Access the full text.
Sign up today, get DeepDyve free for 14 days.
Loading next page...
/lp/unpaywall/activin-inhibin-beta-b-subunit-gene-disruption-leads-to-defects-in-IS67u8dCSD
References
References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.
- Publisher
- Unpaywall
- ISSN
- 0890-9369
- DOI
- 10.1101/gad.8.4.414
- Publisher site
- See Article on Publisher Site
Abstract
Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Activin/inhibin ^B subunit gene disruption leads to defects in eyelid development and female reproduction Anne Vassalli/'^ Martin M. Matzuk,^ Humphrey A.R. Gardner/ Kuo-Fen Lee/ and Rudolf Jaenisch^'^ 'Whitehead Institute for Biomedical Research, Cambridge, Massachusetts 02142 USA; ^Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139 USA; ^Department of Pathology and Institute for Molecular Genetics, Baylor College of Medicine, Houston, Texas 77030 USA Inhibins and activins are dimeric growth factors of the transforming growth factor-p superfamily, a class of peptides that can regulate the growth and differentiation of a variety of cell types. Recently, activins have been implicated in early vertebrate development through their ability to evoke, in Xenopus embryo explants, both morphological and molecular changes characteristic of mesoderm induction. To understand these processes further, we have used homologous recombination in embryonic stem cells to create mouse strains carrying mutations in the gene encoding the activin/inhibin PB subunit. These mice are expected to be deficient in activin B (pB:pB), activin AB (pA:pB), and inhibin B (a:pB). Viable mutant animals were generated, indicating that the PB subunit is not essentia] for mesoderm formation in the mouse. Mutant animals suffered, however, from distinct developmental and reproductive defects. An apparent failure of eyelid fusion during late embryonic development led to eye lesions in mutant animals. Whereas pB-deficient males bred normally, mutant females manifested a profoundly impaired reproductive ability, characterized by perinatal lethality of their offspring. The phenotype of mutant mice suggests that activin 3B (1) plays a role in late fetal development and (2) is critical for female fecundity. In addition, we have found that expression of the related pA subunit of activin is highly upregulated in ovaries of mutant females. Altered regulation of PA activin in pB-deficient mice may contribute to the mutant phenotype. [Key Words: Mouse; activin B/inhibin B; gene targeting; eyelid fusion; female reproduction] Received October 8, 1993; revised version accepted December 28, 1993. Candidate genes and proteins that govern the formation cause of their ability to induce a full range of mesoder and patterning of the mesoderm in vertebrate embryos mal fates, from anterior-most mesoderm (organizer) to have recently emerged. Through studies in amphibians tail, in a concentration-dependent maimer (Green et al. came the realization that specific polypeptide growth 1992). Genes induced by activins as an immediate-early factors were able to reproduce some aspects of mesoderm response were identified and include the organizer-spe induction in embryo explants. These factors include cific goosecoid gene and the Brachyuiy gene (Cho et al. members of the transforming growth factor-p (TGF-p) 1991; Smith et al. 1991). In the Biachymy mouse, me superfamily such as activins A and B, bone morphoge- soderm formation is defective. Cloning of the mouse netic protein 4 (BMP-4) and Vg-1, and members of the Brachyury (T) gene (Herrmann et al. 1990) has provided fibroblast growth factor (FGF) family such as basic FGF the first identification of a gene located on the pathway (Kimelman and Kirschner 1987; Slack et al. 1987; Asash- of mesoderm formation in vertebrates. As inducers of ima et al. 1990; Smith et al. 1990; Thomsen et al. 1990; Brachyury, activins are thought to be part of this path van den Eijnden-Van Raaij et al. 1990; Koster et al. 1991; way as well. Animal caps induced by activin also un dergo morphogenetic cell movements that are character Dale et al. 1992; Jones et al. 1992; Thomsen and Melton istic of gastrulation (Smith and Howard 1992). Thus his 1993). In addition, other factors that do not have the tological, molecular, and cell behavior responses of ability to induce mesoderm from presumptive ectoderm animal cap explants to activin recapitulate many aspects by themselves were found to modify the pattern of me of gastrulation. Hemmati-Brivanlou and Melton (1992) sodermal differentiation evoked by an inducer. These in have provided the most compelling evidence of the in clude members of the Wnt family such as Xwnt-8, and volvement of an activin-like molecule in mesoderm for noggin (Smith andHarland 1991, 1992; Sokol et al. 1991; mation by showing that injection of an mRNA encoding for review, see Kimelman et al. 1992; Sive 1993). Among a truncated activin receptor into Xenopus embryos in- mesoderm-inducing molecules, activins are unique be GENES & DEVELOPMENT 8:414-427 © 1994 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/94 $5.00 414 Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Activin B/inhibin B-deficient mice terferes with activin signaling and prevents mesoderm al. 1991; Hemmati-Brivanlou and Melton 1992), and in induction. duction of erythropoietic differentiation (Murata et al. 1988). In addition to their action on FSH release, their Activins are structurally related to the inhibin class of ability to regulate secretion of other hypothalamic or peptides. Inhibins were initially isolated from mamma pituitary peptides such as oxytocin, ACTH, and GH was lian follicular fluid on the basis of their ability to inhibit shown (Vale et al. 1990). the release of follicle stimulating hormone (FSH) from anterior pituitary cells (for review, see Vale et al. 1990). To study the physiological role of these factors in In the course of inhibin purification, side fractions that mouse development and reproduction, we have used ho instead stimulated FSH release led to the identification mologous recombination in embryonic stem (ES) cells to of activins (Ling et al. 1986; Vale et al. 1986). Inhibins create mutant alleles where most of the first coding exon and activins are dimers assembled from subunits en of the pB gene was deleted, and these mutations were coded by three distinct genes: the a, (3A, and pB subunit introduced into the germ line of mice. Although it re genes. The precursor polypeptides encoded by these mains to be seen whether embryonic development pro genes are proteolytically processed into mature peptides, ceeds normally at a molecular level in pB-deficient ani whose dimers form the biologically active molecules. In mals, they are born without major morphological abnor hibins are a:(B heterodimers, whereas activins are p:p malities, with the exception of a defect in eyelid homo- or heterodimers. The two p subvmits, pA and pB, outgrowth and closure. The generation of viable mutant share within a species —60% sequence identity, whereas animals has, however, revealed a critical role for activin/ the a subunit is more distantly related. Each p subunit is inhibin pB in female reproductive function. highly conserved across species and >98% identical be tween mammalian species (Vale et al. 1990). The pB sub- unit is expressed in Xenopus (Thomsen et al. 1990) and Results chick (Mitrani et al. 1990) late blastula embryos. In the Mice homozygous for a targeted mutation mouse, p subunit mRNAs and p subunit immunoreac- in the activin/inhibin pB subunit gene are viable tivity, in the absence of a subunit expression, were dem onstrated during preimplantation stages, indicating the The mouse activin/inhibin pB subunit gene was isolated presence of activin in the pregastrulation mouse embryo from a genomic DNA library derived from the 129/Sv (van den Eijnden-van Raaij et al. 1992; Albano et al. strain by use of a rat partial cDNA as a probe (rinB-c2-P; 1993). K. Mayo, unpubl.). Location of the coding exons is dia In adult animals, highest expression levels of activin/ gramed in Figure 1, A and B. Two replacement type tar inhibin subunits are found in the gonads. Recent evi geting vectors termed pBBI and pBB2 (Fig. IB) were used dence implicates inhibins and activins as intragonadal to disrupt the pB subunit gene by homologous recombi paracrine and/or autocrine regulators (for review, see de nation (Thomas and Capecchi 1987). Each vector con Jong et al. 1990; Mather et al. 1992; Findlay 1993). The tained 15.5 kb of homology with the cognate gene. These major sites of activin/inhibin expression are the Sertoli constructs were designed to replace a 1.3-kb region of the cells of the testis and the granulosa cells of the ovary. pB subunit locus spanning the translational start codon, Inhibins and activins were found to affect both the germ the signal peptide coding region, as well as most of exon 1-encoded pro-region sequences with a PGK promoter- cells and the somatic components of the gonads. Activin driven neo selection cassette containing a PGK poly(A) can stimulate spermatogonial proliferation, whereas in signal [PGK-NEO-p(A)] (McBumey et al. 1991). The hibin has the opposite activity. Isotypes of activin recep pBBl