Fetus-derived IGF2 matches placental development to fetal demand

20 Growth of a fetus is dependent upon the functional capacity of its placenta, but how the latter is 21 matched to fetal demands is currently unknown. Critically, there is continuous expansion of the feto22 placental microvasculature throughout pregnancy, along with morphogenic modifications in the 23 overlying trophoblast epithelium. Here we demonstrate, through fetal and trophoblast specific 24 genetic manipulations in the mouse, that signalling by IGF2 from the feto-placental endothelium and 25 endocrine actions of circulating fetal IGF2 are required. We provide evidence that endothelial and 26 fetal-derived IGF2 plays an important role in trophoblast morphogenesis, acting through Gcm1 and 27 Synb. The effects on placental microvasculature expansion are mediated through IGF2R and 28 angiopoietin-Tie2/TEK signalling. Thus, our study reveals a direct role for IGF2-IGF2R axis on matching 29 fetal demand to placental supply and establishes the principle that hormone-like signals from the fetus 30 play important roles in the control of placental vascularization and trophoblast morphogenesis, 31 findings that have potential clinical implications. 32

Growth of a fetus is dependent upon the functional capacity of its placenta, but how the latter is 21 matched to fetal demands is currently unknown. Critically, there is continuous expansion of the feto-22 placental microvasculature throughout pregnancy, along with morphogenic modifications in the 23 overlying trophoblast epithelium. Here we demonstrate, through fetal and trophoblast specific 24 genetic manipulations in the mouse, that signalling by IGF2 from the feto-placental endothelium and 25 endocrine actions of circulating fetal IGF2 are required. We provide evidence that endothelial and 26 fetal-derived IGF2 plays an important role in trophoblast morphogenesis, acting through Gcm1 and 27 Synb. The effects on placental microvasculature expansion are mediated through IGF2R and 28 angiopoietin-Tie2/TEK signalling. Thus, our study reveals a direct role for IGF2-IGF2R axis on matching 29 fetal demand to placental supply and establishes the principle that hormone-like signals from the fetus 30 play important roles in the control of placental vascularization and trophoblast morphogenesis, 31 findings that have potential clinical implications. 32 33

Main 34
The mammalian fetus is totally dependent upon the placenta for nutrients and oxygen. Little is known, 35 however, about how placental functional capacity is matched to fetal demands. As gestation 36 progresses, the increase in fetal size requires a higher leveI of demand for nutrients and consequently 37 a higher level of supply. Depending on the species, the surface area for nutrient exchange increases 5 38 to 15 fold between mid and late gestation 1 . This remarkable adaptation is likely to occur, at least in 39 part, in response to fetus-derived signals of demand, but this important principle remains untested. 40 We have proposed that imprinted genes, in particular Igf2, play central roles in controlling both the 41 fetal demand for, and the placental supply of, maternal nutrients 2,3,4 . The Igf2 (insulin-like growth 42 factor 2) gene encodes a small polypeptide that is highly abundant in both fetal tissues and fetal 43 circulation. It is one of the most potent growth factors during intrauterine development, affecting the 44 metabolism, proliferation, survival and differentiation of a wide variety of cell types 5,6,7,8 . In humans,45 reduced Igf2 expression contributes to the intra-uterine growth restriction in patients with  Russell syndrome (SRS) 9 . Conversely, bialellic Igf2 expression caused by loss of Igf2 imprinting is 47 observed in Beckwith-Wiedemann patients (BWS), a syndrome characterized by somatic overgrowth 48 and increased predisposition to tumours 9 . IGF2 exerts its effects by binding to several IGF/INS 49 receptors (IGF1R, INSR, IGF1/INSR hybrids, IGF2R) 10,11 . IGF2 binds to IGF2R with the highest affinity, 50 which leads to either IGF2 degradation in the lysosomes or signalling via G-proteins 10,12,13 . 