Downloaded from genesdev.cshlp.org on October 5, 2021 - Published by Cold Spring Harbor Laboratory Press

Patterns of expression of murine Vgr-1 and BMP-2a RNA suggest that transforming growth factor-p-like genes coordinately regulate aspects of embryonic development

Karen M. Lyons, Ron W. Pehon, and Brigid L.M. Hogan Department of Cell Biology, Vanderbilt University, Nashville, Tennessee 37232 USA

The murine Vgr-1 (Vg-related) and BMP-2a (bone morphogenetic protein 2a) genes are members of the decapentaplegic subgroup of the transforming growth factor-p (TGF3) superfamily. Although genetic and biochemical studies suggest that the members of this subgroup play important roles in development, little is known about their function in mammals. Therefore, we investigated the expression of Vgr-1 and BMP-2a RNAs in embryonic, newborn, and adult tissues by in situ hybridization. Vgr-1 RNA is maternally encoded in ovarian oocytes but declines in fertilized eggs and is undetectable by the two- to four-cell stage. Only low levels of transcripts are seen in blastocysts and early postimplantation stages. From mid-gestation on, Vgr-1 RNA is expressed at high levels in developing skin, especially in the suprabasal cells of the proliferating epidermis but not in the dermis or hair follicles, both of which contain TGPpi and/or TGFp2 RNAs. In contrast, BMP-2a transcripts are seen only in the hair follicles in the cells of the hair bulb cortex. Temporally and spatially distinct patterns of BMP-2a, Vgr-1, TGPpi, and TGFP2 expression are also seen in different populations of mesenchymal cells in the developing skeletal system (cartilage and bone). Our results suggest that the coordinated expression of several members of the TGFp superfamily is required to control the progression of specific cell types through their differentiation pathways. [Key Words: Vgr-1; BMP-2a; TGFp; in sitU; development] Received July 24, 1989; revised version accepted September 4, 1989.

The transforming growth factor-p (TGFp) superfamily Lehnert and Akhurst 1988; Sandberg et al. 1988; contains at least five gene products closely related to Flanders et al. 1989; Pelton et al. 1989; Thompson et al. TGFpi, as well as several more distantly related mole­ 1989). Finally, the observations that porcine TGFpi, in cules such as Miillerian inhibiting substance (MIS), bone combination with basic fibroblast growth factor (bFGF), morphogenetic proteins, the Drosophila decapenta­ and porcine TGFp2 alone can induce the expression of plegic [dpp] gene product, and the Xenopus Vgl gene mesoderm specific markers in isolated Xenopus animal product. Numerous studies have indicated that cap cells provide additional evidence that members of members of both subgroups play important regulatory this subgroup regulate developmental events in vivo roles in growth and development. TGFpi and TGFp2 (Kimmelman and Kirschner 1987; Rosa et al. 1988). have been isolated from a variety of adult tissues and can Several members of the second subgroup, which show either stimulate or inhibit cell proliferation and differ­ greater homology to the Drosophila dpp gene product entiation and have positive or negative effects on spe­ than to TGFpi, also have critical roles in embryo- cific gene expression in model systems in vitro (see, e.g., genesis. Drosophila embryos homozygous for null al­ Moses et al. 1985; Roberts et al. 1985; Ignotz and Mas- leles of dpp are completely ventralized (Irish and Gelbart sague 1986; Edwards et al. 1987; Lund et al. 1987; Coffey 1987). Transcripts of the dpp gene are detected in the et al. 1988; D.M. Rosen et al. 1988; Russell et al. 1988). embryo and larva, predominantly in the dorsal ectoderm In addition, several in vivo studies have demonstrated and epithelial imaginal discs but also in the visceral me­ that TGFpi affects the growth and differentiation of soderm and gut endoderm, suggesting that the dpp gene adult tissues (Roberts et al. 1986; Silberstein and Daniel product participates in a number of morphogenetic 1987; Daniel et al. 1989), and immimohistochemical and events (St. Johnston and Gelbart 1987). The closely re­ in situ hybridization studies further support the notion lated bovine bone morphogenetic protein 2a (BMP-2a) that TGFpi and TGFp2 are involved in regulating the induces ectopic cartilage formation in vivo, raising the morphogenesis of embryonic tissues (Heine et al. 1987; possibility that this protein is involved in bone forma-

GENES & DEVELOPMENT 3:1657-1668 © 1989 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/89 $1.00 1657 Downloaded from genesdev.cshlp.org on October 5, 2021 - Published by Cold Spring Harbor Laboratory Press

Lyons et al.

