Life or Death Decisions in the Corpus Luteum

Introduction

Because of its critical role in establishment and maintenance of pregnancy, there has been much written about the function of the corpus luteum (CL). Domestic ruminants are an excellent choice of animal model for studies of the CL, because each oestrous cycle is characterized by a fully functional CL, the timing and length of the oestrous cycle can be easily manipulated with exogenous hormones, the CL can be easily removed from the ovary for in vitro analyses, and the large size of the CL provides significant amounts of tissue for experimentation. For these reasons, much of what is known about the CL is based on work performed with cattle or sheep, which will be the focus of this review.

Within the size restriction of this article, it is impossible to cover all of the work that has been carried out on steroidogenic mechanisms, signalling pathways, extracellular matrix, reactive oxygen species and some novel regulators of luteal cells. Therefore, emphasis will be placed on regulation of luteolysis, mostly citing recent work that has shed some light on our understanding of the role of immune cells and prostaglandins in luteal regression. For additional information, the reader is referred to other recent reviews on the CL (Miyamoto et al. 2010a; Pate et al. 2010; Shirasuna 2010; Skarzynski and Okuda 2010).

Regulation of Uterine Release of PGF_2α

In ruminants, regression of the CL at the end of the oestrous cycle is caused by uterine release of prostaglandin (PG)F_2α. The synthesis of PGF_2α is regulated by ovarian steroids acting in the endometrium and release into the uterine vein may be stimulated by oxytocin, although the role of oxytocin in luteolysis is controversial (rev. Goodman and Inskeep 2006). Ablation of ovarian follicles delayed luteolysis in heifers, but luteolysis was restored with oestradiol replacement (Araujo et al. 2009). However, the ability of oestradiol to promote luteolysis may be concentration and time-dependent (Bisinotto et al. 2012). An elevated concentration of oestradiol from the developing preovulatory follicle is likely involved in initiating PG release from the uterus, but is not necessary to maintain pulses of PGF_2α (Pugliesi et al. 2012). However, it has become clear that sequential pulses of PGF_2α are necessary for complete luteolysis to occur (Ginther et al. 2009), and that the endocrinological changes associated with luteolysis differ in cattle that receive sequential PGF pulses compared to a bolus treatment of PGF (Shrestha et al. 2010).

During maternal recognition of pregnancy, the embryo secretes numerous molecules that could be involved in alteration of uterine function and rescue of the CL. Considerable evidence has accumulated that the Type I interferon (IFN), IFNτ, is the primary embryonic signal in ruminants (rev. Bazer et al. 2008), whereas members of the IFNδ family may play a role in maternal recognition in horses (Cochet et al. 2009). IFNτ can suppress uterine release of PGF_2α, and it is widely asserted that inhibition of PGF_2α release or alteration of the pulsatile pattern of its release is the mechanism by which luteolysis is prevented. However, in sheep, the concentration and pattern of release of PGF_2α does not differ in pregnant compared to non-pregnant ewes (Lewis et al. 1977, 1978), suggesting that the antiluteolytic effect of IFNτ may involve mechanisms other than, or in addition to, alteration of uterine PGF_2α release. The conceptus secretes more than one Type I IFN (Cochet et al. 2009). Because Type I IFNs are potent activators of immune cells, perhaps one of the primary functions for embryonic IFNs is to communicate with uterine and/or blood-borne immune cells.

A number of interferon-responsive genes are upregulated in the endometrium in response to IFNτ that may affect various aspects of endometrial function, but there are systemic responses to IFNτ as well, most notably an increase in IFN-responsive genes in circulating immune cells (Yankey et al. 2001; Gifford et al. 2007; Ott and Gifford 2010). Interferon-stimulated gene 15 (ISG15) and myxovirus resistance protein 1 (MX1) are upregulated in corpora lutea of pregnant sheep and cows (Oliveira et al. 2008; Yang et al. 2010), and there is now evidence that IFNτ is released into the uterine vein and may itself exert endocrine effects on the CL (Spencer et al. 1999; Bott et al. 2010; Hansen et al. 2010). Yang et al. (2010) failed to observe IFNτ-induced stimulation of ISG15 in cultured bovine luteal cells, raising the possibility that the effects of IFNτ in the CL may be mediated by a cell type that was absent in the cultures, such as immune or endothelial cells. However, Benyo and Pate (1992) and Pate (1995) demonstrated that IFNα, which acts on the same Type I interferon receptor as IFNτ, suppressed both cytokine-stimulated prostaglandin production and expression of class II major histocompatibility molecules in cultured luteal steroidogenic cells, indicating that these cells can respond to type I interferons.

