The role of macrophage migration inhibitory factor in maintaining the immune privilege at the fetal–maternal interface
- First Online:
- Cite this article as:
- Viganò, P., Cintorino, M., Schatz, F. et al. Semin Immunopathol (2007) 29: 135. doi:10.1007/s00281-007-0074-3
- 162 Views
Macrophage migration inhibitory factor (MIF) is a pivotal regulator of the innate and adaptive immunity affecting the response and behavior of macrophages and lymphocytes. MIF is also implicated in other fundamental cellular processes including angiogenesis and cell proliferation. Several studies examined the expression of MIF in reproductive organs and tissues and its involvement in different aspects of human and animal reproduction. The goal of this review was to summarize these findings and discuss, in particular, the role of MIF in the maintenance of the immune privilege at the human fetal–maternal interface.
KeywordsMIFPregnancyImmunosuppressionUterine NK cellsDecidual macrophages
The name macrophage migration inhibitory Factor (MIF) was proposed—about 40 years ago—to identify a factor able to inhibit the random migration of macrophages from cultured guinea pig peritoneal exudates in a capillary tube assay . However, the cellular origin of this activity remained undefined until 1966 when Bloom and Bennet  and David , studying the delayed type hypersensitivity reaction, identified MIF as a factor produced by activated T lymphocytes. Subsequent studies essentially focused on the effects of MIF on the activation of macrophage and the resulting enhancement of their antimicrobial and tumoricidal activity [4, 5]. A substantial improvement in our knowledge of the biological activities of MIF came after cloning of the human MIF cDNA and the subsequent availability of the recombinant protein and specific antibodies. Since the early 1990s, an impressive number of publications depicted the image of a molecule with a complex array of features and involved in multiple biological processes. The present review will summarize these features and describe the most important of the physiological and pathological activities of MIF. This review will then focus on the expression and roles of this cytokine in reproductive tissues, with particular emphasis on the potential involvement of MIF in the maintenance of immune privilege at the fetal–maternal interface.
Gene, protein, and receptor
MIF gene and protein
In 1994, Paralkar and Wistow  reported the human MIF gene as having three exons separated by two short introns and spanning less that 1 Kb. The 5′-flanking region contains multiple Sp1 sites, a cAMP response element (CRE), and a CATT repeat. The human gene is located on chromosome 22 (22q11.2) . The gene for mouse MIF is located on chromosome 10, and its 5′-proximal promoter region possesses NF-kB, CK-1, glucocorticoid response element (GRE), CRE, and Sp1 [8, 9]. While only one MIF gene has been described in human, several pseudogenes have been found in the mouse genome. In both humans and mice, a single mRNA of approximately 0.8 Kb has been described.
MIF is a highly conserved protein, with homologues found in evolutionary divergent organisms including vertebrates, worms, insects, plants, and protists. In mammalians, a particular high degree of homology (90%) has been reported between the species examined . The MIF protein consists of 114 residues and has a predicted MW of 12,345 Da. Our laboratory characterized human MIF as consisting of three different isoforms with pI values of 7.8, 6.98, and 6.23, respectively, with the latter being expressed specifically in the liver .
There has been much conflicting information about the state of oligomerization of MIF in solution. Crystallography studies have indicated MIF as a homotrimer [12, 13], while other studies reported MIF as a monomer , a dimer , a trimer , or as a monomer, dimer, and trimer mixture . Recently, however, Philo et al.  showed convincing data consistent with a strongly associated trimeric quaternary structure.
Although MIF cytokine activity is one of the oldest described, its mechanisms of action are not yet fully understood primarily because of limited information available on its receptor and receptor-mediated response. In 2003, Leng et al.  reported the identification of CD74, the cell surface form of the invariant chain of major histocompatibility factor (MHC) class II, as a cell surface binding protein for MIF. These authors reported a high affinity interaction between MIF and CD74 and the requirement for CD74 expression to promote MIF-mediated cellular response such as signaling to extracellular signal-regulated kinase 1/2 (ERK1/2) mitogen-activated protein kinase (MAPK), prostaglandin (PG)E2 production, and cell replication. The involvement of CD74 in MIF-dependent response has since been extended to include other biological systems [20, 21]. Furthermore, the involvement of CD44, an activator of tyrosine kinase, as a member of the CD74 receptor complex leading to MIF signal transduction was recently proven .
Physiological and pathological functions of MIF
In addition to its first known recognized function—the inhibition of the random migration of macrophages in vitro—MIF has been shown to regulate the immune response by affecting a number of macrophage and lymphocyte features including cytokine synthesis, cell activation, and phagocytosis. MIF has been also implicated in several immune and inflammatory conditions including sepsis , delayed type hypersensitivity , and rheumatoid arthritis .
Subsequent studies have shown that MIF is endowed with features that go beyond those of a cytokine. In particular, MIF exhibits at least two catalytic activities, a tautomerase and an oxidoreductase activity. These findings have led to a definition proposed by Kleemann et al.  of MIF as a “cytozyme” to indicate a cytokine with enzymatic activity. Furthermore, high levels of MIF mRNA and protein have been observed in a number of cancers, including breast , prostate [28, 29], lymphoma , lung , and melanoma . These studies have linked MIF with cell proliferation, angiogenesis, tumor-associated immunity, and tumorigenesis.
Finally, MIF expression has been described in several normal tissues and cells including those of the endocrine, the nervous, and the reproductive system , thereby highlighting the complex and multifaceted physiological roles of this protein. Hereafter, we will summarize the most important physiological and pathological functions of MIF. For more specific information and details, the reader may refer to exhaustive review articles [10, 34–36].
