Fetal access to the maternal circulation is dependent on the type of placentation, particularly the interhaemal barrier separating maternal and fetal circulations. In human placenta, this comprises only fetal tissues as the villous trophoblast faces an intervillous space filled with maternal blood. This is one kind of haemochorial placenta. In a more common type, found in many rodents, the maternal blood flows in trophoblast-lined blood channels. In endotheliochorial placentas, trophoblast reaches the maternal capillaries, and placental hormones need to cross the capillary endothelium. Epitheliochorial placentas appear to offer a greater challenge since several layers of fetal and maternal tissue separate the two blood streams. However, as described below, access to maternal tissues can be gained by trophoblast invasion, as in equids, or through fusion of trophoblasts with uterine epithelium, as in ruminants.

The trophoblast and other placental tissues are fetal in origin (4) and express paternal genes. Therefore, trophoblast invasion of the uterine wall challenges the maternal immune system. Since several placental hormones are thought to modulate immune responses, some authors tie their evolution to the degree of invasiveness (5, 6). It should be remembered, however, that the Grosser classification defines the tissue layers of the interhaemal barrier and is not an index of invasiveness (7, 8).

The hormones under discussion are products of the definitive, chorioallantoic placenta. There are other fetal membranes and they vary across mammals (9). In rodents and some other orders, there is a yolk sac placenta that persists to term. It has an epithelium, endodermal in origin, that faces the uterine cavity. This serves mainly for uptake of maternal secretions and antibodies. Whilst the yolk sac does synthesize hormones and hormone-binding proteins, such as transthyretin, these are secreted across the basolateral surface towards the fetal circulation (10).

Nineteen orders of eutherian mammals are currently recognized (11). Based on genomics, they can be assigned to four lineages or superorders (11). Best characterized from the perspective of placental endocrinology is Euarchontoglires, which includes rodents and primates. For Laurasiatheria, a fair amount is known about placental hormones in domesticated species within the orders of even-toed and odd-toed ungulates (Artiodactyla and Perissodactyla). In contrast little is known about placental hormones in bats although Chiroptera is the second most speciose order of mammal (12). The two other superorders are Afrotheria, which includes elephants and tenrecs, and Xenartha, which comprises sloths, anteaters and armadillos. There are few observations on placental hormones in these mammals although some studies have been made on elephants (13). A guide to taxonomic terms used in this review is given in Table 1.

In ruminants, the signal for pregnancy recognition is interferon tau, which is secreted by the trophectoderm of the elongated blastocyst during its long sojourn in the uterine lumen. The IFNT gene arose in the lineage of pecoran ruminants through duplication of the IFNW gene (20). In the process, IFNT lost the viral control elements of the promotor region and acquired sequences responsible for trophectoderm-specific expression (21, 22). Interferon tau binds to receptors in the endometrium and acts to suppress secretion of PGF_2α thus preventing luteolysis (23, 24). It is also detectable in uterine venous blood and may act on extrauterine tissues (25). Recent work has focussed on the direct effects of interferon tau on the transcriptome of the corpus luteum. Both the luteinized large cells and the luteal endothelial cells respond to interferon tau and the net effect is downregulation of luteolytic factors, promotion of cell survival and vascular stability (26). IFNT is absent in non-ruminant species of Artiodactyla, such as pig (Sus scrofa), hippopotamus (Hippopotamus amphibius) and llama (Lama glama) (27). The pregnancy recognition signal in pigs is estrogen, primarily estradiol-17β, secreted by the trophectoderm of the filamentous blastocyst (28).

In the horse (Equus caballus), the developing blastocyst is enclosed in a glycoprotein capsule that moves about the uterus. Blastocyst motility is essential to maintenance of pregnancy. A hitherto unidentified factor inhibits release of PGF_2α from the uterus (29). A recent overview concluded that maternal recognition of pregnancy in the horse involves a combination of chemical and mechanical signalling through multiple pathways (30). After implantation, between days 36 and 38 of gestation, trophoblast migrates from the chorionic girdle of the developing placenta into the endometrium and there forms the endometrial cups, which secrete equine chorionic gonadotrophin (eCG) (31, 32). Unlike in primates, a single gene codes for the β-subunit of pituitary LH and placental CG. It evolved in the equid lineage through acquisition of an extended carboxy-terminal peptide that is key to placental expression (33, 34). Placental expression occurs in the donkey (Equus asinus) and Plains zebra (Equus quagga) (34, 35), but not in other perissodactyls such as the Central American tapir (Tapirus bairdii) and various species of rhinoceros (36). Together with pituitary FSH, eCG stimulates development of accessory corpora lutea, which support pregnancy until about day 70, after which pregnancy is maintained through placental secretion of progestins such as dihydroprogesterone (DHP). For a fuller consideration of the endocrinology of pregnancy in the mare, including estrogen production by the fetoplacental unit, the reader is referred to an excellent review by Conley (37).