51 Here, we apply novel genetic approaches to define the signalling mechanisms of demand to the 52 placenta by creating mouse models with a growth mismatch between the placenta and the fetus, using 53 genetic manipulations of the IGF system. We first show that circulating IGF2 levels increase in late 54 gestation, thus reflecting fetal size and higher demand. Decreasing the demand by lowering IGF2 levels 55 in both the fetus and circulation abolishes the capacity of the placenta to increase the surface area in 56 late gestation; conversely, increased demand by excess fetal IGF2 has the opposite effect. 57 Mechanistically, we show that fetus-derived and circulating IGF2 signalling is essential for the 58 appropriate growth of the fetal-derived vasculature and the underlying trophoblast. These effects are 59 mediated in part by IGF2-IGF2R signalling in the feto-placental vascular endothelium. Our work 60 demonstrates that the interaction of circulating IGF2 and endothelial IGF2 with the trophoblast is 61 essential for matching the placental surface area for nutrient exchange (supply) to the growth rate of 62 fetal tissues (demand). 63

Expansion of placental labyrinth coincides with elevated levels of circulating and endothelial IGF2 64
The gas and nutrient exchange layer of the mouse placenta (labyrinthine zone -Lz) increased in size 65 with gestational age (Fig. 1a), matching the fetal weight (Fig. 1b). Concomitantly, fetal plasma IGF2 66 increased approximately two-fold between E16 and E19 ( Fig. 1c). At these two developmental stages, 67 we also observed a significant and positive correlation between fetal plasma IGF2 and fetal weights 68 (Fig. 1d). Within the placental Lz, Igf2 expression was the highest in feto-placental endothelial cells 69 (FPEC) (Fig. 1e) and its mRNA levels increased approximately six-fold between E14 and E19 (Fig. 1f). 70 Igf2 ranked as the highest expressed gene in FPEC RNA-Seq transcriptome at E16, and several other 71 known imprinted genes 14 ranked in the top one hundred out of approximately 14,000 genes detected 72 ( Fig. 1g and Supplementary Table 1). IGF2 protein was also highly expressed in FPEC (Fig. 1h), and 73 significantly higher than in the surrounding trophoblast cells (Fig. 1i). 74

Fetal and endothelial IGF2 control placental labyrinthine expansion 75
To explore whether fetus-derived IGF2 plays a direct role in placental development, we first used a 76 conditional allele (Igf2 +/fl ) to delete Igf2 in the epiblast lineage using the Meox2 Cre line 15 ( Fig. 2a and  77 Extended Data Fig. 1). The deletion of Igf2 from embryonic organs and FPEC, but not extra-embryonic 78 tissues, led to placental growth restriction from E14 onwards (Fig. 2b). Stereological analyses indicated 79 that only the placental compartments containing embryonic-derived structures (i.e. Lz and the 80 chorionic plate -Cp) were smaller in the Meox2 Cre/+ ; Igf2 +/fl mutants (referred subsequently as Igf2 EpiKO ) 81 (Fig. 2c). The continuous expansion of the Lz, measured as volume increase, that occurs in late 82 gestation between E14 and E19 was severely compromised in mutants (Fig. 2c). The overall volume, 83 surface area and total length of fetal capillaries (FC) were normal at E14, but became abnormal from 84 E16 onwards ( Fig. 2d and Extended Data Fig. 2a). Notably, all other components of placental Lz, not 85 originating from the embryonic lineage, (i.e. labyrinthine trophoblast -LT, and maternal blood spaces 86 -MBS) were also reduced in volume, to a similar extent as the FC (Fig. 2d). These findings provide 87 evidence for a role of fetus-derived IGF2 on the expansion of placental Lz in late gestation. 88 IGF2 is highly expressed in FPEC as previously shown in Fig. 1e-i. Therefore, we next tested whether 89 endothelial-derived IGF2 plays a role in placental development. Igf2 deletion in the fetal endothelium,  90 including FPEC, using the Tek Cre line 16 ( Fig. 2e and Extended Data Fig. 3) led to a moderate but 91 significant fetal and placental growth restriction, evident from E16 onwards (Fig. 2f). Mutant Tek Cre/+ ; 92 Igf2 +/fl (referred subsequently as Igf2 ECKO ) placentae had reduced volumes of Cp and Lz at both E16 and 93 E19 (Fig 2g), but less striking when compared to Igf2 EpiKO mutants (Fig. 2c). Within the Lz, the LT was 94 reduced at both E16 and E19, while the MBS and FC were comparable to controls at E16, but 95 significantly reduced at E19 ( Fig. 2h and Extended Data Fig. 2b). 96 We conclude that the 'small' labyrinthine phenotype observed in Igf2 EpiKO mutants is more severe than 97 in Igf2 ECKO mutants, which suggests that full placental Lz expansion in late gestation requires both 98 fetus-derived and endothelial-derived IGF2. 99