MIF and the immune system
MIF in adaptive immunity
MIF has been long documented as a product of activated T lymphocytes. However, its role as regulator of T cell response was recognized only in the last decade. Studies of Bacher et al.  showed that neutralizing anti MIF-antibodies inhibit proliferation of T cells and their production of interleukin (IL)-2 in vitro, as well as antigen-driven T cell activation and antibody production in vivo. Furthermore, endogenous MIF was shown to be involved in the mitogenic response of lymphocytes after activation with exotoxins . Results from other studies, however, have challenged the role of MIF as an activator of T-lymphocytes. In particular, in a mouse model of cytotoxic T lymphocyte (CTL) response using the chicken ovalbumin (OVA)-transfected tumor cell line EG.7, Abe et al.  found that splenocytes from EG.7-primed mice treated with neutralizing anti-MIF antibody showed a significant increase in the CTL response compared with control cultures. They also showed augmented CD4(+) and CD8(+) T cells accumulation at the tumor site in anti-MIF treated EG.7 tumor-bearing mice and an increased accumulation of CD8(+) T lymphocytes in tumor tissue when the cells were transferred from the spleens of anti-MIF-treated animals. Recent results have also shown that neuroblastoma cell-derived MIF inhibits T cell activation and that tumor-derived MIF induces cell death in activated T cells . Although MIF is produced by both the T cell helper subsets TH1 and TH2, when TH1 and TH2 clones were stimulated with concanavalin A, MIF mRNA and protein levels were increased upon stimulation in the TH2, but not in the TH1 clones .
MIF and innate immunity
Although several studies have linked MIF to the adaptive immune host response, the predominant role for this cytokine is considered to be in the innate immunity. MIF is thought to play a central role in the host response to Gram-negative and Gram-positive bacteria. Bacterial lipopolysaccaride (LPS) induces MIF release by the pituitary, lung, liver, kidney, spleen, adrenal gland, and skin where significant quantities of MIF protein are stored preformed in the cells and appears to be released as a consequence of endotoxemia [41, 42]. In mice, recombinant MIF (rMIF) enhances LPS-induced toxicity when co-injected with LPS, while anti-MIF antibodies confer protection from LPS- and staphylococcal exotoxins-induced shock [38, 41]. Compared with wild-type mice, MIF deficient mice (−/−) are more resistant to LPS, and their macrophages show a lower production of tumor necrosis factor (TNF)-α in response to LPS and interferon (IFN)-γ . Protection of septic shock by neutralization of MIF has also been reported in two models of bacterial peritonitis .
The name MIF underscores the capacity of this cytokine to target macrophages. However, the effect of MIF on these cells extends beyond inhibition of random migration. Previous reports have demonstrated that rMIF can directly inhibit human macrophage  and monocyte chemotaxis . This latter study, in particular, showed that MIF can prevent monocyte chemoattractant protein-1 (MCP-1)-driven monocyte migration in a dose-dependent manner. Paradoxically, recent results show that MIF is capable of inducing monocyte migration in a murine model most likely through MCP-1 induction . Further investigations are required to establish the physiological relevance of this observation in humans and to relate it to the previously reported inhibitory activities of MIF.
MIF is constitutively expressed by macrophages, with high levels of protein present in resting cells. Endotoxins and cytokines, including TNF-α and IFN-γ, induce MIF release. However, rMIF stimulates the secretion of TNF-α by macrophages, suggesting that these two cytokines may interact in a proinflammatory loop . MIF-deficient macrophages are less responsive to LPS than are wild-type cells. This reduction reflects down-regulation of Toll-like receptor 4, the signal transducer of LPS . In addition, MIF acts on macrophages to increase phagocytosis  and killing of parasites  and tumor cells . A well-defined link exists between MIF and glucocorticoids (GC). In 1995, Calandra et al.  observed that GC induced MIF production in LPS-activated macrophages. Interestingly, dose-dependent studies revealed that this effect conformed a bell-shaped curve, reaching a maximum at 10−14, 10−12 M with lower response at 10−8, 10−6 M GC. These authors demonstrated that exogenously administrated MIF was able to override GC-dependent inhibition of TNF-α, IL-1β, IL-6, and IL-8 secretion by LPS-stimulated monocytes and suggested that MIF can counteract the anti-inflammatory activity of GC. More recent data suggested involvement of MIF-dependent activation of NF-kB in the cytokine counter-regulatory effect of GC .
Studies on natural killer cells (NK) provided further information about the role of MIF in the innate immunity. Apte et al. , attempting to elucidate factors preserving the immune privilege in the eye, found that aqueous humor contains a protein that inhibits NK cell-mediated cytotoxicity in vitro and identified such protein as MIF. They showed that rMIF inhibited NK-cell mediated lysis in a dose-dependent manner and that this effect was exerted by reducing perforin granule exocytosis. These authors subsequently proved that MIF is also responsible for inhibition of NK cell-mediated cytolysis detected in uveal melanoma cell supernatants .
Recent work in our laboratory confirmed that MIF affects NK cells by showing that uterine NK (uNK) cells express MIF and that both exogenous and endogenous MIF modulate uNK-mediated cytolysis of K562 target cells . These results and their potential pathophysiogical implications are detailed below.
