Altered gene expression and spongiotrophoblast differentiation in placenta from a mouse model of diabetes in pregnancy
- First Online:
- Cite this article as:
- Salbaum, J.M., Kruger, C., Zhang, X. et al. Diabetologia (2011) 54: 1909. doi:10.1007/s00125-011-2132-6
- 703 Downloads
Pregnancies complicated by diabetes have a higher risk of adverse outcomes for mothers and children, including predisposition to disease later in life, e.g. metabolic syndrome and hypertension. We hypothesised that adverse outcomes from diabetic pregnancies may be linked to compromised placental function, and sought to identify cellular and molecular abnormalities in diabetic placenta.
Using a mouse model of diabetic pregnancy, placental gene expression was assayed at mid-gestation and cellular composition analysed at various stages. Genome-wide expression profiling was validated by quantitative PCR and tissue localisation studies were performed to identify cellular correlates of altered gene expression in diabetic placenta.
We detected significantly altered gene expression in diabetic placenta for genes expressed in the maternal and those expressed in the embryonic compartments. We also found altered cellular composition of the decidual compartment. In addition, the junctional and labyrinth layers were reduced in diabetic placenta, accompanied by aberrant differentiation of spongiotrophoblast cells.
Diabetes during pregnancy alters transcriptional profiles in the murine placenta, affecting cells of embryonic and maternal origin, and involving several genes not previously implicated in diabetic pregnancies. The molecular changes and abnormal differentiation of multiple cell types precede impaired growth of junctional zone and labyrinth, and of placenta overall. Regardless of whether these changes represent direct responses to hyperglycaemia or are physiological adaptations, they are likely to play a role in pregnancy complications and outcomes, and to have implications for developmental origins of adult disease.
KeywordsDiabetic placenta Diabetic pregnancy Gene expression profiling Gene regulation Hyperglycaemia Junctional layer Maternal diabetes Placental insufficiency Spongiotrophoblast Transcription factors
Database for Annotation, Visualization and Integrated Discovery
Hypoxia-inducible factor 1
Mouse Genome Informatics
Nuclear factor κB
Transcription factor binding site
Trophoblast-specific protein alpha
Diabetes in pregnancy is a well established risk factor for diabetic embryopathy, a spectrum of congenital malformations that include heart defects and neural tube defects [1, 2, 3]. Exposure to maternal diabetes during pregnancy also predisposes to disease later in life, e.g. metabolic syndrome, obesity, diabetes and hypertension [4, 5]. While mechanistic studies on developmental origins of adult disease are still in early stages, substantial progress has been made in identifying causes of the congenital abnormalities. Thus in rodent models, metabolic perturbations and nutrient deficiencies have been demonstrated, including perturbed prostaglandin metabolism and inositol deficiency. Oxidative stress and hypoxia have also been documented as causes , and antioxidant administration reduced the incidence of fetal malformations. In addition to increased apoptosis, developmental regulators  and signalling pathways [8, 9] are perturbed in exposed embryos, implicating these mechanisms in abnormal morphogenesis and congenital malformations in diabetic embryopathy.
There is, however, little knowledge about the basis for increased disease susceptibility later in life in progeny exposed to maternal diabetes. Once an embryo has escaped the critical time window of vulnerability to congenital defects, it nevertheless remains exposed to the diabetic environment. We have recently shown that this leads to reduction of intrauterine growth during later stages of gestation , possibly due to hyperglycaemia-induced changes in the placenta. It has been proposed that placental changes could be central to developmental programming of disease susceptibility of the offspring [11, 12], but the underlying molecular mechanisms are unknown. Altered placental weight has been reported for diabetic rats , and for human type 1  and gestational diabetic  pregnancies. Human type 1 diabetic placenta undergoes increased angiogenesis and exhibits vascular dysfunction, mediated through vascular endothelial growth factor and nitric oxide in early pregnancy, and through hyperinsulinaemia in the later trimesters . Furthermore, enrichment of glycogen  and excess lipid accumulation [18, 19] have been described in rat diabetic placenta.
