Inner histopathologic changes and disproportionate zone volumes in foetal growth plates following gestational hypoglycaemia in rats

Maternal hypoglycaemia throughout gestation until gestation day (GD)20 delays foetal growth and skeletal development. While partially prevented by return to normoglycaemia after completed organogenesis (GD17), underlying mechanisms are not fully understood. Here, we investigated the pathogenesis of these changes and significance of maternal hypoglycaemia extending beyond organogenesis in non-diabetic rats. Pregnant rats received insulin-infusion until GD20 or GD17, with sacrifice on GD20. Hypoglycaemia throughout gestation increased maternal corticosterone levels, which correlated with foetal levels. Growth plates displayed central histopathologic changes comprising disrupted cellular organisation, hypertrophic chondrocytes, and decreased cellular density; expression of pro-angiogenic factors, HIF-1α and VEGF-A increased in surrounding areas. Disproportionately decreased growth plate zone volumes and lower expression of the structural protein MATN-3 were seen, while bone ossification parameters were normal. Ending maternal/foetal hypoglycaemia on GD17 reduced incidence and severity of histopathologic changes and with normal growth plate volume. Compromised foetal skeletal development following maternal hypoglycaemia throughout gestation is hypothesised to result from corticosterone-induced hypoxia in growth plates, where hypoxia disrupts chondrocyte maturation and growth plate structure and volume, decreasing long bone growth. Maternal/foetal hypoglycaemia lasting only until GD17 attenuated these changes, suggesting a pivotal role of glucose in growth plate development.

Maternal hypoglycaemia throughout gestation until gestation day (GD)20 delays foetal growth and skeletal development. While partially prevented by return to normoglycaemia after completed organogenesis (GD17), underlying mechanisms are not fully understood. Here, we investigated the pathogenesis of these changes and significance of maternal hypoglycaemia extending beyond organogenesis in non-diabetic rats. Pregnant rats received insulin-infusion until GD20 or GD17, with sacrifice on GD20. Hypoglycaemia throughout gestation increased maternal corticosterone levels, which correlated with foetal levels. Growth plates displayed central histopathologic changes comprising disrupted cellular organisation, hypertrophic chondrocytes, and decreased cellular density; expression of pro-angiogenic factors, HIF-1α and VEGF-A increased in surrounding areas. Disproportionately decreased growth plate zone volumes and lower expression of the structural protein MATN-3 were seen, while bone ossification parameters were normal. Ending maternal/foetal hypoglycaemia on GD17 reduced incidence and severity of histopathologic changes and with normal growth plate volume. Compromised foetal skeletal development following maternal hypoglycaemia throughout gestation is hypothesised to result from corticosterone-induced hypoxia in growth plates, where hypoxia disrupts chondrocyte maturation and growth plate structure and volume, decreasing long bone growth. Maternal/foetal hypoglycaemia lasting only until GD17 attenuated these changes, suggesting a pivotal role of glucose in growth plate development.
It has previously been established that maternal hypoglycaemia may cause foetal growth restriction and in rodents, even short-term, maternal hypoglycaemia may disturb foetal skeletal development, reflected in decreased ossification and bone malformations [1][2][3][4] . The underlying mechanisms are, however, not clear; especially an understanding of the critical time windows of maternal hypoglycaemia and its duration to allow for normal versus abnormal foetal development is still lacking. We therefore established a model in non-diabetic rats to study the effect of sustained insulin-induced hypoglycaemia on foetal skeletal development 5 . Previously we showed that continuous maternal hypoglycaemia throughout gestation resulted in skeletal malformations, as well as decreased skeletal growth and bone mineral density 5 . In comparison hypoglycaemia lasting only until GD17, when organogenesis is completed, followed by normoglycemia allowed for near-normal development, emphasising that sufficient glucose supply late in gestation is crucial 6 .
To elucidate the principal mechanisms underlying the observed foetal skeletal defects, we have taken a wider approach to assess histological changes and to identify potentially regulatory pathways for bone growth and calcification. We hypothesised that hypoglycaemia disrupts regulation of the skeletal micro-environment by altering growth plate structure, as well as gene expression, and hormones essential for normal development of growth In contrast, when infused only until end of organogenesis (HI-EoOrg) approximately half the foetuses displayed histopathologic changes, and these were of moderate grade (Fig. 1c). Compared to CTRLs, the changes in the HI-EoOrg group were more pronounced.