Determination of the three-dimensional structure of MIF has demonstrated that homologous structures exist between MIF and the bacterial enzymes 4-oxolocrotonate tautomerase, 5-carboxymethyl-2-hydroxymuconate isomerase, and chorismate mutase . Each of these catabolic enzymes is involved in the degradation of aromatic compounds and catalyzes isomerization reactions. The amino acid sequence of human MIF has small but significant similarity to d-dopachrome tautomerase, a protein involved in melanin synthesis . Although the two proteins share less than 30% amino acid identity, MIF can also catalyze tautomerization of d-isomer of dopachrome . In addition, MIF exhibits enzymatic phenylpyruvate tautomerase activity , thiol-protein oxidoreductase activity , and catecholamine oxidase activity . Despite several reports, the exact relationship between the catalytic activities of MIF and the pathophysiological functions of the protein is still undefined. Several studies based on mutational analysis failed to link the enzymatic with the biological activity of MIF [45, 61, 62], with the only exception being an association between glucocorticoids counter-regulating activity and the thiol–protein oxidoreductase activity .
MIF in cell proliferation and angiogenesis
An important aspect of the research on MIF is its involvement in the control of cell proliferation and angiogenesis. In an early report on lens epithelial cell mRNA expression in a cataract rat model, Wen et al.  proposed that MIF over-expressed during cataract formation promote cell proliferation of undifferentiated epithelial cells. In the same year, a study on T lymphocyte by Bacher et al.  showed that anti-MIF antibodies inhibited T cell proliferation in vitro. Perhaps, the first insights into the mechanisms underlying MIF effects on cell growth came from studies by Mitchell et al.  on serum-mediated growth promotion of quiescent murine fibroblasts. These authors showed that both endogenous and exogenous MIF promoted proliferation of NIH/3T3 fibroblasts and that this response was associated with activation of the ERK1/2 MAPK pathway. By contrast, Kleemann et al.  observed that rMIF induced a dose-dependent reduction in fibroblast proliferative activity, while an anti-MIF neutralizing antibody caused a sharp increase in cell proliferation. According to these authors, the anti-proliferative effects of MIF occur through inhibition of Jun activation domain-binding protein 1 (JAB1)-dependent degradation of p27Kip1.
An increasing number of reports on the effect of MIF on cell growth came from studies on cancer cells. Takahashi et al.  demonstrated that the growth of murine colon carcinoma cells transfected with anti-sense MIF plasmid was significantly inhibited compared to control cells. In 1999, Shimizu et al.  showed that reduction of endogenous MIF levels using an antisense strategy in a human melanoma cell line was mirrored by a suppression of cell growth rate. More recently, Meyer-Siegler et al.  reported that reduction of MIF levels by RNA interference or by anti-MIF antibodies decreased proliferation of DU145 human prostate cancer cells and that MIF effects were likely mediated by the cell surface receptor CD74. Similar results were obtained in human neuroblastoma cells in which reduction of endogenous MIF levels was associated with a reduced proliferative activity .
Proof of the involvement of MIF in neo-angiogenesis has been primarily obtained using in vivo tumor models. Chesney et al.  observed that an anti-MIF antibody reduced the neovascularization of B cell lymphoma in vivo. They also showed that microvascular endothelial cells, but not lymphoma cells, produce MIF and that the cytokine is required for proliferation. Shimitzu et al.  and Ogawa et al.  reported that treatment of mice with an anti-MIF antibody resulted in a reduction in cancer cell-induced neovascularization. In the latter study, an anti-MIF antibody suppressed thymidine uptake by human umbilical vein endothelial cells. In another report, MIF increased the production of the vascular endothelial growth factor and IL-8 in hepatocellular carcinoma cells, while anti-MIF antibodies decreased it . More recent observations correlated MIF expression and microvessel density in human tumor specimens. Specifically, a significant correlation between levels of MIF and microvessel numbers was reported in malignancies such as hepatocellular , gastric , and esophageal squamous cell carcinoma . In non-tumor systems, MIF has been described as an endothelial cell growth-promoting factor produced by ectopic human endometrial cells. Yang et al.  showed that conditioned medium from endometriotic epithelial cell promotes human endothelial cell proliferation and identified MIF as one of the proteins responsible for this growth-promoting effect. They also showed that rMIF induced in vitro proliferation on human coronary arteries endothelial cells.
MIF mode of action
MIF-activated signaling pathways
Several studies have examined MIF-activated intracellular signaling events. In 1999, Mitchell et al.  noted that both exogenous and endogenous MIF-enhanced proliferation of quiescent NIH/3T3 fibroblasts by activating the ERK1/2 MAPK pathway. MIF-induced activation of MAPK pathway resulted in the phosphorylation and activation of cytoplasmic phospholipase A2 (cPLA2). This latter effect represents the first link between MIF and the synthesis of prostaglandins and leukotrienes, compounds with known pro-inflammatory and growth-regulating properties. In addition, these authors showed that MIF can override the suppressive effect of GC on TNF-α-induced arachidonic acid release, providing the first explanation of the capacity of MIF to counteract anti-inflammatory effects of GC. Recent studies have better clarified the upstream regulatory mechanisms and kinases of MIF-activated ERK1/2 MAPK pathway. Lue et al.  demonstrated that MIF is able to rapidly and transiently activate the ERK pathway and that this effect involves phosphorylation and activation of Raf-1, mitogen/extracellular signal-regulated kinase kinase (MEK1/2), and ERK1/2 and is dependent on an upstream Src-type tyrosine kinase.