To identify the molecular basis of placental abnormalities in diabetic pregnancies in the mouse, we used a streptozotocin-induced model of diabetes. Our results demonstrate that altered gene expression at mid-gestation precedes cellular and structural abnormalities in diabetic mouse placenta at later stages.
All experiments were conducted in accordance with the United States Public Health Service Policy on Humane Care and Use of Laboratory Animals (http://grants1.nih.gov/grants/olaw/references/phspol.htm) in a facility approved by the American Association for Accreditation of Laboratory Animal Care. Diabetes was induced in 7- to 9-week-old FVB female mice (Charles River Laboratories, Wilmington, MA, USA) by two i.p. injections of streptozotocin (100 mg/kg body weight) and matings were set up as described previously . Dams whose glucose levels in tail vein blood (Contour Glucometer; Bayer, Tarrytown, NY, USA) exceeded 13.9 mmol/l were considered diabetic; average glucose levels were 8.22 mmol/l (±1.0 SD) before streptozotocin treatment, 18.722 ± 4.4 mmol/l in diabetic dams on the day of mating and 29.33 ± 3.9 mmol/l on the day of placenta removal (from dams at gestational day E10.5, counting the day of detection of vaginal plug as E0.5).
Whole placentae were dissected at E10.5; only specimens associated with morphologically normal embryos were used. Decidua were removed from the uterus, and the yolk sac and embryos were dissected out; placenta samples consisted of embryonic and maternal tissue. Placentae were processed as individual samples following published methods . Samples were hybridised to 10 Affymetrix Mouse 430 2.0 arrays (Affymetrix, Santa Clara, CA, USA), which were scanned using a scanner (GeneChip3000, Affymetrix). Quality control was done with GeneChip Operating Software (Affymetrix) and statistical analyses were performed as previously described for embryos . The primary data files are available at the NCBI Gene Expression Omnibus repository (accession number GSE 28277).
Quantitative real-time PCR
Quantitative real-time PCR was performed as described  using the ABI Prism 7900HT device (Life Technologies, Carlsbad, CA, USA). Primers (electronic supplementary material [ESM] Table 1) were positioned to span exon–exon junctions and amplicons were designed to originate from a different transcript region than that detected by microarray probes. Complementary DNA samples were derived from individual E10.5 placentae from different diabetic and control dams than those from which samples were assayed on microarrays. For normalisation, we used measurement of polymerase epsilon 4 (Polε4, also known as Pole4) expression; Polε4 expression proved to be unaffected by maternal diabetes on the arrays.
Annotation for molecular function and mutant mouse phenotypes
Molecular function attributes were based on Gene Ontology annotation (NetAffx www.affymetrix.com/analysis/netaffx/index.affx), supplemented with information from PubMed and Database for Annotation, Visualization and Integrated Discovery (DAVID) (http://david.abcc.ncifcrf.gov/), which was also used to examine for enrichment of specific categories. Information on mouse mutants was obtained from Mouse Genome Informatics (MGI) (www.informatics.jax.org; accessed 20 December, 2010). For transcription factor binding site (TFBS) prediction, Whole Genome rVISTA (http://genome.lbl.gov/vista/index.shtml) was used to identify motifs that are conserved between mouse and human and are over-represented in the 5 kb upstream regions of diabetes-affected genes, relative to all 5 kb upstream regions in the genome (p < 0.005).
Placentae from normal and diabetic pregnancies were embedded in paraffin using a Shandon Excelsior tissue processor (Thermo, Waltham, MA, USA) and an embedder (Histocentre 3, Thermo), and sectioned at a thickness of 6 μm. Deparaffinisation and staining with haematoxylin/eosin was done in a Special Stainer device (ST5020; Leica Microsystems, Bannockburn, IL, USA), followed by coverslipping (CV5030; Leica). Slides were scanned on a digital pathology station (NDP Nanozoomer; Hamamatsu, Bridgewater, NJ, USA).