construct contained the PGK-NEC)-p(A) cassette tors were detected on specific populations of male germ in the same transcriptional orientation as the pB gene, cells (Kaipia et al. 1992, 1993), suggesting that these fac and the pBB2 construct contained it in reverse orienta tors play a role in the regulation of spermatogenesis. Fur tion. }1 ES cells (Li et al. 1992) were electroporated with thermore, androgen biosynthesis in testicular Leydig or linearized DNA and selected with G418. Homologous ovarian thecal cells is inhibited by activin and enhanced recombinant ES cell clones were identified by Southern by inhibin. In females, activin and inhibin were sug blot analysis of genomic DNA from individual clones gested to either stimulate or inhibit, respectively, oocyte (see Materials and methods). Both targeting constructs meiotic maturation, and to affect follicular develop demonstrated similar, high frequencies of recombination ment. The most conclusive evidence for a critical role of at the endogenous pB subunit locus (Table 1). inhibin as a regulator of proliferation of somatic compo nents of the gonad was provided by the finding that in- Three independent ES clones heterozygous for the BBl hibin-deficient mice generated by a subunit gene target mutation and one clone heterozygous for the BB2 muta ing develop gonadal stromal tumors at a very early age tion were injected into host blastocysts to generate chi (Matzuk et al. 1992). meric foimder mice. All four clones contributed to the Other studies have revealed extragonadal sources of germ line of chimeric animals, permitting the derivation inhibin and activin synthesis and demonstrated re of several independent mouse lines carrying the mutant sponses in a variety of cell types (Meunier et al. 1988; alleles. Mice heterozygous for either the BBl or BB2 mu Vale et al. 1990). Thus, activins have been implicated in tation did not display an overt phenotype. To evaluate inhibition of neural differentiation (Hashimoto et al. the possibility that the genetic background influences 1990; Schubert et al. 1990; van den Eijnden-van Raaij et the mutant phenotype, animals heterozygous for the BBl GENES & DEVELOPMENT 415 Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Vassalli et al. Nc JO JQ <VI P»J 1.3 kb I N £t Si (M Si Si deleted in • + JO Si \ \ BBl A BB2 - + + + <oV alleles ^O^ // 100 bp 2.4 kb intron -8. 0 -6. 0 precursoivencoding mature peptide- precursor-encoding region of exon 2 encoding region region of exon 1 Figure 1. Mutagenesis at the activin/m- hibin pB subunit locus of mice. (A) Exon Mouse Activin/Inhibin locus ^yg 5J0P structure of the pB subunit gene showing I B I B B the region deleted in the BBl and BB2 mu- rc r I I I I I L tant alleles. The transcription start and polyadenylation sites have not been 'robe A rinB -c2-P I I mapped; therefore, the corresponding exon Probe E bovmdaries are indicated with dashes. The I i hatched box indicates the coding region for the mature pB peptide. Note that the 6.0 kb 9.5 kb pBB l 5'-splice site at the end of the first coding PGKNEOPCA) exon is left intact in the BBl and BBl re combinant alleles. Not all Narl sites I within the 1.3-kb deleted fragment are rep- pBB2 ^ resented. (B) Targeting scheme. A physical CV)d03N)ISd °-^*P °^ ^^^ locus is shown at the top. The 1 kb structure of the targeting constructs is in- dicated below. Dashed lines delineate re gions of homology with the chromosome. The new £coRI site introduced by the mutation is indicated in each targeting vector. The pBB2 vector is identical to pBBl, except that the orientation of the neo cassette is reversed. Probe E is the external probe used to screen ES cell colonies and to genotype mice. (rinB-c2-P) The rat cDNA probe used to isolate genomic DNA from the pB locus. (C) Southern blot analysis of DNA from offspring produced by heterozygote intercrosses. DNA was digested with £coRI and hybridized to probe E. Molecular weight markers (in kilobases) are shown at the right. Restriction sites: (B) BglU; (R) £coRI; (N) Nail; (Nc) Ncol. mutatio n were bred to homozygosity on three different BB2 mutatio n are viable, although their viability ma y b e geneti c backgrounds. Genotype analysis of weanlin g off affected by th e genetic background. spring (Table 2) revealed that homozygous mutant ani mal s with a 129/SvxC57BL/6 hybrid genetic back- Mutant mice lack wild-type uiRNA and mature pB groimd were only marginally compromised in their via peptide bility, if at all, as they represented 21 % of th e animals genotyped, instead of the expected 25%. In the 129/ T o examine ho w th e BBl mutatio n affects transcription SvxBALB/ c hybrid background, however, th e fraction of at the pB locus, RNA was prepared from ovaries and homozygou s mutants was lower (Table 2), suggesting teste s of immature (3-week) or adult animals and sub tha t alleles in th e BALB/c background ma y interac t with jected to reverse transcription followed by th e polymer th e activin mutatio n and reduc e viability. These results ase chain reaction (RT-PCR). When oligonucleotide therefore indicated tha t mic e homozygous for th e BBl or primer s tha t span th e matur e peptide-coding region were Table 1. Homologous lecombination at the activin/inhibin pB locus Targeting frequency G418' targeted clones/ Targeting Cells G418^ Positive clones construct electroporated colonies screened clones^ 0418"^ clones transfected cells pBBl 5 X 10^ 5100 172 41 1/4.2 2.4 X 10-^ pBB2 5 X 10^ 2900 58 10 1/5.8 1 X 10-^ *As determined by Southern blot analysis using genomic probe E (see Fig. 1). 416 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Activin B/inhibin B-defidetit mice Table 2. Genotypes at weaning age PB genotype of progeny Cross Genetic Total (female x male) Mouse line^ background'' progeny + / + + / m m/m ( + /m) X ( + /m) 129-134 11 15 14 129 40 BALB-210 (129 xBALBjFi 103 30 53 20 BALE-15 (129 XBALB)N1 60 22 28 10 B6-134 (129 xB6)Fi 65 17 35 13 36-43 (129 xB6)N l 223 61 114 48 (+ /m) X (m/m) B6-43 (129 xB6)N l 80 40 40 (m/m) X ( + /m) B6-43 (129 xB6)N l 21 11 10 *A11 lines carry the BBl mutation, except line BALB-210, which carries the BB2 mutation. The number after the dash in a line name refers to the original ES cell clone. "To establish lines B6-43 and BALB-15, heterozygous Fj animals were first backcrossed once with wild-type C57BL/6 or BALB/c animals, respectively, and the resulting heterozygotes (B6, or BALB, Nl animals) were intercrossed. (129) The 129/Sv strain. (B6) The C57BL/6 strain. used, a pB-specific band of the expected size was ob pB readthrough transcript was —1/10 as abundant as the tained both in mutants and in controls. In addition, neo transcript terminated at the PGK poly(A) site. Exam primers that span the intron amplified the expected pB- ination of RNA from animals homozygous for the BB2 specific band in mutant animal tissue, although with allele likewise showed the absence of the wild-type tran markedly reduced efficiency relative to controls (data scripts and revealed the production from this allele of a not shown). These results indicated that transcription new transcript of —3.0 kb that did not hybridize to the downstream of the neo insertion occurred in the BBl neo probe (Fig. 2), consistent with its initiation within allele. This is not unexpected as the PGK poly(A) site the PGK promoter (used in reverse orientation) and its was shown previously to allow readthrough transcript termination at an endogenous pB poly(A) site. accumulation (e.g., Lee et al. 1992). To establish whether the aberrant transcript made by To characterize further the nature of the transcript the BBl allele could produce pB peptide, we then per produced by the BBl mutated allele, Northern blot anal formed protein blot analysis. pB-Specific antibodies ysis was performed (Fig. 2). The two probes used were the (Vaughan et al. 1989) could readily detect a — 12-kD pep mACTlO/lljl S probe, a PCR-generated fragment corre tide in ovarian tissue from heterozygous or wild-type sponding to the mouse pB mature peptide region (Fig. weanling females (Fig. 3, top, lanes 6,7,9,10,12-14), but 2A; see Materials and methods), and a neo probe (Fig. 2B). not in tissue from bbl/bbl mutant females (lanes An ~4.5-kb transcript corresponding to the major wild- 4,5,8,11). A polypeptide of —50 kD, consistent with the type mRNA (Albano et al. 1990; Manova et al. 1992) was pB precursor size, was also detected by the pB-specific detected with the mACT10/ll;13 probe in ( + / + ) and antibodies in gonadal extracts from wild-type or hetero (-H/bbl) but not in (bbl/bbl) ovaries. In wild-type ani zygous animals but was absent in homozygous mutants mals, a minor ovarian transcript of ~3.5 kb could be (not shown). detected but not in (bbl/bbl) animals. The 3.5-kb tran In summary, the results of Figures 2 and 3 indicate script was also seen in wild-type testes, approximately that the BBl mutant allele does not yield productive pB equimolar to the 4.5-kb transcript, as reported previously subunit mRNAs capable of generating the mature pB (Feng et al. 1989a). Neither the 4.5- nor the 3.5-kb tran peptide whose dimer with another p subunit or with an scripts were detected in the (bbl /bbl ) mutant testes. The a subimit forms