Another potential mechanism by which MIF can promote inflammation is via activation of the NF-kB pathway, although recent reports have questioned the physiological relevance of this activity. NF-kB is a ubiquitous transcriptional factor involved in the regulation of many inflammatory genes expression including cytokines, adhesion molecules, and cyclooxygenase. It is activated by inflammatory cytokines, such as TNF-α and IL-1β, and modulated by GC that can repress the interaction of the p65 subunit with the transcription machinery and induce IkB-α synthesis. Several studies examined the activation of NF-kB by MIF as a mechanism underlying the cytokine counter-regulatory effect of GC. Daun and Cannon  have reported that in LPS-stimulated human peripheral blood mononuclear cells, MIF antagonizes the anti-inflammatory effects of GC by decreasing cytosolic IkB-α levels and augmenting nuclear NF-kB DNA binding. However, a recent study of Aeberli et al. , found no direct evidence of MIF effects on LPS-induced NF-kB binding and transcriptional regulation of IkB-α in dexamethasone-treated macrophages. Contradictory results were also obtained in studies evaluating MIF activation of the NF-kB pathway in other experimental systems. For instance, Amin et al.  showed that MIF increases the expression of cell adhesion molecules via NF-kB. On the contrary, Lacey et al.  have reported no effect of MIF on NF-kB on sinoviocytes. Although these conflicting observations may be explained, at least in part, by the different cell types used, they raise the need for a deeper evaluation of the physiological relevance of MIF and NF-kB interactions.
MIF and p53
Hudson et al.  identified MIF as a negative regulator of p53 tumor suppressor activity. These authors showed that MIF suppresses p53-dependent transcriptional activation and apoptosis and also bypass p53-mediated growth arrest. In macrophages, exogenous MIF suppresses nitric oxide (NO)-induced p53 accumulation, and MIF-dependent p53 inhibition involves the serial activation of ERK1/2, cPLA2, cyclooxygenase 2, and PGE2 [78, 79]. The anti-proliferative and pro-apoptotic activities of p53 can be mediated by the E2F family of transcription factors whose cell growth modulation is regulated by the retinoblastoma (Rb) family of proteins. Petrenko et al.  have clarified the molecular mechanisms underlying MIF control of p53 regulation, showing that MIF can modulate the E2F-dependent pathway and that this cytokine interferes with the functions of Rb and E2F in DNA replication .
MIF and JAB1
Jun activation domain-binding protein 1 was initially identified as a coactivator of c-Jun  and was later shown to be a component of the COP9 signalosome complex . Previous work demonstrated that a number of diverse proteins interact with the subunits of the COP9 signalosome, particularly with JAB1. It was found that JAB1 activates c-Jun amino-terminal kinase (JNK) and thereby activates c-Jun . Interestingly, JAB1 was shown to regulate the cell cycle by degrading the cyclin-dependent kinase inhibitor p27Kip1 . The transcriptional coactivator function of JAB1 is also connected with enhancement of the activator protein-1 (AP-1) transcriptional activity . In 2001, Kleeman et al. , using the yeast two-hybrid system, demonstrated that MIF interacts with JAB1. These authors also showed that after the endocytosis of the cytokine, MIF and JAB1 co-localize in the cell cytoplasm, and MIF inhibits the activation of JNK and its activity on AP-1. Binding of MIF to JAB1 was also associated with reduced JAB1-dependent degradation of p27Kip1. Intriguing new data obtained by Lue et al.  showed that over-expression of JAB1 inhibits MIF-stimulated sustained ERK1/2 phosphorylation, leading to an enhancement of inhibition of phospho-ERK1/2 by higher MIF concentrations. At the same time, JAB1 over-expression did not affect transient ERK activation by MIF, although minimum JAB1 levels were required to maintain MIF-stimulated transient MAPK activation.
MIF in the female reproductive system
Distribution of MIF in female reproductive organs, in the placenta, and fetal membranes
Follicular fluid, granulosa cells
mRNA (RT-PCR), protein (ELISA, WB)
Wada et al. 
mRNA (NB), protein (IHC, WB)
Bove et al. 
Suzuki et al. 
Endometrium (epithelium, stromal cells, endothelium)
mRNA (RT-PCR), protein (ELISA, IHC)
Arcuri et al. 
Endometrium (epithelium, stromal cells, macrophages, T lymphocytes)
mRNA (NB), protein (ELISA, IFS, IHC)
Kats et al. 
Endometrium (epithelium, stroma, endothelium)
Paulesu et al. 
mRNA (NB), protein (IHC, WB)
Wang et al. 
mRNA (NB), Protein (IHC, WB)
Suzuki et al. 
Human (first trimester)
Endometrium (epithelium, decidua)
Arcuri et al. 
Human (first trimester)
Endometrium (NK cells)
mRNA (RT-PCR), protein (ELISA, IFS, IHC)
Arcuri et al. 
Endometrium (epithelium, decidua)
Arcuri et al. 
Endometrium (epithelium, stroma, endothelium)
Paulesu et al. 
Human (first trimester)
mRNA (RT-PCR), protein (ELISA, IHC, WB)
Arcuri et al. 
Zeng et al. 
Arcuri et al. 
Protein (IHC, WB)
Paulesu et al. 
Amnion, chorion, Amniotic fluid
Protein (ELISA, IHC)
Ietta et al. 
Amnion, amniotic fluid
Protein (ELISA, IHC)
Chaiworapongsa et al 
mRNA (RT-PCR), Protein (WB)
Zicari et al. 