There is a superficial resemblance between horse and human in that pregnancy maintenance depends initially on hCG, which is secreted by the syncytiotrophoblast, and then on placental secretion of progesterone. Although hCG rescues steroid synthesis by the corpus luteum (38), its role may be short-lived (39). There is no chorionic gonadotrophin in lemurs, lorises or tarsiers. Duplication of the pituitary LH β-subunit gene, which presaged evolution of chorionic gonadotrophin, occurred in the lineage of anthropoid primates (40). Several copies of the CG β-subunit gene are found in anthropoid primates, but many lack placental expression (41). In human gestation, the plasma concentration of hCG rises rapidly 4 weeks after implantation and peaks at 8 to 10 weeks (42). Subsequently, pregnancy is maintained through placental secretion of progesterone. An enduring hypothesis is that human parturition is triggered by progesterone withdrawal (43). The current view favours a functional progesterone withdrawal resulting from an increase in the ratio between the two progesterone receptors (PR-A/PR-B ratio) and mediated by prostaglandins, although the full picture is considerably more complex (44, 45).

It should be remarked that hCG has actions other than maintenance of the corpus luteum (3). For example, hCG induces proliferation of the uterine natural killer cells that play a key role in maternal-fetal interactions within the placental bed (46).

In mouse (Mus musculus), rat (Rattus norvegicus), and golden hamster (Mesocricetus auratus), rescue of the corpus luteum is through pituitary secretion of prolactin (PRL), which requires stimulation of the uterine cervix during coitus. As pregnancy proceeds, PRL is supplemented by placental lactogens (PLs). PRL, PL-I and PL-II act by silencing expression of 20α-hydroxysteroid dehydrogenase, which otherwise would catabolize progesterone (47). The trophoblast giant cells secrete PL-I in mid-gestation and PL-II during the last half of gestation (48). In rodents, placental lactogens arose through duplication of the Prl gene and in mouse they are encoded by Prl3d1 (PL-I) and Prl3b1 (PL-II). A variant form of PL-I in the rat is encoded by Prl3d4. Further functions of PLs and prolactin-like proteins in rat and mouse are discussed in a later section.

Evolution of placental lactogens in rodents has yet to be explored in depth. Pl1, Pl2, and variants thereof, have been documented in cricetid rodents: the golden hamster (49) and two species of deer mouse (Peromyscus maniculatus and P. polionotus) (50). Thus, based on current phylogeny, Pl1 and Pl2 must have been present in the common ancestor of Muridae and Cricetidae, which together account for 94% of muroid diversity (51). Without further data, it is not possible to pinpoint when and where placental lactogens emerged in this lineage of rodents.

The guinea pig and other hystricomorph rodents have a long gestation and give birth to well-developed young. They differ in this respect from mouse and rat. Maintenance of the corpora lutea in the first weeks of guinea pig pregnancy has been ascribed to a chorionic gonadotrophin [evidence summarized in (52)]. In addition, spongiotrophoblast from the interlobular areas of the placenta secretes prolactin-like proteins (53). The ovaries are not required after day 21 of gestation when pregnancy maintenance depends on placental secretion of progesterone (54). As in women, there is no decline in plasma progesterone prior to birth suggesting guinea pig as a promising model for research on parturition (44).

The prolactin gene went through a period of rapid evolution in the African savannah elephant (Loxodonta africana) and rock hyrax (Procavia capensis), but without gene duplication (55). Nonetheless, placental expression of PRL has been suggested for both African and Asian elephants (Elephas maximus) based on immunostaining with an antibody raised against human prolactin (56, 57). Gestation in elephants is maintained by large accessory corpora lutea and the luteotrophic factor may well be PRL derived from the placenta (58).

Duplication and placental expression of the growth hormone gene is a distinctive feature of anthropoid primates. A cluster of five genes is found on human chromosome 17. One codes for pituitary growth hormone (hGH-N), another for placental growth hormone variant (hGH-V) and two for placental lactogens, also known as chorionic somatomammotropic hormones (CSH1/hCS-A, CSH2/hCS-B) (62). The appellation placental lactogen is supported by their much greater affinity for the PRL receptor than the GH receptor (63).