the biologically active activin or inhibin mACT10/ll;13 probe, however, detected a novel tran ligand. bbl/bbl mutant animals are therefore expected script of —4.0 kb in both testes and ovaries of animals to lack activins B and AB, as well as inhibin B, but their containing a mutated BBl allele but not in (+ / +) ani ability to produce activin A and inhibin A should be mals. The neo probe also hybridized to a band of the preserved. same size. Therefore, the 4.0-kb transcript likely corre sponds to a neo-pB chimeric transcript, reading through The ^B-deficient ovary up-iegulates expression the PGK poly(A) site and terminating at one of the pB of the pA subunit gene endogenous poly(A) sites. Because the promoter(s) and poly(A) site(s) of the mouse pB gene have not been Expression of the activin/inhibin pA mature peptide was mapped, it is not possible to predict the exact size of investigated in total ovarian tissue extract from 3-week- such a chimeric transcript. As expected, a major ~1.0-kb old females. This analysis revealed a 3- to 20-fold in band corresponding to a transcript initiated at the PGK crease in levels of pA mature peptide in animals ho promoter and terminated at the PGK poly(A) site was mozygous for the BBl mutation, relative to their hetero detected by the neo probe in bbl/bbl and -I-/bbl ani zygous or wild-type littermates (Fig. 3, middle, e.g., cf. mals. As determined with the neo probe, the 4.0-kb neo - lanes 4 and 5 with lanes 6 and 7). In the same ovarian GENES & DEVELOPMENT 417 Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Vassalli et al. A mACT1011;13 sion was examined by Northern analysis of ovarian ex tract s from immatur e animals. N o consistent alterations in levels of th e ~7.0-kb mRN A were observed in mutan t tissu e (not shown). Therefore, a post-transcriptional mechanis m may operate to up-regulate pA matur e pep tide production in the pB-deficient ovary. — 28S Failure of eyelid fusion at birth Upo n intercrossing animals heterozygous for th e BBI (or BB2) mutatio n we observed that a fraction of the neo nate s in th e resulting litters ha d a distinctiv e phenotype. — 18S Normally , mouse pups are bo m wit h closed eyes due to fusion of the eyelids at embryonic day 16 (E16) (Rugh 1990), and eyes open again only at —13 days of postnatal life. In contrast, a fraction of the offspring from hetero zygous parents were bo m with open eyes, as illustrated in Figure 4, A and B. All of the open-eyed neonates de ovary testis Uterus rived from four independent ES cell clones carrying ei 3-wk 4-wk 6-wk the r the BBI or BB2 mutatio n and genotyped at birth by B neo Southern analysis were homozygous for the mutan t al rsi S s s lele. No t all homozygotes, however, showed this pheno JU £> - O JD X i JO x» ,— 1 1—4 Xi Xi ,- H ~~-> •-^ type, suggesting incomplete penetrance (see below). To + "+ + + "•• ^ •~~ ^ x> ^•^ ^ X X •~-> . X ) x> X I X X ^-^ X> determin e whether the stmcture of the eye was other + + + + + + + wis e affected in the mutan t newborns, histological sec ''^ * tions through the heads of bbl/bb l mutant animals and 28 S -l-/bbl control littermates were examined (Fig. 4 C-E). Th e general anatomy of the eye of th e mutan t newborns appeared normal, indicating a defect confined to th e eye lids. Animals bom with open eyes quickly developed a numbe r of eye defects, which likely reflected traumatic damage due to the lack of eyelid protection. Hyperkera- 18S tinizatio n and squamous metaplasia of the corneal epi thelium , accompanied by massive leukocyte infiltration of all cornea layers and eyelids, was observed as soon as th e first day of life. During the subsequent days, eyelids often sealed. Th e timing of eye reopening the n appeared delayed in mutant s and revealed permanen t damage that ovary included ocular dystrophy and opacification of the cor 3-wk nea. Thi s macroscopic eye patholog y allowed ready iden tification of mutan t animals during adulthood. Figure 2. RNA analysis in pB mutant animals. Total RNAs from ovaries (~10 fxg, lanes 1-5), testes (20 (xg, lanes 6-8], and Th e open eyelid phenotype of the BBI mutation was uterus (20 \3.g, lanes 9-J i) from individual animals were exam found to depend on the genetic background (Table 3). In ined by RNA blot analysis with a probe corresponding to the pB th e original mouse line 43, F^ animals had been back- mature peptide encoding region [mACT1011;13 (A)] or a neo crossed once to C57BL/6 and then intercrossed. In this probe [B]. Genotypes of animals are indicated at the top, and line, th e fraction of homozygous mutan t animals exhib positions of ribosomal RNAs are shown at the right. iting obvious eye defects during adulthood was 0.84 (n = 89). A similar phenotype was observed in a 129/ SvxBALB/c hybrid genetic background, although with extracts, anti-a inhibin subunit antibodies recognized a reduced penetrance. In contrast, whe n the BBI mutation ~45-k D polypeptide w^hose levels were not altered by wa s present in a 129/Sv inbred background, the failure of th e pB mutatio n (Fig. 3 , bottom). eyelid fusion was never observed: Non e of th e pups born Overexpression of the pA subunit in the pB-deficient from heterozygote intercrosses (12 = 137), or bom from ovary may reflect altered development and cellular ar homozygou s mutant males mated to heterozygous fe chitectur e of the mutan t tissue, leading to an over-rep male s (12 = 69), had open eyes at birth; and among 14 resentatio n of pA-expressing cells. Alternatively, regula adul t mutant animals, none had eye defects. This sug tory mechanism s that sense th e pB deficiency may exist gests that the 129/Sv inbred strain carries a gene (or and cause a compensator y increase in pA subuni t expres genes) tha t suppresses the open eyelid phenotyp e caused sion. by the BBI mutation. Segregation of such 129/Sv-spe- To analyze thi s effect further, pA subuni t RNA expres cific suppressor genes may also explain the incomplete GENES & DEVELOPMENT 418 Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Activin B/inhibiti B-deficient mice i + + ^ ii 21.5 - Figure 3. Animals homozygous for the BBl muta tion lack mature pB peptide and up-regulate pA pep tide expression. Ovarian extracts from individual im PB mature (21-day-old) females were run on a SDS- 14.3-1 : PAGE gel under reducing conditions and examined by immunoblot analysis with anti-pB antibodies [top]. These antibodies recognize a peptide in the car- boxy-terminal region of the pB subunit. The same blot was stripped and reanalyzed using anti-pA anti 21.5 - bodies [middle] or anti-a antibodies (bottom). Lanes 1-3 contain 1 ng each of human recombinant inhibin A (lane 1), activin A (lane 2), and activin B (lane 3). Lanes 4-14 contain 100 |xg each of total protein from 14.3 pA ovary. Genotypes are indicated at the top. Lanes 13 and 14 represent a control wild-type 129/Sv and a control wild-type C57BL/6 female, respectively. Ex tracts in lanes 4—7 originate from siblings with a 129/ SvxBALB/c hybrid genetic background (BALB-N2 69 - backcross), lanes 8-10 are from siblings with a 129/ SvxC57BL/6 background (B6-N1), and lanes 11 and 46 — a 12 are from siblings from a (I29/SvxC57BL/6) N4 backcrossed generation. Migrations of molecular mass markers (in kilodaltons) are shown at left. 8 9 10 11 12 13 14 penetranc e of the phenotype in the two hybrid genetic backgrounds . Furthermore^ the expressivity of the phe notyp e varied among affected individuals, M^ith various S|^^ degrees of completion of eyelid fusion at birth or, occa sionally, a single eye being affected. Mutant females manifest a reproductive impairment Activin s and inhibins are thought to be regulators of re productiv e functions through their influence on both the pituitar y and the gonads. Therefore, we then evaluated th e fertility of the BBl homozygous mutant animals. Th e viability of homozygous mutant embryos had also raised th e possibility that maternal supply of pB peptide migh t rescue the potential defect in mutan t embryos. If thi s were the case, mutan t females might be expected to be impaire d in sustaining norma l pregnancy and delivery of live pups. To assess reproductive performance, we bred bbl/bb l mutan t animals of either gender with heterozygous mate s and intercrossed homozygous mutan t animals or ^^f^r.v.' heterozygous littermates. The average numbe r of viable Figure 4. Failure of eyelid closure at birth. [A] A heterozygous offspring per breeding pair and per month of breeding control pup showing fused eyelids covering the eye. (5) Ho was determined for each type of cross. Figure 5 summa mozygous mutant littermate with open eyelids. Photographs in rizes the reproductive performances observed. Homozy A and B were taken within a few hours of birth. (C-E) Trans gous mutan t males bred as well as their heterozygous verse sections through the eyes of a heterozygous control neo littermate s when mated to heterozygous females. In nate (C); an arrow indicates the site of eyelid fusion, a mutant sharp contrast, homozygous mutant females mated to neonate a few hours after birth (D), before inflammation is ev eithe r heterozygous or homozygous mutan t males man ident, and a mutant neonate at —0.5-1 day of age (£), displaying ifested a