Heme Oxygenase is Downregulated in Stress-Triggered and Interleukin-12-Mediated Murine Abortion


Dr P. C. Arck, Biomedizinisches Forschungszentrum, Raum 2.0549, Augustenburger Platz 1, 13353 Berlin, Germany. E-mail: petra.arck@charite.de


Heme oxygenases (HOs) are responsible for heme degradation. Besides their enzymatic activities, HOs are involved in tissue protection. Failing upregulation of HOs has been linked to increased necrosis in inflammatory tissues. Interestingly, previously published data indicated that mice exposed to sonic stress during early gestation show an augmented production of decidual inflammatory T-helper 1 (Th1) cytokines, thus resulting in increased abortion rate. No data linked the Th1-inducer interleukin (IL)-12 with the event of abortion. As little is known about the role of HO in pregnancy maintenance, we evaluated the expression of decidual and placental HO-1 and HO-2 in the abortion-prone murine mating combination CBA/J × DBA/2 J with (1) CBA/J female control mice, (2) CBA/J mice exposed to stress during early gestation and (3) CBA/J females injected with recombinant IL-12. Decidual and placental HOs protein expression was analysed by immunohistochemistry and mRNA levels by real time polymerase chain reaction (PCR).

As expected, an increased abortion rate was present in mice exposed to stress compared with the control. IL-12 injections also boosted the abortion rate compared with control mice, mimicking the effect of stress. HOs' proteins could be detected in placenta and decidua. Real time PCR revealed lower levels of HO-1 and HO-2 mRNA in stress-triggered and IL-12-injected mice. We conclude that increased Th1-cytokine levels during murine pregnancy may result in low expression of HO-1 and HO-2, thus leading to placental necrosis and foetal rejection.


Pregnancy maintenance is a very complex phenomenon, and the mechanisms that allow the survival of the foetus within the maternal uterus are still poorly understood. Since Medawar's hypothesis in 1953 [1], the foetus has been considered as an allograft, and the role of the different molecules of the immune system has been in this context intensively studied. For this reason, investigating mechanisms unravelled in the field of transplantation immunology have often proven to help us to understand the molecular and cellular pathways of either foetal rejection or pregnancy maintenance.

In transplantation immunology, endothelial cells (EC) play a crucial role in graft acceptance. It is well known that EC activation can induce two types of responses: either a ‘pro-inflammatory response’, or a ‘protective response’. During a pro-inflammatory response, EC activation appears to be the major underlying cause of xenograft rejection [2–5]. On the contrary, the activation of protective and noninflammatory genes prevents acute rejection [2, 6, 7]. The sequences encoding for heme oxygenases (HO), A20, Bcl-2 and Bcl-x have been postulated as protective genes, as their products are able to protect the cells from injury by different pathways [7, 8]. Data obtained on the ability of EC to upregulate these protective genes in graft transplantation [5, 9, 10] allow us to speculate that such an upregulation could also be vital for the survival of the ‘allograft foetus’.

The placenta is a unique organ which transports essential nutrients and, owing to the rapid growth of the foetus, also transports huge amounts of oxygen. Therefore, it is very probable that from this panel of protective enzymes which participate in transplantation success, particularly HOs play critical roles during gestation, as they are the limiting enzymes in the catabolism of heme proteins like haemoglobin.

Three isoforms of HO have been identified: HO-1, HO-2 and HO-3 [11]. HO-1 can be highly upregulated by several factors, including some pro-inflammatory T-helper 1 (Th1)-type cytokines [12], ultraviolet radiation [13], oxidative stress [14] and hypoxia [15], and is responsible for the destruction of heme from damaged red blood cells [11]. HO-2, on the contrary, does not seem to be inducible [11]. There is convincing evidence that the upregulation of HO-1 allows the acceptance of mouse allograft [5, 9, 10, 16]. Further, the upregulation of HO-1 seems to protect tissues from oxidative injury [17–20].

