In the present study, a detailed temporal and spatial placental expression map was generated for all murine PRL/PL family members from E7.5 to E18.5 of gestation in three genetic strains. This detailed analysis uncovered several new markers for some trophoblast cell types that will be useful for future analysis of placental structure in mutant mice with placental phenotypes. Moreover, several important conclusions about regulation of the locus are apparent. First, no two family members have the same expression pattern even when complete temporal and spatial data are examined. Second, most genes are expressed in multiple trophoblast cell subtypes though none were detected in the chorion, where trophoblast stem cells reside, or in syncytiotrophoblast of the labyrinth layer. Third, bioinformatic comparisons of upstream regulatory regions identified predicted transcription factor binding site modules that are shared by genes expressed in the same trophoblast subtype. Fourth, further diversification of gene products from the PRL/PL locus occurs through alternative splice isoforms for several genes.
In reviewing the summary data, it was striking that so many genes within the PRL/PL family are expressed in the ectoplacental cone and in the adjacent parietal TGCs early in gestation. It was already reported that TGCs at the periphery of the ectoplacental cone express Prl3d and Prl2c as early as E6.5  and Prl7a1, Prl4a1 and Prl6a1 are expressed in the developing placenta as early as E8.0 [32, 54]. Interestingly, PRL/PL family members found at the periphery of the ectoplacental cone are not uniformly expressed in all TGCs and likely demarcate distinct TGC subpopulations and/or stages of TGC differentiation. Expression of genes in trophoblast cells at the center of the ectoplacental cone was also diverse. Of note, Prl3b1 is expressed in a small subset of cells within the center of the ectoplacental cone at E8.5, possibly in secondary TGC precursors or spongiotrophoblast precursors. Prl3b1 expression, which is broad later in gestation, has not been described before E9 at which time Prl3d expression in TGCs begins to decline . Prl5a1 expression is also interesting as it is biphasic in its expression, expressed in both TGCs and cells in the centre of the ectoplacental cone early in gestation and not again until late in gestation where it is expressed in spongiotrophoblast cells. Further studies are required to understand the heterogeneous ectoplacental cone and parietal TGC cell populations.
The current studies have also provided several markers that are restricted to either the spongiotrophoblast or glycogen cell populations. The junctional zone of the placenta contains both spongiotrophoblast cells and glycogen trophoblast cells. While many genes are known to be expressed within this layer , there had been very few genes definitively characterized as markers of either spongiotrophoblast or glycogen trophoblast cells, although members of the PRL/PL family (Prl2a1, Prl7b1 in mouse and Prl4a1, Prl2a1, Prl7b1 and Prl5a1 in rat) have been specifically localized to invasive/migratory trophoblast within the decidua , suggesting 'migratory' or glycogen trophoblast specifiCity. Pcdh12 [44, 45] and connexin 31  have been recently identified as glycogen trophoblast cell-specific markers, but no spongiotrophoblast-specific markers have been reported. Double in situ hybridization with PRL/PL family members and Pcdh12 showed that expression of Prl3b1, Prl2c, Prl8a6, Prl8a8, Prl8a1, Prl7a2, Prl3a1, Prl2b1, and Prl5a1 is restricted to spongiotrophoblast cells. Prl8a8 appears to be the best spongiotrophoblast-specific marker as it is not expressed in any other trophoblast cell subtypes throughout gestation and appears to be broadly expressed within the entire population of spongiotrophoblast. In contrast, Prl6a1, Prl2a1, Prl7b1 and Prl7c1 all have expression patterns overlapping with Pcdh12 expression (or complementary to Prl7a2 as shown) indicating they are specifically expressed in glycogen trophoblast cells within this layer. Markers that allow the distinction of glycogen trophoblast and spongiotrophoblast cells will be very useful in dissecting the pathology of mouse mutants with placental phenotypes, since the most commonly used marker, Tpbpa, is expressed in both . It is also unclear whether glycogen trophoblast cells differentiate directly from spongiotrophoblast later in gestation  or whether they represent an independent trophoblast population earlier within the ectoplacental cone . Examining both spongiotrophoblast and glycogen trophoblast in mutant mice with defects in the junctional zone will give insights into this question.