profoundly reduced ability to generate live off acute leukocyte infiltration in the cornea and eyelids, corneal spring as compared wit h their heterozygous littermates. edema, and corneal hyperkeratinization. Objective magnifica tion was 10 X. Histological examinatio n of ovaries from pB-deficient fe- GENES & DEVELOPMENT 419 Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Vassalli et al. Table 3. The eyelid closure defect: phenotypic variation with genetic background Cross oeb/Tota P Penetrance' ' Geneti c (female x male) (neonates) (adults) Mous e line background 0.00 [n = 14) ( + /m ) X ( + /m) 129-134 129 0.00 {n = 137) 0.21 (22 = 24) BALB-210 (129 X BALB)Fi 0.04 [n = 90) BALE-15 (129 X BALE) N l 0.12 {n = 102) N.D . B6-134 {129xB6)F i 0.11 (n = 299) 0.33 (22 = 21) B6-43 (129 X B6)N 1 0.20 {n = 367) 0.84 (22 = 89) ( + /m ) X (m/m) B6-43 (129 X B6)N 1 0.46 (22 = 225) (m/m) X (+ /m) 0.53 {n = 36) B6-43 (129 X B6)N 1 (m/m) X (m/m) 0.86 {n = 78) B6-43 (129 X B6)N 1 ^oeb/Total, fraction of total offspring with one or both eye(s) open at birth or displaying resulting eye defects as observed within 1 week of life. The total number of offspring examined for each cross is indicated in parentheses. All pups whose eyes could be examined are entered irrespective of their viability. ''For each mouse line, the penetrance is calculated as the fraction of adult homozygous mutant animals displaying obvious eye defects (ocular dystrophy and/or corneal opacification) of one or both eyes, as determined by simple visual inspection. The total number of homozygous mutant animals examined is in parentheses. (N.D.) Not determined. this, we observed no distortion of genotypes among the male s did not show^ overt abnormalities (data not rare pups that survived the perinatal period: Of 21 wean show^n). Th e reproductive impairmen t of mutan t females ling mice bo m from bbl/bb l mutant females mated to did not result from defective oogenesis, inability to un heterozygous males, 10 were bbl/bbl and 11 were dergo fertilization, or development of embryos in utero, -l-/bbl. Thus, neonatal lethality appears unaffected by because they became pregnant at a frequency similar to th e genotype of th e offspring but rather reflects a mater thei r heterozygous littermates. Reduced fertility ap nal defect. peared rather to result from perinatal loss of the progeny. Offspring were bo m from mutan t females but rarely sur Female reproductive failure was observed in lines of vived beyond 24 h r postpartum. A commo n observation mic e derived from three independent ES cell clones and o n the day of delivery was death of an entire normal- was manifest in th e three genetic backgrounds tested: In sized litter. inbred 129/Sv background and in the two hybrid back grounds (129/SVXC57BL/6 and 129/SvxBALB/c, after Ther e was no excess of phenotypically mutant ani one backcross to C57BL/6 or BALB/c, respectively). 129/ mal s among th e pups dying perinatally. Consistent with (129XB6)N1 {129XBALB)N1 129 O) *^ Figure 5. Reproductive performance of (A XT animals heterozygous or homozygous for the BBl mutation. The number of viable progeny per breeding pair and per month of breeding was averaged for each type of cross in each genetic background. Genetic background is indicated at the top, and the number of breeding pairs (n) is indicated at m/m +/m m/m female +/m +/m m/m m/m +/m the bottom. Several m/ m (129xBALB) fe X X X X X X X X males died during their first pregnancy male +/m m/m +/m +/m +/m +/m m/m +/m from failure to deliver and were not in cluded. n=3 n=2 n=4 n=4 n=6 n=3 n=2 n=6 420 GENES & DEVELOPIVIENT Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Activin B/inhibin B-defident mice SvxBALB/c hybrid mutant females appeared the most lobules and ducts of mutant tissue is evident. These ob seriously affected: A total of 10 pregnancies carried by servations suggest that either milk let-down is impaired four different females failed to generate live offspring. or that the babies fail to suckle. The former possibility is supported by preliminary data showing that offspring bom from mutant females and not nursed can be rescued Mutant females show increased gestation time by fostering to a normal female (not shown). and decreased nursing ability In staged matings where the day of vaginal plug (E0.5) FSH measurements in mutant animals was recorded, pB-deficient females were observed to To analyze further the cause of the reproductive defect in show an increase in the average duration of gestation. mutant females and to determine whether pituitary Among 13 control sibling females (heterozygous or wild function was altered in pB-deficient animals, serum FSH type), 8 had delivered the morning of day E19.5 and all levels were measured (Table 4). The interpretation of but one had delivered by the afternoon of that day. In these data was complicated by the facts that (1) FSH striking contrast, none of 12 mutant females delivered values fluctuate widely within an experimental group on day E19.5. Most mutant females delivered on day and (2) an age-dependent increase in FSH values was ev E20.5 and some in the morning of day E21.5. Occasion ident in adult animals from 11 to 24 weeks of age. Nev ally, initiation of labor appeared to fail altogether, result ertheless, this study revealed an —20% increase in aver ing in sickness of the gravid mother and death of fetuses age serum FSH levels in mutant animals relative to het- in utero. erozygotes, and the increase was consistent in all groups Delivered babies that were about to die were morpho analyzed. It thus appears that in the context of a defi logically unremarkable but were cold and had empty ciency in both activins B, AB (which enhance FSH re stomachs, suggesting that their death was due to a ma lease in vitro) and inhibin B (which decreases it), the ternal failing. Lactogenesis, however, occurred in mu effect of inhibin deficiency is slightly dominant. The tant females. Histological analysis of mammary glands physiological significance of this modest elevation in cir in a pair of control and mutant siblings at —0.5 day post culating FSH is unknown. partum is shov^m in Figure 6. Accumulation of milk in Discussion We have created mouse lines carrying two different mu + /bb l tant alleles for the gene that encodes the activin/inhibin PB subunit. In both alleles the coding capacity (in exon 1) for approximately one-third of the pB precursor polypep tide was deleted at its amino terminus. Gray and Mason (1990) have shown that the pro-region of the activin A precursor is essential for the intracellular assembly of the homodimer and secretion of biological activity. It is not known where the dimerization activity resides within the pro-region and whether the exon 2-encoded pro-region would suffice in this function. It is likely, however, that the alteration generated loss-of-function alleles because the signal sequence and approximately half of the propeptide sequence were deleted. Protein analysis confirmed the lack of mature pB peptide in mu tant tissue. The mutation, therefore, should prevent the synthesis of activins B, AB and of inhibin B. Neverthe less, mutant mice completed embryonic development and were viable. Their birth with open eyes indicated, however, that some aspects of development were altered in the absence of a wild-type pB subunit gene. The gen eration of viable mice has furthermore uncovered a ma ternal effect by which maternal pB chain expression is Figure 6. Postpartum histology of mammary glands in mutant required for efficient perinatal survival of newborns. As a and control heterozygote siblings. {A,C] Control tissue from a result, pB-deficient females had an extremely low fecun heterozygous female. {B,D) Tissue from a mutant pB-deficient dity, despite their ability to sustain gestation up to the sibling. Tissue was taken —0.5 days after delivery. Note the time of birth. increased glandular/adipose tissue ratio, the accumulation of Recent evidence implicates an activin-like molecule milk and the vacuolation of lobules in mutant mammary tissue. as a patteming morphogen during early vertebrate devel The mutant animal had delivered a litter that was not nursed. Both animals were from a BALB-N1F2 generation. Objective opment (for review, see Kimelman et al. 1992; Sive magnification was 4x (A,B], and 20x {C,D]. 