The transcriptional repressor Blimp1/PRDM1 regulates the maternal decidual response in mice

The transcriptional repressor Blimp1 controls cell fate decisions in the developing embryo and adult tissues. Here we describe Blimp1 expression and functional requirements within maternal uterine tissues during pregnancy. Expression is robustly up-regulated at early post-implantation stages in the primary decidual zone (PDZ) surrounding the embryo. Conditional inactivation results in defective formation of the PDZ barrier and abnormal trophectoderm invasion. RNA-Seq analysis demonstrates down-regulated expression of genes involved in cell adhesion and markers of decidualisation. In contrast, genes controlling immune responses including IFNγ are up-regulated. ChIP-Seq experiments identify candidate targets unique to the decidua as well as those shared across diverse cell types including a highly conserved peak at the Csf-1 gene promoter. Interestingly Blimp1 inactivation results in up-regulated Csf1 expression and macrophage recruitment into maternal decidual tissues. These results identify Blimp1 as a critical regulator of tissue remodelling and maternal tolerance during early stages of pregnancy. The transcriptional repressor Blimp1/PRDM1 regulates cell fate decisions in the developing embryo and adult tissues. Here the authors show that conditional inactivation within maternal uterine tissues results in a defective primary decidual zone barrier, increased expression of inflammatory cytokines IFN gamma and Csf1, and early embryonic lethality during pregnancy.

D ivergent patterns of gene expression underlie the establishment of cell identity. For the most part, this reflects the induction of gene expression by transcriptional activators. However, in some cases cell fate decisions, especially those made within an already committed lineage, may be due to selective silencing of an ongoing transcriptional programme that is associated with a developmental switch controlled by a cell-type specific repressive complex. Much has been learned over recent years about the constitutive epigenetic machinery responsible for gene silencing-including the activities of key enzymatic components such as G9a, Lsd-1 and Hdac family members. However, relatively little is known about cell type specific repressors guiding target site selection.
We have been studying the zinc finger SET domain protein Blimp1 encoded by the Prdm1 gene -originally cloned as a postinductive repressor of type I IFNβ gene expression 1 -and subsequently identified as the key transcription factor controlling terminal plasma cell differentiation 2 . Its functional role in the B cell lineage has been extensively characterised. Its ability to silence expression of well-described target genes including c-Myc, Pax5, Bcl6 and CIITA, and consequently cause termination of B cell identity in favour of terminally differentiated plasma cell functionality has been intensely investigated 3 . Similarly in the context of the early embryo, Blimp1 silences the default somatic programme allowing a small subset of primordial germ cell (PGC) progenitors to avoid responsiveness to BMP/Smad signals and become committed to acquire a germ cell fate 4,5 .
Loss of function Blimp1 mutant embryos arrest at around embryonic day (E) 10.5 due to defective placental morphogenesis 5,6 . Blimp1 expression is essential for specification of a distinct sub-set of trophoblast giant cells, the SpA-TGC, that migrate into the uterine maternal tissue to surround, invade and remodel the maternal blood vessels 6,7 . Interestingly, our recent sc-RNA-Seq analysis of specialized cell types at the fetal-maternal interface at mid-gestation stages identified a discrete subpopulation of maternal Blimp1+ cells co-expressing high levels of the decidual stromal marker Prl8a2 7 .
In the present study we perform immunostaining experiments to further investigate Blimp1 expression within the maternal uterine environment. Expression is robustly upregulated at early post-implantation stages in the primary decidual zone (PDZ) surrounding the embryo. To explore Blimp1 functional contributions we exploit the well-characterised progesterone receptor Cre (PR-Cre) strain 8 to selectively eliminate Blimp1 expression in the maternal uterine environment. The loss of function mutation compromises the decidualisation response and results in loss of PDZ barrier formation, ectopic trophoblast expansion, increased macrophage invasion and ultimately, embryonic lethality.

Results
Upregulated Blimp1 expression during implantation. In the virgin uterus, Blimp1 expression is restricted to a few scattered cells within the stroma (Fig. 1a). A discrete population of Blimp1+ stromal cells immediately adjacent to the uterine luminal epithelium (LE) was readily detectable at embryonic day (E3.5) of pregnancy, prior to embryo implantation. Coincident with embryo attachment within the uterine crypts 24 hours (hr) later we observe a marked increase in Blimp1 expression in the uterine LE immediately adjacent to the trophectoderm. Blimp1 expression is strongly upregulated during formation of the PDZ surrounding the embryo. Expression persists in the PDZ at E5.5 and E6.5. Immunostaining results were confirmed by Western blot analysis (Fig. 1b).