Ovary and oviducts
The first evidence of MIF expression in the ovary and the oviduct was obtained by Suzuki et al.  who reported the presence of a MIF transcript in mice. In humans, MIF mRNA was detected in ovarian tissues, and MIF protein was found in the follicular fluid and in the granulosa cells . Insight into the biological functions of ovarian MIF came from studies by Matsuura et al.  who treated mice with an anti-MIF antibody and evaluated the morphological and functional changes associated with neutralization of the cytokine. These authors reported a reduced number of growing follicles and diminished proliferation of granulosa and theca cells, as well as a significant suppression in the number of ovulated oocytes in the ovary of anti-MIF-treated mice. In another study, MIF was detected in large luteal cells of the bovine corpus luteum during the estrous cycle . These authors also reported an increase in mRNA levels after PGF2α-induced luteal regression. Although these observations require further confirmation, they point out the involvement of MIF in key ovarian functions such as follicle growth and ovulation, luteinization and luteal regression.
We demonstrated the presence of MIF in the human endometrium in 2001 . By examining the expression of the cytokine in specimens of uterine mucosa and isolated endometrial glands, we showed that MIF is expressed in the endometrium throughout the menstrual cycle. We detected the protein in specimens from proliferative and secretory phase by Western blot analysis and MIF mRNA in isolated human glands by reverse transcription polymerase chain reaction (RT-PCR). Using immunohistochemistry, we showed that the protein was mainly localized in the epithelial cells, with stronger expression in the surface and in the glandular epithelium. Endometrial stromal cells were homogeneously although weakly stained. Interestingly, the pattern of epithelial cell MIF distribution varied throughout the menstrual cycle. Thus, in proliferative phase specimens, immunostaining was generally homogeneously distributed in the cell cytoplasm, while it was often more prominent in the apical portion of secretory phase epithelial cells, suggesting that these cells can secrete MIF in the luteal phase. We also found immunoreactivity in predecidualized stromal cells and in the vascular endothelium. When we evaluated MIF during the menstrual cycle by enzyme-linked immunosorbent assay (ELISA) on tissue homogenates, we observed no differences in protein levels among proliferative, early secretory, and late secretory endometrium.
Our results were confirmed by Kats et al.  who reported an identical distribution pattern of MIF in specimens of cycling endometrium, with the protein mainly localized in the epithelial cells. By dual immunofluorescence analysis, these authors further identified macrophages and T lymphocytes as MIF-expressing cells in the stroma. When quantification of MIF through the menstrual cycle was carried out, no changes were detected among proliferative, early secretory, and late secretory specimens. However, differences were found when early, mid, and late menstrual phases were examined. Thus, MIF levels increase during the mid-late proliferative phase, reaching a maximum around ovulation, and decrease in the mid-secretory phase before increasing again in late secretory specimens.
The presence and distribution of MIF in the non-human endometrium has been also examined. Paulesu et al.  reported an expression pattern in the pig comparable with that of human, with positive immunostaining present in both epithelial and stromal cells. By contrast, in bovine endometrium, protein and mRNA were detected in the epithelial but not in the stromal cells .
Several studies identified factors controlling MIF expression in the human uterine mucosa. In a series of reports, Akoum et al.  showed that both MIF protein and mRNA increased in cultured human endometrial stromal cells (HESC) exposed to chorionic gonadotropin hormone and to the inflammatory cytokines IL-1β  and TNF-α . In endometrial epithelial cells, factors regulating MIF production seem to act by augmenting secretion of the stored intracellular cytokine more than stimulating de novo synthesis. Schaefer et al. , examining the effect of the TLR3 agonist poly (I:C) on uterine MIF expression, showed that exposure to poly (I:C) resulted in an enhanced apical secretion of MIF not accompanied by a parallel increase in mRNA levels. Similar results were obtained by Wang and Goff  who demonstrated that exposure of cultured bovine endometrial epithelial cells to IFN-τ resulted in increased protein secretion, although the treatment had no effect on MIF mRNA levels.
MIF in pregnancy
To identify factors regulating MIF expression in human decidua, we established a culture system of human decidual cells obtained from first trimester and term specimens. As in vivo, these cells show decidualization-related morphological changes and express elevated levels of plasminogen-activator inhibitor-1 and reduced levels of interstitial collagenase and stromelysin-1 under the influence of progestin. Furthermore, these cultures have a high degree of purity, as they are more than 99% negative for the leukocyte marker CD45, vimentin positive, and cytokeratin negative. In our laboratory, cultured first trimester and term decidua cells were treated with the inflammatory cytokines IL-1β and TNF-α (1 ng/ml) and the levels of MIF protein determined in the culture medium by ELISA and those of mRNA measured by quantitative RT-PCR . In contrast with results obtained with HESC [95, 96], neither cytokine influenced MIF expression in first trimester and term decidual cell cultures. Similar results were obtained when the incubation period with TNF-α and IL-1β was extended to 48 h and the concentration increased up to 10 ng/ml. Intriguingly, no changes in MIF protein levels were observed in cells treated with other inflammatory agents as LPS and IL-6 (unpublished results). However, in these cultured cells, treatment with IL-1β and TNF-α produced an increase in protein and mRNA levels of other chemokines such as MCP-1 , IL-8 , and macrophage-colony stimulating factor (M-CSF) .
