1 Introduction

↓1

Mammalian pregnancy is a complex phenomenon that allows that the maternal immune system supports its “semi-allogeneic fetus” during the gestation period without rejecting it. At the same time, the maternal immune system has to be alert and respond to foreign antigens. Since Medawar in 1953 (Medawar, 1953) postulated that the fetus could be considered as an “antigenic foreign body”, many related hypothesis have been proposed in order to explain the paradoxical success of pregnancy from the immunological point of view. Since then, the fetus is usually compared to an “allograft”, and it is thought that mechanisms leading to successful pregnancy are comparable to those leading to transplant acceptance.

1.1 Tolerance at the feto-maternal interface

Tolerance can be defined as a state of antigen–specific immunological unresponsiveness (Tsokos et al., 2001), as a result of inactivation or death of antigen-specific lymphocytes, induced by the exposure to certain antigens (Abbas and Lichtmann in: Cellular and Molecular Immunology, 2003). Tolerance to self antigens is a common feature of the adaptive immune system. Tolerance mechanisms are initiated during fetal life in the thymus for T-lymphocytes and in peripheral lymphoid organs for B-cell clones by a mechanism of clonal deletion. These early mechanisms eliminate the autoreactive T- and B-cell clones. The autoreactive clones that escape clonal deletion can be neutralized by mechanisms of peripheral tolerance. One of the proposed mechanisms of peripheral tolerance is clonal anergy, which is a process that incapacitates or disables autoreactive clones, and these clones loss the ability to respond to stimulation with the corresponding antigen (Tsokos et al., 2001). Tolerance to foreign antigens may be induced under certain conditions of antigen exposure, e.g. it can be induced if protein antigens are administered systemically at high doses without adjuvants or by oral administration of certain protein antigens (Abbas and Lichtmann in: Cellular and Molecular Immunology, 2003).

Pregnancy is normally considered as a natural state of tolerance against foreign antigens, but how this tolerance is achieved is still controversial. In 1953, the Brazilian Sir Peter Medawar initiated the modern immunology of reproduction by asking: “how does the pregnant mother contrive to nourish within itself, for many weeks or months, a foetus that is an antigenically foreign body?” (Medawar, 1953). In this article, three theories were proposed in order to explain the lack of an immunological reaction from the mother against the fetus:

↓2

  1. The anatomical separation of the fetus from mother: according to this theory, it was thought maternal and fetal blood circulations were separated by a barrier impermeable to cells. This theory was proved to be wrong, since it is now known that the fetal-maternal interface is a bi-directional exchange surface. Not only maternal immune cells are present at the feto-maternal interface, but also fetal cells are found in maternal circulation. This phenomenon is called feto-maternal microchimerism, being microchimerism defined as a small nonhost cell population (or DNA quantity) from one individual harboured by another individual (Adams and Nelson, 2006). Many reports indicate that microchimerism persists in mother and child even decades later (Bianchi et al., 1996; Lo et al., 1996; Maloney et al., 1999), and new studies have shown that feto-maternal chimerism occurs even very early in pregnancy (Khosrotehrani et al., 2005; Diploma thesis from Nadja Ahmad), showing that the anatomical separation between fetus and mother consists of a “permeable” barrier.
  2. The antigenic immaturity of the fetus: this theory pointed out that the fetus is “antigenically immature”, not expressing histocompatibility antigens. This theory collapsed very quickly, since fetal skin dendritic cells that are positive for MHC class I but negative for MHC class II are very potent accessory cells in polyclonal T cells responses (Elbe-Bürger et al., 2000).
  3. The immunological inertness of the mother: although this theory is not fully rejected, it still does not explain why immune responses against foreign pathogens are normal in pregnancy. However, this theory gave rise to the concept of active tolerance mechanisms against the fetus. Nowadays, it is considered that the mother achieves a state of tolerance against the fetus, still being able to elicit normal immune responses against infections. This was nicely shown by a work from Tafuri et al. where it was demonstrated that maternal T cells are aware of the presence of paternal antigens during pregnancy, where they acquire a transient state of tolerance specific for paternal antigens (Tafuri et al., 1995).

1.2 Biology of pregnancy

Despite the differences between the placentas of humans and mice, the immunological mechanisms leading to a successful pregnancy in both species are similar, which validates the use of a mouse model in order to study the role of different molecules in pregnancy. These similarities encourage one to use animal models to find alternative therapies to avoid pregnancies complications.

Many steps are involved in the development of a successful pregnancy, such us proper ovulation and receptiveness of the uterus, fertilization, implantation, placentation, as well as a proper immune response allowing the acceptance of the developing fetus.

1.2.1 Ovulation

↓3

Ovulation (i.e., follicular rupture) is a distinct morphological phenomenon that occurs during the transformation of an ovarian follicle into a corpus luteum. The changes that take place in the tissue are pathophysiological in nature since they require acute disruption of dense layers of collagenous tissues. This local damage includes hemorrhage in the vicinity of the lesion on the surface of the ovary (Espey et al., 2004).

The hormonal regulation of ovulation is regulated by different levels of luteinizing hormone (LH), progesterone (P4) and oestrogen (E2) (red and blue in Fig. 1). In humans, ovulation takes place when LH and E2 expression reach their maximum. In mice, this is also accompanied by an augmentation in P4 levels. A schematic representation of hormonal levels during the menstrual cycle is represented in Fig. 1 (from Wang and Dey, 2006).

Fig. 1 : Hormonal changes controlling the menstrual cycle in humans (A) and mice (B)

Fig. taken from Wang and Dey, 2006
LH: luteinizing hormone; P4: progesterone; E2: oestrogen
 

↓4

Although it is normally assumed that ovulation is a relatively simple phenomenon, it is now evident that the ovulatory process is dependent on the expression of numerous genes, mostly associated with acute inflammatory reactions. These genes include immediate-early genes like tumor necrosis factor-α (TNF-α), cyclooxygenase-2 (COX-2), genes involved in steroidogenesis, pro-inflammatory genes like interleukin-1β (IL-1β) and IL-6 and genes related to oxidative stress. Concomitant with the inflammatory cascade, there is a cluster of genes that yield protein products to counteract the oxidative stress that is generated in inflamed tissues (reviewed in Espey et al., 2004).

1.2.2 Fertilization and blastocyst formation

The first step in the development of the mammalian embryo involves the fertilization of the oocyte with the sperm. Fertilization is followed by continued cell division, the establishment of cell polarity and compaction to form a morula, followed by a lineage differentiation to form a blastocyst. This period is called pre-implantatory phase and occurs in the oviduct and uterine lumen. A schematic representation of this phase is depicted in Fig. 2 (Wang and Dey, 2006). If the fertilization occurred in a receptive phase, this stage will be followed by the implantatory phase, at which the blastocyst adheres and implants in the endometrium.

Fig. 2: Preimplantation embryo development in mice

Fig. taken from Wang and Dey, 2006

1.2.3 Implantation

↓5

Implantation is the process in which a receptive endometrium allows the attachment of the blastocyst. The independently developing pre-implantation blastocyst becomes then dependent on the maternal environment for its continued development. The main purpose of implantation is to ensure that trophoblast cells firmly anchor into the endometrial stroma, and successful implantation is most likely a function of both embryonic and maternal processes (Senturk and Arici in: Immunology of pregnancy). It occurs during a specific period of time called the “implantation window”, that is 96-100 h after fertilization of the ovum in the mouse (Finn and Martin, 1967; Tabibzadeh and Babaknia, 1995), and at between day 19 and day 24 of a normal 28 day cycle in humans (Bergh and Navot, 1992). Females experience physiological changes before implantation takes place: ovulation, copulation and fertilization. These events induce changes in the metabolism as well as in the hormone and cytokine balance, preparing the reproductive tract for the subsequent blastocyst implantation. In order for implantation to be successful, the embryo must have reached a proper stage of development and the endometrium must be receptive (Abrahamsohn and Zorn, 1993). This receptiveness results from adequate exposure of the uterus to progesterone and estrogen (Psychoyos, 1976). During this period, the uterine membrane morphology presents pinopodes, which are large cytoplasmic projections from the uterine epithelium. Although their importance for implantation is unclear, they appear to provide a useful morphological correlate of the receptive state (Sharkey, 1998). Regarding the blastocyst, the loss of the zona pellucida is probably the last event that precedes the beginning of implantation in rodents (Abrahamsohn and Zorn, 1993). The general aspects of the implantation process are similar in humans and rodents, starting with erosion of the uterine epithelium and invasion of the uterine mucosa by trophoblast cells (Arvola and Mattsson, 2001). Mouse trophoblasts penetrate the surface epithelium by displacement penetration, where a number of surface epithelial cells detach from their basement membrane and from each other. These cells degenerate and are then phagocytized by trophoblasts, being the trophoblasts exposed to the bare basement membrane (Tabibzadeh and Babaknia, 1995). When the trophoblasts of the blastocyst contact and attach to endometrial epithelium, it induces the local formation of decidua. The female uterus undergoes changes during this period, process called “decidualization”, a reason why the maternal part at the feto-maternal interface receives the name of decidua during pregnancy. A schematic representation of the implantation process is depicted in Fig. 3 (Wang and Dey, 2006).