tion and remodeling during mammalian embryogenesis indicates that the fertilized eggs contain 50-70% the (Wozney et al. 1988). The Vgl gene product was isolated level of the Vgr-1 RNA present in ovarian oocytes with as a maternally encoded mRNA that becomes progres­ an intact germinal vesicle. Vgr-1 RNA is undetectable at sively localized to the vegetal hemisphere of Xenopus the two- to four-cell stage (Fig. 1F,G), and the 8- to 16- embryos and then declines in level after gastrulation cell stage (data not shown) but a low level of signal is (Rebagliati et al. 1985). Several different signals origi­ present (Fig. 1H,I) in 4.5-day p.c. blastocysts. Hybridiza­ nating in the vegetal hemisphere are thought to induce tion to the Vgr-1 riboprobe is also detectable in 6.5- to mesoderm formation in the overlying animal pole cells 12.5-day p.c. embryos, as judged by comparison of in a complex process known as mesoderm induction overall grain density in sections hybridized with sense (Dale and Slack 1987; for review, see Smith 1989). On and antisense RNA. However, there appears to be no the basis of its temporal and spatial expression, Vgl is a specific localization to a particular germ layer or region candidate for a mesoderm-inducing factor. (data not shown). Begirming at day 13.5 p.c, Vgr-1 tran­ To study the potential roles of members of the dpp scripts could be detected in specific tissues, as discussed subgroup during murine development, an 8.5-day post- in the following sections. coitum [p.c.) embryo cDNA library was screened with a partial cDNA from the Xenopus Vgl gene. A cDNA, des­ ignated Vgr-1 (for Vg-related) was isolated. The carboxy- Expression of Vgr-1 and BMP-2a RNAs in epithelia terminal portion of the predicted protein encoded by the A preliminary survey of embryonic, newborn, and adult Vgr-1 cDNA shows 78% and 77% similarities to the tissues showed that Vgr-1 RNA is localized predomi­ corresponding regions of the Xenopus Vgl and Dioso- nantly, but not exclusively, in stratified squamous epi­ phila dpp gene products, but only 45% similarity to the thelia. For example, it is present in the suprabasal layers corresponding portion of TGFpi (Lyons et al. 1989). of the developing skin (Fig. 2) and in adult cervix, vagina, Thus, Vgr-1 belongs to the subgroup of dpp-like genes and esophagus (data not shown). A particularly high within the TGFp superfamily. level of Vgr-1 RNA expression is seen in the stratified Previous Northern analyses showed that Vgr-1 RNA is squamous epithelium of the adult forestomach (Fig. expressed at many stages throughout development in 3A-D). Furthermore, a sharp discontinuity in Vgr-1 ex­ specific tissues. (Lyons et al. 1989). Here we examine pression occurs within a few cells at the squamo-co- the spatial distribution of these transcripts by in situ hy­ lumnar jimction between the forestomach and the gas­ bridization. We show that Vgr-1 RNA is maternally en­ tric epithelium (Fig. 3E,F, arrowheads). Not all stratified coded and is present in primary oocytes and fertilized squamous epithelia express Vgr-1 RNA at high levels, eggs but is not detected in cleavage-stage embryos. In however, and a much lower level of hybridization, con­ postimplantation embryos, Vgr-1 RNA is present in fined to the uppermost layers of the epithelium, is many of the same tissues shown previously to contain present in the developing palates, lip (Fig. 2G,H,K,L), TGFpi and TGFp2 transcripts but with a temporally and tongue (data not shown). and spatially distinct pattern of expression. Vgr-1 also To investigate the possible roles of members of the has a different pattem of expression from BMP-2a, an­ dpp subfamily in epidermal morphogenesis in more other member of the dpp subfamily. These observations detail, we examined the expression of both Vgr-1 raise the possibility that the coordinated expression of and BMP-2a transcripts in developing skin from 13.5 different members of the TGFp superfamily is required days p.c. to birth, and in adults. Vgr-1 expression is not to regulate, by autocrine and paracrine feedback mecha­ seen at 13.5 days p.c. (Fig. 2A,B) but is first detected at nisms, the progression of specific subpopulations of cells 15.5 days p.c. within the thickening epidermis (Fig. along specific differentiation pathways during mamma­ 2C,D). Expression continues in the suprabasal layers lian development. throughout embryogenesis and in newborns (Fig. 2C,D, M-P). Neither the hair follicles nor the dermis show hy­ Results bridization at any stage of development. Vgr-1 RNA could not be detected in adult dorsal epidermis (data not Expression of Vgr-1 RNA in oocytes and early embryos shown). This pattem of expression is in contrast to that To characterize Vgr-1 expression at early stages of mu­ of TGFP2 (Pelton et al. 1989) and BMP-2a in developing rine development, we localized RNA in oocytes, fertil­ epidermis. As shown in Figure 2, E and F, TGFp2 RNA is ized eggs, and pre- and postimplantation embryos by in first detected at 15.5 days p.c. within the dermis. At 17.5 situ hybridization. In the adult ovary, Vgr-1 RNA is days p.c, TGFp2 expression in the dermis begins to de­ present at high levels in both immature and mature oo­ cline, and by 18.5 days p.c, transcripts appear in the epi­ cytes (Fig. 1A,B). A much lower level of hybridization dermis and in the outer root sheath of the hair follicles. can be detected in the somatic follicle cells surrounding This pattem of TGFp2 expression persists at least the developing oocytes. Northern analysis (Fig. IE) through 3 days postpartum (Pelton et al. 1989). shows that the Vgr-1 transcripts in oocytes and follicle A very different pattem of expression is observed cells are slightly larger than those detected in other cell when in situ hybridization is carried out using a BMP-2a types. Vgr-1 transcripts are present in fertilized eggs at a antisense probe. Expression is not seen at any stage in level somewhat lower than in oocytes (Fig. 1C,D). the interfoUicular epidermis but is confined to devel­ Quantization of the density of autoradiographic grains oping hair and whisker follicles (Lyons et al. 1990; data

1658 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on October 5, 2021 - Published by Cold Spring Harbor Laboratory Press

Expression of Vgr-1 and BMP-2a

GENES & DEVELOPMENT 1659 Downloaded from genesdev.cshlp.org on October 5, 2021 - Published by Cold Spring Harbor Laboratory Press

Lyons et al.

1660 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on October 5, 2021 - Published by Cold Spring Harbor Laboratory Press

Expiession of Vgr-1 and BMP-2a not shown). Expression in the hair folHcles persists regulated during the estrous cycle, because no Vgr-1 throughout aduhhood and appears to be locahzed specif­ RNA is seen in some female mice. However, we have ically to the cortex/inner root sheath of the hair shaft not attempted to correlate the level of Vgr-1 expression (Fig. 4). with specific stages of the estrous cycle.