Placenta, fetal membranes, and amniotic fluid
MIF was first identified in human placenta during studies on the isolation and identification of sarcolectin binding proteins. Zeng et al.  in 1993 observed among placental factors with high affinity for sarcolectin a protein whose partial sequence and functional characteristics corresponded to those of MIF. Further information on the expression on MIF in the human placenta came from our work. In 1999, we demonstrated by Western blot, RT-PCR analysis, and immunohistochemistry that first trimester human placenta expresses MIF . The protein is present in cytotrophoblast of the chorionic villi, in trophoblastic cell islands, and in the extravillous trophoblast (Fig. 1b). In a more recent report, we examined by immunohistochemistry the presence of MIF in specimens of term placenta . As shown in Fig. 1d, we observed the same distribution pattern, with intense immunostaining localized in the cytotrophoblast of villi and in the extravillous trophoblast (not shown).
Presently, studies on the expression of MIF in non-human placenta are limited to the pig. Similar to observations in humans, Paulesu et al.  detected MIF in villous trophoblast, with no gestational age-related differences evident in the villous protein levels.
Other reports have focused on the expression of MIF in the human fetal membranes and in the amniotic fluid. Ietta et al.  localized MIF in the amnion epithelial cells, in isolated cells of the amnion–chorion mesenchymal layer, and in the chorion layer. These authors detected immunoreactive MIF in the amniotic fluid, with its levels increasing at term and peaking at labor. In a more recent study, Chaiworapongsa et al.  identified MIF in amniotic epithelial cells and fluid across gestation. These authors confirmed an increase in amniotic fluid MIF concentration at term, but they did not observe any significant enhancement at the time of parturition.
Only two studies on factors regulating MIF expression in the placenta and fetal membranes have been published up to date. Ietta et al.  recently described oxygen regulation of MIF in placental explants. In villous explants maintained in hypoxic conditions, they observed an increase in MIF expression and secretion. By combining in situ hybridization and immunohistochemistry, they also showed that MIF increased in extravillous trophoblast cells of placental explants maintained in low-oxygen culture conditions. In vivo measurements revealed higher levels of MIF mRNA and protein at 7–10 weeks of pregnancy compared to 11–12 weeks and term and enhanced MIF mRNA levels in placental tissues from high-altitude pregnancies.
In human fetal membranes, a reciprocal regulatory interaction between MIF and NO seems to exist. In term membranes explants, we showed that incubation with the NO donor sodium nitroprusside enhanced MIF release and that this effect was concentration-dependent. However, treatment of membranes with recombinant MIF induced a significant decrease in NO metabolites and a down-regulation of inducible NO synthase (iNOS) mRNA levels. These results were confirmed by the increase in nitrite release and iNOS expression observed in fetal membranes treated with a neutralizing anti-MIF antibody .
Levels of MIF in the blood of pregnant women have been assessed by three independent groups. The first report is that of Todros et al.  that, in a study on MIF blood levels in preeclampsia, found no significant differences between normal pregnant and non-pregnant women. In contrast, Yamada et al.  and Hristoskova et al.  reported increased MIF blood levels in pregnant women compared to non-pregnant controls. When MIF levels were assessed during gestation, no significant changes between the first, second, and third trimester were reported by these three groups.
MIF in gynecological and obstetrical diseases
The potential involvement of MIF in gynecological and obstetrical diseases has been the subject of several publications. The production of MIF by ectopic endometrium and its possible implication in endometriosis was first demonstrated by Yang et al. . These authors showed that the conditioned medium of an immortalized endometriotic cell line stimulated the proliferation of endothelial cells and identified MIF as one of the protein responsible for this stimulatory effect. Later, this group demonstrated that MIF is expressed in endometriotic lesions , with increased levels of MIF present in the peritoneal fluid  and in the peripheral blood  of women with endometriosis. In the last two studies, the authors reported an association between MIF levels and stage of the disease with higher concentrations of peritoneal fluid MIF in women with active, early stage endometriotic lesions and in patients with infertility, and higher peripheral blood levels in women with more advanced disease. Some of these results were confirmed by Mahutte et al.  who observed significantly higher peritoneal fluid MIF levels in women with endometriosis but failed to detect any association with the stage of the disease. Recently, Akoum et al.  reported higher levels of MIF in total eutopic endometrial protein extracts from endometriosis patients compared to controls, with increased concentrations evident in more advanced disease stages.
The observation that preeclampsia (PE) is associated with alteration of maternal immune response  has suggested a possible relationship with aberrant expression of MIF in women with this syndrome. Todros et al.  observed an increase in MIF blood levels in PE women compared to controls. More recently, however, Hristoskova et al.  reported no changes in PE patients compared to women with uncomplicated pregnancy. Although this discrepancy may be explained by the higher number of patients enrolled in the latter study, additional investigations are required to clarify the existence of alterations in MIF levels in PE.
Correlations between MIF levels, microbial infections of the amniotic cavity, and preterm delivery also have been examined. Chaiworapongsa et al.  reported increased MIF levels in the amniotic fluid of patients with intra-amniotic infection (with preterm labor and preterm premature rupture of membranes) compared with those with sterile amniotic fluid. These authors also reported an increased number of MIF-positive amniotic epithelial cells in cases with histologic chorioamnionitis versus those without infection. Interestingly, while amniotic fluid MIF concentration did not increase during parturition at term, higher concentrations of MIF were observed in patients with sterile amniotic fluid with preterm labor delivering preterm compared with those with preterm labor delivering at term. These results suggest a role for MIF in the host response to intrauterine infection and its potential involvement in the mechanism of preterm parturition.
Only one study has examined the relationship between MIF and miscarriage. Yamada et al.  measured MIF in the blood of women with uncomplicated pregnancy and in those with recurrent abortion. Decreased serum MIF concentrations were observed in women experiencing recurrent miscarriage with normal chromosome karyotype compared to controls, thus, indicating the potential involvement of the cytokine in the etiology of abortion.