Immunohistochemistry. In controls, extracellular collagen II staining was observed throughout the growth plates (Fig. 2a); similarly, in the HI-EoGest group, staining was present in the central hypocellular area (Fig. 2b). In controls, collagen X distribution in growth plates clearly defined the boundary between proliferative and hypertrophic zones through positive staining of peri-cellular regions of hypertrophic chondrocytes (Fig. 2c). While the HI-EoGest showed a similar pattern, the cell types were not easily recognized, and the positive zone was markedly disordered (Fig. 2d).
In situ hybridization. HI-EoGest foetal growth plates had pronounced positive staining for the pro-angiogenic factors HIF-1α and VEGF-A, locally restricted to the margin of the histopathologic changes (Fig. 3b,d). No positive staining was observed in controls (Fig. 3a,c), except for a few scattered positive cells in the degenerative zone and peri-articular cartilage in some of the control animals, which were not easily recognisable even at 100x magnification. Additionally, the peri-articular cartilage of foetal tibial and meta-tarsal joints displayed a markedly local increased HIF-1α expression (Fig. 3e-h). VEGF-A showed a similar expression pattern (data not shown).
As some basal expression of HIF-1α and VEGF-A is expected in growth plates, the lack of positive staining is likely due to the sensitivity being too low to detect this; however, since the staining was very marked in the cells surrounding the changes, showing a pronounced increase versus controls, this is not considered to significantly affect the results.
Quantitative histology. Stereologic volume quantification. Growth plates. HI-infusion throughout gestation (HI-EoGest) resulted in decreased total growth plate volume (p = 0.0001, Fig. 4a) and reduced reserve, proliferative, and degenerative zone volumes (p < 0.0001-0.003, Fig. 4b-e). Disproportionately decreased total growth plate and zone volumes were reflected by increased relative volumes of the proliferative and hypertrophic zones (p = 0.0192; p = 0.0239), but conversely decreased the relative volume of the degenerative zone (p = 0.0436,    5). Recovering normoglycaemia after organogenesis (HI-EoOrg) resulted in increased total growth plate volume compared to the HI-EoGest (p = 0.0140), reaching control levels; similarly, the reserve, proliferative, and degenerative zone volumes did not differ from controls (p = 0.0007-0.0490). Relative zone volumes were in-between group HI-EoGest and CTRL levels.
Spongious zone. The absolute spongious zone volume was decreased in HI-EoGest animals (p = 0.0042), whereas volumes for the HI-EoOrg group ranged between HI-EoGest and CTRL groups (Fig. 4f). The ratio of spongious zone: total growth plate volumes was not affected by HI-infusion (mean ratios 0.10-0.11 ± 0.01-0.02 for all three groups; mean CE ≤ 0.05 and SD ≤ 0.01 for all).
Gene expression analysis. RT-qPCR quantification of foetal bone tissue mRNA levels. HI-infusion did not alter mRNA levels of cartilage collagens, hypoxia-related genes, the structurally related gene comp, genes related to the IGF-1 pathway nor osteoblast/osteoclast differentiation and activity (data not shown). In contrast, mRNA level of Matn-3, an additional structurally related gene, was decreased in HI-EoGest animals (p = 0.0222, data no shown), whereas recovering from hypoglycaemia after organogenesis (HI-EoOrg) resulted in normal mRNA level of this gene. www.nature.com/scientificreports www.nature.com/scientificreports/ plasma hormone levels. Maternal and foetal corticosterone. Maternal corticosterone levels were increased in HI-infused animals on GD17 (HI-GD17; p < 0.0082) and GD18 (HI-EoGest, p = 0.0241; HI-EoOrg p = 0.0172. Figure 6a,b), with no difference between these groups. On GD20, HI-EoGest corticosterone levels remained elevated, whereas HI-EoOrg levels had returned to normal (p < 0.0001, Fig. 6c).
As maternal and litter corticosterone levels in the HI-EoGest group were elevated on GD20, we tested the correlation between levels in individual dams and corresponding litter and this was strong (Pearson r = 0.834, p = 0.010, Supplementary Fig. S2).