Fetus-derived IGF2 is essential for placental morphogenesis and microvasculature expansion 100
To uncover the molecular mechanisms responsible for the placental Lz expansion, we first performed 101 microarray analysis in micro-dissected Lz samples from E19 Igf2 EpiKO mutants and controls. 102 Differentially expressed genes (DEG) were enriched in genes implicated in vasculature development 103 and immune responses ( Fig. 3a and Extended Data Fig. 4a,b). We identified a classic molecular 104 signature of impaired angiogenesis -reduced angiopoietin-Tie2/TEK signalling 17 ( Fig. 3b and  105 Supplementary Table 2). Lower levels of Angpt1 and Tek, and increased expression of Angpt2 were 106 validated by qRT-PCR in an independent set of biological samples in late gestation (Fig. 3b). Consistent 107 with the well-established roles of the angiopoietin-Tie2/TEK signalling in the control of endothelial cell 108 survival and proliferation 17 , placental TUNEL staining revealed a six-fold increase in apoptotic cell 109 frequency in mutants at E16, specifically in the Lz (Fig. 3c). CD31-stained (marking endothelial cells) or 110 methylene blue-stained resin sections revealed the presence of feto-placental capillaries lacking 111 endothelial cells, or obstructed and thrombotic capillaries surrounded by highly disorganized and 112 fragmented endothelial cells (Fig. 3d). These observations indicate that a large proportion of the 113 apoptotic cells are FPEC. Furthermore, endothelial cell proliferation measured by flow cytometry was 114 significantly reduced at E16 (Fig. 3e and Extended Data Fig. 4c), and this finding was confirmed by 115 immunofluorescence (Extended Data Fig. 4d). 116 In addition to vascular pathways, the expression microarrays also identified transcriptional 117 upregulation of genes related to immune responses and leukocyte migration (Fig. 3a). Among these 118 was Adgre1, a gene that encodes the glycoprotein F4/80, a highly specific cell-surface marker for 119 murine macrophages 18 . The up-regulation of Adgre1 was confirmed by qRT-PCR in placental Lz also at 120 E16 (Fig. 3f). Immunostaining for F4/80 showed that the total number of macrophages in Lz was 121 significantly higher in mutants than controls (Fig. 3f). Additionally, clusters of macrophages 122 surrounding feto-placental capillaries were found exclusively in mutants (Fig. 3g). Next, we assessed 123 the impact of the described increased cell death, reduced cell proliferation and macrophage 124 infiltration, on capillary remodelling across gestation by CD31 immunostaining. The density of FC was 125 dramatically reduced at E16 and E19, suggestive of a disproportionate loss of FPEC (Fig. 3h). 126 Importantly, the array data indicated downregulation of key genes involved in syncytiotrophoblast 127 differentiation (i.e. Gcm1 and Synb -which are expressed specifically in layer II of the 128 syncytiotrophoblast, SynT-II, which is closest to FC; see Supplementary Table 2). To validate these  129  observations, we performed qRT-PCR for, and confirmed significant transcriptional reductions of,  130 SynT-II-specific genes 19,20 Gcm1, Synb and Slc16a3 (Fig. 3i). However, only the SynT-I specific 19,20,21  131 gene Slc16a1 was modestly down-regulated, but not Ly6e and Syna (Extended Data Fig. 4e). 132 Together, our data show that lack of fetus -derived IGF2 triggers dysregulation of angiopoietin-133  Tie2/TEK signalling in late gestation, with consequent reduced FPEC proliferation and excessive cell  134 death with associated placental macrophage infiltration. It also highlights that fetus-derived IGF2 135 supports normal development of the trophoblast cells, particularly the SynT-II layer, in a 136 paracrine/endocrine manner, with a knock-on effect on the development of maternal blood spaces. 