The CL of early pregnancy is more resistant to the luteolytic effects of PGF_2α than the CL on the same day of the normal oestrous cycle (Pratt et al. 1977; Silvia and Niswender 1986), suggesting that intraluteal factors can influence the life or death fate decisions that take place in the CL. There is a growing list of potential factors that may be involved in the luteolytic process, including endothelial-derived proteins, immune cell-derived cytokines, reactive oxygen species, neuropeptides and lipophilic molecules (for recent reviews, see Miyamoto et al. 2010a,b; Pate et al. 2010; Shirasuna 2010; Skarzynski and Okuda 2010). Although all of these molecules and diverse cell types that secrete them may be involved in the luteolytic cascade, it is often unclear which of the PGF_2α-responsive factors actually commit the CL to regression. The difference in sensitivity of the developing CL to PGF_2α compared to that of the mature CL has also been exploited as a means to understand luteal responses to PGF_2α that result in activation of death pathways vs responses to PGF_2α that are transient (Tsai and Wiltbank 1998; Sayre et al. 2000). These studies have pointed to involvement of endothelial cells, extracellular matrix molecules and cytokines in PGF_2α-induced luteolysis (rev. Pate 2003). More recently, Atli et al. (2012) utilized a model of repeated intrauterine infusions of PGF_2α to enable the direct comparison of CL that underwent luteolysis, to those that exhibited temporary responses to PGF_2α but did not regress. In that model, the luteolytic effect of PGF_2α appeared to depend on the regulation of luteal prostaglandin synthesis and activation of immune response genes.

Heterogeneous Cells Involved in Luteolysis

There is now a wealth of evidence that capillary endothelial cells are intricately involved in the functional cascade of events that result in luteolysis. Most of this work has been reviewed by Townson (2006), Skarzynski et al. (2008), Miyamoto et al. (2010b) and Skarzynski and Okuda (2010). Additional recent work indicates that growth factors may play a role in determining the fate of luteal endothelial cells. The luteolytic effect of PGF_2α may depend on inhibition of fibroblast growth factor 2 effects (Castilho et al. 2008; Zalman et al. 2012), which would promote angiogenesis. In contrast, Maroni and Davis (2011) demonstrated clear effects of transforming growth factor beta (TGFB), which is increased in the CL in response to PGF_2α, on bovine luteal endothelial cells. In their experiments, TGFB caused a loss of cell contacts, increased permeability of endothelial cell monolayers and inhibited cell proliferation, consistent with a role for TGFB in mediating the luteolytic effect of PGF_2α by disrupting the capillary endothelium. Interactions among the heterogeneous cell types that comprise the CL may be required for the full luteolytic effect of PGF. Using cultures of murine luteal cells, Kuranaga et al. (2000) observed greater prolactin-induced apoptosis of steroidogenic cells when immune cells were present in the culture. The inhibitory effect of PGF_2α on cultured bovine luteal steroidogenic cells may be enhanced by the presence of immune cells and endothelial cells (Liptak et al. 2005; Korzekwa et al. 2008a).