Discussion
We have previously used HI-infusion in a pregnant non-diabetic rat model to investigate the effects from continuous hypoglycaemia on foetal skeletal development. We showed that maternal hypoglycaemia throughout gestation delays foetal growth and skeletal development, while hypoglycaemia only until after completion of organogenesis (GD17) allows for near-normal development 6 . The mechanisms underlying these developmental changes are not fully understood. In order to elucidate this, we extended our investigation to assess for histopathologic changes and potential regulatory players in foetal bone development, based on the hypothesis that hypoglycaemia disrupts regulation of the skeletal micro-environment. The present study clearly shows that continuous hypoglycaemia throughout gestation causes pronounced histopathologic changes in tibial growth plates including disproportionately altered zone volumes. These changes are highly similar to changes caused by hypoxia, leading us to propose hypoglycaemia-induced hypoxia as a pivotal mechanism ( Supplementary Fig. S5). When hypoglycaemia lasted only until GD17, histopathologic changes were less pronounced and growth plate volumes normal. The observed severely disrupted ECM architecture in central foetal growth plates following hypoglycaemia throughout gestation may have resulted, at least in part, from decreased expression of the ECM protein MATN-3, essential to structural integrity of the chondrocyte scaffold 7 . In contrast, expression of COMP, collagen IX, and collagen II were unaffected, although similarly important for growth plate structural integrity through the formation of interconnected complexes with MATN-3 7,8 . In murine growth plates, ablation of any one of these proteins causes a longitudinal, central disruption of the columnar chondrocyte organisation and decreased cell numbers, occasionally accompanied by hypertrophic chondrocytes with pyknotic nuclei 9-13 . These changes are very similar to the appearance and confinement of the histopathologic findings in the current study and those in studies by others following growth plate hypoxia [14][15][16][17] , hence the rational for our hypothesis. Reversely, MATN-3 decreases may be secondary to a hypoxia-induced disruption of the ECM architecture. The less pronounced chondrocyte disorganisation and normal MATN-3 expression after hypoglycaemia only until GD17 indicates that changes have either partly recovered after GD17 or the development has been attenuated. From this study, it cannot be concluded which of these scenarios take place, as growth plates were not evaluated on GD17.
In support of our interpretation, extended hypoglycaemia increased expression of the hypoxia-protecting factors, HIF-1α and VEGF-A, at the boundaries immediately surrounding the central histopathologic changes. Strikingly similar histopathologic changes were recorded in growth plates following ablation of either HIF-1α or VEGF-A in mice [14][15][16][17] and vice versa, exposure of cultured chondrocytes to hypoxia increases expression of these factors [18][19][20] . Moreover, in mice, growth plate hypoxia induced by chondrocyte-specific inactivation of the HIF-1α gene increases VEGF-A expression at the periphery of the accompanying central growth plate changes 17 , similar to our findings. In line with this, histological growth plate changes were confined to the longitudinal central area, which is particularly vulnerable to hypoxia, being avascular and relying on passive oxygen diffusion from adjacent blood vessels 17,21 . This suggests that the histologic growth plate changes seen here are due to central hypoxia, which in turn elicits a local counter-regulatory increase of HIF-1α and VEGF-A levels.
On GD20, the increased maternal corticosterone levels following hypoglycaemia throughout gestation are likely caused by the hypoglycaemia, and is a known counter-regulatory response, increasing glucose-output from the liver 22 , also explaining changes in maternal IGF-1 and thyroid levels [23][24][25] . Since maternal and litter plasma corticosterone levels correlated in the hypoglycaemic animals, we speculate that increased maternal corticosterone level in turn exposed the foetus to increased levels through trans-placental transport 26 . Subsequently, foetal corticosterone may have impeded vascularisation in the growth plate adjacent to the metaphyseal area. Corticosteroid exposure has been shown to decrease production of VEGF-A by growth plate chondrocytes in vitro and in vivo 27,28 , and induce growth plate hypoxia by decreasing density of adjacent blood vessels in mice and piglets 27,29 . This is also in line with our observation of increased HIF-1α and VEGF-A mRNA levels in foetal tibial and meta-tarsal joint articular cartilage, which is equally avascular and vulnerable to decreased peri-articular angiogenesis following increased blood corticosteroid levels in mice 27,30,31 . The observed secondary increases in anti-hypoxia factors may be a compensatory response to corticosterone-induced hypoxia. In line with this, www.nature.com/scientificreports www.nature.com/scientificreports/ ending hypoglycaemia on GD17 and normalising corticosterone levels may have interrupted the progression of the histopathologic changes.