137 Endocrine IGF2 is a fetus-derived signal that matches placental supply to fetal demand 138 To provide further insights into the roles of fetus-derived IGF2 in matching supply to fetal demand we 139 analysed five genetic models with either deletion of Igf2 in fetal tissues, endothelium, trophoblast or 140 ubiquitously, or overexpression of Igf2 in fetal tissues (Fig. 4). For these models we used flow 141 cytometry to count FPEC (defined as CD31 + /CD41cells 22 ) and measured labyrinthine weight and 142 circulating IGF2 levels. In Igf2 EpiKO mutants, as expected from the immunostainings shown in Fig. 3h, 143 we observed a severe deficit in the total number and the proportion of FPEC at E16 and E19, but 144 normal values at E14 (Fig. 4a). The linear Lz expansion expected with gestational age was not observed 145 in this model, matching the severe reductions in FPEC numbers and circulating IGF2 (Fig. 4a). In 146 contrast, in Igf2 ECKO mutants lacking endothelial Igf2, circulating levels of IGF2 were only moderately 147 reduced and total numbers of FPEC, but not relative numbers, were only significantly reduced at E19 148 (Fig. 4b). Lz expansion in this model was only blunted at the end of gestation (Fig. 4b). A deletion of 149 Igf2 specifically in the trophoblast cells of the placenta using Cyp19 Cre (Igf2 +/fl ; Cyp Cre/+ referred 150 subsequently as Igf2 TrKO ) 23 ( Fig. 4c and Extended Data Fig. 5a-e) did not result in changes in FPEC 151 numbers and circulating IGF2, demonstrating that FPEC expansion is independent of trophoblast-152 derived IGF2. Consequently, Lz expansion was normal in this model (Fig. 4c). Ubiquitous deletion of 153 Igf2 in embryo and trophoblast using CMV Cre (Igf2 +/fl ; CMV Cre/+ referred subsequently as Igf2 UbKO ) 24 (Fig. 154 4d and Extended Data Fig. 5f) led to a loss of FPEC similar to that observed in the Igf2 EpiKO mutants, 155 further demonstrating that trophoblast-derived IGF2 does not contribute significantly to FPEC 156 expansion. Lz weight was severely reduced from E14, in line with the near complete absence of IGF2 157 in fetal circulation (Fig. 4d). Conversely, reactivating the transcriptionally silent maternal Igf2 allele in 158 H19DMD fl/+ ; Meox2 +/Cre mutants 25 (referred subsequently as H19-DMD EpiKO ) ( Fig. 4e and Extended Data 159 Fig. 5g,h), which led to increased levels of circulating IGF2, was associated with an increase of Lz weight 160 and higher numbers of FPEC at E16 and E19 (Fig. 4e). 161 Taken together, these results show that IGF2 produced by fetal organs and secreted into the fetal 162 circulation stimulates the expansion of placental Lz, matching FPEC numbers to the fetal demand. 163

IGF2 signalling controls expression of FPEC-derived angiogenic factors 164
We hypothesised that the interaction of circulating IGF2 and the trophoblast, via FPEC, are key events 165 underlying the feto-placental microvascular remodelling. To establish the molecular signatures of IGF2 166 effects on FPEC we carried out RNA-Seq analysis on FACS-isolated endothelial cells from E16 placental 167 Lz of Igf2 EpiKO mutants and controls ( Fig. 5 and Extended Data Fig. 6). Gene ontology (GO) analysis of 168 DEGs showed statistical enrichment of biological processes related to immune responses, cell 169 migration, impaired cell proliferation and angiogenesis, extracellular matrix organization and response 170 to hypoxia (Fig. 5a,b and Supplementary Significantly, the four TFs control the expression of several proteins involved in angiogenesis (labelled 179 with * in Fig. 5e and further presented in Supplementary Table 4), some of which are secreted by the 180 endothelial cells into the extracellular space (Supplementary Table 4). This analysis also highlighted 181 several chemokines that were up-regulated in FPEC (such as CCL2 33 and IL15 34 ) that are likely involved 182 in attracting and modulating the activity of macrophages that surround the feto-placental capillaries 183 (as shown in Fig. 3g). Thus, we established that IGF2 signalling is necessary for proliferation and 184 survival of FPEC and modulates their angiogenic properties. 185