Expression of Vgr-1 RNA in mesenchymal tissues Expression of Vgr-1 RNA in the brain Vgr-1 transcripts are present in some tissues of meso­ In addition to the expression of Vgr-1 in the meninges of dermal origin. In particular, high levels are seen in hy­ the brain, Vgr-1 transcripts are detected at low levels pertrophic cartilage throughout the embryonic skeletal throughout the neural tissue of newborn and adult system, including the developing limbs, vertebral brains. However, Vgr-1 transcripts are present at a column, ribs, and head. higher level in the pyramidal layer of the hippocampus, Examination of developing limbs from 11.5 to 18.5 with increased expression in the adult compared to the days p.c. showed strikingly different patterns of localiza­ 16.5-day p.c. brain (data not shown). tion for BMP-2a and Vgr-1. As reported previously (V. Rosen et al. 1988), BMP-2a RNA is present in the precartilaginous mesenchyme of the limbs. Figure 5, A Discussion and B, shows that BMP-2a transcripts are localized in prevertebrae of 12.5-day p.c. embryos. Upon examina­ The observations reported here point to multiple roles tion of 13.5- to 18.5-day p.c. developing limbs, a low for the Vgr-1 gene product in development. Vgr-1 tran­ level of BMP-2a expression is also present in the perios­ scripts are present at many stages of embryogenesis, in teum and osteogenic zone (data not shown). Vgr-1 RNA the derivatives of all three germ layers and in both the is first detected at 13.5 days p.c. and is specifically local­ mesenchymal and epithelial components of different ized to the cells of the hypertrophic cartilage. This local­ tissues. In addition to its potential functions in develop­ ization continues throughout development and in new­ ment, the presence of Vgr-1 RNA in adult tissues sug­ borns (Fig. 5C-F). No Vgr-1 expression is seen in the gests a role in the regulation of the growth and differen­ perichondrium, periosteum, osteoblasts, osteoclasts, os- tiation of established systems, particularly epithelial. teocytes, or tendon, all of which have been shown to It is clear from Figure 1 that the mouse oocyte con­ contain TGFpl and/or TGFp2 transcripts (Lehnert and tains maternally encoded Vgr-1 transcripts. Similarly, Akhurst 1988; Sandberg et al. 1988; Pelton et al. 1989). Vgl RNA is maternally encoded in Xenopus oocytes. Expression of Vgr-1 is also seen in developing and However, in contrast to the persistence of Xenopus Vgl adult meninges. No expression is apparent in the me­ RNA into the gastrulation stage (Rebagliati et al. 1985), ninges of 12.5-day p.c. embryos, which do not contain a the level of Vgr-1 RNA in the mouse begins to decline in distinct arachnoid layer at this stage. In the newborn and fertilized eggs and is undetectable by the two- to four- in adults, Vgr-1 transcripts are abundant in all three cell stage. This observation follows a general pattern of layers of the meninges surrounding the brain and spinal mammalian oocyte RNA turnover (Bachvarova et al. cord (data not shown). 1985) in which meiotic maturation results in the deple­ Vgr-1 transcripts are also detected in the stromal com­ tion of approximately half of the poly(A)"^ RNA. This ponent of the uterine endometrium in adult female mice loss continues rapidly through the two-cell stage (Piko (data not shown). Uterine expression of Vgr-1 may be and Clegg 1982; DeLeon 1983). Substantial decreases in

Figure 2. Expression of Vgr-1 RNA in embryonic and newborn skin. Bright-field (A, C, E, I, K, M, and O) and corresponding dark- ground (B, D, F, H, J, L, N, and P) micrographs of sections hybridized with Vgr-1 antisense probe. [A] Section through dorsal skin of a day-13.5 p.c. embryo hybridized with Vgr-1 antisense probe. Bar, 100 [xm. [B] Same section as in A, showing that no Vgr-1 signal is seen at this stage in either the epidermis (upper arrowhead) or dermis. (C) Section through dorsal skin from a day-15.5 p.c. embryo hybridized with Vgr-1 antisense probe. (D) Same section as in C, showing strong signal in the suprabasal layers of the epidermis (upper arrowhead), whereas no specific signal is seen in the dermis or the hair follicles (lower arrowhead). (£) A dorsal skin section similar to C taken from a different day-15.5 p.c. embryo and hybridized with TGF(32 antisense probe. (F) Same section as in E, showing strong signal in the dermis but no signal in the epidermis (upper arrowhead) or hair follicle (lower arrowhead). (G) Section through the upper snout of a day-16.5 p.c. embryo hybridized with Vgr-1 antisense probe. Bar, 1 mm. [H] Same section as in G, showing strong signal in the epidermis. No specific signal is seen elsewhere. (I) Enlarged region (upper box) from the section seen in G. [J] Same section as in /, showing abundant Vgr-1 signal to be present in the suprabasal layers of the epidermis (upper arrowhead) and absent from basal cells (lower right arrowhead), hair follicles (lower left arrowhead) and dermis. {K) Enlarged region (lower box) from the section seen in G. (L) Same section as in K, showing low Vgr-1 signal confined to the uppermost layers of the lip epidermis (arrowhead). (M) Section of dorsal skin from a newborn pup hybridized with Vgr-1 antisense probe. [N] Same section as in M, showing hybridization in the suprabasal epidermis (upper arrowhead). The basal cells, hair follicles (lower arrowheads), and dermis show no Vgr-1 signal. (O) Adjacent section to M hybridized with TGFp2 sense (negative control) probe. (P) Same section as in O, showing that no specific signal is seen in the negative control. The apparent signal at the edge of the epidermis (upper arrowhead) is due to light scattering by comified cells. Sections were exposed for 10-20 days, (epi) Epidermis; (dm) dermis; (hf) hair follicle; (sm) smooth muscle; (be) basal cells; (nc) nasal cavity; (pi) palate.

GENES & DEVELOPMENT 1661 Downloaded from genesdev.cshlp.org on October 5, 2021 - Published by Cold Spring Harbor Laboratory Press

Lyons et al.

1662 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on October 5, 2021 - Published by Cold Spring Harbor Laboratory Press

Expression of Vgi-1 and BMP-2a

,0r?^.v,: M

'•/i-'

hf :^.

