PDGFB Regulates the Development of the Labyrinthine Layer of the Mouse Fetal Placenta

Developmental Biology 212, 124–136 (1999) Article ID dbio.1999.9306, available online at http://www.idealibrary.com on PDGFB Regulates the Development of the Labyrinthine Layer of the Mouse Fetal Placenta

Rolf Ohlsson,*,1 Pierre Falck,* Mats Hellstro¨m,† Per Lindahl,† Hans Bostro¨m,† Gary Franklin,* Lars A¨ hrlund-Richter,‡ Jeffrey Pollard,§ Philippe Soriano,¶ and Christer Betsholtz† *Department of Animal Development & Genetics, Uppsala University, Norbyva¨gen 18A, S-752 36 Uppsala, Sweden; †Department of Medical Biochemistry, Gothenburg University, Medicinaregatan 9A, S-413 90 Gothenburg, Sweden; ‡Department of Medical Nutrition, Karolinska Institute, Huddinge, Sweden; §Department of Developmental and Molecular Biology, Albert Einstein College of Medicine, Bronx, New York 10461; and ¶Division of Basic Sciences, Fred Hutchinson Cancer Research Center, Seattle, Washington 98109

PDGFB is a growth factor which is vital for the completion of normal prenatal development. In this study, we report the phenotypic analysis of placentas from mouse conceptuses that lack a functional PDGFB or PDGFR␤ gene. Placentas of both types of mutant exhibit changes in the labyrinthine layer, including dilated embryonic blood vessels and reduced numbers of both pericytes and trophoblasts. These changes are seen from embryonic day (E) 13.5, which coincides with the upregulation of PDGFB mRNA levels in normal placentas. By E17, modiﬁcations in shape, size, and number of the fetal blood vessels in the mutant placentas cause an abnormal ratio of the surface areas between the fetal and the maternal blood vessels in the labyrinthine layer. Our data suggest that PDGFB acts locally to contribute to the development of the labyrinthine layer of the fetal placenta and the formation of a proper nutrient–waste exchange system during fetal development. We point out that the roles of PDGFB/R␤ signaling in the placenta may be analogous to those in the developing kidney, by controlling pericytes in the labyrinthine layer and mesangial cells in the kidney. Key Words: trophoblasts; labyrinthine; morphology; blood vessels; pericytes.

INTRODUCTION types elicit potent mitogenic signals, although only PDGFR␤ mediates chemotaxis and actin reorganization in certain cell Platelet-derived growth factor (PDGF) is a potent mitogenic types, such as human fibroblasts and porcine aortic endothe- agent for connective tissue cells such as fibroblasts and lial cells (Hammacher et al., 1989). smooth muscle cells and for glial cells such as oligodendrocyte In adults, most of the functions of PDGF relate to progenitors (Heldin and Westermark, 1990; Raines et al., 1990; different responses to injury, such as inflammation, vascu- Raines and Ross, 1993; Calver et al., 1998). PDGF is synthe- lar injury, and wound healing (Heldin and Westermark, sized by a variety of cell types, including vascular endothelial 1990; Ross et al., 1986; Khachigian et al., 1996). Increased cells, monocytes/macrophages, placental cytotrophoblasts, expression of PDGF and PDGF receptors is seen in a and certain neurons. PDGF is a dimeric molecule composed of number of pathological conditions, including atherosclero- disulfide-bonded A and/or B polypeptide chains, PDGFA and sis, connective tissue overgrowth in conjunction with PDGFB, which are the products of two distinct genes (Heldin chronic inflammatory processes, fibrosis, and tumor stroma and Westermark, 1989). PDGF dimers signal via homo- or formation (Raines and Ross, 1993). The successive activa- heterodimers of two receptor types. The PDGF ␣-receptor tion of the PDGFB and PDGFR␤ genes during the genesis of (PDGFR␣) interacts with both the A and the B chain of PDGF, choriocarcinoma may reflect the establishment of a persis- whereas the PDGF ␤-receptor (PDGFR␤) preferentially binds tent autocrine loop (Holmgren et al., 1994). The most the B chain (Heldin and Westermark, 1990). Both receptor striking illustration of the importance of PDGFB and PDGFR␤ function can be found, however, during prenatal 1 To whom correspondence should be addressed. Fax: ϩ46-18- development. PDGFB-orPDGFR␤-deficient mice die peri- 4712683. E-mail: [email protected]. natally and exhibit specific developmental abnormalities