In situ hybridisation
In situ hybridisations were performed as published  to either 20 μm frozen sections or 6 μm sections of paraffin-embedded tissues. Processing of slides was done together for diabetic and control samples in the same vessel for all fixation, wash, hybridisation, incubation and development steps. Plasmids for probe generation were obtained from Open Biosystems (Huntsville, AL, USA). For comparison of control and diabetic tissues, photography was done with the same magnification and exposure.
Cell counting and statistical analysis
All sections were from the middle of the respective placentae. The sizes of placental compartments were determined as the respective area fractions in ImageJ (National Institutes of Health, Bethesda, MD, USA; http://rsbweb.nih.gov/ij/download.html, accessed 14 March 2011). Cell cluster counting was done by selecting the labyrinth only and counting of pixels with Tpbpa signal above a uniform threshold was performed. The validity of these methods was confirmed manually by two individuals, independently. For each placenta, data from ten sections were averaged and adjusted for size of the whole placenta. We included six placentae each in the control and diabetic groups for gestational day E15.5 and three specimens per group were used for E18.5. Each placenta came from a different pregnancy. Embryo and placenta size were averaged to each dam before averaging for the group. Statistical analysis was done by two-sided t test.
Mouse model of diabetic pregnancy
We employed the well-established streptozotocin-induced mouse model of diabetes, using the FVB inbred strain of mice. Embryos and placentae were collected at embryonic day 10.5 (E10.5) because at this stage, neural tube defects are easily detectable. Samples associated with malformed embryos were excluded from the analysis. Control embryos (n = 65 embryos from nine dams) had a crown–rump length of 5.17 ± 0.34 mm (mean ± SD). In diabetes-exposed embryos (n = 97 embryos from 13 dams) crown–rump length was 5.07 ± 0.39 mm. Differences were not statistically significant (p = 0.54). Placentae looked normal at this time point, but they differed (p = 0.0038) in size, with control placenta (n = 31 placentae from seven dams) having a wet weight of 42.87 ± 2.01 mg and diabetic placentae (n = 61 placentae from nine dams) weighing 37.92 ± 3.96 mg. From five control dams and five diabetic dams, we collected one placenta each of a weight close to the mean for the respective group and performed ten independent microarray assays (Affymetrix 430 2.0). Separate placentae were used for validation studies by quantitative real-time PCR. There were no significant differences in litter size at any of the gestational points investigated, except at E18.5, where control dams on average had 9.8 ± 2.31 embryos and diabetic dams had 8.1 ± 3.04 progeny.
Altered gene expression profiles in placentae from diabetic dams
Genes differentially expressed in diabetic placenta classified by cellular function
Extracellular matrix, cell adhesion, cell migration