IGF2 signalling on FPEC is independent of IGF1R and is mediated by IGF2R in vitro and in vivo 186
To further investigate the role of IGF2 in fetal capillary remodelling and identify the receptors that 187 might mediate its effects on endothelial cells, we isolated primary FPEC from E16 wild-type placental 188 Lz and cultured them ex vivo (Extended Data Fig. 7a-c). Only the type I (Igf1r) and type II (Igf2r) 189 receptors were expressed in FPEC both in vivo and ex vivo (Fig. 6a,b and Extended Data Fig. 7d). 190 Exposure of cultured FPEC, which express low levels of Igf2, to exogenous IGF2 significantly increased 191 their ability to form capillary-like tube structures when seeded on matrigel (Extended Data Fig. 7e and 192 Fig. 6c), demonstrating that IGF2 exerts direct angiogenic effects on FPEC. We also exposed cultured 193 FPEC to IGF2 Leu27 , an analogue previously shown to bind to IGF2R with high selectivity 35 , which 194 stimulated capillary-like tube formation although to a lesser extent compared to IGF2 (Fig. 6b, We further confirmed these in vitro findings by knocking-out these receptors (IGF1R and IGF2R) in 201 vivo. Accordingly, efficient deletion of Igf1r from the endothelium (Igf1r ECKO ) did not have any 202 significant impact on fetal, whole placenta or placental Lz growth kinetics, nor did it alter the total and 203 relative numbers of FPEC/Lz, apart from a slight increase in the percentage of FPEC at E19 (Extended 204 Data Fig. 8a-e). Strikingly, the deletion of Igf2r from the endothelium (Igf2r ECKO -Extended Data Fig.  205 8f,g) resulted in a reduction in the percentage of FPEC/placental Lz at both E16 and E19, further 206 confirmed by a reduced density of CD31 + cells by immunofluorescent staining (Fig. (Fig. 6d, e). The 207 total number of FPEC/Lz was also significantly reduced at E16, but became normal at E19 (Fig. 6d), 208 with Lz being overgrown from E16 onwards (Fig. 6f) coincident with an increase in levels of circulating 209 IGF2 in plasma (Fig. 6g). Together, our in vitro and in vivo experiments demonstrate that IGF2R 210 mediates, at least partially, the signalling actions of IGF2 on FPEC. 211 212

214
The major finding of this study is the demonstration that fetal growth demand signals are major 215 regulators of placental development and function. Although a vast number of genetic pathways have 216 been discovered that are important for the development of different cell types in the placenta and 217 the fetus, there are no functional genetic investigations to date on how the fetus signals demand to 218 the placenta and how the placenta matches the fetal demands. We tackled these questions with an 219 innovative experimental design, which is based on the manipulation of the growth rate of fetal tissues 220 independent of the placenta, and vice-versa, in the mouse. We used conditional targeting of imprinted 221 genes with well-established growth functions (Igf2, Igf2r, H19) as model systems (importantly, due to 222 imprinting, the mother is phenotypically normal). The analysis of these models of mismatch between 223 supply and demand allowed us to establish a number of key mechanistic principles that regulate the 224 cooperative signalling between the fetus and the placenta and, consequently, the control of maternal 225 resources. 226 Firstly, we found that circulating IGF2 correlates positively with fetal size in late gestation, reflecting 227 the growth rate of fetal tissues and the demand for nutrients. Mice with a severe decrease in levels of 228 circulating/fetal IGF2, and thus fetal demand, showed a drastic (and disproportionate) loss of feto-229 placental endothelial cells. This severe placental angiogenesis phenotype was associated with reduced 230 endothelial cell proliferation and increased apoptosis, altered differentiation of the overlying 231 trophoblast and reduced density of maternal blood spaces, ultimately leading to a failure in the 232 expansion of the labyrinthine layer and surface area for nutrient transport. Conversely, increased 233 demand for nutrients caused by bi-allelic Igf2 