As already noted, PRL, PL-1 and PL-2 maintain corpus luteum function, which is a prerequisite for mammary development. In addition, they act through the PRL receptor (coded by Prlr) to promote lobuloalveolar growth, differentiation, and milk protein gene expression (87, 88). Until quite recently, the metabolic effects of rodent PLs were unclear. However, a comparison of mouse mutants lacking either Prl or Prlr indicated that PLs rather than prolactin were essential for maintenance of adequate glucose levels during gestation (89).

A large cluster of PRL-like genes occurs in rat and mouse (90, 91). Many of these are orphan ligands or act through non-classical pathways rather than through the PRL receptor (92). They include genes coding for proliferin (Prl2c2) and proliferin-related protein (Prl7d1). Proliferin promotes angiogenesis and is expressed by the trophoblast giant cells during development of the placental labyrinth (93). In contrast, proliferin-related protein, which is expressed by cytotrophoblasts of the junctional zone, is anti-angiogenic. At midgestation there is decrease in proliferin and increase in proliferin-related protein that may restrict further vascularization (93). Whilst these actions are paracrine, both proteins are secreted to the maternal circulation.

Several of the PRL-like proteins (PRL-A, -B, and -C) are expressed by the trophoblasts that invade the mesometrial triangle in the last week of gestation. This coincides with the disappearance of uterine natural killer (uNK) cells and PLP-A has been shown to bind to uNK cells and suppress their synthesis of interferon-gamma (94).

Rodents do not have a placental growth hormone. However, in the mouse, pituitary secretion of GH rises at mid-gestation coincident with development of the chorioallantoic placenta (95). The increase has been attributed to a placental factor. Subsequent work ruled out placentally derived acyl-ghrelin (96), but placental secretion of GH- releasing hormone (GHRH) remains a possibility (95). Finally, it is noteworthy that Prl is expressed by uterine decidua in murid rodents (76), as in anthropoid primates, although this is another incidence of convergent evolution (76).

Most research on placental hormones of artiodactyls has focussed on domesticated species such as cattle (Bos taurus), sheep, and goat (Capra hircus). These and other ruminants, including the basal tragulids or mouse deer (Tragulus javanicus and T. napu) (97), feature binucleate trophoblast cells (BNCs) that synthesize prolactins and pregnancy-associated glycoproteins (PAGs). As first shown in cattle (98), BNCs can fuse with uterine epithelial cells to form a short-lived trinucleate cell that delivers the hormones to maternal tissues. Trinucleate cells have been demonstrated in species from three further families, including in white-tailed deer (Odocoileus virginianus), Northern giraffe (Giraffa camelopardalis) and pronghorn (Antilocapra americana) (99, 100). In sheep, goat, and blue wildebeest (Connochaetes taurinus), there is a syncytium that Wooding considers to be a hybrid tissue maintained by continual fusion with BNCs (100, 101). This view has been challenged recently, and it is suggested that uterine epithelial cells undergo apoptosis so that the syncytial layer is entirely trophoblastic in origin (102). If that is the case, the placenta is syndesmochorial according to the Grosser classification (7).

Although rodent and primate PSGs are thought to have evolved from a CEACAM-1-like gene, the two gene families arose independently through convergent evolution. Orthologous genes are found only within rodents (mouse and rat) or primates (human and baboon) (136). The Psg locus of the mouse comprises 17 genes and has been explored in detail (137). The rat Psg locus has evolved less rapidly and includes eight genes (136). Trophoblast giant cells express Psg22 in the first half of mouse pregnancy, whereas the spongiotrophoblast expresses Psg16, Psg21 and Psg23 in the second half (138). Mouse Psg23 can activate latent TGFβ1 (139) and likely increases the availability of regulatory T-cells, as shown by administering recombinant human PSG1 to mice (140).

PSGs do not occur in the guinea pig (6), but have not been sought in other rodents so it is not known if they evolved in the murid lineage or a deeper branch.

In the horse, PSGs evolved through expansion from a CEACAM1-like ancestral gene to a cluster of some 17 genes coding for secreted proteins (141, 142). Five are known to be expressed in the trophoblast of the endometrial cups (142). As mentioned above, the cups are formed by invasive trophoblast and are responsible for secretion of eCG. The endometrial stroma surrounding the cups is heavily infiltrated by CD4+ and CD8+ T-cells as well as by macrophages and natural killer cells (143, 144), yet the cups survive until at least 100 days of gestation. Immunosuppression by secreted PSGs is a plausible hypothesis (142). It is thought that regulatory T-cells play a role in tolerance of the invasive trophoblast (145). Since at least one of the equine PSGs can activate TGFβ1, they may contribute to differentiation of regulatory T-cells (146) as shown for human and mouse (130). PSG expression has not been examined beyond 36 days (142) and it would be interesting to know if the putative protection is withdrawn at a later stage when the cups become necrotic and eventually are sloughed off. It is not known if PSGs occur in other perissodactyls, such as tapirs and rhinoceroses.