Figure 4. Expression of BMP-2a RNA in adult hair follicles. Bright-field [left] and corresponding dark-ground [light] photomicro­ graphs of adult sections through skin after in situ hybridization, autoradiography, and staining. [A] Section through adult skin with hair follicles (lower arrowheads) hybridized with BMP-2a antisense probe. Bar, 200 M,m. [B] Same section as in A, showing signal in the hair follicles. (C) Adjacent section to A hybridized with BMP-2a sense (control) probe. (D) Same section as in C, illustrating absence of signal in the hair follicles (lower arrowheads) with the control probe. (£) Same section as in A at higher power. Bar, 100 \im.. (F) Same section as in £, showing signal in the cortex/inner root sheath of the hair follicle. Sections were exposed for 10-20 days, (epi) Epidermis; (hf) hair follicle; (hb) hair bulb. the levels of several specific maternally encoded tran­ be expressed at high levels in mouse embryos, either in scripts, including c-mos (Mutter et al. 1988) and tissue- the extraembryonic endoderm (hypoblast) underlying type plasminogen activator (t-PA) (Huarte et al. 1987), by the primitive ectoderm (epiblast) or in the presumptive the two-cell stage have also been demonstrated. Elucida­ endoderm generated from the primitive ectoderm; how­ tion of the role that the Vgr-1 gene product plays in oo­ ever, Vgr-1 RNA is present at low levels in all identifi­ cytes and fertilized eggs must await the isolation of spe­ able layers (extraembryonic endoderm, embryonic ecto­ cific antibodies and active Vgr-1 protein, as well as the derm, and mesoderm). This observation appears to argue localization of specific receptors. However, it is conceiv­ against a role for Vgr-1 as a natural mesoderm inducer in able that the Vgr-1 gene product regulates the growth early mouse embryos, although the possibility remains and/or final maturation of female germ cells in an auto­ that the active Vgr-1 protein and/or its receptor are more crine maimer. In support of this notion, two other localized. members of the TGFp superfamily, TGF(il and MIS In contrast to its generalized distribution in early pos­ have been shov^m to promote or inhibit, respectively, oo­ timplantation embryos, Vgr-1 RNA has a very localized cyte meiosis (Donahoe et al. 1987; Feng et al. 1988). pattern of expression in later mid-gestation embryos, In early cleavage-stage embryos, Vgr-1 RNA either is newborns, and adults. A primary site of expression is in not present or is produced at levels too low to be de­ stratified squamous epithelia and a sharp discontinuity tected by in situ hybridization, but low levels of tran­ in the level of Vgr-1 RNA is seen at the jimction be­ scripts are seen in 4.5-day p.c. blastocysts. The expres­ tween glandular and squamous epithelia, for example, in sion of Vgr-1 in early postimplantation embryos at a the adult stomach (Fig. 3). Experiments are in progress to time when the three embryonic germ layers are being test whether this difference in expression is regulated by established is consistent with a role for the gene product signals from the underlying mesenchyme. Within strati­ in inductive tissue interactions. By analogy with the dis­ fied squamous epithelia, the level of Vgr-1 RNA appears tribution of Vg-1 RNA in Xenopus embryos and the pos­ to be related to the degree of keratinization. This is il­ sible role of the gene product in mesoderm induction lustrated, for example, in Figure 2, G-L, where the (Rebagliati et al. 1985), one might expect Vgr-1 RNA to dorsal epidermis of the upper snout contains high levels

GENES & DEVELOPMENT 1663 Downloaded from genesdev.cshlp.org on October 5, 2021 - Published by Cold Spring Harbor Laboratory Press

Lyons et al.

Figure 5. Expression of BMP-2a and Vgr-1 RNA in developing bone and cartilage. Bright-field (left) and corresponding dark-ground [light] photomicrographs of embryo sections after in situ hybridization, autoradiography, and staining. [A] Section through developing prevertebrae (arrowheads) of a day-12.5 p.c. embryo hybridized with BMP-2a antisense probe. Bar, 150 M-m. (S) Same section as A, showing hybridization in the condensing mesenchyme, which will give rise to the ribs/vertebrae (arrowheads). (C) Longitudinal section through the limb of a day-15.5 p.c. embryo hybridized with BMP-2a antisense probe. Bar, 1 mm. [D] Same section as in C, showing strong signal in the hypertrophic cartilage of the long bones (double arrowheads) and digits (single arrowhead). Note the strong signal in the epidermis surrounding the limb. (E) A long bone from a 17.5-day p.c. embryo, showing four stages of maturation in the developing bone. Bar, 200 M-m. (P) Same section as in £, showing that only the hypertrophic cartilage (double arrowheads) contains Vgr-1 RNA. The resting chondrocytes, prohfcrating chondrocytes, and osteogenic zone show no specific signal. Sections were exposed for 7-10 days, (re) Resting chondrocytes; (pc) proliferating chondrocytes,- (he) hypertrophic cartilage; (oz) osteogenic zone (pern) peri­ chondrium.

of Vgr-1 RNA; however, the lip epithelium, which has a and N. Jenkins, in prep.). Embryos that are homozygous thicker layer of nucleated suprabasal cells and a thin for the recessive ch mutation die shortly after birth due, stratum comeum, expresses very low levels of Vgr-1 at least in part, to a failure of the subarachnoid space to RNA. Hybridization to the Vgr-1 riboprobe is confined develop (Gruneberg 1953; Green 1970). These mutants in the lip and palate to several highly differentiated also have a number of other defects, including delayed outer layers. This correlation suggests that one function growth of the long bones and cartilaginous skull. The of Vgr-1 may be to regulate the progression of the imme­ expression of Vgr-1 transcripts in meninges and devel­ diate suprabasal cells to a more differentiated and, there­ oping limbs is thus consistent with the theory that the fore, more comified phenotype. The observation that ch mutation may be a lesion in the Vgr-1 gene. We are TGFpi reversibly inhibits the proliferation of human currently exploring this possibility. prokeratinocytes in vitro (Shipley et al. 1986) is consis­ One of the striking findings emerging from this work tent with this proposal and suggests that Vgr-1 may have is that all of the members of the TGFp superfamily that some functions related to those described for TGFpi. have been examined to date by in situ hybridization are Vgr-1 RNA is not confined to epithelia, as we observed expressed in developing skin and limbs in temporally high levels of expression in a variety of mesenchymal and spatially distinct patterns. We note that the expres­ and ectomesenchymal tissues, including hypertrophic sion of TGFP2 and Vgr-1 in skin coincides with the time cartilage, uterine stroma, and meninges. The potential of thickening and differentiation of the epidermis roles of Vgr-1 in the growth and differentiation of these (Pelton et al. 1989; for a review of skin development, see tissues are unknown at present but are considered Kopan and Fuchs 1989). Furthermore, Lehnert and Ak- below. However, the expression of Vgr-1 in the me­ hurst (1988) reported that TGF(31 transcripts are present ninges is particularly interesting because the Vgr-1 gene in developing whisker follicles, and Akhurst et al. (1988) has been mapped recently to a location near the congen­ found that TGFpi RNA is present in the suprabasal ital hydrocephalus (ch) locus on mouse chromosome 13 layers of adult skin treated with the tumor promoter 12- (M. Dickinson, M. Kobrin, C. Silan, D. Kingsley, M. Jus­ tetradecanoyl-phorbol-13-acetate (TPA). Thus, the ex­ tice, D. Miller, J. Ceci, L. Lock, A. Lee, A. Buchberg, L. pression of TGFpi and TGFp2 in the epidermis and epi­ Siracusa, K. Lyons, R. Derynck, B. Hogan, N. Copeland, thelial components of the hair follicles and Vgr-1 in the