MIF and the immune privilege in the uterine mucosa
The mucosa of the pregnant uterus is an immunologically privileged site in which the maternal response to genetically different fetal tissues is effectively restrained and the embryo is allowed to survive, develop, and grow. From the initial invasion of the fetal trophoblast and throughout all of gestation, the trophoblast is in contact with cells of the maternal immune system present in the uterine mucosa. These cells include mediators of innate immunity such as uNK cells and macrophages, and of adaptive immunity, such as T and B lymphocytes. The mechanisms that contribute to maternal tolerance of the fetus remain poorly defined. One likely factor is the lack of expression in the trophoblast cells of human leukocyte antigen (HLA)-A or -B molecules, although they express HLA-C as well as HLA-E and HLA-G, nonclassical MHC class I molecules whose function is still unknown . However, the maternal tolerance of the semiallogenic fetus probably results from the sum of multiple immunological mechanisms active at the maternal–fetal interface. Typical examples are the indoleamine 2,3-dioxygenase-dependent depletion of tryptophan by decidual and trophoblast cells  and the expression by fetal trophoblasts of complement regulatory proteins . Furthermore, different types of suppressor cells are considered involved in pregnancy-related tolerance .
The role of MIF in pregnancy is unclear. MIF does not seem to be essential for the establishment and maintenance of gestation, as MIF−/− mice are fertile, their litter size is normal, and develop normally . Nevertheless, MIF expression at the fetal–maternal interface, together with its demonstrated immunomodulatory effects on leukocytes, indicate that MIF may contribute to the maintenance of the immune privilege in the uterine mucosa by targeting uNK cells and macrophages.
MIF and uterine natural killer cells
The human uterine mucosa of both the cycling and pregnant uterus is largely populated by uNK cells. Although these cells share some properties with the dominant CD56 population of peripheral blood NK, they exhibit specific phenotypic and functional characteristics. Thus, they express CD56 as well as the killer activatory and inhibitory receptors, but lack expression of other typical NK markers such as CD16 or CD57. Moreover, uNK cells express early T cell markers such as CD2 and CD7, integrins, such as CD11a and CD18, and IL-2Rβ. In addition, in contrast to the majority of peripheral NK cells, uNK cells express CD69, an early activation marker . Several changes occur in the uNK cell population during the menstrual cycle and pregnancy. The proportion of uNK cells in the endometrial stromal cell population increases more than threefold between the proliferative phase and the decidua of early pregnancy. Thus, their number increases through the secretory phase, peaking in early gestation. At this point, uNK cells comprise about 75% of bone-marrow-derived cells, with the largest number in the decidua basalis, the region of trophoblast invasion of maternal tissues. From mid-gestation onwards, uNK cells gradually disappear, to be virtually absent in term decidua .
Several studies have addressed the question of the mechanisms underling uNK cell activity in both cycling and pregnant endometrium, and it is generally accepted that cytokines play a critical role. It has been shown that short-term exposure to IL-2 stimulates proliferation and activates uNK cells to kill first trimester cytotrophoblast cells . Verma et al.  have demonstrated that in vitro cytotoxicity and proliferative activity of these cells are both enhanced by IL-15. Moreover, uNK cells express a number of cytokines thought to affect uterine mucosal functions as well as trophoblast invasion and differentiation. Such cytokines include granulocyte M-CSF (GM-CSF), M-CSF, and TNF-α . Interestingly, uNK cells are also known to produce cytokines that normally are not expressed by circulating NK cells, such as leukemia inhibitory factor (LIF) and angiogenic growth factors , supporting the concept that uNK cell-derived cytokines might have specific functions in the uterus.
Although much effort has been directed in laboratories worldwide toward improving the understanding of the physiological role of uNK cells, their specific functions are not clear. One hypothesis considers that uNK cells are involved in the endothelial remodeling that occurs during human implantation and is supported by the presence of a vessel abnormality at implantation sites in mice deficient in NK cells . The tgɛ26 strain mice, lacking both NK and T cells, are fertile, but experience more than 50% fetal loss between 10 and 14 days post-conception. These mice show abnormal decidual vessels in the mesometrial segment of the uterus on days 7 and 8 of gestation. Transplantation of bone marrow from scid donors, resulting in the reconstitution of uNK cells, but not other immune cell types, reverses reproductive abnormalities of the tg26 strain . In humans, the mechanisms involved in spiral artery remodeling are not well established. However, it has been shown that structural changes in placental bed vessels occur early in gestation when uNK cells, but not trophoblast cells, are present in the decidua . Furthermore, a role for uNK cells in vessel remodeling is supported by the observation that these cells are distributed around spiral arteries in the placental bed and that their number decreases after 20 weeks of gestation when vascular changes are generally complete .
A second hypothesis for the functional role of uNK cells is based on their high number in the uterus at the time of implantation and by their intimate contact with the invading placental trophoblast. It has been proposed that uNK cells are important for successful pregnancy by controlling trophoblast invasion of the decidua via cell-mediated cytotoxicity . The activity of NK cells is regulated by the integration of both activating and inhibitory signals. In this context, the main unsolved issue is the outcome of the interaction between uNK class I receptor and trophoblast MHC class I molecules. Initially, it was proposed that the expression of trophoblast HLA molecules (HLA-G, HLA-E, HLA-C) could lead to the inhibition of NK cell lysis . However, recent experimental data have shown that antibodies against trophoblast MHC class I antigens failed to reverse the inhibition of NK cell lysis, suggesting that these molecules are probably less relevant for placental cell survival than previously supposed . Alternatively, locally produced immunological molecules could directly or indirectly control uNK cell activity. Consistent with this hypothesis is the reported inhibition of uNK cell activity by immunoregulatory factors present at the fetal–maternal interface such as IL-4 and transforming growth factor-β (TGF-β) [121, 127].
