Interferon-τ is a major product of ovine and bovine conceptuses during the period before the trophoblast makes firm attachment to the uterine wall and begins to form a placenta. Its primary function is in preventing a return to ovarian cyclicity and hence ensuring the pregnancy continues, although it undoubtedly has other roles in ensuring receptivity of the maternal endometrium. Despite having properties similar to those of other Type 1, IFN-τ is not virally inducible and functions in a constitutive process unrelated to pathogenesis. The genes for IFN-τ (IFNT), which are confined to ruminant ungulate species, would appear to be the most recently evolved mammalian Type 1 gene family and are primarily under the transcriptional control of Ets2 and signal transduction pathways that target that transcription factor. The IFNT provide an illustration of how a gene control region can be commandeered and then refined to provide a radically changed pattern of expression.

My entry into the field of interferon research was quite by chance and brought about by the unexpected discovery that the factor responsible for “rescuing” the corpus luteum (CL) during early pregnancy in sheep and cattle was a Type 1 interferon (IFN) (1-3). The CL is the normally transitory structure that develops on the ovary from the ruptured ovarian follicle following release of the oocyte and would normally regress if that oocyte failed to be fertilized and a pregnancy did not ensue. Its primary function is to produce the steroid hormone, progesterone. In most eutherian mammals, the functional lifespan of the CL must be extended during a pregnancy to provide continued production of progesterone, a steroid hormone that acts on the mucosal lining of the uterus (endometrium) to provide an environment in which the conceptus (defined here as embryo plus its surrounding membranes) can continue to develop. Therefore, the prevention of luteal regression is crucial if a pregnancy is to succeed and depends upon a biochemical signal that originates from the conceptus. In the case of cattle and sheep and related ruminants, the active factor is now known as IFN-τ, which acts locally on the maternal uterine endometrium to prevent the release of the luteolytic factor, prostaglandin F2α, at the end of the ovarian cycle when the CL is poised on the point of regression. IFN-τ probably also induces local changes in the maternal uterus to provide a local environment friendly to subsequent conceptus development. In this short paper, I describe the discovery of IFN-τ, how the genes (IFNT) encoding these IFN appear to have evolved, and the nature of the cis-regulatory control elements on the IFNT responsible for their unique pattern of expression. In closing, I speculate that although the IFN-τ would appear to provide a recently evolved mechanism for conceptus-maternal signaling, there may be analogies in other mammalian groupings, including primates.

By the late 1970s, it had become clear that there was no common mechanism to explain how CL rescue occurred in different species (4, 5). In primates, where the conceptus implants into the uterine wall within a day or two of entering the uterus, a chorionic gonadotrophin (hCG in the case of the human) is secreted by the invading trophoblast, gains access to the maternal blood supply, and provides direct luteotrophic support to the CL. The situation in non-primate species is quite different. No species of mammal outside the primate order, for example, appears to even possess the gene for the β-subunit of CG. In addition, attachment of the conceptus to the uterine wall in cattle, sheep and related pecoran ruminants does not occur until the third week of pregnancy, and implantation, when it occurs, is superficial. Surprisingly, the prospective mother “knows” she is pregnant and her CL is already programmed for extension well before these placentation events are initiated. By 1980, it had become clear that the active factor responsible for CL rescue and extension of the ovarian cycle was a protein produced by the conceptus for a few days prior to when the conceptus attaches to the uterine wall, and only effective if it were introduced into the uterine lumen rather than directly into the ewe’s blood stream (4, 6). At the University of Florida, my colleagues, Fuller Bazer and Jim Godkin, and I reasoned that the protein factor was probably, therefore, secreted and produced transiently during pregnancy. Advantage was taken of the fact that it is possible to flush conceptuses from the surgically exposed uteri of early-pregnant ewes, and then to culture them as intact, whole structures. Accordingly, conceptuses were cultured under serum-free conditions in presence of a radioactive amino acid for about 24 hours, and proteins released into the medium analyzed by two-dimensional PAGE. These experiments immediately revealed that the major secretory product released by the unattached sheep conceptus was a protein of molecular weight approximately 18,000, which consisted of several forms differing slightly in isoelectric point (7). The production of this protein, which was maximal at about day 15, became undetectable after the end of the third week of pregnancy. We were able to purify this conceptus product relatively easily, show that it could be introduced into the uteri of non-pregnant ewes and cause a temporary extension of CL function, and, after raising a specific antiserum, clone its cDNA from a cDNA expression library prepared from polyadenylated conceptus RNA (reviewed in (8)). The nucleotide sequence of this cDNA revealed that it encoded a protein that was likely to be a Type 1 IFN. A parallel series of experiments with cattle embryos showed that a similar IFN-like protein was responsible for maternal recognition of pregnancy in cattle. These IFN eventually became recognized as a new family of Type 1 IFN, known as the IFN-τ (see (8, 9)).