Blimp1 inactivation compromises maternal decidual response.
To investigate Blimp1 functional contributions, we exploited the PR-Cre deleter strain, proven to be a valuable tool for studying gene function in the uterine LE and stroma during pregnancy 8 . To generate females lacking Blimp1 function in progesterone responsive tissues, hereinafter referred to as Blimp1 mutants, we crossed PR-Cre males carrying a Blimp1 null allele 5 with females homozygous for the Blimp1 conditional allele 9 . Loss of Blimp1 expression in maternal decidual tissues was confirmed by immunostaining at E6.5 (Fig. 1c).
We observed normal numbers of decidual swellings in Blimp1 mutant females at E6.5. (Fig. 2a). The average number of decidua/ mouse ± standard error of the mean (SEM) was 9.88 ± 0.42 for wild types (n = 25) and 9.38 ± 0.38 for mutants (n = 24). These results strongly suggest that embryo attachment and implantation proceed normally. However, as judged by reduced levels of Chicago blue uptake (Fig. 2b) mutant deciduae are significantly smaller in comparison to wild type. Alkaline phosphatase staining, a marker of stromal decidualisation was significantly reduced (Fig. 2c, Supplementary Fig. 1a). As assessed by BrdU incorporation and Ki67 staining these differences cannot simply be explained due to reduced proliferative capacity (Supplementary Fig. 1b, 1c). Collectively, these results demonstrate that loss of Blimp1 function compromises the maternal decidual response.
Embryos normally become confined to the anti-mesometrial region of the decidua shortly after implantation. In contrast, Blimp1 mutant embryos were often located in more mesometrial regions of the uterine crypts (Fig. 2d, Supplementary Fig. 1d). In Blimp1 mutant females at E6.5, both the embryos and surrounding PDZ display morphological disturbances. Additionally, the cell density of the PDZ is significantly reduced (Fig. 2e).

Loss of PDZ barrier and abnormal trophectoderm invasion.
To identify the extra-embryonic ectoderm (ExE) and trophoblast cell populations at early post-implantation stages, next we examined expression of the Tbox transcription factor Eomesodermin (Eomes) (Fig. 3a). We observed extensive invasion of Eomes+ trophectoderm giant cells into the maternal stroma in Blimp1 mutants. Whole mount staining of dissected embryos at E5.5 likewise shows that they retain the Oct4+ epiblast and overlying Eomes+ visceral endoderm populations. However, the morphology of extra-embryonic tissues is highly disturbed. Thus, the ExE which normally provides important trophic signals to the underlying epiblast 10,11 is selectively reduced.
To better visualise the behaviour of the ExE and embryonic cell populations, we made use of a paternally inherited ubiquitously expressed Rosa26-membraneTomato knock-in allele 12 . Confocal imaging of thick sections through E5.5 implantation sites demonstrate extensive ectopic growth of the mTomato + -labelled embryonic cells into the Blimp1 mutant maternal decidual tissues. As judged by cell size and keratin-8 immuno-reactivity, these invasive cells mostly represent primary trophoblast giant cells (Fig. 3c). We failed to recover viable intact embryos beyond E6.5.
Formation of the avascular PDZ immediately surrounding the embryo is associated with robust expression of the tight junction protein ZO-1 (Fig. 3d). Consistent with decreased tight junction density, ZO-1 staining is markedly reduced around the mutant E5.5 implantation sites (Fig. 3d, e). Moreover, transmission electron microscopy (TEM) analysis revealed a highly disturbed cellular architecture at E6.5 (Fig. 3f). In contrast to the wild-type PDZ, characterised by densely packed cells and extensive tight junction formation, the mutant decidua cells are very loosely packed and there were only a few sporadic tight junctions. Interestingly, this cellular morphology closely resembles that of pre-decidual stromal cells 13 . Collectively, these experiments demonstrate that Blimp1 function is required in the maternal tissue to establish the densely packed PDZ that normally acts as a barrier to constrain TE invasion.