The immune system plays a central role in reproduction. Immunological imbalances (i.e. Th1/Th2 cytokines imbalance) are involved in reproductive failures in humans and mice [21–28]. A Th2-type response, characterized by the predominance of interleukin-4 (IL-4), IL-3, IL-10, has been associated with successful pregnancy [29–32]. On the contrary, the prevalence of an inflammatory response, or Th1-type response, characterized by enhanced production of IL-2, interferon-gamma (IFN-γ) and tumour necrosis factor-alpha (TNF-α), does not allow the foetus to survive [29, 30, 32–36]. Stress is known to induce a Th1-cytokine response; further, during pregnancy, stressed mice have enhanced abortion rates by means of an increased local production of Th1-type cytokines (i.e. IFN-γ and TNF-α) by cytotoxic T cells and γδT cells in the decidua [24, 25, 37].

Nevertheless, over the past few years, published data indicated that the Th1/Th2 cytokine ratio might not be the overall explanation for successful reproductive outcome; e.g. Svensson et al. demonstrated in experiments combining IL-4 and IL-10 knockout mice that neither maternal, nor foetal production of IL-4 or IL-10 is crucial for the completion of allogeneic pregnancies [38]. In addition, Chaouat et al. analysed the expression of several novel cytokines at the murine foeto–maternal interface and observed that the abortion-prone mating combination CBA/J × DBA/2 J expressed and secreted less IL-18 (a recently described Th1-type cytokine) than the ‘nonabortion’ mating combination CBA/J × BALB/c[39]. These data clearly question the validity of the Th1/Th2 paradigm in reproductive immunology.

IL-12 is of particular interest for the complex cytokine network as it induces the production of IFN-γ and promotes the development of a Th1-like responses [40, 41]. There are to date no published data available on the significance of the Th1-inducing cytokine IL-12 in the onset of murine abortion.

Th1-cytokines are on the one hand known to regulate HO expression and, on the other hand, involved in mediating murine abortion. Therefore, the aims of our study were:

  • 1to boost the abortion rate by employing an established model of Th1-mediated murine abortion via exposure to sonic sound stress;
  • 2to introduce a novel approach of Th1-mediated increase of abortion by injecting the mice with the Th1-inducer IL-12; and
  • 3to investigate the expression of HOs in decidua and placenta in normal murine pregnancy and Th1-mediated pregnancy failures.

Materials and methods

Animals Male DBA/2 J mice and female CBA/J mice were purchased from Charles River (Sulzfeld, Germany) and maintained in a barrier animal facility with a 12 h light/dark cycle. Animal care and experimental procedures were followed according to institutional guidelines and conformed to the requirement of the state authority for animal research conduct (LaGetSi, Berlin). Two-month-old CBA/J females were mated with 3-month-old DBA/2 J males, and the females were checked for vaginal plugs every morning. The day at which the vaginal plug was detected was considered as day 0. The pregnant females were then randomized and divided in three groups: (1) control group (n = 21); (2) stress group (n = 14); and (3) IL-12-injected group (n = 15).

In group (2), stress was induced by exposing the animals at day 5 of pregnancy to 24 h of ultrasonic sound as described previously by our group [26, 30, 37]. Mice of group (3) were injected intraperitoneally (i.p.) with 100 ng of recombinant murine (rm) IL-12 (R&D Systems, Wiesbaden-Nordenstadt, Germany) daily for 4 days (day 5, 6, 7 and 8 of pregnancy) between 8 : 00 and 10 : 00 a.m., and the dosages were suggested by Fantuzzi et al. [41]. The control group (1) received no treatment, as previous experiments revealed that the inoculation of BSA–PBS (bovine serum albumin–phophate buffered saline) buffer (the vehicle for rm IL-12) induced no change in the abortion rate, as compared with untreated animals.

Percentage of resorptions At day 13 of pregnancy, the females were sacrificed, the uteri removed and the implantation sites were documented. The resorption sites were identified by their small size and necrotic, haemorrhagic appearance, compared with normal embryos and placentas. The percentage of resorption was calculated as the ratio of resorption sites and total implantation sites (resorption plus normal implantation sites) as described previously [26, 30, 37].

Tissue preparation For immunohistochemical studies, decidual and placental samples from day 13 of gestation were isolated, washed carefully with PBS pH = 7.40, embedded in medium (Tissue Freezing Medium, Nussloch, Germany), snap frozen in liquid nitrogen and kept at −80 °C until use. Cryostat sections were cut at 8 µm, air-dried, fixed in acetone for 10 min and stored at −20 °C. For molecular biology studies, approximately 100 mg of placental and decidual tissue was washed carefully with sterile PBS pH = 7.40, snap frozen in liquid nitrogen and kept at −80 °C until RNA isolation. We exclusively harvested tissue from healthy implantation sites.