Anxa4, Fam19a5, Ceacam15, Cldn23, Cldn3, Emid1, Emp2, Ggt a1, Gpld1, Igsf11, Lims1, Mmp9, Mpzl2, Pfpl, Pkp2, Pvrl1, Thsd4, Tmem30a, Tmem46
Soluble factors, hormones, ligands, ligand binding proteins
Angpt4, Bmp2, Calcb, Cck, Cx3cl1, Fgfbp3, Fstl3, Osm, Prl5a1, Ren1/Ren, Sfrp5, Tac4, Tgfbi, Thbs1, Thbs2
Cell surface receptors
Chrm1, Csf2rb, Cxcr4, Gpr161, Hhipl1, Ms4a4c, Tnfrsf1b
Ctnnd2, Dab1, Ehd3, Gnao1, Hipk2, Pctk3, Phip, Prkg2, Rhov, Rock1, Socs3, Sphk1
Ascl2, Atoh8, Cdx2, Creb3l3, E2f2, Fosl2, Foxk2, Hand2, Myb, Nfatc1, Prrx2, Smad6, Srebf2, Zbtb20, Zfp458
Ddx31, Fancm, Whsc1l1
Aldh5a1, Car5b, Ckmt1, Cyp1a1, Eno1, Haghl, Pgk1, Fabp4, Gdpd1, Mgam, Pank1, Tdo2
Lpl, Osbp, Star
Atp1a3, Ccbe1, Itpr2, Kcnk2, Kcnn4, Slc28a3, Slc41a3, Slc6a4, Slco2a1, Tmc5
Arl15, Cts7, Gcnt3, Il4i1, Lta4h, Pcsk5, Ppm1j, Prosc, Spsb4, Trip12, Trim63, Usp32
Actc1, Ankrd2, Dctn5, Kif18a, Phr1, Synpo2l, Tsga10
Ambra1, Incenp, Pacs2
Metal ion homeostasis
Hjurp, Hspa1a, Mrps23, Nup88, Rad51c, Rmi1, Sfrs18, Vps11, Vamp5
Bat2d, Bud31, Ccdc34, Ier5, Klhl5, Prdm10, Sbsn, Sfmbt2, Ttc28, Zdhhc14, AI414330, AU015536, BC020535, C76555, C80406, D9Ertd115e, LOC100043487, Rik1500009L16, Rik1810010M01, Rik2610019E17, Rik2610028H07, Rik5033413D22, Rik5330426P16, Rik6820431F20, Rik9130008F23, RikB130019D13, RikD130040H23, RikF630043A04
There was little overlap of the genes we identified as misregulated in diabetic placentae at mid-gestation with those reported by others for late gestation. In fact, no genes were found in common with results from a 151 gene custom array applied to ICR outbred mouse strain diabetic placentae at gestational day 17  or from cDNA arrays of E17.5 diabetic placentae from the C57BL/6, BALB/c, DBA/1 and CBA/J inbred mouse strains . In human term placentae from pregnancies affected by gestational diabetes, TGFß-induced transcript, thrombospondin 1 and heat shock protein A1A had increased and follistatin-like 3 decreased expression , while we saw decreased levels for all four genes. Fos-like antigen 2 and annexin A4 were elevated in an independent study of gestational diabetes mellitus placentae , while annexin A4 was decreased in our assays. Apart from strain or species differences, the most likely reason for the dearth of common discoveries is the early time point of our assays. Between E10.5 and E17 in the mouse, and mid-term to term in humans, respectively, the placenta undergoes fundamental changes in gene expression  in the maternally and embryo-derived compartments . Thus, even in normal placenta at E10.5, we would expect different sets of genes to be expressed compared with later-stage placenta as surveyed by others. Differential gene expression under conditions of maternal diabetes has also been reported for the yolk sac in rats at E12.5 . The only feature common to our results is increased enolase 1 expression. We therefore conclude that yolk sac and placenta undergo distinct responses to maternal diabetes. Taken together, our results show that, placental gene expression is uniquely altered in diabetic pregnancy in mice as early as E10.5.