Gaining a better understanding of the regulation of gene expression from the PRL/PL locus was also a major aim of our study. Placental-specific expression from the expanded human growth hormone locus is regulated by a locus control region 15–32 kb upstream of the gene cluster [26, 27]. The evidence to date has not generally supported the notion that the rodent PRL/PL genes are similarly regulated; small defined upstream regions isolated from the promoters of several PRL/PL family members, including rat Prl3b1 (Pl2) [57–59], mouse Prl3d (Pl1) [60–62] and Prl2c (Plf) , rat Prl4a1 (Plp-A) , rat Prl8a2 (d/tPrp) , rat Prl8a3 (Plp-Cv) , independently drive trophoblast-specific expression in vitro. Furthermore, genes with similar placental expression patterns do not appear to be grouped together within the PRL/PL cluster [2, 4]. Nevertheless, regulatory elements for Prl3b1  and Pl3d and Prl2c (J.C. Cross, unpublished data) that are sufficient in vitro fail to drive trophoblast-specific expression when tested in vivo. This indicates a requirement for additional elements for these genes at least making it difficult to rule out completely the involvement of locus control regions or additional enhancers. In addition, complete temporal and spatial expression data for each PRL/PL family member have not been compiled. In the current study we compiled expression data for each family member over the whole time course of placentation with high resolution, reporting expression patterns in all the characterized trophoblast subtypes, in an effort to uncover unappreciated associations between locus position and expression patterns. The identification of regulatory elements driving trophoblast subtype-specific gene expression would be of particular use for future studies of trophoblast differentiation and placental development. It is now clear from our data set however that each PRL/PL family member has a truly unique expression pattern with no correlation to locus structure, consistent with previous reports [2, 4] and diminishing further the likelihood of control regions driving subtype-specific trophoblast expression. As such, subtype-specific expression patterns are likely driven by elements contained within the local upstream promoters of individual genes, although this does not preclude the possibility of a locus control region regulating placental-specific for the entire cluster, regulating trophoblast-specific transcriptional access to the whole of the region.
We therefore sought another way to investigate trophoblast subtype-specific regulatory elements. Previous studies of PRL/PL gene regulation have rarely investigated what, if any, subtype-specific expression is conveyed by the identified elements. Two notable exceptions are an enhancer element from the rat Prl8a3 promoter shown to drive expression preferentially in spongiotrophoblast cells rather than TGCs  and the demonstration that the transcription factor Gata2 is essential to restrict Prl4a1 expression to secondary TGCs within the TGC population . We used the bioinformatics software program Toucan2  to identify groups, or modules, of putative transcription factor binding sites present in multiple promoter sequences from genes co-expressed in a particular trophoblast subtype, rather than a single promoter in isolation. This approach effectively narrows the focus to elements involved in gene expression within a particular cell population, rather than trophoblast specifiCity in general. Our comprehensive expression data set allowed us for the first time to try such an in silico approach and we compared the promoters of genes co-expressed in the poorly characterized glycogen trophoblast and sinusoidal TGC populations as examples, identifying several modules of putative transcription factor binding sites common to the promoters of each group. The identification of these conserved modules serves not only to bolster the notion of local versus distant regulation of PRL/PL family members, but provides a valuable in silico resource to identify candidate elements for further experimental studies. Interestingly, putative Gata3 sites were identified in all the modules, a factor of known importance for trophoblast-specific expression of Prl3d and Prl2c [60–62]. Also, putative AP-1 sites were present in both modules contained in the promoters of genes co-expressed in glycogen trophoblast cells, motifs important for Plr3d and Plr3b1 expression [59, 62]. The presence of motifs previously been shown to play roles in regulating trophoblast-specific PRL/PL members within the modules imparts a higher degree of confidence to these predicted elements, making them strong candidates for further experimental studies.
We conducted our expression analysis in three different mouse lines, not only to increase the confidence of our data set but also to investigate whether any PRL/PL family members are expressed differently between commonly used strains. We chose the outbred stock CD-1 and two inbred strains C57/B6 and 129svj, all commonly used lines. Average litter sizes are known to vary among different mouse stocks and strains (129svj - average 4.5 pups/litter , C57/B6 - average 6.2 pups/litter , CD-1 - average 11 pups/litter (Charles River) and several PRL/PL family members have been shown to regulate reproductive adaptations to physiological stresses such as hypoxia. Differences in PRL/PL gene expression between lines may therefore be related to differences in litter size and the accompanying changes in physiological adaptation . Interestingly, only a few slight differences in the timing of PRL/PL family expression or the breadth of expression within a certain trophoblast subtype were observed.
Gene duplication followed by the acquisition of novel gene function offers a way for species to adapt to changing environmental challenges or to capitalize on newly available niches. Interestingly, members of the murine PRL/PL family have undergone further diversification by adopting splice variants. Evidence for positive selection within large families of amplified genes, particularly those associated with reproduction, has been accumulating [70–74]. One theory suggests that the gene duplication event itself is positively selected for to allow the amplification of genes that are somewhat pre-adapted to meet a particular environmental challenge or biochemical niche, so that divergence and acquisition of new functionality may follow . It is tempting to visualize the evolution of the PRL/PL genes in rodents and ruminants as a result of the driving environmental challenge of reproductive fitness. Numerous other gene families have been reported in the mouse genome that are also associated with reproduction such as the placentally expressed cathepsins [76–78], Rhox transcription factors  and pregnancy-specific glycoproteins [80, 81], although their specific roles in reproduction remain largely unknown. The functions of the individual PRL/PL family members are just beginning to be revealed through knockout mouse studies. The diverse patterns of expression and evidence of splice variants indicates that the family is rapidly evolving and suggests that the different genes have distinct functions. The majority of PRL/PL genes are expressed later in gestation when it is likely they are acting as hormones to affect feto-maternal adaptations to pregnancy, but it is also apparent from this study that approximately half of the PRL/PL family members are expressed early in the ectoplacental cone and may be involved in decisions affecting cell fate. Whether the PRL/PL locus evolved as a result of positive selection, or some other adaptive force, it is now clear that PRL/PL gene amplification serves as an excellent model for studying the process of genetic diversification.