expression, which drove higher growth rates, led to 234 'overexpansion' of the labyrinthine layer. Secondly, we also found that feto-placental endothelial cells 235 are a significant source of IGF2, with levels increasing with gestational age. Endothelial Igf2-deficient 236 mice show modest reductions in circulating IGF2 and impaired expansion of the microvasculature and 237 labyrinthine layer, but no disproportionate reduction in number of placental endothelial cells (which 238 is only seen when circulating IGF2 is severely reduced). These findings establish the important 239 principle that hormone-like signals from the fetus, such as IGF2, are required for the normal expansion 240 of the labyrinthine layer and surface area of the placenta. 241 Based on the experimental evidence provided in this study, we propose a model ( Fig. 6h) in which 242 fetus-derived IGF2 is the signal that allows matching placental supply capacity to fetal demand. At the 243 placenta interface, circulating IGF2 directly stimulates endothelial cell proliferation and survival, and 244 capillary branching through IGF2R (as shown in vivo and ex-vivo). Circulating IGF2 may also directly 245 control the growth and differentiation of the underlying trophoblast, as it can cross (in free form or in 246 binary complexes) the capillary walls or permeate through the fenestrated endothelium 37 . We suggest 247 that the feto-placental endothelium is a large reservoir of IGF2, boosting further IGF2 signalling, and 248 acting in a paracrine and autocrine manner to control the growth and remodelling of fetal capillaries, 249 and trophoblast morphogenesis. Importantly, the effect of IGF2 signalling on feto-placental 250 microvascular remodelling seems specifically driven by fetus-derived IGF2. Accordingly, we did not 251 find any evidence that IGF2 produced by the trophoblast has a direct role on vascularization, being 252 instead required for trophoblast morphogenesis. We therefore suggest that the key role of circulating 253 IGF2 is to provide fetus-derived angiogenic signals to promote the vascular tree expansion in later 254 gestation, in conjunction with local IGF2, derived from the fetal endothelium of the placenta. 255 Mechanistically, the most likely molecular effectors of fetus-derived IGF2 signalling on 256 microvasculature expansion and trophoblast morphogenesis are the angiopoietin-Tie2/TEK signalling 257 and the key trophoblast differentiation genes Gcm1 and Synb, respectively. 258 Our study has a number of important implications. It provides insights into the complex interplay 259 between trophoblast branching morphogenesis and placental vascularization. To our knowledge, IGF2 260 is the first example of a hormone-like molecule that signals fetal demand to the placenta by adapting 261 the expansion of feto-placental microvasculature and trophoblast morphogenesis to the embryo size. 262 Matching placental supply to fetal demand also involves IGF2R -the other imprinted member of the 263 IGF family 38 . The imprinting of the IGF system is thus likely to have played a key evolutionary role in 264 . CC-BY-NC-ND 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted January 16, 2019. ; https://doi.org/10.1101/520536 doi: bioRxiv preprint the origins of the expansion of the feto-placental microvasculature and surface area for nutrient 265 transport throughout pregnancy -a fundamental biological process that is observed in all eutherian 266 species 1 . In humans, circulating levels of IGF2 in the umbilical cord progressively increase between 29 267 weeks of gestation and term, similarly to our findings in the mouse 39  CC-BY-NC-ND 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted January 16, 2019. ; https://doi.org/10.1101/520536 doi: bioRxiv preprint were loaded with 20µl ELISA Diluent RD5-38 per well, plus 10µl standard or plasma (diluted 50 fold in 308 RIPA buffer, Sigma -R0278). The plates were then sealed and incubated for two hours at room 309 temperature on a plate shaker. After three washes with MSD wash, the wells were plated with 25µl 310 detection antibody (Biotinylated Goat Anti-Mouse IGF-II, R&D Systems -840963), diluted to 0.72 311 µg/ml