Validation of microarray results by quantitative real-time PCR
Validation of microarray results by quantitative RT-PCR
Gene expression significantly changed in RT-PCR
Atonal homologue 8 (Drosophila)
Bone morphogenetic protein 2
Calcitonin-related polypeptide, beta
Eosinophil-associated, ribonuclease A family, member 11
Matrix metallopeptidase 9
Proprotein convertase subtilisin/kexin type 5
Pore forming protein-like
Prolactin family 5, subfamily a, member 1
Solute carrier family 6 (neurotransmitter transporter, serotonin), member 4
Ankyrin repeat domain 2 (stretch responsive muscle)
Achaete-scute complex homologue-like 2 (Drosophila)
Myelin protein zero-like 2
Ras association (ralgds/AF-6) domain family 4
Transforming growth factor, beta induced
Trophoblast specific protein beta
Gene expression not significantly changed in RT-PCR
A disintegrin-like and metallopeptidase with thrombospondin type 1 motif, 6
Colony stimulating factor 2 receptor, beta, low-affinity (granulocyte-macrophage)
Potassium channel, subfamily K, member 2
Secreted frizzled-related sequence protein 5
Solute carrier organic anion transporter family, member 2a1, prostaglandin transporter
MAD homologue 6 (Drosophila)
Cytochrome P450, family 1, subfamily a, polypeptide 1
Kelch-like 5 (Drosophila)
Thrombospondin, type I, domain containing 4
If cohorts of genes in the placenta are dysregulated through shared pathways of transcriptional regulation, one would expect occurrence of common TFBSs in the regulatory regions of these genes. In embryos exposed to hyperglycaemia , we previously found over-representation of binding motifs for forkhead box (FOX)O1, FOXO4 and nuclear factor erythroid 2-related factor 2 (NRF2), which are involved in the oxidative stress response, and for hypoxia-inducible factor 1 (HIF1), which orchestrates the response to hypoxia. Both oxidative stress and hypoxia have been reported for embryos exposed to diabetes . However, when we analysed promoter sequences (5 kb upstream of the transcription start) of diabetes-affected placental genes for the presence of TFBS, we found no enrichment of HIF1 sites. In addition, neither Hif1α expression nor expression of any of its known targets  were perturbed in placentae from diabetic dams, providing scant evidence for a hypoxia response in the placenta.
For FOXO1 and FOXO4, putative binding sites were present in 85 and 82 genes, respectively, of 131 placental genes for which results were returned (this accounts for 65% and 62.6%, respectively, of 131 genes, with 80 genes having both motifs). In addition, there was significant enrichment of these TFBS (−log10p = 6.7809 and − log10p = 5.55, respectively) compared with the entire genome. In placenta, oxidative stress has been associated with activation of nuclear factor κB (NFκB) [29, 30], and we found TFBS motifs for NFκB-enrichment (−log10p = 5.076), with 63 genes (48% of 131) carrying these sites. These results suggest that oxidative stress pathways could be involved in deregulated gene expression in the diabetic placenta.
Functional roles of genes deregulated in diabetic placentae
Conceivably, misregulated genes in diabetic placentae could be involved in altered placental function. This proposition is supported by mouse mutant phenotype information (MGI), which is available for 68 of the 158 genes. Of these 68 genes, seven have been shown to play a role in abnormal development of extra-embryonic tissues, uterus and/or the placenta (ESM Table 4 plus references cited there). Knockout of Incenp and Lims1 affects extra-embryonic tissues, and Bmp2 deficiency delays allantois development, while Star deletion is associated with abnormal uterus morphology. The labyrinth is disorganised in mutants that are heterozygous for a mutant Cdx2 allele on a Cdx4-deficient background. Animals with knockout of Socs3 and Ascl2 exhibited altered placental morphology in the labyrinth layer, with the latter gene also affecting trophoblast giant cells and spongiotrophoblasts. However, for many other diabetes-affected placental genes, functional annotation is not yet available. Interestingly, the number of genes we identified as altered in diabetic placenta is only a small fraction of the 458 genotypes listed in MGI with placental phenotypes. This strongly suggests that diabetes affects specific pathways in the placenta, such as cell adhesion, growth factor responses and nutrient transport. Collectively, our results are consistent with the notion that misregulation of gene expression compromises development and function of the placenta in diabetic mouse pregnancy.
Expression studies identified abnormalities in the maternal and embryonic compartments of the diabetic placenta
Particularly noteworthy are increased hybridisation signals for the Slc6a4 (serotonin transporter), Cyp1a1, calcitonin-related-peptide B and cholecystokinin genes in diabetic placenta (Fig. 3g–r). In all four cases, the number of cells expressing the respective transcript was increased. The identity of cells expressing Slc6a4 (Fig. 3g–i) adjacent to the junctional layer remains to be determined; a weak signal was also detected in the labyrinth. Cyp1a1 (Fig. 3j–l) was expressed in cells lining the decidual blood sinuses, consistent with a report that found the same localisation after administration of a toxic compound . Elevated Cyp1a1 expression and activity in placenta is a well-known indicator of exposure to xenogenic substances, most prominently linked to maternal cigarette smoking . In diabetic placentae, the number of Cyp1a1-expressing cells was increased (Fig. 3k). Although in human diabetic pregnancies, P450 enzyme activities were not elevated  in term placenta, our results from mid-gestation diabetic mice support the notion that elevated Cyp1a1 expression may have a physiological role in diabetic pregnancies, possibly in the removal of undesirable metabolites.