Fig. 3 : Preimplantation embryo development in mice

Fig. taken from Wang and Dey, 2006. T: trophectoderm-derived trophoblast cells; En: blastocyst, represented by the embryonic endoderm; LE: luminal epithelium; S: stroma; D: decidual cells.

1.2.4 Placentation

The implantation period is followed by the formation of the placenta, a process called placentation. The placentas are very different between species, but they all have in common the existence of two separate circulatory systems (maternal and fetal). The placental barrier or interhemal membrane (Benirschke and Kaufmann, 2001) derives from fetal tissues such as the blastocyst wall (trophoblast), which following fusion with the fetal mesenchyme is called chorion, together with the allantois and the yolk sac. All are complemented by the amnion, which forms the second fetally derived membranous sac surrounding the embryo. A representative scheme is showed in Fig. 4 (from: Benirschke and Kaufmann, 2001):

↓6

Fig. 4 : Representation of the fetal membranes that contribute to the formation of a placenta

Fig. taken from Benirschke and Kaufmann, 2001

The classification of the different placenta types (Benirschke and Kaufmann, 2001) is defined according to the outer shape, the structure of the feto-maternal barrier, the kind of fetomaternal interdigitation, and the materno-fetal blood flow interrelations. Regarding the placental shape, both human and mouse placentas are a single disk-like zone of intimate fetomaternal contact, being called discoidal placenta. This type of placenta possesses the highest degree of feto-maternal interdigitation. Human and mouse placenta have also a similar fetomaternal barrier, which is called hemochorial placenta, which has the highest degree of invasion of the trophoblast, that destroy the maternal vessels completely, with the trophoblastic surface directly facing the maternal blood.

However, mouse and human placenta differ in the type of feto-maternal interdigitation and blood flow interrelations. In humans, the placenta is constituted by villi (villous placenta), which fit into corresponding endometrial cripts or are directly surrounded by maternal blood. Mice placentas have the most effective kind of interdigitation, the labyrinthine pl a centa, where the trophoblast is penetrated by web-like channels that are filled with maternal blood or fetal capillaries. Here, the blood flow interaction is countercurrent, being maternal and fetal capillaries arranged in parallel but with flow in different directions, facilitating passive diffusion. In humans, the blood flow interrelation is called multivillous flow.

↓7

A schematic representation of human feto-maternal interface is shown in Fig. 5 (from: Arvola and Mattsson, 2001). The tree-like structure of the chorionic villi constitutes the part of the human placenta at which the exchange of nutrients and waste occurs. The yolk sac can be observed between the amnion and the chorion in early pregnancy. By the end of the third month of the human pregnancy, the amnion and chorion have fused (Arvola and Mattsson, 2001).

Fig. 5 : schematic drawing of the human placenta and organization of the fetal membranes

Fig. taken from Arvola and Mattsson, 2001

A schematic representation of the mouse placenta is depicted in Fig. 6. Here, the labyrinthine trophoblast constitutes the area of maternofetal interdigitation. In contrast to the human fetus, the mouse fetus is enclosed by the yolk sac (Arvola and Mattsson, 2001).

↓8

Fig. 6 : schematic representation of the mouse placenta and organization of the fetal membranes

Fig. taken from Arvola and Mattsson, 2001

1.2.5 Immune cells at the feto-maternal interface

Throughout the pregnancy, many immune cells are present at the feto-maternal interface. The dominant lymphocytes during human and murine healthy implantations and later on, are pregnancy associated uterine natural killer cells (uNK cells) (Ashkar and Croy, 2001). It is thought that these uNK cells are bone marrow-derived leukocytes that migrate in large numbers to the pregnant uterus. After migration, uNK proliferate, differentiate, and accumulate in large numbers in specific areas of the uterus between days 2.5 and 12 of murine pregnancy (Redline, 2000). After day 12, uNK cells undergo extensive apoptosis and are dramatically decreased in number and activation (Linnemeyer and Pollack, 1994; Croy et al., 1997). These uNK produce IFN-γ at the feto-maternal interface, which contributes to the initiation of vascular modification, decidual integrity and uNK maturation and senescence (Ashkar et al., 2000). In both mice and rats, uNK contain inducible nitric oxide synthase (iNOS) (Hunt et al., 1997).

Macrophages are also present in the uterus during implantation. These cells, together with elevated levels of inflammation-associated cytokines such as IL-1 are associated with early implantation (Hunt, 1989). However, other cytokines and growth factors produced mainly by uterine epithelial cells such as TNF-α, TGF-β and GM-CSF are also necessary for facilitation of implantation. All these facts suggest that a limited inflammatory response seems to be a natural component of successful pregnancies (Hunt et al., 2000).

↓9

Dendritic cells are also present at the feto-maternal interface, and are likely to regulate immune responses to both uterine infections and placental trophoblast cells (Liang and Horuzsko, 2003). A model proposed by Juretic et al. points out that the presence of extravillous cytotrophoblast cells in human pregnancies promotes the recruitment of immature dendritic cells (iDC) and NK cells at the implantation site. It is postulated that, if the stimulation of iDC at implantation site is governed by a strong Th2 response, the function of DC would be tolerogenic, but if they are stimulated under a strong Th1 response, they would become potent antigen presenting cells (Juretic et al., 2004). However, it is known that Th1 cytokines are needed at the time of implantation, making this theory contradictory.

Regarding T cells, it was found that activated γ/δ TCR positive cells are significantly enriched in the decidua as well as in peripheral blood of healthy pregnant women (Szekeres-Bartho et al., 2001). In abortion-prone mice, it was found that TGF-β-producing γ/δ T cells presented suppressor activity dependent on the presence of soluble signals from fetal trophoblast (Clark et al., 1997). Two populations of γ/δ T cells have been described in the murine decidua: an early population producing Th1 cytokines, and a Th2/3 cell subset that appears later (Arck et al., 1999). It was also suggested that γ/δ T cells suppress the anti-fetal immune response through TGF-β production (Suzuki et al., 1995).

1.2.6 Apoptosis at the feto-maternal interface

Apoptosis is a physiologically active mode of cell death that mediates the “safe” deletion of unwanted cells (Gobe and Harmon, 2005). This suicidal pathway is characterized by membrane blebbing, the appearance of highly condensed chromatin and activation of an endonucleolytic process, which leads to the sequential cleavage of genomic DNA. As a result, cells shrink and condense into small membrane-bound “apoptotic bodies”, which are then removed by macrophages (Wu et al., 2001). This removal does not produce pro-inflammatory cytokines; on the contrary, it increases the production of anti-inflammatory cytokines (Sepiashvili et al., 2001).

↓10

At the feto-maternal interface, it is known that apoptosis plays a role in placental remodelling (reviewed in Jerzak and Bischof, 2002). However, changes in apoptosis at the feto-maternal interface may lead to failing pregnancies, as shown by decreased expression of Bcl-2 and increased expression of Bax in the decidua of failing human pregnancies (Lea et al., 1999). However, it is impossible to determine in the human system if low levels of Bcl-2 and high levels of Bax are a consequence or a cause of the abortion.