Immunohistochemistry (IHC): immunostaining for HO-1 and HO-2 IHC was performed for both molecules HO-1 and HO-2 on tissue pieces containing decidua and placenta, using our standard protocol. In brief, the sections were washed with Tris buffered saline solution (TBS) pH = 7.4 for 15 min and treated with 3% hydrogen peroxide in methanol for 30 min at room temperature to block the endogenous peroxidase activity. The tissues were then exposed to 10% normal goat serum (Dako, Hamburg, Germany) diluted in TBS for 20 min at room temperature, stained with the primary antibody (Ab) and incubated overnight at 4 °C. The primary Abs were rabbit polyclonal anti-HO-1 (Alexis, GrÏnberg, Germany) diluted at 1/100, or rabbit polyclonal anti-HO-2 (Stress Gen, Victoria, BC, Canada) diluted at 1/500 in 1% foetal calf serum (FCS), and pre-absorbed for 30 min at room temperature to reduce nonspecific binding before usage. The tissues were then washed and further stained with the secondary Ab (goat anti-rabbit horse radish peroxidase (HRP)-conjugated) (Dako) diluted in 10% mouse normal serum and 4% FCS for 1 h at room temperature. Finally, the sections were developed with diaminobenzidine (DAB, Sigma, Taufkirchen, Germany) and hydrogen peroxide, counterstained with Hemalaun (Roth, Karlsruhe, Germany) and mounted. Negative controls were obtained by replacing the primary Ab with 10% of normal goat serum or rabbit normal serum (Dako).

Total RNA Isolation One hundred microgram of tissue was treated with 1 ml Trizol (Gibco, Germany) and processed using a homogenizator (Ultra Turrax T8, Ika, Germany). The RNA was then extracted with chloroform, precipitated with absolute ethanol, washed and finally diluted in diethylpyrocarbonate (DEPC)-water. The RNA was quantified by ultraviolet absorbance at 260 nm. The integrity of the total extracted RNA was evaluated by electrophoresis in ethidium bromide-stained 1.5% agarose gels.

Reverse transcription (RT) First strain cDNA synthesis was performed using Superscript II reverse transcriptase (Perkin Elmer Biosystems, Weiterstadt, Germany) as follows: 2 µg of the extracted RNA was incubated with random hexamer primers (Roche Diagnostics, Mannheim, Germany) for 10 min at 70 °C and denaturalized at 4 °C for 1 min. The mixture was further added with desoxynucleotide triphosphate (dNTPS) (Fermentas, St. Leon-Rot, Germany), first strand buffer (Gibco), dithiothreitol (DTT) (Roche Diagnostics), water and the reverse transcriptase enzyme. The 20 µl tubes were incubated for 10 min at room temperature, 1 h at 42 °C and 10 min at 70 °C. The cDNA obtained was immediately used for traditional polymerase chain reaction (PCR) or real time PCR reactions. Control tubes included omission of Superscript of the template, or of the first strand buffer.

RT-PCR for HO-1 and HO-2 RT-PCR reactions for HO-1 and HO-2 were developed as follows: 2 µl of the RT product (cDNA) was incubated with dNTPs mixture (Fermentas), polymerase buffer, forward and reverse primers, water and the Taq-polymerase (AmpliTaq Gold, Perkin Elmer) in a final volume of 20 µl. The reactions were developed in a thermocycler (Master Cycler 5330, Eppendorf, KÎln, Germany) as follows: hot start for 10 min at 94 °C, followed by 35 cycles at 94 °C for 30 min, 60/62 °C for 45 min, ending with a last cycle of 10 min at 72 °C. As negative controls we processed RNA samples in order to evaluate putative genomic DNA contamination and water. As positive controls RT-PCR for the constitutive protein glyceraldehyde-3-phosphate (GAPDH) was performed for each sample. All the reactions were done in duplicates. The sets of primers in the three cases were designed by the authors, synthesized commercially (MWG Produkts, Ebersberg, Germany) and obtained from published sequences of murine GAPDH [42], HO-1 [43] and HO-2 [44]. The primers sequences, reaction conditions as well as primers and product size are available upon request.