One must exercise caution when interpreting experiments in which any type of luteal cells was treated with PGF_2αin vitro. A review of the literature within the past 10–15 years shows that most investigators are using a pharmacological concentration of PGF_2α for in vitro experiments. The most commonly used concentration is 1 μm, which is 354.5 ng/ml. During luteolysis, the peak concentrations of PGF_2α in the uterine venous drainage of sheep (Pexton et al. 1975; Ottobre et al. 1980) and PGFM in the peripheral circulation of cattle (Kindahl et al. 1976a,b; Garrett et al. 1988) are 0.6–11 ng/ml. Therefore, investigators are using 32–590 times more PGF_2α than the CL would be exposed to in vivo. This is especially disconcerting when one considers that at higher concentrations, PGF_2α can bind to the PGE₂ receptor (Rao 1974). Anderson et al. (1999) reported that the IC₅₀ of PGF_2α for the PGE₂ receptor is 66 nm. Clearly, a 1-μm concentration of PGF_2α is well above the range at which effects could be mediated by the PGE, rather than the PGF, receptor. Activation of the PGE₂ receptor results in stimulation of progesterone production (Speroff and Ramwell 1970), which can explain why some investigators report ‘luteotropic’ effects of PGF_2α on luteal cells in vitro. When luteal steroidogenic cells are treated with 10 ng/ml of PGF_2α, there is a clear inhibition of LH- and lipoprotein-stimulated progesterone synthesis (Pate and Condon 1984). It is therefore suggested that future studies of luteolytic mechanisms using in vitro models utilize physiological concentrations of PGF_2α.

Lobel and Levy (1968) first described the infiltration of immune cells into the CL of the cow, and using detailed ultrastuctural analysis, Paavola (1979) showed that immune cells were involved in removing dead cells and debris during luteal regression in the guinea pig. Since that time, numerous investigators have used immunohistochemical techniques to identify the types of immune cells that accumulate in the CL, with particular focus on T lymphocytes, monocyte/macrophages (rev. Pate 1995; Penny 2000), neutrophils (Shirasuna et al. 2012) and dendritic cells (Spanel-Borowski 2011). Some recent work has focused on the chemokine and adhesion molecules that may be responsible for trafficking of immune cells into the CL, including prokineticin (Kisliouk et al. 2007) and P-selectin (Shirasuna et al. 2012). The type of immune cell that migrates into the CL appears to be regulated by intrinsic and extrinsic factors. Immune cell trafficking in the ovary can be influenced by the presence of seminal plasma in the uterus (O’Leary et al. 2006), and the profile of immune cell types in the fully functional CL differs from that in CL that have been induced to regress (Poole and Pate 2012). This could occur if T cells that reside in the fully functional CL exit the tissue and are replaced with an alternative type during luteal regression. Alternatively, when the CL begins to regress, the resident T cells may receive signals that alter their function, presumably so they can participate in the remodelling events associated with luteal regression. Analysis of luteal-resident T lymphocytes by flow cytometry allowed for distinction of multiple subtypes of T cells, characterized by differential expression of cell surface molecules. The most profound changes were in membrane-associated proteins that are involved in regulation of cell signalling, suggesting alteration in function of existing tissue-resident cells (Poole and Pate 2012). Surprisingly, the changes in T cells that were observed in this study were more reflective of T cells that are involved in homoeostatic mechanisms, as opposed to proinflammatory events, a finding that has changed the way we think about the current dogma of an extensive proinflammatory event during luteolysis.

Cytokine Involvement in Prostaglandin-Induced Luteolysis

The idea that resident immune cells may influence steroidogenic cell function or viability via paracrine regulators may have first been suggested by Bagavandoss et al. (1988, 1990) who reported an association between TNF production and immune cell accumulation in the rabbit corpus luteum. This was followed by identification of TNF and its receptor in bovine CL, as well as the demonstration that TNF and IFNG exerted profound effects on bovine luteal cells in vitro. Specifically, these cytokines inhibited LH-stimulated progesterone production and caused induction of apoptosis of the steroidogenic cells (rev. Pate 2003). The presence and temporal expression of these cytokines and their receptors has been more fully described in the last few years. The mRNA for TNF, TNF receptors and IFNG increases during luteolysis (Neuvians et al. 2004; Korzekwa et al. 2008b). Expression of these molecules in the CL of mares is similar to that of cows, and the effects of TNF and IFNG on equine luteal cells vary with stage of the oestrous cycle (Galvão et al. 2012). Using in situ hybridization and immunohistochemistry, Sakumoto et al. (2011) localized TNF and its receptors to large and small steroidogenic cells, endothelial cells and immune cells, indicating that there is the possibility that locally produced TNF can act as a paracrine mediator of luteal function. An interesting observation from this study is that the TNF- and TNFR-positive cells were sparsely scattered in the functional CL, but most cells are TNF/TNFR-positive during luteal regression, corroborating the quantitative studies of TNF in the CL and indicating that it is not simply increased in a few immune cells, but its expression is widely induced in multiple cell types during luteolysis. In addition to the well-documented effects of TNF/IFNG on steroidogenic cells, these cytokines also modulate synthesis of prostaglandins, leukotrienes and endothelins in bovine luteal endothelial cells (Korzekwa et al. 2011) and can induce endothelial cell apoptosis (Hojo et al. 2010). Endothelial cells are the first to undergo apoptosis in the regressing CL (Sawyer et al. 1990), perhaps indicating that TNF/IFNG play an important role in the very early events of luteolysis.