RNA-Seq identifies gene expression changes. To further characterise cellular defects in Blimp1 mutant decidua, we performed transcriptional profiling experiments. RNA-Seq analysis of wildtype and mutant-decidual tissues demonstrates gene expression changes detectable at E5.5 (Supplementary Data 1). More pronounced differences became evident by E6.5 (Fig. 4a, Supplementary Data 2). Hierarchical clustering shows that E6.5 Blimp1 mutant transcripts are more similar to E5.5 wild-type deciduae than they are to E6.5 wild-type deciduae ( Supplementary Fig. 2a). Based on statistical significance (DESeq2 FDR < 0.05), greater than 2-fold changes in expression, and an expression level confidence filter (FPKM ≥ 1 in all samples of either genotype), 703 genes were upregulated and 458 genes were found to be downregulated in E6.5 Blimp1 mutant decidua in comparison to wild type. Gene ontology (GO) analysis of upregulated genes shows a significant enrichment for genes associated with responses to external biotic stimulus, regulation of the immune response, regulation of the defence response, leukocyte activation, and cell chemotaxis (Fig. 4b    these genes are normally silenced in decidual tissues to maintain an immunologically privileged environment and protect the developing embryo 14 . In contrast, the 458 downregulated genes were enriched for genes involved in negative regulation of peptidase activity, cell adhesion, regulation of body fluid levels, female pregnancy and reproductive structure development (Fig. 4b, Supplementary Data 3). For example, expression of Prl family members selectively (Prl8a2) or predominantly (Prl3c1 and Prl6a1) expressed by decidual cells at early stages of pregnancy 15 is downregulated in Blimp1 mutants (Fig. 4c), strengthening the evidence suggesting decidualisation is compromised. The steroid hormones progesterone and oestrogen are known to play essential roles controlling gene expression during stromal cell decidualisation 16 . When we used gene set enrichment analysis to compare differentially expressed genes in E6.5 Blimp1 mutants with predicted progesterone and oestrogen responsive genes (based on proximal PR and Esr1 binding) we observed a highly significant correlation (FDR q-value > 0.0001) between downregulated genes in Blimp1 mutant decidua and progesterone receptor target genes ( Fig. 4d). Similar but less pronounced correlations were detectable for oestrogen receptor target genes (FDR q-value = 0.024). These results strongly suggest that pregnancy hormonal-driven decidualisation is impaired in Blimp1 mutants.
Consistent with results above that demonstrate increased trophoblast spreading and invasion in mutant decidua (Fig. 3), expression of parietal trophoblast giant cell (P-TGC)-restricted Prl3d1, 3d2 and 3d3 (and Prl7a1 at this stage of gestation) 15 is markedly increased. Additionally, the metallopeptidase (MMP)inhibitor Timp3, selectively expressed within the PDZ immediately surrounding the embryo 17 , is downregulated (Fig. 4c). Blimp1 mutant E6.5 decidua display increased Mmp11 (Supplementary Data 2). It seems likely that increased MMP activity observed in Blimp1 mutant decidua facilitates trophoblast spreading.
It is well known that Blimp1 mediated repression silences interferon signalling and expression of multiple interferonresponsive genes 1,18-20 . Interestingly, here we observe upregulated expression of multiple IFNγ-inducible genes (Fig. 4e, Supplementary Data 4) as well as increased levels of Ifnγ transcripts (Fig. 4f) in mutant decidua. Upregulated gene categories related to leukocyte activation and cell chemotaxis were also over-represented. Moreover, well-known macrophage marker genes including Itgam and Adgre1, encoding Cd11b and F4/80 respectively, were also found to be strongly upregulated ( Supplementary Fig. 2b).    decidualisation and identify candidate Blimp1 target genes, we exploited mice expressing an endogenous N-terminal EGFPtagged Blimp1 protein (hereafter referred to as NEG) in combination with a proven GFP antibody for ChIP-Seq analysis 21 . We identified 5808 high-confidence genome-wide Blimp1 binding sites (Fig. 5a, Supplementary Data 5). Of these, 2846 overlapped with those previously identified in E18.5 small intestine and multiple diverse cell types 21,22 . Among shared ChIP peaks we identified 935 candidate Blimp1 target genes containing at least one peak within 100 kb of the transcriptional start site (TSS). This group included genes involved in JAK/STAT, Wnt and EGFreceptor signalling pathways (Fig. 5b, Supplementary Data 6). On the other hand, 2962 ChIP-Seq peaks associated with 1971 genes (Blimp1 ChIP-Seq peak ± 100 kb TSS) were unique to decidua (Fig. 5a). These included an over-representation of genes involved in gonadotrophin-releasing hormone receptor, integrin, angiogenesis and TGF-β signalling pathways (Fig. 5b, Supplementary  Data 6).