Real time RT-PCR for HO-1 and HO-2 We use the recently described method of Taq Man RT-PCR [45, 46]. This assay exploits the 5′nuclease activity of AmpliTaq Gold (Perkin Elmer) DNA polymerase to cleave a fluorogenic Taq Man probe designed for HO-1 and HO-2 during PCR reaction. To normalize our samples, real time RT-PCR was also performed to detect the mRNA levels of the constitutive protein GAPDH, as suggested by various published protocols [47, 48]. The Taq Man probe contains a reporter dye at the 5′end of the probe and a quencher dye at the 3′end of the probe. During the reaction, cleavage of the probe separates the reporter dye and the quencher dye, which results in increased fluorescence of the reporter. Accumulation of PCR products is detected directly by monitoring the increasing fluorescence intensity of the reporter dye.

The sets of primers and probes were designed by the authors, from the published data for HO-1, HO-2 and GAPDH sequences, and commercially synthesized (Perkin Elmer). The primers sequences, reaction conditions and primers and product size are available upon request. All reactions were done in triplicates. Negative controls were obtained by replacing the RT product by total RNA, showing that the samples were free of contaminating DNA for all the samples, or water. The specificity of the products was confirmed by having them commercially sequenced (Agowa, Berlin, Germany).

The PCR results were obtained as exportable computer data. For each probe, the amplification plots were obtained and analysed. ‘cT’ is the cycle number at which the amplification reaction begins. Each probe was normalized to GAPDH by calculating the difference between the cT for GAPDH and the cT for HO, as:


As a lower ΔcT means more cDNA, which was obtained from mRNA, and taking into account the exponential increase of cDNA in PCR, the initial mRNA quantity was calculated by using the formula described as:


*The source of the amplified cDNA was mRNA; **inverse ratio – as a lower ΔcT means more cDNA; and ***the amplification product increases exponentially.

Microscopic evaluation The pattern of staining in the different cell types of placenta and decidual samples was evaluated using a light microscope (Zeiss Axiophot, Jena, Germany), and a semiquantitative analysis was done as described below.

Data analysis and statistics The percentage of resorptions was expressed as mean ± standard error of the mean (SEM), and the differences between the groups were calculated using the nonparametric Kruskal–Wallis test followed by Mann–Whitney U-test. In the IHC study, the qualitative analysis was done taking into account the following staining patterns: negative staining (–), weak (+), moderate (++), high (+++), intense (++++) in each placental cellular type, as well as in the decidua; the scores were then transferred into numbers for the semiquantitative analysis, scoring (–) as 0, (+) as 1, (++) as 2, etc. The numbers were then compared by using Kruskal–Wallis nonparametric test followed by Mann–Whitney U-test.

The mRNA levels, evaluated by real time RT-PCR for HO-1 and HO-2, normalized to GAPDH, and represented as 1/2ΔcT, were calculated for each sample, so that the mean and SEM were obtained for each group. The differences between group means were also evaluated using Kruskal–Wallis test, followed by Mann–Whitney U-test.

In all cases, P < 0.05 was considered a statistically significant difference.


Influence of stress or IL-12 injection on abortion rate

As evaluated at day 13 of pregnancy, stress significantly enhanced the abortion rate, as compared with unstressed animals. This is in accordance with previously published data on the CBA/J × DBA/2 J model [26, 30, 37]. Interestingly, syngeneic mating combinations, i.e. CBA/J × CBA/J mating, have no increased abortion rate (unpublished observations, A.C.Z.). Injections of rm IL-12 boosted the abortion rate as well; the increase in the abortion rate was comparable to the increase observed after stress application (Fig. 1). It should also be noted that the administration of rm IL-12 produced splenomegaly in all mice, noticed by macroscopic observation at day 13 of pregnancy.

Figure 1.

Abortion rates, as calculated by the number of resorptions to total implantation sites. The stress group, as well as the interleukin (IL)-12 treated group had significantly higher abortion rates in comparison with the control group. * = P < 0.05 and ** = P < 0.01, as evaluated by the nonparametric Kruskal–Wallis H test.