A review by Mor and Abrahams (Mor and Abrahams, 2003) points out that during implantation, there is a constant turnover necessary for the appropriate growth and function of the placenta. Furthermore, during the third trimester of human pregnancies, there is increased placental apoptosis that may be involved in the process of parturition. In pregnancy complications such as pre-eclampsia and fetal growth restriction, it was found that an insufficient trophoblast invasion is accompanied by a greater incidence of placental apoptosis.

In spontaneous abortion, there is no evidence that apoptosis may play a determinant role. In murine failing pregnancies, it was shown that apoptosis is not augmented in failing pregnancies, probably due to higher levels of the anti-apoptotic molecule Bcl-2 at the fetomaternal interface (Bertoja et al., 2005).

1.2.7 Factors determining the success or failure of pregnancy

↓11

There are many hypotheses trying to explain the onset of a successful or failing pregnancy in murine models as well as in the human situation. Some of these theories are:

  1. Th1/Th2 balance: many scientists have given importance to the Th1/Th2 cytokines in the onset of pregnancy. They proposed that an excess of Th1 cytokines would be deleterious for pregnancy outcome, leading to pregnancy complications such as spontaneous abortion or preeclampsia. This was based on the fact that failing murine and human pregnancies are associated with high Th1 and low Th2 levels (Raghupathy et al., 2000; Zenclussen et al., 2001). Lin as well as Wegmann proposed in 1993 (Lin et al., 1993; Wegmann et al., 1993) that the balance of pro-inflammatory Th1 cytokines and anti-inflammatory Th2 cytokines is critical to normal pregnancy. According to this, a higher level of Th2 cytokines locally at the feto-maternal interface from human or mouse is related to normal pregnancy (Wegmann, 1993; Piccinni et al., 1998; Krishnan et al., 1996; Saito, 2000; Zenclussen et al., 2001; Zenclussen et al., 2003). Although a Th1/Th2 balance is very important throughout the whole pregnancy, and this has been confirmed by several groups, this explanation turned out to be not sufficient for the explanation of the success or failure of pregnancy, since IL-4/IL10 genetically deficient mice develop normal pregnancies (Svensson et al., 2002). Accordingly, mice lacking simoultaneously IL-5, IL-19, IL-13 and IL-4 do not present abnormal pregnancies, suggesting that these cytokines are not essential for fetal survival (Fallon et al., 2002). In humans, the Th1/Th2 paradigm has also been questioned, since lymphocytes, monocytes and granulocytes from normal pregnant patients produce more IL-12 than those from spontaneous abortion patients (Zenclussen et al., 2002).
  2. Protective role of asymmetric IgG antibodies: asymmetric antibodies are nonprecipitating antibodies that posses an asymmetric structure, due to a high manose carbohydrate group present in only one of the two Fab regions of the molecule (Margni and Binaghi, 1988). They do not form insoluble complexes with antigen, they do not fix complement, and they do not induce clearance of the specific antigen. When they combine with an antigen, they act in a competitive way when they are mixed with precipitating antibodies of the same specificity (Margni, 1994). It was shown that, during pregnancy, the production of asymmetric IgG is considerable increased, and the IgG is of maternal origin and fixed to the membrane of the placental cells (Malan Borel et al., 1991). It is proposed that the presence of these antibodies protect the fetus from being rejected, as these antibodies are not able to develop an immune response. It is still a valid theory that is in accordance with data pointing out low levels of IL6 (known to stimulate the production of symmetric normal antibodies), in normal pregnant patients (Zenclussen et al., 2000) and may have a connection with the HLA-G expression by the trophoblast (explained later).
  3. Tryptophan catabolism: tryptophan deprivation might reduce or inhibit some immune cell responses (Thellin et al., 2000). Indoleamine 2,3-dioxygenase (IDO) is an enzymatic protein that catabolizes tryptophan. IDO is synthesized and secreted by the syncytiotrophoblast, and acts by catalysing tryptophan destruction in maternal immune cells that are localised in the placental area. Although Munn et al. first proposed that IDO is essential for the success of pregnancy in the mouse (Munn et al., 1998; Mellor and Munn, 1999), the same authors showed later that mice lacking IDO develop normal pregnancies (Baban et al., 2004), pointing out that IDO might be beneficial but not essential for the development of a normal pregnancy. Furthermore, IDO is not expressed in human placentas until 14 weeks of gestation, 13 weeks after implantation (Kamimura et al., 1991) and in mice until day 8 (Munn et al., 1998; unpublished data from our lab). 
  4. HLA-G expression by the human trophoblasts: until the moment, the expression of MHC class II in trophoblasts could not be demonstrated. Human trophoblast cells express one class Ia molecule (HLA-C) and all three class Ib molecules (HLA-E, -F and –G) (Hunt et al., 2005). These “non-classical” molecules are recognized by the mother´s immune system as demostrated by the fact that anti-HLA paternal antibodies are common in pregnant women. It is tempting to speculate that these antibodies are of asymmetric nature, but unfortunately no reports could be found on this aspect. It has been proposed that the expression of HLA-G by trophoblast cells may regulate immune cells, targeting all of the major immune cell subsets and programming them into an immunosuppressive phenotype. Thus HLA-G has been proposed to be essential to immune privilege in pregnancy (Hunt et al., 2005). However, definitive proof that HLA-G is required remains elusive, since in vivo experiments are difficult to design (Hunt et al., 2005).
  5. Regulatory T cells: In both rodent and human systems, there are many evidences that there is a special population of cells with immunoregulatory activity, which are called regulatory T cells (Treg cells). These cells were first described as the population in charge of neutralizing autoimmune reactive cells in the periphery that escaped clonal selection in thymus (Sakaguchi et al., 1995). There are many attempts to characterize this population of cells and to understand the mechanism by which they exert their regulatory action. So far, there is some evidence that the immune suppression by this subset of cells is dependent on IL-10 and TGFβ (Hara et al., 2001; Kingsley et al., 2002). One of the markers found in many populations of cells with regulatory activity is CD25. However, it is not an exclusive and stable marker for Treg cells, because recently activated T effector cells also express CD25. Additionally, there are some models, in which CD4+CD25- cells also have regulatory function (Wood and Sakaguchi, 2003). Nevertheless, the sorting of CD25-expressing cells is still a useful way of enriching Treg cells of this subset, and there are many evidence that this CD4+CD25+ population plays a regulatory role both in vitro and in vivo (Gregori et al., 2001; Chai et al., 2002; Kingsley et al., 2002). There are many other possible candidates as markers for the Treg population, such as Cytotoxic Tlymphocyte antigen-4 (CTLA-4), glucocorticoid-induced tumour necrosis factor (GITR), CD122, CD103 and foxp3 as a possible specific molecular marker (Wood and Sakaguchi, 2003). A novel molecular marker may be also Neuropilin-1 (Bruder et al., 2004).
    In pregnancy, it was first described by Heikkinen et al. in human and by Aluvihare et al. in murine pregnancies (Heikkinen et al., 2004; Aluvihare et al., 2004) that the onset of pregnancy is dependent on the presence of this population of cells. Data from our group showed for the first time that mice developing spontaneous abortion have less number of naturally occuring CD4+CD25+ Treg cells (Zenclussen et al., 2005). Moreover, in abortionprone females receiving Treg cells arising from normal pregnant mice, fetal rejection could be completely reverted, confirming a very important role of Tregs during pregnancy (Zenclussen et al., 2005b). In murine pregnancies, it is thought that Tregs induce a privileged tolerant microenvironment at the fetal-maternal interface (Zenclussen et al., 2006). This microenvironment would be characterized by augmented TGF-β, leukaemia inhibitory factor (LIF), and heme oxygenase-1 (Zenclussen et al., 2006a). In humans, it has been reported that Treg mediate a potent inhibition of CD4+CD25- T cells, and that this effect requires cell to cell contact (Sasaki et al., 2004). However, it is not described in the literature whether they induce the expression of any of these molecules at the feto-maternal interface.
  6. Leukemia inhibitory factor (LIF): LIF is a pleiotropic cytokine of the IL-6 family and has different biological actions in various tissue systems (Hilton and Gough, 1991). Nowadays, there is an accumulation of evidence of the importance of LIF in different stages of reproduction. In a mice model, LIF has been shown to be one of the essential cytokines for implantation, since mice lacking LIF could produce normal embryos, but the embryos failed to implant. Interestingly, when embryos from LIF-deficient mice were transferred into the uteri of wild-type mice, normal implantation occurred, confirming that LIF produced by endometrium is critical for murine implantation (Stewart et al., 1992). LIF seems also to affect sperm survival rates and motility in the fallopian tube, thus playing also a role even in the fertilization process (Attar et al., 2003). There is also evidence suggesting that LIF plays an important role in human reproduction, and is considered as a possible cause of unexplained infertility and multiple failures in implantation (Hambartsoumian, 1998; Steck et al., 2004).

