**3. Discussions**

By combining RNA-seq and ChIP-seq data from the mouse placenta, we identified *Plagl1* as an upregulated transcription factor that has its motif enriched within e9.5-specific enhancers. In mouse, gene ontology analysis showed that PLAGL1 was predicted to associate with genes involved in labyrinth layer development and fetal growth. *Plagl1* was expressed throughout the allantois and within vascular endothelial cells of the labyrinth layer. We also found that *Plagl1* has sex-specific gene expression differences between normal placentas and those from the GDM mouse model. Next, we performed an siRNA-mediated knockdown of *PLAGL1* in HTR-8/SVneo cells to determine if the gene could be important in human placental trophoblast cells. Ontology analysis of genes which decreased in expression upon *PLAGL1* knockdown showed enrichment of terms related to blood vessel remodeling. *PLAGL1* was further implicated in this role by a tube formation assay, where we observed a decrease in cord formation in *PLAGL1* knockdown cells.

In mouse, the importance of PLAGL1 in fetal growth has been established, as pups from global PLAGL1 knockout mice are smaller in size compared to wild type pups. The authors found a slight decrease in placental weight by gestational day 16.5. However, they did not report a difference in the histology of the placenta or its ability to transport glucose [65]. Our analysis identified a PLAGL1 motif in e9.5 placenta enhancers predicted to target *Ppar*γ and *Soc3.* Both genes are associated with 'regulation of cellular response to insulin stimulus' and 'abnormal placental labyrinth vasculature morphology' (Supplemental Figure S4a). These genes were also found to significantly decrease in HTR-8/SVneo cells after *PLAGL1* knockdown and are a part of terms like 'blood vessel development'. Interestingly, PLAGL1 has been associated with the upregulation of each of these genes in beta cell proliferation, and insulin secretion or signaling, and both genes have been suggested as targets of study to better understand the role of PLAGL1 in transient neonatal diabetes [66]. However, such a relationship

has not been established in placental vasculature although PPARγ null placentas, and SOCS3 null placentas both show defects in labyrinth formation and maternal blood sinuses [67,68]. Although we predict other targets of PLAGL1 that may be involved in blood vessel formation, future work, including PLAGL1 ChIP-seq, is necessary to confirm predicted, and identify novel, PLAGL1 binding sites. Given the association between PLAGL1 and angiogenic gene expression in the mouse placenta prior to major endovascular invasion, a role in blood vessel formation within the labyrinth layer is also possible. To test this, placenta-specific knockout of PLAGL1 would need to be generated, since *Plagl1* is highly expressed in other mouse embryonic tissues, including the liver and limb [69].

Our findings also revealed that, amongst male placentas, *Plagl1* is more highly expressed in a mouse model of GDM compared to controls. This adds to the evidence that *Plagl1* methylation and expression are sensitive to the maternal environment. *Plagl1* is hypomethylated and its expression is upregulated in mouse offspring generated by assisted reproductive technologies, and its methylation is associated with maternal folate concentrations, a Mediterranean diet, alcohol, and vitamin B2 consumption in women [70–73]. These *Plagl1* alterations are associated with birthweight and childhood obesity [70,74], and similarly, offspring of the mouse model of GDM utilized here have greater adiposity, and are more sensitive to metabolic disruption of their reproductive systems [75–77]. However, it is not clear whether placental *Plagl1* plays a functional role in these offspring outcomes, or is simply a marker for them; it is a putative metastable epiallele, meaning that environmental alterations in placental gene regulation may be maintained in offspring tissues and affect their adult functions [74]. *Plagl1* knockdown in the human trophoblast cells shows altered expression of *Oas1* and *Polr2g,* which are misexpressed in placentas from women with GDM, in the direction that would be predicted by PLAGL1 overexpression (downregulated and upregulated, respectively), suggesting a functional role for PLAGL1 in GDM-induced placental dysfunction [78]. Moreover, the findings here suggest that *Plagl1* misexpression could contribute to offspring outcomes in GDM by directly regulating placental function. For example, placental capillary density is increased, and branching decreased, in placentas from GDM pregnancies [44], consistent with the potential role of *Plagl1* in angiogenesis that we have uncovered. However, further experiments are needed to directly test whether placental angiogenesis is altered in this murine model of GDM, and whether *Plagl1* is responsible.

Notably, the effect of GDM on placental *Plagl1* expression in this model is sex-specific occurring in males only. In contrast, low birthweight is only associated with *Plagl1* methylation changes in female placentas [74]. Other genes have also been shown to be differentially expressed between male and female placentas associated with GDM, including lipolipase, which is involved in fatty acid transport and uptake [79]. Interestingly, GDM, like PLAGL1, is known to have sex-specific attributes. GDM is more common in pregnancies with a male fetus [80] and is also a risk factor for later type 2 diabetes of the fetus, with males developing it more than females [81]. GDM has also been shown to have different effects on thyroid hormone receptors within the placenta for male and female placentas [82], and affect glucose utilization differently between the sexes [83]. The mouse model of GDM used here affects the metabolism of both male and female offspring, but more dramatically impacts adipose tissue gene expression and substrate utilization in males [76].