Protein expression of HO-1 and HO-2 in decidual and placental tissue

To analyse the expression of HO-1 and HO-2 protein, IHC was performed in samples containing placenta and decidua, and we observed that not only the EC at the foeto–maternal interface were HO-1+ and HO-2+ but also the various murine placental cell types and decidual cells. Figure 2 shows typical fields in placenta and decidua for HO-1. Similar images could be observed when analysing the expression of HO-2. The patterns of expression for both proteins varied in the different cell types analysed, i.e. the giant cells and the spongiotrophoblasts were intensively positive (Fig. 2A,B for HO-1). Labyrinthic trophoblast cells were positive (Fig. 2C for HO-1). Decidual cells, as well as myometrial cell were intensively positive (Fig. 2D for HO-1). We further analysed the intensity of staining separately for each cell type, but no differences could be observed between the three groups, as resumed in Tables 1 and 2. As no differences could be observed using IHC, which may be owing to the variety of cell types in placenta and the low sensitivity of this technique, we decided to evaluate the mRNA levels for both enzymes in placental and decidual tissues from the three groups.

Figure 2.

Immunohistochemistry (IHC) of heme oxygenase (HO)-1+ cells at the murine foeto–maternal interface. All figures (A–E) are representative of the whole tissue staining. (A) HO-1+ spongiotrophoblast cells (SP), brown staining, among HO-1 stromal cells (blue cells). (B) Arrows indicate HO-1+ giant cells (GC) appearing highly positive for HO-1. C depicts a typical field for positive HO-1 staining pattern (brown staining) in labyrinthic trophoblasts (L). (D) Picture of HO-1+ decidual cells (D), and E is the negative control of staining.

Table 1.  Semiquantitative analysis of the heme oxygenase (HO)-1 expression in the different cell types in murine decidua and placental cell types for the three animals groups included in the present study. +, ++, +++, ++++ are the intensity scores. Data are presented as the mean ± standard error of the mean (SEM). No differences could be observed between the groups when analysed with Kruskal–Wallis H nonparametric multiple comparison test
GroupHO-1+ giant cells (placenta)HO-1+ labyrinthic cells (placenta)HO-1+ spongiotrophoblasts (placenta)HO-1+ vessels (placenta+decidua)HO-1+ decidual cells
Control group (n=21)1.22 ± 0.151.50 ± 0.261.80 ± 0.201.45 ± 0.201.27 ± 0.19
Stress group (n=14)1.33 ± 0.331.14 ± 0.262.16 ± 0.301.28 ± 0.182.14 ± 0.14
IL-12-treated group (n=15)1.61 ± 0.251.29 ± 0.161.28 ± 0.301.57 ± 0.351.43 ± 0.21
Table 2.  Semiquantitative analysis of the heme oxygenase (HO)-2 expression in the different cell types in murine decidua and placental cell types for the three animals groups included in the present study. +, ++, +++, ++++ are the intensity scores. Data are expressed as the mean ± standard error of the mean (SEM). No differences could be observed between the groups when using Kruskal–Wallis H nonparametric multiple comparison test
GroupHO-2+ giant cells (placenta)HO-2+ labyrinthic cells (placenta)HO-2+ spongiotrophoblasts (placenta) HO-2+ vessels (placenta+decidua)HO-2+ decidual cells
Control group (n=21)2.00 ± 0.331.80 ± 0.252.20 ± 0.130.82 ± 0.181.27 ± 0.14
Stress group (n=14)1.38 ± 0.181.22 ± 0.152.22 ± 0.281.40 ± 0.161.40 ± 0.22
IL-12-treated group (n=15)2.00 ± 0.361.17 ± 0.272.17 ± 0.452.00 ± 0.381.18 ± 0.28

Decidual and placental HO-1 and HO-2 mRNA detection

Our results indicate that murine placenta and decidua at the beginning of late pregnancy contain mRNA transcripts for both HOs. The intensity of the bands obtained for the samples was variable, and not comparable within the same group. RT-PCR reactions for GAPDH were also included for each sample in order to realize a densitometrical semiquantitative analysis as already described by many other authors [49–52]. Homogeneous and comparable bands were obtained for GAPDH in all samples. However, it was not possible to detect if there was any difference between the groups with respect to HOs owing to the heterogeneity of the results for different amplification bands between a group, as it can be observed in Fig. 3. Therefore, we decided to analyse quantitatively the levels of mRNA for HO-1 and HO-2 by using the novel technique real time RT-PCR.