1.2.8 Spontaneous abortion

Between the possible pregnancy complications that may occur during mammalian gestation, spontaneous abortion is one of the most common. It occurs between the first and third month of pregnancy, and ends with the rejection of the embryo. About 20% of all pregnancies end in spontaneous abortion (Essentials of Clinical Immunology), whereas two-thirds of fetuses are lost even before the woman realizes that she is pregnant, reason why it is thought that the percentage of women suffering spontaneous abortion is even higher.

↓12

Spontaneous abortion can be caused by genetic disorders, infections, endocrine abnormalities, or autoimmune states as i.e. antiphospholipid syndrome (Essentials of Clinical Immunology). When the cause of abortion does not respond to anatomical, genetical or endocrinological causes, it is very probable that the cause of abortion is of immunological origin. Thus, when a woman suffers from three or more consecutive spontaneous abortions and all other factors have been discharged, the causes of the losses are thought to be immunological. In these cases, the immune system of the pregnant woman does not adapt to the situation and rejects the growing embryo.

1.2.8.1 The CBA/J x DBA/2J model of spontaneous abortion

In the mouse, a useful model of abortion was described for the first time by Clark and coworkers in 1980 and further characterized by Chaouat and co-workers in 1988. In this model of immunological abortion, the combination of CBA/J females with DBA/2J males leads to spontaneous abortion (Clark et al., 1980; Chaouat et al., 1988). The mentioned mice combination develops between 20 and 30% of abortion, and these immunological rejections occur spontaneously without being caused by any treatment of the pregnant mice. Because CBA/J females bear H2k antigens and DBA/2J males bear H2d antigens, this model employs mice that differ in the antigens of the major histocompatibility complex. Different to the transplant situation, this does not lead to rejection as it is even thought that an allostimulus is necessary for pregnancy to occur. Interestingly, when mating the same females with a BALB/c male, also bearing H2d antigens like the DBA/2J males, these females undergo normal pregnancies. This fact excludes the possibility that the rejection of the fetus may be due to differences in the major histocompatibility antigens. 

1.3 Heme Oxygenase-1 (HO-1)

In 1964, Wise and Drabkin (Wise and Drabkin, 1964) first described an enzymatic reaction that converted heme to biliverdin and carbon monoxide. They described that this reaction required NAD and ATP, and that, interestingly, the enzyme source was the light mitochondrial fraction of hemophagous organ of the dog placenta. This report was followed by reports of Tenhunen and co-workers (Tenhunen et al., 1968; Tenhunen et al., 1969; Tenhunen et al., 1972) who described a bile-pigment producing system located in the microsomal fraction of the rat liver, and they named it “microsomal heme oxygenase”. This enzyme was described to be responsible for the cleavage of heme at the α-methene bridge resulting in the formation of biliverdin and CO, being biliverdin rapidly reduced to bilirubin by NADPH-dependent biliverdin reductase. A further characterization of the enzyme (Tenhunen et al., 1969) allowed the description of the enzymatic reaction as being dependent on the presence of molecular oxygen and of NADPH or an operational NADPH-generating system. It was also suggested that cytochrome P-450 would play an essential role in the reaction. Further studies (Maines and Kappas, 1974; Yoshida et al., 1974; Maines and Kappas, 1975) led to the identification of the enzyme as a distinct microsomal entity in which its activity does not require cytochrome P-450.

↓13

Nowadays it is widely know that Heme Oxygenase (HO), encoded by the Hmox1 gene, is the enzyme catalyzing the first and rate limiting step in the degradation of heme, to yield equimolecular quantities of biliverdin, CO, and free iron. Biliverdin is then converted to bilirubin via the action of biliverdin reductase, and free iron is sequestered into ferritin, reaction that is schematised in Fig. 7 (Ryter et al., 2002).

Fig. 7 : Scheme of the reaction of heme oxygenase-1

Fig. taken from Ryter et al., 2002

To date, three different mammalian isoforms have been identified: HO-1, HO-2 and HO-3 (Montellano, 2000; Otterbein and Choi, 2000; Morse and Choi, 2002). HO-1, also known as heat-shock protein (HSP) 32, is very sensitive to several stimuli and agents that cause oxidative stress and pathological conditions, such as heat shock, ischemia, radiation, hypoxia, hyperoxia (Maines, 1997), cytokines (IL-1, IL-6, or TNF-α), heavy metals and nitric oxide. The HO-2 isoform (Maines et al., 1986; Trakshel et al., 1986) is normally referred as constitutively expressed, and is not inducible by the agents capable of inducing HO1. The list of inducers of HO-2 is limited and include developmental factors, adrenal glucocorticoids, opiates and possibly nitric oxide (reviewed in Maines and Panahian in: Hypoxia: from genes to the bedside). In human pregnancy, different levels of HO-2 were found between placentas of normal pregnant or abortion patients (Zenclussen et al., 2003b) as well as in patients with preeclampsia and fetal growth restriction (Barber et al., 2001), suggesting HO-2 may be regulable in human placenta. The third isozyme, HO3, was isolated from rat tissues and it presents 90% of aminoacid homology with HO-2 (McCoubrey et al., 1997) but acts as a less efficient heme catalyst (Morse and Choi, 2002).

1.3.1 Tissue distribution and sub-cellular localization

↓14

Heme oxygenase is widely expressed in different tissues in mammals, as well as in various species, including unicellular organisms (Maines in: Heme Oxygenase, Clinical Applications and Functions, 1992). Several reports indicate that at least in the rat, HO-1 specific activity is being the highest in the spleen, followed by bone marrow, liver, brain, kidney, and lung, in decreasing order (Tenhunen et al., 1969). In most other tissues not directly involved in erythrocyte turnover or hemoglobin metabolism, low detectable levels of HO-1 are detectable under basal conditions but responds to rapid transcriptional activation by diverse chemical and physical stimuli (Ryter et al., 2006). HO-2 highest activity occurs in the testes, being also expressed in brain and central nervous system, vasculature, liver, kidney and gut (Maines in: Heme Oxygenase: Clinical Applications and Functions, 1992; Maines, 1986; Maines, 1997).

In cells, HO-1 enzymes have been characterized as endoplasmic reticulum (ER)-associated proteins due to the abundant detection of HO-1 activity in microsomal (104,000 g) fractions (Ryter et al., 2006). Both HO-1 and HO-2 contain a COOH-terminal hydrophobic domain that suggests a general membrane compartmentalization (Shibahara et al., 1985). Other reports have also described the presence of HO-1 in mitochondria (Converso et al., 2006; Slebos et al., 2007) and in caveolae (Kim et al., 2004). However, recent reports showed that HO-1 can also localize to the nucleus and activate transcription factors important in oxidative stress (Lin et al., 2007), and that this form of HO-1 seems to lack the C terminus (Lin et al., 2007). 