1664 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on October 5, 2021 - Published by Cold Spring Harbor Laboratory Press

Expression of Vgi-1 and BMP-2a epidermis occurs at a time of active proliferation and progression through this lineage. Similarly, the expres­ differentiation, whereas the levels of transcripts for sion of the Vgr-1 gene in the hypertrophic cartilage may these grow^th factors are reduced in adult skin (Lehnert promote the terminal differentiation of these cells in an and Akhurst 1988; Pelton et al. 1989, and unpubl.). Simi­ autocrine fashion, as well as affect the proliferation of larly, BMP-2a RNA expression persists in adults in the the chondrocytes from which the hypertrophic cells actively proliferating hair cortex cells. arise by a paracrine mechanism. An additional possi­ A similar example of overlapping but temporally and bility is that the Vgr-1 protein is chemotactic for the os­ spatially distinct patterns of expression of TGF3 super- teoblasts and osteoclasts that eventually replace the car­ family members occurs in the developing limbs. The de­ tilaginous matrix. In our model we indicate that each of velopment of long bones is initiated by the condensation the TGFp-like gene products has autocrine and paracrine of the mesenchyme, proliferation and differentiation of effects on proliferation and differentiation. The precise the chondroblasts into chondrocytes, deposition of a car­ roles and sites of action in cartilage differentiation of tilaginous matrix, hypertrophy and degeneration of the each of these gene products will require the use of spe­ chondrocytes, calcification of cartilage, and replacement cific antibodies and purified proteins in in vivo studies,- of this matrix by invading osteogenic cells, which pro­ however, there is some experimental support for this duce a bone matrix that later becomes calcified. BMP-2a model. For example, BMP-2a has been shown to induce is the first member of the TGFp superfamily known to ectopic bone formation in vivo in a process involving the be expressed in this sequence and is localized in con­ migration, proliferation, and condensation of mesen­ densing precartilaginous mesenchyme but is undetect­ chymal cells (Wozney et al. 1988). In vitro studies show able in the resting, proliferating, and differentiating that TGFpi and TGF32 can induce fetal rat muscle mes­ chondrocytes. TGF(32 is first detected at a later stage in enchymal cells to undergo differentiation and synthesize the chondroblasts of the perichondrium (Pelton et al. cartilage-specific macromolecules in vitro (Seyedin et al. 1989). Vgr-1 is expressed at the final stage of chondro- 1986, 1987). The most extensively studied protein, blast differentiation in the hypertrophic cells. TGFpi, has been shown to have autocrine and paracrine This pattern raises the possibility that the expression effects on the proliferation and differentiation of many of different members of the TGFp superfamily is re­ of the cell types within the chondrocyte lineage in in quired at distinct stages of differentiation within the vitro model systems (Seyedin et al. 1985; Centrella et al. chondrocyte lineage to ensure the regulated progression 1988; Kato et al. 1988; D.M. Rosen et al. 1988). These of the condensing mesenchymal cell through to the ter­ studies thus demonstrate that TGF(3-like gene products minally differentiated hypertrophic cartilage phenotype. can affect growth and differentiation at many stages The in situ hybridization evidence for this proposal is within the lineage, at least in vitro. Different approaches summarized in the model presented in Figure 6. We sug­ will be necessary to elucidate the specific roles each gest that the initial expression of BMP-2a and perhaps TGF32-like gene product has in the progression of the other as yet unidentified members of the TGFp superfa­ chondrocyte lineage in vivo. mily may promote the condensation and differentiation In summary, our work provides strong evidence that of mesenchyme into actively proliferating chondroblasts the Vgr-1 gene plays multiple roles in murine develop­ and chondrocytes, which results in the expression of a ment and in the growth regulation of adult tissues. Fur­ different set of TGF|3-like gene products, including thermore, we have shown that Vgr-1 and BMP-2a, two TGFp2, in the more terminally differentiated cells. The members of the dpp subgroup of the TGFp superfamily, newly induced growth factors then may regulate the pro­ are expressed in temporally and spatially distinct pat­ liferation and differentiation of the cells in which they terns in developing skin and limbs, as are TGFpi and are produced in an autocrine manner, as well as act on TGF32. This observation suggests that the coordinated additional cell types in the chondrocyte differentiation expression of many TGF3-like growth factors is required pathway in a paracrine manner to ensure a coordinated during the development of specific organs.

condensing and/or cell type: precartilaginous , chondroblast chondrocyte' hjrpertrophic cartilage mesenchyme

mRNA: BMP-2a TGFJ32 [TGFU-like]' Vgr-1 Figure 6. [Top line) The progression of cells from condensing mesenchyme through to hypertrophic cartilage; {bottom line) TGFp- like transcripts that have been reported to be expressed in each cell type. Arrows depict potential autocrine and paracrine effects of the various TGFp-like gene products, (a] Data from Pelton et al. (1989). [b] Sandberg et al. (1988) reported small amounts of TGFpi RNA in some chondrocytes of the growth plates and epiphyses. In contrast, Lehnert and Akhurst (1988) did not detect TGFp RNA in the cartilage under conditions of high stringency hybridization, but they reported hybridization within the cartilage with lower strin­ gency. This observation may thus reflect low levels of TGFpi RNA or the presence of a closely related member of the TGFp superfa­ mily. Evidence for the latter possibility is provided by Jakowlew et al. (1988), who showed that TGFp4, which is 85% homologous to TGFpi at the nucleotide level, is abundant in cultured primary chick embryo chondrocytes.

GENES & DEVELOPMENT 1665 Downloaded from genesdev.cshlp.org on October 5, 2021 - Published by Cold Spring Harbor Laboratory Press

Lyons et al.

