich inhibits the differentiation of mu-rine embryonal carcinoma and embry-onic stem cells. Dev Biol 121:1–9.

Theiler K. 1983. Embryology in the mousein biomedical research. New York: Aca-demic Press.

van Straaten HWM, He Y, van Duist MM,Labruyere WT, Vermeulen JLM, van DijkPJ, Ruijter JM, Lamers WH, HakvoortTBM. 2006. Cellular concentrations of glu-tamine synthetase in murine organs. Bio-chem Cell Biol 84:215–231.

Wang ZQ, Kiefer F, Urbanek P, WagnerEF. 1997. Generation of completely em-bryonic stem cell-derived mutant miceusing tetraploid blastocyst injection.Mech Dev 62:137–145.


Documents

Glutamine synthetase is essential in early mouse embryogenesis