1.3.2 Relationship between the HO and the NO system

There is some similarity between the systems that generate the gaseous heme ligands carbon monoxide (CO) and nitric oxide (NO). Both the HO and the nitric oxide synthase (NOS) systems have inducible and constitutive forms. The inducible form of the synthase, iNOS, is responsive to some of the stimuli activating HO-1, such as bacterial endotoxins, cytokines, and reactive oxygen intermediates (Nathan, 1992; North et al., 1996). HO-1 and iNOS are highly induced in inflammatory cells, neutrophils, polymorphonuclear and mononuclear cells (Nath et al., 1992; Willis et al., 1996), whereas HO-2 and neuronal and endothelial forms of NOS are similar in the sense that both are regulated by adrenal glucocorticoids (Weber et al., 1994; Maines et al., 1996), being HO2 upregulated and NOS down-regulated (Weber et al., 1994). CO and NO share affinity for the heme molecule, but, unlike NO, which is a free radical and can react with oxygen-derived radicals, CO is not a free radical and cannot produce tissue damage associated with inflammation and cytotoxicity to invading organisms that is caused by free radical species (Maines, 1997).

↓15

Induction of HO-1 is likely to modulate NO production, whereas NO has been shown to both inhibit and activate HO activity depending on the conditions (reviewed in Maines, 1997). The proposed regulatory interactions between the HO and the NO system are shown in Fig. 8 (from Maines, 1997).

Fig. 8 : schematic representation between the HO and the NOS systems

Fig. adopted from Maines, 1997.

1.3.3 Beneficial effects of HO-1 in different fields of medicine

The most studied isoform of the Heme Oxygenase is the HO-1, a 32 kDa inducible isoform, being its beneficial effects described for many diseases such as atherosclerosis, pulmonary, cardiovascular and renal diseases, as well as in organ transplantation (Morse and Choi, 2002).

↓16

The activity of this enzyme is of great importance in cells to protect themselves against oxidative injury. Additionally, HO has been shown to have anti-inflammatory, anti-apoptotic and anti-proliferative effects.

There are many possible explanations to account for the beneficial effects of HO-1. One of them, proposed by Stocker in 1990, suggests that an augmentation of HO-1 represent an anti-oxidant defence operating at two different stages simultaneously:

  1. it decreases the levels of potential pro-oxidants such as heme and heme proteins
  2. it increases the tissue concentration of antioxidatively active bile pigments

↓17

As it can be seen in Fig. 9 (from Stocker, 1990), the heme molecule serves as a prostetic group of various hemoproteins such as those that transport oxygen or electrons, activate oxygen, or degrade peroxides, which favours a pro-oxidant state. Besides, free heme is capable of catalyzing oxygen radical reactions and is therefore a potent pro-oxidant (Stocker, 1990). Free heme not only stimulates lipid peroxidation (Tappel, 1953), it also causes oxidative damage to DNA and protein (Aft and Müller, 1983, Aft and Müller, 1984) and is cytotoxic to various cell types (Linn and Everse, 1987). Moreover, the iron in the core of the heme structure becomes available to participate in the generation of free radicals (Soares et al., 2001).

Fig. 9 : Pathway of heme metabolism

Fig. taken from Stocker, 1990.

It has been shown that an increase in HO activity is generally accompanied by a decrease in the cellular concentrations of total and free heme (Maines and Kappas, 1974; Granick, 1975; Maines, 1984) and the lowering of this potent prooxidant may already explain some of the beneficial effects associated with HO-1.

↓18

An additional protective effect of the HO-1 reaction is related with the accumulation of bilirubin caused by the rapid conversion of biliverdin (product of the HO-1 reaction) to bilirubin by biliverdin reductase. Bilirubin is considered a physiological important antioxidant and will be explained later.

Besides the fact that HO-1 has beneficial effects by decreasing the heme levels, it is also thought that its protective effects are due to the effect of the three products of the catalyzed reaction (CO, biliverdin, free iron).

Beneficial effects of HO-1 have been described in different fields of medicine, such as transplantation (Soares et al., 1998; Coito et al., 2002; Tullius et al., 2002; Braudeau et al., 2004), atherosclerosis (Juan et al., 2001), sepsis (Fujii et al., 2003), autoimmune neuroinflammation (Chora et al., 2007) and infections such as malaria (Pamplona et al., 2007). In pregnancy, works from our group (Sollwedel et al., 2005) as well as work obtained and presented in this thesis clearly show that up-regulation of HO-1 have beneficial effects in pregnancy, at least in an experimental model(Zenclussen ML et al., 2006).

1.3.4 Products of the HO enzymatic reaction and their effects

1.3.4.1 Carbon monoxide (CO)

↓19

CO is a gaseous second messenger that arises in biological systems during the oxidative catabolism of heme by HO. The endogenous production of CO was described even before the characterization of the HO enzyme (Wise and Drabkin, 1964; Coburn et al., 1967), although it was considered for many years as a waste product with toxic effects but no specific functions. Nowadays, besides its toxic effects at high doses, it is known to mediate salutary effects when applied at low doses. CO is a signal molecule for the generation of cGMP in biological systems (Maines, 1997), and can regulate vasomotor tone as well as neurotransmission, having a regulatory function that may indeed account for the anti-inflammatory effects of HO1 (Otterbein and Choi, 2000). CO is also known to inhibit platelet aggregation and endothelin-1 production by endothelial cells (Buelow et al., 2001). Besides, anti-apoptotic and anti-proliferative effects were described for CO (Brouard et al., 2000; Pae et al., 2004). Moreover, CO has been shown to affect several intracellular signalling pathways, including guanylate cyclase and the mitogen-activated protein kinases (MAPK) (Otterbein et al., 2000), and these pathways mediate, in part, the known vasoregulatory, anti-inflammatory, antiapoptotic and anti-proliferative effects of this gas (Ryter and Otterbein, 2004). Recently, it has been reported that CO alone leads to the generation of reactive oxygen species (ROS) which play a major role as signalling molecules to upregulate peroxisome proliferator-activated receptor-γ (PPARγ), and that PPARγ accounts for the anti-inflammatory effects of CO in vitro and in vivo (Bilban et al., 2006).

CO was shown to mimic the effects the HO-1 in many models (Sass et al., 2003; Akamatsu et al., 2004; Chora et al., 2007; Pamplona et al., 2007), even administered during an ongoing disease process (Hegazi et al., 2005). However, the effects of CO are lost in hmox1 -/- macrophages, suggesting the need for HO-1 up-regulation for the inhibitory effects of exogenous CO (Hegazi et al., 2005). It has been suggested that CO initiates an amplification mechanism that results in the production of another product of HO1, which is the real mediator of the therapeutic effect (Bach, 2005). Alternatively, CO might mediate all of the beneficial effects, but the amplification steps generated by each of the products of HO-1 degradation are essential to generate enough CO at the site of injury to manifest the those effect (Bach, 2006). An interesting theory was proposed by M. Soares at the 5th International Congress Heme Oxygenase 2007, by which it is postulated that complications that are different in nature and outcomes such as artherosclerosis, autoimmune neuroinflammation or malaria present a common feature: the presence of exacerbated inflammation. Interestingly, he proposed that these complications can be reverted by counteracting the exacerbated inflammation by the application of CO, known to have anti-inflammatory properties. Although this theory is based so far only in experimental models, it could be shown e.g. in the case of malaria, that this carbon monoxide may act not only by up-regulating HO-1, but also by binding to free cell hemoglobin, preventing hemoglobin oxidation and the generation of free heme, a potent oxidant that would trigger the pathogenesis of experimental cerebral malaria (Pamplona et al., 2007).

1.3.4.2 Iron (Fe2+)

The iron released from heme by HO potentially enters a pool of “labile” or “chelatable” iron, where it may be available for cellular processes that depend on iron (Ryter and Tyrrel, 2000). Normally, the free iron is rapidly sequestered into the protein ferritin, and such sequestration can itself lower the pro-oxidant state of the cell by removing the free iron (Balla et al., 1992; Otterbein and Choi, 2000), avoiding iron-dependent oxidative stress. Ferritin-mediated iron chelation inhibits cell-cycle progression, leukocyte migration and fibroblast and endothelial cell apoptosis (Buellow et al., 2001). Ferritin has been implied as a cytoprotective molecule in different in vitro models (Balla et al., 1992; Lin and Girotti, 1998), and has been proposed as a contributory mechanism underlying HO-dependent protection (Ryter et al., 2006). In vivo, it has been shown that the over-expression of H-ferritin protects rat livers from ischemia/reperfusion injury and prevents hepatocellular damage upon transplantation and that its protective effect is related with inhibition of endothelial cells and hepatocyte apoptosis in v i tro and in vivo (Berberat et al., 2003).