Our comparison of Blimp1 ChIP-Seq peaks with differentially expressed transcripts identified approximately 26% of upregulated (n = 181) and 22% of downregulated genes (n = 102) in E6.5 Blimp1 mutant decidua as candidate Blimp1 targets. Several downregulated genes potentially contribute to the observed phenotypic abnormalities (Fig. 5c, Supplementary Data 7). For example, as mentioned above, expression of the metalloproteinase inhibitor Timp3 is markedly reduced in Blimp1 mutant decidua at both E6.5 and E5.5. Additionally, Nr2f2, encoding COUP-TF2, is downregulated (2.02-fold) in mutant decidua. Conditional Nr2f2 deletion in uterine tissues results in implantation and decidualisation defects 23 . Dsp, encoding desmoplakin, an essential cytoskeletal linker protein required for the assembly of functional desmosomes, is similarly downregulated (2.72-fold) 24 . Expression of desmin-containing intermediate filaments in endometrial stromal cells is thought to promote homophilic cell adhesion 25 . We speculate that decreased Dsp expression probably contributes to loose stromal cell contacts in Blimp1 mutant decidua.
In contrast, candidate Blimp1 targets with upregulated expression in E6.5 Blimp1 mutant decidua are enriched for genes associated with immune function and responses to other organisms (Fig. 5d, Supplementary Data 8). Previous experiments demonstrate that Blimp1 governs reprogramming of the postnatal intestinal epithelium and, in direct competition with the activator IRF-1, silences the MHC class I peptide-loading pathway to maintain tolerance during the suckling to weaning transition 22,26 . Consistent with its ability to repress expression of key components of the MHC class I peptide loading machinery in other tissues, we also found here that Blimp1 silences Psmb8, Psmb9 and Tap1 transcription. gene promoter in E6.5 mouse decidua (Fig. 6a). Comparison with previously published Blimp1 ChIP-Seq data sets likewise shows Blimp1 occupancy at the Csf1 promoter is widely detectable in other mouse tissue types 21,22,27 . It has been suggested that highly conserved Blimp1 binding sites are largely non-functional 21 . However, here our qPCR analysis confirmed upregulated expression of Csf1 in Blimp1 mutant decidual tissues (Fig. 6b). Blimp1 was previously shown to bind the human Csf1 gene promoter 28 . Moreover, CSF1 was among the most significantly upregulated genes in U266 cells following siRNA-mediated knockdown of Blimp1 28 confirming that occupancy of the human Csf1 promoter silences gene expression. Wild-type decidua is mostly devoid of macrophages 29,30 . As expected, here we found that macrophage marker gene expression is largely restricted to the surrounding myometrium (Fig. 6c). However, upregulated Csf1 expression in Blimp1 mutant decidua was associated with macrophage invasion deep into decidual tissues (Fig. 6c, Supplementary Fig. 4) and upregulated expression of macrophage marker genes ( Supplementary Fig. 2b). Moreover, F4/80, MHC Class II and CD74 immuno-reactive cells were present throughout Blimp1 mutant decidua (Fig. 6c, Supplementary Fig. 3a). Quantification of F4/80 and MHC Class II positive cells demonstrates a significant increase in the spread of macrophages (Supplementary Fig. 3b) as well as increased total macrophage numbers ( Supplementary Fig. 3c). These results strongly suggest that Blimp1 normally silences Csf1 expression to prevent macrophage invasion into the implantation site.
C-fms/Csf1r, the only known Csf1 cell surface receptor, broadly expressed on cells of the myeloid mononuclear phagocytic lineage, mediates cell activation and chemotaxis [31][32][33] . Additionally, c-fms expression has been reported on trophoblasts and a restricted population of decidualised stromal cells at implantation sites [34][35][36] . Immunostaining confirmed abundant c-fms expression within the PDZ immediately adjacent to the embryo at E6.5 in both wild-type and mutant decidua (Fig. 6d). Expression was also observed on invading macrophages in Blimp1 mutant decidua (Supplementary Fig. 5). However, as expected c-fms was un-detectable on trophoblasts at these early stages of post-implantation development 35 .

Discussion
The present experiments demonstrate that the zinc finger transcriptional repressor Blimp1/PRDM1 is a critical regulator of the maternal decidual response during early pregnancy. Interestingly, the gene regulatory networks controlling functionality of the maternal decidual cell type have been shown to closely resemble those during inflammatory and cellular stress responses 37,38 . Similarly, Blimp expression is induced downstream of the unfolded protein response by diverse stress stimuli 39 . Moreover, our transcriptional profiling experiments demonstrate that expression of serum amyloid protein SAA3, an acute phase response protein implicated as a pro-inflammatory mediator 40,41 , is robustly upregulated (272-fold) in mutant decidua. Collectively these observations strongly suggest that Blimp1 normally functions to silence maternal inflammatory responses during early postimplantation stages of pregnancy.