Figure 3.

Reverse transcription polymerase chain reaction (RT-PCR). The lower right and lower left panels are the amplification bands obtained for the positive control glyceraldehyde-3-phosphate (GAPDH). The upper left panel is the amplification bands for heme oxygenase (HO)-1 and upper right is for HO-2. A and B are samples from placental tissue (control group), B is a placental sample of the stress group, D and E are decidual samples from the control group, and F and G are samples from the decidua of the stress group.

Decidual and placental HO-1 and HO-2 mRNA quantification

We observed that placental and decidual tissues from all groups had approximately 16 times more mRNA for HO-1 and HO-2 than murine spleen tissue (data not shown), which was used as positive control because the highest HO levels had been reported for spleen among adult tissue [11, 53].

The real time PCR analysis revealed lower levels of mRNA for HO-1 in placenta and decidua from stressed animals (with high abortion rate) compared with the control, although the difference was not statistically significant, as presented in Fig. 4A. HO-2 mRNA levels were also diminished in placenta and decidua of stress-exposed animals compared with the tissues from the control group (Fig. 4B); again, no significant differences could be observed.

Figure 4.

(A) The mRNA levels for heme oxygenase (HO)-1 from the placental samples (control group n = 11; stress group n = 11; interleukin (IL)-12-treated n = 12) as well as from the decidual samples (the same n as placental samples) were calculated as described by using the formula cited in Materials and Methods. * = P < 0.05 when analysed by the Tukey multiple comparison test. (B) HO-2 mRNA levels from placenta and decidua (the same n as for HO-1 determinations). * = P < 0.05 and ** = P < 0.01 as analysed by the Kruskal–Wallis multiple comparison test.

When analysing the tissues from those animals which received rm IL-12 injections (which had also enhanced abortion rates), we observed decreased HO-1 mRNA levels in the placenta and significantly decreased HO-1 mRNA levels in the decidua. Significantly decreased mRNA HO-2 levels were observed in both placenta and decidual tissue compared with the animals from the control group (Fig. 4A,B).


The presence of HO-1 and HO-2 had previously been reported in rat, guinea pig and human placenta during normally progressing pregnancies; however, very little is known about the function of HO-1 and HO-2 in pregnancy maintenance [49–54]. In the present study, we were able to show that mouse placental and decidual tissue also express both enzymes, HO-1 and HO-2, and that mRNA transcripts for HO-1 and HO-2 are present during late gestation. In our immunohistochemical study, positive staining of HO-1 and HO-2 protein was observed not only in EC, as expected, but also on the different trophoblast types of the placenta and in decidual cells as well. Similar results were obtained by Ihara et al. [53] in an IHC study in rat placenta.

To evaluate the possible regulation of HO during pregnancy, we included in our current study the evaluation of HO expression in mice undergoing stress-triggered foetal resorption and in mice which received high doses of Th1-inducing cytokine IL-12.

Stressed mice presented, as expected, increased abortion rates. The abortion events in our study were in fact accompanied by increased local levels of IFN-γ and TNF-α (unpublished material), supporting previous reports [26, 30, 37]. IL-12 is known as a potent Th1-inducer, and its in vitro and in vivo effect of enhancing TNF-α, IFN-γ and IL-18 expression was already described [40, 41]. In contrast to TNF-α and IFN-γ, to date no convincing evidence that IL-12 could be detrimental for pregnancy outcome has been published. The inoculation of IL-12 induced an augmented abortion rate, confirming for the first time its negative impact on pregnancy outcome. Abortion rate was comparable with those observed after exposing the mice to sonic stress, and was accompanied by high levels of IFN-γ but not of TNF-α mRNA levels in the decidua (unpublished material).

In the present study, we observed that in Th1-mediated abortions (stress, as well as IL-12 injections), the HO-1 and HO-2 mRNA levels were diminished in placental and decidual tissues. No differences were detectable with respect to HOs' protein expression, as identified by IHC, because this method is clearly not sensitive enough. Low levels of HO have a high potential to be harmful in tissues at the foetal–maternal interface, as a failure in the upregulation of HO leads to large amounts of free heme [55], thus damaging the tissue. Unbound heme is per se toxic [56]. In addition, it leads to an increased expression of adhesion molecules in EC, thus allowing a massive migration of activated immune cells into the tissue, which eventually results in tissue inflammation and injury [56, 57]. Additional data [58] support this concept by describing that insufficient amounts of HO-1 render cells unable to defend themselves against a high amount of heme. Heme readily incorporates into EC, leading to oxidative injury and enhanced adhesion molecule expression [59]. This then amplifies the leucocyte-derived oxidant damage, exacerbating the damage to cell membranes and other cell constituents [60].