Methods this, the corresponding regions of the human BMP-2a and BMP-2b cDNAs are 58% identical. A secreted phosphoprotein 1 Isolation of total RNA from primary oocytes and follicle cells or 2ar (SPP-l)(Nomura et al. 1988) antisense riboprobe was used For isolation of primary oocytes and follicle cells, ovaries from as a positive control in bone, uterus, and cumulus cells and as a 12.5-gram (C57BL/6 x DBA/2) Fi females were dissected into negative control in other tissues. Single-stranded sense and an­ Dulbecco's modified Eagle's medium, containing 10% fetal bo­ tisense RNA probes were labeled with a-^^s.^abeled UTP (1400 vine serum, and punctured repeatedly with a fine-gauge needle. Ci/mM, New England Nuclear) to an approximate specific ac­ Released primary oocytes with an intact germinal vesicle and tivity of 2 X 10' dpm/|xg RNA, using either SP6 or T7 poly­ no polar body were separated individually from follicle cells by merase, essentially as described by Hogan et al. (1986). These pipetting and collected manually. Clumps of follicle cells were probes were reduced to an average size of 100-150 bp by lim­ also collected for RNA extraction. Total RNA was isolated by ited alkaline hydrolysis (Cox et al. 1984) and used for hybridiza­ disrupting cells in guanidium isothiocyanate, followed by tion at a final concentration of 2 x 10* dprn/fil. phenol/chloroform extraction and precipitation with ethanol in the presence of 50 fig of carrier tRNA, as described (Huarte et al. 1987). Northern hybridization Total RNA from undifferentiated F9 cells and uteri of 9.5-day Preimplantation embryos p.c. pregnant mice was isolated as described (Lyons et al. 1989). Northern hybridizations on 20 |xg of total RNA or -150 pri­ Prepuberal Fj females (—12.5 g) were superovulated and mated mary oocytes were carried out as described in Lyons et al. with Fi males as described (Hogan et al. 1986). Oviducts con­ (1989). taining fertilized eggs and cumulus cells were fixed ~21 hr post-human chorionic gonadotropin (hCG). Cleavage-stage em­ bryos and blastocysts were recovered from the oviduct and In situ hybridization uterus as described (Hogan et al. 1986). For convenience of han­ In situ hybridizations and washes were carried out at high strin­ dling, they were transferred into oviducts containing fertilized gency, essentially as described by Nomura et al. (1988), with one-cell eggs, and the oviducts were dissected and fixed. The the following modifications. Sections were hybridized for fertilized eggs thus provided a positive control for the in situ 14-16 hr at 50°C in 50% formamide. Following hybridization, hybridizations. the sections were washed for 1 hr in 50% formamide at 50°C, treated with RNase A (Sigma) at a concentration of 10 M-g/ml for Postimplantation embryonic and adult tissues 20 min at 37°C, and washed at 55°C for 1 hr in 2 x SSC and 1 hr in 0.1 X SSC. Exposure times were between 1 and 3 weeks, with Postimplantation embryos were obtained from ICR outbred fe­ Ilford K5 emulsion. After development, the slides were stained males mated with Fi males (Harlan Sprague-Dawley). Noon of with 0.2% toluidine blue and analyzed on a Zeiss Axiophot mi­ the day of the vaginal plug is 0.5 days p.c. croscope using bright-field and dark-grovmd optics. The slides were photographed using Kodak Technical Pan film. Probe construction An 893-bp SacI-£coRI fragment from the 5' region of the Vgr-1 In situ analysis of grain density cDNA (Lyons et al. 1989) was subcloned into pGEM-3Zf(-) Autoradiographic silver grains in slides developed after 10-20 (Promega, Madison, Wisconsin). This fragment does not include days of exposure were counted on photographs of dark-ground the region of high similarity between Vgr-1 and other members images. Grain density was measured without subtraction of of the TGFp superfamily, thus minimizing the probability of background grain counts. Densities over five fertilized eggs cross hybridization. Our Vgr-1 probe shows 41% identity to the were calculated as percent of the grain density observed in two murine BMP-2a cDNA and only 37% identity to the murine primary oocytes from the same set of slides. TGFpl cDNA. The Vgr-1 construct detects a single, ~3.5-kb transcript when hybridized to RNA by Northern analysis from all tissues except testes, where an additional 1.8-kb transcript is Acknowledgments detected (Lyons et al. 1989). A 1124-bp £coRI fragment from the murine BMP-2a cDNA (M. Dickinson, M. Kobrin, C. Silan, We thank Sofie Hashmi for excellent technical assistance, Rik D. Kingsley, M. Justice, D. Miller, J. Ceci, L. Lock, A. Lee, A. Derynck for providing the BMP-2a cDNA clone, and Lyrm Ma- Buchberg, L. Siracusa, K. Lyons, R. Derynck, B. Hogan, N. trisian, Hal Moses, Jeff Holt, and Ed Yang for helpful comments Copeland, and N. Jenkins, in prep.) was subcloned into on the manuscript. K.M.L. was supported by National Cancer pGEM-3Zf(-) and linearized with XmaUl. This construct Institute postdoctoral training grant 5T32-CA0959292. R.W.P. yields a 1-kb riboprobe that detects a single 3.5-kb transcript was supported by National Institutes of Health grant T32- when hybridized to Northern blots (data not shown). We were GM07347. imable to determine the overall identity of this construct to the closely related murine BMP-2b cDNA because the complete se­ quence for the murine BMP-2b cDNA was not available. How­ References ever, 3'-terminal 374 bp of otir probe is only 64% identical with Akhurst, R.J., F. Fee, and A. Balmain. 1988. Localized produc­ the corresponding region of a partial murine BMP-2b cDNA tion of TGFp mRNA in tumour promoter-stimulated mouse clone. Because the remaining 620 bp of the murine BMP-2a con­ epidermis. Nature 331: 363-365. struct encodes a region of the protein that is less highly con­ Bachvarova, R., V. DeLeon, A. Johnson, G. Kaplan, and B.V. served at the amino acid level (Wozney et al. 1988), we expect Paynton. 1985. Changes in total RNA, polyadenylated RNA, that the overall homology of our BMP-2a probe to the BMP-2b and actin mRNA during meiotic maturation of mouse oo­ transcript will be <64% at the nucleotide level. In support of cytes. Dev. Biol. 108: 325-332.