In the above studies, no central histopathologic changes were observed in growth plates following corticosteroid exposure, however, this does not invalidate the present working hypothesis. Unlike the present study, administration was postnatal, and it is probable that the developing skeleton may be more vulnerable to corticosteroid exposure; counter-regulatory measures may also differ. Furthermore, temporary dosing was used, whereas in the present study, foetuses were exposed to high corticosterone levels, presumably, from conception.
Hypoglycaemia throughout gestation decreased total growth plate volume as anticipated based on previously observed shortened tibia length in litter-mates (foetuses from the same litters used for other investigations) 6 , which are known effects of glucocorticoid-dosing in rats and rabbits [32][33][34] . The disproportionate zone volume changes may be linked to reduced local VEGF-A levels. VEGF-A inactivation in growing mice expands the hypertrophic zone in tibial growth plates 35 , and chondrocyte-specific VEGF-A knock-out increases hypertrophic and decreases degenerative zone lengths in mouse embryo tibial growth plates (through delayed maturation/removal of hypertrophic chondrocytes) 15 . This is in line with the hypothesis of a key role of hypoxia in the present changes. Decreased MATN-3 expression may also be implicated, since MATN-3 KO promotes chondrocyte hypertrophy in mouse embryos 10,36 ; this is reflected by an increased hypertrophic zone length in tibial growth plates following premature maturation of proliferative into hypertrophic chondrocytes 10 . In contrast to hypoglycaemia lasting throughout gestation, hypoglycaemia ending on GD17 resulted in only slightly decreased total and separate growth plate volumes, which were not statistically different from controls; this was surprising given the shorter tibial length in litter-mates 6 . However, since this decrease in tibial length was less pronounced than after hypoglycaemia throughout gestation 6 , it possibly reflects delayed bone growth prior to GD17 extending beyond GD17, allowing only for an attenuation of the decrease.
In addition to a potential primary role of hypoxia in the development of the histopathologic growth plate changes, hypoglycaemia may also play an important role. The growth plate is a highly metabolic tissue, relying on anaerobic glycolysis 37 , and glucose serves as an essential source of energy 38 . Consequently, central chondrocytes may die due to glucose insufficiency and impaired glycolysis decreasing the energy supply. Apoptosis was seen despite local up-regulation of HIF-1α, which is known to increase expression of glucose transporters and enzymes of the glycolytic pathway [39][40][41] . However, this does not exclude a concurrent role of hypoxia in the changes; changes may be caused by a combination of hypoglycaemia and hypoxia.
Our data suggests that during foetal development, the growth plates are particularly vulnerable to maternal hypoglycaemia, more so than the ossification process, regardless of duration. High corticosteroid levels are known to decrease osteoblast and increase osteoclast activity and number/life-span in bone, leading to decreased bone formation and increased bone resorption ultimately reducing bone mineral density 30,[42][43][44] . However, following hypoglycaemia throughout gestation, parameters relating to the bone mineralisation process did overall not change, in line with our previous findings of unaffected tibial peak mineral density in litter-mates in this group 6 .
Our investigation has some limitations; firstly, animal models are not necessarily translatable to humans, which should be kept in mind when interpreting the results. Moreover, pregnant diabetic women would not be continuously hypoglycaemic following insulin-dosing. Secondly, the rats are not diabetic, therefore the influence of counter-regulatory responses might differ in diabetics. This model was not intended as an animal model of diabetes, but rather, a model of experimentally insulin-induced hypoglycaemia, to be used as a means of understanding mechanisms involved in normal skeletal development.
This strengthens the study in several aspects including that it provides a stable controlled model in which all animals were similar at the outset and unaffected by disease. It also allows for investigation of changes to foetal skeletal development during hypoglycaemic conditions. Moreover, to better understand duration of continuous hypoglycaemia on foetal skeletal development we included both a group with insulin-induced hypoglycaemia throughout gestation as well as a group with hypoglycaemia only until completion of organogenesis (GD17) in order to also study prevention of foetal changes. Additionally, insulin-infusion was well tolerated and less stressful than the otherwise necessary multiple daily dosing needed to approach continuous hypoglycaemia, allowing for better animal welfare. Finally, the breadth of analytical methods employed provides linkage between observed histopathological changes and potential regulatory factors.