Here we describe Blimp1 expression patterns and essential functional activities within the maternal uterine environment during early stages of pregnancy. Blimp1 expression is robustly upregulated coincident with embryo attachment at E4.5 and maintained in the PDZ surrounding the embryo. To explore Blimp1 functional contributions within the maternal uterine tissues, we made use of the PR-Cre deleter strain 8 to selectively inactivate Blimp1 expression in the progesterone responsive decidualising stroma. Blimp1 functional activity is non-essential for uterine receptivity, blastocyst attachment and implantation. Moreover, decidualisation is correctly initiated at E4.5. However, the process of decidualisation arrests prematurely.
Our recent experiments demonstrate that Blimp1 silences expression of key components of the MHC class I peptide loading pathway and directly blocks IRF-1 occupancy at these sites to prevent premature activation of MHC class I surface expression and maintain neonatal tolerance in the developing intestine 22 . Similarly, here our ChIP-Seq analysis identified interferoninducible components of the MHC class I peptide loading pathway as Blimp1 target genes normally silenced in the maternal decidual tissues. It has been known for many years that Blimp1 silences IFNγ expression during Th1/Th2 T lymphocyte differentiation to shift the developmental program 42 . The present results demonstrate that Blimp1 plays a crucially important functional role dampening expression of the proinflammatory cytokine IFNγ and its downstream effectors in the context of the maternal uterine tissues at early post-implantation stages of pregnancy. Thus, IFNγ and many well-known IFNγ-responsive genes governing innate and adaptive immune responses were markedly upregulated in Blimp1 mutant decidua.
Early immunostaining and in situ hybridisation experiments suggested that uterine epithelial cells, natural killer (NK) cells, macrophages and placental trophoblasts at the maternal-fetal interface during mid-gestation all have the ability to produce IFNγ 43 . However, this signal was barely detectable at early postimplantation stages. Here we found in the absence of Blimp1 that upregulated IFNγ expression was associated with increased macrophage recruitment into the maternal decidual tissue. However, the numbers of uterine NK (uNK) cells remained constant ( Supplementary Fig. 2c). Increased MHC class II staining similarly suggests that Blimp1-mediated repression of IFNγ production helps to maintain an immunologically privileged environment. However, additional studies will be required to clarify the connection between Blimp1 conditional loss in maternal cells expressing the progesterone receptor and those responsible for production of the inflammatory cytokine IFNγ.
Our RNA-Seq analysis demonstrates that expression of numerous genes involved in cell adhesion that normally accompany the decidual response was significantly downregulated in the Blimp1 mutants. Formation of the so-termed PDZ, the avascular region of tightly adherent decidual cells, physically constrains the embryo and prevents invasion of the embryonic trophoblasts into the maternal environment. Thus, a unique feature of Blimp1 mutant maternal decidual cells is their inability to correctly form the PDZ barrier. Formation of the PDZ is accompanied by robust induction of proteinase inhibitors. For example, expression of the metalloproteinase inhibitor Timp3 is normally induced in the PDZ at E6.5 17,44 .
Here we found that Timp3 expression is markedly reduced in Blimp1 mutant decidua at both E5.5 and E6.5. Additionally, our ChIP-Seq analysis identified Timp3 as a candidate Blimp1 target gene suggesting that Blimp1 may also function as a transcriptional activator. Support for this idea also comes from recent experiments showing that Blimp1 functions as a transcriptional activator in the B-cell lineage 45 . Unlike other family members Timp3 is ECM bound and probably has only short-range localised activities 46 . Maternal Timp3 expressed immediately adjacent to the post-implantation embryo is thought to regulate trophoblast invasion into maternal tissues 44 . Mmps have also been implicated in trophoblast invasion 47 and Mmp11 is increased in Blimp1 mutant decidua. We speculate that imbalanced Timp3/Mmp11 activities potentially accounts for the increased trophoblast spreading and migration in Blimp1 mutant decidua.
Stromal cells within Blimp1 mutant decidua are loosely organised and have reduced tight junction-mediated barrier formation within the PDZ (Fig. 3f). Desmosomes in particular are implicated in tight junction formation in stromal cells during decidualisation 25 . Notably, expression of Dsp encoding desmoplakin, a key structural component required for functional desmosome formation, is reduced in Blimp1 mutants 24 . Dsp was also identified as a potential Blimp1 target gene. These findings suggest that activation of Dsp expression by Blimp1 promotes homophilic stromal cell interactions to maintain PDZ barrier function.