1993). Remarkably, interference with the activin path- GENES & DEVELOPMENT 421 Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Vassalli et al. Table 4. Serum FSH values in ^B-deficient mice FSH (ng/ml) Genetic Age Mouse line background Gender (weeks) + / + + /m m/m 129-134 129 male 11-16 N.D. 90.4 ± 14.6 105.7 ± 18.5 (22 = 9) (22 = 6) B6-43 (129 X B6)N1 male 11-25^ 141.6 ±33.2 172.0 ± 42.8 206.0 ± 57.7 (22 = 8) (22 = 14) (22 = 14) 6.5-14.5" 124.1 ± 62.6 110.6 ±33.8 171.0 ± 111.4 (22 = 6) (22 = 11) (22 = 11) B6-43 (129xB6)N l 12-24 78.9 ± 22.4 73.2 ± 18.1 86.4 ± 27.4 female (22 = 6) (22 = 9) (22 = 11) "Two experimental groups of males of the B6-43 line had FSH values determined in separate experiments. way by injection of an mRNA encoding a truncated ac- are bom with closed eyes, suggesting that the open eye tivin receptor into Xenopus embryos prevents mesoderm phenotype specifically results from activin deficiency. induction and formation of axial structures (Hemmati- Unfused eyelids at birth could result from either prema Brivanlou and Melton 1992). Of the two activin genes, ture (prenatal) eye opening or, perhaps more likely, from the pB subunit gene is the first expressed in Xenopus failure of fusion at El6. The event of eyelid closure de (Thomsen et al. 1990; Dohrmann et al. 1993) and chick pends on the proliferation of mesenchymal cells to effect embryos (Mitrani et al. 1990), which has led to the hy apposition and eventual fusion of the eyelids at El6. pothesis that activin B may be involved in mesoderm These cells are neural crest derivatives that populate the induction. Activin activity is already detectable in the first branchial arch (see Juriloff and Harris 1993). The unfertilized Xenopus egg (Asashima et al. 1991), but zy significance of the open eye phenotype is unclear. A va gotic expression occurs only at the late blastula stage riety of mutations are known to cause birth with open (Thomsen et al. 1990). Thus, it is not clear whether ac- eyes in mice (Lyon and Searle 1989). Some reside at xm- tivins are involved in the induction of mesoderm itself known loci and others result from targeted disruption of or, rather, in patterning and regionalization of already specific genes. The latter category includes mutants in induced mesoderm. In the mouse embryo, activin pB- the c-abl and TGF-a genes (Schwartzberg et al. 1991; specific RNA and activin 3 subunit protein have been Luetteke et al. 1993; Mann et al. 1993). The variety of detected in blastocysts and at cleavage stages, respec primary defects causing open eyes at birth suggests an tively (van den Eijnden-van Raaij et al. 1992; Albano et unspecific nature to this phenotype. The process of eye al. 1993). Survival of pB mutant mice to adulthood, lid outgrowth and fusion may be especially sensitive to therefore, indicates that activins B, AB and inhibin B are sublethal developmental perturbations. For example, it dispensable for embryonic development in the mouse. may rely on events that can only occur in a narrow win Matzuk et al. (1992) have shown previously that a sub- dow of time. Alternatively, as suggested by the fact that unit-deficient mice, lacking inhibins A and B, develop to the genes for both the pB subunit and an activin receptor term. These molecules may not play a function in early are expressed in the eye anlagen of Xenopus embryos mouse development, or overlapping pathways may exist (Dohrmann et al. 1993), a more specific effect of pB on allowing compensation for their deficit at these stages. the eye may be operating. A different mutation in the pB The pA subunit gene in particular may cover the pB subunit gene, which results in deletion of part of the deficiency in our mutants. Our finding of up-regulated mature peptide-encoding region, was also found to affect expression of the pA subunit in the pB-deficient ovary eyelid fusion in mouse embryos (H. Schrewe and T. Grid- possibly reflects a regulatory mechanism that also oper ley, pers. comtn.). ates in the mutant embryo and provides a means for Expression of the eye phenotype was affected by the normal morphogenesis in the absence of activin B. Gen genetic background. The defect was never observed in eration of an activin-less mutant by targeting the pA 129/Sv inbred mice but was seen when the mutation gene and intercrossing pA and pB mutant mice should was introduced into either of two hybrid genetic back provide definitive answers to some of these questions. In grounds {129/SVXC57BL/6 or 129/SvxBALB/c). The in addition, examination of the expression of activin imme complete penetrance observed in hybrid mice may there diate-early response genes, as Brachyury and goosecoid, fore be due to the existence of suppressor allele(s) con in the pB-deficient embryos should help to determine tributed by the 129/Sv genome. Further backcrosses of whether development proceeds normally, both tempo the mutant animals to C57BL/6 or BALB/c mice will rally and spatially. clarify this question. In addition, the occasional individ uals displaying a single affected eye suggest that not all Although pB subunit mutant embryos develop to variation is attributable to genetic factors. In contrast to term, the deficiency in either activin B, activin AB, or the eye phenotype, the reproductive failure of mutant inhibin B affects prenatal development, as revealed by a females (see below), appears fully penetrant in the 129/ defect in eyelid fusion at birth. Inhibin-deficient mice GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Activin B/inhibin B-deficient mice Sv inbred genetic background, suggesting that the sup- reported to inhibit milk ejection (Vale et al. 1990). A pressor(s) carried by the 129/Sv genome act(s) specifi local source of oxytocin expression by the epithelium of cally during embryonic development. the uterine endometrium peaks at parturition (Lefebvre et al. 1992) and is a potential target of the pB mutatio n as A large body of evidence exists for the relevance of well. Preliminary analysis of postpartum uteri suggests activins and inhibins in the biology of reproduction of morphological alterations in mutant females (A. Vassalli both males and females. Initially discovered as antago and H. Gardner unpubl.). It is interesting to note that the nistic regulators of FSH secretion from anterior pituitary mutation of another pleiotropic cytokine, leukemia in cells, activins and inhibins were also shown to exert a hibitory factor (LIE), was shown to affect specifically the variety of local effects in the gonads, their primary site of reproductive function of females by interfering with the expression in adults (for review, see Vale et al. 1990; ability of the mutant uterus to allow implantation of Mather et al. 1992). The decidual tissue, forming in the uterine wall in response to nidation of an embryo, has blastocyst embryos (Stewart et al. 1992). Various cyto also been found to express activin/inhibin p subunits. kines may be involved in the timely regulation of the Expression of the pA subunit mRNA has been demon critical physiological changes that occur in the uterus strated in the mouse uterine decidua shortly after im during pregnancy. plantation (Manova et al. 1992), and pB subunit expres Whether the pB subunit gene is active in pathways of sion in human decidua was reported to increase in the early development that display redundancy can best be course of pregnancy (Petraglia et al. 1990). It seemed pos addressed by studying the interaction of the mutated sible, therefore, that maternal peptide secreted by the gene with other, early-acting mutations. For example, decidua functions in the developing embryo to promote the presence of the pB mutation may alter the phenotype its normal development to birth. We have thus tested the of embryos heterozygous or homozygous for mutations reproductive capacity of mutant animals. Histological that affect mesoderm formation, such as the Biachyuiy examination of testes and ovaries from 3B-deficient an mutation at the T locus (Lyon and Searle 1989) or the imals had failed to detect overt abnormalities. However, 413.d mutation in the nodal gene (Zhou et al. 1993). when the fertility of mutant animals mated to heterozy When mutated in mice, the T gene causes axial deficien gous partners was assayed, it became apparent that mu cies whose severity are dependent on the dose of the tant males bred indistinguishably from heterozygous residual wild-type T function (MacMurray and Shin controls, but mutant females suffered an impaired repro 1988). Because activin is an inducer of T, and levels of T ductive performance. Although mutant females were expression affect development in a concentration-depen able to become pregnant and to carry pregnancies to the dent manner, it is possible to envisage that activin B end of gestation, they failed to raise their offspring nor deficiency would cause an altered, perhaps enhanced, T mally. Perinatal lethality appeared to affect equally ho mutant phenotype. mozygous mutant and heterozygous neonates. This pB-deficient mice may also have relevance to the demonstrated that embryonic development can be com study of tumorigenesis. a subunit mutant mice (inhibin pleted in the absence of both maternal and fetal pB pep deficient) develop gonadal tumors early in life with com tide. The viability of pB-deficient embryos is therefore plete penetrance (Matzuk et al. 1992). In contrast the pB not dependent upon maternal supply of pB peptide. Our subunit mutant mice do not develop such tumors. In experiments have, however, revealed a critical role for hibin deficient mice would be expected to have increased maternal pB expression in perinatal survival of the new- levels of effective activin activity due to the lack of an boms. tagonist activity. The involvement of the pB gene in the mechanisms of tumorigenesis can be studied in a/pB The causes that underlie the reproductive failure in double mutants. mutant females are not resolved. The newborn pups ap peared morphologically normal, suggesting that their In summary, the mutations in the pB subunit gene death was due to a defect in the delivery process or in