1666 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on October 5, 2021 - Published by Cold Spring Harbor Laboratory Press

Expression of Vgi-1 and BMP-2a

Centrella, M., T.L. McCarthy, and E. Canalis. 1988. Skeletal Kato, Y., 1. Masahiro, T. Koike, F. Suzuki, and Y. Takano. 1988. tissue and transforming growth factor p. FASEB f. 2: 3006- Terminal differentiation and calcification in rabbit chondro­ 3073. cyte cultures grown in centrifuge tubes: Regulation by Coffey, R.J., C.C. Bascom, N.J. Sipes, R. Graves-Deal, B.E. transforming growth factor p and semm factors. Proc. Natl. Weissman, and H.L. Moses. 1988. Selective inhibition of Acad. Sci. 85: 9552-9556. growth-related gene expression in murine keratinocytes by Kimmelman, D. and M. Kirschner. 1987. Synergistic induction transforming growth factor 3. Mol. Cell. Biol. 8: 3088-3093. of mesoderm by FGF and TGF-p and the identification of an Cox, K.H., D.V. DeLeon, L.M. Angerer, and R.C. Angerer. 1984. mRNA coding form FGF in the early Xenopus embryo. Cell Detection of mRNAs in sea urchin embryos by in situ hy­ 51: 869-877. bridization using asymmetric RNA probes. Dev. Biol. Koplan, R. and E. Fuchs. 1989. A new look into an old problem: 101: 485-502. Keratins as tools to investigate determination, morphogen­ Dale, L. and M.W. Slack. 1987. Regional specification within esis, and differentiation in skin. Genes Dev. 3: 1-15. the mesoderm of early embryos of Xenopus laevis. Develop­ Lehnert, S.A. and R.J. Akhurst. 1988. Embryonic expression ment 100: 279-295. pattem of TGF beta type-1 RNA suggest both paracrine and Daniel, C.W., G.B. Silberstein, K. Van Horn, P. Strickland, and autocrine mechanisms of action. Development 104: 263- S. Robinson. 1989. TGF-beta-1-induced inhibition of mouse 273. mammary ductal growth: developmental specificity and Lund, L.R., A. Riccio, P.A. Andreasen, L.S. Nielsen, P. Kris- characterization. Dev. Biol. 135: 20-30. tensen, M. Laiho, O. Saksela, F. Blasi, and K. Dan0. 1987. DeLeon, V., A. Johnson, and R. Bachvarova. 1983. Half-lives Transforming growth factor-p is a strong and fast acting pos­ and relative amoimts of stored and polysomal ribosomes and itive regulator of the level of type-1 plasminogen activator poly(A)+ RNA in mouse oocytes. Dev. Biol. 98: 400-408. inhibitor mRNA in WI-38 human lung fibroblasts. EMBO J. Donahoe, P.K., R.L. Gate, D.T. MacLaughlin, J. Epstein, A.F. 6: 1281-1286. Fuller, M. Takahashi, J.P. Coughlin, E.G. Ninfa, and L.A. Lyons, K.M. and B.L.M. Hogan. 1990. TGFp-like genes in mam­ Taylor. 1987. Miillerian inhibiting substance: Gene struc­ malian development. In Proceedings of the 48th Symposium ture and mechanism of action of a fetal regressor. Recent of the Society for Developmental Biology, (ed. W.F. Loomis) Prog. Horm. Res. 43: 431-467. Alan R. Liss, New York. (In press). Edwards, D.R., G. Murphy, J.J. Reynolds, S.E. Whitham, A.J.P. Lyons, K.M., J.L. Graycar, A. Lee, S. Hashmi, P.B. Lindquist, Docherty, P. Angel, and J.K. Heath. 1987. Transforming E.Y. Chen, B.L.M. Hogan, and R. Derynck. 1989. Vgr-1, a growth factor beta modulates the expression of coUagenase mammalian gene related to Xenopus Vg-I and a new and metalloproteinase inhibitor. EMBO ]. 6: 1899-1904. member of the transforming growth factor p gene superfa- Feng, P., K.J. Catt, and M. Knecht. 1988. Transforming growth mily. Proc. Natl. Acad. Sci. 86: 4554-4558. factor-p stimulates meiotic maturation of the rat oocyte. Moses. H.L., R.F. Tucker, E.B. Leof, R.J. Coffey, J. Halper, and Endocrinology 111: 181-186. G.D. Shipley. 1985. Type beta transforming growth factor is Flanders, K.C., N.L. Thompson, D.S. Cissel, E. Van Obberghen- a growth stimulator and a growth inhibitor. Cancer Cells Schilling, C.C. Baker, M.E. Kass, R.L. EUingsworth, A.B. 3:65-71. Roberts, and M.B. Spom. 1989. Transforming growth factor- Mutter, G.L., G.S. Grills, and D.J. Wolgemuth. 1988. Evidence pi: Histochemical localization with antibodies to different for the involvement of the proto-oncogene c-mos in mam­ epitopes. /. Cell Biol. 108: 653-660. malian meiotic maturation and possibly very early embryo- Green, M. 1970. The developmental effects of congenital hydro­ genesis. EMBO J. 7: 683-689. cephalus (ch) in the mouse. Dev. Biol. 23: 585-608. Nomura, S., A.J. Wills, D.R. Edwards, J.K. Heath, and B.L.M. Griineberg, H. 1953. Genetical studies on the skeleton of the Hogan. 1988. Developmental expression of 2ar (osteopontin) mouse: VII. Congenital hydrocephalus. /. Genet. 51: 327- and SPARC (osteonectin) RNA as revealed by in situ hybrid­ 358. ization. /. Cell. Biol. 106: 441-450. Heine, U.I., E.F. Mimoz, K.C. Flanders, L.R. EUingsworth, Pelton, R.W., S. Nomura, H.L. Moses, and B.L.M. Hogan. 1989. H.-Y.P. Lam, N.L. Thompson, A.B. Roberts, and M.B. Spom. Expression of transforming growth factor p2 RNA during 1987. Role of transforming growth factor-p in the develop­ murine embryogenesis. Development 106: 759-767. ment of the mouse embryo. J. Cell. Biol. 105: 2861-2876. Piko, L. and K.B. Clegg. 1982. Quantitative changes in total Hogan, B.L.M., F. Constantini, and E. Lacy. 1986. Manipulating RNA, total poly(A), and ribosomes in early mouse embryos. the mouse embryo. A laboratory manual. Cold Spring Dev. Biol. 89: 362-378. Harbor Laboratory, Cold Spring Harbor, New York. Rebagliati, M.R., D.L. Weeks, R.P. Harvey, and D.A. Melton. Huarte, J., D. Belin, A. Vassalli, S. Strickland, and J.-D. VassaUi. 1985. Identification and cloning of localized maternal RNAs 1987. Meiotic maturation of mouse oocytes triggers the from Xenopus eggs. Cell 42: 769-777. translation and polyadenylation of dormant tissue-type Roberts, A.B., M.A. Anzano, L.M. Wakefield, N.S. Roche, D.F. plasminogen activator mRNA. Genes Dev. 1: 1201-1211. Stem, and M.B. Spom. 1985. Type p transforming growth Ignotz, R.A. and J. Massague. 1986. Transforming growth factor: A bifunctional regulator of cellular growth. Proc. factor-p stimulates the expression of fibronectin and col­ Natl. Acad. Sci. 82: 119-123. lagen and their incorporation into the extracellular matrix. Roberts, A.B., M.B. Spom, R.K. Assoian, J.M. Smith, N.S. /. Biol. Chem. 261: 4337-4345. Roche. L.M. Wakefield, U.I. Heine, L.A. Liotta, V. Falanga, Irish, V.G. and W.M. Gelbart. 1987. The decapentaplegic gene is J.H. Kehrl, and A.S. Fauci. 1986. Transforming growth factor required for dorsal-ventral patterning of the Drosophila type p: Rapid induction of fibrosis and angiogenesis in vivo embryo. Genes Dev. 1: 868-879. and stimulation of collagen formation in vitro. Proc. Natl. Jakowlew, S.B., P.J. Dillard, M.B. Spom, and A.B. Roberts. 1988. Acad Sci. 83: 4167-4171. Complementary deoxyribonucleic acid cloning of a mes­ Rosa, F., A.B. Roberts, D. Danielpour, L.L. Dart, M.B. Spom, senger ribonucleic acid encoding transforming growth factor and LB. Dawid. 1988. Mesoderm induction in amphibians: P4 from chick embryo chondrocytes. Mol. Endocrinol. The role of TGFp2-like factors. Science 239: 783-785. 2: 1186-1195. Rosen, D.M., S.A. Stempien, A.Y. Thompson, and S.M.