The activities of both BMP and Wnt ligands induced in the early uterine stroma surrounding the crypts are essential to promote the decidualisation process. Loss of either Bmp2 or the Bmp type 1 receptor Alk3 results in a complete failure of the decidualisation response 48,49 . Wnt signalling down-stream of BMP activity stimulates decidual morphogenesis. The embryo mis-location phenotype observed here is reminiscent of that described for Wnt5a mutants 50 . Interestingly, comparison of Blimp1 ChIP-Seq peaks identified here in maternal decidual cells with those previously reported for multiple cell types 21,22 shows that the common peaks include components of BMP and Wnt signalling pathways. The present results strongly suggest that Blimp1 directly silences expression of the BMP and Wnt antagonists Grem1 and Sfrp1, respectively. Conversely, Blimp1 may also positively regulate transcription of genes upstream of BMP signalling. For example, Nr2f2/COUP-TFII, identified as a candidate Blimp1 target gene by ChIP-Seq, was downregulated in E6.5 mutant decidua. Additionally, we found that many of the peaks unique to decidua include regulators of gonadotrophinreleasing hormone receptors and TGF-β signalling. It will be interesting to learn more about the hierarchy of Blimp1dependent transcriptional networks controlling decidualisation.
Growth factors produced by maternal decidual cells that regulate differentiation and migration of trophoblast cell populations is poorly understood. A unique feature of the Blimp1 mutant phenotype is the striking expansion of parietal trophoblast giant cells. Blimp1 expression in the developing placenta at mid-gestation stages has been previously described 6 . The present experiments strongly suggest that Fgf7 is a direct Blimp1 target gene. Upregulated Fgf expression could potentially lead to ectopic activation of the P-TGC subset. Addition of Fgf7 to cultured blastocysts results in increased P-TGC numbers and precocious differentiation 51 . Blimp1 probably dampens Fgf7 signalling to maintain the balance of TGC proliferation versus differentiation within the decidual stroma. Expression of Prl family members specific to P-TGCs is also markedly elevated. Increased numbers of P-TGCs probably reflect premature differentiation of the ExE trophoblast progenitor population. Likewise, expression of the Blimp1 target gene pregnancy associated plasma protein A (Pappa) is upregulated in the mutant. This metalloproteinase, a component of the IGF signalling pathway, regulates IGF bioavailability via cleavage of IGFBP4 52 . It seems likely that increased Pappa activity, leading to higher levels of IGF, promotes expansion of the primary TGC population 53 .
RNA-Seq experiments demonstrate that upregulated transcripts in Blimp1 mutants are greatly enriched for genes related to immune cell function and chemotaxis. ChIP-Seq analysis identified many of these as candidate Blimp1 targets. The peak located at the Csf1 TSS previously reported in both human and mouse data sets is shared across multiple cell types 21,28 . The cytokine Csf1 controls macrophage recruitment into the cycling mouse uterus 54 . High doses of Csf1 have been shown to induce embryo resorption 55 . Similarly, here we found that upregulated Csf1 expression is associated with increased macrophage invasion into the maternal decidual tissues, loss of the PDZ barrier function and ectopic expansion of fetal extra-embryonic trophoblast cell populations. Additional experiments will be necessary to dissect the contributions made by individual Blimp1 target genes that are collectively responsible for these tissue disturbances.

Methods
Animal care and use. Female C57BL/6 mice (6-10 weeks of age) were used as the wild-type strain. Prdm1 BEH/+ ;PR Cre/+ males were generated by crossing Prdm1 BEH/+ mice 5 with PR Cre/+ mice 8 . To induce Prdm1 gene deletion in the uterus, these males were crossed with female Prdm1 CA/CA mice 9 to generate Prdm1 BEH/CA ;PR Cre females (referred to as Blimp1 mutants throughout the text). In some experiments Blimp1 mutant females were mated with Rosa26 mT/mG males 12 for immunofluorescence analysis. For BrdU labelling experiments, 0.25 mg of 5-bromo-2deoxyuridine (BrdU, Sigma) per gram of body weight in phosphate buffered saline (PBS) was injected intraperitoneally 2 h prior to sacrifice. Genotyping was performed as described in the original reports. All animal experiments were performed in accordance with the UK Home Office regulations and approved by the University of Oxford Local Ethical Committee.
Chicago Blue visualisation and quantification. To visualise implantation sites, mice were injected intravenously with 0.1 ml of 1% (w/v) Chicago Sky Blue 6B (Sigma) in PBS under anaesthetic and sacrificed by CO 2 asphyxiation 10 min later. Individual decidua were removed from the uterus and shaken in 25 μl of formamide for 2 days at room temperature (RT), protected from light to extract the dye. After clearing by centrifugation (16,000 × g for 5 min), dye content in formamide extracts was quantitated spectrophotometrically using an ND1000 Nanodrop at 618 nm relative to a standard curve of Chicago Sky Blue dye dissolved in formamide.