Ungulate species, in general, have placentae that are superficial and minimally invasive compared, for example, to those of rodents and higher primates. Although it may appear counter-intuitive, such placentae are a more recent innovation than the hemochorial placentation of the human and mouse (28) and highly efficient. However, as the structure of the placental barrier has evolved, so has the manner in which the conceptus signals to the mother. Some years ago we suggested that the IFN-τ had arisen in conjunction with the non-invasive synepitheliochorial placentation of ruminant artiodactyl species (10), possibly as an antiviral barrier, initially to protect the conceptus against infection (13), but eventually evolving to act as mediators of maternal recognition of pregnancy. The question that must be posed is whether the role of IFN-τ as a hormone of pregnancy has been achieved by their acquiring unique, or, at the very least, specialized biological properties. Alternatively, is the key to IFN-τ physiology their local production in emerging trophoblast at a critical time that allow them to intervene in the normal ovarian cycle and prevent the loss of CL function that would otherwise end the pregnancy? Certainly, the IFN-τ bind the same IFNAR1/IFNAR2 receptor used by other Type 1 IFN (29). Experiments comparing the potency of an ovine IFN-τ against another sub-type of ovine IFN, however, are not easily performed, but this laboratory has found little difference between an ovine IFN-τ and an ovine IFN-α in their respective abilities to extend estrous cycle length when infused into the uterine lumen of non-pregnant ewes (30). In cattle, limited IFN-α / IFN-τ comparisons have been attempted, but equivalent amounts of protein, rather than comparable quantities of antiviral activity, have been tested (see (31, 32). IFN-τ have excellent antiviral activities, at least the equivalent of corresponding IFN-α from the same species (33, 34). It still remains possible that IFN-τ have some specialized function, an ability for example to regulate a complement of genes in its target, the maternal endometrium, that make the action of this cytokine especially adapted to pregnancy signaling and the shut down of prostaglandin release. At present there is no convincing evidence that the IFN-τ have acquired such a specialized function. One question that has been addressed is whether the six amino acid extensions at the carboxyl tail of the IFN-τ have any significance. A recombinant form of ovine IFN-τ in which this tail had been deleted was no less potent in its ability to extend estrous cycle length than its normal length counterpart (35). The amounts used to achieve CL maintenance were comparable or lower than those secreted daily by individual conceptuses at the time of peak production (36).

Although the IFN-τ are unique to the Ruminantia, a sub-order within the much larger Artiodactyla order, there are hints from transcriptional profiling experiments that a partial interferon-response occurs in humans and mice as the trophoblast makes intimate contact with the maternal endometrium and induces the formation of a deciduum, in which stromal fibroblast proliferate and form an edematous structure adjacent to the site of implantation (37-39). This response does not involve a full-blown up-regulation of many interferon-stimulated genes (ISGs), as noted in the sheep endometrium during early pregnancy (12, 13, 40), but is instead more limited. The notion that there are one or more signaling pathways common to the onset of early pregnancy across diverse taxonomic groups is an attractive one, because a process as fundamental to mammalian development as trophoblast implantation, although subject to maternal-fetal conflict and hence rapid evolutionary change(5, 41), would be expected to exhibit some vestiges of conservation. Krause & Pestka (19) in their recent review suggest that, in species of mammal that do not possess IFNT, production of another cytokine, possibly even a Type 1 IFN, such as IFN-ε, could provide the crucial signal from the growing trophoblast that prompts the local maternal endometrium to become a receptive site for the conceptus to grow and develop. What may be the conserved process in implantation is the downstream signaling pathway rather than the signaling molecule itself.

The discovery of IFN-τ can be considered important for several reasons. First, it described a Type 1 IFN functioning in a constitutive biological process unrelated to pathogenesis. Second, the find provided the first example of a cytokine involved in trophoblast signaling, plus an example of how a defined trophoblast product might intervene in maternal immune responses. Finally the IFNT would appear to be the most recent mammalian Type 1 gene family to have arisen; they provide a clear illustration of how a gene control region can be commandeered and then refined to provide a radically changed pattern of expression.