A role of TNF in luteal regression was conclusively demonstrated in mice, in which removal of TNF action, by using either a TNF neutralizing antibody or TNFRI null mice, inhibited PGF-induced luteolysis (Henkes et al. 2008). In the same study, a similar effect was observed in mice that lacked acid sphingomyelinase, which is a critical enzyme in TNF signalling. It is much more difficult to evaluate in vivo effects in large, domestic animals, but Skarzynski et al. (2007) attempted to do so using intrauterine infusions of TNF in cows. The effect of TNF was dose-dependent; high concentrations of TNF prolonged luteal lifespan but low concentrations stimulated uterine PGF_2α release and promoted luteolysis. The low doses used in this study are likely more closely aligned with physiological concentrations, suggesting that TNF is luteolytic, although the role of intraluteal TNF was not directly tested. However, temporal changes in expression of TNF/IFNG in the CL and effects on luteal cells in vitro provide strong support for a role of these cytokines as paracrine effectors of luteolysis.

Luteal Prostaglandin Synthesis and Metabolism

One of the most profound effects of TNF and IFNG on luteal cells is to stimulate prostaglandin synthesis (Nothnick and Pate 1990; Fairchild and Pate 1991; Benyo and Pate 1992; Townson and Pate 1994). This link between cytokines and luteal prostaglandin synthesis may explain why gene families classified as ‘immune response’ and ‘prostaglandin-related’ are differentially expressed in CL that undergo complete luteolysis compared to CL that receive a luteolytic signal, but do not regress (Atli et al. 2012). Milvae and Hansel (1983) described temporal changes in the ratio of PGI₂:PGF_2α in the bovine CL throughout the oestrous cycle and suggested that intraluteal prostaglandin synthesis is a critical component of the luteolytic cascade. Luteal prostaglandin synthesis would not only be stimulated by the increase in cytokines during luteolysis, but is suppressed prior to luteolysis by the high concentrations of progesterone within the CL (Pate 1988). Diaz et al. (2011) and Luo et al. (2011) showed that treating pigs with epostane to inhibit progesterone production allowed PGF_2α to induce luteolytic responses, including upregulation of chemokines, in CL that had not yet acquired luteolytic capacity. Thus, the intraluteal concentration of prostaglandins and the ratio of luteotropic to luteolytic prostaglandins produced within the CL are likely important for fulfilment of luteolysis, and this may be directly influenced by intraluteal progesterone. The time of maternal recognition of pregnancy is another period in which the CL is less sensitive to PGF_2α, and this is associated with a lower concentration of prostaglandin F synthase (Costine et al. 2007). There are greater concentrations of the mRNA for 15-hydroxyprostaglandin dehydrogenase as well as increased activity of this enzyme in the CL of pregnant compared to non-pregnant ewes (Silva et al. 2000), and more PGF is catabolized to PGFM in the CL of pregnant animals (Costine et al. 2007). Atli et al. (2012) reported that CL that regressed in response to intrauterine infusions of PGF_2α had elevated concentrations of mRNA for prostaglandin-endoperoxide synthase 2 (PTGS2) and prostaglandin F synthase (PTGFS), but hydroxyprostaglandin dehydrogenase (HPGD) mRNA concentrations were not entirely consistent with reduced PGF_2α catabolism to achieve luteolysis. However, Costine et al. (2007) reported that HPGD is post-transcriptionally regulated in the CL, so HPGD mRNA concentrations may not reflect activity. Together, this means that luteal survival during maternal recognition of pregnancy involves two mechanisms to reduce intraluteal concentrations of PGF_2α. The first is to decrease PGF_2α synthesis and perhaps increase the PGE:PGF ratio, and the second is to catabolize PGF_2α. Conversely, luteolysis likely involves an increase in PGF_2α synthesis and suppression of PGF_2α catabolism, resulting in elevated intraluteal concentrations of PGF_2α. Changes in luteal metabolism of prostaglandins during maternal recognition of pregnancy could be due to direct effects of IFNτ on the CL, because IFNα suppressed cytokine-stimulated prostaglandin production by cultured luteal cells (Benyo and Pate 1992).