Western blot analysis. RIPA lysates (30 μg per sample) were analysed by Western blot as previously described 56 . Antibodies used are listed in the Supplementary Table 1. Uncropped blots can be found in the Source data file.
Immunohistochemistry. For immunohistochemistry (IHC), virgin and pregnant (E3.5, E4.5, E5.5 and E6.5) uteri or individual decidua were fixed overnight in 4% paraformaldehyde (PFA) in PBS, dehydrated using an ethanol series, embedded in paraffin wax and sectioned (6 μm). Dewaxed sections were subjected to antigen retrieval by boiling for 1 h in Tris/EDTA (pH 9.0) and permeabilised for 10 min in 0.1% Triton X-100 (Sigma) in PBS. After blocking with 10% normal goat serum in PBS with 0.05% Tween-20 (Sigma) for 1 h at RT, sections were incubated with primary antibodies in blocking solution overnight at 4°C. Rat monoclonal antibodies underwent signal amplification with rabbit anti-rat secondary antibody (AI-4001, Vector Laboratories) for 45 min at RT. All samples were then subjected to peroxidase blocking for 20 min at RT and developed with Envison System-HRP for rabbit antibodies (K4011, DAKO) and Vector Red substrate (SK-4805, Vector Laboratories). Sections were lightly counterstained with haematoxylin, coverslipped and imaged. Haematoxylin and eosin staining was performed as per standard protocols. Antibodies used are listed in the Supplementary Table 1.
Alkaline phosphatase staining. Decidua (E5.5) from Blimp1 mutant and wildtype females were fixed overnight in 1% PFA in PBS at 4°C, washed in PBS, cryoprotected using a sucrose gradient and embedded in OCT. Cryosections (8 μm) were cut using a Leica CM3050 S Research Cryostat, post fixed in 0.2% glutaraldehyde in PBS for 15 min at RT, washed with PBS and stained with BCIP/NBT (11697471001, Sigma) according to the manufacturer's protocol.
Immunofluorescence. Freshly dissected whole decidua and isolated embryos were fixed in 1% PFA in PBS overnight at 4°C followed by washing in PBS. For vibratome processing decidua were embedded in 4% agarose in PBS, manually trimmed and 100 μm sections cut using a Leica VT1000 S vibrating blade microtome. For cryosectioning, decidua were cryoprotected using a sucrose gradient and embedded in OCT before being sectioned at 8 μm on a Leica CM3050 S Research Cryostat.
Decidual sections or whole embryos were washed three times in PBS containing 0.1% Triton X-100 (PBS-T). Samples were permeabilised in PBS containing 0.5% Triton X-100 for 20 min, washed in PBS-T and then blocked in 5% donkey serum plus 0.2% BSA in PBS-T for 1 h at RT. Samples were incubated overnight with primary antibodies in blocking solution at 4°C. After washing, samples were incubated with fluorophore-conjugated secondary antibodies in blocking solution for 2 h at RT. Following three subsequent washes in PBS-T, samples were washed in PBS-T containing 2 μg/ml DAPI for 15 minutes at RT, then washed three more times in PBS-T and mounted in Vectashield with DAPI (H-1200, Vector Laboratories). All samples were imaged on an Olympus Fluoview FV1000 confocal microscope and image data was processed and analysed using ImageJ. Antibodies are listed in Supplementary Table 1.
Transmission electron microscopy. For TEM analysis, E6.5 wild-type and Blimp1 mutant decidua were fixed with 4% PFA plus 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer at pH 7.2 for 4 h at RT then overnight at 4°C. After embedding in 4% agarose in PBS, samples were manually trimmed and thick sections (200 μm) cut using a Leica VT1000 S vibrating blade microtome. The thick sections were processed using a Leica AMW microwave tissue processor, followed by infiltration with TAAB TLV resin over 3 days, and polymerisation for 48 h at 60°C. Ultrathin (90 nm) sections were cut using a Diatome diamond knife on a Leica Ultracut7 ultramicrotome, post-stained with lead citrate for 5 min, and examined on a Tecnai 12 transmission electron microscope (FEI) equipped with a Gatan OneView CMOS camera.
RNA-Seq. Total RNA was extracted from E5.5 and E6.5 decidua from Blimp1 mutant and wild-type littermates (Prdm1 CA/+ ) using an RNeasy Mini kit (Qiagen Cat#74104) with on column DNase treatment according to the manufacturer's protocol. RNA was normalized to 800 ng per sample followed by depletion of cytoplasmic and mitochondrial ribosomal RNA sequences (Ribo-Zero Gold rRNA Removal Kit (H/M/R), Cat: #MRZG12324). Subsequent library preparation was performed using the Illumina TruSeq Stranded Total RNA Library Prep kit (H/M/ R) (Cat: #20020597), followed by sequencing on an Illumina HiSeq4000 (75 bp paired end).