indicate that activins B, AB and inhibin B are dispensable maternal nursing behavior. pB-deficient mothers may for embryonic development in the mouse. The g^ne is, manifest a defect in parturition, in postpartum events however, active during prenatal development, as re such as uterine remodeling, lactation, or of the hormonal vealed by the failure of eyelid closure in mutant neo milieu affecting maternal behavior. Examination of nates. Secondly activins B, AB and/or inhibin B func- mammary tissue indicated that lactogenesis was not im tion(s) later in adult female life to regulate events that paired in mutant females. The abnormal accumulation surround parturition. A deeper understanding of the of milk in mutant mammary glands suggests, however, causes underlying these phenotypes may allow future that milk let-down might be impaired. Oxytocin is a study of processes relevant to both development and re hypothalamic nonapeptide that is very potent in induc production. ing uterine contractions and controlling milk ejection. It is significant that Sawchenko et al. (1988) have demon Materials and methods strated p subunit immunoreactivity in the central neural Cloning of the mouse activin/inhibin pB subunit pathways involved in oxytocin secretion. Furthermore, gene and construction of targeting vectors it was reported that local administration of activin in the hypothalamus of rats leads to a rapid increase in plasma To clone the mouse pB subunit gene, a 129/Sv mouse strain oxytocin level and infusion of anti-p subunit serum was genomic DNA library derived from the D3 ES cell line (Doet- GENES & DEVELOPMENT 423 Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Vassalli et al. schman et al. 1985; gift of Drs. En Li and Doug Gray, Whitehead Transfection and selection of ES cells Institute) was screened with a partial rat 3B cDNA fragment The Jl ES cell line (Li et al. 1992) was established from a male (rinB-c2-P, gift of Dr. Kelly Mayo, Northwestern University, 129/Sv embryo and grown essentially as described by Robertson Evanston, IL). A total of 23 kb of genomic DNA, centered on the (1987). ES cells were cultured on a feeder layer of 7-irradiated pB subunit-coding regions, were isolated in overlapping phages embryonic fibroblasts (EFs) in DMEM supplemented with 15% and restriction-mapped. Exons were mapped by Southern blot heat-inactivated fetal bovine serum (Hyclone), nonessential hybridization with exon-specific probes (see PCR cloning of amino acids (GIBCO), lO"'* M p-mercaptoethanol and antibiot mouse pB subvmit exon probes). To construct the targeting vec ics. Leukemia inhibitory factor (LIF) was added to the medium tors, the 17-kb XI phage insert was cloned into pBlue- at 200-500 U/ml . G418'^ EFs were prepared from E13.5 embryos script(KS-) (Stratagene), digested with Narl enzyme, and reli- carrying a targeted mutation in the Pj-microglobulin gene (Zijl- gated, yielding a construct designated pXlANarl. This results in stra et al. 1989). the deletion of a 1.3-kb Narl fragment that comprises the Two T75 flasks of subconfluent Jl cells at passage 9-10 were mouse-coding regions homologous to the coding regions for trypsinized and resuspended at 2.5 x 10^ ml " Mn PBS. Linearized amino acids 1-137 of the rat precursor chain (Feng et al. 1989b). pBBl plasmid DNA was added to the cell suspension at a con For positive selection, a PGK-NEO-p(A) cassette derived from centration of 30 ii-g/val, and cells were electroporated at 250 V p(KJl) (gift of Dr. Michael Rudnicki, Whitehead Institute; Ty- and 250 fiF by use of a BTX 300 electroporator. Electroporated bulewicz et al. 1991) and ligated to Clal linkers, was inserted cells were plated on feeder layers of G418'^ EF cells at a density into the Narl site of pXlANarl. The resulting targeting vector of 5 X10*^ cells per 9-cm plate. Selection with G418 at 350 (Jig/ml containing the neo gene in the same transcriptional orientation (dry powder, GIBCO) was initiated 20-30 hr later and carried as the pB subunit gene is termed pBBl, and pBB2 contains the out for 7-10 days. At that time individual G418'^ colonies were neo cassette in reverse orientation. Both vectors have the same picked, dissociated in trypsin, and plated on wells of feeder- arms of homology, of 9.5 kb (5') and 6.0 kb (3'). covered 24-well plates. After a 3- to 4-day expansion, two-thirds The 3' breakpoint of the ANarl deletion was confirmed by of the cells in each well were frozen, and the rest further ex PCR amplification on tail DNA of a BBI allele-specific frag panded in the absence of feeders for DNA preparation. ment, "010"/ACT15b, that flanks the breakpoint of the dele The pBB2 targeting plasmid was electroporated in similar tion, using a PGK p(A) primer ("010", 5'-AACGAGATCAG- conditions, but the cells were resuspended at 5xlO^/ml in CAGCCTCTG-3') and an exon I pB primer (ACTlSb) (see PCR the presence of 25 \ig of linearized DNA and electroporated in cloning of mouse pB subunit exon probes). This yielded the one cuvette of a Bio-Rad Gene Pulser set at 800 V and 3 [iF. expected fragment of 127 bp in animals carrying a BBI mutant allele but not in -I- / -I- littermates. Moreover, the ACT 16 primer (5'-ATGGTCACGGCCCTGCGCAA-3'), which lies within the Screening of recombinant ES clones and animal genotyping ANarl deletion, was used in conjunction with the ACT 15b ES cell and tail genomic DNAs were prepared according to Laird primer to amplify a 138-bp wild-type allele-specific fragment in et al. (1991), digested with £coRI, and submitted to Southern -\- /+ and -I- /bb l but not in bbl/bbi animals, further confirm blot analysis, using a 850-bp Bglll-BamHI genomic fragment as ing the 3 ' breakpoint of the deletion. a probe (E in Fig. 1), which lies adjacent 3 ' to the right arm of the targeting constructs. Of 172 G418'^ ES clones transfected with PCR cloning of mouse pB subunit exon probes the pBBl vector, 41 were homologous recombinants, as evi denced by the presence of an 8.2-kb recombinant £coRI restric A mouse exon 1 probe, mACTI4/15, and a mouse exon 2 probe, tion fragment in addition to the ~20-kb wild-type fragment. mACTI0/Il;I3 , were cloned from D3 genomic DNA with the Further characterization of the mutated alleles in 22 targeted use of the PCR. Published rat (Esch et al. 1987; Feng et al. 1989b) clones was done by use of probe E on BamHl or ffindlll genomic and human (Mason et al. 1989) sequences were used in the digests, and by use of an internal probe corresponding to the choice of primers. Limited degeneracy was introduced in the other end of the targeting vectors (probe A, Fig. 1), as well as a primer sequences. For niACTI4/15, 700 ng of D3 genomic neo probe. Probe A is a 1.5-kb Sall-BgRl fragment corresponding DNA was amplified using ACT14 (5'-CGAATTCCAGGA- to the 5' extremity of the left arm of the vectors. These analyses CACCTGTACGTCGTG-3') and ACT15 (5'-GCGGATC- confirmed that 20 of the 22 clones had undergone the predicted CCTCTGCAAAGCTGATGAT(CT)TC-3') primers (0.2 jiM homologous recombination event with no additional event. The each) in a 50-JJL1 PCR reaction comprised of 10 mM Tris-HCl (pH same screening strategy was employed in the pBB2 transfection 9.0), at 25°C, 1.5 mM MgClj, 50 mM KCl, 0.1% Triton X-100, experiment. Ten ES clones among 58 that were screened con 100 |xg/ml of gelatin, 200 |xM dNTPs, and 1.5 units of Taq poly tained the predicted 6.4-kb recombinant £coRI fragment, indic merase (Cetus). Cycling parameters were 95°C for 1 min, 56°C ative of the reverse orientation of the PGK-NEO-p(A) cassette for 30 sec, and 72°C for 1 min (5 cycles) followed by 40 cycles of in this allele. Final washes with both probes A and E were in 95°C for 1 min and 72°C for 2 min. The 318-bp PCR product was 0.2 X SSC, 0.5% SDS at 65°C. gel purified, digested with £coRI and BamHl, and cloned into pKS. All 12 clones analyzed were identical and contained the Narl site at the homologous position to codons for amino acids Blastocyst injections and breeding of chimeras 137-138 in the rat (Feng et al. 1989b). For mACT10/Il;13, ACTIO [5'-CGGATCCGGC(CT)TGGAGTG(CT)GA(CT)GG- Embryo manipulations were carried out as described by Bradley 3'] and ACT 11 primers [5'-CGAATTCCCACACTCCTCCA- (1987). Subconfluent ES cells were trypsinized, resuspended in C(AGT)AT-CAT-3'] (0.5 |xM each) were used with the same re injection medium (45% DMEM, 45% HEPES-buffered saline, action components as above, and amplification was carried out supplemented with 10% fetal bovine serum, 10 "'^ M p-mercap by 40 cycles of 94°C for 1 min, 55°C for 1 min, and 72°C for 1 toethanol, and nonessential amino acids), and injected into ei min. The 350-bp PCR products were purified and cloned into ther C57BL/6 or BALB/c blastocyst embryos. Resulting chi pKS as above. The ACTIO and ACTl 1 primers amplify both pA meric animals of both genders were bred to C57BL/6 or BALB/c and pB gene fragments (see Mitrani et al. 1990; Thomsen et al. mates. Both female and male chimeras transmitted the mutated 1990), which are distinguishable by their restriction maps. allele to some of their progeny. To obtain the BBI mutation in GENES & DEVELOPMENT 424 Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Activin B/inhibin B-deficient mice an inbred genetic