in PBS, sealed, and incubated for one hour at room temperature on a plate shaker. Following 312 three additional washes with MSD wash, the wells were plated with 25µl MesoScale Discovery 313 Streptavidin Sulpho-TAG, diluted 1:1000 in the MSD Diluent 100, sealed and incubated for 30 minutes 314 at room temperature on a plate shaker. After three final washes with MSD wash, the wells were plated 315 with 150µl of MSD Read Buffer T (1x) and the reading was performed on the MSD s600 analyser. Each 316 sample was measured in duplicate and the results were calculated against the standard curve, using 317 the MSD Workbench Software. 318

Western blot analysis 340
Tissues were lysed in ~10μl/mg tissue RIPA buffer (Sigma -R0278), then the lysates were spun at 3,000 341 RPM and 4°C for 15 minutes. The supernatants were transferred into new tubes and protein 342 concentrations were quantified using the Pierce BCA Assay Protein kit (Thermo Scientific -23225). 343 60μg total protein were mixed with SDS gel loading buffer, then denatured at 70°C for 10 minutes and 344 loaded into 12-well NuPAGE® Novex® 4-12% Bis-Tris precast gels. The pre-stained Novex Sharp protein 345 standard (Invitrogen -LC5800) was used as protein marker. After electrophoresis for 40 minutes at 346 200V and 4°C, the proteins were transferred onto nitrocellulose membranes, using the iBlot® Transfer 347 Stacks (Invitrogen IB 3010-01) and the iBlot® Gel Transfer Device set for 7 minutes at 20V. Blocking 348 was performed for one hour at 4°C in 5% semi-skimmed milk (Marvel) dissolved in TBS-T. The 349 membranes were then incubated overnight at 4°C with the primary antibody dissolved in 0.5% milk in 350 TBS-T (goat anti-human IGF2, 1:1,000, R&D AF292-NA or goat anti-mouse SOD1, 1:50,000, R&D 351 AF3787). After 2x10 minutes washes with milliQ water and 2x10 minutes washes with TBS-T, the blots 352 were incubated for one hour at room temperature with the secondary antibody dissolved in TBS-T 353 containing 3% semi-skimmed milk (rabbit anti-goat IgG-HRP, 1:2,500, Santa Cruz sc-2768). The blots 354 were then washed as above, exposed to substrate (Clarity ECL Western Blotting Substrate, Biorad) for 355 5 minutes and imaged with the Biorad GelDoc system. Stripping of antibodies was carried out using a 356 stripping buffer (ThermoFisher -21059) for 15 minutes at room temperature. The band intensities 357 were quantified using the ImageLab software (Biorad) and expressed as IGF2/SOD1 ratios. 358

Placenta stereology 359
Placenta stereology analyses were performed as described 52 in placentae (n=5-7) collected from three 360 litters at each developmental stage. Briefly, the placentae were weighted, then halved and each half 361 placenta weighted again. A half was fixed in 4% paraformaldehyde in PBS at 4°C overnight, then 362 dehydrated and embedded in paraffin wax. The paraffin blocks were exhaustively sectioned using a 363 microtome at 7μm thickness. Placental sections spaced 140 μm apart were hematoxylin-eosin stained 364 and stereological measurements of placental layers were done using the NewCAST system 365 (Visiopharm, Hoersholm, Denmark), using the point counting method 52 . 366 The corresponding placental halves were fixed for 6 hours with 4% glutaraldehyde in 0. quantity and quality were verified using RNA 6000 Nano Kit (Agilent -5067-1511) and an Agilent 2100 395 Bioanalyzer. Only RNA samples with RNA integrity numbers (RIN) >9.0 were used. Array profiling was 396 . CC-BY-NC-ND 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made Cytometry Assay Kit (ThermoFisher -C10420), according to manufacturer's instructions. Flow 430 cytometry analysis was performed using a BD LSRFortessa cell analyser (BD Biosciences). FSC files were 431 analysed with the FlowJo_V10 software, using single-cell discrimination and gating based on single-432 stained controls. Proliferating FPEC were identified as viable EdU + /CD31 + /CD41cells. 433

FPEC isolation by FACS 434