1.3.4.3 Biliverdin and bilirubin

↓20

Biliverdin, the first product of HO-catalyzed heme cleavage, is enzymatically reduced by biliverdin reductase (BVR) to produce bilirubin. Under normal conditions, biliverdin and bilirubin are processed for rapid elimination (Ryter et al., 2006). Bilirubin is known as a waste product, but it was also shown to be a potent anti-oxidant (Maines, 1997). It has been seen that both biliverdin and bilirubin can act as potent in vitro anti-oxidants with possible physiological implications (Stocker et al., 1987; Stocker and Peterhans, 1989; Neuzil and Stocker, 1993). Biliverdin has been also shown to be anti-inflammatory (Vachharajani et al., 2000), and to induce tolerance to cardiac allografts (Yamashita et al., 2004). In ischemia/reperfusion injury, both biliverdin and bilirubin have been shown to be protective in vivo (Nakao et al., 2005; Adin et al., 2005).

1.3.5 Heme oxygenase-1 deficiency

So far, only one human case of total heme oxygenase-1 deficiency was described (Yachie et al., 1999). A schematic representation of the observations done in the case of human Hmox1 deficiency is summarized on Fig. 10.

Briefly, the male patient showed growth retardation, in addition to fever, rash, hepatomegaly, arthralgia without swelling and generalized lymphadenopathy. He also presented fragmented red blood cells in serum, together with giant platelets and dysmorphic monocytes in peripheral blood, with hematuria and proteinuria constantly present. He showed generalized inflammation, disturbances in the coagulation/fibrinolysis system, nephropathy, vascular endothelial cell injury, and asplenia (reviewed in Koizumi, 2007). The patient unfortunately died at the age of 6, but the discovery of this case was of great importance for understanding the importance of HO-1 in the well functioning of the immune system as well as in the metabolism in general. Interestingly, the mother (who presented a heterozygous mutation for HO-1) presented previously two intrauterine fetal deaths (abortions), suggesting that HO-1 plays an important role also in human pregnancies.

↓21

Fig. 10 : Schematic representation of the pathophysiology of Hmox1 deficiency in humans

Adapted from Koizumi, 2007.

Mice deficient for heme oxygenase-1 (Hmox1 -/- mice) were first generated and described by Poss and Tonegawa in 1997, in the (129/Sv x C57BL/6) hybrid strain background. These mice were slightly smaller than Hmox1 +/+ or Hmox1 +/- littermates from birth to early adulthood, but were otherwise indistinguishable (Poss and Tonegawa, 1997). However, after 25 weeks of age they died prematurely or became less active. Besides, after 20 weeks of age Hmox1 -/- mice presented hypoferremia and anemia with accumulation of tissue iron. These observations led to the authors to propose that these mice have a defect in iron reutilization. Furthermore, these mice presented a progressive chronic inflammatory disease, demonstrated by enlarged spleens and lymph nodes, high splenic and lymph node CD4+:CD8+ T-cell ratios with numerous activated CD4+ T cells and hepatic inflammatory cell infiltrates (Poss and Tonegawa, 1997). Regarding the maintenance of the colony it was described that matings between heterozygous mice for Hmox1 did not yield the expected mendelian ratio, being it only 20% of the expected Hmox1 -/- mice (8% instead of 25%). Even more interesting is the fact that no viable litters were obtained when mating Hmox1 -/- females and males, suggesting an essential role of HO-1 in pregnancy.

Another colony of Hmox1 deficient mice was generated by Yet et al. in 1999, in a (129Sv x BALB/c) mixed genetic background. Consistent with the data of Poss and Tonegawa, the matings of Hmox1 +/- did also not yield the expected mendelian ratio, which led the authors to propose that homozygous mutants present partial embryonic or neonatal lethality (Yet et al., 1999). These mice also present enlarged spleen, but are otherwise macroscopically indistinguishable from Hmox1 +/+ or Hmox1 +/- mice (personal observations).

1.3.6 Role of HO-1 in tolerance

↓22

A very interesting paper published in 2005 by Fritz H. Bach (Bach, 2005) points out that many molecules act through a so called “HO-1 amplification funnel”. According to this theory, the beneficial effects associated to molecules such as IL-10, rapamycin, 15-PGJ2, or aspirin, are in fact due to the up-regulation of HO-1 by these molecules. By definition, a given molecule would only function via the “amplification funnel” if both of two conditions obtain:

  1. the molecule functions only when HO-1 is present and induced, and a product of HO-1 can mediate the functions ascribed to the given molecule in the absence of that other molecule.
  2. HO-1 can amplify the therapeutic effects of the other molecule.

Fig. 11 : HO-1 therapeutic amplification funnel

Fig. taken from Bach, 2005.

↓23

There are many molecules that appear to function via the HO-1 funnel, and they are represented in Fig. 11 (Bach, 2005). Although these molecules are not related, they all require the presence of HO-1 to exert its action, and their actions can be mimicked by HO-1 or its products. The effects obtained in all the cases are related to the achievement of a tolerant state.

1.3.6.1 HO-1 in organ transplantation

Tolerance plays a central role in organ transplantation, and in this regard, HO-1 has been shown to play an important role. Almost all transplanted organs suffer certain degree of ischemia/reperfusion injury, characterized by primary microcirculatory flow disturbances caused by the production of oxygen free radicals and cytokine-mediated inflammatory damage (Katori et al., 2002). Besides, after organ transplantation, large amounts of free hemoproteins are released, and it is known that acute exposure to heme is a highly cytotoxic event. In these cases, there are mechanisms to protect the cells against these damages, and in these processes the HO-1 molecule plays a central role. In this regard, many works have already shown that the upregulation of HO-1 is able to avoid graft rejection. The use of HO-1 inducers such as cobalt protoporphyrin (CoPPIX) has shown beneficial effects through the upregulation of the HO-1 expression in a steatotic rat liver model of ex vivo cold ischemia/reperfusion injury (Amersi et al., 1999), and in a mouse model of transplantation of islets of Langerhans (Pileggi et al., 2001). Additionally, in a model where animals tolerated their transplants by means of a treatment with anti-CD40L antibody plus donor-specific transfusion, tolerance could not be achieved in Hmox1 -/- mice or in mice treated with the HO1 inhibitor zinc protoporphyrin (ZnPPIX) (Yamashita et al., 2006), suggesting that HO-1 is essential for the tolerance to transplanted organs.

1.3.6.2 HO-1 in pregnancy

Considering the similarities between an allotransplant and a fetus from an antigenically point of view, it is to expect that HO-1 may play an important role in favouring the tolerance against the allogeneic fetus. Although little was known in the last years about the role of HO1 in pregnancy, many recent studies point out a key role of this enzyme. So far, we could show that mice undergoing abortion presented down regulated levels of HO-1 and HO-2 at the feto maternal interface when compared to normal pregnant mice (Zenclussen et al., 2005). Accordingly, human miscarriage was associated with diminished placental HOs levels (Barber et al., 2001; Zenclussen et al., 2003b). Besides, Kreiser et al. reported that the injection of an adenoviral vector containing HO-1 into 15-day pregnant rats lead to increased pup weight (Kreiser et al. , 2002). As previously explained, mice lacking HO-1 (Hmox1 -/- mice) present a variety of disease symptoms, including anemia, abnormal iron-loading, and chronic inflammation, showing Hmox1 -/- embryos and mice abnormalities like impaired cellular stress response and abnormal responses to endotoxin (Poss and Tonegawa, 1997). Interestingly, Hmox1 -/- mating pairs did not yield viable litters (Poss and Tonegawa, 1997). These data reflex the fact that no successful pregnancy can be achieved without HO-1, pointing out a key role of this enzyme in pregnancy.

↓24

The hypothetical scenario depicted in Fig. 12 (adapted from Zenclussen et al., 2002b) points out that during pregnancy, specially due to the inflammatory nature of blastocyst implantation, there are huge amounts of free heme that need to be degraded.

Fig. 12 : Hypothetical scenario of the role of HOs in pregnancy and Th1-mediated abortion.