In summary, this experimental rat model of insulin-induced hypoglycaemia on foetal skeletal development shows pronounced effects on growth plate organisation; however, discontinuation of insulin-infusion after organogenesis allows for a partly normal development. We speculate that the skeletal changes are induced by hypoxia secondary to increased corticosterone levels (Supplementary Fig. S5). The hypoglycaemia induces a counter-regulatory increase in maternal plasma corticosterone levels, which in turn increases foetal corticosterone levels, decreasing oxygen supply to central growth plates, causing hypoxia, reflected by local counter-regulatory increases in anti-hypoxic factors and hypoxia-like histopathologic changes. Future follow-up investigations to validate this hypothesis could include measurement of tissue oxygen level in the growth plates, as well as histologic evaluation of: i) metaphyseal vascularisation (e.g. by immunostaining for endothelial cell markers), ii) local HIF-1α and VEGF-A protein levels (e.g. by immunohistochemistry), to determine if they mirror the mRNA changes, and iii) the apoptotic phenotype of the growth plate chondrocytes (e.g. by TUNEL assay).
In conclusion, our study underlines the importance of sufficient foetal glucose availability during late gestation for support of normal skeletal development.

Materials and methods
Experimental design. The study design is presented in Fig. 7 and described in detail elsewhere 6 , but briefly, female Sprague-Dawley rats (approximately ten weeks old) were continuously infused with either HI or vehicle, starting one week prior to mating (Day 1).

Scientific RepoRtS |
(2020) 10:5609 | https://doi.org/10.1038/s41598-020-62554-2 www.nature.com/scientificreports www.nature.com/scientificreports/ The two experimental groups were: HI-infusion throughout and until end of gestation i.e. until gestation day (GD)20 (HI-EoGest); HI-infusion only until GD17 (i.e. approximate end of organogenesis) followed by a three-day infusion free period (HI-EoOrg). Control animals were infused with vehicle until GD20. Animals which died or were terminated prematurely were excluded; a total of 44 animals were infused with HI (16 received HI-infusion until GD20 and 20 until GD17); 29 were infused with vehicle. On GD17 eight animals from the HI-infused and control group were sacrificed for maternal blood sampling, these sub-groups were termed HI-GD17 and CTRL-GD17, respectively. At the end of the study (GD20) all remaining animals were sacrificed, and various parameters assessed (described below).
It has been confirmed that insulin-infusion decreased maternal blood glucose levels as intended, from 3 h after infusion-start 6 ; at the end of study/sacrifice HI-EoGest foetal blood glucose levels were also decreased.
All Sampling of foetal tissue. Within one hour of sacrifice, one foetus per litter was taken from a mid-horn position (alternating between left/right uterine horn) from approximately half the dams in each group (Fig. 7). From each foetus, the left hind limb was fixed in 10% neutral buffered formalin and paraffin embedded for histology. The paraffin block was sectioned into 5 μm thick lateral longitudinal sections, mounted on glass slides, and stored at −20 °C until staining. The starting point was set at the first section where tissue appeared. From and including the first section, every fifth section was stained with haematoxylin and eosin (H&E) for histopathologic evaluation and quantification of growth plate zone volumes. The next slide in line after each of these sections was used for quantification of osteoblast-osteoclast ratios in the primary ossification centre. Sections were scanned (Aperio AT2 scanner, Leica Biosystems, Nussloch, Germany) and histopathologic evaluations and volume quantifications were performed blinded. From a subset of animals in the CTRL and HI-EoGest group (see below), the next three adjacent slides (i.e. third, fourth, fifth) were used for either immunohistochemistry or in situ hybridization (see below).
Descriptive histology. Histopathologic evaluation. Foetal skeletal histopathologic changes were evaluated using a four-step grading scale based on severity and extent of changes (x10 and x100 magnification). Features for each grade were: Grade 0, no changes; Grade 1, minimal changes and only a slightly increased amount of www.nature.com/scientificreports www.nature.com/scientificreports/ extracellular matrix (ECM) across all zones, except the reserve zone; Grade 2, moderate changes, increased ECM and mild hypertrophy of chondrocytes across all zones, except the reserve zone; Grade 3, severe changes, characterised by increased though remarkably hypocellular ECM. The affected area extending through all growth plate zones.
Immunohistochemistry. To assess structural integrity and demarcation of the hypertrophic zone in tibial growth plates, distribution of collagen II and X was qualitatively evaluated by light microscopy from a subset of animals displaying severe (HI-EoGest group, n = 6-7 animals) or no changes (CTRL group, n = 4-5 animals). Representative sections, 2-4 sections per animal (slide 3, 4 or 5) were selected for evaluation. Immunohistochemistry was not performed in HI-EoOrg animals.