background, a chimeric male that had dem Acknowledgments onstrated high chimerism of the germ Une was bred to 129/Sv We thank Kelly Mayo, Wylie Vale, Joan Vaughan, Jennie Mather females. and Lyime Krummen for the kind gift of reagents. A.V. thanks En Li for his advice on ES cell culture; Jessie Dausman for her critical, dedicated help in blastocyst injections; Ruth Curry for RNA analysis expert animal care and Wu-teh Tu for histology; Gerry Thom- RNA was isolated from tissue by the acid guanidinium thiocy- sen for initial advice on PCR of the mature peptide coding re anate-phenol-chloroform extraction method (Chomczynski gion; Mike Frohman for initial encouragement and Peter Mo- and Sacchi 1987), electrophoresed on a 2.2 M formaldehyde, 1% mbaerts for advice on targeting; Arlene Sharpe and George Mut agarose gel and transferred onto Zetabind membrane. Blots were ter for interpretation of histology. We thank Jordan Kreidberg hybridized to the inACT10/ll;13 probe (see above) or a neo and Paul Soloway for their useful editorial comments. This re probe, consisting of a 600-bp Pstl fragment of p(KJl). search was supported by a grant from Amgen, Inc., to K.F.L. and National Institutes of Health grant R35 CA 44339-06 to R.J. The publication costs of this article were defrayed in part by Protein analysis payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 USC section Ovaries from individual 3-week-old females were homogenized 1734 solely to indicate this fact. in 50 mM HEPES (pH 7.5), 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1.5 mM MgClj, 1 mM EGTA, in the presence of 2 mM PMSF, 5 M-g/ml of chymostatin, 5 |xg/ml of pepstatin A, 5 References jjLg/ml of leupeptin, and 25 |xg/ml of aprotinin. Protease inhib itors were from Sigma. Extracts were spun and total protein in Albano, R.M., S.F. Godsave, D. Huylebroeck, K. Van Nimmen, supematants was quantitated by BCA assay (Pierce). Extracts H.V. Isaacs, J.M. Slack, and J.C. Smith. 1990. A mesoderm- were run in reducing conditions on 15% poly aery lamide-SD S inducing factor produced by WEHI-3 murine myelomono- gels and transferred to nitrocellulose. Equal loading of protein cytic leukemia cells is activin A. Development 110: 435 - was verified by Ponceau S staining of the membrane after trans 443. fer. All subsequent incubations and washes were done at room Albano, R.M., N. Groome, and J.C. Smith. 1993. Activins are temperature. Blots were blocked with 5% Carnation nonfat dry expressed in preimplantation mouse embryos and in ES and milk, 0.2% NP-40 in Tris-buffered saline [TBS; 50 mM Tris-HCl EC cells and are regulated on their differentiation. Develop (pH 7.4), 150 mM NaCl] for 30 min, and then incubated over ment 117:711-723. night in primary antibodies diluted in blocking buffer. After Asashima, M., H. Nakano, K. Shimada, K. Kinoshita, K. Ishii, H. being washed once in 0.2% NP-40 in TBS and twice in 0.1% Shibai, and N. Ueno. 1990. Mesodermal induction in early Tween 20 in TBS, blots were blocked again, incubated for 1 hr amphibian embryos by activin A (erythroid differentiation in anti-rabbit HRP-conjugate antibodies (Sigma), that had been factor). Wilhelm Roux's Arch. Dev. Biol. 198: 330-335. diluted (1:10,000) in blocking buffer. Blots were then washed as Asashima, M., H. Nakano, H. Uchiyama, H. Sugino, T. Naka- above, rinsed in TBS, and detected by enhanced chemilumines- mura, Y. Eto, D. Ejima, S. Nishimatsu, N. Ueno, and K. cence (Amersham). Kinoshita. 1991. Presence of activin (erythroid differentia tion factor) in unfertilized eggs and blastulae of Xenopus Anti-peptide affinity purified rabbit antibodies were a gener laevis. Proc. Natl. Acad. Sci. 88: 6511-6514. ous gift of Dr. Wylie Vale (Vaughan et al. 1989). The pB-specific antibodies were directed against peptide (80-112)-NH2 of the Bradley, A. 1987. Production and analysis of chimaeric mice. In mature pB subunit, and the pA-specific antibodies recognize Teratocarcinomas and embryonic stem cells: A practical peptide (81-113)-NH2 of the mature pA subunit. Both antibody approach (ed. E. J. Robertson), IRL Press, Oxford, England. samples were used at a dilution (1:300). The anti-a subvmit an Cho, K.W., B. Blumberg, H. Steinbeisser, and E. De Robertis. tibodies were directed against a human inhibin a(l-25)-Gly-Tyr 1991. Molecular nature of Spemann's organizer: The role of peptide and were used at a (1:150) dilution. Specificity was con the Xenopus homeobox gene goosecoid. Cell 67: 111 1-1120. firmed using human recombinant activins A and B, and inhibin Chomczynski, P. and N. Sacchi. 1987. Single-step method of A as standards run in parallel in Western blotting assays. Hu RNA isolation by acid guanidinium thiocyanate-phenol- man recombinant proteins were kindly provided by Drs. Jennie chloroform extraction. Anal. Biochem. 162: 156-159. Mather and Lynne Krummen (Genentech). Dale, L., G. Howes, B.M. Price, and J.C. Smith. 1992. Bone mor- phogenetic protein 4: A ventralizing factor in early Xenopus development. Development 115: 573-585. Histological analysis de Jong, F., A.J. Grootenhuis, LA. Klaij, and W. Van Beurden. 1990. Inhibin and related proteins: Localization, regulation, Neonatal mice were fixed in 10% buffered formalin. Heads were and effects. Adv. Exp. Med. Biol. 274: 271-293. dehydrated, paraffin embedded, and serially sectioned in the Doetschman, T.C., H. Eistetter, M. Katz, W. Schmidt, and R. transverse plane. Sections were stained with hematoxylin and Kemler. 1985. The in vitro development of blastocyst-de- eosin. Abdominal mammary glands were processed similarly rived embryonic stem cell lines: Formation of visceral yolk and sectioned longitudinally. sac, blood islands and myocardium. /. Embryol. Exp. Mor- phol. 87: 27-45. Dohrmann, C.E., A. Hemmati-Brivanlou, G.H. Thomsen, A. PSH measurements Fields, T.M. Woolf, and D.A. Melton. 1993. Expression of Blood (0.5 ml) was taken by retro-ocular bleeding of animals activin mRNA during early development in Xenopus laevis. that had been ether anesthetized. All mice used in the study Dev. Biol. 157: 474-483. shown in Table 4 were nonbreeding animals. Serum was pre Esch, F.S., S. Shimasaki, K. Cooksey, M. Mercado, A.J. Mason, pared and FSH values were determined by radioimmunoassay, S.Y. Ying, N. Ueno, and N. Ling. 1987. Complementary as described by Matzuk et al. (1992). deoxyribonucleic acid (cDNA) cloning and DNA sequence GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Vassalli et al. analysis of rat ovarian inhibins. Mol. Endocrinol. 1: 388 - Zingg. 1992. Oxytocin gene expression in rat uterus. Science 396. 256:1553-1555. Feng, Z.M., C.W. Bardin, and C.L. Chen. 1989a. Characteriza Li, E., T.H. Bestor, and R. Jaenisch. 1992. Targeted mutation of tion and regulation of testicular inhibin beta-subunit the DNA methyltransferase gene results in embryonic le- mRNA. Mol. Endocrinol. 3: 939-948. thahty. Cell 69: 915-926. Feng, Z.M., Y.P. Li, and C.L. Chen. 1989b. Analysis of the 5'- Ling, N., S.Y. Ying, N. Ueno, S. Shimasaki, F. Esch, M. Hotta, flanking regions of rat inhibin alpha- and beta-B-subunit and R. Guillemin. 1986. Pituitary FSH is released by a het- genes suggests two different regulatory mechanisms. Mol. erodimer of the beta-subunits from the two forms of inhibin. Endocrinol. 3: 1914-1925. Nature 321: 779-782. Luetteke, N.C., T.H. Qiu, R.L. Peiffer, P. Oliver, O. Smithies, Findlay, J.K. 1993. An update on the roles of inhibin, activin, and D.C. Lee. 1993. TGF alpha deficiency results in hair and foUistatin as local regulators of foUiculogenesis. Biol. Reprod. 48: 15-23. follicle and eye abnormalities in targeted and waved-1 mice. Cell 73: 263-278. Gray, A.M. and A.J. Mason. 1990. Requirement for activin A and transforming growth factor-beta 1 pro-regions in ho- Lyon, M. and A.G. Searle, eds. 1989. Genetic variants and modimer assembly. Science 247: 1328-1330. strains of the laboratory mouse. Oxford University Press, Green, J.B., H.V. New, and J.C. Smith. 1992. Responses of em Oxford, England. bryonic Xenopus cells to activin and FGF are separated by MacMurray, A. and H.S. Shin. 1988. The antimorphic nature of multiple dose thresholds and correspond to distinct axes of the Tc allele at the mouse T locus. Genetics 120: 545-550. the mesoderm. Cell 71: 731-739. Mann, G.B., K.J. Fowler, A. Gabriel, E.G. Nice, R.L. Williams, Hashimoto, M., S. Kondo, T. Sakurai, Y. Etoh, H. Shibai, and M. and A.R. Duim. 1993. Mice with a null mutation of the TGF Muramatsu. 1990. Activin/EDF as an inhibitor of neural dif alpha gene have abnormal skin architecture, wavy hair, and ferentiation. Biochem. Biophys. Res. Commim. 173: 193- curly whiskers and often develop corneal inflammation. Cell 200. 73: 249-261. Manova, K., B.V. Paynton, and R.F. Bachvarova. 1992. Expres Hemmati-Brivanlou, A. and D.A. Melton. 1992. A truncated activin receptor inhibits mesoderm induction and formation sion of activins and TGF beta 1 and beta 2 RNAs in early of axial structures in Xenopus embryos. Nature 359: 609 - postimplantation mouse embryos and uterine decidua. 614. Mech. Dev. 36: 141-152. Herrmaim, B.G., S. Labeit, A. Poustka, T.R. King, and H. Le- Mason, A.J., L.M. Berkemeier, C.H. Schmelzer, and R.H. Schwall. 1989. Activin B: Precursor sequences, genomic hrach. 