gene expression levels were also increased in diabetic placenta (Fig. 3n), this being again associated with an increase in the number of positive cells, which were dispersed in the decidual part of the placenta (Fig. 3o), and whose identity and physiological role remain to be clarified. Calcitonin-related peptides are known to be potent vasodilators in placenta , and increased blood flow to implantation sites has been documented in streptozotocin-induced diabetic rats . It is thus possible that increased Calcitonin-related peptide B levels could be involved in placental complications in diabetic pregnancies.
Robust induction by maternal diabetes was observed for the gene encoding cholecystokinin, a gastrointestinal peptide and neurotransmitter. In diabetic placentae, the signal for Cck mRNA was clearly elevated (Fig. 3q); again, this was associated with a larger number of Cck-expressing cells. The identity of these cells in the metrial region (Fig. 3r) is currently undetermined; a subset is CD45-positive (data not shown), indicating a haematopoietic origin.
Taken together, our studies demonstrate that maternal diabetes alters gene expression in embryonic and maternal compartments of the placenta, including several genes not previously implicated in diabetic pregnancies. In all cases where increased expression was present, this was linked to a greater number of cells expressing the respective gene, demonstrating that the cellular composition of the placenta is altered in diabetic pregnancy.
Reduced junctional zone and labyrinth size in diabetic placentae
Impaired spongiotrophoblast development has been reported for mutants deficient in the genes encoding the transcription factors achaete-scute complex homologue 2 (ASCL2), Hif1α, HOP homeobox or retinoid X receptor (RXR) α/β combined (ESM Table 4), as well as for transcriptional co-regulators Cbp/p300-interacting transactivator with Glu/Asp-rich carboxy-terminal domain 2 (CITED2) and alkylation repair B homologue 1 (ALKBH1), and for loss of imprinting in the Dnmt3L (also known as Dnmt3l) mutant. Reduced spongiotrophoblast growth was associated with expansion of trophoblast giant cells at mid-gestation  and reduction of giant and glycogen cells at E14.5/15.5 [39, 40]. Where diminished labyrinth size was reported for late pregnancy stages, it was interpreted as secondary to the spongiotrophoblast defect [41, 42]. We therefore consider it likely that the reduced labyrinth in diabetic placentae could be a consequence of impaired spongiotrophoblast growth. Where investigated, reduced spongiotrophoblast growth in mutants was associated with reduced Ascl2 expression at mid-gestation (ESM Table 4), which we also found in diabetic placentae. Collectively, this evidence supports the conclusion that the spongiotrophoblast deficiency in diabetic placentae is the result of aberrant gene expression at mid-gestation.
Conceivably, the thickness of the spongiotrophoblast layer could also be influenced by cell proliferation. In Senp2 mutants, reduction of the spongiotrophoblast layer was associated with delayed cell cycle progression . Mice transgenic for cathepsin 7 also display reduced spongiotrophoblast layer by E12.5, attributed to slower progression through late G2 or M phases of the cell cycle . The possibility that growth factor signalling may be affected by diabetes is suggested by reduced expression of Pappa2, which regulates IGF2 availability (data not shown).