124 0012-1606/99 PDGF and Placental Development 125

(Leveén et al., 1994; Soriano, 1994). These include mal- Peroxidase activity was detected by conventional DAB staining formed kidney glomerular tufts and the formation of mi- (Vectastain). Staining of placenta pericytes was done using croaneurysms due to loss of mesangial cells and microvas- antibodies against ␣-smooth muscle actin (ASMA) or desmin as cular pericytes (Lindahl et al., 1997a, 1998). Most embryos described (Lindahl et al., 1998) develop fatal vascular leakage and rupture of microvessels Morphometric and statistical analysis. Cell numbers within the labyrinthine layer were counted from randomly chosen areas just prior to birth. The hematological status of such mice measuring 0.27 ϫ 0.21 mm. Fetal blood vessels could be distin- includes erythroblastosis, macrocytic anemia, and throm- guished from maternal blood lacunas by the presence of endothelial bocytopenia. The fact that similar phenotypes are observed cells and adjacent isolectin-positive matrix (see Results). The with both mutants suggests failure of PDGFB interaction surface areas (parametric length; 1000 units equivalent to 1 mm2)of with PDGFR␤, rather than with PDGFR␣. fetal blood vessels and maternal blood lacunas were scored from The communication between the fetus and the mother randomly chosen areas (0.27 ϫ 0.21 mm) in the labyrinthine layer occurs via the placenta in all eutherian mammals. In the and analyzed using the computer program NIH Image 1.58. At least mouse placenta, contact between the fetal and the maternal three different specimens from each of the wild-type, PDGFB- ␤ circulatory systems takes place in the labyrinthine layer, in deficient, and PDGFR -deficient placentas were examined in 9–16 which the two intermingled networks of vessels—fetal areas (from nine different glass slides). Although measurements were analyzed using average and standard deviation of error, they capillaries and maternal lacunas—form a large contact were also tested using the ranking nonparametric Mann–Whitney surface for nutrient exchange and excretion. The labyrin- U test. Comparisons of cell and vessel numbers and parametric thine layer is functionally analogous to the region of the length variations between wild-type versus PDGFB Ϫ/Ϫ and human placenta containing the chorionic villi, which carry PDGFR␤ Ϫ/Ϫ showed that the probabilities that the measure- the fetal capillaries, and the intervillus spaces, which con- ments are part of the same populations of values were P Ͻ 0.0001 tain the maternal blood. We show here that PDGFB- and for all the statistical calculations presented in Fig. 6. The volumes PDGFR␤-deficient mouse placentas have a specific defect of the indicated placentas were analyzed by measuring the para- in the labyrinthine layer, characterized by reduced numbers metric surfaces of the labyrinthine and the spongiotrophoblast ␮ of pericytes and labyrinth trophoblasts, dilated fetal blood layers in every third section (each 7 m thin) throughout each vessels, and a reduced contact surface for maternal blood. specimen. The volumes were then calculated using the NIH Image software. Based on the phenotypic similarities between PDGFB- and ␤ ␤ RNA probes. The PDGFR probe was derived from the PDGFR -null placentas and kidney glomeruli, we propose pSVRI clone (Yarden et al., 1986) and was a kind gift from Dr. J. that placental pericytes and kidney mesangial cells may Escobedo. A 5.4-kb insert was subcloned in pBluescript KS have similar functions in the morphogenesis of complex plasmid. RNA polymerases T7 and T3 generated sense and capillary tufts involved in excretion and/or nutrient ex- antisense probes, respectively. The mouse PDGFB probe was change. generated from a cDNA clone isolated from a ␭gt10 cDNA library of mouse fetal mRNA (Stratagene) using a human PDGFB cDNA probe. The 2.2-kb insert, starting 918 bp downstream of MATERIALS AND METHODS the cap site and ending within exon 7, was subcloned into a pBSKS plasmid. RNA polymerases T7 and T3 were used to generate sense and antisense probes, respectively. The PDGFR␣ Production of PDGFB-, PDGFA-, PDGFR␤-, and PDGFR␣- template was a 234-bp segment (Mercola et al., 1990) containing mutant conceptuses. The derivation of mutant mice has already 3Ј untranslated sequences and was a kind gift from Dr. M. been described (Leveén et al., 1994; Soriano, 1994; Bostro¨m et al., Mercola. RNA polymerases T7 and T3 were used to generate 1996; Soriano, 1997). 129SV/C57Bl/6 hybrid offspring or embryos antisense and sense probes, respectively. were genotyped by Southern blot analysis or PCR as described Hybridization analysis. Northern blot hybridization analysis (Leveén et al., 1994; Soriano, 1994; Lindahl et al., 1997a,b). of total cellular RNA, extracted from the dissected placentas and Histological analysis. Placentas were processed for routine juxtaposed decidua, was performed as described (Ohlsson et al., histology by fixation in Bouin’s solution, or in 4% paraformalde- 1989). In situ hybridization analysis was performed on 7-␮m hyde in PBS, followed by paraffin embedding and sectioning. Four- paraffin sections of paraformaldehyde- or Bouin’s-solution-fixed to seven-micrometer sections were stained with PAS–Schiffs re- placentas using 35S-labeled antisense riboprobes as described (Ohls- agents and Mayer’s hematoxylin–eosin. For electron microscopy, son et al., 1989). After being washed, slides were dipped into NTB2 placentas were fixed overnight at 4°C in 3% glutaraldehyde in 0.1 emulsion (Kodak) and exposed for 1–2 weeks at 4°C. After devel- M cacodylate buffer and postfixed at room temperature in 1% opment and fixation, sections were counterstained with Mayer’s osmium tetraoxide in 0.1 M cacodylate buffer. Following fixation, hematoxylin and mounted. Digital images of the sections were the placentas were stained with 1% uranyl acetate and finally collected using a ColourCoolView integrating video camera (Pho- embedded in epoxy resin. Sections were contrasted with lead tonic Science, UK) and a Leica DMRXE light microscope. Images citrate before analysis in a Philips transmission electron micro- were processed and mounted using Adobe PhotoShop software, scope. version 4.0. Placental sections were stained with isolectin to visualize the matrix surrounding the fetal blood vessels (Wood et al., 1979). Sections were incubated in PBS containing 0.6% H O for 10 2 2 RESULTS min; in PBS, pH 6.8, 1% Triton X-100, 0.1 mM CaCl2, 0.1 mM