1990. Cloning of the T gene required in mesoderm structure and in vitro activities. Mol. Endocrinol. 3: 1352- formation in the mouse. Nature 343: 617-622. Jones, CM., K.M. Lyons, P.M. Lapan, C.V. Wright, and B.L. Hogan. 1992. DVR-4 (bone morphogenetic protein-4) as a Mather, J.P., T.K. Woodruff, and L.A. Krummen. 1992. Para posterior-ventralizing factor in Xenopus mesoderm induc crine regulation of reproductive function by inhibin and ac tion. Development 115: 639-647. tivin. Proc. Soc. Exp. Biol. Med. 201: 1-15. Juriloff, D.M. and M.J. Harris. 1993. Retinoic acid, cortisone, or Matzuk, M.M., M.J. Finegold, J.G. Su, A.J. Hsueh, and A. Brad thyroxine suppresses the mutant phenotype of the eyelid ley. 1992. Alpha-inhibin is a tumour-suppressor gene with development mutation, IgMl, in mice. /. Exp. Zool. gonadal specificity in mice. Nature 360: 313-319. 265: 144-152. McBumey, M.W., L.C. Sutherland, C.N. Adra, B. Leclair, M.A. Kaipia, A., T.L. Penttila, S. Shimasaki, N. Ling, M. Parvinen, Rudnicki, and K. Jardine. 1991. The mouse Pgk-1 gene pro and J. Toppari. 1992. Expression of inhibin beta A and beta B, moter contains an upstream activator sequence. Nucleic Ac follistatin and activin-A receptor messenger ribonucleic ac ids Res. 19: 5755-5761. ids in the rat seminiferous epithelimn. Endocrinology Meunier, H., C. Rivier, R.M. Evans, and W. Vale. 1988. Gonadal 131: 2703-2710. and extragonadal expression of inhibin alpha, beta A, and beta B subunits in various tissues predicts diverse functions. Kaipia, A., M. Parvinen, and J. Toppari. 1993. Localization of Proc. Natl. Acad. Sci. 85: 247-251. activin receptor (ActR-IIB2) mRNA in the rat seminiferous Mitrani, E., T. Ziv, G. Thomsen, Y. Shimoni, D.A. Melton, and epithelium. Endocrinology 132: 477-479. A. Bril. 1990. Activin can induce the formation of axial Kimelman, D. and M. Kirschner. 1987. Synergistic induction of structures and is expressed in the hypoblast of the chick. mesoderm by FGF and TGF-beta and the identification of an mRNA coding for FGF in the early Xenopus embryo. Cell CeU 63: 495-501. 51: 869-877. Murata, M., Y. Eto, H. Shibai, M. Sakai, and M. Muramatsu. 1988. Erythroid differentiation factor is encoded by the same Kimelman, D., J.L. Christian, and R.T. Moon. 1992. Synergistic mRNA as that of the inhibin pA chain. Proc. Natl. Acad. Sci. principles of development: Overlapping patterning systems 85: 2434-2438. in Xenopus mesoderm induction. Development 116: 1-9. Koster, M., S. Plessow, J.H. Clement, A. Lorenz, H. Tiedemann, Petraglia, F., L. Calza, G.C. Garuti, M. Abrate, L. Giardino, A.R. and W. Knochel. 1991. Bone morphogenetic protein 4 (BMP- Genazzani, W. Vale, and H. Meunier. 1990. Presence and 4), a member of the TGF-beta family, in early embryos of synthesis of inhibin subunits in human decidua. /. Chn. En docrinol. Metab. 71: 487-492. Xenopus laevis: Analysis of mesoderm inducing activity. Mech. Dev. 33: 191-199. Robertson, E.J., ed. 1987. Embryo-derived stem cell lines. In Laird, P.W., A. Zijderveld, K. Linders, M.A. Rudnicki, R. Jae- Teratocarcinomas and embryonic stem cells: A practical nisch, and A. Bems. 1991. Simplified mammahan DNA iso approach. IRL Press, Oxford, England. lation procedure. Nucleic Acids Res. 19: 4293. Rugh, R. 1990. The mouse: Its reproduction and development. Lee, K.F., E. Li, L.J. Huber, S.C. Landis, A.H. Sharpe, M.V. Chao, Oxford University Press, Oxford, England. and R. Jaenisch. 1992. Targeted mutation of the gene encod Sawchenko, P.E., P.M. Plotsky, S.W. Pfeiffer, E.J. Curmingham, ing the low affinity NGF receptor p75 leads to deficits in the J. Vaughan, J. Rivier, and W. Vale. 1988. Inhibin beta in cen peripheral sensory nervous system. Cell 69: 737-749. tral neural pathways involved in the control of oxytocin se cretion. Nature 334: 615-617. Lefebvre, D.L., A. Giaid, H. Bennett, R. Lariviere, and H.H. 426 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Activin B/inhibin B-deficient mice Schubert, D., H. Kimura, CM . La, J. Vaughan, D. Karr, and W.H. endoderm-specific FGF-like factor and inhibited by activin Fischer. 1990. Activin is a nerve cell survival molecule. Na A. Mech. Dev. 33: 157-165. ture 344:868-870. van den Eijnden-van Raaij, A.J., A. Feijen, K.A. Lawson, and C.L. Schwartzberg, P.L., A.M. Stall, J.D. Hardin, K.S. Bowdish, T. Mummery. 1992. Differential expression of inhibin subunits Humaran, S. Boast, M.L. Harbison, E.J. Robertson, and S.P. and foUistatin, but not of activin receptor type II, during Goff. 1991. Mice homozygous for the ablml mutation show early murine embryonic development. Dev. Biol. 154: 356 - poor viability and depletion of selected B and T cell popula 365. tions. Cell 65: 1165-1175. Vaughan, J., J. Rivier, A.Z. Corrigan, R. McClintock, C.A. Campen, D. JoUey, J.K. Voglmayr, C.W. Bardin, C. Rivier, Sive, H.L. 1993. The frog prince-ss: A molecular formula for and W. Vale. 1989. Detection and purification of inhibin dorsoventral patterning in Xenopus. Genes &. Dev. 7: 1-12. using antisera generated against synthetic peptide frag Slack, J.M., B.G. Darlington, J.K. Heath, and S.F. Godsave. 1987. ments. In Hormone action: Part K neuroendocrine peptides Mesoderm induction in early Xenopus embryos by heparin- binding growth factors. Nature 326: 197-200. (ed. P.M. Conn), pp. 588-617. Academic Press, New York. Smith, J.C. and J.E. Howard. 1992. Mesoderm-inducing factors Zhou, X., H. Sasaki, L. Lowe, B.L. Hogan, and M.R. Kuehn. and the control of gastrulation. Development (Suppl.) 127- 1993. Nodal is a novel TGF-beta-like gene expressed in the 136. mouse node during gastrulation. Nature 361: 543-547. Smith, J.C., B.M. Price, K. Van Nimmen, and D. Huylebroeck. Zijlstra, M., E. Li, F. Sajjadi, S. Subramani, and R. Jaenisch. 1989. 1990. Identification of a potent Xenopus mesoderm-inducing Germ-line transmission of a disrupted Pj-^^icroglobulin factor as a homologue of activin A. Nature 345: 729-731. gene produced by homologous recombination in embryonic Smith, J.C., B.M. Price, J.B. Green, D. Weigel, and B.G. Her stem cells. Nature 342: 435-438. rmann. 1991. Expression of a Xenopus homolog of Brachyury (T) is an immediate-early response to mesoderm induction. Cell 67: 79-87. Smith, W.C. and R.M. Harland. 1991. Injected Xwnt-8 RNA acts early in Xenopus embryos to promote formation of a vegetal dorsalizing center. Cell 67:753-765. . 1992. Expression cloning of noggin, a new dorsalizing factor localized to the Spemann organizer in Xenopus em bryos. Cell 70: 829-840. Sokol, S., J.L. Christian, R.T. Moon, and D.A. Melton. 1991. Injected Wnt RNA induces a complete body axis in Xenopus embryos. Cell 67: 741-752. Stewart, C.L., P. Kaspar, L.J. Brunet, H. Bhatt, I. Gadi, F. Kont- gen, and S.J. Abbondanzo. 1992. Blastocyst implantation de pends on maternal expression of leukaemia inhibitory fac tor. Nature 359: 76-79. Thomas, K.R. and M.R. Capecchi. 1987. Site-directed mutagen esis by gene targeting in mouse embryo-derived stem cells. CeJi 51: 503-512. Thomsen, G.H. and D.A. Melton. 1993. Processed Vgl protein is an axial mesoderm inducer in Xenopus. Cell 74: 433-441. Thomsen, G., T. Woolf, M. Whitman, S. Sokol, J. Vaughan, W. Vale, and D.A. Melton. 1990. Activins are expressed early in Xenopus embryogenesis and can induce axial mesoderm and anterior structures. Cell 63: 485-493. Tybulewicz, V.L., C.E. Crawford, P.K. Jackson, R.T. Bronson, and R.C. Mulligan. 1991. Neonatal lethality and lymphope nia in mice with a homozygous disruption of the c-abl proto- oncogene. Cell 65: 1153-1163. Vale, W., J. Rivier, J. Vaughan, R. McClintock, A. Corrigan, W. Woo, D. Karr, and J. Spiess. 1986. Purification and charac terization of an FSH releasing protein from porcine ovarian follicular fluid. Nature 321: 776-779. Vale, W., A. Hsueh, C. Rivier, and J. Yu. 1990. The inhibin/ activin family of hormones and growth factors. In Peptide growth factors and their receptors: Handbook of experimen tal pharmacology (ed. M.A. Spom and A.B. Roberts), pp. 211-248. Springer-Verlag, Berlin, Germany. van den Eijnden-Van Raaij, A., E. van Zoelent, K. van Nimmen, C.H. Koster, G.T. Snoek, A.J. Durston, and D. Huylebroeck. 1990. Activin-like factor from a Xenopus laevis cell line re sponsible for mesoderm induction. Nature 345: 732-734. van den Eijnden-van Raaij, A., T. van Achterberg, C. van der Kruijssen, A.H. Piersma, D. Huylebroeck, S. de Laat, and C.L. Mummery. 1991. Differentiation of aggregated murine P19 embryonal carcinoma cells is induced by a novel visceral GENES & DEVELOPMENT 427 Downloaded from genesdev.cshlp.org on October 24, 2021 - Published by Cold Spring Harbor Laboratory Press Activin/inhibin beta B subunit gene disruption leads to defects in eyelid development and female reproduction. A Vassalli, M M Matzuk, H A Gardner, et al. Genes Dev. 1994, 8: Access the most recent version at doi:10.1101/gad.8.4.414 This article cites 64 articles, 11 of which can be accessed free at: References http://genesdev.cshlp.org/content/8/4/414.full.html#ref-list-1 License Receive free email alerts when new articles cite this article - sign up in the box at the top Email Alerting right corner of the article or click here. Service Copyright © Cold Spring Harbor Laboratory Press
Journal
Genes & Development – Unpaywall
Published: Feb 15, 1994