Effect of an anti-MIF antibody (α-MIF) on the cytolytic activity of uNK cellsa
Percentage of cytolytic activity (mean±SE)
15.5 ± 2.03
18.3 ± 3.57
18.5 ± 2.2
29.0 ± 3.2
p (t test)
Dose-dependent effect of recombinant human MIF (rhMIF) on the cytolytic activity of uNK cellsa
Percentage of cytolytic activity (mean±SE)
14.7 ± 2.86
12.7 ± 2.86
11.7 ± 2.92
Our data also support the involvement uNK cell-derived MIF in other aspects of human gestation where these cells can play a key role such as the modification of maternal arterial physiology and angiogenesis. Thus, uNK cell MIF could play a relevant role as a mediator of neovascularization. Indeed, MIF has been implicated in tumor growth-associated angiogenesis in vivo and in vitro and in the regulation of vascular endothelial cell proliferation in vitro. The possibility that MIF plays a role in specific pathologies of pregnancy involving both an immune and a vascular maladaptation such as preeclampsia is presently under investigation.
MIF and uterine macrophages
Human monocytes and macrophages are mononuclear phagocytic cells that comprise a major part of both the innate and adaptive immune system . In the innate immune response to infectious pathogens, macrophages promote inflammation through their production of such effector molecules as NO, reactive oxygen species, proteolytic enzymes, and inflammatory cytokines [129, 130]. In acquired immunity, macrophages can express HLA class II molecules, which facilitate presentation of exogenous peptides along with co-stimulatory molecules such as B7 (CD80, CD86) to T lymphocytes. Furthermore, macrophages can control immune responses by secreting anti-inflammatory cytokines such as IL-10 and TGF-β1 [129, 130].
Macrophages comprise approximately 20% of decidual leukocytes , and their numbers remain relatively invariant throughout normal gestation . Although the exact role of these cells is still unknown, they are thought to contribute to host protection against infection, but also to the creation of an immunosuppressive environment that favors the immunological tolerance against fetal tissues . In fact, although decidual macrophages are activated in vivo, as indicated by the expression of HLA class II, CD11c, and CD86, they also express B7-H1, ILT3, and other markers involved in immune evasion or that are associated with macrophage alternative activation, a phenotypic state of immune-suppressive activity . In addition, decidual macrophages elicit smaller allogeneic and autologous T cell responses when compared to their blood counterparts . Finally, although decidua macrophages secrete high levels of TNF-α, they produce low levels of IL-1β and secrete anti-inflammatory cytokines such as IL-10 and TGF-β1 [134, 135]. Increasing evidence suggests that an aberrant activation in the maternal immune system may result in pathological conditions such as preeclampsia and intrauterine growth restriction. Studies have shown that compared with normal pregnancy, an excess of decidual macrophages is present around spiral arteries in these conditions [99, 136]. Secretion of TNF-α by these cells would induce apoptosis in extravillous trophoblast, thereby limiting invasion .
The mechanism controlling macrophages trafficking in the placental bed is poorly understood. Several studies have shown that macrophage-recruiting chemokines such as MCP-1, M-CSF, and GM-CSF are produced at the fetal–maternal interface , supporting the concept that an elaborate network controls macrophages recruitment and maintenance throughout pregnancy. MIF targets macrophages and monocytes [44–46]. We speculate that MIF produced at the fetal–maternal interface can play an important role in regulating monocyte/macrophages chemotaxis thus ensuring the proper environment for the establishment of normal fetal–maternal immunological interactions.
Pregnancy represents an immunological paradox in which the developing semiallogenic embryo is tolerated by the maternal immune system. Since the seminal study of Medawar  in 1953, in which the similarity between the fetus and an allogenic graft was first proposed, a number of mechanisms have been considered to explain the state of immunological tolerance that characterizes a successful pregnancy. Among these, the local secretion of immunomodulatory factors is considered crucial for the establishment of a compatible immune balance.
The fetoplacental unit produces an array of immunosuppressive and anti-inflammatory molecules that exert a profound effect on the maternal immune response. Studies performed in our laboratory and in others indicate that MIF is a novel cytokine expressed at the fetal–maternal interface with the potential to participate in the establishing and maintaining of the immune privilege. MIF action can be exerted at least through two mechanisms. First, MIF can modulate cytolytic activity of uNK cells, which are thought to control trophoblast invasion of the decidua. Second, MIF can contribute to the establishment of normal fetal–maternal immunological interactions by controlling macrophages trafficking in the placental bed. Predictably, further studies examining the relationships between MIF and other immunomodulatory factors expressed at the fetal–maternal interface will shed more light on the physiological and pathological functions of this cytokine during pregnancy.
Studies in the authors’ laboratories were supported by grants from the Italian Ministry of Education and Scientific Research, the University of Siena, and by the National Institutes of Health 2 R01HD 33937-05 and 1 R01 HL070004-03. The authors would like to thank Dr. Paolo Toti, Lynn Buchwalder,and Antonietta Carducci for their invaluable help.