Aberrant spongiotrophoblast differentiation in diabetic placentae Unexpectedly, Tpbpa signal was also detected within the labyrinth of diabetic placentae. At E15.5, numerous clusters of Tpbpa + cells were present (Fig. 5b, e), with lower incidence at E18.5 (Fig. 5h, k). This suggests that diabetic conditions cause ectopic differentiation of cells towards a spongiotrophoblast cell fate, ectopic proliferation of spongiotrophoblast-like cells in the labyrinthine area or aberrant migration into the labyrinth. In normal placentae, some clusters of Tpbpa + cells were observed in the labyrinth at E15.5, but very few at E18.5 (Fig. 5m, o), indicating that such cells are normally either eliminated or migrate out of the labyrinth. In the diabetic placentae, the ectopic Tpbpa + clusters were significantly larger (Fig. 5n) and persisted to E18.5. The mislocalisation of these cells outside the spongiotrophoblast layer might explain its reduced thickness. Interestingly, Yu et al.  reported moderate enlargement of the junctional layer in E17.5 diabetic mouse placentae, based on isolectinB4 staining, but they included glycogen-containing cells in their measurements. These cells, which display light cytoplasms on haematoxylin–eosin-stained sections , exhibit weak Tpbpa labelling (Fig. 5e, f) and were excluded from our analysis because they differ in histological appearance from the cells with the strongest Tpbpa signal at E15.5 (Fig. 5e, f) and at E18.5. The precise relationship of the ectopic clusters in the labyrinth to trophoblast progenitors, spongiotrophoblasts  and glycogen cells  remains to be defined. The persistence of large clusters of inappropriately differentiated cells may affect the functional integrity of the labyrinth and the vasculature, a possibility that deserves further investigation. Taken together, our results show that maternal diabetes impairs growth and alters spongiotrophoblast differentiation in the placenta, which is likely to affect signalling pathways important for placental growth and may have a detrimental effect on nutrient supply to the embryo.
Mislocalisation of spongiotrophoblast cells to the labyrinth has been reported in only a few instances to date. ‘Islands of SP-like cells’  in the labyrinth were detected with targeted disruption of the genes encoding haemoxygenase 1, prolyl hydroxylase domain 2 and protocadherin 12, and aberrant expression of the imprinted gene Phlda2 in mice (ESM Table 4). Effects on labyrinth size were not consistent, but in the latter, fetal growth was delayed, resulting in a 13% smaller size by E18.5. These results implicate ectopic spongiotrophoblast differentiation in intrauterine growth reduction and support the intriguing possibility that the reduced spongiotrophoblast layer and ectopic spongiotrophoblast localisation in diabetic placentae could be responsible for the reduced embryonic growth we previously observed  at later stages of mouse diabetic pregnancies.
Finally, although spongiotrophoblasts originate from extra-embryonic progenitors, their differentiation may be influenced by maternal factors. For example, the spongiotrophoblast layer was reduced in mutants with uterus-specific deletion of the Nr2f2 gene (encoding COUP-TFII), demonstrating a maternal effect on spongiotrophoblast growth . Given that we found altered cell composition in the decidual compartment of diabetic placentae, it will be interesting to investigate which role specific decidual cell types have in spongiotrophoblast differentiation and placenta growth. Collectively, our results show that maternal diabetes causes misregulation of distinct gene repertoires in embryo and placenta, bringing about developmental defects, intrauterine growth reduction and abnormal placental cell differentiation, respectively. Perturbed cellular interactions in the diabetic placenta and subsequent impairment of fetal growth could be the underlying cause for life-long elevated disease risk for mother and offspring of a pregnancy complicated by diabetes.
We are grateful for expert technical assistance by J. MacGowan, who performed the animal work and dissections. This work was funded in part through RO1-HD34706 to C. Kappen with a supplement to G. Pavlinkova, and through RO1-HD055528 to J. M. Salbaum with a supplement to N. Arbour Delahaye, as well as through the Peggy M. Pennington Cole Chair for Maternal Biology (C. Kappen). Histological investigations made use of the Cell Biology and Bioimaging Core Facility at Pennington, and PCR assays were performed in the Genomics Core Facility (both facilities are supported through COBRE P20RR021945 and CNRU P30DK072476). The Comparative Biology Core Facilities received support through G20RR024838.
Duality of interest
The authors declare that there is no duality of interest associated with this manuscript.