To identify osteoblasts and osteoclasts in primary ossification centres, tibial sections were stained for osterix (in osteoblast nuclei) and CD68 (localised in lysosome membranes of osteoclasts). Protocols are included in Supplementary Material.
In situ hybridization. To evaluate hypoxia-related changes in foetal growth plates, distribution of anti-hypoxic factors HIF-1α and VEGF-A was assessed in 2-4 representative sections from animals with severe or no changes (i.e. HI-EoGest and CTRL groups; 5 animals/group). In situ hybridization was not performed in HI-EoOrg animals. mRNA signal was detected using Automated Assay for Ventana Systems. The assay RNAscope 2.5 VS Reagent Kit-RED (322250,) Probe-Rn-Hif1a (432289) and Probe-Rn-Vegfa (315369) from Advanced Cell Diagnostics, Inc., Newark, CA, USA.
Quantitative histology. Stereologic quantification of volumes. The sampling method described above resulted in 11-22 H&E stained sections per foetus that contained growth-plate. Volumes of each growth plate zone and adjacent primary spongious zone were quantified based on the Cavalieri principle 49 . Demarcation of individual zones was defined according to the morphologic criteria of chondrocytes in each zone described by Burdan et al. 50 . NewCAST software (Visiopharm, Hoersholm, Denmark) was used; point-counting was based on preliminary counting on randomized sections, with a minimum 150-200 hit-points in total and a coefficient of error (CE) < 10%. Grid systems used 7 × 8 (reserve zone) and 12 × 12 points (remaining four zones) per view field (on-screen magnification x100). The sampling fraction was 100% of growth plate area on each section (shape factor 4, smooth organ). For each zone, volume and mean CE were calculated according to stereologic principles 49,51 .
The primary ossification centre was included in 6-18 sections/foetus. Osteoblast nuclei and osteoclast cytoplasm total volumes were estimated using a grid system with 2 × 2 and 5 × 5 points per view field (on-screen magnification x400). Sampling fraction was 100% of the primary ossification centre per section (shape factor 6, smooth organ). Volumes and mean CE were calculated as above followed by calculation of volume ratio.
Gene expression analysis. RT-qPCR quantification of foetal bone tissue mRNA levels. One foetus per litter was taken from a mid-horn position from the uterine horn (opposite to the above), hind limbs were sectioned at the hip-joint, soft tissue removed by microdissection, bones placed in aluminium foil envelopes, snap frozen in liquid nitrogen, and stored at −20 °C until RNA isolation. Total RNA was extracted using Trizol (15596026, Thermo Fisher Scientific, Waltham, MA, USA), purified using RNeasy Mini Kit (74104, QIAGEN AB, Sollentuna, Sweden), and reverse transcribed to cDNA using High-Capacity cDNA reverse Transcription Kit (4374966, Thermo Fisher Scientific). mRNA levels were quantified by Real-Time PCR using QuantStudio 7 Flex Real-Time PCR system and Taqman Gene Expression Assays (Applied Biosystems, Foster City, CA, USA).
Triplicate measurements were performed. If the standard deviation (SD) of mean C T values exceeded 0.167 and one measurement was a clear outlier, this was removed. If there was no clear outlier, C T values from all three measurements were included in the mean. The "∆∆C T Method" was used to calculate relative expression of each target gene. First the target gene C T value (C T , target) was normalized to the mean C T value of the two housekeeping (reference) genes, (C T , ref), for each animal and target gene separately (∆C T = C T , target-C T , ref). Thereafter, mean ∆C T and SD were calculated for each target gene within each group. The ∆∆C T value was then calculated by normalizing the group mean ∆C T for each test group to the mean ∆C T of the control group (∆∆C T = mean ∆C T , test group -mean ∆C T , control group). Data are reported as −∆∆C T values.

Quantification of hormones in plasma.
Sampling of blood for quantification of plasma hormone levels was performed on the following days: Maternal: on GD17, groups CTRL-GD17 (n = 8) and HI-GD17 (n = 8) were sacrificed and sampled. On GD18, CTRL, HI-EoGest, and HI-EoOrg groups were sampled (approximately 24 h after infusion-stop in the HI-EoOrg animals). On GD20, at sacrifice, all dams and litters (pooled blood) were sampled.