HO-1 and HO-2 mRNA levels in the placenta, as well as in the decidua, were about 16 times more than in adult murine spleen, where the highest activity of HO was observed in adult tissues [11, 53]. Similar results were reported for rat placenta [53]. These observations confirm that invasive trophoblast cells, where the exchange between foetal and maternal circulation takes place, express HOs responsible for heme degradation. Foetal heme is therefore degraded in the placenta, avoiding heme accumulation or recirculation, which could be extremely toxic for the mother and for the foetus. In humans, diminished HO-2 placental protein expression was described in pre-eclamptic placentas as well as in placenta with foetal growth restriction [49]. All these data strongly suggest the clue role of HOs in the oxygen placental transport.

Oxidative injury of the endothelium – mediated by free heme between others – has been implicated in the pathogenesis of a number of pathological conditions, including haemorrhagic injury and ischaemic/reperfusion injury [61, 62]. Interestingly, the mice abortion is characterized by haemorrhagic sites, and the upregulation of molecules such as the novel prothrombinase fgl-2 could be implicated [63]. It can be suggested that diminished HO expression in reproductive tissues (after stress or Th1-injection) leads to a chain reaction in which many molecules are implicated, which results finally in foetal rejection owing to haemorrhagic events.

Analysing the results obtained in the present study, we believe that the prolonged inflammatory situation in Th1-mediated abortion fails to upregulate the HO expression – or can even downregulate its expression.

Although the HO-1 and HO-2 mRNA levels were lower in both treated groups, stressed mice and mice which received IL-12, only for those Th1-injected mice the difference was highly significant. Stress induces, as has been widely demonstrated [24, 25, 37], a Th1-type response that leads to abortion, but this physiologic response to stress is not comparable with aggressive local and repeated injections of high doses of a Th1-cytokine. Further, the injection of IL-12 also induced splenomegaly, and therefore, the systemic Th1-cytokines levels are expected to be augmented.

In the present work, we observed low levels of HOs during Th1-mediated abortion (IL-12 or stress-triggered abortion). As hypothetically depicted in Fig. 5, the downregulated expression of HOs leads to enhanced levels of free heme, which are known to be cytotoxic for EC, as well as for the foeto–maternal interface cells. Free heme also upregulates the expression of adhesion molecules, allowing the lymphocyte migration into the foeto–maternal interface. Additional work is needed to unravel additional mechanisms that could be implicated in the downregulation of HOs and further leucocyte migration in the inflammatory setting of murine abortion.

Figure 5.

Hypothetical scenario that depicts possible pathways by which the downregulation of haeme oxygenase (HO) could be implicated in abortion. (A) During normal pregnancy, heme from damaged and aged erythrocytes is degraded into bilirubin by the enzyme HO, which is produced by endothelial cells (EC), as well as by decidual and placental cells. The degradation of free heme by HOs avoids its accumulation and further toxic effects at the foeto–maternal interface. (B) During T helper 1 (Th1)-mediated abortion (stress or interleukin (IL)-12 injections), enhanced levels of Th1-cytokines result in the failure of decidual, trophoblast and/or EC to upregulate the expression of HOs, leading to increased levels of free heme at the maternal–foetal interface. Free heme enhances the expression of adhesion molecules, which allow further trafficking of Th1-lymphocytes into the tissue. Free heme itself is also a cytotoxic factor, not only for EC, but also for decidual and placenta cells, leading finally to haemorrhage and foetal rejection.


The work presented in this paper was supported by grants from the German Science Foundation (DFG; Ar232/8–1) and the Charité to PCA. ACZ was funded by the Alexander von Humboldt Foundation. The authors would like to thank Stefan Fest for his excellent assistance in the graphic design and Dr Otto B. Walter for his advice on developing the formula used to describe mRNA levels.