MgCl2, 0.1 mM MnCl2 for 15 min at room temperature; and then with 5 ␮g/ml isolectin B4 from Bandeiraea simplicifolia conju- PDGF and PDGF receptor mRNA expression. PDGFB gated with peroxidase (Sigma) in a humid chamber for 2 h. is temporally controlled during mouse placenta develop-

pathways involving these genes. PDGFR␣ expression was seen only in the spongiotrophoblast layer (Figs. 2D and 2G) and did not overlap with the expression of PDGFB and PDGFR␤. PDGFB- and PDGFR␤-deficient placentas display changes in the maternal–fetal interface. The function of the labyrinthine layer is to provide a pivotal nutrient–waste exchange compartment for the developing fetus. The exten- sive coexpression of the PDGFB and PDGFR␤ genes in the labyrinthine layer might be suggestive of a short-range function for PDGFB at this location. Electron microscopy shows that placental maternal lacunas are lined exclusively with trophoblasts, in contrast to fetal blood vessels which are lined with endothelial cells (Fig. 3A). By light microscopy, maternal lacunas and the fetal blood vessels were easily discriminated by the size of the erythrocytes (small FIG. 1. Developmental profile of PDGFB expression. Northern and uniform in maternal lacunas; large, irregularly shaped blot analysis of total cellular RNA, prepared from different stages of and occasionally nucleated in fetal vessels). Isolectin B4 placental development. The ethidium bromide-stained agarose gel stained the matrix surrounding the fetal vessels and was is included to show that each lane contains comparable amounts (20 ␮g) of input RNA. The numbers at the top depict days therefore used for morphological studies (Fig. 3B). This postconception. analysis revealed that fetal vessels were dilated to a varying degree in PDGFB Ϫ/Ϫ or PDGFR␤ Ϫ/Ϫ placentas, compared with wild-type littermate tissue, in which the vessel lumens were narrower and uniform in diameter (Figs. 3C–3F; and see overviews of HE stainings in Fig. 4). This ment (Fig. 1) with levels increasing significantly after E12, phenotype was clearly seen by at E13.5, whereas at E12.5, in agreement with previous results (Bidwell et al., 1995). the mutant fetal vessels were marginally affected, if at all Given the minute amounts of placental tissue during early (Figs. 5A–5F). embryonic development, the PDGFB signal in microdis- sected placentas at embryonic day (E) 9 may reflect con- We compared the sizes and the numbers of the fetal and tamination with embryonic tissue. We cannot, however, maternal blood compartments within the labyrinthine ␤ formally rule out the existence of a biphasic pattern of layer of normal and PDGFB-orR -mutant placentas in thin PDGFB expression with a first, more modest period of sections of fixed placental tissue by stereological methods expression ending at E10. Reprobing of these blots revealed (which estimates 3D parameters by 2D analyses; DeHoff no detectable PDGFA expression (data not shown). To and Rhines, 1968). The surface of the endothelial and determine in which cell types the PDGFB gene is expressed, trophoblast tissues facing the fetal and maternal blood, we performed in situ hybridization analysis: the bulk of respectively, was determined by examining the parametric PDGFB mRNA is confined to the labyrinthine layer of length (i.e., the length of the tissue lining the fetal or placentas by E17 (Fig. 2B). Closer examination showed that maternal blood compartment) in thin sections. Similarly, the PDGFB transcripts were localized primarily to subpopu- the volume of the blood compartments was assessed by lations of the labyrinth trophoblasts and endothelial cells parametric area determinations, i.e., the surface of the and, to a lesser extent, in some giant trophoblasts (Fig. 2E blood vessel in thin sections (see Materials and Methods for and data not shown). PDGFR␤ expression was likewise further information). Figure 6A shows the size distribution restricted to the labyrinthine layer, but appeared more of individual blood vessels (parametric length and area are evenly distributed than that of PDGFB. These studies do represented by blue and red colors, respectively) within the not conclusively resolve the expression patterns at the level fetal and maternal compartments of the labyrinthine layer of single cells, but the results are compatible with PDGFB of an E17 mouse placenta. An optimal ratio of the fetal/ expression by capillary endothelium and PDGFR␤ expres- maternal compartments would be expected to approach 1:1, sion by trophoblasts and vascular mural cells, as has been to reflect the attainment of optimal nutrient and waste suggested from other studies (Goustin et al., 1985; exchange rates between the fetal and the maternal blood Holmgren et al., 1991, 1992; Lindahl et al., 1997a, 1998). circulatory systems. The size and the number ratios be- There was no detectable hybridization of the PDGFB probe tween fetal and maternal blood vessels in the labyrinth to PDGFB Ϫ/Ϫ placentas (Fig. 2H), even though the tar- trophoblast layer of wild-type E17 placentas are approxi- geted mutation involved deletion of only exon 4 (Leveén et mately 1:1 (Fig. 6B). In PDGFB-deficient- and to a lesser al., 1994). Moreover, the absence of any marked effects on extent in PDGFR␤-deficient placentas, however, the fetal the levels of PDGFR␤ mRNAs in PDGFB-deficient (Fig. 2I) vessel parametric length and area increased relative to the or on PDGFB levels in PDGFR␤-deficient (not shown) maternal lacunas (Figs. 6A and 6B). placentas argues against the existence of any autoregulatory Assessment of nonendothelial labyrinthine cells, which