RNA-Seq data analysis. Paired-end sequencing reads (75 bp) from mouse decidual samples were mapped to the mm10 mouse genome using RNA-STAR in Galaxy (https://usegalaxy.org). Aligned BAM files were then analysed using the RNA-Seq quantitation pipeline in SeqMonk (V1. 45.4). Differentially expressed genes were identified using DeSeq2 with a FDR cutoff of 0.05, >2-fold change in expression and FPKM of ≥1 (in all samples within at least one group). Data was visualised using Seqmonk, PCA, MA plot and datastore tree (hierarchal clustering) functions. GO analysis was performed using WebGestalt (www.webgestalt.org) 57 . Gene set enrichment analysis (GSEA) was performed using all RNA-Seq genes ranked by DESeq2 FDR from most significantly upregulated to most significantly downregulated and compared with genes displaying PR binding 1 hour following P4 treatment 58 (GSE40663) or Esr1 binding 1 h following E2 treatment 59 (GSE36455) in uterine tissues. PR and Esr1 ChIP-Seq regions were converted from mm9 to mm10 using LiftOver 60 and gene/peak associations identified using GREAT 61 based the single nearest gene option ± 10 kb of gene TSSs. IFNγ responsive genes (GO:0034341, filter: Mus musculus) were identified using AMIGO2 (http://amigo.geneontology.org/amigo) 62 .
ChIP-Seq analysis. To identify genome-wide Blimp1 binding sites, triplicate pools (n = 6-8) of freshly dissected decidua (E6.5) from homozygous NEG mice expressing an endogenous N-terminal EGFP-tagged Blimp1 protein 21 were minced using a razor blade in 50 μl of culture media (RPMI-1640, 10% FCS, 0.0004% βmercaptoethanol, 1x Pen/Strep, 2 mM glutamine) and cross-linked in 1% formaldehyde in culture media for 20 min at RT. Pools of decidua from wild-type mice were processed in parallel as a negative control. After quenching in 0.125 M glycine for 5 min, samples were washed in PBS and processed for ChIP using 6 μg of rabbit anti-GFP antibody (Ab290, Abcam). Triplicate test (GFP ChIP of NEG) and control (GFP ChIP of wild type) decidual samples, and their respective input samples were analysed. Sequence reads were mapped to the mm10 genome using bwa-MEM (0.7.15-r1140). After removing PCR duplicate pairs using SAMTools-0.1.19 63 , peak calling was performed with MACS2 64 using default parameters to call areas of enrichment in ChIP samples over input. High-confidence peaks were identified by intersecting peak areas in the triplicate GFP ChIP of NEG samples. Non-specific peaks were removed by subtracting overlapping peaks called in triplicate GFP ChIP of wild-type samples.
For comparison with NEG Blimp1 ChIP-Seq data, mouse embryonic small intestine Blimp1 ChIP-Seq data 22 (GSE66069) was converted from mm9 to mm10 using LiftOver 60 . Blimp1 ChIP peaks in multiple other NEG mouse tissues were previously reported 21 . Gene/peak associations were identified using GREAT 61 based the single nearest gene option ± 100 kb of gene TSSs.
QPCR. RNA (1 μg) was reverse transcribed to cDNA using Superscript III First Strand Synthesis System (Life Technologies, Cat#18080-051) and oligo dT. QPCR was performed with QuantiTect SYBR Green PCR mix (Qiagen) using 50 ng of cDNA per reaction. Relative gene expression was calculated using the ddCT method using Hprt as the housekeeping gene. QPCR primer sequences are listed in Supplementary Table 2.
Statistical analysis. Experiments were repeated at least three times. Exact n is stated for every experiment in figure legends. GraphPad Prism 8.0 software was used for statistical analysis. Student's unpaired two tailed t-tests were used for statistical analyses. P < 0.05 was considered statistically significant. Sample sizes were selected based on current and previous experiments and no statistical method was applied to predetermine sample size. Experiments were not randomized and investigators were not blinded during experimentation or assessment.

Data availability
RNA-Seq and ChIP-Seq data have been deposited in the NCBI GEO database under accession code: GSE141613. All other relevant data is available from the authors.
Publicly available source data in Figs. 4 and 5 were obtained from NCBI GEO under accession codes: GSE40663, GSE36455, GSE66069, and GSE91038.