Chiroptera is the most speciose order after rodents. Based on molecular and morphological evidence it can be divided into two clades: Yinpterochiroptera includes megabats and six families of echo-locating bats; the remaining orders constitute Yangochiroptera (147, 148). Expansion of CEACAM genes has occurred in Yinpterochiroptera, but CEAPSG-like genes have been found only in three families of Yangochiroptera (149). One cluster of putative PSG genes occurs in the Natal long-fingered bat (Miniopterus natalensis), the common moustached bat (Pteronotus parnellii) and four species of vesper bat. A second cluster is restricted to the vesper bats. It must be stressed that bat PSGs were identified from genomic data. Further evaluation will require demonstration of placental expression of these genes.

Placental protein hormones evolved through duplication of existing genes including those coding for pituitary hormones (Table 2). Gene duplication often is preceded by a burst of rapid change in nucleotide sequence (87, 150). These bursts can be interpreted as reflecting adaptation of the gene product for a new function, alternating with reversal towards the original function. Repeated cycles of change would lead to accumulation of substantial changes. The process would end once gene duplication allowed a second protein to adopt the new function (87).

After initial duplication, the genes of many placental hormones have undergone further expansion to create multigene families. This is a contrast to adult hormones, which usually are encoded by a single gene (6). One explanation has been formulated in terms of parent-offspring conflict; a hypothesis based on conflicting priorities for allocation of maternal resources to the offspring. Paternal genes evolve to promote nutrient supply and improve survival of the neonates, while maternal genes evolve to conserve maternal resources for subsequent pregnancies. These ideas were developed to explain the evolution of placental lactogens and their receptors (151). Parent-offspring conflict has also been suggested as a plausible explanation for the expansion of PSG genes in human (eleven genes), mouse (seventeen), and horse (seven genes) (6). It has also been alluded to with respect to the multigene family of PAGs (119).

An alternative hypothesis suggests that bursts of rapid evolution in genes and their repeated duplication reflect the arms race between pathogens and their mammalian hosts. After decades of research on the evolution of growth hormone, prolactin and their receptors, Wallis concluded this to be a plausible explanation for his findings (150). He hypothesized that viruses could gain access to cells by binding to these hormones and then be internalized with the hormone-receptor complex. Thus, the hormones would evolve rapidly to hinder binding of the virus. Gene expansion, as seen with placental lactogens and their kin, would result in multiple species that could act as viral decoys. Human PL reaches very high levels in the second half of gestation, 100 times greater than GH in nonpregnant humans, yet its physiological significance remains unclear. However, a high concentration of hPL would decrease the odds of a virus fastening to a molecule that was to be internalized after receptor binding. The downside to this hypothesis is that viruses capable of binding to growth hormone or prolactin have yet to be identified (150, 152).

Here it is worth noting that CEACAMs are thought to act as decoy receptors and PSGs may act in a similar fashion, so this is an alternative explanation for expansion of PSG gene loci (149). However, as with the PL locus, there is no clear evidence that PSGs bind to microorganisms (6).

Fetal growth is regulated by the insulin-like growth factor system with IGF2 as a key factor (67). IGF2 is expressed in the adults of most amniotes (reptiles, birds and mammals), including human (153), whereas in mouse and rat Igf2 is expressed in placenta but is largely absent in adults. Both IGF2 and the M6P/IGF2R receptor are imprinted genes in most mammals including marsupials (154). In mice the phenotype of the conceptus, including the trophoblast, is determined by the paternal allele of Igf2 and the maternal allele of Igf2r. This reciprocal imprinting has been interpreted in terms of parent-offspring conflict: the kinship theory of genomic imprinting (155). It has limited application to human pregnancy as imprinting of M6P/IGF2R was lost in the lineage of primates, colugos and tree shrews (156).

No gene for human placental protein hormones has orthologous genes in mammals other than anthropoid primates. That does not mean that findings in animal models are without merit for understanding the role of placental hormones in human pregnancy. However, hypotheses generated from models need to be verified in clinical studies or further explored in nonhuman primates (81).