Niswender et al. (2007) tested the theory that luteal prostaglandin synthesis is necessary for luteolysis by using intraluteal implants containing the prostaglandin-synthesis inhibitor, indomethacin. In ewes that received indomethacin-containing implants, progesterone declined at the end of the oestrous cycle, but the weight of the CL was heavier than that of control animals and was comparable to the weight of a midcycle CL. They further demonstrated that this was a local effect of indomethacin, by placing control implants in the CL on the opposite ovary. The weight of the control-implanted CL declined at the end of the oestrous cycle, whereas the weight of the indomethacin-implanted CL was maintained. These results are nearly identical to results obtained in our laboratory in cows, using implants that contained an inhibitor of cytokine synthesis, Cyclosporin A (CsA), rather than an inhibitor of prostaglandin synthesis. In this experiment, the corpus luteum of regularly cyclic heifers received silastic implants containing either saline (n = 4) or the immunosuppressive drug, CsA (n = 5) on day 13 of the oestrous cycle. Daily blood samples were collected for determination of plasma progesterone, and luteal size was monitored by ultrasonography until a discernible CL could no longer be detected. CsA had no effect on plasma progesterone or length of the oestrous cycle. As expected, the size of the CL in saline-implanted animals declined throughout luteolysis and the early part of the following oestrous cycle (Fig. 1). The same pattern of decrease in luteal size was observed in 3 of the 5 CsA-treated animals, whereas in two of the CsA-treated animals, the size of the CL remained unchanged, then very gradually declined at a much later stage than in the control animals. The number of days that a CL could be observed is presented in Table 1. Clearly, the CL in two of the experimental heifers, 587 and 571, exhibited an increased structural lifespan. Unfortunately, the silastic implants did not allow for controlled release of the CsA, and the concentration of CsA in the tissue at the end of the experiment was highly variable. It is possible that sufficient concentrations to inhibit T cell function were achieved in the two animals in which luteal size was maintained, because in our experience, this length of time of structural maintenance of the CL is never observed in regularly cyclic cattle. Therefore, local inhibition of either prostaglandin production or immune cell function/cytokine production may result in more gradual structural regression of the CL, further supporting a role for a cytokine–prostaglandin interaction in ensuring complete luteolysis.

Conclusion

Rescue of the CL during maternal recognition of pregnancy is critical for embryo survival, and therefore, understanding the factors that result in luteal maintenance or luteal death is important in the quest to increase fertility of domestic animals. The ability of PGF_2α to bring about luteolysis almost certainly depends on interactions among the heterogeneous cell types that are found within the CL and may involve synthesis of immune cell cytokines that stimulate PGF_2α production in luteal cells. Embryonic production of IFNτ during maternal recognition of pregnancy may alter the sensitivity of the CL to exogenous PGF_2α by modifying luteal PG metabolism and immune cell responses. Similarly, the resistance of early CL to exogenous PGF_2α may be due to the lack of resident immune cells in those CL as well as to intraluteal synthesis of an abundance of luteotropic prostaglandins during luteal development.

Acknowledgement

The authors thank Ms. Shannon Boone for her assistance with the preparation of the Figure and manuscript.

Conflicts of interest

The authors have no conflicts of interest to declare.