The HTR-8/SVneo cells used for knockdown and subsequent experiments were of female origin [84]. Due to the sex-specific responses and expression of *Plagl1*, it would be advantageous to check the results in a male cell line to determine if PLAGL1 regulates the same genes in both genders. Much work remains in understanding the impact of sex on PLAGL1 functions and regulation.

Not only can PLAGL1 have sex-specific roles, but PLAGL1 is known to have different, or even opposite, functions depending on the type of cells being studied and on the co-factors that are expressed with it. For example, PLAGL1 induces activation of the PAC1 receptor promoter, unless co-transfected with ERA, which then leads it to repress promoter activity [37]. PLAGL1 expression has also been found to be increased in some tumor tissues [85,86], and decreased in others [87,88]. Since PLAGL1 can have different effects in different environments, it would be interesting to compare the effects of *PLAGL1* knockdown in specific subsets of trophoblast cells. Recently, human placental cells were

cultured and differentiated into extravillous and syncytiotrophoblast cells [89], which have varying degrees of *PLAGL1* expression. Knockdown and functional assays on multiple trophoblast subtypes will give us a greater insight into the diverse functions and cell-specific roles of PLAGL1. Other cell lines that could also be more appropriate for studying aspects of blood vessel formation, and that could help better understand the results we observe in mouse, include human placental vascular endothelial cells (HPVEC) or human umbilical vein endothelial cells (HUVEC).

In HTR-8/SVneo cells, we found that genes downregulated upon *PLAGL1* knockdown were strongly enriched for blood vessel development terms. Interestingly, both HTR-8/SVneo cells and primary first trimester trophoblast, but not third trimester trophoblast, are capable of cord formation [90], indicating that this property is characteristic of trophoblast cells capable of endovascular invasion. We found that by knocking down *PLAGL1,* cord formation of HTR-8/SVneo cells was hindered. However, to determine if PLAGL1 has a role in tube formation, other methods such as co-culture assays would need to be assessed, since several non-endothelial cells can also form tubes in the presence of Matrigel [91]. Genes upregulated upon *PLAGL1* knockdown were enriched for cell adhesion terms. Previous research has found that PLAGL1 targets genes involved in cellular adhesion and extracellular matrix composition [92,93]. We also observed that several immune genes were upregulated including interferons (*IRF6*, *IRF7*) [94,95], *TGFB3* [96], and *HLA-DQB1* [97]. However, the role of PLAGL1 in immunity is unknown. Further analysis could reveal a role of PLAGL1 in regulating an immune response in the placenta.

Although we performed experiments and bioinformatics analysis in both mouse placenta and a human trophoblast cell line that represent different aspects of trophoblast-endothelial interactions in placental development, we do observe similarity of PLAGL1 targets and predicted functions, as noted above. In mouse, however, we observed *Plagl1* expression in endothelial cells, whereas in human we investigated trophoblast cells. It is possible PLAGL1 has a role in both cell types in both species, and we could not capture this with the specific timepoint we investigated in mouse, and the specific cell line we used in human. We note, however that there are also similarities in the genes identified in our study and PLAGL1 target genes identified in other studies. For example, a previous study analyzed 15 imprinted genes associated with *Plagl1* and checked if they displayed significantly different expression in the embryonic liver (e18.5) of WT mice compared to the liver in *Plagl1* knockout mice [65]. Of these 15 imprinted genes, we found that *CDKN1C*, *DCN*, *GATM*, *GRB10*, *MEG3*, and *MEST* were upregulated in the HTR-8/SVneo *PLAGL1* knockdown cells, while others, including *SLC38A4*, *IGF2*, and *IGF2R* have downregulated family members in the HTR-8/SVneo cells. We found other known imprinted targets of PLAGL1 to be disrupted in the HTR-8/SVneo *PLAGL1* knockdown cells, such as *RASGRF1* [98]. *TP73* and *WT1*, which were also misregulated upon *PLAGL1* knockdown, were previously found to be highly methylated in prostate cancer, along with *PLAGL1* [99].

Although many roles of PLAGL1 have been well studied in different tissues and cells, including its ability to regulate migration [100] in neurons and proliferation [38] in several tissues, a role in the placenta has not been thoroughly described. We provide evidence that PLAGL1 alters angiogenic gene expression and function. Proper remodeling of maternal blood vessels and formation of fetal blood vessels are critical for efficient transport of nutrients and oxygen between the maternal and fetal blood supply. Errors in this process could lead to several complications. Therefore, PLAGL1 may be an interesting target of analysis when understanding the pathogenesis of pregnancy diseases.