FIG. 2. Spatial distribution of active PDGFB, PDGFR␣, and PDGFR␤ genes in fetal placenta. Dark-field views show that PDGFB (B, E) and PDGFR␤ (C, F) mRNAs can be found only in the labyrinthine layer (lt) of the E17 mouse placenta. In contrast, PDGFR␣ (D, G) is expressed only in the juxtaposed spongiotrophoblast cell layer (st). (A) A bright-field view of an adjacent section. (H) Absence of PDGFB expression in PDGFB Ϫ/Ϫ placenta and (I) expression of PDGFR␤ mRNA in a PDGFB-deficient placenta. Scale bar, 1 mm for A–D and 0.05 mm for E–J.

are dominated by the trophoblasts, showed that their num- It is noteworthy that the spongiotrophoblast layer, which ber was reduced by approximately 40% (Fig. 7A). Although expresses high levels of PDGFR␣ mRNA, does not appear to this would be expected to lead to a smaller placenta, there be affected in PDGFB-orPDGFR␤-deficient placentas. We is no persistent difference in the volumes of the spongio- also analyzed placentas from PDGFA Ϫ/Ϫ (Bostro¨m et al., trophoblast and labyrinthine layers (Fig. 7C), suggesting 1996; Lindahl et al., 1997b) and PDGFA/B double knockout that there is an increased volume of blood in the labyrin- (Hellstro¨m et al., unpublished results) embryos. PDGFA thine layer of PDGFB- or PDGFR␤-deficient fetal placentas. Ϫ/Ϫ placentas had no apparent phenotypic abnormality, This conclusion is supported by the morphometric analysis. and lack of PDGFA did not appreciably add to the effect of The increased parametric area of PDGFB-deficient fetal PDGFB deficiency (data not shown). vessels (Fig. 6A) and the smaller number of vessel cross- Loss and altered organization of pericytes in PDGFB- section profiles per area (Fig. 7B) show that the mutant fetal and PDGFR␤-deficient placentas. PDGFB deficiency blood vessels are grossly dilated and should therefore con- leads to blocked recruitment of pericytes to blood vessels tain a larger volume of blood. sprouting into the developing mouse brain (Lindahl et al.,

1997a). We compared, therefore, the abundance of pericytes in the placentas of wild-type and PDGFB Ϫ/Ϫ or PDGFR␤ Ϫ/Ϫ conceptuses. Antibodies against ASMA labeled cells located between the endothelium and the isolectin-positive matrix surrounding the embryonic vessels (Fig. 8A). These cells had, therefore, the marker expression and distribution expected for pericytes in the placenta. The ASMA-positive cells were found in the labyrinthine layer but were com- pletely absent from the spongiotrophoblast layer, further demonstrating an association with the fetal vessels, rather than the maternal lacunas. Quantification of the density of ASMA-positive profiles (a profile being a continuous area of staining irrespective of size, representing part of a single cell or a cluster of cells) in the labyrinthine layer showed that it was reduced by approximately 50% in the PDGFB Ϫ/Ϫ embryo at E13.5 and E16.5 (Figs. 8B–8E). A similar reduction was seen in PDGFR␤ Ϫ/Ϫ placentas (data not shown). The actual loss of ASMA-positive cells in mutant placentas may be even larger, since in the wild-type, ASMA-positive cells were frequently found in clusters, whereas this was less apparent in the mutants (Figs. 8 and 9). By ASMA and isolectin double labeling, the spatial rela- tionships between the placenta pericytes and the fetal vessels could be studied in further detail (Fig. 9). In wild- type placentas, the pericytes were arranged in core-like structures, around which the fetal vessels were organized (Figs. 9A, 9C, and 9E). In the mutant placentas, the pericytes were still located to structures, which appeared to join individual dilated loops to each other (Fig. 9B). These attachments were regularly thin and appeared to be in the process of breaking up (arrows in Figs. 9D and 9F), however, leading to the displacement of pericytes from multicellular core aggregates to scattered sites in the walls of the dilated vessels.

DISCUSSION

It has been suggested that PDGFB controls vasculariza- tion and the proliferative potential of extravillus trophoblasts in the human placenta (Goustin et al., 1985; Holmgren et al., 1991, 1992, 1993). The role of PDGFB and PDGFR␤ in the formation of blood vessels has been supported by substantial genetic evidence in the mouse FIG. 3. Morphology of fetal placentas devoid of functional PDGFB (Leve´en et al., 1994; Soriano, 1994; Lindahl et al., 1997a, and PDGFR␤ genes. (A) Electron microscopic image of the labyrin- 1998; Crosby et al., 1998). Here, we provide genetic evi- thine layer depicting the absence of endothelial cells (en) in the dence that PDGFB and PDGFR␤ directly or indirectly maternal lacunas (ml) of an E17.5 placenta. Fetal vessel (fv); regulate the development of pericytes and trophoblasts in labyrinth trophoblast (lt). (B) Matrix between the endothelium of the labyrinthine layer of the mouse placenta. High-level the fetal vessels and the labyrinth trophoblasts of the maternal expression of the mouse PDGFB gene is associated with lacunas was stained with isolectin B4. (C and D) Isolectin staining relatively late placental development, coinciding with the of a PDGFB Ϫ/Ϫ (D) and a wild-type littermate (C) E17.5 placenta. expansion of the labyrinthine layer. The defects arising in (E and F) Isolectin staining of a PDGFR␤ Ϫ/Ϫ (F) and a wild-type littermate (E) placenta. Note that in the wild-type placentas, the the mutants are clearly seen at E13.5 (Figs. 5 and 8), which fetal and the maternal blood vessels are equal in size and number, correlates in time with the onset of strong PDGFB expres- whereas in the mutant placentas, the fetal blood vessels are larger sion in the normal placenta. Moreover, the placenta defects and less abundant than the maternal vessels. Scale bars, (A) 0.01 are restricted to the labyrinthine layer, which is the only mm, (B) 0.05 mm, and (C–F) 0.1 mm. part of the placenta showing PDGFB and PDGFR␤ expres-