Adapted from Zenclussen et al., 2002b.

In a spontaneous abortion situation (B), HO-1 would be expressed but not at sufficient levels, thus being insufficient to degrade the free heme present in the system. The accumulation of free heme, a potent prooxidant, would lead to damage of endothelial cells and trophoblasts. In addition, it would enhance the expression of adhesion molecules, which would allow further trafficking of Th1 lymphocytes into the feto-maternal interface, known to be pregnancy-deleterious. In a normal pregnancy situation (A), heme oxygenase would be expressed in sufficient amounts and would be thus able to degrade all the free heme present at the feto-maternal interface, being the previously described cascade avoided.

1.3.7 HO-1 and regulatory T cells (Treg)

↓25

Some works point out that there might be a relationship between HO-1 and Tregs. In 2003, Pae found that human CD4+CD25+ cells from blood constitutively express HO-1, whereas CD4+CD25- did not (Pae et al., 2003). Later in 2005, it was demonstrated that HO-1 expression could be induced by Foxp3 gene transfection, and that HO-1 was involved in Foxp3-mediated immune suppression (Choi et al., 2005).

In pregnancy, as already mentioned, a link between Tregs and HO-1 was also found, as mice receiving Tregs to be rescued from spontaneous abortion showed augmented levels of HO-1 at the feto-maternal interface when compared to mice developing spontaneous abortion (Zenclussen et al., 2006). Additionally, the upregulation of HO-1 by Cobalt Protoporphyrin (CoPPIX) led to augmented levels of mRNA of Neuropilin-1, a suggested marker of Treg (Sollwedel et al., 2005). Interestingly, in a work where tolerance to islet allografts is achieved by inducing HO-1 has shown that if Treg are depleted prior to transplantation, this tolerance is no longer achieved (Lee et al., 2007). All these data suggest a relationship between both systems. However, it has still to be proven that there is in fact a functional important feedback between both systems. Some other authors point out that there is no such relationship between HO-1 and Tregs. For example, the work of Zelenay et al. (Zelenay et al., 2007) showed that the frequency of CD25+ or foxp3+ Tregs does not differ between Hmox1 -/-  and Hmox1 +/+ mice, and that Treg isolated from these animals are equally efficient in controlling the proliferation in vitro and the expansion in vivo of CD4+CD25- T cells.

1.4 Gene Therapy

The basic concept of gene therapy involves the introduction of a foreign gene with the purpose of correcting genetic diseases or in order to deliver new therapeutic functions to target cells (Walther and Stein, 1996). However, the use of gene therapy for clinical trials is still controversial and facing many obstacles before the use of DNA as a drug becomes commonplace (Thyagarajan et al., 2001). Nevertheless, the use of gene therapy is still of great use in animal models especially in helping to understand the function of certain molecules in different experimental models.

1.4.1 Viral expression systems used in gene therapy

↓26

One of the most important aspects of gene therapy is the safe and efficient delivery of DNA, and for that, the choice of the vehicle for delivery. Any ideal vector used as DNA delivery vehicle should have the properties of easy and sustained production and should be immunologically inert. It is desirable that it delivers only in certain cell types and that it can infect both dividing and non dividing cells. An ideal vector should have no size limit regarding the genetic material it can deliver, and in some cases a site-specific integration into the chromosome of the target cell is convenient (Somia and Verma, 2000). There are at the time no ideal vector fullfilling all these properties but it is important that the chosen vector has so many of them as possible.

Although there exist non-viral systems to introduce foreign DNA in tissues or cells, they have as disadvantage an inefficient gene transfer and a transient expression of the foreign gene (Somia and Verma, 2000; Wood and Prior, 2001). Therefore, viral vectors still represent the most suitable vehicles for gene therapy trials.

Viral vectors are obtained by replacing genetic components from the original virus with the therapeutic gene. Depending on the integration of the genetic information into the host genome or not, these vectors are divided in two categories.

↓27

Some of the viral vectors that are most used in gene therapy trials are described below:

a) Retroviral vectors: these vectors are the most used in gene therapy trials and derive from Moloney murine leukaemia virus (MoMuLV). Retroviral vectors derived from MoMuLV promote the efficient transfer of genes into a variety of cell types, and cause no detectable harm as they enter their target cells. The retroviral nucleic acid becomes integrated into chromosomal DNA, ensuring its long-term persistence and stable transmission to all future progeny of the transduced cell (Boulikas, 1998). Retroviruses have three essential genes which are usually provided separately in packaging cells, in order to make them deficient of replication in the absence of the packaging cell line, being therefore safer. These genes are the gene gag, which encodes for viral structural proteins, the gene pol encoding for reverse transcriptase/integrase, and the gene env, which encodes viral envelope glycoprotein. The gag and pol genes are separated from the env gene, making the regeneration of a replication competent virus unlikely (Somia and Verma, 2000). A schematic representation of the production of retroviral vectors is schematized in Fig. 13 (Somia and Verma. 2000).

↓28

Fig. 13 : Retrovirus-based vectors

Adapted from Somia and Verma, 2000. a) The retroviral genome contains normally the gag, pol and env genes. The Ψ is the packaging sequence and is recognized by the viral proteins for packaging. b) The vector genome, where the gag, pol and env are replaced by the therapeutic gene. c) The packaging cell. The vector genome, by virtue of the Ψ sequence are encapsulated along with the Pol and Gag proteins and assembled under the membrane. The virus buds off from the packaging cell, where it is able to infect other cells but not to replicate.

Other advantage of the retroviral vectors, is that by changing the envelope protein it is possible to change the target cells from ecotropic (infecting only rodent cells) to xenotropic (infecting most mammalian cells except rodent cells), amphotropic (infecting all mammalian cells) and pantropic (infecting various species) (Danos and Mulligan, 1988; Markowitz et al., 1988; Burns et al., 1993). One of the disadvantages of these vectors is that they can infect only dividing cells, which constitutes the main limitation of their use. Other mentionable disadvantage is the low viral titers obtained when working with these viruses, and they may be too low to achieve therapeutic levels of gene expression (Boulikas, 1998). Despite these disadvantages of the retroviral vectors, the already mentioned advantages make them one of the most used vectors in gene therapy trials.

b) Lentiviral vectors: they are part of the retrovirus family, and their advantage is that they are able to infect non-dividing cells (Somia and Verma, 2000). One disadvantage that they have is the possibility of recombination and generation of infectious HIV and the non-specific integration in the chromosome. Although there are many trials trying to overcome this disadvantage, their use remains very limited. Additionally, they present the disadvantage of low titer production and decreasing levels of transgene expression over time (Buchschader and Wong-Stall, 2000).

↓29

c) Adeno-Associated Viral Vector (AAV): AAV is a small-single stranded DNA virus, non-pathogenic, which requires a helper virus (adenovirus or herpesvirus) for replication and propagation (Somia and Verma, 2000; Mathieu et al., 2002). The AAV is capable of integrating into the host genome and can transfer genes in both quiescent and replicative cells (Mathieu et al., 2002). The problems related with these vectors are the coding capacity (restricted to 4.5 kb (Somia and Verma, 2000)), and the initial time delay before the beginning of transgene expression (Mathieu et al., 2002).