For sorting, single cell preparation and staining for FPEC markers was performed as above. FACS was 435 done using an Aria-Fusion cell sorter (BD Bioscience), with exclusion of cell duplets and dying cells 436 (7AAD + ). Cell fractions (endothelial and non-endothelial cells) were then spun at 3,000 RPM and 4°C 437 for 3 min, the excess of sorting liquid was removed and cell pellets were flash frozen in liquid N2 and 438 stored at -80°C until used for RNA extraction. 439

Primary FPEC isolation, culture and tube formation assay 440
. CC-BY-NC-ND 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted January 16, 2019. Primary FPEC were isolated as previously described 54  Photographs were taken at 30 min, 4, 6 and 8 hours using an EVOS FL Cell Imaging system 459 (ThermoFisher Scientific). Each experiment was performed on 5-6 litters for every treatment. For each 460 tube formation assay, we used five wells seeded with primary FPEC exposed to the treatment agent 461 with equivalent numbers of the corresponding vehicle. Quantification of tubular network structures 462 was performed using the Angiogenesis Analyzer software in ImageJ 55 . 463  19219-19224 (2005). 526 . CC-BY-NC-ND 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  shared cis acting regulatory region upstream of H19. Genes Dev. 14, 1186-1195 (2000). 575

RNA-sequencing and data analysis
. CC-BY-NC-ND 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made . CC-BY-NC-ND 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made . CC-BY-NC-ND 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted January 16, 2019.  CC-BY-NC-ND 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted January 16, 2019.  CC-BY-NC-ND 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted January 16, 2019. was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made . CC-BY-NC-ND 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted January 16, 2019. ; https://doi.org/10.1101/520536 doi: bioRxiv preprint H19-DMD EpiKO (e). Columns 2 and 3: total numbers (column 2) and proportion of FPEC/placental Lz 720 (column 3), measured by flow cytometry (n conceptuses per group: Whitney tests (fifth column). 728 known regulators of angiogenesis (angiostatic or pro-angiogenic factors) and key references are listed 739 in Supplementary Table 4 representation of IGF2 and IGF receptors. IGF2 Leu27 analogue acts specifically on IGF2R and 747 picropodophyllin (PPP) inhibits phosphorylation of IGF1R. c, Representative images of capillary-like 748 tube formation assay in primary FPEC seeded on matrigel and exposed to exogenous IGF2, IGF2 Leu27 , 749 . CC-BY-NC-ND 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made Tek Cre/+ ) versus control (C -Igf2 +/fl ) placentae (n=5-7 per genotype at each developmental stage). For 800 all graphs, data is shown as averages ± SD; * P < 0.05, *** P < 0.001 calculated by two-way ANOVA 801 plus Sidak's multiple comparisons tests. double transgenic for TeK Cre and Ai9(RCL-tdT) reporter at E16 of gestation. Scale bars are 2 mm (fetus) 808 and 1 mm (placenta). b, The TeK Cre is not expressed in the syncytiotrophoblast layers, as demonstrated 809 by the lack of immunostaining overlap between the tomato protein (red) and MCT4 (Syn-TII layer) or 810 MCT1 (Syn-TI layer). Scale bars are 50 µm. c, Western blot analysis of pro-IGF2 (18 kDa) in cell lysates 811 from placental Lz micro-dissected at E16 and corresponding data quantification (n=3 per genotype). 812 SOD1 (19 kDa) was used as internal control for loading. d, Efficiency of Igf2 deletion evaluated by qRT-813 PCR in fluorescence-activated sorted FPEC (n=5-7 per genotype). e, Flow cytometry analysis shows 814 that the majority (>80%) of Igf2 ECKO mutant FPEC express YFP, thus demonstrating good efficiency of 815 TeK Cre in these cells (n=5-11 per genotype). f, Fetal and placental growth kinetics are not altered in 816 TeK Cre/+ carriers (maternal inheritance) at E19 (n=13-15 conceptuses per genotype from 4 817 independent litters). 818 819 . CC-BY-NC-ND 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted January 16, 2019. is not observed in non-endothelial cells from placental Lz measured by flow cytometry analysis after 828 EdU injections (16 hours exposure; n=4-11 per group CC-BY-NC-ND 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted January 16, 2019. ; https://doi.org/10.1101/520536 doi: bioRxiv preprint