FIG. 4. Overview morphology of placentas stained with PAS–Schiffs reagents and Mayer’s hematoxylin–eosin. For optimal resolution, 24–32 images (each representing a 10-fold magniﬁcation) per placenta specimen were integrated using the Adobe PhotoShop program. (A) Wild-type placenta. (B) PDGFB Ϫ/Ϫ placenta. (C) PDGFR␤ Ϫ/Ϫ placenta. Age E18.5. Labyrinthine layer (lt) and spongiotrophoblast layer (st) are indicated. Scale bars, 1 mm and in the inserts 0.1 mm.

FIG. 5. PDGFB regulates the development of the labyrinthine layer. Isolectin-stained mouse placenta at different developmental stages. (A) E14.5 placenta (PDGFB ϩ/ϩ). (B) E13.5 placenta (PDGFB ϩ/ϩ). (C) E12.5 placenta (PDGFB ϩ/Ϫ). (D) E14.5 placenta (PDGFB Ϫ/Ϫ). (E) E13.5 placenta (PDGFB Ϫ/Ϫ). (F) E12.5 placenta (PDGFB Ϫ/Ϫ). Placentas in (A)/(D), (B)/(E), and (C)/(F) are derived from sibling conceptuses. Although the control placenta of E12.5 is hemizygous for the PDGFB null allele (no wild-type littermate was available), the presence of only one functional PDGFB allele has so far not generated any recognizable phenotype in placental or fetal development (Leve´en et al., 1994; Soriano, 1994; and data not shown). Labyrinthine layer (lt) and spongiotrophoblast layer (st) are indicated. The magniﬁcation is the same for all. The scale bar corresponds to 0.2 mm.

sion. The normal pattern of PDGFB mRNA expression in these cells may be controlled by PDGFB in vivo (Lindahl et the mouse placenta correlates both temporally and spa- al., 1998; M. Hellstro¨m et al., submitted for publication). tially, therefore, with the defects in this organ that develop The loss of brain pericytes and kidney mesangial cells in PDGFB Ϫ/Ϫ and PDGFR␤ Ϫ/Ϫ conceptuses. This correlates with the formation of increased diameter capil- strongly suggests that the placenta phenotype results from lary loops and focal microaneurysms. In both cases, the deficient local PDGFB/R␤ signaling in the developing pla- pericyte/mesangial cell loss has been suggested to be the centa rather than from systemic effects involving the em- trigger for capillary dilation. Pericytes and mesangial cells bryo. have been shown to express PDGFR␤, constituting the The phenotypic alterations in PDGFB- and PDGFR␤- likely primary target for PDGFB released by capillary endo- deficient placentas comprise dilated embryonic blood ves- thelial cells (Lindahl et al., 1997a, 1998). Any effects arising sels and reduced numbers of pericytes and labyrinth tropho- in the endothelial cells because of PDGFB disruption are blasts. It has been demonstrated in vitro that trophoblasts likely to be secondary to the pericyte/mesangial cell loss. It and pericytes may express PDGFR␤ (Goustin et al., 1985; is tempting to speculate that a similar scenario prevails in Holmgren et al., 1992; D’Amore and Smith, 1993; Bernstein the placenta, but the unequivocal establishment of the et al., 1982) and these cells may, therefore, be directly PDGFB/R␤ expression pattern is more difficult in this targeted by mitogenic or other actions of PDGFB in the organ than at many other sites, due to the complex histol- developing placenta. Previous studies of PDGFB- and ogy of the labyrinthine layer. It also remains to be demon- PDGFR␤-null mice have demonstrated pericyte loss in the strated exactly which cellular mechanism(s) is controlled developing brain (Lindahl et al., 1997a) and mesangial cell by PDGFB signaling in the developing placenta. loss in kidney glomeruli (Leveénet al., 1994; Soriano, 1994). Despite these uncertainties, we would like to propose There are data to suggest that both mitosis and migration of that the reduction in pericyte density, and perhaps altered

FIG. 6. Morphometric analyses of mouse fetal placentas. Surface areas (parametric length) surrounding the fetal blood vessels and maternal blood lacunas were measured from different randomly chosen areas in the labyrinth trophoblast layer from wild-type, PDGFB Ϫ/Ϫ, and PDGFR␤ Ϫ/Ϫ placentas as outlined under Materials and Methods. Columns without error bars show a standard deviation too small to be visualized. (A) The frequency distribution of size and numbers of maternal and fetal blood vessels in the labyrinthine layer. The parametric length (surface of blood compartment; 1000 units equivalent to 1 mm) and parametric surface (volume of blood compartment; 1000 units equivalent to 1 mm2) are represented by blue and red colors, respectively. (B) The ratio of the average surface (parametric length) of the endothelial and trophoblast tissues facing the fetal and maternal blood, respectively.