d) Adenoviral vectors: adenoviruses are double-stranded DNA viruses, with a genome flanked by inverted terminal repeats (ITRs), which serve as origins of replication. The wild type adenoviral genome possess 4 major classes of genes: early E1 gene encoding products that activate transcription of the other viral genes; the E2 gene encode proteins that enable DNA replication, while the products of the E3 gene function in counter host cell defence mechanism. Finally, the products of the E4 gene count on the regulation of the viral life cycle (Thyagarajan et al., 2001). The most commonly used replication defective adenoviruses are viruses containing deletions of the E1 and E3 regions, which are also referred as first generation vectors (Graham, 2000). Deletion of E1 renders the adenovirus replication-defective, thus E1 genes are provided in trans to allow the production of adenoviruses in cell lines. The E3 gene is normally replaced by the desired (therapeutic) transgene (reviewed in Thyagarajan et al., 2000). Their advantages for the use in gene therapy are that they allow inserts up to 7.5 kb and that they are produced at high viral titers (Wilson, 1996). Adenoviruses have the ability to transduce dividing and non-dividing cells efficiently, and although they can not integrate into the host genome, they have many advantages that make them very useful in many gene therapy clinical trials, especially in those trials where short-term expression of immunomodulatory molecules can be efficient for the induction of the protective effect, e.g. tolerance. These short-term expression is a consequence of the loss of adenoviral episomes in progeny cells, and the humoral and cellular immune response of the host against the adenoviral antigens, aiming to clear the transduced cells. In adult animals, the transgene is expressed for only a short time, between 5 and 20 days post-infection (Dai et al., 1995). Nevertheless, their ability to transduce dividing and non-dividing cells makes them one of the most used vectors for gene therapy. An overview of the immune response generated after the inoculation of an adenoviral vector is shown in Fig. 14 (from Bangari and Mittal, 2000), in the situation of a first inoculation with an adenoviral vector. Briefly, the first use of an adenoviral vector leads to a strong innate as well as adaptive immune response, resulting in the elimination of transduced cells as well as in the development of neutralizing antibodies. In the case of a high dose of administered vector, a strong innate immune response is obtained, resulting in the proinflammatory cytokines and chemokines that lead to an acute toxic response and hepatotoxicity (reviewed in Bangari and Mittal, 2000).

There are different possibilities to manipulate the immune system through gene therapy. One of them is the ex vivo gene transfer of the target molecule e.g. by using adenoviral or retroviral gene transfer, with different therapeutic genes. The advantage of both systems is the local production of the molecule of interest. The disadvantage of the adenoviral gene transfer is that it produces a transient gene expression, being the expression and effects of the molecule of interest limited. Alternatives to overcome this particular disadvantage are offered by retroviral gene transfer, since the integration of the gene of interest to the host genome allows the long term expression of the target molecule. Other possibility is the modification of certain types of cells, e.g. cells of the immune system and they are used as cellular therapy.

↓30

Fig. 14 : Development of adenoviral vector immunity.

Figure taken from Bangari and Mittal, 2000. Briefly, the application of an adenoviral vector leads to innate and adaptive immune responses that ultimately lead to the neutralization of the adenovirus as well as to the elimination of the transduced cells.

One of the most interesting target cells in this regard are T lymphocytes, since they play a key role in immune responses, and they are able to be genetically modified by retroviral vectors. In this case, the retroviral vectors bring the possibility of permanently modify the genome of the T cells, further enabling the selection of the modified cells through a selection marker. This allows the obtainig of a high number of modified T cells that are able to transmit the genetic information to other generation of T cells. Adenoviral vectors have been used so far in many different approaches such as the transfer of factor IX gene in hemophilia B dogs via vein injection (Kay et al., 1994) and in mice (Smith et al., 1993), for the transfer of the VLDL receptor gene for the treatment of familiar hypercholesterolemia in a mouse model (Kozarsky et al., 1996), and for the ex vivo transduction of T cells from ADA-deficient patients, between others (Blaese et al., 1995). In humans, trials have been overshadowed by the tragic death of a 18-year-old patient in a Phase I clinical trial in 1998 (reviewed in Somia and Verma, 2000), but nevertheless adenoviral vectors are still widely used in animal models.

The route of administration of the vector determines the efficiency of transgene expression in experimental models. In 1995, Huard et al. systematically studied the efficiency of transduction of various adenovirus-recombinants via different routes of administration: intramuscular, intracardiac, intraperitoneal, buccal, gastric, rectal, intravenous and nasal. They found out that the route of administration has a major role in determining the transduction efficiency of the different tissues (Huard et al. 1995). This should be taken into account when planning experiments tending to obtain a molecule over-expressed in a determined tissue.

1.4.2 Gene transfer to pregnant animals

↓31

In reproductive medicine, experimental trials have been made so far only for correcting gene defects in utero. The use of gene therapy for improving pregnancy-rate success or avoiding pregnancy-related diseases i.e. miscarriage or pre-eclampsia remains a very distant goal with unresolved moral and ethical aspects. However, gene therapy may be a useful tool in determining the role of genes with a proposed protective role in supporting fetal growth and/or avoiding its rejection experimentally and might further help to identify new targets of intervention. Gene therapy strategies to avoid fetal rejection may include for example the transfer and expression of cytoprotective molecules locally at the fetal-maternal interface. In addition, the ex-vivo genetic modification of immune cells and further transfer into pregnant animals for tolerance induction represents a novel and tempting approach.

Interestingly, as reported by Koi et al., the placental “receptivity” to adenovirus depends on the trophoblast differentiation stage (Koi et al. 2001). Natural infection of human villous trophoblasts by adenovirus is reduced as the cells differentiate into syncytiotrophoblasts (MacCalman et al. 1996, Parry et al. 1998). As reported by Senoo et al. (2000), the administration of the virus into the amniotic space on 14-days pregnant mice resulted in the expression of the desired gene in the lung, skin surface and epithelium of the digestive tract, while the administration of the recombinant virus into the intra-peritoneal space of the fetus resulted in gene expression in the peritoneum and upper digestive tract. Further, the same authors introduced the adenovirus directly into the systemic circulation of guinea pig fetuses through the umbilical vein and found out that the gene was expressed in multiple organs (Senoo et al. 2000), being this route the most effective one for effective fetal somatic therapy. The highest gene expression after gene transfer was found in the liver (80%). Additionally, the administration of recombinant adenovirus per se did not have a detrimental effect on the fetuses (Seeno et al. 2000). It is also possible that the low lymphocyte infiltration described by these authors (Seeno et al., 2000) is due to the immaturity of the fetus or most probably, to the tolerant state at the fetal-maternal interface. Further publications confirm that adenovirus-mediated gene transfer into the fetal systemic circulation resulted in the transgene expression primarily in the liver (Schachtner et al. 1999; Themis et al. 1999). Although successful, the diffused delivery of vector throughout the fetus, the immune response to the vector and the transfer of genetic material to the mother in this report point out the technical obstacles to be solved if systemic in utero gene therapy should became a reality for correcting fetal diseases (Themis et al. 1999, Walsh 1999).

Regarding murine placenta, Senoo already reported very low placental transgene expression if the vector was applied into the amniotic space or intra-peritoneally into the fetus on day 14. Accordingly, Laurema et al. recently reported that the injection of LacZ-adenovirus into the exocoelomic cavity of rat fetuses led to expression of the gene expression on giant cells while no transduction could be observed in fetuses or rat dams (Laurema et al. 2004). These data confirm that the exocoelomic cavity does not offer a route for gene transfer into the fetus. Fetal membranes may act as a barrier, which may naturally prevent adenoviral particles from passing between embryonic cavities (Laurema et al. 2004). Interestingly, a work by Okada and co-workers showed that placenta specific gene incorporation can be achieved by lentiviral transduction of murine blastocysts only after the removal of the zona pellucida (Okada et al., 2007).

↓32

Ex vivo gene transfer

One tempting approach to be used in this emerging field is the transfer of ex vivo generated immune cells, which are specific for paternal antigens and carry therapeutic genes as i.e. anti-inflammatory, immunomodulatory or cytokine genes. Experimental protocols from the transplantation immunology (Wood and Fry 1999; Hammer et al. 2002) suggest that the use of gene-engineered cells may be very effective in pregnancy models since they would act locally at the fetal-maternal interface and may thus prevent inflammatory events, which are known to end in miscarriage. By transferring gene-modified cells into pregnant animals in abortion or pre-eclampsia models (Chaouat et al. 1988, Zenclussen et al. 2004), gene therapy would be useful to verify the role of certain genes and to unravel novel pathways involved in pregnancy outcome. Gene therapy approaches in these models could further help designing therapies tending to avoid e.g. immunological rejection of the fetus. Other advantages of this method for using it experimentally in pregnancy models would be the possibility of localization of immunosuppressive/immunoregulatory molecules to targeted sites and the possibility of manipulation of the fetal genotype before implantation takes place since the use of retroviral vectors ensures that the genotype of the target cells in question are permanently modified.


© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 4.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML generated:
29.09.2014