GENES & DEVELOPMENT 1667 Downloaded from genesdev.cshlp.org on October 5, 2021 - Published by Cold Spring Harbor Laboratory Press

Lyons et al.

Seyedin. 1988. Transforming growth factor-beta modulates expression of osteoblast and chondroblast phenotypes in vitro, f. Cell Physiol. 134: 337-346. Rosen, V., E.A. Wang, J.M. Wozney, P. Cordes, D. McQuaid, and L. Kurtzberg. 1988. Purification and molecular cloning of a novel group of BMPs and localization of BMP mRNAs in developing bone. Third Int. Conf. on the Chemistry and Bi­ ology of Minerahzed Tissues, p. 76. Russell, W.E., R.J. Coffey, A.J. Ouellette, and H.L. Moses. 1988. Type p transforming growth factor reversibly inhibits the early proliferative response to partial hepatectomy in the rat. Pioc. Natl. Acad. Sci. 85: 5126-5130. St. Johnston, R.D. and W. M. Gelbart. 1987. Decapentaplegic transcripts are localized along the dorsal-ventral axis of the Diosophila embryo. EMBO f. 6: 2785-2791. Sandberg, M., T. Vuorio, H. Hirvonen, K. Alitalo, and E. Vuorio. 1988. Enhanced expression of TGF-p and c-fos mRNAs in the growth plates of developing human long bones. Develop­ ment 102: 461-470. Seyedin, S.A., T.C. Thomas, A.Y. Thompson, D.M. Rosen, and K. A. Piez. 1985. Purification and characterization of two cartilage-inducing factors from bovine demineralized tissue. PIOC. Natl. Acad. Sci. 82: 2267-2271. Seyedin, S.J., P.R. Segarini, D.M. Rosen, A.Y. Thompson, H. Bentz, and J. Graycar. 1987. Cartilage-inducing factor-B is a unique protein structurally and functionally related to transforming growth factor-p. /. Biol. Chem. 262: 1946- 1949. Seyedin, S.M., A.Y. Thompson, H. Bentz, D.M. Rozen, J.M. McPherson, A. Conit, N.R. Siegel, G.R. Gallupi, and K.A. Piez. 1986. Cartilage-inducing factor-A. /. Biol. Chem. 261: 5693-5695. Shipley, G.D., M.R. Pittelkow, J.J. Wille, R.E. Scott, and H.L. Moses. 1986. Reversible inhibition of normal human pro- keratinocyte proliferation by type p transforming growth factor-growth inhibitor in serum-free medium. Cancer Res. 46:2068-2071. Silberstein, G.B. and C.W. Daniel. 1987. Reversible inhibition of mammary gland growth by transforming growth factor p. Science 237: 291-293. Smith, J.D. 1989. Mesoderm induction and mesoderm-inducing factors in early amphibian development. Development 105: 665-677. Thompson, N.L., K.C. Flanders, J.M. Smith, L.R. Ellingsworth, A.B. Roberts, and M.B. Spom. 1989. Expression of trans­ forming growth factor-pi in specific cells and tissues of adult and neonatal mice. /. Cell Biol. 108: 661-669. Wozney, J.M., V. Rosen, A.J. Celeste, L.M. Mitsock, M.J. Whitters, R.W. Kriz, R.M. Hewick, and E.A. Wang. 1988. Novel regulators of bone formation: Molecular clones and activities. Science 242: 1528-1534.

1668 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on October 5, 2021 - Published by Cold Spring Harbor Laboratory Press

Patterns of expression of murine Vgr-1 and BMP-2a RNA suggest that transforming growth factor-beta-like genes coordinately regulate aspects of embryonic development.

K M Lyons, R W Pelton and B L Hogan

Genes Dev. 1989, 3: Access the most recent version at doi:10.1101/gad.3.11.1657

References This article cites 45 articles, 25 of which can be accessed free at: http://genesdev.cshlp.org/content/3/11/1657.full.html#ref-list-1

License

Email Alerting Receive free email alerts when new articles cite this article - sign up in the box at the top Service right corner of the article or click here.

Copyright © Cold Spring Harbor Laboratory Press