pericyte function, underlies the labyrinth blood vessel dila- cally illustrated in Fig. 10). This appears analogous to how tion. The position of placenta pericytes between individual mesangial cells connect the capillary loops in the kidney labyrinth capillary loops suggests that the cells are involved glomerulus (Fig. 10). In the kidney, mesangial cells appar- in attachment of the loops to each other (Fig. 9A; schemati- ently have an important morphogenetic role in promoting

FIG. 7. Cell numbers of labyrinthine layer cell components in mouse fetal placenta. Numbers of trophoblastic cells (A) and blood vessels (B) per counted area (0.27 ϫ 0.21 mm) were estimated as outlined under Materials and Methods. (C) Absolute volumes (mm3)ofthe labyrinthine trophoblast layer and the spongiotrophoblast layer. Placenta age was E17.5. Error bars show standard deviations of the data. Columns without error bars show a standard deviation too small to be visualized.

FIG. 8. Pericyte reduction in PDGFB Ϫ/Ϫ placentas. Placental pericytes were visualized by ASMA staining (black). Pericytes were found in the labyrinthine layer but not in the spongiotrophoblast layer. Micrographs show representative fields from the labyrinthine layer of E12.5, E13.5, or E16.5 placentas (A–F). Note the reduction in pericyte density in the PDGFB Ϫ/Ϫ placentas ages E13.5 and E16.5 in comparison with the wild-type controls. (G) A comparison of pericyte density between placentas of different ages and genotypes by counting ASMA-stained profiles in equal-sized microscopic fields. Error bars indicate the standard deviation. All micrographs are of the same magnification. The scale bar in (F) represents 0.1 mm. Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. PDGF and Placental Development 133

FIG. 9. Spatial location of pericytes in wild-type and PDGFB Ϫ/Ϫ placentas. ASMA (black) and B4 isolectin (blue) double staining of E16.5 placentas. In the wild-type placenta, the ASMA-positive pericytes locate centrally in small clusters of fetal capillary loops and appear to hold these loops together (white asterisks in (C) and (E), which show two high-magniﬁcation views of the section in (A)). A similar spatial location of ASMA-positive cells is seen in PDGFB Ϫ/Ϫ placentas, although many ASMA-positive cells appear have detached from each other and located to the wall of dilated fetal vessels (arrows in (D) and (F)). Scale bars, 0.1 mm.

FIG. 10. Schematic illustration of analogies between blood vessel organization in the placenta and the kidney and effects of PDGFB or PDGFR␤ deﬁciency. Capillary loops in the labyrinthine layer are joined by placental pericytes, which form a core around which the capillaries organize (A). In mutant placentas (B), the pericytes are reduced in number and detached from the core structures, which correlates with (and may be the cause of) vessel dilation. Placental pericytes appear, therefore, to have a function that is highly analogous to that of mesangial cells in kidney glomeruli, which constitute the core around which glomerular capillaries are organized (C). Loss of mesangial cells, as appears in PDGFB-orPDGFR␤-mutant glomeruli, results in glomerular capillary ballooning (D). Placental pericytes and mesangial cells have similar marker expression and mesangial cells have been suggested to be specialized pericytes. By organizing capillary tufts, the placental pericytes and mesangial cells seem to have critical roles in establishing complex large-surfaced structures involved in nutrient exchange, excretion, and ﬁltration.

the formation of a complex tuft of capillaries. In PDGFB-or layer (Fig. 7B). Such trophoblastic lacunas are filled with PDGFR␤-deficient kidneys, which lack mesangial cells, maternal blood, which carries nutrients to, and waste either a single or a few dilated capillary loops develop products from, the fetus. The 1:1 ratio in the surface areas (Leveén et al., 1994; Soriano, 1994; Lindahl et al., 1998). between the trophoblast lacunas and the fetal blood vessels Mesangial cells also have an important mechanical func- would seem to be rational for an efficient nutrient–waste tion in preserving the architecture of the mature tuft, since exchange. According to this reasoning, a reduction in the injury to mesangial cells elicited by Thy-1 antibodies, or surface area of trophoblastic lacunas would lead to a sub- venom toxins, results in the formation of dilated capillary optimally functioning placenta, forcing the conceptus to loops (Bagchus et al., 1986; Uiker and Kriz, 1995). Pericytes adopt either of two avenues. One of these would be to adapt of the placenta labyrinthine layer may have analogous to the reduced supply of nutrients by reducing the growth of morphogenetic and structural functions in the development the embryo/fetus. Consistent with this, PDGFB Ϫ/Ϫ em- or maintenance of the complex network of fetal blood bryos are approximately 20% smaller than PDGFB ϩ/ϩ vessels. Loss of such pericytes may underlie the labyrinth embryos at E14.5 until term (M. Hellstro¨m et al., unpub- fetal vessel dilation (Fig. 10). lished). An alternative avenue is that the relative blood The reduced number of labyrinth trophoblasts in the volume of the placenta increases to compensate for the PDGFB-deficient placenta is likely to cause the reduction in reduced availability of maternal blood. This is also sup- the surface area of the maternal lacunas in the labyrinthine ported by the morphometric analysis showing dilated em-

bryonic blood vessels containing a larger volume of embry- Bidwell, M. C., Eitzman, B. A., Walmer, D. K., McLachlan, J. A., onic blood. In this context, it is interesting to note that and Gray, K. D. (1995). Analysis of messenger ribonucleic acid PDGFB- and PDGFR␤-null embryos have macrocytic ane- and protein for the ligands and receptors of the platelet-derived mia and thrombocytopenia and an increased proportion of growth factor signaling pathway in the placenta, extraembryonic circulating nucleated erythroblasts at late gestation (Leveén membranes, and uterus during the latter half of murine gesta- et al., 1994; Soriano, 1997). This may in part be accounted tion. Endocrinology 136, 5189–5201. Bostro¨m, H., Willetts, K., Pekny, M., Leveén, P., Lindahl, P., for by hemorrhage due to ruptured microvessels in the Hedstrand, H., Pekna, M., Hellstro¨m, M., Gebre-Medhin, S., embryo (Lindahl et al., 1997a), but the present study adds Schalling, M., Nilsson, M., Kurland, S., To¨rnell, J., Heath, J. K., the possibility that embryonic anemia results from the and Betsholtz, C. (1996). PDGF-A signaling is a critical event in accumulation of a larger volume of fetal blood in the lung alveolar myofibroblast development and alveogenesis. Cell placenta. The erythroblastosis may be secondary to the 85, 863–873. anemia, reflecting a mechanism of compensation involving Crosby, J. R., Seifert, R. A., Soriano, P., and Bowen-Pope, D. F. increased release of immature blood cells from the hema- (1998). Chimaeric analysis reveals role of Pdgf receptors in all topoietic organs (e.g., liver and spleen). muscle lineages. Nat. Genet. 18, 385–388. In summary, we have shown here that absence of D’Amore, P. A., and Smith, S. R. (1993). Growth factor effects on PDGFB/R␤ signaling perturbs the structure of the nutrient– cells of the vascular wall: A survey. Growth Factors 8, 61–75. waste exchange system of the fetal placenta. Specifically, DeHoff, R. T., and Rhines, F. N. (1968). “Quantitative Micros- PDGFB- and PDGFR␤-deficient placentas display abnor- copy.” McGraw–Hill, New York. malities confined to the labyrinthine layer, involving peri- Dickson, M. C., Martin, J. S., Cousins, F. M., Kulkarni, A. B., cyte and trophoblast loss and fetal blood vessel dilation. Karlsson, S., and Akhurst, R. J. (1995). Defective haematopoiesis and vasculogenesis in transforming growth factor-␤1 knockout The labyrinthine layer normally expresses both PDGFB and mice. Development 121, 1845–1854. PDGFR␤ during late development. PDGFB signaling is Goustin, A. S., Betzholtz, C., Pfeifer-Ohlsson, S., Persson, H., therefore likely to act locally to contribute to the formation Rydnert, J., Bywater, M., Holmgren, G., Heldin, C. H., Wester- of a properly structured labyrinthine layer. Analogous phe- mark, B., and Ohlsson, R. (1985). Coexpression of the sis and myc ␤ notypes of PDGFB- or PDGFR -null placentas and kidney proto-oncogenes in developing human placenta suggests auto- glomeruli may suggest that the PDGFB/R␤ signaling sys- crine control of trophoblast growth. Cell 41, 301–312. tem has a generic function in the formation of cells (spe- Hammacher, A., Mellstro¨m, K., Heldin, C.-H., and Westermark, B. cialized pericytes) with critical morphogenetic roles in the (1989). Isoform-specific induction of actin reorganization by generation of vascular structures involved in both excretion platelet-derived growth factor suggests that the functionally and nutrient exchange. active receptor is a dimer. EMBO J. 8, 2489–2495. Heldin, C.-H., and Westermark, B. (1989). Platelet-derived growth factor: Three isoforms and two receptor types. Trends Genet. 5, ACKNOWLEDGMENTS 108–111. Heldin, C.-H., and Westermark, B. (1990). Platelet-derived growth We are grateful to Virpi To¨ho¨ nen, Helena Malmikumpu, Gun- factor: Mechanism of action and possible in vivo function. Cell Britt O¨ rlander, and Sara Beckman for excellent technical assis- Regul. 1, 555–566. tance. Valuable advice has been provided by Drs. Lennart Olsson, Holmgren, L., Glaser, A., Pfeifer-Ohlsson, S., and Ohlsson, R. Reinald Fundele, and Stefan Gunnarsson. This work was supported (1991). Angiogenesis of human extraembryonic tissue involves by the Swedish Cancer Research Foundation (RmC/CF), the Natu- the spatiotemporal control of PDGF ligand and receptor genes. ral Science Research Council Sweden (NFR), the Medical Research Development 133, 749–754. Council Sweden (MFR), the von Hofsten Foundations, the Inga Holmgren, L., Claesson-Welsh, L., Heldin, C.-H., and Ohlsson, R. Britt and Arne Lundberg Foundation, the Go¨ran Gustafsons Foun- (1992). The expression of PDGF ␣- and ␤-receptors in subpopu- dation, the Novo Nordisk Foundation, and the National Institutes lations of PDGF-producing cells implicates autocrine stimula- of Health (U.S.A.). tory loops in the control of proliferation of active cytotrophoblasts that have invaded the uterine lining. Growth Factors 6, REFERENCES 219–231. Holmgren, L., Flam, F., Larsson, E., and Ohlsson, R. (1993). Succes- Bagchus, W. M., Hoedemaeker, P. J., Rozing, J., and Bakker, W. W. sive activation of the platelet-derived growth factor beta receptor (1986). Glomerulonephritis induced by monoclonal anti-Thy 1.1 and platelet-derived growth factor B genes correlates with the antibodies. A sequential histological and ultrastructural study in genesis of human choriocarcinoma. Cancer Res. 53, 2927–2931. the rat. Lab. Invest. 55, 680–687. Jollie, W. P. (1964). Fine structure changes in the placental laby- Bernstein, L. R., Antoniades, H., and Zetter, B. R. (1982). Migration rinth of the rat with increasing gestational age. J. Ultrastruct. of cultured vascular cells in response to platelet-derived growth Res. 10, 27–47. factor. J. Cell Sci. 56,71–82. Khachigian, L., Lindner, V., Williams, A. J., and Collins, T. (1996). Betsholtz, C., Johnsson, A., Heldin, C.-H., Westermark, B., Lind, P., Egr-1-induced endothelial gene expression: A common theme in Urdea, M. S., Eddy, R., Shows, T. B., Philpott, K., Mellor, A. L., vascular injury. Science 271, 1427–1431. Knott, T. J., and Scott, J. (1986). cDNA sequence and chromo- Leveén, P., Pekny, M., Gebre-Medhin, S., Swolin, B., Larsson, E., somal localization of human platelet-derived growth factor and Betsholtz, C. (1994). Mice deficient for PDGF-B show renal, A-chain and its expression in tumor cell line. Nature 320, cardiovascular and hematological abnormalities. Genes Dev. 8, 695–699. 1875–1887.

Lindahl, P., Johansson, B. R., Leveén, P., and Betsholtz, C. (1997a). Raines, E. W., and Ross, R. (1993). Platelet-derived growth factor in Pericyte loss and microaneurysm formation in PDGF-B-deficient vivo. In “Biology of Platelet-Derived Growth Factor” (B. West- mice. Science 277, 242–245. mark and C. Sorg, Eds.), pp. 74–114. Karger, Basel. Lindahl, P., Karlsson, L., Hellstro¨m, M., Gebre-Medhin, S., Wil- Roberts, W. M., Look, A. T., and Sherr, C. J. (1988). Tandem linkage letts, K., Heath, J. K., and Betsholtz, C. (1997b). Alveogenesis of human CSF-1 receptor (c-fms) and the PDGF receptor genes. failure in PDGF-A deficient mice is coupled to lack of distal Cell 55, 655–661. spreading of alveolar smooth muscle cell progenitors during lung Ross, R., Raines, E. W., and Bowen, P. D. (1986). The biology of development. Development 124, 3943–3953. platelet-derived growth factor. Cell 46, 155–169. Lindahl, P., Hellstro¨m, M., Kaleń, M., Karlsson, L., Pekny, M., Soriano, P. (1994). Abnormal kidney development and hematologi- Pekna, M., Soriano, P., and Betsholtz, C. (1998). Paracrine PDGF- cal disorders in PDGFB-receptor mutant mice. Genes Dev. 8, B/PDGF-R␤ signaling controls mesangial cell development in 1888–1896. kidney glomeruli. Development 125, 3313–3322. Soriano, P. (1997). The PDGF ␣ receptor is required for neural crest Mercola, M., Wang, C., Kelly, J., Brownlee, C., Jackson-Grusby, L., cell survival and for normal patterning of the somites. Develop- Stiles, C., and Bowen-Pope, D. F. (1990). Selective expression of PDGF-A and its receptor during early mouse embryogenesis. ment 124, 2691–2700. Dev. Biol. 138, 114–122. Uiker, S., and Kriz, W. (1995). Structural analysis of the formation Ohlsson, R., Larsson, E., Nilsson, O., Wahlstro¨m, T., and Sund- of glomerular microaneurysms in the Habu venom model. Vir- stro¨m, P. (1989). Blastocyst implantation precedes induction of chows Arch. 426, 281–293. insulin-like growth factor II gene expression in human tropho- Yarden, Y., Escobedo, J. A., Kuang, W.-J., Yang-Feng, T. L., Daniel, blasts. Development 106, 555–559. T. O., Tremble, P. M., Chen, E. Y., Ando, M. E., Harkins, R. A., Palmieri, S. L., Pagne, J., Stiles, C. D., Biggers, J. D., and Mercola, Francke, U., Fried, V. A., Ullrich, A., and Williams, L. T. (1986). M. (1992). Expression of mouse PDGF-A and PDGF-␣-receptor Structure of the receptor for platelet-derived growth factor helps genes during pre- and post-implantation development. Evidence define a family of closely related growth factor receptors. Nature for a developmental shift from an autocrine to a paracrine mode 323, 226–232. of action. Mech. Dev. 39, 181–191. Raines, E. W., Bowen-Pope, D. F., and Ross, R. (1990). Platelet-derived growth factor. In “Handbook of the Experimental Pharmacology. Received for publication October 19, 1998 Peptide Growth Factors and Their Receptors” (M. B. Sporn and A. B. Revised April 16, 1999 Roberts, Eds.), pp. 173–262. Springer-Verlag, Heidelberg. Accepted April 16, 1999